United States Office of Research EPA 600/P-99/002a
Environmental Protection and Development October 1999
Agency Washington, DC 20460 External Review Draft
Air Quality Criteria for
Particulate Matter
Volume I
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on
its technical accuracy and policy implications.
-------
EPA 600/P-99/002a
October 1999
External Review Draft
Air Quality Criteria for
Particulate Matter
Volume I
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
Disclaimer
This document is an external review draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
October 1999 I-ii DRAFT-DO NOT QUOTE OR CITE
-------
Preface
National Ambient Air Quality Standards (NAAQS) are promulgated by the United States
Environmental Protection Agency (U.S. EPA) to meet requirements set forth in Sections 108 and
109 of the U.S. Clean Air Act (CAA). Sections 108 and 109 require the EPA Administrator:
(1) to list widespread air pollutants that may reasonably be expected to endanger public health or
welfare; (2) to issue air quality criteria for them which assess the latest available scientific
information on nature and effects of ambient exposure to them; (3) to set "primary" NAAQS to
protect human health with adequate margin of safety and to set "secondary" NAAQS to protect
against welfare effects (e.g., effects on vegetation, ecosystems, visibility, climate, manmade
materials, etc); and (5) to periodically (every 5-yrs) review and revise, as appropriate, the criteria
and NAAQS for a given listed pollutant or class of pollutants.
The original U.S. NAAQS for particulate matter (PM), issued in 1971 as "total suspended
particulate" (TSP) standards, were revised in 1987 to focus on protecting against human health
effects associated with exposure to ambient PM less than 10 microns (< 10 |um) that are capable
of being deposited in thoracic (tracheobronchial and alveolar) portions of the lower respiratory
tract. Later periodic reevaluation of newly available scientific information, as presented in the
last previous version of this "Air Quality Criteria for Particulate Matter" document published in
1996, provided key scientific bases for PM NAAQS decisions published in July 1997. More
specifically, the PM10 NAAQS set in 1987 (150 |ug/m3, 24-h; 50 |ug/m3, annual ave.) were
retained in modified form and new standards (65 |ug/m3, 24-h; 15 |ug/m3, annual ave.) for
particles < 2.5 |um (PM25) were promulgated in July 1997.
This First External Review Draft of revised Air Quality Criteria for Particulate Matter
assesses new scientific information that has become available since early 1996 through mid-
1999. Extensive additional pertinent information is expected to be published during the next 6 to
9 months (including results from a vastly expanded U.S. EPA PM Research program and from
other Federal and State Agencies, as well as other partners in the general scientific community)
and, as such, the findings and conclusions presented in this draft document must be considered
only provisional at this time. The present draft is being released for public comment and review
by the Clean Air Scientific Advisory Committee (CASAC) mainly to obtain comments on the
October 1999 I-iii DRAFT-DO NOT QUOTE OR CITE
-------
organization and structure of the document, the issues addressed, and the approaches employed
in assessing and interpreting the thus far available new information on PM exposures and effects.
Public comments and CASAC review recommendations will be taken into account, along with
newly available information published or accepted for peer-reviewed publication by April/May
2000, in making further revisions to this document for incorporation into a Second External
Review Draft. That draft is expected to be released in June 2000 for further public comment and
CASAC review (September 2000) in time for final revisions to be completed by December
2000). Evaluations contained in the present document will be drawn upon to provide inputs to
associated PM Staff Paper analyses prepared by EPA's Office of Air Quality Planning and
Standards (OAQPS) to pose options for consideration by the EPA Administrator with regard to
proposal and, ultimately, promulgation by July 2000 of decisions on potential retention or
revision of the current PM NAAQS.
This document was prepared and reviewed by experts from Federal and State government
agencies, academia, industry, and NGO's for use by EPA in support of decision making on
potential public health and environmental risks of ambient PM. It describes the nature, sources,
distribution, measurement, and concentrations of PM in both the outdoor (ambient) and indoor
environments and evaluates the latest data on the health effects in exposed human populations, as
well as environmental effects on: vegetation and ecosystems; visibility and climate; manmade
materials; and associated economic impacts. Although not intended to be an exhaustive literature
review, this document is intended to assess all pertinent literature through mid-1999.
The National Center for Environmental Assessment - Research Triangle Park, NC
(NCEA-RTP) acknowledges the contributions provided by authors, contributors, and reviewers
and the diligence of its staff and contractors in the preparation of this document.
October 1999 I-iv DRAFT-DO NOT QUOTE OR CITE
-------
Air Quality Criteria for Particulate Matter
VOLUME I
1. EXECUTIVE SUMMARY 1-1
2. INTRODUCTION 2-1
3. PHYSICS, CHEMISTRY, AND MEASUREMENT OF
PARTICULATE MATTER 3-1
4. CONCENTRATIONS, SOURCES, AND EMISSIONS OF
ATMOSPHERIC PARTICLES 4-1
APPENDIX 4A: Composition of Particulate Matter Source
Emissions 4A-1
5. HUMAN EXPOSURE TO AMBIENT PARTICULATE MATTER:
RELATIONS TO CONCENTRATIONS OF AMBIENT AND
NON-AMBIENT PARTICULATE MATTER AND OTHER AIR
POLLUTANTS 5-1
APPENDIX 5A: Nomenclature 5A-1
VOLUME II
6. EPIDEMIOLOGY OF HUMAN HEALTH EFFECTS FROM
AMBIENT PARTICULATE MATTER 6-1
7. DOSIMETRY AND TOXICOLOGY OF PARTICULATE
MATTER 7-1
8. INTEGRATIVE SYNTHESIS OF KEY POINTS: PARTICULATE
MATTER EXPOSURE, DOSIMETRY, AND HEALTH RISKS 8-1
VOLUME III
9. ENVIRONMENTAL EFFECTS OF PARTICULATE MATTER 9-1
October 1999 I-v DRAFT-DO NOT QUOTE OR CITE
-------
Table of Contents
Page
List of Tables I-xi
List of Figures I-xv
Authors, Contributors, and Reviewers I-xxi
U.S. Environmental Protection Agency Project Team for Development of
Air Quality Criteria for Particulate Matter I-xxix
1. EXECUTIVE SUMMARY 1-1
1.1 INTRODUCTION 1-1
1.1.1 Purpose of the Document 1-1
1.1.2 Organization of the Document 1-2
1.2 AIR QUALITY AND EXPOSURE ASPECTS 1-2
1.2.1 Chemistry and Physics of Atmospheric Particles 1-3
1.2.2 Sources of Airborne Particles in the United States 1-5
1.2.3 Atmospheric Transport and Fate of Airborne Particles 1-6
1.2.4 Airborne Particle Measurement Methods 1-6
1.2.5 Ambient U.S. PM Concentrations: Regional Patterns and Trends 1-7
1.2.6 Human PM Exposure 1-7
1.3 DOSIMETRY 1-9
1.4 PARTICULATE MATTER HEALTH EFFECTS 1-11
1.4.1 Epidemiology Findings 1-11
1.4.2 Toxicologic Responses to PM in Animals and Humans 1-13
1.4.3 Population Groups at Risk 1-15
1.5 ENVIRONMENTAL EFFECTS 1-15
1.5.1 Vegetation and Ecosystem 1-15
1.5.2 Particulate Matter-Related Effects on Materials 1-19
1.5.3 Particulate Matter-Related Effects on Visibility 1-20
1.5.4 Environmental and Economic Impacts of PM 1-21
2. INTRODUCTION 2-1
2.1 LEGISLATIVE REQUIREMENTS 2-1
2.2 HISTORY OF PREVIOUS PM CRITERIA/STANDARDS REVIEWS 2-2
2.2.1 The 1997 PM NAAQS Revision 2-3
2.2.2 Presidential Memorandum: Next Particulate Matter Review
and Research 2-5
2.3 CURRENT PM CRITERIA AND NAAQS REVIEW 2-6
2.3.1 Criteria Review and Plans and Schedule 2-6
2.3.2 Methods And Procedures for Document Preparation 2-8
2.3.3 Approach 2-10
2.3.4 Key Issues of Concern 2-11
2.4 DOCUMENT CONTENT AND ORGANIZATION 2-14
REFERENCES 2-16
October 1999 I-vii DRAFT-DO NOT QUOTE OR CITE
-------
Table of Contents
(cont'd)
3. PHYSICS, CHEMISTRY, AND MEASUREMENT OF PARTICULATE MATTER ... 3-1
3.1 PHYSICS AND CHEMISTRY OF PARTICULATE MATTER 3-1
3.1.1 Definitions 3-1
3.1.2 Physical Properties and Processes 3-2
3.1.2.1 Definitions of Particle Diameter 3-2
3.1.2.2 Aerosol Size Distributions 3-3
3.1.2.3 Nuclei-Mode Particles 3-15
3.1.3 Chemistry of Atmospheric PM 3-18
3.1.3.1 Chemical Composition and Its Dependence on Particle Size . . . 3-19
3.1.3.2 Primary and Secondary Particulate Matter 3-19
3.1.3.3 Particle-Vapor Partitioning 3-25
3.1.3.4 Removal Processes 3-29
3.1.3.5 Particulate Matter and Acid Deposition 3-30
3.1.3.6 Particles as Carriers of Toxic Species 3-31
3.1.3.7 Separation of Fine and Coarse Particles 3-33
3.2 MEASUREMENT OF PARTICULATE MATTER 3-34
3.2.1 Problems in Measuring Particulate Matter 3-36
3.2.1.1 Treatment of Semivolatile Components of PM 3-37
3.2.1.2 Upper Cut Point 3-37
3.2.1.3 Separation of Fine-Mode and Coarse-Mode PM 3-39
3.2.1 A Treatment of Pressure, Temperature, and Relative Humidity . . . 3-40
3.2.1.5 No Way to Determine Accuracy 3-41
3.2.2 Why Measure Particles 3-43
3.2.2.1 Attainment of a Standard 3-43
3.2.2.2 Implementation of a Standard 3-43
3.2.2.3 Determination of Health Effects 3-43
3.2.2.4 Determination of Ecological Effects 3-44
3.2.2.5 Determination of Radiative Effects 3-44
3.2.2.6 PM Components/Parameters Which Need To Be Measured .... 3-44
3.2.3 Problems Associated with Semivolatile Particulate Matter 3-44
3.2.3.1 Particulate Nitrates 3-46
3.2.3.2 Semivolatile Organic Compounds 3-49
3.2.3.3 Use of Denuder Systems To Measure Semivolatile
Compounds 3-55
3.2.3.4 Particle-Bound Water 3-61
3.2.4 EPA Monitoring Programs 3-67
3.2A.I The Federal Reference Methods for Equilibrated Mass 3-67
3.2.4.2 Speciation Monitoring 3-69
3.2.5 Continuous Monitoring 3-77
3.2.5.1 TEOM 3-77
3.2.5.2 RAMS 3-78
October 1999 I-viii DRAFT-DO NOT QUOTE OR CITE
-------
Table of Contents
(cont'd)
3.2.5.3 CAMM 3-78
3.2.5.4 Light Scattering 3-80
3.2.5.5 Beta-Gauge Techniques 3-81
3.2.5.6 Measurements of Individual Particles 3-81
3.3 SUMMARY 3-83
REFERENCES 3-90
4. CONCENTRATIONS, SOURCES, AND EMISSIONS OF ATMOSPHERIC
PARTICLES 4-1
4.1 INTRODUCTION 4-1
4.2 TRENDS AND PATTERNS IN AMBIENT PARTICULATE MATTER PM2 5
CONCENTRATIONS AND TRENDS 4-2
4.2.1 Daily and Seasonal Variability 4-7
4.2.2 Relations Between Mass and Chemical Component Concentrations 4-12
4.2.3 Spatial Variability 4-15
4.2.4 Urban Concentrations and Patterns from the New PM2 5 Compliance
Network 4-20
4.2.5 Trends and Patterns 4-20
4.2.5.1 Visual Range/Haziness 4-20
4.2.5.2 Urban Trends 4-22
4.3 SOURCES OF PRIMARY AND SECONDARY PARTICULATE
MATTER 4-22
4.3.1 Source Contributions to Ambient PM 4-27
4.4 EMISSIONS ESTIMATES AND THEIR UNCERTAINTIES 4-33
4.4.1 Emissions Estimates for Primary Particulate Matter and SO2, NOX,
and VOCs in the United States 4-33
4.4.2 Uncertainties of Emissions Inventories 4-39
4.5 LONG RANGE TRANSPORT OF PM FROM SOURCES OUTSIDE
THE UNITED STATES 4-41
4.6 SUMMARY AND CONCLUSIONS 4-42
REFERENCES 4-45
APPENDIX 4A: Composition of PM Source Emissions 4A-1
REFERENCES 4A-20
5. HUMAN EXPOSURE TO AMBIENT PARTICULATE MATTER: RELATIONS
TO CONCENTRATIONS OF AMBIENT AND NON-AMBIENT PM AND
OTHER AIR POLLUTANTS 5-1
5.1 INTRODUCTION AND BASIC CONCEPTS 5-1
5.1.1 The History of Understanding Human Exposure to Particulate Matter .... 5-5
5.1.1.1 Caveat 5-10
5.1.2 Exposure to PM of Ambient Origin and Total PM 5-10
October 1999 I-ix DRAFT-DO NOT QUOTE OR CITE
-------
Table of Contents
(cont'd)
5.2 EXPOSURES TO PM CONCENTRATIONS IN THE COMMUNITY
AMBIENT ATMOSPHERE 5-14
5.3 EXPOSURES TO AMBIENT PM IN INDOOR MICROENVIRONMENTS . . . 5-19
5.4 EXPOSURES TO PM OF INDOOR ORIGIN 5-24
5.5 PERSONAL EXPOSURES TO PM OF ONE'S OWN PERSONAL
ACTIVITIES 5-27
5.6 PERSONAL PM EXPOSURE 5-28
5.7 EXPOSURE TO PM OF AMBIENT ORIGIN IN BOTH INDOOR AND
OUTDOOR MICROENVIRONMENTS 5-29
5.7.1 Estimation of the Daily Exposure to PM of Ambient Origin 5-37
5.8 AMBIENT P ARTICULATE MATTER CONCENTRATION ASA
SURROGATE FOR EXPOSURE TO PARTICULATE MATTER
OF AMBIENT ORIGIN 5-44
5.9 CONCENTRATIONS OF AMBIENT PM FOUND INDOORS AND
IN OTHER NON-AMBIENT ENVIRONMENTS 5-45
5.10 PERSONAL PM MONITORING STUDIES AND FACTORS THAT
INFLUENCE THEIR ABILITY TO ESTIMATE RELATIONSHIPS
TO EXPOSURE TO PM OF AMBIENT ORIGIN 5-53
5.11 EXPOSURES TO AMBIENT PM25 OF PEOPLE BELIEVED TO BE
SUSCEPTIBLE TO THE EFFECTS OF AMBIENT PM25 5-67
5.12 PERSONAL EXPOSURE TO CONSTITUENTS OF PM: SULFATES
AND ACIDITY 5-72
5.13 CORRELATION OF AMBIENT PM AND PM EXPOSURE IN
CROSS-SECTIONAL-TYPE STUDIES 5-77
5.14 EXPOSURE TO AMBIENT GASEOUS POLLUTANTS RELATED TO
AMBIENT PM 5-80
5.15 CONFOUNDING BY INDOOR PM 5-89
5.16 IMPLICATIONS OF THE AMBIENT PM EXPOSURE RELATIONSHIPS
FOR EPIDEMIOLOGIC ANALYSIS AND A SUMMARY OF THE
CHAPTER CONCLUSIONS 5-95
REFERENCES 5-97
APPENDIX 5-A: Nomenclature 5A-1
October 1999 I-x DRAFT-DO NOT QUOTE OR CITE
-------
List of Tables
Number Page
1-1 Provisional Conclusions About Alternative Hypotheses That May Affect the
Synthesis of Epidemiology Studies 1-12
2-1 Schedule for Development of the Current Revised Particulate Matter Air
Quality Criteria Document 2-7
3-1 Comparison of Ambient Particles, Fine Mode and Coarse Mode 3-35
3-2 PM Components/Parameters of Interest for Health, Ecological, or
Radiative Effects; for Source Category Apportionment Studies; or for
Air Quality Validation Studies 3-45
4-1 Concentrations of PM25 and Selected Elements in the PM25 Size Range
and Correlations Between Elements and PM25 Mass 4-14
4-2 Correlation Coefficients for Spatial Variation of PM2 5 Mass and Different
Sources for Pairs of Sampling Sites in SoCAB (1986) 4-18
4-2a Correlation Coefficients for Spatial Variation of PM25 Mass and Different
Components for Pairs of Sampling Sites in Philadelphia (1994) 4-19
4-3 Constituents of Atmospheric Particles and Their Major Sources 4-25
4-4 Receptor Model Source Contributions to PM2 5 4-28
4-5 Receptor Model Source Contributions to PM10 4-29
4A-1 Average Abundances of Major Elements in Soil and Crustal Rock 4A-2
4A-2 Composition of Fine Particles Released by Various Stationary Sources in
the Philadelphia Area 4A-6
4A-3 Fractional Organic and Elemental Carbon Abundances in Motor Vehicle
Emissions 4A-10
4A-4 Phoenix PM25 Motor Vehicle Emissions Profiles 4A-12
4A-5 Emission Rates for Constituents of Particulate Matter from Gasoline
and Diesel Vehicles 4A-13
October 1999 I-xi DRAFT-DO NOT QUOTE OR CITE
-------
List of Tables
(cont'd)
Number Page
4A-6 Mean Aerosol Composition at Tropical Site Affected Heavily by
Biomass Burning Emissions 4A-17
5-1 Summary of the Mean Values of PM Variables From the PTEAM Study 5-33
5-2 Average Fractions of Time Spent Outdoors or in a Vehicle in the PTEAM
Study and NHAPS Study 5-38
5-3 Relative Risk of Nonmalignant Respiratory Mortality From Increasing
Time Spent Outdoors 5-40
5-4 Summary of Correlations Between PM10 Personal Exposures of Seven
Tokyo Residents and the PM10 Measured Outdoors Under the Eaves of
Their Homes, and the PM Measured at the Itabashi Monitoring Station 5-47
5-5 Summary of PM Data During the Summer in an Indoor/Outdoor Study
in Southwest Virginia 5-51
5-6 Seasonal Variation of Regression Coefficients for PM10 With Fine and
Coarse Mode Fractions 5-53
5-7 Average Levels of Personal Exposures and Outdoor Concentrations and
the Correlation Between Them in Longitudinal Exposure Studies 5-59
5-8 Personal Exposures to Particulate Matter of Non-Smoke Exposed Children
in Wageningen, NL, and Stationary Ambient Monitoring Data 5-61
5-9 Personal PM10 Exposure of 13 Adults in Amsterdam, NL 5-63
5-10 Personal PM10 Exposure of 24 Adults in Amsterdam, NL 5-64
5-11 Personal Exposures to PMj 5 of Five Elderly Residents of a Retirement
Center in Baltimore, MD, as a Function of Ambient PM25 Concentration 5-69
5-12A Summary of Correlations Between Daytime Personal PM Exposures
and Daytime Outdoor PM Concentrations for 15 COPD Patients During
Combined Summer and Winter Periods in Boston, MA 5-71
5-12 Example of a Completed Matrix for a Type-1 Analysis of PM Exposure
and Ambient PM Concentration 5-77
October 1999 I-xii DRAFT-DO NOT QUOTE OR CITE
-------
List of Tables
(cont'd)
Number Page
5-13 Example of an Incomplete Matrix for Type-2 Analysis of PM Exposure
and Ambient PM Concentration 5-78
5-14 Results of the Analysis of the PTEAM Daily Average Personal PM10
Exposure Data from Riverside, California 5-79
5-15 Average Annual Correlations Between PM and Criteria Gaseous Pollutants
in the United States 5-87
October 1999 I-xiii DRAFT-DO NOT QUOTE OR CITE
-------
List of Figures
Number Page
3-1 Number of particles as a function of particle diameter 3-5
3-2 Particle volume distribution as a function of particle diameter 3-6
3-3 Distribution of coarse, accumulation, and nuclei or ultrafine mode
particles by three characteristics, volume, surface area, and number
for the grand average continental size distribution 3-7
3-4 Volume size distribution, measured in traffic, showing fine-mode and
coarse-mode particles and the nuclei and accumulation modes within
the fine-particle mode 3-9
3-5 An idealized distribution of ambient particulate matter showing
fine-mode particles and coarse-mode particles and the fractions
collected by size-selective samplers 3-10
3-6 Specified particle penetration through an ideal inlet for five different
size-selective sampling criteria 3-12
3-7 Comparison of penetration curves for two PM10 beta gauge samplers
using cyclone inlets 3-14
3-8 Particle growth curves showing fully reversible hygroscopic growth of
sulfuric acid particles, deliquescent growth of ammonium sulfate
particles at about 80% relative humidity (RH), hygroscopic growth of
ammonium sulfate solution droplets at RH greater than 80%, and hysteresis
until the crystallization point is reached 3-28
3-9 Theoretical predictions and experimental measurements of growth of
NH4HSO4 and particles at relative humidity between 95 and 100% 3-29
3-10 Schematic showing major nonvolatile and semivolatile components of PM2 5 ... 3-38
3-11 Amount of ammonium nitrate volatilized from Teflon filters, expressed
as a percentage of the measured PM25 mass, for the SCAQS and CalTech
studies, for spring and fall sampling periods 3-47
3-12 Aerosol water content expressed as a mass percentage, as a function of
relative humidity 3-64
October 1999 I-xv DRAFT-DO NOT QUOTE OR CITE
-------
List of Figures
(cont'd)
Number Page
3-13 This thermogram, for a sample containing rock dust and diesel exhaust,
shows three traces that correspond to temperature, filter transmittance,
and FID detector response 3-75
3-14 Comparison of mass measurements with collocated RAMS, PC-BOSS,
FRM PM25 sampler, and a conventional TEOM monitor 3-79
3-15 Size distribution of particles divided by chemical classification into
organic, marine, and crustal 3-82
4-la Major constituents of particles measured at sites in the eastern United States .... 4-3
4-lb Major constituents of particles measured at sites in the central United States .... 4-4
4-lc Major constituents of particles measured at sites in the western United States .... 4-5
4-2 Annual average PM25 concentration (1994-96) 4-6
4-3 Concentrations of PM25 measured at the PBY site in southwestern
Philadelphia 4-8
4-4 Frequency distribution of PM25 concentrations measured at the PBY site
in southwestern Philadelphia 4-9
4-5 Concentrations of PM25 measured at the EPA site in Phoenix 4-10
4-6 Frequency distribution of PM2 5 concentrations measured at the EPA site
in Phoenix 4-11
4-7 Concentrations of PM25 measured at the Riverside-Rubidoux site 4-12
4-8 Frequency distribution of PM25 concentrations measured at the
Riverside-Rubidoux site 4-13
4-9 PM25 chemical components in downtown Los Angeles and Burbank (1986)
have similar characteristics 4-16
4-10 Concentrations of PM25 chemical components in Rubidoux and downtown
Los Angeles (1986) . 4-17
October 1999 I-xvi DRAFT-DO NOT QUOTE OR CITE
-------
List of Figures
(cont'd)
Number Page
4-11 Five-year average haze patterns centered on 1980 and 1990 4-21
4-12 Secular haze trends (1940 to 1990) for six eastern U.S. regions, summer
and winter 4-23
4-13 Trend data from Stockton-Hazelton from CARB: fine, coarse, and total
means; fine, coarse, and total 90th percentiles; every sixth-day fine and
coarse mass for 1991; and fine and coarse mass as a fraction of PM10 4-24
4-14 Nationwide emissions of PM25 from various source categories 4-34
4-15 Distribution of primary PM25 emissions across the United States 4-36
4-16 Distribution of primary PM10 emissions across the United States 4-37
4-17 Nationwide emissions of SO2, NOX, VOCs, and NH3 from various source
categories 4-38
4A-1 Size distribution of particles generated in a laboratory resuspension chamber . . 4A-3
4A-2 Size distribution of California source emissions, 1986 4A-4
4A-3 Chemical abundances for PM2 5 emissions from paved road dust in
Denver 4A-5
4-A4 Chemical abundances for PM2 5 emissions from wood burning in
Denver 4A-15
5-1 Sizes of various types of indoor particles 5-3
5-2 Categories of particle exposure outdoors and indoors 5-12
5-3 Two compartment model for PM deposition and resuspension by human
activity in a residential microenvironment 5-20
5-4 Histograms of the estimated fractions of outdoor PM25 and outdoor
PM10 found indoors during the PTEAM study in Riverside, CA 5-24
5-5 Individual indoor versus outdoor relationships of PM10 in Tokyo for the
seven subjects reported on by Tamura et al. (1996a) 5-46
October 1999 I-xvii DRAFT-DO NOT QUOTE OR CITE
-------
List of Figures
(cont'd)
Number Page
5-6 Comparison of light scattering coefficient by PM indoors and
outdoors at a home with an electrostatic precipitator in operation 5-49
5-7 Comparison of light scattering coefficients by PM indoor and outdoors
at a home with a standard in-line filter in the air recirculating system 5-50
5-8 Plot of relationship between average personal PM10 exposure versus
ambient PM10 monitoring data from Phillipsburg, NJ, and regression
line calculated by U.S. EPA 5-57
5-9 Plot of 48-h average personal PM10 exposure and ambient PM10 data from
Japan—linear regression 5-58
5-10 Completed mean personal PM2 5 exposure of children not exposed to tobacco
smoke at home in Wageningen, NL, versus the simultaneous ambient PM2 5
measured in their community 5-62
5-11A,B Completed mean personal PM10 exposures of two groups of adults
not exposed to tobacco smoke at home in Amsterdam, NL, versus
the simultaneous ambient PM10 measured in their community 5-65
5-12 Mean personal exposures to PMj 5 of five elderly residents of a retirement
home in Baltimore, MD, as a function of ambient PM25 concentration 5-70
5-13 Personal versus outdoor SO4= in State College, PA 5-73
5-14 Results from simultaneous measurements of the indoor PM10 concentration
and the immediate outdoor PM10 concentration of an uninhabited apartment
in a building fronting to a busy street in Oslo, Norway 5-75
5-15 PTEAM mean 24-h PM10 data compared for personal PEM and SAM 5-79
5-16 Correlations of PM10-CO for highest urban PM10 site per state;
correlations of PM10-CO for lowest rural PM10 site per state; and
correlations of PM25-CO 5-82
5-17 Correlations of PM10-NO2 for highest urban PM10 site per state;
correlations of PM10-NO2 for lowest rural PM10 site per state; and
correlations of PM25-NO2 5-83
October 1999 I-xviii DRAFT-DO NOT QUOTE OR CITE
-------
List of Figures
(cont'd)
Number Page
5-18 Correlations of PM10-SO2 for highest urban PM10 site per state;
correlations of PM10-SO2 for lowest rural PM10 site per state; and
correlations of PM25-SO2 5-84
5-19 Correlations of PM10-O3 1 -hr for highest urban PM10 site per state;
correlations of PM10-O3 1-hr for lowest rural PM10 site per state; and
correlations of PM25-O3 1-hr 5-85
5-20 Correlations of PM10-O3 8-hr for highest urban PM10 site per state;
correlations of PM10-O3 8-hr for lowest rural PM10 site per state; and
correlations of PM25-O3 8-hr 5-86
5-21 Air-exchange rate versus outdoor PM10 concentration at the home in the
PTEAM study 5-90
5-22 Estimated personal exposure to indoor and personally generated PM10
versus outdoor PM10 at the same home in the PTEAM study 5-92
October 1999 I-xix DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
CHAPTER 1. EXECUTIVE SUMMARY
Principal Authors
Dr. Robert Chapman—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Beverly Comfort—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. J.H.B. Garner—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. David Mage—National Exposure Research Laboratory (MD-56), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
Dr. Allan Marcus—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. William E. Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999 I-xxi DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
CHAPTER 2. INTRODUCTION
Principal Authors
Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
CHAPTER 3. PHYSICS, CHEMISTRY, AND MEASUREMENT
OF PARTICULA TE MA TTER
Principal Authors
Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Candis S. Claiborn—Washington State University, Laboratory for Atmospheric Research,
Department of Civil and Environmental Engineering,, P.O. Box 642910, Pullman, WA 99164
Contributors and Reviewers
Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD 21205
Ms. Lee Byrd—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way, Suite C
Petaluma, CA 94549
Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430
Irvine, C A 92612
Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602
October 1999 I-xxii DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095
Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lynn Hildemann—Stanford University, Civil and Environmental Engineering Department
Stanford, CA 94305
Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130
Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street
Waterbury,VT 05671
Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999 I-xxiii DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA 02461
Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street
Sacramento, CA 95814
CHAPTER 4. CONCENTRATIONS, SOURCES, AND
EMISSIONS OF ATMOSPHERIC PARTICLES
Principal Authors
Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD 21205
Ms. Lee Byrd—Office of Air Quality Planning and Standards (MD-14)
U. S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way, Suite C
Petaluma, CA 94549
Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430
Irvine, C A 92612
Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602
Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999 I-xxiv DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095
Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lynn Hildemann—Civil and Environmental Engineering Department
Stanford University, Stanford, CA 94305
Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130
Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street
Waterbury,VT 05671
Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA 02461
October 1999 I-xxv DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street, Sacramento, CA 95814
CHAPTER 5. HUMAN EXPOSURE TO AMBIENTF'ARTICULATE MATTER:
RELATIONS TO CONCENTRATIONS OF AMBIENT AND
NON-AMBIENTPM, AND OTHER AIR POLLUTANTS
Principal Authors
Dr. David Mage—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD 21205
Ms. Lee Byrd U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way, Suite C
Petaluma, CA 94549
Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430
Irvine, C A 92612
Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602
Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095
Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999 I-xxvi DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Vic Hasselblad—29 Autumn Woods Drive, Durham, NC 27713
Dr. Lynn Hildemann—Civil and Environmental Engineering Department
Stanford University, Stanford, CA 94305
Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130
Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Allan Marcus—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
October 1999 I-xxvii DRAFT-DO NOT QUOTE OR CITE
-------
Authors, Contributors, and Reviewers
(cont'd)
Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street
Waterbury,VT 05671
Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA 02461
Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street, Sacramento, CA 95814
Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Jim Xue—Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115
October 1999 I-xxviii DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
FOR PARTICULATE MATTER
Scientific Staff
Dr. Lester D. Grant—Director, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Randy Brady—Deputy, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lawrence J. Folinsbee—Health Coordinator, Chief, Environmental Media Assessment
Group, National Center for Environmental Assessment (MD-52), U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711
Dr. William E. Wilson—Air Quality Coordinator, Physical Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711
Dr. Dennis J. Kotchmar—Project Manager, Medical Officer, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711
Dr. Robert Chapman—Technical Consultant, Medical Officer, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711
Ms. Beverly Comfort—Health Scientist, National Center for Environmental Assessment (MD-
52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald—Technical Project Officer, Health Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711
Dr. David Mage—Technical Project Officer, Physical Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711
Dr. Allan Marcus—Technical Project Officer, Statistician, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711
October 1999 I-xxix DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
FOR PARTICULATE MATTER
(cont'd)
Dr. James McGrath—Technical Project Officer, Visiting Senior Health Scientist, National
Center for Environmental Assessment (MD-52), U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711
Dr. Joseph P. Pinto—Technical Project Officer, Physical Scientist, National Center for
Environmental Assessment, U.S. Environmental Protection Agency, Research Triangle Park, NC
27711
Technical Support Staff
Mr. Douglas B. Fennell—Technical Information Specialist, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711
Ms. Emily R. Lee—Management Analyst, National Center for Environmental Assessment (MD-
52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Diane H. Ray—Program Analyst, National Center for Environmental Assessment
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Eleanor Speh—Office Manager, Environmental Media Assessment Branch, National Center
for Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711
Ms. Donna Wicker—Administrative Officer, National Center for Environmental Assessment
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Richard Wilson—Clerk, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Document Production Staff
Mr. John R. Barton—Document Processing Coordinator
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Ms. Yvonne Harrison—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
October 1999 I-xxx DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
FOR PARTICULATE MATTER
(cont'd)
Ms. Bettye Kirkland—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Mr. David E. Leonhard—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Ms. Carolyn T. Perry—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Ms. Veda E. Williams—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Technical Reference Staff
Mr. R. David Belton—Reference Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Mr. John Bennett—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Mr. William Hardman—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Ms. Sandra L. Hughey—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
Mr. Jian Ping Yu—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
October 1999 I-xxxi DRAFT-DO NOT QUOTE OR CITE
-------
i 1. EXECUTIVE SUMMARY
2
3
4 1.1 INTRODUCTION
5 1.1.1 Purpose of the Document
6 The purpose of this document, Air Quality Criteria for Particulate Matter, is to present air
7 quality criteria for particulate matter (PM) in accordance with Clean Air Act (CAA) Sections 108
8 and 109, which govern establishment, review, and revision of U.S. National Ambient Air Quality
9 Standards (NAAQS) as follows:
10 • Section 108 directs the U.S. Environmental Protection Agency (EPA) Administrator to list
11 pollutants that may reasonably be anticipated to endanger public health or welfare and to issue
12 air quality criteria for them. The air quality criteria are to reflect the latest scientific
13 information useful in indicating the kind and extent of all exposure-related effects on public
14 health and welfare expected from the presence of the pollutant in ambient air.
15 • Section 109 directs the EPA Administrator to set and periodically revise, as appropriate,
16 (a) primary NAAQS to protect against adverse health effects of listed criteria pollutants among
17 sensitive population groups, with an adequate margin of safety, and (b) secondary NAAQS to
18 protect against welfare effects (e.g., impacts on vegetation, crops, ecosystems, visibility,
19 climate, man-made materials, etc.).
20 • Section 109 of the CAA also requires periodic review and, if appropriate, revision of existing
21 criteria and standards. Also, an independent committee of non-EPA experts, the Clean Air
22 Scientific Advisory Committee (CASAC), is to provide the EPA Administrator advice and
23 recommendations regarding the scientific soundness and appropriateness of criteria and
24 NAAQS.
25 To meet these CAA mandates, this document assesses the latest scientific information
26 useful in deriving criteria as scientific bases for decisions on possible revision of current
27 PM NAAQS. A separate EPA PM Staff Paper will draw upon assessments in this document,
28 together with other information, to develop exposure/risk analyses and to pose options for
29 consideration by the EPA Administrator with regard to possible retention or revision of the
30 PM NAAQS.
October 1999 1-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.1.2 Organization of the Document
2 The present document is organized into nine chapters, as follows:
3 • This Executive Summary (Chapter 1) summarizes key points from the ensuing chapters.
4 • Chapter 2 provides a general introduction, including a brief summary of the history of the PM
5 NAAQS and an overview of issues, methods and procedures used to prepare this document.
6 • Chapters 3 through 5 provide background information on air quality and exposure aspects, to
7 help to place the succeeding discussions of PM effects into perspective.
8 • Chapter 6 discusses community epidemiology information and Chapter 7 discusses dosimetry
9 and toxicology of PM.
10 • Chapter 8 provides an integrative synthesis of key points from those health chapters (6 & 7)
11 and other preceding air quality and exposure chapters.
12 • Chapter 9 deals with environmental effects of PM on vegetation and ecosystems; visibility;
13 climate; and manmade materials, as well as economic impacts of such effects.
14 It should be noted that new research results that have become available since early 1996
15 (when the last previous PM criteria assessment was completed) through mid-1999 are assessed in
16 this First External Review Draft of the revised PM Air Quality Criteria Document (PM AQCD).
17 Extensive further new research results, expected to be published during the next 6 to 9 mo
18 (including many from a vastly expanded EPA PM Research Program) will be assessed in the
19 Second External Review Draft of this document to be released in mid-2000 for public comment
20 and CASAC review. Thus, key findings and conclusions summarized below must be considered
21 to be only provisional and subject to change, as appropriate, due to consideration of new research
22 in the next draft.
23
24
25 1.2 AIR QUALITY AND EXPOSURE ASPECTS
26 The document's discussion of air quality and exposure aspects considers chemistry and
27 physics of atmospheric PM; analytical techniques for measuring PM mass, size, and chemical
28 composition; sources of ambient PM in the United States; temporal/spatial variability and trends
29 in ambient U.S. PM levels; and human exposure relationships. Key findings are summarized in
30 the next six sections. Overall, the atmospheric sciences and air quality information provides
October 1999 1 -2 DRAFT-DO NOT QUOTE OR CITE
-------
1 further evidence substantiating the 1996 PM AQCD conclusion that distinctions between fine
2 and coarse mode particles (in terms of sources of emission, formation mechanisms, atmospheric
3 transformation and transport distances, and air quality patterns) warrant fine and coarse PM being
4 viewed as separate subclasses of ambient PM.
5
6 1.2.1 Chemistry and Physics of Atmospheric Particles
7 • Airborne PM is not a single pollutant, but rather is a mixture of many subclasses of pollutants
8 with each subclass containing many different chemical species. Atmospheric PM occurs
9 naturally as fine-mode and coarse-mode particles that, in addition to falling into different size
10 ranges, differ in formation mechanisms, chemical composition, sources, and exposure
11 relationships.
12 • Fine-mode PM is derived from combustion material that has volatilized and then condensed to
13 form primary PM or from precursor gases reacting in the atmosphere to form secondary PM.
14 New fine-mode particles are formed by the nucleation of gas phase species, and grow by
15 coagulation (existing particles combining) or condensation (gases condensing on existing
16 particles). Fine particles are composed of freshly generated nuclei-mode particles, also called
17 ultrafine or nanoparticles, and an accumulation mode, so called because particles grow into and
18 remain in that mode.
19 • Coarse-mode PM, in contrast, is formed by crushing, grinding, and abrasion of surfaces, which
20 breaks large pieces of material into smaller pieces. These particles are then suspended by the
21 wind or by anthropogenic activity. Energy considerations limit the break-up of large particles
22 and small particle aggregates generally to a minimum size of about 1 //m in diameter. Mining
23 and agricultural activities are examples of anthropogenic sources of coarse-mode particles.
24 Fungal spores, pollen, and plant and insect fragments are examples of natural bioaerosols also
25 suspended as coarse-mode particles.
26 • Within atmospheric particle modes, the distribution of particle number, surface, volume, and
27 mass by diameter is frequently approximated by lognormal distributions. Aerodynamic
28 diameter, dae, which depends on particle density and is defined as the diameter of a particle with
29 the same settling velocity as a spherical particle with unit density (1 g/cm3), is often used to
30 describe particle size. Typical values of the mass median aerodynamic diameters (MMAD) are
October 1999 1 -3 DRAFT-DO NOT QUOTE OR CITE
-------
1 0.05 to 0.07 //m for the nuclei mode, 0.3 to 0.7 //m for the accumulation mode, and 6 to 20 //m
2 for the coarse mode. At high relative humidities or in air containing evaporating fog or cloud
3 droplets, the accumulation mode may be split into a droplet mode (MMAD = 0.5 to 0.8 //m)
4 and a condensation mode (MMAD = 0.2 to 0.3 //m).
5 • Research studies use (a) impactors to determine mass and composition as a function of size
6 over a wide range and (b) particle counting devices to determine number of particles as a
7 function of size. Such studies indicate an atmospheric bimodal distribution of fine and coarse
8 particle mass with a minimum in the distribution between 1 and 3 //m dae. Routine monitoring
9 studies generally measure thoracic particles or PM10 (upper size limited by a 50% cut at 10 //m
10 dae). Research studies and new monitoring studies measure fine particles or PM2 5 (upper size
11 limited by a 50% cut point at 2.5 //m dae) and the coarse fraction of PM10, i.e., the difference
12 between PM10 and PM25 (PM10_2 5). Cut points are not perfectly sharp for any of these PM
13 indicators; some particles larger than the 50% cutpoint are collected and some particles smaller
14 than the 50% cutpoint are not retained.
15 • The terms "fine" and "coarse" were originally intended to apply to the two major atmospheric
16 particle distributions which overlap in the size range between 1 and 3 //m diameter. Now, fine
17 has come to be often associated with the PM25 fraction and coarse is often used to refer to
18 PM10_25. However, PM25 may also contain, in addition to the fine-particle mode, some of the
19 lower-size tail of the coarse particle mode between about 1 and 2.5 //m dae. Conversely, under
20 high relative humidity conditions, the larger fine particles in the accumulation mode may also
21 extend into the 1 to 3 //m dae range.
22 • Three approaches are used to classify particles by size: (1) modes, based on formation
23 mechanisms and the modal structure observed in the atmosphere, e.g., nuclei and accumulation
24 modes (which comprise the fine-particle mode) and the coarse-particle mode; (2) cut point,
25 based on the 50% cut point of the specific sampling device; (3) dosimetry, based on the ability
26 of particles to enter certain regions of the respiratory tract; and (4) regulatory, based on
27 instrument configuration or 50% cut-points, e.g., high volume sampler, PM10, and PM2 5.
28
October 1999 1 -4 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.2.2 Sources of Airborne Particles in the United States
2 • The chemical complexity of airborne particles requires that the composition and sources of a
3 large number of primary and secondary components be considered. Major components of fine
4 particles are: sulfate, strong acid, ammonium, nitrate, organic compounds, trace elements
5 (including metals), elemental carbon, and water.
6 • Primary particles are emitted directly from sources. Secondary particles are formed from
7 atmospheric reactions of sulfur dioxide (SO2), nitrogen oxides (NOX), and certain organic
8 compounds. NO reacts with ozone (O3) to form NO2. SO2 and NO2 react with hydroxy radical
9 (OH) during the daytime to form sulfuric and nitric acid. During the nighttime NO2 reacts with
10 ozone and forms nitric acid through a sequence of reactions involving the nitrate radical (NO3).
11 These acids may react further with ammonia to form ammonium sulfates and nitrates. Some
12 types of higher molecular weight organic compounds react with OH radicals, and olefinic
13 compounds also react with ozone, to form oxygenated organic compounds which nucleate or
14 can condense onto existing particles. SO2 also dissolves in cloud and fog droplets where it may
15 react with dissolved O3, H2O2, or, if catalyzed by certain metals, with O2, yielding sulfuric acid
16 or sulfates, that lead to PM when the droplet evaporates.
17 • Receptor modeling has proven to be a useful method for identifying contributions of different
18 types of sources especially for the primary components of ambient PM. Apportionment of
19 secondary PM is more difficult because it requires consideration of atmospheric reaction
20 processes and rates. Results from western U.S. sites indicate that fugitive dust, motor vehicles,
21 and wood smoke are the major contributors to ambient PM samples there, while results from
22 eastern U.S. sites indicate that stationary combustion, motor vehicles and fugitive dust are
23 major contributors to ambient PM samples in the East. Sulfate and organic carbon are the
24 major secondary components in the East, while nitrates and organic carbon are the major
25 secondary components in the West.
26 • Fine and coarse particles have distinctly different sources, both natural and anthropogenic.
27 Therefore different control strategies are likely to be needed, depending on whether fine or
28 coarse particles (or both) are selected for control.
29
October 1999 1 -5 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.2.3 Atmospheric Transport and Fate of Airborne Particles
2 • Primary and secondary fine particles have long lifetimes in the atmosphere (days to weeks) and
3 travel long distances (hundreds to thousands of kilometers). They tend to be uniformly
4 distributed over urban areas and larger regions, especially in the eastern United States. As a
5 result, they are not easily traced back to their individual sources.
6 • Coarse particles normally have shorter lifetimes (minutes to hours) and only travel short
7 distances (<10's of km). Therefore, coarse particles tend to be unevenly distributed across
8 urban areas and tend to have more localized effects than fine particles. However, dust storms
9 occasionally cause long range transport of the smaller coarse-mode particles.
10
11 1.2.4 Airborne Particle Measurement Methods
12 • Measurements of ambient PM mass and chemical composition are needed to determine
13 attainment of standards; to guide attainment of a standard (including determination of source
14 categories and validation of air quality models); and to determine health, ecological, and
15 radiative effects. A comprehensive approach requires a combination of analytical techniques to
16 assess: (1) mass, (2) crustal and trace elements, (3) water-soluble ionic species including
17 strong acidity, (4) elemental carbon, and (5) organic compounds.
18 • There are no calibration standards for suspended particle mass; therefore, the accuracy of
19 particle mass measurements cannot be definitively determined. The precision of particle mass
20 measurements can be determined by comparing results from collocated samplers. When using
21 different measurement techniques, samplers of different design or manufacturer, and, in some
22 cases, when using identical systems of different age or cleanliness, substantial biases of 50% or
23 more have been observed. Mass concentration measurements with a precision close to 10%
24 have been obtained with collocated samplers of identical design and same time since cleaning.
25 • Available technology allows accurate (±10 to 15%) measurement of several of the major
26 components of coarse and fine particles (crustal and trace elements, sulfates, and strong
27 acidity). However, collection and measurement technologies for elemental carbon, organic
28 carbon, and nitrates are not as well established.
29 • Semivolatile organic compounds and semivolatile ammonium compounds (such as NH4NO3)
30 may be lost by volatilization during sampling. Such losses may be very important in
October 1999 1 -6 DRAFT-DO NOT QUOTE OR CITE
-------
1 woodsmoke impacted areas for organic compounds or in agricultural and other areas where low
2 sulfate and high ammonia lead to high NH4NO3 concentrations. Hence, while the Federal
3 Reference Methods for PM10 and PM25 give precise (± 10%) measurements of "equilibrated
4 mass", loss of semivolatile PM and possible retention of some particle-bound water contribute
5 to uncertainly in measurement of the mass of PM as it exists suspended in the atmosphere.
6 • Intercomparisons, using different techniques and samplers of different designs, coupled with
7 mass balance studies (relating the sum of components to the measured mass), provide a method
8 for gaining confidence in the reliability of PM measurements.
9
10 1.2.5 Ambient U.S. PM Concentrations: Regional Patterns and Trends
11 • Particle mass data have been collected at a number of rural, suburban, and urban sites across
12 the United States by various local, state, and national programs. The data have been stored in
13 the Aerometric Information Retrieval System (AIRS). Data have also been collected at remote
14 sites as part of the IMPROVE and NESCAUM networks. An extensive analysis of this data
15 was reported in the 1996 Air Quality Criteria Document for Particulate Matter (PM AQCD).
16 • Information on trends of PM25 (fine) and PM10_25 (coarse) were also examined for Philadelphia,
17 several AIRS sites, the Harvard Six-City sites, and California sites. However, such data is still
18 not sufficient, either in number of sites or number of years per site, to demonstrate differential
19 trends in fine or coarse PM.
20
21 1.2.6 Human PM Exposure
22 Chapter 5 examines: ambient particulate matter (PM) air quality; that portion of ambient
23 PM which penetrates into indoor microenvironments; and, to a lesser extent, the contributions of
24 sources of non-ambient PM to total PM exposure. This is to aid in interpretation of acute and
25 chronic epidemiology studies discussed in Chapter 6, in which ambient PM concentrations are
26 assumed to be an indicator, or a surrogate, for mean community exposure to PM of ambient
27 origin, or an individual's exposure to ambient PM. Thus, this chapter has three objectives:
28 (a) To provide a review of pertinent studies of personal exposures to total PM (ambient PM plus
29 non-ambient PM).
October 1999 1 -7 DRAFT-DO NOT QUOTE OR CITE
-------
1 (b) To evaluate linkages of human exposure to PM of ambient origin estimated from
2 concentrations of PM measured at a fixed-site monitor located at some central location in a
3 community under study.
4 (c) To quantify the contribution of PM of ambient origin to total personal PM exposure.
5 In meeting those objectives, Chapter 5 has reached the following provisional conclusions as
6 supported by the evidence cited:
7 • Human exposure to PM of ambient origin for individuals in a community is often highly
8 correlated (R2 > 0.5) in time with concentrations of PM of ambient origin of the same size as
9 measured in that community.
10 • Longitudinal correlation coefficients for ambient concentrations of fine PM (AD < 2.5 //m)
11 with personal exposures to ambient fine PM are greater than the corresponding correlations for
12 the coarse fraction of ambient PM10 (2.5 //m < AD < 10 //m), as shown by studies of ambient
13 sulfate concentrations and sulfate exposures.
14 • People in a community surrounding an ambient monitoring station, over time, are exposed to
15 relatively similar mixtures and concentrations of ambient PM2 5.
16 • People in a community are exposed to widely different mixtures and concentrations of
17 non-ambient PM due to the diversity of smoking habits, personal activities such as hobbies,
18 residential furnishings and appliances, and varying occupations.
19 • Exposures to PM of indoor origin appear to be uncorrelated with exposures to PM of ambient
20 origin.
21 • The correlation of a single individual's sequence of daily personal exposures to total PM and
22 ambient PM concentrations will be greater than the correlation that would occur had a different
23 person been monitored on each of the same days (e.g., one person monitored consecutively for
24 n days vs sequentially monitoring n different people, each for one day, over n days).
25 • Ambient PM in the U.S. has average annual correlations with the ambient gaseous pollutants
26 CO, ozone, NO2, and SO2 of order r = 0.25 with a standard deviation of order 0.25.
27 • Although exposures to PM from indoor sources and occupational activities may not be
28 correlated with ambient PM concentrations, these non-ambient PM species may possibly act as
29 effect modifiers by making subjects more or less susceptible to exposure to PM of ambient
30 origin.
October 1999 1 -8 DRAFT-DO NOT QUOTE OR CITE
-------
1 • There are only limited data available, from non-probability samples, to evaluate how well the
2 exposures to PM of ambient origin for susceptible subgroups correlate with the ambient PM
3 concentrations of similar AD size range as measured in their community.
4 The newly available PM exposure information, overall, appears to further substantiate that
5 ambient PM measurements (especially for fine PM) made at community monitoring sites likely
6 index well personal human exposures (both outdoors and indoors) to PM of ambient origin. The
7 community monitoring data, used as the main PM exposure indices in PM community
8 epidemiology studies, is not highly correlated with human exposure to PM from indoor sources,
9 making it unlikely that exposure to PM from indoor sources confound reported ambient
10 PM-health effects associations.
11
12
13 1.3 DOSIMETRY
14 Key findings derived from the assessment of dosimetry information include:
15 • Particles may be deposited in (a) the extrathoracic airways (i.e., mouth, nose, and larynx); (b) in
16 airways of the tracheobronchial region; and (c) in the alveolar region where gas exchange
17 occurs. There are differences in deposition mechanisms and dose distribution in each of these
18 areas that are dependent on particle size and airway geometry. Whereas, impaction is an
19 important deposition mechanism in large extra- and intrathoracic airways at higher flows,
20 sedimentation and diffusion are more important at low flow rates in smaller airways.
21 • Respiratory tract deposition patterns are dependent on particle size and distribution
22 (as indicated by the mass median aerodynamic diameter and the geometric standard deviation)
23 within the inspired air. Biologic effects may be a function not only of particle mass deposition
24 but also of particle number or the total surface area of the particles.
25 • Various host factors have been shown to influence particle deposition patterns, including
26 airway dimensions (size and shape), breathing pattern (flow and volume), and the presence of
27 obstructive or inflammatory airway disease. Particle deposition in the extrathoracic region is
28 highest during nasal breathing, and is greatest in small children and least in adults. Increased
29 total ventilation and increased oral breathing leads to greater deposition of coarse particles in
October 1999 1 -9 DRAFT-DO NOT QUOTE OR CITE
-------
1 the thoracic airways. Obstructive airway disease, such as asthma, emphysema, and chronic
2 bronchitis, results in increased deposition of particles in the lower respiratory tract.
3 • Acute effects of PM are probably best related to deposited dose, whereas chronic effects may
4 be related to cumulative or retained dose. Retention of particles is a function of deposition site,
5 clearance of particles by macrophages or the mucociliary system, and particle characteristics,
6 especially solubility. Chronic effects may also arise from recurring cycles of pulmonary injury
7 and repair.
8 • There are substantial differences among laboratory animal species with regard to the
9 inhalability of different sized particles as well as quantitative and qualitative differences in
10 airway geometry and branching patterns.
11 • Extrathoracic deposition of ultrafine particles (<0.100 //m) is very high, despite their small
12 size. Estimates of deposition range from 50% for oral breathing to >90% for nasal breathing.
13 Enhanced deposition of both ultrafine and coarse particles occurs at branching points within the
14 intrathoracic airways.
15 • Clearance is either absorptive (dissolution) or nonabsorptive (transport of intact particles).
16 Deposited particles may be dissolved in body fluids, taken up by phagocytic cells, or
17 transported by the mucociliary system. Retained particles tend to be small (<2.5 //m) and
18 poorly soluble, e.g., silica, metals).
19 • Tracheobroncial clearance has both a fast and a slow component. Translocation of poorly
20 soluble PM to the lymph lodes takes a few days and is more rapid for smaller (<2 //m)
21 particles; elimination rates of these retained particles are on the order of years. People with
22 COPD have increased particle retention and use cough to augment mucociliary clearance.
23 • In order to extrapolate experimental results between different species, a number of factors must
24 be considered such as: airway size; airway branching asymmetry; inhalability; deposition;
25 deposition/surface area, hot-spots; number deposition versus mass deposition; dose-response;
26 and clearance retention (where and how long).
27 • Mathematical models are available to predict deposition, clearance, and retention of particles.
28 Two models, the ICRP and NCRP, were extensively discussed in the 1996 PM AQCD.
29
30
October 1999 1-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.4 PARTICULATE MATTER HEALTH EFFECTS
2 1.4.1 Epidemiology Findings
3 The epidemiology evidence about the health effects of ambient PM has expanded greatly
4 since the 1996 PM Air Quality Criteria Document (PM AQCD). The most important
5 enhancements in information include:
6 • New studies of health endpoints using ambient PM10 and closely related mass concentration
7 indices such as PM13 and PM7;
8 • New studies on a variety of endpoints have evaluated effects of the ambient coarse PM fraction
9 (PM10_2 5), the ambient fine particle fraction (PM2 5), and even ambient ultrafine particle mass
10 concentrations (PMj and smaller);
11 • New studies in which the relationship of some health endpoints to ambient particle number
12 concentrations were evaluated;
13 • Additional studies which evaluated the sensitivity of estimated PM effects to the inclusion of
14 gaseous co-pollutants in the model;
15 • Preliminary attempts to evaluate the effects of air pollutant combinations or mixtures including
16 PM components, based on empirical combinations (factor analysis) or source profiles;
17 • New studies of infants and children as a potentially susceptible population;
18 • Further studies of cardiovascular endpoints associated with PM exposures;
19 • New studies on asthma and other respiratory conditions exacerbated by PM exposure.
20 This additional information does not yet allow for full resolution of all outstanding key
21 issues in PM air pollution epidemiology. Table 1-1 presents provisional conclusions about
22 alternative hypothesis that may affect the interpretation and synthesis of epidemiology study
23 results. The authors conclude that multiplicity of findings about PM health effects suggest that
24 exposure to ambient PM at current concentrations may cause serious adverse health effects, but
25 that the quantitative magnitude of the effects depends on several environmental and biological
26 factors whose role is not yet known. That is, current levels of ambient PM may be harmful to
27 human health, but not necessarily equally harmful everywhere or at all times.
28 Some more specific provisional key findings emerging thus far from assessment of new
29 epidemiology study results in this draft document are:
30
October 1999 1-11 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1. PROVISIONAL CONCLUSIONS ABOUT
ALTERNATIVE HYPOTHESES THAT MAY AFFECT THE
SYNTHESIS OF EPIDEMIOLOGY STUDIES
Alternative hypotheses
Adverse health effects depend only
on ambient PM size range at
current PM concentrations
Adverse health effects depend on
ambient PM with specific physical
properties or current ambient PM
composition
Adverse health effects depend only
on current ambient PM,
independent of co-factors
Not likely. Adverse health effects
from coarse particles may occur at
some sites, not others.
Possible. Adverse health effects
from ambient PM of a given size
may be different in sites where PM
has different physical properties or
composition with same PM size
range.
Adverse health effects depend
on current ambient PM and on
environmental co-factors
Possible. Adverse health effects
from ambient PM are different in
sites where ambient PM has
different co-factors with same PM
range.
Probable. Adverse health effects
from ambient PM are different in
sites where PM has different
physical properties, composition, or
co-factors, even in the same
ambient PM size range
1 • New studies suggest that infants and children may represent an additional subgroup at special
2 risk for ambient PM exposure effects. The new results most clearly indicate that children
3 appear to be susceptible to respiratory effects associated with ambient PM exposures, including
4 exacerbation of asthma and respiratory symptoms in school-age children. A few studies also
5 report ambient PM to be associated with intrauterine growth reduction and low birth weight
6 (known infant health risk factors) and excess infant mortality. However, some studies found
7 that PM is not as good a predictor of these endpoints as other pollutants (e.g., CO), and no
8 toxicologic evidence has yet been advanced to support biological plausibility of such effects
9 due to ambient PM or to identify pathophysiologic mechanisms involved.
10 • Cardiovascular causes of death and hospitalization in older adults may also be a significant
11 component of PM-attributable mortality, as well as respiratory causes
12 • PM health effects have been reported to be associated with several different ambient PM size
13 fractions (ultrafine, fine, coarse), but some health effects may be absent from some ambient PM
14 mass fractions under some circumstances.
15 • PM health effects may occur at different time scales for exposure to PM10 or PM25, from (a)
16 short -term responses to daily exposure through (b) larger excess mortality associated with
October 1999
1-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 medium term exposures (15 to 120 day averages) to (c) excess morbidity or mortality
2 associated with long-term (multi-year) exposures.
3 • Adverse health effects attributable to PM2 5 in short-term exposure studies are at times seriously
4 confounded by exposure to gaseous co-pollutants making it difficult to estimate, quantitatively
5 that portion of the risk attributable to: PM acting alone; PM acting in combination with
6 gaseous co-pollutants; the gaseous co-pollutants per se; or the overall ambient pollutant mix.
7
8 1.4.2 Toxicologic Responses to PM in Animals and Humans
9 Data on the toxicology of PM are derived from controlled inhalation exposure studies of
10 humans and animals, intratracheal instillation studies in humans and animals, occupational
11 studies, and ex vivo studies of human and animal cells grown in culture. The human or animal
12 populations (cells) studied vary by age, health status, or other host factors. Exposures vary by
13 duration, mass or number concentration, chemical composition and size of the PM in addition to
14 other exposure variables (e.g., temperature, humidity, activity levels, etc.). Responses to PM in
15 the respiratory tract are dependent on the physiological status of the host as well as on
16 translocation of PM or PM constituents to other sites. Ex vivo studies provide important
17 additional information regarding the mechanism of action of PM or PM constituents on cells or
18 cellular components.
19 Responses to acidic aerosols (sulfuric acid, sulfates, nitrates) have been comprehensively
20 reviewed in previous documents. Much of the newly available research focuses on
21 combustion-related PM such as concentrated ambient particles (CAPs) from urban air or various
22 forms of fly ash PM.
23 • Acute exposures to metal particles can cause inflammatory responses in the respiratory tract of
24 humans and animals. The effective exposure levels are typically much higher than ambient air
25 metal concentrations in the U.S. atmosphere. Endotoxin, a lipopolysacharide associated with
26 bacteria, causes inflammation in human occupational exposures at concentrations that are also
27 much higher than in the ambient air.
28 • Combustion-related particles (fly ash and urban air particles) cause a spectrum of responses in
29 the airways of animals. These include inflammation, cellular injury, and increased
30 permeability. Metal components (V, Cu, Zn, Fe, Ni) of combustion particles have been
October 1999 1-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 implicated in the responses, possibly related to oxidant production and release of intercellular
2 signaling molecules (cytokines).
3 • Cells primed by inflammatory mediators show increased cytokine responses to PM.
4 Combustion-related particles may cause increased oxidant production in in vitro systems,
5 presumably related to metal components of particles. Combustion-related particles cause
6 damage to cells in vitro. Responses include impaired macrophage phagocytosis and altered
7 permeability.
8
9 Susceptibility
10 • Chemically or pharmacologically treated rat "models of cardiorespiratory disease" as well as
11 older rats demonstrate increased pulmonary responses to combustion-related particles.
12 • Diesel particulate matter and oil fly ash (ROFA) may augment responses to antigens in allergic
13 animals or humans. These studies provide a plausible mechanism for an association between
14 combustion PM exposure and exacerbation of asthma.
15 • Inhaled or instilled particles can have systemic effects, especially on the cardiovascular system
16 which in certain circumstances can be lethal.
17
18 Mixtures
19 • Mixtures of ozone and PM (urban PM, sulfate aerosols, ultrafine carbon) may cause enhanced
20 effects on lung cells, increased inflammation, and decrements in human lung function.
21
22 Mechanisms
23 • A number of studies indicate that reactive oxidant species (ROS) play a role in PM-induced
24 responses. Catalysis of ROS is likely related to soluble metals in PM.
25 • ROFA and urban PM can induce apoptosis (programmed cell death) of human alveolar
26 macrophages.
27 • Studies on ultrafine compared to fine particles indicate a greater response to ultrafines in terms
28 of airway inflammation, an effect that appears to be related to their greater surface area.
29
October 1999 1-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.4.3 Population Groups at Risk
2 Susceptibility can be affected by factors which influence dosimetry or the response of
3 tissues to particle burdens. Host factors that may increase the susceptibility to PM include both
4 changes in physiologic factors affecting respiratory tract deposition and pathophysiologic factors
5 affecting response.
6 • Susceptible groups most clearly at special risk for PM effects include the elderly and those with
7 cardiopulmonary disease, based on available epidemiology findings.
8 • Epidemiology findings indicate that risk of mortality and morbidity due to lower respiratory
9 disease (e.g. pneumonia) is increased by ambient PM exposure. This may be due to
10 exacerbation, by PM, of already existing respiratory disease. PM may also increase
11 susceptibility to infectious disease by decreasing clearance, impairing macrophage function, or
12 through other specific and nonspecific effects on the immune system. The epidemiologic
13 findings also indicate that individuals with preexisting infectious respiratory disease (e.g.,
14 pneumonia) are at increased risk for PM effects.
15 • Epidemiologic findings indicate that ambient PM exposures are also associated with increased
16 risk for mortality and hospitalization due to cardiovascular causes. Cardiac arrhythmia has
17 been hypothesized as being involved in mortality due to acute PM exposure.
18 • Studies of infants and children indicate that they are a potentially susceptible population.
19 Panel studies on asthma and other respiratory conditions show exacerbation by PM exposure.
20 Children are susceptible to respiratory effects associated with PM exposure from pre-natal and
21 post-natal effects through exacerbation of asthma and respiratory symptoms in school age
22 children.
23
24
25 1.5 ENVIRONMENTAL EFFECTS
26 1.5.1 Vegetation and Ecosystem
27 • Human existence on this planet depends on the life-support service that ecosystems provide.
28 Ecosystem structure and function play an essential role in providing two types of benefits to
29 society. From the structural aspects (biodiversity, abundance, mass and arrangement of
30 species), an ecosystem provides: (1) products with market value such as fish, minerals, forage,
October 1999 1-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 forest products, biomass fuels, natural fiber, and many pharmaceuticals, and the genetic
2 resources of valuable species (e.g., plants for crops and timber and animals for domestication);
3 and (2) the use and appreciation of ecosystem for recreation, aesthetic enjoyment, and study.
4 • Ecosystems maintain clean water, pure air, a green earth, and a balance of creatures, the
5 functions that enable humans to survive. The benefits they impart include absorption and
6 breakdown of pollutants, cycling of nutrients, binding of soil, degradation of organic waste,
7 maintenance of a balance of gases in the air, regulation of radiation balance, climate, and the
8 fixation of solar energy.
9 • Concern has risen in recent years concerning the integrity of ecosystems because there are few
10 ecosystems on planet earth today that are not influenced by humans. For this reason,
11 understanding the effects of PM deposition and its impact on vegetation and ecosystems is of
12 prime importance.
13 • The criteria pollutant presently defined as PM10 has no particular relevance to particulate effects
14 on vegetation for which chemical composition is more relevant than mass. The PM whose
15 effects on vegetation are considered in this chapter is not a single pollutant, but a
16 heterogeneous mixture of particles of differing in size, origin, and chemical constituents.
17 Exposure to a given mass concentration of PM10 may, depending on the particular mix of
18 deposited particles, lead to widely differing phytotoxic responses. This variable has not been
19 adequately characterized.
20 • Atmospheric deposition of particles to ecosystems takes place via both wet and dry processes
21 through the three major routes indicated below:
22 (1) Precipitation scavenging in which particles are deposited in rain and snow
23 (2) Fog, cloud-water, and mist interception
24 (3) Dry deposition, a much slower, yet more continuous removal to surfaces.
25 • Deposition of heavy metal particles to ecosystems occurs by wet and dry processes. Dry
26 deposition is considered more effective for coarse particles of natural origin and elements such
27 as iron and manganese, whereas wet deposition generally is more effective for fine particles of
28 atmospheric origin and elements such as cadmium, chromium, lead, nickel, and vanadium.
29 • The actual importance of wet versus dry deposition, however, is highly variable, depending on
30 the type of ecosystem, location and elevation. The range of particle sizes, the variety of
31 chemical constituents in airborne PM, and the diversity of canopy surfaces, have slowed
October 1999 1-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 progress in both prediction and measurement of dry particulate deposition. Wet deposition
2 generally is confounded by fewer factors and has been easier to quantify.
3 • Emphasis in this chapter has been placed on the effects of PM on natural plants and terrestrial
4 ecosystems. Neither nitrate or sulfate deposition on crops is discussed as they are frequently
5 added in fertilizers. Except for nitrogen and sulfur-containing compounds and their effects in
6 acidic precipitation, information concerning the effects of PM on crops is not readily available.
7 • Particulate matter when transferred from the atmosphere to plant surfaces may cause direct
8 effects when they (1) reside on the leaf, twig or bark surface for an extended period; (2) are
9 taken up through the leaf surface; or indirect effects when (3) removed from the plant via
10 suspension to the atmosphere, washing by rainfall, or by litter-fall with subsequent transfer to
11 the soil.
12 • Deposition of PM on above-ground plant parts can have either a physical and or chemical
13 impact, or both. The effects of "inert" PM are mainly physical, while the effects of toxic
14 particles are both chemical and physical. The effects of dust deposited on plant surfaces or on
15 soil are more likely to be associated with their chemistry than with the mass deposited particles
16 and are usually of more importance than any physical effects. Studies of the direct effects of
17 chemical additions to foliage in particulate deposition have found little or no effects of PM on
18 foliar processes unless exposure levels were significantly higher than would typically be
19 experienced in the ambient environment.
20 • The majority of the easily identifiable direct and indirect effects, other than climate, occur in
21 severely polluted areas around heavily industrialized point sources such as limestone quarries,
22 cement kilns, iron, lead, and various smelting factories. Indirect impacts are usually the most
23 significant because they can alter nutrient cycling in the soil and inhibit plant uptake of
24 nutrients.
25 • Most PM deposited on vegetation eventually enters the soil environment which is one of the
26 most dynamic sites of biological interaction in nature. The impact of particulate deposition on
27 plants results from changes in the soil environment and plant nutrient uptake. These changes
28 determine plant and ecosystem response.
29 • Bacteria and fungi in the soil have an important role in plant nutrition. Bacteria are essential
30 components of the nitrogen and sulfur cycles that make these elements available for plant
31 uptake. Fungi form mycorrhizae, a mutualistic, symbiotic relationship, that is integral in the
October 1999 1-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 uptake of mineral nutrients. The impact of nitrates, sulfates and metals in PM is determined by
2 their affect on the growth and functions of the bacteria and fungi involved in making nutrients
3 available for plant uptake.
4 • The major impact of PM on ecosystems is indirect and occurs in the soil through the deposition
5 of nitrates and sulfates and the acidifying effects of the H+ ion associated with these
6 compounds in wet and dry deposition.
7 • Intensive research over nearly a decade indicates that although the soils of most North
8 American forests are nitrogen limited, there are forests that exhibit symptoms of excess
9 nitrogen. Nitrogen saturation results in a progressive syndrome of concurrent responses to
10 long-term, chronic nitrogen deposition. As nitrogen reaches saturation in temperate-zone
11 forests, there is decrease in nitrogen mineralization and an increase in the trends of foliar
12 Mg:N and Ca:Al ratios. Preliminary evidence suggests some forests may decline in
13 productivity and experience greater mortality as a result of chronic nitrogen deposition.
14 • Increases in soil nitrogen plays a selective role. Plant succession patterns and biodiversity in
15 some ecosystems are significantly affected by chronic nitrogen additions. Long-term nitrogen
16 fertilization studies in both New England and Europe suggest that some forests receiving
17 chronic inputs of nitrogen may decline in productivity and experience greater mortality.
18 Studies also suggest that declining coniferous forest stands with slow nitrogen cycling may be
19 replaced by deciduous fast-growing forests that cycle nitrogen rapidly. Excess nitrogen inputs
20 to unmanaged heathlands in the Netherlands has resulted in nitrophilous grass species replacing
21 slower growing heath species. Over the past several decades the composition of plants in the
22 forest herb layers had been shifting toward species commonly found on nitrogen-rich areas.
23 It also was observed that the fruiting bodies of mycorrhizal fungi had decreased in number.
24 • The effects of excessive deposition of nitrogen, particularly NH3 and NH4+ Deposition have
25 lead to changes in Dutch heathlands via: (1) acidification of the soil and the loss of cations of
26 K+, Ca 2+ and Mg 2+; and (2) nitrogen enrichment which results in increased plant growth rates
27 and altered competitive relationships. Alteration of any of a number of parameters
28 (e-g-> increased nitrogen) can alter ecosystem structure and function.
29 • There is a major concern that soil acidification will result in nutrient deficiency. Growth of tree
30 species can be affected when high aluminum to nutrient ratios limit uptake of calcium and
31 magnesium. Calcium is essential in the formation of wood and the maintenance of cells, the
October 1999 1-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 primary plant tissues necessary for tree growth. Calcium must be dissolved in the soil water to
2 be taken up by plants. A major concern is that soil acidity will lead to nutrient deficiency.
3 Acid deposition can increase the aluminum concentrations in soil water by lowering the pH in
4 aluminum-rich soils through dissolution and ion-exchange processes. Aluminum in soil can be
5 taken up by roots more readily than calcium because of its greater affinity for negatively
6 charged surfaces. Tree species can be adversely affected if high Ca/Al ratios impair Ca and
7 Mg uptake.
8
9 1.5.2 Particulate Matter-Related Effects on Materials
10 The effects of particulate matter and SO2 on materials are related to both aesthetic appeal
11 and physical damage. Studies have demonstrated particles, primarily consisting of carbonaceous
12 compounds, cause soiling of commonly used building materials and culturally important items
13 such as statutes and works of art. Physical damage from the dry deposition of SO2, particles and
14 the absorption or adsorption of corrosive agents on deposited particles can also result in the
15 acceleration of the weathering of manmade building and naturally occurring cultural materials.
16 • The natural process of metal corrosion from exposure to environmental elements (wind,
17 moisture, sun, temperature fluctuations, etc.) is enhanced by exposure to anthropogenic
18 pollutants, in particular SO2.
19 • Dry deposition of SO2 enhances the effects of environmental elements on calcereous stones
20 (limestone, marble, and carbonated cemented) by converting the calcium carbonate (calcite) in
21 the stone to calcium sulphate dihydrate (gypsum). The rate of deterioration is determined by
22 the SO2 concentration, the stone's permeability and moisture content, and the deposition rate.
23 • Sulfur dioxide limits the life expectancy of paints by causing discoloration, loss of gloss, and
24 loss of thickness of the paint film layer.
25 • A significant detrimental effect of particulate pollution is the soiling of painted surfaces and
26 other building materials. Soiling is a degradation process requiring remediation by cleaning or
27 washing, and depending on the soiled surface, repainting. Soiling decreases the reflectance of a
28 material and reduces the transmission of light through transparent materials. Soiling may
29 reduce the life usefulness of the material soiled.
30
October 1999 1-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.5.3 Particulate Matter-Related Effects on Visibility
2 Visibility is the degree to which the atmosphere is transparent to visible light and the clarity
3 (transparency) and the color fidelity of the atmosphere. The farthest distance at which a large
4 black object can be distinguished against the horizontal sky is the visual range. For regulatory
5 purposes, visibility impairment, any humanly perceptible change in visibility (light extinction,
6 visual range, contrast, coloration), is classified into two principal forms: "reasonably
7 attributable" impairment, attributable to a single source/small group of sources, and regional
8 haze, any perceivable change in visibility caused by a combination of many sources over a wide
9 geographical area.
10 • Visibility is measured by human observation, light scattering by particles, the light
11 extinction-coefficient (the sum of the light scattering coefficient and light absorption
12 coefficient) and parameters related to the light-extinction coefficient (visual range and deciview
13 scale), the light scattering coefficient, and fine particulate matter concentrations.
14 • Light scattering by gases is the major component of light extinction. Light absorption by gases
15 is almost entirely due to NO2, and is typically significant only near NO2 sources. Light
16 absorption by particles is primarily caused by elemental carbon.
17 • Visibility impairment or haziness is often associated with fine mass concentrations. Visibility
18 impairment or haziness is greatest in the eastern United States and southern California.
19 Haziness in the southeastern United States, caused by increased atmospheric sulfate, has
20 increased by approximately 80% since the 1950s and is greatest in the summer months,
21 followed by the spring and fall, and winter.
22 • Visibility impairment in southern California is primarily caused by light extinction by nitrates.
23 Nitrates contribute about 40% to the total light extinction in southern California. Nitrates
24 account for 10 to 20% of the total extinction in other areas of the United States.
25 • Some of the visibility impairment in northern California and Nevada, including Oregon,
26 southern Idaho and western Wyoming, results from coarse mass and soil, primarily considered
27 natural extinction. In some areas of the United States, extinction from coarse mass is almost
28 negligible because the overall extinction is so high.
October 1999 1 -20 DRAFT-DO NOT QUOTE OR CITE
-------
1 • High dust concentrations transported from southern California and the subtropics have
2 contributed to regional haze in the Grand Canyon and other class I areas in the southwestern
3 United States.
4 • Organics are the second largest contributors to light extinction in most areas in the United
5 States. Organic carbon is the greatest cause of light extinction in the Pacific Northwest,
6 Oregon, Idaho, and Montana, accounting for 40 to 45% of the total extinction. Organic carbon
7 contributes between 15 to 20% to the total extinction in most of the western United States and
8 20 to 30% in the remaining areas of the United States.
9 • Light absorption by carbon is relatively insignificant but is highest in the Pacific Northwest
10 (up to 15%) and in the eastern United States (3%).
11
12 1.5.4 Environmental and Economic Impacts of PM
13 • The four important categories of environmental costs and benefits of PM are: agriculture and
14 forestry, cleaning and materials damage, visibility, and ecosystem functions. EPA has
15 developed and applied cost methodologies to these cost categories in great detail, and its
16 procedures have passed a number of scientific reviews.
17 • Any given level of particulate matter will be associated with resulting environmental effects
18 that potentially have economic significance. Examples include the level of crop damage or
19 visibility impairment that result from specified levels of PM. Defining the welfare effects of
20 PM changes requires that baseline levels of effects be defined.
21 • Estimating benefits for visibility and for ecosystem services is a more difficult and less precise
22 exercise because the effects are not valued in markets.
23 • Once endpoints reflecting physical and biological outcomes have been defined, several
24 economic methods may be used to estimate economic damages. Some of the results of existing
25 research were summarized for the major categories of endpoints.
26 • The measured economic costs of PM are particularly significant for reduced visibility, both in
27 residential areas and in recreational areas with special value (e.g., the National Parks).
28 • It is possible that the costs imposed on ecosystems are significant as well. Making progress on
29 measuring these ecosystem costs depends on improvements in linking environmental endpoints
October 1999 1 -21 DRAFT-DO NOT QUOTE OR CITE
-------
1 to PM levels, and then on using these endpoints as a basis for improved techniques to elicit
2 willingness to pay for changes in ecosystem quality.
October 1999 1 -22 DRAFT-DO NOT QUOTE OR CITE
-------
i 2. INTRODUCTION
2
3
4 This document is an update of "Air Quality Criteria for Particulate Matter" published by the
5 United States Environmental Protection Agency (EPA) in 1996, and it will serve as the basis for
6 Congressionally-mandated periodic review of the National Ambient Air Quality Standards
7 (NAAQS) for particulate matter (PM). The present document critically assesses the latest
8 scientific information relative to determining the health and welfare effects associated with
9 exposure to various concentrations of PM in ambient air. The document is not intended as a
10 complete and detailed literature review, but rather focuses on thorough evaluation of that
11 information most relevant to PM NAAQS criteria development, based on pertinent literature
12 available through mid-1999. This introductory chapter presents a brief summary of the history of
13 the PM NAAQS, provides an overview of issues addressed and procedures utilized in the
14 preparation of the present document, and provides orientation to the general organizational
15 structure of this document.
16
17
18 2.1 LEGISLATIVE REQUIREMENTS
19 Sections 108 and 109 of the U.S. Clean Air Act (CAA) (U.S. Code, 1991) govern the
20 establishment, review, and revision of National Ambient Air Quality Standards (NAAQS).
21 Section 108 directs the Administrator of the U.S. Environmental Protection Agency (EPA) to list
22 pollutants that may reasonably be anticipated to endanger public health or welfare and to issue air
23 quality criteria for them. The air quality criteria are to reflect the latest scientific information
24 useful in indicating the kind and extent of all exposure-related effects on public health and
25 welfare that may be expected from the presence of the pollutant in ambient air.
26 Section 109(a, b) directs the Administrator of EPA to propose and promulgate "primary"
27 and "secondary" NAAQS for pollutants identified under Section 108. Section 109(b)(l) defines
28 a primary standard as a level of air quality, the attainment and maintenance of which, in the
29 judgment of the Administrator, based on the criteria and allowing for an adequate margin of
30 safety, is requisite to protect the public health. Under Section 109(b) of the CAA, the
October 1999 2-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 Administrator must consider available information to set secondary NAAQS that are based on
2 the criteria and are requisite to protect the public welfare from any known or anticipated adverse
3 effects associated with the presence of such pollutants. Welfare effects include effects on
4 vegetation, crops, soils, water, animals, manufactured materials, weather, visibility, and climate,
5 as well as damage to and deterioration of property, hazards to transportation, and effects on
6 economic value and personal comfort and well-being. Section 109(d) of the CAA requires
7 periodic review and, if appropriate, revision of existing criteria and standards. Also, an
8 independent committee of non-EPA experts, the Clean Air Scientific Advisory Committee
9 (CASAC), is to provide the EPA Administrator advice and recommendations regarding the
10 scientific soundness and appropriateness of criteria and NAAQS for PM and other "criteria air
11 pollutants."
12
13
14 2.2 HISTORY OF PREVIOUS PM CRITERIA/STANDARDS REVIEWS
15 On April 30, 1971 (Federal Register, 1971), EPA promulgated the original primary and
16 secondary NAAQS for particulate matter (PM) under Section 109 of the CAA. The reference
17 method for measuring attainment of these standards was the "high-volume" sampler (Code of
18 Federal Regulations, 1977), which collects PM up to a nominal size of 25 to 45 micrometers
19 (/^m), i.e., so-called "total suspended particulate" or "TSP". Thus, TSP was the original indicator
20 for U.S. PM NAAQS. The primary standards for PM (measured as TSP) were 260 //g/m3 (24-h
21 average) not to be exceeded more than once per year, and 75 //g/m3 (annual geometric mean).
22 The secondary standard (measured as TSP) was 150 //g/m3 (24-h average) not to be exceeded
23 more than once per year.
24 The next review of PM air quality criteria and standards was completed in July 1987 with
25 notice of a final decision to revise the existing standards published in the Federal Register
26 (Federal Register, 1987). In that decision, EPA changed the indicator for PM from TSP to PM10.
27 PM10 refers to particles with an upper 50% cut point of 10 //m aerodynamic diameter. Identical
28 primary and secondary PM10 standards were set for two averaging times: 150 //g/m3 (24-h ave.)
29 with no more than one expected exceedance per year; and 50 //g/m3 (expected annual arithmetic
30 mean) averaged over 3 years.
31
October 1999 2-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 2.2.1 The 1997 PM NAAQS Revision
2 The EPA initiated the last previous review of the air quality criteria and standards for PM
3 in April 1994 by announcing its intention to develop revised Air Quality Criteria for Particulate
4 Matter (henceforth, the "PM Air Quality Criteria Document" or PM AQCD). Thereafter, the
5 EPA presented its plans for review of the criteria and standards for PM under a highly
6 accelerated, court-ordered schedule at a public meeting of the CAS AC in December 1994. A
7 court order entered in American Lung Association v. Browner, CIV-93-643-TUC-ACM (U.S.
8 District Court of Arizona, 1994), as subsequently modified, required publication of EPA's final
9 decision on the review of the PM NAAQS by July 19, 1997.
10 Several workshops were held by EPA's National Center for Environmental Assessment's
11 RTP Division (NCEA-RTP) in November 1994 and January 1995 to discuss important new
12 health effects information useful in preparing initial PM AQCD draft materials. External review
13 drafts of the PM AQCD were then made available for public comment and were reviewed by
14 CAS AC at public meetings held in August 1995, December 1995, and February 1996. The
15 CASAC came to closure in its review of the PM AQCD, advising the EPA Administrator in a
16 March 15, 1996 closure letter (Wolff, 1996) that "although our understanding of the health
17 effects of PM is far from complete, a revised Criteria Document which incorporates the Panel's
18 latest comments will provide an adequate review of the available scientific data and relevant
19 studies of PM." Revisions in response to public and CASAC comments were incorporated as
20 appropriate in the final 1996 PM AQCD (U.S. Environmental Protection Agency, 1996a). A PM
21 Staff Paper (SP), prepared by U.S. EPA's Office of Air Quality Planning and Standards
22 (OAQPS) and drawing upon the 1996 PM AQCD and other exposure and risk assessments to
23 pose options for PM NAAQS decisions, also underwent similar CASAC review and public
24 comment, with consequent revision to its July 1996 final form (U.S. Environmental Protection
25 Agency, 1996b).
26 The SP analyses served as key inputs to subsequently published proposal for revision of the
27 PM NAAQS. Taking into account information and assessments presented in the PM AQCD and
28 the Staff Paper, advice and recommendations of CASAC, and public comments received on the
29 proposal, the EPA Administrator revised the PM NAAQS by adding new PM2 5 standards and by
30 revising the form of the 24-h PM10 standard. Specifically, in July 1997, the Administrator made
31 the following revisions to the PM NAAQS:
October 1999 2-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 (1) The suite of PM standards was revised to include an annual primary PM25 standard and a
2 24-h PM2 5 standard.
3 (2) The annual PM2 5 standard is met when the 3-year average of the annual arithmetic mean
4 PM2 5 concentrations, from single or multiple community-oriented monitors is less than or
5 equal to 15 //g/m3, with fractional parts of 0.05 or greater rounding up.
6 (3) The 24-h PM25 standard is met when the 3-year average of the 98th percentile of 24-h PM25
7 concentrations at each population-oriented monitor within an area is less than or equal to
8 65 //g/m3, with fractional parts of 0.5 or greater rounding up.
9 (4) The form of the 24-h PM10 (150 //g/m3) standard was revised to be based on the 3-year
10 average of the 99th percentile of 24-h PM10 concentrations at each monitor within an area.
11 (5) In addition, the Administrator retained the annual PM10 standard at the level of 50 //g/m3,
12 which is met when the 3-year average of the annual arithmetic mean PM10 concentrations at
13 each monitor within an area is less than or equal to 50 //g/m3, with fractional parts of 0.5 or
14 greater rounding up.
15 The principal focus of the last review of the air quality criteria and standards for PM was on
16 recent epidemiological evidence reporting associations between ambient concentrations of PM
17 and a range of serious health effects. Particular attention was given to several size-specific
18 classes of particles, including PM10 and the principal fractions of PM10, referred to as the fine
19 (PM2 5) and coarse (PM10_2 5) fractions. PM2 5 refers to particles with an upper 50% cutpoint of
20 2.5 //m aerodynamic diameter. PM10_2 5 refers to those particles with an upper 50% cutpoint of
21 10 //m and a lower 50% cut point of 2.5 //m aerodynamic diameter. In other words, the coarse
22 fraction (PM10_25) refers to the inhalable particles that remain if fine (PM25) particles are removed
23 from a sample of PM10 particles. As discussed in the 1996 PM AQCD, fine and coarse fraction
24 particles can be differentiated by their sources and formation processes and their chemical and
25 physical properties, including behavior in the atmosphere. Detailed discussions of atmospheric
26 formation, ambient concentrations, and health and welfare effects of PM, as well as quantitative
27 estimates of human health risks associated with exposure to PM, can be found in the 1996 PM
28 AQCD and in the 1996 OAQPS Staff Paper (U.S. Environmental Protection Agency, 1996b).
29
30
31
October 1999 2-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 2.2.2 Presidential Memorandum: Next Particulate Matter Review
2 and Research
3 On July 18,1997, the EPA published a final rule revising the NAAQS for PM (Federal
4 Register, 1997a), and, on the same day, a final rule revising the Ozone NAAQS (Federal
5 Register, 1997b). A Presidential Memorandum (Federal Register, 1997c) was also published
6 outlining the Administration's goals for implementing the revised PM and Ozone NAAQS. The
7 Memorandum directed EPA to provide to CASAC within 90 days and to publish a notice
8 outlining its schedule for the next periodic review of PM and to complete the next review,
9 including review by CASAC, within 5 years after issuance of the revised standards (i.e., by July
10 2002). Such a schedule would ensure that EPA's review of emerging scientific information,
11 which forms the criteria upon which the standards are based, and of the standards themselves will
12 have been completed prior to any areas being designated as "nonattainment" under the newly
13 established standards for fine particles (i.e., PM25 standards) and prior to the imposition of any
14 new controls related to the revised standards. The Presidential Memorandum also directed EPA
15 and other relevant Federal agencies to develop and implement a greatly expanded, coordinated
16 research plan. These PM research plans are outlined in the following section. To facilitate
17 timely scientific research within this review period, EPA initiated certain activities immediately,
18 as noted below in the discussion of the PM Research Program.
19
20 PM Research Program
21 The EPA has broadened its ongoing PM research activities by developing, in partnership
22 with other Federal agencies, a coordinated interagency PM research program. This interagency
23 program will contribute to expanding scientific knowledge of PM health effects, as well as the
24 development of improved monitoring methods and cost-effective mitigation strategies. The
25 interagency effort is also promoting further coordination with other research organizations
26 including state-, university-, and industry-sponsored research groups. Beginning in the fall of
27 1997, public participation has been and continues to be encouraged through workshops and
28 review of program documentation. Workshops and the availability of relevant documentation are
29 being announced in the Federal Register.
30 To aid identification of needed research efforts, EPA published a particulate matter health
31 risk research needs document (U.S. Environmental Protection Agency, 1998a). That document
October 1999 2-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 identifies research needed to improve scientific information supporting future health risk
2 assessment and review of the PM NAAQS. The document aimed to provide a foundation for PM
3 research coordination among Federal agencies and other research organizations and served as one
4 useful input to National Research Council PM research deliberations. In January 1998, the
5 National Research Council (NRC) established its Committee on Research Priorities for Airborne
6 Particulate Matter in response to a request from Congress in the Fiscal 1998 appropriation to
7 EPA. This Committee is charged to identify the most important research priorities relevant to
8 setting particulate matter standards, to develop a conceptual plan for particulate matter research,
9 and to monitor research progress toward improved understanding of the relationship between
10 particulate matter and public health. The Committee issued its first report in early 1998
11 (National Research Council, 1998) and a second one recently in 1999 (National Research
12 Council, 1999).
13 The EPA's research program includes studies to improve understanding of the formation
14 and composition of fine PM, the characteristics or components of PM that are responsible for its
15 health effects, the mechanisms by which these effects are produced, and improved measurements
16 and estimation of population exposures to PM. Specific EPA research efforts include controlled
17 human exposure studies, in vivo and in vitro toxicology, epidemiology, atmospheric sciences
18 including monitoring and modeling studies, development of data on emissions of fine particles
19 from stationary and mobile sources, and identification and evaluation of risk management
20 options. The results from these efforts, as well as related efforts by other Federal agencies and
21 the general scientific community, will advance the scientific and technical bases for future
22 decisions on the PM NAAQS and for the implementation of PM monitoring and control efforts.
23
24
25 2.3 CURRENT PM CRITERIA AND NAAQS REVIEW
26 2.3.1 Criteria Review Plans and Schedule
27 The EPA's plans for this current review of PM criteria are outlined in Table 2-1 below,
28 together with target dates for key milestones. As with all NAAQS reviews, the purpose is to
29 update the criteria and to determine whether it is appropriate to revise the standards in light of
30 new scientific and technical information. Although the EPA concluded in its recent final rule on
October 1999 2-6 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 2-1. SCHEDULE FOR DEVELOPMENT OF THE CURRENT REVISED
PARTICULATE MATTER AIR QUALITY CRITERIA DOCUMENT (PM AQCD)
Major Milestones
Target Dates
PM NAAQS Review Plan to CASAC
Prepare AQCD Development Plan
Begin Literature Search
Federal Register Call for Information/Sources Sought
CASAC Meeting on AQCD Development Plan
Prepare Workshop Drafts of Chapters
Peer Review Workshop
Prepare External Review Draft AQCD
First External Review Draft to CASAC
Public Comment Period on Draft AQCD
CASAC Meeting on Draft AQCD
Prepare Revised Draft AQCD
Second External Review Draft to CASAC
Public Comment Period on Second Draft
CASAC Meeting on Second Draft
Complete Final PM AQCD
October 1997
November 1997 to January 1998
February 1998
April 1998
May 1998
May to December 1998
April 1999
March to September 1999
October 1999
October 1999 to January 2000
December 1999
January to June 2000
June 2000
July - August 2000
September 2000
December 2000
Source: Modified from Federal Register (1997c).
1 the PM standards (Federal Register, 1997a) that the current scientific knowledge provides a
2 strong basis for the revised PM standards, including the establishment of PM2 5 standards, there
3 remain scientific uncertainties associated with the health effects of PM and with the means of
4 reducing such effects. Recognizing the importance of developing a better understanding of the
5 effects of fine particles on human health, including their causes and mechanisms, as well as the
6 species and sources of PM25, the EPA will continue to sponsor research to address these
7 uncertainties even as this criteria review progresses as per Table 2-1.
October 1999
2-7
DRAFT-DO NOT QUOTE OR CITE
-------
1 As with other NAAQS reviews, a rigorous assessment of relevant scientific information is
2 to be presented in this updated, revised PM AQCD being prepared by EPA's NCEA-RTP
3 Division. The development of the document has and will continue to involve substantial external
4 peer review through public workshops involving the general aerosol scientific community,
5 through iterative reviews of successive drafts by CAS AC, and through comments from the
6 public. The final document will reflect input received through these reviews and will serve to
7 evaluate and integrate the latest available scientific information to ensure that the review of the
8 PM standards is based on sound science. The schedule for this review will allow for
9 consideration of relevant new peer-reviewed scientific studies published or accepted for
10 publication from early 1996 (when the previous PM AQCD was completed) through mid-2000.
11 After the December 1999 CAS AC meeting noted above in Table 2-1, EPA's Office of Air
12 Quality Planning and Standards (OAQPS) will also start to prepare a Staff Paper (SP) for the
13 Administrator, drawing on information in the newly revised PM AQCD. The SP will evaluate
14 the policy implications of the key studies and scientific information contained in the AQCD and
15 identify critical elements that EPA staff believes should be considered in reviewing the
16 standards. The SP is intended to bridge the gap between the scientific review in the AQCD and
17 the public health and welfare policy judgments required of the Administrator in reviewing the
18 PM NAAQS. For that purpose, the SP will present technical analyses, including air quality
19 analyses and a quantitative health risk assessment, and other factors relevant to the evaluation of
20 the PM NAAQS, as well as staff conclusions and recommendations of options for the EPA
21 Administrator's consideration. The SP will also be reviewed by CAS AC and the public, and the
22 final SP will reflect the input received through these reviews.
23 Following completion of the final SP, the Administrator will then announce in the Federal
24 Register proposals for retaining or revising the current PM NAAQS, and opportunities will be
25 provided for public comment and CASAC review of those proposals. Taking into account public
26 comments and CASAC recommendations, final decisions regarding the current PM NAAQS
27 review are to be promulgated by July 2002.
28
29 2.3.2 Methods and Procedures for Document Preparation
30 The procedures followed for developing this revised PM AQCD build on the knowledge
31 and methods derived from the last PM, ozone, and CO AQCD preparation efforts. Briefly, the
October 1999 2-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 respective responsibilities for production of the present PM AQCD are as follows. An NCEA-
2 RTF team has been appointed to be responsible for developing and implementing the project plan
3 for preparation of the PM AQCD and inputs from individuals in other EPA program and policy
4 offices identified as part of the EPA PM Work Group have been obtained. The resulting project
5 plan was then discussed with CAS AC (May 1998) and appropriately revised. An ongoing
6 literature search has continued to be conducted to identify, to the extent possible, all PM
7 literature published since early 1996. Additionally, EPA published (1) a request for information
8 in the Federal Register asking for recently available research information on PM that may not yet
9 be published and (2) a request for individuals with the appropriate type and level of expertise to
10 contribute to the writing of PM AQCD materials to identify themselves (U.S. Environmental
11 Protection Agency, 1998b). Specific authors of chapters or sections of the proposed document
12 were selected on the basis of their expertise on the subject areas and their familiarity with the
13 relevant literature, and these include both EPA and non-EPA scientific experts. The project team
14 defined critical issues and topics to be addressed by the authors and provided direction in order to
15 emphasize evaluation of those studies most clearly identified as important for standard setting.
16 The main focus of this revised criteria document is the evaluation and interpretation of air
17 quality data and health and welfare effects information newly published since that assessed in the
18 1996 PM AQCD and likely to be useful in deriving criteria for PM NAAQS. Draft AQCD
19 chapters were evaluated via expert workshops and/or expert written peer reviews, which focused
20 on the selection of pertinent studies included in the chapters, the potential need for additional
21 information to be added to the chapters, and the quality of the summarization and interpretation
22 of the literature. The authors of the draft chapters then revised them on the basis of the workshop
23 and/or written expert review recommendations. These and other integrative summary materials
24 have been incorporated into this First External Review Draft of the PM AQCD now being
25 released for public comment and CASAC review. Necessary revisions will be made on the basis
26 of the public comments, CASAC recommendations, and newly emerging research results before
27 a further Second External Review will be released in mid-2000 for public comment and CASAC
28 review (Fall 2000). The final version of the newly revised PM AQCD will incorporate changes
29 made in response to public comments and CASAC review of that Second External Review Draft.
30
October 1999 2-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 New research results are being incorporated into this document as they become available.
2 In order to foster presentation and publication of any new research findings not published during
3 1999, NCEA-RTP is working with the Air and Waste Management Association (AWMA) to
4 hold an International Speciality Conference, entitled PM 2000: Particulate Matter and Health, in
5 late January 2000, in Charleston, SC. The conference is being co-sponsored in cooperation with
6 one or more government agencies or private organizations that also fund PM research. Topics to
7 be covered will include new research results concerning the latest advances in PM atmospheric
8 sciences (e.g., PM formation, transport, transformation); PM exposure; PM dosimetry and
9 extrapolation modeling; PM toxicology (e.g., mechanisms, laboratory animal models, human
10 clinical responses); and PM epidemiology. The main purpose of the conference will be to
11 facilitate having the latest scientific information available for incorporation into the final draft of
12 the revised U.S. EPA PM Air Quality Criteria Document (AQCD) in time for the anticipated
13 final CAS AC review of the draft PM AQCD projected for Fall 2000. Arrangements will be
14 made for PM 2000 presenters to submit written manuscripts at the conference and to have
15 professional societies/journals prepared to expedite processing of the submitted papers through
16 their peer-review processes, so that decisions on acceptance for publication can be made by
17 April/May 2000. Given that extensive additional information is expected to be published during
18 the next 6 to 9 mo as a result of the PM 2000 conference and elsewhere, the evaluations and
19 findings set forth in this draft of the revised PM AQCD, overall, must be considered only
20 provisional at this time.
21
22 2.3.3 Approach
23 The approach to organization and content of this revised PM AQCD is somewhat different
24 from those used for previous criteria documents. Since the recent document (U.S. Environmental
25 Protection Agency, 1996a) provides an extensive discussion of most topic areas, this new
26 document focuses more specifically on critical issues that have been identified as areas needed to
27 improve the scientific basis (criteria) for PM NAAQS, particularly for those areas in which the
28 information data base has continued to evolve rapidly.
29 Detailed review of key new research was undertaken as an initial step. However, instead of
30 presenting a comprehensive review of all the literature, emphasis in this revised AQCD is placed
31 on (1) the concise summary of key findings derived from previous PM criteria reviews and
October 1999 2-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 (2) evaluation of the most pertinent new key information, with greater emphasis on more
2 interpretive assessment. This approach reflects recommendations made by CAS AC.
3 An initial step was to review the available scientific literature and to focus on the selection
4 of pertinent issues to include in the document as the basis for the development of PM NAAQS
5 criteria. Preliminary issues were identified by the NCEA Project Team and through input from
6 other EPA program and policy offices. Development of issue topics started with the last AQCD,
7 staff paper, CAS AC and public reviews, EPA's PM Research Needs Document, and from the
8 standard promulgation process. Further identification and clarification of issues resulted from
9 the National Research Council (NRC) review and report on PM research priorities. The CASAC
10 review of the PM AQCD Development Plan and public comments on draft AQCD materials at
11 various stages of their development has also played an important role in issue identification.
12 To aid in development of a concise document, compilation of summary tables of the
13 relevant published literature and selective discussion of the literature was undertaken. Building
14 on the previous PM AQCD, most of the scientific information selected for review and discussion
15 in the text was from the literature published since completion of the previous PM AQCD (U.S.
16 Environmental Protection Agency, 1996a).
17
18 2.3.4 Key Issues of Concern
19 Summarized below are several broad topics related to the main issues of concern to be
20 addressed by this revised PM AQCD. The document reviews and assesses available data bearing
21 on each of the issues identified below.
22 1. Causality. Evaluation of the evidence for or against a causal relationship between health
23 outcomes and ambient PM and/or specific physical-chemical components.
24 • Specific components of interest include size classes such as PM10, PM10_25, PM25, and
25 ultrafine particles. Chemical components include transition metals, acidity, sulphates,
26 nitrates, and organics.
27 • Expand review of foundations of causal inference for associated PM air pollution health
28 effects.
29 • Access new long-term PM exposure and health data to broaden interpretation of long-term
30 exposure findings.
October 1999 2-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 • Review data exploring potential mechanisms of response to PM physical-chemical
2 characteristics, response pathway, and exposure-dose-response relationships (laboratory and
3 clinical research).
4 2. Uncertainties. In carrying out overall assessment, address the following types of uncertainty:
5 • Uncertainties between stationary PM monitoring instruments and personal exposure to PM
6 of ambient origin, especially for susceptible groups and their related activity patterns.
7 Specific topics include: measurement error in outdoor monitors themselves; use of central
8 monitors for estimates of community concentrations; and the use of community
9 concentrations as a surrogate for personal exposure to particles of ambient origin.
10 • Uncertainties related to particulate matter size fraction, particle number, surface area, and
11 content of semi-volatile components.
12 • Uncertainties about the effects of long term PM exposure, such as life shortening, and
13 development and progression of disease.
14 • Uncertainties due to coexposure to other pollutants such as O3, SO2, and CO and
15 meterological factors.
16 • Uncertainties due to confounding in epidemiologic studies (e.g., economic factors,
17 demographic and lifestyle attributes, genetic susceptibility factors, occupational exposure,
18 and medical care).
19 • Uncertainty about shape of concentration-response (CR) relationships and associated
20 community risks (linear and threshold models for CR).
21 • Uncertainty about methods for synthesis of health outcome studies and evaluation of
22 sensitivity and confounding aspects including but not limited to meta-analysis.
23 3. Biological Mechanisms of Action. Evaluate data examining mechanisms for health outcomes
24 ofPM. Mechanistic information aids judgement about causality
25 • New studies have examined mechanisms of action of PM constituents including transition
26 metals, airborne allergens, and the generation of reactive oxygen species. Different cell
27 types have differing responses to PM components.
28 • Newly published studies have also identified potential mechanisms for the production of
29 cardiac arrhythmias by PM constituents, especially in animal models of disease and suggest
30 that particular attention should be accorded to PM metal constituents.
October 1999 2-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 • Although many new animal toxicology studies involve instillation of previously collected
2 particles and this technique is appropriate to study mechanisms of action, extrapolation to
3 human equivalent exposure/doses is uncertain.
4 • Ongoing work on the effects of lung inflammation and PM phagocytosis on subsequent
5 systemic effects, especially cardiac or vascular effects, is needed to provide further
6 information on the relationship between inhaled pollutants and cardiac events.
7 • Interpretation of concentrated ambient particles (CAPs) studies. Newly available
8 information will be examined from toxicology studies using devices which concentrate
9 (to variable extents) ambient PM to determine PM concentration-response relationships.
10 Again, difficulties are encountered regarding extrapolation to comparable human exposures
11 to ambient PM levels.
12 4. Susceptible Populations. Examine health outcome data to determine specific risk groups that
13 are more susceptible than normal healthy adults to adverse effects from PM exposure.
14 • Preexisting respiratory or cardiovascular disease in conjunction with advanced age, appear
15 to be important factors in PM mortality susceptibility.
16 • For morbidity health endpoints, children and asthmatics may potentially display increased
17 sensitivity to PM exposure. Data will be examined for coherence.
18 • Patterns of respiratory tract deposition, clearance, and retention in susceptible populations
19 have been recently studied and provide evidence of increased deposition associated with
20 lung disease.
21 • Animal models of lung disease exposed to PM constituents suggest a role for PM in cardiac
22 death.
23 5. Environmental Effects. Evaluation of several types of PM welfare effects:
24 • Vegetation and ecosystem effects.
25 • Visibility effects.
26 • Materials damage.
27 • Role of PM in atmospheric radiative transfer and potential consequences for penetration of
28 biologically harmful UVB to the earth surface and for climate change.
29 6. Background Information Topics Useful in Evaluating Health Risks. Topics include:
30 • New monitoring methods, especially methods used in epidemiology studies
31 • Indicator topics such as PM2 5 versus PMj 0. ultrafine; and PM2 5 versus PM10_2 5
October 1999 2-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 • New data patterns of daily and annual concentrations for PM25, PM10_25, and PM10
2
3
4 2.4 DOCUMENT CONTENT AND ORGANIZATION
5 The present document critically reviews and assesses relevant scientific literature on PM
6 through mid-1999. The material selected for review and comment in the text generally comes
7 from the more recent literature published since early 1996, with emphasis on studies conducted at
8 or near PM pollutant concentrations found in ambient air. Literature discussed in detail in the
9 previous 1996 EPA PM AQCD (U.S. Environmental Protection Agency, 1996a) is generally not
10 discussed in depth in this document. However, some limited treatment is included of the earlier
11 studies judged to be potentially useful in deriving PMNAAQS. Key literature is discussed in
12 the text and presented in tables as well. Reports of lesser importance for the purposes of this
13 document are typically only summarized in tables.
14 The primary emphasis is on consideration of published material that has undergone
15 scientific peer review. However, in the interest of admitting new and important information
16 expected to become available shortly, some material not yet fully published in the open literature
17 but meeting other standards of scientific reporting (i.e., peer review and quality assurance) are
18 now provisionally included. As noted earlier, emphasis has been placed on studies in the range
19 of current ambient levels. However, studies examining effects of higher concentrations have
20 been included if they contain unique data, documentation of a previously unreported effect or
21 mechanism. In reviewing and summarizing the literature, an attempt is made to present
22 alternative points of view where scientific controversy exists.
23 The present document consists of 9 chapters. The Executive Summary for the entire
24 document is contained in Chapter 1, followed by this general introduction in Chapter 2.
25 Chapters 3 through 5 provide background information on physical and chemical properties of PM
26 and related compounds; sources and emissions; atmospheric transport, transformation, and fate
27 of PM; methods for the collection and measurement of PM; and ambient air concentrations and
28 factors affecting exposure of the general population. Chapters 6 through 8 evaluate information
29 concerning the health effects of PM. Chapter 6 discusses epidemiological studies, and Chapter 7
30 discusses dosimetry of inhaled particles in the respiratory tract information on the toxicology of
31 specific types of PM constituents, including laboratory animal studies and controlled human
October 1999 2-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposure studies. Chapter 8 integrates key information on exposure, dosimetry, and critical
2 health risk issues derived from studies reviewed in the prior chapters. Lastly, Chapter 9
3 describes PM environmental effects on: vegetation and ecosystems; visibility; manmade
4 materials; and climate, as well as economic impacts of such welfare effects.
5 Neither control techniques nor control strategies for the abatement of PM are discussed in
6 this document, although some topics covered may be incidentally relevant to abatement
7 strategies. Technologies for controlling PM emissions are discussed in other documents issued
8 by EPA's Office of Air Quality Planning and Standards (OAQPS). Likewise, issues germane to
9 the scientific basis for control strategies, but not pertinent to the development of NAAQS criteria,
10 are addressed in numerous other documents issued by OAQPS.
11 In addition, certain issues of direct relevance to standard setting are not explicitly addressed
12 in this document, but are instead analyzed in documentation prepared by OAQPS as part of its
13 regulatory analyses materials. Such analyses include (1) delineation of particular adverse effects
14 that the primary and secondary NAAQS are intended to protect against, (2) exposure analyses
15 and assessment of consequent risk, and (3) discussion of factors to be considered in determining
16 an adequate margin of safety. Key points and conclusions from such analyses will be
17 summarized in the PM Staff Paper to be prepared by OAQPS and reviewed by CASAC.
18 Although scientific data contribute significantly to decisions regarding the above issues, their
19 resolution cannot be achieved solely on the basis of experimentally acquired information. Final
20 decisions on items (1) and (3) are made by the Administrator, as mandated by the Clean Air Act.
21 A fourth issue directly pertinent to standard setting is identification of populations at risk,
22 which is basically a selection by EPA of the subpopulation(s) to be protected by the promulgation
23 of a given standard. This issue is addressed only partially in this document. For example,
24 information is presented on factors, such as preexisting disease, that may biologically predispose
25 individuals and subpopulations to adverse effects from exposures to PM. The characterization of
26 population risk, however, requires information above and beyond data on biological
27 predisposition, such as information on estimated exposure, activity patterns, and personal habits.
28 Such information is typically addressed in the Staff Paper developed by OAQPS.
29
October 1999 2-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Code of Federal Regulations. (1977) Appendix B—Reference method for the determination of suspended
3 particulate matter in the atmosphere (high-volume method). C. F. R. 40: §50.
4 Federal Register. (1971) National primary and secondary ambient air quality standards. F. R. (April 30)
5 36:8186-8201.
6 Federal Register. (1987) Revisions to the national ambient air quality standards for particulate matter. F. R. (July 1)
7 52:24,634-24,669.
8 Federal Register. (1997a) National ambient air quality standards for particulate matter; final rule. F. R. (July 18)
9 62:38,652-38,752.
10 Federal Register. (1997b) National ambient air quality standards for ozone; final rule. F. R. (July 18)
11 62:38,856-38,896.
12 Federal Register. (1997c) Implementation of revised air quality standards for ozone and particulate matter. F. R.
13 (July 18) 62: 38,421-38,422.
14 National Research Council. (1998) Research priorities for airborne particulate matter. I. Immediate priorities and a
15 long-range research portfolio. Washington, DC: National Academy Press.
16 National Research Council. (1999) Research priorities for airborne particulate matter. II. Evaluating research
17 progress and updating the portfolio. Washington, DC: National Academy Press.
18 U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air
19 quality standards. U. S. C. 42: §§7408-7409.
20 U.S. District Court of Arizona. (1994) American Lung Association v. Browner. 884 F.Supp. 345
21 (CIV-93-643-TUC-ACM).
22 U.S. Environmental Protection Agency. (1996a) Air quality criteria for particulate matter. Research Triangle Park,
23 NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
24 Available from: NTIS, Springfield, VA; PB96-168224.
25 U.S. Environmental Protection Agency. (1996b) Review of the national ambient air quality standards for particulate
26 matter: policy assessment of scientific and technical information. OAQPS staff paper. Research Triangle
27 Park, NC: Office of Air Quality Planning and Standards; report no. EPA/452/R-96-013. Available from:
28 NTIS, Springfield, VA; PB97-115406REB.
29 U.S. Environmental Protection Agency. (1998a) Particulate matter research needs for human health risk assessment
30 to support future reviews of the national ambient air quality standards for particulate matter. Research
31 Triangle Park, NC: National Center for Environmental Assessment; report no. EPA/600/R-97/132F.
32 U.S. Environmental Protection Agency. (1998b) Review of national ambient air quality standards for particulate
33 matter. Commer. Bus. Daily: February 19. Available online: cbdnet.access.gpo.gov.
34 Wolff, G. T. (1996) Closure by the Clean Air Scientific Advisory Committee (CASAC) on the draft air quality
35 criteria for particulate matter [letter to Carol M. Browner, Administrator, U.S. EPA]. Washington, DC:
36 U.S. Environmental Protection Agency, Clean Air Scientific Advisory Committee.;
37 EPA-SAB-CASAC-LTR-96-005; March 15.
38
October 1999 2-16 DRAFT-DO NOT QUOTE OR CITE
-------
i 3. PHYSICS, CHEMISTRY, AND MEASUREMENT
2 OF PARTICULATE MATTER
3
4
5 An extensive review of the physics and chemistry of particulate matter (PM) was included
6 in Chapter 3 of the 1996 Air Quality Criteria for Particulate Matter (AQC PM 96) (U.S.
7 Environmental Protection Agency, 1996). Section 3.1 of this new version of the Air Quality
8 Criteria for Particulate Matter provides background information on the physics and chemistry of
9 atmospheric particles that may be useful in reading subsequent sections and chapters.
10 New information needed to understand risk assessment will be discussed. Emphasis will be
11 placed on the differences between fine and coarse particles and the differences between the
12 nuclei mode and the accumulation mode within fine particles.
13 Chapter 4 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996) contained a
14 review of the state-of-the-art of PM measurement technology. Since that time there has been
15 considerable progress in understanding problems and errors in the measurement of PM mass,
16 chemical composition, and physical parameters. There has also been some progress in
17 developing new and improved measurement techniques. Therefore, a more extensive survey on
18 measurement errors and newly developed measurement techniques is included in Section 3.2.
19 For more detail and older references the reader is referred to Chapter 3 and 4 of the AQC PM 96
20 (U.S. Environmental Protection Agency, 1996).
21
22
23 3.1 PHYSICS AND CHEMISTRY OF PARTICULATE MATTER
24 3.1.1 Definitions
25 Atmospheric particles originate from a variety of sources and possess a range of
26 morphological, chemical, physical, and thermodynamic properties. Examples include
27 combustion-generated particles such as diesel soot or fly ash, photochemically produced particles
28 such as those found in urban haze, salt particles formed from sea spray, and soil-like particles
29 from resuspended dust. Some particles are liquid, some are solid; others contain a solid core
30 surrounded by liquid. Atmospheric particles contain inorganic ions, metallic compounds,
October 1999 3-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 elemental carbon, organic compounds, and crustal compounds. Some atmospheric particles are
2 hygroscopic and contain particle-bound water. The organic fraction is especially complex,
3 containing hundreds of organic compounds.
4 The composition and behavior of airborne particles are fundamentally linked with those of
5 the surrounding gas. Aerosol is defined as a suspension of solid or liquid particles in air and
6 includes both the particles and all vapor or gas phase components of air. However, the term
7 aerosol is often used to refer to the suspended particles only. Particulate is an adjective and
8 should only be used as a modifier, as in particulate matter.
9 A complete description of the atmospheric aerosol would include an accounting of the
10 chemical composition, morphology, and size of each particle and the relative abundance of each
11 particle type as a function of particle size (Friedlander, 1970). However, most often the physical
12 and chemical characteristics of particles are measured separately. Number size distributions are
13 often determined by physical means, such as electrical mobility or light-scattering of suspended
14 particles. Chemical composition usually is determined by analysis of collected samples although
15 sulfate can be measured in-situ. The mass and average chemical composition of particles,
16 segregated by aerodynamic diameter, by cyclones or impactors, can also be determined,
17 However, recent developments in single particle analysis techniques, by electron microscopy
18 with x-ray analysis of particles collected on a substrate or by mass spectroscopy of suspended
19 particles passing through a sensing volume, provide elemental composition of individual
20 particles by particle size and thus are bringing the description envisioned by Friedlander closer to
21 reality.
22
23 3.1.2 Physical Properties and Processes
24 3.1.2.1 Definitions of Particle Diameter
25 The diameter of a spherical particle may be determined geometrically, from optical or
26 electron microscopy; by light scattering and Mie theory; or by its behavior, such as its electrical
27 mobility or its aerodynamic behavior. However, atmospheric particles often are not spherical.
28 Therefore, their diameters are often described by an "equivalent" diameter, i.e., that of a unit
29 density sphere which would have the same physical behavior. The aerodynamic diameter is
30 important for particle transport, collection, and respiratory tract deposition. The aerodynamic
31 diameter, Da, depends on particle density and is defined as the diameter of a spherical particle
October 1999 3-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 with equal settling velocity but a material density of 1 g/cm3. Particles with the same physical
2 size and shape but different densities will have different aerodynamic diameters. Detailed
3 definitions of the various sizes and their relationships are given in standard aerosol textbooks,
4 e.g., Friedlander (1977), Reist (1984), and Seinfeld and Pandis (1998).
5
6 3.1.2.2 Aerosol Size Distributions
7 Particle size, as indexed by one of the "equivalent" diameters, is an important parameter in
8 determining the properties, effects and fate of atmospheric particles. The atmospheric deposition
9 rates of particles, and therefore, their residence times in the atmosphere, are a strong function of
10 aerodynamic diameter. The aerodynamic diameter also influences deposition patterns of
11 particles within the lung. Light scattering is strongly dependent on the optical particle size.
12 Particle size distributions, therefore, have a strong influence on atmospheric visibility and,
13 through their effect on radiative balance, on climate. Size distribution studies using impactors
14 give direct measurements of the aerodynamic diameter. The diameter of atmospheric particles
15 range from 1 nanometer to 100 // meters, 5 orders of magnitude. A variety of different
16 instruments, measuring a variety of equivalent diameters, are required to cover this range.
17 Older particle counting studies used optical particle counters to cover the range of 0.3 to
18 30 //m diameter. Diameters of particles below 0.5 //m were measured as mobility diameters.
19 The particle diameters used in size distribution graphs from these studies are usually given as
20 geometric diameters rather than aerodynamic diameters. In recent years, aerodynamic particle
21 sizers, which give a direct measurement of the aerodynamic diameter in the range of
22 approximately 0.7 to 10 //m diameter, have been used with electrical mobility analyzers, which
23 measure the mobility diameter from approximately 0.5 //m to very small particles of the order of
24 .005 //m, to cover the range of regulatory interest. Unfortunately, there is no agreed-upon
25 technique for combining the various equivalent diameters. Some workers use various
26 assumptions to combine the various measurements into one presentation; others report each
27 instrument separately. Therefore, the user of size distribution data must be careful to determine
28 exactly which equivalent diameter is reported.
29 Aerodynamic diameter is the most widely used equivalent diameter. Therefore, in future
30 discussions, particle diameters, unless otherwise indicated, will be understood to refer to the
31 aerodynamic diameter.
October 1999 3-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 Particle Size Distribution Functions
2 The distribution of particles with respect to size is an important physical parameter
3 governing their behavior. Because atmospheric particles cover several orders of magnitude in
4 particle size, size distributions are often expressed in terms of the logarithm of the particle
5 diameter, on the X-axis, and the measured differential concentration on the Y-axis:
6 AN/A(logDp) = the number of particles per cm3 of air having diameters in the size range from log
7 Dp to log(Dp + ADp). It is not proper formally to take the logarithm of a dimensional quantity.
8 However, one can think of the distribution as a function of log(Dp/Dp0), where the reference
9 diameter Dp0 = 1 //m is not explicitly stated. If AN/A(logDp) is plotted on a linear scale, the
10 number of particles between Dp and Dp + ADp is proportional to the area under the curve of
11 AN/A(logDp) versus logDp. Similar considerations apply to distributions of surface, volume, and
12 mass. It has been found that atmospheric aerosol size distributions frequently may be
13 approximated by a sum of log-normal distributions corresponding to the various modes or
14 fractions. When approximated by a function, the distributions are usually given as dN/d(log Dp)
15 rather than AN/A(log Dp).
16
17 Atmospheric Aerosol Size Distributions
18 Averaged atmospheric size distributions are shown in Figures 3-1,2, and 3 (Whitby, 1978;
19 Whitby and Sverdrup, 1980). Figure 3-1 describes the number of particles as a function of
20 particle diameter for rural, urban-influenced rural, urban, and freeway-influenced urban aerosols.
21 For some of the same data, the particle volume distribution is shown in Figure 3-2. Figure 3-3
22 show the number, surface, and volume distribution for the grand average continental size
23 distribution. Note that the particle diameter is always shown on a logarithmic scale. The particle
24 number is frequently shown on a logarithmic scale in order to display the wide range in number
25 concentration for different particle sizes and different sites. Volume and surface area, and
26 sometimes number, are shown on an arithmetic scale with the distributions plotted such that the
27 volume, surface area, or number of particles in any specified size range is proportional to the
28 corresponding area under the curve. These distributions show that most of the particles are quite
29 small, below 0.1 //m, while most of the particle volume (and therefore most of the mass) is found
30 in particles > 0.1 //m. The surface area peaks around 0.1 //m.
31
October 1999 3-4 DRAFT-DO NOT QUOTE OR CITE
-------
1,000,000 -
10,000 -
o: 1 -
O,
^3)
| 0.01 H
z
T3
0.0001 -
0.000001 -
• Clean Rural
Urban Influenced Rural
Average Urban
Urban + Freeway
200,000
0.01 0.1 1 10 100
Particle Diameter, Dp (u.m)
O
C)
T3
150,000
100,000 -
50,000 -
0.01 0.1 1 10
Particle Diameter, Dp (pm)
Figure 3-1. Number of particles as a function of particle diameter: (a) number
concentrations are shown on a logarithmic scale to display the wide range by
site and size; (b) number concentrations for the average urban distribution
are shown on a linear scale for which the area under any part of the curve is
proportional to particle number in that size range.
Source: Whitby and Sverdrup (1980).
1 An important feature of the mass or volume size distributions of atmospheric aerosol is
2 their multimodal nature. Volume distributions, measured in ambient air in the United States, are
3 almost always found to be bimodal, with a minimum between 1.0 and 3 //m. The distribution of
4 particles that are mostly larger than the minimum is termed "coarse". The distribution of
5 particles that are mostly smaller than the minimum is termed "fine." Whitby and Sverdrup
6 (1980); Whitby (1978); and Willeke and Whitby (1975) identified three modes: nuclei,
7 accumulation, and coarse. The three modes are most apparent in the freeway-influenced size
8 distribution of Figure 3-2b and in the surface area distribution of Figure 3-3b. However, the
9 nuclei mode, corresponding to particles below about 0.1 //m, may not be noticeable in volume or
October 1999
3-5
DRAFT-DO NOT QUOTE OR CITE
-------
/u -
65 -
60 -
55 -
50 -
_ 45 -
; 40 -
^35 -
i,30 -
7)
| 25 -
3 20 -
15 -
10 -
5 -
• \
• \ Rural
.' * South-Central
r • New Mexico
/ v' A\
/ / \\
J'"'~\\ \
/ _.- --^=^: ^s ^'~'~--..^^
70
65 -
60 -
55 -
50 -
« 45 -
f 40 -
^35 -
3
9.30 -
Ol
=§ 25 "
^ 20 -
15 -
10 -
5 -
0
Average Urban
Urban + Freeway
0.01 0.1 1 10
Particle Diameter, Dp (|jm)
100 0.01 0.1 1 10 100
Particle Diameter, Dp (|jm)
Figure 3-2. Particle volume distribution as a function of particle diameter: (a) for the
averaged rural and urban-influenced rural number distributions shown in
Figure 3-1 and a distribution from south central New Mexico, and (b) for the
averaged urban and freeway-influenced urban number distributions shown in
Figure 3-1.
Source: Whitby and Sverdrup (1980) and Kim et al. (1993).
1 mass distributions. The middle mode, from 0.1 to 1 or 2 //m, is the accumulation mode. Fine
2 particles include both the accumulation and the nuclei modes. The third mode, containing
3 particles larger than 1 or 2 //m, is known as the coarse particle mode. The number concentrations
4 of coarse particles are usually too small to see in arithmetic plots (Figures 3-lb and 3-3a) but can
5 be seen in a logarithmic plot (Figure 3-lb). Whitby and Sverdrup (1980) observed that rural
6 aerosols, not influenced by sources, have a small accumulation mode and no observable nuclei
7 mode. For urban aerosols, the accumulation and coarse particle modes are comparable in
8 volume. The nuclei mode is small in volume but it dominates the number distributions of urban
9 aerosols. Whitby's conclusions were based on extensive studies of size distributions in a number
October 1999
3-6
DRAFT-DO NOT QUOTE OR CITE
-------
Nn = 7.7x 10
DGNn = 0.013
Nc = 4.2
DGNC = 0.97
CO
30 H
o
10H
Vn = 0.33
DGVn = 0.031
0.001
0.01
0.1
1.0
i
10
Dr
100
Figure 3-3. Distribution of coarse (c), accumulation (a), and nuclei or ultrafine (n), mode
particles by three characteristics, volume (V), surface area (S), and number
(N) for the grand average continental size distribution. DGV = geometric
mean diameter by volume; DGS = geometric mean diameter by surface area;
DGN = geometric mean diameter by number; Dp = geometric diameter.
Source: Whitby (1978).
October 1999
3-7
DRAFT-DO NOT QUOTE OR CITE
-------
1 of western and midwestern locations during the 1970's (Whitby, 1978; Whitby and Sverdrup,
2 1980). No size-distribution studies of similar scope have been published since then. Newer
3 results from particle counting techniques and impactor, size-segregation studies, including data
4 from Europe (U.S. Environmental Protection Agency, 1996) and Australia (Keywood et al.,
5 1999), show similar results.
6
7 Definitions of Particle Size Fractions
8 In the preceding discussion several subdivisions of the aerosol size distribution were
9 identified. The aerosol community uses four different approaches or conventions in the
10 classification of particles by size: (1) modes, based on the observed size distributions and
11 formation mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling
12 device; (3) occupational sizes, based on the entrance into various compartments of the respiratory
13 system; and (4) legally-specified, regulatory sizes for air quality standards.
14
15 Modal. The modal classification, first proposed by Whitby (1978), is shown in Figure 3-3.
16 The nuclei mode can be seen clearly in the volume distribution only in traffic or near traffic or
17 other sources of nuclei mode particles (Figure 3-4). The observed modal structure is frequently
18 approximated by several log-normal distributions. Definitions of terms used to describe size
19 distributions in modal terms are given below.
20
21 Coarse Mode: The distribution of particles with diameters mostly greater than the
22 minimum in the particle mass distribution, which generally occurs between 1 and 3 //m.
23 These particles are usually mechanically generated.
24
25 Fine Mode: The distribution of particles with diameters mostly smaller than the minimum
26 in the particle mass distribution, which generally occurs between 1 //m and 3 //m. These
27 particles are generated in combustion or formed from gases. The fine mode includes the
28 accumulation mode and the nuclei mode.
29
30 Nuclei Mode: That portion of the fine particle fraction with diameters below about 0.1 //m.
31 The nuclei mode can be observed as a separate mode in mass or volume distributions only
32 in clean or remote areas or near sources of new particle formation by nucleation.
33
34 Toxicologists use ultrafine to refer to particles, generated in the laboratory, which are in the
35 nuclei-mode size range. Aerosol physicists and material scientists tend to use nanoparticles
36 to refer to particles in this size range generated in the laboratory.
October 1999 3-8 DRAFT-DO NOT QUOTE OR CITE
-------
C?4
O)
3
2
1
Mechanically
Generated
0.002
0.01
Nuclei Mode
0.1 1
Geometric Diameter, Dp, |jm
Accumulation Mode
10
Coarse Mode
100
Fine-Mode Particles
Coarse-Mode Particles
Figure 3-4. Volume size distribution, measured in traffic, showing fine-mode and
coarse-mode particles and the nuclei and accumulation modes within the
fine particle mode. DGV (geometric mean diameter by volume, equivalent to
volume median diameter) and og (geometric standard deviation) are shown for
each mode. Also shown are transformation and growth mechanisms (e.g.,
nucleation, condensation, and coagulation).
Source: Adapted from Wilson and Suh (1997).
1
2
3
4
5
6
7
Accumulation Mode: That portion of the fine particle fraction with diameters above about
0.1 //m. Accumulation-mode particles normally do not grow into the coarse mode.
Nuclei-mode particles grow by coagulation (two particles combining to form one) or by
condensation (low-equilibrium vapor pressure gas molecules condensing on a particle) and
"accumulate" in this size range.
Over the years, the terms fine and coarse, as applied to particle sizes, have lost the precise
meaning given in Whitby's (1978) definition. In any given article, therefore, the meaning of fine
and coarse, unless defined, must be inferred from the author's usage. In particular, PM2 5 and
October 1999
3-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
fine mode particles are not equivalent. In this document, the term mode is used with fine and
coarse when it is desired to specify the distribution of fine-mode particles or coarse-mode
particles as shown in Figures 3-4 and 3-5.
70
E
E
Q
O)
-------
1 Sampler Cut Point. Another set of definitions of particle size fractions arises from
2 considerations of size-selective sampling. Size-selective sampling refers to the collection of
3 particles below or within a specified aerodynamic size range, usually defined by the upper 50%
4 cut point size, and has arisen in an effort to measure particle size fractions with some special
5 significance, e.g., health, visibility, source apportionment, etc. Dichotomous samplers split the
6 particles into smaller and larger fractions, which may be collected on separate filters. Cascade
7 impactors use multiple size cuts to obtain a distribution of size cuts for mass or chemical
8 composition measurements. One-filter samplers with a variety of upper size cuts have also been
9 used.
10
11 Occupational Health Size Cuts. The occupational health community has defined size
12 fractions for use in the protection of human health. This convention classifies particles into
13 inhalable, thoracic, and respirable particles according to their upper size cuts. However, these
14 size fractions may also be characterized in terms of their entrance into various compartments of
15 the respiratory system. Thus, inhalable particles enter the respiratory tract, including the head
16 airways. Thoracic particles travel past the larynx and reach the lung airways and the gas-
17 exchange regions of the lung. Respirable particles are a subset of thoracic particles which are
18 more likely to reach the gas-exchange region of the lung. In the past exact definitions of these
19 terms have varied among organizations. As of 1993 a unified set of definitions was adopted by
20 the American Conference of Governmental Industrial Hygienists (ACGIH) (1994), the
21 International Standards Organization (ISO), and the European Standardization Committee
22 (CEN). The curves which define inhalable (IPM), thoracic (TPM), and respirable (RPM)
23 particulate matter are shown in Figure 3-6.
24
25 Regulatory Size Cuts. In 1987, the NAAQS for PM were revised to use PM10, rather than
26 TSP, as the indicator for the NAAQS for PM (Federal Register, 1987). The use of PM10 as an
27 indicator is an example of size-selective sampling based on a regulatory size cut (Federal
28 Register, 1987). The selection of PM10 as an indicator was based on health considerations and
29 was intended to focus regulatory concern on those particles small enough to enter the thoracic
30 region. The PM25 standard, set in 1997, is also an example of size-selective sampling based on a
31 regulatory size cut (Federal Register, 1997). The PM2 5 standard was based primarily on
October 1999 3-11 DRAFT-DO NOT QUOTE OR CITE
-------
10
• IPM
• TPM
O RPM
v PM
2.5 4 10 20 50
Aerodynamic Diameter (|jm)
100
Figure 3-6. Specified particle penetration (size-cut curves) through an ideal (no-particle-
loss) inlet for five different size-selective sampling criteria. PM 10 is defined in
the Code of Federal Regulations (1991). PM25 is also defined in the Federal
Register (1997). Size-cut curves for inhalable particulate matter (IPM),
thoracic particulate matter (TPM) and respirable particulate matter (RPM)
size cuts are computed from definitions given by American Conference of
Governmental and Industrial Hygienists (1994).
1 epidemiological studies using concentrations measured with PM2 5 samplers as an exposure
2 index. However, the PM25 sampler was not designed to collect respirable particles. It was
3 designed to collect fine-mode particles because of their different sources (Whitby et al., 1974).
4 Thus, the PM25 standard will increase regulatory concern with the sources of fine-mode particles.
5 Prior to 1997, the indicator for the NAAQS for PM was total suspended particulate matter
6 (TSP). TSP is defined by the design of the High Volume Sampler (hivol) which collects all of
7 the fine particles but only part of the coarse particles. The upper cut off size of the hivol depends
October 1999
3-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 on the wind speed and direction, and may vary from 25 to 40 //m. Heroic measures, such as
2 those undertaken with the Wide Range Aerosol Classifier (WRAC), are required to collect the
3 entire coarse mode (Lundgren and Burton, 1995).
4 An idealized distribution showing the normally observed division of ambient aerosols into
5 fine-mode particles and coarse-mode particles, and the size fractions collected by the WRAC,
6 TSP, PM10, PM25 and PM(10_25) samplers, is shown in Figure 3-5. PM10 samplers, as defined in
7 Appendix J to 40 Code of Federal Regulations (CFR) Part 50 (Code of Federal Regulations,
8 199 la, Federal Register, 1987), collect all of the fine particles and part of the coarse particles.
9 The upper cut point is defined as having a 50% collection efficiency at 10±0.5 //m aerodynamic
10 diameter. The slope of the collection efficiency curve is defined in amendments to 40 CFR,
11 Part 53, (Code of Federal Regulations, 1991b). An example of a PM10 size-cut curve is shown in
12 Figure 3-6.
13 An example of a PM25 size-cut curve is also shown in Figure 3-6. The PM25 size-cut
14 curve, however, is defined by the design of the Federal Reference Method Sampler. The basic
15 design of the FRM is given in the Federal Register (1997, 1998) and as 40 CFR Part 50,
16 Appendix L in the Code of Federal Regulations (Code of Federal Regulations, 1997a).
17 Additional performance specifications are given in 40 CFR Parts 53 and 58 (Code of Federal
18 Regulations, 1997b). Each actual PM25 reference method, as represented by a specific sampler
19 design and associated manual operational procedures, must be designated as a reference method
20 under Part 53 (see section 1.2 of Appendix L). Thus there may be many somewhat different
21 PM25 FRMs (currently, 6 have been designated).
22 Papers discussing PM10 or PM2 5 frequently insert an explanation such as PMX (particles less
23 than x //m diameter) or PMX (nominally, particles with aerodynamic diameter x are collected and not all particles
-------
O
'-i—>
03
-I—<
0)
100
90-
80-
70-
60-
50-
S. 40H
30-
20-
10-
0-
Kimoto cyclonic inlet
Manufacturer
• Tsai and Cheng (1996)
Wedding Cyclonic Inlet
O U=2 km/h
D U=8 km/h
A U=24 km/h
Wedding and Weigand
(1993)
1
Figure 3-7.
8 10
20
30
Aerodynamic diameter, |jm
Source: Tsai and Cheng (1996)
Comparison of penetration curves for two PM10 beta gauge samplers using
cyclone inlets. The Wedding PM10 sampler uses the U.S. EPA definition of
PMX as x = 50% cut point. The Kimoto PM10 defines PMX as x = the 100% cut
point (or zero penetration).
Source: Tsai and Cheng (1996).
1 In an analysis reported in 1979, EPA scientists endorsed the need to measure fine and
2 coarse particles separately (Miller et al., 1979). Based on the availability of a dichotomous
3 sampler with a separation size of 2.5 //m, they recommended 2.5 //m as the cut point between
4 fine and coarse particles. Because of the wide use of this cut point, the PM2 5 fraction is
5 frequently referred to as "fine" particles. However, while the PM25 sample contains all of the
6 fine particles it may, especially in dry areas or during dry conditions, collect a small fraction of
7 the coarse particles. A PM10-PM25 size fraction may be obtained from a dichotomous sampler or
8 by subtracting the mass collected by a PM2 5 sampler from the mass collected by a PM10 sampler.
October 1999
3-14
DRAFT-DO NOT QUOTE OR CITE
-------
1 The resulting PM10-PM25 mass, or PM(10_25), is sometimes called "coarse" particles. However,
2 it would be more correct to call PM2 5 an indicator of fine-mode particles (since it contains some
3 coarse-mode particles), PM(10_25) an indicator of the thoracic component of coarse-mode particles
4 (since it excludes some coarse-mode particles below 2.5 //m and above 10 //m). It would be
5 appropriate to call PM10 an indicator of thoracic particles. PM10 and thoracic PM, as shown in
6 Figure 3-6, have the same 50% cut point. However, the thoracic cut is not as sharp as the PM10
7 cut so thoracic PM contains some particles between 10 and 30 //m diameter that are excluded
8 from PM10.
9
10 3.1.2.3 Nuclei-Mode Particles
11 Some further discussion of nuclei-mode particles is justified because of their possible
12 health importance. The current PM25 standard is based largely on the statistical association of
13 health outcomes with particle mass. A recent epidemiologic study in Erfurt, Germany (Peters
14 et al., 1997) found a better statistical association of asthma with particle number than with PM25.
15 Toxicologic studies (Oberdorster et al., 1992, 1994; Li et al., 1997; Lison et al., 1997) suggest
16 that nuclei-size particles (diameter
-------
1 growth of fine particles may be found in the AQC PM 1966 (U.S. Environmental Protection
2 Agency, 1966).
3
4 Equilibrium Vapor Pressures
5 An important parameter in particle nucleation and in particle growth by condensation is the
6 saturation ratio S, defined as the ratio of the partial pressure of a species, p, to its equilibrium
7 vapor pressure above a flat surface, p0: S = p/p0. For either condensation or nucleation to occur,
8 the species vapor pressure must exceed its equilibrium vapor pressure. For particles, the
9 equilibrium vapor pressure is not the same as p0. Two effects are important: (1) the Kelvin
10 effect, which is an increase in the equilibrium vapor pressure above the surface due to its
11 curvature; thus very small particles have higher vapor pressures and will not be stable to
12 evaporation until they attain a critical size and (2) the solute effect, which is a decrease in the
13 equilibrium vapor pressure of the liquid due to the presence of other compounds in solution.
14 For an aqueous solution of a nonvolatile salt, the presence of the salt decreases the
15 equilibrium vapor pressure of the water over the droplet. This effect is in the opposite direction
16 of the Kelvin effect, which increases the equilibrium vapor pressure above a droplet because of
17 its curvature. The existence of an aqueous solution will also influence the vapor pressure of
18 water-soluble species. The vapor pressure behavior of mixtures of several liquids or of liquids
19 containing several solutes is complex.
20
21 New Particle Formation
22 When the vapor concentration of a species exceeds its equilibrium concentration (expressed
23 as its equilibrium vapor pressure), it is considered condensable. Condensable species can either
24 condense on the surface of existing particles or can form new particles. The relative importance
25 of nucleation versus condensation depends on the rate of formation of the condensable species
26 and on the surface or cross-sectional area of existing particles (McMurry and Friedlander, 1979).
27 In ambient urban environments, the available particle surface area is sufficient to rapidly
28 scavenge the newly formed condensable species. Formation of new particles (nuclei mode) is
29 usually not important except near sources of condensable species. Wilson et al. (1977) report
30 observations of the nuclei mode in traffic. New particle formation can also be observed in
31 cleaner, remote regions. Bursts of new particle formation in the atmosphere under clean
October 1999 3-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 conditions correspond to low aerosol surface area concentrations (Covert et al., 1992). High
2 concentrations of nuclei mode particles can occur in regions corresponding to low particle mass
3 concentrations, indicating that new particle formation is inversely related to the available aerosol
4 surface area (Clarke, 1992).
5
6 Sources of Nuclei Mode Particles
7 Nuclei mode particles are the result of nucleation of gas phase species to form condensed
8 phase species with very low equilibrium vapor pressure. In the atmosphere there are four major
9 classes of sources which yield particulate matter with equilibrium vapor pressures low enough to
10 form nuclei mode particles:
11 (1) Particles containing heavy metals. Nuclei mode particles of metal oxides or other
12 metal compounds are generated when metallic impurities in coal or oil are vaporized during
13 combustion and the vapor undergoes nucleation. Metallic ultrafine particles may also
14 formed from metals in lubricating oil or fuel additives that are vaporized during
15 combustion of gasoline or diesel fuels. Nuclei-mode metallic particles were discussed in
16 6.9 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996)
17 (2) Elemental carbon or soot, Ce. Ce particles are formed primarily by condensation of C2
18 molecules generated during the combustion process. Because Ce has a very low
19 equilibrium vapor pressure, ultrafine Ce particles can nucleate even at high temperatures
20 (Kittelson, 1998; Morawska et al., 1998a).
21 (3) Sulfates. Sulfuric acid (H2SO4), or its neutralization products with ammonia (NH3),
22 ammonium sulfate ((NH4)2SO4) or ammonium acid sulfate (NH4HSO4), are generated in the
23 atmosphere by conversion of sulfur dioxide (SO2) to H2SO4. As H2SO4 is formed, it can
24 either nucleate to form new ultrafine particles or it can condense on existing nuclei mode or
25 accumulation mode particles. (Clark and Whitby, 1975; Whitby, 1978).
26 (4) Organic carbon. Recent smog chamber studies and indoor experiments show that
27 atmospheric oxidation of certain organic compounds found in the atmosphere can produce
28 highly oxidized organic compounds with an equilibrium vapor pressure sufficiently low to
29 result in nucleation (Kamens et al., 1999; Weschler and Shields, 1999).
30
October 1999 3-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 Concentration of Nuclei-Mode Particles: A Balance Between Formation and Removal
2 Nuclei-mode particles may be removed by dry deposition or by growth into the
3 accumulation mode. This growth takes place as other low vapor pressure material condenses on
4 the particles or as nuclei-mode particles coagulate with themselves or with accumulation mode
5 particles. Since the rate of coagulation would vary with the concentration of accumulation-mode
6 particles, it might be expected that the concentration of nuclei-mode particles would increase
7 with a decrease in accumulation-mode mass. On the other hand, the concentration of particles
8 would be expected to decrease with a decrease in the rate of generation of particles by reduction
9 in emissions of metal and carbon particles or a decrease in the rate of generation of H2SO4 or
10 condensable organic vapor. The rate of generation of H2SO4 depends on the concentration of
11 SO2 and OH, which is generated primarily by the photolysis of O3. Thus, the reductions in SO2
12 and O3 that are expected to form a major basis for attaining PM25 and O3 standards and
13 implementation of Title II and Title IV Clean Air Act programs should lead to a decrease in the
14 rate of generation of H2SO4 and condensable organic vapor and a decrease in the concentration of
15 nuclei-mode particles. These processes can be modeled using a general dynamic equation for
16 particle size distribution (Friedlander, 1977) or by aerosol dynamics modules in newer air quality
17 models (Binkowski and Shanker, 1995; Binkowski and Ching, 1996).
18 Since preliminary studies of the effects of ultrafine particles suggest the potential for
19 enhanced toxicity of this size range, further research in this area is important. It is possible that
20 freshly generated ultrafine particles relatively near significant sources could present an additional
21 risk to health, above those associated with particle mass. It will, therefore, be important to
22 monitor particle number and surface as well as mass to further delineate the relative effectiveness
23 of strategies for reducing particle mass, surface, and number.
24
25 3.1.3 Chemistry of Atmospheric PM
26 The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
27 ions; particle-bound water; elemental carbon; a great variety of organic compounds; and crustal
28 material. Atmospheric PM also contains a large number of elements in various compounds and
29 concentrations. More information, references, and the composition of PM, measured in a large
30 number of studies in the U.S., may be found in AQC PM 96 (U.S. Environmental Protection
October 1999 3-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 Agency, 1996). The composition and concentrations of PM are discussed in Chapter 4 of this
2 document.
3
4 3.1.3.1 Chemical Composition and Its Dependence on Particle Size
5 Studies conducted in most parts of the U.S. indicate that sulfate, ammonium, and hydrogen
6 ions; elemental carbon, secondary organic compounds and some primary organic compounds;
7 and certain transition metals are found predominantly in the fine particle mode. Crustal materials
8 such as calcium, aluminum, silicon, magnesium, and iron are found predominately in the coarse
9 particles. Some organic materials such as pollen, spores, and plant and animal debris are also
10 found predominantly in the coarse mode. Some components such as potassium and nitrate may
11 be found in both the fine and coarse particle modes but from different sources or mechanisms.
12 Potassium in coarse particles comes from soil. Potassium is also found in fine particles in
13 emissions from burning wood or cooking meat. Nitrate in fine particles comes primarily from
14 the reaction of gas-phase nitric acid with gas-phase ammonia to form particulate ammonium
15 nitrate. Nitrate in coarse particles comes primarily from the reaction of gas-phase nitric acid with
16 pre-existing coarse particles.
17
18 3.1.3.2 Primary and Secondary Particulate Matter
19 Particulate material can be primary or secondary. PM is called primary if it is in the same
20 chemical form in which it was emitted into the atmosphere. PM is called secondary if it is
21 formed by chemical reactions in the atmosphere. Primary coarse particles are usually formed by
22 mechanical processes. This includes material emitted in particulate form such as wind-blown
23 dust, sea salt, road dust, and combustion-generated particles such as fly ash and soot. Primary
24 fine particles are emitted from sources, either directly as particles or as vapors which rapidly
25 condense to form particles. This includes soot from diesel engines as well as compounds of As,
26 Se, Zn, etc., condensed from vapor formed during combustion or smelting. The concentration of
27 primary particles depends on their emission rate, transport and dispersion, and removal rate from
28 the atmosphere.
29 Secondary PM is formed by chemical reactions of free, adsorbed, or dissolved gases. Most
30 secondary fine PM is formed from condensable vapors generated by chemical reactions of
31 gas-phase precursors. Secondary formation processes can result in either the formation of new
October 1999 3-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles or the addition of particulate material to preexisting particles. Most of the sulfate and
2 nitrate and a portion of the organic compounds in atmospheric particles are formed by chemical
3 reactions in the atmosphere. Secondary aerosol formation depends on numerous factors
4 including the concentrations of precursors; the concentrations of other gaseous reactive species
5 such as ozone, hydroxyl radical, or hydrogen peroxide; atmospheric conditions including solar
6 radiation and relative humidity; and the interactions of precursors and preexisting particles within
7 cloud or fog droplets or on or in the liquid film on solid particles. As a result, it is considerably
8 more difficult to relate ambient concentrations of secondary species to sources of precursor
9 emissions than it is to identify the sources of primary particles.
10
11 Formation ofSulfates and Nitrates
12 A substantial fraction of the fine particle mass, especially during the warmer months of the
13 year, is secondary sulfate and nitrate, formed as a result of atmospheric reactions. Such reactions
14 involve the gas phase conversion of SO2 to H2SO4 by OH radicals and aqueous-phase reactions of
15 SO2 with H2O2, O3, or O2 (catalyzed by Fe and Mn). These heterogeneous reactions may occur in
16 cloud and fog droplets or in films on atmospheric particles. The NO2 portion of NOX can be
17 converted to HNO3 by reaction with OH radicals during the day. At night, NOX is also oxidized
18 to nitric acid by a sequence of reactions initiated by O3, that include nitrate radicals (NO3) and
19 dinitrogenpentoxide (N2O5). Both H2SO4 and HNO3 react with atmospheric ammonia (NH3).
20 Gaseous NH3 reacts with gaseous HNO3 to form particulate NH4NO3. Gaseous NH3 reacts with
21 H2SO4 to form acidic HSO4 (in NH4 HSO4) as well as in SO4 in (NH4)2SO4. In addition, acid
22 gases such as SO2 and HNO3 may react with coarse particles to form coarse secondary PM
23 containing sulfate and nitrate. Examples include reactions with basic compounds resulting in
24 neutralization, e.g., CaCo3 + 2NHO3 -> Ca (NO3)2 + H2CO31, or with salts of volatile acids
25 resulting in release of the volatile acid, e.g., SO2 + 2NaCl + H2O - Na2SO3 + 2HC11.
26 Chemical reactions of SO2 and NOX within plumes are an important source of H+, SO4 and
27 NO3. These conversions can occur by gas-phase and aqueous-phase mechanisms. In power-
28 plant or smelter plumes containing SO2 and NOX, the gas-phase chemistry depends on plume
29 dilution, sunlight and volatile organic compounds, either in the plume or in the ambient air
30 mixing into and diluting the plume. For the conversion of SO2 to H2SO4, the gas-phase rate in
31 such plumes during summer midday conditions in the eastern United States typically varies
October 1999 3-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 between 1 and 3% h"1 but in the cleaner western United States rarely exceeds 1% h"1. For the
2 conversion of NOX to HNO3, the gas-phase rates appear to be approximately three times faster
3 than the SO2 conversion rates. Winter rates for SO2 conversion are approximately an order of
4 magnitude lower than summer rates.
5 The contribution of aqueous-phase chemistry to particle formation in point-source plumes
6 is highly variable, depending on the availability of the aqueous phase (wetted aerosols, clouds,
7 fog, and light rain) and the photochemically generated gas-phase oxidizing agents, especially
8 H2O2 for SO2 chemistry. The in-cloud conversion rates of SO2 to SO4 can be several times
9 larger than the gas-phase rates given above. Overall, it appears that SO2 oxidation rates to SO4
10 by gas-phase and aqueous-phase mechanisms may be comparable in summer, but aqueous phase
11 chemistry may dominate in winter.
12 In the western United States, markedly higher SO2 conversion rates have been reported in
13 smelter plumes than in power plant plumes. The conversion is predominantly by a gas-phase
14 mechanism. This result is attributed to the lower NOX in smelter plumes. In power plant plumes
15 NO2 depletes OH and competes with SO2 for OH.
16 In urban plumes, the upper limit for the gas-phase SO2 conversion rate appears to be about
17 5% h"1 under the more polluted conditions. For NO2, the rates appear to be approximately three
18 times faster than the SO2 conversion rates. Conversion rates of SO2 and NOX in background air
19 are comparable to the peak rates in diluted plumes. Neutralization of H2SO4 formed by SO2
20 conversion increases with plume age and background NH3 concentration. If the NH3
21 concentrations are more than sufficient to neutralize H2SO4 to (NH4)2SO4, the HNO3 formed from
22 NOX conversions may be converted to NH4NO3.
23
24 Formation of Secondary Organic PM
25 Atmospheric reactions, involving volatile organic compounds such as alkenes, aromatics,
26 and terpenes (or any reactive organic gas which contains at least seven carbon atoms), yield
27 organic compounds with low ambient temperature vapor pressures which nucleate or condense
28 on existing particles to form secondary organic PM. While the mechanisms and pathways for
29 forming inorganic secondary particulate matter are fairly well known, those for forming
30 secondary organic PM are not as well understood. Ozone and the hydroxyl radical are thought to
31 be the major initiating reactants. However, HO2 and NO3 radicals may also initiate reactions and
October 1999 3-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 organic radicals may be nitrated by HNO2, HNO3, or NO2. Understanding the mechanisms of
2 formation of secondary organic PM is important because secondary organic PM can contribute in
3 a significant way to ambient PM levels, especially during photochemical smog episodes. Studies
4 of the production of secondary organic PM in ambient air have focused on the Los Angeles
5 Basin. Turpin and Huntzicker (1991) and Turpin et al. (1991) provided strong evidence that
6 secondary PM formation occurs during periods of photochemical ozone formation in Los
7 Angeles and that as much as 70% of the organic carbon in ambient PM was secondary in origin
8 during a smog episode in 1987. Schauer et al. (1996) estimated that 20 to 30% of the total
9 organic carbon PM in the <2.1 //m size range in the Los Angeles airshed is secondary in origin
10 on an annually averaged basis.
11 Pandis et al. (1992) identified three mechanisms for formation of secondary organic PM:
12 (1) condensation of oxidized end-products of photochemical reactions (e.g., ketones, aldehydes,
13 organic acids, and hydroperoxides); (2) adsorption of organic gases onto existing solid particles
14 (e.g., polycyclic aromatic hydrocarbons); and (3) dissolution of soluble gases which can undergo
15 reactions in particles (e.g., aldehydes). The first and third mechanisms are expected to be of
16 major importance during the summertime when photochemistry is at its peak. The second
17 pathway can be driven by diurnal and seasonal temperature and humidity variations at any time
18 of the year. With regard to the first mechanism, Odum et al. (1996) suggested that the products
19 produced by the photochemical oxidation of reactive organic gases are semivolatile and can
20 partition themselves onto existing organic carbon at concentrations below their saturation
21 concentrations. Thus, the yield of secondary organic PM depends not only on the identity of the
22 precursor organic gas but also on the ambient levels of organic carbon capable of absorbing the
23 oxidation product.
24 Haagen-Smit (1952) first demonstrated that hydrocarbons irradiated in the presence of NOX
25 produce light scattering aerosols. The aerosol forming potentials of a wide variety of individual
26 anthropogenic and biogenic hydrocarbons were compiled by Pandis et al. (1992) based mainly on
27 estimates made by Grosjean and Seinfeld (1989) and data from Pandis et al. (1991) for p-pinene
28 and Izumi and Fukuyama (1990) for aromatic hydrocarbons. Zhang et al. (1992) examined the
29 oxidation of cc-pinene. Pandis et al. (1991) found no aerosol products formed in the
30 photochemical oxidation of isoprene, although they and Zhang et al. (1992) found that the
31 addition of isoprene to reaction mixtures increased the reactivity of the systems studied. Further
October 1999 3-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 details about the oxidation mechanisms and secondary organic PM yields from various reactive
2 organic gases given in the above studies and estimates of the production rate of secondary
3 organic PM in the Los Angeles airshed are provided in the previous AQCD for PM (U.S.
4 Environmental Protection Agency, 1996).
5 More recently Odum et al. (1997a,b) have found that the aerosol formation potential of
6 whole gasoline vapor can be accounted for solely by summing the contributions of the individual
7 aromatic compounds in the fuel. In general, data for yields for secondary organic PM formation
8 can be broken into two distinct categories. The oxidation of aromatic compounds containing
9 zero or one methyl and ethyl groups (i.e., toluene, ethylbenzene, and ethyltoluenes) and
10 n-propylbenzene produce higher yields of secondary organic PM than did the oxidation of
11 aromatic compounds containing two or more methyl groups (i.e., xylenes, di-, tri-, and
12 tetra-methylbenzenes). Yields in the first group ranged from about 7 to 10% and in the second
13 group were generally between 3 and 4% within a range of existing organic carbon levels between
14 13 and 100 //g/m"3. This grouping is consistent with those found by Izumi and Fukuyama (1990).
15 Reasons for the differences in secondary organic PM yields found between the two classes of
16 compounds are not clear.
17 Hoffmann et al. (1997) using the same framework described above found secondary
18 organic PM yields of ~5% for open-chain biogenic hydrocarbons such as ocimene and linalool,
19 5 to 25% for monounsaturated cyclic monoterpenes such as cc-pinene, d-3 carene and terpinene-
20 4-ol, and ~40% for a cyclic monoterpene with two double bonds such as d-limonene. Secondary
21 organic PM yields of close to 100% were observed during the photochemical oxidation of one
22 sesquiterpene, trans-caryophyllene. These results were all obtained for initial hydrocarbon
23 mixing ratios of 100 ppb.
24 Kamens et al. (1999) observed secondary organic PM yields of 20-40% for cc-pinene.
25 Using information on the composition of secondary PM formed from cc-pinene (Jang and
26 Kamens, 1999), they were able to calculate formation rates with a kinetic model including
27 formation mechanisms for O3 + cc-pinene reaction products. Griffin et al. (1999) introduced the
28 concept of incremental aerosol reactivity, the change in the secondary organic aerosol mass
29 produced (in //g/m3) per unit change of parent organic reacted (in ppb), as a measure of the
30 aerosol-forming capability of a given parent organic compound in a prescribed mixture of other
31 organic compounds. They measured the incremental aerosol reactivity for a number of aromatic
October 1999 3-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 and biogenic compounds for four initial mixtures. Incremental aerosol reactivity ranged from
2 0.133 to 10.352 //gm"3 ppb"1 and varied by almost a factor of two depending on the initial
3 mixture.
4 Kao and Friedlander (1995) examined the statistical properties of a number of PM
5 components in the South Coast Air Basin. They found that the concentrations of non-reactive,
6 primary components of PM10 have approximately log normal frequency distributions and
7 constant values of geometric standard deviations (GSDs) regardless of source type and location
8 within their study area. However, aerosol constituents of secondary origin (e.g., SO4=, NH4+, and
9 NO3") were found to have much higher GSD's. Surprisingly, the GSD's of organic (1.87) and
10 elemental (1.74) carbon were both found to be within lo (0.14) of the mean GSD (1.85) for
11 non-reactive primary species, compared to GSD's of 2.1 for sulfate, 3.5 for nitrate, and 2.6 for
12 ammonium. These results suggest that most of the organic carbon seen in ambient samples is of
13 primary origin. Pinto et al. (1995) found similar results for data obtained during the summer of
14 1994. Further studies are needed to determine if these relations are valid at other locations and to
15 determine to what extent the results might be influenced by the evaporation of volatile
16 constituents after sampling.
17 It must be emphasized that the inferences drawn from field studies in the Los Angeles
18 Basin are unique to that area and cannot be extrapolated to other areas of the country.
19 In addition, there is a high degree of uncertainty associated with all aspects of the calculation of
20 secondary organic PM concentrations which is compounded by the volatilization of organic
21 carbon from filter substrates during and after sampling as well as potential positive artifact
22 formation from the absorption of gaseous hydrocarbon on quartz filters. Significant uncertainties
23 always arise in the interpretation of smog chamber data because of wall reactions. Limitations
24 also exist in extrapolating the results of smog chamber studies to ambient conditions found in
25 urban airsheds and forest canopies. Concentrations of terpenes and NOX are much lower in forest
26 canopies (Altshuller, 1983) than the levels commonly used in smog chamber studies. The
27 identification of aerosol products of terpene oxidation has not been a specific aim of field studies,
28 making it difficult to judge the results of model calculations of secondary organic PM formation.
29 Uncertainties also arise because of the methods used to measure biogenic hydrocarbon emissions.
30 Khalil and Rasmussen (1992) found much lower ratios of terpenes to other hydrocarbons (e.g.,
31 isoprene) in forest air than were expected, based on their relative emissions strengths and rate
October 1999 3-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 coefficients for reaction with OH radicals and O3. They offered two explanations, either the
2 terpenes were being removed rapidly by some heterogeneous process or emissions were
3 enhanced artificially by feedbacks caused by the bag enclosures they used. If the former
4 consideration is correct, then the production of aerosol carbon from terpene emissions could be
5 substantial; if the latter is correct, then terpene emissions could have been overestimated by the
6 techniques used.
7
8 3.1.3.3 Particle-Vapor Partitioning
9 Several atmospheric aerosol species, such as ammonium nitrate and certain organic
10 compounds, are semivolatile and are found in both gas and particle phases. A variety of
11 thermodynamic models have been developed to predict the temperature and relative humidity
12 dependence of the ammonium nitrate equilibria with gaseous nitric acid and ammonia. However,
13 under some atmospheric conditions, such as cool, cold, or very clean air, the relative
14 concentrations of the gas and solid phases are not accurately predicted by equilibrium
15 considerations alone, and transport kinetics can be important. The gas-particle distribution of
16 semivolatile organic compounds depends on the equilibrium vapor pressure of the compound,
17 total particle surface area, particle composition, atmospheric temperature, and relative humidity.
18 Although it is generally assumed that the gas-particle partitioning of semivolatile organics is in
19 equilibrium in the atmosphere, neither the equilibria nor the kinetics of redistribution are well
20 understood. Diurnal temperature fluctuations, which cause gas-particle partitioning to be
21 dynamic on a time scale of a few hours, can cause semivolatile compounds to evaporate during
22 the sampling process. The pressure drop across the filter can also contribute to loss of
23 semivolatile compounds. The dynamic changes in gas-particle partitioning, caused by changes in
24 temperature, pressure and gas-phase concentration, both in the atmosphere and after collection,
25 cause serious sampling problems which are discussed in Section 3.2.3.
26
27 Equilibria with Water Vapor
28 As a result of the equilibrium of water vapor with liquid water in hygroscopic particles,
29 many ambient particles contain liquid water (particle-bound water). Unless removed, this
30 particle-bound water will be measured as a component of the particle mass. Particle-bound water
31 is important in that it influences the size of the particles and in turn their aerodynamic properties
October 1999 3-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 (important for deposition to surfaces, to airways following inhalation, and in sampling
2 instrumentation) and their light scattering properties. The aqueous solution provides a medium
3 for reactions of dissolved gases, including reactions that do not take place in the gas phase. The
4 aqueous solutions may also act as a carrier to convey soluble toxic species to the gas-exchange
5 regions of the respiratory system, including species that would be removed by deposition in the
6 upper airways if the gas phase (Friedlander and Yeh, 1998; Kao and Friedlander, 1995; Wilson,
7 1995). An extensive review of this equilibrium as it pertains to ambient aerosols was given in
8 Chapter 3 of the AQC PM 96 (U.S. Environmental Protection, Agency, 1996).
9 Briefly the interaction of particles with water vapor may be described as follows.
10 As relative humidity increases, crystalline soluble salts in aerosol particles, such as (NH4) 2SO4,
11 NH4HSO4, or NH4NO3, undergo a phase transition to become aqueous solution aerosols.
12 According to the phase rule, for particles consisting of a single component, this phase transition
13 is abrupt, taking place at a relative humidity that corresponds to the vapor pressure of water
14 above the saturated solution (the deliquescence point). With further increase in relative
15 humidity, the particle adds water (and the concentration of the solute decreases) so that the vapor
16 pressure of the solution is maintained equal to that of the surrounding relative humidity; thus the
17 particle tends to follow the equilibrium growth curve. As relative humidity decreases, the
18 particle follows the equilibrium curve to the deliquescence point. However, rather than
19 crystallizing at the deliquescence relative humidity, the particle remains a solution in a
20 supersaturated solution to considerably lower relative humidities. Ultimately the particle
21 abruptly loses its water vapor (efflorescence), returning typically to the initial, stable crystalline
22 form.
23 For particles consisting of more than one component, the solid to liquid transition will take
24 place over a range of relative humidities, with an abrupt onset at the lowest deliquescence point
25 of the several components, and with subsequent growth as crystalline material in the particle
26 dissolves according to the phase diagram for the particular multicomponent system. Under such
27 circumstances a single particle may undergo several more or less abrupt phase transitions until
28 the soluble material is fully dissolved. At decreasing relative humidity such particles tend to
29 remain in solution to relative humidities well below the several deliquescence points. In the case
30 of the sulfuric acid-ammonium sulfate-water system the phase diagram is fairly completely
31 worked out. Mixed anion systems containing nitrate are more difficult due to the equilibrium
October 1999 3-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 between particulate NH4NO3 and gaseous NH3 and HNO3. For particles of composition
2 intermediate between NH4HSO4 and (NH4)2SO4 this transition occurs in the range from 40% to
3 below 10%, indicating that for certain compositions the solution cannot be dried in the
4 atmosphere. At low relative humidities, particles of this composition would likely be present in
5 the atmosphere as supersaturated solution droplets (liquid particles) rather than as solid particles,
6 thus they would exhibit hygroscopic rather than deliquescent behavior during relative humidity
7 cycles.
8 Other pure compounds, such as sulfuric acid (H2SO4), are hygroscopic, i.e. they form water
9 solutions at any relative humidity and maintain a solution vapor pressure over the entire range of
10 relative humidity. Soluble organic compounds may also contribute to the hygroscopicity of the
11 atmospheric aerosol (Saxena et al., 1995; Saxena and Hildeman, 1996), but the equilibria
12 involving organic compounds and water vapor, and especially for mixtures of salts, organic
13 compounds and water, are not so well understood. These equilibrium processes may cause an
14 ambient particle to significantly increase its diameter at relative humidities above about 40%
15 (Figure 3-8). A particle can grow to 5 times its dry diameter as the RH approaches 100%
16 (Figure 3-9). The Federal Reference Methods, for filter measurements of PM2 5 and PM10 mass,
17 require, after collection, equilibration at a specified, low relative humidity (-40% RH) to remove
18 particle-bound water (see 3.2 for details and references). Otherwise, particle mass would be a
19 function of relative humidity and, at higher relative humidities, the particle mass would be
20 largely particle-bound water. Continuous monitoring techniques must remove particle-bound
21 water before measurement, either by heating or dehumidification. Semivolatile material may be
22 lost during sampling, lost during equilibration, and is certainly lost when the collected sample is
23 heated above ambient. In addition to problems due to the loss of semivolatile species, recent
24 studies have shown that significant amounts of particle-bound water are retained in particles
25 collected on impaction surfaces even after equilibration and that the amount of retained particle-
26 bound water increases with relative humidity during collection (Hitzenberger et al., 1997). Large
27 increases in mass with increasing relative humidity were observed for the accumulation mode.
28 The change in particle size with relative humidity also means that particle measurements such as
29 surface area or volume, or composition as a function of size, must all be made at the same RH if
30 the results are to be comparable. These problems are addressed in more detail in Section 3.2 on
31 Measurement of Particulate Matter.
October 1999 3-27 DRAFT-DO NOT QUOTE OR CITE
-------
2.0
O
Q.
Q
1.5-
1.0
0
H
8
- 7
- 6
h 5
(NH4)2 S04
O
Q.
Q_
h 3
- 2
- 1
0
30
50
RH, %
70
90
Figure 3-8. Particle growth curves showing fully reversible hygroscopic growth of
sulfuric acid (H2SO4) particles, deliquescent growth of ammonium sulfate
[(NH4)2 SO4] particles at about 80% relative humidity (RH), hygroscopic
growth of ammonium sulfate solution droplets at RH greater than 80%, and
hysteresis (the droplet remains supersaturated as the RH decreases below
80%) until the crystallization point is reached.
Source: National Research Council (1993) adapted from Tang (1980).
October 1999
3-28
DRAFT-DO NOT QUOTE OR CITE
-------
o
Q
Q
o"
"on
5-
3
o
5
2 -
0
T
T
Theoretical Prediction at 22°c
ooooo Experimental Measurements
RH=99.8%
RH=98%
_^_i^^^^^^—
RH=96%
216
-125
-64
-27
-8
0
50 100 150
NH4 HSO4 Dry Particle Diameter (nm)
200
o
Figure 3-9. Theoretical predictions and experimental measurements of growth of
NH4HSO4 and particles at relative humidity between 95 and 100%.
Source: Lietal. (1992).
1 3.1.3.4 Removal Processes
2 The lifetimes of particles vary with size. Coarse particles can settle rapidly from the
3 atmosphere within hours, and normally travel only short distances. However, when mixed high
4 into the atmosphere, as in dust storms, the smaller-sized coarse-mode particles may have longer
5 lives and travel distances. Nuclei mode particles rapidly grow into the accumulation mode.
6 However, the accumulation mode does not grow into the coarse mode. Accumulation-mode fine
7 particles are kept suspended by normal air motions and have very low deposition rates to
8 surfaces. They can be transported thousands of km and remain in the atmosphere for a number of
October 1999
3-29
DRAFT-DO NOT QUOTE OR CITE
-------
1 days. Coarse mode particles of less than ~10 //m diameter as well as accumulation-mode and
2 nuclei-mode (or ultrafine) particles all have the ability to penetrate deep into the lungs and be
3 removed by deposition in the lungs. Dry deposition rates are expressed in terms of a deposition
4 velocity which varies as the particle size, reaching a minimum between 0.1 and 1.0 //m
5 aerodynamic diameter. Accumulation-mode particles are removed from the atmosphere
6 primarily by cloud processes. Fine particles, especially particles with a hygroscopic component,
7 grow as the relative humidity increases, serve as cloud condensation nuclei, and grow into cloud
8 droplets. If the cloud droplets grow large enough to form rain, the particles are removed in the
9 rain. Falling rain drops impact coarse particles and remove them. Ultrafine or nuclei mode
10 particles are small enough to diffuse to the falling drop and be removed. Falling rain drops,
11 however, are not effective in removing accumulation-mode particles.
12
13 3.1.3.5 Particulate Matter and Acid Deposition
14 The EPA is required by law to set primary standards to product human health and
15 secondary standards to mitigate welfare effects. The role of particles in reducing visibility and
16 affecting radiative balance through scattering and absorption of light is evident as are the effects
17 of particles in soiling and damaging materials. Visibility effects are addressed through regional
18 haze regulations. The direct effects of particles in scattering and absorbing light and the indirect
19 effects of particles on clouds are being addressed in climate change programs in several
20 government agencies with the lead role assigned to the Department of Energy. These welfare
21 effects are discussed briefly in Chapter 9.
22 Concerns over the possible ecological effects of acid deposition in the United States led to
23 the creation of a major research program in 1980 under the new National Acid Precipitation
24 Assessment Program (NAPAP). However, the role of PM in acid deposition has not always been
25 recognized. Acid deposition and PM are intimately related, however, first because particles
26 contribute significantly to the acidification of rain and secondly because the gas phase species
27 that lead to dry deposition of acidity are also precursors of particles. Therefore, reductions in
28 SO2 and NOX emissions will decrease both acid deposition and PM concentrations.
29 Sulfate, nitrate, and some partially oxidized organic compounds are hygroscopic and act as
30 nuclei for the formation of cloud droplets. These droplets provide chemical reactors in which
31 (even slightly) soluble gases can dissolve and react. Thus SO2 can dissolve in cloud droplets and
October 1999 3-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 be oxidized to sulfuric acid by dissolved ozone or hydrogen peroxide. These reactions do not
2 take place in the gas phase but only in solution in water. Sulfur dioxide may also be oxidized
3 more slowly by dissolved oxygen if metal catalysts such as iron or manganese are present in
4 solution. If the droplets evaporate, larger particles are left behind. If the droplets grow large
5 enough to fall out they will fall as rain and the particles will be removed from the atmospheric
6 with potential effects on the materials, plants, or soil on which the rain falls. (Similar
7 considerations apply to dew.) Atmospheric particles which nucleate cloud droplets may also
8 contain other soluble or non-soluble materials such as metal salts and PNA organic compounds
9 which may add to the toxicity of the rain. Thus the adverse effects of acid deposition on soils,
10 plants, and trees as well as lakes, streams and fish must be taken into account in setting
11 secondary PM standards. These effects are discussed in Chapter 9.
12 Sulfuric acid, ammonium nitrate, and organic particles are also deposited on surfaces by dry
13 deposition. The utilization of nitrate by plants leads to the production of acidity. Therefore, dry
14 deposition of particles can also contribute to the ecological damages caused by acid deposition.
15
16 3.1.3.6 Particles as Carriers of Toxic Species
17 Wilson (1995) has suggested that particles could carry toxic species into the deep lung.
18 "It is possible that water-soluble gases, which would be removed by deposition to wet surfaces in
19 the upper respiratory system during inhalation, could dissolve in particle-bound water and be
20 carried with the particles into the deep lung. Water-soluble gases in polluted air include oxidants
21 such as O3, H2O2, and organic peroxides; acid gases such as SO2, HC1, HNO3, HONO, and formic
22 acid; and polar organic species such as formaldehyde." Friedlander and Yeh (1998) have
23 discussed this possibility for peroxides. Wexler and Sarangapani (1998), quoted below, have
24 discussed this process and investigated it quantitatively.
25 Air pollutants are deposited in the human airway via two pathways—particle
26 deposition and vapor deposition. In the absence of particles, vapors deposit at
27 different locations in the lung depending on their solubility in mucus, which is over
28 99% water. High-solubility compounds, such as nitric acid or hydrogen peroxide,
29 are rapidly removed in the upper airways while low-solubility compounds, such as
30 oxygen or ozone, are less well removed and so penetrate to the alveoli. Pollutant
31 deposition in the upper airways is less harmful than in the lower airways because
October 1999 3-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 upper airways clearance is more rapid and the epithelium is protected by a mucus
2 layer. As a result, low-solubility pollutants, such as ozone, may harm the alveoli,
3 while high-solubility pollutants, such as nitric acid, do not reach these tissues.
4 In the presence of aerosol particles, this scenario changes. Undermost
5 ambient conditions, aerosol particles contain some liquid water so that soluble
6 compounds are partitioned between the gas phase and the aerosol liquid-water phase.
7 The degree of deposition via the gas compared to that via the particles is a function of
8 a number of factors including the solubility of the compound and the liquid-water
9 content of the aerosol. Since highly soluble compounds deposit in the upper airways,
10 particles may provide a vector for deposition of these compounds in the lower
11 airways. Lower-solubility compounds may persist in the vapor phase and so may
12 deposit in lower airway segments.
13 As aerosols are inhaled, soluble vapors deposit on the mucus, disrupting the
14 gas-particle equilibrium, and the compound begins to evaporate from the aerosol
15 particles. If the evaporation is rapid, the pattern of deposition of the compound
16 will not be influenced by the presence of the particles, i.e., the deposition pattern is
17 essentially that of the vapor alone. If the evaporation is very slow compared to a
18 breathing cycle, a significant amount of the compound will remain in the particle
19 phase and the pattern of deposition may be shifted toward the pulmonary region by
20 the particles.
21 Wexler and Sarangapani (1998) calculated the concentrations of a water-soluble gas,
22 relative to its initial concentration, as a function of airway generation (a measure of penetration
23 into the lung) for a range of Henry's law coefficients (a measure of solubility) for 0.1, 0.3, and
24 1 //m diameter particles, assuming that there is no resistence to evaporation of the gas from the
25 particle-bound water. Wexler and Sarangapani conclude that, "Particles do not increase vapor
26 deposition in human airways". This conclusion is based largely on the fact that only a very small
27 fraction of the gas is dissolved in the particle-bound water for normal relative humidities.
28 However, their calculations do show that soluble gases are carried to higher generation airways
29 (deeper into the lung) in the presence of particles than in the absence of particles.
30 Underhill (1999) has pointed out that species such as SO2 and formaldehyde either react to
31 form different species (SO2 ^ H+ + HSO7 ) or hydrates (HCHO + H2O ^ CH2 (OH)2}. These
October 1999 3-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 reactions reduce the concentration of the dissolved gas-phase species and provide a kinetic
2 resistence to the evaporation of the dissolved gas. Thus, the concentration of the dissolved
3 species may be greater than that predicted by the equilibrium calculations of Wexler and
4 Sarangapani. Underhill further points out that Amdur (1957, 1960, 1966, 1969), Amdur and
5 Mead (1958), and Amdur and Underhill (1968), in studies of the effects of particles and gases on
6 pulmonary resistence in guinea pigs, observed synergistic effects between NaCl particles and SO2
7 and HCHO, but not between NaCl and formic acid. The reaction rates for converting the reacted
8 species back to the gas-phase compound is fast enough to maintain equilibrium conditions for
9 formic acid, but not for SO2 and HCHO.
10 It is also possible that toxic gases could be absorbed on solid particles thus be carried into
11 the lungs. The gases might be desorbed in the higher temperature and relative humidity of the
12 lung. If the toxic gas remains absorbed on the particle, it could be brought into direct contact
13 with cells in the respiratory system. The thermal release of NO from ambient air and diesel
14 particles has been studied (Ball et al., 1999) as well as the absorption of NO2 on carbon particles
15 (Kalberer et al., 1999). The partitioning of semivolatile organic compounds between the gas-
16 phase and the liquid organic layer of an atmospheric particle has been investigated and modeled
17 (e.g., Pankow et al., 1993; Pankow, 1994a,b; Jang et al., 1997). Less information is available on
18 the kinetics of the partitioning process. However, Kamens and Coe (1997) have measured the
19 rate of evaporation of polycyclic aromatic hydrocarbons (PAHs) from fresh diesel soot particles.
20 For some PAHs the evaporation rates were slow enough (order of seconds) to allow particles to
21 carry PAHs into higher generation airways. Kamens et al. (1999) also report calculations of
22 desorption rates for certain products of the O3 + cc-pinene reaction from the secondary particles
23 formed in the reaction. Muzyka et al. (1998) report that particulate matter from diesel exhaust
24 contains absorbed benzene as well as polycyclic aromatic hydrocarbons (PAH) and nitro-PAH.
25 Recent studies also suggest that a variety of allergens may be absorbed on atmospheric particles
26 and carried into the lung and contribute to aggregation of allergy and asthma (Schappi et al.,
27 1999; Ormstad et al., 1998).
28
29 3.1.3.7 Separation of Fine and Coarse Particles
30 The many reasons for wanting to collect fine and coarse particles separately and
31 considerations as to the appropriate cutpoint for separating fine and coarse particles were
October 1999 3-33 DRAFT-DO NOT QUOTE OR CITE
-------
1 discussed in Chapter 3 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996).
2 A review of atmospheric particle-size-distribution data did not provide a clear or obvious
3 rationale for selection of an appropriate cutpoint. Depending on conditions, a significant amount
4 of either fine or coarse mode material may be found in the intermodal region between 1.0 and
5 3 //m. However, the analysis of the existing data did demonstrate the important role of relative
6 humidity in influencing the size of the fine particle mode and indicated that significant fine mode
7 material is found above 1.0 //m only during periods of very high relative humidity.
8 Thus, a PM2 5 sample will contain most of the fine mode material, except during periods of
9 RH near 100 %. However, especially in conditions of low RH, it may contain 5 to 20 % of the
10 coarse mode material below 10 //m in diameter. A PMLO sample will prevent misclassification of
11 coarse mode material as fine but under high RH conditions could result in some of the fine mode
12 material being misclassified as coarse. A reduction in RH, either intentionally or inadvertently,
13 will reduce the size of the fine mode. A sufficient reduction in RH will yield a dry fine particle
14 mode with very little material above 1.0 //m. Studies of the changes in particle size with changes
15 in relative humidity suggest that only a small amount of accumulation mode particles will be
16 above 1 //m in diameter at RHs below 60% but a substantial fraction will grow above 1 //m for
17 RH above 80% (Hitzenberger et al, 1997; McMurry and Stolzenberg, 1989; U.S. Environmental
18 Protection Agency, 1996). As discussed in Section 3.2, some new techniques have been
19 developed for both integrated and continuous measurement of fine particulate matter minus
20 particle-bound water, but including semivolatile nitrate and organic compounds. These
21 techniques require reduction of RH prior to collection. With such techniques PMj 0 would be an
22 appropriate cut-point.
23
24 Summary
25 The physical and chemical properties of ultrafine mode, accumulation mode, and coarse
26 mode particles are summarized in Table 3-1.
27
28
29 3.2 MEASUREMENT OF PARTICULATE MATTER
30 The 1996 Air Quality Criteria Document for Particulate Matter (AQC PM 96) (U.S.
31 Environmental Protection Agency, 1996) summarized sampling and analytical techniques for
October 1999 3-34 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-1. COMPARISON OF AMBIENT PARTICLES
FINE MODE (Nuclei Mode Plus Accumulation Mode) AND COARSE MODE
Fine
Coarse
Nuclei
Accumulation
Formed from:
Formed by:
Combustion, high temperature
processes and atmospheric reactions
Composed of:
Solubility:
Sources:
Atmospheric
half-life:
Removal
Processes:
Nucleation
Condensation
Coagulation
Sulfates
Elemental carbon
Metal compounds
Organic compounds
with very low,
saturation vapor
pressure at ambient
temperature
Probably less
soluble than
accumulation mode
Combustion
Atmospheric
transformation of
SO2 and some
organic compounds
High temperature
processes
Minutes to hours
Grows into
accumulation mode
Travel distance:
-------
1 particulate matter (PM) and acid deposition that had appeared in the literature since the 1982 Air
2 Quality Criteria for Particulate Matter (U.S. Environmental Protection Agency, 1982). Two
3 other excellent reviews have been published in recent years by Chow (1995) and McMurry
4 (1999). This section will concentrate on problems in measuring PM, new techniques that attempt
5 to alleviate these problems or measure problem species, the current EPA monitoring program
6 (including measurements with Federal Reference Methods, speciation monitors, and continuous
7 monitors), and the importance of intercomparison studies in the absence of any reference
8 standard for suspended atmospheric particles.
9
10 3.2.1 Problems in Measuring Particulate Matter
11 The decision by the US EPA to revise the PM standards by adding daily and yearly
12 standards for PM2 5 has led to a renewed interest in the measurement of atmospheric particles and
13 also to a better understanding of the problems in developing precise and accurate measurements
14 of particles. Unfortunately, it is very difficult to measure and characterize particles suspended in
15 the atmosphere.
16 The US Federal Reference Methods (FRM) for PM25 and PM10 provide relatively precise
17 (±10 %) methods for determining the mass of material remaining on a Teflon filter after
18 equilibration. However, numerous uncertainties remain as to the relationship between the mass
19 and composition of material remaining on the filter, as measured by the FRMs, and the mass and
20 composition of material that exists in the atmosphere as suspended PM. The goal of a PM
21 indicator might be to accurately measure what exists as a particle in the atmosphere. However,
22 this is not currently possible, in part because of the difficulty of creating a reference standard for
23 particles suspended in the atmosphere. As a result, EPA defines accuracy for PM measurements
24 in terms of agreement of a candidate sampler with a reference sampler. Therefore,
25 intercomparisons of samplers become very important in determining how well various samplers
26 agree and how various design choices influence what is actually measured.
27 There are five general areas where choices must be made in designing an aerosol indicator.
28 These include treatment of semivolatile components; selection of an upper cut point; separation
29 of fine-mode and coarse-mode PM; treatment of pressure, temperature, and relative humidity;
30 and how to assess the reliability of the measurement technique. In many cases choices have been
31 made by default and with inadequate understanding of the consequences. As a result,
October 1999 3-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 measurement methods developed by different organizations may give different results when
2 sampling the same atmosphere, even though the techniques appear to be identical.
3
4 3.2.1.1 Treatment of Semivolatile Components of PM
5 Current filtration-based mass measurements lead to significant evaporative losses, during
6 and possibly after collection, of a variety of semivolatile components, i.e., species which exist in
7 the atmosphere in dynamic equilibrium between the condensed phase and gas phase. Important
8 examples include ammonium nitrate, semivolatile organic compounds, and particle-bound water.
9 This problem is illustrated in Figure 3-10.
10 Possible approaches that have been used to address the problem of potentially lost
11 semivolatile components include the following which will be discussed in more detail in
12 subsequent sections:
13 1. Collect/measure all components present in the atmosphere in the condensed phase except
14 particle-bound water. (Examples: Brigham Young absorptive sampler, Harvard pressure
15 drop monitor. Both require pre-concentration of the accumulation mode and reduction of
16 ambient humidity.)
17 2. Stabilize PM at a specified temperature high enough to remove all particle-bound water.
18 This results in loss of most of the semivolatile PM. (Examples: TEOM operated at 50°C,
19 beta gauge with heated inlet.)
20 3. Equilibrate collected material at fixed, near-room temperature and low relative humidity to
21 remove particle-bound water. Accept loss of an unknown but possibly significant fraction of
22 semivolatile PM. (Example: US Federal Reference Method and most filter-weighing
23 techniques. There is also information to suggest that not all particle-bound water is removed
24 by the equilibration process.)
25 The semivolatile artifact is composition dependent and has been shown to be significant in
26 air sheds with high nitrate, wood smoke or secondary organic aerosols.
27
28 3.2.1.2 Upper Cut Point
29 A technique must be used that gives an upper cut-point, and its standard deviation, that is
30 independent of wind speed and direction (the classical high volume sampler head was
31 unsatisfactory because of radial asymmetry).
October 1999 3-37 DRAFT-DO NOT QUOTE OR CITE
-------
Should be
retained
Particle-bound water
should be removed
elemental
Mineral/Metal
1.0
Aerodynamic Diameter, |jm
2.5
Semivolatile components subject to evaporation during or after sampling
Figure 3-10. Schematic showing major nonvolatile and semivolatile components of PM 2_5.
Semivolatile components are subject to partial to complete loss during
equilibration or heating. The optimal technique was remove all
particle-bound water but no ammonium nitrate or semivolatile organic PM.
1 A separation which simulates the removal of particles by the upper part of the human
2 respiratory system would appear to be a good choice, i.e., measure what gets into the lungs. The
3 ACGIH-ISO-CEN penetration curve for thoracic particles, with a 50% cut-point at 10 //m
October 1999
3-38
DRAFT-DO NOT QUOTE OR CITE
-------
1 aerodynamic diameter (AD), would be an appropriate choice. (Thoracic particles are able to pass
2 the larynx and penetrate into the bronchial and alveolar regions of the lung.) Some countries,
3 however, use PM10 to refer not to samplers with a 50% cut at 10 //m AD but samplers with 100%
4 rejection of all particles greater than 10 //m AD. Such samplers miss too much of the thoracic
5 PM. The US PM10 separation curve, while sharper than the thoracic curve, is probably
6 satisfactory both for regulatory and health risk monitoring. It has the advantage of reducing the
7 problem of maintaining the finite collection efficiency specified by the thoracic penetration curve
8 for particles larger than 10 //m AD. (See Figure 3-6 and Section 3.1.2.2.)
9
10 3.2.1.3 Separation of Fine-Mode and Coarse-Mode PM
11 Fine-mode and coarse-mode particles differ not only in size and morphology (e.g., smooth
12 droplets vs rough solid particles) but also in formation mechanisms; sources; and chemical,
13 physical, and biological properties. They also differ in terms of dosimetry (deposition in the
14 respiratory system), toxicity, and health effects as observed by epidemiologic studies.
15 At high relative humidity, such as that found in fog and clouds, hygroscopic fine-mode
16 particles will increase in size due to accretion of particle-bound water. Some, originally
17 sub-micrometer, fine-mode PM may be found with an AD above 1 //m. At very low relative
18 humidity, coarse-mode particles may be fragmented into smaller sizes and small amounts of
19 coarse-mode PM may be found with an AD below 1 //m. It is desirable to separate fine-mode
20 PM and coarse-mode PM as cleanly as possible in order to properly allocate health effects to
21 either fine-mode PM or coarse-mode PM and to correctly determine sources by factor analysis
22 and/or chemical mass balance. For example sulfate in the fine-mode is associated with hydrogen
23 and/or ammonium ions; sulfate in the coarse mode is associated with basic metal ions. The
24 sources are different and the health effects are likely to be different. Transition metals in the
25 coarse mode are likely to be associated with soil and tend to be less soluble than transition metals
26 in the fine mode which may be found in fresh combustion particles.
27 The current practice of separating fine-mode and coarse-mode particles at 2.5 //m AD,
28 while satisfactory for a health-based standard, does not provide an adequate separation for
29 epidemiologic studies, especially in areas where winds cause high concentrations of wind blown
30 soil, or for the determination of source categories to guide control strategy. A possible approach,
31 which would provide much better separation of fine-mode PM and coarse-mode PM, would be to
October 1999 3-39 DRAFT-DO NOT QUOTE OR CITE
-------
1 dehumidify the air stream to some fixed humidity that would remove all or most particle-bound
2 water without evaporating semivolatile components and make the cut near 1 //m AD. (See
3 Section 3.1.3.7.)
4
5 3.2.1.4 Treatment of Pressure, Temperature, and Relative Humidity
6 There are a variety of techniques for defining (or ignoring) the pressure, temperature, and
7 relative humidity during or after sampling.
8 Temperature and Pressure:
9 a. Sample volume based on mass or volumetric flow corrected to standard temperature and
10 pressure (273 °K and 1 atm.) (former EPA technique for PM10).
11 b. Sample volume based on volumetric flow at ambient conditions of temperature and
12 pressure (current EPA technique for PM2 5 and PM10).
13 Temperature During Collection:
14 a. Heat enough to remove all particle-bound water (i.e., TEOM at 50 °C).
15 b. Heat several degrees to prevent condensation of water in sampling system.
16 c. Try to maintain sampler near ambient temperature.
17 d. Maintain sampler at constant temperature inside heated/air conditioned shelter.
18 Temperature After Collection:
19 a. No control
20 b. Constant Temperature (room temperature)
21 c. Store at cool temperature (4 °C)
22 Relative Humidity: Changes in relative humidity cause changes in particle size of hygroscopic or
23 deliquescent particles. Changing relative humidity by adding or removing water vapor affects
24 measurements of:
25 a. Particle number, particle surface area and particle size distribution
26 b. Amount of overlap of fine-mode and coarse-mode particles
27 Changing relative humidity by intentional or inadvertent changes in temperature affects above
28 measurements plus:
29 c. Amount of loss of ammonium nitrate and semivolatile organic compounds.
30 Studies of relationships between personal/indoor/outdoor measurements present special
31 problems. Indoor environments are typically dryer than outdoors and may be warmer or, if
October 1999 3-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 air-conditioned, cooler. These differences may change particle size and the amount of
2 volatilization of semivolatile components. Such changes between indoors and outdoors will
3 complicate the comparison of indoor to outdoor concentrations, the modeling of personal
4 exposure to all particles, and the disaggregation of personal exposure into exposure to particles
5 of ambient origin and exposure to particles of indoor origin.
6
7 3.2.1.5 No Way to Determine Accuracy
8 Precision and accuracy are normally used to describe the quality of a measurement.
9 Precision is typically determined by comparison of collocated samplers or through replicate
10 analyses, while accuracy is determined through the use of traceable calibration standards.
11 Unfortunately, no standard reference material for suspended PM exists. Therefore, it is not
12 possible to establish the accuracy of a PM monitoring technique. Intercomparison studies, to
13 establish the precision of identical monitors and the extent of agreement between different types
14 of monitors are essential for establishing the reliability of PM measurements. Intercomparison
15 studies have contributed greatly to our understanding of the problems in PM measurement. Such
16 studies will be discussed as they apply to specific measurement problems, monitoring
17 instruments, or analytical techniques. Measurement errors of concern in PM10 sampling that arise
18 due to uncertainty tolerances in cutpoint; particle bounce and re-entrainment; impactor surface
19 overloading; and losses to sampler internal surfaces were discussed in detail in the AQC PM 96
20 (U.S. Environmental Protection Agency, 1996). Measurement errors of concern in PM25
21 sampling arise because of our inability to assess accuracy in an absolute sense due to a lack of
22 primary calibration standards, because of the use of an operational definition of PM25 as a
23 surrogate for fine particles, and because of problems associated with trying to measure the mass
24 of particles as they exist in the air rather than what remains after collection on a filter.
25 Because of the difficulties associated with determining the accuracy of PM measurements,
26 the U.S. Environmental Protection Agency has sought to make FRM measurements equivalent by
27 specifying operating conditions and, in the case of PM25 samplers, by specifying details of the
28 sampler design. Thus, both the PM10 as well as PM2 5 standards are defined with consistency of
29 measurement technique, rather than accuracy of the true mass concentration measurement, in
30 mind (McMurry, 1999). It is acknowledged in the Federal Register (1997) that "because the size
31 and volatility of the particles making up ambient particulate matter vary over a wide range and
October 1999 3-41 DRAFT-DO NOT QUOTE OR CITE
-------
1 the mass concentration of particles varies with particle size, it is difficult to define the accuracy
2 of PM25 measurements in an absolute sense..." Thus, accuracy is defined as the degree of
3 agreement between a subject field PM2 5 sampler and a collocated PM2 5 reference method audit
4 sampler (McMurry, 1999). The Federal Reference Method (FRM) for PM2 5 will be discussed in
5 Section 3.2.3.3. The accurate measurement of the mass of PM suspended in the atmosphere was
6 of concern in PM10 sampling and is of even greater concern for PM25. As mentioned earlier,
7 volatilization losses, during sampling or post-sampling handling, of some organics as well as
8 ammonium nitrate can lead to significant underestimation of the true fine particulate mass
9 concentration in some locations. Sources of error in the measurement of true PM25 mass also
10 arise due to adsorption of organic vapors onto collected PM, filter media, or other sampler
11 surfaces; neutralization of acid or basic vapors on either filter media or collected PM; and the
12 role of particle-bound water in PM sampling.
13 The lack of a standard reference material or a primary calibration standard for PM
14 suspended in the atmosphere has prevented any evaluation of accuracy. In the last 25 years, there
15 have been advancements in the generation of monodisperse aerosols and classification, as well as
16 in the development of electron microscopy and imaging analysis, that have contributed to the
17 advancement in aerosol calibration (Chen, 1993). Still, one of the limitations in PM sampling
18 and analysis remains the lack of primary calibration standards for evaluating analytical methods
19 and for intercomparing laboratories. Klouda et al. (1996) examined the possibility of
20 resuspending the NIST Standard Reference Material 1649 (Urban Dust) in air for collection on
21 up to 320 filters simultaneously, using SRI, International's dust generation and collection system,
22 however little additional work in this area has been reported.
23 Methods validation was discussed in the previous AQC PM 96 (U.S. Environmental
24 Protection Agency, 1996), and the usefulness of intercomparisons and "internal redundancy" was
25 emphasized. For example, a number of internal consistency checks are applied to the IMPROVE
26 network (Malm et al., 1994). These include mass balances, sulfur measurements by both proton
27 induced x-ray emission (PIXE) and ion chromatography (1C); and comparison of organic matter
28 by combustion and by proton elastic scattering analysis (PESA) analysis of hydrogen. Mass
29 balances compare the gravimetrically determined mass with the mass calculated from the sum of
30 the major chemical components, i.e. crustal elements plus associated oxygen, organic carbon,
31 elemental carbon, sulfate, nitrate, ammonium, and hydrogen ions. Mass balances are useful
October 1999 3-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 validation techniques, however, they do not check for, or account for, artifacts associated with
2 the absorption of gases during sampling, or the loss of semi-volatile material during sampling.
3 The mass balance check may appear reasonable even if such artifacts are present, since only the
4 material collected on the filter is included in the balance.
5
6 3.2.2 Why Measure Particles
7 3.2.2.1 Attainment of a Standard
8 A critical need for particle measurements is to determine if a location is in compliance with
9 an existing standard and to determine if trends show improvements in air quality. For this
10 purpose, precision of the measurement by the variety of indicators in use is the most important
11 consideration. Therefore, intercomparisons of various potential indicators, under a variety of
12 atmospheric and air quality conditions are essential. PM standards are currently based on
13 24-hour measurements, with measurements every sixth day being used to estimate an annual
14 average.
15
16 3.2.2.2 Implementation of a Standard
17 In order to develop State Implementation Plans, to reduce pollution in order to attain a
18 standard, local agencies and national research organizations need measurements to identify
19 source categories and to develop and validate air quality models. For these purposes PM
20 parameters other than mass, such as chemical composition and size distribution, must be
21 measured. Also measurements are needed with shorter time resolution in order to match changes
22 in pollution associated with diurnal changes in the boundary layer.
23
24 3.2.2.3 Determination of Health Effects
25 PM measurements are needed to determine exposure for use in epidemiological studies, to
26 assess exposure for risk assessment and to determine components of PM to guide planning and
27 interpretation of toxicologic experiments. For these purposes size and chemical composition
28 may be needed. For exposure assessment, different measurement time intervals may be needed.
29 For acute epidemiology, one-hour or continuous measurements may be needed as well as 24-hour
30 measurements. For chronic epidemiologic studies, measurements which integrate over a week to
31 a month may be more cost effective. For dosimetric studies and modeling, information will be
October 1999 3-43 DRAFT-DO NOT QUOTE OR CITE
-------
1 needed on the particle size distribution and on the behavior of particles as the relative humidity
2 and temperature are increased to those in the respiratory system.
3
4 3.2.2.4 Determination of Ecological Effects
5 Measurement of particles, and of the chemical components of particulate matter in rain, fog
6 and dew, are needed to understand the contributions of PM to soiling of surfaces and damage to
7 materials and to understand the wet and dry deposition of acidity and toxic substances to surface
8 water, soil and plants. Some differentiation into particle size is needed to determine dry
9 deposition. Information on chemical composition is also needed to understand materials damage
10 and ecological damage.
11
12 3.2.2.5 Determination of Radiative Effects
13 Particles reduce visibility by scattering and absorbing light. They also have a direct effect
14 on the climate by scattering visible and ultraviolet light back into space and indirectly, as cloud
15 condensation nuclei, by changing the albedo and stability of clouds. For understanding these
16 effects information is needed on refractive index (including ratio of scattering to absorption), size
17 distribution, and change in particle size with change in relative humidity.
18
19 3.2.2.6 PM Components/Parameters Which Need To Be Measured
20 The large variety of components of PM or PM parameters that need to be measured for
21 various purposes are summarized in Table 3-2.
22
23 3.2.3 Problems Associated with Semivolatile Particulate Matter
24 It is becoming increasingly apparent that the semivolatile component of PM may
25 significantly impact the quality of the measurement, and can lead to both positive and negative
26 sampling artifacts. Losses of semivolatile species, like ammonium nitrate and many organic
27 species, may occur during sampling, due to changes in temperature, relative humidity, or
28 composition of the aerosol, or due to pressure drop across the filter (McMurry, 1999).
29 Semivolatile species may adsorb onto, or react with, filter media and/or collected PM, leading to
30 a positive sampling artifact. Tsai and Huang (1995) observed positive sulfate and nitrate artifacts
31 on high volume PM10 quartz filters and attributed the artifacts to interactions between acidic
October 1999 3-44 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-2. PM COMPONENTS/PARAMETERS OF INTEREST FOR HEALTH,
ECOLOGICAL, OR RADIATIVE EFFECTS; FOR SOURCE CATEGORY
APPORTIONMENT STUDIES; OR FOR AIR QUALITY VALIDATION STUDIES
Particle number
Particle surface area
Particle size distribution
PM mass (fine-mode {PM: „} and coarse-mode {PM 1(M} mass as well as PM2 5 and PM10; nonvolatile mass,
Federal Reference mass, and mass including semivolatile components such as ammonium nitrate and semivolatile
organic compounds but not particle-bound water)
Ions (sulfate, nitrate and ammonium)
Strong acidity (H+)
Elemental carbon
Organic carbon (total, nonvolatile and semivolatile; functional groups and individual species)
Transition metals (water soluble, bioavailable, oxidant generation)
Specific toxic elements
Crustal elements
Bioaerosols
Particle refractive index (real and imaginary)
Particle density
Particle size change with changes in relative humidity
1 gases SO2, HONO and HNO3 and both the filter media (either glass fiber or quartz) and the
2 coarse particles collected on the filter. Volatilization losses have also been reported to occur
3 during sample transport and storage (Chow, 1995). Evaporative losses of particulate nitrates
4 have been investigated in laboratory and field experiments (e.g., Wang and John, 1988), and in
5 theoretical studies (Zhang and McMurry, 1992). It has been known for some time that
6 volatilization losses of SVOC can be significant (e.g., Eatough et al., 1993).
7 The theory describing phase equilibria of SVOC continues to be developed. Liang et al.
8 (1997), Jang et al. (1997), and Strommen and Kamens (1997) modeled the gas/particle
9 partitioning of SVOC on inorganic, organic, and ambient smog aerosols.
October 1999 3-45 DRAFT-DO NOT QUOTE OR CITE
-------
1 Adsorption of organic vapors onto quartz filters has also been recognized as a source of
2 positive sampling error. Although original experiments investigating this sampling artifact were
3 typically carried out utilizing two quartz fiber filters deployed in series, the second quartz filter
4 can indicate both gaseous VOC adsorbed on both filters (positive) artifact and SVOC evaporated
5 from particles on the first filter and subsequently adsorbed on the second filter (negative artifact).
6 Unless the individual compounds are identified, the investigator does not know what to do with
7 the loading value on the second filter (i.e., to add or subtract from the first filter loading value).
8 The developing state of the art in which diffusion denuder technology is being applied to SVOC
9 sampling (e.g., Eatough et al., 1993; Gundel et al., 1995a), as well as for sampling of gas and
10 particulate phase organic acids (Lawrence and Koutrakis, 1996a,b), holds promise for improving
11 our understanding of SVOC sampling artifacts.
12 Finally, Eatough et al. (1999) have reported on a batch sampler (the Particle Concentrator -
13 Brigham Young University Organic Sampling System, or PC-BOSS) and a continuous sampler
14 (Real-Time Air Monitoring System or RAMS) which attempt to correct simultaneously for
15 volatilization losses of both nitrate and SVOC. These samplers will be discussed in more detail
16 in Section 3.2.3.2.
17
18 3.2.3.1 Particulate Nitrates
19 It is well known that volatilization losses of particulate nitrates (e.g., Zhang and McMurry
20 [1992]; see also Hering and Cass [1999], and references therein) will occur during sampling on
21 Teflon filters. The impact on the accuracy of atmospheric particulate measurements from these
22 volatilization losses will be even more significant for PM2 5 than they are for PM10. The FRM for
23 PM25 will suffer loss of nitrates, similar to the losses experienced with other simple filter
24 collection systems. Sampling artifacts due to the loss of particulate nitrates will represent a
25 significant problem in areas that experience high amounts of nitrogen species, like southern
26 California.
27 Hering and Cass (1999) examined the errors in PM25 mass measurements due to
28 volatilization of particulate nitrate by looking at data from two field measurement campaigns
29 conducted in southern California - the Southern California Air Quality Study (SCAQS, Lawson,
30 1990), and the 1986 CalTech study (Solomon et al., 1992). In both these studies, side-by-side
31 sampling of PM2 5 was conducted. One sampler collected particles directly onto a Teflon filter.
October 1999 3-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 The second sampler collected particles onto a nylon filter, following a denuder to remove
2 gaseous nitric acid. In both studies, the denuder consisted of MgO coated glass tubes (Appel
3 et al., 1981). Fine particulate nitrate collected on the Teflon filter was compared to fine
4 particulate nitrate collected on the denuded nylon filter. In both studies, the PM2 5 mass lost due
5 to volatilization of ammonium nitrate represented a significant fraction of the total PM25 mass,
6 and the fraction lost was higher during summer than during fall (17% versus 9% during the
7 SCAQS study, and 21% versus 13% during the CalTech study) (Figure 3-11). Hering and Cass
8 (1999) found that nitrate obtained from Teflon filter samples was on average 28% lower than that
9 obtained by denuded nylon filters.
80%
o
60% -
40% -
20% -
0%
SCAQS Data Set
o Summer Measurements
• Fall Measurements
80%
50 100 150 200
PM25 Gravimetric Mass (ng/m3)
250
.i 60% -
40% -
20% -
100
Caltech Data Set
o April - September
• October - March
150
200
250
PM25 Gravimetric Mass (ng/m
Figure 3-11. Amount of ammonium nitrate volatilized from Teflon filters, expressed as a
percentage of the measured PM25 mass, for the SCAQS and CalTech studies,
for spring and fall sampling periods.
Source: Herring and Cass (1999).
1 Hering and Cass (1999) also analyzed these data by extending the evaporative model
2 developed by Zhang and McMurry (1987). The extended model utilized by Hering and Cass
3 (1999) takes into account dissociation of collected particulate ammonium nitrate on Teflon filters
4 into nitric acid and ammonia, via three mechanisms: scrubbing of nitric acid and ammonia in the
October 1999
3-47
DRAFT-DO NOT QUOTE OR CITE
-------
1 sampler inlet (John et al. [1988] showed that clean PM10 inlet surfaces serve as an effective
2 denuder for nitric acid), heating of the filter substrate above ambient temperature by sampling,
3 and pressure drop across the Teflon filter. For the sampling systems modeled, the flow-induced
4 pressure drop was measured to be less than 0.02 atmospheres, and the corresponding change in
5 vapor pressure was 2%, so losses driven by pressure drop were not considered to be significant in
6 this work. Losses from Teflon filters were found to be higher during the summer compared to
7 the winter, and higher during the day compared to night, and were reasonably consistent with
8 modeled predictions.
9 Finally, during the SCAQS study, particulate samples were also collected using a Berner
10 impactor and greased Tedlar substrates, in size ranges from 0.05 to 10 mm in aerodynamic
11 diameter. The Berner impactor PM2 5 nitrate values were much closer to those from the denuded
12 nylon filter than those from the Teflon filter, with the impactor nitrate being approximately 2%
13 lower than the nylon filter nitrate for the fall measurements, and approximately 7% lower during
14 the summer measurements. When the impactor collection was compared to the Teflon filter
15 collection for a nonvolatile species (sulfate), the results were in agreement.
16 It should be noted that during these intercomparison studies, filters or collection surfaces
17 were removed immediately after sampling and placed into vials containing a basic extraction
18 solution. Therefore, losses that might occur during handling, storage, and equilibration of filters
19 or impaction surfaces were avoided. The loss of nitrate observed from Teflon filters and
20 impaction surfaces in this study, therefore, is a lower limit compared to losses that might occur
21 during the normal processes involved in equilibration and weighing of filters and impaction
22 surfaces.
23 In atmospheres with high sulfate and low ammonia, the PM tends to be acidic (NH4HSO4
24 or H2SO4) and nitric acid remains as a gas. In atmospheres with lower sulfate and higher
25 ammonia, there may be sufficient ammonia to fully neutralize the H2SO4 and also react with
26 FfNO3 to form NH4NO3 particles. In the U.S., therefore, loss of nitrate will be a bigger problem
27 in the western U.S. than in the eastern U.S. However, as SO2 emissions are reduced in the
28 eastern U.S., nitrate may become a larger fraction of the suspended PM.
29
October 1999 3-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.2.3.2 Semivolatile Organic Compounds
2 Although there is less information on losses of semivolatile organic compounds (SVOC), it
3 is known that these species can similarly be lost from Teflon filters due to volatilization, and can
4 also cause the PM2 5 mass to be significantly underestimated. Like particulate nitrates, the FRM
5 for PM2 5 will suffer loss of SVOC, similar to the losses experienced with other simple filter
6 collection systems. It has been shown that attempting to correct for the losses of SVOC during
7 sampling by deploying a second quartz filter directly behind either a quartz or Teflon filter can
8 significantly underestimate the volatilization losses (e.g., Eatough et al., 1993). Using their
9 multichannel diffusion denuder sampling system (BOSS), Eatough et al. (1995) reported that, for
10 samples collected at the South Coast Air Quality Management District sampling site at Azusa,
11 CA, changes in the phase distribution of SVOC could result in a loss on average of 35% of the
12 particulate organic material; the fraction of the total fine particulate matter that this loss
13 represents was not given. At present, there are limited data available specifically on the fraction
14 of PM25 mass lost during sampling onto Teflon filters due to volatilization of organic species,
15 and even less on the regional differences in the effects of volatilization losses of SVOC. Cui
16 et al. (1998) found that losses of SVOC from particles in the Los Angeles Basin during the
17 summer were greater during the day than at night. Cui et al. (1998) determined that on average,
18 42 and 62% of the particulate organic material was semivolatile organic compounds lost from
19 particles during sampling for daytime and nighttime samples, respectively.
20 In addition to their contribution to suspended PM mass, SVOC are also of interest because
21 of their possible health effects. SVOC include products of incomplete combustion such as, for
22 example, polycyclic aromatic hydrocarbons (PAHs) and polycyclic organic matter, which has
23 been identified as a hazardous air pollutant. PAHs have also been suggested as alternative
24 particulate tracers for automobile emissions, since the phase-out of organo-lead additives to
25 gasoline means that lead is no longer a good tracer for automobiles (Venkataraman et al., 1994).
26 PAHs are also emitted during biomass burning, including burning of cereal crop residues and
27 wood fuels (Jenkins et al., 1996; Roberts and Corkill, 1998). The semivolatile PAHs are also of
28 interest in ambient PM studies, because of their potential for causing both positive and negative
29 sampling artifacts if not properly accounted for. Several investigators have observed that
30 collection of particles on a filter can result in underestimation of particulate organic compounds
31 due to losses of semivolatile organic material during sample collection (negative sampling
October 1999 3-49 DRAFT-DO NOT QUOTE OR CITE
-------
1 artifact) (Eatough et al, 1993; Tang et al, 1994; Eatough et al, 1995; Gundel et al, 1995a; Finn
2 et al., 1999). Positive sampling artifacts can also occur due to the adsorption of gases onto the
3 filter materials (e.g., Gundel et al., 1995a). There appears to be a larger positive artifact due to
4 adsorption of organic vapor onto quartz fiber filters than to Teflon filters (Turpin et al., 1994;
5 Chow et al., 1994, 1996; Eatough et al., 1996; Finn et al., 1999).
6 Phase distribution of semivolatile organic species has been the subject of several studies
7 that have employed denuder technology (see Gundel et al., 1995a; Gundel and Lane, 1999) to
8 directly determine the phase distributions while avoiding some of the positive and negative
9 sampling artifacts associated with using backup quartz filters. For measuring particulate phase
10 organic compounds, the denuder-based sampling system is a definite improvement over the
11 filter/adsorbent collection method (Turpin et al., 1993). Some researchers have reported that
12 denuder coatings themselves can introduce contamination (Mukerjee et al., 1997), or the
13 adsorbed species may be difficult to remove from the coating (Eatough et al., 1993).
14 In calculating the overall phase distributions of SVOC PAH from a denuder system, the
15 collection efficiency for each compound is needed. The efficiency of silicone-grease-coated
16 denuders for the collection of polynuclear aromatic hydrocarbons was examined by Coutant et al.
17 (1992), who examined the effects of uncertainties in the diffusion coefficients, and in the
18 collisional reaction efficiencies, on the overall phase distributions of SVOC PAH calculated
19 using denuder technology. In their study, they used a single stage, silicone-grease-coated
20 aluminum annular denuder, with a filter holder mounted ahead of the denuder, and an XAD trap
21 deployed downstream of the denuder. In a series of laboratory experiments, they spiked the filter
22 with a mixture of perdeuterated PAH, then swept the system with ultra-high purity air for several
23 hours, and then analyzed the filter and the XAD. They found that the effects of these
24 uncertainties, introduced by using a single compound as a surrogate PAH (in their case,
25 naphthalene) for validation of the denuder collection efficiency, are less significant than normal
26 variations due to sampling and analytical effects. Results on field studies using their sampling
27 system have not been published.
28 Losses of the SVOC fraction of particulate organic matter occurring during sampling were
29 investigated by Eatough et al. (1995), who found that on average losses of 35% of the POM in
30 samples collected at a site in southern California resulted. In this study, the Brigham Young
31 University Organic Sampling System (BOSS) (Eatough et al., 1993) was used for determining
October 1999 3-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 POM composition, and a high-volume version (BIG BOSS) (flowrate 200 Lpm) was utilized for
2 determining the particulate size distribution and the chemical composition of SVOC in fine
3 particles. The BOSS, a multi-channel diffusion denuder sampling system, consists of two
4 separate samplers (each operating at 35 Lpm). The first sampler consists of a multi-parallel plate
5 diffusion denuder with charcoal-impregnated filter papers as the collection surfaces, followed by
6 a two-stage quartz filter pack, followed by a two-stage charcoal-impregnated filter pack. The
7 second sampler operating in parallel with the first consists of a two-stage quartz filter pack,
8 followed by the parallel plate denuder, followed by the two-stage charcoal-impregnated filter
9 pack. The filter samples collected by the BOSS sampler were analyzed by temperature-
10 programmed volatilization analysis. Eatough et al. (1995) also operated a two-stage quartz filter
11 pack alongside the BOSS sampler. The BIG BOSS system (Tang et al., 1994) consists of
12 4 systems (each with a flowrate of 200 Lpm). Particle size cuts of 2.5, 0.8, and 0.4 mm are
13 achieved by virtual impaction, and the sample subsequently flows through a denuder, then is
14 split, with the major flow (150 Lpm) flowing through a quartz filter followed by an XAD-II bed.
15 The minor flow is sampled through a quartz filter backed by a charcoal-impregnated filter paper.
16 The samples derived from the major flow (quartz filters and XAD-II traps) were extracted with
17 organic solvents and analyzed by gas chromatography and GC-mass spectroscopy. The organic
18 material lost from the particles was found to represent all classes of organic compounds.
19 Eatough et al. (1996) operated the BOSS sampler for a year at the IMPROVE site at
20 Canyonlands National Park, Utah, alongside the IMPROVE monitor and alongside a separate
21 sampler consisting of a two-stage quartz filter pack. They found that concentrations of
22 particulate carbon determined from the quartz filter pack sampling system were low on average
23 by 39% due to volatilization losses of SVOC from the quartz filters.
24 Six-hour daytime samples and 9-hour nighttime samples were collected with the BIG
25 BOSS sampler in the LA Basin in September of 1994 (Cui et al., 1998). During this study, Cui
26 et al. (1998) determined that an average of 42 and 62% of organic PM was SVOC lost from the
27 particles during daytime and nighttime sampling, respectively. The negative sampling artifact
28 associated with SVOC losses was an order of magnitude larger than the positive quartz filter
29 artifact that results due to adsorption of gas phase organic material.
30 The BIG BOSS sampler developed by Eatough and colleagues (Eatough et al., 1999) has
31 been used to determine the total carbonaceous material collected by the quartz filters and the
October 1999 3-51 DRAFT-DO NOT QUOTE OR CITE
-------
1 charcoal impregnated filter (GIF) by temperature programmed volatilization (TPV) and for GC
2 analyses of extractable material from the XAD-II sorbent beds. In TPV, a portion of the sample
3 is continuously heated from ambient temperature to 800 C in a mixed nitrogen/oxygen air stream,
4 and the volatilized gases are converted to CO2 in a catalytic furnace and detected as CO2. The
5 TPV analysis gives a measure of the EC and OC. To thermally desorb the GIF, a pure nitrogen
6 stream is used as the carrier gas, to avoid oxidative decomposition of the GIF substrate, as well as
7 the collected organic material. To determine organic compounds collected on the XAD-II
8 substrate, the sorbent is solvent extracted with dichloromethane and analyzed via GC and
9 GC-MS. In this way, aromatic compounds, paraffins, organic acids, and esters have been
10 detected (quantities were not reported) in samples collected at Azusa (Los Angeles Basin, CA)
11 and in Philadelphia (Eatough et al., 1995).
12 Ding et al. (1998a) developed a method for the determination of total n-nitroso compounds
13 in air samples, and used the method to examine organic compounds formed from NOX chemistry
14 in Provo, UT (Ding et al., 1998b). In their method, n-nitroso compounds are selectively
15 decomposed to yield nitric oxide, which is then detected using chemiluminescence. From the
16 samples from Provo, UT, they found that the majority of the n-nitroso and nitrite organic
17 compounds that were present in fine particulate matter were semivolatile organic compounds that
18 could be evaporated from the particles during sampling. They found particulate n-nitroso
19 compound concentrations ranging between <1 and 3 nmoles/m3, and gas-phase n-nitroso
20 compound concentrations in the same range. Particulate organic nitrite concentrations were
21 found in the range of <1 to ~5 nmoles/m3, and gas-phase concentrations as high as 10 nmoles/m3
22 were found.
23 Turpin et al. (1993) developed a sampling system based upon a diffusion separator, which
24 corrects for the loss of semivolatile organic compounds during sampling by removal of most of
25 the gas phase material from the particles in a diffusion separator sampling system. Unlike the
26 previously mentioned systems, wherein the particulate phase is measured directly, in the system
27 of Turpin et al., the gas-phase is measured directly. In their laminar flow system, ambient,
28 particle-laden air enters the sampler as an annular flow. Clean, particle-free air is pushed through
29 the core inlet of the separator. The clean air and ambient aerosol join downstream of the core
30 inlet section, and flow parallel to each other through the diffusion zone. Because of the much
31 higher diffusivities for gases compared to particles, the SVOC in the ambient air diffuse to the
October 1999 3-52 DRAFT-DO NOT QUOTE OR CITE
-------
1 clean, core flow. The aerosol exits the separator in the annular flow, and the core flow exiting
2 the separator now contains a known fraction of the ambient SVOC. Downstream of the diffusion
3 separator, the core exit flow goes into a PUF plug, where the SVOC is collected. The adsorbed
4 gas phase on the PUF plug is extracted with supercritical fluid CO2, and analyzed by gas
5 chromatography/mass-selective detection (GC/MSD). The gas-phase SVOC is thus determined.
6 Ultimately, to determine particulate phase SVOC concentrations, the total compound
7 concentration will also be measured, and the particulate phase obtained by difference. The
8 system has been evaluated for the collection of PAH.
9 Gundel et al. (1995a) recently developed a technique for the direct determination of phase
10 distributions of semivolatile polcyclic aromatic hydrocarbons, rather than phase distribution
11 determination using difference method, using annular denuder technology. That technique has
12 been applied to characterizing PAHs in environmental tobacco smoke (Gundel et al., 1995b), and
13 to ambient air sampling (Lane and Gundel, 1996). The method developed by Gundel et al.
14 (1995a), called the integrated organic vapor/particle sampler (IOVPS), uses a cyclone inlet with a
15 D50 cutpoint of 2.5 mm. The airstream then goes through two or three sandblasted glass annular
16 denuders that are coated with ground adsorbent resin material (XAD-4 was initially examined)
17 that traps vapor-phase organics. The airstream subsequently passes through a filter, followed by
18 a backup denuder.
19 The IOVPS, which operates at 10 Lpm, was tested for sampling semivolatile PAH in
20 laboratory indoor air, and environmental tobacco smoke (ETS). After exposure, the denuders,
21 filters, and sorbent traps were extracted with cyclohexane (Gundel et al., 1995a) and analyzed for
22 PAHs from naphthalene to chrysene using dual-fluorescence detection (Mahanama et al., 1994).
23 Recoveries from both denuders and filters were approximately 70% for 30 samples. Detection
24 limits (lower limits of detection, defined as 3 times the standard deviation of the blanks) for gas
25 phase SVOC PAHs ranged from 0.06 ng for anthracene to 19 ng for 2-methylnaphthalene. The
26 95% confidence interval for reproduction of an internal standard concentration was 6.5% of the
27 mean value. Relative precision as determined either from a propagation of errors analysis, or
28 from the 95% confidence interval from replicate analyses of standard reference material SRM
29 1649 (urban dust/organics) was 12% on average, and ranged from 8% for naphthalene to 22% for
30 fluorene. Sources of error included sampling flow rate, internal standard concentration, and
31 co-eluting peaks.
October 1999 3-53 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gas-phase PAH concentrations in indoor laboratory air ranged from 0.4 (0.1 - 0.6) ng/m3
2 for anthracene to 338 (162 - 545) ng/m3 for naphthalene. In ETS, gas-phase PAHs ranged from
3 0.4 (0.2 -1.1) ng/m3 for benz(a)anthracene to 1099 (784 - 1690) ng/m3 for naphthalene. In ETS,
4 the fraction of PAHs in the particle phase was less than 0.11 for most, except for fluoranthene
5 (38% in particle phase); pyrene (18%); benz(a)anthracene (99%); and chrysene (97%).
6 Particulate phase PAHs have been measured by Gundel and Lane (1999) using sorbent-
7 coated diffusion denuders (IOVPS). A series of 6 IOVPS were used sequentially over a 24-hr
8 period to measure PAH phase distributions near a roadway in Berkeley, CA. Particulate fractions
9 of PAH varied from 7 to 91% and were higher during colder periods. Gundel and Lane also
10 reported that roughly two third of particulate PAH fluoranthene, pyrene, benz [a] anthracene and
11 chrysene were found on the post-filter denuders, so that it is likely that considerable desorption
12 from the collected particles took place. The IOVPS has also been used by Kamens and
13 associates to study semivolatile PAH and nitro-PAH, and dioxins, primarily in smog chamber
14 studies. Fan et al. (1996a) found that particulate phase nitro-PAH decayed quickly in sunlight,
15 and that degradation by ozone was important at night (Fan et al., 1996b). In a study of the
16 atmospheric behavior of dioxins, Penisse and Kamens (1996) found that, under high particle
17 loadings (TSP concentrations ranging from 1 to 7 mg/m"3), the tetra and pentachlorinated dioxins
18 and furans partitioned into the gas phase.
19 Solid adsorbent-based denuder systems have been investigated by other researchers, as
20 well. Bertoni et al. (1984) described the development of a charcoal-based denuder system, for
21 the collection of organic vapors. Risse et al. (1996) developed a diffusion denuder system to
22 sample aromatic hydrocarbons. In their system, denuder tubes with charcoal coating and
23 charcoal paper precede a filter pack for particulate collection, and an adsorption tube to capture
24 particle blow-off from the filter sample. Breakthrough curves for benzene, toluene, ortho-xylene,
25 and meta-xylene were developed for denuder tubes of length 60, 90, and 120 cm. The effects of
26 relative humidity on the adsorption capacities of the denuder system were examined, and it was
27 found that the capacity of the charcoal was not significantly impacted by increases in relative
28 humidity. The feasibility of outdoor air sampling with the system was demonstrated. Risse et al.
29 (1996) developed a diffusion denuder system for sampling aromatic hydrocarbons, in which
30 denuder tubes were coated with charcoal.
October 1999 3-54 DRAFT-DO NOT QUOTE OR CITE
-------
1 Krieger and Kites (1992) designed a diffusion denuder system that uses capillary gas
2 chromatographic columns as the tubes of SVOC collection. The denuder was followed by a filter
3 to collect particles, which in turn was followed by a polyurethane foam (PUF) plug to collect
4 organic material volatilizing off the filter. Denuder samples were analyzed by liquid solvent
5 extraction (CH2C12) followed by GC-MS analysis. The PUF plugs and filters were extracted with
6 supercritical fluid extraction using supercritical N2O. Using this system, an indoor air sample
7 was found to contain primarily chlorinated biphenyls, ranging from trichlorobiphenyls (vapor
8 pressures 10~3 - 10"4 Torr at 25 °C) to octachlorobiphenyls (10~6 - 10~7 Torr), which demonstrated
9 that the sampler collects compounds with a wide range of volatility. They also found that on-line
10 desorption is successful in maintaining good chromatographic peak shape and resolution. The
11 entire method, from sample collection to the end of the chromatographic separation, took
12 2 hours.
13 Organic acids in both the vapor and particulate phases may be important contributors to
14 ambient acidity, as well as representing an important fraction of organic particulate matter.
15 Lawrence and Koutrakis (1996a,b) used a modified Harvard/EPA annular denuder system
16 (HEADS) to sample both gas and particulate phase organic acids in Philadelphia, PA in the
17 summer of 1992. The HEADS sampler inlet had a 2.1 mm cutpoint impactor (at 10 Lpm),
18 followed by two denuder tubes, and finally a filter pack with a Teflon filter. The first denuder
19 tube was coated with KOH to trap gas phase organic acids. The second denuder tube was coated
20 with citric acid to remove ammonia and thus to avoid neutralizing particle phase acids collected
21 on the filter. The KOH-coated denuder tube was reported to collect gas phase formic and acetic
22 acids at better than 98.5% efficiency, and with precisions of 5% or better (Lawrence and
23 Koutrakis, 1994). It was noted that for future field measurements of particulate organic acids,
24 a Na2CO3-coated filter should be deployed downstream of the Teflon filter to trap organic acids
25 that may evaporate from the Teflon filter during sampling.
26
27 3.2.3.3 Use of Denuder Systems To Measure Semivolatile Compounds
28 Much progress has been made to date in the design of diffusion denuder systems for the
29 measurement and characterization of both the particulate and gaseous phases of semivolatile
30 compounds. Some of the recent research has focused upon reduction in the size of the denuder,
31 optimization of the residence time in the denuder, understanding the effect of diffusion denuders
October 1999 3-55 DRAFT-DO NOT QUOTE OR CITE
-------
1 on the positive quartz filter artifact, identifying changes in chemical composition that occur
2 during sampling, determining the effects of changes in temperature and relative humidity, and
3 identifying possible loses by absorption in impactor coatings.
4
5 Reducing the Size ofDenuders
6 Diffusion denuder systems have been used in a number of studies aimed at quantifying both
7 gaseous and particulate pollution. The typical denuder configuration is an annular diffusion
8 denuder tube of significant length (e.g., 26.5 cm for 10 Lpm, Koutrakis et al., 1988). A more
9 compact design based upon a honeycomb configuration was shown to significantly increase the
10 capacity (Koutrakis et al., 1993). However, in intercomparisons with an annular denuder/filter
11 pack system (Koutrakis et al., 1988), significant losses of ammonia and nitric acid were observed
12 for the honeycomb configuration, and attributed to the large inlet surface area and long sample
13 residence time of the honeycomb design, relative to the annular denuder system. Sioutas et al.
14 (1996a) subsequently designed a modified glass honeycomb denuder/filter pack sampler (HDS)
15 with an inlet that minimizes vapor losses on the inlet surfaces. The modified HDS has reduced
16 inlet surfaces and decreased residence time of sampled gases (NH3 and HNO3) compared to its
17 predecessor (Sioutas et al., 1994). Sioutas et al. (1996b) tested various inlet materials (glass,
18 PFA, and PTFE) in laboratory tests and found that a PTFE Teflon coated inlet minimized loss of
19 sampled gases (1 - 8% losses of HNO3 observed, and -4 - 2 % losses of NH3 observed). The
20 highest inlet losses were observed for HNO3 lost to PFA surfaces (14 - 25%). The modified
21 HDS was tested in laboratory and field tests and found to agree within 10% with the annular
22 denuder system.
23
24 Residence Time in the Denuder
25 The efficiency of a diffusion denuder sampler for the removal of gas phase material can be
26 improved by increasing the residence time of the sampled aerosol in the denuder. However, the
27 residence time can only be increased within certain limits. Since the diffusion denuder reduces
28 the concentration of gas phase semivolatile organic material, semivolatile organic matter present
29 in the particles passing through the denuder will be in a thermodynamically unstable environment
30 and will tend to outgas SVOC during passage through the denuder. The residence time of the
31 aerosol in the denuder, therefore, should be short enough to prevent significant loss of particulate
October 1999 3-56 DRAFT-DO NOT QUOTE OR CITE
-------
1 phase SVOC to the denuder. Various studies have suggested that the residence time in the
2 denuder should be less than about 2 seconds (Gundel and Lane, 1999; Kamens and Coe, 1997;
3 Kamens et al., 1995). The residence times in the various denuder designs described by Gundel
4 and Lane (1999) are from 1.5 to 0.2 seconds.
5
6 Effect of Diffusion Denuders on the Positive Quartz Filter Artifact
7 Denuder systems may be useful to reduce artifact reactions between gases and either the
8 filter substrate or collected particles (Durham et al., 1978). Annular denuder systems, coupled
9 with filter packs employing both Teflon and nylon filters, have been used to study ammonium
10 nitrate, nitric acid, and ammonia in the vapor and aerosol phases (e.g., Benner et al., 1991).
11 To account for the volatilization losses of semi-volatile organic compounds, Turpin et al. (1994)
12 recommended that a quartz filter be placed behind a Teflon filter in a parallel sampler. Addition
13 of a vapor trap (e.g., polyurethane foam plug) downstream of the filter was also suggested as a
14 method to collect semi-volatile organic compounds. However, it was noted that these methods
15 (addition of some type of trap behind the Teflon filter) collected both vapor phase organics as
16 well as "blow-off from the Teflon filter i.e., material vaporized from particles collected on
17 Teflon filter (Van Vaeck et al., 1984). At the time of the previous AQC PM 96 (U.S.
18 Environmental Protection Agency, 1996), some investigators were beginning to examine the
19 phase partitioning of SVOC and accounting for the vapor phase SVOC, the particulate organic
20 matter captured on the filter, and any SVOC that was subsequently lost from the filter by
21 volatilization. Eatough et al. (1993) measured both adsorption and volatilization artifacts by
22 using a sampling train that consisted of a diffusion denuder followed by a filter pack followed by
23 a sorbent bed. The effects of face velocity and pressure drop across the filter were noted, and
24 sampling systems optimizing these parameters to minimize artifacts were discussed. Gundel
25 et al. (1995a,b) have also used sorbent-coated diffusion denuders to examine the phase
26 distributions of semi-volatile organic compounds.
27 The adsorption of organic compounds by a second quartz filter has been shown to be
28 reduced, but not eliminated, in samples collected in the Los Angeles Basin, if a multi-channel
29 diffusion denuder with quartz filter material as the denuder collection surface preceded the quartz
30 filters (Fitz, 1990). This artifact can be further reduced by the use of activated charcoal at the
31 denuder surface and use of a particle concentrator to reduce the amount of gas phase organic
October 1999 3-57 DRAFT-DO NOT QUOTE OR CITE
-------
1 compounds relative to condensed phase organic compounds (Cui et al., 1998, 1997; Eatough,
2 1999). Recent experiments (Cotham and Bidleman, 1992; Cui et al., 1998; Eatough et al., 1995,
3 1996) have shown that the quartz filter artifact can result both from the collection of gas phase
4 organic compounds and from the collection of semivolatile organic compounds lost from
5 particles during sampling. Thus, results available to date suggest that both a "positive" and a
6 "negative" artifact can be present in the determination of particulate phase organic compounds,
7 using two tandem quartz filters.
8
9 Changes in Chemical Composition During Sampling
10 The use of sampling systems designed to correctly identify the atmospheric gas and
11 particulate phase distributions of collected organic material has been outlined above.
12 An additional sampling artifact which has received little consideration in the collection of
13 atmospheric sampling is the potential alteration of organic compounds as a result of the sampling
14 process. These alterations appear to result from the movement of ambient air containing
15 oxidants and other reactive compounds past the collected particles. The addition of NO2
16 (
-------
1 1990). The BOSS denuder should be effective in eliminating most of the chemical
2 transformation artifacts since reactive gases are removed by the charcoal denuder which proceeds
3 the particle collection filter. The implication from these studies is that, if the organic material
4 lost from particles is important to respiratory health problems, then the traditional PM
5 measurement techniques based upon integrated filter sampling followed by equilibration and
6 gravimetric analysis may significantly underestimate the concentrations of POM. Moreover, it
7 appears that the negative sampling artifact associated with volatilization losses from quartz filters
8 is usually much greater than the positive sampling artifact that would be associated with
9 adsorption of SVOC vapors onto quartz fiber filters. Because of the redistribution of SVOC
10 between phases and the potential for chemical changes that occur during sampling, it may be
11 necessary to use diffusion denuders or diffusion separation systems like those described in this
12 section to adequately characterize particulate organic matter in ambient air.
13
14 Temperature and Relative Humidity Effects
15 The problems of sampling artifacts associated with SVOC adsorption and evaporation are
16 compounded by temperature and relative humidity effects (Pankow and Bidleman, 1991; Pankow
17 et al., 1993; Falconer et al., 1995; Goss and Eisenreich, 1997). Effects of temperature on the
18 partitioning of PAH was examined by Yamasaki et al. (1982), who found that the partition
19 coefficient (PAHvapo/PAHpart) was inversely related to temperature and could be described using
20 the Langmuir adsorption concept. The dissociation of ammonium nitrate aerosol is also a
21 function of temperature. Bunz et al. (1996) examined the dissociation and subsequent
22 redistribution of NH4NO3 within a bimodal distribution, using a 9-stage low pressure Berner
23 impactor followed by analysis by ion chromatography and found a strong temperature
24 dependency on the redistribution. Bunz et al. found that at lower temperatures (below 10 °C),
25 there was little change in the aerosol size distribution. At temperatures between 25 and 45 °C,
26 however, the lifetime of NH4NO3 particles decreases by more than a factor of 10, and size
27 redistribution, as measured by average ending particle diameter, increased more for higher
28 temperatures than for lower temperatures.
29 The effects of relative humidity on the sorption of SVOC on particles are not well
30 understood. In a series of laboratory experiments, Goss and Eisenreich (1997) examined the
31 sorption of both polar (hydrocarbons and chlorinated hydrocarbons) and non-polar (ethyl ether
October 1999 3-59 DRAFT-DO NOT QUOTE OR CITE
-------
1 and acetone) volatile organic compounds (VOC) onto combustion soot particles as a function of
2 temperature and relative humidity. The soot particles used in their experiments were collected
3 from oil furnaces and contained 60% (w/w) iron sulfate (water-soluble fraction) and 9% (w/w)
4 elemental and organic carbon. The carbon and sulfate contents of their particulate matter are
5 comparable to the chemical composition of ambient fine particles. They found that, for all
6 compounds, the sorption of VOC onto soot particles decreased with increasing relative humidity
7 over the range of 10 to 95%. They also observed hysteresis in the relative humidity dependency,
8 with sorption coefficients at a given relative humidity higher when the RH is being increased
9 than when the RH is being decreased. The sorption coefficients were fit with an exponential
10 function to the RH so that the slope of the regression line would provide a measure of the
11 influence of relative humidity. Based upon the magnitude of the slope, they concluded that the
12 RH-dependency of sorption was stronger for water-soluble organic compounds. In another study
13 by Jang and Kamens (1998), humidity effects on gas-particle partitioning of SVOC were
14 examined using outdoor environmental chambers and the experimentally determined partitioning
15 coefficients were compared to theoretical values. They examined the partitioning of SVOC onto
16 wood soot, diesel soot, and secondary aerosols and concluded that "the humidity effect on
17 partitioning was most significant for hydrophobic compounds adsorbing onto polar aerosols."
18 Although these two studies seem to be contradictory, upon closer examination, it is difficult to
19 compare the two studies for several reasons. The experiments conducted by Jang and Kamens
20 (1998) were conducted in outdoor chambers at ambient temperatures and humidities. Their
21 model was for absorptive partitioning of SVOC on liquid-like atmospheric particulate matter.
22 In contrast, the results of Goss and Eisenreich (1997) were obtained from a gas chromatographic
23 system operated at 70 °C higher than ambient conditions. The model of Goss and Eisenreich
24 (1997) was for adsorptive partitioning of VOC on solid-like atmospheric particulate matter.
25 In the study of Jang and Kamens (1998), calculated theoretical values for water activity
26 coefficients for diesel soot were based upon an inorganic salt content of 1 to 2%, while the
27 combustion particles studied by Goss and Eisenreich (1997) contained 60% water-soluble,
28 inorganic salt content. Jang and Kamens (1998) obtained their diesel soot from their outdoor
29 chamber, extracted it with organic solvent (mixtures of hexane and methylene chloride), and
30 measured the organic fraction. The resulting salt content of 2% of the particulate matter studied
October 1999 3-60 DRAFT-DO NOT QUOTE OR CITE
-------
1 in Jang and Kamens (1998) is enough to affect water uptake but presumably not to affect the
2 sorption partitioning of organics.
3
4 Adsorption by Impactor Coatings
5 There are other sources of error inherent in some of the currently acceptable practices that
6 could potentially affect particulate mass concentration measurements and that will surely become
7 even more important as more emphasis in particulate sampling is placed upon chemical
8 speciation. Allen et al. (1999) reported that the practice of greasing impaction substrates may
9 introduce an artifact from the absorption of semivolatile species from the gas phase by the grease,
10 which could artificially increase the amount of PAHs and other organic compounds attributed to
11 the aerosol. Allen et al. (1999) offer several criteria to ensure that this absorption artifact is
12 negligible, including selecting impaction oils in which analytes of interest are negligibly soluble;
13 and ensuring that species do not have time to equilibrate between the vapor and oil phases
14 (criterion is met for nonvolatile species). They recommend using oiled impaction substrates only
15 if the absorption artifact is negligible as determined from these criteria. Similarly, the use of
16 coating solutions, especially those that contain organic liquids, in denuder systems may lead to
17 adsorption of volatilized coating solution components on the downstream filters (Finn et al.,
18 1999).
19
20 3.2.3.4 Particle-Bound Water
21 It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
22 compounds. However, for many measurement of suspended particle mass, it is desirable to
23 remove the particle-bound water before determining the mass. In other situations it may be
24 important to know how much of the suspended particle's mass or volume is due to particle-
25 bound water. The water content of PM is significant and highly variable. Moreover, there is
26 significant hysteresis in the water adsorption-desorption pathways (Seinfeld and Pandis, 1998),
27 further complicating the mass measurement. Figures 3-8 and 3-9 show the change in diameter of
28 sulfate particles as a function of relative humidity. Figure 3-8 shows the difference between
29 deliquescence and crystallization points.
30 Pilinis et al. (1989) calculated the water content of atmospheric particulate matter above
31 and below the deliquescent point. They predicted that aerosol water content is strongly
October 1999 3-61 DRAFT-DO NOT QUOTE OR CITE
-------
1 dependent upon composition, and concluded from their calculations that liquid water could
2 represent a significant mass fraction of aerosol concentration at relative humidities above 60%.
3 Since then, a few researchers have attempted to measure the water content of atmospheric
4 aerosol. Most techniques have focused on tracking the particle mass as the relative humidity is
5 changed, and are still in the development phase. There have been only a few demonstrations
6 using actual ambient aerosol, to date. Of interest, in particular, is the development of the
7 Tandem Differential Mobility Analyzer (TDM A) and its applications in investigations of the
8 effects of relative humidity on particle growth.
9 Lee et al. (1997) examined the influence of relative humidity on the size of atmospheric
10 aerosol using a TDMA coupled with a scanning mobility particle sizer (SMPS). They reported
11 that the use of the TDMA/SMPS system allowed for the abrupt size changes of aerosols at the
12 deliquescence point to be precisely observed. They also reported that, at relative humidities
13 between 81 and 89%, the water content of ammonium sulfate aerosols (by mass) was 47 to 66%.
14 Andrews and Larson (1993) investigated the interactions of single aerosol particles coated
15 with an organic film with a humid environment. Using an electrodynamic balance, they
16 conducted laboratory experiments in which sodium chloride and carbon black particles were
17 coated with individual organic surfactants, intended to simulate the surface-active, organic films
18 that many atmospheric aerosol particles may exhibit, and their water sorption curves examined.
19 Their results showed that when ordinarily hydrophobic carbon black particles were coated with
20 an organic surfactant, they sorbed significant amounts of water (20 - 40% of the dry mass of the
21 particle).
22 Liang and Chan (1997) developed a fast technique using the electrodynamic balance to
23 measure the water activity of atmospheric aerosols. In their technique, the mass of a levitated
24 particle is determined as the particle either evaporates or grows in response to a step change in
25 the relative humidity. Their technique was demonstrated using laboratory experiments with
26 NaCl, (NH4)2SO4, NaNO3, and (NH4)2SO4/NH4NO3 solutions. They conclude that one of the
27 advantages of their fast method is the ability to measure the water activity of aerosols containing
28 volatile solutes such as ammonium chloride and some organics.
29 Mclnnes et al. (1996) measured aerosol mass concentration, ionic composition, and
30 associated water mass of marine aerosol over the remote Pacific Ocean. The mass of particle-
31 bound water was determined by taking the difference between the mass obtained at 48% RH and
October 1999 3-62 DRAFT-DO NOT QUOTE OR CITE
-------
1 at 19% RH, assuming the aerosol particles were dry at 19% RH. Based upon a comparison of the
2 remote Pacific aerosol to aerosol collected at a site at the marine/continental interface of the
3 Washington coast, the amount of water associated with the aerosol was observed to be a function
4 of the ammonium to sulfate ratio. They found that the amount of water associated with the
5 submicrometer aerosol comprised 29% of the total aerosol mass collected at 47% RH, and 9% of
6 the total mass at 35% RH.
7 Ohta et al. (1998) characterized the chemical composition of atmospheric fine particles
8 (D50 = 2 mm) in Sapporo, Japan, and as part of their measurements, determined the water
9 content using the Karl Fischer method (Meyer and Boyd, 1959). After exposing a Teflon filter, a
10 portion of the filter was equilibrated at 30% RH for 24 hours. Then the filter piece was placed in
11 a water evaporator heated at 150 °C, vaporizing the particle-bound water. The vapor evolved
12 was analyzed for water in an aqua-counter where it was titrated coulometrically in Karl Fischer
13 reagent solution (containing iodine, sulfur, and methanol). The accuracy of the aqua-counter is
14 ±1 mg. Using this technique, they determined that the water content of the particles ranged from
15 0.4 to 3.2% of the total particulate mass (at RH < 30%). This represents a smaller portion of
16 water compared to their previous reported values (Ohta and Okita, 1990) which were determined
17 by calculation at RH of 50%.
18 Speer et al. (1997) developed an aerosol liquid water content analyzer (LWCA), in which
19 aerosol samples are collected on PTFE filters, and then placed in a closed chamber in which the
20 relative humidity is closely controlled. The aerosol mass is monitored using a beta-gauge, as the
21 relative humidity is increased from low RH to high RH, and then as the RH is decreased again.
22 They demonstrated the LWCA on generated aerosol and on an ambient PM2 5 sample collected in
23 RTF, NC. The ambient aerosol sample was also analyzed for chemical constituents. It is
24 interesting to note that while their laboratory-generated (NH4)2SO4 aerosol demonstrated a sharp
25 deliquescent point, their atmospheric aerosol, which was essentially (NH4)2SO4, did not show a
26 sharp deliquescent point.
27 Hygroscopic properties of aerosols have been studied from the viewpoint of their ability to
28 act as condensation nuclei. The hygroscopic properties of fresh and aged carbon and diesel soot
29 particles were examined by Weingartner et al. (1997) who found that fresh, submicron-size
30 particles tended to shrink with increasing relative humidity, due to a crystalline structural change.
31 Lammel and Novakov (1995) found, in laboratory studies, that the hygroscopicity of soot
October 1999 3-63 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
particles could be increased by chemical modification, and that the cloud condensation
nucleation characteristics of diesel soot were similar to those of wood smoke aerosol.
The results of several of the above studies, in which aerosol water content as a function of
relative humidity was determined, are summarized in Figure 3-12. In this figure, the results of
Lee et al. (1997), Mclnnes et al. (1996), and Ohta et al. (1998) are included. Relative humidity
ranged from 9% (at which the aerosol water content was assumed to be zero, Mclnnes et al.,
1996) to 89%, at which the aerosol water content was determined to be 66% by mass (Lee et al.,
1997).
mn
t_ I UU •
CD
-i— •
<> 90-
T3
§ 80-
CD
i 70-
CD
O
•-E 60-
CD
Q_
0 50-
CD
= 40-
1 30-
3 20-
s. 1°-
° n.
• Mclnnes etal., 1996
• Lee etal., 1997
A Ohta etal., 1998
f
•
A
m, ........
10 20 30 40 50 60 70
Relative Humidity, %
80
90
100
Figure 3-12. Aerosol water content expressed as a mass percentage, as a function of
relative humidity.
Source: Mclnnes et al. (1996); Lee et al. (1997); and Ohta et al. (1998).
1 The effects of relative humidity on particle growth were also examined in several studies.
2 Fang et al. (1991) investigated the effects of flow-induced relative humidity (RH) changes on
October 1999
3-64
DRAFT-DO NOT QUOTE OR CITE
-------
1 particle cut sizes for aqueous sulfuric acid particles in a multi-nozzle micro-orifice uniform
2 deposit impactor (MOUDI). Laboratory experiments were conducted in which polydisperse
3 sulfuric acid aerosols were generated and the RH was adjusted. The aerosols were analyzed by a
4 differential mobility analyzer. Fang et al. (1991) observed that for inlet RH less than 80%, the
5 cut sizes for the sulfuric acid aerosols were within 5% of that for nonhygroscopic particles except
6 at the stage for which the cut size was 0.047 mm, where the cut size was 10.7% larger than the
7 nonhygroscopic particle cut size. They concluded that flow-induced RH changes would have
8 only a modest effect on MOUDI cut sizes at RH < 80%.
9 Hitzenberger et al. (1997) collected atmospheric aerosol in the size range of 0.06 - 15 mm
10 in Vienna, Austria using a 9-stage cascade impactor and measured the humidity-dependent water
11 uptake when the individual impaction foils were exposed to high RH. They observed particle
12 growth with varying growth patterns. Calculated extinction coefficients and single scattering
13 albedo increased with humidity.
14 Hygroscopic properties, along with mixing characteristics, of submicrometer particles
15 sampled in Los Angeles, CA during the summer of 1987 SCAQS study and at the Grand Canyon,
16 AZ during the 1990 Navajo Generating Station Visibility Study were reported by Zhang et al.
17 (1993). They used a tandem differential mobility analyzer (TDMA, McMurry and Stolzenburg,
18 1989) to measure the hygroscopic properties for particles in the 0.05 to 0.5 mm range. In their
19 experimental technique, monodisperse particles of a known size are selected from the
20 atmospheric aerosol with the first DMA. Then, the relative humidity of the monodisperse
21 aerosol is adjusted and the new particle size distribution is measured with the second DMA.
22 At both sites, they observed that monodisperse particles could be classified according to "more"
23 hygroscopic and "less" hygroscopic. Aerosol behavior observed at the two sites differed
24 markedly. The "less" hygroscopic particles sampled in Los Angeles did not grow to within the
25 experimental uncertainty (±2%) when the RH was increased to 90%, whereas at the Grand
26 Canyon, the growth of the "less" hygroscopic particles varied from day to day, but ranged from
27 near 0% to 40% when the RH was increased to 90%. The growth of the "more" hygroscopic
28 particles in Los Angeles, CA was dependent upon particles size (15% at 0.05 //m to 60% at
29 0.5 // m) whereas at the Grand Canyon, the "more" hygroscopic particles grew by about 50%,
30 with the growth not varying significantly with particle size. By comparison of the TDMA data to
31 impactor data, Zhang et al. (1993) surmised that the more hygroscopic particles contained more
October 1999 3-65 DRAFT-DO NOT QUOTE OR CITE
-------
1 sulfates and nitrates, while the less hygroscopic particles contained more carbon and crustal
2 components.
3 Although most of the work to date on the hygroscopic properties of atmospheric aerosols
4 has focused on the inorganic fraction, the determination of the contribution of particle-bound
5 water to atmospheric particulate mass is greatly complicated by the presence of organics. The
6 effects of RH on adsorption of semivolatile organic compounds is discussed elsewhere in this
7 chapter. Saxena et al. (1995) observed that particulate organic compounds can also affect the
8 hygroscopic behavior of atmospheric particles. They idealized the organic component of aerosol
9 as containing a hydrophobic fraction (high-molecular weight alkanes, alkanoic acids, alkenoic
10 acids, aldehydes, and ketones) and a hydrophilic fraction (e.g., lower-molecular weight
11 carboxylic acids, dicarboxylic acids, alcohols, aldehydes, etc.) that would be likely to absorb
12 water. They then analyzed data from a tandem differential mobility analyzer in conjunction with
13 particle composition observations from an urban site (Claremont, CA) and from a non-urban site
14 (Grand Canyon, AZ) to test the hypothesis that, by adding particulate organics to an inorganic
15 aerosol, the amount of water absorbed would be affected, and the effect could be positive or
16 negative, depending on the nature of the organics added. They further presumed that the
17 particulate organic matter in nonurban areas would be predominantly secondary and thus
18 hydrophilic, compared to the urban aerosol that was presumed to be derived from primary
19 emissions and thus hydrophobic in nature. Their observations were consistent with their
20 hypothesis, in that at the Grand Canyon, the presence of organics tended to increase the water
21 uptake by aerosols, whereas at the Los Angeles site, the presence of organics tended to decrease
22 water uptake.
23 Non-equilibrium issues may become of important for the TDMA, as well as for other
24 methods of measuring water content. While approach to equilibrium when the RH is increased is
25 expected to be rapid for pure salts, it may be much slower for aerosols containing a complex mix
26 of components (Saxena et al., 1995). For example, if an aerosol contains an organic film or
27 coating, that film may impede the transport of water across the particle surface, thus increasing
28 the time required for equilibrium (Saxena et al., 1995). Insufficient time to achieve equilibrium
29 in the TDMA could result in underestimation of the water content.
30
October 1999 3-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.2.4 EPA Monitoring Programs
2 3.2.4.1 The Federal Reference Methods (FRM) for Equilibrated Mass
3 Federal Reference Methods (FRM) have been specified for measuring PM10 (Federal
4 Register, 1987) and for measuring PM25 (Federal Register, 1997). The FRM for PM10 has been
5 discussed in previous AQC for PM and will only be briefly reviewed. The PM10 FRM defines
6 performance specifications for samplers in which particles are inertially separated with a
7 penetration efficiency of 50% at an aerodynamic diameter of 10 ± 0.5 //m. The collection
8 efficiency increases to ~100% for smaller particles and drops to ~0% for larger particles.
9 Particles are collected on filters, and mass concentrations are determined gravimetrically.
10 Sample volumes are adjusted to standard conditions (1 atm and 25 °C). Measurement precision
11 for 24-hr samples must be ± 5 ug/m3 for PM10 concentrations below 80 ug/m3, and 7% above this
12 (McMurry, 1999).
13 As opposed to the performance-based FRM standard for PM10, the new FRM for PM25
14 specifies certain details of the sampler design, as well as of sample handling and analysis, while
15 other aspects have performance specifications. The PM25 FRM sampler consists of a PM10 inlet,
16 an oil-soaked impaction substrate to remove particles larger than 2.5 mm, and a 47-mm
17 polytetrafiuoroethylene (PTFE) filter with a particle collection efficiency greater than 99.7%.
18 The sample duration is 24 hours, during which the sampler temperature is not to exceed ambient
19 temperatures by more than 5 °C. Filters are weighed before and after sampling at relative
20 humidities in the range of 30 - 40%, but controlled to within ± 5%. For sampling conducted at
21 ambient relative humidity less than 30%, mass measurements at relative humidities down to 20%
22 are permissible (McMurry, 1999).
23 The FRM also allows for Class I, II, and III equivalent methods for PM2 5. Class I
24 equivalent methods use samplers with relatively small deviations from the sampler described in
25 the FRM. Class II equivalent methods include "all other PM2 5 methods that are based upon
26 24-hr integrated filter samplers that are subjected to subsequent moisture equilibration and
27 gravimetric mass analysis." Class III equivalent methods include filter-based methods having
28 other than a 24-hr collection interval or non-filter-based methods such as beta attenuation,
29 harmonic oscillating elements, and nephelometry (McMurry, 1999).
30 The strength of the PM25 FRM is that specification of all details of the sampler design
31 ensures that measurements at all locations should be comparable. However, the FRM requires
October 1999 3-67 DRAFT-DO NOT QUOTE OR CITE
-------
1 maintenance because of the oil-soaked impaction substrate that could otherwise become loaded
2 with coarse particles. Failure to do so could lead to coarse particle bounce, thus artificially
3 increasing the fine particle concentrations. Moreover, the specification of a PM10 inlet requires
4 the oil-soaked impaction substrate to collect all particles between 2.5 and 10 mm - if an inlet
5 with a smaller cutpoint diameter were specified, coarse particle bounce could potentially be
6 reduced, and perhaps the maintenance frequency could be reduced (McMurry, 1999).
7 Since the implementation of the PM10 standard in 1987 (Federal Register, 1987)
8 considerable information has accumulated on the factors that affect the quality of the data
9 gathered from the EPA reference method for PM10. These include inlet losses of coarse fraction
10 particles (e.g., Anand et al., 1992); biases in concentrations due to differences between samplers
11 in large particle cutpoints that are within the EPA's specified acceptable tolerances (Ranade
12 et al., 1990); and particle bounce tolerances and re-entrainment leading to as much as 30% errors
13 (Wang and John, 1988). The sampling issues associated with cutpoint tolerances are predictable,
14 and the particle bounce and re-entrainment problems have since been dealt with voluntarily by
15 the manufacturers by recommending operational procedures including oiling of impact surfaces
16 and regular cleaning. The AQC PM 96 (U.S. Environmental Protection Agency, 1996)
17 concluded that the PM10 sampling systems can be designed such that concentration measurements
18 are within ±10% of the true concentrations. For PM25, cutpoint tolerances are not expected to
19 affect the mass concentration as much as for PM10, since the 2.5 mm cutpoint generally occurs
20 near a minimum in the mass distribution (e.g., Figure 3-5).
21 The PM25 mass concentration will be affected, on the other hand, by other sampling issues
22 mentioned but not discussed extensively in the previous AQC PM 96 (U.S. Environmental
23 Protection Agency, 1996). These included gas/particle and particle/substrate interactions for
24 sulfates and nitrates (e.g., Appel et al., 1984); volatilization losses of nitrates (Zhang and
25 McMurry, 1992); semivolatile organic compound (SVOC) losses (e.g., Eatough et al., 1993); and
26 relative humidity effects (e.g., Keeler et al., 1988). Due to conversion of SO2 and nitrogen
27 oxides to particulate sulfates and nitrates, respectively, on glass fiber filters, according to the
28 previous AQC PM 96 (U.S. Environmental Protection Agency, 1996), TSP concentration
29 measurements could be inflated by as much as 10 to 20 mg/m3. Losses of particulate nitrates,
30 chlorides, and/or ammonium from quartz fiber filters were noted during storage or during
31 sampling by several researchers (e.g., Witz et al., 1990). Although losses of fine particle nitrates
October 1999 3-68 DRAFT-DO NOT QUOTE OR CITE
-------
1 from Teflon filters were reported by several investigators, there was some evidence that the
2 nitrate artifacts were minor except in California (Malm et al., 1994), and unquantifiable with the
3 current theory. Similarly, significant losses of particulate organic compounds on quartz filters
4 were observed by Eatough et al. (1993). Adsorption of organic vapors onto the filter or collected
5 particulate matter was also observed to cause a positive sampling artifact during the
6 Carbonaceous Species Methods Intercomparison Study (Hering et al., 1990), so that in regions
7 where a significant fraction of the ambient PM mass is organic, there may be significant positive
8 or negative errors in the mass concentration measurement.
9
10 3.2.4.2 Speciation Monitoring
11 In addition to FRM sampling to determine compliance with PM standards, EPA requires
12 States to conduct speciation sampling primarily to determine source categories and trends. The
13 current samplers include three filters: Teflon for equilibrated mass and elemental analysis, a
14 Nylon filter with a nitric acid denuder to collect nitrate, and a quartz fiber filter for elemental and
15 organic carbon (but without any correction for positive or negative artifacts due to adsorption of
16 organic gases or the quartz filters or evaporation of semivolatile organics from the collected
17 particles.
18
19 Measurements for Source Category Apportionment
20 Chemical analyses from the speciation network will be used for source category
21 apportionment via receptor modeling of PM. There are two major approaches to receptor
22 modeling: the chemical mass balance (CMB) receptor modeling approach, and statistically based
23 approaches. The CMB approach requires chemical characterization of all relevant sources and
24 receptor sample characterization should be performed using the same analyses. A considerable
25 amount of receptor modeling work has been conducted with CMB and using elemental analyses
26 coupled with OC/EC and some ionic species (e.g., Watson et al., 1994; Hidy and Venkataraman,
27 1996; McLaren and Singleton, 1996; Vega et al., 1997). Recent developments in receptor
28 modeling include using organic analyses for tracers of specific sources (Benner et al., 1995), and
29 very detailed organic analyses for source fingerprinting (Rogge et al., 1991, 1993a,b,c,d, 1994,
30 1997a,b, 1998) and chemical mass balance receptor modeling (Schauer et al., 1996). Further
October 1999 3-69 DRAFT-DO NOT QUOTE OR CITE
-------
1 detail on the organic analyses for these studies is beyond the scope of this chapter and will not be
2 discussed further, here.
3 Statistical models based upon factor analysis or principal component analysis have the
4 advantage of not requiring detailed source characterization but the drawback is that they require a
5 large data set of receptor sample analyses. These statistically based models have an additional
6 benefit in that they can also use other parameters such as meteorology. For a detailed review of
7 factor analysis and PC A, see Henry et al. (1984). In PC A, many intercorrelated variables within
8 a large data set are sorted into a smaller number of independent components, or factors, that
9 account for the variability in the data set. Veltkamp et al. (1996) reported on a principal
10 component analysis (PCA) study conducted at Niwot Ridge, Colorado, during which organic
11 constituents of atmospheric aerosols were measured, along with physical and meteorological
12 data. Organic compounds were thermally desorbed from the aerosol particles at 250 C in a pure
13 helium atmosphere, separated by gas chromatography, and identified by mass spectrometry.
14 Veltkamp et al. (1996) did not report desorption recoveries or calibration procedures. For each
15 of 48 samples, 31 variables, including 18 particulate organic compounds, along with 11 organic
16 and inorganic vapor species (e.g., NO, NO2, FfNO3, HONO, PAN, H2O2, etc.) wind direction and
17 time of day were used as variables in a principle component analysis. Several factors were
18 identified that served to distinguish various sources, and included gas phase internal combustion
19 products, particulate phase, oxygenated biogenic hydrocarbons, high molecular weight n-alkanes,
20 particulate phase anthropogenic products, and particulate phase biogenic aldehydes.
21 Pinto et al. (1998) also used a combination of PM25 chemical speciation and ambient
22 monitoring data in a receptor modeling calculation to determine the relative sources of particulate
23 pollution in an industrial area in the northern Bohemia region of the Czech Republic. During
24 that study, a severe air pollution episode occurred in 1993 during which smoke and SO2
25 concentrations were 1800 and 1600 //g/m3, respectively.
26 In addition to chemical speciation for factor analysis and source apportionment, Norris
27 et al. (1999) showed that meteorological indices could prove useful in identifying sources of
28 particulate matter that are responsible for observed health effects (specifically asthma) associated
29 with exposure to particulate matter. They examined meteorology associated with elevated
30 pollution events in Spokane and in Seattle, WA, and identified a "stagnation index" that was
31 associated with low wind speeds and increases in concentrations of combustion-related
October 1999 3-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 pollutants. Their factor analysis also identified a meteorological index (low relative humidity
2 and high temperatures) that was associated with increases in soil-derived particulate matter, as
3 well as a third factor (low temperatures and high relative humidity) that was associated with
4 increasing concentrations of particulate sulfate and nitrate species (Norris, 1998).
5 Ondov (1996) discussed the feasibility of using sensitive isotopic and elemental tracer
6 materials to determine the contributions of petroleum-fueled sources of PM10 in the San Joaquin
7 Valley, CA. Costs of these experiments are affected not only by the tracer materials cost, but
8 also by the sensitivities of the analytical methods for each, as well as the background levels.
9 Suarez et al. (1996) used iridium tracer to tag emissions from diesel-burning sanitation trucks in
10 Baltimore, MD, and determined the size distribution of soot from the trucks.
11
12 Elemental Analyses
13 X-ray emission, either stimulated by X-rays (X-ray fluorescence, XRF) or by proton beams
14 (Proton Induced X-ray Emission, PIXE) are standard techniques for non-destructive analysis of
15 certain elements. Some newer techniques with some advantages have become available in recent
16 years.
17 Instrumental Neutron Activation Analysis (INAA). Instrumental neutron activation
18 analysis (INAA) was mentioned briefly in the previous AQC PM 96 and will be expanded upon
19 here. INAA has been used to examine the chemical composition of atmospheric aerosols in
20 several studies, either as the only method of analysis, or in addition to XRF (e.g., Yatin et al.,
21 1994; Gallorini, 1995). INAA has the advantage of having a higher sensitivity for many trace
22 species, and it is particularly useful in analyzing for many trace metals. Landsberger and Wu
23 (1993) analyzed air samples collected near Lake Ontario for Sb, As, Cd, In, I, Mo, Si, and V,
24 using INAA. They demonstrated that using INAA in conjunction with epithermal neutrons and
25 Compton suppression produces very precise values with relatively low detection limits.
26 Enriched rare-earth isotopes have been analyzed via INAA and used to trace sources of
27 particulate matter from a coal-fired power plant (Ondov et al., 1992); from various sources in the
28 San Joaquin Valley (Ondov, 1996); from intentially tagged (iridium) diesel emissions from
29 sanitation trucks (Suarez et al., 1996; Wu et al., 1998) and from iridium-tagged emissions from
30 school buses (Wu et al., 1998).
October 1999 3-71 DRAFT-DO NOT QUOTE OR CITE
-------
1 An intercomparison was conducted in which 18 pairs of filters were sent to participants in
2 the Coordinated Research Program (CRP) on Applied Research on Waste Using Nuclear Related
3 Analytical Techniques (Landsberger et al., 1997). As part of that study, participants used PIXE,
4 INAA, XRF, or AAS to analyze the samples. Many of the results for XRF and PIXE in the coarse
5 fraction were observed to be biased low compared to INAA. The authors speculated that there is
6 a systematic error due to self-attenuation of the X-rays due to particle size effect.
7 Atomic Absorption Spectrophotometry (AAS). Atomic Absorption Spectrophotometry
8 (AAS) was used to characterize the atmospheric deposition of trace elements Zn, Ni, Cr, Cd, Pb,
9 and Hg, to the Rouge River watershed by particulate deposition (Pirrone and Keeler, 1996). The
10 modeled deposition rates were compared to annual emissions of trace elements that were
11 estimated from the emissions inventory for coal and oil combustion utilities, iron-steel
12 manufacturing, metal production, cement manufacturing, and solid waste and sewage sludge
13 incinerators. They found generally good agreement between the trend observed in atmospheric
14 inputs to the river (dry + wet deposition) and annual emissions of trace elements, with
15 r2 ~ 0.84 - 0.98. Both atmospheric inputs and emissions were found to have followed downward
16 trends for Pb. For the period of 1987 to 1988, steady increases were observed for Cd (major
17 sources are municipal solid waste incineration, coal combustion, sludge incineration, and iron
18 and steel manufacturing); Cr and Ni (major sources are iron and steel production, and coal
19 combustion); and Hg (major sources are coal, the contribution from which had decreased from
20 53 to 45%, and municipal, solid waste, and medical waste incineration, the contribution from
21 which has increased).
22 Inductively Coupled Plasma - Mass Spectroscopy (ICP-MS). Keeler and Pirrone (1996)
23 also used inductively coupled plasma - mass spectroscopy (ICP-MS) to determine trace elements
24 Cd, Mn, V, As, Se, and Pb in atmospheric particulate fine (PM25) and total suspended particulate
25 samples collected in two Detroit sites. The results were then similarly used in a deposition
26 model to estimate the dry deposition flux of trace elements to Lake Erie.
27
28 Elemental/Organic Carbon, Soot, or Particulate Organic Matter
29 Total carbon in aerosol particles (TC) can be expressed as the sum of organic carbon (OC),
30 elemental carbon (EC), and carbonate carbon (CC), with the contribution of CC to TC usually on
31 the order of 5% or less, for particulate samples collected in urban areas (Appel, 1993). The AQC
October 1999 3-72 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM 96 (U.S. Environmental Protection Agency, 1996) listed several filter-based, thermal
2 methods for measuring OC and EC, and described the thermal/optical reflectance (TOR) method,
3 which was noted, along with thermal manganese oxidation, to be one of the most commonly
4 applied methods in the U.S. at the time. In thermal separation methods, thermally evolved
5 OC and EC are oxidized to CO2 and quantified either by nondispersive infrared detection or
6 electrochemically, or the CO2 can be reduced to CH4 and quantified via flame ionization
7 detection (FID). The various methods give similar results for TC, but not for EC or OC. In a
8 methods comparison study (Countess, 1990), it was shown that it is necessary to minimize or
9 correct for pyrolytically generated EC ("char"), and that CC found in wood smoke and
10 automobile exhaust samples may interfere with some of the thermal methods. Recently,
11 Lavanchy et al. (1999) reported on a study in which the operation of a catalytic oxidation system
12 was modified in an attempt to minimize pyrolysis of OC, and at the same time minimize the
13 oxidation of CaCO3. In the thermal apparatus used by Lavanchy et al. (1999), a filter sample is
14 placed in a moveable sample boat and, prior to insertion into the 340 °C oven, the sample is flash
15 heated in the 650 °C oven for one minute, in order to minimize charring. It then is inserted into
16 the first stage of a two-stage oven, in which OC is oxidized to CO2 at 340 °C in the presence of
17 O2 for 42 minutes. The filter is then moved into the second stage oven, in which EC is oxidized
18 at 650 °C, for 32 minutes. This temperature is reported to be sufficient to completely oxidize
19 EC, but with only about 1% of the CaCO3 being oxidized (Lavanchy et al., 1999; Petzold et al.,
20 1997). In order to test for charring, they challenged their system with atmospheric samples for
21 which duplicates were analyzed via the German reference method for measuring OC and EC in
22 atmospheric samples (Petzold and Niessner, 1995), in which a solvent extraction is used to
23 remove organics before combustion. Lavanchy et al. (1999) reported a high correlation
24 (R2 = 0.97) between their thermal oxidation method and the German method VDI. They also
25 reported detection limits of 1.3 //g for EC and 1.8 //g for OC.
26 The thermal/optical transmission method (TOT) was mentioned briefly in the AQC PM 96
27 (U.S. Environmental Protection Agency, 1996), as being similar to the TOR with the exception
28 that light transmission rather than reflectance is monitored on the filter throughout the analysis.
29 The National Institute for Occupational Safety and Health (NIOSH) Method 5040 for monitoring
30 elemental carbon as a marker for particulate diesel exhaust is based upon a TOT method analyzer
31 (Birch and Gary, 1996) and has gained significantly in popularity since then, and so will be
October 1999 3-73 DRAFT-DO NOT QUOTE OR CITE
-------
1 described in more detail, here. The OC/EC method described by Birch and Gary (1996) is a TOT
2 method similar to the TOR method described in the AQC PM 96 (U.S. Environmental Protection
3 Agency, 1996) in that temperature and atmospheric control are used to accomplish carbon
4 speciation; several temperature steps are utilized; carbon evolved is converted to methane and
5 quantified with FID; and light transmission, rather than reflectance, off the filter is measured
6 during the analysis. In thermal/optical methods, the optical feature allows for correction for
7 pyrolytic char generated during the analysis. The OC/EC method of Birch and Gary (1996)
8 consists of a two stage process, with the first stage being conducted in a pure helium atmosphere,
9 and the second stage conducted in a 10% oxygen-helium mix. The temperature is ramped to
10 about 820 °C in the helium phase, during which organic and carbonate carbon are volatilized
11 from the filter. In the second stage, the oven temperature is reduced, and then raised to about
12 860°C. During the second stage, pyrolysis correction and EC measurement are made.
13 Figure 3-13, an example of a thermogram, shows temperature, transmittance, and FID response
14 traces. Peaks are evident that correspond to OC, CC, EC, and pyrolitic carbon (PC). As can be
15 seen in this figure, the high temperature in the first stage allows for decomposition of CC. The
16 ability to quantify PC is particularly important in high OC/EC regions (like wood smoke
17 -impacted airsheds), allowing for the volatilization of any remaining complex organic
18 compounds so they are not apportioned to the EC phase.
19 Black carbon (BC) is also used, in addition to the thermal and thermal/optical methods, for
20 determining EC as a measure of soot (Penner and Novakov, 1996). Both EC and BC define a
21 similar fraction of aerosol, but EC is determined based upon thermal properties, while BC is
22 based upon light-absorption properties. Optical methods for determining BC tend to suffer from
23 calibration problems (Hitzenberger et al., 1996). Lavanchy et al. (1999) compared their EC
24 concentrations as determined from their catalytic thermal oxidation method to BC concentrations
25 determined using an aethalometer operated at the same site, and found that the instrumental
26 calibration factor provided by the manufacturer was on the order of two times the calibration
27 factor they determined (9.3 ± 0.4 m2 g"1). It is possible to calculate a theoretical specific
28 absorption coefficient (BJ from Mie theory given a known size distribution and refractive index,
29 and typically BC aerosols have values of Ba between 3 and 17m2 g"1 (Hintzenberger et al. [1996]
30 and references therein). The absorption coefficient, (BJ, is defined as absorption per mass
31 concentration and can be calculated given the sample filter area, the total deposited mass, and
October 1999 3-74 DRAFT-DO NOT QUOTE OR CITE
-------
DC . EC spirt
Time/mi ft
Figure 3-13. This thermogram, for a sample containing rock dust (a carbonate source)
and diesel exhaust, shows three traces that correspond to temperature, filter
transmittance, and FID detector response. Peaks correspond to organic
(OC), carbonate (CC), pyrolytic (PC), and elemental (EC) carbon. The final
peak is a methane calibration peak.
Source: Birch and Gary (1996).
1
2
3
4
5
6
absorption signals for both the loaded and unloaded filters. Often, when no direct measurements
are available, values of Ba on the order of 10 m2 g"1 have been used (Hintzenberger et al. (1996),
and references therein). European countries are trying to set air pollution standards that target
diesel vehicles, one of the principal sources of BC in urban areas (Hintzenberger et al. (1996),
and references therein) and so it is essential that accurate values for Ba are available.
Hintzenberger et al. (1996) investigated the feasibility of using an integrating sphere photometer
October 1999
3-75
DRAFT-DO NOT QUOTE OR CITE
-------
1 as an adequate measurement system for the BC content and the absorption coefficient. Based
2 upon samples collected during a 10-day period in May 1994, they determined that the usually
3 assumed value of 10 m2 g"1 was also applicable to aerosol BC occurring in Vienna.
4 In 1986, the Carbonaceous Species Methods Comparison Study (CSMCS) was conducted
5 in Los Angeles, CA, during which a number of methods for the measurement of this species were
6 intercompared. The CSMCS was mentioned in the previous AQC PM 96 (U.S. Environmental
7 Protection Agency, 1996), however it is interesting to add that Hansen and McMurry (1990)
8 specifically compared two very dissimilar methods for aerosol elemental carbon - collection of
9 impactor samples backed by a quartz fiber afterfilter, followed by EC analysis by oxidation in
10 helium over a MnO2 catalyst, and real-time measurements using an aethalometer (an optical
11 absorption technique) - and found good agreement between these two, very different methods.
12 The comparisons between organic carbon measurements exhibited considerably less agreement.
13 Hitzenberger et al. (1999) recently reported on a study in which the integrating sphere
14 method was compared to an aethalometer (Hansen et al., 1984), the thermal method of Cachier
15 et al. (1989), and the thermal/optical method of Birch and Gary (1996). The absorption
16 coefficients that were obtained from both the integrating sphere and the aethalometer were
17 comparable. The BC mass concentration obtained from the aethalometer were 23% of those
18 obtained from the integrating sphere. Compared to the thermal method, the integrating sphere
19 overestimated the BC mass concentrations by 21%. Compared to the thermal/optical method, the
20 integrating sphere was within 5% of the 1:1 line, however the data were not so well-correlated.
21 Turpin et al. (1990) reported on an in-situ, time-resolved analyzer for particulate organic
22 and elemental carbon that could operate on a time cycle as short as 90 minutes. The analyzer is
23 comprised of a filter-based sampling section and a thermal-optical carbon detector. Adsorbed
24 organic material is thermally desorbed from the filter at 650 °C and oxidized at 1000°C over a
25 MnO2 catalyst bed. The evolved CO2 is converted to methane over a nickel catalyst, and the
26 methane is measured in a flame ionization detector. Then the elemental carbon is oxidized in a
27 98% He-2% O2 atmosphere, at 350°C. Correction is made for pyrolytic conversion of some of
28 the organic particulate matter. The instrument was operated with a 2 h time resolution during the
29 Southern California Air Quality Study (SCAQS) in 1987 (Turpin and Huntzicker, 1991), as well
30 as during the Carbonaceous Species Methods Comparison Study (CSMCS), in 1986.
October 1999 3-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 In summary, the state of the art for soot measurements continues to develop, and although
2 advances are being made, the definitions of EC and BC continue to be operational and
3 determined by the method employed.
4
5 3.2.5 Continuous Monitoring
6 The U.S. EPA expects that 100 local agency monitoring sites throughout the States will
7 operate continuous PM monitors. However, EPA has not yet provided any guidance regarding
8 what continuous monitoring technique to use. All currently available continuous measurements
9 of suspended particle mass share the problem of dealing with semivolatile PM components.
10 In order not to include particle-bound water as part of the mass, the particle bound water must be
11 removed by heating or dehumidification. However, heating also causes loss of ammonium
12 nitrate and semivolatile organic components. A variety of potential candidates for continuous
13 measurement of mass or chemical components will be discussed in this section.
14
15 3.2.5.1 TEOM
16 The advantages of continuous PM monitoring, and the designation of the Tapered Element
17 Oscillating Microbalance (TEOM) as an equivalent method for PM10, has led to the deployment
18 of the TEOM at a number of air monitoring sites. The TEOM is also being used to measure
19 PM25. The TEOM differs philosophically from the federal reference methods for particulate
20 mass in that it does not require equilibration of the samples at a specified temperature and
21 relative humidity. Moreover, the TEOM samples at a constant temperature, typically heated to
22 some temperature higher than the ambient temperature (Meyer et al., 1995; Meyer and
23 Rupprecht, 1996), whereas the federal reference methods sample at the ambient temperature.
24 Thus, the TEOM may not provide data equivalent to the FRM due to losses of volatile species.
25 Volatilization losses in the TEOM sampler can be reduced by operating the instrument heated to
26 30 °C rather than the 50 °C specified, during the cooler times of the year, and by using Nafion
27 dryers on the inlet.
28 This philosophical difference in operation and the implications for fine particle
29 measurements were examined by researchers at CSIRO Atmospheric Research in Australia
30 (Ayers et al., 1999). That group compared 24-hr mean PM25 mass concentrations as determined
31 by a TEOM and by two manual, gravimetric samplers (a low-volume filter sampler and a
October 1999 3-77 DRAFT-DO NOT QUOTE OR CITE
-------
1 MOUDI sampler) in four Australian cities, on 15 days in the winter half-year. The TEOM was
2 operated at 50 °C at one location and at 35 °C at the other three locations. A systematically low
3 TEOM response in comparison to the integrated gravimetric methods was observed. In a
4 comprehensive study, Allen et al. (1997) reported results in which TEOM data collected at
5 10 urban sites in the U.S. and Mexico were compared with 24-hr integrated mass concentrations,
6 for both PM10 and PM2 5. They collected a large data set that included both winter and summer
7 seasons. Allen et al. (1997) concluded that, especially for urban areas, a significant portion of
8 PM10 could be semivolatile compounds that could be lost from the heated filter in the TEOM
9 thus leading to a systematic difference between the TEOM and the EPA FRM for PM10.
10 Moreover, they suggested that this difference is likely to be larger for PM25 than it is for PM10
11 (Allen etal, 1997).
12
13 3.2.5.2 RAMS
14 A Real-Time total Ambient Mass Sampler, RAMS, based on diffusion denuder and TEOM
15 monitor technology has been developed, validated, and field tested (Eatough et al., 1999; Obeidi
16 and Eatough, 1999) for the real-time determination of total fine particulate mass, including
17 semivolatile species. The RAMS measures total mass of collected particles, including
18 semivolatile species with a TEOM monitor using a "sandwich filter". The "sandwich" contains a
19 Teflon coated particle collection filter followed by a charcoal impregnated filter (GIF) to collect
20 any semivolatile species lost from the particles during sampling. Since the instrument measures
21 total mass collected by the "sandwich filter," all gas phase compounds that can be adsorbed by a
22 GIF must be removed from the sampling stream prior to the TEOM monitor. Laboratory and
23 field validation data indicate that the precision of fine PM mass determination is better than 10%.
24 The RAMS uses a Nafion dryer to remove particle bound water from the suspended particles and
25 a particle concentrator to reduce the amount of gas phase organics that must be removed by the
26 denuder. An example of data from the RAMS, the TEOM, and the PC-BOSS is shown in
27 Figure 3-14.
28
29 3.2.5.3 CAMM
30 Koutrakis and colleagues (Koutrakis et al., 1996; Wang, 1997) have developed the
31 Continuous Ambient Mass Monitor (CAMM), a technique for the continuous measurement of
October 1999 3-78 DRAFT-DO NOT QUOTE OR CITE
-------
PC-BOSS (Nonvolatile Material)
PC-BOSS (Lost From Particles)
TEOM
at35C
—B— RAMS
at35C
FRM PM25
24 h average
100
Riverside, CA
•Q
&JL 5L_
13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8 9
Time of Day
Figure 3-14. Comparison of mass measurements with collocated RAMS (real-time data),
PC-BOSS (1-hour data), FRM PM25 sampler (24-hour data) and a
conventional TEOM monitor (real-time data). The semivolatile fine
particulate matter is sampled with the RAMS and PC-BOSS, but not with
the TEOM monitor or the FRM PM25 single filter sampler. The PC-BOSS
provides information on both the nonvolatile component (NV) and the
semivolatile organic component (SVOC).
Source: Eatoughetal. (1999).
1 ambient particulate matter mass concentration, based upon the measurement of pressure drop
2 increase with particle loading across a membrane filter. Recently, Sioutas et al. (1999) examined
3 the increase in pressure drop with increasing particle loading on Nuclepore filters. They tested
4 filters with two pore diameters (2 and 5 mm) and filter face velocities ranging from 4 to 52 cm/s,
5 and examined the effects of relative humidity in the range of 10 to 50%. They found that, for
October 1999
3-79
DRAFT-DO NOT QUOTE OR CITE
-------
1 hygroscopic ammonium sulfate particles, the change in pressure drop per unit time and
2 concentration was a strong function of relative humidity, decreasing with increasing relative
3 humidity. These results suggest that particulate concentration measurements like the method of
4 Koutrakis et al. (1996) that use the pressure drop method may be subject to additional
5 uncertainties if used in an environment where the ambient relative humidity cannot be accurately
6 controlled. The current version of the CAMM (Wang, 1997) uses a particle concentrator, a
7 Nafion dryer, and frequent changes of the position on the filter tape were the pressure drop
8 measurement is made to avoid artifacts due to semivolatile components.
9
10 3.2.5.4 Light Scattering
11 The evaporation of ammonium nitrate aerosol in a heated nephelometer was examined by
12 Bergin et al. (1997). This is potentially of concern because the nephelometer operates in part by
13 heating the ambient aerosol to a low reference relative humidity of 40%, in order to measure the
14 light scattering intrinsic to the aerosol rather than including atmospheric relative humidity.
15 Bergin et al. conducted laboratory experiments at low relative humidity (~10%) and as a function
16 of temperature (300 - 320K), mean residence time in the nephelometer, and initial particle size
17 distribution. The evaporation of ammonium nitrate aerosol was also modeled, for comparison,
18 and was found to accurately describe the decrease in aerosol scattering coefficient as a function
19 of aerosol physical properties, and nephelometer operating conditions. Bergin et al. (1997)
20 determined an upper limit estimate of the decrease in the aerosol light scattering coefficient at
21 450 nm due to evaporation for typical field conditions. The model estimates for their worst-case
22 scenario suggest that the decrease in the aerosol scattering coefficient could be roughly 40%.
23 Under most conditions, however, they estimate that the decrease in aerosol scattering coefficient
24 is generally expected to be less than 20%.
25 Morawska et al. (1996) examined the correlations between PM10, visibility, and submicron
26 concentration data in Brisbane, and concluded that the different principles of operation for each
27 instrument and the different aerosol characteristics measured by each technique make it difficult
28 to observe any relationships. Morawska et al. (1998b) reported on a long-term monitoring
29 program that included the criteria pollutants as well as light scattering, number/size distributions,
30 number concentrations, and elemental analysis via inductively coupled plasma mass
31 spectrometry. Particle size classification was conducted using a TSI scanning mobility particle
October 1999 3-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 sizer for the size range of 0.016 to 0.7 mm, and a TSI aerodynamic particle sizer for the size
2 range of 0.7 to 30 mm. They reported correlation coefficients between the light-scattering
3 coefficient and PM10, SMPS concentration, and APS concentration of 0.58, 0.38, and 0.37,
4 respectively. They also reported a correlation coefficient between PM10 and the SMPS
5 concentration of 0.25, which is consistent with the notion that PM10 mass measurements would
6 provide less information about ultrafine particles.
7
8 3.2.5.5 Beta-Gauge Techniques
9 The use of absorption of beta radiation as a indicator of particle mass has been used
10 effectively to measure the mass of equilibrated particulate matter collected on Teflon filters
11 (Jaklevic et al., 1981; Courtney et al., 1982). The technique has also been used to provide near
12 real-time measurements with time intervals on the order of an hour (Wedding and Weigand,
13 1993). However, real-time beta gauge monitors experience the same problems as other
14 continuous or near real-time particular matter mass monitoring techniques. Particle-bound water
15 must be removed to reduce the sensitivity of the indicated mass to relative humidity. However,
16 the simplest technique, mild heating, will remove a portion of the ammonium nitrate and the
17 semivolatile organic compounds as well as the particle-bound water.
18 An intercomparison study of two beta gauges at three sites indicated that the Wedding beta
19 gauge and the Sierra Anderson SA 1200 PM10 samplers were highly correlated (r>0.97) (Tsai and
20 Cheng, 1996). The Wedding beta gauge was not sensitive to relative humidity but was
21 approximately seven percent lower. This suggests that the mild heating in the beta gauge causes
22 losses comparable to those due to equilibration, although the differences could be due to slight
23 differences in the upper cut points. The Kimoto beta gauge, however, which was operated at
24 ambient temperature, was sensitive to relative humidity, yielding significantly higher mass
25 concentrations relative to the Sierra Anderson SA 1200 for RH>80% than for RH<80%, even
26 though the correlation with the SA 1200 was reasonable, r=0.94 for RH>80% and 0.83 for
27 RH<80%.
28
29 3.2.5.6 Measurements of Individual Particles
30 A new technique, aerosol time-of-flight mass spectroscopy (ATOFMS), has demonstrated
31 the ability for real-time measurement of correlated size and composition profiles of individual
October 1999 3-81 DRAFT-DO NOT QUOTE OR CITE
-------
1 atmospheric aerosol particles (Noble and Prather, 1996; Gard et al., 1997). Measurements are
2 made in-situ by combining a dual-laser aerodynamic particle sizing system to size and track
3 individual particles through the instrument and laser desorption/ionization time-of-flight mass
4 spectrometry to obtain correlated single particle composition data. By measuring both positive
5 and negative ions from the same particle, information can be obtained about the chemical
6 composition, not just the elemental composition, of individual particles of know aerodynamic
7 diameter. This information is especially useful in determining sources of particles. An example
8 of the type of information that can be determined is shown in Figure 3-15.
o_
eg
Figure 3-15. Size Distribution of particles divided by chemical classification into organic,
marine, and crustal.
1 Until recently, single particle ATOFMS systems have only been able to characterize
2 particles that are larger than approximately 0.2 to 0.3 mm in diameter. The work of Wexler and
3 colleagues (Carson et al., 1997; Ge et al., 1998) have developed a single particle, TOFMS
October 1999
3-82
DRAFT-DO NOT QUOTE OR CITE
-------
1 instrument that it is able to size, count and provide chemical composition on individual particles
2 ranging in size from 10 nm to 2 //m.
3
4
5 3.3 SUMMARY
6 Atmospheric particles originate from a variety of sources and possess a range of
7 morphological, chemical, physical, and thermodynamic properties. The composition and
8 behavior of airborne particles are linked with those of the surrounding gas. Aerosol is defined as
9 a suspension of solid or liquid particles in air and includes both the particles and all vapor or gas
10 phase components of air. However, the term aerosol is often used to refer to the suspended
11 particles only. Particulate is an adjective and should only be used as a modifier, as in particulate
12 matter.
13 A complete description of the atmospheric aerosol would include an accounting of the
14 chemical composition, morphology, and size of each particle and the relative abundance of each
15 particle type as a function of particle size. Recent developments in single particle analysis
16 techniques are bringing such a description closer to reality.
17 The diameter of a spherical particle may be determined geometrically, from optical or
18 electron microscopy; by light scattering and Mie theory; or by its behavior, such as its electrical
19 mobility or its aerodynamic behavior. However, the various types of diameters may be different
20 and atmospheric particles often are not spherical. Therefore, particle diameters are described by
21 an "equivalent" diameter. Aerodynamic diameter, i.e., the diameter of a unit density sphere
22 which would have the same physical behavior, is the most widely used equivalent diameter.
23 Therefore, in this document, particle diameters, unless otherwise indicated, refer to the
24 aerodynamic diameter.
25 Atmospheric size distributions show that most atmospheric particles are quite small, below
26 0.1 //m, while most of the particle volume (and therefore most of the mass) is found in particles
27 greater than 0.1 //m. The surface area peaks around 0.1 //m. An important feature of the mass or
28 volume size distributions of atmospheric aerosols is their multimodal nature. Volume
29 distributions, measured in ambient air in the United States, are almost always found to be
30 bimodal, with a minimum between 1.0 and 3.0 //m. The distribution of particles that are mostly
31 larger than the minimum is termed the coarse mode. The distribution of particles that are mostly
October 1999 3-83 DRAFT-DO NOT QUOTE OR CITE
-------
1 smaller than the minimum is termed the fine mode. Fine-mode particles include both the
2 accumulation mode and the nuclei mode. Accumulation-mode particles are that portion of the
3 fine particle fraction with diameters above about 0.1 //m. The nuclei mode, that portion of the
4 fine particle fraction with diameters below about 0.1 //m, can be observed as a separate mode in
5 mass or volume distributions only in clean or remote areas or near sources of new particle
6 formation by nucleation. Toxicologists use ultrafine to refer to particles, generated in the
7 laboratory, which are in the nuclei-mode size range. Aerosol physicists and material scientists
8 tend to use nanoparticles to refer to particles in this size range generated in the laboratory.
9 The aerosol community uses four different approaches or conventions in the classification
10 of particles by size: (1) modes, based on the observed size distributions and formation
11 mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling device,
12 i.e., the particle size at which 50% of the particles enter and 50% of the particles are rejected;
13 (3) occupational sizes, based on the entrance into various compartments of the respiratory
14 system; and (4) legally-specified, regulatory sizes for air quality standards. Over the years, the
15 terms fine and coarse, as applied to particle sizes, have lost the original precise meaning of fine
16 mode and the coarse mode. In any given article, therefore, the meaning of fine and coarse, unless
17 defined, must be inferred from the author's usage. In particular, PM25 and fine mode particles
18 are not equivalent. In this document, the term mode is used with fine and coarse when it is
19 desired to specify the distribution of fine-mode particles or coarse-mode particles as shown in
20 Figures 3-4 and 3-5.
21 Several processes influence the formation and growth of particles. New particles may be
22 formed by nucleation from gas phase material. Particles may grow by condensation as gas phase
23 material condenses onto existing particles. Particles may also grow by coagulation as two
24 particles combine to form one. Gas phase material condenses preferential on smaller particles
25 and the rate constant for coagulation of two particles decreases as the particle size increases.
26 Therefore, nuclei mode particles grow into the accumulation mode but accumulation mode
27 particles do not grow into the coarse mode.
28 The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
29 ions; particle-bound water; elemental carbon; a great variety of organic compounds; and crustal
30 material. Atmospheric PM contains a large number of elements in various compounds and
31 concentrations and hundreds to thousands of specific organic compounds. Particulate material
October 1999 3-84 DRAFT-DO NOT QUOTE OR CITE
-------
1 can be primary or secondary. PM is called primary if it is in the same chemical form in which it
2 was emitted into the atmosphere. PM is called secondary if it is formed by chemical reactions in
3 the atmosphere. Primary coarse particles are usually formed by mechanical processes. Primary
4 fine particles are emitted from sources, either directly as particles or as vapors which rapidly
5 condense to form particles.
6 Most of the sulfate and nitrate and a portion of the organic compounds in atmospheric
7 particles are secondary, i.e., they are formed by chemical reactions in the atmosphere. Secondary
8 aerosol formation depends on numerous factors including the concentrations of precursors; the
9 concentrations of other gaseous reactive species such as ozone, hydroxyl radical, or hydrogen
10 peroxide; atmospheric conditions including solar radiation and relative humidity; and the
11 interactions of precursors and preexisting particles within cloud or fog droplets or on or in the
12 liquid film on solid particles. As a result, it is considerably more difficult to relate ambient
13 concentrations of secondary species to sources of precursor emissions than it is to identify the
14 sources of primary particles.
15 The lifetimes of particles vary with particle size. Coarse particles can settle rapidly from
16 the atmosphere within minutes or hours, and normally travel only short distances. However,
17 when mixed high into the atmosphere, as in dust storms, the smaller-sized, coarse-mode particles
18 may have longer lives and travel distances. Nuclei mode particles rapidly grow into the
19 accumulation mode. However, the accumulation mode does not grow into the coarse mode.
20 Accumulation-mode fine particles are kept suspended by normal air motions and have very low
21 deposition rates to surfaces. They can be transported thousands of km and remain in the
22 atmosphere for a number of days. Accumulation-mode particles are removed from the
23 atmosphere primarily by cloud processes. Coarse mode particles of less than ~10 //m diameter as
24 well as accumulation-mode and nuclei-mode (or ultrafine) particles all have the ability to
25 penetrate deep into the lungs and be removed by deposition in the lungs. Dry deposition rates are
26 expressed in terms of a deposition velocity which varies as the particle size, reaching a minimum
27 between 0.1 and 1.0 //m aerodynamic diameter.
28 The role of particles in reducing visibility and affecting radiative balance through scattering
29 and absorption of light is evident as are the effects of particles in soiling and damaging materials.
30 EPA addresses visibility effects through regional haze regulations. The direct effects of particles
October 1999 3-85 DRAFT-DO NOT QUOTE OR CITE
-------
1 in scattering and absorbing light and the indirect effects of particles on clouds are being
2 addressed in climate change programs in several government agencies.
3 The role of PM in acid deposition has not always been recognized. Acid deposition and
4 PM are intimately related, however, first because particles contribute significantly to the
5 acidification of rain and secondly because the gas phase species that lead to dry deposition of
6 acidity are also precursors of particles. Therefore, reductions in SO2 and NOX emissions will
7 decrease both acid deposition and PM concentrations. Sulfuric acid, ammonium nitrate, and
8 organic particles are also deposited on surfaces by dry deposition. The utilization of nitrate by
9 plants leads to the production of acidity. Therefore, dry deposition of particles can also
10 contribute to the ecological damages caused by acid deposition.
11 It has been proposed that particles could act as carriers to transport toxic gases into the deep
12 lung. Water-soluble gases, which would be removed by deposition to wet surfaces in the upper
13 respiratory system during inhalation, could dissolve in particle-bound water and be carried with
14 the particles into the deep lung. Equilibrium calculations indicate that particles do not increase
15 vapor deposition in human airways. However, their calculations do show that soluble gases are
16 carried to higher generation airways (deeper into the lung) in the presence of particles than in the
17 absence of particles. In addition, species such as SO2 and formaldehyde react in water, reducing
18 the concentration of the dissolved gas-phase species, and providing a kinetic resistence to the
19 evaporation of the dissolved gas. Thus, the concentration of the dissolved species may be greater
20 than that predicted by the equilibrium calculations. Toxic species, such as NO, NO2, benzene,
21 polycyclic aromatic hydrocarbons (PAH), nitro-PAH, and a variety of allergens may be absorbed
22 on solid particles and carried into the lungs.
23 The decision by the US EPA to revise the PM standards by adding daily and yearly
24 standards for PM2 5 has led to a renewed interest in the measurement of atmospheric particles and
25 also to a better understanding of the problems in developing precise and accurate measurements
26 of particles. Unfortunately, it is very difficult to measure and characterize particles suspended in
27 the atmosphere.
28 The US Federal Reference Methods (FRM) for PM25 and PM10 provide relatively precise
29 (±10 %) methods for determining the mass of material remaining on a Teflon filter after
30 equilibration. However, numerous uncertainties exist as to the relationship between the mass and
31 composition of material remaining on the filter, as measured by the FRMs, and the mass and
October 1999 3-86 DRAFT-DO NOT QUOTE OR CITE
-------
1 composition of material that exists in the atmosphere as suspended PM. It is currently not
2 possible to accurately measure what exists as a particle in the atmosphere, in part because of the
3 difficulty of creating a reference standard for particles suspended in the atmosphere. As a result,
4 EPA defines accuracy for PM measurements in terms of agreement of a candidate sampler with a
5 reference sampler. Therefore, intercomparisons of samplers become very important in
6 determining how well various samplers agree and how various design choices influence what is
7 actually measured. Reasons for measuring particles include: attainment of a standard,
8 implementation of a standard, determination of health effects, determination of ecological
9 effects, and determination of radiative effects.
10 Current filtration-based mass measurements lead to significant evaporative losses, during
11 and possibly after collection, of a variety of semivolatile components, i.e., species which exist in
12 the atmosphere in dynamic equilibrium between the condensed phase and gas phase. Important
13 examples include ammonium nitrate, semivolatile organic compounds, and particle-bound water.
14 Other areas where choices must be made in designing an aerosol indicator include selection of an
15 upper cut point; separation of fine-mode and coarse-mode PM; and treatment of pressure,
16 temperature, and relative humidity.
17 It is becoming increasingly apparent that the semivolatile component of PM may
18 significantly impact the quality of the measurement, and can lead to both positive and negative
19 sampling artifacts. Negative artifacts, due to loss of ammonium nitrate and semivolatile organic
20 compounds, may occur during sampling, due to changes in temperature, relative humidity, or
21 composition of the aerosol, or due to pressure drop across the filter. Negative artifacts may also
22 occur during handling and storage due to evaporation. Positive artifacts occur when volatile
23 species adsorb onto, or react with, filter media and/or collected PM.
24 The loss of particulate nitrate may be determined by comparing nitrate collected on a
25 Teflon filter to that collected on a nylon filter (which absorbs nitrate) preceded by a denuder to
26 remove nitric acid. In two studies, the PM2 5 mass lost due to volatilization of ammonium nitrate
27 was found to represent a significant fraction of the total PM2 5 mass (9% to 21%. The fraction
28 lost was higher during summer than during fall. The nitrate obtained from Teflon filter samples
29 was on average 28% lower than that obtained from denuded nylon filters. In these studies
30 samples were extracted immediately after sampling. Therefore, losses that might occur during
31 handling, storage, and equilibration were avoided. Semivolatile organic compounds (SVOC) can
October 1999 3-87 DRAFT-DO NOT QUOTE OR CITE
-------
1 similarly be lost from Teflon filters due to volatilization during or after collection. Such losses
2 can cause the PM2 5 mass to be significantly underestimated. The FRM for PM2 5 will suffer loss
3 of particulate nitrates and SVOC, similar to the losses experienced with other single filter
4 collection systems.
5 Much progress has been made to date in the design of diffusion denuder systems for the
6 measurement and characterization of both the particulate and gaseous phases of semivolatile
7 compounds. Some of the recent research has focused upon reduction in the size of the denuder,
8 optimization of the residence time in the denuder, understanding the effect of diffusion denuders
9 on the positive quartz filter artifact, identifying changes in chemical composition that occur
10 during sampling, determining the effects due to changes in temperature and relative humidity,
11 and identifying possible loses by absorption in impactor coatings.
12 It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
13 compounds. However, for many measurement of suspended particle mass, it is desirable to
14 remove the particle-bound water before determining the mass. In other situations it may be
15 important to know how much of the suspended particle's mass or volume is due to
16 particle-bound water. Calculation and measurements indicate that aerosol water content is
17 strongly dependent upon composition but that liquid water could represent a significant mass
18 fraction of aerosol concentration at relative humidities above 60%.
19 Federal Reference Methods (FRM) for equilibrated mass have been specified for PM10 and
20 PM25. In addition to FRM sampling to determine compliance with PM standards, EPA requires
21 States to conduct speciation sampling primarily to determine source categories and trends. The
22 current speciation samplers include three filters: Teflon for equilibrated mass and elemental
23 analysis, a Nylon filter with a nitric acid denuder to collect nitrate, and a quartz fiber filter for
24 elemental and organic carbon (but without any correction for positive or negative artifacts due to
25 adsorption of volatile organic compounds on the quartz filters or evaporation of semivolatile
26 organic compounds from the collected particles.
27 The U.S. EPA expects that 100 local agency monitoring sites throughout the States will
28 operate continuous PM monitors. However, EPA has not yet provided any guidance regarding
29 appropriate continuous monitoring techniques. All currently available continuous measurements
30 of suspended particle mass share the problem of dealing with semivolatile PM components.
31 In order not to include particle-bound water as part of the mass, the particle bound water must be
October 1999 3-88 DRAFT-DO NOT QUOTE OR CITE
-------
1 removed by heating or dehumidification. However, heating also causes loss of ammonium
2 nitrate and semivolatile organic components. Potential candidates for continuous measurement
3 of mass include the Tapered Element Oscillating Microbalance (TEOM), which determines
4 non-volatile mass by measuring the change in frequency of a vibrating quartz tube with a filter on
5 the end; the Real-Time total Ambient Mass Sampler, RAMS, based on diffusion denuder and
6 TEOM monitor technology, which attempts to determine nonvolatile plus semivolatile mass by
7 using a charcoal impregnated filter (GIF) to collect any semivolatile species lost from the
8 particles during sampling (all gas phase compounds that can be adsorbed by a GIF must be
9 removed from the sampling stream prior to the TEOM monitor); the Continuous Ambient Mass
10 Monitor (CAMM), a technique based upon the measurement of pressure drop increase with
11 particle loading across a membrane filter; and a variety of techniques based on light scattering or
12 measurement of particles mass on a filter by absorption of beta radiation. In addition to
13 continuous mass measurement a number of techniques for continuous measurement of sulfate or
14 semicontinuous measurements of nitrate and elements have been demonstrated. More recently, a
15 new technique, aerosol time-of-fiight mass spectroscopy (ATOFMS), that has demonstrated the
16 ability for real-time measurement of correlated size and composition profiles of individual
17 atmospheric aerosol particles.
18
October 1999 3-89 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Allen, G.; Sioutas, C.; Koutrakis, P.; Reiss, R.; Lurmann, F. W.; Roberts, P. T. (1997) Evaluation of the TEOM
4 method for measurement of ambient particulate mass in urban areas. J. Air Waste Manage. Assoc.
5 47:682-689.
6 Allen, J. O.; Paschkewitz, J. S.; Plummer, E. F.; Lafleur, A. L.; Sarofim, A. F.; Smith, K. A. (1999) Absorption of
7 semi-volatile compounds in oiled impaction substrates: measurement of pyrene absorption. Aerosol Sci.
8 Technol. 30: 16-29.
9 Altshuller, A. P. (1983) Review: natural volatile organic substances and their effect on air quality in the United
10 States. Atmos. Environ. 17:2131-2165.
11 Amdur, M. O. (1957) The influence of aerosols upon the respiratory response of guinea pigs to sulfur dioxide.
12 Am. Ind. Hyg. Assoc. Q. 18: 149-155.
13 Amdur, M. O. (1960) The response of guinea pigs to inhalation of formaldehyde and formic acid alone and with a
14 sodium chloride aerosol. Int. J. Air Pollut. 3: 201-220.
15 Amdur, M. O. (1966) Respiratory absorption data and SO 2 dose-response curves. Arch. Environ. Health
16 12:729-732.
17 Amdur, M. O. (1969) Toxicologic appraisal of particulate matter, oxides of sulfur, and sulfuric acid. J. Air Pollut.
18 Control Assoc. 19: 638-644.
19 Amdur, M. O.; Mead, J. (1958) Mechanics of respiration in unanesthetized guinea pigs. Am. J. Physiol.
20 192: 364-368.
21 Amdur, M. O.; Underbill, D. (1968) The effect of various aerosols on the response of guinea pigs to sulfur dioxide.
22 Arch. Environ. Health 16: 460-468.
23 American Conference of Governmental Industrial Hygienists. (1994) Appendix D: particle size-selective sampling
24 criteria for airborne particulate matter. In: 1994-1995 Threshold limit values for chemical substances and
25 physical agents and biological exposure indices. Cincinnati, OH: American Conference of Governmental
26 Industrial Hygienists; pp. 43-46.
27 Anand, N. K.; McFarland, A. R.; Kihm, K. D.; Wong, F. S. (1992) Optimization of aerosol penetration through
28 transport lines. Aerosol Sci. Technol. 16: 105-112.
29 Andrews, E.; Larson, S. M. (1993) Effect of surfactant layers on the size changes of aerosol particles as a function
30 of relative humidity. Environ. Sci. Technol. 27: 857-865.
31 Appel, B. R. (1993) Atmospheric sample analysis and sampling artifacts. In: Willeke, K.; Baron, P. A., eds. Aerosol
32 measurement: principles, techniques, and applications. New York, NY: Van Nostrand Reinhold; pp. 233-259.
33 Appel, B. R.; Tokiwa, Y.; Haik, M. (1981) Sampling of nitrates in ambient air. Atmos. Environ. 15: 283-289.
34 Appel, B. R.; Tokiwa, Y.; Haik, M.; Kothny, E. L. (1984) Artifact particulate sulfate and nitrate formation on filter
35 media. Atmos. Environ. 18: 409-416.
36 Ayers, G. P.; Keywood, M. D.; Gras, J. L. (1999) TEOM vs. manual gravimetric methods for determination of
37 PM2.5 aerosol mass concentrations. Atmos. Environ. 33: 3717-3721.
38 Ball, J. C.; Hurley, M. D.; Straccia, A. M.; Gierczak, C. A. (1999) Thermal release of nitric oxide from ambient air
39 and diesel particles. Environ. Sci. Technol. 33: 1175-1178.
40 Benner, C. L.; Eatough, D. J.; Eatough, N. L.; Bhardwaja, P. (1991) Comparison of annular denuder and filter pack
41 collection of HNO3 (g), HNO2 (g), SO2 (g), and particulate-phase nitrate, nitrite and sulfate in the south-west
42 desert. Atmos. Environ. Part A 25: 1537-1545.
43 Benner, B. A., Jr.; Wise, S. A.; Currie, L. A.; Klouda, G. A.; Klinedinst, D. B.; Zweidinger, R. B.; Stevens, R. K.;
44 Lewis, C. W. (1995) Distinguishing the contributions of residential wood combustion and mobile source
45 emissions using relative concentrations of dimethylphenanthrene isomers. Environ. Sci. Technol.
46 29:2382-2389.
47 Bergin, M. H.; Ogren, J. A.; Schwartz, S. E.; Mclnnes, L. M. (1997) Evaporation of ammonium nitrate aerosol in a
48 heated nephelometer: implications for field measurements. Environ. Sci. Technol. 31: 2878-2883.
49 Bertoni, G.; Febo, A.; Perrino, C.; Possanzini, M. (1984) Annular active diffuse sampler: a new devise for the
50 collection of organic vapours. Ann. Chim. (Rome) 74: 97-104.
51 Binkowski, F. S.; Ching, J. K. S. (1996) Modeling fine particulate mass and visibility using the EPA regional
52 particulate model. In: Proceedings from the ninth joint conference on applications of air pollution
53 meteorology; Jan-Feb 1996; Atlanta, GA. Boston, MA: American Meteorological Society; pp. 565-569.
October 1999 3-90 DRAFT-DO NOT QUOTE OR CITE
-------
1 Binkowski, F. S.; Shankar, U. (1995) The regional particulate matter model. I. Model description and preliminary
2 results. J. Geophys. Res. [Atmos.] 100: 26,191-26,209.
3 Birch, M. E.; Gary, R. A. (1996) Elemental carbon-based method for occupational monitoring of particulate diesel
4 exhaust: methodology and exposure issues. Analyst 121: 1183-1190.
5 Brorstrom, E.; Grennfelt, P.; Lindskog, A. (1983) The effect of nitrogen dioxide and ozone on the decomposition of
6 particle-associated polycyclic aromatic hydrocarbons during sampling from the atmosphere. Atmos. Environ.
7 17:601-605.
8 Bunz, H.; Koyro, M.; Mohler, O.; Saathoff, H. (1996) Temperature dependence of the redistribution mechanisms
9 for volatile components of atmospheric aerosol particles. J. Aerosol Sci. 27(suppl. 1): S653-S654.
10 Cachier, H.; Bremond, M.-P.; Buat-Menard, P. (1989) Determination of soot carbon with a simple thermal method.
11 Tellus 416:379-390.
12 Carson, P. G.; Johnston, M. V.; Wexler, A. S. (1997) Laser desorption/ionization of ultra fine aerosol particles.
13 Rapid Commun. Mass Spectrom. 11: 993.
14 Chen, B. T. (1993) Instrument calibration. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement: principles,
15 techniques, and applications. New York, NY: Van Nostrand Reinhold; pp. 493-520.
16 Chow, J. C. (1995) Measurement methods to determine compliance with ambient air quality standards for
17 suspended particles. J. Air Waste Manage. Assoc. 45: 320-382.
18 Chow, J. C.; Fujita, E. M.; Watson, J. G.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994) Evaluation of filter-based
19 aerosol measurements during the 1987 Southern California Air Quality Study. Environ. Monit. Assess.
20 30:49-80.
21 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Countess, R. J. (1996) Sources and chemistry of PM 10 aerosol in
22 Santa Barbara County, CA. Atmos. Environ. 30: 1489-1499.
23 Clark, W. E.; Whitby, K. T. (1975) Measurements of aerosols produced by the photochemical oxidation of SO 2 in
24 air. J. Colloid Interface Sci. 51:477-490.
25 Clarke, A. D. (1992) Atmospheric nuclei in the remote free-troposphere. J. Atmos. Chem. 14: 479-488.
26 Code of Federal Regulations. (1991a) Appendix J to Part 50—reference method for the determination of particulate
27 matter as PM10 in the atmosphere. C. F. R. 40: §50.
28 Code of Federal Regulations. (1991b) Ambient air monitoring reference and equivalent methods. C. F. R. 40: §53.
29 Code of Federal Regulations. (1997a) Appendix L to Part 50—reference method for the determination of fine
30 particulate matter as PM2 5 in the atmosphere. C. F. R. 40: §50.
31 Code of Federal Regulations. (1997b) Revised requirements for designation of reference and equivalent methods for
32 PM2 5 and ambient air quality surveillance for particulate matter. C. F. R. 40: §§53, 58.
33 Gotham, W. E.; Bidleman, T. F. (1992) Laboratory investigations of the partitioning of organochlorine compounds
34 between the gas phase and atmospheric aerosols on glass fiber filters. Environ. Sci. Technol. 26: 469-478.
35 Countess, R. J. (1990) Interlaboratory analyses of carbonaceous aerosol samples. Aerosol Sci. Technol.
36 12:114-121.
37 Courtney, W. J.; Shaw, R. W.; Dzubay, T. G. (1982) Precision and accuracy of a P gauge for aerosol mass
38 determinations. Environ. Sci. Technol. 16: 236-239.
39 Coutant, R. W.; Callahan, P. J.; Chuang, J. C.; Lewis, R. G. (1992) Efficiency of silicone-grease-coated denuders
40 for collection of polynuclear aromatic hydrocarbons. Atmos. Environ. Part A 26: 2831-2834.
41 Covert, D. S.; Kapustin, V. N.; Quinn, P. K.; Bates, T. S. (1992) New particle formation in the marine boundary
42 layer. J. Geophys. Res. [Atmos.] 97: 20,581-20,589.
43 Cui, W.; Machir, J.; Lewis , L.; Eatough, D. J.; Eatough, N. L. (1997) Fine particulate organic material at Meadview
44 during the project MOHAVE summer intensive study. J. Air Waste Manage. Assoc. 47: 357-369.
45 Cui, W.; Eatough, D. J.; Eatough, N. L. (1998) Fine particulate organic material in the Los Angeles Basin -1:
46 assessment of the high-volume Brigham Young University organic sampling system, BIG BOSS. J. Air
47 Waste Manage. Assoc. 48: 1024-1037.
48 Davis, C. S.; Fellin, P.; Otson, R. (1987) A review of sampling methods for polyaromatic hydrocarbons in air.
49 JAPCA 37: 1397-1408.
50 Ding, Y.; Cui, W.; Lee, M. L.; Eatough, D. J. (1998a) Fine particulate n-nitroso and nitrite organic compounds in
51 the atmosphere. Appl. Occup. Environ. Hyg. 13: 432-438.
52 Ding, Y.; Lee, M. L.; Eatough, D. J. (1998b) The determination of total nitrite and n-nitroso compound in
53 atmospheric samples. J. Environ. Anal. Chem. 69: 243-255.
54 Durham, J. L.; Wilson, W. E.; Bailey, E. B. (1978) Application of an SO 2-denuder for continuous measurement of
55 sulfur in submicrometric aerosols. Atmos. Environ. 12: 883-886.
October 1999 3-91 DRAFT-DO NOT QUOTE OR CITE
-------
1 Eatough, D. J. (1999) BOSS, the Brigham Young University Organic Sampling System: determination of participate
2 carbonaceous material using diffision denuder sampling technology. In: Lane, D., ed. Gas/phase partition
3 measurements of atmospheric organic compounds. New York, NY: Gordon and Breach Science Publishers;
4 in press.
5 Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. (1993)
6 A multiple-system, multi-channel diffusion denuder sampler for the determination of fine-particulate organic
7 material in the atmosphere. Atmos. Environ. Part A 27: 1213-1219.
8 Eatough, D. J.; Tang, H.; Cui, W.; Machir, J. (1995) Determination of the size distribution and chemical
9 composition of fine particulate semi-volatile organic material in urban environments using diffusion denuder
10 technology. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on particulate air pollution
11 and human mortality and morbidity, part II; January 1994; Irvine, CA. Inhalation Toxicol. 7: 691-710.
12 Eatough, D. J.; Eatough, D. A.; Lewis, L.; Lewis, E. A. (1996) Fine particulate chemical composition and light
13 extinction at Canyonlands National Park using organic particulate material concentrations obtained with a
14 multisystem, multichannel diffusion denuder sampler. J. Geophys. Res. [Atmos.] 101: 19,515-19,531.
15 Eatough, D. J.; Obeidi, F.; Pang, Y.; Ding, Y.; Eatough, N. L.; Wilson, W. E. (1999) Integrated and real-time
16 diffusion denuder sampler for PM25. Atmos. Environ. 33: 2835-2844.
17 Falconer, R. L.; Bidleman, T. F.; Gotham, W. E. (1995) Preferential sorption of non- and
18 mono-ortho-polychlorinated biphenyls to urban aerosols. Environ. Sci. Technol. 29: 1666-1673.
19 Fan, Z.; Kamens, R.; Hu, J. X.; Zhang, J. B.; McDow, S. (1996a) Photostability of nitro-polycyclic aromatic
20 hydrocarbons on combustion soot particles in sunlight. Environ. Sci. Technol. 30: 1358-1364.
21 Fan, Z.; Kamens, R. M.; Zhang, J.; Hu, J. (1996b) Ozone—nitrogen dioxide—NPAH heterogeneous soot particle
22 reactions and modeling NPAH in the atmosphere. Environ. Sci. Technol. 30: 2821-2827.
23 Fang, C. P.; McMurry, P. H.; Marple, V. A.; Rubow, K. L. (1991) Effect of flow-induced relative humidity changes
24 on size cuts for sulfuric acid droplets in the microorifice uniform deposit impactor (MOUDI). Aerosol Sci.
25 Technol. 14: 266-277.
26 Federal Register. (1987) Revisions to the national ambient air quality standards for particulate matter. F. R. (July 1)
27 52:24,634-24,669.
28 Federal Register. (1997) National ambient air quality standards for particulate matter; final rule. F. R. (July 18)
29 62:38,652-38,752.
30 Federal Register. (1998) National ambient air quality standards for particulate matter and revised requirements for
31 designation of reference and equivalent methods for PM 2 5 and ambient air quality surveillance for particulate
32 matter; correction. F. R. (February 17) 63: 7710-7718.
33 Finn, D.; Rumburg, B.; Claiborn, C.; Bamesberger, W.; Norris, G.; Koenig, J.; Larson, T. (1999) Sampling artifacts
34 from the use of Teflon filters and denuder tubes during the measurement of airborne carbonaceous particulate
35 matter. J. Air Waste Manage. Assoc.: submitted.
36 Fitz, D. R. (1990) Reduction of the positive organic artifact on quartz filters. Aerosol. Sci. Technol. 12: 142-148.
37 Friedlander, S. K. (1970) The characterization of aerosols distributed with respect to size and chemical composition.
38 J. Aerosol Sci. 1:295-307.
39 Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John Wiley &
40 Sons.
41 Friedlander, S. K.; Yeh. E. K. (1998) The submicron atmospheric aerosol as a carrier of reactive chemical species:
42 case of peroxides. Appl. Occup. Environ. Hyg. 13: 416-420.
43 Gallorini, M. (1995) Trace element monitoring in atmospheric pollution processes by neutron activation analysis.
44 Microchem. J. 51: 127-137.
45 Gard, E.; Mayer, J. E.; Morrical, B. D.; Dienes, T.; Fergenson, D. P.; Prather, K. A. (1997) Real-time analysis of
46 individual atmospheric aerosol particles: design and performance of a portable ATOFMS. Anal. Chem.
47 69:4083-4091.
48 Ge, Z.; Wexler, A. S.; Johnston, M. V. (1998) Laser desorption/ionization of single ultrafine multicomponent
49 aerosols. Environ. Sci. Technol. 32: 3218-3223.
50 Goss, K.-U.; Eisenreich, S. J. (1997) Sorption of volatile organic compounds to particles from a combustion source
51 at different temperatures and relative humidities. Atmos. Environ. 31: 2827-2834.
52 Griffin, R. J.; Cocker, D. R., Ill; Seinfeld, J. H. (1999) Incremental aerosol reactivity: application to aromatic and
53 biogenic hydrocarbons. Environ. Sci. Technol. 33: 2403-2408.
54 Grosjean, D. (1983) Polycyclic aromatic hydrocarbons in Los Angeles air from samples collected on Teflon, glass
55 and quartz filters. Atmos. Environ. 17: 2565-2573.
October 1999 3-92 DRAFT-DO NOT QUOTE OR CITE
-------
1 Grosjean, D.; Parmar, S. S. (1990) Interferences from aldehydes and peroxyacetyl nitrate when sampling urban air
2 organic acids on alkaline traps. Environ. Sci. Technol. 24: 1021-1026.
3 Grosjean, D.; Seinfeld, J. H. (1989) Parameterization of the formation potential of secondary organic aerosols.
4 Atmos. Environ. 23: 1733-1747.
5 Gundel, L. A.; Lane, D. A. (1999) Sorbent-coated diffusion denuders for direct measurement of gas/particle
6 partitioning by semi-volatile organic compounds. In: Lane, D. A., eds. Gas and Particle Phase Measurements
7 of Atmospheric Organic Compounds. Newark, NJ: Gordon and Breach. (Advances in environmental process
8 control technologies, v. 2).
9 Gundel, L. A.; Lee, V. C.; Mahanama, K. R. R.; Stevens, R. K.; Daisey, J. M. (1995a) Direct determination of the
10 phase distributions of semi-volatile polycyclic aromatic hydrocarbons using annular denuders. Atmos.
11 Environ. 29: 1719-1733.
12 Gundel, L. A.; Mahanama, K. R. R.; Daisey, J. M. (1995b) Semivolatile and particulate polycyclic aromatic
13 hydrocarbons in environmental tobacco smoke: cleanup, speciation, and emission factors. Environ. Sci.
14 Technol. 29: 1607-1614.
15 Haagen-Smit, A. J. (1952) Chemistry and physiology of Los Angeles smog. Ind. Eng. Chem. 44: 1342-1346.
16 Hansen, A. D. A.; McMurry, P. H. (1990) An intercomparison of measurements of aerosol elemental carbon during
17 the 1986 carbonaceous species method comparison study. J. Air Waste Manage. Assoc. 40: 894-895.
18 Hansen, A. D. A.; Rosen, H.; Novakov, T. (1984) The aethalometer - an instrument for the real-time measurement
19 of optical absorption by aerosol particles. Sci. Total Environ. 36: 191-196.
20 Henry, R. C.; Lewis, C. W.; Hopke, P. K.; Williamson, H. J. (1984) Review of receptor model fundamentals.
21 Atmos. Environ. 18:1507-1515.
22 Hering, S.; Cass, G. (1999) The magnitude of bias in the measurement of PM 2 5 arising from volatilization of
23 particulate nitrate from Teflon filters. J. Air Waste Manage. Assoc. 49: 725-733.
24 Hering, S. V.; Appel, B. R.; Cheng, W.; Salaymeh, F.; Cadle, S. H.; Mulawa, P. A.; Cahill, T. A.; Eldred, R. A.;
25 Surovik, M.; Fitz, D.; Howes, J. E.; Knapp, K. T.; Stockburger, L.; Turpin, B. J.; Huntzicker, J. J.; Zhang,
26 X.-Q.; McMurry, P. H. (1990) Comparison of sampling methods for carbonaceous aerosols in ambient air.
27 Aerosol Sci. Technol. 12: 200-213.
28 Hidy, G. M.; Venkataraman, C. (1996) The chemical mass balance method for estimating atmospheric particle
29 sources in southern California. Chem. Eng. Commun. 151: 187.
30 Hitzenberger, R.; Dusek, U.; Berner, A. (1996) Black carbon measurements using an integrating sphere. J. Geophys.
31 Res. [Atmos.] 101: 19,601-19,606.
32 Hitzenberger, R.; Berner, A.; Dusek, U.; Alabashi, R. (1997) Humidity-dependent growth of size-segregated aerosol
33 samples. Aerosol Sci. Technol. 27: 116-130.
34 Hitzenberger, R.; Jennings, S. G.; Larson, S. M.; Dillner, A.; Cachier, H.; Galambos, Z.; Roue, A.; Spain, T. G.
35 (1999) Intercomparison of measurement methods for black carbon aerosols. Atmos. Environ. 33: 2823-2833.
36 Hoffmann, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.; Seinfeld, J. H. (1997) Formation
37 of organic aerosols from the oxidation of biogenic hydrocarbons. J. Atmos. Chem. 26: 189-222.
38 Izumi, K.; Fukuyama, T. (1990) Photochemical aerosol formation from aromatic hydrocarbons in the presence of
39 NOX. Atmos. Environ. Part A 24: 1433-1441.
40 Jaklevic, J. M.; Gatti, R. C.; Goulding, F. S.; Loo, B. W. (1981) A p-gauge method applied to aerosol samples.
41 Environ. Sci. Technol. 15: 680-686.
42 Jang, M.; Kamens, R. M. (1998) A thermodynamic approach for modeling partitioning of semivolatile organic
43 compounds on atmospheric particulate matter: humidity effects. Environ. Sci. Technol. 32: 1237-1243.
44 Jang, M.; Kamens, R. M. (1999) Newly characterized products and composition of secondary aerosols from the
45 reaction of cc-pinene with ozone. Atmos. Environ. 33: 459-474.
46 Jang, M.; Kamens, R. M.; Leach, K. B.; Strommen, M. R. (1997) A thermodynamic approach using group
47 contribution methods to model the partitioning of semivolatile organic compounds on atmospheric particulate
48 matter. Environ. Sci. Technol. 31: 2805-2811.
49 Jenkins, B. M.; Jones, A. D.; Turn, S. Q.; Williams, R. B. (1996) Particle concentrations, gas-particle partitioning,
50 and species intercorrelations for polycyclic aromatic hydrocarbons (PAH) emitted during biomass burning.
51 Atmos. Environ. 30: 3825-3835.
52 John, W.; Wall, S. M.; Ondo, J. L. (1988) A new method for nitric acid and nitrate aerosol measurement using the
53 dichotomous sampler. Atmos. Environ. 22: 1627-1635.
54 Kalberer, M.; Ammann, M.; Gaggeler, H. W.; Baltensperger, U. (1999) Adsorption of NO 2 on carbon aerosol
55 particles in the low ppb range. Atmos. Environ. 33: 2815-2822.
October 1999 3-93 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kamens, R. M.; Coe, D. L. (1997) A large gas-phase stripping device to investigate rates of PAH evaporation from
2 airborne diesel soot particles. Environ. Sci. Technol. 31: 1830-1833.
3 Kamens, R.; Odum, J.; Fan, Z.-H. (1995) Some observations on times to equilibrium for semivolatile polycyclic
4 aromatic hydrocarbons. Environ. Sci. Technol. 29: 43-50.
5 Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of cc-pinene and ozone
6 using a gas-phase kinetics-aerosol partitioning model. Environ. Sci. Technol. 33: 1430-1438.
7 Kao, A. S.; Friedlander, S. K. (1995) Frequency distributions of PM 10 chemical components and their sources.
8 Environ. Sci. Technol. 29: 19-28.
9 Keeler, G. J.; Pirrone, N. (1996) Atmospheric transport and deposition of trace elements to Lake Erie from urban
10 areas. Water Sci. Technol. 33: 259-265.
11 Keeler, G. J.; Brachaczek, W. W.; Gorse, R. A., Jr.; Japar, S. M.; Pierson, W. R. (1988) Effect of ambient humidity
12 on dichotomous sampler coarse/fine ratios. Atmos. Environ. 22: 1715-1720.
13 Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
14 segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
15 Environ. 33: 2907-2913.
16 Khalil, M. A. K.; Rasmussen, R. A. (1992) Forest hydrocarbon emissions: relationships between fluxes and ambient
17 concentrations. J. Air Waste Manage. Assoc. 42: 810-813.
18 Kim, Y. J.; Boatman, J. F.; Gunter, R. L.; Wellman, D. L.; Wilkison, S. W. (1993) Vertical distribution of
19 atmospheric aerosol size distribution over south-central New Mexico. Atmos. Environ. Part A
20 27: 1351-1362.
21 Kittelson, D. B. (1998) Engines and nanoparticles: a review. J. Aerosol Sci. 29: 575-588.
22 Klouda, G. A.; Klinedinst, D. B.; Steel, E. B.; Benner, B. A., Jr.; Parish, H. J. (1996) Exploring a method to
23 produce an urban dust particle filter standard. J. Aerosol Sci. 27(suppl. 1): S351-S352.
24 Koutrakis, P.; Wolfson, J. M.; Spengler, J. D. (1988) An improved method for measuring aerosol strong acidity:
25 results from a nine-month study in St. Louis, Missouri and Kingston, Tennessee. Atmos. Environ.
26 22: 157-162.
27 Koutrakis, P.; Sioutas, C.; Ferguson, S. T.; Wolfson, J. M.; Mulik, J. D.; Burton, R. M. (1993) Development and
28 evaluation of a glass honeycomb denuder/filter pack system to collect atmospheric gases and particles.
29 Environ. Sci. Technol. 27: 2497-2501.
30 Koutrakis, P.; Wang, P.-Y.; Wolfson, J. M.; Sioutas, C., inventors. (1996) Method and apparatus to measure
31 particulate matter in gas. U.S. patent 5,571,945. November 5. Available: www.patents.ibm.com.
32 Krieger, M. S.; Hites, R. A. (1992) Diffusion denuder for the collection of semivolatile organic compounds.
33 Environ. Sci. Technol. 26: 1551-1555.
34 Lammel, G.; Novakov, T. (1995) Water nucleation properties of carbon black and diesel soot particles. Atmos.
35 Environ. 29:813-823.
36 Landsberger, S.; Wu, D. (1993) Improvement of analytical sensitivities for the determination of antimony, arsenic,
37 cadmium, indium, iodine, molybdenum, silicon and uranium in airborne particulate matter by epithermal
38 neutron activation analysis. J. Radioanal. Nucl. Chem. 167: 219-225.
39 Landsberger, S.; Wu, D.; Vermette, S. J.; Cizek, W. (1997) Intercomparison of IAEA airborne particulate matter
40 reference material. J. Radioanal. Nucl. Chem. 215: 117-127.
41 Lane, D. A.; Gundel, L. A. (1996) Gas and particle sampling of airborne polycyclic aromatic compounds. In:
42 15th International symposium on polycyclic aromatic compounds: second biennial meeting of International
43 Society for Polycyclic Aromatic Compounds; September 1995; Belgirate, Italy. Polycyclic Aromat. Compd.
44 9:67-73.
45 Lavanchy, V. M. H.; Gaggeler, H. W.; Nyeki, S.; Baltensperger, U. (1999) Elemental carbon (EC) and black carbon
46 (BC) measurements with a thermal method and an aethalometer at the high-alpine research station
47 Jungfraujoch. Atmos. Environ. 33: 2759-2769.
48 Lawrence, J. E.; Koutrakis, P. (1994) Measurement of atmospheric formic and acetic acids: methods evaluation and
49 results from field studies. Environ. Sci. Technol. 28: 957-964.
50 Lawrence, J.; Koutrakis, P. (1996a) Measurement and speciation of gas and particulate phase organic acidity in an
51 urban environment: 1. analytical. J. Geophys. Res. [Atmos.] 101: 9159-9169.
52 Lawrence, J.; Koutrakis, P. (1996b) Measurement and speciation of gas and particulate phase organic acidity in an
53 urban environment: 2. speciation. J. Geophys. Res. [Atmos.] 101: 9171-9184.
54 Lawson, D. R. (1990) The Southern California Air Quality Study. J. Air Waste Manage. Assoc. 40: 156-165.
October 1999 3-94 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lee, W.-M. G.; Shih, P.-M.; Wang, C.-S. (1997) The influence of relative humidity on the size of atmospheric
2 aerosol. J. Environ. Sci. Health Part A: Environ. Sci. Eng. Toxic Hazard. Subst. Control 32: 1085-1097.
3 Li, W.; Montassier, N.; Hopke, P. K. (1992) A system to measure the hygroscopicity of aerosol particles. Aerosol
4 Sci. Technol. 17: 25-35.
5 Li, X. Y.; Gilmour, P. S.; Donaldson, K.; MacNee, W. (1997) In vivo and in vitro proinflammatory effects of
6 particulate air pollution (PM10). In: Driscoll, K. E.; Oberdorster, G., eds. Proceedings of the sixth
7 international meeting on the toxicology of natural and man-made fibrous and non-fibrous particles;
8 September 1996; Lake Placid, NY. Environ. Health Perspect. Suppl. 105(5): 1279-1283.
9 Liang, Z.; Chan, C. K. (1997) A fast technique for measuring water activity of atmospheric aerosols. Aerosol Sci.
10 Technol. 26: 255-268.
11 Liang, C.; Pankow, J. F.; Odum, J. R.; Seinfeld, J. H. (1997) Gas/particle partitioning of semivolatile organic
12 compounds to model inorganic, organic, and ambient smog aerosols. Environ. Sci. Technol. 31: 3086-3092.
13 Lindskog, A.; Brorstrom-Lunden, E. (1987) Chemical transformation of PAH on airborne particles by exposure to
14 NO2 during sampling: a comparison between 2 filter media. Sci. Total Environ. 61: 51-57.
15 Lison, D.; Lardot, C.; Huaux, F.; Zanetti, G.; Fubini, B. (1997) Influence of particle surface area on the toxicity of
16 insoluble manganese dioxide dusts. Arch. Toxicol. 71: 725-729.
17 Lundgren, D. A.; Burton, R. M. (1995) Effect of particle size distribution on the cut point between fine and coarse
18 ambient mass fractions. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on particulate air
19 pollution and human mortality and morbidity; January 1994; Irvine, CA. Inhalation Toxicol. 7: 131-148.
20 Mahanama, K. R. R.; Gundel, L. A.; Daisey, J. M. (1994) Selective fluorescence detection of polycyclic aromatic
21 hydrocarbons in environmental tobacco smoke and other airborne particles. Int. J. Environ. Anal. Chem.
22 56:289-309.
23 Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in particle
24 concentration and optical extinction in the United States. J. Geophys. Res. [Atmos.] 99: 1347-1370.
25 Mclnnes, L. M.; Quinn, P. K.; Covert, D. S.; Anderson, T. L. (1996) Gravimetric analysis, ionic composition, and
26 associated water mass of the marine aerosol. Atmos. Environ. 30: 869-884.
27 McLaren, R.; Singleton, D. L. (1996) Real-world measurements of exhaust and evaporative emissions in the Cassair
28 Tunnel predicted by chemical mass balance modeling. Environ. Sci. Technol. 30: 3001.
29 McMurry, P. H. (1999) A review of atmospheric aerosol measurements. Prepared for the NARSTO assessment.
30 McMurry, P. H.; Friedlander, S. K. (1979) New particle formation in the presence of an aerosol. Atmos. Environ.
31 13:1635-1651.
32 McMurry, P. H.; Stolzenburg, M. R. (1989) On the sensitivity of particle size to relative humidity for Los Angeles
33 aerosols. Atmos. Environ. 23: 497-507.
34 Meyer, A. S., Jr.; Boyd, C. M. (1959) Determination of water by titration with coulometrically generated Karl
35 Fischer reagent. Anal. Chem. 31: 215-219.
36 Meyer, M. B.; Rupprecht, E. G. (1996) Particulate matter sampling methods: the importance of standardization.
37 J. Aerosol. Sci. 27(suppl. 1): S349-S350.
38 Meyer, M. B.; Rupprecht, E.; Patashnick, H. (1995) Considerations for the sampling and measurement of ambient
39 particulate matter. In: Particulate matter: health and regulatory issues: proceedings of an international
40 specialty conference; April; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association;
41 pp. 204-210. (A&WMA publication VIP-49).
42 Miller, F. J.; Gardner, D. E.; Graham, J. A.; Lee, R. E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
43 considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc. 29: 610-615.
44 Morawska, L.; Thomas, S. B.; Bofinger, N. D.; Neale, D.; Wainwright, D. (1996) Characterisation of airborne
45 particulate pollution. Presented at: 13th International clean air conference proceedings.
46 Morawska, L.; Bofinger, N. D.; Kocis, L.; Nwankwoala, A. (1998a) Submicrometer and supermicrometer particles
47 from diesel vehicle emissions. Environ. Sci. Technol. 32: 2033-2042.
48 Morawska, L.; Thomas, S.; Bofinger, N.; Wainwright, D.; Neale, D. (1998b) Comprehensive characterization of
49 aerosols in a subtropical urban atmosphere: particle size distribution and correlation with gaseous pollutants.
50 Atmos. Environ. 32: 2467-2478.
51 Mukerjee, S.; Ellenson, W. D.; Lewis, R. G.; Stevens, R. K.; Somerville, M. C.; Shadwick, D. S.; Willis, R. D.
52 (1997) An environmental scoping study in the lower Rio Grande Valley of Texas—III. residential
53 micro environmental monitoring for air, house dust, and soil. Environ. Int. 23: 657-673.
54 Muzyka, V.; Veimer, S.; Shmidt, N. (1998) Particle-bound benzene from diesel engine exhaust. Scand. J. Work
55 Environ. Health 24: 481-485.
October 1999 3-95 DRAFT-DO NOT QUOTE OR CITE
-------
1 National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
2 National Academy Press. 3v.
3 Noble, C. A.; Prather, K. A. (1996) Real-time measurement of correlated size and composition profiles of individual
4 atmospheric aerosol particles. Environ. Sci. Technol. 30: 2667-2680.
5 Norris, G. (1998) Air pollution and the exacerbation of asthma in an arid, western, US city [dissertation]. Seattle,
6 WA: University of Washington. Available from: University Microfilms International, Ann Arbor, MI;
7 publication no. 9836228.
8 Norris, G.; Larson, T.; Koenig, J.; Claiborn, C.; Sheppard, L.; Finn, D. (1999) Asthma aggravation, combustion, and
9 stagnant air. Thorax: accepted.
10 Obeidi, F. D.; Eatough, D. J. (1999) Continuous measurement of semi-volatile fine particulate mass in Provo, UT.
11 J. Aerosol Sci.: submitted.
12 Oberdorster, G.; Ferin, J.; Gelein, R.; Soderholm, S. C.; Finkelstein, J. (1992) Role of the alveolar macrophage in
13 lung injury: studies with ultrafine particles. Environ. Health Perspect. 97: 193-199.
14 Oberdorster, G.; Ferin, J.; Lehnert, B. E. (1994) Correlation between particle size, in vivo particle persistence, and
15 lung injury. Environ. Health Perspect. 102(suppl. 5): 173-179.
16 Oberdorster, G.; Gelein, R. M.; Ferin, J.; Weiss, B. (1995) Association of particulate air pollution and acute
17 mortality: involvement of ultrafine particles? In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the
18 colloquium on particulate air pollution and human mortality and morbidity; January 1994; Irvine, CA.
19 Inhalation Toxicol. 7:111-124.
20 Oberdorster, G.; Finkelstein, J. N.; Johnston, C.; Gelein, R.; Cox, C.; Baggs, R.; Elder, A. (1999) Ultrafine particles
21 as inducers of acute lung injury: mechanisms and correlation with age and disease. Cambridge, MA: Health
22 Effects Institute: submitted.
23 Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. (1996) Gas/particle partitioning
24 and secondary organic aerosol yields. Environ. Sci. Technol. 30: 2580-2585.
25 Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. (1997a) Aromatics,
26 reformulated gasoline, and atmospheric organic aerosol formation. Environ. Sci. Technol. 31: 1890-1897.
27 Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. (1997b) The atmospheric
28 aerosol-forming potential of whole gasoline vapor. Science (Washington, DC) 276: 96-99.
29 Ohta, S.; Okita, T. (1990) A chemical characterization of atmospheric aerosol in Sapporo. Atmos. Environ. Part A
30 24A: 815-822.
31 Ohta, S.; Hori, M.; Yamagata, S.; Murao, N. (1998) Chemical characterization of atmospheric fine particles in
32 Sapporo with determination of water content. Atmos. Environ. 32: 1021-1025.
33 Ondov, J. M. (1996) Particulate tracers for source attribution: potential for application to California's San Joaquin
34 Valley. J. Aerosol Sci. 27(suppl. 1): S687-S688.
35 Ondov, J. M.; Kelly, W. R.; Holland, J. Z.; Lin, Z. C.; Wight, S. A. (1992) Tracing fly ash emitted from a coal-fired
36 power plant with enriched rare-earth isotopes: an urban scale test. Atmos. Environ. Part B 26B: 453-462.
37 Ormstad, H.; Johansen, B. V.; Gaarder, P. I. (1998) Airborne house dust particles and diesel exhaust particles as
38 allergen carriers. Clin. Exp. Allergy 28: 702-708.
39 Pandis, S. N.; Paulson, S. E.; Seinfeld, J. H.; Flagan, R. C. (1991) Aerosol formation in the photooxidation of
40 isoprene and p-pinene. Atmos. Environ. Part A 25: 997-1008.
41 Pandis, S. N.; Harley, R. A.; Cass, G. R.; Seinfeld, J. H. (1992) Secondary organic aerosol formation and transport.
42 Atmos. Environ. Part A 26: 2269-2282.
43 Pankow, J. F. (1994a) An absorption model of gas/particle partitioning of organic compounds in the atmosphere.
44 Atmos. Environ. 28: 185-188.
45 Pankow, J. F. (1994b) An absorption model of the gas/aerosol partitioning involved in the formation of secondary
46 organic aerosol. Atmos. Environ. 28: 189-193.
47 Pankow, J. F.; Bidleman, T. F. (1991) Effects of temperature, TSP and per cent non-exchangeable material in
48 determining the gas-particle partitioning of organic compounds. Atmos. Environ. Part A 25: 2241-2249.
49 Pankow, J. F.; Storey, J. M. E.; Yamasaki, H. (1993) Effects of relative humidity on gas/particle partitioning of
50 semivolatile organic compounds to urban particulate matter. Environ. Sci. Technol. 27: 2220-2226.
51 Parmar, S. S.; Grosjean, D. (1990) Laboratory tests of KI and alkaline annular denuders. Atmos. Environ. Part A
52 24:2695-2698.
53 Pellizzari, E. D.; Krost, K. J. (1984) Chemical transformations during ambient air sampling for organic vapors.
54 Anal. Chem. 56: 1813-1819.
October 1999 3-96 DRAFT-DO NOT QUOTE OR CITE
-------
1 Penner, J. E.; Novakov, K. (1996) Carbonaceous particles in the atmosphere: a historical perspective to the fifth
2 international conference on carbonaceous particles in the atmosphere. J. Geophys. Res. [Atmos.]
3 101: 19,373-19,378.
4 Pennise, D. M.; Kamens, R. M. (1996) Atmospheric behavior of polychlorinated dibenzo- />-dioxins and
5 dibenzofurans and the effect of combustion temperature. Environ. Sci. Technol. 30: 2832-2842.
6 Peters, A.; Wichmann, H. E.; Tuch, T.; Heinrich, J.; Heyder, J. (1997) Respiratory effects are associated with the
7 number of ultrafine particles. Am. J. Respir. Crit. Care Med. 155: 1376-1383.
8 Petzold, A.; Niessner, R. (1995) Method comparison study on soot-selective techniques. Mikrochim. Acta
9 117:215-237.
10 Petzold, A.; Kopp, C.; Niessner, R. (1997) The dependence of the specific attenuation cross-section on black carbon
11 mass fraction and particle size. Atmos. Environ. 31: 661-672.
12 Pilinis, C.; Seinfeld, J. H.; Grosjean, D. (1989) Water content of atmospheric aerosols. Atmos. Environ.
13 23: 1601-1606.
14 Pinto, J. P.; Pleasant, M.; Kellogg, R.; Torok, Sz.; Stiles, D.; Willis, R. D.; Koutrakis, P.; Allen, G. A. (1995)
15 Chemical characterization of the ambient aerosol in Philadelphia. Presented at: Particulate matter: health and
16 regulatory issues: an A&WMA international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA:
17 Air & Waste Management Association; submitted abstract.
18 Pinto, J. P.; Stevens, R. K.; Willis, R. D.; Kellogg, R.; Mamane, Y.; Novak, J.; Santroch, J.; Benes, I.; Leni cek, J.;
19 Bures, V. (1998) Czech air quality monitoring and receptor modeling study. Environ. Sci. Technol.
20 32: 843-854.
21 Pirrone, N.; Keeler, G. J. (1996) The Rouge River watershed pollution by trace elements: atmospheric depositions
22 and emission sources. Water Sci. Technol. 33: 267-275.
23 Ranade, M. B. (Arun); Woods, M. C.; Chen, F.-L.; Purdue, L. J.; Rehme, K. A. (1990) Wind tunnel evaluation of
24 PM10 samplers. Aerosol Sci. Technol. 13: 54-71.
25 Reist, P. C. (1984) Introduction to aerosol science. New York, NY: Macmillan Publishing Company.
26 Risse, U.; Flammenkamp, E.; Kettrup, A. (1996) Determination of aromatic hydrocarbons in air using diffusion
27 denuders. Fresenius J. Anal. Chem. 356: 390-395.
28 Roberts, R.; Corkill, J. (1998) Grass seed field smoke and its impact on respiratory health. J. Environ. Health
29 60: 10.
30 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine organic
31 aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 25: 1112-1125.
32 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
33 aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
34 Technol. 27:636-651.
35 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
36 aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks. Environ.
37 Sci. Technol. 27: 1892-1904.
38 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine organic
39 aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
40 27:2700-2711.
41 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993d) Sources of fine organic
42 aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27: 2736-2744.
43 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Sources of fine organic
44 aerosol. 6. Cigarette smoke in the urban atmosphere. Environ. Sci. Technol. 28: 1375-1388.
45 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997a) Sources of fine organic
46 aerosol. 7. Hot asphalt roofing tar pot fumes. Environ. Sci. Technol. 31: 2726-2730.
47 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997b) Sources of fine organic
48 aerosol. 8. Boilers burning no. 2 distillate fuel oil. Environ. Sci. Technol. 31: 2731-2737.
49 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1998) Sources of fine organic
50 aerosol. 9. Pine, oak, and synthetic log combustion in residential fireplaces. Environ. Sci. Technol. 32: 13-22.
51 Saxena, P.; Hildemann, L. M. (1996) Water-soluble organics in atmospheric particles: a critical review of the
52 literature and applications of thermodynamics to identify candidate compounds. J. Atmos. Chem. 24: 57-109.
53 Saxena, P.; Hildemann, L. M.; McMurry, P. H.; Seinfeld, J. H. (1995) Organics alter hygroscopic behavior of
54 atmospheric particles. J. Geophys. Res. [Atmos.] 100: 18,755-18,770.
October 1999 3-97 DRAFT-DO NOT QUOTE OR CITE
-------
1 Schappi, G. F.; Taylor, P. E.; Pain, M. C. F.; Cameron, P. A.; Dent, A. W.; Staff, I. A.; Suphioglu, C. (1999)
2 Concentrations of major grass group 5 allergens in pollen grains and atmospheric particles: implications for
3 hay fever and allergic asthma sufferers sensitized to grass pollen allergens. Clin. Exp. Allergy 29: 633-641.
4 Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R. (1996) Source apportionment of
5 airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
6 Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change.
7 New York, NY: John Wiley & Sons, Inc.
8 Sioutas, C.; Koutrakis, P.; Wolfson, J. (1994) Particle losses in glass honeycomb denuder samplers. Aersol Sci.
9 Technol. 21: 137.
10 Sioutas, C.; Koutrakis, P.; Wolfson, J. M.; Azaroff, L. S.; Mulik, J. D. (1996a) Field evaluation of a glass
11 honeycomb denuder/filter pack system to collect atmospheric gases and particles. In: Gay, B. W.; Fuerst,
12 R. G.; Jayanty, R. K. M., eds. Measurement of toxic and related air pollutants: proceedings of a conference;
13 May, 1994; Durham, NC. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of
14 Research and Development; report no. EPA/600/R-94/136; pp. 426-435.
15 Sioutas, C.; Wang, P. Y.; Ferguson, S. T.; Koutrakis, P.; Mulik, J. D. (1996b) Laboratory and field evaluation of an
16 improved glass honeycomb denuder/filter pack sampler. Atmos. Environ. 30: 885-895.
17 Sioutas, C.; Koutrakis, P.; Wang, P.-Y.; Babich, P.; Wolfson, J. M. (1999) Experimental investigation of pressure
18 drop with particle loading in Nuclepore filters. Aerosol Sci. Technol. 30: 71-83.
19 Solomon, P. A.; Salmon, L. G.; Fall, T.; Cass, G. R. (1992) Spatial and temporal distribution of atmospheric nitric
20 acid and particulate nitrate concentrations in the Los Angeles area. Environ. Sci. Technol. 26: 1596-1601.
21 Speer, R. E.; Barnes, H. M.; Brown, R. (1997) An instrument for measuring the liquid water content of aerosols.
22 Aerosol Sci. Technol. 27: 50-61.
23 Strommen, M. R.; Kamens, R. M. (1997) Development and application of a dual-impedance radial diffusion model
24 to simulate the partitioning of semivolatile organic compounds in combustion aerosols. Environ. Sci.
25 Technol. 31:2983-2990.
26 Suarez, A. E.; Caffrey, P. F.; Ondov, J. M.; Thaung, K. C. (1996) Size distribution of Ir-tagged soot from diesel
27 sanitation trucks. J. Aerosol Sci. 27(suppl. 1): S697-S698.
28 Tang, I. N. (1980) Deliquescence properties and particle size change of hygroscopic aerosols. In: Willeke, K., ed.
29 Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann Arbor Science
30 Publishers, Inc.; pp. 153-167.
31 Tang, H.; Lewis, E. A.; Eatough, D. J.; Burton, R. M.; Farber, R. J. (1994) Determination of the particle size
32 distribution and chemical composition of semi-volatile organic compounds in atmospheric fine particles with
33 a diffusion denuder sampling system. Atmos. Environ. 28: 939-947.
34 Tsai, C.-J.; Cheng, Y.-H. (1996) Comparison of two ambient beta gauge PM 10 samplers. J. Air Waste Manage.
35 Assoc. 46: 142-147.
36 Tsai, C.-J.; Huang, H.-Y. (1995) Atmospheric aerosol sampling by an annular denuder system and a high-volume
37 PM10 sampler. Environ. Int. 21: 283-291.
38 Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a
39 descriptive analysis of organic and elemental carbon concentrations. Atmos. Environ. Part A 25: 207-215.
40 Turpin, B. J.; Gary, R. A.; Huntzicker, J. J. (1990) An in situ, time-resolved analyzer for aerosol organic and
41 elemental carbon. Aerosol Sci. Technol. 12: 161-171.
42 Turpin, B. J.; Huntzicker, J. J.; Larson, S. M.; Cass, G. R. (1991) Los Angeles summer midday particulate carbon:
43 primary and secondary aerosol. Environ. Sci. Technol. 25: 1788-1793.
44 Turpin, B. J.; Liu, S.-P.; Podolske, K. S.; Gomes, M. S. P.; Eisenreich, S. J.; McMurry, P. H. (1993) Design and
45 evaluation of a novel diffusion separator for measuring gas/particle distributions of semivolatile organic
46 compounds. Environ. Sci. Technol. 27: 2441-2449.
47 Turpin, B. J.; Huntzicker, J. J.; Hering, S. V. (1994) Investigation of organic aerosol sampling artifacts in the
48 Los Angeles basin. Atmos. Environ. 28: 3061-3071.
49 U.S. Environmental Protection Agency. (1982) Air quality criteria for particulate matter and sulfur oxides. Research
50 Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
51 Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
52 U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
53 NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
54 Available from: NTIS, Springfield, VA; PB96-168224.
October 1999 3-98 DRAFT-DO NOT QUOTE OR CITE
-------
1 Underbill, D. W. (1999) Aerosol synergism and hydrate formation - a possible connection. Inhalation Toxicol.:
2 submitted.
3 Van Vaeck, L.; Van Cauwenberghe, K.; Janssens, J. (1984) The gas-particle distribution of organic aerosol
4 constituents: measurements of the volatilisation artefact in hi-vol cascade impactor sampling. Atmos.
5 Environ. 18:417-430.
6 Vega, E.; Garcia, L; Apam, D.; Ruiz, M.; Barbiaux, M. (1997) Application of a chemical mass balance receptor
7 model to respirable particulater matter in Mexico City. J. Air Waste Manage. Assoc. 47: 524.
8 Veltkamp, P. R.; Hansen, K. J.; Barkley, R. M.; Sievers, R. E. (1996) Principal component analysis of summertime
9 organic aerosols at Niwot Ridge, Colorado. J. Geophys. Res. [Atmos.] 101: 19,495-19,504.
10 Venkataraman, C.; Lyons, J. M.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic hydrocarbons
11 and elemental carbon. 1. Sampling, measurement methods, and source characterization. Environ. Sci.
12 Technol. 28: 555-562.
13 Wang, P. Y. (1997) Continuous aerosol mass measurement by flow obstruction. Boston, MA: Harvard University,
14 School of Public Health.
15 Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
16 Technol. 8: 157-172.
17 Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994)
18 Chemical mass balance source apportionment of PM 10 during the Southern California Air Quality Study.
19 Aerosol Sci. Technol. 21: 1-36.
20 Wedding, J. B.; Weigand, M. A. (1993) An automatic particle sampler with beta gauging. Air Waste 43: 475-479.
21 Weingartner, E.; Burtscher, H.; Baltensperger, U. (1997) Hygroscopic properties of carbon and diesel soot particles.
22 Atmos. Environ. 31:2311 -2327.
23 Weschler, C. J.; Shields, H. C. (1999) Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
24 Environ. 33:2301-2312.
25 Wexler, A. S.; Sarangapani, R. (1998) Particles do not increase vapor deposition in human airways. J. Aerosol Sci.
26 29: 197-204.
27 Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
28 Whitby, K. T.; Sverdrup, G. M. (1980) California aerosols: their physical and chemical characteristics. In: Hidy,
29 G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The character and origins of smog
30 aerosols: a digest of results from the California Aerosol Characterization Experiment (ACHEX). New York,
31 NY: John Wiley & Sons, Inc.; pp. 477-517. (Advances in environmental science and technology: v. 9).
32 Whitby, K. T.; Charlson, R. E.; Wilson, W. E.; Stevens, R. K. (1974) The size of suspended particle matter in air.
33 Science (Washington, DC) 183: 1098-1099.
34 Willeke, K.; Whitby, K. T. (1975) Atmosperic aerosols: size distribution interpretation. J. Air Pollut. Control Assoc.
35 25:529-534.
36 Williams, E. L., II; Grosjean, D. (1990) Removal of atmospheric oxidants with annular denuders. Environ. Sci.
37 Technol. 24: 811-814.
38 Wilson, W. E. (1995) Aerosol exposure, physics, and chemistry. Inhalation Toxicol. 7: 769-772.
39 Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
40 epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
41 Wilson, W. E.; Spiller, L. L.; Ellestad, T. G.; Lamothe, P. J.; Dzubay, T. G.; Stevens, R. K.; Macias, E. S.; Fletcher,
42 R. A.; Husar, J. D.; Husar, R. B.; Whitby, K. T.; Kittelson, D. B.; Cantrell, B. K. (1977) General Motors
43 sulfate dispersion experiment: summary of EPA measurements. J. Air Pollut. Control Assoc. 27: 46-51.
44 Witz, S.; Eden, R. W.; Wadley, M. W.; Dunwoody, C.; Papa, R.; Torre, K. J. (1990) Rapid loss of particulate
45 nitrate, chloride and ammonium on quartz fiber filters during storage. J. Air Waste Manage. Assoc.
46 40:53-61.
47 Wu, C. C.; Suarez, A. E.; Lin, Z.; Kidwell, C. B.; Borgoul, P. V.; Caffrey, P. F.; Ondov, J. M.; Sattler, B. (1998)
48 Application of an Ir tracer to determine soot exposure to students commuting to school on Baltimore public
49 buses. Atmos. Environ. 32: 1911-1919.
50 Yamasaki, H.; Kuwata, K.; Miyamoto, H. (1982) Effects of ambient temperature on aspects of airborne polycyclic
51 aromatic hydrocarbons. Environ. Sci. Technol. 16: 189-194.
52 Yatin, M.; Tuncel, S. G.; Tuncel, G.; Aras, N. K. (1994) Trace element composition of atmospheric aerosols in
53 Ankara, Turkey, determined by instrumental neutron activation analysis. J. Radioanal. Nucl. Chem.
54 181:401-411.
October 1999 3-99 DRAFT-DO NOT QUOTE OR CITE
-------
1 Zhang, X. Q.; McMurry, P. H. (1987) Theoretical analysis of evaporative losses from impactor and filter deposits.
2 Atmos. Environ. 21: 1779-1789.
3 Zhang, X.; McMurry, P. H. (1992) Evaporative losses of fine particulate nitrates during sampling. Atmos. Environ.
4 Part A 26: 3305-3312.
5 Zhang, S.-H.; Shaw, M.; Seinfeld, J. H.; Flagan, R. C. (1992) Photochemical aerosol formation from cc-pinene and
6 p-pinene. J. Geophys. Res. [Atmos.] 97: 20,717-20,729.
7 Zhang, X. Q.; McMurry, P. H.; Hering, S. V.; Casuccio, G. S. (1993) Mixing characteristics and water content of
8 submicron aerosols measured in Los Angeles and at the Grand Canyon. Atmos. Environ. Part A
9 27: 1593-1607.
October 1999 3-100 DRAFT-DO NOT QUOTE OR CITE
-------
i 4. CONCENTRATIONS, SOURCES, AND EMISSIONS
2 OF ATMOSPHERIC PARTICLES
3
4
5 4.1 INTRODUCTION
6 The present chapter incorporates material from Chapters 5 (Sources and Emissions of
7 Atmospheric Particles) and Chapter 6 (Environmental Concentrations) of the previous document,
8 Air Quality Criteria (CD96) for Particulate Matter (U.S. Environmental Protection Agency,
9 1996) and presents updates to these materials where available.
10 Information on ambient concentrations of particles in various size ranges (PM10, PM25,
11 PM10_2 5) and their chemical composition, based on specific field studies, is presented in
12 Section 4.2. The results of field studies will be used to characterize the spatial and temporal
13 variability in PM and its components in selected urban areas in geographically diverse regions of
14 the United States as they become available.
15 Unlike gaseous criteria pollutants (SO2, NO2, CO, O3), which are well defined chemical
16 entities, atmospheric particulate matter (PM) is composed of a variety of particles differing in
17 size and chemical composition. Therefore, sources of each component of the atmospheric
18 aerosol must be considered in turn. Differences in the composition of particles emitted by
19 different sources also will lead to spatial and temporal heterogeneity in the composition of the
20 atmospheric aerosol. The nature of the sources and the composition of the emissions from these
21 sources are discussed in Section 4.3. Since PM is composed of both primary and secondary
22 constituents, emissions of both the primary components and the gaseous precursors of secondary
23 PM must be considered. Nationwide emissions estimates of primary PM and precursors to
24 secondary PM are discussed in Section 4.4. Estimates of contributions of various sources to
25 ambient PM levels given by source apportionment studies are presented in Section 4.5.
26 Information about the composition of emissions from various sources is given in Appendix A.
27
28
29
30
October 1999 4-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.2 TRENDS AND PATTERNS IN AMBIENT PARTICULATE MATTER
2 PM2 5 CONCENTRATIONS AND TRENDS
3 A significant amount of data for characterizing PM10 mass concentrations and trends exists
4 and that available up to about 1994 was presented in CD96. However, data sets for
5 characterizing PM2 5 and PM(10_2 5) mass or trends are not as extensive. Results from the small
6 number of recent studies in which daily mass and composition measurements are available for
7 extended periods will be discussed in this section. Sources of data on PM25 (fine) and PM(10_25)
8 (coarse), which were discussed in CD96, include EPA's Aerometric Information Retrieval
9 System (AIRS) (Aerometric Information Retrieval System, 1995), IMPROVE (Eldred and Cahill,
10 1994; Cahill, 1996), The California Air Resources Board (CARB) (California Air Resources
11 Board, 1995), the Harvard Six-Cities Data Base (Spengler et al, 1986; Neas, 1996), and the
12 Harvard Philadelphia Data Base (Koutrakis, 1995). The Inhalable Particulate Network (IPN)
13 (Inhalable Particulate Network, 1985; Rodes and Evans, 1985) provided TSP, PM15 and PM25
14 data with only a small amount of PM10 data.
15 Summary tables giving the results of field studies which obtained data for the composition
16 of particles in the PM2 5, PM(10_2 5), or PM10 size ranges were presented in Appendix A
17 to Chapter 6 of CD96. The summary tables included data for mass, organic carbon, elemental
18 carbon, nitrate, sulfate, and trace elements. The results of sixty six studies were separated and
19 presented for the eastern, western and central United States. The data for the broad
20 compositional categories given above from these studies are summarized in pie-chart form in
21 Figures 4-la, 4-lb, and 4-lc. The mean ratio of PM(10_25) to PM25 from these studies was 0.33 in
22 the East, 0.92 in the central United States, and 0.89 in the West.
23
24 PM10 Concentrations and Trends
25 Nationwide PM10 levels declined by 20% between 1988 and 1994 (U.S. Environmental
26 Protection Agency, 1996). The United States was divided into seven regions and on a regional
27 basis, reductions ranged from 17% in the Southeast to 33% in the Southwest. The estimated
28 ratio of PM25 to PM10 ranged from 0.38 to 0.70 in the seven regions considered. Darlington et al.
29 (1997) extended this analysis to include 1995. Their analysis indicated a nationwide average
30 PM10 level of 34 //g/m3 in 1988 declining to 26 //g/m3 in 1995, corresponding to an overall
October 1999 4-2 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 Components
Minerals 4.3%
Unknown 22.8%
EC 3.9%
OCx 1.4 20.9%
NO; 1.1%
so; 34.1%
(N 1-0*13%
Nitrate based on 3 studies
Coarse PM(2.5-10) Components
Unknown 41.5%
Minerals 51.8%
(NhQ* 1.8%
SO4" 4.9%
Insufficient Nitrate, OC, and EC data available
Figure 4-1 a. Major constituents of particles measured at sites in the eastern United States.
(NH4+)* represents the concentration of NH4+ that would be required if all
SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore, (NH4+)*
represents an upper limit to the true concentration of NH4+.
Source: U.S. Environmental Protection Agency (1996).
October 1999 4-3
DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 Components
EC 9.0%
Minerals 7.6%
OC x 1.4 44.6%
so: 22.3%
(NH4+)*10.2%
NO; 8.1%
Reconstructed sum = 124.8%
Coarse PM(2.5-10) Components
Unknown 33.0%
so 3.1%
Minerals 62.8%
Insufficient Nitrate, OC, and EC data available
Figure 4-lb. Major constituents of particles measured at sites in the central United States
(NH4+)* represents the concentration of NH4+ that would be required if all
SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore, (NH4+)*
represents an upper limit to the true concentration of NH4+.
Source: U.S. Environmental Protection Agency (1996).
October 1999 4-4
DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 Components
EC 14.7%
OCx 1.4 38.9%
Minerals 14.6%
so; 10.8%
(NhC)* 7.5%
NO3"15.7%
Reconstructed sum = 102.2%
Coarse PM(2.5-10) Components
Unknown 27%
(N HI)* 0.8%
so; 3.1%
Minerals 69.9%
Insufficient Nitrate, OC, and EC data available
Figure 4-lc. Major constituents of particles measured at sites in the western United
States. (NH4+)* represents the concentration of NH4+ that would be required
if all SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore,
(NH4+)* represents an upper limit to the true concentration of NH4+.
Source: U.S. Environmental Protection Agency (1996).
October 1999
4-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
reduction of 24% or 3.4% per year. They also found that the 95th percentile value declined from
69 //g/m3 in 1988 to 52 //g/m3 in 1995.
PM25 data have been collected continuously since 1994 as part of the children's health
study in twelve communities in southern California (Taylor et al., 1998). Data obtained at all
sites show decreases ranging from 2% at Santa Maria to 37% at San Dimas/ Glendora in PM2 5
from 1994 through 1998. These decrease were accompanied by decreases in major components
such as nitrate, sulfate, ammonium, and acids. However, undefined components showed a mixed
pattern of increases and decreases at the same sites. Information regarding PM2 5 levels mainly in
non-urban areas across the United States is shown in Figure 4-2. Sufficient data are not yet
available to permit the calculation of nationwide trends or average levels of PM25 or PM(10_2 5),
however some general conclusions can be reached. Darlington et al. (1997) proposed that since
the consistent reductions in PM10 levels were found in a wide variety of environments ranging
from urban to rural over large areas, that common factors or controls might be responsible and
that these factors affected fine particles more strongly than coarse particles because fine particles
can be transported over longer distances.
- «»«J''
i ,- *V "-"V. ' 4
•' .' , V • •'(>, »*
!> I', • '-"<<> ' V,«"
«
Figure 4-2. Annual average PM25 concentration (1994-96).
Source: CAPITA, 1999.
(Available http://llcapita.wuste.edu/CAPITA/Capita Reports/Capita Activities 98-99/CAPITA Activities 98 99/)
October 1999
4-6
DRAFT-DO NOT QUOTE OR CITE
-------
1 A few attempts to infer "background" levels of PM25 and PM10 have been made. The
2 background levels most relevant to the present criteria document include (1) a "natural
3 background" which excludes all anthropogenic sources anywhere in the world, and (2) a "natural
4 and uncontrollable anthropogenic background" which includes anthropogenic sources outside of
5 North America in addition to (1). Annual average natural background levels (1) of PM10 have
6 been estimated to range from 4 to 8 //g/m3 in the western United States and 5 to 11 //g/m3 in the
7 eastern United States. Corresponding PM25 levels have been estimated to range from 1 to
8 4 //g/m3 in the western United States and from 2 to 5 //g/m3 in the eastern United States (U.S.
9 Environmental Protection Agency, 1996). Twenty-four hour average natural background
10 concentrations may be substantially higher than the annual or seasonal average natural
11 background concentrations. Estimates of levels for background (2) are not yet available.
12 Data for characterizing the daily and seasonal variability of PM25 mass and composition
13 will be discussed in 4.2.1, the interrelations and correlations among the various PM components
14 and parameters will be discussed in 4.2.2, the spatial variability of various PM components will
15 be discussed in 4.2.3, new data on PM2 5 mass from EPA compliance network will be discussed
16 in 4.2.4, and trends of PM2 5 mass in rural air will be discussed in 4.2.5.
17
18 4.2.1 Daily and Seasonal Variability
19 Information, useful for relating ambient concentrations to health effects, can be obtained by
20 analyzing long time series of concentration data at a single site. Collocated 24-hour PM25 and
21 PM10 filter samples were collected at a site in southwestern Philadelphia, PA from May 1992
22 through April 1995 (Koutrakis, 1995). The PM2 5 filters were analyzed by X-ray fluorescence
23 (XRF) spectroscopy at EPA for the three years. This unique data set was collected on a nearly
24 daily basis, thereby allowing an assessment of day-to-day variability in aerosol concentrations
25 and relationships. EPA has also obtained similar data bases for at least one year from Phoenix,
26 AZ and Baltimore, MD. PM2 5 and PM(10_2 5) mass data were also collected at a number of sites in
27 California (California Air Resources Board, 1995). The data collected at Philadelphia and at
28 Riverside-Rubidoux, CA were presented in CD96 and are shown here for completeness.
29 In addition, the compositional data obtained at the Philadelphia and Phoenix sites will be
30 presented in Section 4.2.2 and spatial relations among PM components in Philadelphia will be
31 presented in Section 4.2.3.
October 1999 4-7 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
The data for Philadelphia are presented as box plots showing the lowest, lowest tenth
percentile, lowest quartile, median, highest quartile, highest tenth percentile, and highest PM2 5
values in Figure 4-3. The four three-month averaging periods shown (March-May, June-August,
September-November, December-February) correspond to the so-called climatological or
meteorological seasons.
70-
— 60~
a 5°-
o
o3 40-
8
§ 30-
o
20-
10-
n -
I
i
1
»
i
r
[
t
}
Philadelphia - PBY site
r PM2.5
(n = 1 024)
h
I
A /
j • 4
[ V ^
i
»
t
Mar- May
June - Aug
Sep - Nov
Dec - Feb
Figure 4-3. Concentrations of PM2 5 measured at the PBY site in southwestern
Philadelphia. The data show the lowest, lowest tenth percentile, lowest
quartile, median (black circles), highest quartile, highest tenth percentile, and
highest PM2 5 values.
1 Frequency distributions for PM2 5 are shown in Figure 4-4 for Philadelphia. Concentrations
2 predicted from the log-normal distribution, using mean values and geometric standard deviation
3 derived from the data, are also shown. Frequency distributions of particle concentrations at
4 several sites in the South Coast Air Basin (Kao and Friedlander, 1995) have also been shown to
5 be approximated reasonably well by log-normal distributions.
October 1999
4-8
DRAFT-DO NOT QUOTE OR CITE
-------
geometric mean = 15.2 |jg/m
og= 1.69
|LJ|LJ|LJ|LJ|LJ|LJ|LJ|
10 20 30 40 50 60
Concentration (|jg/m3)
70 80
Figure 4-4. Frequency distribution of PM2 5 concentrations measured at the PBY site in
southwestern Philadelphia. Log-normal distribution fit to the data shown as
solid line.
1 In Philadelphia, the highest PM2 5 values were observed when winds were from the
2 southwest during sunny but hazy high pressure conditions. In contrast, the lowest values are
3 found after significant rainstorms during all seasons of the year. Day-to-day concentration
4 differences in the data set are 6.8 ± 6.5 //g/m3 for PM2 5 and 8.6 ± 7.5 //g/m3 for PM10. Maximum
5 day-to-day concentration differences are 54.7 //g/m3 for PM2 5 and 50.4 //g/m3 for PM10.
6 Different conclusions could be drawn about data collected elsewhere in the United States.
7 PM2 5 concentrations obtained in Phoenix, AZ are summarized in Figure 4-5 and frequency
8 distributions of PM25 concentrations obtained in Phoenix are shown in Figure 4-6. Day-to-day
9 concentration differences in this data set are 2.9 ±3.0 //g/m3 with a maximum day-to-day
October 1999
4-9
DRAFT-DO NOT QUOTE OR CITE
-------
tu •
^ 30-
~B)
o 20-
2
c
Q)
c 10-
o
O
n -
Phoenix, AZ T
PM2.5
(n = 876)
y
\\
_
T
1
0
T ?
Mar - May June-Aug Sep-Nov Dec-Feb
Figure 4-5. Concentrations of PM2 5 measured at the EPA site in Phoenix, AZ. The data
show the lowest, lowest tenth percentile, lowest quartile, median (black
circles), highest quartile, highest tenth percentile, and highest PM2 5 values.
1 concentration difference of 23 //g/m3. PM25 and PM(10_25) data were obtained at a number of sites
2 in California on a sampling schedule of every six days with dichotomous samplers (California
3 Air Resources Board, 1995). Data for PM25 are summarized in box plot form in Figure 4-7.
4 The frequency distribution of PM25 concentrations obtained at Riverside-Rubidoux from 1989
5 to 1994 is shown in Figure 4-8. It can be seen that the data are not as well fit by a log normal
6 distribution as can the data shown in Figures 4-4 and 4-6, mainly as the result of a significant
7 number of days with PM25 >100 //g/m3 (Figure 4-7).
8 An examination of the data from Philadelphia, PA, Phoenix, AZ, and Riverside, CA
9 indicates that substantial differences exist in aerosol properties between widely separated
10 geographic regions. Fine mode particles make up most of the PM10 mass observed in
11 Philadelphia and appear to drive the daily and seasonal variability in PM10 concentrations there.
October 1999
4-10
DRAFT-DO NOT QUOTE OR CITE
-------
200
PM
2.5
geometric mean = 10.5 |jg/m3
afl = 1.70
10 15 20 25 30 35 40
Concentration (|jg/nn3)
Figure 4-6. Frequency distribution of PM2 5 concentrations measured at the EPA site in
Phoenix, AZ.
1
2
3
4
5
6
7
8
9
10
11
Coarse mode particles represent a larger fraction of PM10 mass in Phoenix and Riverside and
drive the seasonal variability in PM10 seen there. The ratio of PM25 to PM10 mass is much larger
in Philadelphia (0.72) than in either Phoenix (0.34) or Riverside (0.49). Differences between
median and maximum concentrations in any size fraction are much larger at the Riverside site
than at either the Philadelphia or Phoenix sites. Many of these differences could reflect the more
sporadic nature of dust suspension at Riverside. These considerations demonstrate the hazards in
extrapolating conclusions about the nature of variability in aerosol characteristics inferred at one
location to another.
October 1999
4-11
DRAFT-DO NOT QUOTE OR CITE
-------
o
O
140-
120-
100-
60-
40-
20-
o-
Fine
(n = 382)
•
r
•
•
ft 1 i
i ii
• T-T V
1 1 1
V I i Y
i i i i
Jan - Mar Apr - Jun Jul - Sept Oct - Dec
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr
Figure 4-7. Concentrations of PM25 measured at the Riverside-Rubidoux site. The data
show the lowest, lowest tenth percentile, lowest quartile, median (black
squares), highest quartile, highest tenth percentile, and highest PM2 5 values.
1 4.2.2 Relations Between Mass and Chemical Component Concentrations
2 Time series of elemental composition data for PM2 5 particles have been obtained at a few
3 locations across the United States. The filter samples that were collected at the PBY site in
4 southwestern Philadelphia and were used in the construction of Figures 4-3 and 4-4 were also
5 analyzed by X-ray fluorescence. Concentrations of the trace elements and correlations between
6 trace elements and the total mass of particles in the PM2 5 size range are shown in Table 4-1.
7 Also shown in Table 4-1 are similar results obtained for filter samples collected in Phoenix, AZ.
8 Filters from both monitoring studies were analyzed by the same X-ray spectrometer at the U.S.
9 EPA facility in Research Triangle Park, NC. As can be seen from inspection of Table 4-1, the
10 analytical uncertainty (given in parentheses next to concentrations) as a fraction of the absolute
11 concentration is highly variable and it exceeds the concentration for a number of trace metals
12 whose absolute concentrations are low, while it is very small for abundant elements such as S.
13
October 1999
4-12
DRAFT-DO NOT QUOTE OR CITE
-------
1OO
8O-
6O-
4O-
2O-
PM
2.5
geometric mean = 26.6 |jg/m3
cjg = 2.10
O 2O 4O 6O 8O 1OO 12O 14O 16O 18O 2OO
Figure 4-8. Frequency distribution of PM25 concentrations measured at the Riverside-
Rubidoux site.
1 There are a number of distinct differences between the two data sets. For instance, sulfate
2 and associated cations and water appear to constitute a major fraction of the composition of the
3 PM in the Philadelphia data set while they appear to constitute a much smaller fraction of the
4 Phoenix data set. The highest PM2 5 values were observed in Philadelphia during episodes driven
5 by high sulfate abundances and are due, at least partly, to higher sulfate concentrations.
6 Correlation coefficients between SO^ and PM25 were 0.97 during the summer of 1993. Similar
7 correlations between SO^ and PM2 5 were found at a site in northeastern Philadelphia (24 km
8 distant from the site under discussion) during the summer of 1993.
9 Concentrations of "crustal elements" (e.g., Al, Si, K, Ca, Ti, Mn, Fe) are higher relative to
10 PM25 mass in the Phoenix data set compared to the Philadelphia data set. Sulfur is very highly
11 correlated with PM2 5 (r = 0.92) in the Philadelphia data set while it is only weakly correlated
12 (r = 0.16) with PM25 in the Phoenix data set. Toxic trace metals e.g., Cr, Co, Ni, Cu, Zn, As, and
13 Pb are not well correlated ( 0.04 < r < 0.25) with PM25 in the Philadelphia data set while they are
14 more variably correlated (0.01 < r < 0.69) with PM25 in the Phoenix data set. The uncertainty in
15 the concentration measurement most probably plays a role in determining a species' correlation
October 1999
4-13
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-1. CONCENTRATIONS OF PM25 (//g/m3) AND SELECTED ELEMENTS
(ng/m3) IN THE PM25 SIZE RANGE AND CORRELATIONS BETWEEN ELEMENTS
AND PM25 MASS. VALUES IN PARENTHESES REFER TO ANALYTICAL
UNCERTAINTY IN X-RAY FLUORESCENCE DETERMINATIONS
PM25
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
As
Se
Br
Pb
Philadelphia, PA1
Concentration
17.0 ± 0.8 Mg/m3
4.0 (3 1.1) ng/m3
116(21.1)
8.6(10.3)
2100(143)
5.1 (3.4)
60.4 (4.7)
46.6 (4.2)
4.9(4.1)
8.8(1.8)
0.7 (0.7)
3.1(0.8)
109(10.5)
0.1 (1.4)
7.3 (1.4)
4.8(1.1)
36.9(3.7)
0.6(1.2)
1.5 (0.6)
5.0 (0.9)
17.6(2.5)
r
1
0.10
0.51
0.31
0.92
-0.01
0.50
0.39
0.44
0.37
0.15
0.39
0.50
0.04
0.22
0.25
0.21
0.18
0.63
0.11
0.19
Phoenix, AZ2
Concentration
9.4 ± 0.5 //g/m3
68.9 (27.2) ng/m3
209 (48.4)
7.6 (4.5)
408 (30.9)
11.4(2.4)
78.6 (8.2)
76.5 (9.7)
7.2(3.3)
0.7(1.0)
0.4 (0.4)
4.3 (0.6)
112(15.1)
-0.2 (0.8)
0.4 (0.4)
3.3 (0.7)
12.7(1.7)
1.3 (0.6)
0.3 (0.3)
3.1 (0.6)
4.5(1.0)
r
1
0.23
0.35
0.52
0.16
0.13
0.67
0.51
0.44
-0.28
0.41
0.64
0.80
-0.01
0.38
0.69
0.64
0.50
0.40
0.57
0.69
'n= 1105
2n = 643
October 1999
4-14
DRAFT-DO NOT QUOTE OR CITE
-------
1 with PM2 5, especially when the analytical uncertainty is high relative to concentration, e.g., for
2 trace metals such as Co.
3
4 4.2.3 Spatial Variability
5 Three methods for comparing the chemical composition of aerosol databases obtained at
6 different locations and times were discussed by Wongphatarakul et al. (1998). Log - log plots of
7 chemical concentrations obtained at pairs of sampling sites accompanied by the coefficient of
8 divergence (COD) were examined as a way to provide an easily visualized means of comparing
9 two data sets1. Examples comparing downtown Los Angeles with Burbank and with Riverside
10 Rubidoux are shown in Figures 4-9 and 4-10, respectively. As the composition of two sampling
11 sites become more similar, the COD approaches zero, as their compositions diverge, the COD
12 approaches one. Cluster analyses based on the COD between individual data sets can be used to
13 determine the degree of similarity among a number of data sets. Correlation coefficients
14 calculated between components can be used to show the degree of similarity between pairs of
15 sampling sites. In addition to calculating correlation coefficients for total mass or for individual
16 components, correlation coefficients for characterizing the spatial variation of the contributions
17 from given source types can also be calculated by averaging the correlation coefficients of the set
18 of chemical components that represent the source type. The first two methods could be applied
19 either to aerosol data sets collected at multiple sites within a given geographic region or to
20 aerosol data sets collected at widely different locations or times while the third method is best
21 used to characterize sites within a particular geographic region.
22 Correlation coefficients showing the spatial relations among PM2 5 and contributions from
23 different source categories obtained at various sites in the South Coast Air Basin (SoCAB) Study
24 are shown in Table 4-2. In Wongphatarakul et al. (1998), crustal material (crust), motor vehicle
The COD for two sampling sites is defined as follows:
j X-: 2
where x;j represents the average concentration for a chemical component i at site j, j and k represent two sampling
sites and p is the number of chemical components.
October 1999 4-15 DRAFT-DO NOT QUOTE OR CITE
-------
102
<£ 10°
c
CO
I 1C'1
QQ
io-2-
10
-3
COD=0.099
so,
2-
nknown
NO;
Source: \Afongphatarakul (1997)
10"
\ I I
1C'2 1C'1 10°
Downtown Los Angeles (|jg/m3)
102
*Jan. 2 - Dec. 28,1986 (63 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 |jm
**Jan. 2 - Dec. 28,1986 (61 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 urn
Figure 4-9. PM25 chemical components in downtown Los Angeles and Burbank (1986)*
have similar characteristics.
Source: Wongphatarakul (1997).
October 1999
4-16
DRAFT-DO NOT QUOTE OR CITE
-------
x
13
O
T3
J2
13
o:
102
E 10°-
10"-
10
-3
COD=0.230
10
,-3
Unknown
Gas Al
Zn
Source: Wongphatarakul(1997)
T
10':
10'
I
10°
102
Downtown Los Angeles (pg/m3)
*Jan. 2 - Dec. 28,1986 (63 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 urn
**Jan. 2 - Dec. 28,1986 (60 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 pm
Figure 4-10. Concentrations of PM25 chemical components in Rubidoux and downtown
Los Angeles (1986)*. The diagram shows a significant spread in the
concentrations for the two sites compared with downtown Los Angeles and
Burbank (Figure 7).
Source: Wongphatarakul (1997).
October 1999
4-17
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-2. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
PM2 5 MASS AND DIFFERENT SOURCES FOR PAIRS OF SAMPLING
SITES IN SoCAB (1986)
R™ Rcmstal Rsec R™ Rresidmloll
Hawthorne and Rubidoux
Long Beach and Rubidoux
Anaheim and Rubidoux
Downtown Los Angeles and Rubidoux
Burbank and Rubidoux
Hawthorne and Anaheim
Long Beach and Anaheim
Burbank and Anaheim
Downtown Los Angeles and Anaheim
Downtown Los Angeles and Hawthorne
Burbank and Hawthorne
Long Beach and Burbank
Long Beach and Hawthorne
Downtown Long Angeles and Long Beach
Downtown Los Angeles and Burbank
-0.027
0.051
0.066
0.095
0.120
0.760
0.852
0.770
0.827
0.808
0.704
0.731
0.880
0.842
0.928
0.034
0.075
0.105
0.143
0.568
0.599
0.633
0.649
0.653
0.825
0.768
0.888
0.749
0.804
0.854
0.790
0.737
0.909
0.817
0.960
0.492
0.504
0.579
0.556
0.669
0.688
0.714
0.861
0.719
0.871
0.170
0.150
0.161
0.233
0.533
0.491
0.295
0.482
0.378
0.606
Source: Wongphatarakul et al. (1998).
1 exhaust (mv), residual oil emissions (residual oil) and secondary PM (sec) were considered as
2 source categories. Al, Si, Fe and Ca were used as markers for crustal material (crustal). V and
3 Ni were used as markers for fuel oil combustion (residual oil). Pb, Br and Mn were used as
4 markers for motor vehicle exhaust (MV). NO3", NH4+ and SO4= represent secondary PM
5 components (sec). The average of the correlation coefficients of marker elements within each
6 source category are shown in Table 4-2. Total PM25 (tot) varies in a similar fashion among
7 Friedlander, 1995). Values of Rsec and Rmv are much higher than those for Rcmstal and
8 throughout the SoCAB suggesting a more uniform distribution of the contributions from
October 1999
4-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
secondary PM formation and automobiles than from crustal material and localized stationary
sources.
In the Philadelphia area PM25 was found to be strongly correlated (r > 0.9) between seven
urban sites and one background site (Valley Forge, PA) during the summer of 1993 (Suh et al.,
1995). The same relations were also found during the summer of 1994 at four monitoring sites
as part of a separate study (Pinto et al., 1995). The results from these studies strongly suggest
that PM2 5 and SO^ concentrations are spatially uniform throughout the Philadelphia area, and
that variability in PM10 levels is caused largely by variability in PM2 5 (Wilson and Suh, 1997).
However, not enough data are available from regional sites to define the total areal extent of the
spatial homogeneity in PM2 5 and SO4= concentrations observed in the urban concentrations SO4=.
Correlation coefficients in Philadelphia air for PM2 5, crustal components (Al, Si, Ca, Fe), the
major secondary component (sulfate), organic carbon (OC) and elemental carbon (EC) are shown
in Table 4-2a. Because these data were obtained after Pb had been phased out of gasoline, a
motor vehicle contribution could not be extracted.
TABLE 4-2a. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
PM2 5 MASS AND DIFFERENT COMPONENTS FOR PAIRS OF SAMPLING
SITES IN PHILADELPHIA (1994)
Castor Ave. and Roxboro
Castor Ave. and NE Airport
Castor Ave. and Broad St.
Roxboro and NE Airport
Roxboro and Broad St.
NE Airport and Broad St.
Rtot
0.92
0.93
0.93
0.98
0.95
0.95
15
^crustal
0.52
0.47
0.57
0.67
0.90
0.69
Rsec
0.98
0.99
0.99
0.98
0.98
0.99
ROC
0.88
0.88
0.85
0.83
0.86
0.84
Rec
0.84
0.77
0.89
0.82
0.79
0.63
Source: Pinto etal. (1995).
October 1999
4-19
DRAFT-DO NOT QUOTE OR CITE
-------
1 4.2.4 Urban Concentrations and Patterns from the New PM2 5 Compliance
2 Network
3 4.2.5 Trends and Patterns
4 4.2.5.1 Visual Range/Haziness
5 Observations of visual range, obtained by the National Weather Service and available
6 through the National Climatic Data Center of the National Oceanic and Atmospheric
7 Administration, provide one of the few truly long-term, daily records of any parameter related to
8 air pollution. After some manipulation, the visual range data can be used as an indicator of fine
9 mode particle pollution. The data reduction process and analyses of resulting trends have been
10 reported by Husar et al. (1994), Husar and Wilson (1993), and Husar et al. (1981).
11 Visual range i.e., the maximum distance at which an observer can discern the outline of an
12 object, is an understandable and for many purposes an appropriate measure of the optical
13 environment. It has the disadvantage, however, of being related inversely to aerosol
14 concentration. It is usual, therefore, to convert visual range to a direct indicator of fine mode.
15 particle concentration. The quantitative measure of haziness is the extinction coefficient, Bext,
16 defined as Bext=K/visual range, where K is the Koschmieder constant. The value of K is
17 determined both by the threshold sensitivity of the human eye and the initial contrast of the
18 visible object against the horizon sky. Husar et al. (1994) use K=1.9 in accordance with the data
19 given by Griffing (1980). The extinction coefficient is in units of km"1 and is proportional to the
20 concentration of light scattering and absorbing aerosols and gases. The radiative transfer
21 characteristics which determine the visual range depend on time of day. Only local noon
22 observations are used.
23
24 Haze Trend Summary
25 The U.S. haze patterns and trends from 1960 to 1992 were presented in the CD96 as
26 16 haze maps representing four time periods (5 year averages centered at 1960, 1970, 1980, and
27 1990) and the four climatological seasons (Quarter 1 is for December of prior year, January and
28 February of indicated year, etc.) seasons. The average haze patterns, centered on 5 year averages
29 for 1980 and 1990 are shown in Figure 4-11 and will be replaced by similar data for 1990 and
30 1995. Haze is indicated by the 75th percentile of the extinction coefficient which is calculated
31 from the visual range, corrected to 60% relative humidity by the Koschmeider relationship.
October 1999 4-20 DRAFT-DO NOT QUOTE OR CITE
-------
1980
*
*
Figure 4-11. Five-year average haze patterns (75th percentile of the extinction coefficient)
centered on 1980 and 1990 (to be replaced with patterns for 1990 and 1995).
The relationship between haze, expressed as extinction coefficient (corrected
to 60% relative humidity) is 1.9/extinction coefficient in km"1 = visual range
in km, i.e., the intensity of shading increases as visual range decreases:
0.2 km-1 = 9.5 km, 0.17 = 11.2, 0.14 = 13.6, 0.11 = 17.3, 0.08 = 23.8, and
0.05 = 38.0. Ql = Dec., Jan., Feb.; Q2 = March, April, May; Q3 = June,
July, Aug.; Q4 = Sept., Oct., Nov.
October 1999
4-21
DRAFT-DO NOT QUOTE OR CITE
-------
1 Regional Pattern
2 Trends for specific regions in the eastern U.S. might be updated, from 1992 to 1998, and be
3 presented in a manner similar to that shown in Figure 4-12 for 1940 to 1990 (CD96). The trend
4 graphs represent the 75th percentile of Bext for the stations located within the designated region.
5 The trends are presented for Quarters 1 (winter) and 3 (summer) separately. The northwestern
6 U.S. exhibits an increase of Quarter 3 haze between 1960 and 1970, and a steady decline between
7 1973 (0.22) and 1992 (0.12). In the winter quarter the haziness has steadily declined from
8 0.15 to 0.10 in the 30-year period. The Mid-Atlantic region that includes the Virginias and
9 Carolinas shows a strong summer increase between 1960 and 1973, followed by a decline. The
10 winter haze was virtually unchanged over the 30-year period. The haziness over the Gulf states
11 increased between 1960 and 1970, and remained virtually unchanged since then. The central
12 Midwest including Missouri and Arkansas exhibit virtually no change during the winter season
13 and a slight increase in the summer (1960-1970). The upper Midwest (Figure 4-12) shows an
14 opposing trend for summer and winter. While summer haze has increased, mostly 1960-1973,
15 the winter haze has declined.
16
17 4.2.5.2 Urban Trends
18 PM2 5 trends will be shown for those cities which have several years of PM2 5 data as they
19 become available. As an example, trend data from Stockton-Hazleton, CA are shown in
20 Figure 4-13.
21
22
23 4.3 SOURCES OF PRIMARY AND SECONDARY PARTICULATE
24 MATTER
25 Information about the nature and relative importance of sources of ambient PM is presented
26 in this section. Table 4-3 summarize anthropogenic and natural sources for the major primary
27 and secondary aerosol constituents of fine and coarse particles. Major sources of each
28 constituent are shown in boldface type. Anthropogenic sources can be further divided into
29 stationary and mobile sources. Stationary sources include fuel combustion for electrical utilities,
30 residential space heating and industrial processes; construction and demolition; metals, minerals,
31 petrochemicals and wood products processing; mills and elevators used in agriculture; erosion
October 1999 4-22 DRAFT-DO NOT QUOTE OR CITE
-------
o
o
r+
O
Tl
H
O
O
2
O
H
O
c
o
H
W
O
W
O
HH
H
W
^-v 0.26
i
f= 0.22
,-x 0.18
j£J 0.14
0.10
Upper Midwest
Midwest
West Gulf
Southeast
Mid Atlantic
Northeast
o>
m
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
CD
CD
3
O
1940 1950 1960 1970 1980 1990
m
0.26
0.22
0.18
0.14
0.10
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 1990
0.06
0.26
0.22
0.18
0.14
0.10
1940 1950 1960 1970 1980 19!
0.06
CD
03
a
1940 1950 1960 1970 1980 1990
Figure 4-12. Secular haze trends (1940 to 1990) for six eastern U.S. regions, summer (Ql) and winter (Q2).
-------
o
o
r+
O
Tl
f
O
O
2
O
H
O
c
o
H
W
O
^
O
HH
H
W
Annual Arithmetic Mean (pg/m3 )
100
80
60
40
20
0
NAAQS
Stockton-Hazelton, CA
100
80
70
eo 60
~D> 50
40
30
20
10
0
89
90 91 92 93
• Total Coarse
90th Percentile (ug/nf )
94 95
•••• Fine
180
160
140
120
100
80
60
40
20
0
(b)
89
90
• Total
91
92 93
— Coarse
94 95
•••• Fine
Every Sixth Day. 1991
1
o.g
0.8
0.7
°'4
0.3
0.2
0.1
01/06/91 Io3/01/91 I 05/06/91 I 07/05/91 I 09/03/91 I 11/02/91 112/26/91
02/11/91 04/06/81 06/05/81 OB/10/81 10/03/91 12/08/81
Date
0 COARSE * FINE
PM2 5 as a Fraction of PM10
01/06/891 01/05/90 I 01/06/911 01/01/92loi/01/93l 01/02/941 01/04/95108/31/95
07/03/89 07/04/90 07/05/91 07/05/92 07/06/93 07/01/94 07/Of —
7/02/95
Date
Figure 4-13. Trend data from Stockton-Hazelton from CARB: (a) fine, coarse, and total means; (b) fine, coarse, and total 90th
percentiles; (c) every-sixth-day fine and coarse mass for 1991; and (d) fine and coarse mass as a fraction of PM10.
-------
o
r+
O
&
CD
H^
>o
>o
>o
i ABLitL to. ^\jrv a 1 1 1 uiLr> i a \JF A i iviuarnjLKi^
rAKii^i^JLa Ar>u I±IJLIK IVIA^JUK auuK^
Jli»
Sources
Primary (PM <2.5 /^m) Primary (PM >2.5 /^m) Secondary PM Precursors (PM <2.5 /^m)
Aerosol
species Natural Anthropogenic Natural
SO4= Sea spray Fossil fuel combustion Sea spray
Sulfate
Anthropogenic Natural
— Oxidation of reduced sulfur
gases emitted by the oceans and
wetlands; and SO2 and H2S
emitted by volcanism and forest
fires
Anthropogenic
Oxidation of SO2 emitted
from fossil fuel
combustion1
O
NO3-
Nitrate
Minerals
Erosion,
re-entrainment
NH4+ —
Ammonium
Organic Wild fires
carbon (OC)
Motor vehicle exhaust2
Fugitive dust; paved,
unpaved roads;
agriculture and forestry
Motor vehicle exhaust2
Open burning, wood
burning, motor
vehicle exhaust,
cooking
Oxidation of NOX produced by
soils, forest fires, and lighting
Oxidation of NOX emitted
from fossil fuel
combustion; and in motor
vehicle exhaust
Erosion, re-entrainment
Fugitive dust; paved,
unpaved road dust,
agriculture and
forestry
Tire and asphalt wear,
paved road dust
Emissions of NH3 from wild
animals, undisturbed soil
Oxidation of hydrocarbons
emitted by vegetation,
(terpenes, waxes); wild fires
Emissions of NH3 from
animal husbandry,
sewage, fertilized land
Oxidation of hydrocarbons
emitted by motor vehicles,
open burning, wood
burning
Tl
H
O
O
2j
O
H
0
O
H
W
0
O
H
W
Elemental Wild fires Motor vehicle
carbon exhaust, wood
(EC) burning, cooking
Metals Volcanic Fossil fuel
activity combustion, smelting,
brake wear
Bioaerosols Viruses, —
bacteria
'Major source of each component shown in boldface type.
2Relatively minor source of substance, included only for the
Erosion, re-entrainment, — — —
organic debris
Plant, insect fragments, — — —
pollen, fungal spores,
bacterial agglomerates
sake of completeness.
-------
1 from tilled lands; waste disposal and recycling; and fugitive dust from paved and unpaved roads.
2 Mobile, or transportation related, sources include direct emissions of primary PM and secondary
3 PM precursors from highway and off-highway vehicles and nonroad sources. Also shown are
4 sources for precursor gases whose oxidation forms secondary particulate matter. A description of
5 the atmospheric chemical processes producing secondary PM is given in Section 3.4.
6 In general, the nature of sources of fine particulate matter is very different from that for
7 coarse particulate matter. A large fraction of the mass in the fine size fraction is derived from
8 material that has been formed during combustion (primary), has been volatilized in combustion
9 chambers and then recondensed to form primary PM, or has been formed in the atmosphere from
10 precursor gases as secondary PM. Since precursor gases and fine particulate matter are capable
11 of traveling great distances, it is difficult to identify individual sources of constituents. The
12 coarse PM constituents have shorter lifetimes in the atmosphere, so their impacts tend to be more
13 localized. Only major sources for each constituent within each broad category shown at the top
14 of Table 4-3 are listed. Chemical characterizations of primary particulate emissions from a wide
15 variety of natural and anthropogenic sources as shown in Table 4-3 were given in Chapter 5 of
16 CD96. Summary tables of the composition of source emissions presented in CD96 and updates
17 are given in Appendix A. These profiles were based in large measure on the results of various
18 studies collecting source signatures for use in source apportionment studies.
19 Natural sources of primary PM include windblown dust from undisturbed land, sea spray,
20 and plant and insect debris. The oxidation of a fraction of terpenes emitted by vegetation and
21 reduced sulfur species from anaerobic environments leads to secondary PM formation.
22 Ammonium (NH4+) ions, which regulate the pH of particles, are derived from emissions of
23 ammonia (NH3) gas. Source categories for NH3 have been divided into emissions from
24 undisturbed soils (natural) and emissions which are related to human activities (e.g., fertilized
25 lands, domestic and farm animal waste). It is difficult to describe emissions from biomass
26 burning as either natural or anthropogenic. Clearly, fuel wood burning is an anthropogenic
27 source of PM, whereas wildfires would be a natural source. Forest fires have been included as a
28 natural source, because of the lack of information on the amount of prescribed burning or
29 accidental fires caused by humans. Similar considerations apply to the biogenic emissions of
30 trace metals which may be remobilized from anthropogenic inputs.
October 1999 4-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 Receptor models are perhaps the primary means used to estimate source category
2 contributions to particulate matter at individual monitoring sites. Receptor models relate source
3 category contributions to ambient concentrations based on composition analysis of ambient
4 particulate and source emissions samples. In addition, receptor models have been developed for
5 apportioning source categories of primary particulate matter and are not formulated to include the
6 processes of secondary particulate matter formation. However, hybrid receptor models which
7 use elements from a chemical transport model have been used to apportion source categories of
8 gaseous precursors to secondary particulate matter (Stevens and Lewis, 1984).
9 A number of specialty conference proceedings, review articles, and books have been
10 published to provide greater detail about source category apportionment receptor models (Cooper
11 and Watson, 1980; Watson et al, 1981; Macias and Hopke, 1981; Dattner and Hopke, 1982;
12 Pace, 1986; Watson et al., 1989; Gordon, 1980, 1988; Stevens and Pace, 1984; Hopke, 1985,
13 1991; Javitz et al., 1988). Watson et al. (1994a) present data analysis plans which include
14 receptor models as an integral part of visibility and PM10 source apportionment and control
15 strategy development. A review of the various methods used to apportion PM in ambient
16 samples among its source categories was given in Section 5.5.2 of CD96. The collection of the
17 source category characterization profiles shown in Appendix 4A has been motivated in many
18 cases by the need to use them in receptor modeling applications.
19
20 4.3.1 Source Contributions to Ambient PM
21 The results of several source apportionment studies will be discussed in this section to
22 provide an indication of different sources of particulate matter across the United States. First,
23 results obtained by using the chemical mass balance (CMB) approach for estimating
24 contributions to PM2 5 from different source categories at monitoring sites in the United States
25 will be discussed and presented in Table 4-4. Results obtained at a number of monitoring sites in
26 the central and western United States by using the CMB model for PM10 are shown in Table 4-5.
27 The sampling sites represent a variety of different source characteristics within different regions
28 of Arizona, California, Colorado, Idaho, Illinois, Nevada and Ohio. Several of these are
29 background sites, specifically Estrella Park, Gunnery Range, Pinnacle Peak, and Corona de
30 Tucson, AZ, and San Nicolas Island, CA. Definitions of source categories also vary from study
October 1999 4-27 DRAFT-DO NOT QUOTE OR CITE
-------
October 1999
-h
oo
o
H
O
0
2
o
H
O
O
H
W
O
O
H
W
TABLE 4-4. RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM2 5
% Contribution3
Gasoline Road Dust; Vegetation Secondary
Sampling Site Time Period Diesel Vehicles soil burning Sulfate
Pasadena, CA (Schauer etal., 1996) 1982 18.8 5.1 12.4 9.6 20.9
DowntownLA, CA (Schauer etal., 1996) 1982 35.7 6.5 11.1 5.8 20.3
WestLA,CA (Schauer etal., 1996) 1982 18.0 5.7 12.2 11.0 24.1
Rubidoux, CA (Schauer et al., 1996) 1982 12.8 0.7 13.1 1.2 13.8
Philadelphia, PA (Dzubay etal., 1988) Summer 1982 8.5e — 4.4 — 81.9f
Camden, NJ (Dzubay etal., 1988) Summer 1982 9.2e — 3.2 — 81.3f
Clarksboro, NJ (Dzubay etal., 1988) Summer 1982 5.8e — 2.7 — 84.6f
Grover City ILENP; (Glover etal., 1991) 1986 to 1987 — — 2.3 — 83.2f
Grover City, IL SSW; (Glover etal., 1991) 1986 to 1987 — — — — 59.0f
Grover City, IL WNW; (Glover etal., 1991) 1986 to 1987 2.4' — 5.1 — 88.5f
Grover City, IL NNW; (Glover etal., 1991) 1986 to 1987 — — 3.1 — 86.6f
Welby, CO (Lawson and Smith, 1998) Winter 1997 10 28 16 5 10"
Brighton, CO (Lawson and Smith, 1998) Winter 1997 10 26 11 2 15°
'secondary and other organic compounds fincluding associated cations and water
'secondary ammonium sincinerators
"meat cooking hoil fly ash
Vegetative detritus 'fluidized catalyst cracker
Value represents sum of diesel and gasoline vehicle exhaust -Kvind direction
Secondary Misc. Misc. Misc. Misc. Measured
Nitrate Source Source Source Source PM2S
1234 Concentration
7.4 5.3' 9.2b 8.5C l.ld
9.2 3.7' 9.2b 5.2C 0.6d
7.8 4.1' 9.4b 8.2C 1.6d
24.7 4.5' 12.1b 4.5C 0.5d
— 2.2s 1.9h 0.41 —
0.4 2.5s 2.5h 0.71 —
— 0.8s 1.5h 0.4s —
— 9.7k 3.01 1.2s —
— 11.6k 11.91 4.1s 4.6m
— 2.8k — — —
— 3.41 3.0" — —
25P 4, T
32P 2° T — —
klead smelter
'iron works
"copper smelter
"coal power plant
°as ammonium sulfate
pas ammonium nitrate
28.2
32.5
24.5
42.1
27.0
28.3
26.0
-------
o
o
r-K
o
VO
VO
VO
TABLE 4-5. RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM,
% Contribution3
-j^
vo
O
Tl
H
o
O
2
O
H
o
H
W
O
!*
O
I-H
H
W
Primary
Motor Primary Secondary Secondary Misc. Misc. Misc. Misc.
Primary Primary Vehicle Vegetative Ammonium Ammonium Source Source Source Source
Measured
Sampling Site
Central Phoenix, AZ (Chow et al., 1991)
Craycroft, AZ (Chow et al., 1992a)
Hayden 1, AZ (Garfield) (Ryan et al., 1988)
Hayden 2, AZ (Jail) (Ryan et al., 1988)
Phoenix, AZ (Estrella Park) (Chow et al., 1991)
Phoenix, AZ (Gunnery Rg.) (Chow et al., 1991)
Phoenix, AX (Pinnacle Pk.) (Chow et al., 1991)
Rillito, AZ (Thanukos et al., 1992)
Scottsdale, AZ (Chow et al., 1991)
West Phoenix, AZ (Chow et al., 1991)
Bakersfield, CA (Magliano, 1988)
Bakerfield, CA (Chow et al., 1992b)
Crows Landing, CA (Chow et al., 1992b)
Fellows, CA (Chow et al., 1992b)
Fresno, CA (Magliano, 1988)
Fresno, CA (Chow et al., 1992b)
Indio, CA (Kim et al., 1992)
Kern Wildlife Refuge, CA (Chow et al., 1992b)
Long Beach, CA (Gray et al., 1988)
Long Beach, CA (Summer) (Watson et al., 1994a)
Long Beach, CA (Fall) (Watson et al., 1994a)
Riverside, CA (Chow et al., 1992c)
Rubidoux, CA (Gray et al., 1988)
Rubidoux, CA (Summer) (Watson et al., 1994a)
Rubidoux, CA (Fall) (Watson et al., 1994a)
Rubidoux, CA (Chow et al., 1992c)
San Nicolas Island, CA (Summer) (Watson et al.,
1994a)
Time Period
Winter 1989-1990
Winter 1989-1990
1986
1986
Winter 1989-1990
Winter 1989-1990
Winter 1989-1990
1988
Winter 1989-1990
Winter 1989-1990
1986
1988-1989
1988-1989
1988-1989
1986
1988-1989
1988-1989
1986
Summer 1987
Fall 1987
1988
1986
Summer 1987
Fall 1987
1988
Summer 1987
Geological
51.6
55.6
4.8
35.6
67.3
74.1
58.3
53.7
45.5
43.5
40.5
53.9
61.3
53.1
35.6
44.5
56.9
31.6
39.9
24.1
11.8
50.9
49.3
30.4
17.1
55.2
9.2
Construction
0.0
0.0
1.9b
6.8b
0.0
0.0
0.0
17.4b
0.0
0.0
4.4
2.0
0.0
2.6
1.5
0.0
5.2
4.2
0.0
0.0
0.0
0.0
4.6"
3.9
14.4
0.0
0.0
Exhaust
39.0
35.5
0.0
0.0
18.2
20.4
24.2
1.5f
34.5
36.2
8.1
9.7
4.2
3.8
8.3
9.5
7.6
4.6
9.8s
13.7
44.5
10.9
6.4s
15.1
27.1
11.7
5.2
Burning
3.6
0.0
0.0
0.0
1.6
0.0
8.3
0.0
13.5
14.5
14.21
8.2
6.5
6.2
19.11
7.1
12.2
8.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Sulfate
0.3
3.0
3.8
6.8
2.9
3.7
7.5
0.0
1.1
0.6
8.3
6.9
5.3
9.3
3.7
5.0
6.2
6.9
15.4
23.6
4.0
7.5
7.3
8.3
1.9
6.1
21.3
Nitrate
4.4
2.6
0.0
0.0
0.0
0.0
0.0
0.0
6.5
4.5
0.0
16.0
12.4
13.7
0.0
14.5
7.1
3.1
17.7
1.7
24.1
33.4
24.4
23.9
28.2
24.9
2.9
1
0.0
5.1°
70.5°
47.5°
0.0
0.0
0.0
14.6s
0.0
0.0
0.7j
1.3"
1.0m
12.8m
0.2j
0.4m
0.3j
1.0m
0.2j
0.2j
O.Oi
0.5j
0.3j
0 (V
0 (V
0.&
o.tf
2
0.0
0.0
4.8d
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.9"
1.9"
2.6"
0.0
1.9"
1.7h
3.1°
3.9h
4.8h
2.8h
2.0h
1.1°
4.4h
1.0h
1.7h
24.7h
3
0.0
0.0
1.0e
l.T
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.8k
2.3k
2.6k
0.0
O.lk
0.0
1.5k
12.3k
0.0
0.0
1.7°
6.8k
0.0
0.0
6.6°
0.0
4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Concentration
64.0
23.4
105.0
59.0
55.0
27.0
12.0
79.5
55.0
69.0
67.6
79.6
52.5
54.6
48.1
71.5
58.0
47.8
51.9
46.1
96.1
64.0
87.4
114.8
112.0
87.0
17.4
-------
o
o
r-K
o
VO
VO
VO
TABLE 4-5 (cont'd). RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM,
u>
o
O
Tl
H
O
O
2
O
H
Sampling Site Time Period
Stockton, CA (Chow et al., 1992b) 1989
Pocatello, ID (Houck et al., 1992) 1990
S. Chicago, IL (Hopke et al., 1988) 1986
S.E. Chicago, IL (Vermette et al., 1992) 1988
Reno, NV (Chow et al., 1988) 1986-1987
Sparks, NV (Chow et al., 1988) 1986-1987
Follansbee, WV (Skidmore et al., 1992) 1991
Mingo, OH (Skidmore et al., 1992) 1991
Steubenville, OH (Skidmore et al., 1992) 1991
"Smelter background aerosol.
bCement plant sources, including kiln stacks, gypsum pile, and kiln area.
"Copper ore.
dCopper tailings.
'Copper smelter building.
fHeavy-duty diesel exhaust emission.
background aerosol.
'Marine aerosol, road salt, and sea salt plus sodium nitrate.
'Motor vehicle exhaust from diesel and leaded gasoline.
Primary
Motor
Primary Primary Vehicle
Geological Construction Exhaust
55.1 0.8 8.3
8.3 7.5' 0.1
34.0 3.0 3.5
35.9¥ 0.0 2.2f
497 0.0 33.3
36.8 0.0 28.3
15.2 0.0 53.0
20.0 0.0 23.3
18.0 0.0 30.4
^Residual oil combustion.
kSecondary organic carbon.
'Biomass burning.
"Primary crude oil.
"NaCl + NaNO3.
"Lime.
FRoad sanding material.
'Asphalt industry.
%
Primary
Vegetative
Burning
7.7
0.0
0.0
0.0
6.3
32.7
0.0
6.8
1.7
Contribution3
Secondary Secondary Misc. Misc.
Ammonium Ammonium Source Source
Sulfate Nitrate 1 2
5.0 11.2 1.1" 2.9°
0.0 0.0 0.0 0.0
19.2s - 18.9' 2.7°
18.8 - 2.0' 0.7h
4.3 2.0 0.0 0.0
6.6 2.2 0.0 0.0
24.2 - 14.1' 0.0
25.0 - 5.7' 18.3X
30.4 - 8.3' 10.91
'Regional sulfate.
'Steel mills.
"Refuse incinerator.
"Local road dust, coal yard road dust,
"Incineration.
"Unexplained mass.
Misc. Misc. Measured
Source Source PMi0
3 4 Concentration
0.0k 0.0 62.4
84.1r 0.0 100.0
0.0 0.0 80.1
2.7W 18.8s 41.0
0.0 0.0 30.0
0.5k 0.0 41.0
0.0 0.0 66.0
0.0 0.0 60.0
0.0 0.0 46.0
steel haul road dust.
Thosphorus/phosphate industry.
o
H
W
O
O
I-H
H
W
-------
1 to study. The results of the PM10 source apportionment studies were given in CD96 and are
2 presented here to allow easy comparison with results of PM25 source apportionment studies.
3 There are several differences between the source categories shown at the tops of Tables 4-4
4 and 4-5. These differences reflect the nature of sources which are important for producing fine
5 and coarse particulate matter shown in Table 4-3. They are also related to improvements in the
6 ability to distinguish between sources of similar nature, e.g., diesel and gasoline vehicles, meat
7 cooking and vegetation burning. It has only been recently that motor vehicle emissions can be
8 broken down into contributions from diesel and gasoline vehicles through the use of organic
9 tracers. Meat cooking is also distinguished from vegetation burning in more recent studies,
10 although both are considered to be part of biomass burning. Vegetation burning consists of
11 contributions from residential fuel wood burning, wild fires, prescribed burning and burning of
12 agricultural waste. Miscellaneous sources of fine particles include contributions from
13 combustion sources while miscellaneous sources of coarse particles consist of contributions from
14 soil and sea spray and industrial processing of geological material (e.g., cement manufacturing).
15 Although a large number of elements and chemical compounds are used to differentiate among
16 source categories, it can be seen from Tables 4-4 and 4-5 that only a relatively small number of
17 sources are needed to account for the mass of PM25 and PM10.
18 Secondary sulfate is the dominant component of PM25 samples collected in the studies of
19 Dzubay et al. (1988) and Glover et al. (1991). Both studies found that sulfate at their monitoring
20 site arose from regionally dispersed sources. Sulfate, associated cations and water also represent
21 the major components of PM25 found in monitoring studies in the eastern United States
22 (Figure 4-la). Motor vehicle emissions, arising mainly from diesels, are other major sources of
23 PM25. Contributions from road dust and soils are relatively minor, typically constituting less
24 than 10% of PM2 5 in the studies shown in Table 4-4. The most notable difference in the relative
25 importance of major source categories of PM25 shown in Table 4-4 and PM10 shown in Table 4-5
26 involves crustal material, (e.g., soil, road dust) which represents about 40% on average of the
27 total mass of PM10 in the studies shown in Table 4-5. The fraction is higher in locations located
28 away from specific sources such as sea spray or smelters. Emissions of fugitive dust are
29 concentrated mainly in the PM(10_2 5) size range. The average fugitive dust source contribution is
30 highly variable among sampling sites within the same urban areas, as seen by differences
31 between the Central Phoenix (33 //g/m3) and Scottsdale (25 //g/m3) sites in Arizona, and it is also
October 1999 4-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 seasonally variable, as evidenced by the summer and fall contributions at Rubidoux, CA. The
2 variability in fugitive dust loadings reflects the sporadic nature of its emissions and its short
3 lifetime in the atmosphere.
4 In Table 4-5, primary motor vehicle exhaust contributions account for up to 40% of average
5 PM10 at many of the sampling sites. Vehicle exhaust contributions are also variable at different
6 sites within the same study area. The mean value and the variability of motor vehicle exhaust
7 contributions reflects the proximity of sampling sites to roadways and traffic conditions during
8 the time of sampling. Many studies were conducted during the late 1980s, when a portion of the
9 vehicle fleet still used leaded gasoline. Pb and Br in motor vehicle emissions facilitated the
10 distinction of motor vehicle contributions from other sources. Vehicles using leaded fuels have
11 higher emission rates than vehicles using unleaded fuels. Pb also poisons automobile exhaust
12 catalysts and produces adverse human health effects. As a result, Pb has been virtually
13 eliminated from vehicle fuels. However, recently organic species have replaced Pb as a source
14 marker for motor vehicle emissions (e.g., Rogge et al., 1993a).
15 Marine aerosol is found, as expected, at coastal sites such as Long Beach (average 3.8% of
16 total mass), and San Nicolas Island (25%). These contributions are relatively variable and are
17 larger at the more remote sites. Individual values reflect proximity to local sources. Of great
18 importance are the contributions from secondary ammonium sulfate in the eastern United States
19 and ammonium nitrate in the western United States. These are especially noticeable at sites in
20 California's San Joaquin Valley (Bakersfield, Crows Landing, Fellows, Fresno, Kern Wildlife,
21 and Stockton) and in the Los Angeles area.
22 Samples selected for chemical analysis are often biased toward the highest PM10 mass
23 concentrations in the studies shown in Table 4-5, so average source contribution estimates are
24 probably not representative of annual averages. Quoted uncertainties in the estimated
25 contributions of the individual sources shown in Table 4-4 range from 10 to 50%. Uncertainties
26 of source contribution estimates are not usually reported with the average values summarized in
27 Table 4-5. Estimates of standard errors are calculated in source apportionment studies, and
28 typically range from 15 to 30% of the source contribution estimate. They are much higher when
29 the chemical source profiles for different sources are highly uncertain or too similar to
30 distinguish one source from another.
31
October 1999 4-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.4 EMISSIONS ESTIMATES AND THEIR UNCERTAINTIES
2 In principle, source contributions to ambient PM could also be estimated on the basis of
3 predictions made by chemistry-transport models (CTM) or even on the basis of emissions
4 inventories alone. Uncertainties in emissions inventories have arguably been regarded as
5 representing the largest source of uncertainty in CTMs (Calvert et al., 1993). A number of
6 factors limit the ability of an emissions inventory driven CTM to determine the effects of various
7 sources on particle samples obtained at a particular location, apart from uncertainties in the
8 inventories given above. Air pollution model predictions represent averages over the area of a
9 grid cell, which in the case of the Urban Airshed Model typically has been 25 km2
10 (5 km x 5 km). The contributions of sources to pollutant concentrations at a monitoring site are
11 controlled strongly by local conditions which cannot be resolved by an Eulerian grid-cell model.
12 Examples would be the downward mixing of tall stack emissions and deviations from the mean
13 flow caused by buildings. The impact of local sources at a particular point in the model domain
14 may not be predicted accurately, because their emissions would be smeared over the area of a
15 grid cell or if the local wind fields at the sampling point deviated significantly from the mean
16 wind fields calculated by the model.
17
18 4.4.1 Emissions Estimates for Primary Particulate Matter and SO2, NOX, and
19 VOCs in the United States
20 Estimated emissions of primary PM25 from different sources in the United States are
21 summarized in Figure 4-14. The estimates are based on information presented in the U.S. EPA
22 National Air Pollutant Emission Trends Report, 1997 (U.S. Environmental Protection Agency,
23 1998) to which the reader is referred for detailed tables showing trends in PM25 emissions from a
24 number of source categories from!990 to 1997; descriptions of the methodology used in the
25 construction of these tables; and descriptions of the uncertainties involved in the estimation
26 process. This document also provides information about emissions of PM10, sulfur dioxide
27 (SO2), nitrogen oxides (NOX ), volatile organic compounds (VOCs), and ammonia (NH3).
28 Estimated total nationwide emissions of primary PM25 were 7.5 Tg yr"1 in 1997. The
29 category of fossil fuel combustion referred to in Figure 4-14 includes fossil fuel burning by
30 electric utilities, industry, and residences. The industry category includes contributions from
31 metals processing, petroleum refining, agricultural products processing, mining, and the storage
October 1999 4-33 DRAFT-DO NOT QUOTE OR CITE
-------
PM25 Total Emissions (1997) = 18 Tg yr1
On-Road
2.5%
Incinerator
.07%
Industry
6.3%
Fossil Fuel
Combustion
5.8%
Biomass
Burning
17.5%
Non-Road
4.9%
Wind Erosion
9.6%
Agriculture
11.1%
Fugitive Dust
41.7%
Figure 4-14. Nationwide emissions of PM2 5 from various source categories.
Source: U.S. Environmental Protection Agency (1998).
1 and transport of industrial goods. Incineration refers to the burning of non-biomass waste by
2 residences and municipalities. The on-road vehicles category includes contributions from gas
3 and diesel powered vehicles. Non-road engines and vehicles include their emissions in
4 transportation, construction, and other commercial, industrial and recreational activities. Wind
5 erosion refers to the raising of mineral dust by the wind. The biomass burning category includes
6 contributions from residential wood burning, open burning of vegetation, and forest fires.
7 The agriculture category includes mainly emissions of soil dust related to the production of
October 1999
4-34
DRAFT-DO NOT QUOTE OR CITE
-------
1 agricultural crops and livestock. Fugitive dust refers mainly to mineral dust raised by on-road
2 and non-road vehicles during their operation. As can be seen from inspection of Figure 4-14, the
3 raising of mineral dust by wind erosion, agriculture, and as fugitive dust constitutes the largest
4 primary source (62.4%) of PM25 on a nationwide basis. Biomass burning constitutes the second
5 largest primary source (17.5%) of PM25. The gross composition of emissions from most of these
6 categories is summarized in Table 4-3. Total emissions of PM25 as well as contributions from
7 individual source categories were relatively constant over the period from 1990 to 1997 (U.S.
8 Environmental Protection Agency, 1998). While crustal dust constitute over 60% of the total
9 PM25 inventory, they constitute less than about 10% of the source strengths inferred from the
10 receptor modeling studies shown in Table 4-4. However, it should be remembered that
11 secondary components often represent the major fraction of ambient samples. Dust sources
12 constitute 88% of the annual average PM10 National Emissions Inventory (U.S. Environmental
13 Protection Agency, 1994), but they average more than 50% of the contribution to average PM10
14 concentrations in only about 40% of the entries shown in Table 4-5. The reasons for this
15 apparent discrepancy are not clear. In addition to errors in inventories or source apportionments,
16 weather-related factors (wind speed and ground wetness) and the dominance of local sources on
17 spatial scales too small to be captured in inventories may be involved. It should be remembered
18 that dust emissions are widely dispersed and highly sporadic. Dust particles also have short
19 atmospheric residence times and as a result their dominance in emissions inventories may not be
20 reflected in samples collected near specific sources.
21 The geographic distribution of primary PM25 emissions is shown in Figure 4-15 and the
22 distribution of primary PM10 emissions is shown in Figure 4-16. As may be expected, high
23 emissions of PM2 5 are centered around many large urban areas. Although mineral dust sources
24 account for most of the emissions, their contributions are distributed much more widely than are
25 those from combustion sources. The nature of the distribution of mineral dust sources is shown
26 more clearly in Figure 4-16 where high densities of PM10 emissions are also found in sparsely
27 populated areas.
28 Estimated contributions from individual sources to emissions of gaseous precursors to
29 secondary PM formation are summarized in Figure 4-16 for SO2, NOX , VOCs, and NH3.
30 Information about the yield of particulate matter formed during the oxidation of VOCs is given in
31 Section 3.4.
October 1999 4-35 DRAFT-DO NOT QUOTE OR CITE
-------
mi.)
11 .11
f-11
4-f
=^'
I
- ,1
Figure 4-15. Distribution of primary PM25 emissions across the United States.
Source: U.S. Environmental Protection Agency (1998).
1 Although total emissions of gaseous precursors (SO2, NOX, VOC's, and NH3) are shown in
2 Figure 4-17, it should be remembered that these values cannot be directly translated into
3 production rates of particulate matter. Dry deposition and precipitation scavenging of some of
4 these gases can occur before they are converted to particulate matter in the atmosphere.
5 In addition, some fraction of these gases are transported outside of the domain of the continental
6 United States before being oxidized. Likewise, emissions of these gases from areas outside the
7 United States can result in the transport of their oxidation products into the United States. While
8 the chemical oxidation of SO2 will lead quantitatively to the formation of SO4=, the yield of
9 particulate matter from the oxidation of VOCs will be much less because only a small fraction of
October 1999
4-36
DRAFT-DO NOT QUOTE OR CITE
-------
rW,
>n
it -11
7-11
4-7
0-4
Figure 4-16. Distribution of primary PM10 emissions across the United States.
Source: U.S. Environmental Protection Agency (1998).
1 VOC's react to form particles, and those that do have efficiencies less than 10% (c.f.
2 Section 3.4).
3 The values shown in this section are based on annually averaged quantities. However,
4 annual averages do not reflect the seasonality of a number of emissions categories. Residential
5 wood burning in fireplaces and stoves, for example, is a seasonal practice which reaches its peak
6 during cold weather. Cold weather also affects motor vehicle exhaust particulate emissions, both
7 in terms of chemical composition and emission rates (e.g., Watson et al., 1990a; Huang et al.,
8 1994). Planting, fertilizing, and harvesting are also seasonal activities. Forest fires occur mainly
9 during the local dry season and during periods of drought. Maximum dust production by wind
10 erosion in the United States occurs during the spring, while the minimum occurs during the
October 1999
4-37
DRAFT-DO NOT QUOTE OR CITE
-------
SO2 Total Emissions (1997)
= "ISTgyr-1
All Other
Non-Road Engines 7%
and Vehicles
5%
Metal Processing
3%
Fuel Combustion - Other
4%
Fuel Combustion - Industrial
17%
Fuel Combustion -
Electrical Utilities
NOX Total Emissions (1997)
= 21 TgyH
Fuel Combustion-
Industrial
14%
Fuel Combustion -
Electric Utility
26%
Fuel Combusion -
Other
5%
On Road Vehicles
30%
Non-Road Engines
and Vehicles
19%
VOCs Total Emissions (1997)
= 17Tgyr-1
All Other
19%
Non-Road Engines
and Vehicles
13%
NH3 Total Emissions (1997)
= 2.9Tgyr1
All Other
On-Road Vehicles 4-2%
7.6%
Waste Water
Treatment
3.1%
Agricultural Chemical
Manufacturini
6.1%
On Road Vehicles
27%
Figure 4-17. Nationwide emissions of SO2, NOX, VOCs, and NH3 from various source
categories.
Source: U.S. Environmental Protection Agency (1998).
1
2
3
summer (Gillette and Hanson, 1989). Effects are being made to account for the seasonal
variations of emissions in the nationwide emissions inventories.
October 1999
4-38
DRAFT-DO NOT QUOTE OR CITE
-------
1 4.4.2 Uncertainties of Emissions Inventories
2 As described in CD96, it is difficult to assign uncertainties quantitatively to entries in
3 emissions inventories. Methods that can be used to verify or place constraints on emissions
4 inventories are sparse. In general, the overall uncertainty in the emissions of a given pollutant
5 includes contributions from all of the terms used to calculate emissions, i.e., activity rates,
6 emissions factors, and control device efficiencies. Additional uncertainties arise during the
7 compilation of an emissions inventory because of missing sources and computational errors.
8 The variability of emissions can cause errors when annual average emissions are applied to
9 applications involving shorter time scales.
10 Activity rates for well-defined point sources (e.g., power plants) should have the smallest
11 uncertainty associated with their use, since accurate production records need to be kept. On the
12 other hand, activity rates for a number of areally dispersed fugitive sources are extremely
13 difficult to quantify. Emissions factors for easily measured fuel components which are released
14 quantitatively during combustion (e.g., CO2 and SO2) should be the most reliable. Emissions of
15 components formed during combustion are more difficult to characterize as the emissions rates
16 are dependent on factors specific to individual combustion units and on combustion stage (i.e.,
17 smoldering or active). Although the AP-42 emissions factors (U.S. Environmental Protection
18 Agency, 1995) contain extensive information for a large number of source types, these data are
19 very limited in the number of sources sampled. The efficiency of control devices is determined
20 by their age, their maintenance history, and operating conditions. It is virtually impossible to
21 assign uncertainties in control device performance due to these factors. It should be noted that
22 the largest uncertainties occur for those devices which have the highest efficiencies (>90%).
23 This occurs because the efficiencies are subtracted from one and small errors in assigning
24 efficiencies can lead to large errors in emissions.
25 Ideally an emissions inventory should include all major sources of a given pollutant. This
26 may be an easy task for major point sources, but becomes problematic for poorly characterized
27 area sources of both primary PM and precursors to secondary PM formation. Further research is
28 needed to better characterize the sources of pollutants in order to reduce this source of
29 uncertainty. Errors can arise from the misreporting of data, and arithmetic errors can occur in the
30 course of compiling entries from thousands of individual sources. A quality assurance program
31 is required to check for outliers and arithmetic errors.
October 1999 4-39 DRAFT-DO NOT QUOTE OR CITE
-------
1 Because of the variability in emissions rates, there can be errors in the application of
2 inventories developed on an annually averaged basis (as are the inventories shown in
3 Figures 4-14 to 4-17) to episodes occurring on much shorter time scales. As an example, most
4 modeling studies of air pollution episodes are carried out for periods of a few days.
5 Uncertainties in annual emissions were estimated to range from 4 to 9% for SO2 and from
6 6 to 11% for NOX in the 1985 NAPAP inventories for the United States (Placet et al, 1991).
7 Uncertainties in these estimates increase as the emissions are disaggregated both spatially and
8 temporally. The uncertainties quoted above are minimum estimates and refer only to random
9 variability about the mean, assuming that the variability in emissions factors was adequately
10 characterized and that extrapolation of emissions factors to sources other than those for which
11 they were measured is valid. The estimates do not consider the effects of weather or variations in
12 operating and maintenance procedures. Fugitive dust sources, as mentioned above, are extremely
13 difficult to quantify, and stated emission rates may represent only order-of-magnitude estimates.
14 As rough estimates, uncertainties in emissions estimates could be as low as 10% for the best
15 characterized source categories, while emissions figures for windblown dust should be regarded
16 as order-of-magnitude estimates. Given (a) uncertainties in the deposition of SO2 and its
17 oxidation rate, (b) the variability seen in OC and EC emissions from motor vehicles along with
18 the findings from past verification studies for NMHC and CO to NOX ratios, (c) ranges of values
19 found among independent estimates for emissions of individual species (NH3, OC), and (d) the
20 predominance of fugitive emissions, PM emissions rates should be regarded as
21 order-of-magnitude estimates.
22 There have been few field studies designed to test emissions inventories observationally.
23 The most direct approach would be to obtain cross-sections of pollutants upwind and downwind
24 of major urban areas from aircraft. The computed mass flux through a cross section of the urban
25 plume can then be equated to emissions from the city chosen. This approach has been attempted
26 on a few occasions. Results have been ambiguous because of contributions from fugitive
27 sources, non-steady wind flows, and general logistic difficulties.
28
29
30
31
October 1999 4-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.5 LONG RANGE TRANSPORT OF PM FROM SOURCES OUTSIDE
2 THE UNITED STATES
3 Apart from sources within the continental United States, particulate matter can be brought
4 in by long range transport from sources outside the United States, as evidenced by the transport
5 of PM from uncontrolled biomass burning in Central America and southern Mexico which
6 resulted in anomalously high PM levels observed in southern Texas and to a lesser extent,
7 throughout the entire central and southeastern United States during the spring and early summer
8 of 1998. Windblown dust from individual dust storms in the Sahara desert has been observed in
9 satellite images as plumes crossing the Atlantic Ocean and reaching the southeast coast of the
10 United States (e.g., Ott et al., 1991). Dust transport from the deserts of Asia across the Pacific
11 Ocean also occurs (Prospero, 1995). Most dust storms in the deserts of China occur in the spring
12 after the snow has melted and before vegetation has grown following the passage of strong cold
13 fronts. Strong winds and unstable conditions result in the rapid transport of dust into the middle
14 and upper troposphere (4-5 km altitude) where it is transported by strong westerly winds out over
15 the Pacific Ocean (Duce, 1995). Satellite images have been used to track the progress of a dust
16 cloud from the Gobi desert to the northwestern United States during the spring of 1998.
17 Satellite images obtained at visible wavelengths cannot track mineral dust across the
18 continents because of a lack of contrast between the plume and the underlying surface. Other
19 means must be used to track the spread of North African dust through the eastern United States.
20 Perry et al. (1997) used two criteria (PM2 5 soil concentration > 3 //g m"3 and Al/Ca > 3.8) to
21 distinguish between soil of local origin from soil originating in North Africa in characterizing the
22 sources of PM in aerosol samples collected in the IMPROVE (Interagency Monitoring of
23 Protected Visual Environments) network. Their analysis indicates that incursions of Saharan
24 dust into the continental United States have occurred, on average, about three times per year from
25 1992 to 1995. These events have persisted for about ten days principally during the summer.
26 As can be expected, the frequency of dust events is highest in the southeastern United States and
27 about half are observed only within the state of Florida and these are associated with dense hazes
28 in Miami during the summer (Prospero et al., 1987) such that African dust is the dominant
29 aerosol constituent in south Florida during the summer (Prospero, 1999). The mass median
30 diameter of mineral dust over the oceans is typically between 2 and 3 //m (Duce, 1995).
31 North African dust has been tracked as far as Illinois (Gatz and Prospero, 1996) and to Maine
October 1999 4-41 DRAFT-DO NOT QUOTE OR CITE
-------
1 (Perry et al., 1997). Larger scale events typically covered from 15 to 30% of the area of the
2 continental United States and resulted in increases of PM25 levels of 8.7 ± 2.3 m m"3 throughout
3 the affected areas with mean maximum dust contributions of 19.7 ± 8.4 mg m"3 during these
4 events, and a peak contribution of 32 mg m"3 to 24-h average PM2 5 levels.
5
6
7 4.6 SUMMARY AND CONCLUSIONS
8 Ambient particulate matter contains both primary and secondary components. The results
9 of ambient monitoring studies and receptor modeling studies in the eastern United States indicate
10 that PM2 5 is dominated by secondary components. Secondary constituents are smaller but still
11 important components of PM25 in the central and western United States. Minerals constitute the
12 largest fraction of PM(10_25) throughout the United States. Data collected in the Los Angeles
13 Basin and Philadelphia suggest that secondary PM components are more uniformly distributed
14 than are primary components. Compositional data obtained at multiple sites in other urban areas
15 are sparse.
16 Due to the complexity of the composition of ambient PM25 and PM(10_25), sources are best
17 discussed in terms of individual constituents of both primary and secondary PM2 5 and PM(10_2 5).
18 Each of these constituents can have anthropogenic and natural sources, as shown in Table 4-3.
19 The distinction between natural and anthropogenic sources is not always obvious. While
20 windblown dust might seem to be the result of natural processes, highest emission rates are
21 associated with agricultural activities in areas that are susceptible to periodic drought. Examples
22 include the dust bowl region of the midwestern United States and the Sahel of Africa. Most
23 forest fires in the United States may ultimately be of human origin, either through prescribed
24 burning or accident.
25 Emissions inventories are generally not the most appropriate way to apportion material in
26 ambient samples. Receptor modeling has proven to be an especially valuable tool in this regard.
27 Compositional profiles developed for receptor modeling applications are perhaps the most
28 accessible and reliable means of characterizing the composition of emissions. The results of
29 receptor modeling studies throughout the United States indicate that the combustion of fossil and
30 biomass fuels is a major source of PM25. Fugitive dust, found mainly in the PM(10_25) range size,
31 represents the largest source of PM10 in many locations in the western United States. Quoted
October 1999 4-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 uncertainties in source apportionments of constituents in ambient aerosol samples typically range
2 from 10 to 50%. It is apparent that a relatively small number of source categories, compared to
3 the total number of chemical species which are typically measured in ambient monitoring-source
4 receptor model studies, are needed to account for most of the observed mass of PM in these
5 studies. Again, it should be emphasized that, because of limitations in receptor modeling
6 methods in treating secondary components, these efforts are more likely to be successful for
7 primary components, although it should be mentioned that methods are being developed to
8 apportion secondary constituents by source categories.
9 Windblown dust from whatever source represents the largest single source of PM25 in the
10 U.S. emissions inventory. Although dust emissions (62% of total U.S. PM25) are far in excess of
11 any other source of primary or secondary PM2 5 in any region of the country, measurements of
12 soil constituents in ambient samples suggest that the overall contribution from this source could
13 be less than 10%. The reasons for this apparent discrepancy are not clear. In addition to errors in
14 inventories or source apportionments, weather-related factors (wind speed and ground wetness)
15 and the dominance of local sources on spatial scales too small to be captured in inventories may
16 be involved. It should be remembered that dust emissions are widely dispersed and highly
17 sporadic. Dust particles also have short atmospheric residence times and, as a result, their
18 dominance in emissions inventories may not be reflected in samples collected away from the
19 sources of the dust.
20 As seen in Table 4-3, emissions of mineral dust, organic debris, and sea spray are
21 concentrated mainly in the coarse fraction of PM10 ( > 2.5 //m aero. diam.). A small fraction of
22 this material is in the PM25 size range ( < 2.5 //m aero. diam.). Nevertheless, concentrations of
23 crustal material can be appreciable especially during dust events. It should also be remembered
24 that virtually all of the Saharan dust reaching the United States is in the PM2 5 size range.
25 Emissions from combustion sources (mobile and stationary sources, biomass burning) are
26 predominantly in the PM2 5 size range.
27 Uncertainties in emissions inventories are difficult to quantify. They may be as low as 10%
28 for well-defined sources (e.g., for SO2) and may range up to a factor of 10 or so for windblown
29 dust. As a rule, total PM emissions rates should be regarded as order-of-magnitude estimates.
30 Because of the large uncertainty associated with emissions of suspended dust, trends of total
October 1999 4-43 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM10 emissions should be viewed with caution and emissions from specific source categories are
2 best discussed on an individual basis.
3 Although most emphasis in this chapter has been on sources within the United States,
4 it should also be remembered that sources outside the United States also contribute to ambient
5 PM levels that can, at times, result in exceedances of the ambient NAAQS for PM. Saharan dust
6 storms contribute routinely to PM loadings in areas east of the Mississippi River. The results of
7 Perry et al. (1997) indicate that highest concentrations of mineral dust in the PM25 fraction are
8 found in the eastern United States during the summer and not in arid areas of the western
9 United States. Large scale dust storms in the deserts of central Asia have recently been found to
10 contribute to PM levels in the Northwest on an episodic basis. Uncontrolled biomass burning in
11 central America and Mexico resulted in exceedances of the daily NAAQS for PM in Texas.
12
October 1999 4-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 AIRS (1995).
4 Cahill, T. A. (1996) [Data from IMPROVE data base.] Fort Collins, CO: Air Resource Specialists, Inc.
5 California Air Resources Board. (1995) [Data base provided by John R. Holmes, Chief, Research Division.]
6 Sacramento, CA: California Air Resources Board.
7 Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality: some
8 reflections on controlling vehicle emissions. Science (Washington, DC) 261: 37-45.
9 CAPITA (1999) Available: http://capita.wustl.edu/CAPITA/.
10 Chow, J. C.; Watson, J. G.; Frazier, C. A.; Egami, R. T.; Goodrich, A.; Ralph, C. (1988) Spatial and temporal
11 source contributions to PM10 and PM2.5 in Reno, NV. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10:
12 implementation of standards, an APCA/EPA international specialty conference; February; San Francisco,
13 CA. Pittsburgh, PA: Air Pollution Control Association; pp. 438-457. (APCA transactions series TR-13).
14 Chow, J. C.; Watson, J. G.; Richards, L. W.; Haase, D. L.; McDade, C.; Dietrich, D. L.; Moon, D.; Sloane, C.
15 (1991) The 1989-90 Phoenix PM10 study, volume II: source apportionment final report. Phoenix, AZ:
16 Arizona Department of Environmental Quality; final report no. 8931.6F1; Table 3-1.
17 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Frazier, C. A.; Hinsvark, B. A.; Prichett, L. C.; Neuroth, G. R.
18 (1992a) Wintertime PM 10 and PM2 5 chemical compositions and source contributions in Tucson, Arizona.
19 In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source controls, an
20 A&WMA/EPA international specialty conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air &
21 Waste Management Association; pp. 231-243. (A&WMA transactions series no. 22).
22 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L. W.
23 (1992b) PM10 source apportionment in California's San Joaquin Valley. Atmos. Environ. Part A
24 26: 3335-3354.
25 Chow, J. C.; Liu, C. S.; Cassmassi, J.; Watson, J. G.; Lu, Z.; Pritchett, L. C. (1992c) A neighborhood-scale study of
26 PM10 source contributions in Rubidoux, California. Atmos. Environ. Part A 26: 693-706.
27 Cooper, J. A.; Watson, J. G., Jr. (1980) Receptor oriented methods of air particulate source apportionment. J. Air
28 Pollut. Control Assoc. 30: 1116-1125.
29 Darlington, T. L.; Kahlbaum, D. F.; Heuss, J. M.; Wolff, G. T. (1997) Analysis of PM10 trends in the United States
30 from 1988 through 1995. J. Air Waste Manage. Assoc. 47: 1070-1078.
31 Dattner, S. L.; Hopke, P. K. (1982) Receptor models applied to contemporary pollution problems: an introduction.
32 In: Frederick, E. R., ed. A specialty conference on: receptor models applied to contemporary pollution
33 problems; October; Danvers, MA. Pittsburgh, PA: Air Pollution Control Association; pp. 1-5. (APCA
34 specialty conference proceedings SP-48).
35 Duce, R. A. (1995) Sources, distributions, and fluxes of mineral aerosols and their relationship to climate.
36 In: Charlson, R. J.; Heintzenberg, J., eds. Aerosol forcing of climate: report of the Dahlem workshop on
37 aerosol forcing of climate; April 1994; Berlin, Federal Republic of Germany. Chichester, United Kingdom:
38 John Wiley & Sons, Ltd.; pp. 43-72.
39 Dzubay, T. G.; Stevens, R. K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E.; Courtney, W. J. (1988) A composite
40 receptor method applied to Philadelphia aerosol. Environ. Sci. Technol. 22: 46-52.
41 Eldred, R. A.; Cahill, T. A. (1994) Trends in elemental concentrations of fine particles at remote sites in the United
42 States of America. Atmos. Environ. 28: 1009-1019.
43 Gatz, D. F.; Prospero, J. M. (1996) A large silicon-aluminum aerosol plume in central Illinois: North African Desert
44 dust? Atmos. Environ. 30: 3789-3799.
45 Gillette, D. A.; Hanson, K. J. (1989) Spatial and temporal variability of dust production caused by wind erosion in
46 the United States. J. Geophys. Res. [Atmos.] 94: 2197-2206.
47 Glover, D. M.; Hopke, P. K.; Vermette, S. J.; Landsberger, S.; DAuben, D. R. (1991) Source apportionment with
48 site specific source profiles. J. Air Waste Manage. Assoc. 41: 294-305.
49 Gordon, G. E. (1980) Receptor models. Environ. Sci. Technol. 14: 792-800.
50 Gordon, G. E. (1988) Receptor models. Environ. Sci. Technol. 22: 1132-1142.
51
October 1999 4-45 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gray, H. A.; Landry, B.; Liu, C. S.; Henry, R. C.; Cooper, J. A.; Sherman, J. R. (1988) Receptor modeling for
2 PM10 source apportionment in the South Coast Air Basin of California. In: Mathai, C. V.; Stonefield, D. H.,
3 eds. PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
4 San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 399-418. (APCA transactions
5 series TR-13).
6 Griffing, G. W. (1980) Relations between the prevailing visibility, nephelometer scattering coefficient and
7 sunphotometer turbidity coefficient. Atmos. Environ. 14: 577-584.
8 Hopke, P. K. (1985) Receptor modeling in environmental chemistry. New York, NY: John Wiley & Sons.
9 Hopke, P. K., ed. (1991) Receptor modeling for air quality management. Amsterdam, The Netherlands: Elsevier.
10 (Vandeginste, B. G. M.; Kvalheim, O. M., eds. Data handling in science and technology: v. 7).
11 Hopke, P. K.; Wlaschin, W.; Landsberger, S.; Sweet, C.; Vermette, S. J. (1988) The source apportionment of PM10
12 in south Chicago. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10: implementation of standards, an
13 APCA/EPA international specialty conference; February; San Francisco, CA. Pittsburgh, PA: Air Pollution
14 Control Association; pp. 484-494. (APCA transactions series TR-13).
15 Houck, J. E.; Rau, J. A.; Body, S.; Chow, J. C. (1992) Source apportionment: Pocatello, Idaho PM 10 nonattainment
16 area. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source controls, an
17 A&WMA/EPA international specialty conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air &
18 Waste Management Association; pp. 219-230. (A&WMA transactions series no. 22).
19 Huang, X.; Olmez, I.; Aras, N. K.; Gordon, G. E. (1994) Emissions of trace elements from motor vehicles:
20 potential marker elements and source composition profile. Atmos. Environ. 28: 1385-1391.
21 Husar, R. B.; Wilson, W. E. (1993) Haze and sulfur emission trends in the eastern United States. Environ. Sci.
22 Technol. 27: 12-16.
23 Husar, R. B.; Holloway, J. M.; Patterson, D. E.; Wilson, W. E. (1981) Spatial and temporal pattern of eastern U.S.
24 haziness: a summary. Atmos. Environ. 15: 1919-1928.
25 Husar, R. B.; Elkins, J. B.; Wilson, W. E. (1994) U.S. visibility trends, 1960-1992, regional and national. Presented
26 at: 87th annual meeting of the Air & Waste Management Association; June; Cincinnati, OH. Pittsburgh, PA:
27 Air & Waste Management Association; paper no. 94-MP3.05.
28 Inhalable Particulate Network. (1985) [Inhalable Particulate Network Database.] Research Triangle Park, NC: U.S.
29 Environmental Protection Agency, National Exposure Research Laboratory; Jose M. Sune, contact.
30 Javitz, H. S.; Watson, J. G.; Guertin, J. P.; Mueller, P. K. (1988) Results of a receptor modeling feasibility study.
31 JAPCA 38: 661-667.
32 Kim, B. M.; Zeldin, M. D.; Liu, C. S. (1992) Source apportionment study for state implementation plan
33 development in the Coachella Valley. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional
34 particulate source controls: an A&WMA/EPA international specialty conference, v. II; January; Scottsdale,
35 AZ. Pittsburgh, PA: Air & Waste Management Association; pp. 979-991. (A&WMA transactions series
36 no. 22).
37 Koutrakis, P. (1995) [PM25 and PM10 data for Philadelphia, 1992-1995]. Boston, MA: Harvard School of Public
38 Health.
39 Lawson, D. R.; Smith, R. E. (1998) The northern front range air quality study: a report to the Governor and General
40 Assembly. Fort Collins, CO: Colorado State University; December.
41 Macias, E. S.; Hopke, P. K., eds. (1981) Atmospheric aerosol: source/air quality relationships; based on a
42 symposium jointly sponsored by the Divisions of Nuclear Chemistry and Technology at the 180th national
43 meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington, DC: American
44 Chemical Society. (Comstock, M. J., ed. ACS symposium series: 167).
45 Magliano, K. L. (1988) Level 1 PM 10 assessment in a California air basin. In: Mathai, C. V.; Stonefield, D. H., eds.
46 PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
47 San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 508-517. (APCA transactions
48 series TR-13).
49 Neas, L. M. (1996) [Dataprovided for 1986 from Harvard six cities database.] Boston, MA: Harvard School of
50 Public Health.
51 Ott, S.-T.; Ott, A.; Martin, D. W.; Young, J. A. (1991) Analysis of a trans-Atlantic Saharan dust outbreak based on
52 satellite and GATE data. Mon. Weather Rev. 119:1832-1850.
53 Pace, T. G., ed. (1986) Receptor methods for source apportionment: real world issues and applications, an APCA
54 international specialty conference; March 1985; Williamsburg, VA. Pittsburgh, PA: Air Pollution Control
55 Association. (APCA transactions no. TR-5).
October 1999 4-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 Perry, K. D.; Cahill, T. A.; Eldred, R. A.; Dutcher, D. D.; Gill, T. E. (1997) Long-range transport of North African
2 dust to the eastern United States. J. Geophys. Res. [Atmos] 102: 11,225-11,238.
3 Pinto, J. P.; Pleasant, M.; Kellogg, R.; Torok, Sz.; Stiles, D.; Willis, R. D.; Koutrakis, P.; Allen, G. A. (1995)
4 Chemical characterization of the ambient aerosol in Philadelphia. Presented at: Particulate matter: health and
5 regulatory issues: an A&WMA international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA:
6 Air & Waste Management Association; submitted abstract.
7 Placet, M.; Battye, R. E.; Fehsenfeld, F. C.; Bassett, G. W. (1991) Emissions involved in acidic deposition
8 processes. In: Irving, P. M., ed. Acidic deposition: state of science and technology, volume I: emissions,
9 atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation Assessment
10 Program. (State of science and technology report no. 1).
11 Prospero, J. M. (1995) The atmospheric transport of particles to the ocean. In: Ittekkot, V.; Honjo, S.; Depetris,
12 P. J., eds. SCOPE report: particle flux in the ocean. New York, NY: John Wiley & Sons, Ltd.; in press.
13 Prospero, J. M. (1999) Long-range transport of mineral dust in the global atmosphere: impact of African dust on the
14 environment of the southeastern United States. Proceed. Natl. Acad. Sci. 96: 3396-3403.
15 Prospero, J. M.; Nees, R. T.; Uematsu, M. (1987) Deposition rate of particulate and dissolved aluminum derived
16 from Saharan dust in precipitation at Miami, Florida. J. Geophys. Res. [Atmos.] 92: 14,723-14,731.
17 Rodes, C. E.; Evans, E. G. (1985) Preliminary assessment of 10 pan particulate sampling at eight locations in the
18 United States. Atmos. Environ. 19: 293-303.
19 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
20 aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
21 Technol. 27:636-651.
22 Ryan, W. M.; Badgett-West, C. R.; Holtz, D. R.; Peters, T. A.; Cooper, J. A.; Ono, D. (1988) Reconciliation of
23 receptor and dispersion modeling impacts of PM 10 in Hayden, Arizona. In: Mathai, C. V.; Stonefield, D. H.,
24 eds. PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
25 San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 419-429. (APCA transactions
26 series TR-13).
27 Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R. (1996) Source apportionment of
28 airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
29 Skidmore, L. W.; Chow, J. C.; Tucker, T. T. (1992) PM 10 air quality assessment for the Jefferson County, Ohio air
30 quality control region. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source
31 controls, an A&WMA/EPA international specialty conference, v. II; January; Scottsdale, AZ. Pittsburgh, PA:
32 Air and Waste Management Association; pp. 1016-1031. (A&WMA transactions series no. 22).
33 Spengler, J. D.; Briggs, S. L. K.; Ozkaynak, H. (1986) Relationship between TSP measurements and
34 size-fractionated particle mass measurements in six cities participating in the Harvard Air Pollution Health
35 Study. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning
36 and Standards; EPA cooperative agreement no. CR-812667-01-0; December 5.
37 Stevens and Lewis (1984).
38 Stevens, R. K.; Pace, T. G. (1984) Overview of the mathematical and empirical receptor models workshop
39 (Quail Roost II). Atmos. Environ. 18: 1499-1506.
40 Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1995) Spatial variation in acidic sulfate and ammonia
41 concentrations within metropolitan Philadelphia. J. Air Waste Manage. Assoc. 45: 442-452.
42 Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM 10 samplers and
43 source apportionment of ambient PM 10 concentrations at Rillito, Arizona. In: Chow, J. C.; Ono, D. M., eds.
44 PM10 standards and nontraditional particulate source controls, an A&WMA/EPA international specialty
45 conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air & Waste Management Association; pp.
46 244-261. (A&WMA transactions series no. 22).
47 U.S. Environmental Protection Agency. (1994) National air quality and emissions trends report, 1993. Research
48 Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-94/026. Available
49 from: NTIS, Springfield, VA; PB95-159893.
50 U.S. Environmental Protection Agency. (1995) Compilation of air pollutant emission factors. Volume I: stationary
51 point and area sources. 5th ed. Research Triangle Park, NC: Office of Air Quality Planning and Standards;
52 report no. AP-42-ED-5. Available from: NTIS, Springfield, VA; PB95-196028.
53 U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
54 NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001 aF-cF. 3v.
55 Available from: NTIS, Springfield, VA; PB96-168224.
October 1999 4-47 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Environmental Protection Agency. (1998) National air quality and emissions trends report, 1997. Research
2 Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-98-016. Available:
3 www.epa.gov/oar/aqtrnd97/.
4 Vermette, S. J.; Williams, A. L.; Landsberger, S. (1992) PM 10 source apportionment using local surface dust
5 profiles: examples from Chicago. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional
6 particulate source controls, an A&WMA/EPA international specialty conference, v. I; January; Scottsdale,
7 AZ. Pittsburgh, PA: Air & Waste Management Association; pp. 262-278. (A&WMA transactions series
8 no. 22).
9 Watson, J. G.; Henry, R. C.; Cooper, J. A.; Macias, E. S. (1981) The state of the art of receptor models relating
10 ambient suspended particulate matter to sources. In: Macias, E. S.; Hopke, P. K., eds. Atmospheric aerosol:
11 source/air quality relationships; based on a symposium jointly sponsored by the Divisions of Nuclear
12 Chemistry and Technology and Environmental Chemistry at the 180th national meeting of the American
13 Chemical Society; August 1980; Las Vegas, NV. Washington, DC: American Chemical Society; pp. 89-106.
14 (Comstock, M. J., ed. ACS symposium series: 167).
15 Watson, J. G.; Chow, J. C.; Mathai, C. V. (1989) Receptor models in air resources management: a summary of the
16 APCA international specialty conference. JAPCA 39: 419-426.
17 Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Burns, S.; Ragazzi, R. A. (1990a) Composite source
18 profiles for particulate motor vehicle exhaust source apportionment in Denver, CO. In: Mathai, C. V., ed.
19 Visibility and fine particles: an A&WMA/EPA international specialty conference; October 1989; Estes Park,
20 CO. Pittsburgh, PA: Air & Waste Management Association; pp. 422-436. (A&WMA transactions series
21 no. TR-17).
22 Watson, J. G.; Chow, J. C.; Lowenthal, D. H.; Pritchett, L. C.; Frazier, C. A.; Neuroth, G. R.; Robbins, R. (1994a)
23 Differences in the carbon composition of source profiles for diesel- and gasoline-powered vehicles.
24 Atmos. Environ. 28: 2493-2505.
25 Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
26 epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
27 Wongphatarakul(1997).
28 Wongphatarakul, V.; Friedlander, S. K.; Pinto, J. P. (1998) A comparative study of PM 2 5 ambient aerosol chemical
29 databases. Environ. Sci. Technol. 32: 3926-3934.
October 1999 4-48 DRAFT-DO NOT QUOTE OR CITE
-------
i APPENDIX 4A
2
3 Composition of PM Source Emissions
4
5
6 This appendix includes figures and tables showing the elemental composition of emissions
7 from various source categories discussed in Table 4A-3. The material is mainly derived from
8 Chapter 5 of CD96, to which the reader is referred for further details. The primary emphasis in
9 the figures and tables is on the source composition of PM2 5 particles.
10 The compositions of soils and average crustal material are shown in Table 4A-1 (adapted
11 from Warneck, 1988). Two entries are shown as representations of average crustal material.
12 Differences from the mean soil composition shown can result from local geology and climate
13 conditions. Major elements in both soil and crustal profiles are Si, Al, and Fe which are found in
14 the form of various minerals. In addition, organic matter constitutes a few percent, on average,
15 of soils. In general, the soil profile is similar to the crustal profiles, except for the depletion of
16 soluble elements such as Ca, Mg, Na, and K. It should be noted that the composition of soils
17 from specific locations can vary considerably from these global averages, especially for elements
18 like Ca, Mg, Na, and K.
19 Fugitive dust emissions arise from paved and unpaved roads, building construction and
20 demolition, parking lots, mining operations, storage piles, and agricultural tilling in addition to
21 wind erosion. Figure 4A-1 shows examples of size distributions in dust from paved and unpaved
22 roads, agricultural soil, sand and gravel, and alkaline lake bed sediments which were measured in
23 a laboratory resuspension chamber as part of a study in California (Chow et al., 1994). This
24 figure shows substantial variation in particle size among some of these fugitive dust sources.
25 The PMj 0 abundance (6.9%) in the total suspended PM (TSP) from alkaline lake bed dust is
26 twice its abundance in paved and unpaved road dust. Approximately 10% of the TSP is in the
27 PM25 fraction and approximately 50% of TSP is in the PM10 fraction. The sand/gravel dust
28 sample shows that 65% of the mass is in particles larger than the PM10 fraction. ThePM25
29 fraction of TSP is approximately 30% to 40% higher in alkaline lake beds and sand/gravel than
30 in the other soil types. The tests were performed after seiving and with a short (<1 min) waiting
31 period prior to sampling. It is expected that the fraction of PMj 0 and PM25 would increase with
October 1999 4A-1 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4A-1. AVERAGE ABUNDANCES OF MAJOR ELEMENTS IN
SOIL AND CRUSTAL ROCK
Elemental Abundances (ppmw)
Element
Si
Al
Fe
Ca
Mg
Na
K
Ti
Mn
Cr
V
Co
Soil
(a)
330,000
71,300
38,000
13,700
6,300
6,300
13,600
4,600
850
200
100
8
Crustal Rock
(b)
277,200
81,300
50,000
36,300
20,900
28,300
25,900
4,400
950
100
135
25
(c)
311,000
77,400
34,300
25,700
33,000
31,900
29,500
4,400
670
48
98
12
Source: (a) Vinogradov (1959); (b) Mason (1966); (c) Turekian (1971), Model A; as quoted in Warneck (1988).
1 distance from a fugitive dust emitter as the larger particles deposit to the surface faster than do
2 the smaller particles.
3 The size distribution of samples of paved road dust obtained from a source characterization
4 study in California is shown in Figure 4A-2. As might be expected, most of the emissions are in
5 the coarse size mode. The chemical composition of paved road dust obtained in Denver, CO,
6 during the winter of 1987-1988 is shown in Figure 4A-3. The chemical composition of paved
7 road dust consists of a complex mixture of particulate matter from a wide variety of sources.
8 Hopke et al. (1980) found that the inorganic composition of urban roadway dust in samples from
9 Urbana, IL could be described in terms of contributions from natural soil, automobile exhaust,
10 rust, tire wear, and salt. Automobile contributions arose from exhaust emissions enriched in Pb;
October 1999 4A-2 DRAFT-DO NOT QUOTE OR CITE
-------
100
80
60
(0
40
20
Paved
Road Dust
Unpaved
Road Dust
Agricultural
Soil
C3<1.0|jm
2.5 |
Soil/Gravel
OTSP
Alkaline
Lake Bed
Figure 4A-1. Size distribution of particles generated in a laboratory resuspension
chamber.
Source: Chow et al. (1994).
1
2
3
4
5
6
7
8
9
10
11
from rust as Fe; tire wear particles enriched in Zn; brake linings enriched in Cr, Ba, and Mn; and
cement particles derived from roadways by abrasion. In addition to organic compounds from
combustion and secondary sources, road dust also contains biological material such as pollen and
fungal spores.
The elemental composition of primary particulate matter emitted in the fine fraction from a
variety of power plants and industries in the Philadelphia area is shown in Table 4A-2 as a
representative example of emissions from stationary fossil combustion sources (Olmez et al.,
1988). Entries for the coal fired power plant show that Si and Al followed by sulfate are the
major primary constituents produced by coal combustion, while fractional abundances of
elemental carbon were much lower and organic carbon species were not detected. Sulfate is the
major particulate constituent released by the oil fired power plants examined in this study; and,
October 1999
4A-3
DRAFT-DO NOT QUOTE OR CITE
-------
100
80
60
0.
(0
0>
Q.
40
20
52.3%
92.8%
82.7%
(<2.5M)
81.6%
95.8%
93.1%
(<2.5M
92.4%
96.2%
(<10M)
92.3%
(<2.5M)
91.8%
99.2%
97.4%
(<2.5M)
87.4%
34.9%
10M)
Road and Agricultural Residential Diesel
Soil Dust Burning Wood Truck
Combustion Exhaust
Crude Oil Construction
Combustion Dust
Code:
Figure 4A-2. Size distribution of California source emissions, 1986.
Source: Houcketal. (1989, 1990).
October 1999
4A-4
DRAFT-DO NOT QUOTE OR CITE
-------
10'
10
10'
8
T5
C
D
.a
10
10
10
2
2
-3
Chemical Species
Figure 4A-3. Chemical abundances for PM25 emissions from paved road dust in Denver,
CO. Solid bars represent fractional abundances, and the error bars
represent variability in species abundances. Error bars represent detection
limits when there are no solid bars.
Source: Watson and Chow (1994).
1
2
3
4
5
6
7
again, elemental and organic carbon are not among the major species emitted. Olmez et al.
(1988) also compared their results to a number of similar studies and concluded that their data
could have much wider applicability to receptor model studies in other areas with some of the
same source types. The high temperature of combustion in power plants results in the almost
complete oxidation of the carbon in the fuel to CO2 and very small amounts of CO. A number of
trace elements are greatly enriched over crustal abundances (in different fuels), such as Se in coal
October 1999
4A-5
DRAFT-DO NOT QUOTE OR CITE
-------
o
o_
o
o
l-l
TABLE 4A-2. COMPOSITION OF FINE PARTICLES RELEASED BY
SOURCES IN THE PHILADELPHIA AREA
Sneries
Fddvstnne Cnal-
Oil-Fired Power Plants
^ (Units) Fired Power Plant N
"vO
[V
^^
1
ON
O
£>
Tl
H
O
O
0
H
0
c
o
H
W
O
O
H
W
C-v (%)
C-e (%)
NH4 (%)
Na (%)
Al (%)
Si (%)
P (%)
S (%)
SO4 (%)
Cl (%)
K(%)
Ca (%)
Sc (ppm)
Ti (%)
V (ppm)
Cr (ppm)
Mn (ppm)
Fe (%)
Co (ppm)
Ni (ppm)
ND
0.89 ±0.12
1.89 ±0.19
0.31 ±0.03
14 ±2
21.8±1.6
0.62 ±0.10
3.4 ±0.6
11.9±1.2
0.022 ±0.11
1.20 ±0.09
1.4 ±0.5
42 ±2
1.1 ±0.2
550 ± 170
390 ±120
290 ±15
7.6 ±0.4
93 ±10
380 ±50
3
3
3
3
9
9
9
3
3
9
3
3
3
3
3
3
3
3
9
Eddystone
2.7 ± 1.2
7.7 ±1.5
3.5 ±1.6
3.0 ±0.8
0.45 ± 0.09
1.9 ±0.6
1.5 ±0.4
11±2
40 ±4
0.019 ±0.009
0.16 ±0.05
3.6±1.0
0.17 ±0.02
0.040 ± 0.044
11500 ±3000
235 ±10
380 ±40
1.6 ±0.2
790 ±150
15000 ±5000
N
3
3
3
3
3
9
9
9
3
2
9
3
3
9
3
3
3
3
3
9
Schuylkill
0.75 ± 0.63
0.22 ±0.1 7
3.7 ±1.7
3.3 ±0.8
0.94 ± 0.08
2.6 ±0.4
1.0 ±0.2
13± 1
45 ±7
ND
0.21 ±0.03
2.3 ±1.0
0.47 ± 0.02
0.12 ±0.02
20000 ± 3000
230 ±70
210±50
1.7 ±0.4
1100 ±200
19000 ±2000
N
4
4
4
3
3
11
11
11
4
11
3
3
11
3
3
3
3
3
11
Secondary
Al Plant
1.6±1.5
0.18 ±0.10
2.2 ± 0.9
16.3 ±0.8
1.74 ±0.09
3.1 ±2.2
0.45 ± 0.27
3±4
5.9 ±2
21±4
10.9 ± 1.5
0.12 ±0.09
0.092 ± 0.039
0.024 ± 0.003
36 ±7
410±20
120± 15
0.31 ±0.02
13±2
300 ±100
VARIOUS STATIONARY
N
2
2
2
1
1
2
2
2
2
1
2
2
1
2
1
1
1
1
1
2
Fluid Cat.
Cracker
ND
0.16 ±0.05
0.43 ± 0.22
0.38 ± 0.05
6.8 ±1.2
9.8 ±20.0
ND
4.2 ± 12.6
38 ±4
ND
0.031 ±0.005
0.030 ± 0.004
2.7 ± 0.4
0.38 ±0.1
250 ±70
59 ±8
14±3
0.20 ± 0.03
15±2
220 ±30
N
3
3
3
3
9
9
3
9
9
3
3
3
3
3
9
3
9
Municipal
Incinerator
0.57 ±0.26
3.5 ±0.2
0.36 ±0.07
6.6 ±3.5
0.25 ±0.10
1.7 ±0.3
0.63 ±0.12
2.9 ±0.8
6.8 ±2.3
29 ±5
7.6 ±2.3
0.23 ±0.10
0.11 ±0.02
0.030 ±0.015
8.6 ±5.3
99 ±31
165 ±40
0.22 ± 0.05
3.7 ±0.8
290 ± 40
N
4
4
4
3
3
10
10
10
4
3
10
10
1
10
2
3
3
3
3
10
-------
0
o
o
o*
fD
•O
•O
•O
£
1
a
§
H
a
0
o
H
O
c
o
H
W
O
o
HH
H
W
TABLE 4A-2 (cont'd).
Species
(Units)
Cu (ppm)
Zn (%)
As (ppm)
Se (ppm)
Br (ppm)
Rb (ppm)
Sr (ppm)
Zr (ppm)
Mo (ppm)
Ag (ppm)
Cd (ppm)
In (ppm)
Sn (ppm)
Sb (ppm)
Cs (ppm)
Ba (ppm)
La (ppm)
Ce (ppm)
Nd (ppm)
Sm (ppm)
Eddystone
y-1 1 T7:,-.-.,J
Coal-rired
Power Plant
290 ± 20
0.041 ± 0.005
640 ± 80
250 ±20
35 ±8
190 ±80
1290 ±60
490 ±190
170 ±60
ND
ND
0.71 ± 0.04
ND
(a)
9.2 ±0.9
ND
120 ±10
180±10
80 ±26
23 ±2
COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS STATIONARY
SOURCES IN THE PHILADELPHIA AREA
Oil-Fired Power Plants
N
9
3
3
3
3
1
9
9
2
2
2
3
2
3
3
Eddystone
980 ±320
1.3 ±0.3
33 ±6
26 ±9
90 ±60
ND
160 ±50
140 ±180
930 ±210
ND
ND
ND
320 ±230
370 ±410
ND
1960 ± 100
130±30
89 ±23
28 ±5
3.7 ±0.7
N
9
3
1
3
9
9
9
3
9
3
3
3
3
2
3
Schuylkill
1100 ±500
0.78 ±0.30
50 ±16
23 ±7
45 ±17
ND
280 ±70
100 ±120
1500 ±300
ND
ND
ND
200 ± 80
1020 ±90
ND
2000 ± 500
450 ±30
360 ± 20
230 ±20
20.5 ±1.5
N
11
3
3
3
11
11
11
3
11
3
3
3
3
3
3
Secondary
Al Plant
450 ±200
0.079 ± 0.006
15±6
66 ±3
630 ± 70
97 ±38
ND
ND
ND
ND
ND
ND
550 ±540
6100 ±300
ND
ND
19±2
ND
ND
ND
N Fluid Cat. Cracker N
2 14 ±8 9
1 0.0026 ± 0.0007 3
1 ND
1 15 ±1 3
2 5.6±1.8 9
1 ND
36 ±6 9
130 ±50 2
ND
ND
ND
ND
2 ND
1 7.7 ±1.5 3
ND
290 ±90 2
1 3300 ±500 3
2700 ± 400 3
1800 ±250 3
170 ±20 3
Municipal
Incinerator
1300 ±500
10.4 ±0.5
64 ±34
42 ± 16
2300 ± 800
230 ±50
87 ± 14
ND
240 ±130
71 ± 15
1200 ± 700
4.9 ±1.4
6700 ± 1900
1300± 1000
5.9 ±3.0
ND
1.1 ±0.5
ND
ND
ND
N
3
3
3
3
10
2
10
10
3
3
3
10
3
3
1
-------
TABLE 4A-2 (cont'd). COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS
o
r+
O
cr
fD
•O
•O
•O
^
i
oo
STATIONARY SOURCES IN THE PHILADELPHIA AREA
Species
(Units)
Eu (ppm)
Gd (ppm)
Tb (ppm)
Yb (ppm)
Lu (ppm)
Hf(ppm)
Ta (ppm)
W (ppm)
Au (ppm)
Pb (%)
Th (ppm)
% mass
Eddystone Coal-
Fired Power
Plant
5.1 ±0.5
ND
3.3 ±0.3
10.3 ±0.5
ND
5. 8 ±0.8
ND
20 ±8
ND
0.041 ± 0.004
24 ±2
24 ±2
Oil-Fired Power Plants
N
3
3
1
3
1
9
3
6
Eddystone N
ND
ND
ND
ND
ND
0.39 ±0.07 1
ND
60 ±5 2
0.054 ±0.017 2
1.8 ±0.6 9
1.9 ±0.5 2
93.5 ±2.5 6
Schuylkill
0.65 ± 0.23
ND
0.90 ± 0.29
ND
ND
ND
ND
ND
ND
1.0 ±0.2
ND
96 ±2
Secondary
N Al Plant
3 ND
ND
3 ND
ND
ND
ND
ND
ND
ND
11 0.081 ±0.014
ND
6 81 ±10
Fluid Cat.
N Cracker
4.9 ±0.7
71 ±10
8.9 ±1.3
3.7 ±0.4
0.59 ±0.17
0.99 ±0.08
0.56 ±0.10
ND
ND
2 0.0091 ± 0.0021
6.2 ± 0.7
2 97 ±2
N
3
3
3
3
3
3
3
9
3
7
Municipal
Incinerator
ND
ND
ND
ND
ND
ND
ND
ND
0.56 ±0.27
5. 8 ±1.2
ND
89 ±2
N
3
10
7
o
O
2
O
H
N = Number of samples.
ND = Not detected.
The "% mass" entries give the average percentage of the total emitted mass found in the fine fraction.
(a) Omitted because of sample contamination.
Source: Adapted from Olmez et al. (1988).
o
H
W
O
W
O
HH
H
W
-------
1 and V and Ni in oil. In fact, the higher V content of the fuel oil than in coal could help account
2 for the higher sulfate seen in the profiles from the oil-fired power plant compared to the
3 coal-fired power plant since V at combustion temperatures is known to catalyze the oxidation of
4 reduced sulfur species. Although Table 4A-2 only gives values of the fine particle composition,
5 measurements of coarse particle composition were also reported by Olmez et al. (1988) which
6 were qualitatively similar.
7 Apart from emissions in the combustion of fossil fuels, toxic trace elements are emitted as
8 the result of various industrial processes such as steel and iron manufacturing and non-ferrous
9 metal production (e.g., for Pb, Cu, Ni, Zn, and Cd). As may be expected, emissions factors for
10 the various trace elements are highly source-specific (Nriagu and Pacyna, 1988). Inspection of
11 Table 4A-3 reveals that the emissions from the catalytic cracker and the oil-fired power plant are
12 greatly enriched in rare-earth elements such as La compared to other sources.
13 Emissions from municipal waste incinerators are heavily enriched in Cl arising mainly from
14 the combustion of plastics and metals that form volatile chlorides. The metals can originate from
15 cans or other metallic objects and some metals such as Zn and Cd are also additives in plastics or
16 rubber. Many elements such as S, Cl, Zn, Br, Ag, Cd, Sn, In, and Sb are enormously enriched
17 compared to their crustal abundances. A comparison of the trace elemental composition of
18 incinerator emissions in Philadelphia, PA (shown in Table 4A-2) with the composition of
19 incinerator emissions in Washington B.C., and Chicago, IL, (Olmez et al., 1988) shows
20 agreement for most constituents to better than a factor of two.
21 The principal components emitted by diesel and gasoline fueled vehicles are organic carbon
22 (OC) and elemental carbon (EC) as shown in Table 4A-3. As can be seen, the variability among
23 entries for an individual fuel type is large and overlaps that found between different fuel types.
24 On average, the abundance of elemental carbon is larger than that of organic carbon in the
25 exhaust of diesel vehicles, while organic carbon is the dominant species in the exhaust of
26 gasoline fueled vehicles. Per vehicle, total carbon emissions from light and heavy duty diesel
27 vehicles can range from one to two orders of magnitude higher than those from gasoline vehicles.
28 There appears to be a tendency for emissions of elemental carbon to increase relative to
29 emissions of organic carbon for gasoline fueled vehicles as simulated driving conditions are
30 changed from a steady 55 km /hr to the various load conditions specified in the Federal Test
31
October 1999 4A-9 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4A-3. FRACTIONAL ORGANIC AND ELEMENTAL CARBON
ABUNDANCES IN MOTOR VEHICLE EMISSIONS
Fuel Type
Diesel
Denver, COa
Los Angeles, CAa
Bakersfield, CAb
Phoenix, AZb
Unleaded gasoline
Denver, COa
Los Angeles, CA°
Los Angeles, CAa
Phoenix, AZb
Leaded gasoline
Denver, COa
Los Angeles, CA°
Los Angeles, CAa
Mixed (tunnel and roadside)
Denver, CO
Los Angeles, CAd
Phoenix, AZ
Organic Carbon
23 ± 8%
36 ± 3%
49 ±13%
40 ± 7%
76 ± 29%
93 ± 52%
49 ± 10%
30 ± 12%
67 ± 23%
52 ± 4%
31 ±20%
50 ± 24%
38 ± 6%
39 ± 19%
Elemental Carbon
74 ±21%
52 ± 5%
43 ± 8%
33 ± 8%
18 ± 11%
5 ± 7%
39 ± %
14 ± 8%
16 ±7%
13 ± 1%
15 ±2%
28 ± 19%
38 ± 5%
36 ± 11%
Ne
3
2
3
8
8
11
11
9
3
3
3
3
Sources
1,2
3,4,5,6
7
8
1,2
3,4,5,6
3,4,5,6
8
1,2
3,4,5,6
3,4,5,6
1,2
3
8
Sources: (1) Watson et al. (1990b), (2) Watson et al. (1990a), (3) Cooper et al. (1987), (4) NBA, Inc. (1990),
(5) Peltier et al. (1990a), (6) Peltier et al. (1990b), and (7) Houck et al. (1989), cited in (8) Watson et al.
(1994b).
Notes: (a) Modified Federal Test Procedures followed in dynamometer tests; (b) Roof monitoring at
inspection station; (c) 55 km/hr steady speed in dynamometer tests; (d) Rt. 1 tunnel at LA airport,
(e) N = Number of samples.
1 Procedures (FTPs). Also shown are the results of sampling from mixed vehicle types along
2 roadsides and in tunnels.
October 1999
4 A-10
DRAFT-DO NOT QUOTE OR CITE
-------
1 The results shown in Table 4A-3 were obtained during the late 1980's, and, so, may not be
2 entirely representative of current vehicles. Examples of data for the trace element composition of
3 motor vehicle emissions obtained in Phoenix, AZ are shown in Table 4A-4. SO2 emissions are
4 also shown in relation to the mass of fine particles emitted. As can be seen, small quantities of
5 soluble ions such as SO4= and NH4+ are emitted. The ammonium may be emitted as the result of
6 an improperly functioning catalytic converter, or may simply be the result of contamination
7 during sample handling and analysis. Four fractions are given for the organic carbon fraction and
8 three for elemental carbon. These refer to abundances measured at different temperatures in a
9 thermographic analysis. Temperatures for OC1, OC2, OC3, and OC4 are 120 °C, 250 °C,
10 450 °C, and 550 °C, respectively; and, for EC1, EC2, EC3, they are 550 °C, 700 °C, and
11 800 °C, respectively, in He/2% O2. The abundances of trace elements are all quite low, with
12 most being less than 1%. It is not clear what the source of the small amount of Pb seen in the
13 auto exhaust profile is. It is extremely difficult to find suitable tracers for automotive exhaust
14 since Pb has been removed from gasoline. However, it should also be remembered that
15 restrictions in the use of leaded gasoline have resulted in a dramatic lowering of ambient Pb
16 levels. Examples of data for the trace elemental composition of the emissions from a number of
17 vehicle classes obtained as part of the North Frontal Range Air Quality Study (NFRAQS) which
18 took place in December of 1997 in Colorado are shown in Table 4A-5. As can be seen from
19 Table 4A-5, emissions of total carbon (TC), which is equal to the sum of organic carbon (OC)
20 and elemental carbon (EC), from gasoline vehicles are highly variable. Emissions from
21 "smokers", i.e., light duty vehicles with visible smoke emitted from their tailpipes, are
22 comparable to those from diesel vehicles. Thus, older poorly maintained gasoline vehicles could
23 be significant sources of PM25 (Sagebiel et al., 1997; Lawson and Smith, 1998) in addition to
24 being significant sources of gaseous pollutants (e.g., Calvert et al., 1993). Durbin et al. (1999)
25 point out that although "smokers" constitute only 1.1 to 1.7% of the light duty fleet in the South
26 Coast Air Quality Management District in California, they contribute roughly 20% of the total
27 PM emissions from the light duty fleet. In general, motor vehicles which are high emitters of
28 hydrocarbons and carbon monoxide will also tend to be high emitters of PM (Sagebiel et al.,
29 1997; Cadle et al., 1997). Particle emission rates are also correlated with vehicle acceleration
30 and emissions occur predominantly during periods of heavy acceleration even in newer vehicles
31 (Maricqetal, 1999).
October 1999 4A-11 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4A-4. PHOENIX PM2S MOTOR VEHICLE EMISSIONS PROFILES (% MASS)
Chemical Species
NO3-
S042'
NH4+
OC
OC1
OC2
OC3
OC4
EC
EC1
EC2
ECS
Al
Si
P
S
Cl
K
Ca
Ti
Cr
Mn
Fe
Cu
Zn
Sb
Ba
La
Pb
SO2a
Auto
3. 9 ±2.9
2.3 ±1.3
1.7±1.0
30.1 ±12.3
11.3±3.5
9.2 ±6.8
4.6 ±2.2
3. 5 ±1.5
13.5 ±8.0
11.7 ±7.2
3.1 ±1.6
0.15 ±0.30
0.41 ±0.20
1.64 ±0.88
0.11 ±0.07
1.01 ±0.48
0.34 ±0.32
0.25 ±0.14
0.71 ±0.41
0.07 ±0.13
0.02 ±0.01
0.10 ±0.04
0.68 ±0.42
0.07 ±0.06
0.27 ±0.22
0.02 ±0.13
0.06 ±0.40
0.15±0.51
0.16 ±0.07
32.8 ±13. 9
Diesel
0.31 ±0.40
2.4 ±1.0
0.87 ±0.13
40.1 ±6.6
21.0 ±6.3
9.1 ±1.9
5.9 ±1.3
4.0 ±1.5
32.9 ±8.0
4.4 ±1.3
27.9 ±5.6
0.69 ±0.82
0.17±0.12
0.46 ±0.18
0.06 ±0.06
1.24 ±0.28
0.03 ±0.06
0.04 ±0.03
0.16 ±0.06
0.00 ±0.15
0.00 ±0.01
0.01 ±0.01
0.16 ±0.07
0.01 ±0.01
0.07 ± 0.02
0.01 ±0.14
0.14 ±0.47
0.18 ±0.59
0.01 ±0.03
66.9 ±24.0
Source: Watson et al. (1994b).
Note: Elemental abundances <0.01% (V, Co, Ni, Ga, As, Se, Br, Rb, Sr. Y, Zr, Mo, Pd, Ag, Cd, In, Sn, Au, Hg, Tl,
U) in XRF analyses excluded; OC = organic carbon; EC = elemental carbon.
aRelative to total PM25.
October 1999 4A-12 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4A-5. EMISSION RATES (MG/MILE) FOR CONSTITUENTS OF
PARTICIPATE MATTER FROM GASOLINE AND DIESEL VEHICLES
TC
OC
EC
NO3-
S04=
Na
Mg
Al
Si
P
S
Cl
K
Ca
Fe
Ni
Cu
Zn
Br
Ba
Pb
Low
9.07 ±0.75
6.35 ±0.54
2.72 ±0.52
0.039 ±0.027
0.158 ±0.036
0.060 ±0.063
0.036 ±0.022
0.083 ±0.016
0.066 ±0.008
0.035 ±0.004
0.085 ±0.006
0.024 ±0.012
0.010 ±0.009
0.060 ±0.010
0.143 ±0.004
0.001 ±0.004
0.002 ± 0.004
0.048 ±0.003
0.001 ±0.002
0.013 ±0.136
0.007 ± 0.006
Gasoline
Medium
41.30±1.68
26.02 ±1.31
15.28 ±0.99
0.057 ±0.028
0.518 ±0.043
0.023 ±0.111
0.068 ± 0.027
0.078 ±0.016
0.279 ±0.011
0.152 ±0.007
0.442 ± 0.009
0.038 ±0.012
0.019 ±0.009
0.212 ±0.011
0.756 ±0.005
0.005 ± 0.004
0.016 ±0.003
0.251 ±0.004
0.016 ±0.002
0.009 ±0.138
0.085 ±0.005
Vehicles
High
207.44 ± 7.29
95.25 ±4.28
112.19±5.82
0.141 ±0.031
0.651 ±0.052
0.052 ± 0.092
0.041 ±0.033
0.057 ±0.014
0.714 ±0.012
0.1 13 ±0.007
0.822 ±0.022
0.081 ±0.020
0.031 ±0.035
0.210 ±0.030
1.047 ±0.010
0.011 ±0.005
0.021 ±0.005
0.265 ± 0.023
0.079 ± 0.003
0.011 ±0.299
0.255 ±0.008
Smoker
456.38 ±16.80
350.24 ± 15.27
106.14 ±5.42
0.964 ±0.051
2.160 ±0.137
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
2.515±0.116
0.140±0.117
0.033 ±0.386
0.362 ±0.250
2.438 ±0.054
0.008 ±0.017
0.071 ±0.018
0.188 ±0.272
0.047 ±0.012
0.380 ±2.175
0.345 ±0.032
Diesel
LDDa
373.43 ±13.75
132.01 ±5. 82
241.42 ±12.11
1.474 ±0.071
2.902 ±0.165
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
2.458 ±0.124
0.228 ±0.1 14
0.000 ± 0.426
0.150 ±0.304
0.515 ±0.057
0.014 ±0.018
0.024 ±0.021
0.000 ± 0.299
0.003 ±0.014
0.428 ±2.390
0.153 ±0.033
Vehicles
HDD"
1570.69 ±58.24
253.94 ±16.12
1316.75 ±55.33
1.833 ±1.285
3. 830 ±1.286
1.288 ±2.160
1.061 ±0.729
0.321 ±0.543
8.018 ±0.221
0.407 ±0.136
3.717±0.111
0.881 ±0.221
0.064 ±0.248
0.716 ±0.107
0.376 ±0.055
0.002 ±0.057
0.001 ±0.062
0.707 ±0.032
0.012 ±0.050
0.493 ±3. 108
0.008 ±0.154
"Light duty.
bHeavy duty.
Source: Lawson and Smith (1998).
October 1999
4 A-13
DRAFT-DO NOT QUOTE OR CITE
-------
1 In addition to fossil fuels, biomass in the form of wood may be burned in forest fires or as
2 fuel for heating or cooking. At first glance these two broad categories might seem to serve to
3 distinguish between natural and anthropogenic sources. However, many forest fires result from
4 human intervention, either deliberately through prescribed burning in forest management or
5 accidentally through the improper disposal of flammable material or fugitive sparks
6 (e.g., Andreae, 1991). On the other hand, human intervention also suppresses lightning triggered
7 fires and can also lead to the buildup of combustible fuel on the forest floor. Not enough data are
8 available to assess the effects of humans on forest fires, except for land clearing for agriculture.
9 In contrast to the mobile and stationary sources discussed earlier, emissions from biomass
10 burning in woodstoves and forest fires are strongly seasonal and can be highly episodic within
11 their peak emissions seasons. The burning of fuelwood is confined mainly to the winter months
12 and is acknowledged to be a major source of ambient air particulate matter in the northwestern
13 United States during the heating season. Forest fires occur primarily during the driest seasons of
14 the year in different areas of the country and are especially prevalent during prolonged droughts.
15 PM produced by biomass burning outside the United States, e.g., in central America during the
16 spring of 1988 can also strongly affect ambient air quality in the United States.
17 An example of the composition of fine particles (PM25) produced by woodstoves is shown
18 in Figure 4A-4. These data were obtained in Denver during the winter of 1987-1988 (Watson
19 and Chow, 1994). As was the case for motor vehicle emissions, organic and elemental carbon
20 are the major components of particulate emissions from wood burning. It should be remembered
21 that the relative amounts shown for organic carbon and elemental carbon vary with the type of
22 stove, the stage of combustion and the type and condition of the fuelwood. Fine particles are
23 dominant in studies of wood burning emissions. For instance, the mass median diameter of
24 wood-smoke particles was found to be about 0.17 //m in a study of the emissions from burning
25 hardwood, softwood and synthetic logs (Dasch, 1982).
26 Measurements of aerosol composition, size distributions, and aerosol emissions factors
27 have been made in biomass burning plumes either on towers (Susott et al., 1991) or aloft on fixed
28 wing aircraft (e.g., Radke et al., 1991) or on helicopters (e.g., Cofer et al., 1988). As was found
29 for woodstove emissions, the composition of biomass burning emissions is strongly dependent
30 on the stage of combustion (i.e., flaming, smoldering, or mixed), and the type of vegetation (e.g.,
31 forest, grassland, scrub). Over 90% of the dry mass in particulate biomass burning emissions is
October 1999 4A-14 DRAFT-DO NOT QUOTE OR CITE
-------
10
10
10
,-4
Chemical Species
Figure 4A-4. Chemical abundances for PM2 5 emissions from wood burning in Denver,
CO. Solid bars represent fractional abundances, and the error bars
represent variability in species abundances. Error bars represent detection
limits when there are no solid bars.
Source: Watson and Chow (1994).
1 composed of organic carbon (Mazurek et al., 1991). Ratios of organic carbon to elemental
2 carbon are highly variable ranging from 10:1 to 95:1, with the highest ratio found for smoldering
3 conditions and the lowest for flaming conditions. Emissions factors for total particulate
4 emissions increase by factors of two to four in going from flaming to smoldering stages in the
5 individual fires studied by Susott et al. (1991).
6 Particles in biomass burning plumes from a number of different fires were found to have
7 three distinguishable size modes, namely a nucleation mode, an accumulation mode, and a coarse
8 mode (Radke et al., 1991). Based on an average of 81 samples, approximately 70% of the mass
9 was found in particles < 3.5 //m in aerodynamic diameter. The fine particle composition was
10 found to be dominated by tarlike, condensed hydrocarbons and the particles were usually
October 1999
4 A-15
DRAFT-DO NOT QUOTE OR CITE
-------
1 spherical in shape. Additional information for the size distribution of particles produced by
2 vegetation burning was shown in Figure 4A-2.
3 An example of ambient data for the composition of PM25 collected at a tropical site that
4 was heavily affected by biomass burning is shown in Table 4A-6. The samples were collected
5 during November of 1997 on the campus of Sriwijaya University, which is located in a rural
6 setting on the island of Sumatra in Indonesia (Pinto et al., 1998). The site was subjected
7 routinely to levels of PM25 well in excess of the U.S. NAAQS as a result of the Indonesian
8 biomass fires from the summer of 1997 through the spring of 1988. As can be seen from a
9 comparison of the data shown in Table 4A-6 with those shown in Figure 4A-4, there are a
10 number of similarities and differences (especially with regard to the heavy metal content) in the
11 abundances of many species. The abundances of some crustal elements (e.g., Si, Fe) are higher
12 in Table 4A-6 than in Figure 4A-4 perhaps reflecting additional contributions of entrained soil
13 dust.
14 Although sea-salt aerosol production is confined to salt water bodies, it is included here
15 because many marine aerosols can exert a strong influence on the composition of the ambient
16 aerosol in coastal areas. In some respects, the production of sea-salt aerosols is like that of
17 windblown dust in that both are produced by wind agitation of the surface. The difference
18 between the two categories arises because sea-salt particles are produced from the bursting of air
19 bubbles rising to the sea surface. Air bubbles are formed by the entrainment of air into the water
20 by breaking waves. The surface energy of a collapsing bubble is converted to kinetic energy in
21 the form of a jet of water which can eject drops above the sea surface. The mean diameter of the
22 jet drops is about 15% of the bubble diameter (Wu, 1979). Bubbles in breaking waves range in
23 size from a few //m to several mm in diameter. Field measurements by Johnson and Cooke
24 (1979) of bubble size spectra show maxima in diameters at around 100 //m, with the bubble size
25 distribution varying as (d/d0)"5 with d0 = 100 //m.
26 Since sea-salt particles receive water from the surface layer, which is enriched in organic
27 compounds, the aerosol drops are composed of this organic material in addition to sea salt (about
28 3.5% by weight in sea water). Na+ (30.7%),Cr (55.0%), SO4= (7.7%), Mg2+ (3.6%), Ca2+ (1.2%),
29 K+ (1.1%), HCO3" (0.4%), and Br" (0.2%) are the major ionic species by mass in sea water
30 (Wilson, 1975). The composition of the marine aerosol also reflects the occurrence of
October 1999 4A-16 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4A-6. MEAN AEROSOL COMPOSITION AT TROPICAL SITE
(SRIWIJAYA UNIVERSITY, SUMATRA, INDONESIA) AFFECTED
HEAVILY BY BIOMASS BURNING EMISSIONS1
Component
OC
EC
scv
Al
Si
Cl
K
Ca
Ti
V
Abundance (%)
76
1.2
11
bd2
9.3 x 1Q-2
4.4
0.7
4.5 x 1Q-2
4.2 x 1Q-3
bd2
Component
Cr
Mn
Fe
Ni
Cu
Zn
As
Se
Br
Pb
Abundance (%)
bd2
bd2
3.9 x 1Q-2
<3.8 x 1Q-5
4.8 x 1Q-4
3.1 x 1Q-3
6.4 x 1Q-4
2.8 x 1Q-4
3.6 x 1Q-2
3.1 x 1Q-3
'The mean PM2 5 concentration during the sampling period (11/5-11/11/97) was 264 //g/m3.
2beneath detection limit.
Source: Pinto etal. (1998).
1 displacement reactions which enrich sea-salt particles in SO4" and NO3" while depleting them of
2 Cr and Br.
3 Seasalt is concentrated in the coarse size mode with a mass median diameter of about 7 //m
4 for samples collected in Florida, the Canary Islands and Barbados (Savoie and Prospero, 1982).
5 The size distribution of sulfate is distinctly bimodal. Sulfate in the coarse mode is derived from
6 sea water but sulfate in the submicron aerosol arises from the oxidation of dimethyl sulfide
7 (CH3SCH3) or DMS. DMS is produced during the decomposition of marine micro-organisms.
8 DMS is oxidized to MSA (methane sulfonic acid) a large fraction of which is oxidized to sulfate
9 (e.g., Hertel etal., 1994).
10 Apart from sea spray, other natural sources of particles include the suspension of organic
11 debris and volcanism. Particles are released from plants in the form of seeds, pollen, spores, leaf
October 1999
4 A-17
DRAFT-DO NOT QUOTE OR CITE
-------
1 waxes and resins, ranging in size from 1 to 250 //m (Warneck, 1988). Fungal spores and animal
2 debris such as insect fragments are also to be found in ambient aerosol samples in this size range.
3 While material from all the foregoing categories may exist as individual particles, bacteria are
4 usually found attached to other dust particles (Warneck, 1988). Smaller bioaerosol particles
5 include viruses, individual bacteria, protozoa, and algae (Matthias-Maser and Jaenicke, 1994).
6 In addition to natural sources, other sources of bioaerosol include industry (e.g., textile mills),
7 agriculture, and municipal waste disposal (Spendlove, 1974). The size distribution of
8 bioaerosols has not been as well characterized as it has for other categories.
9 Trace metals are emitted to the atmosphere from a variety of sources such as sea spray,
10 wind blown dust, volcanoes, wild fires and biotic sources (Nriagu, 1989). Biologically mediated
11 volatilization processes (e.g., biomethylation) are estimated to account for 30-50% of the
12 worldwide total Hg, As, and Se emitted annually, whereas other metals are derived principally
13 from pollens, spores, waxes, plant fragments, fungi, and algae. It is not clear, however, how
14 much of the biomethylated species are remobilized from anthropogenic inputs. Median ratios of
15 the natural contribution to globally averaged total sources for trace metals are estimated to be
16 0.39 (As), 0.15 (Cd), 0.59 (Cr), 0.44 (Cu), 0.41 (Hg), 0.35 (Ni), 0.04 (Pb), 0.41 (Sb), 0.58 (Se),
17 0.25 (V), and 0.34 (Zn), suggesting a not insignificant natural source for many trace elements.
18 It should be noted though that these estimates are based on emissions estimates which have
19 uncertainty ranges of an order of magnitude.
20 The discussion above focused mainly on the elemental composition of emissions. Carbon
21 was treated mainly as organic carbon and elemental carbon. However, there are literally
22 hundreds of organic compounds which have been quantified in ambient and source samples
23 which are lumped routinely into the category of organic carbon. These compounds, many of
24 which have been used in source apportionment studies as signature (e.g., Schauer et al., 1996),
25 consist of high molecular weight alkanes, hopanes and steranes, organic acids, aldehydes,
26 polycyclic aromatic hydrocarbons and steroids. Profiles of organic compounds from meat
27 cooking (Rogge et al., 1991); automobiles and heavy-duty diesel trucks (Rogge et al., 1993a);
28 road dust, tire debris and brake linings (Rogge et al., 1993b); vegetative detritus (Rogge et al.,
29 1993c); natural gas home appliances (Rogge et al., 1993d); cigarette smoke (Rogge et al., 1994);
30 hot asphalt roofing tar pots (Rogge et al., 1997a); distillate fuel oil burning (Rogge et al., 1997b);
31 and pine, oak, and synthetic log burning in residential fire places (Rogge et al., 1998) have been
October 1999 4A-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 obtained. Many individual compounds are present in concentrations much less than 1 ng/m3 in
2 ambient PM samples. They have been used in only a limited number of studies (mainly in
3 Los Angeles) by only a small number of groups. Measurement methods need to be standardized
4 and made more cost-effective to take advantage of the opportunities they offer in studies
5 throughout the United States.
October 1999 4A-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Andreae, M. O. (1991) Biomass burning: its history, use, and distribution and its impact on environmental quality
4 and global climate. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric
5 implications. Cambridge, MA: MIT Press; pp. 1-21.
6 Cadle, S. H.; Mulawa, P. H.; Ball, J.; Donase, C.; Weibel, A.; Sagebiel, J. C.; Knapp, K. T.; Snow, R. (1997)
7 Particulate emission rates from in-use high-emitting vehicles recruited in Orange county, California. Environ.
8 Sci. Technol. 31:3405-3412.
9 Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality: some
10 reflections on controlling vehicle emissions. Science (Washington, DC) 261: 37-45.
11 Chow, J. C.; Watson, J. G.; Houck, J. E.; Pritchett, L. C.; Rogers, C. F.; Frazier, C. A.; Egami, R. T.; Ball, B. M.
12 (1994) A laboratory resuspension chamber to measure fugitive dust size distributions and chemical
13 compositions. Atmos. Environ. 28: 3463-3481.
14 Cofer, W. R., Ill; Levine, J. S.; Sebacher, D. L; Winstead, E. L.; Riggin, P. J.; Brass, J. A.; Ambrosia, V. G. (1988)
15 Particulate emissions from a mid-latitude prescribed chaparral fire. J. Geophys. Res. [Atmos.] 93: 5207-5212.
16 Cooper, J. A.; Redline, D. C.; Sherman, J. R.; Valdovinos, L. M.; Pollard, W. L.; Scavone, L. C.; Badgett-West, C.
17 (1987) PM10 source composition library for the South Coast Air Basin: volume I, source profile development
18 documentation final report. El Monte, CA: South Coast Air Quality Management District; July 15.
19 Dasch, J. M. (1982) Particulate and gaseous emissions from wood-burning fireplaces. Environ. Sci. Technol.
20 16: 639-645.
21 Durbin, T. D.; Smith, M. R.; Norbeck, J. M.; Truex, T. J. (1999) Population density, particulate emission
22 characterization, and impact on the particulate inventory of smoking vehicles in the South Coast Air Quality
23 Management District. J. Air Waste Manage. Asssoc. 49: 28-38.
24 Hertel, O.; Christensen, J.; Hov, 0. (1994) Modelling of the end products of the chemical decomposition of DMS in
25 the marine boundary layer. Atmos. Environ. 28: 2431-2449.
26 Hopke, P. K.; Lamb, R. E.; Natusch, D. F. S. (1980) Multielemental characterization of urban roadway dust.
27 Environ. Sci. Technol. 14: 164-172.
28 Houck, J. E.; Chow, J. C.; Watson, J. G.; Simons, C. A.; Pritchett, L. C.; Goulet, J. M.; Frazier, C. A. (1989)
29 Determination of particle size distribution and chemical composition of particulate matter from selected
30 sources in California: volume I and executive summary (final report). Sacramento, CA: California Air
31 Resources Board.
32 Houck, J. E.; Goulet, J. M.; Chow, J. C.; Watson, J. G.; Pritchett, L. C. (1990) Chemical characterization of
33 emission sources contributing to light extinction. In: Mathai, C. V., ed. Visibility and fine particles:
34 an A&WMA/EPA international specialty conference; October 1989; Estes Park, CO. Pittsburgh, PA: Air &
35 Waste Management Association; pp. 437-446. (A&WMA transactions series no. TR-17).
36 Johnson, B. D.; Cooke, R. C. (1979) Bubble populations and spectra in coastal waters: a photographic approach.
37 J. Geophys. Res. C: Oceans Atmos. 84: 3761-3766.
38 Lawson, D. R.; Smith, R. E. (1998) The northern front range air quality study: a report to the Governor and General
39 Assembly. Fort Collins, CO: Colorado State University; December.
40 Maricq, M. M.; Podsiadlik, D. H.; Chase, R. E. (1999) Gasoline vehicle particle size distributions: comparison of
41 steady state, FTP, and USOG measurements. Environ. Sci. Technol. 33: 2007-2015.
42 Mason, B. (1966) Principles of geochemistry. 3rd ed. New York, NY: John Wiley & Sons, Inc.
43 Matthias-Maser, S.; Jaenicke, R. (1994) Examination of atmospheric bioaerosol particles with radii > 0.2 pan.
44 J. Aerosol Sci. 25: 1605-1613.
45 Mazurek, M. A.; Cofer, W. R., Ill; Levine, J. S. (1991) Carbonaceous aerosols from prescribed burning of a boreal
46 forest ecosystem. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric
47 implications. Cambridge, MA: MIT Press; pp. 258-263.
48 NEA, Inc. (1990) Enhancement of the south coast air basin source profile library for chemical mass balance
49 receptor model applications: volume I, final report. Beaverton, OR: NEA.
50 Nriagu, J. O. (1989) A global assessment of natural sources of atmospheric trace metals. Nature (London)
51 338:47-49.
52 Nriagu, J. O.; Pacyna, J. M. (1988) Quantitative assessment of worldwide contamination of air, water and soils by
53 trace metals. Nature (London) 333: 134-139.
October 1999 4A-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 Olmez, I.; Sheffield, A. E.; Gordon, G. E.; Houck, J. E.; Pritchett, L. C.; Cooper, J. A.; Dzubay, T. G.; Bennett,
2 R. L. (1988) Compositions of particles from selected sources in Philadelphia for receptor modeling
3 applications. JAPCA 38: 1392-1402.
4 Peltier, R. G.; Redline, D. C.; Cooper, J. A.; Aquilizan, A. V.; Miller, E. A.; White, J. L. (1990a) Enhancement of
5 the south coast air basin source profile library for chemical mass balance receptor model applications.
6 Volume II: source profile library supplement. Final report. Beaverton, OR: NEA, Inc.
7 Peltier, R. G.; Redline, D. C.; Cooper, J. A.; Aquilizan, A. V.; Miller, E. A.; White, J. L. (1990b) Enhancement of
8 the south coast air basin source profile library for chemical mass balance receptor model applications.
9 Volume III: appendices. Final report. Beaverton, OR: NEA, Inc.
10 Pinto, J. P.; Grant, L. D.; Hartlage, T. A. (1998) Report on U.S. EPA air monitoring of haze from S.E. Asia biomass
11 fires. Research Triangle Park, NC: U.S. Environmental Protection Agency, National Center for
12 Environmental Assessment; report no. EPA/600/R-98/071.
13 Radke, L. F.; Hegg, D. A.; Hobbs, P. V.; Nance, J. D.; Lyons, J. H.; Laursen, K. K.; Weiss, R. E.; Riggan, P. J.;
14 Ward, D. E. (1991) Particulate and trace gas emissions from large biomass fires in North America. In:
15 Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric implications. Cambridge,
16 MA: MIT Press; pp. 209-224.
17 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine organic
18 aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 25: 1112-1125.
19 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
20 aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
21 Technol. 27:636-651.
22 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
23 aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks. Environ.
24 Sci. Technol. 27: 1892-1904.
25 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine organic
26 aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
27 27:2700-2711.
28 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993d) Sources of fine organic
29 aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27: 2736-2744.
30 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Sources of fine organic
31 aerosol. 6. Cigarette smoke in the urban atmosphere. Environ. Sci. Technol. 28: 1375-1388.
32 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997a) Sources of fine organic
33 aerosol. 7. Hot asphalt roofing tar pot fumes. Environ. Sci. Technol. 31: 2726-2730.
34 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997b) Sources of fine organic
35 aerosol. 8. Boilers burning no. 2 distillate fuel oil. Environ. Sci. Technol. 31: 2731-2737.
36 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1998) Sources of fine organic
37 aerosol. 9. Pine, oak, and synthetic log combustion in residential fireplaces. Environ. Sci. Technol. 32: 13-22.
38 Sagebiel, J. C.; Zielinska, B.; Walsh, P. A.; Chow, J. C.; Cadle, S. H.; Mulawa, P. A.; Knapp, K. T.; Zweidinger,
39 R. B.; Snow, R. (1997) PM-10 exhaust samples collected during IM-240 dynamometer tests of in-service
40 vehicles in Nevada. Environ. Sci. Technol. 31: 75-83.
41 Savoie, D. L.; Prospero, J. M. (1982) Particle size distribution of nitrate and sulfate in the marine atmosphere.
42 Geophys. Res. Lett. 9: 1207-1210.
43 Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R. (1996) Source apportionment of
44 airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
45 Spendlove, J. C. (1974) Industrial, agricultural, and municipal microbial aerosol problems. In: Current problems in
46 aerobiology: proceedings of the thirtieth general meeting of the Society for Industrial Microbiology; August
47 1973; Evanston, IL. Dev. Ind. Microbiol. 15: 20-27.
48 Susott, R. A.; Ward, D. E.; Babbitt, R. E.; Latham, D. J. (1991) The measurement of trace emissions and
49 combustion characteristics for a mass fire. In: Levine, J. S., ed. Global biomass burning: atmospheric,
50 climatic, and biospheric implications. Cambridge, MA: MIT Press; pp, 245-257.
51 Turekian, K. K. (1971) Elements, geochemical distribution of. In: Encyclopedia of science and technology, v. 4.
52 New York, NY: McGraw-Hill Book Company; pp. 627-630.
53 Vinogradov, A. P. (1959) The geochemistry of rare and dispersed chemical elements in soils. 2nd ed. New York,
54 NY: Consultants Bureau, Inc.
55 Warneck, P. (1988) Chemistry of the natural atmosphere. New York, NY: Academic Press, Inc.
October 1999 4A-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 Watson, J. G.; Chow, J. C. (1994) Clear sky visibility as a challenge for society. Annu. Rev. Energy Environ.
2 19:241-266.
3 Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Burns, S.; Ragazzi, R. A. (1990a) Composite source
4 profiles for particulate motor vehicle exhaust source apportionment in Denver, CO. In: Mathai, C. V., ed.
5 Visibility and fine particles: an A&WMA/EPA international specialty conference; October 1989; Estes Park,
6 CO. Pittsburgh, PA: Air & Waste Management Association; pp. 422-436. (A&WMA transactions series
7 no. TR-17).
8 Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Ragazzi, R. A.; Burns, S. (1990b) Chemical source
9 profiles for particulate motor vehicle exhaust under cold and high altitude operating conditions. Sci. Total
10 Environ. 93: 183-190.
11 Watson, J. G.; Chow, J. C.; Blumenthal, D. L.; Lurmann, F. W.; Hackney, R. J.; Magliano, K. L.; Pederson, J. R.;
12 Neff, W. D.; Roth, P. M.; Solomon, P. A.; Thuillier, R. H.; Ziman, S. D. (1994b) Planning for data analysis.
13 In: Solomon, P. A.; Silver, T. A., eds. Planning and managing regional air quality modeling and measurement
14 studies: a perspective through the San Joaquin Valley air quality study and AUSPEX. Boca Raton, FL:
15 Lewis Publishers; pp. 335-349.
16 Wilson, T. R. S. (1975) Salinity and the major elements of sea water. In: Riley, J. P.; Skirrow, G., eds. Chemical
17 oceanography: v. 1. 2nd ed. London, United Kingdom: Academic Press Inc.; pp. 365-413.
18 Wu, J. (1979) Spray in the atmospheric surface layer: review and analysis of laboratory and oceanic results.
19 J. Geophys. Res. C: Oceans Atmos. 84: 1693-1704.
October 1999 4A-22 DRAFT-DO NOT QUOTE OR CITE
-------
i 5. HUMAN EXPOSURE TO AMBIENT PARTICULATE
2 MATTER: RELATIONS TO CONCENTRATIONS
3 OF AMBIENT AND NON-AMBIENT PM AND
4 OTHER AIR POLLUTANTS
5
6
7 5.1 INTRODUCTION AND BASIC CONCEPTS
8 This chapter examines ambient particulate matter (Plvf) air quality and that portion of
9 ambient PM which penetrates into indoor microenvironments. It also examines, to a lesser
10 extent, the contributions of sources of non-ambient PM to total PM exposure. This is to aid in
11 the interpretation of the acute and chronic epidemiology studies discussed in Chapter 6, in which
12 ambient PM concentrations are assumed to be an indicator, or a surrogate, for the mean
13 community exposure to PM of ambient origin, or an individual's exposure to ambient PM.
14 Thus, this chapter has three objectives:
15 (a) To provide a review of pertinent studies of personal exposures to total PM (ambient PM plus
16 non-ambient PM).
17 (b) To evaluate linkages of human exposure to PM of ambient origin estimated from
18 concentrations of PM measured at a fixed-site monitor located at some central location in a
19 community under study.
20 (c) To quantify the contribution of PM of ambient origin to total personal PM exposure.
21 The 1996 Particulate Matter Air Quality Criteria Document (PM AQCD) (U.S.
22 Environmental Protection Agency, 1996) thoroughly reviewed the PM exposure literature
23 through 1995 and early 1996. This chapter reviews the history of PM AQCD developments from
24 1969 to the present. It then thoroughly reviews key pre-1996 studies, the new PM exposure
25 literature from 1995/6 through 1998 to date, and some literature accepted or submitted for
26 publication later in 1999.
27 The U.S. Environmental Protection Agency regulatory authority for PM extends to the
28 ambient air, defined in 40 CFR 50.1(e) as that portion of the atmosphere, external to buildings, to
'in this chapter PM without a subscript refers to PM in general. PMX refers to the mass of PM collected by
a monitor with a penetration fraction of 0.5 for particles with an aerodynamic diameter of X microns (XponAD).
October 1999 5-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 which the general public has access (Code of Federal Regulations, 1994). Therefore, polluted
2 air inside a building, or anywhere on private property owned or controlled by the source of
3 pollution, is not regulated by the National Ambient Air Quality Standards (NAAQS). However,
4 it is necessary to consider total personal exposure to ambient PM, both in regulated ambient air
5 and within non-regulated indoor air. This is because ambient (outdoor) particles penetrate into
6 non-ambient (indoor residential and occupational) micro environments where, on average, people
7 spend 87% of their time (Klepeis et al., 1999). Therefore, when people are indoors, they are
8 exposed to a mixture of PM of ambient origin and particles generated indoors from sources not
9 regulated by EPA (e.g., cigarette smoke or an occupational activity).
10 Particulate matter represents a generic class of pollutants which requires a different
11 interpretation of exposure in contrast to that for the gaseous criteria pollutants, such as CO
12 (Mage, 1985). Whereas a molecule of CO emitted from a motor vehicle is indistinguishable
13 from a molecule of CO emitted from a cigarette, a 1-^m aerodynamic diameter (AD) particle
14 emitted from a motor vehicle and a l-^m AD particle emitted from a cigarette may have a
15 different shape, mass, chemical composition, solubility and toxicity (Siegmann et al., 1999).
16 In the atmosphere, a particle may be a single entity, or an agglomeration of particles, such as a
17 particle from motor vehicle exhaust bound to a particle from cigarette ash. Furthermore, indoor
18 sources of particles also produce a wide variety of particles of varying AD and composition that
19 people are exposed to, as shown in Figure 5-1 (Owen et al., 1992). Most of these particles can be
20 fragmented by mechanical activity and their detritus may exist at smaller ADs than shown.
21 A subject's personal exposure to PM is theoretically measured by sampling the
22 concentration of PM in the inhaled air entering the nose or mouth. The inlet to a personal
23 monitor is normally placed at the outer limit of the breathing zone to avoid a negative sampling
24 bias that could be caused by dilution of the collected air by the exhaled breath which is depleted
25 of PM. However, such placement does not allow sampling of directly inhaled cigarette smoke or
26 the inhaled air that passes through a dust mask. Thus, personal monitoring of PM exposure of
2In this chapter, the term "ambient air" means that portion of the ambient atmosphere that may be
considered representative of a community and not unduly influenced by any specific identifiable source (e.g., not at
the side of a highway, or next to a coke oven). "Outdoor air" is taken to mean that portion of the atmosphere,
external to buildings, where a PM concentration measurement may be unduly influenced by an immediate source of
PM (undiluted cigarette smoke, backyard barbecue, idling motor vehicle, etc.). Thus, all ambient air is outdoor air,
but not all outdoor air is ambient air.
October 1999 5-2 DRAFT-DO NOT QUOTE OR CITE
-------
o
o
r-K
o
cr
o
l-l
^o
^o
'S*O
^-/l
o
Tl
H
O
0
0
H
0
0
H
W
O
O
HH
H
W
0
Plant
Animal
Mineral
Combustion
Home/
Personal
Care
Radioactive
Figure 5-1. Sh
Source: Owen et £
Particle Aerodynamic Diameter (|jm)
.01 0.1 1 10 100 1,0
— Corn starch
-_ Carbon
"^
Bacteriophage ^
-., Viruses ^.
•••| • |^ • Vnllen L "
f ^ Starches ^
K ^ Milled Flour, Ground Corn ^ "" Grain Duste
Black - ^ *•
^ Channel Black ^. Textile Fibers
PuddinofMix ^Com Cob Chaff ^
Snuff 4 Cayenne Pepper ^
•^ anu" ™ »-^ Sawdust ^ p
Droplet Nuclei1 ^ Bacteria
^ Gelatin ^
Spider Web ., Hair ^
^ » p,- • . . i n 9 ogray "nea mill-. rjust M|tes
•< >• Disintegrated Feces •<" Faces ^ Bone Dust * *»,
Asbestos _
^" Talo. ^ •—
^ Carbon Dust from Graphite >
^ Clay
^ Lead. Bromine ^
Metalurgical Dusts and Fumes
^ NI-UCI Fume
4 Sea Salt * ^
^ Tobacco Smoke
l»-Carbon Dust
^ ^
^Calcium, Zinc • Lead Dust ^
Special Use Insulation ^ Fiberglass " Glass Wool ^
Toxtiloa + 4 !*• Fertilizer, Ground Limestone f"
^^ Camant Dust "^.
"" ^ Burning Wood _ *""
^ Kosin smoKe ^
Coal Flue Gas
^ "Anti-Stick Sorav
Paint Pigments
^Alkali Fume ^ —
Face Powder: > Talc
. Radon Progeny
1 Liquid droplets containing bacteria etc., sneezed, etc
2 Man-made mineral fibers
ses of various types of indoor particles.
il (1992).
— ^ Smouldering or Flaming Cooking Oil
^ Oil Smoke Fly Ash
^ Auto Emissions ^ 1
. . . _ > < Ant iperspi rant
=— ^ Spray Pamt Spray Paint Dust
Freshener +.
-, Humidifier £ < Fabric Protector
•^ — ^ Nebulizer Drops fc
™ ^ ~
insecticide Dusts £ __». Emollients
> ^Clumps "" ' MgCO3
Pigment fc _ Binder
f " 4 ^Artificial Textile Fibers
k>
-------
1 non-smokers not using a dust mask can be adequately characterized by a personal monitor worn
2 by the subject with the sampling inlet close to, but not in, the breathing zone.
3 The total PM exposure of an individual during a period of time is composed of exposure to
4 many different particles from various sources while in different microenvironments. Duan
5 (1982) defined a microenvironment as "a [portion] of air space with homogeneous pollutant
6 concentration"; it has also been defined (Mage, 1985) as a volume in space, during a specific
7 time interval, during which the variance of concentration within the volume is significantly less
8 than the variance between that microenvironment and its surrounding microenvironments. For
9 example, a kitchen with a wood stove can constitute a single microenvironment for total PM
10 when the stove is off, and all people in the kitchen would have similar PM exposures. When the
11 stove is on, the kitchen could have a significant vertical PM concentration gradient, and a child
12 on the floor in a far corner and an adult standing at the stove could be exposed to significantly
13 different PM concentrations. When a concentration gradient exists, such as from smoke from a
14 lit cigarette, where the gradient goes from mg/rri to //g/m3, the former definition of a
15 microenvironment breaks down because there is no homogeneous concentration. Alternatively,
16 the latter definition requires a rather impractical specification of a large number of transient
17 microenvironments.
18 In a given microenvironment, such as one in the kitchen example above, the particles may
19 originate from a wide variety of sources. PM may be generated from within (e.g., the stove, deep
20 frying, burning toast), from without (ambient PM entering through an open window), from
21 another indoor microenvironment (cigarette smoke from the living room), or from a personal
22 activity that generates a heterogeneous mix of PM (e.g., sweeping the kitchen floor and
23 resuspending a mixture of PM from both indoor and outdoor sources that had settled out).
24 In general, as a function of space and time, people pass through a series of
25 microenvironments. Thus their average total daily exposure $ //g/m3) to PM can be expressed
26 as the sum of their exposures within the microenvironments they occupy. With appropriate
27 averaging over sets of four classes of microenvironments (e.g.,indoors, ambient-outdoors.
28 occupational, and in-traffic), the average total exposure of a non-smoker can be expressed as
29 follows (Mage, 1985):
30
31 ir = fFt+Et+Et+EtVT f5n
^ \^mlm + ^out1 out ^ r'occLocc ^ ^tra Ltra>'' L P1'
October 1999 5-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 where each value of E is the mean value of total PM concentration in the microenvironment class
2 while the subject is in it, time (t) is the total time the subject is in that microenvironment during
3 the study period, and T is equal to the sum of all times (usually 24-h). Note that ^ is not
4 necessarily a 24-h time-weighted-average (TWA) of PM concentration in an indoor
5 microenvironment. Thatcher and Layton (1995) report that "merely walking into a room
6 increased the particle concentration by 100%". Consequently, an integrated measurement of air
7 quality in an enclosed space that includes time when it is unoccupied, or not occupied by a
8 specific subject, may not be a valid measure that can be used to estimate the exposure of that
9 specific subject while in that microenvironment (Larssen et al., 1993). A measured
10 microenvironmental concentration when the space is unoccupied will tend to be biased low as a
11 measure of the exposure within it during periods of occupancy. For example it is incorrect, as it
12 is for NO2, to associate a PM exposure to a person, while cooking at a gas stove in a kitchen,
13 with a 24-h kitchen PM concentration measurement that is influenced by periods when the stove
14 was off (Smith et al., 1994).
15
16 5.1.1 The History of Understanding Human Exposure to Particulate Matter
17 The 1969 PM AQCD (National Air Pollution Control Administration, 1969) only discussed
18 ambient PM concentrations as indices of a general exposure to PM, to the extent that the
19 document index mentions neither exposure nor indoor. The introduction stated "Air quality
20 criteria are descriptive; that is, they describe the effects that have been observed to occur when
21 the ambient level of a pollutant has been reached or exceeded specific figures for a specific time
22 period. . . .Epidemiologic studies [are presented that] analyze the effects of pollution from
23 ambient exposure of groups of people living in the community." The indoor microenvironment
24 was considered protective and no mention was made of sources of pollution indoors. For
25 example, it was suggested that the decrease in mortality associated with reduction of PM
26 pollution in London, after the 1952 Fog, may possibly have been related to the report that "a
27 great deal of publicity has been given to the harmful effects of smog, and more susceptible
28 individuals have been encouraged to use masks and filters, and to stay indoors."
29 In the context of protection from ambient PM afforded by staying indoors, the document
30 caveats as follows: "There has, however, been a minimum of attention paid to indoorand
31 domestic environments and their potential contribution. Measurement of such indoor exposures
October 1999 5-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 might be difficult, but omission of the information could well modify the appraisal of the
2 importance of particulate pollution" [e.g., ambient PM as measured by a high volume sampler
3 would now appear less potent because susceptible people would stay indoors on high pollution
4 days].
5 The 1982 PM AQCD (U.S. Environmental Protection Agency, 1982) reported on indoor
6 PM concentrations as follows: "Yocom et al. (1971) studied TSP in public buildings, offices and
7 homes using a scaled down version of the hi-vol sampler. . .Indoor levels were about half outdoor
8 levels on the average. . . Alzona et al. (1979) reported elemental analyses for calcium and iron,
9 normally coarse-particle components, and for zinc, lead and bromine, components of fine
10 particles. . . it appears that tracer components of coarse particles do not penetrate any of these
11 structures as readily as the fine components."
12 However, early personal PM exposure monitoring studies indicated that some people's
13 personal activities, along with PM generated by personal and indoor sources (e.g., cigarette
14 smoking), could lead to PM indoors and personal exposures to total PM that exceeded the
15 concentration of the PM found in the immediate outdoor air or in the local ambient air (Binder
16 et al., 1976; Repace and Lowrey, 1980; Spengler et al., 1980). This was reported as follows:
17 "It is apparent that, in the absence of smoking, indoor and outdoor levels of fine particulate mass
18 are almost the same. However, smoking contributes very significantly to indoor level."
19 The section concluded: "Therefore, fine particles readily penetrate buildings and occur
20 inside to about the same extent as outdoors. Indoor activity adds incrementally to outdoor levels
21 and, frequently, somewhat higher levels of fine particles are observed indoors. Smoking adds
22 very materially to indoor levels."
23 The 1982 document also summarized the situation as follows: "Because stationary
24 ambient-air pollution monitors provide general statistics on composite population exposures, it
25 would be extremely difficult (if not impossible) to predict an individual's actual exposure to PM
26 on the basis of community air-monitoring data alone."
27 "Although outdoor concentrations of pollutants can be measured at particular sites, our highly
28 mobile population can be exposed to either higher or lower values than community monitors
29 show. Indoor particle levels can be high because of smoking, cleaning operations, or normal
30 activities. Exposures of individuals to PM can vary more than community monitors show."
31 (Volume H, page 5-138)
October 1999 5-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 In the period between 1982 and 1996, many more studies on personal PM exposure and
2 indoor PM were reported that documented the fact that, in most inhabited domestic
3 environments, indoor PM concentrations, and personal PM exposures of the residents, were
4 greater than the simultaneous ambient PM concentrations (e.g., Sexton et al., 1984; Spengler
5 et al., 1985; Clayton et al., 1993). Clayton et al. (1993) referred to their finding of a 60% excess
6 of daytime personal PM10 exposure in Riverside, CA, relative to a time-weighted-average of
7 indoor and outdoor PM10 concentrations, as a "personal cloud" which is discussed later in this
8 chapter.
9 In 1996 it was known, from personal monitoring and indoor monitoring, that some PM
10 constituents, such as sulfates, are almost always lower indoors than outdoors because of the
11 virtual absence of indoor sources and the presence of sinks for sulfates in indoor settings
12 (exceptions can occur if high sulfur coal or kerosene are used as fuel in a poorly vented stove or
13 space heater). However, this relationship does not hold for most other PM constituents, as the
14 indoor and personal monitoring data show both higher- and lower-than ambient PM
15 concentrations in indoor settings as a function of particle size and human activity patterns.
16 The largest coarse mode particles (>10//m AD), which are generally of non-anthropogenic
17 origin (e.g., wind blown dust), require turbulence to provide vertical velocity components greater
18 than their settling velocity to allow them to be lifted and remain suspended in the air
19 (Figure 5-1). Particles of ambient origin enter into an indoor setting either by bulk flow, as
20 through an open window, in which all particles can enter at the inlet condition, or by pressure
21 driven drafts and diffusional flows through cracks and fissures in the barriers of the building
22 envelope when all windows are closed. In the latter mode of entry, velocities are relatively
23 lower, thereby increasing the settling out of the largest coarse particles (>25//m AD) in the
24 passage through the barriers (Larssen et al., 1993; Thatcher and Layton, 1995).
25 Indoor settings are usually quiescent (Matthews et al., 1989), unless fans or Heating-
26 Ventilation-Air Conditioning (HVAC) systems are in use. Ambient particles that enter indoors
27 quickly settle out by gravity or electrostatic forces, leading to familiar dust layers on horizontal
28 surfaces and vertical TV screens that require constant cleaning (Raunemaa et al., 1989; Kildeso
29 et al., 1999). However, human activity in indoor settings, such as smoking and cooking, does
30 generate fine particles (<2.5//m); cooking, dusting, vacuuming and general activity can generate
October 1999 5-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 coarser particles (>2.5 //m), or resuspend coarse particles that previously had settled out
2 (Litzistorf et al, 1985; Thatcher and Layton, 1995; Abt et al, 1999a,b).
3 The National Research Council (1991) summarized the changing concept of concern for
4 computing a person's total exposure to air pollutants, such as PM, instead of using an ambient
5 concentration as a surrogate, as follows:
6 "Advances in indoor-air exposure studies have demonstrated the significant health effects from
7 indoor emissions and exposures to contaminants that had been regulated only as outdoor
8 pollutants. . .These demonstrations of high indoor contaminant levels showed the importance of
9 accounting for incremental exposures from microenvironments when making risk assessments. . .
10 These data have indicated the potential importance of indoor sources of contaminants."
11 However, from efforts to help resolve the apparently paradoxical situation that health
12 effects were being associated with ambient PM concentrations, but personal exposures to total
13 PM were found to be uncorrelated with ambient PM concentrations, a new realization began to
14 emerge. The early works of Janssen et al. (1995) at Wageningen University, and Tamura et al.
15 (1996a) at Tskuba University, revealed that longitudinal total PM exposures of an individual
16 were highly correlated with ambient PM concentrations because the variance of non-ambient PM
17 exposures (same home and work place from day to day) seemed to be small compared to the
18 variance of ambient PM concentrations. Retrospective analyses of other data (Lioy et al., 1990;
19 Clayton et al., 1993) by Mage and Buckley (1995) also showed that the mean community
20 personal exposure on a given day was much more positively correlated with ambient PM
21 concentrations than the cross-sectional correlation of all the subjects' exposures taken
22 individually and analyzed together.
23 The non-peer-reviewed status of these publications (Tamura et al. [1996a] was in press at
24 the time) led to a very conservative treatment of these findings in the 1996 PM AQCD (U.S.
25 Environmental Protection Agency, 1996). Therefore, the 1996 PM AQCD, Chapter 7,
26 represented a cautious transition away from the 1982 PM AQCD, and, for the first time, it made
27 a distinction between exposure to PM of ambient origin and exposure to total PM of all origins.
28 Consequently, the exposure chapter gave equal emphasis to the presentation and discussion of
29 sources of PM in indoor domestic microenvironments and the infiltration of ambient PM into
30 these indoor microenvironments. However, in summary, several suppositions and conclusions
31 were cited, that, when woven together, could implicitly support the premise that exposures to PM
October 1999 5-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 of ambient origin are highly correlated with ambient PM concentrations and would make a
2 correlation of ambient PM concentration with community health effects plausible. For example:
3 "For the morbidity/mortality studies described in Chapter 12 that use SAM [stationary ambient
4 monitoring] as the independent variable, that SAM can be interpreted to stand as a surrogate for
5 the average community exposure to PM from sources that influence the SAM data." (Volume I,
6 page 7-119).
7 "(4) Personal exposures to outdoor-generated PM of any size fractions PM10 can be estimated
8 from the fraction of time spent indoors and an estimate of the air exchange rate and
9 deposition rate associated with that size fraction.
10 (5) The relationship between ambient concentration and [total] personal exposure [to PM] is
11 better for finer size fractions of ambient PM than for coarser PM. Higher correlations of
12 ambient concentrations and personal exposures have been found for fine PM constituents
13 (such as sulfates) without indoor sources.
14 (6) For a study population of nonsmokers for which there is a significant positive correlation
15 between personal exposures and ambient concentrations, the ambient concentration can
16 predict the mean personal exposure with much less uncertainty than it can predict the
17 exposure of any given individual.
18 (22) Variations in personal exposure due to fluctuations produced by indoor sources of PM are
19 independent of the variations in personal exposure produced by [fluctuations of] ambient
20 PM." (Volume I, pages 7-163 and 7-165)
21 Since 1996, the work of Janssen and colleagues has been peer reviewed and published, and
22 several other related articles discussing it have been published, or accepted for publication [Mage
23 et al., 1999; Wallace, 1999a] - and are reported on later in this document. The net result is that
24 the current literature appears to support the concepts that:
25 (1) ambient PM concentration is an index of exposure to PM of ambient origin, and that this
26 index is currently the most appropriate quantity to relate to human health effects associated
27 with the ambient PM concentrations within communities in epidemiologic studies;
28 (2) a measurements of an individual's personal exposure to total PM from all sources, indoors
29 and outdoors, ambient and non-ambient, is the most appropriate index to use to relate that
30 individual's health effects from that combined exposure to PM from all those different
31 categories of sources.
October 1999 5-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.1.1.1 Caveat
2 In virtually all the experimental studies of personal exposures to PM to be described in the
3 rest of this document, with noteworthy exception of the U.S. Environmental Protection Agency
4 Particle Total Exposure Assessment Methodology (PTEAM) study (Clayton et al., 1993), the
5 Toronto, Ontario study of Pellizzari et al. (1999), and the Expolis exposure study (Jantunen et al.,
6 1998), the studies were all conducted with subjects who were not chosen by a scientific
7 probability-sampling schema (SPSS). Strictly speaking, without an SPSS, the results of all such
8 studies apply only to the subjects sampled on the days that they were sampled, and no valid
9 inference can be made to any other population or period of time. Although such studies may
10 report significant differences, confidence intervals andp values, albeit they were peer reviewed,
11 perhaps not by survey statisticians, they have no statistical meaning.
12 "In many cases researchers are reluctant to face the problems that may be
13 present in the survey. An 'ignorance is bliss' attitude and gratuitous
14 assumptions are made about the quality of the data (the million nonrespondents
15 are adequately represented by the ten respondents). For a one-time survey,
16 conventional wisdom often dictates using methods that are believed to give
17 good results for the funds available without adequate investigation of
18 alternatives." (Lessler and Kalsbeek, 1992).
19 This chapter reports experimental results of such non-SPSS studies for what they are,
20 without the authors' reports of statistical significance where such would not be valid, and it
21 integrates the content of all these studies with the abovecaveat in mind.
22
23 5.1.2 Exposure to PM of Ambient Origin and Total PM
24 Personal exposure to the PM of ambient origin is important for several reasons:
25 (1) The U.S. EPA regulates PM emitted into the atmosphere from mobile sources and stationary
26 industrial or commercial sources, but it does not control PM emissions in any private indoor
27 location.
28 (2) The human body may react differently to PM of ambient origin and PM of non-ambient
29 origin, because such particle mixtures have different chemical composition.
30 (3) Comparison of personal exposures to mixtures of PM of ambient origin and personal
31 exposures to mixtures of PM of non-ambient origin, if possible to differentiate, may provide
October 1999 5-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 clues as to the difference in their acute toxicity on a unit size and mass basis (Siegmann
2 etal, 1999).
3 (4) Often in epidemiologic studies, the lagged daily mean ambient PM concentration level in a
4 community is used as a surrogate to characterize a subject's exposure to ambient PM.
5 Personal exposure to PM of non-ambient origin might act as a confounder or an effect
6 modifier in such epidemiologic studies, if people's exposures to non-ambient PM are
7 correlated with their exposures to ambient PM. This is true even if exposure to non-ambient
8 PM may cause similar but independent health effects.
9 In general one can think of an individual's total exposure to PM as arising from several
10 distinct categories: 1 - Ambient PM; 2 - Outdoor PM; 3- Indoor PM; 4- Personal activity PM;
11 5- Personal PM. Figure 5-2 shows how people can be exposed to combinations of these
12 categories while outdoors as well as indoors. The following describes these categories in detail
13 to clarify their differences and definitions as used in this chapter:
14 (1) Ambient PM: As defined in Section 5.1, PM of ambient origin is that PM that is formed in
15 the ambient atmosphere and emitted into it. Ambient PM is well mixed in the outdoor air so
16 that all people in the community are exposed to it, over time, at approximately the same
17 mean concentration. This is true more so for fine mode PM than for coarse mode PM. The
18 major sources of primary and secondary ambient PM species are industry, traffic, commerce,
19 domestic emissions such as wood smoke, and natural wind blown dust or soil (see
20 Chapter 3).
21 (2) Outdoor PM: PM of outdoor origin differs conceptually from PM of ambient origin. It is
22 measured as the difference between the PM concentration at an outdoor location which is
23 not in the ambient atmosphere (i.e., on private property or by a road side) and the
24 simultaneous ambient PM concentration. Scaperdas and Colvile (1999) give an example of
25 such non-representative air quality monitored outdoors 5-meters from an urban intersection.
26 (3) PM emitted or formed indoors: PM is emitted at home as house dust resuspended by human
27 activity and cleaning procedures, environmental tobacco smoke (ETS), cooking fumes, pets,
28 etc. PM emitted indoors at work varies with type of occupation. Aerosol formation occurs
29 indoors from the dark (no sunlight) reaction of ozone with gaseous terpenes and other
30 hydrocarbon species such as cc-pinene and limonine, often found in household deodorizers.
31 Kamens et al. (1999) reported " Some of the products have subcooled liquid vapor pressures
October 1999 5-11 DRAFT-DO NOT QUOTE OR CITE
-------
People Are Exposed to a Variety
of Types of Particles Outdoors
Local
Outdoors,
Community
Ambient,
Personal
Activity
pact
Individual Only,
pind
People Are Exposed to a Variety of
Types of Particles Indoors
Indoor/Ambient
Interaction, Pa x
Personal
Activity
Pact
Indoor sources,
Pi
Outdoor Ambient
Infiltrated
> Indoors, Pa|
'S£ S S SS SS S 'SS /"^
/ I \
Individual Only,
pind
Figure 5-2. Categories of particle exposure outdoors and indoors.
Source: Wilson and Mage (1999).
October 1999 5-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 which are low enough to initiate self-nucleation." Weschler and Shields (1999) report on a
2 study of these dark ozone reactions, and report "The results demonstrate that ozone/terpene
3 reactions can be a significant source of sub-micron particles in indoor settings", as high as
4 95 //g/m3 under simulated conditions. Indoor generated PM is considered to be well mixed
5 in an indoor microenvironment.
6 (4) Personal activity PM: Personal activity sources can exist either indoors or outdoors. These
7 are microscale PM generating activities that primarily influence the exposure of the person
8 performing the activity, from either a PM generating activity (hobby or occupation) or a
9 physical activity that brings the subject into the undiluted PM plume from a local source
10 (standing on a street corner or holding a lit cigarette between puffs). Thus personal activity
11 PM exposure is only measured by a personal monitor carried by the subject, because a
12 stationary monitor located nearby will not measure the high PM concentration generated by
13 that activity. The difference between a personal monitor measurement and an
14 area-representative measurement several meters away is sometimes called a "personal
15 cloud" (Wallace, 1999a).
16 (5) Exposure to personal PM: This category pertains to all PM exposures that cannot be
17 measured by a personal exposure monitor (PEM). i) Actively smoking creates a stream of
18 smoke with a high PM concentration that is inhaled directly and it is not sampled by a
19 personal monitor carried by the subject. This contribution to a daily exposure can be
20 estimated from the Federal Trade Commission (FTC) ratings of mg delivered divided by the
21 estimated total ventilation volume inhaled during a 24-h period by the smoker (Federal
22 Trade Commission, 1994). For a smoker who breathes at an average rate of 10 Lpm over a
23 day, each 1 mg of tar inhaled represents an exposure increment of 70//g/m3 to their daily
24 PM exposure as measured by a personal PM monitor, ii) Wearing a courtesy-mask for a
25 respiratory infection, or a dust-mask or respirator for an occupation or hobby, removes an
26 unknown fraction of the PM measured by a PEM in the inhaled air. The resulting
27 contribution to daily PM exposure can only be estimated from the efficiency of the filter
28 given by the manufacturer and the PEM data.
29 An important distinction that was developed in the 1996 PM AQCD between exposures to
30 ambient PM25 and non-ambient PM25 (categories 2 to 5 above) is the relative homogeneity of the
31 concentration of PM25 of ambient origin compared to the heterogeneity of exposures to PIV|5 of
October 1999 5-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 non-ambient origin. As discussed in the following section, virtually all people in a community
2 are routinely exposed to a similar mixture of PMj 5 of ambient origin when in the ambient
3 atmosphere, with more heterogeneous exposures to coarse mode PM and semivolatile PM
4 constituents in the western U.S. than the eastern U.S. See the discussion in Chapter 4 of the
5 spatial and temporal variability of chemical composition of ambient PM in rural and urban
6 communities in different regions of the U.S.
7
8
9 5.2 EXPOSURES TO PM CONCENTRATIONS IN THE COMMUNITY
10 AMBIENT ATMOSPHERE
11 This section addresses the question, "How well does a concentration of PM measured at a
12 community ambient monitoring station reflect the PM concentration in the outdoor air elsewhere
13 in that area at both the local and regional levels?"
14 Spatial variation ofPMat the local level. Kotchmar et al. (1987) measured TSP, and fine
15 (PM25) and coarse (PM10 - PM25) using dichotomous samplers, in five cities across the U.S., at
16 Bakersfield, CA; Riverside, CA; Granite City, IL; Owensboro, KY; Tampa, FL. In each city
17 three identical PM monitors of each type were sited in accord with EPA guidelines at separations
18 ranging from 1.6 km to 10 km. The monitors were run simultaneously for periods ranging from
19 two weeks to two months. The means of the PM^ monitoring data within each city were found
20 to be highly uniform but the coarse PM data had a major variability. The authors concluded
21 "Only fine particles were found to have equivalent mean concentrations, suggesting that only one
22 monitoring site in each community is required to provide an adequate exposure estimate of the
23 outdoor component. However, variable concentrations of total inhalable and/or coarse particles
24 were found, which implies a requirement for multiple monitoring sites [for sampling ambient
25 coarse mode particles]".
26 Quackenboss et al. (1991) studied exposure to PM;0 and related health effects in Tuscon,
27 AZ. The ambient PM10 data measured by the Pima County Department of Environmental Quality
28 (PCDEQ), method unspecified, were compared to the outdoor PIV(0 measured at the homes of the
29 subjects in the study, using a Harvard indoor aerosol sampler (Marple et al., 1987). The
30 distances between the homes and the PCDEQ monitors ranged up to 20 km. The Harvard-
31 indoor-sampler data measured outdoors had a statistically significant slope of 0.63 ± 0.03 when
October 1999 5-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 fit through the origin [intercept = 0] versus the ambient PM10 values reported by PCDEQ. The
2 authors do not report any collocated inter-comparison between these two different monitors to
3 document whether the 0.63 factor was more due to differences in instrumental inlet penetration
4 curves, than to higher PM at the ambient monitoring sites. The spread of the daily outdoor PMg
5 about the ambient values was quite large (R2 = 0.185), but no data are provided to show whether
6 this variance is predominantly in either the coarse mode or the fine mode of the PMg.
7 Lillquist et al. (1998) measured outdoor PMj0 at three hospitals in Salt Lake City and
8 compared the measurements from November 29, 1994 through April 29, 1995 with the
9 simultaneous measurements at the Utah Department of Air Quality (DAQ) ambient monitoring
10 station nearby. The three hospitals were 3.4, 5.6 and 12.5 km from the DAQ site, and the mean
11 PM10 values measured were quite different, leading to the conclusions that "Under conditions of
12 high atmospheric PM10 concentrations...Salt Lake City, UT requires more than one monitor.
13 When ambient PM10 concentration data are used as a predictor of individual exposure, more than
14 one centralized monitor is absolutely necessary."
15 These results for the Salt Lake City, UT area are in direct contrast to those reported by Pope
16 et al. (1999) at a similar scale of separation, in the same season and general area (Provo, UT in
17 the Utah Valley) one year later (November 18, 1995 - March 15, 1996). In this period the Utah
18 Valley PM10 data monitored at three sites (at separations of 4 to 12 km) were virtually identical,
19 with Pearson correlations of 0.92 and 0.96. The greater degree of variability in the PMg of the
20 Salt Lake City, UT area, relative to the Provo, UT area, may be related to the higher presence of
21 wind-blown crustal material in the Salt Lake City area. Pope et al. (1999) reported that increased
22 health effects in the Utah Valley were associated with stagnation and thermal inversions leading
23 to a buildup of anthropogenic PMj0, whereas the similarly high concentrations of PM,0 created by
24 high winds picking up crustal materials were not associated with increased health effects in that
25 same community. Thus, this is an indication that it may be important to differentiate exposure to
26 PM of ambient anthropogenic origin from locally-variable high concentrations of wind blown
27 dust, as in Salt Lake City, which may not be as important for health effect prediction as
28 anthropogenic ambient PM.
29 Buzorius et al. (1999) measured short-term aerosol number concentrations by using a
30 condensation particle counter at several locations in metropolitan Helsinki, Finland. They report
31 that number concentration can vary in magnitude with the local traffic intensity, and that "during
October 1999 5-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 the working days concentration averages of 10 min - 1 h are good representatives of
2 concentration variation in relatively large areas of the city." They conclude, "Therefore, by
3 sampling at one point in the [urban] space one can describe changes in relatively large area of the
4 city with correlation coefficient > 0.7".
5 Dubowski et al. (1999) point out that although the relationship of small variation of PM5
6 mass concentration may hold for a community, there may be significant spatial variations of
7 specific components of the total mass on a local scale. An example is given of a study of
8 concentrations of polycyclic aromatic hydrocarbons (PAH) at three indoor locations in a
9 community; urban and semi-urban separated by 1.6 km and a suburban site located further away.
10 The authors found the geometric mean PAH concentrations at these three locations varied
11 respectively as 31 : 19:8 ng/m3, and suggest that the local variations in traffic density were
12 responsible for this gradient. Note that these concentrations are 1000 times lower than the total
13 mass concentration, so that such a gradient of 0.03 to 0.01//g/m3 for these components would not
14 be noticeable for total PM25 mass measurements of order 25//g/m3.
15 Jedrychowski and Flak (1998) report on the spatial variation of suspended particulate
16 matter (SPM) measured as "black smoke" (BS) in Cracow, Poland during the years 1991 - 1995.
17 The authors report that, for both winter and summer, the city could be divided into a central
18 "high pollution zone" and regions of lower pollution concentrations decreasing with distance
19 from the city center. The authors do not report the size fraction of the PM collected or the
20 calibration procedure used to convert the BS reading into SPM//g/m3. Thus, their results may
21 emphasize the effect of the black carbon content related to local traffic density variations which
22 could produce a gradient signal imposed on a more uniform background of well mixed emissions
23 ofPM.
24 Vakeva et al. (1999) measured the vertical gradient of submicron particles in an urban
25 street canyon of Lahti, Finland. They monitored number concentration by using a TSI screen
26 diffusion battery and a condensation particle counter at 1.5 m and 25 m above the street at
27 rooftop level. "It was concluded that dilution and dispersion decreases the concentrations of
28 pollutants emitted at street level by a factor of roughly 5 between the two sampling heights."
29 The presence of such a local vertical gradient of concentrations for people living in high-rise
30 buildings may need to be considered in studies of exposures to PM of urban populations.
October 1999 5-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spatial variation of ambient PM on a regional scale. Burton et al. (1996) report on the spatial
2 variability of sulfates and PM in metropolitan Philadelphia, PA and their data, also discussed by
3 Wilson and Suh (1997), show very little variance in spatial mass concentrations. They showed
4 that there was a gradient of ammonium sulfate across the city, with a maximum in the urban area
5 center, indicative of ammonia generation by human activities, leading to a composition variation
6 of that component of urban PM.
7 Keywood et al. (1999) reported their analysis of PM variability measured in six Australian
8 cities using a MOUDI sampler. They reported that PM^ was more highly correlated with PM2 5
9 than with coarse PM (PM10 - PM2 5), suggesting that "variability in PMj0 is dominated by
10 variability in PM25". Although the authors found that mass of PM^ was highly correlated with
11 mass of PMj (r2 = 0.98), the mass of the ultrafines (nuclei mode), reported both as PIVj 15 and by
12 integration under a fitted curve, had weak correlations with PM.5 of r2 = 0.50 and 0.01,
13 respectively. This suggests that although a single monitoring station may be adequate for
14 characterizing fine mode ambient PM in a community, the ultrafine mode mass of ambient PM,
15 like the coarse mode mass, may require additional monitoring.
16 Leaderer et al. (1999a) monitored 24-h PMj0, PM25 and sulfates during the summers of
17 1995 and 1996 at a regional site in Vinton, VA (6 km from Roanoke, VA). One similar 24-h
18 measurement was made outdoors at residences in the surrounding area, at distances ranging from
19 1 km to > 175 km from the Vinton, VA site, at an average separation distance of 96 km. The
20 authors reported significant correlations forPM2 5 and sulfates between the residential outdoor
21 values and those measured at Vinton, VA on the same day. In addition, the mean values of the
22 regional site and residential sitePM25 and sulfates showed no significant differences in spite of
23 the large distance separations and mountainous terrain intervening in most directions. However,
24 for the concentrations of the coarse mode PM, estimated as PM0 - PM25, no significant
25 correlation among these sites was found (n = 30, r = -0.20).
26 Jantunen et al. (1998) and Koistinen et al. (1999) report on the protocol and quality
27 assurance procedures of the EXPOLIS exposure monitoring study that included measurements of
28 ambient and microenvironmental concentrations of PM,5, and personal exposures to PM^.
29 A planned article (Jantunen et al., 2000) is expected to report the final results of their exposure
30 analyses on these measurements. Their preliminary results show that: (1) in Basel and Helsinki,
31 a single ambient monitoring station was sufficient to characterize the ambient PM5
October 1999 5-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 concentration in each city - only a few simultaneous measurements of ambient PM5 were made
2 in the four other EXPOLIS cities (Athens, Grenoble, Milan and Prague) so no other intra-city
3 comparisons are available at this time; and (2) by using microenvironmental concentration data
4 collected while the subjects were at home, at work, and outdoors, the time-weighted-averages of
5 these data closely match the personal PM, 5 exposure data collected by the monitors carried by
6 most of the subjects, with a few subjects, mostly smokers, being noticeable exceptions.
7 In summary, a PM2 5 measurement at a properly cited ambient monitoring location can
8 represent the mixed mean concentration of ambient PM, 5 that exists in the ambient atmosphere
9 of that local area. Coarse mode PM is apparently influenced more by local sources than PM5 so
10 this finding of a single monitoring site as being sufficient for PIV|5 is not generally applicable to
11 the coarse fraction of PM.
12 White (1998) suggests that the higher random measurement error for the coarse PM
13 fraction compared to the error for the fine PM fraction may be responsible for a major portion of
14 the apparent greater spatial variability of coarse ambient PM concentration compared to fine
15 ambient PM concentration in a community (e.g., Burton et al., 1996; Leaderer et al., 1999a).
16 When PM25 and PM10 are collected independently, and the coarse fraction is obtained by
17 difference, as PM10 _2 5 = PM10 - PM25, then the expected variance in the coarse fraction is the
18 sum of the variances of the PM,0 and PM25 measurements. When a dichotomous sampler
19 collects PM25 and PM10_25 on two separate filters, the coarse fraction also is expected to have a
20 larger error than the fine fraction. There is a possible error due to loss of mass below the
21 cut-point size and a gain of mass above the cut-point size which is created by the asymmetry of
22 the product of the penetration times PM concentration about the cut-point size. Because a
23 dichotomous PM sampler collects coarse mass using an upper and lower cut-point, it is expected
24 to have a larger variance than for the fine mass collected using the same lower cut-point.
25 Carrothers and Evans (1999) also discuss the effect of relative measurement errors (both
26 instrumental error and Berkson-type error) when analyzing the relative toxicity of ambient PM5
27 to that of the coarser mode (PMj0 - PM2 5 ). They present a model that allows an estimate of the
28 relative bias in the regression coefficients of coarse and fine PM on daily mortality, and conclude
29 that "if one pollutant is truly more harmful than the other, then it must be measured more
30 precisely than the other, in order not to bias the ratio of the fine and coarse regression
31 coefficients". The authors note "the need for spatial variability data, i.e., several ambient
October 1999 5-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 monitors situated across a metropolitan area for a period of many months" so that "definitive
2 conclusions can be made regarding the possibility of bias due to differences in measurement
3 [Berkson-type] error among correlated pollutants".
4 Wilson and Suh (1997) review the difference between the chemistry, sources and factors
5 influencing exposures to generic fine and coarse ambient PM (see Chapter 3). They point out
6 "The infiltration factor, which gives the fraction of outdoor particles found indoors, ... is greater
7 for the fine particles than for coarse particles, largely because of the lower indoor lifetimes of
8 coarse particles relative to fine particles." Wilson and Suh (1997) hypothesize that because the
9 ambient monitoring data for PMj 5 are more constant across an urban area than the corresponding
10 concentration data for coarse mode PM, represented by (PM^ - PM25), and PM10 is more highly
11 correlated with PM25 than with (PM10 - PM25), the health effects associated with total PM,0 are
12 more likely to be related to the variations of the fine mode portion of the PIVf5 than the
13 variations of the (PM10 - PM25) measure of the coarse fraction.
14 The authors note that PM25 is most often a mixture of fine mode PM and coarse mode PM
15 because the lower tail of the coarse mode extends below 2.5//m AD, and since 2.5 //m is a 50%
16 cut point, not a 100% cut point, a PM^ sampler collects some PM > 2.5//m. They conclude that
17 "Fine and coarse particles are separate classes of pollutants and should be measured separately in
18 research and epidemiologic studies. PM;0 and PM(10_25) are indicators or surrogates, but not
19 measurements, of fine and coarse particles." Janssen et al. (1999a) report "A method to estimate
20 the distributions of various fractions of PM,0in ambient air in the Netherlands". However, their
21 method estimates the PM25 fraction, but not the fraction of PMj0 that is exclusively in the fine
22 mode aerosol, as recommended by Wilson and Suh (1997).
23
24
25 5.3 EXPOSURES TO AMBIENT PM IN INDOOR
26 MICROENVIRONMENTS
27 Besides the exposure to ambient PM while outdoors, people are also exposed to PM of
28 ambient origin in the non-ambient indoor-type microenvironments of the residence, workplace,
29 school, motor vehicle, etc. Ambient PM enters indoors from the outdoors by both forced and
30 free convection. An equal quantity of indoor air must also leave through the same type of
October 1999 5-19 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
passages to maintain the pressure equal between the indoors and outdoors. Figure 5-3 shows an
idealized indoor micro environment exchanging air with the ambient surroundings.
v air flow out
Co outdoor
ambient PM
concentration
airflow in
P fraction of PM
penetrating into
the indoor region
Volume V
emission rate =XM
C Indoor concentration
of ambient PM
I K deposition rate
Surface Area (A) with Mass (M) of Ambient PM on it
Figure 5-3. Two compartment model for PM deposition and resuspension by human
activity in a residential microenvironment.
1 As described by Alzona et al. (1979), Tung et al. (1999) and Kulmala et al. (1999) the
2 ambient PM intrusion process can be modeled for a well mixed indoor volume f7) by a mass
3 balance equation between start time t = 0, and stop time t = T, with initial condition C = C(0),
4 @ t = 0. Here C(0) is the concentration of PM that originated from the ambient air that is found
5 in the indoor air while the adjacent outdoor air has a PM concentration of Co = Co(0), @ t = 0:
VdC/dt= vPCo -
(5-2)
October 1999
5-20
DRAFT-DO NOT QUOTE OR CITE
-------
1 where V= volume of the well mixed indoor air, m3;
2 v = volumetric air exchange rate between the indoor and outdoor condition, rrl/hr;
3 P= fraction of ambient PM that is not deposited on the inlet surfaces during the entry into
4 the indoor volume from the ambient condition, a function ofv, 0 < P < 1;
5 Co = concentration of ambient PM in the outdoor air that enters into indoor volumeF,
6 //g / m3;
7 C= concentration of the ambient PM in the indoor volume V, //g /m3;
8 k= deposition parameter for the ambient PM upon the interior surfaces within volumeF,
9 1/hr; a function of the distribution of PM aerodynamic diameter (AD), presence of air
10 cleaners or in-line filters in air circulation systems, etc.
11 Qa = A discontinuous rate of resuspension of some of the ambient PM that had entered
12 volume Fat time t > 0 and had been deposited on interior surfaces,//g/hr; Qa > 0.
13 All the parameters, save for V, are assumed to be functions of time and PM AD. For
14 example, the deposition parameter (fc) can increase as a step function if an air cleaner is turned
15 on. Anuszewski et al. (1998) showed that in-line filters in heating systems may remove PM of
16 optical diameters less than 1 //m. The volumetric air exchange rate (v) is always finite, but it
17 varies with wind speed, indoor to outdoor temperature difference, and as windows are opened or
18 closed. For particles larger than 1 //m AD, which can settle by gravity, the penetration factor/3 is
19 expected to decrease with decreasing air exchange flow rate (v) because the time that is available,
20 for gravitational deposition in passage through cracks and fissures, increases as the flow rate
21 decreases. Suh et al. (1993) showed that operation of an air conditioning system leads to usage
22 of a lower air exchange rate, resulting in lower PM concentrations of ambient constituents, such
23 as sulfates. The resuspension parameter (Q) can be zero when the volume Vis unoccupied, or
24 while occupied but people are sleeping or sedentary and Q > 0 when people are active within V.
25 It is important to note that the Koutrakis et al. (1992) reformulation of Equation 5-2,
26 presented as Equation (7-3) in U.S. Environmental Protection Agency (1996), was used to
27 compute average values of deposition rate (fc) independent of any resuspension of deposited
28 material. This deposition parameter (fc) will vary widely for PM10 as its relative amounts of fine
29 mode and coarse mode PM vary. Therefore, the term Qs (emission factor of indoor sources,
30 in units of//g/hr) in Equation (7-3) op. cit., contains within it the resuspension of PM from
October 1999 5-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 ambient sources that was deposited on previous days or earlier on the very day of the
2 measurement.
3 If C(0) is of the order of Co(0) and we assume an average value for Q, we can estimate the
4 average steady state value of C. This is done by setting dC/dt = 0 in Equation 5-2 and solving for
5 C as follows:
6
/ (a+k\ (5-3)
7 where a =v/V, the number of air exchanges per hour.
8 The outdoor PM (Co) at the location of the air inlet to indoor volumeFmay not be exactly
9 equal to the ambient PM concentration (Ca) measured at the neighborhood air monitoring station
10 that can be several kilometers away. Let us assume that there are no major sources of PM
1 1 between the monitoring station where Ca is measured and the location where Co is measured,
12 and that the surrounding urban area is relatively homogeneous in terms of traffic, residential
13 communities and commercial activities including light industry. With this frame of reference, we
14 can model the measured outdoor concentration (Co) as equal to the measured ambient
15 monitoring station concentration (Ca) plus a small random component ^) that has a mean of
16 order zero and a finite variance.
17 This parameter (e) covers the true spatial variation of ambient PM resulting from micro-
18 scale weather variations and local sources. For example, on some days there is a wind vector
19 component from the monitoring station to the modeled indoor location and on other days the
20 vector will be from the indoor location to the monitoring station. Sophisticated methods that
21 account for presence of sources and the topography between ambient monitoring stations are
22 available for estimating e by interpolating ambient pollutant concentrations between monitoring
23 stations (Beyea and Hatch, 1999). The parameter e also allows for random measurement errors
24 from the weighing of the filters and the measurement of the flow rate. This leads to the
25 relationship for the concentration of ambient PM in the indoor micro environment (C), in terms of
26 the measured ambient concentration (Ca) as:
27
+ k) (5-4)
28
October 1999 5-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 Equation 5-4 indicates that an indoor microenvironment will tend to equilibrate with the
2 ambient PM at a fractional value of the ambient PM concentration.
3 Clayton et al. (1993), Ozkaynak et al. (1996a) and U.S. Environmental Protection Agency
4 (1996) report the results of the Particle Total Exposure Assessment Methodology (PTEAM)
5 Study carried out in Riverside, CA in 1990. 178 subjects carried personal PM0 monitors for one
6 day each, while PM10 was being monitored in their home, outside their home (Co) and in their
7 community (Ca). At each home, an air exchange rate was measured using a continuously
8 emitting source of a perfluorocarbon tracer (PFT). Using the procedure of Koutrakis et al.
9 (1992), Ozkaynak et al. (1996a) determined average values ofP and k for PM10 of 1 and 0.65/hr
10 respectively, and 1 and 0.39/hr for PM^. For the 178 homes, 174 values of a were successfully
11 obtained during the daytime period from approximately 7am to 7pm. The histograms of the
12 values of P a I (a+k) for the 174 monitored homes are shown for PM,0 and PM2 5 as Figure 5-4.
13 These values represent an estimate of the average fraction of the outdoor PM that was found
14 inside the home. Because of the positive resuspension term (Q) the actual values will be
15 expected to be higher than the values shown. However, there is minimal resuspension of
16 submicron particles and very little PM2 5 is resuspended (U.S. Environmental Protection Agency,
17 1996) so the reported values of k and Qis in the PTEAM study for PM25 are unlikely to have been
18 appreciably affected by ambient PM resuspension. Exposure to the ambient accumulation mode
19 PM (-0.1 //m < AD < ~1 //m ) is inferred by the relations of exposure to sulfur and sulfates
20 which are predominantly in this size range (Leaderer et al., 1999a). PM in this size range has a
21 deposition parameter of order 0.2/hr (Ozkaynak et al., 1996b).
22 In summary, the indoor microenvironment will have an appreciable amount of the ambient
23 PM25 equilibrated within it. In the case of the PTEAM homes during the fall season in
24 Riverside, CA, where temperatures were moderate and homes had an air exchange rate of
25 approximately a = 1/hr, the fraction of the ambient PMj0 to be found within the indoor residences
26 was approximately !/(!+ 0.65) -0.6. For PM25 the fraction 17(1+0.39) ~ 0.7. For sulfate the
27 fraction 1/(1 + 0.16) = 0.85. The combined exposure to PM of ambient origin both indoors and
28 outdoors is analyzed in Section 5-7.
29
30
October 1999 5-23 DRAFT-DO NOT QUOTE OR CITE
-------
(a)
04 0 45 05 0 55 0 60 0 65 0 70 0.75 D 80 0.85 0 90 0 95
Fraction of PM25 Found Indoors
"to 15'
£
(b)
0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.8!
Fraction of PM10 Found Indoors
Figure 5-4. Histograms of the estimated fractions of outdoor PM2 5 (a) and outdoor PM10
(b) found indoors during the PTEAM study in Riverside, CA (Computed from
data of Ozkaynak et al., 1996a).
1 5.4 EXPOSURES TO PM OF INDOOR ORIGIN
2 U.S. Environmental Protection Agency (1996), Wallace (1996), and Ott and Roberts (1998)
3 review and discuss the pre-1996 literature on PM as found indoors from indoor sources. Most
4 particles generated indoors by human activities have different chemical and physical properties
5 than those generated by anthropogenic ambient sources (Siegmann et al., 1999). In the U.S., in
6 general, combustion product PM from ambient sources is from the burning of fossil fuels (e.g.,
7 coal, gasoline, fuel oil) and wood, and combustion product PM from indoor sources is from
8 biomass burning (e.g., tobacco, wood, foods, etc.). However, some indoor sources of PM, such
9 as cigarette smoking, meat cooking and coal burning (in China), occur both indoors and outdoors
October 1999
5-24
DRAFT-DO NOT QUOTE OR CITE
-------
1 and may constitute an identifiable portion of the measured ambient PM by use of source
2 apportionment techniques (Cha et al., 1996; Kleeman and Cass, 1998). During the PTEAM
3 study (Ozkaynak et al., 1996a) some non-ambient particles that appeared on personal monitoring
4 filters were identified as skin flakes, fibers and carpet debris. These PM species are considered
5 as "inert or nuisance dusts" by the U.S. Department of Labor (Code of Federal Regulations,
6 1998), which has established an 8-hour PM, 5 time-weighted-average (TWA) occupational
7 standard of 5,000 //g/m3 for controlling them. Some other non-inert PM species of indoor or
8 occupational origin may be carcinogenic (e.g., cigarette tars, radon progeny) or produce chronic
9 effects (e.g., silica, asbestos). However, there is no evidence that day-to-day fluctuations of
10 personal exposures to PM species such as these that are not known to produce acute effects at
11 lower concentrations, could cause the day-to-day fluctuations of the acute health effects
12 (mortality and morbidity) that are associated with the day-to-day fluctuations of ambient PM
13 concentrations in epidemiologic studies (Schwartz et al., 1999).
14 The major sources of indoor PM in the residence include cigarette smoking, cooking, and
15 unvented or poorly vented combustion devices such as stoves and kerosine heaters. Human and
16 pet activities also lead to PM detritus production from track-in soil, fabrics, skin and hair, home
17 furnishings, etc., which can all be found in the ubiquitous house dust found on floors and in the
18 lint trapped by the filter of a clothes dryer. This generic house dust and lint is suspended in the
19 indoor air by air movements and, after deposition, it can be resuspended by cleaning activities of
20 sweeping, dusting and vacuuming. Other sources of PM in the home may arise from hobby
21 activity, and from para-occupational materials brought into the home by workers on their persons
22 (Sterling et al., 1995). Biological aerosols commonly found indoors are discussed in Chapter 7
23 of U.S. Environmental Protection Agency (1996).
24 Abt et al. (1999a) studied the PM size distribution and sources of PM in four non-smoking
25 households in the Boston metropolitan area, and confirm previous findings that the major indoor
26 emission sources of PM are cooking, cleaning and human activity. They discuss the size
27 characteristics of these ubiquitous sources and report "The size of the particles generated by these
28 activities reflected their formation processes, with combustion processes (oven cooking, toasting
29 and barbecuing) producing fine particles, and mechanical processes (sauteing, frying, cleaning,
30 and movement of people) creating coarse particles." The authors suggest that at air exchange
October 1999 5-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 rates of less than 1/hr the indoor sources of PM predominate in the indoor micro environment and
2 at air exchange rates above 2/hr, the outdoor sources of PM predominate.
3 The smoking of cigarettes is the major contributor to PM concentrations from indoor
4 sources in the homes where people smoke. Neas et al. (1994) report from the Harvard 6-City
5 Study that the between 1983 and 1986, annual average PM, 5 was 3 l//g/m3 higher in the homes of
6 consistent smokers than in the homes of consistent non-smokers. An extensive investigation has
7 been recently performed on environmental tobacco smoke (ETS) exposures of non-smokers in
8 homes and workplaces in Europe by Phillips et al. (1994, 1996, 1997a,b, 1998a,b,c, 1999). For
9 example, Phillips et al. (1999) sampled non-smokers homes and workplaces in Basel,
10 Switzerland and computed annual total ETS exposures as the mean "Potential inhaled quantity"
11 in mg of RSP. They found that the median exposures of non-workers in smoking households
12 were 15% higher than in smoking households, and at the 90th percentile that smoking
13 households were 100% higher than the non-smoking households.
14 Jenkins et al. (1996a,b) made similar measurements in the U.S. and report that smoker's
15 homes in 1993 and 1993/1994 averaged 17 and 20 ug/m3 above the mean values in non-smoker's
16 homes, respectively. The higher effect of smoking (31 ug/ni) reported by Neas et al. (1994) in
17 1983 - 1986 may have been related to changes in smoking habits in the decade between these
18 studies, such as smoking reduced tar cigarettes and smoking less cigarettes in the home in the
19 more recent Jenkins study.
20 Klepeis et al. (1996) measured PM, 5 due to environmental tobacco smoke (ETS) using a
21 TSI 8510 piezobalance. They measured air exchange rates in two airport glass-enclosed smoking
22 lounges and estimated a rate of emission of 1.43 mg/min per cigarette smoked by means of a box
23 model for a well-ventilated lounge volume. The PM, 5 coefficient of variation (o///) in the room
24 was 0.12 indicating that the lounge was indeed well-mixed. The authors concluded that personal
25 exposures to ETS can be satisfactorily modeled in such microenvironments. Such a model may
26 be useful for calculating the non-ETS PM found in indoor microenvironments where smokers are
27 present, which is often mostly of ambient origin.
28 For smoker's homes in Riverside, CA, the PTEAM study (Ozkaynak et al., 1996a) reported
29 that ETS constituted about 75% of the PM^ generated by indoor sources and 55% of the PM,0
30 generated by indoor sources, which corresponds to approximately 35% of the coarse PM
October 1999 5-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 (PM10-PM2 5) generated by indoor sources. For those homes (smoking and non-smoking) in
2 which cooking took place, cooking fumes were responsible for 66% of the PIV|5 indoor
3 emissions.
4 Occupational exposures to PM generated indoors are a major source of PM exposure for
5 "blue collar" workers involved in the "dusty trades". OHSA standards for PM are both specific,
6 for industries such as coal mining and textile manufacturing which have significantly different
7 toxicities of their PM (e.g., coal dust vs cotton dust), and generic, for all "inert and nuisance
8 dusts" not specifically mentioned (Code of Federal Regulations, 1998). An 8-h time-weighted-
9 average (TWA) PM35 generic standard of 5 mg/m3 has been established, in part, to maintain
10 visibility in work places for personal safety, as well as for respiratory health protection. It is
11 interesting to note that the American Conference of Governmental Industrial Hygienists (1996)
12 has established similar guidelines for generic PM, not specific to an industry ("containing no
13 asbestos and < 1% crystalline silica"), as "Particulates Not Otherwise Classified (PNOC)". Their
14 recommended 8-h TWA for such PNOC as "inhalable particulate" (PM^o) and "respirable
15 particulate" (PMO are 10 mg/m3 and 3 mg/m3, respectively.
16
17
18 5.5 PERSONAL EXPOSURES TO PM OF ONE'S OWN PERSONAL
19 ACTIVITIES
20 Personal activities, such as body motion, cigarette smoking, hobbies and occupational tasks
21 may generate a plume of particles that abruptly decreases in concentration with distance from the
22 person generating the particles. This is especially important in certain occupational settings.
23 Average concentrations of inhalable dust [~PM50, American Conference of Governmental
24 Industrial Hygienists (1996)] over 50 mg/m3 have been measured by personal monitoring of
25 agricultural activities, with an average respirable fraction (PM;) of 4.5 mg/m3 (Nieuwenhuijsen
26 et al, 1999).
27 Teschke et al. (1999) report on personal total PM exposure data (~PM50) from workers in
28 industry involving wood-production, wood-finishing and wood-construction, as collected by the
29 U.S. Occupational Safety and Health Administration Integrated Management Information System
30 (Stewart and Rice, 1990). The data set consisted of 1632 observations over the period from 1979
October 1999 5-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 -1997. The arithmetic mean exposure was 7.93 mg/m3, the geometric mean was 1.86 mg/m3,
2 12 values exceeded 100 mg/m3, and the maximum reported value was 604 mg/m3.
3 During the time of the existence of the activity plume, a subject will be exposed to a much
4 higher PM concentration than would be measured by a stationary PM monitor several meters
5 away. Mage and Ott (1996) analyzed this situation for a cigar being smoked in a large tavern.
6 During the period while the cigar is burning (called thecc period) and during the period
7 immediately after the cigar is extinguished, during which the plume is mixing into the rest of the
8 indoor air (the P period), the person generating that plume will be exposed to a higher
9 concentration of PM than an indoor monitor located in the same room. Then, after the plume is
10 mixed (the y period), the subject and a monitor anywhere in that microenvironment will
11 experience the same concentration of that material.
12 The difference between the reading of a personal monitor and an indoor monitor during the
13 a and P periods represents the PM exposure due to a person's own personal activities. This
14 incremental exposure above the surrounding micro environmental concentration is unique to the
15 subject, save for the case where someone else is immediately next to the monitored person (e.g.,
16 a helper holding parts together to be welded by a monitored welder).
17
18
19 5.6 PERSONAL PM EXPOSURE
20 This is PM exposure that occurs from direct inhalation of tobacco smoke by a smoker.
21 By placing the item (pipe, cigar, cigarette) to the mouth, the smoker inhales the concentrated
22 fumes directly into the lung and completely bypasses the inlet of a personal PM monitor worn by
23 the smoker. This exposure category is distinct from the personal smoking-activity exposure
24 described in the previous section, which can be captured by a personal monitor in the breathing
25 zone of the subject.
26 The magnitude of this source is appreciable and dominates all other categories of exposure
27 when computing the total exposure to PM of a smoker from all sources of ambient and
28 non-ambient PM. The nominal amount of PM (tars) delivered by each brand and type of
29 cigarette smoked are reported by the Federal Trade Commission (1994). The concentration
30 (mg/m3) delivered by each puff is not reported. However, for a person breathing at an average of
31 10 Lpm over a day, or 14.4 m3/day, each 1 mg delivered adds approximately 70//g/m3 to the 24-h
October 1999 5-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 average PM exposure of the smoker. Consequently, a subject smoking 20 cigarettes (one pack)
2 per day, rated by FTC at 10 mg tar each, would have an average daily PM exposure of 14 mg/m.
3 Siegmann et al. (1999) point out a major difference between cigarette smoke particles and
4 other combustion source particles derived from candles, paper, and motor vehicles. "The
5 particles of cigarette smoke are much larger than the other ones, and grow with time to even
6 larger ones" and "particles generated when a cigarette is smoked are known to contain liquid
7 matter which will contract the particles to denser material due to the surface tension of the
8 liquid". The latter effect would also increase the Stokes' Law settling velocity of the particle,
9 which would increase the effective aerodynamic diameter (AD) of the particle.
10
11
12 5.7 EXPOSURE TO PM OF AMBIENT ORIGIN IN BOTH INDOOR
13 AND OUTDOOR MICROENVIRONMENTS
14 Let a subject living in the residence of volume V, at the location where the outdoor PM
15 concentration is Co, spend some fraction of time outdoors (y), and the remaining fraction of time
16 (1 - y) in the residence where Equation 5-4 applies. For simplicity, let us also assume that while
17 outdoors the subject is close to home and exposed to the local outdoor PM concentration Co
18 [Co = Ca + e, where Ca is the ambient monitoring station value ande is a random increment of
19 order zero with a finite variance]. The subject's total exposure to PM of ambient origin (Ea)
20 during a complete day will be as follows:
Ea - y(Ca + e) + (1 - y)[Ca Pa+ e Pa + Qa / V] I (a + k\ (5.5)
/3 (5-6)
where a = y + (1-y) P a I (a + k)
1 Equation 5-6 means that a person's daily exposure to the ambient PM is proportional to the
2 concentration Ca at the monitoring station (a Ca) plus a random variable (p). The variable P has
3 a term with mean zero and finite variance representing the effect of the spatial variation in the
October 1999 5-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 ambient air (e), plus another positive term with a finite variance representing resuspension of that
2 day's settled ambient PM by mechanical activity in volume V (walking on carpets, sitting on
3 stuffed furniture, dusting or vacuuming, etc.).
4 From Equation 5-6 we expect that someone living in a style such that they are not exposed
5 to any appreciable additional occupational PM or indoor generated PM to have a measured
6 personal exposure to total PM (E) that is highly correlated with the ambient PM concentration.
7 However, should someone live in a style in which they are exposed to additional PM generated at
8 work and at home, as from passive cigarette smoke, cooking, fireplace emissions, etc., then this
9 increment to the exposure would appear mathematically as if it were a gross enlargement of the
10 random term p. Because the strengths of these indoor sources are independent of both the
11 ambient concentration Ca and the outdoor concentration Co, the resulting correlation of total
12 personal exposure (to ambient PM plus non-ambient PM) with the ambient PM concentration
13 would be decreased, and approach zero. These are the relationships that have been observed in
14 the PM exposure literature, as discussed in the following section.
15 People are exposed to PM of ambient origin while indoors (C, defined as the concentration
16 of PM of ambient origin in the indoor microenvironment) and outdoors (Co, where Co^ Ca due
17 to possible spatial variation of PM from the monitoring station concentration Ca). Consequently
18 one must include the exposure to ambient PM indoors in any analysis of the total exposure to PM
19 of ambient origin. The analysis below follows the formulation of Mage (1998) and Kulmala
20 etal. (1999).
21 During any given period T, the exposure to PM of ambient origin (Ea) is defined by
22 Equation 5-7.
T T
Ea = 1 / T [ J SoiCo dt + J 8io C dt] (5-7)
o o
23
24 where 8oi = 1 if outdoors and = 0 if indoors; 8io = 1 if indoors and = 0 if outdoors.
25 Because of the current inability to monitor the quantity of ambient PM inside the indoor
26 locations where people spend their time, Equation 5-7 is often modified to estimate Ea by
27 assuming either 8io = 0 or C = Co at all times, resulting in Equation 5-8 as follows:
October 1999 5-30 DRAFT-DO NOT QUOTE OR CITE
-------
1
Ea-1/TjCodt (5-8)
1 Equations 5-7 and 5-8 are both attempts at addressing the question, "How much of the
2 ambient PM that existed in the ambient atmosphere during the previous T = 24-h period are
3 people exposed to during that period?" However, in practice, a calculation complication exists.
4 A portion of the coarse mode ambient PM that enters the home and then is deposited in the home
5 during the interval 0 < t < T is resuspended by human activity during the 24-h measurement
6 period (Kildeso et al., 1999). There is expected only minimal resuspension of fine PM directly,
7 but any fine PM that settled out and adhered to a coarse particle could also be resuspended. It is
8 difficult, if not impossible, to distinguish the resuspended ambient PM from the PM of
9 non-ambient origin that is also resuspended by the same human activities.
10 Figure 5-3 shows the two compartment model that is used for this analysis. The first
1 1 compartment is the indoor air of volume V and the second compartment is the surface area (A)
12 within volume V where ambient PM mass is deposited. The other parameters are as defined
13 previously. The PTEAM study (Ozkaynak et al., 1996a,b) reports widely variable values olfr for
14 night and day conditions which may be influenced by diurnal patterns of resuspension. The
15 fraction of the mass (M) of freshly deposited ambient PM on surfaced that is resuspended per
16 unit time is a function of human activity, such as walking on a surface and creating vibrations
17 and air currents sufficient to levitate a particle. The differential equations that describe the daily
1 8 accumulation of ambient PM on interior surfaces (M) and the time variation of ambient PM in
19 volume V are as follows:
20
VdC/dt = PvCo-vC-kVC + 7 M (5-9)
21
dM/dt = kVC - 7 M (5-10)
22
23 where y Mis the resuspension rate (mass/time) of that day's previously deposited ambient PM
24 that is assumed to be proportional to the mass $f) deposited on the surface, y ^0.
October 1 999 5-3 1 DRAFT-DO NOT QUOTE OR CITE
-------
1 The solutions to equations 5-9 and 5-10 are coupled, and C and M must be solved for
2 numerically using the randomly varying values of Co, P, k, v, and y due to weather and human
3 activity in the indoor location. The initial conditions for the integration are as follows:
4 M = M(0)@t = 0;C = C(0)att = 0.
5 At the start of the integration at t = 0, the ambient PM inside an indoor microenvironment
6 (C) is some unknown fraction of the outdoor PM that developed from air exchange over the
7 previous 24-hrs; if some of that material may already be deposited on the indoor surfaces then
8 M(0) > 0. In the following analysis, the mass of ambient PM on the surfaces that had deposited
9 from the previous weeks' depositions (since the last house cleaning) is assumed to be
10 uncorrelated with the concentration of ambient PM on the measurement day so it does not enter
11 into the equation.
12 Except for the trivial conditions where y is equal to zero, the solutions to coupled
13 Equations 5-9 and 5-10 are beyond the scope of this chapter, because several parameters and
14 initial conditions are unknown. For example, y is close to zero when the indoor occupants are all
15 asleep and/or sedentary; it is larger while the occupants are moving about; it is maximal while an
16 occupant is dusting, sweeping or vacuuming. The value of the air exchange parameter v varies
17 with the wind speed and varying window and door openings, and k varies with the size
18 distribution of the mixture of PM in the ambient air that penetrates into the home. Consequently,
19 the approach chosen here is not to solve them simultaneously because of the unknown
20 parameters. Rather, Equation 5-9 is solved with y = 0, and the result is reported as an inequality
21 because the exposure to PM of ambient origin with resuspension (y > 0) must be greater than the
22 exposure without resuspension (y = 0).
23 Table 5-1 summarizes some of the parameters necessary for creating a solution to
24 Equations 5-7 and 5-9. The air exchange rate (a) is the ratio v I V, and it was measured in the
25 PTEAM study by collection of a continually emitted tracer gas inside the subject homes. These
26 data for a contain two important artifacts that lead to a negative bias (an underestimation) of the
27 average value of a. First: When the air exchange rate was too high, the collected tracer was
28 below the minimum detectable level (MDL) of the analytical procedure. Ozkaynak et al.,
29 (1996a,b) used the MDL values of a as alternative default values for computing k and P in order
30 to maximize the amount of data available for the analysis using a non-linear optimization
31 procedure (Koutrakis et al., 1992). Second: The concentration of tracer gas, and therefore the
October 1999 5-32 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-1. SUMMARY OF THE MEAN VALUES OF
PM VARIABLES FROM THE PTEAM STUDY
Parameter
& (I/ hour)
a* (I/ hour)
J3**
Co (//g/m3)
Qother(mg/hr)***
PM25
7am-7pm
Day
0.27
1.144
1
48.9
1.46
PM25
7pm-7am
Night
0.39
0.98
0.89
50.5
0.784
PM25
7am-7am
Combined
0.39
0.97
1
49.7
1.08
PM10
7am- 7pm
Day
0.91
1.144
1
94.9
14.3
PM10
7pm-7am
Night
0.43
0.98
0.88
86.3
2.82
PM10
7am-7am
Combined
0.65
0.97
1
90.6
5.64
Source of Data: Ozkaynak et al. (1996a).
* Values of air exchange above the maximum level of detection (LOD) were assumed equal to the LOD.
**P was constrained to the range 0
-------
1 locations, or for all homes in a given location. For example, where it is hot during the day
2 and cool at night, people with an HVAC system may keep windows closed during the day
3 and open them at night to cool off.
4 2. Indoor emissions (including resuspended ambient PM) are much greater for coarse PM
5 (PM10 - PM2 5) than for PM2 5, which is consistent with the literature findings that fine PM is
6 held more tightly to surfaces than coarse PM. Resuspension of previously settled PM is
7 expected to be greatest during the day when people are most active (Roorda-Knape et al.,
8 1998). For example, Larssen et al. (1993) report a study of three days in an unoccupied
9 apartment in Oslo on a busy street, showing that the indoor fraction of ambient coarse PM in
10 their study is of order 0.2. In the PTEAM study the PM collected during the second 12-h day
11 period contained some of the PM deposited during the first 12-h night period that was
12 resuspended by daytime activity. This increased the amount of the nighttime PM that the
13 subject was exposed to during the 24-h period, but the analysis treated it as a component of
14 Qother-
15 3. There is an opposite and counter-intuitive relation with the& values. The computed k for
16 PM25 is higher at night than during the day (0.39/hour vs 0.27/hour), which is expected since
17 turbulence (which provides upward velocity components to counter Stoke's Law settling) is
18 less when homes are closed and people are resting and sleeping. Turbulence reduces
19 gravitational settling but also decreases the boundary layer thickness through which PM must
20 diffuse to reach surfaces for deposition which may increase deposition rates for the ultra-fine
21 PM (< 1 //m AD). For PM10 the opposite variation of k is observed. The PM,0 k value
22 decreases from a day time value of 0.91/hour to a nighttime value of 0.43/hour which is
23 counter to the PM25 behavior, and also counter to the deposition increase from day to night
24 during passage of PM10 into the indoor environments. There are several possible
25 explanations for these phenomena:
26 a) There are experimental errors in all measured values, such as air exchange rate fy) and
27 concentrations (Co and C). Such experimental errors can inadvertently cause artifacts to
28 appear like values ofP > 1 as cited earlier (Ozkaynak et al., 1996a). The air exchange rate
29 data set {a} contains two artifacts as previously discussed. The minimum detectable level
30 (MDL) of perfluorocarbon tracer (PFT) mass collected corresponds to a maximum detectable
31 air exchange rate because an increase in air exchange rate lowers the amount of tracer
October 1999 5-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 collected. The
-------
1 that indoor human activity at night is usually much less than daytime activity. In the absence
2 of the influence of human activity the indoor value ofk only has a constant expectation if the
3 relative size distribution, the proportion of PMj5 to (PM10 - PM25), is constant. In PTEAM
4 the day time PM2 5 and (PM10 - PM2 5) were approximately equal and at night there was
5 approximately 50% more PM25 than (PM10 - PM25) (Ozkaynak et al, 1996a).
6 e) Penetration (P) and decay rate (k) may both depend on air exchange rate (a) and, thus, may be
7 correlated to each other. This may lead to unstable or inaccurate analyses.
8 f) The results are real and there is some mechanism operating that is not completely understood
9 causing the opposite behavior of the k values for fine PM (PM25) and coarse PM
10 (PM10-PM25).
11 An alternative procedure to estimate mean values, and to conserve the property that mean
12 24-h values must be approximately intermediate between their constituent 12-h day and 12-h
13 night values, would be to use only the complete data set that constitutes a nominal 24-h average
14 by discarding all daytime sets without a matching nighttime set andvice versa. Then the 24-h
15 mean value of indoor and outdoor concentrations could be used with the 24-h mean value of air
16 exchange and reported in two ways: 1) from averaging the mass of PFT collected using only
17 those data where it was greater than the MDL for both night and day; 2) from averaging the mass
18 of PFT collected using all data, but giving lower weights to the default values ofa computed
19 with the arbitrarily set MDL PFT values. The unknowns in the equation would be the
20 corresponding values off, k and emission strengths (e.g., Q,ther) averaged over 24-hrs. Such
21 procedures should then give mean values ofP, k, and emission strengths approximately
22 intermediate to their day and night values.
23 Abt et al. (1999b) also model their PM size and number density data (Abt et al., 1999a)
24 from four homes in the Boston area using the same Koutrakis et al. (1992) model described
25 above, as used in the PTEAM study. The authors define the "effective penetration efficiency of
26 outdoor air" for PM of various size ranges by the termP a /(a + k) which represents the fraction
27 of the outdoor PM found indoors at equilibrium. They report effective penetration efficiencies
28 ranged from 0.38 to 0.94 for 0.02 - 0.5//m particles with a maximum between 0.1 and 0.2
29 microns.
30 For 0.7 - 10 //m particles the efficiency values ranged from 0.53 to 0.12, decreasing with
31 increasing particle size. Whereas the PTEAM study (Ozkaynak et al., 1996a) only reported mean
October 1999 5-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 values of penetration (P) and deposition (k) values for their 178 subjects, these Abt et al. (1999b)
2 data indicate that "estimated decay rates varied considerably both within and between homes,
3 with variability attributed to factors including differences in air flow rates, house volumes and
4 surface materials." This suggests that the variances of the distributions off* a / (a + K) shown in
5 Figure 5-4a and 5-4b, are underestimated because they were derived using the two PTEAM mean
6 values of k as a constant.
7
8 5.7.1 Estimation of the Daily Exposure to PM of Ambient Origin
9 Let each person spend a fraction x of their time outdoors during the day and a fraction w
10 outdoors during the nighttime periods during which PM is monitored [y~ (x + w)/2]. While
1 1 outdoors, people are exposed to 100% of the ambient concentration and while indoors, they are
12 exposed to a lower fraction of the ambient PM as estimated by setting the derivative equal to zero
13 in Equation 5-2 that resulted in Equation 5-3.
14 The resulting equation for the fraction (z) of the daily ambient PM one is exposed to,
15 assuming 12-h daytime and 12-h nighttime sampling, is as follows:
16
_ [x + (1 - x)/3 a I (a + k)](Co)day + [w + (1 - w)P a I (a + k)](Co) night + F(M)
Z~ (Co)day + (Co)night (5"n)
17
18 where the function of deposited mass of ambient PM that is resuspended [F^/)] is > 0. Because
19 of the difficulty in computingMby solving Equation 5-1 1, as discussed previously, the
20 assumption that [F(M)J is equal to zero for all time t > 0 allows a rewriting of Equation 5-11 as
21 an inequality:
22
Z ~ (Co) day + (Co) night (5"12)
23
24 The estimation of z can be made using parameters from the PTEAM data set shown in
25 Table 5-1 and the fractions of time spent outdoors (x and w) shown in Table 5-2. In this analysis
October 1999 5-37 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-2. AVERAGE FRACTIONS OF TIME SPENT OUTDOORS OR IN A
VEHICLE IN THE PTEAM(DAY) STUDY AND NHAPS STUDY (NIGHT) [PTEAM
NIGHTTIME DATA NOT AVAILABLE]
Location Outdoors Daytime Fraction (8am-8pm) Nighttime Fraction (8pm-8am)
Outdoors at home location 0.040 (See below)
Outdoors at other location 0.087 0.020 (Both locations)
Outdoors in vehicle 0.092 0.032
Total Outdoor Fraction x= 0.219 (PTEAM) w = 0.052 (NHAPS)
* Sources of data: Ozkaynak et al. (1996a) and Klepeis et al. (1999).
1 it is assumed that the ambient PM (away from traffic) is relatively uniform throughout the
2 community. The air exchange rate while driving a closed vehicle is very large (> 13/hour at
3 20 mph) so the subject in a vehicle would be exposed to-100% of the ambient PM measured at
4 the central site plus the PM generated by the surrounding traffic (Ott et al., 1992; Park et al.,
5 1998).
6 A recent study (California Environmental Protection Agency, 1998) reports that PlVf 5 in
7 motor vehicles is intermediate between the roadside concentration and the concentration
8 measured immediately outside the vehicle. The inside/outside ratio for the vehicles in
9 commuting traffic was approximately 2/3. Note that the locally generated PM from the traffic
10 surrounding the vehicle does not directly influence the monitoring station measurement (Ca) or
11 the outside air at the home (Co). Aim et al. (1999) monitored the particle count (by Climet-500
12 laser particle counter) in a commute vehicle in Kuopio, Finland and compared the results with
13 the background values, estimated as the mean count at the start and finish of the trip at an
14 off-road location. The authors found that the excess (vehicle - background) counts of fine PM
15 (optical equivalent diameter < 1 micron OD) increased as the average vehicle and wind speeds
16 decreased, and that the excess counts of coarser PM (> 1 micron OD) increased as wind speed
17 and vehicle speed increased.
18 Substituting the corresponding parameters from Tables 5-1 and 5-2 in Equation 5-12, the
19 mean values for the daily total fraction of the ambient PM that people are exposed to (z) are
20 estimated to be z >0.75 for PM2 5 and z > 0.64 for PM10. The bounds for the PTEAM study can
October 1999 5-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 be estimated from the daytime minimum (0) and maximum (0.976) values of x reported in
2 PTEAM, and assuming that the nighttime minimum exposed subject is indoors (w = 0) and the
3 maximum exposed subject at night is outdoors or in a vehicle for 4 of the 12 hours (w = 0.33).
4 The estimates of the population range then becomes as follows:
5 For PM25: z maximum >0.88; z mean >0.75; z minimum >0.72
6 For PM10: z maximum >0.87; z mean >0.64; z minimum >0.58
7 The mean values, the caveats cited above not withstanding, are consistent with the PTEAM
8 findings that approximately 3/4 of the indoor PM, 5 and 2/3 of the indoor PM10 were of ambient
9 origin (Ozkaynak et al., 1996a). The minimum and maximum values are conservative because
10 identical daytime and nighttime values of Cout,a, k and P are implicitly assumed for all homes
11 in the study. A Monte Carlo analysis, allowing all parameters to vary randomly about their mean
12 values, would increase the estimated variance of z so the maximum would increase and the
13 minimum decrease.
14 Abbey et al. (1999) evaluated the effect of time spent indoors as a surrogate for exposure to
15 PM from indoor sources. They compared the relative risk (RR) of mortality by nonmalignant
16 respiratory disease with ambient PM;0 concentration, for different amounts of time spent
17 outdoors, in the 7th Day Adventist Health Study on Smog (ASHMOG). They assumed that the
18 indoor concentration of PM10 of ambient origin was 70% of the outdoor PM,0 concentration to
19 allow for a protective effect of staying indoors. They found that RR for a 50//g/m3 increase of
20 ambient PM10 increased when the number of hours spent outdoors increased (and the number of
21 hours indoors decreased) during the week, as shown in Table 5-3. Thus, the authors reasoned
22 that increased exposure to indoor generated PM and decreased exposure to PM of ambient origin
23 did not appear to be a confounder in their analysis of the effects of ambient PM. This increasing
24 RR trend with time outdoors is consistent with a higher exposure to particles of ambient origin
25 with more time outdoors.
26 Further in regard to time spent outdoors, Abbey et al. (1999) explain their observed gender
27 difference in health effects in which males appear to be more affected by ambient PM than
28 females, as being related to the males' greater percentage of time outdoors than females of
29 similar ages. However, Brunekreef (1999) points out that this is not a likely explanation: "As the
30 authors (Abbey et al., 1999) note, fine particles readily penetrate indoors, and if it is the fine
31 particles that matter, small differences in time spent outdoors cannot matter all that much."
October 1999 5-39 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-3. RELATIVE RISK (RR) OF NONMALIGNANT RESPIRATORY
MORTALITY FROM INCREASING TIME SPENT OUTDOORS
Hours outdoors per week RR per increase of 50 //g/m3 PM10 95% Confidence Interval
t<4 1.07 0.85-1.34
416 1.32 1.02-1.68
Source: Abbey et al. (1999).
1 The presence of appreciable amounts of PM generated indoors would add to the mass of the
2 ambient PM collected by a personal PM monitor and cause the high correlations of personal
3 exposure to PM of ambient origin with ambient PM concentrations to degrade and approach zero
4 (U.S. Environmental Protection Agency, 1996; Monn et al., 1997; Monn and Junker, 1999). For
5 example, Monn et al. (1997), (with clarification by personal communication [Monn, 1999]),
6 reported that human activity increased the median Indoor/Outdoor ratio by 20% for PM5 and
7 50% for PM10. As a result, some authors (e.g., Gamble, 1998) have misinterpreted an absence of
8 a significant correlation of ambient PM concentration with personal exposure tototal PM as
9 implying an absence of a significant correlation between ambient PM concentration and personal
10 exposure to PM of ambient origin. [See Kiinzli and Tager (1999) comments on Gamble (1998),
11 and the response by Gamble (1999).]
12 All people in a community, when outdoors, are exposed to a heterogeneous mixture of
13 ambient PM2 5 with small variations of composition and concentration from the ambient PlVf 5
14 measured at a central location. Therefore, no great ecological fallacy is produced and no large
15 Berkson-type error is involved, in the use of community ambient PM.5 concentration as a
16 surrogate for exposure to particles of ambient origin. This is because the mean exposure of a
17 random sample of people in a community to PM, 5 of ambient origin, while both outdoors and
18 indoors (via PM25 infiltration from outdoors), is an excellent predictor for the simultaneous
19 exposure to PM25 of ambient origin of virtually everyone in that community - exceptions for
20 unusual circumstances such as home use of high performance air cleaners not withstanding.
21 In stark contrast, the mean of the personal exposure to PM,5 or PM10 of non-ambient origin
October 1999 5-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 (from residential and occupational emissions) of the same randomly chosen subjects would have
2 virtually zero predictive power for estimating the exposure to PM^ or PM10 of non-ambient
3 origin of any person in the community not residing in the home of a sampled person.
4 Exposure to an arbitrary PM fraction (PM,,) from non-ambient sources is not of relevance to
5 the question of personal exposure to PM,, of ambient origin, and an ambient PM^, concentration is
6 not expected to be a surrogate for any arbitrary person's total exposure to P]V| as measured by a
7 personal PMX monitor that collects PIV^ of both ambient and non-ambient origin. In the absence
8 of appreciable indoor sources of PM^, a personal exposure to total PM^ while indoors is primarily
9 to the PMX of ambient origin that infiltrates indoors. Under such conditions a measurement of
10 personal exposures to PIV^ that is mostly of ambient origin is expected to be highly correlated
11 with ambient PMX concentration as described in later sections.
12 People in a community are routinely exposed to widely different mixtures of PM of
13 non-ambient origin with a large variance of composition and concentration. This wide variance
14 arises from each individual's unique combination of occupation, personal habits (e.g., cigarette
15 smoking), social contacts (e.g., living with a smoker) and non-occupational/residential activities
16 (e.g., cooking, cleaning and dusting). A subject's exposure to PM of ambient origin and its
17 chemical composition can be estimated from knowledge of the ambient PM concentration, its
18 composition, the air exchange rate between ambient and indoor locations where the subject
19 spends time, the deposition rate of the PM and the subject's time-activity patterns. In contrast,
20 a subject's exposure to PM of non-ambient origin and its composition cannot be predicted
21 accurately from measurements of other peoples' exposures to non-ambient PM concentration and
22 its composition. Knowing the distributions of ambient PM concentrations (uninfluenced by an
23 immediate source such as a barbecue) in all other backyardsdoes allow the accurate prediction of
24 the ambient PM concentration in the subject's own backyard; knowing the distribution of
25 non-ambient PM concentrations in all other homes does not allow the accurate prediction of the
26 non-ambient PM concentration in the subject's own home.
27 When people are members of a cohort whose individual health outcomes are hypothesized
28 as due to ambient PM exposure (and other possibly co-occurring), and they are being tracked
29 from day to day, it is important to determine each individual's daily total personal exposure to
30 PM of ambient origin (Beyea and Hatch, 1999). However, additional information is needed for
31 such epidemiologic studies to be able to separate the total PM exposure, if measured, into its two
October 1999 5-41 DRAFT-DO NOT QUOTE OR CITE
-------
1 general components of exposure to ambient PM and the exposure to non-ambient PM (e.g., from
2 residential or occupational sources). With this information it will be possible to improve the
3 estimation of the health outcomes related to ambient PM exposure and separate them from those
4 related to non-ambient PM exposure.
5 This chapter examines ambient PM air quality and that portion of ambient PM which
6 penetrates into indoor microenvironments. It also examines, to a lesser extent, the contribution
7 of sources of non-ambient PM to total PM exposure. This is to aid in interpretation of acute and
8 chronic epidemiology studies assessed in Chapter 6, in which ambient PM concentrations are
9 assumed to be an indicator or a surrogate for mean community exposure to PM of ambient origin
10 or an individual's exposure to ambient PM. Thus, this chapter has three objectives:
11 (a) Provide a review of pertinent studies of personal exposures to total PM of various size
12 fractions.
13 (b) Evaluate linkages of human exposures to PM of ambient origin estimated from
14 concentrations of PM measured at a fixed-site monitor located at some central site in a
15 community under study.
16 (c) Quantify the contributions of PM of ambient origin and non-ambient origin to total personal
17 PM exposure.
18 At the present time, little published data are available for time scales of less than 12-h to
19 compare short-term personal PM exposures, short-term peak ambient PM concentrations and
20 short-term indoor PM concentrations found in different locations within a community, all for the
21 same time interval. It has been hypothesized (Michaels, 1997, 1998) that some health effects
22 may be better correlated with short-term 1-h peak PM concentrations than the 12-h or 24-h
23 concentration averages containing the peak value. This has been supported by Delfino et al.
24 (1998) with a finding of "1-h and 8-h maximum PM,0 having larger effects than the 24-h mean"
25 in relating asthma symptoms. Hourly PM data and even 1/2-h PM data (Keary et al., 1998)
26 obtained by use of the TEOM® sampler (see Chapter 4 and Ayers et al. [1999] for a description
27 of loss of semivolatile components) are beginning to become available in the literature. Such
28 data may allow for more study of health in relation to the lagged effect of peak ambient PM
29 concentrations.
30 The PM-mortality literature, to date, is based upon the general linearized model (GLM)
31 assumption that there is a virtually linear relationship between health effects and ambient PM
October 1999 5-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposure characterized by ambient PM concentration in the concentration range below the
2 existing PM10 NAAQS (Liang and Zeger, 1986). The relative risk has generally been modeled as
3 a proportional or percentage increase in non-trauma mortality per unit increase in ambient PM
4 concentration, such as 0.1% per 1 //g/m3 of PM2 5. For a strictly linear relation with PM of
5 ambient origin, going from 10//g/m3 of PM25 to 110 //g/m3 of PM25 would increase the base
6 mortality rate by 10%. However, the actual 'compounding interest' effect of an increased rate of
7 0.1% with each additional 1 //g/m3 of PM2 5 would only increase the rate by (1.001}°° = 10.5%
8 which closely approximates the linear increase. In the realistic range of ambient PM
9 concentrations, the linear approximation is generally acceptable.
10 By the principal of superposition for a linear system, the health effect from exposure to a
11 mixture of ambient PM and non-ambient PM is the sum of the health effect of the ambient PM
12 exposure plus the health effect of the non-ambient PM exposure. This assumed linear-relation
13 implies that the health effects of the PM of ambient origin are independent of the health effects of
14 the PM of non-ambient origin in the current range of ambient and non-ambient PM
15 concentrations in modern society. This is complementary with the finding that sources of indoor
16 PM appear to operate independently of the ambient PM concentration.
17 Given this framework of linear analysis, if PM of non-ambient origin produced health
18 effects similar to those produced by ambient PM, the fluctuations in the health effects of PM
19 from non-ambient sources would be independent of the fluctuations in the health effects
20 produced by PM of ambient origin - and therefore would not act as a confounder. Rather, they
21 would appear as a source of random error in the linear epidemiologic analyses of human health
22 and exposure to PM of ambient origin. Until nonlinear effects of PM exposures are
23 demonstrated in clinical and epidemiologic studies, the sources of non-ambient PM, and the
24 effects they produce on total personal exposure to PM, are treated as errors in the estimate of
25 human exposure to PM of ambient origin. Thus PM of non-ambient origin (e.g., from personal
26 activity and occupational or residential sources) is considered, at the present time, to play a minor
27 role, if any, in the study of human health and its relationship to acute exposure to PM of ambient
28 origin in community time-series studies.
29
30
October 1999 5-43 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.8 AMBIENT PARTICULATE MATTER CONCENTRATION AS A
2 SURROGATE FOR EXPOSURE TO PARTICULATE MATTER OF
3 AMBIENT ORIGIN
4 The health effects due to ambient PM may depend upon an individual's genetic makeup,
5 lung anatomy, and previous health history, and the mass, size and composition of those inhaled
6 particles that can be deposited within various regions of the respiratory tract. The amount of this
7 dose per unit lung area (mg/cnf-day) will depend on the concentration of ambient PM inhaled
8 (e.g., the instantaneous personal exposure to ambient PM); the ventilation rate and respiration
9 frequency (a function of physical activity and basal metabolism); and the fractional PM
10 deposition, which is a function of ventilation rate and respiration frequency, mode of breathing
11 (e.g., oral or nasal), and any alteration of normal pulmonary flows due to lung dysfunction. If all
12 people had identical ventilation rates (L/min - kg body weight), respiration frequency, and
13 deposition patterns, then the potential-dosage distribution (mg/crri-day) could be linearly scaled
14 to the personal ambient PM exposure distribution which would serve as a suitable primary
15 surrogate. The usage of ambient PM concentration in health studies as a surrogate for personal
16 exposure to PM of ambient origin, and thereby a secondary surrogate for the ambient PM dosage,
17 would be suitable if ambient PM concentration is linearly related to the personal exposure to PM
18 of ambient origin and the dose-response relationship is linear (Mage, 1983).
19 Adult ventilation rates are lowest (mean- 6 L/min) during the night while asleep, highest
20 (mean -12 L/min; peak ~ 60 L/min) during the day while awake (Adams, 1993), and in phase
21 with ambient PM concentration, which is also usually lower at night than during the day (Clayton
22 et al., 1993). Consequently, the product of the 24-h average ambient PM concentration, the 24-h
23 average ventilation rate, and the average deposition parameter for the average ventilation could
24 seriously under-predict the amount of ambient PM deposited in the respiratory tract (Mage,
25 1980).
26 In practice, when relating human health to ambient PM pollution variables (as in
27 Chapter 6), one is forced to use time-weighted-average (TWA) ambient PM concentration as a
28 surrogate for ambient PM exposure and ambient PM dosage because typically only fragmentary
29 data are available on personal exposures to PM of ambient origin in populations. Data are also
30 limited on ventilation rates as a function of basal metabolism and physical activities (Adams,
31 1993), and on the pulmonary deposition rates of particles people are inhaling. The size
October 1999 5-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 distribution of PM in the ambient air is usually unknown, and its pulmonary deposition is
2 affected by unmeasured individual physiological parameters.
3 In the sections that follow, the experimentally observed relationships between ambient PM
4 concentration, indoor concentrations of ambient PM and non-ambient PM, total personal
5 exposures to PM, and personal exposures to PM of ambient origin are discussed in detail. The
6 following four caveats, developed in more detail in Chapter 7 of U.S. Environmental Protection
7 Agency (1996), should be kept in mind:
8 1. Ambient PM concentration (times volume inhaled) is a surrogate for the dosage of ambient
9 PM inhaled and deposited in peoples' respiratory tracts.
10 2. The daily dosage of total ambient PM deposited per unit surface area of the sensitive
11 portion(s) of the respiratory tract is in turn a surrogate for the mass of the true (but unknown)
12 species and/or size fraction of the ambient PM that is the specific aetiologic toxic agent(s)
13 that act by a presently unknown mechanism.
14 3. Virtually all analyses and discussions of personal PM exposure presented here are based on
15 personal exposure to PM of self-declared non-smokers.
16 4. A total TWA personal exposure to total PM (ambient PM plus non-ambient PM) is expected
17 to be a poor surrogate for the personal exposure to PM of ambient origin. This will be
18 particularly true for those people whose personal exposures to total PM are dominated by
19 residential and occupational indoor sources, or personal sources such as a hobby activity or
20 active smoking of tobacco.
21
22
23 5.9 CONCENTRATIONS OF AMBIENT PM FOUND INDOORS AND IN
24 OTHER NON-AMBIENT ENVIRONMENTS
25 In the absence of appreciable indoor sources, there is an excellent correlation between
26 ambient PM concentration (ambient PM is virtually all of ambient origin) and the concentration
27 of PM found indoors. Figure 5-5 (Tamura et al., 1996a; U.S. Environmental Protection Agency,
28 1996) shows how the indoor PM,0 correlated with the outdoor PM,0 for a set of seven elderly
29 non-smoke exposed individuals living in traditional Japanese homes. These Tokyo homes
30 (Itabashi ward), where people routinely took their shoes off prior to entering, hadtatami reed mat
31 or carpeting on tatami or wooden flooring, and had gas for cooking. The study was designed to
October 1999 5-45 DRAFT-DO NOT QUOTE OR CITE
-------
110-
100-
E 80-
'§> 70-
tT 60-
8 50-
-o 40-
- 30-
20-
10-
E
m a
DO™
ffl
D On
mm
ffl° a ffl
Dft1^1 ffl Annual r = 0.922
™ffl Winter r = 0.920
Summer r = 0.961
40
80
120
160
200
110-
100-
~ 90-
-S 80-
Cf) ^^
3; 70~
ri 60 —
o 50-
E 40-
30-
20-
10-
F
D Q
r*
B BO • °
n DB B
j^j1 Annual r = 0.897
If Winter r = 0.702
Summer r = 0.970
i i i i i i i i i
40
80
120
160
200
110-
100-
"E so-
"ro 70-
— 60-
o 50-
o
? "-
— 30-
20-
10-
G
a
ffl m
D a
m OB D
gm
,rip |p ffl Annual r = 0.898
0 nyffl Winter r = 0.879
Summer r = 0.919
160
200
Outdoor (M9/m
120-
110-
100-
90-
r so-
5 60-
8 40-
_E 30-
20-
10-
0
Annual r = 0.983
Winter r = 0.980
Summer r = 0.982
40
120
160
40
80
120
160
40
80
120
160
Outdoor (jjg/m3
200
110-
100-
JT 90~
"°E 80-
D) 70-
•^- cn
b(J—
o 50-
E 40-
- 30-
20-
10-
B
B ffl
0° ^aSa°
a gg& Annual r= 0.929
^nSpB" Wnter r= 0.966
Summer r = 0.877
110-
100-
r~>~^
E 80-
ra 70-
•^ 60-
0 50-
? 40-
- 30-
20-
10-
0-
C
ffl
Qa_BBa
Cffl *
D qgff^ Annual r= 0.970
a Wnter r = 0.968
Summer r = 0.964
i i i i i i i i i
200
110-
100-
pT~ 90~
E 80-
ra 70-
— 60—
0 5Q_
o
-0 40-
- 30-
20-
10-
D
ffl ffl
m ffl
ffl
ffl ffl D
en a
tSb
_ ^§Q ffl
gji* Annual r= 0.838
D D Wnter r= 0.775
Summer r = 0.978
I I I I I I I I I
200
Figure 5-5. Individual indoor versus outdoor relationships of PM10 in Tokyo for the seven
subjects (A-G) reported on by Tamura et al. (1996a). El Winter, D Summer
1 monitor the exposures of people to ambient PM so the subjects were purposefully chosen in a
2 non-random manner to eliminate indoor combustion sources of PM. Consequently, these results
3 apply strictly to these seven people and cannot be used to infer a similar relationship in other
October 1999
5-46
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
Tokyo homes. Table 5-4 provides the correlation coefficients reported and the designations
A - G are the identifiers of the subjects with the indoor/outdoor relationships shown in
Figure 5-5.
TABLE 5-4. SUMMARY OF CORRELATIONS BETWEEN PM in PERSONAL
10
EXPOSURES OF SEVEN TOKYO RESIDENTS AND THE PM in MEASURED
10
OUTDOORS UNDER THE EAVES OF THEIR HOMES, AND THE PM MEASURED
AT THE ITABASHI MONITORING STATION
Subject ID
A
B
C
D
E
F
G
A-G
Number of Samples
48-h PM10
9
9
11
9
10
7
9
64
Correlation between Correlation between Personal
Personal and Outdoor at and Itabashi Station (r)
home (r)
0.958
0.874
0.846
0.922
0.960
0.776
0.961
0.834
0.876
0.747
0.848
0.964
0.925
0.801
0.952
0.830
Source: Tamura et al. (1996a).
1 Tamura et al. (1996a) did not report duplicate measurements of the indoor/outdoor
2 monitors so it is not possible to correct these data for the variance component due to
3 experimental errors of filter weighing and flowrate measurement. Such errors prevent two
4 collocated measurements of PM from approaching a perfect correlation of r = 1, and are expected
5 to decrease correlations, such as those shown on Figure 5-5. These data show that in the absence
6 of major sources of indoor PM,0 generation there is a high correlation relationship between
7 indoor and outdoor PM10.
8 Tamura et al. (1996b) performed another study in Osaka, Japan similar in design to
9 their Tokyo study (Tamura et al., 1996a). The authors measured indoor and outdoor PM
10 simultaneously at 26 homes (not chosen randomly) during the autumn seasons from 1990
October 1999
5-47
DRAFT-DO NOT QUOTE OR CITE
-------
1 through 1995. A dichotomous sampler was used with cut points at 2 and 10//m AD, and a
2 greased impactor plate to collect PM > 10//m AD. There were 146 days when indoor PM and
3 outdoor PM concentrations were successfully collected simultaneously in all three size ranges.
4 The authors pooled all 146 pairs of observations and reported the group correlation coefficient
5 between PM10 concentration measured immediately outdoors at the eaves of the home and the
6 concentration of PM10 in the home. The regression line had a slope of 0.645, an intercept of
7 9.5 //g/m3, and a correlation coefficient r = 0.865. It is interesting to note that when the authors
8 pooled the data for only the 24 homes where 4 or more samples were collected (total number of
9 data points not given) the correlation of r = 0.958 was almost identical to the summer correlation
10 (Tamura et al., 1996a) of r = 0.950. The authors concluded "In the present study [Tamura et al.,
11 1996b], similar relationships [to Tamura et al., 1996a] were confirmed with 18 houses even
12 [though] the measurement period was limited in season and number of times [sampled].
13 Anuszewski et al. (1998) measured simultaneous indoor and outdoor hourly PM by light
14 scattering using a portable nephelometer for 18 days at nine homes of non-smokers in Seattle,
15 WA. Although light scattering does not provide a quantitative measure of concentration, the
16 light scattering occurs from particles of optical diameters (not aerodynamic diameters)
17 approximately centered about the wave length of the light source of the instrument. Therefore,
18 the particles that scatter light using this instrument are primarily in the accumulation mode
19 (0.2 < AD < 1 //m). An example was given in the article of how sweeping a patio (which raises
20 primarily coarse PM) next to a kitchen only caused 25% of the increase in light scattering
21 measured in the kitchen created by smoking a cigarette on the same patio.
22 The authors report that the mean indoor to outdoor ratio of hourly light scattering values
23 was 0.98, and that correlations of indoor with outdoor were in the range 0.58 < Pv < 0.99. The
24 minimum of 0.58 occurred in a home with an electrostatic precipitator (Figure 5-6) and the
25 maximum of 0.99 occurred in a home with a standard in-line filter in the heating system
26 (Figure 5-7). Tung et al. (1999) discuss the PM removal by a standard in-line air filter in a
27 heating-ventilation-air conditioning (HVAC) system used in their study. "In the filtering process,
28 particles with diameters less than 0.1 //m were removed by diffusion removal mechanism.
29 Particles with the size greater than 1 //m were removed by interception and impaction."
30 Therefore, operation of the in-line filter in the house B-l may not have made any appreciable
31 impact on the indoor PM concentration in the range of PM optical diameters (0.2//m < AD < 1)
October 1999 5-48 DRAFT-DO NOT QUOTE OR CITE
-------
0.7 -I
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0
House B-2
R2 = 0.58
Slope = 0.92
Intercept = 0.003
0
0.1
0.2
0.3
Out
0.4
0.5
0.6
Figure 5-6. Comparison of light scattering coefficient (km J) by PM indoors and outdoors
at a home with an electrostatic precipitator in operation.
Source: Anuszewski et al. (1998).
1 measured by the nephelometer. The absence of sources of PM in this size range in the home B-1
2 would then explain the almost perfect correlation between the indoor and outdoor PM shown in
3 Figure 5-7. They report no source of PM in House B-2 that can explain the higher values indoors
4 compared with outdoors as opposed to House B-l where indoor and outdoor were virtually equal.
5 House B-2 had "Some" activities in the studied room and "Medium" in/out traffic, whereas
6 House B-l had "Many" activities in the studied room and "Light" in/out traffic.
7 The average air exchange rates in both houses, B-l and B-2, were equal (1.7/hr), which is
8 much higher than the average in the U.S. reported by Murray and Burmaster (1995).
9 The electrostatic charging of PM in the size range monitored by the nephelometer in House
10 B-2, and that PM subsequent precipitation, could explain a portion of the decrease in the
11 indoor/outdoor PM ratio shown in Figure 5-6. The two homes with no in-line filters in the
12 heating system had the highest minimum slopes [1.0, 0.95] and the six homes with in-line filters
13 had intermediate minimum slopes [0.72 - 0.90]. The authors concluded that "For these nine
October 1999
5-49
DRAFT-DO NOT QUOTE OR CITE
-------
0.6 -I
0.5 -
0.4 -
= 0.3 -
0.2 -
0.1 -
0
House B-1
0.1
0.2
0.3
Out
FT = 0.99
Slope = 0.87
Intercept = 0.004
0.4
0.5
0.6
Figure 5-7. Comparison of light scattering coefficients (km J) by PM indoor and outdoors
at a home with a standard in-line filter in the air recirculating system.
Source: Anuszewski et al. (1998).
1 homes, the dominant source of indoor fine particles, as indicated by [light scattering coefficient],
2 appears to be the outside air."
3 Leaderer et al. (1999a) reported on an indoor/outdoor study of PIV(0, PM25 and related
4 species (sulfate, nitrate, H+, NH4+) at 58 residences in southwest Virginia, measured during the
5 summer periods from 1995 to 1997. The residents were all non-smoker families participating in
6 an epidemiologic study of maternal and infant health with respect to indoor air pollution
7 (Leaderer et al., 1999b). Forty nine of the 58 homes were air conditioned (A/C) and 21 of the
8 homes used gas for cooking.3
9 Table 5-5 summarizes these summer data collected in the area surrounding a central
10 monitoring site located at Vinton, VA (6 km east of Roanoke, VA). Because of happenstance,
3 The residence locations were not chosen randomly to represent all residences in the area surrounding
Vinton, VA so the standard statistical tests used by the authors may not apply to the sampled mean as representing
the mean at any other locations in the southwest Virginia region.
October 1999
5-50
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-5. SUMMARY OF PM DATA DURING THE SUMMER IN AN
INDOOR/OUTDOOR STUDY IN SOUTHWEST VIRGINIA
Pollutant
PM10 Cug/m3)
PM2 5 (,ug/m3)
Coarse Mode (//g/m3)
Sulfate (SO4=, nmol/m3)
Acidity (H+, nmol/m3)
Ammonium (NH4+, nmol/m3)
Nitrate (NO3", nmol/m3)
Sampling Site
Regional
Outside all homes
Inside A/C* homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
n
47
43
49
8
50
43
49
9
47
42
48
8
45
42
47
9
47
45
49
9
43
45
49
9
42
42
49
9
Mean (Std Dev)
26.0(11.5)
28.0(17.7)
28.9(18.7)
33.3 (14.2)
20.2 ( 9.9)
21.8(14.8)
18.7(13.2)
21.1 ( 7.5)
6.3 ( 2.7)
7.7 ( 6.2)
10.4 ( 8.5)
11.4( 9.7)
88.4(51.6)
83.7(53.7)
47.8 (36.3)
63.0 (37.3)
41.0(28.5)
33.0 (36.9)
12.4(15.3)
16.7 ( 9.4)
124.6 (59.0)
129.4 (87.8)
78.3 (77.2)
96.7 (68.9)
10.2 ( 5.0)
8.0(5.4)
5.5 ( 8.9)
6.8 ( 4.6)
* A/C = Air conditoned
Source: Leaderer et al. (1999a).
October 1999
5-51
DRAFT-DO NOT QUOTE OR CITE
-------
1 some data were missing so the total numbers of samples for each category are not equal. The
2 main findings of the summer study were that -
3 (1) there were no significant differences between the mean concentrations of all seven PM
4 categories outside the homes and the same quantities at the regional site in Vinton, VA
5 which was at an average distance of 96 km away from the residences.
6 (2) the cross-sectional correlations of Vinton, VA ambient PM with the outdoor residential PM
7 on paired days were r = 0.5 for fine mode (PMj5, n = 34) and r = -0.2 for the coarse mode
8 (PM10 - PM2 5, n = 30).
9 (3) the indoor mean PM concentrations of the six independently measured quantities listed in
10 Table 5-4 (PM coarse is not an independent measure) were all lower in the A/C homes than
11 in the non A/C homes, which is significant atp = (Yz)6 = 1/64.4
12 Leaderer et al. (1999a) also reported on PM sampling during the winter periods of 1995/96,
13 1996/97 and 1997/98, at 20 residences in Connecticut and 223 residences in the southwestern
14 and central Virginia areas, similar to the summer PM sampling study described above. However,
15 no central site PM data were collected during these winter periods in either state. The residents
16 were all non-smoker families participating in an epidemiologic study of respiratory health with
17 respect to indoor air pollution (Leaderer et al., 1999c). Kerosene heaters were used in 74 homes,
18 and 52 of the homes with no kerosene heaters had gas stoves. Because the subjects were not
19 chosen randomly the home characteristics and corresponding results of the study may not
20 necessarily apply to any other residences and locations within the sampled area.
21 One important difference between the reported results for the winter and summer periods is
22 that during the summer, fluctuations in ambient PM^ were more driven by the fluctuations of the
23 ambient fine mode PM mass and during the winter the fluctuations in outdoor PMg were more
24 driven by the fluctuations of the outdoor coarse mass. Table 5-6 shows the regression
25 coefficients (R2) for these comparisons.
26 The authors assumed no generation of indoor sulfates for the homes without kerosene
27 heaters in operation, and no enhanced deposition of sulfates for the homes with no A/C during
4 Because these are not random samples from a larger domain than N = 58, the measured means only
represent the means at the 58 locations on the 58 days they were sampled, so there is a smaller sampling error than
those predicted by the SD values in Table 5-4. The analytical measurement errors had an SD of 1.2 Mg/m3 for the
PM2 5,1.1 Mg/m3 for the PM10, 2 nmol/m3 for the sulfate, and 1.3 nmol/m3 for the H+ ions.
October 1999 5-52 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-6. SEASONAL VARIATION OF REGRESSION COEFFICIENTS FOR
PM10 WITH FINE AND COARSE MODE FRACTIONS
Season
Summer
Summer
Winter
Outdoor Location
Vinton, VA
Southwest VA
Southwest VA and CT
n
50
45
53
R2, PM10 vs Fine
0.96
0.21
0.07
R2, PM10 vs Coarse
0.40
0.12
0.90
Source: Leaderer et al. (1999a)
1 the summer. Then using the assumption that ambient sulfates characterized the fine mode
2 component of ambient PM, and the measured ratios of sulfate indoors to outdoors, the authors
3 estimated that -75% of the ambient fine mode PM was found indoors during the summer in the
4 homes sampled in southwest Virginia, and-70% of the ambient fine mode PM was found
5 indoors during the winter in the homes sampled in southwest Virginia and Connecticut.
6
7
8 5.10 PERSONAL PM MONITORING STUDIES AND FACTORS THAT
9 INFLUENCE THEIR ABILITY TO ESTIMATE RELATIONSHIPS
10 TO EXPOSURE TO PM OF AMBIENT ORIGIN
11 The science aspects of personal exposure monitor (PEM) usage for monitoring exposure to
12 PM were reviewed in Chapter 7 of U.S. Environmental Protection Agency (1996). A PEM
13 strictly measures the total exposure of the person carrying it and it applies to no other person (see
14 the discussion of Figure 5-2 for possible exceptions from actively smoking and wearing of dust
15 masks). A person stirs up PM instantaneously by the very act of arm waving (Bohne and Cohen,
16 1985; Cohen and Positano, 1986; Wallace et al., 1997) and Thatcher and Layton (1995)
17 demonstrated that the effect of simply walking into a room creates a rapid rise of suspended PM,
18 primarily >10 //m AD. These PM increases are created by the air currents of body movement,
19 particles and fibers dislodged from clothes, and the vibrations and mechanical action of stepping
20 or sitting on fabric surfaces.
21 Because of the PM gradients created by such movements and air currents, there is a
22 microscale variation of PM surrounding a subject and this may contribute to what is known as a
October 1999 5-53 DRAFT-DO NOT QUOTE OR CITE
-------
1 'personal cloud'. Ozkaynak et al. (1996a) defined 'personal cloud' as "a local increase in
2 particles due to unknown sources", as evidenced by people in the PTEAM study who stayed
3 home all day having higher personal PM exposure than the fixed indoor sampler recorded. This
4 personal cloud may lead to a much higher personal PEM concentration of a different chemical
5 composition than that of the PM collected by an area monitor located several meters away from
6 the subject or in another connecting room in the same indoor environment. This phenomenon
7 was recognized by Ogden et al. (1993) who noted that, when comparing static to personal
8 sampling for occupational cotton dust exposure, "a mean background level of 0.5 mg/m would
9 correspond to a median personal exposure of about 2.2 mg/mV'
10 McBride et al. (1999) documented the magnitude of the proximity effect by measuring, in a
11 room of a home, the microscale variations in concentrations surrounding local sources of sulfur
12 hexafluoride (SF6), CO, polyaromatic hydrocarbons, and PM generated by burning an incense
13 stick and by human activity of two-or-more people walking about the carpeted room [by Met-
14 One laser particle counter]. Ratios of particle counts at distances of 1.0 m and 5.7 m from the
15 sources of combustion and the center of activity were obtained at a constant height of 1.53 m
16 above floor level. When only walking was performed, the ratio of closer mean to further mean
17 was of order 1.5 for PM between 2.5 and 10 microns, and 1 for the PM less than 2.5 microns;
18 When only combustion occurred, and no one was in the room, the ratio was of order 3 for PM
19 less than 5 microns and more than 0.5 microns, and 1.5 for PM larger than 5 microns. They
20 conclude that the proximity effect may help explain the existence of the personal cloud denoted
21 by the difference between a personal monitor and an area monitor.
22 The 1996 PM AQCD Chapter 7 provided extensive coverage of many published studies,
23 available as of 1996, that compared measurements of personal exposures to total PM and its
24 constituents with the simultaneous PM and its constituents measured in the ambient air. It is
25 shown below that these studies can be divided into two mutually exclusive categories:
26 Type 1. Longitudinal or time-series studies: A group of people are followed
27 simultaneously for a long enough time (to provide statistical power) to determine the
28 correlation in time of each person's personal exposure to PM with ambient PM
29 concentration.
30 Type 2. Cross-sectional studies: A group of people are followed sequentially for a period
31 of time, with each person's personal PM exposure sampled for only a few days during that
October 1999 5-54 DRAFT-DO NOT QUOTE OR CITE
-------
1 period. A single correlation of the group's personal PM exposures with ambient PM
2 concentration is computed from the subjects' pooled exposure data.
3 Because of their importance, two key Type-1 studies (Lioy et al., 1990; Tamura et al.,
4 1996a) that were discussed in U.S. Environmental Protection Agency (1996) are discussed again
5 in this new light (Mage et al., 1999) along with Janssen et al. (1999b). These three studies are
6 unique in that each subject in the study cohort was followed for virtually the same chronological
7 period of time as all the other subjects. This is in contrast to the Type-2 studies in the literature,
8 such as PTEAM (Clayton et al., 1993), Spengler et al. (1985), and Pellizzari et al. (1999), in
9 which different subjects were sampled on different days.
10 The important difference between these two study designs arises from the fact that each
11 subject may have major residential indoor sources of PM that are highly correlated from day to
12 day (the same stove and cooking fuel, the same rugs and furniture as reservoirs for resuspending
13 PM, the same co-occupants of the home with day-to-day similar smoking or non-smoking habits,
14 etc.) Given that these sources may provide a relatively constant increment of PM exposure to the
15 subjects from non-ambient sources, the day-to-day variation of the study subject's total exposure
16 may be driven by the variation of their exposures to PM of ambient origin. This coherence is lost
17 when different subjects, each with different indoor sources, are sampled on different days and
18 then grouped together for the data analysis, as described below.
19 As an example of such a Type-2 study, Pellizzari et al. (1999) monitored personal exposure
20 to PM2 5 of a random sample of several hundred people in Toronto, Canada, either once or twice
21 for 3-day periods over the year from September, 1995 through August, 1996. The outdoor PlVf 5
22 was monitored simultaneously at the participants residences. The mean outdoor PM5 during the
23 measurement periods was 24.3 ug/m3 and the mean personal exposure PM^ was 67.9 ug/m3.
24 The reported correlation between the logarithms of personal exposure and outdoor concentration
25 is 0.23, statistically significant at a level of 0.01. The authors concluded "These results, while
26 statistically significant, indicate that none of the outdoor concentration data types can adequately
27 predict personal exposure to particulate matter."
28 Lioy et al. (1990) measured 24-h personal PM;0 exposures of 14 non-smoking individuals
29 in Phillipsburg, NJ who were not otherwise exposed to cigarette smoke at home, for
30 14 consecutive days. Phillipsburg, NJ is a small city with a major industrial activity (cast iron
31 pipe production). Lioy et al. installed four non-personal PIV(0 monitors at outdoor locations
October 1999 5-55 DRAFT-DO NOT QUOTE OR CITE
-------
1 distributed throughout the city to monitor the ambient PM^. These ambient PM10 monitors,
2 included the Anderson Hi-Volume sampler and the 10 L/min Harvard impactor. Although they
3 are designed to collect the same nominal PM,0 fraction as the personal PM10 monitors, different
4 monitors have slightly different penetration curves and slightly different cut-points than the
5 personal PM10 monitors. Consequently, it is expected that this variation would increase the
6 variance between the PM measured by a collocated personal PM0 monitor and a different type of
7 ambient PM10 monitor.
8 The matrices of daily personal PM exposure and ambient PM data were reported by U.S.
9 Environmental Protection Agency (1996). The matrices contain several estimated values
10 computed by the E-M algorithm to correct for missing data and three obvious outlier values, as
11 described in U.S. Environmental Protection Agency (1996) and by Mage et al. (1999).
12 Figure 5-8 is the average daily personal PM;0 exposure of the 14 subjects vs the average ambient
13 PM concentration from the four ambient monitors. The regression PC value is 0.91, which
14 assumes that the mean of the ambient PM values is the true mean ambient PM average on a given
15 day. If the three personal exposure outliers (453, 809, 971//g/m3) replace their E-M estimated
16 values, the regression would have a value of R2 = 0.34 instead of 0.91. However, the average is
17 an estimate with sampling error, so an orthogonal regression was also made to find the line that
18 minimizes the sum of the squares of the prediction errors in both exposure and ambient
19 concentration. With this framework, 98.4%of the variation of the mean exposure is predicted by
20 the variation of the mean of the ambient concentrations. U.S. Environmental Protection Agency
21 (1996) concluded that if additional subjects with similar non-smoke exposure lifestyles had been
22 sampled on those 14 days in Phillipsburg, NJ and if additional air quality monitors had been
23 placed in the community to more precisely estimate the mean ambient PM in the community,
24 then the correlation coefficient between mean personal PM exposure and mean ambient PM
25 would approach a limit of one with increasing sample size.
26 Tamura et al. (1996a) measured personal PM;0 exposure of a set of seven elderly non-
27 smoke-exposed individuals living in traditional Japanese homes, and this study was discussed in
28 detail within U.S. Environmental Protection Agency (1996). The study was designed to monitor
29 the exposures of people to ambient PM so they were purposefully chosen to minimize indoor
30 sources of PM. Figure 5-5 shows how well the indoor PIV(0 correlated with the outdoor PMj0 for
31 these seven homes, indicating the minimal effect of indoor sources of PM.
October 1999 5-56 DRAFT-DO NOT QUOTE OR CITE
-------
200-
150-
Q_
"CD
c
o
co
CD
Q.
CD
cn
CD
CD
>
100-
50-
y = 42 + 0.55x
50 100 150
Average Site PM10 (|jg/m3)
200
Figure 5-8. Plot of relationship between average personal PM10 exposure versus ambient
PM10 monitoring data from Phillipsburg, NJ, and regression line calculated by
U.S. EPA.
Source: Lioy etal. (1990).
1 Each subject carried the personal exposure monitor (PEM) for 48-hours, for up to
2 11 periods total over four seasons of the year 1992. Some data were missing due to random
3 equipment failure, and the collected data set was screened to remove any observations which
4 were contaminated by identifiable indoor sources of PM, such as cigarette smoking of visitors, or
5 operation of an indoor combustion source such as a mosquito coil or an incense burner. The
6 ambient data, the remaining exposure data, and the E-M algorithm used to estimate missing
7 exposure data were reported in U.S. Environmental Protection Agency (1996). The individual
8 personal exposure correlations with the ambient PM ranged from 0.77 to 0.96. Figure 5-9 shows
9 the completed means of the eleven 48-hr average values versus the simultaneous 48-hr local
10 Itabashi monitoring station values. The reported statistical analysis in U.S. Environmental
11
October 1999
5-57
DRAFT-DO NOT QUOTE OR CITE
-------
80
"/—
E 60
40-
CD
Q_
0)
D)
CD
20-
0
0
y = 11 +0.47x
30 60 90
Ambient PM10 (|jg/m3)
120
Figure 5-9. Plot of 48-h average personal PM10 exposure and ambient PM10 data from
Japan—linear regression.
Source: U.S. Environmental Protection Agency reanalyses of data from Tamura et al. (1996a).
1
2
3
4
5
6
7
Protection Agency (1996) led to the fitted line that has an intercept of 11.3//g/m3, a slope of
0.47, and a regression R2 value of 0.91.
Tamura et al. (1996b) performed another study in Osaka, Japan similar in design to their
Tokyo study (Tamura et al., 1996a). More than 24 subjects, all housewife non-smokers who
were not exposed to smokers living at home, were sampled in a non-random manner during the
autumn seasons between 1990 and 1995. A dichotomous sampler was used with cut points at
2 and 10 //m AD, and a greased impactor plate to collect PM > 10//m AD. There were 77 days
when outdoor PM concentrations and personal exposures were successfully collected
October 1999
5-58
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
simultaneously in all three size ranges. The authors pooled all 77 pairs of observations and
reported the group correlation coefficient between PM measured immediately outdoors at the
eaves of the home and the personal exposure to PM of the housewife resident. The personal
versus outdoor correlation results were for PM,, r = 0.741; for PM10, r = 0.672; for coarse mode
PM10_2 r = 0.343; and for the PM > 10//m AD, r = 0.05. Because these data do not constitute a
random sample, no statistical inferences can be made to other households.
Janssen (1998) reports the work of Janssen et al. (1995, 1997, 1998a,b, 1999b,c) on a study
of personal monitoring of children and adults in the Netherlands. Janssen (1998) found that in
longitudinal studies "Personal PMj0 exposures of both adults (aged 50 to 70 years) and
children(aged 10 to 12 years) were reasonably well correlated in time with ambient PMg
concentrations. Personal fine particle (FP) exposures were highly correlated with ambient FP
concentrations" as shown in (Table 5-7). These cohorts were not random samples from a defined
population so these results apply strictly to only the adults and children sampled.
TABLE 5-7. AVERAGE LEVELS OF PERSONAL EXPOSURES AND OUTDOOR
CONCENTRATIONS AND THE CORRELATION ( r) BETWEEN THEM IN
LONGITUDINAL EXPOSURE STUDIES
Population
All subjects
Adults
Children
Children
Non-ETS
exposed
Adults
Children
Children
Size
fraction
-
PM10
PM10
PM25
-
PM10
PM10
PM25
Number of
subjects
-
37
45
13
-
23
25
9
Mean*
personal
Mg/rn3
-
62
105
28
-
51
89
23
Mean*
ambient
-
42
39
17
-
41
40
18
Median
individual
correlation r
-
0.50
0.63
0.86
-
0.71
0.73
0.92
Cross-sectional
correlation r**
-
0.34
0.28
0.41
-
0.50
0.49
0.84
*Mean of individual averages.
**Mean value. Estimated by randomly selecting one measurement per subject, 1000 times.
Source of Data: Janssen (1998).
October 1999
5-59
DRAFT-DO NOT QUOTE OR CITE
-------
1 These data show that by excluding subjects with environmental tobacco smoke (ETS)
2 exposure, correlation coefficients increased in all cases. For PM. 5 the median correlation was
3 r = 0.75 for all subjects, and r = 0.84 for all non-ETS exposed subjects. This indicates that, in
4 these individual longitudinal time-series cases, the personal exposure to fine PM is highly
5 correlated with the concentration of ambient fine PM. For comparison, Janssen (1998) obtained
6 cross-sectional correlation coefficients by a Monte Carlo analysis. The values shown in the last
7 column of Table 5-7 were obtained for each cohort by taking one random sample from each of
8 the individual subject data sets (with replacement), and repeating the procedure a total of 1000
9 times.
10 Janssen et al. (1999b) report on a personal monitoring study of 13 non-randomly chosen
11 school children, ages 10 to 12 years, in the rural town of Wageningen, the Netherlands. These
12 children were sampled on one school day per week for a total of eight weeks. Nine of these
13 children lived with parents and relatives who did not smoke tobacco in any form, and their data
14 were chosen for analysis. Each of the nine children had five, six or seven 24-hour observations
15 of personal exposure to PM25 within the eight days monitored. The ambient PM, 5 was collected
16 at a central monitoring site on these same days for comparison using an identical monitor.
17 Because there were widely different numbers of children with data on any given day in this time
18 series (1 to 9), the missing data for each child was estimated using the E-M algorithm, and the
19 results are shown in Table 5-8 along with their mean exposure and corresponding ambient
20 concentration.
21 Figure 5-10 shows the relation of the estimated mean total PM of the children as a function
22 of the simultaneous ambient PM^ measured at the central station. The regression R2 of the PM25
23 daily mean exposures with the daily ambient PM is 97.8%, and the intercept is 12//g/m3. This
24 analysis of children who were constrained by design to have the same time-activities at school
25 and at home, where there were no major indoor sources of PIV|5 that could vary from day to day,
26 shows that in the absence of large variability of non-ambient sources of PIV|5, variations in total
27 personal exposure to PM25 is virtually controlled by the variations of ambient PM, 5.
28 Janssen et al. (1998a) report on a study of non-randomly chosen adult's exposure to PMg in
29 the Netherlands during the two winter periods of 1994. A total of 37 nonsmoking adults
30 (18o*, 19?), living with nonsmokers, average age 64-y, were monitored with personal PM,,
31 monitors with simultaneous monitoring of the local ambient PM0. One subset of 13 adults was
October 1999 5-60 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-8. PERSONAL EXPOSURES TO PARTICULATE MATTER (PM 2 5) OF
NON-SMOKE EXPOSED CHILDREN IN WAGENINGEN, NL, AND STATIONARY
AMBIENT MONITORING DATA (SAM). MISSING VALUES ESTIMATED ( )
Day
1
2
3
4
5
6
7
8
Person ID
01
19.8
29.5
15.3
(27.1)
(41.2)
29.3
34.1
(51.3)
03
(8.0)
15.9
11.2
(16.6)
20.2
26.0
29.3
53.7
05
12.5
17.1
28.3
(21.1)
15.8
30.8
35.9
(41.1)
08
13.5
15.0
22.9
20.5
19.6
21.0
26.7
(30.7)
09
14.6
20.7
(40.9)
(29.4)
19.5
24.4
30.7
(35.5)
11
17.2
17.0
(29.2)
(25.9)
19.7
28.7
42.1
(45.2)
13
19.9
13.7
11.9
(18.3)
15.0
22.2
29.4
(49.0)
14
18.3
17.1
(17.0)
(20.0)
15.7
26.7
29.0
34.8
15
13.1
(10.5)
11.2
(15.0)
13.4
(20.1)
26.7
40.7
Mean
Personal
15.21
17.39
20.88
21.54
20.01
25.47
31.54
42.44
SAM
6.24
6.61
12.15
12.51
15.70
20.56
31.67
45.24
Source of Data: Janssen et al. (1999b).
1 monitored in the period January through March 1994, and another subset of 24 adults was
2 monitored in the period October - December, 1994. Because there was no overlap of these two
3 monitoring periods we analyzed each cohort separately.
4 The data sets for the adults are shown in Tables 5-9 and 5-10, for the 13 and 24 adult
5 cohorts, respectively. In both cases the E-M algorithm could not be used to estimate the missing
6 values because of the sparse coverage of these data. The extensive missing data caused the
7 estimation routines to converge for several values in the negative concentration domain, which
8 has no physical meaning. The method chosen for analysis was to use a linear model that was
9 fitted to these data which included terms for days and persons, but no interaction terms.
10 This model we chose is equivalent to running an unbalanced two-way analysis of variance
11 (ANOVA) on these data to estimate the marginal distribution of the means of the daily personal
12 exposure data. Because we are only obtaining estimates of the model parameters (not testing for
13 significance), the correlations between the people can be ignored (Winer, 1962). The last
14 columns in Tables 5-9 and 5-10 show the estimated means that were used in the regression
15 analysis.
October 1999 5-61 DRAFT-DO NOT QUOTE OR CITE
-------
CO
c
o
(/)
s_
-------
o
o
r-K
o
o
\Q
\Q
\Q
Lf\
i
S
o
Tl
H
O
O
o
H
O
c
o
H
W
0
HH
H
W
Day
102
103
202
203
702
902
903
1402
1403
1503
1701
2001
2102
2103
2203
2401
2701
2802
2803
2903
3003
3101
TABLE 5-9. PERSONAL PM10
SOURCE
Ambient 101 102 103 104
80.3
40.4
45.2 45.1
53.3 58.3
36.2 59.5 [68.4] 46.5 40.7
37.2
28.5 40.3 35.3
75.0 [66.1]
30.6 [95.6] 172.9
29.7 65.3
19.4 [70.9]
61.9
63.0 94.3 [95.2]
33.6 [76.8]
32.5 78.1 23.3
25.8 29.9
23.3 116.6 54.6
48.0 75.0 [108] 112.1
41.0 83.2 45.3
47.9
33.1 37.9
35.0 27.5
EXPOSURE 0"g/m3) OF 13 ADULTS IN AMSTERDAM,
OF DATA: JANSSEN ET AL. (1998A)
Person ID
105 107 109 110 111 113 114 116 118
143.4 71.8
94.7
56.2
50.5 125.9 87.1 58.8
172.9 87.5 [93.0]
43.9 75.1 32.8 79.1 38.6
39.7 [45.2] 33.1 57.4
[87.3] 47.0 65.2 88.7 72.9
42.1 40.4 70.4
[117] [68.4] 75.2 38.7
39.9
89.5 139.1 104.3 95.5
149.3 163.2 100.5 80.9
82.6 84.8 28.3 [70.3]
22.5 [58.6] 39.5
58.2 26.4 [99.9]
35.7 54.2 79.3 48.1
[80.0] 62.7 49.0 84.8
103.9 37.8 30.5 57.6 54.8
[69.7]
62.4
123.3 [70.7] 34.7 50.7
NL
Mean of
Personal
Raw Data
107.6
94.7
50.7
78.1
81.8
53.9
41.1
68.5
81.7
44.8
39.9
107.1
117.6
65.2
40.9
38.2
64.8
86.3
59
[69.7]
50.2
59.1
The data marked [ ], representing smoke exposure at home, were not used in the ANOVA. No estimate of the mean could be made for day 103
Mean of
Personal
Estimated*
90.6
-
52.4
70.2
66.4
60.8
55.2
81.5
88.8
58.1
40.7
77.8
124.7
48.1
65.2
41.1
56.7
87.6
58.0
-
68.9
69.4
because it had
only one datum point; person 102 was smoke exposed at home for all sampled days and was excluded from all analyses; on day 2903 there were no non-smoke
exposed data.
* - Estimated mean of all subjects' non-smoke PM exposure as computed by U.S. EPA via a 2-way ANOVA.
-------
TABLE 5-10. PERSONAL PM10 EXPOSURE fag/m3) OF 24 ADULTS IN AMSTERDAM, NL
o
o
•O
•O
•O
01
o^
Pv
o
£>
Tl
H
O
O
O
H
O
c
o
H
W
0
O
H
W
Day
\n
311
5U
6U
1\\
&U
1011
1212
1411
1512
1710
1711
1912
2010
2111
2212
2410
2411
2710
2811
3110
avjUKA^JL vjr jj/vi/v: jAmoojiim ILL p
Person ID
Ambient 301 302 303 306 307 308* 310 312 313 315 316 317 318 321 322
42.5 29.7 73.7 78.3 42.1
46.7 37 38.9 36.1 83.7
29 31.8 58.2 50
24.8 29.2 60.7 55.6
84.5 67 99.6 66.2 50.2 51.7 77.7
22.9 31.3 17.6 61.6 38.3 31 33.9
44.9 43.9 100.6 45.9 85.7 72 80.6
40.1 31.5 43.9 [41.6] 50.4 51.1 42.4 34.6 45.7
35.2 45.3 50.4 [23.3] 32.1 66.1
63.9 70.5 75.6 36.7 45.8 52.2
32.7 51.1 53.8 [92.2] 55.5 54.6
31 27.7 30.8 52.8 36 28.9
20.4 40.8 [90.9] 66.9 49.5 69.6 24 34 51.4
77.5 129.2 58.6 101.4 82.1 73.1
67.5 109.9 68.8 52.4
25.8
34 63.4 59.2
29.8 45.5 34.2 49.4 33.3 49.7 22.7 21.6
32.2 11.5 52.2 [35.8] 88.3 30.8 18.1
27.6 30.2 81.3 60.2 [71] 34.4 57.7
29.5 46.8 42.8 40.7 35.8 79.8
LLI. {lyyoA)
Mean of
Personal
325 327 331 333 334 336 337 338 339 Raw Data
71.7 57.4 61.7 86.0 109.8 67.8
73.9 56.8 86.2 58.9
31.3 40.3 45.0 42.8
48.5
114.4 55.8 75.6 59.6 71.8
44.0 26.5 59.8 68.1 41.2
50.3 [112.4] 68.4
44.2 75.4 50.7 47.0
[86.6] 34.1 61.6 63.4 50.4
69.9 75.7 65.7 85.8 64.2
53.8
69.4 40.9
71.6 42.7 84.9 53.5
80.1 81.8 86.6
87.1 48.3 40.9 68.9 51.8 75.6 67.1
71.5 50.6 92.7 71.6
70.0 35.6 56.6 61.3 34.0 54.3
104.3 52.2 45.9
32.6 46.6 40.0
51.5 51.1 57.1 52.9
31.0 42.8 45.7
Mean of
Personal
Estimated*
63.7
58.2
47.1
54.7
70.1
42.7
76.6
46.8
49.1
69.0
49.4
44.4
51.0
95.4
59.3
54.6
49.3
50.9
49.5
51.1
45.1
[ ] - smoker present at home
* - Estimated mean of all subjects' non-smoke PM exposure as computed by U.S. EPA via a 2 -way AN OVA.
-------
120
A
Ambient PM Concentration (|jg/m )
140-
120-
E 100-
B
£ 80-
to
o
Q.
UJ
60-
CD 40-
§
0
CL
20-
n = 13
y = 36.4 +0.760 x
R2 = 0.424
10 20 30 40 50 60 70 80 90
Ambient PM Concentration (|jg/m3)
Figure 5-llA,B. Completed mean personal PM10 exposures of two groups of adults (n = 24
[A] and n = 13 [B]) not exposed to tobacco smoke at home in Amsterdam,
NL, versus the simultaneous ambient PM10 measured in their community.
Source of data: Janssenetal. (1998a)
October 1999
5-65
DRAFT-DO NOT QUOTE OR CITE
-------
1 ambient PM10 is reflected as an increase in the exposure to PM10 of the subjects in this study.
2 These correlation coefficients for adults who were not smoke-exposed at home, are both higher
3 than the estimated mean cross-sectional correlation of r = 0.50, shown in Table 5-7.
4 The Janssen et al. (1999b) personal exposure intercept of 12//g/m3 for PM25 corresponds to
5 the average portion of the daily PM25 exposure that is uncorrelated with the ambient PM25
6 concentration, primarily while the subjects were indoors. This quantity represents the
7 combination of indoor emissions of PM25 and the children's 'personal cloud' of PM25 stirred up
8 by their activities, and the net effect of experimental errors in ambient PM measurement. If the
9 children's personal clouds for coarse PM were much greater than their personal clouds for fine
10 PM as in the PTEAM study (Ozkaynak et al., 1996a), and there was much more coarse PM
11 generated indoors than fine PM (there was no cigarette smoke exposure at home or in school),
12 then their intercept for personal PM10 exposure could be much greater and of the order of the
13 Lioy et al. (1990) intercept of 41 //g/m3 or the Janssen et al. (1999b) intercepts of 34 and
14 36 //g/m3 for PM10. Values of this order are much larger than the Tamura et al. (1996a) intercept
15 value of 11 //g/m3 for PM10. This difference may be due in part to the relative cleanliness of the
16 Tokyo homes of the seven elderly Japanese subjects compared to the Phillipsburg, NJ and
17 Wageningen, Netherlands homes, and the lower physical activity levels of these elderly people
18 compared to the activities of the monitored children and working adults in the Netherlands.
19 The results of these three regressions show that the variations in personal exposure for
20 these non-randomly chosen subjects who have similar life styles (habits and activities) from day
21 to day, and who are not exposed to tobacco smoke on a routine basis, are driven by the variations
22 in the ambient PM concentration. This finding, originally expounded by Janssen et al. (1995),
23 supports the plausibility of the use of fluctuations of ambient PM concentration as a surrogate for
24 fluctuations in human exposure to PM of ambient origin for subjects residing in the community
25 surrounding an ambient monitoring station. However, this plausibility is weakened by the
26 non-random choice of subjects and sampling periods which strictly limit the finding to the
27 subjects sampled on the days sampled.
28
29
October 1999 5-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.11 EXPOSURES TO AMBIENT PM25 OF PEOPLE BELIEVED TO BE
2 SUSCEPTIBLE TO THE EFFECTS OF AMBIENT PM2 5
3 The study of the historic air pollution episodes reveals that those people most subject to the
4 mortality effects of exposure to ambient PM2 5 are the elderly with pre-existing cardio-pulmonary
5 lesions (U.S. Environmental Protection Agency, 1996). Although it has been established in the
6 previous section that, in general, peoples' exposures to PM of ambient origin are highly
7 correlated with ambient PM in their community, prior to 1996, very few studies of the elderly
8 and apparently susceptible people had been performed. For example, the Tamura et al. (1996a)
9 study was of elderly people with no known cardio-pulmonary disease.
10 The question to be addressed here is 'Do elderly people with cardio-pulmonary disease
11 have the same relationship of their exposure to PM of ambient origin as do the members of the
12 general public who have been studied to date, as reported in the preceding section?' It may be
13 possible that the elderly or infirm modify their physical activities and living conditions in such a
14 manner as to influence the relationship of their exposures to ambient PM with the ambient PM
15 concentrations of their community. If so, differences in exposure estimate errors between
16 susceptible-diseased groups and non-susceptible-healthy groups may result in differential
17 misclassification which has implications for the validity and interpretation of epidemiologic
18 studies (Armstrong et al., 1992). The studies described in the following section address studies
19 of this apparently susceptible cohort.
20 Williams et al. (1999a) measured the personal exposures to PMj 5 of five elderly subjects
21 living at a retirement center in Baltimore, MD in January-February, 1997 for ten consecutive
22 days. Some of these subjects were under medical treatment for cardio-pulmonary conditions at
23 the time of this study, and they represent a non-random sample of elderly subjects living in an
24 environment of a retirement center. Each subject carried a personal PMj 5 monitor described by
25 Williams et al. (1999b), and the ambient PM2 5 was monitored simultaneously outside the
26 retirement center by a dichotomous sampler. Three subjects (1-3) lived apart in separate
27 efficiency apartments in a new unit and the two other subjects (4, 5) lived apart in an older unit,
28 in two separate smaller rooms, each with an attached bathroom. The three newer apartments had
29 either a picture window or double windows and the two older single-rooms had only one
30 window. Thus, even though none of the residents reported opening windows during the study,
31 the newer units may have had a larger window-perimeter leakage area for air exchange directly
October 1999 5-67 DRAFT-DO NOT QUOTE OR CITE
-------
1 with the outdoor air. The new and old units also had different air handling systems in their two
2 general living areas which were connected by staircases and elevators. The authors reported that
3 there were no special air cleaners installed in either of these two air handling systems.
4 The collected data are shown in Table 5-11 with the missing values estimated by the E-M
5 algorithm (Mage et al., 1999) shown within parentheses ( ). Because of the different air
6 handling systems for their living areas, the E-M algorithm was applied separately to these data
7 for subjects 1-3, and then to these data for subjects 4 and 5. These different air handling
8 systems, and the difference in window leakage areas, may be a partial explanation of why
9 subjects 1-3 had positive correlations of their exposures with the ambient PM ( r = 0.91, 0.12,
10 0.70 respectively), and subjects 4 and 5 had negative correlations of their exposures with the
11 ambient PM ( r = -0.55, -0.05 respectively). The mean PMj 5 exposure vs ambient PM2 5 for all
12 five subjects is shown as Figure 5-12, with an R2 of 27%. The range of nine observations of
13 ambient PM25 is fairly narrow (8 to 23 //g/m3) and inclusion of the last day of data with an
14 ambient PM2 5 value of 32 //g/m3 raises the correlation of their mean PM exposure with the
15 ambient PM from r = 0.269 to r = 0.522. It is expected that all these correlations would have
16 been higher than shown if both the ambient monitor and the personal monitor measured PM with
17 the same nominal size cut.
18 Bahadori (1998) and Bahadori et al. (1999) report on a pilot study of the PM exposure of
19 ten non-randomly chosen Chronic Obstructive Pulmonary Disease (COPD) patients in Nashville,
20 TN, during the summer of 1995. Each subject alternately carried a personal PM2 5 or PM10
21 monitor for a 12-h daytime period (8 a.m. - 8 p.m.), for six consecutive days. These same
22 pollutants were monitored simultaneously indoors and outside their homes. These homes were
23 all air conditioned with lower air exchange rates (mean a = 0.57/h), which may have contributed
24 to the finding that mean indoor PM2 5 was 66% of the mean ambient PM2 5. This can be
25 contrasted to the PTEAM study in Riverside, CA, where no air conditioners were in use and the
26 air exchange rate was much higher (mean a > 0.97/h). In the PTEAM study the mean indoor
27 PM25 was 98% of the mean ambient PM25 (Clayton et al., 1993).
28 Because each person carried a PM2 5 or a PM10 personal monitor for only three 12-h periods,
29 no Pearson correlations of personal PM vs ambient PM were reported for the individual subjects.
30 However, these data were combined into two groups of 30 and Bahadori et al. (1999) report the
31 correlations found between personal and outdoor concentrations when data were analyzed
October 1999 5-68 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-11. PERSONAL EXPOSURES TO PM15 OF FIVE ELDERLY RESIDENTS
OF A RETIREMENT CENTER IN BALTIMORE, MD, AS A FUNCTION OF
AMBIENT PM2 5 CONCENTRATION
Day
1
2
o
3
4
5
6
1
8
9
10
Correlation
Ambient
PM25
14.8
9.2
14.2
14.0
22.6
21.2
11.6
8.0
14.5
32.0
Person 1
Exposure
21
22
17
25
33
32
27
12
29
50
r = 0.907
Person 2
Exposure
31
16
35
20
38
29
33
35
45
(29.5)
r = 0.122
Person 3
Exposure
(32.0)
(38.7)
18
(39.5)
(39.6)
44
37
30
(35.2)
57
r = 0.702
Person 4
Exposure
(33.1)
39
40
18
32
26
51
39
38
(24.7)
r = -0.550
Person 5
Exposure
31
27
43
30
30
35
19
45
58
(34.2)
r = -0.050
Mean
Exposure
29.6
28.5
30.6
26.5
34.5
33.2
33.4
32.2
41.0
39.1
r = 0.522
( ) Missing data estimated by U.S. EPA using E-M algorithm (U.S. Environmental Protection Agency, 1996; Mage et al., 1999).
Source of data: Williams et al. (1999a).
1 cross-sectionally were r = 0.09 for PM2 5 and r = -0.08 for PM10. The authors suggest that
2 "personal-outdoor correlations would be stronger if data were to be analyzed longitudinally.
3 However, this analysis could not be performed due to insufficient data." That relationship, if
4 true, would be consistent with that in U.S. Environmental Protection Agency (1996) where it was
5 shown how combining personal exposures for an individual with those for other individuals,
6 leads to a degradation of the correlation coefficient.
7 Rojas-Bracho et al. (1999) report the results of a 1996/1997 study of the personal exposures
8 to PM25 and PM10 of 18 COPD patients in the Boston, MA metropolitan area. The study design
9 was based on the pilot study by Bahadori et al. (1999) in Nashville, TN. The subjects were also
10 not chosen as a random sample from the population of COPD patients in the Boston area, but
11 were recruited via local physicians, from COPD exercise groups, as well as by newspaper
12 advertisements. Therefore the results of the study strictly only apply to the 18 subjects on the
13 days sampled, and cannot be inferred to relate to Boston area COPD patients in general.
14
October 1999 5-69 DRAFT-DO NOT QUOTE OR CITE
-------
Ambient PM25 Concentration (|jg/m3)
Figure 5-12. Mean personal exposures to PMj 5 of five elderly residents of a retirement
home in Baltimore, MD, as a function of ambient PM2 5 concentration.
Source of data: Williams et al. (1999a).
1 Each subject carried a personal monitor that sampled both PM25 and PM10 for 12-hours per
2 day for one, two or three consecutive 6-day periods between 8 a.m. and 8 p.m., while PM25 and
3 PM10were also measured inside and outside their homes during the same time period. Subjects
4 were monitored for either summer only (2), winter only (2) or in both summer and winter(14).
5 The regression coefficients (R2) between the daytime personal exposure and daytime outdoor PM
6 measures for the 15 subjects with more than one week of sampling are shown in Table 5-12a.
7 The 0.85 maximum value in the set is for the one subject who was only sampled in the summer
8 period. As expected, the combining of summer and winter data reduced the correlation between
9 personal and outdoor PM data because air exchange rates in Boston are decreased in the winter to
10 save heat. The article does not report the difference between summer and winter correlations for
11 the other 14 subjects.
October 1999
5-70
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-12A. SUMMARY OF CORRELATIONS BETWEEN DAYTIME
PERSONAL PM EXPOSURES AND DAYTIME OUTDOOR PM CONCENTRATIONS
FOR 15 COPD PATIENTS DURING COMBINED SUMMER AND WINTER
PERIODS IN BOSTON, MA
Person ID
1
2
3
4
5
6
9
10
11
12
13
14
16
17
18
Weeks sampled in
Summer (S) and
Winter (W)
S 1,W1
S2, Wl
S 1, Wl
S 1,W1
S 1,W1
S 1, Wl
S2, Wl
S2, Wl
S2, Wl
S 1, Wl
S2
S2, Wl
S1,W1
S2, Wl
S2, Wl
Number of
valid paired
observations
10
10
17
9
12
11
12
17
17
17
10
16
12
17
17
R2 PM25
Personal vs Outdoor
0.58
0.21
0.13
0.44
0.28
0.01
0.65
0.49
0.32
0.02
0.85 (maximum value)
0.83
0.52
0.01*
0.37
R2 PM10-PM25
Personal vs Outdoor
0.11*
0.10
0.06*
0.26*
0.02*
0.31
0.00
0.00
0.00*
0.09
0.00* (9 observations)
0.00
0.12
0.05*
0.39
* Correlation coefficient (r) is negative.
Source: Rojas-Bracho et al. (1999).
1 Table 5-12a shows that the personal exposures to PM2 5 were all positively correlated with
2 the ambient PM, except for subject 17 (the slope of the regression line is negative). However, for
3 the difference between PM10 and PM2 5, representing the coarse mode fraction between 2.5 and
4 10 microns AD (neglecting cutpoint imprecision), there were 7 negative correlations, one zero
5 correlation, and 7 positive correlations, which is an expected result if there is little or no
6 correlation, on average, between ambient coarse PM and personal exposure to coarse PM for
October 1999
5-71
DRAFT-DO NOT QUOTE OR CITE
-------
1 such people who spend the majority of their time indoors where coarse PM is generated by daily
2 activities.
3
4
5 5.12 PERSONAL EXPOSURE TO CONSTITUENTS OF PM: SULFATES
6 AND ACIDITY
7 PM sulfates are species that exist in the ambient atmosphere from primary emissions of
8 combustion products of fossil fuels containing sulfur, from formation by photochemical
9 oxidation of gaseous sulfur species, and from non-anthropogenic sources (e.g. volcanic activity
10 and wind-blown soil). In the indoor environment, where ammonia is present, the only common
11 major source of sulfate may be resuspension by human activity of PM containing ammonium
12 sulfates and soil sulfates that were tracked into the home. In some homes an unvented kerosine
13 heater using a high-sulfur fuel may be a major winter contributor (Leaderer et al., 1999a).
14 Trijonis et al. (1980) reported surface monitoring sulfate data from the U.S. EPA Regional
15 Air Pollution Study (RAPS) of the greater St. Louis, MO metropolitan area including portions of
16 Illinois. The authors reported that all site-to-site correlation coefficients of 24-h sulfate data were
17 greater than 0.84. Suh et al. (1993) reported a study of personal exposure to PM sulfate of
18 children living in State College, PA during a summer season. Figure 5-13 shows the excellent
19 agreement between the personal exposure and ambient sulfate concentration. The regression
20 slope of 0.87 falls below the 1:1 line because sulfates, although primarily of < 1 //m AD, deposit
21 on inner home surfaces with a value of k of order 0.16/hr (Ozkaynak et al., 1996b). This leads to
22 a value of [Pa / (a + k)] of slightly less than 1 in the summer when homes are not air conditioned
23 and sealed tightly (a »0 and P ~ 1).
24 If the home is sealed to conserve energy with air conditioning use, the air exchange
25 parameter a is decreased. If the return-air to the air conditioner is filtered, or passes through an
26 electrostatic precipitator, then the deposition parameter k is increased as some sulfates may be
27 removed. This leads to a decrease in the parameter [Pa/(a + k)], as shown by the data on
28 Figure 5-13 where the open circles representing the air conditoned homes fall below the data for
29 the non air conditioned homes. The overall regression coefficient for these data from non air
30 conditioned homes is R2 = 0.92, indicating how well personal exposure to this material of almost
31 exclusively ambient origin correlates with ambient concentration.
October 1999 5-72 DRAFT-DO NOT QUOTE OR CITE
-------
600
500
0)
O 400
E
^
-------
1 The effect of air conditioning may be important to consider when comparing responses to
2 air pollution in cities with different percentages of air conditioned homes. For example, if cities
3 A and B have identical ambient air pollution concentrations and city A is completely air
4 conditioned while city B has no need for air conditioning, the subjects in city A will be exposed
5 to less ambient PM than the subjects in city B, and consequently, if the subjects in city A have
6 less symptoms, the ambient PM in city A may appear to be less toxic. See the discussion of the
7 effect of assumed protection by going indoors when smog warnings are issued, from the 1969
8 PM AQCD (National Air Pollution Control Association, 1969) mentioned in Section 5.1.1.
9 Ozkaynak et al. (1996a,b) and Janssen et al. (1999c) report that in the U.S. PTEAM and
10 Netherlands studies respectively, that XRF analyses of indoor PM and the immediate outdoor
11 PM show that sulfur is the only element reported with virtually identical mass concentration in
12 both the indoor and outdoor air. Therefore, where there are no indoor sources of fine mode
13 sulfates, and no enhanced air cleaning systems operating, one may deduce the same relationship
14 between personal exposure to PM of ambient origin (C) and ambient concentrations (Co) in an
15 indoor microenvironment for other materials in the same aerodynamic size range as the fine
16 mode sulfates found in the atmosphere.
17 Larssen et al. (1993) report such a study of traffic generated PM in Oslo, Norway. They
18 made measurements of indoor and outdoor fine fraction PM25 and coarse fraction (PM10 - PM25)
19 in a closed and uninhabited apartment facing on a busy street. Figure 5-14 presents three
20 examples of their daily data. In the absence of indoor sources, these data show that with low air
21 exchange rates (not monitored), the PM equilibrium parameter [P a I (a + k)] is of order 5-10%
22 for the coarse PM fraction and about 80% for the fine PM fraction.
23 Tung et al. (1999) developed a methodology to measure the particulate penetration
24 coefficient (P) through a building shell by measuring simultaneously the air exchange rate (a),
25 and the indoor and outdoor PM concentrations as the indoor PM equilibrates with the outdoor
26 PM in the absence of any PM sources within the indoor microenvironment (microenvironment).
27 The PM deposition parameter (K) is then estimated by the measured decay rate of the PM
28 (e.g., a + K) minus the measured air exchange rate (a), and the penetration is computed as
29 P ~ C (a + k) I (Ca a). The authors demonstrated their technique using a large enclosed office as
30 the simulated building microenvironment, while it was unoccupied at night with all mechanical
31 ventilation shut off. The office had two closed doors leading to a corridor with closed louvers
October 1999 5-74 DRAFT-DO NOT QUOTE OR CITE
-------
250-1
50-|
Coarse Fraction (2.5—10 |jm)
Outside
Inside
Day Night Day Night Day Night
Fine Fraction (< 2.5 |jm)
Outside
Inside
Day Night Day Night Day Night
Figure 5-14. Results from simultaneous measurements of the indoor PM10 concentration
and the immediate outdoor PM10 concentration of an uninhabited apartment
in a building fronting to a busy street in Oslo, Norway.
Source: Larssen et al. (1993).
October 1999
5-75
DRAFT-DO NOT QUOTE OR CITE
-------
1 above them, and the corridor PM simulated the ambient PM. In three tests of the procedure the
2 penetration of PM10into the office had an average value of P ~ 0.8 corresponding to an average
3 air exchange rate of a ~ 0.4/h.
4 Ormstad et al. (1997) measured PM10 indoors and outdoors for 29 occupied homes in Oslo,
5 Norway. Using a 60 Lpm open face sampler, pointing 45° downward, they found that the
6 median indoor/outdoor mass ratio was 1.43. PM2 5 was not monitored, but the authors used
7 transmission electron microscopy (TEM) to estimate that the vast majority of countable
8 particles(> 90%) were under 2.5//m optical diameter. It is interesting to compare these data with
9 the Larssen et al. (1993) data, cited just above. For a closed and unoccupied apartment in Oslo,
10 the indoor/outdoor ratio was of order 0.2. The difference between these indoor/outdoor ratios
11 (~ 1.4 vs -0.2) may represent the effects of indoor PM generated and resuspended by occupants,
12 and an increase in air exchange rate with door and window openings and closings by the
13 occupants.
14 Brauer et al. (1989) measured personal exposures and ambient concentrations of acidic
15 aerosols and gases using a personal annual denuder/filter pack sampling system. They found that
16 personal exposures to aerosol strong acidity (H+) were slightly lower than concentrations
17 measured at a stationary site, and concentrations of sulfate and ammonium ions were similar to
18 those measured at the stationary site.
19 In the absence of personal activity to resuspend or generate PM, and with a very low air
20 exchange rate, the indoor PM fine fraction is virtually all of ambient origin and the PM coarse
21 fraction is minimal due to its higher settling values (k) and lower penetration values (P). The
22 caveat to this analogy is that while indoors, charged ambient particles and ultra-fine ambient PM
23 (< 0.1 //m AD) may have greater electrostatic deposition and diffusion deposition than the larger
24 sulfate particles (> 0.1 //m AD), which would inflate their deposition parameter k. Although the
25 correlation between the ambient concentrations of such particles and the related exposure
26 measurements may be much lower due to the fluctuations of the transient forces causing
27 deposition, it remains that any exposure to these PM constituents which have no indoor sources,
28 will be entirely caused by the PM of ambient origin.
29
30
October 1999 5-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.13 CORRELATION OF AMBIENT PM AND PM EXPOSURE IN
2 CROSS-SECTIONAL-TYPE STUDIES
3 The results for personal sulfate exposure and longitudinal (Type-1) personal PM exposure
4 studies stand in contrast to the lack of correlation found in cross-sectional studies (Type-2)
5 described in Chapter 7 of U.S. Environmental Protection Agency (1996). Typical of these
6 studies is Spengler et al. (1985), which found virtually no cross-sectional correlation (r ~ 0)
7 between personal exposure to PM3 5 and ambient PM3 5 concentration in the adjacent
8 communities of Kingston and Harriman, in Eastern Tennessee. The reason for this difference
9 can be visualized by comparison of the example data matrices for the Type-1 and Type-2 studies
10 shown in Tables 5-12 and 5-13: The longitudinal Type-1 studies of Tamura et al. (1996a),
11 Janssen et al. (1999b) and Lioy et al. (1990) correspond to Table 5-12. However, the Spengler
12 et al. (1985) study and the PTEAM study (Clayton et al., 1993) were of Type-2, the class covered
13 by Table 5-13.
14
15
TABLE 5-12. EXAMPLE OF A COMPLETED MATRIX FOR A TYPE-1 ANALYSIS
OF PM EXPOSURE AND AMBIENT PM CONCENTRATION
Day 1
Day 2
Day 3
Day 4
Person 1
PM Exposure
X
X
*(X)
X
Person 2
PM Exposure
*(X)
X
X
X
Person 3
PM Exposure
X
*(X)
X
X
Person 4
PM Exposure
X
X
X
*(X)
Ambient PM
Concentration
X
X
X
X
*(X) missing data filled in by the E-M algorithm from assumption of bivariate normal distributions of daily personal exposures and each
individual's exposure time series.
1 In the Spengler et al. (1985) study, subsets of people were sampled on subsets of days
2 without any overlap. In the PTEAM study four different people were to be sampled on each
3 consecutive day with no repeat sampling of any person, but some people's data were missed due
4 to happenstance. Without any overlapping of sampling between all people one is unable to
5
October 1999 5-77 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-13. EXAMPLE OF AN INCOMPLETE MATRIX FOR TYPE-2 ANALYSIS
OF PM EXPOSURE AND AMBIENT PM CONCENTRATION
Day 1
Day 2
Day 3
Day 4
Person 1
PM Exposure
X
X
_*
.*
Person 2
PM Exposure
X
X
_*
.*
Person 3
PM Exposure
.*
_*
X
X
Person 4
PM Exposure
.*
_*
X
X
Ambient PM
Concentration
X
X
X
X
*Missing data that cannot be estimated by E-M algorithm because there is no comparison data on how the
personal PM exposures of Persons -3 and -4 relate to the simultaneous personal PM exposures of Persons
- 1 and -2.
1 estimate the missing data by the E-M algorithm in these two studies. For example, if person-3 in
2 an extensive data set always has the mean personal exposure of person-1 and person-2 we can
3 estimate person-3's exposure as the mean of the person-1 and person-2 exposures on the day
4 when person-3 was not sampled. However, if person-3 was never sampled simultaneously with
5 persons -1 and -2 this relationship cannot be discovered. Consequently, the E-M algorithm
6 cannot be applied to complete the Type-2 matrices of the form shown as Table 5-13.
7 In the PTEAM study all subjects were a set of randomly sampled residents of Riverside,
8 CA aged 10 and above, who were self-declared nonsmokers. The daily PM exposure data from
9 the subjects were averaged to create a time series of mean exposure values that were an estimate
10 of the mean personal exposure of the Riverside cohort on those days. Most days had four
11 observations but some days had only two or three observations, so the means of the four values
12 are expected to have smaller confidence intervals than the means of the other values. Each
13 missing personal PM exposure datum was estimated using the E-M algorithm. This amounts to
14 finding a set of missing values that maintain an assumed bivariate normal correlation structure of
15 the time series of completed (sets of 4) mean PM exposure data and ambient PM data.
16 The resulting table of average PM10 personal exposure and average ambient PM10
17 concentration of 43 days of data is as provided in Table 7-48 in U.S. Environmental Protection
18 Agency (1996). Figure 5-15 shows these data, and the regression parameters are given in
19 Table 5-14.
October 1999 5-78 DRAFT-DO NOT QUOTE OR CITE
-------
200-
150-
• •
CO
c
o
cu
Q.
100-
50-
50 100 150
Ambient SAM |jg/m3
200
Figure 5-15. PTEAM mean 24-h PM10 data compared for personal PEM and SAM.
Source: U.S. EPA reanalyses of data reported on by Ozkaynak et al. (1996a).
TABLE 5-14. RESULTS OF THE ANALYSIS OF THE PTEAM DAILY AVERAGE
PERSONAL PM10 EXPOSURE DATA FROM RIVERSIDE, CALIFORNIA
Regression Model
Linear, normal error
Linear, lognormal error
Orthogonal
Linear adjusted for person error
Measures of Association
Correlation of averages
Fraction of variation explained by orthogonal regression
Slope
0.6174
0.6185
0.8071
0.9675
Intercept
59.7
57.1
44.2
31.0
Value
0.721
0.864
Source: U.S. Environmental Protection Agency (1996) reanalysis of PTEAM data (Ozkaynak et al.,1996a,b).
October 1999
5-79
DRAFT-DO NOT QUOTE OR CITE
-------
1 The regression R2 value for these averages is 0.52 which is much less than the regression
2 R2 values for the longitudinal time-series data averages of order 0.9 corresponding to a completed
3 matrix (Table 5-12). This, in effect, is caused by the wider variance in the increment of non-
4 ambient PM added to the PTEAM subjects' exposure to ambient PM, produced by the day-to-day
5 variation of the indoor PM sources and personal activities of the different people who were
6 sampled for only a single 24-hr period. For example, on day 1, the four people chosen may not
7 live with smokers and have personal PM exposures at or moderately above the ambient PM
8 value. On day 2, if some of the new subjects live with smokers, their personal exposures may be
9 well above the ambient PM value.
10 The coherence of the time-series can be reduced by treating all such individuals together
11 and the underlying high correlation of individuals with similar day-to-day exposure to PM of
12 non-ambient origin is lost, as was shown by Figure 7-26 of the previous PM AQCD (U.S.
13 Environmental Protection Agency, 1996). Janssen (1998) combined data from subjects with high
14 longitudinal correlations, by randomly sampling from each individual's data set and combining
15 them to create a cross-sectional data set. In each case, as shown in Table 5-7, the correlation
16 coefficients were significantly reduced. It is this process of combining sets of people with
17 assumed similar exposure to ambient PM but widely different exposures to non-ambient PM on
18 different days that leads to the published studies that report little or no statistically significant
19 correlation between personal exposure to the total PM and the corresponding ambient PM
20 concentrations (e.g., Sexton et al, 1984; Spengler et al, 1985; Pellizzari et al, 1999).
21
22
23 5.14 EXPOSURE TO AMBIENT GASEOUS POLLUTANTS RELATED
24 TO AMBIENT PM
25 Several gaseous pollutants, such as SO2, NO (rapidly converted to NO2) and CO, are often
26 emitted at the same time as ambient PM is emitted from combustion sources. In addition,
27 photochemical oxidants, such as ozone (O3) are also formed in the atmosphere by the same
28 processes that lead to formation of some of the photochemical aerosol species found in the fine-
29 mode PM fraction (Seinfeld and Pandis, 1998). Weather also introduces a correlation effect as a
30 stagnation inversion may cause all pollutant concentrations to rise, and a rain storm or frontal
31 passage may cause all pollutant concentrations to fall. Consequently, these gaseous pollutants
October 1999 5-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 may have ambient concentrations that are correlated with the concentrations of various fractions
2 of the ambient PM mixture [ultra-fine PM, fine-mode PM, or coarse-mode PM less than 10//m
3 AD].
4 If these gaseous species are capable of producing the same effect as an ambient PM
5 fraction, and their ambient concentrations are correlated with ambient PM concentration, it
6 becomes difficult to separate out which factor, if any, is causing the observed effect (Morris and
7 Naumova, 1998; Chen et al., 1999). It is assumed that the generation of the primary pollutants
8 CO, NO2 and SO2 in residential and occupational microenvironments is independent of ambient
9 concentrations of those gases. However, there may be some locations where indoor emissions
10 may be correlated with ambient concentrations as in the winter heating season. Assuming a
11 linear system and no differential measurement errors, the health effects of indoor generated CO,
12 NO2 and SO2 would also be independent of, and uncorrelated with, the health effects of the
13 ambient PM. If these assumptions are valid, the presence of indoor sources of these gases should
14 not confound the ambient PM mortality relationships reported in the literature.
15 An investigation (as yet unpublished) has been performed on the general correlation
16 behavior of ambient PM and the ambient gaseous criteria pollutants in the U.S. (Shadwick et al.,
17 1999). In order to examine the general geographic variability of the correlation coefficient, the
18 U.S. EPA public Aerometric Information Retrieval System (AIRS) data base for the years
19 1992 - 1996 was reviewed (http://www.epa.gov/airs/airs2.html). Shadwick et al. (1999) chose
20 the highest PM urban site and lowest rural PM site from each of the 50 states, Puerto Rico and
21 Washington B.C. irrespective of the years of available data. The distributions of correlations
22 between ambient PM10 and the 24-h concentrations of CO, NO2, and SO2, and ozone (8-h and
23 1-h), for urban and rural sites respectively, are shown in Figures 5-16a,b to 5-20a,b for the
24 chosen urban and rural sites. For PM25 only 38 sites were available for the entire U.S. so the
25 urban and rural stations were combined for a joint analysis. The results for ambient PM^ and
26 the gaseous species CO, NO2, SO2 and ozone (8-h and 1-h), for combined urban and rural sites
27 are shown in Figures 5-16c to 5-20c. These data are summarized in Table 5-15.
28 To estimate the relationship between exposure to ambient PM and exposures to the primary
29 gases of ambient origin, their hourly averages along with an air exchange rate estimate would be
30 required. For the highly reactive gases with strong sinks on indoor surfaces (such as ozone
October 1999 5-81 DRAFT-DO NOT QUOTE OR CITE
-------
A
9-
8-
7
0 Q_
£Z D
0) E_
3 5
5f 4-
0) H
LJ_ o
2-
1-
0-
Whitefish
i i i i
MN
1
|
Denver, CO
1
| |
i i i i i i
-07-06-05-04 -03-02-01 00 01 02 03 04 05 06 07 08 09
Urban Correlation Between PM10 and CO
15-
14-
13-
12-
11-
>, 10-
c 9-
0) 7-
5-
4-
3-
2-
1-
Columbia Falls, MN
1
1 1 1 1 1
| |
^_^^
—
1
— 1
^^ Central Point, OR
|
ll— 1
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rural Correlation Between PM10 and CO
10-
9-
8-
7
I1 «-
^ 5-
§- 4-
l_ o _
LJ_ °
2-
1-
^^^^^
_
Blue Island (Chicago), IL
i i i i i
i
— — 1 1
1 — 1
Mammoth Lake, CA
1
rnr~i
i i i i i i i
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Correlation Between PM,, and CO
Figure 5-16. (a) Correlations of PM10-CO for highest urban PM10 site per state;
(b) correlations of PM10-CO for lowest rural PM10 site per state; and
(c) correlations of PM2 5-CO.
Source: Shadwick et al. (1999).
October 1999
5-82
DRAFT-DO NOT QUOTE OR CITE
-------
11-
10-
9-
>, 8-
0) 6-
Ag- 5-
lt 4-
3-
2-
1-
Philadelphia, PA
1
U 1 1 1 1 II 1 1 1 1
—
Portland, OR
1
i i i i i
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Urban Correlation Between PMin and NO,
B
11-
10-
9-
8-
l°-
CD 6-
I 5-
LL 4-
3-
2-
1-
Christian County, KY
1
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.
3 0
1 Salt Lake County, UT
, , i
1
n
4 0.5 0.6 0.7 0.8 0.9
Rural Correlation Between PM10 and NO2
0
(D
11-
10-
9-
8_
7-
6-
5-
4-
3-
2-
1-
Columbia, SC
1
rnrn
^~
^
—
^_^_
Calexico, CA
|
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Correlation Between PM25 and NO2
Figure 5-17. (a) Correlations of PM10-NO2 for highest urban PM10 site per state;
(b) correlations of PM10-NO2 for lowest rural PM10 site per state; and
(c) correlations of PM2 5-NO2.
Source: Shadwick et al. (1999).
October 1999
5-83
DRAFT-DO NOT QUOTE OR CITE
-------
15-
14-
13-
12-
11-
10-
9-
8-
6-
5-
4-
3-
2-
1-
Philadelphia, PA
1
n r~i
1 1
—
—
l~~l — ~l
Sitka, AK
1
i—iH
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Urban Correlation Between PM10 and S02
11-
10-
9-
B| e-
-------
15-
14-
13-
12-
11-
>, 10-
c 9-
n 13 8-
Ac^ 7-
iV" 6-
u_ u
5-
4-
3-
2-
1-
Salt Lake City, UT 1 1
1 1 1
1 II
n 1 1
— — |
__
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0
__
Atlanta, GA
1
____
i i i i i iii
1 0.2 0.3 0.4 n 5 D R 07 n R 09
14 Urban Correlation Between PM10 and O3 Maximum 1-hr Average
13-
12-
11-
10-
O1 9-
g 8-
B §- 7-
CD 6-
5-
4-
3-
2-
1-
Iberville Parish, LA
I
nn
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0
1 1
—
Kent County, Rl
i
i
i i i i i i
1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
10
Rural Correlation Between PM10 and 03 Maximum 1-hr Average
Frequency
D_iK)W^UlC»-
1 1 1 1 1 1 1
Adams County, CO
i
nn
nf
n
n
Columbia, SC
1
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Correlation Between PM25 and Maximum 031-hr Average
Figure 5-19. (a) Correlations of PM10-O31-hr for highest urban PM10 site per state;
(b) correlations of PM10-O31-hr for lowest rural PM10 site per state and
(c) correlations of PM2 5-O31-hr.
Source: Shadwick et al. (1999).
October 1999
5-85
DRAFT-DO NOT QUOTE OR CITE
-------
A
(D
15-
14-
13-
12-
11-
10-
9-
8-
6-
5-
4-
3-
2-
1-
^^-^^
Salt Lake City, UT
1
nH
n
—
^^
Providence, Rl
1
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Urban Correlation Between PM10 and O3 Maximum 8-hr Average
15-f
14-
13-
12-
11-
B
cr
0)
7-
6-
5-
4-
3-
2-
1-
Kent County, Rl
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rural Correlation Between PM10 and O3 Maximum 8-hr Average
0)
10-
9-
8-
7-
5-
4-
3-
2-
1-
Ad
r
amsCoun
~|
ty, CC
fin
nfln
n
—
Columbia, SC
n
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Correlation Between PM25 and O3 Maximum 8-hr Average
Figure 5-20. (a) Correlations of PM10-O3 8-hr for highest urban PM10 site per state;
(b) correlations of PM10-O3 8-hr for lowest rural PM10 site per state; and
(c) correlations of PM2 5-O3 8-hr.
Source: Shadwick et al. (1999).
October 1999
5-86
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-15. AVERAGE ANNUAL CORRELATIONS BETWEEN PM AND
CRITERIA GASEOUS POLLUTANTS IN THE UNITED STATES (Std. Dev.)
Correlation PM vs Gas vs CO vs NO2 vs SO2 vs O3_lh vs O3_8h
Urban PM10 0.27(0.22) 0.41(0.21) 0.27(0.21) 0.31(0.25) 0.29(0.26)
Rural PM10 0.09(0.18) 0.25(0.21) 0.10(0.13) 0.49(0.17) 0.46(0.18)
Combined PM,, 0.44(0.29) 0.52(0.18) 0.20(0.20) 0.08(0.36) 0.03(0.38)
Source of Data: Shadwick et al. (1999).
1 which also oxidizes terpenes) their fractions of the ambient concentrations found indoors can be
2 of order zero. In the absence of hourly PM data and air exchange rate data the estimation of a
3 correlation between the ambient PM exposure and the exposure to the gases of ambient origin
4 cannot be made without introducing assumptions about the hourly variability of ambient PM and
5 air exchange rate. However, it is evident that the correlation between the personal exposures to
6 ambient PM and personal exposures to ambient ozone, NO2 and SO2 must be closer to zero than
7 the corresponding correlation of their respective atmospheric concentrations, because their
8 different indoor deposition rates will create different and variable fractional exposures.
9 Carbon monoxide (CO), however, is virtually non-reactive in indoor micro-environments
10 and it has no known surface sinks indoors. On average, a subject will be exposed to virtually
11 100% of the ambient CO concentration. Consequently the correlation between exposure to CO
12 of ambient origin and exposure to PM of ambient origin will closely approach the correlation of
13 the ambient PM concentration and the ambient CO concentration. However, the presence of
14 indoor sources of CO, such as tobacco smoking, and outdoor in-vehicle exposures to traffic CO,
15 will make the correlation between personal exposure to total CO and ambient PM much lower, in
16 the same way that indoor sources of PM degrade the correlation of total PM exposure and
17 ambient PM concentration.
18 Table 5-15 shows that on an annual basis PM10 is generally correlated positively with
19 almost all gaseous criteria pollutants. The urban correlations are higher than the rural
20 correlations for the daily averages of primary pollutants (CO, NO2 and SO2) and lower for the
21 secondary pollutant ozone (1-h and 8-h). Examination of the corresponding histograms, with
October 1999 5-87 DRAFT-DO NOT QUOTE OR CITE
-------
1 standard deviations of order 0.2, indicate that there is no general relationship; there are locations
2 that can have no significant correlation (of order zero) and appreciable correlation (of order 0.5).
3 Table 5-15 shows that PM25 is, on the average, most strongly correlated with CO and NO2,
4 moderately correlated with SO2, and not at all correlated with ozone in the combined urban and
5 rural locations we have analyzed.
6 The correlations for PM25 with ozone (maximum 1-h and maximum 8-h) for both urban
7 and rural sites combined has the largest variance of all the data sets. The mean correlation is
8 close to zero; however, the standard deviation of approximately 0.4 resulted from the fact that
9 some locations were highly positively correlated, while others had low negative correlations.
10 A possible explanation for a portion of this large variation in correlation between PM and ozone,
11 may arise from the difference in number of years of data at each site, and the influence of motor
12 vehicle traffic emissions of PM and NO. For example, Bernard et al. (1999) report that ozone
13 and NO2 in urban Montpellier, France had a highly negative correlation of r = -0.96,p < 0.01 in
14 the short time period studied (November 13 -20, 1995) as expected from the rapid gas phase
15 reduction of ozone by NO. Consequently, if the U.S. monitoring stations for both PM and ozone
16 are close to local traffic emissions that have the same diurnal pattern, then the PM may be a
17 surrogate for accompanying NO emissions that would reduce the correlation of ozone with the
18 measured PM.
19 These annual correlations between PM and criteria gaseous pollutants are relevant with
20 respect to epidemiologic studies which analyze PM and health data over complete years of
21 record. U.S. Environmental Protection Agency (1996) and Chen et al. (1999) discuss the
22 variations in these gas-PM correlations between the summer and winter periods. Because of the
23 different seasonal emissions and weather patterns, summer corresponds to a period of ozone
24 maxima, and the primary combustion products (CO, SO2) have their maxima during the winter,
25 which often leads to seasonal variations of correlation. Therefore the results cited above may not
26 be applicable to epidemiologic studies that analyze data by season, and a non-significant annual
27 correlation of order zero may mask the presence of significant positive and negative correlations
28 in the different seasons of the year.
29 In summary it appears that there is no general rule that applies to PM and the gaseous
30 criteria pollutants. The possibility of confounding of a relation to ambient PM concentration by a
October 1999 5-88 DRAFT-DO NOT QUOTE OR CITE
-------
1 gaseous ambient pollutant with a positive correlation in time to the ambient PM should be
2 considered in epidemiologic studies.
3
4
5 5.15 CONFOUNDING BY INDOOR PM
6 It has been hypothesized that human exposure to indoor generated PM may be correlated
7 with the ambient PM so that the health effects associated with the ambient PM concentration are
8 in fact caused by the indoor generated PM and not the ambient PM (Crandall et al., 1996; Vedal,
9 1997). "For example, if study subjects closed their windows on days with higher levels of
10 pollution, exposure to indoor pollutants might increase and actually be responsible for the
11 increase in adverse health outcomes, and therefore confound the particle and health association"
12 (Vedal, 1997). However this scenario may not apply to the situations in these studies because the
13 ambient PM was virtually always below the PM NAAQS, and there were no episode warnings
14 issued to the public to stay indoors to avoid high ambient PM pollution.
15 Hoek et al. (1989) reported respirable PM data (96% efficient collection of PM with a 4//m
16 AD) indoors and outdoors before-during-after a moderate air pollution episode in the
17 Netherlands in 1985, for which no smog warnings were issued. They found that indoor PM
18 increased and then fell similarly in homes with 0, 1 and 2 smokers, in phase with the increase in
19 ambient PM during the episode. Because Dutch homes with smokers have higher air exchange
20 rates than homes without smokers (Lebret, 1985), Hoek et al. (1989) reasoned that "the increases
21 in indoor concentrations observed during the episode were largely due to penetration of outdoor
22 air, and not due to decreased ventilation, leading to increased concentrations of pollutants
23 generated indoors, during the episode."
24 Mage et al. (1999) also tested this hypothesis of a decrease in air exchange rate leading to
25 higher indoor PM from indoor sources by using the PTEAM data set from Riverside, CA
26 (Clayton et al., 1993). When the PTEAM study was conducted in 1990, Riverside, CA was the
27 region of the country with the highest annual average concentration of ambient PM10. The
28 PTEAM study utilized a constant emission source of perfluorocarbon tracer (PFT) and integrated
29 charcoal PFT collectors for estimating air exchange rates. The measurements of the collected
30 PFT were used to estimate the average number of air exchanges per hour (a) in each home during
31 the resident's PM sampling period. Figure 5-21 shows that the measured air exchange rate had
October 1999 5-89 DRAFT-DO NOT QUOTE OR CITE
-------
0)
CD
0)
0)
I
o
X
LJJ
0
0
R2 = 0.001
.
* *
100
200
300
400
500
600
Outdoor PM Concentration at the Home
10
Figure 5-21. Air-exchange rate versus outdoor PM10 concentration at the home in the
PTEAM study.
Source: Mage et al. (1999).
1 no statistically significant relation to the ambient PM10 during the daytime (r = - 0.04).
2 Furthermore, each subject kept a time-activity diary so that the fraction of the time spent indoors
3 and outdoors has been computed for each person. Ozkaynak et al. (1996a) have determined the
4 deposition parameter (A=0.65/h) and the penetration parameter (P ~ 1) for PM10 from these very
5 same PTEAM data. Consequently we can estimate the exposure concentration of PM10 generated
6 by indoor sources and personal activities (Ei) as,
Ei = E - Co (a + ky) /(a + k)
(5-7)
Where E is the measured total personal exposure; Co (a + ky)/(a + K) is the estimated personal
exposure (Ea) to PM10 of ambient origin that also penetrates indoors (P ~ 1).
October 1999
5-90
DRAFT-DO NOT QUOTE OR CITE
-------
1 Figure 5-22 shows the quantity Ei for each subject vs the ambient concentration of PM10
2 (Co) that was measured outside their home at the same time. The regression has a slightly
3 negative slope which is not significantly different from zero. Note the negative values for Ei in
4 Figure 5-22 may be caused by the exposure and concentration measurement errors, and errors in
5 the estimated parameters P, a and k. This suggests, at least for that Riverside cohort, that the
6 personal exposure to a PM10 concentration of non-ambient origin does not appear to be positively
7 correlated with the ambient PM10 concentration measured at that same time. We propose that
8 this is the underlying general relationship between non-ambient generated PM and ambient PM
9 concentration.
10 In order for there to be a positive correlation between exposure to indoor generated PM (Ei)
11 and ambient PM (Co or Ca), people would have to smoke more cigarettes, dust and vacuum
12 more, cook more, and perform more hobby-type activities indoors on days with higher ambient
13 PM concentrations than on days with lower ambient PM concentrations. We expect that, where
14 air quality standards are being met, people make their decisions to perform these personal
15 activities without any conscious or unconscious consideration of the ambient PM concentration
16 which is unknown to them (Hoek et al., 1989). Thus, the amount of non-ambient PM that people
17 generate through their daily activities is expected to be independent of, and uncorrelated with, the
18 ambient PM concentration in their community.
19 Because the concentrations of PM due to such indoor sources were not observed to be
20 correlated with ambient PM concentration in either the Netherlands or Riverside, CA (the region
21 of the U.S. with the highest ambient PM concentration at the time of the PTEAM study), it is
22 reasoned that human exposure to non-ambient PM is likely to be independent of the
23 concentration of PM of ambient origin. Therefore it would be independent of the human
24 exposure to PM of ambient origin and not a confounder in the epidemiologic analysis. However,
25 these indoor generated PM species can be effects modifiers as discussed in the following
26 material.
27 Some PM species, and gaseous pollutants for which U.S. EPA has not established a
28 short-term National Ambient Air Quality Standard (NAAQS), may also play a role as acute effect
29 modifiers. Such a condition may occur with long-term exposure to a pollutant with a chronic
30 effect without any apparent acute effects. That chronic exposure may make a subject more
31 susceptible to the acute effect of a given exposure to PM of ambient origin. Chronic exposure to
October 1999 5-91 DRAFT-DO NOT QUOTE OR CITE
-------
500
o
Q.
X
LU
Q)
"03
E
400-
SOD-
CD
2
-------
1 cigarette smoke, and chronic exposure to radon progeny and asbestos in PM of indoor origin,
2 may cause malignant and non-malignant pulmonary disease which can make people more
3 susceptible to the acute effects of ambient PM than otherwise healthy individuals (United
4 Kingdom Ministry of Health, 1954). Therefore the daily concentration of such agents with
5 chronic toxicity may be a surrogate for an underlying susceptibility to acute effects of ambient
6 PM exposure in older subjects who lived in the monitored residence for a long period of time and
7 who developed an underlying pulmonary insufficiency.
8 There are also possible short-term effects of exposure to "nuisance and inert" PM,
9 occupationally generated PM, and unmonitored ambient PM in the upper portion of the coarse
10 mode of PM with an AD > 10 //m (see Figure 5-1). PM species with biogenic or allergenic
11 properties could irritate and sensitize the lung to the ambient PM10 (see Section 7.2 of U.S.
12 Environmental Protection Agency, 1996) or increase the residence time of ambient PM10in the
13 lung. For example, if alveolar macrophages removed non-ambient particles and ambient
14 particles of AD > 10 //m, instead of deposited ambient PM10 species, the rate of pulmonary
15 clearance of the deposited ambient PM10 particles would be decreased.
16 The presence of effect-modification on mortality and morbidity is difficult to discern except
17 perhaps in animal testing where controlled dose-response studies can be used to pick up
18 modifications of responses to ambient PM exposures in the presence and absence of candidate
19 agents. In epidemiologic studies, these effects would appear as statistically significant
20 interaction terms, but at the present time such interaction effects have not been reported (see
21 Chapter 12 of U.S. Environmental Protection Agency, 1996). The general-linearized-model
22 (GLM) described by Liang and Zeger (1986) has been used for most all the epidemiologic studies
23 discussed in Chapter 7 of this document. The GLM implies that the effect of exposure to
24 2 //g/m3 of ambient PM is twice the effect of exposure to 1 //g/m3 of the same mixture of ambient
25 PM, and the effect of an exposure to a mixture of 1 //g/m3 of ambient PM plus 1 //g/m3 of
26 non-ambient PM is equal to the effect of exposure to 1 //g/m3 of the same ambient PM mixture,
27 plus the effect of an exposure to 1 //g/m3 of the same non-ambient PM mixture.
28 Semi-volatile constituents of PM exist in the atmosphere in a dynamic equilibrium between
29 their vapor and condensed phases. For example, water vapor in an air mass at a given relative
30 humidity (RH) is in dynamic equilibrium with the water content of the aerosol PM within that air
31 mass. After a particle is collected, fluctuations in relative humidity cause fluctuations in the
October 1999 5-93 DRAFT-DO NOT QUOTE OR CITE
-------
1 amount of the water adsorbed on the particle, so consequently, the Federal Reference Method for
2 Particulate Matter (40CFR50 Appendices B, L, M, J) calls for filters to be weighed, following a
3 24-h dessication and equilibration period within a temperature and humidity controlled room, to
4 reduce the water content of the collected particles.
5 Other species, such as some inorganic nitrates and organic compounds, have high enough
6 vapor pressures, in the range of ambient temperatures, that they, like water, can coexist in both
7 the vapor and aerosol phase. Consequently, a variable fraction of such species may be lost from
8 the collected mass on filters during equilibration. Continuous monitors must also remove water
9 associated with particles (particle-bound water). For example, the TEOM® monitor is operated
10 with the filter at 50°C to remove particle-bound water. The TEOM® can record negative PM if
11 mass volatilizes from the filter collecting the PM, faster than new PM adheres to it.
12 Because ambient concentrations of these semi-volatile species are not currently measured,
13 and personal exposures to these constituents have not been reported in the literature, it is not
14 known whether these effervescent PM species could contribute to the health effects associated
15 with the non-volatile PM mass that is represented by the PM data in the epidemiology literature.
16 Although one might not expect the health effects associated with the non-volatile PM mass to be
17 caused by the unmeasured semi-volatile PM species, there may be a correlation between these
18 two masses so the captured non-volatile PM mass may in some aspects be a surrogate for the
19 mass of semi-volatile compounds that escaped from the collected PM.
20 In summary, this discussion supports the conclusion that the average concentrations of PM
21 measured at a community ambient monitoring station are reasonably good surrogate-estimates of
22 the average concentrations of PM of ambient origin to which people residing in that community
23 are exposed, more so for fine mode PM than coarse mode PM. This agreement between
24 exposure to ambient PM and ambient PM concentrations supports the plausibility of PM of
25 ambient origin, or a constituent thereof, being responsible for the fluctuations of health effects
26 that are correlated with fluctuations of ambient PM concentrations. The size distributions, the
27 chemical compositions, and the related toxicities per unit mass of ambient PM may vary from
28 city to city in a season, and from season to season in a city. In addition, different varieties of
29 climates between cities may influence the amount of time people spend outdoors and the air
30 exchange rates between their homes and the outdoor air, resulting in some cities having higher
31 percentages of the ambient PM penetrating into the residences than the others (Gamble, 1998).
October 1999 5-94 DRAFT-DO NOT QUOTE OR CITE
-------
1 The resulting inter-urban differences of exposure to PM of ambient origin may explain a portion
2 of the variance between the observed differences in mortality/morbidity per unit concentration of
3 ambient PM that are reported in the literature and are discussed in Chapter 7 of this document.
4
5
6 5.16 IMPLICATIONS OF THE AMBIENT PM EXPOSURE
7 RELATIONSHIPS FOR EPIDEMIOLOGIC ANALYSIS AND
8 A SUMMARY OF THE CHAPTER CONCLUSIONS
9 The main conclusion of this chapter is that (a) exposures to PM of ambient origin are highly
10 correlated with ambient PM concentrations; and; (b) exposures to emissions of PM of
11 non-ambient origin (e.g., tobacco smoke, residential activity, occupational activity) have very
12 low correlations with ambient PM concentrations. Therefore, it follows that the exposures to the
13 concentrations of PM produced by non-ambient sources (e.g., indoor and/or occupational) are
14 also uncorrelated with exposures to PM of ambient origin and the corresponding ambient PM
15 concentrations. Consequently the finding that personal exposures to total PM (both ambient
16 origin and non-ambient origin) are uncorrelated with ambient PM concentrations is not
17 important. This is because the acute health effects, if any, caused by the exposure to
18 non-ambient PM will have an equally low correlation with the acute health effects created by the
19 personal exposure to PM of ambient origin.
20 As described in Section 5.14, ambient PM concentrations in the U.S. have a wide range of
21 correlations in time with the ambient gaseous criteria pollutants that may produce or influence
22 cardiac and pulmonary effects. It appears that there is no a priori reason to neglect their ability
23 to confound the ambient PM vs health effect relationships. Therefore, these correlations need to
24 be evaluated at each location where an epidemiologic study is conducted to determine whether
25 they are significantly different from zero. The chapter conclusions are as follows:
26 1. Human exposure to PM of ambient origin for individuals in a community is often highly
27 correlated (R2 > 0.5) in time with concentrations of PM of ambient origin of the same size as
28 measured in that community.
29 2. The longitudinal correlation coefficient for the ambient concentration of fine PM
30 (AD < 2.5 //m) with personal exposure to ambient fine PM is greater than the corresponding
October 1999 5-95 DRAFT-DO NOT QUOTE OR CITE
-------
1 correlation for the coarse fraction of ambient PM (2.5 //m < AD < 10 //m) as shown by
2 studies of ambient sulfate concentrations and sulfate exposures.
3 3. People in a community surrounding an ambient monitoring station, over time, are exposed to
4 relatively similar mixtures and concentrations of ambient PM2 5.
5 4. People in a community are exposed to widely different mixtures and concentrations of
6 non-ambient PM due to the diversity of smoking habits, personal activities such as hobbies,
7 residential furnishings and appliances, and varying occupations.
8 5. Exposures to PM of indoor origin appear to be uncorrelated with exposures to PM of ambient
9 origin.
10 6. The correlation of a single individual's sequence of daily personal exposures to total PM and
11 ambient PM concentrations will be greater than the correlation that would occur had a
12 different person been monitored on each of the same days [e.g., one person monitored
13 consecutively for n days vs sequentially monitoring n different people, each for one day, over
14 ndays].
15 7. Ambient PM in the U.S. has average annual correlations with the ambient gaseous pollutants
16 CO, ozone, NO2, and SO2 of order r = 0.25 with a standard deviation of order 0.25.
17 8. Although exposures to PM from indoor sources and occupational activities may not be
18 correlated with ambient PM concentrations, these non-ambient PM species may possibly act
19 as effect modifiers by making subjects more or less susceptible to exposure to PM of ambient
20 origin.
21 9. There are only limited data available, from non-probability samples, to evaluate how well the
22 exposures to PM of ambient origin for susceptible subgroups correlates with the ambient PM
23 concentrations of similar AD size range as measured in their community.
24 In conclusion, day-to-day variation in the ambient concentration of fine particles is a good
25 surrogate for the day-to-day variation in the community average personal exposure to fine
26 particles of ambient origin. This relationship is not as clear for coarser particles (PM10 - PM2 5)
27 which do have more local sources. This supports the plausibility of ambient PM25 concentration
28 as a surrogate measure for personal exposure to PM25 of ambient origin in time series studies,
29 and provides the linkage necessary to more completely evaluate the impact of ambient PM
30 regulation under Title I of the Clean Air Act Amendments.
31
October 1999 5-96 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Abbey, D. E.; Nishino, N.; McDonnell, W. F.; Burchette, R. J.; Knutsen, S. F.; Beeson, W. L.; Yang, J. X. (1999)
3 Long-term inhalable particles and other air pollutants related to mortality in nonsmokers. Am. J. Respir. Crit.
4 Care Med. 159:373-382.
5 Abt, E.; Suh, H. H.; Allen, G.; Koutrakis, P. (1999a) Characterization of indoor particle sources: a study conducted
6 in the metropolitan Boston area. Environ. Health Perspect: submitted.
7 Abt, E.; Suh, H. H.; Catalano, P.; Koutrakis, P. (1999b) The relative contribution of outdoor and indoor particle
8 sources to indoor concentrations. Environ. Sci. Technol.: submitted.
9 Adams, W. C. (1993) Measurement of breathing rate and volume in routinely performed daily activities [final
10 report]. Sacramento, CA: California Environmental Protection Agency, Air Resources Board; contract no.
11 A033-205.
12 Aim, S.; Jantunen, M. J.; Vartiainen, M. (1999) Urban commuter exposure to particle matter and carbon monoxide
13 inside an automobile. J. Exposure Anal. Environ. Epidemiol. 9: 237-244.
14 Alzona, J.; Cohen, B. L.; Rudolph, H.; Jow, H. N.; Frohliger, J. O. (1979) Indoor-outdoor relationships for airborne
15 particulate matter of outdoor origin. Atmos. Environ. 13: 55-60.
16 American Conference of Governmental Industrial Hygienists. (1996) Threshold limit values for chemical
17 substances and physical agents: biological exposure indices. Cincinnati, OH: American Conference of
18 Governmental and Industrial Hygienists.
19 Anuszewski, J.; Larson, T. V.; Koenig, J. Q. (1998) Simultaneous indoor and outdoor particle light-scattering
20 measurements at nine homes using a portable nephelometer. J. Exposure Anal. Environ. Epidemiol. 8:
21 483-493.
22 Armstrong, B. K.; White, E.; Saracci, R., eds. (1992) Principles of exposure measurement in epidemiology. New
23 York, NY: Oxford University Press. (Monographs in epidemiology and biostatistics: v. 21).
24 Ayers, G. P.; Keywood, M. D.; Gras, J. L. (1999) TEOM vs. manual gravimetric methods for determination of
25 PM2.5 aerosol mass concentrations. Atmos. Environ. 33: 3717-3721.
26 Bahadori, T. (1998) Human exposure assessment: relationship between outdoor, indoor, and personal measurements
27 [dissertation]. Boston, MA: Harvard University.
28 Bahadori, T.; Suh, H. H.; Rojas-Bracho, L.; Koutrakis, P. (1999) Personal exposure to particulate matter of
29 individuals with chronic obstructive pulmonary disease (COPD). Submitted.
30 Bernard, N. L.; Gerber, M. J.; Astre, C. M.; Saintot, M. J. (1999) Ozone measurement with passive samplers:
31 validation and use for ozone pollution assessment in Montpellier, France. Environ. Sci. Technol. 33:
32 217-222.
33 Beyea, J.; Hatch, M. (1999) Geographic exposure modeling: a valuable extension of geographic information
34 systems for use in environmental epidemiology. Environ. Health Perspect. Suppl. 107(1): 181-190.
35 Binder, R. E.; Mitchell, C. A.; Hosein, H. R.; Bouhuys, A. (1976) Importance of the indoor environment in air
36 pollution exposure. Arch. Environ. Health 31: 277-279.
37 Bohne, J. E., Jr.; Cohen, B. S. (1985) Aerosol resuspension from fabric: implications for personal monitoring in the
38 beryllium industry. Am. Ind. Hyg. Assoc. J. 46: 73-79.
39 Brauer, M.; Koutrakis, P.; Spengler, J. D. (1989) Personal exposures to acidic aerosols and gases. Environ. Sci.
40 Technol. 23: 1408-1412.
41 Brunekreef, B. (1999) All but quiet on the particulate front [editorial]. Am. J. Respir. Crit. Care Med. 159: 354-356.
42 Burton, R. M.; Suh, H. H.; Koutrakis, P. (1996) Spatial variation in particulate concentrations within metropolitan
43 Philadelphia. Environ. Sci. Technol. 30: 400-407.
44 Buzorius, G.; Hameri, K.; Pekkanen, J.; Kulmala, M. (1999) Spatial variation of aerosol number concentration in
45 Helsinki city. Atmos. Environ. 33: 553-565.
46 California Environmental Protection Agency. (1998) Measuring concentrations of selected air pollutants inside
47 California vehicles [final report]. Sacramento, CA: Air Resources Board, Research Division; contract no.
48 95-339
49 Carrothers, T. J.; Evans, J. S. (1999) Assessing the impact of differential measurement error on estimates of fine
50 particle mortality. J. Air Waste Manage. Assoc.: accepted.
51 Cha, Q.; Chen, Y.-Z.; Du, Y.-X. (1996) Influence of indoor air pollution on quality of outdoor air in city of
52 Guangzhou, China. In: Indoor air '96: the 7th international conference on indoor air quality, volume I; July;
53 Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 1131-1135.
October 1999 5-97 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chen, C.; Chock, D. P.; Winkler, S. L. (1999) A simulation study of confounding in generalized linear models for
2 air pollution epidemiology. Environ. Health Perspect. 107: 217-222.
3 Clayton, C. A.; Perritt, R. L.; Pellizzari, E. D.; Thomas, K. W.; Whitmore, R. W.; Wallace, L. A.; Ozkaynak, H.;
4 Spengler, J. D. (1993) Particle total exposure assessment methodology (PTEAM) study: distributions of
5 aerosol and elemental concentrations in personal, indoor, and outdoor air samples in a southern California
6 community. J. Exposure Anal. Environ. Epidemiol. 3: 227-250.
7 Code of Federal Regulations. (1994) National primary and secondary ambient air quality standards; definitions. C.
8 F.R. 40: part 50, §50.1.
9 Code of Federal Regulations. (1998) Occupational safety and health standards, subpart Z—toxic and hazardous
10 substances. Table Z-3, mineral dusts. C. F. R. 29: §1910.1000.
11 Cohen, B. S.; Positano, R. (1986) Resuspension of dust from work clothing as a source of inhalation exposure. Am.
12 Ind. Hyg. Assoc. J. 47: 255-258.
13 Crandall, R. W.; Rueter, F. H.; Steger, W. A. (1996) Clearing the air: EPA's self-assessment of clean-air policy.
14 Regulation 19(4): 35-46.
15 Delfino, R. J.; Zeiger, R. S.; Seltzer, J. M.; Street, D. H. (1998) Symptoms in pediatric asthmatics and air pollution:
16 differences in effects by symptom severity, anti-inflammatory medication use and particulate averaging time.
17 Environ. Health Perspect. 106: 751-761.
18 Duan, N. (1982) Models for human exposure to air pollution. Environ. Int. 8: 305-309.
19 Dubowsky, S. D.; Wallace, L.; Buckley, T. J. (1999) The contribution of traffic to indoor concentrations of
20 polycyclic aromatic hydrocarbons (PAH). J. Exposure Anal. Environ. Epidemiol.: accepted.
21 Federal Trade Commission. (1994) Tar, nicotine, and carbon monoxide of the smoke of 93 3 varieties of domestic
22 cigarettes. Washington, DC: Federal Trade Commission.
23 Gamble, J. F. (1998) PM2 5 and mortality in long-term prospective cohort studies: cause-effect or statistical
24 associations? Environ. Health Perspect. 106: 535-549.
25 Gamble, J. (1999) Reply to Kunzli and Tager regarding causality in PM 2 5 cohort studies [author's reply]. Environ.
26 Health Perspect. 107: A235-A236.
27 Hoek, G.; Brunekreef, B.; Hofschreuder, P. (1989) Indoor exposure to airborne particles and nitrogen dioxide
28 during an air pollution episode. JAPCA 39: 1348-1349.
29 Janssen, N. (1998) Personal exposure to airborne particles: validity of outdoor concentrations as a measure of
30 exposure in time series studies [dissertation]. Wageningen, Netherlands: Landbouwuniversiteit Wageningen.
31 Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1995) A relation between personal and ambient PM 10.
32 Epidemiology 6(suppl.): S45.
33 Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1997) Childhood exposure to PM 10: relation between
34 personal, classroom, and outdoor concentrations. Occup. Environ. Med. 54: 888-894.
35 Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H.; Mensink, L; Zuidhof, A. (1998a) Personal sampling of
36 particles in adults: relation among personal, indoor, and outdoor air concentrations. Am. J. Epidemiol. 147:
37 537-547.
38 Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1998b) Personal sampling of airborne particles: method
39 performance and data quality. J. Exposure Anal. Environ. Epidemiol. 8: 37-49.
40 Janssen, L. H. J. M.; Buringh, E.; van der Meulen, A.; van den Hout, K. D. (1999a) A method to estimate the
41 distribution of various fractions of PM10 in ambient air in the Netherlands. Atmos. Environ. 33: 3325-3334.
42 Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1999b) Personal exposure to fine particles in children
43 correlates closely with ambient fine particles. Arch. Environ. Health 54: 95-101.
44 Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H. (1999c) Mass concentration and elemental composition
45 of PM10 in classrooms. Occup. Environ. Med. 56: 482-487.
46 Jantunen, M. J.; Hanninen, O.; Katsouyanni, K.; Knoppel, H.; Kuenzli, N.; Lebret, E.; Maroni, M.; Saarela, K.;
47 Sram, R.; Zmirou, D. (1998) Air pollution exposure in European cities: the " EXPOLIS" study. J. Exposure
48 Anal. Environ. Epidemiol. 8: 495-518.
49 Jantunen et al. (2000) [Manuscript in preparation.]
50 Jedrychowski, W.; Flak, E. (1998) Separate and combined effects of the outdoor and indoor air quality on chronic
51 respiratory symptoms adjusted for allergy among preadolescent children. Int. J. Occup. Med. Environ. Health
52 11:19-35.
53 Jenkins, R. A.; Palausky, A.; Counts, R. W.; Bayne, C. K.; Dindal, A. B.; Guerin, M. R. (1996a) Exposure to
54 environmental tobacco smoke in sixteen cities in the United States as determined by personal breathing zone
55 air sampling. J. Exposure Anal. Environ. Epidemiol. 6: 473-502.
October 1999 5-98 DRAFT-DO NOT QUOTE OR CITE
-------
1 Jenkins, R. A.; Palausky, M. A.; Counts, R. W.; Guerin, M. R.; Dindal, A. B.; Bayne, C. K. (1996b) Determination
2 of personal exposure of non-smokers to environmental tobacco smoke in the United States. Lung Cancer
3 14(suppl. 1): S195-S213.
4 Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of a-pinene and ozone
5 using a gas-phase kinetics-aerosol partitioning model. Environ. Sci. Technol. 33: 1430-1438.
6 Keary, J.; Jennings, S. G.; O'Connor, T. C.; McManus, B.; Lee, M. (1998) PM 10 concentration measurements in
7 Dublin City. Environ. Monitor. Assess. 52: 3-18.
8 Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
9 segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
10 Environ. 33:2907-2913.
11 Kildes0, J.; Vallarino, J.; Spengler, J. D.; Brightman, H. S.; Schneider, T. (1999) Dust build-up on surfaces in the
12 indoor environment. Atmos. Environ. 33: 699-707.
13 Kleeman, M. J.; Cass, G. R. (1998) Source contributions to the size and composition distribution of urban
14 particulate air pollution. Atmos. Environ. 32: 2803-2816.
15 Klepeis, N. E.; Ott, W. R.; Switzer, P. (1996) A multiple-smoker model for predicting indoor air quality in public
16 lounges. Environ. Sci. Technol. 30: 2813-2820.
17 Klepeis, N. E.; Nelson, W. C.; Tsang, A. M.; Robinson, J. P. Hern, S. C.; Engelmann, W. H.; Behar, J. V. (1999)
18 The national human activity pattern survey (NHAPS): data collection methodology and selected results. J.
19 Exposure Anal. Environ. Epidemiol.: submitted.
20 Koistinen, K. J.; Kousa, A.; Tenhola, V.; Hanninen, O.; Jantunen, M. J. (1999) Fine particle (PM 25) measurement
21 methodology, quality assurance procedures and pilot results of the EXPOLIS study. J. Air Waste Manage.
22 Assoc.: accepted.
23 Kotchmar, D. J.; McMullen, T.; Hasselblad, V. (1987) Adequacy of a single monitoring site for defining mean
24 outdoor concentrations of fine particles in a demarcated residential community. JAPCA 37: 3 77-3 81.
25 Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. (1992) Source apportionment of indoor aerosols in Suffolk and
26 Onondaga Counties, New York. Environ. Sci. Technol. 26: 521-527.
27 Kulmala, M.; Asmi, A.; Pirjola, L. (1999) Indoor air aerosol model: the effect of outdoor air, filtration and
28 ventilation on indoor concentrations. Atmos. Environ. 33: 2133-2144.
29 Kiinzli, N.; Tager, I. B. (1999) Comments on "PM 2 5 and mortality in long-term prospective cohort studies:
30 cause-effect or statistical associations?" Environ. Health Perspect. 107: A234-A235.
31 Larssen, S.; T0nnesen, D.; Clench-Aas, J.; Aarnes, M. J.; Arnesen, K. (1993) A model for car exhaust exposure
32 calculations to investigate health effects of air pollution. Sci. Total Environ. 134: 51-60.
33 Leaderer, B. P.; Naeher, L.; Jankun, T.; Balenger, K.; Holford, T. R.; Toth, C.; Sullivan, J.; Wolfson, J. M.;
34 Koutrakis, P. (1999a) Indoor, outdoor, and regional summer and winter concentrations of PM 10, PM25, SO42",
35 H+, NH4+, NO3", NH3, and nitrous acid in homes with and without kerosene space heaters. Environ. Health
36 Perspect. 107:223-231.
37 Leaderer, B. P.; Balenger, K.; Naeher, L.; Triche, E. W.; Hellenbrand, K.; Holford, T. R.; Zheng, H.; Bracken, M.
38 B.; Beckett, W. S. (1999b) Daily summer respiratory symptoms in infants associated with exposure to PM 10,
39 PM2.5, sulfate and ozone. Submitted.
40 Leaderer, B. P.; Balenger, K; Triche, E. W.; Hellenbrand, K.; Holford, T. R.; Zheng, H.; Bracken, M. B.; Beckett,
41 W. S. (1999c) Daily winter respiratory symptoms in infants associated with exposure to particulate and
42 sulfate air pollution. Submitted.
43 Lebret, E. (1985) Air pollution in Dutch homes: an exploratory study in environmental epidemiology. Wageningen,
44 The Netherlands: Department of Air Pollution, Department of Environmental and Tropical Health; report
45 R-138; report 1985-221.
46 Lessler, J. T.; Kalsbeek, W. D. (1992) Nonsampling error in surveys. New York, NY: John Wiley & Sons, Inc.;
47 p. 7.
48 Liang, K-Y.; Zeger, S. L. (1986) Longitudinal data analysis using generalized linear models. Biometrika 73: 13-22.
49 Lillquist, D. R.; Lee, J. S.; Ramsay, J. R.; Boucher, K. M.; Walton, Z. L.; Lyon, J. L. (1998) A comparison of
50 indoor/outdoor PM 10 concentrations measured at three hospitals and a centrally located monitor in Utah.
51 Appl. Occup. Environ. Hyg. 13: 409-415.
52 Lioy, P. J.; Waldman, J. M.; Buckley, T.; Butler, J.; Pietarinen, C. (1990) The personal, indoor and outdoor
53 concentrations of PM-10 measured in an industrial community during the winter. Atmos. Environ. Part B 24:
54 57-66.
October 1999 5-99 DRAFT-DO NOT QUOTE OR CITE
-------
1 Litzistorf, G.; Guillemin, M. P.; Buffat, P.; Iselin, F. (1985) Influence of human activity on the airborne fiber level
2 in paraoccupational environments. J. Air Pollut. Control Assoc. 35: 836-837.
3 Mage, D. T. (1980) The statistical form for the ambient particulate standard annual arithmetic mean vs annual
4 geometric mean. J. Air Pollut. Control Assoc. 30: 796-798.
5 Mage, D. T. (1983) Public health aspects of air quality surveillance. Public Health Rev. 11: 5-54.
6 Mage, D. T. (1985) Concepts of human exposure assessment for airborne particulate matter. Environ. Int. 11:
7 407-412.
8 Mage, D. T. (1998) How much PM of ambient origin are people exposed to? In: Measurement of toxic and related
9 air pollutants. Volume II: proceedings of a specialty conference; September; Gary, NC. Pittsburgh, PA: Air &
10 Waste Management Association; pp. 666-681. (A&WMA publication VIP-85).
11 Mage, D. T.; Buckley, T. J. (1995) The relationship between personal exposures and ambient concentrations of
12 particulate matter. Presented at: 88th annual meeting of the Air & Waste Management Association; June; San
13 Antonio, TX. Pittsburgh, PA: Air & Waste Management Association; paper no. 95-MP18.01.
14 Mage, D. T.; Ott, W. R. (1996) Accounting for nonuniform mixing and human exposure in indoor environments. In:
15 Tichenor, B. A., ed. Characterizing sources of indoor air pollution and related sink effects. Philadelphia, PA:
16 American Society for Testing and Materials; pp. 263-278.
17 Mage, D. T.; Wilson, W.; Hasselblad, V.; Grant, L. (1999) Assessment of human exposure to ambient particulate
18 matter. J. Air Waste Manage. Assoc.: in press.
19 Marple, V. A.; Rubow, K. L.; Turner, W.; Spengler, J. D. (1987) Low flow rate sharp cut impactors for indoor air
20 sampling: design and calibration. JAPCA 37: 1303-1307.
21 Matthews, T. G.; Thompson, C. V.; Wilson, D. L.; Hawthorne, A. R.; Mage, D. T. (1989) Air velocities inside
22 domestic environments: an important parameter in the study of indoor air quality and climate. Environ. Int.
23 15: 545-550.
24 McBride, S. J.; Ferro, A.; Ott, W.; Switzer, P.; Hildemann, L. M. (1999) Investigations of the proximity effect for
25 pollutants in the indoor environment. J. Exposure Anal. Environ. Epidemiol.: accepted.
26 Michaels, R. A. (1997) Particulate matter policy [letter]. Science (Washington, DC) 278: 1,696.
27 Michaels, R. A. (1998) Permissible daily airborne particle mass levels encompass brief excursions to the "London
28 Fog" range, which may contribute to daily mortality and morbidity in communities. Appl. Occup. Environ.
29 Hyg. 13:385-394.
30 Monn, C. (1999) [Personal communication with D. Mage concerning article appearing in Science of the Total
31 Environment 208: 15-21 (1997)]. Zurich, Switzerland: Institute for Hygiene und Applied Physiology; June 2.
32 Monn, C.; Junker, M. (1999) Indoor/outdoor ratios, deposition and resuspension of fine and coarse particles. In:
33 Indoor air '99; August; Edinburgh, United Kingdom.
34 Monn, C.; Fuchs, A.; Hogger, D.; Junker, M.; Kogelschatz, D.; Roth, N.; Wanner, H.-U. (1997) Particulate matter
35 less than 10 ^ (PM10) and fine particles less than 2.5 ^ (PM2 5): relationships between indoor, outdoor and
36 personal concentrations. Sci. Total Environ. 208: 15-21.
37 Morris, R. D.; Naumova, E. N. (1998) Carbon monoxide and hospital admissions for congestive heart failure:
38 evidence of an increased effect at low temperatures. Environ. Health Perspect. 106: 649-653.
39 Murray, D. M.; Burmaster, D. E. (1995) Residential air exchange rates in the United States: empirical and estimated
40 parametric distributions by season and climatic region. Risk Anal. 15: 459-465.
41 National Air Pollution Control Administration. (1969) Air quality criteria for particulate matter. Washington, DC:
42 U.S. Department of Health, Education, and Welfare, Public Health Service; NAPCA publication no. AP-49.
43 Available from: NTIS, Springfield, VA; PB-190251/BA.
44 National Research Council. (1991) Human exposure assessment for airborne pollutants: advances and opportunities.
45 Washington, DC: National Academy of Sciences.
46 Neas, L. M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr.; Speizer, F. E. (1994) Concentration of
47 indoor particulate matter as a determinant of respiratory health in children. Am. J. Epidemiol. 139:
48 1088-1099.
49 Nieuwenhuijsen, M. J.; Noderer, K. S.; Schenker, M. B.; Vallyathan, V.; Olenchock, S. (1999) Personal exposure to
50 dust, endotoxin and crystalline silica in California agriculture. Ann. Occup. Hyg. 43: 35-42.
51 Ogden, T. L.; Bartlett, I. W.; Purnell, C. J.; Wells, C. J.; Armitage, F.; Wolfson, H. (1993) Dust for cotton
52 manufacture: changing from static to personal sampling. Ann. Occup. Hyg. 37: 271-285.
53 Ormstad, H.; Gaarder, P. L; Johansen, B. V. (1997) Quantification and characterisation of suspended particulate
54 matter in indoor air. Sci. Total Environ. 193: 185-196.
55 Ott, W. R.; Roberts, J. W. (1998) Everyday exposure to toxic pollutants. Sci. Am. 278(2): 86-91.
October 1999 5-100 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ott, W.; Langan, L.; Switzer, P. (1992) A time series model for cigarette smoking activity patterns: model validation
2 for carbon monoxide and respirable particles in a chamber and an automobile. In: Measuring, understanding,
3 and predicting exposures in the 21st century: proceedings of the conference; November 1991; Atlanta, GA. J.
4 Exposure Anal. Environ. Epidemiol. 2(suppl. 2): 175-200.
5 Owen, M. K.; Ensor, D. S.; Sparks, L. E. (1992) Airborne particle sizes and sources found in indoor air. Atmos.
6 Environ. Part A 26: 2149-2162.
7 Ozkaynak, H.; Xue, J.; Weker, R.; Butler, D.; Koutrakis, P.; Spengler, J. (1996a) The particle TEAM (PTEAM)
8 study: analysis of the data: final report, volume III. Research Triangle Park, NC: U.S. Environmental
9 Protection Agency, Atmospheric Research and Exposure Assessment Laboratory; report no.
10 EPA/600/R-95/098. Available from: NTIS, Springfield, VA; PB97-102495.
11 Ozkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins, P. (1996b) Personal exposure to airborne
12 particles and metals: results from the particle TEAM study in Riverside, California. J. Exposure Anal.
13 Environ. Epidemiol. 6: 57-78.
14 Park, J.-H.; Spengler, J. D.; Yoon, D.-W.; Dumyahn, T.; Lee, K.; Ozkaynak, H. (1998) Measurement of air
15 exchange rate of stationary vehicles and estimation of in-vehicle exposure. J. Exposure Anal. Environ.
16 Epidemiol. 8: 65-78.
17 Pellizzari, E. D.; Clayton, C. A.; Rodes, C. E.; Mason, R. E.; Piper, L. L.; Fort, B.; Pfeifer, G.; Lynam, D. (1999)
18 Particulate matter and manganese exposures in Toronto, Canada. Atmos. Environ. 33: 721-734.
19 Phillips, K.; Howard, D. A.; Browne, D.; Lewsley, J. M. (1994) Assessment of personal exposures to environmental
20 tobacco smoke in British nonsmokers. Environ. Int. 20: 693-712.
21 Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G. (1996) Assessment of air quality in Stockholm by personal
22 monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Scand. J.
23 Work Environ. Health 22(suppl. 1): 1-24.
24 Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G.; Huici, A. (1997a) Assessment of air quality in Barcelona by
25 personal monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke.
26 Environ. Int. 23: 173-196.
27 Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1997b) Assessment of air quality in Turin by personal
28 monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
29 23:851-871.
30 Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G. (1998a) Assessment of air quality in Paris by personal
31 monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
32 24:405-425.
33 Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1998b) Assessment of environmental tobacco smoke and
34 respirable suspended particle exposures for nonsmokers in Lisbon by personal monitoring. Environ. Int. 24:
35 301-324.
36 Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1998c) Measured exposures by personal monitoring for
37 respirable suspended particles and environmental tobacco smoke of housewives and office workers resident
38 in Bremen, Germany. Int. Arch. Occup. Environ. Health 71: 201-212.
39 Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1999) Assessment of environmental tobacco smoke and
40 respirable suspended particle exposures for nonsmokers in Basel by personal monitoring. Atmos. Environ.
41 33: 1889-1904.
42 Pope, C. A.; Hill, R. W.; Villegas, G. M. (1999) Particulate air pollution and daily mortality on Utah's Wasatch
43 Front. Environ. Health Perspect: submitted.
44 Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment approaches to evaluate
45 respiratory health effects of particulate matter and nitrogen dioxide. J. Exposure Anal. Environ. Epidemiol. 1:
46 83-107.
47 Raunemaa, T.; Kulmala, M.; Saari, H.; Olin, M.; Kulmala, M. H. (1989) Indoor air aerosol model: transport indoors
48 and deposition of fine and coarse particles. Aerosol Sci. Technol. 11: 11-25.
49 Repace, J. L.; Lowrey, A. H. (1980) Indoor air pollution, tobacco smoke, and public health. Science (Washington,
50 DC) 208: 464-472.
51 Rojas-Bracho, L.; Suh, H. H.; Koutrakis, P. (1999) Relationship among personal, indoor, and outdoor fine and
52 coarse particle concentrations for individuals with COPD. Boston, MA: Harvard University, School of Public
53 Health.
54 Roorda-Knape, M. C.; Janssen, N. A. H.; De Hartog, J. J.; Van Vliet, P. H. N.; Harssema, H.; Brunekreef, B. (1998)
55 Air pollution from traffic in city districts near major motorways. Atmos. Environ. 32: 1921-1930.
October 1999 5-101 DRAFT-DO NOT QUOTE OR CITE
-------
1 Scaperdas, A.; Colvile, R. N. (1999) Assessing the representativeness of monitoring data from an urban intersection
2 site in central London, UK. Atmos. Environ. 33: 661-674.
3 Schwartz, J.; Norris, G.; Larson, T.; Sheppard, L.; Claiborne, C.; Koenig, J. (1999) Episodes of high coarse particle
4 concentrations are not associated with increased mortality. Environ. Health Perspect. 107: 339-342.
5 Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change. New
6 York, NY: John Wiley & Sons, Inc.
7 Sexton, K.; Spengler, J. D.; Treitman, R. D. (1984) Personal exposure to respirable particles: a case study in
8 Waterbury, Vermont. Atmos. Environ. 18: 1385-1398.
9 Shadwick, D.; Glen, G.; Smith, L. (1999) Correlations of particulate matter with gaseous pollutants. Research
10 Triangle Park, NC: U.S. Environmental Protection Agency; National Exposure Research Laboratory; Office
11 of Research and Development; October 1998.
12 Siegmann, K.; Scherrer, L.; Siegmann, H. C. (1999) Physical and chemical properties of airborne nanoscale
13 particles and how to measure the impact on human health. J. Mol. Struct. (Theochem) 458: 191-201.
14 Smith, K. R.; Apte, M. G.; Ma, Y.; Wongsekiarttirat, W.; Kulkarni, A. (1994) Air pollution and the energy ladder in
15 Asian cities. In: Tyler, S.; Sathaye, J.; Goldman, N., eds. Energy use in Asian Cities. Energy (Oxford) 19:
16 587-600.
17 Spengler, J. D.; Dockery, D. W.; Reed, M. P.; Tosteson, T.; Quinlan, P. (1980) Personal exposures to respirable
18 particles. Presented at: 73rd annual meeting of the Air Pollution Control Association; June; Montreal, PQ,
19 Canada. Pittsburgh, PA: Air Pollution Control Association; paper no. 80-61.5b.
20 Spengler, J. D.; Treitman, R. D.; Tosteson, T. D.; Mage, D. T.; Soczek, M. L. (1985) Personal exposures to
21 respirable particulates and implications for air pollution epidemiology. Environ. Sci. Technol. 19: 700-707.
22 Sterling, T. D.; Rosenbaum, W. L.; Weinkam, J. J. (1995) Concentration of indoor particulate matter as a
23 determinant of respiratory health in children [letter]. Am. J. Epidemiol. 141: 581-582.
24 Stewart, P. A.; Rice, C. (1990) A source of exposure data for occupational epidemiology studies. Appl. Occup.
25 Environ. Hyg. 5: 359-363.
26 Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993) Validation of personal exposure models for sulfate and aerosol
27 strong acidity. J. Air Waste Manage. Assoc. 43: 845-850.
28 Tamura, K.; Ando, M.; Sagai, M.; Matsumoto, Y. (1996a) Estimation of levels of personal exposure to suspended
29 particulate matter and nitrogen dioxide in Tokyo. Environ. Sci. (Tokyo) 4: 37-51.
30 Tamura, K.; Ando, M.; Miyazaki, T.; Funasaka, K.; Kuroda, K. (1996b) Indoor concentrations and personal
31 exposure to particulate matter at roadside houses. In: Indoor air '96: the 7th international conference on
32 indoor air quality, volume I; July; Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 599-602.
33 Teschke, K.; Marion, S. A.; Vaughan, T. L.; Morgan, M. S.; Camp, J. (1999) Exposures to wood dust in U.S.
34 industries and occupations, 1979 to 1997. Am. J. Ind. Med. 35: 581-589.
35 Thatcher, T. L.; Layton, D. W. (1995) Deposition, resuspension, and penetration of particles within a residence.
36 Atmos. Environ. 29: 1487-1497.
37 Trijonis, J.; Eldon, J.; Gins, J.; Berglund, G. (1980) Analysis of the St. Louis RAMS ambient particulate data.
38 Volume I: final report. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air
39 Quality Planning and Standards; report no. EPA-450/4-80-006a. Available from: NTIS, Springfield, VA;
40 PB80-203359.
41 Tung, T. C. W.; Chao, C. Y. H.; Burnett, J. (1999) A methodology to investigate the particulate penetration
42 coefficient through building shell. Atmos. Environ. 33: 881-893.
43 U.S. Environmental Protection Agency. (1982) Air quality criteria for particulate matter and sulfur oxides. Research
44 Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
45 Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
46 U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
47 NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
48 Available from: NTIS, Springfield, VA; PB96-168224.
49 United Kingdom Ministry of Health. (1954) Mortality and morbidity during the London fog of December 1952.
50 London, United Kingdom: Her Majesty's Stationery Office. (Reports on public health and medical subjects
51 no. 95).
52 Vakeva, M.; Hameri, K.; Kulmala, M.; Lahdes, R.; Ruuskanen, J.; Laitinen, T. (1999) Street level versus rooftop
53 concentrations of submicron aerosol particles and gaseous pollutants in an urban street canyon. Atmos.
54 Environ. 33: 1385-1397.
55 Vedal, S. (1997) Ambient particles and health: lines that divide. J. Air Waste Manage. Assoc. 47: 551-581.
October 1999 5-102 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wallace, L. (1996) Indoor particles: a review. J. Air Waste Manage. Assoc. 46: 98-126.
2 Wallace, L. (1999a) Correlations of personal exposure to particles with outdoor air measurements: a review of
3 recent studies. Aerosol. Sci. Technol.: accepted.
4 Wallace, L. (1999b) Penetration of PM [email to D. Mage]. Reston, VA: U.S. Environmental Protection Agency;
5 September 1.
6 Wallace, L.; Quackenboss, J.; Rodes, C. (1997) Continuous measurements of particles, PAH, and CO in an
7 occupied townhouse in Reston, VA. In: Measurement of toxic and related air pollutants. Volume II:
8 proceedings of a specialty conference; April-May; Research Triangle Park, NC. Pittsburgh, PA: Air & Waste
9 Management Association, pp. 860-871. (A&WMA publication VIP-74).
10 Weschler, C. J.; Shields, H. C. (1999) Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
11 Environ. 33:2301-2312.
12 White, W. H. (1998) Statistical considerations in the interpretation of size-resolved particulate mass data. J. Air
13 Waste Manage. Assoc. 48: 454-458.
14 Williams, R. W.; Creason, J.; Shy, C.; Zweidinger, R.; Watts, R. R. (1999a) Indoor, outdoor and personal exposure
15 monitoring of particulate air pollution: the Baltimore elderly epidemiology-exposure pilot study. Submitted.
16 Williams, R. W.; Watts, R. R.; Stevens, R. K.; Stone, C. L.; Lewtas, J. (1999b) Evaluation of a personal air sampler
17 for twenty-four hour collection of fine particles and semivolatile organics. J. Exposure Anal. Environ.
18 Epidemiol. 9: 158-166.
19 Wilson, W. E.; Mage, D. T. (1999) Reinventing personal exposure to particulate matter. In: Third colloquium on
20 particulate air pollution and human health: abstracts; June; Durham, NC.
21 Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
22 epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
23 Winer, B. J. (1962) Statistical principles in experimental design. New York, NY: McGraw-Hill Book Company; pp.
24 96-111.
25 Yocom, J. E.; Clink, W. L.; Cote, W. A. (1971) Indoor/outdoor air quality relationships. J. Air Pollut. Control
26 Assoc. 21:251-259.
October 1999 5-103 DRAFT-DO NOT QUOTE OR CITE
-------
i APPENDIX 5-A
2
3 Nomenclature
4
5 a air exchange rate between indoors and outdoors
6 A surface area for PM deposition in an indoor microenvironment
7 AD aerodynamic diameter (not to be confused with actual dimensions or optical diameter)
8 C indoor concentration of PM of ambient origin
9 Ca ambient concentration of PM (virtually all PM of ambient origin) at a monitoring station
10 Co outdoor concentration of PM at a residence, not necessarily equal to Ca
11 E total exposure to PM of ambient origin and PM of non-ambient origin.
12 Ea exposure to PM of ambient origin
13 Ei exposure to PM of non-ambient origin (PM generated indoors)
14 ETS environmental tobacco smoke (PM)
15 F(M) a function of ambient PM mass deposited indoors (M)
16 HVAC Heating-Ventiliation-Air Conditioning
17 k deposition rate of PM onto indoor surfaces, I/time
18 LOD limit of detection
19 M mass of PM of ambient origin deposited indoors
20 MDL minimum detectable level
21 P penetration factor for ambient PM from outdoors to an indoor microenvironment
22 PM^ particles collected by a monitor with a 50% collection efficiency at the given AD
23 Qa rate of resuspension of ambient PM in an indoor microenvironment
24 Qother PTEAM estimate of indoor PM emission rate from sources other than smoking and
25 cooking
26 r Pearson correlation coefficient
27 R2 regression coefficient
28 SPSS Scientific Probability Sampling Schema
29 T time period of the study or the analysis
30 t time, as a variable
31 v alveolar ventilation rate
October 1999 5A-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 v volumetric flow rate of air exchanged between indoors and outdoors
2 V volume of well mixed indoor microenvironment
3 w fraction of time spent outdoors at night in PTEAM (7pm - 7am)
4 x fraction of time spent outdoors during the day in PTEAM (7am - 7pm)
5 y fraction of time spent outdoors during a complete 24-h day
6 z fraction of ambient concentration (Ca) to which a person is exposed during a 24-hour
7 period
8 Y resuspension rate of PM of ambient origin previously deposited in an indoor
9 microenvironment
10 8io delta function equal to 1 when subject is indoors and 0 when subject is outdoors, and8oi
11 is vice versa
12 e spatial variation of ambient PM (Co - Ca)
13 a arithmetic standard deviation
14 // arithmetic mean
15
October 1999 5A-2 DRAFT-DO NOT QUOTE OR CITE
------- |