vvEPA
United States
Environmental Protection
Agency
Office of Research and
Development
DC 20460
EPA/600/AP-95/001a
April 1995
External Review Draft
C.I
Air Quality
Criteria for
Paniculate
Matter
Volume I of III
Review
Draft
(Do Not
Cite or
Quote)
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.
-------
DRAFT-DO NOT QUOTE OR CITE EPA/eoo/AP-95/ooia
April1995
External Review Draft
Air Quality Criteria for
Particulate Matter
i
Volume I of
'' i
"*"*' i
'l
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.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Printed on Recycled Paper
-------
DISCLAIMER
This document is an external 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.
April 1995 I-ii DRAFT-DO NOT QUOTE OR CITE
-------
Air Quality Criteria for Participate Matter
TABLE OF CONTENTS
Volume I
1. EXECUTIVE SUMMARY 1-1
2. INTRODUCTION 2-1
3. PHYSICS AND CHEMISTRY OF PARTICULATE MATTER 3-1
4. SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
ACID DEPOSITION 4-1
5. SOURCES AND EMISSIONS OF SUSPENDED PARTICLES 5-1
6. ENVIRONMENTAL CONCENTRATIONS 6-1
Appendix 6A: Tables of Chemical Composition of PM 6A-1
7. EXPOSURE: AMBIENT AND INDOOR 7-1
Volume II
8. EFFECTS ON VISIBILITY AND CLIMATE 8-1
9. EFFECTS ON MATERIALS 9-1
10. DOSIMETRY OF INHALED PARTICLES IN THE
RESPIRATORY TRACT 10-1
11. TOXICOLOGY OF PARTICULATE MATTER CONSTITUENTS 11-1
Volume III
12. EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
WITH EXPOSURE TO AIRBORNE PARTICLES/ACID
AEROSOLS 12-1
Appendix 12A: Effects of Weather and Climate on Human Mortality and Their
Roles as Confounding Factors for Air Pollution 12A-1
13. INTEGRATIVE HEALTH SYNTHESIS 13-1
April 1995 I-iii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS
"Jase
LIST OF TABLES I-xiv
LIST OF FIGURES I-xviii
AUTHORS, CONTRIBUTORS, AND REVIEWERS I-xxxi
U.S. ENVIRONMENTAL PROTECTION AGENCY PROJECT TEAM
FOR DEVELOPMENT OF AIR QUALITY CRITERIA FOR
PARTICULATE MATTER I-xl
U.S. ENVIRONMENTAL PROTECTION AGENCY SCIENTIFIC
ADVISORY BOARD, CLEAN AIR SCIENTIFIC ADVISORY
COMMITTEE I-xlv
1. EXECUTIVE SUMMARY 1-1
1.1 PURPOSE OF DOCUMENT 1-1
1.2 INTRODUCTION 1-1
1.3 PHYSICS AND CHEMISTRY OF PARTICULATE MATTER 1-2
1.4 SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
ACID DEPOSITION 1-4
1.5 SOURCES AND EMISSIONS OF SUSPENDED PARTICLES 1-5
1.6 ENVIRONMENTAL CONCENTRATIONS 1-7
1.7 EXPOSURE: AMBIENT AND INDOOR 1-13
1.8 EFFECTS ON VISIBILITY AND CLIMATE 1-16
1.9 EFFECTS ON MATERIALS 1-18
1.10 DOSIMETRY MODELING OF INHALED PARTICLES IN THE
RESPIRATORY TRACT 1-20
1.11 TOXICOLOGY OF PARTICULATE MATTER CONSTITUENTS 1-22
1.12 EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
WITH EXPOSURE TO AIRBORNE PARTICLES/ACID
AEROSOLS 1-31
1.13 BIOLOGICAL PLAUSIBILITY POTENTIAL MECHANISMS
OF ACTION 1-70
1.14 IDENTIFICATION OF POPULATION GROUPS
POTENTIALLY SUSCEPTIBLE TO HEALTH EFFECTS FROM
PM EXPOSURE 1-83
1.15 IMPLICATIONS OF RELATIVE RISK ESTIMATES 1-87
2. INTRODUCTION 2-1
2.1 LEGISLATIVE REQUIREMENTS 2-1
2.2 REGULATORY BACKGROUND 2-2
2.3 SCIENTIFIC BASIS FOR THE EXISTING PARTICULATE MATTER
STANDARDS 2-4
2.3.1 Rationale for the Primary Standards 2-4
2.3.2 Pollutant Indicator 2-5
2.3.3 Averaging Time and Form of the Standards 2-6
2.3.4 Level of the Standards 2-8
April 1995 I_v DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
2.3.5 Welfare Effects 2-17
2.4 TOPICS/ISSUES OF CONCERN FOR CURRENT CRITERIA
DEVELOPMENT 2-17
2.4.1 Air Quality and Exposure 2-18
2.4.2 Health Effects 2-22
2.4.3 Welfare Effects 2-28
2.5 DOCUMENT CONTENT AND ORGANIZATION 2-29
REFERENCES 2-32
3. PHYSICS AND CHEMISTRY OF PARTICULATE MATTER 3-1
3.1 INTRODUCTION 3-1
3.1.1 Overview 3-1
3.1.2 Major Chemical Constituents 3-4
3.1.3 Atmospheric Aerosol Size Distributions 3-6
3.1.4 Chemical Composition and Its Dependence on Particle Size 3-9
3.1.5 Particle-Vapor Partitioning 3-11
3.1.6 Single Particle Characteristics 3-12
3.1.7 Definitions 3-13
3.1.7.1 Definitions of Particle Diameter 3-13
3.1.7.2 Definitions of Particle Size Fractions 3-15
3.1.7.3 Other Terminology 3-16
3.1.8 Field Studies 3-17
3.1.9 Dry Deposition 3-18
3.1.10 Atmospheric Scavenging 3-18
3.2 PHYSICAL PROPERTIES 3-18
3.2.1 Aerosol Size Distributions 3-18
3.2.1.1 Particle Size Distribution Functions 3-18
3.2.1.2 Log-Normal Size Distributions 3-19
3.2.1.3 Ambient Aerosol Size Distributions 3-20
3.2.1.4 Coagulation of Spherical Particles 3-20
3.2.2 Particle Formation and Growth 3-22
3.2.2.1 Equilibrium Vapor Pressures 3-22
3.2.2.2 New Particle Formation 3-23
3.2.2.3 Particle Growth 3-23
3.2.2.4 Resuspension 3-24
3.2.3 Particle Removal Mechanisms and Deposition 3-30
3.3 CHEMISTRY AND CHEMICAL COMPOSITION 3-33
3.3.1 Fine Particle Chemistry 3-33
3.3.1.1 Acid Aerosols and Paniculate Sulfates 3-33
3.3.2 Formation of Sulfates in Clouds 3-35
3.3.2.1 Particle Formation in Clouds 3-35
3.3.3 Aqueous-Phase Oxidation of SO2 in Clear-Air Aerosols 3-43
3.3.3 Physical and Chemical Considerations in Particulate Sampling and
Analysis 3-44
April 1995 I-vi DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
3.3.3.1 Semi-Volatile Organic Compounds 3-44
3.3.4 Paniculate Nitrates 3-49
3.3.4.1 Sources 3-49
3.3.4.2 Gas-Phase 3-49
3.3.5 Water Content and Aerosol Equilibria 3-51
3.3.5.1 Water Content of Atmospheric Aerosols, and Its
Dependence on Ambient Humidity 3-51
3.3.5.2 Equilibria with Water Vapor 3-52
3.3.5.3 Ammonium Nitrate Vaporization Equilibria 3-54
3.3.6 Carbon-Containing Paniculate Matter 3-56
3.3.6.1 Introduction 3-56
3.3.6.2 Elemental Carbon 3-56
3.3.6.3 Organic Carbon 3-59
3.3.6.4 Primary Organic Carbon 3-61
3.3.7 Metals and Other Trace Elements 3-69
3.3.8 Removal Processes 3-75
3.4 TRANSPORT AND TRANSFORMATIONS TO SECONDARY
PARTICULATE MATTER 3-76
3.4.1 Aqueous-Phase Chemical Equilibria and Chemical Kinetics of
Transformations to Secondary Paniculate Matter 3-76
3.4.1.1 Aqueous-Phase Equilibria 3-76
3.4.1.2 Aqueous-Phase Transformation of S02 to Sulfate 3-80
3.4.1.3 Aqueous-Phase Transformation of NO2 to HNO3 and
NH4N03 3-87
3.4.2 Transport and Transformations in Plumes 3-87
3.4.2.1 Field Studies of Transport Processes 3-88
3.4.3 Transformations in Plumes 3-98
3.4.3.1 Gas-to-Particle Conversion in Plumes 3-98
3.5 DRY DEPOSITION 3-112
3.5.1 Theoretical Aspects of Dry Deposition 3-112
3.5.2 Field Studies of Dry Deposition 3-116
3.5.2.1 Measured Deposition Velocities 3-119
3.6 WET DEPOSITION 3-121
3.6.1 Introduction 3-121
3.6.2 Field Studies of Wet Deposition 3-123
3.6.2.1 Overview of SO2 and NOX Wet Scavenging 3-128
3.7 PHYSICAL AND CHEMICAL CONSIDERATIONS IN PARTICULATE
MATTER SAMPLING AND ANALYSIS 3-129
3.7.1 Size Cut-Point for Separating Fine and Coarse Particulate
Matter 3-129
3.7.1.1 Background 3-129
3.7.1.2 Size Measurements . 3-129
3.7.1.3 Appropriate Display of Size-Distribution Data 3-131
April 1995 I_vii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
Page
3.7.1.4 Comparison of Particle-Counting and Particle-Collection
Techniques 3-134
3.7.1.5 Review of Size Distribution Data 3-138
3.7.1.6 Intermodal Region 3-144
3.7.1.7 Conclusions 3-169
3.8 SUMMARY 3-169
REFERENCES 3-172
4. SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
ACID DEPOSITION , 4-1
4.1 INTRODUCTION 4-1
4.2 SAMPLING FOR PARTICULATE MATTER 4-5
4.2.1 Background 4-5
4.2.2 Large Particle Separators 4-6
4.2.2.1 Cutpoint Considerations 4-6
4.2.2.2 Total Suspended Particulates (TSP) 4-11
4.2.2.3 Total Inhalable 4-12
4.2.2.4 PM10 4-12
4.2.3 Fine Particle Separators 4-20
4.2.3.1 Cutpoint Considerations 4-20
4.2.3.2 Virtual Impactors 4-21
4.2.3.3 Cyclones 4-23
4.2.3.4 Impactors 4-25
4.2.4 Sampling Considerations 4-26
4.2.4.1 Siting Criteria 4-26
4.2.4.2 Averaging Time/Sampling Frequency 4-27
4.2.4.3 Collection Substrates 4-28
4.2.4.4 Chemical Speciation Sampling 4-30
4.2.4.5 Data Corrections/Analyses 4-33
4.2.5 Performance Specifications 4-34
4.2.5.1 Approaches 4-34
4.2.5.2 Critiques 4-36
4.2.6 Reference and Equivalent Method Program 4-39
4.2.7 Determination of Size Distribution 4-43
4.2.7.1 Cascade Impactors 4-43
4.2.7.2 Single Particle Samplers 4-45
4.2.8 Automated Sampling 4-48
4.2.8.1 TEOM 4-48
4.2.8.2 Beta Gauge 4-50
4.2.8.3 Nephelometer 4-52
4.2.9 Specialized Sampling 4-57
4.2.9.1 Personal Exposure Sampling 4-57
4.2.9.2 Receptor Model Sampling 4-60
4.2.9.3 Particle Acidity 4-61
April 1995 I-viii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
4.3 ANALYSIS OF PARTICULATE MATTER 4-63
4.3.1 Mass Measurement Methods 4-67
4.3.2 Physical Analysis 4-69
4.3.2.1 X-Ray Fluorescence of Trace Elements 4-69
4.3.2.2 Particle Induced X-Ray Emission of
Trace Elements 4-78
4.3.2.3 Instrumental Neutron Activation Analysis of
Trace Elements 4-80
4.3.2.4 Microscopy Analysis of Particle Size, Shape,
and Composition 4-84
4.3.3 Wet Chemical Analysis 4-86
4.3.3.1 Ion Chromatographic Analysis for Chloride,
Nitrate, and Sulfate 4-87
4.3.3.2 Automated Colorimetric Analysis for Ammonium,
Nitrate, and Sulfate 4-90
4.3.3.3 Atomic Absorption Spectrophotometric and
Inductive Coupled Plasma Atomic Emission
Spectrophotometry Analyses for Trace Elements 4-92
4.3.4 Organic Analysis 4-93
4.3.4.1 Analysis of Organic Compounds 4-93
4.3.4.2 Analysis of Organic and Elemental Carbon 4-97
4.3.5 Quality Assurance 4-99
REFERENCES 4-101
5. SOURCES AND EMISSIONS OF SUSPENDED PARTICLES 5-1
5.1 INTRODUCTION 5-1
5.2 SUMMARY OF 1982 CRITERIA DOCUMENT EMISSIONS REVIEW ... 5-3
5.3 SOURCE CONTRIBUTIONS TO SUSPENDED PARTICLES 5-5
5.4 NATIONAL EMISSION RATES AND ANNUAL TRENDS 5-11
5.5 EMISSIONS PROCESSES AND ESTIMATION METHODS 5-19
5.5.1 Fugitive Dust 5-19
5.5.2 Mobile Source Emissions 5-28
5.6 SIZES DISTRIBUTIONS OF PRIMARY PARTICLE
EMISSIONS 5-31
5.7 CHEMICAL COMPOSITIONS OF PRIMARY PARTICLE
EMISSIONS 5-35
5.8 EMISSIONS MODELS AND EMISSIONS INVENTORIES 5-42
5.9 SUMMARY AND CONCLUSIONS 5-44
REFERENCES 5-48
6. ENVIRONMENTAL CONCENTRATIONS 6-1
6.1 BACKGROUND, PURPOSE, AND SCOPE 6-1
6.1.1 Dimensionality and Structuring of the Aerosol Data Space 6-2
6.1.2 Spatial Pattern and Scales 6-4
April 1995 I_ix DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
6.1.3 Temporal Pattern and Scales 6-4
6.1.4 Space-Time Relationships 6-5
6.1.5 Particle Size Distribution 6-6
6.1.6 Aerosol Chemical Composition 6-8
6.1.7 Chapter Organization and Approach 6-10
6.2 CONTINENTAL AND GLOBAL AEROSOL PATTERNS 6-11
6.3 U.S. NATIONAL AEROSOL PATTERN AND TRENDS 6-18
6.3.1 Non-urban National Aerosol Pattern 6-18
6.3.1.1 Non-urban PM2 5 Mass Concentrations 6-18
6.3.1.2 Non-urban PM Coarse Concentrations 6-18
6.3.1.3 Non-urban PM10 Mass Concentrations 6-21
6.3.1.4 PM25/PM10 Ratio at Non-urban Sites 6-21
6.3.1.5 Non-urban Fine Particle Chemistry 6-21
6.3.1.6 Seasonally of the Non-urban Chemistry 6-27
6.3.2 Urban National Aerosol Pattern—AIRS 6-34
6.3.2.1 National Pattern and Trend of AIRS PM10 6-37
6.3.2.2 Eastern U.S. PM10 Pattern and Trend 6-39
6.3.2.3 Western U.S. PM10 Pattern and Trend 6-41
6.3.2.4 Short-term Variability of PM10 Concentrations 6-45
6.3.2.5 AIRS PM2 5 Concentrations 6-45
6.3.2.6 Other National Surveys 6-48
6.3.3 Comparison of Urban and Non-urban Concentrations 6-51
6.4 Regional Patterns and Trends 6-55
6.4.1 Regional Aerosol Pattern in the Northeast 6-56
6.4.1.1 Non-urban Size and Chemical Composition in the
Northeast 6-57
6.4.1.2 Urban Aerosols in the Northeast 6-57
6.4.2 Regional Aerosol Pattern in the Southeast 6-61
6.4.2.1 Non-urban Size and Chemical Composition in the
Southeast 6-61
6.4.2.2 Urban Aerosols in the Southeast 6-63
6.4.3 Regional Aerosol Pattern in the Industrial Midwest 6-66
6.4.3.1 Non-urban Size and Chemical Composition in the
Industrial Midwest 6-66
6.4.3.2 Urban Aerosols in the Industrial Midwest 6-69
6.4.4 Regional Aerosol Pattern in the Upper Midwest 6-71
6.4.4.1 Non-urban Size and Chemical Composition in the Upper
Midwest 6-73
6.4.4.2 Urban Aerosols in the Upper Midwest 6-73
6.4.5 Regional Aerosol Pattern in the Southwest 6-75
6.4.5.1 Non-urban Size and Chemical Composition in the
Southwest 6-79
6.4.5.2 Urban Aerosols in the Southwest 6-79
6.4.6 Regional Aerosol Pattern in the Northwest 6-80
April 1995 I-x DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
6.4.6.1 Non-urban Size and Chemical Composition in the
Northwest 6-84
6.4.6.2 Urban Aerosols in the Northwest 6-84
6.4.7 Regional Aerosol Pattern in Southern California 6-88
6.4.7.1 Non-urban Size and Chemical Composition in
Southern California 6-88
6.4.7.2 Urban Aerosols in Southern California 6-90
6.5 SUBREGIONAL AEROSOL PATTERNS AND TRENDS 6-93
6.5.1 Subregional Aerosol Pattern in the Northeast 6-93
6.5.1.1 Shenandoah National Park 6-93
6.5.1.2 Washington, DC 6-95
6.5.1.3 Comparison of Non-urban (Shenandoah) to Urban
(Washington, DC) Aerosols 6-96
6.5.1.4 New York City, NY 6-100
6.5.1.5 Philadelphia, PA 6-105
6.5.1.6 Whiteface Mountain, NY 6-108
6.5.2 Subregional Aerosol Pattern in the Southeast 6-109
6.5.2.1 Winston-Salem, NC, and Florida 6-109
6.5.2.2 Large Southeast Metropolitan Areas 6-111
6.5.2.3 Great Smoky Mountains 6-112
6.5.3 Subregional Aerosol Pattern in the Industrial Midwest 6-112
6.5.3.1 Pittsburgh, PA 6-112
6.5.3.2 St. Louis, MO 6-117
6.5.3.3 Chicago, IL 6-119
6.5.3.4 Detroit, MI 6-122
6.5.5 Subregional Aerosol Pattern in the Southwest 6-123
6.5.5.1 El Paso, TX 6-123
6.5.5.2 Phoenix and Tucson, AZ 6-123
6.5.6 Subregional Aerosol Pattern in the Northwest 6-128
6.5.6.1 South Lake Tahoe 6-128
6.5.6.2 Salt Lake City, UT, Subregion 6-130
6.5.6.3 Denver, CO 6-132
6.5.6.4 Northern Idaho-Western Montana Subregion 6-132
6.5.6.5 Washington-Oregon Subregion 6-135
6.5.7 Subregional Aerosol Pattern in Southern California 6-137
6.5.7.1 San Joaquin Basin 6-137
6.5.7.2 Los Angeles-South Coast Air Basin 6-140
6.6 CHEMICAL COMPOSITION OF PM AEROSOLS AT URBAN AND
NON-URBAN SITES 6-147
6.7 ACID AEROSOLS 6-152
6.7.1 Introduction 6-152
6.7.2 Geographical Distribution 6-153
6.7.3 Spatial Variation (City-Scale) 6-155
6.7.4 Spatial Variation (Regional-Scale) 6-156
April 1995 I-xi DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
6.7.5 Seasonal Variation 6-157
6.7.6 Diurnal Variation 6-157
6.7.7 Indoor and Personal 6-160
6.8 PARTICLE NUMBER CONCENTRATION 6-160
6.8.1 Introduction 6-160
6.8.2 Ultrafine Particle Number-Size Distribution 6-161
6.8.3 Relation of Particle Number to Particle Mass 6-164
6.8.4 Conclusion 6-169
6.9 AMBIENT CONCENTRATIONS OF ULTRA-FINE METALS 6-169
6.9.1 Introduction 6-169
6.9.2 Formation of Ultra-Fine Particles 6-170
6.9.3 Techniques for Collecting and Analyzing Ultra-Fine Metals . . . 6-172
6.9.4 Observations of Ultra-Fine Metals; Stack and Source-Enriched
Aerosols 6-176
6.10 SUMMARY 6-185
REFERENCES 6-190
APPENDIX 6A: TABLES OF CHEMICAL COMPOSITION OF PM 6A-1
7. HUMAN EXPOSURE TO PARTICULATE MATTER AMBIENT AND
INDOOR CONCENTRATIONS 7-1
7.1 INTRODUCTION 7-1
7.1.1 Ambient PM Concentration as a Surrogate for PM Dosage 7-2
7.1.2 General Concepts for Understanding PM Exposure and
Microenvironments 7-4
7.1.3 Review of State-of-Knowledge Recorded in the 1982
PM-SOX AQCD 7-8
7.2 DIRECT METHODS OF MEASUREMENT OF HUMAN EXPOSURE
TO PM BY PERSONAL MONITORING 7-11
7.2.1 Personal Monitoring Artifacts 7-11
7.2.1.1 "The Hawthorne Effect" 7-11
7.2.1.2 The Monitor Effect 7-12
7.2.1.3 Subject Effect 7-12
7.2.1.4 Non-Response Error 7-12
7.2.2 Characterization of PM Collected by Personal Monitors 7-13
7.2.3 Microscale Variation and the Personal Cloud Effect 7-13
7.3 NEW LITERATURE ON PARTICLE EXPOSURES SINCE 1981 7-14
7.3.1 Review of the Literature 7-16
7.3.1.1 Results of U.S. Studies 7-16
7.3.2 Personal Exposures in International Studies 7-21
7.3.3 The Particle TEAM Study 7-22
7.3.3.1 Pilot Study 7-23
7.3.3.2 Main Study 7-27
7.3.4 Personal Exposures to Constituents of PM 7-36
April 1995 I-xii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
"Jage
7.4 INDIRECT MEASURES OF EXPOSURE 7-39
7.4.1 Personal Exposure Models Using Time-Weighted-Averages
of Indoor and Outdoor Concentrations 7-41
7.5 DISCUSSION 7-46
7.5.1 Relation of Individual Exposures to Ambient
Concentration 7-46
7.5.2 Relation of Community Exposures to Ambient Concentrations . . . 7-51
7.5.3 Implications for PM and Mortality Modeling 7-57
7.5.4 Relative Toxicity of Ambient PM and Indoor PM 7-59
7.5.5 Conclusions 7-62
7.6 INDOOR CONCENTRATIONS AND SOURCES OF PARTICULATE
MATTER 7-63
7.6.1 Introduction 7-63
7.6.2 Concentrations of Particles in Homes and Buildings 7-64
7.6.2.1 Concentrations in Homes 7-65
7.6.2.2 Studies in Buildings 7-107
7.6.2.3 Studies in Locations Other Than Homes and Buildings . 7-114
7.6.3 Indoor Air Quality Models and Supporting Experiments 7-114
7.6.3.1 Mass Balance Models 7-114
7.6.3.2 Models of ETS 7-122
7.6.4 Summary and Conclusions 7-126
REFERENCES 7-129
April 1995 I-xiii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES
Number Page
1-1 Characterization of Urban PM10 Data from AIRs Network by Region
for the United States 1-9
1-2 PM10 Levels by Annual Average for Selected U.S. SMSAs for 1993 1-10
1-3 Selected U.S. PM10 Levels by Second Max PM10 for 1993 1-11
1-4 Concentration Ranges of Various Elements Associated with
Paniculate Matter in the United States Atmosphere 1-12
1-5 Comparison of Relative Risk Estimates for Total Mortality
from 50 jug/m3 Change in PM10 1-37
1-6 Prospective Cohort Mortality Studies 1-48
1-7 Hospital Admissions Studies for Respiratory Disease 1-51
1-8 Hospital Admissions Studies for COPD 1-52
1-9 Hospital Admissions Studies for Pneumonia 1-53
1-10 Hospital Admissions Studies for Heat Disease 1-53
1-11 Acute Respiratory Disease Studies 1-55
1-12 Chronic Respiratory Disease Studies 1-59
1-13 Acute and Chronic Pulmonary Function Changes 1-61
1-14 Estimated Excess Mortality per Day In a Population of One Million
For Which An Increase of 50 /zg/m3 PM10 (24-H) Could Be a
Contributing Factor 1-95
1-15 Estimated Number of Deaths per Day In Cities of Ten Thousand
To Ten Million for Which An Increase of 50 />ig/m3 PM10
Could Be a Contributing Factor 1-95
1-16 Association Between Cigarette Smoking Status and Excess Mortality Risk
from Air Pollution In the Six Cities Study 1-96
1-17 Estimated Hospital Admissions per Day In a Population of
One Million for Which An Increase of 50 jug/m3 (24-H) PM10
Could Be a Contributing Factor 1-98
April 1995 I_xiv DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
1-18 Estimated Number of Hospital Admissions for Respiratory and
Cardiovascular Causes per Day In Cities of Ten Thousand to
Ten Million for Which An Increased of 50 ^g/m3 PM10 (24-H)
Could Be a Contributing Factor 1-98
3-1 Lognormal Parameters for Ambient Aerosols 3-21
3-2 Values of Log P£ for Various PAHs at 20 °C 3-46
3-3 mp Values for PAHs Sorbing to UPM in Osaka, Japan 3-47
3-4 Effects of Three Types of Artifacts on Volume-Averaged Values of
Measured Using a Filter/Adsorbent Sampler 3-48
3-5 Some Secondary Organic Compounds Identified in Ambient Particles in
Urban Air 3-64
3-6 Predicted Percent Contribution to Secondary Organic Aerosol
Concentrations at Los Angeles 3-65
3-7 Amount of Secondary Aerosol Produced in a Typical Los Angeles Smog
Episode According to Functional Groups 3-65
3-8 Reactivity Scale for the Electrophilic Reactions of PAH 3-68
3-9 Concentration Ranges of Various Elements Associated with Paniculate
Matter in the Atmosphere 3-70
3-10 Compounds Observed in Aerosols by a Roadway at Argonne National
Laboratory 3-72
3-11 Compounds Observed in Aerosols in a Forested Area, State
College, PA 3-72
3-12 Henry's Law Coefficients of Some Atmospheric Gases Dissolving in
Liquid Water 3-77
3-13 Recent Field Studies of a-Mesoscale Transport and Trajectory Model .... 3-93
3-14 Scavenging Ratios 3-124
3-15 Relative Humidity of Deliquescence and Crystallization for Several
Atmospheric Salts 3-153
3-16 Summary of Hygroscopic Growth Factors 3-157
April 1995 I-xv DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
3-17 Comparison of Sulfate Concentration and Mass Mean Diameters of
Aerosols for Days with Higher and Lower Relative Humidity 3-163
4-1 EPA-designated Reference and Equivalent Methods for PM10 4-41
4-2 Instrumental Detection Limits for Particles on Filters 4-65
4-3 Minimum Detectable Limits for XRF Analysis of Air Filters 4-73
4-4 INAA Counting Scheme and Elements Measured 4-81
5-1 Receptor Model Source Contributions to PM10 5-7
5-2 Typical Chemical Abundances in Source Emissions 5-36
6-1 Aerosol Information Needs for Assessing Effects 6-2
6-2 Spatial Regions and Scales 6-4
6-3 Maximum SO^" and H+ Concentrations Measured in North American
Cities 6-154
6-4 Regulated Metals and the Volatility Temperature 6-171
6-5 Composition of the Aerosols Present at Grand Canyon National Park
in the Summer of 1984, for the Two Sulfate Episodes of
August 15 and 16 6-179
6-6 Measurements of fine and ultra-fine metals 6-180
6-7 Measurements of fine and ultra-fine metals—lead and nickel 6-183
6-8 Comparison of Selected Species, Shenandoah National Park, and
Washington, DC, San Gorgonio Wilderness, CA, and Grand Canyon
National Park, Summer, 1993 6-187
7-1 Quantile Description of Personal, Indoor, and Outdoor PM3 5
Concentrations by Location in Two Tennessee Communities 7-17
7-2 Summary of WHO/UNEP GEMS/HEAL PM, Personal Exposure Pilot
Study Results 7-22
7-3 Summary of Daily Indoor, Outdoor, and Personal Exposures to PM10
During Cooking as a Function of Fuel Type in Three Cities in Asia 7-23
April 1995 I-xvi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number
7-4 PTEAM Prepilot Study: Mean Outdoor Particle Concentrations (/*g/m3) . . . 7-25
7-5 PTEAM Prepilot Study: 24-Hour Particle Concentrations 7-25
7-6a PTEAM Prepilot Study: 24-Hour PM10 Concentrations 7-26
7-6b PTEAM Prepilot Study: 24-Hour PM2 5 Concentrations 7-27
7-7 Population-Weighted Concentrations and Standard Errors 7-30
7-8 Comparison of PEM Exposure of Individuals to the Simultaneous Ambient
PM Concentration in 10 U.S. Cities and Four Foreign Cities 7-35
7-9 Concentrations of Particles (PM2 5) in Homes of Children Participating
in the Harvard Six-City Study 7-67
7-10a Reconstructed Source Contributions to PM2 5 Mass for Steubenville 7-74
7-1 Ob Reconstructed Source Contributions to PM2 5 Mass for Portage 7-74
7-11 Weighted Summary Statistics by County for Respirable Suspended
Paniculate (PM2 5) Concentrations 7-75
7-12 Weighted Analysis of Variance of Respirable Suspended Particulate
(PM2 5) Concentrations in the Main Living Area of Homes
Versus Source Classification 7-76
7-13 Respirable Suspended Particulate Concentration 7-76
7-14 Weighted Distributions of Personal, Indoor, and Outdoor Particle
Concentrations 7-79
7-15 Weighted Distributions of PM2 5/PM10 Concentration Ratio 7-80
7-16 Stepwise Regression Results for Indoor Air Concentrations of PM10,
PM25, and Nicotine: Coefficients (Standard Errors of Estimates) 7-84
7-17 Penetration Factors, Decay Rates, and Source Strengths: Nonlinear
Estimates 7-87
7-18 Indoor-Outdoor Mean Concentrations of Fine Particles in
Three Large-Scale Studies 7-93
7-19 Influence of Recent Cigarette Smoking on Indoor Concentrations of
PM (Size Unspecified) 7-98
April 1995 I_xvii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
7-20 Indoor and Outdoor PM in Buildings in Helsinki, Finland, as a Function
of Season and Location 7-99
7-21 Indoor Average PM2 5 and PM10 by Reported Smoking in the
Home and Evaporative Cooler Use During Sampling Week 7-100
7-22 Indoor PM10 and PM2 5 by Season and Environmental Tobacco
Smoke ' 7-101
7-23 Smoking, Nonsmoking, and Outdoor RSP Concentrations and Ratios .... 7-108
April 1995 I-xviii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES
Number Page
3-1 Particle size related to RH 3-5
3-2 Number of particles as a function of particle diameter 3-7
3-3 Particle volume distribution as a function of particle diameter 3-7
3-4 Particle volume distribution as a function of particle diameter in a
freeway-influenced area . 3-8
3-5 In concentration as a function of particle size 3-10
3-6 Efficiency values for size-selective sampling criteria 3-16
3-7 Model dust emissions for the United States 3-30
3-8 Diffusion constants and settling velocities for particles 3-31
3-9 Particle deposition from wind tunnel studies 3-32
3-10 Sedimentation and inertia effects on large particle deposition 3-32
3-11 Comparison of observed H2O2 depletions and observed
sulfate yields 3-41
3-12 Extrapolations from correlations of wind tunnel measured deposition
velocities for z = 1 m, densities of 1, 4, and 11.5 g cm"3 3-117
3-13 An example of histogram display and fitting to log-normal functions for
particle-counting size distribution data 3-132
3-14 An example of an effective display of impactor data 3-133
3-15 Size distributions of sulfate, Long Beach, June, 1987, showing use of
fitted log-normal distributions to describe diurnal variations in size
and concentration 3-134
3-16 Effect of changing endpoints 3-135
3-17 These size distributions, obtained during an EPA study of the Denver
brown cloud represent one of the few efforts to compare particle-counting
and particle-collection size-distribution measurements 3-136
April 1995 I_xix DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number
3-18 Grand average volume-size distributions from the Aerosol Characterization
Experiment in 1972 3-139
3-19 Volume-size distribution taken in the midwestern U.S. near the
Cumberland Power Plant in Tennessee 3-140
3-20 Examples of size distribution histograms for total mass, sulfate, and
iron obtained at two visibility levels using an Anderson impactor 3-141
3-21 Impactor size distribution measurement generated by Lundgren et al.
with the Wide Range Aerosol Classifier: Philadelphia and Phoenix 3-143
3-22 Example of aged and fresh coarse mode particle size distributions 3-144
3-23 Size distributions reported by Noll from the Chicago area using an
Anderson impactor for the smaller particles and a Noll Rotary Impactor
for the larger particles 3-145
3-24 Size distribution of dust generated by driving a truck over an unpaved
test track 3-147
3-25 Size distribution of emissions from a pulverized-coal power plant and
the particle size distributions remaining after several types of control
devices 3-149
3-26 Size distributions from a fluidized-bed, pulverized coal combustor,
after initial cleanup by a cyclone collector and after final cleanup
by a baghouse 3-150
3-27 Particle growth curves showing fully reversible hygroscopic growth
of sulfuric acid particles 3-152
3-28 Theoretical predictions and experimental measurements of growth
of NH4HSO4 and (NH4)2SO4 particles at RH between 95 and 100% .... 3-154
3-29 Tandem Differential Mobility Analyzer measurements of the sensitivity
of particle size to relative humidity at Claremont, CA 3-156
3-30 Example of growth in particle size due primarily to increases in
relative humidity from Uniontown, PA 3-159
3-31 Mass size distribution of non-volatile aerosol material 3-160
April 1995 I-xx DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
3-32 Example of particle-counting volume distribution obtained in
Claremont, CA 3-162
3-33 Relative humidity versus sulfur, during the 1986 Carbonaceous
Species Methods Comparison Study, for particles with Dp>0.56 /mi .... 3-165
3-34 Data from the South Coast Air Quality Study 3-166
3-35 Log-log plot of sulfate mode concentration versus mode diameter
from Claremont during summer SCAQS (John et al., 1990) 3-167
3-36 Typical results of size-distribution measurements taken with a Berner
impactor in a Vienna street with heavy automotive traffic (Berner et al.,
1993) 3-168
4-1 Characteristics of aerosol measurement instruments 4-3
4-2 American Conference of Governmental Industrial Hygienists,
British Medical Research Council, and International Organization
for Standardization size-selective sampling criteria 4-7
4-3 Sampling efficiency of IOM ambient inhalable aerosol sampler for three
different types of test aerosol 4-9
4-4 Liquid particle sampling effectiveness curves with solid particle
points superimposed for the Wedding IP10 and the Andersen Samplers
Model 321A inlets at 8 km/h 4-14
4-5 Andersen sampler 4-15
4-6 Sampling characteristics of two-stage size-selective inlet for liquid
aerosols 4-17
4-7 Penetration of particles for 16.67 1 pm PM10 inlets 4-18
4-8 Collection performance variability as a function of wind speed 4-19
4-9 Calibration of a 2.5 /xm impactor 4-22
4-10 Percent collection as a function of aerodynamic diameter 4-24
4-11 Performance of glass fiber filters compared to greased substrate 4-27
4-12 Schematic diagram of an annular denuder system 4-32
April 1995 I-xxi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number
4-13 Aerosol calibration of a cascade impactor 4-44
4-14 TEOM 4-49
4-15 Beta gauge 4-51
4-16 Integrating nephelometer 4-53
4-17 Particle-scattering coefficient as a function of particle size 4-54
4-18 Correlation of bsn and fine fraction mass 4-56
SH
4-19 Collection efficiency of the MSP personal aerosol sampler 4-59
4-20 Modified dichotomous sampler 4-62
4-21 Schematic of a typical X-ray fluorescence system 4-72
4-22 Example of an X-ray fluorescence spectrum 4-75
4-23 Schematic of a PIXE/PESA analysis system 4-79
4-24 Typical gamma-ray spectra observed for long counts 4-82
4-25 Typical gamma-ray spectra observed for short counts 4-83
4-26 Schematic representation of an ion chromatography system 4-88
4-27 Example of an ion chromatogram showing the separation of fluoride,
chloride, nitrite, nitrate, phosphate, and sulfate ions 4-89
4-28 Schematic of a typical automated colorimetric system 4-91
5-1 Primary PM10 emissions estimated for 1983 to 1992 5-12
5-2 Sub-categories of non-fugitive dust emissions, 1983 to 1992 5-13
5-3 National emissions of sulfur dioxide, 1983 to 1992 5-14
5-4 National emissions for oxides of nitrogen, 1983 to 1992 5-15
5-5 National emissions for volatile organic compounds, 1983 to 1992 5-16
5-6 Size distribution of California source emissions, 1986 5-32
April 1995 I-xxii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
5-7 Size distribution of California particle emissions, 1986 5-33
5-8 Particle size distribution in laboratory resuspension chamber 5-34
5-9 Chemical abundances for PM2 5 profiles of road dust 5-37
5-10 Chemical abundances for PM2 5 profiles of vehicle exhaust 5-37
5-11 Chemical abundances for PM25 profiles of wood burning 5-38
5-12 Chemical abundances for PM25 profiles of coal-fired power plant 5-38
6-1 Time scales for particle emissions 6-5
6-2 Relationship of spatial and temporal scales for coarse and fine
Particles 6-7
6-3 Residence time in the lower troposphere for atmospheric particles
between 0.1 and 1.0 pirn 6-8
6-4 Space-time relationship in urban and mountainous areas 6-9
6-5a Continental scale pattern of aerosols derived from visibility
observations over land and satellite monitoring over the
oceans: eastern North America 6-12
6-5b Continental scale pattern of aerosols derived from visibility
observations over land and satellite monitoring over the
oceans: western North America 6-13
6-5c Continental scale pattern of aerosols derived from visibility
observations over land and satellite monitoring over the
oceans: southern North America 6-14
6-6 Global pattern of oceanic aerosols derived from satellite observations 6-15
6-7 Seasonal pattern of oceanic aerosols derived from satellite
observations 6-17
6-8 Fine mass concentration derived from non-urban IMPROVE/NESCAUM
networks 6-19
6-9 Coarse mass concentration derived from non-urban IMPROVE/NESCAUM
networks 6-20
April 1995 I-xxiii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-10 PM10 mass concentration derived from non-urban IMPROVE/NESCAUM
networks 6-22
6-11 Fine fraction of pm10 derived from non-urban IMPROVE/NESCAUM
networks 6-23
6-12 Chemical mass balance of fine particles derived from non-urban
IMPROVE/NESCAUM networks 6-24
6-13a,b Yearly average absolute and relative concentrations for sulfate,
nitrate, organics, and soot 6-25
6-13c,d Yearly average absolute and relative concentrations for sulfate, nitrate,
organics, and soot 6-26
6-14 Seasonal pattern of non-urban aerosol concentrations for the entire
United States 6-28
6-15 Seasonal pattern of non-urban aerosol concentrations for the eastern
United States 6-31
6-16 Seasonal pattern of non-urban aerosol concentrations for the western
United States 6-33
6-17 Trend of valid pm10 monitoring stations in the AIRS database 6-35
6-18 AIRS PM10 quarterly concentration maps using all available data 6-36
6-19 AIRS PM10 and PM2 5 concentration pattern for the conterminous
United States . . . 6-38
6-20 AIRS PM10 and PM2 5 concentration pattern for east of the Rockies 6-40
6-21 AIRS PM10 and PM2 5 concentration pattern for west of the Rockies 6-42
6-22 Short-term PM10 concentration time series for Missoula, MT, and
Knoxville, TN 6-46
6-23 Logarithmic standard deviation airs PM10 concentrations 6-47
6-24 Annual PM2.5 concentration pattern obtained from IMPROVE/NESCAUM
and AIRS networks 6-49
April 1995 I-xxiv DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-25 Monthly mean concentrations in /ig/m3 of IP, fine fraction, and S
as (NH4)2SO4 in Portage, WI; Topeka, KS; Harriman, TN; Watertown, MA;
St. Louis, MO; and Steubenville, OH 6-50
6-26 Spatial maps of PM10 concentration difference between AIRS and
IMPROVE/NESCAUM networks 6-52
6-27 Urban excess concentrations (AIRS minus IMPROVE) for the United
States, eastern United States, and western United States 6-54
6-28 Aerosol regions of the conterminous United States 6-56
6-29 IMPROVE/NESCAUM concentration data for the Northeast 6-58
6-30 AIRS concentration data for the Northeast 6-60
6-31 Short-term variation of PM10 average for the Northeast 6-62
6-32 Urban excess concentration (AIRS minus IMPROVE) for the Northeast . . . 6-62
6-33 IMPROVE/NESCAUM concentration data for the Southeast 6-64
6-34 AIRS concentration data for the Southeast 6-65
6-35 Short-term variation of PMj0 average for the Southeast 6-67
6-36 Urban excess concentration (AIRS minus IMPROVE) for the Southeast . . . 6-67
6-37 IMPROVE/NESCAUM concentration data for the industrial Midwest 6-68
6-38 AIRS concentration data for the industrial Midwest 6-70
6-39 Short-term variation of PM10 average for the industrial Midwest 6-72
6-40 Urban excess concentration (AIRS minus IMPROVE) for the industrial
Midwest 6-72
6-41 IMPROVE/NESCAUM concentration data for the Upper Midwest 6-74
6-42 AIRS concentration data for the Upper Midwest 6-76
6-43 Short-term variation of PM10 average for the Upper Midwest 6-77
April 1995 I-xxv DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-44 Urban excess concentration (AIRS minus IMPROVE) for the Upper
Midwest 6-77
6-45 IMPROVE/NESCAUM concentration data for the Southwest 6-78
6-46 AIRS concentration data for the Southwest 6-81
6-47 Short-term variation of PMj0 average for the Southwest 6-82
6-48 Urban excess concentration (AIRS minus IMPROVE) for the
Southwest 6-82
6-49 IMPROVE/NESCAUM concentration data for the Northwest 6-83
6-50 AIRS concentration data for the Northwest 6-86
6-51 Short-term variation of PM10 average for the Northwest 6-87
6-52 Urban excess concentration (AIRS minus IMPROVE) for the
Northwest 6-87
6-53 IMPROVE/NESCAUM concentration for Southern California 6-89
6-54 AIRS concentration for Southern California 6-91
6-55 Short-term variation of PM10 average for Southern California 6-92
6-56 Urban excess concentration (AIRS minus IMPROVE) for Southern
California 6-92
6-57 IMPROVE/NESCAUM concentration for Shenandoah National Park 6-94
6-58 IMPROVE/NESCAUM concentration for Washington, DC 6-97
6-59 Excess aerosol concentration at Washington, DC compared to
Shenandoah National Park 6-99
6-60 Daily concentration of fine mass and fine sulfur at Washington, DC
and Shenandoah National Park 6-101
6-61 Aerosol concentration map, trend and seasonality in the New York City
region 6-102
6-62 Fine, coarse, and PM10 particle concentration near New York City 6-103
April 1995 I-xxvi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-63 Aerosol concentration map, trend and seasonality in the Philadelphia
region 6-106
6-64 Fine, coarse, and PM10 particle concentration near Philadelphia 6-107
6-65 PM10 concentration seasonality at Whiteface Mountain and neighboring
low elevation sites 6-109
6-66 Aerosol concentration pattern at North Carolina and Florida sites ...... 6-110
6-67 Aerosol concentration pattern in Texas and Gulf states 6-113
6-68 Aerosol concentration pattern and trends in the Pittsburgh subregion .... 6-114
6-69 Fine, coarse, and PMIO concentration near Pittsburgh 6-116
6-70 Aerosol concentration pattern and trends in the St. Louis subregion 6-118
6-71 Fine, coarse, and PM10 concentration pattern near St. Louis 6-120
6-72 Aerosol concentration pattern and trends in the Chicago subregion 6-121
6-73 Aerosol concentration pattern and trends in the El Paso subregion 6-124
6-74 Fine, coarse, and PM10 concentration pattern near El Paso 6-125
6-75 Aerosol concentration pattern and trends in the Phoenix-Tucson
subregion 6-126
6-76 Excess aerosol concentration at South Lake Tahoe compared to Bliss
State Park 6-129
6-77 Aerosol concentration pattern near Salt Lake City 6-131
6-78 Aerosol concentration pattern and trends in the Northern
Idaho-Northwestern Montana subregion 6-133
6-79 PM10 concentration pattern at sites in Northern Idaho-Northwestern
Montana subregion 6-134
6-80 Aerosol concentration pattern in Washington State and Oregon 6-136
6-81 Aerosol concentration pattern and trends in the San Joaquin Valley 6-138
April 1995 I-XXVii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-82 Fine, coarse, and PM10 concentration pattern in the
San Joaquin Valley 6-139
6-83 Aerosol concentration pattern and trends at Los Angeles 6-141
6-84 Fine, coarse, and PM10 concentration pattern near Los Angeles 6-142
6-85a Mass apportionment: eastern United States 6-149
6-85b Mass apportionment: central United States 6-150
6-85c Mass apportionment: western United States 6-151
6-86 Mean air pollutant concentrations for days when winds were from
the southerly direction, plotted versus population density 6-156
6-87 Average monthly aerosol strong acidity for Year 1 sites of the Harvard
24-city study 6-158
6-88 Diurnal Pattern of sulfate and hydrogen ion at Harriman, TN 6-159
6-89 Aerosol number and volume size distributions from an urban site
at Long Beach, CA 6-162
6-90 Aerosol number and volume size distributions from a background site
in the Rocky Mountains, CO 6-163
6-91 Number concentrations as a function of time of day at Long Beach,
CA 6-164
6-92 Number and volume size distributions at the Rocky Mountain site
showing an intrusion of urban air 6-165
6-93 Number and volume size distributions from Los Angeles, CA,
showing comparison of three measurement techniques 6-166
6-94 Relationship between particle number and particle volume 6-168
6-95 Impact of treatment temperature on the enrichment of metals
in the fly ash after the thermal treatment of soils from a
Superfund site 6-173
April 1995 I-xxviii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
6-96 Average normalized concentrations as a function of stage number,
for Se, S, Ca, Al, Si, K, Mo, W, Ni, and Cr for five BLPI samples
from a coal-fired power plant 6-177
6-97 Fine and ultra-fine sulfur at Grand Canyon National Park,
Summer, 1984 6-178
6-98 Fine and ultra-fine metals, nickel, selenium, and lead, in
Long Beach, CA, December 10 through 13, 1987, in
four-hour increments 6-182
6-99 Patterns of zinc, arsenic, sulfur, and selenium in the United States 6-186
6-100 Apparent deposition of automotive lead aerosol in the respiratory
tract of one of the authors as determined by cascade impactor and
PIXIE, as a function of aerodynamic diameter for > 4, 4 to 2, 2 to 1,
1 to 0.5, 0.5 to 0.25, and <0.25 jum particles of size classes 1, 2, 3,
4, 5, and 6, respectively 6-188
7-1 Sizes of indoor particles 7-5
7-2 An example of personal exposure to respirable particles 7-10
7-3 Central-site mean of two dichotomous samplers versus residential
outdoor monitors 7-31
7-4 Personal exposures versus residential (back yard) outdoor PM10
concentrations 7-32
7-5 Increased concentrations of elements in the personal versus the
indoor samples 7-33
7-6 Source apportionment of PTEAM PM10 Personal Monitoring data 7-34
7-7 Personal versus outdoor SO4= 7-37
7-8 Estimated ("best fit" model) versus measured personal SO4= 7-38
7-9 Personal activity cloud and exposure 7-42
7-10 Components of personal exposure ,„.,..„...... 7-47
7-11 Personal exposure to PM in Phillipsburg, NJ (Winter, 1988) 7-52
April 1995 I_xxix DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
7-12 Personal exposure to PM in Bejing, China (Winter, 1985) 7-53
7-13 Personal exposure to PM in Azusa, CA (Spring, 1989) 7-54
7-14 Personal exposure to PM in Riverside, CA 7-55
7-15 Personal exposure to PM in Phillipsburg, NH, with concentration
outliers removed 7-60
7-16 Venn diagram (Mage, 1985) showing focusing of information to more
completely specify toxicity of a given PM mixture 7-61
7-17 Percentage of time spent in different microenvironments by U.S.
residents 7-64
7-18 The annual mean concentration of respirable particles for the
highest and lowest site from the network of indoor and outdoor monitors
in each city 7-67
7-19 Distribution of numbers of children living in households with varying
respirable paniculate matter (PM2 5) as a function of parental smoking
status 7-69
7-20 Distribution percentiles for annual average concentrations of indoor
respirable particulate matter (PM2 5) by household smoking status and
estimated number of cigarette packs smoked in the home 7-71
7-21 PM2 5 in smoking and non-smoking homes in three of the
Harvard Six-Citys Study sites 7-73
7-22 Cumulative frequency distribution of 24-h personal, indoor, and outdoor
PM10 concentrations in Riverside, CA 7-82
7-23 Cumulative frequency distribtuion of 24-h indoor and outdoor PM2 5
concentrations in Riverside, CA 7-82
7-24 Forty-eight-day sequence of PM10 and PM2 5 in Riverside, CA, PTEAM
study 7-83
7-25 Average indoor and outdoor 12-h concentrations of PM10 during the
PTEAM study in Riverside, CA 7-83
7-26 Sources of fine particles (PM2 5) and respirable particles (PM10)
in all homes (Riverside, CA) 7-89
April 1995 I-xxx DRAFT-DO NOT QUOTE OR CITE
-------
\
LIST OF FIGURES (cont'd)
Number Page
7-27 Sources of fine particles (PM2 5) and respirable particles (PM10)
in homes with smokers (Riverside, CA) 7-90
7-28 Sources of fine particles (PM2 5) and respirable particles (PM10)
for homes with cooking during data collection (Riverside, CA) 7-91
7-29 Comparison of respirable particles in smoking and nonsmoking areas
of 38 buildings in the Pacific Northwest 7-109
7-30 Respirable particles in smoking and nonsmoking areas of homes for
the elderly (arithmetic mean for 72 hours) 7-113
7-31 Predicted fate of particles penetrating into buildings of three
California museums as a function of particle size 7-120
April 1995 I_xxxi DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHAPTER 1. EXECUTIVE SUMMARY
Principal Authors
Ms. Beverly Comfort-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lester D. Grant-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dennis Kotchmar-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Annie M. Jarabek-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. James McGrath-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. William Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. Judith Graham-Environmental Criteria and Assessment Office (MD-52), Environmental
Protection Agency, Research Triangle Park, NC 27711
Dr. Jeanette Wiltse, Office of Health and Environmental Assessment, Office of Research and
Development (8601), Waterside Mall, 401 M. St. S.W., Washington, DC 20460
CHAPTER 2. INTRODUCTION
Principal Author
Dr. Lester D. Grant-Environmental Criteria and Assessment Office (MD-52), U.S.
Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dennis Kotchmar-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
April 1995 I-xxxii DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
CHAPTER 3. PHYSICS AND CHEMISTRY OF PARTICULATE MATTER
Principal Authors
Dr. Paul Altshuller-Environmental Criteria and Assessment Office, U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dale Gillette-Atmospheric Research and Exposure Assessment Laboratory (MD-81),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Suzanne Hering-Aerosol Dynamics, Inc., 2319 Fourth Street, Berkeley, CA 94710
Dr. Paul Lioy-Environmental Occupational Health and Science Institute, Exposure
Measurement and Assessment Division, 681 Frelinghuysen Road, Piscataway,
NJ 08855-1179
Dr. Kenneth Noll-Illinois Institute of Technology, Chicago, IL 60616
Dr. Spyros Pandis-Carnegie-Mellon University, Pittsburgh, PA 15146
Dr. James Pankow-Oregon Graduate Center, Beaverton, OR 97229-3678
Dr. Steven Schwartz-Brookhaven National Laboratory, Upton, NY 11934
Dr. William Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. Michael Barnes-Atmospheric Research and Exposure Assessment laboratory (MD-46),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Thomas Cahill-University of California at Davis, Crocker Nuclear Lab., Davis, CA
95616
Ms. Beverly Tilton-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
April 1995 I-xxxiii DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
Contributors and Reviewers (cont'd)
Dr. Sheldon Friedlander-University of California, 5531 Boelter Hall, Los Angeles, CA
90095
Dr. Peter McMurray-University of Minnesota, Department of Mechanical Engineering, 111
Church Street, SE, Minneapolis, MN 55455-0111
Dr. Sidney Soderholm-NIOSH, 1095 Willowdale Rd., Morgantown, WVA 26505
Dr. Barbara Turpin-Rutgers University, Box 231, New Brunswick, NJ 08903
CHAPTER 4. SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
ACID DEPOSITION
Principal Authors
Dr. Judith Chow-Desert Research Institute, P.O. Box 60220, Reno, NV 89506-0220
Dr. Steven McDow-University of North Carolina at Chapel Hill, Chapel Hill, NC 27599
Dr. Charles Rodes-Research Triangle Institute, Center for Aerosol Technology,
P.O. Box 12194, Research Triangle Park, NC 27709-2194
Contributors and Reviewers
Dr. Robert Burton-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Thomas Cahill-University of California - Davis, Crocker Nuclear Laboratory, Davis,
CA 95616
Dr. Delbert Eatough-Brigham Young University, Department of Chemistry, Provo, UT
84602
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Sheldon Friedlander-University of California, Department of Chemical Engineering,
5531 Boelter Hall, Los Angeles, CA 90095
April 1995 I-xxxiv DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
Contributors and Reviewers (cont'd)
Dr. Peter McMurry-University of Minnesota, Department of Mechanical Engineering,
111 Church Street, S.E., Minneapolis, MN 55455-0111
Dr. Sidney Soderholm-National Institute of Occupational Safety and Health, 1095
Williowdale Road, Room 111, Morgantown, WV 26505
Dr. Barbara Turpin-Rutgers University, Environmental Sciences Building, Box 231,
College Farm Road, New Brunswick, NJ 08903
Dr. Russell Wiener-Atmospheric Research and Exposure Assessment Laboratory (MD-77),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Nancy Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
CHAPTER 5. SOURCES AND EMISSIONS
Principal Authors
Dr. Judith Chow-Desert Research Institute, P.O. Box 60220, Reno, NV 89506-0220
Dr. John Watson-Desert Research Institute, P.O. Box 60220, Reno, NV 89506-0220
Contributors and Reviewers
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Timothy Larson-University of Washington, Department of Civil Engineering, Seattle,
WA 98195
Dr. Barbara Turpin-Rutgers University, Environemntal Sciences Building, Box 231, College
Farm Road, New Brunswick, NJ 08903
April 1995 I-XXXv DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
CHAPTER 6. ENVIRONMENTAL CONCENTRATIONS
Principal Authors
Dr. Thomas Cahill-University of California, Davis
Dr. Suzanne Her ing-Aerosol Dynamics, Inc., Berkeley, CA
Dr. Rudolf Husar-Washington University, Center for Air Pollution and Impact and Trend
Analyses, St. Louis, MO 63130
Mr. Joseph Pinto-Atmospheric Research and Exposure Assessment Laboratory (MD-84),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Robert Stevens-Atmospheric Research and Exposure Assessment Laboratory (MD-47),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Robert Willis-ManTech Environmental Technology, Inc., P.O. Box 12313, Research
Triangle Park, NC 27709.
Dr. William Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Michael Zelenka-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. John Core-WESTAR Council, 1001 SW 5th Avenue, Suite 1100, Portland, OR 97204
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Timothy Larson-University of Washington, Department of Civil Engineering, Seattle,
WA 98195
Dr. Brian Leaderer-John B. Pierce Laboratory, 290 Congress Avenue, New Haven,
CT 06519
Dr. Helen Sub-Harvard University, School of Public Health, 665 Huntington Avenue,
Boston, MA 02115
April 1995 I-xxxvi DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
CHAPTER 7. EXPOSURES
Principal Authors
Dr. Dave Mage-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Lance Wallace-Office of Research and Development, U.S. Environmental Protection
Agency, Wallington, VA 22186
Dr. Robert Ziegenfus-Kutztown University, Department of Geography, Kutztown, PA 19550
Contributors and Reviewers
Dr. Gerald Akland-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Timothy Buckley-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Robert Burton-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Paul Lioy-Environmental Occupational Health Science Institute, Exposure Measurement
and Assessment Division, 681 Frelinghuysen Road, Piscataway, NJ 08855-1179
Dr. Peter McMurry-University of Minnesota, Department of Mechanical Engineering,
111 Church Street, SW, Minneapolis, MN 55455
Dr. William Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
April 1995 I-xxxvii DRAFT-DO NOT QUOTE OR CITE
-------
FORWARD
As explained in Chapter 2 (Introduction), the present draft document, Air Quality
Criteria for Particulate Matter, is being prepared to meet Clean Air Act provisions which
require that the Administrator of the United States Environmental Protection Agency (U.S.
EPA) periodically review and revise, as appropriate, the criteria and National Ambient Air
Quality Standards (NAAQS) for paniculate matter and other listed "criteria air pollutants".
This criteria assessment, therefore, contains evaluations of new scientific information that has
become available since the last prior criteria reviews for paniculate matter were carried out
by U.S. EPA in the 1980's, as also described in Chapter 2.
Responsibility for preparation of this External Review Draft of the subject Particulate
Matter Air Quality Criteria Document (PM AQCD) falls within the mission of U.S. EPA's
Environmental Criteria and Assessment Office in Research Triangle Park, NC. That office
(ECAO/RTP) is a component of the Office of Health and Environmental Assessment
(OHEA) within the Office of Research and Development (ORD), the scientific arm of
U.S. EPA. Members of the U.S. EPA Project Development Team for Development of the
present draft document are listed in ensuing pages of the front matter for this volume (I of
III) of the document, and include both ECAO/RTP staff and a few other scientists on
temporary assignment to ECAO/RTP.
The U.S. EPA Project Development team has carried out preparation of an overall
Project Development Plan for preparation of the subject PM AQCD including identification
of key issues to be addressed, planned content of the document, description of the process
and schedule for preparation and review of draft materials, and identification of U.S. EPA
staff scientists and non-EPA consultants expected to serve as authors of sections of the
document. That Project Development Plan was reviewed by members and consultants of the
Clean Air Act Scientific Advisory Committee (CASAC) listed on Pages xlv-xlvii, and it was
appropriately modified in response to their comments and recommendations. The EPA
Project Team has also coordinated and implemented the planning and execution of specific
logistical arrangements, by which the writing, word processing, editing, and assembly of
draft chapter materials and their preliminary peer review and subsequent revision to the
present External Review Draft version have been accomplished.
April 1995 I-xxxviii DRAFT-DO NOT QUOTE OR CITE
-------
FORWARD (cont'd)
Initial drafts of various chapters (other than Chapters 1 and 13) were prepared mainly
during summer/fall of 1994 and subsequently underwent preliminary peer review in early
1995, either (a) as part of public expert workshops held in January 1995 where most key
chapters (4, 6, 7, 10, 11, 12) were discussed in open forum, or (b) via receipt of oral or
written comments on the other chapters from internal EPA scientific staff and/or selected
non-EPA experts. Revisions made in response to such preliminary reviews were then
incorporated into revised chapters to produce this External Review Draft of the document.
The principal authors of materials contained in each chapter of the document are
identified in the front matter for each of the three volumes of this External Review Draft, as
shown on pages preceding this forward for Volume I chapters. Other contributors and
reviewers, who provided comments on initial drafts of particular chapters or other
information importantly considered as inputs to revisions incorporated into the chapters, are
also identified under "Contributors and Reviewers" in the front matter materials for each
volume. However, the evaluations and conclusions contained in this External Review Draft
do not necessarily reflect the individual views of all identified authors, contributors, and
reviewers.
The present External Review Draft of the subject PM AQCD has been prepared for
release for public comment and review by CAS AC, as mandated by the Clean Air Act. The
public comment period, extending from May 1, 1995 to August 1, 1995, will be followed
soon after by a public meeting of CASAC in early August, 1995 (specific site and dates to be
announced in the Federal Register). Further revisions will then be incorporated into this
draft document in response to public comments and CASAC peer review to produce a final
version of this document by no later than January, 1996.
April 1995 I-xxxix 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, Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dennis Kotchmar—Project Manager, Medical Officer, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Ms. Beverly Comfort—Deputy Project Manager/Technical Project Officer, Health Scientist,
Environmental Criteria and Assessment Office (MD-52), U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711
Mr. Norman Childs—Branch Chief, Environmental Media Assessment Branch,
Environmental Criteria and Assessment Office (MD-52), U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711
Dr. A. Paul Altshuller—Technical Consultant, Environmental Criteria and Assessment Office
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. William Ewald—Technical Project Officer, Health Scientist, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Dr. Jasper Garner—Technical Project Officer, Ecologist, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Dr. Judith A. Graham—Associate Director, Environmental Criteria and Assessment Office
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Allan Marcus—Technical Project Officer, Statistician, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Ms. Beverly Tilton—Technical Project Officer, Physical Scientist, Environmental Criteria
and Assessment Office (MD-52), U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711
April 1995 I_xl 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. Annie M. Jarabek—Technical Project Officer, Toxicologist, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Dr. James McGrath—Technical Project Officer, Visiting Senior Health Scientist,
Environmental Criteria and Assessment Office (MD-52), U.S. Environmental Protection
Agency, Research Triangle Park, NC 27711
Dr. William Wilson—Technical Consultant, Physical Scientist, Atmospheric Research and
Exposure Assessment Laboratory (MD-75), U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711
Technical Support Staff
Mr. Douglas B. Fennell—Technical Information Specialist, Environmental Criteria and
Assessment Office (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Ms. Emily R. Lee—Management Analyst, Environmental Criteria and Assessment Office
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Diane H. Ray—Program Analyst, Environmental Criteria and Assessment Office
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Eleanor Speh—Office Manager, Environmental Media Assessment Branch,
Environmental Criteria and Assessment Office (MD-52), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
Ms. Donna Wicker—Administrative Officer, Environmental Criteria and Assessment Office
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Mr. Richard Wilson—Clerk, Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
April 1995 I_xli DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
FOR PARTICULATE MATTER
(cont'd)
Document Production Staff
Ms. Marianne Barrier—Graphic Artist, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Mr. John R. Barton—Document Production Coordinator, ManTech Environmental
Technology, Inc., P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Suzanne Borneman—Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Lynette D. Cradle—Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Jorja R. Followill—Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Sheila R. Lassiter—Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Wendy B. Lloyd—Lead Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. Cheryl B. Thomas—Word Processor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Mr. Peter J. Winz—Technical Editor, Mantech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Technical Reference Staff
Mr. John A. Bennett—Bibliographic Editor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
Ms. S. Blythe Hatcher—Bibliographic Editor, Information Organizers, Inc.,
P.O. Box 14391, Research Triangle Park, NC 27709
Ms. Susan L. McDonald—Bibliographic Editor, Information Organizers, Inc.,
P.O. Box 14391, Research Triangle Park, NC 27709
April 1995 I-xlii 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. Carol J. Rankin—Bibliographic Editor, Information Organizers, Inc., P.O. Box 14391,
Research Triangle Park, NC 27709
Ms. Deborah L. Staves—Bibliographic Editor, Information Organizers, Inc.,
P.O. Box 14391, Research Triangle Park, NC 27709
Ms. Patricia R. Tierney—Bibliographic Editor, ManTech Environmental Technology, Inc.,
P.O. Box 12313, Research Triangle Park, NC 27709
April 1995 I.xiiii DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
CLEAN AIR SCIENTIFIC ADVISORY COMMITTEE
PARTICULATE MATTER CRITERIA DOCUMENT
PROJECT DEVELOPMENT PLAN REVIEW
Chairman
Dr. George T. Wolff—General Motors Corporation, Environmental and Energy Staff,
General Motors Bldg., 12th Floor, 3044 West Grand Blvd., Detroit, MI 48202
Members
Dr. Stephen Ay res—Office of International Health Programs, Virginia Commonwealth
University, Medical College of Virginia, Box 980565, Richmond, VA 23298
Dr. Jay Jacobson-Boyce Thompson Institute, Tower Road, Cornell University, Ithaca,
NY 14853
Dr. Benjamin Liu-Department of Mechanical Engineering, Institute of Technology,
111 Church Street, S.E., Minneapolis, MN 55455
Dr. Joseph Mauderly—Inhalation Toxicology Research Institute, Lovelace Biomedical and
Environmental Research Institute, P.O. Box 5890, Albuquerque, NM 87185
Dr. Paulette Middleton—University Cooperation for Atmospheric Research, P.O. Box 3000,
Boulder, CO 80307
Dr. James H. Price, Jr.—Research and Technology Section, Texas Natural Resources
Conservation Commission, P.O. Box 13087, Austin, TX 78711
Invited Scientific Advisory Board Members
Dr. Morton Lippmann—Institute of Environmental Medicine, New York University Medical
Center, Long Meadow Road, Tuxedo, NY 10987
Dr. Roger O. McClellan—Chemical Industry Institute of Toxicology, P.O. Box 12137,
Research Triangle Park, NC 27711
Consultants
Dr. Stanley Auerbach—Environmental Sciences Division, Oak Ridge National Laboratory,
Bldg. 1505, MS 6036, Oak Ridge, TN 37831
April 1995 I-xlv DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
CLEAN AIR SCIENTIFIC ADVISORY COMMITTEE
(cont'd)
Consultants (cont'd)
Dr. Petros Koutrakis-Harvard School of Public Health, 665 Huntington Avenue, Boston,
MA 02115
Dr. Kinley Larntz-Department of Applied Statistics, University of Minnesota, 352 COB,
1994 Beuford Avenue, St. Paul, MN 55108-6042
Dr. Allan Legge-Biosphere Solutions, 1601 llth Avenue, N.W., Calgary, Alberta T2N
1H1, Canada
Dr. Daniel Menzel-Department of Community and Environmental Medicine, University of
California, 19172 Jamboree Boulevard, Irvine, CA 92717-1825
Dr. Frederick J. Miller-Chemical Industry Institute of Toxicology, P.O. Box 12137,
Research Triangle Park, NC 27709
Dr. William R. Pierson-Energy and Environmental Engineering Center, Desert Research
Institute, P.O. Box 60220, Reno, NV 89506-0220
Dr. Christian Seigneur-ENSR Consulting and Engineering, 1320 Harbor Bay Parkway,
Suite 210, Alameda, CA 94502
Dr. Carl M. Shy-Department of Epidemiology, School of Public Health, University of North
Carolina, CB #7400 McGravran-Greenberg Hall, Chapel Hill, NC 27599-7400
Dr. Jan Stolwijk-Epidemiology and Public Health, Yale University, 60 College Street,
New Haven, CT 06510
Dr. Warren White-Washington University, Campus Box 1134, 684 Waterman Avenue,
St. Louis, MO 63130-4899
Dr. Ron Wyzga-Electric Power Research Institute, 3412 Hill view Avenue, P.O. Box 10412,
Palo Alto, CA 94303
Dr. Mark J. Utell—Pulmonary Disease Unit, Box 692, University of Rochester Medical
Center, 601 Elmwood Avenue, Rochester, NY 14642
April 1995 I-xlvi DRAFT-DO NOT QUOTE OR CITE
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
CLEAN AIR SCIENTIFIC ADVISORY COMMITTEE
(cont'd)
Designated Federal Official
Mr. Randall C. Bond—Science Advisory Board (1400), U.S. Environmental Protection
Agency, 401 M Street, SW, Washington, DC 20460
Staff Assistant
Ms. Janice M. Cuevas—Science Advisory Board (1400), U.S. Environmental Protection
Agency, 401 M Street, SW, Washington, DC 20460
Secretary
Ms. Lori Anne Gross—Science Advisory Board (1400), U.S. Environmental Protection
Agency, 401 M Street, SW, Washington, DC 20460
April 1995 I-xlvii DRAFT-DO NOT QUOTE OR CITE
-------
i 1. EXECUTIVE SUMMARY
2
3
4 1.1 PURPOSE OF THE DOCUMENT
5 The purpose of this document is to present air quality criteria for paniculate matter
6 (PM) in accordance with the Clean Air Act (CAA). Two sections of the CAA (Sections 108
7 and 109) govern the establishment, review, and revision of National Ambient Air Quality
8 Standards (NAAQS). Section 108 directs the Administrator of the U.S. Environmental
9 Protection Agency (EPA) to list pollutants that may reasonably be anticipated to endanger
10 public health or welfare and to issue air quality criteria for them. The air quality criteria are
11 to reflect the latest scientific information useful in indicating the kind and extent of all
12 exposure-related effects on public health and welfare that may be expected from the presence
13 of the pollutant in ambient air.
14
15
16 1.2 INTRODUCTION
17 Air Quality Criteria for Particulate Matter evaluates the latest scientific information
18 useful in deriving criteria that form the scientific basis for U.S. Environmental Protection
19 Agency (EPA) decisions regarding the National Ambient Air Quality Standards (NAAQS) for
20 paniculate matter (PM). This Executive Summary concisely summarizes key conclusions
21 from the document which comprises thirteen chapters. The Executive Summary is followed
22 by a general introduction in Chapter 2. Chapters 3 through 7 provide background
23 information on physical and chemical properties of PM and related compounds; sources and
24 emissions; atmospheric transport, transformation, and fate of PM; methods for the collection
25 and measurement of PM; and ambient air concentrations and factors affecting exposure of the
26 general population. Chapter 8 describes effects on visibility, and Chapter 9 describes
27 damage to materials attributable to PM. Chapters 10 through 13 evaluate information
28 concerning the health effects of PM. More specifically, Chapter 10 discusses dosimetry of
29 inhaled particles in the respiratory tract; Chapter 11 summarizes information on the
30 toxicology of specific types of PM constituents, and includes experimental toxicological
31 studies of animals and human clinical studies. Chapter 12 discusses epidemiological studies.
April 1995 1-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chapter 13 integrates information on critical health issues derived from studies reviewed in
2 the prior chapters. The following sections conform to the chapter organization of the criteria
3 document.
4
5
6 1.3 PHYSICS AND CHEMISTRY OF PARTICULATE MATTER
7 Chapter three first describes the physical properties, chemistry, and chemical
8 composition of PM. Next, the transport and transformation to secondary paniculate matter
9 are presented. Then, both dry and wet deposition are discussed. Finally, the physical and
10 chemical considerations in paniculate matter sampling and analysis are examined. The
11 following brief comments present some basic characteristics of PM.
12 Atmospheric particles originate from a variety of sources and possess a range of
13 morphological, chemical, physical, and thermodynamic properties. Examples include
14 combustion-generated particles such as diesel soot or fly ash, photochemically produced
15 particles such as those found in urban haze, salt particles formed from sea spray, and soil-
16 like particles from resuspended dust. Some particles are liquid, some are solid; others
17 contain a solid core surrounded by liquid. Atmospheric particles contain inorganic ions and
18 elements, elemental carbon, organic compounds, and crustal compounds. Some atmospheric
19 particles are hygroscopic and contain particle-bound water. The organic fraction is especially
20 complex. Hundreds of organic compounds have been identified in atmospheric aerosols,
21 including alkanes, alkanoic and carboxylic acids, polycyclic aromatic hydrocarbons, and
22 nitrated organic compounds.
23 Particle diameters span more than four orders of magnitude, from a few nanometers to
24 one hundred micrometers. Combustion-generated particles, such as those from power
25 generation, from automobiles, and in tobacco smoke, can be as small as 0.01 /mi and as
26 large as 1 /urn. Particles produced in the atmosphere by photochemical processes range in
27 diameter from 0.05 to 2 ^m. Fly ash produced by coal combustion ranges from 0.1 to
28 50 jum or more. Wind-blown dust, pollens, plant fragments, and cement dusts are generally
29 above 2 /xm in diameter. Particles as small as a few nanometers and as large as 100
30 have been measured in the atmosphere.
April 1995 1-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 The composition and behavior of airborne particles are fundamentally linked with those
2 of the surrounding gas. Although the term aerosol is often used to refer to suspended
3 particles, aerosol is defined as a dilute suspension of solid or liquid particles in gas.
4 Paniculate material can be primary or secondary.
5 Primary particles are those emitted in particulate form and include wind-blown dust, sea
6 salt, road dust, mechanically generated particles and combustion-generated particles such as
7 fly ash and soot. The concentration of primary particles depends on their emission rate,
8 transport and dispersion, and removal rate from the atmosphere.
9 Secondary particulate material may form from condensation of high temperature vapor
10 or from vapors generated at as a result of chemical reactions involving gas-phase precursors.
11 Secondary formation processes can result in either the formation of new particles or the
12 addition of particulate material to preexisting particles. Most atmospheric sulfate is formed
13 from atmospheric oxidation of sulfur dioxide. Atmospheric nitrate is also essentially
14 secondary, formed from reactions involving oxides of nitrogen to form nitric acid. A portion
15 of the organic aerosol is also attributed to secondary processes. Secondary aerosol formation
16 can depend on concentrations of other gaseous reactive species such as ozone or hydrogen
17 peroxide, atmospheric conditions including solar radiation and relative humidity, and the
18 interactions of precursors and preexisting particles with cloud or fog droplets. As a result, it
19 is considerably more difficult to relate ambient concentrations of secondary species to sources
20 of precursor emissions than it is to identify the sources of primary particles.
21 Airborne particulate matter can be anthropogenic or biogenic in origin. Both
22 anthropogenic and biogenic particulate material can occur from either primary or secondary
23 processes. Anthropogenic refers to particulate matter which is directly emitted or formed
24 from precursors which are emitted as a result of human activity. Primary anthropogenic
25 sources include fossil fuel combustion, fireplace emissions, and road dust. Secondary
26 anthropogenic particulate material can be generated photochemically from anthropogenic
27 SO2, NOX, or organic gases. Primary biogenic sources include leaf waxes and other plant
28 fragments from plants. In addition, plants emit gaseous species such as terpenes. Terpenes
29 are photochemically reactive, and in the presence of nitrogen oxides can form secondary
30 organic particles. Other types of primary particulate material such as sea salt and wind-
31 generated dust from soil undisturbed by man also are of non-anthropogenic origin.
April 1995 1-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.4 SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
2 ACID DEPOSITION
3 This chapter is intended to provide supplemental information to Chapter 3 and other
4 discussions of aerosol measurement methodologies used in support of the existing PM10
5 standards and/or potentially useful in considerations related to the possible need for
6 development of a "fine particle" standard with an upper cut-point in the 1 to 3 /^m range.
7 The discussion of ambient PM monitoring methods is also included to enhance understanding
8 of exposure data (and their interpretation) used in epidemiology analyses assessed in Chapter
9 12 of this document. An important contribution of the sampling and analytical sections is the
10 extensive compilation of salient peer-reviewed technical references that can be consulted by
11 the reader for more detailed information.
12 Chapter four briefly describes the technical capabilities and limitations of aerosol
13 sampling and analytical procedures focusing on those that were used: (1) to collect data
14 supporting other sections in this document; (2) to support the existing PM10, TSP, and Pb
15 regulations; (3) to support health and welfare effects studies; and/or (4) have application to
16 development of a possible fine particle standard; or (5) illustrate the attributes of several new
17 technologies. The discussion of aerosol separation technologies is divided between (1)
18 devices used to mimic the larger particle (> 10 /im) penetration rationales for the upper
19 respiratory tract airways, and (2) those devices generally used to mimic smaller particle
20 penetration (< 10 pm) to the thoracic regions. These device descriptions are followed by
21 sampling considerations for their applications.
22 The applications of performance specifications to define these measurement systems for
23 regulatory purposes are discussed, along with a number of critical observations suggesting
24 that the current specification process does not always ensure the accuracy or
25 representativeness necessary in the field.
26 The EPA program designating PM10 reference and equivalent sampling systems is
27 briefly described, along with a current list of designated devices. Selected measurement
28 systems used to provide more detailed characterization of aerosol properties for research
29 studies are discussed, with a focus on the determination of particle size distributions.
30 Aerosol sampling systems for specialty applications, including automated samplers, personal
April 1995 1-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposure samplers, and the sampling systems used in aerosol apportionment studies are
2 briefly described.
3
4
5 1.5 SOURCES AND EMISSIONS OF SUSPENDED PARTICLES
6 Chapter five is organized to present first a concise summary of key information on PM
7 emissions derived from the previous criteria review in the 1980's and then to provide a more
8 extensive discussion of newer information appearing in recent years.
9 The main objectives of Chapter 5 discussions are:
10 • To identify the sources that are major contributors to suspended particle
11 concentrations in the United States.
12
13 • To describe the particle sizes and chemical properties of source emissions.
14
15 • To evaluate the limitations and uncertainties of emissions rate estimates and
16 source contributions for suspended particles and their gaseous precursors.
17 The ambient atmosphere contains both primary and secondary particles; the former are
18 emitted directly by sources, and the latter are formed from gases (SO2, NOX, NH3, VOCs).
19 Fugitive dust is a primary pollutant. Major sources of particle emissions are classified as
20 major point sources, mobile sources, and area sources; these are anthropogenic. Natural
21 sources also contribute to ambient concentrations.
22 The 1982 Criteria Document emphasized emissions from industrial sources, especially
23 primary particles. SO2 was the only precursor of secondary particles considered. Since
24 1982, many of these sources have been controlled, yet particle standards are exceeded in
25 many areas.
26 Source and receptor models are used to quantify major contributions to excess PM10
27 concentrations. Source models use emissions inventories and meteorological data to predict
28 particle formation dispersion and particle concentrations measured at receptors. Receptor
29 models use the chemical composition of emissions (finger points) and concentrations at
30 receptor sites to estimate the contribution of sources. The latter are used to identify sources
31 in non-attainment areas.
April 1995 !_5 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fugitive dust is a major contribution to PM10 at nearly all sampling sites, although the
2 average fugitive dust source contribution is highly variable among sampling sites within the
3 same areas, and is highly variable between seasons.
4 Primary motor vehicle exhaust makes up as much as 40% of average PM10 at many
5 sampling sites. Vegetative burning outdoor and residential wood burning are significant
6 sources in residential areas. Fugitive dust from paved and unpaved roads, agricultural
7 operations, construction, and soil erosion constitute —90% of nationwide primary emissions.
8 Fugitive dust consists of geological material that is suspended into the atmosphere by natural
9 wind and by anthropogenic activities from sources such as paved and unpaved roads,
10 construction and demolition of buildings and roads, storage piles, wind erosion, and
11 agricultural tilling. There are obvious discrepancies between the proportion of fugitive dust
12 in primary emissions and geological contributions to PM10 calculated by receptor models,
13 due to contributions from secondary aerosols, which are not included in the primary PM10
14 emission estimates. Even when secondary aerosol is subtracted, however, other sources such
15 as vegetative burning and wood combustion make larger relative contributions to ambient
16 concentrations than is indicated by the emissions inventories. Fugitive dust estimates are
17 especially affected by the general limitations of emissions inventories. All of the emissions
18 have remained relatively constant over the 8-year period except for those from soil erosion.
19 The major non-fugitive dust emitters are other industrial processes and exhaust from
20 highway vehicles. Fuel combustion from utilities, industrial, and other sources together
21 contribute between 1 to 2% to total primary particle emissions. Industrial fuel combustion
22 emissions were reduced by one-third and other fuel combustion emissions were reduced by
23 one half between 1983 and 1992. On-highway vehicle emissions increased by 50%,
24 primarily due to large increases in the number of vehicle miles traveled. Electric utilities
25 account for the largest fraction of sulfur dioxide, nearly 70% of total emissions. These
26 emissions have not changed substantially over the 10 years reported. Annual averages do not
27 reflect the seasonality of certain emissions, residential wood burning in fireplaces and stoves,
28 for example. Cold weather also affects motor vehicle exhaust emissions, both in terms of
29 chemical composition and emission rates. Planting, fertilizing, and harvesting are also
30 seasonal.
April 1995 1-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mobile sources are major emitters of primary particles, oxides of nitrogen, and volatile
2 organic compounds. They are also minor emitters of sulfur dioxide and ammonia. On-road
3 motor vehicles using gasoline-and diesel-fueled engines are by far the largest component of
4 mobile source emissions, and the emissions estimation methods are most highly developed
5 for these vehicles. Studies show that while vehicle emissions models may function well
6 under idealized conditions, they underestimate the effects of high emitting vehicles that may
7 be major sources of VOCs. Motor vehicle exhaust contains high concentrations of organic
8 and elemental carbon, but their ratios are much different from those found in wood
9 combustion with the abundance of elemental carbon being nearly equal to the organic carbon
10 abundance.
11 There are major discrepancies between the relative amounts of emissions and
12 contributions to suspended particles found in many areas. Some major re-design is needed to
13 create more accurate emissions models that can improve the quantification of source-receptor
14 relationships. Emissions models are intended to estimate the emissions rates as a function of
15 space and time of selected pollutants from point, area, and mobile sources. In contrast to an
16 emissions inventory, which is a static catalogue of emissions estimates for a given
17 geographical area and averaging time, an emissions model is capable of accessing activity
18 data bases from a multitude of information-gathering agencies and determining actual
19 emissions for relatively small regions and averaging times.
20
21
22 1.6 ENVIRONMENTAL CONCENTRATIONS
23 Chapter six summarizes PM concentrations over the United States, including the
24 spatial, temporal, size, and chemical aspects. This chapter mainly aims to provide
25 background information on U.S. PM concentrations to help set a context for discussions in
26 later chapters on the characterization and quantification of PM health effects. The general
27 approach in the chapter is to organize, evaluate, and summarize the existing large scale
28 aerosol data sets over the United States. Emphasis is placed on complete national coverage
29 as well as the fusion and reconciliation of multiple data sets.
30 The main organizing dimension used to structure Chapter 6 is space. Accordingly, PM
31 concentrations are presented on global, continental, national, regional, and sub-regional or
April 1995 1-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 urban scales. Within each spatial domain, the spatial-temporal structure, aerosol size and
2 chemical composition are also presented. The presentation of aerosol pattern begins with a
3 global and continental perspective. Next, national U.S. aerosol patterns are examined, as
4 derived from non-urban and urban PM10 and PM2 5 monitoring networks, and the aerosol
5 characteristics over seven subregions of the contiguous United States are examined in more
6 detail. Ten year trends, seasonal patterns, as well as the PM2 5/PM10 relationship and fine
7 particle chemical composition are examined for each region. An ensuing section then
8 focuses further on the sub-regional and urban-scale aerosol patterns over representative areas
9 of the United States.
10 The aerosol concentration pattern over the United States has been reported by many
11 aerosol researchers over the past decade. In particular the research groups associated with
12 the IMPROVE aerosol monitoring networks have been prolific producers of high quality
13 data, reports, and analysis of non-urban data. This section draws heavily on their
14 contribution but the maps, charts, and computations have been re-done for sake of
15 consistency with other (urban) data from the AIRS network. Each of the sections are
16 augmented by suitable but not exhaustive references to the pertinent literature.
17 Table 1-1 summarizes, for illustrative purposes, annual average urban PM10 data from
18 the AIRS Network by U.S. region. The annual means for all regions show declines from
19 1985 to 1993. Tables 1-2 and 1-3 provide more specific illustrative information on 24-h
20 mean, second highest maximum, and annual average values for PM10 concentrations in 1993
21 for selected U.S. cities from various U.S. regions.
22 Next, a section on chemical composition of PM aerosols at urban and non-urban sites
23 summarizes available data on the composition of atmospheric particles. Emphasis has been
24 placed on the Harvard six-city study and the inhalable paniculate network (1980-1981).
25 However, data for fine particle mass and elemental composition only were available from
26 these studies. Data for sulfate, nitrate, and elemental and organic carbon content are
27 included from other studies to provide an overview of the chemical composition of the
28 atmospheric aerosol in the United States. Extensive tables in this section provide detailed
29 representation of atmospheric properties of aerosols to which U.S. populations are exposed.
30 Unfortunately, data this complete are generally collected over limited time periods and are
31 not of sufficient duration to be useful for most epidemiological
April 1995 1-8 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1. CHARACTERIZATION OF URBAN PM10 DATA FROM AIRS NETWORK BY REGION FOR THE
UNITED STATES
Region
Northeast
Southeast
Industrial Midwest
Upper Midwest
Southwest
Northwest
Southern California
PM10
1993
22
24
25
25
26
25
32
(/*g/m3)
1985
36
32
38
31
52
50
45
SDa
30%
17%
28%
19%
45%
45%
40%
Seasonality
Summer Peak July
Summer High
July, August
High June, August
Low November, February
Slightly Lower Levels
December/January
April-June Peak October-
November Peak
August-September Dip
Peak December
Low March-May
Peak November
Low March
Seasonal
Variation %
20 %b
37%
37%
19%
36%
27%
PM2.5/PM10
62%
58%
59%
38%
37%
59%
50%
Influences
Canadian and Gulf
airmasses, local sources,
long range transport
Flat, poor regional
ventilation
Winter cold Canadian
airmasses. Summer moist
Gulf Coast masses
Agricultural Heartland
windblown dust influence
low precipitation, coarse
particle dominant, dust
contribution to PM10
Meteorology highly
variable
Air flow from Pacific, dry
summer, low in remote
Basin wide elevation
O
O
aStandard deviation among monitoring stations within regions.
bSeasonal range expressed as percent.
O
H
O
c;
o
H
w
o
n
HH
H
W
-------
TABLE 1-2. PM10 LEVELS BY ANNUAL
AVERAGE FOR SELECTED U.S.
SMSAs FOR 1993
Metropolitan Statistical Area
Santa Fe, NM
Amarillo, TX
Santa Rosa, CA
Springfield, MD
Casper, WY
Danbury, CT
Glens Falls, NY
Titusville Area, FL
New London Area, CT/RI
Bridgeport, CT
Fort Lauderdale, FL
Asheville, NC
Montgomery, AL
Honolulu, HI
Oakland, CA
Charleston, SC
San Francisco, CA
Dallas, TX
Louisville, KY
Baltimore, MD
Birmingham, AL
Mobile, AL
Orange County, CA
Phoenix, AZ
New York, NY
1990
Population
117,043
187,547
388,222
239,971
61,226
187,867
118,539
398,978
266,819
443,722
1,255,480
174,821
292,517
836,231
2,082,914
506,875
1,603,678
2,553,362
952,662
2,382,172
907,810
476,923
2,410,556
2,122,101
8,546,846
PM10
WTD AM1
0*g/m3)
15
16
18
18
18
19
19
19
19
21
21
22
23
24
26
26
29
30
33
35
36
38
38
44
47
PM10
2nd Max2
Og/m3)
35
29
52
39
41
46
44
57
41
50
71
58
48
58
71
58
72
74
73
70
85
71
80
92
86
O3 (ppm)3
—
-
—
—
—
0.14
—
—
0.13
0.17
—
—
—
—
0.13
—
—
0.14
0.14
0.15
0.13
—
0.17
0.13
—
'Weighted Annual Mean
2Highest Second Maximum 24-hour Concentration
3Highest O3 Second Daily Maximum 1-hour Concentration
April 1995
1-10
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-3. SELECTED U.S. PM10 LEVELS BY 2ND MAX PM10 FOR 1993
Metropolitan Statistical Area
St. Louis, MO
Los Angeles, CA
San Diego, CA
El Paso, TX
Medford, OR
Seattle, WA
Gary, IN
Flint, MI
Bakerville, CA
Fresno, CA
Denver, CO
Chicago, IL
Eugene, OR
Salt Lake City, UT
Spokane, WA
Pittsburgh, PA
Riverside, CA
Steubenville, OH
New Haven, CT
Provo, UT
Philadelphia, PA
1990
Population
2,444,099
8,863,164
2,498,016
591,610
146,389
1,972,961
604,526
430,459
543,477
667,490
1,622,980
6,069,974
282,912
1,072,227
361,364
2,056,705
2,588,793
142,523
638,220
263,590
4,856,881
PM10
2nd Max2
(Mg/m3)
101
102
105
106
106
119
122
127
128
131
142
147
151
156
166
167
172
177
178
209
531
PM10
WTD AM1
(Mg/m3)
44
47
34
37
41
35
34
24
54
53
41
47
28
42
46
38
73
40
52
40
34
Note
O33 0.13
O3 - 0.25, CO4-
14
03 0.16
03 0.14, CO-11
—
—
—
—
O3 0.16
O3 0.14
—
—
—
—
CO 12
SO25 0.155
O3 0.23
S02 0.244
O3 0.15
CO 10
O3 0.14
1 Weighted Annual Mean
2Highest Second Maximum 24-hour Concentration
3Highest O3 Second Daily Maximum 1-hour Concentration
4Highest CO Second Maximum Non-overcapping 8-hour Concentration
5Highest SO2 Second Maximum 24-hour Concentration
1 investigations. The tables do, however, provide insights as to the types of information that
2 could be collected as part of future monitoring efforts in support of human exposure
3 investigations. Table 1-4 provides illustrative data from an earlier 1987 study with regard to
April 1995
1-11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-4. CONCENTRATION RANGES OF VARIOUS ELEMENTS
ASSOCIATED WITH PARTICULATE MATTER IN THE
UNITED STATES ATMOSPHERE (ng/m3)
Elements
As
Cd
Ni
Pb
V
Zn
Co
Cr
Cu
Fe
Hg
Mn
Se
Sb
Remote
0.007 to 1.9
0.003 to 1.1
0.01 to 60
0.007 to 64
0.001 to 14
0.03 to 460
0.001 to 0.9
0.005 to 11.2
0.029 to 12
0.62 to 4160
0.005 to 1.3
0.01 to 16.7
0.0056 to 0.19
0.0008 to 1.19
Rural
1.0 to 28
0.4 to 1000
0.6 to 78
2 to 1700
2.7 to 97
11 to 403
0.08 to 10.1
1.1 to 44
3 to 280
55 to 14530
0.05 to 160
3.7 to 99
0.01 to 3.0
0.6 to 7
Urban (USA)
2 to 2320
0.2 to 7000
1 to 328
30 to 96270
0.4 to 1460
15 to 8328
0.2 to 83
2.2 to 124
3 to 5140
130 to 13800
0.58 to 458
4 to 488
0.2 to 30
0.5 to 171
Source: Schroeder et al., 1987.
1 concentration ranges of various metals found in remote, rural, and U.S. urban areas as
2 common specific airborne PM constituents.
3 Chapter 6 also includes a section on acid aerosols. Acid aerosols are secondary
4 pollutants formed primarily through oxidation of sulfur dioxide (SO2), a gas emitted by the
5 combustion of fossil fuels. Oxidation of SO2 forms sulfuric acid (H2SO4), the major
6 component of acid aerosols. Sulfuric acid is formed to a lesser extent through the oxidation
7 of sulfur species (H2S and CH3SCH3) from natural sources. H+ is found in the fine particle
8 size fraction i.e., particles with aerodynamic diameter (Dp) < 1.0 pirn. Although recent
9 research has shown a typically high correlation between 804 and acidity, data from
10 summertime sampling have shown that SO4= is not always a reliable predictor of H+ for
11 individual events at a given site.
12 A major determinant of the lifetime of H+ in the atmosphere is the rate of
13 neutralization by ammonia (NH3). Ammonia reacts with H2SO4 to form ammonium sulfate
14 [(NH4)2SO4] and ammonium bisulfate (NH4HSO4). The major sources of ammonia in the
April 1995
1-12
DRAFT-DO NOT QUOTE OR CITE
-------
1 environment are animals and humans. In North America, ambient concentrations of H+ tend
2 to be regional in nature with the highest concentrations found in the northeastern United
3 States and southwestern Canada. Recent research has shown that regional transport is
4 important to acid sulfate concentrations, as elevated levels of ambient H+ were measured
5 simultaneously during a regional episode at multiple sites located from Tennessee to
6 Connecticut.
7 Recent work has suggested that ultrafine particles may be responsible for some of the
8 health effects associated with exposure to paniculate matter, leading to an interest in the
9 number concentration of ambient particles. A Chapter 6 section examines data on particle
10 number concentration and the relationship between particle number and particle mass or
11 volume. In some situations the ultrafine mode can be the dominant size range for selected
12 components of atmospheric aerosol particles. One example is the case of metallic aerosols
13 for which fine particles (Dp<2.5/>im) concentrations can be dominated by the ultrafine mode
14 despite the strength of the processes that tend to remove particles from this mode. While
15 there is consensus that ultrafine metals are abundantly produced and emitted into the
16 atmosphere, there are not much data on ambient concentrations of ultrafine metals. The few
17 direct measurements available can be extended with some confidence using indirect methods;
18 i.e., from particle counting techniques that have size information but no chemical
19 information, or from filter collection methods that have limited size information but detailed
20 compositional information.
21
22
23 1.7 EXPOSURE: AMBIENT AND INDOOR
24 Chapter seven focuses on studies which include information on measurement of
25 simultaneous personal PM exposures, indoor-residential PM concentrations, and ambient PM
26 concentrations. The literature on concentrations of PM in indoor settings are also presented.
27 For any air pollutant, the total exposure of an individual consists of a variety of
28 sequential exposures to a variety of microenvironments. They are typically, outdoor, indoors
29 at-home, at-work, in-traffic, and many other indoor microenvironments. The principle of
30 superposition is a useful mechanism to visualize the summation process. For any identified
31 air pollutant, the ambient environment is one source of indoor pollution due to air exchange
April 1995 M3 DRAFT-DO NOT QUOTE OR CITE
-------
1 and infiltration. Whether the ambient is a significant or dominant source of indoor pollution
2 depends on the relative strength of indoor sources and sinks.
3 Paniculate matter studies have detected a "personal cloud" related to the activities of an
4 individual which may generate significant levels of airborne PM in his/her vicinity which
5 may not be picked up by an indoor PM monitor at a distance. Other studies have identified
6 significant sources in the home, e.g. cooking and smoking.
7 In PM of size fractions that include coarse particles, some studies have identified
8 statistically significant relationships between personal exposures and particle concentrations
9 from fixed-site ambient or indoor monitors, and other studies have not, probably due to
10 overwhelming effects of indoor sources, "personal clouds" and other individual activities.
11 For PM of a fine size fraction - such as sulfates, there seems to be more of a relationship
12 between ambient concentration and personal exposure, than for coarser PM, perhaps because
13 of the ability of fine PM to penetrate into indoor settings.
14 For a study population in which there is a detectable correlation between personal
15 exposures and ambient concentrations, the ambient concentration can predict the mean
16 personal exposure with much less uncertainty than it can predict the personal exposure of any
17 given individual in the population.
18 The three largest studies of indoor air particles in the U.S. have all found that the
19 single largest indoor source of fine (PM3 5 or PM2.5) particles is cigarette smoke. The
20 estimate of the impact of smoking on a home PM levels ranges from about 30 to 45 /*g/m3,
21 and of a single cigarette from 1 to 2 jug/m3 for a 24-h period. Homes without smoking have
22 indoor particle concentrations (both PM10 and PM2 5) that are sometimes below and
23 sometimes above the outdoor levels. At low outdoor levels (as in most of the cities in the 6-
24 City and New York State studies) indoor concentrations are generally higher—at high
25 outdoor levels, they are slightly lower. Indoor concentrations are considerably higher during
26 the day, when people are active, than at night.
27 The second largest source of indoor particles, as determined by the PTEAM Study, is
28 cooking. Estimates of the effect of cooking ranged from about 10 to 20 /xg/m3. A few small
29 studies confirm the effect of cooking on indoor particle levels, both PM10 and PM2 5. The
30 two other large-scale studies did not directly test for the effect of cooking, although the
April 1995 1-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 residual indoor concentrations in multivariate calculations led the authors to suggest that
2 cooking could have contributed to the residual.
3 Kerosene heater use was determined to contribute about 15 /^g/m3 to indoor
4 concentrations in one county but not the other in the New York State study. Also, a later
5 effort using mass-balance calculations was unsuccessful in determining a contribution to
6 particle mass from kerosene heater use in either county, although a somewhat smaller set of
7 homes may have been responsible for this result. Gas stoves, wood stoves, and fireplaces
8 were found to have no noticeable impact on total concentrations of particles, although many
9 studies show an increase in PAH concentrations and some show an increase in mutagenicity
10 of indoor air due to these combustion sources.
11 Vacuuming, dusting, and sweeping were found to contribute slightly but with doubtful
12 significance to indoor levels in the PTEAM Study. House volume had a significant but small
13 effect on particle concentrations, with values of —1 to —2 /zg/m3 per 1,000 cubic feet. Air
14 exchange rates were also significant at times, but with different impacts depending on the
15 relative indoor and outdoor concentrations—at high outdoor concentrations, increased air
16 exchange resulted in increases in the indoor air particle levels.
17 Unknown indoor sources were found to account for a substantial fraction (25%) of
18 indoor concentrations (both PM2 5 and PM10) in the PTEAM study. This suggests a need for
19 further research to determine the source or sources of these particles.
20 Decay rates for fine (PM2 5) particles were determined to be about 0.4 h"1 compared to
21 1 h'1 for coarse particles, with an intermediate estimate of 0.65 h"1 for PM10. For a home
22 with no indoor sources and a typical air exchange rate of about 0.75, this would imply that
23 fine particles indoors would be about 0.757(0.4+0.75) = 65% of the outdoor value at
24 equilibrium, indoor PM10 would be about 54% of outdoor levels, and indoor coarse particles
25 would be about 43% of outdoor levels. Since few homes were observed to have
26 concentrations this low, it can be inferred that few homes are free of important indoor
27 sources of particles.
28 Studies in buildings also indicated that smoking was the major indoor source of
29 particles, with a geometric mean of 44 versus 15 /ig/m3 (arithmetic mean of 70 versus
30 18 /*g/m3) observed for smoking versus nonsmoking areas in 38 Pacific Northwest buildings.
April 1995 1_15 DRAFT-DO NOT QUOTE OR CITE
-------
1 This difference of 29 to 52 /xg/m3 is similar to the difference of 30 to 45 /^g/m3 estimated
2 from the three major studies of U.S. homes.
3 Indoor air quality models have been employed with increasing success to estimate
4 source emission rates and particle decay rates. Cigarettes smoked in homes with normal
5 activities emit about 14 mg/cigarette, a result that agrees well with various chamber studies
6 with smokers or smoking machines. Cooking was estimated to emit 4 mg/min, a result that
7 needs confirmation. Elemental emission profiles have been determined for both smoking and
8 cooking, with potassium and chloride being major markers for smoking, and iron and
9 calcium for cooking. Particle decay rates have been estimated for homes to range between
10 0.4 and 1.0 h"1. Studies in telephone equipment buildings and museums have established
11 particle deposition velocities for sulfates and other ions, although differences in the estimates
12 suggest that further research is needed.
13
14
15 1.8 EFFECTS ON VISIBILITY AND CLIMATE
16 Chapter eight discusses factors affecting visibility, ways to measure it, historical trends,
17 and methods to determine its value. Paniculate matter effects on climate are also discussed.
18 Much of the information contained in the section on visibility is a summary of information
19 from the previous 1982 Criteria Document for Paniculate Matter and Sulfur Oxides.
20 Traditionally, visibility has been defined in terms of the distance from an object that is
21 necessary to produce a minimum detectable contrast between that object and its background.
22 Although visibility is often defined by this "visual range," it includes not only being able to
23 see or not see a target, but also seeing targets at shorter distances and appreciating the details
24 of the target, including its colors. Visibility impairment can manifest itself in two ways: (1)
25 as a layer of haze (or a plume), which is visible because it has a visual discontinuity between
26 itself and its background, or (2) as a uniform haze which reduces atmospheric clarity. The
27 type and degree of impairment are determined by the distribution, concentrations, and
28 characteristics of atmospheric particles and gases, which scatter and absorb light traveling
29 through the atmosphere. Scattering and absorption determine light extinction.
30 On a regional scale, the extinction of light is generally dominated by particle scattering.
31 In urban areas, absorption by particles becomes important and occasionally dominant.
April 1995 1-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 Extinction by particles is usually dominated by particles of diameter 0.1 to 2 pm (fine par-
2 tides). In general, scattering by particles accounts for 50 to 95% of extinction, depending
3 on location, with urban sites in the 50 to 80% range and nonurban sites in the 80 to 95%
4 range.
5 Available visibility monitoring methods measure different aspects of visibility
6 impairment. Generally, contrast-type measurements (such as photography, telephotometry,
7 and human eye observations) relate well to the perception of visual air quality, while
8 extinction or scattering measurements (such as transmissometry and nephelometry) relate to
9 the cause of visibility degradation. The above measurement methods can be used to approx-
10 imate visual range.
11 Current knowledge indicates that fine paniculate matter is composed of varying
12 amounts of sulfate, ammonium, and nitrate ions, elemental carbon, organic carbon
13 compounds, water, and smaller amounts of soil dust, lead compounds, and trace species.
14 Sulfate often dominates the fine mass and light scattering, while elemental carbon is the
15 primary light-absorbing species. Ammonium ion typically accounts for 5 to 15% of the fine
16 mass and often correlates well with sulfate levels. Mean nitrate concentrations can represent
17 up to 37% of the total fine particle mass in urban cities.
18 Visibility has value to individual economic agents primarily through its impact upon
19 activities of consumers and producers. Most economic studies of the effects of air pollution
20 on visibility have focused on the aesthetic effects which are believed to be the most
21 significant economic impacts of visibility degradation caused by air pollution in the U.S. It
22 is well established that people notice changes in visibility and that visibility conditions affect
23 the well-being of individuals.
24 Paniculate matter of submicron size in the earth's atmosphere perturbs the radiation
25 field. There is no doubt that anthropogenic aerosols have the potential to affect climate; the
26 question is by how much. There are two chief avenues through which aerosols impact the
27 radiation budget of the earth. The direct effect is that of enhanced reflection of solar
28 radiation by particles in a cloud-free atmosphere. Since aerosols, even those containing some
29 absorptive component, are primarily reflective, their impact is felt as a negative radiative
30 forcing (i.e., a cooling) on the climate system. Although there is some uncertainty in the
31 global distribution of such aerosols and in the chemical and radiative properties of the
April 1995 1_17 DRAFT-DO NOT QUOTE OR CITE
-------
1 aerosols, the radiative effects can still be modeled within certain bounds. Estimates of this
2 forcing range from — 0.3 W m"2 to about twice that value for current conditions over pre-
3 industrial times.
4 The indirect forcing results from the way in which aerosols, in their role as cloud
5 condensation nuclei (CCN), affect cloud microphysical properties. The most important is the
6 effective radius of cloud droplets, which decrease as CCN concentrations increase. This
7 effect is most pronounced when the concentration, N, is very low, and clouds are moderately
8 reflective. Other effects are the enhancement of cloud lifetimes and also changes in the
9 nucleating ability of CCN through chemical changes. Although estimates of the indirect
10 effect are uncertain by at least a factor of 2, it appears to be potentially as important as the
11 direct effect. On a global mean basis, anthropogenic emissions of anthropogenic aerosols
12 could have offset substantially the positive radiative forcing due to greenhouse gas emissions.
13 The one crucial difference between aerosol forcing and greenhouse (gas) forcing is the
14 atmospheric lifetime of aerosols and gases and hence, forcing. The aerosol forcing is fairly
15 regional, whereas the greenhouse forcing is global. One should, therefore, expect
16 inter-hemispheric differences in the forcing and perhaps climate response. However, climate
17 models are not currently at the level of sophistication needed to determine climate response
18 unambiguously. Global observations of surface temperature cannot separate natural and
19 anthropogenic causal mechanisms, with few exceptions.
20
21
22 1.9 EFFECTS ON MATERIALS
23 Chapter nine briefly discusses the effects of paniculate matter exposure on the aesthetic
24 appeal and physical damage to different types of building materials and economic
25 consequences, including background information on the physics and chemistry of atmospheric
26 corrosion. Where possible, the chapter discusses only those effects associated with particle
27 exposure; however, most of the data are on the effects of particles in combination with SO2.
28 A significant detrimental effect of particulate matter pollution is the soiling of painted
29 surfaces and other building materials. Soiling is defined as a degradation mechanism that can
30 be remedied by cleaning or washing, and depending on the soiled surface, repainting.
31 Available data on pollution exposure indicates that particulate matter can result in increased
April 1995 1-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 cleaning frequency of the exposed surface, and may reduce the life usefulness of the material
2 soiled. Data on the effects of particulate matter on other surfaces are even less well
3 understood. Some evidence also shows damage to fabrics, electronics, and works of art
4 composed of one or more materials, but this evidence is largely qualitative and sketchy.
5 The damaging and soiling of materials by airborne pollutants have an economic impact,
6 but this impact is difficult to measure. The accuracy of economic damage functions is
7 limited by several factors. One of the problems has been to separate costs related to
8 particulate matter-related materials from other pollutants, as well as from those related to
9 normal maintenance. Cost studies typically involve broad assumptions about the kinds of
10 materials that are exposed in a given area and then require complex statistical analysis to
11 account for a selected number of variables. Attitudes regarding maintenance may vary
12 culturally, further confounding the problem of quantifying economic impact.
13 The nature and extent of damage to materials by particulate matter have been
14 investigated by field and laboratory studies. Both physical and economic damage functions
15 have been developed for specific damage/effect parameters associated with exposure to
16 particulate matter. To date, only a few of these functions are relatively reliable in
17 determining damage, while none has been generally accepted for estimating costs.
18 In recent years, fairly reliable damage functions for soiling of exterior wall paints have
19 been developed. The available damage functions are few in number but represent a major
20 fraction of the total surface that is exposed and sensitive to pollution damage.
21 Although there still remains a lack of sensitive materials distribution data, the
22 geographic resolution of available data is about as good as that of environmental monitoring
23 data. These limitations may hinder accurate estimates of total material damage and soiling,
24 but they do not prevent estimates within ranges of error. Studies have used various
25 approaches to determine pollutant-related costs for extra cleaning, early replacement, more
26 frequent painting, and protective coating of susceptible materials, as well as other indicators
27 of the adverse economic effects of pollutants. No study has produced completely satisfactory
28 results, and estimates of cost vary widely.
29
30
April 1995 1-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.10 DOSIMETRY MODELING OF INHALED PARTICLES IN THE
2 RESPIRATORY TRACT
3 Particles are deposited in the respiratory tract by mechanisms of impaction,
4 sedimentation, interception, diffusion, and electrostatic precipitation. Ventilation rates differ
5 for various activity patterns in humans, for different ages, and among species. These
6 ventilation differences coupled with differences in upper respiratory tract structure and in
7 size, branching pattern, and structure of the lower respiratory tract among species and
8 healthy versus diseased states result in significantly different patterns of airflow that in turn
9 affect particle deposition in the respiratory tract regions. For a given aerosol, the two most
10 important parameters determining deposition are the mean aerodynamic diameter and the
11 distribution of particles about that mean. Subsequent clearance of a deposited dose is
12 dependent on the initial site of deposition, physicochemical properties of the particles (e.g.,
13 dissolution half-time), and on time since deposition.
14 An accurate description of the exposure-dose-response relationship for the observed
15 health effects of PM should account, to the extent possible, for these mechanistic
16 determinants of particle disposition. Deposited dose may be an appropriate metric for
17 "acute" effects, (e.g., mortality), especially if the particles exert their primary action on the
18 surface contacted. An alternative to consider is dose rate (jwg/min) per unit surface area
19 because insoluble particles deposit and clear along the surface of the respiratory tract.
20 "Chronic" effects (e.g., certain types of morbidity) may be better described by retained dose
21 estimates because clearance is affected by the time since deposition and the aerosol solubility,
22 characterized by dissolution-absorption half-times.
23 The human model chosen to make deposited and retained dose predictions is a semi-
24 empirical compartmental model that is able to describe particle deposition and clearance by
25 three routes (absorption into blood, transport to gastrointestinal tract, and transport to
26 lymphatics). Two different models were used to model particle deposition and clearance in
27 laboratory animals.
28 The predictions of deposited and retained doses show anticipated differences due to the
29 influence of aerosol particle diameter and distribution, minute ventilation, and species-
30 specific morphometry. For example, mouth breathing alters the deposition fraction of
31 ambient aerosols in the tracheobronchial and alveolar regions when compared to nasal
April 1995 1_20 DRAFT-DO NOT QUOTE OR CITE
-------
1 breathing. The differences shown in the these predictions for deposition point to the
2 importance of characterizing the differences between genders and the impact of age on
3 deposition. The chosen model has predicted differences between children of 1 year and
4 adults across particle diameters ranging from the diffusion to aerodynamic range of
5 approximately 2.5-fold in the tracheobronchial region and 2-fold in the alveolar region. The
6 direction and magnitude are a function of aerosol particle diameter and distribution.
7 Differences in ventilation and morphometry for diseased states can also be anticipated and
8 recent studies and other model predictions show an increased deposition in subjects with
9 chronic obstructive pulmonary disease.
10 The various species used in inhalation toxicology studies that serve as the basis for
11 exposure-dose-response assessment do not receive identical doses in a comparable respiratory
12 tract region when exposed to the same aerosol. Such interspecies differences are important
13 because the toxic effect is likely more related to the quantitative pattern of deposition within
14 the respiratory tract than to the exposure concentration; this pattern determines not only the
15 initial respiratory tract tissue dose but also the specific pathways by which the inhaled
16 material is cleared and redistributed. Thus, accounting for differences in dosimetry can
17 change the apparent effect levels among species. To illustrate, for the same aerosol of 0.5
18 jim MM AD and a of 1.3, using deposition normalized to surface area for an effect in the
6
19 tracheobronchial region, an exposure concentration of 100 /xg/m3 to rats and guinea pigs
20 would predict a human equivalent exposure concentration of 939 and 79 /ig/m3, respectively,
21 assuming species sensitivity to the deposited tracheobronchial dose were equal. However,
22 for chronic exposures to the same aerosol, retained alveolar dose (/-ig/g lung tissue) may be
23 more appropriate as a dose metric. Assuming it is a relatively insoluble aerosol (i.e.,
24 assuming a dissolution-absorption half-time of 1,000 days), a human equivalent exposure
25 concentration would be predicted to be 22 and 784 /xg/m3 based on the rat versus guinea pig,
26 respectively.
27 These examples show that relevance of a particular animal model should be considered
28 together with dosimetry and the appropriateness of the metric for a given health endpoint. In
29 general, the objective should be to provide a metric that is mechanistically motivated by the
30 observed health effect of interest for extrapolation.
April 1995 1-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dosimetry modeling at the moment can address important mechanistic factors of
2 particle deposition and clearance including the aerosol particle diameter and distribution,
3 intra and interspecies differences in deposition as a function of ventilation and morphometry,
4 and intra and interspecies differences in clearance rates. Use of dosimetry modeling and
5 judicious choice of appropriate dose metrics should be used to interpret the observed health
6 effects data related to PM10 exposures.
7 Further, these predictions were based on the use of mass as the exposure metric.
8 Recent data suggest that particle number, or possibly particle surface area, may be a more
9 appropriate exposure metric because the fine mode aerosols are small in mass but have
10 extremely high concentrations of particle numbers. Also, normalizing factors such as
11 number of alveoli or number of macrophages may be more appropriate for certain
12 pathogenesis mechanisms. Creating these dose metrics for various species will depend on the
13 availability of morphometric information.
14
15
16 1.11 TOXICOLOGY OF PARTICIPATE MATTER CONSTITUENTS
17 Chapter eleven reviews results on exposure to specific PM constituents, based on
18 controlled human clinical studies, selected occupational studies, and animal toxicology
19 studies. It focuses on those studies published since the 1982 PM Criteria Document and
20 includes coverage of specific PM species selected for discussion based on their being
21 commonly present in ambient aerosols at concentrations > 1 ng/m3.
22 Paniculate matter is a broad term that encompasses thousands of chemical species,
23 many of which have not been investigated in controlled animal or human studies. However,
24 even a full discussion of all the types of particles that have been studied is well beyond the
25 scope of the chapter. Most of the animal toxicological and occupational epidemiological
26 studies summarized used very high paniculate concentrations, relative to ambient, even when
27 animal-to-human dosimetric differences are considered. In spite of these difficulties, the
28 array of animal studies does illustrate certain toxicological principles for particles. To
29 identify but a few here, the data base clearly shows that the site of respiratory tract
30 deposition (and hence particle size) clearly influences the health outcome and that toxicity is
31 dependent on the chemical species.
April 1995 1-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 Effects of Controlled Exposure to Acid Aerosols
2 The effects of acid aerosols are primarily related to strong acidity (i.e., H+ content).
3 That is, H2SO4 is more potent than NH4HSO4 which is more potent than NH4(H)2SO4. The
4 size of acid aerosols also affects their potency, but the exact relationships are dependent upon
5 the endpoint being examined.
6 Sulfuric acid affects pulmonary function. Healthy subjects are only affected by very high
7 levels of acute exposure (around 2,000 ng/m3), even if they exercise and gargle with acidic
8 material to reduce neutralization by oral ammonia. Asthmatics, especially adolescent
9 asthmatics, appear to be more sensitive. For example, a few, but not all studies, found that
10 an acute exposure to around 70 ^g/m3 caused small decrements in the pulmonary function of
11 adolescent asthmatics.
12 Sulfuric acid affects mucociliary clearance in humans and animals. The direction (i.e.,
13 increase or decrease) and the magnitude of the effect is dependent on the concentration and
14 duration of exposure, as well as the specific region of the lung being measured. Humans
15 exposed to levels of H2SO4 as low as 100 jiig/m3 (Ih) experienced a decrease in mucociliary
16 clearance. Animal studies have shown that H2SO4 can also affect alveolar clearance.
17 Chronic exposure to H2SO4 causes a variety of structural changes in the lung. For
18 example, mucus-secreting cells are affected and can be found in deeper regions of the lung
19 than usual. Pulmonary function is also altered.
20 Several studies have sought to define interactions of acids with other pollutants,
21 especially ozone. Work with animals has demonstrated additivity, synergism, and
22 antagonism, depending upon the species, exposure, and endpoint. More recent human
23 clinical studies found that acute exposure to 100 ptg/m3 H2SO4 may potentiate the response
24 to ozone on pulmonary function.
25
26 Complex Mixtures
27 The 1982 Air Quality Criteria Document for Particulate Matter and Sulfur Dioxide
28 concluded from its review of studies on the genotoxicity and carcinogenicity of atmospheric
29 particles that "all the major types of airborne paniculate matter may contain adsorbed
30 compounds that are mutagenic and/or carcinogenic to animals and may contribute in some
31 degree to the human cancer associated with exposure to urban air pollution." Recent
April 1995 1-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 research activity has added data that support this conclusion, but do not warrant that it be
2 changed significantly.
3 The direct relevance of the evidence for the mutagenicity and tumorigenicity of extracts
4 of particulate matter in experimental systems to exposure scenarios experienced by humans is
5 uncertain at this time. Recent analytical epidemiological studies, that adjusted for tobacco
6 smoking and other major potential risk factors, have found a weak to non-existent association
7 between human lung cancer and indices of exposure to air pollution including particulate
8 matter. Most investigators believe that the epidemiological evidence obtained thus far does
9 not substantiate causality, although the hypothesis remains credible.
10
11 Diesel Emissions
12 Acute toxic effects caused by exposure to diesel exhaust are mainly attributable to the
13 gaseous components (i.e., mortality from carbon monoxide intoxication and lung injury from
14 respiratory irritants). When the exhaust is diluted to limit the concentrations of these gases,
15 acute effects are not seen.
16 Ten different long-term (> 1 year) animal inhalation studies of diesel engine emissions
17 have been conducted. The focus of these studies has been on respiratory tract effects in the
18 alveolar region. Effects in the upper respiratory tract and in other organs were not found
19 consistently in chronic animal exposures. The pathogenic sequence following the inhalation
20 of diesel exhaust as determined histopathologically and biochemically begins with the
21 phagocytosis of diesel particles by alveolar macrophages (AMs). These activated
22 macrophages release chemotactic factors that attract neutrophils and additional AMs. As the
23 lung burden of diesel particles increases, there is an aggregation of particle-laden AMs in
24 alveoli adjacent to terminal bronchioles, increases in the number of Type 2 cells lining
25 particle-laden alveoli, and the presence of particles within alveolar and peribronchial
26 interstitial tissues and associated lymph nodes. The PMNs and macrophages release
27 mediators of inflammation and oxygen radicals and particle-laden AMs are functionally
28 altered resulting in decreased viability and impaired phagocytosis and clearance of particles.
29 There is a substantial body of evidence for an impairment of particulate clearance from the
30 bronchio-alveolar region of rats following exposure to diesel exhaust. The latter series of
April 1995 1-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 events may result in the presence of pulmonary inflammatory, fibrotic, or emphysematous
2 lesions.
3 The noncancer toxicity of diesel emissions is considered to be due to the particle rather
4 than the gas phase, since the long-term effects seen with whole diesel are not found or are
5 found to a much lesser extent in animals exposed to similar dilutions of diesel exhaust
6 filtered to remove most of the particles. Chronic studies in rodents have demonstrated
7 pulmonary effects at 200 to 700 jug/m3. No-effect levels have been reported ranging from 60
8 to 260 Mg/m3.
9 Several epidemiologic studies have evaluated the effects of chronic exposure to diesel
10 exhaust on occupationally exposed workers. None of these studies are useful for a
11 quantitative evaluation of noncancer toxicity because of inadequate exposure characterization,
12 either because exposures were not well defined or because the possible confounding effects of
13 concurrent exposures could not be evaluated.
14 The U.S. Environmental Protection Agency has developed a draft qualitative and
15 quantitative cancer assessment for diesel emissions. The summary to follow was drawn from
16 that document. This draft is currently undergoing external review by the public and the
17 Clean Air Scientific Advisory Committee. As a result of limited evidence from
18 epidemiological data, supported by adequate evidence for carcinogenicity of diesel engine
19 emissions in animal studies, as well as positive evidence for mutagenicity, it was concluded
20 that diesel engine emissions best fit into cancer weight-of-evidence Category Bl. Diesel
21 engine emissions are thus considered to be probable human carcinogens. This is in
22 agreement with a 2A classification by the International Agency for Research on Cancer.
23 Using a dosimetry model that accounted for animal-to-human differences in lung
24 deposition efficiency, lung particle clearance rates, lung surface area, ventilation, metabolic
25 rate, as well as elution rates of organic chemicals from the particle surface, equivalent human
26 doses were calculated on the basis of particle concentration per unit lung surface area.
27 Following dosimetric adjustment, risk estimates were derived using a linearized multistage
28 model. A unit risk estimate of 3.4 x 10"5 (the upper 95% bound of the cancer risk from
29 lifetime exposure to 1 /ig/m3 diesel paniculate matter) is recommended. This estimate is
30 based on the geometric mean of estimates derived from three separate animal bioassays using
31 Fischer 344 rats.
April 1995 1-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 Metals
2 A summary of the health effects of airborne metals follows. The descriptions are
3 qualitative in nature. Most of the literature on these compounds deals with high
4 concentration animal toxicological studies or occupational epidemiological studies.
5 The primary target for inhaled aluminum compounds is the respiratory tract.
6 Commonly reported symptoms include asthma, cough, and decreased pulmonary function;
7 fibrosis has also been reported. Laboratory animal studies support findings from human
8 studies that aluminum acts via an irritant, rather than by an allergic, mechanism.
9 The respiratory tract is the primary target organ for antimony (trioxide) following
10 inhalation exposure. Respiratory effects have been reported in workers chronically exposed
11 to mg levels of antimony dust. Other reported effects include altered ECGs, gastrointestinal
12 symptoms, ocular and dermal effects, and reproductive effects.
13 The toxicity data on inhalation exposures to arsenic are limited. Long-term
14 occupational exposure leads to lung cancer and causes skin changes and peripheral nerve
15 damage in workers. Respiratory tract tumors occurred in hamsters exposed to intratracheal
16 doses of arsenic combined with a carrier dust.
17 Data on barium are extremely limited, with no epidemiological data available and no
18 standard inhalation toxicity studies in animals. Occupational case studies, supported by
19 histopathological studies in rats, indicate that the respiratory tract is a target for barium
20 compounds.
21 The kidney and the respiratory tract are the primary target organs for cadmium by
22 inhalation exposure in the human; toxicity is dependent on cumulative exposure, with renal
23 tubular dysfunction and associated increased excretion in urine of proteins, amino acids, and
24 essential metals being key outcomes of long-term exposures. Acute high-level exposure in
25 humans causes intense respiratory tract irritation, and milder effects on pulmonary function
26 follow chronic low-level exposure. Rat studies show that cadmium can cause lung cancer;
27 there is evidence that lung cancer has been observed in humans following high occupational
28 exposure.
29 The respiratory tract is the primary target for inhaled chromium compounds. Human
30 and animal data agree on the nature of nasal effects. Laboratory animal studies have
31 reported lung lesions and evidence of inflammation. Human and animal data agree that
April 1995 1-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cr(VI) compounds cause lung cancer. Human studies have also reported early signs of renal
2 damage with exposure to Cr(VI) compounds.
3 The respiratory tract is the major target for inhaled cobalt compounds. In humans, two
4 major types of effects are observed, interstitial lung disease and asthma. Inflammation and
5 decreased lung function have been observed in laboratory animals.
6 Limited data support the respiratory system as a major target for inhaled copper and
7 copper compounds. In humans the data are limited to subjective reporting of respiratory
8 symptoms and radiographic evidence of pulmonary effects after acute and chronic inhalation.
9 The respiratory tract is the primary target for iron oxides by inhalation exposure. In
10 humans, respiratory effects have been reported in workers chronically exposed to iron dust.
11 In laboratory animals, hyperplasia and alveolar fibrosis have been reported after inhalation or
12 intratracheal administration of iron oxide.
13 The nervous system is the most sensitive target for elemental mercury following acute
14 or chronic inhalation exposures. Effects range from reversible neurological symptoms to
15 psychomotor and neurobehavioral changes and peripheral nerve dysfunction. Respiratory,
16 gastrointestinal, and cardiovascular symptoms have also been reported in case reports and
17 occupational studies with exposure to high concentrations of mercury. The kidney is a
18 sensitive target toxicity following elemental mercury exposure in humans.
19 The nervous system and the respiratory tract are primary targets for inhaled manganese.
20 Acute occupational exposures are associated with pneumonitis, while chronic exposures
21 mainly impact the central nervous system. Limited information suggests that prenatal and/or
22 postnatal exposure of laboratory rodents to inhaled manganese oxide may depress
23 neurobehavioral activity.
24 Limited data on the inhalation of magnesium and its compounds support the respiratory
25 tract as a target. Acute high-level exposure of humans or laboratory animals to magnesium
26 oxide fume results in a reaction similar to zinc oxide metal fume fever. Suggestive evidence
27 indicates chronic exposure to magnesium dusts may produce pneumoconiosis. In laboratory
28 animals, fibrosis is observed with chronic exposure to high levels of magnesium dusts.
29 The respiratory tract appears to be the main target in humans and animals after
30 inhalation exposure to molybdenum compounds; however, inhalation exposure to
31 molybdenum has also been associated with nonspecific effects in humans including general
April 1995 1-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 weakness and dizziness. Animal inhalation studies indicate that toxicity varies with the
2 molybdenum compound.
3 The respiratory tract is the primary target for nickel compounds following inhalation
4 exposure. In humans, respiratory effects include asthma and altered pulmonary function. In
5 laboratory animals, inflammatory responses suggest an immunological response in the lungs.
6 Immunological changes have been reported in refinery workers exposed to nickel. The
7 potential for lung and nasal cancer was evident in occupational and laboratory animal studies.
8 When ingested in relatively low concentrations, potassium is an essential metal, but
9 available data on inhaled potassium compounds are insufficient to assess toxicity. Similarly,
10 with inhalation exposure, the respiratory tract is the target for selenium, another essential
11 metal. In humans, respiratory effects have been reported in workers chronically exposed to
12 selenium; and similar effects have been reported in laboratory animal studies.
13 Inorganic tin is relatively inert lexicologically, and effects are limited to mild
14 respiratory effects, along with the formation of radio-opaque nodules in the lungs. No other
15 target systems for inhalation exposure to inorganic tin have been reported. Limited data
16 indicate the nervous, hepatic, renal, and respiratory systems are targets for inhalation
17 exposure to organotin compounds. The respiratory tract is the primary target for titanium
18 following inhalation exposure. No histopathology of other organs was found in rats
19 chronically exposed to titanium tetrachloride at up to 6,000 /igTi/m3. Titanium is not
20 translocated in the body, even with chronic exposure and high concentrations. Titanium
21 dioxide inhalation results in pneumoconiosis in humans and signs of inflammation in
22 laboratory animals.
23 The respiratory tract is the primary target for inhaled vanadium compounds. Vanadium
24 damages alveolar macrophages, and toxicity is related to compound solubility and valence.
25 Human occupational case studies and epidemiological studies indicate symptoms of
26 respiratory distress. Symptoms of systemic effects have been observed following chronic
27 occupational exposure and in laboratory animal studies.
28 Following inhalation exposure, the respiratory tract is the primary target for zinc,
29 another essential metal. In humans, metal fume fever, characterized by respiratory
30 symptoms and pulmonary dysfunction, was observed in workers and experimental subjects
April 1995 1_28 DRAFT-DO NOT QUOTE OR CITE
-------
1 during acute exposures to high levels of zinc oxide. Zinc can produce inflammatory
2 responses in both human and animal species.
3
4 Silica
5 Silica can occur in two chemical forms, amorphous and crystalline. Crystalline forms
6 include quartz, which is the most prevalent; cristobalite; tridymite; and a few other rare
7 forms. Freshly fractured crystalline silica is more lexicologically reactive than aged forms of
8 crystalline silica or forms that may be coated with other chemical compounds. Amorphous
9 silica is less well studied and may have similar toxic endpoints but is less potent than
10 crystalline silica. With sufficient exposure, crystalline silica is toxic to the respiratory
11 system. Acute high exposure in both humans and animals causes lung inflammation and, if
12 the exposure is high enough, rapid onset of a fibrotic lung disease (acute silicosis) which can
13 be fatal. Occupational studies show that chronic exposure to crystalline silica causes
14 inflammation of the lung which is followed by fibrosis and a human fibrotic disease called
15 silicosis which can lead to early mortality. Silocotic individuals are also at higher risk for
16 other diseases, e.g., tuberculosis. Some occupational studies also show a concurrent
17 incidence of lung cancer.
18 The role, if any, of silica-induced lung inflammation, fibrosis, and silicosis in the
19 development of lung cancer is hypothesized but not adequately demonstrated. Crystalline
20 silica interaction with DNA has been shown. Chronic exposure animal studies in rats also
21 show a similar pattern of lung inflammation, fibrosis, and lung cancer. In 1987, the
22 International Agency for Research on Cancer classified crystalline silica as a "possible"
23 human carcinogen owing to a sufficient level of evidence in animal studies but inadequate
24 evidence in human studies. The evidence for amorphous silica carcinogenicity was said to be
25 inadequate for both humans and animals, placing it in Group 3 (agent not classifiable).
26 While active surveillance of the U.S. population for fibrosis and silicosis is not standard
27 practice, U.S. health statistics do not reveal a general population increase of crystalline silica
28 diseases. However, there is an increase in these diseases among the occupational work
29 force.
30 An assessment of the occupational risk of silicosis was made using recent studies from
31 South Africa and Canada, both of which examined medical histories of over 2000 miners.
April 1995 1-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 Both predicted zero risk for cumulative silica exposures of 0.6 mg/m3 • years (equivalent to
2 a 20-year workplace exposure to an average concentration of 30 /zg/m3). At higher
3 exposures, excess risk was observed in these workers (e.g., 2% risk at 1.6 mg/m3 • years).
4 These effective occupational exposures are greater and the particle sizes smaller than those
5 likely to be experienced by the public; however, the public would be expected to include
6 susceptible subpopulations. Information gaps still exist for both the exposure-response
7 relationship (especially in potentially susceptible subgroups) for levels of exposure within the
8 general population.
9
10 Asbestos
11 The mechanisms underlying the development of asbestos-induced pulmonary fibrosis in
12 rats is complex. While the acute response to asbestos results in pulmonary inflammation and
13 cell proliferation, the pattern of fibrosis following chronic exposures becomes more complex.
14 It is likely that the retention of inhaled fibers and consequent accumulation of interstitial
15 fibers concomitant with prolonged inflammation will contribute to the development of a
16 diffuse and progressive pattern of pulmonary fibrosis. The pathogenesis of asbestos-related
17 lung tumors clearly is a complex process and requires further investigation.
18
19 Ultrafine Particles
20 Certain freshly-generated ultrafine particles when inhaled as singlets at very low mass
21 concentrations (10 to 50 /xg/m3) can be highly toxic to the lung. Mechanisms responsible for
22 this high toxicity could include (1) high pulmonary deposition efficiencies of these particles
23 (2) the large numbers per unit mass of these particles, (3) their increased surface area
24 available for reaction, and (4) the presence of radicals on the particle surface, depending on
25 the process of generation of the particles. Results of studies with ultrafine model particles
26 indicate that particle number may be of more import as a dose parameter, than just particle
27 mass.
April 1995 1-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.12 EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
2 WITH EXPOSURE TO AIRBORNE PARTICLES/ACID AEROSOLS
3 Chapter twelve assesses epidemiological evidence relating human health effects to
4 exposure to airborne particles, which constitutes some of the most crucial information
5 assessed in this latest PM criteria review. Much new information has appeared since EPA's
6 publication of the 1982 document on Air Quality Criteria for Paniculate Matter, and sulfur
7 oxides (PM/SOX), its second Addendum (1986), and a later Acid Aerosol Issue Paper (1989).
8 A rapidly growing body of epidemiologic literature examines relationships between PM
9 concentrations and human health effects, ranging from respiratory function changes and
10 symptoms to exacerbation of respiratory disease and excess mortality associated with
11 premature death.
12 The time-series mortality studies reviewed in this and past criteria documents provide
13 evidence that PM-containing ambient air pollution can cause increases in daily human
14 morbidity and mortality. The newly available epidemiology studies provide indications that
15 very small increases in relative risk for such effects are associated with ambient air pollutant
16 mixtures containing low or moderate concentrations of PM, as indexed by a variety of
17 monitoring methods (e.g., black smoke, TSP, COH, PM10,PM25). This includes emerging
18 new evidence indicating likely associations of health effects with PM across a wide range of
19 routine ambient concentration levels seen in the United States and other countries, including
20 levels that extend below present U.S. PM air quality standards.
21 The 1982 EPA criteria document earlier concluded that the most clearly defined effects
22 on mortality arising from exposure to PM were sudden increases in the number of deaths
23 occurring, on a day-to-day basis, during episodes of high pollution, as occurred in the Meuse
24 Valley in 1930, in Donora in 1948, and in London in 1952. During the December, 1952
25 London episode 3,000 to 4,000 excess deaths were attributable to air pollution, with the
26 greatest increase in the death rate most dramatic for those > 45 years old and occurring
27 most notably in those with chronic lung disease and heart disease due to pollution-induced
28 cardio-respiratory problems. Other episodes with associated notable increases in mortality
29 occurred in London during various winters from 1948 to 1962. Collectively, studies of these
30 and other early episodes left little doubt that airborne particles contribute to mortality
April 1995 1-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 associated with very high concentrations of urban aerosol mixes dominated by combustion
2 products (e.g., from burning coal) and/or their transformation products (e.g., H2SO4).
3 Besides evaluating mortality associated with major episodes, the 1982 criteria
4 document also focused on studies of more moderate day-to-day variations in mortality within
5 large cities in relation to PM pollution. Various methodological problems were identified for
6 most of the studies, precluding quantitative conclusions regarding exposure-response
7 relationships of importance for deriving air quality standards. Among the main problems
8 were inadequate measurement or control for potentially confounding variables and inadequate
9 quantification of exposure to airborne particles and other associated pollutants (e.g., sulfates
10 or acid aerosols). Despite such problems, the 1982 document concluded that the then
11 available studies collectively indicated that mortality was clearly and substantially increased
12 when airborne particle 24-h concentrations exceeded 1,000 /-ig/m3 (as measured by the black
13 smoke, or BS, method) in conjunction with sulfur dioxide (SO2) elevations in excess of 1,000
14 Mg/m3 (with the elderly or others with severe preexisting cardiovascular or respiratory
15 disease mainly being affected).
16 The 1986 addendum to the 1982 criteria document later considered several additional,
17 then-new analyses of acute PM exposure mortality in London during the 1958-1959 through
18 1971-1972 winter periods. After reviewing the new data analyses, and taking into account
19 the previously reviewed London results and the above noted methodological considerations,
20 the following conclusions were drawn:
21
22 (1) Markedly increased mortality occurred, mainly among the elderly and chronically
23 ill, in association with BS and SO2 concentrations above 1,000 /xg/m3, especially
24 during episodes when such pollutant elevations occurred for several consecutive
25 days;
26
27 (2) During such episodes, coincident high humidity or fog was also likely important,
28 possibly by providing conditions leading to formation of sulfuric acid (H2SO4) or
29 other acidic aerosols;
30
31 (3) Increased risk of mortality is associated with exposure to BS and SC^ levels in
32 the range of 500 to 1,000 /xg/m3, for SO2 most clearly at concentrations in
33 excess of =700 /ig/m3; and
34
35 (4) Convincing evidence indicates that relatively small, but statistically significant,
36 increases in the risk of mortality exist at BS (but not SO2) levels below 500
37 Mg/m3. with no indications of any specific threshold level having been
April 1995 1-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 demonstrated at lower concentrations of BS (e.g., at < 150 /*g/m3). However,
2 precise quantitative specification of the lower PM levels associated with mortality
3 is not possible, nor can one rule out potential contributions of other possible
4 confounding variables at these low PM levels.
5
6 In setting the current U.S. PM standards, the BS levels noted above were taken as indexing
7 particles roughly in the same size range as inhalable particles reaching tracheobronchial or
8 alveolar regions of the respiratory tract; and, taking into account other evidence of morbidity
9 effects (e.g., worsening of chronic bronchitis symptoms), the U.S. 24 h primary NAAQS
10 was set as 150 jig/m3 PM10.
11 The decade or so since the 1986 EPA Addendum has seen the publication of numerous
12 new time series analyses of associations between human mortality or morbidity and acute
13 exposures to PM concentrations at or below the lower end of the range indexed by the above
14 studies of London mortality or the level of the current U.S. 24-h standard. Some utilized
15 TSP or other measures (e.g., COH, BS, etc.) as an indices of PM exposure, but during the
16 last few years, the analyses have mainly focused on PM10 as a measure of PM.
17
18 Short-Term PM Exposure Mortality Studies
19 Based on the new time-series analyses, numerous investigators have reported very
20 small, but statistically significant associations between increased relative risk for mortality
21 and various indices of PM (e.g., BS, COH, TSP, PM10, PM2.5, etc.) for many different
22 cities in the United States and in other countries, as well. The elderly (>65 yr old),
23 particularly those with preexisting cardiopulmonary disease, are found to have distinctly
24 higher risks than younger age groups. The small relative risk estimates for PM are generally
25 reduced when other likely important (potentially confounding) factors are also controlled for
26 in the models, but the PM association still usually remains statistically significant, although
27 typically accounting for much less of the variance in mortality than temperature or
28 combinations of variables used to index contributions of weather-related mortality. Thus,
29 qualitatively, the newly emerging database appears to provide indications that polluted
30 atmospheres containing relatively low concentrations of particles may contribute (along with
31 other factors) to a very small increase in relative risk for human mortality, especially in the
32 elderly with preexisting cardiopulmonary diseases.
April 1995 1_33 DRAFT-DO NOT QUOTE OR CITE
-------
1 It is important to note that differences in opinion exist within the scientific community
2 with regard to: (a) how adequately other likely important confounding factors (including
3 weather and copollutants) and/or other seasonal factors were controlled for across the various
4 new analyses; and (b) interpretation of reported associations between increased relative risk
5 estimates and indices of ambient PM concentrations. For example, introduction of one or
6 more other commonly-present ambient air pollutants (e.g., SO2, O3, CO, NOX) into models
7 of PM effects generally reduces the estimated PM effect, often by as much as 50% and, at
8 times, to statistically non-significant levels. In a few studies, however, the size of the PM
9 effect remained essentially the same or increased slightly with other copollutants in the
10 model. Similarly, analyses of PM-mortality effects by season (winter, spring, summer, fall),
11 as in a few studies, so far have yielded varying patterns of PM-mortality effects being
12 significant in one or another season(s) but not all, with specific effective seasons differing
13 from one locale to another. The copollutant and seasonality analyses results, in particular,
14 have led to considerable debate in the scientific community, typified on the one hand by (a)
15 skepticism about the size and the "realness" of reported low-level PM effects and, on the
16 other hand, (b) countervailing views asserting that the effect of PM (or any other weakly
17 contributing factor) on mortality can be made to "disappear" by overspecification of
18 applicable models (i.e., by introduction of sufficient other, possibly extraneous, variables
19 into the models or by more detailed breakdowns of data (e.g., by season) that may reduce
20 the power to detect a PM effect).
21 No clear resolution of this debate or "consensus" opinion in the scientific community
22 has yet crystallized, but some agreement appears to be emerging that the results for models
23 containing only PM and no other copollutants may provide upper bound estimates for effects
24 of ambient particle-containing mixes of pollutants, whereas results derived from analyses
25 including other copollutants and extensive controls for weather, seasonality, and/or other
26 likely important contributing factors should be viewed as lower-bound estimates of PM
27 effects (which may be 50% or more lower than the upper bound or even include zero). Key
28 points regarding derivation of quantitative estimates of PM-related mortality and morbidity
29 effects, taking into account the types of uncertainties and scientific debate just noted are
30 summarized below. Major emphasis is placed first on results derived from studies of PM10
April 1995 1-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 as the PM indicator of most interest in relation to the current U.S. PM standards and, then,
2 additional key quantitative points for PM2 5 and acid aerosols are summarized.
3
4 PM10 Relative Risk Analyses
5 This section discusses key findings from studies conducted since the 1986 PM criteria
6 document addendum that have employed PM10 in their analyses of the human mortality
7 effects of acute exposures to PM, as discussed in more detail in Chapter 12. Some studies
8 considered daily mortality in the entire population (i.e., all ages) and some by cause; some
9 also considered subpopulations (e.g, the elderly).
10 Two earlier published summaries of the PM literature converted all results to a
11 PM10-equivalence basis and provided quantitative intercomparisons and after such summaries
12 used TSP as the reference PM metric. The results from such summaries suggest about a 1
13 percent change in acute total mortality for a 10 /xg/m3 change in PM10, but the estimates
14 range from 0.3 to 1.6% (i.e., a factor of 5). While most of the 95% confidence intervals
15 (CIs) of these estimates overlap, CIs of the highest and lowest estimates do not overlap,
16 indicating significant differences between these estimates. Note that the effects indicated for
17 a 10 /ig/m3 PM10 change cannot be reliably converted to other PM increments (e.g., 50 or
18 100 /xg/m3 PM10), as differences in model specification (e.g., linear versus log models) will
19 cause them to differ in their conversions to other particle concentration ranges. The reasons
20 for the approximately five-fold effect estimate difference noted among studies are not
21 obvious, but one factor appears to be the PM exposure averaging time, as estimates using
22 multiple day PM10 averages are all 1 % or higher. This is not unexpected, given that any
23 lagged effects from prior days of PM10 exposure will be added to the effects estimate when a
24 multi-day average is employed, increasing the estimated effect on a per /xg/m3 basis.
25 It is also important to note that other air pollutants were generally not addressed in
26 deriving the coefficients reported by the above summaries. Differences among coefficients
27 are to be expected, given that the composition (and, potentially toxicity) of the PM, as well
28 as the demographic characteristics in each city, can be expected to differ. Moreover, the
29 conversions from other PM metrics to PM10 necessarily introduce much additional
30 uncertainty. However, though not all of these results may therefore be the most appropriate
31 available for quantifying a PM10 effect, they do consistently indicate that there is an
April 1995 1_35 DRAFT-DO NOT QUOTE OR CITE
-------
1 association between acute exposure to ambient air mixes containing PM and increased daily
2 mortality. Moreover, the by-cause results also reported in the summaries indicate that PM
3 effect estimates are greater for respiratory causes, which lends support to the biological
4 plausibility of the noted PM associations.
5 In an effort to more clearly quantify daily PM10-total acute mortality associations,
6 Table 1-5 summarizes total mortality relative risk (RR) estimates reported for a 50 /xg/m3
7 increase in PM10 estimated from nine studies reviewed in Chapter 12, which employed PM10
8 data in their analysis of total mortality data (or which had on-site PM10 reference data to
9 convert other PM metrics with more certainty). The studies listed were selected for this
10 analysis mainly because they can most readily be intercompared and provide direct data
11 related to PM10 levels. The RR's calculated were based upon a 50 /ig/m3 increase above the
12 mean PM10 24-h concentration, which is approximately the order of magnitude of the typical
13 difference between the mean and maximum in these cities evaluated and is fairly
14 representative of the mean/max differences seen for most of the U.S. cties in Tables 1-2 and
15 1-3. This is noted because in non-linear models as often employed in the studies discussed in
16 Chapter 12, the RR estimate associated with a given /ng/m3 PM10 increase will vary
17 depending upon the baseline concentration to which it is added.
18 From results presented in Chapter 12, it is apparent that these studies generally have
19 yielded at least marginally significant PM10 coefficients, but the resultant excess risk
20 estimates differ by a factor of five across these studies (from 1.5% to 8.5% per 50 /ig/m3).
21 The mean and maximum PM10 concentration data are noted for each study. If the PM10
22 coefficient decreased as the mean level of PM10 decreased, then confounding as a function of
23 varying PM level would be suggested. However, the data presented indicate that the
24 variability in coefficients is not a function of PM10 level, as sites with high or low PM10
25 concentrations can report either high or low RR's. In Chapter 12, the statistical
26 methodology characteristics of each study were concisely summarized, in order to determine
27 if any factors are important to help explain the variability observed from study to study in the
28 PMio RR estimate. As noted earlier, the RR estimate for acute mortality associated with
29 short-term exposure to PM10 is usually larger when other pollutants are not simultaneously
30 considered in the model. Those studies which considered PM10 both alone and with other
31 pollutants in the model often yielded smaller, and usually more marginally significant, PM10
April 1995 1_36 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-5. COMPARISON OF RELATIVE RISK (RR) ESTIMATES FOR TOTAL MORTALITY FROM 50 jtg/m3
CHANGE IN PM10, USING STUDIES WHERE PM10 WAS MEASURED OR WAS CALIBRATED FOR THE SITE
Co
H
O'
O
1
H
O
e:
H
tn
O
Q
H
W
Study
Utah Valley, UT
St. Louis, MO
Kingston, TN
Birmingham, AL
Athens, Greece
Toronto, ON Canada
Los Angeles, CA
Chicago, IL
Santiago, Chile
Chicago, IL
'Calculated on a basis of 50
Reference PM10 (/ig/m3)
Mean Maximum
Pope et al.(1992) 47 297
Dockery et al. (1992) 28 97
Dockery et al. (1992) 30 67
Schwartz (1993) 48 163
Touloumi et al. (1994) 78 306
Ozkaynak et al. (1994) 40 96
Kinney et al. (1995) 58 177
Ito et al. (1995) 38 128
Ostro et al. (1995a) 115 367
Styer et al. (1995) 37 365
/ig/m3 increase from 50 to 100 jug/m3.
Other Pollutants
In Model
None
None, Winter
None, Summer
Max O3, Summer
Avg O3, Summer
None
03
None
03
None
None
SO2, CO
None
None
03, CO
O3, CO
None
None
None, Poisson
SO2, Poisson
NO2, Poisson
O3, Poisson
None
Lag Times, d
< 4 d
< 4d
< 4 d
< 4 d
< 4 d
< 3 d
< 3 d
< 3d
< 3d
< 3 d
1 d
1 d
Od
1 d
1 d
< 3 d
1 d
< 4 d
1 d
1 d
1 d
1 d
3 d
RRper
50 /ig/m3
1.08
1.085
1.11
1.19
1.14
1.08
1.06
1.085
1.09
1.05
1.034
1.015
1.025
1.025
1.017
1.025
1.04
1.07
1.0221
1.0261
1.0431
1.0261
1.04
95 Percent
Confidence Interval
(1.05, 1.11)
(1.03, 1.14)
(0.92, 1.35)
(0.96, 1.47)
(0.92, 1.41)
(1.005, 1.15)
(0.98, 1.15)
(0.94, 1.25)
(0.94, 1.26)
(1.01, 1.10)
(1.025, 1.044)
(1.00, 1.03)
(1.015, 1.034)
(1.00, 1.055)
(0.99, 1.036)
(1.005, 1.05)
(1.005, 1.06)
(1.04, 1.10)
(1.003, 1.042)
(1.005, 1.047)
(1.020, 1.066)
(1.005, 1.047)
(1.00, 1.08)
-------
1 relative risks when other pollutants were also considered. This ranges from roughly a 20 to
2 50 percent reduction in the estimate of excess risk associated with PM10 (e.g., in Athens,
3 Greece, the PM10 RR declines from 1.07 to 1.03 per 100 jug/m3 when other pollutants are
4 considered). Such a reduction is to be expected when co-linear variables are added.
5 Older studies using BS or TSP often found high correlations between SO2 and the PM
6 indicator which reduced the apparent PM effect and attenuated its statistical significance.
7 However, studies using a variety of PM indicators at cities were SO2 levels were so low as
8 to have little likelihood of SO2 being a significant confounder of a PM effect found
9 quantitatively similar significant PM effects. While there is some possibility that
10 summertime PM effects may be partially confounded with those of other pollutants (e.g., O3)
11 derived from motor vehicle fuel combustion or transformation products, winter effects of PM
12 are clearly detectable when O3 levels are much lower. If PM effects on mortality were so
13 completely confounded with those co-pollutants so as to be undetectable, then one would
14 need to invoke many different confounders in different studies of communities. While this
15 explanation is not impossible, it appears highly unlikely, but cannot be precluded altogether
16 since PM may derive from different sources in these studies, have varying size and chemical
17 composition from one locale to another, and therefore may have different characteristics that
18 affect health outcomes such as mortality.
19 Another factor clearly affecting the PM10 RR estimates is the PM10 averaging period.
20 Most of the studies which utilized multi-day averages of PMjo in their regressions (i.e., for
21 Utah Valley; St. Louis; eastern TN; Santiago; Chicago; and Birmingham) yielded higher RR
22 estimate studies. However, the increase indicated for these studies is not proportional to the
23 averaging time. Indeed, in sub-analyses for Utah Valley data, the PM10 mortality risk is
24 indicated to be roughly doubled by using a five day average versus a single day
25 concentration, and sub-analyses for Santiago also indicate approximately a doubling in the
26 PMio RR when a 3 day average is considered (i.e., from RR = 1.04 for a single day PM10
27 value to RR = 1.07 for a 3d average PM10 value). This may be due to the fact that, since
28 autocorrelation exists in the PM10 concentrations from day to day, the single day
29 concentration is "picking up" some of the effect of multi-day pollution episodes, even though
30 they are not explicitly modeled. These results suggest that a multi-day rather than a
April 1995 1-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 single-day average PM10 concentration may provide a more relevant index to gauge the
2 effects of short-term PM exposures over several consecutive days.
3 Table 1-5 shows that total acute mortality relative risk estimate associated with a 50
4 Mg/m3 increase in the one-day 24-h average PM10 ranged from 1.015 to 1.085, depending
5 upon the site (i.e., the PM10 composition and population demographics) and whether PM10 is
6 modeled as the sole index of air pollution. Relative Risk estimates with PM10 as the only
7 pollutant index in the model range from RR = 1.025 to 1.085, while the PM10 RR with
8 multiple pollutants in the model range from 1.015 to 1.025. As noted earlier, the former
9 range might be viewed as approximating an upper bound of the best estimate, as any
10 mortality effects of co-varying pollutants are likely to be "picked up" by the PM10 index.
11 On the other hand, the latter multiple pollutant model range might be viewed as
12 approximating a lower bound of the best estimate, as the inclusion of highly correlated
13 covariates may weaken the PM10 estimate. Overall, consistently positive PM-mortality
14 associations are seen throughout these analyses, even with the use of various modeling
15 approaches and after controlling for major confounders such as season, weather, and
16 co-pollutants. The 24-h 50 jig/m3 PM10 total mortality effect estimate most typically falls in
17 approximately the RR = 1.025 to 1.05 range (representing an expected 2.5 to 5.0% increase
18 in risk of death over daily background mortality rates for which a 50 /xg/m3 increment in
19 ambient PM10 concentration could be a contributing factor).
20 It is logical to assume that the bulk of the total mortality effects suggested by these
21 studies are among the elderly. During the historic London, 1952 pollution episode the
22 greatest increase in the mortality rate was among older citizens and those having respiratory
23 diseases. An analysis of mortality in Philadelphia, PA during 1973 through 1980 comparing
24 mortality during the 5% highest versus the 5% lowest TSP days also found the greatest
25 increase in risk of death to be among those aged 65 to 74 and those > 74 year of age
26 (mortality risk ratios = 1.09 and 1.12, respectively, between high and low TSP days). Also,
27 in a time series analyses of Philadelphia daily mortality during this period, the TSP-mortality
28 coefficient was significantly higher (6 = 0.000910 ± 0.000161) for persons > 65 yrs old
29 than for the younger population (6 = 0.000271 + 0.000206). These coefficients indicate an
30 effect size for the elderly roughly three times that for the younger population (10% versus
31 3%, respectively, for a 100 pig/m3 increase in TSP). Also, two other recent PM10 analyses
April 1995 1_39 DRAFT-DO NOT QUOTE OR CITE
-------
1 (one for Sao Paulo, Brazil and the other for Santiago, Chile) directly considered the question
2 of PM10-mortality associations among the elderly population (> 65 years of age). The first
3 of these two analyses evaluated PM10-mortality associations for the elderly in Sao Paulo
4 during 1990 through 1991 in Sao Paulo. The reported PM10 relative risk (RR = 1.13 for a
5 100 jitg/m3 increase) is higher than noted above for total mortality analyses addressing
6 multiple pollutants (100 /xg/m3 RR = 1.03 to 1.05), supporting past observations that the
7 elderly represent a population especially sensitive to the health effects of air pollution. The
8 second study in Santiago, Chile found a 24-h PM10 100 peg/in3 RR estimate of 1.08, for the
9 overall population but the RR estimates for a 100 /ig/m3 increase in PM10 rose to an RR =
10 1.11 for the elderly (aged 65+) using the same model specification. Thus, these directly
11 comparable estimates (i.e., using the same model specification and population) suggest that
12 the elderly experience roughly a 40 percent higher excess risk from exposure to PM air
13 pollution than the overall population.
14 Overall, considering the historical pollution episode evidence and the results of recent
15 PM10-mortality analyses evaluating elderly populations, it seems evident that elderly adults
16 represent a population especially at risk for mortality implications of acute exposure to air
17 pollution, including PM.
18 Relatively few studies have directly examined the PM-mortality association in children.
19 It is difficult, given the limited and somewhat conflicting results available at this time, to
20 ascribe any such association to low-level PM pollution in particular. This is an area where
21 further research is clearly needed to broaden the base upon which to assess the potential for
22 PM to increase mortality among children.
23 In studies alluded to above and in others disused in more detail in Chapter 12, a
24 consistent trend was for acute PM exposure effect estimates to be higher for the respiratory
25 mortality category than for total mortality from all (non-accident) causes. This lends support
26 to the biological plausibility of a PM air pollution effect, as the breathing of toxic particles
27 would be expected to most directly affect the respiratory tract. Of particular interest is
28 comparison of relative risk values from those studies that made most direct and appropriate
29 comparisons. In a Santa Clara study, the PM-respiratory mortality RR was 4.3 times as
30 large as for deaths as a whole (i.e., 3.5/0.8); for Philadelphia, the PM (TSP)-respiratory
31 mortality RR was 2.7 times as large as for total mortality (i.e., 3.3/1.2); for Utah Valley,
April 1995 1-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 the PM10-respiratory mortality RR was 2.5 times as large as for deaths as a whole (i.e.,
2 3.7/1.5); for Birmingham, AL, the respiratory mortality RR of PM10 was 1.5 times as large
3 as for deaths as a whole (i.e., 1.5/1.0); and for Santiago, Chile the reported excess
4 respiratory mortality RR of PM10 was 1.8 times as large as for deaths as a whole (comparing
5 1.15/1.08 RR per 100 /*g/m3). Thus, in these studies, the PM RR for respiratory diseases is
6 indicated to range from 50 to over 400% higher for respiratory disease categories than for all
7 causes of death, indicating that increases in respiratory deaths are a major contributor to the
8 overall PM-mortality associations noted previously. Moreover, since evidence suggests that
9 an acute pollution episode is most likely be inducing its primary effects by stressing already
10 compromised individuals (rather than, for example, inducing chronic respiratory disease from
11 a single air pollution exposure episode), the above results indicate that persons with
12 pre-existing respiratory disease represent a population especially at risk to the mortality
13 implications of acute exposures to air pollution, including PM.
14 In overall summary, the time-series mortality studies reviewed in this and the previous
15 1982 and 1986 PM criteria assessments provide reasonably strong evidence that increases in
16 daily human mortality are associated with short-term exposures to air pollution mixes
17 containing elevated PM levels. Recent studies provide indications that small increases in
18 such risk occur in association with air pollution indexed by moderate increases of 24-h PM10
19 (~50 Mg/m3) above routine ambient levels averaging around 50 to 100 /-tg/m3. Overall, the-
20 PM10 relative risk estimates derived from the most recent PM10 total mortality studies
21 suggest an acute exposure effect on the order of RR = 1.025 to 1.05 in the general
22 population for increases in ambient air pollution indexed by a 24-h average 50 /*g/m3 PM10
23 increment, with higher (30-40%) relative risks indicated for the elderly sub-population and
24 for those with pre-existing respiratory conditions.
25
26 Fine Particles/Acid Aerosols Relative Risks
27 As noted earlier and in both Chapters 11 and 12, some epidemiologic and experimental
28 toxicology data point toward fine particles as a class or certain constituents (e.g., acidic
29 aerosols) as possibly being key contributors to observed PM-mortality and/or PM-morbidity
30 associations. Only a few epidemiologic studies provide direct comparisons between various
31 PM indices, including fine particle and acidity measurements.
April 1995 1-41 DRAFT-DO NOT QUOTE OR CITE
-------
1 One such study investigated relationships between multiple air pollutants (including
2 PM2.s) and total daily mortality during a one year period in St. Louis, MO and
3 Kingston/Harrirman, TN and surrounding counties. In Poisson regressions controlling for
4 weather and season, previous day's PM10 was the only significant predictor of daily mortality
5 (B = 0.00175 ± 0.00067), but the association dropped off at 3 days (fl = 0.00042 +
6 0.00063). Size-fractionated PM data were examined to determine whether this association
7 could be attributed to either the fine (PM2 5, aerodynamic diameter d& < 2.5 ju.m) or the
8 coarse (2.5 /mi < Ja < 10 ^im) component of the PM10 mass. The fine fraction (PM2 5)
9 was positively associated with mortality (6 = 0.00171 + 0.00096, P = 0.075). Coarse
10 particles were also positively associated (6 = 0.00247 + 0.00129, P = 0.056). Neither fine
11 nor coarse particles showed a stronger association than the other when considered
12 simultaneously.
13 Both daily SO4 and H+ concentrations were significantly correlated with PM10 (Pearson
14 correlations 0.52 and 0.76, respectively). Sulfate (SO42~) as measured by the sulfur fraction
15 of PM10 (/? = 0.00608 + 0.00577) and H+ (6 = 0.00086 + 0.00118) were positively, but
16 not significantly, associated with daily mortality. Among other PM elements measured,
17 those correlated with PM10 concentrations were also associated with mortality. In particular,
18 aluminum, calcium, chromium, iron, and silica all had correlations with PM10 of 0.5 or
19 higher and had positive associations with mortality. Neither SO2, NO2, nor O3 was
20 significantly associated (P > 0.30) with total mortality.
21
22 Long-Term PM10/PM2 5 Exposure Mortality Studies
23 Population-Based Cross Sectional Mortality Studies
24 Ecological cross-sectional studies employing averages across various geopolitical units
25 (cities, SMS As, etc.) present data that examine relationships between community-wide PM
26 levels and mortality. Such community-based studies seek to define (average) community
27 characteristics associated with overall average health status-in this case, annual mortality rate.
28 One study analyzed 1980 total mortality in 98 SMSAs, using data on PM15 and PM2 5
29 from the EPA inhalable particle (IP) monitoring network for 38 of these locations, ranked the
30 importance of the pollutants by relative statistical significance in separate regressions, and
31 concluded that the results were "suggestive" of an effect of particles on mortality decreasing
April 1995 1-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 with particle size; although in the basic model only SO42" was statistically significant. In
2 some other models tested, PM2 5 was also significant, and PM15 nearly so. However, if the
3 effects are judged by elasticities rather than significance levels, SO42"' PM2 5, and PM15
4 would be judged as equivalent, with TSP ranking somewhat lower. Also, based on source
5 apportionment techniques, particles from coal combustion and from the metals industry
6 appeared to be the most important. The specific coefficients and significance levels obtained
7 for TSP may be the result of the particular TSP data used, being based on a single
8 monitoring station in each SMSA and thus not clearly fully representative of population
9 exposures. Thus, alternative interpretations of these findings are certainly possible. In
10 addition, because smoking, diet, and other socioeconomic or lifestyle variables were not
11 considered in the regression model, the pollution coefficients may have been biased. Finally,
12 the study did not specifically address the question of acute vs. chronic responses by exploring
13 lagged pollution variables.
14 Data from up to 149 metropolitan areas (mostly SMSAs) were analyzed in another
15 study of relationships between community air pollution and "excess" mortality due to various
16 causes for the year 1980. Several socioeconomic models were used in cross- section multiple
17 regression analyses to account for non- pollution effects. Two different sources of
18 (measured) air quality data were used: data from the EPA AIRS database (TSP, SO42, Mn,
19 and ozone) and data from the inhalable paniculate (IP; PM15) network; the latter data
20 (PM15,PM2.5 and SO4= from the IP filters) were only available for 63 locations. All PM
21 data were averaged across all monitoring stations available for each SMSA, with TSP data
22 restricted to 1980 and based on an average of about 10 sites per SMSA. Using these models,
23 statistically significant associations were found between TSP and mortality due to non-
24 external causes with the log-linear models evaluated, but not with a linear model. Sulfates,
25 manganese, inhalable particles (PM15), and fine particles (PM2 5) were not significantly (P <
26 0.05) associated with mortality with any of the parsimonious models, although PM2 5 and
27 manganese were close with linear models (p=0.07) and significance may have been affected
28 by the use of smaller data sets. This study found PM2 5 to be the "strongest" PM index with
29 linear models, but TSP with log-linear models. This study support previous findings of
30 associations between TSP and premature mortality.
31
April 1995 1_43 DRAFT-DO NOT QUOTE OR CITE
-------
1 Prospective Mortality Studies
2 Prospective studies consider data on the relative survival rates of individuals, as
3 affected by age, sex, race, smoking habits, and certain other individual risk factors. This
4 type of analysis has a substantial advantage over the above population-based studies, because
5 identification of actual decedents allows stratification according to important risk factors such
6 as smoking. However, since none of the prospective cohort studies had data on personal
7 exposures to air pollution, these studies are also considered to be "ecological."
8 Several newer prospective studies are of most interest. One such study followed
9 approximately 6,000 white, non-hispanic, nonsmoking, long-term California residents for 6
10 to 10 years, beginning in 1976. The study was designed to test the use of cumulative
11 exposure data as an explanatory factor for disease incidence and chronic effects. Only TSP
12 and ozone data were used to index pollution exposures; with reliance on oxidant
13 concentrations in the early part of the monitoring record. In a follow-up analysis, SO42",
14 PM10 (estimated from site-specific regressions on TSP), PM2 5 (estimated from visibility),
15 and visibility per se (extinction coefficient) were used to index PM exposure. No significant
16 associations with nonexternal mortality were reported, and only high levels of TSP or PM10
17 were associated with symptoms of asthma, chronic bronchitis or emphysema. The finding of
18 no association between long-term cumulative exposure to TSP or O3 and all natural-cause
19 mortality may be interpreted as showing the absence of chronic responses after 10 years but
20 not necessarily the absence of (integrated) acute responses, since coincident air pollution
21 exposures were not considered.
22 Another prospective study analyzed survival probabilities among 8,111 adults first
23 recruited in the mid-1970s in six cities in the eastern portion of the United States. The cities
24 are: Portage, WI, Topeka, KS; St. Louis, MO; Steubenville, OH, Watertown, MA, and
25 Kingston-Harriman, TN, two small towns southwest of Knoxville. These locations thus
26 comprise a transect across Northeastern and Northcentral United States, from suburban
27 Boston, through Appalachia, and into the upper Midwest. The adults were white, aged 25 to
28 74 at enrollment, and the final cohorts numbered 1,400 to 1,800 persons in each city.
29 Follow-up periods ranged from 14 to 16 years, during which from 13 to 22% of the
30 enrollees died. Of the 1,430 death certificates, 98% were located, including those for
31 persons who had moved away and died elsewhere. The bulk of the analysis was based on
April 1995 1-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 all-cause mortality and included individual characteristics of the members (and thus of the
2 decedents), e.g. smoking habits, an index of occupational exposure, body mass index, and
3 completion of a high school education. Air pollution effects were evaluated in two ways: by
4 evaluating the relative risks of residence in each city relative to Portage (the city with the
5 lowest pollution levels for most indices), and by including the community-average air quality
6 levels directly in the models.
7 Based on statewide mortality data, substantial differences in survival rates would be
8 expected across this transect of the Northeastern U.S. and were in fact observed. The long-
9 term average mortality rate in Steubenville was 16.2 deaths per 1,000 person-years; in
10 Topeka, it was 9.7, yielding a 67% variation in the range of annual average (crude) relative
11 risk across the six cities. After individual adjustment for age, smoking status, education, and
12 body-mass index, the range in average relative risk was reduced to 26%. The authors
13 reported that "mortality was more strongly associated with the levels of fine, inhalable, and
14 sulfate particles" than with the other pollutants (e.g., NOX, SO2, TSP), which they attributed
15 primarily to factors of particle size. For those three PM indices, relative risk estimates and
16 confidence limits based on the differences between air quality in Steubenville and in Portage
17 were calculated. Only small differences were found between many pollutants, including S02
18 and NO2, owing in part to the strong collinearity present. Neither mortality associations with
19 TSP nor with coarse particle fractions created by subtracting PM15 from TSP or PM2 5 from
20 PM15 were significant, suggesting that particles > 15 um may be less important. This
21 outcome may reflect in part greater spatial variability within the communities for these
22 measures. The non-sulfate portion of PM2 5 had the tightest confidence limits (SO42~ was
23 multiplied by 1.2 before subtraction, assuming an average composition of NH4HSO4).
24 However, all of the differences in relative risks and their confidence limits could have
25 occurred due to chance, given the availability of only 6 observations. No relationship was
26 found for aerosol acidity (H+), but only limited data were available.
27 The authors of this study appear to have made the most of the available individual data
28 on some of the most important mortality risk factors. They were quite cautious in their
29 conclusions, stating that the results only suggest that fine-particulate air pollution "contributes
30 to excess mortality in certain U.S. cities." There are several other important outcomes-
31
April 1995 !_45 DRAFT-DO NOT QUOTE OR CITE
-------
1 • None of the population subgroups examined appeared to be stastically significantly
2 more sensitive to air pollution than any other, although smokers had notably higher
3 relative risk estimates (with wide confidence intervals) than non-smokers.
4
5 • The implied regression coefficients are much larger (about an order of magnitude)
6 than those found in either type of population-based study noted earlier. This could
7 be interpreted as evidence that the chronic effects of air pollution far exceed the
8 acute effects, or that not all of the spatial confounding has been controlled. Use of
9 linear models for non-linear effects (body-mass index) and failure to control for
10 alcohol consumption, diet, exercise and migration may have contributed to the
11 relatively large effects indicated for air pollution.
12
13 • If the measured responses to air pollution truly are due to chronic PM exposure it is
14 logical to expect that cumulative exposure would be the preferred metric. However,
15 pollution levels 10 years before this study began were much higher in Steubenville
16 and St. Louis, as indexed by TSP from routine monitoring networks. For example,
17 annual average TSP in 1965 in Steubenville was about three times the value used to
18 index chronic exposure in the study. Estimates of previous levels of fine particles
19 are more difficult, but atmospheric visibility data suggest that previous levels may
20 have been higher in winter, but not necessarily in summer. These uncertainties
21 make it difficult to accept quantitative regression results based solely on coincident
22 monitoring data.
23
24 Because it seems unlikely that any of the above-noted shortcomings of this study could have
25 resulted in bias sufficient to reduce the risk estimates to levels less than those found in acute
26 mortality studies, the Six City study appears to provide support for the hypothesis that effects
27 indexed by results of long-term air pollution studies must also reflect the presence of acute
28 effects on mortality as integrated over the long term. Or, it may also be concluded that
29 support has been shown for the existence of chronic effects; these two possibilities are not
30 mutually exclusive. However, these conclusions must be qualified by the realization that not
31 all of the relevant socioeconomic factors may have been properly controlled in this study.
32 In a very recently reported 1995 study, 7-year survival data (1982 to 1989) for about
33 550,000 adult volunteers obtained by the American Cancer Society (ACS) were analyzed.
34 The Cox proportional hazards model was used to define individual risk factors for age, sex,
35 race, smoking (including passive smoke exposure), occupational exposure, alcohol
36 consumption, education, and body-mass index. The deaths, about 39,000 in all, were
37 assigned to geographic locations using the 3-digit zip codes listed at enrollment into the ACS
38 study in 1982. Relative risks were then computed for 151 metropolitan areas defined by
39 these zip codes and were compared to the corresponding air quality data (ca. 1980) derived
April 1995 1-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 from the EPA AIRS system for sulfates, as obtained from high-volume sampler filters for
2 1980, and the Inhalable Particulate Network for fine particles (PM2 5), measured by
3 dichotomous samplers during 1979-81. Causes of death considered included: all causes,
4 cardiopulmonary causes, lung cancer, and all other causes.
5 The adjusted total mortality risk ratios for the ACS study (computed for the range of
6 the pollution variables) were 1.15 (95% CL = 1.09 to 1.22) for sulfates and 1.17 (95% CL
7 = 1.09 to 1.26) for PM2 5. When expressed as log-linear regression coefficients, these
8 values were quite similar for both pollution measures: 0.0070 (0.0014) per /ig/m3 for SO42
9 and 0.0064 (0.0015) for PM2 5, suggesting that particle chemistry may be relatively
10 unimportant as an independent risk factor (it is possible that the SO42" results have been
11 biased high by the presence of filter artifacts). However, the pollution coefficients were
12 reduced by 10 to 15% when variables for climate extremes were^dded to the model.
13 This study took great care to control for those potential confounding factors for which
14 data were available. Several different measures of active smoking were considered, as was
15 the time exposed to passive smoke. The occupational exposure variable was specific to any
16 of: asbestos, chemicals/solvents, coal or stone dusts, coal tar/pitch/asphalt, diesel exhaust,
17 or formaldehyde. The education variable was an indicator for having less than a high-school
18 education. However, the possible influences of other air pollutants were not discussed, and
19 other risk factors not considered included income, employment status, dietary factors,
20 drinking water hardness and physical activity levels, all of which have been shown to affect
21 longevity. Another important caveat is that the ACS cohort is by no means a random sample
22 of the U.S. population; it is 94% white and better educated than the general public, with a
23 lower percentage of smokers than in the Six City Study. The (crude) death rate during the
24 7.25 years of follow-up was just under 1% per year, which is about 20% lower than
25 expected for the white population of the U.S. in 1985, at the average age reported for the
26 study cohort. In contrast, the corresponding rates for the Six-City study discussed above
27 tended to be higher than the U.S. average.
28 The results of the long-term prospective cohort studies are compared in Table 1-6. The
29 results of the American Cancer Society (ASC) prospective study were qualitatively consistent
30 with those of the Six City study with regard to their findings for sulfates and fine particles;
31 but relative standard errors were smaller, as expected because of the substantially larger
April 1995 1-47 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-6. PROSPECTIVE COHORT MORTALITY STUDIES
§
u\
Source
Abbey et
al. (1991)
Dockery et
al. (1993)
p. 1758
Health
Outcome
Total
mortality
from
disease
Total
mortality
Population
Calif. 7th
Day
Adventist
White
adult
volunteers
in 6 U.S.
cities3
Time
Period/ PM
No. Units Indicators
1977-82 24 h TSP
Defined by > 200
air
monitoring
sites
1974-91 PM15
PM2.5
SO4
PM
Mean
(Mg/m3)
102
29.9
18
7.6
PM Range/
(Std. Dev.)
25-175
(annual avg)
18-47
11-30
5-13
Sites
Per Total
City Deaths Model Type
NA 845 Cox
proportional
hazards
1 1429 Cox
proportional
hazards
PM Lag Other
Structure Pollutants Other Factors
10 yrs none age, sex, race,
smoking,
education,
airway disease
none none age, sex,
smoking,
education,
body mass,
occup.
exposure
hypertension4.
diabetes4
Relative
Risk1 at
S04 = 15,
PM15 = 50,
PM2 5 = 25
0.99 TSP1
1.42 PM15
1.31 PM25
1.46SO/
RR.
Confidence
Interval
(0.87-1.13)'
(1.16-2.01)
(1.11-1.68)
(1.16-2.16)
Elasticity
NS2
0.25
0.22
0.23
OO
Popeetal. Total American 1982-89 PM25 18.2 9-34
(1995) mortality Cancer PM25 50
Table 2 Society, cities
adult SO4 151 SO4 II5 4-24
volunteers cities
in U.S.
20,765 Cox
proportional
hazard
38,963
age, sex, race, 1.17PM25 (1.09-1.26) 0.117
smoking,
education,
body mass, 1.10SO4 (1.06-1.16) 0.077
occup.
exposure,
alcohol
consumption,
passive
smoking,
climate"
O
o
o
H
O
O
3
o
o
»—<
a
'For l,000h/yr > 200pg/m3.
2NS = non significant, confidence limits not shown.
3
Portage, WI; Topeka, KS; Watertown, MA; Harrisman-Kingston, TN; Steubenville, OH.
4Used in other regression analyses not shown in this table.
5Value may be affected by filter artifacts.
-------
1 database. However, no other non-PM pollutant measures were investigated in the ACS
2 analysis, so that no further progress was made in attempting to clarify relative contributions
3 of PM or other possible "responsible" pollutants. Also of note were that the ACS regression
4 coefficients were about 1/4 to 1/2 of the corresponding Six City values and were much closer
5 to the corresponding values obtained in various acute mortality studies. Thus it is not clear
6 to what extent chronic effects (as opposed to integrated acute effects) are indicated by these
7 results and to what extent the limited air quality data base used was responsible for this
8 outcome.
9 The California and Six-City studies both suffer from small sample sizes and inadequate
10 degrees of freedom, which partially offset the specificity gained by considering individuals
11 instead of population groups. All of them may have neglected some important risk factors.
12 The studies of California nonsmokers by Abbey et al. (1991, 1994) that had the best
13 cumulative exposure estimates found no significant mortality effects of previous air pollution
14 exposure. The Six Cities and ACS studies agree in their findings of strong associations
15 between fine particles and excess mortality. At this time, the long-term studies appear
16 mainly to provide support for the existance of short-term PM-related mortality increases,
17 which are not subsequently offset by decreases below normal rates. However, they do not
18 exclude the possible existence of additional chronic exposure effects; nor do they provide
19 convincing evidence as to the specific pollutant(s) involved; and they do not rule out the
20 possible existence of pollutant thresholds.
21
22 Morbidity Outcomes Associated With PM Exposure
23 Dockery and Pope (1994) reviewed the effects of PM on respiratory mortality and
24 morbidity. The authors considered five primary health endpoints: mortality, hospital usage,
25 asthma attacks, respiratory symptoms and lung funtion. In order to include as many studies
26 as possible, they converted both British smoke and TSP measurements to PM10. Results
27 from each study were converted to an estimated percent change in the health endpoint per
28 10 /xg/m3 PM10. These converted results were then combined across studies of similar
29 endpoints using the standard inverse variance weighted method (fixed effects model). The
30 authors concluded that there was a coherence of effects across the endpoints, with most
31 endpoints showing a one to three percent change per 10 fig/m3 PM10. Pulmonary function
April 1995 ^49 DRAFT-DO NOT QUOTE OR CITE
-------
1 showed a smaller change of 0.15 percent for FEV and 0.08 percent for PEFR. These
2 smaller percent changes are to be expected because there is much less variation in pulmonary
3 function measurements than in the other measures. The limitations of the methodological
4 considerations as they pertain to quantitative assessment of the subject individual studies are
5 discussed in Chapter 12. Dockery and Pope (1994) also noted such limitations in their
6 review.
7 The primary difficulties in combining studies can be summarized as follows. Most
8 studies used several endpoints and it is not clear that results for all of the different endpoints
9 were reported. Most studies used different lag times or moving averages for the pollutants,
10 and in some cases reported only those which gave positive results. For those studies which
11 did report results for similar endpoints, many were analyzed with different statistical models.
12 The short-term studies must take into account serial correlation, and this was done in a
13 variety of ways in those studies which did adjust for it. For these reasons, key findings from
14 most of the studies are only summarized here rather than combined formally.
15
16 Short-Term PM Exposure Hospital Admission Studies
17 Hospitalization data can provide a measure of the morbidity status of a community
18 during a specified time frame. Hospitalization data specific for respiratory illness diagnosis,
19 or more specifically for COPD and pneumonia, index respiratory health status and provide
20 outcome measures which relate to mortality studies for total and specified respiratory
21 measures. Tables 1-7 through 1-10 summarize studies that associate hospitalization data with
22 various measures of PM. Some of the same factors and concerns related to the mortality
23 studies are at issue for these studies also.
24 Both COPD and pneumonia hospitalization studies show moderate, but statistically
25 significant relative risks in the range of 1.06 to 1.25 in association with an increase of 50
26 Mg/m3 m PMIO or its equivalent. There are also indications of a relationship with heart
27 disease, admissions, but the evidence is less clear. Overall, these hospitalization studies are
28 indicative of health outcomes related to PM. They are also supportive of the mortality
29 studies, especially with the more specific diagnosis relationships.
April 1995 1-50 DRAFT-DO NOT QUOTE OR CITE
-------
H
6
o
2
o
H
O
e;
o
H
w
o
Study
Burnett et al. (1994)
All ages in Ontario,
Canada, 1983-1988
Thurstonet al. (1994)
All ages in Ontario,
Canada, July and August,
1986-1988
Thurston et al. (1992)
All ages in Buffalo,
Albany, New York City,
July and August, 1988-1989
Schwartz (in press)
Elderly in New Haven,
1988-1990
Schwartz (in press)
Elderly in Tacoma, 1988-
1990
PM Type &
No. Sites
9 monitoring
stations
measuring
sulfate
3 monitoring
stations
measuring
sulfate, TSP,
and PM10
3 monitoring
stations (one
per city)
measuring
sulfate, H +
PM10
monitoring
stations
averaged, no.
of stations not
given
PM10
monitoring
stations
averaged, no.
of stations not
given
PM Mean Ave. Count
& Range per Day
sulfate means 108
ranged from 3 . 1
to 8.2 jtg/m3
mean sulfate 14.4
ranged 38 to 124
(nmole/m3), PM10
30 to 39 jig/m3,
TSP 62 to 87
/ig/m3
(values not given) Buffalo, 24
Albany, 12,
New York,
137
mean = 41, 8.1
10% tile = 19,
90% tile = 67
mean = 37, 4.2
10% tile = 14,
90% tile = 67
Model Type
&Lag
Structure
Lin. regress.
on filtered
data, 1-d lag
best
Linear
regression on
filtered data,
0-d lag best
Linear
regression on
filtered data
Poisson log-
linear
regression,
19 day mov.
ave. filter,
0-d lag best
Poisson log-
lin. regress.
19 day mov.
ave. filter,
0-d lag best
Other pollutants
measured
Ozone
Ozone, H + , SO2,
NO2
Ozone, H +
Weather &
Other Factors
Temperature
Temperature
Temperature
Result*
Pollutants (Confidence
in model Interval)
none 1 .03
(1.02,
none PM10
1.09
(0.96,
ozone PM10
1.01
(0.87,
1.04)
1.22)
1.15)
ozone (not given for
PM measures)
Ozone (ppb): mean
= 29; 10% tile =
16; 90% tile = 45;
SO2 (ppb): mean =
30; 10% tile = 9;
90% tile = 61
Ozone (ppb): mean
= 25; 10% tile =
13; 90% tile = 36;
SO2 (ppb): mean =
17; 10% tile = 6;
90% tile = 28
Temperature
and dew point
adjusted for in
the moving
average
Temperature
and dew point
adjusted for in
the moving
average
none 1 .06
(1.00,
SO2(2day 1.07
lag) (1.01,
none 1.10
(1.03,
SO2 (2 day 1.11
lag) (1.02,
1.13)
1.14)
1.17)
1.20)
* Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 /xg/m3 increase in TSP.
-------
TABLE 1-8. HOSPITAL ADMISSIONS STUDIES FOR COPD
l-t
vo
vo
>_>
N>
0
!>
Tl
O1
o
2
o
H
0
o
H
W
O
Q
W
Study
Sunyer et al. (1993)
Adults in Barcelona,
1985-1989
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
Schwartz (1994e)
Elderly in Birmingham,
1986-1989
Schwartz (1994d)
Elderly in Detroit
1986-1989
* Relative risk calculated
PM Type &
No. Sites
15 monitoring
stations measuring
black smoke
6 monitoring
stations measuring
PM10
1 to 3 monitoring
stations measuring
PM10
2 to 11 PM10
monitoring stations,
data available for
82% of possible
days
from parameters given
Ave.
PM Mean Count
& Range per Day
winter 33% tile 12
= 49, 67% tile
= 77, summer
33% tile = 36,
67% tile = 55
mean = 36, 10% 2.2
tile = 18, 90%
tile = 58
mean = 45, 2.2
10% tile = 19.
90% tile = 77
mean = 48, 5.8
10% tile = 22,
90% tile = 82
by author assuming a 50 ^ig
Model Type
&Lag
Structure
Autoregressive
linear
regression
analysis, 0-d
lag best
Autoregressive
Poisson model,
1 -d lag best
Autoregressive
Poisson model,
0-d lag best
Poisson auto-
regressive
model using
GEE, 0-d lag
best
;/m increase in PM
Other
pollutants
measured
Sulfur dioxide,
winter 33% tile
= 49/tg/m3,
67%tile = 77,
summer
33%tile = 36,
67%tile = 55
Ozone, mean
= 26 ppb,
10%tile =11,
90% tile = 41
Ozone, mean
= 25 ppb,
10%tile = 14,
90% tile = 37
Ozone, mean
= 21 ppb,
10%tile = 7,
90% tile = 36
10 or 100 £ig/m3
Weather & Result*
Other Pollutants (Confidence
Factors in model Interval)
min temp, none winter: 1.15
dummies for day (1.09,1.21)
of week and summer: 1.05
year (0.98,1.12)
SO2 winter: 1.05
(1.01,1.09)
summer:
1.01
(0.97,1.05)
8 categories of none 1.25
temp. & dew (1.10,1.44)
pt., month,
year, lin. &
quad, time trend
7 categories of none 1.13
temp. &dew (1.04,1.22)
pt., month,
year, lin. &
quad, time trend
Dummy vars. ozone 1.11(1.04,
for temp, 1.17)
month, lin. &
quad, time trend
increase in TSP.
-------
TABLE 1-9. HOSPITAL ADMISSIONS STUDIES FOR PNEUMONIA
U)
Tl
U
o
2!
o
H
O
d
o
H
w
Study
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
Schwartz (1994e)
Elderly in Birmingham,
1986-1989
Schwartz (1994d)
Elderly in Detroit
1986-1989
PM Type & PM Mean
No. Sites & Range
6 monitoring stations mean = 36,
measuring PM10 10% tile =18,
90% tile = 58
1 to 3 monitoring mean = 45,
stations measuring 10% tile = 19.
PM10 90% tile = 77
2 to 11 PM10 mon. mean = 48,
stations, data for 10% tile = 22,
82% of possible days 90% tile = 82
Ave.
Count
per Day
6.0
5.9
15.7
Model Type
& Lag Other pollutants
Structure measured
Autoregressive Ozone: mean 26
Poisson mod., ppb; 10% tile 11;
1-d lag best 90% tile 41
Autoregressive Ozone: mean 25
Poisson mod., ppb; 10% tile 14;
0-d lag best 90% tile 37
Poisson auto- Ozone: mean 21
regress, mod. ppb; 10% tile 7;
using GEE, 90% tile 36
0-d lag best
Weather &
Other Pollutants
Factors in model
8 categories of temp. &none
dew pt., month, year,
lin. & quad, time trend
7 cat. of temp. & dew none
pt., month, year, lin.
& quad, time trend
Dummy variables for ozone
temp, month, lin. &
quad, time trend
Result*
(Confidence
Interval)
1.08
(1.01,1.15)
1.09
(1.03, 1.15)
1.06
(1.02, 1.10)
TABLE 1-10. HOSPITAL ADMISSIONS STUDIES FOR HEART DISEASE
Study
Schwartz and Morris
(in press)
Elderly in Detroit
1986-1989
Ischemic Heart Disease
Burnett et al. (in press)
All ages in Ontario,
Canada, 1983-1988
Cardiac disease admission
PM Type & PM Mean
No. Sites & Range
2 to 11 PM10 mean = 48,
monitoring 10% tile = 22,
stations, data 90% tile = 82
available for
82% of possible
days
22 sulfate station means
monitoring ranged from 3.0 to
stations 7.7 in the summer
and 2.0 and 4.7 in
the winter
Ave.
Count
per Day
44.1
14.4
Model Type
& Lag Other pollutants
Structure measured
Weather &
Other Pollutants
Factors in model
Poisson auto- SO2, mean = 25 Dummy vars. for none
regressive ppb, 10% tile = temp, month, lin.
model using 11, 90% tile = 44 & quad, time trend
GEE, 0-d lag CO, mean 2.4 ozone,
best ppm, 10% tile 1.2, CO, SO2
90% tile = 3.8
Linear Ozone averaged 36 Temperature none
regression on a ppb included in
19 day linear separate analyses
filter, 1-d lag by summer and ozone
best winter
Result*
(Confidence
Interval)
1.06
(1.02, 1.10)
1.06
(1.02, 1.10)
1.04
(1.03, 1.06)
1.04
(1.03, 1.05)
2 * Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 ^g/m3 increase in TSP.
O
3
-------
1 Short and Long-Term Exposure Respiratory Disease Studies
2 Respiratory illness and the factors determining its occurrence and severity are
3 important public health concerns. This effect is of public health importance because of the
4 widespread potential for exposure to PM and because of the very common occurrence of
5 respiratory illness. Of added importance is the fact that recurrent childhood respiratory
6 illness may be a risk factor for later susceptibility to lung damage. The occurrence of lower
7 respiratory morbidity in early childhood may be associated with impaired lung function and
8 growth that appears to persist through adolescence, and certain physicians assert that
9 infections, reactive airways, and inhaled pollutants (mostly cigarette smoke) are the most
10 important risk factors in the development of chronic lung disease. Thus, factors such as the
11 presence of PM (which increases the risk for respiratory symptoms and related respiratory
12 morbidity) are important because of associated public health concern with regard to both the
13 immediate symptoms produced and the longer term potential for increases in the development
14 of chronic lung disease.
15
16 Acute Respiratory Disease Studies
17 Acute respiratory disease studies include several different endpoints, but the majority of
18 authors reported results on at least two of: (1) upper respiratory illness; (2) lower
19 respiratory illness; or (3) cough (See Table 1-11). These relative risks are all estimated for
20 an increase of 50 /ig/m3 in PM10 (24-h) or its equivalent. The results for upper respiratory
21 illness are very inconsistent: two studies estimate a relative risk near 1.00 whereas four
22 others obtain estimates between 1.14 and 1.55. The relative risks for lower respiratory
23 illness are spread between 1.01 and 2.03, but all are positive. The relative risks for cough
24 include two below 1.0 and go as high as 1.51. All of these are generally suggestive of a PM
25 effect, but whereas the hospital admission studies were all done in a similar manner and
26 resulted in very similar results, these studies used different designs and yield very
27 inconsistent results.
28
29 Chronic Respiratory Disease Studies
30 The three studies listed (Table 1-12) are based on a similar type of questionnaire but
31 were done by two different groups of researchers. All three studies suggest a chronic effect
April 1995 1-54 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-11. ACUTE RESPIRATORY DISEASE STUDIES
H
6
o
z
o
H
O
G
O
H
W
Study
Schwartz et al. (1994) 300
elementary school children
in Six-Cities in U.S.,
1984-1988
Popeet al. (1991),
students in the Utah
Valley, winter 1989-1990
Popeet al. (1991),
asthmatic children in the
Utah Valley, winter 1989-
1990
Pope and Dockery (1992),
symptomatic children in
the Utah Valley, winter
1990-1991
PM Type &
No. Sites
PM10
monitoring in
each city
PM10
monitoring
stations at 3
sites
PM10
monitoring
stations at 3
sites
PM10
monitoring
stations at 2
sites
Ave.
PM Mean Rate
& Range per Day
median 30 /ig/m3 3.1
10th percentile =
13, 90th
percentile = 53
mean = 46 (not given)
Hg/m3,
range = 11 to
195
mean = 46 (not given)
^g/m3,
range = 11 to
195
mean = 76 (not given)
fig/m3,
range = 7 to 251
Model Type
& Lag Structure
Autoregressive
logistic
regression using
GEE
Fixed effects
logistic
regression
Fixed effects
logistic
regression
Autoregressive
logistic
regression using
GEE
Other
pollutants
measured
Ozone, NO2,
S02
Limited
monitoring of
NO2, SO2, and
ozone. Values
were well
below the
standard
Limited
monitoring of
NO2, SO2, and
ozone. Values
were well
below the
standard
none
Weather & Other
Other pollutants
Factors in model
Temperature none
S02
ozone
Variables for none
temperature and
time trend
Variables for none
low
temperature and
time trend
Variable for none
low
temperature
Result*
(Confidence
Interval)
1.51 (1.12,
1.39(0.98,
1.49(1.10,
Upper resp.
1.20(1.03,
Lower resp.
1.28(1.06,
Upper resp.
0.99(0.81,
Lower resp,
1.01 (0.81,
Upper resp.
1.20(1.03,
Lower resp
1.27(1.08,
Cough
1.29(1.12,
2.05)
1.96)
2.01)
1.39)
1.56)
1.22)
1.27)
1.39)
1.49)
1.48)
n
HH
m
-------
TABLE 1-11 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
2:
VO
S
H-
Ov
O
£
H
6
o
z
o
H
0
H
W
O
Study
Pope and Dockery (1992),
asymptomatic children in
the Utah Valley, winter
1990-1991
Hoek and Brunekreef
(1993), respiratory disease
in school children aged 7
to 12 in Wageningen,
Netherlands, winter
1990-1991
Schwartz et al. (1991)
Study of acute respiratory
illness in children in
5 German communities,
1983-1985
PM Type &
No. Sites
PM10
monitoring
stations at
2 sites
Two to 4
monitoring
stations
measured
PM10
Two to 4
monitoring
stations in
each area
measured
TSP
Ave.
PM Mean Rate
& Range per Day
mean = (not given)
76 /ig/m3,
range = 7 to
251
max = (not given)
1 10 /ig/m3
medians ranged 0.5 to 2.9
from 17 to
56 /ig/m3,
10% tiles from
5 to 34, 90%
tiles from 41 to
118
Model Type
&Lag
Structure
Autoregressive
logistic regression
using GEE
Autoregressive
logistic regression
using GEE
Autoregressive
Poisson regression
using GEE
Other
pollutants
measured
none
Max SO2 = 105
ftg/m3, max NO2
= 127 /tg/m3
median SO2
levels ranged
from 9 to
48 /tg/m3,
median NO2
levels ranged
from 14 to
5 /xg/m3
Weather & Other
Other pollutants
Factors in model
Variable for none
low temperature
Variable for none
ambient
temperature and
day of study
Most none (TSP
significant was not
terms of day significant
of week, time when NO2
trend, and added to model)
weather
(terms not
listed)
Result*
(Confidence
Interval)
Upper resp.
0.99
(0.78, 1.26)
Lower resp.
1.13
(0.91, 1.39)
Cough
1.18
(1.00, 1.40)
Upper resp.
1.14
(1.00, 1.29)
Lower resp.
1.06
(0.86, 1.32)
Cough
0.98
(0.86, 1.11)
1.26
(1.12, 1.42)
n
HH
H
W
-------
TABLE 1-11 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
d:
h-t
u?
K-t
1
O
>
2
O1
0
^
o
o
H
W
O
n
H
Study
Schwartz et al. (1994)
Study of respiratory
symptoms in 6 U.S.
cities, 1984-1988
Braun-Fahrlander et al.
(1992)
Study of preschool
children in four areas
of Switzerland
Roemer et al. (1993)
Study of children with
chronic respiratory
symptoms in
Wageningen, The
Netherlands
Dusseldorf et al.
(1994)
Study of adults near a
steel mill in The
Netherlands
PM Type &
No. Sites
Daily
monitoring
of PM10,
PM2 5 at
each city
Daily
monitoring of
TSP
Daily
monitoring of
PM10
Daily
monitoring of
PM[0, iron,
sodium,
silicon, and
manganese
Ave.
PM Mean Rate
& Range per Day
median PM10 (not given)
= 30 iKg/m3,
10% tile = 13,
90% tile = 53
median PM, 5
= 18 /ig/m ,
10% tile = 7,
90% tile = 37
(not given) 4.4
6 days above .094
110 ng/m3 incidence
rate
mean PM10 (not given)
= 54 pig/m3,
range =
4 to 137)
Model Type
&Lag
Structure
Autoregressive
logistic
regression
using GEE
Logistic
regression
Autoregressive
logistic
regression
Logistic
regression
Other Weather &
pollutants Other
measured Factors
SO2, median = 4 temperature,
ppb, 10% tile = day of week,
1, 90% tile = 18 city or
NO2, median = residence
13 ppb, 10% tile
5, 90% tile = 24,
ozone
SO2, NO2, and city, risk
ozone levels not strata, season,
given temperature
SO2 and NO2 (not given)
means not given
Geometric mean (not given)
iron = 501
ng/m3, manganese
= 17 ng/m3,
silicon =
208 ng/m3
Other Result*
pollutants (Confidence
in model Interval)
all two Cough
pollutant (PMIO lag 1)
models were 1.51
fitted with (1.12,2.05)
minimal effect Upper resp.
on PM (PMIO lag 2)
1.39
(0.97, 2.01)
Lower resp.
(PM10 lag 1)
2.03
(1.36,3.04)
none Upper resp.
1.55
(1.10,2.24)
none Cough
(not given,
probably less than
one)
none Cough
1.14
(0.98, 1.33)
-------
TABLE 1-11 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
Ui
oo
o
O
2J
O
H
O
d
o
H
W
Study
PM Type &
No. Sites
PMMean
& Range
Ave.
Rate
per Day
Model Type
&Lag
Structure
Other
pollutants
measured
Weather &
Other
Factors
Other
pollutants
in model
Result*
(Confidence
Interval)
Ostro et al. (1991)
Study of adult asthmatics
in Denver, Colorado
November 1987 to
February 1988
Ostro et al. (1993)
Study of non-smoking
adults in Southern
California
Two monitors
provided daily
measurements of
PM2.5
22 /ug/m3, range 15 (out of 108) Autoregressive nitric acid,
= 0.5 to 73 logistic sulfates,
regression nitrates, SO2,
andH+
day of survey, none
day of week, gas
stove, minimum
temperature
Apparently one mean sulfate = 84.2/person for
site (Azusa). /*g/m3, range = lower,
PM measurements 2 to 37 mean 10.2/person,
included sulfate COHS = 12 per upper
fraction and 100 ft, range
COHS = 4 to 26
Logistic ozone, mean temperature, rain none
regression = 7 pphm, humidity
range = 1 to 28
Cough
1.09
(0.57,2.10)
Sulfates:
Upper resp.
0.91
(0.73, 1.15)
Lower resp.
1.48
(1.14, 1.91)
H- * Relative risk calculated from parameters given by author assuming a 50 /zg/m3 increase in PM10 on 100 /*g/m3 increase in TSP.
n
HH
H
W
-------
TABLE 1-12. CHRONIC RESPIRATORY DISEASE STUDIES
eL
Co
Ul
1
Ui
VO
O
g
^
1
O
O
z
o
H
O
C!
0
H
W
O
Study
Ware et al. (1984)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1974-1977
Dockery et al. (1989)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1980-1981
Chapman et al. (1985)
Study of persistent cough
and phlegm (bronchitis) in
adults in four
communities in Utah.
Survey done in 1976
Neas et al. (1994)
Study of children aged
7 to 1 1 from six cites in
U.S. Survey done
1983-1986.
* Estimates calculated from
PM Type &
No. Sites
Daily monitoring
of TSP, SO2,
NO2, and ozone at
each city
Daily monitoring
of PM15, sulfate
fraction at each
city
Daily monitoring
of TSP, and
sulfate fraction at
each city
PM2.5
PM Mean
& Range
City TSP means
ranged from 39
to 1 14 /tg/m3
City PM15
means ranged
from 20 to 59
/*g/m3
Previous 5 year
TSP ranged from
11 to 115 /xg/m3
Not given
Overall
Symptom
Rate
Cough, .08,
Bronchitis
.08,
Lower resp.
.19
Cough, .02 to
.09, Bronchitis
.04 to .10,
Lower resp.
.07 to .16
.02 to .05 by
city
Not given
data tables assuming a 50 /*g/m3 increase in PM]0 on
Model
Type
&Lag
Structure
Logistic
regression
Logistic
regression
Logistic
regression
Logistic
regression
100 /ig/m3
Other
Other Other pollutants
pollutants Covariates in model
measured
SO2, NO2, age, gender, none
and ozone parental
education,
maternal
smoking
S02, NO2, age, gender, none
and ozone maternal
smoking
SO2, NO2 smoking none
NO2 household none
smoking, gas
stove, age,
gender none
none
increase in TSP.
Result*^
(Confidence
Interval)
Cough
2.75 (1.92,
Bronchitis
2.80(1.17,
Lower resp.
2.14(1.06,
Cough
5.39(1.00,
Bronchitis
3.26(1.13,
Lower resp.
2.93 (0.75,
Mothers
1.75(1.21,
Fathers
1.94(1.16,
Cough
1.08(0.76,
Bronchitis
1.32(0.98,
Lower resp
1.23 (0.98,
3.94)
7.03)
4.31)
28.6)
10.28)
11.60)
2.54)
3.25)
1.53)
1.79)
1.55)
O
a
-------
1 of paniculate matter on respiratory disease, but the studies suffer from the usual difficulty of
2 cross sectional studies. The PM effect estimates are based on variations in exposure which
3 are determined by different numbers of locations. In the first two studies there were six
4 locations and, in the second, four. The results seen were consistent with a PM gradient, but
5 it is impossible to separate out the effect of PM any other factors or pollutants which have
6 the same gradient.
7
8 Short and Long-Term Exposure Pulmonary Function Studies
9 Pulmonary function studies are part of a comprehensive investigation of the possible
10 effects of any air pollutant. Measurements can be made in the field, they are noninvasive,
11 and their reproductibility has been well documented; and guidelines for reference values and
12 interpretative strategies of lung function tests have been prepared. Various factors are
13 important determinants of lung functions. Lung function in children has been related to
14 genetic factors that exert their greatest influence through general stature as measured by
15 height and age. Growth patterns in children differ by gender and lung function declines with
16 age among adults. Studies of the growth of pulmonary function and generalized growth
17 models consider factors of how growth is statistically dependent on initial measures of
18 function, and how it is related to respiratory illness in childhood. The effects of active
19 smoking and passive smoking are also considered. Epidemiological studies relating ambient
20 PM measures to decrements in pulmonary function represent a potentially important health
21 effect.
22 The acute pulmonary function studies (Table 1-13) are suggestive of a short term effect
23 resulting from paniculate pollution. Peak flow rates show decreases in the range of 30 to 40
24 ml/sec to be associated with an increase of 50 ^g/m3 in PM10 (24-h) or its equivalent. The
25 results appear to be larger in symptomatic groups such as asthmatics. The effects are seen
26 across a variety of study designs, authors, and analysis methodologies. Effects using FEVj
27 or FVC as endpoints are less consistent.
28 The chronic pulmonary function studies are less numerous than the acute studies
29 (Table 1-13). The one study with good monitoring showed no effect from paniculate
30 pollution. Cross sectional studies require very large sample sizes to detect differences
31 because the studies cannot eliminate person to person variation which is much larger than the
April 1995 1-60 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-13. ACUTE AND CHRONIC PULMONARY FUNCTION CHANGES
3:
vJD
3
a
%^
o
z
g
r— 5
O
|
w
o
Study Period,
Population
Dockery et al. (1982)-
School age children in
Steubenville, OH,
measured at three times
between 1978 and 1980
Dassen et al. (1986)
School age children in
The Netherlands,
measured in November,
1984 and January, 1985
Quackenboss et al.
(1991) - Asthmatic
children aged 6 to 15
years in Tuscon, AZ,
measured in May and
November, 1988
Pope et al. (1991) -
Study of asthmatic
children in the Utah
Valley
Pope and Dockery
(1992) - Study of non-
asthmatic symptomatic
and asymptomatic
children in the Utah
Valley
PM Type & PM Mean
No. Sites & Range
Single station up to 455 /Kg/m3
measuring TSP
Six station network TSP and RSP both
measuring TSP, exceeded 200
RSP (PM10) /ig/m3
Individual
monitoring of
homes of PM2 5,
PM10
PMIO monitors in PM10 ranged from
Orem and Lindon, 11 to 195 /ig/m3
Utah
PM10 monitors in PM10 ranged from
Orem and Lindon, 11 to 195 pig/m3
Utah
Model Type Other
& Lag pollutants
Structure measured
individual SO2
regression analyses
for each child,
coefficients pooled
across time
multiple linear SO2
regression
Random effects NO2
linear model
Weighted least SO2, NO2, ozone
squares regression
Weighted least SO2, NO2, ozone
squares regression
Weather &
Other Pollutants
Factors in model
average TSP
temperature
technician, RSP
appliance,
presence of
colds
temperature, PM2 5
wind speed,
dew point
low PM10
temperature
low PM10
temperature
Decrease*
(Confidence
Interval)
FVC: 8.1 ml
FEV075: 1.8ml
Note: decreases
were statistically
significant
slopes not given but
FVC, FEVj, and
PEFR were
significantly reduced
during episodes
PEFR: 375 ml/s
Note: these are
diurnal rather than
daily changes
PEFR: 55 ml/s
(24, 86)
Symptomatic
PEFR
30 ml/s
(10, 50)
Asympto-
matic PEFR
21 ml/s
(4, 38)
O
H
W
-------
TABLE 1-13 (cont'd). ACUTE AND CHRONIC PULMONARY FUNCTION CHANGES
i
H- '
OS
O
!>
3
a'
0
2
O
H
G
O
H
W
O
^-\
Study Period,
Population
Koenig et al. (1993) -
Study of asthmatic and
non-asthmatic elementary
school children in Seattle,
WA in 1989 and 1990
Hoek and Brunekreef
(1993) - Study of children
aged 7 to 12 in
Wageningen, Netherlands
Roemer et al. (1993) -
Study of children with
chronic respiratory
symptoms in The
Netherlands
Pope and Kanner (1993) -
Study of adults in the Utah
Valley from 1987 to 1989
Neas et al. (1994) -
Study of lung function in
children in 6 cities in the
U.S. Data collected from
1983-1988.
PM Type &
No. Sites
PM2 5 calibrated
from light
scattering
Single site
measure black
smoke. PM10 was
measured during
episodes
Single site
measure black
smoke. PM10 was
measured using an
Anderson dichot
PM10 was collected
daily from the north
Salt Lake site
Daily monitoring of
PM2 5, sulfate
fraction at each city
PM Mean
& Range
PM2 5 ranged
from 5 to
45 /tg/rn
range of PM10
was 30 to
144 ftg/m3
range of PM10
was 30 to
144 /ig/rn3
PM10 daily
mean =
55 /ig/m ,
ranged from 1
to 181 /ig/m3
not given
Model Type Other
& Lag pollutants
Structure measured
Random effects none
linear regression
SAS procedure SO2, NO2
AUTOREG
multiple linear SO2, NO2
regression
analysis
Linear regression Limited
on difference in monitoring of
PFT as a function SO2, NO2, and
of PM10 ozone
Linear regression SO2, NO2, and
using logarithm of ozone
PFT value
Weather &
Other Pollutants
Factors in model
height, PM25
temperature
day of study PM10
none PM10
low PM2 5
temperature
city, gender PM2 5
parental
education,
history of
asthma, age,
height, weight
Decrease*
(Confidence
Interval)
Asthmatics
FEVj 42 ml
(12, 73)
FVC 45 ml
(20, 70)
Non-asthmatics
FEVj 4 ml
(-7, 15)
FVC -8 ml
(-20, 3)
PEFR
41 ml/s (-8, 90)
PEFR
34 ml/s (9, 59)
FEV!
29 ml (7, 51)
FVC 15 ml (-
15, 45)
FVC and FEVi not
changed. Values
could not be
converted to mis.
Decreases in lung function calculated from parameters given by author assuming a 50
increase in PM10 or 100 /ig/m3 increase in TSP.
-------
1 within person variation. Thus the lack of statistical significance cannot be taken as proof of
2 no effect.
3
4 Comparison of Effects of PM10 Versus PM2>5 on Respiratory Disease and Pulmonary
5 Function
6 The most direct comparison of effects of PM10 versus PM2 5 are possible when studies
7 include both exposure measures in their analyses. This occurred in the Six City study for
8 Steubenville children, the Tuscon study, and the Uniontown study for pulmonary function
9 changes (Table 1-13). None of these studies directly showed that one of these PM measures
10 was a significantly better predictor than the other for pulmonary function effects. The study
11 did suggest that PM2 5 may be better predictor of lung function change.
12 A few other studies used PM2 5 as a measure of paniculate exposure. One on lung
13 function in asthmatic and non-asthmatic school children in Seattle (Table 1-13) found a
14 slightly larger effect of PM2 5 for asthmatics, but a slightly smaller effect for non-asthmatics
15 when compared with the PM10 studies. Also, the Ostro et al. (1991) study of respiratory
16 disease in Denver found an effect that was in the middle of the range of effects found by the
17 PM10 studies.
18 Based on the above information, there is currently no obvious way by which to clearly
19 distinguish morbidity effects of PM10 versus PM2 5. Even the suggestive evidence leaves the
20 scales in a balanced position.
21
22 Mortality Effects of Acid Aerosols
23 Few epidmiological studies have examined mortality data for an association with
24 ambient paniculate strong acid aerosol (H+) exposures. The scarcity of the analyses is due
25 to the absence of adequate ambient acid measurement techniques in the past, and to the lack
26 of routine acid aerosol monitoring in more recent years. Some studies now exist which
27 suggest that human health effects may be associated with expsoures to ambient acid aerosols,
28 both: (1) as derived from reexamination of older, historically important data on air pollution
29 episode events in North America and Europe and (2) as can be deduced from limited recent
30 epidemiology studies carried out in the U.S., Canada, and Europe.
April 1995 1_63 DRAFT-DO NOT QUOTE OR CITE
-------
1 Historical and present-day evidence suggests that strongly acidic PM can be associated
2 with both acute and chronic human health effects. Evidence from historical pollution for
3 episodes, notably the London Fog episodes of the 1950's and early 1960's, indicates that
4 extremely elevated daily acid aerosol concentrations (on the order of 400 /xg/m3 as H2SO4, or
5 roughly 8,000 nmoles/m3 H+) may be associated with excess acute human mortality when
6 present as a copollutant with elevated concentrations of PM and SO2. In addition, Thurston
7 et al. (1989) and Ito et al. (1993) both found significant associations between acid aerosols
8 and mortality in London during non-episode pollution levels (< 30 /xg/m3 as H2SO4, or
9 < approximately 600 nmoles/m3 H+), although these associations could not be separated
10 from those for BS or SO2. The only attempts to date to associate present-day levels of acidic
11 aerosols with acute and chronic mortality (Dockery et al., 1992; Dockery et al., 1993b,
12 respectively) were unable to do so, but weaknesses in these analyses (in particular, too
13 limited H+ data for the analysis) may have made associations undetectable. At very high
14 concentrations that do not occur in the ambient air, mortality in laboratory animals can occur
15 following acute exposure, due primarily to laryngeal or bronchoconstriction; larger particles
16 are more effective in this regard than are smaller ones.
17
18 Respiratory Illness Effects of Acid Aerosols
19 Historical and present-day evidence suggests that there can be both acute and chronic
20 effects of strongly acidic PM on human health. Increased hospital admissions for respiratory
21 causes were documented during the London Fog episode of 1952, and this association has
22 now been observed under present-day conditions, as well. Thurston et al. (1992) and
23 Thurston et al. (1994) have noted associations between ambient acidic aerosols and
24 summertime respiratory hospital admissions in both New York State and Toronto, Canada,
25 respectively, even after controlling for potentially confounding temperature effects. In the
26 latter of these studies, significant independent H+ effects remained even after simultaneously
27 considering the other major copollutant, O3, in the regression model. In these studies, H+
28 effects were estimated to be the largest during acid aerosol episodes (H+ > 10 )ug/m3 as
29 H2SO4, or =200 nmoles/m3 H+), which occur roughly 2 to 3 times per year in eastern
30 North America. These studies provide evidence that present-day strongly acidic aerosols
April 1995 1-64 DRAFT-DO NOT QUOTE OR CITE
-------
1 may represent a portion of PM which is particularly associated with significant acute
2 respiratory disease health effects in the general public.
3 Results from recent acute symptoms studies of healthy children indicate the potential for
4 acute acidic PM effects in this population. While the 6-City study of diaries kept by parents
5 of children's respiratory and other illness did not demonstrate H+ associations with lower
6 respiratory symptoms except at H+ above 110 nmoles/m3 (Dockery et al., 1994), upper
7 respiratory symptoms in two of the cities were found to be most strongly associated with
8 daily measurements of H2SO4 (Schwartz, et al. 1991).
9 Studies of the effects of chronic H+ exposures on children's respiratory health and lung
10 function are generally consistent with effects as a result of chronic H+ exposure.
11 Preliminary analyses of bronchitis prevalence rates as reported across the 6-City study locales
12 were found to be more closely associated with average H+ concentrations than with PM in
13 general (Speizer, 1989). A follow-up analysis of these cities and a seventh locality which
14 controlled the analysis for maternal smoking and education and for race, suggested
15 associations between summertime average H+ and chronic bronchitic and related symptoms
16 (Damokosh et al., 1993). The relative odds of bronchitic symptoms with the highest acid
17 concentration (58 nmoles/m3 H+) versus the lowest concentration (16 nmoles/m3) was 2.4
18 (95% CI: 1.9 to 3.2). Furthermore, in a follow-up study of children in 24 U.S. and
19 Canadian communities (Dockery et al., 1993a) in which the analysis was adjusted for the
20 effects of gender, age, parental asthma, parental education, and parental allergies, bronchitic
21 symptoms were confirmed to be significantly associated with strongly acidic PM (relative
22 odds = 1.7, 95% CI: 1.1 to 2.4). It was also found in the 24-Cities study that mean FVC
23 and FEV1-0 were lower in locales having high particle strong acidity (Raizenne et al., 1993).
24 Thus, chronic exposures to strongly acidic PM may have effects on measures of respiratory
25 health in children.
26 The respiratory tract has an array of defense mechanisms to kill, detoxify, and
27 physically remove inhaled material, and these defenses may be altered by exposure to H2SO4
28 at levels < 1,000 /xg/m3. Acid aerosols alter mucociliary clearance in human and laboratory
29 animals, with effects dependent on exposure concentration and the region of the lung being
30 studied. For example 1- to 2-h resting exposures of humans to 100 ^ig/m3 accelerate
31 clearance in large bronchi, but slows clearance in smaller more peripheral airways.
April 1995 1-65 DRAFT-DO NOT QUOTE OR CITE
-------
1 Clearance in asthmatics is also affected, but the results are not clearly interpretable. Long-
2 term exposure also affects mucociliary clearance in animals. For example, in rabbits
3 exposed intermittently for 125 /ug/m3 H2SO4 for 1 year, clearance was accelerated during
4 exposure but was depressed 6 months after exposure ceased. These responses are complex
5 and are accompanied by histological and chemical changes in mucus and epithelial secretory
6 cells. Defenses, such as resistance to bacterial infection, may be altered by acute exposure
7 to concentrations of H2SO4 around 1,000 jug/m3.
8 Severe morphologic alterations in the respiratory tracts of animals occur at high acid
9 levels. At low levels and with chronic exposure, the main response seems to be hypertrophy
10 and/or hyperplasia of mucus secretory cells in the epithelium; these alterations may extend to
11 the small bronchi and bronchioles, where secretory cells are normally rare or absent.
12 Limited data also suggest that exposure to acid aerosols may affect the phagocytic
13 functioning of alveolar macrophages; the lowest level examined to date is 500 /ig/m3 H2SO4.
14 Alveolar region particle clearance is accelerated by repeated H2SO4 exposures to as low as
15 250 jug/m3; higher levels retard clearance. Acute exposure of rabbits to lower concentrations
16 (e.g., 75 /ig/m3 H2SO4) can affect other alveolar macrophage functions.
17
18 Pulmonary Function Effects of Acid Aerosols
19 Both acute and chronic exposure of laboratory animals to H2SO4 at levels well below
20 lethal ones will produce functional changes in the respiratory tract. The pathological
21 significance of some of these are greater than for others. Acute exposure will alter
22 pulmonary function, largely due to bronchoconstrictive action. However, attempts to
23 produce changes in airway resistance in healthy animals at levels below 1 mg/m3 have been
24 largely unsuccessful, except when the guinea pig has been used. The lowest effective level
25 of H2SO4 producing bronchoconstriction to date in the guinea pig is 100 /ig/m3 (1-h
26 exposure). In general, smaller size droplets are more effective in altering pulmonary
27 function, especially at low concentrations. Yet even in the guinea pig, there are
28 inconsistencies in the type of response exhibited towards acid aerosols. Chronic exposure to
29 H2SO4 is also associated with alterations in pulmonary function (e.g., changes in the
30 distribution of ventilation and in respiratory rate in monkeys). But, in these cases, effective
31 concentrations are >500 jug/m3. Hyperresponsive airways have been induced with repeated
April 1995 1-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposures to 250 ptg/m3 H2SO4 in rabbits, and have been suggested to occur following single
2 exposures at 75 /xg/m3.
3 Ten human clinical studies since 1988 have confirmed previous findings that healthy
4 subjects do not experience decrements in lung function following single exposures to H2SO4
5 at levels up to 2,000 ^tg/m3 for 1 h, even with exercise and use of acidic gargles to minimize
6 neutralization by oral ammonia. Mild lower respiratory symptoms occur at exposure
7 concentrations in the mg/m3 range, particularly with larger particle sizes.
8 There is no clearly established exposure-response relationship across studies. Asthmatic
9 subjects appear to be more sensitive than healthy subjects to the effects of acid aerosols on
10 lung function, but reported effective concentrations differ widely among studies. Adolescent
11 asthmatics may be more sensitive than adult asthmatics, and may experience small
12 decrements in lung function in response to H2SO4 at exposure levels only slightly above peak
13 ambient levels (e.g., less than 100 /xg/m3). Although the reasons for the inconsistency
14 among studies remain largely unclear, individual variability in sensitivity and subject
15 selection may be an important factors. Even in studies reporting an overall absence of
16 effects on lung function, occasional asthmatic subjects appear to demonstrate clinically
17 important effects. Two studies from different laboratories have suggested that responsiveness
18 to acid aerosols may correlate with the degree of baseline airway hyperresponsiveness.
19 However, based on very limited studies, the elderly and individuals with chronic obstructive
20 pulmonary disease do not appear to be particularly more susceptible to the effects of acid
21 aerosols on lung function than healthy adults.
22 Two recent studies have examined the effects of exposure to both H2SO4 and ozone on
23 lung function in healthy and asthmatic subjects. Both studies found evidence that 100 /xg/m3
24 H2SO4 may potentiate the response to ozone, in contrast with previous studies. Recent
25 summer camp (and schoolchildren) studies of lung function have also indicated a significant
26 association between acute exposures to acidic PM and decreases in the lung function of
27 children independent of those associated with O3 (Studnicka et al., 1995; Neas et al, 1995).
28 In view of uncertainties about differences between high acid concentrations needed to
29 produce effects in animal studies and low concentrations found in the human environment,
30 the epidemiologic evidence does not establish a clear role for acid aerosols as a primary
31 agent contributing to ambient PM exposure effects on pulmonary function.
April 1995 1_67 DRAFT-DO NOT QUOTE OR CITE
-------
1 Coherence of Epidemiologic Findings
2 Factors involved in evaluating both the data and the entire group of epidemiological
3 studies, include the strength of association, the consistence of the association, as evidenced
4 by its repeated observation by different persons, in different places, circumstances and time,
5 and the consistency with other known facts (Bates, 1992). One can look for
6 interrelationships between different health indices to provide a stronger and more consistent
7 synthesis of available information. The various findings that support a picture of coherence
8 would provide a stronger case with quantitative studies as opposed to qualitative studies.
9 Other studies may be inappropriate to use in such a discussion, and the quality of the study
10 should be considered. Bates (1992) states that the difficulty with discussing any index of
11 internal coherence is that this requires a series of judgements on the reliability of the
12 individual findings and observations. The outcome of a coherence discussion then is a
13 qualitative presentation in the end, not quantitative. Thus, coherence cannot be formally
14 measured.
15 Bates (1992) also noted that the strength of different health indexes are important as are
16 difficulties in assessing exposure. Bates (1992) also suggests three areas to look for
17 coherence: (1) within epidemiological data, (2) between epidemiological and animal
18 toxicological data, and (3) between epidemiological, controlled human and animal data.
19 Coherence by its nature considers biological relationships of exposure to health
20 outcome. The biologic mechanism underlying an acute pulmonary function test reduction in
21 children is most likely not part of the acute basis for a change in the mortality rate of a
22 population exposed in an older group of individuals. In looking for coherence one should
23 compare outcomes that look at similar time frames—daily hospitalizations compared to daily
24 mortality rather than monthly hospitalizations. Overall the data indicates that PM has a
25 relationship with a continuum of health outcomes, but the studies may not establish a
26 coherence between them. The underlying mechanisms may be different.
27 The principal health outcome for which coherence is desirable is mortality, the death
28 rate in a population. This can be considered within the endpoint and/or in other endpoints.
29 Of the various morbidity outcomes studied and discussed in the earlier part of the chapter,
30 hospitalization studies reviewed in the chapter support this notion. The mortality studies
31 suggest that these specific causes provide stronger relationships (i.e., larger RR estimates)
April 1995 1_68 DRAFT-DO NOT QUOTE OR CITE
-------
1 than total mortality. The outcome potentially most related is hospital admission for
2 respiratory or cardiovascular causes in the older age group (i.e., > 65 years old). In a
3 qualitative sense, the increased mortality found in that age group should also be paralled by
4 increased hospital admissions.
5 Partial coherence is established by those studies in which increased incidence of
6 different health outcomes associated with PM are found in the same population, as is the case
7 for the following examples, based on currently published studies:
8 •Detroit: Mortality mainly in elderly populations, hospital admissions for respiratory
9 causes and for cardiovascular causes in the elderly;
10 •Birmingham: Mortality mainly in the elderly, hospital admissions for the elderly;
11 •Philadelphia: Mortality and hospital admissions for pneumonia in the elderly;
12 •Utah Valley: Mortality and hospital admissions for respiratory causes in adults.
13
14 Also, pulmonary function, respiratory symptoms, and medication use in asthmatic subjects of
15 all ages; hospital admissions for respiratory symptoms, pulmonary function, respiratory
16 symptoms, and medication use in healthy school children, pulmonary function in
17 symptomatic and asymptomatic children; and elementary school absences in children were
18 found to be associatied with PM exposures in Utah Valley. A similar study found a PM
19 effect on pulmonary function in smokers with COPD in Salt Lake Valley. The Utah Valley
20 population was largely non-smoking, so smoking was not likely to be a source of
21 confounding.
22 While these multiple outcomes did not occur in strictly identical subgroups of each
23 population, there was probably a sufficient degree of overlap to indicate that PM was a
24 significant predictor of a wide range of health outcomes within a specific community. The
25 symptoms serious enough to warrant hospitalization and the major part of the excess
26 mortality occurred in the elderly sub-group of the population. However, a significant
27 decrement in pulmonary function and increased incidence of symptoms associated with daily
28 increases in PM occurred in children in Utah Valley, along with a "quality of life" effect
29 measured by lost school days. Thus, there is evidence for increased risk of health effects
30 related to PM exposure ranging in seriousness from asymptomatic pulmonary function
31 decrements, to respiratory symptoms and cardiopulmonary symptoms sufficiently serious to
April 1995 1-69 DRAFT-DO NOT QUOTE OR CITE
-------
1 warrant hospitalization, and to excess mortality from respiratory and cardiovascular causes,
2 especially in those older than 65 years of age.
3 Children may also be at increased risk of pulmonary function changes and increased
4 incidence of symptoms associated with PM exposure. While we have arrayed these health
5 outcomes in order of increasing severity, there is as yet little indication that there is a
6 progression of effects in any single individual associated with increasing exposure to PM.
7 The "exposure-response" relationship that is derived in most studies must be understood as
8 characterizing population risk from population exposure. Additional studies are needed to
9 define the relationship(s) among individual exposure to PM and other stress factors,
10 individual risk, and individual progression among disease states. Differences in PM
11 dosimetry in the developing, aged, or diseased respiratory tract may also contribute to
12 increased susceptability.
13 The coherence of the various health effects in humans could be established more
14 conclusively from epidemiology studies if there were better evidence. We cannot prove that
15 the people that suffered respiratory symptoms in response to PM exposure were among the
16 same people who suffered pulmonary function decrements from PM exposure in the past,
17 that those who were admitted to hospital for respiratory or cardiopulmonary causes in
18 response to PM exposure were among those who had suffered respiratory symptoms or
19 pulmonary function decrements from earlier PM exposures, nor that those who died from
20 PM exposure were among those who had earlier shown other health endpoints associated
21 with PM exposures. Such information could, in principle, be extracted from longitudinal
22 data bases such as those collected by health care providers; however, although some such
23 efforts are now being considered, the preferred design for such a study is a prospective
24 design rather than a retrospective design. If and when these studies are completed, they
25 could be useful in future PM health assessments.
26
27
28 1.13 BIOLOGICAL PLAUSIBILITY: POTENTIAL MECHANISMS OF
29 ACTION
30 Chapter 13, the Integrative Health Synthesis Chapter, incorporates key information of
31 the types summarized in the several preceeding sections of this chapter. It also importantly
April 1995 1-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 discusses key points relating to evaluation of the biological plausibility of the new
2 epidemiologic findings, the identification of special risk groups, and the interpretation of
3 implications of reported relative risk estimates for associations between ambient PM exposure
4 indices and mortality/morbidity effects. Each of these latter three topics are accorded
5 separate sections in this Executive Summary, starting with this one on biological plausibility.
6 Epidemiologic studies have suggested that ambient paniculate exposure may be
7 associated with increased mortality and morbidity at PM concentrations below those
8 previously thought to affect human health (Section 1.12 and Chapter 12). However, the
9 biological plausibility of a causal relationship between low concentrations of PM and daily
10 mortality and morbidity rates is neither intuitively obvious nor expected based on
11 experimental studies of the toxicity of inhaled particles. As indicated in Chapter 11, chronic
12 toxicity from poorly soluble particles has been observed based on the slow accumulation of
13 large lung burdens of particles, not due to small daily fluctuations of one or another of the
14 specific PM constituents discussed in that chapter. Two possible exceptions can be noted.
15 Acute toxicity from inhaled particles has been demonstrated with acidic particles, but only at
16 much higher particle concentrations than those observed in the recent epidemiology studies
17 reporting an association between low-level PM concentrations and morbidity/mortality.
18 Acute toxicity resulting in death has also been reported in rats inhaling singlet ultrafine
19 particles (<0.05 jum) formed in the pyrolysis of perfluorinated compounds at concentrations
20 of 60 to 200 /ig/m3' (Oberdorster et al., 1995; Warheit et al., 1990), but the significance of
21 these findings for ambient human exposures is yet to be determined.
22 To approach the difficult problem of determining if reported associations between low-
23 level PM concentrations and daily morbidity and mortality are biologically plausible, one
24 must consider: the chemical and physical characteristics of the particles in the inhaled
25 atmospheres; the characteristics of the morbidity/mortality observed and the affected
26 population; as well as potential mechanisms that might link the two. Several salient
27 considerations related to the evaluation of biological plausibility of the epidemiology findings
28 are discussed below.
29
30
April 1995 1_71 DRAFT-DO NOT QUOTE OR CITE
-------
1 Characteristics of Observed Morbidity and Mortality
2 If daily mortality rates are increased in association with elevated ambient paniculate
3 concentrations, what are the people dying of? Schwartz (1994) addressed this question by
4 comparing causes of death in Philadelphia on high pollution days (average = 141 /*g/m3)
5 with causes of deaths on lower pollution days ( average = 47 /^g/m3). On the high pollution
6 days there was a higher relative increase in deaths due to chronic obstructive pulmonary
7 disease (COPD) (RR = 1.25), pneumonia (RR = 1.13), cardiovascular disease (RR = 1.09)
8 and stroke (RR = 1.15). There was also an increase in reports of respiratory factors being
9 contributing causes in the deaths and a higher relative age of those dying. The patterns of
10 causes of death and age of those dying were found to be similar to the patterns observed in
11 the London smog deaths of 1952.
12 Other studies on associations of morbidity with paniculate pollution noted small
13 decreases (2 to 2.5%) in pulmonary function (FVC or FEVj) in smokers on high pollution
14 days (100 /ig/m3; Salt Lake City; Pope and Kanner, 1993) and in nonsmokers (>60 /zg/m3;
15 NHANES I data, Chestnut et al., 1991). An increased number of asthma attacks among
16 working age adults was correlated with increases in paniculate pollution over a 3-year period
17 (average particle level = 76 jug/m3) in Helsinki (Ponka, 1991). Thus, the characteristics of
18 health effects on high particle pollution days are mainly cardiopulmonary in nature and are
19 the types of effects that can be considered plausibly related to airborne toxicants.
20 It is also of interest to consider the health status of the people affected. People with
21 previously existing health conditions (such as COPD, asthma, or other chronic debilitating
22 conditions) are logically likely to be more susceptible to effects from exposure to paniculate
23 pollutants than would be healthy persons. Such a situation might result in an increased daily
24 mortality rate on days with higher PM10, followed by a decreased daily mortality rate so that
25 the average mortality rate over a longer time period would not be affected.
26 Data on the relative effect of particle exposures in persons with pre-existing pulmonary
27 disease compared to healthy persons do not yield a clear picture. Pope and Kanner (1°93)
28 reported an approximate 2% decline in FEVj in smokers with mild to moderate COPD
29 during an increased concentration in PM10 of 100 fig/m3 in Salt Lake City. However,
30 persons with severe COPD (average FEVj equal to 50% of predicted) had no further
31 reduction in pulmonary function upon acute (2 h) exposure to 90 fig/m3 H2SO4 in clinical
April 1995 1-72 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies (Morrow et al., 1994). Exercising asthmatics experienced mild bronchoconstriction
2 following the same exposures. In a separate study, exercising adolescent asthmatics exposed
3 to 68 £tg/m3 H2SO4 experienced reduced pulmonary function (average of 6% decrease in
4 FEVt) (Koenig et al., 1989), but in another study, exercising asthmatics did not respond to
5 exposures to as high as 130 jug/m3 H2SO4 (Avol et al., 1990). Using an elastase-induced rat
6 model of emphysema, Mauderly et al. (1990) found that exposure to diesel exhaust, which
7 contains aggregates of ultrafine soot particles, resulted in less particle deposition in the lungs
8 of emphysematous rats than in normal rats, thus sparing the emphyseniatous rats the health
9 effects induced by the soot particles in normal animals.
10
11 Influence of Particle Size, Chemical Composition, and Respiratory Tract
12 Deposition/Clearance
13 The PM10 standard is the only U.S. national ambient air quality standard that is not
14 chemical-specific. The chemical composition of a particle will greatly affect its toxicity and,
15 if possible, should be considered in determining if the observed associations between
16 atmospheric PM concentrations and increases in morbidity /mortality are causal. For
17 example, alpha-quartz particles are more toxic than TiO2 particles (Driscoll and Maurer,
18 1991); and acid sulfate aerosols are more likely to cause acute health effects than are neutral
19 sulfate aerosols (Fine et al., 1987).
20 Size is also important in defining the toxicity of particles. Recent studies indicate that
21 ultrafine particles (<20 nm) are much more toxic than larger inhalable particles (Oberdorster
22 et al., 1992; Driscoll and Maurer, 1991). The ultrafine particles have a greater number and
23 surface area per unit mass than fine or coarse particles, which may account, in part, for their
24 greater toxicity. Fine particles tend to have a different chemical composition than larger
25 particles, because their source is often combustion processes. A study of the chemical
26 composition of PM2 5 particles versus PM10 particles in Los Angeles indicated that nitrates,
27 sulfates, ammonium and organic and elemental carbon were the most abundant species in the
28 PM2 5 fraction, while the coarser particles contained soil-related species, such as aluminum,
29 silicon, calcium, and iron (Chow et al., 1994). Chemical composition of PM10 is discussed
30 in Chapter 3 and summarized earlier in this chapter (Section 13.2).
April 1995 1_73 DRAFT-DO NOT QUOTE OR CITE
-------
1 In a few epidemiology studies, the investigators attempted to determine what size
2 and/or chemical form of particles had the strongest association with health effects. For
3 example, in the Harvard 6-cities study (Dockery et al., 1993), the excess chronic mortality
4 was most strongly associated with the ambient fine particles, including sulfates. However, in
5 a study of daily air pollution in St. Louis and eastern Tennessee by Dockery et al. (1992),
6 the strongest association of particulate pollution with daily mortality rates was PM10, with
7 progressively weaker associations with PM2 5, sulfate, and aerosol acidity. This is the
8 opposite of what one would expect if aerosol acidity were the main cause of increased
9 mortality, as has been suggested (Lippmann, 1989). The Six Cities study investigators state,
10 however, that the low daily death counts, the short study period, and the large geographic
11 areas considered in the St. Louis/Eastern Tennessee study limited the statistical power of the
12 study, and they could not conclude that the acidity of the aerosol was not associated with
13 mortality.
14 If the chemical and physical forms of the PM are important in determining the health
15 effects induced by PM, one would expect different concentration-response curves to be
16 observed in different epidemiology studies, depending on the type of aerosol present in the
17 atmosphere. Spurny (1993) in his analysis of studies conducted in the south-western part of
18 Germany found differences in concentrations, composition, and cell-toxic effects among
19 urban, residential, and remote areas. The different slopes of the concentration-response
20 curves for the different cities could be due to several factors, including differences in
21 physicochemical properties and resultant potency of the PM in the different cities.
22 It is also worth noting that considerations of dosimetry could potentially provide insight
23 on plausible mechanisms or alter the exposure-dose-response relationships evaluated. To
24 date, most analyses have used the exposure concentration (/xg/m3) of particles. Because
25 deposition of particles in the respiratory tract are determined by particle diameter and
26 distribution, calculation of the RR estimates based on various internal dose metrics (e.g.,
27 deposited dose (mass) rate per tracheobronchial or alveolar surface area or deposited particle
28 number rate per surface area), could alter some of these relationships. Different dose
29 metrics may be more appropriate to characterize acute effects (e.g., mortality) versus chronic
30 effects (e.g., morbidity). Certainly dosimetry can provide insight on the variability of
31 inhaled dose due to differences in airway morphometry and ventilation rates among species,
April 1995 1-74 DRAFT-DO NOT QUOTE OR CITE
-------
1 age, genders, and disease status of the respiratory tract. For example, it has been shown that
2 patients with COPD have increased deposited particle burdens when compared to healthy
3 subjects (Anderson et al., 1992). To the extent that particle composition alters the particle
4 diameter and distribution of a given aerosol, dosimetry will also be effected. Solubility of an
5 aerosol influences clearance rates and subsequent retained dose estimates. The potential for
6 dosimetry to influence the exposure-dose-relationship should be considered to the extent that
7 mathematical modeling andd norphometric data allow.
8
9 Potential Mechanisms of Causality Between Low Levels of Particulate
10 Pollution and Health Effects
11 Pathophysiologic mechanisms by which various specific PM constituents can cause
12 health effects are discussed in detail in Chapter 11. Here, the focus is on mechanisms by
13 which airborne particles are known to cause health effects and the extent to which such
14 mechanisms provide plausible evidence or explanations for the reported epidemiologic
15 findings of increases in morbidity and daily mortality rates at low PM concentrations. For
16 purposes of this discussion, health effects of particle inhalation are discussed below in terms
17 of: clinical considerations, acute lung injury, chronic pulmonary toxicity from accumulation
18 of particles in the lung, effects on pulmonary function, effects on pulmonary defense
19 mechanisms, and pathophysicologic mechanisms. Also considered in this section is the
20 potential for interactive mechanisms among air pollutants that might influence the health
21 effects induced by airborne particles. This is an area in which there is little information;
22 most studies have been directed toward determining the toxicity of single compounds.
23
24 Clinical Considerations
25 Potential mechanisms which might help explain the phenomenon of particle related
26 mortality have been considered by Frampton and Utell (1995). These mechanisms include:
27 (1) "premature" death, that is the hastening of death for individuals already near death (i.e.,
28 hastening of an already certain death by hours or days); (2) increased susceptibility to
29 infectious disease; and (3) exacerbation of chronic underlying cardiac or pulmonary disease.
30 Particulate pollution could contribute to daily mortality rates by affecting those at
31 greatest risk of dying, i.e., those individuals for whom death is already very imminent.
April 1995 1-75 DRAFT-DO NOT QUOTE OR CITE
-------
1 Acute exposure to moderately elevated concentrations which might only be a minor irritant to
2 healthy people could be the "last straw" that tips over the precariously balanced physiology
3 of a dying patient. Other studies suggest that the full effect of particles on mortality cannot
4 be explained solely by acute PM exposure death-bed effects (Frampton and Utell, 1995), i.e.,
5 some studies also indicate an effect on annual mortality rates which cannot be explained
6 simply by the hastening of death for individuals already near death.
7 Particle exposure could also increase susceptibility to respiratory infection with bacteria
8 or viruses, leading to an increased incidence of (and death from) pneumonia in susceptable
9 members of the population. However, pneumonia rarely results in death within 24 h of
10 onset; serious infections of the lower respiratory tract generally develop and evolve over
11 weeks, and would not explain effects on daily mortality. If pollutant exposure increased
12 susceptibility to infectious disease, it should be possible to detect differences in the incidence
13 of such diseases in communities with low vs. high particle concentrations. Emergency room
14 visits and hospitalizations for pneumonia caused by the relevant agent should also be
15 measurably higher on days with elevated ambient particle concentrations. However, no such
16 relationship has been observed, and laboratory animal data to support such a mechanism are
17 weak.
18 Particulate air pollution might also aggravate the severity of underlying chronic lung
19 disease. This mechanism could explain increases in daily mortality (through effects on those
20 near death from their disease) and longitudinal increases in mortality (if individuals with
21 chronic airways disease experienced more frequent or severe exacerbation of their disease, or
22 more rapid loss of function as a result of paniculate exposure).
23 What chronic disease processes are most likely to be affected by inhaled paniculate
24 matter? To explain the daily mortality statistics, there must be common conditions that
25 contribute significantly to overall mortality from respiratory causes. The most likely
26 candidates are the chronic airways diseases, particularly chronic obstructive pulmonary
27 disease (COPD). This group of diseases encompasses both emphysema and chronic
28 bronchitis, however, information on death certificates does not allow differentiation between
29 these diagnoses. The pathophysiology includes chronic inflammation of the distal airways as
30 well as destruction of the lung parenchyma. There is loss of supportive elastic tissue, so that
31 the airways collapse more easily during expiration, obstructing outflow of air. Processes that
April 1995 1-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 enhance airway inflammation or edema, increase smooth muscle contraction in the
2 conducting airways, or slow mucociliary clearance could adversely affect gas exchange and
3 host defenses. Moreover, the uneven ventilation-perfusion matching characteristics of this
4 disease, with dependence on fewer functioning airways and alveoli for gas exchange, means
5 inhaled particles may be directed to the few remaining functioning lung units in higher
6 concentration than in normal lungs (Bates, 1992)
7 Particulate pollutants have been associated with increases in cardiovascular mortality
8 both in the major air pollution episodes and in the more recent time-series analyses. Bates
9 (1992) has postulated three ways in which pollutants could affect cardiovascular mortality
10 statistics. These include: (1) acute airways disease misdiagnosed as pulmonary edema;
11 (2) increased lung permeability, leading to pulmonary edema in people with underlying heart
12 disease and increased left atrial pressure and (3) acute bronchiolitis or pneumonia induced by
13 air pollutants precipitating congestive heart failure in those with pre-existing heart disease.
14 Moreover, the pathophysiology of many lung diseases is closely intertwined with cardiac
15 function. For example, one postulated cause of the increasing mortality rate in asthma is
16 overuse of adrenergic agonist medications leading to fatal cardiac arrhythmias. Many
17 individuals with COPD also have cardiovascular disease caused by smoking, aging, or
18 pulmonary hypertension accompanying COPD. Terminal events in patients with end-stage
19 COPD are often cardiac complications, and may therefore be misclassified as cardiovascular
20 deaths. Hypoxemia associated with abnormal gas exchange can precipitate cardiac
21 arrhythmias and sudden death.
22
23 Acute Lung Injury
24 The acute toxicity of particles in the respiratory tract has been the topic of numerous
25 studies to determine the potential pulmonary toxicity of dusts, particularly those of concern in
26 industrial processes. Toxic particles that deposit in the lung can induce an inflammatory
27 response that, if it persists, may lead to pulmonary fibrosis and impaired pulmonary function.
28 The response of the respiratory tract to such particles includes the release of numerous
29 cytokines from alveolar macrophages and epithelial lining cells that promote healing and
30 repair or, if healing does not occur because of the persistence of toxic particles, may
31 promote development of fibrosis. Although such acute responses are well known, they
April 1995 ^77 DRAFT-DO NOT QUOTE OR CITE
-------
1 typically only occur after several days or weeks of exposure to airborne particle
2 concentrations many fold higher than those that have been shown to be associated with
3 increased mortality and morbidity in epidemiology studies. Recently, however, it has been
4 observed in experimental animal studies that certain types of particles are acutely toxic to the
5 lung at low exposure concentrations. A half-hour exposure of rats to freshly generated
6 ultrafine polytetrafluoroethylene particles at a concentration of 64 ^g/m3 resulted in severe
7 pulmonary inflammation and death (Oberdorster et al, 1995). Warheit et al. (1990) also
8 found that fresh ultrafine aerosols resulted in mortality in rats by causing severe lung injury.
9 The significance for environmental exposures of the highly toxic fresh aerosols formed from
10 pyrolysis of perfluorinated materials is unknown at this time, because of the rapid loss of
11 toxicity of the aerosols with time and the lack of information on the concentration of those
12 specific aerosols in the ambient atmosphere. Although it is known that combustion processes
13 emit ultrafine aerosols into the environment (Cantrell and Whitby, 1978), it is not clear how
14 much ultrafine particulate matter is present as the product of pyrolysis of perfluorinated
15 compounds. Nor is there much information on typical ambient concentrations of other
16 ultrafine particles (e.g., metals from high temperature smelting) or their persistance as
17 ultrafines in urban aerosol mixes.
18
19 Toxicity Resulting from Accumulation of Particles in the Lung
20 The accumulation of large lung burdens of poorly soluble particles can lead to
21 decreased clearance of subsequently inhaled particles and an enhanced rate of accumulation
22 of particles in the lung (Morrow, 1992). Large lung burdens of particles of even relatively
23 low inherent toxicity have been shown to induce lung cancer in animal models such as the rat
24 (Mauderly et al., 1994). But how much prior exposure to particles is required to accumulate
25 enough particles to impair clearance of subsequently inhaled particles? Rats exposed to 350
26 Mg/m3 diesel soot (aggregated ultrafine carbon particles) for 24 months did not accumulate
27 enough particles to induce pulmonary inflammation (as measured by both histopathology and
28 analysis of lung lavage fluid) or to impair particle clearance, but rats exposed to 3500 ^g/m3
29 for the same length of time did. Rats that inhaled carbon black particles at an 8-h
30 time-weighted concentration of 10,000 jig/m3 5 days a week for 12 weeks also accumulated
31 enough particles to induce an inflammatory response by 6 weeks (Henderson et al., 1992).
April 1995 1-78 DRAFT-DO NOT QUOTE OR CITE
-------
1 In general, the toxicity resulting from accumulation of large burdens of particles in the
2 lung does not likely provide a plausible biological basis for reported associations between
3 acute exposures to low level PM concentrations (ca, 30 to 200 Mg/m3) of inhalable particles
4 (PM10) and daily mortality/morbidity rates. One possible exception that stands out as a
5 relatively sizeable segment of the general population would be smokers or former smokers
6 among the elderly. In such individuals, particle overload from 40 to 60 years of directly
7 inhaled tobacco smoke particles could make them more vulnerable to the impacts of
8 relatively small additional acute increments in their lung particle burdens, as would the
9 preexisting chronic cardiorespiratory diseases caused by smoking.
10 A second possible exception might be elderly persons who experienced notable past
11 exposures over many years to very high ambient or workplace PM concentrations, as would
12 be the case for individuals who resided or worked in heavily industrialized cities before
13 effective occupational and air pollution control measures were introduced in the 1950s to
14 1970s to reduce such exposures. For example, in the Harvard 6-cities study, an association
15 was found between daily mortality rates and PM levels across a few rural communities,
16 lightly industrialized cities, and some heavily industrialized cities. Because the ranking of
17 the cities in terms of air-pollution levels did not change during the study period, it is not
18 possible to distinguish completely between effects due to past historical exposures and those
19 due to recent exposures. Therefore, the elevation in daily mortality rates in industrialized
20 cities such as Steubenville compared to less industrialized cities (such as Topeka or Portage)
21 may be in part based on accumulated past exposures to higher particle levels and
22 consequently larger lung particle burdens in the former.
23
24 Impaired Respiratory Function
25 Very few of the specific PM constituents discussed in Chapter 11 have acute exposure
26 effects on respiratory function, except possibly at very high concentrations (in the /ng/m3
27 range). One possible exception is acid aerosols, which appear to have acute effects on
28 pulmonary function among some sensitive individuals at levels below 1,000 Mg/m3.
29 Exposures to acid particles are known to induce hyperreactive airways and in some cases,
30 bronchoconstriction, but at concentrations in the mg/m3 range, well above peak U.S. ambient
31 acidity levels of 50 to 75 /xg/m3. In healthy humans, inhalation of 1,000 Mg/m3 H2SO4
April 1995 1_79 DRAFT-DO NOT QUOTE OR CITE
-------
1 aerosol for 3 h did not cause any influx of inflammatory cells into the lung based on analysis
2 of lung lavage fluid obtained 18 h after the exposures (Frampton et al., 1992). However,
3 mild bronchoconstriction has been reported after brief exposures to as low as 68 jug/m3
4 H2SO4 in exercising adolescent asthmatics and 90 /ig/m3 in excersing adult asthmatics
5 (Morrow et al., 1994); (Koenig et al., 1989), although this has not always been observed
6 (Avol, et al., 1990). Also of interest is the finding that hyperresponsive airways developed
7 after exposure of healthy rabbits to as little as 75 jug/m3 H2SO4 for 3 h (El-Fawal and
8 Schlesinger, 1994). Additional studies have also found that acid-coated particles were more
9 potent than the acid or particles alone. Therefore, under some circumstances, one possible
10 mechanism for increased mortality among some elderly persons with a debilitating disease
11 (asthma) on days with moderately high PM pollution might be that acid aerosols place a
12 stress on their cardiopulmonary system, leading to death.
13
14 Impaired Pulmonary Defense Mechanisms
15 The ability of paniculate exposures to reduce pulmonary defense mechanisms has been
16 documented for aerosols of H2SO4 and trace metals. Trace metals have been shown to be
17 cytotoxic to alveolar macrophages (AMs) and immunosuppressive, but only at much higher
18 concentrations than encountered in ambient atmospheres (Zelikoff et al., 1993). Sulfuric
19 acid aerosols have also been shown to alter resistance to bacterial infection in mice after
20 acute exposures to 1,000 fj.g/m3; repeated exposures to 100 /jg/m3 reduced mucociliary
21 transport rates in animals. Even these levels of H2SO4 are much higher than have been
22 reported in atmospheres of cities evaluated in the recent epidemiology studies. Also, one
23 would expect effects from impaired pulmonary defense mechanisms to develop over an
24 extended period of continuing exposure, not within a few days.
25
26 Synergistic Effects
27 An area for which there is little information is the potential interactive effects of
28 mixtures of air pollutants and/or with other factors (e.g., aging). The potential significance
29 of mixtures is illustrated by the studies of Amdur and Chen (1989), in which a repeated daily
30 3-h exposure for 5 days of guinea pigs to 20 /ng/m3 of H2SO4 coated on metal particles
31 resulted in decrements in lung volume and pulmonary diffusing capacity and elevations of
April 1995 1-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 lung weight/body weight ratio, protein, and number of neutrophils in pulmonary lavage fluid.
2 For example, A 1-h exposure to 20 /ag/m3 H2SO4 coated on metallic particles increased
3 bronchial reactivity in guinea pigs; a 10-fold higher concentration of H2SO4 alone was
4 required to produce the same response (Chen et al., 1992b). However, such synergistic
5 effects were not observed by Anderson et al. (1992), who studied the effects on 15 healthy
6 and 15 asthmatic volunteers of 1-hr exposures to 100 /ng/m3 H2SO4 (0.5 /^m) or 250 /-ig/m3
7 carbon black (0.5 /xm) separately or with the H2SO4 coated on the particles. The exposures
8 did not result in changes in symptoms or pulmonary function, except for an equivocal
9 response in one person.
10 The population segment most susceptible to elevations in ambient PM are the elderly
11 (> 65 years old) with preexisting respiratory disease. Aging, in the absence of pathology, is
12 an extremely complex biological phenomenon and is described as being a multifactorial
13 process composed of both genetic and environmental components (Cristofalo et al., 1994).
14 While the physiological characteristics of the healthy older population is an area of active
15 research, significant decrements in key physiological parameters including lung volumes,
16 FEVj, flow velocity/volume curves, resting cardiac output, and cardiac output reserve with
17 age have been reported (Kenney, 1989). However, there is controversy concerning
18 decrements in physiological function associated with the aging process alone as well as with
19 accompanying disease processes or with other environmental stressors. Moreover, there is
20 little information on the extent to which an older population might be more susceptible to the
21 effects of ambient paniculate pollution (Cooper et al., 1991). It is possible that the elderly
22 are more susceptible to ambient particles because of numerous changes in the body's
23 protective mechanisms and protracted exposures to particles over a life time. This could
24 allow time for latent effects from earlier life time exposures to manifest themselves, and for
25 potential cumulative effects to emerge. Virtually nothing is known of the possibilities for
26 interaction among toxicants over a long life time or the possibilities for interaction between
27 medications and ambient pollutants.
28
29
April 1995 1-81 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pathophysiologic mechanisms
2 The respiratory system may be compromised and become less efficient in older people
3 or as a result of disease, and inhaled particles could, conceivably, further compromise their
4 respiratory function. Because small increases in environmental particle concentrations would
5 not be lethal to most people, the effect must result from initiating or promoting a lethal
6 failing of a critical function, such as ventilation, gas exchange, pulmonary circulation, or
7 cardiorespiratory control in subjects brought to the limits of tolerance by preexisting
8 conditions (Mauderly, 1995).
9 Inhaled particles or their pathophysiological reaction products could further impair
10 ventilation in the chronically ill individual by further reducing airway caliber. For example,
11 particles may activate airway smooth muscle, constricting airways, or may influence various
12 airway secretions which could add to and thicken the mucous blanket. Inhaled particles or
13 their pathophysiological reaction products could decrease the diffusing capacity of the lungs
14 by decreasing the area of the respiratory membrane available for diffusion, by increasing
15 diffusion distances across the respiratory membrane, and/or by causing abnormal ventilation-
16 perfusion ratios in some parts of the lung. Particles or their products could also act at the
17 level of the pulmonary vasculature to elicit changes in pulmonary vasculature resistance,
18 which could further alter ventilation-perfusion abnormalities in people with respiratory
19 disease. Furthermore, particles could conceivably alter respiratory and cardiovascular
20 control by affecting local control mechanisms located in the endothelial cells or other sites.
21 This could produce changes in peripheral and central control mechanisms and directly affect
22 the respiratory and cardiovascular control centers. Little evidence is currently available that
23 directly addresses the above speculative possibilities.
24
25 Biological Plausibility Conclusions
26 Having considered the characteristics of the paniculate exposure atmospheres and the
27 types of morbidity and mortality associated with the polluted atmospheres, what can be
28 concluded about the biological plausibility of the epidemiological results? The types of
29 morbidity and mortality reported to be associated with increased ambient particle
30 concentrations are consistent with the types of health effects that one might expect from
31 exposures to high levels of PM. Therefore, the type of response seems plausible, if one
April 1995 1-82 DRAFT-DO NOT QUOTE OR CITE
-------
1 accepts the temporal relationships modeled in the epidemiological studies. The analyses
2 found associations with 1-day or multi-day (usually 3 to 5 day) lags. The concentrations of
3 paniculate matter reported to be associated with such health responses, however, are much
4 lower than would be expected based on animal and human clinical studies of responses to
5 single particulate pollutants. This is true even when one considers that there is evidence that
6 the people who make up the excess mortality population may be susceptible subpopulations.
7 Moreover, it is not clear what portion of the inhalable particulate matter constitutes the
8 delivered dose that is associated with the observed morbidity or mortality. There are
9 suggestions from both animal toxicology data and epidemiology data that ultrafine acid
10 aerosols may be of greater health significance than the rest of the particulate mass. Finally,
11 the potential for interactive effects between PM of different types and 'PM and other air
12 pollutants is not known.
13 Thus, although there are several hypotheses as outlined above, little clear or convincing
14 evidence is available at this time to support the biological plausibility of a causal relationship
15 for the reported epidemiologic associations between low ambient concentrations of PM and
16 daily mortality and morbidity rates.
17
18
19 1.14 IDENTIFICATION OF POPULATION GROUPS POTENTIALLY
20 SUSCEPTIBLE TO HEALTH EFFECTS FROM PM EXPOSURE
21 Certain groups within the population may be more susceptible to the effects of PM
22 exposure, including persons with preexisting respiratory disease, children, and the elderly.
23 The reasons for paying special attention to these groups is that (1) they may be affected by
24 lower levels of PM than other subpopulations and (2) the impact of an effect of given
25 magnitude may be greater. Some potential causes of heightened susceptibility are better
26 understood than others. Subpopulations that already have reduced ventilatory reserves (e.g.,
27 the elderly and persons with asthma, emphysema, and chronic bronchitis) would be expected
28 to be more impacted than other groups by a given decrement in pulmonary function. For
29 example, a healthy young person may not even notice a small percentage change in
30 pulmonary function, but a person whose activities are already limited by reduced lung
31 function may not have the reserve to compensate for the same percentage change.
April 1995 1-83 DRAFT-DO NOT QUOTE OR CITE
-------
1 Based on Chapter 12 discussions, it is clear that the bulk of the total mortality effects
2 suggested by the epidemiology studies discussed earlier are among the elderly. During the
3 historic London, 1952 pollution episode the greatest increase in the mortality rate was among
4 older citizens and those having respiratory diseases. An analysis by Schwartz (1994c) of
5 mortality in Philadelphia, PA found the greatest increase in risk of death in those aged 65 to
6 74 and those >74 year of age (mortality risk ratios = 1.09 and 1.12, respectively, between
7 high and low TSP days). Other studies also suggest that the elderly experience a higher
8 excess risk from exposure to PM air pollution than the population overall.
9 Other potentially susceptible groups include patients with COPD, such as emphysema
10 and chronic bronchitis. Some of these patients have airway hyperresponsiveness to physical
11 and chemical stimuli. A major concern with COPD patients is the absence of an adequate
12 ventilatory reserve, a susceptibility factor described above. In addition, altered distribution
13 of respiratory tract ventilation in COPD may lead to a greater delivery of PM to the segment
14 of the lung that is well ventilated, thus resulting in a greater regional tissue dose. Also, PM
15 exposure may alter already impaired defense mechanisms, making this population potentially
16 more susceptible to respiratory infection. It is estimated (U.S. Department of Health and
17 Human Services, 1990; Collins, 1988) that 14 million persons (-6%) suffer from COPD in
18 the United States. Bronchial mucous transport clearance may be impaired in people with
19 chronic bronchitis, asthma, and in association with various acute infections. Rates of
20 alveolar region clearance appear to be reduced in humans with chronic obstructive lung
21 disease.
22 Throughout the results and discussions presented above and in Chapter 12 regarding the
23 effects of acute PM exposure on human mortality, a consistent trend was for the effect
24 estimates to be higher for the respiratory mortality category. This lends support to the
25 biological plausibility of a PM air pollution effect, as the breathing of toxic particles would
26 be expected to most directly affect the respiratory tract, and these results are consistent with
27 this expectation. For example, the estimates of relative risk for PM-induced mortality due to
28 respiratory causes discussed in Chapter 12 are all higher than the risks for the population as
29 a whole and for other causes. More specifically, the PM RR for respiratory diseases ranged
30 from 50 to more than 400% higher for respiratory disease categories than for all causes of
31 death, indicating that increases in respiratory deaths are a key major contributor to the
April 1995 1-84 DRAFT-DO NOT QUOTE OR CITE
-------
1 overall PM-mortality associations noted previously. PM relative risk estimates for
2 cardiovascular causes were also notably elevated. Moreover, since evidence suggests that an
3 acute pollution episode is most likely be inducing its primary effects by stressing already
4 compromised individuals (rather than, for example, inducing chronic respiratory disease from
5 a single air pollution exposure episode), the above results indicate that persons (especially the
6 elderly) with pre-existing cardiovascular or respiratory disease constitute a population
7 segment especially at risk for mortality implications of acute ambient PM exposures.
8 Apropos to the identification of individuals with pre-existing respiratory and
9 cardiovascular diseases as being at special risk for ambient PM exposure effects, it is
10 important to highlight smoking as a key etiological agent for such diseases. The U.S.
11 Environmental Protection Agency (1992) report on environmental tobacco smoke indicates
12 that smoking is the major cause of chronic obstructive pulmonary disease (COPD), which
13 includes emphysema, and is thought to be responsible for approximately 61,000 COPD
14 deaths yearly, i.e., about 82% of U.S. COPD deaths (U.S. DHHS, 1989). Tobacco use is
15 also a major risk factor for cardiovascular diseases, the leading cause of death in the United
16 States. It is estimated that each year 156,000 heart disease deaths and 26,000 deaths from
17 stroke are attributable to smoking (CDC, 1991). Smoking is also a risk factor for various
18 respiratory infections, such as influenza, bronchitis, and pneumonia. An estimated 20,000
19 influenza and pneumonia deaths per year are attributable to smoking (CDC, 1991).
20 The U.S. Environmental Protection Agency report also indicates that in children, ETS
21 exposure is causally associated with an increased risk of lower respiratory tract infections
22 such as bronchitis and pneumonia. It is estimated that 150,000 to 300,000 cases annually in
23 infants and young children up to 18 months of age are attributable to ETS. ETS exposure is
24 also causally associated with additional episodes and increased severity of symptoms in
25 children with asthma. It is estimated that 200,000 to 1,000,000 asthmatic children have their
26 condition worsened by exposure to ETS. ETS is also a risk factor for new cases of asthma
27 in children who have not previously displayed symptoms (U.S. Environmental Protection
28 Agency).
29 Lastly, the EPA report also indicates that environmental tobacco smoke (ETS) is a
30 human lung carcinogen, responsible for approximately 3,000 lung cancer deaths annually in
31 U.S. nonsmokers (U.S. Environmental Protection Agency, 1992).
April 1995 !_85 DRAFT-DO NOT QUOTE OR CITE
-------
1 Overall, then, the most susceptible population segment that can be most clearly
2 identified as being at likely increased risk for low-level ambient PM exposure-induced
3 mortality or morbidity are elderly individuals with pre-existing cardiovascular respiratory
4 diseases, the majority of which are likely either current or former smokers. Smoking may
5 also be a key ancillary contributor to any low-level PM exposure-induced exacerbation of
6 respiratory infections among other adults and children and to any increased cancer mortality
7 attributable to chronic ambient PM exposures.
8 Asthmatic subjects appear to be more sensitive than healthy subjects to the effects of
9 acid aerosols on lung function, but the effective concentration differs widely among studies.
10 Adolescent asthmatics may be more sensitive than adults, and may experience small
11 decrements in lung function in response to H2SO4 at exposure levels only slightly above peak
12 ambient levels. Although the reasons for the inconsistency among studies remain largely
13 unclear, subject selection may be an important factor. Even in studies reporting an overall
14 absence of effects on lung function, occasional asthmatic subjects appear to demonstrate
15 clinically important effects. Studies from different laboratories suggest that responsiveness to
16 acid aerosols may correlate with degree of baseline airway hyperresponsiveness. On the
17 other hand, based on very limited studies, elderly and individuals with chronic obstructive
18 pulmonary disease do not appear to be particularly susceptible to the effects of acid aerosols
19 on lung function.
20 Alveolar deposition at different flow rates was lower (26% versus 48% thoracic
21 deposition) in subjects after induced bronchoconstriction. In asthmatics, thoracic deposition
22 of particles was higher than healthy subjects (83% versus 73% of total deposition).
23 Trachial/bronchial deposition was also found to be higher in asthmatics. The results are
24 similar to those found in subjects with obstructive lung disease. The buffering capacity of
25 mucus may be altered in persons with compromised lungs. For example, sputum from
26 asthmatics had a lower pH than that from normals and a reduced buffering capacity, and so
27 may represent a population segment especially sensitive to inhaled acidic particles.
28 The National Institutes of Health (1991) estimates that approximately 10 million persons
29 in the United States have asthma. In the general population, asthma prevalence rates
30 increased by 29% from 1980 to 1987. For those under 20 years old, asthma rates increased
31 from approximately 35 to 50 per 1,000 persons, a 45% increase. The airways of asthmatics
April 1995 1-86 DRAFT-DO NOT QUOTE OR CITE
-------
1 may be hyperresponsive to a variety of inhaled materials, including pollens, cold-dry air,
2 allergens, and air pollutants. The potential addition of an PM-induced increase in airway
3 response to the already heightened responsiveness to other substances raises the possibility of
4 exacerbation of this pulmonary disease by PM.
5
6
7 1.15 IMPLICATIONS OF RELATIVE RISK ESTIMATES
8 Preceding sections of this Chapter concluded that the newly emerging epidemiologic
9 data base on PM-mortality/morbidity effects provides reasonably consistent results indicative
10 of increased risk of mortality and morbidity effects being associated with exposures of the
11 general population to ambient air pollutant mixes containing PM concentrations currently
12 found in many U.S. urban areas. This includes effects associated with ambient air exposures
13 to pollutant mixes having 24-h PM10 concentrations falling in the range of 30 Mg/m3 to 200
14 /*g/m3, including evidence suggestive of effects below 150 /*g/m3 (the level of the current 24-
15 h U.S. PM10 NAAQS).
16 It was also noted in Chapter 12 that the relative risk (RR) estimates for both the
17 mortality and morbidity effects associated with short-term (ca. 24-h or a few days) exposures
18 to ambient PM are very small compared to RR values typically viewed in epidemiologic
19 literature as providing strong evidence for a likely causative association. Section 1.12 further
20 noted the relatively limited evidence directly demonstrating coherence between the mortality
21 and morbidity effects findings from epidemiologic studies, with the most compelling evidence
22 for coherence now being findings of both increased hospital admissions (for cardiopulmonary
23 endpoints) and increased mortality in relation to increments in 24-h PM concentrations in the
24 same population group (the elderly) within several U.S. urban areas (Detroit, Birmingham,
25 Philadelphia) and the Utah Valley. However, only very limited evidence for the biological
26 plausibility of acute low-level PM exposure effects at the above-stated PM10 concentration
27 range now appears to exist to support several hypotheses discussed in Section 1.13 with
28 regard to possible mechanisms of action. A key point emerging from the plausibility
29 discussion and the ensuing section (1.14) was the identification of elderly individuals (65 yr.
30 old) with preexisting chronic cardiovascular and respiratory disease conditions (the majority
31 of whom are likely current or former smokers) as being the most susceptible general
April 1995 1-87 DRAFT-DO NOT QUOTE OR CITE
-------
1 population segment most clearly at special risk for mortality and morbidity effects associated
2 with exposures to ambient air mixes containing moderately elevated PM concentrations.
3 The meaning or interpretation of quantitative estimates of PM-related effects (i.e.,
4 relative risk estimates) discussed earlier as having been generated by the newly available PM
5 epidemiology studies remains a subject of controversy, with divided opinions still existing in
6 the scientific community as noted earlier in this chapter. Thus, in attempting to interpret
7 such risk estimates, several caveats should be kept in mind. First, caveats analogous to those
8 made in point (4) at the top of page 1-33 for key conclusions drawn from the last previous
9 PM criteria review still apply. That is, although new evidence has emerged which points
10 toward very small, but statistically significant increases in risk of human mortality and
11 morbidity effects being associated with exposures to ambient air mixes containing moderately
12 elevated PM (with no evident thresholds being identified in the studied range of PM
13 concentrations), precise quantitative specification of relative contributions of such low-level
14 concentrations of ambient PM to reported mortality and morbidity effects is not possible at
15 this time. Nor can one now separate out with confidence potential relative contributions to
16 the reported PM effects of several other likely important confounding or interacting
17 variables.
18 With regard to the latter, it is as of yet very difficult, for example, to sort out with
19 confidence relative contributions of weather versus PM per se. It is clear that temperature
20 extremes (very hot or very cold days in relation to typical ranges of temperature for any
21 given locale) have notable effects on variations in daily mortality, with temperature or other
22 combinations of variables indexing weather impacts usually being found to be significant
23 predictors of daily human mortality in modeling of PM effects and to account for distinctly
24 larger proportions of the variance in daily mortality than do indices of PM pollution. On the
25 other hand, in most of the newer PM studies, small elevations in relative risk attributable to
26 PM still remained even after control for temperature extremes and/or other weather indices;
27 and PM effects were found to be significant in several analyses (e.g., for London) restricted
28 to days not involving wide variations in temperature that would constitute geographic-specific
29 extremes. It is also not yet clear to what extent any given relative risk estimate derived from
30 any of the newer analyses represent actual risk due to an increase in ambient PM or to what
31 extent the elevations in risk attributed to modeled PM indices more broadly represent
April 1995 1-88 DRAFT-DO NOT QUOTE OR CITE
-------
1 increased mortality or morbidity risks due to human exposure to the overall pollutant mix in
2 the particular airshed evaluated (including not only the ambient PM aerosols present but
3 other copollutants, such as SO2, CO, O3, NOX or non-particulate organic air toxics).
4 Other caveats bear on the issue of how generalizable the reported PM relative risk
5 estimates are. It is not yet possible to determine the extent to which the risk estimates for
6 PM mortality or morbidity effects are generalizable to other geographic areas or are highly
7 site-specific, i.e., narrowly applicable to the specific cities from which they were derived or,
8 at least, most credibly confined for use in projecting any estimates of likely PM risk to other
9 airsheds with fairly similar ambient aerosol mixes in terms of particle size distribution and
10 chemical composition. Thus, it is not clear, for example, how credible the use of PM-
11 related relative risk estimates derived from Philadelphia, St. Louis, or other midwestern or
12 eastern U.S. conurbanations (or foreign cities such as Sao Paulo, Santiago, or Athens) with
13 high percentages of particles from combustion processes might be in attempting to estimate
14 PM-related risks for other cities, e.g., in the western U.S., with much greater proportions of
15 crustal materials in the ambient air pollutant mix. Use of presently available PM-related risk
16 estimates to attempt to quantify potential PM-related risks across various seasons in locales
17 where widely varying seasonal mixes of particles of different sizes/chemical composition may
18 also be of dubious scientific credibility at this time.
19 Another issue of much interest and debate has been that of "threshold" for the estimated
20 PM effects derived from the newly available analyses. As noted earlier and discussed in
21 Chapter 12, no evident thresholds have yet been demonstrated for reported PM-related
22 mortality or morbidity effects, based on the presently available published analyses. On the
23 other hand, as also discussed in Chapter 12, only very limited efforts have been made to date
24 to undertake statistical analyses by which to more definitively address the issue; and serious
25 doubt exists as to whether any thresholds, if they do exist, even in the range of the observed
26 data (i.e., roughly from 30 to 200 /ug/m3 PM10) can be demonstrated, given notable
27 statistical power limitations associated with necessary breaking down of data into more
28 refined intervals as part of any threshold "search". Nor is there now any scientifically
29 credible basis by which to make a "no-threshold" argument in support of extrapolating
30 currently available PM relative risk estimates to ambient PM concentrations below the range
31 of observed data used in the reported analyses. This is especially true in view of the lack of
April 1995 1-89 DRAFT-DO NOT QUOTE OR CITE
-------
1 any well demonstrated evidence for one or another hypothesized potential mechanisms of
2 action that might plausibly explain the elevated risk of mortality or morbidity at the very low
3 PM concentrations implied by the results of the newly available epidemiology studies.
4 It is also clear from the available analyses that the occurrence of any increased risk of
5 mortality or morbidity due to short-term moderate elevations in PM (either alone or in
6 concert with other copollutants) likely represents the outcome of a combination of risk factors
7 culminating in relatively rare health events (as clarified further by the ensuing quantitative
8 discussion below). By far the greatest risk is posed for the elderly over 65 years old and
9 especially those with preexisting cardiopulmonary diseases, with very distinctively lower risk
10 estimates having been derived for younger individuals and those without chronic respiratory
11 or cardiovascular diseases. Thus, in order for notable health effects to occur in association
12 with short-term exposures to ambient PM (and/or copollutants), it appears that other
13 predisposing conditions and/or contributing risk factors must be present, as well. That is,
14 low-level ambient PM exposures alone do not typically appear to be sufficient per se to
15 induce increased morbidity or mortality, but may contribute to such health outcomes under
16 conditions when one or more other contributing risk factors also converge. Thus, for
17 example, short-term low-level exposures to ambient PM at concentrations in the ranges
18 evaluated in most of the newer epidemiology studies are extremely unlikely alone to cause
19 lung function decrements or respiratory symptoms of much note (except possibly for some
20 highly sensitive asthmatic patients), based on currently available epidemiologic and controlled
21 human exposure study results. Nor are such exposures likely to markedly reduce or impair
22 respiratory tract defenses (e.g., alveolar macrophage numbers or function, retrociliary
23 clearance mechanisms, lung immune response, etc.) sufficiently so as to cause increased
24 susceptibility to respiratory infections, based on available experimental toxicology findings.
25 On the other hand, once a respiratory infection were to occur due to other causes, then it is
26 conceivable that added stress due to low-level PM exposure in terms of small further
27 decrements in pulmonary function or exacerbation of respiratory symptoms could lead to
28 worsening of the acute illness and, possibly, the need for medical attention and/or hospital
29 admission in some cases.
30 Still additional converging risk factors appear to be necessary for exposures to ambient
31 air pollution mixes containing low concentrations of typical outdoor urban aerosols to
April 1995 1-90 DRAFT-DO NOT QUOTE OR CITE
-------
1 contribute to increased mortality. By far the most important are the cooccurrence of
2 advanced age (> 65 yr old) and already compromised cardiopulmonary function. In older
3 individuals with preexisting COPD, emphysema, chronic heart disease, etc. resulting from
4 other predisposing risk factors (e.g., long-term earlier high-level particle exposures from
5 smoking or past occupational or ambient PM exposures before effective control measures
6 were introduced), it appears conceivable that additional stress from low-level ambient PM
7 exposures might cause further complications that might lead to terminal consequences in
8 some cases. Several possibilities were discussed earlier as having been hypothesized, e.g.,
9 increased air flow to and consequent greater particle deposition/retention in remaining
10 functioning areas of the compromised lung, possible tipping over by small additional particle
11 burdens of already saturated lung defenses due to particle overloads from past long-term high
12 level particle exposures, and/or the induction of cascading inflammatory or other immune
13 responses (due to particularly toxic specific PM constituents e.g., possibly certain transition
14 metals) that overwhelm remaining lung reserves and oxygen exchange mechanisms.
15 However, at this time, no clearly convincing scientific evidence has yet been reported by
16 which to either compellingly substantiate or refute such hypothesized possibilities. Thus,
17 considerable uncertainties still exist with regard to what the relative risk estimates from the
18 newly available epidemiologic studies may imply.
19 In evaluating the potential public health significance of the relative risk increases for
20 mortality or morbidity effects reported in the newer PM epidemiology studies, much recent
21 interest has focussed on use of such relative risk estimates to generate projections of numbers
22 of excess deaths or morbidity events likely to be associated with ambient PM exposures at
23 concentrations currently found in the United States or other countries. Given the above-
24 noted caveats and uncertainties pertaining both to the derivation of the relative risk estimates
25 and their interpretation, there are substantial reasons to caution against attempting such
26 calculations at this time and to have major reservations about accepting any such projections
27 as credible quantitative estimates of additional deaths or morbidity events likely to actually
28 occur with current or future exposures in the United States or elsewhere. At best, such
29 projections might be associated with exposures to PM-containing ambient air mixes in cities
30 with closely similar particle size/chemical composition characteristics and population
31 demographics to those cities from which the relative risk estimates were derived. It is
April 1995 j_91 DRAFT-DO NOT QUOTE OR CITE
-------
1 currently questionable as to whether widely generalizable, broadly applicable projections can
2 be made based on some single "best estimate" of PM-related relative risk and, also, whether
3 such projections can be credibly aggregated across PM exposure variations during different
4 seasons and/or across geographic locales with widely disparate mixes of PM aerosols and/or
5 other copollutants.
6 Despite the above caveats and reservations, however, it may be useful to provide
7 illustrative examples of possible quantitative implications of relative risk estimates of the type
8 generated by the recent PM epidemiology analyses. Table 1-5 earlier showed that total acute
9 mortality relative risk estimates associated with exposure to ambient air pollution having a 50
10 Mg/m3 increase in one-day 24-h average PM10 can range from 1.015 to 1.085, depending
11 upon the site (i.e., the PM10 composition and population demographics) and also upon
12 whether PM10 is modeled as the sole index of air pollution or not. Relative Risk estimates
13 with PM10 as the only pollutant index in the model range from RR = 1.025 to 1.085, while
14 the PM10 RR with multiple pollutants in the model range from 1.015 to 0.025. The former
15 range, as noted earlier, might be viewed as approximating an upper bound of the best
16 estimate, as any mortality effects of co-varying pollutants may be "picked up" by the PM10
17 index, whereas the latter multiple pollutant model range might be viewed as approximating a
18 lower bound of the best estimate, as the inclusion of highly correlated covariates might
19 weaken the PM10 estimate. Thus, "typical" total mortality effect estimates (per 50 /xg/m3
20 PMio increase) most likely fall within an approximate RR = 1.025 to 1.06 range, based on
21 the various coefficients reported in the published studies. Formal EPA meta-analyses results
22 discussed in Chapter 12 yielded a best estimate of 1.031 with 95% confidence intervals (CI)
23 of 1.025 to 1.038 for PM10 studies using models that include 0-1 day lags but no
24 copollutants. For those analyses with longer lag times (3-5 days) and no copollutants in the
25 models, the EPA meta-analyses yielded a best estimate of 1.064 (CI = 1.047 to 1.082).
26 Thus, the very small increased risks of about 3.1 to 6.4% over baseline mortality levels (per
27 50 Mg/m3 increase in 24-h PM10 concentration in the 30 to 200 /xg/m3 range) derived from
28 the EPA meta-analyses probably represent currently best available upper bound estimates for
29 reported PM10-related total mortality effects. Lower bound estimates, from analyses that
30 included other copollutants in the models for acute PM-mortality effects, could be as much as
April 1995 1-92 DRAFT-DO NOT QUOTE OR CITE
-------
1 50% lower than the above upper bound estimates or, possibly, even include zero (i.e.,
2 represent no increased risk) especially during some seasons in different locales.
3 To help place these findings into a context by which to better understand the potential
4 implications of such relative risk estimates, Table 1-14 summarizes important information by
5 which to project potential increases in excess mortality in a city of one million people that
6 might be associated with exposure to ambient air mixes in which an increment in 24-h PM10
7 exposure of 50 /zg/m3 may be a contributing factor. First it is important to note that the
8 typical general population baseline or background risk of death on any given day in the U.S.
9 is only about 23.6 in a million (23.6 x 10"6) or 23.6 deaths per day in a city of 1 million
10 people. If the 24-h PM10 concentration increased by 50 /xg/m3 on a given day (e.g., from a
11 usual level of about 50 /ug/m3 to around 100 /xg/m3) then risk for mortality in the total
12 general population would be expected to increase by about 3.0 to 6.0% over baseline, i.e., to
13 increase from 23.6 in a million to about 24 or 25 in a million as an upper bound estimate.
14 In other words, exposure to the ambient mix of pollutants indexed by the 50 jwg/m3 increase
15 in 24-h PM10 levels might contribute to as much as an additional 0.7 to 1.5 deaths per
16 million people exposed, as shown in the Table 1-14 far right column.
17 Of the 23.6 baseline deaths per day expected in a city of 1 million, about 17 would be
18 attributable to elderly individuals (aged 65 or over), who only constitute about 12.6% of the
19 1991 U.S. population but for whom the background risk of dying on a given day is much
20 higher than for the total general population. For such individuals, the upper bound estimate
21 for increased numbers of excess deaths possibly contributed to by the 50 /jg/m3 PM10
22 increase would be projected to be approximately 1.0 (more than half of the higher estimate
23 for total mortality among the entire population), assuming that the specific city has a typical
24 demographic distribution of percentages of people in different age brackets. In other cities
25 or locations with notably higher elderly populations (e.g., some retirement communities or
26 cities left with higher percentages of the elderly possibly due to outmigration of younger
27 people), then the overall risk and expected deaths per day would be higher. Conversely, in
28 other locations with much younger than average populations and lower percentages of elderly
29 residents, the risk and expected numbers of PM-related excess deaths would be lower.
30 If the increment in PM10 concentration continued to average about 50 /xg/m above
31 routine ambient levels for 3-5 days in the given city of 1.0 million people, then relative risk
April 1995 1.93 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimates derived from PM]0-mortality models using 3-5 d lags might more appropriately
2 apply. Then, the expected number of deaths to which the 3-5 day 50 pig/m3 PM elevation
3 might be projected to contribute could range up to about 1.5 deaths per day among the total
4 general population per 1 million people exposed; or up to about 4.5 to 7.5 deaths during the
5 full 3-5 days of elevated PM10. Of the 1.50 excess death per day attributed to the 3-5 day 50
6 /xg/m3-increment in PM10 24-h concentrations, an estimated 0.34 would likely be due to
7 respiratory causes and about 0.91 to cardiovascular causes. Obviously, both the increased
8 deaths due to respiratory and cardiovascular causes would mainly occur in elderly persons
9 having preexisting chronic respiratory or cardiovascular disease conditions. Note that small
10 numerical inconsistencies in Table 1-14 and in succeeding tables on morbidity arise from the
11 fact that the excess risk estimates are based on different studies in a number of different
12 populations, with different baseline death or hospital admissions rates.
13 Table 1-15 simply takes the information from the far right column of Table 1-14 on
14 upper bound estimates of the number of possible PM-contributory deaths per day (for the
15 total population, for the elderly 65 + , and for respiratory and cardiovascular causes), and
16 depicts ranges of lower and upper bound estimates for comparable numbers of estimated
17 possible deaths per day contributed to by exposure to ambient air pollution mixes containing
18 50 jug/m3 increments in PM10 concentrations in cities ranging from 10 thousand to 10 million
19 in size. Table 1-14 is extremely informative in showing that no appreciable risk for
20 mortality is expected to occur with exposure to such ambient air mixes for cities less than 1
21 million population, even if the PM10 elevation lasts for 3-5 days or occurs several times a
22 year; nor is there much appreciable risk for the elderly in smaller population cities, unless
23 perhaps a particular city with less than 1 million population has an extraordinarily high
24 percentage of elderly residents. This applies even for days when 100 /xg/m3-increments in
25 PM10 might occur for 3-5 days in a row. Even for cities of 1 million population, the
26 projected upper bound risk may be of dubious public health significance unless 50-100 /xg/m3
27 PM10 elevations were to occur numerous times per year, especially in view of such tiny
28 increased risk likely mainly being posed for elderly individuals with preexisting
29 cardiopulmonary disease conditions that predominantly arise from voluntary smoking. Any
30 risk of excess mortality associated with short-term, acute exposures to ambient air pollutant
April 1995 1-94 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-14. ESTIMATED EXCESS MORTALITY PER DAY IN A POPULATION
OF ONE MILLION FOR WHICH AN INCREASE OF 50 jtg/m3 PM10 (24-h)
COULD BE A CONTRIBUTING FACTOR
Health
Outcome
Total
Mortality
Total
Mortality
Respiratory
Mortality
Cardiovascular
Mortality
Age
Group
All
65 +
All
All
All
Population
Baseline
Annual
Mortality
S^OB1
6,2013
8,603
6761
3,635
Population
Baseline
Daily
Mortality
23.6
17.0
23.6
1.85
10.0
PM10
Lag
Time
< 2d
2d
3-5d
3-5d
3-5d
Upper Bound
Excess Risk
Per PM,r,
50 pg/rr?
0.032
0.064
0.062
0.195
0.095
Possible
Number of
PM-Related
Deaths/Day
0.7
1.0
1.5
0.3
0.9
'From Monthly Vital Statistics Report for 1991 (U.S. CD 1993).
2From EPA meta-analyses, Table 12-25; all models without co-pollutants
3Elderly as 12.6% of 1991 U.S. population
4From Saldiva and Bohn (1994) and Ostro et al. (1995), variance-weighted average (TWA); Section 12.3.1.3
5From Pope, et al. (1991), Schwartz (1993) for Utah Valley and Birmingham TWA, Table 12-4
TABLE 1-15. ESTIMATED NUMBER OF DEATHS PER DAY IN CITIES OF
10,000 to 10 MILLION^OR WHICH AN INCREASE OF 50 jtg/m3 PM10
COULD BE A CONTRIBUTING FACTOR
Expected Number of PM-Related Excess Deaths Per Day
Population
of City
10 Million
5 Million
1 Million
500,000
100,000
50,000
10,000
Whole Pop.
All Causes
< 2d Lag
~4 -7
~2 -4
-0.4 -0.7
-0.2 -0.4
-0.05
-0.03
-0.005
65+ Pop.
All Causes
< 2d Lag
-5 - 10
-2.5 - 5
-0.5 - 1
-0.3 - 0.5
-0.07
-0.04
-0.01
Whole Pop.
All Causes
3-5 Day Lag
-7 - 15
-4 - 8
-0.8 - 1.5
-0.4 -0.8
-0.01
-0.05
-0.01
Whole Pop.
Respiratory
3-5 Day Lag
-2 - 3
-1 - 2
-0.2 - 0.3
-0.1 - 0.15
-0.03
-0.02
-0.002
Whole Pop.
Cardiovascular
3-5 Day Lag
-5 -9
-2 -5
-0.5-0.9
-0.2-0.5
-0.07
-0.04
-0.008
'Upper end of range for each city size calculated from upper bound estimates in Table 13-17 for population of 1
million. Lower end of range derived as lower bound estimate roughly 50% less than the upper bound, as per
text.
April 1995
1-95
DRAFT-DO NOT QUOTE OR CITE
-------
1 mixes having 50-lOOjug/m3 increments in 24-h PM10 levels would most likely be projected as
2 possibly causing meaningful numbers of excess deaths mainly as such exposures occur for
3 large segments of the elderly population (age 65 + yrs) with preexisting cardiopulmonary
4 diseases in rather large cities exceeding 1-2 million population. The level of public concern,
5 however, even in such cases may be tempered by the likelihood that a majority of those at
6 special risk are most probably current or former smokers, given the predominant role
7 attributed (U.S. EPA, 1992) to smoking in the etiology of preexisting cardiopulmonary
8 diseases that distinguish those identified as being at greater risk.
9 There is some limited direct evidence for an interaction between smoking status and
10 excess mortality attributable to PM exposure. Based on the Six Cities Study, Dockery, et.
11 al. (1993) reported an increased RR for PM 2 5 between the least polluted city (Portage) and
12 the most polluted city (Steubenville) that is substantially (albeit not statistically significantly)
13 higher in individuals who are current or former smokers, compared to never-smokers. This
14 is shown in Table 1-16. Prospective studies that have individual data on smoking status
15 probably offer the best opportunity for detecting differential effects of smoking status on PM-
16 related mortality and morbidity for use in future criteria assessments.
17
18
TABLE 1-16. ASSOCIATION BETWEEN CIGARETTE SMOKING STATUS AND
EXCESS MORTALITY RISK FROM AIR POLLUTION IN THE SIX CITIES STUDY
Relative Risk for Worst PM2 5 City (Steubenville)
Versus Lowest Best PM2 5 City (Portage)
Smoking Status
Never Smoker
Former Smoker
Current Smoker
M+F
1.19
1.35
1.32
(95% CI)
(0.90,1.57)
(1.02,1.77)
(1.04,1.68)
M
1.29
1.31
1.42
(95% CI)
(0.80,2.09)
(0.96,1.80)
(1.05,1.92)
F
1.15
1.48
1.23
(95% CI)
(0.82,1.62)
(0.82,2.66)
(0.83,1.83)
on Table 3 from Dockery, et. al (1993)
1 The prematurity of the excess deaths is also a matter of considerable importance, but
2 there is as yet little firm evidence from acute mortality studies by which to judge whether
3 PM-related excess deaths generally represent highly compromised elderly individuals dying a
April 1995 1-96 DRAFT-DO NOT QUOTE OR CITE
-------
1 few days or weeks sooner than they would have otherwise versus several months or years of
2 prematurity of death for some.
3 Morbidity effects demonstrated as likely being associated with short-term exposures to
4 ambient U.S. PM exposures include increased hospital admissions for respiratory and
5 cardiovascular disease conditions, increased respiratory symptoms (including exacerbation of
6 asthma), and small pulmonary function decrements (e.g., 2-3% decreases in FEVj or FVC).
7 Probably of most immediate public health concern are the hospital admissions, which are also
8 more readily quantifiable and understandable as an index of obviously serious health impacts.
9 Table 13-20 summarizes key types of information by which one might attempt to project
10 increments in hospital admissions for which exposure to an increment in ambient PM10
11 (24-h) of 50 /ig/m3 might contribute per 1 million people exposed. Table 1-17 can be
12 interpreted in an analogous fashion to Table 1-14. Note from Table 1-17 that the typical
13 number of PM10-related hospital admissions for cardiovascular causes would be projected to
14 be only about 2.5 times as high as the number of potential deaths during the same event, and
15 the number of respiratory admissions about 6 times as high as the possible number of deaths
16 from respiratory causes shown in Table 1-14. However, many deaths from cardiovascular or
17 respiratory causes occur without a prior hospital admission. There is, nonetheless, a
18 reasonable numeric consistency between the rough estimates of potential hospital admissions
19 or discharges and possible total deaths contributed to by exposure in a community to PM-
20 containing ambient-air mixes.
21 Table 1-18 then scales expected daily hospital admissions potentially associated with
22 exposures to ambient air mixes having a 50 ^g/m3 increase in PM10 (24-h) for towns and
23 cities with populations of 10 thousand to 10 million (analogous to what was done earlier in
24 Table 1-15 based on Table 1-14 calculations). However, in this case, both Tables 1-17 and
25 1-18 provide only upper bound estimates for hospital admissions based on available analyses,
26 which did not include copollutants in the models. Essentially the same types of statements
27 as made with regard to the very small increases in excess risk depicted in Tables 1-17 and
28 1-18 for mortality also generally apply here for hospital admissions, except to note somewhat
29 larger projected numbers for possible hospital admission cases for which the ambient PM
30 exposure might be a contributing factor.
April 1995 1-97 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-17. ESTIMATED HOSPITAL ADMISSIONS PER DAY IN A POPULATION OF
ONE MILLION FOR WHICH AN INCREASE OF 50 /tg/m3 (24-h) PM10 COULD BE A
CONTRIBUTING FACTOR
First
Listed
Diagnosis
All conditions
Respiratory
Conditions (all)
Pneumonia
COPD
Heart
Disease
Age
Group
All
65 +
All
65 +
All
65 +
All
65 +
All
65 +
Population
Baseline
Annual
Discharges
124, HO1
42,8452
12, ISO1
54011
4,3401
2,3352
3,3377
2,5607
12,310
13,502
Population
Baseline
Daily Hospital
Discharges
340.0
117.4
33.4
14.0
11.9
6.4
9.1
7.0
58.4
37.0
Excess Risk
per PMJj?
50 /ig/m3
(Lag <. 1 d)
—
0.063
0.083
0.084
0.155
0.163
0.046
0.066
Possible Number
of PM-Related
Hosp. Admissions
Per Day
—
2.0
1.1
0.5
1.4
1.1
2.3
2.2
Table 12-7
3From Table 12-9, average
5From Table 12-10, average 6From Table 12-12
7From 1992 detailed Tables; excludes asthma (ICD9 493-493.9)
2From Table 12-7, assuming 12.6% age 65 +
4From Table 12-11, average
TABLE 1-18. ESTIMATED NUMBERS OF HOSPITAL ADMISSIONS FOR RESPIRATORY
AND CARDIOVASCULAR CAUSES PER DAY IN CITIES OF 10,000 to 10 MILLION FOR
WHICH AN INCREASE OF 50 jtg/m3 PM10 (24-h) COULD BE A CONTRIBUTING FACTOR
Population
of City
10 Million
5 Million
1 Million
500,000
100,000
50,000
10,000
All Respiratory
Conditions
Whole 65 +
Pop. Pop.
20.0
10.0
2.0
1.0
0.2
0.1
0.02
11.0
5.5
1.1
0.55
0.11
0.05
0.01
Pneumonia
Whole 65 +
Pop. Pop.
5.0
2.5
0.5
0.25
0.05
0.02
0.01
COPD
Whole 65 +
Pop. Pop.
14.0
7.0
1.4
0.7
0.14
0.07
0.02
11.0
5.5
1.1
0.55
0.11
0.05
0.01
Heart Disease
Whole 65 +
Pop. Pop.
23.0
11.5
2.3
1.15
0.23
0.12
0.02
22.0
11.0
2.2
1.1
0.22
0.11
0.02
April 1995
1-98
DRAFT-DO NOT QUOTE OR CITE
-------
1 Overall, based on the foregoing discussions, there appears to exits credible evidence for
2 a likely very small, but real PM effect on human health in some susceptible subpopulations
3 (including contributing along with other risk factors to premature deaths among the elderly
4 with preexisting cardiopulmonary diseases) at PM10 24-h concentrations in the range of 30 to
5 200 jug/m3. However, the biological mechanisms by which such effects occur are as yet not
6 well understood and remain to be delineated, as is the case for clearer characterization and
7 interpretation of relative risk estimates for PM-related effects and their appropriate use in
8 projecting potential public health impacts.
April 1995 1-99 DRAFT-DO NOT QUOTE OR CITE
-------
i 2. INTRODUCTION
2
3
4 This document is an update revision of "Air Quality Criteria for Particulate Matter and
5 Sulfur Oxides" published by the United States Environmental Protection Agency (EPA) in
6 1982, an Addendum to that document published in 1986, and an Acid Aerosols Issue Paper
7 published in 1989, and it will serve as the basis for reevaluating the current National
8 Ambient Air Quality Standard (NAAQS) for particulate matter (PM). The present document
9 critically evaluates and assesses the scientific information relative to determining the health
10 and welfare effects associated with exposure to various concentrations of PM in ambient air.
11 Although the document is not intended as a complete and detailed literature review, it is
12 intended to cover pertinent literature through its publication date. The literature through that
13 time is reviewed thoroughly for information relevant to criteria development. Though the
14 emphasis is on the presentation of data on health and welfare effects, other scientific data are
15 also discussed in order to provide a better understanding of the pollutants in the environment.
16
17
18 2.1 LEGISLATIVE REQUIREMENTS
19 Two sections of the CAA (Sections 108 and 109, 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
22 list pollutants that may reasonably be anticipated to endanger public health or welfare and to
23 issue air quality criteria for them. The air quality criteria are to reflect the latest scientific
24 information useful in indicating the kind and extent of all exposure-related effects on public
25 health and 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
27 "primary" and "secondary" NAAQS for pollutants identified under Section 108. Section
28 109(b)(l) defines a primary standard as a level of air quality, the attainment and maintenance
29 of which, in the judgment of the Administrator, based on the criteria and allowing for an
30 adequate margin of safety, is requisite to protect the public health. Section 109(d) of the
31 CAA requires periodic review and, if appropriate, revision of existing criteria and standards.
April 1995 2-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 Under Section 109(b) of the CAA, the Administrator must consider available information to
2 set secondary NAAQS that are based on the criteria and are requisite to protect the public
3 welfare from any known or anticipated adverse effects associated with the presence of such
4 pollutants. The welfare effects included in the criteria are effects on vegetation, crops, soils,
5 water, animals, manufactured materials, weather, visibility, and climate, as well as damage
6 to and deterioration of property, hazards to transportation, and effects on economic value and
7 personal comfort and well-being.
8
9
10 2.2 REGULATORY BACKGROUND
11 "Particulate matter" is the generic term for a broad class of chemically and physically
12 diverse substances that exist as discrete particles (liquid droplets or solids) over a wide range
13 of sizes. Particles originate from a variety of stationary and mobile sources. They may be
14 emitted directly or formed in the atmosphere by transformation of gaseous emissions such as
15 sulfur oxides (SOX), nitrogen oxides (NOX), and volatile organic substances. The chemical
16 and physical properties of PM vary greatly with time, region, meteorology, and source
17 category, thus complicating the assessment of health and welfare effects. Particles in
18 ambient air usually occur in two somewhat overlapping bimodal size distributions: (1) fine
19 (diameter less than 2.5 jum) and (2) coarse (diameter larger than 2.5 ji«n). The two size
20 fractions tend to have different origins and composition.
21 On April 30, 1971 (Federal Register, 1971), EPA promulgated the original primary and
22 secondary NAAQS for paniculate matter (PM) under Section 109 of the CAA. The
23 reference method for measuring attainment of these standards was the "high-volume" sampler
24 (Code of Federal Regulations, 1986), which collects PM up to a nominal size of 25 to 45 /xm
25 (so-called "total suspended paniculate," or "TSP"). Thus, TSP was the original indicator for
26 the PM standards. The primary standards for PM (measured as TSP) were 260 jwg/m3, 24-h
27 average not to be exceeded more than once per year, and 75 pig/m3, annual geometric mean.
28 The secondary standard (measured as TSP) was 150 ptg/m3, 24-h average not to be exceeded
29 more than once per year.
30 On October 2, 1979 (Federal Register, 1979a), EPA announced that it was in the
31 process of revising the AQCD and reviewing the existing air quality standards for possible
April 1995 2-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 revisions. External review drafts of that revised AQCD were made available for public
2 comment and peer review by the Clean Air Scientific Advisory Committee (CASAC) of
3 EPA's Science Advisory Board (SAB). CASAC prepared a "closure" memorandum to the
4 Administrator indicating its satisfaction with the final draft of the AQCD. After closure,
5 minor technical and editorial refinements were made to the AQCD (U.S. Environmental
6 Protection Agency, 1982). The final draft (December 1981) of the document was issued
7 simultaneously with the proposal of revisions to the PM standards.
8 On March 20, 1984 (Federal Register, 1984), EPA proposed a number of revisions to
9 the primary and secondary PM standards. Following publication of the proposal, EPA held a
10 public meeting in Washington, DC, on April 30, 1984, to receive comments on the proposed
11 standards revisions. After the close of the original public comment period (June 5, 1985),
12 CASAC met on December 16 and 17, 1985, to review the proposal and to discuss the
13 relevance of certain new scientific studies on the health effects of PM that had emerged since
14 CASAC completed its review of the AQCD and staff paper in January 1982. Based on its
15 preliminary review of these new studies, CASAC recommended that EPA prepare separate
16 addenda to the AQCD and staff paper for the purpose of evaluating relevant new studies and
17 discussing their potential implications for standard-setting. The EPA announced its
18 acceptance of these recommendations on April 1, 1986 (Federal Register, 1986a). On July
19 3, 1986, EPA announced (Federal Register, 1986b) the availability of the external review
20 draft document, entitled Second Addendum to Air Quality Criteria for Particulate Matter and
21 Sulfur Oxides (1982): Assessment of Newly Available Health Effects Information (U.S.
22 Environmental Protection Agency, 1986). At the same time, on July 3, 1986, EPA
23 announced a supplementary comment period to provide the public an opportunity to comment
24 on the implications of the new studies and addenda for the final standards. The CASAC held
25 a public meeting on October 15 and 16, 1986, to review the AQCD addendum. At this
26 meeting, CASAC members, as well as representatives of several organizations, provided
27 critical review of the EPA documents.
28 The CASAC sent a closure letter on the AQCD addendum to the Administrator dated
29 December 15, 1986, which stated that the 1986 addendum and the 1982 AQCD, previously
30 reviewed by CASAC, represent a scientifically balanced and defensible summary of the
31 extensive scientific literature on PM and SOX (Lippmann, 1986b).
April 1995 2-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 On July 1, 1987 (Federal Register, 1987), EPA published final revisions to the NAAQS
2 for PM. The principal revisions in 1987 included (1) replacing TSP as the indicator for the
3 ambient standards with a new indicator that includes only particles with an aerodynamic
4 diameter less than or equal to a nominal 10 pirn ("PM10"), (2) replacing the 24-h primary
5 TSP standard with a 24-h PM10 standard of 150 /xg/m3, (3) replacing the annual primary TSP
6 standard with an annual PM10 standard of 50 ^g/m3, and (4) replacing the secondary TSP
7 standard with 24-h and annual PM10 standards identical in all respects to the primary
8 standards.
9
10
11 2.3 SCIENTIFIC BASIS FOR THE EXISTING PARTICIPATE
12 MATTER STANDARDS1
13 The following discussion describes the bases for the existing PM NAAQS set in 1987.
14 The discussion includes the rationale for the primary standards, the pollutant indicator for
15 paniculate matter, the averaging time and form of the standard, and finally a discussion of
16 EPA's assessment that led to the standard set in 1987.
17
18 2.3.1 Rationale for the Primary Standards
19 In selecting primary standards for PM, the Administrator must specify (1) the particle
20 size fraction that is to be used as an indicator of paniculate pollution, (2) the appropriate
21 averaging times and form(s) of the standards, and (3) the numerical levels of the standards.
22 Based on the assessment of relevant scientific and technical information in the earlier 1982
23 PM AQCD and addenda, the staff paper and staff paper addendum outlined a number of key
24 factors considered in making decisions in each of these areas. The following discussion of
25 the 1987 revisions of the standards focuses mainly on the considerations that were most
26 influential in the Administrator's selection of particular options.
27
28
29 Adapted from Federal Register (1987) National Ambient Air Quality Standard for Paniculate
30 Matter.
31
April 1995 2-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 2.3.2 Pollutant Indicator
2 Based on the assessment of the available scientific information, EPA concluded in 1987
3 that (1) a separate PM standard (as opposed to a combination standard for PM and SOX)
4 remained a reasonable public health policy choice, and (2) given current scientific knowledge
5 and uncertainties, a size-specific (rather than chemical-specific) indicator should be used. In
6 assessing the information in the AQCD, EPA reached several conclusions summarized here.
7
8 (1) Health risks posed by inhaled particles are influenced by both the penetration and
9 deposition of particles in the various regions of the respiratory tract and the biological
10 responses to these deposited materials. Smaller particles penetrate furthest in the
11 respiratory tract. The largest particles are deposited predominantly in the
12 extrathoracic (head) region, with somewhat smaller particles depositing in the
13 tracheobronchial region; still smaller particles can reach the deepest portion of the
14 lung, the pulmonary region.
15
16 (2) The risks of adverse health effects associated with deposition of typical ambient fine
17 and coarse particles in the thoracic region (tracheobronchial and pulmonary deposition)
18 are markedly greater than those associated with deposition in the extrathoracic region.
19 Maximum particle penetration to the thoracic region occurs during oronasal or mouth
20 breathing.
21
22 (3) The size-specific indicator for primary standards should represent those particles small
23 enough to penetrate to the thoracic region. The risks of adverse health effects from
24 extrathoracic deposition of typical ambient PM are sufficiently low that particles
25 depositing only in that region can safely be excluded from the indicator.
26
27 Considering the above conclusions, together with information on air quality
28 composition, the need to provide protection for sensitive individuals who may breathe by
29 mouth or oronasally and the similar convention on particles penetrating the thoracic region
30 adopted by the International Standards Organization (1981), EPA staff recommended that the
31 size-specific indicator include particles of diameters less than or equal to a nominal 10 nm
32 "cut point" generally referred to as "PM10". In terms of collection efficiency, this represents
33 a 50% cut point or diameter (D50) the aerodynamic particle diameter for which the efficiency
34 of particle collection is 50%. With such a cut point, larger particles are not excluded
35 entirely but are collected with substantially decreasing efficiency, and smaller particles are
36 collected with increasing (up to 100%) efficiency. Ambient samplers with this cut point
37 provide a reliable estimate of the total mass of suspended PM of aerodynamic size less than
April 1995 2-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 or equal to 10 /mi. Such an indicator (PM10) is conservative with respect to health
2 protection in that it includes all of the particles small enough to penetrate to the sensitive
3 pulmonary region and includes approximately the same proportion of the coarse-mode
4 fraction that would be expected to reach the tracheobronchial region. It places substantially
5 greater emphasis on controlling smaller particles than does a TSP indicator, but does not
6 completely exclude larger particles from all control.
7 The assessment of then-available information on respiratory tract deposition in the 1986
8 AQCD and staff paper addenda reinforced the conclusions reached in the original EPA
9 assessment. In particular, (1) the data do not provide support for an indicator that excludes
10 all particles larger than 10 /zm in diameter; (2) the analysis used to support an alternative
11 indicator with a nominal size cut point of 6 /mi (Swift and Proctor, 1982) significantly
12 underestimated thoracic deposition of particles larger than 6 /mi in diameter under natural
13 breathing conditions; (3) the PM10 indicator generally includes a similar or larger fraction of
14 the range of particles that can deposit in the tracheobronchial region, although it appears to
15 be somewhat less conservative in this regard than previously thought with respect to large
16 (> 10 /mi) particle deposition under conditions of natural mouthbreathing; and (4) the studies
17 of tracheobronchial deposition generally involved adult subjects (other information indicating
18 even greater tracheobronchial deposition of particles in children than in adults provides an
19 additional reason for an indicator that includes particles capable of penetration to the
20 tracheobronchial region). Consideration of these and the earlier conclusions led EPA to
21 reaffirm its recommendation for a PM10 indicator. The CASAC also restated its support for
22 PM10 in its review of the proposal and the closure letter to the Administrator (Lippmann,
23 1986a,c).
24 In 1987 the Administrator accepted the recommendations of the staff and CASAC, as
25 well as their underlying rationale, and decided to replace TSP as the particle indicator for the
26 primary standards with a new indicator that includes only those particles less than a nominal
27 10 /xm in diameter (PM10) as specified in the Federal Reference Method.
28
29 2.3.3 Averaging Time and Form of the Standards
30 The EPA's assessment at that time of scientific information available prior to 1987
31 confirmed the need for both short- and long-term standards for PM. The alternative of a
April 1995 2-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 single averaging time would not provide adequate protection against potential effects from
2 both long- and short-term exposures without being unduly restrictive. The forms for the
3 24-h and annual standards are discussed below.
4
5 24-Hour Standard
6 The Environmental Protection Agency proposed in 1987 that the 24-h standard be stated
7 in a statistical form that uses more than 1 year of data and accounts for variations in
8 sampling frequency in order to predict the actual number of exceedances to be expected in an
9 average year. When used with an appropriate standard level, the statistical form can provide
10 improved health protection that is less sensitive to changes in sampling frequency than the
11 deterministic form and also can offer a more stable target for control programs. Recognition
12 of the limitations of the deterministic form also led EPA to promulgate a statistical form for
13 the ozone standard (Federal Register, 1979b).
14
15 Annual Standard
16 The Administrator decided to change the form of the annual standard in 1987 from the
17 previous annual geometric mean form to a statistical form expressed as an expected annual
18 arithmetic mean. The expected annual arithmetic mean is equivalent to the long-term
19 arithmetic average concentration level, assuming no changes in underlying emissions. The
20 expected arithmetic mean is more directly related to the available health effects information
21 than is the annual geometric mean, which was the previous form of the standard. Because
22 the arithmetic mean concentration is proportional to the sum of the daily means, it reflects
23 the total cumulative exposure of PM to which an individual is exposed. Therefore, it is an
24 appropriate indicator to protect against any health effect that depends on chronic, total
25 exposure. It is also a reasonable indicator for protecting against health effects that depend on
26 repeated short-term high concentrations (short-term peaks have an influence on the arithmetic
27 mean that is proportional to their frequency, magnitude, and duration). The geometric mean,
28 on the other hand, deemphasizes the effect of short-term peak concentrations and is heavily
29 influenced by days of relatively clean air. For these reasons, EPA staff and CAS AC
30 recommended the change to an arithmetic mean.
April 1995 2-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 Under the statistical form, the expected annual arithmetic average is determined by
2 averaging the annual arithmetic averages from 3 successive years of data. The prior
3 deterministic form of the standard did not adequately take into account the random nature of
4 meteorological variations. In general, annual mean PM concentrations will vary from year to
5 year, even if emissions remain constant, due to the random nature of meteorological
6 conditions that affect the formation and dispersion of particles in the atmosphere. If only 1
7 year of data is considered, compliance with the standard and, consequently, emission control
8 requirements, may be determined on the basis of a year with unusually adverse or unusually
9 favorable weather conditions. The problem of year-to-year variability is, however, reduced
10 by averaging 3 years of data.
11
12 2.3.4 Level of the Standards
13 The original Office of Air Quality Planning and Standards (OAQPS) PM Staff Paper
14 and CASAC recommendations set forth a framework for determining the levels for the
15 proposed PM standards that would protect public health with an adequate margin of safety.
16 The discussion that follows relies heavily on that framework and on the supporting material
17 in the staff paper and its addendum, as well as the CASAC closure letters. The essential
18 steps in this framework are summarized here.
19
20 Assessment of the Quantitative Epidemiological Studies
21 The 1982 AQCD and its 1986 addendum identified a small number of community
22 epidemiological studies that are useful in determining concentrations at which PM is likely to
23 adversely impact public health. The EPA staff used these quantitative studies to examine
24 concentration-response relationships and to develop numerical "ranges of interest" for
25 possible PM10 standards.
26 A number of uncertainties associated with the use of these studies had to be considered
27 in selecting an appropriate margin of safety. As discussed in the staff paper and the AQCD,
28 and the addenda to those documents, epidemiological studies generally are limited in
29 sensitivity and are subject to inherent difficulties involving confounding variables.
30 Moreover, many of the quantitative studies were conducted in times and places where
31 pollutant composition may have varied considerably from current U.S. atmospheres. Also,
April 1995 2-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 most of the studies used British Smoke—British Smoke (BS) is a pseudo-mass indicator
2 related to small particle (aerodynamic diameter less than a nominal 4.5 /*m) darkness—or
3 TSP as particle indicators. None of the published studies used the proposed PM10 indicator.
4 Thus, assumptions had to be used to convert the various results to common (PM10) units.
5
6 Identification of Margin of Safety Considerations
7 The 1982 AQCD and its addendum identified an additional substantial body of scientific
8 literature that, although it did not provide reliable concentration-response relationships for
9 ambient exposures, did provide important qualitative insights into the health risks associated
10 with human exposure to particles. This literature included both quantitative and qualitative
11 epidemiological studies, controlled human exposure experiments, and animal toxicological
12 studies. The EPA staff assessed this literature to identify additional factors and uncertainties
13 that should be considered in selecting the most appropriate margin of safety.
14 Experience has shown that it is difficult to identify, with confidence, the lowest
15 pollution level at which an adverse effect will occur. Moreover, in cases such as the
16 present one, the evidence suggests that there is a continuum of effects, with the risk,
17 incidence, or severity of harm decreasing, but not necessarily vanishing, as the level of
18 pollution is decreased.
19 The requirement for an adequate margin of safety for primary standards addresses
20 uncertainties associated with inconclusive scientific and technical information available at the
21 time of standard setting. It also aims to provide a reasonable degree of protection against
22 hazards that research has not yet identified. Both kinds of uncertainties are components of
23 the risk associated with pollution at levels below those at which human health effects can be
24 said to occur with reasonable scientific certainty. Thus, by selecting primary standards that
25 provide an adequate margin of safety, the Administrator sought not only to prevent pollution
26 levels that have been demonstrated to be harmful, but also to prevent lower pollutant levels
27 that may pose an unacceptable risk of harm, even if that risk is not precisely identified as to
28 nature or degree.
29 In the absence of clearly identified thresholds for health effects, the selection of a
30 standard that provides an adequate margin of safety requires an exercise of informed
31 judgment by the Administrator. The level selected will depend on the expected incidence
April 1995 2-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 and severity of the potential effects and on the size of the population at risk, as well as on
2 the degree of scientific certainty that the effects will in fact occur at any given level of
3 pollution.
4 The 1986 EPA staff paper recommended a range of potential standards for the
5 Administrator's consideration. The recommended range was below the levels at which EPA
6 staff, with the concurrence of CAS AC, had concluded from the available data that adverse
7 health effects were "likely", but in the domain where the data suggested that such effects
8 were "possible". The Administrator proposed refined ranges of standard levels that were
9 based on the 1984 staff and CAS AC recommendations. After consideration of the then new
10 scientific evidence contained in the AQCD addendum, the staff revised its recommended
11 range of standards. The Administrator considered the revised assessments and the
12 recommendations of CASAC (Lippmann, 1986b) in making the final decision on the standard
13 levels. The rationales for the levels of the 24-h and annual standards are presented below.
14
15 24-Hour Standard
16 The 1987 assessment of the short-term epidemiological data expresses PM levels in
17 both the BS or TSP and PMIO units. The term "effects likely" denoted concentration ranges
18 derived from the 1982 AQCD and its addendum at or above which a consensus judgment
19 suggests the greatest certainty that the effects studied would occur, at least under the
20 conditions that occurred in the original studies. In the "effects possible" range, EPA found
21 credible scientific evidence suggesting the existence of adverse health effects in sensitive
22 populations, but substantial uncertainty exists regarding the conclusions to be drawn from
23 such evidence.
24 The 1987 review of the data did not provide evidence of clear thresholds in exposed
25 populations. Instead, they suggested a continuum of response for a given number of exposed
26 individuals, with both the likelihood (risk) of any effects occurring and the extent (incidence
27 and severity) of any potential effect decreasing with concentration (this is particularly true for
28 the statistical analyses of daily mortality in London). Substantial agreement existed that
29 wintertime pollution episodes produced premature mortality in elderly and ill populations, but
30 the range and nature of association provide no clear basis for determining lowest
31 effects-likely levels or for defining a concentration below which no association remained.
April 1995 2-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 The lung function studies in children also provided evidence of effects at concentrations over
2 a range, but the relationships were not certain enough to derive effects-likely levels for
3 PM10. The lung function studies did, however, suggest levels below which detectable
4 functional changes were unlikely to occur in exposed populations. Following CAS AC
5 recommendations, EPA used the combined range of effects-possible studies as a starting
6 point for developing alternative standards.
7 The original range proposed by the Administrator, drawn from the 1984 staff analysis,
8 was 150 to 250 /xg/m3 PM10 24-h average, with no more than one expected exceedance per
9 year. The lower bound of this range was derived from the original assessment of the London
10 mortality studies. As a result of reanalyses of the London mortality data and the findings
11 from the then current U.S. morbidity studies, the staff reduced the level of the lower bound
12 of the range of interest to 140 /ig/m3, and noted that the difference between it and the
13 original lower bound (150 /tig/m3) was within the range of uncertainty associated with
14 converting the morbidity study results from TSP to PM10.
15 At that tune the study of Lawther et al. (1970) was judged to provide evidence that
16 health effects are likely at PM concentrations above 250 Mg/m3 (as BS). The effects
17 observed in this study (related to aggravation of bronchitis) are of concern because of both
18 their immediate impact and their potential for inducing longer term deterioration of health
19 status in a significant sensitive group. Based on the uncertain conversion between smoke and
20 PM10, the lowest effects likely level derived from the Lawther study (250 jig/m3 as BS)
21 should be in the range of 250 to 350 /ig/m3 in PM10 units.
22 The 1987 assessment of the Lawther et al. (1970) study formed the basis for the upper
23 bound of the range of PM10 standards proposed by the Administrator in 1984. Considering
24 this study alone, a PM10 standard of 250 /xg/m3 might appear to contain some margin of
25 safety, even for the sensitive bronchitics studied, because it incorporated a conservative PM10
26 conversion factor and because of differences between exposure conditions in the British study
27 and current U.S. air quality. Because bronchitics are identified as a group particularly
28 sensitive to particulate pollution, a standard of 250 /xg/m3 (as PM^) also might provide some
29 margin of safety for other, less sensitive groups. Nevertheless, this study of bronchitics in
30 London has inherent limitations in sensitivity that preclude derivation of unequivocal "effects
31 thresholds" at 250 ^ig/m3 as BS and, by extension, PM10. The 1982 AQCD noted that
April 1995 2-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 associations between pollution and health status persisted at lower BS concentrations in
2 selected, more sensitive individuals. Although the lead author of the study objected to
3 attaching any importance to these latter findings (Lawther, 1986), EPA, with CASAC
4 concurrence, found no basis for asserting that this study demonstrated a population threshold
5 at 250 /zg/m3.
6 In evaluating the margin of safety for a 24-h standard, it was also important to consider
7 the London mortality studies. A standard at the upper portion of the proposed range (250
8 /^g/m3) would be well below the levels (500 to 1,000 /zg/m3 as BS) of the historical London
9 episodes in which the scientific consensus indicated that pollution was responsible for excess
10 mortality. The portions of the population at greatest risk of premature mortality associated
11 with PM exposures in those episodes included the elderly and persons with preexisting
12 respiratory or cardiac disease. Although the extent of life shortening could not be specified,
13 the seriousness of the effect strongly justified a margin of safety for it (below the consensus
14 effects levels) that was larger than that warranted for the effects on bronchitis.
15 The staff assessment at that time of several reanalyses of London mortality suggested,
16 however, that the risk of premature mortality to sensitive individuals extended to
17 concentrations substantially lower than those that occurred in the "episodes". Other analyses
18 (Mazumdar et al., 1982; Ostro, 1984; Shumway et al., 1983) provided no objective support
19 for a population threshold below which such a risk no longer existed. Although the risk to
20 individuals may be small at concentrations of 250 /xg/m3 and below, the number of people
21 exposed to lower concentrations, given U.S. levels, was substantially larger than the number
22 exposed to higher levels. The increased number of individuals exposed increased the risk
23 that effects would occur in the total population exposed.
24 Differences in the composition of particles and gases among U.S. cities and between
25 conditions in the United States and London at the time the mortality and morbidity data were
26 gathered added to the complexity of assessing the risk associated with PM in the United
27 States. In the case of the mortality studies, however, the staff found that at least one of the
28 studies (Ozkaynak and Spengler, 1985) provided qualitative support for an association
29 between daily mortality and particle concentrations in then nearly contemporary U.S.
30 atmospheres.
April 1995 2-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 The 1982 assessment of the mortality studies and related factors prompted the EPA
2 Administrator to consider standard levels that extended from 250 ^g/m3 to the lower bound
3 of the original staff range (150 Mg/m3) and even lower. Reanalyses of the London mortality
4 data prior to 1987 provided additional evidence that serious adverse health effects may occur
5 at PM concentrations below 250 /^g/m3. These analyses addressed a number of the
6 uncertainties associated with the earlier studies and reinforced the Administrator's concern
7 that a 24-h standard at the upper end of the proposed range may not provide an adequate
8 margin of safety. However, given the uncertainties in converting from BS to PM10
9 measurements, particularly at lower concentrations, and the possible differences in particle
10 composition between London at the time the data were gathered and the contemporary United
11 States, it was difficult to use these studies to set a precise level for a PM10 standard.
12 Given these difficulties, it was important to examine studies contemporary with the
13 other studies that utilize gravimetric measurements of particulate concentrations. The staff
14 found the studies of Dockery et al. (1982) and Dassen et al. (1986) to be useful. The
15 Dockery study observed physiologically small but statistically significant decreases in lung
16 function in a group of children exposed to peak PM10 levels of 140 to 250 ^g/m3. The
17 decrements persisted for 2 to 3 weeks following the exposures. The study also suggested the
18 possibility of larger responses in a subset of the children, including those with existing
19 respiratory symptoms. The Dassen study recorded similar decrements in children in the
20 Netherlands following exposure to PM10 levels estimated at 200 to 250 /ig/m3, but no
21 observable effects 2 days after exposure to PM10 levels estimated at 125 /xg/m3. The particle
22 composition, at least in the Dockery study, was more representative of contemporary U.S.
23 cities in that time period, and the associated aerometry provided a more reliable estimate of
24 PMio levels than did the measurements used in the London studies. It was reasonable to
25 expect the endpoints observed (small reversible reductions in lung function in children) to be,
26 in most cases, more sensitive to air pollution than those observed in the London studies.
27 These effects are, of themselves, of uncertain significance to health, but might be associated
28 with aggravation of respiratory symptoms in children with preexisting illness. Long-term
29 examination of respiratory health in the same community studied by Dockery et al. (1982)
30 suggested that the children in that community had a higher incidence of respiratory illness
31 and symptoms than children in communities with lower particle levels, but the data showed
April 1995 2-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 no evidence for any persistent reduction in lung function (Ware et al., 1986). Uncertainties
2 with respect to the effects of other pollutants (e.g., sulfur dioxide), the consistency of the
3 changes, and exposures precluded specifying unequivocal "effects likely" levels based on this
4 study. The EPA assessment therefore suggested that short-term lung function effects in
5 children were possible across a range of 140 to 250 /^g/rn3 or more as PM10.
6 In making a decision on a final standard level, the Administrator also considered
7 information from the more qualitative studies of PM. These studies suggested increased risks
8 for sensitive groups (asthmatics) and risks of potential effects (morbidity in adults) not
9 demonstrated in the more quantitative epidemiological literature. The qualitative studies did
10 not provide clear information on effect levels but did justify consideration of effects of PM
11 that have not been sufficiently investigated.
12 Based on the 1982 assessment of the available scientific data, in 1984, the
13 Administrator expressed an inclination to select a 24-h level from the lower portion of the
14 proposed range of 150 to 250 /tg/m3. The addendum to the 1982 assessment supported the
15 original findings and, if anything, suggested an even wider margin of safety was warranted.
16 The Administrator, therefore, decided to set the final standard at the extreme lower bound of
17 the range originally proposed (i.e., 150 Mg/m3)- This standard provided a substantial margin
18 of safety below the levels at which there was a scientific consensus that PM caused
19 premature mortality and aggravation of bronchitis. Such a margin was necessary because of
20 the seriousness of the effects and because of the analyses of daily mortality studies that
21 suggested that adverse effects may occur at PM levels well below the consensus levels. The
22 standard was in the lower portion of the range where sensitive, reversible physiological
23 responses of uncertain health significance possibly, but not definitely, are observed in
24 children. Using a conservative assessment of the lung function/particle relationship from
25 Dockery et al. (1982), a change in concentration from background levels ( = 20 /xg/m3) to
26 150 /xg/m3 would produce lung function changes of at most 10 to 15% in less than 5% of
27 exposed children. Based on the results of Dassen et al. (1986), it appeared unlikely that any
28 functional changes would be detected 1 or 2 days following such exposures. Thus, the
29 maximum likely changes in lung function appeared to present little risk of significant adverse
30 responses. Standards set at a somewhat higher level would, however, present an
31 unacceptable risk of premature mortality and allow the possibility of more significant
April 1995 2-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 functional changes. Furthermore, a standard level of 150 /xg/m3 was fully consistent with the
2 recommendations of CAS AC on the 24-h standard (Lippmann, 1986c).
3
4 Annual Standard
5 The long-term epidemiological studies examined in 1987 were subject to confounding
6 variables that reduce the studies' sensitivity and make their interpretation difficult. No clear
7 thresholds could be identified for the effects-likely levels, and evidence existed for effects at
8 lower levels (the effects-possible levels); however, the evidence was inconclusive, and the
9 effects were difficult to detect.
10 Based on an EPA assessment of PM10/TSP ratios in areas with elevated TSP levels, the
11 effects-likely levels from the Ferris et al. (1973) study were revised to'80 to 90 pig/m3 as
12 PM10. Because of limitations in sampling duration and the conversion to PM10, this estimate
13 was particularly uncertain, with effects possible at lower concentrations. Of greatest concern
14 was the possibility of long-term deterioration of the respiratory system in exposed
15 populations, the potential for which is indicated by lung function (mechanical pulmonary)
16 changes and increased incidence of respiratory disease. One set of studies (Ferris et al.,
17 1973, 1976) provided some evidence for a "no-observed-effect level" for those effects at or
18 below 60 to 65 /xg/m3 as PM10 (130 /ng/m3 as TSP), whereas another study (Bouhuys et al.,
19 1978) suggested some possibility of symptomatic responses in adults at long-term median
20 levels at or below about 50 to 55 /xg/m3 as PM10. The importance of these symptomatic
21 responses, which were unaccompanied by lung function changes, to long-term respiratory
22 health was unclear.
23 The most important study of long-term effects at that time was an ongoing examination
24 of six U.S. cities (Ware et al., 1986). The study indicated the possibility of increased
25 respiratory symptoms and illnesses in children at multiyear levels across a range of 40 to
26 more than 58 /^g/m3 as PM10 but found no evidence of reduced lung function at these
27 concentrations. This study did not find similar gradients in symptoms and illness within
28 some of the cities, which had somewhat smaller localized pollution gradients. The results of
29 a separate series of studies of long- and intermediate-term (2- to 6-week) exposures in a
30 number of U.S. metropolitan areas (Ostro, 1987; Hausman et al., 1984) were more
31 supportive of the possibility of effects within cities (respiratory-related activity restrictions in
April 1995 2-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 adults) at comparable U.S. exposure levels. The results of these studies were generally
2 consistent with the earlier U.S. studies. In particular, the finding of symptomatic responses
3 in children with no change in lung function (Ware et al., 1986) was consistent with similar
4 findings in adults (Bouhuys et al., 1978) at estimated long-term PM10 levels down to 50
5 Mg/m3. However, the information available to support the existence of significant adverse
6 effects at annual PM10 levels below 50 /zg/m3 (especially when 24-h levels are maintained
7 below 150 jig/m3) was quite limited and uncertain.
8 Because of the uncertainties and the limited scope and number of long-term quantitative
9 studies available for review in 1987, it was important to examine the results of qualitative
10 data from a number of epidemiological, animal, and ambient particle composition studies in
11 determining what constitutes an adequate margin of safety for an annual standard. These
12 studies justified concern for serious effects not directly evaluated in the above studies. Such
13 effects included damage to lung tissues contributing to chronic respiratory disease, cancer,
14 and premature mortality. Substantial segments of the population may be susceptible to one
15 or more of these effects. Although the qualitative data did not provide evidence for major
16 risks of these effects at the annual PM levels in most U.S. cities at that time, the
17 Administrator believed, that the seriousness of the potential effects and the large population at
18 risk warranted caution in setting the standard.
19 Based on the findings discussed in the 1982 AQCD, in 1984, the Administrator
20 proposed to select an annual standard level from a range of 50 to 65 jig/m3. In the proposal,
21 the Administrator favored a standard in the lower portion of the range. The evidence
22 discussed in the 1986 addendum, although subject to substantial uncertainty, served to
23 reinforce this inclination. In light of the 1986 assessment, and in accordance with the
24 recommendation of CAS AC, the Administrator decided to set the level of the annual standard
25 at the lower bound of the original range (50 /ng/m3, expected annual arithmetic mean). This
26 standard provided a reasonable margin of safety against long-term degradation in lung
27 function, which was judged likely to occur at estimated PM10 levels above 80 to 90 pig/m3
28 and for which there was some evidence at PM10 levels above 60 to 65 /*g/m3. Such a
29 standard also provided reasonable protection against the less serious symptomatic effects
30 (bronchitis) for which only inconclusive evidence was available. Moreover, the staff and
31 CASAC recommended that the combined protection afforded by both 24-h and annual
April 1995 2-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 standards be considered in selecting the final standard level. In this regard, analyses of air
2 quality data showed that implementation of the 24-h standard would reduce substantially the
3 annual levels in a number of areas to below 50 /ig/m3, adding to the protection afforded by
4 the annual standard in areas with higher 24-h peak-to-mean ratios. Based on the then
5 available information with respect to risks associated with annual exposures, the
6 Administrator believed that the annual and 24-h standards provided an adequate margin of
7 safety.
8
9 2.3.5 Welfare Effects
10 No convincing evidence existed indicating significant adverse soiling and nuisance at
11 TSP levels below 90 to 100 ptg/m3, and, on that basis, the Administrator concluded that
12 secondary standards different from the primary standards were not requisite to protect the
13 public welfare against soiling and nuisance. This conclusion was supported by CASAC's
14 determination that there was no scientific support for a TSP-based secondary standard
15 (Lippmann, 1986c). Therefore, the Administrator decided to set 24-h and annual secondary
16 PM10 standards that are equal in all respects to the primary standards.
17 The other welfare effects of principal interest were impairment of visibility, potential
18 modification of climate, and contribution to acidic deposition. All three of these effects were
19 believed to be related to regional-scale levels of fine particles, and control programs designed
20 to ameliorate them would likely involve region-wide reductions in emissions of sulfur oxides.
21 The Administrator also concurred with the staff suggestions that a separate secondary particle
22 standard was not needed to protect vegetation or to prevent adverse effects on personal
23 comfort and well-being.
24
25
26 2.4 TOPICS/ISSUES OF CONCERN FOR CURRENT CRITERIA
27 DEVELOPMENT
28 Based on the available scientific evidence, several critical topics and associated issues
29 are addressed in this document, as part of the current CAA-mandated periodic review of
30 criteria and NAAQS for PM. Some of the most critical topics and issues addressed are as
31 follows.
April 1995 2-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 2.4.1 Air Quality and Exposure
2 Physics and Chemistry of Atmospheric Aerosols
3 The atmospheric aerosols of interest because of their potential health and welfare effects
4 consist of two principal components: a gas phase ("air" in this case) and a solid or liquid
5 particle phase in suspension. Fine particles in the atmosphere consist mainly of (1) sulfate,
6 nitrate, ammonium ions, and water; (2) photochemically formed organic aerosols; and (3)
7 carbon, organic matter, and metallic components emitted directly into the atmosphere.
8 Coarse particles in the atmosphere are composed mainly of silica, calcium carbonate, clay
9 minerals, soot, and, sometimes, organic substances. A general relationship exists between
10 chemical composition and particle diameter, with particles of <2.5 /mi in diameter
11 containing most of the SO42", H+, and NH4+, as well as a significant fraction of the NO3"
12 and Cl~. The particle volume (mass) frequency function is often multimodal. The fine-
13 volume fraction may have two or more modes below 1.0. The coarse fraction generally has
14 one mode within the range ~ 5 to 50 /xm. The particle volume frequency functions for the
15 fine and coarse fractions often behave independently, (i.e., vary in relative proportion of the
16 total ambient particle mix from location to location or from one time or season to another at
17 the same location).
18 Previous documentation has shown that hydroxy, hydroperoxy, and alkoxy radicals are
19 probably important in the oxidation of SO2 to SO3", although the rate constants for some of
20 these reactions are not well established. The hydroxy radical dominates the gas-phase
21 oxidation of SO2 in the clean troposphere, and H2O2 is effective in the formation of SO42~ in
22 particles, mists, fogs, and rain. Transition metals and soot have been shown to be effective
23 catalysts for atmospheric oxidation of SO2. Oxidation rates for NO and NO3~ are known but
24 have been considered too low to be important. The oxidation rate for NO2~ is known, but
25 the tropospheric concentration of HNO2 is probably too low for this reaction to be important.
26 Except for reactions of carbon (soot), solid surface reactions do not appear to be effective
27 pathways for H2SO4 formation in the troposphere.
28 The physical properties of particles are physical configuration, bulk material properties,
29 and surface properties. The bulk material properties that affect aerosol behavior include
30 chemical composition, vapor pressure, hygroscopicity and deliquescence, and index of
31 refraction. These properties control the physical state and growth of particles and result in
April 1995 2-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 scattering and absorption of light by tropospheric particles. Hygroscopicity, deliquescence,
2 and efflorescence are critical properties in the growth of particles, but there is a paucity of
3 thermodynamic data to permit prediction of deliquescence and hygroscopic behavior and
4 vapor pressures of multicomponent systems, especially for relative humidities below about
5 90%. Few studies of desorption under atmospheric conditions have been reported. Of
6 greater concern, desorption may prove to be important in biological systems. Shape,
7 structure, and density are physical configuration properties that are important parameters in
8 the equations of motion for particles. Because of irregularities in particle geometry or
9 because the particles are agglomerates, the three configuration properties are usually defined
10 in terms of an aerodynamic diameter. Surface properties of importance include electrostatic
11 charging, adhesion, and the influence of surface films.
12 The physical properties of particles and their modal distributions are important
13 considerations (1) in the sampling and analysis of atmospheric particles and (2) in predicting
14 or determining the flux to biological and nonbiological materials and deposition in the human
15 and experimental animal respiratory tracts.
16 Advances in understanding the properties and behavior of atmospheric paniculate matter
17 have been made since publication of the previous criteria document (U.S. Environmental
18 Protection Agency, 1982). In the current revision of the document, newer literature and data
19 on the above topics are reviewed and discussed. For example, chemical pathways and rates
20 of atmospheric particle formation and of removal from the atmosphere, by dry deposition and
21 by precipitation scavenging, are examined. Likewise, the physical processes of nucleation,
22 condensation, and coagulation by which condensible material is converted into particles are
23 discussed, along with the size distribution of the resulting particles. The physical properties
24 relevant to sampling considerations and deposition on surfaces, including those of the
25 respiratory tract, are also discussed, including coverage of several newer areas of expanded
26 research: aerosol equilibria, the unique properties of semi-volatile aerosols, and the role of
27 particle-bound water.
28
29 Measurement Methodology
30 Techniques available for measurement of mass and specific components of aerosols are
31 examined. Special attention is given to the suitability of current technology for the
April 1995 2-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 measurement of aerosol mass with sufficient accuracy and precision to determine compliance
2 with one or another possible type of a new PM standard (i.e., a PM10 standard with a lower
3 level or a fine-particle standard). The need for continuous or at least daily PM
4 measurements, the difficulty of removing particle-bound water without losing NH4NO3 or
5 semivolatile organic matter, problems in defining and maintaining a precise cut at 10 ^im or
6 lower (e.g., at 2.5 pirn), and techniques for maintaining good quality control in monitoring
7 networks are also addressed.
8
9 Ambient Levels
10 The present draft of the revised PM AQCD describes ambient PM data for the United
11 States, with characterization as available by size (fine/coarse) and chemical composition.
12 Data that focus on the current U.S. PM10 standard are emphasized, but information is also
13 provided on PM2 5, PM2 5.10, and other similar cut points, as data are available. Ambient
14 patterns are discussed, to include daily, seasonal, regional, etc. Acid aerosol data are also
15 described as above as a separate aspect of PM. Key questions addressed include: What
16 information is available on the distribution of PM in regard to: geographic, seasonal,
17 diurnal, size, composition, sources, and trends? How important are uncertainties introduced
18 by variations in the position and shape of the 10-/mi cut point in various PM10 monitors?
19 How important are measurement uncertainties due to volatilizable/condensible components
20 (e.g., loss of ammonium nitrate and, possibly, other ammonium salts) and loss of
21 semivolatile organics or retention of particle-bound water? How do these uncertainties vary
22 geographically and seasonally? How do these uncertainties differ for filter collection and
23 subsequent weighing as compared to continuous indicators?
24
25 Cut Points
26 Information helpful in evaluating the possible need for a new particle standard (PM-
27 Fine) in addition to or instead of a PM10 NAAQS is presented. This information includes
28 discussion of sources, composition, lung deposition, sampling problems, epidemiology,
29 biochemistry, and toxicology of fine and coarse particles. Other considerations include
30 techniques for separating fine particles from coarse particles. Can fine and coarse particles
31 be separated adequately by a single size cut-point in all areas of the country or will the
April 1995 2-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 optimal cut point differ in time and space, especially between very dry areas where coarse
2 particles may be found below 2.5 /urn and very humid areas where fine particles may be
3 found above 1.0 /xm? If a single fine-particle cut point is chosen, which is best: 2.5, 1.0, or
4 something in between? Is size an adequate separation or will chemical composition
5 measurements also be needed?
6
7 Exposure
8 Particulate matter exposure estimates for most epidemiology studies are based on data
9 from ambient monitoring sites. Relationships between such measurements and personal
10 exposure are important in evaluating epidemiology studies. Aspects evaluated and discussed
11 in the present draft document include: urban scale PM exposure models, indoor/outdoor
12 characteristics and relationships, and the validity of ambient measurements to provide
13 appropriate estimates to relate to health effect endpoints. Two exposure estimates are of
14 concern, individual and population estimates of PM exposure. The type of epidemiology
15 study determines which estimate is appropriate. Additionally, other factors (such as exposure
16 durations) that may determine health effects are considered. Human exposure patterns to
17 ambient and indoor air particles, including consideration of activity patterns and various
18 microenvironments, are characterized.
19 Actual human exposure differs from outdoor concentrations due to infiltration of
20 ambient aerosols indoors, indoor sources, and human activity patterns. Human exposure can
21 be determined through measurements and models. For PM the indoor and personal
22 monitoring data show both higher than ambient and lower than ambient PM concentrations in
23 indoor settings as a function of varying particle size and human activity patterns.
24 Coarse-mode particles (>2.5 ^m), which are generally of nonanthropogenic origin
25 (windblown dust, etc.), require turbulence to provide vertical velocity components greater
26 than their settling velocity to allow them to remain suspended in the air. Outdoor particles
27 enter into an indoor setting either by bulk flow (e.g., through an open window) in which all
28 particles can enter at the inlet condition, or by diffusional flow (e.g., through cracks and
29 fissures in the barrier of the building envelope) in which velocities are relatively lower and
30 therefore capable of settling out the coarser particles in the passage through the barrier. Fine
April 1995 2-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles, however, are not easily removed by settling or impaction and penetrate indoors
2 with high efficiency.
3 Indoor settings are usually quiescent, and the larger ambient particles that do enter
4 indoors quickly settle out, leading to the presence of the familiar dust layers that require
5 indoor settings to be cleaned constantly. However, human activity in indoor settings does
6 generate fine particles (<2.5 /xm) from smoking, vacuuming, cooking, etc., and resuspends
7 coarse particles that previously had settled out.
8 Two major factors influencing the relationship of ambient to indoor PM air quality are
9 (1) the variability of indoor concentrations of PM compared to outdoor concentrations as a
10 function of particle size (e.g., fine indoor > fine outdoor and coarse indoor < coarse
11 outdoor) and (2) the variation of exposures of individuals related to the different activities
12 that are involved with the local generation of particles in their immediate surroundings
13 (smoking, traffic, dusting and vacuuming at home, etc.).
14 Long-term personal exposures to coarse-fraction PM (>2.5 pirn) can be less than half
15 the ambient concentrations. Long-term personal exposures to fine-fraction PM (<2.5 jum) of
16 ambient origin may be estimated by ambient measurements of the <2.5-^tm PM fraction.
17 However, personal activities and indoor concentrations cause personal exposures to PM to
18 vary substantially raising the issue of how well ambient measurements can serve as predictors
19 of human exposures, either on an individual personal level or on a community-wide level.
20
21 2.4.2 Health Effects
22 A rapidly growing body of epidemiologic data examines relationships between PM
23 concentrations and human health effects, ranging from respiratory function changes and
24 symptoms to exacerbation of respiratory disease and excess mortality associated with
25 premature death. These effects appear to lie along an increasing gradient of severity of
26 effects in different subpopulations. Although the exact biological mechanisms underlying
27 such effects are poorly understood, there seems to be an emerging pattern of findings that
28 increases the plausibility that the observed relationships may reflect a real, causal relationship
29 between paniculate matter and human health. This revised PM criteria document assesses
30 evidence suggesting that this overall pattern of effects may extend to concentrations of PM10
31 below the current NAAQS or may be associated with other PM size fractions (e.g., fine
April 1995 2-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles < 2.5 ^m). Controlled human exposure and animal toxicologic studies are also
2 evaluated, and the overall coherence/consistency of findings in relationship to the
3 epidemiologic database is assessed. These include, for example: (1) studies of respiratory
4 tract disposition and clearance of particles, (2) experimental studies (animal and human)
5 evaluating mechanisms of action of various particles (by size, chemical composition, etc.) in
6 order to evaluate biological plausibility of effects reported by epidemiology studies, and (3)
7 other experimental studies that demonstrate various toxic effects of PM in humans or in
8 animal models.
9
10 Respiratory Tract Dosimetry
11 The biological endpoint or health effect of an aerosol exposure is likely more directly
12 related to the quantitative pattern of deposition within the respiratory tract than just to the
13 external exposure concentration. The regional deposition pattern determines not only the
14 initial respiratory tract dose but also the specific pathways and rates by which the inhaled
15 material is cleared and redistributed. Thus, in order to evaluate different toxic responses to
16 inhaled particles across species and to accurately extrapolate such laboratory animal data to
17 humans, or to evaluate differences that sex, age, or disease may have on human variability,
18 the various physicochemical, anatomic, and physiologic factors described must be integrated
19 to estimate a deposited dose or perhaps a retained dose (deposition — clearance = retention).
20 Delineation of the dose to each respiratory tract region (extrathoracic, tracheobronchial, and
21 pulmonary) is desired because each region has different dominant factors controlling
22 deposition and clearance, and different defense mechanisms. A theoretical model to describe
23 particle deposition and clearance would require detailed information on all the influential
24 parameters mentioned above (e.g., respiratory rates, exact airflow patterns, complete
25 measurements of the branching structure of the respiratory tract, pulmonary region
26 mechanics) for men, women, children, and across the various species used in toxicity
27 studies. An empirical model (i.e., equations fit to experimental data) may adequately
28 describe regional deposition and require much less data to develop the model structure.
29 Within the dosimetry chapter (Chapter 10) of this document, the anatomy of the
30 respiratory tract and the physicochemical, anatomical, and physiological factors controlling
31 particle deposition, clearance, and retention are reviewed. Other factors that modify
April 1995 2-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 deposition, including sex, age, disease state, and exposure to irritants also are discussed.
2 The available human and laboratory data on deposition and clearance and their positive and
3 negative attributes for use in quantitative model development are discussed. Available
4 validated model structures to estimate deposition and clearance in humans and laboratory
5 animals are described and evaluated. The application of these models to quantitative
6 extrapolation of the human and animal toxicity data also are discussed. Consideration is
7 given to uncertainties in input parameters and the variability of model predictions when
8 evaluating the usefulness of models for quantitative dose extrapolation.
9
10 Epidemiology Studies
11 Epidemiologic analyses are expected to provide some of the most crucial information
12 useful in deriving health criteria upon which to base Agency decisions regarding possible
13 revision of the current PM standards, and such studies are accorded extensive attention in
14 this PM criteria document as evaluated in Chapter 12 and elsewhere.
15 One useful distinction is to separate short- and long-term effects. The short-term
16 effects include changes in respiratory function, symptom indicators, hospital admissions
17 associated with exacerbation of respiratory or cardiovascular disease, and excesses of daily
18 death rates in urban areas associated with concurrent 24-h PM measurements on the same or
19 preceding few days. The short-term effects studies are typically longitudinal in nature and
20 are specific to a community or metropolitan area with reasonably homogeneous PM
21 exposures. The analyses of data in short-term studies use time-series analysis methods. The
22 long-term or chronic exposure effects studies typically use annual PM concentrations and
23 annual symptom or death rates and are more likely to involve comparisons across several
24 communities rather than within a single community. Although both kinds of epidemiologic
25 analysis are useful, it is important to assess the consistency of conclusions based on different
26 kinds of studies. Coherence of effects at lower concentrations is a useful criterion for
27 assessing diverse studies with different endpoints or effects, different populations, and
28 different exposure metrics (Bates et al., 1990) and is considered as part of the evaluation of
29 the available epidemiology literature.
30
April 1995 2-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mortality Studies. Studies examining the relationship between ambient measures of PM and
2 mortality were examined during the last review process (U.S. Environmental Protection
3 Agency, 1982, 1986) and contributed to the key scientific bases underlying the current PM10
4 NAAQS. However, given the uncertainties in converting from British Smoke to PM10
5 measurements, particularly at lower concentrations, and the possible differences in paniculate
6 composition between London at the time the data were gathered and the contemporary United
7 States, it was difficult to determine a precise level for a relationship between PM10 and
8 mortality. Since that time, numerous contemporary U.S. mortality studies using either PM10
9 or TSP measurements have been published that examine short-term measurements. Also,
10 long-term PM ambient measurements and mortality have been examined in some recent
11 studies. These and other newly emerging PM-mortality studies are summarized and critically
12 evaluated.
13 Issues of greatest concern so far relate primarily to the use and interpretation of the
14 short-term mortality studies. Almost all analyses of the relationship between PM and excess
15 mortality require statistical adjustment for mortality excesses associated with other potential
16 confounding factors, including other environmental stressors such as temperature and relative
17 humidity or other pollutants (co-pollutants) associated with PM and with mortality. For
18 example, weather-related effects may be directly related to excess mortality, but may also be
19 indirectly related when weather affects PM emissions and atmospheric concentrations.
20 Statistical and conceptual approaches to estimating the direct and indirect effects of
21 confounding variables, and specification of the form of the statistical adjustment for
22 confounding factors are evaluated in interpreting the PM effects on mortality calculated from
23 each study. Comparison of studies using different exposure metrics is considered. In
24 characterizing the relationship between excess mortality and PM in different cities, evaluation
25 of differences in particle size distribution or particle composition between cities, is done as
26 the data allow.
27 Specification of "exposure-effect" relationship(s) between mortality and PM is also
28 important. A number of studies have reported no evident threshold for effects, even at
29 relatively low concentrations, but the ability to carry out meaningful threshold evaluations
30 may be greatly limited by the power of the various available studies. Estimates of the
31 relationship between PM and mortality may depend on differences in model specification.
April 1995 2-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 Even with similar model specifications (exposure-response relationship, adjustment for
2 weather, copollutants, and other factors) there may be differences in the effects of PM at a
3 given concentration, possibly related to particle size/composition and/or differences in
4 climate or demographics among different cities. An important component of the health
5 effects assessment in the criteria document is identification of susceptible subpopulations and
6 other variables such as weather, climate, or other pollutants, potentially contributing to
7 increased mortality risk.
8
9 Morbidity Studies. Decreased pulmonary function in predominantly healthy children was
10 been reported in some earlier epidemiology studies. More recent studies add to this
11 database. Earlier long-term exposure studies provided no evidence for an effect from PM
12 exposure on level of pulmonary function, whereas some recent studies report reductions in
13 pulmonary function associated with chronic exposure to paniculate pollution. An evaluation
14 of the epidemiologic database relating short-term (24-h) and long-term (annual) ambient
15 measurement of PM10 and other measures of PM to changes in pulmonary function test
16 results in children and adults is presented. The strength and consistency of epidemiologic
17 databases that relate short-term (24-h) and long-term (annual) PMjQ and other ambient PM
18 indicator measurements to changes in the rate and/or severity of respiratory symptoms and
19 disease are also critically reviewed. Studies examining exacerbation of respiratory (i.e.,
20 COPD and asthma) and cardiovascular diseases that lead to increased medical care utilization
21 (such as emergency room visits and hospital admissions) in relation to ambient PM exposure
22 are also evaluated. As appropriate, other factors and copollutants are also examined in
23 relation to findings on each of the above types of health endpoints.
24
25 Toxicology of PM Constituents
26 In addition to the chapter evaluating epidemiologic studies of PM differentiated mainly
27 in terms of various size indicators (TSP, PM10, etc.), the toxicology of various major
28 subclasses of PM constituents is also summarized and discussed in a separate chapter. That
29 toxicology chapter focuses on acid aerosols, metals, ultrafine particles, diesel particles,
30 silica, and other types of particles that make up ambient air mixes of particles in the broad
31 class designated in toto as "particulate matter". Animal inhalation toxicology and other types
April 1995 2-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 of studies reviewed are useful in improving understanding of several key overall health
2 issues, especially: (1) the influence of particle size, number, and mass on health responses;
3 (2) the differential influence of varying particle chemistry on the health effects observed;
4 (3) the array of health effects that can be caused by specific PM constituent; (4) exposure-
5 response relationships for various exposure durations (acute and chronic); (6) mechanisms of
6 toxicity; and (6) pollutant interactions. Some of the information from these studies provides
7 a background for evaluating the biological plausibility of the mortality and morbidity
8 associations reported in epidemiological studies. For example, whether chronic bronchitis
9 can be caused by sulfuric acid exposure, as hypothesized from animal studies, is a significant
10 issue. The data on the relationship between particle size, mass, and number elucidates the
11 appropriateness of various exposure indicators of potential human effects. The document
12 reviews toxicological studies examining hypotheses related to health outcome and the physical
13 and chemical characterizations of PM. For chemistry, this includes acidity, surface coatings
14 (i.e., soluble metals), and particle-bound organics. Particle size is examined (PM10, PM2 5,
15 other [fine versus ultrafine]). Aerosol concentration examines particle number and mass.
16 Evaluation of the controlled human exposure (clinical) studies database concerning PM
17 and health outcomes is presented as a subsection of the overall PM constituent toxicology
18 chapter. This includes critical review of PM effects on pulmonary function in healthy and
19 asthmatic individuals, pulmonary clearance mechanisms, airway reactivity, and immunologic
20 defense especially in relation to particle size but only to a limited extent in relation to
21 chemical composition. There remains an almost complete absence of controlled experiment
22 data on exposure of humans to particles other than acid aerosols.
23 Human clinical studies of PM constituents have been almost completely limited to
24 measuring effects on symptoms, lung function, and airway reactivity, in addition to a few
25 studies of effects on mucociliary clearance. Few have used bronchoalveolar lavage to study
26 effects on airway inflammation and host defense; nor have many, if any, examined effects of
27 acid aerosols or other particle exposures on airway inflammation in asthmatic subjects or on
28 exacerbation of effects of antigen challenge in allergic or asthmatic subjects.
29
30
April 1995 2-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sensitive Groups
2 Available data are also evaluated for insight concerning human population groups
3 potentially having increased susceptibility to PM exposure. Preexisting respiratory or
4 cardiovascular disease, in conjunction with advanced age, appear to be important factors in
5 PM mortality susceptibility. For morbidity health endpoints, children and asthmatic
6 individuals potentially may display increased sensitivity to PM exposure, and, as such, this
7 topic is discussed.
8
9 2.4.3 Welfare Effects
10 Effects on Materials
11 All manmade materials exposed to the outdoor environment undergo degradation by
12 heat, moisture, and some bacteria and fungi. For many years, air pollution has been
13 suspected of accelerating the natural degradation processes. For example, acidic pollutants
14 have been associated with accelerated degradation of paints such as water-based paint and
15 alkyd coatings containing titanium dioxide, lead minium, or ferric oxide red. Other
16 researchers have reported acidic pollution-related effects on automotive paint and steel
17 coating. Particulate matter has also been reported to produce paint soiling. Also, acid
18 aerosols and other particles containing acids also have been reported to affect building stones,
19 cement, and concrete. Acidic aerosols change the physical characteristics of some stones,
20 cement, and concrete by changing the chemical composition. Studies examining the effects
21 on materials of PM pollution (primary and secondary particles and aerosol precursor gases)
22 are reviewed and summarized; where possible, changes in material damage are correlated
23 with changes in PM concentrations.
24
25 Visibility Effects
26 Airborne PM in the form of varying amounts of sulfates, ammonium and nitrate ions,
27 elemental carbon and organic carbon compounds, water and smaller amounts of soil dust,
28 lead compounds, and other trace species reduce visibility, thereby affecting transportation
29 safety and creating a loss in aesthetic appeal. The natural background visibility range is 150
30 ± 45 kilometers for the east and 230 ± 35 kilometers for the west. When current visibility
31 data are compared to background visibility data, manmade contributions account for
April 1995 2-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 approximately one-third of the average extinction coefficient in the rural west and over 80%
2 of the average extinction coefficient in the rural east. The effects of aerosol concentration,
3 composition, and size and pollutant emission trends on visibility are evaluated. Existing PM
4 models are discussed in relation to how well such models can be used to predict changes in
5 visibility.
6
7 Climate Change
8 It has been suggested that fine particles released into the atmosphere may alter the
9 climate through a reduction in the amount of solar radiation reaching the earth's surface, thus
10 cooling the surface while heating the aerosol layer. The scattering and absorbing properties
11 of aerosols and their vertical distribution are briefly reviewed and reference made to other
12 assessments of their effects on radiative balance and how changes in radiative balance may
13 affect weather and climate. Aerosols also affect weather and climate through their role as
14 cloud condensation nuclei. The concentration, composition, size, and number of aerosols can
15 influence the structure, stability, and albedo of clouds, possible changing the location and
16 amount of rainfall and the rate of global and regional warming due to greenhouse gases.
17
18
19 2.5 DOCUMENT CONTENT AND ORGANIZATION
20 The present document includes review and critical evaluation of relevant scientific
21 literature on PM through early 1995. The material selected for review and comment in the
22 text generally comes from the more recent literature published since 1982, with emphasis on
23 studies conducted at or near PM pollutant concentrations found in ambient air. Older
24 literature cited in the previous criteria document for PM and Addendum (U.S. Environmental
25 Protection Agency, 1982, 1986) is generally not discussed. However, as appropriate, some
26 limited discussion is included of older studies judged to be significant because of their
27 potential usefulness in deriving a NAAQS. An attempt has been made to discuss key
28 literature in the text and present it in tables as well. Reports of lesser importance for the
29 purposes of this document are typically only summarized in tables.
30 Generally, main emphasis is placed on consideration of published material that has
31 undergone scientific peer review. In the interest of admitting new and important
April 1995 2-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 information, however, some material not yet published in the open literature but meeting
2 other standards of scientific reporting may be included as reviewed by CAS AC. Emphasis
3 has been placed on studies in the range of current ambient levels. On this basis, studies in
4 which the lowest concentration employed exceeded this level have been included if they
5 contain unique data, such as documentation of a previously unreported effect or of
6 mechanisms of effects, or if they were multiple-concentration studies designed to provide
7 information on concentration-response relationships. Results of studies conducted at higher
8 levels have been included because of the potential importance of these effects to public
9 health. In reviewing and summarizing the literature, an attempt is made to present
10 alternative points of view where scientific controversy exists. As warranted, considerations
11 bearing on the quality of studies are noted.
12 The present document consists of 13 chapters. The Executive Summary for the entire
13 document is contained in Chapter 1, followed by this general introduction in Chapter 2.
14 Chapters 3 through 7 provide background information on physical and chemical properties of
15 PM and related compounds; sources and emissions; atmospheric transport, transformation,
16 and fate of PM; methods for the collection and measurement of PM; and ambient air
17 concentrations and factors affecting exposure of the general population. Chapter 8 describes
18 effects on visibility, and Chapter 9 describes damage to materials attributable to PM.
19 Chapters 10 through 13 evaluate information concerning the health effects of PM. More
20 specifically, Chapter 10 discusses dosimetry of inhaled particles in the respiratory tract and
21 Chapter 11 summarizes information on the toxicology of specific types of PM constiuents,
22 including experimental toxicological studies of animals and human clinical studies.
23 Chapter 12 discusses epidemiological studies and Chapter 13 characterizes information on
24 critical health issues derived from studies reviewed in the prior chapters.
25 Neither control techniques nor control strategies for the abatement of PM are discussed
26 in this document, although some topics covered may be incidentally relevant to abatement
27 strategies. Technologies for controlling PM emissions are discussed in other documents
28 issued by EPA's Office of Air Quality Policy and Standards (OAQPS). Likewise, issues
29 germane to the scientific basis for control strategies, but not pertinent to the development of
30 criteria, are addressed in numerous other documents issued by OAQPS.
April 1995 2-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 In addition, certain issues of direct relevance to standard setting are not explicitly
2 addressed in this document, but are instead analyzed in documentation prepared by OAQPS
3 as part of its regulatory analyses materials. Such analyses include (1) discussion of what
4 constitutes an "adverse effect" and delineation of particular adverse effects that the primary
5 and secondary NAAQS are intended to protect against, (2) exposure analyses and assessment
6 of consequent risk, and (3) discussion of factors to be considered in determining an adequate
7 margin of safety. Key points and conclusions from such analyses are summarized in a Staff
8 Paper prepared by OAQPS and reviewed by CAS AC. Although scientific data contribute
9 significantly to decisions regarding the above issues, their resolution cannot be achieved
10 solely on the basis of experimentally acquired information. Final decisions on items (1) and
11 (3) are made by the Administrator, as mandated by the CAA.
12 A fourth issue directly pertinent to standard setting is identification of populations at
13 risk, which is basically a selection by EPA of the subpopulation(s) to be protected by the
14 promulgation of a given standard. This issue is addressed only partially in this document.
15 For example, information is presented on factors, such as preexisting disease, that may
16 biologically predispose individuals and subpopulations to adverse effects from exposures to
17 PM. The identification of a population at risk, however, requires information above and
18 beyond data on biological predisposition, such as information on levels of exposure, activity
19 patterns, and personal habits. Such information is included in the Staff Paper developed by
20 OAQPS and reviewed by CAS AC as a separate item from this document.
April 1995 2-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Bates, D. V.; Baker-Anderson, M.; Sizto, R. (1990) Asthma attack periodicity: a study of hospital emergency
4 visits in Vancouver. Environ. Res. 51: 51-70.
5
6 Beloin, N. J.; Haynie, F. H. (1975) Soiling of building materials. J. Air Pollut. Control Assoc. 25: 399-403.
7
8 Bouhuys, A.; Beck, G. J.; Schoenberg, J. B. (1978) Do present levels of air pollution outdoors affect respiratory
9 health? Nature (London) 276: 466-471.
10
11 Braun-Fahrlander, C.; Ackermann-Liebrich, U.; Schwartz, J.; Gnehm, H. P.; Rutishauser, M.; Wanner, H. U.
12 (1992) Air pollution and respiratory symptoms in preschool children. Am. Rev. Respir. Dis. 145: 42-47.
13
14 Brunekreef, B.; Dockery, D. W.; Speizer, F. E.; Ware, J. H.; Spengler, J. D.; Ferris, B. G. (1989) Home
15 dampness and respiratory morbidity in children. Am. Rev. Respir. Dis. 140: 1363-1367.
16
17 Brunekreef, B.; Kinney, P. L.; Ware, J. H.; Dockery, D.; Speizer, F. E.; Spengler, J. D.; Ferris, B. G., Jr.
18 (1991) Sensitive subgroups and normal variation in pulmonary function response to air pollution episodes.
19 Environ. Health Perspect. 90: 189-193.
20
21 Chestnut, L. G.; Schwartz, J.; Savitz, D. A.; Burchfiel, C. M. (1991) Pulmonary function and ambient
22 paniculate matter: epidemiological evidence from NHANES I. Arch. Environ. Health 46: 135-144.
23
24 Code of Federal Regulations. (1986) Appendix B—reference method for the determination of suspended
25 paniculate matter in the atmosphere (high-volume method). C. F. R. 40: § 50.
26
27 Dassen, W.; Brunekreef, B.; Hoek, G.; Hofschreuder, P.; Staatsen, B.; De Groot, H.; Schouten, E.; Biersteker,
28 K. (1986) Decline in children's pulmonary function during an air pollution episode. J. Air Pollut. Control
29 Assoc. 36: 1223-1227.
30
31 Dockery, D. W.; Ware, J. H.; Ferris, B. G., Jr.; Speizer, F. E.; Cook, N. R.; Herman, S. M. (1982) Change
32 in pulmonary function in children associated with air pollution episodes. J. Air Pollut. Control Assoc.
33 32: 937-942.
34
35 Dockery, D. W.; Speizer, F. E.; Stram, D. O.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr. (1989) Effects
36 of inhalable particles on respiratory health of children. Am. Rev. Respir. Dis. 139: 587-594.
37
38 Dockery, D. W.; Schwartz, J.; Spengler, J. D. (1992) Air pollution and daily mortality: associations with
39 particulates and acid aerosols. Environ. Res. 59: 362-373.
40
41 Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.;
42 Speizer, F. E. (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J.
43 Med. 329: 1753-1759.
44
45 Edney, E. O.; Cheek, S. F.; Stiles, D. C.; Corse, E. W.; Wheeler, M. L.; Spence, J. W.; Haynie, F. H.;
46 Wilson, W. E. (1988) Effects of acid deposition on paints and metals: result of a controlled field study.
47 Atmos. Environ. 22: 2263-2274.
48
49 Edney, E. O.; Cheek, S. F.; Corse, E. W.; Spence, J. W.; Haynie, F. H. (1989) Atmospheric weathering
50 caused by dry deposition of acidic species. J. Environ. Sci. Health Part A 24: 439-457.
51
52 Federal Register. (1971) National primary and secondary ambient air quality standards. F. R. (April 30)
53 36: 8186-8201.
54
April 1995 2-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 Federal Register. (1979a) National ambient air quality standards; review of criteria and standards for paniculate
2 matter and sulfur oxides. F. R. (October 2) 44: 56730-56731.
3
4 Federal Register. (1979b) National primary and secondary ambient air quality standards: revisions to the national
5 ambient air quality standards for photochemical oxidants. F. R. (February 8) 44: 8202-8221.
6
7 Federal Register. (1984) Proposed revisions to the national ambient air quality standards for paniculate matter. F.
8 R. (March 20) 49: 10408-10435.
9
10 Federal Register. (1986a) National ambient air quality standards; review of criteria and standards for paniculate
11 matter and sulfur oxides. F. R. (April 1) 51: 11058.
12
13 Federal Register. (1986b) National ambient air quality standards: review of criteria and standards for paniculate
14 matter and sulfur oxides. F. R. (July 3) 51: 24392-24393.
15
16 Federal Register. (1987) Revisions to the national ambient air quality standards for paniculate matter. F. R.
17 (July 1) 52: 24634-24669.
18
19 Federal Register. (1994) Air quality criteria for paniculate matter. F. R. (April 12) 59: 17375.
20
21 Ferris, B. G., Jr.; Higgins, I. T. T.; Higgins, M. W.; Peters, J. M. (1973) Chronic nonspecific respiratory
22 disease in Berlin, New Hampshire, 1961 to 1967: a follow-up study. Am. Rev. Respir. Dis.
23 107: 110-122.
24
25 Ferris, B. G., Jr.; Chen, H.; Puleo, S.; Murphy, R. L. H., Jr. (1976) Chronic nonspecific respiratory disease in
26 Berlin, New Hampshire, 1967 to 1973: a further follow-up study. Am. Rev. Respir. Dis. 113: 475-485.
27
28 Gauri, K. L. (1979) Effect of acid rain on structures. In: Acid rain: a session at the national convention; April;
29 Boston, MA. New York, NY: American Society of Civil Engineers, Environmental Impact Analysis
30 Research Council; pp. 55-75.
31
32 Hausman, J. A.; Ostro, B. D.; Wise, D. A. (1984) Air pollution and lost work. Cambridge, MA: National
33 Bureau of Economic Research; NBER working paper no. 1263.
34
35 He, Q.-C.; Lioy, P. J.; Wilson, W. E.; Chapman, R. S. (1993) Effects of air pollution on children's pulmonary
36 function in urban and suburban areas of Wuhan, People's Republic of China. Arch. Environ. Health
37 48:382-391.
38
39 Hoek, G.; Brunekreef, B. (1993) Acute effects of a winter air pollution episode on pulmonary function and
40 respiratory symptoms of children. Arch. Environ. Health 48: 328-335.
41
42 International Standards Organization. (1981) Size definitions for particle sampling: recommendations of ad hoc
43 working group appointed by Committee TC 146 of the International Standards Organization. Am. Ind.
44 Hyg. Assoc. J. 42: A64-A68.
45
46 Johnson, J. B.; Elliott, P.; Winterbottom, M. A.; Wood, G. C. (1977) Short-term atmospheric corrosion of mild
47 steel at two weather and pollution monitored sites. Corros. Sci. 17: 691-700.
48
49 Johnston-Feller, R.; Osmer, D. (1977) Exposure evaluation: quantification of changes in appearance of pigmented
50 materials. J. Coat. Technol. 49: 25-36.
51
52 Knotkova-Cermakova, D.; Vlckova, J.; Honzak, J. (1982) Atmospheric weathering steels. In: Dean, S. W.;
53 Rhea, E. C., eds. Atmospheric corrosion of metals. Philadelphia, PA: Aerican Society for Testing and
54 Materials; special technical publication 767.
April 1995 2-33 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lampa, K.; Saarnak, A. (1986) Acid rain test: short term test on paint systems for corrosion protection. Farbe
2 Lacke 92: 692-696.
3
4 Lawther, P. J. (1986) [Letter to John Bachmann]. Washington, DC: Office of Air Quality Planning and
5 Standards; August 22. Available for inspection at: U.S. Environmental Protection Agency, Central
6 Docket Section, Washington, DC; docket no. A-82-37, IV-D-319.
7
8 Lawther, P. J.; Waller, R. E.; Henderson, M. (1970) Air pollution and exacerbations of bronchitis. Thorax
9 25: 525-539.
10
11 Lippmann, M. (1986a) [Letter to EPA Administrator Lee Thomas]. Washington, DC: U.S. Environmental
12 Protection Agency, Clean Air Scientific Advisory Committee; January 2. Available for inspection at:
13 U.S. Environmental Protection Agency, Central Docket Section, Washington, DC; docket no. A-82-37,
14 IV-D-315.
15
16 Lippmann, M. (1986b) [Letter to EPA Administrator Lee Thomas]. Washington, DC, U.S. Environmental
17 Protection Agency, Clean Air Scientific Advisory Committee; December 15. Available for inspection at:
18 U.S. Environmental Protection Agency, Central Docket Section, Washington, DC; docket no. A-82-37,
19 IV-D-339.
20
21 Lippmann, M. (1986c) [Letter to EPA Administrator Lee Thomas]. Washington, DC: U.S. Environmental
22 Protection Agency, Clean Air Scientific Advisory Committee; December 16. Available for inspection at:
23 U.S. Environmental Protection Agency, Central Docket Section, Washington, DC; docket no. A-82-37,
24 IV-D-338.
25
26 Malm, W. C.; Bell, P.; McGlothin, G. E. (1984) Field testing a methodology for assessing the importance of
27 good visual quality. Presented at: the 77th annual meeting of the Air Pollution Control Association; June;
28 San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association.
29
30 Mazumdar, S.; Schimmel, H.; Higgins, I. T. T. (1982) Relation of daily mortality to air pollution: an analysis of
31 14 London winters, 1958/59-1971/72. Arch. Environ. Health 37: 213-220.
32
33 Nriagu, J. O. (1978) Deteriorative effects of sulfur pollution on materials. In: Nriagu, J. O., ed. Sulfur in the
34 environment: part II, ecological effects. New York, NY: Wiley and Sons.
35
36 Ostro, B. (1984) A search for a threshold in the relationship of air pollution to mortality: a reanalysis of data on
37 London winters. Environ. Health Perspect. 58: 397-399.
38
39 Ostro, B. D. (1987) Air pollution and morbidity revisited: a specification test. J. Environ. Econ. Manage.
40 14: 87-98.
41
42 Ostro, B. D.; Lipsett, M. J.; Wiener, M. B.; Seiner, J. C. (1991) Asthmatic responses to airborne acid aerosols.
43 Am. J. Public Health 81: 694-702.
44
45 Ozkaynak, H.; Spengler, J. D. (1985) Analysis of health effects resulting from population exposures to acid
46 precipitation precursors. Environ. Health Perspect. 63: 45-55.
47
48 Pope, C. A., III.; Dockery, D. W. (1992) Acute health effects of PM10 pollution on symptomatic and
49 asymptomatic children. Am. Rev. Respir. Dis. 145: 1123-1128.
50
51 Pope, C. A., Ill; Dockery, D. W.; Spengler, J. D.; Raizenne, M. E. (1991) Respiratory health and PM10
52 pollution: a daily time series analysis. Am. Rev. Respir. Dis. 144: 668-674.
53
April 1995 2-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pope, C. A., Ill; Schwartz, J.; Ransom, M. R. (1992) Daily mortality and PM10 pollution in Utah valley. Arch.
2 Environ. Health 47: 211-217.
3
4 Roemer, W.; Hoek, G.; Brunekreef, B. (1993) Effect of ambient winter air pollution on respiratory health of
5 children with chronic respiratory symptoms. Am. Rev. Respir. Dis. 147: 118-124.
6
7 Ross, D. M.; Malm, W. C.; Loomis, R. J. (1985) The psychological valuation of good visual air quality by
8 national park visitors. Presented at: the 78th annual meeting of the Air Pollution Control Association;
9 June; Detroit, MI. Pittsburgh, PA: Air Pollution Control Association.
10
11 Ross, D. M.; Malm, W. C.; Loomis, R. J. (1987) An examination of the relative importance of park attributes at
12 several national parks. In: Bhardwaja, P. S., ed. Visibility protection: research and policy aspects.
13 Pittsburgh, PA: Air Pollution Control Association.
14
15 Schwartz, J. (1989) Lung function and chronic exposure to air pollution: a cross-sectional analysis of NHANES
16 II. Environ. Res. 50: 309-321.
17
18 Schwartz, J. (1991) Paniculate air pollution and daily mortality in Detroit. Environ. Res. 56: 204-213.
19
20 Schwartz, J. (1992) Air pollution and the duration of acute respiratory symptoms. Arch. Environ. Health
21 47: 116-122.
22
23 Schwartz, J. (1993) Air pollution and daily mortality in Birmingham, Alabama. Am. J. Epidemiol.
24 137: 1136-1147.
25
26 Schwartz, J.; Dockery, D. W. (1992) Increased mortality in Philadelphia associated with daily air pollution
27 concentrations. Am. Rev. Respir. Dis. 145: 600-604.
28
29 Schwartz, J.; Marcus, A. (1990) Mortality and air pollution in London: a time series analysis. Am. J. Epidemiol.
30 131: 185-194.
31
32 Schwartz, J.; Wypij, D.; Dockery, D.; Ware, J.; Zeger, S.; Spengler, J.; Ferris, B., Jr. (1991) Daily diaries of
33 respiratory symptoms and air pollution: methodological issues and results. Environ. Health Perspect.
34 90: 181-187.
35
36 Shumway, R. H.; Tai, R. Y.; Tai, L. P.; Pawitan, Y. (1983) Statistical analysis of daily London mortality and
37 associated weather and pollution effects. Sacramento, CA: California Air Resources Board; contract no.
38 Al-154-33.
39
40 Silver-man, F.; Hosein, H. R.; Corey, P.; Holton, S.; Tarlo, S. M. (1992) Effects of paniculate matter exposure
41 and medication use on asthmatics. Arch. Environ. Health 47: 51-56.
42
43 Skerry, B. S.; Alavi, A.; Lindgren, K. I. (1988) Environmental and electrochemical test methods for the
44 evaluation of protective organic coatings. J. Coat. Technol. 60: 97-106.
45
46 Stebbings, J. H., Jr.; Hasselblad, V.; Chapman, R. S.; McClain, K. E. (1974) Ventilatory function in school
47 children Chattanooga 1971-1972. Research Triangle Park, NC: U.S. Environmental Protection Agency,
48 National Environmental Research Center, Human Studies Laboratory; December 4.
49
50 Swift, D. L.; Proctor, D. F. (1982) Human respiratory deposition of particles during oronasal breathing. Atmos.
51 Environ. 16: 2279-2282.
52
53
April 1995 2-35 DRAFT-DO NOT QUOTE OR CITE
-------
1 Trijonis, J. C.; Pitchford, M.; McGown, M. (1987) Preliminary extinction budget results from the RESOLVE
2 program. In: Bhardwaja, P. S., ed. Visibility protection: research and policy aspects, an APCA specialty
3 conference; September 1986; Grand Teton National Park, WY. Pittsburgh, PA: Air Pollution Control
4 Association; pp. 872-883. (APCA transactions no. TR-10).
5
6 Trijonis, J.; McGown, M.; Pitchford, M.; Blumenthal, D.; Roberts, P.; White, W.; Macias, E.; Weiss, R.;
7 Waggoner, A.; Watson, J.; Chow, J.; Flocchini, R. (1988) Visibility conditions and causes of visibility
8 degradation in the Mojave Desert of California: executive summary, RESOLVE project final report.
9 China Lake, CA: Department of the Navy, Naval Weapons Center; document no. NWC TP 6869.
10 Available from: NTIS, Springfield, VA; AD-A206 322.
11
12 Trijonis, J. C.; Malm, W. C.; Pitchford, M.; White, W. H. (1991) Visibility: existing and historical
13 conditions—causes and effects. In: Irving, P. M., ed. Acidic deposition: state of science and technology,
14 volume III, terrestrial, materials, health and visibility effects. Washington, DC: The U.S. National Acid
15 Precipitation Assessment Program. (State of science and technology report no. 24).
16
17 U.S. Code. (1991) Clean Air Act, section 108, air quality criteria and control techniques, § 109, national ambient
18 air quality standards. U.S. C. 42: § 7408-7409.
19
20 U.S. Environmental Protection Agency. (1982) Air quality criteria for particulate matter and sulfur oxides.
21 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
22 and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
23 VA; PB84-156777.
24
25 U.S. Environmental Protection Agency. (1986) Second addendum to air quality criteria for particulate matter and
26 sulfur oxides (1982): assessment of newly available health effects information. Research Triangle Park,
27 NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office;
28 EPA report no. EPA-600/8-86-020F. Available from: NTIS, Springfield, VA; PB87-176574.
29
30 U.S. Environmental Protection Agency. (1989) An acid aerosols issue paper: health effects and aerometrics.
31 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
32 and Assessment Office; EPA report no. EPA-600/8-88-005F. Available from: NTIS, Springfield, VA;
33 PB91-125864.
34
35 Vedal, S.; Schenker, M. B.; Munoz, A.; Samet, J. M.; Batterman, S.; Speizer, F. E. (1987) Daily air pollution
36 effects on children's respiratory symptoms and peak expiratory flow. Am. J. Public Health 77: 694-698.
37
38 Ware, J. H.; Ferris, B. G., Jr.; Dockery, D. W.; Spengler, J. D.; Stram, D. O.; Speizer, F. E. (1986) Effects
39 of ambient sulfur oxides and suspended particles on respiratory health of preadolescent children. Am.
40 Rev. Respir. Dis. 133: 834-842.
41
42 White, J.; Rothschild (1987) Defects in the finish of motor cars. Met. Finish. 85: 15-18.
43
44 Xu, X.; Wang, L. (1993) Association of indoor and outdoor particulate level with chronic respiratory illness.
45 Am. Rev. Respir. Dis. 148: 1516-1522.
46
April 1995 2-36 DRAFT-DO NOT QUOTE OR CITE
-------
i 3. PHYSICS AND CHEMISTRY OF
2 PARTICULATE MATTER
3
4
5 3.1 INTRODUCTION
6 3.1.1 Overview
7 Atmospheric particles originate from a variety of sources and possess a range of
8 morphological, chemical, physical, and thermodynamic properties. Examples include
9 combustion-generated particles such as diesel soot or fly ash, photochemically produced
10 particles such as those found in urban haze, salt particles formed from sea spray, and soil-
11 like particles from resuspended dust. Some particles are liquid, some are solid; others
12 contain a solid core surrounded by liquid. Atmospheric particles contain inorganic ions and
13 elements, elemental carbon, organics compounds, and crustal compounds. Some atmospheric
14 particles are hygroscopic and contain particle-bound water. The organic fraction is especially
15 complex. Hundreds of organic compounds have been identified in atmospheric aerosols,
16 including alkanes, alkanoic and carboxcylic acids, poly cyclic aromatic hydrocarbons, and
17 nitrated organic compounds (Rogge et al., 1993; Kaplan and Gordon, 1994; Mazurek et al.,
18 1989; Standley and Simoneit, 1987; Ip et al., 1984; Simoneit and Mazurek, 1982; Schuetzle
19 etal.,1975).
20 Particle diameters span more than four orders of magnitude, from a few nanometers to
21 one hundred micrometers. Combustion-generated particles, such as those from power
22 generation, from automobiles, and in tobacco smoke, can be as small as 0.01 /^m and as
23 large as 1 /xm. Particles produced in the atmosphere by photochemical processes range in
24 diameter from 0.05 to 2 pirn. Fly ash produced by coal combustion ranges from 0.1 to
25 50 ^m or more. Wind-blown dust, pollens, plant fragments, and cement dusts are generally
26 above 2 /*m in diameter. Particles as small as a few nanometers (Covert et al., 1992;
27 Clarke, 1992) and as large as 100 /*m have been measured in the atmosphere (Lin et al.,
28 1993).
29 Particles are ubiquitous in the atmosphere. The lowest concentrations are found in
30 background marine environments, where particle number concentrations range from 100/cm3
31 to 400/cm3. In background continental environments, particle concentrations vary from
April 1995 3_! DRAFT-DO NOT QUOTE OR CITE
-------
1 100/cm3 to 5,000/cm3; while in urban areas of the United States concentrations may be as
2 high as 4,000,000/cm3 (Willeke and Whitby, 1975; Whitby and Sverdrup, 1980). Particles
3 account for a mass of a few /zg/m3 near the surface over dry continental areas to several
4 hundred /ig/m3 in polluted urban areas.
5 The composition and behavior of airborne particles are fundamentally linked with those
6 of the surrounding gas. Although the term aerosol is often used to refer to suspended
7 particles, aerosol is defined as a dilute suspension of solid or liquid particles in gas.
8 Paniculate material can be primary or secondary.
9 Primary particles are those emitted in paniculate form and include wind-blown dust, sea
10 salt, road dust, mechanically generated particles and combustion-generated particles such as
11 fly ash and soot. The concentration of primary particles depends on their emission rate,
12 transport and dispersion, and removal rate from the atmosphere.
13 Secondary particulate material may form from condensation of high temperature vapor
14 or from vapors generated at as a result of chemical reactions involving gas-phase precursors.
15 Secondary formation processes can result in either the formation of new particles
16 (Wiedensohler et al., 1994; Covert et al., 1992; Clarke et al., 1991, 1993; Frick and
17 Hoppel, 1993; Hoppel et al., 1994; Weber et al., 1994) or the addition of particulate
18 material to preexisting particles (Andreae et al., 1986; Wall et al., 1988; Wu and Okada,
19 1994). Most atmospheric sulfate is formed from atmospheric oxidation of sulfur dioxide.
20 Atmospheric nitrate is also essentially secondary, formed from reactions involving oxide of
21 nitrogen to form nitric acid. A portion of the organic aerosol is also attributed to secondary
22 processes (Hildemann et al., 1994; Turpin and Huntzicker, 1991; Mylonas et al., 1991;
23 Pickle et al., 1990; Gray et al., 1986). Secondary aerosol formation can depend on
24 concentrations of other gaseous reactive species such as ozone or hydrogen peroxide,
25 atmospheric conditions including solar radiation and relative humidity, and the interactions of
26 precursors and preexisting particles with cloud or fog droplets (Meng and Seinfeld, 1993;
27 McMurry and Wilson, 1983; Hoppel and Frick, 1990). As a result, it is considerably more
28 difficult to relate ambient concentrations of secondary species to sources of precursor
29 emissions than it is to identify the sources of primary particles.
30 Airborne particulate matter can be anthropogenic or biogenic in origin. Both
31 anthropogenic and biogenic particulate material can occur from either primary or secondary
April 1995 3-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 processes. Anthropogenic refers to participate matter which is directly emitted or formed
2 from precursors which are emitted as a result of human activity. Primary anthropogenic
3 sources include fossil fuel combustion, fireplace emissions, and road dust. Secondary
4 anthropogenic paniculate material can be generated photochemically from anthropogenic
5 SO2, NOX, or organic gases. Primary biogenic sources include leaf waxes and other plant
6 fragments from plants (Simoneit and Mazurek, 1982). In addition, plants emit gaseous
7 species such as terpenes (Lamb et al., 1987). Terpenes are photochemically reactive, and in
8 the presence of nitrogen oxides can form secondary organic particles (Kamens et al., 1981;
9 Pondis et al., 1991, 1993). Other types of primary particulate material such as sea salt and
10 wind-generated dust from soil undisturbed by man also are of non-anthropogenic origin.
11 In addition to secondary formation, volatilization and sorption processe's also affect
12 concentrations and compositions of airborne particles. Some aerosol constituents are
13 semivolatile and exist in both gas and particle phases. Their gas-particle distribution depends
14 on atmospheric conditions such as temperature, the concentrations of other aerosol species
15 including water vapor, and the vapor pressure of the constituent. Some inorganic compounds
16 such as ammonium nitrate (Stelson and Seinfeld, 1982; Bassett and Seinfeld, 1983, 1984) and
17 organic compounds, including many polycyclic aromatic hydrocarbons (Yamasaki et al.,
18 1982; Ligocki and Pankow, 1989; Pankow, 1987, 1994a,b) are semivolatile. Diurnal
19 temperature fluctuations can cause substantial changes in the particle-phase concentrations of
20 semivolatile constituents as a result of gas-particle redistribution. Evidence exists suggesting
21 that this volatilization-sorption cycle results in the redistribution of semivolatile material
22 among particles of differing origins (Venkataraman and Hildemann, 1994).
23 Sampling semi-volatile species requires special techniques, such as the use of denuder
24 systems (Kautrukis et al., 1988). The processing of atmospheric particles also occurs in
25 clouds. For example, in-cloud processes can lead to the combination of many small particles
26 (Andreae et al., 1986).
27 A complete description of the atmospheric aerosol would include an accounting of the
28 chemical composition, morphology, and size of each particle and the relative abundance of
29 each particle type as a function of particle size (Friedlander, 1970). However, most often
30 the physical and chemical characteristics of particles are measured separately. Number size
31 distributions are often determined by physical means, such as electrical mobility or light-
April 1995 3-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 scattering. Chemical composition is determined by analysis of collected samples. The mass
2 size distribution and the average chemical composition of the aerosol as a function of size
3 can be determined by collection of size-segregated samples (Countess et al., 1980; Hering
4 and Friedlander, 1982; John et al., 1990; Sloane et al., 1991). Recent developments in
5 single particle analysis techniques coupled with multivariate classification methods (Van
6 Grieken and Xhoffer, 1992; Germani and Buseck, 1991; Mansoori et al., 1994) are bringing
7 the description envisioned by Friedlander closer to reality. This introductory section
8 describes some of the measurements that have been made on atmospheric particles, and the
9 insights thus provided on the nature, origins, and atmospheric processes that affect particle
10 composition.
11
12 3.1.2 Major Chemical Constituents
13 The major constituents of atmospheric aerosol are sulfates, nitrates, carbonaceous
14 compounds, water, ammonium ions and materials of crustal origin. Inorganic ions, including
15 sulfate and nitrate, are typically analyzed by ion chromatography. Crustal elements are
16 analyzed by x-ray fluorescence and/or proton-induced x-ray emission. Average compositions
17 vary with particle size, by location and season. The equilibrium models for inorganic ions
18 predict that water is an important constituent of atmospheric particles, but measurements are
19 limited. McMurry and coworkers (McMurry and Stolzenburg, 1989; Zhang et al., 1993)
20 measured the sensitivity of particle size to relative humidity (RH) for Los Angeles and Grand
21 Canyon aerosols. They found that atmospheric particles of a single size exhibited two
22 distinct hygroscopicities. These were described as "more" and "less" hygroscopic, as shown
23 in Figure 3-1. For example, the diameters of more hygroscopic 0.2 /mi particles humidified
24 to approximately 90% RH increased by factors of 1.23 ± 0.08 and 1.49 ± 0.11 for Los
25 Angeles and Grand Canyon particles, respectively. For relative humidities above 85 or 90%,
26 water was the most abundant particulate species both in Los Angeles and at the Grand
27 Canyon.
28 Because of the multitude of carbonaceous compounds present in atmospheric aerosols,
29 carbonaceous material is often categorized as organic or elemental carbon (OC or EC). Most
30 measurements of aerosol carbon are made using one of a variety of thermal techniques that
April 1995 3-4 DRAFT-DO NOT QUOTE OR CITE
-------
^
5
Q,
cf
^
2
O,
cf
<
s.
cf
*a
I
l.OU
1.55
1.50
1.45
1.40
1.35
1.30
1.25
1.20
1.15
1.10
1.05
1f\f\
.00
One
.yo
(
1Cf\
.60
1.55
1.45
1.40
1.35
1.30
1.25
1.20
1.15
1.10
1.05
1.00
A nc
i i i i i i i i i
o More Hygroscopic Particles
• Less Hygroscopic Particles
-
-
-
o o -
%
0°
o §S8
0 °
O ^ m
S& °& <$B^° i <*T i i 1*1
) 10 20 30 40 50 60 70 80 90 10
DMA2 Relative Humidity
1 I 1 1 1 1 1 1 Of
„ o More Hygroscopic Particles Oo -^ -
• Less Hygroscopic Particles 0 <§g3o
o
12N»
^K d> •
0 @% • ..
; Ose#& -
4, "WT -
° g ro0j*«1j'. -
» .^M«« %^° f*?1** ..:
1 1 f 1 ) |_ L 1 1
10 20 30 40 50 60 70
DMA2 Relative Humidity
Figure 3-1. Particle size related to RH.
80 90 100
1 report paniculate organic and elemental carbon concentrations (Huntzicker et al., 1982;
2 Mueller, 1982; Turpin et al., 1990). The split between organic and elemental carbon is
3 somewhat operationally defined, but the term elemental generally refers to the nonvolatile,
4 optically absorbing (black) portion of the carbon aerosol. Elemental carbon is associated
5 with soot emissions from combustion. The remaining, more volatile portion is termed
6 organic. Various methods of further classifying the organic fraction include: selective
April 1995
3-5
DRAFT-DO NOT QUOTE OR CITE
-------
1 solvent extraction (Lioy and Daisey, 1986), functional group identification (Allen et al.,
2 1994; Gordon et al., 1988), and division into neutral and acidic fractions (Hildemann et al.,
3 1994a). Radiocarbon dating techniques have been used to distinguish fossil and
4 contemporary carbon (Currie et al., 1994; Kaplan and Gordon, 1994; Hildemann et al.,
5 1994b).
6
7 3.1.3 Atmospheric Aerosol Size Distributions
8 Size is one of the most important parameters in determining the atmospheric lifetime
9 and deposition of particles. As a result, the environmental, visual, and health effects of
10 atmospheric aerosols and the fate of the compounds that they contain are strongly dependent
11 on the particle size distribution. Particle size largely determines deposition patterns of
12 particle-phase compounds within the lung. Light scattering is also strongly dependent on
13 particle size, and thus particle size distributions have a strong influence on atmospheric
14 visibility and radiative balance (i.e., climate).
15 Atmospheric size distributions for averaged continental background, urban-influenced
16 background, averaged urban, and freeway-influenced urban aerosols are shown in
17 Figures 3-2 to 3-4 (Whitby and Sverdrup, 1980). Figure 3-2 describes the number of
18 particles as a function of particle diameter. For the same data, the particle volume
19 distribution with respect to size is shown in Figures 3-3 and 3-4. Number and volume
20 distributions are defined such that the number (or volume) of particles in a specified size
21 range is proportional to the corresponding area under the curve. These distributions show
22 that most of the particles are quite small, below 0.1 /mi, while most of the particle volume
23 (and therefore most of the mass) is found in particles > 0.1.
24 An important feature of atmospheric aerosol size distributions is their multimodal
25 nature. Volume distributions in ambient air are almost always bimodal, with a minimum
26 between 1 and 3 /on. Particles in the larger mode are termed "coarse" and those in the
27 smaller made, "fine". Whitby and Sverdrup (1980) and Willeke and Whitby (1975)
28 identified three modes: nuclei, accumulation, and coarse. The three modes are most
29 apparent in the freeway-influenced size distribution of Figure 3-4. The smallest mode,
30 corresponding to particles below about 0.08 /mi, is the nuclei mode. The middle mode,
April 1995 3-6 DRAFT-DO NOT QUOTE OR CITE
-------
\
Q.
Q
J£
z
TJ
1,000,000--
10,000- -
100--
0.01 1
0.0001 - -
0.000001 - -
Average Background
Urban Influenced Background
Average Urban
Urban + Freeway
0.001
0.01
0.1
1
10
100
Particle Diameter, Dp (urn)
Figure 3-2. Number of particles as a function of particle diameter.
f
TO
•^
"S
Q,
o>
o
TJ
1
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
0.(
I \ \ 7! 1 1
Background .' '.
Average •' '•
Background
Urban Influenced ,' '•
Background
South-Central • •
New .
1
-
-
/ A\
' : ! ',\
i . 1 t i
'• •' / Vi
1 '"''/ A
* i / '• \\
/ / \\\
:/'"^-/'J/""^^>^
X)1 0.01 0.1 1 10 1C
Particle Diameter, Dp
Figure 3-3. Particle volume distribution as a function of particle diameter.
April 1995
3-7
DRAFT-DO NOT QUOTE OR CITE
-------
70
65
60--
55--
—, 50- •
40- •
§ 30+
0) 25-
^
"O
15-•
la-
s'-
Average Urban
Urban + Freeway
0.001 0.01 0.1 1 10
Particle Diameter, Dp
100
Figure 3-4. Particle volume distribution as a function of particle diameter in a
freeway-influenced area.
Source: Whitby and Sverdrup (1980).
1 from 0.08 to 1 or 2 /mi, is the accumulation mode. The largest particles (> 1 or 2 /mi)
2 comprise the coarse mode. Formal delineation of these modes arises from the fitting of a
3 trimodal, lognormal distribution function to the data.
4 Whitby and coworkers observed that continental background aerosols not influenced by
5 sources have a small accumulation mode and no nuclei mode. For urban aerosols, the
6 accumulation and coarse particles modes are comparable in volume. The nuclei mode is
7 small in volume but dominates the number distributions of urban aerosols. More recent
8 measurements of fine-particle size distributions (Eldering and Cass, 1994) and species-size
9 distributions for sulfates, nitrates, and ammonium ion, as discussed below, indicate that the
10 accumulation mode can be further divided into a "condensation" and "droplet" mode (John
11 et al., 1990). Measurements over remote areas (Hoppel et al., 1986, 1990) indicate that the
12 nuclei mode can also be divided into two separate modes.
13 Many measurements indicate that the chemical compositions of coarse and fine particles
14 are distinct. The processes that affect the formation and removal of these two size fractions
April 1995
3-8
DRAFT-DO NOT QUOTE OR CITE
-------
1 of atmospheric aerosols are also distinct. Coarse particles are generated by mechanical
2 processes and consist of soil dust, sea spray, plant fragments, particles from tire wear, and
3 emissions from rock-crushing operations. These particles are removed primarily by
4 impaction and settling. Nuclei and accumulation mode particles contain primary particles
5 from combustion sources and secondary particles that result from condensation of low-
6 volatility vapors formed from chemical reactions. Particles in the nuclei mode may be
7 transferred into the accumulation mode by coagulation, but cloud coalescence and liquid
8 phase cloud droplet transformations may be more important in atmosphere. In contrast,
9 accumulation mode particles do not ordinarily grow into the coarse mode, because number
10 concentrations are too low for coagulation to be effective. Nuclei are readily removed by
11 diffusion to surfaces. However, accumulation mode particles are not easily removed from
12 the airstream. They have long atmospheric lifetimes and are able to penetrate deep into the
13 lungs. The nuclei and accumulation modes are fairly independent of the coarse mode, both
14 in formation and removal (Willeke and Whitby, 1975; Whitby and Sverdrup, 1980).
15 Fine and coarse particles are best differentiated by their formation mechanism (Wilson,
16 1995). Fine particles are formed by nucleation with gases while coarse particles are formed
17 by mechanical processes from larger particles or bulk materials. The most appropriate size
18 cut for separating fine from coarse particles is still under consideration.
19
20 3.1.4 Chemical Composition and Its Dependence on Particle Size
21 Since the work of Whitby, several studies have been conducted that provide chemical or
22 elemental composition data on the coarse and fine fractions of the atmospheric aerosol.
23 Generally this is done by separate collection of particles less than 2.5 ^m in diameter (fine or
24 PM2 5) and particles less than 10 /^m (PM10). Coarse-particle concentrations are obtained by
25 difference. Alternatively, particles can be collected in two or more size fractions, using
26 impactor methods.
27 Detailed size distributions of the inorganic ions in Los Angeles are shown in Figure 3-5
28 (Wall et al., 1988; John et al., 1990). These data show two modes for sulfate and nitrate
29 aerosols between 0.1 and 1 /xm. Similar results for sulfate aerosols were reported by Hering
30 and Freidlander (1982). The smaller mode, corresponding to particles near 0.2 pirn in
31 diameter, is attributed to gas-phase formation of condensible species and is referred to as the
April 1995 3.9 DRAFT-DO NOT QUOTE OR CITE
-------
500-
o
o>
_g
TJ
300-
m
£
O
200 H
100-
Legend
Ammonium, NH
Nitrate, NO2"
-2
Sulfate,SO4
Hydrogen Ion, H+
Sodium, Na+
Chloride, Cl"
0.01 0.1 1
Aerodynamic Diameter, Dae (n,m)
Figure 3-5. In concentration as a function of particle size.
1 condensation mode. The larger mode has a peak near 0.6 fim and is called the droplet
2 mode. Its existence is attributed to secondary formation through heterogeneous, aqueous-
3 phase transformations. McMurry and Wilson (1983) found 0.6 /^m sulfate particles in power
4 plant plumes and attributed their existence to formation by heterogeneous processes. Further
5 analysis of the data by Meng and Seinfeld (1994) indicate that these aqueous reactions most
6 likely occur in cloud or fog droplets.
1 The data of Figure 3-5 in Los Angeles show that paniculate nitrate is found in both
2 coarse and fine particles. Nitrate near the coast was predominantly in the coarse mode.
3 Coarse mode nitrate was less prominent for inland sites. Several investigators (Wall et al.,
4 1988; John et al., 1990; Andreae et al., 1986) proposed that the coarse particle nitrate results
5 from the heterogeneous reaction of nitric acid with sea salt. On the basis of single particle
6 analysis by electron microscopy-energy dispersive spectroscopy, Wu and Okada (1994)
7 concluded that coarse-particle nitrate in a coastal region of Japan formed on sea salt. Course
8 nitrate collected at an inland site was associated with soil dust. These data suggest that a
9 heterogeneous chemical reaction on the surface of a mechanically generated, primary particle
April 1995 3-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 may provide a mechanism for adding secondary material to the coarse particle mode. They
2 also show that secondary paniculate material can be formed by the interaction of a natural
3 constituent (sea salt) with a species derived from anthropogenic emissions (nitric acid).
4
5 3.1.5 Particle-Vapor Partitioning
6 Several atmospheric aerosol species, such as ammonium nitrate and certain organic
7 compounds, are semi volatile and are found in both gas and particle phases. The gas-particle
8 distribution of semivolatile organic compounds depends on compound vapor pressure, total
9 particulate surface area and composition, and atmospheric temperature (Pankow, 1987;
10 Junge, 1977; Bidleman, 1988). Junge (1977) modeled this relationship using a linear form of
11 a Langmuir adsorption isotherm. Measurements of semivolatile organic compounds show
12 that gas-particle distributions are highly correlated with total suspended particulate matter,
13 temperature, and the sub-cooled liquid vapor pressure of the pure compound (Foreman and
14 Bidleman, 1990; Ligocki and Pankow, 1989; Yamasaki et al., 1982). Yamasaki et al. (1982)
15 used this information to model an empirical relationship between the gas-particle distribution,
16 total suspended particulate matter and temperature. Pankow showed that the expressions of
17 Junge (1977) and Yamasaki et al. (1982) are consistent and continued the theoretical
18 development of equilibrium gas-particle partitioning (Pankow, 1987; 1991; 1994a,b).
19 Although it is generally assumed that the gas-particle partitioning of semivolatile
20 organics is in equilibrium in the atmosphere, the kinetics of redistribution are not well
21 understood. Gerde and Scholander (1989) and Rounds and Pankow (1993) predicted that
22 redistribution in the ambient air could take minutes to hours. Since changes in atmospheric
23 conditions (i.e., temperature) will drive redistribution, it is not clear whether equilibrium
24 conditions are maintained. However, the gas and particle data agree reasonably well with
25 equilibrium theories. The development of an understanding of gas-particle partitioning of
26 semivolatile organic compounds is hampered by the difficulty associated with measuring the
27 multitude of compounds, all present in small concentrations, for which diurnal temperature
28 fluctuations cause gas-particle partitioning to be dynamic on a time scale of a few hours.
29 Stelson and Seinfeld (1982) developed a thermodynamic model to predict the
30 temperature and relative humidity dependence of the ammonium nitrate equilibrium
31 dissociation constant, which has been supported by ambient data at inland sites in the Los
April 1995 3_H DRAFT-DO NOT QUOTE OR CITE
-------
1 Angeles Basin (Hildemann et al., 1984; Doyle et al., 1979). Bassett and Seinfeld extended
2 the equilibrium model to include sulfates (1983) and the effect of particle size (1984). With
3 the inclusion of sodium chloride in the equilibrium model, Pilinis and Seinfeld (1987) were
4 able to predict observations at coastal sites. Atmospheric models based on equilibrium
5 considerations have been successful in accounting for the gas-particle partitioning of
6 inorganic species measured in Phoenix, Arizona (Watson et al., 1994b), and Uniontown,
7 Pennsylvania (Saxena et al., 1994). Wexler and Seinfeld (1992) found that under some
8 atmospheric conditions the size distributions of ammonium ion and nitrate are not accurately
9 predicted by equilibrium considerations alone, and that transport kinetics can be important.
10
11 3.1.6 Single Particle Characteristics
12 The "mixing characteristics" of the aerosol describes the distribution of chemical
13 species among particles. An aerosol in which all particles contain the same homogeneous
14 blend of chemical species is internally mixed. In an externally mixed aerosol each chemical
15 species is found in a distinct set of particles. Experiments measuring atmospheric aerosol
16 properties for single-particle size ranges (Hering and McMurry, 1991; Covert et al., 1990;
17 Zhang et al., 1993) and single-particle analyses (Bock et al., 1994; Sheridan et al., 1993;
18 Van Borm et al., 1989; Anderson et al., 1988) indicate that atmospheric aerosols are to some
19 degree both internally and externally mixed. Single particle analyses provide descriptions of
20 individual particle compositions. These are then categorized either manually or through
21 multivariate methods such as cluster analysis (Kim and Hopke, 1988) to give an accounting
22 of the relative number of particles of each chemically defined particle type. Morphological
23 information can also be included in particle type definitions.
24 Single-particle composition and morphology provide insights into the sources and
25 atmospheric processes affecting airborne particles. For example, a priori one expects that
26 particles emitted from different sources would in fact be distinct. However, Andreae et al.
27 (1986) observed that over remote ocean areas between 80 and 90% of silicon-rich particles
28 (presumably originating from silicate mineral particles) were also rich in sodium, chlorine,
29 and variable amounts of potassium, magnesium, calcium, and sulfur (attributed to sea salt
30 particles). The internal mixing of silicates with sea salt, particles originating from different
31 sources and externally mixed when emitted into the atmosphere, suggests the processing of
April 1995 3-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 aerosol particles within clouds (see Section 3.2.1.4). The hypothesis is that a single cloud
2 droplet takes up two or more particles and that these particles remain together after droplet
3 evaporation. Other mechanisms of particle coalescence, such as differential settling,
4 Brownian coagulation, and electrostatic attraction, are too slow to account for the large
5 fraction of internal mixing observed. Andreae et al. (1986) also found enrichment of sulfur
6 (presumably sulfate) on sea salt particles. This also was attributed to the interaction of
7 clouds with particles. Gas-to-particle conversion in cloud droplets or by condensation can
8 also lead to mixtures of aerosol species.
9 Particle morphology has many effects on atmospheric particle properties and processes.
10 Chain agglomerates, for example, have much larger surface areas on which adsorption and
11 chemical reactions can take place than spherical particles of identical volumes. In addition,
12 the atmospheric lifetime is longer, and the optical absorption per unit mass is greater for
13 chain agglomerates than for more compact particles. Combustion-generated soot particles are
14 often chain agglomerates composed of a large number of small primary spherules.
15 Laboratory experiments conducted by Huang et al. (1994) and Colbeck et al. (1990)
16 demonstrated that condensation-evaporation processes can cause chain agglomerates to
17 become more compact. Colbeck et al. (1990) also showed that the collapse of the soot
18 aggregates resulting from humidification results in a decrease in both the optical scattering
19 and extinction of the particles.
20
21 3.1.7 Definitions
22 3.1.7.1 Definitions of Particle Diameter
23 The diameter of a particle may be determined geometrically, from optical or electron
24 microscopy; by light scattering and Mie theory, or by its behavior, such as its electrical
25 mobility, its settling velocity, or its aerodynamic behavior. Although atmospheric particles
26 are often not spherical, their diameters are described by an "equivalent" diameter, that of a
27 sphere which would have the same physical behavior. Two parameters that are often used to
28 describe particle diameter are the Stokes and aerodynamic diameters. The Stokes diameter,
29 Dp, describes particle size based on the aerodynamic drag force imparted on a particle when
30 its velocity differs from that of the surrounding fluid. For a smooth, spherically shaped
31 particle, Dp exactly equals the physical diameter of the particle. For irregularly shaped
1995 3_13 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles, Dp is the diameter of an equivalent sphere that would have the same aerodynamic
2 resistance. Particles of equal Stokes diameters that carry the same electric charge will have
3 the same electrical mobility. Particles of equal density and equal Stokes diameter have the
4 same settling velocity.
5 Aerodynamic diameter, Da, depends on particle density and is defined as the diameter
6 of a particle with equal settling velocity but a material density of 1 g/cm3. Particles with the
7 same physical size and shape but different densities will have the same Stokes diameter but
8 different aerodynamic diameters. For particles greater than about 0.5 pm, the aerodynamic
9 diameter is generally the quantity of interest because it is the parameter that is important to
10 particle transport, collection, and respiratory tract deposition. Respirable, thoracic, and
11 inhalable particle sampling are based on particle aerodynamic diameter.
12
13 Aerodynamic diameter, Da, is related to the Stokes diameter, Dp, by:
14
P C
1/2
(3-D
15
16 where p is the particle density, and C and Ca are the Cunningham slip factors evaluated for
17 the particle diameters Dp and Da respectively. The slip factor is a function of the ratio
18 between particle diameter and mean free path of the suspending gas; it is given by the
19 expression (Hinds, 1982):
20
C = 1 + — {2.514 + 0.800 exp(-0.55 _?)} (3-2)
DP X
21
22 where X is the mean free path of the air. C is an empirical factor that accounts for the
23 reduction in the drag force on particles due to the "slip" of the gas molecules at the particle
24 surface. It is important for particles less than 1 /nm in diameter, for which the surrounding
25 air cannot be modeled by a continuous fluid. At normal atmospheric conditions (temperature
26 = 20 °C, pressure = 1 atmosphere) X = 0.066 /xm. For large particles (Dp > 5 pim)
27 C = 1; while for smaller particles C > 1.
April 1995 3-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 For particles with diameters greater than the mean free path, the aerodynamic diameter
2 given by equation (3-1) is approximated by:
Da=(p)1/2Dp (Dp>X) (3-3)
3
4 This expression, which shows that aerodynamic diameter is directly proportional to the
5 square root of the particle density, is often used for particles as small as 0.5 /mi. For
6 particles with diameters much smaller than the mean free path, the slip factor must be taken
7 into account. In this case the aerodynamic diameter is directly proportional to the particle
8 density [Da = (p) Dp for Dp«X].
9
10 3.1.7.2 Definitions of Particle Size Fractions
11 In the preceding discussion several modes of the aerosol size distribution were
12 identified; they are defined as follows:
13
14 Nuclei Mode: that portion of the fine mode particles with diameters below about
15 0.08 /mi;
16
17 Accumulation Mode: particles formed from gases;
18
19 Condensation Mode: that portion of the accumulation mode with a volume (mass)
20 median diameter near 0.2 /mi;
21
22 Droplet Mode: that portion of the accumulation mode with a volume (mass) median
23 diameter at 0.5 to 0.8 jim;
24
25 Fine Particles: The combination of the modes listed above.
26
27 Coarse Mode or Coarse Particles: mechanically generated particles.
28
29
30 There is some overlap between fine and coarse particles in the 1 to 3 jim region. For
31 further discussion see Chapter 4. PM2 5 refers to particles less than 2.5 /xm diameter and is
32 frequently referred to as "fine" PM. A discussion of the best size to differentiate fine from
33 coarse particles in given in Section 3.7.
34 Another set of definitions of particle size fractions arises from considerations of size-
35 selective sampling. Size-selective sampling refers to the collection of particles below a
April 1995 3-15 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
specified aerodynamic size cut, and has arisen in an effort to measure the particle size
fractions of importance to human health. The PM10 standard set by the U.S. Environmental
Protection Agency in 1987 is one example of size-selective sampling criteria, and it was
designed to match the penetration of particles into the thoracic region. It is a fairly sharp
cutpoint with a 50% efficiency near 10 ^m in particle aerodynamic diameter. The exact
definition is given by a table of efficiency values (Federal Register, 1988) and is shown in
Figure 3-6.
100r
A PM-10
• IPM
• TPM
O RPM
10
Aerodynamic Particle Diameter (urn)
Figure 3-6. Efficiency values for size-selective sampling criteria.
100
1
2
3
4
5
6
7
8
3.1.7.3 Other Terminology
Other terminology that has been introduced in this section is summarized below:
Primary Particles: those directly emitted to the atmosphere from either natural sources
or sources derived from human activity;
Secondary Particulate Material: material formed in the atmosphere as the result of
chemical and physical conversion of gaseous species;
April 1995
3-16
DRAFT-DO NOT QUOTE OR CITE
-------
1 Internal Mixture: an aerosol for which the chemical composition of each individual
2 particle is the same, that is, equal to the bulk composition;
3
4 External Mixture: an aerosol for which different chemical species comprise separate
5 particles;
6
7 Anthropogenic: derived from human activities;
8
9 Biogenic: derived from plants;
10
11 Bioaerosols: airborne microorganisms and aeroallergens;
12
13 Fossil: derived from fossil fuel combustion; and
14
15 Contemporary carbon: derived from non-fossil fuel sources such as plants, wood
16 burning, and cooking oils.
17
18 3.1.8 Field Studies
19 Appropriate and reliable field measurements play a central role in shaping our under-
20 standing of atmospheric processes, in providing key model inputs, and in the evaluation of
21 models. Real-world observations are all the more important in the case of atmospheric
22 aerosols, which, on the one hand, are the end product of many complex processes and, on
23 the other hand, are key precursors of important microphysical cloud processes. Field studies
24 include short-term, 3-D, high-resolution intensive research campaigns, as well as longer-term
25 surface and upper-air monitoring programs (in routine mode, or in more comprehensive and
26 higher-resolution research mode). Research studies are generally mechanistic (targeted at
27 understanding of process rates and mechanisms), and/or diagnostic (aimed at development
28 and testing of individual process modules or subgrid-scale parameterizations for use in
29 complex models). Routine monitoring studies are aimed more at operational evaluation of
30 overall model performance, or at generation of model input data including those (e.g.,
31 meteorological) which, through dynamic assimilation into the computations, can improve the
32 realism of the simulations. Since atmospheric fine particles (FP) are substantially of
33 secondary origin, measurements of their gaseous precursors and other reactants are also
34 important. In North America, most of the anthropogenic emissions of FP and their
35 precursors are from large point sources (power plants and smelters) and from urban-
April 1995 3-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 industrial complexes including vehicle emissions. Consequently, special attention is given in
2 this section to measurements in the plumes of such emissions.
3
4 3.1.9 Dry Deposition
5 Dry deposition is the process whereby airborne gases and particles are transported
6 down to the surface of the earth where they are removed. Atmospheric turbulent mixing
7 continually brings airborne gases and particles into close proximity to the earth's surface,
8 where they may diffuse across a thin layer of stagnant air to the surface itself. Actual
9 removal at the surface depends on the affinity between the airborne substance and the surface
10 element (ground, body of water, vegetation surface, etc.). Dry deposition is a complex
11 process but it is represented as occurring in three steps: 1) transport down to the vicinity of
12 the earth by turbulent mixing processes; 2) diffusion across a thin quasi-laminar layer of air;
13 and 3) attachment to the surface itself.
14
15 3.1.10 Atmospheric Scavenging
16 Atmospheric gases are scavenged directly by absorption in droplets and by chemical
17 reactions in clouds. The direct absorption of gases in falling droplets depends on the
18 solubility of the gas in water, and may be affected by the presence of other species in
19 solution (Seinfeld, 1986). Particles are scavenged when they serve as cloud condensation
20 nuclei (CCN) and when they are intercepted by falling hydrometers. The wet removal of
21 particles depends on the air trajectories through clouds, the supersaturation to which the air
22 mass is exposed, and the tune for which droplets are present before arriving at the ground.
23
24
25 3.2 PHYSICAL PROPERTIES
26 3.2.1 Aerosol Size Distributions
27 3.2.1.1 Particle Size Distribution Functions
28 The distribution of particles with respect to size is perhaps the most important physical
29 parameter governing their behavior. The concentration of the number of particles as a
30 function of their diameter is given by a particle number distribution.
April 1995 3_18 DRAFT-DO NOT QUOTE OR CITE
-------
1 Because the sizes of atmospheric particles cover several orders of magnitude in particle
2 size, and because atmospheric size distributions tend to be log-normally distributed (see
3 Section 3.2.2), size distributions are often expressed in terms of the logarithm of the particle
4 diameter, on the X-axis and differential concentration on the Y-axis:
dn
d(logDp)
6
7 = the number of particles per cm3 of air having diameters in the size range from log Dp to
8 log(Dp + dDp).
9
10 Formally, it is not proper to take the logarithm of a dimensional quantity, but one can think
11 of the distribution as a function of log(Dp/Dp0) where the reference diameter Dp0 = 1 jum is
12 not explicitly stated.
13 The number of particles is proportional to the area under the curve of n(logDp) versus
14 logDp. Similar considerations apply to distributions of surface, volume, and mass.
15
16 3.2.1.2 Log-Normal Size Distributions
17 As presented in Section 3.1, atmospheric aerosols tend to follow a sum of log-normal
18 distributions. A log-normal distribution is a specific functional form of the size distribution
19 function for which the population of particles follows a Gaussian distribution function with
20 respect to the logarithm of the particle diameter. The geometric standard deviation ag is the
21 standard deviation of the quantity logDp and defines the width of the distribution. For a
22 monodisperse aerosol, that is, one for which all particles are the same diameter, oe = 1.
o
23 For polydisperse aerosols, ag > 1. Typical values for one of the modes of the atmospheric
24 aerosol, such as the accumulation mode discussed above, are 1.8 < ag <2.8. For log-
25 normal distributions, 84.1% of the particles are below the size ffg-Dgn, 15.9% lie above the
26 size Dgn/ag, and 95% of the particles lie within two standard deviations of the mean, that is,
27 the range from Dgn/2ag to Dp-2ag.
April 1995 349 DRAFT-DO NOT QUOTE OR CITE
-------
1 One of the properties of the log-normal distribution is that if the number distribution is
2 log-normal, the surface and volume distributions are also log-normal, and their geometric
3 standard deviation ag is the same as for the number distribution.
o
4
5 3.2.1.3 Ambient Aerosol Size Distributions
6 Ambient aerosol size distributions are completely described by the geometric number
7 mean diameter Dgn, geometric standard deviation crg, and number concentration N for each
8 mode, as listed in Table 3-1. Also given are the parameters of the lognormal volume
9 distributions, geometric mean diameters Dgv, and the corresponding total particle volume for
10 each mode V. Because the distributions are lognormal, the geometric standard deviation is
11 the same for both number and volume distributions. The results from more recent
12 measurements in a nonurban area of New Mexico, for which the distribution is described by
13 the sum of two lognormal distributions, were obtained with laser light scattering
14 instrumentation, and were fitted to a bimodal lognormal form, corresponding to the
15 accumulation and coarse particle modes. The lognormal fit parameters characterizing these
16 distributions are also listed in Table 3-1. Note that the volume geometric mean diameters for
17 the accumulation mode vary from 0.2 /xm to 0.4 mm, those for the coarse mode from 5 to
18 12 urn. The standard deviations for the coarse particle mode tend to be larger than for the
19 accumulation mode.
20
21 3.2.1.4 Coagulation of Spherical Particles
22 Many processes affect the size distribution of an aerosol, including addition of volume
s
23 by gas-to-particle conversion, and losses by deposition. Even without these processes, under
24 conditions in which the total volume of an aerosol is conserved the number of particles will
25 decrease by coagulation while the average volume per particle increases. The coalescence of
26 two particles always reduces the total surface area and therefore is favored
27 thermodynamically. Thus, in this sense, aerosols are inherently unstable. In some cases
28 coagulation leads to the formation of chain agglomerates, such as for soot and some metals.
April 1995 3_20 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-1. LOGNORMAL PARAMETERS FOR AMBIENT AEROSOLS
K A. Parameters of
i§
Site of Measurement
Clean continental background
Average continental background
Urban influenced background
Urban average
Urban and freeway
South central NM - February 1989
South central NM - July 1989
Nuclei
Num.
(cm3)
N: 1,000
N
N
N
N
N
N
: 6,400
: 6,600
: 106,000
: 2,120,000
Model
Dgn
0.016 1.6
0.015 1.7
0.014 1.6
0.014 1.8
0.013 1.74
the Number
Accumulation Mode
Num.
(cm3)
800.00
2,300.00
9,600.00
32,000.00
37,000.00
706.00
253.00
Dgn
(/mi)
0.067
0.076
0.120
0.054
0.032
0.13
0.13
B. Parameters of
u>
to
DRAFT-DO 1
3
O
a
o
Site of Measurement
Clean continental background
Average continental background
Urban influenced background
Urban average
Urban and freeway
Sources: (1) Whitby and Sverdrup
V
V
V
V
V
(1980);
Nuclei
Volume
(/*m3 cm3)
: 0.01
: 0.04
: 0.03
: 0.63
: 9.20
(2) Kim et al.
Model
Dgv
0.030 1.6
0.034 1.7
0.028 1.6
0.038 1.8
0.032 1.74
(1993).
a
2.1
2.0
1.84
1.98
1.72
1.71
the Volume
Accumulation Mode
Volume
(/Ltm3 cm3)
1.50
4.45
44.00
38.40
37.50
Dgv
0.35
0.32
0.36
0.32
0.25
a
2.1
2.0
1.84
2.16
1.98
Distribution
Coarse Mode
Num.
(cm3)
0.72
3.2
7.2
5.4
4.9
0.42
0.72
Distribution
Dgn
(/im)
0.93
1.02
0.83
0.86
1.08
2.45
1.59
Og
2.2
2.16
2.12
2.25
2.13
1.91
2.27
Reference
(1)
(1)
(1)
(1)
(1)
(2)
(2)
Coarse Mode
Volume
(/mi3 cm3)
5.0
25.9
27.4
30.8
42.7
Dgv
6.0
6.04
4.51
5.7
6.0
ffg
2.0
2.16
2.12
2.25
2.13
Reference
(1)
(1)
(1)
(1)
(1)
n
-------
1 3.2.2 Particle Formation and Growth
2 A significant portion of the fine atmospheric aerosol is secondary, it is material added
3 to the particle phase as the result of gas-to-particle conversion processes. For example, fine-
4 particle sulfate and nitrate particles are mostly formed by secondary processes. One
5 mechanism of gas-to-particle conversion is homogeneous gas-phase chemical reactions to
6 form a condensible species, such as the oxidation of sulfur dioxide to form sulfuric acid.
7 Condensible species can either nucleate to form a new particle (nucleation), or can condense
8 onto the surface of an existing particle (condensation). Another important class of gas-to-
9 particle conversion mechanisms is heterogeneous chemical reactions, which are chemical
10 reactions involving both gas-phase and particle-phase constituents. Transformation on the
11 surface of particles, such as the uptake of nitric acid on the surface of sodium chloride (sea
12 salt) particles to produce nitrate is one type of heterogeneous reaction. Chemical reactions
13 within aerosol and cloud droplets, such as when sulfur dioxide dissolves within an aqueous
14 droplet and is subsequently oxidized to sulfate, are another important heterogeneous gas-to-
15 particle mechanism. Heterogeneous reactions lead to addition of aerosol material to existing
16 particles. Nucleation results in an increase in particle number as well as an increase in
17 particle mass. In this section we consider the physical aspects of these gas-to-particle
18 conversion mechanisms, and their effects on the particle size distribution.
19
20 3.2.2.1 Equilibrium Vapor Pressures
21 An important parameter in particle nucleation and in particle growth by condensation is
22 the saturation ratio S, defined as the ratio of the partial pressure of a species, p, to its%
23 equilibrium vapor pressure above a flat surface, p0: S = p/p0. For either condensation or
24 nucleation to occur, the species vapor pressure must exceed its equilibrium vapor pressure.
25 For particles, the equilibrium vapor pressure is not the same as p0. Two effects are
26 important: (1) the Kelvin effect, which is an increase in the equilibrium vapor pressure
27 above the surface due to its curvature; thus very small particles have higher vapor pressures
28 and will not be stable to evaporation until they attain a critical size and (2) the solute effect,
29 which is a decrease in the equilibrium vapor pressure due to the presence of other
30 compounds.
April 1995 3-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 For an aqueous solution of a nonvolatile salt, the presence of the salt decreases the
2 equilibrium vapor pressure of the drop. This effect is in the opposite direction as the Kelvin
3 effect, which increases the equilibrium vapor pressure above a droplet because of its
4 curvature.
5
6 3.2.2.2 New Particle Formation
7 When the vapor concentration of a species exceeds its equilibrium concentration
8 (expressed as its equilibrium vapor pressure), it is considered condensible. Condensible
9 species can either condense on the surface of existing particles or can form new particles.
10 The relative importance of nucleation versus condensation depends on the rate of formation
11 of the condensible species and on the surface area of existing particles. An analytical
12 relation for the relative importance of each pathway is dependent on the ratio of the square of
13 the available surface area to the rate of formation (McMurry and Friedlander, 1979). In
14 urban environments, it was found that new particle formation is found only near sources of
15 nuclei such as freeways because the available surface area is sufficient to rapidly scavenge
16 the newly formed condensible species. Wilson et al. (1977) report observations of nuclei
17 mode in traffic. New particle formation can also be observed in cleaner, remote regions.
18 Bursts of new particle formation in the atmosphere under clean conditions correspond to low
19 aerosol surface area concentrations (Covert et al., 1992). The highest concentration of
20 volatile ultrafine particles occur in regions corresponding to the lowest particle mass
21 concentrations, indicating that new particle formation is inversely related to the available
22 aerosol surface area Clarke (1992). In contrast to continental aerosols where sulfate
23 formation is the result of conversion of sulfur dioxide, the sulfur particles over the oceans
24 are formed from the oxidation of dimethylsulfide emitted by phytoplankton (Charlson et al.,
25 1987).
26
27 3.2.2.3 Particle Growth
28 When material is added to the particle phase by condensation or by heterogeneous
29 chemical reactions, particles of different sizes may grow at different rates, depending on the
30 mechanism involved. Condensational growth can have a different effect on the size
31 distribution of the aerosol than the effect of heterogeneous conversion through chemical
April 1995 3_23 DRAFT-DO NOT QUOTE OR CITE
-------
1 reactions within a droplet. The relative rates at which the size of particles change depends
2 on whether the rate-limiting step in the growth process is transport to the particle, chemical
3 reactions at the surface of the particle, or chemical reactions within the particle. These are
4 referred to as transport-limited, surface-reaction rate-limited or volume-reaction rate-limited.
5 These different physical mechanisms give rise to a different form of the growth law for the
6 particle. Growth laws are the expressions for dv/dt or dD /dt as a function of particle size
7 (where v is single particle volume and Dp is particle diameter).
8 For condensational growth, the rate-limiting step relevant to the rate at which particles
9 of different size grow is transport of condensible species to the particle surface. For
10 particles much smaller than the mean free path of air, transport is governed by single
11 molecular bombardment of the surface, and the volume (or mass) of these particles grows in
12 proportion to their surface area. For particles larger than the mean free path, transport is
13 governed by diffusion. In this regime the loss of diffusing species at the surface of the
14 particle causes a gradient in the concentration of the diffusing species near the surface of the
15 particle such that the volume of the particle grows in proportion to particle diameter rather
16 than surface area.
17
18 3.2.2.4 Resuspension
19 The resuspension of deposited material as well as the suspension of material which has
20 not been previously airborne can be an important source of particulate contamination. This
21 discussion will use "resuspension" to include both resuspension and suspension. Surface
22 contamination may result from the atmospheric deposition of a number of materials; for some
23 of these (e.g., plutonium), resuspension has been considered to be the most important
24 exposure pathway. Likewise, resuspended soil particles have the greatest atmospheric mass
25 over continents of any single particle type (Peterson and Junge, 1971). Despite this
26 importance, the literature shows relatively few experimental or theoretical studies for the
27 resuspension mechanism compared to other aerosol generation mechanisms. The following
28 summarizes work on the physics of resuspension, physical/chemical properties of
29 resuspension generated particles, and levels of production and transport of resuspended
30 particles.
31
April 1995 3_24 DRAFT-DO NOT QUOTE OR CITE
-------
1 Resuspension Mechanics
2 Resuspension studies may be divided into applied research and detailed studies of
3 mechanisms. Applied studies are usually motivated by atmospheric deposition of hazardous
4 substance (i.e., radionuclides from the Chernobyl 1982 accident [Cambray, 1989]) and the
5 need to predict the spreading of contamination and the lifetime of hazardous air
6 concentrations. Resuspension experiments have been conducted over a wide range of surface
7 types. Many experiments have been conducted in dry or arid regions, simply because many
8 contamination events have occurred in such locations (i.e., the Nevada Test Site). Of the
9 experiments conducted over vegetation, most have been related to short grass. Alternately,
10 applied studies may be motivated by mitigation efforts for soil erosion by wind or by need
11 for measurement of high atmospheric paniculate concentrations caused by resuspension, so-
12 called "fugitive dust". Experiments concerning wind erosion have largely occurred in
13 locations where wind erosion is prevalent, i.e., in the "Dust Bowl" area of the central United
14 States).
15
16 Applied Studies
17 Resuspension can occur due to the action of wind or by mechanical stresses. Applied
18 research considers resuspension factors, K (air concentration divided by surface
19 concentration) (units of length"1) and resuspension rates (flux of contaminant divided by
20 surface concentration) (unit of time"1). Mechanical stresses, such as disturbances by traffic
21 or agricultural operations, might result in large amounts of resuspension over short intervals
22 in specific localities. For example, Sehmel (1984) quotes K values of 4 X 10 m"1 (for
23 beryllium particles by vigorous sweeping in an unventilated room) to 7 x 10"3 m"1 for
24 plutonium particles in extensive traffic at the Nevada Test Site to 3 x 10"7 m"1 for gamma-
25 radioactive-fallout by walking on the deposit in an Australian desert.
26 Wind generated resuspension is considered to be of major importance because it can be
27 relatively continuous and can occur over large regions. Resuspension has been found to
28 increase as a power of wind speed (with the resuspension rate being related to the second or
29 third power of wind speed). Examples of resuspension factors from wind stresses quoted by
30 (Sehmel, 1984) range from 3 x 10"4 m"1 for uranium at Maralinga trials to 9 x 10"11 m"1 for
31 yttrium chloride on a cleared, sandy soil. Part of the range of K's quoted above might be
April 1995 3_25 DRAFT-DO NOT QUOTE OR CITE
-------
1 caused by the aging of deposits, although a lack of understanding of the mechanisms
2 dominant in the resuspension process has precluded identifying any reasons for the wide
3 range of results.
4 Nicholson's (1993) data verify previous work, giving an approximate I/time decrease of
5 the resuspension rate. Makhon'ko's (1986) data for resuspension from grass suggest a
6 relationship between relative resuspension rate K' versus phytomass m in grams per square
7 meter,
8
9 K' = 2.9 x 1(T8 m-1-4 [sec'1]. (3-4)
10
11 Aerodynamic Resuspension
12 Aerodynamic Models include (1) balance of forces models and (2) statistical
13 mechanisms. Balance of forces models account for forces holding the particles to the
14 surfaces versus those forces acting to remove the particles from the surfaces. Experimental
15 studies of particle motions show that particles being entrained into a turbulent fluid tend to
16 move vertically into the stream with unsteady motions (Sutherland, 1967). Braaten et al.
17 (1990) and Braaten and Paw U (1992) stressed the importance of bursts of a sweeping eddy
18 having the characteristics of large shear stress near the wall where particles are sparsely
19 deposited, breaking up the viscous sublayer and transporting fluid forces to the particles.
20 This mechanism removes particles from a surface in short bursts followed by periods of little
21 resuspension activity. Observations of Lycopodium spores placed on the flat floor of a wind
22 tunnel were used to verify the model.
23 Reeks et al. (1988) proposed a different aerodynamic mechanism that would account for
24 sudden random injections of particles into the air, the injections taking place more randomly
25 in time than in the above force balance model. Their mechanism calls for the individual
26 particles to accumulate energy from the turbulent stream (most efficiently at a resonant
27 frequency for the particle). Accumulation of energy takes place because energy dissipation is
28 limited by the local fluid and substrate. Once sufficient energy has accumulated to overcome
29 the potential energy well holding it in place, the particle is resuspended. Slow motion
30 movies of saltating sand surfaces showed such a vibrating motion of a particle before it
31 becomes airborne (Willetts, 1992).
April 1995 3_26 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mechanical Resuspension
2 The importance of mechanical disturbance is seen in the differences of resuspension
3 factors given by Sehmel (1984) for mechanical activities over contaminated soil versus those
4 for wind. Another example of the comparison of resuspension by mechanical disturbance
5 with resuspension by the wind was given by Garland (1979) as a two-order-of-magnitude
6 increase of the resuspension factor for the mechanical disturbance of a full 5-liter bottle
7 dragged along the grass 20 times in 5 minutes in wind compared to the 10 m/s wind alone.
8 Sehmel (1984) conducted experiments to determine the fraction of tracer particles
9 resuspended by driving cars and trucks through the deposited tracer or near the deposited
10 tracer. The fraction increases with speed and size of vehicle. The fraction resuspended per
11 vehicle pass increased as the first power of vehicle speed for the truck driven through the
12 tracer, the fourth power of vehicle speed for a car driven through the tracer, and the third
13 power of the vehicle speed for a car driven near the tracer.
14 The emission of PM-10 particles in wind erosion is driven by the mechanical process of
15 sandblasting, although Shinn et al. (1983) have pointed out the importance of direct
16 aerodynamic emission for low emission rates below erosion threshold. Threshold velocities
17 for particles smaller than 10 micrometer diameter are several times greater than that for 100
18 micrometer particles (Bagnold, 1941). Nonetheless, one observes submicrometer to 10-
19 micrometer particles in wind erosion events for winds very much below the threshold
20 velocity for the above mentioned particles. Gillette and Walker (1977) interpreted this to be
21 caused by the mechanical suspension (sandblasting) of fine particles by more-easily-eroded
22 sand particles. Shao et al. (1993) showed that sand-grain bombardment (saltation) is the
23 overwhelmingly dominant mechanism in maintaining fine particle emissions from the surface.
24 To derive an expression for the emission of dust, Shao et al. (1993) assumed that the number
25 of dust particles dislodged from a surface per sand grain impact was proportional to the ratio
26 for the kinetic energy loss of the impacting sand grain to the binding potential energy holding
27 a dust particle to the surface. This assumption led to the prediction that the dust flux is
28 proportional to the sand grain mass flux, which was in turn proportional to the friction
29 velocity cubed. Dust emission is highly sporadic. After the wind stress threshold is
30 surpassed, the vertical flux increases with the third power of friction velocity.
31
April 1995 3_27 DRAFT-DO NOT QUOTE OR CITE
-------
1 Physical and Chemical Properties of Resuspended Particles
2 The physical and chemical properties of resuspended particles depend partly on the
3 properties of the particles that were deposited on the surface in the initial stage of
4 resuspension. But,
5 "the deposited particles probably lose their individual identity by becoming
6 attached to host (soil) particles. When the pollutant particle is transported
7 downwind, it is usually attached (aggregated) to this host particle" (Sehmel,
8 1978).
9 Furthermore, the host particle is most likely an aggregate itself. Studies of the cross section
10 of particles, mineralogy, and scanning electron microscope analysis of dust samples show
11 that particles suspended from the soil are aggregated. For these reasons, this section
12 describes physical properties of the aggregated (host plus pollutant) particles.
13 The size distribution of resuspended soil particles may be described as lognormal
14 bimodal with one mode at 2 to 5 micrometers and another mode at 30 to 60 micrometers
15 (Sviridenkov et al., 1993; Patterson and Gillette, 1977; Gillette and Nagamoto, 1993;
16 Gillette, 1974). Because the mass mode of the distribution for particles smaller than 10
17 micrometers is roughly at 2.5 micrometers, a rough approximation is that half the PM10
18 mass is smaller than 2.5 micrometers and half is larger. The parameter
19
^ <0.1 (3-5)
u*
20
21 defines the upper size of suspended dust, where vsed is the sedimentation velocity of the
22 upper size limit, and u* is friction velocity. Data from Pinnick (1985) shows that very
23 similar size distributions result from resuspension by traffic.
24 Mineralogically (chemically) the dust consists of (in order of the most abundant) for
25 particles 1 to 10 micrometers: quartz, mica, kaolinite, mixed layer phyllosilicates and
26 feldspars. For particles smaller than 1 micrometers: mica, kaolinite, quartz, and mixed layer
27 phyllosilicates (Gillette et al., 1978). Studies of elemental composition show that
28 composition of the resuspended material compared to that of the total sediment is enriched in
April 1995 3-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 elements associated with the smallest particles (i.e., titanium) and impoverished in elements
2 associated with the coarsest materials (i.e., silicon).
3
4 Levels of Production and Transport of Resuspended Aerosols
5 Airborne dust measurements in the southern and central Great Plains states of the
6 United States were made in the early 1970's. The total mass of dust produced by individual
7 dust storms was 300,000 to 500,000 million Tg (Gillette et al., 1978). Individual dust storm
8 production rates may be compared to the global production rate estimated by d'Alameida
9 (1989) of 1,800 to 2,000 Tg per year. The Great Plains study, part of a severe storm study,
10 showed that the dust storms were typically associated with vigorous frontal activity, and that
11 the dust travels great distances (many 100's of km) as tracked by jet aircraft. Estimates of
12 transport distance for dust of well over 1,000 km (from West Texas dust storms to deposition
13 sites in northern Minnesota) were supported by isentropic trajectories, positions of rainclouds
14 and elevated concentrations of calcium in collections of rainwater in the National Acid
15 Deposition Program/National Trends Network. Even greater transport distances of
16 resuspended dust are shown by oxygen isotopic 18 to 16 ratios (618) in quartz (parts per
17 thousand). By matching the S18 value for deposited quartz and source areas for the quartz
18 (wind credible soils) the following long-range transport paths were found: Asian deserts to
19 Hawaii; Sahara desert to the Caribbean, South America, and Florida; and U.S. sources to
20 Greenland and northern Europe (Jackson et al., 1973).
21 A model developed for national acid rain and decertification/paleoclimate studies
22 (Gillette and Passi, 1988) expressed the emission of dust for a given study area as an integral
23 over friction velocity (expressing the forcing function), and the threshold friction velocity
24 (expressing the resistance of the soil and environment to ablation). Results from the model
25 for the contiguous United States (Figure 3-7) show a strong agreement of the model dust
26 emissions with known dusty areas (Gillette and Hanson, 1989). Predicted alkaline emissions
27 also agree in many respects with observed wet deposition patterns of alkaline elements
28 (Gillette et al., 1993). A considerable fraction of wind emitted dust is from dust devils
29 (Gillette and Sinclair, 1989).
April 1995 3-29 DRAFT-DO NOT QUOTE OR CITE
-------
10 15 20 25 30 35 40 45 50 55 60
45
>10'3g/cm2
10-4-1Q-3g/cm2
0 5 10 15 20 25 30 35 40 45 50 55 60
Figure 3-7. Model dust emissions for the United States.
Source: Gillette and Manson, 1989.
1 3.2.3 Particle Removal Mechanisms and Deposition
2 Particles in the air are in constant motion. They are subject to Brownian motion, which
3 is the constant random movement along an irregular path caused by the bombardment by
4 surrounding air molecules. This process is most important for small particles, and is related
5 to the particle diffusion coefficient. Particles are also subject to the earth's gravitational
6 force, as characterized by a sedimentation velocity. Gravitational settling is most important
7 for larger particles. Both of these processes involve the motion of the particle relative to its
8 surrounding air medium.
9 Brownian diffusion is important for small particles, whereas gravitational settling is
10 important for large ones. During a time period of 1 s a 0.1 pun particle will travel a distance
11 of about 40 pirn from Brownian motion, while it will fall about 1 pun due to gravity. In the
12 same 1 s time period a 1 jum particle will diffuse about 8 jum and will fall 35 pirn. Note that
13 the diffusion constant is directly proportional to the particle mobility B, while the settling
April 1995
3-30
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
velocity depends on the product of particle mass and mobility, mB. Diffusion constants and
settling velocities are plotted in Figure 3-8.
0.01
Figure 3-8. Diffusion constants and settling velocities for particles.
1 The deposition of particles in the atmosphere is not easily modeled, and is characterized
2 by a deposition velocity, which is defined as the ratio of the flux of particles to the surface to
3 the ambient concentration. Results from wind tunnel studies, shown in Figure 3-9, show
4 characteristic minima. Small particles are collected by diffusion, larger particles are
5 collected by impaction and sedimentation. Deposition models which account for these
6 mechanisms are given by Sehmel (1982), Fernandez de la Mora and Friedlander (1982) and
7 Fernandez de la Mora (1986). Atmospheric data from Lin et al. (1994), shown in
8 Figure 3-10 show that inertial mechanisms, as well as sedimentation, are important for the
9 deposition of large particles.
April 1995
3-31
DRAFT-DO NOT QUOTE OR CITE
-------
10
1
*
10"
10
Grass about 10cm high
(Chamberlain, Clough, Little)
Filter paper (Clough)
Smooth surface (Sehmel)
10~1 1 10
Particle Diameter (urn)
Figure 3-9. Particle deposition from wind tunnel studies.
10'
•a
lio-*
i
2 10"
x
£
10-
10
o Flux • Mass
7 Calculated deposition velocity
101
10'
10'3
0.1
1 10
Particle diameter, urn
100
I
EE
8
o>
O
Figure 3-10. Sedimentation and inertia effects on large particle deposition.
April 1995
3-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 This section examines the present state of knowledge regarding the loading, size
2 distribution, and chemical composition of the fine particle component of tropospheric
3 aerosols and the processes that govern these properties. Reasons why it is necessary to know
4 the physical properties and composition of these aerosols include the following:
5 1. Identification of the processes and sources responsible for the aerosol.
6 2. Development of a data base of measurements to be used in setting standards
7 3. Relating aerosol loadings and composition to putative deleterious consequences,
8 e.g. epidemiological studies.
9 4. Development of models whose objective is to relate aerosol loading and chemical
10 and physical properties to sources of the aerosols and their precursors and
11 evaluation the performance of such models.
12
13
14 3.3 CHEMISTRY AND CHEMICAL COMPOSITION
15 3.3.1 Fine Particle Chemistry
16 3.3.1.1 Acid Aerosols and Paniculate Sulfates
17 Sulfuric acid and its neutralization products with ammonia constitute a major
18 anthropogenic contribution to fine particle aerosol. This section reviews recent advances in
19 understanding of the sources, removal processes, loadings and properties of tropospheric
20 sulfate aerosols. Emphasis is given to properties and processes pertinent to these aerosols in
21 regions influenced by anthropogenic emissions as distinguished from remote locations
22 influenced primarily by natural sources.
23
24 Sources
25 Aerosol sulfate in the troposphere consists of particles emitted directly from sources
26 (primary sulfate) and of sulfate formed by atmospheric oxidation of gaseous sulfur
27 compounds, mainly SO2. Knowledge of the sources of this particulate material is important
28 to understanding the processes responsible for observed loading, composition, and size
29 distribution of the material and to developing effective methods to control its concentration.
30 Principal sources of ambient sulfate may be distinguished into primary emissions (that is
31 material emitted into the atmosphere as particulate sulfate or as gas-phase SO3 and/or H2SO4,
April 1995 3.33 DRAFT-DO NOT QUOTE OR CITE
-------
1 both of which readily form condensed-phase species) and gas-to-particle conversion in the
2 atmosphere that produces SO3 and/or H2SO4.
3
4 Atmospheric oxidation of SO2
5 Atmospheric oxidation of SO2 takes place both by gas-phase reaction and by aqueous-
6 phase reaction. The principal gas-phase mechanism is thought to be the OH-initiated
7 reaction. The principal aqueous-phase reactions are thought to be oxidation by H2O2 and O3.
8 Aqueous-phase reactions followed by cloud evaporation can result in formation of clear-air
9 aerosol. Evaporation can be a major production route for atmospheric sulfate aerosols. The
10 relative proportion of sulfate aerosol produced by the aqueous and gas-phase routes is not
11 well established.
12
13 Gas-phase oxidation of SO2. Gas phase oxidation of SO2 is thought to occur largely, if
14 not entirely, by a sequence of reactions initiated by the reaction of OH with SO2.
15
S02 + OH + M -» HS05 + M (3-6)
16
HSO 5 + O2 -* SO3 + HO2 (3-7)
17
(3-8)
18
19 The gaseous H2SO4 subsequently adds to existing particles or may nucleate to form new
20 particles.
21 Until recently the evidence for the occurrence of this reaction in the atmosphere has
22 relied on modeled OH concentrations and on laboratory-determined reaction rate coefficient
23 (Gleason et al., 1987) for the OH + SO2 reaction. However, recent measurements of OH
24 and H2SO4 in the atmosphere provide empirical evidence for this mechanism (Eisele and
25 Bradshaw, 1993; Eisele and Tanner, 1993). Simultaneous measurements of OH and SO2
26 allow the gas-phase reaction production rate of H2SO4 to be calculated at the time and
27 location of the measurement. Likewise, measurements of particle size distribution allow the
28 effective first-order rate coefficient for diffusive uptake of H2SO4 monomer by aerosol
April 1995 3.34 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles to be calculated, and measurement of the concentration of H2SO4 monomer allows
2 the loss rate by this mechanism to be calculated. Comparison of the calculated production
3 and loss rates of H2SO4 monomer show them to be equal, consistent with the observed steady
4 state concentration of this species. This study lends substantial confidence to the applicability
5 of the laboratory mechanism and rate to evaluation of the rate of sulfuric acid formation in
6 the ambient atmosphere. At night, however, the calculated loss rate substantially exceeded
7 the apparent production rate. This suggests an additional source of H2SO4 monomer, either
8 from some hitherto unrecognized reaction, or from release of H2SO4 from the particles back
9 to the gas phase. The investigators suggest that the explanation is the latter, in view of
10 correlation of particle concentration and H2SO4 monomer concentration.
11 The substantial progress made in the past few years in measurement of OH leads to the
12 expectation of increased confidence in models that calculate the concentration of this species
13 from local photolysis rate constants and abundances. This may be expected to lead in turn to
14 enhanced confidence in OH concentrations and sulfuric acid production rates calculated by
15 regional scale transport models.
16 The reaction of SO3 has recently been reexamined by Kolb et al. (1994), who find that
17 the reaction is second order in water vapor and propose that the reaction takes place by
18 interaction of SO3 with water vapor dimer:
19
S03 + (H2O)2 - H2SO4 + H2O (3-9)
20
21 The investigators note that it is probable that sufficient water dimer exists in the
22 atmosphere to allow the reaction to efficiently form sulfuric acid vapor. Other processes
23 may involve H2SO3 + H2O. The complex H2O x SO3 also may be involved in sulfuric
24 formation (Leopold et al., 1985).
25
26 3.3.2 Formation of Sulfates in Clouds
27 3.3.2.1 Particle Formation in Clouds
28 3.3.2.LI Introduction
29 The atmospheric aqueous phase (clouds, fogs) can be viewed as a processor of the
30 aerosol size/composition distribution (Pandis et al., 1990a,b). Precipitating clouds are well
April 1995 3.35 DRAFT-DO NOT QUOTE OR CITE
-------
1 known to be the major removal mechanism of aerosol particles from the atmosphere. At the
2 same time, the liquid droplets provide the reacting medium for aqueous-phase reactions
3 (Graedel and Weschler, 1981; Chameides and Davis, 1982; Graedel and Goldberg, 1983;
4 Jacob and Hoffmann, 1983; Munger and Hoffman, 1983; Chameides, 1984; Seigneur and
5 Saxena, 1984; Hoffman and Jacob, 1984; Fuzzi et al., 1984; Hong and Carmichael, 1986;
6 Hill et al., 1986; Jacob, 1986; Jacob et al., 1986; Johnson et al., 1987; Fuzzi et al., 1988;
7 Dlugi, 1989; Pandis and Seinfeld, 1989; Munger et al., 1990; Forkel et al., 1990; Bott,
8 1991; Joos and Baltensperger, 1991; Earth, 1994; De Valk, 1994). Several gaseous species
9 dissolve in cloudwater and react giving products that remain in the aerosol phase after the
10 cloud dissipates; for example, the dissolution of SO2, its ionization, and subsequent oxidation
11 to sulfate. These species can attract additional gaseous species, such as ammonia and water
12 into the aerosol phase and thereby increase further the aerosol mass. Therefore, aerosol
13 processing by nonprecipitating clouds represents a mechanism by which atmospheric particles
14 can grow during their residence time in the atmosphere. A detailed review of the state of
15 science in 1990 has been presented by United States National Acid Precipitation Assessment
16 Program (U.S. NAPAP) (1991).
17 A cyclical relationship between the occurrence of smog and fog in polluted areas has
18 been proposed by Munger et al. (1983) and was termed the smog-fog-smog cycle. In a
19 polluted atmosphere with high aerosol concentration, the formation of late night and early
20 morning fogs is augmented enhancing smog production, visibility reduction, and aerosol
21 sulfate the next day (Cass, 1979; Cass and Shair, 1984; Pandis et al., 1990). Processing of
22 aerosol by clouds can result in similar cyclical relationships and enhanced contribution of the
23 aerosol produced in clouds to ground-level paniculate concentrations (Altshuller, 1987).
24 This processing cycle accelerates the production of atmospheric acidity through aqueous-
25 phase reactions (Schwartz, 1989).
26
27 3.3.2.1.2 Cloud Effects On Particle Number Concentration
28 There has been a series of observations of enhanced aerosol number concentrations in
29 the vicinity of clouds (Saxena and Hendler, 1983; Hegg et al., 1990; Radke and Hobbs,
30 1991; Hegg et al., 1991). Saxena and Hendler (1983) suggested that the observed high
31 aerosol number concentrations near clouds could be due to the shattering of rapidly
April 1995 3_36 DRAFT-DO NOT QUOTE OR CITE
-------
1 evaporating droplets. Hegg et al. (1991) proposed that the high actinic radiation fluxes near
2 cloud tops could lead to high OH concentrations and nucleation of new H2SO4/H2O particles.
3 The high relative humidity areas around clouds often have total particle number
4 concentrations about twice those in the air at the same level but well removed from the cloud
5 boundaries (Radke and Hobbs, 1991). Kerminen and Wexler (1994) have demonstrated that
6 there is high nucleation probability associated with these high relative humidity areas,
7 especially near relatively clean clouds. All these speculated mechanisms for production of
8 new particles produce negligible new aerosol mass, but may influence the shape of the
9 aerosol distribution, especially in remote regions. Aqueous-phase reactions producing sulfate
10 and nitrate increase the aerosol mass, but do not influence directly the aerosol number
11 concentration. The removal of gas-phase SO2, H2SO4, and NH3, due to their transfer to
12 aqueous-phase, indirectly slows down the production of new particles in the vicinity the
13 cloud.
14
15 3.3.2.1.3 Cloud Effects On Aerosol Mass Concentration
16 Significant production of sulfate has been detected in clouds and fogs in different
17 environments (Hegg and Hobbs, 1987, 1988; Pandis and Seinfeld, 1989b; Husain et al.,
18 1991; Swozdziak and Swozdziak, 1992; Pandis et al., 1992; De Valk, 1994; Liu et al.,
19 1994). The detection of sulfate-producing reactions is often hindered by the variability of
20 cloud liquid water content and the temporal instability and spatial variability in concentrations
21 of reagents and product species (Kelly et al., 1989). The production of sulfate has also been
22 detected and investigated in laboratory clouds (Hansen et al., 1991). Aqueous-phase
23 oxidation of HSO3" by H2O2 is particularly fast, as illustrated by the mutual exclusivity of
24 SO2 and H2O2 observed in clouds (Daum et al., 1984; 1987). Other reactions, including
25 oxidation of dissolved SO2 by ozone and oxidation by O2 catalyzed by Fe3+ and Mn2+ may
26 also contribute, significantly in some cases, to sulfate production (Pandis et al., 1989; Earth
27 et al., 1992; Earth, 1994). During aqueous-phase sulfate production the reactants including
28 SO2, H2O2, O3, and OH are transferred from the gas-phase to the cloud droplets. This
29 transport includes a series of steps (gas-phase diffusion, transport across the gas-liquid
30 interface, dissociation and aqueous-phase diffusion) that ultimately couple the gas and
April 1995 3.37 DRAFT-DO NOT QUOTE OR CITE
-------
1 aqueous phases and in some cases control the overall sulfate production rate (Schwartz,
2 1988).
3 Hydrogen peroxide is the most important oxidant for the conversion of SO2 in cloud
4 water at pH 4.5 or lower (Calvert et al., 1985) and dominates the aqueous sulfate formation
5 pathways (McHenry and Dennis, 1994) in the northeastern United States. The measured
6 H2O2 gas-phase mixing ratio over the northeastern and central United States has been
7 reported to vary from 0.2 to 6.7 ppb (Sakugawa et al., 1990) with the highest values during
8 the summer and the lowest during the winter months. The H2O2 concentrations usually
9 increase with decreasing latitude and increasing altitude (Sakugawa et al., 1990). The
10 availability of hydrogen peroxide is often limiting the sulfate formation in clouds. This
11 limitation is more pronounced near SO2 sources and during the winter months. The seasonal
12 contribution of clouds to sulfate levels depends on both the availability of oxidants and on the
13 cloud cover. In cases where the sulfate cloud production is oxidant limited, changes in
14 aerosol sulfate levels will be less than proportional to SO2 emission changes, with the
15 relationship being more nonlinear in winter than in spring or summer (U.S. NAPAP, 1991).
16 Evaluations of the rate of the SO2-H2O2 reaction in cloudwater indicate that the
17 characteristic time for this reaction is a few minutes to an hour, depending on conditions
18 (Schwartz, 1984; Meagher et al., 1990). Since such a reaction time is shorter than the
19 lifetime of stratiform clouds in the troposphere it is anticipated that the reaction of SO2 and
20 H2O2 will proceed to completion in liquid water stratiform clouds. Evidence of this
21 occurring would be that only one or the other of these species would be present in such
22 clouds, but not both at the same time. This expectation has been borne out in field
23 measurements supporting the inference of rapid reaction given by the model estimates.
24 Daum and colleagues (Daum et al, 1984; Daum, 1988) have presented results of
25 simultaneous aircraft measurements of H202 in collected cloudwater samples and SO2 in air
26 (filter pack measurements) in nonprecipitating stratiform clouds indicating that in almost all
27 instances either one or the other species was at very low concentrations, and by inference
28 that the reaction has proceeded essentially to completion in the clouds. A rather different set
29 of results was reported by Husain et al. (1991) who conducted measurements of gas-phase
30 SO2 and H2O2 during cloud events at Whiteface Mountain, NY. Although a general negative
April 1995 3-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 correlation between the two species concentrations was exhibited, the data indicated
2 substantial periods of apparent coexistence of these species.
3 There is the possibility of spatial inhomogeneities in the clouds that are not resolved in
4 the sampling period (typically 30 min in the Daum studies; an hour or more for the Husain
5 studies), in which one region was H2O2 rich and another SO2 rich. In such instances a lack
6 of coexistence of the two species would be masked by the extended duration of sampling.
7 Such spatial inhomogeneities might also account for the few instances reported by Daum in
8 which SO2 and H2O2 apparently coexisted in clouds. Additionally, local patches of
9 subsaturated air in the clouds during the sampling period might also account for these
10 observations, although Daum took efforts to exclude such instances from their data base.
11 Yet another possible explanation of the Husain results is that the cloud was relatively newly
12 formed, and the material had not had time to react. An obvious improvement in this
13 approach is to measure the species, as well as cloud liquid water content, with greater time
14 resolution. Burkhard et al. (1994) have present aircraft measurements of gas-phase SO2 and
15 H2O2 during in-cloud flights; traces of liquid water content are also shown. These data
16 support a strong anticorrelation of SO2 and H2O2 in clouds on various time (location) scales,
17 with numerous instances of peaks of S02 coincident with valleys of H2O2 and vice versa.
18 A quantitative estimate of the amount of cloudwater sulfate that is formed by in-cloud
19 reaction can be gained by inferring the amount of cloudwater sulfate that derives from
20 preexisting sulfate aerosol. Husain et al. (1991) has used selenium as a tracer to allow such
21 inferences to be drawn. By measuring the sulfate to selenium ratio in clear air aerosol that is
22 representative of the aerosol that is the pre-cloud aerosol of the clouds under investigation,
23 and by assuming that the fractional incorporation of the sulfate and selenium into cloudwater
24 is identical (and/or by measuring this ratio), it is possible to infer the amount of cloudwater
25 sulfate derived from preexisting sulfate aerosol and by difference, the amount formed by in-
26 cloud reaction. A series of such studies carried out at Whiteface Mountain, NY, indicates
27 that assumption of identical scavenging of sulfate and selenium is valid (1.04 + 0.29; 1.04
28 ± 0.19 in two separate cloud systems). Evidence of enhanced sulfate in cloudwater,
29 attributed to sulfate formed by in-cloud reaction, was found in five of six cloud systems
30 studied; amounts formed were consistent with ambient SO2 concentrations. Examination of
31 the pH dependence of the concentration of in-cloud produced sulfate inferred by this
April 1995 3.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 technique indicated that sulfate was produced by in-cloud reaction only at pH values below
2 4.0, consistent with oxidation by H2O2, but not with oxidation by O3.
3 Recently Snider and Vali (1994) reported studies of oxidation of SO2 in winter
4 orographic clouds in which SO2 was released and the extent of increased concentrations of
5 sulfate in cloudwater (relative to the unperturbed cloud) were compared to decreased
6 concentrations of H2O2 (sum of gaseous plus aqueous, inferred from aqueous concentrations).
7 Despite considerable scatter, the data fall fairly close to the one-to-one line, indicative of the
8 expected stoichiometry of reaction, Figure 3-11. The investigators also modeled the reaction
9 kinetics. The rate of reaction is sensitive to the liquid water content (LWC) of the cloud
10 during the time between the point of cloud condensation to the point of sampling. Since this
11 profile was not known the investigators assumed a linear profile for LWC versus time. The
12 resulting model predictions agreed closely with the extent of reaction inferred from changes
13 in H2O2 and sulfate concentrations, supporting the applicability of the model.
14 In contrast to the H2O2 reaction, oxidation of SO2 by O3 exhibits a strong pH
15 dependence. The reaction is quite rapid at high pH (~6) but is expected to greatly slow
16 down as strong acid is produced in the course of the reaction. However, if concentrations of
17 NH3 or other basic materials are sufficiently high to maintain a pH above 5, the reaction can
18 continue to proceed at a high rate.
19 Walcek et al. (1990) calculated that, during the passage of a midlatitude storm system,
20 over 65 % of the sulfate in the troposphere over the northeastern United States was formed in
21 cloud droplets via aqueous-phase chemical reactions. The same authors indicated that,
22 during a 3-day springtime period, chemical reactions in clouds occupying 1 to 2% of the
23 tropospheric volume were responsible for sulfate production comparable to the gas-phase
24 reactions throughout the entire tropospheric volume under consideration. McHenry and
25 Dennis (1994) estimated that annually more than 60% of the ambient sulfate in Central and
26 Eastern United States is produced in clouds with the non-precipitating cloud production
27 dominating over precipitating cloud production. Spatial variability of emissions and ambient
28 H2O2 concentrations induces spatial variability in the contribution of in-cloud sulfate
29 production, making it highest in the south. These conclusions are in quantitative agreement
30 with similar calculations of Dennis et al. (1994) and Karamachdani and Venkatram (1992).
April 1995 3-40 DRAFT-DO NOT QUOTE OR CITE
-------
0.25 -
0.20 -
0.15 -
m
O
>
f 0.10
0.05 -
-
0.00
L 1 | 1 1 1 1 | 1 1 1
0.10
0.05
-<
0.00
2 D94
F+
D66
_
-
.•
o
-
8F
— <
c
0.00
0.05 0.10 0.15
% . Q94 -
8B 8J
o I
84
0.05 0.10
-
•
•
™
0.20 0.25
DH,O, (PPbv), OBSERVATION
0.00 - _ _
Figure 3-11. Comparison of observed H2O2 depletions (D^O^ abscissa) and observed
sulfate yields (YSO4, ordinate). Errors associated with experiments 84, 8B,
8F, and 8J are indicated and data values from these experiments are
labeled in the inset figure. Data values corresponding to experiments 94
and 66 are also labeled. The slope of the best fit line, forced through the
origin, and calculated using only those data values indicated by circles is
1.21 (±0.13).
Source: Snider and Vali (1994).
1 Aqueous-phase oxidation in clouds is also the most important pathway for the conversion of
2 SO2 to sulfate on the global scale (Hegg, 1985; Langner and Rodhe, 1991).
3 Clouds could under some conditions also be a significant source of aerosol nitrate
4 during the night. Choularton et al. (1992) and Colvile et al. (1994) observed production of
5 around 0.5 mg m~3 of nitrate during the processing of an air parcel by a cloud. They
6 speculated that the sources of this nitrate were gaseous N2O5 and NO3.
7 Chemical heterogeneities in the droplet population affect significantly the overall sulfate
8 production rate and the produced sulfate size distribution (Seidl, 1989; Twohy et al., 1989;
9 Lin and Chameides, 1990; Pandis et al., 1990a,b; Ayers and Larson, 1990; Hegg and
April 1995
3-41
DRAFT-DO NOT QUOTE OR CITE
-------
1 Larson, 1990; Bower et al., 1991; Ogren and Charlson, 1992; Roelofs, 1992a,b; 1993;
2 Carter and Borys, 1993; Bott and Carmichael, 1993; Collett et al., 1993b). Neglecting these
3 chemical concentration differences could result in significant underestimations of the sulfate
4 production rates in some cases (Hegg and Larson, 1990; Roelofs, 1993). Ice-related
5 microphysical processes can also have a significant impact on cloud chemistry (Taylor, 1989;
6 Wang and Chang, 1993; Collett et al., 1993a).
7 Fogs in polluted environments have the potential to increase aerosol sulfate
8 concentrations but at the same time to cause reductions in the aerosol concentrations of
9 nitrate, chloride, ammonium and sodium as well as in the total aerosol mass concentration
10 (Pandis et al., 1990a). Pandis et al. (1992) calculated that more than half of the sulfate in a
11 typical Los Angeles air pollution episode was produced inside a fog layer the previous night.
12 This heterogeneously produced sulfate represented 5 to 8% of the measured PM10 mass.
13
14 3.3.2.1.4 Cloud Effects On Aerosol Size/Composition Distribution
15 Several measurements of the aerosol mass distributions in urban areas have shown that
16 two distinct modes can exist in the 0.1 to 1 /^m diameter range (Hering and Friendlander,
17 1982; McMurry and Wilson, 1983; Wall et al., 1988; John et al., 1990). These are referred
18 to as the condensation mode (approximate aerodynamic diameter 0.2 jum) and the droplet
19 mode (aerodynamic diameter around 0.7 /mi). Hering and Friedlander (1982) and John et al.
20 (1990) postulated that the larger mode could result from aqueous-phase chemistry. Meng and
21 Seinfeld (1994) proposed that growth of condensation mode particles by accretion of water
22 vapor or by gas-phase or aerosol-phase sulfate production cannot explain the existence of the
23 droplet mode. Activation of condensation mode particles, formation of cloud/fog drops
24 followed by aqueous-phase chemistry, and aqueous droplet evaporation was shown by these
25 authors to be a plausible mechanism for formation of the urban and regional aerosol droplet
26 mode. The sulfate formed during fog/cloud processing of an air mass favors the aerosol
27 particles that had access to most of the fog/cloud liquid water content, which are usually the
28 particles with dry diameters around 1 /mi (Pandis et al., 1990b). These two submicron
29 mass-distribution modes have been also observed in non-urban continental locations
30 (McMurry and Wilson, 1983; Hobbs et al., 1985; Radke et al., 1989), but the frequency of
31 their co-existence remains unknown. Thus, cloud processing of an air parcel can clearly
April 1995 3-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 impact the scattering efficiency and in general the radiative properties of the corresponding
2 aerosol (Hegg et al., 1992; Bower and Choularton, 1993).
3 The aerosol distribution is also modified during in-cloud processing by collision-
4 coalescence of droplets and impaction scavenging of aerosols (Pruppacher and Klett, 1980).
5 The aerosol scavenging by droplets is a relatively slow process, and collision coalescence
6 among droplets of different sizes causes a redistribution of aerosol mass in such a manner
7 that the main aerosol mass is associated with the main water mass (Flossmann et al., 1985).
8 The processing of the remote aerosol distribution by clouds has been clearly demonstrated in
9 a series of field studies (Frick and Hoppel, 1993). This multiple processing of remote
10 aerosol by nonprecipitating clouds results in an extra mode in the aerosol number distribution
11 (Hoppel et al., 1986; Frick and Hoppel, 1993).
12 Clouds and fogs can influence the atmospheric aerosol number and mass concentration
13 and chemical composition, the shape of the aerosol size distribution, the aerosol acididity and
14 radiative properties. These effects can be important in all environments (urban, rural and
15 remote) and all seasons. Our qualitative understanding of the aerosol-cloud interactions has
16 improved significantly, but, with few exceptions, the quantification of these effects remains
17 uncertain (Altshuller, 1987; Kelly et al., 1989; Pandis et al., 1992).
18
19 3.3.3 Aqueous-Phase Oxidation Of SO2 In Clear-Air Aerosols
20 Until recently it was thought that the low amount of liquid water associated with clear-
21 air aerosols (volume fraction on the order of 1 x 10-10 , compared to clouds, for which the
22 volume fraction is the order of 1 x 10-7) precluded significant aqueous-phase conversion of
23 SO2 in such droplets. However Sievering and colleagues have called attention to the
24 possibility of rapid rate of oxidation of SO2 by O3 in aqueous sea-salt aerosols, which are
25 buffered by the alkalinity of sea salt particles,. Indeed it appears that such a rate may
26 initially be quite rapid, 1 /jM s-1 corresponding to 8% hr-1, in the example given by
27 Sievering et al. (1991) for liquid water content 50 /xg m-3 and SO2 concentration 2 n mol m-
28 3 (mixing ratio 0.05 ppb). Despite this rapid initial rate, it would appear that the extent of
29 such oxidation may be quite limited. For the example given by Sievering et al. (1991), the
30 sea-salt sodium concentration is given as 100 n mol m-3. Based on the concentrations of
31 (HCO3~ + CO32") and Na+ in seawater (2.25 and 454 m mol kg-1, respectively), the
April 1995 3.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 alkalinity of the sea salt aerosol is expected to be 0.5 n mol m-3. Consequently, after only
2 0.25 n mol m-3 of SO2 is taken up in solution and oxidized (i.e., 12% of the initial SO2), the
3 initial alkalinity would be exhausted, and the reaction rapidly quenched.
4 Sievering et al. 1994 have presented field measurements over Lake Michigan of coarse-
5 mode sulfate (diameter 5-20 /xm), which they ascribe at least in part to oxidation of SO2 in
6 such particles derived from wind driven spray of lake water, in which the pH is maintained
7 high by alkalinity present in the lake water. Calculations were carried out for liquid water
8 volume fraction of 13 x 10-12 (13 /^g m-3). The alkalinity was inferred from the measured
9 cation minus anion difference (cations NH4 + , Mg2 + , Ca2 + ; anions SO42-, NO3-) in the
10 coarse mode, which averaged 26 neq m-3, corresponding to an aqueous alkalinity of 2 x 10-5
11 M. In the absence of mass transport limitation the rate of the aqueous-phase O3-SO2 reaction
12 was calculated to be 7 ± 3 x 10-4 M s-1; however, mass transport limitation reduced this
13 rate by a factor of 20 to 40 at pH 7. The conversion rate referred to gas-phase SC^ was
14 calculated as 0.5 to 1.7 % hr-1. The investigators concluded that this mechanism is a
15 significant contributor to the SO2 oxidation under these conditions. Again, however, concern
16 may be raised with that conclusion, namely that the indicated oxidation rate, 2 x 10 -5 M s-1
17 after taking mass transport limitation into account, would quickly produce an acidity equal to
18 the initial alkalinity, thereby quenching the reaction.
19
20 3.3.3 Physical and Chemical Considerations in Particulate Sampling and
21 Analysis
22 3.3.3.1 Semi-Volatile Organic Compounds (SOCs)
23 General
24 SOCs are defined here to be organic compounds with intermediate pure compound, sub-
25 cooled liquid vapor pressures (p^). Definition ranges vary somewhat, but SOCs can be
26 thought to include compounds with/j£ values in the range 10"2 down to 10"9 torr. For SOCs
27 and for semi-volatile inorganic materials, there are health and sampling reasons for under-
28 standing the factors controlling the relative amounts that are in the gaseous (G) and aerosol
29 particulate (P) phases. G/P partitioning of SOCs has often been considered to involve mainly
30 simple physical adsorption to particle surfaces (e.g., Junge, 1977; Yamasaki et al., 1982;
31 Pankow, 1987). However, absorptive phase partitioning must also often play some role
April 1995 3.44 DRAFT-DO NOT QUOTE OR CITE
-------
1 (Pankow, 1994) as into aerosol particles containing plant wax as well as organic carbon from
2 primary emissions (Turpin et al., 1991) and from secondary organic aerosol formation
3 (Pandis et al., 1992).
4
5 Theory
6 A useful parameterization of G/P partitioning is (Yamasaki et al., 1982; Pankow, 1991)
*, - (3-9)
8 where: K (m3 /zg"1) = partitioning constant; TSP (/jg m~3) = concentration of total
9 suspended paniculate matter; and F (ng m"3) and A (ng m"3) = the P-associated and G
10 concentrations of the compound of interest, respectively. The symbols F and A originate in
11 the common usage of a filter followed by an adsorbent to collect the P and G portions,
12 respectively. With urban particulate matter (UPM), a given SOC at a given temperature T
13 tends to exhibit similar K values from sampling event to sampling event. The fraction 0 of
14 the total compound that is on/ in the P phase is given by
A + F Kp TSP + 1
15
16 Though not yet used in practice, it may also prove useful to define Kp 10 = (F10 / PM10 ) / A
17 where PM10 (pg m"3) = concentration of particles with aerodynamic diameters smaller than
18 10 pirn, and F10 (ng m"3) = PM70-associated concentration of the compound of interest.
19 Theory (Pankow, 1994) predicts that the values of Kp for a given compound class will
20 be given by a relation of the form Kp = [C1 + C2] I p£ , where C;/p£ and C2/p£ represent
21 the adsorptive and absorptive contributions to Kp, respectively. Log Kp values measured
22 under given conditions (e.g. , I) for a compound class such as the polycyclic aromatic
23 hydrocarbons (PAHs) will thus tend to be linearly correlated with the corresponding log /?£
24 values according to log Kp = mr log p£ + br For PAHs sorbing to UPM in Osaka, Japan,
25 mr « -1.028 and br ~ -8.11 (Pankow and Bidleman, 1992). (Table 3-2 gives /?£ values for
26 several PAHs at 20 °C.) This correlation allows Kp to be predicted for a compound that is
27 within the compound class of interest, but was not examined in a given study. K for a
April 1995 3.45 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-2. VALUES OF LOG P£ FOR VARIOUS PAHS AT 20 °C
Compound log /?£ (torr)
Fluorene -2.72
Phenanthrene -3.50
Anthracene -3.53
Fluoranthene -4.54
Pyrene -4.73
Benzo [a] fluorene -5.24
Benzo[b]fluorene -5.22
Benz[a]anthracene -6.02
Chrysene -6.06
Triphenylene -6.06
Benzo [b] fluoranthene -7.12
Benzo [kjfluoranthene -7.13
Benzo[a]pyrene -7.33
Benzo[e]pyrene -7.37
1 given compound depends on T (Kelvin) according to log Kp = mpIT + bp where mp depends
2 on the enthalpy of desorption; values of the intercept bp will be similar within a given
3 compound class (Table 3-3). Increasing the relative humidity from 40 to 90% appears to
4 cause K values to decrease by a factor of about two for PAHs sorbing to UPM (Pankow et
5 al., 1993).
6 For constant K , then will increase as TSP increases. For constant TSP and T, as
7 volatility increases (i.e., as /?£ increases), then Kp and <£ will decrease. When <£ « 0, one
8 can sample just the G phase when determining the atmospheric concentration of an SOC;
9 when ~ I, one can sample just the P phase; when is between 0 and 1, one must sample
10 both phases.
April 1995 3-46 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-3. mp Values for PAHs Sorbing to UPM in Osaka, Japan. Obtained by
Fitting to a Common j-Intercept br of -18.48.
Compound mp
Phenanthrene & Anthracene 4,124
Methylphenanthrene & Methylanthracene 4,240
Fluoranthene 4,412
Pyrene 4,451
Benzo[a]fluorene & Benzo[b]fluorene 4,549
Benz[a]anthracene, Chrysene, & Triphenylene 4,836
Benzo[b]fluoranthene & Benzo[k]fluoranthene 5,180
Benzo[a]pyrene & Benzo[e]pyrene 5,301
Source: Pankow, 1991.
1 Sampling Methods and Associated Sampling Artifacts
2 Atmospheric SOCs can be determined using a filter followed by an adsorbent. These
3 collect the P and G portions, respectively. Filter types include glass fiber filters (GFFs),
4 quartz fiber filters (QFFs), and teflon membrane filters (TMFs). Adsorbent types include
5 polyurethane foam (PUF), Tenax, and XAD resins. Safe sampling volumes for G-phase
6 SOCs on Tenax and PUF can be predicted based on studies of retention volumes on these
7 adsorbents (Pankow, 1988 and 1989). Volatilization losses from particles (i.e., "blow-off")
8 can occur from a filter/adsorbent when T increases during sampling, when the general level
9 of air contamination decreases during sampling, and/or when a large pressure drop develops
10 across the filter (Zhang and McMurry, 1991). In the first case, K for a given compound
11 and the already-filtered particles will decrease, leading to desorption from the sampled P-
12 phase. In the second case, even with T constant, if A in the air being sampled decreases,
13 then desorption losses from the collected particles can occur. Volatilization is of particular
14 concern with long sampling times since large overnight T cycles and/or large changes in the
15 level of contamination are then more likely. Material volatilized from the filter will be
16 collected on the adsorbent following the filter. Adsorption gains to particles from the gas
17 phase due to decreases in T and/or increases in A during sampling is a second possible
April 1995 3.47 DRAFT-DO NOT QUOTE OR CITE
-------
1 artifact type with filter/adsorbent samplers. Adsorption to the filter from the G phase is a
2 third artifact type. In this last case, a portion of the value of A for an SOC of interest sorbs
3 directly to the filter and so incorrectly contributes to the measured value of F for the
4 compound. It is difficult to generalize regarding the magnitudes of the first two artifact
5 types. One can attempt to correct for the third artifact type through the use of a backup-
6 filter (Hart et al., 1994). For sampling of UPM in Portland, Oregon, Hart et al. (1994)
7 estimated that failure to correct for gas adsorption to the filter caused F values for PAHs to
8 be overestimated by a factor of —1.4. Correction of the G-adsorption effect through the use
9 of a backup filter is subject to two possible confounding effects: 1) the atmospheres to which
10 the front and back filters are exposed may differ, making for different G-adsorption to the
11 two filters; 2) organic compounds sorbed to a backup filter could have in part volatilized
12 from the front filter. Table 3-4 summarizes how the three artifact types act to cause
13 measured values of F, A, and to deviate from the true, volume-averaged values.
14
15
TABLE 3-4. Effects of Three Types of Artifacts on Volume-Aver aged Values of
Measured Using a Filter/Adsorbent Sampler
Artifact
Volatilization from collected particles
Adsorption to collected particles
Gas adsorption to filter itself
Artifact Effect
On A On F and 0
Too large Too small
Too small Too large
Too small Too large
1 A sampler employing a diffusion denuder may avoid some of the artifact problems of
2 filter/adsorbent samplers. Air drawn into a diffusion denuder can be stripped of G-phase
3 SOCs by a sorbent that coats the walls of the denuder: G-phase SOCs diffuse from the core
4 of the air flow toward the walls. Sorbent coatings that have been used include silicones, gas
5 chromatographic stationary phases (Krieger and Kites, 1992 and 1994), and finely divided
6 XAD resin (Gundel et al., 1994; Kamens et al., 1995). The majority of the P-phase SOCs
7 do not deposit on the walls of the denuder because aerosol particles have much smaller dif-
April 1995 3-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 fusion coefficients than do gaseous molecules. The particles exiting the denuder are collected
2 on a filter. Because the air stream flowing onto the filter has been largely stripped of G-
3 phase SOCs, some desorption of the filtered P-phase SOCs can occur, and so an adsorbent is
4 often placed after the filter to collect any such desorbed SOCs. F for a given compound is
5 taken as the sum of the amounts on the filter and the subsequent adsorbent. Analysis of the
6 denuder sorbent provides A. When the denuder sorbent cannot be analyzed (as with silicone
7 rubber), A can be determined by difference using a second, total (A + F) determination for
8 SOCs (Lane et al., 1988; Coutant et al., 1988 and 1992; and Eatough et al., 1989 and 1993).
9 Although sampling artifacts are not often discussed for denuder-based samplers, artifacts
10 cannot be assumed to be absent. In the denuder section, less than 100% efficiency for G-
11 phase collection will tend to make measured A values too small and F and 0 values too large;
12 greater than 0% efficiency for P-phase collection will tend to make measured A values too
13 large and F and <£ values too small. Turpin et al. (1993) have presented a new denuder
14 design which does not use a sorbent-coated wall. Rather, a laminar flow separator is used to
15 separate a portion of the G phase from a mixed G+P flow; collection of the G-phase
16 compounds on a sorbent like PUF allows determination of the G-phase concentrations. P-
17 phase concentrations are determined by difference. Other sampling and analysis issues are
18 discussed in Chapter 4 of this document.
19
20 3.3.4 Particulate Nitrates
21 3.3.4.1 Sources
22 By analogy to the sulfur system sources of aerosol nitrates might be distinguished into
23 primary, gas-phase, and aqueous-phase. However, as primary nitric acid emissions are
24 considered to be small, the present discussion focuses on in-situ production mechanisms in
25 the atmosphere.
26
27 3.3.4.2 Gas-phase
28 The principal mechanism for gas-phase production of nitrates is reaction of OH with
29 NO2 to form HNO3.
April 1995 3.49 DRAFT-DO NOT QUOTE OR CITE
-------
OH+NO2 + M -> HNO5 (3-10)
1
2 Here, as with SO2, the mechanism and rate of the gas-phase reaction is well established from
3 laboratory studies (see NAPAP SOST, 1990), and the principal source of uncertainty in
4 describing the reaction rate is the concentrations of the reagent species, mainly OH. As
5 noted above substantial progress has been made in the past few years in measurement of
6 OH. It may thus be expected that improved knowledge of the concentration of this species
7 will allow more confident evaluation of the rate of this reaction in specific situations and
8 ultimately in regional-scale models.
9 A second key pathway for formation of nitric acid is the reaction sequence:
10
+ O2 (3-11)
11
+* N205 (3-12)
12
N2O5+H2O(1) -* 2HNO3(aq) (3-13)
13
14
15 In addition, in daytime, photolysis of NO3 must be considered:
16
NOJ^NO + 02 (3-14)
17
18 Other reactions of NO3 and/or N2O5, for example N2O5 with aromatics (Pitts et al.,
19 1985) must also be considered. Reaction of N2O5 with liquid water appears to be rapid and
20 irreversible. Studies of the uptake of N2O5 on aqueous sulfuric droplets give mass
21 accommodation coefficients of about 0.1 (Mozurkewich and Calvert, 1988; Van Doren et al.,
22 1990; Fried et al., 1994) . Thus the overall rate and yield of this reaction can be evaluated
April 1995 3-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 from the pertinent gas-phase rate constants and the mass transfer rate constant for uptake of
2 N2O5 by aqueous aerosol or cloud droplets.
3 Recently a study has been reported which claims to provide indication of uptake of
4 nitrogen oxides to nitrate during passage of an airstream through a hill cap cloud (Colvile et
5 al. 1994). However, as noted above, claims of this sort must be viewed with caution. In
6 this particular study it was necessary to invoke corrections for entrainment and dry deposition
7 of magnitude comparable to the measured differences.
8
9 3.3.5 Water Content and Aerosol Equilibria
10 3.3.5.1 Water Content of Atmospheric Aerosols, and Its Dependence on Ambient
11 Humidity
12 Water is an important ingredient of atmospheric aerosols (AA). The water content of
13 AA and the behavior of A A with respect to changes in ambient humidity are of great
14 importance in the global water cycle, the global energy budget, and also in atmospheric
15 chemistry and optics. Understanding the relationship between AA and water has proven to
16 be a difficult problem. Most of the water associated with AA is "unbound" (Pilinis et al.,
17 1989) i.e., it can increase or decrease with ambient humidity in a non-linear manner. This
18 non-linear relationship depends on particle size and composition, indeed on size-dependent
19 composition. More recent studies have included monitoring of particle size distributions
20 (either directly, or indirectly through light scattering and use of Mie theory) and size-
21 dependent chemical composition under controlled RH (e.g., Covert and Heintzenberg, 1984;
22 Rood et al., 1985). Such studies have presented increasing evidence in favor of external
23 mixtures in particles. Covert and Heintzenberg (1984) found that size spectra of sulfur-
24 bearing species were sensitive to RH while those of EC were not, and concluded that sulfur
25 and EC are, to some extent, externally mixed. Harrison (1985) segregated the particles into
26 CCN (cloud condensation nuclei) and non-CCN fractions and measured their chemical
27 compositions. Both fractions contained sulfate, nitrate and soot, but sulfate was 15% of the
28 CCN mass and only 5.8% of the non-CCN mass. Again, this was taken as evidence of
29 external mixture to some extent. The differential mobility analyzer (DMA) has been a useful
30 tool permitting study of particle properties for monodispersed size classes. Using the DMA,
31 Covert et al. (1990) and Hering and McMurry (1991) showed that monodispersed particles
April 1995 3.51 DRAFT-DO NOT QUOTE OR CITE
-------
1 scatter varying amounts of light in a single particle optical counter, indicating different
2 refractive indices, and hence, different chemical composition. Using a Tandem
3 DMA (TDMA), McMurry and Stolzenberg (1989) showed that hygroscopic and hydrophobic
4 particles of the same size co-exist frequently in Los Angeles, again an indication of external
5 mixing.
6
7 3.3.5.2 Equilibria with water vapor
8 The principal equilibrium of concern pertinent to ambient aerosols is that with water
9 vapor. This equilibrium is important as it influences the size of the particles and in turn
10 their aerodynamic properties (important for deposition to the surface, to airways, following
11 inhalation, and to sampling instrumentation) and their light scattering properties. This
12 section reviews recent work describing this equilibrium as it pertains to ambient aerosols.
13 Briefly the interaction of particles with water vapor may be described as follows. As
14 relative humidity increases, crystalline soluble salts in aerosol particles undergo a phase
15 transition to become aqueous solution aerosols. According to the phase rule, for particles
16 consisting of a single component, this phase transition is abrupt, taking place at a relative
17 humidity that corresponds to the vapor pressure of water above the saturated solution (the
18 deliquescence point). With further increase in relative humidity the particle growth is such
19 that the vapor pressure of the solution (concentration of which decreases as additional water
20 is accreted) is maintained equal to that of the surrounding relative humidity; the particle thus
21 tends to follow the equilibrium growth curve. As relative humidity decreases, the particle
22 follows the equilibrium curve to the deliquescence point. However, rather than crystallizing
23 at the deliquescence relative humidity, the particle remains a solution (supersaturated
24 solution) to considerably lower relative humidities. Ultimately the particle abruptly loses its
25 water vapor (efflorescence), returning typically to the initial, stable crystalline form. This
26 behavior has been amply demonstrated in numerous laboratory studies (Tang and
27 Munkelwitz, 1977; Tang, 1980). Recently Tang and Munkelwitz (1994) have presented data
28 for water activity (equilibrium relative humidity) as a function of composition for several
29 sulfate salts.
30 For particles consisting of more than one component, the solid to liquid transition will
31 take place over a range of relative humidities, with an abrupt onset at the lowest
April 1995 3-52 DRAFT-DO NOT QUOTE OR CITE
-------
1 deliquescence point of the several components, and with subsequent growth as crystalline
2 material in the particle dissolves according to the phase diagram for the particular
3 multicomponent system. Under such circumstances a single particle may undergo several
4 more or less abrupt phase transitions until the soluble material is fully dissolved. At
5 decreasing relative humidity such particles tend to remain in solution to relative humidities
6 well below the several deliquescence points; such behavior has been amply demonstrated. In
7 the case of the sulfuric acid-ammonium sulfate-water system the phase diagram is fairly
8 completely worked out, but in the case of the mixed anion system with nitrate, there are
9 remaining uncertainties (Tang et al, 1978, 1981; Spann and Richardson, 1985). Spann and
10 Richardson also give the compositional dependence of the relative humidity of efflorescence.
11 For particles of composition intermediate between NH4HSO4 and (NH4)2SO4 this transition
12 occurs in the range from 40% to below 10%, indicating that for certain compositions the
13 solution cannot be dried in the atmosphere. Particles of this composition would likely be
14 present at low relative humidities in the atmosphere as supersaturated salts and exhibit
15 apparent hygroscopic rather than deliquescent behavior.
16 Evidence of the interaction of ambient aerosol particles with water vapor has been
17 obtained by several investigators. Koutrakis et al. (1989) found systematically increasing
18 aerosol mean diameter with increasing relative humidity, which they attributed to water
19 accretion on sulfates. Rood et al. (1989) examined the response of light scattering coefficient
20 of ambient aerosols to increase in temperature (effectively, reduced relative humidity) and
21 established that metastable supersaturated aerosols were essentially "ubiquitous". More
22 detailed information regarding the size dependence of hygroscopic properties has been
23 obtained examining the change in particle size of a monodisperse size cut selected with a
24 mobility analyzer, subjecting that aerosol to an increase or decrease in relative humidity, and
25 reanalyzing the size at the new humidity. Studies of this phenomenon in the Los Angeles
26 area indicate this phenomenon, but also frequently indicate the presence of externally mixed
27 aerosol, in which some of the aerosol exhibits the growth expected of soluble salts, where
28 another, apparently hydrophobic, fraction does not exhibit such growth (McMurry and
29 Stolzenburg, 1989). Such bimodal RH growth is exhibited for particles present at Hopi Point
30 Arizona (Pitchford and McMurry (1994). In the latter study the relative humidity
31 dependence of the size of the more hygroscopic fraction was found to be consistent with that
April 1995 3.53 DRAFT-DO NOT QUOTE OR CITE
-------
1 expected for sulfate salts. Such external mixtures have also been commonly observed in
2 European aerosols (Hansson and Svenningsson, 1994).
3 The time constant that characterizes the rate of exchange of water vapor between the
4 gas phase and a solution droplet is of interest relative to the rate of response of particles to
5 changes in relative humidity in the ambient environment, especially in the vicinity of
6 surfaces, and relative to changes experienced by particles following inhalation or during
7 sampling. It is generally assumed that the rate of this water exchange is rapid. The
8 characteristic time for diffusional growth in response to a change in relative humidity was
9 calculated by Pilinis et al. (1989) to be about 1 x 10'7 s. However Klystov et al. (1993)
10 noted that this estimate was erroneously low by several orders of magnitude. The latter
11 investigators examined the characteristic time for establishment of phase equilibrium in
12 response to a change in relative humidity for (NH4)2SO4 aerosol particles (dry radius 0.5
13 /xm). The characteristic time increases from ca 1 ms at 8% RH to 1.6 s at 99% RH. Above
14 99% RH the characteristic time can become much longer because of the large change in
15 droplet radius at such relative humidities. These calculations indicate that the water
16 equilibrium can be expected to be rapidly achieved in the ambient environment. A possible
17 but important exception is near 100% RH, pertinent to dry deposition of particles to
18 vegetation or to water, where the equilibrium size might not be reached in the time required
19 for the particle to traverse the diffusive layer adjacent to the surface.
20 The lability of water associated with ambient aerosol has been evidenced in
21 comparisons by Malm et al. (1994) of measured paniculate light scattering coefficient
22 obtained with an integrating nephelometer with values reconstructed from aerosol
23 composition, taking into account the relative humidity dependence of light scattering
24 coefficients of the aerosol components. The reconstructed values were found to
25 systematically exceed the measured value. However when in the reconstruction the relative
26 humidity was taken as that in the nephelometer chamber (invariably lower than ambient
27 because of heating in the chamber) the reconstruction was markedly improved.
28
29 3.3.5.3 Ammonium Nitrate Vaporization Equilibria
30 In the sulfate system the vapor pressure of H2SO4 is negligible, so that all sulfate may
31 be considered present in the particles. Also, at least for acidic sulfates (that is, not fully
April 1995 3.54 DRAFT-DO NOT QUOTE OR CITE
-------
1 neutralized) the vapor pressure of NH3 is likewise negligible. Even for fully neutral
2 (NH4)2SO4 any hydrolysis of NH4+ to form NH3 that might escape to the vapor phase is
3 suppressed by the resultant acidity. In contrast, nitrates in aerosols are distinguished from
4 sulfates because of the volatility of NO3- (as HNO3) and of NH4NO3 (as NH3 -I- HNO3).
5 The equilibrium
6
NH^N05 (s) or (aq) ?* NH5(g) + HNO/g)
7
8 is such that at ambient conditions the partial pressures of NH3 and/or HNO3 are appreciable
9 above crystalline NH4NO3 and likewise above solutions containing NH4 + and NO3- ions (of
10 not necessarily equal concentrations). It is thus necessary to consider these equilibria not just
11 for the crystalline material but also for solutions, in the latter case as a function of
12 concentration or, equivalently, water activity. Such a treatment has been given in detail by
13 Stelson and Seinfeld (1982a,b), and that study is the basis for much subsequent interpretation
14 of field measurements.
15 As an example of such a study, Harrison and Msibi (1994) show comparison of
16 measured concentration product of HNO3 and NH3 versus the equilibrium constant for the
17 reaction. Agreement is found roughly within a factor of 2 or so based on assumption of
18 equilibrium with pure NH4NO3 (crystal or solution). However when the observations were
19 stratified by RH, no strong trend of measured concentration product with RH was evidenced.
20 As noted above, the time scale of reaching this equilibrium is of interest, for example
21 as it may influence dry deposition or accommodation to changing gaseous environments, as
22 in human airways. Wexler and Seinfeld (1990) modeled the time dependence of achieving
23 this equilibrium and concluded that equilibrium is generally reached within seconds to
24 minutes for typical aerosol loadings. However they caution that at low temperatures and low
25 aerosol loadings the time constant for achieving this equilibrium can be a day or more.
26 An important implication of the high vapor pressure of ammonium nitrate (as NH3 +
27 HNO3) is that ammonia will distill from any ammonium nitrate if there is an acidic site
28 present, for example acidic sulfate that is less than fully neutralized by ammonia. As a
29 consequence ammonium nitrate aerosol is not expected to coexist with acidic aerosol. As
April 1995 3.55 DRAFT-DO NOT QUOTE OR CITE
-------
1 noted above this situation appears generally to obtain, for example in the work of Gebhart
2 et al. (1994).
3 A further consequence of this equilibrium is the influence it may exert on dry
4 deposition. Sievering et al. 1994 report high rates of deposition of paniculate nitrate 2 to 9
5 cm s-1, comparable to that of HNO3, to forest canopies, inferred from steep gradients of
6 NH4NO3 concentration with height above the canopy. They attribute this to the large
7 particle size of the nitrate, 2 to 2.5 pim mean diameter, citing calculation of Peters and Eiden
8 (1992). An alternative explanation of the observations, which does not appear to be ruled
9 out, is that the deposition is actually of HNO3; that deposition of HNO3 perturbs the
10 equilibrium of NH4NO3 with NH3 + HNO3, leading to decrease of NH4NO3 in the vicinity
11 of the surface and apparent deposition of this species.
12
13 3.3.6 Carbon-containing Particulate Matter
14 3.3.6.1 Introduction
15 The carbonaceous fraction of ambient paniculate matter consists of both elemental (EC)
16 and organic carbon (OC). Elemental carbon, also called black carbon or graphitic carbon,
17 has a chemical structure similar to impure graphite and is emitted directly into the
18 atmosphere predominantly during combustion. Organic carbon is either emitted directly by
19 sources (primary OC) or can be formed in situ by condensation of low volatility products of
20 the photooxidation of hydrocarbons (secondary OC). The primary carbonaceous aerosol
21 (sum of primary EC and OC) is traditionally called soot. Small additional quantities of
22 aerosol carbon may exist either as carbonates (e.g., CaCO3) or CO2 adsorbed onto
23 paniculate matter such as soot (Appel et al., 1989; Clarke and Karani, 1992).
24
25 3.3.6.2 Elemental Carbon
26 Elemental carbon is a strong absorber of visible radiation and is the major species
27 responsible for light absorption by atmospheric particles (Novakov, 1984; Goldberg, 1985;
28 Fmlayson-Pitts and Pitts, 1986; Japar et al., 1986; Sloane et al., 1991; Hamilton and
29 Mansfield, 1991). Elemental carbon found in atmospheric particles is a complex three
30 dimensional array of carbon with small amounts of other elements such as oxygen, nitrogen,
31 and hydrogen incorporated in its graphitic hexagonal structure (Chang et al., 1982).
April 1995 3.56 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wood-burning fireplaces and diesels are major sources of EC (Mulhbaier and Williams,
2 1982; Dasch and Cadle, 1989; Brown et al., 1989; Dod et al., 1989; Hansen and Rosen,
3 1990; Burtscher, 1992). In areas where wood burning is significant, more particulate
4 graphitic carbon is expected in winter than in summer. Tracer techniques have been
5 developed for the calculation of the source contribution to the EC concentrations, including
6 use of K as a woodsmoke tracer (Currie et al., 1994) and use of the carbon isotopic tracers
7 14C and 12C (Lewis et al., 1988; Klouda et al., 1988; Currie et al., 1989). Around 47% of
8 the EC in Detroit, 93% in Los Angeles and 30 to 60% in a rural area in Pennsylvania has
9 been attributed to motor vehicle sources (Wolff and Korsog, 1985; Pratsinis et al., 1988;
10 Keeler et al., 1990), The corresponding contribution of diesel emissions to EC
11 concentrations in Western Europe is estimated to be 70 to 90% (Hamilton and Mansfield,
12 1991). Elemental carbon was also a major constituent of the Kuwait oil fires, with
13 concentrations as high as 178 mg m"3 inside the plume (Cofer et al., 1992; Daum et al.,
14 1993; and references therein). Global emissions of EC were estimated by Penner et al.
15 (1993) to be 12.6 to 24 Tg C yr'1, while the EC emission for the US was 0.4 to 1.1 Tg yr1
16 and for the rest of North America 0.2 Tg yr"1.
17 Elemental carbon also scatters light (Appel et al., 1985) although its light scattering
18 efficiency is smaller than the efficiencies of the other aerosol principal components (Sloane
19 et al., 1991). Because EC both absorbs and scatters light, its contribution to total light
20 extinction exceeds its contribution to fine particle mass. For example, in Los Angeles, EC
21 was found to represent 8.5 to 10% of the fine particulate mass, but to account for 14 to 21%
22 of the total light extinction (Pratsinis et al., 1984). A significant fraction of the dark colored
23 fine EC particles is able to penetrate the indoor atmosphere of buildings and may constitute a
24 soiling hazard of objects like works of art (Ligocki et al., 1993).
25 The concentration of EC varies significantly depending on location and season.
26 Elemental carbon concentrations in rural and remote areas usually vary from 0.2 to 2.0 mg
27 m'3 (Wolff, 1981; Clarke et al., 1984; Goldberg, 1985; Cadle and Dash, 1988; Japar et al.,
28 1986; Shah et al., 1986; Pinnick et al., 1993) and from 1.5 to 20 mg m'3 in urban areas
29 (Wolff, 1981; Delumyea and Kalivretenos, 1987; Pratsinis et al., 1984, 1986; Grosjean,
30 1984; Heitzenberg and Winkler, 1984; Goldberg, 1985; Shah et al., 1986; Rau, 1989). The
31 concentration of EC over the remote oceans is approximately 5 to 20 ng m"3 (Clarke, 1989).
April 1995 3.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 Average EC concentration values are around 1.3 and 3.8 mg m"3 for U.S. rural and urban
2 sites respectively (Shah et al., 1986). Average PM10EC values exceeding 10 mg m~3 are
3 common for some urban locations (Chow et al., 1994). The ratio of EC to total carbon has
4 been observed to vary from 0.15 to 0.20 in rural areas, to 0.2 to 0.6 in urban areas (Wolff et
5 al., 1982; Gray et al., 1984; Grosjean, 1984; Pratsinis et al., 1984; Chow et al., 1993). The
6 annual mean of this ratio was approximately 0.4 for the Los Angeles basin in 1982 (Gray et
7 al., 1986), while this ratio in the same area decreases to 0.2 during summer midday periods
8 (Larson et al., 1989; Wolff et al., 1991). Aging of an air mass results in lowering of the EC
9 fraction of the aerosol due to its mixing with non-combustion particles or by condensation of
10 material from the gas phase (Burtscher et al., 1993).
11 The distribution of EC emitted by automobiles is unimodal with over 85 % of the mass
12 in particles smaller that 0.12 mm aerodynamic diameter (Venkataraman et al. 1994). The
13 ambient distribution of EC is bimodal with peaks in the 0.05 to 0.12 mm (mode I) and 0.5 to
14 1.0 mm (mode II) size ranges (Nunes and Pio, 1993; Venkataraman and Friedlander, 1994).
15 The creation of mode II is mainly the result of accumulation of secondary aerosol products
16 on primary aerosol particles.
17 The degree of mixing of EC particles with the rest of the aerosol components remains a
18 controversial issue. Particles emitted from spark-ignition engines have been found to consist
19 of a core of EC covered with a layer of PAHs and an outermost shell of volatile compounds
20 (Steiner et al., 1992). Ambient carbonaceous aerosol in urban areas has been found to
21 consist of aggregated spherules, with a range of carbon structures from amorphous (OC) to
22 graphitic (EC) within aggregates (Katrinak et al., 1992). These aggregates are often
23 (especially during summer months) coated with sulfates and nitrates (Katrinak et al., 1992,
24 1993). However, often sulfate and EC are externally mixed (Covert and Heintzenberg,
25 1984). Coating of EC with organic compounds may alter its hygroscopicity and its lifetime
26 in the atmosphere (Andrews and Larson, 1993). Noone et al. (1992) reported that the
27 interstitial aerosol inside urban fogs is enriched in EC, something that would tend to increase
28 its lifetime in the atmosphere with respect to other species like sulfate or OC (Nunes and
29 Pio, 1993). However, the degree of incorporation of EC in droplets is highly variable (0 to
30 80%) and its behavior appears to vary from hygroscopic to hydrophobic (Hansen and
April 1995 3-58 DRAFT-DO NOT QUOTE OR CITE
-------
1 Novakov, 1988). Our luck of understanding of these processes makes a quantitative estimate
2 of the atmospheric lifetime of EC problematic.
3 The participation of EC and soot in atmospheric chemical reactions with SO2, O3 and
4 NO2 has been the subject of a series of studies (Baldwin, 1982; Dlugi and Giinsten, 1983;
5 Akhter et al., 1985; 1985; Jassim et al., 1989; Sergides et al., 1987; Gundel et al., 1989;
6 Chughtai et al., 1991). The strong dependence of the often conflicting results of these
7 studies on the nature of the samples is inhibiting the extrapolation of their conclusions to the
8 atmosphere. Chughtai et al. (1991) reported that oxidation and hydrolysis of accessible
9 reactive sites on the soot surface result in particle solubilization and accelerated particle
10 removal from the atmosphere. DeSantis and Allegrini (1992) suggested that NO2 reactions in
11 the presence of SO2 on carbon-containing particles could be a source of HNO2 in the urban
12 environment. The reaction of soot with ozone is faster than its reaction with NO2 that is in
13 turn faster than the reaction with SO2 (Smith et al., 1989).
14
15 3.3.6.3 Organic Carbon
16 The organic component of ambient aerosol both in polluted and remote areas is a
17 complex mixture of hundreds of organic compounds (Cass et al., 1982; Seinfeld, 1986;
18 Rogge, 1993; Hahn, 1980; Simoneit and Mazurek, 1982; Zafiriou et al., 1985; Graedel,
19 1986). Compounds identified in the ambient aerosol include n-alkanes, n-alkanoic acids, n-
20 alkanals, aliphatic dicarboxylic acids, diterpenoid acids and retene, aromatic polycarboxylic
21 acids, polycyclic aromatic hydrocarbons, polycyclic aromatic ketones and quinones, steroids,
22 N-containing compounds, regular steranes, pentacyclic triterpanes, iso- and anteiso-alkanes,
23 etc. (Graedel, 1986; Mazurek et al., 1989; Hildemann et al., 1993; Rogge, 1993). OC does
24 not strongly absorb light, but its light scattering efficiency in urban hazes is similar to that of
25 nitrate and sulfate.
26 Aerosol OC measurements are often subject to sampling artifacts due to adsorption of
27 organic vapors on the filters used or evaporation of the collected mass. These sampling
28 problems are discussed in Section 3.3.3.1. Wolff et al. (1991) found that this sampling error
29 represented roughly 20% of the measured OC under urban polluted conditions. McMurry
30 and Zhang (1989) observed in ambient and smog chamber measurements that a consistently
31 large fraction of the OC (40 to 70%) was collected on the quartz filters following their
April 1995 3.59 DRAFT-DO NOT QUOTE OR CITE
-------
1 impactors. The strong possibility of sampling artifacts in the laboratory and field
2 measurements presented below, increases the uncertainty of our current knowledge about
3 aerosol OC. Most of the investigators report the OC concentration as concentration of
4 carbon. These values neglect the contribution to the aerosol mass of the other elements
5 (namely oxygen, hydrogen and nitrogen) of the organic aerosol compounds. Wolff et al.
6 (1991) suggested that the measured OC values should be multiplied by a factor of 1.5 for the
7 calculation of the total organic mass associated with the OC.
8 The concentration of OC is around 3.5 mg C m~3 in rural locations (Stevens et al.,
9 1984) and 5 to 20 mg C m"3 in polluted atmospheres (Grosjean, 1984a; Wolff et al., 1991).
10 Wolff et al. (1991) and Chow et al. (1994) summarizing measurements during the summer
11 and fall of 1987 in the Los Angeles basin, reported that OC represented on average 10 to
12 18% of the PM10 mass and 11 to 24% of the PM2 5 mass during the summer and 15 to 25%
13 of the PM10 and 16 to 25% of the PM2 5 during the fall. Wolff et al. (1991) suggested that
14 these values should be reduced by roughly 20% to correct for the sampling bias and then
15 multiplied by 1.5 to account for the non-carbon mass of the organic aerosol compounds (an
16 overall increase by roughly a factor of 1.3). In rural areas of the western U.S. paniculate
17 OC concentrations are comparable to sulfate (White and Macias, 1989). In other areas OC
18 contributes roughly 10 to 15% of the PM2 5 and PM10 mass (Stevens et al., 1984). Organic
19 compounds accumulate mainly in the submicrometer aerosol size range (Finlayson-Pitts and
20 Pitts, 1986; McMurry and Zhang, 1989) and their mass distribution is typically bimodal with
21 the first peak around diameter of 0.2 jum and the second around 1 jiim (Pickle et al., 1990;
22 Mylonas et al., 1991).
23 The contribution of the primary and secondary components of aerosol OC have been
24 difficult to quantify. The lack of a direct chemical analysis method for the identification of
25 either of these OC components has led researchers to the employment of several indirect
26 methods. These methods include the use of tracer compounds for either the primary or the
27 secondary OC (Larson et al., 1989; Turpin and Huntzicker, 1991; Turpin et al., 1991), the
28 use of models describing the emission and dispersion of primary OC (Gray, 1986; Gray et
29 al., 1986; Larson et al., 1989; Hildemann, 1990) and the use of models describing the
30 formation of secondary OC (Pilinis and Seinfeld, 1988; Pandis et al., 1992; Pandis et al.,
31 1993). The above studies concluded that the secondary OC contribution is maximized in the
April 1995 3-60 DRAFT-DO NOT QUOTE OR CITE
-------
1 early afternoon of summer days, varying from 30 to 60% of the total OC depending on
2 location. The yearly averaged contribution of secondary OC is smaller, accounting for 10 to
3 40% of the OC.
4 The interactions of the OC compounds with each other and the inorganic aerosol
5 species are poorly understood. The compounds have the potential to form organic films
6 around the inorganic and EC core of the aerosol (Gill et al., 1983). Goschnich et al. (1990)
7 provided evidence for such formation by reporting that carbon compounds and organic
8 hydrogen were enriched within the particles' outer layer, while inorganics like NH4NO3 were
9 enriched inside the particles. The presence of such films can inhibit the transport of water
10 and other inorganic components between the gas and aerosol phases (Otani and Wang, 1984;
11 Rubel and Gentry, 1984).
12
13 3.3.6.4 Primary Organic Carbon
14 Primary carbonaceous particles (OC) are produced by combustion (pyrogenic), chemical
15 (commercial products), geological (fossil fuels), and natural (biogenic) sources. The
16 complexity of the mixture molecular composition of OC is such that tracer compounds are
17 still necessary to decouple the contributions of the various sources. Rogge et al. (1991)
18 suggested that fine aerosol cholesterol could be used as a tracer for meat smoke. An
19 alternative proposed meat smoke tracer set consists of myristic acid (n-tetradecanoic acid),
20 palmitic acid (n-hexadecanoic acid), stearic acid (n-octadecanoic acid), oleic acid (cis-9-
21 octadecenoic acid), nonanal and 2-decanone (Rogge et al., 1991). Benzothiazole has been
22 used as a tracer for tire wear contributions to ambient aerosol (Kim et al., 1990; Rogge
23 et al., 1993b). Steranes and pentacyclic triterpanes (hopanes) can be used as tracer
24 compounds for the vehicular sources (Rogge et al., 1993a). The odd carbon number n-
25 alkanes ranging from C27 to C33 can serve as a molecular tracer assemblage for biogenic
26 primary OC (green, dead, and degraded plant wax material directly emitted or resuspended
27 from soil and road dust) (Mazurek and Simoneit, 1984; Simoneit, 1989; Rogge et al.,
28 1993c). The iso- and anteiso- alkanes can be used to trace the cigarette smoke contribution
29 to the outdoor atmosphere (Rogge et al., 1994),
30 Primary biogenic organic matter consists predominantly of lipids, humic and fulvic
31 acids, and often represents a major fraction of the carbonaceous aerosol mass (Duce et al.,
April 1995 3.51 DRAFT-DO NOT QUOTE OR CITE
-------
1 1983; Gagosian et al., 1987; Mazurek et al., 1989, 1991; Simoneit, 1984, 1986, 1989).
2 Mamane et al. (1990) reported that most coarse OC in the Great Lakes region is of biologic
3 origin while most fine OC is anthropogenic.
4
5 Secondary Organic Carbon
6 Secondary organic aerosol material is formed in the atmosphere by the condensation on
7 already existing particles of low vapor pressure products of the oxidation of organic gases.
8 As the hydrocarbons are oxidized in the gas-phase by species such as the hydroxyl radical
9 (OH), ozone (O3) and the nitrate radical (N03), their oxidation products accumulate in the
10 gas-phase. If the concentration of such a product is smaller than its saturation concentration,
11 the species remains mainly in the gas-phase. Small amounts of the species can be adsorbed
12 on aerosol surfaces or dissolved in the aerosol phase at this stage (Yamasaki et al., 1982;
13 Pankow, 1987; Ligocki and Pankow, 1989; Pankow and Bidleman, 1991; Pankow, 1994a, b;
14 Pandis et al., 1992). If the gas-phase concentration of a species exceeds its saturation
15 concentration, the species condenses on the available aerosol surface so that at equilibrium its
16 gas-phase concentration equals its saturation concentration. If this gas-phase concentration is
17 reduced to less than the saturation value as a result of dilution, deposition or chemical
18 reaction, the aerosol species evaporates in an effort to maintain thermodynamic equilibrium
19 (Pilinis and Seinfeld, 1988). Many volatile organic compounds (VOC) do not form aerosol
20 under atmospheric conditions due to the high vapor pressure of their products (Grosjean and
21 Seinfeld, 1989). These VOC include all alkanes with up to six carbon atoms (from methane
22 to hexane isomers), all alkenes with up to six carbon atoms (from ethylene to hexene
23 isomers), benzene and many low-molecular-weight carbonyls, chlorinated compounds and
24 oxygenated solvents.
25 Organic aerosols formed by gas-phase photochemical reactions of hydrocarbons, ozone
26 and nitrogen oxides have been identified in both urban and rural atmospheres (Grosjean,
27 1977). Most of these species are di- or poly-functionally substituted alkane derivatives.
28 These compounds include aliphatic organic nitrates (Grosjean and Friedlander, 1975),
29 dicarboxylic acids (adipic and glutaric acids) (O'Brien et al., 1975), carboxylic acids derived
30 from aromatic hydrocarbons (benzoic and phenylacetic acids), polysubstituted phenols and
31 nitroaromatics from aromatic hydrocarbons (Kawamura et al., 1985; Satsumakayashi et al.,
April 1995 3_62 DRAFT-DO NOT QUOTE OR CITE
-------
1 1989, 1990). Some species that have been identified in ambient aerosol and are believed to
2 be secondary in nature are depicted in Table 3-5. Despite the above studies, the available
3 information about the molecular composition of atmospheric secondary OC and about the
4 composition of the OC produced during the oxidation of specific hydrocarbons remains
5 incomplete. The reaction mechanisms leading to the observed products are to a great extent
6 speculative at present (Finlayson-Pitts and Pitts, 1986). Natural hydrocarbons like the
7 monoterpenes (C10H16) and isoprene (C5H8) are emitted by various types of trees and plants.
8 In the United States the biogenic hydrocarbon sources are estimated to produce 30 to 60 Mt
9 of carbon per year (isoprene and monoterpenes combined) whereas anthropogenic
10 hydrocarbon sources have been estimated to account for 27 Mt of carbon per year (Lamb et
11 al., 1987; Zimmerman, 1979; Altshuller, 1983). Laboratory investigations have indicated
12 that biogenic hydrocarbons are very reactive under typical atmospheric conditions (Arnts and
13 Gay, 1979). The aerosol forming potential of biogenic hydrocarbons has been investigated in
14 a series of smog chamber studies (Kamens et al., 1981, 1982; Hatakeyama et al., 1989;
15 1991; Pandis et al., 1991; Zhang et al., 1992) and it has been suggested that isoprene
16 photooxidation does not contribute to the production of secondary aerosol under ambient
17 conditions. On the contrary, pinenes and other monoterpenes form secondary aerosol in their
18 reactions with O3 and OH and have the potential to contribute significantly to aerosol in
19 areas with high vegetation coverage. Monoterpenes were estimated to contribute around
20 15% of the secondary organic aerosol (SOA) in urban areas with low vegetation like Los
21 Angeles, while they are expected to dominate the SOA in areas with high vegetation
22 coverage like Atlanta (Pandis et al., 1991, 1992). The chemical composition of the majority
23 of the aerosol products of the monoterpene photooxidation remains unknown or is speculative
24 (Paulson et al., 1990; Palen et al., 1992). The few products that have been identified
25 include nopinone, pinanediol, pinonic acid and 5-(l-hydroxy-l-methylethyl)-2-methyl-2-
26 cyclohexen-1-one. Several investigators have studied the SOA formation from selected
27 anthropogenic hydrocarbons. The literature data up to 1976 have been reviewed by Grosjean
28 (1977). Other studies focused on toluene and other aromatic hydrocarbons (Leone et al.,
29 1985; Stern et al., 1987; Gery et al., 1985, 1987; Izumi and Fukuyama, 1990), styrenes
30 (Izumi and Fukuyama, 1990), cyclic olefins (Hatakeyma et al., 1985, 1987; Izumi et al.,
31 1988), cresols and nitrocresols (Grosjean, 1985) and alkenes with more than six carbon
April 1995 3_63 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-5. SOME SECONDARY ORGANIC COMPOUNDS IDENTIFIED IN
AMBIENT PARTICLES IN URBAN AIR
Compound n
HOOC(CH2)nCOOH 1-8
HOOC(CH2)nCHO 3-5
HOOC(CH2)nCH2OH 3-5
HOOC(CH2)nCH2ONO or CHO(CH2)nCH2ONO2 3-5
CHO(CH2)nCHO 3-5
CHO(CH2)nCHO 3-5
HOOC(CH2)nCOONO or HOOC(CH2)nCOONO2 3-5
CHO(CH2)nCOONO 3-4
HOOC(CH2)nCOONO 3-4
HOOC(CH2)nCOONO2 4-5
HOOC(CH2)nCH2ON02 3-4
(C6H6)-(CH2)nCOOH 1-3
HOOC-(C6H6)-(CH2)n 1-3
1 atoms (Grosjean, 1984b; McMurry and Grosjean, 1985; Wang et al., 1992). Measured and
2 estimated aerosol yields from a variety of SOA precursors have been tabulated by Grosjean
3 and Seinfeld (1989) and Pandis et al. (1992).
4 The calculated contribution of the main individual secondary organic aerosol precursors
5 to the secondary organic aerosol concentration in Los Angeles on August 28, 1987 is
6 presented in Table 3-6 (Grosjean and Seinfeld, 1989; Pandis et al., 1992). Toluene, the
7 nonmethane hydrocarbon with the highest emission rate in the Los Angeles area (165 t d"1 )
8 was predicted to contribute 28% of the secondary organic aerosols. Differences were
9 attributed to sampling artifacts and calibration uncertainties during the interpretation of the
10 ambient data.
11 Grosjean (1992) calculated the daily production rates of various chemical functionalities
12 of the secondary organic aerosol formed in situ during a smog episode in Los Angeles using
13 the precursor hydrocarbon emission inventory and the results of smog chamber studies. His
April 1995 3.54 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-6. PREDICTED PER CENT CONTRIBUTION TO SECONDARY
ORGANIC AEROSOL CONCENTRATIONS AT LOS ANGELES
Species
Contribution
Grosjean and Seinfeld (1989) Pandis et al. (1992)
1
2
3
4
Aromatics
Biogenic Hydrocarbons
Alkanes
Olefins
58 65
10 16
21 15
11 4
estimates are presented in Table 3-7. These predictions were compared with the available
measurements of ambient
OC functional group relative abundances (Grosjean,
TABLE 3-7. AMOUNT OF SECONDARY AEROSOL PRODUCED IN
LOS ANGELES SMOG EPISODE ACCORDING TO FUNCTIONAL
(GROSJEAN, 1992)
1
2
3
Precursor
Alkenes
Cyclic olefins
Terpenes
Alkanes
Cycloalkanes
Aromatics
TOTAL
Pickle et al. (1990)
Aerosol produced (kg day"1)
1992).
A TYPICAL
GROUPS
Carbonyls Aliphatic Acids Nitrophenols Aliphatic Nitrate
608
62 131
295 623
243
72
3118
672 1362 3118
and Mylonas et al. (1991) argued that the SO A mass
-
9
41
121
72
-
243
size
distribution in urban areas like Los Angeles is typically bimodal with maxima in the 0.1 and
1.0 /j,m size ranges. Our
understanding of the mechanisms of creation of these
two modes
April 1995
3-65
DRAFT-DO NOT QUOTE OR CITE
-------
1 remains tentative (Pandis et al., 1993). The effect of relative humidity in the SO A
2 partitioning between gas and aerosol phases is generally not understood. Thibodeaux et al.
3 (1991) developed a theoretical model based on classical adsorption theory and predicted that
4 as air relative humidity increases (remaining less than 60%) the equilibrium secondary
5 organic carbon content on the aerosol particles decreases due to competition for adsorption
6 sites with water molecules. This theoretical result seems to be supported by the
7 littleavailable experimental information, but the necessary experimental data for the
8 incorporation of these relative humidity effects on SOA partitioning in aerosol model do not
9 exist. Knowledge of the saturation concentrations of the organic condensable species remains
10 incomplete. These concentrations are expected to vary significantly with temperature. The
11 few available relevant measurements include the saturation vapor concentrations of
12 monocarboxylic and dicarboxylic acids (Tao and McMurry, 1989) and the b-pinene aerosol
13 products (Pandis et al., 1991). The saturation vapor concentrations of condensable products
14 from the oxidation of some aromatic hydrocarbons (toluene, m-xylene, and 1,3,5-
15 trimethylbenzene) were estimated to lie in the range 3 to 30 ppt (Seinfeld et al., 1987).
16 McMurry and Grosjean (1985) estimated saturation vapor concentrations for condensable
17 products from the oxidation of 1-heptene (0.14 to 1.28 ppb), o-cresol (0.06 to 1.6 ppb) and
18 nitrocresol (1.7 to 2.2 ppb).
19
20 Polycyclic Aromatic Hydrocarbons (PAH)
21 Polycyclic aromatic hydrocarbons are formed during the incomplete combustion of
22 organic matter, for example, coal, oil, wood and gasoline fuel (National Academy of
23 Sciences, 1983; Bjorseth, 1983). Stationary sources (residential heating, industrial processes,
24 open burning, power generation) are estimated to account for roughly 80% of the annual
25 total PAH emissions in the US with the remainder produced by mobile sources (Peters et al.,
26 1981; Ramdahl et al., 1983). Mobile sources are however the major contributors in urban
27 areas (National Academy of Sciences, 1983; Freeman and Cattell, 1990). More than one
28 hundred PAH compounds have been identified in urban air. The PAH observed in the
29 atmosphere range from bicyclic species such as naphthalene, present mainly in the gas phase,
30 to PAH containing seven or more fused rings, such as coronene, which are present
31 exclusively in the aerosol phase (Finlayson-Pitts and Pitts, 1986). Intermediate PAH such as
April 1995 3-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 pyrene and athracene are distributed in both the gas and aerosol phases (see also
2 Section 3.3.3.1).
3 Measurements of the size distribution of PAH indicate that while they are found
4 exclusively in the 0.01 to 0.5 jum diameter mode of fresh combustion emissions
5 (Venkataraman et al., 1994) they exhibit a bi-modal distribution in ambient urban aerosol,
6 with an additional mode in the 0.5 to 1.0 /im diameter range (Venkataraman and Friedlander,
7 1994). The growth of nuclei-mode particles by condensation of secondary aerosol species
8 like nitrate, sulfate and secondary organic aerosol has been proposed as an explanation of this
9 distribution.
10 Polycyclic aromatic hydrocarbons adsorbed on the surfaces of combustion generated
11 particles are released into an atmosphere containing gaseous co-pollutants including O3, NO2,
12 SO2, HNO3, PAN, radicals and are exposed to sunlight. Under these conditions PAH
13 undergo chemical transformations that might lead to significant degradation and formation of
14 products more polar than the parent PAH (National Academy of Sciences, 1983). Several
15 studies have focused on the reaction rates and products of reactions of PAH adsorbed on
16 specific substrates and exposed in the dark or in the light to other pollutants. However, the
17 extrapolation of these laboratory results to real atmospheric conditions remains difficult.
18 Benzo(a)pyrene (BaP) and other PAH on a variety of aerosol substrates react with
19 gaseous NO2 and HNO3 to form mono- and dinitro-PAH (Finlayson-Pitts and Pitts, 1986).
20 The presence of HNO3 along with NO2 is necessary for PAH nitrification. The reaction rate
21 depends strongly on the nature of the aerosol substrate (Ramdahl et al., 1984), but the
22 qualitative composition of the products does not. The aerosol water is also a favorable
23 medium for heterogeneous PAH nitration reactions (Nielsen et al., 1983). Nielsen (1984)
24 proposed a reactivity classification of PAH based on chemical and spectroscopic parameters
25 (Table 3-8). The PAH nitration rate under typical urban conditions remains poorly
26 understood. Bjorseth et al. (1979) observed a lack of significant PAH reactions during their
27 transport from central to northern Europe and suggested that these reactions are slow in most
28 environments. However, this may not be the case in heavily polluted areas with high NO2
29 and HNO3 concentrations and acidic particles (Finlayson-Pitts and Pitts, 1986). Reactions of
30 fluoranthene and pyrene with NO2 in the gas phase and condensation the 2-nitro-PAH
April 1995 3_67 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-8. REACTIVITY SCALE FOR THE ELECTROPHILIC REACTIONS OF
PAH (REACTIVITY DECREASES IN THE ORDER I TO VI)
I Benzo(a)tetracene, dibenzo(a,h)pyrene, pentacene, tetracene
II Anthanthrene, anthracene, benzo(a)pyrene, cyclopenta(cd)pyrene, dibenzo(a,l)pyrene,
dibenzo(a,i)pyrene, dibenzo(a,c)tetracene, perylene
III Benz(a)anthracene, benzo(g)chrysene, benzo(ghi)perylene, dibenzo(a,e)pyrene,
picene, pyrene
IV Benzo(c)chrysene, benzo(c)phenanthrene, benzo(e)pyrene, chrysene, coronene,
dibenzanthracene, dibenzo(e, l)pyrene
V Acenaphthylene, benzofluoranthenes, fluranthene, indeno(l,2,3-cd)fluoranthene,
indeno(l,2,3-cd)pyrene, naphtalene, phenanthrene, triphenylene
VI Biphenyl
1 derivatives on the aerosol surface have been proposed as an alternative reaction pathway for
2 the production of the observed aerosol nitro-PAH (Pitts et al., 1985a).
3 Nitrogen oxide (N2O5) has been proposed as an additional nitrating agent for certain
4 PAH during nighttime (Kamens et al., 1990). Pitts et al. (1985b) exposed six PAH to N2O5
5 and proposed the following reactivity order: pyrene > fluoranthene > BaP >
6 benz(a)anthracene > perylene > chrysene. Nitro-PAH photodecompose into quinones and
7 possibly phenolic derivatives. For example 6-NO2-BaP on silica gel photolyses to the 1,6-,
8 3,6-, and 6,12- isomers of BaP quinones and a host of other oxy-PAH (Finlayson-Pitts and
9 Pitts, 1986). These reactions are expected to depend strongly on the chemical composition
10 and structure of the aerosol substrate and are not well understood for ambient particles.
11 Aerosol PAH react with O3 to produce oxidized PAH. Pyrene, BaP and athracenes
12 react rapidly and the benzofluoranthenes slowly (Finlayson-Pitts and Pitts, 1986; Alebic-
13 Juretic et al., 1990). Reaction rates of 15 to 30% hr"1 were observed for the most reactive
14 PAH adsorbed on filters during exposure to 200 ppb of O3 (Pitts et al., 1986). However,
15 other researchers (Atkinson and Aschmann, 1986; Coutant et al., 1988; De Raat et al., 1990)
16 have suggested that the PAH-O3 reaction is of negligible importance for typical atmospheric
17 conditions. Relatively little is known about the full ranges of products and the mechanisms
18 of their formation. Polycyclic aromatic hydrocarbons exposed to sunlight have been found to
April 1995 3-68 DRAFT-DO NOT QUOTE OR CITE
-------
1 photodegrade in a series of laboratory studies (Valerio and Lazzarotto, 1985; Behymer and
2 Kites, 1988). The photodegradation rates depend strongly on the chemical composition and
3 the pH of the aerosol substrate (Dlugi and Giisten, 1983; Valerio and Lazzarotto, 1985;
4 Behymer and Hites, 1988). Polycyclic aromatic hydrocarbons appear to be more stable when
5 adsorbed on ambient aerosol than when present in pure form or in solution or on artificial
6 surfaces (Baek et al., 1991). The occurrence of PAH-SOX reactions remains uncertain
7 (Baek etal., 1991).
8
9 3.3.7 Metals and Other Trace Elements
10 The major components of fine particles are sulfate, nitrate, organic and elemental
11 carbon, ammonium ions and a variety of trace elements (Godish, 1985; Pitts and Pitts,
12 1986). Trace elements that are found predominantly in the fine particle size distribution are
13 Na, Cs, Cl, Br, Cu, Zn, As, Ag, Cd, In, Sn, Sb, W, and Pb, and greater than 75% of their
14 mass is associated with particles of diameter less than 2 um. Metals which are found in both
15 fine and coarse modes are V, Cr, Mn, Fe, Co, and Se, while elements found primarily
16 within large particle distributions are Ca, Al, Ti, Sc, and La (Klee, 1984). The
17 concentrations and the relative proportions of these species in the various particle size ranges
18 depend on a number of factors such as the nature of the emissions, the photochemical activity
19 and the meteorology (Pitts and Pitts, 1986). The concentration ranges of various elements
20 associated with particulate matter in the atmosphere are shown in Table 3-6. For most
21 elements the range in concentrations is greater than three orders of magnitude. This reflects
22 the different sources and the different pollution control strategies that exist in each area.
23 This information was compiled by Schroeder et al. (1987), and includes a large number of
24 studies from the United States, and abroad, which indicates the need to complete site specific
25 evaluations for high end concentrations (references can be found in the original paper by
26 Schroeder et al., 1987).
27 In general, remote areas recorded measurable concentrations of some elements
28 associated with crustal origin, as well as some elements indicative of anthropogenic sources.
29 This supports hypotheses which suggest that long range transport occurs in these remote
30 areas (Schroeder et al., 1987). The urban data (Table 3-9) reflect elemental concentrations
31 in different parts of the world. Elements such as lead, iron, and copper are measured in
April 1995 3_69 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-9 CONCENTRATION RANGES OF VARIOUS ELEMENTS ASSOCIATED
WITH PARTICIPATE MATTER IN THE ATMOSPHERE (ng/m3)
Elements
As
Cd
Ni
Pb
V
Zn
Co
Cr
Cu
Fe
Hg
Mn
Se
Sb
Remote
0.007
0.003
0.01 -
0.007
0.001
- 1.9
- 1.1
60.0
-64
- 14
0.03 - 460
0.001
0.005
0.029
0.62-
0.005
0.01 -
0.0056
0.0008
-0.9
- 11.2
- 12
4,160
- 1.3
16.7
-0.19
- 1.19
Rural
1.0
0.4-
0.6
2- 1
2.7
11 -
0.08-
1.1
3-
-28
1,000
-78
,700
-97
403
- 10.1
-44
280
55 - 14,530
0.05
3.7
0.01
0.6-
- 160
-99
-3.0
-7.0
Urban (USA)
2 - 2,320
0.2 - 7,000
1 -328
30 - 96,270
0.4- 1,460
15 - 8,328
0.2 - 83
2.2 - 124
3 - 5,140
130 - 13,800
0.58 -458
4-488
0.2 - 30
0.5 - 171
Source: Schroeder et al., 1987
1 greatest abundance in paniculate matter from all locations, while elements such as cobalt,
2 mercury and antimony are found in the smallest quantities (Schroeder et al., 1987).
3 Potential sources of trace metals found in fine airborne particles are primarily
4 anthropogenic and include combustion of coal and oil, wood burning, waste incineration, and
5 metal smelting operations. Biomass burning which includes residential wood combustion and
6 forest fires, is another source for the release of trace elements in the atmosphere. In a
7 profile of biomass burning, zinc was the characteristic trace element present in the fine
8 particles in concentration (0.0866 ± 0.0355 %) of primary mass emitted. Other trace
9 elements present were Cl (1.9083 + 0.6396 %), K (3.9926 ± 1.2397 %) and S (0.5211 +
10 0.1761 %) (Chow et al., 1992).
11 The chemical composition of paniculate matter analyzed in New Jersey as part of the
12 Airborne Toxic Element and Organic Substances project (ATEOS), identified the trace
April 1995
3-70
DRAFT-DO NOT QUOTE OR CITE
-------
l elements Pb, Fe, Zn, V and As (Daisey, 1987; Morandi et al., 1991). The main source for
2 atmospheric lead concentration is the combustion of leaded gasoline in motor vehicles.
3 However with increased use of unleaded gasoline, levels of atmospheric lead have been
4 reduced, and other sources of lead tend now to be more significant components of the
5 residual lead. Morandi (1985) has reported evidence of contributions to airborne lead from
6 resuspended soil, oil burning and small scale smelting, which taken together accounted for
7 more than half of the airborne lead at a New Jersey site. Vanadium levels were derived
8 from oil burning for space heating and power production, while Zn is attributed to a zinc
9 smelter in the area (Daisey, 1987).
10 Road dust aerosols are analyzed for trace elements in a variety of studies (Barnard
11 et al., 1987; Barnard et al, 1988; Warren et al., 1987). Recent source apportionment studies
12 in California's South Coast Air Basin, provide additional information on trace element
13 concentrations in roadside dusts as well as in motor vehicle exhaust for particle sizes
14 < 2.5 urn (Watson et al., 1994). In addition to elemental carbon, Al, Si, K, Ca, Ti and
15 Fe were present in paved road dust in abundances which exceeded 1%. Elevated
16 concentrations of Pb and Br were detected, which illustrated the deposition from the tailpipe
17 exhaust from vehicles that burned leaded fuels (Watson et al. 1994; Chow et al., 1992).
18 Significant amounts of SO=4, Br, Cl", and Pb were detected in the motor vehicle exhaust
19 profile, though Pb levels were much lower than those reported in earlier tests (Watson et al.,
20 1994; Pierson and Brachaczek, 1983).
21 Ambient measurements of the mass and chemical composition of PM10 and PM2 5, and
22 associated source profiles have been taken through the years. Data base summaries identify
23 locations, sampling times and chemical species of data available since 1988, complementing
24 previous existing databases (Watson and Chow, 1992; Lioy et al., 1980). Size specific
25 measurements show that over 90% of the mass from geological material is in the coarse
26 particle size fraction, while the combustion related source categories contained —90% of
27 their mass concentrations in the PM2-5 size fraction (Chow et al., 1992). In a municipal
28 incinerator profile, elements in the fine particle fraction include Cu, Zn, As, Cd, Sb, Pb and
29 Ba, while trace elements in the coarse particle fraction include Ca, Cr, Mn, and Ni (Olmez
30 et al., 1988). In an oil-fired power plant, trace elements such as V, Ni, Co, Ba and Cu are
31 present in both fine and coarse particles (Olmez et al., 1988).
April 1995 3.7 j DRAFT-DO NOT QUOTE OR CITE
-------
1 Although a knowledge of the elemental and ionic composition of ambient particles is
2 necessary in order to understand their sources and chemistry, the chemical forms in which
3 important species exist are not known. For example, sulfates, nitrates and ammonium ions,
4 which are the main constituents of fine particles, may exist in forms other than simple
5 ammonium salts (Pitts and Pitts, 1986). Table 3-10 lists some compounds identified in
6 aerosols by a roadway at Argonne National Laboratory, and Table 3-11 lists compounds
7 observed in aerosols in a forested area, at State College, Pennsylvania (Tani et al., 1983).
8 However, there are uncertainties associated with the compounds shown in Tables 3-10 and
9 3-11. Tani et al. pointed out that both physical and chemical changes may occur during or
10 following impaction of aerosol particles on a collector, which would lead to the formation of
11 compounds not initially present in the ambient aerosols (Tani et al., 1983).
12
13
TABLE 3-10. COMPOUNDS OBSERVED IN AEROSOLS BY A ROADWAY AT
ARGONNE NATIONAL LABORATORY
K2Sn(S04)2
CaCO3 (NH4)2Co(SO4)2 . 6H2O
CaMg(CO3)2 (NH4)3H(SO4)2 (letovicite)
CaS04.2H20 3(NH4N03).(NH4)2S04
(NH4)2Pb(S04)2 2(NH4N03) . (NH4)2SO4
(NH4)2Ca(SO4)2.H2O NH4MgCl3.6H2O
(NH4)HSO4 NaCl
(NH4)2S04 _ (NH4)2Ni(S04)2 . 6H2O
Source: Tani et al., 1983.
TABLE 3-11. COMPOUNDS OBSERVED IN AEROSOLS IN A FORESTED AREA,
STATE COLLEGE, PA.
(NH4)2S04
(NH4)3H(SO4)2 (letovicite)
(NH4)HSO4
2(NH4N03).(NH4)2SO4
(NH4)2Pb(S04)2
Source: Tani et al., 1983
April 1995 3_72 DRAFT-DO NOT QUOTE OR CITE
-------
1 Metals such as Al, Ca, Fe, Mg and Pb known to be present in atmospheric aerosols,
2 also exist in uncertain chemical forms (Pitts and Pitts, 1986). This is partially due to the use
3 of analytical techniques that normally provide information on total metal content (Schroeder
4 et al., 1987). It is generally assumed that many of the elements, especially from combustion
5 sources, are present in the form of oxides (Olmez et al. 1988), while trace elements in
6 incinerator emissions may be in the form of chlorides (Schroeder, 1987). Data from Los
7 Angeles indicate that arsenic may be present in two chemical forms in atmospheric aerosols,
8 as arsenite and arsenate. Both forms were identified in both the fine and coarse particle
9 fractions (Rabano et al., 1989). Fe2O3, Fe3O4, A12O3, and A1P04 have been identified in
10 roadside particulate matter (Biggins and Harrison, 1980). Ca and Mg may exist in the form
11 of oxides (i.e., CaO, MgO), although in the presence of water, Stelson and Seinfeld (1981)
12 suggest that, on equilibrium considerations, CaO and MgO should react to form their
13 hydroxides, Ca(OH)2 and Mg(OH)2, respectively. Similarly the oxides Na2O and K2O
14 should form NaOH and KOH when water is present. Lead has been observed in roadside
15 particulate matter in a wide variety of forms, such as PbSO4, Pb3O4, PbSO4.(NH4)2SO4,
16 PbO.PbS04, 2PbC03.Pb(OH)2, 2PbBrCl.NH4Cl, PbBrCl, (PbO)2PbBrCl,
17 3Pb3(PO4)2.PbBrCl, and elemental lead (Biggins and Harrison, 1980; Post and Buseck,
18 1985).
19 Heterogeneous oxidation of sulphur dioxide in air can be catalyzed by species such as
20 iron, manganese (Barrie and Georgii, 1976) and cadmium, while vanadium is suspected to
21 catalyze the formation of sulfuric acid during oil combustion. Oxides of iron, manganese
22 and lead are reported to absorb SO2 (Schroeder, 1987).
23 It has been suggested that the elements arsenic, cadmium, manganese, nickel, lead,
24 antimony, selenium, vanadium and zinc volatilize at high temperatures during fossil fuel
25 combustion and condense uniformly on surfaces of entrained flyash particles as the
26 temperature falls beyond the combustion zone (Linton et al., 1976;). Accumulation of trace
27 metals in the fine fraction of airborne dust sampled in iron foundries showed Pb and Zn
28 localized on the surface of the fine particles (Michaud et al., 1993). From the viewpoint of
29 toxicity, such emissions are more important than natural sources where trace elements are
30 usually bound within the matrix of natural aerosols and thus less mobile and bioavailable
31 (Schroeder, 1987).
April 1995 3.73 DRAFT-DO NOT QUOTE OR CITE
-------
1 Trace metal compounds found in road dust can accumulate from anthropogenic or
2 natural sources. Subsequently these can become re-entrained in the atmosphere. In such
3 samples lead and zinc were found to be strongly associated with carbonate and iron-
4 manganese oxide phases, with small amounts being associated with an organic phase. Half
5 of cadmium was associated with carbonate and iron-manganese oxide phases, while copper
6 was mainly associated with the organic phase. These associations influence the relative
7 mobility and bioavailability of trace metals in the environment (Harrison et al., 1981).
8 Resuspension of particles from contaminated surfaces may also contribute to an
9 increase in the toxic trace elements in airborne particles (Kitsa et al., 1992, Kitsa and Lioy,
10 1992; Pastuszka and Kwapulinski, 1988; Falerios et al., 1992). Kitsa et al. (1992) measured
11 elemental concentrations in particles resuspended from a waste site in New Jersey. Close to
12 the resuspension source, coarse particles were dominant, but farther downwind from the site,
13 fine particles were prevailing. The fine particles were enriched in chromium and lead,
14 indicating the potential for elevated human exposure through inhalation. Chromium may
15 exist in different valence states, but the most stable and abundant are the trivalent and
16 hexavalent states. Hexavalent chromium is classified as a known respiratory carcinogen in
17 humans.
18 Oxidation of the species present in aerosols results from interaction with various
19 atmospheric oxidants, such as molecular oxygen, ozone or hydrogen peroxide. The presence
20 of oxides of As, Cd, Co, Cr, Cu, Hg, Mn, Ni, Pb, Sb, Se, V and Zn has been measured in
21 emissions of cement plants, blast furnace and sintering operations, secondary iron foundries,
22 non-ferrous smelting of arsenic-bearing ores, zinc and lead smelters and many other sources
23 (Schroeder et al., 1987).
24 Sulphation, and possibly nitration, of metallic oxides can be surmised to be an
25 important transformation as particles age. A statistical assessment of multielemental
26 measurements in a study in the rural and urban atmospheres of Arizona showed strong
27 correlations of lead, copper, cadmium and zinc with sulfates in the rural atmosphere and
28 moderate correlation of lead and copper with sulfates and nitrates in urban atmosphere
29 (Moyers et al., 1977). Nickel has also great affinity for sulfur which may lead to the
30 emission of nickel sulfate containing particulates from combustion sources. In the absence of
April 1995 3-74 DRAFT-DO NOT QUOTE OR CITE
-------
1 sulfur, nickel oxides or complex metal oxides containing nickel may form (U.S.
2 Environmental Protection Agency, 1986a).
3 Lead is emitted in the air from automobiles as lead halides and as double salts with
4 ammonium halides (e.g. PbBrC1.2NH4Cl). From mines and smelters, the dominant species
5 are PbSO4, PbO.PbSO4, and PbS. In the atmosphere lead is present as sulfate with minor
6 amounts of halides. Lead sulfide is the main constituent of samples associated with ore
7 handling and fugitive dust from open mounds of ore concentrate. The major constituents
8 from sintering and blast furnace operations appeared to be PbSO4 and PbO.PbSO4
9 respectively (U.S. Environmental Protection Agency, 1986b).
10
11 3.3.8 Removal Processes
12 Removal of accumulation mode aerosol particles from the atmosphere occurs largely
13 by the precipitation process (e.g., Slinn, 1983). These particles are the dominant particles on
14 which cloud droplets form (cloud condensation nuclei, CCN); once a cloud droplet (of
15 diameter of a few up to about 20 micrometers) is formed, it is much more susceptible to
16 scavenging and removal in precipitation than is the original submicrometer particle. The
17 fraction of aerosol particles incorporated in cloud droplets on cloud formation is the subject
18 of active current research. Earlier work yielded a fairly wide spread in this fractional
19 incorporation, based in part on limitations of then existing techniques and in part on
20 definitions of incorporation efficiencies (based on number, mass, light scattering efficiency,
21 or amount of specific compounds; ten Brink et al., 1987). More recent work indicates a
22 high fractional incorporation at low concentrations of aerosol particles decreasing as the
23 aerosol particle loading increases (Leaitch et al., 1992; Gillani et al., 1992). Model
24 calculations of the efficiency of incorporation of accumulation-mode aerosol particles into
25 cloud droplets and precipitation are highly sensitive to assumptions and approach (Jensen and
26 Charlson, 1984; Flossmann et al., 1985; Hanel, 1987; Ahr et al., 1989; Albeit et al., 1990).
27 The dominance of precipitation removal processes for accumulation mode particles
28 results in high variabity in temporal patterns of aerosol loadings, that may be attributed to the
29 episodicity of precipitation events and synoptic scale meteorology that delivers air of
30 differing origins to a given location (e.g., Waldman et al,, 1990). This variability leads to
31 difficulties in attempts to estimate mean residence times based on budget considerations
April 1995 3.75 DRAFT-DO NOT QUOTE OR CITE
-------
1 (Junge, 1963; Schwartz, 1979). A unique approach to estimation of the mean residence time
2 of accumulation mode aerosol particles was presented by measurement of the decay of
3 atmospheric concentrations of Cei37 at several mid-latitude surface stations in Europe and
4 Asia in the weeks following the Chernobyl accident (Cambray et al., 1987); the Ce137 was
5 present largely in this size range. This study led to an estimate for the mean residence time
6 of 7 days , consistent with other estimates. It may be noted, however, that this residence
7 time is applicable to particles in the free troposphere, where the Ce137 was mainly present
8 during the several week period. The mean residence time of accumulation mode particles in
9 the boundary layer is expected to be somewhat less (Benkovitz et al., 1994).
10
11
12 3.4 TRANSPORT AND TRANSFORMATIONS TO SECONDARY
13 PARTICIPATE MATTER
14 3.4.1 Aqueous-Phase Chemical Equilibria and Chemical Kinetics of
15 Transformations to Secondary Particulate Matter
16 3.4.1.1 Aqueous-Phase Equilibria
17 The liquid water content of the atmosphere, WL, is usually expressed either in g of
18 water per m3 of air or as a dimensionless volume fraction L (e.g., m3 of liquid water per
19 m3 of air). Typical liquid water content values are 0.1 to 1 g m"3 (L= 10"7- 10"6) for
20 clouds, 0.05 to 0.5 g nV3 (L= 5 x 10'7 - 5 x 10"6) for fogs, and only 10'5 to 10'4g in3
21 (L= 10-n-10-10) for aerosols.
22 For dilute solutions the equilibrium distribution of a reagent gas A between the gas and
23 aqueous phases is given by Henry's law
24 [A] = H^A (3-15)
25
26 where pA is the partial pressure of A in the gas-phase, [A] is the equilibrium aqueous-phase
27 concentration of A and HA is the Henry's law coefficient for species A. The customary units
28 of //A are mole I"1 atm"1. HA can be viewed as the equilibrium constant of the reaction
29
30 A(g) £> A(aq) (3-16)
31
April 1995 3-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 Table 3-12 gives the Henry's law coefficients of some atmospheric gases in liquid water
2 at 298 K. The values given reflect only the
3 subsequent fate of the dissolved species A.
physical solubility of the gas regardless of the
Some of the species included in Table 3-12
4 dissociate after dissolution or react with water. Henry's law constants do not account for
5 these processes, and the modifications necessary will be discussed in the next paragraph.
6 Henry's law coefficients generally decrease
7 solubilities at higher temperatures (Seinfeld,
8
9
for increasing temperatures, resulting in lower
1986).
TABLE 3-12. HENRY'S LAW COEFFICIENTS OF SOME ATMOSPHERIC GASES
DISSOLVING
Species
02
NO
C2H4
NO2
03
N2O
CO2
H2S
S02
CH3ONO2
CH3C(O)O2NO2
CH302
OH
HNO2
NH3
CH3OH
CH3OOH
CH3C(O)OOH
HC1
HO2
HCOOH
HCHO
CH3COOH
H202
HN03
NO3
IN LIQUID WATER
H (M/atm) (298 K)
1.3xlQ-3
1.9xlO-3
4.8xlO-3
l.OxlQ-2
1.13xl(T2
2.5xlO-2
3.4xlQ-2
0.12
1.23
2.6
2.9
6.0
25.
49.
75.
220.
227.
473.
727.
2.0xl03
3.5xl03
6.3xl03
8.7xl03
7.45xl04
2.1xl05
2.1xl05
April 1995
3-77
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
Several gases, after dissolving in the aqueous-phase, ionize and establish an
aqueous-phase chemical equilibrium system. For example for SO2,
S02(g) * S02- H20
(3-17)
SO2- H2O ^ HSO3 + H +
(3-18)
HSO *» SO
2"
(3-19)
5
1
2
3
4
5
6
7
8
with
H
so.
[SO2- H2O]
Pso,
[SO2- H2O]
[HSO~]
(3-20)
ATsl and Ks2 are the first and second dissociation constants for SO2. It is convenient to
consider the total dissolved sulfur in oxidation state IV as a single entity and refer to it as
= [SO2- H2O] + [HSO3] =
2-,
(3-21)
The three sulfur species are in rapid equilibrium and therefore [S(IV)] changes only
when SO2 is transferred between the gas and aqueous phases. The total dissolved sulfur,
S(IV), can be expressed as a function of only the pH and the partial pressure of SO2 over the
solution by:
(3-22)
[H + ]
April 1995
3.78 DRAFT-DO NOT QUOTE OR CITE
-------
1 The above equation can be expressed in a form similar to Henry's law as
= Hs*IV)Ps02 (3-23)
2
3 where Hf(IV) is the effective (or modified) Henry's law coefficient given for S(IV) by
(3-24)
4
5 The modified Henry's law coefficient relates the total dissolved S(IV) (and not only with the
6 SO2 vapor pressure over the solution. The effective Henry's law coefficient always exceeds
7 the Henry's law coefficient, indicating that the dissociation of a species enhances its
8 solubility in the aqueous phase.
9 Several of the species that are in rapid equilibrium can be also considered as single
10 entities:
[S(IV)] = [H2S04(aq)] + [HSO4] + [SO42~]
11
[N(V)] = [HN03(aq)] + [NO~]
12
[H02T] = [HN02(aq)] + [O~]
[HCHOT] = [HCHO] + [H2C(OH)2]
13
14 Equations relating the total concentrations of these aqueous-phase species with the
15 corresponding equilibrium concentrations of the gas-phase species can be derived similarly to
16 those for S(IV).
April 1995 3.79 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.4.1.2 Aqueous-Phase Transformation of SO2 to Sulfate
2 The aqueous-phase conversion of dissolved SO2 to sulfate is thought to be the most
3 important chemical transformation in cloudwater. Dissolution of SO2 in water results in the
4 formation of three chemical species: hydrated SO2 (SO2 • H2O), the bisulfite ion (HSO^) and
5 the sulfite ion (S0>3~). At the pH range of atmospheric interest (pH =2-7) most of the S(IV)
6 is in the form of HSOJ, whereas at low pH (pH <2), all of the S(IV) occurs as SO2 • H2O.
7 At higher pH values (pH >7), (SO^~) is the preferred S(IV) state (Seinfeld, 1986). The
8 individual dissociations are fast, occurring on timescales of milliseconds or less (Martin,
9 1984; Schwartz and Freiberg, 1981; Seinfeld, 1986). Therefore, during a reaction
10 consuming one of the three species, SO2 • H2O, HSO^, or SO^", the corresponding
11 aqueous-phase equilibria are re-established instantaneously. The dissociation of dissolved
12 SO2 enhances its aqueous solubility and the total amount of dissolved S(IV) always exceeds
13 that predicted by Henry's law for SO2 alone and is quite pH dependent. The Henry's law
14 coefficient for SO2 alone, Hso , is 1.23 M atm"1 at 298 K, while for the same temperature,
15 the effective Henry's law coefficient for S(IV), HS(IV> is 16.4 M atm"1 for pH=3,
16 152 M atm"1 for pH=4 and 1,524 M atm"1 for pH=5. Equilibrium S(IV) concentrations for
17 SO2 gas-phase concentrations of 0.2-200 ppb, and over a pH range 1-6 vary approximately
18 from 0.001 to 1000 mM.
19 Several pathways for S(IV) transformation to S(VI) have been identified involving
20 reactions of S(IV) with O3, H2O2, O2 (catalyzed by Mn + and Fe3+), OH, SO5~, HSO5~
21 SO^, PAN, CH3OOH, CH3C(O)OOH, HO2, NO3, NO2, N(III), HCHO and Cl^ (Pandis
22 and Seinfeld, 1989a).
23 Although ozone reacts very slowly with SO2 in the gas phase, the aqueous-phase
24 reaction is rapid. The possible importance of O3 as an aqueous-phase oxidant for S(IV) was
25 first suggested by Penkett (1972) and the kinetics of
26
S(IV) + O3 -» S(VI) + O2 (3-25)
27
28 have been studied by several investigators (Erickson et al., 1977; Penkett et al., 1979;
29 Maahs, 1983). Hoffmann and Calvert (1985), after a detailed investigation of existing
April 1995 3-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 experimental kinetic and mechanistic data, suggested the following expression for the rate of
2 the reaction of S(IV) with dissolved ozone:
3
H20] + kjtHSOj] + k2[SO32 "])[O3] (3-26)
4
5 recommending the values kQ =2.4 x 104 M'1 s'1, k\ =3.7 x 105 M'1 sl and, k2 = l.5 x 109
6 M"1 s"1. They also proposed that this reaction proceeds by nucleophilic attack on ozone by
7 SO2 • H2O, HSO^~, and 803". An increase in the aqueous-phase pH results in an increase
8 of all three, [SO2 • H2O], [HSO^~] and [SO^"], equilibrium concentrations and therefore in
9 an increase of the overall reaction rate. For an ozone gas-phase mixing ratio of 30 ppb, the
10 reaction rate varies from less than 0.001 mM h"1 (ppb SO^"1 at pH=2 (or less than 0.01%
11 SO2 (g) h'1 (g water /m3 air)'1) to 3,000 mM IT1 (ppb SC^)'1 at pH=6 (7,000% SO2 (g) h'1
12 (g water /m3 air)"1). The gas-phase SO2 oxidation rate is of the order of 1% h"1 and
13 therefore the S(IV) heterogeneous oxidation by ozone is significant for pH values greater
14 than 4. The strong positive dependence of the reaction rate on the pH renders this reaction
15 self limiting. The production of sulfate by this reaction lowers the pH and effectively
16 decreases the rate of further reaction. The availability of atmospheric ozone guarantees that
17 this reaction will play an important role both as a sink of gas-phase SO2 and as a cause of
18 cloudwater acidification as long as the pH of the atmospheric aqueous phase exceeds 4.
19 Hydrogen peroxide, H2O2, is one of the most effective oxidants of S(IV) in clouds and
20 fogs (Pandis and Seinfeld, 1989a). H2O2 is very soluble in water and under typical ambient
21 conditions its aqueous-phase concentration is approximately six orders of magnitude higher
22 than that of ozone. This reaction has been studied in detail by several investigators
23 (Hoffmann and Edwards, 1975; Penkett et al., 1979; Martin and Damschen, 1981; Cocks et
24 al., 1982; Kunen et al., 1983; McArdle and Hoffmann, 1983) and the reproducibility of the
25 measurements suggests a lack of susceptibility of this reaction to influence of trace
26 constituents. The proposed rate expression is (Hoffmann and Calvert, 1985)
27
April 1995 3.81 DRAFT-DO NOT QUOTE OR CITE
-------
R = _ d[S(IV)] = k[H+][H202][HS03] (
2 dt i + K[H+]
1
2 with k=1.45 x 107 M'1 s'1 and K=13 M'1 at 298 K. Noting that H2O2 is a very weak
3 electrolyte, and from 6.3, 6.4, and 6.5 that [H+] [HSOJ] = Hso Kslpso and that for
4 pH>2,l+/sf[H+]—1, one concludes that the rate of this reaction is practically pH
5 independent in the pH range of atmospheric interest. For a H2O2(g) mixing ratio of 1 ppb
6 the rate is roughly 300 mM tr1 (ppb SO^'1 (700% SO2(g)h'1 (g water /m3 air)"1). The near
7 pH independence can also be viewed as the result of the cancellation of the pH dependence
8 of the S(IV) solubility and the reaction rate constant. The reaction is very fast and indeed
9 both field measurements (Daum et al., 1984) and theoretical studies (Pandis and Seinfeld,
10 1989b) have suggested that H2O2(g) and SO2(g) rarely coexist in clouds and fogs. The
11 species with the lowest concentration before the cloud or fog formation is the limiting
12 reactant, and is rapidly depleted inside the cloud or fog layer.
13 Organic peroxides have been also proposed as potential aqueous-phase oxidants of
14 dissolved sulfur (Graedel and Goldberg, 1983; Lind and Lazrus, 1983; Hoffmann and
15 Calvert, 1985).
16 Simulations for typical continental clouds suggest that these reactions are of minor
17 importance for the S(IV) oxidation and represent small sinks for the gas-phase hydroperoxide
18 (0.2% CH3OOH h-1) and peracetic acid (0.7% CH3C(O)OOH h'1). The S(IV) oxidation
19 by O2 is known to be catalyzed by Fe3+ and Mn2+
20
21 Mn2+,Fe3+
22 S(IV) + 1 O2 > S(VI) (3-28)
23
24 This reaction has been the subject of considerable interest (Hoffmann and Boyce, 1983;
25 Martin, 1984; Hoffmann and Jacob, 1984; Hoffmann and Calvert, 1985; Clarke and
26 Radojevic, 1987) and significantly different measured reaction rates, rate laws and pH
27 dependencies have been reported (Hoffmann and Jacob, 1984). Martin and Hill (1987a,b)
28 have demonstrated that this reaction is inhibited as ionic strength increases. They explained
April 1995 3-82 DRAFT-DO NOT QUOTE OR CITE
-------
1 most of the literature discrepancies by differences in these factors during the various
2 laboratory studies.
3 In the presence of oxygen, iron in the ferric state, Fe(III), catalyzes the oxidation of
4 S(IV) in aqueous solutions. Fe(II) appears not to catalyze directly the reaction and is first
5 oxidized to Fe(III) before S(IV) oxidation can begin (Huss et al., 1982a,b).
6 For pH values from 0 to 3.6 the iron-catalyzed S(IV) oxidation rate is first order in
7 iron, first order in S(IV) and is inversely proportional to [H+] (Martin and Hill, 1987a),
8
r _ _ d[S(IV)] _ k [Fe3+][S(IV)] (3_29)
dt l [H+]
9
10
11 This reaction is inhibited by ionic strength and sulfate and these effects are described by:
12
(3-30)
13
14 and
(3-31)
1 + 150[S(VI)]2/3
15
16 where / is the ionic strength of the solution and [S(VI)] is in M. A rate constant k = 6 s
17 has been recommended by Martin and Hill (1987a). Sulfite appears to be almost as equally
18 inhibiting as sulfate.
19 The rate expression for the same reaction changes completely above pH 3.6. This
20 suggests that the mechanism of the reaction differs in the two pH regimes, and is probably a
21 free radical chain at high pH and a non radical mechanism at low pH (Martin et al., 1991).
22 The low solubility of Fe(III) above pH 3.6 presents special experimental problems. At high
23 pH the reaction rate depends on the amount of iron in solution, rather than on the total
24 amount of iron present. At this range the reaction is second order in dissolved iron (zero
25 order above the solution iron saturation point) and first order in S(IV). The reaction is still
April 1995 3-83 DRAFT-DO NOT QUOTE OR CITE
-------
1 not very well understood and Martin et al. (1991) proposed the following phenomeno logical
2 expressions (in M s"1)
pH4.0: - = lxl09[S(IV)][Fe3+]2
dt
pH5.0-6.0: d[S(IV)] = lxlO-3[S(IV)]
dt
pHT.O: -d[S(IV)] = 1 X KT4[S(IV)]
dt
3
4
5 for the following conditions:
[S(IV)] « 10/xM,[Fe3+]>0.1//M, K0.01M, [S(VI)] < 100M, and T=298K.
6
7 Note that iron does not appear in the pH 5-7 rates because it is assumed that a trace of iron
8 will be present under normal atmospheric conditions. This reaction is important in this high
9 pH regime (Pandis and Seinfeld, 1989a, b; Pandis et al., 1992).
10 Martin et al. (1991) also found that non-complexing organic molecules (e.g. acetate,
11 trichloroacetate, ethylalcohol, isopropyl alcohol, formate, allyl alcohol, etc.) are highly
12 inhibiting at pH values of 5 and above, and are not inhibiting at pH values of 3 and below.
13 They calculated that, for remote clouds, formate would be the main inhibiting organic, but
14 by less than 10%. In contrast, near urban areas formate could reduce the rate of the
15 catalyzed oxidation by a factor of 10-20 in the high pH regime.
16 The manganese catalyzed S(IV) oxidation was initially thought to be inversely
17 proportional to the H+ concentration. Martin and Hill (1987b) suggested that ionic strength,
18 not hydrogen ion, accounts for the pH dependence of the rate. These authors were also able
19 to explain some unusual behavior described in the literature on this reaction and to partially
20 reconcile some of the literature rates. The manganese catalyzed reaction obeys zero-order
21 kinetics in S(IV) in the concentration regime above 100 mM S(IV),
22
April 1995 3-84 DRAFT-DO NOT QUOTE OR CITE
-------
_d[SOV)] = ko[Mn2+]2 (3.32)
-4.07 J F
k0=k0*10 ^
1 with A:* = 680 M'1 s'1 (Martin and Hill, 1987b). For S(IV) concentrations below 1 mM the
2 reaction is first order in S(IV),
= k0[Mn2+][S(IV)] (3-33)
dt
-4.07J T
k = k0*10 1+Jr
3
4 with k*0 = 1,000 M'1 s"1 (Martin and Hill, 1987b). It is still not clear which rate law is
5 appropriate for use in atmospheric calculations, although Martin and Hill (1987b) suggested
6 the provisional use of the first order, low S(IV) rate.
7 When both Fe3+ and Mn2+ are present in atmospheric droplets, the overall rate of the
8 S(IV) reaction is enhanced over the sum of the two individual rates. Martin (1984) reported
9 that the rates measured were 3 to 10 times higher than expected from the sum of the
10 independent rates. Martin and Good (1991) obtained at pH 3.0 and for [S(IV)] < 10 mM
11 the following rate law
12
d[S(IV)]
75o[Mn(II)][S(IV)] + 2600[Fe(III)][S(IV)]
13 (3-34)
14 and a similar expression for pH 5.0 in agreement with the work of Ibusuki and Takeuchi
15 (1987).
16 Free radicals, such as OH and HO2, either heterogeneously scavenged by the aqueous
17 phase or produced in the aqueous phase, participate in a series of aqueous phase reactions
18 (Graedel and Weschler, 1981; Chameides and Davis, 1982; Graedel and Goldberg, 1983;
19 Schwartz, 1984; Jacob, 1986; Pandis and Seinfeld, 1989a).
20 Pandis and Seinfeld (1989a) proposed that under typical remote continental conditions
21 there are two main radical pathways resulting in the conversion of S(IV) to S(VI):
April 1995 3.35 DRAFT-DO NOT QUOTE OR CITE
-------
S(IV)(+OH) -» SOgC+Oj) -* HSO^+HSOg) -* S(VI) (3-35)
S(IV)(+OH) -^ SO^ SO4(+Cr,HSC>3) -* S(VI) (3-36)
1
2 with the first of these two pathways typically being faster that the second.
3 Nitrogen dioxide has a low water solubility and therefore its low resulting
4 aqueous-phase concentrations suggests that its oxidation of S(IV)
5
H0
2 -
6 2NO2 + HSO3- + - > 3H+ + 2NO2~ + SO|~ (3-37)
7
8 should be of minor important in most cases. This reaction has been studied by Lee and
9 Schwartz (1983) at pH 5.0, 5.8 and 6.4 and was described as a reaction that is first order in
10 NO2 and first order in S(IV), with a pH-dependent rate constant. The evaluation of this rate
1 1 expression was considered tentative by Lee and Schwartz, in view of evidence for the
12 formation of a long-lived intermediate species. The apparent rate constant was found to
13 increase with pH. This reaction is considered of secondary importance at the concentrations
14 and pH values representative of clouds. However, Pandis and Seinfeld (1989b) reported that
15 for fogs occurring in urban polluted areas with high NO2 concentrations this reaction could
16 be a major pathway for the S(IV) oxidation, if the atmosphere has enough neutralizing
17 capacity, e.g. high NH3 (g) concentrations.
18 Sulfite and bisulfite can form complexes with various dissolved aldehydes. One
19 important example is the reaction of sulfite or bisulfite with formaldehyde to produce
20 hydroxymethanesulfonate ion (HMS) (Boyce and Hoffmann, 1984; Munger et al., 1984,
21 1986; Olson and Hoffman, 1989; Faccini et al., 1992).
22 The HMS formed acts as a S(IV) reservoir protecting it from further oxidation, and its
23 formation has been advanced to explain high S(IV) concentrations that have been observed in
24 clouds and fogs. The rates of S(IV) complexation and oxidation are highly dependent on
25 cloud pH and on the concentrations of HCHO and oxidants. Characteristic tunes for S(IV)
26 depletion through complexation and oxidation can be compared for typical ranges of HCHO,
27 H2O2, and pH. At pH values below about 4, the rate of reactions 6.26 and 6.27 are several
28 orders of magnitude slower than the reaction of S(IV) with dissolved H2O2. Thus, in this
April 1995 3-86 DRAFT-DO NOT QUOTE OR CITE
-------
1 range oxidation predominates over complexation. The characteristic times of the two
2 reactions become approximately comparable at pH around 5 so that complexation with
3 HCHO becomes competitive with oxidation by H2O2. When pH exceeds 6, the reactions of
4 S(IV) with HCHO became more important than reaction with H2O2. HMS formation can
5 inhibit S(IV) oxidation if the S(IV) complexation rate is comparable to, or greater than, the
6 S(IV) oxidation rate and the rate of SO2 mass transport into the drop controls the rate of
7 S(IV) oxidation. The effectiveness of HMS as a S(IV) reservoir depends critically on its
8 resistivity to OH attack.
9
10 3.4.1.3 Aqueous-phase Transformation of NO2 to HNO3 and NH4NO3
11 Aside from reaction of N2O5 with liquid water, there does not appear to be any other
12 aqueous-phase reaction of nitrogen oxides that contributes substantially to atmospheric
13 nitrate.
14 In contrast to the sulfate system, the nitrate system exhibits a gaseous equilibrium that
15 admits to a substantial gas-phase fraction (as nitric acid vapor) under ambient conditions.
16 Thus, the chemical kinetics of the aqueous-phase oxidation of NO by O2 has been
17 reexammed by two groups (Lewis and Deen, 1994; Pires et al.,1994), with confirmation of a
18 third-order rate law,
R = k[NO(aq)]2[02(aq)], (3-38)
19
20 analogous to the gas-phase reaction, with k = (7 ± 1) x 106 M"2 s"1 at 296 K. Evaluation of
21 the rate of this reaction in cloudwater confirms that the reaction rate is negligible under
22 atmospheric conditions, as indicated earlier by Schwartz and White (1983).
23
24 3.4.2 Transport and Transformations in Plumes
25 In the 1970s, many field studies were plume studies or urban-scale studies, and most
26 models were Lagrangian and limited to linearized treatment of chemistry and other non-linear
27 processes. Some of these field studies, along with regional visibility information and back-
28 trajectories from local pollution episodes, pointed to the existence of long range transport and
29 to the regional nature of air pollution and haze (Hall et al., 1973; Gillani and Husar, 1976;
30 Wolff et al., 1977). In response, some of the major field studies in the 1980s had a regional
April 1995 3.37 DRAFT-DO NOT QUOTE OR CITE
-------
1 scope with focus on acidic depositions, oxidants, or aerosols and visibility. That decade also
2 saw major strides in measurement technology and in the development of increasingly
3 sophisticated Eulerian air quality models with explicit treatment of non-linear processes.
4 In these models, however, the treatment of plumes, particularly point-source plumes, was
5 grossly distorted by varying degrees depending on the spatial resolution of the grid. New
6 interest also began to emerge in global climate change, global data, and global modeling.
7 In the decade of the 1990s, the principal interests in modeling and measurements appear to
8 be in two areas: global-scale issues, with particular focus on clouds and aerosols; and,
9 regional and sub-regional issues, with special interests in comprehensive linked study of
10 oxidants, aerosols and acidic depositions, and in multi-scale interactions (e.g., nested
11 gridding and the treatment of subgrid-scale processes related to plumes, clouds, and air-
12 surface interactions).
13 Topics related to field measurements are also covered in other parts of this document:
14 methodologies for sampling and analysis of PM and acidic deposition in Chapter 4; ambient
15 air measurements of PM concentrations and properties in Chapter 7; and field studies of
16 visibility and PM in Chapter 10. The focus in this section is on North American field
17 studies of the past 15 years or so, particularly as they relate to the following objectives:
18 better understanding of atmospheric processes (formation, transformation, transport, and
19 removal) which modify the concentration, size and composition of PM; evaluation of source-
20 or receptor- oriented models of PM air quality; and generation of model inputs.
21
22 3.4.2.1 Field Studies of Transport Processes
23 Except for the gravitational settling of coarse particles (included in dry deposition), the
24 transport of PM is similar to that of gases. Following their emissions, gases and fine
25 aerosols rise due to buoyancy effects, are advected downwind by the prevailing mean flow
26 field, and are dispersed horizontally and vertically by ambient turbulence, wind-shear effects,
27 and cloud processes. These dispersive mechanisms result from the interaction of large air
28 masses, or from the disturbance of the larger-scale flow in a given air mass by insolation-
29 driven surface fluxes of heat and moisture, and by surface drag effects. The influence of
30 these surface effects is largely confined to the atmospheric boundary layer (ABL), the height
31 of which varies diuraally and seasonally, peaking typically at between 1 and 3 km on
April 1995 3-88 DRAFT-DO NOT QUOTE OR CITE
-------
1 summer afternoons over the continental U.S.A. Pollutant emissions may be within the ABL
2 or above it (depending on emission height and buoyancy), and their dispersion is markedly
3 different in the two cases, being much more rapid and vigorous in the daytime convective
4 boundary layer (CBL) than in the stable layers aloft or in the stable nocturnal boundary
5 layer. Quantitative study of these transport and dispersion processes requires, ideally,
6 simultaneous measurements of a large number of variables related to insolation and clouds,
7 surface characteristics and surface fluxes of heat and moisture, and dynamic 3-D fields of
8 flow, temperature, humidity and concentrations of trace pollutants in the ambient atmosphere.
9 Transport and dispersion processes also have a critical influence on plume chemistry and dry
10 deposition, which are often diffusion-limited. Meteorological measurements must therefore
11 be an integral part of any plume study, even when the focus is on chemistry or deposition.
12 The shift to Eulerian grid modeling in the 1980s did not include measures to preserve the
13 essence of the sub-grid-scale features of plumes, which were instantaneously dispersed over
14 the entire emission grid cell (a volume of ~ 1012 m3 in RADM with 80 km horizontal
15 resolution), thereby also grossly distorting plume chemistry, aerosol formation, and pollutant
16 budgets. There is growing awareness now of the need for more realistic treatment of plumes
17 in grid models.
18 A large body of literature exists on studies (including field studies) of ABL structure
19 and dynamics, and on the characteristics of the wind, temperature and moisture fields in the
20 ABL and, to a lesser extent, in the free troposphere aloft. Those studies are outside the
21 present scope. Some of the recent major advances in the knowledge about the ABL are
22 reviewed by Briggs and Binkowski (1985). Our scope here is limited to field studies of the
23 transport and dispersion of PM and their precursors (e.g., SOX and NOX). Prior to 1975,
24 most such field studies were limited to the behavior of point-source plumes in the
25 7-mesoscale range (120 km), i.e., on plume rise and short-range dispersion. Such behavior
26 is well understood qualitatively; quantitatively, it is well enough represented in models at the
27 time scales characteristic of most commonly-used plume dispersion models (~ 1 h), but not
28 at the much shorter time scales of relevance to plume chemistry and plume visibility. In this
29 near-source range, instantaneous plume behavior is very different from the larger scale
30 average behavior. In an intercomparison of four plume visibility models, it was concluded
31 that much of the variation in visibility observed in the Navajo power plant plume in northern
April 1995 3.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 Arizona was probably due to fluctuations in source emissions and plume dispersion at scales
2 below those resolvable by the models (White et al., 1985). Since the atmospheric residence
3 of fine PM in the lower troposphere is on the order of days, our interest here is more on the
4 transport and dispersion of plumes over the 0- and a- mesoscale ranges (« 20 to 200 and
5 200 to 2,000 km). Quantitative determination of transport over the mesoscale requires
6 special field studies with controlled tracer releases. Such studies are relatively recent and
7 very few, and they represent only a few isolated meteorological scenarios.
8
9 3.4.2.1.1 Field Measurements Related to Transport Modeling
10 Routine meteorological field measurements include surface weather observations of a
11 broad variety of meteorological variables made every three hours at several thousand sites
12 across the country by the National Weather Service, as well as upper-air soundings
13 (radiosondes) of wind, temperature and relative humidity twice a day (noon and midnight) at
14 a much more limited number of sites which, on average, are about 400 km apart. These
15 data constitute the principal raw meteorological information used in regional transport
16 models, which are either Lagrangian trajectory models or dynamic 3D Eulerian grid models.
17 Most trajectory models are two-dimensional, with atmospheric flow patterns being analyzed
18 on isobaric or terrain-following surfaces, or in bulk transport layers confined to the mixed
19 boundary layer. These simplifying assumptions concerning vertical motions lead to large
20 transport errors on the regional scale (Kuo et al., 1985). The vertical velocity can be
21 calculated at grid points in a regional model domain from the continuity equation, but the
22 temporal and spatial resolutions of the radiosonde data are so coarse in most areas that the
23 result would be a gross approximation only. 3D flows may be best simulated by moist
24 adiabatic trajectories, but since analysis methods cannot always resolve the stratified nature
25 of the required moisture fields, the most reasonable simulations of 3D transport are probably
26 dry adiabatic (isentropic) trajectories. Danielsen (1961) presented a case study showing a
27 separation of =1,300 km after only 12 h of transport as simulated by isobaric and isentropic
28 trajectories. It was probably an extreme case. The gridded wind field in regional Eulerian
29 air quality models is typically generated by the application of dynamic 3D mesoscale
30 meteorological models (e.g., PSU-MM5 and CSU-RAMS) which incorporate the routine
31 NWS observations through a dynamic Four Dimensional Data Assimilation (FDDA)
April 1995 3-90 DRAFT-DO NOT QUOTE OR CITE
-------
1 technique. The NWS surface weather database also includes a measure of prevailing
2 visibility as determined by human observers. A number of field studies have established the
3 reliability of such subjective visibility observations (e.g., Horvath and Noll, 1969; Hoffman
4 and Kuehnemann, 1979). They have proved to be a very useful indicator of regional haze
5 and its long-range transport (Gillani and Husar, 1976), and have been used to study the long-
6 term trends of the spatial-temporal variability of regional haze and air quality in the eastern
7 U.S. over many decades (Husar et al., 1981; Sloane, 1982).
8 Special field studies of transport and dispersion are based on observations of the
9 transport of pressurized (constant density) balloons (called tetroons if their shape is
10 tetrahedral), and of the evolution of plumes resulting from pollutant emissions or controlled
11 releases of artificial tracers. Balloons have been used in mesoscale studies in three ways: as
12 isolated Lagrangian markers of pollutant emissions (e.g., Clarke et al., 1983); in sequential
13 releases to provide one-particle diffusion estimates (e.g., Thomas and Vogt, 1990); and in
14 cluster releases to study relative diffusion (e.g., Er-El and Peskin, 1981). Tetroons generally
15 carry a transponder which permits continuous tracking with a radar, thus providing the
16 complete detailed 3D trajectory. The range of the tetroon experiment is normally limited by
17 the tracking range of the radar (< 100 km). This range can be extended to the full range of
18 tetroon transport by including a tag which the finder can return with information about at
19 least the terminal location. In some studies (e.g., Clarke et al., 1983), tetroons have been
20 tracked continuously over much longer ranges by sequential tracking from the network of
21 FAA radars used in support of aviation. Studies based on tracers and air pollutants also
22 provide information about plume dispersion. Most early tracer studies were limited to a
23 range of about 100 km due to the nature of the tracers then available and limitations of
24 technology. Development of new tracers (e.g., the PFTs or perfluorocarbon tracers) and
25 new sampling and analysis techniques have not only extended the range in more recent
26 experiments by more than an order of magnitude, but the new data are also more reliable.
27 Pack et al. (1978) presented a detailed review of many early studies in which
28 observations of the transport of pollutant plumes, tracers, or balloons were compared with
29 results of diagnostic trajectory calculations. The models commonly used then were based on
30 the kinematic approach (using objectively-analyzed wind fields based on measured winds) and
31 a single transport layer. The observed winds were used as input in different ways: for
April 1995 3_91 DRAFT-DO NOT QUOTE OR CITE
-------
1 example, surface winds or adjusted surface winds representing average winds in the whole
2 transport layer; or, upper air winds averaged over the transport layer. The adjustment of
3 surface winds included enhancement of the speed by as much as a factor of two, and a veer
4 of the wind direction by as much as 40°, to account for the real-world wind speed shear and
5 directional veer with height. The advantage of using surface winds was due to their much
6 higher spatial and temporal resolution, compared to the much coarser resolution of the upper-
7 air radiosonde winds. The early results of comparisons of calculated and observed
8 trajectories evidenced a broad range of discrepancy (10 to 54% of the trajectory length after
9 only 100 km, and 55 to 60% after 650 km), and also the presence of large systematic errors,
10 not always in the same direction, depending on the presence of complex flows due to fronts,
11 complex terrain, etc. The best simulations were often obtained by the use of adjusted surface
12 winds, and such adjustments varied between studies. The errors were found to be lowest for
13 transport in the daytime CBL, and substantially larger for transport in stably-stratified layers.
14 Moran (1992) has tabulated (his Table 2-4) basic information about a number of formal
15 /3- and a- mesoscale tracer experiments since 1973, in which the release was at surface level
16 and the measured transport range was at least 25 km (and up to 3,000 km). Table 3-16
17 summarizes, in chronological order, some of the major field studies of the past 20 years with
18 measurements and modeling of transport extending into the a-mesoscale. It includes the
19 major tracer studies as well as air quality and tetroon studies. The transport models in these
20 studies were driven either by routine meteorological observations or by additional
21 measurements made as part of the field studies. The following important observations are
22 based on the studies listed in Table 3-13:
23
24 • The routine data of the radiosonde network («400 km, 12 h) are too coarse both
25 spatially (Kahl and Samson, 1986, 1988) and temporally (Rolph and Draxler, 1990;
26 Kuo et al., 1985) for accurate simulation of long range transport.
27
28 • The error in calculated trajectories is greatest under conditions of high speeds which
29 generally accompany complex mesoscale systems (Rolph and Draxler, 1990).
30
April 1995 3-92 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-13. RECENT FIELD STUDIES OF a-MESOSCALE TRANSPORT AND TRAJECTORY MODEL
VO
OJ
Study
INEL Study
Idaho Nat'l Eng. Lab
MISTT
Midwest Interstate
Sulfur Transport and
Transformation Study
VISTTA
Visibility Impairment
due to Sulfur Transport
and Transformation
in the Atmosphere
TPS
Tennessee Plume
Study
NEROS
Northeast Regional
Oxidant Study
Mt. Isa
Smelters Plume Study
Period
Feb-May 74
Summer 75
Summer 76
Jun, Jul,
Dec-79
Aug-78
15-Aug-78
Summer 79
Summer 80
Jul-79
Tracer(s)
Kr-85
Plume sulfur
Anthropogenic
aerosol, ozone
Tetroons (1 cu. m)
with transponder
"
Tetroons (6 cu. m)
Tetroons (1 and 6 cu. m)
Excess plume S
and Aitken Nuclei Count
(ANC)
Release Sites(s)
INEL (Idaho) fuel
reprocessing plant
(76 m stack)
Labadie Power
Plant near
St. Louis, MO
Los Angeles Basin
TVA
Cumberland
Steam Plant, TN
••
MD, OH, PA, TN
Columbus, OH
Mt. Isa, Australia
(Sulphide smelters,
-0.6 km apart)
Tracking/Sampling
Samplers at
1 1 Midwestern NWS
sites; 10-h day and
night samples
In-situ aircraft
measurements
Detailed air quality
and aerosol
measurements at a
Grand Canyon site
Radar to - 75 km;
terminal point based
on return tag.
«
Continued FAA
radars Radar and
return tag.
Aircraft measts. of
Total S, ANC,
COSPEC-SO2
Maximum Range
(Airshed)
-1,500 km
-300 km
-750 km
-1,000km
(KY,IN,OH,ONT)
~300km(KY)
-500 km to NE
-1,500 km to NE
-1,000km
(Semi-and region
in N. Australia)
Model
Comparison(s)
NOAA-ARL
trajectory
model with 300 m
vertical resolution
Simple particle
trajectory model
CAPITA
Monte Carlo particle
transport model
• NOAA-ATAD
• NCAR isentropic
• CAPITA Monte
Calo
3D reg'l dyn. model
NOAA, NCAR,
CAPITA, as above
Simple layered wind
trajectory model;
Ref(s)
Draxler(1982)
Gillani et al. (1978)
Gillani (1986)
Maciaset al. (1981)
Clarke et al. (1983)
Warner (1981)
Clarke et al. (1983)
Carras and
Williams (1981)
Comments
Small signal above b/g;
300 m layered approach
to permit spread by wind
dir'l shear necessary.
Quasi-Lagrangian pibal
measurements of winds
along plume transport
Evidence also of long-
range impact of Copper
smelter plumes
Part of a large plume
transport/ chemistry
study, including aircraft
measts.
Part of a large urban and
reg'l oxidant study
Exceptionally clean
plume b/g.
o
o
z
o
H
O
c
o
n
HH
H
m
-------
1
'O
VO Study
Great Plains
Mesoscale Tracer
Expt.
CAPTEX
Cross-Appalachian
Tracer Expt.
ANATEX
Across North
America
Tracer Expt.
MISERS GOLD
TABLE 3-13 (cont'd). RECENT FIELD STUDIES OF a-MESOSCALE TRANSPORT
AND TRAJECTORY MODELS
Period Tracer(s)
Jul-80 Two PFTs (PMCH
and PDCH) and two
heavy meyhanes
(ME-20, ME-21)
Sep/Oct83 PFT(PMCH)
Jan-Mar 87 3 PFTs (PMCP,
PMCH, PDCH)
l-Jun-89 Indium oxide
(vapor deposits on
particles)
Release Sites(s)
Norman, OK
(1 m AGL)
Dayton, OH
Sudbury, ONT
Glasgow, MT
St. Cloud, MN
White Sands
Missile Range,
NM
Maximum Range
Tracking/Sampling (Airshed)
Surface samplers: 17600kmtoNNE
on arc at 100 km 38
on arc at 600 km and
aircraft sampling
Surface array of — 1 , 100 km
>80 samplers at arcs(NE U.S.)
from 300-1, 100 km
and aircraft sampling
Surface network (77); -3,000 km
Towers (5); and (Eastern U.S.)
aircraft sampling.
In-situ aircraft: filter ~ 1 ,400 km
samples analysed for NM to MO
tracer and particles.
Model Comparison(s)
Different 3D regional
models
Different 3D regional
models; also
MESOPUFF II
3 single-layer LAGR,
6 multi-layer LAGR,
2 multi-layer Eulenan
Gifford's random-force
diffusion theory
Ref(s) Comments
Ferber et al. (1981) Important role of wind
Moran(1992) shear effects of nocturnal
jet.
Ferber et al. (1986) Terrain-effects found
Moran (1992) important. Enhanced
Godowitch (1989) upper air met measts.
Draxler et al. Enhanced upper air met
(1991) measts.
Rolph and Draxler
(1990)
Clark and Cohn
(1990)
Kahl et al. (1991) Dust plume from a
Mason and military test explosion.
Gifford (1992)
•^
%
9
O
d
I
O
-------
1 • Initial errors in trajectory simulations (both in direction and vertical spread) play a
2 critical role in overall model uncertainty (Draxler et al., 1991).
3
4 • Single-layer Lagrangian trajectory models do not spread the "plume" adequately,
5 while Eulerian models spread it too much. Multi-layer Lagrangian models perform
6 the best in terms of dispersion of point-source emissions (Clark and Cohn, 1990).
7
8 • Vertical information about tracer trajectories, based on continuously-tracked tetroons
9 and aircraft measurements, contain much useful information not captured by surface
10 sampling alone (Clarke et al., 1983). There is, for example, evidence of cloud
11 venting of ABL pollutants into the free troposphere, where their residence time is
12 longer and the flow field may be quite different.
13
14 • Terrain-induced effects played an important role in CAPTEX, and effects related to
15 the nocturnal jet were important in the Great Plains Experiment (Moran, 1992).
16 Nocturnal wind directional shear plays a major role in effectively dispersing plumes
17 which have been dispersed vertically during the preceding daytime CBL.
18
19 * Directional wind shear plays an important role in plume dispersion even in the CBL
20 during (3-mesoscale transport (Gillani, 1986).
21
22
23 The issue of substantial overdispersion by Eulerian models is important because the
24 state-of-the-art as well as the future direction in mesoscale modeling (meteorological/air
25 quality/aerosol) appear to favor the Eulerian approach. A significant source of the problem
26 must be related to the gross initial overdispersion of plumes in regional Eulerian models,
27 particularly of elevated point-source plumes (carriers of most of the U.S. anthropogenic
28 emissions of sulfur). The instantaneous false dilution of fresh emissions of NOX into the
29 NOx-limited surrounding environment (e.g., in the eastern U.S.) greatly distorts plume
30 chemistry and aerosol formation. Proper sub-grid-scale treatment of plumes remains an
31 important outstanding issue in regional modeling. Other sub-grid-scale effects in need of
32 more attention pertain to complex mesoscale flows (e.g., storms, fronts, cloud venting,
33 complex terrain effects, etc.). They too are an important source of model errors. A few
34 special field studies have been carried out to investigate such flows: for example, VENTEX
35 (Ching and Alkezweeny, 1986) and PRESTORM (Dickerson et al., 1987) for cloud venting,
36 and ASCOT (Allwine, 1993) and the NGS Visibility Study (Richards et al., 1991) for flows
37 over complex terrain. Thermal effects and drainage flows also evidently play an important
38 role in influencing paniculate air quality, as in the occurrence of the Denver "brown cloud"
39 phenomenon (Sloane and Groblicki, 1981).
April 1995 3.95 DRAFT-DO NOT QUOTE OR CITE
-------
1 There is considerable field evidence also of synoptic scale transport (12,000 km) of
2 airborne particles (see, for example, Gordon, 1991). The impact of such transport is
3 important on the global scale. That subject is beyond the present scope.
4
5 3.4.2.1.2 Field Measurements Related to Dispersion Modeling
6 Gaussian semi-empirical models have been the basis of most applied diffusion modeling
7 since their development around 1960. These models were based on Taylor's diffusion theory
8 of stationary homogeneous turbulence (Taylor, 1921), and were built on a few field
9 experiments that were quite limited in scope and technology. The results have been
10 extrapolated far beyond the intended range of downwind distance and ambient conditions.
11 Some of the extrapolations were guided by statistical theory, but most were freehand
12 extrapolations (Briggs and Binkowski, 1985). Many research-grade field studies of
13 atmospheric dispersion have since been performed, but most have been limited to the
14 -y-mesoscale range. These have been reviewed by Draxler (1984), Irwin (1983), Briggs and
15 Binkowski (1985) and others. /3- and a-mesoscale studies, based on observations of the
16 dispersion of pollutant and tracer plumes have been reviewed by Moran (1992). Pollutant
17 plumes remain vertically narrow in stable flows (e.g., elevated power plant plumes released
18 at night), but rapidly fill up the CBL after fumigation in the daytime (see, for example,
19 Gillani et al., 1984). Information about spreads of plumes in the elevated stable layers is
20 particularly limited. The most common basis for estimation of such spreads (expressed as
21 ay and az, the RMS variances of lateral and vertical plume spreads) over distances under
22 100 km or so is the well-known Pasquill-Gifford (P-G) curves for different stability classes
23 (Gifford, 1961), which make use of the routine meteorological measurements to determine
24 applicable stability class. The P-G curves were developed mostly from data collected within
25 the mixing layer. Another set of parameterizations of elevated plume spreads was developed
26 by TVA (Carpenter et al., 1971) based on twenty years of experience in plume observations
27 and aerial monitoring. These require the temperature profile to establish atmospheric
28 stability. More recently, Smith (1981) analysed aircraft measurements in elevated power
29 plant plumes in different parts of the U.S., mostly in the stable layers, and determined that
30 the P-G curves overestimated plume spread in stable layers quite substantially both vertically
31 and horizontally, and that the TVA approach tended to underestimate the horizontal spread,
April 1995 3-96 DRAFT-DO NOT QUOTE OR CITE
-------
1 but possibly overestimate the vertical spread. Evidently, there continues to be uncertainty
2 about plume spreads even at distances under 100 km. Of particular interest is horizontal
3 plume dispersion, both because it is generally far greater over the mesoscale, and because it
4 is highly variable. Close to the source, plume spread is largely by progressively larger
5 turbulent eddies, but after the plume dimension substantially exceeds the scale of these eddies
6 (typically less than 1 km), dispersion is increasingly by directional wind shear with height.
7 Such shear is small for the vertically thin nocturnal plume, moderate for the plume in the
8 CBL, but maximum for the daytime plume which, after maximum vertical spread in the
9 CBL, enters the nocturnal regime which is often characterized by strong directional shear
10 effects (Gillani et al., 1984). The average crosswind spread rates of plumes from a large tall
11 stack power plant emitted within the CBL on summer days in the Midwest were observed to
12 be in the range 0.25 to 1.0 km per km of downwind transport until the plume attained a
13 width of about 30 km. Thereafter, further plume spread within the CBL was typically much
14 slower (Gillani and Pleim, 1994).
15 A common approach in Lagrangian studies of dispersion over long distances has been
16 to use semi-empirical "mesoscale" dispersion coefficients by analogy with parameterizations
17 of microscale turbulent spread. An important consequence of Taylor's statistical theory was
18 that, in stationary homogeneous turbulence, sy grew linearly with time at first (for t ~ TL,
19 the Lagrangian time scale, = 1 to 2 min in the CBL), and then asymptotically as t1/2 within a
20 few kilometers. Observations of a few a-mesoscale field studies have been interpreted to
21 suggest that the regime of linear time dependence may apply also at long distances (see, for
22 example, Pack et al., 1978), with the characteristic time scale (TL) here being related to the
23 diurnal and/or inertial scale («24 h). Others have proposed parameterizations of mesoscale
24 ay which use powers of t ranging from 0.85 to 1.5 (see, for example, Carras and Williams,
25 1988). Thus, there is no consensus about simplistic modeling of mesoscale diffusion over
26 scales exceeding 24 h. Given the wide range of conditions that plumes can experience
27 during long range transport in different air masses, over a variety of terrain types, and over
28 multiple diurnal cycles during different seasons, such a controversy is not surprising. For
29 transport in the first 24 h, the time and height of emission are critical influencing variables.
30 Thus, for example, crosswind spreads after 24 h of transport of two plumes released from
31 the same tall-stack power plant at 0800 and 2000 on a given day are likely to be very
April 1995 3.97 DRAFT-DO NOT QUOTE OR CITE
-------
1 different. During the next diurnal cycle, however, these spreads, as a fraction of downwind
2 distance travelled, are likely to converge. Alternate semi-empirical approaches of
3 representing mesoscale plume dispersion include simulation of relative dispersion of
4 hypothetical co-emitted conservative particles. In conjunction with instantaneous wind data
5 (e.g., pibal soundings), such models have proved to be satisfactory over /3-mesoscale
6 distances (e.g., Gillani, 1986), but more work is needed to establish their application over
7 long distances based on hourly-average gridded wind data such as are produced by the
8 meteorological preprocessors of regional Eulerian models. Overall, based on field evidence,
9 paniculate air quality is significantly influenced by regional transport and dispersion, but
10 quantitative simulation of these processes is still subject to considerable error.
11
12 3.4.3 Transformations in Plumes
13 3.4.3.1 Gas-to-Particle Conversion in Plumes
14 A number of field studies of gas-to-particle conversion have been conducted in the
15 plumes of large point-sources of SOX and NOX (e.g., coal- and oil-fired power plants and
16 metal smelters). Fewer studies have focused on urban-industrial plumes. These studies have
17 focused principally on quantifying the rates of aerosol formation and, to a lesser extent, on
18 investigating the mechanisms. Mechanistic studies are more difficult, particularly when
19 multiple mechanisms are co-active, as is commonly the case. In the NAPAP emissions
20 inventory for base year 1985 (Placet et al., 1991), about 70% of the U.S. anthropogenic
21 emissions of SO2, and about 25% of the corresponding emissions of NOX, were attributed to
22 large point-sources with stack heights exceeding 120 m (probably less than 150 individual
23 sources). The contribution of such sources is even higher in the eastern U.S., particularly in
24 the Ohio and Tennessee River Valleys. Clearly, these large emissions are very important in
25 the context of regional aerosols. Fortunately, many of these sources are located in rural
26 areas, and their plume chemistry can be studied in isolation from the complications of
27 interactions with other plumes. Much of the remaining anthropogenic emissions of SOX and
28 NOX are contributed by urban-industrial area sources.
29
30
April 1995 3-98 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.4.3.1.1 Plume Studies
2 Power plant and urban plume studies of SO2-to-sulfate transformations published before
3 1980 have been reviewed by Newman (1981) and in the earlier 1982 PM/SOX Air Criteria
4 Document (U.S. Environmental Protection Agency, 1982). Only a brief overview of those
5 studies is provided here; the main focus here is on plume studies published after 1980. Since
6 the plume mass is airborne, the most meaningful plume studies are based on measurements
7 made from instrumented aircraft. Early studies (pre-1975) often reported SO2 oxidation rates
8 as high as 50% h"1. They are now generally considered to be flawed due to limitations in the
9 measurement technology then available. This technology has made major strides since. For
10 example, the development of the filter pack (Forrest and Newman, 1973) has proved to be a
11 useful method of simultaneous collection of high-volume samples of SO^ and paniculate
12 sulfur. Such samples, however, only provide average concentrations over entire plume
13 cross-sections or, at best, over long crosswind plume traverses. The development of
14 continuous monitors for both SO2 and paniculate sulfur (Huntzicker et al., 1978; Cobourn
15 et al., 1978) made it possible to study sulfate formation with crosswind plume detail. Such
16 detail during a single plume traverse contains a nearly instantaneous snapshot of the full
17 spectrum of chemistry between the high-NOx regime in plume core to the low-NOx regime at
18 plume edge (Gillani and Wilson, 1980). With cross-sectionally averaged measurements, such
19 a spectrum can only be discerned in measurements ranging from near-source to far
20 downwind. The technology of continuous measurements of nitrogen species with high
21 sensitivity has also evolved greatly since 1980.
22 The period between 1974 and 1981 was very active in terms of plume studies focused
23 particularly on estimating the rate of oxidation of SO2. Studies by Brookhaven National
24 Laboratory (Newman et al., 1975a,b; Forrest and Newman, 1977a,b) and TVA (Meagher
25 et al., 1978) in coal- and oil-fired power plant plumes as well as a nickel smelter plume
26 generally yielded low oxidation of SO2 (seldom exceeding 5% over 50 km and several hours
27 of plume transport, with an uncertainty of about a factor of two). These investigators found
28 the oxidation rate to be highest close to the source, where it appeared to be correlated with
29 plume paniculate loading, and interpreted the oxidation to be due to a heterogeneous second-
30 order mechanism which became quenched as the plume diluted (Schwartz and Newman,
31 1978). These results were in sharp contrast to those of Husar et al. (1976) for a coal-fired
April 1995 3.99 DRAFT-DO NOT QUOTE OR CITE
-------
1 power plant plume, also over about 50 km of plume transport, which showed the oxidation
2 rate to be slow during an early induction period, increasing thereafter to as much as 5 % h'1.
3 No mechanistic interpretation was proposed by these authors. This controversy was resolved
4 by the subsequent findings of Gillani et al. (1978) resulting from two case studies which
5 were remarkable for their coverage of downwind range exceeding 300 km and 10 to 12 h of
6 transport of a coal-fired power plant plume during daylight as well as dark. The authors
7 found the oxidation rate of SO2 to be strongly correlated with sunlight, and also with the
8 extent of plume dilution, and background ozone concentration (considered to be a surrogate
9 for background reactivity). Maximum measured paniculate sulfur as a fraction of total
10 plume sulfur ranged as high as 18%. The daytime conversion rate in the plume was slow at
11 first, but increased as the plume diluted, reaching maximum values on the two days of
12 1.8 and 3.0% h"1 in the afternoon. Such rates are consistent with theoretical rates based on
13 the SO2-OH reaction (Calvert et al., 1978; Hov and Isaksen, 1981). The entire plume
14 transport on both occasions was in fairly dry environment (R.H. < 70%). Presumably, the
15 mixing of plume NOX and background VOC led to photochemistry which generated the
16 necessary oxidants for gas-phase oxidation of SO2. The measurements of VOC in the
17 background were both sparse and of limited reliability. The study also found the formation
18 of substantial excess of ozone in aged plumes. The interpretation based on plume-
19 background interaction satisfactorily explained the results of the BNL and TV A studies in
20 which the measurements of low oxidation of SO2 were all in coherent stable elevated plumes
21 during early morning and evening hours (low sunlight and little plume dilution), as well as of
22 Husar et al., whose measurements were in the more polluted and convective summer daytime
23 CBL.
24 As of the end of the 1970s, a number of factors had been implicated as being relevant
25 to plume sulfur chemistry. Gillani and Wilson (1980) conducted a systematic investigation of
26 the dependence of ozone and aerosol formation in power plant plumes on a variety of
27 possible influencing factors, based on the plume data of five case studies. They found that
28 temperature variations in the range 28 to 33 °C, and R.H. variations in the range 50 to 80%
29 did not have an appreciable influence; the importance of sunlight, plume dilution and
30 background composition was reconfirmed. Eatough et al. (1981, 1982) have observed a
April 1995 3-100 DRAFT-DO NOT QUOTE OR CITE
-------
1 positive temperature dependence of a linear SO2 oxidation rate in power plant and smelter
2 plumes in western U.S. in the temperature range 0 to 30 °C.
3 Gillani and Wilson (1980) also presented direct evidence and interpretation of the role
4 of plume-background interactions in plume photochemistry within the context of a common
5 pattern of diffusion-limited plume chemical evolution through three stages in a moderately
6 polluted environment. In the "early" stage, the plume is narrow and dominated by a high-
7 NOX regime in which ozone and other oxidants are sharply depleted by reaction with plume
8 NO and SO2; the VOC-NOX chemistry, SO2 oxidation, and aerosol formation are inhibited in
9 the plume in this stage. As the plume spreads and dilutes with a background characterized
10 by relatively high VOC/NOX ratio, the VOC/NOX ratio increases also in plume edges. This
11 "intermediate" stage of plume chemistry is characterized by rapid formation of ozone and
12 aerosols in plume edges, leading to an observed excess there of ozone over the background
13 (ozone "wings") while the plume core still has an ozone deficit. Sharp "wings" of Aitken
14 nuclei concentration have also been observed in plume edges at times, indicating directly the
15 nucleation of new aerosol (Wilson, 1978; Gillani et al., 1981). With continuing dilution, the
16 plume ultimately develops a condition of low-NOx, high VOC/NOX ratio and, in the summer,
17 an ozone "bulge" throughout. In this "mature" stage, the rate of oxidation of SO2 to sulfates
18 (and presumably also of NOX to secondary products) reaches its peak.
19 Gillani et al. (1981) provided a quantitative interpretation of the above observations by
20 developing an empirical parameterization of the gas-phase conversion rate of SO2 to sulfate
21 in terms of measured variables representing sunlight, mixing and background reactivity. The
22 parameterization was verified based on the "dry" data of three different power plant plumes
23 over ten days of measurements in two different summer periods. Crosswind-resolved
24 reactive plume models capable of facilitating plume-background interactions and including
25 detailed simulation of chemical kinetics have been developed and applied by Hov and Isaksen
26 (1981), Stewart and Liu (1981), and Gillani (1986). These models can depict the observed
27 behavior of ozone in the three plume stages. Their applications have shown that the
28 evolution of OH in the plume (a measure of oxidation potential) mimics the above description
29 of ozone evolution (Hov and Isaksen, 1981), and that plume oxidant and aerosol formation
30 are very sensitive to background VOC and their ingestion into the plume (Gillani, 1986).
31 However, these models continue to remain unevaluated adequately owing to a continuing lack
April 1995 3_10i DRAFT-DO NOT QUOTE OR CITE
-------
1 of data characterizing the composition of plume background (especially VOC) and the
2 crosswind detail of important intermediate and secondary species (e.g., OH, HO2, HNO3,
3 etc.).
4 A number of plume studies have verified the sunlight dependence of the SO2 oxidation
5 process, observing higher seasonal conversion rates during summer, and higher diurnal rates
6 during midday (Husar et al., 1978; Lusis et al., 1978; Roberts and Williams, 1979; Meagher
7 et al., 1981; Hegg and Hobbs, 1980; Gillani et al., 1981; Forrest et al., 1981; Williams
8 et al., 1981; Wilson, 1981; Wilson and McMurry, 1981; Liebsch and de Pena, 1982).
9 In these studies, the peak daytime conversion rate was typically between 1 and 5% h"1 in the
10 summer (higher under humid conditions), and much lower in winter. Wilson (1981)
11 reviewed the data of twelve power plant and smelter plumes in the U.S., Canada and
12 Australia, covering measurements during day and night, and summer and winter. The main
13 conclusion was that diurnally, midday conversion rates were relatively high and quite
14 variable (1 to 10% h"1), while the nighttime conversion rates were generally low (under
15 0.5% h"1). Also, the rates were found to be lower in winter than in summer.
16 Geographically, the measured plume conversion rates in the arid and relatively clean
17 southwestern U.S. environment were found to be particularly low (0.5% h"1) at all times,
18 including summer midday. Williams et al. (1981) also found the rates to be low in a smelter
19 plume in the arid, clean environment of north central Australia ( = 0.15% h"1 averaged over
20 24 h of transport).
21 Gillani et al. (1981) were able to formulate the parameterization of the gas-phase
22 conversion rate by isolating case studies performed entirely in dry conditions when liquid-
23 phase contributions were negligible. They also observed that for all cases when the plume
24 had any history of wet exposure (clouds, fogs or high humidity), the oxidation of SO2
25 invariably proceeded at a rate faster than that predicted by the gas-phase parameterization.
26 Whereas the typical range of the peak summer daytime conversion rate was 1 to 5 % h"1 in
27 Project MISTT (Missouri, Illinois), it was closer to 1 to 10% h"1 in the more humid
28 conditions of the Tennessee Plume Study (Tennessee, Kentucky). In the wetter daytime
29 situations, evidently, liquid-phase chemistry was superposed over the underlying gas-phase
30 chemistry. Gillani and Wilson (1983) focused their study on the plume data of such "wet"
31 situations. They attributed to liquid-phase chemistry the part of the total measured
April 1995 3-102 DRAFT-DO NOT QUOTE OR CITE
-------
1 conversion rate which was in excess of the rate estimated by the gas-phase parameterization.
2 The liquid phase was found to be due to clouds, fogs and light rain, or due to wetted
3 aerosols under conditions of high ambient humidity (R.H. > 75%). The liquid-phase
4 contribution to the conversion rate was found to be in excess of 40% of the total in two-
5 thirds of the cases analysed, being as high as 8% hf1 averaged over the whole plume over
6 6 h of transport in the most extreme case (clouds and light rain). Similar increases in
7 conversion rates in power plant plumes interacting with high humidity have also been
8 observed by others (e.g., Dittenhofer and de Pena, 1978; Eatough et al., 1984).
9 Determination of the liquid-phase conversion rate involves quantification not only of the
10 kinetics, but also of the discrete and variable extent of plume-cloud interaction. Gillani et al.
11 (1983) formulated a parameterization of the conversion rate for plume-cloud interaction in
12 which the physical extent of such interaction was represented probabilistically, and the higher
13 liquid-phase conversion rate was applied only for the in-cloud portion of the plume. The
14 application of the parameterization to a case study corresponding to summer daytime plume
15 transport within the CBL, in patchy contact with fair-weather cumulus above, permitted
16 estimation of the average in-cloud conversion rate averaged over 7 h (1000 to 1700) to be
17 12% h"1. Considering that the corresponding average liquid water content in the clouds was
18 certainly less than 1 g m"3 (1 ppm), much higher actual oxidation rates within individual
19 droplets are indicated. Gas-phase photochemistry at a much slower rate was concurrently
20 quite active in the more extensive drier parts of the plume below, producing ozone and other
21 oxidants which contributed to gas-phase as well as liquid-phase sulfur chemistry. It was not
22 possible to relate the in-cloud kinetic rate to the critical variables controlling it, such as cloud
23 liquid water content, H2O2 concentration, or droplet pH, because such measurements were
24 not made. The role of concurrent gas-phase photochemistry is indeed essential to provide the
25 oxidizing agents of liquid-phase chemistry. Clark et al. (1984) found the contribution of
26 liquid-phase chemistry in a power plant plume to be negligible during long-range transport
27 over water in a shallow stratocumulus-filled boundary layer, with limited plume dilution, low
28 insolation, and little photochemistry.
29 A quite different approach based on aerosol growth laws applied to aerosol size
30 distribution data was taken by McMurry et al. (1981) and McMurry and Wilson (1982) to
31 study relative contributions of the principal mechanisms of gas-to-particle conversion.
April 1995 3_103 DRAFT-DO NOT QUOTE OR CITE
-------
1 Theory predicts different growth laws for different chemical mechanisms of aerosol
2 formation. The authors examined the functional dependence of calculated particle diameter
3 growth rate on particle diameter. By matching field data with theoretical growth laws, it was
4 possible to differentiate between mechanisms. Application of this approach indicated gas-
5 phase chemistry and condensation of the product to be the predominant mechanism of aerosol
6 formation in several power plant plumes in eastern and western U.S., with increasing
7 contribution of heterogeneous mechanisms with increasing humidity (McMurry et al., 1981);
8 in a case study of the urban plume of St. Louis, 75% and 25% of the aerosol formation were
9 attributed to homogeneous and heterogeneous mechanisms, respectively, while most of the
10 aerosol formation in the ambient air in the Great Smokey Mountains where relative
11 humidities were high (up to 95%) was attributed to the droplet-phase mechanism (McMurry
12 and Wilson, 1982).
13 In an overview of empirical parameterizations of sulfur transformations in power plant
14 plumes, Gillani (1985) estimated that on a 24-h average basis, sulfate formation rates in a
15 large power plant plume in the U.S. Midwest in July 1976 were likely to be 0.8 ± 0.3% h'1
16 by gas-phase reactions (midday peak ~ 2.6% h"1) and at least half as much by liquid-phase
17 reactions. Winter rates were estimated to be an order of magnitude lower than the summer
18 rates for the gas-phase mechanism, but comparable for the liquid-phase mechanism. Since
19 1981, no new field studies of chemistry in large point-source plumes have been conducted in
20 the eastern U.S. A comprehensive plume study with state-of-the-art aircraft measurements of
21 primary and secondary sulfur and nitrogen species, as well as VOC and ozone, is planned to
22 occur in the summer of 1995 as part of the Southern Oxidant Study (SOS) Nashville Field
23 Measurement Program.
24 Information about field measurements of nitrate formation in point-source combustion
25 plumes is much more meager. Summertime plume measurements suggest that nitrate
26 formation is principally in the form of nitric acid vapor (Hegg and Hobbs, 1979; Richards
27 et al. 1981), and that oxidation of NOX to HNO3 may proceed about three times faster than
28 the rate of oxidation of SO2 (Richards et al., 1981; Forrest et al., 1981). Richards et al.
29 (1981) observed that along the transport of the Navajo Generating Station in Arizona, there
30 was adequate ammonia to neutralize the sulfate formed in the plume, but not enough to form
31 ammonium nitrate. Forrest et al. (1981) found NH^/SO^f to increase with downwind
April 1995 3-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 distance and was mostly less than 2 (not enough to fully neutralize the sulfate), but
2 sometimes more than 2, indicating a possibility of the formation of some ammonium nitrate.
3 Eatough et al. (1981) observed that in the western desert region, the neutralization of sulfuric
4 acid in plumes was due not only to ammonia, but also to other basic material (e.g., metal
5 oxides and CaCO3).
6 Field information about secondary formations in urban plumes is scantier than for
7 power plant plumes for sulfur compounds, but possibly slightly more for nitrogen
8 compounds. White et al. (1976, 1983) reported slow formation of ozone and aerosols at first
9 in the St. Louis urban plume, but faster rates farther downwind. Average sulfate formation
10 rates between successive downwind measurement locations on summer days were estimated
11 at 2 to 4% h"1. Isaksen et al. (1978) applied a reactive plume model to a subset of the
12 St. Louis data, and estimated peak rates for the formation of sulfuric and nitric acid of 5 and
13 20% h"1, respectively. Based on the same data set, Whitby (1980) estimated that about
14 1,000 tons of secondary fine aerosol may be produced in the plume in one summer
15 irradiation day. Alkezweeny and Powell (1977) estimated peak sulfate formation rates in the
16 St. Louis plume at 10 to 14% h"1. Miller and Alkezweeny (1980) reported sulfate formation
17 rates in the Milwaukee urban plume on two summer days in very different air masses to
18 range from 1% h"1 (clean background) to 11% h"1 (polluted background). The most
19 extensive studies of NOX chemistry in urban plumes have been reported by Spicer and
20 co-workers. They have reported results for the Los Angeles, Phoenix, Boston and
21 Philadelphia urban plumes. In the Los Angeles studies, the transformation rate of NO2-to-
22 products was estimated at 5 to 15% h"1 (Spicer, 1977) and 5 to 10% h:1 (Spicer et al., 1979).
23 The sum of transformation plus removal rates was estimated for the Phoenix and Boston
24 plumes at <5% h"1 and 14 to 24% h"1, respectively. The low rate for Phoenix was
25 attributed partly to thermal decomposition of PAN after its formation in the plume. In a
26 study of the Detroit plume, Kelly (1987) estimated the NOX transformation rate at 10% h"1,
27 with 67 to 84% of the products being in the form of HNO3. Measured concentrations of
28 nitric acid, however, were much lower because of its higher removal rate. All of the above
29 urban plume studies, and most of the power plant plume studies, have been daytime studies.
30 Field measurements of nighttime chemistry of nitrogen oxides in plumes are almost non-
31 existent.
April 1995 3_105 DRAFT-DO NOT QUOTE OR CITE
-------
1 3,4.3.1.2 Background Field Studies
2 Attention is now focused on studies of aerosol formation in background air. The plume
3 studies have shown that the rates of oxidation of SO2 and NOX in the background represent
4 approximately the upper limit of the conversion rates in the plume. In non-humid,
5 moderately polluted conditions, they range typically between 1 and 5% h"1 for midday SO2
6 oxidation in summer in the eastern U.S. (depending on the variability of VOC/NOX and the
7 composition of VOC), and up to 1 % h"1 in the cleaner parts of the Western U.S. Winter
8 rates are about an order of magnitude lower. By contrast, observed NOX to nitrate
9 conversion rates are about three times faster in summer than in winter (Parrish et al., 1986).
10 Aerosol nitrate formation depends strongly on availability of NH3 and on temperature.
11 Background aerosol is generally more aged and its acidity more neutralized than plume
12 aerosol.
13 The situation is more complex in humid conditions. Field measurements of the
14 compositions of cloudwater, rainwater and the precursor clear-air aerosol have shown that
15 strong acidity is substantially greater in cloud and rain water than in the clear-air aerosol
16 (Daum et al., 1984; Lazrus et al., 1983; Weathers et al., 1988). This is indicative of the
17 contribution of aqueous-phase chemistry to cloudwater acidity in excess of that due to
18 scavenged aerosol. Based on climatological data of clouds and SO2 distribution, and
19 assuming aqueous-phase oxidation of SO2 by ozone, Hegg (1985) estimated contribution of
20 the aqueous mechanism to global tropospheric sulfate production to be at least 10 to 15 times
21 greater than that due to the gas-phase mechanisms. Applications of more comprehensive
22 global models have given estimated aqueous-phase contributions of 40 to 95 % of the total
23 sulfate production (Langner and Rodhe, 1991 and references therein). Regional models for
24 North America suggest 50 to 80% of the sulfate deposited in precipitation to be formed in
25 clouds (Fung et al., 1991; McHenry and Dennis, 1991).
26 A number of ambient studies have attempted to study aqueous chemistry based on in
27 situ measurements in clouds. Determination of the rates and mechanisms of aqueous-phase
28 chemistry is particularly ambiguous for several reasons. First, it is difficult to distinguish
29 between the contributions of in situ chemistry and aerosol scavenging to the observed
30 concentration of the solute in the droplet phase. Also, aqueous chemistry rate depends not
31 only on the change in concentration, but also on the change in time. It is difficult enough to
April 1995 3-106 DRAFT-DO NOT QUOTE OR CITE
-------
1 determine the difference in concentration of even one reactant or product species, but
2 determining the corresponding time difference is even more difficult (Schwartz, 1987; Gervat
3 et al., 1988; Kelly et al., 1989). In stratiform clouds, in particular, it is not always possible
4 to determine what constitutes pre-cloud air corresponding to specific cloud water samples
5 (Gillani et al., 1994). Finally, it is difficult, based on field data, to attribute the inferred
6 chemistry to specific mechanisms (oxidation by H2O2 or O3, etc.). The conclusions
7 regarding rates and mechanisms of aqueous chemistry based on measurements in clouds are
8 therefore quite uncertain, and have been a source of considerable controversy (e.g., Hegg
9 and Hobbs, 1982, 1983a,b versus Schwartz and Newman, 1983). One important finding in
10 support of in-cloud oxidation of SO2 by H2O2 , however, is the almost universal mutual
11 exclusion of these two species in non-precipitating stratiform clouds (Daum et al., 1984;
12 Daum, 1988). In such clouds, there is generally enough time available for the species to
13 react fully until the one with the lower concentration in the precursor air is depleted. The
14 implication is that the aqueous-phase oxidation of SO2 by H2O2 takes precedence over other
15 competing reactions.
16 Most field studies have been limited to estimating the amount or fraction of sulfate
17 formed by the aqueous pathway, rather than the rate of formation. Liu et al. (1993) have
18 summarized the results of a number of cloud studies between 1979 and 1991. In these
19 studies, a number of different approaches have been used to resolve the contributions of
20 aerosol scavenging and in situ chemistry to the observed cloudwater sulfate. The study of
21 Liu et al. (1993), which was part of the first intensive (summer 1988) of the Eulerian Model
22 Evaluation Field Study (EMEFS), used three different approaches for estimating the
23 scavenged fraction of observed sulfate, and attributed 27 to 55% of cloudwater sulfate to in
24 situ production. The inferred results for the aqueous-phase production of sulfate in the
25 collective studies vary widely. In winter studies, such production is low (e.g., Strapp et al.,
26 1988), while in summer studies, it is generally higher (e.g., Mohnen and Kedlacek, 1989).
27 Many studies implicate H2O2 as the principal oxidant (e.g., Van Valin et al., 1990), while
28 others implicate ozone (e.g., Hegg and Hobbs, 1986).
29 There is a variety of evidence for and against the formation of HNO3 in the cloud
30 environment (e.g., Lazrus et al., 1983; Daum et al., 1984; Hegg and Hobbs, 1986; Leaitch
31 et al., 1986a). The heterogeneous mechanism involving N2O5 has received attention mostly
April 1995 3_107 DRAFT-DO NOT QUOTE OR CITE
-------
1 as the "nighttime" mechanism (Lazrus et al., 1983; Richards, 1983) owing to the short life of
2 the NO3 radical (precursor of N2O5) in sunlight. To account for the comparable measured
3 amounts of sulfate and nitrate deposited in winter storms in Ontario, Barrie (1985) suggested
4 the possibility of the N2O5 mechanism for wintertime formation of nitrate in clouds. Leaitch
5 et al. (1988) found substantial enhancement of NO"3 in and near clouds on 8 of 12 days of
6 winter measurements in central Ontario under freezing conditions and low insolation.
7 On these occasions, variations in NOySO^" were associated with H+/SO4~ in the cloud water,
8 implicating HNO3. Also, the observed levels of NO"3 could not be simulated in a model
9 without invoking the N2O5 mechanism. Based on a detailed examination of the nighttime
10 behavior of the NO3 radical, Noxon (1983) concluded that there was a significant loss of
11 NO3 compared to N2O5 by an unknown scavenger (wet particles?). In measurements at a
12 rural site in central Ontario in August 1988 as part of EMEFS, Li et al. (1993) observed a
13 gradual increase in the concentration of aerosol nitrate (NO3~) from 1800 to midnight, and
14 then a gradual decrease. In a diagnostic model study, they concluded that the observations
15 could be explained by heterogeneous reactions of NO3 and N2O5 on wet particles. They
16 attributed more than 80% of the NO3" formation to NO3 and about 10% to N2O5 , and less
17 than 5% to HNO3.
18 In visibility studies, the water content of aerosols is of crucial importance. The
19 estimation of visibility impairment involves use of models in conjunction with ambient data
20 of both aerosols and relative humidity. Frequently, both sets of data are not available
21 concurrently for all stations in a monitoring network such as IMPROVE (Interagency
22 Monitoring of PROtected Visual Environments). In such cases, gaps in information must be
23 filled by the use of empirical relationships between average visibility impairment caused by
24 soluble aerosols and average relative humidity derived from the available concurrent data.
25 Such an application based on data at the 36 national IMPROVE sites is described by Sisler
26 and Malm (1994).
27 Another important area which critically involves water uptake by soluble aerosols
28 relates to aerosol-cloud interactions. Such interactions are a critical link in cloud formation
29 and the global water cycle, in cloud optics and the global energy budget, in pollutant
30 redistribution by clouds, in pollutant wet removal from the atmosphere, and in atmospheric
31 chemistry. Of particular importance is the process of aerosol scavenging by clouds.
April 1995 3-108 DRAFT-DO NOT QUOTE OR CITE
-------
1 Interstitial aerosols in clouds may become incorporated into cloud droplets by "activation"
2 (droplet nucleation), Brownian diffusion, inertial impaction, coalescence, and phoretic
3 effects. Of these microphysical cloud processes, aerosol activation is by far the most
4 important. A soluble particle (the CCN) is activated when water vapor supersaturation
5 around it (S) exceeds a critical value (Sc) which depends principally on particle dry size (D0)
6 and composition (commonly expressed in terms of the water-soluble solute fraction, s). The
7 works of Kohler (1936), Junge and McLaren (1971) and Hanel (1976) provide the underlying
8 theory for condensation of water on aerosols based on assumptions of internally mixed
9 aerosols. Based on properties of representative continental and marine CCN, Junge and
10 McLaren predicted that Sc would be sensitive to CCN size, but to CCN composition only for
11 10.1. Fitzgerald (1973) confirmed the insensitivity to e in the range 0.15 to 0.35 based on
12 simultaneous measurements of CCN size, e and CCN activation spectra (functional
13 dependence of activated fraction of aerosol on S) for S between 0.35 and 0.75%.
14 More recently, based on extensive year-long measurements of CCN spectra for
15 continental aerosols (representative of eastern U.S. background), separated into narrow size
16 bands within the accumulation mode, Alofs et al. (1989) derived a simple semi-empirical
17 expression relating Sc to D0 and e applicable down to S = 0.014%. They also showed,
18 based on their own data and a literature review, that for continental aerosols in industrialized
19 regions, e » 0.5 is a reasonable approximation, indicating that the activation of such
20 aerosols is unlikely to be sensitive to particle composition. Based on their expression for Sc
21 and using s = 0.5, a supersaturation of about 0.1% (characteristic for stratiform clouds)
22 would be adequate to activate most of the accumulation mode particles exposed to a cloud.
23 Cumuliform clouds with higher S would activate many Aitken mode particles also. In
24 cumulus clouds, peak supersaturation is typically attained near cloud base, which is where
25 maximum activation is likely to occur. The cloud module of the Regional Acid Deposition
26 Model (RADM) is based primarily on a cumulus parameterization, and makes the assumption
27 of 100% cloud scavenging efficiency for sulfates formed from the oxidation of SO2 (Chang
28 etal., 1990).
29 The principal interest in quantitative field studies of aerosol-cloud interactions is the
30 scavenging of acidic aerosol mass by clouds. The focus of measurements in these studies
31 (from aircraft or at fixed mountain sites) was on gross spatial averages (over 10s of km) of
April 1995 3.109 DRAFT-DO NOT QUOTE OR CITE
-------
1 species mass concentrations (mostly of sulfate and nitrate) based on batch samples collected
2 in cloud water, and in cloud and clear air (Scott and Laulainen, 1979; Sievering et al., 1984;
3 Daum et al., 1984, 1987; Hegg et al., 1984; Hegg and Hobbs, 1986; Leaitch et al., 1986b;
4 Pueschel et al., 1986). In some studies, continuous measurements of aerosol size spectra
5 were used to derive spatially-averaged aerosol volume concentrations (Leaitch et al., 1983;
6 Hegg et al., 1984; Heintzenberg et al., 1989) based on which, aerosol volume scavenging
7 efficiency was inferred. In one study, continuous measurements of light scattering
8 coefficient were used as a surrogate for aerosol mass concentration (ten Brink et al., 1987).
9 In these studies, inferences of the efficiency of aerosol scavenging were generally based on
10 comparisons of species mass or volume concentrations (or their surrogates) in cloud water
11 and/or cloud interstitial air with those in putative pre-cloud air. Such inferences can be
12 confounded by incorrect identification of pre-cloud air, non-Lagrangian sampling, extended
13 sampling periods and resultant averaging of spatial inhomogeneities (including clear air
14 pockets within clouds), and inadequately resolved contributions of aqueous-phase chemistry.
15 Not surprisingly, the results of the above studies varied quite widely. Most commonly,
16 however, mass scavenging efficiency was found to be high (>0.8).
17 The above studies based on spatially-averaged particle mass concentrations could not
18 address the issue of main concern with respect to radiative transfer, namely, the partitioning
19 of cloud particles between droplets and interstitial aerosol in terms of their local number
20 concentrations. Field studies focused on aerosol scavenging based on particle number
21 concentrations are relatively scarce. In the study of Leaitch et al. (1986) for stratiform and
22 cumuliform clouds, the authors took special care to ensure Lagrangian adiabatic
23 interpretation by comparing the instantaneous cloud droplet number concentration at a single
24 location within the adiabatic updraft core near cloud base with the below-cloud aerosol
25 number concentration. They found that activation efficiencies so defined were generally high
26 when pre-cloud AMP concentrations were less than about 750 cm"3, but dropped off non-
27 linearly at higher particle loading. Raga and Jonas (1993) made a similar observation when
28 comparing droplet concentrations near cloud top with the sub-cloud aerosol concentrations on
29 the assumption that the latter represented the pre-cloud condition.
30 Gillani et al. (1994) demonstrated that such an assumption was not generally valid in
31 stratiform clouds which are layered and may include sharp inversions decoupling the layers
April 1995 3-110 DRAFT-DO NOT QUOTE OR CITE
-------
1 from each other and from the sub-cloud layer. For such clouds, the adiabatic assumption
2 made in 1-D cloud models is not generally valid. To circumvent this difficulty with respect
3 to identification of pre-cloud air, Gillani et al. defined fractional activation (F) in terms of
4 local variables only, as the ratio of cloud droplet concentration (activated particles) to total
5 particle concentration (droplet concentration + concentration of unactivated accumulation-
6 mode particles, 0.17 to 2.07 jum diameter). In their study (aircraft measurements in and near
7 stratiform clouds near Syracuse, NY in the Fall of 1984), continuous in situ measurements
8 were available for particle number concentrations in 15 size classes each for the droplets and
9 for dried (by heating the probe inlet air) interstitial aerosols. Thus, they were able to
10 determine F at a high spatial resolution throughout the clouds studied (continental stratiform).
11 It was determined that accumulation-mode particles larger than 0.37 ^m were efficiently
12 activated in the cloud under all measurement conditions, but that particles in the range 0.17
13 to 0.37 jum were often activated only partially. Partial activation generally correlated with
14 high local total particle concentration (> 600 cm"3) and with low temperature lapse rate
15 (surrogate for cooling rate with ascent, dT/dt = w. dT/dz, where w= the mean long-wave
16 updraft speed), the two conditions most responsible for limiting supersaturation. It is
17 important to note that w is a most difficult quantity to measure, and is not generally available
18 in field measurements. Under the most polluted conditions in a stable stratus, fractional
19 activation of the accumulation-mode particles was as low as 0.1 in the core of the cloud.
20 Statistically, based on ten days of measurements in the Syracuse study, it exceeded 0.9 in
21 36% of the data in cloud interior, but was below 0.6 in 28% of such data. It was generally
22 quite low in cloud edges. Evidently, the assumption made in RADM of total activation is
23 questionable for stratiform clouds.
24 Simple parameterizations of fractional activation in clouds have been developed based
25 on 1-D adiabatic Lagrangian models (e.g., Twomey, 1959; Ghan et al., 1993), and generally
26 highlight the significance of particle loading and updraft speed (model calculated). The 1-D
27 adiabatic approach is useful near cloud base and in updraft cores, but it breaks down near
28 cloud edges and in the upper portions of clouds where entrainment and mixing effects are
29 substantial. It is also questionable in the presence of additional complexities such as cloud
30 layering (Gillani et al., 1994) and lifting and sinking motions (Baker and Latham, 1979;
31 Pruppacher and Klett, 1978). These complex effects result in three-dimensional spatial
April 1995 34 H DRAFT-DO NOT QUOTE OR CITE
-------
1 inhomogeneities and multi-modal droplet size spectra which are uncharacteristic of the simple
2 adiabatic model.
3 Noone et al. (1992) studied activation in ground fogs. They were able to infer size-
4 segregated volume and number scavenging efficiencies of aerosols (using a counterflow
5 virtual impactor) in the fog under conditions of very high particle loading and extremely low
6 supersaturations. For such highly-polluted fog conditions, they found high activation
7 efficiencies (>0.8) only for particles larger than 0.8 pcm.
8 In most cloud and fog studies which include considerations of particle composition, use
9 is made of the concept of water-soluble mass fraction (s). This implicitly assumes internally
10 mixed particles. As was shown by Zhang et al. (1993), there may really be two e's, one
11 (£m) for the "more" hygroscopic particles, and one (gj) for the "less" hygroscopic aerosols.
12 In the diagnostic modeling study of Pitchford and McMurry (1994), the two-e concept was
13 implemented. For clouds and fogs, this implies that Sc may be different for different
14 particles in the same size range.
15 The interaction between aerosols and clouds modifies not only the clouds, but also the
16 aerosols. The condensation-evaporation cycling of aerosols through non-precipitating clouds
17 generally results in growth of the nuclei due to microphysical and chemical processes during
18 their in-cloud residence (Hoppel, 1988; Hoppel et al., 1990).
19
20
21 3.5 DRY DEPOSITION
22 3.5.1 Theoretical Aspects of Dry Deposition
23 Dry deposition is commonly parameterized by the deposition velocity, Vrf (m s"1) which
24 is defined as the coefficient relating the pollutant deposition flux F (g m'V1) and the
25 pollutant concentration c (g/m3) at a certain reference height above the surface, i.e.,
26
F = Vdc (3-39)
27
28 The deposition velocity can be expressed as the inverse of a sum of "resistances" in three
29 layers adjacent to the surface (Sehmel, 1980; Hicks, 1982):
30
April 1995 3-112 DRAFT-DO NOT QUOTE OR CITE
-------
1 1. The aerodynamic layer, i.e., the layer in which atmospheric turbulent fluxes are
2 constant (typically extending to about 20 m above the ground). In this layer,
3 pollutant transfer, whether gas or particle, is controlled by atmospheric
4 turbulence.
5 2. The surface (or quasi-laminar) layer, a thin layer (~ 1 mm) just above the
6 surface in which transport occurs by molecular diffusion. In this layer, gases
7 transfer to the surface by molecular diffusion and particles undergo Brownian
8 diffusion and inertial impaction.
9 3. The earth/canopy/vegetation surface, the actual pollutant sink
10
11 For gases, the deposition velocity is a function of these three types of resistance as
12 follows:
VH = (r + r + r V1
d v a s CY
13
14 where ra is the atmospheric resistance through the aerodynamic layer, rs is the surface layer
15 resistance, and rc is the canopy/ vegetation resistance. All resistances are in units of s m "J.
16 The aerodynamic resistance ra can be expressed (Wesley and Hicks, 1977) by:
17
18
19 where zs is the reference height (m) (~ 10 m), z0 is the roughness length (m), k is the von
20 Karman constant (0.4), «* is the friction velocity (m s"1), and fh is the stability correction
21 factor. Roughness lengths vary from about 10"5 m for very smooth surfaces (ice, mud flats)
22 to 0.1 m for fully grown root crops, to 1 m for a forested area, to 5-10 m for an urban core
23 (Seinfeld, 1986).
24 The surface layer resistance can be parameterized as a function of the Schmidt number
25 Sc = v/D, where v is the kinematic viscosity of air (m2/s) and D is the molecular diffusivity
26 (m2/s)of the species, as
April 1995 3-113 DRAFT-DO NOT QUOTE OR CITE
-------
rs = d,- (3-42)
s 'ku*
1
2 where dlt d2 are empirical parameters (d1 =~1.6 - 16.7, and d2 =0.4-0.8, with a
3 suggested choice of d\ = 5, d2 = 0.66).
4 The canopy resistance rc for a gaseous species can be parameterized (Yamartino et al.,
5 1989) as:
rc = [LAAI/rf + LAI/rcut + l/r^"1 (3-43)
6
7 where LAI is the leaf area index (i.e., the ratio of leaf surface area divided by ground surface
8 area), /y is the internal foliage resistance, rcut is the cuticle resistance, and r is the ground
9 or water surface resistance. Values for /y are discussed by O'Dell et al. (1977). The
10 resistance rcut is parameterized by Pleim et al. (1984).
11 For gaseous pollutants, solubility and reactivity are the major factors affecting surface
12 resistance and net deposition velocity. For particles, the factor most strongly influencing the
13 deposition velocity is the particle mass or, assuming similar densities, the particle size.
14 Particles are transported toward the surface by turbulent diffusion, which for larger particles
15 is enhanced by gravitational settling. Across the quasi-laminar surface layer very small
16 particles (< 0.05 /urn diameter) are transported primarily by Brownian diffusion, analogous
17 to the molecular diffusion of gases. The larger particles possess inertia, which may enhance
18 the flux through the quasi-laminar sublayer.
19 The downward pollutant flux is the sum of the turbulent diffusive flux and a flux due to
20 gravitational settling, i.e.,
F(z) = Fd + VgC = VdC (3-44)
21
22 where Vg is the gravitational settling velocity of the particle. Whereas in the formulation of
23 the algorithm for gases the analogy with electrical resistance is straightforward, it is less so
24 for particles. This is because at any height within the aerodynamic layer and surface layer
25 the flux of trace gases is diffusive only and hence a function of the concentration gradient.
26 Consequently, when equating the fluxes through each layer under the steady-state
27 assumption, the deposition velocity may be cast in a form proportional to the inverse of a
April 1995 3414 DRAFT-DO NOT QUOTE OR CITE
-------
1 sum of resistances. Nevertheless, the electrical resistance analogy can still be employed for
2 particles. The gravitational settling velocity is merely represented by the reciprocal of an
3 additional resistance acting in parallel with the diffusive resistance.
4 As noted earlier, for particles, the resistance in the vegetation layer (rc) is usually
5 assumed to be zero, since particles that penetrate the surface layer are assumed to stick to the
6 surface. The expression for deposition velocity in terms of the resistances, modified to
7 include gravitational settling, is
8
Vd = (ra + rs + r^Vg)'1 + Vg (3-45)
9
10 Therefore, the deposition velocity of particles may be viewed in terms of electrical
11 resistance as the reciprocal of three resistances in series (ra, rs, and /y^V) and one in
12 parallel (1/VJ. The third resistance in series is denoted here as a virtual resistance in view
o
13 of the fact that it is a mathematical artifact of the equation manipulation and not a physical
14 resistance. Equation (3-109) is usually implemented with ra (particles) equal to ra (gases), in
15 which ra is computed by Equation 3-105, and the surface layer resistance is
16
rs = (Sc~2/3 - l(T3'Vu *-1 (3-42)
17
18 where Sc is the Schmidt number based on D, the Brownian diffusivity of the particle in air,
19 and St is the Stokes number, St=F'u2lgn. The surface layer resistance incorporates the
o
20 effects of both Brownian diffusion, through the Schmidt number, and inertial impaction
21 effects, through the Stokes number.
22 The gravitational settling velocity Vg is a function of the particle size, shape and
23 density. For spherical particles (Seinfeld, 1986),
24
v - dp2g(pp - pa)C (3.46)
g "
25
April 1995 3415 DRAFT-DO NOT QUOTE OR CITE
-------
1 where d is the particle diameter (m), p is the particle density (g /m ), p& is density of the
2 air (g /m3), m is the viscosity of air (g rrf's"1), and C is the slip correction factor
3
C = 1 + (2X/dp)[1.257 + 0.4exp(-0.55dp/A)] (3-47)
4
5 where X. is the mean free path of air molecules (k = 6.53 x 10"6 cm at 298K)
6 Figure 3-12 shows particle deposition velocities based on wind tunnel measurements.
7 Deposition velocities are presented as a function of particle diameter, particle density, and
8 surface roughness height. Particle deposition velocities exhibit a characteristic minimum as
9 a function of particle size. For the smallest particles, deposition velocity increases as
10 particle size decreases because diffusion by Brownian motion increases as particles get
11 smaller. For the largest particles, gravitational settling becomes important as particles get
12 larger so the deposition velocity increases as particles increase in size. A characteristic
13 minimum in deposition velocity results in the range of 0.1 to 1.0 jim diameter where neither
14 Brownian diffusion nor gravitational settling is strong enough to control removal.
15 It is possible to obtain a rapid estimate of the atmospheric lifetime of particles with
16 respect to removal by dry deposition. If the aerosol can be assumed to have a uniform
17 concentration between the ground and a height h, then the residence time relative to
18 removal by dry deposition is h/Vd. For example, for a 1000 m atmospheric layer, and a
19 particle deposition velocity of 0.1 cm/s, the estimated residence time is 11.5 days.
20
21 3.5.2 Field Studies of Dry Deposition
22 In spite of many field measurements and considerable progress since 1980 in our
23 understanding of dry deposition processes and their quantification, uncertainties remain
24 substantial. The problem is extremely complex involving a large multiplicity of factors,
25 and their complex interactions, which influence dry deposition of atmospheric particles and
26 their precursors (see, for example, a tabulation of some of these in Davidson and Wu,
27 1990). These factors relate to characteristics of the atmosphere, nature of the
28 depositionsurface, and properties of the depositing species. It is impossible in field studies
29 to measure all the pertinent variables over large enough spatial and temporal domains. In
April 1995 3-116 DRAFT-DO NOT QUOTE OR CITE
-------
_ 5 ton truck (8-11 ms'1 speed)
0.1 0.2 0.4 124 10 20 40
Diameter, urn
100 200
Figure 3-12. Extrapolations from correlations of windtunnel measured deposition
velocities for z = 1 m, densities of 1, 4, and 11.5 g cm"3. VT represents
terminal settling velocity.
Source: Sehmel, 1980, as presented by Nicholson, 1988.
April 1995
3-117
DRAFT-DO NOT QUOTE OR CITE
-------
1 essence, knowledge of dry deposition is limited by the inability to make the necessary
2 measurements in other than special circumstances. This was a key statement of the NAPAP
3 Workshop on Dry Deposition in Harpers Ferry, West Virginia (Hicks et al., 1986). The
4 Workshop report also noted that there is presently a lack of fundamental knowledge
5 concerning the chemical and biological processes influencing dry deposition, and there are
6 serious hazards associated with scaling input information down from grid level to local, and
7 scaling up the results of local measurements to broader domains. Information contained in
8 the Workshop report and in subsequent research publications on the subject were reviewed
9 by Davidson and Wu (1990, henceforth to be referred to as "DW90"). That review
10 summarizes the results of a large number of field studies published since earlier reviews by
11 McMahon and Denison (1979), Sehmel (1980), Hosker and Lindberg (1982) and Galloway
12 et al. (1982). It also includes summaries of dry deposition processes, wind tunnel studies
13 and empirical models, techniques for measuring deposition in the field, and comparisons of
14 field data and model results. The summary presented in this section is based largely on
15 DW90.
16 A large number of techniques have been used in measurements of dry deposition.
17 They are generally grouped into two classes: surface analysis methods, which are based on
18 examination of contaminant accumulations on natural or surrogate surfaces, and
19 atmospheric flux methods, which involve ambient measurements of the species of interest
20 and other related variables. These methods provide the deposition flux out of which the
21 deposition velocity is inferred. Surface analysis methods include foliar extraction (by
22 washing individual leaves), throughfall and stemflow (wet measurements above and within
23 the canopy), watershed mass balance, tracer techniques, snow sampling, collection on
24 surrogate surfaces, etc. These methods may provide useful data on the flux of coarse
25 particles, but fail to simulate the physical processes which control the deposition of sub-
26 micron particles to natural surfaces, and to give meaningful data on trace gas deposition.
27 Deposition on surrogate surfaces may not mimic that on natural surfaces. Atmospheric flux
28 methods include micrometeorological methods (eddy correlation and vertical gradients),
29 aerometric mass balance in a box over the depositing surface, tracer techniques, etc.
30 Micrometeorological methods also include what has come to be known as the inferential
31 approach in which measured concentrations are combined with specified or calculated
April 1995 3-118 DRAFT-DO NOT QUOTE OR CITE
-------
1 deposition velocities based on meteorological data and surface information. This approach
2 is used in long-term monitoring programs in which only simple measurements are possible
3 at remote sites (e.g. weekly average species concentrations and routine meteorological
4 measurements). For details of the various methods, see DW90 and the Workshop report.
5
6 3.5.2.1 Measured Deposition Velocities
7 Measurements of dry deposition in the field and in chambers have primarily involved
8 six categories of contaminants: sulfur species, nitrogen species, chloride species, ozone,
9 trace elements and atmospheric particles. The results of many of these studies published
10 between 1978 and 1987 are reviewed in DW90, which includes extensive tabulations of the
11 studies and their results. Of the reported studies on Sulfur Species, 20 pertain to SO2.
12 They give deposition velocities ranging from nearly 0 to 3.4 cm/s. The variations are due
13 to differences in seasonal and diurnal conditions, aerodynamic transfer, surface
14 characteristics (especially stomatal resistance), measurement methods, etc. Daytime values
15 are generally higher, as expected (lower aerodynamic and stomatal resistances).
16 Micrometeorological methods were used in 16 studies whose average values of vd gave a
17 grand average of 0.95 ± 0.62 cm/s. Four studies provided an average value of 0.13 ± 0.09
18 cm/s for deposition velocity on snow. For particulate sulfur, 34 studies are included, with
19 10 also including particle size measurements. A graph also includes results of earlier
20 studies, and gives values of vd in the range 0.01 to 10 cm/s. Results for vd in cm/s based
21 on different methods are as follows: 0.55 ± 0.65 for micromet methods, 0.26 ± 0.25 for
22 surrogate surface exposures, 0.23 + 0.24 for foliar extraction, and 1.00 ± 0.41 for
23 throughfall. Since the micromet method is believed to be more specific for submicron
24 particles while the surrogate surface method is biased in favor of larger particles, the
25 difference in the results of those methods is opposite to that expected. The surrogate
26 surface and foliar extraction results are close, but each has a large variance. Throughfall
27 values are the largest probably partly because they include deposition of SO2. Evidently,
28 measurement methods themselves are an important variable because they do not measure
29 the same thing.
30 Twenty two species are reported for Nitrogen Species, including NO2, NOX, HNO3,
31 NO3''NH3, and NH4+. The inferred values of vd (cm/s) are: 0.012 to 0.5 for NO2 (2
April 1995 3_H9 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies), -2.6 to 0.3 for NOX (4 studies), 0 to 2.9 for HNO3 (4 studies), 0.13 to 1.3 for NO3
2 ' (7 studies), 1.9 ± 1.55 for NH3 (1 study), and 0.06 to 1.0 for NH4+ (4 studies). The zero
3 value for nitric acid was for snow in a chamber study; otherwise, the values for nitric acid
4 are the highest, indicating low surface resistance. The values for particulate nitrate are
5 somewhat larger than for sulfate; this may reflect larger particle size associated with nitrate.
6 Four studies are reported for chloride-containing particles, giving values of 1.0 to 5.1 cm/s;
7 a value for HC1 of 0.73 cm/s on dew was obtained in one study. The highest values for
8 chloride were in winter, related to road salt. Based on 11 studies using micromet methods,
9 vd of ozone on vegetation ranged between nearly 0 and 1.5 cm/s (average of 15 values =
10 0.39 ± 0.21). Nighttime values were lower, but the day-night difference was less for ozone
11 than for NO2.
12 Results of 19 studies included measurements for 21 trace elements, with particle size
13 data in 15 studies. For these data, crustal element enrichment factors (EF) were
14 determined. Values of EF « 1 indicate crustal sources, while EF > 1 (enriched) indicate
15 non-crustal sources such as anthropogenic, natural combustion (volcanism, forest fires),
16 biogenic, sea-spray, etc.. Large enrichment factors were found for Ag, As, Cd, Cu, In, Pb,
17 Sb, Se and Zn. Ni and V were marginally enriched. Other elements were mainly soil-
18 derived. vd for these elements were generally higher (>1 cm/s), while they were generally
19 less than 1 cm/s for the enriched elements (smaller, submicron particles). A figure
20 including these as well as data of earlier studies is presented, showing a positive correlation
21 between vd and MMD (mass median diameter). For Pb, the values ranged between 0.1 and
22 1.0 cm/s. Friedlander et al. (1986) have used CO as a tracer for automobile emissions to
23 estimate the deposition velocity for Pb, by comparing the ratio Pb/CO in ambient air to that
24 in a tunnel. They found the former to be lower, indicating deposition compared to its value
25 in fresh emission (tunnel). Based on these data, they estimated vd for Pb to be 0.26 cm/s,
26 which is consistent with the range given above. DW90 also report the results of 5 other
27 field studies with micromet measurements of dry deposition for submicron particles, and
28 particle size measurements also. vd was generally less than 1 cm/s, in general agreement
29 with results for sulfate and the enriched trace elements.
30 DW90 have also presented results of comparisons between measured values ofvd
31 with predicted values based on six model calculations. These results are from published
April 1995 3_120 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies with size distribution data for aerosol sulfate and trace elements. The measured
2 values of vd are for the full size range; the model value is the concentration-weighted
3 average of the calculated values for all size classes. For sulfate, the predicted values were
4 generally smaller than the measured values. Good agreement was, however, not expected
5 because of differences in ambient conditions and surface conditions between values used in
6 the model compared to the corresponding measured values. Similar comparisons for 24
7 trace elements were also tenuous: out of 11 of the 24 elements for which more than one or
8 two data points only were available, the measured values were in the predicted range; for
9 Al, Ca and Fe, the predictions were low, while for Zn, the predictions were too high. For
10 the other 13 elements with sparse measured data, the agreement was generally much poorer.
11
12
13 3.6 WET DEPOSITION
14 3.6.1 Introduction
15 Although detailed physico-chemical models are needed to describe the details of
16 in-cloud and below-cloud scavenging of particles, there has been a benefit in using
17 comparatively simple formulations of precipitation scavenging that provide a convenient
18 picture of the process as a whole. These simple methods are not designed to explain
19 detailed variations in wet deposition with time or space, but they are useful in describing
20 average deposition rates over large areas. Two alternative techniques have become popular.
21 The first relates concentrations of material in precipitation to the quantity available in the
22 air, thus describing the overall efficiency of precipitation as a removal path. By relating
23 concentrations in precipitation to those in the air, dimensionless scavenging ratios can be
24 determined. The second common method is based on the first-order removal of airborne
25 gases or particles as rain falls through the atmosphere. Concentrations in the air will
26 decrease exponentially and a scavenging rate can then be determined.
27 Below-cloud scavenging rates for particles of about 3x10"5 s"1 appear to be typical;
28 in-cloud scavenging leads to rates typically ten times larger (Hicks and Meyers, 1989).
29 Hygroscopic particles are scavenged more readily than hydrophobic ones.
30 Based on the wet flux W the wet deposition velocity may be defined as
31
April 1995 3-121 DRAFT-DO NOT QUOTE OR CITE
-------
Vw = A ~Ah (3-48)
w c(x,y,0,t)
2 where the last equality assumes that the pollutant is uniformly distributed between z - 0
3 and z =h. The wet deposition velocity Vw can be computed by
4
Vw = wrPo (3-49)
5
6
7 where wr is the washout ratio (i.e., the dimensionless ratio of the concentration of material
8 in surface-level precipitation to the concentration of the material in surface-level air) and p0
9 is the precipitation intensity (mm hr"1). For example, if \vr - 106 and p0 = 1 mm h"1, then
10 Vw = 28 cm s"1, which gives, for h = 1,000 m, A = 2.8 x 10'4 s'1. Seinfeld (1986)
11 provides a detailed discussion of precipitation scavenging of particles, including the
12 calculation of collision efficiencies and scavenging rates.
13 Scavenging ratios relate concentrations in precipitation to those in air. Although such
14 ratios depend on many factors, they provide a simple way to include wet deposition
15 processes in air quality models. The washout (or "scavenging") ratio is
16
Wr = [c]fainPa (3-50)
LCJaerosol
17
18 with [c]rain in mg g"1, [c]aerosol in mg m"3, and ra (=1,200 g m"3), the density of air. The
19 definition of this ratio presumes that the aerosol measured at ground level is vertically
20 uniform and that there are no factors limiting the collection of aerosol by the droplets, such
21 as solubility. Scavenging ratios of about 400 appear to be appropriate in the case of
22 particles well mixed in the lower atmosphere but originating near the surface, while values
23 of about 800 appear characteristic of material derived from the free troposphere (Hicks and
24 Meyers, 1989).
25
26
April 1995 3-122 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.6.2 Field Studies of Wet Deposition
2 The removal of pollutants from the atmosphere by precipitation is the terminal step of
3 a three-step sequence. In the first step, the pollutant or its precursor(s) must be delivered
4 by transport processes to the precipitating cloud .or to the air below it; in the second step,
5 the species must become incorporated into the precipitating droplets either within the cloud
6 (in-cloud scavenging) or below it (washout). We have already addressed field
7 measurements of transport and in-cloud scavenging in liquid water clouds. The focus here
8 will be on field studies of the third step. Wet deposition measurements are made
9 principally to meet three objectives: (1) to determine the regional spatial-temporal
10 distribution and chemistry of wet deposition; (2) to study pathways and mechanisms of
11 pollutant wet removal from the atmosphere; and, (3) to generate data for diagnostic
12 evaluations of precipitation scavenging modules. The first of these objectives is best
13 studied based on data of routine monitoring programs. These are reviewed in detail in
14 SOST Report No. 6 of NAPAP (1991), and are not covered here. Our focus here is on
15 recent research field studies aimed at objectives (1) and (2) above.
16 The scavenging coefficient and the scavenging ratio, in common use in the Lagrangian
17 models of the 1970s, represent highly lumped representations of the complex of processes
18 involved in wet removal. They are empirical entities which, by themselves, contain little
19 mechanistic information. While reporting their measurements of scavenging ratio during a
20 year-long study in Paris, Jaffrezo and Colin (1988) included a table (Table 3-14) which
21 summarized not only their own data but also those of other earlier studies. The various
22 results are not directly comparable owing, at least partly, to differences in measurement
23 methods. Of particular interest in their study is the interpretation of elemental composition
24 data. They were able to separate the measured elements into three groups which differed in
25 terms of their solubility and also, by the mechanisms of their scavenging. The measured
26 concentrations in precipitation and in air were nearly proportional for the insoluble species
27 Al, Si, and Fe; this was interpreted to imply that their scavenging was mostly a local
28 mechanism (below-cloud impaction). At the other extreme, the local concentrations of the
29 very soluble species Na and Cl in the two phases were least correlated, indicating a more
30 complex and progressive process of enrichment of one medium relative to the other (in-
April 1995 3.123 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-14. SCAVENGING RATIOS (DIMENSIONSLESS)
t— '
1
?3
-t*.
O
J>
H
6
o
^
s
o
CJ
O
H
W
Reference n G.M.
Cl 78 2,941
S 82 743
Na 81 444
K 82 951
Mg 81 596
Ca 82 1,048
Zn 69 767
Al 82 291
Si 82 373
Fe 82 184
Ti 9 305
Mn 7 146
1. Jaffrezo and Colin, 1988
2. Harrison and Pio, 1983
3. Arimotoet al., 1985
4. Buat-Menard and Duce, 1986
5. Lindberg, 1982
6. Gatz, 1977
7. Chanet al., 1986
8. Peirsonet al., 1973
9. Cawse, 1981
10. Savoieet al., 1987
* Non-sea sulfate
G.M. Geometric mean
A.M. Arithmetic mean
Med. Median
S.D. Geometric standard deviation
Med.
(1)
2,917
753
530
970
682
1,097
707
283
405
194
S.D.
4.73
1.98
3.17
2.30
2.39
2.49
2.65
2.72
2.35
2.51
1.30
1.36
A.M.
7,710
940
744
1,325
816
1,579
1,226
459
533
267
378
171
A.M.
(2)
600
700
560
620
850
1,890
G.M. G.M. A.M. A.M. G.M. A.M. A.M. A.M.
(3) (4) (5) (6) (7) (8) (9) (10)
350 1,400 2,300 4,100
1,000 370*
360 2,100 2,900 5,500 490
300 2,000 548
400 457
320 1,100 352 2,100
790 820 179 612 1,050 1,030
580 1,300 756 620 430
390 600 253 468 890 270
250 2,100 3,600 370 756 760
n
i—i
H
W
-------
1 cloud processes). The remaining soluble species (SO", K, Ca, Zn, and Mg) showed an
2 intermediate behavior. Earlier data at the same site of the relationship between scavenging
3 ratio and particle mass median diameter (MMD), which showed a minimum in the
4 scavenging ratio for MMD = 1 to 2 ptm (reported as Figure 6-1), were judged to be
5 supportive of the above interpretation.
6 A significant effort in NAPAP in the 1980s was devoted to development of wet
7 removal characterizations that directly reflected the cloud physics, attachment, reaction, and
8 precipitation processes (Hales, 1991). The PLUVIUS II models, prepared under the
9 auspices of NAPAP, was a reactive storm model based on multi-phase material balance,
10 and served as the basis for the development of the one-dimensional RADM Scavenging
11 Module, RSM. A parallel activity in NAPAP was DOE's PRECP (Processing of Emissions
12 by Clouds and Precipitation) field measurements program which comprised a series of six
13 individual intensive field studies with the objective of systematically measuring scavenging
14 characteristics for different classes of storm systems important to regional acid deposition.
15 In these, studies, the emphasis was on in situ aircraft measurements. What follows is a
16 brief review of such research field studies. It is based substantially on Hales (1991). In the
17 context of precipitation scavenging studies, it is useful to bear in mind that pollutant
18 particles, on average, undergo a number of repeated cycles in and out of non-precipitating
19 clouds before finally being removed by precipitation.
20 In situ aircraft measurements in clouds and precipitation are of crucial importance in
21 mechanistic/diagnostic studies. Current technology permits continuous aircraft
22 measurements of NO, NO2, NOy, HNO3, PAN, SO2, O3, H2O2, liquid water content
23 (LWC), and size-segregated aerosol and cloud/rain droplet concentrations with quite high
24 sensitivity and precision. In addition, filter samples and cloudwater samples can provide
25 mass concentrations of the major ions in aerosols and droplets at a temporal resolution of a
26 few minutes. Ground monitoring of precipitation in recent studies has included use of the
27 NAPAP-developed Computer-Controlled Automated Rain Sampler (CCARS) which is a
28 combination rain gauge and sequential precipitation chemistry sampler, controlled and
29 monitored by a programmable microprocessor. Such samplers permit capture of statistically
30 valid footprints (multiple sequential event samples) of deposition during the course of a
31 storm. Upper-air meteorological measurements with fine vertical resolution of wind
April 1995 3_125 DRAFT-DO NOT QUOTE OR CITE
-------
1 components, temperature and moisture are also important. These can be made using radar
2 profilers and doppler radars.
3 Field studies have been conducted in and below point-source plumes (meso-y scale)
4 and urban plumes (meso-p scale). In the former, precipitation scavenging of S and N
5 compounds was found to be minimal (Granat and Soderland, 1975; Dana et al., 1976;
6 Drewes and Hales, 1982), indicating low precipitation scavenging efficiency for SO2 and
7 NOX from fresh plumes. Hales and Dana (1979) found appreciable removal of S and N
8 compounds from the urban plume of St. Louis by summer convective storms. Patrinos and
9 Brown (1984), Patrinos (1985) and Patrino et al. (1989) found efficient scavenging of these
10 compounds from the urban plumes of Philadelphia and Washington, DC by frontal storms.
11 H2O2 data in rain showed considerable spatial variability in the plumes.
12 The major regional-scale field studies include OSCAR (Oxidation and Scavenging by
13 April Rains, April 1981), PRECP (mid-1980s), and the DOE-FBS (Frontal Boundary
14 Study). OSCAR (Chapman et al., 1987) included a nested array of ground level sampling
15 (an extended regional precipitation chemistry network in northeastern United States, with an
16 embedded high-density network in northeast Indiana) as well as three research aircraft. The
17 focus was on scavenging by extratropical cyclonic storms. The aircraft made clear air
18 measurements before and after frontal passage, as well as measurements within the storm,
19 in the vicinity of the high-density network. Measurements were made during four storms.
20 OSCAR data have been used for regional model development and evaluation.
21 The six PRECP studies, conducted between 1984 and 1988, were targeted at
22 scavenging measurements in different types of storm systems. Three studies were focussed
23 on convective storms (II, V, and VI) in summer, and the other three on extratropical
24 cyclonic and frontal storms during other seasons; five were conducted east of the
25 Mississippi River, and one in the Oklahoma-Kansas-Colorado area. All of them included
26 two or more research aircraft, and all also included at least limited area precipitation
27 chemistry networks (PRECP IV had three multiscale networks ranging from a coastal "rain-
28 band" network to a truly regional scale network). The network in PRECP VI was a highly-
29 density network within an 80-km RADM grid cell, designated to provide information about
30 regional sub-grid scale variability. Two of the studies were conducted jointly with other
31 meteorology-focussed measurement programs; PRECPII with the NSF PRESTORM study,
April 1995 3-126 DRAFT-DO NOT QUOTE OR CITE
-------
1 and PRECP IV with the NSF-NASA-NOAA Genesis of Atlantic Lows Experiment
2 (GALE). Such synergism resulted in particularly strong meteorological data in these two
3 studies. PRECP I was intended to be an exploratory study, but generated a database of
4 which at least one storm has been extensively studied (Saylor, 1989). PRECP VI, on the
5 other hand was designed as the grand finale aimed at generating a definitive database for
6 evaluation of the RADM Scavenging Module, but failed to meet its main objective owing
7 to the extreme drought of the summer of 1988.
8 Overall, the studies have developed a substantial database of mechanistic-diagnostic
9 information suitable for diagnostic model studies. PRECP II definitively demonstrated the
10 cloud venting phenomenon transporting boundary layer pollutants to considerable heights in
11 the free troposphere (Dickerson et al., 1987). PRECP III provided a significant new
12 mechanistic insight regarding scavenging in orographically enhanced storms, e.g., the
13 observation of an unexpected entrainment mechanism that occurs as orographic lifting
14 occurs, and which enhances chemical wet removal appreciably (Hales, 1991). PRECP V,
15 focussed on studying vertical profiles of chemical species in and around convective storms,
16 resulted in one study (Daum et al., 1990) which showed that while SO2 was more
17 concentrated in the lower parts of the ABL, H2O2 was concentrated near the top,
18 underscoring the importance of mixing in facilitating aqueous-phase of SO2 by H2O2. The
19 same study also found that in the low-NOx background, H2O2 was correlated with humidity.
20 The Frontal Boundary Study (DOE) was conducted in fall 1989 as part of a global
21 study of the fate of energy-related pollutants. The focus was on pollutant redistribution and
22 removal by stable frontal storms occurring subsequent to pollution episodes associated with
23 high-pressure stagnation. Aircraft soundings ahead of, within, and following the passage of
24 the front showed considerable spatial variability in precipitation amount and composition
25 (Hales, 1991).
26 The data of the above studies constitute a substantial mechanistic-diagnostic database
27 for model evaluation. In addition to these research studies, a number of research-grade
28 precipitation chemistry networks were also operated in the 1980s. They include the
29 Canadian CAPMON, and the U.S. MAP3S and UAPSP, as well as the shorter-term EPRI-
30 OEN and the EPA-ME35. Applications of the research network measurements for source-
31 receptor pathway studies are discussed by Hales et al. (1987).
April 1995 3_127 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.6.2.1 Overview of SO2 and NOX Wet Scavenging
2 Hales (1991) has presented a useful overview of our understanding of SO2 and NOX
3 wet scavenging based on field measurements which is very pertinent here, and is recapped
4 below. SOX: SO2 is emitted principally from point sources. It is moderately soluble in
5 water, and its solubility decreases with increasing acidity of the solution. It is not
6 efficiently scavenged from concentrated fresh plumes, but this efficiency improves as the
7 plumes dilute. It is essentially insoluble in ice and cold snows, but tends to be more
8 efficiently scavenged by wet slushy snow and snows composed of graupel formed by
9 rimming of supercooled cloud water. Only a small fraction of the SO2 emission is removed
10 as unreacted S(IV) which constitutes about 20% of S in precipitation in the eastern U.S. in
11 sold seasons (significantly in the form of hydroxymethane sulfonate ions), and virtually
12 none in summer (high acidity of droplets). Sulfate removal is also small from fresh plumes
13 (not much there), but increases substantially with plume dilution as more is formed in the
14 plume. It is scavenged efficiently by clouds and rain. Roughly 1/3 of the S emitted
15 annually in North America is believed to be removed by precipitation.
16 NOX: Point sources are a relatively smaller contributor of NOX, but still quite
17 substantial. Both NO and NO2 have low solubility in water. Virtually no NOX is removed
18 from fresh plumes. HNO3 formed by gas-phase oxidation of NO2 is very soluble in water
19 and is the principal source of NOjin precipitation. NO3, N2O5, and HO2NO2 are also
20 believed to be significant intermediates. Since all of the intermediates are secondary
21 products, NOX scavenging increases with plume dilution and oxidation. Mesoscale studies
22 show much variation in the efficiency of wet scavenging of SOX and NOX, depending on
23 storm type and history of plume chemistry. About 1/3 of the anthropogenic NOX emissions
24 in the U.S. are estimated to be removed by wet deposition. The distinct seasonal character
25 of SOX wet deposition is absent in the case of NOX wet deposition. Some likely reasons are
26 as follows: HNO3 has a strong affinity for ice as well as liquid water; its formation has no
27 direct dependence on H2O2 which peaks in summer; and, there are mechanisms for the
28 formation of HNO3 in low winter sunlight.
April 1995 3-128 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.7 PHYSICAL AND CHEMICAL CONSIDERATIONS IN
2 PARTICULATE MATTER SAMPLING AND ANALYSIS
3 3.7.1 Size Cut-Point For Separating Fine and Coarse Particulate Matter
4 3.7.1.1 Background
5 In 1979 EPA scientists, in a paper entitled "Size Considerations for Establishing a
6 Standard for Inhalable Particles" recommended that total suspended particulate matter
7 (TSP), as defined by the high volume sampler, be replaced by the fraction obtained with a
8 sampler having a precise upper cut-point (originally 15 /^m, but later changed to 10 /*m);
9 and that "a second particle size cut-point of < 2.5 /xm diameter be incorporated in the air
10 sampling devices" (Miller et al., 1979). This study found that "the existence of a bimodal
11 distribution with fine and coarse modes has been clearly demonstrated by.... mass-size
12 distribution studies and by number distribution studies. These size distribution studies
13 suggest 1 to 3 /^m as the most appropriate range for a cut-point for fine and coarse
14 aerosols. However, practical considerations of reducing plugging of impactor orifices
15 indicate that 2.5 /xm is a more appropriate cut-point, especially for particle size fractionating
16 devices such as the dichotomous sampler" (Miller et al, 1979).
17 The cut-point of 2.5 jum, which has been used in many studies since 1979, was chosen
18 not because it was ideal but because it was the smallest cut-point deemed feasible for a
19 dichotomous sampler at that time. Current technology has demonstrated the feasibility of
20 dichotomous samplers with cut-points at 1 pirn, or even lower if desired. Impactor and
21 cyclone technology can also be used for cut-points below 2.5 fj.m. Therefore, it is
22 appropriate at this time to review existing data on size distribution of ambient aerosols so
23 that policy makers may consider whether a change to a smaller cut-point should be
24 considered. This is especially important in view of the possibility of a major increase in
25 both research measurements, exposure assessment, and regulatory monitoring of fine
26 particles, as well as of PM]0.
27
28 3.7.1.2 Size Measurements
29 Information on the size of fine and coarse particles comes from two basic techniques,
30 (1) particle-counting techniques that measure the size of individual particles and convert the
31 particle number distribution to a particle volume distribution and (2) particle-collecting
April 1995 3429 DRAFT-DO NOT QUOTE OR CITE
-------
1 techniques that use aerodynamic separation, collection of material in specific size ranges,
2 and gravimetric or chemical analysis to determine the total mass or the mass of specific
3 components in the size ranges collected. Particle counting has the potential advantages of
4 not causing as much disturbance to the gas/particle equilibrium. However, considerable
5 care must be taken to avoid heating the sample or diluting it with clean or drier air than
6 that present in the atmosphere. With particle counting techniques it may also be possible to
7 avoid problems of particle bounce. However, several expensive and complex instruments
8 are required to cover the desired range of 0.001 to 100 /zm. Because sizes can be measured
9 very precisely, the size ranges covered can be very small and an almost continuous function
10 of number versus size can be obtained.
11 Particle collecting techniques have the advantage of obtaining size-differentiated
12 samples for chemical analysis. The equipment used is simpler and less expensive.
13 However, aerodynamic separation does not provide as distinct a classification by size.
14 Large particles may bounce from their intended collection surface and be counted in smaller
15 size ranges. Also, the requirement for long sampling times may result in averages of
16 distributions that change with time. Particle collection techniques provide a limited number
17 of size cuts and yield discontinuous functions of mass versus particle size.
18 Both techniques, however, clearly indicate the natural division of ambient air particles
19 into fine and coarse modes with a minimum between 1.0 and 3.0 /nm diameter. Size
20 distributions obtained with particle counting techniques tend to show a lower, broader, and
21 more distinct minimum than distributions obtained with particle collection techniques such
22 as impactors. The position of the minimum between the accumulation and coarse mode
23 may vary from study to study. The peak of the fine particle mode tends to increase in size
24 with increasing concentration and with increasing relative humidity. Several good reviews
25 of particle size distribution are available: on physical properties of sulfur aerosols (Whitby,
26 1978), on the size distribution of urban aerosols (Lippmann, 1980) on sizes of particulate
27 sulfate and nitrate in the atmosphere (Milford and Davidson, 1987); and on the size
28 distribution of coarse mode aerosol (Lundgren and Burton, 1995).
29
30
April 1995 3-130 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.7.1.3 Appropriate Display of Size-distribution Data
2 Size-distribution data, if not properly displayed, can give misleading information on
3 the position and shape of peaks and valleys and can lead to incorrect conclusions, especially
4 in regard to the position, width, and separation of fine and coarse modes. For this reason
5 many workers use a histogram display obtained as follows. The mass, number, surface, or
6 volume in each size range is divided by the difference of the logarithms of the diameters at
7 the upper D; and the lower DJ.J ends of the size range, and plotted as rectangles of width
8 log Dj-log DJ.J and height, i.e. mass/ (log Du-log De) on a log diameter scale. This is
9 normally shown as AC/A log D , dM/d log D , or normalized, for example, as AM/M*A log
10 D . Such histogram plots are especially useful for impactor data, which normally yield
11 fewer size intervals than particle-counting techniques. Examples of such displays are
12 shown in Figure 3-13 (Wilson et al., 1977) and Figure 3-14 (John et al., 1990).
13 It is frequently desirable to draw a smooth line through the data in order to identify
14 modes and the mass mean diameters (MMD) and widths (o ) of modes. This can be done
&
15 by fitting the data to two or more lognormal distributions, as was done in Figure 3-13 (also
16 see Dzubay and Hasan, 1989; and Whitby-DISFIT (TSI, 1993); or by using an inversion
17 process such as originally developed by Twomey, as was done in Figure 3-14 (John et al.,
18 1990; Winklmayr et al., 1989). In this type of presentation the area in each rectangle or the
19 area under a portion of a curve is proportional to the mass in that size range (or the
20 quantity of any other parameter plotted on a linear scale). Plotting mass per impactor stage
21 versus impactor stage number, or drawing lines connecting the midpoints of size range at
22 the heights of the mass in each size range, does not provide such quantitative information.
23 Once the characteristics of the impactor have been demonstrated, and once good fits to
24 lognormal distributions have been obtained, repeated measurements of the same species
25 may be shown by curves fitted to inversion or lognormal distributions such as the example
26 in Figure 3-15 (John et al., 1990).
27 In impactor measurements, the maximum size of the upper stage and the minimum
28 size of the lower stage (or after filter) are not well defined. Therefore, an arbitrary choice
29 must be made in order to define the A log Dp. This choice can have a remarkable influence
30 on the perceptions of the positions, height, and width of modes. A particularly
April 1995 3-131 DRAFT-DO NOT QUOTE OR CITE
-------
Q."
O
O
Electrical aerosol analyzer
0.002 0.01
100
Figure 3-13. An example of histogram display and fitting to log-normal functions for
particle-counting size distribution data. Instruments used and the
range covered by each are shown. Counts are combined into
reasonably-sized bins and displayed. Lognormal functions, fitted to the
data, are shown with geometric mean sizes (MMD) of each mode and
the width (
-------
•fe
*
£
Q
8s
T3
8
300
250
200
•1 CO
lOU
100
50
n
-
-
_
H
n
(a)
' — •— T~
i 1 —
0.01
0.1 1 10
Aerodynamic Diameter (urn)
100
300
250
200
§" 100
•O
(J 50
•O
0
0.01
Inverted Size Distribution
(b)
0.1 1 10
Aerodynamic Diameter (urn)
100
300
250
200
o 100
I 50
0
0.01
Lognormal Fit
(c)
0.1 1 10
Aerodynamic Diameter (urn)
100
Figure 3-14. An example of an effective display of impactor data: (a) histograph
showing mass found on each impactor stage and upper and lower cut
points of each stage, (b) inverted size distribution, (i.e., a smooth
distribution that would give the observed distribution considering the
actual efficiency of each stage; cut points are not exact; each stage allows
some large particles, which it should collect, to pass through to the next
stage and collects some small particles which it should pass on to the next
stage), (c) the solid line is the distribution obtained by fitting a sum of
several lognormal functions to the inverted distribution. The dashed lines
show the lognormal functions obtained from the fitting process.
Source: John et al., 1990.
April 1995
3-133
DRAFT-DO NOT QUOTE OR CITE
-------
a1
o>
300
250
200
150
Q
O)
I 1°°
50
0
•0600-0930
1000-1330
1400-1730
1800-0100
0.01
0.1 1 10
Aerodynamic Diameter (|im)
100
Figure 3-15. Size distributions of sulfate, Long Beach, June, 1987, showing use of fitted
log-normal distributions to describe diurnal variations in size and
concentration.
Source: John et al., 1990.
1 3.7.1.4 Comparison of Particle-counting and Particle-collection Techniques
2 Unfortunately, there have been few efforts to compare results of the two particle-sizing
3 techniques. One such effort is shown in Figure 3-17 (Durham et al., 1975). The differences
4 between the two techniques, as evident in the figure, are qualitatively observed in individual
5 studies using either of the two techniques. Particle counting techniques usually give a lower
6 and wider minimum. Typically particle counting leads to volume distributions plotted versus
7 geometric size (or more properly, geometric size inferred from mobility or optical size);
8 whereas impactor separations give mass versus aerodynamic size. In Figure 3-17 both
9 geometric and aerodynamic scales are given. This figure illustrates the problems involved in
10 defining particle "size" and serves as a reminder that each particle sizing technique gives a
11 different "size". The upper scale, used for impactor data, is given in aerodynamic diameter.
April 1995
3-134
DRAFT-DO NOT QUOTE OR CITE
-------
a) Yugoslavia, Winter B, Author's Original Endpoints, 0.1 and 20 |im
300
MMD aa %Mass
0.256 1.28 39.8
1.93 1.09 19.9
13.1 2.31 40.3
Authors
original-,
curve 1
1.0 10.0
Aerodynamic Diameter, Dp (urn)
100.0
b) Yugoslavia, Winter B, Replotted with New Endpoints, 0.4 and 11 u,m
300
Mode MMD oa %Mass
1 0.671 1.09 44.6
2 1.83 1.28 25.4
3 7.89 1.20 30.1
1.0 10.0
Aerodynamic Diameter, Dp (u,m)
100.0
3-16. Effect of changing endpoints. This example of impactor data shows how
the lack of a well-defined upper and lower size limit can affect the
perception of location of fine and coarse particle modes. The curve drawn
by authors of the report, and a histogram with an upper limit of 20 /tin
and a lower limit of 0.1 urn diameter, are shown in Figure 4a. In Figure
4b a histogram with a lower limit of 0.4 /tin and an upper limit of 10 /*m
is shown. Notice how the author's free hand curve and histogram suggest
a fine particle MMD around 1.5 /im diameter. A quite different idea of
the location of the modes is given when different endpoints are chosen and
when the data is fitted to a 3-lognormal mode distribution. Much of the
material found between 1.0 and 5.0 /tni is probably smaller particles
caught on the glass fiber impactor stages which have very poor separation
efficiencies.
Source: Sega and Fugas, 1984.
1 The aerodynamic diameter of a particle is the diameter of a particle of density, p = 1.0,
2 which would behave similarly with respect to impaction as the particle in question. For
April 1995
3-135
DRAFT-DO NOT QUOTE OR CITE
-------
CO
E
O)
_o
c
.o
.n
*h-
w
b
o
1
§
I
I
I
0.01
140 -
120 -
100 -
80 -
60 -
40 -
Aerodynamic Diameter, Dae, urn
0.1 1.0
10
TIM
Denver-Welby, Nov.12,1971.
Pollution Decay
-O- MAAS
A*y ,»„ 1 CFM Andersen
• After Filter I
?Lage|-,. } 2.54 CFM Andersen
After Filter I
0.1 1.0
Particle Diameter, Dp, urn
10
140
120
100
80
60
40
20
O)
c
o
I
b
i
I
o
O
GO
Figure 3-17. These size distributions, obtained during an EPA study of the Denver
brown cloud represent one of the few efforts to compare particle-counting
and particle-collection size-distribution measurements. Note that impactor
data is given in aerodynamic diameter and particle-counting data is given
in geometric diameter derived from the number distribution and estimated
density.
Source: Durham et al., 1975.
1 spheres, the aerodynamic diameter, Da, equals \[p Dp, where p is the density of the particle
2 and D is the geometric diameter. Since coarse particles are expected to have a greater
3 density than fine particles, converting the volume, geometric-size distribution to a mass,
4 aerodynamic-size distribution would increase the apparent size of the volume distribution
5 above 1 /zm and widen the minimum. For small particles, below 0.5 pm, or at reduced
6 pressures where the mean free path of the gas molecules is of the same order, or larger than
7 the particle diameter, the Stokes diameter, which is more closely related to the diffusion
April 1995
3-136
DRAFT-DO NOT QUOTE OR CITE
-------
1 coefficient, is a more useful parameter. The relationships between Stokes, aerodynamic and
2 geometric diameter are discussed in Section 3.1.7.1.
3 The particle size distribution shown on the bottom of the graph was derived from a
4 combination of a mobility counter and several optical counters. The "mobility size",
5 obtained from the electrical aerosol analyzer (EAA) in earlier studies and the differential
6 mobility analyzer (DMA) in more recent studies is dependent on the particle shape but not
7 the density. For irregularly shaped particles the "mobility" size gives the Stokes diameter,
8 which is the geometric diameter of a sphere with the same aerodynamic drag. For a sphere
9 the Stokes diameter and the geometric diameter are the same. By comparing the mobility or
10 Stokes diameter to the aerodynamic diameter it is possible to measure the density of spherical
11 particles (Stein et al., 1994).
12 The "optical" size of a particle depends on the particles shape and refractive index, and
13 on the characteristics of the optical counter. The amount of light scattered by a particle at a
14 wavelength near the particle size varies rapidly with changes in size, wavelength, refractive
15 index, and scattering angle. Therefore, several different optical counters may be needed to
16 cover the range of atmospheric particle sizes. Because of non-linearities in the response of
17 laser or narrow wavelength optical counters to size changes it is especially difficult to
18 measure particles in the 0.5 to 1.0 size range (Hering and McMurray, 1991; Kim, 1995).
19 Since the amount of scattered light depends strongly on the refractive index it would be
20 useful to calibrate optical counters with particles of the same refractive index as those in the
21 atmosphere. Hering and McMurray (1991) used a differential mobility analyzer to select
22 particles of a uniform geometric diameter. The light scattering of these monodispersed
23 atmospheric particles, as measured by a Particle Measuring System LAS-X optical counter,
24 was compared to that of spheres of polystyrene latex (a substance frequently used to calibrate
25 optical counters) and oleic acid of the same geometric diameter. The atmospheric aerosols
26 scattered less light than polystyrene latex sphere (refractive index = 1.9 - O.Oi), but about
27 the same amount of light as oleic acid spheres (refractive index = 1.46 - O.Oi) of the same
28 geometric size. Relating the variety of sizes measured by particle counters and impactors,
29 and combining them into a single size indicator, is a major task which has not yet been
30 adequately addressed.
April 1995 3-137 DRAFT-DO NOT QUOTE OR CITE
-------
1 The greater width of the coarse modes, as measured by the impactor in Figure 3-17,
2 may be attributed to the use of glass fiber filter paper for the impactor surface. It is now
3 recognized that the use of glass fiber filter material, as contrasted to a flat surface, causes a
4 severe reduction in the effectiveness of the cut. Large particles bounce off the glass fiber
5 (Vanderpool et al., 1987) giving much reduced collection efficiencies; whereas fine particles
6 penetrate into the fiber and some are captured in stages that should have near zero collection
7 efficiencies (Rao and Whitby, 1978). Many studies that used the Anderson High Volume
8 Fractionating Sampler also used glass fiber filters. The use of glass fiber filters as impaction
9 collection surfaces causes any given size range to contain both larger and smaller particles
10 than predicted and thus tends to spread out the modes and fill in the minima. An example of
11 the smoothing effect of glass fiber collection surfaces, and especially the collection of fine
12 particles on upper stages, can be seen in Figure 3-16. Nevertheless, the bimodal nature of
13 the ambient aerosol is still captured.
14
15 3.7.1.5 Review of Size Distribution Data
16 Early Studies
17 In 1978, when EPA scientists debated the best cut-point to separate fine particles from
18 coarse particles, there was limited information available. Particle-counting data from
19 California studies had been summarized by Whitby and Sverdrup (1980) and are shown in
20 Figure 3-18. With the exception of one distribution from Pomona, all distributions showed a
21 minimum near 1 /zm and indications of significant amounts of coarse particle material
22 between 1.0 and 2.5 jum. (The region between 1 and 2.5 /zm will be referred to as the
23 intermodal region.) Other studies of size distribution (McMurry et al., 1981) in the
24 Southeastern United States, provided similar information (Figure 3-19).
25 Results from several impactor studies were also available, some of which suggested two
26 modes. However, much of the impactor data were considered unreliable in regard to the
27 existence and position of modes (Whitby et al., 1974). However, one of the more extensive
28 and reliable studies available (Patterson and Wagman, 1977) provided confirmation of the
29 Whitby bimodal observations. In this study, mass and composition measurements were made
30 for four different levels of visibility. The histograms for mass, sulfate, and iron for two
31 levels of visibility are shown in Figure 3-20. It is clear that the major portion of the fine
April 1995 3-138 DRAFT-DO NOT QUOTE OR CITE
-------
Richmond
San Francisco Airport
Fresno
Hunter Liggett
Harbor Freeway
Pomona
Goldstone
Clean Continental
Background
1.0 2.5
0.01
Figure 3-18.
Grand average volume-size distributions from the Aerosol
Characterization Experiment (ACHEX) in 1972. A size distribution for
clean continental aerosol is shown for comparison. Note that with the
exception of the Pomona size distribution, all distributions show a distinct
minima near 1.0 fim diameter. A line has been added at 1.0 /on, 2.5 /on,
and 10 /on diameter to indicate how much of the coarse particle mode is
observed between 1.0 and 2.5 jtm diameter.
Source: Whitby and Sverdrup, 1980.
1 mass is below 0.6 ^m and the major portion of the coarse mass is greater than 3 pm in
2 diameter. These impactor data, coupled with the more extensive number-size distributions
3 data of Whitby and Sverdrup (1980) led to a preference for a 1 ^m cut-point but an
April 1995 3-139 DRAFT-DO NOT QUOTE OR CITE
-------
70
60
50
= 40
Q.
Q
S 30
20
10
© 16 Km Downwind-13:23 15:21
Average of 18 Distributions
5O2 = 78 ppb
® 23 Km Downwind-16:18 17:07
Average of 8 Distributions
SD2 = 34 ppb
A Background
0.01
0.1
10
Dp,
Figure 3-19. Volume-size distribution taken in the midwestern U.S. near the
Cumberland Power Plant in Tennessee. Note that coarse mode decreases
and fine mode increased as the mobile sampling van moved downwind
farther from urban influence but allowing more time for reaction as the
power plant plume mixed with background air and SO2 was converted to
sulfate and NOX to nitrate.
Source: McMurry et al., 1981.
1 acceptance of 2.5 /zm on the assumption, then considered to be the case, that 2.5
2 represented the minimum cut-point that was attainable with a dichotomous sampler (Miller et
3 al., 1979).
April 1995
3-140
DRAFT-DO NOT QUOTE OR CITE
-------
Background visibility
M.ot=44.8^g/nnP
Visibilit level A
2.0
g> 1.5
5 1.0
0.5
1 III
Mass
-
-
• n
Illll 1 IIIMlll 1 Illllll
-
-
-
-
1 1 1 1 1 'i "i i 1 1 1 1 1 1 i IIIIIM
z.o
2.0
1.5
1.0
0.5
1 1 1 ' 1'"' ' ' ' ' 1 1 1 Illl
Mass
• — -
-
-
-
^r— n
1 1 1 1 1 1 III 1 IIIMlll 1 Illllll
0.1 0.20.5 125 10 2050
Diameter, urn
100 0.1 0.20.5 1
25 10 2050
Diameter, urn
100
2.80
|2.10
-; 1.40
& 0.70
n
' Illllll "1111 1 1 1 1 1 III
L
-
1
I I I i I I II
Sulfate ~
-
-
r~i i i IIIMII i i i i 1 1 ii
2.80
2.10
1.40
0.70
n
i i i i 1 1 in i i i i 1 1 M i i iiiliil
n Sulfate ~
1 1
1
I | Illl Ml "' 1 1 llflllll 1 Illllll
*0.1 0.20.5 1 25 10 2050 100 T).1 0.20.5 1 25 10 2050
Diameter, urn Diameter, \im
100
1.20
1.00
0.80
ff 0.60
0.40
0.20
0
- Iron (Fe)
1.20
1.00
0.80
0.60
0.40
0.20
- Iron (Fe)
0.1 0.20.5 1 25 10 2050 100 0.1 0.20.5 1 25 10 2050 100
Diameter, urn Diameter, \im
Figure 3-20. Examples of size distribution histograms for total mass, sulfate, and iron
obtained at two visibility levels using an Anderson impact or. Arbitrary
choice of 0.1 and 100 for lower and upper limits cause the extreme
rectangles to be long and low. Note the separation into fine and coarse
modes in mass and that sulfate and iron clearly belong in the fine and
coarse mode respectively.
Source: Patterson and Wagman, 1977.
April 1995
3-141
DRAFT-DO NOT QUOTE OR CITE
-------
1 Recent Work
2 In the intervening 15 years, there has been very little additional work in which
3 particle-counting techniques, covering the entire size range, have been used to measure
4 ambient aerosols. Most of the particle-counting studies have focused on fine and ultrafine
5 particles, diameter < 1.0 urn. There have however been a number of impactor studies that
6 provide insight into the size of the fine and coarse modes and into what material is found
7 between them.
8 There are only a few impactor size distribution studies that cover the full size range
9 from 0.01 to 100 am (Lundgren and Hausknecht, 1982; Lundgren et al., 1984; Burton and
10 Lundgren, 1987; Vanderpool et al, 1987). Lundgren and co-workers used a mobile unit, the
11 wide range aerosol classifier (WRAC), to measure mass-size distribution in ten size ranges
12 from <0.4 to >50 pun. Two distributions, averages for Philadelphia and Phoenix, are
13 shown in Figure 3-21. Both clearly indicate a fine particle mode with an MMD near 0.5 am
14 for Philadelphia and 0.3 um for Phoenix. Both show a coarse particle mode with an MMD
15 near 20 am in diameter. However, there is a significant amount of material found in the
16 intermodal region, 1 to 2.5 am. Although the intermodal mass is not a significant fraction of
17 the total suspended particulate mass or even of TSP, as would be measured by a high-volume
18 sample (upper cut-point around 25 am), it does represent a major portion of the coarse
19 fraction of PM10.
20 The existing size-distribution data were recently reviewed by Lundgren and Burton
21 (1995), with emphasis on the coarse mode. They concluded that the coarse mode could be
22 reasonably well described by a lognormal distribution with a mass mean diameter (MMD) of
23 15 to 25 am and a mode spread (ae) of approximately two. This allows one to calculate, for
&
24 a freshly-generated coarse mode aerosol, that about 1% of the mass would be less than
25 2.5 am and only about 0.1 % would be less than 1.0 /xm in diameter. This conclusion is
26 confirmed by data from Whitby in which a wind change allowed a measurement of fresh
27 coarse mode aerosol (National Research Council, 1979). As can be seen in Figure 3-22, the
28 intermodal mass, 1.0 to 2.5 am, was not affected, even though the mass at 20 am increased
29 substantially.
April 1995 3-142 DRAFT-DO NOT QUOTE OR CITE
-------
90.0
Philadelphia-WRAC
o.
Q
o
T3
45.0-
1.0 10.0
Aerodynamic Diameter, Dp
100.0
o
•a
90.0'
45.0-
Phoenix-WRAC
Mode
1
2
3
MMD og %Mass
0.188 1.54 22.4
1.70 1.90 13.8
16.4 2.79 63.9
1.0 10.0
Aerodynamic Diameter, Dp (urn)
100.0
Figure 3-21. Impactor size distribution measurement generated by Lundgren et al.
with the Wide Range Aerosol Classifier: (a) Philadelphia and
(b) Phoenix. Note the much larger, small size tail to the coarse mode in
the dryer environment of Phoenix.
Source: Lundgren et al., EPA Report.
1 Another extensive set of studies covering the full size range, but limited to the Chicago
2 area, has been reported by Noll and coworkers (Lin et al., 1993, 1994). They used an
3 Anderson impactor for smaller particles and a Noll Rotary Impactor for larger particles.
4 Results of Lin et al. also indicate a bimodal mass distribution. For the shorter time interval
5 measurements (8 or 16 h), the average MMD for the fine mode was 0.42 /mi, with a ag
6 around two. The average MMD of the coarse mode was 26+8 /mi, with a oa varying from
&
7 2.0 to 3.5. As shown in Figure 3-23, the results of Noll and coworkers (Lin et al., 1993,
8 1994) also suggest that some coarse mode material is found in the intermodal region, 1.0 to
9 2.5 urn.
April 1995
3-143
DRAFT-DO NOT QUOTE OR CITE
-------
Hunter-Liggett
9-14-72
0.1
1
Particle Diameter,
2.5
10
Figure 3-22. Example of aged and fresh coarse mode particle size distributions.
A sudden wind change brought fresh wind-blown dust to the sampler,
operated as part of the South Coast Air Quality Study. Note that there
is only a very small change in the intermodal mass, 1.0 to 2.5 urn
diameter, although there is a major increase in the mass between 2.5
and 10 fim in diameter.
Source: National Research Council, 1979.
1 3.7.1.6 Intermodal Region
2 Coarse mode
3 The question then arises, what portion of the coarse mode material found in the
4 intermodal region is real and what portion is artifact? As discussed in Section 3.3.3.2.4, the
5 optical size may differ from the geometric or aerodynamic size. Optical counters are
6 normally calibrated with latex particles, or other particles of a specific refractive index.
7 Atmospheric particles with different refractive indices would be incorrectly sized if the
April 1995
3-144 DRAFT-DO NOT QUOTE OR CITE
-------
20.0
20.0-
1.0
10.0
100.0
Aerodynamic Diameter, Dp (urn)
Aerodynamic Diameter, Dp (urn)
40.0
30.0-
0.0
0.1
10.0
100.0
Aerodynamic Diameter, Dp
Aerodynamic Diameter, Dp
Figure 3-23. Size distributions reported by Noll from the Chicago area using an
Anderson impactor for the smaller particles and a Noll Rotary Impactor
for the larger particles.
Source: Lin et al., 1993.
1 difference in refractive index resulted in a difference in the amount of light scattered by the
2 particles (Wilson et al., 1988; Liu et al., 1992; Hering and McMurry, 1991). For particle
3 counters using lasers, particles of different sizes within the 0.5 to 1.0 /mi range may give the
4 same light scattering (Hering and McMurry, 1991; Kim 1995).
5 In the case of impactors, it is possible that it an artifact may arise from particle bounce,
6 from fragmentation of larger agglomerates, or from loosening of material from other surfaces
7 by impacting particles. The problem of particle bounce in impactors has been treated
8 theoretically and practically in many studies (Wang and John, 1987, 1988). Most workers
9 coat the coarse particle stages with a grease or oil to reduce bounce. However, as the
April 1995
3-145 DRAFT-DO NOT QUOTE OR CITE
-------
1 surface becomes covered with aerosols, a particle may impact another particle instead of the
2 surface and either bounce to a lower stage or cause deagglomeration and reentrainment of
3 previously collected particles (John et al., 1991; John and Sethi, 1993). As impactor
4 collection plates become loaded or as inlet upper size cut surfaces become dirty, the
5 magnitude of the effect increases (Ranade et al., 1990; John and Wang, 1991). One result is
6 a lowering of the effective cut point of the inlet and the impactor stages. Thus, it is
7 uncertain how much of the mass found in the intermodal size range is real and how much is
8 due to artifacts.
9 There are several reasons to believe, however, that some of the intermodal mass may
10 be real. For example, Lundgren and Burton (1995) point out that the lifetime of particles in
11 the atmosphere is a strong function of their aerodynamic size. Thus, while freshly generated
12 coarse mode aerosol may have a MMD of 20 /mi, with only 1% below 2.5 jum, as the
13 aerosol ages the larger particles will rapidly fall out, leaving a distribution enriched with
14 particles in the small-size tail of the distribution.
15 A second explanation has to do with the possible multimodal nature of dust generated
16 by wind or vehicular traffic. A study by the U.S. Army (Pinnick et al., 1985) measured the
17 size distribution of dust generated by heavy vehicles driven on unpaved roadways in the arid
18 southwestern United States. A variety of light-scattering instruments were used and were
19 recalibrated for the refractive index of the soil particles. The occurrence of strong surface
20 winds (gusts of 15 to 20 m s"1) during the study permitted, in addition to the vehicular-
21 generated dust, some measurements of windblown dust. There were some differences
22 between sandy soil and silty soil, and between dust generated by vehicular traffic and by
23 wind. However, all situations produced a bimodal size distributions. The upper mode had
24 an MMD that ranged from 35 to 53 /mi, with ag from 1.37 to 1.68. Of particular interest,
25 however, was a second mode having an MMD that varied from 6.2 to 9.6 /mi, with a #„
6
26 from 1.95 to 2.20. (One measurement in silty soil had an MMD of 19.4 /mi.) The MMDs
27 of the smaller coarse particle modes are significantly smaller than those coarse mode MMD's
28 observed by Lundgren or Noll. An example of vehicular generated dust is shown in
29 Figure 3-24. Note that the differential mass is plotted on a logarithmic scale. These results
30 suggest that in arid areas, significant soil material, generated by traffic or wind, may be
31 found in the intermodal region.
April 1995 3-146 DRAFT-DO NOT QUOTE OR CITE
-------
O\
2.
8
U)
I
O
O
z
o
H
O
G
O
H
W
O
90
n
^-^
H
W
s
§•
I'
en
n>
(H
O
O
cn
m
a
nT
en
.
Q.
8
a
en_
N'
R" s- B. »
^- O Mi SI"
O
5
8
»•
en
0)
O
3
en
£
o
2
13
GO
a
o
en
cy
0*
o>
s?
ode
3- 3
« tr
cr
i
ffi
O
CD
O)
T3
s»
a.
Rg
en
en
>— a a-
^^ ^2 ^3
O
n>
3
o l—
I I
3 M
O en
^ a,
< g1
o
R
S'
en
S
€
>O
s- 3
rt N
»-• fB
«• a
y n-
S
n> c.
2 g a
O
1
o.
n>
o'
en
sr
en
O)
D-
en
O
O)
§
S.
n>
O
3
CD
P>
p— a
S ^
a g
c*. ^
O o
S. I
§ S
en _
O "
OSB9
rT
I
O
D.
en
en
CL
O
I ^
en p
3
2— f^
Average peak differential mass concentration
dM/dr(ngnv3nm3)
-------
1 Efficiencies typically reach a minimum between 0.1 and 1 /wm and increase for particles
2 smaller than 0.1 /mi. Thus, although most of the paniculate mass is captured, the particles
3 that do escape are in the smaller size range. Data from U.S. EPA, plotted in Figure 3-25,
4 (U.S. Environmental Protection Agency, 1993) show the size distribution of fly ash from a
5 pulverized coal power plant and the size distribution of the material escaping from the
6 various control devices. The small-size tail of the coarse mode escapes preferentially and
7 likely contributes material in the intermodal region.
8 Cheng et al. (1985) reported experimental measurements from an atmospheric fluidized-
9 bed coal combustor. Size distribution measurements, made downstream of a cyclone and
10 again downstream from baghouse filtration of the material leaving the cyclone, are shown in
11 Figure 3-26 (Cheng et al., 1985). Electron microscope photographs confirmed a fine particle
12 mode of spherical particles between 0.1 and 0.25 /mi, presumably formed from evaporation
13 and condensation of volatile species from the coal matrix; and irregular-shaped chunks from
14 the coarse mode with a peak concentration between 1 and 3 /tin in diameter.
15 A fourth reason comes from a study of the size of particles collected in various types of
16 samplers. Burton et. al (1995) used two techniques to measure the size of individual
17 particles collected on filters. Particles with diameters between 1 and 2.5 /mi were found to
18 account for 18 to 20% of the coarse fraction of PM10.
19 A fifth piece of evidence comes from studies in which measurements are made of the
20 elemental composition of PM2 5 and PM10 or the coarse fraction of PM10. Elements
21 representative of soil type material have been found in the PM2 5 fraction. In a study in
22 Philadelphia that used dichotomous samplers, an amount of soil-type material equal to 5% of
23 the coarse mode fraction of PM10 was found in the PM2 5 fraction (Dzubay et al., 1988).
24 Since the virtual impactor used in the dichotomous sampler minimizes particle bounce and
25 reintrainment, this would appear to be the small-size tail of the coarse mode in the 1 to 2.5
26 pirn size range. Similar results have been reported from the IMPROVE network (Eldred et
27 al., 1995). Elemental analysis suggested that soil-derived material, equal to 20% of the
28 coarse fraction of the PMj0 sample, was found in the PM2 5 sample.
29 Thus, one can conclude that coarse mode material is found in the intermodal region.
30 There are reasons to suspect that a portion of this material is an artifact but that a portion is
31 real coarse mode material having an aerodynamic diameter between 1.0 and 2.5 /mi. In
April 1995 3-148 DRAFT-DO NOT QUOTE OR CITE
-------
o.0-9
O 0.8
OJ „-,
O 0.7
< 0.6
*
in 0.5
Vt **•**
Ui
CO 0.4
CO 0.2
5 0.1
^]
No Controls
. 100% Emitted
•— I
~~*
No Controls
Cyclones
Scrubber
ESP
Baghouse
% of Flyash
in 1.0 to 2.5 |im
of Controlled
Emissions
6
3
51
29
53
of Total
Flyash
6
0.59
3.06
0.23
0.11
Scrubber
6% Emitted
1 2.5 10
Log diameter, Dp
1 2.5 10
Log diameter, Dp
100
Figure 3-25. Size distribution of emissions from a pulverized-coal power plant and
the particle size distributions remaining after several types of control
devices (EPA, AP-42, 1993).
1 either event, this can lead to a misunderstanding of the source of the particles, to
2 inappropriate control strategies, or to exposure studies that fail to differentiate correctly
3 between fine and coarse particles.
April 1995
3-149
DRAFT-DO NOT QUOTE OR CITE
-------
0.05 0.1 1.0 2.5
Stokes diameter,
0.05 0.1 1.0 2.5
Stokes diameter, urn
Figure 3-26. Size distributions from a fluidized-bed pulverized coal combustor, a
after initial clean up by a cyclone collector, and (b) after final clean
up by a baghouse.
Source: Cheng et al., 1985.
1 Fine Mode
2 This section discusses the source of fine mode material found in the intermodal region.
3 Early particle-counting data suggested that, with a few exceptions, significant mass of fine
4 mode material would not be found above 1 /xm (see Figures 3-13, 3-18, 3-19, and 3-20).
5 However, impactor studies, on some occasions, have observed significant mass on stages
6 with a cut point of 1 /zm. In some instances, the MMD of the fine mode was as large as
7 1 /*m (John et al., 1990). The change in relative humidity produced by a few degrees
April 1995
3-150
DRAFT-DO NOT QUOTE OR CITE
-------
1 change in temperature can significantly modify the MMD of an ambient aerosol size
2 distribution. As the RH approaches 100%, accumulation mode aerosols, with dry sizes
3 below 1.0 pm in diameter, may grow larger than 2.5 /zm in diameter, be rejected by PM2 5
4 samples, and be counted as coarse particles.
5 Before examining additional field data demonstrating the effect of relative humidity on
6 particle size, it is useful to review some basic information on the interaction of water vapor
7 with the components of fine particles. Sulfuric acid (H2SO4) is a hygroscopic substance.
8 When exposed to water vapor a H2SO4 droplet will absorb water vapor and grow in size
9 until an equilibrium exists between the liquid water concentration in the H2SO4 solution
10 droplet and the water vapor concentration in the air. The amount of water in the sulfuric
11 acid droplet will increase and decrease smoothly as the RH increases and decreases.
12 Ammonium sulfate, (NH4)2SO4, however, is deliquescent. If initially a crystal in dry air, it
13 will remain a crystal until a certain RH is reached; at this point it will absorb water and
14 become a solution droplet. The RH at which this happens, =80% RH in the case of
15 NH4)2SO4, is called the deliquescent point. At RH's above the deliquescent point the
16 (NH4)2SO4 droplets are hygroscopic, gaining or losing water reversibly as the RH increases
17 or decreases. If the RH decreases below the deliquescent point the solution droplet becomes
18 supersaturated and unstable to crystallization. However, sub-micron sized droplets will
19 remain supersaturated until a significantly lower RH, known as the crystallization or
20 efflorescent point is reached. The crystallization point decreases with decreasing droplet size
21 and decreasing purity (Whitby, 1984). Thus, for a deliquescent substance, a plot of droplet
22 diameter or water content as a function of RH will have two lines, one for increasing RH
23 and another for decreasing RH. An example of this phenomena, known as hysteresis, is
24 shown in Figure 3-27. Table 3-15 shows the RH at the deliquescent and crystallization
25 points for some compounds found in the atmosphere.
26 Much experimental and theoretical effort has gone into understanding this process. The
27 basic theory was elucidated by Hanel (1976). Much experimental work has been done on
28 atmospheric species (e.g., Tang and Munkelwitz, 1977, 1993; Richardson and Spann, 1984).
29 Ammonium nitrate, NH4NO3, because of its volatility, is difficult to handle but has been
30 studied successfully by Richardson and Hightower (1987). The aerosol equilibria models
April 1995 3-151 DRAFT-DO NOT QUOTE OR CITE
-------
2.0
o
Q.
Q
1.5
1.0
0
H
L
(NH4)2 S04
30
50 70
RH, %
8
7
6
5
4
3
1
0
90
o
Q.
Figure 3-27. Particle growth curves showing fully reversible hygroscopic growth of
sulfuric acid (H2SO4) particles, deliquescent growth of ammonium
sulfate [(NH4)2 SOJ particles at about 80% RH, hygroscopic growth
of ammonium sulfate solution droplets at RH greater than 80%; and
hysteresis as the droplet remains supersaturated as the RH decreases
below 80% RH until the crystallization point is reached.
Source: National Research Council, 1979.
April 1995
3-152
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-15. RELATIVE HUMIDITY OF DELIQUESCENCE AND
CRYSTALLIZATION FOR SEVERAL ATMOSPHERIC SALTS3
Compound Deliquescence Crystallization0
(NH4)2S04 79.9 ± 0.5 37 + 2
NH4HSO4 39.0 ± 0.5
NH4NO3 61.8
NaCl 75.3 + 0.1 42
"Taken from Tang and Munkelwitz (1993) unless otherwise indicated.
bTang and Munkelwitz (1977).
cShaw and Rood (1990) and references therein.
1 developed by Seinfeld and co-workers allow calculation of the water content of bulk solution
2 as a function of relative humidity. (Kim et al., 1993a,b).
3 The water content of a sub-micron sized droplet, and therefore it size, depends not only
4 on the dry size but is a result of a balance between surface tension and solute concentration
5 (Li et al., 1992). Pure water is in equilibrium with its vapor when the RH equals 100% and
6 is therefore, stable, i.e. the rate of evaporation equals the rate of condensation. The water in
7 a solution will be in equilibrium with water vapor at a lower water vapor concentration
8 because the presence of solute molecules or ions lower the rate of evaporation. Therefore, a
9 solution will absorb water and become more dilute, increasing the water vapor concentration
10 needed for equilibrium until the solution water vapor concentration required for equilibrium
11 matches the ambient water vapor concentration or RH. As the droplet size decreases the
12 surface tension increases and the vapor pressure of water required to maintain equilibrium
13 increases. Therefore, the smaller the dry size of the particle, the less the amount of growth
14 as RH increases.
15 Theoretical calculations of the growth of various sizes of ammonium sulfate particles
16 and an experimental verification of such calculations, using a simulation of the humidification
17 process in the human lung, are shown in Figure 3-28. Note the very rapid increase in the
18 amount of water and in the diameter of the aerosol particle as the relative humidity
19 approaches 100% RH. Considering the difficulty of measuring relative humidity accurately
20 between 99 and 100%, theory and experiment are in reasonable agreement. As can be seen
April 1995 3-153 DRAFT-DO NOT QUOTE OR CITE
-------
O" 4
.g"
co
DC
2
CD
I I
Theoretical Prediction at 22° C
o o o o o Experimental Measurements
RH-99.8%
216
125
64
27
.. I I I I I I I I I I I I I I I I I I I I lQ
200
50 100 150
NH4 HSO4 Dry Particle Diameter (nm)
tf
O 4
.g"
rr
I 3
(5
Theoretical Prediction at 22° C
o o o o o Experimental Measurements
RH-99.5%
i i i i
I
I i i i i I i i i i
125
64
27
50 100 150
NH4 HSO4 Dry Particle Diameter (nm)
200
Figure 3-28. Theoretical predictions and experimental measurements of growth of
NH4HSO4 and (NH4)2SO4 particles at RH between 95 and 100% RH.
Source:
April 1995
3-154 DRAFT-DO NOT QUOTE OR CITE
-------
1 the effect of surface tension is most important for particles with dry size less than 100 nm
2 (0.1 fj.m). This phenomena may be of importance in considering the biological effect of
3 water-soluble pollutants. Accumulation mode particles will be diluted when exposed to
4 humidification in the lungs. Ultrafine or nuclei mode particles will not be diluted as much.
5 In the atmospheric aerosol the number distribution will almost always be dominated by
6 particles below 100 nm (see Section 3.2.1). However, aerosols generated in the laboratory
7 for exposure studies probably lack the smaller particles found in the atmosphere. This
8 provides a hypothesis for the difference in effects observed in epidemiological studies and
9 laboratory exposure studies. The importance of this more concentrated, ultrafine droplet
10 component of the atmospheric atmosphere may have been neglected because most of the
11 experimental studies of hygroscopicity have used near-micron-sized particles. However, in
12 the modeling of deposition of hygroscopic particles, workers, such as Martonen (1993), have
13 corrected the experimental curves of particle size as a function of RH, based on
14 measurements of near micron-sized particles, to account for the effects of surface tension on
15 ultrafine particles.
16 In addition to the laboratory studies discussed above there are some measurements on
17 the effect of RH changes on atmospheric aerosol. McMurray and co-workers have made use
18 of a Tandem Differential Mobility Analyzer (TDMA) system (Rader and McMurry, 1986) to
19 measure the change in particle size with changes in relative humidity at Claremont, CA, as
20 part of the Southern California Air Quality Study (SCAQS) (McMurry and Stolzenberg,
21 1989) and at the Grand Canyon National Park, AZ, as part of the Navajo Generating Station
22 Visibility Study (Zhang and McMurray, 1993; Pitchford and McMurry, 1994). One mobility
23 analyzer is used to isolate a narrow size distribution. After humidification the size
24 distribution of this fraction is measured. An example is shown in Figure 3-29. Note that
25 Figure 3-29 is a number size distribution not a mass size distribution. Particle growth with
26 increasing RH is evident. However, between 70 and 91% RH the distribution splits into
27 less-hygroscopic and more-hygroscopic components. Pitchford and McMurry (1994)
28 attribute this splitting to external mixing, i.e. there are two relatively distinct classes of
29 particles, both containing some soluble and some non-soluble material, with the more
30 hygroscopic component containing significant more soluble and hygroscopic material. A
31 summary of the results of these studies is given in Table 3-16 (Zhang and McMurray, 1993).
APril 1995 3_155 DRAFT-DO NOT QUOTE OR CITE
-------
0
0
"^
-------
t
>— *
vo
U>
1
-J
O
?
H
6
0
25
O
H
O
O
3
g
O
H
m
TABLE 3-16. SUMMARY OF HYGROSCOPIC GROWTH FACTORS
Dry Size (/xm)
0.05
0.2
0.4
0.5
Dry Size (/^m)
0.05
0.10
0.20
0.30
0.40
Values are mean + standard deviations.
1987 SCAQS, Claremont,
More hygroscopic peak
Dp(90 + 3% RH)
Dp(0% RH)
1.14 + 0.05
1.23 ± 0.08
1.63 + 0.11
1.59 ± 0.08
1990 NGS Visibility Study, Grand
More hygroscopic peak
Dp(89 + 4% RH)
Dp(0% RH)
1.36 ± 0.08
1.42 ± 0.08
1.49 + 0.11
15.1 ± 0.09
1.43 + 0.10
CA
Less hygroscopic peak
Dp(87 + 2% RH)
Dp(0% RH)
1.03 + 0.03
1.02 ± 0.02
1.04 + 0.05
1.07 + 0.03
Canyon, AZ
Less hygroscopic peak
Dp(89 + 4% RH)
Dp(0% RH)
1.14 ± 0.10
1.17 + 0.09
1.17 + 0.10
1.14 ± 0.10
1.07 + 0.03
-------
1 The difference in growth rates may be due both to size and to variation in composition as a
2 function of size. The lower growth factor for 0.2 yum particles in Claremont relative to the
3 Grand Canyon may be due to a higher concentration of non-soluble organic material in
4 Claremont.
5 Some experimental examples of the significant effect of relative humidity on ambient
6 aerosol size distributions are shown in Figure 3-30 (Lowenthal et al., 1995). In this work,
7 supported by the Electric Power Research Institute, impactor collection and ion
8 chromatographic analysis were used to measure sulfate size distributions over short enough
9 periods to demonstrate the effects of changing relative humidities. The results suggest that
10 the lognormal distribution is preserved as relative humidity increases, but that the MMD
11 increases. This effect is especially pronounced as the relative humidity approaches 100%.
12 There are also studies of the behavior of ambient aerosols as the relative humidity is
13 reduced by heating the sampled air. Shaw and Rood (1990) report a study using a heated
14 integrating nephelometer in which crystallization RHs of 4 to 67% were observed. Similar
15 studies in Washington, D.C. by Fitzgerald et al. (1982) found no evidence of crystallization
16 or efflorescence when RH was reduced to 30% RH.
17 Further experimental evidence of the effect of decreasing relative humidity on aerosol
18 size distribution is provided by impactor data reported by Berner (1989). One impactor
19 sampled aerosol in its humidified state directly from the atmosphere. The inlet of a second
20 impactor was warmed «7 °C above the ambient temperature of ~5 °C in order to
21 evaporate most of the particle-bound water before collecting the aerosol. The water and
22 other volatile material in both the "wet" and the "dry" samples would evaporate in the
23 laboratory prior to weighing the impactor stages. As can be seen in Figure 3-31 in the
24 ambient air most of the non-volatile mass was above 1.0 /xm with significant amounts above
25 2.5 nm. However, after heating the size of the aerosol was reduced so that most of the non-
26 volatile mass was below 1.0 /mi. Berner treated the distributions as monomodal and derived
27 growth factors of 4.9 for fog and 4.1 for haze. If the observations are treated as
28 multimodal, good bimodal, or as shown in Figure 3-31, trimodal fits are obtained. This
29 splitting into "more" and "less" hygroscopic modes at high relative humidity has been
30 observed by McMurry and co-workers (McMurry and Stolzenberg, 1989; Zhang and
31 McMurry, 1993) (Figure 3-29) and Lowenthal et al., (1995) (Figure 3-30). In some cases,
April 1995 3_158 DRAFT-DO NOT QUOTE OR CITE
-------
Q
O)
o
O
0
• RH - 99% 8/12/90, 0200 hr
+ RH < 50%
Sulfate Size Distributions
0.01
Diameter
00
Q
D)
_g
O
•D
0
RH = 95% 8/4/90, 0200 hr
H- RH < 50%
Sulfate Size Distributions + •+ •
i i
0.01
0.1 1
Diameter (|im)
10
Figure 3-30. Example of growth in particle size due primarily to increases in
relative humidity from Uniontown, PA.
Source: Lowenthal et al., 1995.
April 1995
3-159 DRAFT-DO NOT QUOTE OR CITE
-------
80.0
Bologna Haze, Wet (Berner, 1989)
Mode
1
2
3
MMD
0.204
1.95
3.50
oa %Mass
1.69 9.9
1.97
2.65
23.5
66.5
0.1 1.0 10.0
Aerodynamic Diameter, Dp (um)
100.0
100.0
0" 50.0-]
o
o.o
Bologna Haze, Dry (Berner, 1989)
Mode MMD og %Mass
1 0.130 1.42 10.8
2 0.589 1.34 57.4
3 1.65 1.36 31.8
% Dry mass lost
upon heating
0.01
0.1 1.0
Aerodynamic Diameter, Dp (urn)
10.0
100.0
70.0
|
§"
T3
35.0-
Bologna Fog, Wet (Berner, 1989)
Mode MMD oa %Mass
1 0.310 2.09 30.8
2 1.34 1.93 36.4
3 5.31 1.91 32.8
0.01
0.1 1.0
Aerodynamic Diameter, Dp
10.0
100.0
200.0
E
o
I
•
100.0-
0.0
Bologna Fog, Dry (Berner, 1989)
Mode MMD aa %Mass
1 0.145 1.39 17.8
2 0.524 1.36 65.4
3 1.56 1.32 13.9
% Dry mass lost
upon heating
10.9%
0.01
0.1 1.0
Aerodynamic Diameter, Dp
10.0
100.0
Figure 3-31. Mass size distribution of non-volatile aerosol material. The aerosol
was collected at ambient conditions, "wet", or after evaporation of
water, "dry". A - haze; B - fog.
Source: Berner, 1989.
April 1995
3-160
DRAFT-DO NOT QUOTE OR CITE
-------
1 reported by Pitchford and McMurray (1994), splitting into three modes of varying
2 hygroscopicity was observed. However, the separation into two "more" hygroscopic modes
3 may represent, as suggested by Berner, variations in relative humidity extremes during
4 different parts of the overnight sampling period.
5 In measuring light scattering with the integrating nephelometer, the aerosol community
6 has been very concerned about the difference in relative humidity and temperature in the
7 ambient air and in the volume of air in which particle scattering is actually measured (Covert
8 et al., 1972; Fitzgerald et al., 1982). Temperature differences between the measurement
9 volume and ambient air of 1 or 2 °C can change the relative humidity and change the
10 observed light scattering. Great efforts have been made to minimize this temperature
11 difference. However, researchers have not been nearly as careful in considering temperature
12 and relative humidity effects when measuring size distribution, either with impactors or
13 particle counters, even though effects have been reported in the early literature (Wagman
14 et al., 1967; Sverdrup et al., 1980).
15 A recent paper by Cass and coworkers (Eldering et al., 1994) provides some insight
16 into how differences in RH resulting from heating can cause differences between
17 particle-counting distributions and impactor distributions. Particle size distributions were
18 obtained by counting particles by mobility (electrical aerosol analyzer) and light scattering
19 (optical particle counter). An example is shown in Figure 3-32. Almost no particles were
20 found between 1.0 and 2.5 /^m diameter. When these particle number data were converted
21 to total expected light scattering, they agreed with measurements made by a heated, but not
22 an unheated, integrating nephelometer; and when converted to expected mass, agreed with
23 filter measurements of dry mass. Eldering et al. (1994) conclude that even the moderate
24 heating occurring in mobility and optical counters was enough to change the size of the
25 particles, especially when the ambient air was close to 100% RH. It seems likely that most
26 particle counting systems produce some heating of the aerosol, and thus some reduction of
27 the measured particle size from that existing in the ambient air. On the other hand, if
28 particle-size measuring devices were located in air conditioned or heated trailers or
29 laboratories, the temperature of the sampled air would be changed and the measured particle
30 size distribution would be different from that existing in the ambient air (Sverdrup et al.,
31 1980).
April 1995 3_16l DRAFT-DO NOT QUOTE OR CITE
-------
100.00
80.00--
"E eo.oo
0.
Q
§> 40.00
T3
>
20.00--
0.00
Figure 3-32.
0.01
Diameter ((im)
Example of particle-counting volume distribution obtained in
Claremont, CA. Compare to Figures 3-14 and 3-31. Heating of the
sampled air by the mobility and optical counters are believed to have
resulted in a distribution representative of a lower than ambient
relative humidity.
Source: Eldering et al., 1994,
1 During the high relative humidities that occur at nighttime, growth of hygroscopic
2 components can result in the growth of some fine mode aerosol to diameters greater than
3 1.0 /mi and perhaps even above 2.5 /mi. As can be seen in Figure 3-28, dry ammonium
4 sulfate particles having a dry diameter of 0.5 /mi will grow to =2.5 /mi at a relative
5 humidity between 99 and 100%. When the relative humidity actually reaches 100%, the
6 particles will continue to grow to maintain the relative humidity at 100%, and eventually
7 become fog droplets that are large enough to be collected in the fraction larger than 2.5 /mi.
8 Ammonium sulfate particles with dry sizes greater than 0.5 /mi would also grow into the
9 larger than 2.5 urn size range.
10 The addition of water to hygroscopic particles, discussed in the previous section, is a
11 reversible process. Particles absorb water and grow as RH increases; as RH decreases some
April 1995
3-162
DRAFT-DO NOT QUOTE OR CITE
-------
1 of the particle-bound water evaporates and the particles shrink. However, the large amount
2 of liquid water associated with hygroscopic particles at high relative humidity provides a
3 medium for liquid phase transformation process. A number of atmospheric process, which
4 convert SO2 to sulfate or NOX to nitrate, can take place in water solutions but not in the gas
5 phase. These processes are not reversible but lead to an accumulation of sulfate or nitrate
6 and lead to an increase in the dry size of the particle. Of course as more sulfate or nitrate is
7 added to the particle it will absorb more water so that the wet size will also increase.
8 The first observation and clear discussion of these combined effects of relative humidity
9 on growth and SO2 conversion to sulfate are given by Hering and Friedlander (1982) as
10 shown in Table 3-17. Using a low pressure impactor, they observed that days with higher
11 relative humidity had higher sulfate concentration and higher MMD's compared to days with
12 lower relative humidity. Hering and Friedlander (1982) named the small mode the
13 condensation mode and suggested that it was formed by the gas phase conversion of SO^ to
14 sulfate and subsequent nucleation, coagulation, and growth by condensation. They named
15 the larger mode the droplet mode. They discussed possible mechanisms for formation of this
16 mode. They ruled out coagulation as being too slow. Reactions in fog droplets were ruled
17 out on the basis that fog has too few particles per unit volume to give the number of particles
18 found in the droplet mode. They concluded that growth occurred due to reaction of SO2 in
19 the liquid water associated with the particle.
20
21
TABLE 3-17. COMPARISON OF SULFATE CONCENTRATION AND MASS MEAN
DIAMETERS OF AEROSOLS FOR DAYS WITH HIGHER AND LOWER
RELATIVE HUMIDITY
Minimum RH, %
Maximum RH, %
Sulfate Concentration, /ig/m3
Mass Mean Diameter, pm
Low RH Days
17-35
45 -68
3-9
0.20 + 0.02
High RH Days
26-66
69 - 100
3 -52
0.54 ± 0.07
1 In a series of papers McMurray and co-workers make use of the aerosol growth law,
2 originally developed by Heisler and Friedlander (1977), to study the mechanism and rates of
April 1995 3_163 DRAFT-DO NOT QUOTE OR CITE
-------
1 sulfate formation in ambient air (McMurry et al., 1981; McMurry and Wilson, 1982, 1983).
2 They were able to apportion growth to condensation and droplet mechanisms and observed
3 droplet growth in particles up to 3 /nm in diameter.
4 A process of aerosol growth due to increasing relative humidity (Figure 3-33) has also
5 been utilized by Cahill et al. (1990) to explain observations of sulfate size changes during the
6 1986 Carbonaceous Species Methods Comparison Study in Glendora, CA. Cahill used a
7 DRUM sampler to measure sulfate in nine size ranges. By tracking the mass of sulfate in the
8 0.56 to 1.15 /xm size range Cahill et al. could follow the expansion and contraction of
9 aerosol particles containing sulfate. Because of the relative high time resolution of the
10 DRUM sampler (4 h except for an 8-h increment each night from midnight to 8 a.m.), Cahill
11 et al. (1990) could follow this process as the relative humidity increased during the night and
12 decreased during the day. This data indicates that during the "Poor Period" (low visibility)
13 particles grow as relative humidity increases. However, they did not return to the smaller
14 size observed during the "Fair Period" (good visibility). This could be due to a combination
15 of growth due to reaction of SO2 to sulfate within the particles or failure of the droplet to
16 crystallize thus maintaining particle-bound water in a supersaturated state.
17 John et al. (1990), in studies in the Los Angeles area, observed a number of sulfate size
18 distributions with MMD near 1.0 ^m. A histogram of the sulfate MMDs from his study is
19 shown in Figure 3-34. John et al. (1990) have provided a qualitative explanation to account
20 for these large MMDs for fine mode aerosol. In analyzing their data John et al. plotted
21 sulfate mass as a function of sulfate MMD and found two distinct regions, as shown in
22 Figure 3-35. Distributions with particles near 0.2 /*m diameter are probably still dry; the
23 particles have not reached their deliquescent point. As the relative humidity increases they
24 reach their deliquescent point and grow rapidly into the 0.5 to 0.7 pirn size range. During
25 the formation of fog, the hygroscopic particles act as fog condensation nuclei, and with
26 relative humidity at 100%, grow into 1 to 10 fim fog droplets. Sulfur dioxide dissolves in
27 the fog droplets and is rapidly oxidized to sulfate by atmospheric oxidants such as H2O2 or
28 O3, or by catalysis by Fe or Mn. These particles lose some of their water as the relative
29 humidity decreases below 100% RH, but will have substantially more sulfate than prior to
30 activation. Similar processes occur in clouds (Swartz, 1984).
April 1995 3-164 DRAFT-DO NOT QUOTE OR CITE
-------
50
40
E
(O
S 30
2°
10
0
D Poor Period, 8/12-8/16
• 0000 - 0800
a
D
D
on
n
n
n
o
o°-
o o
o
o
O Fair Period, 8/17-8/20
•0000-0800
20
40 60
Relative Humidity (%)
80
100
Figure 3-33.
Relative humidity versus sulfur, during the 1986 Carbonaceous Species
Methods Comparison Study, for particles with Dp>0.56 /im. The
approximate trajectories followed during each day by the Dp > 0.56
sulfur size fraction are shown for period P and period F. Note that
even when the humidities are low, 30 to 50 %, the period P aerosols
remain coarser by a factor of three than those of period F. The water
content incorporated in the aerosols during the 0000- to 0800-h time
periods is lost only slowly, giving a strong hysteresis effect in sulfur
size.
Source: Cahill et al., 1990.
1 In an analysis of data from the IMPROVE network Cahill and co-workers (Eldred et
2 al., 1995) report that 20% of the total sulfate is found in the coarse fraction of PM10.
3 Studies in Philadelphia using dichotomous samplers have also reported that 20% of the total
4 sulfate is found in the coarse fraction (Dzubay et al., 1988). Cahill and coworkers suggest
April 1995
3-165
DRAFT-DO NOT QUOTE OR CITE
-------
CD
£
35
30
25
20
15
10
5
0
Summer Atl Sites SO*'
(a)
0.1 1
Aerodynamic Mode Diameter
10
400
300
o
§
O 200
(D
Summer All Sites SOn •
(b)
0.1 1
Aerodynamic Mode Diameter (urn)
10
Figure 3-34. Data from the South Coast Air Quality Study (John et al., 1990).
Plots show (a) frequency of sulfate modes of various sizes as a function
of mode mass mean diameter (MMD) and (b) average sulfate mode
concentration as a function of mode MMD. Note that although there
are only a few instances when the MMD is near 1.0 /an diameter, it is
these situations that give rise to the highest sulfate concentrations.
Modes with MMDs above 2.5 /im diameter may be due to collection of
fog droplets containing sulfate or reaction of SO2 in liquid droplets of
NaCl due to NaCl sea spray droplets in which SO2 has dissolved and
reacted to form sulfate and release HC1 gas.
April 1995
3-166
DRAFT-DO NOT QUOTE OR CITE
-------
1,000 -
<0
^
0)
•-
I
o
O
o>
Figure 3-35.
0.1 1
Aerodynamic Mode Diameter (jim)
Log-log plot of sulfate mode concentration versus mode diameter from
Claremont during summer SCAQS (John et al., 1990). The solid lines
have slopes corresponding to mode concentration increasing with the
cube of the mode diameter. A transition between the two modes is
believed to occur at approximately the sulfate mode concentration
indicated by the horizontal dashed line.
1 that sulfate particles may grow larger than 2.5 pm in diameter and thus be sampled in the
2 PM10 fraction but not the PM2 5 fraction. It is possible for SO2 to react with basic
3 carbonatecoarse particles to form a sulfate coating or to dissolve in wet NaCl particles, from
4 oceans, lakes, or salt placed on streets to dissolve ice, and be converted to sulfate with the
5 release of HC1. However, there is substantial evidence that some fine sulfate, and therefore
6 possibly other fine mode material, may be found in the size range above 1.0 p,m and even
7 above 2.5 pm diameter, due to the growth of hygroscopic particles at very high relative
8 humidity.
April 1995
3-167
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
A similar process probably accounts for the large size of the fine mode observed in
Vienna (Berner et al., 1993). Winter and summer size distribution are shown in Figure 3-36.
Berner et al. reported that fog occurred during the night time during the winter study. In
this European study, as in American studies, instances of fine mode size distributions with
MMDs near or above 1 ^m seem to occur only when fog or very high relative humidity
conditions have been present.
40.0
Q.
Q
O)
.o
TJ
20.0"
Vienna, Summer
Aerodynamic Diameter, Dp (^
50.0
Q.
Q
CD
_O
T3
25.0-
Vienna, Winter
10
Aerodynamic Diameter, Dp (urn)
Figure 3-36. Typical results of size-distribution measurements taken with a Berner
impactor in a Vienna street with heavy automotive traffic (Berner et al.,
1993): (a) measurements taken during summer at three different
elevations, (b) measurements taken during winter at three different
elevations, fog was frequently present during the winter sampling
period.
April 1995
3-168
DRAFT-DO NOT QUOTE OR CITE
-------
1 3.7.1.7. Conclusions
2 This review of atmospheric particle-size-distributions was undertaken to provide
3 information which could be used to determine what cut-point; 1.0 /urn, 2.5 pirn, or something
4 in between; would give the best separation between the fine and coarse particle modes. The
5 data do not provide a clear or obvious answer. Depending on conditions, a significant
6 amount of either fine or coarse mode material may be found in the intermodal region
7 between 1.0 and 2.5 pun. However, the analysis does demonstrate the important role of
8 relative humidity in influencing the size of the fine particle mode and indicates that
9 significant fine mode material is found above 1.0 ptm only during periods of very high
10 relative humidity.
11 Thus, a PM2.5 pirn sample will contain most of the fine mode material, except during
12 periods of RH near 100 %. However, especially in conditions of low RH, it may contain
13 5 to 20 % of the coarse mode material below 10 />im in diameter. A PM1.0 /an sample will
14 prevent misclassification of coarse mode material as fine but under high RH conditions will
15 result in some of the fine mode material being misclassified as coarse.
16 A reduction in RH, either intentionally or inadvertently, will reduce the size of the fine
17 mode. A sufficient reduction in RH will yield a dry fine particle mode with very little
18 material above 1.0 /mi. However, reducing the RH by heating will result in loss of
19 semivolatile components such as ammonium nitrate and semivolatile organic compounds. No
20 information was found on techniques designed to remove particle-bound water without loss of
21 other semivolatile components.
22
23
24 3.8 SUMMARY
25 • The scales of transport of most interest with respect to atmospheric aerosols are the
26 13- and a-mesoscales, which have been emphasized in this section.
27
28 • Field measurements and model evaluations of transport and dispersion on these scales
29 are quite limited. Quantitative simulations of transport processes on these scales are
30 believed to be subject to considerable error.
31
32 • The main sources of this error are believed to be related to sub-grid-scale processes
33 related to plumes, clouds, complex terrain and complex mesoscale flow systems.
April 1995 3-169 DRAFT-DO NOT QUOTE OR CITE
-------
1 Routine surface meteorological measurements are extensive enough, but upper level
2 measurements are too coarse.
3
4 • New multi-scale models currently under development are expected to provide
5 substantially finer spatial resolution where most needed, as well as special treatment
6 of some of the remaining sub-grid-scale effects (e.g., plume-in-grid treatment for
7 major point source emissions). Transport simulation should be much improved in
8 these new models.
9
10 • Meteorology influences paniculate air quality in a variety of other ways also (Sloane,
11 1983). Over a period of a week or so, air masses stagnating in high pressure cells
12 over source regions may develop considerable haze which may then be transported
13 over a long range (Gillani and Husar, 1976; Samson, 1977). Also, UV radiation,
14 temperature and humidity are well known to be important factors in atmospheric
15 chemistry and aerosol formation.
16
17 • The primary focus in this section has been on field measurements related to two
18 important transformation processed: gas-to-particle conversation (particle formation),
19 and growth of hygroscopic particles in humid air, clouds, and fogs.
20
21 • Another major focus has been on transformations in plumes of major point sources
22 and urban-industrial complexes which are the carriers of most particles of
23 anthropogenic origin.
24
25 • Field measurements of homogeneous gas-phase and heterogenous aqueous-phase
26 chemistry are reviewed for sulfur, nitrogen, and organic compound.
27
28 • Gas-phase chemistry in (NOX) plumes depends principally on plume dilution with
29 background hydrocarbons and oxidants, and on sunlight. Large diurnal and seasonal
30 variations exist in the rates of oxidation of SO2 to sulfate and NO2 to inorganic
31 nitrate. For SO2, the gas-phase rate in diluted point-source plumes varies typically
32 between 1 and 3% h"1 during summer midday conditions in the eastern United States,
33 and up to about 1% h"1 in the cleaner conditions of southwestern United States. In
34 urban plumes, the upper limit appears to be closer t 5% h"1 under the more polluted
35 conditions. For NO2, the rates appear to be about three times faster for both types of
36 plumes. Winter rates are about an order of magnitude lower, on average.
37 Conversion rates in the background are comparable to the peak rates in diluted
38 plumes. Neutralization of H2SO4 formed from SO2 oxidation increases with plume
39 age and background NH3 concentration. If the NH3 concentration is more than
40 sufficient to completely neutralize H2SO4 to (NH4)2SO4, then some of the HNO3 may
41 be converted to NH4NO3, depending also on temperature. NH4NO3 is observed
42 commonly in summer in Los Angeles at Riverside, downwind of a major source of
43 ammonia.
44
45 • Contributions of aqueous-phase chemistry in plumes are highly variable, depending
46 on availability of the aqueous phase (wetted aerosols, clouds, fogs, and light rain) and
47 the photochemically generated oxidizing agents (particularly H2O2 in the case of
April 1995 3-170 DRAFT-DO NOT QUOTE OR CITE
-------
1 sulfur). In-cloud conversion rates for sulfur can be several fold larger than the
2 underlying gas-phase rates, and both are low in stable and low-sunlight conditions.
3 Nitrate in the aqueous phase is due mainly to dissolution of nitric acid formed in the
4 gas phase during daylight; there are indications of its formation under dark conditions
5 by heterogeneous processes involving the NO3 radical and N2O5.
6
7 • Variable amounts of secondary organic aerosols have been observed in urban smog,
8 particularly in the Los Angeles basin. They are most common on summer days at
9 downwind sites, and have been observed to comprise as much as 70% of the total
10 OC. More typically, however, primary OC is dominant.
11
12 • There is considerable recent evidence suggesting that aerosol composition is
13 externally mixed, at least partially, with "more" hygroscopic and "less" hygroscopic
14 components co-existing in monodisperse populations. This observation has important
15 implications concerning the water content of atmospheric particles, and their growth
16 in humid conditions.
17
18 • One- to three-quarters of the aerosol mass in the eastern United States atmosphere is
19 estimated to be water-soluble. At RH > 75%, the water content of such aerosol in
20 creases with RH. Also, the critical supersaturation for activation of such aerosol is
21 relatively insensitive to its chemical composition. The accumulation mode of such
22 aerosol is expected to be fully activated in convective clouds, but evidently only
23 partially in stratiform clouds and ground fogs, depending on particle dry size,
24 concentration, and cooling rate. In stratiform clouds under polluted conditions,
25 activation efficiency decreases non-linearly with increasing particle concentration.
26
27
April 1995 3_171 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Ahr, M.; Flossman, A. I.; Pruppacher, H. R. (1989) A comparison between two formulations for nucleation
4 scavenging. Contrib. Atmos. Phys. 62: 321-326.
5
6 Air Quality Group. (1993) Anual report: aerosol collection and compositional analysis for improvement. Davis,
7 CA: University of California at Davis.
8
9 Akhter, M. S.; Chughtai, A. R.; Smith, D. M. (1984) J. Phys. Chem. 88: 5334-5342.
10
11 Akhter, M. S.; Chughtai, A. R.; Smith, D. M. (1985) Appl. Spectrosc. 39: 143-153.
12
13 Alheit, R. R.; Flossman, A. I.; Pruppacher, H. R. (1990) A theoretical study of the wet removal of
14 atmospheric pollutants—Part IV: the uptake and redistribution of aerosol particles through nucleation and
15 impaction scavenging by growing cloud drops and ice particles. J. Atmos. Sci. 47: 870-887.
16
17 Alkezweeny, A.; Powell, D. (1977) Estimation of transformation rate of SO2 to SO4 from atmospheric
18 concentration data. Atmos. Environ. 11: 179-182.
19
20 Allen, D. T.; Palen, E. J.; Haimov, M. I.; Hering, S. V.; Young, J. R. (1994) Fourier transform infrared
21 spectroscopy of aerosol collected in a low pressure impactor (LPI / FTIR): method development and field
22 calibration. Aerosol Sci. Technol. 21: 325-342.
23
24 Allwine, K. J. (1993) Atmospheric dispersion and tracer ventilation in a deep mountain valley. J. Appl.
25 Meteorol. 32: 1017-1037.
26
27 Alofs, D. J.; Hagen, D. E.; Trueblood, M. B. (1989) Measured spectra of the hygroscopic fraction of
28 atmospheric aerosol particles. J. Appl. Meteorol. 28: 126-136.
29
30 Altshuller, A. P. (1983) Review: natural volatile organic substances and their effect on air quality in the United
31 States. Atmos. Environ. 17: 2131-2165.
32
33 Altshuller, A. P. (1987) Potential contribution of sulfate production in cumulus cloud droplets to ground level
34 particle sulfur concentrations. Atmos. Environ. 21: 1097-1105.
35
36 American Conference of Governmental Industrial Hygienists. (1994) Appendix D: particle size-selective
37 sampling criteria for airborne paniculate matter. Cincinnati, OH: American Conference of Governmental
38 Industrial Hygienists. Threshold limit values for chemical substances and physical agents and biological
39 exposure indices; pp. 43-46.
40
41 Anderson, J. R.; Aggett, F. J.; Buseck, P. R.; Germani, M. S.; Shattuck, T. W. (1988) Chemistry of
42 individual aerosol particles from Chandler, Arizona, an arid urban environment. Environ. Sci. Technol.
43 22: 811-818.
44
45 Andreae, M. O.; Charlson, R. J.; Bruynseels, F.; Storms, H.; Van Grieken, R.; Maenhaut, W. (1986) Internal
46 mixture of sea salt, silicates, and excess sulfate in marine aerosols. Science (Washington, DC)
47 232: 1620-1623.
48
49 Andrews, E.; Larson, S. M. (1993) Effect of surfactant layers on the size changes of aerosol particles as a
50 function of relative humidity. Environ. Sci. Technol. 27: 857-865.
51
52 Appel, B. R.; Tokiwa, Y.; Hsu, J.; Kothny, E. L.; Hahn, E. (1985) Visibility as related to atmospheric aerosol
53 constituents. Atmos. Environ. 19: 1525-1534.
54
April 1995 3-172 DRAFT-DO NOT QUOTE OR CITE
-------
1 Appel, B. R.; Cheng, W.; Salaymeh, F. (1989) Sampling of carbonaceous particles in the atmosphere—II.
2 Atmos. Environ. 23: 2167-2175.
3
4 Arimoto, R.; Duce, R. A.; Ray, B. J.; Unni, C. K. (1985) Atmospheric trace elements at Enewetak Atoll.
5 2. Transport to the ocean by wet and dry deposition. J. Geophys. Res. 90: 2391-2408.
6
7 Arnts, R. R.; Gay, B. W., Jr. (1979) Photochemistry of some naturally emitted hydrocarbons. Research
8 Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Sciences Research Laboratory;
9 EPA report no. EPA-600/3-79-081. Available from: NTIS, Springfield, VA; PB80-131980.
10
11 Ashbaugh, L. L.; Myrup, L. O.; Flocchini, R. G. (1984) A principal component analysis of sulfur
12 concentrations in the western United States. Atmos. Environ. 18: 783-791.
13
14 Ayers, G. P.; Larson, T. V. (1990) Numerical study of droplet size dependent chemistry in oceanic, wintertime
15 stratus clouds at southern mid-latitudes. J. Atmos. Chem. 11: 143-167.
16
17 Bagnold, R. A. (1941) The physics of blown sand and desert dunes. London, United Kingdom: Methuen & Co.
18
19 Baker, M. B.; Latham, J. (1979) The evolution of droplet spectra and the rate of production of embryonic
20 raindrops in small cumulus clouds. J. Atmos. Sci. 36: 1612-1615.
21
22 Baldwin, A. C. (1982) Heterogeneous reactions of sulfur dioxide with carbonaceous particles. Int. J. Chem.
23 Kinet. 14: 269-277.
24
25 Barnard, W. R.; Stensland, G. J.; Gatz, D. F. (1986) Alkaline materials flux from unpaved roads: source
26 strength, chemistry and potential for acid rain neutralization. Water Air Soil Pollut. 30: 285-293.
27
28 Barnard, W. R.; Gatz, D. F.; Stensland, G. J. (1987) Chemical characterization of aerosols emitted from
29 vehicle traffic on unpaved roads. Presented at: the 80th annual meeting of the Air Pollution Control
30 Association; June; New York, NY.
31
32 Barnard, W. R.; Stansland, G. J.; Gatz, D. F. (1988) Flux of alkaline materials from unpaved roads in the
33 southwestern United States. In: Particulate matter/fugitive dusts: measurement and control in western arid
34 regions. Proceedings of an Air Pollution Control Association meeting. Pittsburgh, PA: Air Pollution
35 Control Association; pp. 27-39.
36
37 Barrie, L. A. (1985) Scavenging ratios, wet deposition and in-cloud oxidation: an application to the oxides of
38 sulfur and nitrogen. J. Geophys. Res. 90: 5789-5799.
39
40 Barrie, L. A.; Georgii, H. W. (1976) An experimental investigation of the absorption of sulphur dioxide by
41 water drops containing heavy metal ions. Atmos. Environ. 10: 743-749.
42
43 Barth, M. C. (1994) Numerical modeling of sulfur and nitrogen chemistry in a narrow cold frontal rainband:
44 the impact of meteorological and chemical parameters. J. Appl. Meteorol. 33: 855-868.
45
46 Barth, M. C.; Hegg, D. A.; Hobbs, P. V. (1992) Numerical modeling of cloud and precipitationchemistry
47 associated with two rainbands and some comparisons with observations. J. Geophys. Res. 97: 5825-5845.
48
49 Bassett, M. E. (1990) Episodic PM10 model development and application for the South Coast air basin.
50 El Monte, CA: South Coast Air Quality Management District; draft technical report no. V-E.
51
52 Bassett, M.; Seinfeld, J. H. (1983) Atmospheric equilibrium model of sulfate and nitrate aerosols. Atmos.
53 Environ. 17: 2237-2252.
54
April 1995 3473 DRAFT-DO NOT QUOTE OR CITE
-------
1 Bassett, M. E.; Seinfeld, J. H. (1984) Atmospheric equilibrium model of sulfate and nitrate aerosols—II.
2 particle size analysis. Atmos. Environ. 18: 1163-1170.
3
4 Batterman, S. A.; Dzubay, T. G.; Baumgardner, R. E. (1988) Development of crustal profiles for receptor
5 modeling. Atmos. Environ. 22: 1821-1828.
6
7 Benkovitz, C. M.; Berkowitz, C. M.; Easter, R. C.; Nemesure, S.; Wagener, R.; Schwartz, S. E. (1994)
8 Sulfate burdens over the North Atlantic and adjacent continental regions: evaluations for October and
9 November, 1986 using a three dimensional model driven by observation-derived meteorology. J. Geophys.
10 Res [Atmos.] 99: 20725-20756.
11
12 Bennett, R. L.; Stockburger, L.; Barnes, H. M. (1994) Comparison of sulfur measurements from a regional
13 fine particle network with concurrent acid modes network results. Atmos. Environ. 28: 409-419.
14
15 Berner, A. (1989) Haze and its relation to atmospheric accumulation aerosol. Sci. Total Environ. 86: 251-263.
16
17 Berner, A.; Liirzer, C.; Pohl, F.; Preining, O.; Wagner, P. (1979) The size distribution of the urban aerosol in
18 Vienna. Sci. Total Environ. 13: 245-261.
19
20 Bernstein, D. M.; Rahn, K. A. (1979) New York Summer Aerosol Study: trace element concentrations as a
21 function of particle size. In: Kneip, T. J.; Lippmann, M., eds. The New York Summer Aerosol Study,
22 1976. Ann. N. Y. Acad. Sci. 322: 87-97.
23
24 Bidleman, T. F. (1988) Atmospheric processes. Environ. Sci. Technol. 22: 361-367.
25
26 Biggins, P. D. E.; Harrison, R. M. (1980) Chemical speciation of lead compounds in street dusts. Environ. Sci.
27 Technol. 14: 336-339.
28
29 Binkowski, F. S.; Shankar, U. (1994) The regional paniculate model, part I: model description and preliminary
30 results. Research Triangle Park, NC: U.S. Environmental Protection Agency.
31
32 Biswas, P.; Jones, C. L.; Flagan, R. C. (1987) Distortion of size distributions by condensation and evaporation
33 in aerosol instruments. Aerosol Sci. Technol. 7: 231-246.
34
35 Bott, A. (1991) On the influence of the physicochemical proerties of aerosol on the life cycle of radiation fogs.
36 Boundary Layer Meteorol. 56: 1-31.
37
38 Bott, A.; Carmichael, G. R. (1993) Multiphase chemistry in a microphysical radiation fog model-A numerical
39 study. Atmos. Environ. 27: 503-522.
40
41 Bower, K. N.; Choularton, T. W. (1993) Cloud processing of the cloud condensation nucleus spectrum and its
42 climatological consequences. Q. J. R. Meteorol. Soc. 119: 655-679.
43
44 Bower, K. N.; Hill, T. A.; Coe, H.; Choularton, T. W. (1991) SO2 oxidation in an entraining cloud model
45 with explicit microphysics. Atmos. Environ. Part A 25: 2401-2418.
46
47 Boyce, S. D.; Hoffmann, M. R. (1984) Kinetics and mechanism of the formation of hydroxymethanesulfonic
48 acid at low pH. J. Phys. Chem. 88: 4740-4746.
49
50 Braaten, D. A.; Paw U, K. T. (1992) A stochastic particle resuspension and deposition model. In: Schwartz,
51 S.; Slinn, W. G. N., eds. Precipitation scavenging and atmosphere-surface exchange, v. 2. New York, NY:
52 Hemisphere Publishing Corp.; pp. 1143-1152.
53
April 1995 3-174 DRAFT-DO NOT QUOTE OR CITE
-------
1 Braaten, D. A.; Paw U, K. T.; Shaw, R, H. (1989) Particles resuspension in a turbulent boundary layer:
2 observed and modeled. J. Aerosol Sci. 12: 613-628.
3
4 Briggs, G. A.; Binkowski, F. S. (1985) Research on diffusion in atmospheric boundary layers: a position paper
5 on status and needs. Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric
6 Sciences Research Laboratory; EPA report no. EPA/600/3-85/072. Available from: NTIS, Springfield, VA;
7 PB86-122587.
8
9 Brown, N. J.; Dod, R. L.; Mowrer, F. W.; Novakov, T.; Williamson, R. B. (1989) Smoke emission factors
10 from medium-scale fires: part 1. Aerosol Sci. Technol. 10: 2-19.
11
12 Buat-Menard, P.; Duce, R. A. (1986) Precipitation scavenging of aerosol particles over remote marine regions.
13 Nature (London) 321: 508-510.
14
15 Burkhard, E. G.; Dutkiewicz, V. A.; Husain, L. (1994) A study of SO2, S02"4 and trace elements in clear air
16 and clouds above the midwestern United States. Atmos. Environ. 28: 1521-1533.
17
18 Burnett, R. T.; Dales, R. E.; Raizenne, M. E.; Krewski, D.; Summers, P. W.; Roberts, G. R.; Raad-Young,
19 M.; Dann, T.; Brook, J. (1994) Effects of low ambient levels of ozone and sulfates on the frequency of
20 respiratory admissions to Ontario hospitals. Environ. Res. 65: 172-194.
21
22 Burton, R. M.; Lundgren, D. A. (1987) Wide range aerosol classifier: a size selective sampler for large
23 particles. Aerosol Sci. Technol. 6: 289-301.
24
25 Burtscher, H. (1992) Measurement and characteristics of combustion aerosols with special consideration of
26 photoelectric charging and charging by flame ions. J. Aerosol Sci. 23: 549-595.
27
28 Burtscher, H.; Leonardi, A.; Steiner, D.; Baltensperger, U.; Weber, A. (1993) Aging of combustion particles
29 in the atmosphere: results from a field study in Zurich. Water Air Soil Pollut. 68: 137-147.
30
31 Byers, H. R. (1974) General meteorology. New York, NY: McGraw Hill.
32
33 Cadle, S. H.; Dasch, J. M. (1988) Wintertime concentrations and sinks of atmospheric paniculate carbon at a
34 rural location in northern Michigan. Atmos. Environ. 22: 1373-1381.
35
36 Cahill, T. A.; Surovik, M.; Wittmeyer, I. (1990) Visibility and aerosols during the 1986 Carbonaceous Species
37 Methods Comparison Study. Aerosol Sci. Technol. 12: 149-160.
38
39 Calvert, J. G.; Su, F.; Bottenheim, J. W.; Strausz, O. P. (1978) Mechanism of the homogeneous oxidation of
40 sulfur dioxide in the troposphere. In: Sulfur in the atmosphere: proceedings of the international symposium;
41 September 1977; Dubrovnik, Yugoslavia. Atmos. Environ. 12: 197-226.
42
43 Calvert, J. G.; Lazrus, A.; Kok, G.; Heikes, B.; Walega, J.; Cantrell, C. A. (1985) Chemical mechanisms of
44 acid generation in the troposphere. Nature (London) 317: 27-38.
45
46 Cambray, R. S. (1989) Radioactive fallout in air and rain: results to the end of 1987. Harwell, United
47 Kingdom: Atomic Energy Establishment; report no. AERE-R 13226.
48
49 Cambray, R. S.; Cawse, P. A.; Garland, J. A.; Gibson, J. A. B.; Johnson, P.; Lewis, G. N. J.; Newton, D.;
50 Salmon, L.; Wade, B. O. (1987) Observations on radioactivity from the Chernobyl accident. Nucl. Energy
51 (Br. Nucl. Energy Soc.) 26: 77-101.
52
53 Carpenter, S. B.; Montgomery, T. L.; Leavitt, J. M.; Colbaugh, W. C.; Thomas, F. W. (1971) Principal
54 plume dispersion models: TVA power plants. J. Air Pollut. Control Assoc. 21: 491-495.
April 1995 3.175 DRAFT-DO NOT QUOTE OR CITE
-------
1 Carras, J. N.; Williams, D. J. (1981) The long-range dispersion of a plume from an isolated point source.
2 Atmos. Environ. 15: 2205-2217.
3
4 Carras, J. N.; Williams, D. J. (1988) Measurements of relative a up to 1800 km from a single source. Atmos.
5 Environ. 22: 1061-1069.
6
7 Carter, E. J.; Borys, R. D. (1993) Aerosol cloud chemical fractionation: enrichment factor analysis of cloud
8 water. J. Atmos. Chem. 17: 277-292.
9
10 Cass, G. R. (1977) Methods for sulfate air quality management with applications to Los Angeles
11 [Ph.D. dissertation]. Pasadena, CA: California Institute of Technology; EQL report no. 16-2.
12
13 Cass, G. R. (1979) On the relationship between sulfate air quality and visibility with examples in Los Angeles.
14 Atmos. Environ. 13: 1069-1084.
15
16 Cass, G. R.; Shair, F. H. (1984) Sulfate accumulation in a sea breeze/land breeze circulation system.
17 J. Geophys. Res. 89: 1429-1438.
18
19 Cass, G. R.; Boone, P. M.; Macias, E. S. (1982) In: Wolff, G. T.; Klimisch, R. L., eds. Particulate carbon:
20 atmospheric life cycle. New York, NY: Plenum Press; pp. 207-240.
21
22 Cawse, P. A. (1981) Trace elements in the atmosphere of the United Kingdom. Presented at: ESNA meeting;
23 September-October; Aberdeen, Scotland.
24
25 Chamberlain, A. C. (1983) Deposition and resuspension. In: Pruppacher et al., eds. Precipitation scavenging,
26 dry deposition, and resuspension, v. 2. New York, NY: Elsevier Science Publishing Co., Inc.;
27 pp. 731-751.
28
29 Chameides, W. L. (1984) The photochemistry of a remote stratiform cloud. J. Geophys. Res. [Atmos.]
30 89: 4739-4755.
31
32 Chameides, W. L.; Davis, D. D. (1982) The free radical chemistry of cloud droplets and its impact upon the
33 composition of rain. J. Geophys. Res. C: Oceans Atmos. 87: 4863-4877.
34
35 Chan, W. H.; Tang Al, J. S.; Chung, D. H. S.; Lusis, M. A. (1986) Concentration and deposition of trace
36 metals in Ontario, 1982. Water Air Soil Pollut. 29: 373-389.
37
38 Chang, S. G.; Brodzinsky, R.; Gundel, L. A.; Novakov, T. (1982) Chemical and catalytic properties of
39 elemental carbon. In: Wolff, G. T.; Klimisch, R. L., eds. Particulate carbon: atmospheric life cycle.
40 New York, NY: Plenum Press; pp. 158-181.
41
42 Chang, J. S.; Middleton, P. B.; Stockwell, W. R.; Walcek, C. J.; Pleim, J. E.; Lansford, H. H.; Madronich,
43 S.; Binkowski, F. S.; Seaman, N. L.; Stauffer, D. R. (1991) The regional acid deposition model and
44 engineering model. In: Irving, P. M., ed. Acidic deposition: state of science and technology, volume I:
45 emissions, atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation
46 Assessment Program. (State of science and technology report no. 4).
47
48 Charlson, R. J.; Lovelock, J. E.; Andreae, M. O.; Warren, S. G. (1987) Oceanic phytoplankton, atmospheric
49 sulfphur cloud, albedo, and climate. Nature (London) 326: 655-661.
50
51 Cheng, Y.-S.; Carpenter, R. L.; Barr, E. B.; Hobbs, C. H. (1985) Size distribution of fine particle emissions
52 from a steam plant with a fluidized-bed coal combustor. Aerosol Sci. Technol. 4: 175-189.
53
April 1995 3-176 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cheng, M.-D.; Lioy, P. J.; Opperman, A. J. (1988) Resolving PM10 data collected in New Jersey by various
2 multivariate analysis techniques. In: Mathai, C. V.; Stonefield, D. H., eds. Transactions: PM10
3 implementation of standards. Pittsburgh, PA: Air Pollution Control Association; pp. 472-483.
4
5 Ching, J. K. S.; Alkezweeny, A. J. (1986) Tracer study of vertical exchange by cumulus clouds. J. Clim.
6 Appl. Meteorol. 25: 1702-1711.
7
8 Choularton, T. W.; Wicks, A. J.; Downer, R. M.; Gallagher, M. W.; Penkett, S. A.; Bandy, B. J.; Dollard,
9 G. J.; Jones, B. M. R.; Davies, T. D.; Gay, M. J.; Tyler, B. J.; Fowler, D.; Cape, J. N.; Hargreaves,
10 K. J. (1992) A field study of the generation of nitrate in a hill cap cloud. Environ. Pollut. 75: 69-73.
11
12 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards,
13 L. W. (1992) PM10 source apportionment in California's San Jaoquin Valley. Atmos. Environ. Part A
14 26: 3335-3354.
15
16 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards,
17 L. W. (1993) PM10 and PM2 s compositions in California's San Joaquin Valley. Aerosol Sci. Technol.
18 18: 105-128.
19
20 Chow, J. C.; Watson, J. G.; Fujita, E. M.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994) Temporal and
21 spatial variations of PM2 5 and PM10 aerosol in the Southern California Air Quality Study. Atmos. Environ.
22 28: 2061-2080.
23
24 Chughtai, A. R.; Jassim, J. A.; Peterson, J. H.; Stedman, D. H.; Smith, D. M. (1991) Spectroscopic and
25 solubility characteristics of oxidized soots. Aerosol Sci. Technol. 15: 112-126.
26
27 Clark, T. L.; Cohn, R. D. (1990) The Across North America Tracer Experiment (ANATEX) model evaluation
28 study. Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Research and
29 Exposure Assessment Laboratory; EPA report no. EPA/600/3-90/051. Available from: NTIS, Springfield,
30 VA; PB90-261454.
31
32 Clark, P. A.; Fletcher, I. S.; Kallend, A. S.; McElroy, W. J.; Marsh, A. R. W.; Webb, A. H. (1984)
33 Observations of cloud chemistry during long-range transport of power plant plumes. Atmos. Environ.
34 18: 1849-1858.
35
36 Clarke, A. D. (1989) Aerosol light absorption by soot in remote environments. Aerosol Sci. Technol.
37 10: 161-171.
38
39 Clarke, A. D. (1992) Atmospheric nuclei in the remote free-troposphere. J. Atmos. Chem. 14: 479-488
40
41 Clarke, A. G.; Karani, G. N. (1992) Characterization of the carbon content of atmospheric aerosols. J. Atmos.
42 Chem. 14: 119-128.
43
44 Clarke, A. G.; Radojevic, M. (1987) Oxidation of SO2 in rainwater and its role in acid rain chemistry. Atmos.
45 Environ. 21: 1115-1123.
46
47 Clarke, J. F.; Clark, T. L.; Ching, J. K. S.; Haagenson, P. L.; Husar, R. B.; Patterson, D. E. (1983)
48 Assessment of model simulation of long-distance transport. Atmos. Environ. 17: 2449-2462
49
50 Clarke, A. D.; Weiss, R. E.; Charlson, R. J. (1984) Elemental carbon aerosols in the urban, rural and
51 remote-marine troposphere and in the stratosphere: interferences from light absorption data and
52 consequences regarding radiative transfer. Sci. Total Environ. 36: 97-102.
53
April 1995 3.177 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cobourn, W. G.; Husar, R. B.; Husar, J. D. (1978) Continuous in situ monitoring of ambient paniculate sulfur
2 using flame photometry and thermal analysis. In: Husar, R. B.; Lodge, J. P., Jr.; Moore, D. J., eds.
3 Sulfur in the atmosphere: proceedings of the international symposium; September 1977; Dubrovnik,
4 Yugoslavia. Atmos. Environ. 12: 89-98.
5
6 Cocks, A. T.; McElroy, W. L.; Wallis, P. G. (1982) The oxidation of sodium sulphite solutions by hydrogen
7 peroxide. Central Electricity Research Laboratories; report no. RD/L/2215N81.
8
9 Cofer, W. R.; Stevens, R. K.; Winstead, E. L.; Pinto, J. P.; Sebacher, D. L.; Abdulraheem, M. Y.; Alsahafi,
10 M.; Mazurek, M. A.; Rasmussen, R. A. (1992) Kuwaiti oil fires: composition of source smoke. J.
11 Geophys. Res. 97: 14,521-14,525.
12
13 Colbeck, L; Appleby, L.; Hardman, E. J.; Harrison, R. M. (1990) The optical properties and morphology of
14 cloud-processed of Denver's ambient paniculate. Atmos. Environ. 21: 527-538.
15
16 Coles, D. G.; Ragaini, R. G.; Ondov, J. M.; Fisher, G. L.; Silberman, D.; Prentice, B. A. (1979) Chemical
17 studies of stack fly ash from a coal-fired power plant. Environ. Sci. Technol. 13: 455-459.
18
19 Collett, J. L.; Oberholzer, B.; Mosimann, L.; Staehelin, J.; Waldrogel, A. (1993a) Contributions of cloud
20 processes to precipitation chemistry in mixed-phase clouds. Water Air Soil Pollut. 68: 43-57.
21
22 Collett, J., Jr.; Oberholzer, B.; Staehelin, J. (1993b) Cloud chemistry in Mt. Rigi Switzerland: dependence on
23 drop size and relationship to precipitation chemistry. Atmos. Environ. Part A 27: 33-42.
24
25 Colvile, R. N.; Choularton, T. W.; Gallagher, M. W.; Wicks, A. J.; Downer, R. M.; Tyler, B. J.;
26 Storeton-West, K. J.; Fowler, D.; Cape, J. N.; Dollard, G. J.; Davies, T. J.; Jones, B. M. R.; Penkett,
27 S. A.; Bandy, B. J.; Burgess, R. A. (1994) Observation on Great Dun Fell of the pathways by which
28 oxides of nitrogen are converted to nitrate. Atmos. Environ. 28: 397-408.
29
30 Cooper, J. A.; Watson, J. G., Jr. (1980) Receptor oriented methods of air paniculate source apportionment.
31 J. Air Pollut. Control Assoc. 30: 1116-1125.
32
33 Cotton, W. R.; Anthes, R. A. (1989) Storm and cloud dynamics. New York, NY: Academic Press, Inc.
34 (Dmowska, R.; Holton, J. R., eds. International geophysics series: v. 44).
35
36 Countess, R. J.; Wolff, G. T.; Cadle, S. H. (1980) The Denver winter aerosol: a comprehensive chemical
37 characterization. J. Air Pollut. Control Assoc. 30: 1194-1200.
38
39 Countess, R. J.; Cadle, S. H.; Groblicki, P. J.; Wolff, G. T. (1981) Chemical analysis of size-segregated
40 samples of Denver's ambient particulate. J. Air Pollut. Control Assoc. 31: 247-252.
41
42 Coutant, R. W.; Brown, L.; Chuang, J. C.; Riggin, R. M.; Lewis, R. G. (1988) Phase distribution and artifact
43 formation in ambient air sampling for polynuclear aromatic hydrocarbons. Atmos. Environ. 22: 403-409.
44
45 Coutant, R. W.; Callahan, P. J.; Chuang, J. C.; Lewis, R. G. (1992) Efficiency of silicone-grease-coated
46 denuders for collection of polynuclear aromatic hydrocarbons. Atmos. Environ. Part A 26: 2831-2834.
47
48 Covert, D. S.; Heintzenberg, J. (1984) Measurement of the degree of internal/external mixing of hygroscopic
49 compounds and soot in atmospheric aerosols. Sci. Total Environ. 36: 347-352.
50
51 Covert, D. S.; Charlson, R. J.; Ahlquist, N. C. (1972) A study of the relationship of chemical composition and
52 humidity to light scattering by aerosols. J. Appl. Meteorol. 11: 968-976.
53
April 1995 3-178 DRAFT-DO NOT QUOTE OR CITE
-------
1 Covert, D. S.; Heintzenberg, J.; Hansson, H.-C. (1990) Electro-optical detection of external mixtures in
2 aerosols. Aerosol Sci. Technol. 12: 446-456.
3
4 Covert, D. S.; Kapustin, V. N.; Quinn, P. K.; Bates, T. S. (1992) New particle formation in the marine
5 boundary layer. J. Geophys. Res. [Atmos.] 97: 20581-20589.
6
7 Cowherd, C., Jr.; Axtell, K., Jr.; Guenther, C. M.; Jutze, G. A. (1974) Development of emission factors for
8 fugitive dust sources. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air
9 Quality Planning and Standards; EPA report no. EPA-450/3-74-037. Available from: NTIS, Springfield,
10 VA; PB-238262.
11
12 Currie, L. A.; Stafford, T. W.; Sheffield, A. E.; Klouda, G. A.; Wise, S. A.; Fletcher, R. A. (1989)
13 Microchemical and molecular dating. Radiocarbon 31: 448-463.
14
15 Currie, L. A.; Sheffield, A. E.; Riederer, C.; Gordon, G. E. (1994) Improved atmospheric understanding
16 through exploratory data analysis and complementary modeling: the urban K-Pb-C system. Atmos. Environ.
17 28: 1359-1369.
18
19 d'Alameida, G. (1989) Desert aerosol: characteristics and effects on climate. In: Leinen, M.; Sarnthein, M.
20 Paleoclimatology and paleometeorology: modern and past patterns of global atmospheric transport.
21 Dordrect, The Netherlands: Kluwer Academic Publishers; pp. 311-338.
22
23 Daisey, J. M. (1987) Chemical composition of inhalable paniculate matter—seasonal and inters ite comparisons.
24 In: Lioy, P. J.; Daisey, J. M., eds. Toxic air pollution: a comprehensive study of non-criteria air
25 pollutants. Chelsea, MI: Lewis Publishers, Inc.; pp. 45-65.
26
27 Danielsen, E. F. (1961) Trajectories: isobaric, isentropic and actual. J. Meteorol. 18: 479-486.
28
29 Dasch, J. M.; Cadle, S. H. (1989) Atmospheric carbon particles in the Detroit urban area: wintertime sources
30 and sinks. Aerosol Sci. Technol. 10: 236-248.
31
32 Dattner, S. L.; Hopke, P. K. (1982) Receptor models applied to contemporary pollution problems.
33 In: Proceedings: receptor models applied to contemporary pollution problems. Pittsburgh, PA: Air Pollution
34 Control Association; pp. 1-5.
35
36 Daum, P. H. (1988) Processes determining cloudwater composition: inferences from field measurements.
37 In: Unsworth, M. H.; Fowler, D., eds. Acid deposition at high elevation sites. Dordrecht, The
38 Netherlands: Kluwer; pp. 139-153.
39
40 Daum, P. H.; Kelly, T. J.; Schwartz, S. E.; Newman, L. (1984) Measurements of the chemical composition of
41 stratiform clouds. Atmos. Environ. 18: 2671-2684.
42
43 Daum, P. H.; Schwartz, S. E.; Newman, L. (1984) Acidic and related constituents in liquid-water clouds.
44 J. Geophys. Res. 89: 1447-1458.
45
46 Daum, P. H.; Kelly, T. J.; Strappe, J. W.; Leaitch, W. R.; Joe, P.; Schemenauer, R. S.; Isaac, G. A.;
47 Anlauf, K. G.; Wiebe, H. A. (1987) Chemistry and physics of a winter stratus cloud layer: a case study.
48 J. Geophys. Res. 92: 8426-8436.
49
50 Daum, P. H.; Kleinman, L. I.; Hill, A. J.; Lazrus, A. L.; Leslie, A. C. D.; Bushess, K.; Boatman, J. (1990)
51 Measurement and interpretation of concentrations of H2O2 and related species in the upper midwest during
52 summer. J. Geophys. Res. D: Atmos. 95: 9857-9871.
53
April 1995 3.179 DRAFT-DO NOT QUOTE OR CITE
-------
1 Daum, P. H.; Al-Sunaid, A.; Busness, K. M.; Hales, J. M.; Mazurek, M. (1993) Studies of the Kuwait oil fire
2 plume during midsummer 1991. J. Geophys. Res. 98: 16,809-16,827.
3
4 Davidson, C. I.; Wu, Y.-L. (1990) Dry deposition of panicles and vapors. In: Lindberg, S. E.; Page, A. L.;
5 Norton, S. A., eds. Acidic precipitation: v. 3, sources, deposition, and canopy interactions. New York,
6 NY: Springer-Verlag; pp. 103-216.
7
8 Davis, E. J.; Ray, A. K. (1978) Submicron droplet evaporation in the continuum and non-continuum regimes.
9 J. Aerosol Sci. 9: 411-422.
10
11 Dayal, P.; Neuroth, G. R.; Lowe, T.P. (1992) Toxic metal characterization of coal-fired power plant
12 emissions. Transactions AWMA 1: 199-210.
13
14 De Bock, L. A.; Van Malderen, H.; Van Grieken, R. E. (1994) Individual aerosol particle composition
15 variations in air masses crossing the North sea. Environ. Sci. Technol. 28: 1513-1520.
16
17 De Valk, J. P. J. M. M. (1994) A model for cloud chemistry: a comparison between model simulations and
18 observations in stratus and cumulus. Atmos. Environ. 28: 1665-1678.
19
20 DeSantis, F.; Allegrini, I. (1992) Heterogeneous reactions of SO2 and NO2 on carbonaceous surfaces. Atmos.
21 Environ. 26: 3061-3064.
22
23 Delumyea, R. D.; Kalivretenos, A. (1987) Elemental carbon and lead content of fine particles from American
24 and French cities of comparable size and industry, 1985. Atmos. Environ. 21: 1643-1647.
25
26 Dennis, R. L.; McHenry, J. N.; Barchet, W. R.; Binkowski, F. S.; Byun, D. W. (1993) Correcting RADM's
27 sulfate underprediction: discovery and correction of model errors and testing the corrections through
28 comparisons against field data. Atmos. Environ. Part A 27: 975-997.
29
30 Dickerson, R. R.; Huffman, G. J.; Luke, W. T.; Nunnermacker, L. J.; Pickering, K. E.; Leslie, A. C. D.;
31 Lindsey, C. G.; Slinn, W. G. N.; Kelly, T. J.; Daum, P. H.; Delany, A. C.; Greenberg, J. P.;
32 Zimmerman, P. R.; Boatman, J. F.; Ray, J. D.; Stedman, D. H. (1987) Thunderstorms: an important
33 mechanism in the transport of air pollutants. Science (Washington, DC) 235: 460-465.
34
35 Dickson, R. J.; Wilkinson, J. G.; Bruckman, L.; Tesche, T. W. (1991) Conceptual formulation of the
36 SARMAP emissions modeling system. Presented at: the 84th annual meeting of the Air and Waste
37 Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air and Waste Management
38 Association.
39
40 Dittenhoefer, A. C.; de Pena, R. G. (1978) A study of production and growth of sulfate particles in plumes
41 from a coal-fired power plant. Atmos. Environ. 12: 297-306.
42
43 Dlugi, R. (1989) Chemistry and deposition of soot particles in moist air and fog. Aerosol Sci. Technol.
44 10: 93-105.
45
46 Dlugi, R.; Glisten, H. (1983) The catalytic and photocatalytic activity of coal fly ashes. Atmos. Environ.
47 17: 1765-1771.
48
49 Dod, R. L.; Brown, N. J.; Mowrer, F. W.; Novakov, T.; Williamson, R. B. (1989) Smoke emission factors
50 from medium-scale fires: part 2. Aerosol Sci. Technol. 10: 20-21.
51
52 Dodd, J. A.; Ondov, J. M.; Tuncel, G.; Dzubay, T. G.; Stevens, R. K. (1991) Multimodal size spectra of
53 submicrometer particles bearing various elements in rural air. Environ. Sci. Technol. 25: 890-903.
54
April 1995 3-180 DRAFT-DO NOT QUOTE OR CITE
-------
1 Doyle, G. J.; Tuazon, E. C.; Graham, R. A.; Mischke, T. M.; Winer, A. M.; Pitts, J. N., Jr. (1979)
2 Simultaneous concentrations of ammonia and nitric acid in a polluted atmosphere and their equilibrium
3 relationship to paniculate ammonium nitrate. Environ. Sci. Technol. 13: 1416-1419.
4
5 Draxler, R. R. (1982) Measuring and modeling the transport and dispersion of Krypton-85 1500 km from a
6 point source. Atmos. Environ. 16: 2763-2776.
7
8 Draxler, R. R. (1984) Diffusion and transport experiments. In: Randerson, D., eds. Atmospheric science and
9 power production. Washington, DC: U.S. Department of Energy, Office of Science and Technological
10 Information; pp. 367-422; DOE report no. DOE/TIC-27601.
11
12 Draxler, R. R.; Dietz, R.; Lagomarsino, R. J.; Start, G. (1991) Across North America Tracer Experiment
13 (ANATEX): sampling and analysis. Atmos. Environ. Part A 25: 2815-2836.
14
15 Duce, R. A.; Mohnen, V. A.; Zimmerman, P. R.; Grosjean, D.; Cautreels, W.; Chatfield, R.; Jaenicke, R.;
16 Ogren, J. A.; Pellizzari, E. D.; Wallace, G. T. (1983) Organic material in the global troposphere. Rev.
17 Geophys. Space Phys. 21: 921-952.
18
19 Dulac, F.; Buat-Menard, P.; Ezat, U.; Melki, S.; Bengametti, G. (1989) Atmospheric input of trace metals to
20 the western Mediterranean: uncertainties in modeling dry deposition from cascade impactor data. Tellus
21 Ser. B41: 362-378.
22
23 Durham, J. L.; Wilson, W. E.; Ellestad, T. G.; Willeke, K.; Whitby, K. T. (1975) Comparison of volume and
24 mass distributions for Denver aerosols. Atmos. Environ. 9: 717-722.
25
26 Dzubay, T. G.; Mamane, Y. (1989) Use of electron microscopy data in receptor models for PM-10. Atmos.
27 Environ. 23: 467-476.
28
29 Dzubay, T. G.; Stevens, R. K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E.; Courtney, W. J. (1988)
30 A composite receptor method applied to Philadelphia aerosol. Environ. Sci. Technol. 22: 46-52.
31
32 Eatough, D. J.; Richter, B. E.; Eatough, N. L.; Hansen, L. D. (1981) Sulfur chemistry in smelter and power
33 plant plumes in the western U.S. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr. Plumes and visibility:
34 measurements and model components: proceedings of the symposium; November 1980; Grand Canyon
35 National Park, AZ. Atmos. Environ. 15: 2241-2253.
36
37 Eatough, D. J.; Arthur, R. J.; Eatough, N. L.; Hill, M. W.; Mangelson, N. F.; Richter, B. E.; Hansen,
38 L. D.; Cooper, J. A. (1984) Rapid conversion of SO2(g) to sulfate in a fog bank. Environ. Sci. Technol.
39 18: 855-859.
40
41 Eatough, D. J.; Sedar, B.; Lewis, L.; Hansen, L. D.; Lewis, E. A.; Farber, R. J. (1989) Determination of
42 semivolatile organic compounds in particles in the Grand Canyon area. Aerosol Sci. Technol. 10: 438-449.
43
44 Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. (1993) A
45 multiple-system, multi-channel diffusion denuder sampler for the determination of fine-particulate organic
46 material in the atmosphere. Atmos. Environ. Part A 27: 1213-1219.
47
48 Eisele, F. L.; Bradshaw, J. D. (1993) The elusive hydroxyl radical: measuring OH in the atmosphere. Anal.
49 Chem. 65: 927A-939A.
50
51 Eisele, F. L.; Tanner, D. J. (1993) Measurement of the gas phase concentration of H2SO4 and methane sulfonic
52 acid and estimates of H2SO4 production and loss in the atmosphere. J. Geophys. Res. [Atmos.] 98:
53 9001-9010.
54
April 1995 3_18l DRAFT-DO NOT QUOTE OR CITE
-------
1 Eisele, F. L.; Mount, G. H.; Fehsenfeld, F. C.; Harder, J.; Marovich, E.; Parrish, D. D.; Roberts, J.;
2 Trainer, M.; Tanner, D. (1994) Intercomparison of tropospheric OH and ancillary trace gas measurements
3 at Fritz Peak Observatory, Colorado. J. Geophys. Res. [Atmos.] 99: 18605-18626.
4
5 Eisenreich, S. (1980) Atmospheric input of trace metals to Lake Michigan. Water Air Soil Pollut. 13: 287-301.
6
7 Eldering, A.; Cass, G. R. (1994) A source-oriented model for air pollutant effects on visibility. J. Geophys.
8 Res. [Atmos.]: submitted.
9
10 Eldering, A.; Cass, G. R.; Moon, K. C. (1994) An air monitoring network using continuous particle size
11 distribution monitors: connecting pollutant properties to visibility via Mie scattering calculations. Atmos.
12 Environ. 28: 2733-2749.
13
14 Eldred, R. A.; Cahill, T. A.; Flocchini, R. G. (1994) Composition of PM10 and PM2 5 aerosols in the
15 IMPROVE network. Proceedings of the international specialty conference on aerosol and atmospheric
16 optics: radiative balance and visual air quality, volume A. Air Waste: submitted.
17
18 Er-El, J.; Peskin, R. L. (1981) Relative diffusion of constant-level balloons in the southern hemisphere.
19 J. Atmos. Sci. 38: 2264-2274.
20
21 Erickson, R. E.; Yates, L. M.; Clark, R. L.; McEwen, D. (1977) The reaction of sulfur dioxide with ozone in
22 water and its possible atmospheric significance. Atmos. Environ. 11: 813-817.
23
24 Facchini, M. C.; Fuzzi, S.; Lind, J. A. (1992) Phase-partitioning and chemical reactions of
25 low-molecular-weight organic compounds. Tellus Ser. B 44B: 533-544.
26
27 Falerios, M.; Schild, K.; Sheehan, P.; Paustenbach, D. (1992) Hexavalent chromium: airborne concentrations.
28 J. Air Waste Manage. Assoc. 42: 40-48.
29
30 Ferber, G. J.; Telegadas, K.; Heffter, J. L.; Dickson, C. R.; Dietz, R. N.; Krey, P. W. (1981) Demonstration
31 of a long-range atmospheric tracer system using perfluorocarbons. Silver Springs, MD: National Oceanic
32 and Atmospheric Administration; NOAA report no. NOAA TM ERL ARL-101.
33
34 Ferber, G. J.; Heffter, J. L.; Draxler, R. R.; Lagomarsino, R. J.; Thomas, F. L.; Dietz, R. N.; Benkovitz,
35 C. M. (1986) Cross-Appalachian Tracer Experiment (CAPTEX-83) final report. Silver Springs, MD:
36 National Oceanic and Atmospheric Administration; NOAA report no. NOAA TM ERL ARL-142.
37
38 Fernandez de la Mora, J. (1986) Inertia and interception in the deposition of particles from boundary layers.
39 Aerosol Sci. Technol. 5: 261-266.
40
41 Fernandez de la Mora, J.; Friedlander, S. K. (1982) Int. J. Heat Mass Transfer 25: 1725-1735.
42
43 Finlayson-Pitts, B. J.; Pitts, J. N. (1986) Paniculate matter in atmosphere: primary and secondary particles.
44 In: Finlayson-Pitts, B. J.; Pitts, J. N., eds. Atmospheric chemistry: fundamentals and experimental
45 techniques. New York, NY: John Wiley and Sons; pp. 783-802.
46
47 Finlayson-Pitts, B. J.; Pitts, J. N., Jr. (1986) Atmospheric chemistry: fundamentals and experimental
48 techniques. New York, NY: John Wiley & Sons.
49
50 Fisher, G. L. (1980) Size-related chemical and physical properties of power plant fly ash. In: Willeke, K., ed.
51 Generation of aerosols and facilities for exposure experiments; April 1979; Honolulu, HI. Arm Arbor, MI:
52 Ann Arbor Science Publishers, Inc.; pp. 203-214.
53
April 1995 3-182 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fitzgerald, J. W. (1973) Dependence of the supersaturation spectrum of CCN on aerosol size distribution and
2 composition. J. Atmos. Sci. 30: 628-634.
3
4 Fitzgerald, J. W.; Spyers-Duran, P. A. (1973) Changes in cloud nucleus concentration and cloud droplet size
5 distribution associated with pollution from St. Louis. J. Appl. Meteorol. 12: 511-516.
6
7 Flossman, A. I.; Hall, W. D.; Pruppacher, H. R. (1985) A theoretical study of the wet removal of atmospheric
8 pollutants—Part I. the redistribution of aerosol particles captured through nucleation and impaction
9 scavenging by growing cloud drops. J. Atmos. Sci. 42: 583-606.
10
11 Foreman, W. T.; Bidleman, T. F. (1990) Semivolatile organic compounds in the ambient air of Denver,
12 Colorado. Atmos. Environ. Part A 2405-2416.
13
14 Forkel, R.; Seidl, W.; Dlugi, R.; Deigle, E. (1990) A one dimensioal numerical model to simulate formation
15 and balance of sulfate during radiation fog events. J. Geophys. Res. 95: 18501-18515.
16
17 Forrest, J.; Newman, L. (1973) Sampling and analysis of atmospheric sulfur compounds for isotope ratio
18 studies. Atmos. Environ. 7: 561-573.
19
20 Forrest, J.; Newman, L. (1977a) Further studies on the oxidation of sulfur dioxide in coal-fired power plant
21 plumes. Atmos. Environ. 11: 465-474.
22
23 Forrest, J.; Newman, L. (1977b) Oxidation of sulfur dioxide in the Sudbury smelter plume. Atmos. Environ.
24 11:517-520.
25
26 Forrest, J.; Garber, R.; Newman, L. (1979) Formation of sulfate, ammonium and nitrate in an oil-fired power
27 plant plume. Atmos. Environ. 13: 1287-1297.
28
29 Forrest, J.; Garber, R. W.; Newman, L. (1981) Conversion rates in power plant plumes based on filter pack
30 data: the coal-fired Cumberland plume. Atmos. Environ. 15: 2273-2282.
31
32 Freeman, D. L.; Chow, J. C.; Egami, R. T.; Watson, J. G. (1989) A receptor and dispersion modeling
33 software package. In: Watson, J. G., ed. Transactions: receptor models in air resources management.
34 Pittsburgh, PA: Air and Waste Management Association; pp. 243-268.
35
36 Frick, G. M.; Hoppel, W. A. (1993) Airship measurements of aerosol size distributions, cloud droplet spectra,
37 and trace gas concentrations in the marine boundary layer. Bull. Am. Meteorol. Soc. 74: 2195-2202.
38
39 Fried, A.; Henry, B. E.; Calvert, J. G.; Mozurkewich, M. (1994) The reaction probability of N2O5 with
40 sulfuric acid aerosols at stratospheric temperatures and compositions. J. Geophys. Res. [Atmos.]
41 99: 3517-3532.
42
43 Friedlander, S. K. (1970) The characterization of aerosols distributed with respect to size and chemical
44 composition. J. Aerosol Sci. 1: 295-307.
45
46 Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John
47 Wiley & Sons, Inc.
48
49 Friedlander, S. K.; Turner, J. R.; Hering, S. V. (1986) J. Aerosol Sci. 17: 240-244.
50
51 Fuchs, N. A. (1964) The mechanics of aerosols. New York, NY: Pergamon Press.
52
53 Fuchs, N. A.; Stutugin, A. G. (1971) Highly dispersed aerosols. In: Hidy, G. M.; Brock, J. R., eds. Topics in
54 current aerosol research. New York, NY: Pergamon; pp. 4-60.
April 1995 3483 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fung, C. S.; Misra, P. K.; Bloxam, R.; Wong, S. (1991) A numerical experiment on the relative importance of
2 H2O2 and O3 in aqueous conversion of SO2 to SO2"/4. Atmos. Environ. Part A 25: 411-423.
3
4 Fuzzi, S.; Castillo, R. A.; Jiusto, J. E.; Lala, G. G. (1984) Chemical composition of radiation fog water at
5 Albany, New York, and its relationship to fog microphysics. J. Geophys. Res. [Atmos.] 89: 7159-7164.
6
7 Fuzzi, S.; Orsi, G.; Nardini, G.; Fachini, M. C.; Mariotti, M.; McLaren, S.; McLaren, E. (1988)
8 Heterogeneous processes in the Po Valley (Italy) radiation fog. J. Geophys. Res. 93: 11141-11151.
9
10 Gagosian, R. B.; Peltzer, E. T.; Merrill, J. T. (1987) Long range transport of terrestrially derived lipids in
11 aerosols from South Pacific. Nature (London) 325: 800-803.
12
13 Galloway, J. N.; Thornton, J. D.; Norton, S. A.; Volchok, H. L.; McLean, R. A. N. (1982) Trace metals in
14 atmospheric deposition: a review and assessment. Atmos. Environ. 16: 1677-1700.
15
16 Garland, J. A. (1979) Resuspension of paniculate matter from grass and soil. Harwell, United Kingdom:
17 UK Atomic Energy Authority; report no. AERE-R9452.
18
19 Garland, J.; Nicholson, K. (1991) A review of methods for sampling large airborne particles and associated
20 radioactivity. J. Aerosol Sci. 22: 479-499.
21
22 Gatz, D. F. (1977) Scavenging ratio measurements in METROMEX. In: Semonin, R. G.; Beadle, R. W., eds.
23 Precipitation scavenging (1974): proceedings of a symposium; October 1974; Champaign, IL. Oak Ridge,
24 TN: Energy Research and Development Administration; pp. 71-87. Available from: NTIS, Springfield,
25 VA; CONF-741003. (ERDA symposium series no. 41).
26
27 Gebhart, K. A.; Lattimer, D. A.; Sisler, J. F. (1990) Empirical orthogonal function analysis of the particulate
28 sulfate concentrations measured during WHITEX. In: Mathai, C. V., ed. Transactions: visibility and fine
29 particles, Pittsburgh, PA: Air and Waste Management Association; pp. 860.
30
31 Gebhart, K. A.; Malm, W. C.; Day, D. (1994) Examination of the effects of sulfate acidity and relative
32 humidity on light scattering at Shenandoah National Park. Atmos. Environ. 28: 841-849.
33
34 Gelbard, F.; Seinfeld, J. H. (1978) Numerical solution of the dynamic equation for particulate systems.
35 J. Comp. Phys. 28: 357-375.
36
37 Gelbard, F.; Seinfeld, J. H. (1980) Simulation of multicomponent aerosol dynamics. J. Colloid Interface Sci.
38 78:485-501.
39
40 Gerde, P.; Scholander, P. (1989) Mass transfer rates of polycyclic aromatic hydrocarbons between micron-size
41 particles and their environment—theoretical estimates. Environ. Health Perspect. 79: 249-258.
42
43 Germani, M. S.; Buseck, P. R. (1991) Automated scanning electron microscopy for atmospheric particle
44 analysis. Anal. Chem. 63: 2232-2237.
45
46 Gervat, G. P.; Clark, P. A.; Marsh, A. R. W. (1988) Controlled chemical kinetics experiment in cloud:
47 a review of the CERL/UMIST Great Dunn Fell Project. In: Unsworth, M. H.; Fowler, D., eds. Acid
48 deposition at high elevation sites. Hingham, MA: Kluwer Academic; pp. 283-298.
49
50 Gery, M. W.; Fox, D. L.; Jeffries, H. E.; Stockburger, L.; Weathers, W. S. (1985) A continuous stirred tank
51 reactor investigation of the gas-phase reaction of hydroxyl radicals and toluene. Int. J. Chem. Kinet.
52 17: 931-955.
53
April 1995 3-184 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gery, M. W.; Fox, D. L.; Kamens, R. M.; Stockburger, L. (1987) Investigation of hydroxyl radical reactions
2 with o-xylene and m-xylene in a continuous stirred tank reactor. Environ. Sci. Technol. 21: 339-348.
3
4 Ghan, S. J.; Chuang, C. C.; Penner, J. E. (1994) A parameterization of cloud droplet nucleation, part I: single
5 aerosol type. Atmos. Res.: submitted.
6
7 Gifford, F. (1961) Use of routine meteorological observations for estimating atmospheric dispersion. Nucl. Saf.
8 2: 47-57.
9
10 Gill, P. S.; Graedel, T. E.; Wechsler, C. J. (1983) Organic films on atmospheric aerosol particles, fog
11 droplets, raindrops and snowflakes. Rev. Geophys. Space Phys. 21: 903.
12
13 Gillani, N. V. (1985) Modeling of chemical transformations of SOx and NOX in the polluted atmosphere: an
14 overview of approaches and current status. In: De Wispelaere, C, ed. Air pollution modeling and its
15 application IV. New York, NY: Plenum Publishers.
16
17 Gillani, N. V. (1986) Ozone formation in pollutant plumes: a reactive plume model with arbitrary crosswind
18 resolution. Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Sciences
19 Research Laboratory; EPA report no. EPA/600/3-86/051. Available from: NTIS, Springfield, VA;
20 PB86-236973.
21
22 Gillani, N. V.; Husar, R. B. (1976) Synoptic-scale haziness over the eastern US and its long range transport.
23 Presented at: 4th National AMS/SAF conference on fire and forest meteorology Boston MA
24
25 Gillani, N. V.; Pleim, J. E. (1994) Subgrid-scale features of anthropogenic emissions of VOC and NOX in the
26 context of regional Eulerian models. Atmos. Environ.: submitted.
27
28 Gillani, N. V.; Wilson, W. E. (1980) Formation and transport of ozone and aerosols in power plant plumes.
29 Ann. N. Y. Acad. Sci. 338: 276-296.
30
31 Gillani, N. V.; Wilson, W. E. (1983) Gas-to-particle conversion of sulfur in power plant plumes—II.
32 observations of liquid-phase conversions. Atmos. Environ. 17: 1739-1752.
33
34 Gillani, N. V.; Husar, R. B.; Husar, J. D.; Patterson, D. E.; Wilson, W. E., Jr. (1978) Project MISTT:
35 kinetics of paniculate sulfur formation in a power plant plume out to 300 km. Atmos. Environ.
36 12: 589-598.
37
38 Gillani, N. V.; Kohli, S.; Wilson, W. E. (1981) Gas-to-particle conversion of sulfur in power plant plumes—I.
39 parametrization of the conversion rate for dry, moderately polluted ambient conditions. Atmos Environ
40 15: 2293-2313.
41
42 Gillani, N. V.; Colby, J. A.; Wilson, W. E. (1983) Gas-to-particle conversion of sulfur in power plant
43 plumes—III. parameterization of plume-cloud interactions. Atmos. Environ. 17: 1753-1763
44
45 Gillani, N. V.; Shannon, J. D.; Patterson, D. E. (1984) Transport processes. In: Altshuller, A. P.; Linthurst,
46 R. A., eds. The acidic deposition phenomenon and its effects: critical assessment review papers, v. 1.
47 Research Triangle Park, NC: U.S. Environmental Protection Agency; EPA report no.
48 EPA/600/8-83/016AF. Available from: NTIS, Springfield, VA; PB85-100030
49
50 Gillani, N. V.; Schwartz, S. E.; Daum, P. H.; Leaitch, W. R.; Strapp, J. W.; Isaac, G. A. (1992) Fractional
51 activation of accumulation-mode aerosols in continental stratiform clouds. Presented at: The World
52 Meterological Organization workshop on cloud microphysics and global change; August; Toronto, ON,
53 Canada.
54
April 1995 3_185 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gillani, N. V.; Leaitch, W. R.; Strapp, J. W.; Isaac, G. A. (1994) Field observations in continental stratiform
2 clouds: partitioning of cloud particles between droplets and unactivated interstitial aerosols. J. Geophys.
3 Res.: submitted.
4
5 Gillette, D. A. (1974) On the production of soil wind erosion aerosols having the potential for long range
6 transport. J. Rech. Atmos. 8: 735-744.
7
8 Gillette, D. (1980) Major contributions of natural primary continental aerosols: source mechanisms. In: Kniep,
9 T. J.; Lioy, P. J., eds. Aerosols: anthropogenic and natural, sources and transport; January 1979; New
10 York, NY. Ann N. Y. Acad. Sci. 338: 348-358.
11
12 Gillette, D. A.; Hanson, K. J. (1989) Spatial and temporal variability of dust production caused by wind erosion
13 in the United States. J. Geophys. Res. 94: 2197-2206.
14
15 Gillette, D.; Nagamoto, C. (1992) [Size distribution and single particle composition for two dust storms in
16 Soviet Central Asia in September 1989 and size distributions and chemical composition of local soil].
17 In: Golitsyn, G., ed. Soviet-American experiment for the investigation of arid aerosols. St. Petersburg,
18 USSR: Typhoon; pp. 130-140.
19
20 Gillette, D. A.; Passi, R. (1988) Modeling dust emission caused by wind erosion. J. Geophys. Res.
21 93: 14,233-14,242.
22
23 Gillette, D. A.; Sinclair, P. C. (1990) Estimation of suspension of alkaline material by dust devils in the United
24 States. Atmos. Environ. Part A 24: 1135-1142.
25
26 Gillette, D.; Walker, T. (1977) Characteristics of airborne particles produced by wind erosion of sandy soil,
27 high plains of west Texas. Soil Sci. 123: 97-110.
28
29 Gillette, D. A.; Clayton, R. N.; Mayeda, T. K.; Jackson, M. L.; Sridhar, K. (1978) Tropospheric aerosols
30 from some major dust storms of the Southwestern United States. J. Appl. Meteorol. 17: 832-845.
31
32 Gillette, D. A.; Stensland, G. J.; Williams, A. L.; Barnard, W.; Gatz, D.; Sinclair, P. C.; Johnson, T. C.
33 (1992) Emissions of alkaline elements calcium, magnesium, potassium, and sodium from open sources in
34 the contiguous United States. In: Global biogeochemical cycles.
35
36 Gleason, J. F.; Sinha, A.; Howard, C. J. (1987) J. Phys. Chem. 91: 719-724.
37
38 Godish, T. (1985) Air quality. Chelsea, MI: Lewis Publishers, Inc.
39
40 Godowitch, J. M. (1989) Evaluation and sensitivity analyseis results of the MESOPUFF II model with
41 CAPTEX measurements. Research Triangle Park, NC: U.S. Environmental Protection Agency,
42 Atmospheric Research and Exposure Assessment Laboratory; EPA report no. EPA/600/3-89/056. Available
43 from: NTIS, Springfield, VA; PB89-198253.
44
45 Goldberg, E. D. (1985) Black carbon in the environment: properties and distribution. New York, NY: John
46 Wiley & Sons. (Metcalf, R. L.; Stumm, W., eds. Environmental science and technology series).
47
48 Gordon, G. E. (1980) Receptor models. Environ. Sci. Technol. 14: 792-800.
49
50 Gordon, G. E. (1988) Receptor models. Environ. Sci. Technol. 22: 1132-1142.
51
52 Gordon, G. E. (1991) Airborne particles on global and regional scales. Environ. Sci. Technol. 25: 1822-1828.
53
April 1995 3-186 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gordon, R. J.; Trivedi, N. J.; Singh, B. P.; Ellis, E. C. (1988) Characterization of aerosol organics by diffuse
2 reflectance Fourier transform infrared spectroscopy. Environ. Sci. Technol. 22: 672-677.
3
4 Goschnich, J.; Fichtner, M.; Lipp, M.; Schuricht, J.; Ache, M. (1990) Depth resolved chemical analysis of
5 environmental microparticles by secondary mass spectrometry. Appl. Surface Sci. 70: 63-67.
6
7 Graedel, T. E.; Goldberg, K. I. (1983) Kinetic studies of raindrop chemistry: 1. inorganic and organic
8 processes. J. Geophys. Res. 88: 10865-10882.
9
10 Graedel, T. E.; Weschler, C. J. (1981) Chemistry within aqueous atmospheric aerosols and raindrops. Rev.
11 Geophys. Space Phys. 19: 505-539.
12
13 Grams, G. W.; Blifford, I. H., Jr.; Gillette, D. A.; Russell, P. B. (1974) Complex index of refraction of
14 airborne soil particles. J. Appl. Meteorol. 13: 459-471.
15
16 Gray, H. A. (1986) Control of atmospheric fine primary carbon particle concentration [Ph.D. thesis]. Pasadena,
17 CA: California Institute of Technology; EQL report no. 23.
18
19 Gray, H. A.; Cass, G. R.; Huntzicker, J. J.; Heyerdahl, E. K.; Rau, J. A. (1984) Elemental and organic
20 carbon particle concentrations: a long-term perspective. Sci. Total Environ. 36: 17-25.
21
22 Gray, H. A.; Cass, G. R.; Huntzicker, J. J.; Heyerdahl, E. K.; Rau, J. A. (1986) Characteristics of
23 atmospheric organic and elemental carbon particle concentrations in Los Angeles. Environ. Sci. Technol.
24 20: 580-589.
25
26 Greenberg, R. R.; Zoller, W. H.; Gordon, G. E. (1978) Composition and size distributions of particles released
27 in refuse incineration. Environ. Sci. Technol. 12: 566-573.
28
29 Grosjean, D. (1984a) Particulate carbon in Los Angeles' air. Sci. Total Environ. 32: 133.
30
31 Grosjean, D. (1984b) Photooxidation of 1-heptane. Sci. Total Environ. 37: 195-211.
32
33 Grosjean, D. (1985) Reactions of o-cresol and nitrocresol with NOX in sunlight and with ozone-nitrogen dioxide
34 mixtures in the dark. Environ. Sci. Technol. 19: 968-974.
35
36 Grosjean, D. (1992) In situ organic aerosol formation during a smog episode: estimated production and
37 chemical functionality. Atmos. Environ. Part A 26: 953-963.
38
39 Grosjean, D.; Friedlander, S. K. (1975) Gas-particle distribution factors for organic and other pollutants in the
40 Los Angeles atmosphere. J. Air Pollut. Control Assoc. 25: 1038-1044.
41
42 Grosjean, D.; Seinfeld, J. H. (1989) Parameterization of the formation potential of secondary organic aerosols.
43 Atmos. Environ. 23: 1733-1747.
44
45 Gundel, L. A.; Guyot-Sionnest, N. S.; Novakov, T. (1989) A study of the interaction of NC^ with carbon
46 particles. Aerosol Sci. Technol. 10: 343-351.
47
48 Gundel, L. A.; Lee, V. C.; Mahanama, K. R. R.; Stevens, R. K.; Daisey, J. M. (1994) Direct determination
49 of the phase distributions of semi-volatile polycyclic aromatic hydrocarbons using annular denuders. Atmos.
50 Environ.: in press.
51
52 Gunz, D. W.; Hoffmann, M. R. (1990) Atmospheric chemistry of peroxides: a review. Atmos. Environ. Part A
53 24: 1601-1633.
54
April 1995 3-187 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hahn, J. (1980) Organic constituents of natural aerosols. Ann. N. Y. Acad. Sci. 338: 359-376.
2
3 Hall, F. P., Jr.; Duchon, C. E.; Lee, L. G.; Hagen, R. R. (1976) Long-range transport of air pollution: a case
4 study, August 1970. Mon. Weather Rev. 101: 404.
5
6 Haltiner, G. J. (1971) Numerical weather prediction. New York, NY: John Wiley & Sons, Inc.
7
8 Hamilton, R. S.; Mansfield, T. A. (1991) Airborne paniculate elemental carbon: its sources, transport and
9 contribution to dark smoke and soiling. Atmos. Environ. Part A 25: 715-723.
10
11 Ha'nel, G. (1976) The properties of atmospheric aerosol particles as functions of the relative humidity at
12 thermodynamic equilibrium with the surrounding moist air. In: Landsberg, H. E.; Van Mieghem, J., eds.
13 Advances in geophysics: v. 19. New York, NY: Academic Press; pp. 73-188.
14
15 Ha'nel, H. (1987) The role of aerosol properties during the condensational growth of cloud: a reinvestigation of
16 numerics and metaphysics. Contrib. Atmos. Phys. 60: 321-339.
17
18 Hansen, A. D. A.; Novakov, T. (1988) Real time measurements of the size fractionation of ambient black
19 carbon aerosols at elevated humidities. Aerosol Sci. Technol. 106-110.
20
21 Hansen, A. D. A.; Rosen, H. (1990) Individual measurements of the emission factor of aerosol black carbon in
22 automobile plumes. J. Air Waste Manage. Assoc. 40: 1654-1657.
23
24 Hansen, A. D. A.; Benner, W. H.; Novakov, T. (1991) S02 oxidation in laboratory clouds. Atmos. Environ.
25 25: 2521-2530.
26
27 Hansson, H.-C.; Svenningsson, B. (1994) Aerosols and clouds. In: Angeletti, G.; Restelli, G., eds.
28 Physico-chemical behaviour of atmospheric pollutants, v, 2. Proceedings of the sixth European symposium;
29 October 1993; Varese, Italy. Luxembourg, Office of Official Publications of the European Commission;
30 pp. 837-846.
31
32 Harms, D. E.; Raman, S.; Madala, R. V. (1992) An examination of four-dimensional data-assimilation
33 techniques for numerical weather prediction. Bull. Am. Meteorol. Soc. 73: 425-440.
34
35 Harrison, L. (1985) The segregation of aerosols by cloud nucleating activity: II. observation of urban aerosol.
36 J. Clim. Appl. Meteorol. 24: 312-321.
37
38 Harrison, R. M.; Msibi, I. M. (1994) Validation of techniques for fast response measurement of HNO3 and
39 NH3 and determination of the [NH3][HNO3] concentration product. Atmos. Environ. 28: 247-255.
40
41 Harrison, R. M.; Pio, C. A. (1983) Size-differentiatedl composition of inorganic atmospheric aerosols of both
42 marine and polluted continental origin. Atmos. Environ. 17: 1733-1738.
43
44 Harrison, R. M.; Pio, C. A. (1983) A comparative study of the ionic composition of rainwater and atmospheric
45 aerosols: implications for the mechanism of acidification of rainwater. Atmos. Environ. 17: 2539-2543.
46
47 Harrison, R. M.; Laxen, D. P. H.; Wilson, S. J. (1981) Chemical associations of lead, cadmium, copper, and
48 zinc in street dusts and roadside soils. Environ. Sci. Technol. 15: 1378-1383.
49
50 Hart, K. M.; Pankow, J. F. (1994) High-volume air sampler for particle and gas sampling. 2. Use of backup
51 filters to correct for the adsorption of gas-phase polycyclic aromatic hydrocarbons to the front filter.
52 Environ. Sci. Technol. 28: 655-661.
53
April 1995 3-188 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hasan, H.; Dzubay, T. G. (1987) Size distributions of species in fine particles in Denver using a microorifice
2 impactor. Aerosol Sci. Technol. 6: 29-39.
3
4 Hatakeyama, S.; Tanonaka, T.; Weng, J.; Bandow, H.; Takagi, H.; Akimoto, H. (1985) Ozone-cyclohexene
5 reaction in air: quantitative analysis of paniculate products and the reaction mechanism. Environ. Sci.
6 Technol. 19: 935-942.
7
8 Hatakeyama, S.; Ohno, M.; Weng, J.; Takagi, H.; Akimoto, H. (1987) Mechanism for the formation of
9 gaseous and paniculate products from ozone-cycloalkene reactions in air. Environ. Sci. Technol. 21: 52-57.
10
11 Hatakeyama, S.; Izumi, K.; Fukuyama, T.; Akimoto, H. (1989) Reactions of ozone with a-pinene and /3-pinene
12 in air: yields of gaseous and paniculate products. J. Geophys. Res. [Atmos.] 94: 13,013-13,024.
13
14 Hatakeyama, S.; Izumi, K.; Fukuyama, T.; Akimoto, H.; Washida, N. (1991) Reactions of OH with a-pinene
15 and 0-pinene in air: estimate of global CO production from the atmospheric oxidation of terpenes.
16 J. Geophys. Res. [Atmos.] 96: 947-958.
17
18 Hegg, D. A. (1985) The importance of liquid-phase oxidation of SO2 in the troposphere. J. Geophys. Res.
19 90: 3773-3779.
20
21 Hegg, D. A. (1991) Particle production in clouds. Geophys. Res. Lett. 18: 995-998.
22
23 Hegg, D. A.; Hobbs, P. V. (1979) Some observations of paniculate nitrate concentrations in coal-fired power
24 plant plumes. Atmos. Environ. 13: 1715-1716.
25
26 Hegg, D. A.; Hobbs, P. V. (1980) Measurements of gas-to-particle conversion in the plumes from five
27 coal-fired electric power plants. Atmos. Environ. 14: 99-116.
28
29 Hegg, D. A.; Hobbs, P. V. (1982) Measurements of sulfate production in natural clouds. Atmos. Environ.
30 16: 2663-2668.
31
32 Hegg, D. A.; Hobbs, P. V. (1983a) Errata [Hegg and Hobbs (1982)]. Atmos. Environ. 17: 1059.
33
34 Hegg, D. A.; Hobbs, P. V. (1983b) Author's reply: measurements of sulfate production in natural clouds.
35 Atmos. Environ. 17: 2632-2633.
36
37 Hegg, D. A.; Hobbs, P. V. (1986) Sulfate and nitrate chemistry in cumuliform clouds. Atmos. Environ.
38 20: 901-909.
39
40 Hegg, D. A.; Hobbs, P. V. (1987) Comparisons of sulfate production due to ozone oxidation in clouds with a
41 kinetic rate equation. Geophys. Res. Lett. 14: 719-721.
42
43 Hegg, D. A.; Hobbs, P. V. (1988) Comparisons of sulfate and nitrate production in clouds on the mid-Atlantic
44 and Pacific Northwest coast of the United States. J. Atmos. Chem. 7: 325-333.
45
46 Hegg, D. A.; Larson, T. V. (1990) The effects of microphysical parameterization on model predictions of
47 sulfate predictions in clouds. Tellus Ser. B 42: 272-284.
48
49 Hegg, D. A.; Hobbs, P. V.; Radke, L. F. (1984) Measurements of the scavenging of sulfate and nitrate in
50 clouds. Atmos. Environ. 18: 1939-1946.
51
52 Hegg, D. A.; Radke, L. F.; Hobbs, P. V. (1990) Particle production associated with marine clouds.
53 J. Geophys. Res. [Atmos.] 95: 13917-13926.
54
April 1995 3.139 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hegg, D. A.; Radke, L. F.; Hobbs, P. V. (1991) Measurements of Aitken nuclei and cloud condensation nuclei
2 in the marine atmosphere and their relationship to the DMS-cloud-climate hypothesis. J. Geophys. Res.
3 96: 18727-18733.
4
5 Hegg, D. A.; Yuen, P. F.; Larson, T. V. (1992) Modeling the effects of heterogeneous cloud chemistry on the
6 marine particle size distribution. J. Geophys. Res. 97: 12927-12933.
7
8 Heintzenberg, J. (1989) Fine particles in the global troposphere: a review. Tellus Ser. B 41: 149-160.
9
10 Heintzenberg, J.; Winkler, P. (1984) Elemental carbon in the urban aerosol: results of a seventeen month study
11 in Hamburg, FRG. Sci. Total Environ. 36: 27-38.
12
13 Heintzenberg, J.; Ogren, J. A.; Noone, K. J.; Gardneus, L. (1989) The size distribution of submicrometer
14 particles within and about stratocumulus cloud droplets on Mt. Areskutan, Sweden. Atmos. Res.
15 24:89-101.
16
17 Henry, R. C. (1991) Multivariate receptor models. In: Hopke, P. K., ed. Receptor modeling for air quality
18 management. Amsterdam, The Netherlands: Elsevier.
19
20 Henry, R. C.; Hidy, G. M. (1979) Multivariate analysis of paniculate sulfate and other air quality variables by
21 principal components—part I: annual data from Los Angeles and New York. Atmos. Environ.
22 13: 1581-1596.
23
24 Henry, R. C.; Hidy, G. M. (1982) Multivariate analysis of particulate sulfate and other air quality variables by
25 principal components—part II: Salt Lake City, UT and St. Louis, MO. Atmos. Environ. 16: 929-943.
26
27 Henry, R. C.; Kim, B. M. (1989) A factor analysis receptor model with explicit physical constraints.
28 In: Watson, J. G., ed. Transactions: receptor models in air resources management. Pittsburgh, PA: Air and
29 Waste Management Association; pp. 214-225.
30
31 Henry, R. C.; Wang, Y.-J.; Gebhart, K. A. (1990) The relationship between empirical orthogonal functions and
32 sources of air pollution. Atmos. Environ. Part A 24: 503-509.
33
34 Henry, R. C.; Lewis, C. W.; Collins, J. F. (1994) Vehicle-related source compositions from ambient data: the
35 GRACE/SAFER method. Environ. Sci. Technol. 28: 823-832.
36
37 Hering, S. V.; Friedlander, S. K. (1982) Origins of aerosol sulfur size distributions in the Los Angeles basin.
38 Atmos. Environ. 16: 2647-2656.
39
40 Hering, S. V.; McMurry, P. H. (1991) Optical counter response to monodisperse atmospheric aerosols. Atmos.
41 Environ. Part A 25: 463-468.
42
43 Hicks, B. B. (1984) Dry deposition processes. In: Altshuller, A. P.; Linthurst, R. A., eds. Acidic deposition
44 phenomenon and its effects: critical assessment review papers, volume I, atmospheric sciences. Washington,
45 DC: U.S. Environmental Protection Agency, Office of Research and Development; pp. 7-1 to 7-70; EPA
46 report no. EPA/600/8-83/016AF. Available from: NTIS, Springfield, VA; PB85-100030.
47
48 Hicks, B. B.; Meyers, T. P. (1989) Atmosphere-surface exchange processes. Presented at: 82nd annual meeting
49 of the Air & Waste Management Association; June; Anaheim, CA. Pittsburgh, PA: Air & Waste
50 Management Association; paper no. 89-113.4.
51
52 Hicks, B. B.; Wesely, M. L.; Lindberg, S. E.; Bromberg, S. M., eds. (1986) Proceedings of the NAPAP
53 workshop on dry deposition; March; Harpers Ferry, WV. Oak Ridge, TN: National Oceanic and
54 Atmospheric Administration.
April 1995 3.190 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hicks, B. B.; Draxler, R. R.; Dodge, M.; Hales, J. M.; Albritton, D. L.; Schwartz. S. E.; Meyers, T. P.;
2 Fehsenfeld, F. C.; Tanner, R. L.; Vong, R. J. (1991) Atmospheric processes research and process model
3 development. In: Irving, P. M., ed. Acidic deposition: state of science and technology, volume I:
4 emissions, atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation
5 Assessment Program. (State of science and technology report no. 2).
6
7 Hidy, G. M.; Appel, B. R.; Charlson, R. J.; Clark, W. E.; Friedlander, S. K.; Hutchison, D. H.; Smith,
8 T. B.; Suder, J.; Wesolowski, J. J.; Whitby, K. T. (1975) Summary of the California Aerosol
9 Characterization Experiment. J. Air Pollut. Control Assoc. 25: 1106-1114.
10
11 Hildemann, L. M. (1990) A study of the origin of atmospheric organic aerosols [Ph.D. thesis]. Pasadena, CA:
12 California Institute of Technology.
13
14 Hildemann, L. M.; Russell, A. G.; Cass, G. R. (1984) Ammonia and nitric acid concentrations in equilibrium
15 with atmospheric aerosols: experiment vs theory, Atmos. Environ. 18: 1737-1750.
16
17 Hildemann, L. M.; Cass, G. R.; Mazurek, M. A.; Simoneit, B. R. T. (1993) Mathematical modeling of urban
18 organic aerosol: properties measured by high-resolution gas chromatography. Environ. Sci. Technol.
19 27: 2045-2055.
20
21 Hildemann, L. M.; Klinedinst, D. B.; Klouda, G. A.; Currie, L. A.; Cass, G. R. (1994) Sources of urban
22 contemporary carbon aerosol. Environ. Sci. Technol. 28: 1565-1576.
23
24 Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Seasonal trends in Los Angeles
25 ambient organic aerosol observed by high-resolution gas chromatography. Aerosol Sci. Technol.
26 20: 303-317.
27
28 Hill, T. A.; Choularton, T. W.; Penkett, S. A. (1986) A model of sulphate production in a cap cloud and
29 subsequent turbulent deposition onto the hill surface. Atmos. Environ. 20: 1763-1771.
30
31 Hinds, W. C. (1982) Aerosol technology. New York, NY: John Wiley and Sons.
32
33 Hoffmann, M. R.; Boyce, S. D. (1983) Catalytic autooxidation of aqueous sulfur dioxide in relationship to
34 atmospheric systems. Adv. Environ. Sci. Technol. 12: 148-149.
35
36 Hoffmann, M. R.; Calvert, J. G. (1985) Chemical transformation modules for Eulerian acid deposition models.
37 In: Lee, Y.-N., ed. Chemical transformations in acid rain: volume II. investigation of kinetics and
38 mechanism of aqueous-phase peroxide formation. Research Triangle Park, NC: U.S. Environmental
39 Protection Agency, Atmospheric Sciences Research Laboratory; EPA report no. EPA/600/3-85/017.
40 Available from: NTIS, Springfield, VA; PB85-173433.
41
42 Hoffmann, M. R.; Edwards, J. O. (1975) Kinetics of the oxidation of sulfite by hydrogen peroxide in acidic
43 solution. J. Phys. Chem. 79: 2096-2098.
44
45 Hoffmann, M. R.; Jacob, D. J. (1984) Kinetics and mechanisms of the catalytic oxidation of dissolved sulfur
46 dioxide in aqueous solution: an application to nighttime fog water chemistry. In: Calvert, J. G., ed. SO2,
47 NO and NO2 oxidation mechanisms: atmospheric considerations. Boston, MA: Butterworth Publishers;
48 pp. 101-172. (Teasley, J. I., ed. Acid precipitation series: v. 3).
49
50 Hoffmann, H. E.; Kuehnemann, W. (1979) Comparison of the results of two measuring methods determining
51 the horizontal standard visibility with the visual visibility range. Atmos. Environ. 13: 1629-1634
52
53 Holsen, T. M.; Noll, K. E. (1992) Dry deposition of atmospheric particles: application of current models to
54 ambient data. Environ. Sci. Technol. 26: 1807-1815.
April 1995 3_191 DRAFT-DO NOT QUOTE OR CITE
-------
1 Holsen, T. M.; Noll, K. E.; Fang, G.-C.; Lee, W.-J.; Lin, J.-M.; Keeler, G. J. (1993) Dry deposition and
2 particle size distributions measured during the Lake Michigan Urban Air Toxics Study. Environ. Sci.
3 Techno!. 27: 1327-1333.
4
5 Hong, M.-S.; Carmichael, G. R. (1986) An investigation of sulfate production in an orographic storm using a
6 detailed transport/chemistry model coupled with a detailed cloud scavenging model. Atmos. Environ.
7 20: 1989-1997.
8
9 Hopke, P. K. (1985) Receptor modeling in environmental chemistry. New York, NY: John Wiley and Sons.
10
11 Hopke, P. K., ed. (1991) Receptor modeling for air quality management. Amsterdam, The Netherlands:
12 Elsevier Publishers.
13
14 Hoppel, W. A. (1988) The role of non-precipitating cloud cycles and gas-to-particle conversion in the
15 maintainence of the submicron aerosol size distribution over the tropical oceans. In: Hobbs, P. V.;
16 McCormick, M. P., eds. Aerosols and climate. Hampton, VA: A. Deepak Publishers; pp. 9-19.
17
18 Hoppel, W. A.; Frick, G. M. (1990) Submicron aerosol size distributions measured over the tropical and South
19 Pacific. Atmos. Environ. Part A 24: 645-659.
20
21 Hoppel, W. A.; Frick, G. M.; Larson, R. E. (1986) Effect of nonprecipitating clouds on the aerosol size
22 distribution in the marine boundary layer. Geophys. Res. Lett. 13: 125-128.
23
24 Horvath, H.; Noll, K. E. (1969) The relationship between atmospheric light scattering coefficient and visibility.
25 Atmos. Environ. 3: 543-550.
26
27 Hosker, R. P., Jr.; Lindberg, S. E. (1982) Review: atmospheric deposition and plant assimilation of gases and
28 particles. Atmos. Environ. 16: 889-910.
29
30 Hov, O.; Isaksen, I. S. A. (1981) Generation of secondary pollutants in a power plant plume: a model study.
31 Atmos. Environ. 15: 2367-2376.
32
33 Huang, P.-F.; Turpin, B. J.; Pipho, M. J.; Kittelson, D. B.; McMurry, P. H. (1994) Effects of water
34 condensation and evaporation on diesel chain-agglomerate morphology. J. Aerosol Sci. 25: 447-459.
35
36 Huie, R. E.; Neta, P. (1987) Rate constants for some oxidations of S(IV) by radicals in aqueous solutions.
37 Atmos. Environ. 21: 1743-1747.
38
39 Huntzicker, J. J.; Hoffman, R. S.; Ling, C.-S. (1978) Continuous measurement and speciation of
40 sulfur-containing aerosols by flame photometry. Atmos. Environ. 12: 83-88.
41
42 Huntzicker, J. J.; Johnson, R. L.; Shah, J. J.; Gary, R. A. (1982) Analysis of organic and elemental carbon in
43 ambient aerosols by a thermal-optical method. In: Wolff, G. T.; Klimisch, R. L., eds. Particulate
44 carbon—atmospheric life cycle: proceedings of an international symposium; October 1980; Warren, MI.
45 New York, NY: Plenum Press; pp. 79-88.
46
47 Husain, L.; Dutkiewicz, V. A.; Hussain, M. M.; Khwaja, H. A.; Burkhard, E. G.; Mehmood, G.; Parekh,
48 P. P.; Canelli, E. (1991) A study of heterogeneous oxidation of SO2 in summer clouds. J. Geophys. Res.
49 [Atmos.] 96: 18789-18805.
50
51 Husar, R. B.; Gillani, N. V.; Husar, J. D. (1976) Particulate sulfur formation in power plant: urban and
52 regional plumes. Presented at: Symposium on aerosol science and technology at the 82nd national meeting
53 of the AIChE; August-September; Atlantic City, MD.
54
April 1995 3-192 DRAFT-DO NOT QUOTE OR CITE
-------
1 Husar, R. B.; Patterson, D. E.; Husar, J. D.; Gillani, N. V.; Wilson, W. E., Jr. (1978) Sulfur budget of a
2 power plant plume. In: Husar, R. B.; Lodge, J. P., Jr.; Moore, D. J., eds. Sulfur in the atmosphere:
3 proceedings of the international symposium; September 1977; Dubrovnik, Yugoslavia. Atmos. Environ.
4 12: 549-568.
5
6 Husar, R. B.; Holloway, J. M.; Patterson, D. E.; Wilson, W. E. (1981) Spatial and temporal pattern of eastern
7 U.S. haziness: a summary. Atmos. Environ. 15: 1919-1928.
8
9 Huss, A., Jr.; Lim, P. K.; Eckert, C. A. (1982a) Oxidation of aqueous sulfur dioxide. 1. Homogeneous
10 manganese(II) and iron(II) catalysis at low pH. J. Phys. Chem. 86: 4224-4228.
11
12 Huss, A., Jr.; Lim, P. K.; Eckert, C. A. (1982b) Oxidation of aqueous sulfur dioxide. 2. High-pressure studies
13 and proposed reaction mechanisms. J. Phys. Chem. 86: 4229-4233.
14
15 Ibusuki, T.; Takeuchi, K. (1987) Sulfur dioxide oxidation by oxygen catalyzed by mixtures of manganese(II)
16 and iron(III) in aqueous solutions at environmental reaction conditions. Atmos. Environ. 21: 1555-1560.
17
18 Ip, W. M.; Gordon, R. J.; Ellis, E. C. (1984) Characterization of organics in aerosol samples from a
19 Los angeles receptor site using extraction and liquid chromatography methodology. Sci. Total Environ.
20 36: 203-208.
21
22 Irwin, J. S. (1983) Estimating plume dispersion: a comparison of several sigma schemes. J. Clim. Appl.
23 Meteorol. 22: 92-114.
24
25 Irving, P. M., ed. (1991) Acidic deposition: state of science and technology, volume 1: emissions, atmospheric
26 processes, and deposition. Washington, DC: The U.S. National Acid Precipitation Asssessment Program.
27
28 Izumi, K.; Fukuyama, T. (1990) Photochemical aerosol formation from aromatic hydrocarbons in the presence
29 of NOX. Atmos. Environ. Part A 24: 1433-1441.
30
31 Izumi, K.; Murano, K.; Mizuochi, M.; Fukuyama, T. (1988) Aerosol formation by the photooxidation of
32 cyclohexene in the presence of nitrogen oxides. Environ. Sci. Technol. 22: 1207-1215.
33
34 Jackson, M. L.; Gillette, D. A.; Danielsen, E. F.; Blifford, I. H., Jr.; Bryson, R.; Syers, T. K. (1973) Global
35 dustfall during the quaternary as related to environments. Soil Sci. INQUA Issue 166: 135-145.
36
37 Jacob, D. J. (1986) Chemistry of OH in remote clouds and its role in the production of formic acid and
38 peroxymonosulfate. J. Geophys. Res. [Atmos.] 91: 9807-9826.
39
40 Jacob, D. J.; Hoffmann, M. R. (1983) A dynamic model for the production of H+, NO3 , and SO42" in urban
41 fog. J. Geophys. Res. 88: 6611-6621.
42
43 Jacob, D. J.; Waldman, J. M.; Munger, J. W.; Hoffmann, M. R. (1985) Chemical composition of fogwater
44 collected along the California coast. Environ. Sci. Technol. 19: 730-736.
45
46 Jacob, D. J.; Waldman, J. M.; Munger, J. M.; Hoffmann, M. R. (1986a) The H2SO4-HNO3-NH3 system at
47 high humidities and in fogs, I: spatial and temporal patterns in the San Joaquin Valley of California.
48 J. Geophys. Res. [Atmos.] 91: 1073-1088.
49
50 Jacob, D. J.; Munger, J. W.; Waldman, J. M.; Hoffman, M. R. (1986b) The H2SO4-HN03-NH3 system at
51 high humidities and in fogs: II. comparison of field data with the thermodynamic calculations. J. Geophys
52 Res. [Atmos.] 91: 1089-1096.
53
April 1995 3.193 DRAFT-DO NOT QUOTE OR CITE
-------
1 Jaffrezo, J.-L.; Colin, J.-L. (1988) Rain-aerosol coupling in urban area: scavenging ratio measurement and
2 identification of some transfer processes. Atmos. Environ. 22: 929-935.
3
4 Japar, S. M.; Brachaczek, W. W.; Gorse, R. A., Jr.; Norbeck, J. M.; Pierson, W. R. (1986) The contribution
5 of elemental carbon to the optical properties of rural atmospheric aerosols. Atmos. Environ. 20: 1281-1289.
6
7 Javitz, H. S.; Watson, J. G.; Guertin, J. P.; Mueller, P. K. (1988) Results of a receptor modeling feasibility
8 study. JAPCA38: 661.
9
10 Jensen, J. B.; Charlson, R. J. (1984) On the efficiency of nucleation scavenging. Tellus Ser. B 36: 367-375.
11
12 Jessium, J. A.; Lu, H. P.; Chughtai, A. R.; Novakov, T. (1986) Appl. Spectrosc. 40: 113-119.
13
14 John, W.; Sethi, V. (1993) Threshold for resuspension by particle impaction. Aerosol Sci. Technol. 19: 69-79.
15
16 John, W.; Wang, H.-C. (1991) Laboratory testing method for PM-10 samplers: lowered effectiveness from
17 particle loading. Aerosol Sci. Technol. 14: 93-101.
18
19 John, W.; Wall, S. M.; Ondo, J. L.; Winklmayr, W. (1990) Modes in the size distributions of atmospheric
20 inorganic aerosol. Atmos. Environ. Part A 24: 2349-2359.
21
22 John, W.; Winklmayr, W.; Wang, H.-C. (1991) Particle deagglomeration and reentrainment in a PM-10
23 sampler. Aerosol Sci. Technol. 14: 165-176.
24
25 Johnson, C. A.; Sigg, L.; Zobrist, J. (1987) Case studies on the chemical composition of fogwater: the
26 influence of local gaseous emissions. Atmos. Environ. 21: 2365-2374.
27
28 Joos, F.; Baltensperger, U. (1991) A field study on chemistry, S(IV) oxidation rates and vertical transport
29 during fog conditions. Atmos. Environ. Part A 25: 217-230.
30
31 Junge, C. E. (1963) Air chemistry and radioactivity. New York, NY: Academic Press. (Van Mieghem, J.;
32 Hales, A. L., eds. International geophysics series: v. 4).
33
34 Junge, C. E. (1977) Basic considerations about trace constituents in the atmosphere as related to the fate of
35 global pollutants. In: Suffet, I. H., ed. Fate of pollutants in the air and water environments: part I,
36 mechanism of interaction between environments and mathematical modeling and the physical fate of
37 pollutants, papers from the 165th national American Chemical Society meeting; April 1975; Philadelphia,
38 PA. New York, NY: John Wiley & Sons; pp. 7-25. (Advances in environmental science and technology:
39 v. 8).
40
41 Junge, C.; McLaren, E. (1971) Relationship of cloud nuclei spectra to aerosol size distribution and composition.
42 J. Atmos. Sci. 28: 382-390.
43
44 Kadowaki, S. (1977) Size distribution and chemical composition of atmospheric paniculate nitrate in the Nagoya
45 area. Atmos. Environ. 11: 671-675.
46
47 Kadowaki, S. (1990) Characterization of carbonaceous aerosols in the Nagoya urban area. 1. Elemental and
48 organic carbon concentrations and the origin of organic aerosols. Environ. Sci. Technol. 24: 741-744.
49
50 Kadowaki, S. (1994) Characterization of carbonaceous aerosols in the Nagoya urban area. 2. Behavior and
51 origin of paniculate «-alkanes. Environ. Sci. Technol. 28: 129-135.
52
53 Kahl, J. D.; Samson, P. J. (1986) Uncertainty in trajectory calculations due to low-resolution meteorological
54 data. J. Clim. Appl. Meteorol. 25: 1816-1831.
April 1995 3-194 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kahl, J. D.; Samson, P. J. (1988) Trajectory sensitivity to rawinsonde data resolution. Atmos. Environ.
2 22: 1291-1299.
3
4 Kahl, J. D.; Schnell, R. C.; Sheridan, P. J.; Zak, B. D.; Church, H. W.; Mason, A. S.; Heffter, J. L.;
5 Harris, J. M. (1991) Predicting atmospheric debris transport in real-time using a trajectory forecast model.
6 Atmos. Environ. Part A 25: 1705-1713.
7
8 Kamens, R. M.; Jeffries, H. E.; Gery, M. W.; Wiener, R. W.; Sexton, K. G.; Howe, G. B. (1981) The
9 impact of *-pinene on urban smog formation: an outdoor smog chamber study. Atmos. Environ.
10 15: 969-981.
11
12 Kamens, R. M.; Gery, M. W.; Jeffries, H. E.; Jackson, M.; Cole, E. I. (1982) Ozone-isoprene reactions:
13 product formation and aerosol potential. Int. J. Chem. Kinet. 14: 955-975.
14
15 Kamens, R.; Odum, J.; Fan, Z.-H. (1995) Some observations on times to equilibrium for semivolatile
16 polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 29: 43-50.
17
18 Kaplan, C. R.; Gentry, J. W. (1988) Agglomeration of chain-like combustion aerosols due to Brownian motion.
19 Aerosol Sci. Technol. 8: 11-28.
20
21 Kaplan, I. R.; Gordon, R. J. (1994) Non-fossil-fuel fine-particle organic carbon aerosols in Southern California
22 determined during the Los Angeles Aerosol Characterization and Source Apportionment Study. Aerosol Sci.
23 Technol. 21: 343-359.
24
25 Karamchandani, P.; Venkatram, A. (1992) The role of non-precipitating clouds in producing ambient sulfate
26 during summer: results from simulations with the Acid Deposition and Oxidant Model (ADOM) Atmos.
27 Environ. Part A 26: 1041-1052.
28
29 Katrinak, K. A.; Rez, P.; Buseck, P. R. (1992) Structural variations in individual carbonaceous particles from
30 an urban aerosol. Environ. Sci. Technol. 26: 1967-1976.
31
32 Katrinak, K. A.; Rez, P.; Perkes, P. R.; Buseck, P. R. (1993) Fractal geometry of carbonaceous aggregates
33 from an urban aerosol. Environ. Sci. Technol. 27: 539-547.
34
35 Kawamura, K.; Ng, L.-L.; Kaplan, I. R. (1985) Determination of organic acids (CrC10) in the atmosphere,
36 motor exhausts, and engine oils. Environ. Sci. Technol. 19: 1082-1086.
37
38 Keeler, G. J.; Japar, S. M.; Brachaczek, W. W.; Gorse, R. A., Jr.; Norbeck, J. M.; Pierson, W. R. (1990)
39 The sources of aerosol elemental carbon at Allegheny Mountain. Atmos. Environ. Part A 24: 2795-2805.
40
41 Kelly, N. A. (1987) The photochemical formation and fate of nitric acid in the metropolitan Detroit area:
42 ambient, captive-air irradiation and modeling results. Atmos. Environ. 21: 2163-2177.
43
44 Kelly, T. J.; Schwartz, S. E.; Daum, P. H. (1989) Detectability of acid producing reactions in natural clouds.
45 Atmos. Environ. 23: 569-583.
46
47 Kerminen, V. M.; Wexler, A. S. (1994) Particle formation due to SO2 oxidation and high relative humidity in
48 the remote marine boundary layer. In press.
49
50 Kerminen, V.; Wexler, A. S. (1994) Post-fog nucleation of H2SO4-HO2 particles in smog. Atmos. Environ.
51 28: 2399-2406.
52
April 1995 3_195 DRAFT-DO NOT QUOTE OR CITE
-------
1 Khlystov, A.; ten Brink, H. M.; Wyers, G. P. (1993) Hygroscopic growth rates of aerosols at high relative
2 humidity. Petten, The Netherlands: Netherlands Energy Research Foundation ECN; report no.
3 ECN-C-93-011.
4
5 Kim, Y. J. (1995) Response of the Active Scattering Aerosol Spectrometer Probe (ASASP-IOOX) to particles of
6 different chemical composition. Aerosol Sci. Technol. 22: 33-42.
7
8 Kim, D.; Hopke, P. K. (1988) Classification of individual particles based on computer-controlled scanning
9 electron microscopy data. Aerosol Sci. Technol. 9: 133-151.
10
11 Kim, M. G.; Yagawa, K.; Inoue, H.; Lee, Y. K.; Shirai, T. (1990) Measurement of tire tread in urban air by
12 pyrolysis-gas chromatography with flame photometric detection. Atmos. Environ. Part A 24: 1417-1422.
13
14 Kim, Y. P.; Seinfeld, J. H.; Saxena, P. (1993a) Atmospheric gas-aerosol, equilibrium: I. thermodynamic
15 model. Aerosol Sci. Technol. 19: 157-181.
16
17 Kim, Y. P.; Seinfeld, J. H.; Saxena, P. (1993b) Atmospheric gas-aerosol equilibrium: II. analysis of common
18 approximations and activity coefficient calculation methods. Aerosol Sci. Technol. 19: 182-198.
19
20 Kim, Y. J.; Boatman, J. F.; Gunter, R. L.; Wellman, D. L.; Wilkison, S. W. (1993c) Vertical distribution of
21 atmospheric aerosol size distribution over south-central New Mexico. Atmos. Environ. Part A 27:
22 1351-1362.
23
24 Kirchner, W.; Welter, F.; Bongartz, A.; Kames, J.; Schweighoefer, S.; Schurath, U. (1990) Trace gas
25 exchange at the air/water interface: measurements of mass accomodation coefficients. J. Atmos. Chem.
26 10: 427-449.
27
28 Kitsa, V.; Lioy, P. J. (1992) Near field dispersion of mechanically resuspended dust from an unpaved road.
29 Transactions AWMA 1: 199-210.
30
31 Kitsa, V.; Lioy, P. J.; Chow, J. C.; Watson, J. G.; Shupack, S.; Howell, T.; Sanders, P. (1992) Particle-size
32 distribution of chromium: total and hexavalent chromium in inspirable, thoracic, and respirable soil particles
33 from contaminated sites in New Jersey. Aerosol Sci. Technol. 17: 213-229.
34
35 Klee, A. J. (1984) Source control: municipal solid waste incinerators. In: Calvert, J. G.; Englund, eds.
36 Handbook of air pollution technology. New York, NY: John Wiley and Sons; pp. 513-550.
37
38 Kleinman, M. T.; Pasternack, B. S.; Eisenbud, M.; Kneip, T. J. (1980) Identifying and estimating the relative
39 importance of sources of airborne particulates. Environ. Sci. Technol. 14: 62-65.
40
41 Klouda, G. A.; Currie, L. A.; Verkouteren, R. M.; Eifield, W.; Zak, B. D. (1988) Advances in
42 microradiocarbon dating and the direct tracing of environmental carbon. J. Radioanal. Nucl. Chem.
43 123: 191-197.
44
45 Knapp, K. T.; Bennett, R. L. (1990) Procedures for chemical characterization of sized particles in stationary
46 source emissions. Aerosol Sci. Technol. 12: 1067-1074.
47
48 Koenig, L. R.; Murray, F. W. (1976) Ice-bearing cumulus cloud evolution, numerical simulation and general
49 comparison against observations. J. Appl. Meteorol. 15: 747-766.
50
51 Kohler, H. (1936) The nucleus in, and the growth of, hygroscopic droplets. Trans. Faraday Soc.
52 32: 1152-1161.
53
April 1995 3-196 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kolb, C. E.; Jayne, J. T.; Worsnop, D. R.; Molina, M. J.; Meads, R. F.; Viggiano, A. A. (1994) The gas
2 phase reaction of sulfur trioxide with water vapor. J. Am. Chem. Soc. 116: 10314-10321.
3
4 Kopcewicz, B.; Nagamoto, C.; Parungo, F.; Harris, J.; Miller, J.; Sievering, H.; Rosinski, J. (1991)
5 Morphological studies of sulfate and nitrate particles on the east coast of North America and over the North
6 Atlantic Ocean. Atmos. Res. 26: 245-271.
7
8 Korzun, E. A.; Heck, H. H. (1990) Sources and fates of lead and cadmium in municipal solid waste. J. Air
9 Waste Manage. Assoc. 40: 1220-1226.
10
11 Koutrakis, P.; Wolfson, J. M.; Slater, J. L.; Brauer, M.; Spengler, J. D.; Stevens, R. K.; Stone, C. L. (1988)
12 Evaluation of an annular denuder/filter pack system to collect acidic aerosols and gases. Environ. Sci.
13 Technol. 22: 1463-1468.
14
15 Koutrakis, P.; Wolfson, J. M.; Spengler, J. D.; Stern, B.; Franklin, C. A. (1989) Equilibrium size of
16 atmospheric aerosol sulfates as a function of the relative humidity. J. Geophys. Res. [Atmos.]
17 94: 6442-6448.
18
19 Kowalczyk, G. S.; Gratt, L. B.; Ricci, P. F. (1987) An air emission risk assessment for benzo(a)pyrene and
20 arsenic from the Mt. Tom Power Plant. JAPCA 37: 361-369.
21
22 Krieger, M. S.; Kites, R. A. (1992) Diffusion denuder for the collection of semivolatile organic compounds.
23 Environ. Sci. Technol. 26: 1551-1555.
24
25 Krieger, M. S.; Kites, R. A. (1994) Measurement of polychlorinated biphenyls and polycyclic aromatic
26 hydrocarbons in air with a diffusion denuder. Environ. Sci. Technol. 28: 1129-1133.
27
28 Kunen, S. M.; Lazrus, A. L.; Kok, G. L.; Heikes, B. G. (1983) Aqueous oxidation of SC^ by hydrogen
29 peroxide. J. Geophys. Res. C: Oceans Atmos. 88: 3671-3674.
30
31 Kuo, Y.-H.; Skumanich, M.; Haagenson, P. L.; Chang, J. S. (1985) The accuracy of trajectory models as
32 revealed by the observing system simulation experiments. Mon. Weather Rev. 113: 1852-1867.
33
34 Lamb, B.; Guenther, A.; Gay, D.; Westberg, H. (1987) A national inventory of biogenic hydrocarbon
35 emissions. Atmos. Environ. 21: 1695-1705.
36
37 Lane, D. A.; Johnson, N. D.; Barton, S. C.; Thomas, G. H. S.; Schroeder, W. H. (1988) Development and
38 evaluation of a novel gas and particle sampler for semivolatile chlorinated organic compounds in ambient
39 air. Environ. Sci. Technol. 22: 941-947.
40
41 Langner, J.; Rodhe, H. (1991) A global three-dimensional model of the tropospheric sulfur cycle. J. Atmos.
42 Sci. 13: 225-263.
43
44 Larson, S. M.; Cass, G. R. (1989) Characteristics of summer midday low-visibility events in the Los Angeles
45 area. Environ. Sci. Technol. 23: 281-289.
46
47 Larson, S. M.; Cass, G. R.; Gray, H. A. (1989) Atmospheric carbon particles and the Los Angeles visibility
48 problem. Aerosol Sci. Technol. 10: 118-130.
49
50 Lazrus, A. L.; Haagenson, P. L.; Kok, G. L.; Huebert, B. J.; Kreitzberg, C. W.; Likens, G. E.; Mohnen,
51 V. A.; Wilson, W. E.; Winchester, J. W. (1983) Acidity in air and water in a case of warm frontal
52 precipitation. Atmos. Environ. 17: 581-591.
53
AP»1 1995 3.!97 DRAFT-DO NOT QUOTE OR CITE
-------
1 Leaitch, W. R.; Strapp, J. W.; Wiebe, H. A.; Isaac, G. A. (1983) Measurements of scavenging and
2 transformation of aerosol inside cumulus. In: Pruppacher, H. R.; Seronin, R. G.; Slinn, W. G. N., eds.
3 Precipitation scavenging, dry deposition and resuspension, v. 1. New York, NY: Elsevier; pp. 53-69.
4
5 Leaitch, W. R.; Bottenheim, J. W.; Strapp, J. W. (1988) Possible contribution of N2O5 scavenging to HNO3
6 observed in winter stratiform cloud. J. Geophys. Res. 93: 12569-12584.
7
8 Leaitch, W. R.; Strapp, J. W.; Wiebe, H. A.; Anlauf, K. G.; Isaac, G. A. (1986a) Chemical and
9 microphysical studies of non-precipitating summer cloud. J. Geophys. Res. [Atmos.] 91: 11821-11831.
10
11 Leaitch, W. R.; Strapp, J. W.; Hudson, J. G.; Isaac, G. A. (1986b) Cloud droplet nucleation and cloud
12 scavenging of aerosol sulfate in polluted atmospheres. Tellus Ser. B 38: 328-344.
13
14 Leaitch, W. R.; Isaac, G. A.; Strapp, J. W.; Banic, C. M.; Wiebe, H. A. (1992) The relationship between
15 cloud droplet number concentrations and anthropogenic pollution: observations and climatic implications.
16 J. Geophys. Res. [Atmos.] 97: 2463-2474.
17
18 Lee, Y. N.; Schwartz, S. E. (1983) Kinetics of oxidation of aqueous sulfur(IV) by nitrogen dioxide.
19 In: Pruppacher, H. R.; Semonin, R. G.; Slinn, W. G. N., eds. Precipitation scavenging, dry deposition and
20 resuspension: v. 1. New York, NY: Elsevier.
21
22 Lee, Y.-N.; Shen, J.; Klotz, P. J.; Schwartz, S. E.; Newman, L. (1986) Kinetics of the hydrogen
23 peroxide-sulfur(IV) reaction in rainwater collected at a northeastern U.S. site. J. Geophys. Res. [Atmos.]
24 91: 13264-13274.
25
26 Lelieveld, J.; Crutzen, P. J. (1990) Influences of cloud photochemical processes on tropospheric ozone. Nature
27 (London) 343: 227-233.
28
29 Lelieveld, J.; Crutzen, P. J. (1991) The role of clouds in tropospheric photochemistry. J. Atmos. Chem.
30 12: 229-267.
31
32 Lelieveld, J.; Heintzenberg, J. (1992) Sulfate cooling effect on climate through in-cloud oxidation of
33 anthropogenic SO2. Science (Washington, DC) 258: 117-120.
34
35 Leone, J. A.; Flagan, R. C.; Grosjean, D.; Seinfeld, J. H. (1985) An outdoor smog chamber and modeling
36 study of toluene-NOx photooxidation. Int. J. Chem. Kinet. 17: 177-216.
37
38 Lesaffre, F. (1989) Characterization of aerosol aggregates through fractal parameters: effects due to humidity.
39 J. Aerosol. Sci. 20: 857-860.
40
41 Lewis, R. S.; Deen, W. M. (1994) Kinetics of the reaction of nitric oxide with oxygen in aqueous solutions.
42 Chem. Res. Toxicol. 7: 568-574.
43
44 Lewis, C. W.; Stevens, R. K. (1987) Hybrid receptor models. In: Johnson, R. W.; Gordon, G., eds. The
45 chemistry of acid rain: sources and atmospheric processes. Washington, DC: American Chemical Society;
46 pp. 58-65. (ACS symposium series 349).
47
48 Lewis, C. W.; Baumgardner, R. E.; Stevens, R, K.; Claxton, L. D.; Lewtas, J. (1988) Contribution of
49 woodsmoke and motor vehicle emissions to ambient aerosol mutagenicity. Environ. Sci. Technol.
50 22: 968-971.
51
52 Li, W.; Montassier, N.; Hopke, P. K. (1992) A system to measure the hygroscopicity of aerosol particles.
53 Aerosol Sci. Technol. 17: 25-35.
54
April 1995 3-198 DRAFT-DO NOT QUOTE OR CITE
-------
1 Li, S.; Anlauf, K. G.; Wiebe, H. A. (1993) Heterogeneous nighttime production and deposition of paniculate
2 nitrate at a rural site in North America during summer 1988. J. Geophys. Res. 98: 5139-5157.
3
4 Liebsch, E. J.; de Pena, R. G. (1982) Sulfate aerosol production in coal-fired power plant plumes. Atmos.
5 Environ. 16: 1323-1331.
6
7 Ligocki, M. P.; Pankow, J. F. (1989) Measurements of the gas/particle distributions of atmospheric organic
8 compounds. Environ. Sci. Technol. 23: 75-83.
9
10 Ligocki, M. P.; Salmon, L. G.; Fall, T.; Jones, M. C.; Nazaroff, W. W.; Cass, G. R. (1993) Characteristics
11 of airborne particles inside southern California museums. Atmos. Environ. Part A 27: 697-711.
12
13 Lin, X.; Chameides, W. L. (1991) Model studies of the impact of chemical inhomogeneity on SC«2 oxidation in
14 warm stratiform clouds. J. Atmos. Chem. 13: 109-129.
15
16 Lin, J.-M.; Fang, G.-C.; Holsen, T. M.; Noll, K. E. (1993) A comparison of dry deposition modeled from
17 size distribution data and measured with a smooth surface for total particle mass, lead and calcium in
18 Chicago. Atmos. Environ. Part A 27: 1131-1138.
19
20 Lin, J. J.; Noll, K. E.; Holsen, T. M. (1994) Dry deposition velocities as a function of particle size in the
21 ambient atmosphere. Aerosol Sci. Technol. 20: 239-252.
22
23 Lind, J. A.; Kok, G. L. (1986) Henry's law determinations for aqueous solutions of hydrogen peroxide,
24 methylhydroperoxide, and peroxyacetic acid. J. Geophys. Res. [Atmos.] 91: 7889-7895.
25
26 Lind, J. A.; Kok, G. L. (1994) Correction to "Henry's law determinations for aqueous solutions of hydrogen
27 peroxide, methylhydroperoxide, and peroxyacetic acid" by John A. Lind and Gregory L. Kok. J. Geophys.
28 Res. [Atmos.] 99: 21119.
29
30 Lind, J. A.; Lazrus, A. L. (1983) Aqueous-phase oxidation of sulfur(IV) by some organic peroxides. EOS
31 Trans. 64: 670.
32
33 Lindberg, S. E. (1982) Factors influencing trace metal, sulfate and hydrogen ion concentrations in rain. Atmos.
34 Environ. 16: 1701-1709.
35
36 Linton, R. W.; Loh, A.; Natusch, D. F. S.; Evans, C. A.; Williams, P. (1976) Surface predominance of trace
37 elements in airborne particles. Science (Washington, DC) 191: 852.
38
39 Lioy, P. J.; Daisey, J. M. (1986) Airborne toxic elements and organic substances. Environ. Sci. Technol.
40 20: 8-14.
41
42 Lioy, P. J.; Watson, J. G.; Spengler, J. D. (1980) APCA specialty conference workshop on baseline data for
43 inhalable paniculate matter. J. Air Pollut. Control Assoc. 30: 1126-1130.
44
45 Lioy, P. J.; Mallon, R. P.; Lippmann, M.; Kneip, T. J.; Samson, P. J. (1982) Factors affecting the variability
46 of summertime sulfate in a rural area using principal component analysis. J. Air Pollut. Control Assoc.
47 32: 1043-1047.
48
49 Lipfert, F. W. (1994) Filter artifacts associated with paniculate measurements: recent evidence and effects on
50 statistical relationships. Atmos. Environ. 28: 3233-3249.
51
52 Lippmann, M. (1980) Size distributions in urban aerosols. In: Kneip, T. J.; Lioy, P. J., eds. Aerosols:
53 anthropogenic and natural, sources and transport. Ann. N. Y. Acad. Sci. 338: 1-12
54
April 1995 3499 DRAFT-DO NOT QUOTE OR CITE
-------
1 Liu, P. S. K.; Leaitch, W. R.; Strapp, J. W.; Wasey, M. A. (1992) Response of particle measuring systems
2 airborne to ASASP and PCASP to NaCl and latex particles. Aerosol Sci. Technol. 16: 83-95.
3
4 Liu, P. S. K.; Leaitch, W. R.; McDonald, A. M.; Isaac, G. A.; Strapp, J. W.; Wiebe, H. A. (1994) Sulfate
5 production in a summer cloud over Ontario, Canada. Tellus 45: 368-389.
6
7 Lowenthal, D. H.; Rogers, C. F.; Saxena, P.; Watson, J. G.; Chow, J. C. (1994) Sensitivity of estimated light
8 extinction coefficients to model assumptions and measurement errors. Atmos. Environ: accepted.
9
10 Lundgren, D. A.; Hausknecht, B. J. (1982) Ambient aerosol mass distribution of 1 to 100 /*m particles in five
11 cities. Presented at: 75th annual meeting, Air Pollution Control Association; June; New Orleans, LA.
12 Pittsburgh, PA: Air Pollution Control Association; paper no. 82-45.4.
13
14 Lundgren, D. A.; Paulus, J. J. (1975) The mass distribution of large atmospheric particles. J. Air Pollut.
15 Control Assoc. 25: 1227-1231.
16
17 Lundgren, D.; Burton, R. M. (1995) The effect of particle size distribution on the cut point between fine and
18 coarse ambient mass fractions. Inhalation Toxicol.: accepted.
19
20 Lundgren, D.; Haasknecht, B.; Burton, R. (1984) Large particle size distribution in five U.S. cities and the
21 effect on the new ambient paniculate matter standard (PM10). Aerosol Sci. Technol. 7: 467-473.
22
23 Lurmann, F. W.; Collins, J.; Coyner, L. (1988) Development of a chemical transformation submodel for
24 annual PM10 dispersion modeling. El Monte, CA: South Coast Air Quality Management District; final
25 report no. P-6190-003.
26
27 Lusis, M. A.; Anlauf, K. G.; Barrie, L. A.; Wiebe, H. A. (1978) Plume chemistry studies at a northern
28 Alberta power plant. Atmos. Environ. 12: 2429-2437.
29
30 Lyons, W. A.; Tremback, C. J.; Tesche, T. W. (1991) Lake Michigan ozone study prognostic modeling: model
31 performance evaluation and sensitivity testing. Crested Butte/ Fort Collins, CO: Alpine Geophysics,
32 ASTeR, Inc.
33
34 Maahs, H. G. (1983) Kinetics and mechanism of the oxidation of S(IV) by ozone in aqueous solution with
35 particular reference to SO2 conversion in nonurban tropospheric clouds. J. Geophys. Res. 88: 10721-10723.
36
37 Macias, E. S.; Hopke, P. K., eds. (1981) Atmospheric aerosol: source/air quality relationships. Washington,
38 DC: American Chemical Society. (ACS symposium series 167).
39
40 Macias, E. S.; Zwicker, J. O.; White, W. H. (1981) Regional haze case studies in the southwestern U.S.—II.
41 source contributions. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes and visibility:
42 measurements and model components: proceedings of the symposium; November 1980; Grand Canyon
43 National Park, AZ. Atmos. Environ. 15: 1987-1997.
44
45 Makhon'ko, K. P. (1986) Vetrovoy zakhvat pweel c podstilaioushey poberkhnosty, pokreetoy travot. Meteorol.
46 Gidrol. 10: 61-65.
47
48 Malm, W. C.; Gebhart, K. A.; Molenar, J.; Cahill, T.; Eldred, R.; Huffman, D. (1994) Examining the
49 relationship between atmospheric aerosols and light extinction at Mount Rainier and North Cascades
50 National Parks. Atmos. Environ. 28: 347-360.
51
52 Mamane, Y.; Noll, K. E. (1985) Characterization of large particles at a rural site in the eastern United States:
53 mass distribution and individual particle analysis. Atmos. Environ. 19: 611-622.
54
April 1995 3-200 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mamane, Y.; Miller, J. L.; Dzubay, T. G. (1986) Characterization of individual fly ash particles emitted from
2 coal- and oil-fired power plants. Atmos. Environ. 20: 2125-2135.
3
4 Mamane, Y.; Stevens, R. K.; Dzubay, T. G. (1990) On the sources of fine and coarse carbon particles in the
5 Great Lakes. J. Aerosol Sci. 21: S353-S356.
6
7 Mamane, Y.; Dzubay, T. G.; Ward, R. (1992) Sulfur enrichment of atmospheric minerals and spokes. Atmos.
8 Environ. Part A 26: 1113-1120.
9
10 Mansoori, B. A.; Johnston, M. V.; Wexler, A. S. (1994) Quantitation of ionic species in single microdroplets
11 by on-line laser desorption/ionization. Anal. Chem. 66: 3681-3687.
12
13 Marlia, J. C.; Kim, B. M.; Wu, R. (1990) Annual PM10 dispersion model application to the South Coast air
14 basin. El Monte, CA: South Coast Air Quality Management District; technical report V-D.
15
16 Marple, V. A.; Rubow, K. L.; Behm, S. M. (1991) A microorifice uniform deposit impactor (MOUDI):
17 description, calibration, and use. Aerosol Sci. Technol. 14: 434-446.
18
19 Martin, L. R. (1984) Kinetic studies of sulfite oxidation in aqueous solution. In: Calvert, J. G., ed. SO2,
20 NO and NO2 oxidation mechanisms: atmospheric considerations. Boston, MA: Butterworth Publishers; pp.
21 63-100. (Teasley, J. I., ed. Acid precipitation series: v. 3).
22
23 Martin, L. R.; Damschen, D. E. (1981) Aqueous oxidation of sulfur dioxide by hydrogen peroxide at low pH.
24 Atmos. Environ. 15: 1615-1621.
25
26 Martin, L. R.; Hill, M. W. (1987a) The iron catalyzed oxidation of sulfur: reconciliation of the literature rates.
27 Atmos. Environ. 21: 1487-1490.
28
29 Martin, L. R.; Hill, M. W. (1987b) The effect of ionic strength on the manganese catalyzed oxidation of
30 sulfur(IV). Atmos. Environ. 21: 2267-2270.
31
32 Martin, L. R.; Hill, M. W.; Tai, A. F.; Good, T. W. (1991) The iron catalyzed oxidation of sulfur(IV) in
33 aqueous solution: differing effects of organics at high and low pH. J. Geophys. Res. 96: 3085-3097.
34
35 Martonen, T. B.; Zhang, Z. (1993) Deposition of sulfate acid aerosols in the developing human lung. Inhalation
36 Toxicol. 5: 165-187.
37
38 Martonen, T. B.; Barnett, A. E.; Miller, F. J. (1985) Ambient sulfate aerosol deposition in man: modeling the
39 influence of hygroscopicity. Environ. Health Perspect. 63: 11-24.
40
41 Mason, A. S.; Gifford, F. A. (1992) Atmospheric tracer dispersion over a 24-h time span. Atmos. Environ.
42 Part A 26: 3203-3205.
43
44 Matsoukas, T.; Friedlander, S. K. (1991) Dynamics of aerosol agglomerate formation. J. Colloid Interface Sci.
45 146: 495-506.
46
47 Mazurek, M. A.; Simoneit, B. R. T. (1984) Characterization of biogenic and petroleum-derived organic matter
48 in aerosols over remote, rural and urban areas. In: Identification and analysis of organic pollutants in air:
49 an ACS symposium. Woburn, MA: Ann Arbor Science/Butterworth Publishers; pp. 353-370.
50
51 Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1989) Interpretation of high-resolution gas chromatography
52 / mass spectrometry data aquired from atmospheric organic aerosol samples. Aerosol Sci. Technol. 10:
53 408-420.
54
April 1995 3-201 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Biological input to visibility-reducing aerosol particles
2 in the remote arid southwestern United States. Environ. Sci. Technol. 25: 684-694.
3
4 McArdle, J. V.; Hoffmann, M. R. (1983) Kinetics and mechanism of the oxidation of aquated sulfur dioxide by
5 hydrogen peroxide at low pH. J. Phys. Chem. 87: 5425-5429.
6
7 McHenry, J. N.; Dennis, R. L. (1991) Partitioning of the sulfate budget into gas- and aqueous-phase
8 components in the regional acid deposition model (RADM). Presented at: Proceedings of the 7th joint
9 AMS-AWMA conference: applications of air pollution meteorology; New Orleans, LA. pp. 143-147.
10
11 McHenry, J. N.; Dennis, R. L. (1994) The relative importance of oxidation pathways and clouds to atmospheric
12 ambient sulfate production as predicted by the regional and deposition model. J. Appl. Meteorol.
13 33: 890-905.
14
15 McHenry, J. N.; Binkowski, F. S.; Dennis, R. L.; Chang, J. S.; Hopkins, D. (1992) The tagged species
16 engineering model (TSEM). Atmos. Environ. Part A 26: 1427-1443.
17
18 McMahon, T. A.; Denison, P. J. (1979) Empirical atmospheric deposition parameters—a survey. Atmos.
19 Environ. 13: 571-585.
20
21 McMurry, P. H.; Friedlander, S. K. (1979) New particle formation in the presence of an aerosol. Atmos.
22 Environ. 13: 1635-1651.
23
24 McMurry, P. H.; Grosjean, D. (1985) Photochemical formation of organic aerosols: growth laws and
25 mechanisms. Atmos. Environ. 19: 1445-1451.
26
27 McMurry, P. H.; Stolzenburg, M. R. (1989) On the sensitivity of particle size to relative humidity for Los
28 Angeles aerosols. Atmos. Environ. 23: 497-507.
29
30 McMurry, P. H.; Wilson, J. C. (1982) Growth laws for the formation of secondary ambient aerosols:
31 implications for chemical conversion mechanisms. Atmos. Environ. 16: 121-134.
32
33 McMurry, P. H.; Wilson, J. C. (1983) Droplet phase (heterogeneous) and gas phase (homogeneous)
34 contributions to secondary ambient aerosol formation as functions of relative humidity. J. Geophys. Res.
35 88: 5101-5108.
36
37 McMurry, P. H.; Zhang, X. Q. (1989) Size distributions of ambient organic and elemental carbon. Aerosol Sci.
38 Technol. 10: 430-437.
39
40 McMurry, P. H.; Rader, D. J.; Stith, J. L. (1981) Studies of aerosol formation in power plant plumes—I.
41 growth laws for secondary aerosols in power plant plumes: implications for chemical conversion
42 mechanisms. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes and visibility: measurements
43 and model components: proceedings of the symposium; November 1980; Grand Canyon National Park, AZ.
44 Atmos. Environ. 15: 2315-2327.
45
46 Meagher, J. F.; Stockburger, L.; Bailey, E. M.; Huff, O. (1978) The oxidation of sulfur dioxide to sulfate
47 aerosols in the plume of a coal-fired power plant. Atmos. Environ. 12: 2197-2203.
48
49 Meagher, J. F.; Stockburger, L., Ill; Bonanno, R. J.; Bailey, E. M.; Luria, M. (1981) Atmospheric oxidation
50 of flue gases from coal-fired power plants—a comparison between conventional and scrubbed plumes.
51 Atmos. Environ. 15: 749-762.
52
53 Meagher, J. F.; Olszyna, K. J.; Weatherford, F. P.; Mohnen, V. A. (1990) The availability of H2O2 and O3
54 for aqueous phase oxidation of SO2. The question of linearity. Atmos. Environ. Part A 24: 1825-1829.
April 1995 3-202 DRAFT-DO NOT QUOTE OR CITE
-------
1 Meng, Z.; Seinfeld, J. H. (1994) On the source of the submicrometer droplet mode of urban and regional
2 aerosols. Aerosol Sci. Technol. 20: 253-265.
3
4 Meng, Z.; Seinfeld, J. H.; Saxena, P.; Kim, Y. P. (1995) Contribution of water to paniculate mass in the
5 South Coast Air Basin. Aerosol Sci. Technol. 22: 111-123.
6
7 Michaud, D.; Baril, M.; Perrault, G. (1993) Characterization of airborne dust from cast iron foundries by
8 physico-chemical methods and multivariate statistical analyses. Air Waste 43: 729-735.
9
10 Middleton, P.; Brock, J. (1976) Simulation of aerosol kinetics. J. Colloid Interface Sci. 54: 249-264.
11
12 Midwest Research Institute. (1976) Section 11.2, fugitive dust sources: an AP-42 update of open source fugitive
13 dust emissions. Research Triangle Park, NC: U.S. Environmental Protection Agency; report no.
14 EPA-450/2-76-029.
15
16 Midwest Research Institute. (1983) Section 11.2, fugitive dust sources: an AP-42 update of open source fugitive
17 dust emissions. Research Triangle Park, NC: U.S. Environmental Protection Agency; report no.
18 EPA-450/4-836-010.
19
20 Milford, J. B.; Davidson, C. I. (1985) The sizes of paniculate trace elements in the atmosphere—a review.
21 J. Air Pollut. Control Assoc. 35: 1249-1260.
22
23 Milford, J. B.; Davidson, C. I. (1987) The sizes of paniculate sulfate and nitrate in the atmosphere—a review.
24 JAPCA 37: 125-134.
25
26 Miller, F. J.; Gardner, D. E.; Graham, J. A.; Lee, R. E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
27 considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc. 29: 610-615.
28
29 Mohnen, V. A.; Kedlacek, J. A. (1989) Cloud chemistry research at Whiteface Mountain. Tellus Ser.
30 B 41: 79-91.
31
32 Moran, M. D. (1992) Numerical modeling of mesoscale atmospheric dispersion [Ph.D. dissertation].
33 Ft. Collins, CO: Colorado State University, Department of Atmospheric Science.
34
35 Morandi, M. (1985) Development of source apportionment models for inhalable particluate matter and its
36 extractable organic fractions in urban areas of New Jersey [Ph.D. dissertation]. Syracuse, NY: New York
37 University.
38
39 Morandi, M. T.; Lioy, P. J.; Daisey, J. M. (1991) Comparison of two multivariate modeling approaches for
40 the source apportionment of inhalable particulate matter in Newark, NJ. Atmos. Environ. Part A
41 25: 927-937.
42
43 Mountain, R. D.; Mulholland, G. W.; Baum, H. (1986) Simulation of aerosol agglomeration in the free
44 molecular and continuum flow regimes. J. Colloid Interface Sci. 114: 67-81.
45
46 Moyers, J. L.; Ranweiler, L. E.; Hopf, S. B.; Korte, N. E. (1977) Evaluation of particulate trace species in
47 southwest desert atmosphere. Environ. Sci. Technol. 11: 789-795.
48
49 Mozurkewich, M.; Calvert, J. G. (1988) Reaction probability of N2O5 on aqueous aerosols. J. Geophys. Res.
50 [Atmos.] 93: 15889-15896.
51
52 Mozurkewich, M.; McMurry, P. H.; Gupta, A.; Calvert, P. G. (1987) Mass accomodation coefficient for HO2
53 radicals on aqueous particles. J. Geophys. Res. [Atmos.] 92: 4163-4170.
54
April 1995 3_203 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mueller, P. K. (1982) Atmospheric participate carbon observations in urban and rural areas of the United
2 States. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate carbon—atmospheric life cycle: proceedings of
3 an international symposium; October 1980; Warren, MI. New York, NY: Plenum Press.
4
5 Mulhbaier, J. L.; Williams, R. L. (1982) Fireplaces, furnaces, and vehicles as emission sources of paniculate
6 carbons. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate carbon: atmospheric life cycle. New York,
7 NY: Plenum Press; pp. 185-205.
8
9 Munger, J. W.; Jacob, D. J.; Waldman, J. M.; Hoffmann, M. R. (1983) Fogwater chemistry in an urban
10 atmosphere. J. Geophys. Res. C: Oceans Atmos. 88: 5109-5121.
11
12 Munger, J. W.; Jacob, D. J.; Hoffmann, M. R. (1984) The occurence of bisulfite-aldehyde addition products in
13 fog- and cloudwater.. J. Atmos. Chem. 1: 335.
14
15 Munger, J. W.; Tiller, C.; Hoffman, M. R. (1986) Identification of hydroxymethanesulfonate in fog water.
16 Science (Washington, DC) 231: 247-249.
17
18 Munger, J. W.; Collett, J., Jr.; Daube, B., Jr.; Hoffmann, M. R. (1990) Fogwater chemistry at Riverside,
19 California. Atmos. Environ. Part B 24: 185-205.
20
21 Mylonas, D. T.; Allen, D. T.; Ehrman, S. H.; Pratsinis, S. E. (1991) The sources and size distributions of
22 organonitrates in Los Angeles aerosol. Atmos. Environ. Part A 25: 2855-2861.
23
24 National Research Council. (1979) Airborne particles. Baltimore, MD: University Park Press.
25
26 National Research Council. (1991) Rethinking the ozone problem in urban and regional air pollution.
27 Washington, DC: National Academy Press.
28
29 Natusch, D. F. S.; Wallace, J. R.; Evans, C. A., Jr. (1974) Toxic trace elements: preferential concentration in
30 respirable particles. Science (Washington, DC) 183: 202-204.
31
32 Newman, L. (1981) Atmospheric oxidation of sulfur dioxide: a review as viewed from power plant and smelter
33 plume studies. Atmos. Environ. 15: 2231-2239.
34
35 Newman, L.; Forrest, J.; Manowitz, B. (1975a) The application of an isotopic ratio technique to a study of the
36 atmospheric oxidation of sulfur dioxide in the plume from an oil-fired power plant. Atmos. Environ.
37 9: 959-968.
38
39 Newman, L.; Forrest, J.; Manowitz, B. (1975b) The application of an isotopic ratio technique to a study of the
40 atmospheric oxidation of sulfur dioxide in the plume from a coal fired power plant. Atmos. Environ.
41 9: 969-977.
42
43 Nicholson, K. W. (1988) The dry deposition of small particles: a review of experimental measurements. Atmos.
44 Environ. 22: 2653-2666.
45
46 Noll, K. E.; Pilat, M. J. (1971) Size distribution of atmospheric giant particles. Atmos. Environ. 5: 527-540.
47
48 Noll, K. E.; Pontius, A.; Frey, R.; Gould, M. (1985) Comparison of atmospheric coarse particles at an urban
49 and non-urban site. Atmos. Environ. 19: 1931-1943.
50
51 Noll, K. E.; Yuen, P.-F.; Fang, K. Y.-P. (1990) Atmospheric coarse paniculate concentrations and dry
52 deposition fluxes for ten metals in two urban environments. Atmos. Environ. Part A 24: 903-908.
53
April 1995 3-204 DRAFT-DO NOT QUOTE OR CITE
-------
1 Noone, K. J. et al. (1992) Changes in aerosol size and phase distributions due to physical and chemical
2 processes in fog. Tellus Ser. B 44: 489-504.
3
4 Noone, K. J.; Ogren, J. A.; Hallberg, A.; Hansson, H. C.; Wiedensohler, A.; Swietlicki, E. (1992)
5 A statistical examination of the chemical differences between interstitial and scavenged aerosol. Tellus Ser.
6 B 44: 581-592.
7
8 Novakov, T. (1984) The role of soot and primary oxidants in the atmospheric chemistry. Sci. Total Environ.
9 36: 1-10.
10
11 Noxon, J. F. (1983) NO3 and NO2 in the mid-Pacific troposphere. J. Geophys. Res. C: Oceans Atmos.
12 88: 11017-11021.
13
14 Nriagu, J. O. (1990) Global metal pollution: poisoning the biosphere? Environment 32(7): 6-11, 28-33.
15
16 Nunes, T. V.; Pio, C. A. (1993) Carbonaceous aerosols in industrial and coastal atmospheres. Atmos. Environ.
17 Part A 27: 1339-1346.
18
19 O'Brien, R. J.; Crabtree, J. H.; Holmes, J. R.; Hoggan, M. C.; Bockian, A. H. (1975) Formation of
20 photochemical aerosol from hydrocarbons: atmospheric analysis. Environ. Sci. Technol. 9: 577-582.
21
22 O'Dell, R. A.; Taheri, M.; Kabel, R. L. (1977) A model for uptake of pollutants by vegetation. J. Air Pollut.
23 Control Assoc. 27: 1104-1109.
24
25 Ogren, J. A.; Charlson, R. J. (1992) Implications for models and measurements of chemical inhomogeneities
26 among cloud droplets. Tellus 44: 208-225.
27
28 Olmez, I.; Sheffield, A. E.; Gordon, G. E.; Houck, J. E.; Pritchett, L. C.; Cooper, J. A.; Dzubay, T. G.;
29 Bennett, R. L. (1988) Composition of particles from selected sources in Philadelphia for receptor modeling
30 applications. JAPCA 38: 1392-1402.
31
32 Olson, T. M.; Hoffman, M. R. (1989) Hydroxyalkylsulfonate formation: its role as a S(IV) reservoir in
33 atmospheric water droplets. Atmos. Environ. 23: 985-997.
34
35 Oppelt, E. T. (1987) Incineration of hazardous waste: a critical review. JAPCA 37: 558-586.
36
37 Otani, Y.; Wang, C. S. (1984) Growth and deposition of saline droplets covered with a monolayer of
38 surfactant. Aerosol Sci. Technol. 3: 155-166.
39
40 Pace, T. G., ed. (1986) Transactions, receptor methods for source apportionment: real world issues and
41 applications. Pittsburgh, PA: Air Pollution Control Association.
42
43 Pack, D. H.; Ferber, G. J.; Heffter, J. L.; Telegadas, K.; Angel 1, J. K.; Hoecker, W. H.; Machta, L. (1978)
44 Meteorology of long-range transport. Atmos. Environ. 12: 425-444.
45
46 Pandis, S. N.; Seinfeld, J. H. (1989a) Sensitivity analysis of a chemical mechanism for aqueous-phase
47 atmospheric chemistry. J. Geophys. Res. [Atmos.] 94: 1105-1126.
48
49 Pandis, S. N.; Seinfeld, J. H. (1989b) Mathmatical modeling of acid deposition due to radiation fog.
50 J. Geophys. Res. 94: 12911-12923.
51
52 Pandis, S. N.; Seinfeld, J. H.; Pilinis, C. (1990a) The smog-fog-smog cycle and acid deposition. J. Geophys.
53 Res. 95: 18489-18500.
54
April 1995 3-205 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pandis, S. N.; Seinfeld, J. H.; Pilinis, C. (1990b) Chemical composition differences in fog and cloud droplets
2 of different sizes. Atmos. Environ. Part A 24: 1957-1969.
3
4 Pandis, S. N.; Paulson, S. E.; Seinfeld, J. H.; Flagan, R. C. (1991) Aerosol formation in the photooxidation of
5 isoprene and /?-pinene. Atmos. Environ. Part A 25: 997-1008.
6
7 Pandis, S. N.; Harley, R. A.; Cass, G. R.; Seinfeld, J. H. (1992a) Secondary organic aerosol formation and
8 transport. Atmos. Environ. Part A 26: 2269-2282.
9
10 Pandis, S. N.; Seinfeld, J. H.; Pilinis, C. (1992b) Heterogeneous sulfate production in an urban fog. Atmos.
11 Environ. Part A 26: 2509-2522.
12
13 Pandis, S. N.; Wexler, A. S.; Seinfeld, J. H. (1993) Secondary organic aerosol formation and transport—II.
14 predicting the ambient secondary organic aerosol size distribution. Atmos. Environ. Part A 27: 2403-2416.
15
16 Pankow, J. F. (1987) Review and comparative analysis of the theories on partitioning between the gas and
17 aerosol paniculate phases in the atmosphere. Atmos. Environ. 21: 2275-2283.
18
19 Pankow, J. F. (1988) Gas phase retention volume behavior of organic compounds on the sorbent
20 poly(oxy-/n-terphenyl-2'-5'-ylene). Anal. Chem. 60: 950-958.
21
22 Pankow, J. F. (1989) Overview of the gas phase retention volume behavior of organic compounds on
23 polyurethane foam. Atmos. Environ. 23: 1107-1111.
24
25 Pankow, J. F. (1991) Common y-intercept and single compound regressions of gas-particle partitioning data
26 vs. l/T. Atmos. Environ. Part A: 25: 2229-2239.
27
28 Pankow, J. F. (1994a) An absorption model of gas/particle partitioning of organic compounds in the
29 atmosphere. Atmos. Environ. 28: 185-188.
30
31 Pankow, J. F. (1994b) An absorption model of the gas/aerosol partitioning involved in the formation of
32 secondary organic aerosol. Atmos. Environ. 28: 189-193.
33
34 Pankow, J. F.; Bidleman, T. F. (1991) Effects of temperature, TSP and per cent non-exchangeable material in
35 determining the gas-particle partitioning of organic compounds. Atmos. Environ. Part A 25: 2241-2249.
36
37 Pankow, J. F.; Bidleman, T. F. (1992) Interdependence of the slopes and intercepts from log-log correlations of
38 measured gas-particle partitioning and vapor pressure-I. Theory and analysis of available data. Atmos
39 Environ. Part A 26: 1071-1080.
40
41 Pankow, J. F.; Storey, J. M. E.; Yamasaki, H. (1993) Effects of relative humidity on gas/particle partitioning
42 of semivolatile organic compounds to urban paniculate matter. Environ. Sci. Technol. 27: 2220-2226.
43
44 Parrish, D. D.; Fahey, D. W.; Williams, E. J.; Liu, S. C.; Trainer, M.; Murphy, P. C.; Albritton, D. L.;
45 Fehsenfeld, F. C. (1986) Background ozone and anthropogenic ozone enhancement at Niwot Ridge,
46 Colorado. J. Atmos. Chem. 4: 63-80.
47
48 Pastuszka, J. S.; Kwapulinski, J. (1988) The change in mass size distribution of aerosol near dumps as a result
49 of resuspension of dust. Presented at: the 81st annual meeting of Air Pollution Control Association; June;
50 Dallas, TX.
51
52 Patterson, E. M.; Gillette, D. A. (1977a) Commonalities in measured size distributions for aerosols having a
53 soil-derived component. J. Geophys. Res. 82: 2074-2082.
54
April 1995 3-206 DRAFT-DO NOT QUOTE OR CITE
-------
1 Patterson, E. M.; Gillette, D. A. (1977b) Measurements of visibility vs. mass-concentration for airborne soil
2 particles. Atmos. Environ. 11: 193-196.
3
4 Patterson, R. K.; Wagman, J. (1977) Mass and composition of an urban aerosol as a function of particle size
5 for several visibility levels. J. Aerosol Sci. 8: 269-279.
6
7 Patterson, E. M.; Gillette, D. A.; Grams, G. W. (1976) On the relation between visibility and concentration for
8 airborne soil particles. J. Appl. Meteorol. 15: 470-478.
9
10 Patterson, E. M.; Gillette, D. A.; Stockton, B. H. (1977) Complex index of refraction between 300 and 700 nm
11 for Saharan aerosols. J. Geophys. Res. 82: 3153-3159.
12
13 Peirson, D. H.; Cawse, P. A.; Salmon, L.; Cambray, R. S. (1973) Trace elements in the atmospheric
14 environment. Nature (London) 241: 252-256.
15
16 Penkett, S. A. (1972) Oxidation of SO2 and other atmospheric gases by ozone in aqueous solution. Nature
17 (London) 240: 105-106.
18
19 Penkett, S. A.; Jones, B. M. R.; Brice, K. A.; Eggleton, A. E. J. (1979) The importance of atmospheric ozone
20 and hydrogen peroxide in oxidising sulphur dioxide in cloud and rainwater. Atmos. Environ. 13: 123-137.
21
22 Penner, J. E.; Eddleman, H.; Novakov, T. (1993) Towards the development of a global inventory for black
23 carbon emissions. Atmos. Environ. Part A 27: 1277-1295.
24
25 Perry, G. B.; Chai, H.; Dickey, D. W.; Jones, R. H.; Kinsman, R. A.; Morrill, C. G.; Spector, S. L.;
26 Weiser, P. C. (1983) Effects of paniculate air pollution on asthmatics. Am. J. Public Health 73: 50-56.
27
28 Peters, K.; Eiden, R. (1992) Modelling the dry deposition velocity of aerosol particles to a spruce forest.
29 Atmos. Environ. Part A 26: 2555-2564.
30
31 Peterson, J. T. (1970) Distribution of sulfur dioxide over metropolitan St. Louis, as described by empirical
32 eigenvectors, and its relation to meteorological parameters. Atmos. Environ. 4: 501-518.
33
34 Peterson, J.; Junge, C. (1971) Sources of paniculate matter in the atmosphere. In: Matthews, W.; Kellogg, W.;
35 Robinson, G. D., eds. Man's impact on the climate. Cambridge, MA: MIT Press.
36
37 Phalen, R., ed. (1985) Particle size-selective sampling in the workplace: report of the ACGIH technical
38 committee on air sampling procedures. Cincinnati, OH: American Conference of Governmental Industrial
39 Hygienists.
40
41 Pickle, T.; Allen, D. T.; Pratsinis, S. E. (1990) The sources and size distributions of aliphatic and carbonyl
42 carbon in Los Angeles aerosol. Atmos. Environ. Part A 24: 2221-2228.
43
44 Pielke, R. A. (1984) Mesoscale meteorological modeling. Orlando, FL: Academic Press.
45
46 Pielke, R. A. (1989) Status of subregional and mesoscale models, v. 2: mesoscale meteorological models in the
47 United States. Fort Collins, CO: Electric Power Research Institute; report no. EN-6649, v. 2.
48
49 Pierson, W. R.; Brachaczek, W. W. (1983) Paniculate matter associated with vehicles on the road. II. Aerosol
50 Sci. Technol. 2: 1-40.
51
52 Pierson, W. R.; Gertler, A. W.; Bradow, R. L. (1990) Comparison of the SCAQS tunnel study with other
53 on-road vehicle emission data. J. Air Waste Manage. Assoc. 40: 1495-1504.
54
April 1995 3_207 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pilinis, C. (1990) Derivation and numerical solution of the species mass distribution equations for
2 multicomponent paniculate systems. Atmos. Environ. Part A 24: 1923-1928.
3
4 Pilinis, C.; Seinfeld, J. H. (1987) Continued development of a general equilibrium model for inorganic
5 multicomponent atmospheric aerosols. Atmos. Environ. 21: 2453-2466.
6
7 Pilinis, C.; Seinfeld, J. H. (1988) Development and evaluation of an Eulerian photochemical gas-aerosol model.
8 Atmos. Environ. 22: 1985-2001.
9
10 Pilinis, C.; Seinfeld, J. H.; Grosjean, D. (1989) Water content of atmospheric aerosols. Atmos. Environ.
11 23: 1601-1606.
12
13 Pinnick, R. G.; Fernandez, G.; Hinds, B. D.; Bruce, C. W.; Schaefer, R. W.; Pendleton, J. D. (1985) Dust
14 generated by vehicular traffic on unpaved roadways: sizes and infrared extinction characteristics. Aerosol
15 Sci. Technol. 4: 99-121.
16
17 Pinnick, R. G.; Fernandez, G.; Martinez-Andazola, E.; Hinds, B. D.; Hansen, A. D. A.; Fuller, J. (1993)
18 Aerosol in the arid southwestern United States: measurements of mass loading, volatility, size distribution,
19 absorption characteristics, black carbon content, and vertical structure to 7 km above sea level. J. Geophys.
20 Res. 98: 2651-2666.
21
22 Pires, M.; Rossi, M. J.; Ross, D. S. (1994) Kinetic and mechanistic aspects of the NO oxidation by O2 in
23 aqueous phase. Int. J. Chem. Kinet. 26: 1207-1227.
24
25 Pitchford, M. L.; McMurry, P. H. (1994) Relationship between measured water vapor growth and chemistry of
26 atmospheric aerosol for Grand Canyon, Arizona, in winter 1990. Atmos. Environ. 28: 827-839.
27
28 Pitts, J. N., Jr.; Atkinson, R.; Sweetman, J. A.; Zielinska, B. (1985) The gas-phase reaction of naphthalene
29 with N2O5 to form nitronaphthalenes. Atmos. Environ. 19: 701-705.
30
31 Placet, M.; Battye, R. E.; Fehsenfeld, F. C.; Bassett, G. W. (1991) Emissions involved in acidic deposition
32 processes. In: Irving, P. M., ed. Acidic depostion: state of science and technology, volume I: emissions,
33 atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation Assessment
34 Program. (State of science and technology report no. 1).
35
36 Pleim, J.; Venkatram, A.; Yamartino, R. J. (1984) The dry deposition model, v. 4. Rexdale, ON, Canada:
37 Ontario Ministry of the Environment, ADOM/TADAP Model Development Program.
38
39 Post, J. E.; Buseck, P. R. (1985) Quantitative energy-dispersive analysis of lead halide particles from the
40 Phoenix urban aerosol. Environ. Sci. Technol. 19: 682-685.
41
42 Pratsinis, S.; Ellis, E. C.; Novakov, T.; Friedlander, S. K. (1984) The carbon containing component of the Los
43 Angeles aerosol: source apportionment and contributions to the visibility budget. J. Air Pollut. Control
44 Assoc. 34: 643-650.
45
46 Pratsinis, S. E.; Zeldin, M. D.; Ellis, E. C. (1988) Source resolution of the fine carbonaceous aerosol by
47 principal component-stepwise regression analysis. Environ. Sci. Technol. 22: 212-216.
48
49 Pruppacher, H. R.; Klett, J. D. (1980) Microphysics of clouds and precipitation. Boston, MA: D. Reidel.
50
51 Pueschel, R. F.; Van Valin, C. C.; Castillo, R. C.; Kadlecek, J. A.; Ganor, E. (1986) Aerosols in polluted
52 versus nonpolluted airmasses: long-range transport and effects on clouds. J. Clim. Appl. Meteorol.
53 25: 1908-1917.
54
April 1995 3-208 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rabano, E. S.; Castillo, N. T.; Torre, K. J.; Solomon, P. A. (1989) Speciation of arsenic in ambient aerosols
2 collected in Los Angeles. JAPCA 39: 76-80.
3
4 Rader, D. J.; McMurry, P. H. (1986) Application of the tandem differential mobility analyzer to studies of
5 droplet growth or evaporation. J. Aerosol Sci. 17: 771-787.
6
7 Radke, L. F.; Hobbs, P. V. (1991) Humidity and particle fields around some small cumulus clouds. J. Atmos.
8 Sci. 48: 1190-1193.
Q
10 Raga, G. B.; Jonas, P. R. (1993) On the link between cloud radiative properties and subcloud aerosol
11 concentrations. Q. J. R. Meteorol. Soc. 119: 1419-1425.
12
13 Ranade, M. B. (Arun); Woods, M. C.; Chen, F.-L.; Purdue, L. J.; Rehme, K. A. (1990) Wind tunnel
14 evaluation of PM10 samplers. Aerosol Sci. Technol. 13: 54-71.
15
16 Rao, A. K.; Whitby, K. T. (1978) Non-ideal collection characteristics of inertial impactors—I. single-stage
17 impactors and solid particles. J. Aerosol Sci. 9: 77-86.
18
19 Rau, J. A. (1989) Composition and size distribution of residential wood smoke particles. Aerosol Sci. Technol.
20 10: 181-192.
21
22 Reeks, M. W.; Reed, J.; Hall, D. (1988) On the resuspension of small particles by a turbulent flow. J. Phys.
23 D: Appl. Phys. 21: 574-589.
24
25 Richards, L. W. (1983) Comments on the oxidation of NO2 to nitrate—day and night. Atmos. Environ.
26 17: 397-402.
27
28 Richards, L. W.; Anderson, J. A.; Blumenthal, D. L.; Brandt, A. A.; McDonald, J. A.; Waters, N.; Macias,
29 E. S.; Bhardwaja, P. S. (1981) The chemistry, aerosol physics, and optical properties of a western
30 coal-fired power plant plume. Atmos. Environ. 15: 2111-2134.
31
32 Richards, L. W.; Blanchard, C. L.; Blumenthal, D. L. (1991) Navajo generating station visibility study, draft
33 #2. Santa Rosa, CA: Sonoma Technology, Inc.; final report STI-90200-1124-FRD2.
34
35 Richardson, C. B.; Spann, J. F. (1984) Measurement of the water cycle in a levitated ammonium sulfate
36 particle. J. Aerosol Sci. 15: 563-571.
37
38 Roberts, D.; Williams, D. (1979) The kinetics of oxidation of sulphur dioxide within the plume from a sulphide
39 smelter in a remote region. Atmos. Environ. 13: 1485-1499.
40
41 Rodes, C.; Holland, D.; Purdue, L.; Rehme, K. (1985) A field comparison of PM10 inlets at four locations.
42 J. Air Pollut. Control Assoc. 35: 345-354.
43
44 Roelofs, G. J. (1992a) Drop size dependent sulfate distribution in a growing cloud. J. Atmos. Chem.
45 14: 109-118.
46
47 Roelofs, G. J. H. (1992b) On the drop and aerosol size dependence of aqueous sulfate formation in a
48 continental cumulus cloud. Atmos. Environ. Part A 26: 2309-2321.
49
50 Roelofs, G. J. H. (1993) A cloud chemistry sensitivity study and comparison of explicit and bulk cloud model
51 performance. Atmos. Environ. Part A 27: 2255-2264.
52
53 Rogge, W. F. (1993) Molecular tracers for sources of atmospheric carbon particles: measurements and model
54 predictions [Ph.D. thesis]. Pasadena, California: California Institute of Technology.
April 1995 3-209 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine
2 organic aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 25: 1112-1125
3
4 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine
5 organic aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ.
6 Sci. Technol. 27: 636-651.
7
8 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine
9 organic aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks.
10 Environ. Sci. Technol. 27: 1892-1904.
11
12 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine
13 organic aerosol. 4. Paniculate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
14 27:2700-2711.
15
16 Rogge, W. F.; Mazurek, M. A.; Hildemann, L. M.; Cass, G. R.; Simoneit, B. R. T. (1993d) Quantification of
17 urban organic aerosols at a molecular level: identification, abundance and seasonal variation. Atmos.
18 Environ. Part A 27: 1309-1330.
19
20 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R. (1994) Sources of fine organic aerosol:
21 6. cigarette smoke in the urban atmosphere. Environ. Sci. Technol. 28: 1375-1388.
22
23 Rojas, C. M.; Injuk, J.; Van Grieken, R. E.; Laane, R. W. (1993) Dry and wet deposition fluxes of Cd, Cu,
24 Pb and Zn into the Southern Bight of the North Sea. Atmos. Environ. Part A 27: 251-259.
25
26 Rolph, G. D.; Draxler, R. R. (1990) Sensitivity of three-dimensional trajectories to the spatial and temporal
27 densities of the wind field. J. Appl. Meteorol. 29: 1043-1054.
28
29 Rood, M. J.; Larson, T. V.; Covert, D. S.; Ahlquist, N. C. (1985) Measurement of laboratory and ambient
30 aerosols with temperature and humidity controlled nephelometry. Atmos. Environ. 19: 1181-1190.
31
32 Rood, M. J.; Shaw, M. A.; Larson, T. V.; Covert, D. S. (1989) Ubiquitous nature of ambient metastable
33 aerosol. Nature (London) 337: 537-539.
34
35 Roth, P. M. (1992) [Personal ommunication].
36
37 Rounds, S. A.; Pankow, J. F. (1990) Application of a radial diffusion model to describe gas/particle sorption
38 kinetics. Environ. Sci. Technol. 24: 1378-1386.
39
40 Rubel, G. O. (1991) Partitioning of partially soluble volatiles between the vapor and liquid aerosol phase.
41 Atmos. Environ. Part A 25: 1009-1012.
42
43 Rubel, G. 0.; Gentry, J. W. (1984) Measurement of the kinetics of solution droplets in the presence of
44 adsorbed monolayers: determination of water accommodation coefficients. J. Phys. Chem. 88: 3142-3148.
45
46 Russell, A. G.; Cass, G. R. (1986) Verification of a mathematical model for aerosol nitrate and nitric acid
47 formation and its use for control measure evaluation. Atmos. Environ. 20: 2011-2025.
48
49 Samson, P. J.; Ragland, K. W. (1977) Ozone and visibility reduction in the Midwest: evidence for large-scale
50 transport. J. Appl. Meteorol. 16: 1101-1106.
51
52 Samson, R. J.; Mulholland, G. W.; Gentry, J. W. (1987) Structural analysis of soot agglomerates. Langmuir
53 3: 272-281.
54
April 1995 3.210 DRAFT-DO NOT QUOTE OR CITE
-------
1 Satsumakayashi, H.; Kurita, H.; Yokouchi, Y.; Ueda, H, (1989) Mono and dicarboxylic acids under long-range
2 transport of air pollution in central Japan. Tellus Ser. B. 41: 219-229.
3
4 Satsumakayashi, H.; Kurita, H.; Yokouchi, Y.; Ueda, H. (1990) Photochemical formation of paniculate
5 dicarboxylic acids under long-range transport in central Japan. Atmos. Environ. Part A 24: 1443-1450.
6
7 Savoie, D. L.; Prospero, J. M.; Nees, R. T. (1987) Washout ratios of nitrate, non-sea-salt sulfate and sea-salt
8 on Virginia Key, Florida and on American Samoa. Atmos. Environ. 21: 103-112.
9
10 Saxena, V. K.; Hendler, A. H. (1983) In-cloud scavenging and resuspension of cloud active aerosols during
11 winter storms over Lake Michigan. In: Pruppacher, H. R.; Semonin, R. G.; Slinn, W. G. N., eds.
12 Precipitation scavenging, dry deposition and resuspension. New York, NY: Elsevier; pp. 91-102.
13
14 Saxena, P.; Peterson, T. W. (1981) Thermodynamics of multicomponent electrolytic aerosols. J. Colloid
15 Interface Sci. 79: 496-510.
16
17 Saxena, P.; Hudischewskyj, A. B.; Seigneur, C.; Seinfeld, J. H. (1986) A comparative study of equilibrium
18 approaches to the chemical characterization of secondary aerosols. Atmos. Environ. 20: 1471-1483.
19
20 Saxena, P.; Mueller, P. K.; Kim, Y. P.; Seinfeld, J. H.; Koutrakis, P. (1993) Coupling thermodynamic theory
21 with measurements to characterize acidity of atmospheric particles. Aerosol Sci. Technol. 19: 279-293.
22
23 Schaefer, D. W.; Hurd, A. J. (1990) Growth and structure of combustion aerosols: fumed silica. Aerosol Sci.
24 Technol. 12: 876-890.
25
26 Schroeder, W. H.; Dobson, M.; Kane, D. M.; Johnson, N. D. (1987) Toxic trace elements associated with
27 airborne paniculate matter: a review. JAPCA 37: 1267-1285.
28
29 Schuetzle, D.; Cronn, D.; Crittenden, A. L.; Charlson, R. J. (1975) Molecular composition of secondary
30 aerosol and its possible origin. Environ. Sci. Technol. 9: 838-845.
31
32 Schwartz, S. E. (1979) Residence times in reservoirs under non-steady state conditions: application to
33 atmospheric S02 and aerosol sulfate. Tellus 31: 530-547.
34
35 Schwartz, S. E. (1984) Gas-aqueous reactions of sulfur and nitrogen oxides in liquid-water clouds. In: Calvert,
36 J. G., ed. SO2, NO and NO2 oxidation mechanisms: atmospheric considerations. Boston, MA: Butterworth
37 Publishers; pp. 173-208. (Teasley, J. I., ed. Acid precipitation series: v. 3).
38
39 Schwartz, S. E. (1984) Gas- and aqueous-phase chemistry of HO2 in liquid water clouds. J. Geophys. Res.
40 89: 11589-11598.
41
42 Schwartz, S. E. (1986) Mass-transport considerations pertinent to aqueous-phase reactions of gases in
43 liquid-water clouds. In: Jaeschke, W., ed. Chemistry of multiphase atmospheric systems. Heidelberg,
44 Federal Republic of Germany: Springer; pp. 415-471.
45
46 Schwartz, S. E. (1987) Both sides now: the chemistry of clouds. Ann. N. Y. Acad. Sci. 502- 83-144
47
48 Schwartz, S. E. (1988) Mass-transport limitation to the rate of in-cloud oxidation of SO2: re-examination in the
49 light of new data. Atmos. Environ. 22: 2491-2499.
50
51 Schwartz, S. E. (1989) Acid deposition: unraveling a regional phenomenon. Science (Washington, DC)
52 243: 753-763.
53
April 1995 3_2H DRAFT-DO NOT QUOTE OR CITE
-------
1 Schwartz, S. E. (1991) Aqueous phase reactions. In: Irving, P. M., ed. Acidic deposition: state of science and
2 technology, volume I: emissions, atmospheric processes, and deposition. Washington, DC: The U.S.
3 National Acid Precipitation Assessment Program; pp. 2-93—2-104. (Hicks, B. B.; Draxler, R. R.; Dodge,
4 M.; Hales, J. M.; Albritton, D. L.; Schwartz, S. E.; Meyers, T. P.; Fehsenfeld, F. C.; Tanner, R. L.;
5 Vong, R. J. Atmospheric processes research and process model development: state of science and
6 technology report no. 2).
7
8 Schwartz, S. E. (1994) Cloud studies seen from a physical chemist's perspective. In: Angeletti, G.; Restelli,
9 G., eds. Physico-chemical behaviour of atmospheric pollutants, v. 2. Proceedings of the sixth European
10 symposium; October 1992; Varese, Italy. Luxembourg: Office of Official Publications of the European
11 Commission; pp. 891-900.
12
13 Schwartz, S. E.; Freiberg, J. E. (1981) Mass-transport limitation to the rate of reaction of gases in liquid
14 droplets: application to oxidation of SO2 in aqueous solutions. Atmos. Environ. 15: 1129-1144.
15
16 Schwartz, S. E.; Newman, L. (1978) Processes limiting oxidation of sulfur dioxide in stack plumes. Environ.
17 Sci. Technol. 12: 67-73.
18
19 Schwartz, S. E.; Newman, L. (1983) Measurements of sulfate production in natural clouds [discussion of Hegg
20 and Hobbs (1982)]. Atmos. Environ. 17: 2629-2632.
21
22 Schwartz, S. E.; White, W. H. (1983) Kinetics of reactive dissolution of nitrogen oxides into aqueous solution.
23 In: Advances in environmental science and technology, v. 12. New York, NY: John Wiley and Sons;
24 pp. 1-116.
25
26 Scire, J. S.; Lurmann, F. W.; Bass, A.; Hanna, S. R. (1983) Development of the MESOPUFF II dispersion
27 model. Research Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Sciences
28 Research Laboratory; EPA report no. EPA-600/3-84-057. Available from: NTIS, Springfield, VA;
29 PB84-184753.
30
31 Scott, B. C.; Laulainen, N. S. (1979) On the concentration of sulfate in precipitation. J. Appl. Meteorol.
32 18: 138-147.
33
34 Seaman, N. L. (1990) Meteorological modeling applied to regional air-quality studies using four-dimensional
35 data assimilation. In: Proceedings of the IBM summer institute on environmental modeling; July; Oberlech,
36 Austria.
37
38 Sega, K.; Fugas, M. (1984) Seasonal and spatial differences in mass concentration levels and particle size
39 distribution of aerosols over an urban area. Atmos. Environ. 18: 2433-2437.
40
41 Sehmel, G. A. (1973) Particle resuspension from an asphalt road caused by car and truck traffic. Atmos.
42 Environ. 7: 291-309.
43
44 Sehmel, G. A. (1980) Particle and gas dry deposition: a review. Atmos. Environ. 14: 983-1011.
45
46 Sehmel, G. (1984) Deposition and resuspension. In: Atmospheric science and power production. Washington,
47 DC: U.S. Department of Energy, Office of Science and Technical Information; pp. 533-583; report no.
48 DOE/TIC-27601.
49
50 Seidl, W. (1989) Ionic concentrations and initial S(IV) oxidation rates in droplets during the condensation stage
51 of a cloud. Tellus Ser. B 41: 32-50.
52
53 Seigneur, C.; Saxena, P. (1984) A study of atmospheric acid formation in different environments. Atmos.
54 Environ. 18: 2109-2124.
April 1995 3-212 DRAFT-DO NOT QUOTE OR CITE
-------
1 Seinfeld, J. H. (1986) Atmospheric chemistry and physics of air pollution. New York, NY: John Wiley and
2 Sons.
3
4 Seinfeld, J. H. (1988) Ozone air quality models: a critical review. JAPCA 38: 616-645.
5
6 Seinfeld, J. H.; Bassett, M. (1982) Effect of the mechanism of gas-to-particle conversion on the evolution of
7 aerosol size distributions. In: Heterogeneous atmospheric chemistry. Washington, DC: American
8 Geophysical Union; pp. 6-12.
9
10 Seinfeld, J. H.; Flagan, R. C.; Petti, T. B.; Stern, J. E.; Grosjean, D. (1987) Aerosol formation in aromatic
11 hydrocarbon-NOx systems. Pasadena, CA: California Institute of Technology; project no. AP-6.
12
13 Sergides, C. A.; Jassim, J. A.; Chughtai, A. R.; Smith, D. M. (1987) Appl. Spectrosc. 41: 482-492.
14
15 Shah, J. J.; Johnson, R. L.; Heyerdahl, E. K.; Huntzicker, J. (1986) Carbonaceous aerosol at urban and rural
16 sites in the United States. J. Air Pollut. Control Assoc. 36: 254-257.
17
18 Shao, Y.; Raupach, M.; Findlater, P. (1993) The effect of saltation bombardment on the entrainment of dust by
19 wind. J. Geophys. Res.: 12719-12726.
20
21 Shaw, G. E. (1989) Production of condensation nuclei in clean air by nucleation of H2SO4. Atmos. Environ.
22 23: 2841-2846.
23
24 Shaw, M. A.; Rood, M. J. (1990) Measurement of the crystallization humidities of ambient aerosol particles.
25 Atmos. Environ. Part A 24: 1837-1841.
26
27 Sheridan, P. J.; Schnell, R. C.; Kahl, J. D.; Boatman, J. F.; Garvey, D. M. (1993) Microanalysis of the
28 aerosol collected over south-central New Mexico during the ALIVE field experiment, May-December 1989.
29 Atmos. Environ. Part A 27: 1169-1183.
30
31 Shi, B.; Seinfeld, J. H. (1991) On mass transport limitation to the rate of reaction of gases in liquid droplets.
32 Atmos. Environ. Part A 25: 2371-2383.
33
34 Shinn, J.; Homan, D.; Gay, D. (1983) Plutonium aerosol fluxes and pulmonary exposure rates during
35 resuspension from base soils near a chemical separation facility. In: Pruppacher, H.; Sermonin, R.; Slinn,
36 W., eds. Precipitation scavenging, dry deposition and resuspension. Amsterdam, The Netherlands: Elsevier;
37 pp. 1131-1143.
38
39 Showman, R. E.; Hendricks, J. C. (1989) Trace element content of Flavoparmelia caperata (L.) hale due to
40 industrial emissions. JAPCA 39: 317-320.
41
42 Sievering, H.; Van Valin, C. C.; Barrett, E. W.; Pueschel, R. F. (1984) Cloud scavenging of aerosol sulfur:
43 two case studies. Atmos. Environ. 18: 2685-2690.
44
45 Sievering, H.; Boatman, J.; Galloway, J.; Keene, W.; Kim, Y.; Luria, M.; Ray, J. (1991) Heterogeneous
46 sulfur conversion in sea-salt aerosol particles: the role of aerosol water content and size distribution. Atmos.
47 Environ. Part A 25: 1479-1487.
48
49 Sievering et al. (1994) dry deposition.
50
51 Sievering, H.; Gorman, E.; Kim, Y.; Ley, T.; Seidl, W.; Boatman, J. (1994) Heterogeneous conversion
52 contribution to the sulfate observed over Lake Michigan. Atmos. Environ. 28: 367-370
53
April 1995 3_213 DRAFT-DO NOT QUOTE OR CITE
-------
1 Simoneit, B. R. T. (1984) Organic matter of the troposphere—III. characterization and sources of petroleum and
2 pyrogenic residues in aerosols over the western United States. Atmos. Environ. 18: 51-67.
3
4 Simoneit, B. R. T. (1986) Characterization of organic constituents in aerosols in relation to their origin and
5 transport: a review. Presented at: the 2nd workshop on chemistry analysis of hydrocarbons in the
6 environment; November 1985; Barcelona, Spain. Int. J. Environ. Anal. Chem. 23: 207-237.
7
8 Simoneit, B. R. T. (1989) Organic matter in the troposphere: V. application of molecular marker analysis to
9 biogenic emissions into the troposphere for sources reconciliations.. J. Atmos. Chem.: 251-278.
10
11 Simoneit, B. R. T.; Mazurek, M. A. (1982) Organic matter of the troposphere—II. natural background of
12 biogenic lipid matter in aerosols over the rural western United States. Atmos. Environ. 16: 2139-2159.
13
14 Sisler, J. F.; Malm, W. C. (1994) The relative importance of soluble aerosols to spatial and seasonal trends of
15 impaired visibility in the United States. Atmos. Environ. 28: 851-862.
16
17 Sisterson, D. L.; Bowersox, V. C.; Olsen, A. R.; Meyers, T. P.; Olsen, A. R.; Vong, R. L. (1991)
18 Deposition monitoring: methods and results. In: Irving, P. M., ed. Acidic deposition: state of science and
19 technology, volume I: emissions, atmospheric processes, and depostition. Washington, DC: The U.S.
20 National Acid Precipitation Assessment Program. (State of science and technology report no. 6).
21
22 Slinn, W. G. N. (1983) Air-to-sea transfer of particles. In: Liss, P. S.; Slirm, W. G. N., eds. Air-to-sea
23 exchange of gases and particles. Hingham, MA: D. Reidel; pp. 299-405.
24
25 Sloane, C. S. (1982) Visibility trends—II. mideastern United States 1948-1978. Atmos. Environ.
26 16:2309-2321.
27
28 Sloane, C. S. (1983) Summertime visibility declines: meteorological influences. Atmos. Environ. 17: 763-774.
29
30 Sloane, C. S.; Groblicki, P. J. (1981) Denver's visibility history. Atmos. Environ. 15: 2631-2638.
31
32 Sloane, C. S.; Watson, J.; Chow, J.; Pritchett, L.; Richards, L. W. (1991) Size-segregated fine particle
33 measurements by chemical species and their impact on visibility impairment in Denver. Atmos. Environ.
34 Part A 25: 1013-1024.
35
36 Smith, T. B. (1981) Some observations of pollutant transport associated with elevated plumes. Atmos. Environ.
37 15: 2197-2203.
38
39 Smith, D. M.; Akhter, M. S.; Jassim, J. A.; Sergides, C. A.; Welch, W. F.; Chughtai, A. R. (1989) Studies
40 of the structure and reactivity of soot. Aerosol Sci. Technol. 10: 311-325.
41
42 Snider, J. R.; Vali, G. (1994) Sulfur dioxide oxidation in winter orographic clouds. J. Geophys. Res. [Atmos.]
43 99: 18713-18733.
44
45 Solomon, P. A.; Larson, S. M.; Fall, T.; Cass, G. R. (1988) Basinwide nitric acid and related species
46 concentrations observed during the Claremont Nitrogen Species Comparison Study. Atmos. Environ.
47 22: 1587-1594.
48
49 Solomon, P. A.; Fall, T.; Salmon, L.; Cass, G. R.; Gray, H. A.; Davidson, A. (1989) Chemical
50 characteristics of PM10 aerosols collected in the Los Angeles area. JAPCA 39: 154-163.
51
52 Solomon, P. A.; Salmon, L. G.; Fall, T.; Cass, G. R. (1992) Spatial and temporal distribution of atmospheric
53 nitric acid and paniculate nitrate concentrations in the Los Angeles area. Environ. Sci. Technol.
54 26: 1596-1601.
April 1995 3-214 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spann, J. F.; Richardson, C. B. (1985) Measurement of the water cycle in mixed ammonium acid sulfate
2 particles. Atmos. Environ. 19: 819-825.
3
4 Squires, P. (1958a) The microstructure and colloidal stability of warm clouds: I. the relation between structure
5 and stability. Tellus 10: 256-261.
6
7 Squires, P. (1958b) The microstructure and colloidal stability of warm clouds: II. the causes of the variations in
8 microstructure. Tellus 10: 262-271.
9
10 Standley, L. J.; Simoneit, B. R. T. (1987) Characterization of extractable plant wax, resin, and thermally
11 matured components in smoke particles from prescribed burns. Environ. Sci. Technol. 21: 163-169.
12
13 Stein, S. W.; Turpin, B. J.; Cai, X.; Huang, P.-F.; McMurry, P. H. (1994) Measurements of relative
14 humidity-dependent bounce and density for atmospheric particles using the DMA-impactor technique.
15 Atmos. Environ. 10: 1739-1746.
16
17 Steiner, D.; Burtscher, H.; Gross, H. (1992) Structure and disposition of particles from a spark-ignition engine.
18 Atmos. Environ. Part A 26: 997-1003.
19
20 Stelson, A. W.; Seinfeld, J. H. (1981) Chemical mass accounting of urban aerosol. Environ. Sci. Technol.
21 15: 671-679.
22
23 Stelson, A. W.; Seinfeld, J. H. (1982a) Relative humidity and temperature dependence of the ammonium nitrate
24 dissociation constant. Atmos. Environ. 16: 983-992.
25
26 Stelson, A. W.; Seinfeld, J. H. (1982b) Relative humidity and pH dependence of the vapor pressure of
27 ammonium nitrate-nitric acid solutions at 25 °C. Atmos. Environ. 16: 993-1000.
28
29 Stern, J. E.; Flagan, R. C.; Grosjean, D.; Seinfeld, J. H. (1987) Aerosol formation and growth in atmospheric
30 aromatic hydrocarbon photooxidation. Environ. Sci. Technol. 21: 1224-1231.
31
32 Stevens, R. K.; Pace, T. G. (1984) Overview of the mathematical and empirical receptor models workshop
33 (Quail Roost II). Atmos. Environ. 18: 1499-1506.
34
35 Stevens, R. K.; Dzubay, T. G. ; Russwurm, G.; Rickel, D. (1978) Sampling and analysis of atmospheric
36 sulfates and related species. Atmos. Environ. 12: 55-68.
37
38 Stevens, R. K.; Dzubay, T. G.; Shaw, R. W., Jr.; McClenny, W. A.; Lewis, C. W.; Wilson, W. E. (1980)
39 Characterization of the aerosol in the Great Smoky Mountains. Environ. Sci. Technol. 14: 1491-1498.
40
41 Stevens, R. K.; Dzubay, T. G.; Lewis, C. W.; Shaw, R. W., Jr. (1984) Source apportionment methods applied
42 to the determination of the origin of ambient aerosols that affect visibility in forested areas. Atmos.
43 Environ. 18: 261-272.
44
45 Stewart, D. A.; Liu, M.-K. (1981) Development and application of a reactive plume model. Atmos. Environ.
46 15: 2377-2393.
47
48 Strapp, J. W. et al. (1988) Winter cloud water and air composition in central Ontario. J. Geophys. Res.
49 93: 3760-3772.
50
51 Stumm, W.; Morgan, J. J. (1970) Aquatic chemistry. New York, NY: Wiley-Interscience.
52
53 Suh, H. H.; Allen, G. A.; Aurian-Blajeni, B.; Koutrakis, P.; Burton, R. M. (1994) Field method comparison
54 for the characterization of acid aerosols and gases. Atmos. Environ. 28: 2981-2989.
April 1995 3-215 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sutherland, A. J. (1967) Proposed mechanism for sediment entrainment by turbulent flows. J. Geophys. Res.
2 72: 6183-6194.
3
4 Sverdrup, G. M.; Whitby, K. T. (1980) The effect of changing relative humidity on aerosol size distribution
5 measurements. In: Hidy, G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The
6 character and origins of smog aerosols: a digest of results from the California Aerosol Characterization
7 Experiment (ACHEX). New York, NY: John Wiley & Sons; pp. 527-558. (Advances in environmental
8 science and technology: v. 10).
9
10 Sviridenkov, M. A.; Gillette, D. A.; Isakov, A. A.; Sokolik, I. N.; Smirnov, V. V.; Belan, B. D.; Pachenko,
11 M. V.; Andronova, A. V.; Kolomiets, S. M.; Zhukov, V. M.; Zhukovsky, D. A. (1993) Size distributions
12 of dust aerosol measured during the Soviet-American experiment in Tadzhikistan, 1989. Atmos. Environ.
13 Part A 27: 2481-2486.
14
15 Sweet, C.; Vermette, S. (1993) Sources of toxic trace elements in urban air in Illinois. Environ. Sci. Technol.
16 27: 2502-2510.
17
18 Swozdziak, J. W.; Swozdziak, A. B. (1990) Sulfate aerosol production in the Sudely Range, Poland. J. Aerosol
19 Sci.: S369-S372.
20
21 ten Brink, H. M.; Schwartz, S. E.; Daum, P. H. (1987) Efficient scavenging of aerosol sulfate by liquid-water
22
clouds. Atmos. Environ. 21: 2035-2052.
23
24 Tang, I. N. (1980) Deliquescence properties and particle size change of hygroscopic aerosols. In: Willeke, K.,
25 ed. Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann Arbor Science;
26 pp. 153-167.
27
28 Tang, I. N.; Lee, J. H. (1987) Accomodation coefficients of ozone and sulfur dioxide: their implications on
29 S02 oxidation in cloud water. In: Johnson, R. W.; Gordon, G. E., eds. The chemistry of acid rain: sources
30 and atmospheric processes. Washington, DC: American Chemical Society; pp. 109-117.
31
32 Tang, I. N.; Munkelwitz, H. R. (1977) Aerosol growth studies—III. ammonium bisulfate aerosols in a moist
33 atmosphere. J. Aerosol Sci. 8: 321-330.
34
35 Tang, I. N.; Munkelwitz, H. R.; Davis, J. G. (1978) Aerosol growth studies—IV. phase transformation of
36 mixed salt aerosols in a moist atmosphere. J. Aerosol Sci. 9: 505-511.
37
38 Tang, I. N.; Munkelwitz, H. R. (1993) Composition and temperature dependence of the deliquescence
39 properties of hygroscopic aerosols. Atmos. Environ. Part A 27: 467-473.
40
41 Tang, I. N.; Munkelwitz, H. R. (1994) Water activities, densities, and refractive indices of aqueous sulfates and
42 sodium nitrate droplets of atmospheric importance. J. Geophys. Res. [Atmos.] 99: 18801-18808.
43
44 Tang, I. N.; Wong, W. T.; Munkelwitz, H. R. (1981) The relative importance of atmospheric sulfates and
45 nitrates in visibility reduction. Atmos. Environ. 15: 2463-2471.
46
47 Tani, B.; Siegel, S.; Johnson, S. A.; Kumar, R. (1983) X-ray diffraction investigation of atmospheric aerosols
48 in the 0.3-1.0 pan aerodynamic size range. Atmos. Environ. 17: 2277-2283.
49
50 Tao, Y.; McMurry, P. H. (1989) Vapor pressures and surface free energies of C14-C18 monocarboxylic acids
51 and C5 and C6 dicarboxylic acids. Environ. Sci. Technol. 23: 1519-1523.
52
53 Taylor, G. I. (1921) Diffusion by continuous movements. Proc. London Math. Soc. 20: 196-212.
54
April 1995 3-216 DRAFT-DO NOT QUOTE OR CITE
-------
1 Taylor, G. R. (1989) Sulfate production and deposition in midlatitude continental cumulus clouds: II. chemistry
2 model formulation and sensitivity analysis. J. Atmos. Sci. 46: 1991-2007.
3
4 Taylor, D. D.; Flagan, R. C. (1982) The influence of combustor operation on fine particles from coal
5 combustion. Aerosol Sci. Technol. 1: 103-117.
6
7 Tesche, T. W.; Haney, J. L.; Morris, R. E. (1987) Performance evaluation of four grid-based dispersion
8 models in complex terrain. Atmos. Environ. 21: 233-256.
9
10 Tesche, T. W.; Balentine, H.; Fosdick, E. (1990) Ozone modeling protocol for the San Diego air quality study.
11 San Diego, CA: San Diego County Air Pollution Control District.
12
13 Tesche, T. W.; Roth, P. M.; Reynolds, S. D.; Lurmann, F. W. (1992) Scientific assessment of the urban
14 airshed model (UAM-IV). Washington, DC: American Petroleum Institute; report no. API-300-92.
15
16 Thibodeaux, L. J.; Nadler, K. C.; Valsaraj, K. T.; Reible, D. D. (1991) The effect of moisture on volatile
17 organic chemical gas-to-particle partitioning with atmospheric aerosols—competitive adsorption theory
18 predictions. Atmos. Environ. Part A 25: 1649-1656.
19
20 Thomas, P.; Vogt, S. (1990) Mesoscale atmospheric dispersion experiments using tracer and tetroons. Atmos.
21 Environ. Part A 24: 1271-1284.
22
23 TSI. (1993) Model 390045 DISTFIT ® aerosol data fitting software. St. Paul and Golden Valley, MN: TSI, Inc.
24 and Chimera Software.
25
26 Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a
27 descriptive analysis of organic and elemental carbon concentrations. Atmos. Environ. Part A 25: 207-215.
28
29 Turpin, B. J.; Huntzicker, J. J. (1994) Identification of secondary organic aerosol episodes and quantitation of
30 primary and secondary organic aerosol concentrations during SCAQS. Atmos. Environ.: accepted.
31
32 Turpin, B. J.; Cary, R. A.; Huntzicker, J. J. (1990) An in situ, time-resolved analyzer for aerosol organic and
33 elemental carbon. Aerosol Sci. Technol. 12: 161-171.
34
35 Turpin, B. J.; Huntzicker, J. J.; Larson, S. M.; Cass, G. R. (1991) Los Angeles summer midday paniculate
36 carbon: primary and secondary aerosol. Environ. Sci. Technol. 25: 1788-1793.
37
38 Turpin, B. J.; Liu, S.-P.; Podolske, K. S.; Gomes, M. S. P.; Eisenreich, S. J.; McMurry, P. H. (1993)
39 Design and evaluation of a novel diffusion separator for measuring gas/particle distributions of semivolatile
40 organic compounds. Environ. Sci. Technol. 27: 2441-2449.
41
42 Twohy, C. H.; Austin, P. H.; Charlson, R. J. (1989) Chemical consequences of the initial diffusional growth of
43 cloud droplets: a clean marine case. Tellus Ser. B 41: 51-60.
44
45 Twomey, S. (1959) The nuclei of natural cloud formation. Part II. The supersaturation in natural clouds and the
46 variation of cloud droplet concentration. Geofis. Pura Appl. 43: 243-249.
47
48 Twomey, S.; Warner, J. (1967) Comparison of measurements of cloud droplets and cloud nuclei. J. Atmos.
49 Sci. 24: 702-703.
50
51 U.S. Environmental Protection Agency. (1982) Air quality criteria for paniculate matter and sulfur oxides.
52 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and
53 Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield VA-
54 PB84-156777.
April 1995 3.217 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Environmental Protection Agency. (1986a) Health assessment document for nickel and nickel compounds.
2 Research Triangle Park, NC: Environmental Criteria and Assessment Office; report no.
3 EPA/600/8-83/012FF. Available from: NTIS, Springfield, VA; PB86-232212.
4
5 U.S. Environmental Protection Agency. (1986b) Air quality criteria for lead. Research Triangle Park, NC:
6 Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office; EPA
7 report no. EPA-600/8-83/028aF-dF. 4v. Available from: NTIS, Springfield, VA; PB87-142378.
8
9 U.S. Environmental Protection Agency. (1987a) PM10 SIP development guideline. Research Triangle Park, NC:
10 U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; EPA report no.
11 EPA-450/2-86-001. Available from: NTIS, Springfield, VA; PB87-206488.
12
13 U.S. Environmental Protection Agency. (1987b) Protocol for reconciling differences among receptor and
14 dispersion models. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air
15 Quality Planning and Standards; EPA report no. EPA-450/4-87-008. Available from: NTIS, Springfield,
16 VA; PB87-206504.
17
18 U.S. Environmental Protection Agency. (1994) Guidelines for PM-10 sampling and analysis applicable to
19 receptor modeling. Research Triangle Park, NC: Office of Air Quality Planning and Standards; report no.
20 EPA/452/R-94/009. Available from: NTIS, Springfield, VA; PB94-177441.
21
22 U.S. Environmental Protection Agency. (1995) Air quality criteria for ozone and related photochemical
23 oxidants. Washington, DC: Office of Research and Development; EPA report nos. EPA/600/AP-93/004a-c.
24 3v.
25
26 Van Borm, W. A.; Adams, F. C.; Maenhaut, W. (1989) Characterization of individual particles in the Antwerp
27 aerosol. Atmos. Environ. 23: 1139-1151.
28
29 Van Doren, J. M.; Watson, L. R.; Davidovits, P.; Worsnop, D. R.; Zahniser, M. S.; Kolb, C. E. (1990)
30 Termperature dependence of the uptake coefficients of HNO3, HC1, and N2O5 by water droplets. J. Phys.
31 Chem. 94: 3265-3269.
32
33 Van Grieken, R.; Xhoffer, C. (1992) Microanalysis of individual environmental particles. J. Anal.
34 At. Spectrom. 7: 81-88.
35
36 Van Malderen, H.; Rojas, C.; Van Grieken, R. (1992) Characterization of individual giant aerosol particles
37 above the North Sea. Environ. Sci. Technol. 26: 750-756.
38
39 Van Valin, C. C.; Luria, M.; Ray, J. D.; Boatman, J. F. (1990) Hydrogen peroxide and ozone over
40 northeastern United States in 1987. J. Geophys. Res. 95: 5689-5695.
41
42 Vanderpool, R. W.; Lundgren, D. A.; Marple, V. A.; Rubow, K. L. (1987) Cocalibration of four
43 large-particle impactors. Aerosol Sci. Technol. 7: 177-185.
44
45 Venkataraman, C.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic hydrocarbons and
46 elemental carbon. 2. Ambient measurements and effects of atmospheric processes. Environ. Sci. Technol.
47 28: 563-572.
48
49 Venkataraman, C.; Hildemann, L. M. (1994) factors affecting the distribution of primary semivolatile organi
50 compounds of urban aerosol particles. Environ. Sci. Technol.: submitted.
51
52 Venkataraman, C.; Lyons, J. M.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic
53 hydrocarbons and elemental carbon. 1. Sampling, measurement methods, and source characterization.
54 Environ. Sci. Technol. 28: 555-562.
April 1995 3-218 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wagman, D. D.; Evans, W. H.; Parker, V. B.; Schumm, R. H.; Halow, I.; Bailey, S. M.; Churney, K. L.;
2 Nuttall, R. L. (1982) The NBS tables of chemical thermodynamic properties, selected values for inorganic
3 and Cl and C2 organic substances in SI units. J. Phys. Chem. Ref. Data II(suppl. 2): 1-392.
4
5 Walcek, C. J.; Taylor, G. R. (1986) A theoretical method for computing vertical distributions of acidity and
6 sulfate production within cumulus clouds. J. Atmos. Sci. 43: 339-355.
7
8 Walcek, C. J.; Stockwell, W. R.; Chang, J. S. (1990) Theoretical estimates of the dynamic radiative and
9 chemical effects of clouds on tropospheric gases. Atmos. Res. 25: 53-69.
10
11 Waldman, J. M.; Lioy, P. J.; Thurston, G. D.; Lippmann, M. (1990) Spatial and temporal patterns in
12 summertime sulfate aerosol acidity and neutralization within a metropolitan area. Atmos. Environ. Part B
13 24: 115-126.
14
15 Wall, S. M.; John, W.; Ondo, J. L. (1988) Measurement of aerosol size distributions for nitrate and major
16 ionic species. Atmos. Environ. 22: 1649-1656.
17
18 Wang, C.; Chang, J. S. (1993) A three-dimensional numerical study of cloud dynamics, micrphysics, and
19 chemistry. 4. Cloud chemistry and precipitation chemistry. J. Geophys. Res. 98: 16799-16808.
20
21 Wang, H.-C.; John, W. (1987) Comparative bounce properties of particle materials. Aerosol Sci. Technol.
22 7: 285-299.
23
24 Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
25 Technol. 8: 157-172.
26
27 Wang, S.-C.; Paulson, S. E.; Grosjean, D.; Flagan, R. C.; Seinfeld, J. H. (1992a) Aerosol formation and
28 growth in atmospheric organic/NOx systems-I. Outdoor smog chamber studies of C7- and C8- hydrocarbons.
29 Atmos. Environ. Part A 26: 403-420.
30
31 Wang, S.-C.; Flagan, R. C.; Seinfeld, J. H. (1992b) Aerosol formation and growth in atmospheric organic/NOx
32 systems—II. aerosol dynamics. Atmos. Environ. Part A 26: 421-434.
33
34 Warneck, P. (1991) Chemical reactions in clouds. Fresenius' J. Anal. Chem. 340: 585-590.
35
36 Warneck, P. (1992) Chemistry and photochemistry in atmospheric water drops. Ber. Bunsen Ges. Phys. Chem.
37 96: 454-460.
38
39 Warner, T. T. (1981) Verification of a three-dimensional transport model using tetroon data from Project
40 STATE. Atmos. Environ. 15: 2219-2222.
41
42 Warner, J.; Twomey, S. (1967) The production of cloud nuclei by can fires and the effect on cloud droplet
43 concentration. J. Atmos. Sci. 24: 704-706.
44
45 Warren, R. S.; Birch, P. (1987) Heavy metal levels in atmospheric particulates, roadside dust and soil along a
46 major urban highway. Sci. Total Environ. 59: 253-256.
47
48 Watson, J. G.; Chow, J. C. (1992) Data bases for PM10 and PM2 5 chemical compositions and source profiles.
49 Transactions AWMA 1: 61-91.
50
51 Watson, J. G.; Henry, R. C.; Cooper, J. A.; Macias, E. S. (1981) The state of the art receptor models relating
52 ambient suspended paniculate matter to sources. In: Macias, E. S.; Hopke, P. K., eds. Atmospheric
53 aerosol: source/air quality relationships. Washington, DC: American Chemical Society. (ACS symposium
54 series no. 167).
April 1995 3-219 DRAFT-DO NOT QUOTE OR CITE
-------
1 Watson, J. G.; Cooper, J. A.; Huntzicker, J. J. (1984) The effective variance weighting for least squares
2 calculations applied to the mass balance receptor model. Atmos. Environ. 18: 1347-1355.
3
4 Watson, J. G.; Chow, J. C.; Mathai, C. V. (1989) Receptor models in air resources management: a summary
5 of the APCA international specialty conference. JAPCA 39: 419-426.
6
7 Watson, J. G.; Chow, J. C.; Pace, T. G. (1991) Chemical mass balance. In: Hopke, P. K., ed. Data handling
8 in science and technology: v. 7, receptor modeling for air quality management. New York, NY: Elsevier
9 Press; pp. 83-116.
10
11 Watson, J. G.; Chow, J. C.; Blumenthal, D. L.; Lurmann, F. W.; Hackney, R. J.; Magliano, K. A.;
12 Pederson, J. R.; Neff, W.; Roth, P. M.; Solomon, P. A.; Thuillier, R. H.; Ziman, S. D. (1994a) Planning
13 for data analysis. In: Solomon, P. A.; Silver, T., eds. Planning and managing air quality modeling and
14 measurement studies: a perspective through SJVAQS/AUSPEX. Pittsburgh, PA: Air and Waste
15 Management Association; pp. 335-349.
16
17 Watson, J. G.; Chow, J. C.; Lurmann, F. W.; Musarra, S. P. (1994b) Ammonium nitrate, nitric acid, and
18 ammonia equilibrium in wintertime Phoenix, Arizona. Air Waste 44: 405-412.
19
20 Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L.
21 (1994c) Chemical mass balance source apportionment of PM10 during the Southern California Air Quality
22 Study. Aerosol Sci. Technol. 21: 1-36.
23
24 Weathers, K. C.; Likens, G. E.; Bormann, F. H.; Bicknell, S. H.; Bormann, B. T.; Daube, B. C., Jr.; Eaton,
25 J. S.; Galloway, J. N.; Keene, W. C.; Kimball, K. D.; McDowell, W. H.; Siccama, T. G.; Smiley, D.;
26 Tarrant, R. A. (1988) Cloudwater chemistry from ten sites in North America. Environ. Sci. Technol.
27 22: 1018-1026.
28
29 Went, F. W. (1960) Blue hazes in the atmosphere. Nature (London) 187: 641-643.
30
31 Wesely, M. L.; Hicks, B. B. (1977) Some factors that affect the deposition rates of sulfur dioxide and similar
32 gases on vegetation. J. Air Pollut. Control Assoc. 27: 1110-1116.
33
34 Wexler, A. S.; Lurmann, F. W.; Seinfeld, J. H. (1994) Modelling urban and regional aerosols—I. model
35 development. Atmos. Environ. 28: 531-546.
36
37 Wexler, A. S.; Seinfeld, J. H. (1990) The distribution of ammonium salts among a size and composition
38 dispersed aerosol. Atmos. Environ. Part A 24: 1231-1246.
39
40 Wexler, A. S.; Seinfeld, J. H. (1991) Second-generation inorganic aerosol model. Atmos. Environ. Part A
41 25: 2731-2748.
42
43 Wexler, A. S.; Seinfeld, J. H. (1992) Analysis of aerosol ammonium nitrate: departures from equilibrium
44 during SCAQS. Atmos. Environ. Part A 26: 579-591.
45
46 Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
47
48 Whitby, K. T. (1980) Aerosol formation in urban plumes. Ann. N. Y. Acad. Sci. 338: 258-275.
49
50 Whitby, K. T. (1984) Physical and optical behavior of sulfate-nitrate aerosols in equilibrium with atmospheric
51 water vapor, ammonia and nitric acid. In: Ruhnke, L. H.; Deepak, A., eds. Hygroscopic aerosols:
52 technical proceedings of the workshop on hygroscopic aerosols in the planetary boundary layer; April 1982;
-53- - Vail, CO. Hampton, VA: A. Deepak Publishing; pp. 45-63.
54
April 1995 3-220 DRAFT-DO NOT QUOTE OR CITE
-------
1 Whitby, K. T.; Sverdrup, G. M. (1980) California aerosols: their physical and chemical characteristics.
2 In: Hidy, G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The character and
3 origins of smog aerosols: a digest of results from the California Aerosol Characterization Experiment
4 (ACHEX). New York, NY: John Wiley & Sons, Inc.; pp. 477-517. (Advances in environmental science
5 and technology: v. 9).
6
7 Whitby, K. T.; Husar, R. B.; Liu, B. Y. H. (1972) The aerosol size distribution of Los Angeles smog. J.
8 Colloid Interface Sci. 39: 177-204.
9
10 Whitby, K. T.; Charlson, R. E.; Wilson, W. E.; Stevens, R. K. (1974) The size of suspended particle matter in
11 air. Science (Washington, DC) 183: 1098-1099.
12
13 Whitby, E. R.; McMurry, P. H.; Shankar, U.; Binkowski, F. S. (1991) Modal aerosol dynamics modeling.
14 Research Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Research and Exposure
15 Assessment Laboratory; EPA report no. EPA/600/3-91/020. Available from: NTIS, Springfield, VA;
16 PB91-161729.
17
18 White, W. H.; Macias, E. S. (1989) Carbonaceous particles and regional haze in the western United States.
19 Aerosol Sci. Technol. 10: 111-117.
20
21 White, W. H.; Anderson, J. A.; Blumenthal, D. L.; Husar, R. B.; Gillani, N. V.; Husar, J. D.; Wilson,
22 W. E., Jr. (1976) Formation and transport of secondary air pollutants: ozone and aerosols in the St. Louis
23 urban plume. Science (Washington, DC) 194: 187-189.
24
25 White, W. H.; Seigneur, C.; Heinold, D. W.; Eltgroth, M. W.; Richards, L. W.; Roberts, P. T.; Bhardwaja,
26 P. S.; Conner, W. D.; Wilson, W. E., Jr. (1985) Predicting the visibility of chimney plumes: an
27 intercomparison of four models with observations at a well-controlled power plant. Atmos. Environ.
28 19: 515-528.
29
30 Willeke, K.; Whitby, K. T. (1975) Atmosperic aerosols: size distribution interpretation. J. Air Pollut. Control
31 Assoc. 25: 529-534.
32
33 Willetts, B. (1992) [Personal communication]. Aberdeen, Scotland: University of Aberdeen.
34
35 Williams, D. J.; Carras, J. N.; Milne, J. W.; Heggie, A. C. (1981) The oxidation and long-range transport of
36 sulphur dioxide in a remote region. Atmos. Environ. 15: 2255-2262.
37
38 Williams, A. L.; Stensland, G. J.; Gatz, D. F.; Barnard, W. R. (1988) Development of a PM10 emission factor
39 from unpaved roads. In: Gatz, D. F.; Stensland, G. J.; Miller, M. V.; Chu, L. C., eds. Alkaline aerosols:
40 an initial investigation of their role in determining precipitation acidity, preprint no. 88-71.4.
41
42 Wilson, W. E. (1978) Sulfates in the atmosphere: a progress report on project MISTT. Atmos. Environ.
43 12: 537-547.
44
45 Wilson, W. E., Jr. (1981) Sulfate formation in point source plumes: a review of recent field studies. Atmos.
46 Environ. 15: 2573-2581.
47
48 Wilson, J. C.; McMurry, P. H. (1981) Studies of aerosol formation in power plant plumes—II. secondary
49 aerosol formation in the Navajo generating station plume. In: White, W. H.; Moore, D. J.; Lodge, J. P.,
50 eds. Plumes and visibility: measurements and model components: proceedings of the symposium; November
51 1980; Grand Canyon National Park, AZ. Atmos. Environ. 15: 2329-2339.
52
53
April 1995 3-221 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wilson, W. E.; Spiller, L. L.; Ellestad, T. G.; Lamothe, P. J.; Dzubay, T. G.; Stevens, R. K.; Macias, E. S.;
2 Fletcher, R. A.; Husar, J. D.; Husar, R. B.; Whitby, K. T.; Kittelson, D. B.; Cantrell, B. K. (1977)
3 General Motors sulfate dispersion experiment: summary of EPA measurements. J. Air Pollut. Control
4 Assoc. 27: 46-51.
5
6 Wilson, J. C.; Gupta, A.; Whitby, K. T.; Wilson, W. E. (1988) Measured aerosol light scattering coefficients
7 compared with values calculated from EAA and optical particle counter measurements: improving the utility
8 of the comparison. Atmos. Environ. 22: 789-793.
9
10 Winkler, P. (1973) The growth of atmospheric aerosol particles as a function of relative humidity: II.
11 an improved concept of mixed nuclei. J. Aerosol Sci. 4: 373-387.
12
13 Winkler, P.; Junge, C. (1972) The growth of atmospheric aerosol particles as a function of relative humidity:
14 I. methods and measurements at different locations. J. Rech. Atmos. 6: 617-638.
15
16 Winklmayr, W.; Wang, H.-C.; John, W. (1990) Adaptation of the Twomey algorithm to the inversion of
17 cascade impactor data. Aerosol Sci. Technol. 13: 322-331.
18
19 Witten, T. A.; Sander, L. M. (1983) Diffusion limited aggregation. Phys. Rev. B: Condens. Matter
20 27: 5686-5697.
21
22 Wolff, G. T. (1981) Paniculate elemental carbon in the atmosphere. J. Air Pollut. Control Assoc. 31: 935-938.
23
24 Wolff, G. T. (1984) On the nature of nitrate in coarse continental aerosols. Atmos. Environ. 18: 977-981.
25
26 Wolff, G. T.; Korsog, P. E. (1985) Estimates of the contributions of sources to inhalable paniculate
27 concentrations in Detroit. Atmos. Environ. 19: 1399-1409.
28
29 Wolff, G. T.; Lioy, P. J.; Wight, G. D.; Meyers, R. E.; Cederwall, R. T. (1977) An investigation of
30 long-range transport of ozone across the midwestern and eastern United States. Atmos. Environ.
31 11:797-802.
32
33 Wolff, G. T.; Groblicki, P. J.; Cadle, S. H.; Countess, R. J. (1982) Paniculate carbon at various locations in
34 the United States. In: Wolff, G. T.; Klimsch, R. L., eds. Paniculate carbon: atmospheric life cycle. New
35 York, NY: Plenum Press; pp. 297-315.
36
37 Wolff, G. T.; Ruthkosky, M. S.; Stroup, D. P.; Korsog, P. E. (1991) A characterization of the principal
38 PM-10 species in Claremont (summer) and Long Beach (fall) during SCAQS. Atmos. Environ. Part A
39 25: 2173-2186.
40
41 Worsnop, D. R.; Zahniser, M. S.; Kolb, C. E.; Gardner, J. A.; Watson, L. R.; Van Doren, J. M.; Jayne,
42 J. T.; Davidovits, P. (1989) Temperature dependence of mass accomodation of SC^ and H2O2 on aqueous
43 surfaces. J. Phys. Chem. 93: 1159-1172.
44
45 Wu, P.-M.; Okada, K. (1994) Nature of coarse nitrate particles in the atmosphere—a single particle approach.
46 Atmos. Environ. 28: 2053-2060.
47
48 Xu, G. B.; Yu, C. P. (1985) Theoretical lung deposition of hygroscopic NaCl aerosols. Aerosol Sci. Technol.
49 4: 455-461.
50
51 Yamartino, R. J.; Scire, J. S.; Hanna, S. R.; Carmichael, G. R.; Chang, Y. S. (1989) CALGRID: a mesoscale
52 photochemical grid model. Sacramento, CA: California Air Resources Board; report no. A049-1.
53
April 1995 3-222 DRAFT-DO NOT QUOTE OR CITE
-------
1 Yamasaki, H.; Kuwata, K.; Miyamoto, H. (1982) Effects of ambient temperature on aspects of airborne
2 polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 16: 189-194.
3
4 Zafiriou, O. C.; Gagosian, R. B.; Peltzer, E. T.; Alford, J. B.; Loder, T. (1985) Air to sea fluxes of lipids at
5 Enewetak Atoll. J. Geophys. Res.: 2409.
6
7 Zeldin, M. D.; Bregman, L. D.; Wylie, B. (1988) Meteorological conditions during the 1985 Southern
8 California Nitrogen Special Methods Comparison Study. Atmos. Environ. 22: 1541-1543.
9
10 Zemaitis, J. F.; Clark, D. M.; Rafal, M.; Scrivner, N. C. (1986) Handbook of aqueous electrolyte
11 thermodynamics. New York, NY: American Institute of Chemical Engineers.
12
13 Zeng, Y.; Hopke, P. K. (1989) Three-mode factor analysis: anew multivariate method for analyzing spatial and
14 temporal composition variation. In: Watson, J. G., ed. Transactions: receptor models in air resources
15 management. Pittsburgh, PA: Air and Waste Management Association; pp. 173-189.
16
17 Zhang, X.; McMurry, P. H. (1991) Theoretical analysis of evaporative losses of adsorbed or absorbed species
18 during atmospheric aerosol sampling. Environ. Sci. Technol. 25: 456-469.
19
20 Zhang, H. X.; Sorensen, C. M.; Ramer, E. R.; Olivier, B. J.; Merklin, J. F. (1988) in situ optical structure
21 factor measurements of an aggregating soot aerosol. Langmuir 4: 867-871.
22
23 Zhang, S.-H.; Shaw, M.; Seinfeld, J. H.; Flagan, R. C. (1992) Photochemical aerosol formation from a-pinene
24 and 0-pinene. J. Geophys. Res. [Atmos.] 97: 20,717-20,729.
25
26 Zhang, X. Q.; McMurry, P. H.; Hering, S. V.; Casuccio, G. S. (1993) Mixing characteristics and water
27 content of submicron aerosols measured in Los Angeles and at the Grand Canyon. Atmos. Environ. Part A
28 27: 1593-1607.
29
30 Zimmerman, P. R. (1979) Testing of hydrocarbon emissions from vegetation, leaf litter and aquatic surfaces,
31 and development of a methodology for compiling biogenic emission inventories: final report. Research
32 Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards;
33 EPA report no. EPA-450/4-79-004. Available from: NTIS, Springfield, VA; PB-296070.
April 1995 3_223 DRAFT-DO NOT QUOTE OR CITE
-------
i 4. SAMPLING AND ANALYSIS OF
2 PARTICULATE MATTER AND ACID DEPOSITION
3
4
5 4.1 INTRODUCTION
6 Assessment of the risks associated with airborne aerosols implies that measurements be
7 made defining the aerosol characteristics, concentrations and exposures that contribute to, or
8 simply correlate with, adverse health effects. The proper selection of an aerosol sampling or
9 analysis methodology to accomplish these measurements requires that rationales be applied
10 that consider how the resulting data will be applied and interpreted, in addition to the data
11 quality required. As an example, integrated collection of an aerosol sample on a heated
12 substrate may help to stabilize a subsequent measurement technique, but in the process may
13 dramatically change the character of the aerosol as it existed in the air. Similarly, integrated
14 collection of acidic fine aerosols, without selectively removing the larger, more basic
15 particles, will cause neutralization (i.e., modification) of the sample on the substrate. The
16 same logic applies to the selective removal of gas phase components during sampling that
17 might react with the deposited aerosol sample, in a manner inconsistent with naturally
18 occurring transformation processes. The assumption that fixed-location measurements are
19 representative of inhalation exposure implies that the effects of local spatial and temporal
20 gradients are understood and appropriately applied to the sampler siting criteria.
21 Development of relationships between aerosol characteristics and health or ecological
22 responses requires that the aerosol sampling and analysis processes are truly representative
23 and adequately defined.
24 The application of sampling and analytical systems for aerosols must recognize that
25 particles exist modally as size distributions (see Section 3.3.3), generated by distinctively
26 different source categories and having distinctly different chemistries. The primary reasons
27 for making size-specific aerosol measurements are (a) to relate the in situ aerosol character to
28 the potential deposition sites, and thus toxicity, of the respiratory system, and (b) separation
29 of the size distribution modes to identify sources, transformation process or aerosol
30 chemistry. The interpretation of particle size must be made based on the diameter definition
31 inherent in the measurement process. Since the respiratory system classifies particles of
April 1995 4-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 unknown shapes and densities based on aerodynamic diameter, development of aerosol
2 relationships with health responses requires that sampling techniques either incorporate
3 inertial aerodynamic sizers or provide mechanisms to accurately convert the measured
4 diameters (e.g., optical) to an aerodynamic basis. All particle diameters described in this
5 chapter are aerodynamic, unless otherwise specified.
6 Friedlander (1977a) provided the descriptive matrix shown in Figure 4-1 for placing
7 measurement techniques that define aerosol characteristics into perspective, in terms of their
8 particle sizing capabilities, resolution times and chemical identification attributes. This
9 approach defined these characteristics by resolution (single particle or greater), discretizing
10 ability, and averaging process. The author notes that the "perfect" aerosol sampler
11 characterizes particle size with "perfect" resolution, determines the chemistry "perfectly" of
12 each particle, and operates in real-time with no "lumping" of classes. These characteristics
13 could be amended in "real-world" terms by suggesting that the "perfect" sampler would also
14 have minimal cost and operator intervention. Additionally, if the aerosol measurement
15 design goal is mimicking the respiratory system, physiological averaging characteristics must
16 be considered. Size-specific, integrated aerosol measurements have improved significantly
17 and their capabilities are better characterized since the 1987 PMi0 standard, but a "perfect"
18 aerosol sampling system has not been devised. As discussed subsequently, the methodologies
19 required to adequately define the performance specifications of aerosol samplers have yet to
20 be devised.
21 Many recent developmental efforts in aerosol measurement technologies have addressed
22 the need to perfect the chemical characterization of reactive or volatile species collected on
23 filtration substrates. Some of the most significant recent advances in aerosol measurement
24 technologies have come in the form of analysis system "protocols", rather than individual
25 pieces of hardware. Recognizing that there is no single "perfect" sampler, these protocols
26 attempt to merge several aerosol sampling and analysis technologies into an adaptable and
27 analytically versatile system. System attributes typically include one or more size-specific
28 aerosol inlets, subsequent fractionators to separate the fine and coarse particle modes, and
29 denuders and/or sequential filter packs to selectively account for reactive gas phase species.
30 Examples include EPA's Versatile Air Pollution Sampler (VAPS), (Conner et al., 1993), the
31 Southern California Air Quality Study (SCAQS) sampler, (Fitz et al., 1989) and the
April 1995 4-2 DRAFT-DO NOT QUOTE OR CITE
-------
Instrument
Perfect Single
Particle Counter
Analyzer
Optical Single
Particle Counter
Electrical
Mobility
Analyzer
Condensation
Nuclei
Counter
Impactor
Impactor
Chemical
Analyzer
Whole Sample
Chemical Analyzer
1 Resolution
Size Time Chemical
Composition
^\ » ^\ • ^\ •
"" """^J * ^W ' ^J
•
-------
1 during normal activities. Miniaturization of aerosol separators stretches the limits of current
2 technologies to maintain required sampling precisions and accuracies. One of the most
3 significant limitations imposed by the low flowrates inherent in personal exposure samplers is
4 the extremely small sample size available for chemical analysis.
5 This chapter briefly describes the technical capabilities and limitations of aerosol
6 sampling and analytical procedures in Sections 4.2 and 4.3, respectively, focusing on those
7 that (1) were used to collect data supporting other sections in this document, (2) those
8 supporting the existing PM10, TSP and Pb regulations, (3) those that were used to support
9 health and welfare response studies, (4) those having application in development of a possible
10 fine particle standard, and (5) discussing the attributes of several new technologies. The
11 discussion of aerosol separation technologies is divided between (a) devices used to mimic
12 the larger particle (> 10 ^m) penetration rationales for the upper airways, and (b) those
13 devices generally used to mimic smaller particle penetration (< 10 /*m) to the sub-thoracic
14 regions. These device descriptions are followed by sampling considerations for their
15 applications. The applications of performance specifications to define these measurement
16 systems for regulatory purposes are discussed, along with a number of critical observations
17 suggesting that the current specification process does not always assure the accuracy or
18 representativeness necessary in the field. The EPA program designating PM10 reference and
19 equivalent sampling systems is then briefly described, along with a current list of designated
20 devices. Selected measurement systems used to provide more detailed characterization of
21 aerosol properties for research studies are discussed, with a focus on the determination of
22 particle size distributions. Aerosol sampling systems for specialty applications, including
23 automated samplers, personal exposure samplers and the sampling systems used in aerosol
24 apportionment studies are briefly described. This chapter is intended to provide supplemental
25 information to Section 3.3.3 and other discussions of aerosol methodologies in support of the
26 existing standards and possible development of a "fine particle" standard in the 1 to 3 /^m
27 range. An important contribution of the sampling and analytical sections is the extensive
28 compilation of salient peer-reviewed technical references that can be obtained by the reader
29 for more detailed information.
April 1995 4.4 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.2 SAMPLING FOR PARTICIPATE MATTER
2 4.2.1 Background
3 The development of relationships between airborne paniculate matter and human or
4 ecological effects requires that the aerosol1 measurement process be accurately, precisely
5 and representatively defined. Improvements in sampling methodologies since the 1982 Air
6 Quality Criteria Document for Particulate Matter and Sulfur Oxides2 was released, have
7 resulted from improved sensor technologies, and more importantly, a better understanding of
8 the aerosol character in situ3. Additionally, health studies and atmospheric chemistry
9 research in the past decade have focused more closely on smaller, better-defined aerosol size
10 fractions of known integrity, collected specifically for subsequent chemical characterization.
11 The system of aerosols in ambient air is a continuum of particle sizes in a gas phase
12 carrier formed as the summation of all size distributions produced by individual sources and
13 secondary transformations. Portions of the composite distributions are often found to exist
14 lognormally (Baron and Willeke, 1993; see also Section 3.3.3). Aerosol systems also exists
15 as a continuum of particle "ages", resulting from loss and transformation mechanisms such
16 as agglomeration, settling, volatilization, gas-particle reaction, and rain-out affecting freshly
17 generated particles. The chemical compositions of the various portions (modes) of the
18 aerosol size distribution are more discreet, and sampling strategies must consider a specific
19 range of sizes for a given chemical class. The constantly changing character of the
20 atmosphere (or indoor air) places a premium on a sampling strategy to remove a
21 representative aerosol sample from the air and protect its integrity until analyzed.
22 The 1982 Criteria Document provided basic descriptions of many of the aerosol
23 measurement techniques still used today, and these methods will be briefly mentioned here,
24 but not described in detail. This section will highlight the more recent peer-reviewed
25 Consistent with recent literature (e.g., see Willeke and Baron, 1993), the term "aerosol" will refer to the continuum
26 of suspended particles and the carrier gas.
27 2U.S. Environmental Protection Agency (1982a), referenced subsequently as an entity as the "1982 PM Criteria
28 Document".
29 3The in situ characteristics of particles in the ambient air medium can be substantially modified by the sampling
30 and analysis processes. For example, a particle counter which draws particles through a restrictive or heated inlet
31 before reaching the sensing volume, may perceive the particle properties (e.g. scattering coefficients, size
32 distributions) differently from those that existed in the ambient.
April 1995 4.5 DRAFT-DO NOT QUOTE OR CITE
-------
1 research on aerosol measurement technologies since 1982 and provide salient points that
2 should be considered in their application. The aerosol sampling section is not intended to be
3 an exhaustive treatise, but is structured to highlight important concepts and technologies
4 relevant to the development of aerosol measurement/response relationships, or supporting
5 existing and potential EPA aerosol regulations. Ancillary reference texts, describing basic
6 aerosol mechanics (e.g., Hinds, 1982; Reist, 1984) and applied aerosol mechanics and
7 measurements (e.g., Willeke and Baron, 1993; Hering, 1989; Lundgren et al., 1979;
8 Friedlander, 1977b, Liu, 1976) should be consulted for more fundamental details.
9
10 4.2.2 Large Particle Separators
11 4.2.2.1 Cutpoint Considerations
12 The collection of an aerosol sample is defined by the penetration characteristics of the
13 inlet, overlaid on the existing in situ size distribution. Cooper and Guttrich (1981) describe
14 this process mathematically, and estimate the influences of non-ideal penetration
15 characteristics. Miller et al. (1979) described the considerations for the possible selection of
16 15 ptm (designated "inhalable") as a standard for size-selective particle sampling with upper
17 airway respiratory deposition as the primary consideration. The selection of the most
18 appropriate aerodynamic criteria for ambient aerosol sampling was only partially resolved by
19 the 1987 EPA designation (U.S. Environmental Protection Agency, 1987b) of a 10 jum
20 cutpoint, PM10 cutpoint. The "ideal" PM10 inlet was referenced to the thoracic penetration
21 model of Lippmann and Chan (1979). Ogden (1992) noted that the standardization for
22 aerosol cutpoint sizes and separation sharpness is still under debate across settings (ambient
23 air, occupational) and across national and international governmental entities. As shown in
24 Figure 4-2 (from Jensen and O'Brien, 1993), the international conventions for cutpoints have
25 been roughly categorized as Respirable, Thoracic and Inhalable (previously, Inspirable).
26 These cutpoints are related to the penetration, respectively, to the gas exchange region of the
27 lung, the larynx, and the nasal/oral plane. The influences of physiological variables on these
28 cutpoints are described by Soderholm (1989). The British Standard EN 481:1993 describes
29 size fraction definitions for workplace aerosol sampling, and identifies inhalable
30 "conventions" relative to thoracic, respirable, extra-thoracic and tracheobronchial penetration
31 (but not necessarily deposition) in the respiratory system. They define a thoracic cumulative
April 1995 4-6 DRAFT-DO NOT QUOTE OR CITE
-------
100
Inhalable
ACGIH(1994)
Proposed ISO (1992)
ISO (1983)
Thoracic
ACGIH(1994)
Proposed ISO(1992)
ISO (1983)
=5 0/100
Respirable
ACGIH(1994)
Proposed ISO(1992)
ISO (1983)
BMRC(1959)
Aerodynamic Diameter
Figure 4-2. American Conference of Governmental Industrial Hygienists
(ACGIH), British Medical Research Council (BMRC), and
International Organization for Standardization (ISO) size-selective
sampling criteria.
April 1995
4-7
DRAFT-DO NOT QUOTE OR CITE
-------
1 lognormal distribution with a median of 11.64 /*m and a geometric standard deviation of 1.5,
2 such that 50% of airborne particles with Da = 10 ^m are in the thoracic region. The
3 American Conference of Governmental and Industrial Hygienists (ACGIH, 1994) also
4 adopted these convention definitions as applied to chemical substance TLV's. Owen et al.
5 (1992) provides an extensive list of the outdoor and indoor particles by type and source
6 category that are found in or overlap these ranges. Willeke et al. (1992) describe the
7 sampling efficiencies and test procedures for bioaerosol monitors.
8 The concept of using an inlet that has the same sampling (penetration) characteristics as
9 portions of the respiratory system has been discussed by a number of researchers, including
10 Lippmann and Chan (1979), Vincent and Mark (1981), Soderholm (1989), Liden and Kenny
11 (1991) and John (1993). They describe sampler design considerations for matching
12 penetration models for respirable, thoracic and inhalable fractions that have been proposed by
13 a number of governing bodies. Since all models proposed for the same fraction do not
14 necessarily coincide, given the variability and differences in interpretation of respiratory
15 system data, Soderholm (1989) proposed compromise conventions for each fraction. Watson
16 et al. (1983), Wedding and Carney (1983) and van der Meulen (1988) mathematically
17 evaluated the influences of inlet design parameters on collection performance relative to
18 proposed sampling criteria. These analyses suggested that factors such as extremes in wind
19 speed and Coarse particle concentration could pose significant problems in meeting
20 performance specifications.
21 An analysis of the human head as an aerosol sampler was discussed by Ogden and
22 Birkett (1977), who noted that breathing is an anisokinetic sampling process. The concept of
23 a "total inhalable" fraction that passes the oral and nasal entry planes was refined by Mark
24 and Vincent (1986) with the development of a personal aerosol sampling inlet that mimicked
25 this penetration as a function of aerodynamic size. The inlet was designated the IOM for the
26 Institute for Occupational Medicine in Edinburgh, Scotland, where it was developed with the
27 cutpoint as a function of wind speed and aerosol type shown in Figure 4-3. The total
28 inhalable approach has been adopted by the International Standards Organization (ISO, 1983
29 and ISO, 1992)4, European Committee for Standardization (CEN, 1993) and by the
30 American Conference on Governmental and Industrial Hygienists (ACGIH, 1985; ACGIH,
31 4 A proposed ISO convention is described by Soderholm (1989)
April 1995 4-8 DRAFT-DO NOT QUOTE OR CITE
-------
»**•
2.
VO
<-"
.£.
t
O
T)
H
b
O
o
H
O
s
w
o
n
120
100
^ 80
&
%^
c
9m/s •'
V A 0 \ /T
"'*(., N. •
— , *'*'*•* x / ^fc~
O '•--,fc - 4m/s *
"""'«•. N. •'
n Xx \ m/
J(*°\ •! *'*!• ^^t'"'~'~'"'^^B"" ^^A ^™"
£\J *<^f ^H J^ ^B
. ~]JL —
v n /
— V / ^ A —
<37 / *
n im/s x
O
* o
, 1 , , , , 1 , , ,,,,,,
1 10 100
Aerodynamic Diameter (|im)
Inhalable
Convention
Solid Particles
Aloxite
• 1m/s
A 2m/s
• 4m/s
4 6m/s
T 9m/s
Sodium Fluorescein
O 1m/s
A 2m/s
n 4m/s
O 6m/s
V 9m/s
Droplets
DEHS
!_
-------
1 1994) for workplace aerosol sampling. The total inhalable fraction using the IOM inlet was
2 selected for a total human exposure study (Pellizzari et al., 1994) to provide the total body
3 burden for metals (lead and arsenic), by the air exposure route.
4 Similar thoracic penetration conventions have been adopted by ISO, CEN, ACGIH and
5 EPA, each with D50 values of 10.0 Mm (ISO, 1992, CEN, 1993, ACGIH, 1994, and EPA,
6 1987). The EPA definition was based primarily on the data of Chan and Lippmann (1980).
7 The exact shapes of each efficiency curve were mathematically defined by Soderholm (1989)
8 and are slightly different for each convention.
9 The respirable conventions have had D50 values ranging from 3.5 to 5.0 /mi, but a
10 compromise convention has been accepted internationally be several organizations. It has a
11 D50 of 4.0 /mi (Soderholm, 1989). ISO (1992) calls this the "healthy adult respirable
12 convention". Liden and Kenny (1992) discuss the performance of currently available
13 respirable samplers. EPA's emphasis on the 2.5 /xm cutpoint was more closely associated
14 with separating the Fine and Coarse atmospheric aerosol modes, rather than mimicking a
15 respiratory deposition convention. The exact location of this minimum in the atmospheric
16 size distribution is currently under debate. It is noteworthy that ISO (1992) defines a "high
17 risk" respirable convention which is claimed to relate to the deposition of particles in the
18 lungs of children and adults with certain lung diseases. The respirable "high risk"
19 convention has a D50 of 2.4 pm, so it could be identified closely with the EPA samplers
20 having a cutpoint of 2.5 /mi.
21 The PM10 size fraction has become nearly universal for ambient air sampling in the
22 U.S., with the implementation of the 1987 standard (U.S. Environmental Protection Agency,
23 1987a). The setting of performance specifications, even with their limitations, has provided
24 a more consistent PM10 data base, with better definition of the data quality. As additional
25 information becomes available on the sources of biases in aerosol collection methodologies,
26 further characterizations of older methods may be needed to better define the quality of
27 collected data. Factors that affect bias, and especially representativeness, should be
28 identified and their influences determined as a function of particle size. As an example,
29 volatilization losses of nitrates were reported by Zhang and McMurry (1992), while losses
30 for organics were reported by Eatough et al. (1993). Because of the prevalence of these
31 chemical classes in the Fine fraction, the effect of the losses on larger fractions (e.g., PM10,
April 1995 4-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 TSP) would be proportionately smaller and can now be estimated. The losses of larger
2 particles through aerosol inlet sampling lines (Anand et al., 1992) has a substantial influence
3 on Coarse fraction samples. This was demonstrated for the British smoke shade sampler
4 inlet line by McFarland et al. (1982). Inlet losses would be expected to play only a minor
5 role in sampling the Fine particle fraction. Biases in concentration for samplers with large
6 particle cutpoints are exacerbated by the large amount of mass present near the cutpoints and
7 the steep slope of mass versus aerodynamic size. Thus, small changes in cutpoint can give
8 significant and hard-to-predict mass biases.
9 :._
10 4.2.24 Total Suspended Particulates (TSP)
11 The gable roof inlet and sampling system for the TSP high volume sampler have
12 remained essentially unchanged since the sampler's identification as a reference ambient
13 sampling device in 1971 (U.S. Environmental Protection Agency, 1971), The sampling
14 performance (e.g., wind speed and direction sensitivity) was described in detail in the 1982
15 Criteria Document, and shown by McFarland and Ortiz (1979) to collect particles with
16 aerodynamic diameters exceeding 40 /mi. More importantly, its particle collection
17 characteristics were shown to be significantly wind speed (2 to 24 km/hr) and wind direction
18 sensitive. Only minor technical updates have been incorporated in commercially available
19 units, such as in the types of available sequence and elapsed timers (mechanical, electronic)
20 and in the types of flow controllers (mass flow, volumetric). Cassettes are now available
21 that protect the fragile glass or quartz fiber filters during handling and transport. Size
22 fractionating inlets for smaller size cutpoints (e.g., 2.5, 6.0 and 10.0 ^m) and cascade
23 impactors have been developed that can be retro-fitted in place of the gable roof. Similar to
24 the Pb strategy of using the TSP high volume sampler to collect a "total" sample, asbestos
25 sampling utilizes an aerosol inlet that attempts to collect a "total" sample, by using an open-
26 faced filter holder with a conductive inlet cowling. Baron (1993) discusses the potential
27 anisokinetic problems that can occur with such a simple inlet, but notes that the small Stokes
28 number for typical asbestos fibers provides efficiencies close to 100%.
29
30
April 1995 4_H DRAFT-DO NOT QUOTE OR CITE
-------
1 4.2.2.2 Total Inhalable
2 The toxicity of contaminants such as lead pose health concerns as total body burdens,
3 suggesting that penetration of all aerosols inhaled into the nose and mouth must be
4 considered, rather than just thoracic penetration. The TSP sampler for atmospheric lead is
5 considered (U.S. Environmental Protection Agency, 1978) to more closely capture this larger
6 size fraction than would a PM10 counterpart, but was not specifically designed to mimic
7 inhalability. The ISO "inhalable" draft sampling convention (ISO, 1993) is intended to apply
8 to such situations, defining collection of all particles passing the oral/nasal entry planes. The
9 total inhalable cutpoint is currently available only in a personal sampler version. Mark and
10 Vincent (1986) described the development of an inhalable inlet (designated as the IOM)
11 meeting the ISO (1992), CEN (1993) and ACGIH (1994) conventions for inpirable dust.
12 This inlet was improved by Upton et al. (1992) and tested by Mark et al. (1992) and shown
13 to satisfy the ACGIH criteria for wind speeds of 0.5 and 1.0 m/s.
14
15 4.2.2.3 PM10
16 The penetration of ambient aerosols through a size-fractionating inlet to the collection
17 substrate must be characterized over the ranges of operating conditions (typically,
18 meteorology and aerosol types) that may be encountered. The range of conditions currently
19 required by EPA PM10 performance specifications were given in U.S. Environmental
20 Protection Agency (1987b). Ranade et al. (1990) and John (1993) described the required
21 testing, which specifies a controlled flow wind tunnel, mono-dispersed, fluorescently-tagged
22 wet and dry aerosols, and an iso-kinetic nozzle aerosol sampling reference to determine
23 aerodynamic penetration through candidate PM10 inlets.
24 Marple and Rubow (1976) suggested an alternate approach to fluorescent tracer
25 chemistry, using a representative poly-dispersed aerosol and monitoring the size distributions
26 of the challenge aerosol entering and exiting the inlet in a static chamber with an optical
27 particle counter (OPC). Buettner (1990) showed that this technique is only accurate if the
28 OPC particle responses are aerodynamically calibrated to account for factors affecting the
29 optical response, including particle shape and refractive index. Maynard (1993) used this
30 approach to determine the penetration of a respirable cyclone to poly-disperse glass micro-
31 spheres, using the TSI, Inc. Aerodynamic Particle Sizer. John and Wall (1983) noted that
April 1995 4-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 inaccurate inlet sizing results may be obtained using poly-disperse AC test dust, as the result
2 of agglomeration. Kenny and Liden (1991) used the APS to characterize personal sampler
3 inlets, and observed that on theoretical grounds, calm air sampling would be expected to
4 provide unity aspiration efficiencies for particles below about 8 jum. Tufto and Willeke
5 (1982) used an OPC to monitor monodispersed aerosols in a wind tunnel setting to determine
6 the performance of aerosol sampling inlets relative to an iso-kinetic nozzle. Yamada (1983)
7 proposed using electron microscopy to determine the size distributions of poly dispersed
8 particles using manual counting techniques before and after a candidate aerosol separator.
9 Penetration data from this technique were found to be significantly less precise and difficult
10 to interpret compared with data for the same separators using fluorometric methods.
11 The aerosol cutpoint performance of two PM10 samplers that have met the EPA
12 performance specifications are illustrated (see Figure 4-4) by the data for the Andersen 321A
13 and Wedding IP10 high volume sampler inlets at 8 km/hr from Ranade et al. (1990). The
14 data show that the cutpoint requirements, defined as a D50 of 10.0 /zm + 0.5 /xm and
15 mimicking a modeled cutpoint sharpness (ae), were met for each of the tested wind speeds.
&
16 These performance results were verified by repeating the tests in wind tunnels located at two
17 other research facilities. A diagram (U.S. Environmental Protection Agency, 1990) of the
18 two-stage Sierra-Andersen PM10 high volume sampler inlet with a design flowrate of 1.13
19 m3/min is shown in Figure 4-5. The buffer chamber of this inlet serves to dampen the
20 influences of variable wind speeds and directions. Aerodynamic separation occurs as the
21 particle-laden air stream passes through two sets of acceleration nozzles, which deposit the
22 particles larger than PM10 on internal collection surfaces. The PM10 fraction is typically
23 collected by a glass fiber filter. An oiled impaction shim was incorporated into the first
24 stage fractionator of the 321A to minimize reentrainment of deposited particles during field
25 sampling. This modified version (Sierra-Andersen 32IB) was designated as an EPA
26 reference method for PM10 in 1987. A subsequent single-stage fractionator (Sierra-Andersen
27 1200) was developed5 and designated as an EPA reference method, with a D50 of 9.5
28 and a hinged design to facilitate cleaning and oiling of the oiled impaction shim.
29 5Graseby-Andersen, Inc., Atlanta, GA.
April 1995 4_13 DRAFT-DO NOT QUOTE OR CITE
-------
100
S- 80
^
¥
0)
I 60
1
LLJ
40
20
• Wedding IP10
0 Model 321A
J_L
4 5 6 7 8 910 15 20
Aerodynamic Diameter (jj.m)
Figure 4-4. Liquid particle sampling effectiveness curves with solid particle points
superimposed for the Wedding IP10 (•) and the Andersen Samplers Model
321A inlets at 8 km/h.
1
2
3
4
5
6
1
A diagram of the cyclone-based Wedding6 PM10 high volume sampler inlet (U.S.
Environmental Protection Agency, 1990) with a design flowrate of 1.13 m3/min is shown in
Figure 4-6. This inlet uses an omni-directional cyclone to accelerate the particle-laden air
stream to deposit particles larger than PM10 on an oiled collection surface. Two additional
turns are made to alter the flow into a downward trajectory toward the collection filter. A
brush is used to clean the deposited aerosol from the absorber surface through an access port.
This inlet was designated as an EPA reference method for PM10 in 1987.
6Wedding and Associates, Fort Collins, CO.
April 1995
4-14
DRAFT-DO NOT QUOTE OR CITE
-------
VO
0
o
z
o
H
O
c;
o
H
w
o
5«
O
HH
H
W
Figure 4-5. Andersen sampler.
Buffer Chamber
V^ Air Flow
Acceleration Nozzle
Impaction Chamber
Acceleration Nozzle
Impaction Chamber
Vent Tubes
Filter Cassette
Filter
Filter Support
Screen
Motor Inlet
-------
1 The aerosol collection performances for 16.67 1pm PM10 i^618 for tne dichotomous
2 sampler are described by Wedding et al. (1982) and McFarland and Ortiz (1984), and
3 illustrated by the penetration data in Figure 4-7. The variability of the performance as a
4 function of wind speed for the Andersen 321A PM10 inlet is shown in Figure 4-8 from data
5 by McFarland et al. (1984). This is a dramatic improvement over the variability shown by
6 the TSP high volume sampler (McFarland and Ortiz, 1979) for the same speed range. An
7 attempt to simplify the complexity and improve the availability of wind tunnels to test PM10
8 inlets was addressed by Teague et al. (1992), who describe a compact tunnel 6 m long by
9 1.2 m high that is capable of testing inlets against the EPA PM10 specifications.
10 Watson and Chow (1993) noted that the EPA PM10 performance specifications allowed
11 a tolerance range around the D50 that permitted inlets to be undesirably "fine tuned" to
12 provide a cutpoint on the lower or upper end of the range. Since a significant amount of
13 mass in the atmospheric aerosol may be associated with particles in the allowable tolerance
14 range, a "reduction" in reported concentrations could be achieved by simply using a lower
15 (e.g., 9.6 pim) cutpoint inlet that is still within the acceptable D50 range. The biases
16 between acceptable samplers have been apparent in the data from field aerosol comparison
17 studies (e.g., Rodes et al., 1985; Purdue et al., 1986; Thanukos et al., 1992). Most of the
18 reported biases between samplers were less than 10%, although differences of up to 30%
19 were reported. The data suggested that the high volume sampler PM10 inlets based on
20 cyclonic separation (Wedding, 1985) were consistently lower, while those based on low
21 velocity impaction (McFarland et al., 1984) were consistently higher. Sweitzer (1985)
22 reported results of a field comparison of these two high volume sampler types at an industrial
23 location, and reported average biases of 15%. It was noted that this amount of bias was
24 unacceptable for compliance monitoring and more stringent performance requirements should
25 be used. Rodes et al. (1985) observed that the PM10 concentration data from the
26 dichotomous sampler (regardless of the inlet design) gave the most predictable results.
27 Wang and John (1988) were critical of the EPA PM10 performance specification on
28 allowable particle bounce (U.S. Environmental Protection Agency, 1987b), stating that the
29 criteria can lead to a 30% overestimation of mass under worst-case conditions. In a related
30 paper, John et al. (1991) reported that although reentrainment by air flow alone of particles
31 deposited in an aerosol inlet is typically negligible, reentrainment caused from subsequent
April 1995 4-16 DRAFT-DO NOT QUOTE OR CITE
-------
Maintenance Access Port
Housing
Deflector
Spacing
Perfect
Absorber
No-Bounce
Surface
Middle
Tube
Plug
Flow
Y
Vanes
Vane
Assembly
Base
Insect
Screen
Protective
Housing
Aerodynamic
Inlet
Pathway
Aerodynamic Flow
Deflector
Outer Tube
Figure 4-6. Sampling characteristics of two-stage size-selective inlet for liquid aerosols.
April 1995
4-17 DRAFT-DO NOT QUOTE OR CITE
-------
100
80
0)
0
I
*= 40
HI HU
20
0
4 6 8 10 20
Aerodynamic Particle Diameter (jim)
Figure 4-7. Penetration of particles for 16.67 1 pm PM10 inlets.
40
1 particle "bombardment" can be substantial. John and Wang (1991) suggested that particle
2 loading on oiled deposition surfaces can bias the collection 2.2%/gram deposited, and
3 strongly suggested that a periodic cleaning schedule should be required for PM10 inlets.
April 1995
4-18
DRAFT-DO NOT QUOTE OR CITE
-------
100 -
80
60
o
I 40
20
o 2 km/h
A 8 km/h
n 24 km/h
I I
U2 4 6 8 10 20
Aerodynamic Particle Diameter (urn)
Figure 4-8. Collection performance variability as a function of wind speed.
1 The EPA PM10 performance specification program should be considered successful
2 (John, 1993) in providing consistent aerosol collection results during field sampling. As
3 noted by Thanukos et al. (1992), the cases of greatest concern were those where the
4 measured concentrations were near an exceedance level. A review of the current PM10
April 1995
4-19
DRAFT-DO NOT QUOTE OR CITE
-------
1 performance requirements and possible amendments of the existing specifications may be
2 appropriate, given the information base now available.
3
4 4.2.3 Fine Particle Separators
5 4.2.3.1 Cutpoint Considerations
6 Although a particle separation at 2.5 /im has been utilized by the dichotomous sampler
7 for a number of years, the 1987 standard reassessment (U.S. Environmental Protection
8 Agency, 1987a) did not specifically require routine monitoring for Fine particles. It has
9 become apparent (see Chapter 14) that certain health and ecological responses are most
10 strongly correlated with fine particles, significantly smaller than 10 /*m, and their related
11 chemistry. Since the mass of a particle is proportional to the cube of its diameter, larger
12 particles (especially above 10 /*m) can totally dominate the mass of PM10 and TSP samples.
13 The 2.5 /im cutpoint generally occurs near a minimum in the mass distribution, minimizing
14 mass concentration differences between samplers with cutpoint biases. The development of
15 control strategies based on mass concentrations from a smaller cutpoint standard must be
16 carefully constructed, especially if large particle interference problems (e.g., particle bounce)
17 cannot be appropriately minimized. This issue was highlighted by an EPA workshop in
18 May, 1994 that focused on the implications of introducing a "fine" particle standard and
19 possibly changing the current 2.5 /zm cutpoint for fine fraction sampling to 1.0 /un. A
20 background paper by Lundgren and Burton (1994) notes that the size fraction less than 1.0
21 /xm typically contains only 0.1% of the total aerosols by mass for particles less than 100 /mi,
22 while the less than 2.5 /mi fraction contains ~ 1%. By comparison, it was estimated that at
23 least 50% of the aerosol particles by number are less than 1.0 jum, while at least 80% are
24 less than 2.5 pm.
25 Practical considerations would be the time and expense required to develop inlets with
26 1.0 nm cutpoints that meet required specifications and retrofit existing samplers. Given the
27 body of data available at 2.5 /xm, a focused effort may prove practical that defines the
28 characteristics of the particle mass and chemistry between 1.0 and 2.5 /mi. This would add
29 to the technical knowledge base, allow interpretive corrections between cutpoints to be made,
30 and permit continued sampling at 2.5 /mi with a minimum of additional resources.
31 Compositional analysis of the PM2 5 to PM10 coarse fraction at eastern U. S. sites show that
April 1995 4-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 more than one half of anthropogenically -produced sulfates would be lost by adoption of a
2 1.0 /xm cutpoint during the important summer haze period (Eldred et al., 1995). Similar
3 losses would also be suggested at some California sites.
4
5 4.2.3.2 Virtual Impactors
6 The dichotomous sampler utilizes virtual impaction to separate the Fine and Coarse
7 fractions into two separate flowstreams (see Novick and Alvarez, 1987). The calibration of a
8 nominal 2.5 /mi impactor, including wall loss data, is shown in Figure 4-9 (from Loo and
9 Cork, 1988). A virtual impactor has been designed with a 1.0 jum cutpoint (Marple et al.,
10 1989), and for cutpoints as small as 0.12 /-tm (Sioutas et al., 1994). After applying a cross-
11 channel correction factor for the Coarse mode, the mass concentrations of each fraction and
12 total (using a PM10 inlet) can be determined gravimetrically. An inherent consideration with
13 virtual separation is contamination of the Coarse fraction by a portion of the Fine fraction,
14 equivalent to the ratio of the Coarse channel flow to the total flow (typically 10%). This can
15 influence subsequent chemical and physical characterizations, if significant differences exist
16 between the chemistry of each fraction (e.g., acidic Fine fraction and basic Coarse fraction).
17 Stevens et al. (1993) utilized this limited addition of Fine particles to the Coarse fraction to
18 advantage in the SEM analysis of samples collected on Nuclepore filters. The current
19 separator design provides a relatively sharp cutpoint with minimal internal losses. Keeler
20 et al. (1988) showed that the growth of Fine aerosols at elevated relative humidities can
21 significantly alter the ratio of Fine to Coarse collection for the dichotomous sampler. During
22 early morning periods when the humidity approached 100%, an apparent loss of up to 50%
23 of the Fine mass (to the Coarse channel) was observed. They commented that analyzing
24 only the fine fraction of the measured aerosol may not be appropriate, especially for short
25 integration intervals.
26 A high volume (1.13 mVmin) virtual impactor assembly was developed by Marple, Liu
27 and Burton (1990) that can be placed on an existing high volume sampler to permit larger
28 total collections than the dichotomous sampler for chemical speciation by size fraction. By
29 placing a number of virtual impactors in parallel, a separation can be achieved at higher
30 flows, while reducing the total pressure drop. Marple et al. (1993) provide a list of
April 1995 4-21 DRAFT-DO NOT QUOTE OR CITE
-------
I
I
IT)
I
I
U
5
s
April 1995
4-22
DRAFT-DO NOT QUOTE OR CITE
-------
1 commercially available virtual impactors by flowrate and available cutpoints. They also note
2 that virtual separators inherently concentrate the particles in the Coarse fraction (typically by
3 a factor of 10), making them useful as pre-concentrators for sensors with marginal
4 sensitivities. John et al. (1983) found that an oiled Nuclepore filter with a nominal 8 /^m
5 porosity could provide a D50 cutpoint of 2.5 /xm, similar to that of a virtual impactor, if
6 operated at the appropriate face velocity.
7
8 4.2.3.3 Cyclones
9 Cyclones have been used as aerosol separators in personal exposure sampling in
10 occupational settings for many years. Lippmann and Chan (1979) summarized the cyclones
11 for sampling aerosol sizes below 10 /xm, and note that the aerosol penetration through a
12 cyclone can be designed to closely mimic respiratory deposition. An intercomparison of
13 three cyclone-based personal exposure samplers under occupational conditions (concentrations
14 typically > 1 mg/m3) was described by Groves et al. (1994). They reported that even
15 though the cyclones were reportedly design to mimic similar respirable conventions, biases as
16 large as a factor of two were noted, possibly attributable to over-loading problems. Marple
17 et al. (1993) provided a list of commercially available air sampling cyclones, by sampling
18 flowrate and D50 range. Cyclones can be used individually or in a cascade arrangement to
19 provide a size distribution. Hartley and Breuer (1982) describe methods to reduce biases
20 when using a 10 mm (diameter) personal air sampling cyclone, especially as related to
21 cutpoint shifts caused by flowrate changes. Saltzman (1984) provided a similar analysis for
22 atmospheric sampling cyclones. Sass-Kortsak et al. (1993) observed that substantial
23 uniformity-of-deposition problems can occur on the filters downstream of personal sampling
24 cyclones. Wedding (1983) used a cyclone within a high volume aerosol inlet to provide a
25 PM6 cutpoint for ambient sampling that did not allow penetration of particles greater than
26 10.0 nm.
27 The simplicity of cyclones has prompted their use as inlets and subsequent separators in
28 samplers designed to fractionate the aerosol sample for chemical analysis. The "Enhanced
29 Method" employed by EPA for sampling acidic aerosols, uses a glass cyclone with a 2.5 /im
30 cutpoint as the sampler inlet (EPA, 1992). The percent collection as a function of
31 aerodynamic diameter is shown in Figure 4-10 (Winberry et al., 1993). Bering et al. (1990)
April 1995 4-23 DRAFT-DO NOT QUOTE OR CITE
-------
to
o
o
2
o
o
C
O
s
o
»
o
HH
a
IUU
SO
Cp*
- 60
C
o
ts
0)
o 40
O
20
n
C
/
X
0^^^
: (
/
/
y4
,/ !
k !
) ;
;/(.
s
s
^
X"
)
r— ' '
— i •-
1
2 2.5 4
Aerodynamic Diameter (jam)
S
10
Figure 4-10. Percent collection as a function of aerodynamic diameter.
-------
1 describe several validated aerosol systems for sampling carbonaceous particles that utilize
2 cyclones with 2.5 jum cutpoints to sample the Fine fraction on either Teflon or quartz
3 substrates. Spagnolo and Paoletti (1994) describe a dual cyclone ambient aerosol sampler
4 with a 15 /mi inlet (described by Liu and Piu, 1981). This sampler was designed to collect
5 an "inhalable" 0 to 15 /im fraction, an "extra-thoracic"7 0 to 4.0 /mi fraction, and a
6 "respirable" 0 to 2.5 /im fraction. Malm et al. (1994) describe a sampling system with a
7 PMIO inlet and three parallel channels following a 2.5 /mi cutpoint cyclone that was used for
8 the 40 site IMPROVE network. Over 120,000 fine particle filter substrates of Teflon6,
9 nylon and quartz were collected for chemical analysis over a 6 year period.
10
11 4.2.3.4 Impactors
12 Impactors have been developed for a wide range of cutpoints and flowrates. In cascade
13 arrangements (see Section 4.2.7.1.1) with a characterized inlet, impactors provide
14 distributional information over a range of aerodynamic sizes. Impactors used as components
15 of inlets or as in-line fractionators, stop and retain the aerosol on a surface (e.g., oil-soaked,
16 sintered metal or glass) that hopefully provides consistent performance (primarily minimal
17 bounce) over the entire sampling interval. Recovery and analysis of the deposited particles
18 in these situations are usually not considerations. Koutrakis et al (1990) described the design
19 of an 2.1 /mi cutpoint impactor for a single stage annular denuder system that exhibited
20 internal losses of less than 3 %.
21 Marple et al. (1993) noted that the three primary limitations of impactors are particle
22 bounce, overloading of collection stages and interstage losses. Particles can bounce from a
23 stage after impaction if the surface forces are not adequate for their retention. Wang and
24 John (1988) described the effects of surface loading and relative humidity on particle bounce
25 and growth, and noted that if less than 6% of the impact area was covered by deposited
26 particles, particle-to-particle collisions (and bounce) could be neglected. They also showed
27 that ammonium sulfate aerosol growth with increasing humidity resulted in a 25% shift in
28 cutpoint as the relative humidity increased to 64%. Biswas et al. (1987) showed that,
29 especially in low pressure zones, the relative humidity and temperature can change rapidly
30 within a cascade impactor, significantly altering cutpoints and losses. Turner and Hering
31 7The term "extra-thoracic" is most often used to refer to those panicles > 10 ^m.
April 1995 4_25 DRAFT-DO NOT QUOTE OR CITE
-------
1 (1987) noted that the stage substrate materials (Mylar*, stainless steel and glass) with the
2 same grease (Vaseline*) could produce substantially different particle adhesion
3 characteristics. Vanderpool et al. (1987) showed (see Figure 4-11) that using glass fiber
4 filters as impactor surfaces can produce drastically reduced performance as compared to a
5 greased substrate. Markowski (1987) suggested that adding a duplicate (same cutpoint),
6 serial impactor stage can permit reasonable bounce and re-entrainment corrections to be
7 made.
8
9 4.2.4 Sampling Considerations
10 4.2.4.1 Siting Criteria
11 The selection of an aerosol sampling location is partially guided by siting criteria
12 provided as part of the 1987 PM10 regulation (U.S. Environmental Protection Agency,
13 1987c), which provided limited guidance for Pb and PM10 samplers. The details behind
14 these guidelines for PM10 are provided by EPA in a guidance document (U.S. Environmental
15 Protection Agency, 1987d), which relates the physical and chemical characteristics of
16 aerosols to the spatial scales (regional, urban, neighborhood, middle and micro) required to
17 define the influences of sources on various populations. Guidance was also provided on the
18 influences of nearby point, line and area sources on sampling location as a general function
19 of particle size. Only limited information was noted to be available on specific influences of
20 local obstructions and topography (e.g., trees, buildings) on the measured aerosol
21 concentrations. The primary focus was establishment of the degree that a sampling location
22 was representative of a specific scale.
23 The high purchase cost, and occasionally physical size, of aerosol samplers have tended
24 to restrict the number of sampling sites used in air monitoring studies. In an attempt to
25 address the biases resulting from too few aerosol samplers in a field study, a "saturation"
26 sampler approach has been used, utilizing an inexpensive, miniature and battery-powered
27 PM10 sampler that can be deployed at a large number of sites. Phillips et al. (1994) reported
28 the application of this approach using 15 PM10 saturation samplers in conjunction with one
29 dichotomous sampler to study the contribution of diesel emissions to the total paniculate
30 levels in Philadelphia. Although the mean PM10 concentrations of the saturation samplers
April 1995 4-26 DRAFT-DO NOT QUOTE OR CITE
-------
100
O Greased Substrate
D Glass-Fiber Filter
5 10 20
Aerodynamic Particle Diameter
Figure 4-11. Performance of glass fiber filters compared to greased substrate.
1
2
3
4
5
6
7
8
9
10
11
12
13
was essentially identical to that of the dichotomous sampler, the saturation data showed site-
to-site mean differences of as much as 30
4.2.4.2 Averaging Time/Sampling Frequency
The collection frequency for samples to support the EPA PM10 standard has typically
been on an every-6th-day schedule. A statistically-based concern (Shaw et al., 1984) was
raised that infrequent collection increases the coefficient of variation about the overall mean
concentration value. They observed that the variability of computed Fine mass concentration
means increased as the square root of the number of intervals between individual
measurements. Symanski and Rappaport (1994), using time series analyses, described the
influences of autocorrelation and non-stationary behavior in occupational settings on the
concentration distributions constructed from infrequent sampling. They recommended a
random sampling design where a sufficient number of locations are sampled repeatedly over
April 1995 4-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 an adequate period of time to account for the full range of exposure possibilities. Hornung
2 and Reed (1990) described a method of estimating non-detectable (or missing) values to
3 improve the variance about the estimate of the geometric mean, by assuming the
4 concentration distribution is log-normal.
5 Insufficient sample collections can be remedied by more frequent operation of manual
6 samplers. The recent PM10 equivalency designations (see section 4.2.5) of two beta gauge
7 samplers and the TEOM sampler can provide the necessary information, with hourly, rather
8 than daily, resolution. The initial cost of an automated sampler is typically 2-3 times that of
9 a manual, single channel PM10 sampler, but can be offset by the savings in operator labor
10 costs. If the inherent biases described in section 4.2.3.4 for the beta and TEOM samplers
11 can be accommodated (and they are field reliable), these approaches should prove very useful
12 in routine regulatory and research monitoring studies. The potential also exists that the
13 integrating nephelometer may be an acceptable exceedance monitor8, using site specific
14 calibrations relating the measured scattering coefficient, bsp, to Fine aerosol mass
15 concentrations (e.g., Larson et al., 1992).
16 Another consideration for setting the sampling interval concerns the setting of start and
17 stop clock times. Daily 24-h sampling is most often accomplished from midnight-to-
18 midnight, but occasionally from noon-to-noon to either reduce the number of samplers
19 required or reduce operator burden. Sampling locations with highly variable diurnal aerosol
20 concentration patterns (e.g., from night time wood smoke influence or day time traffic dust),
21 or substantial differences between week days and weekend days may require special
22 consideration. These influences can be especially significant for sampling periods less than
23 24 h.
24
25 4.2.4.3 Collection Substrates
26 The selection of a filtration substrate for integrated collection of particles must be made
27 with some knowledge of the expected particle characteristics and a pre-determined analytical
28 protocol. The expected sampled size distribution places a requirement on the porosity of the
29 filter media to effectively trap a reasonably high percentage of the particles with a minimum
30 8A Pollutant Standard Index (PSI) monitor used to estimate when a pre-determined exceedance level has been
31 reached or exceeded, to potentially trigger the operation of an equivalent PM10 gravimetrically-based sampler.
April 1995 4-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 of pressure drop. The most common filter types used in air sampling are fiber and
2 membrane. Fiber filters tend to be less expensive than membrane filters, have low pressure
3 drops, and have high efficiencies for all particle sizes. They are most commonly available in
4 glass fiber, Teflon coated glass fiber and quartz materials. Membrane filters retain the
5 particles on the surface for non-depth analyses (e.g., X-Ray Fluorescence), can have specific
6 porosity's, and are available in a wide variety of materials. Teflon is a popular membrane
7 material because of its inertness, but is 2 to 4 times as expensive as more common materials.
8 Liu et al. (1978) summarize the effective penetration characteristics as a function of particle
9 size and the pressure drops for a wide variety of fiber and membrane filters. Polycarbonate
10 filters with well defined porosity's (e.g., Nuclepore0) have been used in "stacked"
11 arrangements as fine particle separators. John et al. (1983) describe using an 8 ^m porosity
12 filter in series with a back-up filter to effectively provide a 3.5 /xm separation of fine and
13 coarse particles in a small, inexpensive package. Samplers based on this principle were
14 widely used in the early 1980's (Cahill et al., 1990) and their performance under field
15 conditions was shown to be equivalent to later cyclone base PM2 5 samplers in the
16 IMPROVE network.
17 The reactivities of filter substrates with the aerosol have been reported extensively. A
18 common problem with glass fiber filters used on high volume samplers is the basic pH of the
19 glass material and its effective conversion of acid gases to particulate sulfates (e.g., Pierson
20 et al., 1976). Appel et al. (1984) also reported similar conversions of nitrogen oxides to
21 particulate nitrates on glass fiber filters. Witz et al. (1990) reported losses of particulate
22 nitrates, chlorides and ammonium (19, 51 and 65 %, respectively) from quartz fiber filters
23 during storage. No significant losses of sulfates were reported from quartz filters.
24 Similarly, Zhang and McMurry (1992) reported the anomalous loss of fine particle nitrates
25 from Teflon filters and noted that predictive loss theories were insufficiently accurate to
26 permit corrections. Lipfert (1994) also observed that nitrate artifacts on glass fiber filters
27 were difficult to quantify on a routine basis. Measurements of particulate nitrate using nylon
28 filters by the IMPROVE protocols show, however, that such effects are minor except in
29 California (Malm et al., 1994). Eatough et al. (1993) found significant losses of particulate
30 organic compounds on quartz filters due to volatilization, such that the ambient
31 concentrations of particulate carbon may be underestimated substantially. Lipfert (1994)
April 1995 4_29 DRAFT-DO NOT QUOTE OR CITE
-------
1 investigated filter artifacts in a field study in New York, and concluded that positive sulfate
2 artifacts inflated PM10 values from glass fiber filters by 6 jwg/m3. It was noted that the
3 combination of sulfate and nitrate artifacts on glass fiber filters may inflate TSP
4 concentrations by as much as 10 to 20 jug/m3.
5
6 4.2.4.4 Chemical Speciation Sampling
7 The collection of aerosol samples for chemical speciation analysis adds another
8 dimension to the complexity of the sampling protocol (also see Section 4.3 for additional
9 discussion). The simplest approach utilizes a characterized inlet or separator to define a size
10 fraction, provides an aerosol collection substrate compatible with the analytical technique,
11 and collects an adequate quantity of sample for analysis. This approach is applicable for
12 relatively nonreactive and stable components such as heavy metals. An important
13 consideration is the potential reactivity of the sampling substrate with either the collected
14 aerosols or the gas phase. Appel et al. (1984) predicted the effect of filter alkalinity on the
15 conversion of acid gases to sulfates and nitrates. They provided an upper limit estimate on
16 artifact sulfate formation (added mass) for TSP high volume sampling of 8-15 /xg/m3 for a
17 24-h sample.
18 Analyses for semi-volatile organics which are found in both the particle and vapor
19 phases must be collected by adding a vapor trap (e.g., polyurethane foam plug) downstream
20 of the sampling filter. Arey et al. (1987) noted that this arrangement of sequential sampling
21 reservoirs may account for the total mass of organics, but not accurately describe their phase
22 distribution in situ, due to "blow-off" from the filter during sampling. Van Vaeck et al.
23 (1984) measured the volatilization "blow-off" losses of organic species from cascade
24 impactor sampling to be up to 30%, while the loss of total mass was only 10%. McDow and
25 Huntzicker (1990) characterized the face velocity dependence for organic carbon sampling
26 and provided correction models, based on adsorption losses to a backup filter. Turpin et al.
27 (1994) examined organic aerosol sampling artifacts and highlighted the distinction between
28 "organic carbon" and individual organic species. They observed that organic carbon
29 sampled from the atmosphere is unlikely to attain equilibrium between that in the gas phase
30 and that adsorbed on a quartz fiber back-up filter. They also noted that under typical
31 sampling conditions, adsorption is the dominant artifact in the sampling of particulate organic
April 1995 4-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 carbon, and longer sampling periods reduce the percentage of collected material that is
2 adsorbed vapor. It was recommended that collection of aerosols for carbon analyses by
3 made on a pre-fired quartz filter, with estimates of the adsorption artifact made from a quartz
4 filter placed behind a Teflon filter in a parallel sampler.
5 For more highly reactive and unstable species, the recognition of the in situ character
6 of the aerosol in the air must be identified and preserved during all facets of the sampling
7 process to provide a representative and accurate sample. Durham et al. (1978) described a
8 denuder to remove sulfur dioxide while sampling for sub-micron aerosols. Spicer and
9 Schumacher (1979) observed that many artifact reactions may occur if stripping of nitric
10 acid, sulfuric acid and ammonia is not performed during speciated aerosol sampling. Appel
11 et al. (1988) described the various loss mechanisms that apply to the aerosol and vapor
12 phases while sampling for nitric acid. They noted that residence time, surface material
13 compositions, and conditioning prior to sampling were the predominant variables affecting
14 transmission efficiency.
15 The determination of strong acidity for atmospheric aerosols (U.S. Environmental
16 Protection Agency, 1992) describes an "Enhanced" method that recognizes the inter-
17 relationships between the vapor and aerosol phases for each constituent and the potential
18 interferences. An inlet cyclone or impactor is used to provide a 2.5 /*m cutpoint to exclude
19 the higher pH aerosols found in the Coarse fraction. As shown in Figure 4-12, denuders are
20 used in the flowstream which selectively remove gas phase components with minimal,
21 characterized losses of aerosol. Ye et al. (1991) determined the aerosol losses through an 10
22 1pm annular denuder system as a function of particle size. They noted that total particle
23 losses were less than a few percent whether the denuders were coated or uncoated. Forrest
24 et al. (1982) using parallel annular denuders, found aerosol losses of only 0.2-2.2% for 0.3-
25 0.6 jum particles, and 4-5% for 1-2 /mi particles.
26 Filter packs have been developed, consisting of a sandwich of filters and collection
27 media of various types in series, to collect aerosols and selectively trap gases and aerosol
28 volatilization products. Benner et al. (1991) described an annular denuder sampling system
29 using Teflon and nylon filter packs and annular denuders to quantitatively collect the
30 distributed ammonium nitrate, nitric acid and ammonia in the vapor and aerosol species.
31 They observed that volatile nitrates were 71 % ±27% of the total nitrates during the day and
April 1995 4.31 DRAFT-DO NOT QUOTE OR CITE
-------
Filter Pack
Nylon
Teflon
H
+3
d4
1
d3
i
d2
Coupler (Typical) — — [
d1
£
'tk
ff'>r
OBI
nu
v4
;!,(!
8
i
Hj||
yarn
mm
m
4
'ff,
n^**
X5
m
MS
"V?*
*
-tlli
•O
3
o
*k_
+-I
b
f
NH3
V
m
O
-------
1 Vossler et al. (1988) reported the results of improvements in an annular denuder
2 system, including Teflon coating of the internal glass surfaces. They found an apparent
3 particle bounce problem with the cyclone inlets (with or without Teflon coating), and
4 proposed adding an additional in-line, greased impactor. John et al. (1988) found that
5 anodized aluminum surfaces absorb nitric acid efficiently and irreversibly. A number of
6 method comparison studies have been reported for systems utilizing annular denuder/filter
7 pack technologies, including Harrison and Kitto (1990), Sickles et al. (1990), and Benner
8 et al. (1991).
9
10 4.2.4.5 Data Corrections/Analyses
11 Aerosol concentration data are reported in units of mass per volume (e.g., /zg/m3). The
12 current EPA regulations for sampling TSP, PM10 and Pb require that sampler flowrates be
13 controlled and the sampled volumes be standardized to 760 mm Hg and 25 °C. These
14 requirements may pose problems in the interpretation of concentrations from aerosol
15 samplers. The flowrate through inertial impactors should be maintained at "local"
16 temperatures and pressures to retain the separator's aerodynamic calibration. Mass flow
17 controllers may significantly affect the separator flow velocity during large diurnal
18 temperature changes, excessively biasing the resulting cutpoint diameter.
19 Subsequent correction of the sampled aerosol volume to "standard" conditions by
20 mathematically compensating for average meteorological conditions may improperly report
21 the aerosol concentration measurement. If the rationale for aerosol sampling was to mimic
22 respiratory penetration (which occurred at local conditions), a correction after-the-fact may
23 not be appropriate. These corrections are typically small (less than a few percent) except in
24 locations at higher altitudes and those with large diurnal or seasonal temperature changes.
25 The basis for mandating flowrate controller performance for aerosol samplers is sound, but
26 and the subsequent requirements for concentration corrections for temperature and pressure
27 are complex. Although the issue of sampled volume correction for local temperature and
28 pressure is beyond the scope of this document, the scientific bases should be reassessed for
29 aerosol sampling to determine if this requirement is consistent with EPA goals.
30 The matching of aerosol measurement capabilities with data quality requirements is
31 discussed by Baron and Heitbrink (1993). They note that although aerosol sampler precision
April 1995 4.33 DRAFT-DO NOT QUOTE OR CITE
-------
1 can be determined from collocated measurements, field sampling accuracy is more difficult to
2 define. Generation of mono- or polydisperse calibration aerosols are rarely done in field
3 settings because of the complexity of the calibration process. Typically, only the aerosol
4 sampler flowrate accuracy is determined in the field. Biases between the means from
5 collocated aerosol samplers using different separation techniques, may result from sampler
6 operational errors, or from inadequacies in determining the performance specifications during
7 laboratory testing.
8
9 4.2.5 Performance Specifications
10 4.2.5.1 Approaches
11 A significant step in the standardization process for aerosol sampling was the EPA
12 definition (U.S. Environmental Protection Agency, 1987a) of the PM10 size fraction, based
13 on the aerodynamic diameter of particles capable of penetrating to the thoracic region of the
14 respiratory system. This definition was followed by implementation of the PM10 provisions
15 of EPA's Ambient Air Monitoring Reference and Equivalent Methods regulation (U.S.
16 Environmental Protection Agency, 1987b). The format of the latter regulation (see section
17 4.2.5 for specifics) was the adoption of performance specifications for aerosol samplers,
18 based on controlled wind tunnel testing with mono-dispersed aerosols. Controlled laboratory
19 testing is followed by limited field testing, including tests of candidate equivalent methods to
20 demonstrate comparability to designated reference methods. This approach was chosen,
21 rather than the design specification approach taken in 1971 (U.S. Environmental Protection
22 Agency, 1971), which identified the high volume sampler and associated operational
23 procedures as the reference method for Total Suspended Particulates (TSP)9. The 1971
24 regulation had no provisions for the use of alternative or equivalent methods. Subsequent to
25 this design designation, significant problems of the TSP high volume sampler, such as wind
26 speed and direction dependency (McFarland and Rodes, 1979) and off-mode collection (Sides
27 and Saiger, 1976), were reported. These inherent biases complicated the interpretation of
28 TSP concentration data (U.S. Environmental Protection Agency, 1982c) and weakened
29 Subsequent identifications in this section: "TSP" for Total Suspended Particulates by high volume sampler,
30 "PM10" for the fraction less than 10 /xm, "Fine" (capitalized) for the fraction less than 2.5 /mi, and "Coarse" for
31 the fraction between 2.5 and 10 /j.m.
April 1995 4-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 correlations with other measures. The problems were estimated to have induced biases of
2 less than 10% for most situations, but occasionally as high as 30%. The subsequent
3 development of aerosol testing programs for size selective aerosol samplers (e.g., McFarland
4 and Ortiz, 1979; Wedding, 1980; John and Wall, 1983; Ranade et al., 1990; Hall et al.,
5 1992) more rapidly identified weaknesses in existing technologies and facilitated the
6 development of better methods. No reference standard exists for aerosol concentration
7 measurements in air. The calibration of aerosol samplers relies primarily on
8 characterizations under controlled conditions of the sampler sub-systems, including the size
9 selective inlet, sample conditioning and transmission system, the flow control system, and, if
10 used, subsequent size separators, sample collection and storage elements, and sensors and
11 associated electronics. Although the precision of an aerosol sampler is readily obtained by
12 using replicate, collocated samplers, the accuracy can only be estimated by comparison with
13 either designated "reference" samplers or with computations of expected aerosol mass
14 collections. Performance specification limits are used to control the overall aerosol sampling
15 accuracy. As noted by John (1993) the selection of a comprehensive list of sampling
16 elements requiring inclusion and the setting of the performance limits for each element is a
17 difficult task, especially when the range of "real-world" sampling situations is considered.
18 Performance specifications were utilized for the PM10 standard to allow the broadest
19 spectrum of measurement technologies, hopefully encouraging the development of new and
20 better methods. A research program was implemented by EPA in parallel with the 1982
21 Criteria Document to identify the critical specifications and understand the inter-relationships
22 among the parameters influencing the aerosol sampling process. Studies of the influences of
23 factors such as wind velocity, particle character, flow rate stability, particle bounce and wall
24 losses on precision and accuracy substantially advanced the science of large particle
25 sampling. The performance specification approach was a significant improvement over the
26 design specification approach used for the TSP high volume sampler, in that it fostered the
27 development of new information and technologies, and provided for the use of alternative
28 methods. In retrospect, the primary weakness of the design specification approach for the
29 TSP reference method was not the process per se, but the technical inadequacy of the
30 development and testing program that produced the high volume sampler design.
April 1995 4.35 DRAFT-DO NOT QUOTE OR CITE
-------
1 The utilization of a performance specification approach requires that a minimum level
2 of knowledge be available about the measurement process and the associated test procedures.
3 Some significant drawbacks subsequently observed in the performance specification approach
4 for PM10 included the complexity, expense and scarcity of aerosol wind tunnel test facilities,
5 and the difficulty in defining comprehensive specifications that considered all of the nuances
6 of aerosol sampling. Wind tunnel evaluation and limited field tests do not always identify
7 sampler related problems encountered during extended periods of ambient sampling (e.g.,
8 John and Wang, 1991).
9
10 4.2.5.2 Critiques
11 Aerosol sampling research studies since the 1982 Criteria Document have identified a
12 number of factors that influence the precision and accuracy of both wind tunnel sampler
13 performance testing and individual aerosol samplers, demonstrated to meet performance tests.
14 Rodes et al. (1985) and Purdue et al. (1986) showed in field evaluations that PM10 samplers
15 meeting the EPA performance specifications did provide consistent aerosol concentration
16 measurements within 10%, under a variety of sampling situations, but reported that
17 significant biases were evident. The biases were based on expected collections computed
18 from a knowledge of aerosol inlet penetration by particle size and the size distributions of
19 ambient aerosol mass by particle size. They also noted that sampler precisions (coefficients
20 of variation) were better than +10%, with several samplers better than ±5%.
21 Mark et al. (1992) reviewed the attributes of wind tunnel testing, and noted that tests
22 using controlled conditions are a necessity to determine whether an aerosol sampler meets a
23 basic cadre of established performance specifications. Hollander (1990) suggested that
24 sampler performance criteria should be evaluated in controlled outdoor tests, given the
25 inability of wind tunnels to accurately mimic the influences of outdoor meteorological
26 conditions on sampling. The current EPA PM10 performance testing requires field tests to
27 demonstrate sampler precision and flow rate stability, and the comparability of equivalent
28 methods to designated reference methods. The stringency of such tests are highly dependent
29 on the sampling location chosen, local aerosol sources, the existing meteorology and the
30 season.
April 1995 4-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kenny and Liden (1989) noted that the EPA PM10 sampler performance specifications
2 (U.S. Environmental Protection Agency, 19875) provided inadequate consideration for
3 defining the uncertainty in each parameter, and suggested that bias mapping approaches be
4 considered. This approach relates the allowable precision of a parameter to the critical
5 values of expected bias that just meet the specifications. Botham et al. (1991) recommended
6 that the wind tunnel test system duplicate the expected field sampling scenarios as closely as
7 possible, including characteristic flow obstructions. They described the wind tunnel testing
8 of personal aerosol samplers mounted on an anthropogenically consistent (e.g., breathing,
9 heated) mannequin. Hoffman et al. (1988) and John et al. (1991) described the adverse
10 influence of internal surface soiling on aerosol collection performance during extended field
11 operation, and noted that the existing EPA PM10 performance specifications only considered
12 clean samplers.
13 Significant new innovations in aerosol sensing technologies, that met the PM10
14 performance specification and earned designations as equivalent methods (see section 4.2.6)
15 have occurred since the 1982 Criteria Document. These indirect10 methods include
16 automated beta attenuation monitors (e.g., Merrifield, 1989; Wedding and Weigand, 1993),
17 and the automated Tapered Element Oscillating Microbalance (TEOM) technology
18 (Patashnick and Rupprecht, 1991). These designations added automated sampling capabilities
19 to the previously all-manual list of sampling methods. Recent field tests of both the beta and
20 TEOM methodologies suggest that biases compared to gravimetrically-based samplers may
21 exist that were not identified by the EPA performance test requirements. Arnold et al.
22 (1992) provide data suggesting that the mass concentration data from a Wedding beta gauge
23 averaged 19% lower than a collocated Wedding PM10 gravimetric sampler. Several
24 researchers, including Hering (1994) and Meyer (1992), have suggested that the TEOM
25 method can exhibit biases (not identified by performance testing) caused by excessive heating
26 (and desiccation) of the sampled aerosol due to operation at an elevated reference
27 temperature (30 or 50 °C) during the measurement process. Devising comprehensive
28 performance specifications and test procedures for aerosol samplers, given the complexities
29 of aerosol mechanics, is a demanding task.
30 10An alternate technology used instead of direct gravimetric analysis to infer mass concentrations from developed
31 relationships.
April 1995 4.37 DRAFT-DO NOT QUOTE OR CITE
-------
1 The size-selective, gravimetrically-based, 24-h manual aerosol concentration
2 measurement has been the mainstay of compliance sampling for at least two decades.
3 Although several new sensor technologies have been designated as Equivalent methods for
4 PM10 by EPA, no superior technology has been developed that is a better reference method
5 than that based on collection of a discreet aerosol sample followed by gravimetric analysis.
6 Improvements have been made since 1982 in the accuracy and precision of integrated,
7 manual aerosol sampling. Some of the most significant advances have occurred in aerosol
8 size separation technologies, improved performance characterization test methods, and
9 speciation sampling techniques.
10 As discussed by Lippmann (1993), there may be no threshold for health responses
11 down to the lowest aerosol concentrations. This implies that the precision and lower
12 detection limit requirements will continue to be important for aerosol measurements across
13 the concentration spectrum. These factors become even more critical as the size fraction of
14 interest becomes smaller and fewer total particles are collected. At low concentrations
15 (especially with small size fractions), normally insignificant factors can become important
16 contributors to biases. Witz et al. (1990) reported rapid and substantial losses of nitrates,
17 chlorides and ammonium ion (19, 65 and 51%, respectively) from quartz high volume
18 sampler filters during storage periods of one week prior to analyses. Transformations can
19 also occur on glass fiber substrates during sampling, as reported by Sickles and Hodson
20 (1989) for the rapid conversion of collected nitrites to nitrates in the presence of ozone.
21 Zhang and McMurry (1992) showed that nearly complete evaporative losses of Fine particle
22 nitrate can occur during sampling on Teflon filters. Lioy et al. (1988), in a study using
23 PM10 samplers, reported 25-34% lower concentration values resulting from losses of glass
24 fibers from the filter to the filter holder gasket during sampling. Feeney et al. (1984)
25 reported weight gains in Teflon filters used in contaminated ring cassettes, that posed
26 significant problems for light aerosol loadings. Grinshpun et al. (1993) suggest that if
27 unavoidable changes in the aerosol occur during sampling, development of a model that
28 permits back-calculation of the in situ characteristics is required.
29
30
April 1995 4-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.2.6 Reference and Equivalent Method Program
2 Ambient air PM10 measurements are used (among other purposes) to determine whether
3 defined geographical areas are in attainment or non-attainment with the national ambient air
4 quality standards (NAAQS) for PM10. These measurements are obtained by the States in
5 their state and local air monitoring station (SLAMS) networks as required under 40 CFR Part
6 58. Further, Appendix C of Part 58 requires that the ambient air monitoring methods used
7 in these EPA-required SLAMS networks must be methods that have been designated by the
8 EPA as either reference or equivalent methods.
9 Monitoring methods for paniculate matter (i.e., PM10) are designated by the EPA as
10 reference or equivalent methods under the provisions of 40 CFR Part 53, which was
11 amended in 1987 to add specific requirements for PM10 methods. Part 53 sets forth
12 functional specifications and other requirements that reference and equivalent methods for
13 each criteria pollutant must meet, along with explicit test procedures by which candidate
14 methods or samplers are to be tested against those specifications. General requirements and
15 provisions for reference and equivalent methods are also given in Part 53, as are the
16 requirements for submitting an application to the EPA for a reference or equivalent method
17 determination. The distinction between reference and equivalent methods is a technical one.
18 On one hand, it provides for detailed, explicit specification of a selected measurement
19 technology for reference methods. On the other hand, it allows alternative (including
20 innovative and potentially improved) methodologies for equivalent methods, based only on
21 meeting specified requirements for functional performance and for comparability to the
22 reference method. For purposes of determining attainment or non-attainment with the
23 NAAQS, however, the distinction between reference and equivalent methods is largely, if not
24 entirely, immaterial.
25 Under the Part 53 requirements, reference methods for PM10 must be shown to use the
26 measurement principle and meet the other specifications set forth in 40 CFR 50, Appendix J.
27 They must also include a PM10 sampler that meets the requirements specified in Subpart D of
28 40 CFR 53. Appendix J specifies a measurement principle based on extracting an air sample
29 from the atmosphere with a powered sampler that incorporates inertial separation of the PM10
30 size range particles followed by collection of the PM10 particles on a filter over a 24-h
31 period. The average PM10 concentration for the sample period is determined by dividing the
April 1995 4.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 net weight gain of the filter over the sample period by the total volume of air sampled.
2 Other specifications are prescribed in Appendix J for flow rate control and measurement,
3 flow rate measurement device calibration, filter media characteristics and performance, filter
4 conditioning before and after sampling, filter weighing, sampler operation, and correction of
5 sample volume to EPA reference temperature and pressure. In addition, sampler
6 performance requirements in Subpart D of Part 53 include wind tunnel tests for "sampling
7 effectiveness" (the efficacy of the PM10 particle size separation capability) at each of three
8 wind speeds and "50 percent cutpoint" (the accuracy of the primary 10-micron particle size
9 separation). Field tests for sampling precision and flow rate stability are also specified. In
10 spite of the instrumental nature of the sampler, this method is basically a manual procedure,
11 and all designated reference methods for PM10 are therefore defined as manual methods.
12 Equivalent methods for PM10, alternatively, need not be based on the measurement
13 principle specified in Appendix J nor meet the other Appendix J requirements. Instead,
14 equivalent methods must meet the "sampler" performance specifications set forth in Subpart
15 D of Part 53 and demonstrate comparability to a reference method as required by Subpart C
16 of Part 53. The provisions of Subpart C specify that a candidate equivalent method must
17 produce PM10 measurements that agree with measurements produced by collocated reference
18 method samplers at each of two field test sites. For this purpose, agreement means a
19 regression slope of 1 +0.1, a regression intercept of 0 +5 /ig/m3, and a correlation >0.97.
20 These requirements allow virtually any type of PM10 measurement technique, and therefore
21 an equivalent method for PM10 may be either a manual method or a fully automated
22 instrumental method (i. e., analyzer).
23 As of this writing, the EPA has designated seven reference methods and three
24 equivalent methods for PM10, as listed in Table 4-1. The reference methods include four
25 methods featuring high-volume samplers from two manufacturers, with one using a cyclone-
26 type size separator and the others using an impaction-type separator. The other reference
27 methods include a low-volume sampler (from a third manufacturer), a low-volume sampler
28 featuring a secondary size separation at 2.5 microns (dichotomous sampler), and a medium-
29 volume, non-commercial sampler. The three designated equivalent methods are all
30 automated PM10 analyzers and include two operating on the beta-attenuation principle and
31 one based on a tapered element oscillating microbalance (TEOM™). It should be noted that
April 1995 4-40 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-1. EPA-DESIGNATED REFERENCE AND EQUIVALENT METHODS FOR PM10
Method No.
Identification
Description
Type
Date
RFPS-1087-062
Wedding & Associates3 PM
Critical Flow High-Volume
Sampler.
10 High-volume (1.13 m3/min) sampler with cyclone-
type PM10 inlet; 203 x 254 cm (8 x 10 in) filter.
Manual reference
method
10/06/87
RFPS-1287-063
RFPS-1287-064
Sierra-Andersen5 or General Metal
Works0 Model 1200 PM10 High-
Volume Air Sampler System
Sierra-Andersenb or General Metal
Worksc Model 321-B PM,0 High-
Volume Air Sampler System
High-volume (1.13 m3/min) sampler with
impaction-type PM10 inlet; 203 x 254 cm (8 x 10
in) filter.
High- volume (1.13 nrVmin) sampler with
impaction-type PM10 inlet; 203 x 254 cm (8 x 10
in) filter. (No longer available.)
Manual reference
method
Manual reference
method
12/01/87
12/01/87
RFPS-1287-065
Sierra-Andersenb or General Metal
Works0 Model 321-C PM10 High-
Volume Air Sampler System
High-volume (1.13 m3/min) sampler with
impaction-type PM10 inlet; 203 x 254 cm (8 x 10
in) filter. (No longer available.)
Manual reference
method
12/01/87
RFPS-0389-071
Oregon DEQ Medium Volume
PM10 Sampler
Non-commercial medium-volume (110 L/min)
sampler with impaction-type inlet and automatic
filter change; two 47-mm diameter filters.
Manual reference
method
3/24/89
RFPS-0789-073
T)
H
6
O
z
s
o
c
s
w
o
h-H
H
W
EQPM-0990-076
EQPM-1090-079
Sierra-Andersenb Models SA241 or Low-volume (16.7 L/min) sampler with impaction- Manual reference 7/27/89
SA241M or General Metal Works0 type PM10 inlet; additional particle size separation method
Models G241 and G241M PM10
Dichotomous Samplers
Andersen Instruments'5 Model
FH62I-N PM10 Beta Attenuation
Monitor
Rupprecht & Patashnickd TEOM
Series 1400 and Series 1400a PM-
10 Monitors
at 2.5 micron, collected on two 37-mm diameter
filters.
\
Low-volume (16.7 L/min) PM10 analyzers using
impaction-type PM10 inlet, 40 mm filter tape, and
beta attenuation analysis.
Low-volume (16.7 L/min) PM10 analyzers using
impaction-type PM10 inlet, 12.7 mm diameter
filter, and tapered element oscillating microbalance
analysis.
Automated
equivalent method
Automated
equivalent method
9/18/90
10/29/90
-------
> TABLE 4-1 (cont'd). EPA-DESIGNATED REFERENCE AND EQUIVALENT METHODS FOR PM10
_, ====1^=:^===:=:^=^============:^=^=^====^=^=:===========:^=^=========^===:^=:^===^====^=^=====^==:^===:^=^=========:^=^=^=:^==^=^=^=^=^=^=
~ Method No. Identification Description Type Date
Jg EQPM-0391-081 Wedding & Associates'1 PM10 Beta Low-volume (16.7 L/min) PM10 analyzer using Automated 3/5/91
^> Gauge Automated Particle cyclone-type PM10 inlet, 32 mm filter tape, and equivalent method
Sampler beta attenuation analysis.
RFPS-0694-098 Rupprecht & Patashnickd Partisol Low-volume (16.7 L/min) PM10 samplerwith Manual reference 7/11/94
Model 2000 Air Sampler impaction-type inlet and 47 mm diameter filter. method
-f.
N)
H
6
o
Z!
S
O
d
s
m
n
HH
H
W
-------
1 although these latter three automated PM10 analyzers may be capable of providing continuous
2 or semi-continuous PM10 concentration measurements, only 24-h average PM10
3 measurements are recognized as approved under their equivalent method designations.
4
5 4.2.7 Determination of Size Distribution
6 The determination of aerosol size distributions can be a powerful research tool when
7 studying source contributions and transformation processes. A number of techniques are
8 available as described by texts such as Willeke and Baron (1993) to make near real-time,
9 single particle aerosol measurement in addition to cascade impactors.
10
11 4.2.7.1 Cascade Impactors
12 In cascade applications, the aerosol is impacted and trapped onto a series of removable,
13 coated substrates (e.g., greased foils), including a final total stage collection on a filter for
14 gravimetric analysis. Marple et al. (1993) list over 30 single stage and cascade impactors
15 that are either commercially available or still commonly used. The design and calibration of
16 a miniature eight-stage cascade impactor for personal air sampling in occupational settings is
17 described by Rubow et al. (1987), operating at 2.0 1pm. Evaluations of the most commonly
18 used cascade impactor systems have been reported by Vaughan (1989) for the Andersen MKl
19 and MK2 7-stage cascade impactors, Marple et al. (1991) for the 10-stage Micro-Orifice
20 Uniform Deposit Impactor (MOUDI), and Wang and John (1988) and Hillamo and
21 Kauppinen (1991) for the 6-stage Berner, low pressure cascade impactor. The smallest
22 particle stages of these impactors can have very small diameter jets and/or very low total
23 pressures to achieve the sub-micron separations. The MOUDI impactor has 2000 holes on
24 the lowest cutpoint stage. Raabe et al. (1988) describe an 8 stage cascade slit impactor with
25 slowly rotating impactor drums instead of flat plates. This arrangement, in combination with
26 a PIXIE analyzer, permitted aerodynamic sizing of elemental components, with temporal
27 resolution. The skill and care required in the operation of cascade impactors suggests that
28 they are research rather than routine samplers.
29 The importance of the aerosol calibration of a cascade impactor is illustrated by
30 Vaughan (1989) in Figure 4-13, which compares the experimental data with the
31 manufacturer's calibrations and indicates biases as large as 1.0 /*m. Marple et al. (1991)
April 1995 4.43 DRAFT-DO NOT QUOTE OR CITE
-------
2J
8
O
d
tr)
O
n
t—i
H
W
100
80
60
W
0
'€ 40
co
20
n Liquid Particles
• Solid Particles
10 9
Stage —=
Cut-Points
8
StageNumber
7654
1 Inlet
0.01 0.1 1
Aerodynamic Particle Diameter (|im)
Figure 4-13. Aerosol calibration of a cascade impactor.
10
100
-------
1 provided a similar type of stage calibration for the MOUDI impactor, and included data on
2 the internal particle losses (see Figure 4-9). These loss data showed that an improperly
3 designed inlet to the impactor, combined with the inertial and interception losses of the larger
4 particle sizes, can substantially bias the first stage collections. This was also demonstrated
5 for the inlet to the Andersen impactor by McFarland et al. (1977).
6 Cascade impactors can be used to construct distributions of mass and speciated
7 constituents as a function of aerodynamic diameter. These distributions can be constructed
8 graphically or using matrix inversion techniques (e.g., Crump and Seinfeld, 1982,
9 Wolfenbarger and Seinfeld, 1990). Marple et al. (1993) notes that impactor stage
10 calibrations which do not demonstrate sharp cutoffs can cause significant between-stage
11 sizing errors if not accommodated. John et al. (1990) measured distributions over the 0.08
12 to 16 fj,m range for mass and inorganic ions for several sites in Southern California. They
13 identified the standard Coarse mode, and two separate and previously unreported modes in
14 the 0.1 to 1.0 ftm range. This range was referred to by Whitby (1978) as a single
15 "accumulation" mode. John et al. (1990) described a "condensation" mode at 0.2 ±0.1 fj.m
16 containing gas phase reaction products, and a "droplet" mode at 0.7 ±0.2 /zm which grows
17 from the "condensation" mode by the addition of water and sulfates. Fang et al. (1991)
18 described the effects of flow-inducted relative humidity changes on the sizing of acid aerosols
19 in the MOUDI impactor. They noted that it may not be possible to measure size
20 distributions of small (less than about 0.2 to 0.5 |um) particles with impactors at relative
21 humidities exceeding 80%.
22
23 4.2.7.2 Single Particle Samplers
24 Aerosol size distribution data are useful in studies of particle transport and
25 transformation processes, source characterization, and particle sizing and collection device
26 performance. In addition to cascade impactors, a number of real time or near real time
27 sizing instruments are available and described in texts such as Willeke and Baron(1993).
28 While cascade impactors provide distributions in terms of aerodynamically sized mass, single
29 particle sampling devices can produce optically sized distributions as a function of particle
30 number (count), with surface area and volume distributions computed during the data
31 reduction, assuming spherical particles. Particle density and shape information as a function
April 1995 4.45 DRAFT-DO NOT QUOTE OR CITE
-------
1 of size are required to convert from volume distributions to an estimated mass basis.
2 Individual particle sizing and counting instruments are generally limited to a particle
3 detection range of a decade or so, but several devices can overlap to cover the range of
4 approximately 0.001 to 10 pm. The principle of detection of an instrument restricts the
5 particle sizes which can be detected. For example, instruments using electrical mobility
6 analysis are limited to particle sizes less than about 1 jum. Optical methods are typically
7 used to measure particles larger than about 0.1 to 0.3 /mi. Inlet and transport system losses
8 of coarse particle above about 2 /mi, prior to the sensing volume, must be factored into
9 reported size distributions.
10 The three most commonly used single particle sampler types are aerodynamic particle
11 sizers, electrical mobility analyzers and optical particle counters (OPC's). Aerodynamic
12 particle sizers use laser doppler anemometry to measure the velocity of particles in a jet.
13 The acceleration of the particle is related to the aerodynamic particle diameter. This
14 technique is typically applied to particles larger than about 0.5 /mi. In electrical mobility
15 analysis, aerosol with a known charge distribution flows through an electric field. The
16 particles migrate according to their mobility which can be related to size. The original TSI
17 electrical aerosol analyzer (EAA) performed this separation in an integrated manner over the
18 total size distribution and detected the particles by unipolar diffusion charging. A more
19 versatile approach is the differential mobility analyzer, or DMA (Knutson and Whitby, 1975;
20 Liu et al., 1978) is able examine a narrow slice of the size distribution in an equilibrium
21 charge state, detected by a condensation nucleus counter (CNC). Differential mobility
22 analyzers have been employed in pairs (Tandem Differential Mobility Analyzer, or TDMA)
23 to examine particle characteristics such as NH3 and H2SO4 reaction rates (McMurry et al.,
24 1983), and the sensitivity of the size distributions of Los Angeles aerosol to relative humidity
25 (McMurry and Stolzenburg, 1989). The latter research used the first DMA to select particles
26 of known mobility from the input aerosol, a humidification system to condition the selected
27 particles, and the second DMA to determine mobility changes. Optical particle counters pass
28 a jet of aerosol through an optical system. Light scattered from individual particles is
29 detected and the signal in processed in a multi-channel analyzer. Discreet signals are
30 counted and sorted by intensity by optical size. An example forward-scattering counter with
31 an open sensing volume (for use on aircraft), is the Particle Measuring Systems, Inc., FSSP-
April 1995 4-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 300, which can provide high resolution (31 channel) count distributions over the size range
2 of 0.3 to 20 urn (Rader and O'Hern, 1993). Gebhart (1993) described currently available
3 OPC's and their counting efficiencies over a range of diameters.
4 Single particle samplers have common considerations.
5 Calibration: They are calibrated with reference aerosol either by the manufacture or by the
6 user. If the properties of the aerosol measured are quite different than the calibration, the
7 indicated size distribution may be quite different than actual distribution. Brockman et al.
8 (1988) demonstrated that the APS calibration can vary significantly with the type of test
9 aerosol, and showed substantial response biases between oleic acid and polystyrene latex
10 spheres above 10 /mi. Wang and John (1989) described a procedure to correct the APS
11 response for aerosol particle density. Particle shape can also provide serious sizing errors,
12 and specific calibrations are needed for particles with shape factors significantly different
13 from unity (spherical). Yeh (1993) commented that the calculated geometric standard
14 deviations (#„) determined by the EAA and DMA are generally larger than 1.3, even if the
&
15 correct value is significantly closer to unity. Woskie et al. (1993) observed, as did Willeke
16 and Degarmo (1988), that optical particle counting devices must be appropriately calibrated
17 using realistic aerosols, especially for low concentration applications. Harrison and Harrison
18 (1982) suggested that the ratio of fine particle mass concentration to optical scattering
19 extinction will be more variable when a significant contribution is made by irregular (shaped)
20 particles - an event likely to occur when the mean mass diameter exceeds 1 pirn.
21
22 Particle Concentration Effects: Gebhart (1993) noted that the response of single particle
23 counters may be influenced by extremely high particle concentrations. Wake (1989) and
24 Heitbrink et al. (1991) described the coincidence problems of the APS when sampling high
25 total particle concentrations, especially for sizes greater than 1 /an. Baron et al. (1993)
26 reported that the concentration levels giving 1 % coincidence in an aerodynamic particle sizer
27 for 0.8, 3 and 10 jum particles, are the relatively low values of 558, 387 and 234
28 particles/cm3, resectively. Optical particle counters experience coincidence errors (two
29 particles are detected as a single particle) and counter saturation at high particle
30 concentrations. Hinds and Kraske (1986) described the performance of the PMS, Inc. LAS-
31 X and noted a sizing accuracy of ±2 channel widths, and coincidence errors of less than
April 1995 4.47 DRAFT-DO NOT QUOTE OR CITE
-------
1 10% for concentrations below 10,000 particles/cm3. Clearly, typical particle concentrations
2 found in the atmosphere may produce significant errors if sample dilution is not utilized.
3
4 4.2.8 Automated Sampling
5 Automated methods to provide measures of aerosol concentrations in the air have
6 existed for decades in an attempt to provide temporal definition of suspended particles with a
7 minimum labor expense. Some of the automated samplers described in the 1982 Criteria
8 Document (e.g., British Smoke Shade and AISI tape samplers), were indicator measures of
9 aerosol concentration, using calibrations relating aerosol concentrations to reflected or
10 absorbed light. Tape samplers were used in the U. S. primarily as exceedance (index)
11 monitors.
12 The beta attenuation and integrating nephelometer techniques were described in 1982
13 primarily as research methods. Refinements to the beta gauge sampling approach and the
14 addition of the Tapered Element Oscillating Microbalance (TEOM) principle have resulted in
15 their designation as equivalent methods for PM10.
16 Presently there are no commercially available, automated high volume (> 1 m3/min
17 flowrate) aerosol samplers, excluding the possibility of the timed operation of an array of
18 manual samplers. The physical size of such a sampling system using 8 x 10 inch filters is
19 impractical. The dichotomous sampler is currently the only low volume, gravimetrically-
20 based sampler, commercially available in an automated version.
21
22 4.2.8.1 TEOM
23 The Tapered Element Oscillating Microbalance (R & P, Inc.) sensor was described by
24 Patashnick and Rupprecht (1990), and consists of an oscillating tapered tube with a filter on
25 its free end (see the diagram in Figure 4-14). The change in mass of the filter and collected
26 aerosol produces a shift in the oscillation frequency of the tapered tube that can be directly
27 related to mass. Rupprecht et al. (1992) suggested that the filter can be archived after
28 sampling for subsequent analysis. The sampler inlet has a PM10 cutpoint and operates at
29 16.67 1pm. A flow splitter samples a 3 1pm portion of this flow to be filtered. Since the
30 fraction of volatile species (e.g., water, nitrates, organics) in the aerosol is a function of
31 ambient temperature, the TEOM heats the inlet air stream to a constant 50 °C to keep
April 1995 4-48 DRAFT-DO NOT QUOTE OR CITE
-------
Flow
Flow
Sampling Head
Heated Air Inlet
Filter Cartridge
Tapered Element
Electronic
Feedback System
Microprocessor
to Flow Controller
Figure 4-14. TEOM
1 moisture in the vapor phase. The mass transducer is also heated to 50 °C to stabilize the
2 measurement process. Operation with the flow stream heated to a lower temperature (e.g.,
3 30 °C) is possible, but care must be taken to avoid moisture condensation that will confound
April 1995 4.49 DRAFT-DO NOT QUOTE OR CITE
-------
1 the measurement. The transducer is also heated to 50 °C to stabilize the mass measurement.
2 A factory calibration regression is used to electronically correct the computed mass from the
3 TEOM to that measured by a reference PM10 sampler.
4 As previously described, several researchers, including Cahill et al. (1994), Hering
5 (1994) and Meyer et al. (1992) have reported that the modification of the aerosol by the
6 elevated operating temperature appears to have a significant effect (loss) on mass
7 concentration. Meyer et al. (1992) collocated a TEOM sampler with an PM10 SA1200
8 gravimetric sampler in Mammoth Lakes, CA during a winter heating season (heavy wood
9 stove usage). The regressions between the TEOM and PM10 sampler gave strong
10 correlations (r2 > 0.98), with slopes of 0.55 for operation at 50 °C, and 0.66 for operation
11 at 30 °C. The negative bias of the TEOM was attributed primarily to losses of semi-volatile
12 organics from the filter. Cahill et al. (1994) reported that the TEOM showed biases on the
13 order of 30% low and poor correlations with PM10 samplers in dry, dusty conditions. The
14 reasons for this discrepancy were unknown. The field comparison data of Patashnick and
15 Rupprecht (1990) showed near unity (1±0.06) regression slopes for the TEOM with the
16 Wedding IP10 and Sierra-Andersen dichotomous samplers in El Paso, TX and Birmingham,
17 AL. Since aerosol composition is highly dependent on local sources and meteorology,
18 volatilization losses could be expected to be site- and season-dependent. More data are
19 needed to determine the implications of these problems on the ability of the TEOM to be
20 used in a regulatory setting.
21
22 4.2.8.2 Beta Gauge
23 The Andersen FH 62I-N beta attenuation sampler was described by Merrifield (1989)
24 and uses a 30 mCi Krypton-85 source and detector to determine the attenuation caused by
25 deposited aerosols on a filter (see diagram in Figure 4-15). To improve the stability over
26 time, a reference reading is periodically made of a foil with an attenuation similar to that of
27 the filter and collected aerosol. The Wedding beta attenuation sampler was described by
28 Wedding and Weigand (1993) and uses a 100 mCi 14C source. Both samplers have inlets
29 with a PM10 cutpoint, with the Andersen sampler operating at 16.67 1pm and the Wedding at
30 18.9 1pm. The filter material is contained on a roll and advances automatically on a time
31 sequence, or when a preset aerosol loading is reached. An automatic beta gauge sampler
April 1995 4.50 DRAFT-DO NOT QUOTE OR CITE
-------
\s
?
Tl
H
I
O
O
o
H
O
cj
O
H
W
O
MC -
CC -
C -
Measuring Chamber
Compensation Chamber
Chamber for Dust Precipitation
and Measurement
30 m Ci KR-85 Source
Filter Feed Spool
Filter Takeup
High-Voltage Power Supply
Bit
I/O
50-Pin
Connector
jmperature / Pressure
Rotary Vane Pump
V24/RS232
Figure 4-15. Beta gauge.
n
-------
1 was also described by Spagnolo (1989), using a 15 /j,m inlet and a 14C source. The
2 calibration of a beta gauge is site specific, and a calibration regression must be processed
3 electronically to provide accurate mass readings. Rupprecht et al. (1992) suggested that the
4 closer link between deposited mass and frequency shift for the TEOM principle should
5 provide less site-specific response, compared to the aerosol compositional sensitivity of the
6 beta gauge technique.
7 Arnold et al. (1992) provided data over a 2 year period in Denver, CO for the mass
8 concentration regression data from a Wedding beta gauge, showing a range of correlations
9 (r2 from 0.72 to 0.86), varying by sampler and season. The authors suggested that
10 installation of a newer technology beta gauge accounted for the higher correlations, but noted
11 that unexplained outliers resulted in poorer than expected results. The regression slopes
12 between the two sampler types showed that the beta gauge averaged 19% lower than a
13 collocated Wedding PM10 gravimetric sampler. Field data from Wedding and Weigand
14 (1993) at two sites (Fort Collins, CO and Cleveland, OH) using the same samplers produced
15 regressions exhibiting strong correlations (r2 = 0.99) with no apparent outliers and a
16 composite slope of 1.00. Arnold et al. (1992) operated the PM10 high volume samplers on
17 the required every-6th-day schedule and the beta attenuation monitors continuously, and
18 noted that only 22.5% of the exceedance days, as measured by the beta monitor, were
19 operational days for the high volume samplers.
20
21 4.2.8.3 Nephelometer
22 The integrating nephelometer is commonly used as a visibility monitor, measuring the
23 light scattered by aerosols, integrated over as wide a range of angles as possible. A
24 schematic diagram of the integrating nephelometer is shown in Figure 4-16 (from Hinds,
25 1982). The measured scattering coefficient of particles, bsp, can be summed with the
26 absorption coefficient, b , and the comparable coefficients for the gas phase, to compute the
27 overall atmospheric extinction coefficient, bext. The atmospheric extinction has been related
28 to visibility as visual range. The particle scattering coefficient is particle size dependent, as
29 shown by Charlson et al. (1968) in Figure 4-17, while the absorption coefficient is relatively
30 independent of size. The field calibration of nephelometers has historically been based on
31 the refractive index of Freon-12 (and occasionally carbon dioxide), but newer calibration
April 1995 4-52 DRAFT-DO NOT QUOTE OR CITE
-------
VO
O
O
Z
O
H
O
g
H
W
O
90
n
»—i
3
Power
Supply
Flash Tube
Power Supply
Clean Air
Purge
Aerosol
Outlet
A
Photomultiplier
Tube
Scattering
Volume
Collimating
Disks
Aerosol
Inlet
Clean Air
Purge
Amplifier
Recorder
Figure 4-16. Integrating nephelometer.
-------
10
o
tl
H
6
o
as
s
co
CO
E «
o £
o
I"E
*•"•» "*^
T- "•"»«, C
» T- t)
"o
"b
S
10
Diameter (urn)
4.00
s
w
Figure 4-17. Particle-scattering coefficient as a function of particle size.
O
>— I
H
W
-------
1 procedures using atomized sugar aerosols have been proposed (Horvath and Kaller, 1994) as
2 more environmentally conscious. Another class of optical instruments which do not resolve
3 the particle size spectrum, measure the laser light scattered from a volume of aerosol
4 containing a number of particles. Gebhart (1993) described devices such as the MIE,
5 Inc.11. MINIRAM, often used in portable applications to estimate real-time aerosol
6 concentrations. Woskie et al. (1993) described the calibration and performance of a
7 MINIRAM (using the manufacturer's calibration) against gravimetric borate concentrations,
8 and found significant biases (a regression slope = 4.48).
9 The relative insensitivity of the nephelometer to particles above ~ 2 /xm provides for
10 poor correlations with PM10 mass. Larson et al. (1992) showed strong correlations (r2 =
11 0.945) between bsp and Fine fraction mass (see Figure 4-19) for a woodsmoke impacted
12 neighborhood near Seattle, WA, with a slope of 4.89 m2/g. They noted that this slope fell
13 within the range of values reported by others and was predicted by Mie scattering theory.
14 The slope of the Larson et al. (1992) data could be compared with other site-specific
15 calibrations, such as the data of Waggoner and Weiss (1980), which gave a composite slope
16 of 3.13 m2/g, characterized by the authors as representative of a "wide range" of sites.
17 Lewis (1981) provided an analysis of the relationships of the features of the ambient size
18 distribution to bsp. The inlet air stream to the nephelometers for the latter data was heated
19 from 5 to 15 °C above background. Rood et al. (1987) conducted a controlled comparison of
20 the influence of aerosol properties on bsp in Riverside, CA, and reported a regression slope
21 against fine mass (defined as less than 2.0 /mi) of 2.1 m2/g with an r2 value of 0.92. In this
22 experiment the relative humidity for bsp determinations was controlled to less than 35% and
23 the gravimetric filter substrate was nylon. They attributed the smaller than normal slope
24 reading to possible nitrate evaporation from the filtered aerosol and artifact reactions with the
25 nylon substrate material.
26 The data scatter in Figure 4-18 (if assumed to be typical of such comparisons) would
27 suggest that fine particle mass concentration estimates from bsp values were typically within 5
28 to 7 /ig/m3 of the gravimetrically determined values. To be useful as a surrogate measure
29 for mass concentration, the site-specific nephelometer calibration should be valid for a wide
30 range of situations, especially during episodes where the concentration levels approach or
31 "Bedford, MA.
April 1995 4.55 DRAFT-DO NOT QUOTE OR CITE
-------
2.2
2.0
1.8
1.6
1.4 -
1.2 -
1.0 -
0.8 -
0.6 -
0.4 ~
0.2 ~
0
Lake Forest Park
Weekly Average Values
January 17,1991 to December 19,1991
0
Slope = 4.89 m /g
R2= 0.945
5 10 15 20 25 30 35 40
PM25(ngm3)
Figure 4-18. Correlation of bsp and fine fraction mass.
45
1 exceed an action limit. The scattergram of bsp versus fine particle mass provided by Rood
2 et al. (1987), showed much greater variability, with a given bsp value providing an estimated
3 20 to 25 /xg/m3 concentration range. They noted that metastable H2O contributed 5 to 20%
4 of the total particle light scattering coefficient, especially during the late afternoon and early
5 evening. The precisions and biases of the dependent and independent variables between bsp
6 and Fine mass concentration are not constants, since at least one factor - moisture content of
7 the aerosol - affects both measures. The gravimetric sample filters are typically equilibrated
8 to a specific relative humidity range (e.g., 40 to 60%) to normalize the tare weighings.
April 1995
4-56
DRAFT-DO NOT QUOTE OR CITE
-------
1 Sloane (1986) and others have noted that light scattering from particles is not solely a
2 function of mass, but very dependent on a summation of the scattering coefficients of each
3 specie. The scattering cross section of a particle is dependent on the water content, and
4 hence the relative humidity in situ.. Pre-heating of the inlet air of the nephelometer
5 normalizes the response to water content, but biases the reading relative to the in situ case.
6 Sloane (1986) also gave the computed and measured scattering coefficients for ammonium
7 sulfate, and noted that chemical interactions can cause a two-fold variation in scattering
8 response to a change in the mass of hygroscopic constituents. It was also observed that the
9 light scattering efficiency of an aerosol such as ammonium acid sulfate is not a constant, but
10 varies with the overall aerosol composition. Eldering et al. (1994) developed and validated a
11 predictive model for bsp in Southern California. This model used composite size
12 distributions constructed from a TSI, Inc.12 EAA, a PMS, Inc.13 LAS-X and a Climet,
13 Inc.14 multi-channel OPC, and filter-based estimates of refractive indices for ammonium
14 sulfate, ammonium nitrate, organic carbon, elemental carbon and residual aerosol mass
15 concentrations as independent variables. The quality of their comparisons with nephelometer
16 data, suggested that this approach could be used to test models that predict visual range from
17 source emissions. Further research is needed to determine the role of the integrating
18 nephelometer as a predictor of Fine particle mass concentrations.
19
20 4.2.9 Specialized Sampling
21 4.2.9.1 Personal Exposure Sampling
22 The application of aerosol measurement technologies to smaller and less obtrusive
23 samplers have resulted in devices used as fixed-location indoor aerosol samplers and personal
24 exposure monitors (PEMs) worn on the body to estimate exposure. The reduction in
25 physical size of personal aerosol sampling systems to reduce participant burden sometimes
26 results in poorer aerosol collection performance as compared to the outdoor counterparts.
27 Wiener and Rodes (1993) noted that personal sampling systems generally have poorer
28 precisions than outdoor aerosol samplers, due to the smaller sampler collections (from lower
29 ^Minneapolis, MN.
30 "Boulder, CO.
31 14Redlands, CA.
April 1995 4.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 flowrates) and poorer flow controllers. Wallace (1994) reported biases for the Particle Total
2 Exposure Assessment Methodology study averaging a factor of two between personal
3 exposure measurements and fixed location PM10 concentrations. He was unable to
4 completely account for the biases, but attributed portions to proximity to indoor sources, a
5 difference in inlet cutpoints (11.7 /urn vs 10.0 jum) and the collection of aerosols from the
6 "personal cloud" caused by body dander. Rodes et al. (1991) showed that the ratio of
7 personal to indoor aerosol measurements for the EPA PTEAM study appeared to be log-
8 normally distributed with a median value of 1.98 and an unexpectedly high value of 3.7 at
9 the 90th ("most exposed") percentile. Ingham and Yan (1994) suggested that the
10 performance of a personal aerosol sampling inlet in an isolated mode (without mounting on a
11 representative humanoid bluff body) can result in substantial under-sampling for larger
12 particles. The relationship between measured aerosol exposure at some external location on
13 the body and actual uptake through oral and nasal entry is very complex.
14 Buckley et al. (1991) described the collection efficiency of an MSP, Inc.15 personal
15 aerosol sampler at 4.0 1pm as shown in Figure 4-19. They evaluated this sampler in a field
16 comparison study with collocated PM10 high volume and dichotomous samplers. The
17 precision for the personal sampler was found to be very good (CV = ±3.2%) with strong
18 correlations (r2 = 0.970) with the dichotomous samplers. Lioy et al. (1988) described a
19 similar comparison for a 10 1pm Air Diagnostics and Engineering, Inc.16 indoor air
20 sampler, with a PM10 inlet characterized by Marple et al. (1987). Correlations against the
21 PM10 dichotomous sampler were also described as very strong (r2 > 0.970), but noted a
22 substantial bias caused by the loss of fragments from indoor air sampler's glass fiber filters.
23 They recommended that exposure studies using samplers that collect small total volumes
24 should utilize filters with greater integrity, such as Teflon. Colome et al. (1992) describe an
25 indoor/outdoor sampling study using an impactor [characterized by Marple et al. (1987)] with
26 a PM10 cutpoint that had duplicate impactors with the same cutpoint in series. This
27 sequential arrangement, in combination with a coating of 100 /xl of light oil, was used to
28 minimize particle bounce at 4.0 1pm for 24 h period.
29 15Minneapolis, MN.
30 16Naples, ME.
April 1995 4-58 DRAFT-DO NOT QUOTE OR CITE
-------
100
80
'o
it
LJU
"o
o
60
20
1
10
Aerodynamic Particle Diameter (jim)
Figure 4-19. Collection efficiency of the MSP personal aerosol sampler.
1 Personal aerosol sampler systems have typically been characterized as burdensome
2 (excessive weight, size, noise). The success of passive detector badges for gaseous
3 contaminants has recently prompted research into passive aerosol samplers. Brown et al.
4 (1994) described a prototype aerosol sampler utilizing electrostatic charge to move the
5 particles to a collection substrate. They noted that preliminary results are encouraging, but
6 the effective sampling rate and size-selectivity of the sampler was dependent on the electrical
7 mobility of the aerosol. This posed calibration problems for real aerosols with a distribution
8 of electrical mobility's. Hollander (1992) described a passive pulsed-corona sampler that has
9 similar collection characteristics as a PM10 inlet, with only modest wind speed dependence.
10 The performance characterization of PEMs has been considered for occupational
11 settings by Kenny and Liden (1989), who reviewed the ACGIH, National Institute for
12 Occupational Safety and Health (NIOSH), and EPA PM10 aerosol sampler performance
13 programs. They proposed that an international consensus be reached on the basic principles
April 1995 4.59 DRAFT-DO NOT QUOTE OR CITE
-------
1 underlying the experimental protocols for testing personal samplers, as an essential
2 prerequisite to the setting of standards. An ISO working group has made progress in
3 developing such a consensus (Kenny, 1992). As EPA becomes more focused on exposure
4 assessment and personal exposure sampling, it will become even more important for the
5 agency to consider establishing performance specifications for personal aerosol samplers.
6 Models have become powerful tools in understanding aerosol behavior in the vicinity of
7 personal exposure samplers. This is demonstrated by particle trajectory models that can
8 predict the influences of the geometries and flow field on aerosol capture and losses (e.g.,
9 Okazaki and Willeke, 1987, Ingham and Yan, 1994, and Tsai and Vincent, 1993). These
10 models have not only permitted more rapid design changes to accommodate new cutpoints
11 and flowrates, but have added insights as to the influence of air flow obstructions on
12 sampling efficiencies. Vincent and Mark (1982) suggested that there is a critical particle
13 trajectory that determines whether a particle is sampled or rejected by an inlet worn on the
14 body. An extension of this model applicable to personal exposure sampling by Ingham and
15 Yan (1994) suggested that testing the performance of a personal aerosol sampling inlet in an
16 isolated mode (without mounting the inlet on a representative bluff body) can result in under-
17 sampling for larger particles by a factor of two. Validation of this model may explain a
18 portion of the bias reported by Wallace et al. (1994) between personal and indoor sampler
19 measurements.
20
21 4.2.9.2 Receptor Model Sampling
22 Receptor modeling has become an established tool to relate ambient concentrations of
23 pollutants to major source categories, by apportioning the components in collected ambient
24 aerosol samples using complimentary source "signatures". Various approaches developed for
25 constructing source/receptor relationships were described by Henry et al. (1984), who also
26 provided a review of modeling fundamentals. They listed the advantages and disadvantages
27 of multivariate models and discussed multi-collinearity problems associated with the presence
28 of two or more sources with nearly identical signatures. Javitz et al. (1988) described the
29 basic Chemical Mass Balance (CMB) approach and showed the influence of the variance in
30 identifying a component in the source signature sample on the projected apportionment.
31 Dzubay et al. (1984) described aerosol source and receptor collection schemes that permitted
April 1995 4-60 DRAFT-DO NOT QUOTE OR CITE
-------
1 the separation of ambient samples into Fine and Coarse fractions for mass, elemental and
2 volatile carbon, and metals analyses. Stevens and Pace (1984) suggested the addition of
3 Scanning Electron Microscopy to permit additional categorization using x-ray diffraction
4 analysis. The most widely used aerosol receptor model is the EPA CMB 7.0 model
5 described by Watson et al. (1990). This paper describes the structure of the model and
6 computer code and the data requirements to evaluate the validity of the estimates. Numerous
7 papers have been published describing the applications of receptor models to the
8 apportionment of the sources of aerosols, with the receptor modeling conference summary by
9 Watson et al. (1989) descriptive of the state-of-the-art.
10 Stevens et al. (1993) described (see Figure 4-20) a modified dichotomous sampler with
11 a PM10 inlet, two Fine channels operating at 15 1pm and one Coarse channel operating at 2.0
12 1pm, designated as the Versatile Air Pollution Sampler (VAPS). The additional Fine fraction
13 channel permitted sampling on a 47 mm Teflon filter for elemental analysis and a 47 mm
14 quartz filter for carbon speciation (elemental and volatile). A Nuclepore filter was used on
15 the Coarse channel for Scanning Electron Microscopy (SEM) evaluation and energy
16 dispersive x-ray diffraction analysis for selected particles.
17
18 4.2.9.3 Particle Acidity
19 An emphasis was placed on sampling sulfuric acidic aerosols in the 1982 Criteria
20 Document. This was followed by a number of research efforts (e.g., Ferm, 1986; Koutrakis
21 et al., 1988) to identify and study the in situ rate reactions, develop sampling strategies to
22 representatively remove the acid particle from the air, identify the co-existing reactive
23 species (e.g., ammonia, nitric acid aerosol, aerosol sulfates and nitrates), and protect the
24 collected aerosol prior to analysis. A "Standard" and an "Enhanced" method were
25 subsequently described (U.S. Environmental Protection Agency, 1992) for the determination
26 of aerosol acidity using annular denuder technology. The "Standard" method did not account
27 for potential interferences from nitric acid, ammonium nitrate aerosol, or other ammonium
28 salts. The "Enhanced" method added an additional denuder prior to filtration, with nylon
29 and treated glass fiber backup filters to account for these species. These sampling
30 technologies utilized either an inlet impactor or a cyclone with 2.5 /^m cutpoints to sample
April 1995 4.51 DRAFT-DO NOT QUOTE OR CITE
-------
- o-
Q> _^ *^ E
S,2tS§
CO 3 0)0-
E
II
N w
ll
o"-
Q. CD
if
o
00
co
CO
0)
I
i
Of)
April 1995
4-62
DRAFT-DO NOT QUOTE OR CITE
-------
1 the Fine fraction. This technology has recently been extended to other reactive aerosol
2 systems, including semi-volatile organics (e.g., Vossler et al., 1988). Bennett et al. (1994)
3 describe a PM2 5 cyclone-based, filter pack sampling system designed fine particle network
4 sampling and acidity measurements, as part of the Acid MODES program. The sampler
5 operated at 8.8 1pm, and was designed to selectively remove ammonia, speciate gas and
6 particle phase sulfur compounds, as well as collect gas phase nitric acid. An intercomparison
7 of 18 nitric acid measurement methods was reported by Hering et al. (1988), who noted that
8 measurements differed by as much as a factor of four and biases increased as nitric acid
9 loadings increased. In general the filter pack systems reported the highest acidity
10 measurements, while the denuder-difference techniques reported significantly lower
11 measurements. Benner et al. (1991) in a comparison of the SCENES filter pack sampler
12 with a denuder-based sampler found excellent agreement between sampler types for both
13 nitric acid and total nitrates. They attributed the close agreement to limited positive artifact
14 formations, since the test field site had high nitric acid gas to paniculate nitrate ratios. John
15 et al. (1988) noted that internal aluminum sampler surface denude nitric acid, and describe
16 the design of an aluminum denuder for the inlet of a commercially available dichotomous
17 sampler to quantitatively remove nitric acid for extended periods.
18 Brauer et al. (1989) describe the design of a miniature personal sampler to collect acid
19 aerosols and gases. A significant finding was the lower than expected personal acidity
20 levels, attributed to the "personal cloud" production of ammonia by the body. Personal
21 exposure levels of acid aerosols were reported to be lower than indoor measurements.
22
23
24 4.3 ANALYSIS OF PARTICIPATE MATTER
25 The interest in the composition of aerosol particles lies in the areas of: (1) explaining
26 and inventorying the observed mass, (2) establishing the effect of aerosols on health and
27 welfare, and (3) attributing ambient aerosols to pollution sources. While any compositional
28 measurement will address one or more of these goals, certain methods excel for specific
29 tasks. In general, no single method can measure all chemical species, and comprehensive
30 aerosol characterization programs use a combination of methods to address complex needs.
31 This allows each method to be optimized for its objective, rather than be compromised to
April 1995 4-63 DRAFT-DO NOT QUOTE OR CITE
-------
1 achieve goals unsuitable to the technique. Such programs also greatly aid quality assurance
2 objectives, since confidence may be placed in the accuracy of a result when it is obtained by
3 two or more methods on different substrates and independent samplers.
4 In the sections that follow, some of the more commonly used methods that address the
5 goals state above are described. The sections are designed to be illustrative rather than
6 exhaustive, since new methods are constantly appearing as old methods are being improved.
7 These chemical analysis methods for the following section are divided into four categories:
8 1) mass, 2) elements, 3) water-soluble ions, and 4) organics. Material balance comparing
9 the sum of the chemical species to the PM mass concentrations show that elements, water
10 soluble ions, and organic and elemental carbon typically explain 65 to 85% of the measured
11 mass and are adequate to characterized the chemical composition of measured mass for filter
12 samples collected in most urban and non-urban areas. Some of these chemical analysis
13 methods are non-destructive, and these are preferred because they preserve the filter for
14 other uses. Methods which require destruction of the filter are best performed on a section
15 of the filter to save a portion of the filter of other analyses or as a quality control check on
16 the same analysis method. Table 4-2 identifies the elements and chemical compounds
17 commonly found in air using these methods with typical detection limits.
18 Less common analytical methods, which are applied to a small number of specially-
19 taken samples, include isotopic abundances (Jackson, 1981; Currie, 1982; Hirose and
20 Sugimura, 1984); mineral compounds (Davis, 1978, 1980; Schipper et al., 1993); and
21 function groups (Mylonas et al., 1991; Palen et al., 1992; 1993; Allen et al., 1994). Recent
22 advances in infrared optics and detectors have resulted in the quantitative determination of
23 the major functional groups (e.g., sulfate, nitrate, aliphatic carbons, carbonyl carbons,
24 organonitrates, and alcohols) in the atmospheric aerosol (Allen et al., 1994). The advantages
25 of functional analysis in source apportionment are that the number of functional groups is
26 much less than the number of organic compounds to be classified. The cited references
27 provide information on sampling and analysis methods for these highly-specialized methods.
28 The following section focuses on:
29 • Physical analysis of elements and single particle size, shape, and composition,
30
31 • Wet chemical analysis of anions and cations, and
April 1995 4-64 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4.2. INSTRUMENTAL DETECTION LIMITS FOR
PARTICLES ON FILTERS
Minimum Detection Limit in ng/m3a
Species
Be
Na
Mg
Al
Si
P
S
Cl
K
Ca
Sc
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Y
Zr
Mo
Pd
ICP/
AESM
0.06
NA
0.02
20
3
50
10
NA
NA
0.04
0.06
0.3
0.7
2
0.1
0.5
1
2
0.3
1
42
50
25
NA
NA
0.03
0.1
0.6
5
42
AA
FlameM
2d
0.2d
0.3
30
85
100,000
NA
NA
2d
ld
50
95
52
2
1
4
6d
5
4
1
52
100
100
NA
NA
4
300
1000
31
10
AA
Furnaceb
0.05
< 0.05
0.004
0.01
0.1
40
NA
NA
0.02
0.05
NA
NA
0.2
0.01
0.01
0.02
0.02
0.1
0.02
0.001
NA
0.2
0.5
NA
NA
0.2
NA
NA
0.02
NA
INAAb'f
NAh
2
300
24
NA
NA
6,000
5
24
94
0.001
65
0.6
0.2
0.12
4
0.02
NA
30
3
0.5
0.2
0.06
0.4
6
18
NA
NA
NA
NA
FIXES
NA
60
20
12
9
8
8
8
5
4
NA
3
3
2
2
2
NA
1
1
1
1
1
1
1
2
2
NA
3
5
NA
XRFC
NA
NA
NA
5
3
3
2
5
3
2
NA
2
1
1
0.8
0.7
0.4
0.4
0.5
0.5
0.9
0.8
0.6
0.5
0.5
0.5
0.6
0.8
1
5
ICb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
ACb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
TORb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
April 1995
4-65
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4.2 (cont'd). INSTRUMENTAL DETECTION LIMITS FOR
PARTICLES ON FILTERS
Minimum Detection Limit in ng/m3a
Species
Ag
Cd
In
Sn
Sb
I
Cs
Ba
La
Au
Hg
Tl
Pb
Ce
Sm
Eu
Hf
Ta
W
Th
U
Cl-
NO3-
S04=
NH4+
ICP/
AESM
1
0.4
63
21
31
NA
NA
0.05
10
2.1
26
42
10
52
52
0.08
16
26
31
63
21
NA
NA
NA
NA
AA
FlameM
4
1
31
31
31
NA
NA
8d
2,000
21
500
21
10
NA
2,000
21
2,000
2,000
1,000
NA
25,000
NA
NA
NA
NA
AA
Furnaceb
0.005
0.003
NA
0.2
0.2
NA
NA
0.04
NA
0.1
21
0.1
0.05
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
INAAb-f
0.12
4
0.006
NA
0.06
1
0.03
6
0.05
NA
NA
NA
NA
0.06
0.01
0.006
0.01
0.02
0.2
0.01
NA
NA
NA
NA
NA
PIXEg
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
3
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
XRFC
6
6
6
8
9
NA
NA
25
30
2
1
1
1
NA
NA
NA
NA
NA
NA
NA
1
NA
NA
NA
NA
ICb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
50
50
50
NA
ACb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
50
TORb
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
April 1995
4-66
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4.2 (cont'd). INSTRUMENTAL DETECTION LIMITS FOR
PARTICLES ON FILTERS
Minimum Detection Limit in ng/m
,3a
ICP/ AA AA
Species AESb-d FlameM Furnace5 INAAb-f PIXR8 XRFC ICb ACb TORb
oc
EC
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
100
100
"Minimum detection limit is three times the standard deviation of the blank for a filter of 1 mg/cm2 areal
density.
ICP/AES = Inductively Coupled Plasma with Atomic Emission Spectroscopy.
AA = Atomic Absorption Spectrophotometry.
PIXE = Proton Induced X-ray Emissions Spectrometry.
XRF = Non-Dispersive X-ray Fluorescence Spectrometry.
INAA = Instrumental Neutron Activation Analysis.
1C = Ion Chromatography.
AC = Automated Colorimetry.
TOR = Thermal Optical Reflectance.
bConcentration is based on the extraction of 1/2 of a 47 mm quartz-fiber filter in 15 ml of deionized-distilled
water, with a nominal flow rate of 20 L/min for 24-h samples.
cConcentration is based on 13.8 cm2 deposit area for a 47 mm ringed teflon-membrane filter, with a nominal
flow rate of 20 L/min for 24-h samples with 100 sec radiation time.
dHarman (1989).
Ternandez (1989).
f01mez (1989).
SEldred (1993).
hNot Available.
1 • Organic analysis of organic compounds and elemental/organic carbon.
2
3 4.3.1 Mass Measurement Methods
4 Paniculate mass concentration is the most commonly made measurement on aerosol
5 samples. It is used to determine compliance with PM10 standards and to select certain
6 samples for more detailed, and more expensive, chemical analyses. As noted in Section 2,
7 the beta attenuation and inertial microbalance methods have been incorporated into in situ
8 measurement systems which acquire real-time mass measurements. Gravimetric analysis is
9 used almost exclusively to obtain mass measurements of filters in a laboratory environment.
10 U.S. Environmental Protection Agency (1976) and Watson et al. (1989a) have published
11 detailed procedures for mass analyses associated with 20.32 cm x 25.40 cm fiber filters, but
12 the guidance for other types of filters used for chemical analyses is less well documented.
April 1995 4.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gravimetry measures the net mass on a filter by weighing the filter before and after
2 sampling with a balance in a temperature- and relative humidity-controlled environment.
3 PM10 reference methods require that filters be equilibrated for 24 h at a constant (within
4 +5%) relative humidity between 20 and 40% and at a constant (within +3 °C) temperature
5 between 15 and 30 °C. These are intended to minimize the liquid water associated with
6 soluble compounds and to minimize the loss of volatile species. Nominal values of 30% RH
7 and 15 to 20 °C best conserve the particle deposits during sample weighing.
8 Balances used to weigh 20.32 cm x 25.40 cm filters from high volume PM10 samples
9 must have a sensitivity of at least 100 /ig. Balances used for medium volume PM10 samples
10 should have a sensitivity of at least 10 ;wg, and those used for low-volume PM10 samples
11 should have a sensitivity of at least 1 /*g. Modifications to the balance chamber are
12 sometimes needed to accommodate filters of different sizes. All filters, even those from
13 high-volume PM10 samplers, should be handled with gloved hands when subsequent chemical
14 analyses are a possibility.
15 Balance calibrations should be established before and after each weighing session using
16 Class M and Class S standards, and they should be verified with a standard mass every ten
17 filters. Approximately one out of ten filters should be re-weighed by a different person at a
18 later time. These re-weights should be used to calculate the precision of the measurement as
19 outlined by Watson et al. (1989b).
20 Feeney et al. (1984) examined the gravimetric measurement of lightly loaded membrane
21 filters and obtained excellent precision and accuracy. The sensitivity of the electrobalance is
22 about ± 0.001 mg, though tolerances on re-weights of Teflon-membrane filters are typically
23 ± 0.010 mg. The main interference in gravimetric analysis of filters results from
24 electrostatic effects. Engelbrecht et al. (1980) found that residual charge on a filter could
25 produce an electrostatic interaction between the filter on the pan and the metal casing of the
26 electrobalance. This charge can be removed by exposing the filter to a radioactive polonium
27 source before and during sample weighing.
28 Beta attenuation methods have been applied in the laboratory as well as in the field, and
29 the results are comparable to those of gravimetric measurements. The precision of
30 beta-gauge measurements has been shown to be ± 5 /xg/m3 or better for counting intervals of
31 one minute per sample, which translates into ± 32 /xg/filter for 37 mm diameter substrates.
April 1995 4-68 DRAFT-DO NOT QUOTE OR CITE
-------
1 This is substantially higher than the ± 6 jig/filter precision determined by gravimetric
2 analysis using an electrobalance (Feeney et al., 1984). Jaklevic et al. (1981) found
3 equivalent accuracy and precision for both techniques as they were used in that study.
4 Courtney et al. (1982) found beta attenuation and gravimetric mass measurements to differ
5 by less than + 5%. Patashnick and Rupprecht (1991) examine results from TEOM samplers
6 operated alongside filter-based PM10 samplers and Barnes et al. (1988) and Shimp (1988)
7 report comparisons with beta attenuation field monitors; these comparisons all show good
8 agreement for mass measurements.
9
10 4.3.2 Physical Analysis
11 The most common interest in elemental composition derives from concerns about health
12 effects and the utility of these elements to trace the sources of suspended particles.
13 Instrumental neutron activation analysis (INAA), photon-induced x-ray fluorescence (XRF),
14 particle-induced x-ray emission (PIXE), atomic absorption spectrophotometry (AAS),
15 inductively-coupled plasma with atomic emission spectroscopy (ICP/AES), and scanning
16 electron microscopy with x-ray fluorescence (SEM/XRF) have all been applied to elemental
17 measurements of aerosol samples. AAS and ICP/AES are also appropriate for ion
18 measurements when the particles are extracted in deionized-distilled water (DDW). Since air
19 filters contain very small particle deposits (20 to 100 jug/cm2), preference is given to methods
20 that can accommodate small sample sizes. XRF and PIXE leave the sample intact after
21 analysis so that it can be submitted to additional examinations by other methods. Excellent
22 agreement was found for the intercomparison of elements acquired form the XRF and PIXE
23 analyses (Cahill, 1980). The analytical measurement specifications of air filter samples for
24 the different elemental analysis is shown in Table 4.2.
25
26 4.3.2.1 X-Ray Fluorescence (XRF) of Trace Elements
27 In x-ray fluorescence (XRF) (Dzubay and Stevens, 1975; Jaklevic et al., 1977; Torok
28 and Van Grieken, 1994), the filter deposit is irradiated by high energy x-rays that eject inner
29 shell electrons from the atoms of each element in the sample. When a higher energy electron
30 drops into the vacant lower energy orbital, a fluorescent x-ray photon is released. The
31 energy of this photon is unique to each element, and the number of photons is proportional to
April 1995 4-69 DRAFT-DO NOT QUOTE OR CITE
-------
1 the concentration of the element. Concentrations are quantified by comparing photon counts
2 for a sample with those obtained from thin-film standards of known concentration.
3 Emitted x-rays with energies less than =4 KeV (affecting the elements sodium,
4 magnesium, aluminum, silicon, phosphorus, sulfur, chlorine, and potassium) can be absorbed
5 in the filter, in a thick particle deposit, or even by large particles in which these elements are
6 contained. Very thick filters also scatter much of the excitation radiation or protons, thereby
7 lowering the signal-to-noise ratio. For these reasons, thin membrane filters with deposits in
8 the range of 10 to 50 pig/cm2 provide the best accuracy and precision for XRF.
9 XRF methods can be broadly divided into two categories: wavelength dispersive x-ray
10 fluorescence (WDXRF), which utilizes crystal diffraction for observation of fluorescent
11 x-rays, and energy dispersive x-ray fluorescence (EDXRF), which uses a silicon
12 semiconductor detector. The WDXRF method is characterized by high spectral resolution,
13 which minimizes peak overlaps. It requires high power excitation to overcome low
14 sensitivity, resulting in excessive sample heating and potential degradation. Conversely,
15 EDXRF features high sensitivity but less spectral resolution, requiring complex spectral
16 deconvolution procedures.
17 XRF methods can be further categorized as direct/filtered excitation, where the x-ray
18 beam from the tube is optionally filtered and then focused directly on the sample, or
19 secondary target excitation, where the beam is focused on a target of material selected to
20 produce x-rays of the desired energy. The secondary fluorescent radiation is then used to
21 excite the samples. The direct/filtered approach has the advantage of delivering higher
22 incident radiation flux to the sample for a given x-ray tube power, since about 99% of the
23 incident energy is lost in a secondary fluorescence. However, the secondary fluorescence
24 approach, produces a more nearly monochromatic excitation that reduces unwanted scatter
25 from the filter, thereby yielding better detection limits.
26 XRF is usually performed on Teflon-membrane filters for sodium, magnesium,
27 aluminum, silicon, phosphorus, sulfur, chlorine, potassium, calcium, titanium, vanadium,
28 chromium, manganese, iron, cobalt, nickel, copper, zinc, gallium, arsenic, selenium,
29 bromine, rubidium, strontium, yttrium, zirconium, molybdenum, palladium, silver,
30 cadmium, indium, tin, antimony, barium, lanthanum, gold, mercury, thallium, lead, and
31 uranium.
April 1995 4-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 A typical XRF system is schematically illustrated in Figure 4-21. The x-ray output
2 stability should be within ±0.25% for any 8-h period within a 24-h duration. Typically,
3 analyses are controlled, spectra are acquired, and elemental concentrations are calculated by
4 software on a computer that is interfaced to the analyzer.
5 Separate XRF analyses are conducted on each sample to optimize detection limits for
6 the specified elements. A comparison of the minimum detectable limits of Teflon-membrane
7 and quartz-fiber filters is listed in Table 4-3. Figure 4-22 shows an example of an XRF
8 spectrum.
9 Three types of XRF standards are used for calibration, performance testing, and
10 auditing: 1) vacuum-deposited thin-film elements and compounds (Micromatter); 2) polymer
11 films (Dzubay et al., 1981); and 3) National Institute of Science and Technology (NIST,
12 formerly NBS) thin-glass films. The thin film standards cover the largest number of
13 elements and are used to establish calibration curves, while the polymer film standards are
14 used to verify the accuracy of the thin film standards. The NIST standards are used to
15 validate the accuracy of the calibration curves. NIST produces the definitive standard
16 reference materials, but these are only available for the species of aluminum, silicon,
17 calcium, iron, cobalt, copper, manganese, and uranium (SRM 1832), and silicon, potassium,
18 titanium, iron, zinc, and lead (SRM 1833). One or more separate Micromatter thin-film
19 standards are used to calibrate the system for each element.
20 Sensitivity factors (number of x-ray counts per jug/cm2 of the element) are determined
21 for each excitation condition. These factors are then adjusted for absorption of the incident
22 and emitted radiation in the thin film. These sensitivity factors are plotted as a function of
23 atomic number and a smooth curve is fitted to the experimental values. The calibration
24 sensitivities are then read from these curves for the atomic numbers of each element in each
25 excitation condition. NIST standards are analyzed on a periodic basis to verify the sensitivity
26 factors. A multi-layer thin film standard prepared by Micromatter is analyzed with each set
27 of samples to check the stability of the instrument response. When deviations from specified
28 values are greater than ±5%, the system should be re-calibrated.
29 The sensitivity factors are multiplied by the net peak intensities yielded by ambient
30 samples to obtain the /*g/cm2 deposit for each element. The net peak intensity is obtained
April 1995 4.7! DRAFT-DO NOT QUOTE OR CITE
-------
Sample
^Characteristic Silicon detector
_x-rays / FET
X-ray excitation
Secondary
target
Be
window
' - ,/
Analog-to-
digital
converter
Anode
Electron beam
X-ray tube
Data output
Video
display
Signal
processing
Data
handling
Figure 4-21. Schematic of a Typical X-Ray Fluorescence (XRF) System.
April 1995
4-72
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-3. MINIMUM DETECTABLE LIMITS3 FOR XRF ANALYSIS
OF AIR FILTERS
Element
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Y
Zr
Mo
Pd
Ag
Cd
In
Condition
Numberd
5
5
5
5
4
4
4
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
2
4
1
1
1
1
Quartz-Fiber
Filter13
Protocol QA-
A ng/cm2 e
NAf
NA
NA
40g
30
40
100
50
20
8
7
15
5
4
4
6
8
9
5
5
5
8
8
10
20
20
20
25
30
Teflon Membrane Filter0
Protocol A
ng/cm2 d
10
6.3
5.6
5.0
10
6.1
4.5
2.9
2.5
1.9
1.6
1.5
0.88
0.89
1.1
1.1
1.9
1.6
1.2
1.0
1.0
1.1
1.3
1.7
2.7
11
12
12
13
Protocol B
ng/cm2
7.2
4.4
4.0
3.5
7.4
4.3
3.2
2.1
1.7
1.4
1.1
1.1
0.62
0.63
0.76
0.76
1.4
1.1
0.86
0.72
0.68
0.78
0.92
1.2
1.9
7.6
8.6
8.6
9.5
Protocol C
ng/cm2
3.6
2.2
2.0
1.8
3.7
2.2
1.6
1.0
0.87
0.67
0.56
0.54
0.31
0.31
0.38
0.38
0.68
0.56
0.43
0.36
0.34
0.39
0.46
0.59
0.95
3.8
4.3
4.3
4.8
Protocol D
ng/cm2
2.5
1.4
1.4
1.2
2.6
1.5
1.1
0.73
0.62
0.48
0.40
0.38
0.22
0.22
0.27
0.27
0.48
0.39
0.31
0.25
0.24
0.28
0.33
0.42
0.67
2.7
3.0
3.0
3.4
April 1995
4-73
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-3. (cont'd) MINIMUM DETECTABLE LIMITS3 FOR XRF ANALYSIS
OF AIR FILTERS
Element
Sn
Sb
Ba
La
Au
Hg
Tl
Pb
U
Condition
Numberd
1
1
1
1
2
2
2
2
2
Quartz-Fiber
Filterb
Protocol QA-
A ng/cm2 e
40
50
170
190
NA
20
NA
14
NA
Teflon Membrane Filter0
Protocol A
ng/cm2 d
17
18
52
62
3.1
2.6
2.5
3.0
2.3
Protocol B
ng/cm2
12
13
37
44
2.2
1.8
1.8
2.2
1.7
Protocol C
ng/cm2
6.2
6.4
18
22
1.1
0.91
0.88
1.1
0.83
Protocol D
ng/cm2
4.4
4.5
13
16
0.77
0.65
0.62
0.76
0.59
aMDL defined as three times the standard deviation of the blank for a filter of 1 mg/cm2
areal density.
bAnalysis times are 100 sec. for Conditions 1 and 4, and 400 sec. for Conditions 2 and 3.
Actual MDL's for quartz filters vary from batch to batch due to elemental contamination
variability.
Standard protocol, developed at the Desert Research Institute, University and Community
College System of Nevada, Reno, NV, analysis times are 100 sec. for Conditions 1, 4 and
5, and 400 sec. for Conditions 2 and 3 for Protocol A; 200 sec. for Conditions 1, 4 and
5 and 800 sec. for Conditions 2 and 3 for Protocol B; 800 sec. for Conditions 1,4 and 5
and 3,200 sec. for Conditions 2 and 3 for Protocol C; and 1600 sec. for Conditions 1, 4
and 5 and 6400 sec. for Conditions 2 and 3 for Protocol D.
dCondition 1 is direct mode excitation with a primary excitation filter of 0.15 mm thick Mo.
Tube voltage is 50 KV and tube current is 0.6 mA. Condition 2 is direct mode excitation
with a primary excitation filter of 0.13 mm thick Rh. Tube voltage is 35 KV and tube
voltage is 2.0 mA. Condition 3 uses Ge secondary target excitation with the secondary
excitation filtered by a Whatman 41 filter. Tube voltage is 30 KV and tube current is
3.3 mA. Condition 4 uses Ti secondary target excitation with the secondary excitation
filtered by 3.8 /tm thick mylar film. Tube voltage is 30 KV and tube current is 3.3 mA.
Condition 5 uses direct mode excitation with a primary excitation filter consisting of
3 layers of Whatman 41 filters. Tube voltage is 8 KV and tube current os 0.6 mA.
Multi-channel analyzer energy range is 0 to 40 KeV for condition 1, 0 - 20 KeV for
condition 2, and 0 to 10 KeV for conditions 3,4, and 5.
eTypical exposed area is 406 cm2 for standard high-volume filters; 6.4 cm2 for 37 mm ringed
Teflon-membrane filters; and 13.8 cm2 for 47 mm ringed Teflon-membrane filters.
Information not available.
gFor condition 4.
April 1995 4.74 DRAFT-DO NOT QUOTE OR CITE
-------
J_
_L
26-Oct-1992 18:09:56
SJTT046
Vert= 2000 counts Disp= 1
Preset= 100 sees
Comp= 2 Elapsed= 400 sees
San Jose, 1/21/92, PM 10
18:01 - 06:00
Excitation Condition 3
Fe
Ge i sec.
target
scatter
Si
S K Ca
& Cl II ;!
Zn
Al
! / I
Ti
Fe
Cu
V Cr
Mn
Ni
0.320 Range= 10.230 keV
Integral 0 -
10.230 >
243425
I I I I | I I I
5
Figure 4-22. Example of an X-Ray Fluorescence (XRF) Spectrum.
Source: Chow et al. (1990).
I
10
1 by: (1) subtracting background radiation; (2) subtracting spectral interferences; and
2 (3) adjusting for x-ray absorption.
3 The elemental x-ray peaks reside on a background of radiation scattered from the
4 sampling substrate. A model background is formed by averaging spectra obtained from
5 several blank filters of the same type used in ambient sampling. It is important to retain
6 blank filters for this purpose when XRF analysis is anticipated. This model background has
7 the same shape and features of the sample spectra (minus the elemental peaks) if the deposit
8 mass is small relative to the substrate mass (Russ, 1977). This model background is
April 1995
4-75
DRAFT-DO NOT QUOTE OR CITE
-------
1 normalized to an excitation radiation scatter peak, or the background areas of the spectrum
2 that have no elemental peaks, in each sample spectrum to account for the difference in scatter
3 intensity due to different masses.
4 The number and spacing of the characteristic x-ray lines relative to detector resolution
5 are such that the peaks from one element can interfere with a peak from another element
6 (Dzubay, 1986). A variety of methods has been used to subtract these peak overlaps (Arinc
7 et al., 1977; Parkes et al., 1979; Drane et al., 1983), including least squares fitting to library
8 spectra, Gaussian and other mathematical functions, and the use of peak overlap coefficients.
9 Peak overlap coefficients are applied to aerosol deposits. The most important of these
10 overlaps are the K-beta to K-alpha overlaps of elements that increase in atomic number from
11 potassium to zirconium, the lead L-alpha to arsenic K-alpha interference, and the lead M line
12 to sulfur K line interference. The ratios of overlap peaks to the primary peak are determined
13 from the thin film standards for each element for the spectral regions of the remaining
14 elements. These ratios are multiplied by the net peak intensity of the primary peak and
15 subtracted from the spectral regions of other elements.
16 The ability of an x-ray to penetrate matter depends on the energy of the x-ray and the
17 composition and thickness of the material. In general, lower energy x-rays, characteristic of
18 light elements, are absorbed in matter to a much greater degree than higher energy x-rays.
19 XRF analysis of air paniculate samples has had widest application to samples collected on
20 membrane-type filters such as Teflon- or polycarbonate-membrane filter substrates. These
21 membrane filters collect the deposit on their surfaces, which eliminates biases due to
22 absorption of x-rays by the filter material. These filters also have a low areal density which
23 minimizes the scatter of incident x-rays, and their inherent trace element content is very low.
24 Quartz-fiber filters used for high-volume aerosol sampling do not exhibit these features.
25 As noted earlier, blank elemental concentrations in quartz-fiber filters that have not
26 undergone acceptance testing can be several orders of magnitude higher than the
27 concentrations in the paniculate deposits. The concentrations vary substantially among the
28 different types of quartz-fiber filters, and even within the same filter type and manufacturing
29 lot. Blank impurity concentrations and their variabilities decrease the precision of
30 background subtraction from the XRF spectral data, resulting in higher detection limits.
31 Impurities observed in various types of glass- and quartz-fiber filters include aluminum,
April 1995 4.75 DRAFT-DO NOT QUOTE OR CITE
-------
1 silicon, sulfur, chlorine, potassium, calcium, iron, nickel, copper, zinc, rubidium, strontium,
2 molybdenum, barium, and lead. Concentrations for aluminum, silicon, phosphorus, sulfur,
3 and chlorine cannot be determined for quartz-fiber filters because of the large silicon content
4 of the filters.
5 Quartz-fiber filters also trap particles within the filter matrix, rather than on the
6 surface. This causes absorption of x-rays within the filter fibers yielding lower
7 concentrations than would otherwise be measured. The magnitude of this absorption
8 increases exponentially as the atomic number of the measured element decreases, and varies
9 from sample to sample. Absorption factors generally are "1.2" or less for iron and heavier
10 elements, but can be from "2" to "5" for sulfur.
11 Quartz-fiber filters are much thicker than membrane filters resulting in an increased
12 scattering of x-rays and a consequent increase in background and degradation of detection
13 limits. The increased x-ray scatter also overloads the x-ray detector which requires samples
14 to be analyzed at a lowered x-ray intensity. These effects alone can result in degradation of
15 detection limits by up to a factor of 10 with respect to Teflon-membrane substrates.
16 Larger particles collected during aerosol sampling have sufficient size to cause
17 absorption of x-rays within the particles. Attenuation factors for fine particles (PM2 5,
18 particles with aerodynamic diameters equal to or less than 2.5 jum) are generally negligible
19 (Criss, 1976), even for the lightest elements, but these attenuations can be significant for
20 coarse fraction particles (particles with aerodynamic diameters from 2.5 to 10 /mi).
21 Correction factors for XRF have been derived using the theory of Dzubay and Nelson (1975)
22 and should be applied to coarse particle measurements.
23 During XRF analysis, filters are removed from their Petri slides and placed with their
24 deposit sides down into filter cassettes. These cassettes are loaded into a mechanism that
25 exposes the filter deposits to x-rays. The sample chamber is evacuated and a computer
26 program controls the positioning of the samples and the excitation conditions. The vacuum in
27 the x-ray chamber and the heat induced by the absorption of x-rays can cause certain
28 materials to volatilize. Therefore, labile species such as nitrate and organic carbon are better
29 measured on a quartz-fiber filter that samples simultaneously with the Teflon-membrane
30 filter.
April 1995 4.77 DRAFT-DO NOT QUOTE OR CITE
-------
1 Quality control standards and replicates from previous batches should be analyzed for
2 every 10 to 20 samples. When quality control results differ from specifications by more than
3 +5%, or if the replicate concentrations differ from the original values (assuming they are at
4 least 10 times detection limits) by more than ±10%, the samples should be re-analyzed.
5
6 4.3.2.2 Particle Induced X-Ray Emission (PIXE) of Trace Elements
7 Particle Induced X-Ray Emission (PIXE) is another form of elemental analysis based on
8 the characteristics of x-rays and the nature of x-ray detection (Cahill et al., 1975; 1987;
9 1989; 1993; Cahill, 1990). PIXE uses beams of energetic ions, consisting of protons at an
10 energy level of 2 to 5 MeV, to create inner electron shell vacancies. As inner electron shell
11 atomic vacancies are filled by outer electrons, the emitted characteristics of x-rays can be
12 detected by wavelength dispersion, which is scattering from a crystal, or by energy
13 dispersion, which involves direct conversion of x-rays. The development of focusing
14 energetic proton beams (proton microprobes) has expanded the application of PIXE from
15 environmental and biological sciences to geology and material sciences. Figure 4-23
16 illustrates a typical PIXE setup in a thin target mode (Cahill, 1989). PIXE analysis is often
17 used for impactor samples or small filter substrates, since proton beams can be focused to a
18 small area with no loss of sensitivity (Cahill, 1993).
19 Very thick filters or thick particle deposits on filter substrates scatter the excitation
20 protons and lower the signal-to-noise ratio for PIXE. X-ray analysis methods, such as PIXE
21 and XRF, require particle size diameter corrections (for low atomic number targets)
22 associated with a spherical particle of a given diameter (typically particles with aerodynamic
23 diameters >2.5 ^m), and compositions typical in ambient aerosol studies. These analyses
24 also require correction for sample loadings that reflect the passage of x-rays through a
25 uniform deposit layer. Procedures for instrument calibration, spectrum process, and quality
26 assurance are similar to those documented in Section 4.3.1.2 for XRF.
27 PIXE analysis can provide one of the widest range of elements in a single analysis,
28 since x-ray results require two or three separate anodes. Attempts to improve sensitivity of
29 PIXE analysis may result in damage to Teflon-membrane filters, however. Recent
30 developments (Malm et al., 1994) using PIXE analysis at moderate sensitivity plus single
April 1995 4-78 DRAFT-DO NOT QUOTE OR CITE
-------
OJ
53
f
•§
C5
§
•g
C/l
SJ
£
April 1995
4-79
DRAFT-DO NOT QUOTE OR CITE
-------
1 anode XRF analysis at high sensitivity for transition/heavy metals have achieved the
2 minimum detectable limits of less than 0.01 ng/m3. With the addition of hydrogen analysis
3 (a surrogate for organic matter), almost all gravimetric mass concentrations can be explained
4 (Cahill, 1987).
5 XRF and PIXE are the most commonly used elemental analysis methods owing to its
6 nondestructive multi-element capabilities, relatively low cost, high detection limits, and
7 preservation of the filter for additional analyses. XRF sometimes needs to be supplemented
8 with INAA when extremely low detection limits are needed, but the high cost of INAA
9 precludes this method from being applied to large numbers of samples. AAS is a good
10 alternative for water-soluble species, especially for low atomic number . ICP/AES analysis
11 is a viable alternative, but it is less desirable because of the sample extraction elements such
12 as sodium and magnesium, but it requires large dilution factors to measure many different
13 elements expense and the destruction of the filter.
14
15 4.3.2.3 Instrumental Neutron Activation Analysis of Trace Elements
16 Instrumental neutron activation analysis (INAA) (Dams et al., 1970; Zoller and
17 Gordon, 1970; Olmez, 1989; Ondov and Divita, 1990), irradiated the thin membrane filter
18 sample in the core of a nuclear reactor for periods ranging from a few minutes to several
19 hours. Bombardment of the sample with neutrons induces a nuclear reaction of the stable
20 isotopes in the sample. The energies of the gamma rays emitted by the decay of this induced
21 radioactivity are used to identify them, and therefore, their parents. With the use of
22 prepared elemental standards, the amount of parent element in the sample can be determined
23 since the intensity of these gamma rays are proportional to their number.
24 The gamma-ray spectra of radioactive species are usually collected with a high
25 resolution germanium detector utilizing commercially available amplifiers and multi-channel
26 analyzers. Typical detector efficiencies range from 10 to 40% relative to a 3 x 3 in. sodium
27 iodide detector. Detector system resolution, measured as the full-width at half-maximum for
28 Table 4-4, the 1,332 KeV gamma-ray peak of 60Co, should be less than 2.3 KeV in order to
29 provide adequate resolution between isotopes of neighboring energies.
April 1995 4-80 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 4-4. INAA COUNTING SCHEME AND ELEMENTS MEASURED
Counting
Period
Irradiation
Time
Cooling
Time
Counting
Time
Elements Measured
Short-Lived 1 10 min
Short-Lived 2
Long-Lived 1 4-6 h
Long-Lived 2
5 min
20 min
30 days
5 min
20 min
3-4 days 6-8 h
12-24 h
Mg, Al, S, Ca, Ti, V, Cu
Na, Mg, Cl, K, Ca, Mn,
Zn, Ga, Br, Sr, In, I, Ba
Na, K, Ga, As, Br, Mo,
Cd, Sb, La, Nd, Sm, Yb,
Lu, W, Au, U
Sc, Cr, Fe, Co, Zn, Se,
Sr, Ag, Sb, Cs, Ba, Ce,
Nd, Eu, Gd, Tb, Lu, Hf,
Ta, Th
1 In order to obtain a full suite of elemental analysis results (often over 40 elements),
2 multiple counting periods and irradiations are performed on the same sample (e.g., two
3 irradiations would produce elements separated into short- and long-lived decay products).
4 An example of the elements determined from multiple irradiations and counting periods, and
5 the irradiation, cooling, and counting times used for ambient particulate samples collected on
6 Teflon-membrane filter material is summarized in Table 4-4 (Divita, 1993). These
7 irradiations were performed at the 20-MW NIST Research Reactor operated at 15-MW
8 (neutron flux of 7.7 x 1013 and 2.7 x 1013 neutron/cm2 x s). Typical gamma-ray spectra
9 resulting from the counting scheme described in Table 4-4 are shown in Figures 4-24
10 and 4-25.
11 The power of INAA is that it is not generally subject to interferences like XRF or
12 PIXE due to a much better ratio of gamma ray peak widths to total spectral width, by a
13 factor of about 20. INAA does not quantify some of the abundant species in ambient
14 particulate matter such as silicon, nickel, tin, cadmin, mercury, and lead. While INAA is
15 technically nondestructive, sample preparation involves folding the samples tightly and
16 sealing it in plastic, and the irradiation process makes the filter membrane brittle and
17 radioactive. These factors limit the use of the sample for subsequent analyses by other
18 methods. The technique also suffers from the fact that a nuclear reactor is usually used as a
April 1995
4-81
DRAFT-DO NOT QUOTE OR CITE
-------
3113 >IO 31OAO ION
5661
Figure 4-24. Typical Gamma-Ray Spectra Observed for Long Counts.
Count Count
Thousands Thousands
— * ~* IN) 10 CD — * -* r\i FO tj
oenoenoeno ocnocnoenc
§
o
O
03
CD.J-*
c 8
1
i
O
IN)
O
O
O
PC
8
o
ro
en
o
0°
13
cp_co
i§
o-
CD
co
en
o
o
o
fe
r-
•
Q
(Q
ro
>
Count
Thousands
[
(O
ro
CD
en
8
o
1
CD -*
= 8
1°
1
_*
Oi
8
ro
8
IN)
en
0
0°
Q)
CD.CO
I8
CT
CD
CO
0
1 1 1 f 1
I
i.
" cT
_ ->j
/_
I
—
^^__M "7
(O
i
CD
-------
aio HO aiono ION
£661 I
Count
Thousands
3
o
ro
O
0>
-c
3
CO
Count
Thousands
Count
en
00
r
i i i r
03
Oi
O
O)
CT
CD
co
o
3.
1
00
-------
1 source of neutrons. However, since the advent of high-resolution gamma-ray detectors,
2 individual samples can be analyzed for numerous elements simultaneously, most at
3 remarkably trace levels without the need for chemical separation. This greatly diminishes the
4 danger of contamination due to excessive sample handling and introduction of chemical
5 reagents used for separation procedures.
6
7 4.3.2.4 Microscopy Analysis of Particle Size, Shape, and Composition
8 Morphological and chemical features of particles can be used to identify the sources and
9 transport mechanism of airborne particles. The chemical analysis of individual particles
10 allows the attribution of specific pollution sources more straightforward while the abundance
11 of a specific group is a representative of the source strength. Both light (optical) and
12 scanning electron microscopy have been applied in environmental studies to examine the
13 single particles (e.g., Casuccio et al., 1983; Bruynseels et al., 1988; Van Borm and Adams,
14 1988; Javitz and Watson, 1989; Van Borm et al., 1989; CorniUe et al., 1990; Hopke and
15 Casuccio, 1991; Hoffer et al., 1991; Cheng et al., 1992; Turpin et al., 1993; 1994; 1995).
16 Light microscopy has been used for providing particle size information regarding the
17 morphology of microscopic features (Crutcher, 1982). The practical resolution of optical
18 microscopes is limited by the wavelengths associated with light of the visible spectrum.
19 When features of interest occur in micron and submicron size ranges, detailed resolution
20 cannot be obtained. The practical resolution of light microscopy is typically 1 to 2 /mi
21 (Meyer-Arendt, 1972).
22 The use of accelerated electrons in electron microscopy allows for the formation of
23 magnified images and an increased depth of field, and provides the resolution of a few
24 angstroms (10"4 ^m). Electron microscopy has now evolved to include: 1) the transmission
25 electron microscope (TEM); 2) the scanning electron microscope (SEM), and; 3) the
26 scanning transmission electron microscope (STEM) (Hearle et al. 1972; Lee et al., 1979;
27 Scott and Chatfield, 1979; Lee and Fisher, 1980; Lee and Kelly, 1980; Lee et al., 1981;
28 Johnson et al., 1981; Mclntyre and Johnson, 1982; Casuccio et al., 1983; Wernisch, 1985,
29 1986; Kim et al., 1987, 1988; Dzubay and Mamane, 1989; Henderson et al., 1989;
30 Schamber, 1993).
April 1995 4-84 DRAFT-DO NOT QUOTE OR CITE
-------
*
1 The SEM and STEM use accelerated electrons to strike the sample. As the electron
2 beam strikes the samples, various signals (e.g., secondary, backscattered, and Anger
3 electrons, characteristic x-rays, photons, and cathodoluminescence) are generated. These
4 signals can be collected to provide highly detailed information on a point-by-point basis. The
5 secondary electron signal yields a sample image with three-dimensional prospective, high
6 depth of field, and illuminated appearance. Back scattered electron images are used to
7 separate phases containing elements of different atomic number.
8 The information obtained from light and scanning microscopy analyses are usually
9 considered to be qualitative, due to the limited number of particles counted. To achieve a
10 quantitative analysis, a sufficient number of particles must be properly sized and identified by
11 morphology and/or chemistry to represent the entire sample. The selection of filter media,
12 optimal particle loadings, and sample handling methods are also of importance. In this
13 manner, the microscopic characteristics can be directly and reliably related to the bulk or
14 macroscopic properties of the sample (Casuccio et al., 1983).
15 Microscopic analysis requires a high degree of skill and extensive quality assurance to
16 provide quantitative information. The techniques is complex and expensive when quantitative
17 analysis is required. The evolution of computer technology has allowed for quantitative
18 analysis of particle samples of an entire population of features. With advanced pattern
19 recognition methods, data from individual particle features can be sorted and summarized by
20 size and composition, permitting improved quantitative source apportionment (Bruynseels
21 et al., 1988; Hopke and Casuccio, 1991). Casuccio et al. (1983) summarized the pros and
22 cons of automatic scanning electron microscopy.
23 Recent development of the SEM/XRF allows analysis of elemental compositions and
24 morphological information on small quantities of material (Bruynseels et al., 1988). Coupled
25 with statistical data analysis, computer controlled scanning electron microscopy shows great
26 promise for identifying and quantifying complex pollution sources in the field of receptor
27 modeling source apportionment (e.g., Griffin and Goldberg, 1979; Janocko et al., 1982;
28 Johnson et al, 1982; Massart and Kaufman, 1983; Hopke, 1985; Derde et al., 1987, Saucy
29 et al., 1987; Mamane, 1988a, 1988b; Cheng and Hopke, 1989; Dzubay and Mamane, 1989).
April 1995 4-85 DRAFT-DO NOT QUOTE OR CITE
-------
1 4.3.3 Wet Chemical Analysis
2 Aerosol ions refer to chemical compounds that are soluble in water. The water-soluble
3 portion of suspended particles associates itself with liquid water in the atmosphere when
4 relative humidity increases, thereby changing the light scattering properties of these particles.
5 Different emissions sources may also be distinguished by their soluble and non-soluble
6 fractions. Gaseous precursors can also be converted to their ionic counterparts when they
7 interact with chemicals impregnated on the filter material.
8 Several simple ions, such as soluble sodium, magnesium, potassium, and calcium are
9 best quantified by atomic absorption spectrometry (AAS) as described above. In practice,
10 AAS has been very useful for measuring water-soluble potassium and sodium, which are
11 important in apportioning sources of vegetative burning and sea salt, respectively.
12 Polyatomic ions such as sulfate, nitrate, ammonium, and phosphate must be quantified by
13 other methods such as ion chromatography (1C) and automated colorimetry (AC). Simple
14 ions, such as chloride, chromium III, and chromium IV, may also be measured by these
15 methods along with the polyatomic ions.
16 All ion analysis methods require filters to be extracted in DDW and then filtered to
17 remove the insoluble residue. The extraction volume needs to be as small as possible, lest
18 the solution become too dilute to detect the desired constituents. Each square centimeter of
19 filter should be extracted in no more than 2 ml of solvent for typical sampler flow rates of
20 20 to 30 L/min and sample durations of 24 h. This often results in no more than 20 ml of
21 extract that can be submitted to the different analytical methods, thereby giving preference to
22 those methods which require only a small sample volume. Sufficient sample deposit must be
23 acquired to account for the dilution volume required by each method.
24 When other analyses are to be performed on the same filter, the filter must first be
25 sectioned using a precision positioning jig attached to a paper cutter. For rectangular filters
26 (typically 20.32 cm by 25.40 cm), a 2.0 cm by 20.32 cm wide strip is cut from the center
27 two-thirds of the filter. Circular filters of 25-, 37-, and 47-mm diameters are usually cut in
28 half for these analyses, so the results need to be multiplied by two to obtain the deposit on
29 the entire filter. Filter materials that can be easily sectioned without damage to the filter or
30 the deposit must be chosen for these analyses.
31
April 1995 4-86 DRAFT-DO NOT QUOTE OR CITE
-------
i» 4.3.3.1 Ion Chromatographic Analysis for Chloride, Nitrate, and Sulfate
2 Ion Chromatography (1C) can be used for both anions (fluoride [F~] chloride [Cl~],
3 nitrite [NO^ bromide [Br~], nitrate [NO^], phosphate [PO43], sulfate [SO4=]) and cations
4 (soluble potassium [K+], ammonium [NH4+], soluble sodium [Na+]) with separate columns.
5 Applied to aerosol samples, the anions are most commonly analyzed by 1C with the cations
6 being analyzed by a combination of atomic absorption spectrophotometry (AAS) and
7 automated colorimetry (AC) (Chow and Watson, 1994). In 1C (Small et al., 1975; Mulik
8 et al., 1976, 1977; Butler et al., 1978; Mueller et al., 1978; Rich et al., 1978; Small, 1978),
9 the sample extract passes through an ion-exchange column that separates the ions in time for
10 individual quantification, usually by a electroconductivity detector. Figure 4-26 shows a
11 schematic representation of the 1C system. Prior to detection, the column effluent enters a
12 suppressor column where the chemical composition of the eluent is altered, resulting in a
13 lower background conductivity. The ions are identified by their elution/retention times and
14 are quantified by the conductivity peak area or peak height. 1C is especially desirable for
15 particle samples because it provides results for several ions with a single analysis and it uses
16 a small portion of the filter extract with low detection limits. Water-soluble chloride (Cl~),
17 nitrate (NOp, and sulfate (SO4=) are the most commonly measured anions in aerosol
18 samples. Figure 4-27 shows an example of an 1C anion chromatogram. 1C analyses can be
19 automated by interfacing to an automatic sampler that can conduct unattended analysis of as
20 many as 400 samples (Tejada et al., 1978).
21 In 1C, approximately 2 to 3 ml of the filter extract are injected into the 1C system. The
22 resulting peaks are integrated and the peak integrals are converted to concentrations using
23 calibration curves derived from standard solutions. For instance, the Dionex system
24 (Sunnyvale, CA) for the analysis of Cl", NO^, NO~3 PO4=, and SO4 contains a guard column
25 (AG4A column, Cat. No. #37042) and an anion separator column (AS4A column, Cat. No.
26 #37041) with a strong-basic anion-exchange resin, and an anion micro-membrane suppressor
27 column (250 x 6 mm ID) with a strong-acid ion-exchange resin. The 4 x 250 mm
28 analytical column is composed of 16 micron polystyrene/divinylbenzene substrate
29 agglomerated with anion exchange latex that has been completely aminated. The 0.5%
30 crosslinked latex particles have a diameter of approximately 0.175 ^im and carry the ion
31 exchange sites. The ion exchange capacity of the 4 X 250 mm analytical column is
April 1995 4-87 DRAFT-DO NOT QUOTE OR CITE
-------
Delivery Module—
Chromatography Module —
Detector Module —
Eluent
Reservoir
Pump
Sample
Injector
Guard
Column
Separator
Column
Suppressor
Device
Conductivity
Cell
\Waste J
Figure 4-26. Schematic Representation of an Ion Chromatography (1C) System.
April 1995
4-i
DRAFT-DO NOT QUOTE OR CITE
-------
18,000-
15,500-
13,000-
10,500-
8,000-
5,500-
3,000-
500-
-2,000-
0.
Fluoride
I Chloride
I I
-^
I
30
Nitrite
lj
Nitrate
\ A Dk u . Sulfate
\ \ Phosphate /x
\ / \_ /\ / \
5.00 10.00
Minutes
Figure 4-27. Example of an Ion Chromatogram Showing the Separation of Fluoride,
Chloride, Nitrite, Nitrate, Phosphate, and Sulfate Ions.
1 20 /ueq/column. The column is stable between pH 0 and 14. The anion eluent consists of
2 sodium carbonate (Na2CO3) and sodium bicarbonate (NaHCO3) prepared in DDW. The
3 DDW is verified to have a conductivity of less than 1.8 x 10"5 ohm"1 cm"1 prior to
4 preparation of the eluent. For quantitative determinations, the 1C is operated at a flow rate
5 of 2.0 L/min. The system can also analyze fluoride with an eluent concentration of 1.8 M
6 Na2CO3/1.7MNaHCO3.
7 The primary standard solution is prepared annually and stored in a refrigerator. It is
8 prepared from the reagent grade sodium salts (e.g., NaF, NaCl, NaNO2, NaBr, NaNO3,
9 Na2HPO, and Na2SO4. These anhydrous salts are dried in an oven at 105 °C for 1 h and
10 then cooled to room temperature in a dessicator. They are weighed to the nearest 0.10 mg
11 on a routinely calibrated analytical balance under controlled temperature ( — 20° C) and
12 relative humidity (± 30%) conditions. These salts are diluted in precise volumes of DDW.
13 Calibration standards are prepared at least once within each month by diluting the primary
14 standard to concentrations covering the range of concentrations expected in the filter extracts.
April 1995
4-89
DRAFT-DO NOT QUOTE OR CITE
-------
1 The normal calibration concentrations prepared are 0.1, 0.2, 0.5, 1.0, and 2.0 ptg/ml for
2 each of the analysis species.
3 Several independent quality assurance (QA) standards should be used to check the
4 calibration curve. The standards that are traceable to NIST simulated rainwater standards
5 are: Environmental Resource Associates (ERA, Arvada, CA) custom standards containing
6 the anions measured at a concentration of 100 /xg/ml, ERA Waste Water Nutrient Standard,
7 ERA Waste Water Mineral Standard, and Alltech individual standards at 200 /ig/ml. The
8 QA standards are diluted in DDW to concentrations that are within the range of the
9 calibration curve.
10 Calibration curves are performed weekly. Chemical compounds are identified by
11 matching the retention time of each peak in the unknown sample with the retention times of
12 peaks in the chromatograms of the standards. The QA standards are analyzed at the
13 beginning of each sample run to check calibrations. A DDW blank is analyzed after every
14 20 samples and a calibrations standard is analyzed after every 10 samples. These quality
15 control (QC) checks verify the baseline and calibration respectively.
16
17 4.3.3.2 Automated Colorimetric Analysis for Ammonium, Nitrate, and Sulfate
18 Automated Colorimetry (AC) applies different colorimetric analyses to small sample
19 volumes with automatic sample throughput. The most common ions measured are
20 ammonium, chloride, nitrate, and sulfate (Butler et al., 1978; Mueller et al., 1978; Fung
21 et al., 1979; Pyen and Fishman, 1979). Since 1C provides multi-species analysis for the
22 anions, ammonium is most commonly measured by AC.
23 The AC system is illustrated schematically in Figure 4-28. The heart of the automated
24 colorimetric system is a peristaltic pump, which introduces air bubbles into the sample
25 stream at known intervals. These bubbles separate samples in the continuous stream. Each
26 sample is mixed with reagents and subjected to appropriate reaction periods before
27 submission to a colorimeter. The ion being measured usually reacts to form a colored liquid.
28 The liquid absorbance is related to the amount of the ion in the sample by Beer's Law. This
29 absorbance is measured by a photomultiplier tube through an interference filter specific to the
30 species being measured.
April 1995 4-90 DRAFT-DO NOT QUOTE OR CITE
-------
18,000-
15,500-
13,000-
10,500-
8,000-
5,500-
3,000-
500-
-2,000-
0.
Fluoride
Chloride
I 1
-^
Nitrite
111
)0
Nitrate
\ |\ Phosphate /^e
\ / V s\ / \
5.00 10.00
Minutes
Figure 4-28. Schematic of a Typical Automated Colorimetric (AC) System.
1 The standard AC technique can analyze »60 samples per hour per channel, with
2 minimal operator attention and relatively low maintenance and material costs. Several
3 channels can be set up to simultaneously analyze several ions. The methylthymol-blue
4 (MTB) method is applied to analyze sulfate. The reaction of sulfate with MTB-barium
5 complex results in free ligand, which is measured colorimetrically at 460 nm. Nitrate is
6 reduced to nitrite that reacts with sulfanilamide to form a diazo compound. This compound
7 is then reacted to an azo dye for colorimetric determination at 520 nm. Ammonium is
8 measured with the indophenol method. The sample is mixed sequentially with potassium
9 sodium tartrate, sodium phenolate, sodium hypochlorite, sodium hydroxide, and sodium
10 nitroprusside. The reaction results in a blue-colored solution with an absorbance measured at
11 630 nm.
12 The system determines carry-over by analysis of a low concentration standard following
13 a high concentration. The percent carry-over is then automatically calculated and can be
14 applied to the samples analyzed during the run.
15 Formaldehyde has been found to interfere with ammonium measurements when present
16 in an amount exceeding 20% of the ammonium content, and hydrogen sulfide interferes in
April 1995
4-91
DRAFT-DO NOT QUOTE OR CITE
-------
1 concentrations exceeding 1 mg/ml. Nitrate and sulfate are also potential interferents when
2 present at levels exceeding 100 times the ammonium concentration. These levels are rarely
3 exceeded in ambient samples. The precipitation of hydroxides of heavy metals such as
4 magnesium and calcium is prevented by the addition of disodium ethylenediamine-tetracetate
5 (EDTA) to the sample stream (Chow et al., 1980; Chow, 1981-). It was learned in the
6 Sulfate Regional Experiment (SURE) (Mueller et al., 1983) that the auto-sampler should be
7 enclosed in an atmosphere that is purged of ammonia by bubbling air through a phosphoric
8 acid solution.
9 The automated colorimetric system requires a periodic standard calibration with the
10 daily prepared reagents flowing through the system. Lower quantifiable limits of AC for
11 sulfate and nitrate are an order of magnitude higher than those obtained with 1C.
12 Intercomparison studies between AC and 1C have been conducted by Butler et al.
13 (1978); Mueller et al. (1978); Fung et al. (1979); and Pyen and Fishman (1979). Butler
14 et al. (1978) found excellent agreement between sulfate and nitrate measurements by AC and
15 1C. The accuracy of both methods is within the experimental errors, with higher blank
16 values observed for AC techniques. Comparable results were also obtained between the two
17 methods by Fung et al. (1979). The choice between the two methods for sample analysis is
18 dictated by sensitivity, scheduling, and cost constraints.
19 Two milliliters of extract in sample vials are placed in an autosampler that is controlled
20 by a computer. Five standard concentrations (e.g., (NH4)2SO4, Na2SO4, NaNO3) are
21 prepared from American Chemical Society reagent-grade chemicals following the same
22 procedure as that for 1C standards. Each set of samples consists of two DDW blanks to
23 establish a baseline, five calibration standards and a blank, then sets of ten samples followed
24 by analysis of one of the standards and a replicate from a previous batch. The computer
25 control allows additional analysis of any filter extract to be repeated without the necessity of
26 loading the extract into more than one vial.
27
28 4.3.3.3 Atomic Absorption Spectrophotometric (AAS) and Inductive Coupled Plasma
29 Atomic Emission Spectro (ICP/AES) Photometry Analyses for Trace Elements
30 In AAS (Ranweiler and Moyers, 1974; Fernandez, 1989), the sample is first extracted
31 in a strong solvent to dissolve the solid material; the filter or a portion thereof is also
32 dissolved during this process. A few milliliters of this extract are introduced into a flame
April 1995 4.92 DRAFT-DO NOT QUOTE OR CITE
-------
1 where the elements are vaporized. Most elements absorb light at certain wavelengths in the
2 visible spectrum, and a light beam with wavelengths specific to the elements being measured
3 is directed through the flame to be detected by a monochrometer. The light absorbed by the
4 flame containing the extract is compared with the absorption from known standards to
5 quantify the elemental concentrations. AAS requires an individual analysis for each element,
6 and a large filter or several filters are needed to obtain concentrations for a large number of
7 the elements specified in Table 4.3.1. AAS is a useful complement to other methods, such
8 as XRF and PIXE, for species such as beryllium, sodium, and magnesium that are not well-
9 quantified by XRF and PIXE. Airborne particles are chemically complex and do not
10 dissolve easily into complete solution, regardless of the strength of the solvent. There is
11 always a possibility that insoluble residues are left behind and soluble species may
12 co-precipitate on them or on container walls.
13 In ICP/AES (Fassel and Kniseley, 1974; McQuaker et al., 1979; Lynch et al., 1980;
14 Harman, 1989), the dissolved sample is introduced into an atmosphere of argon gas seeded
15 with free electrons induced by high voltage from a surrounding Tesla coil. The high
16 temperatures in the induced plasma raise valence electrons above their normally stable states.
17 When these electrons return to their stable states, a photon of light is emitted which is unique
18 to the element which was excited. This light is detected at specified wavelengths to identify
19 the elements in the sample. ICP/AES acquires a large number of elemental concentrations
20 using small sample volumes with acceptable detection limits for atmospheric samples.
21 As with AAS, this method requires complete extraction and destruction of the sample.
22
23 4.3.4 Organic Analysis
24 4.3.4.1 Analysis of Organic Compounds
25 Organic compounds comprise a major portion of airborne particles in the atmosphere,
26 thus contributing to visibility degradation, and affecting the properties of clouds into which
27 these particles are scavenged. Specific groups of organic compounds (e.g., polycyclic
28 aromatic hydrocarbons, PAHs) have also been implicated in human health effects. However,
29 due to the very complex composition of the organic fraction of atmospheric aerosols, the
30 detailed composition and atmospheric distributions of organic aerosol constituents are still not
31 well understood. .
April 1995 4.93 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sampling techniques for atmospheric participate matter have been extensively
2 investigated, resulting in the development of collection methods for ambient aerosol particles
3 in a wide range of particle sizes. Particles are most frequently collected on glass or quartz-
4 fiber filters that have been specially treated to achieve low "carbon blanks". Ambient
5 organic paniculate matter has also been collected on a variety of particle sizing devices, such
6 as low pressure impactors and Micro Oriface Uniforms Deposit Impactors("MOUDI"). Very
7 recently, diffusion based samplers have been used as well (i.e., Eatough et al., 1994; Tang
8 et al., 1994). However, the task of sampling organic compounds in airborne particles is
9 complicated by the fact that many of these compounds have equilibrium vapor pressures
10 (gaseous concentrations) that are considerably larger than their normal ambient
11 concentrations. This implies a temperature- and concentration-dependent distribution of such
12 organics between paniculate and vapor phases. It also suggests that artifacts may occur due
13 to volatilization during the sampling process (Coutant et al., 1988). Such volatilization
14 would cause the under-estimation of the particle-phase concentrations of organics.
15 Conversely, the adsorption of gaseous substances on deposited particles, or on the filter
16 material itself, a process driven by the lowered vapor pressure over the sorbed material,
17 would lead to over-estimation of the particle-phase fraction (Bidleman et al., 1986; Ligocki
18 and Pankow, 1989; McDow and Huntzicker, 1990). In addition, several studies have
19 suggested that chemical degradation of some organics may occur during the sampling
20 procedure (Lindskog et al., 1985; Arey et al., 1988; Parmar and Grosjean, 1990).
21 The partitioning of semi-volatile organic compounds (SOC) between vapor and particle
22 phases has received much attention recently (Ligocki and Pankow, 1989; Gotham and
23 Bidleman, 1992; Lane et al., 1992; Kaupp and Umlauf, 1992; Pankow, 1992; Turpin et al.,
24 1993). Most estimates of partition have relied on high-volume (hi-vol) sampling, using a
25 filter to collect particles followed by a solid adsorbent trap to collect the gaseous portion of
26 SOC (e.g., Kaupp and Umlauf, 1992, Foreman and Bidleman, 1990). It has been shown
27 recently (Kaupp and Umlauf, 1992) that this approach, although not absolutely free from
28 sorption and desorption artifacts, produces reliable results. The maximum differences
29 observed between hi-vol filter-solid adsorbent sampling and impactor sampling (the latter
30 believed to be less susceptible to these sampling artifacts) did not exceed a factor of two.
April 1995 4-94 DRAFT-DO NOT QUOTE OR CITE
-------
1 There is good theoretical and experimental evidence that use of a diffusion denuder
2 technique significantly improves measurements of vapor-particle phase partitioning (Coutant
3 et al., 1988, 1989, 1992; Lane et al., 1988). However, at the present state of their
4 technological development, the reliability of denuders for investigation of atmospheric
5 partitioning of non-polar SOC needs to be improved, as suggested by contradictions in
6 published field data (e.g., Kaupp and Umlauf, 1992). A new, improved sampler has recently
7 been introduced (Gundel et al., 1992) which uses a proprietary XAD-4-coated tube for vapor
8 collection, followed by filter collection of organic aerosol particles and a sorbent bed to
9 quantitatively retain desorbed (volatilized) organic vapors. Preliminary results from the use
10 of this device look very promising for direct measurements of the phase distribution of semi-
11 volatile organic aerosol constituents. Another promising application of denuder technology
12 has been their use to remove ozone from an air sampling stream before it reaches the filter,
13 reducing the potential for artifact formation on the captured paniculate material during the
14 sampling period (Williams and Grosjean, 1990).
15 Since the organic fraction of airborne particulate matter is typically a complex mixture
16 of hundreds to thousands of compounds distributed over many organic functional groups, its
17 chemical analysis is an extremely difficult task (Appel et al., 1977; Simoneit, 1984; Flessel
18 et al., 1991; Hildemann et al., 1991; Li and Kamens, 1993; Rogge et al., 1993a, 1993b,
19 1993c). Analyses of organics generally begin with solvent extraction of the particulate
20 sample. A variety of solvents and extraction techniques have been used in the past. One
21 common method is sequential extraction with increasingly polar solvents, which typically
22 separates the organic material into nonpolar, moderately polar, and polar fractions (Daisey
23 et al., 1982, 1987). This step is usually followed by further fractionation using open-column
24 liquid chromatography and/or high-performance liquid chromatography (HPLC) in order to
25 obtain several less complicated fractions (e.g., Schuetzle and Lewtas, 1986; Atkinson et al.,
26 1988). These fractions can then be analyzed further with high resolution capillary-column
27 gas chromatography (GC), combined with mass spectrometry (GC/MS), Fourier transform
28 infrared (GC/FTIR/MS) or other selective detectors.
29 Much of the recent work on the identification of nonpolar and semi-polar organics in
30 airborne samples has used bioassay-directed chemical analysis (Scheutzle and Lewtas, 1986),
31 and has focused on identification of fractions and compounds that are most likely to be of
April 1995 4.95 DRAFT-DO NOT QUOTE OR CITE
-------
1 significance to human health. In particular, PAHs and their nitro-derivatives (nitroarenes)
2 attracted considerable attention due to their mutagenic and, in some cases, carcinogenic
3 properties. More than 100 PAHs have been identified in the PM2 5 fraction of ambient
4 particulate matter (Lee et al., 1981). While most of the nitroarenes found in ambient
5 particles are also present in primary combustion-generated emissions, some are formed from
6 their parent PAH in the atmospheric nitration reactions (e.g., Arey et al., 1986; Zielinska
7 et al., 1989, Ramdahl et al., 1986).
8 Little work has been done to date to chemically characterize the polar fraction in detail,
9 even though polar material accounts for up to half the mass and mutagenicity of soluble
10 ambient particulate organic matter (Atherholt et al., 1985; Gundel et al., 1994). Until
11 recently the polar fraction has remained analytically intractable, since very polar and labile
12 species interact with conventional fractionation column packing materials and cannot be
13 recovered quantitatively. Recently very polar particulate organic matter has been
14 successfully fractionated using cyanopropyl-bonded silica (Gundel et al., 1994), with good
15 recovery of mass and mutagenicity (Kado et al., 1991). However, ambient particulate polar
16 organic material cannot be analyzed with conventional GC/MS because of GC column losses
17 resulting from adsorption, thermal decomposition, and chemical interactions. New analytical
18 techniques, such as HPLC/MS and MS/MS, need to be applied if the chemical constituents
19 of polar particulate organic matter are to be identified and quantified.
20 Most of the recent work on the identification of particulate organic matter has focused
21 on mutagenic and carcinogenic compounds that are of significance to human health.
22 Relatively little work has been done to characterize individual compounds or classes of
23 compounds that are specific to certain sources of organic aerosol. In urban and rural
24 atmospheres, as well as in the remote troposphere, organic composition corresponding to
25 fingerprints of plant waxes, resin residues, and long-chain hydrocarbons from petroleum
26 residues have been found (e.g., Gagosian et al., 1981; Simoneit, 1984; Mazurek et al., 1987,
27 1989, 1991; Simoneit et al., 1991; Rogge et al., 1994). However, a variety of smaller,
28 multi-functional compounds characteristic of gas-to-particle conversion have also been
29 observed (e.g., F inlay son-Pitts and Pitts, 1986). These compounds tend to be present in the
30 polar fraction of ambient organic aerosol particles, having been formed from atmospheric
April 1995 4.96 DRAFT-DO NOT QUOTE OR CITE
-------
1 chemical reactions of less polar precursors. Little is currently known about the chemical
2 composition of this polar fraction due to the serious analytical difficulties mentioned above.
3
4 4.3.4.2 Analysis of Organic and Elemental Carbon
5 Three classes of carbon are commonly measured in aerosol samples collected on
6 quartz-fiber filters: 1) organic, volatilized, or non-light absorbing carbon; 2) elemental or
7 light-absorbing carbon; and 3) carbonate carbon. Carbonate carbon (i.e., K2CO3, Na2CO3,
8 MgCO3, CaCO3) can be determined on a separate filter section by measurement of the
9 carbon dioxide (C02) evolved upon acidification (Johnson et al., 1981). Though progress
10 has been made in the quantification of specific organic chemical compounds in suspended
11 particles (e.g., Rogge et al., 1991), sampling and analysis methods have not yet evolved for
12 use in practical monitoring situations.
13 Many methods have been applied to the separation of organic and elemental carbon in
14 ambient and source particulate samples (McCarthy and Moore, 1952; Mueller et al., 1971;
15 Lin et al., 1973; Patterson, 1973; Gordon, 1974; Grosjean, 1975; Smith et al., 1975; Appel
16 et al., 1976, 1979; Kukreja and Bove, 1976; Merz, 1978; Rosen et al., 1978; Dod et al.,
17 1979; Johnson and Huntzicker, 1979; Macias et al., 1979; Malissa, 1979; Weiss et al., 1979;
18 Cadle et al., 1980a, 1980b; Heisler et al., 1980a, 1980b; Johnson et al., 1980, 1981;
19 Pimenta and Wood, 1980; Daisey et al., 1981; Mueller et al., 1981; Novakov, 1981, 1982;
20 Cadle and Groblicki, 1982; Gerber, 1982; Heintzenberg, 1982; Huntzicker et al., 1982;
21 Muhlbaier and Williams, 1982; Rosen et al., 1982; Tanner et al., 1982; Stevens et al., 1982;
22 Wolff et al., 1982; Japar et al., 1984). Comparisons among the results of the majority of
23 these methods show that they yield comparable quantities of total carbon in aerosol samples,
24 but the distinctions between organic and elemental carbon are quite different (Countess,
25 1990; Heringetal., 1990).
26 The definitions of organic and elemental carbon are operational and reflect the method
27 and purpose of measurement. Elemental carbon is sometimes termed "soot", "graphitic
28 carbon," or "black carbon." For studying visibility reduction, light-absorbing carbon is a
29 more useful concept than elemental carbon. For source apportionment by receptor models,
30 several consistent but distinct fractions of carbon in both source and receptor samples are
31 desired, regardless of their light-absorbing or chemical properties. Differences in ratios of
April 1995 4.97 DRAFT-DO NOT QUOTE OR CITE
-------
1 the carbon concentrations in these fractions form part of the source profile that distinguishes
2 the contribution of one source from the contributions of other sources.
3 Light-absorbing carbon is not entirely constituted by graphitic carbon, since there are
4 many organic materials that absorb light (e.g., tar, motor oil, asphalt, coffee). Even the
5 "graphitic" black carbon in the atmosphere has only a poorly developed graphitic structure
6 with abundant surface chemical groups. "Elemental carbon" is a poor but common
7 description of what is measured. For example, a substance of three-bond carbon molecules
8 (e.g., pencil lead) is black and completely absorbs light, but four-bond carbon in a diamond
9 is completely transparent and absorbs very little light. Both are pure, elemental carbon.
10 Chow et al. (1993) document several variations of the thermal (T), thermal/optical
11 reflectance (TOR), thermal/optical transmission (TOT), and thermal manganese oxidation
12 (TMO) methods for organic and elemental carbon. The TOR and TMO methods have been
13 most commonly applied in aerosol studies in the United States.
14 The TOR method of carbon analysis developed by Huntzicker et al. (1982) has been
15 adapted by several laboratories for the quantification of organic and elemental carbon on
16 quartz-fiber filter deposits. While the principle used by these laboratories is identical to that
17 of Huntzicker et al. (1982), the details differ with respect to calibration standards, analysis
18 time, temperature ramping, and volatilization/combustion temperatures. In the TOR method
19 (Chow et al., 1993), a filter is submitted to volatilization at temperatures ranging from
20 ambient to 550°C in a pure helium atmosphere, then to combustion at temperatures between
21 550 to 800°C in a 2% oxygen and 98% helium atmosphere with several temperature ramping
22 steps. The carbon that evolves at each temperature is converted to methane and quantified
23 with a flame ionization detector. The reflectance from the deposit side of the filter punch is
24 monitored throughout the analysis. This reflectance usually decreases during volatilization in
25 the helium atmosphere owing to the pyrolysis of organic material. When oxygen is added,
26 the reflectance increases as the light-absorbing carbon is combusted and removed. Organic
27 carbon is defined as that which evolves prior to re-attainment of the original reflectance, and
28 elemental carbon is defined as that which evolves after the original reflectance has been re-
29 attained. By this definition, "organic carbon" is actually organic carbon that does not absorb
30 light at the wavelength (632.8 nm) used, and "elemental carbon" is light-absorbing carbon
31 (Chow et al., 1993). The TOT method applies the same thermal/optical carbon analysis
April 1995 4-98 DRAFT-DO NOT QUOTE OR CITE
-------
1 method except that transmission instead of reflectance of the filter punch is measured.
2 Thermal methods apply no optical correction and define elemental carbon as that which
3 evolves after the oxidizing atmosphere is introduced.
4 The TMO method (Mueller et al., 1982; Fung, 1990) uses manganese dioxide (MnO2),
5 present and in contact with the sample throughout the analysis, as the oxidizing agent, and
6 temperature is relied upon to distinguish between organic and elemental carbon. Carbon
7 evolving at 525°C is classified as organic carbon, and carbon evolving at 850°C is classified
8 as elemental carbon.
9 Carbon analysis methods require a uniform filter deposit because only a small portion
10 of each filter is submitted to chemical analysis. The blank filter should be white for light
11 reflection methods, and at least partially transparent for light transmission methods. The
12 filter must also withstand very high temperatures without melting during combustion.
13 Since all organic matter contains hydrogen as the most common elemental species,
14 analysis of hydrogen by proton elastic scattering analysis (PESA) has been developed by
15 Cahill (1987). A correction must be made for hydrogen in sulfates and nitrates, but since the
16 analysis is done in a vacuum, water is largely absent. The method has excellent sensitivity
17 which is approximately 20 times better than combustion techniques. This method requires
18 knowledge of the chemical state of sulfates, nevertheless, reasonable agreement was found as
19 compared to the combustion techniques.
20
21 4.3.5 Quality Assurance
22 The use of multiple methods and parallel samplers achieves both optimum performance
23 and quality assurance. While this has been a part of major research studies since the 1970's,
24 its extension to routine monitoring of aerosols was most extensively introduces in the
25 SCENES and IMPROVE visibility programs (Eldred, 1989). The concept was labeled,
26 "Integral Redundancy," and was recently adopted by the United Nation's Global Atmospheric
27 Watch Program.
28 The internal consistency checks applied to the IMPROVE network are listed as follows:
29 1) Mass (gravimetric) is compared to the sum of all elements on the Teflon-membrane
30 filter of Channel A (PIXE, PESA, XRF analysis; Internally XRF and PIXE are
31 compared for elements around iron). This was pioneered in the SCENES program
32 and is now the standard practice for many aerosol studies.
April 1995 4.99 DRAFT-DO NOT QUOTE OR CITE
-------
1 2) Sulfate, by ion chromatography on Channel B's nylon filter, after an acidic vapor
2 denuder, is compared to sulfur (X3) from Channel A's Teflon-membrane filter by
3 PIXE. Agreement is excellent, except for summer.
4
5 3) Organic matter, by combustion on Channel C's quartz-fiber filter stack, is
6 compared to organic matter via PESA analysis of hydrogen on Channel A's Teflon-
7 membrane filter. This is an exceptionally sever test due to the nature of organics.
8 These comparisons are made for every IMPROVE analysis, yielding about 25,000
9 comparisons to date (Malm et al., 1994).
10 These types of data validation checks should be carried out in every PM measurement
11 program to ensure the accuracy, precision, and validity of the chemical analysis data.
April 1995 4-100 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Allen, D. T.; Palen, E. J.; Haimov, M. I.; Hering, S. V.; Young, J. R. (1994) Fourier transform infrared
4 spectroscopy of aerosol collected in a low pressure impactor (LPI/FTIR): method development and field
5 calibration. Aerosol Sci. Technol. 21: 325-342.
6
7 American Conference of Governmental Industrial Hygienists. (1985) Particle size-selective sampling in the
8 workplace. Cincinnati, OH: American Conference of Governmental Industrial Hygienists.
9
10 Appel, B. R.; Hoffer, E. M.; Haik, M.; Wall, S. M.; Kothny, E. L. (1977) Characterization of organic
11 paniculate matter. Sacramento, CA: California Air Resources Board; document no. ARB-R-5-682-77-72.
12
13 Appel, B. R.; Hoffer, E. M.; Kothny, E. L.; Wall, S. M.; Haik, M.; Knights, R. L. (1979) Analysis of
14 carbonaceous material in southern California atmospheric aerosols. 2. Environ. Sci. Technol. 13: 98-104.
15
16 Appel, B. R.; Povard, V.; Kothny, E. L. (1988) Loss of nitric acid within inlet devices intended to exclude
17 coarse particles during atmospheric sampling. Atmos. Environ. 22: 2535-2540.
18
19 Appel, B. R.; Cheng, W.; Salaymeh, F. (1989) Sampling of carbonaceous particles in the atmosphere—II.
20 Atmos. Environ. 23: 2167-2175.
21
22 Arey, J.; Zielinska, B.; Atkinson, R.; Winer, A. M.; Ramdahl, T.; Pitts, J. N., Jr. (1986) The formation of
23 nitro-PAH from the gas-phase reactions of fluoranthene and pyrene with the OH radical in the presence
24 of NOX. Atmos. Environ. 20: 2339-2345.
25
26 Arey, J.; Zielinska, B.; Atkinson, R.; Winer, A. M. (1988) Formation of nitroarenes during ambient
27 high-volume sampling. Environ. Sci. Technol. 22: 457-462.
28
29 Arinc, F.; Wielopolski, L.; Gardner, R. P. (1977) The linear least-squares analysis of X-ray fluorescence spectra
30 of aerosol samples using pure element library standards and photon excitation. In: Dzubay, T. G., ed.
31 X-ray fluorescence analysis of environmental samples. Ann Arbor, MI: Ann Arbor Science Publishers;
32 p. 227.
33
34 Arnold, S.; Hague, W.; Pierce, G.; Sheetz, R. (1992) The use of beta gauge monitors for PSI and every day SIP
35 monitoring: an overview of the Denver experience. In: Chow, J. C.; Ono, D. M., eds. PM10 standards
36 and non-traditional paniculate source controls: proceedings of AWMA/EPA international specialty
37 conference. Pittsburgh, PA: Air & Waste Management Association; pp. 13-23.
38
39 Askne, C.; Brosset, C.; Fern, M. (1973) Determination of the proton-donating property of airborne particles.
40 Goteborg, Sweden; NL publication B157.
41
42 Atherholt, T. B.; McGarrity, T. B.; Louis, J. B.; McGeorge, L. J.; Lioy, P. J.; Daisey, J. M.; Greenberg, A.;
43 Darack, F. (1985) In: Waters, M. D.; Sandhu, S. S.; Lewtas, J.; Claxton, L.; Strauss, G.;
44 Nesnow, S., eds. Short-term bioassays in the analysis of complex environmental mixtures IV. New York,
45 NY: Plenum Press; pp. 211-231.
46
47 Atkinson, R.; Arey, J.; Winer, A. M.; Zielinska, B. (1988) A survey of ambient concentrations of selected
48 polycyclic aromatic hydrocarbons (PAH) at various locations in California [final report]. Sacramento,
49 CA: California Air Resources Board; contract no. A5-185-32.
50
51 Baron, P. A. (1986) Calibration and use of the aerodynamic particle sizer (APS-3300). Aerosol Sci. Technol.
52 5: 55-67.
53
April 1995 4-101 DRAFT-DO NOT QUOTE OR CITE
-------
1 Baron, P. A. (1993) Measurement of asbestos and other fibers. In: Willeke, K.; Baron, P. A., eds. Aerosol
2 measurement principles, techniques and applications. New York, NY: Van Nostrand Reinhold; p. 574.
3
4 Baron, P. A.; Willeke, K. (1993) Aerosol fundamentals. In: Willeke, K.; Baron, P. A., eds. Aerosol
5 measurement principles, techniques and applications. New York, NY: Van Nostrand Reinhold; p. 16.
6
7 Baron, P. A.; Mazumder, M. K.; Cheng, Y. S. (1993) Direct reading techniques using optical particle detection.
8 In: Willeke, K.; Baron, P. A., eds. Aerosol measurement principles, techniques and applications. New
9 York, NY: Van Nostrand Reinhold; p. 381.
10
11 Bartley, D. L.; Breuer, G. M. (1982) Analysis and optimization of the performance of the 10 mm cyclone. Am.
12 Ind. Hyg. Assoc. J. 43: 520-528.
13
14 Benner, C. L.; Eatough, D. J.; Eatough, N. L.; Bhardwaja, P. (1991) Comparison of annular denuder and filter
15 pack collection of HNO3 (g), SO2 (g), and particulate-phase nitrate, nitrite and sulfate in the south west
16 desert. Atmos. Environ. Part A 25: 1537-1545.
17
18 Bidleman, T. F.; Billings, W. N.; Foreman, W. T. (1986) Vapor-particle partitioning of semivolatile organic
19 compounds: estimates from field collections. Environ. Sci. Technol. 20: 1038-1043.
20
21 Botham, R. A.; Hughson, G. W.; Vincent, J. H.; Mark, D. (1991) The development of a test system for
22 investigating the performances of personal aerosol samplers under actual workplace conditions. Am. Ind.
23 Hyg. Assoc. J. 52: 423-427.
24
25 Brauer, M.; Koutrakis, P.; Spengler, J. D. (1989) Personal exposures to acidic aerosols and gases. Environ. Sci.
26 Technol. 23: 1408-1412.
27
28 Brockmann, J. E.; Yamano, N.; Lucero, D. (1988) Calibration of the aerodynamic particle sizer (AOS-3310)
29 with polystyrene latex monodisperse spheres and oleic acid monodisperse particles. Aerosol Sci. Technol.
30 8: 279-281.
31
32 Broddin, G.; Cautreels, W.; Van Cauwenberghe, K. (1980) On the aliphatic and polyaromatic hydrocarbon levels
33 in urban and background aerosols from Belgium and the Netherlands. Atmos. Environ. 14: 895-910.
34
35 Brosset, C.; Perm, M. (1978) Man-made airborne acidity and its determination. Atmos. Environ. 12: 909-916.
36
37 Brown, R. C.; Wake, D.; Thorpe, A.; Hemingway, M. A.; Roff, M. W. (1994) Preliminary assessment of a
38 device for passive sampling of airborne paniculate. Ann. Occup. Hyg. 38: 303-318.
39
40 Bruynseels, F.; Storms, H.; Van Grieken, R. (1988) Atmos. Environ. 22: 2593-2602.
41
42 Buckley, T. J.; Waldman, J. M.; Freeman, N. C. G.; Lioy, P. J.; Marple, V. A.; Turner, W. A. (1991)
43 Calibration, intersampler comparison, and field application of a new PM-10 personal air-sampling
44 impactor. Aerosol Sci. Technol. 14: 380-387.
45
46 Buettner, H. (1990) Measurement of the size distribution of fine nonspherical particles with a light-scattering
47 particle counter. Aerosol Sci. Technol. 12: 413-421.
48
49 Butler, F. E.; Jungers, R. H.; Porter, L. F.; Riley, A. E.; Toth, F. J. (1978) Analysis of air particulates by ion
50 chromatography: comparison with accepted methods. In: Sawicki, E.; Mulik, J. D.; Wittgenstein, E.,
51 eds. Ion chromatographic analysis of environmental pollutants. Ann Arbor, MI: Ann Arbor Science
52 Publishers, Inc.; pp. 65-76.
53
April 1995 4-102 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cadle, S. H.; Groblicki, P. J. (1982) An evaluation of methods for the determination of organic and elemental
2 carbon in paniculate samples. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate carbon: atmospheric
3 life cycles. New York, NY: Plenum Press; pp. 89-109.
4
5 Cadle, S. H.; Groblicki, P. J.; Stroup, D. P. (1980) An automated carbon analyzer for paniculate samples. Anal.
6 Chem. 52: 2201-2206.
7
8 Cadle, S. H.; Nebel, G. J.; Williams, R. L. (1980) Measurements of unregulated emissions from General
9 Motors' light duty vehicles. Soc. Automotive Engineers Trans. 87: 2381-2401.
10
11 Cadle, S. H.; Groblicki, P. J.; Mulawa, P. A. (1983) Problems in the sampling and analysis of carbon particles.
12 Atmos. Environ. 17: 593-600.
13
14 Cahill, T. A. (1975) Ion-excited X-ray analysis of environmental samples. In: Ziegler, J., ed. New uses for ion
15 accelerators. New York, NY: Plenum Press.
16
17 Cahill, T. A. (1980) Proton microbes and particle-induced X-ray analytical systems. Ann. Rev. Nucl, Particle
18 Sci. 30: 211-252.
19
20 Cahill, T. A. (1990) Particle-induced X-ray emission. In: Mills, K., ed. Metals handbook. 9th ed. American
21 Society for metals; pp. 102-108.
22
23 Cahill, T. A.; Wakabayashi, P. (1993) Compositional analysis of size-segregated aerosol samples. In:
24 Newman, L., ed. Measurement challenges in atmospheric chemistry. Washington, DC: American
25 Chemical Society; pp. 211 -228.
26
27 Cahill, T. A.; Eldred, R. A.; Kusko, B. H.; Feeney, P. J.; Malm, W. C. (1987) Concentrations of natural
28 hydrocarbon particles at National Park Service sites as derived from mass/hydrogen/sulfur correlations.
29 In: Bhardwaja, P. S., ed. Visibility protection: research and policy aspects. Pittsburgh, PA: Air Pollution
30 Control Association; pp. 407-417.
31
32 Cahill, T. A.; Eldred, R. A.; Motallebi, N.; Malm, W. C. (1989) Indirect measurement of hydrocarbon aerosols
33 across the United States by nonsulfate hydrogen-remaining gravimetric mass correlations. Aerosol Sci.
34 Technol. 10: 421-429.
35
36 Cahill, T. A.; Eldred, R. A.; Feeney, P. J.; Beveridge, P. J.; Wilkinson, L. K. (1990) The stacked filter unit
37 revisited. In: Mathai, C. V., ed. Visibility and fine particles: transactions. Pittsburgh, PA: Air & Waste
38 Management Association; pp. 213-218.
39
40 Cahill, T,; Gill, T.; Gillette, D.; Gearhart, E.; Reid, J.; Yau, M. (1994) Generation, characterization and
41 transport of Owens (dry) Lake dust. Sacramento, CA: California Environmental Protection Agency, Air
42 Resources Board; contract report no. A132105.
43
44 Casuccio, G. S.; Janocko, P. B.; Lee, R. J.; Kelly, J. F.; Dattner, S. L.; Mgebroff, J. S. (1983) The use of
45 computer controlled scanning electron microscopy in environmental studies. J. Air Pollut. Control Assoc.
46 33: 937-943.
47
48 Charlson, R. J.; Ahlquist, N. C.; Horvath, H. (1968) On the generality of correlation of atmospheric aerosol
49 mass concentration and light scatter. Atmos. Environ. 2: 455-464.
50
51 Chen, B. T.; Cheng, Y. S.; Yeh, H. C. (1985) Performance of a TSI aerodynamic particle sizer. Aerosol Sci.
52 Technol. 4: 89-97.
53
April 1995 4403 DRAFT-DO NOT QUOTE OR CITE
-------
1 Code of Federal Regulations. (1979) National primary and secondary ambient air quality standards; appendix
2 B—reference method for the determination of suspended particulates in the atmosphere (high volume
3 method). C.F.R. 40: § 50; pp. 12-17.
4
5 Commins, B. T. (1962) Interim report on the study of techniques for determination of polycyclic aromatic
6 hydrocarbons in air. In: Sawicki, E.; Cassel, K., Jr., eds. Analysis of carcinogenic air pollutants:
7 [proceedings of a symposium]; August 1961; Cincinnati, OH. Bethesda, MD: National Cancer Institute;
8 pp. 225-233. (National Cancer Institute monograph no. 9).
9
10 Conner, T. L.; Miller, J. L.; Willis, R. D.; Kellogg, R. D.; Dann, T. F. (1993) Source apportionment of fine
11 and coarse particles in southern Ontario, Canada. Presented at: 86th annual meeting and exhibition of the
12 Air & Waste Management Association; June; Denver, CO. Pittsburgh, PA: Air & Waste Management
13 Association; paper no. 93-TP-58.05.
14
15 Cooper, D. W.; Guttrich, G. L. (1981) A study of the cut diameter concept for interpreting particle sizing data.
16 Atmos. Environ. 15: 1699-1707.
17
18 Cornille, P.; Maenhaut, W. (1990) Atmos. Environ. Part A 24: 1083-1093.
19
20 Gotham, W. E.; Bidleman, T. F. (1992) Laboratory investigations of the partitioning of organochlorine
21 compounds between the gas phase and atmospheric aerosols on glass fiber filters. Environ. Sci. Technol.
22 26: 469-478.
23
24 Countess, R. J. (1990) Inter-laboratory analyses of carbonaceous aerosol samples. Aerosol Sci. Technol.
25 12: 114-121.
26
27 Courtney, W. J.; Shaw, R. W.; Dzubay, T. G. (1982) Precision and accuracy of a beta-gauge for aerosol mass
28 determinations. Environ. Sci. Technol. 16: 236-239.
29
30 Coutant, R. W.; Brown, L.; Chuang, J. C.; Riggin, R. M.; Lewis, R. G. (1988) Phase distribution and artifact
31 formation in ambient air sampling for polynuclear aromatic hydrocarbons. Atmos. Environ. 22: 403-409.
32
33 Coutant, R. W.; Callahan, P. J.; Kuhlman, M. R.; Lewis, R. G. (1989) Design and performance of a
34 high-volume compound annular denuder. Atmos. Environ. 23: 2205-2211.
35
36 Coutant, R. W.; Callahan, P. J.; Chuang, J. C.; Lewis, R. G. (1992) Efficiency of silicone-grease-coated
37 denuders for collection of polynuclear aromatic hydrocarbons. Atmos. Environ. Part A 26: 2831-2834.
38
39 Criss, J. W. (1976) Particle size and composition effects in X-ray fluorescence analysis of pollution samples.
40 Anal. Chem. 48: 179-186.
41
42 Crutcher, E. R. (1982) Light microscopy as an analytical approach to receptor modeling. In: Dattner, S. L.;
43 Hopke, P. K., eds. Receptor models applied to contemporary pollution problems. Pittsburgh, PA: Air
44 Pollution Control Association; pp. 266-284.
45
46 Currie, L. A. (1982) Contemporary paniculate carbon. In: Wolff, G. T.; Klimisch, R. L., eds. Paniculate
47 carbon: atmospheric life cycle. Chelsea, MI: Lewis Publishers, Inc.; pp. 47-65.
48
49 Daisey, J. M. (1987) Chemical composition of inhalable paniculate matter—seasonal and intersite comparisons.
50 In: Lioy, P. J.; Daisey, J. M., eds. Toxic air pollution: a comprehensive study of non-criteria air
51 pollutants. Chelsea, MI: Lewis Publishers, Inc.; pp. 45-65.
52
53
April 1995 4-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 Daisey, J. M.; Kneip, T. J. (1981) Atmospheric paniculate organic matter: multivariate models for identifying
2 sources and estimating their contributions to the ambient aerosol. In: Macias, E. S.; Hopke, P. K., eds.
3 Atmospheric aerosol: source/air quality relationships; based on a symposium jointly sponsored by the
4 Divisions of Nuclear Chemistry and Technology and Environmental Chemistry at the 180th national
5 meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington, DC: American
6 Chemical Society; pp. 197-221. (Comstock, M. J., ed. ACS symposium series: 167).
7
8 Daisey, J. M.; McCaffrey, R. J.; Gallagher, R. A. (1981) Polycyclic aromatic hydrocarbons and total extractable
9 paniculate organic matter in the Arctic aerosol. Atmos. Environ. 15: 1352-1363.
10
11 Daisey, J. M.; Herschman, R. J.; Kneip, T. J. (1982) Ambient levels of paniculate organic matter in New York
12 City in winter and summer. Atmos. Environ. 16: 2161-2168.
13
14 Dams, R.; Robbins, J. A.; Rahn, K. A.; Winchester, J. W. (1970) Non-destructive neutron activation analysis of
15 air pollution particulates. Anal. Chem. 42: 861-867.
16
17 Davis, B. L. (1978) Additional suggestions for X-ray quantitative analysis of high-volume filters. Atmos.
18 Environ. 12: 2403-2406.
19
20 Davis, B. L. (1980) Standardless' X-ray diffraction quantitative analysis. Atmos. Environ. 14: 217-220. -
21
22 Delia Fiorentina, H.; De Wiest, F.; De Graeve, J. (1975) Determination par spectrometrie infrarouge de la
23 matiere organique non volatile associee aux particules en suspension dans 1'air—II. Facteurs influencant
24 1'indice aliphatique. [Determination of the nonvolatile organic matter associated with suspended panicles
25 by infrared spectrometry—II. Factors influencing the aliphatic index]. Atmos. Environ. 9: 517-522.
26
27 Derde, M. P.; Buydens, L.; Guns, C.; Massart, D. L.; Hopke, P. K. (1987) Comparison of rule-building expert
28 systems with pattern recognition for the classification of analytical data. Anal. Chem. 59: 1868-1871.
29
30 Divita, F., Jr. (1993) Size distributions and sources od submicrometer atmospheric particles in Washington, D.C.
31 and Philadelphia, PA [dissertation]. College Park, MD: University of Maryland. Available from: Xerox
32 University Microfilms, Ann Arbor, MI; publication no. AAD94-25018.
33
34 Dod, R. L.; Rosen, H.; Novakov, T. (1979) Atmospheric aerosol research annual report for 1977-1978.
35 Berkeley, CA: Lawrence Berkeley Laboratories; document no. 8696.
36
37 Drane, E. A.; Branton, D. G.; Tysinger, S. H.; Courtney, W. J. (1983) Data processing procedures for
38 elemental analysis of atmospheric aerosols by X-ray fluorescence. Research Triangle Park, NC: Northrop
39 Services, Inc.; document no. TR-83-01.
40
41 Durham, J. L.; Wilson, W. E.; Bailey, E. B. (1978) Application of an SO2 denuder for continuous measurement
42 of sulfur in submicrometric aerosols. Atmos. Environ. 12: 883-886.
43
44 Dzubay, T. G. (1986) Analysis of aerosol samples by X-ray fluorescence. Research Triangle Park, NC: U.S.
45 Environmental Protection Agency.
46
47 Dzubay, T. G.; Clubb, K. W. (1981) Comparison of telephotometer measurements of extinction coefficients with
48 scattering and absorption coefficients. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes
49 and visibility: measurements and model components, proceedings of the symposium; November 1980;
50 Grand Canyon National Park, AZ. Atmos. Environ. 15: 2617-2624.
51
52 Dzubay, T. G.; Mamane, Y. (1989) Use of electron microscopy data in receptor models for PM-10. Atmos.
53 Environ. 23: 467-476.
54
April 1995 4.105 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dzubay, T. G.; Nelson, R. O. (1975) Self absorption corrections for X-ray fluorescence analysis of aerosols.
2 Adv. X-ray Anal. 18: 619-631.
3
4 Dzubay, T. G.; Stevens, R. K. (1975) Ambient air analysis with dichotomous sampler and X-ray fluorescence
5 spectrometer. Environ. Sci. Technol. 9: 663-668.
6
7 Dzubay, T. G.; Morosoff, N.; Whitaker, G. L.; Yasuda, H. (1981) Evaluation of polymer films as standards for
8 X-ray fluorescence spectrometers. In: Electron microscopy and X-ray applications to environmental and
9 occupational health analysis. Ann Arbor, MI: Ann Arbor Science Publishers, Inc.
10
11 Dzubay, T. G.; Stevens, R. K.; Balfour, W. D.; Williamson, H. J.; Cooper, J. A.; Core, J. E.;
12 De Cesar, R. T.; Crutcher, E. R.; Datiner, S. L.; Davis, B. L.; Heisler, S. L.; Shah,'J. J.;
13 Hopke, P. K.; Johnson, D. L. (1984) Interlaboratory comparison of receptor model results for Houston
14 aerosol. Atmos. Environ. 18: 1555-1566.
15
16 Eatough, D. J. (1995) BOSS, the Brigham Young University Organic Sampling System: determination of
17 paniculate carbonaceous material using diffision denuder sampling technology. In: Lane, D.,
18 ed. Gas/phase partition measurements of atmospheric organic compounds. New York, NY: Gordon and
19 Breach Science Publishers; in press.
20
21 Eatough, D. J.; Sedar, B.; Lewis, L.; Hansen, L. D.; Lewis, E. A.; Farber, R. J. (1989) Determination of
22 semivolatile organic compounds in particles in the Grand Canyon area. Aerosol Sci. Technol.
23 10: 438-449.
24
25 Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. (1993)
26 A multiple-system, multi-channel diffusion denuder sampler for the determination of fine-particulate
27 organic material in the atmosphere. Atmos. Environ. Part A 27: 1213-1219.
28
29 Eatougrs D.J.; Tang, H.; Cui, W.; Machir, J. (1995) Determination of the size distribution and chemical
30 composition of fine paniculate semi-volatile organic material in urban environments using diffusion
31 denuder technology. Inhalation Toxicol. 7: 691-710.
32
33 Eichmann, R.; Neuling, P.; Ketseridis, G.; Hahn, J.; Jaenicke, R.; Junge, C. (1979) n-alkane studies in the
34 troposphere—I. Gas and paniculate concentrations in North Atlantic air. Atmos. Environ. 13: 587-599.
35
36 Eldering, A.; Cass, G. R.; Moon, K. C. (1994) An air monitoring network using continuous particle size
37 distribution monitors: connecting pollutant properties to visibility via Mie scattering calculations. Atmos.
38 Environ. 28: 2733-2749.
39
40 Eldred, R. A.; Cahill, T. A. (1994) Trends in elemental concentrations of fine particles at remote sites in the
41 United States of America. Atmos. Environ. 28: 1009-1019.
42
43 Eldred, R. A.; Cahill, T. A.; Feeney, P. (1989) Regional patterns of selenium and other trace elements in the
44 IMPROVE network.
45
46 Eldred, R. A.; Cahill, T. A.; Malm, W. C. (1989) Composition of fine aerosols during the summer of 1987
47 determined by the IMPROVE and National Service networks. In: Proceedings of the 8th annual AAAR
48 meeting; Reno, NV.
49
50 Eldred, R. A.; Cahill, T. A.; Feeney, P. J. (1993) Comparison of independent measurements of sulfur and
51 sulfate in the IMPROVE network. Presented at: 86th annual meeting and exhibition of the Air & Waste
52 Management Association; June; Denver, CO. Pittsburgh, PA: Air & Waste Management Association;
53 paper no. 93-RA-l 10.02.
54
April 1995 4-106 DRAFT-DO NOT QUOTE OR CITE
-------
1 Eldred, R. A.; Cahill, T. A.; Flocchini, R. G. (1994) Composition of PM10 and PM2 5 aerosols in the
2 IMPROVE network. Proceedings of the international specialty conference on aerosol and atmospheric
3 optics: radiative balance and visual air quality, volume A. Air Waste: submitted.
4
5 Engelbrecht, D. R.; Cahill, T. A.; Feeney, P. J. (1980) Electrostatic effects on gravimetric analysis of
6 membrane filters. J. Air Pollut. Control Assoc. 30: 391-392.
j 7
; 8 Federal Register. (1978) Reference method for the determination of lead in suspended paniculate matter collected
| 9 from ambient air. F. R. 43: 46258.
! 10
11 Federal Register. (1987) Procedures for testing performance characteristic of methods for PM10. F. R.
12 52: 24724.
13
14 Federal Register. (1987) Revisions to the national ambient air quality standards for paniculate matter. F. R. (July
15 1) 52: 24634-24669.
16
17 Feeney, P. J.; Cahill, T. A.; Olivera, J.; Guidara, R. (1984) Gravimetric determination of mass on lightly loaded
18 membrane filters. J. Air Pollut. Control Assoc. 34: 367-378.
19
20 Ferek, R. J.; Lazrus, A. L.; Haagenson, P. L.; Winchester, J. W. (1983) Strong and weak acidity of aerosols
21 collected over the northeastern United States. Environ. Sci. Technol. 17: 315-324.
22
23 Ferek, R. J.; Hegg, D. A.; Hobbs, P. V.; Orphan, M.; Farber, R. J.; Bhardwaja, P. S. (1987) Airborne
24 measurements of atmospheric trace constituents in the Grand Canyon region. In: Mathai, C. V., ed.
25 Visibility and fine particles: transactions. Pittsburgh, PA: Air & Waste Management Association; p. 328.
26
27 Perm, M. (1986) Concentration measurements and equilibrium studies of ammonium, nitrate and sulphur species
28 in air and precipitation [Ph.D. thesis]. Goteborg, Sweden: Goteborg University, Department of Inorganic
29 Chemistry.
30
31 Fernandez, F. J. (1989) Atomic absorption spectroscopy. In: Lodge, J. P., Jr., ed. Methods of air sampling and
32 analysis. 3rd ed. Chelsea, MI: Lewis Publishers; pp. 143-150.
33
34 Finlayson-Pitts, B. J.; Pitts, J. N., Jr. (1986) Atmospheric chemistry: fundamentals and experimental techniques.
35 New York, NY: John Wiley & Sons.
36
37 Fitz, D. R. (1990) Reduction of the positive organic artifact on quartz filters. Aerosol. Sci. Technol.
38 12: 142-148.
39
40 Flessel, P.; Wang, Y. Y.; Chang, K. I.; Wesolowski, J. J.; Guirguis, G. N.; Kim, I. S.; Levaggi, D.; Wayman,
41 S. (1991) Seasonal variations and trends in concentrations of filter-collected polycyclic aromatic
42 hydrocarbons (PAH) and mutagenic activity in the San Francisco Bay area. J. Air Waste Manage. Assoc.
43 41: 276-281.
44
45 Fletcher, R. A.; Small, J. A. (1993) Analysis of individual collected particles. In: Willeke, K.; Baron, P. A.,
46 eds. Aerosol measurement principles, techniques and applications. New York, NY: Van Nostrand
47 Reinhold; pp. 260-295.
48
49 Foreman, W. T.; Bidleman, T. F. (1990) Semivolatile organic compounds in the ambient air of Denver,
50 Colorado. Atmos. Environ. Part A 2405-2416.
51
52 Forrest, J.; Spandau, D. J.; Tanner, R. L.; Newman, L. (1982) Determination of atmospheric nitrate and nitric
53 acid employing a diffusion denuder with a filter pack. Atmos. Environ. 16: 1473-1485
54
April 1995 4.107 DRAFT-DO NOT QUOTE OR CITE
-------
1 Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John Wiley
2 & Sons, Inc.
3
4 Fung, K. K. (1990) Paniculate carbon speciation by MnO2 oxidation. Aerosol Sci. Technol. 12: 122-127.
5
6 .Fung, K. K.; Heisler, S. L.; Price, A.; Nuesca, B. V.; Mueller, P. K. (1979) Comparison of ion
7 chromatography and automated wet chemical methods for analysis of sulfate and nitrate in ambient
8 paniculate filter samples. In: Mulik, J. D.; Sawicki, E., eds. Ion chromatographic analysis of
9 environmental pollutants: volume 2. Ann Arbor, MI: Ann Arbor Science Publishers Inc.; pp. 203-209.
10
11 Gagosian, R. B.; Peltzer, E. T.; Zafiriou, O. C. (1981) Atmospheric transport of continentally derived lipids to
12 the tropical North Pacific. Nature (London) 291: 321-324.
13
14 Galloway, J. N.; Cosby, B. S.; Likens, G. E. (1979) Acid precipitation: measurement of pH and acidity.
15 Limnol. Oceanogr. 24: 1161.
16
17 Gebhart, J. (1993) Optical direct-reading techniques: light intensity systems. In: Willeke, K.; Baron, P. A., eds.
18 Aerosol measurement principles, techniques and applications. New York, NY: Van Nostrand Reinhold;
19 pp. 313-344.
20
21 Gerber, H. E. (1982) Optical techniques for the measurement of light absorption by particulates. In:
22 Wolff, G. T.; Klimisch, B. L., eds. Particulate carbon: atmospheric life cycles. New York, NY: Plenum
23 Publishing Corporation; pp. 1-411.
24
25 Gordon, R. J. (1974) Solvent selection in entrance of paniculate matter. Atmos. Environ. 8: 189-191.
26
27 Griffin, J. J.; Goldberg, E. D. (1979) Morphologies and origin of elemental carbon in the environment. Science
28 (Washington, DC) 206: 563-565.
29
30 Grinshp'un, S.; Willeke, K.; Kalatoor, S. (1993) A general equation for aerosol aspiration by thin-walled
31 sampling probes in calm and moving air. Atmos. Environ. Part A 27: 1459-1470.
32
33 Grosjean, D. (1975) Solvent extraction and carbon and organic carbon determination in atmospheric paniculate
34 matter: the organic extraction-organic carbon analyzer technique. Anal. Chem. 47: 797.
35
36 Gundel, L. A.; Stevens, R. K.; Daisey, J. M.; Lee, V.; Mahanama, K. R. R. (1992) Annular denuders for
37 sampling semi-volatile polycyclic aromatic hydrocarbons and other organic species. Presented at:
38 llth annual meeting of the American Association for Aerosol Research; October; San Francisco, CA.
39
40 Gundel, L. A.; Mahanama, K. R. R.; Daisey, J. M. (1994) Fractionation of polar organic extracts of airborne
41 paniculate matter using cyanopropyl-bonded silica in solid -phase extraction. J. Chromatogr.: in press.
42
43 Hall, D. J.; Mark, D.; Upton, S. L. (1992) A new large wind tunnel for dust and aerosol studies. J. Aerosol Sci.
44 23(suppl. 1): S591-S594.
45
46 Harrison, R. M.; Kitto, A. M. N. (1990) Field intercomparison of filter pack and denuder sampling methods for
47 reactive gaseous and paniculate pollutants. Atmos. Environ. Part A 24: 2633-2640.
48
49 Hart, K. M.; Pankow, J. F. (1990) Comparison of n-alkane and PAH concentrarions collected on quartz fiber
50 and Teflon membrane filters in an urban environment. J. Aerosol Sci. 21(suppl. 1): S377-S380.
51
52 Hearle, J. W. S.; Sparrow, J. R.; Cross, P. M. (1972) The use of the scanning electron microscope. London,
53 United Kingdom: Pergamon Press.
54
April 1995 4-108 DRAFT-DO NOT QUOTE OR CITE
-------
1 Heitbrink, W. A.; Baron, P. A.; Willeke, K. (1991) Coincidence in time-of-flight aerosol spectrometers:
2 phantom particle creation. Aerosol Sci. Technol. 14: 112-126.
3
4 Henderson Sellers, A.; McGuffie, K. (1989) Sulphate aerosols and climate: scientific correspondence. Nature
5 (London) 340: 436-437.
6
7 Henry, R. C.; Lewis, C. W.; Hopke, P. K.; Williamson, H. J. (1984) Review of receptor model fundamentals.
8 Atmos. Environ. 18: 1507-1515.
9
10 Hering, S. V., ed. (1989) Air sampling instruments for evaluation of atmospheric contaminants. 7th ed.
11 Cincinnati, OH: American Conference of Governmental Industrial Hygienists.
12
13 Hering, S. V. (1994) Particle measurements for the children's health study: development of a two-week sampler.
14 Presented at: International Society for Environmental Epidemiology/International Society for Exposure
15 Analysis joint conference; September; Research Triangle Park, NC. paper no. 259.
16
17 Hering, S. V.; Appel, B. R.; Cheng, W.; Salaymeh, F.; Cadle, S. H.; Mulawa, P. A.; Cahill, T. A.;
18 Eldred, R. A.; Surovik, M.; Fitz, D.; Howes, J. E.; Knapp, K. T.; Stockburger, L.; Turpin, B. J.;
19 Huntzicker, J. J.; Zhang, X.-Q.; McMurry, P. H. (1990) Comparison of sampling methods for
20 carbonaceous aerosols in ambient air. Aerosol Sci. Technol. 12: 200-213.
21
22 Hildemann, L. M.; Markowski, G. R.; Cass, G. R. (1991) Chemical composition of emissions from urban
23 sources of fine organic aerosol. Environ. Sci. Technol. 25: 744-759.
24
25 Hillamo, R. E.; Kauppinen, E. I. (1991) On the performance of the Berner low pressure impactor. Aerosol Sci.
26 Technol. 14: 33-47.
27
28 Hinds, W. C. (n.d.) Basis for particle size-selective sampling for wood dust. In: Advances in air sampling.
29
30 Hinds,*W. C. (1982) Aerosol technology. New York, NY: John Wiley and Sons.
31
32 Hinds, W. C.; Kraske, G. (1986) Performance of PMS model LAS-X optical particle counter. J. Aerosol Sci.
33 17: 67-72.
34
35 Hirose, K.; Sugimura, Y. (1984) Excess 228th in the airborne dust: an indicator of continental dust from the east
36 Asian deserts. Earth Planet. Sci. Lett. 70: 110-114.
37
38 Hoffman, A. J.; Purdue, L. J.; Rehme, K. A.; Holland, D. M. (1988) 1987 PM10 sampler intercomparison
39 study. In: Mathai, C. V.; Stonefield, D. H., eds. PM10—implementation of standards: transactions.
40 Pittsburgh, PA: Air Pollution Control Association; pp. 138-149.
41
42 Hollander, W. (1990) Proposed performance criteria for samplers of total suspended paniculate matter. Atmos.
43 Environ. Part A 24: 173-177.
44
45 Hollander, W. (1992) Ambient aerosol sampling without pumps. J. Aerosol Sci. 23(suppl. 1): S619-S622.
46
47 Hopke, P. K. (1985) Receptor modeling in environmental chemistry. New York, NY: John Wiley and Sons.
48
49 Hopke, P. K., Jr. (1985) The use of chemometrics in apportionment of air pollution sources. Trends Anal.
50 Chem. 4: 4.
51
52 Hopke, P. K.; Casuccio, G. S. (1991) Scanning electron microscopy. In: Hopke, P. K., ed. Receptor modeling
53 for air quality management: v. 7. Amsterdam, The Netherlands: Elsevier; pp. 149-212.
54
April 1995 4_109 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hornung, R. W.; Reed, L. D. (1990) Estimation of average concentration in the presence of non-detectable
2 values. Appl. Occup. Environ. Hyg. 5: 46-51.
3
4 Ingham, D. B.; Yan, B. (1994) The effect of a cylindrical backing body on the sampling efficiency of a
5 cylindrical sampler. J. Aerosol Sci. 25: 535-541.
6
7 International Standards Organization. (1983) Air quality: particle size fraction definitions for health-related
8 sampling. Geneva, Switzerland: International Standards Organization; technical report ISO/TR7708-1983.
9
10 Jackson, M. L. (1981) Oxygen isotopic ratios in quartz as an indicator of provenance of dust. Geol. Soc. Am.
11 Spec. Pap.: 27-36.
12
13 Jaklevic, J. M.; Loo, B. W.; Goulding, F. S. (1977) Photon-induced X-ray fluorescence analysis using
14 energy-dispersive detection and dichotomous sampler. In: Dzubay, T. G., ed. X-ray fluorescence analysis
15 of environmental samples. 2nd ed. Ann Arbor, MI: Ann Arbor Science Publishers; pp. 3-18.
16
17 Jaklevic, J. M.; Gatti, R. C.; Goulding, F. S.; Loo, B. W. (1981) A |8-gauge method applied to aerosol samples.
18 Environ. Sci. Technol. 15: 680-686.
19
20 Janocko, P. B.; Casuccio, G. S.; Dattner, S. L.; Johnson, D. L.; Crutcher, E. R. (1982) The El Paso Airshed:
21 source apportionment using complementary analyses and receptor models. In: Dattner, S. L.;
22 Hopke, P. K., eds. Receptor models applied to contemporary pollution problems. Pittsburgh, PA: Air
23 Pollution Control Association.
24
25 Javitz, H. S.; Watson, J. G.; Robinson, N. (1988) Performance of the chemical mass balance model with
26 simulated local scale aerosols. Atmos. Environ. 22: 2309-2322.
27
28 Jenson, P. A.; O'Brien, D. (1993) Industrial hygiene. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement
29 principles, techniques and applications. New York, NY: Van Nostrand Reinhold; p. 540.
30
31 John, W.; Wall, S. M. (1983) Aerosol testing techniques for size-selective samplers. J. Aerosol Sci.
32 14: 713-727.
33
34 John, W.; Wang, H.-C. (1991) Laboratory testing method for PM-10 samplers: lowered effectiveness from
35 particle loading. Aerosol Sci. Technol. 14: 93-101.
36
37 John, W.; Hering, S.; Reischl, S.; Sasaki, G.; Goren, S. (1983) Characterization of nuclepore filters with large
38 pore size: II. filtration properties. Atmos. Environ. 17: 373-382.
39
40 John, W.; Wall, S. M.; Ondo, J. L. (1988) A new method for nitric acid and nitrate aerosol measurement using
41 the dichotomous sampler. Atmos. Environ. 22: 1627-1635.
42
43 John, W.; Wall, S. M.; Ondo, J. L.; Winklmayr, W. (1990) Modes in the size distributions of atmospheric
44 inorganic aerosol. Atmos. Environ. Part A 24: 2349-2359.
45
46 John, W.; Winklmayr, W.; Wang, H.-C. (1991) Particle deagglomeration and reentrainment in a PM-10 sampler.
47 Aerosol Sci. Technol. 14: 165-176.
48
49 Johnson, D. L.; Mclntyre, B. L. (1982) A particle class balance receptor model for aerosol apportionment in
50 Syracuse, NY. In: Receptor models applied to contemporary pollution problems: proceedings. Pittsburgh,
51 PA: Air Pollution Control Association; pp. 238-247.
52
53 Johnson, D. L.; Mclntyre, B. L.; Fortmann, R.; Stevens, R. K.; Hanna, R. B. (1981) A chemical element
54 comparison of individual particle analysis and bulk analysis methods. Scanning Electron Microsc. 1: 469.
April 1995 4-110 DRAFT-DO NOT QUOTE OR CITE
-------
1 Johnson, R. L.; Shah, J. J.; Gary, R. A.; Huntzicker, J. J. (1981) An automated thermal-optical method for the
2 analysis of carbonaceous aerosol. In: Macias, E. S.; Hopke, P. K., eds. Atmospheric aerosol: source/air
3 quality relationships; based on a symposium jointly sponsored by the Divisions of Nuclear Chemistry and
4 Technology at the 180th national meeting of the American Chemical Society; August 1980; Las Vegas,
5 NV. Washington, DC: American Chemical Society; pp. 223-233. (Comstock, M. J., ed. ACS
6 ' symposium series: 167).
7
8 Junge, C. E. (1977) Basic considerations about trace constituents in the atmosphere as related to the fate of global
9 pollutants. In: Suffet, I. H., ed. Fate of pollutants in the air and water environments: part I, mechanism
10 of interaction between environments and mathematical modeling and the physical fate of pollutants,
11 papers from the 165th national American Chemical Society meeting; April 1975; Philadelphia, PA. New
12 York, NY: John Wiley & Sons; pp. 7-25. (Advances in environmental science and technology: v. 8).
13
14 Kado, N. Y.; Kuzmicky, P. A.; Hsieh, D. P. H.; Gundel, L. A.; Daisey, J. M.; Mahanama, K. R. R.;
15 Schuetzle, D. (1989) Mutagenic activity of the polar organic matter and its fractions from airborne
16 particles. Presented at: AAAR meeting; October; Traverse City, MI.
17
18 Kaupp, H.; Umlauf, G. (1992) Atmospheric gas-particle partitioning of organic compounds: comparison of
19 sampling methods. Atmos. Environ. Part A 26: 2259-2267.
20
21 Kim, Y. J.; Boatman, J. F. (1990) The effects that the optical properties have on atmospheric aerosol
22 measurements with optical particle spectrometers. J. Aerosol Sci. 21(suppl. 1): S551-S554.
23
24 Kim, D.; Hopke, P. K. (1988) Classification of individual particles based on computer-controlled scanning
25 electron microscopy data. Aerosol Sci. Technol. 9: 133-151.
26
27 Kim, D.; Hopke, P. K. (1988) Source apportionment of the El Paso aerosol by particle class balance analysis.
28 Aerosol Sci. Technol. 9: 221-235.
29
30 Kim, D. S.; Hopke, P. K.; Massart, D. L.; Kaufman, L.; Casuccio, G. S. (1987) Analysis of CCSEM auto
31 emission data. Sci. Total Environ. 59: 141-155.
32
33 Klippel, W.; Warneck, P. (1980) The formaldehyde content of the atmospheric aerosol. Atmos. Environ.
34 14: 809-818.
35
36 Knollenberg, R. G.; Veal, D. L. (1992) Optical particle monitors, counters, and spectrometers. J. EIS
37 (March/April): 64-81.
38
39 Konig, J.; Funcke, W.; Balfanz, E.; Grosch, B.; Pott, F. (1980) Testing a high volume air sampler for
40 quantitative collection of polycyclic aromatic hydrocarbons. Atmos. Environ. 14: 609-613.
41
42 Koutrakis, P.; Wolfson, J. M.; Slater, J. L.; Brauer, M.; Spengler, J. D.; Stevens, R. K.; Stone, C. L. (1988)
43 Evaluation of an annular denuder/filter pack system to collect acidic aerosols and gases. Environ. Sci.
44 Technol. 22: 1463-1468.
45
46 Koutrakis, P.; Wolfson, J. M.; Brauer, M.; Spengler, J. D. (1990) Design of a glass impactor for an annular
47 denuder/filter pack system. Aerosol Sci. Technol. 12: 607-612.
48
49 Lamb, S. I.; Petrowski, C.; Kaplan, I. R.; Simoneit, B. R. T. (1980) Organic compounds in urban atmospheres:
50 a review of distribution, collection and analysis. J. Air Pollut. Control Assoc. 30: 1098-1115.
51
52 Lane, D. A.; Johnson, N. D.; Hanley, M. J. J.; Schroeder, W. H.; Ord, D. T. (1992) Gas- and particle-phase
53 concentrations of a-hexachlorocyclohexane, g-hexachlorocyclohexane and hexachlorobenzene in Ontario
54 air. Environ. Sci. Technol. 26: 126-132.
April 1995 4_1H DRAFT-DO NOT QUOTE OR CITE
-------
1 Larson, T. V.; Yuen, P. F.; Maykut, N. N. (1992) Weekly composite sampling of PM10 for total mass and trace
2 elemental analysis. In: Chow, J. C.; Ono, D. M., eds. PM10 standards and non-traditional paniculate
3 source controls: proceedings of an AWMA/EPA international specialty conference. Pittsburgh, PA: Air
4 & Waste Management Association; pp. 39-50.
5
6 -Lee, R. E., Jr.; Duffield, F. V. P. (1979) Sources of environmentally important metals in the atmosphere. In:
7 Risby, T. H., ed. Ultrace metal analysis in biological sciences and environment: proceedings of the 174th
8 meeting of the American Chemical Society; August 1977; Chicago, IL. Washington, DC: American
9 Chemical Society; pp. 146-171. (Advances in chemistry series 172).
10
11 Lee, R. J.; Fisher, R. M. (1980) Quantitative characterization of particulates by scanning and high voltage
12 electron microscopy. Washington, DC: National Bureau of Standards; special publication.
13
14 Lee, R. J.; Kelly, J. F. (1980) Overview of SEM-based automated image analysis. Scanning Electron Microsc.
15 1: 303.
16
17 Lee, R. J.; Fasiska, E. J.; Janocko, P. B.; McFarland, D. G.; Penkala, S. J. (1979) Electron-beam paniculate
18 analysis. Ind. Res. Dev.
19
20 Lee, P. H.; Hoffer, T. E.; Schorran, D. E.; Ellis, E. C.; Moyer, J. W. (1981) Laser transmissometer—a
21 description. Sci. Total Environ. 23: 321.
22
23 Lee, R. J.; Spitzig, W. A.; Kelly, J. F.; Fisher, R. M. (1981) Quantitative metallography by computer
24 controlled scanning electron microscopy. J. Metals 33: 3.
25
26 Li, C. K.; Kamens, R. M. (1993) The use of polycyclic aromatic hydrocarbons as a source signature in receptor
27 modeling. Atmos. Environ. Part A 27: 523-532.
28
29 Liden, G.; Kenny, L. C. (1991) Comparison of measured respirable dust sampler penetration curves with
30 sampling conventions. Ann. Occup. Hyg. 35: 485-504.
31
32 Liden, G.; Kenny, L. C. (1992) The performance of respirable dust samplers: sampler bias, precision and
33 inaccuracy. Ann. Occup. Hyg. 36: 1-22.
34
35 Ligocki, M. P.; Pankow, J. F. (1989) Measurements of the gas/particle distributions of atmospheric organic
36 compounds. Environ. Sci. Technol. 23: 75-83.
37
38 Lindskog, A.; Brorstrom-Lunden, E.; Sjodin, A. (1985) Transformation of reactive PAH on particles by
39 exposure to oxidized nitrogen compounds and ozone. Environ. Int. 11: 125-130.
40
41 Lioy, P. J.; Wainman, T.; Turner, W.; Marple, V. A. (1988) An intercomparison of the indoor air sampling
42 impactor and the dichotomous sampler for a 10-^m cut size. JAPCA 38: 668-670.
43
44 Lippmann, M. (n.d.) Environmental toxicology and exposure limits for ambient air. Appl. Occup. Environ. Hyg.
45 8: 847-858.
46
47 Lippmann, M.; Chan, T. L. (1979) Cyclone sampler performance. Staub Reinhalt. Luft 39: 7-11.
48
49 Lippmann, M.; Chan, T. L. (1980) Experimental measurements and empirical modelling of the regional
50 deposition of inhaled particles in humans. Am. Ind. Hyg. Assoc. J. 41: 399-409.
51
52 Liu, B. Y. H., ed. (1976) Fine particles aerosol generation, measurement, sampling and analysis. New York,
53 NY: Academic Press.
54
April 1995 4-112 DRAFT-DO NOT QUOTE OR CITE
-------
1 Liu, B. Y. H.; Pui, D. Y. H. (1981) Aerosol sampling inlets and inhalable particles. Atmos. Environ.
2 15: 589-600.
3
4 Loo, B. W.; Cork, C. P. (1988) Development of high efficiency virtual impactors. Aerosol Sci. Technol.
5 9: 167-176.
5
7 Lundgren, D.; Burton, R. M. (1995) The effect of particle size distribution on the cut point between fine and
3 coarse ambient mass fractions. Inhalation Toxicol.: accepted.
9
10 Lundgren, D. A.; Lippmann, M.; Harris, F. S., Jr.; Clark, W. E.; Marlow, W. H.; Durham, M. D., eds.
11 (1979) Aerosol measurement: [papers from a workshop]; March 1976; Gainseville, FL. Gainesville, FL:
12 University Presses of Florida.
13
14 Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in
15 particles concentration and optical extinction in the United States. J. Geophys. Res. 99: 1347-1370.
16
17 Mamane, Y. (1988) Estimate of municipal refuse incinerator contribution to Philadelphia aerosol. Atmos.
18 Environ. 22: 2411-2418.
19
20 Mark, D.; Lyons, C. P.; Upton, S. L.; Hall, D. J. (1992) A review of the rationale of current methods for
21 determining the performance of aerosol samplers. J. Aerosol Sci. 23(suppl. 1): S611-S614.
22
23 Markowski, G. R. (1987) On identifying and correcting for reentrainment in cascade impactor measurements.
24 Aerosol Sci. Technol. 7: 143-159.
25
26 Marple, V. A.; Rubow, K. L. (1976) J. Aerosol Sci. 6: 425-433.
27
28 Marple, V. A.; Liu, B. Y. H.; Olson, B. A. (1989) Evaluation of a cleanroom concentrating aerosol sampler. In:
29 Proceedings of the 35th annual meeting of the Institute of Environmental Sciences. Anaheim, CA:
30 Institute of Environmental Sciences; pp. 360-363.
31
32 Marple, V. A.; Liu, B. Y. H.; Burton, R. M. (1990) A high volume dichotomous sampler: design and
33 evaluation. Presented at: 83rd annual meeting of the Air and Waste Management Association; June;
34 Pittsburgh, PA. Pittsburgh, PA: Air and Waste Management Association; paper no. 90-84.2.
35
36 Marple, V. A.; Rubow, K. L.; Olson, B. A. (1993) Inertial gravitational, centrifugal, and thermal collection
37 techniques. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement principles, techniques and
38 applications. New York, NY: Van Nostrand Reinhold; p. 209.
39
40 Marshall, T. A.; Mitchell, J. P.; Griffiths, W. D. (1991) The behaviour of regular-shaped, non-spherical
41 particles in a TSI aerodynamic particle sizer. J. Aerosol Sci. 22: 73-89.
42
43 Masia, P.; Di Palo, V.; Possanzini, M. (1994) Uptake of ammonia by nylon filters in filter pack systems.
44 Atmos. Environ. 28: 365-366.
45
46 Massart, D. L.; Kaufman, L. (1983) The interpretation of analytical chemical data by the use of cluster analysis.
47 New York, NY: John Wiley & Sons.
48
49 Maynard, A. D. (1993) Respirable dust sampler characterisation: efficiency curve reproducibility. J. Aerosol Sci.
50 24(suppl. 1): S457-S458.
51
52 Mazurek, M. A.; Simoneit, B. R. T.; Cass, G. R.; Gray, H. A. (1987) Quantitative high-resolution gas
53 chromatography and high-resolution gas chromatography/mass spectrometry analyses of carbonaceous fine
54 aerosol particles. Int. J. Environ. Anal. Chem. 29: 119-139.
April 1995 4-113 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1989) Interpretation of high-resolution gas chromatography /
2 mass spectrometry data aquired from atmospheric organic aerosol samples. Aerosol Sci. Technol.
3 10: 408-420.
4
5 Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Biological input to visibility-reducing aerosol particles
6 in the remote arid southwestern United States. Environ. Sci. Technol. 25: 684-694.
7
8 McDow, S. R.; Huntzicker, J. J. (1990) Vapor absorption artifact in the sampling of organic aerosol: face
9 velocity effects. Atmos. Environ. Part A 24: 2563-2571.
10
11 McDow, S. R.; Huntzicker, J. J. (1993) Vapor adsorption artifact in the sampling of organic aerosol. In:
12 Winegar, E. D.; Keith, L. H., eds. Sampling and analysis of airborne pollutants. Boca Raton, FL: Lewis
13 Publishers, Inc.; pp. 191-208.
14
15 McFarland, A. R.; Ortiz, C. A. (1979) Aerosol characterization of ambient paniculate samplers used in
16 environmental monitoring studies. College Station, TX: Texas A&M Research Foundation; contract no.
17 68-02-2720.
18
19 McFarland, A. R.; Ortiz, C. A. (1983) Characterization of Sierra Anderson PM-10 inlet model 246b. College
20 Station, TX: Texas A&M University, Department of Civil Engineering; Air Quality Laboratory report
21 4716/02/02/84/ARM.
22
23 McFarland, A. R.; Wedding, J. B.; Cermak, J. E. (1977) Wind tunnel evaluation of a modified Andersen
24 impactor and an all weather sampler inlet. Atmos. Environ. 11: 535-539.
25
26 McFarland, A. R.; Ortiz, C. A.; Rodes, C. E. (1982) Wind tunnel evaluation of the British smoke shade
27 sampler. Atmos. Environ. 16: 325-328.
28
29 McFarlaftd, A. R.; Ortiz, C. A.; Bertch, R. W. (1984) A 10 /xm cutpoint size selective inlet for hi-vol samplers.
30 J. Air Pollut. Control Assoc. 34: 544-547.
31
32 Mclntyre, B. L.; Johnson, D. L. (1982) A particle class balance receptor model for aerosol apportionment in
33 Syracuse, NY. In: Receptor models applied to contemporary pollution problems. Pittsburgh, PA: Air
34 Pollution Control Association.
35
36 Merrifield, T. (1989) The measurement of PM10 by /3-absorption. Presented at: 82nd annual meeting and
37 exhibition of the Air & Waste Management Association; June; Anaheim, CA. Pittsburgh, PA:
38 Air & Waste Management Association; paper no. 89-66A.2.
39
40 Meyer, M. B.; Lijek, J.; Ono, D. (1992) Continuous PM10 measurements in a woodsmoke environment. In:
41 Chow, J. C.; Ono, D. M., eds. PM10 standards and non-traditional paniculate source controls:
42 proceedings of an AWMA/EPA international specialty conference. Pittsburgh, PA: Air & Waste
43 Management Association; pp. 24-38.
44
45 Meyer-Arendt, J. R. (1972) Introduction to classical and modern optics. Prentice Hall, Inc.; p. 133.
46
47 Miller, F. J.; Gardner, D. E.; Graham, J. A.; Lee, R. E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
48 considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc.
49 29: 610-615.
50
51 Mulholland, G. W.; Pui, D. Y. H.; Kapadia, A.; Liu, B. Y. H. (1980) Aerosol number and mass concentration
52 measurements: a comparison of the electrical aerosol analyzer with other measurement techniques.
53 J. Colloid Interface Sci. 77: 57-67.
54
April 1995 4_114 DRAFT-DO NOT QUOTE OR CITE
-------
1 Mylonas, D. T.; Allen, D. T.; Ehrman, S. H.; Pratsinis, S. E. (1991) The sources and size distributions of
2 organonitrates in Los Angeles aerosol. Atmos. Environ. Part A 25: 2855-2861.
3
4 Novick, V. J.; Alvarez, J. L. (1987) Design of a multistage virtual impactor. Aerosol Sci. Technol. 6: 63-70.
5
6 Ogden, T. L. (1992) Inhalable and respirable dust: moving from aerosol science to legislation. J. Aerosol Sci.
7 23(suppl. 1): S445-S448.
8
9 Ogden, T. L.; Birkett, J. L. (1977) The human head as a dust sampler. In: Walton, W. H., eds. Inhaled particles
10 IV. Oxford, United Kingdom: Pergamon Press; pp. 93-105.
11
12 Okazaki, K.; Willeke, K. (1987) Transmission and deposition behavior of aerosols in sampling inlets. Aerosol
13 Sci. Technol. 7: 275-283.
14
15 Olmez, I. (1989) Instrumental neutron activation analysis of atmospheric paniculate matter. In: Lodge, J. P., Jr.,
16 ed. Methods of air sampling and analysis. 3rd ed. Chelsea, MI: Lewis Publishers; pp. 143-150.
17
18 Olmez, I. (1989) Trace element signatures in groundwater pollution. In: Watson, J. G., ed. Receptor models in
19 air resources management: transactions. Pittsburgh, PA: Air & Waste Management Association;
20 pp. 3-11.
21
22 Palen, E. J.; Allen, D. T. (1992) Fourier transform infrared analysis of aerosol formed in the photo-oxidation of
23 1-octene. Atmos. Environ. Part A 27: 1471-1477.
24
25 Palen, E. J.; Allen, D. T. (1992) Fourier transform infrared analysis of aerosol formed in the photo-oxidation of
26 isoprene and b-pinene. Atmos. Environ. Part A 26: 1239-1251.
27
28 Pankow, J. F. (1987) Review and comparative analysis of the theories on partitioning between the gas and
29 aerosol particulate phases in the atmosphere. Atmos. Environ. 21: 2275-2283.
30
31 Pankow, J. F. (1992) Application of common Y-intercept regression parameters for log Kp vs 1/T for predicting
32 gas-particle partitioning in the urban environment. Atmos. Environ. Part A 26: 2489-2497.
33
34 Parkes, J.; Rabbitt, L. J.; Hamshire, M. J. (1979) Live peak-stripping during X-ray energy-dispersive analysis.
35 Anal. Chem. 46: 1830-1831.
36
37 Patashnick, H.; Rupprecht, E. G. (1991) Continuous PM10 measurements using the tapered element oscillating
38 microbalance. J. Air Waste Manage. Assoc. 41: 1079-1083.
39
40 Pellizzari, E.; Lioy, P.; Quackenboss, J.; Whitmore, R.; Thomas, K.; Clayton, A.; Freeman, N.; Zelon, H.;
41 Waldman, J.; Rodes, C. (1994) NHEXAS: a pilot study in EPA region V. Presented at: International
42 Society for Epidemiology/International Society for Exposure Analysis joint conference; September;
43 Research Triangle Park, NC. paper 415.
44
45 Pickle, T.; Allen, D. T.; Pratsinis, S. E. (1990) The sources and size distributions of aliphatic and carbonyl
46 carbon in Los Angeles aerosol. Atmos. Environ. Part A 24: 2221-2228
47
48 Pupp, C.; Lao, R. C.; Murray, J. J.; Pottie, R. F. (1974) Equilibrium vapour concentrations of some polycyclic
49 aromatic hydrocarbons, As4O6 and SeO2 and the collection efficiencies of these air pollutants. Atmos
50 Environ. 8: 915-925.
51
52 Purdue, L. J.; Rodes, C. E.; Rehme, K. A.; Holland, D. M.; Bond, A. E. (1986) Intercomparison of
53 high-volume PM10 samplers at a site with high particulate concentrations. J. Air Pollut. Control Assoc
54 36: 917-920.
April 1995 4_H5 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rader, D. J.; O'Hera, T. J. (1993) Optical direct-reading techniques: in situ sensing. In: Willeke, K.;
2 Baron, P. A., eds. Aerosol measurement principles, techniques and applications. New York, NY: Van
3 Nostrand Reinhold; pp. 345-380.
4
5 Ramdahl, T.; Zielinska, B.; Arey, J.; Atkinson, R.; Winer, A.; Pitts, J. N., Jr. (1986) Ubiquitous occurrence of
6 2-nitrofluoranthene and 2-nitropyrene in air. Nature (London) 231: 425-427.
7
8 Ranade, M. B. (Arun); Woods, M. C.; Chen, F.-L.; Purdue, L. J.; Rehme, K. A. (1990) Wind tunnel
9 evaluation of PM10 samplers. Aerosol Sci. Technol. 13: 54-71.
10
11 Reist, P. C. (1984) Introduction to aerosol science. New York, NY: Macmillan Publishing.
12
13 Rodes, C.; Holland, D.; Purdue, L.; Rehme, K. (1985) A field comparison of PM10 inlets at four locations.
14 J. Air Pollut. Control Assoc. 35: 345-354.
15
16 Rodes, C. E.; Kamens, R. M.; Wiener, R. W. (1991) The significance and characteristics of the personal
17 activity cloud on exposure assessment measurements for indoor contaminants. Indoor Air 2: 123-145.
18
19 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine
20 organic aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ.
21 Sci. Technol. 27: 636-651.
22
23 Rogge, W. F.; Mazurek, M. A.; Hildemann, L. M.; Cass, G. R.; Simoneit, B. R. T. (1993b) Quantification of
24 urban organic aerosols at a molecular level: identification, abundance and seasonal variation. Atmos.
25 Environ. Part A 27: 1309-1330.
26
27 Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine
28 organic aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and
29 sinks. Environ. Sci. Technol. 27: 1892-1904.
30
31 Rondia, D. (1965) Sur la volatilite des hydrocarbures polycycliques [Volatility of polycyclic hydrocarbons]. Int.
32 J. Air Water Pollut. 9: 113-121.
33
34 Rood, M. J.; Covert, D. S.; Larson, T. V. (1987) Hygroscopic properties of atmospheric aerosol in Riverside,
35 California. Tellus 39B: 383-397.
36
37 Rubow, K. L.; Marple, V. A.; Olin, J.; McCawley, M.A. (1987) A personal cascade impactor: design,
38 evaluation and calibration. Am. Ind. Hyg. Assoc. J. 48: 532-538.
39
40 Rupprecht, E.; Meyers, M.; Patashnick, H. (1992) The tapered element oscillating microbalance as a tool for
41 measuring ambient paniculate concentrations in real time. J. Aerosol Sci. 23(suppl. 1): S635-S638
42
43 Russ, J. C. (1977) Processing of energy-dispersive X-ray spectra. X-ray Spectrom. 6: 37-55.
44
45 Saltzman, B. E. (1984) Generalized performance characteristics of miniature cyclones for atmospheric paniculate
46 sampling. Am. Ind. Hyg. Assoc. J. 45: 671-680.
47
48 Sass-Kortsak, A. M.; O'Brien, C. R.; Bozek, P. R.; Purdham, J. T. (1993) Comparison of the 10 mm nylon
49 cyclone, horizontal elutriator, and aluminum cyclone for silica and wood dust measurements. Appl.
50 Occup. Environ. Hyg. 8: 31-37.
51
52 Saucy, D. A.; Anderson, J. R.; Buseck, P. R. (1987) Cluster analysis applied to atmospheric aerosol samples
53 from the Norwegian Arctic. Atmos. Environ. 21: 1649-1657.
54
April 1995 4_116 DRAFT-DO NOT QUOTE OR CITE
-------
1 Schamber, F. H. (1993) SEM for nontraditional users. Am. Lab. (June): 34-38.
2
3 Schipper, L. B., Ill; Chow, J. C.; Frazier, C. A. (1993) Particulate air toxic emission estimation of the PM10
4 fraction in natural aggregate processing facilities. Presented at: 86th annual meeting and exhibition of the
5 Air & Waste Management Association; June; Denver, CO. Pittsburgh, PA: Air & Waste Management
6 Association; paper no. 93-MP-6.03.
7
8 Schuetzle, D.; Lewtas, J. (1986) Bioassay-directed chemical analysis in environmental research. Anal. Chem.
9 58: 1060A-1076A.
10
11 Schwartz, G. P.; Daisey, J. M.; Lioy, P. J. (1981) Effect of sampling duration on the concentration of
12 paniculate organics collected on glass fiber filters. Am. Ind. Hyg. Assoc. J. 42: 258-263.
13
14 Scott, W. R.; Chatfield, E. J. (1979) A precision SEM image analysis system with full feature EDXA
15 characterization. Scanning Electron Microsc. 2: 53.
16
17 Shimp, D. R. (1988) Field comparison of beta attenuation PM10 sampler and high-volume PM10 sampler. In:
18 Mathai, C. V.; Stonefield, D. H., eds. PM10: implementation of standards, transactions. Pittsburgh, PA:
19 Air & Waste Management Association; pp. 171-178.
20
21 Sickles, J. E., II; Hodson, L. L. (1989) Fate of nitrous acid on selected collection surfaces. Atmos. Environ.
22 23: 2321-2324.
23
24 Sickles, J. E., II; Hodson, L. L.; McClenny, W. A.; Paur, R. J.; Ellestad, T. G.; Mulik, J. D.; Anlauf, K. G.;
25 Wiebe, H. A.; Mackay, G. I.; Schiff, H. I.; Bubacz, D. K. (1988) Field comparison of methods for the
26 measurement of gaseous and particulate contributors to acidic dry deposition. Atmos. Environ. Part A
27 24: 155-165.
28
29 Sides, J. O.; Saiger, H. F. (1976) Effects of prolonged static exposures of filters to ambient air on high volume
30 sampling results. Topeka, KS: Department of Health and Environment.
31
32 Simoneit, B. R. T. (1984) Organic matter of the troposphere—III. characterization and sources of petroleum and
33 pyrogenic residues in aerosols over the western United States. Atmos. Environ. 18: 51-67.
34
35 Simoneit, B. R. T.; Sheng, G.; Chen, X.; Fu, J.; Zhang, J.; Xu, Y. (1991) Molecular marker study of
36 extractable organic matter in aerosols from urban areas of China. Atmos. Environ. Part A
37 25:2111-2129.
38
39 Sioutas, C.; Koutrakis, P.; Olson, B. A. (1994) Development and evaluation of a low cutpoint virtual impactor.
40 Aerosol Sci. Technol. 21: 223-235.
41
42 Sloane, C. S. (1986) Effect of composition on aerosol light scattering efficiencies. Atmos. Environ.
43 20: 1025-1037.
44
45 Spagnolo, G. S. (1989) Automatic instrument for aerosol sampler using the beta-particle attenuation. J. Aerosol
46 Sci. 20: 19-27.
47
48 Spagnolo, G. S.; Paoletti, D. (1994) Automatic system for three fractions sampling of aerosol particles in
49 outdoor environments. J. Air Pollut. Control Assoc. 44: 702-706
50
51 Spicer, C. W.; Schumacher, P. M. (1979) Particulate nitrate: laboratory and field studies of major sampling
52 interferences. Atmos. Environ. 13: 543-552.
53
April 1995 4_H7 DRAFT-DO NOT QUOTE OR CITE
-------
1 Stevens, R. K.; Pace, T. G. (1984) Overview of the mathematical and empirical receptor models workshop
2 (Quail Roost II). Atmos. Environ. 18: 1499-1506.
3
4 Stevens, R. K.; Dzubay, T. G.; Shaw, R. W., Jr.; McClenny, W. A.; Lewis, C. W.; Wilson, W. E. (1980)
5 Characterization of the aerosol in the Great Smoky Mountains. Environ. Sci. Technol. 14: 1491-1498.
6
7 Stevens, R. K.; Pinto, J.; Mamane, Y.; Ondov, J.; Abdulraheem, M.; Al-Majed, N.; Sadek, M.; Cofer, W.;
8 Ellenson, W.; Kellogg, R. (1993) Chemical and physical properties of emissions from Kuwaiti oil fires.
9 Water Sci. Technol. 27: 223-233.
10
11 Sweitzer, T. (1985) A field evaluation of two PM10 inlets in an industrialized area of Illinois. J. Air Pollut.
12 Control Assoc. 35: 744-746.
13
14 Tang, H.; Lewis, E. A.; Eatough, D. J.; Burton, R. M.; Farber, R. J. (1994) Determination of the particle size
15 distribution and chemical composition of semi-volatile organic compounds in atmospheric fine particles
16 with a diffusion denuder sampling system. Atmos. Environ. 28: 939-947.
17
18 Teague, S. V.; Raabe, O. G.; Elred, R. A. (1992) A bench top wind tunnel for testing PM10 sampling inlets. In:
19 Chow, J. C.; Ono, D. M., eds. PM10 standards and non-traditional particulate source controls:
20 proceedings of an AWMA/EPA international specialty conference. Pittsburgh, PA: Air & Waste
21 Management Association; pp. 2-12.
22
23 Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM10 samplers
24 and source apportionment of ambient PM10 concentrations at Rillito, Arizona. In: Chow, J. C.;
25 Ono, D. M., eds. PM10 standards and non-traditional particulate source controls: proceedings of an
26 AWMA/EPA international specialty conference. Pittsburgh, PA: Air & Waste Management Association;
27 pp. 244-261.
28
29 Torok, S..B.; Van Grieken, R. E. (1994) X-ray spectrometry. Anal. Chem. 66: 186R-206R.
30
31 Tsai, P. J.; Vincent, J. H. (1993) Impaction model for the aspiration efficiencies of aerosol samplers at large
32 angles with respect to the wind. J. Aerosol Sci. 24: 919-928.
33
34 Tufto, P. A.; Willeke, K. (1982) Dynamic evaluation of aerosol sampling inlets. Environ. Sci. Technol.
35 16: 607-609.
36
37 Turner, J. R.; Hering, S. V. (1987) Greased and oiled substrates as bounce-free impaction surfaces. J. Aerosol
38 Sci. 18: 215-224.
39
40 Turpin, B. J., et al. (1993) Elemental analysis of single atmospheric particles influencing visibility at the Grand
41 Canyon. In: Proceedings of the 51st annual meeting of the Microscopy Society of America. San
42 Francisco, CA: San Francisco Press.
43
44 Turpin, B. J.; Liu, S.-P.; Podolske, K. S.; Gomes, M. S. P.; Eisenreich, S. J.; McMurry, P. H. (1993) Design
45 and evaluation of a novel diffusion separator for measuring gas/particle distributions of semivolatile
46 organic compounds. Environ. Sci. Technol. 27: 2441-2449.
47
48 Turpin, B. J., et al. (1994) Single particle analysis of atmospheric aerosols by transmission electron microscopy:
49 technique development and application to the Mohave visibility study. To be submitted.
50
51 Turpin, B. J., et al. (1994) Description of a cluster analysis technique to categorize single particle compositions
52 and application to Grand Canyon aerosols. Aerosol Sci. Technol.: submitted.
53
April 1995 4-118 DRAFT-DO NOT QUOTE OR CITE
-------
1 Turpin, B. J.; Huntzicker, J. J.; Hering, S. V. (1994) Investigation of organic aerosol sampling artifacts in the
2 Los Angeles basin. Atmos. Environ. 28: 3061-3071.
3
4 U.S. Environmental Protection Agency. (1976) Laboratory procedures for the analysis of ammonia in particulates
5 collected by means of HI-VOL samplers-Technicon autoanalyzer II procedures. Research Triangle Park,
6 NC: Analytical Chemistry Branch.
7
8 U.S. Environmental Protection Agency. (1982a) Air quality criteria for particulate matter and sulfur oxides: v. II.
9 Research Triangle Park, NC: Environmental Criteria and Assessment Office; EPA report no.
10 EPA-600/8-82-029b. Available from: NTIS, Springfield, VA; PB84-120419.
11
12 U.S. Environmental Protection Agency. (1982b) Air quality criteria for particulate matter and sulfur oxides:
13 v. II. Research Triangle Park, NC: Environmental Criteria and Assessment Office; pp. 3-40; EPA report
14 no. EPA-600/8-82-029b. Available from: NTIS, Springfield, VA; PB84-120419.
15
16 U.S. Environmental Protection Agency. (1982c) Air quality criteria for particulate matter and sulfur oxides: v. II.
17 Research Triangle Park, NC: Environmental Criteria and Assessment Office; pp. 14-112; EPA report no.
18 EPA-600/8-82-029b. Available from: NTIS, Springfield, VA; PB84-120419.
19
20 U.S. Environmental Protection Agency. (1992) Determination of the strong acidity of atmospheric fine-particles
21 (<2.5 urn) using annular denuder technology. Washington, DC: Atmospheric Research and Exposure
22 Assessment Laboratory; EPA report no. EPA/600/R-93/037.
23
24 Upton, S. L.; Mark, D.; Hall, D. J. (1992) Improvements to the design and performance of an ambient inhalable
25 aerosol sampler. J. Aerosol Sci. 23(suppl. 1): S595-S598.
26
27 Van Borm, W. (1989) Source apportionment of atmospheric particles by electron probe X-ray microanalysis and
' 28 receptor models. Belgium: University of Antwerp.
29
30 Van Borm, W. A.; Adams, F. C. (1988) Cluster analysis of electron microscopy analysis data of individual
31 particles for the source apportionment of air paniculate matter. Atmos. Environ. 22: 2297-2307.
32
. 33 Van Vaeck, L.; Van Cauwenberghe, K.; Janssens, J. (1984) The gas-particle distribution of organic aerosol
34 constituents: measurements of the volatilisation artefact in hi-vol cascade impactor sampling. Atmos.
35 Environ. 18: 417-430.
36
37 Vanderpool, R. W.; Lundgren, D. A.; Marple, V. A.; Rubow, K. L. (1987) Cocalibration of four large-panicle
38 impactors. Aerosol Sci. Technol. 7: 177-185.
39
40 Vaughan, N. P. (1989) The Andersen impactor: calibration, wall losses and numerical simulation. J. Aerosol Sci.
41 20: 67-90.
42
43 Vossler, T. L.; Stevens, R. K.; Paur, R. J.; Baumgardner, R. E.; Bell, J. P. (1988) Evaluation of improved
44 inlets and annular denuder systems to measure inorganic air pollutants. Atmos. Environ. 22: 1729-1736.
45
46 Waggoner, A. P.; Weiss, R. E. (1980) Comparison of fine particle mass concentration and light scattering in
47 ambient aerosol. Atmos. Environ. 14: 623-626.
48
49 Wake, D. (1989) Anomalous effects in filter penetration measurements using the aerodynamic particle sizer (APS
50 3300). J. Aerosol Sci. 20: 13-17.
51
52 Wang, H.-C.; John, W. (1987) Comparative bounce properties of particle materials. Aerosol Sci. Technol.
53 7: 285-299.
54
April 1995 4_119 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
2 Technol. 8: 157-172.
3
4 Wang, H.; John, W. (1989) A simple iteration procedure to correct for the density effect in the aerodynamic
5 particle size. Aerosol Sci. Technol. 10: 501-505.
6
7 Watson, J. G.; Chow, J. C. (1993) Ambient air sampling. In: Willeke, K.; Baron, P. A., eds. Aerosol
8 measurement principles, techniques and applications. New York, NY: Van Nostrand Reinhold; p. 627.
9
10 Watson, J. G.; Chow, J. C.; Shah, J. J.; Pace, T. G. (1983) The effect of sampling inlets on the PM-10 and
11 PM-15 to TSP concentration ratios. J. Air Pollut. Control Assoc. 33: 114-119.
12
13 Watson, J. G.; Bowen, J. L.; Chow, J. C.; Rogers, C. F.; Ruby, M. G.; Rood, M. J.; Egami, R. T. (1989a)
14 Method 501: high volume measurement of size classified suspended paniculate matter. In: Lodge, J. P.,
15 ed. Methods of air sampling and analysis. 3rd ed. Chelsea, MI: Lewis Publishers, Inc.; pp. 427-439.
16
17 Watson, J. G.; Lioy, P. J.; Mueller, P. K. (1989b) The measurement process: precision, accuracy, and validity.
18 In: Hering, S. V., ed. Air sampling instruments for evaluation of atmospheric contaminants. 7th ed.
19 Cincinnati, OH: American Conference of Governmental Industrial Hygienists; pp. 51-57.
20
21 Wedding, J. B. (1980) Ambient aerosol sampling: history, present thinking and a proposed inlet for inhalable
22 paniculate matter. Presented at: 73rd annual meeting of the Air Pollution Control Association; June;
23 Montreal, Canada. Pittsburgh, PA: Air Pollution Control Association; paper no. 80-38.1.
24
25 Wedding, J. B.; Carney, T. C. (1983) A quantitative technique for determining the impact of non-ideal ambient
26 sampler inlets on the collected mass. Atmos. Environ. 17: 873-882.
27
28 Wedding, J. B.; Weigand, M. A. (1983) A thoracic panicle inlet (D50 = 6 ^m, D0 = 10 /*m) for high volume
29 sampler. Atmos. Environ. 17: 1203-1204.
30
31 Wedding, J. B.; Weigand, M. A. (1993) An automatic particle sampler with beta gauging. J. Air Waste Manage.
32 Assoc. 43: 475-479.
33
34 Wedding, J. B.; Weigand, M. A.; Carney, T. C. (1982) 10 m Cutpoint inlet for the dichotomous sampler.
35 Environ. Sci. Technol. 16: 602-606.
36
37 Wernisch, J. (1985) Quantitative electron microprobe analysis without standard. X-ray Spectrom. 14: 109-119.
38
39 Wernisch, J. (1986) Application of different (p-z) distributions in quantitative electron probe microanalysis with
40 standard samples. Radex-Rundschau 2/3: 110-113.
41
42 Wiener, R. W.; Rodes, C. E. (1993) Indoor aerosols and aerosol exposure. In: Willeke, K.; Baron, P. A., eds.
43 Aerosol measurement principles, techniques and applications. New York, NY: Van Nostrand Reinhold;
44 pp. 659-689.
45
46 Willeke, K.; Baron, P. A., eds. (1993) Aerosol measurement: principles, techniques and applications. New
47 York, NY: Van Nostrand Reinhold Publishers.
48
49 Willeke, K.; Degarmo, S. J. (1988) Passive versus active aerosol monitoring. Appl. Ind. Hyg. 3: 110-114.
50
51 Willeke, K.; Grinshpun, S. A.; Chang, C.; Juozaitis, A.; Liebhaber, F.; Nevalainen, A.; Thompson, M. (1992)
52 Inlet sampling efficiency of bioaerosol samplers. J. Aerosol Sci. 23(suppl. 1): S651-S654.
53
April 1995 4-120 DRAFT-DO NOT QUOTE OR CITE
-------
1 Williams, E. L., II; Grosjean, D. (1990) Removal of atmospheric oxidants with annular denuders. Environ. Sci.
2 Technol. 24: 811-814.
3
4 Wilson, J. C.; Gupta, A.; Whitby, K. T.; Wilson, W. E. (1988) Measured aerosol light scattering coefficients
5 compared with values calculated from EAA and optical particle counter measurements: improving the
6 utility of the comparison. Atmos. Environ. 22: 789-793.
7
8 Winberry, W. T.; Forehand, L.; Murphy, N. T.; Ceroli, A.; Phinney, B.; Evans, A. (1993) Determination of
9 reactive acidic and basic gases and paniculate matter in indoor air (annular denuder technique, method
10 IP-9). In: Methods for determination of indoor air pollutants: EPA methods. Park Ridge, NJ: Noyes Data
11 Corporation.
12
13 Woskie, S. R.; Shen, P.; Finkel, M.; Eisen, E. A.; Smith, T. J.; Smith, R.; Wagman, D. H. (1993) Calibration
14 of a continuous-reading aerosol monitor (MINIRAM) to measure borate dust exposures. Appl. Occup.
15 Environ. Hyg. 8: 38-45.
16
17 Yamada, Y. (1983) A new method for the determination of collection efficiency of an aerosol sampler by
18 electron microscopy. Atmos. Environ. 17: 369-372.
19
20 Yamasaki, H.; Kuwata, K.; Miyamoto, H. (1982) Effects of ambient temperature on aspects of airborne "
21 polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 16: 189-194.
22
23 Ye, H. (1993) Electrical technique. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement principles,
24 techniques and applications. New York, NY: Van Nostrand Reinhold; pp. 410-426.
25
26 Ye, Y.; Tsai, C.; Pui, D. Y. H.; Lewis, C. W. (1991) Particle transmission characteristics of an annular
27 denuder ambient sampling system. Aerosol Sci. Technol. 14: 102-111.
28
29 Zhang, X!; McMurry, P. H. (1992) Evaporative losses of fine paniculate nitrates during sampling. Atmos
30 Environ. Part A 26: 3305-3312.
31
32 Zielinska, B.; Arey, J.; Atkinson, R.; Winer, A. M. (1989) The nitroarenes of molecular weight 247 in ambient
33 paniculate samples collected in southern California. Atmos. Environ. 23: 223-229.
34
35 Zoller, W. H.; Gordon, G. E. (1970) Instrumental neutron activation analysis of atmospheric pollutants utilizing
36 Ge(Li) 7-ray detectors. Anal. Chem. 42: 257-265.
April 1995 4_12i DRAFT-DO NOT QUOTE OR CITE
-------
i 5. SOURCES AND EMISSIONS OF
2 SUSPENDED PARTICLES
3
4
5 5.1 INTRODUCTION
6 Excessive particulate concentrations in the atmosphere result from pollutant emissions,
7 meteorological transport of those emissions between source and receptor, and chemical and
8 physical changes during the transport period. Importantly, both primary and secondary
9 particles contribute to ambient PM mass concentrations.
10 Primary particles are those which are directly emitted by sources, and these particles
11 often undergo few changes between source and receptor. Atmospheric concentrations of
12 primary particles are, on average, proportional to the quantities that are emitted. Primary
13 particles are emitted in several size ranges, the most common being less than 1 ^m in
14 aerodynamic diameter from combustion sources and larger than 1 /zm in aerodynamic
15 diameter from dust sources. Particles larger than 10 ^im in aerodynamic diameter usually
16 deposit to the surface within a few hours after being emitted and do not have a large effect
»
17 on urban or regional scales. These larger particles may make significant contributions at
18 receptors located within a few kilometers of the emissions source, however. Coarse particles
19 with aerodynamic diameters between 2 and 10 /xm may make significant contributions at
20 distances exceeding 10 km from their emissions sources, while particles with aerodynamic
21 diameters < 2 /xm may affect receptors that are more than 100 km distant from emissions
22 sources.
23 Secondary particles are those that form in the atmosphere from gases that are directly
24 emitted by sources. Sulfur dioxide, ammonia, and oxides of nitrogen are the precursors for
25 sulfuric acid, ammonium bisulfate, ammonium sulfate, and ammonium nitrate particles
26 (Seinfeld, 1986). Several volatile organic compounds (VOC) may also change into particles;
27 the majority of these transformations result from intense photochemical reactions that also
28 create high ozone levels (Grosjean and Seinfeld, 1989). Secondary particles usually form
29 over several hours or days, attain aerodynamic diameters between 0.1 and 1 /xm, owing to
30 the complex chemical processes that form them, secondary particle concentrations are not
31 necessarily proportional to the quantities that are emitted, and may affect receptors more than
April 1995 5-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 1,000 km distant from the sources of precursor gas emissions. Several of these particles,
2 notably those containing ammonium nitrate, are volatile and transfer mass between the gas
3 and particle phase to maintain a chemical equilibrium.
4 Ambient concentrations of secondary aerosol are not necessarily proportional to primary
5 emissions quantities since the rate at which they form may be limited by factors other than
6 concentrations of the precursor gases. For example, secondary ammonium nitrate is not a
7 stable compound; its equilibrium with gaseous ammonia and nitric acid is influenced strongly
8 by temperature and relative humidity (Watson et al., 1994b). Fugitive dust is predominantly
9 a primary pollutant, but it does play a role in secondary particle formation. Some
10 components of dust, such as ammonium nitrate fertilizer, may volatilize into ammonia and
11 nitric acid gases, thereby contributing to secondary aerosol. Alkaline particles, such as
12 calcium carbonate, may react with nitric and hydrochloric acid gases while on the ground, in
13 the atmosphere, or on filter samples to form coarse-mode particle nitrates and chlorides.
14 Ammonium sulfate fertilizer components and minerals such as gypsum (calcium sulfate) may
15 be mistaken for secondary sulfates when PM10 filters are chemically analyzed.
16 Due to the complex and varying composition of suspended particles, it is necessary to
17 consider sulfur dioxide, oxides of nitrogen, ammonia, VOCs, and primary particle emissions
18 as potential sources of suspended particles in ambient air. Major sources of these emissions
19 are classified into categories of: (1) major point sources; (2) mobile sources; and (3) area
20 sources (U.S. Environmental Protection Agency, 1993).
21 Major point sources are ducted emissions that are subject to permit. Emissions of a
22 single pollutant usually must exceed 25 tons/year to qualify for this category, though some
23 jurisdictions track individual thresholds at lower levels. This category includes stack
24 emissions from most industrial facilities in the United States such as steel mills, smelters,
25 cement plants, electric utilities, refineries, and incinerators. Non-ducted emissions from
26 these industries, such as VOCs from leaking valves and primary particles from materials
27 handling are termed "fugitive emissions".
28 Mobile sources include on-road and off-road motor vehicles, trains, aircraft, and ships,
29 with the majority of emissions resulting from on-road vehicles in most areas. Area sources
30 include many small stationary emitters that, in their aggregate, can be significant contributors
31 to suspended particles. These include residential wood and coal combustion, prescribed
April 1995 5-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 burning, space heating, cooking, paved and unpaved roads, construction and demolition,
2 agricultural operations, and wind erosion.
3 Point, mobile, and area source emissions are anthropogenic. Natural sources of
4 suspended particles such as sea salt, volcanic emissions, wild fires, and aeolian dust from
5 undisturbed surfaces can also contribute to ambient PM concentrations in certain situations.
6 One of the major challenges to air quality sciences is to distinguish among contributions from
7 these different sources when suspended particle concentrations are high.
8 This chapter is organized to present first a concise summary of key information on PM
9 emissions derived from the previous criteria review in the 1980's and then to provide a more
10 extensive discussion of newer information appearing in recent years.
11 The main objectives of the ensuing discussion this section are:
12 • To identify the sources that are major contributors to suspended particle
13 concentrations in the United States.
14
15 • To describe the particle sizes and chemical properties of source emissions.
16
17 • To evaluate the limitations and uncertainties of emissions rate estimates and
18 source contributions for suspended particles and their gaseous precursors.
19
20
21 5.2 SUMMARY OF 1982 CRITERIA DOCUMENT EMISSIONS
22 REVIEW
23 The U.S. Environmental Protection Agency (EPA) (1982) examined 109 open-literature
24 references related to particulate emissions published between 1951 and 1981. This
25 comprehensive review discussed the sources of emissions data and their accuracy, global
26 particulate emissions from natural sources, U.S. particulate and sulfur dioxide emissions
27 from manmade sources, and the size and chemical composition of emitted particles. The
28 major findings of this review were the following:
29
30 • Particulate emissions rates are very uncertain. Two independent EPA estimates
31 yielded 6.4 (U.S. Environmental Protection Agency, 1980a) and 3.9 (U.S.
32 Environmental Protection Agency, 1980b) million metric tons per year for the same
33 industrial process emissions of primary particles in 1977. Particle emissions
34 estimates from mobile sources were believed to have even greater uncertainty.
35
April 1995 5.3 DRAFT-DO NOT QUOTE OR CITE
-------
1 • Global participate emissions include 100 to 500 million metric tons/year (Robinson
2 and Robbins, 1971; Vandegrift et al., 1971) from aeolian dust, 900 million metric
3 tons/year from sea spray (Robinson and Robbins, 1971), and 0.75 to 3.75 million
4 tons/year from volcanoes (Robinson and Robbins, 1971; Granat et al., 1976). Much
5 of these emissions are small particles that do not attain great elevation above ground
6 level and deposit close to their emissions points. Significant fractions of volcanic
7 emissions, however, are injected high into the atmosphere and can have long
8 atmospheric residence times. Primary particle emissions from U.S. wildfires were
9 estimated to be 0.5 to 4.5 million metric tons/year, with most of these emissions
10 being small particles that do not deposit close to the emissions point (Robinson and
11 Robbins, 1971; Yamate, 1973).
12
13 • Naturally-emitted volatile organic compounds that might be secondary paniculate
14 precursors were estimated at 200 million metric tons/year for the world and
15 20 million metric tons/year for the United States (Went, 1960). Global emissions of
16 reduced sulfur compounds were estimated at 37 to 91 million metric tons per year
17 (Robinson and Robbins, 1971; Granat et al., 1976).
18
19 • Manmade particle emissions in the United States during 1978 included 10.5 million
20 metric tons/year from stationary point sources, 3.3 million metric tons/year from
21 non-ducted industrial processes, 110 to 370 tons/year from non-industrial fugitive
22 dust, and 1.3 million metric tons/year from mobile sources. Stationary point sources,
23 primarily electric utilities, accounted for 26.2 million tons/ year of sulfur dioxide
24 emissions, with the remaining 0.8 million tons/year emitted by mobile sources.
25
26 • Primary particle emissions from stationary fuel combustion, industrial processes,
27 solid waste disposal, mobile sources, and burning in the United States decreased from
28 24.8 million metric tons per year in 1940 to 12.5 million metric tons/year in 1978.
29 All categories decreased expect emissions from transportation, which nearly tripled
30 from 0.5 to 1.3 million metric tons/year over this period. Sulfur dioxide emissions
31 increased from 19.5 to 27.0 million metric tons/year between 1940 and 1978, with
32 the major increase in stationary fuel combustion.
33
34 • Coal combustion was the major U.S. industrial emitter, with 3,090 out of 10,460
35 metric tons/year total for primary particles and 17,890 out of 26,180 metric tons per
36 year total for sulfur dioxide during 1978. More than 80% of coal combustion
37 emissions derived from electricity generation.
38
39 • Emissions estimates differed substantially among different parts of the United States,
40 especially between the eastern and western regions. The midwest contained more
41 than 70% of particle and sulfur dioxide emissions from industrial sources.
42
43 • Fugitive dust source emissions were found primarily in particles sizes larger than
44 2.5 /mi. Seventy to 100% of the primary particle emissions from coal and oil
45 combustion and other ducted industrial emissions were smaller than 2.5 /j.m in
46 aerodynamic diameter.
47
April 1995 5.4 DRAFT-DO NOT QUOTE OR CITE
-------
1 • Aluminum, silicon, calcium, potassium, and titanium were found to be abundant in
2 many geologically-related emissions. Several sources of primary particles showed
3 enrichments with respect to crustal abundances for certain elements: (1) copper,
4 lead, and arsenic in smelter emissions; (2) selenium and arsenic in coal combustion
5 emissions; (3) iron and manganese in steel mill emissions; (4) calcium and sulfate in
6 cement emissions; and (5) vanadium and nickel in residual oil combustion emissions.
7
8 The U.S. EPA (1982) emphasized emissions from industrial sources, and especially
9 primary particles emitted by these sources. Sulfur dioxide was the only precursor of
10 secondary aerosol that was considered. Emission rates were quantified in terms of Total
11 Suspended Paniculate (TSP), i.e., particles with aerodynamic diameters < -50 /im, since
12 PM10 had not yet been defined as the size fraction relevant to public health. Measured
13 chemical compositions included many elements and sulfate, but did not include nitrate,
14 ammonium, and carbon.
15 At the time, this emphasis was appropriate owing to the then available information and
16 to the lack of sufficient emissions controls on many of these sources. Since that time, many
17 ducted emissions have been controlled, yet air quality standards are still exceeded in many
18 areas. Non-sulfate secondary aerosol is a major component of suspended particles in many
19 areas. *The particle sizes relevant to health effects are now believed to be much smaller than
20 50 /-on, and possibly substantially smaller than the PM10 fraction which is currently subject to
21 NAAQS regulation. Much new information has been published since 1982 that enhances and
22 expands upon the conclusions about source contributions to suspended particles.
23
24
25 5.3 SOURCE CONTRIBUTIONS TO SUSPENDED PARTICLES
26 Much of the current knowledge related to source contributions derives from evaluations
27 used in development of control strategies for non-attainment areas where PM10 concentrations
28 exceed 50 jug/m3 for an annual arithmetic average and/or 150 /ng/m3 for a 24-h average.
29 Seventy-five areas have been designated in non-attainment. The 1990 Clean Air Act
30 Amendments and several years of ambient PM10 monitoring resulted in designation of
31 75 U.S. areas as "moderate" non-attainment areas for PM10 (Federal Register, 1991; 1994),
32 5 areas as "serious" non-attainment (Federal Register, 1993), and the remaining areas as
33 unclassifiable. Each state containing non-attainment areas must develop and submit State
April 1995 5-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 Implementation Plans (SIPs) that specify the means for reducing PM10 concentrations below
2 the standards. Extensive efforts were expended in many areas to apportion excessive PM10
3 concentrations to their sources.
4 The U.S. EPA (1987) recommends the application of both source and receptor models
5 to quantify the major contributors to excess PM10 concentrations, and linear rollback of
6 emissions to estimate the effects of alternative emission controls (Pace and Watson, 1987;
7 U.S. Environmental Protection Agency, 1987). Source models use emissions inventories and
8 meteorological measurements to predict the PM10 concentrations measured at receptors.
9 Receptor models use the chemical composition of source emissions and receptor
10 concentrations to infer the source contributions that constitute the measured PM10 mass.
11 Source models are most useful when sources have been identified and emissions rates are
12 known. In many non-attainment areas, however, the majority of emissions emanate from
13 mobile and area sources that have highly sporadic and often unknown emissions at different
14 locations and different times. Several non-attainment areas in the mountainous western
15 United States experience highly variable meteorology induced by complex terrain, and most
16 dispersion models are not applicable. In these situations, receptor-oriented source
17 apportionment models were found to the be the most appropriate methods to identify sources
18 and quantify their contributions to measured PM10.
19 Source contributions to average PM10 for several areas where the Chemical Mass
20 Balance (CMB) receptor model (Watson et al., 1990; 1991) was applied are shown in
21 Table 5-1. The values in Table 5-1 are not entirely comparable in an absolute sense, since
22 published source apportionment studies usually report averages for different sample selections
23 (often biased toward high PM10 levels) and different seasons. The sampling sites represent a
24 variety of different source characteristics within different regions of Arizona, California,
25 Colorado, Idaho, Illinois, Nevada and Ohio. Several of these are background sites,
26 specifically Estrella Park, Gunnery Range, Pinnacle Peak, and Corona de Tucson, AZ,
27 Anacapa Island, CA, San Nicolas Island, CA, Vandenberg Air Force Base, CA, and Verdi,
28 NV. Definitions of source categories also vary from study to study. In spite of these
29 differences, several features can be observed from the values in this table.
30 Fugitive dust (geological material) from roads, agriculture and erosion is a major
31 contributor to PM10 at nearly all sampling sites, often contributing up to, but not generally
April 1995 5-6 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-1. RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM10.
t— '•
I—1
VO
1
-J
O
!>
^
H
6
o
o
H
O
e;
O
H
tn
§
o
H
m
Sampling Site
Central Phoenix, AZ (Chow et al , 1991)
Corona de Tucson, AZ (Chow et al., 1992a)
Craycroft. AZ (Chow et al., 1992a)
Downtown Tucson, AZ (Chow et al., 1992a)
Hayden 1, AZ (Garfietd) (Ryan et al , 1988)
Hayden 2, AZ (Jail) (Ryan et al , 1988)
Orange Grove, AZ (Chow et al., 1992a)
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 (Thanukoset al., 1992)
Scottsdale, AZ (Chow et al.. 1991)
West Phoenix, AZ (Chow et al., 1991)
Anacapa Island, CA (Chow et al., 1994b)
Anaheim, CA (Gray et al., 1988)
Anaheim, CA (Summer) (Watson et al , 1994a)
Anaheim, CA (Fall) (Watson et al , 1994a)
Azusa, CA (Summer) (Watson et al , 1994a)
Bakersfield, CA (Magliano, 1988)
Bakerfield, CA (Chow el al , 1992b)
Burbank, CA (Gray et al , 1988)
Burbank, CA (Summer) (Watson et al , 1994a)
Burbank, CA (Fall) (Watson et al., 1994a)
Chula Vista 1, CA (Bayside) (Cooper et al., 1988)
Chula Vista 2, CA (De! Ray) (Cooper et a!., 1988)
Chula Vista 3, CA (Cooper et al., 1988)
Claremont, CA (Summer) (Watson et al , 1994a)
Crows Landing, CA (Chow et al , 1992h)
Primary
Geological
33.0
17.0
13.0
26 0
5.0
21 0
20.0
37.0
20 0
7 0
42.7
25.0
30.0
2.2
21.2
11 4
13 2
34 9
27.4
42 9
21.3
14.0
11.0
6.7
82
9.7
19 4
32 2
Primary
Construction
0.0
0.0
0.0
5 1
2 Ob
4 Ob
00
0.0
00
0 0
13 8b
00
00
0.0
0.0
00
0.0
0 0
3 0
1.6
00
0.0
0.0
0.0
0 3
03
0 0
!) 0
Primary*
Motor
Vehicle
Exhaust
25.0
1.6
8.3
14 0
00
0.0
15 0
10.0
55
2 9
1.2'
19.0
25.0
4 9
4.1'
8.5
37.2
15 9
5.5
7.7
6 I1
17.0
39.1
0.8
1.5
1 4
14 4
2 2
Primary
Vegetative
Burning
2.3
00
00
00
0.0
0.0
0 0
0.9
0.0
1 0
00
7.4
10.0
0.0
00
0.0
00
00
9.6'
65
0.0
0.0
00
0.0
0.0
0.0
0.0
3 4
Secondary
Ammonium
Sulfate
0.2
1 9
0.7
1 0
40
4 0
0 4
1.6
1 0
0.9
0.
0.6
0.4
3 4
7.0
9.0
3.7
11 4
5 6
5.5
7 2
12.4
3 1
7 5
8.9
8.2
9 5
9 y
/ig/m3
Secondary
Ammonium
Nitrate
2.8
0.0
0.6
0.2
0.0
0.0
0 4
0.0
00
0 0
0.0
3.6
3.1
1 0
9.8
2.9
38 5
6 1
0.0
12.7
10.2
6 5
25 1
00
00
0.0
6 3
6 5
Misc.
Source 1
0.0
00
1.2*
1 3a
74 Oc
28 (f
0.0
0.0
0.0
0 0
11.6s
0.0
00
9.6h
0 41
OO1
00'
0 O1
0.5J
1 Om
0.1J
0.0'
0.0-'
0.41
0.61
0.6>
0 0'
0 5m
Misc.
Source 2
0.0
00
00
0.0
5.0d
00
00
0.0
0.0
0.0
00
0.0
0.0
0.0
1.4h
6.5"
3 lh
5 7h
0.0
1.5"
0911
5.7h
1.9h
2.7h
1.8h
1.7h
47h
1 5n
Misc.
Source 3
0.0
00
0.0
0.0
1.0e
1 Oc
0.0
0.0
0.0
0.0
0.0
0.0
0.0
00
8.2k
0.0
0.0
0.0
0.0
0.6k
9.8k
0.0
0.0
2.0k
0.0
- o.o
00
1.2k
Misc.
Source 4
0.0
0.0
00
00
00
0.0
0 0
0.0
0 0
0.0
0.0
0.0
0.0
0.0
0.0
00
0.0
0.0
0.0
00
00
0.0
0.0
0.0
0.0
0.0
0.0
0 0
Measured
PM10
Concentration
64.0
19.1
23 4
48.0
105 0
59 0
34 2
55 0
27.0
12 0
79 5
55.0
690
26.0
52 1
51 3
104 0
92 1
67.6
79.6
56 6
72.3
94.8
28 8
31 1
29.6
70 0
52 5
-------
TABLE 5-1 (cont'd). RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM
10.
N±>'
>o
^o
Ul
i
00
o
>
H
6
o
z
o
H
O
o
H
tn
O
O
H
W
Sampling Site
Downtown Los Angeles, CA (Gray et al., 1988)
Downtown Los Angeles, CA (Summer) (Watson et al ,
1994a)
Downtown Los Angeles, CA (Fall) (Watson et al ,
1994a)
Fellows, CA (Chow et al., 1992b)
Fresno, CA (Magliano, 1988)
Fresno, CA (Chow et al., 1992b)
Hawthorne, CA (Summer) (Watson et al , 1994a)
Hawthorne, CA (Fall) (Watson et al., 1994a)
Indio, CA (Kim et al., 1992)
Kern Wildlife Refuge, CA (Chow et al., 1992b)
Lennox, CA (Gray et al., 1988)
Long Beach, CA (Gray et al., 1988)
Long Beach, CA (Summer) (Watson et al., 1994a)
Long Beach, CA (Fall) (Watson et al., 1994a)
Magnolia, CA (Chow et al., 1992c)
Palm Springs, CA (Kim et al., 1992)
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 Jose, CA (4th St.) (Chow et al., 1994a)
San Jose, CA (San Carlos St.) (Chow et al., 1994a)
San Nicolas Island, CA (Summer) (Watson et al , 1994a)
Santa Barbara, CA (Chow et al., 1994b)
Santa Barbara, CA (GTC) (Chow et al , 1994b)
Santa Maria, CA (Chow et al., 1994b)
Primary
Geological
23.8
12.7
9.4
29.0
17.1
31.8
7.5
8.9
33.0
15.1
16.0
20.7
11.1
11.3
31.7
16.4
32.6
43.1
34.9
19.2
48.0
13.1
11.8
1.6
9.5
3.2
7.4
Primary
Construction
0.0
0.0
0.0
1.4
07
0.0
0.0
0.0
3.0
2.0
0.1
0.0
0.0
0.0
0.0
1.4
0.0
4.01
4 5
16.1
00
0.0
00
0.0
0.0
0.0
0.0
Prirfiary
Motor
Vehicle
Exhaust
6.4'
16.2
41.1
2.1
4 0
6.8
5.6
35.1
4.4
2.2
4.6'
5.1'
6.3
42.8
11.2
2.3
7 0
5.6'
17.3
30.3
10.2
92
8.9
0.9
14 7
5.1
7.6
Primary
Vegetative
Burning
0.0
0.0
00
3.4
9.21
5.1
0.0
0 0
7 1
4.0
0.0
0.0
0.0
0.0
0.0
5.1
0.0
0.0
00
0.0
00
31.3
31.3
0.0
0.0
0.0
00
Secondary
Ammonium
Sulfate
7.6
13.0
3.9
5.1
1.8
3.6
15.0
5.1
3.6
3 3
7.6
8.0
10.9
3.8
4.9
3 7
4.8
6.4
9 5
2 1
5.3
2.3
2.1
3.7
3.2
2 8
3 1
//g/m3
Secondary
Ammonium
Nitrate
11.2
4.4
27.5
7 5
0.0
104
0.6
204
4.1
1.5
7.9
9.2
0.8
23.2
19.7
4.2
21.4
21.3
27.4
31.6
21 7
13 3
12.8
0 5
1 0
0 5
1 4
Misc
Source 1
0.0
00>
O.QJ
7.0m
0 1J
0.3m
O.O1
O.O1
0.2J
0.5m
02J
0.1J
0.1J
O.O1
0.3J
O.V
o.y
oy
O.QJ
O.QJ
0.41
0.9h
0.7h
00'
6.4h
63h
5.7h
Misc.
Source 2
1.3"
6.5h
1.8h
1.4"
0.0
1.0"
7.0h
3.7h
1.0h
1.5"
3.1h
2.0h
2.2h
2.7h
1.2"
0.5h
1.3h
1.0"
5.1"
1.1"
1.5h
00
0.0
4.3h
0.0 .
0.0
00
Misc.
Source 3
7.9k
0.0
0.0
1.4k
00
O.lk
0.0
0.0
00
0.7k
76k
6.4k
0.0
0.0
1.2°
0.0
1 1°
5.9k
0.0
0.0
5.7°
0.0
0.0
0.0
0.0
0.0
0.0
Misc.
Source 4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
00
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
00
0.0
0.0
0.0
00
00
0.0
0.0
0.0
0.0
Measured
PM10
Concentration
60.2
67 6
98.6
546
48.1
71.5
45.9
85.1
58 0
47.8
46.9
51 9
46.1
96 1
66.0
35.1
64.0
87.4
114.8
112.0
87.0
684
64.9
17 4
34.0
20 5
27 0
-------
TABLE 5-1 (cont'd). RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM10.
Pj
h— *
VO
Primary
^ Motor Primary Secondary
vb
O
3
H
t
o
o
2;
o
H
O
cl
O
m
H
w
Sampling Site
Santa Ynez, CA (Chow et al., 1994b)
Stockton, CA (Chow et al., 1992b)
Upland, CA (Gray et al., 1988)
Vandenberg AFB, CA (Watt Road) (Chow et al., 1994b)
Telluride 1, CO (Central) (Dresser and Baird, 1988)
Telluride 2, CO (Society Turn) (Dresser amd Baird, 1988)
Pocatello, ID (Houck et al , 1992)
S. Chicago, IL (Hopke et al , 1988)
S.E. Chicago, IL (Vermette et al., 1992)
Reno, NV (Non-sweeping) (Chow et al , 1990)
Reno, NV (Sweeping) (Chow et al., 1990)
Reno, NV(Chow et al., 1988)
Sparks, NV (Chow et al., 1988)
Verdi, NV (Chow et al., 1988)
Follansbee, OH (Skidmore et al., 1992)
Mingo, OH (Skidmore et al , 1992)
Sewage Plant, OH (Skidmore et al., 1992)
Steubenville, OH (Skidmore et al., 1992)
WTOV Tower, OH (Skidmore et al., 1992)
aSmelter background aerosol.
bCement plant sources, including kiln stacks, gypsum pile, and
cCopperore.
dCopper tailings.
'Copper smelter building
fHeavy-duty diesel exhaust emission.
Background aerosol.
hManne aerosol, road salt, and sea salt plus sodium nitrate
'Motor vehicle exhuast from diesel and leaded gasoline.
Primary
Geological
4.6
34 4
25 4
4.5
32.0
12 1
8.3
27.2
14 7V
9.7
11 8
14 9
15 1
7.8
10.0
120
22.0
8.3
7.4
kiln area.
Primary
Construction
00
0.5
0.*
0.0
00
00
7.5,
2 4
00
0.0
0.0
0.0
0.0
00
0.0
0.0
0.0
00
00
Vehicle Vegetative Ammonium
Exhaust Burning
6.8 0.0
5.2 48
41' 00
3.2 0.0
00 98.7
00 73
0 1 0.0
2.8 0.0
09f 00
8.7 0.1
11.0 1.2
10.0 1.9
11.6 134
4.0 1 1
35.0 0.0
14.0 4.1
12.0 0.0
14.0 0 8
160 0.2
JResidual oil combustion
Sulfate
2 2
3.1
6 4
1.9
0 0
0 0
0 0
15.4s
7 7
06
0 8
1 3
2.7
0 9
160
15 0
13.0
14.0
15.0
/xg/m3
Secondary
Ammonium
Nitrate
0.6
7.0
14.5
1 0
0.0
0.0
00
00
0.0
0.2
0.2
0.6
0.9
0 1
-
--
--
-
^Secondary organic carbon.
'Biomass burning
mPnmary crude oil
"NaCl + NaNO3
°Lime
PRoad sanding material
^Asphalt industry.
Misc.
Source 1
4.0h
0.7m
0.6)
9 3h
61. 3p
7.3p
00
15.1'
08'
0.0
0.0
0.0
0.0
0.0
9.3'
3.4'
6.6'
3.8'
3.4'
sRegional
Misc.
Source 2
0.0
1.8"
0.6h
0.0
00
00
00
2.2"
0.3h
0.0
0.0
0.0
0.0
00
0.0
11 Ox
8.7X
5.0X
7.9X
sulfate.
Misc.
Source 3
0.0
0.0"
7.8k
0.0
0.0
0.0
84 lr
00
i.r
0.0
0.0
0.0
0.2k
0.0
0.0
0.0
0.0
0.0
0.0
Misc.
Source 4
0.0
0.0
0.0
0.0
0.0
0.0
0 0
0.0
7.78
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Measured
PM10
Concentration
19.0
62.4
58.0
20.6
208 0
27.0
100.0
80.1
41 0
20.4
24 9
300
41.0
15.0
66.0
60.0
62.0
46.0
490
'Steel mills.
"Refuse incinerator.
"Local road dust, coal
yard road
dust, steel
haul road dust.
"Incineration.
"Unexplained mass.
rPhosphorus/phosphate industry
-------
1 more than 50% of the average PM10 mass concentration. The average fugitive dust source
2 contribution is highly variable among sampling sites within the same urban areas, as seen by
3 differences between the Central Phoenix (33 ng/m3) and Scottsdale (25 /zg/m3) in Arizona.
4 • It is also highly variable between seasons, as evidence by the summer and winter
5 contributions at Rubidoux, CA. In general, these studies found that fugitive dust was
6 chemically similar, even though it came from different emitters, so that further
7 apportionment into sub-categories was not possible. An exception was for road sanding in
8 Telluride, CO. Road sand often contains salts that allow it to be distinguished from other
9 fugitive dust sources. It is usually the only exposed fugitive dust source when other sources
10 are covered by snowpack. Dust from some construction activities and cement plants can also
11 be separated from other sources due to enrichments in calcium content of these emissions, as
12 seen in studies at Rubidoux, CA and Rillito, AZ (near cement plants), in Pocatello, ID (near
13 chemical and fertilizer production plants), and Tucson, AZ (where a nearby community
14 center was undergoing renovation).
15 Primary motor vehicle exhaust contributions account for up to approximately 40% of
16 average PM10 at many of the sampling sites. Vehicle exhaust contributions are also variable
17 at different sites within the same non-attainment area. Vegetative biomass burning, which
»
18 includes agricultural fires, wildfires, prescribed burning, and residential wood combustion,
19 was found to be significant at residential sampling sites such as: Craycroft, Scottsdale, and
20 West Phoenix, AZ; San Jose, Fresno, Bakersfield, and Stockton, CA; Telluride, CO; Sparks,
21 NV; and Mingo, OH. The predominance of these contributions during winter months and
22 the local rather than regional coverage indicates that residential wood combustion was the
23 major sub-category, even though chemical profiles are too similar to separate residential
24 combustion from other vegetative burning sources. For example, Chow et al. (1988) show
25 substantial differences between the residential Sparks, NV and urban-commercial Reno, NV
26 burning contributions even though these sites are separated by less than 10 km.
27 Sites near documented industrial activity show evidence of that activity, but not
28 necessarily from primary particles emitted by point sources. Hayden, AZ, for example,
29 contains a large smelter, but the major smelter contributions appear to arise from fugitive
30 emissions of or copper tailings rather than stack emissions. Secondary sulfate contributions
31 at Hayden were low, even though sulfur dioxide emissions from the stack were substantial
April 1995 5-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 during the time of the study. Fellows, CA, is in the midst of oilfields facilities that burn
2 crude oil for tertiary oil extraction. The Follansbee, Mingo, Sewage Plant, Steubenville, and
3 Tower sites in Ohio are all close to each other in the Ohio River Valley and show evidence
4 of the widespread steel mill emissions in that area.
5 Marine aerosol is found, as expected, at coastal sites such as Long Beach, San Nicholas
6 Island, and Anacapa Island, CA, but these contributions are relatively low compared to
7 contributions from manmade sources.
8 Of great importance are the contributions from secondary ammonium sulfate and
9 ammonium nitrate. These are especially noticeable at sites in California's San Joaquin
10 Valley (Bakersfield, Crows Landing, Fellow, Fresno, Kern Wildlife, and Stockton), in the
11 Los Angeles area, and in the Ohio River Valley. Nitrate was not measured at the Ohio sites,
12 but there was a large portion of unexplained mass in the CMB source apportionments that
13 could be composed in part by ammonium nitrate.
14 Other aerosol characterization and receptor model source apportionment studies have
15 been performed for PM10 and PM2 5 that could be added to Table 5-1. The general
16 conclusions drawn from this table would not change substantially.
17
18
19 5.4 NATIONAL EMISSION RATES AND ANNUAL TRENDS
20 Figure 5-1 (U.S. Envrionmental Protection Agency, 1993) shows the primary PM10
21 emissions estimated for the period of 1985 through 1992 using the National Trends data base.
22 PM10 fugitive dust emissions were not estimated prior to 1985. Figure 5-1 shows fugitive
23 dust from paved and unpaved roads, agricultural operations, construction, and soil erosion to
24 constitute —90% of nationwide primary emissions. All of the emissions have remained
25 relatively constant over the 8-year period except for those from soil erosion.
26 A more detailed geographical breakdown of the erosion emissions shows that the
27 majority of wind erosion occurs in the "dustbowl" region of the United States that includes
28 the Oklahoma and Texas panhandles (Barnard and Stewart, 1992; Fryrear, 1992). Wind-
29 induced erosion estimates are also highly influenced by annual precipitation and wind-speed
30 distributions. Erosion emissions estimates during 1986, 1989, and 1991 were approximately
31 twice those determined in other years owing to changes in these variables.
April 1995 5-11 DRAFT-DO NOT QUOTE OR CITE
-------
70
60
50 i
.-40
CO
£.v«Wi$~:
30
20
10
I Non-Fugitive Dust Sources
g] Wind Erosion
fH Unpaved Roads
| Paved Roads
[31 Mining and Quarrying |
| Construction
| [33 Agricultural Tilling
1985 1986 1987 1988 1989 1990 1991 1992
Year
Figure 5-1. Primary PM10 emissions estimated for 1983 to 1992.
1 Figure 5-2 (U.S. Environmental Protection Agency, 1993) expands the non-fugitive
2 dust portion of Figure 5-1 into the indicated sub-categories. PM10 emissions from these have
3 been extrapolated from early TSP estimates back to 1983. The major non-fugitive dust
4 emitters are other industrial processes (several of which include materials handling which are
5 sources of industrial fugitive PM10) and exhaust from highway vehicles. Each of these
6 emitters is only —2% of the total emissions noted in Table 5-1. Fuel combustion from
7 utilities, industrial, and other sources together contribute between 1 to 2% to total primary
8 particle emissions. Solvent use and petroleum storage and transport are included for
9 comparability to gaseous emissions inventories and have no primary particle emissions.
10 Industrial fuel combustion emissions were reduced by one-third and other fuel
11 combustion emissions were reduced by one half between 1983 and 1992, mostly owing to
April 1995
5-12
DRAFT-DO NOT QUOTE OR CITE
-------
6 r
Miscellaneous
Off-Highway Vehicles
Highway Vehicles
Waste Disposal and Recycling
Storage and Transport
Solvent Utilization
Other Industrial Processes
Petroleum and Related Industries
Metals Processing
Chemical and Allied Product Manufacture
Other Fuel Conbustion
Industrial Fuel Combustion
Electnc Utility Fuel Combustion
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
Year
Figure 5-2. Sub-categories of non-fugitive dust emissions, 1983 to 1992.
1 increased use of natural gas and the addition of particle removal devices as part of source
2 permitting. On-highway vehicle emissions increased by 50%, despite lower emissions and
3 better gas mileage in newer vehicles. This is primarily due to large increases in the number
4 of vehicle miles traveled. To the greatest extent possible, U.S. EPA (1993) adjusted
5 previous years emissions to conform to current emissions estimation methods, so the year-to-
6 year changes reflect real changes in emissions rather than changes in the estimation methods.
7 Figures 5-3 through 5-5 show national emissions for sulfur dioxide, oxides of nitrogen,
8 and VOCs for the 1983 through 1992 period. As found by U.S. EPA (1982), electric
9 utilities account for the largest fraction of sulfur dioxide, nearly 70% of total emissions.
10 These emissions have not changed substantially over the 10 years reported, and when
11 translated into metric tons, they are lower than the 19.4 million short tons/year estimated for
April 1995
5-13
DRAFT-DO NOT QUOTE OR CITE
-------
25
20
15
CO
o
I
o
10
| Miscellaneous
j§ Off-Highway Vehicles
j§ Highway Vehicles
f Waste Disposal and Recycling
§1 Storage and Transport
| Solvent Utilization
I ] Other Industrial Processes
[7] Petroleum and Related Industries
H Metals Processing
| Chemical and Allied Product Manufacture
B Other Fuel Conbustion
| Industrial Fuel Combustion
[3 Electric Utility Fuel Combustion
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
Year
Figure 5-3. National emissions of sulfur dioxide, 1983 to 1992.
1 utilities in 1978 by U.S. EPA (1982). This difference may be due to methodological
2 changes. Emissions from industrial fuel combustion increased by approximately 20% from
3 1983 to 1985, then leveled off at about 3.1 million short tons/year. A similar increase
4 between 1983 and 1985 is found for chemical manufacturing, with a leveling off at 0.42
5 million short tons/year after 1985. Sulfur dioxide emissions from highway vehicles increased
6 by 60% between 1983 and 1992, while off-highway vehicle emissions decreased to a low of
7 0.23 million short tons/year in 1986, then slowly increased to 0.27 million short tons per
8 year by 1990. Major sulfur dioxide emissions reductions are observed for petroleum
9 processing and other industrial processes, with decreases of 40% to 50% over the ten year
10 period. In total, however, sulfur dioxide emissions estimates in 1992 are identical to those
11 found in 1983 at 22.73 million short tons/year.
April 1995
5-14
DRAFT-DO NOT QUOTE OR CITE
-------
25
20
15 ]
8
I
o
10
Miscellaneous
Off-Highway Vehicles
Highway Vehicles
Waste Disposal and Recycling
Storage and Transport
Solvent Utilization
Other Industnal Processes
Petroleum and Related Industries
Metals Processing
Chemical and Allied Product Manufacture
Other Fuel Conbustion
Industrial Fuel Combustion
Electric Utility Fuel Combustion
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
Year
Figure 5-4. National emissions for oxides of nitrogen, 1983 to 1992.
1 Figure 5-4 shows a less than 5 % increase in total nitrogen oxides emissions over the ten
2 year period. Utility and motor vehicle emissions are about equal at —7.5 million short
3 tons/year, and together these account for two-thirds of total emissions. Industrial and other
4 fuel combustion and off-highway vehicles account for nearly all of the remaining third.
5 There is little change in any of the source categories from year to year, with slight reductions
6 for highway vehicles and slight increases for utility, industrial and other fuel combustion, and
7 off-highway vehicles.
8 Volatile organic compound (VOC) emissions in Figure 5-5 are dominated by highway
9 vehicles and solvent use that account for nearly 60% of total emissions. Off-highway
10 vehicles, petroleum-related industries, chemical manufacturing, and petroleum storage and
11 transport account for most of the remaining amounts. VOC emissions from highway vehicles
April 1995
5-15
DRAFT-DO NOT QUOTE OR CITE
-------
Miscellaneous
Off-Highway Vehicles
Highway Vehicles
Waste Disposal and Recycling
Storage and Transport
Solvent Utilization
Other Industrial Processes
Petroleum and Related Industries
Metals Processing
Chemical and Allied Product Manufacture
Other Fuel Conbustion
Industrial Fuel Combustion
Electric Utility Fuel Combustion
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
Year
Figure 5-5. National emissions for volatile organic compounds, 1983 to 1992.
1 were reduced by 40%, in spite of increased vehicle mileage. Most of this is due to the
2 presumed effectiveness of emissions controls on newer vehicles. VOC emissions from
3 petroleum industries were also reduced substantially, as were those from miscellaneous
4 sources. Emissions from other categories increased slightly or remained approximately the
5 same. Although U.S. EPA (1993) includes a category for "natural sources," there are
6 entries of zero emissions for all years for the category and it has, therefore, been omitted
7 from Figure 5-5.
8 The spatially and temporally averaged emissions rates presented above were compiled
9 according to standardized procedures and provide a good starting point for the identification
10 of potential contributors to ambient paniculate concentrations which exceed the PM NAAQS.
April 1995
5-16
DRAFT-DO NOT QUOTE OR CITE
-------
1 These estimates are insufficient, however, to develop effective emissions reductions strategies
2 for specific standard exceedances that might affect public health.
3 Several potentially important source categories are not included in current inventories.
4 For example, Hildemann et al. (1991) identify charbroiling and frying of meat in homes and
5 restaurants as a potential PM source neglected in previous PM10 inventories. Their studies
6 show almost all of these emissions to be in the PM2 5 size fraction. Emissions rates
7 measured for regular and extra-lean hamburger meat which was charbroiled or fried on a
8 restaurant-style grill (with commonly used grease traps) were: 40 g/kg for charbroiled
9 regular meat; 7.1 g/kg for charbroiled extra-lean meat; 1.1 g/kg for fried regular meat; and
10 1.4 g/kg for fried extra-lean meat. Fugitive dust emitters such as golf course turf
11 replacement, feedlots and dairies, equestrian events, off-road vehicle competitions, and
12 parking lot sweeping are not quantified. These emissions may be small on a national basis
13 but could be important in specific non-attainment areas. Vegetative burning sources are also
14 omitted, as are potential precursor VOC emissions from biogenic sources. Lamb et al.
15 (1993) estimate national VOC emissions from a variety of vegetative sources (based on land
16 use maps) to be =47 million short tons/year. This is twice the emissions from all other
17 sources combined in Figure 5-5.
18 As noted above, ammonia is a major participant in atmospheric reactions which form
19 ammonium sulfate and ammonium nitrate, but it is not included in the National Trends
20 inventory. Only a small fraction of total ammonia emissions are related to activity levels
21 which are quantified by current emissions inventory methods. Russell and Cass (1986)
22 attributed 52% of all ammonia emissions in the Los Angeles area to livestock; 23% to dogs,
23 cats, and humans (allocated by population); 15% to bare soil surfaces; 9% to sewage
24 treatment; and 5% to fertilizer applications. Less than 4% of ammonia emissions were
25 attributed to stationary and mobile fuel combustion sources. For 27 European countries,
26 Buijsman et al. (1987) estimated that 81% of anthropogenic ammonia is emitted by livestock
27 waste, 17% by fertilizers, and less than 2% by industrial sources.
28 Annual averages do not reflect the seasonality of certain emissions. Residential wood
29 burning in fireplaces and stoves, for example, is a seasonal practice which usually reaches its
30 peak during December in response to very cold weather. Cold weather also affects motor
April 1995 5_17 DRAFT-DO NOT QUOTE OR CITE
-------
1 vehicle exhaust emissions, both in terms of chemical composition and emission rates (e.g.,
2 Watson et al., 1990c). Planting, harvesting, and fertilizing and harvesting are also seasonal.
3 Several of the sources in Figures 5-1 through 5-5 are episodic rather than continuous
4 emitters. This is especially true of prescribed and structural fires and fugitive dust
5 emissions. For example, Engineering Science (1988) based windblown dust estimates for
6 Phoenix, AZ, on high wind speeds that occur from 5 to 10 times per year. Although
7 windblown dust emissions are low on an annual average, they are likely to be quite large
8 during those few episodes when wind speeds are high.
9 Source categories are not adequately desegregated. This is especially true of the mobile
10 sources, in which paniculate emissions from several vehicle types and emissions points are
11 combined in emissions models. Construction dust estimates, for example, are based on acres
12 under construction rather than on the individual processes involved in construction. This
13 complicates the evaluation of specific control strategies which might target specific vehicles
14 or processes.
15 As shown in Table 5-1, there is substantial variation in contributions to suspended
16 particles even for nearby sources. The national averages in Figures 5-1 through 5-5 do not
17 take into account the regionality and zones of influence of different emissions sources.
18 Finally, the emissions factors applied to activity data are of questionable accuracy. For
19 example, Engineering Science (1988) took many samples from paved and unpaved roadways
20 in Phoenix, AZ, and they found silt loadings ranging from 0.02 to 0.75 grains/ft2. An
21 average of these measurements was used to represent the loadings for all paved roads. It is
22 very likely that the loadings of suspendable dust on paved roads varies over time in ways
23 which can never be known.
24 Several of these deficiencies may be unimportant, and the source apportionments
25 provided by receptor modeling allows their importance to be evaluated. Others of these
26 deficiencies can, and should be, corrected by better emissions modeling methods. It is
27 revealing, for example, that fugitive dust constitutes 90% of the annual average emissions
28 inventory, but it seldom averages more than 50% of the contribution to average PM10
29 concentrations as evidenced in Table 5-1. The contributions from primary motor vehicle
30 exhaust, residential wood combustion, and industrial sources are definitely underestimated by
31 the relative emissions from these sources in the national emissions inventory. Some of these
April 1995 5-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 deficiencies, such as fugitive dust emissions, are probably intractable and the best that can be
2 done is to estimate the uncertainties in these emissions and to value the data accordingly
3 when decisions are made. These issues are examined in greater detail for major source
4 categories in the next section.
5
6
7 5.5 EMISSIONS PROCESSES AND ESTIMATION METHODS
8 The national emissions estimates shown in Figures 5-1 through 5-5 must be considered
9 within the context of their estimation methods. Many of these methods are empirically
10 derived, rather than process-related, and they are often extrapolated beyond the original
11 empirical data sets used to derive them. The U.S. EPA (1982) did a credible job of
12 describing estimation methods for industrial and natural sources, but it did not describe
13 estimation methods for emissions from area and mobile sources. Since Table 5-1 shows that
14 fugitive dust and motor vehicle are major contributors to suspended particles in several areas,
15 this ensuing review will focus on methods to estimate such emissions.
16
17 5.5.1 Fugitive Dust
18 Fugitive dust consists of geological material that is suspended into the atmosphere by
19 natural wind and by anthropogenic activities from sources such as paved and unpaved roads,
20 construction and demolition of buildings and roads, storage piles, wind erosion, and
21 agricultural tilling. Though qualitative descriptions of fugitive dust emissions are easy to
22 understand, translating these descriptions into quantitative estimates of emission rates,
23 locations, temporal variability, and contributions to suspended particles measured at receptors
24 has been a scientific and engineering challenge.
25 The movement of soil, and especially its suspension from the surface of the earth into
26 the atmosphere, has been studied in many branches of science. "Aeolian dust," named for
27 Aeolia, the Greek goddess of the wind, is a major discipline in the fields of geology and
28 archaeology (Pye, 1987). Wind erosion is of great concern to agriculturists and soil
29 scientists. Meteorologists study interactions between the atmosphere and the earth's surface.
30 Air pollution scientists devise methods to estimate contributions from dust suspended by the
31 wind and other mechanisms to particles which might cause adverse health effects. The
April 1995 5-19 DPxAFT-DO NOT QUOTE OR CITE
-------
1 published literature on the mechanisms for dust suspension are widely dispersed and not
2 entirely comparable in terms of the properties studied and the units of measure. Much of the
3 work on dust suspension has been done by agriculturists to minimize soil erosion. Erosion
4 estimates include all of the mechanisms which might affect the removal of topsoil from a
5 given area, one of which is suspension into the atmosphere.
6 Air pollution scientists are only concerned with that portion of eroded soil which is
7 removed by suspension into the atmosphere and transported reasonable distances (typically
8 greater than 100 m, the nominal dimension of significant fugitive dust emissions sources)
9 without deposition to the surface. This suspension depends on: (1) particle size of the
10 erodible material; (2) surface loadings; (3) moisture; (4) surface roughness; (5) wind speed
11 and direction; (6) suspension height; and (7) vehicular ejection mechanisms. The current
12 status of knowledge of each of these characteristics is discussed below.
13 Very little is known about the respirable size fractions in dust, despite their adverse
14 health potential, long residence times, and high potential for vertical mixing. The most
15 comprehensive information on particle sizes in geological material is contained in soil
16 surveys compiled by the Soil Conservation Service (SCS). These surveys provide detailed
17 boundaries for different soil types on 7.5-minute maps corresponding to U.S. Geological
18 Survey (USGS) maps. The codes are associated with data in a printed summary which
19 accompanies the maps for each survey area.
20 Particle sizes in soil surveys are indicated by qualitative descriptions in terms of the
21 amount of sand (50 to 2,000 /*m geometric diameter), silt (2 to 50 jum geometric diameter),
22 and clay (< 2 /*m geometric diameter). These particle size fractions in the soil survey are
23 estimated by the individuals conducting the survey based on the visual similarity of the
24 observed soils to a sub-set of soil samples which are submitted to particle size analysis in a
25 laboratory. AP-42 emissions factors for fugitive dust often contain silt loading as one of
26 their input variables largely because of the availability of SCS soil survey data for large
27 portions of the United States.
28 The particle sizing procedure American Society for Testing and Materials, 1990a;
29 1990b) which is most commonly followed for soil surveys creates a soil/water suspension in
30 which soil aggregates are broken into their component parts prior to sieving. While this
31 disaggregation is useful for agricultural, construction, and other land uses, it is not especially
April 1995 5-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 useful for estimating air pollution emissions because it does not accurately estimate the size
2 of aggregates which are suspended by surface winds. Gillette (1980) applied two methods to
3 determine the small particle content of soil which might be entrained by winds and cause
4 pollution problems. The first method (i.e., gentle sieve) consists of drying the soil sample
5 and sieving it gently with about twenty circular gyrations parallel to the plane of the sieve
6 (Chepil, 1952). The second method (i.e., hard sieve) consists of up to one-half hour of
7 vigorous shaking (usually using a shaking machine). The gentle sieve method best represents
8 the suspension properties of the soil in the state in which it was sampled. The hard sieve
9 method represents the potential of that soil for resuspension when disaggregating activities
10 (e.g., vehicle traffic) occur. Gillette's threshold suspension velocity measurements apply to
11 soil characteristics obtained by the gentle sieve method.
12 The size distribution of dust particles affects the suspension process. A flat bed of
13 particles with diameters less than 20 /xm is very difficult to suspend by wind, as Bagnold
14 (1937) showed by blowing wind in excess of 100 cm/sec over a bed of fine Portland cement.
15 In this situation, there is no large cross section for wind to act on. A bed of large particles
16 with diameters exceeding 1 mm interspersed with fine particles also mitigates suspension.
17 Particles larger than 0.5 mm cannot be lifted by wind, but they absorb wind energy as they
18 roll along the surface. They also shelter smaller particles on their lee sides from the effects
19 of high wind speeds. Gillette and Stockton (1989) sprinkled glass spheres with diameters
20 ranging from 2.4 to 11.2 mm onto a bed of glass spheres with sizes from 0.107 to 0.575 /mi
21 and found major reductions in suspension of the smaller particles. Bagnold (1941) estimated
22 that 80 jum particles are the most susceptible to suspension by wind, even though their large
23 masses cause them to settle to the surface very rapidly.
24 The amount of suspendable dust on a surface influences how much might be suspended.
25 Most surfaces are limited reservoirs, and the suspendable dust is depleted after a short time
26 period. On exposed land, fine particle which are blown away often expose larger rocks
27 which then shield the suspendable particles from the wind. When surfaces are continually
28 disturbed, however, by very intense winds or by vehicular movement, they may become
29 unlimited reservoirs which emit dust whenever winds exceed threshold suspension velocities.
30 There are few (< 500 for the entire U.S.) reported data on the surface loadings of silt
April 1995 5_2i DRAFT-DO NOT QUOTE OR CITE
-------
1 (< 75 /nm) and absolutely no data on surface loadings of respirable particle sizes for any
2 surface included in fugitive dust emissions inventories.
3 Water adheres to individual soil particles, thus increasing their mass and mitigating
4 suspension and transport. It also increases the cohesive forces among individual particles,
5 and these forces often persist after the water has evaporated as evidenced by aggregates and
6 surface crusts. Chepil (1956), Belly (1964), Bisal and Hsieh (1966), and Svasek and
7 Terwindt (1974) show that substantially greater wind forces are needed when soil surface
8 moisture is increased by less than 1 % from its dry state.
9 Soil surveys include plastic limits and moisture limits to the ability of soils to absorb
10 moisture. The "plastic limit" is the moisture content at which a soil changes from a semi-
11 solid to a plastic. The plastic limit is determined by adding water to dry soil sample until it
12 can be rolled into a coherent cylinder. Soil surveys also report liquid limits (the quantity of
13 water required to create a slurry with the consistency of water), the infiltration rate (the
14 movement of water through soil layers), and field moisture capacity.
15 Kinsey and Cowherd (1992) show how watering might reduce emissions at a
16 construction site. A large pollution control benefit might be derived from initially doubling
17 the area which is watered with lower benefits achieved as more water is applied to the site.
18 Control efficiency is ultimately limited because grading operations are continually exposing
19 dry earth and burying the moistened topsoil. A portion of this moistened soil adheres to the
20 construction vehicles and can be carried out to paved and unpaved roads for subsequent
21 resuspension.
22 While the moisture capacities and retentions of different geological materials are well
23 documented in the soil surveys, the actual moisture content at a given time or place is not
24 recorded. Thornthwaite (1931) proposed the ratio of precipitation to evaporation as an
25 indicator of the availability of moisture for soils. Thornthwaite's major concern was the
26 agricultural potential of land in different areas. The precipitation-evaporation effectiveness
27 index (P-E index) is 10 times the sum of the monthly precipitation to evaporation ratios.
28 Using precipitation, evaporation, and temperature data taken prior to 1921 at twenty-one
29 U.S. monitoring sites, Thornthwaite (1931) devised an empirical precipitation-evaporation
30 index (P-E index) to classify all North America as wet (P-E index > 128), humid
31 (64 < P-E index < 128), sub-humid (32 < P-E index < 64), semi-arid (16 < P-E index
April 1995 5-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 < 32), or arid (P-E index < 16). Much of the western United States is in the arid and
2 semi-arid categories, while much of the eastern United States is in the humid category. The
3 P-E index is cited for several AP-42 emission factors to estimate the moisture content of
4 different soils.
5 Moisture content of soils will vary throughout the year depending on the frequency and
6 intensity of precipitation events, irrigation, and relative humidity and temperature of the
7 surrounding air. Large amounts of rain falling during 1 mo of a year will not be as effective
8 in stabilizing dust as the same amount of rain interspersed at intervals throughout the year.
9 Moisture also causes dust to adhere to vehicle surfaces so that it can be carried out of
10 unpaved roads, parking lots, and staging areas. Carry out also occurs when trucks exit
11 heavily watered construction sites. This dust is deposited on paved roadway surfaces as it
12 dries, where it is available for suspension far from its point of origin. Fugitive dust
13 emissions from paved roads are often higher after rainstorms in areas where unpaved
14 accesses are abundant, even though the rain may have flushed existing dust from many of the
15 paved streets.
16 Windblown dust is a major contribution from all exposed surfaces. Each surface has a
17 threshold velocity that depends on the cohesiveness of the particles and the surface
18 roughness. Surface roughness height is the distance above average ground level at which the
19 average wind velocity approaches zero. Surface roughness is also a meteorological concept;
20 it is related to, but not identical to, measured heights of obstructions in an open area. Larger
21 surface roughness shelters suspendable dust from suspension. Surface roughness depends
22 both on the height and spacing of roughness elements, but is not well characterized for most
23 surfaces. Friction velocity is the slope of the logarithm of wind speed versus elevation above
24 ground level.
25 Dust arises due to suspension of the disturbed surface by wind. Chepil and Woodruff
26 (1963) and Gillette and Hanson (1989) show that the amount of soil which can be suspended
27 by wind depends on the particle size distribution, wind velocity at the soil surface, the
28 roughness of the surface, the relative fractions of erodible (< 2 /mi diameter) and non-
29 credible (> 2 jum diameter) material, and the cohesion of the soil particles with one another.
30 Values for each of these variables affect other variables. For example, a higher moisture
31 content increases cohesion among particles and shifts the size distribution to larger particles.
April 1995 5-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 Larger agglomerations of small particles increase surface roughness which decreases wind
2 speeds at the surface.
3 All of these variables collectively affect threshold friction velocity, which is
4 experimentally determined by placing a wind tunnel over an example of the affected soil and
5 measuring the velocity near the surface at which visible soil movement is first observed.
6 Wind speeds must approach 0 near the earth's surface, and experiments show that wind
7 speed increases exponentially with height. Both friction velocity and surface roughness are
8 determined experimentally for different situations by taking measurements at different
9 elevations. When the actual friction velocity is less than the threshold friction velocity for
10 soil erosion, no particles are suspended into the atmosphere. Most ambient wind speed
11 measurements are made at elevations between 5 and 10 m above ground level, and these
12 must be translated to surface friction velocities to determine suspension. For a given
13 threshold friction velocity, paniculate emissions factors for windblown dust use the fastest
14 mile wind speed as reported in the National Weather Service (NWS) Local Climatological
15 Summaries. Gillette (1980) shows that threshold wind speeds vary from 0.19 to 1.82 m/s for
16 disturbed soils. Even though emissions are initiated at these velocities, the wind force
17 contains insufficient energy to suspend much of the erodible soil mass. The amount of dust
18 suspended increases at approximately the cube of the wind speed above the threshold
19 velocity.
20 Particles suspended into the atmosphere are acted upon by gravity in a downward
21 direction and by atmospheric resistance in an upward direction. Every particle attains an
22 equilibrium between these forces at its terminal settling velocity. The settling velocity
23 increases as the square of the particle diameter or when the particle density increases. For
24 very small particles (< 10 ptm diameter), vertical air movements caused by turbulence can
25 counteract the gravitational settling velocity and such particles can remain suspended for long
26 times. Particle deposition for particles larger than ~20 /*m diameter is dominated by the
27 force of gravity, however. Transport distance depends on the initial elevation of a particle
28 above ground level, the horizontal wind velocity component in the direction of interest at the
29 particle elevation, and the gravitational settling velocity. For this reason, large TSP
30 emissions rates do not give a good impression of contributions to suspended particles that are
31 very distant from the receptor site. Pye (1987) shows vertical profiles for different sized
April 1995 5-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles which might be elevated through a 100 m depth during a wind storm. The particles
2 smaller than 10 /*m are nearly uniformly distributed through this depth, while the larger
3 particles exhibit much higher concentrations closer to the surface.
4 Dust on paved roads, unpaved roads, parking lots, and construction sites is suspended
5 by natural winds and vehicular movement. Vehicular traffic in these areas adds to
6 suspension because tire contact with the road lifts particles into the air. Vehicle wakes
7 create turbulent eddies which act much like natural winds to raise particles. The grinding of
8 particles by tires against the road surface shifts the size distribution toward smaller particles,
9 especially those in respirable size fractions.
10 Unpaved roads and other unpaved areas with vehicular activity are unlimited reservoirs
11 of dust loading when vehicles are moving. These surfaces are always being disturbed, and
12 wind erosion seldom has an opportunity to increase their surface roughness sufficiently to
13 evade particle suspension. The U.S. EPA AP-42 emission factor (U.S. Environmental
14 Protection Agency, 1988) for unpaved road dust emissions contains variables which account
15 for silt loading, mean vehicle speed, mean vehicle weight, mean number of wheels, and
16 number of days with detectable precipitation, to determine annual PM10 dust emissions for
17 each vehicle-kilometer-traveled. These relationships are derived from imprecise correlations
18 of variables, however, and a full physical understanding of the vehicle suspension process is
19 lacking for unpaved roads. Muleski and Stevens (1992) note that more than 90% of the tests
20 which acquired data for the AP-42 factor were conducted with vehicle speeds slower than
21 56 km/h (35 mph), and more than 80% was derived from industrial haul roads. AP-42
22 emission factors may not be applicable to publicly maintained unpaved roads, desert
23 shortcuts, and agricultural roads which are common in most PM10 non-attainment areas.
24 Dust on paved roads must be continually replenished, however, and reducing the
25 deposition of fresh dust onto these surfaces is a viable method for reducing particulate
26 emissions. Dust loadings on a paved road surface build up by being tracked out from
27 unpaved areas such as construction sites, unpaved roads, parking lots, and shoulders; by
28 spills from trucks carrying dirt and other particulate materials; by transport of dirt collected
29 on vehicle undercarriages; by wear of vehicle components such as tires, brakes, clutches, and
30 exhaust system components; by wear of the pavement surface; by deposition of suspended
31 particles from many emissions sources; and by water and wind erosion from adjacent areas.
April 1995 5_25 DRAFT-DO NOT QUOTE OR CITE
-------
1 The relative contribution from each of these sources is unknown. Axetell and Zell
2 (1977) estimated typical deposition of 240 Ib/mile of curb/day for particles of all sizes from
3 the following sources: (1) 42% from mud and dirt carryout; (2) 17% from litter;
4 (3) 8% from biological debris; (4) 8% from ice control compounds (in areas with cold
5 winters); (5) 8% from erosion of shoulders and adjacent areas; (6) 7% from motor vehicles;
6 (7) 4% from atmospheric dustfall; (8) 4% from pavement wear; and (9) less than 1% from
7 spills. These proportions are highly uncertain because they apply to the TSP size fraction
8 (rather than to the respirable size fractions) and because these investigators did not consider
9 all of the sources cited above. Axetell and Zell (1977) cite these fractions without describing
10 the methodology used to estimate them. No other published quantitative apportionments of
11 paved road dust loadings to their sources were found.
12 Nicholson et al. (1989) and Mollinger et al. (1993) identify the turbulent wake of
13 vehicles on roads as a major cause of dust suspension from roads; their research suggests
14 that vehicle shapes might be altered to reduce emissions potential. Mollinger et al. (1993)
15 mounted a cylinder, an elliptical cylinder, and a rectangular solid on a pendulum which
16 swung back and forth over dust-covered test areas. After 20 passes by the cylinder and
17 elliptical cylinder, 65% and 45% of the dust remained in the test area, respectively. After
18 20 passes by the rectangular solid traveling at the same velocity, less than 20% of the dust
19 remained.
20 Other than the information inferred from the chemical composition of road dust and
21 from multivariate relationships between downwind concentrations and vehicle variables, there
22 is no detailed physical understanding of the effects of tire contact with particles and their
23 suspension into the atmosphere. This knowledge is essential to understanding how these
24 particles are suspended and how far they are transported.
25 There are obvious discrepancies between the proportion of fugitive dust in primary
26 emissions and geological contributions to PM10 calculated by receptor models. To some
27 extent, this is due to contributions from secondary aerosols, which are not included in the
28 primary PM10 emission estimates. Even when secondary aerosol is subtracted, however,
29 other sources such as vegetative burning and wood combustion make larger relative
30 contributions to ambient concentrations than is indicated by the emissions inventories.
April 1995 5-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fugitive dust estimates are especially affected by the general limitations of emissions
2 inventories described above. Annual and spatial averages do not reflect the seasonality of
3 certain emissions. Planting and harvesting are seasonal, as are high winds which might cause
4 erosion. Paved road dust emissions might be much higher after rain storms when dirt is
5 tracked from unpaved areas onto paved roads.
6 Many fugitive dust sources are episodic rather than continuous emitters. For example,
7 Engineering Science (1988) based windblown dust estimates for Phoenix, AZ, on high wind
8 speeds that occur 5 to 10 times per year. Though windblown dust emissions are low on an
9 annual average, they can be quite large during those few episodes when wind speeds are
10 high. Construction activities are also episodic in nature. Reeser et al. (1992) reported
11 fugitive dust emissions during wintertime in Denver, CO to be 44% higher than those found
12 in the annual inventory using standard emissions inventory methods. In Coachella Valley,
13 CA, the South Coast Air Quality Management District (SCAQMD) (1994) calculated 24-h
14 emissions based on a worst windy day. When wind gust speeds exceeded 96 km/h, fugitive
15 dust emissions could account for 20% of the entire annual average emission rates.
16 As noted earlier, many fugitive dust emitters are included in the inventories. For
17 example, Chow et al. (1992) identified two cement plants and many roads with unpaved
18 shoulders near the Rubidoux, CA, site. The cement plants were not included in the
19 SCAQMD emissions inventory, and there is no distinction in any of the inventories between
20 curbed and swept roads and those with no shoulders that may be dirtier than others.
21 Finally, the spatial disaggregation for fugitive dust emissions is poorer than that for all
22 other source categories. Whereas most mobile sources are confined to established roadways
23 and most area sources correspond to population density, suspendable dust is everywhere.
24 Most fugitive dust emissions are compiled on a county-wide basis and are not allocated to
25 specific fields, streets, unpaved roads, and construction sites possibly contributing to high
26 airborne PM concentrations. Several of these limitations may be impossible to overcome,
27 but many result from old methods being applied to the problem. Modern data bases,
28 computer systems, and information management software can be applied to improve existing
29 inventories without major additional costs beyond initial investments in establishing an
30 inventory methodology.
31
April 1995 5-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.5.2 Mobile Source Emissions
2 Mobile sources are major emitters of primary particles, oxides of nitrogen, and VOCs.
3 They are also minor emitters of sulfur dioxide and ammonia. On-road motor vehicles using
4 gasoline- and diesel-fueled engines are by far the largest component of mobile source
5 emissions, and the emissions estimation methods are most highly developed for these
6 vehicles. The U.S. EPA (1994) has published the MOBILE model to estimate carbon
7 monoxide, oxides of nitrogen, VOC, and primary particle emissions from on-road vehicles.
8 The other model in common use is the California Air Resources Board's Emissions FACtor
9 (EMFAC) model (CARB, 1993). These models use as their inputs variables such as vehicle
10 speed, vehicle age distribution, vehicle classification, ambient temperature, and laboratory
11 emissions test data. The laboratory test data generally use dynamometers with Federal Test
12 Procedure (FTP) driving cycles.
13 MOBILE and EMFAC model estimates of carbon monoxide, oxides of nitrogen, and
14 VOC emissions are variable as a function of the input variables. Primary particle emissions
15 from tire and break wear and from exhaust are constant, however, regardless of assumed
16 driving conditions. Emissions estimates from these models have recently been called into
17 question by a series of on-road emissions measurement experiments, which consist of
18 measurements made in tunnels, along roadsides, and by pulling over vehicles for on-road
19 inspection and dynamometer testing.
20 The Southern California Air Quality Study (SCAQS) Van Nuys Tunnel Study (Ingalls et
21 al., 1989), conducted in October and December, 1987 in the Sherman Way tunnel under the
22 Van Nuys Airport, first called modeled emissions into question by noting that the emissions
23 models underestimated measured VOC emissions by nearly a factor of four. Nitrogen oxide
24 emissions in the tunnel, however, were consistent with those calculated by the models.
25 Pierson et al. (1990) concluded that the results were, in the main, correct. Even if the
26 absolute values of the emission factors determined were somewhat biased due to inaccurate
27 air flow measurements (one of the major difficulties pointed out by Pierson et al.) the
28 emission factor ratios are still valid.
29 Other tunnel experiments were carried out in the Ft. McHenry Tunnel in Baltimore,
30 Maryland, in June 1992 and in the Tuscarora Mountain Tunnel on Interstate 76 in south-
31 central Pennsylvania, in September 1992 (Pierson et al., 1995). Both tunnels have
April 1995 5-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 considerable fetches of freeway before them; thus, all vehicles should have been in the hot-
2 stabilized mode when they entered. In both tunnels the traffic maintained steady speeds with
3 low run-to-run and vehicle-to-vehicle variability. This is in distinct contrast to the Van Nuys
4 tunnel where there was considerable variation in vehicle speeds and many different speeds.
5 Both MOBILE4.1 and MOBILES modeled running losses fell within the range of
6 source contributions determined by the CMB model from VOC source profiles for tailpipe
7 exhaust and gasoline headspace (Gertler et al., 1995) For the driving mode in these tunnels,
8 failure to model hydrocarbon emissions is not exclusively due to the method which estimates
9 running loss evaporative emissions. Fuel volatilities were lower at both Ft. McHenry and
10 Tuscarora than during the Van Nuys tunnel experiment. While this may lower evaporative
11 emissions, it is not enough to account for the discrepancy between the observed and modeled
12 emission factors at Van Nuys.
13 The emissions models tend to overestimate at Tuscarora and only slightly underestimate
14 emissions at Ft. McHenry, which are reasonable results given the sources of the emission
15 factors in the models and traffic conditions at each tunnel. As stated previously, the models
16 derive their emission factors from the FTP, which has many accelerations and decelerations
17 and very little steady-speed driving. At Tuscarora, there was little acceleration or
18 deceleration. The tunnel is virtually flat and contains no turns. It is more than 10 km from
19 the nearest interchange and many, if not most, commuter vehicles travel for much longer
20 distances. These vehicles were relatively new (median model year was 1989 during the 1992
21 experiment) and presumably well maintained. The average speed of light-duty vehicles in the
22 tunnel (determined by a hand-held radar gun) was 59.4 mph with a vehicle-to-vehicle
23 variability of 5.6 mph, which is negligible variability compared to that seen at the Van Nuys
24 tunnel. Traffic in the Tuscarora tunnel is expected to emit at low levels because of:
25 (1) a flat roadway, (2) steady driving speeds, (3) relatively new and well maintained
26 vehicles, and (4) vehicles in hot-stabilized driving mode. These conditions are consistent
27 with basic assumptions of the emissions models.
28 The Ft. McHenry tunnel is a different situation, with up- and down-grades reaching
29 +3.76% and more speed variability. The vehicle average speeds were 51 mph upon entering
30 the tunnel and 43 mph at the exit. The median model year was 1989 for automobiles, and
31 given the tunnel's location on a major freeway, it is assumed that all vehicles were in hot-
April 1995 5_29 DRAFT-DO NOT QUOTE OR CITE
-------
1 stabilized driving mode. Though Ft. McHenry is a more complex driving situation than
2 Tuscarora, it is still considerably less complex than the Van Nuys situation. At Ft.
3 McHenry, MOBILE4.1 underestimates and MOBILES overestimates emissions, but only
4 slightly in either case. None of the discrepancies are as severe as the underestimates at Van
5 Nuys. The models agree well with the Tuscarora and Ft. McHenry tunnel observations,
6 even when deviations from assumptions such as road grades are significant, but when
7 challenged by a complex urban driving situation such as Van Nuys, it is uncertain at this
8 time how well the models perform.
9 Stedman et al. (1994) and Ashbaugh et al. (1992) conducted a study in the Los Angeles
10 area in 1991 using remote sensing devices for the detection of VOC and carbon monoxide
11 emissions from individual vehicles. During the study, more than 60,000 vehicles' VOC and
12 CO emissions were measured by remote sensing. Of that group, more than 300 high-
13 emitting vehicles were identified by the remote sensors and pulled over for inspection.
14 Sixty-seven percent of the inspected vehicles had emission control systems that were
15 defective or had been tampered with; more than 90% of them failed the Smog Check
16 inspection.
17 During the same time, approximately 80 of those inspected vehicles were given an
18 IM240 test (a 240 second, loaded mode test on a dynamometer). These vehicles previously
19 had been stopped because of high remote sensing CO and/or VOC readings. All but one of
20 the tested vehicles exceeded the California certification emission standards. The ten highest
21 CO and VOC emitters had mass emission rates that were 24 to 70 times higher than the
22 standards. NOX emissions also were measured in the IM240 tests, and three of the vehicles
23 had NOX mass emission rates greater than 10 g/mi. Even though the vehicles had been
24 stopped because of their CO or VOC remote sensing readings, high NOX emitters were
25 encountered. This unexpected finding suggests the presence of high NOX emitters in the fleet
26 (Knapp, 1992) and has important implications for the accuracy of NOX emissions in the
27 mobile source emissions inventory. Though particle emissions were not measured in these
28 experiments, it is likely that inefficient combustion and pollution controls on these vehicles
29 enhance the production of carbonaceous aerosols.
30 Ashbaugh and Lawson (1991) analyzed data collected in 1985, 1987 and 1989 random
31 roadside surveys and showed that, for the low-idle test, 20% of the vehicles were responsible
April 1995 5-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 for 80% of the CO emissions and 20% (not necessarily the same vehicles) were responsible
2 for 80% of the VOC hydrocarbon (HC) emissions.
3 While several tests have examined on-road emissions of both nitrogen oxide and VOC
4 precursors, only Hansen and Rosen (1990) report individual on-road vehicle measurements of
5 particulate emissions. They measured the ratio of light-absorbing carbon to CO2 in the
6 exhausts of 60 gasoline-fueled vehicles in Berkeley, CA. The ratio of carbon to excess CO2
7 above background provides an estimate of emissions per unit of fuel combusted. By making
8 reasonable assumptions about speed and gasoline mileage, these ratios can be translated into
9 grams/vehicle mile traveled. Their experiment found a factor of 250 between the highest and
10 lowest ratio of light-absorbing carbon to CO2 for the 60 vehicles tested. This is at odds with
11 use of a single value for primary particle emissions, regardless of vehicle type or operating
12 condition, typically employed in current emissions models.
13 These studies show that while vehicle emissions models may function well under
14 idealized conditions, they underestimate the effects of high emitting vehicles that may be
15 major sources of VOCs. Though data are lacking to verify the primary particle emissions in
16 these models, it is very unlikely that the values currently in use account for the variability in
17 emissions from different types of vehicles.
18
19
20 5.6 SIZES DISTRIBUTIONS OF PRIMARY PARTICLE EMISSIONS
21 Recent measurements of the size distributions of primary particles confirm U.S.
22 Environmental Protection Agency (1982a) conclusions that most fugitive dust emissions are
23 in particles larger than 2.5 pm and that the majority of emissions from combustion sources
24 are in sizes smaller than 2.5 /mi. Figures 5-6 and 5-7 from Houck et al. (1989, 1990) were
25 derived from a major characterization of different source emissions in California conducted
26 during 1986. Hot exhaust samples were diluted to ambient temperatures prior to sampling
27 onto filter media through impactor inlets with 50% cut-points of 1, 2.5, 10, and —30 /mi.
28 These figures show that combustion products are nearly always less than 2.5 /im in size.
29 Figure 5-8 shows examples of size distributions in dust from paved and unpaved roads,
30 agricultural soil, sand and gravel, and alkaline lake bed sediments which were measured in a
31 laboratory resuspension chamber as part of the California study (Chow et al., 1994). This
April 1995 5-31 DRAFT-DO NOT QUOTE OR CITE
-------
100
'
80
60
Q_
1
0)
a.
40
20
0
~
-
—
|H
in
1
f
fK^:--:':~':.
y//M'/3aftf-'faf>'#.
52.3% (<10^)
10.7% (<2.5n)
^^^g
,^\>.
x V"V,
\^4
r -U' " •
•;';.i"* V
1|
92.8% (<10n)
82.7% (<1 n)
81.6%(<2.5n)
••••••
-^"^x?"
,;|
\>^:
;^%-
IS
1
||
IsJ
1
95.8% (<10n)
93.1%(<2.5n)
92.4% (<1 n)
Code
^I^\ >1Q\i
2.5u - 10u
•^^ ™ • ** "
^ 1^-2.5(1
E3
-------
100
80
60
Q_
P
I
o>
Q_
40
20
96.2%
92.3% (<2.5\i
91.8%
Diesel
Truck
Emissions
99.2%
97.4%
87.4%
Crude Oil
Combustion,
Mean of 2
Composites
Code
34.9% (<1 Oji)
5.8%
4.6%
Fresno
Construction
Figure 5-7. Size distribution of California particle emissions, 1986.
April 1995
5-33
DRAFT-DO NOT QUOTE OR CITE
-------
100
Paved Unpaved Agriculture Soil/Gravel
Road Dust Road Dust Soil
EI3<1.0nm [ZI|<2.5nm ••< 10 urn
Alkaline
Lake Bed
Figure 5-8. Particle size distribution in laboratory resuspension chamber.
1
2
3
4
5
6
7
8
9
10
11
figure shows substantial variation in primary particle size among some of these sources. The
PMj o abundance (6.9%) in the alkaline lake bed dust is twice its abundance in paved and
unpaved road dust. Approximately 10% of TSP is in the PM2 5 fraction and approximately
50% of TSP is in the PM10 fraction. The sand/gravel dust sample shows that 65% of the
mass consists of particles larger than the PM10 fraction. The PM2 5 fraction of TSP in
alkaline lake beds and sand/gravel is approximately 30% to 40% higher than the other soil
types. Particle emissions in the respirable fraction can be expected to vary substantially
among these different fugitive dust sources.
April 1995
5-34
DRAFT-DO NOT QUOTE OR CITE
-------
1 5.7 CHEMICAL COMPOSITIONS OF PRIMARY PARTICLE
2 EMISSIONS
3 The widespread use of receptor models since U.S. EPA (1982a) has resulted in many
4 published chemical compositions for different paniculate sources. Table 5-2 shows the
5 relative amounts of different chemical species in the profiles from different source types.
6 These are deduced from many different profiles compiled by Radian (1989) and by Chow
7 et al. (1994) for EPA's SPECIATE source composition library.
8 Figures 5-9 through 5-12 show examples of the chemical abundances in specific PM2 5
9 profiles for paved road dust, motor vehicle exhaust, residential wood combustion, and a coal-
10 fired power plants. These were measured in Denver, CO during 1987 (Watson and Chow,
11 1994). Substantial differences in chemical composition exist for these emitters, while these
12 differ from site to site.
13 The road dust profile in Figure 5-9 contains large abundances of aluminum, silicon,
14 potassium, calcium, and iron. Though total potassium is abundant in road dust,
15 water-soluble potassium constitutes less than one-tenth of the total. Strontium and lead are
16 also present at detectable levels. Paved road dust is much like an ambient PM10 sample,
17 with a complex combination of particulate matter from a wide variety of sources, especially
18 other geological source types. This complexity is evident in the comparison of a paved road
19 dust profile reported by Chow et al. (1991) for Phoenix, AZ, with profiles from other
20 geological sources in the area. Chow et al. (1991) noted that the abundance of organic
21 carbon in the profile was 11+9%, larger and more variable than its abundance in profiles
22 from agricultural land, construction sites, and vacant lots. The presence of tire wear,
23 detritus, and engine oils can account for this enrichment. This organic carbon content
24 places an effective upper limit on the contribution from tire wear and other carbon sources to
25 suspendable paved road dust. Approximately 25 % of tire material consists of styrene-
26 butadiene rubber (SBR) (Pierson and Brachaczek, 1974). Ondov (1974) measured elemental
27 components of tire material and found minor concentrations for most species, with S (-2%),
28 Cl (~ 1.5%), and Zn (~ 1 %) being the most abundant components.
29 The abundances of Pb and Br in Phoenix paved road dust were more than double the
30 concentrations in the other geological profiles, indicating the presence of tailpipe exhaust
April 1995 5.35 DRAFT-DO NOT QUOTE OR CITE
-------
1
1— 1
VO
i
0-v
O
6
O
O
H
O
Cl
O
H
W
O
50
n
TABLE 5-2. TYPICAL CHEMICAL ABUNDANCES IN SOURCE EMISSIONS.
Chemical Abundances
Source Type
Geological Material
Motor Vehicle
Vegetative Burning
Residual Oil Combustion
Incinerator
Coal Fired Power Plant
Marine
Dominant
Particle Size
Coarse
Fine
Fine
Fine
Fine
Fine
Coarse
< 0.1%
Cr, Zn, Rb, Sr, Zr
Cr, Ni, Y+
Ca, Mn, Fe, Zn,
Br, Rb, Pb
K+, OC, Cl, Ti,
Cr, Co, Ga, Se
V, Mn, Cu, Ag, Sn
Cl, Cr, Mn, Ga,
As, Se, Br, Rb, Zr
Ti, V, Ni, Sr, Zr,
Pd, Ag, Sn, Sb, Pb
0.1 to 1%
Cr, N03-, S04=,
NH4+, P, S, Cl, Ti,
Mn, Ba, La
NH4+, Si, Cl, Al,
Si, P, Ca, Mn, Fe,
Zn, Br, Pb
NOj, S04=, NH4+,
Na+
Na+, NH4+, Zn,
Fe, Si
K+, Al, Ti, Zn,
Hg
NH4+, P, K, Ti, V,
Ni, Zn, Sr, Ba, Pb
K, Ca, Fe, Cu, Zn,
Ba, La, Al, Si
1 to 10% > 10%
OC, EC, Al, K, Si
Ca, K, Fe
S, Cl, NO§, OC, EC
S04=, NH4+
K+, K, Cl, Cl- OC, EC
Ni, OC, EC, V S, SO4=
NO§, Na+, EC, SO4=, NH4+,
Si, S, Ca, Fe, OC, Cl
Br, La, Pb
SO4=, OC, EC, Si
Al, S, Ca, Fe
NOj, S04=, Na + , Na, Cl",
OC, EC Cl
-------
<3> v <3*
Chemical Compound
Figure 5-9. Chemical abundances for PM2 5 profiles of road dust.
^w- ^
-------
^^^^^^^/^^^f
VN
Chemical Compound
Figure 5-11. Chemical abundances for PM2 5 profiles of wood burning.
10
Coal-fired Power Plant
Chemical Compound
Figure 5-12. Chemical abundances for PM2 5 profiles of coal-fired power plant.
April 1995
5-38
DRAFT-DO NOT QUOTE OR CITE
-------
1 from vehicles burning leaded fuels. These will not be good indicators of exhaust
2 contributions today because tetraethyl lead is no longer used as a fuel additive. Enrichments
3 in species from clutch and brake wear were absent in the Phoenix paved road dust profiles.
4 These are often composed of asbestos and/or semi-metal carbon composites. Ondov (1974)
5 measured abundances of ~ 14% Mg, -2% Ca, ~4% Fe, and -1% Ba in asbestos brake
6 shoes, while Anderson et al. (1973) reported Si abundances of ~ 10%. Cooper et al. (1987)
7 examined the elemental composition of semi-metal brake shoes and found abundances of
8 -45% Fe, -2% Cu, -0.5% Sn, -3% Ba, and -0.5% Mo. None of these species was
9 found in the Phoenix paved road dust profiles at levels significantly in excess of their
10 abundances in other geological sub-types.
11 The motor vehicle exhaust profile contains high concentrations of organic and elemental
12 carbon; but their ratios are much different from those found in wood combustion with the
13 abundance of elemental carbon being nearly equal to the organic carbon abundance. Bromine
14 and lead are also much larger components in vehicle exhaust than in other source profiles.
15 However, Br and Pb have been phased out of most U.S. gasolines, and this example does not
16 represent current motor vehicle emissions profiles found in the United States today.
17 Pierson and Brachaczek (1976; 1983) pioneered the chemical characterization of
18 paniculate emissions from motor vehicles. These results, acquired from 1970 through 1981,
19 were used extensively in the early days of CMB modeling of total suspended particulate
20 matter (e.g., Watson, 1979; Kowalczyk et al., 1978). Little additional work on the chemical
21 characterization of particulate motor vehicle emissions was conducted until promulgation of
22 the PM10 NAAQS in 1987. It was soon recognized that fleet composition, emissions
23 controls, fuels, engine designs, and vehicle maintenance had changed considerably since the
24 tests of Pierson and Brachaczek (1976; 1983), and that new tests, and types of testing, were
25 needed to obtain chemical source profiles.
26 Watson et al. (1988b) obtained six roadside samples under a freeway overpass, in a city
27 bus yard, and near busy intersections in Reno, NV, in 1986 and 1987 as part of the State of
28 Nevada Air Pollution Study (SNAPS) (Chow et al., 1988). Cooper et al. (1987) and NBA
29 (1990a; 1990b; 1990c), in studies conducted for SCAQMD (Gray et al., 1988; Zeldin et al.,
30 1990), measured exhaust from eleven unleaded-gasoline-fueled vehicles, 3 leaded-gasoline-
31 fueled vehicles, and 2 heavy-duty diesel-fueled trucks operating on laboratory dynamometers
April 1995 5.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 in 1986. The FTP and a steady-state test at 35 mph were followed for the gasoline-fueled
2 vehicle tests. Diesel-fueled trucks were tested under modified FTP and steady-state
3 conditions. Cooper et al. (1987) also took three roadside samples in a tunnel under the Los
4 Angeles Airport.
5 Watson et al. (1990b; 1990c) report the results from dynamometer tests of eight
6 unleaded-gasoline-fueled vehicles, three leaded-gasoline-fueled vehicles, and three light- to
7 medium-duty diesel-fueled vehicles conducted in 1988 during a Denver visibility study. The
8 FTP was applied to vehicles under conditions (temperatures < 40 °F) similar to those found
9 in wintertime Denver. In 1987, Houck et al. (1989) took three samples of heavy-duty diesel-
10 fueled truck exhaust at a roof monitor over the Wheeler weigh station near Bakersfield, CA.
11 These measurements were used with SCAQMD source profiles for PM10 source
12 apportionment in California's San Joaquin Valley (Chow et al., 1992; 1993d). The TOR
13 method described by Chow et al. (1993b) was applied in all of these tests except that of
14 Cooper et al. (1987). The Cooper et al. (1987) carbon analysis monitored light transmission
15 instead of reflectance to implement the pyrolysis correction (Gary, 1990). Watson et al.
16 (1994) report profiles derived in Phoenix, AZ during 1989.
17 There are significant similarities and differences among the chemical compositions of
18 these different motor vehicle profiles measured in different areas, at different times, and by
19 different methods. The Denver diesel-fueled profiles have a much higher abundance of
20 elemental carbon (74+21 %) than the SCAQMD (52+5%), Wheeler Station (43 +8%), or the
21 PHDIES (33±8%) profiles. The organic carbon in these profiles is 23±8% for Denver,
22 36±3% for SCAQMD, 49 + 13% for Wheeler Station, and 40±7% for PHDIES.
23 For leaded-gasoline-fueled vehicles, the OC and EC abundances were found to be
24 67±23% and 16±7% in the Denver tests, 52±4% and 1.3 + 1% in the SCAQMD steady-
25 state tests, and 31+20% and 15±2% in the SCAQMD FTP tests, respectively. For
26 unleaded-gasoline-fueled vehicles, the OC and EC abundances were found to be 76+29%
27 and 18±11% for Denver tests, 93 ±52% and 5 ±7% for SCAQMD steady-state tests, and
28 49 + 10% and 39+8% for SCAQMD FTP tests, respectively. These compare to OC and EC
29 abundances in PHAUTO of 30 + 12% and 14±8%, respectively.
30 In roadside and tunnel tests which included mixtures of diesel-, leaded-, and unleaded-
31 fueled vehicles, Watson et al. (1988a;1990c) found abundances of 50 + 24% OC and
April 1995 5-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 28 ± 19% EC, and Cooper et al. (1987) found abundances of 38 ± 6% OC and 38 + 5%
2 EC. These are similar to the 39 + 19% OC and 36 + 11% EC abundances found in this study.
3 All of these profiles represent emissions from a small number of vehicles relative to the
4 entire vehicle population. In each of these studies, as in this one, there were individual
5 samples which deviated in their compositions from the majority of profiles. Further study is
6 needed to determine why these deviations exist and how they are related to the mixture of
7 vehicles tested. For wood burning in fireplaces and stoves, organic carbon is by far the most
8 abundant constituent, followed by elemental carbon. Chlorine and potassium also approach
9 an abundance of 1% in these emissions, as do sulfate and nitrate. The water soluble
10 potassium equals total potassium in wood burning emissions.
11 The coal-fired power plant profile in Figure 5-12 has several chemical abundances that
12 are similar to those of the road dust profile. The sulfate level is much higher in these
13 emissions than in the other profiles, and the organic and elemental carbon fractions are much
14 lower. Selenium is clearly detectable in the power plant profile, whereas it is below
15 quantifiable limits in the other profiles.
16 The chemicals identified in Figures 5-9 through 5-12 represent the major components
17 that contribute to atmospheric light extinction (i.e., crustal, sulfate, nitrate, ammonium,
18 organic carbon, elemental carbon) as well as chemical patterns that help to distinguish one
19 source of primary emissions from others (e.g., selenium, potassium, aluminum, silicon,
20 organic carbon, elemental carbon). In addition to these commonly measured components, it
21 is possible that isotopic ratios in source emissions may vary in an informative way with the
22 nature of the combustion process and with the geologic age and character of the source input
23 material. Carbon-14, for example, has been widely used to separate contemporary carbon
24 due to vegetative burning from carbon emitted by fossil fuel combustion (Currie et al.,
25 1984). Certain isotopic ratios might also distinguish coal-fired power plant emissions from
26 other sources. Fuels from different mines may have isotopic differences that allow otherwise
27 similar source emissions from different power plants to be distinguished.
28
29
30
April 1995 5_41 DRAFT-DO NOT QUOTE OR CITE
-------
1 5.8 EMISSIONS MODELS AND EMISSIONS INVENTORIES
2 As noted above, major discrepancies exist between relative amounts of emissions and
3 contributions to suspended particles found in many areas. Major re-design is needed to
4 create more accurate emissions models to improve quantification of source-receptor
5 relationships.
6 Emissions models are intended to estimate the emissions rates as a function of space
7 and time of selected pollutants from point, area, and mobile sources. In contrast to an
8 emissions inventory, which is a static catalogue of emissions estimates for a given
9 geographical area and averaging time, an emissions model is capable of accessing activity
10 data bases from a multitude of information- gathering agencies and determining actual
11 emissions for relatively small regions and averaging times.
12 Emissions models have, at their base, several activity surrogates that are considered to
13 have some relationship to emissions. Activities relevant to the major source contributions
14 identified in Table 5-1 include: (1) number of acres burned, locations, and durations of
15 wildfires and prescribed burns; (2) cords of wood and tons of coal sold for residential
16 heating; (3) vehicle miles traveled for mobile source emissions; (4) heads of livestock,
17 pounds of fertilizer, and number sewage treated for ammonia; and (5) amount of product
18 produced for industrial sources (e.g. power generated for power plant emissions; tons of
19 coke produced for a coke oven). Wherever possible, different activity data for the same
20 sources and emitted species are used to evaluate uncertainties. For example, Watson et al..
21 (1990) showed for Denver, CO, that proportional differences between gasoline and diesel
22 powered mobile source activities might be separately estimated from vehicle counts at
23 selected roadways, areawide fuel sales, and vehicle registrations. Differences between
24 separate activity estimates can be used to quantify the uncertainty of emissions rates.
25 Upon these activity data are imposed emissions/activity relationships (commonly termed
26 "emissions factors") that may require meteorological inputs such as temperature, relative
27 humidity, or wind speed. These relationships are easily replaceable and include the ability to
28 propagate the precisions specified for the input data. The emissions/activity relationships are
29 specific for different source sub-types. As an example, certain roads may have greater or
30 lesser proportions of diesel truck traffic, older and newer vehicles, and vehicles which were
31 recently started and ones which are fully warmed up. In the idealized emissions model, each
April 1995 5-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 road segment is assigned a classification for a specific distribution of vehicles, and a separate
2 emissions/activity relationship is determined for each classification.
3 Emissions models should have several attributes, i.e., they should: (1) be based on
4 documented scientific and engineering principles; (2) be composed of special purpose
5 modules which can be updated with new information and new science when available;
6 (3) have activity levels and emissions/activity relationships specific to a non-attainment area;
7 (4) contain error propagation algorithms to provide precision estimates on outputs; (5) use
8 independent activity data bases and emissions/activity relationships of equivalent quality to
9 estimate accuracy; (6) adjust the emissions/activity relationship in response to environmental
10 variables, especially meteorology; (7) allow the addition, subtraction, or modification of
11 emissions for special events; (8) retain traceability of all information to allow quality
12 auditing; (9) provide output displays, statistics, and data bases which can be used for
13 modeling, data analysis, control strategy development, and quality assurance; and
14 (10) calculate the effects of changes, such as population growth, implementation of PM10,
15 control measures, and changes in land use.
16 Unfortunately, there are no emissions models used or even currently available that
17 attain all these attributes. Attempts have been made to improve the current state of the art
18 with creation of the Flexible Response Emissions Data System (FREDS) (Lebowitz et al.,
19 1987) for the National Acid Precipitation Assessment Program (NAPAP) and the Emissions
20 Preprocessor System (EPS) for the Urban Airshed Model (SAI, 1990). The closest approach
21 is the Geographical Emissions Modeling and Assessment Program (GEMAP; Dickson and
22 Oliver, 1993) developed for the San Joaquin Valley Air Quality Study and Atmospheric
23 Utility Signatures, Predictions and Experiments (SJVAQS/AUSPEX) in California (Solomon,
24 1994).
25 Assumptions inherent in using a static emissions inventory to represent short-term
26 events are: (1) emissions rates are constant, typically averaged over a year and sometimes
27 over a season; (2) emissions factors relating activities to emissions apply to all emitters at all
28 locations at all times; (3) the values of activity variables are highly correlated with actual
29 emissions; and (4) all major emitters have been identified and included.
30 It is not difficult to find major deviations from these assumptions, even in a small area
31 over a short time period. As might be expected averaging over large spatial scales and time
April 1995 5.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 periods evens out much of the variability inherent in applying these assumptions to individual
2 episodes and locations. However, short-term episodes of a few hours or a few days duration
3 and source influences over spatial scales of less than 100 km are exactly what is needed to
4 evaluate source contributions to excessive 24-h values of paniculate concentrations in non-
5 attainment areas.
6 Emissions inventories are often used to develop emissions reductions strategies by the
7 process of linear rollback Earth (1970). This approach assumes that pollutant concentrations
8 and the effects of those concentrations (e.g., health effects) are directly proportional to
9 emissions within a selected geographical region. For this approach, an area is defined that
10 receives emissions from the sources to be controlled and ambient concentrations of the
11 pollutants of interested are measured within and outside this area. If there are contributions
12 from the suspected sources, the within-area concentrations will be significantly higher than
13 the outside-area concentrations, and the outside-area concentrations are subtracted to obtain
14 the incremental amount contributed by in-area sources. Emissions of the pollutant or the
15 precursor within the study area are reduced by an amount proportional to the desired
16 reduction in ambient concentrations.
17 The linear rollback method is simple and its application in urban areas has resulted in
18 reduced ambient concentrations for primary particles when emissions estimates are accurate,
19 spatial scales are large, and averaging times are long. It is currently being tested for U.S.
20 utility sulfur dioxide emissions that are to be reduced to 8.95 million tons per year (by more
21 than 50% from current emissions) by the year 2002. The rollback method, however, has not
22 been proven to be accurate for short duration events over small spatial scales for secondary
23 aerosol when emissions estimates are uncertain.
24
25
26 5.9 SUMMARY AND CONCLUSIONS
27 The ambient atmosphere contains both primary and secondary particles; the former are
28 emitted directly by sources, and the latter are formed from gases (SO2, NOX, HN4, VOCs)
29 that are directly emitted by sources. Fugitive dust is a primary pollutant, and also has a role
30 in secondary particle formation. Major sources of particle emissions are classified as major
April 1995 5-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 point sources, mobile sources, and area sources; these are anthropogenic. Natural sources
2 also contribute to ambient concentrations.
3 The 1982 Criteria Document emphasized emissions from industrial sources, especially
4 primary particles. SO2 was the only precursor of secondary particles considered. Since
5 1982, many of these sources have been controlled, yet particle standards are exceeded in
6 many areas.
7 Source and receptor models are used to quantify major contributions to excess PM10
8 concentrations. Source models use emissions inventories and meteorological data to product
9 particle dispersion and PM10 concentrations measured at receptors. Receptor models use
10 chemical composition of emissions and receptor concentrations to estimate the contribution of
11 sources. The latter are more appropriate to identify sources in non-contaminant areas.
12 Fugitive dust is a major contribution to PM10 at nearly all sampling site, although the
13 average fugitive dust source contribution is highly variable among sampling sites within the
14 same areas, and is highly variable between seasons.
15 Primary motor vehicle exhaust makes up as much as 40% of average PM10 at many
16 sampling sites. Vegetative burning outdoor and residential wood burning are significant
17 sources in residential areas. Fugitive dust from paved and unpaved roads, agricultural
18 operations, construction, and soil erosion constitute -90% of nationwide primary emissions.
19 All of the emissions have remained relatively constant over the 8-year period except for those
20 from soil erosion.
21 The majority of wind erosion occurs in the dustbowl region; estimates are influenced by
22 annual precipitation and wind-speed distribution. The major non-fugitive dust emitters are
23 other industrial processes and exhaust from highway vehicles. Fuel combustion from
24 utilities, industrial, and other sources together contribute between 1 to 2% to total primary
25 particle emissions. Industrial fuel combustion emissions were reduced by one-third and other
26 fuel combustion emissions were reduced by one half between 1983 and 1992. On high-way
27 vehicle emissions increased by 50%, primarily due to large increases in the number of
28 vehicle miles traveled. Electric utilities account for the largest fraction of sulfur dioxide,
29 nearly 70% of total emissions. These emissions have not changed substantially over the 10
30 years reported. Annual averages do not reflect the seasonality of certain emissions,
31 residential wood burning in fireplaces and stoves, for example. Cold weather also affects
April 1995 5.45 DRAFT-DO NOT QUOTE OR CITE
-------
1 motor vehicle exhaust emissions, both in terms of chemical composition and emission rates.
2 Planting, harvesting, and fertilizing and harvesting are also seasonal. Fugitive dust consists
3 of geological material that is suspended into the atmosphere by natural wind and by
4 anthropogenic activities from sources such as paved and unpaved roads, construction and
5 demolition of buildings and roads, storage piles, wind erosion, and agricultural tilling. There
6 are obvious discrepancies between the proportion of fugitive dust in primary emissions and
7 geological contributions to PM10 calculated by recptor models, due to contributions from
8 secondary aerosols, which are not included in the primary PM10 emission estimates. Even
9 when secondary aerosol is subtracted, however, other sources such as vegetative burning and
10 wood combustion make larger relative contributions to ambient concentrations than is
11 indicated by the emissions inventories. Fugitive dust estimates are especially affected by the
12 general limitations of emissions inventories. Annual and spatial averages do not reflect the
13 seasonality of certain emissions. Planting and harvesting are seasonal, as are high winds
14 which might cause erosion. Paved road dust emissions might be much higher after rain
15 storms when dirt is tracked from unpaved areas onto paved roads. The spatial disaggregation
16 of fugitive dust emissions is poorer than that for all other source categories. Whereas most
17 mobile sources are confined to established roadways, and most area sources correspond to
18 population density, suspendable dust is everywhere. Modern data bases, computer systems,
19 and information management software could be applied to improve existing inventories
20 without major additional costs after the initial investment in establishing an inventory
21 methodology. Mobile sources are major emitters of primary particles, oxides of nitrogen,
22 and volatile organic compounds. They are also minor emitters of sulfur dioxide and
23 ammonia. On-road motor vehicles using gasoline- and diesel-fueled engines are by far the
24 largest component of mobile source emissions, and the emissions estimation methods are
25 most highly developed for these vehicles. Studies show that while vehicle emissions models
26 may function well under idealized conditions, they underestimate the effects of high emitting
27 vehicles that may be major sources of VOCs. Motor vehicle exhaust contains high
28 concentrations of organic and elemental carbon, but their ratios are much different from
29 those found in wood combustion with the abundance of elemental carbon being nearly equal
30 to the organic carbon abundance. There are major discrepancies between the relative
31 amounts of emissions and contributions to suspended particles found in many areas. Some
April 1995 5-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 major re-design is needed to create more accurate emissions models that can improve the
2 quantification of source-receptor relationships. Emissions models are intended to estimate
3 the emissions rates as a function of space and time of selected pollutants from point, area,
4 and mobile sources. In contrast to an emissions inventory, which is a static catalogue of
5 emissions estimates for a given geographical area and averaging time, an emissions model is
6 capable of accessing activity data bases from a multitude of information- gathering agencies
7 and determining actual emissions for relatively small regions and averaging times.
8
9
April 1995 5.47 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 American Society for Testing and Materials. (1990a) Standard test method for amount of material in soils finer
4 than the no. 200. In: Annual book of ASTM standards. Philadelphia, PA: American Society for Testing
5 and Materials; pp. 193-194; document no. D1140.
6
7 American Society for Testing and Materials. (1990b) Standard test method for particle-size analysis of soils. In:
8 Annual book of ASTM standards. Philadelphia, PA: American Society for Testing Materials; pp. 94-100;
9 document no. D422.
10
11 Anderson, A. E.; Gealer, R. L.; McCune, R. C.; Sprys, J. W. (1973) Asbestos emissions from brake
12 dynamometer tests. Warrendale, MI: Society of Automotive Engineers; document SAE 7305-49.
13
14 Ashbaugh, L.; Lawson, D. R. (1991) A comparison of emissions from mobile sources using random roadside
15 surveys conducted in 1985, 1987, and 1989. Presented at: the 84th annual meeting of the Air & Waste
16 Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air & Waste Management
17 Association.
18
19 Ashbaugh, L. L.; Lawson, D. R.; Bishop, G. A.; Guenther, P. L.; Stedman, D. H.; Stephens, R. D.; Groblicki,
20 P. J.; Parikh, J. S.; Johnson, B. J.; Huang, S. C. (1992) On-road remote sensing of carbon monoxide
21 and hydrocarbon emissions during several vehicle operating conditions. In: Chow, J. C.; Ono, D. M.,
22 eds. PM10 standards and nontraditional paniculate source controls, v. II. Pittsburgh, PA: Air & Waste
23 Management Association; pp. 885-898.
24
25 Axetell, K.; Zell, J. (1977) Control of reentrained dust from paved streets. Kansas City, MO: U.S.
26 Environmental Protection Agency, Region VII; EPA report no. EPA-907/9-77-007. Available from:
27 NTIS, Springfield, VA; PB-280 325.
28
29 Bagnold, R. A. (1937) The transport of sand by wind. Geogr. J. 89: 409-438.
30
31 Bagnold, R. A. (1941) The physics of blown sand and desert dunes. London, United Kingdom: Methuen & Co.
32
33 Barnard, W. R.; Stewart, M. A. (1992) Modification of the 1985 NAPAP wind erosion methodology to assess
34 annual PM10 emissions for EPA's emissions trends report. In: Chow, J. C.; Ono, D. M., eds. PM10
35 standards and nontraditional paniculate source controls. Pittsburgh, PA: Air and Waste Management
36 Association; pp. 121-130.
37
38 Barnard, W. R.; Carlson, P. M.; Stewart, M. A. (1992) Fugitive PM10 emissions estimates developed for EPA's
39 emission trends report. In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional paniculate
40 source controls. Pittsburgh, PA: Air & Waste Management Association; pp. 110-120.
41
42 Earth, D. S. (1970) Federal motor vehicle emission goals for CO, HC and NOX based on desired air quality
43 levels. J. Air Pollut. Control Assoc. 20: 519-523.
44
45 Belly, P. Y. (1964) Sand movement by wind. U.S. Army Coastal Engineering Research Center; pp. 1-38;
46 technical memo 1.
47
48 Bisal, F.; Hsieh, J. (1966) Influence of moisture on erodibility of soil by wind. J. Soil Sci. 102: 143-146.
49
50 California Air Resources Board. (1993) EMFAC7F workshop jointly sponsored by the Technical Support
51 Division and the Mobile Source Division; June; Sacramento, CA. Sacramento, CA: California Air
52 Resources Board.
53
April 1995 5-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gary, R. (1990) Sunset Laboratory thermal-optical analysis for organic/elemental carbon analysis—analysis
2 description. Forest Grove, OR: Sunset Laboratory.
3
4 Chepil, W. S. (1952) Improved rotary sieve for measuring state and stability of dry soil structure. Soil Sci. Soc.
5 Am. Proc. 16: 1113-1117.
6
7 Chepil, W. S. (1956) Influence of moisture on erodibility of soil by wind. Soil Sci. Soc. Am. Proc. 20: 288-292.
8
9 Chepil, W. S.; Woodruff, N. P. (1963) The physics of wind erosion and its control. Adv. Agron. 15: 211-299.
10
11 Chow, J. C.; Watson, J. G. (1994) Contemporary source profiles for geological material and motor vehicle
12 emissions [final report]. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of
13 Air Quality Planning and Standards.
14
15 Chow, J. C.; Watson, J. G.; Frazier, C. A.; Egami, R. T.; Goodrich, A.; Ralph, C. (1988) Spatial and
16 temporal source contributions to PM10 and PM2 5 in Reno, NV. In: Mathai, C. V.; Stonefield, D. F.,
17 eds. PMi0: implementation of standards. Pittsburgh, PA: Air Pollution Control Association; pp. 438-457.
18
19 Chow, J. C.; Watson, J. G.; Richards, L. W.; Haase, D. L.; McDade, C.; Dietrich, D. L.; Moon, D.; Sloane,
20 C. (1991) The 1989-90 Phoenix PM10 study, v. II: source apportionment [final report]. Phoenix,
21 AZ: Arizona Department of Environmental Quality; report no. 8931.6F1.
22
23 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Frazier, C.; Hinsvark, B.; Prichett, L.; Neuroth, G. (1992a)
24 Wintertime PM10 and PM2 5 mass and chemical compositions in Tucson, Arizona. In: Chow, J. C.; Ono,
25 D. M., eds. Transactions: PM10 standards and nontraditional paniculate source controls. Pittsburgh,
26 PA: Air and Waste Management; pp. 231-243.
27
28 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L.
29 W. (1992b) PM10 source apportionment in California's San Joaquin Valley. Atmos. Environ. Part A
30 26: 3335-3354.
31
32 Chow, J. C.; Liu, C. S.; Cassmassi, J.; Watson, J. G.; Lu, Z.; Pritchett, L. C. (1992c) A neighborhood-scale
33 study of PM10 source contributions in Rubidoux, California. Atmos. Environ. Part A 26: 693-706.
34
35 Chow, J. C.; Watson, J. G.; Ono, D. M.; Mathai, C. V. (1993) PM10 standards and nontraditional paniculate
36 source controls: a summary of the A&WMA/EPA international specialty conference. Air Waste
37 43: 74-84.
38
39 Chow, J. C.; Watson, J. G.; Houck, J. E.; Pritchett, L. C.; Rogers, C. P.; Frazier, C. A.; Egami, R. T.; Ball,
40 B. M. (1994) A laboratory resuspension chamber to measure fugitive dust size distributions and chemical
41 compositions. Atmos. Environ. 28: 3463-3481.
42
43 Cooper, J. A.; Redline, D. C.; Sherman, J. R.; Valdovinos, L. M.; Pollard, W. L.; Scavone, L. C.;
44 Badgett-West, C. (1987) PM10 source composition library for the South Coast Air Basin: volume I,
45 source profile development documentation final report. El Monte, CA: South Coast Air Quality
46 Management District; July 15.
47
48 Cooper, J. A.; Sherman, J. R,; Miller, E.; Redline, D.; Valdonovinos, L.; Pollard, W. (1988) CMB source
49 apportionment of PM10 of an oil-fired power plant in Chula Vista, California. In: Mathai, C. V.;
50 Stonefield, D. H., eds. PM10: implementation of standards. Pittsburgh, PA: Air & Waste Management
51 Association; pp. 495-507.
52
53 Federal Register. (1991) Designations and classifications for initial PMIO nonattainment areas: 40 CFR Part 81.
54 F. R. 56: 11101.
April 1995 5.49 DRAFT-DO NOT QUOTE OR CITE
-------
1 Federal Register. (1993) Reclassification of moderate PM10 nonattainment areas to serious areas: 40 CFR Part
2 81. F. R. 58: 3334.
3
4 Federal Register. (1994) Designations of areas for air quality planning purposes: 40 CFR Part 81. F. R.
5 58: 67334.
6
7 Fryrear, D. (1992) Measured wind erosion of agricultural lands. In: Chow, J. C.; Ono, D. M., eds. PM10
8 standards and nontraditional paniculate source controls. Pittsburgh, PA: Air & Waste Management
9 Association; pp. 433-439.
10
11 Gertler, A. W.; Fujita, E. M.; Pirson, W. R.; Wittorff, D. N. (1995) Apportionment of NMHC tailpipe vs. non-
12 tailpipe emissions in the Fort McHenry and Tuscarora tunnels. Atmos. Environ.: accepted.
13
14 Gillette, D. (1980) Major contributions of natural primary continental aerosols: source mechanisms. In: Kniep,
15 T. J.; Lioy, P. J., eds. Aerosols: anthropogenic and natural, sources and transport; January 1979; New
16 York, NY. Ann N. Y. Acad. Sci. 338: 348-358.
17
18 Gillette, D. A.; Hanson, K. J. (1989) Spatial and temporal variability of dust production caused by wind erosion
19 in the United States. J. Geophys. Res. 94: 2197-2206.
20
21 Gillette, D. A.; Stockton, P. H. (1989) The effect of nonerodible particles on wind erosion of erodibile surfaces.
22 J. Geophys. Res. [Atmos.] 94: 12885-12893.
23
24 Gillette, D. A.; Adams, J.; Endo, A.; Smith, D.; Kihl, R. (1980) Threshold velocities for input of soil particles
25 into the air by desert soils. J. Geophys. Res. 85: 5621-5630.
26
27 Granat, L.; Rodhe, H.; Hallberg, R. O. (1976) The global sulphur cycle. In: Svensson, B. H.; Soderlund, R.,
28 eds. Nitrogen, phosphorus and sulphur: global cycles. Ecol. Bull. 22: 89-134. (SCOPE report 7).
29
30 Grosjean, D.; Seinfeld, J. H. (1989) Parameterization of the formation potential of secondary organic aerosols.
31 Atmos. Environ. 23: 1733-1747.
32
33 Hansen, A. D. A.; Rosen, H. (1990) Individual measurements of the emission factor of aerosol black carbon in
34 automobile plumes. J. Air Waste Manage. Assoc. 40: 1654-1657.
35
36 Hildemann, L. M.; Markowski, G. R.; Jones, M. C.; Cass, G. R. (1991) Submicrometer aerosol mass
37 distributions of emissions from boilers, fireplaces, automobiles, diesel trucks, and meat-cooking
38 operations. Aerosol Sci. Technol. 14: 138.
39
40 Houck, ]. E.; Chow, J. C.; Watson, J. G.; Simons, C. A.; Pritchett, L. C.; Goulet, J. M.; Frazier, C. A.
41 (1989) Determination of particle size distribution and chemical composition of particulate matter from
42 selected sources in California. Sacramento, CA; California Air Resources Board.
43
44 Houck, J. E.; Goulet, J. M.; Chow, J. C.; Watson, J. G.; Pritchett, L. C. (1990) Chemical characterization of
45 emission sources contributing to light extinction. In: Mathai, C. V., ed. Visibility and fine particles.
46 Pittsburgh, PA: Air & Waste Management Association; pp. 437-446.
47
48 Ingalls, M. N.; Smith, L. R.; Kirksey, R. E. (1989) Measurement of on-road vehicle emission factors in the
49 California South Coast Air Basin: volume I, regulated emissions. Atlanta, GA: Coordinating Research
50 Council, Inc.; report no. CRC-APRAC-AP-4-SCAQS-l. Available from: NTIS, Springfield, VA; PB89-
51 220925.
52
53 Kinsey, J. S.; Cowherd, C. (1992) Fugitive emissions. In: Buonicore, A. J.; Davis, W. T., eds. Air pollution
54 engineering manual. Pittsburgh, PA: Air & Waste Management Association; pp. 133-146.
April 1995 5-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 Knapp, K. T. (1992) Dynamometer testing of on-road vehicles from the Los Angeles in-use emissions study.
2 In: Chow, J. C.; Ono, D. M., eds. PM10 standards and nontraditional particulate source controls, v. II.
3 Pittsburgh, PA: Air & Waste Management Association; pp. 871-884.
4
5 Kowalczyk, G. S.; Choquette, C. E.; Gordon, G. E. (1978) Chemical element balances and identification of air
6 pollution sources in Washington, D.C. Atmos. Environ. 12: 1143.
7
8 Lamb, B.; Gay, D.; Westberg, H.; Pierce, T. (1993) A biogenic hydrocarbon emission inventory for the U.S.A.
9 using a simple forest canopy model. Atmos. Environ. Part A 27: 1673-1690.
10
11 Landry, B.; Liu, C. S.; Henry, R. C.; Cooper, J. A.; Sherman, J. R. (1988) Receptor modeling for PM10
12 source apportionment in the south coast air basin of California. In: Mathai, C. V.; Stonefield, D. H.,
13 eds. PM10: implementation of standards. Pittsburgh, PA: Air Pollution Control Association; pp. 399-418.
14
15 Lebowitz, L. G.; Ackerman, A. S. (1987) Flexible regional emissions data system (FREDS) documentation for
16 the 1980 NAPAP (National Acid Precipitation Assessment Program) emissions inventory. Research
17 Triangle Park, NC: U.S. Environmental Protection Agency, Air and Energy Engineering Research
18 Laboratory; EPA report no. EPA/600/7-87/025A. Available from: NTIS, Springfield, VA; PB88-129499.
19
20 Magliano, K. L. (1988) Level I PM10 assessment of a California air basin. In: Mathai, C. V.; Stonefield, D. H.,
21 eds. Transactions, PMi0: implementation standards. Pittsburgh, PA: Air and Waste Management
22 Association,; pp. 508-517.
23
24 Mollinger, A. M.; Nieuwstadt, F. T. M.; Scarlett, B. (1993) Model experiments of the resuspension caused by
25 road traffic. Aerosol Sci. Technol. 19: 330-338.
26
27 Muleski, G.; Stevens, K. (1992) PM10 emissions from public unpaved roads in rural Arizona. In: Chow, J. C.;
28 Ono, D. M., eds. Transactions, PM10 standards and nontraditional particulate source controls. Pittsburgh,
29 PA: Air and Waste Management Association; pp. 324-334.
30
31 NBA. (1990) Enhancement of the south coast air basin source profile library for chemical mass balance receptor
32 model application. Beaverton, OR: NBA.
33
34 Nicholson, K. W.; Branson, J. R.; Giess, P.; Cannell, R. J. (1989) The effects of vehicle activity on particle
35 resuspension. J. Aerosol Sci. 20: 1425-1428.
36
37 Ondov, J. M. (1974) A study of trace elements on particles from motor vehicles. College Park, MD: University
38 of Maryland.
39
40 Pace, T. G. (1987) Protocol for applying and validating the CMB model. Research Triangle Park, NC: U.S.
41 Environmental Protection Agency; report no. EPA-450/4-87-010.
42
43 Parmar, S. S.; Grosjean, D. (1990) Laboratory tests of KI and alkaline annular denuders. Atmos. Environ. Part
44 A 24: 2695-2698.
45
46 Pierson, W. R.; Brachaczek, W. W. (1974) Airborne particulate debris from rubber tires. Rubber Chem.
47 Technol. 47: 1275.
48
49 Pierson, W. R.; Brachaczek, W. W. (1976) Particulate matter associated with vehicles on the road. Presented at:
50 the automotive engineering congress and exposition; February; Detroit, MI. Warrendale, PA: Society of
51 Automotive Engineers; SAE technical paper no. 760039.
52
53 Pierson, W. R.; Brachaczek, W. W. (1983) Particulate matter associated with vehicles on the road. II. Aerosol
54 Sci. Technol. 2: 1-40.
April 1995 5.51 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pierson, W. R.; Gertler, A. W.; Bradow, R. L. (1990) Comparison of the SCAQS tunnel study with other on-
2 road vehicle emission data. J. Air Waste Manage. Assoc. 40: 1495-1504.
3
4 Pye, K. (1987) Aeolian dust and dust deposits. London, United Kingdom: Academic Press.
5
6 Reeser, W. K.; Zimmer, R. A.; Cummins, P.; Briggs, K. R. (1992) An episodic PM10 emissions inventory for
7 Denver, Colorado. In: Chow, J. C.; Ono, D. M., eds. Transactions, PM10 standards and nontraditional
8 paniculate source controls. Pittsburgh, PA: Air and Waste Management Association; pp. 131-145.
9
10 Robinson, E.; Robbins, R. C. (1971) Emissions, concentrations, and fate of paniculate atmospheric pollutants.
11 Final report. Washington, DC: American Petroleum Institute; API publication no. 4076.
12
13 Ryan, W. M.; Badgett-West, C. R.; Cooper, J. A.; Ono, D. M. (1988) Reconciliation of the receptor and
14 dispersion modeling impacts of PM10 in Hayden, AZ. In: Mathai, C. V.; Stonefield, D. H., eds.
15 Transactions, PM10: implementation of standards. Pittsburgh, PA: Air and Waste Management
16 Association; pp. 419-429.
17
18 Seinfeld, J. H. (1986) Air pollution: physical and chemical fundamentals. 2nd ed. New York, NY: McGraw Hill.
19
20 Solomon, P. A. (1994) Planning and managing air quality modeling and measurements studies: a perspective
21 through SJVAQS/AUSPEX. Chelsea, MI: Lewis Pubishers; San Ramon, CA: Pacific Gas and Electric
22 Company; pp. 369-382.
23
24 South Coast Air Quality Management District. (1994) State implementation plan for PM10 in the Coechella
25 Valley: 1994 "BACM" revision. Diamond Bar, CA: South Coast Air Quality Management District.
26
27 Stedman, D. J.; Bishop, G. A.; Beaton, S. P.; Peterson, J. E.; Guenther, P. L.; McVey, I. F.; Zhang, Y.
28 (1994) On-road remote sensing of CO and HC emissions in California. Sacramento, CA: California Air
29 Resources Board; final report under contract A032-093.
30
31 Svasek, T. N.; Terwindt, J. H. (1974) Measurement of sand transport by wind on a natural beach.
32 Sedimentology 21:311 -322.
33
34 Systems Applications, Inc. (1990) User's guide for the urban airshed model volume IV: user's guide for the
35 emissions preprocessor system. San Rafael, CA: Systems Applications, Inc.
36
37 Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM10 samplers
38 and source apportionment of ambient PM10 concentrations in Rillito, Arizona. In: Chow, J. C.; Ono, D.
39 M., eds. Transactions, PM10 standards and nontraditional particulate source controls. Pittsburgh, PA: Air
40 and Waste Management Association; pp. 244-261.
41
42 Thornwaite, C. W. (1931) the climates of North America: according to a new classification. Geophys. Rev.
43 21:633-655.
44
45 U.S. Environmental Protection Agency. (1980a) National air pollutant emission estimates, 1970-1978. Research
46 Triangle Park, NC: Office of Air Quality Planning and Standards; EPA report no. EPA-450/4-80-002.
47 Available from: NTIS, Springfield, VA; PB82-232687.
48
49 U.S. Environmental Protection Agency. (1980b) 1977 National emissions report: national emissions data system
50 of the Aerometric and Emissions Reporting System. Research Triangle Park, NC: Office of Air Quality
51 Planning and Standards; EPA report no. EPA-450/4-80-005. Available from: NTIS, Springfield, VA;
52 PB80-222532.
53
April 1995 5-52 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Environmental Protection Agency. (1982) Air quality criteria for paniculate matter and sulfur oxides.
2 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
3 and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
4 VA; PB84-156777.
5
6 U.S. Environmental Protection Agency. (1987a) PM10 SIP development guideline. Research Triangle Park, NC:
7 U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; EPA report no.
8 EPA-450/2-86-001. Available from: NTIS, Springfield, VA; PB87-206488.
9
10 U.S. Environmental Protection Agency. (1987b) Protocol for reconciling differences among receptor and
11 dispersion models. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air
12 Quality Planning and Standards; EPA report no. EPA-450/4-87-008. Available from: NTIS, Springfield,
13 VA; PB87-206504.
14
15 U.S. Environmental Protection Agency. (1988) Compilation of air pollutant emission factors. Volume I:
16 stationary point and area sources. Research Triangle Park, NC: Office of Air Quality Planning and
17 Standards, Office of Air and Radiation.
18
19 Watson, J. G., Jr. (1979) Chemical element balance receptor methodology for assessing the sources of fine and
20 total suspended paniculate matter in Portland, Oregon [Ph.D. thesis]. Beaverton, OR: Oregon Graduate
21 Center.
22
23 Watson, J. G.; Chow, J. C.; Richards, L. W.; Neff, W. D.; Andersen, S. R.; Dietrich, D. L.; Houck, J. E.;
24 Olnez, I. (1988a) The 1987-88 metro Denver brown cloud study: v. I, II, III. Reno, NV: Desert
25 Research Institute; final report no. 8810.1F(l-3).
26
27 Watson, J. G.; Chow, J. C.; Egami, R. T.; Frazier, C. A.; Goodrich, A.; Ralph, C. (1988b) PM10 source
28 apportionment in Reno and sparks, Nevada for state implemenation plan development. Volume I:
29 modeling methods and results. Reno, NV: Desert Research Institute; document 8086.2F1.
30
31 Watson, J. G.; Chow, J. C.; Mathai, C. V. (1989) Receptor models in air resources management: a summary of
32 the APCA international specialty conference. JAPCA 39: 419-426.
33
34 Watson, J. G.; Chow, J. C.; Pritchett, J. A.; Houch, R. A.; Ragazzi, R. A.; Burns, S. (1990) Chemical source
35 profiles for paniculate motor vehicle exhaust under cold and high altitude operating conditions. Sci. Total
36 Environ. 93: 183-190.
37
38 Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Ragazzi, R. A.; Burns, S. (1990) Composite source
39 profiles for paniculate motor vehicle exhaust source apportionment in Denver, CO. In: Mathai, C. V.,
40 ed. Transaction, visibility and fine particles. Pittsburgh, PA: Air and Waste Management Association;
41 pp. 422-436.
42
43 Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B. M.; Pace, T. G.; Meyer, E. L.; Nguyen,
44 Q. (1990) The USEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ. Software
45 5: 38-49.
46
47 Watson, J. G.; Chow, J. C.; Pace, T. G. (1991) Chemical mass balance. In: Hopke, P. K., ed. Data handling in
48 science and technology: v. 7, receptor modeling for air quality management. New York, NY: Elsevier
49 Press; pp. 83-116.
50
51 Watson, J. G.; Chow, J. C.; Lowenthal, L. C.; Pritchett, C. A.; Frazier, C. A.; Neuroth, G. R.; Robbins, R.
52 (1994) Differences in the carbon compostion of source profiles for diesel- and gasoline-powered vehicles.
53 Atmos. Environ. 28: 2493-2505.
54
April 1995 5.53 DRAFT-DO NOT QUOTE OR CITE
-------
1 Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994)
2 Chemical mass balance source apportionment of PM10 during the Southern California Air Quality Study.
3 Aerosol Sci. Technol. 21: 1-36.
4
5 Watson, J. G.; Chow, J. C.; Lurmann, F. W.; Musarra, S. P. (1994) Ammonium nitrate, nitric acid, and
6 ammonia equilibrium in wintertime Phoenix, Arizona. Air Waste 44: 405-412.
7
8 Watson, J. G.; Chow, J. G.; Roth, P. M. (1994) Clear sky visibility as a challenge for society. Annu. Rev.
9 Energy Environ. 19: 241-266.
10
11 Went, F. W. (1960) Organic matter in the atmosphere, and its possible relation to petroleum formation. Proc.
12 Natl. Acad. Sci. U.S. A. 46: 212-221.
13
14 Yamate, G. (1973) Development of emission factors for estimating atmospheric emissions from forest fires.
15 Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and
16 Standards; EPA report no. EPA-450/3-73-009. Available from: NTIS, Springfield, VA; PB-230 889.
17
18 Zeldin, M.; Kim, B.-M.; Lewis, R.; Marlia, J. C.; Chan, S.; Hogo, H.; Uesugi, K. (1991) PM10 source
19 apportionment for the South Coast Air Basin: final air quality management plan, 1991 revision. Diamond
20 Bar, CA: South Coast Air Quality Management District; final technical report V-F.
April 1995 5-54 DRAFT-DO NOT QUOTE OR CITE
-------
i 6. ENVIRONMENTAL CONCENTRATIONS
2
3
4 6.1 BACKGROUND, PURPOSE AND SCOPE
5 This chapter summarizes the concentrations of particulate matter over the United States,
6 including the spatial, temporal, size and chemical aspects. This chapter is aimed to support
7 the quantification of particulate matter effects and to aid the PM standard setting process.
8 The information needs for assessing the major aerosol effects of concern is summarized in
9 Table 6.1. Depending on the effect, different aspects (dimensions) of aerosol concentrations
10 are important. The effects on human health are considered most serious and this chapter is
11 to provide relevant aerosol concentration information to help in quantifying these effects.
12 Concern also exists for aerosol effects on visibility as well as damage to manmade materials.
13 Health effects are concerned with people, and the geographic areas of importance are
14 those with high population densities. The spatial resolution and detail needed for health
15 effect assessment is rather high, because of strong gradients in population densities. Health
16 effects are believed to occur both as a consequence of short-term acute episodic exposure, as
17 well as through cumulative chronic long-term exposure. The relevant particle sizes are in the
18 inhalable size range (< 10 /mi), but it is known that submicron particles penetrate deeper
19 into the lungs. It is not well known which ambient aerosol chemical species are most potent
20 in causing health effects. However, sulfates, particularly in acidic form are believed to be
21 important along with toxic trace metals and carcinogenic organic substances. Also, health
22 damage is usually the consequence of the combined influence of multiple, coexisting
23 pollutants, weather, and other environmental conditions.
24 Preventing the degradation of visibility, particularly in pristine national parks has been
25 one of the provisions of the Clean Air Act. The main cause of visibility degradation is
26 atmospheric haze that is contributed mainly by fine particles, except during dust events.
27 Sulfates, organics, nitrates, absorbing carbon and to some extent dust particles are the
28 contributors to visibility degradation. Both short term fine particle episodes, seasonal pattern
29 and long term fine particle trends are relevant to visibility degradation. The potential
30 climatic effects of aerosols are influenced by roughly the same factors that determine the
31 visibility degradation except on a global scale.
April 1995 6-1 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6.1 AEROSOL INFORMATION NEEDS FOR ASSESSING EFFECTS
Concern
Space Resolution
Time Resolution
Particle Size
Chemistry
Health
People
Local
Short, Long
Inhalable (Fine?)
Acidity, Toxicity
Visibility
National Parks
Regional
Short, Long
Fine
SO4, NO3, Organics
Materials
Damage
Costs
Local
Long
Fine
Acidity, Soot
1 Aerosol effects on man-made materials include soiling and corrosion. These materials
2 are located mainly in populated areas and high spatial resolution for concentrations is needed.
3 Soiling is due to carbonaceous smoke and soot and settling dust while corrosion increases are
4 due to acidifying sulfurous aerosols. The materials damage occurs over the period of years,
5 but daily and seasonal cycles are also important. The quantification of materials damage is
6 particularly sensitive to the interaction with weather elements, particularly moisture.
7 Other regional and global aerosol effects include acid deposition and effects on
8 climate. Acid deposition and its relationship to aerosols is treated extensively elsewhere
9 (NAPAP, 1991). Direct aerosol perturbation of the radiative climate and the indirect aerosol
10 influence through changing the cloud properties and pattern is beyond the scope of this
11 chapter.
12 The commonality among these effects (Table 6.1) is that the overall damage is driven
13 by the concentration of relevant aerosol parameters, the spatial pattern and density of
14 receptors and by the receptors' sensitivity. The receptor densities, e.g. population densities,
15 national parks, etc. are not discussed here. The sensitivity (damage functions) are treated in
16 the respective chapters on aerosol effects.
17
18 6.1.1 Dimensionality and Structuring of the Aerosol Data Space
19 Aerosol concentration patterns contain endless detail and complexity in space, time,
20 size, and chemical composition. Chemically analyzed aerosol samples over the conterminous
21 United States reveal the coexistence of sulfates, secondary organics, nitrates, smoke, soot,
22 soil dust, sea salt, and trace metals in most aerosol samples. Each of these chemicals is also
23 distributed in different size particles. This chemically rich aerosol mix arises from the
April 1995
6-2
DRAFT-DO NOT QUOTE OR CITE
-------
1 multiplicity of contributing aerosol sources, each having a unique chemical mixture for the
2 primary aerosol at emission time. The primary aerosol chemistry is further enriched by the
3 addition of species during atmospheric chemical processes. Finally, the immensely effective
4 mixing ability of the lower troposphere stirs these primary and secondary particles into a
5 mixed batch with various degrees of homogeneity, depending on location and time. The
6 result is a spatial temporal, size, and chemically heterogeneous aerosol pattern that is
7 probably unparalleled in the domain of atmospheric sciences.
8 In order to characterize the aerosol pattern that is useful for effects assessment, it is
9 necessary to organize and structure the aerosol pattern analysis. This "user-driven" aspect of
10 aerosol concentration structuring demands that it be consistent with the information needs
11 stated in Table 6.1.
12 Another consideration in structuring the aerosol pattern analysis is that it has to be
13 consistent with physical and chemical processes that determine the concentrations. The
14 principles of atmospheric sciences state that the concentration of particulate matter, (C) at
15 any given location and time is determined by the combined interaction of emissions, (E),
16 dilution, (D), and chemical transformation and removal, (T), processes expressed as:
17
18 C = f (D,T,E)
19
20 Dilution, transformation/removal and emissions, D,T,E are generic operators and can,
21 in principle, be determined from suitable measurements and models. However, for
22 consideration of aerosol pattern analysis it is sufficient to recognize and separate these three
23 major causal factors influencing the aerosol concentration pattern.
24 This section outlines the main organizing principles for the analysis of PM pattern. It
25 is convenient to categorize the highly variable aerosol signal along the following major
26 dimensions: space, tune, size and chemical composition. The space and time dependence of
27 concentrations area common to all pollutants. However, both the distribution with respect to
28 particle size as well as the chemical distribution within a given size range constitute unique
29 dimensions of particulate matter that is not present for other pollutants. The concentration of
30 single-compound gaseous pollutants can be fully characterized by their spatial and temporal
31 pattern. This classification by dimensions is consistent with the size-chemical composition
April 1995 6-3 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
distribution function introduced by Friedlander (1976). It could be said that particulate
matter is a composite of hundreds of different pollutants having a distribution in space and
time.
6.1.2 Spatial Pattern and Scales
The spatial dimension covers the .geographic scale and pattern of aerosols. Based on
consideration of emissions, meteorology, and political boundaries, the spatial dimension can
be classified as global, national, regional-synoptic, meso, urban, and local. Some of the
characteristics of these spatial scales are illustrated in Table 6-2.
TABLE 6-2. SPATIAL REGIONS AND SCALES
Global
Continent
10,000 -
50,000 km
National
Country
5,000 -
10,000 km
Regional
Multi-state
1,000-
5,000 km
Meso
State
100-
1,000km
Urban
County
10 - 100 km
Local
City-center
1 - 10km
1 6.1.3 Temporal Pattern and Scales
2 The time dimension of aerosols extends over at least six different scales (Figure 6-1).
3 A significant, unique feature of the temporal domain is the existence of periodicities. The
4 secular time scale extends over several decades or centuries. Given climatic and chemical
5 stability of the atmosphere the main causes of secular concentration trends are changes in
6 anthropogenic emissions. The yearly scale is imposed by seasonal variation of solar
7 radiation. Emissions, atmospheric dilution, as well as chemical/removal processes are
8 influenced by the seasonal cycle. The weekly periodicity is unique among the time scales in
9 that it is imposed exclusively by human-induced emission changes. The synoptic scale
10 covers the duration of synoptic meteorological events (3-5 days). Its role is primarily
11 reflected in dilution and chemical/removal processes. The daily cycle is again imposed by
12 solar radiation and it strongly influences the emissions, dilution, and chemical/removal
13 processes. Microscale defines variation of the order of an hour caused by short-term
14 atmospheric phenomena. In the analysis that follows we will emphasize secular trends and
April 1995
6-4
DRAFT-DO NOT QUOTE OR CITE
-------
Emissions
X
Dilution
X
Chemistry /Removal
=
Concentration
Secular
Yearly
Weekly
Synoptic
Daily
Micro-scale
10-100 yrs
Minutes
1
2
3
4
5
6
Figure 6-1. Time scales for particle emissions.
yearly cycles, with some consideration of daily aerosol pattern. The microscale patterns will
be largely ignored.
6.1.4 Space-Time Relationships
The spatial time scales of aerosol pattern are linked by the atmospheric residence time
of particles. Short residence tunes restrict the aerosol to a short transport distance from a
April 1995
6-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 source, causing strong spatial and temporal gradients. Longer residence times yield more
2 uniform regional pattern caused by long range transport. The relationship between spatial
3 and temporal scales for coarse and fine particles is illustrated in Figure 6-2.
4 The aerosol residence time itself is determined by the competing rates of chemical
5 transformations and removal rates. Secondary aerosol formation tends to be associated with
6 multi-day long range transport because of the time delay necessary for the formation. For
7 sulfates, for example, the residence time is 3-5 days. For fine particles, 0.1 jurn and above,
8 the main removal mechanism involves cloud processing, while coarse particles above 10 ptm
9 are deposited by sedimentation. Ultrafine particles, below 0.1 /*m also coagulate to form
10 particles in the 0.1 to 1.0 /mi size range. As a consequence of low removal rates, aerosols
11 in 0.1-1.0 pirn size range reside in the atmosphere for longer periods than either smaller or
12 larger particles (Figure 6-3). If aerosols are lifted into the mid- or upper-troposphere their
13 residence time will increase to several weeks. Large scale aerosol injections into the
14 stratosphere through volcanoes or deep convection extend their atmospheric residence to 1-2
15 years.
16 In the context of the specific analysis that follows, the space-time-concentration
17 relationship in urban and mountainous areas is of particular importance (Figure 6-4). Urban
18 areas have strong spatial emission gradients and also corresponding concentration gradients,
19 particularly in the winter under poor horizontal and vertical transport conditions. In the
20 summer most urban areas have similar concentrations to their non-urban background.
21 In mountainous regions, the strong concentration gradients are caused by both
22 topography that limits transport as well as the prevalence of emissions in valley floors.
23 Strong wintertime inversions tend to amplify the valley-mountain top concentration
24 difference. Fog formation also accelerates the formation of aerosols in valleys.
25
26 6.1.5 Particle Size Distribution
27 The aerosol size distribution is of importance in quantifying both the formation
28 (generation) as well as the effects of aerosols. Condensation of gaseous substances during
29 combustion in the atmosphere generally produces fine particles below 1 jum in diameter.
30 Forced resuspension of soil dust and dispersion of sea spray produces coarse particles above
31 1
April 1995 6-6 DRAFT-DO NOT QUOTE OR CITE
-------
108-
c/)
az
uu
i-
LJ
UJ
u
I
CO
cr
o
Q.
ce
i 1
Global
C02,CH4
Synoptic-
Regional
SOx.NOx.Os
ine Particles
<2um)
Mesoscale
NO.NOz, 03
Coarse
Particles
20pm)
101
101 102 103 104 105 106
RESIDENCE TIME, SECONDS
107
Figure 6-2. Relationship of spatial and temporal scales for coarse and fine particles.
1 The size distribution of particles also influences both the atmospheric behavior and the
2 effects of aerosols. Atmospheric coagulation, cloud scavenging, and removal by impaction
3 and settling are strongly size dependent. The effects on human health depend on
4 size-dependent lung penetration. Light scattering hi visibility and climatic effects is also
5 strongly dependent on particle size.
6 Measurements over the past decades (Whitby et al, 1972, Whitby 1978) show that
7 atmospheric aerosols may be classified as fine or coarse particles. The size distribution of
8 atmospheric particles is discussed hi Section 3.7. The sources, formation mechanisms, and
9 the chemical composition of these two aerosol modes are different. In general, the two
10 aerosol size modes have independent spatial and temporal pattern as described throughout this
11 chapter. Coarse, dust particles tend to be more variable hi space and tune and can be
12 suspended through natural or human-induced activities. Fine particles are largely of
April 1995
6-7
DRAFT-DO NOT QUOTE OR CITE
-------
10'3 10'2 10'1 10° 101 102 103
RADIUS .
Figure 6-3. Residence time in the lower troposphere for atmospheric particles between
0.1 and 1.0 /im.
Source: Jaenicke, 1980
1 secondary origin and their spatial-temporal pattern is more regional. Notable exceptions are
2 urban-industrial hotspots and mountain valleys where primary submicron size smoke particles
3 can prevail.
4
5 6.1.6 Aerosol Chemical Composition
6 The chemical composition of atmospheric aerosol is believed to influence the effect on
7 human health. While the causal mechanisms are not fully understood, the acidity,
8 carcinogenicity, and other forms of toxicity are chemical properties considered relevant to
9 human health.
10 The aerosol chemical composition has also become an important property for
11 identifying source types based on chemical "fingerprints" in the ambient aerosol. Since
April 1995 6-8 DRAFT-DO NOT QUOTE OR CITE
-------
Summer
rural
Winter
urban
rural
urban
rural
rural
Summer
mountain
Winter
valley
mountain
valley
mountain
mountain
Figure 6-4. Space-time relationship in urban and mountainous areas.
1 aerosols reside in the atmosphere for days and weeks, there is a substantial amount of mixing
2 that takes place among the contributions of many sources. At any given "receptor" location
3 and time, the aerosol is a mixture of tens or hundreds of source contributions each having a
4 chemical signature for possible source type identification.
5 Fine particles are generally composed of sulfates, organics, nitrates, elemental carbon
6 (soot), as well as trace metals (Section 6.6). Each major chemical species have sub-species
7 such as acidic and neutral sulfates, light and heavy organics, ammonium and sodium nitrates,
8 etc.
9 The chemical composition of coarse particles is dominated by the elements of the earth
10 crust, Si, Al, Fe, suspended from soil. Near roadways, coarse particles may be
11 contaminated by lead and other trace metals. At ocean shores, coarse particles may consist
12 of sea salt arising from breaking of waves. Both resuspended dust and sea salt are primary
13 particles, carrying the chemical signature of then- sources.
April 1995 6-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.1.7 Chapter Organization and Approach
2 The general approach in preparing this chapter was to organize, evaluate, and
3 summarize the existing large scale aerosol data sets over the United States. Emphasis was
4 placed on complete national coverage as well as the fusion and reconciliation of multiple data
5 sets. The aerosol concentrations are considered from the point of view of spatial, temporal,
6 size and chemical pattern.
7 The main organizing dimension used to structure this chapter is space. The
8 concentrations are presented on global, continental, national, regional, and
9 sub-regional/urban scales. Within each spatial domain, the spatial-temporal structure, aerosol
10 size and chemical composition is presented. The presentation of aerosol pattern begins with a
11 global and continental perspective (Section 6.2). Next, the national aerosol patterns are
12 examined (Section 6.3) as derived from non-urban and urban PM10 and PM2 5 monitoring
13 networks. In Section 6.4 the aerosol characteristics over seven subregions of the
14 conterminous United States are examined in more detail. The ten year trends, seasonal
15 patterns, as well as the PM2 s/PM10 relationship and fine particle chemical composition is
16 examined for each region. Section 6.5 focuses further on the sub-regional and urban-scale
17 aerosol pattern over representative areas of the United States.
18 The aerosol concentration pattern over the United States has been reported by many
19 aerosol researchers over the past decade. In particular the research groups associated with
20 the IMPROVE aerosol monitoring networks have been prolific producers of high quality
21 data, reports, and analysis of non-urban data. This section draws heavily on their
22 contribution but the maps, charts, and computations have been re-done for sake of
23 consistency with other (urban) data from the AIRS network. Each of the sections are
24 augmented by suitable but not complete references to the pertinent-literature.
25
26
27
28
29
30
31
April 1995 6-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.2 CONTINENTAL AND GLOBAL AEROSOL PATTERNS
2 There are two data sets which can be used to provide information on fine particles
3 concentration patterns on a continental and global scale. Routine visibility distance
4 observations, recorded hourly at many U. S. airports by the U. S. Weather Service,
5 provide an indication of fine particles pollution over the United States. The visibility
6 distance data has been converted to aerosol extinction coefficient and used to access
7 patterns and trends of aerosol pollution over the U.S. (Husar et al., 1994). Routine
8 satellite monitoring of backscattered solar radiation over the oceans by the Advanced
9 Very High Resolution Radiometer sensors on polar orbiting meteorological satellites
10 provides a data set which can be used to give an indication of aerosol pollution over the
11 world's oceans (Husar and Stowe, 1994).
12 Aerosol detection over the oceans is facilitated by the fact the ocean reflectance
13 at 0.6 fjm is only 2%. Hence, even small backscattering from aerosols produces a
14 measurable aerosol signal. The backscattering is converted to a vertically integrated
15 equivalent aerosol optical thickness assuming a shape for the aerosol size distribution
16 or phase function. Clouds are eliminated by a cloud mask, so the data are biased toward
17 clear-sky conditions. The oceanic aerosol maps represent a two-year average (July
18 1989-June 1991) prior to the eruption of Mt. Pinatubo, while the stratosphere was
19 unusually clear from aerosol. Consequently, the images represent the spatial pattern of
20 tropospheric aerosol.
21 These two data sets may then be merged to provide a continental-scale
22 perspective. Some results for North America are shown in Figures 6-5a, 6-5b, and 6-5c.
23 The oceanic aerosol for the entire globe is shown seasonally in Figure 6-6. The average
24 aerosol map of Eastern North America for June, July and August (Figure 6-5a) shows
25 areas of high optical depth over the Mid-Atlantic States and over the Atlantic Ocean.
26 The oceanic aerosol concentration is higher near the coast and declines with distance
27 from the coast. This indicates that the aerosol is of continental origin and represents the
28 plume of Eastern North America heading north-east across the Atlantic ocean. This
29 plume can also be seen in the spring and summer season oceanic aerosol patterns
30 shown in Figure 6-6.
April 1995 6_n DRAFT-DO NOT QUOTE OR CITE
-------
JUNE, JULY, AUGUST
HUSAR AND STOWE, 1994
X
Figure 6-5a. Continental scale pattern of aerosols derived from visibility
observations over land and satellite monitoring over the oceans:
Eastern North America.
Source: Husar and Stowe, 1994
April 1995
6-12 DRAFT-DO NOT QUOTE OR CITE
-------
Mar., Apr., May.
Figure 6-5b. Continental scale pattern of aerosols derived from visibility
observations over land and satellite monitoring over the oceans:
Western North America.
Source: Husar and Stowe, 1994
April 1995
6-13
DRAFT-DO NOT QUOTE OR CITE
-------
2.
c/i
ON
H
6
o
o
c!
§
Mar., Apr., May.
n
HH
H
W
Figure 6-5c. Continental scale pattern of aerosols derived from visibility observations over land and satellite monitoring
over the oceans: Southern North America.
Source: Husar and Stowe, 1994
-------
1 The continental aerosol extinction coefficient data for the southwest coast of North
2 America indicate elevated aerosol extinction over southern California. The area includes
3 the hazy South Coast and the San Joaquin Valley air basins. It is interesting to note that
4 somewhat elevated aerosol optical thickness is also recorder over the Pacific near
5 Southern California. However, the low aerosol signal and the semi-quantitative satellite
6 data preclude a clear cause-effect association.
7 The average aerosol map for Southern North America (Figure 6-5c) April, May
8 and June shows that the oceans adjacent to southern Mexico have high aerosol optical
9 thickness, both on the Gulf side and the Pacific side. The aerosol concentration is higher
10 near the coasts and declines toward the sea. This indicates that the aerosol is of
11 continental origin, over southern Mexico. The haze off the Mexican coasts is most
12 pronounced in the spring season. Visibility observations at meteorological stations also
13 indicate a spring maximum in horizontal extinction. The region is known for extensive
14 springtime slash burning. Photographs taken by astronauts show numerous areas of
15 biomass burning in the spring season. Visible geostationary satellite images taken in the
16 spring also show the haze plumes emanating from southern Mexico. However the
17 composition and sources of the Mexican haze are not established.
18 The seasonal aerosol pattern over the oceans reveals that the highest aerosol
19 signal is near the tropics, where wind-blown dust and biomass combustion from Africa
20 and southern Asia produce 5,000 km long aerosol plumes (Figure 6.6). Further aerosol
21 belts of marine origin are observed just north of the Equator and at 30 to 60° latitudes
22 in both hemispheres. The backscattering in the summer hemispheres exceed the winter
23 values by a factor of 5 to 10. There is a pronounced seasonality in each aerosol region
24 (Figure 6-7); the higher aerosol levels appear in the summer hemispheres although
25 many continental and marine aerosol regions show a spring maximum. Thus, the global
26 tropospheric aerosol is a dynamic collection of independent aerosol regions, each having
27 unique sources and temporal pattern.
28 The seasonal oceanic aerosol maps show two distinctly different spatial patterns:
29 aerosol plumes originating from continents, and oceanic aerosol patches that are
30 detached from the continents. The continental aerosol plumes are characterized by high
April 1995 6_16 DRAFT-DO NOT QUOTE OR CITE
-------
70 Q^ N. Zeland
N Hemisphere
f
Figure 6-7. Seasonal pattern of oceanic aerosols derived from satellite observations.
Source: Husar and Stowe, 1994
April 1995
6-17
DRAFT-DO NOT QUOTE OR CITE
-------
1 values near the coastal areas and a decline with distance from the coast. The most
2 prominent aerosol plume is seen over the equatorial Atlantic, originating from West Africa
3 and crossing the tropical Atlantic. It is the well known Sahara dust plume. Additional
4 continental plumes emanate form Southwest Africa, Indonesia, China-Japan, Central
5 America and eastern North America. Aerosols of marine origin dominate large zonal
6 belts (30 to 60° N and S) in the summer hemispheres as well as near the Equator. In
7 summary, the global tropospheric aerosol is a collection of largely in dependent aerosol
8 regions, each having a bio-geochemically active source and unique spatial temporal
9 pattern.
10 Based on the above global and continental-scale observations, it can be
11 concluded that the continental plume from eastern North America is not as intense as
12 those from other industrial and non-industrial regions of the world. However, quantitative
13 aerosol comparisons of global regions are not available.
14
15 6.3 U.S. NATIONAL AEROSOL PATTERN AND TRENDS
16 Our current understanding of the U.S. national aerosol pattern arises from non-urban,
17 regional background monitoring networks the Interagency Monitoring of Protected Visual
18 Environments (IMPROVE) and the Northeast States for Coordinated Air Use Management
19 (NESCAUM) and from the mainly urban network, the Aerometric Information Retrieval
20 System (AIRS). The non-urban and urban networks yield markedly different national
21 patterns, particularly over the western US. For this reason the results from the two sets of
22 observations are presented separately and the differences between two networks are
23 evaluated.
24 An early compilation of the chemical and size resolved aerosol studies list 31 aerosol
25 data sets gathered since the 1970's. However, these databases are widely dispersed, and are
26 not generally available for study or evaluation (Chow and Watson, 1988).
27
28 6.3.1 Non-Urban National Aerosol Pattern
29 Non-urban aerosol concentrations are measured at remote sites, away from
30 urban-industrial activities. Size-segregated aerosol mass and chemical composition data are
31 available for 50 sites, through the IMPROVE (Eldred et al, 1988) and NESCAUM (Poirot
April 1995 6-18a DRAFT-DO NOT QUOTE OR CITE
-------
1 et al., 1990; Flocchini et al., 1990) networks. These are located mostly in national parks
2 and wilderness areas. The PM10 and PM2 5 mass concentrations are sampled and analyzed
3 on separate filters. The sampling frequency was generally twice a week (Wednesdays and
4 Saturdays) for 24-hours. The PM2 5 samples are analyzed for chemical composition which
5 make the data sets suitable for chemical mass balance computations (e.g. Sisler et al., 1993).
6 The IMPROVE/NESCAUM aerosol data are available from 1988 through 1993.
7 The results of the national spatial and temporal pattern analysis are presented in
8 quarterly contour maps and monthly seasonal time charts. The contours drawn for the
9 eastern United States are derived from only 15 to 20 stations. As a consequence, these
10 contour lines are to be taken as guides to the eye and not as actual pattern. The quarters of
11 the year are calendrical.
12
13 6.3.1.1 Non-urban PM2 5 Mass Concentrations
14 Maps of seasonal average non-urban PM2 5 concentrations are shown in Figure 6-8.
15 The maps show that the country can be divided roughly into east and west halves. The
16 eastern United States is covered by large, contiguous PM2 5 concentrations that range from
17 10 jug/m3 in Quarter 1, and 17 /xg/m3 in Quarter 3. During the transition seasons (Quarters
18 2 and 4) the eastern U.S. non-urban PM2 5 concentrations are at about 12 /xg/m3. Within the
19 eastern US, there are subregions such as New England that have lower concentrations
20 ranging between 8 to 12 ^g/m3.
21 The lowest non-urban PM2 5 concentrations are measured over the central mountainous
22 western states. The low winter concentrations are at about 3 /xg/m3, while the summer
23 valuesare around 6 ftg/m3. Somewhat elevated PM2 5 concentrations are observed over the
24 southwestern border adjacent to Mexico as well as in California and the Pacific Northwest.
25 The non-urban fine particle mass clearly show multiple aerosol regions over the
26 conterminous US, each exhibiting a unique spatial and seasonal characteristics.
27
28 6.3.1.2 Non-urban PMCoarse Concentrations
29 The non-urban coarse aerosol mass concentration in the size range 2.5 to 10 fj.m is
30 given in the seasonal maps Figure 6-9. It is plotted on the same scale as the non-urban
31 PM2.s and PM10 maps to show that the non-urban coarse mass concentration is less than the
April 1995 6-18b DRAFT-DO NOT QUOTE OR CITE
-------
1 fine mass concentration over most of the country. The lowest non-urban coarse particle
2 concentration is recorded during the first and fourth calendar quarters when virtually the
3 entire conterminous United States showed values < 10 ^tg/m3. It is remarkable that during
4 the quarters 1 and 4, the industrialized Midwest, adjacent to the Ohio River shows low
5 PMCoarse concentration (< 10 jug/m3) comparable to the pristine mountainous Rocky
6 Mountains states. The highest non-urban coarse mass concentrations is shown during
7 quarters 2 and 3. In quarter 2, the southwestern United States adjacent to the Mexican
8 border shows the highest non-urban coarse mass concentrations. In quarter 3, the monitoring
9 sites in Florida and Great Smoky Mountains exhibit high concentrations (> 12 /ng/m3).
10
11 6.3.1.3 Non-urban PM10 Mass Concentrations
12 Maps of seasonal average non-urban PM10 concentrations are shown in Figure 6-10.
13 PM10 is the sum of the PM2 5 and PMCoarse. The spatial pattern, including the delineation
14 of aerosol regions is similar to the PM2 5. However, the PM10 concentrations exceed the
15 PM2 5 by up to factor of two depending on region and season.
16 The eastern U.S. PM10 concentrations range between 12 /*g/m3 in Quarter 1, and 25
17 Mg/m3 in Quarter 3. During the transition seasons (Quarters 2 and 4) the eastern U.S.
18 non-urban PM10 concentrations are at about 15 jug/m3, except in New England. The lowest
19 non-urban PM10concentrations are measured over the central mountainous states, 5 /xg/m3 in
20 Quarter 1, 10 ng/m3 in Quarter 3, and 7 /ig/m3 during the transition seasons. Higher PM10
21 concentrations, between 10 to 20 ^tg/m3 were measured over the southwestern United States
22 as well as over the Pacific states from California to the Northwest.
23
24 6.3.1.4 PM2 5/PM10 Ratio at Non-urban Sites
25 The PM10 aerosol mass is composed of fine mass (PM2 5) and coarse mass, below
26 10/xm (Figure 6-10). Both the sources and the effects of fine particles differ markedly from
27 those of coarse particles. For this reason it is beneficial to examine the relative contribution
28 of PM2 5 and PM10 concentrations. Figure 6-11 shows the seasonal fine mass as a fraction
29 of PM10.
30 Nationally, the fine fraction at non-urban sites ranges between 0.4 and 0.8. The
31 highest fine fraction is recorded east of the Mississippi River, where 75% of the PM10 mass
April 1995 6_2ia DRAFT-DO NOT QUOTE OR CITE
-------
1 is in particles <2.5 pm in size. This is also the region that shows the highest PM10
2 concentrations, thus fine particles dominate the non-urban aerosol concentrations east of the
3 Mississippi River. The fine fraction also exceeds the coarse fraction at the non-urban
4 northwestern sites. The fine fraction is the lowest in the southwestern United States
5 (< 50%) particularly in the spring season (Quarter 2). Evidently, the southwestern PM10 is
6 dominated by coarse particles in the spring season.
7 Spatial and seasonal variation of the fine fraction is a further indication for the
8 existence of different aerosol regions over the conterminous US. This is further illuminated
9 in Section 6.4 where the aerosol characteristics over different regions of the United States are
10 discussed.
11
12 6.3.1.5 Non-urban Fine Particle Chemistry
13 The chemical composition of non-urban fine particles over the conterminous United
14 States is now reasonably well understood. The IMPROVE/NESCAUM network provides
15 over 5 years of aerosol mass and chemical composition data. The detailed and almost
16 complete fine particle chemistry data from these networks allows the chemical apportionment
17 of the fine particle mass into aerosol types such as sulfates, nitrates, organics, soot, and fine
18 soil (Schichtel and Husar, 1992; Sisler et al., 1993, Sisler and Malm, 1994). The
19 quantification of these aerosol types is relevant to both the determination of the aerosol
20 effects as well as for source apportionment of fine particle species. It should be emphasized
21 that the chemical composition as well as the absolute concentrations of the chemical species
22 is likely to be different in urban areas and mountain valleys, than at the remote monitoring
23 sites. Also, the quantification of organics, nitrates, and other metastable species is subject to
24 major uncertainties.
25 The following discussion is a summary of the national fine particle chemistry derived
26 from non-urban monitoring networks. The national spatial pattern for fine particle sulfate,
27 nitrate, organics and soot will be presented. These aerosol types along with wind blown dust
28 account for virtually all aerosol fine mass in the conterminous US. Presentation of the
29 detailed chemical pattern in urban-industrial areas as well as in poorly ventilated air sheds
30 over the mountainous western states would be desirable but it is unavailable at this time.
April 1995 6-21b DRAFT-DO NOT QUOTE OR CITE
-------
1 In the non-urban East the fine aerosol is dominated by sulfur aerosol types (sulfate and
2 ammonium ions and associated water) and organics which together constitute approximately
3 80% of the fine paniculate mass. Over the Northeast, organics dominate the fine particle
4 mass, particularly during the winter season. In the Southwest, fine soil is also a major
5 component accounting for 25 to 30% of fine mass while sulfates are a less dominant
6 component (as shown in Figure 6-12) (Schichtel and Husar, 1992).
7 The national pattern of annual fine particle sulfate, nitrate, organics, and soot
8 concentrations from the IMPROVE network is shown in Figure 6-13 (Sisler et al., 1993).
9 The station density, particularly over the eastern United States is limited. The contour lines
10 in the annual average maps are to be used as guides to the eye, rather then actual values.
11 The eastern U.S. sulfate (Figure 6-13a) exceeds the concentrations over the mountainous
12 western states by factor of five or more. Elevated sulfate in excess of 1 /xg/m3 is also
13 reported over the Pacific coast states. Sulfates also contribute over 50% of the eastern fine
14 particulate mass, while in the West sulfates contribute 30% or below.
15 Fine particle nitrates (Figure 6-13b) are most prevalent over California, exceeding 4
16 Mg/m3 at most sites- Their share of the fine mass at several California sites exceeds 20%. It
17 is well known, however, that accurate measurement of nitrate concentration has been plagued
18 by numerous sampling problems, and uncertainties in chemical analysis.
19 Organic carbon concentrations are high over California, northwestern sites, as well as
20 at the eastern U.S. sites. Relative to fine particle mass organics contribute over 50% over
21 the Northwest, and about 30% throughout the eastern US. Sampling and chemical analysis
22 problems of organics are comparable to that of the nitrates. For this reason, the
23 concentration estimates of these meta-stable species are continuously being revised.
24 Light absorbing elemental carbon/soot concentrations are high over the Northwest,
25 southern California, as well as at the Washington DC site. In the Northwest, soot exceeds
26 10% of the fine mass concentration, but over most of the country it is 5% or less.
27 The chemical composition of PM10 and PM2 5 aerosols in the IMPROVE network
28 (Eldred et a., 1994) revealed that the average coarse mass does not differ significantly
29 between the East and Wet, however, the fine mass is higher in the East. Also about 80% of
30 soil elements and 20 % of sulfur were found in the coarse fraction. Most trace elements
31 were found in the fine fraction, both in the East and in the West. The spatial and seasonal
April 1995 6.24a DRAFT-DO NOT QUOTE OR CITE
-------
<»
•a
o
•£
-------
o
T3
OS
CD
1/5
O
1
•«->
O)
o
I
O)
•a
a
O)
ai
£8
0)
/
o
6-25 DRAFT-DO NOT QUOTE OR CITE
-------
03
ex
o
aT
I*
.2
CA
03
I
8
tu
o
en
en
&
>>
~
en
O\
O\
April 1995
6-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 pattern in particle concentration and optical extinction in the United States from the
2 IMPROVE network were also summarized by Malm et al., 1994.
3 Studying the regional patterns of non-urban trace metals in the IMPROVE network
4 (Eldred et al., 1994) found a good correlation between selenium and sulfur at all sites in the
5 East. The correlation in the West is lower. Comparison of the S/Se rations for summer and
6 winter shows that there is approximately twice the sulfur relative to selenium in summer
7 compared to winter. Zinc is highest at the sites in the central East. It does not correlate
8 well with sulfur. Lead and bromine are relatively uniform, with slightly higher mean
9 concentrations in the East. There is poor correlation between lead and bromine. Copper and
10 arsenic are highest n Arizona copper smelter region. Copper is also higher in the central
11 East.
12 Trends (1982 to 1992) of non-urban fine particle sulfur, zinc, lead, and soil elements
13 were reported by Eldred et al. (1994) using the IMPROVE network data. They observe that
14 in the southwest, sulfur trends in spring, summer, and fall decreased, while most of the
15 winter trends increased. The trends in the Northwest increase slightly. The two eastern sites
16 (Shenandoah and Great Smokey Mountains), have increased almost 4% per year in summer,
17 increased 1 to 3 % in spring and fall, and decreased 2% in winter. The annual increase was
18 between 2 to 3%. Generally, there were no significant trends in zinc and the soil elements.
19 Lead at all sites decreased sharply through 1986, corresponding to the shift to unleaded
20 gasoline. The ten year trends reported by Eldred et al. (1994) have not been compared and
21 reconciled with other compatible data.
22
23 6.3.1.6 Seasonality of the Non-urban Chemistry
24 This section discusses the seasonality of size segregated chemical composition at
25 non-urban monitoring sites (IMPROVE/NESCAUM) over the entire U.S. (Figure 6-14).
26 The nationally aggregated average PM10, PM2 5 and PMCoarse is shown in Figure
27 6-14b. The non-urban PM10 concentration ranges from 8 /xg/m3 in the winter, December
28 through February to about 15 jug/m3 in June to August. On the national scale the PM10
29 seasonality is clearly sinusoidal with a peak. Fine particles over the non-urban conterminous
30 United States account for about 50 to 60% of the PM10 mass concentration throughout the
April 1995 6-27 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
03
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
0.0
1989 Feb Mar Apr May Jim Jul Aug Sep Od Hov Dec
-A- Sulfate -a Organics
•+• Soil ; -o- Soot r
-0- SulffOrfl+SoiKSoot
PM10, PM2.5 and PMC Monthly Avg.
U.S. - IMPROVE/NESCAUM Networks
60,000, . . . . . . , . , . ,
55,000
50,000
45,000
40,000
35,000
30,000
26,000
20,000
15,000
10,000
5,000
-B- PM10
1989 Fob Mar Apr May Jun Jul Aug Sep Oct Nov Dec
I -4- PM2.5
-A- PM Coarse
US
4,000
3,500
3,000
2,500
•5 2,000
c
1,500
1,000
500
\j
1989 Feb Mar Apr May Jun Jul Aug Sep OM Nov Dec
-A- Sulfur -B- Selenium -4- Vanadium I-o- S/Se !
Scale 0^000 0-4 0-10 0^000
Figure 6-14. Seasonal pattern of non-urban aerosol concentrations for the entire
U.S. a) Average concentration, b) PM10, PM2 5, and PMCoarse. c)
Chemical fraction of sulfate, soil, organics, and soot, d) Tracer
concentrations.
April 1995
6-28
DRAFT-DO NOT QUOTE OR CITE
-------
1 year. The coarse mass accounts for 40 to 50% throughout the year. Hence, the fine-coarse
2 aerosols ratio does not change dramatically for the average non-urban aerosol.
3 The relative chemical composition of the aggregated non-urban aerosol pattern is
4 shown in Figure 6-14c, including sulfates, organics, soil, and soot as a fraction of the fine
5 particle mass concentration. The Figure also shows the sum of these four aerosol species to
6 indicate the fraction of the fine aerosol mass that is not accounted for. Most notable among
7 the missing species is the contribution of nitrates.
8 Throughout the year, sulfate aerosol, including the ammonium cation accounts for
9 30 to 40% of the fine mass. There is a remarkably mild seasonality in the nationally
10 aggregated sulfate fraction. Organics also contribute 30 to 40% of the nationally averaged
11 fine particle mass. Thus, sulfates and organics are the two dominant species contributing to
12 about 70% of the fine aerosol mass.
13 The contribution of soil dust to the fine mass ranges between 4% in the winter months
14 to 12% during April through July. Soot, i.e. elemental carbon is about 2% during the
15 summer and 5% during the winter.
16 The sum of the four measured fine mass components, sulfates, organics, soil and soot
17 add up to about 80% of the measured fine mass throughout the year. The remaining,
18 unaccounted fine mass may be contributed by nitrates, trace metals (e.g. Pb, Br), sea salt
19 (NaCl), etc.
20 The seasonal pattern of concentration of primary emission tracers, selenium, Se and
21 vanadium, V is shown Figure 6-14d. Se is a known tracer for coal combustion (Miller and
22 Friedlander, 1992?), while V is a trace constituent of fuel oil. The Figure also shows the
23 monthly average concentration of fine particle sulfur as well as the S/Se ratio. If all the fine
24 particle sulfur was contributed by coal combustion then S/Se ratio would be a measure of
25 secondary sulfate formation.
26 The national average Se concentration is rather uniform over the seasons, ranging
27 between 400 to 600 ng/m3. Since Se is a primary pollutant, the seasonal invariance means
28 that the combined effect of emissions and dilution is seasonally invariant over the year.
29 The concentration of V is between 500 to 700 ng/m3, with the higher concentrations
30 occurring in the winter season. Evidently, emission of V bearing fuel oil is more
31 pronounced during the cold season. The monthly average sulfur aerosol exhibits the highest
April 1995 6_29 DRAFT-DO NOT QUOTE OR CITE
-------
1 concentrations 1.5 /^g/m3, during June, July, and August, and the lowest values 0.9 /-ig/m3,
2 during November, December, and January.
3 The S/Se ratio is about 700 during November to January and climbs to about 1,500
4 during April through September. The higher S/Se ratio during the warm season is an
5 indication of secondary sulfate production. Alternatively, the summer sulfates could be
6 contributed by sources other than coal combustion.
7
8 Eastern United States
9 The seasonal pattern of the eastern U.S. aerosol chemistry is shown in Figure 6-15.
10 The concentration of PM10, PM2 5, PMCoarse (Figure 6-15b) indicates a similar
11 seasonality, highest concentrations in the summer, and lowest in the winter. The PM10 levels
12 range between 12 to 24 ^g/m3, the PM2 5 are between 8 to 12 /ig/m3, while PMCoarse are 4
13 to 7 /zg/m3. The size segregated aerosol data for the non-urban East show that the fine mass
14 concentration (8 to 12 /xg/m3) is higher than the national average (4 to 8 ^g/m3), while the
15 coarse mass concentration is comparable to the national average. Consequently, eastern U.S.
16 non-urban fine particles contribute 60 to 70% of the fine mass throughout the year.
17 The apportionment of the fine particle mass into its chemical components
18 (Figure 6-15c) favors sulfates which amount to 40 to 50% of the fine mass throughout the
19 year, compared to about 30% of organics. The contribution of soil dust is about 5%
20 throughout the year, while soot is more important in the winter (6%) than in the summer
21 (3%). The above three aerosol chemical components account for 85 to 90% of the measured
22 fine particle mass, leaving only marginal contribution to nitrates, trace metals, and sea salt.
23 The coal tracer selenium (Figure 6-15d) exhibits a modest winter peaked seasonality
24 between 600 to 800 ng/m3. Vanadium on the other hand, is factor of two higher in the
25 winter (1,500 ng/m3) compared to the summer (750 ng/m3). Evidently, the primary
26 contribution from fuel oil is winter peaked. The S/Se ratio is about 1,000 in the winter, and
27 it is over 2,000 in the summer months. This suggests the seasonality of secondary sulfate
28 formation during the summer months, but other factors can not be excluded.
April 1995 6-30 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.S - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
East of Rockies - IMPROVEyNESCAUM NetworKs
60,000, , , ^ , ,
55,000
50,000
45,000
40,000
36,000
30,000
25,000
20,000
15,000
10,000
5,000
&--A.
tt--£r--'
i-e- PM10
1989 Fob Mar Apr May Jun Jul Aug Sep Oct Nov Dec
! -4- PM2.5 -A- PMCoQrse
Eastern US
Eastern US
03
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
19!
-&- Sulfi
4 Soil
— Sulf
,
- -;---,••''' ~~~' •*<.... fr. .-—-'
-
- »s*^-+ s^*^
^^ \/ ^s^_.
^~~~ -Q. /* '~^' — ^
-
.:.5^-*"-*^L-i:*,-*-^.-..t.-S--:
9 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
ite -B- Organics
-o- Soot
•Org+SoiH-Soot
4,000
3,600
3,000
2,600
ro
c
1,500
1,000
500
0
19t
-A- Surfu
Scale 0-400(
-
-
-
c
''ft
A 4. _
t '""*"• S^ ~~~*^ *-^ * •'
. ^^^D-^B-^*-*
9 Fob MM Apt May Jun Jul Auj Sep Oe> Hoy Dec
-a- Selenium -+- Vanadium I -o- S/Se ~1
3 0-4 0-10 0-4000
Figure 6-15. Seasonal pattern of non-urban aerosol concentrations for the Eastern
U.S. a) Average concentration, b) PM10, PM2 5, and PMCoarse. c)
Chemical fraction of sulfate, soil, organics, and soot, d) Tracer
concentrations.
April 1995
6-31
DRAFT-DO NOT QUOTE OR CITE
-------
1 Western United States
2 The aggregated western U.S. aerosol seasonality is presented in Figure 6-16. The
3 non-urban aerosol concentrations for PM10, PM2 5, and PMCoarse are well below the
4 concentrations over the eastern United States (Figure 6-16b). Evidently, the non-urban
5 western United States differs from the eastern United States, having lower fine mass
6 concentrations, which range between 3 to 5 ^ig/m3. The coarse mass concentration (4 to
7 8 jug/m3) and seasonality is similar over the East and the West. It is worth emphasizing,
8 however, that these measurements are at remote national parks and wilderness areas in both
9 East and West. The examination of monitoring data in urban areas and confined airsheds
10 (Sections 6.4 and 6.5) reveals a highly textured pattern in space and time.
11 The fine particle chemical mass balance (Figure 6-16c) for the aggregated western
12 United States shows the dominance of organics, which account for 30 to 45% of the fine
13 mass. The higher organic fraction occurs in the November through January season. Sulfates
14 hover at 20 to 25% throughout the year. Soil dust plays a prominent role in the western fine
15 mass balance, contributing 20% in April through May, but declining to 5% by January. Soot
16 ranges between 5% in the winter and 2 % during the summer. About 25% of the fine mass
17 over the western United States is not accounted for by sulfates, organics, soil , and soot. It
18 is known (Cahill???) that nitrates are major contributors to the fine particle mass in the South
19 Coast Basin, as well as other western regions.
20 The concentration of the trace substances (Figure 6-16d) selenium and vanadium
21 shows both low concentrations and weak seasonality. The sulfur concentrations are also less
22 than half of the eastern U.S. values. The S/Se ratio is about 500 in the winter months and
23 1,000 during the summer. The low summer S/Se values would indicate less sulfate yield per
24 selenium in the summer. However, these observations need to be tempered by the fact that
25 selenium emitting coal-fired power plants are not the only sources of western U.S. sulfur.
26 The S/Se ratio is included here for sake of completeness.
27 The above discussion of national pattern of chemical and size dependence hide a rich
28 spatial and temporal texture of the U.S. aerosol pattern discussed in the following sections.
29 However, it provides the national scale gross features and serves as a broader context for the
30 more detailed examinations.
April 1995 6-32 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 • Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
West or Rockies - IMPROVE/NESCAUM Networks
55,000
50,000
45,000
40,000
35,000
c
§ 30,000
c
26,000
20,000 -
15,000
10,000
5,000
1989 Feb Mar Apr May Jun Jul Aug Sep OO Nov Dec
j -4- PM2.5 -A- PM Course
Western US
Western US
03
0.8
0.7
0.6
05
0.4
0.3
02
0.1
n n
1
_
"- _.-,--~
..-•'"'""" --., ,- --• ~
t
i .
-\ 0"
"s, ^ c- -a
'^^^^^^--^-V
r^~^rtr^~^ ^%^-^
" ,-•'' + "" '""•*•-
>•- lj o. - -o- -*---
4.UUU
3,500
3,000
2,500
„
I 2,000
c
1,500
1,000
500
D.O
1989 fell Mar Anr May Jun Jul Aug Sen Oct Nov Dec n
-
-
-
o - '"'- '•* ,
l^:^^^^^^^
-+ Soil | -o- Soot
-<% Sufft-Org+Soil+Soot
1989 Feb MM Aft Uay Jun Jul Aug Sep On Nov Dec
-A- Sulfur -g- Selenium -4- Vanadium }-e- S/Se ;
Scale 0-4000 0-4 0-10 0-4000
Figure 6-16. Seasonal pattern of non-urban aerosol concentrations for the Western
U.S. a) Average concentration, b) PM10, PM2-5, and PMCoarse. c)
Chemical fraction of sulfate, soil, organics, and soot, d) Tracer
concentrations.
April 1995
6-33
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.3.2 Urban National Aerosol Pattern - AIRS
2 The urban monitoring network is operated by state and local agencies as mandated by
3 the Clean Air Act. The data from this network constitute the sensory input for the detection
4 of exceedences over the paniculate matter standards. Federal regulations also require that
5 these monitoring data be submitted to the EPA Aerometric Information and Retrieval System
6 (AIRS). In what follows, AIRS PM10 refers to the PM10 mass concentration extracted from
7 the AIRS database.
8 The AIRS PM10 stations are mostly in urban areas but some suburban and non-urban
9 sites are also reported. The aerosol mass concentrations for PM10 and PM2 5 are sampled and
10 weighed by a variety of devices. The sampling frequency is generally every sixth day for
11 24-hours. The analysis presented in this section is based on PM10 and PM2 5 data retrieved
12 from AIRS in October 1994.
13 The PM10 station density has been increasing over time. Particulate matter sampling
14 with the size cut-off of 10/xm, PM10 begun in the early 1980s. By 1985 about 200 samplers
15 were operational and the number of sampling stations has grown to 1,350 by 1994 as shown
16 in Figure 6-17. The emergence of new stations appeared in rough proportion to the final
17 station density shown in Figure 6-17. In other words, in 1985, the national coverage had a
18 similar pattern to 1994, except less dense.
19 The results of AIRS PM10 aerosol pattern analysis are presented in quarterly contour
20 maps, as well as seasonal time charts. For valid monthly and quarterly aggregation, it was
21 required to have at least two samples a month, and six samples per quarter. For the seasonal
22 maps all the available data between 1985 to 1994 were used.
23 The seasonal contour maps also show the location of the PM10 monitoring sites. The
24 size of the rectangle at each site is proportional to the quarterly average PM10 concentration
25 using all available data between 1985 to 1994. Hence, sampling biases due to station density
26 that changed over time can not be excluded.
27 The quarterly concentration pattern of PM10 is shown in Figure 6-18. The high
28 sampler density allows the resolution of spatial texture on the scale of 100 km, particularly
29 over major metropolitan areas. However, remote regions in the central and western states
30 have poor spatial density. In the absence of rural monitoring data computerized contour
April 1995 6-34 DRAFT-DO NOT QUOTE OR CITE
-------
o\
O
O
2
3
O
c
O
3
H
W
i
Valid PM10 Stations
US, All Stations
Figure 6-17.
1984 1986
Trend of valid PM10 monitoring stations in the AIRS database.
1988
1990
1992
-------
OS
H
b
o
O
G
O
H
W
i
n
,,rw^u
PM10 Mass
Figure 6-18. AIRS PM10 quarterly concentration maps using all available data.
-------
1 plotting of PM10 is biased toward extrapolating (spreading) high concentrations over large
2 areas. This bias is particularly evident in the maps for Quarters 1 and 4 in the western
3 states, where the area of high concentration hot spots is exaggerated.
4 The AIRS PM10 concentrations over the eastern United States are lowest during
5 Quarter 1, ranging between 20 to 30 /xg/m3. The higher concentrations exceeding 30 /ng/m3
6 are confined to metropolitan areas.
7
8 6.3.2.1 National Pattern and Trend of AIRS PM10
9 The trend (1985 to 1993) of national average PM10 concentrations is shown in
10 Figure 6-19b. During the decade there was a remarkable reduction in PM10 concentrations
11 from 48/ig/m3 to 25 ptg/m3. Between 1986 and 1993, the reduction was 38%. The
12 Figure 6-19b also shows the standard deviation among the yearly average PM10
13 concentrations for each year. On the national scale the standard deviation of yearly average
14 concentrations is about 40% of the mean.
15 The concentration of PM2 5 and PM10 are compared in the scatter chart in
16 Figure 6-19c. Each point represents a pair of PM2 5-PM10 monthly average concentrations.
17 The diagonal line is the 1:1 line and shows the fine particle concentration ranges between
18 20 and 85% of PM10. The heavy solid line is derived from linear best fit regression. The
19 detailed correlation statistics is reproduced in the upper-left corner of the scatter charts.
20 The ratio of overall average PM2 5 and overall average PM10 is also indicated. For the data
21 when both PM2 5 and PM10 data were available, nationally aggregated PM2 5 particles
22 accounted for 57% of the PM10 mass.
23 The seasonal pattern of the national PM10 concentration is also depicted in
24 Figure 6-19d, utilizing all available data in AIRS. The national average PM10 seasonality
25 ranges between 27 jiig/m3 in March and April, and 33 /xg/m3 in July and August, yielding a
26 modest 16% seasonal modulation. There is also evidence of slight bimodality with the
27 December through January peak.
28 The seasonal chart also shows the annual variation of PM2 5, and PM10-PM2 5
29 (i.e. coarse particles). The national fine particle concentration shows clear evidence of
30 bimodality with peaks in July and December. It is shown below that the fine particle winter
31 peak arises from western sites, while the summer peak is due to eastern U.S. contributions.
April 1995 6-37 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM2.5 vs. PM10 - Monthly Avgerages
Conterminous U.S.
Avg X
Avg Y
Avg Y/Avg X
Corr Coeff
Slops
Y offset
Data Points
0 30 60 90 120
PM10JVVG UWCU METER (25 C)
AIRS PM10 Concentration Trends
U.S.
u
U>
60
65
50
45
40
35
30
26
20
15
10
6
1985 1986 1987 1988 1989 1990 1991 1992
I -Q- PM10AVC-SIG -+ PM10AVC + SIG
PM10 AVG
AIRS PM10, PM2.5 AND PM2.5-10 MONTHLY CONC,
Conterminous U.S.
60
3
5
55 -
50 -
45
«°
35
30
25
20
16
10
5
198S Mar Apr May J«n Jul Aug S«p Oct Nov D«c
-a- PM10 -+- PMZ6 | -A- PM2.6-10 ~|
Figure 6-19. AERS PM10 and PM2 5 concentration pattern for the conterminous
US.
April 1995
6-38
DRAFT-DO NOT QUOTE OR CITE
-------
1 The national average coarse particle concentration has a 50 % yearly modulation with a
2 single peak in July.
3 Summarizing the national PM10 concentrations one can observe that the country has
4 several major aerosol regions. Each region has a discernible geographic extent as well as
5 seasonal pattern. Over the plains of the eastern United States the spatial texture of PM10 is
6 driven by the pattern of the emission fields, while the seasonality of concentrations is likely
7 to be determined by the chemical transformation and removal processes, as well as by the
8 regional dilution. In the mountainous western and Pacific states US, pockets of wintertime
9 PM10 concentrations exist that well exceed the eastern U.S. values. It is believed that haze
10 and smoke in confined mountain valleys and air basins are strongly influenced by topography
11 which in turn influences the emission pattern, dilution, as well as the chemical transformation
12 and removal rate processes.
13 Given the regionality of the aerosol concentration pattern much of the discussion that
14 follows will be focused on the characteristics of these aerosol regions. The Rocky Mountains
15 produce a natural division between the eastern and western aerosol regimes which will be
16 discussed next.
17
18 6.3.2.2 Eastern U.S. PM10 Pattern and Trend
19 The eastern U.S. PM10 concentration (Figure 6-20b) shows a 29% downward trend of
20 yearly average PM10 concentrations from 35 /zg/m3 in 1985 to 25 /wg/m3 in 1993. The
21 decline is rather steady over time.
22 The highest eastern U.S. AIRS PM10 concentrations are recorded in Quarter 3 (Figure
23 6-20d). The peak concentrations are over the Ohio River Valley stretching from Pittsburgh
24 to West Virginia, southern Indiana and St. Louis. In this region, the PM10 concentration
25 over the industrialized Midwest exceeds 40 pig/m3. Additional hot-spots with > 40 /ug/m3
26 are recorded in Birmingham, AL, Atlanta, GA, Nashville, TN, Philadelphia, PA and
27 Chicago. IL. The summer time PM10 concentrations in New England and upstate Michigan
28 are < 20 jiig/m3.
29 The transition seasons Quarters 2 and 4 (Figure 6-20d) show about 30 jwg/m3 over
30 much of the eastern US, with concentration hot-spots over the industrial Midwest as well as
31 in the Southeast, Atlanta, GA and Birmingham, AL. It is quite remarkable, however,
April 1995 6-39 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM10 Concentration Trends
East of Rockies
AIRS PM2.5 vs. PWI10 - Monthly Avgerages
East of Rockies
160rCOkREt.ATiON' SfATi:
Corr Coeff
120|-siapB
Y offset
1101-Data Points
1985 1986 1987 1988 1989 1990 1991 1992
-A- PM10AVG -a PM10AVO-SIO +PM10AVO + SIG
AIRS PM10, PM2.5 AND PM2.5-10 MONTHLY CONC.
East of Rockies
60
65
50
45
40
35
.30
25 k_ -e-
20
15
10
6
\
30 60 90 120
PM10_AVG UGJCU METER (26 C)
1385 M»r Apr May Jun Jul Aug B*p Oct Nov D*c
-B PM10 -+ PM2.6 \-tr- PM2.8-10~ |
Figure 6-20. AIRS PM10 and PM2 5 concentration pattern for east of the Rockies.
April 1995
6-40
DRAFT-DO NOT QUOTE OR CITE
-------
1 that the PM10 concentrations in urban-industrial "hot-spots" exceed their rural surrounding
2 by less than a factor of two.
3 The spatial variability of PM10 over the eastern United States is driven primarily by
4 the varying aerosol emission density. This can be deduced from the coincidence of high
5 concentrations within urban industrial areas. The atmospheric dilution, i.e. horizontal and
6 vertical dispersion is not likely to be spatially variable. Also, the chemical aerosol formation
7 and removal processes are likely to have weak spatial gradients when averaged over a
8 calendrical quarter. Hence, the main factor that is believed to be responsible for the spatial
9 variability is the emission field of primary PM10 particles and the precursors of secondary
10 aerosols.
11 PM10 concentration in excess of 30 /*g/m3 is recorded over the agricultural states of
12 Iowa, Kansas, Nebraska, and South Dakota. The elevated PM10 concentrations over this
13 region tend to persist over all four seasons. The spread of yearly average concentrations east
14 of the Rockies is only 28%.
15 The eastern PM10 seasonality (Figure 6-20d) is rather pronounced, with winter
16 concentrations (December through March) of 24 /^g/m3, and July through August peak of 35
17 Mg/m3- The amplitude of the PM10 seasonal concentrations is about 30%.
18 The scatter chart of PM2 5-PM10 relationship shows significant amount of scatter, with
19 a slope of 0.58. The ratio of the overall average PM2 5 and PM10 concentration is 0.6 such
20 that 60% of PM10 in the sub 2.5 /im size range. The seasonality of the fine particle
21 concentration over the East is bimodal with a major peak in July and smaller winter peak in
22 January. The coarse particle concentration shows a single broad peak over the warm season,
23 April through October (Figure 6-20d). It is therefore evident that fine and coarse particles
24 have different seasonal dynamics in the East.
25
26 6.3.2.3 Western U.S. PM10 Pattern and Trend
27 The mountainous states, west of the Rockies (Figure 6-21) show high PM10
28 concentrations (>50 jug/m3) at localized hot-spots during the cold season, Quarters 1 and 4.
29 These high concentrations occur over both metropolitan areas such as Salt Lake City, as well
April 1995 6-41 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM10 Concentration Trends
West of Rockies
AIRS PM2.5 vs. PM10 - Monthly Avgerages
West of Rockies
150
1985 1986 1987 1988 1989 1990 1991 1992
-O PM10AVG-SIO -4- PM10AVG + SIG
AIRS PM10, PM2.5 AND PM2.5-10 MONTHLY CONC.
West of Rockies
60
Mar Apr May Jun Jul Aug S«p Oet N«v Dec
4- PM2.5
-B- PM10
PM10_AVG UGTCU METER (25 C)
Figure 6-21. AIRS PM10 and PM2.5 concentration pattern for west of the Rockies.
April 1995
6_42 DRAFT-DO NOT QUOTE OR CITE
-------
1 as in smaller towns in mountain valleys of Colorado Wyoming, Montana, Idaho, Oregon,
2 and Washington.
3 The main geographic features of California PM10 aerosols are the Los Angeles basin
4 and the San Joaquin Valley. Both basins show concentrations in excess of 50 /^g/m3. These
5 basins are also confined by surrounding mountains that limit the dilution, facilitate cloud
6 formation, and have emissions that are confined to the basin floor. Accordingly, they
7 represent airsheds with characteristic spatial and temporal pattern. It is likely that the actual
8 spatial texture of the PM10 concentration field in the mountainous western states has much
9 higher texture than depicted in Figure 6-2la.
10 It appears that the spatial pattern of these high concentration hot spots is driven by
11 both emissions as well as by the restricted wintertime ventilation due to mountainous terrain.
12 Over the mountainous western states the atmospheric dilution by horizontal and vertical
13 dispersion is severely restricted by mountain barriers and atmospheric stratification due to
14 strong and shallow inversions. Radiative cooling also causes fog formation which enhances
15 the production rate of aerosols in the valleys. As a consequence, mountain tops are generally
16 extruding out of haze layers. Emissions arising from industrial, residential, agricultural,
17 unpaved roadways and other sources are generally confined to mountain valleys. In the
18 wintertime the mountain valleys are frequently filled with fog which also influences the
19 chemical transformation and removal processes. As a consequence all three major factors
20 that determine the ambient concentrations, i.e. emissions, dilution, and chemical rate
21 processes are strongly influenced by the topography. For this reason, many of the maps
22 depicting the regional pattern use shaded topography as a backdrop.
23 The western half of the US, west of and including the Rockies, show a more
24 pronounced downward PM10 concentration trend (Figure 6-2 Ib). The reduction between
25 1985 (57 /ig/m3) and 1993 (26 /ug/m3) is a remarkable 55%. The reduction between 1986
26 (42 jtig/m3) and 1993 (26 jug/m3) is 38%. Standard deviation among the western stations of
27 yearly average PM10 concentrations is about 40%.
28 The PM2 5-PM10 relationship (Figure 6-21c) shows that on the average about 50% of
29 the PM10 is contributed by fine particles. The scatter chart (Figure 6-2Ic) also shows that
30 during high concentration PM10 episodes the fine fraction dominates.
April 1995 6-43 DRAFT-DO NOT QUOTE OR CITE
-------
1 The western PM10 seasonality (Figure 6-2Id) is also rather pronounced, having about
2 30% amplitude. However, the lowest concentrations (26 /ig/m3) are reported in the late
3 spring (April through June), while the highest values occur in late fall (October through
4 January).
5 The seasonality of PM2 5 west of the Rockies (Figure 6-21d) is strongly peaked in
6 November through January (40 ^ug/m3). In fact, it is about factor of four higher than the
7 summertime values. On the other hand, the coarse fraction shows a broad peak during late
8 summer, July through October. It is to be noted that in Figures 6-20 and 6-21, the fine and
9 coarse particle concentrations do not add up to PM 10, because size resolved samples were
10 only available for tens of sites, while the PM10 concentrations were obtained from hundreds
11 of monitoring stations.
12 In summary, there is a remarkable 40 to 50% reduction of national PM10
13 concentrations between 1985 and 1993. On the national average the PM10 seasonality is
14 insignificant. Desegregation of the national averages into east and west of the Rockies,
15 shows that the downward trend west of the Rockies is more pronounced than over the eastern
16 half of the US. The east west desegregation also shows that the lack of national PM10
17 seasonality arises from two strong seasonal signals that are phase shifted, the eastern United
18 States has a summer peak, the West fall and winter peak, and the sum of two signals is a
19 weakly modulated seasonal pattern. Nationally, PM2 5 mass accounts for about 57% of PM10
20 mass. The East and West show comparable fine fraction (60% in the East and 50% in the
21 West), and fine particles tend to dominate during the winter season particularly in the
22 western US.
23 It is evident that further examination discussed in the next sections will show that the
24 East-West division itself is rather crude and that dividing the conterminous United States into
25 additional subregions is beneficial in explaining the PM10 concentration pattern and trends.
26 A cautionary note on a possible sampling bias is in order. The national average
27 concentrations were calculated utilizing all of the available data since 1985, when more than
28 200 monitoring stations were operational. Since that time, the number of monitoring stations
29 has risen to more than 1,300. The implications of the changing stations density to the above
30 described national PM10 trend is not well studied. Also, changes in sampling equipment and
April 1995 6-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 monitoring protocols are also possible causes of systematic errors in the reported spatial
2 pattern and trends.
3
4 6.3.2.4 Short-term Variability of PM10 Concentrations
5 The previous aerosol concentration patterns were expressed as quarterly averages.
6 However, for health and other effects, the variance of the concentration, in particular the
7 occurrence of extreme high concentrations is of importance. The PM10 concentrations
8 exhibit marked differences in the shape of their distribution functions around the mean
9 values. For example in Figure 6-22, the day to day variations of PM10 concentrations in
10 Knoxville, TN are about 40% of the mean value of 35 jig/m3. On the other hand, the
11 concentration time series for Missoula, MT shows a coefficient of variation of 60% over the
12 mean of 34/ig/m3. During the winter season the coefficient of variation is even higher. It
13 is therefore evident, that for comparable mean concentrations the Missoula, MT site exhibits
14 significantly higher short-term variations.
15 The variability of concentration is examined spatially and seasonally by computing
16 logarithmic standard deviation (ratio of 84/50 concentration percentiles) for each monitoring
17 site. These deviations were then contoured for each season. The results are depicted in the
18 seasonal maps of the logarithmic standard deviation (Figure 6-23). The highest
19 logarithmicstandard deviation is recorded over the northern and northwestern states during
20 the cold season, Quarters 1 and 4. Regionally, the logarithmic standard deviation in the
21 north-northwest is about 2.0 with pockets of high winter variability such as Salt Lake City,
22 UT and Missoula, MT. The lowest variability prevails over the warm season, Quarters
23 2 and 3, covering the southeastern and southwestern states. Over multistate regions in the
24 southern states the summertime logarithmic standard deviation is below 1.5. This means that
25 these areas are covered more or less uniformly by summertime PM10, while the northern
26 states are more episodic.
27
28 6.3.2.5 AIRS PM2 5 Concentrations.
29 The mass concentration of fine particles in urban areas is not well known. Sampling
30 and analysis of PM2 5 is limited by small number of stations (<50), sampling period
31 restricted to few years, and different, non-standard sampling equipment was utilized for
April 1995 6-45 DRAFT-DO NOT QUOTE OR CITE
-------
o
o in
Iln
°- S
I
Mean : 34
CoVa : 60.64
-Min : 1 !
Max : 239
Points: 1660
1988 1989 1990 1991
—-PM10 300630031 MISSOULA
1992
1993
200
180
160
140
120
100
80
60
40
20
STATISTICS:
Mean : 35
CoVa : 39.92
-Min : 9
Max : 73
Points: 258
1988 1989 1990 1991
—=PM10 470931015 KNOXVILLE
1992
1993
Figure 6-22. Short-term PM10 concentration time series for Missoula, MT, adn
Knoxville, TN.
April 1995
6-46
DRAFT-DO NOT QUOTE OR CITE
-------
, Quarter 2
VO
0\
6
o
2
O
o
c;
o
PM10 Log Standard Deviation
PM10 Log Standard Deviation -source IMPROVE and NESCAU
3
^ -v -. Quarter 4
•lav:.:-".-.
PM10 Log Standard Deviation
' SJUICB IMPROVE and NESCAU
PM10 Log Standard Deviation
n
Figure 6-23. Logarithmic standard deviation AIRS PM10 concentrations.
-------
1 PM2 5 monitoring. Consequently, it is not possible to perform a detailed spatial and
2 temporal mapping and trend analysis for AIRS PM2 5.
3 The yearly average AIRS PM2 5 concentrations are shown in Figure 6-24. Figure
4 6-24 also shows the location and magnitude of PM2 5 concentrations arising from the
5 IMPROVE/NESCAUM monitoring networks. The fine particle data from the
6 IMPROVE/NESCAUM shows a smooth pattern with uniformly high concentrations
7 (> 15 /xg/m3) occurring over the eastern United States and uniformly low concentrations of
8 <5 jug/m3 between Sierra and Cascade Mountains. This pattern of non-urban fine particle
9 concentrations was discussed in Section 6.3.1.
10
11 6.3.2.6 Other National Surveys
12 A summary of urban PM10, PM2 5, PMCoarse at eight urban areas, Birmingham, AL,
13 Buffalo, NY, Houston, TX, Philadelphia, PA, Phoenix, AZ, Pittsburgh, PA, Rubidoux, CA
14 and Steubenville, OH was reported by Rhodes and Evans (1985). The overall ratio of the 10
15 jum fraction to Total Suspended Paniculate (TSP) was 0.486. The relationships between
16 PM10 and the 15 /on fraction (IP) are very linear for all sites. With exception of Phoenix,
17 AZ and Houston, TX, PM2 5 exceeded the PMCoarse mass concentration in all six urban
18 areas.
19 Spengler and Thurston (1983) reported inhalable paniculate matter (IP) concentrations
20 in six United State cities, Portage, WI Topeka, KS, Kingston, TN, Watertown, MA St.
21 Louis, MO and Steubenville, OH using dichotomous virtual impactors in the two size ranges,
22 PM2 5 , having dp<2.5 /xm and PMCoarse with 2.5
-------
c
o
w
•g
0.
o
o
ft
I
o
g
a:
a.
I
T3
CS
U
O
85
O
^
•a
-------
30
20
1
10
• IP mats
• Fine mats
• CoirM man
• Total aulfaw man
Portage, Wl
.*"•>
Topeki, KN
i i «,
I / •
*-\ I I
• > "
yVvfxVH
J FMAMJ JASONOJ FMAMJJASONOJ FMAMJJASONO
1979 1980 1901
• IP mats
* Fine mass
* Coarse man
* Total sulfate mas*
JFMAMJJASONDJFMAMJJASONDJ FMAMJ J ASOND
1970 1900 1981
60
60
3°
20
10
Harriman, TN
• IP mass
* Fine mass
* Coarse mass
• Total tulfaie mass
J FMAMJ JASONDJ FMAMJ J ASONO
1980 1981
60
Walertown, MA
30
20
10
• IP mass
* Fine mass
• Coarse mass
* Total sulfatc mass
.ANv\"'X/
JFMAMJJASONOJFMAMJJASONDJFMAMJJASONO
1979 1980 1981
70
60
50
30
20
10
Si. Louis, MO
• IP mass
« Finf mass
• Coarse mass
* Total sulfato mass
Steutxjnvillr. OH
80
70
60
1 5°
* 40
30
20
10
• IP mass
* F me mass
» Coarse mass
* Totat sulfalc mass
J FMAMJJASONOJ FMAMJJASONOJ FMAMJ JASONO
1978 1980 1981
J FMAMJJASONDJ FMAMJ JASONDJ FMAMJJASOND
1979 1980 1981
Figure 6-25. Monthly mean concentrations in (/tg/m3) of IP, Fine Fraction, and S
as (NH4)2SO4 in Portage, WI; Topeka, KS; Harriman, TN;
Watertown, MA; St. Louis, NO; and Steubenville, OH.
Source: Spengler and Thurston, 1983
April 1995
6-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 sites, an average of only 4.9 weight per cent of the coarse particle mass that was measured
2 by the dichotomous samplers consists of quartz and 0.4 weight per cent as quartz in fine
3 mass. Continental interior sites show the highest average quartz content as well as the
4 greatest variability. The coastal regions and eastern interior sites reveal the lowest quartz
5 concentrations. The complete X-ray spectra from some samples in Portland, OR, show that
6 Si comes primarily from minerals such as feldspars, where the Si in the Buffalo, NY
7 aerosols comes from quartz.
8
9 6.3.3 Comparison of Urban and Non-Urban Concentrations
10 Seasonal maps of the AIRS PM10-IMPROVE/NESCAUM PM10 spatial concentrations
11 are given in Figure 6-26. The AIRS PM2 5 concentrations everywhere exceed their adjacent
12 IMPROVE/NESCAUM concentrations. The highest AIRS PM2 5 are reported over the
13 eastern urban industrial centers, such as Philadelphia and Pittsburgh, where the
14 concentrations of 50 /xg/m3 exceed their non-urban counterparts by a factor of 2 to 3.
15 However, the excess urban PM2 5 concentrations are evidently confined to the immediate
16 vicinity of urban centers. The PM2 5 concentrations at remote New England, over the
17 southeastern US, and over the upper Midwest are within about 50% for AIRS PM2 5 and
18 IMPROVE/NESCAUM PM2 5. This indicates that over the eastern United States a
19 regionally homogeneous background of PM2 5 concentration exists that has smooth spatial
20 gradients. Superimposed on the smooth regional pattern are local hot-spots with excess
21 concentrations of factor of 2 to 3 that are confined to few miles of urban industrial centers.
22 The regional homogeneity is an indication that the eastern U.S. PM2 5 is composed of
23 secondary aerosols that is produced several days after the emission of its gaseous precursors.
24 The excess PM2 5 concentration in urban centers suggests that primary emissions such as
25 automobile exhaust, heating furnaces, and are responsible for much the urban PM2 5
26 hot-spots.
27 The reported AIRS PM2 5 concentrations over the Pacific states are generally higher
28 and average at 20 to 50 /-tg/m3. This is 5 to 10 times higher than their companion
29 IMPROVE PM2 5 concentrations. The dramatic difference is attributable to the pronounced
30 concentration differences between urban-industrial-agricultural centers that occur in
April 1995 6-51 DRAFT-DO NOT QUOTE OR CITE
-------
SO
to
H
6
o
z
o
H
O
d
o
PM10 AIRS - PM10 IMPROVE
PM10 AIRS - PM10 IMPROVE
PM10 AIRS - PM10 IMPROVE
PM10 AIRS - PM10 IMPROVE
* Figure 6-26.
n 6
H
W
Spatial maps of PM10 concentration difference between AIRS and IMPROVE/NESCAUM networks.
-------
1 mountainous air basins and the concentrations monitored at remote national parks and
2 wilderness areas that are generally at higher elevations. However, it is fair to presume that
3 the AIRS and IMPROVE PM2 5 data sets represent the extreme of aerosol concentration
4 ranges that exist over the western US. The challenging task of filling in the details,
5 i.e. spatially and temporally extrapolating the aerosol concentrations over the rugged western
6 United States is discussed in further detail in later regionally and locally focused sections
7 below.
8 It is also instructive to compare the seasonality of the urban (AIRS) concentrations to
9 the non-urban (IMPROVE/NESCAUM) data. In Figure 6-27 the difference in PM10, PM2 5,
10 and PMCoarse between AIRS and IMPROVE/NESCAUM sites, using all available data, is
11 used to indicate the urban excess particle concentration compared to the rural concentration.
12 Nationally, the urban excess fine particle concentration ranges between 18 /-tg/m3 in
13 December through February and 10 /ug/m3 in April through June (Figure 6-27a). The urban
14 excess coarse mass concentration is less seasonal ranging between 10 to 7 />tg/m3. The sum
15 of the fine and coarse national urban excess mass concentration is about 25 /zg/m3 in the
16 winter season, and 18 ^g/m3 during spring season. Hence, the nationally aggregated urban
17 and non-urban data confirm that urban area have excess concentrations on the order of 20
18 Mg/m3, and well over half is due to fine particles, particularly in the winter season.
19 The urban excess (AIRS-IMPROVE/NESCAUM difference) over the eastern United
20 States (Figure 6-27c) shows fine particles excess of 8 to 12 /*g/m3, with higher value
21 occurring during both winter and summer. The urban excess coarse mass in the eastern
22 United States is only 5 to 8 /ig/m3, peaking during spring and summer. The sum of fine and
23 coarse urban excess is 15 to 18 jug/m3 throughout the year.
24 The excess urban (AIRS-IMPROVE/NESCAUM) aggregated over the western United
25 States is much more pronounced in magnitude and seasonality. The urban excess fine mass
26 is about 30 /-tg/m3 in November through January and drops to 8 to 10 /xg/m3 in April through
27 August. The urban excess coarse mass is less in magnitude and seasonality 15 to 18 /-ig/m3
28 in July through December, and 10 to 12 /ng/m3 in March through May. The sum of the
29 urban excess fine and coarse mass is 40 to 50 /ng/m3 in November through January and about
30 20 fJLg/m3 in the spring March through June. The urban AIRS and non-urban IMPROVE)
31 networks in the western United States clearly monitor distinctly different aerosol types, as
April 1995 5.53 DRAFT-DO NOT QUOTE OR CITE
-------
US urban excess
Eastern US urban excess
Western US urban excess
60
a
D
O
Z
O
H
O
a
o
H
W
O
>0
n
h-H
H
W
)
3
O
r™
s
a.
50 --
40 --
30
10 -
-a
H h
60
TO ™ CO"
-> 5 5
•Fine+Coarse Mass
CL
(1)
CO
>
O
50 -
40
n
B)
3. 30 -
o
r-
5
Q.
c
03
H 1 1 1 h-
•3 Q. >
•2 m o
co z
c
ro
Fine - a - Coarse
• Fine+Coarse Mass
Fine - o - Coarse •
• Fine+Coarse Mass
Fine a - Coarse
Figure 6-27. Urban excess concentrations (AIRS minus IMPROVE) for the U.S., Eastern U.S., and Western U.S.
-------
1 stated by their respective goals and mandates. The urban non-urban difference is such that
2 the western non-urban concentrations are virtually irrelevant for the much higher urban
3 values, particularly in the winter season. On the other hand, the eastern urban sites are
4 greatly influenced by the non-urban, regionally representative concentrations, particularly in
5 the summer season.
6
7
8 6.4 Regional Patterns and Trends
9 This section describes the spatial, temporal, size, and chemical characteristics of seven
10 aerosol regions of the conterminous US. The size and location of these regions were chosen
11 based mainly on the characteristics of their aerosol pattern. The main criteria for delineating
12 a region were 1) the region had to posses some uniqueness in aerosol trends, seasonality, size
13 distribution, or chemical composition; 2) each territory of conterminous United States had to
14 belong to one of the regions; 3) for reasons of computational convenience the shape of the
15 regions were selected to be rectangular on unprojected latitude longitude maps. The resulting
16 criteria yielded seven rectangular aerosol regions as shown in Figure 6-28. It is recognized
17 that this selection is arbitrary and for future analysis additional regional definition criteria
18 would be desirable.
19 For sake of consistency and intercomparisons each region is described using maps
20 delineating the spatial pattern and the sampling locations (Figure section a). Monthly
21 concentrations for a given region were computed by averaging all the available data for the
22 specific month. In case of non-urban aerosol chemistry some regions only had 2 to 4
23 monitoring stations. The monthly PM2 5, PMCoarse and PM10 (Figure section b) over
24 regions illustrate the relative seasonality of each aerosol type. The non-urban regional
25 average chemical composition is presented as seasonal charts of chemical aerosol components
26 as a fraction of the fine mass concentration (Figure section c). The role of some primary
27 sources, such as coal and fuel oil combustion is indicated through seasonal charts of selenium
28 (coal) and vanadium (fuel oil) trace metals (Figure section d).
29 In addition, for each region figures will be provided showing short term variability of
30 PM10 concentrations and PM10 urban excess.
April 1995 6-55 DRAFT-DO NOT QUOTE OR CITE
-------
PM10 = 01
PM2.5 = 12.
PM2.5/10 = Q« 38
17
Figure 6-28 Aerosol regions of the conterminous US.
1 6.4.1 Regional Aerosol Pattern in the Northeast
2 The Northeast aerosol region covers the New England states, including eastern
3 Pennsylvania and eastern Virginia to the south (Figure 6-29a). In the Northeast, terrain
4 features that significantly influence regional ventilation occur over the mountainous upstate
5 New York, Vermont and New Hampshire. Throughout the year, the Northeast is influenced
6 by Canadian as well as Gulf airmasses. The region includes the Boston-New York
7 megalopolis, as well as other urban-industrial centers. It is known that the Northeast is
8 influenced by both local sources, as well as long range transport of fine particle haze from
9 other regions.
10 Data from a two year fine particle network in the Northeast (Bennett et al., 1994)
11 yielded a geometric mean concentration of PM2 5 of 12.9 and paniculate sulfur (1.4
April 1995
6-56
DRAFT-DO NOT QUOTE OR CITE
-------
1 equivalent to 4.2 ptg/m3 of sulfate), which is somewhat lower than other comparable rural
2 data.
3
4 6.4.1.1 Non-urban size and chemical composition in the Northeast
5 The summary of the non-urban aerosol chemical composition in the Northeast is
6 presented in Figure 6-29c. The region has 14 monitoring sites, 7 of which are part of
7 NESCAUM in the New England states.
8 The PM10 concentration exhibits a factor of two seasonal amplitude between 12 /ug/m3
9 in the winter, and 25 /xg/m3 in June and July (Figure 6-29b). About 60% of PM10 is
10 contributed by fine particles throughout the year, and also contribute to the summer-peaked
11 seasonality.
12 Sulfates are the most important contributors of the fine particle mass in the Northeast,
13 particularly in the summer season when they account for half of the fine mass. The organics
14 account for 30 to 40%, with the higher fractions occurring in the fall and winter, September
15 through January. In fact, during the late fall the sulfate and organic contributions are
16 comparable at 40%. Fine particle soil is remarkably unimportant throughout the year
17 (<5%). Soot on the other hand, is more significant particularly during the fall when it
18 contributes about 10% of the fine mass. The sum of the above four non-urban fine particle
19 aerosol components, account for over 90% of the measured fine particle mass throughout the
20 year. This indicates that nitrates, trace metals and sea salt are of minor importance in the
21 northeastern U.S. fine particle chemical mass balance.
22 The seasonality of both selenium and vanadium indicates a winter peak (Figure 6-29d).
23 In particular, the vanadium concentration increases by factor of two for December and
24 January compared to the summer values. Also, the V concentration is higher than over any
25 other region indicating the strongest contribution of fuel oil emissions. The S/Se ratio is
26 strongly seasonal with the winter value of 1,000 and the summer peak of 2,000 to 2,500.
27
28 6.4.1.2 Urban aerosols in the Northeast
29 The Northeast region shows a substantial (39%) decline in PM10 concentration, from
30 36 /ig/m3 in 1985 to 22 jug/m3 in 1993 (Figure 6-30b). The standard deviation among the
April 1995 6-57 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Northeast - IMPROVE/NESCAUM Networks
60,000, . , . . , , , . . . .
55,000
50,000
45,000
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
, -e- PMIO
1989 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
|~f-PM2.5 -A- PMCoffrs*
Northeast
Northeast
u.
•8
S
•8
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
1989 Feb Mai Api May Jun Jul Aug Sep Od Nov Due
Sulfate -& Organics
Soil ;V" SooT'"_'_ ~~_"
Sulf^Qra+Soil+Soot
1989 Feb Mat Apr May Jun Jul Aug Sep
', -A- Sulfur i -3 Selenium -I- Vanadium
Scale 0-4000
0-4
0-10
Oa Nov Dec
SJSe
0-4000
Figure 6-29. IMPROVE/NESCAUM concentration data for the Northeast.
(a) Monitoring locations, (b) PM10, PM2.5, and PMCoarse. (c)
Chemical fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-58
DRAFT-DO NOT QUOTE OR CITE
-------
1 monitoring stations for any given year is about 30%. The map of the Northeast shows the
2 magnitude of PM10 concentrations in proportion of circle radius. It is evident, that the
3 highest AIRS PM10 concentrations generally occur in urban centers (Figure 6-30a).
4 The seasonality of the Northeast PM10 concentration (Figure 6-30d) is a modest 20%,
5 ranging from 25 to 31 /xg/m3. There is a summer peak in July, and a rather uniform
6 concentration between September and May showing only a slight winter peak.
7 The PM2 5-PM10 relationship in the scatter charts (Figure 6-30c) show that on the
8 average 62% of PM10 is contributed by fine particles. During high levels of PM10, fine
9 particles dominate some of the months, while coarse particles during other months. It is
10 likely that urban and rural sampling locations exhibit different PM2 s-PM10 relationship.
11 In general, the regional scale emissions are not expected to vary significantly from one
12 day to another. However, both meteorological transport, i.e. dilution, as well as aerosol
13 formation and removal processes are important modulators of daily aerosol concentration.
14 Daily concentration of particulate matter exhibits strong fluctuation from one day to another,
15 mainly due to the role of the meteorological transport variability. The AIRS PM10 database
16 reports the concentrations every sixth day, synchronously over the entire country. The
17 sample duration is one day which, over the long run provides the concentration distribution
18 function of daily samples. For determination of the effects (human health, visibility, acid
19 deposition) the concentration has to be known at the specific location where the sensitive
20 receptors reside. Also the concentrations have to be known at a short, e.g. daily time scale,
21 as well as over the long-term.
22 In order to characterize the one day-scale temporal variation over a given region, the
23 entire available data aggregated over the entire region for each monitoring day are plotted as
24 time series. It is recognized that during the other five non-monitored days, the
25 concentrations may be different from the reported value. The six day sample increment
26 ensures that both weekday and weekend data are properly taken into account. The physical
27 interpretation of regionally averaged daily concentration is a measure of the regional scale
28 meteorological ventilation. High regionally averaged concentrations indicate poor
29 ventilation, i.e. combination of low wind speeds and low mixing heights and the absence of
30 fast aerosol removal rates, such as cloud scavenging and precipitation. Low regional
31 concentrations, on the other hand, represent either strong horizontal transport, deep mixing
April 1995 6-59 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1986-1993
AIRS PM10 Concentration Trends
Northeast
1985 1986 1987 1988 1989 1990 1991 1992
-5 PM10AYG-SIG --,, PM10AVG + SIG
AIRS PM2.5 vs. PM10 - Monthly Avgerages
Northeast
ISO
140
AIRS PM10, PM2.S and PM Coarse Cone.
Northeast
Corr Coeff
-Slope
Y offset
-Data Points
u
Si.
60
55
60
45
40
35
30
25
20
15
10
£
30 60 90
PM10_AVG UG/CU METER (25 C)
1986 Mar Apr May Jun Jul Aug Sep oct Nov Oec
!T!r PM10 -3- Fine -+ Coarse
Figure 6-30. AIRS concentration data for the Northeast, (a) Monitoring locations.
(b) Regional PM10 concentration trends, (c) PM10, PM2-5
relationship, (d) PM10, PM2>5, and PMCoarse seasonal pattern.
April 1995
6-60
DRAFT-DO NOT QUOTE OR CITE
-------
1 heights, or high regional removal rates. Advection of high aerosol content air masses from
2 neighboring regions may also be a cause of elevated concentration in a given region.
3 The daily variation of the regional averaged urban PM10 concentration for the
4 Northeast is shown in Figure 6-31. As a guide to the eye the single day concentration data
5 for every sixth day are connected by a line between the data points, although five in-between
6 days are not monitored. The lowest regionally averaged daily urban PM10 is about 10 /xg/m3,
7 while the highest is about 55 /ig/m3, with a regional average of 25? /ig/m3. The highest
8 concentrations (> 40 /ng/m3) occur primarily in the summer season. The tune series also
9 indicate that the high concentration episodes do not persist over multiple six day periods.
10 This is consistent with the notion that the regional ventilation that is caused by synoptic scale
11 airmass changes, which typically occur every four to seven days over eastern US. The daily
12 time series also convey the fact that day to day variation in PM10 is higher than the seasonal
13 amplitude. In fact, visual inspection of Figure 6-31 the concentration seasonally is barely
14 discernible. It can be stated, therefore, that the PM10 concentration in the Northeast is
15 highly episodic, i.e. the temporal concentration variation is both substantial and irregular.
16 The excess urban PM10 (AIRS-IMPROVE) is shown in Figure 6-32.
17
18 6.4.2 Regional Aerosol Pattern in the Southeast
19 The Southeast rectangle stretches from North Carolina to eastern Texas (Figure 6-33).
20 From the point view of regional ventilation the Southeast terrain is flat, with the exception
21 ofmildly rolling the southern Appalachian Mountains. The region is known for increasing
22 population over the past decades, high summertime humidity, and poor regional ventilation,
23 due to stagnating high pressure systems.
24
25 6.4.2.1 Non-urban Size and Chemical Composition in the Southeast
26 The non-urban PM10 concentration in the Southeast (Figure 6-33b) is roughly
27 comparable to the Northeast, exhibiting about factor of two seasonal concentration amplitude
28 between 12 /*g/m3 in the winter, and 25 /ig/m3 in the summer. An anomalous high PM10
29 concentration is recorded for July which is contributed exclusively by excess coarse particle
30 concentrations of about 10 /ig/m3. With exception of July, the fine particle mass accounts
April 1995 6-61 DRAFT-DO NOT QUOTE OR CITE
-------
o
O UJ
D
O
o
Northeast
Every Sixth Day
1991
1992
1993
Figure 6-31. Short term variation of PM10 average for the Northeast. Data are
reported every sixth day.
Northeast urban excess
Jan Mar May Jul Sep Nov
Figure 6-32. Urban excess concentration (AIRS minus IMPROVE) for the
Northeast.
April 1995
6-62
DRAFT-DO NOT QUOTE OR CITE
-------
1 for about 70% of the non-urban PM10, leaving the coarse mass of 30% or less throughout the
2 year (Figure 6-33b).
3 The most prominent aerosol species in the Southeast are sulfates contributing 40 to
4 50% of the fine mass (Figures 6-33c). The anomalous sulfate fraction (35%) coincides with
5 the high (20%) soil contribution during July. During other months, soil contribution is
6 <5% of the fine mass. The relative role of the organics in the non-urban Southeast is most
7 pronounced during the winter (40%), but declines to 25% during the summer months. The
8 soot contribution varies between 2% in the summer to 6% in the winter months.
9 The trace element concentrations of selenium and vanadium (Figure 6-33d) are
10 constant throughout the year, implying that the combined role of emissions and dilution is
11 seasonally invariant. The concentration of sulfur, on other hand shows a definite summer
12 peak, that is 2 to 3 times higher than the winter concentrations. Consequently, the S/Se ratio
13 is strongly seasonal. In fact, the warm season S/Se ratio of 2,500 is higher than over any
14 other region of the country. If Se-bearing coal combustion is the dominant source of sulfur
15 in the Southeast, than the high S/Se ratio implies that the sulfate production per coal
16 production in the summer is 2.5 times that in the winter.
17
18 6.4.2.2 Urban Aerosols in the Southeast
19 There is evidence of significant (34%) PM10 decline over the past decade (Figure
20 6-34b). By 1993, the average PM10 was 24 t*g/m3. It is worth noting that this value is
21 higher than the corresponding 1993 concentration for the Northeast (22 /ig/m3). It is also
22 remarkable that the Southeast concentration trends and patterns most closely resemble the
23 industrial Midwest described below. The unique feature of the Southeast is the uniformity of
24 the aerosol concentration among the monitoring stations. In fact the 17% station to station
25 standard deviation is by far the lowest among the aerosol regions (Figure 6-34b).
26 The Southeast is also characterized by high seasonal amplitude of 37 %, ranging
27 between 22 /xg/m3 in December through February and 35 ptg/m3 in July through August
28 (Figure 6-34d). There is no evidence of a winter peak for the southeastern US.
29 The scattergram of PM2 5-PM10 for the Southeast (Figure 6-34c) shows an average of
30 58% fine particle contribution, with considerable scatter. It should be noted, however, that
April 1995 6-63 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Southeast - IMPROVE/NESCAUM Networks
60,000 r . . . . , . , . ,—, ,
55,000
50,000
45,000
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
1983 Feb Mat Apr May Jun Jul Aug Sep Oct Nov Dec
i -B- PM10 I -I- PM2.5 -A- PM CoQrse
Southeast
Southeast
0.9
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
0.0
*-.i --
-''" '-
15(19 Feb Mar Apr May Jun Jul Aug Sep Oc» Nov Dec
-6- Sulfate 43- Organics
4 Soil |~-r- Soot !
-c. Stiif+Oro+Soll+Soot
4,000
3,600
3,000
2,500
•| 2,000
c
1,500
1,000
600
Scale 0-4000
1909 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Sulfur I -Q- Selenium -4- Vanadium -9- S/S«
n-4 0-10 0-4000
Figure 6-33. IMPROVE/NESCAUM concentration data for the Southeast.
(a) Monitoring locations, (b) PM10, PM2 5, and PMCoarse. (c)
Chemical fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-64
DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1986-1993
AIRS PM2.5 vs. PM10 - Monthly Avgerages
Southeast
160
140 -
30 60 90
PM10_AVG UG/CU METER (25 C)
60
AIRS PM10 Concentration Trends
Southeast
1985 1986 1987 1988 1989 1990 1991 1992
,-A- PM10AVG I -O- PM10AV6-SIC -4- PM10AVG + SIG
AIRS PM10, PM2.5 and PM Coarse Cone.
Southeast
o
a:
|
y
1*88 Mar Apr May Jun Jul Aug s«p Get Nov D«c
I-A- PM10 I -a- Fine -+ Coarse
Figure 6-34. AIRS concentration data for the Southeast, (a) Monitoring locations.
(b) PM10, PM2 5, and PMCoarse. (c) Chemical fraction of sulfate,
soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-65
DRAFT-DO NOT QUOTE OR CITE
-------
1 size segregated samples were only available briefly for two monitoring sites which may not
2 be representative for the large southeastern region.
3 The regionally averaged daily PM10 concentrations over the Southeast (Figure 6-35)
4 shows a clearly discernible seasonality. The concentrations during the winter months are
5 about factor of two lower than during the summer. Overall, the lowest concentrations are
6 about 14 /xg/m3, and the highest about 50 /xg/m3, which is about factor of three. However,
7 seasonality of the temporal signal accounts to about half of the variation. Hence, within a
8 given season the sixth day to sixth day variation is only about 50%. It can be concluded that
9 the PM10 concentration over the southeastern United States region is quite uniform in time,
10 although it exhibits a substantial seasonality. It is to be noted that the southeastern United
11 States also exhibits the highest spatial homogeneity, i.e. the average deviation of average
12 concentrations between the stations. A further index of the short-term temporal variation is
13 given in the seasonal maps of concentration standard deviation (Figure 6-34b). The PM10
14 urban excess (AIRS-IMPROVE) for the southeast region is given in Figure 6-36.
15
16 6.4.3 Regional Aerosol Pattern in the Industrial Midwest
17 This aerosol region stretches between Illinois and western Pennsylvania, including
18 Kentucky on the south (Figure 6-37a). The industrial Midwest is covered by flat terrain with
19 the exception of the central Appalachian Mountains in the Virginias. In the winter the region
20 is under the influence of cold Canadian air masses, while during the summer moist airmass
21 transported from the Gulf Coast prevail. This region includes the Ohio and Mississippi
22 River Valleys that are known for high sulfur emission densities. The region also includes
23 major metropolitan areas.
24
25 6.4.3.1 Non-urban Size and Chemical Composition in the Industrial Midwest
26 The seasonal pattern of the non-urban aerosol in the Industrial Midwest is shown in
27 Figure 6-37b. It is worth noting that the regional observations are based on a few
28 monitoring sites and their representativeness is questionable. The PM10 concentrations range
29 between 10 and 22 /zg/m3, comparable to the non-urban levels in other eastern U.S. regions.
30 It is quite remarkable that 70 to 80% of PM10 is contributed by fine particles throughout the
31 year. In fact, the coarse particle concentrations are 4 to 5 /ng/m3, which is lower than over
April 1995 6-66 DRAFT-DO NOT QUOTE OR CITE
-------
0!
O UJ
is
0- 5
3
O
O
3
Southeast
Every Sixth Day
1991
1992
1993
Figure 6-35. Short term variation of PM10 average for the Southeast. Data are
reported every sixth day.
Southeast urban excess
0 H 1 1 1 1 1
Jan Mar May Jul Sep Nov
Figure 6-36. Urban excess concentration (AIRS minus IMPROVE) for the
Southeast.
April 1995
6-67
DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Industrial - IMPROVE/NESCAUM Networks
80,0001 . . . . , , , , , ,
1983 Feb Mai Apt May Jun Jul Aug Sep Oct Nov Dec
PM10 ] -4- PM2.5 -£- PM CoQrs«
Industrial Midwest
Industrial Midwest
S
c
0.9
0.8
0.7
0.6
0.6
0.4
0.3
0.2
0.1 -
0.0
I- -Q.
1989 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
-A- Sulfate -a- Organics
•f Soil tj:_???!. ~Z :
-(v SulftOrg+SoiHSoot
1983 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Oac
-A- Sulfur -3. Selenium -4- Vanadium ;_-3_&fSe ;
Scale 0-4000 O^t 0-10 0-4000
Figure 6-37. IMPROVE/NESCAUM concentration data for the Industrial
Midwest, (a) Monitoring locations, (b) PM10, PM2 5, and PMCoarse.
(c) Chemical fraction of sulfate, soil, organics, and soot. d)Tracer
concentrations.
April 1995
6-68
DRAFT-DO NOT QUOTE OR CITE
-------
1 any other region of the US. Hence, the contribution of wind blown dust, fly ash, or other
2 man-induced dust entrainment is not a significant factor in the non-urban areas of the
3 Industrial Midwest.
4 The chemical mass balance (Figure 6-37c) shows that sulfates are 45 to 55 % of the
5 fine mass which is higher than the sulfate fractions in other regions. Organics exhibit a
6 variable contribution that is high (40%) during the cold season (October through February)
7 and remarkably low (20%) in July and August. The strong winter peak for the organic
8 fraction differs markedly from the Northeast where the organics are aseasonal. Another
9 unusual feature of the chemical mass balance is that the sum of sulfate, organics soil and soot
10 is about 75 % during the summer and 95 % in the winter. It is not known what is the
11 composition of the missing 25% during the summer time.
12 Chemical tracer data is shown in Figure 6-37d. The chemical tracer for coal
13 combustion, selenium ranges between 1,000 and 1,500 ng/m3, which is higher than in any
14 other region. There is a sizeable month to month variationin Se concentration (partly due to
15 a small number of data points) and the seasonality is not appreciable. This means that the
16 combined effects of coal combustion source strength and meteorological dilution is seasonally
17 invariant over the industrial Midwest. The concentration of vanadium, which is a tracer for
18 oil combustion is low throughout the year. The concentration of fine particle sulfur exhibits
19 random monthly variation but indicates a summer peak. The S/Se ratio is a rather smooth
20 seasonal curve ranging between 1,000 in the winter and 2,000 during the summer months.
21 Hence, the sulfate yield is about twice during the summer than during winter months. For
22 comparison both the Northeast and Southeast exhibit higher seasonality (factor of 2.5) in
23 S/Se ratio. A possible explanation for this change in S/Se ratio is that over the industrial
24 Midwest the average age of the aerosol producing emissions is less than over the Northeast
25 or Southeast. Alternatively, the sulfate formation rate may be higher over the Northeast and
26 Southeast.
27
28 6.4.3.2 Urban Aerosols in the Industrial Midwest
29 The PM10 concentration trends for the industrial Midwest (Figure 6-38b) show a
30 decline of 34% (from 38 to 25 /zg/m3) between 1985 and 1993. There is also a 28%
April 1995 6-69 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM10 Concentration Trends
Industrial
60
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PM2.5 vs. PWI10 -Monthly Avgerages
Industrial Midwest
160
PM10AVG i -Q- PM10AVC-SIG -+ PM10AVG + SIC
AIRS PM10, PM2.5 and PM Coarse Cone.
Industrial Midwest
LU
tn
CJ
s
30 60 90 120
PM10 JVVG UG/CU METER (25 C)
1986 Mar Apr May Jun Jul Aug 8«p Oct Nov Dec
j-A- PM10 i -a- Fine -+• Coars*
Figure 6-38.
AIRS concentration data for the Industrial Midwest, (a) Monitoring
locations, (b) PM10, PM2 5, and PMCoarse. (c) Chemical fraction of
sulfate, soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-70
DRAFT-DO NOT QUOTE OR CITE
-------
1 deviation among the stations within the region. As in the Northeast, the higher
2 concentrations occur within urban-industrial areas (Figure 6-3 8a).
3 The PM10 seasonality (Figure 6-38d) is virtually identical (37% amplitude) to the
4 seasonality of the Southeast: the lowest concentrations (25 /-tg/m3) occur between November
5 and February, while the highest values are recorded in June through August (40 jug/m3). It
6 is quite remarkable that throughout the 1980s of PM10, the trends and the seasonality of the
7 midwestern PM10 aerosols is comparable to that of the Southeast. At this time, the only
8 rationale for separating into a region is the belief that the source types and the aerosol
9 composition may possibly be different from the Southeast.
10 Fine particles contribute 59% of the PM10 concentration on the average (Figure
11 6-38c), and high PM10 can occur when either fine or coarse particles dominate. It is notable
12 that size segregated samples are available primarily from urban-industrial sites over the
13 industrialized Midwest.
14 Daily concentration over the industrial Midwest (Figure 6-39) varies between 50 and
15 75 />ig/m3. The lowest regional concentrations occur during the winter months, while the
16 highest values (in excess of 40 jug/m3) occur during the summer. It is evident, that
17 seasonality is an important component of the time series, accounting for about half of the
18 variance. The elevated concentrations occur only one sixth day observation at the time,
19 indicating general absence of prolonged episodes that last 12 days or more. The industrial
20 Midwest also show substantial spatial variability. The urban excess PM10 (AIRS-IMPROVE)
21 for the industrial midwest is given in Figure 6-40.
22
23 6.4.4 Regional Aerosol Pattern in the Upper Midwest
24 The upper Midwest covers the agricultural heartland of the country (Figure 6-40). The
25 region is void of any terrain features that would influence the regional ventilation. Industrial
26 emissions and the population density are comparatively low. However, the relatively high
27 PM10 concentrations in this region warrant a more detailed examination. In the winter, the
28 region is covered by cold Canadian airmasses, while in the summer moist Gulf air alternates
29 with drier Pacific airmasses.
April 1995 6-71 DRAFT-DO NOT QUOTE OR CITE
-------
O UJ
it
0- 5
D
O
O
Industrial Midwest
Every Sixth Day
1991
1992
1993
Figure 6-39. Short term variation of PM10 average for the Industrial Midwest.
Data are reported every sixth day.
In. Midwest urban excess
Jan Mar May Jul Sep Nov
Figure 6-40. Urban excess concentration (AIRS minus IMPROVE) for the
Industrial Midwest.
April 1995
6-72
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.4.4.1 Non-urban Size and Chemical Composition in the Upper Midwest
2 Over the non-urban areas of upper Midwest the PM10 concentration is about 8 /ug/m3
3 during November through April winter season, and increases to 15 /xg/m3 during the
4 summer. Fine and coarse particles have a comparable contribution to the PM10 mass (Figure
5 6-41b).
6 The chemical mass balance (Figure 6-4 Ic) indicates that during the March through
7 May spring season sulfates dominate, but during July through October season organics
8 prevail. This is a rather unusual pattern not observed over any other region. The
9 contribution of fine particle soil exceeds 10% in the spring as well as in the fall season.
10 Chemical tracers are shown in Figure 6-41d. Selenium concentration is low
11 throughout the year (400 to 600 ng/m3), but the highest concentrations are observed during
12 the summer. This suggest that either the Se sources or the Se transport into the Upper
13 Midwest from other regions is strongest in the summer. The concentration of the fine
14 particle sulfur is < 500 ng/m3 throughout the year, but somewhat higher during March and
15 April. The spring peak for fine particle sulfur has not been observed in any other region. It
16 is also worth noting that S/Se ratio is the highest during the spring and lowest in July
17 through September. This hints on the possibility that over the Upper Midwest additional
18 sources of fine particle sulfur are present for which Se is not a tracer. Here again, it needs
19 to be pointed out that the above chemical patterns are based on only two monitoring stations.
20
21 6.4.4.2 Urban Aerosols in the Upper Midwest
22 The agricultural upper Midwest (Figure 6-42) shows the smallest decline among the
23 aerosol regions. Over the past decade the region average PM10 concentration ranged
24 between 25 and 31 /-ig/m3. Some reduction (19%) is evident since 1989. As over the
25 eastern US, the highest concentrations occur in the vicinity of urban areas. Some of the
26 station to station concentration spread arises from low concentrations over western North
27 Dakota. On the average, the deviation among the stations over the region is a moderate 30%
28 (Figure 6-39). The upper Midwest is also unique in that it shows the regionally lowest
29 seasonal amplitude of 19%, with the slightly lower concentrations occurring in December and
30 January. The sparse size segregated data indicate that only 38% of PM10 is contributed by
April 1995 5.73 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Upper Midwest - IMPROVE/NESCAUM Networks
1
O)
c
OU,VW
55,000
50,000
45,000
40,000
35,000
30,000
26,000
20,000
15,000
10,000
6,000
n
-
-
-
-
-
-
/"-%^^
£^-*~-3>^
1989 Fob Mar Apr May Jun Jul Aug Sep Oct Nov Dec
JH PM2.5 -&- PM Coarse
Upper Midwest
Upper Midwest
10
s
5
V
\L
•6
g
C
""
03
0.8
0.7
0.6
05
0.4
03
02
0.1
-
_•' * ' '-•
-^ :- ~
''
-
/'X &-~a
/ ^\^ /
(^^ &/ Ch- B ^s— ^
x°\ ' \/f^
H' X^6 ^a^is- -^ —
^-•s- a-' *
T3
-
.-4-- a. •-'*'
.- ''•+--. 4. ___..-' '^..4.
>"-"«-" 1^ - 0- . <: . _r,_ ^ _..- -Q- -* -9 -'"?- "
3,500
3,000
2,500
1 2,000
c
1,600
1,000
500
0.0' •
inn1! Fpb Mar Anr Uav Jun Jul Aurj Spp r^ri Nnw npr A
_
-
-
-
..,
-
:•- —
^
'-'-
a •' — •> •'
A-^^^^ „ '^
~"~~&-& -^'"^^^'^^^^^'^sfe^i^^^
^-.+- -M — h---1--.^.-^" ^~a»:-«-i;:S'-
1989 Feb Mar Apr May Jun Jul Aug Sep Ocl Hov Dec
-&- Sulfate -B- Organics __
i — -— ] -A- sum.
+ Soil 1-0- Soot J
r -a Selenium -4- Vanadium I -B S/Se
-0 SurfVOrg+Soll+Soot Scale 0-4000 O^t 0-10 0^000
Figure 6-41.
EMPROVE/NESCAUM concentration data for the Upper Midwest.
(a) Monitoring locations, (b) PM10, PM2 5, and PMCoarse.
(c) Chemical fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-74
DRAFT-DO NOT QUOTE OR CITE
-------
1 fine particles. This is and indication that wind blown dust from natural or man-induced
2 sources prevails. In this sense, the region is similar to the Southwest.
3 The daily regionally averaged PM10 concentrations in the upper Midwest (Figure 6-43)
4 range between 15 and 45 ^g/m3. The highest values (>40 jug/m3) generally occur in the
5 summer season, while the low regional concentrations occur mainly in the cold season, but
6 low values also occur in the summer. It is interesting that the lowest PM10 concentrations
7 over the upper Midwest (15 /ug/m3) are comparable to the Southeast and the industrial
8 Midwest, but differ from these regions by the absence of high concentration events or
9 episodes. In fact, the PM10 "episodes" over the upper Midwest are all in 40 to 45 /ug/m3
10 concentration range, compared to 50 to 75 /*g/m3 in the Midwest. The seasonality is barely
11 discernible from the time series confirming that the day to day variation exceeds the seasonal
12 modulation. The urban excess PM10 (AIRS-IMPROVE) for the upper midwest is given in
13 Figure 6-44.
14
15 6.4.5 Regional Aerosol Pattern in the Southwest
16 The Southwest covers the arid states from western Texas to Arizona (Figure 6-45a).
17 The Southwest is characterized by mountainous terrain features between the southern Rockies
18 and the Colorado Plateau. The industrial activity and agriculture is minor compared to other
19 regions. Major population centers are El Paso, Phoenix, and Tucson. The meteorology of
20 the region is characterized by low annual precipitation, except during the summer monsoon,
21 July through September, when moist air penetrates from the Gulf of Mexico toward the
22 western states, bringing moisture and precipitation.
23 The geographic pattern Figure 6-45a of hours of blowing dust shows that western
24 Texas and southern California deserts are the dustiest regions of the country. A forty-year
25 trend of dust hours shows that the 1950s had several times the dust occurrence of the 1980s
26 (Patterson et al., 1994). It is likely, that during the severe droughts of the 1930s the dust
27 frequency was even higher.
28 Wind erosion is linked to relative humidity as well as wind speed, the combined
29 changes in both variables can sometimes trigger dust storms. An example is the dust storm
30 in November, 1991 that caused the severe accident on Interstate 5 in California. Seventeen
31 people died in this 164-car accident with low visibility from dust (Gregory et al., 1994).
April 1995 6-75 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1986-1993
AIRS PM10 Concentration Trends
Upper Midwest
AIRS PM2.5 vs. PM10 - Monthly Avgerages
Upper Midwest
0 30 60 90 120
PM10.AVG Ufi/CU METER (25 C)
1985 1986 1987 1988 1989 1990 1991 1992
j -£] PM10AVG-SIG -4- PM10AVO + SIO
!-&- PM10AVG
AIRS PM10, PM2.5 and PM Coarse Cone.
Upper Midwest
u
5
60
55
50
46
40
30
20
15
10
6
1886 Mir Apr May Jun Jul Aug s«p Oct Nov 0«c
I-A- PM10 I -a- Fin* -4- Coarw
Figure 6-42. AIRS concentration data for the Upper Midwest, (a) Monitoring
locations, (b) PM10, PM2.5, and PMCoarse. (c) Chemical fraction of
sulfate, soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-76
DRAFT-DO NOT QUOTE OR CITE
-------
o
m
CM
O UJ
It;
=>
O
O
Upper Midwest
Every Sixth Day
1991
1992
1993
Figure 6-43. Short term variation of PM10 average for the Upper Midwest. Data
are reported every sixth day.
Up. Midwest urban excess
Jan Mar May Jul Sep Nov
Figure 6-44. Urban excess concentration (AIRS minus IMPROVE) for the Upper
Midwest.
April 1995
6-77 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.S . Nonurfaan PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Southwest - IMPROVE/NESCAUM Networks
60,000
1389 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
I- PM2.5 -A- PM CoQrse
Southwest
Southwest
u.
•8
§
0.9
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
0.0
1989 Feb Mai Apr May Jun Jul Aug Sep Oct Nov Dec
-A- Sulfate -a Organics
-+• Soil
-«•• SulfHDrg+Soil+Soot
4,000
3,600
3,000
2,600
1,600
1,000
600
1983 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
i-A- Sulfur | -a- Selenium -4- Vanadium -e- S/Se
Scale 0-4000 0-4 0-10 0-4000
Figure 6-45. IMPROVE/NESCAUM concentration data for the Southwest, (a)
Monitoring locations, (b) PM10, PM2 5, and PMCoarse. (c) Chemical
fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-78
DRAFT-DO NOT QUOTE OR CITE
-------
1 Gillette and Sinclair, 1990 estimated that dust devils (strongly spiraling updrafts) in
2 arid regions of southwestern United States have comparable contribution to dust
3 re-suspension as wind-driven soil erosion.
4
5 6.4.5.1 Non-urban Size and Chemical Composition in the Southwest
6 The PM10 concentrations at non-urban southwestern sites show a double peak, one
7 during the late spring (April through July), and another in October. This bimodal seasonality
8 is imposed by the coarse particle mode. The PM2 5 mass concentration is unimodal with a
9 summer peak. Overall, the non-urban PM10 concentrations are comparatively low (8 to
10 15 pig/m3) and over 60% contributed by coarse particles (Figure 6-45b).
11 The chemical mass balance (Figure 6-45c) shows sulfates to be the main contributors
12 during the winter (December through March) as well as in late summer (July through
13 October). However, sulfate and organics contributions are comparable during March through
14 June as well as during November through December. Fine particle soil plays a prominent
15 role in the spring fine particle chemical mass balance reaching 25%. However, the role of
16 fine particle soil dust during December through February dwindles to below 10%.
17 The selenium and vanadium trace elements concentrations (Figure 6-45d) are very low
18 and rather invariant throughout the year. The fine particle sulfur concentration is low and
19 exhibits a peak during August, which is the period of the summertime monsoon, when air
20 masses of Gulf of Mexico penetrate deep into the southwestern US. The S/Se ratio is
21 comparatively low and bimodal, with peaks in April through May as well as August through
22 October.
23
24 6.4.5.2 Urban Aerosols in the Southwest
25 The downward PM10 trend of the Southwest is a remarkable 50% between 1985
26 (52 /*g/m3) and 1993 (26 /xg/m3). The decline was quite steady throughout the period.
27 Another notable feature of the Southwest is the large concentration spread of 45% among the
28 monitoring sites (Figure 6-46). Sites with low concentrations (<20 ng/m3) occur adjacent to
29 high concentration sites (>50 /xg/m3).
30 Seasonally, the Southwest PM10 concentration shows two peaks, one in late spring
31 April through June, and another during the fall October through November. The
April 1995 6.79 DRAFT-DO NOT QUOTE OR CITE
-------
1 concentration dip in August and September has not been observed for any other region. The
2 late summer concentration drop coincides with the occurrence of the moist monsoonal air
3 flows from the Gulf of Mexico. It is also notable, that the rather high 1993 PM10
4 concentration of 26 /ng/m3 is only exceeded by the southern California region.
5 The size segregated aerosol samples from the Southwest clearly show that coarse
6 particles dominate the PM10 concentration, fine particles contributing only 37%. The scatter
7 chart also reveals that high PM10 concentration months occur without the presence of fine
8 particles. Itis evident, therefore, that in the Southwest natural and man-induced coarse
9 particle dust is the dominant contributor to PM10 aerosols.
10 The short term PM10 concentration over the Southwest (Figure 6-47) exhibits a highly
11 irregular pattern, that ranges between 12 to 52 /xg/m3 regional average for any given day.
12 Both the lowest (10 to 15 /xg/m3) as well as the highest values are dispersed throughout the
13 year. The seasonality is virtually indiscernible, being much smaller than the sixth day to
14 sixth day variation. It is worth noting (Figure 6-46b) that the southwestern region has rather
15 low logarithmic standard deviation compared to other regions.
16 The urban excess PM10 (AIRS-IMPROVE) for the Southwest is given in Figure 6-48.
17
18 6.4.6 Regional Aerosol Pattern in the Northwest
19 The Northwest is defined to cover the bulk of the western United States north of the
20 Arizona border (Figure 6-49a). It is covered by mountainous terrain of the Rockies, as well
21 as the Sierra-Cascade mountain ranges. It is clear that the Northwest is actually a collection
22 of many aerosol subregions. The meteorology is highly variable between the Pacific
23 Northwest and the Rocky Mountains with prevailing winds generally from the west. The
24 main feature of the Northwest is pronounced elevation ranges between mountain tops and
25 valleys, and the resulting consequences on emission pattern (confined to the valleys) and
26 limited ventilation. The mountainous Northwest has also industrial population centers, such
27 as Seattle, Portland, Salt Lake City and Denver.
28 Examining the carbonaceous particles and regional haze in the western and
29 northwestern US, White and Macias, 1989 concluded that in the rural areas the
30 concentrations of paniculate carbon are comparable to those of sulfate. Examining
31 paniculate nitrate (White and Macias, 1987) showed that the paniculate nitrate concentration
April 1995 6-80 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM10 Concentration Trends
Southwest
1985 1986 1987 1988 1989 1990 1991 1992
PM10 AVO "I -Q. PM10AVO-SIO -+ PM10AVG + SIC
AIRS PM2.5 vs. PM10 - Monthly Avgerages
Southfetst
150
140 -
6
65
60
45
40
35
30
25
20
15
10
6
AIRS PM10, PM2.5 and PM Coarse Cone.
Southwest
-Q— i
30 60 90
PM10_AVG UGTCU METER (25 C)
120
1886 w»r Apr May Jun Jul Aug Sep Oct Nov D«c
_J -B- Fin* -4- Gears*
Figure 6-46. AIRS concentration data for the Southwest, (a) Monitoring locations.
(b) PM10, PM2 5, and PMCoarse. (c) Chemical fraction of sulfate,
soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-81
DRAFT-DO NOT QUOTE OR CITE
-------
o
m
e^
oT
o 111
i&
Q- 5
D
O
O
Southwest
Every Sixth Day
1991
1992
1993
Figure 6-47. Short term variation of PM10 average for the Southwest. Data are
reported every sixth day.
Southwest urban excess
CO
"5)
3
40
35
30 <
25
20
15
H 1-
Jan Mar May Jul Sep Nov
Figure 6-48. Urban excess concentration (AIRS minus IMPROVE) for the
Southwest.
April 1995
6-82
DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 - Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Northwest - IMPROVE/NESCAUM Networks
60,000, ,—-, , , , .
1989 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
•&PM10 .; H- PM2.5 -i- PM Coarse
Northwest
Northwest
c:
•6
0.9
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
0.0
-
13 —-— e-
1989 Feb Mai Apr May Jun Jul Aug Sep del Nay Dec
-&- Sulfate -e Organics
-+ Soil !'-5-' Soot '
-r> Sulf+Org+SolH-Soot
a,wu
3,500
3,000
2,500
2,000
1,500
1,000
500
n
-
-
-
-
, -*--' - -
^^^-^^-^^^^-^
Scale
1389 Peb Uar Apr May Jun Jul Aug Sep Oct Nov
ts- Sulfur -3- Selenium -+• Vanadium -Q- S/Se
0-4000 0-4 0-10 0-tOOO
Figure 6-49. IMPROVE/NESCAUM concentration data for the Northwest.
(a) Monitoring locations, (b) PM10, PM2-5, and PMCoarse.
(c) Chemical fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-83
DRAFT-DO NOT QUOTE OR CITE
-------
1 in the northern states, (MT, ID, WY) were substantially higher, both in absolute terms and
2 relative to sulfate concentrations. Aerosol particulate nitrates over rural mountainous West
3 were also episodic, i.e. few samples contributed a large fraction of the fine particle
4 integrated dosage.
5
6 6.4.6.1 Non-urban Size and Chemical Composition in the Northwest
7 The non-urban PM10 concentrations show low values ranging between 7 to 14 /^g/m3
8 in the northwestern US. The seasonally shows a peak in the summer which is contributed
9 by both fine and coarse particles. Coarse particles account for more than half fine mass,
10 particularly during March through June spring season (Figure 6-49b).
11 The chemical mass balance (Figure 6-49c) shows roughly comparable contributions
12 from sulfates and organics, but their seasonality is phase shifted. Sulfates prevail during the
13 spring season while organics dominate during late fall (October through January). Fine
14 particle soil dust contributes 20% during April and May, but decline well below 10% during
15 the winter months (November through February). Overall, about 80% of the fine mass is
16 accounted for by the sulfates organics, soil and soot.
17 Both selenium and vanadium concentrations (Figure 6-49d) are low in the Northwest,
18 but there is an indication of a summer peak of Se. The S/Se ratio is between 500 to 1,000,
19 which is the lowest among the regions. It is interesting that this ratio has both spring peak
20 as well as fall peak, similar to the pattern observed for the southwestern United States.
21
22 6.4.6.2 Urban Aerosols in the Northwest
23 The time trend of aggregated PM10 concentrations conveys a significant decline of
24 50% from 1985 to 1993 (from 50 to 25 /xg/m3). However, from 1986 to 1993 the decline
25 was only 37%, which is believed to be more representative for the region because of the low
26 station density in the early years. Once again, the average 1993 concentration is 25 pig/m3
27 which is comparable to the 1993 concentrations of the eastern U.S. regions. The spread of
28 concentration among the Northwest stations is large, with standard deviation of
29 45% (Figure 6-50b). This spread in the concentration values is also evident from the various
30 circle sizes of the Northwest map. It is believed but not formally confirmed that low
31 concentrations occur primarily at high elevation sites that are above the daily reach of surface
April 1995 6-84 DRAFT-DO NOT QUOTE OR CITE
-------
1 based haze and smoke layers. Inspection of the circles in the northwestern map also reveals
2 that the highest PM10concentrations in the Northwest occur at more remote mountainous
3 valleys, rather than in the center of urban-industrial areas.
4 The seasonality of the northwestern United States has an amplitude of 36% which is
5 comparable to the strong seasonality of the eastern US. However, the shape of the seasonal
6 signal is phase shifted to a winter peak. The lowest concentration occurs during March
7 through May and gradually increases to a peak in December through January, falling sharply
8 between January and March.
9 The limited PM2 5-PM10 data for the Northwest indicate that on the average 59% of
10 PM10 particles are PM2 5. The scatter chart also indicates that the extreme PM10
11 concentrations are contributed mainly by fine particles. Furthermore, the extreme PM10
12 concentration also occur in the winter season.
13 The daily concentration when averaged over the large and heterogeneous northwestern
14 region exhibits a remarkably small sixth day to sixth day variation (Figure 6-51).
15 Furthermore, there is clear seasonality with a strong winter peak. Within a given season, the
16 regionally averaged concentrations only vary by 20 to 40% from one sixth day to another.
17 This low temporal variance would indicate homogeneity of aerosol over the region.
18 However, examination of the logarithmic standard deviation (Figure 6.3.12) shows that the
19 Northwest is spatially more heterogeneous and has the highest logarithmic standard deviation
20 among all regions. This apparent discrepancy shows the advantages of aggregations
21 performed on multiple spatial and temporal scales. Evidently, in the Northwest high
22 concentration PM10 pockets, in topographically confined airsheds result in strong spatial and
23 temporal variations. However, the sensory evidence suggests that large scale elevated PM10
24 concentrations that cover the entire Northwestern region do not exist because high
25 concentrations are not "synchronized" between the different airsheds. In this sense, the
26 Northwest differs markedly from the eastern US, where large regional scale airmasses with
27 elevated PM10 determine the regionally averaged values. The urban excess PM10 (AIRS-
28 IMPROVE) for the Northwest is given in Figure 6-52.
April 1995 6-85 DRAFT-DO NOT QUOTE OR CITE
-------
Average AJRS PM10 • 1985-1993
AIRS PM10 Concentration Trends
Northwest
60, p • • . 1 . r
AIRS PM2.5 vs. PM10 -Monthly Avgerages
Northwest
1986 1985 1987 1988 1989 1990 1991 1992
\-tr PM10AVG i -Q- PM10AVO-SIO -+ PM10AVG + SIC
AIRS PM10, PM2.5 and PM Coarse Cone.
Northwest
u
&
0 30 60 90
PM10JWG UGJCU METER {26 C)
1S8« M»r Apr M»y Jun Jul AUJ S«p 0«t Ncv Dee
\-ts- PM10 I -B- Fin* -+• Coars*
Figure 6-50. AIRS concentration data for the Northwest, (a) Monitoring locations.
(b) PM10, PM2 5, and PMCoarse. (c) Chemical fraction of sulfate,
soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-86
DRAFT-DO NOT QUOTE OR CITE
-------
o
O UJ
IS
OL 5
D
O
O
Northwest
Every Sixth Day
1991
1992
1993
Figure 6-51. Short term variation of PM10 average for the Northwest. Data are
reported every sixth day.
Northwest urban excess
"5)
» 20
O
i
a.
Jan Mar May Jul Sep Nov
Figure 6-52. Urban excess concentration (AIRS minus IMPROVE) for the
Northwest.
April 1995
6-87
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.4.7 Regional Aerosol Pattern in the Southern California
2 The region covers California, south of San Francisco Bay (Figure 6-53a). It was
3 declared as a separate region primarily because of the known high aerosol concentrations in
4 the Los Angeles and San Joaquin basins. Meteorologically the region is exposed to the air
5 flows from the Pacific that provide the main regional ventilation toward the south and
6 southeast. The precipitation in the region occurs in the winter season, with the summer
7 being hot and dry. The regional ventilation of the San Joaquin Valley is severely restricted
8 by Sierra Mountain ranges. Also, the San Gabriel Mountains constitute an air flow barrier
9 east of the Los Angeles basin. Both basins have high population, as well as industrial and
10 agricultural activities. Hence, human activities are believed to be the main aerosol sources
11 of the region.
12
13 6.4.7.1 Non-urban Size and Chemical Composition in the Southern California
14 The PM10 concentration at the non-urban sites over southern California ranges
15 between 10 /ig/m3 during December through February, and 20 to 25 /*g/m3 in April through
16 October. Coarse particles contribute more than 50% of PM10 during the warm season May
17 through October. Both the fine and coarse aerosol fractions are lowest during the winter
18 months (December through March). The summer peak fine particle seasonality at non-urban
19 southern California sites is in marked contrast to the strongly winter peaked urban fine
20 particle concentrations (Figure 6-53b).
21 The chemical mass balance (Figure 6-53c) of non-urban southern California aerosol is
22 clearly dominated by organics, which contribute 30 to 40% throughout the year. Sulfates
23 account for only 10 to 15% of the fine mass in the winter, and about 20% in the summer
24 months. The sulfate fraction of the non-urban southern California fine mass is the lowest
25 among the regions. Fine particle soil dust is about 10% between April through November
26 and drops to 5% during the winter months. A notable feature of the southern California
27 chemical mass balance is that 45% of the winter, and 35% of the summer fine mass
28 concentration is not accounted by sulfates, organics, soils, and soot. Undoubtedly nitrates
29 are the major contributors to the southern California aerosols.
30 Both selenium and vanadium (Figure 6-53d) show low values throughout the year
31 without significant seasonality. On the other hand the fine particle sulfur concentration
April 1995 6-88 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 . Nonurban PM Monitoring Sites
PM10, PM2.5 and PMC Monthly Avg.
Northwest - IMPROVE/NESCAUM Networks
60,000, . . . . . . . , , , ,
55,000
50,000
45,000
40,000
35,000
30,000
25,000
20,000
16,000
10,000
5,000
•*---
1989 Feb Mar Apr May Jun Jul Aug Sep Od Nov Dec
:~iEHPM?0; -4- PM2.5 -A PMCoQrse
S. California
S. California
vt
Uu
g
1
LL
4000, r- , r— -, . . . . • • . 1
OS
0.8
0.7
0.6
0.5
0.4
0.3
02
0.1
-
-
•'''""" ,-i*-
f- ""• '''' "" •*"•
~" "^ *- -
^ ''
-
"-• r^'B^Q~L
'TZL £3r~ ~~^~^ ~l-J^-~ __ ^.Q 'U
A.
^ ^
— &^~^ ^^\- _^A~~^~^
^, ft\ _f* — *>" ^-j
fc...-^--~*- o- , 4,. -i- 0- - -S- -S- •'
3,500
3,000
2,500
n
I 2,000
c
1,500
1,000
600
0.0 ' ' ~
*««.* i- t u_- A ••_.. !..» lul Ann Can Art Nnw Hpr n
-
-
_
-
.-, - '" ""
N
.-•'" .,,---1^-*^^-^,^^
'^Sr^ ^ ^
.
4- Soil l-o- Soot
-o- Sulf+Org+SoiH-Soot
1989 feb Mar Apr May Jun Jul Aug Sep Ocl Nov Dec
Scale 0-4000 0^ 0-10 0-4000
Figure 6-53. IMPROVE/NESCAUM concentration for the Southern California, (a)
Monitoring locations, (b) PM10, PM2 5, and PMCoarse. (c) Chemical
fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-89
DRAFT-DO NOT QUOTE OR CITE
-------
1 shows a definite summerpeak at 500 ng/m3, compared to 200 ng/m3 during the winter.
2 Consequently, the S/Se ratio increase from 500 in the winter 1,000 to 1,500 in the summer.
3
4 6.4.7.2 Urban Aerosols in the Southern California
5 The downward concentration trend of southern California between 1985 and 1993 was
6 41%. However, between 1986 and 1993 the reduction was 27%. By 1993, the southern
7 California annual average PM10 concentration was reduced to 32 ^g/m3. However, there is a
8 sizable concentration spread among the stations (40% standard deviation). Inspection of the
9 circle sizes in the map points to uniformly high concentrations in the San Joaquin Valley as
10 well as in the Los Angeles basin. The low concentration sites are located either on the
11 Pacific coast or in the Sierra Mountains. Thus there are clear patterns of basin-wide elevated
12 PM10 concentrations and lower values in more remote areas (Figure 6-54b).
13 The seasonality of the PM10 pattern in southern California is significant at 27%.
14 Furthermore the seasonal pattern is unique that the highest concentrations occur in
15 November, while the lowest in March. However, it is a see saw rather than a sinusoidal
16 pattern.
17 On the average, about half of southern California PM10 is contributed by fine particles
18 as shown in PM2 5-PM10 scattergram. Most of the high PM10 concentration months are
19 dominated by fine particles and tend to occur in the fall, season.
20 The sixth day average time series for the southern California region (Figure 6-55)
21 shows remarkably high sixth daily variance, between 12 and 75 /ig/m3. The lowest values
22 tend to occur between January and April, while the highest concentrations (>50 /ig/m3) tend
23 to occur during October through December. Concentration excursions factor of two are
24 common between two consecutive six day time periods. However, visual inspection of the
25 sixth daily signal also reveals a substantial seasonality that peaks in the fall September
26 through December and lowest in the spring.
27 The urban excess PM10 (AIRS-IMPROVE) for Southern California is given in
28 Figure 6-56.
April 1995 6-90 DRAFT-DO NOT QUOTE OR CITE
-------
Average AIRS PM10 -1985-1993
AIRS PM10 Concentration Trends
S. California
60
1985 1986 1987 1988 1909 1990 1991 1992
-Q- PM10 AVO - SIO -+- PM10 MG + SIO
AIRS PM2.5 vs. PM10 - Monthly Avgerages
S. California
CO&RELATiOH'
AIRS PM10, PM2.5 and PM Coarse Cone.
S. California
Avg X
Y
Avg Y/Avg X
Corr Coeff
Slope
Y offset
Data Points
0 30 60 90 120
PM10JVVG UGJCU METER (25 C)
I
1986 Mar Apr Mjy Jun Jul Aug S«p Oct Nov
I -4- PM10 I -Q- Fine -+• Coarse
D«c
Figure 6-54. AIRS concentration for the Southern California, (a) Monitoring
locations, (b) PM10, PM2 5, and PMCoarse. (c) Chemical fraction of
sulfate, soil, organics, and soot, (d) Tracer concentrations.
April 1995
6-91
DRAFT-DO NOT QUOTE OR CITE
-------
o
m
O UJ
it
o- 5
^
O
o
Southern
California
Every Sixth
Day
1991
1992
1993
Figure 6-55. Short term variation of PM10 average for the Southern California.
Data are reported every sixth day.
S. California urban excess
10
Jan Mar May Jul Sep Nov
Figure 6-56. Urban excess concentration (AIRS minus IMPROVE) for the
Southern California.
April 1995
6-92
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.5 SUB-REGIONAL AEROSOL PATTERNS AND TRENDS
2 The health and other effects of aerosols are imposed on individuals, and the density of
3 population varies greatly in space. Consequently, the evaluation of effects requires the
4 knowledge of aerosol concentrations over specific locations where sensitive receptors reside.
5 The purpose of this section is to characterize the aerosol pattern at specific sites, small
6 airsheds or sub-regions. The discussions is organized by region and then by monitoring site
7 within a region. Most urban aerosol sampling is confined to PM10 or in some instances to
8 PM2 5 and PMCoarse. However, detailed chemical composition data are reviewed for
9 several urban areas.
10
11 6.5.1 Sub-regional Aerosol Pattern in the Northeast
12 In the northeastern region, the Shenandoah National Park and Washington, DC
13 constitute a natural urban-non-urban pair of size and chemically resolved aerosol data. New
14 York City and Philadelphia are also major metropolitan areas with substantial aerosol data.
15 Whiteface Mountain site distinguishes itself from its background by high elevation.
16
17 6.5.1.1 Shenandoah National Park
18 The PM10 concentration at the Shenandoah National Park IMPROVE site
19 (Figure 6-57a) exhibits a pronounced summer peak (27 /zg/m3), which is factor of three
20 higher than the low winter value of 9 /xg/m3. The strong seasonality is driven by the
21 seasonal modulation of the fine mass which accounts for 70 to 80% of PM10 mass
22 (Figure 6-57a). The coarse particle concentration ranges between 3 to 6 /ig/m3, which is
23 small compared to the fine particle mass, particularly in the summer season, when it accounts
24 to < 25% of the PM10. It is clear that at this remote site, in the vicinity of industrial source
25 regions, fine particles determine the magnitude of PM10.
26 The chemical mass balance for the Shenandoah IMPROVE monitoring site (Figure
27 6-57b) clearly documents the dominance of sulfate aerosols which contributes about 60% of
28 the fine mass during April through September, and about 50% during the winter months.
29 Organics, on the other hand, range from 20% in summer to 30% in the winter months. The
30 contribution of fine particle soil and soot is well below 5%. Throughout the year about 90%
31 of the fine mass is accounted for by these measured substances.
April 1995 6.93 DRAFT-DO NOT QUOTE OR CITE
-------
u.
•8
|
1
PM10, PM2.5 and PMC Monthly Avg.
Shenandoah NP
60,000
65,000
60,000
45,000
40,000
35,000
30,000
25,000
20,000
16,000
10,000
6,000
1969 Mar May Jul S«p Nov
-B- PM10 H-- PM2S I-A- PMCoare* I
Shenandoah NP
Shenandoah NP
OJ
o^
0.7
0.6
0.6
0.4
03
02
0.1
0.0
I--&-Q
B'
-Q.
4,000
3,600
3,000
2,000
1,600
1,000
600
1989 Fed Mar Apr M*y Jan M Aug Sap Od Hov Dec
-A- Sulfato -& Orfltnlcs
1S89 F«b Hw Apr Miy J« M A«| S*p Oct No* 0*c
.+. SOD 1-0- SOOt
-cv- SulfKJro+SoiHSoot
J
Scale
0-4000
-Q- 6*)«flhjRi --»- V«nt5, and PMCoarse.
(c) Chemical fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-94
DRAFT-DO NOT QUOTE OR CITE
-------
1 Chemical Tracer data is shown in Figure 6-57rc. The concentration of coal-tracer
2 selenium shows a mild double peaks, one during December through March, and another in
3 June through September. Vanadium is relatively constant throughout the year. The fine
4 particle sulfur concentration is almost factor of five higher in August (3,300 ng/m3) than the
5 December values(700 ng/m3). This extreme sulfur seasonality is unique to the Shenandoah
6 site. The S/Se ratio has a remarkably smooth but highly seasonal variation that varies by
7 about factor of four between the winter (700) and summer (2,600) values. If Se-bearing coal
8 combustion was the exclusive source of sulfur at the Shenandoah National Park , then the
9 sulfate yield per emission would be 3 to 4 higher in the summer than in the winter.
10 An examination of the nature and sources of haze in the Shenandoah Valley/Blue
11 Ridge Mountains area (Ferman et al., 1981) showed that sulfate aerosols were the most
12 important visibility reducing species. Averaging 55% of the fine particle mass, sulfates (and
13 associated water) accounted for 78% of the total light extinction. The second most abundant
14 fine particles, accounting for 29% of the fine mass, was organic carbon. The remaining
15 particle mass and extinction were due to crustal materials.
16 Using an in-situ rapid response measurement of H2SO4/(NH4)SO4 aerosol in
17 Shenandoah NP, VA, Weiss et al, 1982, found that the summer sulfate and ammonium ions
18 average 58% of particle mass smaller than 1 mm. The particle composition in terms of
19 NH4+/SO42" molar ratio ranged from 0.5 to 2.0 with strong diurnal variation. The particles
20 were most acidic at 1500 EOT and least acidic in the period 0600 to 0900 EDT. The water
21 contained in ambient aerosol particles was more strongly associated with sulfate and
22 ammonium ions than with the remainder of fine particle mass.
23
24 6.5.1.2 Washington, DC
25 The PM10 concentration at Washington DC (at the top of the National Park Service
26 Headquarters building) is virtually constant over the seasons at 25 to 30 /*g/m3. Fine
27 particles contribute over 70% of PM10 throughout the year (Figure 6-58a). The lack of
28 seasonality in the fine particle mass is in sharp contrast to the factor of three seasonal fine
29 mass modulation at the Shenandoah National Park. The coarse particle concentration in
30 Washington, DC is 8 to 10 ^g/m3 throughout the year, exhibiting virtually ho seasonality. In
April 1995 6-95 DRAFT-DO NOT QUOTE OR CITE
-------
1 fact, compared to Shenandoah National Park, the excess coarse particle concentration in
2 urban Washington, DC is only 5 j*g/m3 throughout the year.
3 The chemical mass balance at the urban Washington, DC site (figure 6-5 8b) is
4 dominated by sulfates during the summer months (50% of fine mass), which declines to 30%
5 in January. Organics, on the other hand, are more important during October through
6 January (40%) but only 30% during May through August. Thus, the relative roles of
7 organics and sulfates at the Washington, DC urban site is fully phase shifted by half a year.
8 Soot is a substantial (9 to 12%) contributor to the fine particle mass, particularly during
9 October through December. Fine particle soil contributes a remarkably low 2 to 5% of the
10 fine particle mass at this urban site. Hence, neither coarse particle, nor fine particle dust is
11 an important factor in Washington, DC.
12 The chemical tracer species are shown in Figure 6-58c. The coal tracer, selenium
13 concentration ranges between 1.5 to 2.0 ng/m3 without appreciable seasonality. The urban
14 Se is also higher than the Se at any of the non-urban site, including the industrialized
15 Midwest. Vanadium, the tracer for fuel oil, varies by factor of two between the high winter
16 values (> 8 ng/m3) and low summer values (3 ng/m3). The pronounced V concentration
17 seasonality is a clear indication of that the emissions from fuel oil and other vanadium
18 sources are seasonal. The fine particle sulfur concentration varies by about factor of two
19 between 1,400 ng/m3 winter concentration, and about 3,000 ng/m3 summer peak. The
20 seasonal modulation of sulfur in Washington, DC is only factor of two compared to the
21 factor of four fine sulfur modulation at Shenandoah National Park. The difference is
22 primarily due to the elevated winter sulfur in Washington, DC. The S/Se ratio is about 0.6
23 in the winter and about 1.5 in the summer. It differs from Shenandoah by the lower summer
24 S/Se ratios.
25
26 6.5.1.3 Comparison of Non-urban (Shenandoah) to Urban (Washington, DC) Aerosols
27 The Washington, DC urban site and the companion non-urban Shenandoah monitoring
28 site constitute a rare data pair that allows the quantification of urban-rural differences in fine
29 and coarse particle concentration, and chemical composition. Within Washington, DC,
30 industrial emissions are moderate compared to the industrial midwestern cities. However,
31 both automobile emission density and emissions from winter time heating are expected to be
April 1995 6-96 DRAFT-DO NOT QUOTE OR CITE
-------
PM10, PM2.5 and PMC Monthly Avg.
60,000
65,000
60,000
45,000
40,000
35,000
J 30,000
c
25,000
20,000
0.9
0.8
0.7
tn
Fraction of Fine Mi
o o o o o
to b !>. bi b»
0.1
0.0
19
-Ar SM
-+• Soil
15,000
10,000
5,000
0
19
-B- PM10
Washington DC
Washington DC
-
-
-
-
ja-e" H\_
J3— S¥^ '^St ^V
i— e— B— s-^1 Nj^/N
: _^ x^^^v ,J
-,.-*---.
^
i
89 Mar May Jul Sep Nov
-(-- PM25 i-A- PM Coarse |
a t> '*' ' ~~ w
.- '%'" '^— *-'
/ ^
*r *^ ^Q
^< /^X*
/^"^B-e X ^^
•0"
t ^ > - -o. ^ _4
j,--'"*"— «f.".I^~— 4--.".'°' 4,
39 F«b Hw Apt Hay Jim Jul Aug Sep Oct Hov Dec
ate -B- Organic:
)-o- Soot |
4,000
3,600
3,000
2,600
c
1,000
500
0
m
-frSulft«
Washington DC
-
-'••• --f A f'
/ 15 \ /
/ * \ * •' '"k
_B */P;"' \ •' "• ; ^
^ \ y / v' \ ,°A-e-v' ~
v^-/\ / / ^\^// \ ^
>r • ' /• <& / \
- >r x -S" ;•' \ 'V \.
J-~ - i'v
\>
:-* "^^.:
19 Felt Mar Apr
r -a- Selenl
May Jun Jul Aug Sep Oa HOY Dec
urn -•)- Vanadium | -e- SISt }
-0- SulffOrg+SoBfrSoot
Scale
0-4000
0-4
0-10
0-4000
Figure 6-58. BMPROVE/NESCAUM concentration for Washington, D.C. (a)
Monitoring locations, (b) PM10, PM2i5, and PMCoarse. (c) Chemical
fraction of sulfate, soil, organics, and soot, (d) Tracer
concentrations.
April 1995
6-97
DRAFT-DO NOT QUOTE OR CITE
-------
1 high. In this section the excess aerosol concentrations in Washington, DC over the
2 Shenandoah site are examined to elucidate the urban influence.
3 The Washington, DC excess PM10 concentration (Figure 6-59a) ranges between 15-20
4 Mg/m3 in the winter, and <3 ^cg/rn3 in the summer. Hence, there is almost an order of
5 magnitude higher urban excess during the winter, compared to the summer. The seasonality
6 of the excess PM10 is driven by the winter peak excess fine particle concentration of 10-12
7 Mg/m3- The excess coarse particles hover in the 3 to 6 pig/m3 range throughout the year.
8 Thus, the urban Washington, DC concentration exceeds its non-urban regional aerosol values
9 during the winter season, and the excess winter time urban aerosol is largely contributed by
10 fine particle mass. This is remarkable, and indicates the diminishing role of coarse particle
11 fly ash, road dust resuspended by automobile or construction, road salt and all other sources
12 of urban coarse particles.
13 The chemical composition of the excess fine particle concentration over the
14 Shenandoah non-urban background is also shown in Figure 6-59b. Fine particle organics
15 dominate the urban excess chemical mass balance, ranging between 1 ^g/m3 during the
16 summer, and 5.5 jug/m3 during the winter. The seasonality of excess organics also drives the
17 seasonality of excess fine mass. There is an excess sulfate concentration of 1 to 2 /ug/m3 in
18 Washington, DC, except during July, August, and September. In fact, in August in
19 Washington, DC sulfate concentration is about 0.3 /xg/m3 below the Shenandoah values. The
20 urban excess soot concentration is 1 to 2 /ig/m3 throughout the year. The soil contribution to
21 the fine particle mass is identical to the values of the Shenandoah National Park, yielding
22 virtually no excess fine soil contribution in the urban area. It is worth noting, however, that
23 there is a modest excess coarse concentration of 2 to 5 /*g/m3 in Washington, DC
24 (Figure 6-59). This indicates that soil dust aerosol components that make up the coarse mass
25 are large particles without appreciable mass below 2.5 /mi.
26 The short-term fine mass concentration at Washington, DC and Shenandoah National
27 Park for the year 1992 is shown in Figure 6-60. Although the sampling is conducted
28 Wednesdays and Saturdays for 24-hours, the data points are connected as a guide to the eye.
29 The figure also compares the daily (Wednesdays and Saturdays) fine particle sulfur
30 concentrations at the two monitoring sites. The fine mass concentration time series shows
31 elevated concentrations (>30 /xg/m3) that occur throughout the year. On the other hand,
April 1995 6-98 DRAFT-DO NOT QUOTE OR CITE
-------
fl>
o
1
1
Lai
o.
z
«
o
c
ro
c
«
£
CO
1
U
o
c
o
ri
O)
"?-
— i 1 : 1 I i i TV i na — i — *n
' X I
n 4 '• -
* q V -
v n ii -i
'"--. \ '/^
* - ^ p-j \ j
- ^;. |
\& -
' \
f 'ffl'
* ' ^-f
'* ' ' !
' , '
£- 0 < ..• -
-' / ' ('
<> ffl < f .
/ ' [ / ">
* m «/•••'' -
/ f
/ ? fH
1 1 1 1 1 1 1 l.-x" 1 In,' 1 -JJ I
ooooooooooooc
oooooooooooo
oooooooooooo
U
0)
0
g
z
ts
o
g^
U
M
B)
3
_
3
c
3
"5
S1
a.
<
b.
(0
^
0)
LL
S
> 5
(/)
I
M
to *-
? §
O CO
q: 9
M
M
(0
r
£
3 "5
-h 4-
I
I
«
o
"O
c
«
c
a>
-a
c
•a s
a> o
S "3
« ^
il
« So
U fe
P ,T
O)
o
I
S
10
0>
U
C
%
I
o
O
c
U)
10
10
o o
o o
o o.
o
\o
.
10
^
§o
o
o o
8 8
o
o
o
.5
J3
18!
a.
t
a.
g
*j PS
t« ^
W ^j
II
C ^
o
o ^
11
« U
ej
April 1995
6-99
DRAFT-DO NOT QUOTE OR CITE
-------
1 high fine mass levels at Shenandoah are recorded only during the summer season.
2 Paniculate sulfur concentration at the urban and non-urban site are comparable and co-vary
3 throughout the year. This indicates that paniculate sulfur is part of the regional airmass that
4 at any given day influences both Washington, DC and Shenandoah. Fine particle mass, on
5 the other hand, shows and excess concentration at Washington, DC, particularly during the
6 winter months. Visual inspection of the fine mass daily time series clearly indicates that the
7 concentration change from one daily sample to another can be an order of magnitude
8 different. Consequently, most of the concentration variance is due to random synoptic
9 airmass changes, and to a lesser degree due to periodic seasonal variations. Further
10 concentration variance would exist if hourly data were available.
11
12 6.5.1.4 New York City, NY
13 The New York City metropolitan area is characterized by high population density,
14 moderate industrial activity, and relatively flat terrain. The PM10 concentration over the
15 metropolitan area is shown in Figure 6-6la. The circles in the map show the locations of the
16 monitoring sites and the magnitude of each circle is proportional to the average PM10
17 concentration at that site using all available data. The observed average concentrations
18 change by about of factor of two to three from one location to another. Higher average
19 concentrations tend to occur near the center of the metropolitan area.
20 The long-term trend of PM10 averaged over the New York City metropolitan area
21 (Figure 6-61b) shows a decline from about 35 /*g/m3 in 1986 to 25 jiig/m3 in 1992. This
22 PM10 reduction is comparable to the reduction over the entire Northeastern region. The
23 average seasonal pattern over the New York City metropolitan area (Figure 6-61c) is 25 to
24 30 /ig/m3 throughout the year, but rises to about 40 /ig/m3 in July.
25 The seasonal pattern at three different individual monitoring sites in the New York
26 City metropolitan area is shown in Figure 6-62. The three sites all show similar seasonally
27 with a summer peak, but with elevated concentrations closer to the city center.
28 Size segregated aerosol samples in New York City (Figure 6-62e,f) show that at both
29 sites, PM10 concentrations are contributed primarily by fine particles. Furthermore, the
30 seasonality at the Manhattan site shows two peaks, the summer peak and the winter peak.
31 Based on the discussion of the more extensive Washington, DC (Section 6.5.1.2)
April 1995 6-100 DRAFT-DO NOT QUOTE OR CITE
-------
M
re
•
•K
re
a.
>a
c
o
£.
re
o
-
o £:
13
§
•a
o
•a
I
O)
T3
O
|D
1
I
C/J
U 4J
o re
§-c
re
i1 I
i- re
in a)
•p
V}
en
i
c
_o
'•£
2
I
>>
^
U)
£tu/6u
r,
April 1995
6-101 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10-1985-1993
AIRS PM10 Concentrantion Trends
New York
60,- . . . . , . ,_
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PW110 SEASONAL CONCENTRATION
Washington-New York Megalopolis
P )r
5H!
60
55
50
45
40
35
30
25
20
16
10
6
1985 Mar Apr May Jun Jul Aug 8«p Oet N«v D«c
Figure 6-61.
Aerosol concentration map, trend and seasonality in the New York
City region.
April 1995
6-102 DRAFT-DO NOT QUOTE OR CITE
-------
Mar May Jul Sep Nov
NEW YORK CITY, NY - Manhattan
£]- =PM10_AV360610010 NEWYORK CITY
- =PM10_AV360610069 NEWYORK CITY
NEW YORK CITY, NY
90
60
70
60
50
40
30
20
10
19
-dr
-^ 90
80
70
60
50
A
^A / v, -. 40
f" "S ** Q - J"
., ------ -'• "' -'- - - ,. -4___. 10
34 Mar May Jul Sep Nov 19
=PM10_AV360610056 NEW YORK CITYj [^r
-
-
-
-
' /A '
^A A^/a-QX\X^\ /PN
^ -B- -ef XQ^ B^ H
34 Mar May Jul Sep Nov
=PM10_AV 360610079 NEWYORK CfTYj
-Q- =PM26_AV 360610056 NEW YORK CITY
" -PMC AVG36061005fiNPWYnRkrm
-Q- "PM26JW360610079 NEW YORK CPTY
=PMC AVR 3«ofiion7a NFw YORK rrn
Figure 6-62. Fine, coarse and PM10 particle concentration near New York City.
April 1995
6-103 DRAFT-DO NOT QUOTE OR CITE
-------
1 measurements it can be inferred that the summer peak in the fine mass is mainly due to the
2 regional haze aerosol, while the winter peak is contributed by the local sources, confined to
3 the inner metropolitan area.
4 As part of the New York Summer Aerosol study (Leaderer et al., 1978) continuous
5 size monitoring confirmed the expected bimodal volume distribution with modes between
6 0.1 to 1.0 /mi and over >3.0 /mi. A number of interesting patterns were observed when the
7 size distribution was averaged by hour of day. The diurnal average total number
8 concentration showed a pattern which corresponded closely with the normalized diurnal
9 traffic pattern. Particles <0.1 /mi showed the most marked diurnal variation, following the
10 total number curve. Moreover, particles in size ranges >0.1 /mi showed little variation in
11 the diurnal pattern. Analysis of samples processed by the diffusion battery indicated that
12 approximately 54%+ 18% of the sulfate measured was in the suboptical range (approximately
13 0.04 /mi to 0.3 /mi) with the remainder above 0.3 /mi. Little sulfate mass was found in
14 particles in the nuclei range (<0.04 /mi). Analysis of impactor samples for sulfates
15 consistently showed that more than 85% of all water soluble sulfates were <2.0 jam in size.
16 Virtually no nitrate was present in the nuclei size range while the suboptical size range
17 accounted for approximately 30% of the total nitrate. 70% of the total nitrate was found in
18 the size range >0.3 /mi. Analysis of large stages of Anderson impactor showed that
19 approximately 50% of particulate nitrate was greater than 5.5 jum in size.
20 Urban and rural particulate sulfur monitoring near New York City in the summer
21 (Leaderer et al., 1982) indicated that sulfate concentration distributions were regionally
22 homogeneous and increased with increasing ozone levels and covariant with several other
23 pollutant and meteorological parameters. Sulfate concentrations correlated strongly with
24 ammonium and strong acid at all sites. No significant diurnal patterns for sulfate or
25 ammonium were seen at any site for low and high ozone levels. Strong acid concentrations
26 were highest at the rural and semi-rural sites, lowest at the urban sites, increased with
27 increasing ozone levels and exhibited diurnal patterns which matched the ozone diurnal
28 patterns.
29 Sulfate acidity measurements (Waldman et al., 1991) at Chestnut Ridge, PA (east of
30 Pittsburgh) suggest that higher acidity occurred in the overnight period (0000 to 0800) in the
31 late fall, while sulfate had its highest levels in the morning to afternoon period. Size
April 1995 6-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 dependent, mass and composition of New York aerosol for low, medium, and high visibility
2 levels was reported by Patterson and Wagman, 1977. At all levels of visibility, bimodal or
3 multimodal particle size distribution were observed for total mass and for individual
4 components. Decreased visibility corresponded to increased particle mass concentrations
5 especially in the fine particle fraction (ranging in size from about 0.1 to 1.00 MM).
6 Increases in the proportion of particulate sulfate and to a lesser extent of nitrate, chloride,
7 ammonium, and carbon were also associated with decreased visibility.
8 Aerosol pattern analysis of a major wintertime (1983) pollution episode in northern
9 New Jersey (Lioy et al., 1985) revealed that the intensity of the episode was the greatest in
10 the area of highest commercial, residential and industrial activities, and that the atmospheric
11 stagnation conditions resulted in the significant accumulation of aerosol mass. The aerosol
12 mass was primarily fine aerosols, and the extractable organic matter compromised about 50%
13 of the particle mass.
14 An analysis of the winter aerosol chemistry data in the Northeast (Poirot et al., 1990)
15 found that the winter PM2 5 apportionment consisted of 48% sulfates, 23 % organic matter,
16 15% soot, 4% soil, and 10% unexplained.
17
18 6.5.1.5 Philadelphia, PA
19 The metropolitan area of Philadelphia includes urban-industrial emissions over flat
20 terrain. The map of the area shows relatively uniform PM10 concentrations throughout the
21 metropolitan area, with the exception of a single site (AIRS #421010149) in the middle of the
22 urban area.
23 The 1985 to 1992 trend of PMi0 averaged over the metropolitan area is somewhat
24 downward but not significantly. The seasonal concentration of PM10 (Figure 6-63a) is about
25 30 to 35 /ig/m3 throughout the year, except during the summer months when it rises above
26 40 jug/m3.
27 The seasonal average PM10 concentration for three sites near the center of Philadelphia
28 is shown in Figure 6-64. One of the sites is the high concentration site noted on the
29 metropolitan map. The average PM10 concentration at that site ranges between 100 to 150
30 Mg/m3 which is a factor of 2 to 3 higher than the concentration at the neighboring sites. The
31 daily concentration at this anomalous monitoring site (AIRS #421010149) correlates poorly
April 1995 6-105 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1985-1993
AIRS PM10 Concentrantion Trends
Philadelphia
1986 1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
Philadelphia
60
66
60
46
40
36
30
25
20
16
10
6
1985 Mar Apr May Jun Jul Aug Eep Oct Nsv 0«c
Figure 6-63.
Aerosol concentration map, trend and seasonality in the Philadelphia
region.
April 1995
6_106 DRAFT-DO NOT QUOTE OR CITE
-------
o\
100
90
80
70
60
50
40
30
20
10
PHILADELPHIA, PA
1984 Mar May Jul Sep Nov
|-A- -PM10_AV 421010004 PHILADELPHIA |
d- -PM25_AV 421010004 PHILADELPHIA
-4- -PMC JWG 421010004 PHILADELPHIA
250
200
150
100
50
Philadelphia - Suburban 15 km Separation
ilL
50
100
150
200
PhiladelpNa - Urban, 4km Separation
200
150 ••
100 ••
50
0
4PMKM210100382
200
250
Figure 6-64.
Fine, coarse and PM10 particle concentration near Philadelphia.
-------
1 with an adjacent site (AIRS #4210104492) which is only 4 km apart (Figure 6-64e). This is
2 an indication that the concentration at the anomalous site is under the influence of strong
3 local source of PM10. In contrast, two sites in suburban Philadelphia that are 15 km apart
4 (Figure 6-64d) show a strong correlation of daily measurements indicating that spatially
5 uniform regional haze influences the daily values at both sites.
6 Size segregated aerosol samples (Figure 6-64a) at Philadelphia (AIRS #421010004)
7 show the fine paniculate mass is the main contributor of PM10 in this city. It is possible,
8 however, that at other sampling sites, e.g. the anomalous #4210104492 site, coarse particles
9 prevail.
10 Outdoor summertime sulfate (SO42~) concentrations were found to be uniform within
11 metropolitan Philadelphia (Suh, et al 1995). However, aerosol strong acidity (H+)
12 concentrations were found to vary spatially. Also, the wintertime sulfate pattern was likely
13 to be more heterogeneous in space and time as discussed for Washington, DC This variation
14 generally were independent of wind direction, but were related to local factors, such as the
15 NH3 concentration, population density, and distance from the center of the city.
16
17 6.5.1.6 Whiteface Mountain, NY
18 The AIRS sampling location at the Whiteface Mountain in Upstate New York, is a
19 high mountain top site, elevated from the surrounding terrain. The monitoring site offers the
20 possibility of comparing mountain tops concentrations to the surrounding lower elevation
21 sites.
22 The seasonal pattern of PM10 concentration for Whiteface Mountain and the
23 surrounding low elevation sites, Saranac Lake, and Saratoga Spr. are shown in Figure 6-65.
24 The concentration at the three sites is virtually identical during June through September.
25 However, during the winter the mountain top site at Whiteface has a PM10 concentration
26 which is only one third of the surrounding sites. This indicates that during the winter, the
27 Whiteface mountain top is above the surface-based aerosol layer, while during the summer
28 the height of the well mixed aerosol layer rises above the mountain top producing are
29 resulting in a uniform concentration at all sites. At Whiteface Mountain, NY (Webber et al.,
30 1985) have found direct microscopic evidence of flyash particles during a summertime (June
31 23, 1983) sulfate episode.
April 1995 6-108 DRAFT-DO NOT QUOTE OR CITE
-------
14
13C
12(
110
100
90
80
70
60
50
40
30
20
10
PM10 -Average
& B s-'s-^a-
0
1984
Mar May Jul Sep Nov
£1 I=PM10_AV360330002SARANACLAKE
H =PM10_AV 360910002 SARATOGA SPR
Figure 6-65. PM10 concentration seasonality at Whiteface Mountain and
neighborhing low elevation sites.
1 6.5.2 Sub-regional Aerosol Pattern in the Southeast
2 6.5.2.1 Winston-Salem, NC, Florida
3 Winston-Salem is a small urban area imbedded in the relatively homogeneous region
4 of the Southeast. In fact, the main emphasis of the illustrations below is on the remarkable
5 uniformity of aerosols in that region.
6 Comparison of three AIRS PM10 monitoring sites in North Carolina, Winston-Salem,
7 Greensboro, and Raleigh-Durham (Figure 6-66) shows virtually identical concentrations
8 (within 10%), both in absolute magnitudes, and in the seasonality. This is an indication that
9 these sites are exposed to the same regional airmass throughout the year. It also indicates
10 that the excess PM10 concentrations due to local urban sources are insignificant.
April 1995
6-109 DRAFT-DO NOT QUOTE OR CITE
-------
o
150
140
130 -
120 -
110 -
100
90
80
70
60
50
40
30
20
10
1984 Mar May Jul Sep Nov
-A- -PM10JW 120950007 ORLANDO
-H - "PM10_AV 120570095 TAMPA
[-4. -PM10_AV 120251016 MIAMI ~~|
-tr -PM10.AV 370670009 WINSTON-SALEf
•3 -PM10_AV370810009 GREENSBORO
-- •PMib_AV37f830003~RALEIGH~ ~
W1NSTON-SALEM, NC
100
90
80)-
70
60
SO
40
30
20
10
1984 Mar May Jul Sep Nov
\-tr 'PM10JVV 370670009 WINSTON-SALE(
-B- -PM2S.AV 370670009 WINSTON^ALEI
- -PMC.AVG 370670009 WINSTON-SALE
Figure 6-66.
Aerosol concentration pattern at North Carolina and Florida sites.
April 1995
6-110 DRAFT-DO NOT QUOTE OR CITE
-------
1 Size segregated monitoring data for Winston-Salem show that fine particles contribute
2 70 to 80% of the PM10 mass of 25 to 35 /*g/m3. Coarse particles are seasonally invariant at
3 about 10 /-ig/m3 which is typical for eastern US.
4 The PM10 concentration at monitoring sites in Florida (Orlando, Miami, Tampa) show
5 virtually identical concentrations ranging between 25 to 30 pig/m3 throughout the year,
6 without appreciable seasonality.
7
8 6.5.2.2 Large Southeast Metropolitan Areas
9 The relative homogeneity of southeastern aerosol concentrations is consistent with the
10 observations at the regional perspective. However, the concentrations in large metropolitan
11 areas such as Atlanta, GA, Birmingham, AL, Mobile, AL, New Orleans, LA, and Houston,
12 TX, Dallas-Fort Worth, TX, show somewhat elevated values. It is evident, however, that
13 these southeastern urban areas do not have high concentration hot spots.
14 The seasonal PM10 concentration at sites in New Orleans, LA, Mobile and
15 Birmingham, AL show uniformity (20 to 40 jug/m3) with modest seasonality. At sites in
16 Houston, Austin, and San Antonio, TX show similar values, but more spread in
17 concentration.
18 The size segregated aerosol samples collected in the cities of the Gulf states, Corpus
19 Christi, Forth Worth, Houston, TX and New Orleans, LA all show that fine particle
20 concentrations are relatively low (10 to 20 /ig/m3) and are seasonal (Figure 6-67). Coarse
21 particle concentrations, on the other hand, account for more than half of the PM10 mass and
22 their contribution is most pronounced during the summer season.
23 In Houston, TX Dzubay et al., 1982 found that in summertime fine particle mass
24 contained 58% sulfate and 18 % of carbonaceous material. They also found that the coarse
25 fraction (2.5 to 15 /*m) consisted of 69% crustal matter, 12 % carbon, and 7 % nitrate
26 species.
27 Characterization of the Atlanta area aerosol (Marshall et al., 1986) show that
28 elemental carbon and paniculate sulfur represent, respectively 3.1 to 9.9% and 1.9 to 9.4%
29 of the total suspended particulate mass. The concentration of elemental carbon, sulfur, and
30 TSP exhibit strong seasonal variations, with elemental carbon decreasing from winter to
31 summer, and sulfur and TSP increasing. Elemental carbon appears to be statistically separate
April 1995 6-111 DRAFT-DO NOT QUOTE OR CITE
-------
1 from sulfur, indicating that the sources for elemental carbon and paniculate sulfur are
2 distinct.
3
4 6.5.2.3 Great Smoky Mountains
5 Size segregated fine and coarse aerosol concentrations were measured at the Great
6 Smoky Mountains National Park in September of 1980 (Stevens et al., 1980). Sulfate and its
7 associated ions contributed to 61% of the fine particle mass, followed by organics (10%) and
8 elemental carbon (5%).
9
10 6.5.3 Sub-regional Aerosol Pattern in the Industrial Midwest
11 The chemical composition of summertime (July, 1981) fine particles in Detroit (Wolff
12 et al., 1982) was found as 52% sulfates, 27% organics, 4% elemental carbon, 8% soil dust
13 Nitrate was found to absent from fine mass. Fine particles, themselves contributed about
14 64% of the aerosol mass balance.
15 Since the turn of the century, the major cities in the industrial midwestern states had
16 air pollution problems due to smoke and dust. Pittsburgh, St. Louis, Chicago, and Detroit
17 were among the formerly notorious air pollution hot spots. The recently acquired PM10
18 database now allows the re-examination of these metropolitan areas in the industrial Midwest
19 for their concentration pattern in the 1990s.
20
21 6.5.3.1 Pittsburgh, PA
22 The average PM10 concentration over the extended metropolitan area is shown in
23 Figure 6-68a. The map also includes the industrial cities in the eastern Ohio, Steubenville,
24 OH, and Weirton, OH, located on the Ohio River. The average PM10 concentration at the
25 80 sites shown on the map varies only by about 20% from site to site. Outstanding high
26 concentration hot spots are also absent. It is thus evident that during the 1985 to 1993
27 period, the PM10 concentrations in the Pittsburgh sub-region was spatially rather uniform.
28 The PM10 concentration trend shows declining values from about 40 /ig/m3 to 30
29 Mg/m3 (Figure 6-68b) over the 1985 to 1993 period. Figure 6-68b also marks the
30 concentration standard deviation among the monitoring sites for each year, which is about
31 15 to 20% and shrinking over time.
April 1995 6-112 DRAFT-DO NOT QUOTE OR CITE
-------
13
2.
vo
0\
H
m
n
i—i
s
160
140
130
120
(10
100
9C
80
70
60
60
40
30
20
1
. • •
-
-
-
A
-_^_V \.
' JJr--"^ "^ a •1-VX*\__A_^
- ' ' V ^^ * i-"*""1
i B a' B """* a
. . . _. —
16O
140
IX
120
110
100
90
80
70
60
60
40
30
20
10
0
•
•
-
•
.
•
•
«- 4_ *"*\' •*
i__J,*^M*-*r" ^=^^
.
1984 M«r M«y Jul !«P
4r -PM10.AV 482011038 HOUSTON
-O -PM10.AV4846300W AUSTIN
1984
HOUSTON, TX
Mir Miy Jul Sep
A- -PWtO_AV220710010 NEW ORLEANS
a -PM10.AV010970016 MOBILE
' H -PMiojWOIoYjjOMBiRMINCJHAM \
NEWOLRLEANS, LA
1M4 M.r M«y Jut S.p Nov
|T!r -PM10JW493660012 CORPUS CHRIS1
•El- -PM2t^V 483660012 CORPUS CHRIS"
.+ -PMCJW6483S60012 CORPUS CHRIS
•-tr -PM10JW 464390060 FORT WORTH .
D "PM26_AV 4843*0060 FORT WORTH
•I -PMC AVG 484390060 FORT WORTH
M«V Jul
-6- -PM10_AV 462010024 NOT IN A CITY
0 -PM2S_AV 482010024 NOT IN A OTY
-, -PMC_AVG 482010024 NOT IN A CfTY
1984 Mir M»y Jul Scp Nov
-ft- -PM10_AV220TiboiO NEW ORLEANS
-f3 -PM26.AV 220710010 NEW ORLEANS
"PMC_AVG 220710010 NEW ORLEANS
Fisure 6-67. Aerosol concentration pattern in Texas and Gulf states.
6
-------
AIRS PM10 Concentration Trends
Pittsburgh
w
55
60
45
0 «°
O ^ 36
'Sis 30
;is
kg 26
3 20
16
10
5
_
-.---•K
' " " , - +-
'"*"''
' *\ ' ""- '*
\»_ __ A "*''' "'-.
>-~D "X. _-^^
^ ^ ^^^A— -^>
\ /
^-^t/ VV^"Q^--Q--:I
-
-
-
-
1985 1986 1987 1988 1989 1990 1991 1992
j"-A- PM10AVG -Q- PM10AYG-SIG -+ PM10AVG + BIG
AIRS PIU110 SEASONAL CONCENTRATION
Pittsburgh
1986 Mar Apr May Jun Jui Aug Sep Oct Nov Dec
Figure 6-68. Aerosol concentration pattern and trends in the Pittsburgh subregion.
April 1995
6-114 DRAFT-DO NOT QUOTE OR CITE
-------
1 The seasonality of the PM10 pattern (Figure 6-68c) is dominated by a summer peak
2 (45 pig/m3), which is about 50% higher than the winter concentrations (30 jug/m3). The
3 seasonality at sites in Pittsburgh, PA, Weirton, OH, and Steubenville, OH show comparable
4 values and a seasonal pattern that is slightly above the sub-regional average. Hence, these
5 formerly highly polluted locations are now virtually identical to their rural background.
6 Size segregated aerosol samples in Pittsburgh, PA and Steubenville, OH (Figure 6-69)
7 show that fine particles contribute 70 to 80% of the PM10 mass, and also dictate the summer-
8 peak seasonality of the PM10 concentrations. As in other urban monitoring sites in the
9 eastern US, the coarse particle concentration in Pittsburgh is about 10 pig/m3 and seasonally
10 invariant. The size segregated seasonal data for Steubenville, OH exhibit more random
11 fluctuations as well as discrepancy between the sum of fine and coarse at one hand, and
12 PM10 on the other. The discrepancy is attributed to the small number of size segregated
13 aerosol samples.
14 The remarkable uniformity of fine particles mass and elemental composition from site
15 to site in the Ohio River Valley was also shown by Shaw and Paur, 1983. Sulfur was the
16 predominant element in fine particles. Factor analysis of element concentrations indicated
17 three clusters throughout the year (1) coarse particle crustal elements (2) fine particle sulfur
18 and selenium (3) fine particle manganese, iron and zinc.
19 The chemical mass balance of Weirton-Steubenville aerosol was examined by
20 Skidmore et al., 1992. Primary motor vehicle and secondary ammonium sulfate were
21 dominant contributors to the PM2 5 aerosol. Steel emissions were also significant
22 contributors to PM2 5. Wood burning and oil combustion were occasionally detected.
23 Geological material was the major contributor to the coarse aerosol fraction. Primary
24 geological material, primary motor vehicle exhaust, and secondary sulfate were the major
25 contributors to PM10 at all five monitoring sites.
26 The composition of size-fractionated summer aerosol in nearby Charleston, West
27 Virginia was reported by Lewis and Macias, 1980. Ammonium sulfate was the largest single
28 chemical component (41%) of the fine aerosol mass. Carbon was also a large component of
29 both fine and coarse particle mass constituting 16% and 12% respectively. Factor analysis
30 indicated that four factors were sufficient to satisfactorily represent the variance of
April 1995 6_H5 DRAFT-DO NOT QUOTE OR CITE
-------
VO
-PM10_AV 420030031 PITTSBURGH
-U -PM10_AV 640291004 WEIRTON
-H^PMlb~AV''39081l6l2STEUBENVILLE!
STEUBENWILLE, OH
PITTSBURGH, PA
PITTSBURGH, PA -2
o\
1
^ 90
80
70
S «o
S
£ 60
H
i 40
O
O 30
o
H
^ 10
0
(~~^ Q
-
-
-
.
tf^eS*~~*~~\
~^* \
ft~~T5r"^~~^iSx^*' \ /
a, ^«^-
i / \ A i
' ^ - + ^Q •"*" ^4. - *^ • -•• \y •* Vi ''
- B • ^S— / ' Q , ' *
+ ""'"•--f'^Q 4 V
IUU
90
80
70
60
50
40
30
20
10
A
/ q \
y / \ v.
»^^^ 'B \ ^N/^
i. ^~* • , » n -£
TS . n n- H /
s- B , \ / -
.(• - t^
-1- •+ '• +
••-!+• ' s...-*-' -
1 UU
90
80
70
60
60
40
30
20
10
n
-
-
_
/^-
X ^^
^^~S ^*~^ ^^^^
^^ ^ ^ "^
'- -0- » B' ^ & -e -*
. -4-..4. -1- -->•-+-•+• -4..^..
--•+ 4 -S--.+ -'
O 198'1 Mar May Jul Sep Nov 1984 Mar May Jul Sep Nov 1984 Mar MaV J"1 Sep Nov
H
t"d . -6- -pMio_AV39oeiioi2 STEUBENVILLE; -A- -PMIO_AV 420030021 PITTSBURGH . -A- »PMIO_AV 420030027 PITTSBURGH '
O Q "PM2S_AV 390811012 STEUBENVILLE -Q- -PM2S_AV 420030021 PITTSBURGH -B =PM25_AV 420030027 PITTSBURGH
_ H -PMC AVG 39081 1012 STEUBENVILLE -4- -PMC AVG 420030021 PITTSBURGH -+ =PMC Av(5 420030027 PITTSBURGH
n ~
W Figure 6-69. Fine, coarse and PM10 concentration near Pittsburgh.
-------
1 26 measured parameters. The factors were characteristic of crustal material, ammonium
2 sulfate, automotive emissions, and unidentified anthropogenic sources.
3
4 6.5.3.2 St. Louis, MO
5 Historically, the St. Louis metropolitan area has been known for high particulate
6 concentrations, particularly on the east side of the Mississippi River. The map of the
7 metropolitan area (Figure 6-70a) shows about factor of 2 to 3 concentration differences
8 among the PM10 monitoring stations. The monitoring sites east of the river tend to be higher
9 than the western sites of this sub-region.
10 The average PM10 in the St. Louis metropolitan area (Figure 6-70b) has been
11 declining from 40 to 45 ng/m3 to 25 to 30 jug/m3 by 1993. This decline is comparable to the
12 average reductions over the industrial midwestern region. The seasonality of the
13 sub-regionally averaged concentrations (Figure 6-70c) shows the summer peak with 40 to 50
14 Mg/m3 which is about 50% higher than the winter averages.
15 Seasonal comparison of the individual monitoring sites in the area shows that Granite
16 City, IL and East St. Louis, IL have higher PM10 concentrations throughout the year
17 compared to western St. Louis, MO sites.
18 Size segregated aerosol samples at three sites west of the Mississippi River (Ferguson,
19 MO, Affton, MO and Clayton, MO) show that fine particles are mostly responsible for
20 PM10, including the seasonality (Figure 6-71). Coarse particles contribute 10 jig/m3 or less
21 throughout the year, although corresponding size segregated aerosol data for more polluted
22 eastside of the Mississippi River are not available.
23 Monitoring the diurnal and seasonal patterns of particulate sulfur and sulfuric acid in
24 St. Louis (Cobourn and Husar, 1982) noted an afternoon increase in particulate sulfur
25 concentration of about 20%. For the summertime, particulate sulfur concentration was
26 higher than the annual mean by 40%.
April 1995 6_117 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1986-1993
AIRS PM10 Concentrantion Trends
St Louis
1986 1987 1988 1989 1990 1991 1992
AIRS PWI10 SEASONAL CONCENTRATION
1985 Mir Apr M«y Jun Jul Aug S«p Oct N»v Dec
Figure 6-70.
Aerosol concentration pattern and trends in the St. Louis subregion.
April 1995
6-118
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.5.3.3 Chicago, IL
2 Chicago has also been known for industrial dust, smoke, and haze, particularly in East
3 Chicago and Gary, IN. The average PM10 concentrations over the Chicago sub-region
4 (Figure 6-72a) shows that concentrations vary by a factor of two or less throughout the
5 subregion. The downward trend (from 40 to 25 /ig/m3) is comparable to the PM10
6 reductions over the industrial Midwest. The seasonality of PM10 is also typical with the
7 summer peak of 40 /-ig/m3 and winter values of 20 to 30 />ig/m3.
8 Superposition of seasonal PM10 data at Chicago, IL, East Chicago, IL, and Gary, IN
9 conveys a remarkable spatial uniformity, as well as comparatively low PM10 concentrations
10 in this area that has historically been a smoky and dusty industrial sub-region.
11 Chemical composition measurements in Chicago (Lee et al., 1993) showed that me for
12 an concentrations for SO42' (5.55 /*g/m3), NH4+ (2.74 ^g/m3), NH3 (1.63 /xg/m3),
13 HNO3(0.81 ng/m3), HNO2 (0.99 pig/m3), for SO2 (21.2 pcg/m3), for NO3- (4.21 /ig/m3), and
14 for H+ (7.7 nmol/m3). The highest values occurred in the summer, except for HNO2 and
15 NO3" which had the highest values in the winter.
16 Comparison of atmospheric coarse particles at an urban and non-urban site near
17 Chicago, IL show that the concentration were 50% higher during mid-day than at night. Dry
18 ground samples were 30 % higher than wet ground and 90% higher than frozen ground
19 samples. (Noll et al., 1985).
20 The analysis of coarse particles in Chicago, IL (Noll et al., 1990) show that the coarse
21 particle mass could be divided into two categories: material that was primarily of crustal
22 origin (Al, Ca, Fe, and Si) and material that was primarily of anthropogenic origin (Cd, Cu,
23 Mn, Ni, Pb, and Zn). The mass of crustal material varied between 15 and 50% of the total
24 coarse mass, while the mass of anthropogenic material was < 1%.
25 The composition of atmospheric coarse particles at urban (Chicago, IL) and non-urban
26 (Argonne, IL) were reported by Noll et al., 1987. Limestone and silicates were the main
27 source of material at the non urban site. Anthropogenic sources, represented by fly ash and
28 coal, were present in the industrial sector sample and rubber tire was present in the
29 commercial sector sample. The mass median diameters (MMD) for different components
30 were as follows: limestone (20/mi, silicates 12jum, coal, flyash and iron oxide (12^m) and
31 rubber tire (25/mi).
April 1995 6-119 DRAFT-DO NOT QUOTE OR CITE
-------
150
140
130
120
110
90
80
70
60
40
30
20
10
1984 Mar May Jul Sep
-&- -PM10_AV29S100080STLOUIS
Nov
-=PM10_AV171S30010 EAST ST LOUIS
ST LOUIS, MO
ST LOUIS, MO -Clayton
ST LOUIS, MO -Affion
1984 Mar
^A- -PM10_AV29189500l"FERGUSON
-B- »PM25_AV291895O01 FERGUSON
-4- -PMC.AVG 291895001 FERGUSON
-PM10JW 291692003 CLAYTON
-Q- -PM25_AV 291892003 CLAYTON
•+ "PMC.AVG 291892003 CLAYTON
[-A- •PM10_AV29189000fgFTQJr
-Q- -PM26_AV291890001 AFFTON
•4- "PMC_AVG 291890001 AFFTON
Figure 6-71.
Fine, coarse, and PM10 concentration pattern near St. Louis.
April 1995
6-120 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1985-1993
o
^ 01
< ui
O UJ
s*
O- P
150
140
130
120
110
100
90
80
70
60
60
40
AIRS PM10 Concentrantlon Trends
Chicago
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
Chicago-Gary
60, . . , . . . -. r- . , ,
1984 Mar May Jul Sep Nov
-A- =PM10_AV 170310014 CHICAGO
-fl "PM10_AV 180890006 EAST CHICAGO
i + -=PM10_AV 180890022 GARY |
55
50
45
40
35
30
26
20
15
10
6
1985 Mar Apr May Jun Jut Aug Sep Oet Nov D«e
Figure 6-72.
Aerosol concentration pattern and trends in the Chicago subregion.
April 1995
6-121
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.5.3.4 Detroit, MI
2 Detroit is the center of the automotive industry and has been studied by several
3 investigators.
4 In Detroit, July, 1981 (Wolff and Korsog, 1985) fine mass average was found to be
5 42.4 /ig/m3. A major contribution (50%) of sulfate source which appears to be coal
6 combustion was identified. The coarse fraction which averaged as 25.8 /ig/m3 was
7 dominated by crustal material which accounted for about two-thirds of the coarse material.
8 Significant contributions were also identified from motor vehicles (mostly due to reintrained
9 road dust) and iron and steel industry emissions.
10 The seasonal variations in nitric acid, nitrate, strong aerosol acidity, and ammonia in
11 urban are, Warren, MI was examined by Cadle, 1985. The greatest variations was for
12 ammonia, which was 8.5 times higher in summer than winter. The least variation was for
13 paniculate nitrate which had a summer maximum only 1.8 times higher than in spring
14 minimum. It was noted that ammonium nitrate volatilization from filters and impactors can
15 cause large errors in summertime measurements, but the errors are not significant during the
16 winter.
17 The influence of local and regional sources on the concentration of particulate matter
18 in urban and rural sites near Detroit, MI was investigated by Wolff et al., 1985. Analysis of
19 spatial variations of the various particulate components revealed: (1) at all four sites the
20 PM2 5 was dominated by regional influences rather than local sources . The site in industrial
21 sector had the largest impact of local sources, but even at his site the local influences appears
22 to be smaller than he regional ones. (2) the regional influences were most pronounced on the
23 sulfate levels which accounted for the largest fraction (40-59=0%) of the PM2 5. (3) organic
24 carbon compounds were the second most abundant PM2 5 species accounting for 20 to 49%
25 of the mass. Organic carbon seems to be controlled by both local and regional organic
26 carbon influences. Vehicular emissions and possibly secondary reactions appear to affect the
27 organic carbon concentrations (4) elemental carbon appears to be dominated by local
28 emission (5) PMCoarse was dominated by local sources, but at the industrial site unknown
29 non-crustal elements were significant components of coarse mass.
April 1995 6-122 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.5.5 Sub-regional Aerosol Pattern in the Southwest
2 Southwest is a dusty region and much of the discussion below pertains to coarse
3 particles and soil dust.
4 The arid southwestern U.S. metropolitan areas (El Paso, TX, Phoenix-Tucson, AZ)
5 has modest industry and national parks (grand Canyon) where the prevention of visibility
6 degradation has been stated as a national goal.
7
8 6.5.5.1 El Paso, TX
9 The PM10 concentration in the El Paso, TX subregion shows that the high and low
10 concentration sites can be found in close vicinity of each other (Figure 6-73a). This is an
11 indication that local sources of PM10 with limited range of impact are important. The PM10
12 trend since 1985 (Figure 6-73b) shows a remarkable reduction from 60 /xg/m3 to 30 /xg/m3,
13 although the downward trend is not monotonic. This substantial reduction parallels the factor
14 of two PM10 decline over the entire southwestern region.
15 The seasonality of PM10 over the El Paso, TX sub-region (Figure 6-73c) is bimodal
16 with peaks in the spring time, March through July, as well as another peak, October through
17 November. This double peak seasonality at El Paso, TX also parallels the seasonality of the
18 entire region. The double peak is further illustrated by superimposed seasonal charts for
19 Tucson, AZ, Albuquerque, NM and El Paso, TX. It is evident, that these three widely
20 spaced sampling sites all show the concentration reduction in August which coincides with
21 the arrival of moist "monsoon" flow from the Gulf of Mexico toward Arizona.
22 Size segregated aerosol samples for El Paso, TX (AIRS #481410037) shows that
23 coarse particles dominate the PM10 concentrations, accounting for about 70% of the PM10
24 mass (Figure 6-74a). This is consistent with the important role of coarse particles over the
25 arid Southwest. In comparison, size segregated data for San Antonio, TX (Figure 6-74b)
26 located closer to the Gulf Coast in Texas, shows that fine and coarse mass have comparable
27 contributions, similar to Houston, TX.
28
29 6.5.5.2 Phoenix and Tucson, AZ
30 The Phoenix-Tucson sub-region (Figure 6-75a) shows substantial PM10 concentration
31 range. Samplers within Phoenix or Tucson area indicate 2 to 3 times higher concentrations
April 1995 6_123 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1985-1993
AIRS PM10 Concentrantion Trends
El Paso
1986 1387 1988 1989 1990 1991 1992
AIRS PWI10 SEASONAL CONCENTRATION
El Paso
80, . . . , . . , , . ,
3!
|
75
70
65
60
55
60
45
40
35
30
25
20
15
10
5
1985 Mar Apr May Jun Jul Aug s«p Oet Nov D«e
Figure 6-73. Aerosol concentration pattern and trends in the El Paso subregion.
April 1995
6-124 DRAFT-DO NOT QUOTE OR CITE
-------
100
90 -
80 -
70
60
50
40
30
20
10
EL PASO, TX
SAN ANTONIO, TX
H --Q
3"
5-
1984 Mar May Jul Sep Nov
-A- =PM10_AV481410037ELJPASO
-O- =PM25_AV 481410037 EL PASO
~4- =PMC AVG 481410037 EL PASO
100
90
80
70
60
50
40
30
20
10
1984 Mar May Jul Sep Nov
-A- =PM10_AV 480290036 SAN ANTONIO
-ED- =PM25_AV480290036 SAN ANTONIO
-(- =PMC_AVG 480290036 SAN ANTONIO
Figure 6-74. Fine coarse, and PM10 concentration pattern near El Paso.
1 than the more remote sites, particularly the ones in the mountains. There is a general decline
2 of PM10 level between 1985 and 1993, but it is in the presence of substantial year to year
3 variation (Figure 6-75b). The average PM10 seasonality of the Phoenix-Tucson sub-region
4 (Figure 6-75c) shows the bimodal spring and fall peak pattern which is characteristic for the
5 entire Southwest region.
6 The wintertime aerosol chemical pattern in Phoenix was reported by Chow et al.,
7 1990. Solomon and Moyers (1986). They found fine particle crustal species, sulfates,
8 nitrates, and organic and elemental carbon to be at least five times higher hi concentration
9 when comparing samples during a haze episode to samples taken during good visibility.
April 1995
6-125
DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1986-1993
AIRS PM10 Concentrantion Trends
Phoenix-Tucson
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
Phoenix-Tucson
cr
10_AVG
/1ETER[25
1
Ol/
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
_
-
-
-
-
.
-
x/^^Xv>/'^
~
~
.
.
_
1888 Mar Apr May Jun Jul Aug S«p Oct Nev Dee
Figure 6-75. Aerosol concentration pattern and trends in the Phoenix-Tucson
subregion.
April 1995
6-126
DRAFT-DO NOT QUOTE OR CITE
-------
1 During the Phoenix Urban Haze Pilot Study during the winter 1988 to 1989 (Frazier
2 C.A., 1989) a definite diurnal cycle in PM2 5 concentrations was observed. The maximum,
3 generally but not always, occurred at night, which is consistent with the meteorological
4 observations or poor dispersion and dilution.
5 A chemical characterization of wintertime fine particles in Phoenix, AZ (Solomon and
6 Moyers, 1986) showed a dominance of organic and nitrate aerosols. The composition of the
7 Phoenix wintertime haze is most like that of Denver, CO, a city which also experiences
8 wintertime inversions (Pierson and Russell, 1969; Countess et al., 1980; Groblicki et al.,
9 1981). In both cities the average measured NO3" concentrations was about 1 to 2 times that
10 of average SO42" concentration. In addition, the average SO42"concentration measured in
11 Phoenix was much lower than those observed at other locations throughout the US, but
12 similar to the regional values observed in the Southwest (Moyers, 1981).
13 Wintertime PM10 and PM2 5 chemical compositions and source contributions in
14 Tucson, AZ (Chow et al., 1992) show that the major contributors to the highest PM10
15 concentrations were geological material (>50%) and primary motor vehicle exhaust
16 (> 30%) at three urban sampling sites. Secondary ammonium sulfate, secondary ammonium
17 nitrate, and copper smelter aerosols were found to contribute less than 5% to elevated PM10
18 concentrations.
19 It is instructive to compare chemical concentrations in Phoenix with wintertime values
20 in other mountainous, arid communities (Denver (commercial), CO, Reno (commercial),
21 NV, and Sparks (residential), NV. Organic carbon (OC) at the Phoenix site was twice the
22 elemental carbon at the Denver and Sparks sites, while the OC/EC ratio was one to one at
23 Phoenix and Reno sites. Wood burning contribution at Reno site was very low. The
24 average arsenic concentrations in Phoenix was four times higher than observed in other
25 cities, which indicates the potential influence of Arizona smelters located within 100 miles of
26 Phoenix. Average sulfate levels in Phoenix were higher than they were in Denver, which
27 has less local emissions of SO2. Nitrates were major source of visibility reduction in
28 Denver. In Phoenix, nitrates were significant, but the carbonaceous species appear to have
29 much larger role in visibility impairment. The average light absorption (babs in Phoenix was
30 nearly a factor of three higher than the averages obtained in Denver.
31
April 1995 6-127 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.5.6 Sub-regional Aerosol Pattern in the Northwest
2 The mountainous northwestern United States has many aerosol regions with different
3 characteristics. The discussion below will examine South Lake Tahoe, as a case study for
4 mountain-valley difference, Salt Lake City, UT, Denver, CO, Idaho-Montana sites, and
5 several Washington-Oregon sites.
6 Dresser 1988 investigated the winter PM10 concentrations in a small ski resort town,
7 Telluride, CO and found that the street dirt and sand are major contributors, particularly
8 during the dry post snow period.
9 Wintertime aerosol characterization and source apportionment was also conducted for
10 San Jose, CA, attributing 45% of the PM10 mass to residential wood combustion (Chow
11 etal.,1995).
12 In Portland, OR, carbonaceous aerosol was found to account for about 50% of fine
13 aerosol mass (Shah et al., 1984)
14
15 6.5.6.1 South Lake Tahoe
16 South Lake Tahoe IMPROVE monitoring site is located in a in a populated area on the
17 south shore of Lake Tahoe. The Bliss State Park IMPROVE monitoring site is to the
18 northwest, elevated (700ft) and removed from the populated areas. The pair of sites
19 illustrates the populate-remote difference in aerosol pattern. The aerosol and visibility at the
20 two lake Tahoe sites were also examined (Molenar et al., 1994)
21 The concentration of all aerosol components is substantially higher on the south lake
22 shore compared to the more remote site. The seasonality and chemical composition is also
23 substantially different. The excess PM10 concentration at the S. Lake Tahoe site compared
24 to Bliss State Park (Figure 6-76) is about 5 jLtg/m3 during the warm season, May through
25 September, and it climbs to 28 /^g/m3 excess in January. The factor of five seasonal
26 modulation for valley excess PM10 is likely contributed by winter time emission sources,
27 poor dispersion compared to the summer, as well as fog, that tend to enhance the aerosol
28 formation. Fine and coarse particles contribute roughly equally to excess PM10 mass
29 concentration. However, fine particles contribute about 60% during the fall season and
30 coarse particles prevail (>60%) during the spring. Both fine and coarse particles show a
April 1995 6-128 DRAFT-DO NOT QUOTE OR CITE
-------
03
0
C
£
^B
J£
5
_^
0.
«
s
(A
S
o
(0
^
09
5
to
— 1 1 1 1 "---I. 1 J-.J 1 1 1 1 — • 1 ' '
~ - • ^*
" '- "E3 •-•' +
•'-,._ tk^ .1 \
" ^^~ ^ra
• >-. 13 , • "^x^ I
.X*"' 0 ^«
^..y^eT ^.
. -.;-• ' ' ^O^ 4 +«
---' — ^" ' i
i i i i -_ -r i —i, -f"" i i i i v i ,'
ooooooooooooc
oooooooooooo
oooooooooooo
O co u.
4 a 9 ^
i n
s
«
0
1 2
2 «
m i i o>
§i i ^
s T- £\ o
i£ i — i
3 ! O =>
CO i CO CO
4> ! t t
9U\J JO UOflOBJJ
•8
8
•S
H
I
Oi
a.
t
o
r-
a.
§
!
I
"3
3
£UJ/6ll
3,
April 1995
6-129 DRAFT-DO NOT QUOTE OR CITE
-------
1 winter peak concentration. The chemical composition of the coarse mass is not known but
2 both soil dust or the fine particle species are plausible.
3 The chemical composition of the valley excess fine particle mass concentration also
4 shows a strong seasonality for organics and soot. In fact, the excess organics concentration
5 in the winter (13 /xg/m3) is almost an order of magnitude higher than the summer values.
6 The seasonal concentration of excess fine particle soot is similar to that of the organics.
7 However, the relative magnitude of winter organics compared to soot is higher in the winter
8 (factor of five) than in the summer (factor of two). The concentration of fine particle sulfate
9 is virtually identical for South Lake Tahoe and Bliss State Park. This implies that the South
10 Lake Tahoe aerosol sources do not contain sulfur. It is also worth noting that the excess fine
11 particle soil at South Lake Tahoe is below 1 /jg/m3, which is a small fraction of the coarse
12 mass. Thus, the crustal component of the South Lake Tahoe aerosol contributes to the
13 coarse mass but not appreciably to the fine mass concentration.
14 In summary, there is a significant excess PM10 aerosol concentration S. Lake Tahoe
15 compared to the adjacent Bliss State Park remote site, particularly during the winter season
16 (28 jug/m3). The excess mass is about equally distributed between fine and coarse particles.
17 The fine mass is largely composed of organics.
18
19 6.5.6.2 Salt Lake City, UT Sub-region
20 Salt Lake City, Ogden, and Provo, UT are part of an airshed that is confined by tall
21 mountains to the East, limiting the dispersion by westerly winds.
22 The seasonal average PM10 concentration at three AIRS sites in Salt Lake City,
23 Ogden, and Provo, UT is shown in Figure 6-77b. All three sites show virtually identical
24 seasonality, having peak concentrations during December through January. This confirms
25 that the three sites belong to the same airshed with similar source pattern, meteorological
26 dispersion and chemical transformation and removal processes.
27 The size segregated fine and coarse concentration data exhibit a dynamic seasonal
28 pattern. Fine particles clearly dominate the high winter concentrations reaching 40 to
29 50 pig/m3, compared to 10 /ig/m3 of summer values. This magnitude of fine mass
30 concentration is among the highest recorded in the AIRS data system. Coarse particles are
31 less seasonal and they are more important during the dry summer season. The formation of
April 1995 6-130 DRAFT-DO NOT QUOTE OR CITE
-------
1984 Mar May Jul Sep Nov
-6- =PM10_AV 490350012 SALT LAKE CITY
100
90
80
70
60
60
4O
30
20
10
SALT LAKE, UT
100
J3 -PM1 0JW 490S700(H OgDENI ~ I
-4- «PM10_AV490490002 PROVO
SALT LAKE, UT - 2
1984 Mar May Jul S«p Nov
| -&- -PM10_AV 490360003 NOT IN A CHY^
-B- -PM26_AV 490360003 NOT IN A COY
-+ -PMC.AVG 490360003 NOT IN ACHY
1984 Mar May Jul Sep Nov
[•A- -PM10.AV490363001 SALTLAKECrg
-O- -PM2S_AV49036300t SALTLAKE CfTY
-+ -PMOVG 490363001 SALT LAKE Crr
Figure 6-77. Aerosol concentration pattern near Salt Lake City.
April 1995
6-131 DRAFT-DO NOT QUOTE OR CITE
-------
1 sulfate and nitrate during winter inversion fogs near Salt Lake City, UT were studied by
2 Mangelson et al., 1994.
3
4 6.5.6.3 Denver, CO
5 The Denver "brown cloud is a manifestation of high wintertime concentration of
6 particles and gases. Several recent studies have focused on the characterization of the
7 Denver brown cloud aerosols.
8 Size distribution measurements of winter Denver aerosol (Countess et al., 1981) show
9 that on high pollution days that the MMAD of the accumulation mode aerosol was about
10 0.31 jum with ag±2.0. Wolff et al., 1980 found that on the average motor vehicles were
11 responsible for 27% of the elemental carbon while wood burning was responsible for 39% of
12 the elemental carbon.
13 The chemical composition of wintertime Denver fine aerosol mass (16.4 ptg/m3)
14 (Sloane et al., 1991) shows the dominance of organic (8.1 /wg/m3) and elemental carbon
15 (2.6/ug/m3) oversulfa evidence that the fine particle sulfate and nitrates are bimodal,
16 composed of 0.2 to 0.3/mi and 0.4 to 0.6 /mi modes.
17
18 6.5.6.4 Northern Idaho-Western Montana Sub-region
19 The mountainous northern Idaho and western Montana is characterized by deep
20 valleys, absence of major industrial sources, or large urban-metropolitan areas.
21 Nevertheless, PMj0 monitoring sites in northern Idaho and western Montana report
22 concentrations that are among the highest in the nation, as illustrated in Figure 6-78a, while
23 adjacent sites are among the lowest. The large spatial concentration variability is evidently
24 related to the rugged terrain. Most of the monitoring sites are located in the flat valleys.
25 The PM10 concentration trend (Figure 6-78b) is strongly downward, with a factor of
26 two reduction (from 54 to 27 /ig/m3) between 1985 and 1993. The average seasonality of
27 the sub-region is strongly winter peaked (Figure 6-78c) with a factor of two modulation
28 between 25 and 45 Mg/m3.
29 A typical example in northern Idaho (Figure 6-79a), where three adjacent sites show
30 winter monthly averaged peak concentrations of 50 to 85 jug/m3. This is higher than the
31 monthly average PM10 concentration anywhere in the eastern US.
April 1995 6-132 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1985-1993
AIRS PM10 Concentrantion Trends
N. Idaho - NW.Montana
60 P . • • • 1 . -
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
N Idaho-NW Montana
65 I • • • • • • • • • • r
o
60
55
50
45
40
35
; s so
3 26
20
15
10
6 -
1985 Mar Apr May Jun Jui Aug s«p Oct Nov D«c
Figure 6-78. Aerosol concentration pattern and trends in the N. Idaho-NW
Montana subregion.
April 1995
6-133 DRAFT-DO NOT QUOTE OR CITE
-------
S"!
O
g
o
§
o
April 1995
6-134 DRAFT-DO NOT QUOTE OR CITE
-------
1 during December through February the concentrations are elevated to about 50 /-ig/m3. The
2 third site (AIRS#300630034) shows the highest winter peak (> 100 Mg/m3), but summer
3 values that are comparable to the other two sites. The latter site is the closest to the city
4 center. It is evident that in Missoula, MT high concentration gradients exist between the
5 populated areas and remote sites. Boise and Salmon, ID (Figure 6-79e) also show elevated
6 PM10 concentrations during the cold season. Idaho Falls, ID on the other hand, is seasonally
7 uniform at about 30 yug/m3, which is comparable to the background (AIRS#300630020)
8 Missoula, MT site.
9 Remarkably low PM10 concentrations of 10 jug/m3 are reported at three PM10
10 monitoring sites near Anaconda-Deer, ID (Figure 6-79f). This is remarkable because the
11 sites are in a valley, and that the characteristic winter peak is completely absent. This
12 suggests that pristine, low, PM10 sites can exist in the northwestern valleys, and hence the
13 region is not uniformly covered by wintertime haze or smoke.
14
15 6.5.6.5 Washington-Oregon Sub-region
16 The Pacific Northwest is also a mountainous sub-region that exhibits unique aerosol
17 characteristics. PM10 monitoring sites in Seattle, Bellevue, and Tacoma, WA (Figure 6-80)
18 show relatively low concentrations between 20 to 40 jiig/m3, with the higher values occurring
19 during the winter months. A much more pronounced seasonality of PM10 concentrations is
20 recorded in southern Oregon, Medford, Grants Pass, and Klamath Falls, OR. These sites
21 evidently belong to an airshed of which emissions, dispersion, and aerosol formation
22 mechanisms are conducive for the formation of winter time aerosol (60 to 80 /xg/m3).
23 Fine and coarse particle data collected over limited period in 1987 show that the
24 winter peak of PM10 is entirely due to the strong winter peak of fine particle mass (50 to 100
25 Mg/m3). Coarse mass, on the other hand, is seasonally invariant at about 10 to 20 /wg/ni3.
26 Fine particles clearly are responsible for the winter peak. This is somewhat different from
27 the observations at South Lake Tahoe, where the winter peak was attributed to both fine and
28 coarse particles.
29 The size segregated aerosol data for Bend and Central Point, OR, show diminishing
30 concentrations compared to Bedford, where the reduction of PM10 is mainly due to the
31 decrease of the fine particle mass during the winter season.
April 1995 6-135 DRAFT-DO NOT QUOTE OR CITE
-------
ON
160
140 -
130 -
120 -
110
100
90
80
70
60
60
40
30 f=
20
10
1984 Mar May Jul Stp
-ir -PM10.AV 630330073 SEATTLE
j -a- •PM10_AV 630330004 BELLJVUE"
Nov
.4. -PM10_AV 630630021 TACOMA
MEDFORD, OR
100
Il-
ia:
-£r -PM10.AV 410293001 MEDFORD
-Q -PM10JW 410330008 GRANTS PASS
'.+ 'ipwioJw'41036ob04'KLAM'AT>l FALL.)
BEND, OR
CENTRAL POINT, OR
100
90
80
70
60
60
40
30
20
10
1984 Mar May Jul Sep Nov
|-A- •PWr6_AV'41029300l"MEDFORD''' j
-B- «PM2S_AV 410293001 MEDFORD
-I- -PMC_AVG 410293001 MEDFORD
1984 Mar May Jul S.p Nov
1984 Mar May Jul S«p Nov
;-&- -PM10_AV 410170001 BEND
-B- -PM2SJW 410170001 BEND
•+ -PMC.AVO 410170001 BEND
hi- •PM10_AV410291001 CENTRAL
-a- "PM26_AV 410291001 CENTRAL POINT
•+ -PMC.AVG 410291001 CENTRAL POIN-
Figure 6-80.
Aerosol concentration pattern in Washington State and Oregon.
-------
1 6.5.7 Sub-regional Aerosol Pattern in the Southern California
2 The southern California region has two sub-regions, the San Joaquin Valley and the
3 Los Angeles-South Coast Air Basin, discussed separately in sections below.
4
5 6.5.7.1 San Joaquin basin
6 The wide air basin between the coastal mountain ranges of California to the west and
7 the Sierra Nevada Mountains to the east shows a remarkably uniform PM10 concentrations as
8 indicated on the map (Figure 6-8la).
9 There is evidence of PM10 concentration reduction from 55 to 40 pg/m3, but the trend
10 is not conclusive (Figure 6-81b). The seasonal modulation amplitude over the San Joaquin
11 Valley (Figure 6-8Ic) is about factor of 2.5 between the low spring concentration 30 to 35
12 Atg/m3, and high fall concentration (60 to 70 /-ig/m3). The unique feature of this seasonality
13 is the fall peak which differs from the summer peak in the eastern United States and winter
14 peak over the mountainous northwestern states.
15 The AIRS database contains valuable size segregated fine and coarse particle
16 concentration data within the San Joaquin Valley, as shown in Figure 6-82 for Fresno,
17 Madera, Visalia, and Bakersfield, CA. These monitoring sites show virtually identical
18 concentration pattern for fine and coarse mass. Both coarse and fine particles are important
19 contributors to the San Joaquin Valley PM10 aerosol. However, their respective prevalence
20 is phase shifted. Fine particles are most important during November through February
21 winter season, while coarse particles prevail during June through November. As a
22 consequence, in November, both coarse and fine particles are present causing the seasonal
23 peak of PM10. Conversely, during March through May, neither fine or coarse particles are
24 abundant and the PM10 concentration is lowest during the spring season.
25 The temporal dynamics of the emissions, ventilation and aerosol formation in the San
26 Joaquin Valley was a subject of detailed aerosol monitoring, and source apportionment
27 studies.
28 The aerosol composition at non-urban sites (Chow et al., 1995) provide further
29 characteristics of the central California aerosol pattern (Figure 6-82). A PM10 aerosol study
30 was carried out at six sites in California's San Joaquin Valley from 14 June 1988 to 9 June
31 1989, as part of the 1988 to 1989 Valley Air Quality Study (VAQS). Concentrations of
April 1995 6_137 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1985-1993
60
AIRS PM10 Concentration Trends
San Joaquin Valley
1985 1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
San Joaquin Valley
1985 Mar Apr May Jun Jul Aug S«p oet Nov D«e
Figure 6-81. Aerosol concentration pattern and trends at San Joaquin Valley.
April 1995
6-138 DRAFT-DO NOT QUOTE OR CITE
-------
FRESNO, CA
MADERA, CA
a
10
o^
> a:
< LU
o't
100
90
80
70
60
50
40
30
20
10
1984 Mar May Jul Sep Nov
PM10_AV 060190005 FRESNO
VISALIA, CA
1984 Mar May Jul Sep Nov
"PM10JW 061072002 VISALIA
o ui
100
90
80
70
60
50
O 40
S
30
20
10
1984 Mar May Jul Sep
Nov
|-A- ePM10_AV 060390002 MADERA
BAKERSFIELD, CA
o
<
'
UJ
o
100
90
80
70
60
50
40
30
20
10
1984 Mar May Jul Sep Nov
PM10_AV 060290004 BAKERSFIELD
Figure 6-82. Fine, coarse and PM10 concentration pattern in the San Joaquin
Valley.
April 1995
6-139 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM10 and PM2 5 mass, organic and elemental carbon, nitrate, sulfate, ammonium, and
2 elements were determined in 24-h aerosol samples collected at three urban (Stockton, Fresno,
3 Bakersfield) and three non-urban (Crows Landing, Fellows, Kern Wildlife Refuge) locations
4 (Chow et al., 1993). The VAQS data indicate the federal 24-h PM10 standard of 150 ^g/m3
5 was exceeded at four out of the six sites and for reasons which differ by season and by
6 spatial region of influence. The annual average source contributions to PM10 at Bakersfield,
7 the site with the highest annual average, were 54% from primary geological material, 15%
8 from secondary ammonium nitrate, 10 % from primary motor vehicle exhaust, 8% from
9 primary construction, the remaining 4% is unexplained. The results of the source
10 apportionment at all sites show that geological contributions dominate in summer and fall
11 months, while secondary ammonium nitrate contributions derived from direct emissions of
12 ammonia and oxides of nitrogen from agricultural activities and engine exhaust are largest
13 during winter months. (Chow et al., 1992).
14
15 6.5.7.2 Los Angeles-South Coast Air Basin
16 The Los Angeles basin is confined by the San Gabriel Mountains which limit the
17 ventilation during westerly winds. Intensive emissions from automotive and industrial sources
18 produce the notorious Los Angeles smog as a secondary photochemical reaction product of
19 primary emissions. The map of the Los Angeles sub-region shows (Figure 6-83a) that the
20 highest PM10 concentrations are measured in the eastern half of the LA basin.
21 There has been a substantial reduction of sub-region average PM10 concentration from
22 60 to 37 pig/m3 from 1985 and 1993 (Figure 6-83b). The seasonality of the basin averaged
23 PM10 concentration shows a 50% amplitude, with the peak concentration (60 /-ig/m3) during
24 October and the lowest values (40 /xg/m3) during January through March (Figure 6-83c).
25 Hence, this fall peaked seasonality is similar to the fall peak over the San Joaquin Valley.
26 Some unique characteristics of the Los Angeles basin are depicted in Figure 6-84. It
27 shows that monitoring sites at different parts of the basin have markedly different seasonal
28 concentration pattern. Hawthorne, near the Pacific Coast and Burbank in an inland valley
29 have the highest concentration in late fall (November through January). On the other hand,
30 Rubidoux in the eastern part of the basin exhibits the highest concentration in the late
31 summer, July through October. It is likely, that the main cause of different seasonalities can
April 1995 6-140 DRAFT-DO NOT QUOTE OR CITE
-------
Average PM10 -1986-1993
AIRS PM10 Concentrantion Trends
Los Angeles
1986 1987 1988 1989 1990 1991 1992
AIRS PM10 SEASONAL CONCENTRATION
Los Angeles
80
75
70
65
60
65
O 60
'5 45
40
35
30
25
20
15
10
5
1985 Mir Apr Mty Jun Jul Aug S*p Oct Nov D*e
Figure 6-83.
Aerosol concentration pattern and trends at Los Angeles.
April 1995
6-141 DRAFT-DO NOT QUOTE OR CITE
-------
Nov
LOS ANGELES, CA • Long Beach
100
LOS ANGELES, CA -Azuza
-A- •°PM10_AV 660376001HAWTHORNE
•B- =PM10_AV 060668001 RUBIDOUX
--(- =PM10_AV 060371002 BURBANK
LOS ANGELES, CA - Rubldoux
100
1984 Mar May Jul Sep Nov
rJT-PMIOJW 060374002 LONG BEACH
-Q- -PM26_AV 060374002 LONG BEACH
•Jr -PMC.AVG 06037*002 LONG BEACH
1984 Mar May Jul Sep Nov
I TJr-PM10_AV 060370002 AZUSA |
-0- -PM25.AV 060370002 AZUSA
-4- -PMC.AVG 060370002 AZUSA
1984 Mar May Jul Sep Nov
& -PM10.AV0606S8001 RUBJOOUX
-Q- "PM26_AV 060668001 RUBIDOUX
-4- "PMC_AVG 060668001 RUBIDOUX
Figure 6-84. Fine, coarse and PM10 concentration pattern near Los Angeles.
April 1995
6-142 DRAFT-DO NOT QUOTE OR CITE
-------
1 be found in the seasonally varyingmeteorological, transport, and chemical transformation
2 pattern, rather than of emissions. The role of coarse and fine particles in the Los Angeles
3 basin is also illustrated in Figure 6-84. At Long Beach, near the coast (adjacent to
4 Hawthorne), the concentration of fine particles dominate the PM10 during the November
5 through February winter season (40 to 50 /*g/m3). Coarse particles at Long Beach are
6 constant throughout the year at about (20 ^g/m3). The size segregated aerosol samples at
7 Azusa and Rubidoux in the eastern part of the basin, both show a PM10 peak during
8 September through October, although the concentration are higher at Rubidoux. At both
9 sites fine and coarse particles contribute roughly equally to the high PM10 concentrations.
10 Thus, the PM10 aerosols over the smoggiest parts of the Los Angeles basin are not dominated
11 by fine secondary aerosols but contributed by both fine and coarse particles.
12 The Los Angeles smog has been subject of extensive spatial, temporal, size and
13 chemical composition studies even before the late 1960s (J. Colloid Interface Sci. (1972)
14 volume 39, Hidy et al., 1980). More recently the LA aerosol characteristics have been
15 further elucidated by Southern California Air Quality Study (SCAQS) (Watson et al., 1994;
16 Chow etal., 1994; and other SCAQS studies).
17 PM2.5 constituted one-half to two-thirds of PM10 at all sampling sites. PM10 mass
18 concentrations were highest during the fall and were dominated by PM25. Nitrate, sulfate,
19 ammonium, and organic and elemental carbon were the most abundant species in the PM2 5
20 fraction. The coarse particle fraction was composed of soil-related elements (e.g. aluminum,
21 silicon, calcium, iron) at the inland sites and with marine-related elements (e.g. sodium,
22 chloride) at the coastal sits. Average concentrations for most chemical compounds were
23 higher during the fall than during the summer, except for sulfate which was more abundant
24 in summer. PM2 5 nitrate and ammonium concentrations were negatively biased for daytime
25 samples compared to nighttime samples, consistent with diurnal changes in temperature and
26 the effect of these changes on the equilibrium between paniculate ammonium nitrate and
27 gaseous ammonia and nitric acid. (Chow et al., 1994).
28 In situ, time resolved analysis for aerosol organic and elemental carbon in Glendora,
29 CA (Turpin et al., 1990) showed strong diurnal variations with peaks occurring the daylight
30 hours. Comparison of the diurnal profile of organic carbon with those of elemental carbon
31 providedevidence for the secondary formation of organic aerosol in the atmosphere. Turpin
April 1995 6-143 DRAFT-DO NOT QUOTE OR CITE
-------
1 et al., 1991 observed that secondary organic aerosol appears to have contributed roughly half
2 of the organic aerosol in Pasadena during midday summer conditions.
3 Turpin and Huntzicker (1991) also found that the organic and elemental carbon
4 concentrations exhibit strong diurnal variations. Peak concentrations occur during the
5 daylight hours in the summer and at night in the fall. The maximum concentrations observed
6 in the fall. The maximum concentrations observed in the fall (maximum total carbon,
7 88 /ig/m3) were two to three times higher than the summer maxima (maximum total carbon,
8 36 Aig/m3).
9 Gaseous nitric acid and fine particulate nitrate at Claremont, CA (Pierson and
10 Brachaczek, 1988) both showed pronounced (~ 10 fold) diurnal variations, however coarse
11 particles showed little diurnal variations.
12 Wolff et al., 1991 measured the smog aerosol pattern at Claremont, CA and Long
13 Beach, CA, in the eastern and western Los Angles basin, respectively. Claremont's air
14 quality during the summer was characterized by high concentrations of photochemically
15 produced pollutants including ozone, nitric acid, particulate nitrate, and particulate organic
16 carbon (OC). The highest concentration of these species were experienced during the
17 daytime sampling period (0600 to 1800) and were associated with transport from the western
18 part of the basin. Long Beach' air quality during the fall was characterized by frequent
19 periods of air stagnation that resulted in high concentrations of primary pollutants including
20 PM10, OC and elemental carbon (EC) as well as particulate nitrate. Night -time levels of
21 most constituents exceeded day tune levels due to poorer night-time dispersion conditions. At
22 Claremont, OC and nitrate compounds accounted for 52% of PM10 , while at Long Beach
23 they accounted for 67% of PM10. On the average, there appears to be sufficient particulate
24 ammonium to completely neutralize the nitrate and acidic sulfates.
25 Careful size distribution measurements in the Los Angeles basin (John et al., 1990)
26 shed light on the size spectrum dynamics for ammonium, sulfate and nitrate. Three modes,
27 two submicron and one coarse, were sufficient to fit all of the size distributions. The
28 smallest mode, at 0.2 ±0.1 /ma, aerodynamic diameter, is probably a condensation mode
29 containing gas phase reaction products. A larger mode 0.7+0.2 /mi, is defined as a droplet
30 mode. Most of the inorganic particle mass was found in the droplet mode. The observed
31 condensation and droplet modes characterize the overall size distribution in the 0.1 to 1.0 /mi
April 1995 6-144 DRAFT-DO NOT QUOTE OR CITE
-------
1 range, previously described by Whitby and coworkers as a single accumulation mode
2 (Whitby et al., 1972; Whitby, 1978). Wall et al., 1988 also found that in September 1985 at
3 Claremont, CA fine particle nitrate was associated with ammonium, while coarse mode
4 nitrate was associated with both ammonium and sodium. Sulfate was primarily in two
5 submicrometer modes. Strong acid was associated with the smaller sulfate mode.
6 A clear demonstration of the effect of relative humidity and aerosol loading on
7 atmospheric sulfate size distributions is given by Hering and Friedlander, 1982. Days of
8 high relative humidity and aerosol loading correspond to high mass median diameters
9 (0.54+0.07 jitm) for the sulfate while low relative humidity and low aerosol loadings
10 correspond to small mass median diameters (0.2±0.02 /mi). According to their
11 interpretation, the large (0.54 /mi) sulfate particles resulted from aqueous phase reactions of
12 SO2. The fine (0.2 /mi) sulfate resulted from homogeneous gas phase reactions leading to
13 the nucleation of sulfuric acid particles.
14 McMurry and Stolzenburg, 1989 provide evidence that Los Angeles smog aerosols are
15 externally mixed. Monodisperse ambient aerosols were often found to split into
16 nonhygroscopic (nowater uptake) and hygroscopic portions when humidified. An average of
17 30% of the particles in the 0.2 to 0.5 /mi range were nonhygroscopic. However, the
18 proportion of particles that were nonhygroscopic varied considerably from day to day and on
19 occasions was 70 to 80% of the particles. The data show that for the hydrophilic aerosol,
20 the larger particles (0.4 to 0.5 /mi) grew more when humidified than did smaller particles
21 (0.05 to 0.2 /mi).
22 Murray and Zhang, 1989 reported the size distribution of ambient organic and
23 elemental carbon near the Grand Canyon and in the Los Angeles basin. Virtually all of the
24 carbon was found in the submicron range, some below 0.1 /mi. However, positive sampling
25 artifacts for sub O.lptm organics were considered significant.
26 At the Grand Canyon National Park, Zhang et al., 1994, showed that sulfates and
27 carbonaceous particles were the major contributor to PM2 5 particle scattering during the
28 three winter months and that their contributions were comparable. Scattering by nitrates and
29 soil dust was typically a factor of five to ten smaller. The low pressure impactor
30 measurements also showed that sulfur size distributions varies considerably (0.07 to
31 0.66 /mi).
April 1995 6-145 DRAFT-DO NOT QUOTE OR CITE
-------
1 Size distributions of aerosol phase aliphatic and carbonyl groups at Claremont, CA
2 (Pickle et al., 1990) showed maxima in the 0.12 to 0.26 mm and the 0.5 to 1.0 mm size
3 functions. For the aliphatic carbon absorbency, the ambient samples generally showed
4 maxima in the 0.076 to 0.12 mm size fraction, the authors attribute the carbonyl absorbance
5 almost entirely attributed to products of atmospheric reactions and the aliphatic absorbencies
6 in particles smaller than 0.12 mm to automotive emissions.
7 Cahill et al., 1990 found that the sulfate aerosol size at Glendora, CA is smaller,
8 0.33 /mi (MMD) during clear days compared to 0.5 /mi on smoggy days.
9 The size distributions of organic nitrate groups in ambient Los Angeles aerosol were
10 typically bimodal (Mylonas et al., 1991). During periods of high photochemical activity, the
11 maxima in the mass loadings were in the 0.05 to 0.075 /mi and the 0.12 to 0.26 /mi size
12 fractions. During periods of low-moderate ozone concentrations in the distributions were
13 shifted to slightly larger sizes, with maxima appearing in the 0.075 to 012 /mi and the 05 to
14 1.0 /mi size fractions. A principal component analysis of the organonitrate loadings
15 revealed strong correlations with ozone concentrations and with aerosol phase carbonyl
16 loadings.
17 The analysis of coarse particles in Claremont, CA (Noll et al., 1990) show that the
18 coarse particle mass could be divided into two categories: material that was primarily of
19 crustal origin (Al, Ca, Fe, and Si) and material that was primarily of anthropogenic origin
20 (Cd, Cu, Mn, Ni, Pb, and Zn). The mass of crustal material varied between 15 and 50% of
21 the total coarse mass, while the mass of anthropogenic material was < 1%.
22 Chow et al., 1992 also conducted a neighborhood-scale study of PM10 source
23 contributions in Rubidoux, CA elucidating the role of local soil dust.
24 The daily frequency distribution of the chemical components of the Los Angeles
25 aerosol measured over a 1-year period were approximately lognormal (Kao and Friedlander,
26 1994). For nonreactive aerosol components, the geometric standard deviation (GSD) is
27 nearly constant at 1.85±0.14 even for components from different source types. An apparent
28 bimodal frequency distribution for sulfates probably corresponds to the two differing reaction
29 pathways by which gas-to-particle conversion occurs. However, the bimodal sulfate
30 distribution function was not found at other Los Angeles sites (Kao and Friedlander, 1995).
31 The authors suspect a relationship between GSD and the level of complexity of the stochastic
April 1995 5.145 DRAFT-DO NOT QUOTE OR CITE
-------
1 physical and chemical processes affecting the distributions of the individual species. They
2 also point out that the chemical composition of the probable Los Angeles aerosol that
3 corresponded to the peak in the (nearly) lognormal frequency distribution of the total mass is
4 lower than he simple average chemical composition.
5 Twenty-four year (1958 to 1982) average elemental C concentrations at seven
6 monitoring sites in the Los Angeles area are estimated to range from 6.4 /ng/m3 at downtown
7 Los Angeles to 4.5 /*g/m3 at West Los Angeles. (Cass et al., 1984). At most monitoring
8 sites studied, elemental C concentration were lower in recent years than during the late 1950s
9 and early 1960s.
10
11
12 6.6 CHEMICAL COMPOSITION OF PM AEROSOLS AT URBAN AND
13 NON-URBAN SITES
14 This section summarizes available data for the composition of atmospheric particles in
15 suburban, urban, and a few rural areas for comparison purposes. Emphasis has been placed
16 on the Harvard six-city study and the inhalable particulate network (1980-1981). However,
17 data for fine particle mass and elemental composition were available from these studies.
18 Data for sulfate, nitrate, and elemental and organic carbon content are included from other
19 studies to provide an overview of the chemical composition of the atmospheric aerosol in the
20 United States. Tables, presented in Appendix 6A, provide relatively detailed representation
21 of atmospheric properties of aerosols to which U.S. populations are exposed. Unfortunately,
22 data this complete are generally collected over limited time periods and are not of sufficient
23 duration to be useful for most epidemiological investigations. The tables do, however,
24 provide insights as to the types of information that could be collected as part of future
25 monitoring efforts in support of human exposure investigations.
26 A summary of all the aerosol sampling studies included in this compilation is given in
27 Table 6A-la and 6A-lb. Sampling studies have been grouped by geographical region
28 roughly corresponding to the eastern, central and western U.S. Data are tabulated for the
29 PM2.5 (d <2.5 pm), the coarse fraction of PM10 (2.5 ^m < d < 10 /*m) and PM-10 (d
30 < 10 jum) size fractions of the ambient aerosol in Tables 6A-2a, 6A-2b, and 6A-2c.
31 Compositional data for all size fractions were broken down into the following major
32 components: sulfate, represented here as ammonium sulfate, (NH4)2SO4; carbon, as organic
April 1995 6-147 DRAFT-DO NOT QUOTE OR CITE
-------
1 carbon (Cv) which has been multiplied by a factor of 1.4 to account for the presence of
2 oxidized species), and elemental carbon (Ce); nitrate as NO3"; and remaining trace elements
3 which have been determined by XRF and or INAA. The masses of the trace elements from
4 sodium through lead have been calculated by assuming they are in their most stable forms for
5 conditions at the earth's surface. Reconstructed masses calculated in this way are shown by
6 the entry, Sum, along with measured masses, and the ratio of the two are shown at the
7 bottom of the individual summaries for each size fraction. Not all compositional categories
8 were measured in the studies included in the Tables. The data are shown in graphical form
9 in Figures 6-85a, 6-85b, and 6-85c.
10 As can be seen from inspection of Figure 6-85a for the eastern U.S., sulfate is the
11 major identified component of mass for fine particles (46.9%), followed by carbon (24.8%),
12 minerals (4.3%), and nitrate (1.1%). However, this last inference is based on only two
13 studies in which nitrate was measured. Coarse particles are seen to consist mainly of mineral
14 forming elements (51.8%) and sulfate (6.7%). Not enough data were available to determine
15 abundances of carbon species and nitrate in the coarse fraction. A sizable fraction of both
16 the fine (23.0%) and coarse (41.5%) particle mass is shown as unknown. This unknown
17 mass is assumed to be mainly water, either bound as water of hydration or associated with
18 hygroscopic particles. A small fraction of the mass, especially in the coarse fraction, may be
19 present as carbonates. Carbonates are difficult to quantify, in part because of artifact
20 forming reactions on filters with atmospheric CO2. Stable carbonates can be identified by
21 SEM, especially in regions where they are known to represent a substantial fraction of soil
22 composition.
23 Fine particles in the central U.S. (Figure 6-85b) are seen to consist mainly of sulfate
24 (37.9%) and minerals (9.4%) and elemental and organic carbon (66.1%) abundances. The
25 reconstructed mass percentages sum to more than 100%. This is probably because of an
26 overestimation of the carbon content which was based on only a few samples collected during
27 winter in woodsmoke impacted areas. Coarse particles were found to consist mainly of
28 minerals (62.8%), sulfate (4.2%) and an unknown fraction (33.0%). No nitrate or carbon
29 data were available for the coarse fraction from the studies in the central U.S. However,
30 during the spring and fall pollen becomes a significant fraction of the coarse particle
31 composition.
April 1995 6-148 DRAFT-DO NOT QUOTE OR CITE
-------
EASTERN U.S.
PM2.5 Mass Apportionment
-Minerals (4.3%)
• i r
Unknown (23.0%)
EC (3.9%)
OCxl.4 (20.9%)-
Nitrate (1.1%)-
Nitrate based on 3 studies.
•NH42SO4 (46.9%)
Coarse Mass Apportionment
Unknown (41.5%)
Minerals (51.8%)
NH42S04 (6.7%)-
Insufficient Nitrate, OC, and EC data available.
PM10 Mass Apportionment
Unknown (29.2%)
EC (3.3%)
OCx1.4 (8.5%)
^-Minerals (19.6%)
Nitrate (1.2%)
-NH42SO4 (38.2%)
Nitrate based on 2 studies.
Figure 6-85a. Mass apportionment: Eastern U.S.
April 1995
6-149 DRAFT-DO NOT QUOTE OR CITE
-------
CENTRAL U.S.
PM2 5 Mass Apportionment
EC(11.1%)-\ ^-Minerals (9.4%)
OCx1.4 (55.0%)
NH42SO4 (37.9%)
-Nitrate (10.0%)
Reconstructed sum =123%
Coarse Mass Apportionment
Unknown (33.0%)
NH42SO4 (4.2%)-\ffl
(62.8%)
Insufficient Nitrate, OC, and EC data available.
PM10 Mass Apportionment
Unknown (1.4%)-i
V—Minerals (35.8%)
OCx1.4 (29.6%)
EC (5.0%)
-Nitrate (4.5%)
T -"
Sulfate (23.7%)-
Nitrate based on 3 studies; OC and EC based on 4 studies.
Figure 6-85b. Mass apportionment: Central U.S.
April 1995
6-150 DRAFT-DO NOT QUOTE OR CITE
-------
WESTERN U.S.
PM2.5 Mass Apportionment
Unknown (1.3%)-
EC (14.7%)
OCx1.4 (38.9%)
Minerals (14.6%)
NH42S04 (14.6%)
Nitrate (15.7%)
Coarse Mass Apportionment
Unknown (27.0%)
NH42SO4(3.1%)
r--Minerals (70.0%)
Insufficient Nitrate, OC, and EC data available.
PM10 Mass Apportionment
EC (5.0%)-
OCx1.4(29.6%)
V—Minerals (35.8%)
NH42S04 (23.7%)-- ^-Nitrate (4.5%)
Reconstructed sum = 106%
Figure 6-85c. Mass apportionment: Western U.S.
April 1995
6-151 DRAFT-DO NOT QUOTE OR CITE
-------
1 While gross fine particle composition appears to be broadly similar between the
2 eastern and central U.S. on the basis of the few studies available, the fine particle
3 composition is seen to be distinctly different in the western U.S. (Figure 6-85c). Elemental
4 plus organic carbon species (53.6%) are the major identified component of mass, instead of
5 sulfate (14.8%), and minerals and nitrate account for a larger fraction of total mass. While
6 minerals are seen to account for most of the coarse particle mass (70.0%), insufficient data
7 were available for elemental and organic carbon species to estimate their contributions to the
8 coarse mass. Table 6A-3 shows a comparison of selected ratios of mass components for
9 each of the three geographical regions of the U.S.
10 Many of the studies listed in Table 6A-3 involved data collected at more than one site
11 within an airshed. Information about the variability of particle mass within an airshed can
12 yield information about the nature of sources of the particles. The variability of mean
13 concentrations measured at multiple sites within a study area is used as a measure of the
14 intersite variability in fine particle composition and is shown in Table 6A-4.
15 Data for the chemical composition of the ambient aerosol has been summarized from
16 the Harvard six-city study, the inhalable particle (IP) network and for a number of other
17 studies around the country. As can be seen from inspection of the tables, data are not
18 available to characterize the carbon or nitrate content of the ambient aerosol for many of the
19 studies listed. Over the past 15 years woodstove emissions have become a significant
20 contributor to fine particle mass during the winter (Stevens 1990).
21
22
23 6.7 ACID AEROSOLS
24 6.7.1 Introduction
25 Acid aerosols are secondary pollutants formed primarily through oxidation of sulfur
26 dioxide (SO2), a gas emitted by the combustion of fossil fuels. Oxidation of SO2 forms
27 sulfate (SO4=), the major component of acid aerosols. Sulfate is formed to a lesser extent
28 through the oxidation of sulfur species (H2S and CH3SCH3) from natural sources. The
29 oxidation of SO2 occurs through a series of heterogeneous (gas-particle) or homogeneous (gas
30 or aqueous) phase oxidation reactions that convert SO2 to sulfuric acid (H2SO4) particles.
31 The sulfate species are typically expressed in terms of total 804, with the acidic fraction
April 1995 6-152 DRAFT-DO NOT QUOTE OR CITE
-------
1 expressed in terms of titratable H+ and referred to as aerosol strong acidity. H+ is found in
2 the fine particle size fraction (aerodynamic diameter (Dp) < 1.0 /mi) (Koutrakis and Kelly,
3 1993). Although recent research has shown a high correlation between SO4 and acidity,
4 data from summertime sampling have shown that SO4= is not always a reliable predictor of
5 H+ for individual events at a given site (Lipfert and Wyzga, 1993).
6 A major determinant of the lifetime of H+ in the atmosphere is the rate of
7 neutralization by ammonia (NH3). Ammonia reacts with H2SO4 to form ammonium sulfate
8 [(NH4)2SO4] and ammonium bisulfate (NH4HSO4). The major sources of ammonia in the
9 environment are animals and humans (Fekete and Gyenes, 1993). The then current state-of-
10 knowledge regarding acid aerosols was reviewed by EPA in 1989 (U.S. Environmental
11 Protection Agency, 1989) and Spengler et al. in 1990 (Spengler et al., 1990).
12
13 6.7.2 Geographical Distribution
14 In North America, ambient concentrations of H+ tend to be regional in nature with
15 the highest concentrations found in the northeastern United States and southwestern Canada
16 Spengler et al. (1990) have collected information of maximum values of SO^ and H+ found
17 across the U.S. and southern Canada. This information is shown in Table 6-3. Recent
18 research has shown that regional transport is important to acid sulfate concentrations, as
19 elevated levels of ambient H+ were measured simultaneously during a regional episode at
20 multiple sites located from Tennessee to Connecticut (Keeler et al., 1991). It is commonly
21 believed that the source region for most of the H+ precursors (primary inorganic pollutant
22 gases — SO2 and NOX) is the Ohio River Valley (Lioy et al., 1980). The conversion of the
23 primary gases to secondary pollutants takes place as the prevailing winds carry the precursors
24 from the source region, northeastward to the northeastern United States and southwestern
25 Canada. This type of northeasterly wind flow occurs on the backside (western side) of mid-
26 latitude anti-cyclones (high pressure systems).
April 1995 6-153 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-3. MAXIMUM SOj AND H+ CONCENTRATIONS MEASURED IN
NORTH AMERICAN CITIES. H+ CONCENTRATIONS EXPRESSED AS "H2SO4"
EQUIVALENTS. "SC" INDICATES SEMI-CONTINUOUS MEASUREMENTS
Location
Lennox, CA
Smoky Mountains
High Point, NJ
Brookhaven, NY
Tuxedo, NY
St. Louis, MO
St. Louis, MO
Los Angeles, Ca
Harriman, TN
Watertown, MA
Fairview Lake, NJ
Warren, MI
Whiteface Mt., NY
Toronto, Ontario, Canada
Allegheny Mt., PA
Laurel Mt., PA
Harriman, TN
St. Louis, MO
Topeka, KS
Watertown, MA
Steubenville, OH
Portage, WI
Kanawha Valley, WV
Dunville, Ontario, Canada
Hendersonville, TN
Livermore, CA
Morehead, KY
Monroeville, PA
Pembroke, ON, Canada
Springdale, AR
Newtown, CT
Allegheny Mt. , PA
Uniontown, PA
State College, PA
Philadelphia, PA
Pittsburgh, PA
Sample Duration (h)
2-8
12
6
3
1-12
SC
SC
12
SC
SC
SC,4
24
24
8,16
7,10
7,10
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
12
12,24
12
24
6,24
Maximum
SC£ Oig-m-3)
18
17
37
24
41
25
43
10
47
31
27
37
59
75
45
56
28
40
14
23
56
33
46
31
23
9
23
42
29
11
26
33
52
47
39
27
Concentration
H2SO4 (pig-m"3)
0.1
10
18
10
9
7
34
3
18
14
12
9
14
19
31
42
14
6
3
9
18
4
22
15
11
2
14
18
14
2
8
20
39
25
9
15
April 1995 6-154 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.7.3 Spatial Variation (City-scale)
2 A study of acid aerosols and ammonia (Suh et al., 1992) found no significant spatial
3 variation of H+ at Uniontown, Pennsylvania, a suburb of Pittsburgh. Measurements at the
4 central monitoring site accounted for 92% of the variability in outdoor concentrations
5 measured at various homes throughout the town. There was no statistical difference (p >
6 0.01) between concentrations of outdoor H+ among five sites (a central site and four satellite
7 sites) in Newtown, Connecticut (Thompson et al., 1991). However, there were differences
8 in peak values which were probably related to the proximity of the sampling sites to
9 ammonia sources. These studies suggest that long-term averages should not substantially
10 differ across a suburban community, although peak values may differ significantly.
11 In small suburban communities outdoor concentrations of H+ are fairly uniform,
12 suggesting that minor differences in population density do not significantly affect outdoor H+
13 or NH3 concentrations (Suh et al., 1992). In urban areas, however both H+ and NH3 exhibit
14 significant spatial variation. Waldman et al. (1990) measured ambient concentrations of H+,
15 NH3, and SOJ at three locations in metropolitan Toronto. The sites, located up to 33 km
16 apart, had significant differences in outdoor concentrations of H+. Waldman and co-workers
17 reported that the sites with high NH3 measured low H+ concentrations. However, the
18 limited number of sampling sites did not allow for a conclusive determination of the
19 relationship between population density, ammonia concentrations, and concentrations of acid
20 aerosols.
21 An intensive monitoring study has been conducted during the summers of 1992 and
22 1993 in Philadelphia (Suh et al., 1994a). Twenty-four hour measurements of aerosol acidity
23 (H+) sulfate and NH3 were collected simultaneously at 7 sites in metropolitan Philadelphia
24 and at Valley Forge, 30 km northeast of the city center. The researchers reported that SO^
25 was evenly distributed throughout the measurement area but H+ concentrations varied
26 spatially within metropolitan Philadelphia. This variation was related to local NH3
27 concentrations and the local population density (Figure 6-86). The amount of NH3 available
28 to neutralize H+ increased with population density, resulting in lower H+ concentrations in
29 more densely populated areas. The extent of the spatial variation in H+ concentrations did
30 not appear to depend on the overall H+ concentration. It did, however, show a strong
31 inverse association with local NH3 concentrations.
April 1995 6-155 DRAFT-DO NOT QUOTE OR CITE
-------
0.6 r
0.4
0.3
0.2
0.1
0.0
_ 3
5 r
120
90
CO
09
O
E
.5. 60
o
CO
30
SO
NH
0 5000 10000 15000 20000
POPULATION DENSITY (persons/sq.mile)
Figure 6-86. Mean air pollutant concentrations for days when winds were from the
southerly direction plotted vs. population density. The solid line
represents H+ concentrations; the long dashed line represents SO|"
concentrations; the dashed and dotted line represents the ration of H+ to
SO|" levels; and the dotted line represents NH3 concentrations. All data
collected in Philadelphia, PA, during the summers of 1992 and 1993.
Figure adapted from Suh et al. (1994a).
1 6.7.4 Spatial Variation (Regional-scale)
2 Recent evidence has shown that meteorology and regional transport are extremely
3 important to acid sulfate concentrations. Lamborg et al. (1992) measured H+ concentrations
4 to investigate the behavior of regional and urban plumes advecting across Lake Michigan.
5 Results suggested that aerosol acidity is maintained over long distances (up to 100 km or
6 more) in air masses moving over large bodies of water. Lee et al. (1993) reported that H+
April 1995
6-156
DRAFT-DO NOT QUOTE OR CITE
-------
1 and SO 4 concentrations measured in Chicago over a year were similar to levels measured in
2 St. Louis. In an analysis of acid sulfate concentrations measured at Pittsburgh, State
3 College, and Uniontown, Pennsylvania, Burton et al. (1995) reported high correlations for
4 H+ between all three locations. The three locations are separated by large distances
5 (approximately 60 to 240 km) and have vastly different population densities.
6
7 6.7.5 Seasonal Variation
8 An analysis of results from Harvard's 24-city study (Thompson et al., 1991), which
9 measured acid aerosols concentrations at 8 different sites across North America each year
10 during a three year period, revealed that the summer H+ mean concentrations were
11 significantly higher than the annual means at all sites. The results not only showed that at
12 the sites with high H+ concentrations approximately two-thirds of the aerosol acidity
13 occurred from May through September (Figure 6-87). Little or no seasonal variation was
14 observed at sites with low acidity. These findings were supported by those of Thurston et al.
15 (1992) in which H+ concentrations measured at Buffalo, Albany, and White Plains, New
16 York were found to be highest during the summertime. Thurston and co-workers also
17 reported that moderate concentrations of H+ could occur during non-summer months.
18
19 6.7.6 Diurnal Variation
20 Evidence exists of a distinct diurnal pattern in outdoor H+ concentrations. Wilson
21 et al. (1991) examined concentration data for H+, NH3, and 804 from the Harvard 24-City
22 Study for evidence of diurnal variability (Figure 6-88). This investigation found a distinct
23 diurnal pattern for H+ concentrations and the H+/SOJ ratio, with daytime concentrations
24 being substantially higher than nighttime levels. Both H+ and SO4 concentrations peaked
25 between noon and 6:00 pm. No such diurnal variation was found for NH3. Wilson and co-
26 workers concluded that the diurnal variation in H+ and SO4= was probably due to
27 atmospheric mixing. Air containing high concentrations of H+ and SOJ mix downward
28 during daylight hours when the atmosphere is unstable and well-mixed. During the night,
29 ammonia emitted from ground-based sources neutralize the acid in the nocturnal boundary
30 layer, the very stable lower part of the atmosphere, but a nocturnal inversion prevents the
April 1995 6-157 DRAFT-DO NOT QUOTE OR CITE
-------
I
o
E
c
1
a
i
w
o
180
150-
120-
90-
60 J
30-
• MOREHEAD
0 HENDERSONVILLE
m PENN HILLS
0 DUNNVILLE
D NEWTOWN
• PEMBROKE
H SPRINGDALE
m LIVERMORE
JAN FEE MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
Figure 6-87. Average monthly aerosol strong acidity for Year 1 sites of the Harvard 24-
city study.
Source: Thompson et al., 1991
1 ammonia from reacting with the acid aerosols aloft. Then in the morning as the nocturnal
2 inversion dissipates, the acid aerosols mix downward again as the process begins anew.
3 Spengler et al. (1986) also noted diurnal variations in sulfate and sulfuric acid concentrations
4 and suggested atmospheric dynamics as the cause.
April 1995
6-158
DRAFT-DO NOT QUOTE OR CITE
-------
Harriman, TN, 8/12/84-8/19/84
20 40 60 80 100 120 140 160 180 200
T3
CO
o
£
CO
I
J3>
o
c
Vertical lines drawn at noon.
3.2
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Harriman, TN Average
August 12,1984-August 19,1984
-O- Sulfate
-*— Hydrogen Ion
10 12 14
Hour
16 18 20 22 24
Figure 6-88. Diurnal pattern of sulfate and hydrogen ion at Harriman, TN; (a) weekly
pattern, (b) daily average.
April 1995
6-159
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.7.7 Indoor and Personal
2 Several studies have examined indoor concentrations of acid aerosols and
3 personal monitoring. Brauer et al. (1989) monitored personal exposures to particles
4 (including acidic sulfates) and gases in metropolitan Boston in the summer of 1988, and
5 compared these to measurements collected at a centrally located ambient monitor. They
6 found that personal concentrations of acidic aerosols and gases differed significantly from
7 those measured at the centrally located site. Summer and winter concentrations of acid
8 aerosols and gaseous pollutants also collected in Boston (Brauer et al., 1991) showed
9 indoor/outdoor ratios of H+ to be 40-50% of the indoor/outdoor SO4~ ratio indicating
10 neutralization of the acid by the higher indoor NH3 levels, which were reported.
11 Indoor, outdoor, and personal acid aerosol monitoring was performed for
12 children living in Uniontown, Pennsylvania, during the summer of 1990 (Suh et al., 1992).
13 The indoor, outdoor, and personal measurements were compared to outdoor measurements
14 collected from a centrally located ambient monitor. Personal concentrations were lower than
15 corresponding outdoor levels but higher than indoor levels. Air conditioning was found to be
16 an important predictor of indoor H+, while NH3 was found to influence indoor and personal
17 H+ concentrations. Similar results were obtained in a study of the relationships between
18 indoor/outdoor concentrations of H+ and NH3 conducted in State College, Pennsylvania in
19 1991 (Suh etal., 1994b).
20 In a study characterizing H+ concentrations at child and elderly care facilities,
21 Liang and Waldman (1992) measured indoor and outdoor acid aerosol concentrations.
22 Results from this study showed that indoor/outdoor H+ and SO 4 ratios were comparable to
23 those measured inside residential buildings. Air conditioner use and indoor NH3
24 concentrations were again identified as important determinants of indoor H+ concentrations.
25
26
27 6.8 PARTICLE NUMBER CONCENTRATION
28 6.8.1 Introduction
29 Recent work has suggested that ultrafine particles may be responsible for some
30 of the health effects associated with exposure to particulate matter (Section XX). The
31 hypothesis for explaining a biological effect of ultrafine particles is based on the number,
April 1995 6-160 DRAFT-DO NOT QUOTE OR CITE
-------
1 composition and size of particles rather than their mass (Seaton et al., 1995). This has led to
2 an interest in the number concentration of ambient particles. This section examines data on
3 particle number concentration and the relationship between particle number and particle mass
4 or volume.
5
6 6.8.2 Ultrafine Particle Number-Size Distribution
7 In the context of ambient particles, the term ultra fine particles refers to those
8 particles with diameters below about 0.1 /mi. Ultrafine aerosol size distributions from an
9 urban site at Long Beach, California (Karch et al., 1987), and from a background site in the
10 Rocky Mountains, Colorado (Kreidenwies and Brechtel, personal communication) are shown
11 in Figures 6-89 and 6-90. Both of these sets of data were obtained by electrical mobility
12 measurements. For the urban aerosols of Long Beach, the geometric mean number diameter
13 can vary from 0.012 /mi to 0.043 /xm. Some of the ultrafine distributions, such as that
14 shown for the 1,200 to 1,400 PST time period, are bimodal. The number concentrations
15 were higher midday, as shown in Figure 6-91. For the background aerosols from Rocky
16 Mountains the geometric mean diameter of the ultrafine aerosols was somewhat larger than
17 for Long Beach, with geometric mean diameters ranging from 0.047 to 0.075 jum for periods
18 without urban influence. A bimodal character for the ultrafine distribution was also observed
19 for some measurements, as seen in Figure 6-90.
20 The contrast between urban and background ultrafine aerosol size distribtution
21 is demonstrated in Figure 6-92, where a change in the wind direction brought transport from
22 an urban area to the background site at Rocky Mountains. Within a 2-h period, the number
23 concentration increased from 850 cm"3 to 19,000 cm"3, an increase of more than a factor of
24 20. In contrast, the volume distribution increased by less than a factor of 5. The number
25 geometric mean diameter decreased from 0.052 />un for the background aerosol to 0.024 /jm
26 for the urban influenced aerosol. For the urban influenced size distributions, over 96% of
27 the particle number was measured in particles below 0.1 /zm, while 80% of the particle
28 volume was associated with particles above that size.
April 1995 6-161 DRAFT-DO NOT QUOTE OR CITE
-------
Long Beach, CA
0.00
0.01
Particle Diameter (|im)
0.10
0.00
0.01
Particle Diameter (|im)
0.10
Figure 6-89. Aerosol number and volume size distributions from an urban site at Long
Beach, CA.
April 1995
6-162 DRAFT-DO NOT QUOTE OR CITE
-------
Rocky Mountains, CO
11/23/941304
11/23/941804
11/24/941205
0 ^os-
0.01
0.1
Particle Diameter, Dp (|im)
0.6 T
^ 0.5
CO
E
0 0.4
CO
0.3 -f
Q.
Q
o>
° 0.2 +
0.1 4
0.01
11/23/941304
11/23/941804
11/24/941205
0.1
Particle Diameter, Dp (urn)
Figure 6-90. Aerosol number and volume size distributions from a background site in
the Rocky Mountains, CO.
April 1995
6-163 DRAFT-DO NOT QUOTE OR CITE
-------
90,000 -I
80,000 -
|_ 70,000 -
° 60,000 -
° co" 50,000 •
w E
o> o 40,000 -
o —
t:
to
Q.
I
30,000 •
20,000 •
10,000 •
0
12
14
16
18
Time of Day
20
22
24
Figure 6-91. Number concentrations as a function of time of day at Long Beach, CA.
l 6.8.3 Relation of Particle Number to Particle Mass
2 In general, the majority of airborne particle mass is associated with particles above
3 0.1 /mi, while the highest number concentration of particles is found in particles below
4 0.1 /*m. This can be seen in the recent data collected in the Los Angeles, CA shown in
5 Figure 6-93. As with the data of Whitby and Sverdrup, the size distributions of Figure 6-93
6 show data collected by several instruments. Physical size distributions were measured with
7 an electrical aerosol analyzer for particles between 0.01 and 0.4 /zm, and with a laser optical
8 particle counter for particles between 0.14 and 3 jum (Eldering and Cass, 1994).
9 Additionally, Berner (John et al., 1989, 1990) and MOUDI impactors (McMurry, 1987)
10 were used to measure the size distribution of inorganic ion species and carbonacous species,
1 1 and these data have been combined to give a total mass distribution from which the number
12 distribution has been calculated assuming an effective aerosol density of 1.6 g/cm3, and
April 1995
6-164
DRAFT-DO NOT QUOTE OR CITE
-------
Rocky Mountains, CO
60,000
12/25/941452
12/25/941550
12/25/941648
0.01
Particle Diameter, Dp ((im)
3.0 T
2.5 -
D —
O
- 12/25/9414:53
12/2b/94 lb.4b
— 12/25/9416:53
0.01
0.1
Particle Diameter, Dp (|im)
Figure 6-92. Number and volume size distributions at the Rocky Mountain site showing
an intrusion of urban air.
April 1995
6-165
DRAFT-DO NOT QUOTE OR CITE
-------
Downtown Los Angeles
150,000 -,
_ 125,000
co"
E 100,000
0.01
0.1 1
Particle Diameter, Dp (urn)
10
CO
p
o
CO
E
•o
^
•o
O)
0 -t-OMIMCMMMCiili
0.1 1
Particle Diameter, Dp ((im)
i
10
OPC
EAA
Figure 6-93. Number and volume size distributions from Los Angeles, CA, showing
comparison of three measurement techniques.
April 1995
6-166
DRAFT-DO NOT QUOTE OR CITE
-------
1 assuming that the water associated with the aerosol is 15% of the measured dry particle mass
2 (see McMurry and Stolzenburg, 1989). The optical particle counter data were reduced using
3 calibration with dioctyl sebacate aerosol, which has a refractive index (n = 1.45) more
4 closely related to that of ambient aerosol than does polystyrene latex (n = 1.59) (Hering and
5 McMurry, 1990). No fitting has been applied to match the different size distributions in the
6 region of overlap.
7 Figure 6-93 shows the average of distributions collected over a six different days in the
8 fall of 1987 in downtown Los Angeles, as part of the Southern California Air Quality Study.
9 Particle number distributions emphasize the ultrafine particles, or "nuclei" mode. Volume
10 distributions place importance on 0.1 to 1 /xm particles which are associated with the
11 "accumulation" mode. For this average distribution 88% of the particle number is associated
12 with particle below 0.1 jiim, but 99% of the particle volume is from particle above that size.
13 Both the impactor and optical counter data indicate a weakly bimodal character for the
14 accumulation mode aerosol.
15 For unimodal, log normal size distributions, the particle volume V is simply related to
16 the particle number Nby the relation:
17
18
19 where D is the geometric number mean diameter, and agis the geometric standard
20 deviation. However, because of the multimodal character of ambient aerosol size
21 distributions, one does not expect this simple relationship to hold in the atmosphere. The
22 relationship between particle number and particle volume was examined for data from the
23 Southern California Air Quality Study collected at Riverside, CA over 11 days in the
24 summer of 1987, and at downtown Los Angeles in the fall of 1987 using the methods
25 described above. As shown in Figure 6-94, particle number concentrations are correlated
26 with the volume associate with particles below 0.1 /^m, but are not correlated with the total
27 fine particle volume. Similar results are found for the data reported from Rocky Mountains,
28 CO and for the data reported by Whitby and Sverdrup (1980).
April 1995 6-167 DRAFT-DO NOT QUOTE OR CITE
-------
CO
0
160,000 T
140,000 --
120,000
w 100,000 --
CD
E 80,000 --
^D 60,000 -•
o
rc 40,000 --
O.
0
0.00 2.00 4.00 6.00
Volume <.1u,m (iim3cm-3)
8.00
• Los Angeles
D Riverside
* Whitby Background
o Whitby Urban
A Rocky Mts.
160,000 -
140,000 -
CO
4- 120,000 -
E
0
,_ 100,000 -
CD
.a
E 80,000 -
D
z
CD 60,000 -
rt 40,000 1
a.
•
O
•
* *
• m •
•
• •
* n cj *
• • n
3 • D a U a
20.000 k LJnritin 7" tr a
• Los Angeles
D Riverside
A Whitby Background
o Whitby Urban
A Rocky Mts.
p QA D
r\ AA. i- > — i 1
50 100 150 200
Volume < 2.5 (im (u.m3 cm-3)
Figure 6-94. Relationship between particle number and particle volume; (a) volume
<0.1 /on, (b) volume <2.5
April 1995
6-168
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.8.4 Conclusion
2 The size distribution data for ultrafine aerosols in urban and continental backgroun
3 regions have geometric mean diameters which vary from 0.01 to 0.08, with the larger values
4 found in background regions. Particle number concentrations may vary from less than
5 1,000/cm3 at clean, background sites to over 100,000/cm3 in polluted urban areas. Particle
6 number concentrations are dominated by the ultrafine, "nuclei" mode aerosols. In contrast,
7 the volume (or mass) of fine particles is associated with particles above 0.1 yum, which are
8 associated with the "accumulation" mode identified by Whitby and coworkers (Willeke and
9 Whitby, 1975; Whitby and Sverdrup, 1980). Particle number concentrations are correlated
10 with the volume of particles below 0.1 ju,m. The number concentration of ultrafine particles
11 results from a balance between formation and removal. The rate of removal by coagulation
12 with accumulation mode particles will increase as the number (and mass and volume) of
13 accumulation mode particles increases. Therefore, a correlation between number and
14 accumulation mode volume or mass would not be anticipated. As expected no correlation is
15 found between the total number concentration and the total fine particle mass or volume.
16
17
18 6.9 AMBIENT CONCENTRATIONS OF ULTRA-FINE METALS
19 6.9.1 Introduction
20 Numerous pathways result in "ultra-fine" atmospheric aerosols, particles in the size
21 range around 0.1 m diameter. These include both primary production processes, such as
22 combustion, and secondary processes involving gas-to-particle conversion and subsequent
23 growth by condensation and coagulation to larger particles. There are also numerous
24 pathways that readily remove such particles from this mode, both by deposition and by
25 growth in size into the "accumulation mode", broadly defined as the maximum in particle
26 mass or volume that normally occurs in the range of particle diameters between 0.2 and 0.7
11 m (Whitby et al., 1978). The result is that in ambient conditions, the "ultra-fine" mode is
IS generally indistinct or absent from mass or volume profiles of aerosol particles versus size.
19 However, in some situations the ultra-fine mode can be the dominant size range for selected
50 components of the atmospheric aerosol particles. This is the case for metallic aerosols in
April 1995 6-169 DRAFT-DO NOT QUOTE OR CITE
-------
1 which fine (Dp<2.5 m) concentrations can be dominated by the ultra-fine mode despite the
2 strength of the processes that tend to remove particles from this mode.
3 Removal processes are driven by the ability of ultra-fine particles to rapidly diffuse to
4 surfaces, enhancing loss by deposition and by processes such as coagulation. The high
5 surface area of ultra-fine particles, which is 5 times greater than an equal mass of particles at
6 0.5 m diameter, also enhances growth to the accumulation mode by forming nuclei for
7 condensation of volatile species. For these and other reasons, the mass of ultra-fine particles
8 in the ambient atmosphere is generally much smaller than that of the accumulation mode,
9 where removal rates of particles reach a minimum in non-cloud conditions. However, a
10 distinct ultra-fine mode below 0.1 m diameter has been observed in quasi-ambient samples
11 taken close to combustion sources, sometimes referred to as the "combustion mode" (Whitby
12 etal., 1978).
13 While there is consensus that ultra-fine metals are abundantly produced and emitted into
14 the atmosphere, there are not a lot of data on ambient concentrations of ultra-fine metals.
15 The few direct measurements available can be extended with some confidence using indirect
16 methods; i.e., from particle counting techniques that have size information but no chemical
17 information, or from filter collection methods that have limited size information but detailed
18 compositional information. Nevertheless, it is clear that more data on ultra-fine metals is
19 urgently needed to gain confidence in the spatial and temporal concentration profiles of this
20 key atmospheric component.
21
22 6.9.2 Formation of ultra-fine particles
23 Combustion theory establishes that high temperature processes are generally required to
24 form ultra-fine metallic aerosols. Such processes are usually anthropogenic, although natural
25 fires, volcanic eruptions, and other such events can contribute to ultra-fine transition and
26 heavy metals in some circumstances. Table 6-4, taken from Seeker (1990), gives the
27 temperature of formation of EPA-regulated metals (Federal Register, 1986) as a function of
28 temperature, with and without chlorine available in the combustion process.
April 1995 6-170 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-4. REGULATED METALS AND THE VOLATILITY TEMPERATURE
(SEEKER, 1990)
Metal
Chromium
Nickel
Beryllium
Silver
Barium
Thallium
Antimony
Lead
Selenium
Cadmium
Osmium
Arsenic
Mercury
With
Volatility
temp, (°F)
2935
2210
1930
1660
1560
1330
1220
1160
605
417
105
90
57
no chlorine
Principal
species
CrO2/CrO3
Ni(OH)2
Be(OH)2
Ag
Ba(OH)2
T1203
Sb2O3
Pb
SeO2
Cd
OsO4
As2O3
Hg
With 10%
Volatility
temp, (°F)
2930
1280
1930
1160
1660
280
1220
5
605
417
105
90
57
chlorine in waste
Principal
species
CrO2/CrO3
NiCl2
Be(OH)2
AgCl
BaCl2
T1OH
Sb203
Pb
SeO2
Cd
OsO4
As2O3
Hg
1 Note the dramatic shift in temperature for several elements, including lead, for the
2 chlorine-rich combustion scenario. A similar process has been used to prevent lead from
3 coating surfaces in internal combustion engines using leaded gasoline. The process used
4 chlorine and bromine-containing additives to form compounds such as PbBrCl which then
5 leave the vehicles as ultra-fine aerosols.
6 Numerous theoretical and laboratory studies have shown that the typical size of metals
7 derived from combustion is ultra-fine, (Friedlander, S.K., 1977; Senior, et al., 1982; Seeker
8 et al., 1990). Analysis of particles from coal combustion by Natusch et al. (1974a, b)
9 showed an additional aspect. There is a tendency for the condensing metal vapors to form
10 relatively uniform thickness surface coatings on more refractory particles present in the
11 combustion effluent stream. If the particles upon which the metals coat themselves are
12 crustal, as in coal fly ash, this results in a final particle whose enrichment factor compared to
13 crustal averages depends upon the initial size of the refractory particle-minor for large
April 1995
6-171
DRAFT-DO NOT QUOTE OR CITE
-------
1 particles, extreme for ultra-fine particles (Davison et al., 1975). This result also places the
2 (potentially) toxic metals on the biologically-accessible surface.
3 Thus, the presence of metals in a combustion process such as incineration of biological
4 and chemical wastes or treatment of contaminated soils poses a problem. Raising the
5 temperature of combustion high enough to completely (> 99.99%) destroy the biological and
6 chemical species will also enhance the volatilization of metallic components in the feed stock,
7 requiring more efficient removal methods for ultra-fine and accumulation mode metals.
8 Figure 6-95 shows the enhanced volatilization of metals as the combustion temperature is
9 raised from 1000 degrees F (540 C) to 1800 degrees F (980 C) (Seeker, 1990).
10 The combustion effluent can be partitioned into three components (Seeker, 1990; Barton
11 et al., 1990); emitted (as fly ash), captured (assuming there is an attempt to capture fine
12 particles), and collected in the bottom ash. Assuming no particle removal equipment is in
13 place on the combustion process, emitted particles will include both the "emitted" component
14 and most of the "captured" component. In an uncontrolled incineration facility, 96% of
15 mercury, 88% of cadmium, 58% of lead, and 11% of copper might by emitted into the
16 atmosphere. If control is attempted, the capture efficiency is only 25% for mercury, but is
17 better for most other metals, ranging from 86% for cadmium to 91% for copper (Barton
18 et al., 1990). In addition, the chemical state of the metals in the ultra-fine mode can vary
19 from the more toxic phases (for example, arsenite versus arsenate) as a function of
20 combustion conditions (Chesworth et al. 1994). Thus, we must expect that ultra-fine metallic
21 components will be emitted from high temperature processes in both toxic and less toxic
22 forms.
23
24 6.9.3 Techniques for collecting and analyzing ultra-fine metals
25 Relatively little information exists on concentrations of ultra-fine metal particles in
26 ambient air samples away from combustion sources. There are many reasons. The ultra-fine
27 mode falls off rapidly away from the combustion source,due to the rapid migration of some
28 types of ultra-fine particles into the accumulation mode, and increased dispersion as one
29 moves away from the source. Many sources of ultra-fine metals use tall exhaust stacks,
30 which enhances dispersion. The largest of the ultra-fine particles can overlap the smallest
April 1995 6-172 DRAFT-DO NOT QUOTE OR CITE
-------
M _
E
T
A
L 4Q
E
N 30
R
I
C 20
H
M
E 10
N
T
(%)
•
-
-
m
m
" 1000 F
•
-AsSbPbCuCrZnCd
rh-i-.
Cd
••
pbiaooF
Ml
Sb
TCu
Tzn
~i
|Cr
Figure 6-95. Impact of treatment temperature on the enrichment of metals in the fly
ash after the thermal treatment of soils from a superfund site.
1 particles of the much more abundant accumulation mode, roughly 0.2 to 0.7 m aerodynamic
2 diameter. Particles must be size-separated using a device with a sharp cut point, ususally a
3 multistage physical impactor, that entails problems in particle collection and analysis. Since
4 ultra-fine particles may hard and dry, adhesive coatings are essential in order to avoid
5 particle bounce in the impactors. Particle bounce typically translates coarser particles onto
6 finer stages, contaminating the ultra-fine particles with the enormously more abundant
7 coarser particles. Finally, one can collect only a few monolayers of particles (at most) on
8 the adhesive stages before particle bounce becomes important, assuming the particles
9 themselves are not "sticky". A few monolayers of particles of 0.1 m diameter amounts to
10 only about 50 g/cm2 of total deposit. If one then desires to perform minor or trace
11 elemental analysis of the deposit, one is then faced with analytical requirements that reach
12 picogram (10~12 gm) sensitivities. This clearly limits analytical options.
13 For these reasons, much of the data available on the "ultra-fine" mode does
14 not depend on compositional analysis. Most information on the presence of the ultra-fine
April 1995
6-173
DRAFT-DO NOT QUOTE OR CITE
-------
1 mode is derived from particle counting techniques such as the Electrical Mobility Analyzer
2 (EMA), in situations in which the source is well known (source-enriched). This was the
3 method pioneered in the 1972 ACHEX studies of Los Angeles (Whitby, 1978). Particle
4 counting devices do not normally result in collection of particles in the ultra-fine mode in a
5 manner suitable for compositional analysis, although some of the devices ("particle
6 classifiers") could be modified to provide samples for subsequent compositional analysis, if
7 desired. The same can be argued for devices such as diffusion batteries, but to date little has
8 been done along this line in ambient conditions.
9 Integrated samples of fine particles can be collected on substrates suitable for
10 analysis. While some optical information is available as one approaches the ultra-fine mode,
11 most optical techniques do not work in the ultra-fine size range, which is well below the
12 wavelength of light. A Scanning Electron Microscope (SEM) beam can still resolve particles
13 in the ultra-fine mode although some details are lost. The ultra-fine mode can then be
14 derived by particle counting techniques, either manual or automated, and metal composition
15 can be found by x-ray analysis of the single particles. The enormous gain in signal to noise
16 ratio by selecting individual particles off sets the loss of x-ray sensitivity (typically part per
17 thousand) caused by use of the electron beams to induce the x-rays. SEM and electron
18 microprobe analyses rarely achieve any better than one part per thousand sensitivity, but for
19 single particles, this is often enough to classify them by source. Proton microprobes are, at
20 present, not quite able to operate in the 0.1 m diameter region, but can perform Proton
21 Induced X-ray Emission (PIXE) analysis to one part per million by mass on single particles
22 as small as 0.3 m (Cahill, 1980).
23 Impactors are designed to separate particles by aerodynamic size in such a way
24 as to allow compositional analysis. Yet here, too, the ultra-fine mode poses problems.
25 First, most impactors can not operate effectively in the ultra-fine mode. The Stokes number
26 for separation of a 0.1 m diameter particle from an air stream requires either extremely high
27 jet velocities, extremely low pressures in the gas stream, or both. While such performance
28 can be achieved in a physical impactor, most impactors used for ambient particle collection
29 in the 1970's and early 1980's did not possess this capability. For example, the very popular
30 cyclones and virtual impactors are ineffective below about 0.5 m diameter, and at ambient
31 pressure, are unlikely to ever achieve performance in the ultra-fine regime. The Lundgren-
April 1995 6-174 DRAFT-DO NOT QUOTE OR CITE
-------
1 type impactors widely used in California studies (Lundgren et al., 1972; Flocchini et al.,
2 1976; Barone et al. 1978) used 0.5 m as the lowest cut point. Everything smaller was
3 collected on a filter. The Battelle-type samplers (Mercer, 1964) favored by other groups
4 (Van Grieken et al., 1975) used a lowest cut point of 0.25 m diameter. Thus, while both
5 these units generated copious information on aerosol composition, they could not separate
6 ultra-fine aerosols from accumulation mode aerosols.
7 In the mid-1980's four new impactors were developed capable of resolving
8 ultra-fine aerosols in ambient conditions; the Low Pressure Impactor, (LPI, Hering et al.,
9 1978), the Berner Low Pressure Impactor (BLPI, Berner and Lurzer, 1980), the Davis
10 Rotating-drum Unit for Monitoring impactor, (DRUM. Cahill et al., 1985; Raabe et al.,
11 1988), and the Multiple Orifice Uniform Deposit Impactor (MOUDI, Marple et al. 1986).
12 Battelle-type impactors were also modified to add two size cuts below 0.25 m diameter, but
13 unlike the other four units, no certification of performance has been published to date on its
14 performance in the ultra-fine region. The development of reliable, clean adhesive coatings
15 such as Apiezon™-L grease was also a major advance in the field (Wesolowski et al., 1978,
16 Cahill, 1979), allowing separation of abundant soils from ultra-fine size ranges even in dry,
17 dusty conditions. For nominally PM-10 soils, for example, a ratio of coarse to ultra-fine
18 soils was measured at 6,600:1 at a temperaturesabove 30 °C and low relative humidity, RH
19 below 20 % (Cahill et al., 1985). Performances and specifications of all these units is
20 included in a recent review paper (Cahill and Wakabayashi, 1993)
21 It is important to mention, however, that the motivation for development of
22 this ultra-fine capability was not for extensive studies of ultra-fine metals, but rather to get a
23 more complete picture of the accumulation mode behavior of sulfates, nitrates, organics, and
24 other major components of the fine aerosol mix. Thus, compositional analysis was often
25 limited to these species even when suitable samples had been collected. For example, many
26 LPI samples were collected on stainless steel substrates, ideal for combustion analysis of
27 sulfur, but unsuitable for analysis of transition metals by x-ray techniques.
April 1995 6-175 DRAFT-DO NOT QUOTE OR CITE
-------
1 6.9.4 Observations of ultra-fine metals; Stack and source-enriched
2 aerosols
3 Observations of ultra-fine metals in source or source-enriched situations
4 lessens problems with dilution of the sample and identification of the source. This eases both
5 particle collection and analysis. Figure 6-96 shows the results of such a study on a coal fired
6 power plant (Maenhaut et al., 1993) using the Berner Low Pressure Impactor (BLPI). The
7 extreme volatilization of selenium is clearly seen, which is also confirmed in aircraft
8 sampling of power plant stacks. Note, however, that the enrichment factor is rather constant
9 as a function of particle size for both sulfur and its chemical analog selenium. Other more
10 refractory elements, on the other hand, are strongly enhanced in the ultra-fine mode as
11 compared to coarser modes.
12 The BLPI cuts are as follows: Stage number 1-0.011 m diameter, 2-0.021,
13 3-0.032, 4-0.07, 5-0.17, 6-0.30, 7-0.64, 8-1.4, 9-2.6, 10-5.5, 11-10.7 m. All are for
14 particle density 2.45 g/cm3 and a temperature 120° C, the conditions of stack sampling in the
15 coal fired power plant. Both these figures were normalized to Earth crustal averages. Thus,
16 even a two order of magnitude rise in the normalized concentration may not result in a
17 visible "combustion mode" since the mass of soil falls very rapidly as one moves towards
18 ultra-fine particles. This is exactly what is predicted by the results of Natusch et al. (1974).
19 Thus, source testing confirms combustion theory and the laboratory studies and predicts
20 emissions of metals into the ultra-fine mode from many types of high temperature
21 combustion sources.
22
23 Observations of ultra-fine metals: Ambient aerosols
24 Direct observations
25 Because of the difficulties in sampling and analysis, there are relatively few
26 data on concentrations of ultra-fine (Dp 0.1 m diameter) metals in ambient aerosols. Some
27 quantitative determinations of ambient concentrations have become available in the past
28 15 years, however, generally as a result of a number of short but intensive aerosol studies.
29 Examples include the extensive studies near the Grand Canyon, 1979 (Macias et al., 1981) to
30 the Mohave Studies near the Grand Canyon NP, 1993, the Southern California Air Quality
31 Study (SCAQS), 1985-1987 (Hering et al., 1990, Cahill et al., 1990, Cahill et al., 1992);
April 1995 6-176 DRAFT-DO NOT QUOTE OR CITE
-------
lOOOr
100
8
T3
10
0.1
S«x0.1
S
Ca
-)
Al
-H
SI
-i
K
1000T
"S
o
234587(010
Stage number
Stage number
Figure 6-96. Average normalized concentrations as a function of stage number, for Se,
S, Ca, al., Si, K, Mo, W, Ni, and Cr for five BLPI samples from a coal
fired power plant. The smallest size mode is to the left, Stage number 1,
0.011 to Stage number 11, 10.7 m diameter. Normalization is to average
crustal composition.
1 studies at Shenandoah (1991) and Mt. Rainier (1992) National Parks (Malm et al., 1993,
2 Malm et al. 1994 b, Cahill and Wakabayashi, 1994), and others. While almost all of these
3 studies used several different types of impactors with ultra-fine capabilities, relatively few
4 were analyzed for trace metal content.
5 An example of a persistent ambient ultra-fine mode is shown in Figure 6-97 from data
6 collected at Grand Canyon NP 1984 (Cahill et al., 1987). The ultra-fine mode behaves
7 independently from the accumulation mode, in fact often showing a net anti-correlation in
8 concentrations of sulfur as well as dramatic differences in metals (Table 6-5). The ultra-fine
9 mode in Table 6-6 can be attributed to non-ferrous metal smelting activities in the region
10 (Eldred et al. 1983, Small et al. 1981), which puts the nearest important sources a hundred
11 miles away from the sampling site. The completely different behavior of the accumulation
April 1995
6-177
DRAFT-DO NOT QUOTE OR CITE
-------
400
200
STAGE 8 (FINE)
400- STAGE?
200-
STAGE 6
400
zoo-
A..
STAGE 5
. STAGE 4
i IT
10
i i rrn r
20
I'M i11 r i i
00
AUGUST 1984
Figure 6-97. Fine and ultrafine sulfur at Grand Canyon NP, summer, 1984. The sulfur
peaks on August 15 and August 16 were used for the compositional
analysis in Table 3. Note that the size fractions are inverted, with the
finest, Stage 8, at the top, 0.088-0.15 m diameter. The succeeding stages
are at 0.24, 0.34, 0.56, to the coarsest, Stage 4, 1.15 to 2.4 m. The
first three cut points are somewhat uncertain due to altitude and flow rate
corrections. Final stage configurations are given in Raabe et al. (1989),
which were used for all later studies using the DRUM.
1 and ultra-fine modes in this arid site also shows that mis-sizing by particle bounce is not
2 significant.
3 Table 6-6 presents a summary of more recent data for major EPA-regulated metals
4 (lead, nickel) and other metals, at Long Beach, CA, December, 1987 (SCAQS) and at
5 Shenandoah NP, 1991. The elements span the range from refractory metals like nickel and
6 vanadium to metals with low melting temperatures such as zinc and lead. These data were
7 all taken with the same unit, the Davis Rotating-drum Unit for Monitoring (DRUM) using
8 greased stages and a single orifice impactor (Cahill et al., 1985). The last two stages were
April 1995
6-178
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-5. COMPOSITION OF THE AEROSOLS PRESENT AT GRAND CANYON
NP IN SUMMER 1984, FOR THE TWO SULFATE EPISODES OF AUGUST 15
(ACCUMULATION MODE, STAGE 6) AND AUGUST 16 (ULTRA-FINE MODE,
STAGE 8)
Elements
Sodium
Silicon and Aluminum
Sulfur
Chlorine
Potassium
Calcium
Titanium
Vanadium
Iron and nickel
Copper
Zinc
Arsenic
Bromine
Lead
Stage 8,
0.088-0. 15m
(ng/m3)
420
8
204
208
59
150
2
2
2
100
931
13
2
63
Stage6,
0. 15-0.60 m
(ng/m3)
10
6
392
5
3
5
4
3
2
1
2
2
2
4
1 modified form the Gand Canyon configuration as a result of theoretical and laboratory studies
2 (Raabe et al., 1989), yielding 0.069 to 0.24 m for Stage 8, and 0.24 to 0.34 m diameter for
3 Stage 7.
4 The DRUM data were used for several reasons: the DRUM'S slowly rotating greased
5 stages have a documented ability to handle large amounts of coarse, dry soils without
6 contaminating the ultra-fine stages, (Cahill et al. 1985; Cahill and Wakabayashi, 1992), the
7 elemental data are of unprecedented sensitivity for ambient ultra-fine trace metals (PIXE and
8 synchrotron-XRF), there is a consistency of sampler type and protcols at very different
9 locations, and there is more trace element data from the DRUM than from any other type of
10 unit. These advantages outweigh its disadvantages; the DRUM does not have the ultra-fine
11 sizing detail of either the LPI or BLPI impactor, or the ability to measure mass, ions and
12 organic matter of the MOUDI or BLPI.
April 1995 6-179 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-6. MEASUREMENTS OF FINE AND ULTRA-FINE METALS
Site Name Particle Ultra-fine
Duration Aerodynamic mode
Frequency Diameters From
Dates To
(Dae, micro- 0.069
meters) 0.24
Long Beach, CA Maximum
6 days Element Values
6 samples/day ng/m3
(11,12/87) Vanadium 6.6
Nickel 3.4
(Mean min. Zinc 51
detectable Selenium MDL
limit - 0.3 ng/m3) Lead 199
Sulfur (est)
Shenandoah NP Maximum
21 days Values
6 samples/day ng/m3
(9/91) Vanadium 1.2
Nickel 1.2
(Mean min. Zinc 3.8
detectable Selenium 2.7
limit -0. 15 ng/m3) Lead 50
Sulfur
(est) - estimated from graphs
Ultra-fine Accumulation
mode Mode
From From From From From
To To To To To
0.069 0.24 0.34 0.56 1.15
0.24 0.34 0.56 1.15 2.5
Mean
Values
ng/m3 ng/m3 ng/m3 ng/m3 ng/ni3
2.5 6.1 10.5 12.2 8.6
1.3 4.4 7.7 4.5 0.5
17.6 46.3 140.4 189.4 39
MDL 0.32 3.00 1.40 0.65
71.4 47.6 59.9 69.9 25.4
200 250 350 500 250
Mean
Values
ng/m3 ng/m3 ng/m3 ng/m3 ng/m3
0.24 0.67 0.52 0.30 0.80
0.58 0.48 0.13 0.03 0.01
1.42 2.16 2.60 1.92 1.66
0.14 0.11 0.52 0.35 0.14
5.38 5.49 3.01 10.87 16.06
334 929 1235 1727 101
April 1995
6-180
DRAFT-DO NOT QUOTE OR CITE
-------
1 The analyses were done both by PIXE and by synchrotron-XRF (Cahill et al. 1992),
2 with most of the trace metal data from the latter source. In order to obtain sulfate, multiple
3 sulfur by 3.0. These average values, however, obscure a great deal of structure as a function
4 of time.
5 The variability as a function of size and time is shown in Figure 6-98 for nickel,
6 selenium, and lead in Long Beach, CA as part of the SCAQS studies of 1987. By 1987,
7 much of the lead was no longer automotive, and there are significant changes in the ultra-fine
8 fraction over periods of four to twelve hours. Note the behavior of ultra-fine metals; almost
9 total absence for selenium, partial absence for nickel, and constant presence for lead.
10 Almost all elements at almost every site shows similarly complex behavior. Thus, the
11 summary of Table 6-11 can only include the most basic types of information on fine and
12 ultra-fine metals in the atmosphere.
13 In addition to the limited US data, comparison data have also become available from
14 foreign sources such as from the Kuwaiti oil fires (1991) and a study in Santiago, Chile,
15 (1993). While the former is a unique situation, the Santiago data are especially useful since
16 leaded gasoline is still routinely used in Chile and other countries, generating data impossible
17 to obtain in the United States. Table 6-7 summarizes some of these data for a refractory
18 element, nickel, and a volatile metal, lead.. However, the full data set includes 450 samples
19 of four to six hour duration, each analyzed in five fine size fractions, generally with about
20 20 elements found in each fraction, or approximately 40,000 individual elemental values.
21 Some general observations can be made from the data; first, there is an enormous
22 variation in the concentration of fine and ultra-fine metals, sometimes spanning 4 or 5 orders
23 of magnitude in a few days. Such behavior can be modeled by plumes of particles that
24 sweep over the site episodically, as opposed to area or regional sources. Second, one often
25 finds a mixture of ultra-fine and accumulation mode behaviors. However, these may be
26 physically separated in time and size.
27 Lead in the United States follows a variety of very different patterns. In the rural
28 samples, lead tends to be bimodal, with a coarse component above 1.0 m diameter and a
29 very fine and ultra-fine mode below 0.34 m diameter. This can be modeled by a very fresh
30 ultra-fine mode and a coarser mode associated with resuspended soil. Urban sites, however,
31 both in the U.S. and in Santiago, show a strong ultra-fine mode and an accumulation mode.
April 1995 6-181 DRAFT-DO NOT QUOTE OR CITE
-------
00-
200
00-
20.0
0.0-
10.0
rt
LONG BEACH. CA
SELENIUM
LONG BEACH, CA
LEAD
Figure 6-98. Fine and ultra-fine metals, nickel, selenium, and lead, in Long Beach,
CA, December 10-13, 1987, in four hour increments. Stage 8 is ultra-76-
fme, 0.069-0.24 m, then 0.34, 0.56, 1.15, 2.5 m D(ae).
April 1995
6-182
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-7. MEASUREMENTS OF FINE AND ULTRA-FINE METALS
LEAD AND NICKEL
Site Name Particle Ultra-fine.
Duration Aerodynamic mode
Frequency Diameters
Date Element
(Dae, micrc
meters)
Lead Mean
Maximum
Nickel Mean
Maximum
Accumulation
mode
Long Beach
6 days
4 samples/day
(11/87)
Shenandoah NP Lead Mean
21 days Maximum
6 samples/day Nickel Mean
(9/91) Maximum
Mt. Rainier NP Lead Mean
28 days Maximum
6 samples/day Nickel Mean
(7,8/92) Maximum
From
To
0.069
0.24
71.4
199
1.3
3.4
5.4
50
0.58
1.20
2.3
6
From
To
0.24
0.34
47.6
95
4.4
11.4
5.5
20
0.48
1.60
6.5
15
From
To
0.34
0.56
59.9
129
7.7
15.0
3.0
16
0.13
0.80
2.0
21
From
To
0.56
1.15
69.9
164
4.5
13.4
10.9
70
0.03
1.00
3.4
14
From
To
1.15
2.5
25.4
58
0.5
3.7
16.1
130
0.01
0.14
6.7
29
MDL
0.45
0.22
MDL
0.2
0.09
MDL
0.5
Always less than MDL
MDL 0.4 0.8 0.4
0.7
Santiago, Chile Lead Mean 101 53 38 108 41
14 days
6 samples/day
(9/93)
Kuwait Lead
14 days
4 samples/day Nickel
(6/91)
Maximum 920 340 320 640 270
Maximum 5
18 11 8
0.07
MDL
8
Mean
Maximum
Mean
429.9
2580
1.5
154.2
580
2.5
84.7
128
4.3
44.7
86
3.7
38.1
70
6.0
MDL
0.35
0.22
MDL = minimum dectable limit at 95% confidence level, in ng/m^
April 1995
6-183
DRAFT-DO NOT QUOTE OR CITE
-------
1 Any resuspended soil mode is hidden under the extension of the accumulation mode above
2 1.0 m diameter.
3 Other metals at Long Beach, however, lack a distinct ultra-fine mode all the
4 time (selenium) or part of the time (nickel), merely possessing an accumulation mode that
5 closely mimics sulfates and other secondary species (Cahill et al., 1990). It is well known
6 that nickel and vanadium were derived from high temperature combustion sources, and since
7 each is highly refractory, they will occur primarily in an ultra-fine mode at the source.
8 Thus, the similarity between the distributions of these elements and less refractory elements
9 such as zinc can be understood through a rapid condensation and coagulation of the abundant
10 secondary species around these metals, leading to an accumulation mode distribution as the
11 secondary acidic species hydrate. Clearly, such processes are weaker at dry sites such as the
12 arid west in summer (Table 6-6). On the other hand, Shenandoah NP has a mixture of urban
13 and rural behavior, with occasional sharp peaks of ultra-fine metals (nickel) superimposed on
14 an accumulation mode behavior (sulfur, selenium) but some coarse contribution (lead,
15 vanadium). Only with a detailed study of meteorology and knowledge of emission sources
16 can such ambient behavior be understood.
17
18 Indirect methods
19 Lacking a large body of direct data on ultra-fine metallic aerosols, there are
20 indirect ways to increase our knowledge of such aerosols;
21 1. Combustion studies have established the modes of formation of ultra-fine metallic
22 aerosols, and,
23
24 2. Considerable ambient data exist that, when combined with known combustion
25 processes, yield estimates for the concentration of ultra-fine metallic aerosols by
26 time and locations.
27
28 3. In conditions of low ambient concentrations of particles and low humidity and , the
29 ultra-fine mode has been shown to persist for many hours. (Cahill et al., 1985).
30
31 Thus, the numerous observations of fine (Dp < 2.5 m) metallic aerosols in low
32 humidity conditions can yield estimates of the presence of such metals in the ultra-fine mode
33 and set upper limits on their concentrations. The relatively small number of actual
34 measurements can then serve as tests or as confirmation of our level of understanding of
April 1995 6-184 DRAFT-DO NOT QUOTE OR CITE
-------
1 these biologically important aerosols. As an example, Figure 6-101 shows concentration
2 profiles of sulfur, selenium, zinc, and arsenic, all of which can have ultra-fine modes in the
3 western United States. Arsenic and zinc are annual averages, March, 1993 to February,
4 1994, while the sulfur (for sulfate, times 3.0) and selenium are for summer, 1993. This was
5 done to exhibit the correlation of these elements, which are chemically akin, during the
6 eastern U.S. sulfate maximum each summer. The regional nature of the elements is very
7 evident, as are certain strong sub-regional sources such as the copper smelter region of
8 Arizona and New Mexico (arsenic).
9 The non-urban values shown in Figure 6-99, which are derived from the cleanest areas
10 of the United States, are surprisingly relevant to urban areas in the same region for some of
11 the species. Table 6-8 compares major and minor fine elements at Shenandoah NP, where
12 there are detailed measurements of particle size, and Washington, DC, where such size
13 information is lacking. Summer 1993 is the comparison period. Finally, two western sites
14 are compared, both downwind of Los Angeles; San Gorgonio Wilderness, and Grand Canyon
15 NP.
16
17 Inhalation of ultra-fine metals
18 An extensive literature exists on the deposition of fine metals in the human lung, much
19 of which was derived from laboratory studies, some using radioactive tracer isotopes. But an
20 example of one of the few direct measurement of lung capture by ambient ultra-fine metals is
21 found in Dasaedeleer et al., 1977 and shown in Figure 6-100. The lower cut point is only
22 0.25 m but even so, the increased capture efficiency of the lung for very fine and ultra-fine
23 particles is clearly shown.
24
25
26 6.10 SUMMARY
27 There are few data on ambient concentrations of ultra-fine metals. The few direct
28 measurements can be extended with some confidence using indirect methods; i.e., particle
29 counting techniques that have size information but no chemical information, or filter
30 collection methods that have limited size information but detailed compositional information.
April 1995 6_!85 DRAFT-DO NOT QUOTE OR CITE
-------
I
S
I
.5
O 3
U <
cw
•o
-------
TABLE 6-8. COMPARISON OF SELECTED SPECIES, SHENANDOAH NP, AND
WASHINGTON, DC, SAN GORGONIO WILDERNESS, CA, AND GRAND CANYON
NP, SUMMER, 1993
Shenandoah
Concentration (ng/m^)
Mass - PM-10
Mass - PM-2.5
Composition PM-2.5 Mass
Ammonium sulfate
Ammonium nitrate
Organic matter
Soil
Trace composition (ng/m^)
Nickel
Copper
Zinc
Arsenic
Selenium
Bromine
Lead
Bio-smoke tracer
NP
31.00
22.50
11.80
0.40
2.84
1.41
0.24
1.06
7.93
0.22
1.58
2.14
2.17
8.33
Washington
DC
34.90
26.50
14.60
1.47
5.42
1.55
0.97
3.37
13.90
0.56
2.48
4.18
4.48
<2.00
San Gorgonio
Wilderness
21.70
10.30
2.55
4.44
3.88
0.86
0.18
0.76
3.72
0.16
0.44
3.67
1.36
10.00
Grand Canyon
NP
9.37
4.50
1.09
0.25
1.22
0.63
0.09
0.30
0.63
0.18
0.18
2.11
0.51
32.30
(non-soil fine potassium)
Optical Absorption
(b(abs), 10-6 m-l)
19.60
41.90
13.90
5.40
April 1995 6_187 DRAFT-DO NOT QUOTE OR CITE
-------
c
0
TJ
QJ
O
.C
x
UJ
O
rr
c
g
"o
c
u
c
o
O
05--
02--
01--
005--
0.02
0
•20
-•40
-•60
i
^ Pb
D Br
C!
c
o
80^
to
O
CL
O)
O
--90
92 a;
O
-•94
4, 4-2, 2-1, 1-0.5, 0.5-
0.25, and <0.25 /on particles of size classes 1, 2, 3, 4, 5, and 6,
respectively. Extension of the curve to particles of diameter > 2 pm
(classes 2 and 1) is supported by separate experiments using chalk dust
aerosol.
Source: Dasaedeleer et al., 1977.
April 1995
6-188
DRAFT-DO NOT QUOTE OR CITE
-------
1 Nevertheless, it is clear that more data on ultra-fine metals is urgently needed to gain
2 confidence in the spatial and temporal concentration profiles of this key atmospheric
3 component.
4 Ultra-fine metals are produced by a wide variety of anthropogenic activities and emitted
5 into the ambient air. Ambient concentrations of such metals have been seen not only in urban
6 settings but also at the cleanest sites in the United States. Concentrations are highly variable
7 as a function of site and time. While ultra-fine metals have been seen to persist for many
8 hours, or more, in the clean, dry environment of the arid west, they appear to be removed
9 and/or transformed into the accumulation mode in polluted urban or humid rural sites.
April 1995 6489 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Adams, K. M.; Davis, L. I., Jr.; Japar, S. M.; Finley, D. R.; Cary, R. A. (1990) Measurement of atmospheric
4 elemental carbon: real-time data for Los Angeles during summer 1987. Atmos. Environ. Part A
5 24: 597-604.
6
7 Altshuller, A. P. (1980) Seasonal and episodic trends in sulfate concentrations (1963-1978) in the eastern United
8 States. Environ. Sci. Technol. 14: 1337-1349.
9
10 Appel, B. R.; Colodny, P.; Wesolowski, J. J. (1976) Analysis of carbonaceous materials in southern California
11 atmospheric aerosols. Environ. Sci. Technol. 10: 359-363.
12
13 Appel, B. R.; Kothny, E. L.; Hoffer, E. M.; Hidy, G. M.; Wesolowski, J. J. (1978) Sulfate and nitrate data
14 from the California Aerosol Characterization Experiment (ACHEX). Environ. Sci. Technol. 12: 418-425.
15
16 Appel, B. R.; Hoffer, E. M.; Kothny, E. L.; Wall, S. M.; Haik, M.; Knights, R. L. (1979) Analysis of
17 carbonaceous material in southern California atmospheric aerosols. 2. Environ. Sci. Technol. 13: 98-104.
18
19 Appel, B. R.; Tokiwa, Y.; Haik, M. (1981) Sampling of nitrates in ambient air. Atmos. Environ. 15: 283-289.
20
21 Appel, B. R.; Tokiwa, Y.; Hsu, J.; Kothny, E. L.; Hahn, E. (1985) Visibility as related to atmospheric aerosol
22 constituents. Atmos. Environ. 19: 1525-1534.
23
24 Ashbaugh et al. (1989) Estimating fluxes from California's dry deposition monitoring data. Presented at: 82nd
25 annual meeting of the Air & Waste Management Association; June; Anaheim, CA. Pittsburgh, PA: Air &
26 Waste Management Association; paper no. 89-65.3.
27
28 Barone, J. B.; Cahill, T. A.; Eldred, R. A.; Flocchini, R. G.; Shadoan, D. J.; Dietz, T. M. (1978) A
29 multivariate statistical analysis of visibility degradation at four California cities. Atmos. Environ.
30 12: 2213-2221.
31
32 Barton, R. G.; Clark, W. D.; Seeker, W. R. (1990) In: Proceedings of the 1st international conference on toxic
33 combustion byproducts.
34
35 Bennett, R. L.; Stockburger, L.; Barnes, H. M. (1994) Comparison of sulfur measurements from a regional fine
36 particle network with concurrent acid modes network results. Atmos. Environ. 28: 409-419.
37
38 Berner, A.; Lurzer, C. (1980) J. Phys. Chem. 84: 2079-2083.
39
40 Bowman, H. R.; Conway, J. G.; Asaro, F. (1972) Atmospheric lead and bromine concentration in Berkeley,
41 Calif. (1963-70). Environ. Sci. Technol. 6: 558-560.
42
43 Brauer, M.; Koutrakis, P.; Spengler, J. D. (1989) Personal exposures to acidic aerosols and gases. Environ. Sci.
44 Technol. 23: 1408-1412.
45
46 Brauer, M.; Koutrakis, P.; Keeler, G. J.; Spengler, J. D. (1991) Indoor and outdoor concentrations of inorganic
47 acidic aerosols and gases. J. Air Waste Manage. Assoc. 41: 171-181.
48
49 Burton, R. M.; Liu, L. J. S.; Wilson, W. E.; Koutrakis, P. (1995) Comparison of aerosol acidity in urban and
50 semi-rural environments. Atmos. Environ.: submitted.
51
52 Cadle, S. H. (1985) Seasonal variations in nitric acid, nitrate, strong aerosol acidity, and ammonia in an urban
53 area. Atmos. Environ. 19: 181-188.
54
April 1995 6-190 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cadle, S. H.; Dasch, J. M. (1988) Wintertime concentrations and sinks of atmospheric paniculate carbon at a
2 rural location in northern Michigan. Atmos. Environ. 22: 1373-1381.
3
4 Cadle, S. H.; Countess, R. J.; Kelly, N. A. (1982) Nitric acid and ammonia in urban and rural locations.
5 Atmos. Environ. 16: 2501-2506.
6
7 Cahill, T. A. (1979) Comments on surface coating for Lundgren-type impactors. In: Lundgren, D. A.;
8 Lippmann, M.; Harris, F. S., Jr.; Clark, W. E.; Marlow, W. H.; Durham, M. D., eds. Aerosol
9 measurement: [papers from a workshop]; March 1976; Gainesville, FL. Gainesville, FL: University
10 Presses of Florida; pp. 131-134.
11
12 Cahill, T. A. (1980) Proton microbes and particle-induced X-ray analytical systems. Ann. Rev. Nucl. Particle
13 Sci. 30: 211-252.
14
15 Cahill, T. A.; Wakabayashi, P. (1993) Compositional analysis of size-segregated aerosol samples. In: Newman,
16 L., ed. Measurement challenges in atmospheric chemistry. Washington, DC: American Chemical Society;
17 pp. 211-228. (Advances in chemistry series no. 232).
18
19 Cahill, T. A.; Wakabayashi, P. H.; James, T. A. (n.d.) Chemical states of sulfate at Shenandoah National Park
20 during summer, 1991.
21
22 Cahill, T. A.; Goodart, C.; Nelson, J. W.; Eldred, R. A.; Nasstrom, J. S.; Feeney, P. J. (1985) In: Ariman,
23 T.; Veziroglu, T. N., eds. Proceedings of the international symposium on particulate and multi-phase
24 processes. Washington, DC: Hemisphere Publishing Corporation; pp. 319-325.
25
26 Cahill, T. A.; Eldred, R. A.; Feeney, P. A. (1986) Particulate monitoring and data analysis for the National
27 Park Service 1982-1985. Davis, CA: University of California.
28
29 Cahill, T. A.; Feeney, P. J.; Eldred, R. A. (1987) In: Proceedings of the 4th international PIXE conference;
30 June 1986; Tallahassee, FL. Nucl. Instrum. Methods Phys. Res. B22: 344-348.
31
32 Cahill, T. A.; Eldred, R. A.; Motallebi, N.; Malm, W. C. (1989) Indirect measurement of hydrocarbon aerosols
33 across the United States by nonsulfate hydrogen-remaining gravimetric mass correlations. Aerosol Sci.
34 Technol. 10: 421-429.
35
36 Cahill, T. A.; Surovik, M.; Wittmeyer, I. (1990) Visibility and aerosols during the 1986 Carbonaceous Species
37 Methods Comparison Study. Aerosol Sci. Technol. 12: 149-160.
38
39 Cahill, T. A.; Matsumura, R. T.; Wakabayashi, P. (1992) Presented at: 85th annual meeting and exhibition of
40 the Air & Waste Management Association; June; Kansas City, MO. Pittsburgh, PA: Air & Waste
41 Management Association.
42
43 Cahill, T. A.; Wilkinson, K.; Schnell, R. (1992) J. Geophys. Res. [Atmos.] 97: 14513-14520.
44
45 Cass, G. R. (1979) On the relationship between sulfate air quality and visibility with examples in Los Angeles.
46 Atmos. Environ. 13: 1069-1084.
47
48 Cass, G. R.; McRae, G. J. (1983) Source-receptor reconciliation of routine air monitoring data for trace metals:
49 an emission inventory assisted approach. Environ. Sci. Technol. 17: 129-139
50
51 Cass, G. R.; Conklin, M. H.; Shah, J. J.; Huntzicker, J. J.; Macias, E. S. (1984) Elemental carbon
52 concentrations: estimation of an historical data base. Atmos. Environ. 18: 153-162.
53
April 1995 6_191 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chesworth, S.; Yang, G.; Chang, D.; Jones, A. D.; Kelley, P. B.; Kennedy, I. M. (1994) Combust. Flame
2 98: 259-266.
3
4 Chow, J. C. (1985) Development of a composite modeling approach to assess air pollution source/receptor
5 relationship [dissertation]. Cambridge, MA: Harvard University.
6
7 Chow, J. C.; Watson, J. G. (1988) Summary of paniculate databases for receptor modeling in the United States.
8 In: Watson, J. G., ed. Receptor models in air resources management. Pittsburgh, PA: Air and Waste
9 Management Association.
10
11 Chow, J. C.; Watson, J. G.; Frazier, C. A.; Egami, R. T.; Goodrich, A.; Ralph, C. (1988) Spatial and
12 temporal source contributions to PM10 and PM2 5 in Reno, NV. In: Mathai, C. V.; Stonefield, D. F.,
13 eds. PM10: implementation of standards. Pittsburgh, PA: Air Pollution Control Association; pp. 438-457.
14
15 Chow, J. C.; Watson, J. G.; Pritchett, L.; Lowenthal, D.; Frazier, C.; Neuroth, G.; Evans, K.; Moon, D.
16 (1990) Wintertime visibility in Phoenix, Arizona. Presented at: the 83rd annual meeting & exhibition;
17 June; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association; paper no. 90-66.6.
18
19 Chow, J. C.; Liu, C. S.; Cassmassi, J.; Watson, J. G.; Lu, Z.; Pritchett, L. C. (1992) A neighborhood-scale
20 study of PM10 source contributions in Rubidoux, California. Atmos. Environ. Part A 26: 693-706.
21
22 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Frazier, C.; Hinsvark, B.; Prichett, L.; Neuroth, G. (1992)
23 Wintertime PM10 and PM2 5 mass and chemical compositions in Tucson, Arizona. In: Chow, J. C.; Ono,
24 D. M., eds. Transactions: PM10 standards and nontraditional paniculate source controls. Pittsburgh, PA:
25 Air and Waste Management; pp. 231-243.
26
27 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L.
28 W. (1992) PM10 source apportionment in California's San Joaquin Valley. Atmos. Environ. Part A
29 26: 3335-3354.
30
31 Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L.
32 W. (1993) PM10 and PM2 5 compositions in California's San Joaquin Valley. Aerosol Sci. Technol.
33 18: 105-128.
34
35 Chow, J. C.; Watson, J. G.; Ono, D. M.; Mathai, C. V. (1993) PM10 standards and nontraditional paniculate
36 source controls: a summary of the A&WMA/EPA international specialty conference. Air Waste
37 43: 74-84.
38
39 Chow et al. (1994) Planning for SJVAQS/AUSPEX paniculate matter and visibility sampling and analysis.
40 In: Solomon, P., ed. Planning and managing regional air quality. Boca Raton, FL: CRC Press, Inc.
41
42 Chow, J. C.; Watson, J. G.; Fujita, E. M.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994) Temporal and
43 spatial variations of PM2 5 and PM10 aerosol in the Southern California Air Quality Study. Atmos.
44 Environ. 28: 2061-2080.
45
46 Chow, J. C.; Fairley, D.; Watson, J. G.; DeMandel, R.; Fujita, E. M.; Lowenthal, D. H.; Lu, Z.; Frazier, C.
47 A.; Long, G.; Cordova, J. (1995) Source apportionment of wintertime PM10 at San Jose, CA. J. Am.
48 Soc. Civ. Eng.: in print.
49
50 Chow, J. C.; Watson, J. G.; Lu, Z.; Lowenthal, D. H.; Frazier, C. A.; Solomon, P. A.; Thuiller, R. H.;
51 Magliano, K. (1995) Descriptive analysis of PM2 5 and PM10 regionally representative locations during
52 SJVAQS/AUSPEX. Atmos. Environ.: in press.
53
April 1995 6-192 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chu, L.-C.; Macias, E. S. (1981) Carbonaceous urban aerosol—primary or secondary? In: Macias, E. S.;
2 Hopke, P. K., eds. Atmospheric aerosol: source/air quality relationships; based on a symposium jointly
3 sponsored by the Divisions of Nuclear Chemistry and Technology and Environmental Chemistry at the
4 180th national meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington,
5 DC: American Chemical Society; pp. 251-268. (Comstock, M. J., ed. ACS symposium series: 167).
6
7 Coboura, W. G.; Husar, R. B. (1982) Diurnal and seasonal patterns of paniculate sulfur and sulfuric acid in
8 St. Louis, July 1977-June 1978. Atmos. Environ. 16: 1441-1450.
9
10 Conklin, M. H.; Cass, G. R.; Chu, L.-C.; Macias, E. S. (1981) Wintertime carbonaceous aerosols in Los
11 Angeles: an exploration of the role of elemental carbon. In: Macias, E. S.; Hopke, P. K., eds.
12 Atmospheric aerosols: source/air quality relationships; based on a symposium jointly sponsored by the
13 Divisions of Nuclear Chemistry and Technology and Environmental Chemistry at the 180th national
14 meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington, DC: American
15 Chemical Society; pp. 235-250. (Comstock, M. J., ed. ACS symposium series: 167).
16
17 Conner, T. L.; Miller, J. L.; Willis, R. D.; Kellogg, R. D.; Dann, T. F. (1993) Source apportionment of fine
18 and coarse particles in southern Ontario, Canada. Presented at: 86th annual meeting and exhibition of the
19 Air & Waste Management Association; June; Denver, CO. Pittsburgh, PA: Air & Waste Management
20 Association; paper no. 93-TP-58.05.
21
22 Countess, R. J.; Wolff, G. T.; Cadle, S. H. (1980) The Denver winter aerosol: a comprehensive chemical
23 characterization. J. Air Pollut. Control Assoc. 30: 1194-1200.
24
25 Countess, R. J.; Cadle, S. H.; Groblicki, P. J.; Wolff, G. T. (1981) Chemical analysis of size-segregated
26 samples of Denver's ambient particulate. J. Air Pollut. Control Assoc. 31: 247-252.
27
28 Cronn, D. R.; Charlson, R. J.; Knights, R. L.; Crittenden, A. L.; Appel, B. R. (1977) A survey of the
29 molecular nature of primary and secondary components of particles in urban air by high-resolution mass
30 spectrometry. Atmos. Environ. 11: 929-937.
31
32 Davidson, C. L; Osborn, J. F. (1984) The sizes of airborne trace metal-containing particles. In: Nriagu, J. O.;
33 Davidson, C. I., eds. Toxic metals in the atmosphere. New York, NY: John Wiley & Sons. (Advances in
34 environmental science and technology: v. 17).
35
36 Davis, B. L.; Johnson, L. R.; Stevens, R. K.; Courtney, W. J.; Safriet, D. W. (1984) The quartz content and
37 elemental composition of aerosols from selected sites of the EPA inhalable particulate network. Atmos.
38 Environ. 18: 771-782.
39
40 Davison, R. L.; Natusch, D. F. S.; Wallace, J. R.; Evans, C. A., Jr. (1975) Environ. Sci. Technol. 9: 862.
41
42 Desaedeleer, G. G.; Winchester, J. W.; Akselsson, K. R. (1977) Nucl. Instrum. Methods 142: 97-99.
43
44 Desert Research Institute, (n.d.) [Unpublished data].
45
46 Dockery, D. W.; Schwartz, J.; Spengler, J. D. (1992) Air pollution and daily mortality: associations with
47 particulates and acid aerosols. Environ. Res. 59: 362-373.
48
49 Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.;
50 Speizer, F. E. (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J.
51 Med. 329: 1753-1759.
52
53 Dresser, A. L. (1988) A dispersion and receptor model analysis of western community's PM-10 problem.
54 JAPCA38: 1419-1421.
April 1995 6-193 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dronamraju, M.; Peters, L. K.; Carmichael, G. R.; Kasibhatla, P.; Cho, S.-Y. (1988) An Eulerian
2 transport/transformation/removal model for SO2 and sulfate—III. comparison with the July 1974 SURE
3 database. Atmos. Environ. 22: 2003-2011.
4
5 Dzubay, T. G. (1980) Chemical element balance method applied to dichotomous sampler data. Ann. N. Y. Acad.
6 Sci. 338: 126-144.
7
8 Dzubay, T. G.; Stevens, R. K.; Lewis, C. W.; Hern, D. H.; Courtney, W. J.; Tesch, J. W.; Mason, M. A.
9 (1982) Visibility and aerosol composition in Houston, Texas. Environ. Sci. Technol. 16: 514-525
10
11 Dzubay, T. G.; Stevens, R. K.; Haagenson, P. L. (1984) Composition and origins of aerosol at a forested
12 mountain in Soviet Georgia. Environ. Sci. Technol. 18: 873-883.
13
14 Dzubay, T. G.; Stevens, R. K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E.; Courtney, W. J. (1988) A
15 composite receptor method applied to Philadelphia aerosol. Environ. Sci. Technol. 22: 46-52
16
17 Einfeld, W.; Dattner, S. (1988) The Dallas winter visibility study. Austin, TX: Texas Air Control Board.
18
19 Eldering, A.; Cass, G. R. (1994) A source-oriented model for air pollutant effects on visibility. J. Geophys. Res.
20 [Atmos.]: submitted.
21
22 Eldred, R. A. (1994) Comparison of selenium and sulfur in the IMPROVE network. J. Air Waste Manage.
23 Assoc.: submitted.
24
25 Eldred, R. A.; Cahill, T. A. (1994) Trends in elemental concentrations of fine particles at remote sites in the
26 United States of America. Atmos. Environ. 28: 1009-1019.
27
28 Eldred, R. A.; Cahill, T. A. (1994) J. Air Waste Manage. Assoc.: submitted.
29
30 Eldred, R. A.; Ashbaugh, L. L.; Cahill, T. A.; Flocchini, R. G.; Pitchford, M. L. (1983) Sulfate levels in the
31 southwest during the 1980 copper smelter strike. J. Air Pollut. Control Assoc. 33: 110-113.
32
33 Eldred, R. A.; Cahill, T. A.; Feeney, P. J.; Malm, W. C. (1987) Regional patterns in paniculate matter from
34 the National Park Service Network, June 1982 to May 1986. In: Bhardwaja, P. S., ed. Visibility
35 protection research and policy aspects, pp. 386-396.
36
37 Eldred, R. A.; Cahill, T. A.; Pitchford, M.; Malm, W. C. (1988) IMPROVE—a new remote area paniculate
38 monitoring system for visibility studies. Presented at: the 81st annual meeting of the Air Pollution
39 Control Association; June; Dallas, TX. Pittsburgh, PA: Air Pollution Control Association; paper no. 88-
40 54.3.
41
42 Eldred, R. A.; Cahill, T. A.; Wilkinson, L. K.; Feeney, P. J.; Chow, J. C.; Malm, W. C. (1989) Measurement
43 of fine particles and their chemical components in the IMPROVE/NPS networks. In: Mathai, C. V., ed.
44 Visibility and fine particles; pp. 187-196.
45
46 Eldred, R. A.; Cahill, T. A.; Feeney, P. J. (1994) Regional patterns of selenium and other trace elements in the
47 IMPROVE network. In: Proceedings of the international specialty conference on aerosol and atmospheric
48 optics: radiative balance and visual air quality: v. A. Pittsburgh, PA: Air and Waste Management
49 Association.
50
51 Eldred, R. A.; Cahill, T. A.; Flocchini, R. G. (1994) Comparison of PM10 and PM2 5 aerosols in the IMPROVE
52 network. Proceedings of the international specialty conference on aerosol and atmospheric optics:
53 radiative balance and visual air quality, volume A. Air Waste: submitted.
54
April 1995 6-194 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ellestad, T. G.; Knapp, K. T. (1988) The sampling of reactive atmospheric species by transition-flow reactor:
2 application to nitrogen species. Atmos. Environ. 22: 1595-1600.
3
4 Fairley, D. (1990) The relationship of daily mortality to suspended particulates in Santa Clara county, 1980-86.
5 Environ. Health Perspect. 89: 159-168.
6
7 Federal Register. (1986) Hazardous waste management system; land disposal restrictions. F. R. (November 7)
8 51: 40572-40654.
9
10 Federal Register. (1989) F. R. (February 6) 54: 5746.
11
12 Fekete, K. E.; Gyenes, L. (1993) Regional scale transport model for ammonia and ammonium. Atmos. Environ.
13 Part A 27: 1099-1104.
14
15 Ferman, M. A.; Wolff, G. T.; Kelly, N. A. (1981) The nature and sources of haze in the Shenandoah
16 Valley/Blue Ridge Mountains area. J. Air Pollut. Control Assoc. 31: 1074-1082.
17
18 Flocchini, R. G.; Cahill, T. A.; Shadoan, D. J.; Lange, S. J.; Eldred, R. A.; Feeney, P. J.; Wolfe, G. W.;
19 Simmeroth, D. C.; Suder, J. K. (1976) Monitoring California's aerosols by size and elemental
20 composition. Environ. Sci. Technol. 10: 76-82.
21
22 Flocchini, R. G.; Cahill, T. A.; Eldred, R. A.; Feeney, P. J. (1989) Paniculate sampling in the northeast, a
23 description of the Northeast States for Coordinated Air Use Management (NESCAUM) network. In:
24 Mathai, C. V., ed. Visibility and fine particles, pp. 197-206.
25
26 Forrest, J.; Tanner, R. L.; Spandau, D.; D'Ottavio, T.; Newman, L. (1980) Determination of total inorganic
27 nitrate utilizing collection of nitric acid on NaCl-impregnated filters. Atmos. Environ. 14: 137-144.
28
29 Frazier, C. A. (1989) Letter report on mass and chemical concentrations and babs measurements for Phoenix
30 Pilot Study samples. Phoenix, AZ: Arizona Department of Environmental Quality.
31
32 Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John Wiley
33 & Sons, Inc.
34
35 Gebhart, K. A.; Malm, W. C. (1987) Source apportionment of paniculate sulfate concentrations at three National
36 Parks in the Eastern United States. In: Bhardwaja, P. S., ed. Visibility protection research and policy.
37 pp. 898-913.
38
39 Gebhart, K. A.; Malm, W. C.; Day, D. (1994) Examination of the effects of sulfate acidity and relative humidity
40 on light scattering at Shenandoah National Park. Atmos. Environ. 28: 841-849.
41
42 Gillani, N. V.; Kohli, S.; Wilson, W. E. (1981) Gas-to-particle conversion of sulfur in power plant plumes—I.
43 parametrization of the conversion rate for dry, moderately polluted ambient conditions. Atmos. Environ.
44 15: 2293-2313.
45
46 Gillette, D. A.; Sinclair, P. C. (1990) Estimation of suspension of alkaline material by dust devils in the United
47 States. Atmos. Environ. Part A 24: 1135-1142.
48
49 Gregory, J. M.; Peterson, R. E.; Lee, J. A.; Wilson, G. R. (1994) Modeling wind and relative humidity effects
50 on air quality. In: Proceedings of the international specialty conference on aerosol and atmospheric
51 optics: radiative balance and visual air quality: v. A. Pittsburgh, PA: Air and Waste Management
52 Association.
53
April 1995 6-195 DRAFT-DO NOT QUOTE OR CITE
-------
1 Groblicki, P. J.; Wolff, G. T.; Countess, R. J. (1981) Visibility-reducing species in the Denver "brown
2 cloud"—I. relationships between extinction and chemical composition. In: White, W. H.; Moore, D. J.;
3 Lodge, J. P., Jr., eds. Plumes and visibility: measurements and model components, proceedings of the
4 symposium; November 1980; Grand Canyon National Park, AZ. Atmos. Environ. 15: 2473-2484.
5
6 Grosjean, D. (1983) Polycyclic aromatic hydrocarbons in Los Angeles air from samples collected on Teflon,
7 glass and quartz filters. Atmos. Environ. 17: 2565-2573.
8
9 Grosjean, D.; Bytnerowicz, A. (1993) Nitrogeneous air pollutants at a southern California mountain forest smog
10 receptor site. Atmos. Environ. Part A 27: 483-492.
11
12 Gundel, L. (1979) Discussion. In: Novakov, T., ed. Proceedings: carbonaceous particles in the atmosphere;
13 March 1978; Berkeley, CA. Berkeley, CA: University of California, Lawrence Berkeley Laboratory; p.
14 91; report no. LBL-9037.
15
16 Harrison, P. R.; Draftz, R. G.; Murphy, W. H. (1976) Identification and impact of Chicago's ambient suspended
17 dust. In: Atmosphere-surface exchange of particulate and gaseous pollutants (1974): proceedings of a
18 symposium; September 1974; Richland, WA. Oak Ridge, TN: Energy Research and Development
19 Administration; pp. 540-556. (ERDA symposium series 38). Available from: NTIS, Springfield, VA;
20 CONF 740921.
21
22 Hering, S. V.; Friedlander, S. K. (1982) Origins of aerosol sulfur size distributions in the Los Angeles basin.
23 Atmos. Environ. 16: 2647-2656.
24
25 Hering, S. V.; McMurry, P. H. (1991) Optical counter response to monodisperse atmospheric aerosols. Atmos.
26 Environ. Part A 25: 463-468.
27
28 Hering, S. V.; Flagan, R. C.; Friedlander, S. K. (1978) Design and evaluation of new low-pressure impactor. I.
29 Environ. Sci. Technol. 12: 667-673.
30
31 Hering, S. V.; Appel, B. R.; Cheng, W.; Salaymeh, F.; Cadle, S. H.; Mulawa, P. A.; Cahill, T. A.; Eldred,
32 R. A.; Surovik, M.; Fitz, D.; Howes, J. E.; Knapp, K. T.; Stockburger, L.; Turpin, B. J.; Huntzicker,
33 J. J.; Zhang, X.-Q.; McMurry, P. H. (1990) Comparison of sampling methods for carbonaceous aerosols
34 in ambient air. Aerosol Sci. Technol. 12: 200-213.
35
36 Hidy, G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. (1980) The character and
37 origins of smog aerosols: a digest of results from the California Aerosol Characterization Experiment
38 (ACHEX). New York, NY: John Wiley & Sons.
39
40 Hildemann, L. M.; Markowski, G. R.; Cass, G. R. (1991) Chemical composition of emissions from urban
41 sources of fine organic aerosol. Environ. Sci. Technol. 25: 744-759.
42
43 Hildemann, L. M.; Cass, G. R.; Mazurek, M. A.; Simoneit, B. R. T. (1993) Mathematical modeling of urban
44 organic aerosol: properties measured by high-resolution gas chromatography. Environ. Sci. Technol.
45 27: 2045-2055.
46
47 Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Seasonal trends in Los Angeles
48 ambient organic aerosol observed by high-resolution gas chromatography. Aerosol Sci. Technol.
49 20: 303-317.
50
51 Hopke, P. K. (1985) Receptor modeling in environmental chemistry. New York, NY: John Wiley and Sons.
52
April 1995 6-196 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Houck et al. (1992) Source apportionment—Pocatello, Idaho PM10 nonattainment area. In: Chow; Ono, eds.
PM10 standards and nontraditional paniculate source controls. Pittsburgh, PA: Air & Waste Management
Association; p. 219.
Husar, R. B. (1991) Historical visibility trends. In: Irving, P. M., ed. Acid deposition: state of science and
technology, volume III: terrestrial, materials, health and visibility effects. Washington, DC: The National
Acid Precipitation Assessment Program; pp. 24-68—24-76. (Trijonis, J. C.; Malm, W. C.; Pitchford,
M.; White, W. H. Visibility: existing and historical conditions—causes and effects: state of science and
technology report no. 24).
Husar, R. B.; Frank, N. H. (1991) Interactive exploration and analysis of EPA's Aerometric Information and
Retrieval System (AIRS) data sets. Presented at: the 84th annual meeting of the Air and Waste
Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air & Waste Management.
Husar, R. B.; Holloway, J. M. (1984) The properties and climate of atmospheric haze. In: Ruhnke, L. H.;
Deepak, A., eds. Hygroscopic aerosols. Hampton, VA: A. Deepak Publishing.
Husar, R. B.; Poirot, R. (1992) Exploration of the AIRS database: weekly cycle of ozone and PM10 aerosols.
Presented at: 85th annual meeting of the Air & Waste Management Association; June; Kansas City, MO.
Pittsburgh, PA: Air & Waste Management Association; paper no. 92-77.06.
Husar, R. B.; Stowe, L. L. (1994) Tropospheric aerosols over the oceans, submitted.
Husar, R. B.; Wilson, W. E., Jr. (1986) Climatic trends over North America: potential role of aerosols. In: Lee,
S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk assessment and control
strategies, proceedings of the second U.S.-Dutch international symposium; May 1985; Williamsburg,
VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 821-833.
Husar, R. B.; Wilson, W. E. (1993) Haze and sulfur emission trends in the eastern United States. Environ. Sci.
Technol. 27: 12-16.
Husar, R. B.; Patterson, D. E.; Holloway, J. M.; Wilson, W. E.; Ellestad, T. G. (1979) Trends of eastern U.S.
haziness since 1948. In: Preprints for the fourth symposium on turbulence, diffusion, and air pollution;
January; Reno, NV. Boston, MA: American Meteorological Society; pp. 249-256.
Husar, R. B.; Holloway, J. M.; Patterson, D. E.; Wilson, W. E. (1981) Spatial and temporal pattern of eastern
U.S. haziness: a summary. Atmos. Environ. 15: 1919-1928.
Husar, R. B.; Oberman, T.; Hutchins, E. A. (1990) Environmental Informatics: implementation through the
Voyager data exploration software. Presented at: the 83rd annual meeting of the Air and Waste
Management Association; June; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association.
Husar, R. B.; Elkins, J. B.; Wilson, W. E. (1994) US visibility trends, 1960-1992. Presented at: the 87th annual
meeting of the Air and Waste Management Association; June; Cincinnati, OH. Pittsburgh, PA: Air &
Waste Management Association.
Irving, P. M., ed. (1991) Acid deposition: state of science and technology, volume III: terrestrial, materials,
health and visibility effects. Washington, DC: The National Acid Precipitation Assessment Program.
Jaenicke, R. (1980) Washington, DC: National Academy of Science; pp. 97.
Javitz, H. S. (1988) Feasibility study of receptor modeling for apportioning utility contributions to air
constituents, deposition quality and light extinction [draft report]. Palo Alto, CA: Electric Power
Research Institute.
April 1995
6-197
DRAFT-DO NOT QUOTE OR CITE
-------
1 John, W.; Kaifer, R.; Rahn, K.; Wesolowski, J. J. (1973) Trace element concentrations in aerosols from the San
2 Francisco Bay area. Atmos. Environ. 7: 107-118.
3
4 John, W.; Wall, S. M.; Ondo, J. L. (1988) A new method for nitric acid and nitrate aerosol measurement using
5 the dichotomous sampler. Atmos. Environ. 22: 1627-1635.
6
7 John, W.; Wall, S. M.; Ondo, J. L.; Wmklmayr, W. (1989) Acidic aerosol size distributions during SCAQS.
8 Sacramento, CA: California Air Resources Board.
9
10 John, W.; Wall, S. M.; Ondo, J. L.; Winklmayr, W. (1990) Modes in the size distributions of atmospheric
11 inorganic aerosol. Atmos. Environ. Part A 24: 2349-2359.
12
13 Johnson, R. L.; Shah, J. J.; Gary, R. A.; Huntzicker, J. J. (1981) An automated thermal-optical method for the
14 analysis of carbonaceous aerosol. In: Macias, E. S.; Hopke, P. K., eds. Atmospheric aerosol: source/air
15 quality relationships; based on a symposium jointly sponsored by the Divisions of Nuclear Chemistry and
16 Technology at the 180th national meeting of the American Chemical Society; August 1980; Las Vegas,
17 NV. Washington, DC: American Chemical Society; pp. 223-233. (Comstock, M. J., ed. ACS
18 symposium series: 167).
19
20 Johnson, B. J.; Huang, S. C.; LeCave, M.; Porterfield, M. (1994) Seasonal trends of nitric acid, paniculate
21 nitrate, and paniculate sulfate concentrations at a southwestern U.S. mountain site. Atmos. Environ.
22 28: 1175-1179.
23
24 Joseph, J. H.; Metza, J.; Malm, W. C.; Pitchford, M. L. (1987) Plans for IMPROVE: a federal program to
25 monitor visibility in Class I areas. In: Bhardwaja, P. J., ed. Visibility protection: research and policy
26 aspects. Pittsburgh, PA: Air Pollution Control Association.
27
28 Kadowaki, S. (1976) Size distribution of atmospheric total aerosols, sulfate, ammonium and nitrate particulates in
29 the Nagoya area. Atmos. Environ. 10: 39-43.
30
31 Kao, A. S.; Friedlander, S. K. (1994) Chemical signatures of the Los Angeles aerosol (d,2<3.5 /zm). Aerosol
32 Sci. Technol. 21: 283-293.
33
34 Kao, A. S.; Friedlander, S. K. (1995) Frequency distributions of PM10 chemical components and their sources.
35 Environ. Sci. Technol. 29: 19-28.
36
37 Karch, R.; Kruisz, C.; Reischl, G. P.; Winklmayr, W. (1987) Continuous measurement of ultrafine particle size
38 distributions in the range from 2 to 200 nm from the Southern California Air Quality Study, June-July
39 1987 [preliminary report]. Vienna, Austria: Institut fur Experimentalphysik der Universitat Wein.
40
41 Keeler, G. J.; Spengler, J. D.; Koutrakis, P.; Allen, G. A.; Raizenne, M.; Stern, B. (1990) Transported acid
42 aerosols measured in southern Ontario. Atmos. Environ. Part A 24: 2935-2950.
43
44 Keeler, G. J.; Spengler, J. D.; Castillo, R. A. (1991) Acid aerosol measurements at a suburban Connecticut site.
45 Atmos. Environ. Part A 25: 681-690.
46
47 Kim et al. (1992) Source apportionment study for state implementation plant development in the Coachella
48 Valley. In: Chow; Ono, eds. PM10 standards and nontraditional paniculate source controls. Pittsburgh,
49 PA: Air & Waste Management Association; p. 979.
50
51 Kinney, P. L.; Ozkaynak, H. (1991) Associations of daily mortality and air pollution in Los Angeles County.
52 Environ. Res. 54: 99-120.
53
April 1995 6-198 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kleinman, M. T.; Tomczyk, C.; Leaderer, B. P.; Tanner, R. L. (1979) Inorganic nitrogen compounds in New
2 York City air. Ann. N. Y. Acad. Sci. 322: 114-123.
3
4 Knapp, K. T.; Ellestad, T. G. (1990) Nitric acid and paniculate measurements by transition-flow reactor during
5 the carbonaceous species methods comparison study. Aerosol Sci. Technol. 12: 39-43.
6
7 Koutrakis, P.; Kelly, B. P. (1993) Equilibrium size of atmospheric aerosol sulfates as a function of particle
8 acidity and ambient relative humidity. J. Geophys. Res. [Atmos.] 98: 7141-7147.
9
10 Koutrakis, P.; Spengler, J. D. (1987) Source apportionment of ambient particles in Steubenville, OH using
11 specific rotation factor analysis. Atmos. Environ. 21: 1511-1519.
12
13 Koutrakis, P.; Wolfson, J. M.; Spengler, J. D.; Stern, B.; Franklin, C. A. (1989) Equilibrium size of
14 atmospheric aerosol sulfates as a function of the relative humidity. J. Geophys. Res. [Atmos.]
15 94: 6442-6448.
16
17 Lamborg, C.; Keeler, G. J.; Evans, G. (1992) Atmospheric acidity measurements during the Lake Michigan
18 Urban Air Toxics Study. Research Triangle Park, NC: U.S. Environmental Protection Agency,
19 Atmospheric Research and Exposure Assessment Laboratory; EPA report no. EPA/600/A-92/250.
20 Available from: NTIS, Springfield, VA; PB93-121069.
21
22 Larson, S. M.; Cass, G. R. (1989) Characteristics of summer midday low-visibility events in the Los Angeles
23 area. Environ. Sci. Technol. 23: 281-289.
24
25 Larson, S. M.; Cass, G. R.; Gray, H. A. (1989) Atmospheric carbon particles and the Los Angeles visibility
26 problem. Aerosol Sci. Technol. 10: 118-130.
27
28 Leaderer, B. P.; Bernstein, D. M.; Daisey, J. M.; Kleinman, M. T.; Kneip, T. J.; Knutson, E. O.; Lippmann,
29 M.; Lioy, P. J.; Rahn, K. A.; Sinclair, D.; Tanner, R. L.; Wolff, G. T. (1978) Summary of the New
30 York Summer Aerosol Study (NYSAS). J. Air Pollut. Control Assoc. 28: 321-327.
31
32 Leaderer, B. P.; Tanner, R. L.; Holford, T. R. (1982) Diurnal variations, chemical composition and relation to
33 meteorological variables of the summer aerosol in the New York subregion. Atmos. Environ.
34 16: 2075-2087.
35
36 Lee, H. S.; Wadden, R. A.; Scheff, P. A. (1993) Measurement and evaluation of acid air pollutants in Chicago
37 using an annular denuder system. Atmos. Environ. Part A 27: 543-553.
38
39 Lewis, C. W.; Dzubay, T. G. (1986) Measurement of light absorption extinction in Denver. Aerosol Sci.
40 Technol. 5: 325-336.
41
42 Lewis, C. W.; Macias, E. S. (1980) Composition of size-fractionated aerosol in Charleston, West Virginia.
43 Atmos. Environ. 14: 185-194.
44
45 Lewis, C. W.; Baumgardner, R. E.; Stevens, R. K.; Russwurm, G. M. (1986) Receptor modeling study of
46 Denver winter haze. Environ. Sci. Technol. 20: 1126-1136.
47
48 Lewis, C. W. et al. (1989) Comparison of three measures of visibility extinction in Denver, CO. Presented at:
49 the 82nd annual meeting of the Air Pollution Control Association; June; Anaheim, CA. Pittsburgh, PA:
50 Air Pollution Control Association; paper no. 89-151.4.
51
52 Liang, C. S. K.; Waldman, J. M. (1992) Indoor exposures to acidic aerosols at child and elderly care facilities.
53 Indoor Air 2: 196-207.
54
April 1995 6-199 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lioy, P. J.; Waldman, J. M. (1989) Acidic sulfate aerosols: characterization and exposure. In: Symposium on
2 the health effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health Perspect.
3 79: 15-34.
4
5 Lioy, P. J.; Samson, P. J.; Tanner, R. L.; Leaderer, B. P.; Minnich, T.; Lyons, W. (1980) The distribution and
6 transport of sulfate "species" in the New York metropolitan area during the 1977 Summer Aerosol Study.
7 Atmos. Environ. 14: 1391-1407.
8
9 Lioy, P. J.; Daisey, J. M.; Greenberg, A.; Harkov, R. (1985) A major wintertime (1983) pollution episode in
10 northern New Jersey: analyses of the accumulation and spatial distribution of inhalable paniculate matter,
11 extractable organic matter and other species. Atmos. Environ. 19: 429-436.
12
13 Lipfert, F. W.; Wyzga, R. E. (1993) On the spatial and temporal variability of aerosol acidity and sulfate
14 concentration. Air Waste 43: 489-491.
15
16 Lundgren, D. A. (1967) An aerosol sampler for determination of particle concentration as a function of size and
17 time. J. Air Pollut. Control Assoc. 17: 225-229.
18
19 Macias, E. S.; Zwicker, J. O.; Ouimette, J. R.; Hering, S. V.; Friedlander, S. K.; Cahill, T. A.; Kuhlmey, G.
20 A.; Richards, L. W. (1981) Regional haze case studies in the southwestern U.S.—I. aerosol chemical
21 composition. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes and visibility:
22 measurements and model components, proceedings of the symposium; November 1980; Grand Canyon
23 National Park, AZ. Atmos. Environ. 15: 1971-1986.
24
25 Macias, E. S.; Zwicker, J. O.; White, W. H. (1981) Regional haze case studies in the southwestern U.S.—II.
26 source contributions. In: White, W. H.; Moore, D. J.; Lodge, J. P., Jr., eds. Plumes and visibility:
27 measurements and model components: proceedings of the symposium; November 1980; Grand Canyon
28 National Park, AZ. Atmos. Environ. 15: 1987-1997.
29
30 Macias, E. S.; Vossler, T. L.; White, W. H. (1987) Carbon and sulfate particles in the western U.S. In:
31 Bhardwaja, P. J., ed. Visibility protection: research and policy aspects. Pittsburgh, PA: Air Pollution
32 Control Association.
33
34 Maenhaut, W.; Royset, 0.; Vadset, M.; Kauppinen, E. I.; Lind, T. M. (1993) Nucl. Instrum. Methods
35 B75: 266-272.
36
37 Malissa, H. (1979) Some analytical approaches to the chemical characterization of carbonaceous particulates. In:
38 Novakov, T., ed. Proceedings: carbonaceous particles in the atmosphere; March 1978; Berkeley, CA.
39 Berkeley, CA: University of California, Lawrence Berkeley Laboratory; pp. 3-9; report no. LBL-9037.
40
41 Malm, W. C. et al. (1987) Comparison of atmospheric extinction measurements made by a transmissometer,
42 integrating nephelometer, and teleradiometer with natural and artificial black targets. In: Bhardwaja, P.
43 J., ed. Visibility protection: research and policy aspects. Pittsburgh, PA: Air Pollution Control
44 Association.
45
46 Malm, W. C. et al. (1987) An eigenvector analysis of paniculate data in the western United States. In:
47 Bhardwaja, P. J., ed. Visibility protection: research and policy aspects. Pittsburgh, PA.: Air Pollution
48 Control Association.
49
50 Malm, W.; Gebhard, K.; Cahill, T.; Eldred, R.; Pielke, R.; Watson, J.; Latimer, D. (1989) Winter haze
51 intensive experiment: draft final report. Fort Collins, CO: National Park Service.
52
53 Malm, W. C.; Gebhart, K. A.; Henry, R. C. (1990) An investigation of the dominant source regions of fine
54 sulfur in the western United States and their areas of influence. Atmos. Environ. Part A 24: 3047-3060.
April 1995 6-200 DRAFT-DO NOT QUOTE OR CITE
-------
1 Malm, W. C.; Gebhart, K. A.; Molenar, J.; Cahill, T.; Eldred, R.; Huffman, D. (1994) Examining the
2 relationship between atmospheric aerosols and light extinction at Mount Rainier and North Cascades
3 National Parks. Atmos. Environ. 28: 347-360.
4
5 Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in
6 particles concentration and optical extinction in the United States. J. Geophys. Res. 99: 1347-1370.
7
8 Mamane, Y.; Noll, K. E. (1985) Characterization of large particles at a rural site in the eastern United States:
9 mass distribution and individual particle analysis. Atmos. Environ. 19: 611-622.
10
11 ManTech. (1994) LRGVES report. Research Triangle Park, NC: ManTech.
12
13 Mangelson, N. F.; Rees, L. B.; Lewis, L.; Joseph, J. M.; Cui, W.; Machir, J.; Eatough, D. J.; Wilkerson, T.
14 D.; Jensen, D. T. (1994) The formation of sulfate and nitrate during winter inversion fogs in Cache
15 Valley, Utah. In: Proceedings of an international specialty conference on aerosol and atmospheric optics:
16 radiative balance and visual air quality: v. A. Pittsburgh, PA: Air and Waste Management Association.
17
18 Marple, V.; Rubow, K.; Anath, G.; Fissan, H. J. (1986) J. Aerosol Sci. 17: 489-494.
19
20 Marshall, B. T.; Patterson, E. M.; Grams, G. W. (1986) Characterization of the Atlanta area aerosol elemental
21 composition and possible sources. Atmos. Environ. 20: 1291-1300.
22
23 McDade, C. E.; Tombach, I. H. (1987) Goals and initial findings from SCENES. In: Bhardwaja, P. J., ed.
24 Visibility protection: research and policy aspects. Pittsburgh, PA: Air Pollution Control Association.
25
26 McLaughlin, S. B.; Schorn, V. J.; Jones, H. C. (1976) A programmable exposure system for kinetic dose-
27 response studies with air pollutants. J. Air Pollut. Control Assoc. 26: 132-135.
28
29 McMurry, P. (1989) Organic and elemental carbon size distributions of Los Angeles aerosols measured during
30 SCAQS [final report]. Atlanta, GA: Coordinating Research Council; CRC project no. SCAQS-6-1;
31 Particle Technology Laboratory report no. 713.
32
33 McMurry, P. H.; Stolzenburg, M. R. (1989) On the sensitivity of particle size to relative humidity for Los
34 Angeles aerosols. Atmos. Environ. 23: 497-507.
35
36 McMurry, P. H.; Zhang, X. Q. (1989) Size distributions of ambient organic and elemental carbon. Aerosol Sci.
37 Technol. 10: 430-437.
38
39 Mercer, T. T. (1964) The stage constants of cascade impactors. Ann. Occup. Hyg. 7: 115-124.
40
41 Milford, J. B.; Davidson, C. I. (1985) The sizes of particulate trace elements in the atmosphere—a review. J.
42 Air Pollut. Control Assoc. 35: 1249-1260.
43
44 Miller, D. F.; Schorran, D. E.; Hoffer, T. E.; Rogers, D. P.; White, W. H.; Macias, E. S. (1990) An analysis
45 of regional haze using tracers of opportunity. J. Air Waste Manage. Assoc. 40: 757-761.
46
47 Molenar, J. V.; Dietrich, D. L.; Cismoski, D. S.; Cahill, T. A.; Wakabayashi, P. H. (1994) In: Proceedings of
48 an international specialty conference on aerosol and atmospheric optics: radiative balance and visual air
49 quality, volume A. Pittsburgh, PA: Air & Waste Management Association.
50
51 Morandi, M. T.; Kneip, T. J.; Cobourn, W. G.; Husar, R. B.; Lioy, P. J. (1983) The measurement of H2SO4
52 and other sulfate species at Tuxedo, New York with a thermal analysis flame photometric detector and
53 simultaneously collected quartz filter samples. Atmos. Environ. 17: 843-848.
54
April 1995 6-201 DRAFT-DO NOT QUOTE OR CITE
-------
1 Moskowitz, A. H. (1977) Particle size distribution of nitrate aerosols in the Los Angeles Air Basin. Research
2 Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Sciences Research
3 Laboratory; EPA report no. EPA-600/3-77-053. Available from: NTIS, Springfield, VA; PB-269 349.
4
5 Mosteller, P.; Tukey, J. W. (1977) Data analysis and regression. Reading, PA: Addison-Wesely Publishing
6 Company.
7
8 Moyers, J. L. (1981) 1981-1982 aerosol studies for nine southwest national parks and monuments [final report].
9 Washington, DC: U.S. Department of the Interior, National Park Service.
10
11 Mueller, P. K.; Hansen, D. A.; Watson, J. G., Jr. (1986) The Subregional Cooperative Electric Utility,
12 Department of Defense, National Park Service, and EPA Study (SCENES) on visibility: an overview.
13 Palo Alto, CA: Electric Power Research Institute; report no. EA-4664-SR.
14
15 Mukerjee et al. (1993) A methodology to apportion ambient air measurements to investigate potential effects on
16 air quality near waste incinerators. In: Proceedings of the 1993 incineration conference; Knoxville, TN.
17 p. 527.
18
19 Mylonas, D. T.; Allen, D. T.; Ehrman, S. H.; Pratsinis, S. E. (1991) The sources and size distributions of
20 organonitrates in Los Angeles aerosol. Atmos. Environ. Part A 25: 2855-2861.
21
22 National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
23 National Academy of Science.
24
25 Natusch, D. F. S.; Wallace, J. R. (1974) Urban aerosol toxicity: the influence of particle size. Particle size,
26 adsorption, and respiratory deposition profiles combine to determine aerosol toxicity. Science
27 (Washington, DC) 186: 695-699.
28
29 Natusch, D. F. S.; Wallace, J. R.; Evans, C. A., Jr. (1974) Toxic trace elements: preferential concentration in
30 respirable particles. Science (Washington, DC) 183: 202-204.
31
32 Newman, J. E.; Abel. M. D.; Harrison, P. R.; Yost, K. J. (1976) Wind as related to critical flushing speed
33 versus reflotation speed by high-volume sampler paniculate loading. In: Atmospheric-surface exchange of
34 particles and gaseous pollutants. Washington, DC: U.S. Department of Energy; pp. 466-496. Available
35 from: NTIS, Springfield, VA; CONF 740-921.
36
37 Nifong, G. D. (1970) [Ph.D. thesis]. Ann Arbor, MI: Univeristy of Michigan.
38
39 Noll, K. E.; Draftz, R.; Fang, K. Y. P. (1987) The composition of atmospheric coarse particles at an urban and
40 non-urban site. Atmos. Environ. 21: 2717-2721.
41
42 Noll, K. E.; Yuen, P.-F.; Fang, K. Y.-P. (1990) Atmospheric coarse paniculate concentrations and dry
43 deposition fluxes for ten metals in two urban environments. Atmos. Environ. Part A 24: 903-908.
44
45 Ouimette, J. R.; Flagan, R. C.; Kelso, A. R. (1981) Chemical species contributions to light scattering by
46 aerosols at a remote arid site: comparison of statistical and theoretical results. In: Macias, E. S.; Hopke,
47 P. K., eds. Atmospheric aerosols: source/air quality relationships; based on a symposium jointly
48 sponsored by the Divisions of Nuclear Chemistry and Technology and Environmental Chemistry at the
49 180th national meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington,
50 DC: American Chemical Society; pp. 125-156. (Comstock, M. J., ed. ACS symposium series: 167).
51
52 Patterson, R. K.; Wagman, J. (1977) Mass and composition of an urban aerosol as a function of particle size for
53 several visibility levels. J. Aerosol Sci. 8: 269-279.
54
April 1995 6-202 DRAFT-DO NOT QUOTE OR CITE
-------
1 Patterson, D. E.; Holloway, J. ML; Husar, R. B. (1980) Historical visibility over the eastern U.S.: daily and
2 quarterly extinction coefficient contour maps. Research Triangle Park, NC: U.S. Environmental
3 Protection Agency, Environmental Sciences Research Laboratory; EPA report no. EPA-600/3-80-043a.
4 Available from: NTIS, Springfield, VA; PB81-196974.
5
6 Patterson, D. E.; Husar, R. B.; Wilson, W. E.; Smith, L. F. (1981) Monte Carlo simulation of daily regional
7 sulfur distribution: comparison with SURE sulfate data and visual range observations during August
8 1977. J. Appl. Meteorol. 20: 404-420.
9
10 Patterson, R. E.; Bernier, S. A.; Gregory, J. (1994) Blowing dust across the Texas high plains: spatial and
11 temporal variations. In: Proceedings of the international specialty conference on aerosol and atmospheric
12 optics: radiative balance and visual air quality, volume A. Pittsburgh, PA; Air & Waste Management
13 Association.
14
15 Pickle, T.; Allen, D. T.; Pratsinis, S. E. (1990) The sources and size distributions of aliphatic and carbonyl
16 carbon in Los Angeles aerosol. Atmos. Environ. Part A 24: 2221-2228.
17
18 Pierson, W. R.; Brachaczek, W. W. (1988) Coarse- and fine-particle atmospheric nitrate and HNO3(g) in
19 Claremont, California, during the 1985 Nitrogen Species Methods Comparison Study. Atmos. Environ.
20 22: 1665-1668.
21
22 Pierson, W. R.; Russell, P. A. (1979) Aerosol carbon in the Denver area in November 1973. Atmos. Environ.
23 13: 1623-1628.
24
25 Pierson, W. R.; Brachaczek, W. W.; Korniski, T. J.; Truex, T. J.; Butler, J. W. (1980) Artifact formation of
26 sulfate, nitrate, and hydrogen ion on backup filters: Allegheny Mountain experiment. J. Air Pollut.
27 Control Assoc. 30: 30-34.
28
29 Pinto, J. (n.d.) [Unpublished data].
30
31 Poirot, R. L.; Flocchini, R. G.; Husar, R. B. (1990) Winter fine particle composition in the northeast:
32 preliminary results from the NESCAUM monitoring network. Presented at: 83rd annual meeting of the
33 Air & Waste Management Association; June; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management
34 Association; paper no. 90-84.5.
35
36 Pope, C. A., Ill; Schwartz, J.; Ransom, M. R. (1992) Daily mortality and PM10 pollution in Utah valley. Arch.
37 Environ. Health 47: 211-217.
38
39 Raabe, O. G.; Braaten, D. A.; Axelbaum, R. L.; Teague, S. V.; Cahill, T. A. (1988) J. Aerosol Sci.
40 192: 183-195.
41
42 Rahn, K. A.; Harrison, P. R. (1976) The chemical composition of Chicago street dust. In: Atmosphere-surface
43 exchange of paniculate and gaseous pollutants (1974): proceedings of a symposium; September 1974;
44 Richland, WA.. Oak Ridge, TN: Energy Research and Development Administration; pp. 557-570.
45 (ERDA symposium series 38). Available from: NTIS, Springfield, VA; CONF 740921.
46
47 Rahn, K. A.; McCaffrey, R. J. (1980) On the origin and transport of the winter Arctic aerosol. Ann. N. Y.
48 Acad. Sci. 388: 503.
49
50 Rahn, K. A.; Wesolowski, J. J.; John, W.; Ralston, H. R. (1971) Diurnal variation of aerosol trace element
51 concentrations in Livermore, California. J. Air Pollut. Control Assoc. 21: 406-409
52
53 Richards, L. W. (1983) Comments on the oxidation of NO2 to nitrate—day and night. Atmos. Environ
54 17: 397-402.
April 1995 6_203 DRAFT-DO NOT QUOTE OR CITE
-------
1 Richards, L. W.; Markowski, G. R.; Waters, N. (1981) Comparison of nephelometer, telephotometer, and
2 aerosol data in the southwest. Presented at: the 74th annual meeting of the Air Pollution Control
3 Association; June; Philadelphia, PA. Pittsburgh, PA: Air Pollution Control Association; paper
4 no. 81-54.5.
5
6 Rodes, C. E.; Evans, E. G. (1985) Preliminary assessment of 10 /urn paniculate sampling at eight locations in the
7 United States. Atmos. Environ. 19: 293-303.
8
9 Rolph, G. D.; Draxler, R. R.; de Pena, R. G. (1992) Modeling sulfur concentrations and depositions in the
10 United States during ANATEX. Atmos. Environ. Part A 26: 73-93.
11
12 Russell, A. G.; Cass, G. R. (1984) Acquisition of regional air quality model validation data for nitrate, sulfate,
13 ammonium ion and their precursors. Atmos. Environ. 18: 1815-1827.
14
15 Saldiva, P. H. N.; Lichtenfels, A. J. F. C.; Paiva, P. S. O.; Barone, I. A.; Martins, M. A.; Massad, E.;
16 Pereira, J. C. R.; Xavier, V. P.; Singer, J. M.; Bohm, G. M. (1994) Association between air pollution
17 and mortality due to respiratory diseases in children in Sao Paulo, Brazil: a preliminary report. Environ.
18 Res. 65: 218-225.
19
20 Savoie, D. L.; Prospero, J. M. (1982) Particle size distribution of nitrate and sulfate in the marine atmosphere.
21 Geophys. Res. Lett. 9: 1207-1210.
22
23 Saxena, P. () [Personal communication]. Palo Alto, CA: Electric Power Research Institute.
24
25 Schichtel, B. A.; Husar, R. B. (1991) Apportionment of light extinction by aerosol types. Air Force Systems
26 Command, Phillips Laboratory.
27
28 Schroeder, W. H.; Dobson, M.; Kane, D. M.; Johnson, N. D. (1987) Toxic trace elements associated with
29 airborne paniculate matter: a review. JAPCA 37: 1267-1285.
30
31 Schwartz, J. (1994) Air pollution and hospital admissions for the elderly in Birmingham, Alabama. Am. J.
32 Epidemiol. 139: 589-598.
33
34 Schwartz, J.; Dockery, D. W. (1992) Increased mortality in Philadelphia associated with daily air pollution
35 concentrations. Am. Rev. Respir. Dis. 145: 600-604.
36
37 Seaton, A.; MacNee, W.; Donaldson, K.; Godden, D. (1995) Paniculate air pollution and acute health effects.
38 Lancet (8943): 176-178.
39
40 Seeker, W. R. (1990) In: Waste combustion: proceedings of the 23rd international symposium on combustion;
41 pp. 867-885.
42
43 Senior, C. L.; Flagan, R. C. (1982) Ash vaporization and condensation during combustion of a suspended coal
44 particle. Aerosol Sci. Technol. 1: 371-383.
45
46 Shah, J. J.; Watson, J. G., Jr.; Cooper, J. A.; Huntzicker, J. J. (1984) Aerosol chemical composition and light
47 scattering in Portland, Oregon: the role of carbon. Atmos. Environ. 18: 235-240.
48
49 Shaw, R. W., Jr.; Paur, R. J. (1983) Composition of aerosol particles collected at rural sites in the Ohio River
50 Valley. Atmos. Environ. 17: 2031-2044.
51
52 Simoneit, B. R. T.; Mazurek, M. A. (1982) Organic matter of the troposphere—II. natural background of
53 biogenic lipid matter in aerosols over the rural western United States. Atmos. Environ. 16: 2139-2159.
54
April 1995 6-204 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J.; Shields, H. C. (1990) Measurement and modeling of
2 airborne concentrations and indoor surface accumulation rates of ionic substances at Neenah, Wisconsin.
3 Atmos. Environ. Part A 24: 627-638.
4
5 Sirois, A.; Fricke, W. (1992) Regionally representative daily air concentrations of acid-related substances in
6 Canada; 1983-1987. Atmos. Environ. Part A 26: 593-607.
7
8 Sisler, J. F.; Malm, W. C. (1994) The relative importance of soluble aerosols to spatial and seasonal trends of
9 impaired visibility in the United States. Atmos. Environ. 28: 851-862.
10
11 Sisler, J. F.; Huffman, D.; Lattimer, D. A.; Malm, W. C.; Pitchford, M. (1993) Spatial and temporal patterns
12 and the chemical composition of the haze in the United States: an anlysis of data from the IMPROVE
13 network, 1988-1991. Fort Collins, CO: Colorado State University, CIRA.
14
15 Skidmore, L. W.; Chow, J. C.; Tucker, T. T. (1992) PM10 air quality assessment for the Jefferson County,
16 Ohio air quality control region. In: Chow, J. C.; Ono, D. M., eds. Transactions: PM10 standards and
17 nontraditional paniculate source controls. Pittsburgh, PA: Air & Waste Management Association;
18 pp. 1016-1031.
19
20 Sloane, C. S. (1982) Visibility trends—II. mideastern United States 1948-1978. Atmos. Environ. 16: 2309-2321.
21
22 Sloane, C. S. (1982) Visibility trends—I. methods of analysis. Atmos. Environ. 16: 41-51.
23
24 Sloane, C. S. (1986) Effect of composition on aerosol light scattering efficiencies. Atmos. Environ.
25 20: 1025-1037.
26
27 Sloane, C. S.; Wolff, G. T. (1985) Prediction of ambient light scattering using a physical model responsive to
28 relative humidity: validation with measurements from Detroit. Atmos. Environ. 19: 669-680.
29
30 Sloane, C. S.; Watson, J.; Chow, J.; Pritchett, L.; Richards, L. W. (1991) Size-segregated fine particle
31 measurements by chemical species and their impact on visibility impairment in Denver. Atmos. Environ.
32 Part A 25: 1013-1024.
33
34 Small, M.; Germani, M. S.; Small, A. M.; Zoller, W. H.; Moyers, J. L. (1981) Airborne plume study of
35 emissions from the processing of copper ores in southeastern Arizona. Environ. Sci. Technol.
36 15: 293-299.
37
38 Solomon, P. A.; Moyers, J. L. (1986) A chemical characterization of wintertime haze in Phoenix, Arizona.
39 Atmos. Environ. 20: 207-213.
40
41 Solomon, P. A.; Larson, S. M.; Fall, T.; Cass, G. R. (1988) Basinwide nitric acid and related species
42 concentrations observed during the Claremont Nitrogen Species Comparison Study. Atmos. Environ.
43 22: 1587-1594.
44
45 Spengler, J. D.; Thurston, G. D. (1983) Mass and elemental composition of fine and coarse particles in six U.S.
46 cities. J. Air Pollut. Control Assoc. 33: 1162-1171.
47
48 Spengler, J. D.; Allen, G. A.; Foster, S.; Severance, P.; Ferris, B., Jr. (1986) Sulfuric acid and sulfate aerosol
49 events in two U.S. cities. In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols:
50 research, risk assessment and control strategies—proceedings of the second U.S.-Dutch international
51 symposium; May 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 107-120.
52
April 1995 6-205 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spengler, J. D.; Keeler, G. J.; Koutrakis, P.; Ryan, P. B.; Raizenne, M.; Franklin, C. A. (1989) Exposures to
2 acidic aerosols. In: Symposium on the health effects of acid aerosols; October 1987; Research Triangle
3 Park, NC. Environ. Health Perspect. 79: 43-51.
4
5 Spengler, J. D.; Brauer, M.; Koutrakis, P. (1990) Acid air and health. Environ. Sci. Technol. 24: 946-956.
6
7 Spicer, C. W. (1977) The fate of nitrogen oxides in the atmosphere. In: Pitts, J. N., Jr.; Metcalf, R. L.; Lloyd,
8 A. C., eds. Advances in environmental science and technology: v. 4. New York, NY: John Wiley and
9 Sons, Inc.; pp. 163-261.
10
11 Spicer, C. W.; Schumacher, P. M. (1979) Paniculate nitrate: laboratory and field studies of major sampling
12 interferences. Atmos. Environ. 13: 543-552.
13
14 Spix, C.; Heinrich, J.; Dockery, D.; Schwartz, J.; Volksch, G.; Schwinkowski, K.; Collen, C.; Wichmann, H.
15 E. (1993) Air pollution and daily mortality in Erfurt, East Germany, 1980-1989. Environ. Health
16 Perspect. 101: 518-526.
17
18 Stevens, R. K., ed. (1979) Current methods to measure atmospheric nitric acid and nitrate artifacts: proceedings
19 of a workshop on measurement of atmospheric nitrates; October 1978; Southern Pines, NC. Research
20 Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Sciences Research
21 Laboratory; EPA report no. EPA-600/2-79-051. Available from: NTIS, Springfield, VA; PB-294 098.
22
23 Stevens. (1985) Sampling and analysis methods for use in source apportionment studies to determine impact of
24 wood burning on fine particle mass. Environ. Int. 11: 271.
25
26 Stevens et al. (n.d.) A comparison of air quality measurements in Roanoke, VA, and other integrated air cancer
27 project monitoring locations.
28
29 Stevens, R. K.; Dzubay, T. G.; Shaw, R. W., Jr.; McClenny, W. A.; Lewis, C. W.; Wilson, W. E. (1980)
30 Characterization of the aerosol in the Great Smoky Mountains. Environ. Sci. Technol. 14: 1491-1498.
31
32 Stevens, R. K.; Dzubay, T. G.; Lewis, C. W.; Shaw, R. W., Jr. (1984) Source apportionment methods applied
33 to the determination of the origin of ambient aerosols that affect visibility in forested areas. Atmos.
34 Environ. 18: 261-272.
35
36 Struempler, A. W. (1975) Trace element composition in atmospheric particulates during 1973 and the summer of
37 1974 at Chadron, Neb. Environ. Sci. Technol. 9: 1164-1168.
38
39 Suh, H. H.; Spengler, J. D.; Koutrakis, P. (1992) Personal exposures to acid aerosols and ammonia. Environ.
40 Sci. Technol. 26: 2507-2517.
41
42 Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993) Validation of personal exposure models for sulfate and aerosol
43 strong acidity. J. Air Waste Manage. Assoc. 43: 845-850.
44
45 Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1994a) Spatial variation in acidic sulfate and ammonia
46 concentrations within metropolitan Philadelphia. J. Air Waste Manage. Assoc.: submitted.
47
48 Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1994b) The relationship between airborne acidity and ammonia in
49 indoor environments. J. Exposure Anal. Environ. Epidemiol. 4: 1-23.
50
51 Sunyer, J.; Anto, J. M.; Murillo, C.; Saez, M. (1991) Effects of urban air pollution on emergency room
52 admissions for chronic obstructive pulmonary disease. Am. J. Epidemiol. 134: 277-286.
53
April 1995 6-206 DRAFT-DO NOT QUOTE OR CITE
-------
1 Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM10 samplers
2 and source apportionment of ambient PM10 concentrations at Rillito, Arizona. In: Chow, J. C.; Ono, D.
3 M., eds. PM]0 standards and non-traditional paniculate source controls: proceedings of an AWMA/EPA
4 international specialty conference. Pittsburgh, PA: Air & Waste Management Association; pp. 244-261.
5
6 Thompson, K. M.; Koutrakis, P.; Brauer, M.; Spengler, J. D.; Wilson, W. E.; Burton, R. M. (1991)
7 Measurements of aerosol acidity: sampling frquency, seasonal variability, and spatial variation. Presented
8 at: the 84th annual meeting of the Air & Waste Management Association; June; Vancouver, BC, Canada.
9 Pittsburgh, PA: Air & Waste Management Association; paper no. 91-89.5.
10
11 Thurston, G. D.; Gorczynski, J. E.; Jaques, P.; Currie, J.; He, D. (1992) An automated sequential sampling
12 system for paniculate aerosols: description, characterization, and field sampling results. J. Exposure
13 Anal. Environ. Epidemiol. 2: 415-428.
14
15 Tombach, I. H. et al. (1987) SCENES light extinction methods study, September -October 1985. In: Bhardwaja,
16 P. J., eds. Visibility protection: research and policy aspects. Pittsburgh, PA: Air Pollution Control
17 Association.
18
19 Trijonis, J. (1979) Visibility in the southwest—an exploration of the historical data base. Atmos. Environ.
20 13: 833-843.
21
22 Trijonis, J.; McGown, M.; Pitchford, M.; Blumenthal, D.; Roberts, P.; White, W.; Macias, E.; Weiss, R.;
23 Waggoner, A.; Watson, J.; Chow, J.; Flocchini, R. (1988) Visibility conditions and causes of visibility
24 degradation in the Mojave Desert of California: executive summary, RESOLVE project final report.
25 China Lake, CA: Department of the Navy, Naval Weapons Center; document no. NWC TP 6869.
26 Available from: NTIS, Springfield, VA; AD-A206 322.
27
28 Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a
29 descriptive analysis of organic and elemental carbon concentrations. Atmos. Environ. Part A
30 25: 207-215.
31
32 Turpin, B. J.; Gary, R. A.; Huntzicker, I. J. (1990) An in situ, time-resolved analyzer for aerosol organic and
33 elemental carbon. Aerosol Sci. Technol. 12: 161-171.
34
35 Turpin, B. J.; Huntzicker, J. J.; Larson, S. M.; Cass, G. R. (1991) Los Angeles summer midday particulate
36 carbon: primary and secondary aerosol. Environ. Sci. Technol. 25: 1788-1793.
37
38 U.S. Environmental Protection Agency. (1982) Air quality criteria for particulate matter and sulfur oxides.
39 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
40 and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
41 VA; PB84-156777.
42
43 U.S. Environmental Protection Agency. (1989) An acid aerosols issue paper: health effects and aerometrics.
44 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
45 and Assessment Office; EPA report no. EPA-600/8-88-005F. Available from: NTIS, Springfield, VA;
46 PB91-125864.
47
48 U.S. Environmental Protection Agency. (1989) Washington, DC: Office of Solid Waste and Emergency
49 Response.
50
51 U.S. Environmental Protection Agency. (1991) MOHAVE.
52
53 U.S. Environmental Protection Agency. (1995) Visibility research related to source assessment {draft report].
54
April 1995 6-207 DRAFT-DO NOT QUOTE OR CITE
-------
1 Van Grieken, R. G.; Johansson, J.; Winchester, J.; Odom, A. L. (1975) Anal. Chem. 275: 343.
2
3 Vasconcelos, L. A.; Macias, E. S.; White, W. H. (1993) Aerosol composition as a function of haze and
4 humidity levels in the southwestern U.S. [personal communication].
5
6 Vermette et al. (1992) PMJO source apportionment using local surface dust profiles: examples from Chicago. In:
7 Chow; Ono, eds. PM10 standards and nontraditional particulate source controls. Pittsburgh, PA: Air &
8 Waste Management Association; p. 262.
9
10 Vossler, T. L.; Lewis, C. W.; Stevens, R. K.; Dzubay, T. G.; Gordon, G. E.; Tuncel, S. G.; Russwurm, G.
11 M.; Keeler, G, J. (1989) Composition and origin of summertime air pollutants at Deep Creek Lake,
12 Maryland. Atmos. Environ. 23: 1535-1547.
13
14 Waldman, J. M.; Lioy, P. J.; Thurston, G. D.; Lippmann, M. (1990) Spatial and temporal patterns in
15 summertime sulfate aerosol acidity and neutralization within a metropolitan area. Atmos. Environ. Part B
16 24: 115-126.
17
18 Waldman, J. M.; Liang, S.-K. C.; Lioy, P. J.; Thurston, G. D.; Lippmann, M. (1991) Measurements of sulfate
19 aerosol and its acidity in the SO2 source region of Chestnut Ridge, PA. Atmos. Environ. Part A
20 25: 1327-1333.
21
22 Waldman, J. M.; Koutrakis, P.; Allen, G. A.; Thurston, G. D.; Burton, R. M.; Wilson, W. E. (1995) Human
23 exposures to particle strong acidity, submitted.
24
25 Wall, S. M.; John, W.; Ondo, J. L. (1988) Measurement of aerosol size distributions for nitrate and major ionic
26 species. Atmos. Environ. 22: 1649-1656.
27
28 Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
29 Technol. 8: 157-172.
30
31 Watson, J. G.; Cooper, J. A.; Huntzicker, J. J. (1984) The effective variance weighting for least squares
32 calculations applied to the mass balance receptor model. Atmos. Environ. 18: 1347-1355.
33
34 Watson, J. G.; Chow, J. C.; Richards, L. W.; Neff, W. D.; Andersen, S. R.; Dietrich, D. L.; Houck, J. E.;
35 Olnez, I. (1988) The 1987-88 metro Denver brown cloud study: v. I, II, III. Reno, NV: Desert Research
36 Institute; final report no. 8810.1F(l-3).
37
38 Watson, J. G.; Chow, J. C.; Mathai, C. V. (1989) Receptor models in air resources management: a summary of
39 the APCA international specialty conference. JAPCA 39: 419-426.
40
41 Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994)
42 Chemical mass balance source apportionment of PM10 during the Southern California Air Quality Study.
43 Aerosol Sci. Technol. 21: 1-36.
44
45 Webber, J. S.; Dutkiewicz, V. A.; Husain, L. (1985) Identification of submicrometer coal fly ash in a high-
46 sulfate episode at Whiteface Mountain, New York. Atmos. Environ. 19: 285-292.
47
48 Weiss, R. E.; Larson, T. V.; Waggoner, A. P. (1982) In situ rapid-response measurement of HjSO^NH^jSC^
49 aerosols in rural Virginia. Environ. Sci. Technol. 16: 525-532.
50
51 Wesolowski, J. J.; John, W.; Devor, W.; Cahill, T. A.; Feeney, P. J.; Wolfe, G.; Flocchini, R. (1978) In:
52 Dzubay, T. G., ed. X-ray fluorescence analysis of environmental samples. Ann Arbor, MI: Ann Arbor
53 Science Publishers; pp. 121-130.
54
April 1995 6-208 DRAFT-DO NOT QUOTE OR CITE
-------
1 Whitby, K. T. (1977) Physical characterization of aerosols, methods and standards for environmental
2 measurement. In: Kirchoff, W. H., ed. Washington, DC: National Bureau of Standards; pp. 165-173;
3 NBS special publication no. 464.
4
5 Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
6
7 Whitby, K. T.; Sverdrup, G. M. (1980) California aerosols: their physical and chemical characteristics. In: Hidy,
8 G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The character and origins
9 of smog aerosols: a digest of results from the California Aerosol Characterization Experiment (ACHEX).
10 New York, NY: John Wiley & Sons, Inc.; pp. 477-517. (Advances in environmental science and
11 technology: v. 9).
12
13 Whitby, K. T.; Husar, R. B.; Liu, B. Y. H. (1972) The aerosol size distribution of Los Angeles smog. J.
14 Colloid Interface Sci. 39: 177-204.
15
16 White, W. H. (1986) On the theoretical and empirical basis for apportioning extinction by aerosols: a critical
17 review. Atmos. Environ. 20: 1659-1672.
18
19 White, W. H. (1990) The components of atmospheric light extinction: a survey of ground-level budgets. Atmos.
20 Environ. Part A 24: 2673-2679.
21
22 White, W. H.; Husar, R. B. (1976) A Lagrangian model of the Los Angeles smog aerosol.
23
24 White, W. H.; Macias, E. S. (1987) Paniculate nitrate measurements in rural areas of the western United States.
25 Atmos. Environ. 21: 2563-2571.
26
27 White, W. H.; Macias, E. S. (1987) On measurements error and the empirical relationship of atmospheric
28 extinction to aerosol composition in the non-urban west. In: Bhardwaja, P. J., ed. Visibility protection:
29 research and policy aspects. Pittsburgh, PA: Air Pollution Control Association.
30
31 White, W. H.; Macias, E. S. (1989) Carbonaceous particles and regional haze in the western United States.
32 Aerosol Sci. Technol. 10: 111-117.
33
34 White, W. H.; Macias, E. S.; Miller, D. P.; Schorran, D. E.; Hoffer, T. E.; Rodgers, D. P. (1990) Regional
35 transport of the urban work week: methylchloroform cycles in the Nevada-Arizona desert. Geophys. Res.
36 Lett. 17: 1081-1084.
37
38 White, W. H.; Macias, E. S.; Nininger, R. C.; Schorran, D. (1994) Size-resolved measurements of light
39 scattering by ambient particles in the southwestern U.S.A. Atmos. Environ. 28: 909-921.
40
41 Willeke, K.; Whitby, K. T. (1975) Atmosperic aerosols: size distribution interpretation. J. Air Pollut. Control
42 Assoc. 25: 529-534.
43
44 Wilson, W. E.; Koutrakis, P.; Spengler, J. D. (1991) Diurnal variations of aerosol acidity, sulfate, and ammonia
45 in the atmosphere. Presented at: 84th annual meeting and exhibition; June; Vancouver, BC, Canada.
46 Pittsburgh, PA: Air and Waste Management Association; paper no. 91-89.9.
47
48 Wolff, G. T. (1984) On the nature of nitrate in coarse continental aerosols. Atmos. Environ. 18: 977-981.
49
50 Wolff, G. T.; Korsog, P. E. (1985) Estimates of the contributions of sources to inhalable paniculate
51 concentrations in Detroit. Atmos. Environ. 19: 1399-1409.
52
April 1995 6-209 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wolff, G. T.; Countess, R. J.; Groblicki, P. J.; Ferman, M. A.; Cadle, S. H.; Muhlbaier, J. L. (1981)
2 Visibility-reducing species in the Denver "brown cloud"—II. sources and temporal patterns. Atmos.
3 Environ. 15: 2485-2502.
4
5 Wolff, G. T.; Ferman, M. A.; Kelly, N. A.; Stroup, D. P.; Ruthkosky, M. S. (1982) The relationships between
6 the chemical composition of fine particles and visibility in the Detroit metropolitan area. J. Air Pollut.
7 Control Assoc. 32: 1216-1220.
8
9 Wolff, G. T.; Korsog, P. E.; Stroup, D. P.; Ruthkosky, M. S.; Morrissey, M. L. (1985) The influence of local
10 and regional sources on the concentration of inhalable paniculate matter in southeastern Michigan.
11 Atmos. Environ. 19: 305-313.
12
13 Wolff, G. T.; Ruthkosky, M. S.; Stroup, D. P.; Korsog, P. E. (1991) A characterization of the principal PM-10
14 species in Claremont (summer) and Long Beach (fall) during SCAQS. Atmos. Environ. Part A
15 25: 2173-2186.
16
17 Zak, B. D.; Einfeld, W.; Church, H. W.; Gay, G. T.; Jensen, A. L.; Trijonis, J.; Ivey, M. D.; Homann, P. S.;
18 Tipton, C. (1984) The Albuquerque winter visibility study. Volume 1. Overview and data analysis.
19 Albuquerque, NM: U.S. Department of Energy, Sandia National Laboratories; Sandia report SAND84-
20 0173/1. Available from: NTIS, Springfield, VA; DE84014356.
21
22 Zemba, S. G.; Golomb, D.; Fay, J. A. (1988) Wet sulfate and nitrate deposition patterns in eastern North
23 America. Atmos. Environ. 22: 2751-2761.
24
25 Zhang, X.; Turpin, B. J.; McMurry, P. J.; Herring, S. V.; Stolzenburg, M. R. (1994) Mie theory evaluation of
26 species contributions to the 1990 wintertime visibility reduction in the Grand Canyon. J. Air Waste
27 Manage. Assoc. 44: 153-162.
28
29 Zoller, W. H.; Gordon, G. E. (1970) Instrumental neutron activation analysis of atmospheric pollutants utilizing
30 Ge(Li) 7ray detectors. Anal. Chem. 42: 257-265.
April 1995 6-210 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX 6A:
TABLES OF CHEMICAL COMPOSITION OF PM
April 1995 6A-1 DRAFT-DO NOT QUOTE OR CITE
-------
Table 1. Summary of PM2.5 Studies
EAST
Smoky Mtns
Shenandoah
Camden
Philadelphia
Deep Creek
Roanoke
Raleigh
Watertown
Hartford
Boston
Res.Tr. Pk.
Charlotte
Ref
1
1
2
3
4
5
5
6,7
8
8
8
20
Note WEST
Boise
Tarrant CA
b Five Points CA
Riverside CA
c San Jose
d Honolulu
d Winnemucca NV
Portland
a Seattle
a Southern California
a San Joaquin Valley
e Phoenix
Nevada
Ref
5
8
8
8
8
8
8
8
8
9,31
10
11
12
Note
d
a
a
a
a
a
a
a
a
g-n
i
j
f
Table 1 . Summary of Coarse Fraction
EAST
Smoky Mtns
Shenandoah
Camden
Philadelphia
Watertown
Hartford
Boston
Res.Tr. Pk.
Ref
1
1
2
3
6,7
8
8
8
Note WEST
o Tarrant CA
o Five Points CA
b Riverside CA
ab San Jose
o,p Honolulu
a,o Winnemucca NV
a,o Portland
a,o Seattle
Southern California
San Joaquin Valley
Phoenix
Ref
8
8
8
8
8
8
8
8
9,31
10
11
Note
a,o
a,o
a,o
a,o
a,o
a,o
a,o
a,o
g
i
j
CENTRAL
Albuquerque
St. Louis
Steubenville
Harriman
Portage
Topeka
Inglenook AL
Braidwood IL
Kansas City KS
Minneapolis
St. Louis
Kansas City MO
Akron
Cincinnati
Buffalo
Dallas
El Paso
Denver
Urban Denver
Non-urban Denver
Chicago
Houston
St. Louis
Harriman
St. Louis
Steubenville
Brownsville
Ontario
Studies
CENTRAL
St. Louis
Steubenville
Harriman
Portage
Topeka
Inglenook AL
Braidwood IL
Kansas City KS
Minneapolis
St. Louis
Kansas City MO
Cincinnati
Buffalo
Ref
5
6,7
6,7
6,7
6,7
6.7
8
8
8
8
8
8
8
8
8
8
8
13
14
14
15
16
17
17
18
21
24
37
Ref
6,7
6,7
6,7
6,7
6,7
8
8
8
8
8
8
8
8
Note
d
a
a
a
a
a
a
a
a
a
a
a
m
aa
k
n
I
Note
o-P
o,P
o.P
o,P
o,P
a,o
a,o
a,o
a,o
a,o
a,o
a,o
a,o
April 1995
6A-2
DRAFT-DO NOT QUOTE OR CITE
-------
Table 1. Summary of Coarse Fraction Studies (Cont.)
EAST
EAST
Smoky Mtns
Shenandoah
Camden
Philadelphia
Kingston
Watertown
Hartford
Boston
Res.Tr. Pk.
Ref Note
Table
Ref Note
1 o,q
1 o,q
2 b
3 ab
6,7 p,q
6,7 p,q
8 a,q
8 a,q
8 a,q
WEST
1. Summary of PM10
WEST
Tan-ant CA
Five Points CA
Riverside CA
San Jose CA
Honolulu HI
Winnemucca NV
Portland OR
Seattle
Southern California
San Joaquin Valley
Phoenix
San Fran. Bay
San Jose
Palm Springs
Pocatello, ID
Tuscon
Rillito, AZ
Ref
Note
CENTRAL
Dallas
El Paso
Denver
Chicago
Houston
St Louis
Harriman
St. Louis
Brownsville
Ontario
Ref
8
8
13
15
16
17
17
18
24
37
Note
a,o
a,o
o
s
o
k,r
n
I
Studies
Ref
8
8
8
8
8
8
8
8
9,31
10
11
29
29
38
39
40
42
Note
a.q
a,q
a,q
a,q
a,q
a,q
a,q
a,q
g,h
i
i
V
w
t
u
CENTRAL
St. Louis
Harriman
Steubenville
Portage
Topeka
Inglenook AL
Braidwood IL
Kansas City KS
Minneapolis
St. Louis
Kansas City MO
Akron
Cincinnati
Buffalo
Dallas
El Paso
Denver
Chicago
Houston
St. Louis
Harriman
St. Louis
Brownsville
Utah Valley
Ontario
SE Chicago, IL
Ohio
Ref
6,7
6,7
6,7
6,7
6,7
8
8
8
8
8
8
8
8
8
8
8
13
15
16
17
17
18
24
26
37
41
43
Note
p,q
p,q
p-q
p,q
p-q
a,q
a,q
a,q
a,q
a-q
a,q
a,q
a,q
a,q
a,q
a,q
q
s
q
X
I
y
April 1995
6A-3 DRAFT-DO NOT QUOTE OR CITE
-------
Table 1 Footnotes
a. Inhalable Particle Network (IPN) Data. Only represents days of elevated
concentrations, i.e., dichot filter loadings >50 ug/cm2.
b. Data from Site 28 only.
c. Average of all 6-hr samples.
d. Avg over all day/nite samples.
e. Average of all 12-hr samples at 2 incin. sites and 2 background sites.
Only XRF values which exceeded associated uncertainties more than half the time
at all four sites were included.
f. Average from Sparks site and Reno site.
g. Sampling only during intensive episodes.
h. Averages based on 12-hr day/nite samples. There were 59 sampling days at
Claremont and 23 sampling days at Long Beach.
i. Avg over all sites: Stockton, Crow's Landing, Fresno, Kern, Fellows, and Bakersfield.
j. Average of Central Phoenix, West Phoenix, and Scottsdale sites.
k. Average of RAPS site 106.
I. Average from Walpole, Windsor 1, and Windsor 2 sites.
m. Average of 3 urban sites: Auraria, Federal, and Welby.
n. Median VAPS values from Central site.
o. 2.5-1 Sum.
p. Coarse concentrations may be 30% or more underestimated
due to losses from handling filters.
q. PM15.
r. 2.4-20 um.
s. No upper size cutoff on VAPS inlet.
t. Average of Palm Springs and Indio, CA.
u. Avg. of Downtown Tuscon, Orange Grove, Craycroft, and Corona de Tuscon sites.
v. Mean of annual avgs (1988-1992) from ~ 9 sites in Alameda, San Francisco,
and Santa Clara counties.
w. 24-hr average of day/nite concentrations at two sites in San Jose.
x. PM20. Average from RAPS site 106.
y. Avg. of Follansbee, Mingo, Sewage Plant, Steubenville, and WTOV Tower sites.
z. Average of urban sites: Fresno, Bakersfield and Stockton.
aa. Average of non-urban sites: Brighton and Tower.
ab. Castor Avenue site only.
April 1995 6A-4 DRAFT-DO NOT QUOTE OR CITE
-------
mparison
osition fo
Co
1)
co
*"*
8
§
EJ
u
8
}
Pi n _ 7
* 73 5
Q 5 CO
/•^V /^N ^**\
—i tS CO
to
*— <
oo
4
CO
oo
ON
in so oo
IT) OO OO
^*- OO ^** "**-
tS ^ CS O
/-•v X*v /-N /-*\
« tS CO it
c
4
00
00
*
•i
2 S
fl J3
00 00
co
.2
i
fi-S
C4
*r*
•a
*s
1
I
*
"1 tS
ts
April 1995
6A-5 DRAFT-DO NOT QUOTE OR CITE
-------
-
H >
o
50
|o
Ji
8
S
4)
u
t
t-i ,
— • o
FP & IP(2.5-15). 24-hr
(midnite-midnite), every
other No Carbon
day.
24-hr
6th day
F+C(2.5-15)
sample every
ere included. No
§
OO .i
» i3
Fl
3
\
n -^ «N - «
»-< CU (^ ,_, U i ;*3
X-S«« 5-vJ§ ^ x^ S»
^ H tS Pi OH CO 00
5'll'lS
3
•a
1
a
1
09
j§ a
VO
OO
Ov
April 1995
6A-6
DRAFT-DO NOT QUOTE OR CITE
-------
>
-La
H
6
o
•z
o
H
O
CJ
n
h— <
a
10
11
12
13
14
San Joaquin Valley
6 sites
Aerosol
Composition
Phoenix PM Study
Phoenix
4 sites
Also comparison
aerosol data from
Denver, Reno, and
Sparks
Denver
Denver (SCENIC)
6/88-6/89
10/89 - 1/90
Sept'89 -
Jan'90
1/11-1/30/82
Nov'87-Jan'88
24-hr PM10 & PM2.5
every 6 days.
Mass, elements, ions
(K+,SO4=,NH4+,Na+),
EC, OC
F&C mass, elements,
uncertainties from 6 sites
6-hr samples, 2x/day,
(0600-1200, 1300-1900)
PM10 & PM2.5: mass,
elements, HNO3, SO2,
NH3, FP N03 and SO4,
ionic species, OC, EC.
Dichotomous sampler,
OC, EC, nitrate, sulfate
2x daily (0900-1600,
1600-0900). PM2.5 mass,
comp, sulfate, nitrate,
OC, EC, ionic species,
gases
1) Summary of annual geometric
avg, arith. avg, max 24-h PM10
and PM2.5 mass by site.
2) Ann. Avg Mass and comp. for
PM10 and PM2.5 by site.
1) temporal variation of PM2.5
mass at 4 sites.
2) Mean, SD, & Max: PM2.5,
EC, OC, NO3, SO4-, NH4+ and
elements for 3 Phoenix sites
3) Same for Denver (11/87-1/88)
4) Same for Reno (11/86-1/87)
5) Same for Sparks (11/86-1-87)
1) Measured PM2.5 and Coarse,
elements, OC, EC, nitrate,
day /night samples; light
extinction.
1) Avg, SD, Min, Max PM2.5
mass for 6 sites.
2) Avg, SD, Min, Max, for
PM2.5 mass, ionic species, EC,
OC, elements for 3 sites.
3) Source profiles
4) SCE for 4 sites by day and
night
PM10 highest hi
winter and
dominated by F
mass; C >50% of
PM10 in summer
and fall. Data show
spatial and temporal
variations of PM10
and PM2.5
Moudi size-resolved
(0-5.6 fim in 9 bins)
mass, NO3, SO4,
OC, EC.
Source
apportionment for
F&C particles and
extinction.
Source
Apportionment study
-------
ON
>
00
O
O
z
s
O
C
s
w
O
I— I
H
tfl
15
16
17
18
19
20
Chicago
Houston
St. Louis &
Harriman
St. Louis
1) Albuquerque
2) Denver
Charlotte (2 incin
sites and 2 control
sites).
July, 1994
10-19
September
1980
Sept'85-
Aug'86
7/76-8/76
(St.Louis)
RAPS data for
St. Louis exist
for 5/75-3/77
but were not
in this article
1) Jan 3-4/83
2) Jan 19-20,
'82
4/30-6/4/92 &
9/21-9/28/92.
VAPS & Dichot. FM,
CM, OC, EC, elements,
SO2, MONO, HNO3.
Dichotomous sampler:
0.1-2.5, 2.5-15. 4 sites.
Consecutive 12h samples.
Daily F & C (2.5-10/mi).
Also SO2, NO2, and O3.
F(< 2.4) &C (2.4-20) 6-
12 hr. No Carbon.
F & C (2.5-10) +
Carbon, Nitrate & Sulfate
(1C) 12-h samples,
Day/Night:
0700-1900,1900-0700.
VAPS F&C + Acid
gases.
no carbon. 12-hr samples
1) Avg VAPS mass, SD, uncert.
for F&C, OC, EC.
1) Average F&C mass, elements,
Carbon, NH4, NO3, Sulfate
1) Mean, SD, range for PM10,
PM2.5, SO4, H+, SO2, NO2, O3
for both sites.
1) 2-mo avg of F-f-C mass,
metals, sulfate, for one site.
2) F+C composition of selected
samples (different sites) during
events.
3) CMB apportionment of F+C
fractions to 6 components (crustal
shale, crustal limestone,
ammonium sulfate, motor
vehicles, steel, paint).
4) Plots ofintercity variations in
source component concentrations
1) Mean daytime and nightime
comp. of F&C, EC, OC, nitrate,
sulfate, for each site.
2) Source app. of Denver winter
FP composition.
l)'Mean ambient FP cone. +-
XRF unc. @ 4 sites
2) CMB results for FP.
Source
apportionment.
1) Crude CMB
source apportionment
of FP with 6
sources.
More complete
source app results in
Lewis & Enfield
paper.
-------
$
1) avg F mass 4- comp.
2) avg source contributions by
SRFA
3) SRFA-derived source profil
|
u
o
^
o"
uT
S
oo
0
Q
S3
p— .
«
J3
>
^
^ 53 .^3
"g w? «J
^ ?i ^5
1 1 g
i-H
1
We
CO
3
O
S
ex,
CO
W CO
u .a
0
OH
O\
i
o
o.
<4-l
o
E $
••H *^^
> "2
£ ^
PC CQ
rs
cs
&
Q
6
1
+ 60 °
C^U rT
.
00S'gH
«N 1
A w
•*—S W
O *
is
o?oo
^ o
^-^ oo
UH -M
OO
r-
r~-"
i— i
X
1
o
fi
CO
CN
•^
O
8
1
> .
M G
^j
o "8
Z 6
S
£
«c
^
i?l
lH
FPMES
door/outdoc
'~> C
*"^ .1^
1/3
+
.I* OS
W< •
1 1
0> ct
H3 "5
> •-
11
ll
•k
C ^
8|
u -5
+ =3
8S|
rT 6 8
•& 'S
•a "a
H tl
!§
§1
> b
rTco'
-*— »
OS
•a
1
u
rs
1
*^
s
«J
Ul
® .
0 ja'S
1) Seasonal avg SCE for PMl
3 sites, (geological, motor ve
construction, vegetative, sulfa1
nitrate, OC, EC)
I.S ff
9> O..S
« wl S
^ Ul
£j m "O
2 "^ o
PH JC **
^VO^
"r-
00
oo
1
(V ^
t"H ^^ r3^
|
It
SO
(S
SI
§5 ffi
.s S o
c3 55 ^i
pSll
.^
^
1
•"a
-o
ea
II
o c^
O2 U4
•gSri
|o*
"3 ^ T3
tfc rt S
fefl
1 * o
OO >^ j-
OS C ^ co
-H 0 Q 3
C3
3
O
U
c3
^™3
u
2
C/3
r-
ts
April 1995
6A-9 DRAFT-DO NOT QUOTE OR CITE
-------
us
45%
mb
co
tes
1. Highest PMIO
mass during Nov,
Dec, Jan.
2. Wood
contribu
PM
of
_
•a .8
Sw
.2 a
•° «> oi
*.S 3
U&2
•B ^ S
Q-SE
5
S
00
- 8 2
Is |g
G
S
8 a
II
~H eu ja
^h »^H '"^ . » gj . **iH
°rtob^g8o».
*7? **^ S *" "^ "M 3 W
3 0 ^
55«ts
A.
>>
0)
.
G
— <
^ r-
I o
i*§
|u
«>
re
* 1
IT) ft-
il
• M » *
•
ass, elements, ions,
lfate, nitrate, Carbc
.a s s
£
oo
oo
vo
(O
o
fa .52 -o .S2 oo
3 CL, C P-ON
4)
'a
>
a
I
11
a «
b? \O
el
§ 8:
s
B S3
•u ^ O
$
PQ §
'
•< o
§ I.
Si
oo
n
ON
«S
April 1995
6A-10 DRAFT-DO NOT QUOTE OR CITE
-------
VO
Ui
31
1) Claremont
(SCAQS)
2) Long Beach
(SCAQS)
1) Summer '87
(59d)
2) Fall '87
(23d)
Continuous 12-hr
PMIO and PM2.5.
Mass, elements, ionic
species, EC, OC
1) Mean, SD, & Max: PMIO,
FPM, CPM, EC, OC, NO3,
SO4=, NH4+.
2) Mean values of above species
during intensive and non-intensive
periods.
3) Day/nite values of above
4) PMIO and PM2.5 mass
balances
5) Summary of EC, OC data.
32
CADMP ~ 8 sites:
Gasquet, Fremont,
Bakersfield,
Yosemite, Sequoia,
Long Beach, Los
Angeles, Azusa
Summer '88
2 samples every 6th day.
0600-01800,1800-0600.
PM2.5, PMIO. Mass,
ionic species,
1) Graph of avg PMIO & PM2.5
mass and ratio @ 8 sites
2)Graphs of PMIO & PM2.5 ionic
concentrations.
Ask Chow/Watson
for raw data.
•n
H
6
o
s
o
H-
w
o
90
n
t— i
3
33
Central California -
53 sites in SF Bay
area, Sacramento
Valley, San Joaquin
Valley, North and
South Central
Coast, Mountain
Counties
1) 1989
2) July &
August, 1988
PMIO every 6th day.
Sulfate and nitrate
measured on a subset of
these samples.
1) 1989 Max and Avg PMIO
mass, Sulfate, and Nitrate for
-53 sites.
2) Summertime '88 Avg, SD, and
Max PMIO and PM2.5 Mass,
comp, OC,EC, Ionic species, for
3 SJVAQS sites. [Annual data
summary is in ref 20].
34
Birmingham
1986-1989
Daily 24-hr PMIO mass.
Also Ozone data.
No composition data.
1) Table of percentile points of the
distribution of PMIO, 03, T,
DewPoint, Pneumonia, Chronic
obstructive pulmonary disease.
2) Avg PMIO and O3 by season
Aside:
Indoor/Outdoor
ratios of 0.63 for
PMIO were reported
in Tuscon.
-------
35
36
37
38-43
Philadelphia
State College, PA
Southern Ontario
3 sites
Miscellaneous sites
14 sites
1973-1980
summer 1990
Jan-Nov, 1991
1984-1990
24-hr (midnite-midnite)
TSP.
No composition data.
Indoor, outdoor, personal
SO4=,H+, andNH3
24-hr, midnite-midnite,
every 6th day. PMIO
dichot sampler.
PMIO concentrations.
1) Table of percentile points of the
distribution of TSP, SO2, T,
DewPoint, Mortality.
l)Avg mass, elements, for F&C
fractions, for 3 sites. NO OC,
EC.
1) Measured PMIO mass and avg
source contributions (up to 10
source categories).
Validation of
personal exposure
models
Primary reference is
Ref 10.
0\
T1
H
6
o
z
o
H
o
H
w
o
h-H
H
tn
-------
List of References
1. Stevens, R.K., Dzubay, T.G., Lewis, C.W., and Shaw, R.W. (1984). Source apportionment
methods applied to the determination of the origin of ambient aerosols that affect visibility in
forested areas. Atmos. Environ. 18 261.
2. Dzubay, T.G. and Stevens, R.K. (1988). A composite receptor model applied to Philadelphia
aerosol. Environ. Sci. Techn. 22 46.
3. Unpublished data from J. Pinto, U.S. EPA, Research Triangle Park, NC (1995).
4. Vossler, T.L., Lewis, C.W., Stevens, R.K., Dzubay, T.G., Gordon, G.E., Tuncel,
S.G.,Russworm, G.M., and Keeler, G.J. (1989). Composition and origin of summertime air
pollutants at Deep Creek Lake, Maryland. Atmos. Environ. 23_ 1535.
5. Stevens, R.K., Hoffman, A.J., Baugh, J.D., Lewis, C.W., Zweidinger, R.B., Cupitt, L.T.,
Kellogg, R.B. and Simonson, J.H. (1995). A comparison of air quality measurements in
Roanoke, Va, and other integrated air cancer project monitoring locations. In: Measurement of
Toxic and Related Air Pollutants. Proceedings of the 1993 U.S. EPA/A&WMA International
Symposium. A&WMA, Pittsburgh, PA, p!85.
6. Spengler, J.D. and Thurston, G.D. (1983). Mass and elemental composition of fine and coarse
particles in six US cities. JAPCA 33 1162.
7. Dockery, D.W., Pope, C.A., Xu, X., Spengler, J.D., Ware, J.H., Fay, M.E., Ferris, B.G. and
Speizer, F.E. (1993). An association between air pollution and mortality in six U.S. cities. New
Eng. J. Medicine 329 1753.
8. Davis, B.L. Johnson, L.R., Stevens, R.K., Courtney, W.J. and Safriet, D.W. (1984). The quartz
content and elemental composition of aerosols from selected sites of the EPA inhalable particulate
network. Atmos. Environ. 18 771.
9. Chow, J.C., Watson, J.G., Fujita, E.M., Lu, Z., Lawson, D.R. and Ashbaugh, L.L. (1994).
Temporal and spatial variations of PM2.5 and PM10 aerosol in the Southern California Air
Quality Study. Aerosol Sci. and Tech. 21 2061.
10. Chow, J.C., Watson, J.G., Lowenthal, D.H., Solomon, P.A., Magliano, K.L. Ziman, S.D. and
Richards, L.W. (1993). PM10 and PM2.5 compositions hi California's San Joaquin Valley.
Aerosol Sci & Tech. 18 105.
11. Unpublished data from the Desert Research Institute (1995).
12. Chow, J.C., Watson, J.G., Pritchett, L., Lowenthal, D.H., Frazier, C., Neuroth, G. and Evans
K. (1990). Wintertime visibility in Phoenix, Arizona. Paper 90-66.6 in Proceedings of the 83rd
National Meeting of the Air & Waste Management Association, Pittsburgh, PA, 24-29 June,
1990.
13. Lewis, C.W., Baumgardner, R.E., Stevens, R.K., and Russworm, G.M. (1986). Receptor
modeling study of Denver whiter haze. Environ. Sci. Techn. 20 1126.
April 1995 6A.13 DRAFT-DO NOT QUOTE OR CITE
-------
14. Watson, J.G., Chow, J.C., Richards, L.W., Neff, W.D., Andersen, S.R., Dietrich, D.L. Houck,
J.E. and Olmez, I. (1988). The 1987-88 metro Denver brown cloud study, vol 3: data
interpretation. Desert Research Institute. Document No. 8810 1F3.
15. Unpublished data (1995).
16. Johnson, D.L., Davis, B.L., Dzubay, T.G., Hasan, H., Crutcher, E.R., Courtney, W.J.,
Jaklevic, J.M., Thompson, A.C., and Hopke, P.K. (1984). Chemical and physical analyses of
Houston aerosol for interlaboratory comparison of source apportionment procedures. Atmos.
Environ. 18, 1539.
17. Dockery, D.W., Schwartz, J. and Spengler, J.D. (1992). Air pollution and daily mortality:
associations with particulates and acid aerosols. Environ. Research 5.9 362.
18. Dzubay, T.G. (1980). Chemical elements balance method applied to dichotomous sampler data.
In: Annals of the New York Academy of Sciences 338 126.
19. Stevens, R.K. (1985). Sampling and analysis methods for use in source apportionment studies
to determine impact of wood burning on fine particle mass. Environment International H 271.
20. Mukerjee, S., Stevens, R.K., Vescio, N., Lumpkin, T.A., Fox, D.L., Shy, C. andKellogg, R.B.
(1993). A methodology to apportion ambient air measurements to investigate potential effects on
air quality near waste incinerators. In: Proceedings of the 1993 Incineration Conference,
Knoxville, TN. 527.
21. Koutrakis, P. and Spengler, J.D. (1987). Source apportionment of ambient particles in
Steubenville, OH using specific rotation factor analysis. Atmos. Environ. 2_1 1511.
22. Chow, J.C., Watson, J.G., Ono, D.M. and Mathai, C.V. (1993). PM10 standards and
nontraditional paniculate source controls: a summary of the A&WMA/EPA international specialty
conference. Air & Waste 43 74.
23. Solomon, P.A. and Moyers, J.L. (1986). A chemical characterization of wintertime haze hi
Phoenix, Arizona. Atmos. Environ. 20 207.
24. Ellenson, W.D., Schwab, M., Egler, K.A., Shadwick, D. and Willis, R.D. (1994). Draft
technical report for the pilot project of Lower Rio Grande Valley environmental study. ManTech
Environmental Technology, Inc. Submitted to the U.S. EPA.
25. Chow, J.C., Watson, J.G., Frazier, C.A., Egami, R.T., Goodrich, A. and Ralph, C. (1988).
Spatial and temporal source contributions to PM10 and PM2.5 hi Reno, NV. In: Transactions:
PM10 Implementation of Standards, an APCA/EPA international specialty conference. Air
Pollution Control Association, 439.
26. Pope, C.A., Schwartz, J. and Ransom, M.R. (1992). Daily mortality and PM10 Pollution in Utah
Valley. Archives of Environ. Health 47 211.
27. Fairley, D. (1990). The relationship of daily mortality to suspended particulates hi Santa Clara
county 1980-1986. Environ. Health Perspectives £2 159.
April 1995 6A-14 DRAFT-DO NOT QUOTE OR CITE
-------
28. Chow, J.C., Watson, J.G., Lowenthal, D.H., Solomon, P.A., Magliano, K.L. Ziman, S.D. and
Richards, L.W. (1992). PM10 source apportionment in California's San Joaqin Valley. Atmos.
Environ. 26A 3335.
29. Chow, J.C., Fairley, D., Watson, J.G., De Mandel, R., Fujita, E., Lowenthal, D.H., Lu, Z.,
Frazier, C.A., Long, G. and Cordova, J. (1994). Source apportionment of wintertime PM10 at
San Jose, CA. J. Environ Engineers, in press.
30. Watson, J.G., Chow, J.C., Lu, Z., Fujita, E.M., Lowenthal, D.H., Lawson, D.R. and
Ashbaugh, L.L. (1994). Chemical mass balance source apportionment of PM10 during the
Southern California Air Quality Study. Atmos. Environ. 12 2061.
31. Wolff, G.T., Ruthkosky, M.S., Stroup, D. adn Korsog, P.E. (1991). A characterization of the
principal PM10 species in Claremont (summer) and Long Beach (fall) during SCAQS. Atmos.
Environ. 25A 2173.
32. Ashbaugh, L.L., Watson, J.G. and Chow, J. (1989). Estimating fluxes from California's dry
deposition monitoring data. Paper 89-65.3 in: Proceedings of the 82nd Annual Meeting of the
Air & Waste Management Association, Anaheim, CA.
33. Chow, J.C., Watson, J.G., Solomon, P.A., Thuillier, R.H., Magliano, K.L. Ziman, S.D.,
Blumenthal, D.L. and Richards, L.W. (1994). Planning for SJVAQS/AUSPEXParticulate matter
and visibility sampling and analysis. In: Planning and Managing Regional Air Quality, ed. by
Paul Solomon. CRC Press, Inc.
34. Schwartz, J. (1994). Air pollution and hospital admissions for the elderly in Birmingham,
Alabama. Am. J. Epidemiology. 139 589.
35. Schwartz, J. and Dockery, D.W. (1992). Increased mortality in Philadelphia associated with daily
air pollution concentrations. Am. Rev. Respir. Dis. 145 600.
36. Suh, H.H., Koutrakis, P. and Spengler, J.D. (1993). Validation of personal exposure models for
sulfate and aerosol strong acidity. J. Air Waste Manage. Assoc. 43 845.
37. Conner, T.L., Miller, J.M., Willis, R.D,, Kellogg, R.B. and Dann, T.F. (1993). Source
apportionment of fine and coarse particles in Southern Ontario, Canada. In: Proceedings of the
86th Annual Meeting of the Air & Waste Management Association, Denver, CO. Paper 93-TP-
58.05.
38. Kim, B.M., Zeldin, M. and Liu, C. (1992). Source apportionment study for state implementation
plan development in the Coachella Valley. In: PM10 Standards and nontraditional paniculate
source controls. Chow and Ono, Eds., Air & Waste Management Association, Pittsburgh, PA,
979.
39. Houck, J.E., Rau, J.A., Body, S. and Chow, J.C. (1992). Source apportionment - Pocatello,
Idaho PM10 nonattainment area. Ibid, 219.
40. Chow, J.C., Watson, J.G., Lowenthal, D.H., Frazier, C.A., Hinsvark, B.A., Pritchett, L.C. and
Neuroth, G.R. (1992). Wintertime PM10 and PM2.5 chemical compositions and source
contributions in Tuscon, Arizona. Ibid, 231.
April 1995 6A-15 DRAFT-DO NOT QUOTE OR CITE
-------
41. Vermette, S.J., Williams, A.L. and Landsberger, S. (1992). PMIO source apportionment using
local surface dust profiles: examples from Chicago. Ibid, 262.
42. Thanukos, L.C., Miller, T., Mathai, C.V., Reinholt, D. and Bennett, J. (1992). Intercomparison
of PMIO samplers and source apportionment of ambient PMIO concentrations in Rillito, Arizona.
Ibid, 244.
43. Skidmore, L.W., Chow, J.C. and Tucker, T.T. (1992). PMIO air quality assessment for the
Jefferson County, Ohio air quality control region. Ibid, 1016.
April 1995 6A-16 DRAFT-DO NOT QUOTE OR CITE
-------
PM2.5 COMPOSITION (24-h AVG)
EASTERN U.S.
Units = ug/m3
>
I
I—*
^1
O
o
2
o
H
O
a
o
o
H- (
H
m
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
1
Smoky Mtns
9/20-26/78
12
12
24.00
2.22
1.10
0.30
12.00
<0.054
<0.003
0.018
0.016
<0.010
0.003
0.028
0.040
0.097
3.744
0.001
0.038
<0.006
<0.004
0.009
1
Shenandoah
7/23-5/08/80
12
28
27.00
0.44
1.12
13.60
<0.105
<0.003
0.008
0.035
0.010
0.005
0.054
0.061
0.052
4.539
0.001
0.116
<0.010
<0.010
0.011
2(b)
Camden
7/14-8/13 '82
6am-6pm-6am
12
50
28.70
2.05
1.87
<0.48
11.20
0.053
0.001
0.029
0.040
0.002
0.003
0.002
0.091
0.101
0.006
0.001
0.146
0.011
0.249
4.200
0.079
0.002
0.103
<0.012
< 0.002
<0.027
0.013
0.082
3
Philadelphia
7/25-8/14/94
24
21
32.18
4.51
0.76
0.114
0.009
0.058
0.026
0.007
0.127
0.060
0.023
0.003
0.070
0.007
0.015
0.019
3.251
<0.002
0.165
<0.042
<0.013
0.041
4(c)
Deep Creek
8/83
4x daily
6
98
40.00
1.45
0.18
0.57
0.001
0.005
0.048
0.058
0.044
0.003
0.034
0.048
6.700
0.001
0.003
0.150
0.001
0.013
5(d)
Raleigh
1/85-3/85
7am-7pm-7am
12
30.30
10.00
0.50
0.009
0.001
0.028
0.018
0.007
0.000
0.020
0.044
0.159
0.003
0.001
0.096
1.729
0.002
0.076
0.003
0.015
5(d)
Roanoke
10/88-2/89
7am-7pm-7am
12
19.90
7.30
1.50
0.176
0.002
0.005
0.047
0.053
0.001
0.007
0.114
0.177
0.012
0.001
0.027
1.177
0.002
0.077
0.004
0.083
6,7
Watertown
5/79-6/81
0000-0000
24
354
14.90
6.50
20.3
0.088
0.041
0.084
0.074
0.004
0.009
0.329
1.800
0.001
0.100
0.022
8(a)
Hartford
1980
24
2
26.75
0.035
0.036
0.070
0.003
0.043
0.125
0.171
0.007
0.010
0.510
2.219
0.001
0.177
0.002
0.017
0.079
8(a)
Boston
1980
24
1
34.80
0.002
0.020
0.070
0.004
0.035
0.121
0.096
0.001
0.012
0.009
0.285
3.869
0.001
0.144
0.020
0.046
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
oo
o
T1
o
H
O
C
o
H
W
n
HH
H
W
PM2.5 COMPOSITION (24-h
Ret
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a)
Res.Tr.Pk
1980
24
3
28.77
0.073
0.002
0.007
0.035
0.016
0.120
0.148
0.003
0.001
0.042
0.106
2.835
0.002
0.350
0.018
9(9)
Los Angeles
Summer'87
4,5 and 7
11 days
41.10
8.27
2.37
4.34
9.41
0.035
0.022
0.015
0.013
0.022
0.093
0.022
0.063
0.099
0.041
0.024
0.016
0.202
0.005
0.060
0.038
2.832
0.013
0.052
0.019
0.005
0.006
0.090
AVG)
9(9)
Los Angeles
Fall'87
4 and 6
6 days
90.20
18.46
7.28
22.64
4.38
0.250
0.015
0.043
0.065
0.335
0.453
0.025
0.273
0.557
0.217
0.075
0.043
0.466
0.007
0.046
0.185
1.998
0.011
0.520
0.028
0.060
0.007
0.298
WESTERN U.S.
10(i)
San Joaquin Valley
6'88-6'89
24
~35
29.89
4.87
3.24
8.17
3.00
0.152
0.012
0.010
0.096
<0.007
0.094
0.003
0.096
0.180
0.188
0.006
0.016
0.007
0.029
0.001
1.242
<0.002
0.001
0.460
<0.015
0.002
0.017
0.015
0.078
11(j)
Phoenix
10/13/89-1/17/90
6 h, 2x/day
~ 100 days
29.37
10.10
7.47
3.60
1.33
0.130
< 0.020
<0.106
0.011
0.170
<0.018
0.365
0.003
0.015
0.216
0.207
0.023
<0.006
0.003
<0.051
0.039
<0.0025
0.437
<0.033
< 0.002
0.430
< 0.028
<0.030
<0.016
0.056
5(d)
Boise
12/86-3/87
7am-7pm-7a
12
35.70
12.70
1.70
0.102
0.002
0.014
0.026
0.122
0.001
0.011
0.022
0.145
0.002
0.002
0.045
0.603
0.001
0.069
0.001
0.019
Units =
12(f)
Nevada
ug/m3
8(a)
Tarrant CA
11/86-1/87 1980
00-2400
24 24
24 6
56.92
19.97
15.17
2.43
1.67
0.275
0.001
0.013
0.033
0.215
0.145
0.002
0.010
0.310
0.280
0.015
0.006
0.041
0.115
0.001
0.765
0.000
0.860
0.004
0.043
0.009
0.033
57.05
0.177
0.102
0.455
0.002
0.047
0.316
0.186
0.032
0.003
0.619
2.578
0.583
0.010
0.095
8(a)
Five Points C
1980
24
3
31.80
0.239
0.015
0.150
0.004
0.001
0.024
0.216
0.244
0.005
0.025
0.007
0.087
1.129
0.001
0.656
0.005
0.006
0.016
8(a)
Riverside CA
1980
24
4
35.18
0.036
0.037
0.301
0.009
0.040
0.127
0.120
0.007
0.007
0.376
1.653
0.001
0.234
0.003
0.029
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM2.5 COMPOSITION (24-h AVG)
CENTRAL U.S.
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a)
San Jose CA
1980
24
6
36.28
0.123
0.001
0.188
0.089
0.050
0.003
0.043
0.148
0.248
0.006
0.006
0.013
0.891
0.852
0.292
0.002
0.061
8(a)
Honolulu
1980
24
1
21.10
1.127
0.017
1.024
0.518
0.004
0.018
0.726
0.371
0.020
0.002
0.002
0.071
0.313
2.363
0.063
0.001
0.011
8(a)
Winnemucca
1980
24
5
9.68
0,361
0.006
0.243
0.026
0.231
0.149
0.003
0.001
0.042
0.358
0.914
0.009
0.011
8(a)
Portland
1980
24
4
37.18
0.581
0.012
0.093
0.154
0.021
0.009
0.072
0.270
0.218
0.052
0.027
0.017
0.422
1.944
0.001
0.377
0.005
0.014
0.081
8(a)
Seattle
1980
24
1 .
10.70
0.002
0.006
0.019
0.037
0.002
0.024
0.098
0.080
0.004
0.006
0.006
0.215
0.831
0.001
0.092
0.059
5(d)
Albuquerque
12/84-3/85
7am-7pm-7am
12
20.60
13.20
2.10
0.077
0.085
0.059
0.036
0.045
0.074
0.000
0.237
0.507
0.076
0.007
13
Denver
1/11-30/82
6am-6pm-6a
12
~26
20.73
7.11
2.15
2.22
2.06
0.394
<0.002
0.031
0.103
0.047
0.006
0.052
<0.009
0.010
0.079
0.079
0.011
0.003
0.043
0.326
<0.003
0.709
0.277
<0.003
<0.027
0.046
14(m)
Urban Denver
11/87-1/88
9am-4pm-9am
7&17
~136
19.67
7.25
4.41
3.96
1.55
0.037
0.018
0.058
0.005
0.141
0.003
0.017
0.111
0.077
0.012
0.002
0.075
0.642
0.004
0.001
0.272
0.006
0.001
0.009
0.031
14(aa) 15
Non-urban Denver Chicago
11/87-1/88 7/94
9am-4pm-9am 0800-0800
7&17 24
~150 16
10.35 13.57
5.39
1.31
0.046
< 0.003
< 0.091
0.004
0.045
<0.029
0.011
<0.005
0.011
0.089
0.061
0.012
0.005
<0.002
0.022
<0.001
0.008
0.027
1.321
<0.042
<0.001
0.074
<0.049
<0.029
<0.009
0.052
D
O
O
2
O
H
O
G
O
H
W
O
*J
n
H-i
H
W
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM2.5 COMPOSITION (24-h AVG)
CENTRAL U.S.
Units = ug/m3
vo
vo
ON
to
o
"fl
H
6
O
z
o
H
O
d
o
H
m
o
h-H
H
W
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
8
Sb
Se
Si
Sn
Sr
Ti
V
Zn
16
Houston
9/10-19/80
12
20
38.60
5.68
1.42
0.59
14.61
0.123
<0.005
0.048
0.055
0.155
<0.003
0.032
<0.005
0.028
0.162
0.119
0.014
<0.38
0.004
0.028
0.465
<0.002
4.834
0.006
<0.002
0.210
<0.005
<0.002
<0.014
<0.008
0.084
6,7
Harriman
5/80-5/81
0000-0000
24
256
20.80
8.10
36.1
0.038
0.150
0.021
0.120
0.017
BQL
0.180
2.500
0.002
0.120
BQL
17 6,7
Harriman Kingston
9/85-8/86 5/80-6/81
0000-0000
24 24
330 169
21 .00 24.60
8.70
36.1
0.044
0.120
BQL
0.097
0.010
BQL
0.194
2.400
0.002
0.200
BQL
6,7
Portage
3/79-5/81
0000-0000
24
271
11.00
6.81
10.5
0.011
0.045
0.027
0.049
0.003
BQL
0.061
1.400
0.001
0.075
BQL
6,7
Topeka
8/79-5/81
0000-0000
24
286
12.50
6.05
11.6
0.045
0.250
0.031
0.090
0.004
BQL
0.163
1.100
0.000
0.190
BQL
8(a)
El Paso
1980
24
10
27.16
0.155
0.025
0.070
0.332
0.001
0.036
0.134
0.127
0.004
0.001
0.481
0.823
0.002
0.436
0.003
0.055
8(a)
Inglenook
1980
24
8
32.03
0.082
0.001
0.040
0.326
0.003
0.002
0.032
0.281
0.408
0.037
0.001
0.008
0.309
2.655
0.001
0.685
0.133
8(a)
Braidwood
1980
24
1
28.20
0.089
0.003
0.084
0.024
0.071
0.052
0.001
0.001
0.041
2.060
0.001
0.220
0.011
8(a)
Kansas City KS
1980
24
8
25.66
0.091
0.003
0.027
0.519
0.004
0.032
0.189
0.311
0.006
0.002
0.013
0.180
1.816
0.001
0.434
0.004
0.034
References are listed in Table 1 Appendix,
* Values for this size fraction are calculated
Associated notes are explained in Table 1.
from the average measured values reported for the other two size fractions.
-------
PM2.5 COMPOSITION (24-h AVG)
CENTRAL U.S.
Units = ug/m3
>
o
o
2
o
H
O
G
O
H
W
O
»
O
HH
m
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a)
Minneapolis
1980
24
6
15.50
0.004
0.047
0.103
0.001
0.035
0.087
0.092
0.005
0.001
0.308
0.907
0.001
0.169
0.045
8(a)
Kansas City MO
1980
24
3
16.77
0.007
0.064
0.213
0.002
0.021
0.140
0.142
0.006
0.001
0.369
0.763
0.177
0.046
8(a)
Akron
1980
24
7
36.09
0.046
0.012
0.039
0.110
0.010
0.037
0.609
0.268
0.085
0.006
0.059
0.412
3.419
0.008
0.522
0.009
0.150
8(a)
Cincinnati
1980
24
2
29.80
0.062
0.013
0.024
0.062
0.003
0.024
0.174
0.136
0.011
0.004
0.043
0.343
2.876
0.005
0.328
0.003
0.053
8(a)
Buffalo
1980
24
14
38.75
0.192
0.009
0.003
0.218
0.002
0.026
0.671
0.310
0.033
0.008
0.060
0.359
3.706
0.005
0.241
0.001
0.078
8(a)
Dallas
1980
24
4
28.93
0.111
0.033
0.223
0.691
0.005
0.043
0.248
0.125
0.015
0.002
0.018
1.066
1.514
0.442
0.007
0.002
0.054
8(a)
St. Louis
1980
24
5
23.06
0.119
0.003
0.025
0.090
0.018
0.076
0.126
0.002
0.002
0.020
0.277
2.333
0.002
0.170
0.023
18(k)
St. Louis
8-9/76
6-12
34.00
0.203
0.002
0.020
0.132
0.132
0.004
0.087
0.006
0.029
0.275
0.261
0.036
0.004
0.001
0.688
0.000
4.655
0.006
0.004
0.458
0.009
0.002
0.112
0.002
0.101
6,7 17
St. Louis St. Louis
9/79-6/81 9/85-8/86
0000-0000
24 24
306 311
19.00 17.70
8.10 8.00
10.3 9.7
0.078
0.101
0.052
0.190
0.021
0.003
0.327
2.100
0.002
0.160
BQL
6,7
Steubenville
4/79-4/81
0000-0000
24
499
29.60
12.80
25.2
0.042
0.097
0.092
0.590
0.029
0.005
0.216
4.700
0.005
0.290
0.011
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM10 COMPOSITION (24-hr AVG)
EASTERN U.S.
>
to
to
O
O
Z
O
H
O
c:
o
H
w
o
&
o
t—t
H
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
1(o,q)*
Smoky Mtns
9/20-26/78
12
12
29.60
2.22
1.10
0.30
12.00
BQL
BQL
0.023
0.338
BQL
0.003
0.146
0.148
BQL
0.111
3.744
0.001
0.618
0.018
BQL
0.009
1(o,q)*
Shenandoah
7/23-5/08/80
12
28
34.40
0.44
1.12
14.38
0.311
0.011
0.339
0.189
0.011
0.212
0.190
BQL
BQL
0.061
4.539
0.001
0.929
0.017
BQL
0.017
2(b)*
Camden
7/14-8/13 '82
6am-6pm-6am
12
50
40.10
2.05
2.29
0.57
11.20
0.603
0.001
0.044
0.400
0.002
0.072
0.002
0.581
0.252
0.017
0.001
0.146
0.015
0.303
4.430
0.260
0.002
1.713
BQL
0.002
0.065
0.020
0.112
3(ab)* 4(c) 5(d)
Philadelphia Deep Creek Raleigh
7/25-8/14/94 8/83 1/85-3/85
4x daily 7am-7pm-7a
24 6 12
21 98
40.60
4.51
0.76
0.439
0.012
0.479
0.073
0.021
0.479
0.160
0.126
0.010
0.206
0.009
0.042
0.032
3.251
1.098
0.030
0.092
5(d) 6,7(p,q)
Roanoke Watertown
10/88-2/89 5/79-6/81
7am-7pm-7a 0000-0000
12 24
354
24.20
8.94
0.110
0.250
0.389
0.350
0.009
0.011
0.405
2.000
0.001
1.100
0.022
8(a.q)
Hartford
1980
24
2
54.60
1.910
0.082
0.934
0.302
0.011
0.069
1.195
0.481
0.028
0.015
0.033
0.681
2.647
0.001
4.694
0.096
0.025
0.133
8(a,q)
Boston
1980
24
1
140.40
3.458
0.003
0.045
1.139
0.301
0.008
0.058
1.733
0.629
0.030
0.034
0.025
0.462
4.371
0.001
6.904
0.154
0.028
0.100
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM10 COMPOSITION (24-hr AVG)
WESTERN U.S.
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Suifate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,q)
Res.Tr.Pk
1980
24
3
36.93
0.679
0.002
0.010
0.121
0.002
0.026
0.302
0.216
0.006
0.001
0.042
0.119
3.058
0.002
1.737
0.021
0.025
9(g)
Los Angeles
Summer'87
4,5 and 7
1 1 days
67.40
11.61
3.19
9.47
11.28
0.758
0.007
0.070
0.016
0.585
1.119
0.023
0.022
0.836
0.237
0.335
0.033
1.632
0.005
0.187
0.084
3.353
0.008
2.040
0.018
0.077
0.005
0.114
9(9)
Los Angeles
Fall'87
4 and 6
6 days
98.70
23.35
8.49
27.50
5.39
0.847
0.019
0.127
0.072
1.190
0.880
0.042
0.178
2.192
0.460
0.287
0.063
0.518
0.005
0.099
0.251
2.262
0.010
2.162
0.024
0.165
0.009
0.293
10(i)
San Joaquin Valley
6'88-6'89
24
~35
74.05
10.59
5.62
10.55
3.62
3.570
0.051
0.015
1.057
0.487
0.010
0.087
1.633
0.820
0.037
0.010
0.059
0.061
0.004
1.463
0.001
8.037
0.014
0.147
0.014
0.094
11(D* 5(d)
Phoenix Boise
10/13/89-1/17/90 12/86-3/87
7am-7pm-7a
6h,2x/day 12
~ 100 days
62.45
14.56
8.30
4.46
2.34
2.67
BQL
0.01
0.01
2.10
BQL
0.56
0.01
0.04
1.47
0.88
BQL
0.05
BQL
BQL
0.01
0.05
0.06
BQL
0.62
BQL
BQL
7.44
BQL
0.01
0.14
BQL
0.09
12(f) 8(a,q)*
Nevada Tarrant CA
11/86-1/87 1980
00-2400
24 24
24 6
100.90
2.407
0.149
4.543
0.007
0.077
1.257
0.441
0.067
0.006
0.002
0.786
2.888
5.791
0.093
0.147
8(a,q)*
Five Points CA
1980
24
3
124.37
7.317
0.019
1.786
0.026
0.007
0.037
3.275
1.437
0.055
0.037
0.155
0.105
1.422
0.001
16.657
0.277
0.013
0.032
8(a,q)*
Riverside CA
1980
24
4
106.20
3.549
0.065
5.082
0.173
0.005
0.061
2.015
1.081
0.049
0.013
0.144
0.489
2.373
0.001
7.778
0.182
0.003
0.059
ON
>
to
O
O
Z
O
H
O
C
O
S
o
References are listed
* Values for this size
in Table 1 Appendix. Associated notes are explained in Table 1.
fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM10 COMPOSITION (24-hr AVG)
CENTRAL U.S.
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,q)*
San Jose CA
1980
24
6
66.68
2.053
0.001
0.250
0.771
0.480
0.009
0.071
1.214
0.508
0.027
0.014
0.045
1.119
1.109
5.506
0.086
0.002
0.105
8(a,q)*
Honolulu
1980
24
1
46.90
2.992
0.023
1.981
1.456
0.009
0.025
1.384
0.665
0.034
0.005
0.002
0.093
0.571
6.129
0.130
0.001
0.019
8(a,q)*
Winnemucca
1980
24
5
65.42
6.925
0.010
2.177
0.176
0.006
0.043
1.995
1.200
0.044
0.003
0.063
0.573
12.817
0.173
0.026
8(a,q)*
Portland
1980
24
4
117.55
6.932
0.014
0.121
1.459
0.197
0.019
0.109
2.059
0.805
0.108
0.036
0.028
0.537
2.371
0.001
12.505
0.191
0.018
0.119
8(a,q)*
Seattle
1980
24
1
36.00
2.296
0.008
0.033
0.585
0.228
0.005
0.041
1.001
0.231
0.022
0.007
0.006
0.292
0.952
0.001
4.424
0.091
0.093
8(a,q)* 13(q)*
Albuquerque Denver
12/84-3/85 1/11-30/82
7am-7pm-7am 6am-6pm-6am
12 12
~26
56.46
7.11
2.15
2.22
2.45
3.294
<0.004
0.089
0.127
0.705
0.018
1.287
<0.018
0.018
1.033
0.727
0.031
0.008
0.155
0.424
0.005
0.709
< 0.004
<0.004
7.737
<0.004
0.009
0.09
<0.004
0.085
14(m) 14(aa) 15(s)*
Urban Denver Non-urban Denver Chicago
1 1 /87-1 /88 1 1 /87-1 /88 7/94
9am-4pm-9am 9am-4pm-9am 0800-0800
7&17 7&17 24
~136 ~150 16
28.54
5.39
1.31
5.46
0.269
<0.0043
<0.130
0.011
0.761
<0.041
0.047
<0.0073
0.017
0.432
0.161
0.118
0.013
<0.0041
0.022
<0.0018
0.035
0.032
1.363
<0.059
<0.0017
0.813
<0.070
0.019
<0.013
0.090
to
H
6
O
2
O
H
O
c:
O
H
W
O
HH
H
W
References are listed
* Values for this size
in Table 1 Appendix. Associated notes are explained in Table 1.
fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM10 COMPOSITION (24-hr AVG1
CENTRAL U.S.
Units = ug/m3
o\
>
to
o
o
*
o
H
o
3
o
h-H
H
W
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
16(q)*
Houston
9/10-19/80
12
20
63.40
8.78
1.42
2.22
15.52
1.216
<0.015
0.139
0.091
2.935
<0.012
0.398
0.007
0.046
0.766
0.289
0.035
<1.49
0.008
0.128
0.589
<0.006
4.83
0.006
<0.003
3.200
0.036
< 0.045
0.142
6,7(p,q)
Harriman
5/80-5/81
0000-0000
24
256
32.50
11.14
0.052
1.800
0.050
0.690
0.038
0.001
0.237
2.500
0.002
2.000
ND
17* 6,7(p,q)
Harriman Kingston
9/85-8/86 5/80-6/81
0000-0000
24 24
330 169
30.00 35.40
8.70 13.63
36.1
0.056
0.960
0.018
0.360
0.027
ND
0.234
2.400
0.002
1.900
ND
6,7(P,q)
Portage
3/79-5/81
0000-0000
24
271
18.20
7.29
0.014
0.380
0.083
0.230
0.009
0.001
0.074
1.500
0.001
0.980
ND
6,7(p,q)
Topeka
8/79-5/81
0000-0000
24
286
26.40
6.60
0.055
2.400
0.031
0.580
0.020
0.001
0.203
1.200
0.000
2.500
ND
8(a,q)*
El Paso
1980
24
10
76.21
2.903
0.037
0.103
3.964
0.043
0.004
0.083
0.946
0.623
0.027
0.002
0.672
1.072
0.003
5.813
0.080
0.112
8(a,q)*
Inglenook
1980
24
8
72.45
2.508
0.001
0.061
2.924
0.003
0.006
0.059
1.474
0.717
0.078
0.003
0.030
0.388
2.969
0.001
6.997
0.116
0.188
8(a,q)*
Braidwood
1980
24
1
56.90
2.020
0.002
0.006
1.490
0.002
0.044
0.727
0.355
0.018
0.002
0.014
0.054
2.632
0.002
5.987
0.083
0.023
8(a,q)*
Kansas City KS
1980
24
8
70.33
2.144
0.003
0.036
4.371
0.010
0.048
0.989
0.660
0.026
0.005
0.013
0.237
2.031
0.001
4.976
0.076
0.060
References are listed
Values for this size
in Table 1 Appendix. Associated notes are explained in Table 1.
fraction are calculated from the average measured values reported for the other two size fractions.
-------
PM10 COMPOSITION (24-hr AVG)
CENTRAL U.S.
Units = ug/m3
Ui
to
ON
H
6
o
z
o
H
O
d
o
H
ffl
0
H-t
H
W
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,q)*
Minneapolis
1980
24
6
46.35
2.191
0.005
0.069
1.674
0.293
0.003
0.057
0.831
0.402
0.031
0.002
0.406
1.131
0.001
4.848
0.062
0.072
8(a,q)*
Kansas City MO
1980
24
3
58.43
2.284
0.010
0.093
3.967
0.530
0.006
0.036
1.119
0.503
0.031
0.003
0.478
1.043
4.986
0.074
0.086
8(a,q)*
Akron
1980
24
7
70.90
2.555
0.015
0.064
1.541
0.572
0.024
0.055
2.249
0.592
0.129
0.011
0.059
0.509
3.870
0.008
5.531
0.116
0.219
8(a,q)*
Cincinnati
1980
24
2
62.95
2.972
0.013
0.041
1.374
0.103
0.005
0.038
1.057
0.499
0.032
0.007
0.080
0.442
3.265
0.005
6.961
0.099
0.201
8(a,q)*
Buffalo
1980
24
14
83.32
3,000
0,009
0.015
2.768
0.728
0.017
0.048
2.711
0.516
0.111
0.017
0.060
0.467
4.471
0.005
2.916
0.051
0.001
0.121
8(a,q)*
Dallas
1980
24
4
61.55
1.405
0.039
0.274
4.127
0.029
0.010
0.066
0.968
0.335
0.035
0.004
0.018
1.318
1.754
3.652
0.058
0.002
0.084
8(a,q)*
St. Louis
1980
24
5
56.82
3.956
0.004
0.046
1.874
0.053
0.001
0.032
0.663
0.417
0.019
0.004
0.020
0.372
2.612
0.002
4.638
0.058
0.044
18(x)*
St. Louis
8-9/76
6-12
62.00
1.412
0.003
0.054
0.179
2.949
0.005
0.344
0,015
0.043
1.493
0.653
0.071
0.009
0.099
0.877
0.002
5.188
0.007
0.005
4.928
0.010
0.009
0.587
0.006
0.175
6,7(p,q) 17*
St. Louis St. Louis
9/79-6/81 9/85-8/86
0000-0000
24 24
306 311
31 .40 27.60
11.14 8.00
9.7
0.099
1.600
0.145
0.770
0.040
0.005
0.415
2.300
0.002
2.100
ND
6,7(p,q)
Steubenville
4/79-4/81
0000-0000
24
499
46.50
17.60
0.052
1.120
0.303
2.200
0.068
0.008
0.259
5.500
0.005
2.300
0.013
References are listed
* Values for this size
in Table 1 Appendix. Associated notes are explained in Table 1.
fraction are calculated from the average measured values reported for the other two size fractions.
-------
EASTERN U.S.
>
K)
H
6
o
2
o
H
O
G
O
H
W
O
»
n
H-H
H
W
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
1(0)
Smoky Mtns
9/20-26/78
12
12
5.60
< 0.300
<0.001
0.005
0.322
<0.012
<0.005
0.118
0.108
<0.002
0.014
<0.560
<0.0006
0.580
0.018
<0.004
1(0)
Shenandoah
7/23-5/08/80
12
28
7.40
0.78
0.311
<0.002
0.003
0.304
0.179
0.006
0.158
0.129
<0.006
<0.003
0.009
<0.711
<0.001
0.813
0.017
0.006
2(b)
Camden
7/1 4-8/1 3 '82
6am-6pm-6am
12
50
H.4u
<3.00
0.42
0.57
<0.90
0.550
0.015
0.360
<0.006
0.069
<0.009
0.490
0.151
0.011
0.004
0.054
0.230
0.181
<0.0015
1.610
<0.009
0.002
0.065
0.007
0.030
3(ab) 4(c) 5(d) 5(d)
Philadelphia Deep Creek Raleigh Roanoke
7/25-8/14/94 8/83 1/85-3/85 10/88-2/89
4x daily 7am-7pm-7a 7am-7pm-7am
24 6 12 12
21 98
8.42
0.325
0.003
0.421
0.047
0.014
0.352
0.100
0.104
0.006
0.136
0.002
0.027
0.013
BQL
BQL
0.933
0.030
BQL
0.052
6,7(o,p)*
Watertown
5/79-6/81
0000-0000
24
354
9.30
2.44
0.022
0.209
0.305
0.276
0.006
0.076
0.200
1.000
8(a,o)
Hartford
1980
24
2
27.85
1.875
0.046
0.864
0.302
0.008
0.026
1.070
0.310
0.021
0.005
0.033
0.171
0.428
4.517
0.094
0.008
0.054
8(a,o)
Boston
1980
24
1
105.60
3.458
0.001
0.025
1.069
0.301
0.004
0.023
1.612
0.533
0.029
0.022
0.016
0.177
0.502
6.760
0.154
0.008
0.054
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
OS
N>
oo
H
6
O
O
H
w
O
fo
O
H
W
COARSE COMPOSITION (24-hr AVG)
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,o)
Res.Tr.Pk
1980
24
3
8.17
0.606
0.003
0.086
0.002
0.010
0.182
0.068
0.003
0.013
0.223
1.387
0.021
0.007
9(9)*
Los Angeles
Summer'87
4,5 and 7
1 1 days
26.30
3.34
0.82
5.13
1.87
0.723
BQL
0.055
0.003
0.563
1.026
0.002
BQL
0.737
0.196
0.311
0.017
1.431
BQL
0.127
0.046
0.520
BQL
1.988
BQL
0.072
BQL
0.024
9(9)*
Los Angeles
Fall'87
4 and 6
6 days
8.50
4.89
1.21
4.86
1.01
0.597
0.004
0.084
0.006
0.854
0.426
0.017
BQL
1.635
0.243
0.212
0.021
0.052
BQL
0.053
0.066
0.264
BQL
1.642
BQL
0.106
0.003
BQL
10(i)*
San Joaquin Valley
6' 88-6' 89
24
~35
44.17
5.71
2.38
2.38
0.62
3.418
0.000
0.040
0.006
0.961
0.393
0.007
BQL
1.453
0.632
0.000
0.031
0.000
BQL
0.052
0.032
0.222
0.000
7.577
0.012
0.130
BQL
0.016
WESTERN U.S.
11 (i) 5(d)
Phoenix Boise
10/13/89-1/17/90 12/86-3/87
7am-7pm-7a
6h,2x/day 12
~ 100 days
33.09
4.46
0.84
0.86
0.37
2.539
<0.002
<0.077
0.002
1.929
<0.016
0.194
0.008
0.021
1.259
0.669
0.032
< 0.005
0.003
0.038
0.022
0.003
0.178
<0.030
< 0.002
7.013
< 0.026
0.014
0.121
<0.014
0.034
Units =
ug/m3
12(f) 8(a,o) 8(a,o)
Nevada TarrantCA Five Points CA
11/86-1/87 1980
00-2400
24 24
24 6
43.85
2.230
0.047
4.088
0.005
0.030
0.941
0.255
0.035
0.003
0.002
0.167
0.310
5.208
0.083
0.052
1980
24
3
92.57
7.078
0.004
1.636
0.022
0.006
0.013
3.059
1.193
0.050
0.012
0.148
0.018
0.293
16.001
0.272
0.007
0.016
8(a,o)
Riverside CA
1980
24
4
71.63
3.513
0.028
4.781
0.164
0.005
0.021
1.888
0.961
0.042
0.006
0.144
0.113
0.720
7.544
0.182
0.030
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
VO
VO
(Ji
0\
O
O
1
O
H
O
c
O
H
n
i— i
H
W
COARSE COMPOSITION (24-hr AVG)
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,o)
San Jose CA
1980
24
6
30.40
1.930
0.062
0.682
0.430
0.006
0.028
1.066
0.260
0.021
0.008
0.032
0.228
0.257
5.214
0.086
0.044
8(a,o)
Honolulu
1980
24
1
25.80
1.865
0.006
0.957
0.938
0.005
0.007
0.658
0.294
0.014
0.003
0.022
0.258
3.766
0.067
0.008
8(a,o)
Winnemucca
1980
24
5
55.74
6.564
0.004
1.934
0.176
0.006
0.017
1.764
1.051
0.041
0.002
0.021
0.215
11.903
0.164
0.015
8(8,0)
Portland
1980
24
4
80.38
6.351
0.002
0.028
1.305
0.176
0.010
0.037
1.789
0.587
0.056
0.009
0.011
0.115
0.427
12.128
0.186
0.004
0.038
CENTRAL U.S.
8(a,o)
Seattle
1980
24
1
25.30
2.294
0.002
0.014
0.548
0.228
0.003
0.017
0.903
0.151
0.018
0.001
0.077
0.121
4.332
0.091
0.034
5(d) 13(o)
Albuquerque Denver
12/84-3/85 1/11-30/82
7am-7pm-7am 6am-6pm-6a
12 12
~26
35.73
0.39
2.900
0.058
0.024
0.658
0.012
1.235
<0.009
0.008
0.954
0.648
0.021
0.005
0.113
0.099
0.005
<0.48
7.460
0.009
0.090
0.039
Units = ug/m3
14(m) 14(ab) 15(s)
Urban Denver Non-urban Denver Chicago
1 1 /87-1 /88 1 1 /87-1 /88 7/94
9am-4pm-9am 9am-4pm-9am 0800-0800
7&17 7&17 24
~136 ~150 16
14.97
0.223
< 0.001 3
<0.038
0.007
0.716
<0.012
0.036
< 0.0024
0.006
0.344
0.101
0.106
0.008
<0.0017
<0.017
< 0.0007
0.027
0.005
0.043
<0.017
< 0.0006
0.739
<0.021
0.019
<0.004
0.038
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
COARSE COMPOSITION (24-hr AVG)
CENTRAL U.S.
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
8
Sb
Se
Si
Sn
Sr
Ti
V
Zn
16(o)
Houston
9/10-19/80
12
20
24.80
3.10
1.63
0.91
1.093
<0.006
0.091
0.036
2.780
<0.006
0.366
0.007
0.018
0.604
0.170
0.021
<0.74
0.004
<0.1
0.124
<0.003
<1.29
<0.009
2.990
<0.009
<0.008
0.036
<0.03
0.058
6,7(o,p)* 17
Harriman Harriman
5/80-5/81 9/85-8/86
0000-0000
24 24
256 330
1 1 .70 9.00
3.04
0.014
1.650
0.029
0.570
0.021
0.001
0.057
BQL
1.880
6,7(o,p)*
Kingston
5/80-6/81
0000-0000
24
169
10.80
0.012
0.840
0.018
0.263
0.018
BQL
0.040
BQL
1.700
6,7(o,p)*
Portage
3/79-5/81
0000-0000
24
271
7.20
0.48
0.003
0.335
0.056
0.181
0.006
0.001
0.013
BQL
0.905
6,7(o,p)*
Topeka
8/79-5/81
0000-0000
24
286
13.90
0.55
0.010
2.150
0.000
0.490
0.016
0.001
0.040
BQL
2.310
8(a,o)
El Paso
1980
24
10
49.05
2.748
0.012
0.033
3.632
0.043
0.003
0.047
0.812
0.496
0.023
0.001
0.191
0.249
0.001
5.377
0.077
0.057
8(a,o)
Inglenook
1980
24
8
40.43
2.426
0.021
2.598
0.004
0.027
1.193
0.309
0.041
0.002
0.022
0.079
0.314
6.312
0.116
0.055
8(a,o)
Braidwood
1980
24
1
28.70
1.931
0.002
0.003
1.406
0.002
0.020
0.656
0.303
0.017
0.001
0.014
0.013
0.572
0.001
5.767
0.083
0.012
8(a,o)
Kansas City KS
1980
24
8
41.67
2.284
0.003
0.029
3.754
0.530
0.004
0.015
0.979
0.361
0.025
0.002
0.109
0.280
4.809
0.074
0.040
References are listed in Table 1 Appendix. Associated notes are explained in Table 1.
* Values for this size fraction are calculated from the average measured values reported for the other two size fractions.
-------
COARSE COMPOSITION (24-hr AVG)
CENTRAL U.S.
Os
U)
H
a
o
2
o
o
a
o
H
Q
W
Units = ug/m3
Ref
Site
Dates
Hours
Dur
Number
Mass
OC
EC
Nitrate
Sulfate
Acidity
Al
As
Ba
Br
Ca
Cd
Cl
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
Rb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
8(a,o)
Minneapolis
1980
24
6
30.85
2.191
0.001
0.022
1.571
0.293
0.002
0.022
0.744
0.310
0.026
0.001
0.098
0.224
4.679
0.062
0.027
8(a,o)
Kansas City MO
1980
24
3
41.67
2.284
0.003
0.029
3.754
0.530
0.004
0.015
0.979
0.361
0.025
0.002
0.109
0.280
4.809
0.074
0.040
8(a,o)
Akron
1980
24
7
34.81
2.509
0.003
0.025
1.431
0.572
0.014
0.018
1.640
0.324
0.044
0.005
0.097
0.451
5.009
0.107
0.069
8(a,o)
Cincinnati
1980
24
2
33.15
2.910
0.017
1.312
0.103
0.002
0.014
0.883
0.363
0.021
0.003
0.037
0.099
0.389
6.633
0.096
0.148
8(a,o)
Buffalo
1980
24
14
44.57
2.808
0.012
2.550
0.728
0.015
0.022
2.040
0.206
0.078
0.009
0.108
0.765
2.675
0.051
0.043
8(a,o)
Dallas
1980
24
4
32.63
1.294
0.006
0.051
3.436
0.029
0.005
0.023
0.720
0.210
0.020
0.002
0.252
0.240
3.210
0.051
0.030
8(a,o)
St. Louis
1980
24
5
33.76
3.837
0.001
0.021
1.784
0.053
0.001
0.014
0.587
0.291
0.017
0.002
0.095
0.279
4.468
0.058
0.021
18(k,r)
St. Louis
8-9/76
6-12
28.00
1.209
0.001
0.034
0.047
2.817
0.001
0.257
0.009
0.014
1.218
0.392
0.035
0.005
0.098
0.189
0.002
0.533
0.001
0.001
4.470
0.001
0.007
0.475
0.004
0.074
6,7(o,p)* 17
St. Louis St. Louis
9/79-6/81 9/85-8/86
0000-0000
24 24
306 311
12.40 9.90
3.04
0.021
1.499
0.093
0.580
0.019
0.002
0.088
0.200
1.940
SQL
6,7(o,p)*
Steubenville
4/79-4/81
0000-0000
24
499
16.90
4.80
0.010
1.023
0.211
1.610
0.039
0.004
0.043
0.800
2.010
0.002
References are listed in Table
* Values for this size fraction
1 Appendix. Associated notes are explained in Table 1.
are calculated from the average measured values reported for the other two size fractions.
-------
Table 6A-3
Selected Ratios of Mass Components
FM/CM
FM/PM10
Tot Carbon/FM
NH42SO4/FM
EAST
Mean N
2.59 8
0.65 8
0.25 7
0.47 12
WEST
Mean N
0.89 11
0.41 11
0.54 5
0.15 13
CENTRAL
Mean N
1 .06 25
0.51. 25
0.64 5
0.39 28
N = number of studies contributing to the calculated means.
Tot Carbon = (OCx1.4 + EC).
April 1995
6A-32 DRAFT-DO NOT QUOTE OR CITE
-------
Table 4a. Site-to-Site Variability of PM2.5 Concentrations (ug/m3)
OS
u>
O
O
Z
O
o
c
o
H
W
§
o
HH
H
W
Study Area
No. of Sites
Study Dates
Reference
Fine Mass
OC
EC
Nitrate
Sulfate
Al
Br
Ca
Cl
Cr
Cu
Fe
K
Mn
Ni
Pb
S
Si
Ti
V
Zn
Denver Metropolitan
3,a
11/2/87-1/31/88
14
Mean
19.67
7.25
4.41
3.96
1.55
0.037
0.018
0.058
0.141
0.003
0.017
0.111
0.077
0.012
0.002
0.075
0.642
0.272
0.009
0.031
Spread
2.89
0.79
0.78
0.93
0.16
0.005
0.006
0.001
0.013
0.002
0.008
0.023
0.009
0.003
0.002
0.017
0.077
0.009
0.001
0.008
Phoenix
3,b
10/13/89-1/17/90
11
Mean
29.38
10.09
7.49
3.60
1.33
0.131
0.011
0.167
0.366
0.003
0.015
0.216
0.209
0.023
0.003
0.039
0.436
0.430
0.056
Spread
3.49
2.69
1.71
0.37
0.24
0.015
0.003
0.033
0.356
0.001
0.003
0.035
0.020
0.010
0.001
0.009
0.038
0.066
0.030
Philadelphia
4,c
7/25/94-8/14/94
3
Mean
32.18
4.16
0.69
13.43
0.114
0.009
0.058
0.026
0.007
0.127
0.060
0.003
0.007
0.019
3.251
0.165
0.019
0.041
Spread
2.17
0.94
0.21
0.33
0.009
0.005
0.014
0.007
0.001
0.037
0.008
0.000
0.002
0.010
0.081
0.022
0.003
0.018
San Joaquin Valley
6,d
6/14/88-6/9/89
10
Mean
29.89
4.87
3.24
8.17
3.00
0.152
0.010
0.096
0.094
0.003
0.096
0.180
0.188
0.006
0.016
0.029
1.242
0.460
0.017
0.015
0.078
Spread
10.02
2.70
2.58
2.27
1.33
0.055
0.006
0.050
0.070
0.002
0.036
0.060
0.080
0.003
0.030
0.021
0.565
0.245
0.004
0.028
0.027
Mean = Mean over all sites of the average concentrations determined at each site for the sampling period.
Spread = ABS(Highest Mean Cone. - Lowest Mean Conc.)/2) for all the sites.
a. Federal, Auraria, and Welby sites in urban Denver.
b. Central Phoenix, Scottsdale, and Western Phoenix sites.
c. Broad Street, Castor Avenue, Roxboro, and Northeast Airport sites.
d. Stockton, Crow's Landing, Fresno, Kern, Fellows, and Bakersfield sites.
-------
Table 4b. Site-to-Site Variability of PM10 Concentrations (ug/m3)
Study Area
No. of Sites
Study Dates
Reference
Fine Mass
OC
EC
Nitrate
Sulfate
Al
Br
Ca
Cl
Cr
Cu
Fe
K
Mn
Ni
Pb
S
Si
Ti
V
Zn
San Jose
2,a
12/16/91 -2/24/92
29
Mean
64.95
19.39
9.02
10.90
2.24
0.845
0.012
0.670
0.728
0.003
0.029
0.834
0.823
0.014
0.003
0.035
1.147
2.905
0.088
0.007
0.065
Spread
1.65
0.15
0.42
0.60
0.09
0.035
0.001
0.048
0.032
0.001
0.002
0.027
0.021
0.001
0.000
0.004
0.091
0.045
0.024
0.003
0.005
Phoenix
3,b
10/13/89-1/17/90
11
Mean
62.47
14.55
8.33
4.46
1.70
2.670
0.014
2.096
0.559
0.011
0.036
1.475
0.878
0.054
0.006
0.062
0.615
7.442
0.121
0.090
Spread
7.06
3.48
1.78
0.45
0.29
0.273
0.003
0.317
0.349
0.002
0.009
0.170
0.083
0.014
0.002
0.013
0.041
0.862
0.024
0.034
San Joaquin Valley
6,c
6/14/88-6/9/89
10
Mean
62.92
7.87
3.51
9.44
3.57
2.993
0.012
0.950
0.388
0.009
0.084
1.413
0.720
0.030
0.019
0.039
1.472
7.517
0.128
0.022
0.085
Spread
17.28
4.15
2.76
3.02
1.46
1.570
0.005
0.390
0.225
0.003
0.046
0.445
0.220
0.011
0.032
0.027
0.605
1.765
0.033
0.030
0.029
Mean = Mean over all sites of the average concentrations determined at each site for the sampling period.
Spread = ABS(Highest Mean Cone. - Lowest Mean Conc,)/2) for all the sites.
a. San Carlos St. and Fourth St. sites.
b. Central Phoenix, Soottsdale, and Western Phoenix Sites.
c. Stockton, Crow's Landing, Fresno, Kern, Fellows, and Bakersfield sites.
April 1995
6A-34 DRAFT-DO NOT QUOTE OR CITE
-------
i 7. HUMAN EXPOSURE TO PARTICULATE MATTER
2 AMBIENT AND INDOOR CONCENTRATIONS
3
4
5 7.1 INTRODUCTION
6 The 1982 United States Environmental Protection Agency (U.S. EPA) Air Quality
7 Criteria Document for Particulate Matter and Sulfur Oxides (PM-SOX AQCD) thoroughly
8 reviewed the PM exposure literature through 1981. The later 1986 "Second Addendum to
9 Air Quality Criteria for Particulate Matter and Sulfur Oxides (1982)" added coverage of
10 newly available health effects information with references up to 1986. Consequently,
11 literature directly concerning human exposure to PM has only been previously reviewed
12 thoroughly in an AQCD through 1981 and partially reviewed through 1986.
13 This new analysis first summarizes key points from the exposure section of the 1982
14 PM-SOX AQCD, and then reviews thoroughly the PM exposure literature from 1982 through
15 1993 and includes 1994 literature published and in press through approximately September
16 1994. Some additional literature available in 1994 and 1995 has also been included.
17 The U.S. Environmental Protection Agency regulatory authority for PM only extends to
18 the ambient air, defined in 40CFR as that portion of the atmosphere, external to buildings,
19 available to the general public. One major objective of this chapter is to examine the utility
20 of centralized ambient PM monitoring data as a reasonable surrogate for the average of
21 personal exposures to ambient PM of people in the surrounding community. A secondary
22 objective is to quantify the contribution of ambient air to the personal exposure.
23 By the operative definition of ambient air, air inside a building or on private property is
24 not regulated by the NAAQS. However, it is important to consider total personal exposures
25 to PM both from the regulated ambient air and non-regulated air. This is because a variable
26 fraction of ambient PM penetrates into different non-ambient settings where exposure to PM
27 of ambient origin also takes place and, independently, toxic PM can be generated within a
28 non-ambient setting (e.g. cigarette smoke).
29 Personal exposure to PM is important in itself, because it may give us clues as to which
30 components of PM may be active or inactive biologically. In addition, personal exposure
31 can act as a confounder in epidemiological studies which use an inferred community exposure
April 1995 7-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 to ambient PM as a parameter to correlate with community health parameters. On the other
2 hand, an individual's personal exposure to total PM is a critical parameter for analysis if that
3 person is a member of a cohort whose health outcomes are being tracked individually.
4 Therefore, the chapter examines not only personal exposures to ambient PM, but also
5 community and individual exposures to PM, which include that portion of ambient PM which
6 penetrates into indoor microenvironments (pEs). This is to aid in the interpretation of those
7 acute and chronic epidemiology studies of Chapter 14, in which ambient PM concentrations
8 are assumed to be an indicator or a surrogate for mean community exposure to PM or an
9 individual exposure to ambient PM.
10 This chapter focuses on studies which include information on measurements of
11 simultaneous personal PM exposures, indoor-residential PM concentrations and ambient PM
12 concentrations. Because people spend an average of 21 h per day indoors (Robinson and
13 Nelson, 1995) the indoor environment is the major exposure category. Therefore, studies on
14 concentrations of PM indoors are discussed below in Section 7.6 on Indoor Air. The reason
15 for this separation of indoor concentration from personal exposure is explained below in the
16 general concepts Section 7.1.2.
17
18 7.1.1 Ambient PM Concentration as a Surrogate for PM Dosage
19 The health effects of PM on an individual depend upon the mass and composition of
20 those particles which are deposited within the various regions of the respiratory tract during
21 the time interval of interest. The amount of this potential-dose (FR, Part VI, EPA,
22 Guidelines for Exposure Assessment, May 29, 1992) will be the product of the concentration
23 inhaled (e.g., the instantaneous personal exposure) times the ventilation rate (a function of
24 activity and basal metabolism) times the fractional deposition, which is a function of
25 ventilation rate and mode of breathing (e.g., oral or nasal). If all people had identical
26 ventilation rates and deposition patterns, then the potential-dosage distribution could be
27 linearly scaled to the personal exposure distribution which would serve as a suitable primary
28 surrogate. The usage of ambient PM concentration in health studies as a surrogate for
29 personal PM exposure, and thereby a secondary surrogate for the PM dosage, would be
30 suitable if ambient concentration was also linearly related to the personal exposure (Mage,
31 1983).
April 1995 7-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 The ventilation rate, which is at a minimum during the night while asleep (~ 6 L/m)
2 and at a maximum during the day while awake (~ 12 L/m) is in phase with PM exposure
3 which is also lower at night than during the day (Clayton et al., 1993). Consequently the
4 product of the 24-h average exposure, the 24-h average ventilation rate, and the average
5 deposition parameter for the average ventilation, would seriously underpredict the amount of
6 particulate matter deposited in the respiratory tract (Mage, 1980).
7 In practice, when relating human health to pollution variables, one is forced to use
8 ambient concentration as a surrogate for exposure and dosage because there are typically only
9 fragmentary data on personal exposures in populations. Data are also limited on ventilation
10 rates as a function of basal metabolism and physical activities. Furthermore, there are
11 virtually no applicable data on the deposition rates of the particles which people are inhaling
12 since the size distribution is unknown and deposition is influenced by individual physiological
13 parameters which are unmeasured. According to Hodges and Moore (1977), "even when an
14 explanatory variable (ambient PM concentration) can be measured with negligible error it
15 may often be standing as a proxy for some other variable (dosage) which cannot be measured
16 directly, and so it (dosage) is subject to measurement error". This measurement error can
17 produce a negative bias on the relationship between health effects and PM dosage, which
18 may be a partial explanation of why many previous studies have found a positive but not
19 statistically significant relationship between health effects and ambient PM concentration
20 (Pickles, 1982).
21 In the sections of this Chapter that follow, the relationships between ambient PM
22 concentration, indoor PM concentrations and personal exposures to PM are discussed in
23 detail. The reader should keep in mind the following two caveats while going through this
24 chapter:
25
26 1. Both ambient PM concentration and personal exposure to PM are surrogates for the
27 amount of PM deposited in people's respiratory tracts. Even this quantity is a
28 surrogate for the true (but unknown) species and/or fraction of total PM that is the
29 specific etiological toxic agent(s) that act by an unknown mechanism. This
30 unknown quantity should be the independent variable for developing the underlying
31 relationship of ambient PM and PM exposure to the health indices used as the
32 dependent variables.
33
34 2. Virtually all analyses and discussions presented below are based upon the personal
35 exposure to PM of non-smokers. Only Dockery and Spengler (1981) included 6
April 1995 7.3 DRAFT-DO NOT QUOTE OR CITE
-------
1 smokers out of 37 subjects. Smokers are often excluded from these studies because
2 a personal exposure monitor (PEM) on a smoker will not capture the main-stream
3 tobacco smoke that is directly inhaled. In Section 7.6 on indoor air pollution, it is
4 shown that side-stream environmental tobacco smoke (ETS) is the largest
5 identifiable indoor source of PM. For the average smoker, the amount of direct
6 inhalation (several milligrams of PM per cigarette) can be two-to-three orders of
7 magnitude greater than the microgram amounts of ETS which the PEM captures
8 (Repace and Lowery, 1980). The relationships presented below, of ambient PM to
9 individual PM exposure, only apply to non-smokers.
10
11 7.1.2 General Concepts for Understanding PM Exposure and
12 Microenvironments
13 PM represents a generic class of pollutants which requires a different interpretation of
14 exposure in contrast to that for the other specific criteria gaseous pollutants, such as CO
15 (Mage, 1985). Whereas a molecule of CO emitted from a motor vehicle is indistinguishable
16 from a molecule of CO emitted from a fireplace, a 1-pim aerodynamic diameter (A.D.)
17 particle emitted from the motor vehicle and a l-pim A.D. particle emitted from the fireplace
18 can have different shape, different mass, different chemical composition, and different
19 toxicity. A l-/^m "particle" can be a single entity, or an agglomeration of smaller particles,
20 such as a small Pb particle bound to a larger crustal particle. Furthermore, indoor sources
21 of particles produce a wide variety of particles of varying size and composition that people
22 will be exposed to, as shown in Figure 7-1 (Owen et al., 1992). Given that the health effects
23 of inhalation of any particle can depend upon its mass and chemical composition, it would be
24 of use to measure PM exposure in terms of mass and chemical composition as a function of
25 size distribution (Mage, 1985).
26 The total exposure of an individual to PM during a period of time is composed of
27 exposure to a variety of different particles from a variety of different sources in a variety of
28 different microenvironments (/*E). A /iE was defined by Duan (1982) as "a chunk of air
29 space with homogeneous pollutant concentration"; it has also been defined (Mage, 1985) as a
30 volume in space, during a specific time interval, during which the variance of concentration
31 within the volume is significantly less than the variance between that /xE and its surrounding
32 pEs.
April 1995 7-4 DRAFT-DO NOT QUOTE OR CITE
-------
p
-------
1 For example, a kitchen with a wood stove can constitute a single ^E for total PM,
2 when the stove is off, and all people in the kitchen would have similar PM exposures. When
3 the stove is in operation, the kitchen could have a significant vertical PM concentration
4 gradient and a child on the floor in a far corner and an adult standing at the stove could be
5 exposed to significantly different PM concentrations.
6 In a given /zE, such as one in the kitchen example, the particles may come from a wide
7 variety of sources. PM may be generated from within (e.g. the stove, deep frying, burning
8 toast), from without (ambient PM entering through an open window), from another indoor
9 juE (cigarette smoke from the living room), or from a personal activity that generates a
10 heterogeneous mix of PM (sweeping the kitchen floor and resuspending a mixture of PM
11 from indoor and outdoor sources that had settled out).
12 In general, as a person moves through space and time, they pass through a series of
13 /xEs and their average total exposure (X /iig/m3) PM for the day can be expressed by the
14 following equation,
15
16 X = EXjtj/Etj (7-1)
17
18 where Xj is the total exposure to PM in the ith /xE, visited in sequence by the person for a
19 time interval tj (Mage, 1985). Individual human activity patterns determine the time-
20 sequence in which these /^Es may be visited and, therefore, the magnitude of the overall
21 concentration to which a person is exposed. Let two people on a given day spend 1-h
22 outdoors. If one person is outside from 7 to 8 a.m and the other is outside from 7 to 8 p.m.
23 they can have significantly different PM exposures, neither of which would be characterized
24 by the midnight-to-midnight 24-h average.
25 With appropriate averaging over sets of 4 classes of /xEs (e.g. indoors, ambient-
26 outdoors, occupational, and in-traffic we can simplify the Equation 7-1 as follows (Mage,
27 1985):
28
29 X = (Xin tin + Xout tout + Xocc tocc + Xtra !,„) / T (7-2)
30
April 1995 7-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 where each value of X is the mean value of total PM concentration in the pE class while the
2 subject is in it, time (t) is the total time the subject is in that pE during the day, and T is
3 equal to the sum of all times (usually 1-day). Similar equations may be written for personal
4 exposures to particles from specific sources (e.g. Diesel soot), for specific chemicals (e.g.
5 Pb), or for specific size intervals (PM < 2.5^ AD).
6 In reference to the situation cited above, of two people on the same day spending an
7 identical 1-h as tout at different times during the day, they would have to have two different
8 values of Xout in their exposure calculation. This is in accordance with the precept of Ott
9 (1982) that an air pollutant exposure requires the confluence of two variables - the
10 concentration of PM X(x,y,z,t) at a location (x,y,z) and time t, and the position of an
11 individual's breathing zone at x,y,z,t.
12 In the literature, many excellent studies have reported data on air quality concentrations
13 in /iE settings that do not meet a rigorous definition of an exposure, which requires
14 occupancy by a person (Ott, 1982). Section 7.6 on Indoor Concentrations and Sources, cites
15 Thatcher and Layton (1994) who report that "merely walking into a room increased the
16 particle concentration by 100% (from 10 to 20 jug/m3)", perhaps by air currents reentraining
17 PM. Consequently, a measurement of air quality in a space that includes time when it is
18 unoccupied may not be a valid measure that can be used to estimate an exposure. If this
19 measure includes the periods of time when the space was unoccupied it will tend to be biased
20 low as a measure of the exposure within it during periods of occupancy.
21 In the context of exposure, it may be inappropriate to associate an average exposure to
22 a person while cooking at a stove in a kitchen with a concentration measurement that is
23 influenced by periods when people were not in the kitchen or when the stove was not in
24 operation. It is therefore understood that an average concentration measured in an indoor
25 setting - including periods when people are not present - may not have direct relevence for
26 computing personal exposure because it is not necessarily the concentration during the
27 portion of time that the subject was inside that /nE.
28 The literature on 24-h average PM concentrations in indoor juEs, such as those in
29 residential settings, are treated separately in Section 7.6, as is done for 24-h average ambient
30 PM concentrations in Chapter 6. In the exposure portion of this chapter, specific reference
31 will be made to some of those studies where simultaneous personal exposures and indoor
April 1995 7.7 DRAFT-DO NOT QUOTE OR CITE
-------
1 measurements have been made so that the relationship between indoor concentration and
2 personal exposure can be examined.
3 In practice a cascade sampler can collect ambient PM samples by size fractionation for
4 separate chemical analyses, but such a complete definition of personal exposure to PM by
5 chemistry and size is impossible to obtain. Although some personal monitors can be
6 equipped with a cyclone or impactor separator and two filters to capture two sizes (e.g.,
7 <2.5 />im and 2.5 to 10 /mi), because of the current size of a two filter sampler and the low
8 mass collected in the two fractions, one almost exclusively obtains a single integrated
9 measurement of particle mass collected (e.g., <2.5 /mi or < 10 jum). Consequently, health
10 studies on individuals are usually only able to develop associations between their observed
11 health effects and their observed exposure expressed as an integral mass of PM collected and
12 its average chemical composition.
13 Health studies on populations can make multiple measurements of ambient and indoor
14 PM concentrations simultaneously (e.g., PM2 5, PM10, TSP) along with components of PM,
15 such as polycyclic aromatic hydrocarbons (PAHs), to help understand the size distribution
16 and chemistry of the particles in the ambient and indoor atmospheres. However, these data
17 may be weakly correlated with simultaneous personal PM exposure measurements.
18
19 7.1.3 Review of State-of-knowledge Recorded in the 1982 PM-SOX AQCD
20 In 1982 it was known, from personal monitoring and indoor monitoring, that SO2 is
21 almost always lower indoors than outdoors because of the virtual absence of indoor sources
22 for SO2 and the presence of sinks for SO2 in indoor settings (Exceptions can occur if high
23 sulfur coal or kerosene are used as fuel in a poorly vented stove or space heater). However,
24 this relationship does not hold for PM as the indoor and personal monitoring data show both
25 higher- and lower-than ambient PM concentrations in indoor settings as a function of particle
26 size and human activity patterns.
27 The largest coarse mode particles (>W /mi), which are generally of nonanthropogenic
28 origin (wind blown dust, etc.), require turbulence to provide vertical velocity components
29 greater than their settling velocity to allow them to remain suspended in the air (Figure 7-1).
30 Outdoor particles enter into an indoor setting either by bulk flow, as through an open
31 window, in which all particles can enter at the inlet condition, or by pressure driven drafts
April 1995 7-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 and diffusional flows through cracks and fissures in the barriers of the building envelope
2 when all windows are closed. In the latter mode of entry, velocities are relatively lower,
3 thereby settling out the largest coarse particles in the passage through the barriers.
4 Indoor settings are usually quiescent (Matthews et al., 1989), and the larger ambient
5 particles that do enter indoors quickly settle out by gravity and electrostatic forces, leading to
6 the presence of the familiar dust layers on horizontal surfaces and vertical TV screens that
7 require constant cleaning (Raunemaa et al., 1989). However, human activity in indoor
8 settings, such as smoking, dusting, vacuuming and cooking, does generate fine particles
9 (<2.5 ^m) and coarse particles (>2.5 ^m), and resuspends particles that previously had
10 settled out (Thatcher and Layton, 1994; Litzistorf et al., 1985).
11 There were only three studies of personal PM exposures, compared to ambient PM
12 concentrations, that were referenced in the PM-SOX criteria document. Binder et al. (1976)
13 reported that "outdoor air measurements do not accurately reflect the air pollution load
14 experienced by individuals who live in the area of sampling", in a study in Ansonia, CT,
15 where personal exposures to PM5 were double the outdoor PM concentrations measured as
16 TSP (PM27) (115 versus 58 /xg/m3). Spengler et al. (1980) was cited as reporting that "there
17 was no correlation [R2 = 0.04] between the outdoor level [of respirable particles] and the
18 personal exposure of individuals" in a study in Topeka, KS. Figure 7-2, from Repace et al.
19 (1980), was presented as an example of the variability of PM exposures which show very
20 little influence of ambient concentration. Consequently, at the time of writing the 1982 PM-
21 SOX AQCD, two major factors were known to influence the relationship of ambient to indoor
22 PM air quality. They were (1) the variability of indoor concentrations of PM compared to
23 outdoor concentrations as a function of particle size (e.g., fine indoor > fine outdoor and
24 coarse indoor < coarse outdoor) and (2) the variation of exposures of individuals related to
25 the different activities that are involved with the local generation of particles in their
26 immediate surroundings (smoking, traffic, dusting and vacuuming at home, etc.). This
27 understanding was summarized as follows (pg. 5-136, PM AQCD, 1982):
28
29
30 • long term personal exposures to fine fraction PM (<2.5 /mi) of outdoor origin, may
31 be estimated by ambient measurements of the <2.5 /xm PM fraction.
April 1995 7.9 DRAFT-DO NOT QUOTE OR CITE
-------
Tl
H
6
o
z
o
H
O
c!
O
H
W
n
HH
H
W
280
260
240
220
oo
E 200
§
O
160
140
120
100
80
60
40
20
1 I I I I I I I I
• Indoors
• In Transit
O Outdoors
1 I T
V I I I I
i i
Well-Ventilated Kitchen
Cafeteria, Smoking Section'
Behind Smoky Diesel Truck I
Office
Commuting • ^ ^
im
Outside Cigar
Smoker's Office
Cafeteria,
-Nonsmoking
Section
IO..K,,^B Sidewalk
iSuburbs BusExhaust
Vehicle v^
I-In City
Office"
I I I
Bedroom..
Street Suburbs, OutdoorCN' / city Outdoor
Library Unoccupied Cafeteria | y> | | |
Commuting Room
Suburbs
ogging
LyingjRoOjm | |
12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Midnight A.M. Noon P.M.
Time of Day
Figure 7-2. An example of personal exposure to respirable particles.
2 Source: Repace et al. (1980)
-------
1 • Personal activities and indoor concentrations cause personal exposures to PM to vary
2 substantially. Ambient measurements appear to be a poor predictor of personal
3 exposure to PM.
4
5 • Tobacco smoke is an important contributor to indoor concentrations and personal
6 exposures where smoking takes place.
7
8
9 7.2 DIRECT METHODS OF MEASUREMENT OF HUMAN
10 EXPOSURE TO PM BY PERSONAL MONITORING
11 Human exposure to air pollution can be measured by placing a personal exposure
12 monitor (PEM) close to the breathing zone of an individual. However, the very act of
13 studying the subjects can influence the measured value of their exposure and create an
14 erroneous reading. These influences arise because the subjects become conscious of the
15 study objectives from the indoctrination required to obtain their written informed consent to
16 participate, and the presence of the PEM on their body is a constant reminder.
17 The physical location of the monitor inlet, as worn by the subject, can also influence
18 the subject's PM exposure and the recorded PM. The movements of the subject's body and
19 sampling flow rate can alter the air currents in the subjects breathing zone. "The presence of
20 the body and its movement affect what a personal sampler collects" (Ogden, 1993). When in
21 close proximity to a source actively emitting PM (within a meter) a small change in PEM
22 position (e.g. from left side to right side) can vary the PM measurement.
23 These unquantifiable 'errors' in a PM PEM measurement may be greater than the filter
24 weighing errors and flow rate measurement errors that can be quality controlled through
25 calibration procedures. This may be important for interpretation of the PM PEM data in the
26 literature because the expectation is that these errors inflate the variance of the
27 measurements. In the following section, the individual error components that arise from the
28 measurement process are discussed.
29
30 7.2.1 Personal Monitoring Artifacts
31 7.2.1.1 'The Hawthorne Effect'
32 If subjects carry a personal exposure monitor (PEM) they may change their behavior,
33 subconsciously or consciously, which is known as "The Hawthorne Effect" (Last, 1988).
April 1995 7-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 The name derives from early studies of worker productivity at Western Electric's Hawthorne
2 plant in Cicero, IL (Mayo, 1960). For instance a subject may choose not to go to the opera
3 because the pump noise of the monitor would be disturbing to others. Even though the
4 exposure measured in the alternative activity may be quite accurate, it would be an artifact in
5 the context of exposure assessment, because the activity not performed (opera attendance)
6 would have been the correct one to monitor.
7
8 7.2.1.2 The Monitor Effect
9 The filtration of the breathing zone air by a PEM can reduce the PM concentration in
10 the breathing zone by "self dilution" (Cohen et al., 1984). The placement of the personal
11 monitor in the breathing zone, as well as its flow rate, can alter the air flow stream lines in
12 the area of the nose or mouth that would exist in its absence. There may be an electrostatic
13 charge on the plastic cassette filter holder which can possibly affect the collection of charged
14 particles (Cohen et al., 1982).
15
16 7.2.1.3 Subject Effect
17 A subject may contaminate the personal monitoring data by an inadvertent action, such
18 as forgetting to put on the PEM upon awakening in the morning, or purposefully choosing
19 either not to wear it when going to the opera (anti-'Hawthorne Effect') or placing it close to
20 a source. If such actions are not recorded in a diary or reported to the investigator during a
21 verbal debriefing, the exposure data, although valid per se as an actual concentration
22 measurement, could be treated as a valid exposure of a subject in his/her daily life and
23 related to an incorrect classification.
24 For example, Sexton et al (1984) reported that one male subject with a personal exposure
25 mean of 77 /xg/m3 had a spouse with a simultaneous personal exposure of 37 pig/m3, with no
26 recorded activity that could explain the higher exposure values.
27
28 7.2.1.4 Non-Response Error
29 In performance of a personal monitoring study, people often refuse to participate. The
30 refusal rate increases with the burden on the respondents due to the time required to
31 complete questionnaires, diaries and the need to carry the personal monitor with them
April 1995 7-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 throughout the study. If the cohort of people who refuse to participate have significantly
2 different personal PM exposures than the participants, then the study will produce a biased
3 estimate of the exposures of the total population.
4
5 7.2.2 Characterization of PM Collected by Personal Monitors
6 The amount of PM collected by different types of personal monitors with the identical
7 nominal cut-point can be variable. Small differences in the cut point (AD at 50% efficiency)
8 and in other portions of the penetration curve can exist under calm wind conditions when the
9 monitor is stationary, and these differences can also be a function of the air velocity relative
10 to the monitor. Consequently, the difference between two PM measurements made by two
11 nominally identical monitors of different design, can be a function of the wind speed and the
12 size distribution of the PM in the air mass being sampled. Given the knowledge of the
13 sampled size distribution (as collected) and the complete penetration curve, the concentration
14 in the free atmosphere can be predicted (Mage, 1985). For example, if a sampler collects 1
15 Mg/m3 °f PM in a size interval that has an overall efficiency of 0, then we can estimate that
16 the atmosphere contained 1/0 /xg/m3 of PM of the same size range. A recent field
17 comparison of different types of respirable dust samplers used in occupational settings where
18 coarse mode paniculate is predominating, by Groves et al. (1994), shows that there is a
19 considerable difference between the mass collected by sets of paired cyclones sampling in a
20 concentration range of 0.5 to 6.6 mg/m3, which is much higher than normally seen in non-
21 occupational personal exposure studies. This type of comparison study has not been done for
22 personal monitors used in nonoccupational studies at the ambient and indoor respirable
23 concentrations on the order of 0.01 to 0.1 mg/m3 where the fine mode is more important.
24
25 7.2.3 Microscale Variation and the Personal Cloud Effect
26 The tendency for human activity in the home or at work to generate a 'personal activity
27 cloud' of particles from clothing and other items (carpet, stuffed furniture, etc.), that will be
28 intense in the breathing zone, and diluted near an area monitor located several meters away,
29 has also been cited as a contributing factor to the discrepancy between personal measures of
30 exposure and time weighted average (TWA) exposures using microenvironmental
31 measurements (Martinelli et al., 1983; Cohen et al., 1984; Rodes et al., 1991). The
April 1995 7_13 DRAFT-DO NOT QUOTE OR CITE
-------
1 difference between sampling for PM at the nose, on the lapel or at a point several meters
2 away from a person can be appreciable. The microscale variation in PM between a personal
3 monitor sampling at the nose, versus an identical monitor sampling at the lapel, has been
4 investigated by Cohen et al. (1982, 1984) who found no net bias resulted from sampling at
5 either location in a concentration field of uniformly dispersed aerosol. However, they noted
6 that spatial concentration variability and resuspended dust from clothing, and to a lesser
7 extent electrostatic charges on plastic filter cassette holders (and plastic eyeglass frames),
8 could lead to different exposure measurements in the facial region, with three times as much
9 mass collected by a filter attached to the clothing as from the air sampled directly in the
10 breathing zone.
11 Fletcher and Johnson (1988) also measured metal concentrations (measurement method
12 and size unspecified) in an occupational exposure situation (metal spraying of spindles on a
13 lathe) and found 50% higher concentrations measured from the left lapel compared to the
14 right lapel, which reflected the orientation of the operator to the lathe. When a neutrally
15 buoyant tracer gas mixture was released 0.5 m from an operator in a different work setting
16 there was no variation between the left and right lapels, and the nose, as found by Cohen et
17 al. (1982). However, when the experiment was performed with a heavier-than-air tracer gas
18 mixture, the nose measurement was approximately 25% less than the lapel concentrations.
19 This implies that if submicron particles, which behave like a gas, are emitted close to the
20 subject in a buoyant plume, there may also be significant microscale variations in the
21 breathing zone.
22 Parker et al. (1990) measured the aerosol distribution in a small test room resulting
23 from a nozzle-jet injection, using a "heated phantom" (mannequin), and found larger
24 discrepancies between chest mounted monitors and area monitors up to three meters away,
25 by up to a factor of ten difference.
26
27
28 7.3 NEW LITERATURE ON PARTICLE EXPOSURES SINCE 1981
29 The following sections review studies that measured PEM PM in the general non-
30 smoking population. In these studies, the subjects spent time at home and in other indoor
31 environments that include time at work. In the USA, recent data indicate that on a daily
April 1995 7-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 basis, an average US resident spends approximately 21 h indoors (87.2%), 100 minutes in
2 (or near) a vehicle (7.2%), and 80 minutes outdoors (5.6%) [Robinson and Nelson, 1995].
3 The concentration of PM in residential and commercial indoor settings is thoroughly
4 reviewed and presented in Section 7.6. However, the concentrations of PM and personal
5 exposures to PM in 'indoor' industrial workplaces (e.g. coal mines) and the "dusty trades"
6 (e.g. carpentry or machinist), which are covered by occupational standards for air quality,
7 are not covered in this document.
8 Almost all the studies of PM exposure in the general public have been conducted on
9 urban and suburban residents. These subjects are often working in occupations that do not
10 require PM monitoring to assure that occupational standards are being met (e.g. in an
11 office). However, PM monitoring in an industrial workplace by a subject - independently of
12 an official corporate industrial hygiene program - can have legal or security implications for
13 an employer. For example, in a study in Tennessee (Spengler et al., 1985) some potential
14 subjects were unable to participate because their employer (Oak Ridge National Laboratory)
15 would not allow them to wear PM monitors at work. Such exclusion of subjects from
16 exposure studies can negatively bias measured exposure distributions if the reason for the
17 exclusion is related to their potential for high PM exposure.
18 A further complication arises from the fact that industrial exposures tend to be
19 dominated by a specific type of particle. Coal miners are exposed to coal dust, textile
20 workers are exposed to cotton dust and welders are exposed to metal fumes. An additional
21 chapter on personal exposures to industrial PM would be needed to describe the various
22 industries and trades that have their own individual PM problems, and the usage pattern and
23 efficiencies of respirators and masks that are required to be worn. Therefore, occupational
24 PM studies are not presented, and only some selected studies are cited to illustrate a
25 particular point that is applicable to exposure studies in general, such as microscale variation
26 in PM when close to a source of PM.
27 It may be useful to keep in mind that the baseline exposure of nonsmoking workers in
28 the "dusty trades" or industrial workplaces may be similar to that of other nonsmoking
29 people in their communities. Their total working day exposure would then be approximated
30 by that baseline exposure plus their incremental workplace exposure for 8 h, in a similar
April 1995 7-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 manner to a smoking increment for smokers. On a day off from work, these worker's
2 exposures may be similar to that of the general nonsmoking public.
3
4 7.3.1 Review of the Literature
5 7.3.1.1 Results of U.S. Studies
6 Dockery and Spengler (1981) compared personal PM3 5 exposures and ambient PM3 5
7 concentrations in Watertown, MA, and in Steubenville, OH. In Watertown, 24-h personal
8 samples were collected on a 1-in 6-day schedule, and in Steubenville, 12-h personal samples
9 (8 a.m. to 8 p.m.) were collected on a Monday-Wednesday-Friday schedule. A correlation
10 coefficient of 0.692 between the mean personal and the mean ambient concentration for 37
11 subjects, 18 in Watertown and 19 in Steubenville, was reported for the pooled data.
12 However, this appears to be an artifact of two separate clusters formed by these data, each
13 with considerably lower correlation. When these data are analyzed separately, the regression
14 coefficient between personal and ambient for Watertown is R = 0.01 and for Steubenville it
15 is R = 0.43.
16 Sexton, Spengler and Treitman (1984) studied personal exposures to respirable particles
17 (PM3 5) for 48 nonsmokers during a winter period in Vermont, where firewood was either
18 the primary or secondary heating source for the subject. Their results showed that personal
19 exposures were 45% higher than indoor averages (36 /xg/m3 versus 25 ^g/m3) and indoor
20 averages were 45% higher than outdoor averages (25 /xg/m3 versus 17 /xg/m3). This
21 relationship is consistent with those reported in the 1981 PM-SOX AQCD (Spengler et al.,
22 1980). Ambient air pollution, measured by an identical stationary ambient monitor (SAM)
23 outside each residence (a pump contained in a heated box was connected to an external
24 cyclone and filter), had no correlation with the resident's personal exposures (R2 = 0.00)
25 and 95% of the subjects had personal exposures greater than the median outdoor
26 concentration. This would not contradict the first conclusion cited from the 1982 PM-SOX
27 AQCD if there were sources of PM3 5 indoors.
28 Spengler et al. (1985) reported a study of PM3 5 exposures in the non-industrial cities of
29 Kingston and Harriman, Tennessee, during the winter months of February through March,
30 1981. A large TV A coal burning power plant (Watts Bar) with very tall stacks in the
31 immediate area was not a local source of paniculate pollution. In this study, two
April 1995 7-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 Harvard/EPRI PM3 5 monitors were used for each person. One stationary indoor monitor
2 (SIM) remained indoors in the home, in the open air of a first floor activity room, and the
3 second monitor (PEM) was carried for 24-h to obtain the personal exposure. In each
4 community, identical Harvard/EPRI samplers (SAM) were placed at a central site to
5 represent ambient PM3 5 concentrations. The results of the study are shown in Table 7-1. In
6 both communities, 95% of the subjects had personal exposures to PM3 5 greater than the
7 average ambient concentrations. The mean personal exposure and indoor concentrations
8 (44 ± 3 /ig/m3 and 42 + 3 jug/m3) were more than 100% greater than the mean ambient
9 average of 18 +2 /xg/m3 sampled on the same days.
10
11
TABLE 7-1. QUANTILE DESCRIPTION OF PERSONAL, INDOOR,
AND OUTDOOR PM3 5 CONCENTRATIONS (in jtg/m3), BY LOCATION IN
TWO TENNESSEE COMMUNITIES
City
Kingston
Harriman
Total K&Ha
Group
Personal
Indoor
Outdoor
Personal
Indoor
Outdoor
Personal
Indoor
Outdoor
N
133
138
40
93
106
21
249
266
71
95%
99
110
28
122
129
34
113
119
33
75%
47
47
22
54
45
23
48
46
23
50%
34
31
16
35
27
15
34
29
17
25%
26
20
12
24
18
13
26
20
13
5%
19
10
6
15
10
9
17
10
7
Mean
42
42
17
47
42
18
44
42
18
S.E.
2.5
3.5
2.7
4.8
4.1
4.0
2.8
2.6
2.1
"Includes samples from 13 subjects living outside Kingston and Harriman town limits and from four field
personnel residing in these communities.
N = number of samples.
S.E. = Standard error.
Source: Spengler et al. (1985).
1 For the complete cohort, the correlation between PM PEM and PM SAM was r = 0.07
2 (p = 0.30), and between PM PEM and PM SIM was r = 0.70 (p = 0.0001). The
3 correlation between simultaneous PM PEM and PM SAM was r = 0.15 for 162 nonsmoke
April 1995 7_17 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposed individual observations (p = 0.06). For 63 observations on smoke exposed
2 individuals, the correlation r = 0.16 was not significant (p = 0.16).
3 An important finding was that in nonsmoking households, the PM PEM is always
4 higher than SIM and SAM. "This implies that individuals encounter elevated concentrations
5 away from home, and/or that home concentrations are elevated while they are at home and
6 reduced while they are away". This observation is supported by the findings of Thatcher and
7 Layton (1994), reported in Section 7.6 - merely walking into a room can raise the
8 concentrations of PM by 100% (from 10 to 20 /xg/m3).
9 This study is relevant to the analyses by Dockery et al. (1992) of PM mortality in St.
10 Louis, MO, and in Eastern Tennessee counties surrounding the cities of Kingston and
11 Harriman, which are discussed in Chapter 12 on epidemiology. Although the Spengler et al.
12 (1985) and Dockery et al. (1992) studies are not directly comparable, because different years
13 of data were used (1981 versus 1985/1986), this might call into question the meaning of the
14 correlation of ambient PM and mortality, if individual PEM data are uncorrelated with the
15 SAM data. One possible explanation may be that even though individual PEM of non-
16 smokers as a group are not well correlated with SAM data (r = 0.15, p = 0.06) the mean
17 PEM of the nonsmoking population may be better correlated with the mean SAM on any
18 given day as discussed in Section 7.5.2 (Mage and Buckley, 1995).
19 Morandi et al. (1988) investigated the relationship between personal exposures to PM
20 and indoor and outdoor PM concentrations, using a TSI Model 3500 piezobalance that
21 measures respirable particles in the range <3.5 /xm. For the group of 30 asthmatics in
22 Houston, TX, that were studied, outdoor concentrations averaged 22 /xg/m3, indoor
23 concentrations averaged 22% higher than outdoor (27 /xg/m3) and, in motor vehicles, the
24 average concentration of particles was 60% higher than the average outdoors (35/xg/m3). As
25 for correlations between the various measurement categories, personal 12-h (7 a.m. to
26 7 p.m.) daytime exposures to PM were not predicted as well by fixed site dichotomous
27 sampler ambient monitors (R2 = 0.34) as by the indoor exposures (R2 = 0.57). However,
28 for 1-h exposures, they found no correlation (R2 = 0.00) between the personal exposures to
29 PM5 and the indoor exposures measured with a TSI model 5000 stationary continuous
30 piezobalance located in the 'den' area of the home. The authors noted that use of home air
31 conditioning and recirculation tended to increase the PM exposures, and that misclassification
April 1995 7-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 of human activities and microenvironments likely explained the inability to predict personal
2 exposures from microenvironmental data.
3 Lioy et al. (1990) reported a study done during the winter (January 1988) in the
4 industrial community of Phillipsburg, NJ, where personal PM10 was monitored along with
5 indoor and outdoor PM10. They collected PM10 (fine plus coarse particles on a single filter).
6 In this study of eight residences of 14 nonsmoking nonsmoke-exposed individuals, the
7 geometric mean 24-h concentrations were 68, 48 and 42 jug/m3 for personal, outdoor and
8 indoor sites, respectively. The arithmetic mean personal PM exposure of 86 jwg/m3 was 45%
9 higher than the mean ambient concentration of 60 /xg/m3. The higher ambient than indoor
10 concentrations in this study, a reversal of the relationships found in the Sexton et al. (1984),
11 Spengler et al. (1985) and Morandi et al. (1988) studies, may be caused by the local
12 industrial source of coarse particles in that community and the absence of cigarette smokers
13 in the residences sampled. This difference also may be partially explained by the 10 /xm
14 particle sizes in the NJ study and the 3.5 /xm particle sizes in the other studies, which would
15 suggest that the NJ homes had less influence from the locally generated coarse particles that
16 tend to settle out in the home. The regression coefficient between personal and ambient
17 PM10 for all 14 people on the 14 days of the study (n = 191 valid personal values) was 0.19
18 (R2 = 0.037, p = 0.008). With three personal exposure extreme values removed (n = 188
19 personal values), the coefficient was 0.50 (R2 = 0.25, p = 0.007).
20 In all five studies, the personal PM was measured to be higher than either the indoor or
21 the outdoor PM measurements. This relationship has also been found in the two PTEAM
22 studies (Perritt et al., 1991; Clayton et al., 1993) described in detail later in Section 7.3.3.
23 For these PTEAM studies during the day (7 a.m. to 7 p.m.) average personal PM10 exposure
24 data (150 pig/m3) were 57% higher than the average indoor and outdoor concentrations,
25 which were virtually equal (95 /ig/m3). At night (7 p.m. to 7 a.m.) average PM10 personal
26 exposures (77 /xg/m3) were higher than the average indoor concentrations (63 /ig/m3) but
27 lower than the average outdoor concentration (86 ptg/m3). Consequently, considering that the
28 PTEAM subjects were overwhelmingly mostly indoors at night, a time-weighted-average
29 (TWA) of the indoor and outdoor PM concentrations appears to always underestimate the
30 personal exposures to PM.
April 1995 7-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 It has been proposed (WHO, 1982a; Spengler et al., 1985; Mage, 1985) that such a
2 discrepancy between the TWA and the personal monitoring measurements may be caused by
3 two factors described as follows: (1) Human exposure to PM at work and in traffic are only
4 partially accounted for in a TWA of indoor and outdoor ambient PM values; and (2) Indoor
5 and outdoor averages reflect periods of low concentration during which the subject is not
6 present.
7 With regard to the first factor, the PM of occupational exposures, and exposures in
8 traffic that also reflect the vehicular emissions of PM plus the resuspension of roadway dust
9 from the turbulence of the vehicles, are not well represented by ambient PM measurements.
10 Ambient PM monitors are usually sited several meters above ground level, at a location
11 uninfluenced by a single local source, so that the data recorded can represent an average
12 community type concentration (Mage, 1983). The complexity of the situation is exemplified
13 by the PTEAM study reported by Clayton et al. (1993). People who were employed had
14 lower daytime exposures to PM than nonemployed people (n = 59, 127 pig/m3 versus
15 n = 111, 162 jiig/m3), and people who had time in traffic had lower PM exposures than
16 those who stayed at home (n = 31, 97/ig/m3 versus n = 121, 163 /*g/m3). A possible
17 explanation may be that people who stay home generate PM by doing housework and people
18 who go off to work may be driving to a location where the indoor PM and ambient PM are
19 lower than at their home.
20 With regard to the second factor, the PM pollution generating activities in a home
21 usually occur only when a person is at home, as discussed in Section 7.1.2. Therefore, the
22 PM in a home will be higher when a person is present than when the home is unoccupied. A
23 24-h average of the indoor concentration thereby underestimates the average exposure of a
24 person while in that home.
25 Ambient PM is also higher during the day (when industry and traffic are active, and
26 wind speeds are high) than at night when PM generating activities are at a minimum and the
27 air is still. Consequently, a 24-h average ambient PM value generally underpredicts the
28 concentrations during the daylight hours and the exposures of people going outdoors during
29 that period.
30 Therefore, a 24-h TWA personal exposure will always tend to be underpredicted by a
31 simple TWA of 24-h residential-indoor and ambient-outdoor PM concentration data that fails
April 1995 7-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 to account for occupational exposures, in-traffic exposures, and the intensity of personal
2 exposures during human activity that cannot be recorded by area monitors several meters
3 away from the subject.
4
5 7.3.2 Personal Exposures in International Studies
6 The Global Environment Monitoring System (GEMS) of the World Health
7 Organization/United Nations Environment Programme (WHO/UNEP) commissioned a series
8 of four pilot studies of personal exposure to PM in Zagreb (WHO, 1982a), Toronto (WHO,
9 1982b), Bombay (WHO, 1984) and Beijing (WHO, 1985). In these studies, people who
10 worked in the participating scientific institutes were recruited to carry a PM sampler and
11 their exposures were matched to the ambient concentrations measured outside their home or
12 in their communities. The results of these studies of PM, expressed as mean personal
13 exposure (PEM) and mean ambient (SAM), and the regression R2 between them are
14 presented in Table 7-2.
15 In addition to the institute personnel, the Toronto study also measured exposures of
16 asthmatics to PM25 and obtained R2 values of 0.07 and 0.00 in the summer and winter
17 respectively. The net result of these international studies is that they confirm the lack of a
18 consistent relationship between individual personal PM exposures and ambient concentrations
19 as found in the U.S. studies. The results of the Beijing study are of importance because of
20 (a) the recent paper by Xu et al. (1994), which reports a correlation between mortality in
21 Beijing and ambient TSP and SO2 and (b) the very low correlations between personal
22 exposure to PM and ambient PM concentrations (not statistically different from zero) found
23 in the WHO study (R2 = 0.03 and 0.07). The major fuel in Beijing is coal, and in the
24 winter the ambient PM27 (TSP) averages up to 500 )ug/m3 (WHO/UNEP, 1992). During the
25 GEMS PM exposure study of 1985, the 24-h ambient concentrations of PM3 5 measured
26 outside the subjects homes averaged 420 jug/m3, the indoor concentrations in the homes
27 averaged 364 /xg/m3, and the personal exposures averaged 191 /ug/m3. The subjects who
28 were workers at the institute conducting the study spent little time outdoors and their days
29 were spent at the institute. The lower relative values of their exposure appear to be caused
30 by lower values at home during the evenings. Meals are usually prepared during the day
31 when the workers are not there and the indoor exposures during cooking can be quite high.
April 1995 7_2i DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-2. SUMMARY OF WHO/UNEP GEMS/HEAL PM, PERSONAL
EXPOSURE PILOT STUDY RESULTS
Location
Season PM /mi n
Toronto 25 13
winter
summer
Zagreb 5 12
summer
winter
Bombay 3.5 15
winter
summer
monsoon
Beijing 3.5 20
winter
summer
PEM Mean
m
72
78
12
12
105
102
101
71
40
Time
8-h
1-wk
24-h
24-h
1-wk
+ SE
122+9
124+4
114+?
187+?
127+6
67+3
58±3
177+?
66±?
SAM Mean R2 PEM
+ SE
68+9
78+4
55+?
193±?
117±5
65 ±3
51+2
421 ±?
192+?
vs SAM
0.15
0.10
0.00
0.50
0.26
0.20
0.02
0.07
0.03
P
NS
NS
NS
NR
NR
NR
NS
NS
NS
n = number of subjects carrying PEM.
m = total number of observations.
NR = Not Reported, but listed as significant.
NS = Not significantly different from 0.
? = Not reported.
*25 pm A.D. computed from flow rate and open filter design.
1 Smith et al. (1994) reported the cooking exposures to PM10 listed in Table 7-3 for Beijing,
2 Bangkok and Pune. In Beijing, personal exposures of the cooks during cooking were 4 to 20
3 times higher than the 24-h outdoor values on the days that cooking took place. The presence
4 of high levels of coarse particles in the ambient air, which do not readily penetrate into the
5 institute and the homes, may contribute to the significantly higher ambient values that are
6 uncorrelated with the personal exposures.
7
8 7.3.3 The Particle TEAM (PTEAM) Study
9 In 1986, the U. S. Congress mandated that EPA's Office of Research and Development
10 "carry out a TEAM Study of human exposure to particles." EPA's Atmospheric Research
11 and Exposure Assessment Laboratory (AREAL) joined with California's Air Resources
12 Board (CARB) to sponsor a study in the Los Angeles Basin. The study was carried out
13 primarily by the Research Triangle Institute and the Harvard School of Public Health, with
April 1995
7-22
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-3. SUMMARY OF DAILY INDOOR, OUTDOOR, AND PERSONAL
EXPOSURES TO PM10 DURING COOKING AS A FUNCTION OF
FUEL TYPE IN THREE CITIES IN ASIA
City
Beijing, PRC
Pune, India
Bangkok Thailand
Cooking Fuel
Coal (vented)
Gas
LPG
Coal Gas
Natural Gas
Agric. Residue
Wood
Biomass
Kerosene
LPG
Charcoal
LPG near traffic
LPG far from traffic
Indoor
(/ig/m3)
550
400
370
420
410
2800
2000
2100
480
250
330
390
300
Outdoor
(/ig/m3)
550
430
410
440
440
2600
920
1000
340
250
330
450
285
Personal
Cooking
Only (^g/m3)
1900
5000
3300
9100
1600
900
1100
1100
530
420
550
850
3900
Source: Smith et al. (1994).
1 additional support from Lawrence Berkeley Laboratory, Acurex, and AREAL. The main
2 goal of the study was to estimate the frequency distribution of exposures to particles for
3 nonsmoking Riverside residents. Another goal was to determine particle concentrations in
4 the participants' homes and immediately outside the homes.
5 7.3.3.1 Pilot Study
6 7.3.3.1.1 Study design
1 A pilot study was undertaken in nine homes in Azusa, CA in March of 1989 to test the
8 sampling equipment (Ozkaynak, et al., 1990). Newly-designed personal exposure monitors
9 (PEMs) were equipped with inhalable (PM10) and fine (PM2 5) particle inlets. The PEMs
April 1995
7-23
DRAFT-DO NOT QUOTE OR CITE
-------
1 were impactors with 4-Lpm Casella pumps (Wiener, 1988). Two persons in each household
2 wore the PEMs for two consecutive 12-h periods (night and day). Each day they alternated
3 inlet nozzles. The first five households were monitored concurrently for seven consecutive
4 days (March 6 to 13, 1989); the last four households were then monitored concurrently for
5 four consecutive days (March 16 to 20, 1989). This resulted in approximately 100 PEM
6 samples for each size fraction.
7 Indoor and outdoor particle concentrations were monitored using microenvironmental
8 exposure monitors (MEMs). These monitors were the Harvard "black boxes" (Wiener,
9 1989) employing a 10 Lpm pump. Several indoor MEMs were placed in different rooms in
10 each house to determine the magnitude of room-to-room variation. These monitors were
11 capable of monitoring both fine and inhalable particles simultaneously.
12 A central site with a PEM, MEM, and two EPA reference methods (dichotomous
13 samplers and high-volume samplers with a 10 /xm size-selective inlet) was also operated
14 throughout the 11 days (22 12-h periods) of the study.
15 7.3.3.1.2 Results
16 Side-by-side comparisons indicated good agreement of all four monitors (Table 7-4).
17 Good agreement was also noted between outdoor concentrations at the homes and at the
18 central site (Wiener et al., 1990). Room-to-room variation of particle levels was generally
19 less than 10%. Therefore the several indoor MEM values in a particular house were
20 averaged to provide a single mean indoor value to compare to the corresponding outdoor
21 value (Table 7-5). It was decided that this finding would justify using only one indoor
22 monitor in the subsequent full-scale study (Clayton et al., 1991).
23 The personal exposures were about twice as great as the indoor or outdoor
24 concentrations for both PM10 (Table 7-6a) and PM2 5 (Table 7-6b). Considerable effort was
25 expended to determine whether this was a sampling artifact, due for example to the constant
26 motion of the sampler; however, no evidence could be found for an artifactual effect.
27 Nonetheless, to reduce chances for an artifactual finding in the main study, it was decided to
28 use identical PEMs for both the personal and fixed (indoor-outdoor) samples in the main
29 study.
April 1995 7-24 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-4. PTEAM PREPILOT STUDY: MEAN OUTDOOR
PARTICLE CONCENTRATIONS (/tg/m3)
Central Site PM-10
Mean
SD
Mean
SD
MEM
56.9
28.8
MEM
39.4
23.3
PEM
59.1
25.2
Central
PEM
46.0
23.7
DICOT HIVOL
61.7 56.6
27.6 31.5
Site PM-2.5
DICOT
41.8
20.6
Mean
58.4
28.8
SD
5.7
3.3
Residential
MEM
61.4
26.0
SD
7.4
4.4
Residential
Mean
41.4
22.9
SD
6.2
4.6
MEM
41.7
21.6
SD
7.1
5.8
Each sampler collected 22 12-h samples over 11 days.
MEM: Microenvironmental monitor: 10 Lpm impactor.
PEM: Personal exposure monitor: 4 Lpm impactor.
DICOT: Dichotomous sampler: 16.67 Lpm virtual impactor.
HIVOL: High-volume (1130 Lpm) impactor.
TABLE 7-5. PTEAM PREPILOT STUDY: TWENTY FOUR-HOUR (24-h)
PARTICLE CONCENTRATIONS Gig/m3)
Mean
SD
SE
Indoor
58.7
24.6
3.4
PM-10
Outdoor
62.6
24.9
3.5
Indoor
36.3
18.6
2.6
PM-2.5
Outdoor
42.6
21.6
3.0
1
2
3
4
5
6
7
Regressions of outdoor on indoor concentrations showed low R2 values (1 to 30%) for
both PM10 and PM2 5 size fractions, as did regressions of daytime indoor on personal
concentrations (R2 = 0 to 18%). Overnight indoor concentrations had somewhat better
ability to explain personal exposures (R2 = 14 to 58%), as might be expected from the fact
that the personal monitor was placed on the bedside table during the sleeping period.
Personal exposures were essentially uncorrelated with outdoor concentrations (R2 = 0 to 2%)
(Ozkaynaket al., 1993).
April 1995
7-25
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-6a. PTEAM PREPILOT STUDY: TWENTY FOUR HOUR (24-h) PM-10
CONCENTRATIONS (/*g/m3)
House
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
5
5
5
6
6
7
7
8
8
9
9
Mean
SD
SE
Day
1
3
5
7
1
3
5
7
1
3
5
7
2
4
6
2
4
6
8
10
9
11
9
11
8
10
Person 1
102
142
158
92
109
99
131
62
98
100
143
76
109
90
99
80
70
80
130
150
209
80
135
97
136
273
117.2
44.9
8.8
Person 2
86
125
150
127
158
140
87
56
107
141
132
103
92
77
122
104
77
78
152
102
126
71
178
151
102
91
112.9
30.8
6.0
Indoors
54
38
49
34
122
37
41
32
86
39
71
36
77
34
36
76
62
54
114
106
46
29
73
38
63
121
60.3
28.5
5.6
Outdoors
132
49
70
49
112
48
70
46
115
45
79
44
102
47
37
99
65
50
39
51
72
39
59
28
43
48
63.0
27.1
5.3
April 1995
7-26
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-6b. PTEAM PREPILOT STUDY: TWENTY-FOUR HOUR (24-h) PM-2.5
CONCENTRATIONS (/tg/m3)
House
1
1
1
2
2
2
3
3
3
4
4
4
4
5
5
5
5
6
6
7
7
8
8
9
9
Mean
SD
SE
Day
2
4
6
2
4
6
2
4
6
1
3
5
7
1
3
5
7
9
11
8
10
8
10
9
11
Person 1
44
55
55
58
46
51
53
62
109
75
46
118
40
65
59
40
34
71
77
64
111
53
110
178
105
71.2
32.7
6.5
Person 2
96
88
382
53
100
50
66
94
88
61
43
94
40
69
70
56
53
81
75
135
67
100
1453
48
58
140.8
275.5
55.1
Indoors
22
25
21
31
27
28
48
30
39
33
19
31
17
62
35
42
25
56
53
17
32
27
35
70
42
34.7
13.7
2.7
Outdoors
67
39
33
52
43
40
58
35
39
71
29
46
26
96
38
55
28
33
18
27
35
27
35
40
28
41.6
16.8
3.4
1 7.3.3.2 Main Study
2 7.3.3.2.1 Study design
3 A three-stage probability sampling procedure was adopted (Pellizzari et al., 1993a).
4 Ultimately 178 residents of Riverside, CA took part in the study in the fall of 1990.
April 1995 7-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 Respondents represented 139,000 ± 16,000 (S.E.) nonsmoking Riverside residents aged ten
2 and above. Their homes represented about 60,000 Riverside homes.
3 Each participant wore the PEM for two consecutive 12-h periods. Concurrent PM10
4 and PM2 5 samples were collected by the stationary indoor monitor (SIM) and stationary
5 ambient monitor (SAM) at each home. The SIM and SAM were identical to the PEM except
6 for the pump, which was a Medo pump operated off line current. A total of ten particle
7 samples were collected for each household (day and night samples from the PEM10, SIM10,
8 SIM2 5, SAM10, and SAM2 5). Air exchange rates were also determined for each 12-h
9 period. Participants were asked to note activities that might involve exposures to increased
10 particle levels (nearby smoking, cooking, gardening, etc.). Following each of the two 12-h
11 monitoring periods, they answered an interviewer-administered questionnaire concerning their
12 activities and locations during that time.
13 Up to four participants per day could be monitored, requiring 48 days in the field. A
14 central outdoor site was maintained over the entire period (September 22, 1990 through
15 November 9, 1990). The site had two high-volume samplers (Wedding & Assoc.) with
16 lO-pim inlets (actual cutpoint about 9.0 /mi), two dichotomous PM10 and PM2 5 samplers
17 (Sierra-Andersen) (actual cutpoint about 9.5 /*m), one PEM, one PM10 SAM, and one PM2 5
18 SAM.
19 7.3.3.2.2 Results
20 Of 632 permanent residences contacted, 443 (70%) completed the screening interview.
21 Of these, 257 were asked to participate and 178 (69%) agreed.
22 7.3.3.2.3 Quality of the Data
23 More than 2,750 particle samples were collected, about 96% of those attempted. All
24 filters were analyzed by X-ray fluorescence (XRF) for a suite of 40 metals. More than
25 1,000 12-h average air exchange rate measurements were made. A complete discussion of
26 the quality of the data is found in Pellizzari et al., 1993b, and in Thomas et al., 1993. Blank
27 PEM and SIM/SAM filters (N = 51) taken to the field increased in mass by an average of
28 9.5 pig; this value was subtracted from each field sample. Limits of detection (LODs), based
29 on three times the standard deviation of the blanks, were on the order of 7 to 10 /ig/m3. All
April 1995 7-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 field samples exceeded the LOD. Duplicate samples (N = 363) showed excellent precision
2 for all types of particle samplers at all locations, with median relative standard deviations
3 ranging from 2 to 4%.
4 The collocated particle samplers at the central site showed good agreement, with
5 correlations ranging from 0.96 to 0.99. The central-site PEMs collected about 12% more
6 mass than the dichotomous samplers, which in turn collected about 7% more than the
7 modified Wedding hi-vol samplers. These relative relationships had also been noted in the
8 pre-pilot study in Azusa. It was noted that the Wedding samplers collected about the same
9 mass as the dichotomous samplers during the day, but about 13% less mass at night.
10 Multivariate tests indicated that the Wedding samplers appeared to have a temperature
11 dependency, amounting to an increase of about 1% per °F (Ozkaynak et al., 1993, Appendix
12 A).
13 7.3.3.2.4 Concentrations
14 Concentrations of particles and elements have been reported (Clayton et al., 1993;
15 Ozkaynak et al., 1993; Pellizzari et al., 1993; Wallace et al., 1993). Population-weighted
16 daytime personal PM10 concentrations averaged about 150 pig/m3, compared to concurrent
17 indoor and outdoor mean concentrations of about 95 ^g/m3 (Table 7-7). The overnight
18 personal PM10 mean was much lower (77 /*g/m3) and more similar to the indoor (63 jug/m3)
19 and outdoor (86 /xg/m3) means. About 25% of the population was estimated to have
20 exceeded the 24-h National Ambient Air Quality Standard for PM10 of 150 £ig/m3. Over
21 90% of the population exceeded the California Ambient Air Quality Standard of 50 /-ig/m3.
22 7.3.3.2.5 Correlations
23 The central site appeared to be a moderately good estimator of outdoor particle
24 concentrations throughout the city. Spearman correlations of the central-site concentrations
25 measured by all three methods (PEM-SAM, dichot, Wedding) with outdoor near-home
26 concentrations as measured by the SAMs ranged from 0.8 to 0.85 (p< 0.00001). Linear
27 regressions indicated that the central-site 12-h readings could explain 57% of the variance
28 observed in the near-home 12-h outdoor concentrations (Figure 7-3).
April 1995 7-29 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-7. POPULATION-WEIGHTED3 CONCENTRATIONS AND
STANDARD ERRORS (/tg/m3)
Sample type
Daytime PM10
Personal
Indoor
Outdoor
Overnight PM10
Personal
Indoor
Outdoor
Daytime PM2 s
Indoor
Outdoor
Overnight PM2 5
Indoor
Outdoor
N
171
169
165
168
163
162
173
167
166
161
Geom.
Mean
129
78
83
68
53
74
35
38
27
37
GSD
1.75
1.88
1.68
1.64
1.78
1.74
2.25
2.07
2.21
2.23
Arith.
Mean + SE
150 + 9
95 + 6
94 + 6
77+4
63 + 3
87 + 4
48+4
49 + 3
36 + 2
51+4
Percentile
90% + SE
260 + 12
180 + 11
160 + 7
140 + 10
120 + 5
170 + 5
100 + 7
100 + 6
83+6
120 + 5
98%
380
240
240
190
160
210
170
170
120
160
aPersonal samples weighted to represent nonsmoking population of 139,000 Riverside residents aged 10 or
above. Indoor-outdoor samples weighted to represent 61,500 homes with at least one nonsmoker aged 10 or
above.
1 Outdoor 12-h concentrations of PM10 could explain about 25 to 30% of the variance
2 observed in indoor concentrations of PM10, but only about 16% of the variance in 12-h
3 personal exposures to PM10 (Figure 7-4). This is understandable in view of the importance
4 of indoor activities such as smoking, cooking, dusting, and vacuuming on exposures to
5 particles. The higher daytime exposures were even less well represented by the outdoor
6 concentrations.
7 Indoor concentrations accounted for about half of the variance in personal exposures.
8 However, neither the indoor concentrations alone, nor the outdoor concentrations alone, nor
9 time-weighted averages of indoor and outdoor concentrations could do more than explain
April 1995
7-30
DRAFT-DO NOT QUOTE OR CITE
-------
O)
600
500
§ 400
<•*
O
E 300
"2
tC
£
(0
200
Backyard = 1.03*Central + 17.6
#-0.57 N-323
0
50 100 150 200
Central site reference monitor mean (ng/m3)
250
Figure 7-3. Central-site mean of two dichotomous samplers versus residential outdoor
monitors. R2 = 57%.
1 about two-thirds of the observed variance in personal exposures. The remaining portion of
2 personal exposure is assumed to arise from personal activities or unmeasured
3 microenvironments that are not well represented by fixed indoor or outdoor monitors.
4 7.3.3.2.6 Discussion
5 The more than 50% increase in daytime personal exposures compared to concurrent
6 indoor or outdoor concentrations suggested that personal activities were important
7 determinants of exposure. However, the nature of this "personal cloud" of particles has not
8 yet been determined. Scanning electron microscopy was undertaken on 138 personal filters
9 (Mamane, 1992). Skin flakes were common on many filters. A preliminary analysis
10 suggested that the average number of skin flakes per filter was 120,000 to 150,000. The
11 mass of a small number of personal filters may have been considerably increased by
12 unusually large numbers of skin flakes. However, attempts to calculate the mass of skin
April 1995
7-31
DRAFT-DO NOT QUOTE OR CITE
-------
500
400
V)
o
2> 300
(0
o
8-
75
c
-------
20
40 60 80 100
Percent increase in personal cloud
Figure 7-5. Increased concentrations of elements in the personal versus the indoor
samples.
1
2
3
4
5
6
7
8
9
10
11
12
13
fact that the participants were sleeping for much of the 12-h overnight monitoring period,
and were thus not engaging in these particle-generating or reentraining activities.
A source apportionment of the personal PM-10 mass during the daytime period is shown
on Figure 7-6. This chart is derived by subtracting the average SIM and SAM (95 /xg/m3)
from the mean PEM (150 /ig/m3) given on Table 7-7. The 55 /xg/m3 difference is shown as
the 37% fraction of the total of 150 ;ug/m3 labelled Personal 37%. The source of this
'personal cloud' is indeterminable from the SIM, SAM and PEM data. As discussed
previously, it is likely to consist primarily of resuspended dust that would have a composition
of a mixture of all the other sources. The 15% other-indoor PM represents the indoor mass
that could not be assigned to ETS, cooking or ambient PM. It is likely that the 52% of
other-indoor plus personal-cloud categories contains an appreciable amount of ambient PM
that came indoors over a long period of time and is resuspended by activity. If so, then the
PEM would be at least 50% of ambient origin.
April 1995
7-33
DRAFT-DO NOT QUOTE OR CITE
-------
Outdoor
42%
Personal
37%
Smoking
3%
N = 166 Samples
Cooking
3%
Other Indoor
15%
Figure 7-6. Source apportionment of PTEAM PM-10 Personal Monitoring (PEM) Data.
'Other indoor' represents PM found by the indoor monitor (SIM) for which
the source is unknown. 'Personal PM' represents the excess PM captured
by the PEM which cannot be attributed to either indoor (SIM) or outdoor
(SAM).
Source: Clayton et al. (1993).
1 The PTEAM Study and the 13 PEM studies discussed in this chapter so far are
2 summarized in Table 7-8. This table shows that many of the early studies reported no
3 statistically significant correlation between PEM and SAM. However, these early studies
4 were all characterized by a non-probability sample and a relatively small sample size. The
5 PTEAM study in Riverside was a probability sample (Clayton et al., 1993) and the Lioy et
6 al. (1991) study in Phillipsburg, which was not a probability sample, have large sample sizes
7 and achieved significance. The other studies, such as WHO/UNEP (1982a,b) or Morandi et
8 al. (1988) are equivocal. Consequently it is not clear yet what the general pattern is and
9 why. In the following sections, PEM/SAM comparisons of some constituents and two means
10 of visualizing the complex relationships of SAM and PEM are discussed.
April 1995
7-34
DRAFT-DO NOT QUOTE OR CITE
-------
I
VO
Lf»
71
&
w
£>
H
1
0
O
t— ^
2;
O
H
O
O
H
W
TABLE 7-8. COMPARISON OF PEM EXPOSURE OF INDIVIDUALS TO THE SIMULTANEOUS AMBIENT (SAM)
PM CONCENTRATION IN TEN U.S. CITIES AND FOUR FOREIGN CITIES
Reference
Binder et al.
Dockery & Spengler
Dockery & Spengler
Spengler et al.
WHO/UNEP
Spengler et al.
WHO/UNEP
Sexton & Spengler
WHO/UNEP
WHO/UNEP
Morandi et al.
Lioy et al
Perritt et al.
Clayton et al.
Year
1973
1975
1976
1979
1981
Winter
Winter
Summer
Summer
1981
1982
Summer
Winter
1982
1982
Winter
Summer
Monsoon
1985
Winter
Summer
1988
1988
1989
1990
Location
Ansonia
Watertown
Steubenville
Topeka
Toronto
non-asthmatic
non-asthmatic
asthmatic
asthmatic
Kingston/
Harriman
Zagreb
Waterbury
Bombay
Beijing
Houston
Phillipsburg
Azusa
Riverside
PM /xm
5
3.5
3.5
3.5
25
3.5
5
3.5
3.5
3.5
3.5
10
2.5
10
10
n
20
18
19
46
13
13
13
13
97
12
48
15
20
30
14#
14*
9
9
141
Time
24-h
24-h
12-h
12-h
8-h
8-h
8-h
8-h
24-h
1-wk
24-h
24-h
24-h
1-wk
12-h
24-h
24-h
24-h
24-h
24-h
Mean PEM
115
35
57
30
122
124
91
124
44
114
IS7
36
127
67
58
177
66
27
86
76
79
115
113
Mean SAM
59
n
64
13
68
7S
54
80
18
55
193
17
m
65
51
421
192
16
60
60
43
62
84
R2 PEM vs SAM
NS
0.00
0.19
0.04
0.15
0.10
0.00
0.07
0.00
0.00
0.50
0.00
0.26
0.20
0.02
0.07
0.03
0.34
0.04
0.25
0.01
0.01
0.23
P
NS
NS
NR
NS
NS
NS
NS
NS
NS
NS
NR
NS
NR
NR
NS
.09
NS
<.05
.008
.001
NS
NS
NR
n
H
w
n = Number of individuals carrying personal monitors
NS = Not statistically significant from 0
NR = p Value not reported, but mentioned as significant
# = 14 subjects carried PEMS for 14 days for 191 valid measurements
* = Three outliers are removed and regression is for 188 measurements
Year = year study was performed
-------
1 7.3.4 Personal Exposures to Constituents of PM
2 The absence of an apparent correlation between ambient PM concentrations and
3 personal exposures to PM found in most early US and international studies does not
4 necessarily hold for specific chemical constituents of the PM that have predominantly outdoor
5 sources. Suh et al. (1993) measured personal exposures to sulfate (SO4=) and acidity (H+),
6 and ambient and indoor concentrations in State College, PA, summer 1991. The correlations
7 between personal and ambient values of sulfate and acidity were R2= 0.92 and
8 0.38 respectively, which is in marked contrast to the R2 ~ 0 between earlier reported
9 ambient PM and personal PM (Sexton et al., 1984, Spengler et al., 1985; Morandi et al.,
10 1988).
11 Figure 7-7 shows the consistent relation between ambient and personal sulfate
12 measurements (slope = 0.78 ± 0.02), and Figure 7-8 shows the improvement in prediction
13 by using the TWA with a correction factor (estimated personal sulfate = 0.885*TWA,
14 R2 = 0.95 with slope = 0.96 + 0.02). Personal acidity was also computed by the same
15 equation with a correction for personal ammonia (NH3) exposure that gave an R2 = 0.63.
16 As opposed to PM which has both indoor and outdoor sources, the sulfate and acidity are
17 virtually all of outdoor origin. Consequently, the characteristics of the indoor environment,
18 such as air conditioning and ammonia sources, modify the personal exposures indoors.
19 Another important consideration in estimating personal exposures, from the indoor and
20 outdoor environmental measurements, is that the chemical composition of the excess in
21 personal exposure compared to the TWA exposure calculation may be significantly different
22 than that predicted from the indoor and ambient data alone. For example, the excess
23 personal PM found by Morandi et al. (1988) appeared to be related to in-traffic exposures
24 which would have a different chemical composition compared to either the average ambient
25 or average indoor compositions. Exposures to particles in vehicles are quite variable,
26 perhaps as a function of traffic density, and do not constitute a simple microenvironment for
27 estimation purposes. For example, Roemmelt et al. (1993) reported an average in-bus
28 concentration of 570 Mg/m3 of TSP (open filter collection at 1 m3/h), with a minute value
29 peak-to-mean ratio of 3:1 using a MINIRAM monitor (« PM5), during an 8.5-h period on
30 16 March 1993 in downtown Munich. Simultaneous ambient TSP data were not reported.
April 1995 7-36 DRAFT-DO NOT QUOTE OR CITE
-------
CO
7S"
~o
I
CD
Q_
600
500
400
300
200
100
100
200
300
400
Outdoor (nmoles/rrr)
500
600
Figure 7-7. Personal versus outdoor SO4 . Open circles represent children living in air
conditioned homes; the solid line is the 1:1 line.
1
2
3
4
5
6
1
8
9
In addition to the two factors cited just above, a microscale 'personal cloud' can be
generated by the person's activities which complicates the exposure measurement process.
This effect is most important in occupational settings where personal exposures are not
readily comparable to weighted area sampling measurements. For example, Lehmann et al.
(1990) measured workers exposure to diesel engine exhaust by personal monitoring of PM10
with a range of 0.13 to 1.2 mg/m3, compared to an area estimate range of 0.02 to
0.80 mg/m3. MMWR (1988) reports the exposures of nurses and respiratory therapists to
the aerosols of ribavirin during treatment of patients by ribavirin aerosols administered inside
an oxygen tent. Bedside area monitors averaged 317 /ng/m3 while personal exposures ranged
April 1995 7.37 DRAFT-DO NOT QUOTE OR CITE
-------
500
500
Measured (nmoles/nrr)
Figure 7-8. Estimated ("best fit" model) versus measured personal SO4 . Model
includes indoor and outdoor concentration and activity data. Open circles
are air conditioned homes; the solid line is the 1:1 line.
1 from 69 to 316 /zg/m3 with an average of 161 /xg/m3. The implications of these differences
2 in exposure with position relative to a source are discussed in previous Section 7.2.3.
3 Environmental Tobacco Smoke (ETS) is a category of PM that is found in many indoor
4 settings where smoking is taking place or recently had occurred. As stated in the Indoor Air
5 Section 7.6, ETS is the major indoor source of PM where smoking occurs. Because of the
6 depth of the discussion and coverage of ETS in Section 7.6, no further discussion will be
7 made here other than to note that ETS adds on the order of 25 - 30 /zg/m3 to 24-h average
April 1995
7-38
DRAFT-DO NOT QUOTE OR CITE
-------
1 personal exposures and residential indoor environments where smoking takes place
2 (Holcomb, 1993; Spengler et al., 1985).
3 ETS represents an appreciable contribution to the personal PM exposure of the
4 nonsmoker which is uncorrelated to the ambient PM concentration. In many of the PM PEM
5 studies, smoking status and exposure to ETS was self-identified by the subjects in their
6 questionnaires and diary responses. In the absence of an independent verification by a
7 measure of nicotine or cotinine, a subject could be misclassified (Leaderer and Hammond,
8 1991). For example, smoking of a non-tobacco product may not be reported. In the
9 USEPA Denver and Washington CO exposure studies (Wallace et al., 1988) there were
10 several subjects who had high breath CO but their diary and CO PEM record gave no
11 evidence of an equivalent exposure to CO or methylene chloride (which metabolizes to CO).
12 A biological measure of cotinine or nicotine could indicate whether the excessive breath CO
13 was from nonreported smoking or endogenously produced.
14 The random ETS increment will tend to reduce the correlation between PEM and SAM.
15 If one were able to subtract out the ETS from the PEM PM data, the correlation of SAM
16 with the non-ETS PEM PM might be improved (Dockery and Spengler, 1981).
17 As stated as a caveat in the introductory section 7.1, the inhalation of main-stream
18 tobacco smoke will be a major additive exposure to PM for the smokers, which dwarfs the
19 nonsmoker's PEM PM. Therefore the results presented so far apply only to nonsmokers,
20 and a major proportion of the US population (e.g. smokers) has a total exposure to PM that
21 is at least one order of magnitude greater than that of the nonsmokers.
22 7.4 INDIRECT MEASURES OF EXPOSURE
23 The early air pollution literature related health to ambient particulate matter (TSP)
24 concentrations as a surrogate for personal exposures to PM. Although this relationship has
25 been shown to be highly questionable for specific individuals, it still is used in studies such
26 as Pengelly et al. (1987) who estimated TSP exposures of school children in Hamilton,
27 Ontario, by interpolation of ambient TSP concentrations to the school locations.
28 The first usage of a time-weighted-average (TWA) of environmental exposures to
29 estimate total human personal exposure to an air pollutant (Pb) was by Fugas et al. (1973).
April 1995 7.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 In theory, a human exposure to PM could be estimated by use of Equation 7-2 and
2 knowledge of the average PM concentration in each microenvironment (/xE) that a person
3 experiences and the duration of the exposure in each such ^iE (Duan, 1982; Mage, 1985).
4 For a room with no source in operation, the whole room could be treated as a single /xE.
5 However, when a PM source is in operation and gradients exist, that very same room may
6 need to be described by multiple piEs. These pEs could have dimensions of an order of a
7 few centimeters close to the source, and dimensions of an order of several meters, farther
8 away from the source.
9 Under research conditions, the complete spatial variation of a gaseous tracer
10 concentration in a test chamber can be mapped (Yost et al., 1994) and measurements at
11 multiple fixed points can be made (Baughman et al., 1994). Baughman et al., 1994,
12 proposed that a /xE could be considered to have a uniform concentration if the coefficient of
13 concentration variation within it was less than 10% (standard deviation/mean). In the
14 presence of a source of PM, these data indicate that tens of such /iEs would have to be
15 defined when a subject is moving about in its immediate vicinity. A simultaneous video
16 recording of an individual's exact position and activity, while measuring the instantaneous
17 concentration, can be used to visualize a concentration field (O'Brien et al., 1989; Gray
18 et al., 1992) and could be used to measure the concentration field distortion that occurs from
19 the presence of the person. These new techniques establish the variability of concentration
20 over small distances and their relations to human activity. It is clearly impossible to use
21 these research techniques routinely in an exposure survey or to maintain multiple area
22 monitors throughout a normal setting of daily activity while recording human activity in their
23 vicinities.
24 Ogden et al. (1993) compared exposures from personal sampling and static area
25 sampling data for cotton dust exposures. The British cotton dust standard specifies static
26 sampling, because the 1960 dose-response study used to set the standard used static sampling
27 data to compute worker exposure and dosage. Ogden et al. (1993) found median personal
28 exposures of 2.2 mg/m3 corresponding to a mean static background concentration of
29 0.5 mg/m3. They concluded that "The presence of the body and its movement affect what a
30 personal sampler collects, so static comparisons cannot be used to infer anything about the
31 relationship of the (static) method with personal sampling." Ingham and Yan (1994)
April 1995 7-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 confirmed this finding by modelling the human body as a cylinder and showing that unless
2 the personal monitor length/diameter ratio was greater than four, the aspiration efficiency
3 (the fraction of particles sampled that would be sampled in the absence of the body) could be
4 greatly affected.
5 Rodes et al., 1991, compared the literature relationships of personal exposure
6 monitoring (PEM) to /*£ area monitoring (MEM) for PM, as shown in Figure 7-9, to which
7 Ogden et al., 1993 is added as a single point. The authors found that PEM/MEM ratios
8 ranged from 3 to 10 in occupational settings, and from 1.2 to 3.3 in residential settings.
9 These combined data show that approximately 50% of all measured PEM PM values are
10 more than 100% greater than the estimated simultaneous MEM values using the TWA
11 approach. Their explanation points to this excess PM as due to the spatial gradient about
12 indoor sources of PM which are usually well away from area monitors which thus fail to
13 capture the high exposures individuals may get when in close proximity to a source. They
14 suggest that clothing lint and skin dander could only add, at most, a few percent to the total
15 PM mass collected by a personal exposure monitor.
16 7.4.1 Personal Exposure Models Using Time-Weighted-Averages (TWA) of
17 Indoor and Outdoor Concentrations
18 Several studies have used the relationship of Equation 7-2 to compute the time-
19 weighted-average (TWA) exposure of subjects. The procedure calls for a time-activity diary
20 to be kept so that the time at-home, outdoors, at-work, in-traffic, etc., can be defined. By
21 use of fiE monitoring data from the study itself (or literature values of PM concentrations in
22 similar /*Es) and concurrent ambient monitoring, one can predict the concentration that would
23 be measured if the subject had carried a PEM.
24 Because people in the USA spend 21 h indoors each day (Robinson and Nelson, 1995),
25 the concentration in indoor /^Es is a most important quantity for usage within a TWA PM
26 model. The important articles on indoor air qualtity for PM have been reviewed extensively
27 in Section 7.6. Many of these articles, such as Quackenboss et al. (1991), estimated TWA
28 PM10 exposures from SIM and SAM measured at subject's homes without collecting
29 simultaneous PEM for validation of the TWA model.
April 1995 7.4! DRAFT-DO NOT QUOTE OR CITE
-------
100
10
CO
DC
LJJ
LLl
Q_
0.1
5 1.0
30
° Stevens (1969)
A Fletcher and Johnson
O Parker etal. (1990)
o Lioyetal. (1990)
A EPA PTEAM data
© Ogden et al.
50
(1988)
70
90 95 98
Data median og
13.40 1.98
1.78
a
A
O
o
A
©
5.70
1.58
1.98
4.40
1.17
3.40
1.53
1.62
2 5 10 30 50 70
Cumulative % less than
Figure 7-9. Personal activity cloud and exposure.
90 95 98
Source: Rodes et al. 1991
April 1995
7-42
DRAFT-DO NOT QUOTE OR CITE
-------
1 The articles that are discussed here predict PM exposures of non-smokers that include
2 ETS, and most provide PEM data for comparison. This may also be interpreted as the
3 exposure of smokers minus their exposure to direct main-stream smoke and incompletely
4 diluted side-stream smoke.
5 As opposed to the gaseous pollutants for which continuous hour-to-hour time series of
6 SAM data are available, PM SAM monitoring data have been often only available as a time
7 series of 24-h SAM measurements. Consequently, in much of the early PM TWA
8 literature, the modelers assumed, by necessity, the same ambient PM in the morning and
9 evening, which might not be accurate (Dockery and Spengler, 1981).
10 Spengler et al. (1980) in a study of PEM, SAM and SIM in Topeka, Kansas, found the
11 averages of PEM = 30 Mg/m3, SIM = 24 Aig/m3 and SAM = 13 /*g/m3. They note "It
12 suggests that somewhere in an individual's daily activities, they are being exposed to PM at
13 concentrations higher than what is measured either indoors or outdoors". This relationship
14 has been found in almost all other studies, such as PTEAM (Clayton et al., 1993) where
15 daytime PEM averaged 150 ^g/m3 and SIM and SAM averaged just under 100 /ig/m3.
16 Spengler et al.(1985) measured 24-h PEM, SIM and SAM. The resulting relationship based
17 on Equation 1 was: PEM = 17.7 /ig/m3 + 0.9 TWA. The authors noted, in addition to the
18 previous suggestion, that the excess of PEM over TWA may be due to an incorrect
19 assumption that the indoor and outdoor are constant during the 24-h sampling period.
20 Morandi et al., (1988) compared PM3 5 PEM with simultaneous SIM and SAM data.
21 They found that their TWA model overestimated PEM below 27 /zg/m3 SAM and
22 underestimated PEM above 27 /ig/m3 SAM. The authors concluded that "this result indicates
23 that there were differences between PM mean concentrations in ^iEs with similar
24 characteristics ... The implication for air pollution health effect studies is that, for
25 contaminants with significant indoor sources, PEM may be the only adequate measure of
26 exposure when using short-term averaging times".
27 Koutrakis et al. (1992), in a study discussed in Section 7.6 on Indoor Air report that
28 their source-apportionment mass-balance model predicts penetration from outdoors to
29 indoors of order 85-90% for Pb and sulfur compounds. The authors claim that:
April 1995 7.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 "We can satisfactorily predict indoor fine aerosol mass and elemental concentrations
2 using the respective outdoor concentrations, source type and usage, house volume and
3 air exchange rate."
4 The authors note that this may be a cost-effective approach to estimating peoples'
5 exposure while indoors since the necessary ambient data may be available and the housing
6 profile may be collected with a simple interview. This technique could possibly correct the
7 shortcomings noted above by Morandi et al. (1988).
8 Colome et al. (1992) measured indoor and outdoor PM-10 at homes of asthmatics in
9 California. Their personal monitoring data, limited to three individuals, confirmed that
10 "some protection from higher outdoor concentration is afforded by shelter if smokers and
11 other particulate sources are not present". This observation may be important for estimating
12 the exposure of elderly and infirm people who are assumed to be the susceptible cohort.
13 Klepeis et al. (1994) present an up-to-date TWA PM Model that uses, as an input, real-
14 time hourly PM SAM data and a mass balance equation to predict exposures of nonsmokers
15 in various indoor settings based on ambient PM data, presence of PM sources such as
16 smokers, and other variables relating to air exchange rates. The addition of the additive
17 terms that allow for sources, such as cooking and presence of smokers adds to the TWA of
18 Equation 7-2, which in effect is a correction for the underprediction of the /*E concentration.
19
20 In summary, as described by several authors, the PM PEM exposure of individuals who
21 are not smoke exposed has been shown to be higher than their corresponding TWA of SIM
22 and SAM. The exact reason for this excess in PM, sometimes called a 'personal cloud', is
23 not known (Rodes et al., 1991). It has been thought to reflect the fact that the person's
24 presence itself can stir up loosely settled-dust by induced air motion and vibration (Ogden,
25 1993). Thatcher and Layton (1994) gave an example where merely walking into a room
26 raised the total suspended dust from 10 to 20 /*g/m3. A study by Litzistorf et al. (1985) of
27 asbestos type fibers in a classroom showed how fibers (f) were stirred up when it was
28 occupied. The levels rose from below the detectable level of 10000 f/m3 to 80000 f/m3
29 when occupied, and they returned to below detectable levels within 1 h after the end of the
30 class. Millette and Hays (1994) present a detailed discussion of the general topic of
31 resuspended dust in their text on settled asbestos dust.
April 1995 7-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 It may not be a proper procedure to use a 24-h average concentration in a physical
2 setting, such as a kitchen, to estimate a person's exposure while in the kitchen. As described
3 previously in the discussion of the definition of a microenvironment, the same kitchen can
4 constitute one or more /^Es depending on the source operation pattern. In many studies, such
5 as Spengler et al. (1985), the SIM sampled the indoor residential setting for 24-h in phase
6 with the PEM. The resulting average SIM will often understimate the person's exposure
7 while they are at home and may contribute to the difference between a TWA exposure and
8 the PEM.
9 In a similar manner, a person's workplace exposure may be more or less than that in
10 their home. In the PTEAM study (Clayton et al., 1993), there was a general decrease in
11 exposure for those who were employed outside their home. However, employment in a
12 "dusty trade", such as welding, may increase their PM PEM. Lioy et al. (1990) give an
13 example of a subject with a hobby involving welding which led to a 24-h PEM reading of
14 971 jig/m3. The variables influencing the contribution to PEM PM from industrial
15 exposures have not been discussed in this chapter because of their complexity. The reader
16 should bear in mind that application of a TWA model to a subject with such an exposure
17 may create a high TWA estimate with a large uncertainty.
18 Another exposure category that is important for TWA analysis is that within a vehicle
19 in transit (Rudolf, 1994). In California, people spend approximately 100 minutes per day in
20 or near a vehicle (Jenkins et al., 1992). In vehicles people are exposed to auto exhaust, road
21 dust resuspended from vehicle turbulence, and PM generated within the vehicle as ETS or
22 exhaust leakage. Roemelt et al. (1993) reported a range of TSP in an urban bus up to 1500
23 Aig/m3 with a mean of 570 /xg/m3 as measured with an optical monitor during an 8.5-h
24 daytime period.
25 Indirect estimation of a person's time-weighted-average (TWA) PM exposure may be a
26 cost-effective alternative to direct PEM PM measurement. Mage (1992) compared the
27 advantages and disadvantages of the TWA indirect method compared to the direct PEM
28 method. The primary advantages of the indirect method are the low cost and low burden on
29 the subject, because it uses only a time-activity diary and no PM PEM is required; the
30 disadvantage is the low accuracy. The primary advantage of the PEM PM method is that it
31 is a high accuracy direct measurement; the disadvantage is the high cost and high burden on
April 1995 7.45 DRAFT-DO NOT QUOTE OR CITE
-------
1 the subject (see Section 7.2.1.4). Mage (1992) proposed a combined study design in which
2 direct measurements on a subset of subjects can be used to calibrate the TWA estimates of
3 the other subjects. Duan and Mage (1995) present an expression for that includes: the
4 optimum fraction of subjects to carry the PEM as a function of the relative cost of the PM
5 PEM to the TWA PM estimate and the correlation coefficient between the PEM and the
6 TWA.
7 7.5 DISCUSSION
8 7.5.1. Relation of Individual Exposures to Ambient Concentration
9 The previous sections discussed the individual PM PEM vs PM SAM relationships of
10 14 studies listed in Table 7-8. In many of the PM studies there is no statistically significant
11 linear relationship between PEM and SAM, and in other studies the relationship is positive
12 and statistically significant. This section discusses these data in terms of understanding the
13 complex relationship between the SAM concentrations and the individual PEM exposures. In
14 the following section, the relationship of the SAM to the mean PEM in the community
15 surrounding the SAM will be presented.
16 The principle of superposition is offered as a basis for visualization of the process
17 involved in creating a total exposure. A linear system will exist for respirable-PM PEM
18 exposures if the expected PEM response to a source emitting 2 mg/min of PM is exactly
19 twice the PEM response to that identical source emitting 1 mg/min of identical PM. If
20 superposition applies, then we can construct the total exposure by adding all the increments
21 of exposures from the various source classes and activities that a subject performs on a given
22 day.
23 Let the SAM be representative of the macroscale ambient PM concentration in the
24 community as shown on Figure 7-10a. This is the exposure that would be measured for a
25 homeless person if they spent 24-h per day outdoors near the SAM site. Neglecting local
26 microscale variation (e.g. backyard barbecue or leaf burning), while people are outdoors they
27 are exposed to 100% of the SAM value (Figure 7-10b). Assume that this exposure is also
28 the baseline PM for a location in traffic which occurs outdoors. The increment produced by
29 the local traffic is considered later.
April 1995 7-46 DRAFT-DO NOT QUOTE OR CITE
-------
SAM
SAM
Outdoors
SAM
' Indoors "•
6 12 18 24
Time - Hours
6 12 18 24
Time - Hours
6 12 18 24
Time - Hours
SAM
Traffic
Increment
to SAM
6 12 18 24
Time - Hours
Occupational
Exposure
Increment
to SIM
SAM
6 12 18 24
Time - Hours
ETS
Exposure
SAM
6 12 18 24
Time - Hours
SAM
T / X
/ \
SIM
/ Indoors \
/ non-ETS
\ / non-SAM N x
X s | -
- T / -
—if r h—
£.
o>
c
..
°^E
1
20 -*•
Cigarettes
Smoked
;
SAM
- •
-
*
___
6 12 18
Time - Hours
24
6 12 18 24
Time - Hours
Figure lOa-h. Components of personal exposure.
April 1995
7-47
DRAFT-DO NOT QUOTE OR CITE
-------
1 While people are indoors, they are exposed to a variable fraction of time-lagged SAM
2 PM. This constitutes an amount of (1) the fresh PM which depends on recent SAM and the
3 air exchange rate between indoors and outdoors, and the PM deposition sinks (filtration of
4 recirculated air, surfaces, etc.), and (2) PM from outdoor sources that had been deposited in
5 the past but is resuspended due to human activity and air currents. PTEAM (Clayton et al.,
6 1993), as cited in Section 7.6, found that outdoor air was the major source of indoor
7 particles, accounting for 75% of the fine fraction (<2.5 pirn AD) and 67% of the coarse
8 fraction (2.5 pm AD to 10 ^im AD) in indoor air. It is noted that these average fractions
9 will be lower in communities with lower average SAM values. Lewis (1991) reported an
10 apportionment of indoor air PM in 10 homes within a wood burning community in Boise,
11 Idaho. The results showed that 50% of the fine PM was of outdoor origin (SAM), and in 9
12 of 10 homes, 90% of the sulfur was from outdoors (one home had an anomolous sulfate
13 injection from a humidifier using tap water). This is consistent with indoor sources varying
14 independently of the SAM in a stationary manner (constant mean and variance), so that the
15 relative contribution of indoor sources to indoor exposures decreases as SAM increases.
16 Figure 7-10c represents the increment to PEM from outdoor sources of SAM while the
17 subjects are indoors at home and at work. The SAM value is shown as the dotted line for
18 reference in this and all the following Figures 7-10c and 7-10h.
19 While people are indoors, at home and at work, they are also exposed to PM emitted
20 by indoor sources - other than ETS from passive smoking and specific occupational sources.
21 These sources, such as cooking, lint from clothing and furnishings, mold, insects, etc.,
22 create PM that agglomerates and deposits as visible dust that can be continuously
23 resuspended, which constitutes an additional PEM increment. Figure 7-10d shows the
24 additive effect of this source.
25 In traffic, or near vehicles in a parking garage or parking lot, people are exposed to an
26 increment of PM over and above the SAM value for that location. Figure 7-10e shows the
27 additive PM for this setting that would be added to Figure 7-1 Ob for the local vehicular
28 emissions.
29 In an indoor setting, in the presence of a smoker or the wake of a smoker, a PEM will
30 record an increment of ETS associated with the act of smoking. Figure 7-1 Of shows the
31 added PM increment for this source.
April 1995 7-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 At work in a 'dusty trade' such as welder, mechanic, or miner, there will be an
2 increment of exposure associated with these occupational activities that generate PM.
3 Figure 7-10g represents the additive PM for these activities which are assumed to take place
4 'indoors'.
5 Last, but not least, is the physical act of smoking itself. As described previously, the
6 main stream smoke from a cigarette, cigar, or pipe, bypasses the PM monitor and is inhaled
7 directly. The mass of PM inhaled from smoking one-pack-per-day of the lowest Tar 'king
8 soft pack' cigarettes, rated as delivering "1 mg 'tar' per cigarette by FTC method" is 20 mg
9 per day (Woman's Dav. 14 March, 1995). If this were distributed into a nominal 20 m3 of
10 air inhaled per day, it would be an additive increment on the order of 1 mg/m3 to a 24-h
11 PEM reading. Other cigarettes advertised in the same popular magazine are rated at "16 mg
12 'tar' per cigarette by FTC method". Therefore one-pack-per-day smokers can have a PM
13 exposure standard deviation that is much larger than the mean exposure to PM of non-
14 smokers, simply from choice of brand. Figure 7-10h represents the act of smoking as
15 creating exposures represented by the vertical spikes with an integral area > 1 mg-day/m3
16 per day.
17 For all subjects, by the principle of superposition, the sum of the areas in Figures 7-
18 lOb and 7-10c represents the exposure of an individual to the PM constituents that are
19 characterized by a SAM PM concentration. The additional exposure categories that are
20 independent of the SAM concentration (Figures 7-10d through 7-10g) and are appropriate for
21 that subject would represent the portion of 24-h PEM PM that is not associated with SAM.
22 Variance of SAM should explain much of the variance in the SAM related PEM fraction as
23 defined by Figures 7-10b and 7-10c. The summation over a full day for all categories 7-10b
24 to 7-10g would be the PEM for any subject, such as is shown in Figure 7-2 (Repace and
25 Lowery, 1980).
26 Although there are no data for PEM PM exposures of individuals living in homes
27 without any indoor sources of PM, there are data for PEM sulfate as discussed previously in
28 Section 7.3.4. Given that there are negligible sources of sulfur (S) that originate in the home
29 (matches, low-grade kerosine, humidifiers using tap water) the high correlation of PEM S
30 and SAM S (R2 = 0.92) of Figure 7-7 reported by Suh et al. (1993) where no appreciable
31 sources of S were present, is an indication that the same relationship should hold for all
April 1995 7.49 DRAPT-DO NOT QUOTE OR CITE
-------
1 SAM PM of that size range. The data of Anuszewski et al. (1992) show that light scattering
2 particles measured by nephelometry have very high correlation between indoor and outdoor
3 concentrations (R2 > 0.9). Lewis (1991) and Cupitt et al. (1994) report that PM10 appears
4 to penetrate with an average factor of 0.5 in Boise homes without woodburning. The factor
5 goes up to 0.7 with woodburning, and the authors assume that the factor would go up to 0.9
6 in the summer when homes are less tightly sealed.
7 If the variance of the PEM PM portion which is uncorrelated to SAM (lOd to lOg) is
8 very large, the percentage of the variance of the PEM PM that can be explained by the
9 variance of SAM PM will be very small.
10 It may be possible that the 14 different populations sampled, cited in the 14 studies of
11 Table 7-8, have widely different home characteristics, occupations, mode of commuting, and
12 smoking exposures that contribute to the different PEM vs SAM relationships. In some of
13 the cleaner communities, such as Watertown MA, Topeka KS, Waterbury VT, and Kingston
14 and Harriman TN, SAM averaged less than 20 jug/m3. The non-SAM increments to PEM
15 exposure in these locales were greater than the SAM and may have been so variable that the
16 PEM PM became insignificantly correlated with the SAM PM data. The exception is
17 Houston, with a SAM = 16 /ng/m3 and a significant R2= 0.34 (0.005 < p < 0.05).
18 However, Morandi et al. (1988) note that deletion of 2 outlier observations would reduce R2
19 and make it nonsignificantly different from 0 (p > 0.2). This is in contrast to the two large
20 studies in communities with high SAM levels (Clayton et al., 1993, Lioy et al, 1990), where
21 the relations between PEM and SAM were significant.
22 All discussions above relate to nonsmokers. As for the smoker, the exposure from
23 Figure 7-10h would outweigh the sum of all the other exposures, 7-10b through 7-10g. This
24 smoking increment may have an important implication for interpretation of epidemiology
25 studies that relate a surrogate of PEM PM to mortality.
26 In the epidemiology studies of PM and mortality of Chapter 12, the death counts are
27 usually culled to remove suicides and trauma victims. This is because the SAM PM is not
28 considered to have been a possibly contributing cause to the accident, violence or voluntary
29 act that resulted in death. Consequently, smokers and nonsmokers alike are in the residual
30 mortality counts that are regressed against SAM and weather related variables. The
31 community SAM data are representative of the exposure to PM of outdoor origin expressed
April 1995 7-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 by Figures 7-1 Ob and 7-10c. For non-smokers this reflects about 50-75% of their
2 exposure, which is an appreciable amount. However, for one-pack-per-day smokers, if their
3 smoking exposure is more than an order of magnitude higher, the SAM may reflect less than
4 5.0 - 7.5% of their daily exposure to PM.
5 7.5.2 Relation of Community Exposures to Ambient Concentrations
6 Studies of the relationship between ambient PM and mortality/morbidity implicitly
7 assume that the PM concentration at an ambient monitoring station (SAM) is a surrogate for
8 the mean PM exposure of people in the local community. It can be shown that if
9 individuals' probability of mortality from PM exposure is linearly proportional to their PM
10 exposure, then the expectation of total PM-related mortality in the community is proportional
11 to the mean personal exposure to PM in the community. Therefore, it may be appropriate to
12 ask the question, "how well does SAM PM characterize the mean PEM PM in the
13 community?"
14 If all N people in a community carried a PEM, the mean PEM value is obtained by
15 summing all PM PEM values and dividing by N. If indoor and outdoor PM are equally
16 toxic on a ^tg basis, then no further information would be contained in PM SAM. The mean
17 of a random sample of PEM PM measurements on subjects in the community would be an
18 unbiased estimator of the actual community mean PEM PM, and such a mean may be more
19 appropriate for use than a SAM measurement. Mage and Buckley (1995) tested the
20 relationship of the mean PEM PM exposure to the SAM in several locations, and their
21 results are given in the following section.
22 Figures 7-1 la, 7-12a, 7-13a, and 7-14a show the individual personal PEM PM and the
23 corresponding ambient SAM PM from four (4) studies cited in Table 7-8 for which
24 individual data were available. For example, Figure 7-1 la (Lioy et al, 1990) shows a set of
25 PEM samples obtained from 14 nonsmoke exposed individuals on 14 consecutive days.
26 Because these 14 subjects were not selected as a probability sample from the community of
27 Phillipsburg, NJ, we consider their exposures as a biased sample from the exposure
28 distribution that we would measure, had every person in the community carried a PM PEM.
29 The outdoor average is estimated as the mean of four (4) ambient SAM PM values obtained
30 on each day the people carried the PEM for PM10.
April 1995 7-51 DRAFT-DO NOT QUOTE OR CITE
-------
Phillipsburg, NJ (Winter 1988)
(all data included, n-191)
Figure a
Figure b
1000
•800
i
o 600
Q.
400
£. 200
200
§150
o
5:100
CO
I
0)
°- 50
I
R2-0.333 (p-0.031)
m - 0.478 (Mdwr- 0.20)
b - 65.0 (Md«r- 29.3)
n-14
0 50 100 150 200
Mean Outdoor PM-10, ug/m3 (4 Sites)
0 50 100 150 200
Mean Outdoor, ug/m 3 (4 Sites)
Figure 7-lla,b. Personal exposure to PM in Phillipsburg, NJ (Winter, 1988)
Source: Lioy et al. (1990).
1 The data plot is a vertical profile corresponding to up-to-14 valid PEM values obtained
2 on that day. In the regression, each point is weighted equally, and the R2 value of 0.037
3 (p = 0.008) would be significant if the sample were unbiased. The bias of a nonrandom
4 sample has an expectation of zero and a finite variance if the choice of subjects is not based
5 on factors related to exposure to PM. In Figure 7-1 Ib, the mean of the daily PEM values is
6 plotted. The R2 now has increased from 0.037 to 0.333 (p=0.031). The interpretation is
7 that, on the average, variation in SAM only explains on the order of 4% of the variation hi
8 an arbitrary individuals PEM, but that same variation in SAM explains 33% of the variation
9 in their mean exposure.
10 Because both these PEM and SAM values are measured with error, an orthogonal
11 regression may be more appropriate, with inverse variance weighting for each of the PEM
April 1995
7-52
DRAFT-DO NOT QUOTE OR CITE
-------
400
mE300
1?
o
^200
Q.
I 100
0
Beijing, PRO (Winter 1985)
Figure a
R'- 0.064 (p-0.004)
m-0.14(*Merr-0.08)
b-116(std«T-37)
n-45
0 100200300400500600700
Outdoor PM-10 (ug/m3)
400
TO
^3,
O
200
CO
o
y>
0>
Q. 100
I
0
Figure b
0 100200300400500600700
Outdoor PM-10 (ug/m3)
Figure 7-12a,b. Personal exposure to PM in Beijing, China (Winter, 1985).
1 and SAM means. This would be expected to change the regression slightly but not the
2 general conclusion that there is much more influence of SAM variation on the variation in
3 the mean community exposure.
4 In Beijing (Figure 7-12a,b) with a nonprobability sample of 20, the slope remains
5 relatively constant at 0.14 as the R2 value increases from 0.064 to 0.23 with the usage of
6 mean PM PEM exposure.
7 Figure 7-13a,b for Azusa, California, with a nonprobability sample of 9, the correlation
8 between PM PEM and PM SAM is negative (-0.01) and R2 is 0.0001. In this case, taking the
9 mean of personal exposures shows no significant improvement in the R2 value. Such a low
10 value of R2, as in several of the studies cited in Table 7-8, may be caused by several factors,
11 such as sampling error (too few observations), a biased (nonprobability) sample, very strong
12 indoor sources of PM, or commuting of the people during the day to locations with
13 significantly different ambient pollution than in their home community.
April 1995 7.53 DRAFT-DO NOT QUOTE OR CITE
-------
AZUSA.CA (Spring 1989)
Figure a
Figure b
ouu
^ 250
rt
E
? 200
o
2 150
0.
1C
o 100
®
50
n
' • R2-0.00
, •
- •
'Jtf-Ii "
"& . f ' :
t •? 1 ! •
1,1,1,1,!.!.
ouu
^»,
eo
E 250
o>
n
o 200
i
S
5: 150
cO
O
! 100
0.
c
§ 50
2
0
R2- 0.00
n-11
-
_ _
^ •• '
'I. ' "
^
. I , [ , I , i i I , I ,
0 20 40 60 80 100 120 1'
0 20 40 60 80 100 120 140
Mean Outdoor PM-10 (ug/m3)
Outdoor mean represents 2 to 4 sites
Mean Outdoor PM-10 (ug/m3)
Outdoor mean represents 2 to 4 sites
Personal Mean represents 4 to 6 persons
Figure 7-13a,b. Personal exposure to PM in Azusa, CA (Spring, 1989).
Source: Perritt et al. (1991).
1 In Figure 7-14a,b, Riverside, California, with a probability sample of 178 people, the
2 R2 values are significant, and they improved by a factor of three, from 0.16 to 0.49, while
3 the regression equation remained essentially the same.
4 The Riverside CA study and the Phillipsburg NJ study differ in three important aspects.
5 In Phillipsburg, NJ, 14 nonsmoke-exposed at-home people carried a PEM for 14 days; in
6 Riverside, CA, each day up to four different people, some of them who were smoke-exposed
7 at-home, carried a PEM for one day over the 48-day study period.
8 1. The Phillipsburg NJ data have a (potentially) 2-fold more precise estimate of the
9 mean given by 14 PEM measurements as compared to 4 in Riverside, CA
10 (1/V13 = 0.28, < 1/V3 = 0.57).
April 1995
7-54
DRAFT-DO NOT QUOTE OR CITE
-------
300
i 250
o
.c
S 200
1
§ 150
CL
o 50
a.
_a
jffO go O O
'i*7: ° " lW
o~ o o «
50 100 150 200 250 300 350
Backyard Concentration (24 hours)
I
o
250
200
S?
1 150
100
50
LLJ
o
-------
1 2. The lack of smoke exposure to the subjects in NJ would reduce the variance of the
2 PEM measurements, and
3 3. The use of a nonprobability sample in NJ (and all other studies in which
4 nonprobability sampling occurs) limits the statistical applicability of the results.
5 The improvement in the regression R2 value by taking the mean of the PEM PM data is
6 not an important finding of itself. This improvement in the regression coefficient is
7 predictable from the Central Limit Theorem and the process of regression to the mean of the
8 observations - as when random measurement errors are removed. The higher correlation of
9 categorical exposure assignments has been noted in epidemiological studies.
10 The value of the improvement of the mean PEM relationship to SAM is that it
11 provides a better visualization that helps in understanding how mean PEM varies with SAM.
12 It thus provides a measure of the validity of the use of a daily PM SAM as a surrogate for
13 the mean PM PEM in the community. It is clear that the uncertainty in predicting mean
14 personal exposure PM is much smaller than the uncertainty in predicting the personal
15 exposure PM for an individual when we note that the means have a much smaller variability
16 about the line as shown in Figures 7-1 Ib, 7-12b, 7-13b, and 7-14b.
17 There appears to be two distinct categories of exposure studies that are examined:
18 In the first type of study, such as Lioy et al. (1990) and Clayton et al. (1993), there is
19 a significant R2 between individual PM PEM and PM SAM. In this category, there is an
20 appreciable improvement in correlation between the mean PEM and SAM. It has been
21 suggested that these cases with high correlation of PEM PM and SAM PM may arise where
22 the fine portion of the ambient PM (PM2 5) is highly variable from day-to-day, and the
23 ambient coarse fraction is relatively constant. In some locations, the fine portion of the
24 ambient PM (PM-2.5) is more variable from day-to-day than that of the ambient coarse
25 fraction. In an urban area, the fine particle composition and the fine particle concentration
26 are highly correlated from site-to-site on any given day. This is due, in part, to the
27 homogeneous gas phase reactions of SOx and NOx to produce sulfates and nitrates, and
28 aerosol droplet formation with the condensation nuclei, such as metals, which are emitted
29 from ubiquitous sources, such as automobiles.
30 On the other hand, ambient coarse particles are generated locally, and they have higher
31 deposition velocities than the fine particles. Their impact may then be limited by fallout to a
April 1995 7-56 DRAFT-DO NOT QUOTE OR CITE
-------
1 locality downwind of their emission point, as they are not readily transported across an urban
2 area. Therefore, during an air pollution episode, people living in an urban area may be
3 exposed to fine PM of similar chemical composition and concentrations, whereas they will be
4 exposed to coarse PM with a chemical composition that can depend on the location of the
5 exposure. Because PM2 5 could penetrate readily into a nonambient setting, the correlation
6 between the mean PM PEM and PM SAM would be high because all the people would have
7 similar exposure to the ambient fine PM - plus exposure to indoor generated PM and ambient
8 coarse PM which may have less fluctuation.
9 In the second type of study, such as Sexton et al. (1982) and Spengler et al. (1985),
10 there is negligible correlation between individual PEM PM and SAM PM, and consequently
11 there will be little correlation between their mean PEM and the SAM. In these cases, if the
12 fine fraction is not an appreciable portion of the total PM, or there are significant indoor
13 sources, then the correlations between mean PM PEM and PM SAM may not be as
14 impressive as for the other case.
15 7.5.3 Implications for PM and Mortality Modeling
16 PM related mortality may be specific to the most highly susceptible portion of the
17 population. Such a cohort may be the elderly people with the most serious chronic
18 obstructive lung disease (COLD) and cardiac insufficiency. Smithard (1954) relates the
19 findings of Dr. Arthur Davies (Lewisham coroner) who autopsied 44 people who died
20 suddenly during the 1952 London Fog:
21 "The great majority of deaths occurred in people who had pre-existing heart and lung
22 trouble, that is to say they were chronic bronchitic and emphysematous people with
23 consequent commencing myocardial damage. The suddenness of the deaths, Dr. Davies
24 thought, was due to a combination of anoxia and myocardial degeneration resulting in
25 acute right ventricular dilation"
26 Mage and Buckley (1995) hypothesize that these people with compromised cardio-pulmonary
27 systems may be relatively inactive, while selecting to live in homes or institutional settings
28 without sources of indoor pollution. If their time is spent in clean settings (e.g. where
29 smoking is prohibited), then they would have little exposure to PM other than from the
30 ambient pollution that intrudes into their living quarters. The exposure to PM of this cohort
April 1995 7.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 would be highly correlated with PM SAM, and so would be their mortality, if this PM was
2 indeed highly reactive in their pulmonary tracts. However, there have been no exposure
3 studies done with people who correspond to the Lewisham mortality cohort. Individual PM
4 PEM of people outside this cohort, who could be relatively insensitive to PM, might not be
5 significantly correlated with PM SAM, as reported in most of the 14 studies cited in
6 Table 7-8. This suggests a model to relate PM and mortality as follows. Let any person (j)
7 on a given day have a probability of mortality, p(m) = kj Xj, where kj is the unit probability
8 of mortality per /zg/m3 of PM per day, Xj is the daily average exposure to PM, jig/m3,
9 independent of kj. Let us assume that each individual (j) has their own personal value of kj
10 that can vary from day-to-day.
11 The expectation of total mortality (M) in a community of size N can be shown to be the
12 summation of k X over all individuals (j = 1 to N) as follows:
13 M = E kj Xj (7-3)
14 If kj is independent of Xj, then we can define K as (1/N) E kj, and the mean community
15 exposure X as (1/N) £ Xj, and it follows
16 M = N K X (7-4)
17 This implies that, given a linear relationship of mortality with PM PEM exposure (X),
18 the expected mortality is proportional to the mean community personal exposure to PM. The
19 individual in the community, on any given day, with the highest probability of dying from a
20 PM exposure related condition is that individual with the highest product kj Xj, not
21 necessarily the highest exposed individual with the maximum value of Xj.
22 The Phillipsburg, NJ, data set is a case in point. In this study, three subjects had
23 excessively high PEM PM (shown by the three maxima on Figure 7-11 a). These values
24 were caused by a hobby involving welding in a detached garage (971 /ig/m3), a home
25 remodeling activity (809 Mg/m3) and usage of an unvented kerosine heater (453 /xg/m3).
26 Excessive PM generating activities are not expected of elderly people who may have
27 compromised pulmonary systems. In fact, the elderly and infirm husband of the remodeler
28 had a personal exposure of 45 /ug/m3 on the day of the remodeling activity. The indoor
April 1995 7-58 DRAFT-DO NOT QUOTE OR CITE
-------
1 monitors in the homes of the welder and remodeler only recorded 55 jug/m3 and 19 /-ig/m3,
2 respectively, during those events, indicating the specificity of the high exposure to only the
3 individual involved. If we remove these three 'outliers', as being unrepresentative of the
4 magnitude of exposures of those nonsmoke-exposed people most at risk from high values of
5 kj, as defined by Smithard (1954), then as shown on Figures 7-15a,b the correlation R2
6 improves markedly, from 0.250 to 0.914.
7 It is this relation of the PM PEM exposure to PM SAM concentration, as shown in
8 Figures 7-15a,b that may be a better representation of the true situation underlying the PM vs
9 mortality relationships because of the "healthy worker" effect. Chronically ill people who
10 are sensitive to PM might change their behavior to minimize their exposure to irritants.
11 Consequently, healthy people with high PEM PM measures in occupations and indoor
12 settings can cause the regression R2 between PEM and SAM to be low, but they may not be
13 the individuals at highest risk of the acute effects of PM exposure.
14 7.5.4 Relative Toxicity of Ambient PM and Indoor PM
15 In the previous sections the SAM PM was evaluated as a predictor of PEM PM on the
16 implied basis that the health effects of PM were only mass dependent, and independent of
17 chemical composition. It was shown in Table 7-8 that many early PM studies of PEM had a
18 low correlation between PEM and SAM on an individual basis that was often not
19 significantly different from zero. But, in the later studies (Clayton et al., 1993; Lioy et al.,
20 1990), a significant relationship was observed between PEM and SAM. Further analysis
21 showed that on a daily basis, SAM would appear to be a good predictor of mean community
22 PEM from the results of the Riverside, CA, and Phillipsburg, NJ studies.
23 However, there may be a significant difference in toxicity of PM per unit mass, as a
24 function of source type and composition, such that some of the PM of indoor origin is less
25 toxic than the ambient PM. If so, then the SAM might be a better choice of surrogate for
26 the toxicity of an individual's exposure than the PEM which may be influenced by less toxic
27 materials. There is some indication that on a unit mass basis, combustion products of fossil
28 fuels (coal and oil) may be more acutely toxic to the pulmonary system than combustion
29 products of biomass origin (tobacco, wood). Furthermore, soil constituents and other
April 1995 7.59 DRAFT-DO NOT QUOTE- OR CITE
-------
Phillipsburg, NJ (Winter 1988)
Three 'outliers' removed (971,809, & 453)
Figure a
Figure b
300
250
I 200
150
§ 100
2
S.
50
0 50 100 150 200 250 300
Mean Outdoor PM-10, ug/m3 (4 Sites)
300
250
o 200
i
S
| 150
o
® 100
1 50
R!-0.914 (p-0.001)
m-0.561 (stdwr-0.050)
b - 38.6 (std err - 7.45)
n-14
0 50 100 150 200 250 300
Mean Outdoor PM-10, ug/m3 (4 Sites)
Figure 7-15a,b. Personal exposure to PM in Phillipsburg, NJ with concentration
outliers removed.
Source: Lioy et al. (1990).
1 nonanthropogenic materials (iron oxide, alumina, Mt. St. Helens volcanic ash) also appear to
2 be less toxic than combustion products in general. See Chapter 11 regarding comparative
3 toxicity aspects.
4 In summary, there is evidence that not all PM constituents have the same
5 toxicity per unit mass. These differences are due to differences in aerodynamic diameter and
6 chemical composition. As shown on a Venn diagram [Figure 7-16, Mage (1985)], the
7 focusing of the description of a PM exposure increases the ability to estimate the potential
8 toxicity of the exposure. In the sequential description given below, the uncertainty in the
9 toxicity of the mixture is decreased as more information is provided.
April 1995
7-60
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 7-16. Venn diagram (Mage, 1985) showing focusing of information to more
completely specify toxicity of a given PM mixture: (1) universe of all
possible mixtures of PM with concentration of 2 ftg/m3; (2) subuniverse of
all combinations of PM with concentration of 2 /tg/m3 in size interval 2.0
to 2.5 /-on; (3) subuniverse of all combinations of PM with concentration of
2 jig/m3 in size interval 2.0 to 2.5 fim AD with 50% of automotive origin
and 50% from indoor sources; and (4) subuniverse of all combinations of
PM with concentration of 2 jig/m3 in size interval 2.0 to 2.5 /tm AD with
50% of automotive origin and 50% from indoor sources; 25% Pb, 25%
BaP and 50% unspecified inorganic materials.
1
2
3
4
5
6
7
8
9
10
1. 2 />ig/m3 of TSP.
2. 2 /-tg/m3 of TSP in the size interval 2 to 2.5 /mi.
3. 2 /ig/m3 of TSP in the size interval 2 to 2.5 /mi, 50% of automotive origin and
50% of indoor source origin.
4. 2 /ig/m3 of TSP in the size interval 2 to 2.5 /mi, 50% of automotive origin and
50% of indoor source origin, 0.5 /zg/m3 of Pb, 0.5 /*g/m3 of BaP and 1 jtig/m3 of
unspecified inorganic material.
April 1995
7-61
DRAFT-DO NOT QUOTE OR CITE
-------
1 As applied to human exposure to PM, this concept of differential toxicity suggests that
2 data collections might benefit by providing data that would allow the toxicity of a PM
3 exposure to be evaluated in terms of information, in addition to the mass collected per unit
4 volume.
5
6 7.5.5 Conclusions
7 (1) For any air pollutant, the total exposure of an individual consists of a variety of
8 sequential exposures to a variety of microenvironments. They are typically,
9 outdoor, indoors at-home, at-work, in-traffic and many other indoor
10 microenvironments. The principle of superposition is a useful mechanism to
11 visualize the summation process.
12
13 (2) For any identified air pollutant, the ambient environment is one source of indoor
14 pollution due to air exchange and infiltration. Whether the ambient is a significant
15 or dominant source of indoor pollution depends on the relative strength of indoor
16 sources and sinks.
17
18 (3) For PM, studies have detected a 'personal cloud' related to the activities of an
19 individual which may generate significant levels of airborne PM in his/her vicinity
20 which may not be picked up by an indoor PM monitor at a distance.
21
22 (4) For PM, some studies have identified significant sources in the home, e.g. due to
23 cooking and smoking.
24
25 (5) For PM of size fractions that include coarse particles, some studies have identified
26 statistically significant relationships between personal exposures and other studies
27 have not, probably due to overwhelming effects of indoor sources, 'personal
28 clouds' and other individual activities.
29
30 (6) For PM of a fine size fraction - such as sulfates, there seems to be more of a
31 relationship between ambient concentration and personal exposure, than for
32 coarser PM, perhaps because of the ability of fine PM to penetrate into indoor
33 settings.
34
35 (7) For a study population in which there is a detectable correlation between personal
36 exposures and ambient concentrations, the ambient concentration can predict the
37 mean personal exposure with much less uncertainty than it can predict the personal
38 exposure of an individual.
39
40 (8) For fine PM constituents, such as sulfates, high correlations between ambient
41 concentration and personal exposures have been identified.
42
April 1995 7-62 DRAFT-DO NOT QUOTE OR CITE
-------
1 (9) For Riverside, CA, where 25% of the population was estimated to have personal
2 exposures exceeding the 24-h National Ambient Air Quality Standard for PM10 of
3 150 ^g/m3, at least 50% of this mass is of ambient origin.
4
5
6 7.6 INDOOR CONCENTRATIONS AND SOURCES OF PARTICULATE
7 MATTER
8 7.6.1 Introduction
9 Although EPA regulates particles in outdoor air, not indoors, it is still important to
10 consider indoor air. For one thing, most people spend most of their time indoors. The most
11 recent nationwide study of time budgets (Robinson and Nelson, 1995), based on interviews
12 with 9,386 respondents in 1993 to 1994, indicates that U.S. residents spend 87.2% of their
13 time indoors, 7.2% in or near a vehicle, and only 5.6% outdoors (Figure 7-17). Secondly,
14 we need to understand how outdoor particles are affected as they cross building envelopes.
15 For a home with no indoor sources, how much protection is offered against particles of
16 various size ranges? How do parameters such as the volume of the house, the air exchange
17 rate, cleaning frequency and methods, and materials in the home affect particle
18 concentrations? Indoor air studies have grappled with these topics and have the potential to
19 answer these and other important questions ultimately affecting the health of the general
20 public.
21 This section has two parts. The first part deals with field studies of particles indoors
22 and outdoors, concentrating particularly on large-scale surveys of many homes and buildings.
23 Besides presenting the observed indoor and outdoor particle concentrations, contributions of
24 these studies toward understanding important parameters such as air exchange rates, source
25 emission rates, and decay rates are also reported. This section will also discuss a few studies
26 dealing with inorganic and organic constituents of particles (e.g., elements and PAHs) as
27 well as other important considerations such as mutagenicity and the role of house dust in
28 exposure to metals and pesticides. However, each of these topics is an entire field of study
29 in itself, and can only be touched on in this section.
30 The second part of the chapter deals with indoor air quality models and the experiments
31 and chamber studies performed to validate them. A crucial parameter for particle studies is
32 the decay rate on surfaces, and a series of recent studies that have given information on this
April 1995 7.53 DRAFT-DO NOT QUOTE OR CITE
-------
69.22
2.03
Resid. (Indoors)
Resid. (Outdoor)
I" Vehicle
Near Vehicle
Other Outdoor
Office/Factory
Mall or Store
School/Public Bldg
Bar/Restaurant
Other Indoor
3.56
2.05
5.57 1-67
Figure 7-17. Percentage of time spent in different microenvironments by U.S. residents.
Source: Robinson and Nelson (1995).
1 point are reviewed. Since major modeling efforts have been aimed specifically at cigarette
2 smoking, a special section is devoted to these models.
3 In keeping with EPA's regulatory responsibilities, we omit the many studies in
4 industrial workplaces and the "dusty trades". We also omit studies whose main focus is lead
5 in indoor locations, since lead is a separate criteria pollutant and such studies are reviewed in
6 the lead criteria document. Finally, although particle concentrations indoors are of crucial
7 importance in determining the impact of radon daughters on health (smokers are at much
8 higher risk from radon than nonsmokers), we omit studies focused on radon due to the fact
9 that a different branch of EPA has regulatory authority over radon.
10
11 7.6.2 Concentrations of particles in homes and buildings
12 At least six major reviews of field studies of indoor particles have been published since
13 1980. However, all of these reviews, three of which were sponsored by tobacco companies,
April 1995
7-64
DRAFT-DO NOT QUOTE OR CITE
-------
1 were concerned with particles mainly as they relate to environmental tobacco smoke (ETS).
2 Sterling et al. (1982) reviewed studies of ETS byproducts. The National Academy of
3 Sciences (National Research Council, 1986) reviewed 16 ETS-related studies, including 8 in
4 residences and 5 in offices or buildings. Repace (1987) reviewed 13 ETS-related studies,
5 including three in residences and three in offices or hospitals. Guerin et al. (1992) reported
6 on 21 field studies of particles worldwide, including 10 in residences and 10 in offices and
7 public buildings. The U.S. Environmental Protection Agency (1992) reviewed 27 ETS-
8 related studies (24 published since 1980) including 10 in residences and 5 in offices.
9 Holcomb (1993) updated Sterling's review, including 41 studies published in the U.S. or
10 Canada since 1980, with 14 studies in homes and 20 in offices or public buildings.
11 Since the last of these reviews, several important studies have been completed,
12 including EPA's major probability-based PTEAM Study. Other large studies were not
13 included in the earlier reviews for unknown reasons. And some studies, such as the Harvard
14 6-City study, have had very recent summaries of all the data, heretofore scattered in a
15 number of publications. Therefore it is appropriate to provide a new review of all major
16 studies at this time.
17 Since the two environments where people spend the most time are home (68 to 70%:
18 Chapin 1974; Szalai et al., 1972; Robinson and Nelson, 1995) and work or school (17 to
19 20%), we will summarize the studies in these environments in turn.
20
21 7.6.2.1 Concentrations in homes
22 There have been three large-scale studies (greater than 150 homes) of airborne particles
23 inside U.S. homes. In chronological order, these are:
24 1. The Harvard 6-City study, carried out by the Harvard School of Public Health
25 beginning in 1979 and continuing through 1988, with measurements taken in
26 1,273 homes;
27
28 2. The New York State ERDA study, carried out by Research Triangle Institute in
29 433 homes in two New York State counties in 1986;
30
31 3. The EPA Particle TEAM (PTEAM) Study, carried out by Research Triangle
32 Institute and Harvard University School of Public Health in 178 homes in
33 Riverside, California in 1990.
April 1995 7.65 DRAFT-DO NOT QUOTE OR CITE
-------
1 We shall discuss the findings of each in some detail, since these studies present the
2 most complete investigations to date of indoor and outdoor concentrations of particles.
3
4 The Harvard 6-City Study
5 The Harvard 6-city study is a prospective epidemiological study of the health effects of
6 particles and sulfur oxides. Focused on children, it has included pulmonary function
7 measurements on more than 20,000 persons in the 6 cities, chosen to represent low (Portage,
8 WI and Topeka, KN), medium (Watertown, MA and Kingston-Harriman, TN), and high
9 (St. Louis, MO and Steubenville, OH) outdoor particle and sulfate concentrations.
10 The study took place in two measurement phases. The first phase involved monitoring
11 about 10 homes in each city for respirable particles (PM3 5). The homes were measured
12 every sixth day (24-h samples) for one to two years. In the second phase, a larger sample of
13 200 to 300 homes was selected from each city, with week-long PM2 5 samples collected both
14 indoors and outdoors. Two weeks of sampling in summer and in winter were provided.
15 Ultimately over 1,200 homes were monitored in this way.
16 Spengler et al. (1981) described the first 5 years of the Harvard 6-city study, During
17 that Phase I period, pulmonary function measurements were administered to 9,000 adults and
18 11,000 children in grades 1 through 6. A questionnaire asks about living conditions, type of
19 fuel and heating systems, occupation, and smoking habits of parents. Homes were selected
20 on a volunteer basis, so that no extrapolations to a wider community are warranted. In each
21 home a 24-h sample (beginning at midnight) was collected every sixth day. The cyclone
22 sampler has a cut point of about 3.5 pm at a flow rate of 1.71 Lpm. About 10 sites in each
23 city were kept in operation for two years. The annual mean indoor and outdoor RSP
24 concentrations are provided in Figure 7-18. As can be seen, the indoor concentrations
25 exceeded the outdoor levels in all cities except Steubenville, where the outdoor levels of
26 about 46 /xg/m3 slightly exceeded the indoor mean of about 43 /xg/m3. The authors noted
27 that the major source of indoor particles is cigarette smoke, and categorized their data by
28 the number of smokers in the home (Table 7-9).
29 Dockery and Spengler (198la) provided additional data analysis drawn from the same
30 6-city study but including data from 68 homes compared to the 55 reported on in Spengler
31 et al. (1981). Annual (every sixth day) mean indoor PM3 5 concentrations (in |ug/m3) were
April 1995 7-66 DRAFT-DO NOT QUOTE OR CITE
-------
II IVI
140£
100
« 80
1
3.
tt 60
S
40
20
—
—
—
(376)/r
~
X
V*
L/W
MV
jUj
X
I ywwfcy
Q
(330)
(342)
•i\
J)
X
-
X
S.
>^
pnd
highest
(183)
X
^^M *V< V^VI
( ) No. of samples
B«— Highest site mean
4 — Composite overall mean
•*— Lowest site mean
(186)
^
^
(274)
(355) n |_
(f (|3)0 0
P T K W SL S
P T K W SL S
Figure 7-18. The annual mean concentration of respirable particles (MRP) for the
highest and lowest site from the network of indoor and outdoor monitors
in each city. (P-Portage, T-Topeka, K-Kingston/Harriman, W-
Watertown, SL-St. Louis, S-Steubenville). Overall composite mean and
the number of samples are also shown.
TABLE 7-9. CONCENTRATIONS OF PARTICLES (PM2 5) IN HOMES OF
CHILDREN PARTICIPATING IN THE HARVARD SIX-CITY STUDY
Location
Indoors
No smokers
One smoker
Two or more
Outdoors
No. of homes
35
15
5
55
No. of samples
1,186
494
153
1,676
Mean (SD) (Mg/nr*)
24.4(11.6)
36.5 (14.5)
70.4 (42.9)
21.1 (11.9)
Source: Spengler et al. (1981).
1 20 and 23 in the two "clean" locations (Portage and Topeka); 31 and 36 in the two
2 "medium" locations (Watertown and Kingston-Harriman); and 39 and 47 in the two "dirty"
3 locations (Steubenville and St. Louis). Outdoor PM10 concentrations (/zg/m3) measured by
April 1995
7-67
DRAFT-DO NOT QUOTE OR CITE
-------
1 dichotomous samplers running every other day in the cities were 20.1 ± 0.6 (SE) in Portage,
2 31.9 + 1.1 in Topeka, 35.4 + 1.2 in Kingston, 41.1 ± 1.0 in Harriman, 25.8 ± 0.7 in
3 Watertown, 37.4 + 1.0 in St. Louis, and 56.6 ± 1.7 in Steubenville (Spengler and
4 Thurston, 1983). Corresponding outdoor PM2 5 concentrations were 12.5 ± 0.4,
5 12.9 + 0.4, 24.6 ± 0.8, 24.3 ± 0.7, 17.3 ± 0.5, 20.5 ± 0.5, and 36.1 ± 1.2. A mass
6 balance model allowed estimation of the impact of cigarette smoking on indoor particles.
7 Long-term mean infiltration of outdoor PM3 5 was estimated to be 70% for homes without air
8 conditioners, but only 30% for homes with air conditioners. An estimate of 0.88 ^g/m3 per
9 cigarette (24-h average) was made for homes without air conditioning, while in homes with
10 air conditioning the estimate increased to 1.23 /ig/m3 per cigarette. A residual amount of
11 15 /ig/m3 not explained by the model was attributed to indoor sources such as cooking,
12 vacuuming and dusting.
13 Letz et al. (1984) developed a model of personal exposure to particles based on
14 88 participants who wore personal monitors in the Kingston-Harriman portion of the 6-city
15 study. From the 1 to 2 years of indoor-outdoor data on 57 homes in the 6 cities, they
16 developed an equation relating indoor particle concentrations to those measured outdoors:
17 Cin = 0.385 Cout + 29.4 (Smoking) + 13.8.
18
19 Thus homes with smokers had a PM3 5 ETS component of 29.4 pig/m3. The residual of
20 13.8 /ig/m3 was assumed to be due to other household activities.
21 Neas et al. (1995) presented summary results for the entire second phase of the 6-city
22 Study (1983 to 1988). In Phase 2, a total of 1,237 homes containing white never-smoking
23 children 7 to 11 years old at enrollment completed three questionnaires and completed two
24 weeks of summer and winter monitoring indoors and outdoors for PM2 5 using the Harvard
25 impactor. At the beginning of the indoor monitoring study, 55% of the children were
26 exposed to ETS in the home, and 32% were exposed to two or more smokers. Household
27 smoking status changed for 173 children, (13% of smoking households ceased to smoke,
28 15% of the nonsmoking households became smoking households.) The annual (winter and
29 summer) household PM2 5 mean concentration for the 580 children living in consistently
30 smoking households was 48.5 + 1.4 (SE) /ig/m3 compared to 17.3 ± 0.5 /^g/m3 for the 470
31 children in consistently nonsmoking households (Figure 7-19). Among the 614 exposed
April 1995 7-68 DRAFT-DO NOT QUOTE OR CITE
-------
c
1_
-a
Never and Former Smoking Households
Mean 17.3|ig/m3
60
80 100 120
Changed Smoking Status Households
Mean 25.6 jig/m3
40
60
80 100 120
Consistantly Smoking Households
Mean 48.5 jig/m3
20 40 60 80 100
Respirable Paniculate Matter (|ig/m3)
120
Figure 7-19. Distribution of numbers of children living in households with varying
respirable particulate matter (PM2.5) as a function of parental smoking
status.
Source: Neas et al. (1995).
April 1995
7-69
DRAFT-DO NOT QUOTE OR CITE
-------
1 children for whom complete information on smoking consumption was available, 36% were
2 exposed to less than one-half pack daily, 40% to between 1/2 and 1 pack daily, and 25% to
3 > 1 pack daily. The distribution of household concentrations for children in these smoking
4 categories is shown in Figure 7-20. (The authors stated that the estimated number of
5 cigarette packs smoked daily in the home was "highly predictive" of the annual average
6 PM2 5 concentration, but did not provide the results of any tests for significance.
7 Spengler et al. (1985) reported on the Kingston-Harriman (TN) portion of the 6-city
8 study. An initial study design meant to compare exposures among individuals categorized by
9 residence (Kingston or Harriman), ETS exposure, and occupational status (office worker,
1 blue collar, or nonworking) was abandoned due to poor response rates of 30%; the sample
2 was filled out with volunteers and thus cannot be extrapolated to the population of the two
3 towns. 101 participants took part, with 28 having cigarette smoke exposure at home. Each
4 participant had an indoor and personal monitor with cutpoints of 3.5 pm. Each town had a
5 centrally located outdoor dichotomous sampler providing two size fractions, with cutpoints of
6 2.5 /im and 15 /j.m. Both towns had similar outdoor PM2 5 concentrations of 18 /*g/m3, so
7 the values were pooled for subsequent analyses. Indoor concentrations averaged 42 +
8 2.6 (SE) /ig/m3. Indoor values in homes with smoking averaged 74 ± 6.6 /ig/m3, compared
9 to 28 + 1.1 /xg/m3 in homes without smoking (p < 0.0001). No difference in exposure
10 between the unemployed and employed population was noted.
11 Lebret et al. (1987) reported on the Watertown MA portion of the Harvard 6-city
12 study. 265 homes were monitored for two one-week periods. Homes with smoking
13 averaged 54 pig/m3 (N = 147 and 152 during weeks 1 and 2), while homes without smoking
14 averaged 21.6 /ig/m3 (N = 70 and 74). The effect of smoking one cigarette/day was
15 estimated at 0.8 /ig/m3 of PM2 5.
16 Spengler et al. (1987) reported on a new round of measurements in three communities
17 within the 6-city study. In each community, about 300 children are selected to take part in a
18 year-long diary and indoor air quality study. Measurements of PM2 5 were taken indoors at
19 home for two consecutive weeks in winter and again in summer. The sampler was the
20 automated Harvard sampler, which collected an integrated sample for the week except for the
21 8 a.m. to 4 p.m. weekday period when the child was at school. During this 40-h period,
22 samples were taken in one classroom in each of the elementary schools involved. The three
April 1995 7-70 DRAFT-DO NOT QUOTE OR CITE
-------
120-
100
CO
CD
€
CD
DC
80-
CD
to 60
o
"t
ctf
0.
S 40-
20 H
90th %tile
75th %tile
50th %tile
25th %tile
10th%tile
i i
Never Changed
and Status
Former
i i i
Consistantly Smoking
Pack
1/2-1
Pack
Packs
Figure 7-20. Distribution percentiles for annual average concentrations of indoor
respirable particulate matter (PM2 5) by household smoking status and
estimated number of cigarette packs smoked in the home.
Source: Neas et al. (1995).
April 1995
7-71
DRAFT-DO NOT QUOTE OR CITE
-------
1 communities were Watertown, MA, St. Louis, MO., and Kingston-Harriman, TN. Results
2 were presented for smoking and non-smoking homes in each city by season (Figure 7-21);
3 the authors noted that mean concentrations in homes with smokers were about 30 fjLg/m"
4 greater than homes without smokers. The difference was greater in winter than in summer in
5 all cities.
6 Santanam et al. (1990) reported on a more recent and larger-scale monitoring effort in
7 Steubenville and Portage as part of the Harvard 6-city study; 140 homes in each city were
8 monitored for one week in summer and in winter. The Harvard impactor sampler was used
9 with an automatic time unit to collect PM2 5 samples between 4 p.m. and 8 a.m. on
10 weekdays and all day on weekends, corresponding to likely times of occupancy for school-
11 age children. Outdoor samples were collected from one site in each city. Elements were
12 determined by XRF. A source apportionment using principal components analysis (PCA)
13 and linear regressions on the elemental data was carried out. Cigarette smoking was the
14 single largest source, accounting for 20 to 27 /xg/m3 indoor PM2 5 in Steubenville and 10 to
15 25 /xg/m3 in Portage (Table 7-10); unfortunately, the authors do not state the number of
16 homes in the smoking and nonsmoking categories. Wood smoke was estimated to account
17 for about 4 /xg/m3 indoors and outdoors in Steubenville in winter, but only for about 1 /-ig/m3
18 indoors and outdoors in Portage. Sulfur-related sources accounted for 8 to 9 jig/m3 indoors
19 and 16 /ig/m3 outdoors in Steubenville in the summer, but were apparently not important in
20 winter. Auto-related sources accounted for 2 to 5 /xg/m3 in the two cities. Surprisingly, soil
21 sources accounted for only about 1 to 3 jwg/m3 indoor and outdoor PM2 5 concentrations.
22 Nonsmoking homes in both cities had indoor mean PM2 5 concentrations very close to the
23 outdoor mean concentrations (ratios of 1.00 and 1.04 in Steubenville, 1.02 and 1.4 in
24 Portage). Homes with smokers exceeded outdoor levels by 25 and 20 pg/m3 in Steubenville,
25 and 24 and 11 /xg/m3 in Portage.
26
27 The New York State ERDA Study
28 Sheldon et al. (1989) studied PM2 5 and other pollutants in 433 homes in two New
29 York State counties. One goal of the study was to determine the effect of kerosene heaters,
30 gas stoves, wood stoves or fireplaces, and cigarette smoking on indoor concentrations of
April 1995 7-72 DRAFT-DO NOT QUOTE OR CITE
-------
iou -
150-
140-
130-
120-
~ 110-
| 100-
3 90-
"? 80-
2 70-
Q- 60-
50-
40-
30-
20-
10-
0J
S - smoking
N - non-smoking
- - - outdoors
i
h I
1
U
•
T
S N S N S
1
"^
N
1
1-
B
^
I
[•
S N S N
•
1
S N
Winter Summer Winter Summer Winter Summer
Watertown
St. Louis Kingston
Figure 7-21. PM2-5 (ng/m3) in smoking (S) and non-smoking (N) homes in three of the
Harvard Six-City Study sites.
Source: Spongier et al. (1987).
1 combustion products. A stratified design to include all 16 combinations of the four
2 combustion sources was implemented, requiring about 22,000 telephone calls.
3 The sampler was a portable dual-nozzle impactor developed at Harvard University.
4 Two oiled impactor plates in series were employed to reduce the probability that some
5 particles larger than 2.5 ^m would reach the filter. Samples were collected in the main living
6 area and in one other room (containing a combustion source if possible) using a solenoid
7 switch to collect alternate 15-min samples over a 7-day period. Outdoor samples were
8 collected at a subset of 57 homes. All samples were collected during the winter (January to
9 April) of 1986.
10 PM2 5 mean concentrations indoors were approximately double those outdoors in both
11 counties (Table 7-11). Of the four combustion sources, only smoking created significantly
12 higher indoor PM2 5 concentrations in both counties (Table 7-12). Use of kerosene heaters
13 was associated with significantly higher concentrations in Suffolk (N = 22) but not in
April 1995
7-73
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-10a. RECONSTRUCTED SOURCE CONTRIBUTIONS TO PM2 s
MASS FOR STEUBENVILLE
Source
Soil
Wood smoke
O.C.-I
Tobacco Smoke
Sulfur-related
Auto-related
O.C.-II
Indoor dust
Unexplained
Total
TABLE
Source
Sulfur-related
Auto-related
Soil
Tobacco Smoke
Wood smoke
Unexplained
Total
Smokers'
homes
7.9 (3.45)
9.5 (4.15)
10.3 (4.47)
45.6 (19.9)
NA
NA
NA
NA
26.7(11.6)
100 (43.57)
WINTER
Non-smokers
homes
17.6 (3.45)
21.2(4.15)
22.9 (4.47)
NA
NA
NA
NA
NA
38.3 (7.47)
100 (19.54)
Outdoor
site
9.6(1.79)
23.0(4.31)
24.8 (4.65)
NA
NA
NA
NA
NA
42.6 (7.95)
100 (18.7)
7-10b. RECONSTRUCTED SOURCE
Smokers'
homes
13.2 (4.56)
5.1 (1.78)
3.8(1.31)
71.0(24.6)
2.7 (0.94)
4.2(1.38)
100 (34.6)
MASS
WINTER
Non-smokers
homes
30.7 (4.56)
12.0(1.78)
8.8(1.31)
NA
6.3 (0.94)
42.2 (6.23)
100 (14.8)
Smokers'
homes
NA
NA
NA
53.7 (26.8)
17.8 (8.90)
7.3 (3.65)
8.8 (4.40)
7.4 (3.70)
5.0 (2.4)
100 (49.85)
SUMMER
Non-
Smokers'
homes
NA
NA
NA
NA
33.3 (8.23)
14.8 (3.65)
16.5 (4.07)
15.0 (3.70)
20.4 (5.05)
100 (29.5)
Outdoor
site
NA
NA
NA
NA
52.5 (15.5)
5.3(1.55)
26.0 (7.67)
NA
16.2 (4.78)
100 (29.5)
CONTRIBUTIONS TO PM2 *
FOR PORTAGE
Outdoor
site
39.2 (4.04)
17.3 (1.78)
13.4(1.38)
NA
13.0(1.34)
17.1 (1.80)
100 (10.3)
Smokers'
homes
23.3 (5.80)
18.1 (4.50)
7.5 (1.86)
40.1 (9.99)
NA
11.0(2.75)
100 (24.9)
SUMMER
Non-
Smokers'
homes
38.1 (5.30)
29.6(4.12)
13.4(1.86)
NA
NA
18.9 (2.62)
100 (13.9)
Outdoor
site
45.8 (6.23)
35.6 (4.84)
16.5 (2.25)
NA
NA
2.10(0.28)
100 (13.6)
All figures in % (fig m-3)
O.C.-I: Iron and steel, and auto-related sources.
O.C.-II: Iron and steel, and soil sources.
Source: Santanna et al. (1990)
1 Onondaga (N = 13). Use of wood stoves/fireplaces and gas stoves did not elevate indoor
2 concentrations in either county.
3 Leaderer et al. (1990) extended the analysis of these data by collapsing the gas stove
4 category, reducing the number of categories from 16 to 8 (Table 7-13). By inspection of the
April 1995
7-74
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-11. WEIGHTED SUMMARY STATISTICS BY COUNTY FOR
RESPIRABLE SUSPENDED PARTICIPATE (PM2 5) CONCENTRATIONS (jig/m3)
Percent Detected
Sample Size
Population Estimate
Arithmetic Mean (/*g/m3)
Arithmetic Standard Error (ng/m
Geometric Mean (/ig/m3)
Geometric Standard Error
Minimum (/ig/m3)
Maximum (jtg/m3)
Percentiles
10th
16th
25th
50th (median)
75th
84th
90th
95th
99th
Main Living
Onondaga
98.9
224
94,654
36.7a
i3) 2.14
25. T
1.07
0.72
172
9.93
11.2
13.5
23.9
48.4
68.0
85.2
112
136
Area
Suffolk
99.6
209
286,580
46.4
2.77
35.9
1.06
2.18
284
13.8
16.8
18.9
33.6
62.8
76.6
89.4
112
155
Outdoors
Onondaga
100
37
16.8
1.00
15.8
1.06
6.32
28.4
12.8
15.1
20.5
Suffolk
100
20
21.8
4.54
18.6
1.11
12.0
106
13.6
16.7
22.3
Significantly different between counties at 0.05 level.
Source: Sheldon et al. 1989.
1 table, it is clear that smoking was the single most powerful source of indoor fine particles,
2 with geometric means ranging from 28.5 to 61.4 jug/m3, whereas the four nonsmoking
3 categories ranged from 14.1 to 22.0 /ig/m3.
4 Leaderer and Hammond (1991) continued their analysis of the New York State data by
5 selecting a subset of 96 homes for which both nicotine and PM2 5 data were obtained. In the
6 47 homes in which nicotine was detected (detection limit = 0.1 ^g/m3), the mean
7 concentration of RSP was 44.1 (± 25.9 SD) /*g/m3 compared to 15.2 (± 7.4) /-ig/m3 in the
8 49 homes where no nicotine was detected. Thus homes with smoking had an increased
9 weekly geometric mean PM2 5 concentration of about 29 /xg/m3. Imperfect agreement with
10 reported smoking was observed, with nicotine being measured in 13% of the residences that
11 reported no smoking, while nicotine was not detected in 28% of the residences that reported
April 1995 7-75 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-12. WEIGHTED ANALYSIS OF VARIANCE OF RESPIRABLE
SUSPENDED PARTICULATE (PM2 5) CONCENTRATIONS Gig/m3) IN THE MAIN
LIVING AREA OF HOMES VERSUS SOURCE CLASSIFICATION
F Value
Probability
Coefficient
Onondaga
Model 20.5
Independent Variables:
Intercept
Gas Stove 1.87
Kerosene Heater 1.06
Tobacco 81.6
Wood Stove/Fireplace 2.42
R2 = 0.17
0.00
0.17
0.30
0.00
0.12
20.3
5.25
5.05
45.1
7.81
Suffolk
Model 36.9
Independent Variables:
Intercept
Gas Stove 0.13
Kerosene Heater 12.0
Tobacco 114
Wood Stove/Fireplace 0.71
R2 = 0.21
0.00
0.72
0.00
0.00
0.40
26.1
-1.52
30.1
46.8
9.88
Source: Sheldon et al. 1989.
TABLE 7-13. RESPIRABLE SUSPENDED PARTICULATE CONCENTRATION
Suffolk
Onondaga
Source
None
W
K
S
KW
SW
SK
SKW
Outdoor
N
30
15
7
61
29
23
6
19
Mean
17.3
18.1
22.0
49.3
38.0
61.4
30.3
16.9
Standard
1.7
1.6
1.6
1.8
1.8
2.0
1.4
1.3
N
45
16
4
80
4
31
4
4
36
Mean
14.1
19.1
21.2
36.5
19.7
33.9
35.3
28.5
15.8
Standard
1.7
1.7
1.0
2.4
1.5
2.2
1.5
1.6
1.5
Abbreviations: W = woodstove; K = kerosene heater; S = smoking.
Source: Leaderer et al. 1990.
April 1995
7-76
DRAFT-DO NOT QUOTE OR CITE
-------
1 smoking. A regression of PM2 5 on total number of cigarettes smoked during the week (T)
2 gave the result:
3 PM2 5 = 17.7 + 0.322T (N = 96; R2 = 0.55)
4
5 For the homes with measured nicotine, the regression gave the result:
6
7 PM2 5 = 24.8 + 0.272T (N = 47; R2 = 0.40)
8
9 Thus each cigarette produces about a 0.3 (±0.03) jug/m3 increase in the weekly mean PM2 5
10 concentration, equivalent to a 2.1 (+0.2) /xg/m3 increase in the daily concentration.
11 Koutrakis et al. (1992) also analyzed the New York State data, using a mass-balance
12 model to estimate PM2 5 and elemental source strengths for cigarettes, wood burning stoves,
13 and kerosene heaters. Homes with cigar or pipe smoking and fireplace use were eliminated,
14 resulting in 178 indoor air samples. PM2 5 source strength for smoking was estimated at
15 12.7 ± 0.8 (SE) mg/cigarette; PM2 5 source strengths could not be estimated for wood
16 burning or kerosene heater usage, but only 7 homes in each category were available for
17 analysis. For a final category of all other residual indoor sources, a source strength of
18 1.16 mg/h was calculated. For nonsource homes (N = 49) the authors estimated that 60%
19 (9 Mg/m3) of the total PM2 5 mass was from outdoor sources, and 40% (6 ^g/m3) from
20 unidentified indoor sources. For smoking homes, they estimated that 54% (26 /xg/m3) of the
21 PM2.5 mass was fr°m smoking, 30% (15 ^g/m3) from outdoor sources, and 16% (8 /zg/m3)
22 from unidentified sources. These authors also developed an elemental emissions profile for
23 cigarettes, woodburning, and kerosene heaters. For cigarettes, the elemental profile included
24 potassium (160 /xg/cig), chlorine (69 /ig/cig), and sulfur (65 /ng/cig), as well as smaller
25 amounts of bromine, cadmium, vanadium, and zinc. The woodburning profile included three
26 elements: potassium (92 /xg/h), silicon (44 /xg/h) and calcium (38 /xg/h). The kerosene heater
27 profile included a major contribution from sulfur (1500 /xg/h) and fairly large inputs of
28 silicon (195 /xg/h) and potassium (164 /xg/h). A drawback of the mass-balance model was an
29 inability to separately estimate the value of the penetration coefficient P and the decay rate
30 k for particles and elements; Koutrakis et al. (1992) assumed a constant rate of 0.36 h"1 for
31 k, and then solved for P.
April 1995 7.77 DRAFT-DO NOT QUOTE OR CITE
-------
1 The EPA Particle TEAM (PTEAM) Study
2 The EPA sponsored a study of personal, indoor, and outdoor concentrations of PM10
3 particles, and indoor and outdoor concentrations of PM2 5 particles in Riverside, CA
4 (Pellizzari et al., 1993a,b; Clayton et al., 1993; Thomas et al., 1993). The personal
5 exposure results of this study are discussed in Section 7.3.3. The main goal of the study was
6 to estimate the frequency distribution of exposures to PM10 particles for all nonsmoking
7 Riverside residents aged 10 and above. 178 households were selected using probability
8 sampling to represent about 61,000 households throughout most of the City of Riverside.
9 Homes were sampled at the rate of four per day between Sept. 22 and Nov. 9, 1990. Each
10 home had two 12-h samples for both size fractions. A central site operated throughout the
11 48 days of the study, producing 96 12-h samples collected by side-by-side reference samplers
12 (dichotomous samplers and modified hi-volume samplers) along with the low-flow (4 Lpm)
13 impactor designed for this study. The impactors had very sharp cutpoints at 2.5 and 11 /*m.
14 A second filter treated with citric acid to collect nicotine was placed behind the main Teflon
15 filter.
16 A subset of the homes was monitored for PAHs (Sheldon et al., 1992). 125 homes
17 were monitored indoors and 65 of those were monitored outdoors for two consecutive 12-h
18 periods using a 20 1pm pump with an XAD cartridge.
19 Precision of the three types of particle samplers at the central site was excellent, with
20 median RSDs of about 4 to 5% (Wallace, 1991a). The low-flow sampler was noted to
21 produce estimates about 12% greater than the dichotomous sampler, which in turn was about
22 7% greater than the hi-vol sampler (Wallace, 1991b). Part of the difference may be due to
23 the different cutpoints, which are estimated to be 11 ptm for the new sampler, 9.5 for the
24 dichot, and 9.0 for the hi-vol. Part of the difference may also be due to particle bounce
25 (large particles bouncing off the impactor and being reentrained in the flow to the filter),
26 such that the PM2 5 and PM10 fractions in the low-flow sampler may be contaminated with a
27 small number of larger-size particles; however, particle bounce was found in laboratory tests
28 to account for less than 7% of the total mass.
29 The population-weighted distributions of personal, indoor, and outdoor particle
30 concentrations are provided in Table 7-14. PM10 mean concentrations (150 /*g/m3) were
31 more than 50% higher than either indoor or outdoor levels (95 pig/m3). Overnight mean
April 1995 7-78 DRAFT-DO NOT QUOTE OR CITE
-------
3.
O
TABLE 7-14. WEIGHTED DISTRIBUTIONS OF PERSONAL, INDOOR, AND
OUTDOOR3 PARTICLE CONCENTRATIONS (/tg/m3)
CD DAYTIME
<2
4j
VO
O
§>
3
H
6
o
H
Sample Size
Minimum
Maximum
Mean
(Std. Error)
Geometric Mean
(Std. Error)
Std. Deviation
Geometric Std. Deviation15
Percentiles
10th
25th
50th (median)
75th
90th
Std. Errors of Percentiles
10th
25th
50th
75th
90th
**Qtatictir»e r»tVn»r than tVi#> canrnlp ci-7P
PM
SAM
167
7.4
187.8
48.9
(3.5)
37.7
(2.5)
37.6
2.07
14.9
23.4
35.5
60.1
102.2
1.6
2.1
4.0
3.9
4.6
• minimum and ti
2.5
SIM
173
2.8
238.3
48.2
(4.1)
35.0
(3.3)
41.2
2.25
11.5
19.3
33.5
61.5
101.0
3.4
1.4
4.5
3.3
6.7
naYimiim arp i
SAM
165
16.2
506.6
94.9
(5.5)
82.7
(4.1)
57.2
1.68
42.8
56.9
84.1
110.8
157.2
2.3
4.5
4.7
4.0
7.2
ralr.nlatfirl nsin
PM10
SIM
169
16.6
512.8
94.7
(5.7)
78.2
(5-0)
61.4
1.88
30.9
49.5
81.7
127.2
180.7
3.4
4.3
8.3
9.4
11.0
IP weighted
NIGHTTIME
PM25
PEM
171
35.1
454.8
149.8
(9.2)
128.7
(8.5)
84.3
1.75
59.9
86.1
129.7
189.1
263.1
4.0
9.4
7.5
10.8
12.0
data: thev mi
SAM
161
3.4
164.2
50.5
(3.7)
37.2
(3.1)
40.3
2.23
14.5
23.0
35.0
64.9
120.7
2.1
2.7
2.4
4.6
5.8
wide estimate
SIM
166
2.9
133.3
36.2
(2.2)
26.7
(1.9)
29.5
2.21
10.0
14.8
25.9
48.9
82.7
0.9
1.3
2.4
5.3
5.8
s for the taree
SAM
162
13.6
222.9
86.3
(4.4)
74.5
(4.0)
47.7
1.74
39.3
53.6
74.1
103.7
167.8
7.4
3.4
4.8
5.1
4.3
:t population c
PM10
SIM
163
14.1
180.3
62.7
(3.2)
53.1
(3.1)
37.4
1.78
25.2
33.5
51.6
84.8
116.9
1.5
2.4
3.5
4.7
5.3
)f person-d
PEM
168
19.1
278.3
76.8
(3.5)
67.9
(3-D
39.7
1.64
36.6
48.1
66.2
98.8
135.0
1.5
3.1
4.3
8.2
10.1
ays
(PEM) or of household-days (SIM, SAM).
bln contrast to the other statistics, the gsd is a unitless quantity.
n
3
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
personal PM10 concentrations (77 /zg/m3) were similar to the indoor (63 /ig/m3) and outdoor
(86 Mg/ni3) levels. The reason for the higher daytime personal exposures is not completely
understood; it may be due
to the fact that the person is often close to the source of particles,
such as cooking, dusting, or vacuuming. It may also be due to re-entrainment of household
dust. It appears not to be
due to skin flakes or clothing fibers; many skin flakes were found
on filters but their mass does not appear to account for more than 10% of the excess personal
exposure (Mamane 1992).
Mean PM2 5 daytime
concentrations were similar indoors (48 /xg/m3) and outdoors
(49 Mg/m3), but indoor concentrations fell off during the sleeping period (36 /xg/rn3)
compared to 50 /ig/m3 outdoors. Thus the fine particle contribution to PM10 concentrations
averaged about 51% during the day and 58% at night both indoors and outdoors. The
distributions of these ratios are provided in Table 7-15.
TABLE 7-15
Sample Size
Mean
(Std. Error)
Geometric Mean
(Std. Error)
Percentiles
10th
25th
50th (median)
75th
90th
Std. Errors of Percentiles
10th
25th
50th
75th
90th
. WEIGHTED DISTRIBUTIONS3 OF PM2 5/PM10
CONCENTRATION RATIO
DAYTIME NIGHTTIME
Outdoor Indoor Outdoor Indoor
160 167 154 160
0.470 0.492 0.522 0.550
(0.016) (0.021) (0.017) (0.014)
0.444 0.455 0.497 0.517
(0.017) (0.022) (0.019) (0.016)
0.274 0.250 0.308 0.301
0.371 0.347 0.406 0.440
0.469 0.498 0.515 0.556
0.571 0.607 0.646 0.694
0.671 0.735 0.731 0.771
0.018 0.030 0.023 0.023
0.018 0.046 0.028 0.017
0.015 0.020 0.022 0.015
0.019 0.024 0.027 0.023
0.012 0.028 0.016 0.012
"Statistics other than sample size are calculated using weighted data; they provide estimates for the target
population of household-days.
April 1995
7-80
DRAFT-DO NOT QUOTE OR CITE
-------
1 Unweighted distributions are displayed in Figures 7-22 and 7-23 for 24-h average PM10
2 and PM2 5 personal, indoor, and outdoor concentrations. Most of the distributions were not
3 significantly different from log-normal distributions, as determined by a chi-square test.
4 About 25 % of the population of Riverside was estimated to have 24-h personal PM10
5 exposures exceeding the 150 /xg/m3 24-h NAAQS for ambient air.
6 The 48-day sequence of outdoor PM10 and PM2 5 concentrations is shown in
7 Figure 7-24 (Wallace et al., 199la). At least two extended episodes of high fine-particle
8 concentrations occurred. Also about 6 days of high Santa Ana winds, with correspondingly
9 high coarse-particle concentrations from desert sand, were observed.
10 Central-site PM2 5 and PM10 concentrations agreed well with back yard concentrations.
11 Pearson correlations of the log-transformed data were 0.96 and 0.92 for overnight and
12 daytime PM25, and 0.93 for the overnight PM10 values (Ozkaynak et al., 1993). The
13 correlation dropped to 0.64 for the daytime PM10 values; however on this day two homes in
14 one part of Riverside showed very high outdoor concentrations of 380 and 500 jug/m3 while
15 two homes in another part of Riverside and the central-site monitor showed more typical
16 concentrations. It is believed that a local event produced the higher concentrations at those
17 two homes. If they are removed from the data set, the correlation improves to 0.90. This
18 suggests that a single central-site monitor can represent well the PM2 5 and PM10
19 concentrations throughout a wider area such as a town or small city, at least in the Los
20 Angeles basin.
21 Daytime indoor PM10 and PM2 5 concentrations showed low-to-moderate Pearson
22 correlations of 0.46 and 0.55, respectively, with outdoor concentrations (N = 158 to 173).
23 At night, the correlations improved somewhat to 0.65 and 0.61, respectively (N = 50 to
24 168). Outdoor PM10 concentrations explained about 27% of the variance of indoor levels
25 (Figure 7-25).
26 Simple regressions of outdoor on indoor PM10 and PM2 5 resulted in the following
27 equations:
28 Indoor PM10 = 51 + 0.49 x Outdoor PM10 (day) R2 = 0.20
29 Indoor PM10 = 20 + 0.51 x Outdoor PM10 (night) R2 = 0.41
30 Indoor PM2 5 = 10 + 0.81 X Outdoor PM2 5 (day) R2 = 0.49
31 Indoor PM2 5 = 9 + 0.56 x Outdoor PM2 5 (night) R2 = 0.55
April 1995 7-81 DRAFT-DO NOT QUOTE OR CITE
-------
300
30
- o
-a- Personal
-A- Indoor
o Outdoor
J L
J L
300
25 50 75 90 95 98 99
Cumulative Frequency (%)
30
Figure 7-22. Cumulative frequency distribution of 24-h personal, indoor, and outdoor
PM10 concentrations in Riverside, CA.
200
|
o
o
20' *
25
~A~ Indoor
° Outdoor
i i i i
200
50 75 90 95 98 99
Cumulative Frequency (%)
20
Figure 7-23. Cumulative frequency distribution of 24-h indoor and outdoor PM2 5
concentrations in Riverside, CA.
April 1995
7-82
DRAFT-DO NOT QUOTE OR CITE
-------
200
150
,100
D Dichot coarse —Dichot-10
20 40 60 80
12-hour period beginning Sept. 22,1990
100
Figure 7-24. Forty eight day sequence of PM10 and PM2.5 in Riverside, CA, PTEAM
study.
600
~ 500
S 400
8
300
8
TJ
~ 200
CVJ
100
Indoor - 0.54*Outdoor + 32
R2-27% (n-309)
100 200 300 400 500
Average 12-h outdoor concentration (ng/m3)
600
Figure 7-25. Average indoor and outdoor 12-h concentrations of PM10 during the
PTEAM study in Riverside, CA.
Source: Ozkaynak et al. (1993).
April 1995
7-83
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Stepwise regressions resulted in smoking, cooking, and either air exchange rates or
house volumes being added to outdoor concentrations as significant variables (Table 7-16).
Homes with smoking added about 27 to 32 /ig/m3 to the total PM2 5 concentrations and about
29 to 37 /ig/m3 to the PM10 values. Cooking added 13 to 26 /ig/m3 to the daytime PM10
concentration and about 13 /ig/m3 to the daytime PM2 5 concentration, but was not significant
during the overnight period. At night, air exchange added about 4.5 /ig/m3 to the PM2 5
concentration per unit increase (in air changes per hour) and about 12 /ig/m3 to the PM10
concentration, but was not significant during the day. By contrast, the house volume was not
significant at night, but was significant during the day, with larger homes resulting in smaller
PM concentrations.
TABLE 7-16. STEPWISE REGRESSION RESULTS FOR INDOOR AIR
CONCENTRATIONS OF PM10, PM2 5, AND NICOTINE: COEFFICIENTS
(STANDARD ERRORS OF ESTIMATES)
PM10 PM25
Variable
N
R2
Intercept
Outdoor air
Smoking3
No. cigarettes15
Cooking0
Air exchange
House volume*1
All
310
41%
0.52
(0.05)
37
(6)
3.2
(0.7)
20
(5)
5.2
(2.0)
-0.08
(0.02)
Day
158
39%
57
(21)
0.66
(0.09)
29
(8)
3.0
(1.0)
26
(9)
-2.7
(1)
Night
147
58%
0.45
(0.05)
38
(11)
3.9
(0.9)
12
(5)
12
(5)
All
324
55%
0.64
(0.04)
28
(3.5)
2.5
(0-4)
9.4
(2.9)
Day
156
53%
21
(7.8)
0.71
(0.07)
27
(7)
2.4
(0.6)
13
(5)
-2.0
(0.6)
Night
149
71%
0.53
(0.04)
32
(10)
4.0
(0.6)
4.5
(2)
All
222
34%
1.1
(0.1)
0.11
(0.01)
Nicotine
Day
93
28%
1.1
(0.3)
0.1
(0.03)
Night
109
35%
0.28
(0.07)
1.0
(0.3)
0.2
(0.06)
All listed coefficients significantly different from zero at p < 0.05.
"Binary variable: 1 = at least 1 cigarette smoked in home during monitoring period.
bThis variable was interchanged with the smoking variable in alternate regressions to avoid colinearity problems.
cBinary variable: 1 = cooking reported for at least 1 min in home during monitoring period.
d Volume in thousands of cubic feet.
April 1995
7-84
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
21
22
A model developed by Koutrakis et al. (1992) was solved using nonlinear least squares
to estimate penetration factors, decay rates, and source strengths for particles and elements
from both size fractions in the PTEAM study. In this model, which assumes perfect
instantaneous mixing and steady-state conditions throughout each 12-h monitoring period, the
indoor concentration of particles or elements is given by
Qis/v
a. + k
(7-5)
7
8
9
10
11
12
13
14
15
16
17
18
19
20
where
Qn
P
a.
COU
Qis
V
k
Fro
smoking
expressio
= indoor concentration (ng/m3 for elements, jug/m3 for particles)
= penetration coefficient
= air exchange rate (h'1)
= outdoor concentration (ng/m3 or jig/m3)
= mass flux generated by indoor sources (ng/h or jug/h)
= volume of room or house (m3)
= decay rate due to diffusion or sedimentation (h"1)
From initial multivariate analyses, the most important indoor sources appeared to be
smoking and cooking. Therefore the indoor source term Qls was replaced by the following
Qis = (NcigScig
'other
(7-6)
where
23
24
25
26
27
28
29
30
31
t
Nc\g
sci
T
•*cook
c
°cook
Gother
duration of the monitoring period (h)
number of cigarettes smoked during monitoring period
mass of elements or particles generated per cigarette smoked (ng/cig or
Mg/cig)
time spent cooking (min) during monitoring period
mass of elements or particles generated per hour of cooking (ng/min or
/zg/min)
mass flux of elements or particles from all other indoor sources (ng/h or
April 1995
7-85
DRAFT-DO NOT QUOTE OR CITE
-------
1 With these changes, the equation for the indoor concentration due to these indoor
2 sources becomes
3
_ t NcigScig + TcookScook Qother
in
a+k (a + k)Vt (a + k)V
4
5 The indoor and outdoor concentrations, number of cigarettes smoked, monitoring
6 duration, time spent cooking, house volumes, and air exchange rates were all measured or
7 recorded. The penetration factor, decay rates, and source strengths for smoking, cooking,
8 and all other indoor sources (2other) were estimated using a nonlinear model (NLIN in SAS
9 software). The Gauss-Newton approximation technique was chosen to regress the residuals
10 onto the partial derivatives of the model with respect to the unknown parameters until the
11 estimates converge. On the first run, the penetration coefficients were allowed to "float"
12 (no requirement was made that they be < 1). Since nearly all coefficients came out close to
13 one, a second run was made bounding them from above by one. The NLIN program
14 provides statistical uncertainties (upper and lower 95 % confidence intervals) for all parameter
15 estimates. However, it should be noted that these uncertainties assume perfect measurements
16 and are therefore underestimates of the true uncertainties.
17 Results are presented in Table 7-17 for the combined day and night samples.
18 Penetration factors are very close to unity for nearly all particles and elements. The
19 calculated decay rate for fine particles is 0.39 ± 0.16 h'1, and for PM10 is 0.65 + 0.28 h'1.
20 Since PM10 contains the PM2 5 fraction, a separate calculation was made for the coarse
21 particles (PM10 to PM2 5) with a resulting decay rate of 1.0 h"1. Each cigarette emits 22 + 8
22 mg of PM10 on average, about two-thirds of which (14 ± 4 mg) is in the fine fraction.
23 Cooking emits 4.1 ± 1.6 mg/min of inhalable particles, of which about 40% (1.7 + 0.6
24 mg/min) is in the fine fraction. All elements emitted by cooking were limited almost
25 completely to the coarse fraction. Sources other than cooking and smoking emit about 5.6 ±
26 3.1 mg/h of PM10, of which only about 1.1 mg/h ± 1.0 (20%) is in the fine fraction.
27 Similar calculations were carried out for day and night samples separately, and for the
28 logarithms of the data as well as the untransformed data, to obtain more information on the
29 sensitivity of the estimates. For PM10, the lowest estimate of the source strength of the
April 1995 7-86 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-17. PENETRATION FACTORS, DECAY RATES, AND SOURCE STRENGTHS: NONLINEAR ESTIMATES
Decay Rate (1/h)
S cook (fig/rain)
S_smoke (/ig/cig)
i— VAR
tS PM2 5*
Al
Mn
Br
Pb
Ti
Cu
Sr
P
Si
Ca
Fe
K
S
Zn
Cl
PM10"
Al
^nj
i Mn
oo ,,
•Cj Br
Pb
Ti
Cu
0 Sr
53 P
>> Si
hrl
Ca
6 ?
O IT
s
o ^f
2 C1
^ aMass units in
mean
Too
1.00
0.87
0.90
1.00
0.97
1.00
0.98
1.00
1.00
1.00
1.00
0.71
0.50
1.00
1.00
1.00
1.00
1.00
1.00
0.83
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.94
195
0~89
0.95
0.78
0.81
0.56
0.93
0.98
0.75
0.65
0.76
0.81
0.97
0.57
0.28
0.85
0.80
0.80
0.90
0.89
0.80
0.62
0.83
0.81
0.68
0.80
0.83
0.96
0.81
0.44
u95 mean
1.11 0.39
1.05 0.03
0.95 0.23
0.99 0.28
fail to converge
fail to converge
1.44 1.63
1.01 0.07
1 .02 0.04
1 .20 0.54
1.35 0.61
1.24 0.70
1.19 0.16
1.03 0.16
0.86 0.78
0.72 0.64
1.15 0.65
1.20 0.80
1 .20 0.69
1.10 0.21
1.11 0.14
1.20 0.60
1 .05 0.77
1.16 0.62
fail to converge
1.19 0.62
1.32 0.63
1.20 0.66
1.17 0.46
1.04 0.21
1.19 0.37
1.43 2.36
195"
0.22
-0.03
0.07
0.15
0.38
0.01
0.02
0.04
-0.02
0.11
-0.04
0.12
0.31
0.05
0.36
0.38
0.30
0.11
0.01
0.22
0.18
0.28
0.26
0.06
0.26
0.17
0.17
0.10
0.48
u95
0.55
0.09
0.38
0.41
2.88
0.12
0.06
1.05
1.25
1.29
0.37
0.19
1.25
1.24
0.93
1.21
1.07
0.32
0.26
0.98
1.36
0.97
0.97
1.20
1.06
0.75
0.26
0.64
4.24
mean 195b
1.7 1.0
0.9 -1.4
0.1 -0.1
0.1 0.0
0.6 0.0
0.0 0.0
0.1 -0.1
6.1 -8.6
11.9 -0.6
4.5 -3.3
0.0 -4.4
1.0 -3.9
0.4 -0.5
5.9 0.1
4.1 2.6
69.5 16.6
0.9 0.1
0.1 0.0
0.0 -0.3
4.0 0.3
0.5 0.0
0.3 0.0
149.3 26.9
118.7 37.3
46.7 8.5
17.6 0.1
6.8 -0.7
1.2 -0.2
45.7 17.6
u95
2.3
3.1
0.2
0.2
1.2
0.0
0.3
20.9
24.4
12.3
4.4
5.9
1.2
11.6
5.7
122.4
1.7
0.3
0.3
7.8
1.1
0.5
271.8
200.1
84.8
35.2
14.3
2.5
73.9
mean
13.8
9.0
0.2
1.9
3.7
0.1
2.0
14.4
165.6
23.8
121.3
27.1
2.9
102.6
21.9
97.6
1.1
1.8
2.1
10.0
3.5
2.6
296.4
800.0
73.0
215.7
68.0
4.0
320.2
195"
10.2
-2.5
-0.4
1.3
0.2
-0.1
1.1
-58.3
72.0
-16.3
85.7
2.4
-1.5
54.0
13.6
-159.0
-2.7
1.2
0.4
-8.4
0.4
1.2
-293.9
329.0
-109.8
116.9
29.3
-3.0
107.0
u95
17.3
20.5
0.8
2.5
7.2
0.2
2.9
87.2
259.1
63.9
156.9
51.7
7.4
151.2
30.2
354.2
4.9
2.5
3.9
28.4
6.5
3.9
886.6
1271.0
255.9
314.5
106.7
11.0
533.4
mean
1.1
3.0
0.5
0.6
3.8
0.1
0.8
57.3
34.1
23.8
8.9
4.0
7.5
20.6
5.6
154.5
1.2
0.4
0.0
10.3
3.2
0.9
237.8
107.6
51.5
43.6
22.7
7.4
148.4
195b u95
0.0 2.1
-3.7 9.8
0.2 0.9
0.3 0.9
1.4 6.3
0.0 0.2
0.2 1.3
12.5 102.0
3.4 64.8
1.8 45.7
-0.5 18.3
-3.7 11.7
4.2 10.9
7.2 34.0
2.6 8.7
52.0 257.0
-0.2 2.6
0.1 0.6
-0.6 0.6
2.6 18.1
1.3 5.1
0.3 1.5
16.1 459.6
-27.0 242.3
-15.5 118.5
8.6 78.5
10.4 34.9
3.4 11.4
49.4 247.4
mg for PM2 5 and PM10 only.
O bA negative lower confidence
H
6
n
H
W
interval implies a nonzero mean is not statistically
significant.
-------
1 cigarettes was 9 + 4 mg/cig and the highest estimate was 29 + 7 mg/cig. The estimates for
2 cooking ranged from a low of 1.5 ± 0.6 mg/min to a high of 4.9 ± 1.3 mg/min. The
3 estimate for other sources ranged from 2.5 + 0.9 mg/h to 12 ±4 mg/h.
4 Decay rates for elements associated with the fine fraction were generally lower than for
5 elements associated with the coarse fraction, as would be expected. For example, sulfur,
6 which has the lowest mass median diameter of all the elements, had calculated decay rates of
7 0.16 ± 0.04 and 0.21 + 0.04 h"1 for the PM2 5 and PM10 fractions, respectively. The
8 crustal elements (Ca, Al, Mn, Fe), on the other hand, had decay rates ranging from 0.6 to
9 0.8 h4.
10 Based on the mass-balance model, outdoor air was the major source of indoor particles,
11 providing about 3/4 of fine particles and 2/3 of inhalable particles in the average home.
12 It was also the major source for most elements, providing 70 to 100% of the observed indoor
13 concentrations for 12 of the 15 elements. Only copper and chlorine were predominantly due
14 to indoor sources in both the fine particle and inhalable particle fractions. It should be noted
15 that these conclusions are applicable only to Riverside, CA. In five of the six cities studied
16 by Harvard and in both New York counties, outdoor air could not have provided as much as
17 half of the indoor air particle mass, because the observed indoor-outdoor ratios of the mean
18 concentrations were > 2.
19 Unidentified indoor sources accounted for most of the remaining particle and elemental
20 mass collected on the indoor monitors. The nature of these sources is not yet understood.
21 They do not include smoking, other combustion sources, cooking, dusting, vacuuming,
22 spraying, or cleaning, since all these sources together account for less than the unidentified
23 sources. For example, the unidentified sources accounted for 26% of the average indoor
24 PMio particles, whereas smoking accounted for 4% and cooking for 5% (Figure 7-26).
25 Of the identified indoor sources, the two most important were smoking and cooking
26 (Figures 7-27 and 7-28). Smoking was estimated to increase 12-h average indoor
27 concentrations of PM10 and PM2 5 by 2 to 4 ^g/m3 per cigarette, respectively. Homes with
28 smokers averaged about 30 /xg/m3 higher levels of PM10 than homes without smokers. Most
29 of this increase was in the fine fraction. Cooking increased indoor concentrations of PM10
30 by about 0.6 /*g/m3 per minute of cooking, with most of the increase in the coarse particles.
April 1995 7-88 DRAFT-DO NOT QUOTE OR CITE
-------
Cooking
4%
Other Indoor
14%
Smoking
5%
Outdoor
76%
N - 352 Samples from 178 homes
Outdoor
66%
Cooking
5%
Other Indoor
26%
Smoking
4%
N - 350 Samples from 178 homes
Figure 7-26. Sources of fine particles (PM2 s) and respirable particles (PM10) in all
homes (Riverside, CA).
April 1995
7-89 DRAFT-DO NOT QUOTE OR CITE
-------
Outdoor
60%
Indoor
Smoking
30%
N - 61 Samples from 31 homes
Cooking
3%
Outdoor
56%
Other Indoor
16%
Smoking
24%
N - 61 Samples from 31 homes
Figure 7-27. Sources of fine particles (PM2-5) and respirable particles (PM10) in homes
with smokers (Riverside, CA).
April 1995 7-90 DRAFT-DO NOT QUOTE OR CITE
-------
Cooking
Other Indoor
^ , . , 8%
Outdoor
62%
Smoking
5%
N = 62 Samples from 33 homes
Cooking
Outdoor
56%
\ V /
Other Indoor
16%
Smoking
4%
N = 62 Samples from 33 homes
Figure 7-28. Sources of fine particles (PM2 5) and respirable particles (PM10), top and
bottom panels respectively, for homes with cooking during data collection
(Riverside, CA).
Source: Ozykaynak et al. (1993).
April 1995 7-91 DRAFT-DO NOT QUOTE OR CITE
-------
1 Emission profiles for elements were obtained for smoking and for cooking. Major
2 elements emitted by cigarettes were potassium, chlorine, and calcium. Elements associated
3 with cooking included aluminum, iron, calcium, and chlorine.
4 Other household activities such as vacuuming and dusting appeared to make smaller
5 contributions to indoor particle levels. An interesting finding was that commuting and
6 working outside the home resulted in lower particle exposures than for persons staying at
7 home.
8 As with the particle mass, daytime personal exposures to 14 of 15 elements were
9 consistently higher than either indoor or outdoor concentrations. At night, levels of the
10 elements were similar in all three types of samples.
11
12 Comparison of the three large-scale studies
13 The three studies had somewhat different aims and therefore different study designs.
14 The Harvard study selected homes based on various criteria, in particular the requirement
15 that a school-age child be in the home, but did not employ a probability-based study.
16 Therefore the results strictly apply only to the homes in the sample and not to a wider
17 population; however, the very large number of homes suggests that the results should be
18 broadly applicable to homes with school-age children in the six cities. The New York study
19 used a probability-based sample, but stratified on the basis of combustion sources. Therefore
20 there are likely to be a higher number of homes with kerosene heaters, wood stoves, and
21 fireplaces in the sample than in the general population. The PTEAM study used a fully
22 probability-based procedure, and therefore its results are the most broadly applicable to the
23 entire population of Riverside households. However, the participants were limited to
24 nonsmokers, and therefore homes with only smokers were excluded; therefore indoor
25 concentrations are likely to be slightly underestimated.
26 The three studies employed different monitors with different cutpoints; therefore exact
27 comparisons are not possible. However, large differences between the PM3 5 and PM2 5
28 cutpoints are not likely, and therefore these results can be more readily compared. In what
29 follows, we will use the term "fine particles" to refer to the PM3 5 and PM2 5 size fractions
30 collected in the three studies.
31
April 1995 7-92 DRAFT-DO NOT QUOTE OR CITE
-------
1 Indoor-outdoor relationships. Outdoor concentrations of fine particles in five of the
2 six cities and the two New York counties were relatively low, typically in the range of 15 to
3 25 /-ig/m3 (Table 7-18). Only Steubenville, with an annual mean of 40 /xg/m3 (but a range
4 among the outdoor sites of 20 to 60 /Ltg/m3) approached the mean outdoor level of 50 jug/m3
5 observed in Riverside. It is interesting to note that indoor concentrations exceeded outdoor
6 concentrations in the seven sites with low outdoor levels, (indoor/outdoor ratios were
7 contained in a small range between 1.9 and 2.4), but were slightly less than outdoor
8 concentrations in the two sites with high
9
10
outdoor levels (ratios of 0.9).
TABLE 7-18. INDOOR-OUTDOOR MEAN CONCENTRATIONS (jtg/m3) OF FINE
PARTICLES IN THREE LARGE-SCALE STUDIES
Study Name
Harvard 6-City Study
Portage, WI
Topeka, KN
Kingston-Harriman, TN
Watertown, MA
St. Louis, MO
Steubenville, OH
New York State ERDA Study
Onondaga County
Suffolk County
EPA Particle TEAM Study
Riverside, CA
Homes Out In
11 10 20
10 10 22
8 18 44
8 15 29
10 18 42
8 45 42
224 17 37
209 22 46
178 50 43
In/Out
2.0
2.2
2.4
1.9
2.3
0.9
2.2
2.1
0.9
Harvard: PM3 5 measured using cyclone sampler. Samples collected every sixth day for one year (May 1986
to April 1987).
NYS: PM2 5 measured using impactor developed at Harvard. Samples collected for one week at each
household between January and April 1986.
PTEAM: PM2 5 measured using Marple-Harvard-EPA sampler. Samples collected for two 12-h periods at each
home between September and November 1990.
Source: Harvard data—Spengler et al., 1981.
NYS data—Sheldon et al., 1989.
PTEAM data—Pellizzari et al., 1993.
1 Effect of smoking. All three studies found cigarette smoking to be a major source of
2 indoor fine particles. All three studies compared fine particle concentrations in homes with
April 1995 7.93 DRAFT-DO NOT QUOTE OR CITE
-------
1 smokers to those in homes without smokers. The annual mean (every sixth day)
2 concentration in the 35 six-city homes with no smokers was 24.4 /xg/m3, compared to a
3 mean of 36.5 /ig/m3 in 15 homes with one smoker and a mean of 70 /-tg/m3 in five homes
4 with two or more smokers. Neas et al. (1990) derived a difference of 31.2 /ug/m3 in
5 smoking homes compared to nonsmoking homes based on all 1237 homes in Phase II of the
6 6-City Study. In the New York State study, 62 Onondaga homes with smokers had a mean
7 indoor PM2 5 concentration of 69 jiig/m3 compared to 24 jug/m3 in 182 nonsmoking homes;
8 the indoor value in 80 Suffolk homes with smokers was 76 /xg/m3 compared to 30 jug/m3 in
9 129 homes without smokers. Thus homes with smokers had PM2 5 weekly mean increases of
10 about 45 to 46 /*g/m3. A similar calculation was done for the PTEAM Study, resulting in a
11 mean of 36 /xg/m3 in 119 homes with no smokers, 63 /xg/m3 in 18 homes with one smoker
12 and at least one cigarette smoked during the 24-h monitoring period, and 69 ptg/m3 in 15
13 homes with two or more smokers and at least one cigarette smoked during the monitoring
14 period. The latter result is somewhat muddied by the fact that outdoor concentrations were
15 somewhat higher at the homes with smokers (about 60 /xg/m3 in 33 homes with smokers and
16 only 49 ^g/m3 in 119 homes without smokers). Thus homes with smokers were about 20
17 ng/m3 higher (after allowing for higher outdoor concentrations) than homes without smokers.
18 A similar correction for outdoor concentrations cannot be made in the New York State study,
19 due to the lack of sufficient matched outdoor measurements. These bivariate calculations are
20 not highly trustworthy, since they assume that all other sources affecting indoor air are
21 independent of smoking status, which may not be the case. For example, smokers are more
22 likely to be found in the lower socioeconomic brackets. Therefore they may have smaller
23 homes, which would lead to higher indoor concentrations in general. They may also be
24 located in less desirable (more polluted) parts of town, which would lead to higher outdoor
25 concentrations, as was observed in the PTEAM Study. This would lead to overestimates of
26 the effect of smoking on indoor concentrations. On the other hand, persons in homes with
27 smokers may open windows or otherwise ventilate more extensively to reduce perceived
28 smoke or particle loadings. This would lead to underestimates of the effect of smoking.
29 Therefore multivariate regressions accounting for the effect of volume, air exchange, and
30 outdoor concentrations are more trustworthy than the above bivariate calculations.
April 1995 7-94 DRAFT-DO NOT QUOTE OR CITE
-------
1 Multivariate calculations in all three studies result in rather similar estimates of the
2 effect of smoking on fine particle concentrations. Spengler et al. (1981) estimated an
3 increase of about 20 /xg/m3 per smoker based on 55 homes from all six cities. Since homes
4 with at least one smoker probably average at least 1.3 smokers per home, this corresponds to
5 about 26 /xg/m3 per smoking home. Spengler et al. (1985) found a smoking effect of about
6 32 /xg/m3 for smoking homes in multivariate models based on the Kingston-Harriman data.
7 Sheldon et al. (1989) found an increase of 45 (Onondaga) and 47 (Suffolk) /xg/m3 per
8 smoking home in a multivariate model of the New York State data. Ozkaynak et al. (1993)
9 found an increase of about 30 to 35 /xg/m3 in smoking homes in a multivariate regression
10 model of the PTEAM data. Thus the estimates of the effect of a smoking home on indoor
11 fine particle concentrations range from about 26 to about 47 /xg/m3.
12 Dockery and Spengler (1981) found an effect of 0.88 /xg/m3 per cigarette for homes
13 without air conditioning, and 1.23 /xg/m3 per cigarette for homes with air conditioning, based
14 on 68 homes from all six cities. Lebret found an effect of 0.8 /xg/m3 per cigarette for homes
15 in the Watertown area. Leaderer et al. (1991) found an effect ranging between 0.27 and
16 0.33 /xg/m3 per cigarette smoked over a week's time, corresponding to 1.9 to 2.3 /xg/m3 per
17 cigarette contribution to a 24-h average. In a series of stepwise regressions on the PTEAM
18 data, Ozkaynak et al. (1993) found an effect ranging between 2.5 and 4.7 /xg/m3 per cigarette
19 per 12-h monitoring period; this corresponds to an effect ranging between 1.2 and 2.4 /xg/m3
20 per cigarette smoked during a 24-h period. Taking the midpoint of these ranges leads to
21 estimates for the three studies of about 1.1, 2.1, and 1.8 /xg/m3 increases in fine particle
22 concentrations per cigarette smoked in the home over a 24-h period.
23 Both the New York State study and the PTEAM study were able to estimate source
24 strengths for different variables using a mass-balance model. The estimates for PM2 5
25 emissions from cigarettes were very comparable, with Koutrakis et al. (1992) estimating
26 12.7 mg/cig compared to the PTEAM estimate of 13.8 mg/cig (Ozkaynak et al., 1993).
27 Both studies also found similar elemental profiles for smoking, with potassium and chlorine
28 being emitted in substantial amounts.
29
30 Effect of other variables. In the PTEAM Study, the second most powerful indoor
31 source of PM10, and possibly PM2 5 particles, was cooking. Quite large emission strengths
April 1995 7.95 DRAFT-DO NOT QUOTE OR CITE
-------
1 of several mg/minute of cooking were determined from the mass-balance model, while
2 multiple regressions indicated that cooking could contribute between 10 and 20 jwg/m3 PM10,
3 and somewhat smaller amounts of PM2 5, to the 12-h concentration. Apparently neither of
4 the other studies analyzed for the effect of cooking, although Dockery and Spengler (1981)
5 suggested cooking as one possible reason for the observed 15 /ig/m3 residual in their indoor
6 particle model.
7 In the New York State study, homes with kerosene heaters had increased PM2 5
8 concentrations of 15.8 /xg/m3 in Suffolk (p < 0.03) and 3.65 ^g/m3 in Onondaga (p < 0.3,
9 not significant). Both the New York State and PTEAM studies also measured air exchange
10 in every home, and both studies found that air exchange significantly affected indoor particle
11 concentrations. In the PTEAM study, increased air exchange led to increased indoor air
12 concentrations for both PM2 5 and PM10 at night only, perhaps because outdoor
13 concentrations were larger than indoor levels at night. In the New York State study,
14 increased air exchange led to decreased RSP concentrations in Onondaga (p < 0.02) but no
15 effect was noted in Suffolk (p < 0.90). In both of these counties, indoor levels generally
16 exceeded outdoor levels, so increased air exchange would generally reduce indoor
17 concentrations.
18 Both the New York State and PTEAM studies found a very small but significant effect
19 of house volume. In Onondaga, PM25 concentrations decreased by —1.1 /zg/m3 per
20 1,000 cubic foot increase in volume; in Suffolk concentrations decreased by —0.75 /-ig/m3
21 per 1,000 cubic feet. In the PTEAM study, PM2 5 daytime concentrations decreased by
22 -2.0 /xg/m3 per thousand cubic feet. Probably because of a significant negative correlation
23 between house volume and air exchange rate, the two variables did not both reach
24 significance in the same regression.
25
26 Other Studies
27 Several other large-scale studies of homes have taken place in other countries, and a
28 number of smaller studies have occurred in the U.S. These will be discussed in order of the
29 number of homes included in the study.
30 Lebret et al. (1990) carried out week-long RSP measurements (cutpoint not described)
31 in 260 homes in the cities of Ede and Rotterdam, the Netherlands, during the winters of 1981
April 1995 7-96 DRAFT-DO NOT QUOTE OR CITE
-------
1 to 1982 and 1982 to 1983, respectively. 60% of the Ede homes and 66% of the Rotterdam
2 homes included smokers. Diary information collected during the measurement period
3 indicated that, on average, 1 to 2 cigarettes were smoked during the week, presumably by
4 guests, even in the nonsmoking homes. Homes with one smoker averaged 7 cigarettes
5 smoked per day at home in Ede (N = 53) and 11 per day in Rotterdam (N = 35). Homes
6 with two smokers averaged 21 cigarettes per day in Ede (N = 23) and 25 cigarettes per day
7 in Rotterdam (N = 15).
8 Geometric means for the combined smoking and nonsmoking homes were similar in the
9 two cities (61 and 56 jug/m3, respectively), with maxima of 560 and 362 /-ig/m3. Outdoor
10 concentrations averaged about 45 /ig/m3 (N not given). Indoor concentrations in the homes
11 with smokers averaged about 70 /*g/m3 (calculated from data in the paper), compared to
12 levels in the nonsmoking homes of about 30 /xg/m3. Multiple regression analysis indicated
13 that the number of smoking occupants explained about 40% of the variation in the log-
14 transformed RSP concentrations—family size, frequency of vacuuming, volume of the living
15 room, type of space heating, and city (Ede versus Rotterdam) had no significant effect on
16 RSP concentrations. In a second regression, the number of smoking occupants was replaced
17 by the number of cigarettes and cigars smoked during the week. The regression equation
18 was
19
20 log(RSP) =1.4 + 0.37 log(# cigarettes) + 0.53 log(# cigars)
21 +0.03 log(family size)
22 R2 = 0.49; d.f. =250 F = 83.7 p < 0.0001
23
24 From this equation, the authors estimated that one cigarette smoked per day would increase
25 weekly average indoor RSP concentrations by 2 to 5 jig/m3, whereas one cigar smoked per
26 day would increase indoor levels by 10 pig/m3.
27 Instantaneous RSP concentrations were made using a TSI Piezobalance on the day the
28 technicians were setting up the equipment. The influence of smoking on these measurements
29 may be seen in Table 7-19:
30
April 1995 7-97 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-19. INFLUENCE OF RECENT CIGARETTE SMOKING ON INDOOR
CONCENTRATIONS OF PM (SIZE UNSPECIFIED)
Time Since Smoking N RSP (geom. mean)
No smoking 98 41
More than 1 h ago 18 52
Between 1/2 and 1 h ago 7 76
Less than 1/2 an hour ago 27 141
During the measurements 54 191
Source: Lebret et al. (1990).
1 Diemel et al. (1981) measured particles in 101 residences as part of an epidemiological
2 study related to a lead smelter in Arnhem, the Netherlands. The indoor sampler was adapted
3 from a small aquarium-type pump, collecting samples at a flowrate of 1 to 1.5 1pm. The
4 authors state that particles smaller than 3 to 4 jum in diameter should have been sampled
5 efficiently, but present no laboratory data on measured cutpoint size. The outdoor samplers
6 (number not given) were high-volume samplers. The 28-day average levels indoors ranged
7 from 20 to 570 /*g/m3, with an arithmetic mean of 140 jig/m3 (SD not presented) and a
8 geometric mean of 120 ng/m3; corresponding outdoor concentrations (2-mo averages of 24-h
9 daily samples) ranged from 53.7 to 73.3 /ig/m3 (N not given), with nearly identical
10 arithmetic and geometric means of 64 /xg/m3.
11 Kulmala et al. (1987) measured indoor and outdoor air in approximately 100 dwellings
12 (including some office buildings) in Helsinki, Finland between 1983 and 1986. Samples
13 were collected on Nuclepore filters using a stacked foil technique. Mean concentrations were
14 presented in Table 7-20 for fine (< 1 /xm) and coarse (> 1 /mi) particles (standard deviations
15 not provided):
16
17
18
April 1995 7-98 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-20 INDOOR AND OUTDOOR PM IN BUILDINGS IN HELSINKI,
FINLAND, AS A FIUNCTION OF SEASON AND LOCATION.
Fine particles
Coarse particles
Location
Urban
Urban
Suburban
Suburban
Urban
Urban
Suburban
Suburban
Season
Summer
Winter
Summer
Winter
Summer
Winter
Summer
Winter
Outdoors
40
30
15
30
75
30
20
25
Indoors
25
30
20
20
20
15
25
20
Source: Kulmala et al. (1987).
1 The authors noted that the geometric mean for the combined fine particle samples indoors
2 was 16 ^g/m3, with a 95% range of 4 to 67 /xg/m3. Corresponding values for the indoor
3 coarse particles were 13 ^ig/m3 with a range of 3 to 63 ^g/m3. Outdoors, the fine particles
4 had a geometric mean of 20 /zg/m3 with a 95% range of 5 to 82 jug/m3, and the coarse
5 particles had a geometric mean of 16 /xg/m3 with a range of 3 to 91 /ig/m3.
6 Quackenboss et al. (1989) reported PM10 and PM2 5 results from 98 homes in the
7 Tucson, Arizona area. Homes were selected as part of a nested design for an
8 epidemiological study. The Harvard-designed dual-nozzle indoor air sampler (Marple et al.,
9 1987) was employed for indoor air measurements. Outdoor air was measured within each
10 geographic cluster by the same instrument; supplementary data were obtained from the Pima
11 County Air Quality control District, but these data did not include PM2 5 measurements, and
12 some data were apparently PM15. Homes were classified by tobacco smoking and by use of
13 evaporative ("swamp") coolers; these coolers apparently act as a significant removal
14 mechanism for particles (Table 7-21). Homes without smoking averaged about 16 /xg/m3
15 PM2 5, compared to 24 iig/m3 for homes reporting less than a pack a day, and 51 iig/m3 for
16 homes reporting more than a pack a day. PM2 5 particles formed the bulk of the PM10
17 fraction even in nonsmoking homes, ranging from nearly 80% in those homes to nearly
18 90% in homes with heavy smoking. This is somewhat surprising in view of the statement
19 made in the paper that the bulk of the PM in Tucson is silica quartz averaging around 5 /xm
April 1995 7.99 DRAFT-DO NOT QUOTE OH CITE
-------
TABLE 7-21. INDOOR AVERAGE PM2 5 AND PM10 (/tg/m3) BY REPORTED
SMOKING IN THE HOME AND EVAPORATIVE COOLER USE
DURING SAMPLING WEEK
Smoking
Cigarettes/day
None
1-20
>20
Evaporative
Cooler
Yes
No
Total
Yes
No
Total
Yes
No
Total
Mean
8.8
20.3
15.2
19.3
32.3
27.3
36.2
82.7
60.8
PM2.5
S.D.
5.0
19.0
15.5
8.8
28.5
23.6
32.9
55.4
50.8
Homes
(20)
(25)
(45)
(10)
(16)
(26)
(8)
(9)
(17)
Mean
21.0
38.4
30.3
33.9
53.4
46.2
47.4
102.5
75.0
PM10
S.D.
9.7
22.9
19.9
12.0
33.9
29.1
39.6
60.6
57.2
Homes
(20)
(23)
(43)
(10)
(17)
(27)
(9)
(9)
(18)
PM2 5: Significant (p < 0.01) main effects for smoking and evaporative cooler use; two-way interaction nearly
significant (p = 0.06).
PM10: Significant (p < 0.01) main effects for evaporative cooler and smoking.
Source: Quackenboss et al. (1989)
1 in mean aerodynamic diameter; that should result in the coarse fraction being larger than the
2 fine fraction. Outdoor PM10 particles were not strongly correlated with indoor levels
3 (R2 = 0.18; N about 90). Although about 30% of homes in the larger sample reported using
4 fireplaces, no estimate of an effect on indoor air quality was made in the paper.
5 Quackenboss et al. (1991) extended the analysis of the Tucson homes over three
6 seasons. Median indoor PM2 5 levels in homes with smokers were about 20 /xg/m3 in the
7 summer and spring/fall seasons compared to about 10 /zg/m3 in homes without smokers in
8 those seasons (Table 7-22). In winter, however, the difference was considerably increased,
9 with the median level in 24 homes with smokers at about 36 /ig/m3 compared to 13 /zg/m3 in
10 26 homes without smokers.
11 Revsbech et al. (1987) studied 44 apartments in Aarhus, Denmark. All were retrofitted
12 or tight apartments in three-story brick buildings. Particles were measured by an open filter
13 surface directed downwards. The authors considered that the low suction rate and the
14 downward facing surface would result in collecting predominantly the respirable fraction, but
April 1995 7-100 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-22. INDOOR PM10 AND PM2 5 (jig/m3) BY SEASON AND
ENVIRONMENTAL TOBACCO SMOKE (ETS)
Smokers at home
PM10
PM2.5
Season3
Summer
Spring/Fall
Winter
Summer
Spring/Fall
Winter
Median
35.
43.
80.
20.
20.
35.
3
3
3
5
1
7
Qi
29.3
28.7
40.8
13.7
12.8
26.7
Q3
49.8
64.3
104.9
27.0
43.6
77.8
N
49
38
24
49
39
24
No smokers at home
Median
17
30
31
8
10
13
.5
.0
.4
.9
.6
.4
Qi
14.1
20.5
24.3
5.9
8.7
10.2
Q3
24.4
40.1
42.9
11.9
14.8
19.9
N
49
37
26
50
37
26
"Seasons: Summer = May through September;
Spring/Fall = March, April, October, November;
Winter = December through February
QI, Q3: 1st and 3rd quartiles of distribution
N: number of households
1 presented no evidence of work done to determine the sampling efficiency curve or cutpoints.
2 Sampling occurred at the rate of 3 1pm for 16 h beginning in the evening and including the
3 overnight sleeping period. Outdoor levels were not recorded, although existing outdoor
4 measurements were quoted as having yearly averages of 30 to 38 Mg/m3> with 84th
5 percentiles at 41 to 51 /zg/m3. Ventilation rates were measured by introducing CO2 at
6 5,000 ppm and recording the decay with an infrared gas analyzer for at least 1 h. The
7 median number of cigarettes smoked was 8; in eleven homes (25% of the total) no cigarettes
8 were smoked. The median ventilation rate was 0.23 h"1, with an interquartile range of
9 0.19 to 0.31 h"1. In the 11 dwellings without smoking, the median concentration was
10 91 /ig/m3 (IQR 57 to 107 /*g/m3); in the homes with < 10 cigarette equivalents, the median
11 (IQR) was 169 (49 to 338) /zg/m3; and in homes with more than 10 cigarette equivalents, the
12 median (IQR) was 475 (309 to 587) /ig/m3. The particle concentration correlated positively
13 with tobacco consumption (rs = 0.716, p< 0.001) and length of daily residence (rs =
14 0.405, p < 0.01) but not with the frequency of cleaning (rs = 0.203, n.s.), time of airing of
15 living rooms (n.s.), or the ventilation rate (rs = 0.277, n.s.).
16 Sexton et al. (1984) reported on a study in Waterbury VT. This study included 24
17 homes, 19 with wood-burning appliances, and none with smokers. 24-h samples (0800 to
18 0800) were collected in each home every other day for two weeks, providing 163 valid
April 1995 7_10l DRAFT-DO NOT QUOTE OR CITE
-------
1 indoor samples. Indoor RSP levels ranged from 6 to 69 /*g/m3 with a mean value of 25
2 fJig/m3. Outdoor levels ranged from 6 to 30 |Ltg/m3 with a mean value of 19 /zg/m3. Indoor
3 concentrations were not correlated with outdoor concentrations (r = 0.11, p> 0.16.)
4 Kim and Stock (1986) reported results from 11 homes in the Houston area. (The year
5 and the season were not supplied in the paper.) For most homes, two 12-h PM2 5 samples
6 (day and night) were collected for approximately one week. Sampling methods were not
7 fully discussed, but apparently involved samples collected using a mobile van near each
8 home. The mean weekly concentrations in the five smoking homes averaged 33.0 +
9 4.7 (SD) /ig/m3, versus mean outdoor concentrations averaging 24.7 ± 7.4 jug/m3 (calculated
10 from data presented in paper). Indoor concentrations in the six nonsmoking homes averaged
11 10.8 + 4.9 /ig/m3 compared to outdoor levels of 12.0 ± 5.9 jig/rn3.
12 Morandi, Stock and Contant (1986) reported on 13 Houston homes monitored during
13 1981 as part of a larger personal monitoring study of 30 nonsmoking participants. The TSI
14 Piezobalance (cutpoint at about PM3 5) was employed for the personal monitoring, with
15 technicians "shadowing" the participants and taking consecutive 5-min readings. At the
16 homes, dichotomous samplers (cutpoints at PM2 5 and PM10) were used for two 12-h daytime
17 samples (7 a.m. to 7 p.m.) both inside and outside the homes for seven consecutive days.
18 Little difference was noted in the indoor concentrations at homes (25 ± 30 (SD) /zg/m3) and
19 at work or school (29 ± 25 /Ag/m3). The authors noted that the highest overall respirable
20 suspended particle (RSP) concentrations occurred in the presence of active smoking
21 (89 /xg/m3), significantly different from mean RSP values measured in the absence of
22 smokers (19 /xg/m3; p < 0.0001). Among homes with smokers, those homes with central
23 air conditioning were significantly (p< 0.0001) higher (114 versus 52 /zg/m3) than those with
24 no air conditioning; in this case, the outdoor differences only increase this contrast. Cooking
25 was associated with significantly higher RSP concentrations (27 pig/m3 compared to 20
26 Mg/m3, p < 0.01). The single highest RSP concentration (202 /*g/m3) was found in a home
27 with no smokers and no air conditioning but with active cooking. The authors concluded that
28 cooking was a more important source of indoor RSP than smoking, at least in the small
29 number of homes they studied.
30 Coultas et al. (1990) measured PM2 5 and nicotine in 10 homes containing at least one
31 smoker. The authors used the Harvard aerosol impactor with sodium bisulfate-treated filters
April 1995 7-102 DRAFT-DO NOT QUOTE OR CITE
-------
1 for nicotine collection, with analysis by GC-FID using a modified Hammond method.
2 Samples were collected for 24 h every other day for 10 days and then for 24 h every other
3 week for 10 weeks, resulting in 10 samples per household. The mean concentrations of
4 PM25 ranged from 32.4 + 13.1 (SD) to 76.9 + 32.9 /xg/m3; of nicotine, from 0.6 ± 0.7 to
5 6.9 ± 8.2 ,ug/m3. Outdoor particle concentrations were not reported; thus it is difficult to
6 calculate the portion of the observed PM2 5 that might be due to ETS. The authors employed
7 a regression technique together with questionnaire variables indicating the periods of smoking
8 to derive an estimate of 17 /xg/m3 as the contribution of smoking to indoor PM2 5; however,
9 the 95% confidence bounds for this estimate ranged from —3 to 38 ^g/m3, indicating that it
10 is not significantly different from zero. The Spearman correlation between total PM2 5 and
11 nicotine was 0.54 (N = 99).
12 Kamens et al. (1991) measured indoor particles in three homes without smokers in
13 North Carolina in November and December 1987 (no measurements of outdoor particles
14 were taken). Two dichotomous samplers (PM2 5 and PM10), several prototype personal
15 samplers (also PM2 5 and PM10, and three particle sizing instruments including a TSI
16 electrical aerosol mobility analyzer (EAA) with 10 size intervals between 0.01 and 1.0 /mi,
17 and two optical scattering devices covering the range of 0.09 to 3.0 and 2.6 to 19.4 /mi were
18 employed. Air exchange measurements were made using SF6 decay over the course of the
19 seven 8-h (daytime) sampling periods. There were also three 13-h (evening and overnight)
20 sampling periods. For the entire study, 37% of the estimated total mass collected was in the
21 fine fraction, and another 37% in the fraction greater than 10 /mi. The remainder (26%)
22 was in the coarse (PM10 - PM2 5) fraction. However, considerable variation was noted in
23 these size distributions. For example, on one day with extensive vacuuming, cooking, and
24 vigorous exercising of household pets, 52% of the total mass appeared in the fraction larger
25 than 10 /mi, with only 18% in the fine fraction. The peak in particle mass on that day
26 coincided with vacuuming and sweeping of the carpets and floors. On another day, stir-fried
27 vegetables and rice produced a large number of small particles, with those less than 0.1 /mi
28 accounting for 30% of the total EAA particle volume, much larger than the normal amount.
29 The cooking contribution of that one meal to total 8-h daytime particle volume exposure was
30 calculated to be in the range of 5 to 18%. The authors concluded that the most significant
31 source of small particles (<2.5 ^m) in all three of these nonsmoking homes was cooking,
April 1995 7_103 DRAFT-DO NOT QUOTE OR CITE
-------
1 while the most significant source of large particles (> 10 /xm) was vacuum sweeping. Coarse
2 particles (PM10 - PM2 5) appeared to be of largely biological (human dander and insect parts)
3 and mineral (clay, salt, chalk, etc.) origin.
4 In a test of a new sampling device (a portable nephelometer), Anuszewski, Larson and
5 Koenig (1992) reported results from indoor and outdoor sampling at nine Seattle homes
6 sampled for an average of 18 days each during the winter of 1991 to 1992. The
7 nephelometer is a light-scattering device with rapid (1-min) response to various household
8 activities such as sweeping, cigarette smoking, frying, barbecuing, and operating a fireplace.
9 Homes with fewer activities showed high correlations of indoor and outdoor light-scattering
10 coefficients, both between hourly averages and 12-h averages. However, homes with
11 electrostatic precipitators, with weather-stripped windows or doors, and with gas cooking or
12 heating devices showed weak 12-h indoor-outdoor correlations. One home with a fireplace
13 produced the strongest indoor-outdoor light scattering relationship (R2 = 0.99); this is
14 presumably due to the very high air exchange rate produced by the fireplace.
15 Chan et al. (1995) studied particles and nicotine in seven homes with one smoker each
16 in Taiwan. Sampling was carried out in summer and winter of 1991. Each home had one
17 indoor PM5 sampler in the living room and another in the yard. In the winter study, two
18 homes had PM10 samplers added inside and outside and at two central sites. Indoor mean
19 PM5 concentrations averaged 44 + 32 (SD) ^ig/m3 in summer compared to outdoor
20 concentrations of 27 ± 15 /ig/m3. Corresponding winter values were 107 ± 44 /ig/m3 and
21 92 + 40 Mg/ni3. Mean cigarette butt counts were only 11.1 in the summer and 6.1 in the
22 winter, and nicotine levels were only 0.4 to 0.5 ^ig/m3 in the two seasons; thus the homes
23 appeared to include fairly light smokers. It was calculated that these smokers produced an
24 average of about 16 /ig/m3 of PM5 daily.
25 Daisey et al. measured RSP, PAH, and extractable organic matter (EOM) in seven
26 Wisconsin homes with wood stoves; one 48h (1,000 m3) sample was collected during
27 woodburning and a second sample was collected when no woodburning occurred. Five of
28 seven homes had somewhat higher RSP levels during woodburning, but the mean difference
29 was not significant. On the other hand, all homes had from 2 to 40 times higher levels of
30 PAH during woodburning, with the means being significantly different. Mean values of
31 EOM were also significantly higher during woodburning.
April 1995 7-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 Highsmith et al. (1991) reported on the results of two 20-home studies in Boise, ID and
2 Roanoke, VA. The Boise study was designed to assess the effects of wood burning on
3 ambient and indoor concentrations in the area. Ten homes with wood burning stoves were
4 matched with 10 homes without wood burning stoves. One matched pair of homes was
5 monitored from Saturday through Tuesday for eight consecutive 12-h periods. Ambient
6 PM2 5 concentrations increased by about 50% at night, suggesting an influence of
7 woodburning. Indoor PM2 5 concentrations also were increased (by about 45%) in the homes
8 with the wood burning stoves compared to those without (26.3 versus 18.2 /xg/m3), although
9 coarse particles showed no increase (10.2 versus 9.7 Mg/m3). The Roanoke study, designed
10 to assess the effects of residential oil heating, showed no effects on indoor levels of fine or
11 coarse particles.
12 Lofroth et al. (1991) measured particle emissions from cigarettes, incense sticks,
13 "mosquito coils," and frying of various foods. Emissions were 27 and 37 mg/g for two
14 brands of Swedish cigarettes, 51 and 52 mg/g for incense sticks and cones, and 61 mg/g for
15 the mosquito coil. Emissions from pork, hamburgers, herring, pudding, and Swedish
16 pancakes ranged from 0.07 to 3.5 mg/g. Extracts of some of the foods showed mutagenic
17 activity. Other studies of mutagenicity of foods from cooking were referenced: Berg et al.
18 (1988) and Teschke et al. (1989). The authors concluded that indoor air pollution from
19 cooking requires further study.
20 Mumford et al. (1991) measured PM10, PAH, and mutagenicity in eight mobile homes
21 with kerosene heaters. Each home was monitored for 2.6 to 9.5 h/day (mean of 6.5 h) for
22 three days a week for two weeks with the kerosene heaters off and for two weeks with them
23 on (average on-time of 4.5 h). Mean PM10 levels were not significantly increased when the
24 heaters were on (73.7 ± 7.3 (SE) ^g/m3 versus 56.1 ± 5.7 ^tg/m3), but in two homes levels
25 increased to 112 and 113 /ig/m3 when the heaters were on. Outdoor concentrations averaged
26 18.0 ± 2.1 jiig/m3. PAH and CO concentrations and mutagenicity were significantly
27 increased when the heaters were on.
28 Colome et al. (1990) measured particles using PMi0 and PM5 (cyclone) samplers inside
29 and outside homes of 10 nonsmokers, including eight asthmatics, living in Orange County,
30 CA. Indoor PM10 samples were well below outdoor levels for all homes (mean of 42.5 ±
April 1995 7-105 DRAFT-DO NOT QUOTE OR CITE
-------
1 3.7 (SE) ^g/m3 indoors versus 60.8 ± 4.7 /tg/m3 outdoors). No pets, wood stoves,
2 fireplaces, or kerosene heaters were present in any of these homes.
3 Lioy et al. (1990) measured PM10 at eight homes (no smokers) for 14 days in the
4 winter of 1988 in the town of Phillipsburg, NJ, which has a major point source consisting of
5 a grey iron pipe manufacturing company. The Harvard impactor was used indoors to collect
6 14 24-h samples beginning at 4:30 p.m. each day; Wedding hi-vol PM10 samples were
7 deployed at three outdoor sites. A fourth outdoor site was located on a porch directly across
8 the street from the pipe sampler. The first three sites showed little difference from one
9 another, whereas on day 4 and day 6 of the study, the outdoor sampler on the porch had
10 readings that were considerably (about 40 ng/m3) higher than the other outdoor samplers,
11 suggesting an influence of the nearby point source. The geometric mean outdoor PM10
12 concentration was 48 ^g/m3 (GSD not provided) compared to 42 /*g/m3 indoors. A simple
13 regression equation for all homes (N = 101 samples) explained 45% of the variance in
14 indoor PM10:
15
16 Indoor PM10 = 0.496 Outdoor PM10 + 21.5
17
18 Thatcher and Lay ton (1994) measured particle size distributions inside and outside a
19 residence in the summer. Measured deposition velocities for particles between 1 and 5 jim
20 closely matched the calculated gravitational settling velocities; however, for particles
21 >5 fjim, the deposition velocity was less than the calculated settling velocity, perhaps due to
22 the non-spherical nature of these particles. The authors calculated a penetration factor of 1,
23 agreeing with the findings of the PTEAM Study. They also determined that resuspension of
24 particles had a significant impact on indoor particle concentrations. Merely walking into a
25 room increased the particle concentration by 100% for particles > 1 jim. The authors
26 calculated a resuspension rate of 2 to 8 X 10~3 h"1 for four persons in a house performing
27 normal activities (i.e., close to 0.01%) of the total collected house dust could be resuspended
28 in an hour. For a surface mass of house dust per unit area of 220 /xg/cm3, this rate resulted
29 in an airborne concentration due to resuspension of about 10 /zg/m3, compared to an outdoor
30 contribution of similar magnitude.
April 1995 7-106 DRAFT-DO NOT QUOTE OR CITE
-------
1 Because house dust can be resuspended, it will contribute to total airborne exposure to
2 particles and constituents such as metals and pesticides. Roberts et al. (1990) studied
3 42 homes in Washington State. Geometric mean lead concentration in 6 homes where shoes
4 were removed on entry was 240 /xg/m2 on carpets, compared to 2,900 /ig/m2 on carpets in
5 homes where shoes were kept on. This finding suggests that most of the carpet dust in a
6 home enters via track-in on shoes rather than infiltration by air.
7
8 7.6.2.2 Studies in buildings
9 The single largest study of particles in buildings was carried out by the Lawrence
10 Berkeley Laboratory (LBL) for the Bonneville Power Administration (BPA) (Turk et al.,
11 1987, 1989). 38 buildings were chosen from two climatic regions in the Pacific Northwest:
12 Portland-Salem, Oregon (representing mild coastal conditions), and Spokane-Cheney,
13 Washington (representing extreme inland conditions). The buildings were studied for a
14 variety of pollutants to determine how ventilation rates affect indoor air quality. Buildings
15 were measured in winter (21 buildings in both regions), spring (10 buildings in both regions)
16 and summer (nine buildings in the inland region only). All but four buildings were
17 government or public properties, and therefore cannot be considered to represent the full mix
18 of building types.
19 Each building was monitored for 10 working days over a two-week period. From 4 to
20 8 particle sampling sites were chosen in each building according to size. The sampler was
21 an LBL-developed flow controlled device with a 3/mi cutpoint. The pumps sampled only
22 during hours the building was occupied. If filters had to be changed due to excessive
23 loading, the combined weight of all filters from one site was determined—thus all values are
24 approximately 10 working-day (80-h) averages.
25 Buildings had varied types of smoking policies, from relatively unrestricted to very
26 tightly controlled, as in one elementary school. In most buildings, an attempt was made to
27 site at least one monitor in an area where smoking was allowed. Data was obtained from
28 smoking areas in about 30 of the 38 buildings.
29 Results comparing smoking and non-smoking areas are provided in Table 7-23 and
30 Figure 7-29. Mean RSP concentrations in the smoking areas were more than three times
31 higher than in the non-smoking areas (70 versus 19 jwg/m3). Since these arithmetic means
April 1995 7-107 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-23. SMOKING, NON-SMOKING, AND OUTDOOR RSP
CONCENTRATIONS AND RATIOS
Outdoor
(/igm 3)
Building No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30"
31
32
33
34
35
36"
37
38
39
40
AM
ASD
GM
GSD
ND
ND
ND
8
BD
35
35
8
8
9
8
ND
10
6
BD
10
7
7
7
18
17
20
11
11
68
32
52
65
29
33
13
ND
ND
16
18
20
19
14
11
11
19
16
14
2.2
Indoor
0*gnr3)
Arithmetic Mean (Range)
Non-Smoking Smoking0 Meand
25(19-36)
19(18-21)
ND
7(6-8)
13(13)
12(11-13)
38(32-44)
7(7-8)
11(11)
65(53-74)
23(9-49)
10(10)
5(5-6)
ND
11(7-14)
9(8-11)
11(10-13)
ND
ND
11(10-11)
11(9-12)
18(18)
9(BD-20)
44(10-77)
35(32-38)
45(20-70)
36(33-38)
36(29-43)
10(8-12)
24(20-30)
12(8-18)
13(10-17)
ND
13(10-16)
20(6-35)
14(9-18)
21(12-32)
7(BD-9)
8(8-9)
10(8-12)
19
14
15
1.9
ND
ND
20(16-25)
ND
14(14)
35(23-59)
39(39)
ND
16(13-20)
95(67-127)
209(209)
63(63)
ND
30(26-34)
12(12)
73(73)
105(105)
19(19)
20(11-29)
ND
ND
57(22-165)
ND
24(24)
109(109)
82(55-123)
61(33-89)
BD
144(144)
113(113
268(268)
36(21-52)
29(12-74)
54(13-117)
50(50)
72(17-127)
27(11-62)
308(308)
13(11-14)
26(11-40)
70
73
44
2.7
"Repeat test of building #1 1 .
bRepeat test of building #17.
cSmoking within 10 m radius of site.
25(19-36)
19(18-21)
20(16-25)
7(6-8)
13(13-14)
28(11-59)
38(32-44)
7(7-8)
15(11-20)
86(53-127)
63(9-209)
36(10-63)
5(5-6)
30(26-34)
11(7-14)
31(8-73)
40(10-105)
19(19)
20(11-29)
11(10-11)
11(9-12)
50(18-165)
9(BD-20)
37(10-77)
60(32-109)
67(20-123)
48(33-89)
24(BD-43)
32(8-144)
37(20-113)
64(8-268)
21(10-52)
29(12-74)
28(10-117)
23(6-50)
28(9-127)
25(11-62)
46(BD-308)
11(8-14)
15(8-40)
30
19
24
2.0
NA
ND
BD
Ratios
Indoor Non- Indoor
Smoking •*• Smoking •*•
Outdoor Outdoor
NA
NA
NA
0.9
NA
0.3
1.1
0.9
1.3
7.0
2.9
NA
0.5
NA
NA
0.9
1.6
NA
NA
0.6
0.7
0.9
0.8
4.0
0.5
1.4
0.7
0.6
0.3
0.7
0.9
NA
NA
0.8
1.1
0.7
1.1
0.5
0.7
0.9
1.2
1.3
0.9
2.0
= Not applicable.
= No data collected.
= Below detection limit.
NA
NA
NA
NA
NA
1.0
1.1
NA
2.0
11.0
26.1
NA
NA
5.0
NA
7.3
15.0
2.7
2.9
NA
NA
2.9
NA
2.2
1.6
2.6
1.2
NA
5.0
3.4
20.6
NA
NA
3.4
2.8
3.6
1.4
22.0
1.3
2.4
6.0
7.2
3.6
2.6
Indoor
Mean -H
Outdoor
NA
NA
NA
0.9
NA
0.8
1.1
0.9
1.9
9.6
7.9
NA
0.5
5.0
NA
3.1
6.1
2.7
2.9
0.6
0.7
2.5
0.8
3.4
0.9
2.1
0.9
0.4
1.1
1.1
4.9
NA
NA
.8
.3
.4
.3
.3
.0
.4
2.3
2.2
1.7
2.3
dArithmetic average of all sites in building.
1 showed evidence of being driven by one or two high values, the geometric mean (averaged
2 across all sites in a building) may be a better comparison. Here the ratio is very close to
April 1995
7-108
DRAFT-DO NOT QUOTE OR CITE
-------
80
70
60
50
CO
I
I 30
20
10
0
Mean Concentrations
Smoking areas Nonsmoking areas Outdoors
CO
"oi
50
40
30
fe 20
DC
10
Geometric Means
Smoking areas Nonsmoking areas Outdoors
Figure 7-29. Comparison of respirable particles in smoking and non-smoking areas of
38 buildings in the Pacific Northwest.
Source: Turk et al. (1987).
April 1995
7-109
DRAFT-DO NOT QUOTE OR CITE
-------
1 3 to 1 (44 versus 15 /ig/m3). If each of the 70 smoking and 106 nonsmoking sites is allowed
2 to contribute equally to the average, the ratio of geometric means declines (34 versus
3 14 /ig/m3). However, the geometric standard deviation (GSD) for the smoking sites is
4 2.5 compared to only 1.9 for the nonsmoking sites—thus, the 97.7th percentile is likely to be
5 more than 6 (2.52) times the geometric mean in the smoking areas (i.e., about 200 /ig/m3)
6 compared to only about 4 (1.92) times the geometric mean in the nonsmoking areas (i.e.,
7 about 56 /ig/m3). Outdoor results at 30 sites had the identical arithmetic mean as the indoor
8 non-smoking sites: 18.9 /ig/m3.
9 Repace and Lowrey (1980) sampled 19 establishments allowing smoking (seven
10 restaurants, three bars, church bingo games, etc.) and 14 where no smoking occurred
11 (including five residences and four restaurants) between March and early May of 1978.
12 Sampling occurred for short periods of time (2 to 50 min) using a TSI Piezobalance to
13 measure PM3 5. Indoor concentrations ranged from 24 to 55 /ig/m3 in the areas without
14 smoking, and from 86 to 697 /ig/m3 in places with active smoking. Five of the locations
15 with active smoking were sampled at a time or place when no smoking was occurring; the
16 matched concentrations (smoking/no smoking) were 279/30; 110/55; 109/30; 86/51; and
17 107/30 /ig/m3. Because ventilation conditions may have changed between visits, these values
18 cannot be considered fully quantitative.
19 Miesner et al. (1989) sampled particles and nicotine in 57 locations within 21 indoor
20 sites in Metropolitan Boston between July 1987 and Feb. 1988. PM2 5 was sampled using
21 Harvard aerosol impactors. Nicotine was sampled using sodium bisulfate-impregnated filters
22 placed downstream from the Teflon filters for the particles. Sampling times ranged from
23 about 3 h in a bus station to 16 h in a library, depending partly on how "clean" the
24 environment was perceived to be. PM2 5 concentrations ranged from 6 /ig/m3 (in the library)
25 to 521 /ig/m3 in a smoking room in an office building. For 42 measurements in
26 non-smoking areas, the mean PM2 5 concentration was 25 + 30 (SD) /ig/m3. Six of these
27 measurements included a classroom with visible levels of chalk dust on the impactor, four
28 measurements in subways, and the bus station. The remaining 36 nonsmoking areas had a
29 mean PM2 5 concentration of 15 ± 7 /ig/m3. The 15 smoking areas ranged from 20 to
30 520 /ig/m3 with a mean of 110 ± 120 /ig/m3.
April 1995 7-110 DRAFT-DO NOT QUOTE OR CITE
-------
1 Turner, Cyr, and Gross (1992) measured RSP using the TSI Piezobalance (cutpoint
2 about 3.5 /*m) in 585 office environments during 1989. The offices were selected because
3 they had contracted with a commercial group (Healthy Buildings International) to perform
4 indoor air quality evaluations— thus they cannot be considered a probability sample. Each
5 office was measured for 1 h (ten 6-min measurements). If smoking was observed during the
6 hour, the office was so classified. Mean RSP concentrations were 46 + 57 (SD) pig/m3 for
7 331 smoking offices, 20 + 17.6 /ig/m3 for 254 nonsmoking offices. Further discriminant
8 analysis reclassified the smoking offices into light smoking (mean smoking density of
9 0.075 cig/m2h) and heavy smoking (0.30 cig/m2h). This analysis suggested that particle
10 concentrations in the light smoking offices were very similar to those in the nonsmoking
11 offices (19 ± 9.2 versus 17 ± 9.5 /xg/m3) while concentrations were much higher in the
12 heavy smoking offices (85 + 72 /*g/m3). It must be pointed out that funding for this work
13 was supplied ultimately by tobacco companies, and that serious allegations have been made
14 by Congressional staff (U.S. House of Representatives, 1994) regarding data irregularities,
15 including possible data alteration and fabrication, systematic misreporting of room size,
16 misclassification of smoking and nonsmoking rooms, and other charges.
17 Vaughan and Hammond (1990) measured particles and nicotine before and after a large
18 corporation adopted a smoking policy limiting smoking to the cafeteria on the 32nd floor.
19 Nicotine levels in offices on various floors ranged from 1.6 to 24 /xg/m3 before the policy
20 was instituted, but dropped to 0.1 to 0.5 /^g/m3 afterwards, an improvement by 84 to 98%.
21 Particle levels dropped from a range of 20 to 270 jig/m3 before the policy to ND-35 jug/m3
22 afterwards (only three measurements due to loss of several samples). The authors noted
23 some evidence that ETS vapors were spilling over from the snack bar to offices on the same
24 floor and two adjacent floors on the same air handler (offices on these floors had nicotine
25 levels about 4 times higher than those on more distant floors).
26 Sheldon (1988a,b) reported on the EPA 10-building study of hospitals, homes for the
27 elderly, schools, and office buildings. Although the main focus of the study was VOCs,
28 particle measurements were taken in all buildings. Measurements were taken in six buildings
29 using a National Bureau of Standards portable particle sampler (NBS 1982; McKenzie et al.,
30 1982) to collect two size fractions: PM3 and a coarse fraction between PM3 and PM15. The
31 sampler employed two filters in series: an 8.0 pm Nuclepore filter for PM15 and a 3
April 1995 7_1H DRAFT-DO NOT QUOTE OR CITE
-------
1 Ghia Zefluor Teflon filter for fine particles. The flow rate was 6 L/min for a 24-h sample.
2 Three consecutive 24-h samples were collected at each building. Additional particle
3 monitoring was provided at certain locations (e.g., smoking lounge, cafeteria) using a
4 Piezobalance (PM3 5) and a dichotomous sampler (PM2 5 and PM10).
5 In areas without smoking, indoor concentrations of both size fractions were generally
6 lower than outdoor levels; for example, the coarse fraction ranged from 0.2 to 0.66 of the
7 outdoor level (13 to 17 /xg/m3) in the three buildings with no smoking. The fine fraction was
8 present at higher indoor-outdoor ratios, ranging from 0.56 to 0.99 in the same three
9 buildings (outdoor fine fraction ranged from 16 to 33 /xg/m3). The fine fraction was elevated
10 in the regions of smoking (range of 14 to 56 /xg/m3). Piezobalance results for several
11 buildings showed uniformly low (7 to 29 /xg/m3) for 800 min of monitoring in nonsmoking
12 areas.
13 Concentrations in the areas allowing smoking were more often in the 40 to 60 /xg/m3,
14 with short-term peaks as high as 345 itg/m3 (Figure 7-30). It was possible to use the
15 observed declines in PM3 5 following cessation of smoking to calculate an effective air
16 exchange rate and thus a source strength for PM3 5 emissions from cigarettes. Four
17 estimates gave an average value of about 6 mg/cigarette, somewhat below the chamber study
18 estimates of 10 to 15 mg/cig. An estimate due to Repace and Lowrey (1980) of
19 concentrations of respirable particulates due to smoking was also tested, with good
20 agreement. The Repace and Lowrey equation is
21
22 C = 25.6 Pala
23
24 where Pa is smoking occupancy in persons per 1,000 square feet and a is the air exchange
25 rate. The equation was developed assuming 1/3 of the occupants are smokers who smoke
26 two cigarettes per hour. Assuming a background concentration of 15 /xg/m3, the estimates
27 for the smoking lounge for 0, 3, and 9 smokers were 10, 78, and 284 /xg/m3, respectively.
28 Repace's equation predicts 0, 99, and 296 /xg/m3, respectively. In two of the homes for the
29 elderly, apartments with smokers and nonsmokers were measured for three consecutive days
30 using the NBS samplers. In one building, the smoker's apartment had a 3-day PM3 average
April 1995 7-112 DRAFT-DO NOT QUOTE OR CITE
-------
100
Home #1
Home #2
Smoker's Room
Nonsmoker's Room
Figure 7-30. Respirable particles in smoking and non-smoking areas of homes for the
elderly (arithmetic mean for 72 h).
1
2
3
4
of 39 £ig/m3, compared to 9.2 /xg/m3 in the nonsmoker's apartment; in the other home for
the elderly, where two smokers shared one apartment, the average 3-day PM3 concentration
was 89 /ig/m3 compared to 8.6 mg/m3 in the nonsmoking apartment.
April 1995
7-113
DRAFT-DO NOT QUOTE OR CITE
-------
1 7.6.2.3 Studies in locations other than homes and buildings
2 Nagda et al. (1990) measured RSP in aircraft cabins using both gravimetric and optical
3 methods. Although the methods did not agree well, they were averaged to produce an
4 estimate of 75 /ig/m3 in the smoking sections, 54 /xg/m3 in the several "boundary rows"
5 bordering the smoking sections, and 31 to 35 jig/m3 in the middle and remote seats.
6 Average concentrations on nonsmoking flights were 35 to 40 jug/m3. Nicotine concentrations
7 were 13.4 pig/m3 in the smoking section, but very low in all other sections (0.04 to 0.26 in
8 nonsmoking sections of smoking flights, 0.00 to 0.08 in all areas on nonsmoking flights).
9 Oldaker et al. (1990) measured PM3 5 and nicotine in 33 restaurants in the Winston-
10 Salem area during the summer of 1986 and the winter of 1988 to 1989. (In the winter
11 season, the cutpoint was changed to PM2 5.) A wide range of particle concentrations was
12 noted, from 18 to 1,374 /xg/m3 in the summer, and <25 to 281 /ig/m3 in winter. Nicotine
13 concentrations also ranged over wide intervals, from 0.9 to 25.6 /xg/m3 in the summer, and
14 <0.1 to 35.2 jug/m3 in the winter.
15 Lowrey et al. (1993) measured PM3 5 using a TSI Piezobalance in a number of outdoor
16 and indoor locations in Budapest. Outdoor concentrations measured over 6 to 16-min periods
17 ranged from 28 to 150 /ig/m3, with three of the four values above 100 /ig/m3 associated with
18 roadways (tram or bus stops). Indoor concentrations in seven areas without visible smoking
19 ranged from 42 to 100 /zg/m3; in 19 areas with active smoking the range was from 56 to
20 650 ^g/m3.
21
22 7.6.3 Indoor air quality models and supporting experiments
23 Indoor concentrations of particles are a function of penetration of outdoor particles and
24 generation of particles indoors. The concentrations are modified by air exchange rates and
25 decay rates of the particles onto indoor surfaces.
26
27 7.6.3.1 Mass Balance Models
28 Mass balance models have been used for more than a century in various branches of
29 science. All such models depend on the law of the conservation of mass. They simply state
30 that the change in mass of a chemically inert substance in a given volume is equal to the
31 amount of mass entering that volume minus the amount leaving the volume. Usually they
April 1995 7-114 DRAFT-DO NOT QUOTE OR CITE
-------
1 are written in the form of first-order linear differential equations. That is, consider a volume
2 V filled with a gas of mass m. The change in mass Am over a small time At will simply be
3 the difference between the mass entering the volume (mgain) and the mass leaving the volume
4 fatoJ:
5
Am = mgain - mloSS (7'8)
6
7
8 Taking the limit as At approaches zero, we have the differential equation for the rate of
9 change of the mass:
10
dm/dt = d/dt(mgain-mloss) (7-9)
11
12
13 If we require that the mass be uniformly distributed throughout the volume at all times,
14 we have a condition that the physical chemists call "well-mixed". We assume that any mass
15 gained or lost in the volume V is instantaneously distributed evenly throughout the volume.
16 We may then replace the mass terms by the concentration C = m/V:
17
18 VdC/dt = d/dt (mgain - m[oj (7-10)
19
20 The above equation is the basis for all such mass-balance models. It takes on many
21 forms depending on the type of processes involved in transporting mass into or out of the
22 volume being considered. A large class of models assume that the volume is a single
23 compartment. More complex models assume multiple compartments. As an example of a
24 single compartment model, we may consider a room of volume V that exchanges air with the
25 outside at a constant flow rate Q. We also assume that a mass of gas has been released in
26 the room at time t — 0, and that the outdoor concentration of this gas is 0. (This is the
27 situation, for example, when a tracer gas such as SF6 is released to determine the air
April 1995 7-115 DRAFT-DO NOT QUOTE OR CITE
-------
1 exchange a.) In this case, the gain in mass mgain is zero and the loss in mass is equal to the
2 flow rate Q out of the house times the concentration C, so that equation (7-10) becomes:
3
4 VdC/dt = -aVC (7-11)
5 Integrating this equation by separation of variables, we have
6
7 C = C0 eat. (7-12)
8
9 where C0 is the concentration at time / = 0, and a = Q/V is the air exchange rate.
10 Thus we find that the original concentration of tracer in the room decays with a time
11 constant a: the air exchange rate.
12 For a nonreactive gas with a nonzero outdoor concentration (e.g., carbon monoxide),
13 the mass balance equation takes the form
14
15 dCfr/dt = a(COM - Cin) (7-13)
16
17 where Cin is the indoor concentration, and cout is the outdoor concentration.
18
19 Depending on the variation with time of Cout, this equation has a number of solutions.
20 If Cout is constant, for example, and the initial indoor concentration is zero, then the indoor
21 concentration rises at a rate determined by the air exchange rate to approach an asymptotic
22 value equal to the outdoor concentration:
23
24 Cin = Cout(l - eat) (7-14)
25
26 An early effort at developing an indoor air quality model was made by Shair and
27 Heitner (1974). This was a mass balance model in which the building was represented as a
28 single well-mixed chamber. A single first-order linear differential equation represented the
29 change in mass of a pollutant due to infiltration, exfiltration, recirculation, source generation,
30 and removal due to filters in the circulation system or deposition on surfaces. Shair and
April 1995 7-116 DRAFT-DO NOT QUOTE OR CITE
-------
1 Heitner (1974) solved the equation for certain simple inputs, such as a linearly increasing or
2 decreasing outdoor concentration:
3
4 Cout = mt + b (7-15)
5
6 Since the outdoor concentration normally is a slowly-varying function, Shair and
7 Heitner's linear approximation is actually an excellent approximation for time intervals of
8 moderate length.
9 If an indoor source 5(0 exists, it enters the mass balance model in the following way:
10
11 dCin/dt = a(Cout - Cin) + S(t)/V (7-16)
12
13 where S(t) has the units of mass per unit time.
14 If the source has a constant generation rate (e.g., CO2 emissions from a person at rest),
15 then S(t) is a constant value 50 and the equation becomes
16
17 dC-Jdt = a(Cout - Cin) + So/V (7-17)
18
19 If the substance of interest reacts or is adsorbed on surfaces while indoors, the equation
20 becomes
21
22 dCin/dt = aCout - (a+k)Cin + S(t)/V (7-18)
23
24 where k represents the loss of the substance due to chemical reaction, adsorption on surfaces,
25 sedimentation, etc. The decay rate k has the same units as the air exchange rate a (I/time);
26 their sum (a+k) may be thought of as an effective air exchange rate. The decay rate k is
27 often used to apply to particles, which disappear faster indoors than a nonreactive gas such as
28 CO. Since particles experience more difficulty than of a gas in penetrating the building
29 envelope, a penetration factor/(f < 1) is often applied that multiplies the outdoor
30 concentration in Equation 7-18 above.
April 1995 7-117 DRAFT-DO NOT QUOTE OR CITE
-------
1 Alzona et al. (1979) applied the mass-balance equation with terms including a
2 penetration coefficient/through the building envelope, adsorption on surfaces given by a rate
3 k proportional to the indoor concentration Cin, and a resuspension rate R:
4
5 dCin/dt = af Cout - aCin - kCin + R. (7-19)
6
7 Setting dCin/dt = 0, the equilibrium solution is
8
9 Cin = (afCout + R)/(a+k) (7-20)
10
11 The equation was tested against a series of experiments in which elements known to be of
12 outdoor origin were collected under various experimental conditions and analyzed using
13 X-rays. The authors concluded that/was of the order of 0.3 for many elements, and that
14 resuspension did not appear to be particularly important. Measurement errors were fairly
15 large (15 to 25%) and limited their ability to estimate values of these parameters.
16 As described above, Koutrakis et al. (1991) used least-squares analysis to solve a
17 simplified form of the mass-balance model to determine source emission rates for particles
18 and elements due to cigarettes, woodsmoke, and kerosene heater use. Koutrakis assumed a
19 value for k in order to solve the equation for/and the source emission rates. Ozkaynak
20 et al. (1993) improved on Koutrakis' approach by using least-squares analysis of the PTEAM
21 results to solve the equation simultaneously for k, / and source emission rates for PM2 5 and
22 PM10 particles and associated elements for smoking and for cooking.
23 Axley and Lorenzetti (1991) developed an indoor model using an element-assembly
24 computer language (STELLA) that is capable of handling any number of compartments and
25 air flows between compartments. The model is based on an earlier model (CONTAM88)
26 developed under EPA sponsorship.
27 Sparks et al. (1991) developed a more user-friendly version of the Axley model, with
28 menus prompting the user to insert the necessary parameters.
29 Traynor et al. (1989) developed a "macromodel" based on Monte Carlo simulations
30 using global input data such as house volumes, air exchange rates, and emissions from
31 combustion sources to assess residential concentrations of combustion-source pollutants such
April 1995 7-118 DRAFT-DO NOT QUOTE OR CITE
-------
1 as CO, NO2, and respirable suspended particles. For a home with only one combustion
2 source during winter in upstate New York, at an outdoor temperature of 0 °C , and an
3 outdoor RSP geometric mean concentration of 19 /*g/m3, the model predicted geometric
4 mean concentrations of about 80 jug/m3 in a home with smoking, 75 /zg/m3 for a radiant
5 kerosene heater, about 60 /-ig/m3 for a convective unvented gas space heater and a non-
6 airtight wood stove, and about 40 /ig/m3 for a radiant unvented gas space heater. An airtight
7 wood stove was predicted to produce a geometric mean about 30 ptg/m3. Gas ovens, dryers,
8 hot water heaters, boilers, and forced-air furnaces were predicted to result in low indoor
9 concentrations of 10 to 15 /-ig/m3, unless the gas oven was used for heating, in which case
10 the predicted geometric mean was about 20 ^g/m3.
11 At present, one of the most complete forms of the mass-balance indoor air quality
12 model has been presented by Nazaroff and Cass (1989). These authors developed the model
13 to allow for changes in particle size and chemical composition, including terms for
14 homogenous turbulence, natural convection, thermophoresis, advection, and Brownian
15 motion. Coagulation of particles is also included. The computer form of the model required
16 40 to 60 min of CPU time to determine an 11-h evolution of an aerosol mixture of 16
17 different sizes. The model was partially validated by checking it against the results of a
18 chamber study using cigarette-generated aerosol to determine the effectiveness of air cleaners
19 (Offermann et al., 1985).
20 A simplified form of the model was employed in a study of indoor air soiling potential
21 in three California museums, two with new HVAC systems and one with only natural
22 ventilation (Nazaroff et al., 1990a). Measured values of elemental carbon in fine (PM2)
23 particles were 0.63, 0.16, and 5.6 /xg/m3 in the three museums compared to model estimates
24 of 0.62 to 0.83, 0.22 to 0.23, and 4.9 jug/m3- The authors were able to predict the fate
25 (i.e., the main removal process and the rate of removal) of particles of various sizes in the
26 three museums (Figure 7-31). In the two newer museums, most particles below 10 pirn were
27 removed by filters; in the older museum, most PM10 particles were removed by ventilation.
28 In all three museums, the dominant fate of the larger particles was gravitational settling onto
29 upward-facing surfaces. The authors concluded that perceptible soiling would occur in less
30 than a year for the older museum, but would require 10 to 40 years for the newer museums.
April 1995 7-119 DRAFT-DO NOT QUOTE OR CITE
-------
Norton Simon Museum
*'* • i H-
• Filtration
Ventilation
Deposition
x Coagulation (-)
+ Coagulation (+)
0.001
0.1
Particle Diameter (urn)
Scott Gallery
i—i i i i r i • i i i i i i 11
• Filtration
Ventilation
• Deposition
Coagulation (-)
+ Coagulation (+]
Ventilation
• Deposition
x Coagulation (-)
+ Coagulation (+
Particle Diameter (u,m)
Figure 7-31. Predicted fate of particles penetrating into buildings of three California
museums as a function of particle size. The ordinate of each point
represents the fraction of the mass that is removed by the indicated
process.
Source: Nazaroff et al. (1990a).
April 1995
7-120 DRAFT-DO NOT QUOTE OR CITE
-------
1 The hanging of pictures at a slight downward-facing angle was predicted to greatly decrease
2 the rate of soiling.
3 A crucial unknown parameter in the mass-balance model for particles is the rate of
4 decay to surfaces. This rate of decay (k) may be expressed as the product of a deposition
5 velocity kd with the surface to volume ratio in the room or building:
6 k = kdS/V (7-21)
7
8 The deposition velocity will vary with particle size.
9 Both the Nazaroff study above and the series of studies by Weschler and colleagues
10 below have provided useful data on deposition velocities for important anions such as
11 sulfates.
12 A series of studies, also concerned with the effects of indoor particles on materials,
13 were carried out by Weschler and colleagues at AT&T Bell Laboratories (Weschler et al.,
14 1989; Sinclair et al., 1988, 1990, 1992). Studies of buildings with low occupancy, large
15 amounts of electronic equipment, and high-quality filtering and HVAC systems succeeded in
16 determining deposition velocities for coarse particles and various fine particle ions. For
17 coarse particles, these velocities were about equal to velocities predicted for gravitational
18 settling, similar to the results of Nazaroff et al. (1990a) described above. For fine particles,
19 however, the deposition velocity was greater than that predicted for gravitational settling
20 alone. For sulfates, the dominant anion in fine particles, deposition velocities at four
21 buildings in Wichita, Lubbock, Newark, and Neenah were 0.004, 0.005. 0.005, and
22 0.004 cm/s, respectively (Sinclair et al., 1992).
23 Nazaroff et al. (1993) reviewed these and other studies of deposition velocity. The
24 authors pointed out that the studies by Weschler and colleagues and also one study in
25 Helsinki (Raunemaa et al., 1989) had produced values of 0.003 to 0.005 cm/s for fine-mode
26 sulfate, but that studies by Nazaroff and colleagues (Ligocki et al., 1990; Nazaroff et al.,
27 1990) resulted in much smaller values of 0.00002 to 0.001 cm/s. It is not clear whether the
28 differences are due to the many differences in surface materials and filtration systems in the
29 different types of buildings (museums versus telephone equipment buildings) or to the
30 different methods of determining deposition velocities. However, the discrepancy is clear
31 evidence that further work is needed.
April 1995 7-121 DRAFT-DO NOT QUOTE OR CITE
-------
1 Because all large-scale studies of indoor air quality agree that the most important indoor
2 source of fine particles is tobacco smoke, a brief review of models and chamber studies
3 dealing with ETS is provided here.
4
5 7.6.3.2 Models of ETS
6 Using the mass balance approach, a number of efforts have been undertaken to model
7 mathematically the pollutant concentrations from tobacco smoke in indoor locations. For
8 example, Brief (1960) proposed a simple graph to determine transient concentrations for
9 pollutants in indoor settings that is based on an exponential decay as a function of time.
10 Turk (1963) proposed a general equation for calculating the concentrations in a chamber that
11 includes both exterior and interior sources, as well as the removal effect of pollutants by air
12 treatment systems. Bridge and Corn (1972) reported that a solution to the equations
13 proposed by Turk (1963) adequately predicts tobacco smoke in occupied spaces. Jones and
14 Pagan (1974) used Turk's equation to calculate carbon monoxide (CO) concentrations versus
15 time from cigarette smoke in an office building and a single-family dwelling. Ishizu (1980)
16 examined experimentally the inclusion of a mixing factor in these models, and Repace and
17 Lowrey (1980) developed a modification of the Turk equation incorporating a mixing factor.
18 They model the concentration as a function of time (assuming an initial concentration of
19 zero) as :
20
C = CM [(l-exp(-(a+k)mt)] (7-22)
21
22 where
23 CQO = G/Vm (a+k) is the equilibrium concentration of ETS particles;
24 m = a "mixing factor" to account for imperfect mixing'
25 G = the generation rate, a function of the number of cigarettes being smoked and the total
26 particle mass emitted from sidestream and exhaled mainstream smoke.
27 Using an estimated value for G of 24 mg/cigarette, Repace (1978b) arrived at an
28 estimate for the steady-rate concentration of PM3 5 (in /ng/m3) due to smoking:
29
April 1995 7-122 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM35 = 217Dhs/a (7-23)
2
3 where Dhs is the density of "habitual smokers" (i.e., those who smoke two cigarettes per
4 hour) per 100 m3, and the value of 217 incorporates a mixing factor and a particle decay rate
5 based on empirical observations in a number of locations.
6 As reported in Repace and Lowrey (1982), equation 7-23 was partially validated by a
7 chamber experiment reported later by Leaderer et al. (1984), in which the measured
8 equilibrium concentration was 620 /ng/m3 compared to 478 /^g/m3 predicted by the equation,
9 a difference of about 23 %. Since equation 7-23 assumes a nation-wide sales-weighted
10 average tar content, a small experiment using one or two brands of cigarettes would be
11 expected to give somewhat different values (sse discussion on page 7-49).
12 The concentrations of pollutants from ETS in a large mixing volume, such as a room,
13 have been observed to increase once a cigarette starts to burn and to decay exponentially
14 once the cigarette is put out (Brief, 1960; Ishizu, 1980; Repace and Lowrey, 1980, 1982;
15 Leaderer et al., 1984; Repace, 1987). These exponential functions are solutions to the mass
16 balance equation for the case of a source that emits at a fixed rate when it is on—and at zero
17 rate when it is off—with a fixed air exchange rate. This source can be viewed as a
18 "rectangular" input "tune series" (concentration as a function of time) to the mass balance
19 model.
20 Smokers ordinarily engage in a sequential smoking "activity pattern" over time: one
21 cigarette is smoked after another, with a recovery period between each cigarette. A person
22 in a room with a smoker (an office, an automobile, a smoking lounge, a restaurant) is
23 exposed to a time series of concentrations resulting from a succession of cigarettes reflecting
24 the smoking activity patterns of the smoker.
25 In recent research, the basic mass-balance model was adapted to the case of a sequence
26 of cigarettes smoked one after another, and its effectiveness in predicting the pollutant
27 concentrations as a function of time (the "time series" of concentrations) was tested using
28 real-time monitoring instruments (Ott et al., 1992). This work has developed a model for
29 computing the time series of pollutants generated by sequential cigarette smoking activity
30 patterns, the Sequential Cigarette Exposure Model (SCEM). If tj is the time at which
April 1995 7-123 DRAFT-DO NOT QUOTE OR CITE
-------
1 cigarette i begins, and S; is the duration of cigarette i, then during the time the cigarette is
2 being smoked, the particle concentration rises according to:
3
S(l-exp(-(a+k)(t-ti))) (
C = u^ 1— +C2i_2exp(-(a+k)(t-tj)) C
4
5 whereas between cigarettes the concentration declines:
6
C = C^expHa+kXt-Sj-tj)) (7-25)
7
8
9 In these equations, C2i_2 = C(f,) and C2i.! = C(f; + ^).
10 A computer program for the SCEM has been written in Microsoft QuickBASIC Version
11 4.5 programming language to apply to equations 7-24 and 7-25. This research also has
12 derived theoretical equations for the minimum, maximum, and mean pollutant concentration
13 in a well-mixed microenvironment for any cigarette smoking activity pattern. General
14 expressions also have been derived for the case of the habitual smoker (uniform cigarette
15 duration and same time between cigarettes) and for the case of multiple habitual smokers.
16 The equations used to derive the SCEM are general and are consistent with earlier ETS
17 indoor air quality models that were derived for special cases (for example, Repace and
18 Lowrey, 1980). Repace (1987), for example, described a person with uniform smoking
19 activity (a constant rate of smoking per unit time) as an "habitual smoker." He considers the
20 special case in which the habitual smoker smokes two cigarettes per hour, which is based on
21 a national average smoking rate. The SCEM considers the general case in which each
22 habitual smoker can have any smoking rate, and concentration is measured on a "real-tune"
23 (that is, continuous) basis.
24 Solutions to the mass balance equation provide a theoretical basis for calculating all
25 parameters of the model—air exchange rate, source strength, and sink removal terms—in a
26 single experiment. Because of the SCEM's fine time resolution, experiments to validate the
27 model require monitoring instruments that operate with fine tune resolution (minutes or
28 seconds). The air exchange rate is determined from the exponential decay of concentrations
April 1995 7-124 DRAFT-DO NOT QUOTE OR CITE
-------
1 seconds). The air exchange rate is determined from the exponential decay of concentrations
2 in the microenvironment (Ott et al., 1992). The source strength is determined from the
3 equilibrium concentration with continuous smoking. The sink removal term for pollutants
4 that adhere to surfaces, such as particles, is determined by subtracting the particle decay rate
5 from the decay rate for a pollutant that has no surface sinks, such as CO.
6 Klepeis et al. (1995) applied an extension of the SCEM model (now called the Multiple
7 Cigarette Exposure Model, or MCEM) to smoking lounges at the San Francisco, CA and San
8 Jose, CA airports. Three Piezobalances were placed at each end and the center of the
9 lounges. One investigator counted the number of lit cigarettes every minute while the other
10 took readings from the Piezobalances. A Langan CO monitor took carbon monoxide
11 readings continuously and logged them automatically. Five visits to each airport were made.
12 During the fifth visit to each airport, an experiment was done to calculate the air exchange
13 rate when there were few or no smokers present: several cigars were smoked and the decay
14 of the CO level was measured, giving air exchange rates of 10.7 and 13.0 ach at the two
15 airports. The calculated PM3 5 source strengths for cigarettes during these two visits were
16 identical at 1,340 ptg/min. Air exchange rates were not determined at the other visits, but
17 assuming the same rates resulted in an average PM3 5 emission rate of 1,450 /xg/min. The
18 decay rate of the ETS-related PM3 5 was estimated to be 0.048 and 0.034 min"1, or 2.88 and
19 2.04 h-1.
20 Ott et al. (1995a) tested the model in a tavern before and after smoking was prohibited.
21 During 26 visits over a period of two years while smoking was allowed, indoor
22 concentrations averaged 56.9 /xg/m3 above outdoor concentrations, compared to 5.9 /ig/m3
23 above outdoor levels on 24 visits in the first six weeks after smoking was prohibited.
24 A second set of follow-up visits (matched by time of day, day of week, and season to the
25 earlier visits) yielded an average concentration 13.1 pig/m3 above outdoor levels. Using
26 cigarette emission rates from the literature, the measured tavern volume of 521 m3, and a
27 measured air exchange rate under "typical" conditions, a mass-balance model predicted 42.5
28 jug/m3 for an average "continuous smoking" count of 1.17 cigarettes, comparing favorably
29 with the observed average of 43.8 /ig/m3.
30 Ott et al. (1995b) used Laplace transforms to apply the model to an experiment in
31 which three Kentucky reference 2R1 cigarettes were smoked one after another in a 25.7 m3
April 1995 7-125 DRAFT-DO NOT QUOTE OR CITE
-------
1 bedroom. Resulting PM3 5 concentrations were measured in the bedroom and living room
2 with a Piezobalance. Peak values reached 5,500 /*g/m3, and about 2,000 /xg/m3 in the living
3 room. The living room window was opened 2 h later, but concentrations were still in excess
4 of 100 /ig/m3 after 4 h.
5
6 7.6.4 Summary and Conclusions
7 At low outdoor levels of fine (PM3 5 or PM2 5) particles (as in most of the cities in the
8 Six-City and New York State studies), mean indoor concentrations have been found to be
9 twice as high as outdoor levels. At high outdoor levels, mean indoor concentrations have
10 been about 10% lower than the mean outdoor concentrations in the two areas studied
11 (Steubenville and Riverside). Indoor concentrations are considerably higher during the day,
12 when people are active, than at night. Based on a mass-balance model, outdoor air was the
13 major source of indoor particles in the PTEAM study, providing about 3/4 of fine particles
14 (PM2 5) and 2/3 of inhalable particles (PM10) in the average home.
15 The three largest studies of indoor air particles in U.S. homes have all found that the
16 single largest indoor source of fine particles is cigarette smoke, for homes with smokers.
17 (EPA's NHAPS data show that 31% of U.S. homes have a smoker, down from 50 to 60% in
18 years past). Estimates of the impact of a smoking home range from about 30 to 45 /ig/m3,
19 and a of a single cigarette from 1 to 2 /ig/m3 for a 24-h period. Homes without smoking
20 have indoor particle concentrations (both PM10 and PM2 5) that are sometimes below and
21 sometimes above the outdoor levels. At low outdoor levels (as in most of the cities in the 6-
22 City and New York State studies) indoor concentrations are generally higher—at high
23 outdoor levels, they are slightly lower. Indoor concentrations are considerably higher during
24 the day, when people are active, than at night.
25 The second largest identified indoor source of particles, as determined by the PTEAM
26 Study and several smaller studies, is cooking. Slightly less than half of the PTEAM homes
27 reported cooking on the day they were monitored. Estimates of the effect of cooking ranged
28 from about 10 to 20 ^ig/m3. A few small studies confirm the effect of cooking on indoor
29 particle levels, both PM10 and PM2 5. The two other large-scale studies did not directly test
30 for the effect of cooking, although the residual indoor concentrations in multivariate
31 calculations led the authors to suggest that cooking could have contributed to the residual.
April 1995 7-126 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kerosene heater use was determined to contribute about 15 jug/m3 to indoor
2 concentrations in one county but not the other in the New York State study. Also, a later
3 effort using mass-balance calculations was unsuccessful in determining a contribution to
4 particle mass from kerosene heater use in either county, although a somewhat smaller set of
5 homes may have been responsible for this result. Gas stoves, wood stoves, and fireplaces
6 were found to have no noticeable impact on total concentrations of particles, although many
7 studies show an increase in PAH concentrations and some show an increase in mutagenicity
8 of indoor air due to these combustion sources.
9 Vacuuming, dusting, and sweeping were found to contribute slightly but with doubtful
10 significance to indoor levels in the PTEAM Study. House volume had a significant but small
11 effect on particle concentrations, with values of -1 to -2 /xg/m3 per 1,000 cubic feet. Air
12 exchange rates were also significant at times, but with different impacts depending on the
13 relative indoor and outdoor concentrations—at high outdoor concentrations, increased air
14 exchange resulted in increases in the indoor air particle levels.
15 Unknown indoor sources were found to account for a substantial fraction (25%) of
16 indoor concentrations (both PM2 5 and PM10) in the PTEAM study. This suggests a need for
17 further research to determine the source or sources of these particles.
18 Decay rates for fine (PM2 5) particles were determined to be about 0.4 h"1 compared to
19 1 h"1 for coarse particles, with an intermediate estimate of 0.65 h"1 for PM10. For a home
20 with no indoor sources whatever and a typical air exchange rate of about 0.75 h-1, this
21 would imply that fine particles indoors would be about 0.757(0.4+0.75) = 65% of the
22 outdoor value at equilibrium, indoor PM10 would be about 54% of outdoor levels, and indoor
23 coarse particles would be about 43% of outdoor levels. Since very few homes were
24 observed to have concentrations this low, it can be inferred that very few homes are free of
25 important indoor sources of particles.
26 Studies in buildings also indicated that smoking was the major indoor source of
27 particles, with a geometric mean of 44 versus 15 jig/m3 (arithmetic mean of 70 versus
28 18 jiig/m3) observed for smoking versus nonsmoking areas in 38 Pacific Northwest buildings.
29 This difference of 29 to 52 pig/m3 is similar to the difference of 30 to 45 pig/m3 estimated
30 from the three major studies of U.S. homes.
April 1995 7-127 DRAFT-DO NOT QUOTE OR CITE
-------
1 Indoor air quality models have been employed with increasing success to estimate
2 source emission rates and particle decay rates. Cigarettes in homes with normal activities
3 appear to emit about 14 mg/cigarette, a result that agrees well with various chamber studies
4 using smokers or smoking machines. Cooking was estimated to emit 4 mg/min, a result that
5 needs confirmation by other studies. Elemental emission profiles have been determined for
6 both smoking and cooking, with potassium and chloride being major markers for smoking,
7 and iron and calcium for cooking. Particle decay rates have been estimated for homes to
8 range between 0.4 and 1.0 h"1. Studies in telephone equipment buildings and museums have
9 established particle deposition velocities for sulfates and other ions, although differences in
10 the estimates suggest that further research is needed.
11
April 1995 7-128 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Alzona, J.; Cohen, B. L.; Rudolph, H.; Jow, H. N.; Frohliger, J. O. (1979) Indoor-outdoor relationships for
3 airborne particulate matter of outdoor origin. Atmos. Environ. 13: 55-60.
4 Anuszewski, J.; Larson, T. V.; Koenig, J. Q. (1992) Simultaneous indoor and outdoor particle light scattering
5 measurements at nine homes using a portable nephelometer. Presented at: meeting of the American
6 Association for Aerosol Research; paper no. 3A.5.
7 Axley, J. W.; Lorenzetti, D. (1991) IAQ modeling using STELLA™: a tutorial introduction building technology
8 program. Cambridge, MA: Massachusetts Institute of Technology.
9 Baughman, A. V.; Gadgil, A. J.; Nazaroff, W. W. (1994) Mixing of a point source pollutant by natural
10 convection flow within a room. Indoor Air 4: 114-122.
11 Berg, I.; Overvik, E.; Nord, C.-E.; Gustafsson, J.-A. (1988) Mutagenic activity in smoke formed during broiling
12 of lean pork at 200, 250, and 300 °C. Mutat. Res. 207: 199-204.
13 Binder, R. E.; Mitchell, C. A.; Hosein, H. R.; Bouhuys, A. (1976) Importance of the indoor environment in air
14 pollution exposure. Arch. Environ. Health 31: 277-279.
15 Bridge, D. P.; Corn, M. (1972) Contribution to the assessment of exposure of nonsmokers to air pollution from
16 cigarette and cigar smoke in occupied spaces. Environ. Res. 5: 192-209.
17 Brief, R. S. (1960) A simple way to determine air contaminants. Air Eng. 2: 39-51.
18 Chan, C.-C.; Huang, S.-K.; Chen, Y.-C.; Wang, J.-D. (1995) Personal, indoor, and outdoor concentrations of
19 respirable suspended particulates (RSP) and nicotine in seven smoker's families in Taiwan. Atmos.
20 Environ.: submitted.
21 Chapin, F. S., Jr. (1974) Human activity patterns in the city. New York, NY: Wiley-Interscience Publishers.
22 Clayton, C. A.; Pellizzari, E. D.; Wiener, R. W. (1991) Use of a pilot study for designing a large-scale
23 probability study of personal exposure to aerosols. J. Exposure Anal. Environ. Epidemiol. 1: 407-422.
24 Clayton, C. A.; Perritt, R. L.; Pellizzari, E. D.; Thomas, K. W.; Whitmore, R. W.; Ozkaynak, H.; Spengler,
25 J. D.; Wallace, L. A. (1993) Particle total exposure assessment methodology (PTEAM) study:
26 distributions of aerosol and elemental concentrations in personal, indoor, and outdoor air samples in a
27 southern California community. J. Exposure Anal. Environ. Epidemiol. 3: 227-250.
28 Cohen, B. S.; Chang, A. E.; Harley, N. H.; Lippmann, M. (1982) Exposure estimates from personal lapel
29 monitors. Am. Ind. Hyg. Assoc. J. 43: 239-243.
30 Cohen, B. S.; Harley, N. H.; Lippmann, M. (1984) Bias in air sampling techniques used to measure inhalation
31 exposure. Am. Ind. Hyg. Assoc. J. 45: 187-192.
32 Colome, S. D.; Kado, N. Y.; Jacques, P.; Kleinman, M. (1990) Indoor-outdoor relationships of particles less
33 than 10 /*m in aerodynamic diameter (PM10) in homes of asthmatics. In: Indoor air '90: precedings of
34 the 5th international conference on indoor air quality and climate, volume 2, characteristics of indoor air;
35 July-August; Toronto, ON, Canada. Ottawa, ON, Canada: International Conference on Indoor Air
36 Quality and Climate, Inc.; pp. 275-280.
April 1995 7-129 DRAFT-DO NOT QUOTE- OR CITE
-------
1 Colome, S. D.; Kado, N. Y.; Jaques, P.; Kleinman, M. (1992) Indoor-outdoor air pollution relations: paniculate
2 matter less than 10 /mi in aerodynamic diameter (PM10) in homes of asthmatics. Atmos. Environ. Part A
3 26: 2173-2178.
4 Commission of the European Communities. (1992) Exposure assessment report number 1, COST 613/2.
5 Luxembourg: Commission of the European Communities; report no. EUR 14345 EN. (Series on air
6 pollution epidemiology).
7 Coultas, D. B.; Samet, J. M.; McCarthy, J. F.; Spengler, J. D. (1990) Variability of measures of exposure to
8 environmental tobacco smoke in the home. Am. Rev. Respir. Dis. 142: 602-606.
9 Cupitt, L. T.; Glen, W. G.; Lewtas, J. (1994) Exposure and risk from ambient particle-bound pollution in an
10 airshed dominated by residential wood combustion and mobile sources. In: Symposium of risk assessment
11 of urban air: emissions, exposure, risk identification, and risk quantitation; May-June 1992; Stockholm,
12 Sweden. Environ. Health Perspect. 102(suppl. 4):75-84.
13 Daisey, J. M.; Spengler, J. D.; Kaarakka, P. (1987) A comparison of the organic chemical composition of
14 indoor aerosols during woodburning and non-woodburning periods. In: Seifert, B.; Esdorn, H.; Fischer,
15 M.; Rueden, H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th international conference on
16 indoor air quality and climate, v. 1, volatile organic compounds, combustion gases, particles and fibres,
17 microbiological agents; August; Berlin, Federal Republic of Germany. Berlin, Federal Republic of
18 Germany: Institute for Water, Soil and Air Hygiene; pp. 215-219.
19 Diemel, J. A. L.; Brunekreef, B.; Boleij, J. S. M.; Biersteker, K.; Veenstra, S. J. (1981) The Arnhem lead
20 study: II, indoor pollution, and indoor/outdoor relationships. Environ. Res. 25: 449-456.
21 Dietz, R. N.; Cote, E. A. (1982) Air infiltration measurements in a home using a convenient perfluorocarbon
22 tracer technique. Environ. Int. 8: 419-433.
23 Dockery, D. W.; Spengler, J. D. (1981a) Indoor-outdoor relationships of respirable sulfates and particles.
24 Atmos. Environ. 15: 335-343.
25 Dockery, D. W.; Spengler, J. D. (1981b) Personal exposure to respirable particulates and sulfates. J. Air Pollut.
26 Control Assoc. 31: 153-159.
27 Dockery, D. W.; Schwartz, J.; Spengler, J. D. (1992) Air pollution and daily mortality: associations with
28 particulates and acid aerosols. Environ. Res. 59: 362-373.
29 Duan, N. (1982) Models for human exposure to air pollution. Environ. Int. 8: 305-309.
30 Duan, N.; Mage, D. T. (1995) Combination of direct and indirect approaches for exposure assessment. J.
31 Exposure Anal. Environ. Epidemiol.: accepted.
32 Federal Register. (1992) Guidelines for exposure assessment. F. R. (May 29) 57: 22887-22938.
33 Fletcher, R. A.; Bright, D. S. (1983) NBS portable ambient paniculate sampler. Washington, DC: U.S.
34 Department of Commerce, National Bureau of Standards; report no. NBSIR 82-2561. Available from:
35 NTIS, Springfield, VA; PB83-165019.
36 Fletcher, B.; Johnson, A. E. (1988) Comparison of personal and area concentration measurements and the use of
37 a manikin in sampling. In: Vincent, J. H., ed. Ventilation '88. New York, NY: Pergamon Press; pp.
38 161-166.
April 1995 7-130 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fugas , M.; Wilder, B.; Paukovic, R.; Hrsak, J.; Steiner-Skreb, D. (1973) Concentration levels and paniculate
2 size distribution of lead in the air of an urban and an industrial area as a basis for the calculation of
3 population exposure. In: Barth, D.; Berlin, A.; Engel, R.; Recht, P.; Smeets, J., eds. Environmental
4 health aspects of lead: proceedings [of an] international symposium; October 1972; Amsterdam, The
5 Netherlands. Luxembourg: Commission of the European Communities; pp. 961-968; report no. EUR
6 5004 d-e-f.
7 Gray, M. I.; Unwin, J.; Walsh, P. T.; Worsell, N. (1992) Factors influencing personal exposure to gas and dust
8 in workplace air: application of a visualization technique. Safety Sci. 15: 273-282.
9 Groves, W. A.; Hahne, R. M. A.; Levine, S. P.; Schork, M. A. (1994) A field comparison of respirable dust
10 samplers. Am. Ind. Hyg. Assoc. J. 55: 748-755.
11 Guerin, M. R.; Jenkins, R. A.; Tomkins, B. A. (1992) The chemistry of environmental tobacco smoke:
12 composition and measurement. Chelsea, MI: Lewis Publishers.
13 Hammond, S. K.; Leaderer, B. P.; Roche, A. C.; Schenker, M. (1986) A method to measure exposure to
14 passive smoking. In: Proceedings of the 1986 EPA/APCA symposium on measurement of toxic air
15 pollutants; April; Raleigh, NC. Research Triangle Park, NC: U.S. Environmental Protection Agency,
16 Environmental Monitoring Systems Laboratory; pp. 16-24; EPA report no. EPA-600/9-86-013. Available
17 from: NTIS, Springfield, VA; PB87-182713.
18 Highsmith, V. R.; Rodes, C. E.; Hardy, R. J. (1988) Indoor particle concentrations associated with use of tap
19 water in portable humidifiers. Environ. Sci. Technol. 22: 1109-1112.
20 Highsmith, V. R.; Zweidinger, R. B.; Merrill, R. G. (1988) Characterization of indoor and outdoor air
21 associated with residences using woodstoves: a pilot study. Environ. Int. 14: 213-219.
22 Highsmith, V. R.; Hoffman, A. J.; Zweidinger, R. B.; Cupitt, L. T.; Walsh, D. (1991) The IACP: overview of
23 the Boise, Idaho and the Roanoke, Virginia field studies. Presented at: the 84th annual meeting of the Air
24 & Waste Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air & Waste
25 Management Association; paper no. 91-131.1.
26 Hodges, S. D.; Moore, P. G. (1977) Data uncertainties and least square regression. Appl. Stat. 44: 185-195.
27 Holcomb, L. C. (1993) Indoor air quality and environmental tobacco smoke: concentration and exposure.
28 Environ. Int. 19: 9-40.
29 Ingebrethsen, B. J.; Heavner, D. L.; Angel, A. L.; Conner, J. M.; Steichen, T. J.; Green, C. R. (1988) A
30 comparative study of environmental tobacco smoke paniculate mass measurements in an environmental
31 chamber. JAPCA 38: 413-417.
32 Ishizu, Y. (1980) General equation for the estimation of indoor pollution. Environ. Sci. Technol. 14: 1254-1257.
33 Jenkins, P. L.; Phillips, T. J.; Mulberg, E. J.; Hui, S. P. (1992) Activity patterns of Californians: use of and
34 proximity to indoor pollutant sources. Atmos. Environ. Part A 26: 2141-2148.
35 Jones, R. M.; Pagan, R. (1974) Application of mathematical model for the buildup of carbon monoxide from
36 cigarette smoking in rooms and houses. ASHRAE J. 16: 49-53.
37 Kamens, R.; Lee, C.-T.; Wiener, R.; Leith, D. (1991) A study to characterize indoor particles in three non-
38 smoking homes. Atmos. Environ. Part A 25: 929-948.
April 1995 7_131 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kim, Y. S.; Stock, T. H. (1986) House-specific characterization of indoor and outdoor aerosols. In: Berglund,
2 B.; Berglund, U.; Lindvall, T.; Spengler, J.; Sundell, J., eds. Indoor air quality: papers from the third
3 international conference on indoor air quality and climate; August 1984; Stockholm, Sweden. Environ.
4 Int. 12: 75-92.
5 Klepeis, N. E.; Ott, W. R.; Switzer, P. (1994) A total human exposure model (THEM) for respirable suspended
6 particles (RSP). Presented at: the 87th annual meeting of the Air & Waste Management Association;
7 June; Cincinnati, OH. Pittsburgh, PA: Air & Waste Management Association; paper no. 94-WA75A.03.
8 Kleipis, N. E.; Ott, W. R.; Switzer, P. (1995) Modeling the time series of respirable suspended particles and
9 carbon monoxide from multiple smokers: validation in two public smoking lounges. Presented at: the
10 88th annual meeting and exhibition of the Air & Waste Management Association; June. Pittsburgh, PA:
11 Air & Wsate Management Association; paper no. A-1233.
12 Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. (1992) Source apportionment of indoor aerosols in Suffolk and
13 Onondaga Counties, New York. Environ. Sci. Technol. 26: 521-527.
14 Langan, L. (1992) Portability in measuring exposure to carbon monoxide. J. Exposure Anal. Environ. Epidemiol.
15 2(suppl. 1): 232-239.
16 Last, J. M., ed. (1983) A dictionary of epidemiology. New York, NY: Oxford University Press; p. 43
17 Last, J. M., ed. (1988) A dictionary of epidemiology. 2nd ed. New York, NY: Oxford University Pressjp. 56.
18 Leaderer, B. P.; Hammond, S. K. (1991) Evaluation of vapor-phase nicotine and respirable suspended particle
19 mass as markers for environmental tobacco smoke. Environ. Sci. Technol. 25: 770-777.
20 Leaderer, B. P.; Cain, W. S.; Isseroff, R.; Berglund, L. G. (1984) Ventilation requirements in buildings—II.
21 paniculate matter and carbon monoxide from cigarette smoking. Atmos. Environ. 18: 99-106.
22 Leaderer, B.; Koutrakis, P.; Briggs, S.; Rizzuto, J. (1990) Impact of indoor sources on residential aerosol
23 concentrations. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and
24 climate, volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON,
25 Canada: International Conference on Indoor Air Quality and Climate, Inc.; pp. 269-274.
26 Lebret, E. (1985) Air pollution in Dutch homes: an exploratory study in environmental epidemiology.
27 Wageningen, The Netherlands: Department of Air Pollution, Department of Environmental and Tropical
28 Health; report R-138; report 1985-221.
29 Lebret, E.; McCarthy, J.; Spengler, J.; Chang, B.-H. (1987) Elemental composition of indoor fine particles. In:
30 Seifert, B.; Esdorn, H.; Fischer, M.; Rueden, H.; Wegner, J., eds. Indoor air '87: proceedings of the
31 4th international conference on indoor air quality and climate, v. 1, volatile organic compounds,
32 combustion gases, particles and fibres, microbiological agents; August; Berlin, Federal Republic of
33 Germany. Berlin, Federal Republic of Germany: Institute for Water, Soil and Air Hygiene; pp. 569-574.
34 Lebret, E.; Boleij, J.; Brunekreef, B. (1990) Environmental tobacco smoke in Dutch homes. In: Indoor air '90:
35 precedings of the 5th international conference on indoor air quality and climate, volume 2, characteristics
36 of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON, Canada: International Conference on
37 Indoor Air Quality and Climate, Inc.; pp. 263-268.
38 Lehmann, E.; Rentel, K.-H.; Allescher, W.; Hohmann, R. (1990) Measurement of diesel exhaust at the
39 workplace. Zentralbl. Arbeitsmed. Arbeitsschutz Ergon. 40: 2-11.
April 1995 7-132 DRAFT-DO NOT QUOTE OR CITE
-------
1 Letz, R.; Ryan, P. B.; Spengler, J. D. (1984) Estimated distributions of personal exposure to respirable particles.
2 Environ. Monit. Assess. 4: 351-359.
3 Lewis, C. W. (1991) Sources of air pollutants indoors: VOC and fine particulate species. J. Exposure Anal.
4 Environ. Epidemiol. 1: 31-44.
5 Lewtas, J.; Mumford, J.; Everson, R. B.; Hulka, B.; Wilcosky, T.; Kozumbo, W.; Thompson, C.; George, M.;
6 Dobias, L.; Sram, R.; Li, X.; Gallagher, J. (1993) Comparison of DMA adducts from exposure to
7 complex mixtures in various human tissues and experimental systems. Environ. Health Perspect. 99:
8 89-97.
9 Ligocki, M. P.; Liu, H. I. H.; Cass, G. R.; John, W. (1990) Measurements of particle deposition rates inside
10 Southern California museums. Aerosol Sci. Technol. 13: 85-101.
11 Lioy, P. J. (1990) Assessing total human exposure to contaminants. Environ. Sci. Technol. 24: 938-945.
12 Lioy, P. J.; Waldman, J. M.; Buckley, T.; Butler, J.; Pietarinen, C. (1990) The personal, indoor and outdoor
13 concentrations of PM-10 measured in an industrial community during the winter. Atmos. Environ. Part B
14 24: 57-66.
15 Litzistorf, G.; Guillemin, M. P.; Buffat, Ph.; Iselin, F. (1985) Influence of human activity on the airborne fiber
16 level in paraoccupational environments. J. Air Pollut. Control Assoc. 35: 836-837.
17 Lofroth, G.; Burton, R. M.; Forehand, L.; Hammond, S, K.; Seila, R. L.; Zweidinger, R. B.; Lewtas, J.
18 (1989) Characterization of environmental tobacco smoke. Environ. Sci. Technol. 23: 610-614.
19 Lofroth, G.; Stensman, C.; Brandhorst-Satzkorn, M. (1991) Indoor sources of mutagenic aerosol particulate
20 matter: smoking, cooking and incense burning. Mutat. Res. 261: 21-28.
21 Lowrey, A. H.; Kantor, S.; Repace, J. L. (1993) Outdoor and indoor respirable suspended particulates in
22 Budapest, Hungary. In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.;
23 Vuorelma, H., eds. Indoor air '93: proceedings of the 6th international conference on indoor air quality
24 and climate: v. 3, combustion products, risk assessment, policies; July; Helsinki, Finland. Helsinki,
25 Finland: Indoor Air '93; pp. 93-96.
26 Mage, D. T. (1980) The statistical form for the ambient particulate standard annual arithmetic mean vs annual
27 geometric mean. J. Air Pollut. Control Assoc. 30: 706-708.
28 Mage, D. T. (1983) Public health aspects of air quality surveillance. Public Health Rev. 11: 5-54.
29 Mage, D. T. (1985) Concepts of human exposure assessment for airborne particulate matter. Environ. Int.
30 11:407-412.
31 Mage, D. T. (1991) A comparison of the direct and indirect methods of human exposure. In: Gledhill; Mauro,
32 F., eds. New horizons in biological dosimetry. New York, NY: Wiley-Liss, Inc.; pp. 443-454. (Progress
33 in clinical and biological research: v. 372).
34 Mage, D. T.; Buckley, T. J. (1995) The relationship between personal exposures and ambient concentrations of
35 particulate matter. Presented at: the 88th annual meeting of the Air & Waste Management Association;
36 June; San Antonio, TX. Pittsburgh, PA: Air & Waste Management Association; paper no. 95-MP18.01.
37 Mamane, Y. (1992) Characterization of PTEAM indoor aerosol samples (electron microscopy analysis). Research
38 Triangle Park, NC: U.S. Environmental Protection Agency, Atmospheric Research and Exposure
39 Assessment Laboratory.
April 1995 7-133 DRAFT-DO NOT QUOTE OR CITE
-------
1 Marple, V. A.; Rubow, K. L.; Turner, W.; Spengler, J. D. (1987) Low flow rate sharp cut impactors for indoor
2 air sampling: design and calibration. JAPCA 37: 1303-1307.
3 Martinelli, C. A.; Harley, N. M.; Lippmann, M.; Cohen, B. S. (1983) Monitoring real-time aerosol distribution
4 in the breathing zone. Am. Ind. Hyg. Assoc. J. 44: 280-285.
5 Matthews, T. G.; Thompson, C. V.; Wilson, D. L.; Hawthorne, A. R.; Mage, D. T. (1989) Air velocities
6 inside domestic environments: an important parameter in the study of indoor air quality and climate.
7 Environ. Int. 15: 545-550.
8 Mayo, E. (1960) Hawthorne and the Western Electric Company. In: Merrill, H. F., ed. Classics in management.
9 New York, NY: American Management Association; p. 237.
10 McKenzie, R. L.; Bright, D. S.; Fletcher, R. A.; Hodgeson, J. A. (1982) Development of a personal exposure
11 monitor for two sizes of inhalable particulates. Environ. Int. 8: 229-233.
12 Miesner, E. A.; Rudnick, S. N.; Hu, F-C.; Spengler, J. D.; Ozkaynak, H.; Preller, L.; Nelson, W. (1989)
13 Paniculate and nicotine sampling in public facilities and offices. JAPCA 39: 1577-1582.
14 Millette, J. R.; Hays, S. M. (1994) Resuspension of settled dust. In: Settled asbestos dust sampling and analysis.
15 Boca Raton, FL: CRC Press, Inc.
16 Morandi, M. T.; Stock, T. H.; Contant, C. F. (1986) Characterization of indoor microenvironmental exposures
17 to respirable paniculate matter. Presented at: 79th annual meeting of the Air Pollution Control
18 Association; June; Minneapolis, MN. Pittsburgh, PA: Air Pollution Control Association; paper
19 no. 86-67.2.
20 Morandi, M. T.; Stock, T. H.; Contant, C. F. (1988) A comparative study of respirable paniculate
21 microenvironmental concentrations and personal exposures. Environ. Monit. Assess. 10: 105-122.
22 Mumford, J. L.; Williams, R. W.; Walsh, D. B.; Burton, R. M.; Svendsgaard, D. J.; Chuang, J. C.; Houk, V.
23 S.; Lewtas, J. (1991) Indoor air pollutants from unvented kerosene heater emissions in mobile homes:
24 studies on particles, semivolatile organics, carbon monoxide, and mutagenicity. Environ. Sci. Technol.
25 25: 1732-1738.
26 Nagda, N.; Fortmann, R.; Koontz, M.; Konheim, A. (1990) Investigation of cabin air quality aboard commercial
27 airliners. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and
28 climate, volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON,
29 Canada: International Conference on Indoor Air Quality and Climate, Inc.; pp. 245-250.
30 National Research Council. (1986) Environmental tobacco smoke: measuring exposures and assessing health
31 effects. Washington, DC: National Academy Press.
32 Nazaroff, W. W.; Cass, G. R. (1989) Mathematical modeling of indoor aerosol dynamics. Environ. Sci.
33 Technol. 23: 157-166.
34 Nazaroff, W. W.; Salmon, L. G.; Cass, G. R. (1990a) Concentration and fate of airborne particles in museums.
35 Environ. Sci. Technol. 24: 66-77.
36 Nazaroff, W. W.; Ligocki, M. P.; Ma, T.; Cass, G. R. (1990b) Particle deposition in museums: comparison of
37 modeling and measurement results. Aerosol Sci. Technol. 13: 332-348.
April 1995 7-134 DRAFT-DO NOT QUOTE OR CITE
-------
1 Nazaroff, W. W.; Gadgil, A. J.; Weschler, C. J. (1993) Critique of the use of deposition velocity in modeling
2 indoor air quality. In: Nagda, N. L., ed. Modeling of indoor air quality and exposure. Philadelphia, PA:
3 American Society for Testing and Materials; ASTM STP no. 1205; pp. 81-104.
4 Neas, L. M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr.; Speizer, F. E. (1994)
5 Concentration of indoor particulate matter as a determinant of respiratory health in children. Am. J.
6 Epidemiol. 139: 1088-1099.
7 O'Brien, D. M.; Fischbach, T. J.; Cooper, T. C.; Todd, W. F.; Gressel, M. F.; Martinez, K. F. (1989)
8 Acquisition and spreadsheet analysis of real time dust exposure data: a case study. Appl. Ind. Hyg. 4:
9 238-243.
10 Offermann, F. J.; Sextro, R. G.; Fisk, W. J.; Grimsrud, D. T.; Nazaroff, W. W.; Nero, A. V.; Revzan, K. L.;
11 Yater, J. (1985) Control of respirable particles in indoor air with portable air cleaners. Atmos. Environ.
12 19: 1761-1771.
13 Ogden, T. L.; Bartlett, I. W.; Puraell, C. J.; Wells, C. J.; Armitage, F.; Wolfson, H. (1993) Dust for cotton
14 manufacture: changing from static to personal sampling. Ann. Occup. Hyg. 37: 271-285.
15 Oldaker, G. B., Ill; Ogden, M. W.; Maiolo, K. C.; Conner, J. M.; Conrad, F. W., Jr.; DeLuca, P. O. (1990)
16 Results from surveys of environmental tobacco smoke in restaurants in Winston-Salem, North Carolina.
17 In: Indoor air '90: precedings of the 5th international conference on indoor air quality and climate,
18 volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON, Canada:
19 International Conference on Indoor Air Quality and Climate, Inc.; pp. 281-285.
20 Ott, W. R. (1982) Concepts of human exposure to air pollution. Environ. Int. 7: 179-196.
21 Ott, W. R. (1990) A physical explanation of the lognormality of pollutant concentrations. J. Air Waste Manage.
22 Assoc. 40: 1378-1383.
23 Ott, W. R.; Langan, L.; Switzer, P. (1992) A time series model for cigarette smoking activity patterns: model
24 validation for carbon monoxide and respirable particles in a chamber and an automobile. J. Exposure
25 Anal. Environ. Epidemiol. 2(suppl. 2): 175-200.
26 Ott, W. R.; Switzer, P.; Robinson, J. (1995a) Particle concentrations inside a tavern before and after prohibition
27 of smoking: evaluating the performance of an indoor air quality model. Submitted.
28 Ott, W. R.; Klepeis, N. E.; Switzer, P. (1995b) Modeling environmental tobacco smoke in the home using
29 transfer functions. Presented at: the 88th annual meeting and exhibition of the Air & Waste Management
30 Association; June. Pittsburgh, PA: Air & Waste Management Association; paper no. A-1043.
31 Owen, M. K.; Ensor, D. S.; Hovis, L. S.; Tucker, W. G.; Sparks, L. E. (1990) Particle size distributions for an
32 office aerosol. Aerosol Sci. Technol. 13: 486-492.
33 Owen, M. K.; Ensor, D. S.; Sparks, L. E. (1992) Airborne particle sizes and sources found in indoor air.
34 Atmos. Environ. Part A 26: 2149-2162.
35 Ozkaynak, H.; Spengler, J. D.; Ludwig, J. F.; Butler, D. A.; Clayton, C. A.; Pellizzari, E.; Wiener, R. W.
36 (1990) Personal exposure to particulate matter: findings from the Particle Total Exposure Assessment
37 Methodology (PTEAM) prepilot study. In: Indoor air '90: precedings of the 5th international conference
38 on indoor air quality and climate, volume 2, characteristics of indoor air; July-August; Toronto, ON,
39 Canada. Ottawa, ON, Canada: International Conference on Indoor Air Quality and Climate, Inc.;
40 pp. 571-576.
April 1995 7_135 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ozkaynak, H.; Xue, J.; Weker, R.; Butler, D.; Spengler, J. (1993a) The particle TEAM (PTEAM) study:
2 analysis of the data, volume III [draft final report]. Research Triangle Park, NC: U.S. Environmental
3 Protection Agency; EPA contract no. 68-02-4544.
4 Ozkaynak, H.; Spengler, J. D.; Xue, J.; Koutrakis, P.; Pellizzari, E. D.; Wallace, L. (1993b) Sources and
5 factors influencing personal and indoor exposures to particles, elements and nicotine: findings from the
6 particle TEAM pilot study. In: Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.;
7 Vuorelma, H., eds. Indoor air '93: proceedings of the 6th international conference on indoor air quality
8 and climate, v. 3, combustion products, risk assessment, policies; July; Helsinki, Finland. Helsinki,
9 Finland: Indoor Air '93; pp. 457-462.
10 Parker, R. C.; Bull, R. K.; Stevens, D. C.; and Marshall, M. (1990) "Studies of aerosol distributions in a small
11 laboratory containing a heated phantom", Annals of Occupational Hygiene, 34 (1), 34-44.
12 Pellizzari, E. D.; Thomas, K. W.; Clayton, C. A.; Whitmore, R. W.; Shores, R. C.; Zelon, H. S.; Perritt, R.
13 L. (1992) Particle total exposure assessment methodology (PTEAM): Riverside, California pilot study,
14 volume I [final report]. Research Triangle Park, NC: U.S. Environmental Protection Agency,
15 Atmospheric Research and Exposure Assessment Laboratory; EPA report no. EPA/600/R-93/050.
16 Available from: NTIS, Springfield, VA; PB93-166957/XAB.
17 Pellizzari, E. D.; Thomas, K. W.; Clayton, C. A.; Whitmore, R. C.; Shores, H.; Zelon, S.; Peritt, R. L.
18 (1993) Particle total exposure assessment methodology (PTEAM): Riverside, California pilot
19 study—volume I [project summary]. Research Triangle Park, NC: U.S. Environmental Protection
20 Agency, Atmospheric Research and Exposure Assessment Laboratory; EPA report no. EPA/600/SR-
21 93/050.
22 Pengelly, L. D.; Goldsmith, C. H.; Kerigan, A. T.; Furlong, W.; Toplack, S. (1987) The Hamilton study:
23 estimating exposure to ambient suspended particles. JAPCA 37: 1421-1428.
24 Penkala, S. J.; de Oliveira, G. (1975) The simultaneous analysis of carbon monoxide and suspended paniculate
25 matter produced by cigarette smoking. Environ. Res. 9: 99-114.
26 Perritt, R. L.; Clayton, C. A.; Pellizzari, E. D.; Thomas, K. W.; Wallace, L. A.; Spengler, J. D.; Ozkaynak,
27 H. (1991) Particle Total Exposure Assessment Methodology (PTEAM) Pilot Study: personal, indoor, and
28 outdoor paniculate concentration distributions for southern California fall 1990—preliminary results. In:
29 Measurement of toxic and related air pollutants: proceedings of the 1991 U.S. EPA/A&WMA
30 international symposium, v. 2; May; Durham, NC. Pittsburgh, PA: Air & Waste Management
31 Association; pp. 665-671. (A&WMA publication VIP-21).
32 Pickles, J. H. (1982) Air pollution estimation error and what it does to epidemiological analysis. Atmos. Environ.
33 16: 2241-2245.
34 Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment approaches to evaluate
35 respiratory health effects of particulate matter and nitrogen dioxide. J. Exposure Anal. Environ.
36 Epidemiol. 1: 83-107.
37 Raunemaa, T.; Kulmala, M.; Saari, H.; Olin, M.; Kulmala, M. H. (1989) Indoor air aerosol model: transport
38 indoors and deposition of fine and coarse particles. Aerosol Sci. Technol. 11: 11-25.
39 Repace, J. L. (1987a) Indoor concentrations of environmental tobacco smoke: field surveys. In: O'Neill, I. K.;
40 Brunnemann, K. D.; Dodet, B.; Hoffmann, D., eds. Environmental carcinogens—methods of analysis
41 and exposure measurement: v. 9, passive smoking. Lyon, France: World Health Organization;
42 pp. 141-162. (IARC scientific publications no. 81).
April 1995 7-136 DRAFT-DO NOT QUOTE OR CITE
-------
1 Repace, J. L. (1987b) Indoor concentrations of environmental tobacco smoke: models dealing with effects of
2 ventilation and room size. In: O'Neill, I. K.; Brunnemann, K. D.; Dodet, B.; Hoffmann, D., eds.
3 Environmental carcinogens—methods of analysis and exposure measurement: v. 9, passive smoking.
4 Lyon, France: World Health Organization; pp. 25-41. (IARC scientific publications no. 81).
5 Repace, J. L.; Lowrey, A. H. (1980) Indoor air pollution, tobacco smoke, and public health. Science 208: 464-
6 472.
7 Repace, J. L.; Lowrey, A. H. (1982) Tobacco smoke, ventilation, and indoor air quality. ASHRAE Trans.
8 88: 895-914.
9 Repace, J. L.; Ott, W. R.; Wallace, L. A. (1980) Total human exposure to air pollution. Presented at: 73rd
10 annual meeting of the Air Pollution Control Association; June; Montreal, PQ, Canada. Pittsburgh, PA:
11 Air Pollution Control Association; paper no. 80-61.6.
12 Revsbech, P.; Korsgaard, J.; Lundqvist, G. R. (1987) Suspended particulate matter in dwellings—the impact of
13 tobacco smoking. Environ. Int. 13: 147-150.
14 Rickert, W. S.; Robinson, J. C.; Collishaw, N. (1984) Yields of tar, nicotine, and carbon monoxide in the
15 sidestream smoke from 15 brands of Canadian cigarettes. Am. J. Public Health 74: 228-231.
16 Roberts, J. W.; Camann, D. E.; Spittler, T. M. (1990) Monitoring and controlling lead in house dust in older
17 homes. In: Indoor air '90: precedings of the 5th international conference on indoor air quality and
18 climate, volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada. Ottawa, ON,
19 Canada: International Conference on Indoor Air Quality and Climate, Inc.; pp. 435-440.
20 Robinson, J.; Nelson, W. C. (1995) National human activity pattern survey data base. Research Triangle Park,
21 NC: U.S. Environmental Protection Agency.
22 Rodes, C. E.; Kamens, R. M.; Wiener, R. W. (1991) The significance and characteristics of the personal
23 activity cloud on exposure assessment measurements for indoor contaminants. Indoor Air 2: 123-145.
24 Roemmelt, H.; Hoeppe, P.; Prami, G.; Schierl, R.; Zielinsky, M. (1993) Measurement of exposure of drivers
25 and passengers to dust and exhaust gases in public transportation in Munich. Berlin, Germany: GSF;
26 pp. 83-90; technical report 31/93.
27 Rudolf, W. (1994) Concentration of air pollutants inside cars driving on highways and and in downtown areas.
28 Sci. Total Environ. 146/147: 433-444.
29 SAS Institute. (1990) SAS language and procedures: SYNTAX, version 6. 1st ed. Gary, NC: SAS Institute.
30 Santanam, S.; Spengler, J. D.; Ryan, P. B. (1990) Particulate matter exposures estimated from an indoor-outdoor
31 source apportionment study. In: Indoor air '90: precedings of the 5th international conference on indoor
32 air quality and climate, volume 2, characteristics of indoor air; July-August; Toronto, ON, Canada.
33 Ottawa, ON, Canada: International Conference on Indoor Air Quality and Climate, Inc.; pp. 583-588.
34 Sexton, K.; Spengler, J. D.; Treitman, R. D. (1984) Personal exposure to respirable particles: a case study in
35 Waterbury, Vermont. Atmos. Environ. 18: 1385-1398.
36 Shair, F. H.; Heitner, K. L. (1974) Theoretical model for relating indoor pollutant concentrations to those
37 outside. Environ. Sci. Technol. 8: 444-451.
April 1995 7-137 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sheldon, L. S.; Handy, R. W.; Hartwell, T. D.; Whitmore, R. W.; Zelon, H. S.; Pellizzari, E. D. (1988a)
2 Indoor air quality in public buildings: v. I. Washington, DC: U.S. Environmental Protection Agency,
3 Office of Acid Deposition, Environmental Monitoring, and Quality Assurance; report no. EPA
4 600/6-88/009a. Available from: NTIS, Springfield, VA; PB89-102503.
5 Sheldon, L. S.; Zelon, H. S.; Sickles, J.; Eaton, C.; Hartwell, J. (1988b) Indoor air quality in public buildings,
6 v. II. Research Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Monitoring
7 Systems Laboratory; report no. EPA 600/6-88/009b. Available from: NTIS, Springfield, VA;
8 PB89-102511.
9 Sheldon, L. S.; Hartwell, T. D.; Cox, B. G.; Sickles, J. E., II; Pellizzari, E. D.; Smith, M. L.; Perritt, R. L.;
10 Jones, S. M. (1989) An investigation of infiltration and indoor air quality: final report. Albany, NY:
11 New York State Energy Research and Development Authority; New York State ERDA contract no.
12 736-CON-BCS-85.
13 Sheldon, L.; Clayton, A.; Keever, J. Perritt, R. L.; Whitaker, D. (1992) PTEAM: monitoring of phthalates and
14 PAHs in indoor and outdoor air samples in Riverside, California, volume II [final report]. Sacramento,
15 CA: California State Air Resources Board. Available from: NTIS, Springfield, VA; PB93-205649/XAB.
16 Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J. (1988) Indoor/outdoor ratios and indoor surface
17 accumulations of ionic substances at Newark, New Jersey. Atmos. Environ. 22: 461-469.
18 Sinclair, J. D.; Psota-Kelty, L. A.; Weschler, C. J.; Shields, H. C. (1990) Measurement and modeling of
19 airborne concentrations and indoor surface accumulation rates of ionic substances at Neenah, Wisconsin.
20 Atmos. Environ. Part A 24: 627-638.
21 Sinclair, J. D.; Psota-Kelly, L. A.; Peins, G. A.; Ibidunni, A. O. (1992) Indoor/outdoor relationships of
22 airborne ionic substances: comparison of electronic equipment room and factory environments. Atmos.
23 Environ. Part A 26: 871-882.
24 Smith, K. R.; Apte, M. G.; Yuqing, M.; Wongsekiarttirat, W.; Kulkarni, A. (1994) Air pollution and the energy
25 ladder in asian cities. Energy (Oxford) 19: 587-600.
26 Smithard, E. H. R. (1954) The 1952 fog in a metropolitan borough. Mon. Bull. Minist. Health Public
27 Health Serv. (G. B.) 13: 26-35.
28 Sparks, L. E.; Tichenor, B. A.; White, J. B.; Chang, J.; Jackson, M. D. (1991) Verification and uses of the
29 Environmental Protection Agency (EPA) indoor air quality model. Presented at: 84th annual meeting of
30 the Air and Waste Management Association; June; Vancouver, BC, Canada. Pittsburgh, PA: Air and
31 Waste Management Association; paper no. 91-62.12.
32 Spengler, J. D.; Soczek, M. L. (1984) Evidence for improved ambient air quality and the need for personal
33 exposure research. Environ. Sci. Technol. 18: 268A-280A.
34 Spengler, J. D.; Thurston, G. D. (1983) Mass and elemental composition of fine and coarse particles in six U.S.
35 cities. J. Air Pollut. Control Assoc. 33: 1162-1171.
36 Spengler, J. D.; Dockery, D. W.; Turner, W. A.; Wolfson, J. M.; Ferris, B. G., Jr. (1981) Long-term
37 measurements of respirable sulfates and particles inside and outside homes. Atmos. Environ. 15: 23-30.
38 Spengler, J. D.; Treitman, R. D.; Tosteson, T. D.; Mage, D. T.; Soczek, M. L. (1985) Personal exposures to
39 respirable particulates and implications for air pollution epidemiology. Environ. Sci. Technol.
40 19: 700-707.
April 1995 7-138 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spengler, J. D.; Ware, J.; Speizer, F.; Ferris, B.; Dockery, D.; Lebret, E.; Brunekreef, B. (1987) Harvard's
2 indoor air quality respiratory health study. In: Seifert, B.; Esdorn, H.; Fischer, M.; Rueden, H.;
3 Wegner, J., eds. Indoor air '87: proceedings of the 4th international conference on indoor air quality and
4 climate, v. 2, environmental tobacco smoke, multicomponent studies, radon, sick buildings, odours and
5 irritants, hyperreactivities and allergies; August; Berlin, Federal Republic of Germany. Berlin, Federal
6 Republic of Germany: Institute for Water, Soil and Air Hygiene; pp. 218-223.
7 Sterling, T. D.; Dimich, H.; Kobayashi, D. (1982) Indoor byproduct levels of tobacco smoke: a critical review
8 of the literature. J. Air Pollut. Control Assoc. 32: 250-259.
9 Stevens, D. C. (1969) "The particle size and mean concentration of radioactive aerosols measured by personal
10 and static air samples", Annals of Occupational Hygiene, 12, 33-40.
11 Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993) Validation of personal exposure models for sulfate and aerosol
12 strong acidity. J. Air Waste Manage. Assoc. 43: 845-850.
13 Switzer, P.; Ott, W. R. (1992) Derivation of an indoor air averaging time model from the mass balance equation
14 for the case of independent source inputs and fixed air exchange rates. J. Exposure Anal. Environ.
15 Epidemiol. 2(suppl. 2): 113-135.
16 Szalai, A., ed. (1972) The use of time: daily activities of urban and suburban populations in 12 countries. The
17 Hague, The Netherlands: Mouton and Co.
18 Teschke, K.; Hertzman, C.; Van Netten, C.; Lee, E.; Morrison, B.; Cornista, A.; Lau, G.; Hundal, A.(1989)
19 Potential exposure of cooks to airborne mutagens and carcinogens. Environ. Res. 50: 296-308.
20 Thatcher, T. L.; Layton, D. W. (1994) Deposition, resuspension, and penetration of particles within a residence.
21 Riverside, CA: University of California, Lawrence Livermore National Laboratory; report no.
22 UCRL-JC-116597.
23 Thomas, K. W.; Pellizzari, E. D.; Clayton, C. A.; Whitaker, D. A.; Shores, R. C.; Spengler, J. D.; Ozkaynak,
24 H.; Wallace, L. A. (1993) Particle total exposure assessment methodology (PTEAM) study: method
25 performance and data quality for personal, indoor, and outdoor aerosol monitoring at 178 homes in
26 southern California. J. Exposure Anal. Environ. Epidemiol. 3: 203-226.
27 Traynor, G. W.; Aceti, J. C.; Apte, M. G.; Smith, B. V.; Green, L. L.; Smith-Reiser, A.; Novak, K. M.;
28 Moses, D. O. (1989) Macromodel for assessing residential concentrations of combustion-generated
29 pollutants: model development and preliminary predictions for CO, NO2, and respirable suspended
30 particles. Berkeley, CA: U.S. Department of Energy, Lawrence Berkeley Laboratory; report no.
31 LBL-25211. Available from: NTIS, Springfield, VA; DE89013396/XAB.
32 Turk, A. (1963) Measurements of odorous vapors in test chambers: theoretical. ASHRAE J. 5: 55-58.
33 Turk, B. H.; Brown, J. T.; Geisling-Sobotka, K.; Froehlich, D. A.; Grimsrud, D. T.; Harrison, J.;
34 Koonce, J. F.; Prill, R. J.; Revzan, K. L. (1987) Indoor air quality and ventilation measurements in
35 38 Pacific Northwest commercial buildings. Volume I: Measurement results and interpretation [final
36 report]. Berkeley, CA: Lawrence Berkeley Laboratory; report no. LBL-22315.
37 Turk, B. H.; Grimsrud, D. T.; Brown, J. T.; Geisling-Sobotka, K. L.; Harrison, J.; Prill, R. J. (1989)
38 Commercial building ventilation rates and particle concentrations. ASHRAE Trans. 95: 422-433.
39 Turner, S.; Cyr, L.; Gross, A. J. (1992) The measurement of environmental tobacco smoke in 585 office
40 environments. Environ. Int. 18: 19-28.
April 1995 7^39 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Centers for Disease Control. (1988) Assessing exposures of health care personnel to aerosols of
2 ribavirin—California. Morb. Mortal. Wkly. Rep. 37: 560-563.
3 U.S. Environmental Protection Agency. (1992) Respiratory health effects of passive smoking: lung cancer and
4 other disorders [review draft]. Washington, DC: Office of Research and Development, Office of Health
5 and Environmental Assessment; EPA report no. EPA/600/6-90/006B. Available from: NTIS, Springfield,
6 VA; PB92-182344.
7 U.S. House of Representatives. (1994) Environmental tobacco smoke investigation: staff report. Washington,
8 DC: Subcommittee on Health and the Environment.
9 Vaughan, W. M.; Hammond, S. K. (1990) Impact of "designated smoking area" policy on nicotine vapor and
10 particle concentrations in a modern office building. J. Air Waste Manage. Assoc. 40: 1012-1017.
11 Wallace, L.; Thomas, J.; Mage, D.; Ott, W. (1988) Comparison of breath CO, CO exposure, and Coburn model
12 predictions in the U.S. EPA Washington-Denver (CO) study. Atmos. Environ. 22: 2183-2193.
13 Wallace, L.; Clayton, A.; Whitmore, R.; Shores, R.; Thomas, K.; Whitaker, D.; Reading, P.; Pellizzari, E.;
14 Spengler, J.; Ozkaynak, H.; Froehlich, S.; Jenkins, P.; Ota, L.; Westerdahl, D. (1991a) Initial results
15 from the PTEAM study: survey design, population response rates, monitor performance and quality
16 control. In: Measurement of toxic and related air pollutants: proceedings of the 1991 U.S.
17 EPA/A&WMA international symposium, v. 2; May; Durham, NC. Pittsburgh, PA: Air & Waste
18 Management Association; pp. 659-664. (A&WMA publication VIP-21).
19 Wallace, L. A.; Pellizzari, E.; Sheldon, L.; Whitmore, R.; Zelon, H.; Clayton, A.; Shores, R.; Thomas, K.;
20 Whitaker, D.; Reading, P.; Spengler, J.; Ozkaynak, H.; Froehlich, S.; Jenkins, P.; Ota, L.; Westerdahl,
21 D. (1991b) The TEAM study of inhalable particles (PM10): study design, sampler performance, and
22 preliminary results. Presented at: 84th annual meeting of the Air & Waste Management Association;
23 June; Vancouver, British Columbia, Canada. Pittsburgh, PA: Air & Waste Management Association;
24 paper no. 91-171.3.
25 Wallace, L. A.; Ozkaynak, H.; Spengler, J. D.; Pellizzari, E. D.; Jenkins, P. (1993) Indoor, outdoor, and
26 personal air exposures to particles, elements, and nicotine for 178 southern California residents. In:
27 Jantunen, M.; Kalliokoski, P.; Kukkonen, E.; Saarela, K.; Seppanen, O.; Vuorelma, H., eds. Indoor air
28 '93: proceedings of the 6th international conference on indoor air quality and climate, v. 3, combustion
29 products, risk assessment, policies; July; Helsinki, Finland. Helsinki, Finland: Indoor Air '93;
30 pp. 445-450.
31 Weschler, C. J.; Shields, H. C.; Kelty, S. P.; Psota-Kelly, L. A.; Sinclair, J. D. (1989) Comparison of effects
32 of ventilation, filtration and outdoor airat telephone office buildings. In: Nagda, N. L.; Harper, J. P.,
33 eds. Design and protocol for monitoring indoor air quality. Philadelphia, PA: American Society for
34 Testing and Materials; pp. 9-34. (ASTM special technical publication 1002).
35 Wiener, R. W. (1988) Measurement and evaluation of personal exposure to aerosols. In: Measurement of toxic
36 and related air pollutants: proceedings of the 1988 EPA/APCA international symposium; May; Research
37 Triangle Park, NC. Pittsburgh, PA: Air Pollution Control Association; pp. 84-88. (APCA publication no.
38 VIP-10).
39 Wiener, R. W. (1989) Particle total exposure assessment methodology—an overview of planning and
40 accomplishments. In: Measurement of toxic and related air pollutants: proceedings of the 1989 U.S.
41 EPA/A&WMA international symposium. Pittsburgh, PA: Air & Waste Management Association;
42 pp. 442-448. (A&WMA publication VIP-13).
April 1995 7-140 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wiener, R. W.; Wallace, L.; Pahl, D.; Pellizzari, E.; Whittaker, D.; Spengler, J.; Ozkaynak, H. (1990) Review
2 of the particle TEAM 9 home field study. In: Measurement of toxic and related air pollutants:
3 proceedings of the 1990 EPA/A&WMA international symposium, v. 2; May; Raleigh, NC. Pittsburgh,
4 PA: Air and Waste Management Association; pp. 452-460. (A&WMA publication VIP-17).
5 World Health Organization. (1982a) Human exposure to carbon monoxide and suspended paniculate matter in
6 Zagreb, Yugoslavia. Geneva, Switzerland: World Health Organization; GEMS (Global Environmental
7 Monitoring System) report no. EFP/82.33.
8 World Health Organization. (1982b) Human exposure to SO2, NO2 and suspended paniculate matter in Toronto,
9 Canada. Geneva, Switzerland: World Health Organization; GEMS (Global Environmental Monitoring
10 System) report no. EFP/82.38.
11 World Health Organization. (1984) Human exposure to suspended paniculate matter and sulfate in Bombay,
12 India. Geneva, Switzerland: World Health Organization; GEMS (Global Environment Monitoring
13 System) report no. EFP/84.86.
14 World Health Organization. (1985) Human exposure to carbon monoxide and suspended paniculate matter in
15 Beijing, People's Republic of China. Geneva, Switzerland: World Health Organization; GEMS (Global
16 Environmental Monitoring System) report no. EFP/84.66.
17 World Health Organization. (1992) Urban air pollution in the megacities of the world. Oxford, United Kingdom:
18 Blackwell Publishers.
19 Xu, X.; Gao, J.; Dockery, D. W.; Chen, Y. (1994) Air pollution and daily mortality in residential areas of
20 Beijing, China. Arch. Environ. Health 49: 216-222.
21 Yost, M. G.; Gadgil, A. J.; Drescher, A. C.; Zhou, Y.; Simonds, M. A.; Levine, S. P.; Nazaroff, W. W.;
22 Saisan, P. A. (1994) Imaging indoor tracer-gas concentrations with computed tomography: experimental
23 results with a remote sensing FTIR system. Am. Ind. Hyg. Assoc. J. 55: 395-402.
April 1995 *u.s. G.p.o.:i995-652-486 7-141 DRAFT-DO NOT QUOTE OR CITE
------- |