United States            Office of Research       EPA 600/P-99/002a
 Environmental Protection    and Development        October 1999
 Agency                Washington, DC 20460     External Review Draft
 Air Quality  Criteria for
 Particulate  Matter

 Volume I
                 Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on
its technical accuracy and policy implications.

-------
                                         EPA 600/P-99/002a
                                             October 1999
                                       External Review Draft
  Air Quality Criteria  for
      Particulate  Matter
                Volume I
                   Notice
This document is a preliminary draft.  It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
     National Center for Environmental Assessment
        Office of Research and Development
        U.S. Environmental Protection Agency
         Research Triangle Park, NC 27711

-------
                                    Disclaimer

     This document is an external review draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
October 1999                            I-ii        DRAFT-DO NOT QUOTE OR CITE

-------
                                        Preface

     National Ambient Air Quality Standards (NAAQS) are promulgated by the United States
Environmental Protection Agency (U.S. EPA) to meet requirements set forth in Sections 108 and
109 of the U.S. Clean Air Act (CAA).  Sections 108 and 109 require the EPA Administrator:
(1) to list widespread air pollutants that may reasonably be expected to endanger public health or
welfare; (2) to issue air quality criteria for them which assess the latest available scientific
information on nature and effects of ambient exposure to them; (3) to set "primary" NAAQS to
protect human health with adequate margin of safety and to set "secondary" NAAQS to protect
against welfare effects (e.g., effects on vegetation, ecosystems, visibility, climate, manmade
materials, etc); and (5) to periodically (every 5-yrs) review and revise, as appropriate, the criteria
and NAAQS for a given listed pollutant or class of pollutants.
     The original U.S. NAAQS for particulate matter (PM), issued in 1971 as "total suspended
particulate" (TSP) standards, were revised in 1987 to focus on protecting against human health
effects associated with exposure to ambient PM less than 10 microns  (< 10 |um) that are capable
of being deposited in thoracic (tracheobronchial and alveolar) portions of the lower respiratory
tract.  Later periodic reevaluation of newly available scientific information, as presented in the
last previous version of this "Air Quality Criteria for Particulate Matter" document published in
1996, provided key scientific bases for PM NAAQS decisions published in July 1997. More
specifically, the PM10 NAAQS set in 1987 (150 |ug/m3, 24-h; 50 |ug/m3, annual ave.) were
retained in modified form and new standards (65 |ug/m3, 24-h; 15 |ug/m3, annual ave.) for
particles < 2.5 |um (PM25) were promulgated in July 1997.
     This First External Review Draft of revised Air Quality Criteria for Particulate Matter
assesses new scientific information that has become available since early 1996 through mid-
1999. Extensive additional pertinent information is expected to be published during the next 6 to
9 months (including results from a vastly expanded U.S. EPA PM Research program and from
other Federal and State Agencies, as well as other partners in the general scientific community)
and, as such, the findings and conclusions presented in this draft document must be considered
only provisional  at this time. The present draft is being released for public comment and review
by the Clean Air Scientific Advisory Committee (CASAC) mainly to obtain comments on the

October 1999                               I-iii         DRAFT-DO NOT QUOTE OR CITE

-------
organization and structure of the document, the issues addressed, and the approaches employed
in assessing and interpreting the thus far available new information on PM exposures and effects.
Public comments and CASAC review recommendations will be taken into account, along with
newly available information published or accepted for peer-reviewed publication by April/May
2000, in making further revisions to this document for incorporation into a Second External
Review Draft.  That draft is expected to be released in June 2000 for further public comment and
CASAC review (September 2000) in time for final revisions to be completed by December
2000). Evaluations contained in the present document will be drawn upon to provide inputs to
associated PM Staff Paper analyses prepared by EPA's Office of Air Quality Planning and
Standards (OAQPS) to pose options for consideration by the EPA Administrator with regard to
proposal and, ultimately, promulgation by July 2000 of decisions on potential retention or
revision of the current PM NAAQS.
     This document was prepared and reviewed by experts from Federal and State government
agencies, academia, industry, and NGO's for use by EPA in support of decision making on
potential public health and environmental risks of ambient PM. It describes the nature, sources,
distribution, measurement, and concentrations of PM in both the outdoor (ambient) and indoor
environments and evaluates the latest data on the health effects in exposed human populations, as
well as environmental effects on: vegetation and ecosystems; visibility and climate; manmade
materials; and associated economic impacts. Although not intended to be an exhaustive literature
review, this document is intended to assess all pertinent literature through mid-1999.
     The National Center for Environmental Assessment - Research Triangle Park, NC
(NCEA-RTP) acknowledges the contributions provided by authors, contributors, and reviewers
and the diligence of its staff and contractors in the preparation of this document.
October 1999                              I-iv        DRAFT-DO NOT QUOTE OR CITE

-------
           Air Quality Criteria for Particulate Matter
                            VOLUME I


   1.  EXECUTIVE SUMMARY 	1-1

   2.  INTRODUCTION 	2-1

   3.  PHYSICS, CHEMISTRY, AND MEASUREMENT OF
      PARTICULATE MATTER	3-1

   4.  CONCENTRATIONS, SOURCES, AND EMISSIONS OF
      ATMOSPHERIC PARTICLES  	4-1
      APPENDIX 4A:  Composition of Particulate Matter Source
                   Emissions	 4A-1

   5.  HUMAN EXPOSURE TO AMBIENT PARTICULATE MATTER:
      RELATIONS TO CONCENTRATIONS OF AMBIENT AND
      NON-AMBIENT PARTICULATE MATTER AND OTHER AIR
      POLLUTANTS 	5-1
      APPENDIX 5A:  Nomenclature	 5A-1
                           VOLUME II

   6.  EPIDEMIOLOGY OF HUMAN HEALTH EFFECTS FROM
      AMBIENT PARTICULATE MATTER 	6-1

   7.  DOSIMETRY AND TOXICOLOGY OF PARTICULATE
      MATTER	7-1

   8.  INTEGRATIVE SYNTHESIS OF KEY POINTS: PARTICULATE
      MATTER EXPOSURE, DOSIMETRY, AND HEALTH RISKS	8-1
                           VOLUME III

   9.  ENVIRONMENTAL EFFECTS OF PARTICULATE MATTER  	9-1
October 1999                      I-v        DRAFT-DO NOT QUOTE OR CITE

-------
                                Table of Contents

                                                                              Page

List of Tables	I-xi
List of Figures  	 I-xv
Authors, Contributors, and Reviewers	I-xxi
U.S. Environmental Protection Agency Project Team for Development of
  Air Quality Criteria for Particulate Matter	I-xxix
1.   EXECUTIVE SUMMARY 	1-1
    1.1   INTRODUCTION 	1-1
         1.1.1  Purpose of the Document	1-1
         1.1.2  Organization of the Document	1-2
    1.2   AIR QUALITY AND EXPOSURE ASPECTS	1-2
         1.2.1  Chemistry and Physics of Atmospheric Particles	1-3
         1.2.2  Sources of Airborne Particles in the United States  	1-5
         1.2.3  Atmospheric Transport and Fate of Airborne Particles  	1-6
         1.2.4  Airborne Particle Measurement Methods	1-6
         1.2.5  Ambient U.S. PM Concentrations: Regional Patterns and Trends	1-7
         1.2.6  Human PM Exposure	1-7
    1.3   DOSIMETRY	1-9
    1.4   PARTICULATE MATTER HEALTH EFFECTS	1-11
         1.4.1  Epidemiology Findings	1-11
         1.4.2  Toxicologic Responses to PM in Animals and Humans  	1-13
         1.4.3  Population Groups at Risk	1-15
    1.5   ENVIRONMENTAL EFFECTS  	1-15
         1.5.1  Vegetation and Ecosystem	1-15
         1.5.2  Particulate Matter-Related Effects on Materials  	1-19
         1.5.3  Particulate Matter-Related Effects on Visibility  	1-20
         1.5.4  Environmental and Economic Impacts of PM	1-21

2.   INTRODUCTION  	2-1
    2.1   LEGISLATIVE REQUIREMENTS	2-1
    2.2   HISTORY OF PREVIOUS PM CRITERIA/STANDARDS REVIEWS	2-2
         2.2.1  The 1997 PM NAAQS Revision 	2-3
         2.2.2  Presidential Memorandum: Next Particulate Matter Review
               and Research	2-5
    2.3   CURRENT PM CRITERIA AND NAAQS REVIEW	2-6
         2.3.1  Criteria Review and Plans and Schedule 	2-6
         2.3.2  Methods And Procedures for Document Preparation	2-8
         2.3.3  Approach 	2-10
         2.3.4  Key Issues of Concern 	2-11
    2.4   DOCUMENT CONTENT AND ORGANIZATION	2-14
    REFERENCES	2-16

October 1999                            I-vii        DRAFT-DO NOT QUOTE OR CITE

-------
                                 Table of Contents
                                       (cont'd)
3.   PHYSICS, CHEMISTRY, AND MEASUREMENT OF PARTICULATE MATTER ... 3-1
    3.1   PHYSICS AND CHEMISTRY OF PARTICULATE MATTER  	3-1
         3.1.1   Definitions  	3-1
         3.1.2   Physical Properties and Processes  	3-2
                3.1.2.1  Definitions of Particle Diameter	3-2
                3.1.2.2  Aerosol Size Distributions	3-3
                3.1.2.3  Nuclei-Mode Particles	3-15
         3.1.3   Chemistry of Atmospheric PM	3-18
                3.1.3.1  Chemical Composition and Its Dependence on Particle Size  . . . 3-19
                3.1.3.2  Primary and Secondary Particulate Matter 	3-19
                3.1.3.3  Particle-Vapor Partitioning	3-25
                3.1.3.4  Removal Processes  	3-29
                3.1.3.5  Particulate Matter and Acid Deposition  	3-30
                3.1.3.6  Particles as Carriers of Toxic Species	3-31
                3.1.3.7  Separation of Fine and Coarse Particles  	3-33
    3.2   MEASUREMENT OF PARTICULATE MATTER  	3-34
         3.2.1   Problems in Measuring Particulate Matter	3-36
                3.2.1.1  Treatment of Semivolatile Components of PM	3-37
                3.2.1.2  Upper Cut Point	3-37
                3.2.1.3  Separation of Fine-Mode and  Coarse-Mode PM	3-39
                3.2.1 A  Treatment of Pressure, Temperature, and Relative Humidity  . . . 3-40
                3.2.1.5  No Way to Determine Accuracy	3-41
         3.2.2   Why Measure Particles	3-43
                3.2.2.1  Attainment of a Standard	3-43
                3.2.2.2  Implementation of a  Standard	3-43
                3.2.2.3  Determination of Health Effects 	3-43
                3.2.2.4  Determination of Ecological Effects	3-44
                3.2.2.5  Determination of Radiative Effects	3-44
                3.2.2.6  PM Components/Parameters Which Need To Be Measured .... 3-44
         3.2.3   Problems Associated with Semivolatile Particulate Matter	3-44
                3.2.3.1  Particulate Nitrates  	3-46
                3.2.3.2  Semivolatile Organic Compounds	3-49
                3.2.3.3  Use of Denuder Systems To Measure Semivolatile
                        Compounds 	3-55
                3.2.3.4  Particle-Bound Water  	3-61
         3.2.4   EPA Monitoring Programs	3-67
                3.2A.I  The Federal Reference Methods for Equilibrated Mass 	3-67
                3.2.4.2  Speciation Monitoring	3-69
         3.2.5   Continuous Monitoring 	3-77
                3.2.5.1  TEOM  	3-77
                3.2.5.2  RAMS  	3-78

October 1999                             I-viii        DRAFT-DO NOT QUOTE OR CITE

-------
                              Table of Contents
                                   (cont'd)
               3.2.5.3  CAMM  	3-78
               3.2.5.4  Light Scattering	3-80
               3.2.5.5  Beta-Gauge Techniques	3-81
               3.2.5.6  Measurements of Individual Particles	3-81
    3.3   SUMMARY	3-83
    REFERENCES	3-90

4.   CONCENTRATIONS, SOURCES, AND EMISSIONS OF ATMOSPHERIC
    PARTICLES	4-1
    4.1   INTRODUCTION  	4-1
    4.2   TRENDS AND PATTERNS IN AMBIENT PARTICULATE MATTER PM2 5
         CONCENTRATIONS AND TRENDS	 4-2
         4.2.1   Daily and Seasonal Variability	4-7
         4.2.2   Relations Between Mass and Chemical Component Concentrations	4-12
         4.2.3   Spatial Variability	4-15
         4.2.4   Urban Concentrations and Patterns from the New PM2 5 Compliance
               Network 	4-20
         4.2.5   Trends and Patterns  	4-20
               4.2.5.1  Visual Range/Haziness 	4-20
               4.2.5.2  Urban Trends	4-22
    4.3   SOURCES OF PRIMARY AND SECONDARY PARTICULATE
         MATTER	4-22
         4.3.1   Source Contributions to Ambient PM 	4-27
    4.4   EMISSIONS ESTIMATES AND THEIR UNCERTAINTIES 	4-33
         4.4.1   Emissions Estimates for Primary Particulate Matter and SO2, NOX,
               and VOCs in the United States	4-33
         4.4.2   Uncertainties of Emissions Inventories 	4-39
    4.5   LONG RANGE TRANSPORT OF PM FROM SOURCES OUTSIDE
         THE UNITED STATES	4-41
    4.6   SUMMARY AND CONCLUSIONS	4-42
    REFERENCES	4-45
    APPENDIX 4A: Composition of PM Source Emissions 	  4A-1
    REFERENCES	  4A-20

5.   HUMAN EXPOSURE TO AMBIENT PARTICULATE MATTER: RELATIONS
    TO CONCENTRATIONS OF AMBIENT AND NON-AMBIENT PM AND
    OTHER AIR POLLUTANTS	5-1
    5.1   INTRODUCTION AND BASIC CONCEPTS  	5-1
         5.1.1   The History of Understanding Human Exposure to Particulate Matter .... 5-5
               5.1.1.1  Caveat  	5-10
         5.1.2   Exposure to PM of Ambient Origin and Total PM 	5-10

October 1999                          I-ix       DRAFT-DO NOT QUOTE OR CITE

-------
                          Table of Contents
                               (cont'd)
   5.2  EXPOSURES TO PM CONCENTRATIONS IN THE COMMUNITY
       AMBIENT ATMOSPHERE 	5-14
   5.3  EXPOSURES TO AMBIENT PM IN INDOOR MICROENVIRONMENTS . . . 5-19
   5.4  EXPOSURES TO PM OF INDOOR ORIGIN	5-24
   5.5  PERSONAL EXPOSURES TO PM OF ONE'S OWN PERSONAL
       ACTIVITIES	5-27
   5.6  PERSONAL PM EXPOSURE  	5-28
   5.7  EXPOSURE TO PM OF AMBIENT ORIGIN IN BOTH INDOOR AND
       OUTDOOR MICROENVIRONMENTS	5-29
       5.7.1  Estimation of the Daily Exposure to PM of Ambient Origin	5-37
   5.8  AMBIENT P ARTICULATE MATTER CONCENTRATION ASA
       SURROGATE FOR EXPOSURE TO PARTICULATE MATTER
       OF AMBIENT ORIGIN	5-44
   5.9  CONCENTRATIONS OF AMBIENT PM FOUND INDOORS AND
       IN OTHER NON-AMBIENT ENVIRONMENTS 	5-45
   5.10 PERSONAL PM MONITORING STUDIES AND FACTORS THAT
       INFLUENCE THEIR ABILITY TO ESTIMATE RELATIONSHIPS
       TO EXPOSURE TO PM OF AMBIENT ORIGIN	5-53
   5.11 EXPOSURES TO AMBIENT PM25 OF PEOPLE BELIEVED TO BE
       SUSCEPTIBLE TO THE EFFECTS OF AMBIENT PM25	5-67
   5.12 PERSONAL EXPOSURE TO CONSTITUENTS OF PM: SULFATES
       AND ACIDITY  	5-72
   5.13 CORRELATION OF AMBIENT PM AND PM EXPOSURE IN
       CROSS-SECTIONAL-TYPE STUDIES	5-77
   5.14 EXPOSURE TO AMBIENT GASEOUS POLLUTANTS RELATED TO
       AMBIENT PM	5-80
   5.15 CONFOUNDING BY INDOOR PM	5-89
   5.16 IMPLICATIONS OF THE AMBIENT PM EXPOSURE RELATIONSHIPS
       FOR EPIDEMIOLOGIC ANALYSIS AND A SUMMARY OF THE
       CHAPTER CONCLUSIONS	5-95
   REFERENCES	5-97
   APPENDIX 5-A: Nomenclature	  5A-1
October 1999                        I-x       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Tables

Number                                                                           Page

1-1       Provisional Conclusions About Alternative Hypotheses That May Affect the
          Synthesis of Epidemiology Studies	1-12

2-1       Schedule for Development of the Current Revised Particulate Matter Air
          Quality Criteria Document	2-7

3-1       Comparison of Ambient Particles, Fine Mode and Coarse Mode	3-35

3-2       PM Components/Parameters of Interest for Health, Ecological, or
          Radiative Effects; for Source Category Apportionment Studies; or for
          Air Quality Validation Studies	3-45

4-1       Concentrations of PM25 and Selected Elements in the PM25 Size Range
          and Correlations Between Elements and PM25 Mass  	4-14

4-2       Correlation Coefficients for Spatial Variation of PM2 5 Mass and Different
          Sources for Pairs of Sampling Sites in SoCAB (1986)	4-18

4-2a      Correlation Coefficients for Spatial Variation of PM25 Mass and Different
          Components for Pairs of Sampling Sites in Philadelphia (1994) 	4-19

4-3       Constituents of Atmospheric Particles and Their Major Sources 	4-25

4-4       Receptor Model Source Contributions to PM2 5	4-28

4-5       Receptor Model Source Contributions to PM10	4-29

4A-1      Average Abundances of Major Elements in Soil and Crustal Rock  	  4A-2

4A-2      Composition of Fine Particles Released by Various Stationary Sources in
          the Philadelphia Area	  4A-6

4A-3      Fractional Organic and Elemental Carbon Abundances in Motor Vehicle
          Emissions	 4A-10

4A-4      Phoenix PM25 Motor Vehicle Emissions Profiles	 4A-12

4A-5      Emission Rates for Constituents of Particulate Matter from Gasoline
          and Diesel Vehicles	 4A-13
October 1999                             I-xi        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables
                                       (cont'd)

Number                                                                         Page
4A-6      Mean Aerosol Composition at Tropical Site Affected Heavily by
          Biomass Burning Emissions  	 4A-17

5-1       Summary of the Mean Values of PM Variables From the PTEAM Study	5-33

5-2       Average Fractions of Time Spent Outdoors or in a Vehicle in the PTEAM
          Study and NHAPS Study   	5-38

5-3       Relative Risk of Nonmalignant Respiratory Mortality From Increasing
          Time Spent Outdoors	5-40

5-4       Summary of Correlations Between PM10 Personal Exposures of Seven
          Tokyo Residents and the PM10 Measured Outdoors Under the Eaves of
          Their Homes, and the PM Measured at the Itabashi Monitoring Station	5-47

5-5       Summary of PM Data During the Summer in an Indoor/Outdoor Study
          in Southwest Virginia  	5-51

5-6       Seasonal Variation of Regression Coefficients for PM10 With Fine and
          Coarse Mode Fractions  	5-53

5-7       Average Levels of Personal Exposures and Outdoor Concentrations and
          the Correlation Between Them in Longitudinal Exposure Studies	5-59

5-8       Personal Exposures to Particulate Matter of Non-Smoke Exposed Children
          in Wageningen, NL, and Stationary Ambient Monitoring Data  	5-61

5-9       Personal PM10 Exposure of 13 Adults in Amsterdam, NL   	5-63

5-10      Personal PM10 Exposure of 24 Adults in Amsterdam, NL   	5-64

5-11      Personal Exposures to PMj 5 of Five Elderly Residents of a Retirement
          Center in Baltimore, MD,  as a Function of Ambient PM25 Concentration	5-69

5-12A     Summary of Correlations Between Daytime Personal PM Exposures
          and Daytime Outdoor PM Concentrations for 15 COPD Patients During
          Combined Summer and Winter Periods in Boston, MA	5-71

5-12      Example of a Completed Matrix for a Type-1 Analysis of PM Exposure
          and Ambient PM Concentration  	5-77
October 1999                            I-xii        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Tables
                                       (cont'd)

Number                                                                         Page
5-13      Example of an Incomplete Matrix for Type-2 Analysis of PM Exposure
          and Ambient PM Concentration  	5-78

5-14      Results of the Analysis of the PTEAM Daily Average Personal PM10
          Exposure Data from Riverside, California	5-79

5-15      Average Annual Correlations Between PM and Criteria Gaseous Pollutants
          in the United States  	5-87
October 1999                            I-xiii       DRAFT-DO NOT QUOTE OR CITE

-------
                                    List of Figures

Number                                                                            Page

3-1       Number of particles as a function of particle diameter  	3-5

3-2       Particle volume distribution as a function of particle diameter	3-6

3-3       Distribution of coarse, accumulation, and nuclei or ultrafine mode
          particles by three characteristics, volume, surface area, and number
          for the grand average continental size distribution	3-7

3-4       Volume size distribution, measured in traffic, showing fine-mode and
          coarse-mode particles and the nuclei and accumulation modes within
          the fine-particle mode	3-9

3-5       An idealized distribution of ambient particulate matter showing
          fine-mode particles and coarse-mode particles and the fractions
          collected by size-selective samplers	3-10

3-6       Specified particle penetration through an ideal inlet for five different
          size-selective sampling criteria  	3-12

3-7       Comparison of penetration curves for two PM10 beta gauge samplers
          using cyclone inlets 	3-14

3-8       Particle growth curves showing fully reversible hygroscopic growth of
          sulfuric acid particles, deliquescent growth of ammonium sulfate
          particles at about 80% relative humidity (RH), hygroscopic growth of
          ammonium sulfate solution droplets at RH greater than 80%, and hysteresis
          until the crystallization point is reached  	3-28

3-9       Theoretical predictions and experimental measurements of growth of
          NH4HSO4 and particles at relative humidity between 95 and 100%	3-29

3-10      Schematic showing major nonvolatile and semivolatile components of PM2 5  ... 3-38

3-11      Amount of ammonium nitrate volatilized from Teflon filters, expressed
          as a percentage of the measured PM25 mass, for the SCAQS and CalTech
          studies, for spring and fall sampling periods	3-47

3-12      Aerosol water content expressed as a mass percentage, as a function of
          relative humidity  	3-64
October 1999                              I-xv         DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                        (cont'd)

Number                                                                           Page
3-13      This thermogram, for a sample containing rock dust and diesel exhaust,
          shows three traces that correspond to temperature, filter transmittance,
          and FID detector response	3-75

3-14      Comparison of mass measurements with collocated RAMS, PC-BOSS,
          FRM PM25 sampler, and a conventional TEOM monitor	3-79

3-15      Size distribution of particles divided by chemical classification into
          organic, marine, and crustal	3-82

4-la      Major constituents of particles measured at sites  in the eastern United States .... 4-3

4-lb      Major constituents of particles measured at sites  in the central United States .... 4-4

4-lc      Major constituents of particles measured at sites  in the western United States .... 4-5

4-2       Annual average PM25 concentration (1994-96)	4-6

4-3       Concentrations of PM25 measured at the PBY site in southwestern
          Philadelphia	4-8

4-4       Frequency distribution of PM25 concentrations measured at the PBY site
          in southwestern Philadelphia	4-9

4-5       Concentrations of PM25 measured at the EPA site in Phoenix  	4-10

4-6       Frequency distribution of PM2 5 concentrations measured at the EPA site
          in Phoenix  	4-11

4-7       Concentrations of PM25 measured at the Riverside-Rubidoux site	4-12

4-8       Frequency distribution of PM25 concentrations measured at the
          Riverside-Rubidoux site 	4-13

4-9       PM25 chemical components in downtown Los Angeles and Burbank (1986)
          have similar characteristics  	4-16

4-10      Concentrations of PM25 chemical components in Rubidoux and downtown
          Los Angeles (1986)  .	4-17
October 1999                             I-xvi        DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                       (cont'd)

Number                                                                          Page

4-11      Five-year average haze patterns centered on 1980 and 1990	4-21

4-12      Secular haze trends (1940 to 1990) for six eastern U.S. regions, summer
          and winter	4-23

4-13      Trend data from Stockton-Hazelton from CARB: fine, coarse, and total
          means; fine, coarse, and total 90th percentiles; every sixth-day fine and
          coarse mass for 1991; and fine and coarse mass as a fraction of PM10	4-24

4-14      Nationwide emissions of PM25 from various source categories 	4-34

4-15      Distribution of primary PM25 emissions across the United States  	4-36

4-16      Distribution of primary PM10 emissions across the United States	4-37

4-17      Nationwide emissions of SO2, NOX, VOCs, and NH3 from various source
          categories	4-38

4A-1      Size distribution of particles generated in a laboratory resuspension chamber . .  4A-3

4A-2      Size distribution of California source emissions, 1986	  4A-4

4A-3      Chemical abundances for PM2 5 emissions from paved road dust in
          Denver	  4A-5

4-A4      Chemical abundances for PM2 5 emissions from wood burning in
          Denver	  4A-15

5-1       Sizes of various types of indoor particles	5-3

5-2       Categories of particle exposure outdoors and indoors	5-12

5-3       Two compartment model for PM deposition and resuspension by human
          activity in a residential microenvironment	5-20

5-4       Histograms of the estimated fractions of outdoor PM25 and outdoor
          PM10 found indoors during the PTEAM study in Riverside, CA	5-24

5-5       Individual indoor versus outdoor relationships of PM10 in Tokyo for the
          seven subjects reported on by Tamura et al. (1996a)	5-46


October 1999                            I-xvii       DRAFT-DO  NOT QUOTE OR CITE

-------
                                   List of Figures
                                       (cont'd)

Number                                                                           Page
5-6       Comparison of light scattering coefficient by PM indoors and
          outdoors at a home with an electrostatic precipitator in operation  	5-49

5-7       Comparison of light scattering coefficients by PM indoor and outdoors
          at a home with a standard in-line filter in the air recirculating system  	5-50

5-8       Plot of relationship between average personal PM10 exposure versus
          ambient PM10 monitoring data from Phillipsburg, NJ, and regression
          line calculated by U.S. EPA	5-57

5-9       Plot of 48-h average personal PM10 exposure and ambient PM10 data from
          Japan—linear regression	5-58

5-10      Completed mean personal PM2 5 exposure of children not exposed to tobacco
          smoke at home in Wageningen, NL, versus the simultaneous ambient PM2 5
          measured in their community	5-62

5-11A,B  Completed mean personal PM10 exposures of two groups of adults
          not exposed to tobacco smoke at home in Amsterdam, NL, versus
          the simultaneous ambient PM10 measured in their community	5-65

5-12      Mean personal exposures to PMj 5 of five elderly residents of a retirement
          home in Baltimore, MD, as a function of ambient PM25 concentration 	5-70

5-13      Personal versus outdoor SO4= in State College, PA	5-73

5-14      Results from simultaneous measurements of the indoor PM10 concentration
          and the immediate outdoor PM10 concentration of an uninhabited apartment
          in a building fronting to a busy street in Oslo, Norway	5-75

5-15      PTEAM mean 24-h PM10 data compared for personal PEM and SAM	5-79

5-16      Correlations of PM10-CO for highest urban PM10 site per state;
          correlations of PM10-CO for lowest rural PM10 site per state; and
          correlations of PM25-CO	5-82

5-17      Correlations of PM10-NO2 for highest urban PM10 site per state;
          correlations of PM10-NO2 for lowest rural PM10 site per state; and
          correlations of PM25-NO2	5-83
October 1999                            I-xviii       DRAFT-DO NOT QUOTE OR CITE

-------
                                   List of Figures
                                        (cont'd)

Number                                                                           Page
5-18      Correlations of PM10-SO2 for highest urban PM10 site per state;
          correlations of PM10-SO2 for lowest rural PM10 site per state; and
          correlations of PM25-SO2  	5-84

5-19      Correlations of PM10-O3 1 -hr for highest urban PM10 site per state;
          correlations of PM10-O3 1-hr for lowest rural PM10 site per state; and
          correlations of PM25-O3 1-hr	5-85

5-20      Correlations of PM10-O3 8-hr for highest urban PM10 site per state;
          correlations of PM10-O3 8-hr for lowest rural PM10 site per state; and
          correlations of PM25-O3 8-hr	5-86

5-21      Air-exchange rate versus outdoor PM10 concentration at the home in the
          PTEAM study  	5-90

5-22      Estimated personal exposure to indoor and personally generated PM10
          versus outdoor PM10 at the same home in the PTEAM study	5-92
October 1999                             I-xix        DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                        CHAPTER 1. EXECUTIVE SUMMARY
Principal Authors

Dr. Robert Chapman—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Beverly Comfort—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. J.H.B. Garner—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. David Mage—National Exposure Research Laboratory (MD-56), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711

Dr. Allan Marcus—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. William E.  Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999                            I-xxi       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
                           CHAPTER 2. INTRODUCTION
Principal Authors

Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
            CHAPTER 3. PHYSICS, CHEMISTRY, AND MEASUREMENT
                            OF PARTICULA TE MA TTER
Principal Authors

Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Candis S. Claiborn—Washington State University, Laboratory for Atmospheric Research,
Department of Civil and Environmental Engineering,, P.O. Box 642910, Pullman, WA 99164

Contributors and Reviewers

Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD  21205

Ms. Lee Byrd—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way,  Suite C
Petaluma, CA 94549

Dr. Steven Colome—Integrated Environmental Services, 5319  University Drive, #430
Irvine, C A 92612

Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602
October 1999                            I-xxii        DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)
Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095

Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lynn Hildemann—Stanford University, Civil and Environmental Engineering Department
Stanford, CA  94305

Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130

Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Richard Poirot—VT Air Program, Building 3 South,  103 South Main Street
Waterbury,VT 05671

Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999                           I-xxiii       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA  02461

Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street
Sacramento, CA 95814
                 CHAPTER 4. CONCENTRATIONS, SOURCES, AND
                   EMISSIONS OF ATMOSPHERIC PARTICLES
Principal Authors

Dr. Joseph P. Pinto—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors and Reviewers

Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD 21205

Ms. Lee Byrd—Office of Air Quality Planning and Standards (MD-14)
U. S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way, Suite C
Petaluma, CA 94549

Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430
Irvine, C A 92612

Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602

Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

October 1999                           I-xxiv       DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)
Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095

Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lynn Hildemann—Civil and Environmental Engineering Department
Stanford University, Stanford, CA  94305

Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130

Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52), U.S. Environmental
Protection Agency, Research Triangle Park, NC  27711

Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street
Waterbury,VT 05671

Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA  02461
October 1999                            I-xxv        DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street, Sacramento, CA 95814
     CHAPTER 5. HUMAN EXPOSURE TO AMBIENTF'ARTICULATE MATTER:
              RELATIONS TO CONCENTRATIONS OF AMBIENT AND
                NON-AMBIENTPM, AND OTHER AIR POLLUTANTS
Principal Authors

Dr. David Mage—National Center for Environmental Assessment (MD-52)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Contributors and Reviewers

Dr. Timothy Buckley—Johns Hopkins University, Department of Environmental Health
Sciences, 615 North Wolfe Street,, Baltimore, MD  21205

Ms. Lee Byrd U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lyle Chinkin—Sonoma Technology, 1360 Redwood Way, Suite C
Petaluma, CA 94549

Dr. Steven Colome—Integrated Environmental Services, 5319 University Drive, #430
Irvine, C A 92612

Dr. Delbert Eatough—Brigham Young University, E 114 BNSN,
Department of Chemistry and Biochemistry, Provo, UT 84602

Mr. William Ewald—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Sheldon Friedlander—University of California at Los Angeles, Department of Chemical
Engineering, 5531 Boelter Hall, Los Angeles, CA 90095

Dr. Lawrence J. Folinsbee—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
October 1999                            I-xxvi      DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                       (cont'd)
Dr. Judith Graham—National Exposure Research Laboratory (MD-75)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lester D. Grant—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Vic Hasselblad—29 Autumn Woods Drive, Durham, NC 27713

Dr. Lynn Hildemann—Civil and Environmental Engineering Department
Stanford University, Stanford, CA  94305

Mr. Jim Homolya—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Rudolf Husar—CAPITA, Washington University, Campus Box 1124
One Brookings Drive, St. Louis, MO 63130

Dr. Dennis J. Kotchmar—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Charles W. Lewis—National Exposure Research Laboratory (MD-47)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Allan Marcus—National  Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom McCurdy—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Frank McElroy—National Exposure Research Laboratory (MD-46)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Haluk Ozkaynak—National Exposure Research Laboratory (MD-56)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Tom Pace—Office of Air Quality Planning and Standards (MD-14)
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Joseph Pinto—National Center for Environmental Assessment (MD-52), U.S. Environmental
Protection Agency, Research Triangle Park, NC  27711
October 1999                           I-xxvii      DRAFT-DO NOT QUOTE OR CITE

-------
                     Authors, Contributors, and Reviewers
                                      (cont'd)
Dr. Richard Poirot—VT Air Program, Building 3 South, 103 South Main Street
Waterbury,VT 05671

Dr. Linda Sheldon—National Exposure Research Laboratory (MD-77)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Helen Suh—Harvard School of Public Health, 665 Huntington Avenue
Boston, MA 02461

Mr. Robert Wayland—Office of Air Quality Planning and Standards (MD-15)
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Mr. Dane Westerdahl—California Air Resources Board, 2020 L Street, Sacramento, CA 95814

Dr. William Wilson—National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Dr. Jim Xue—Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115
October 1999                           I-xxviii      DRAFT-DO NOT QUOTE OR CITE

-------
             U.S. ENVIRONMENTAL PROTECTION AGENCY
  PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                        FOR PARTICULATE MATTER
Scientific Staff

Dr. Lester D. Grant—Director, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Randy Brady—Deputy, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Dr. Lawrence J. Folinsbee—Health Coordinator, Chief, Environmental Media Assessment
Group, National Center for Environmental Assessment (MD-52), U.S. Environmental Protection
Agency, Research Triangle Park, NC  27711

Dr. William E. Wilson—Air Quality Coordinator, Physical Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711

Dr. Dennis J. Kotchmar—Project Manager, Medical Officer, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Dr. Robert Chapman—Technical Consultant, Medical Officer, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711

Ms. Beverly Comfort—Health Scientist, National Center for Environmental Assessment (MD-
52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. William Ewald—Technical Project Officer, Health Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711

Dr. David Mage—Technical Project Officer, Physical  Scientist, National Center for
Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle
Park,NC 27711

Dr. Allan Marcus—Technical Project Officer, Statistician, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711
October 1999                           I-xxix       DRAFT-DO NOT QUOTE OR CITE

-------
             U.S. ENVIRONMENTAL PROTECTION AGENCY
 PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                       FOR PARTICULATE MATTER
                                      (cont'd)
Dr. James McGrath—Technical Project Officer, Visiting Senior Health Scientist, National
Center for Environmental Assessment (MD-52), U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711

Dr. Joseph P. Pinto—Technical Project Officer, Physical Scientist, National Center for
Environmental Assessment, U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Technical Support Staff

Mr. Douglas B. Fennell—Technical Information Specialist, National Center for Environmental
Assessment (MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC
27711

Ms. Emily R. Lee—Management Analyst, National Center for Environmental Assessment (MD-
52), U.S. Environmental Protection Agency, Research Triangle Park, NC  27711

Ms. Diane H. Ray—Program Analyst, National Center for Environmental Assessment
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Ms. Eleanor Speh—Office Manager, Environmental Media Assessment Branch, National Center
for Environmental Assessment (MD-52), U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711

Ms. Donna Wicker—Administrative Officer, National Center for Environmental Assessment
(MD-52), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Mr. Richard Wilson—Clerk, National Center for Environmental Assessment (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Document Production Staff

Mr. John R. Barton—Document Processing Coordinator
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713

Ms. Yvonne Harrison—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC 27713
October 1999                            I-xxx       DRAFT-DO NOT QUOTE OR CITE

-------
             U.S. ENVIRONMENTAL PROTECTION AGENCY
 PROJECT TEAM FOR DEVELOPMENT OF AIR QUALITY CRITERIA
                       FOR PARTICULATE MATTER
                                     (cont'd)
Ms. Bettye Kirkland—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. David E. Leonhard—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Carolyn T. Perry—Word Processor
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Veda E. Williams—Graphic Artist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Technical Reference Staff

Mr. R. David Belton—Reference Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. John Bennett—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. William Hardman—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Ms. Sandra L. Hughey—Technical Information Specialist
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713

Mr. Jian Ping Yu—Reference Retrieval and Database Entry Clerk
OAO Corporation, Chapel Hill-Nelson Highway, Beta Building, Suite 210, Durham, NC  27713
October 1999                           I-xxxi       DRAFT-DO NOT QUOTE OR CITE

-------
 i                            1.  EXECUTIVE SUMMARY
 2
 3
 4      1.1 INTRODUCTION
 5      1.1.1 Purpose of the Document
 6           The purpose of this document, Air Quality Criteria for Particulate Matter, is to present air
 7      quality criteria for particulate matter (PM) in accordance with Clean Air Act (CAA) Sections 108
 8      and 109, which govern establishment, review, and revision of U.S. National Ambient Air Quality
 9      Standards (NAAQS) as follows:
10      • Section 108 directs the U.S. Environmental Protection Agency (EPA) Administrator to list
11       pollutants that may reasonably be anticipated to endanger public health or welfare and to issue
12       air quality criteria for them. The air quality criteria are to reflect the latest scientific
13       information useful in indicating the kind and extent of all exposure-related effects on public
14       health and welfare expected from the presence of the pollutant in ambient air.
15      • Section 109 directs the EPA Administrator to set and periodically revise, as appropriate,
16       (a) primary NAAQS to protect against adverse health effects of listed criteria pollutants among
17       sensitive population groups, with an adequate margin of safety,  and (b) secondary NAAQS to
18       protect against welfare effects (e.g., impacts on vegetation, crops, ecosystems, visibility,
19       climate, man-made materials, etc.).
20      • Section 109 of the CAA also requires periodic review and, if appropriate, revision of existing
21       criteria and standards. Also, an independent committee of non-EPA experts, the Clean Air
22       Scientific Advisory Committee (CASAC), is to provide the EPA Administrator advice and
23       recommendations regarding the scientific soundness and appropriateness of criteria and
24       NAAQS.
25           To meet these CAA mandates, this document assesses the latest scientific information
26      useful in deriving criteria as scientific bases for decisions on possible revision of current
27      PM NAAQS.  A separate EPA PM Staff Paper will draw upon assessments in this document,
28      together with other information, to develop exposure/risk analyses and to pose options for
29      consideration by the EPA Administrator with regard to possible retention or revision of the
30      PM NAAQS.

        October 1999                              1-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.1.2  Organization of the Document
 2          The present document is organized into nine chapters, as follows:
 3      • This Executive Summary (Chapter 1) summarizes key points from the ensuing chapters.
 4      • Chapter 2 provides a general introduction, including a brief summary of the history of the PM
 5       NAAQS and an overview of issues, methods and procedures used to prepare this document.
 6      • Chapters 3 through 5 provide background information on air quality and exposure aspects, to
 7       help to place the succeeding discussions of PM effects into perspective.
 8      • Chapter 6 discusses community epidemiology information and Chapter 7 discusses dosimetry
 9       and toxicology of PM.
10      • Chapter 8 provides an integrative synthesis of key points from those health chapters (6 & 7)
11       and other preceding air quality and exposure chapters.
12      • Chapter 9 deals with environmental effects of PM on vegetation and ecosystems; visibility;
13       climate; and manmade materials, as well as economic impacts of such effects.
14          It should be noted that new research results that have become available since early 1996
15      (when the last previous PM criteria assessment was completed) through mid-1999 are assessed in
16      this First External Review Draft of the revised PM Air Quality Criteria Document (PM AQCD).
17      Extensive further new research results, expected to be published during the next 6 to 9 mo
18      (including many from a vastly expanded EPA PM Research Program) will be assessed in the
19      Second External Review Draft of this document to be released in mid-2000 for public comment
20      and CASAC review. Thus, key findings and conclusions summarized below must be considered
21      to be only provisional and subject to change, as appropriate, due to consideration of new research
22      in the next draft.
23
24
25      1.2  AIR QUALITY AND EXPOSURE ASPECTS
26          The document's discussion of air quality and exposure aspects considers chemistry and
27      physics of atmospheric PM; analytical techniques for measuring PM mass, size, and chemical
28      composition; sources of ambient PM in the United States; temporal/spatial variability and trends
29      in ambient U.S. PM levels; and human exposure relationships. Key findings are summarized in
30      the next six sections. Overall, the atmospheric sciences and air quality information provides

        October 1999                             1 -2         DRAFT-DO NOT QUOTE OR CITE

-------
 1      further evidence substantiating the 1996 PM AQCD conclusion that distinctions between fine
 2      and coarse mode particles (in terms of sources of emission, formation mechanisms, atmospheric
 3      transformation and transport distances, and air quality patterns) warrant fine and coarse PM being
 4      viewed as separate subclasses of ambient PM.
 5
 6      1.2.1 Chemistry and Physics of Atmospheric Particles
 7      • Airborne PM is not a single pollutant, but rather is a mixture of many subclasses of pollutants
 8       with each subclass containing many different chemical species. Atmospheric PM occurs
 9       naturally as fine-mode and coarse-mode particles that, in addition to falling into different size
10       ranges, differ in formation mechanisms, chemical composition, sources, and exposure
11       relationships.
12      • Fine-mode PM is derived from combustion material that has volatilized and then condensed to
13       form primary PM or from precursor gases reacting in the atmosphere to form secondary PM.
14       New fine-mode particles are formed by the nucleation of gas phase species, and grow by
15       coagulation (existing particles combining) or condensation (gases condensing on existing
16       particles). Fine particles are composed of freshly generated nuclei-mode particles, also called
17       ultrafine or nanoparticles, and an accumulation mode, so called because particles grow into and
18       remain in that mode.
19      • Coarse-mode PM, in contrast, is formed by crushing, grinding, and abrasion of surfaces, which
20       breaks large pieces of material into smaller pieces.  These particles are then suspended by the
21       wind or by anthropogenic activity. Energy considerations limit the break-up of large particles
22       and small particle aggregates generally to a minimum size of about 1 //m in diameter. Mining
23       and agricultural activities are examples of anthropogenic sources of coarse-mode particles.
24       Fungal spores, pollen, and plant and insect fragments  are examples of natural bioaerosols also
25       suspended as coarse-mode particles.
26      • Within atmospheric particle modes, the distribution of particle number, surface, volume, and
27       mass by diameter is frequently approximated by lognormal distributions.  Aerodynamic
28       diameter, dae, which depends on particle density and is defined as the diameter of a particle with
29       the same settling velocity as a spherical particle with unit density (1 g/cm3), is often used to
30       describe particle size. Typical values of the mass median aerodynamic diameters (MMAD) are

        October 1999                               1 -3        DRAFT-DO NOT QUOTE OR CITE

-------
 1       0.05 to 0.07 //m for the nuclei mode, 0.3 to 0.7 //m for the accumulation mode, and 6 to 20 //m
 2       for the coarse mode.  At high relative humidities or in air containing evaporating fog or cloud
 3       droplets, the accumulation mode may be split into a droplet mode (MMAD = 0.5 to 0.8 //m)
 4       and a condensation mode (MMAD = 0.2 to 0.3 //m).
 5      • Research studies use (a) impactors to determine mass and composition as a function of size
 6       over a wide range and (b) particle counting devices to determine number of particles as a
 7       function of size. Such studies indicate an atmospheric bimodal distribution of fine and coarse
 8       particle mass with a minimum in the distribution between 1 and 3 //m dae. Routine monitoring
 9       studies generally measure thoracic particles or PM10 (upper size limited by a 50% cut at 10 //m
10       dae).  Research studies and new monitoring studies measure fine particles or PM2 5 (upper size
11       limited by a 50% cut point at 2.5 //m dae) and the coarse fraction of PM10, i.e., the difference
12       between PM10 and PM25 (PM10_2 5).  Cut points are not perfectly sharp for any of these PM
13       indicators;  some particles larger than the 50% cutpoint are collected and some particles smaller
14       than the 50% cutpoint are not retained.
15      • The terms "fine" and "coarse" were originally intended to apply to the two major atmospheric
16       particle distributions which overlap in the size range between 1 and 3 //m diameter. Now, fine
17       has come to be often associated with the PM25 fraction and coarse is often used to refer to
18       PM10_25.  However, PM25 may also contain, in addition to the fine-particle mode, some  of the
19       lower-size tail of the coarse particle mode between about 1 and 2.5 //m dae. Conversely, under
20       high relative humidity conditions, the larger fine particles in the accumulation mode may also
21       extend into the 1 to 3 //m dae range.
22      • Three approaches are used to classify particles by size: (1) modes, based on formation
23       mechanisms and the modal structure observed in the atmosphere,  e.g., nuclei and accumulation
24       modes (which comprise the fine-particle mode)  and the coarse-particle mode; (2) cut point,
25       based on the 50% cut point of the specific  sampling device; (3) dosimetry, based on the ability
26       of particles to enter certain regions of the respiratory tract; and (4) regulatory, based on
27       instrument configuration or 50% cut-points, e.g., high volume sampler, PM10, and PM2  5.
28
        October 1999                              1 -4        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.2.2 Sources of Airborne Particles in the United States
 2      • The chemical complexity of airborne particles requires that the composition and sources of a
 3       large number of primary and secondary components be considered. Major components of fine
 4       particles are: sulfate, strong acid, ammonium, nitrate, organic compounds, trace elements
 5       (including metals), elemental carbon, and water.
 6      • Primary particles are emitted directly from sources. Secondary particles are formed from
 7       atmospheric reactions of sulfur dioxide (SO2), nitrogen oxides (NOX), and certain organic
 8       compounds. NO reacts with ozone (O3) to form NO2. SO2 and NO2 react with hydroxy radical
 9       (OH) during the daytime to form sulfuric and nitric acid. During the nighttime NO2 reacts with
10       ozone and forms nitric acid through a sequence of reactions involving the nitrate radical (NO3).
11       These acids may react further with ammonia to form ammonium sulfates and nitrates. Some
12       types of higher molecular weight organic compounds react with OH radicals, and olefinic
13       compounds also react with ozone, to form oxygenated organic compounds which nucleate or
14       can condense onto existing particles. SO2 also dissolves in cloud and fog droplets where it may
15       react with dissolved O3, H2O2, or, if catalyzed by certain metals, with O2, yielding sulfuric acid
16       or sulfates, that lead to PM when the droplet evaporates.
17      • Receptor modeling has proven to be a useful method for identifying contributions of different
18       types of sources especially for the primary components of ambient PM. Apportionment of
19       secondary PM is more difficult because it requires consideration of atmospheric reaction
20       processes and rates.  Results from western U.S. sites indicate that fugitive dust, motor vehicles,
21       and wood smoke are the major contributors to ambient PM samples there, while results from
22       eastern U.S. sites indicate that stationary combustion, motor vehicles and fugitive dust are
23       major contributors to ambient PM samples in the East. Sulfate and organic carbon are the
24       major secondary components in the East, while nitrates and organic carbon are the major
25       secondary components in the West.
26      • Fine and coarse particles have distinctly different sources, both natural and anthropogenic.
27       Therefore different control strategies are likely to be needed, depending on whether fine or
28       coarse particles (or both) are selected for control.
29
        October 1999                              1 -5        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.2.3 Atmospheric Transport and Fate of Airborne Particles
 2      • Primary and secondary fine particles have long lifetimes in the atmosphere (days to weeks) and
 3       travel long distances (hundreds to thousands of kilometers). They tend to be uniformly
 4       distributed over urban areas and larger regions, especially in the eastern United States. As a
 5       result, they are not easily traced back to their individual sources.
 6      • Coarse particles normally have shorter lifetimes (minutes to hours) and only travel short
 7       distances (<10's of km). Therefore, coarse particles tend to be unevenly distributed across
 8       urban areas and tend to have more localized effects than fine particles. However, dust storms
 9       occasionally cause long range transport of the smaller coarse-mode particles.
10
11      1.2.4 Airborne Particle Measurement Methods
12      • Measurements of ambient PM mass and chemical composition are needed to determine
13       attainment of standards; to guide attainment of a standard (including determination of source
14       categories and validation of air quality models); and to determine health, ecological, and
15       radiative effects. A comprehensive approach requires a combination of analytical techniques to
16       assess: (1) mass, (2) crustal and trace elements, (3) water-soluble ionic species including
17       strong acidity, (4) elemental carbon, and (5) organic compounds.
18      • There are no calibration standards for suspended particle mass; therefore, the accuracy of
19       particle mass measurements cannot be definitively determined.  The precision of particle mass
20       measurements can be determined by comparing results from collocated samplers. When using
21       different measurement techniques, samplers of different design or manufacturer, and, in some
22       cases, when using identical systems of different age or cleanliness, substantial biases of 50% or
23       more have been observed. Mass concentration measurements with a precision close to  10%
24       have been obtained with collocated samplers of identical design and same time since cleaning.
25      • Available  technology allows accurate (±10 to 15%) measurement of several of the major
26       components  of coarse and fine particles (crustal and trace elements, sulfates, and strong
27       acidity). However, collection and measurement technologies for elemental carbon, organic
28       carbon, and nitrates are not as well established.
29      • Semivolatile organic compounds and semivolatile ammonium compounds (such as NH4NO3)
30       may be lost by volatilization during sampling.  Such losses may be very important in

        October 1999                              1 -6         DRAFT-DO NOT QUOTE OR CITE

-------
 1       woodsmoke impacted areas for organic compounds or in agricultural and other areas where low
 2       sulfate and high ammonia lead to high NH4NO3 concentrations. Hence, while the Federal
 3       Reference Methods for PM10 and PM25 give precise (± 10%) measurements of "equilibrated
 4       mass", loss of semivolatile PM and possible retention of some particle-bound water contribute
 5       to uncertainly in measurement of the mass of PM as it exists suspended in the atmosphere.
 6      • Intercomparisons, using different techniques and samplers of different designs, coupled with
 7       mass balance studies (relating the sum of components to the measured mass), provide a method
 8       for gaining confidence in the reliability of PM measurements.
 9
10      1.2.5  Ambient U.S. PM Concentrations: Regional Patterns and Trends
11      • Particle mass data have been collected at a number of rural, suburban, and urban sites across
12       the United States by various local, state, and national programs. The data have been stored in
13       the Aerometric Information Retrieval System (AIRS).  Data have also been collected at remote
14       sites as part of the IMPROVE and NESCAUM networks.  An extensive analysis of this data
15       was reported in the  1996 Air Quality Criteria Document for Particulate Matter (PM AQCD).
16      • Information on trends of PM25 (fine) and PM10_25 (coarse) were also examined for Philadelphia,
17       several AIRS sites, the Harvard Six-City sites, and California sites. However, such data is still
18       not sufficient, either in number of sites or number of years per site, to demonstrate differential
19       trends in fine or coarse PM.
20
21      1.2.6  Human PM Exposure
22          Chapter 5 examines:  ambient particulate matter (PM)  air quality; that portion of ambient
23      PM which penetrates into indoor microenvironments; and, to a lesser extent, the contributions of
24      sources of non-ambient PM to total PM exposure. This is to aid in interpretation of acute and
25      chronic epidemiology studies discussed in Chapter 6, in which ambient PM concentrations are
26      assumed to be an indicator, or a surrogate, for mean community exposure to PM of ambient
27      origin, or an individual's exposure to ambient PM. Thus, this chapter has three objectives:
28      (a) To provide a review of pertinent studies of personal exposures to total PM (ambient PM plus
29         non-ambient PM).
        October 1999                              1 -7        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (b) To evaluate linkages of human exposure to PM of ambient origin estimated from
 2         concentrations of PM measured at a fixed-site monitor located at some central location in a
 3         community under study.
 4      (c) To quantify the contribution of PM of ambient origin to total personal PM exposure.
 5          In meeting those objectives, Chapter 5 has reached the following provisional conclusions as
 6      supported by the evidence cited:
 7      • Human exposure to PM of ambient origin for individuals in a community is often highly
 8       correlated (R2 > 0.5) in time with concentrations of PM of ambient origin of the same size as
 9       measured in that community.
10      • Longitudinal correlation coefficients for ambient concentrations of fine PM (AD < 2.5 //m)
11       with personal exposures to ambient fine PM are greater than the corresponding correlations for
12       the coarse fraction of ambient PM10 (2.5 //m < AD < 10 //m), as shown by studies of ambient
13       sulfate concentrations and sulfate exposures.
14      • People in a community surrounding an ambient monitoring station, over time, are exposed to
15       relatively similar mixtures and concentrations of ambient PM2 5.
16      • People in a community are exposed to widely different mixtures and concentrations of
17       non-ambient PM due to the diversity of smoking habits, personal activities such as hobbies,
18       residential furnishings and appliances, and varying occupations.
19      • Exposures to PM of indoor origin appear to be uncorrelated with exposures to PM of ambient
20       origin.
21      • The correlation of a single individual's sequence of daily personal  exposures to total PM and
22       ambient PM concentrations will be greater than the correlation that would occur had a different
23       person been monitored on each of the same days (e.g., one person monitored consecutively for
24       n days vs sequentially monitoring n different people, each for one day, over n days).
25      • Ambient PM in the U.S. has average annual correlations with the ambient gaseous pollutants
26       CO,  ozone, NO2, and SO2 of order r = 0.25 with a standard deviation of order 0.25.
27      • Although exposures to PM from indoor sources and occupational activities may not be
28       correlated with ambient PM concentrations, these non-ambient PM species may possibly act as
29       effect modifiers by making subjects more or less susceptible to exposure to PM of ambient
30       origin.

        October 1999                               1 -8         DRAFT-DO NOT QUOTE OR CITE

-------
 1      • There are only limited data available, from non-probability samples, to evaluate how well the
 2       exposures to PM of ambient origin for susceptible subgroups correlate with the ambient PM
 3       concentrations of similar AD size range as measured in their community.
 4           The newly available PM exposure information, overall, appears to further substantiate that
 5      ambient PM measurements (especially for fine PM) made at community monitoring sites likely
 6      index well personal human exposures  (both outdoors and indoors) to PM of ambient origin. The
 7      community monitoring data, used as the main PM exposure indices in PM community
 8      epidemiology studies, is not highly correlated with human exposure to PM from indoor sources,
 9      making it unlikely that exposure to PM from indoor sources confound reported ambient
10      PM-health effects associations.
11
12
13      1.3 DOSIMETRY
14           Key findings derived from the assessment of dosimetry information include:
15      • Particles may be deposited in (a) the extrathoracic airways (i.e., mouth, nose, and larynx); (b) in
16       airways of the tracheobronchial region; and (c) in the alveolar region where gas exchange
17       occurs.  There are differences in deposition mechanisms and dose distribution in each of these
18       areas that are dependent on particle size and airway geometry. Whereas, impaction is an
19       important deposition mechanism in large extra- and intrathoracic airways at higher flows,
20       sedimentation and diffusion are more important at low flow rates in smaller airways.
21      • Respiratory tract deposition patterns are dependent on particle size and distribution
22       (as indicated by the mass median aerodynamic diameter and the geometric standard deviation)
23       within the inspired air.  Biologic effects may be a function not only of particle mass  deposition
24       but also of particle number or the total surface area of the particles.
25      • Various host factors have been shown to influence particle deposition patterns, including
26       airway dimensions (size and shape),  breathing pattern (flow and volume), and the presence of
27       obstructive or inflammatory airway disease. Particle deposition in the extrathoracic  region is
28       highest during nasal breathing, and is greatest in small children and least in adults. Increased
29       total ventilation and increased oral breathing leads to greater deposition of coarse particles in
        October 1999                              1 -9         DRAFT-DO NOT QUOTE OR CITE

-------
 1       the thoracic airways.  Obstructive airway disease, such as asthma, emphysema, and chronic
 2       bronchitis, results in increased deposition of particles in the lower respiratory tract.
 3      • Acute effects of PM are probably best related to deposited dose, whereas chronic effects may
 4       be related to cumulative or retained dose. Retention of particles is a function of deposition site,
 5       clearance of particles by macrophages or the mucociliary system, and particle characteristics,
 6       especially solubility.  Chronic effects may also arise from recurring cycles of pulmonary injury
 7       and repair.
 8      • There are substantial differences among laboratory animal species with regard to the
 9       inhalability of different sized particles as well as quantitative and qualitative differences in
10       airway geometry  and branching patterns.
11      • Extrathoracic deposition of ultrafine particles (<0.100 //m) is very high, despite their small
12       size.  Estimates of deposition range from 50% for oral breathing to >90% for nasal breathing.
13       Enhanced deposition of both ultrafine and coarse particles occurs at branching points within the
14       intrathoracic airways.
15      • Clearance is either absorptive (dissolution) or nonabsorptive (transport of intact particles).
16       Deposited particles may be dissolved in body fluids, taken up by phagocytic cells, or
17       transported by the mucociliary system. Retained particles tend to be small (<2.5 //m) and
18       poorly soluble, e.g., silica, metals).
19      • Tracheobroncial clearance has both a fast and a slow component.  Translocation of poorly
20       soluble PM to the lymph lodes takes a few days and is more rapid for smaller (<2 //m)
21       particles; elimination rates of these retained particles are on the order of years. People with
22       COPD have increased particle retention and use cough to augment mucociliary clearance.
23      • In order to extrapolate experimental results between different species, a number of factors must
24       be considered such as: airway size; airway branching asymmetry; inhalability; deposition;
25       deposition/surface area, hot-spots; number deposition versus mass deposition; dose-response;
26       and clearance retention (where and how long).
27      • Mathematical models are available to predict deposition, clearance, and retention of particles.
28       Two models, the  ICRP and NCRP, were extensively discussed in the 1996 PM AQCD.
29
30

        October 1999                              1-10        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.4 PARTICULATE MATTER HEALTH EFFECTS
 2      1.4.1  Epidemiology Findings
 3          The epidemiology evidence about the health effects of ambient PM has expanded greatly
 4      since the 1996 PM Air Quality Criteria Document (PM AQCD).  The most important
 5      enhancements in information include:
 6      • New studies of health endpoints using ambient PM10 and closely related mass concentration
 7       indices such as PM13 and PM7;
 8      • New studies on a variety of endpoints have evaluated effects of the ambient coarse PM fraction
 9       (PM10_2 5), the ambient fine particle fraction (PM2 5), and even ambient ultrafine particle mass
10       concentrations (PMj and smaller);
11      • New studies in which the relationship of some health endpoints to ambient particle number
12       concentrations were evaluated;
13      • Additional studies which evaluated the sensitivity of estimated  PM effects to the inclusion of
14       gaseous co-pollutants in the model;
15      • Preliminary attempts to evaluate the effects of air pollutant combinations or mixtures including
16       PM components, based on empirical combinations (factor analysis) or source profiles;
17      • New studies of infants and children as a potentially susceptible  population;
18      • Further studies of cardiovascular endpoints associated with PM exposures;
19      • New studies on asthma and other respiratory conditions exacerbated by PM exposure.
20          This additional information does not yet allow for full resolution of all outstanding key
21      issues in PM air pollution epidemiology. Table 1-1 presents provisional conclusions about
22      alternative hypothesis that may affect the interpretation and synthesis of epidemiology study
23      results. The authors conclude that multiplicity  of findings about PM health effects suggest that
24      exposure to ambient PM at current concentrations may cause serious adverse health effects, but
25      that the quantitative magnitude of the effects depends on several  environmental and biological
26      factors whose role is  not yet known.  That is, current levels of ambient PM may be harmful to
27      human health, but not necessarily equally harmful everywhere or  at all times.
28          Some more specific provisional key findings emerging thus far from assessment of new
29      epidemiology study results in this draft document are:
30

        October 1999                             1-11         DRAFT-DO NOT QUOTE OR CITE

-------
                         TABLE 1-1.  PROVISIONAL CONCLUSIONS ABOUT
                       ALTERNATIVE HYPOTHESES THAT MAY AFFECT THE
                              SYNTHESIS OF EPIDEMIOLOGY STUDIES
         Alternative hypotheses
         Adverse health effects depend only
         on ambient PM size range at
         current PM concentrations
         Adverse health effects depend on
         ambient PM with specific physical
         properties or current ambient PM
         composition
Adverse health effects depend only
on current ambient PM,
independent of co-factors

Not likely.  Adverse health effects
from coarse particles may occur at
some sites, not others.
Possible. Adverse health effects
from ambient PM of a given size
may be different in sites where PM
has different physical properties or
composition with same PM size
range.	
      Adverse health effects depend
      on current ambient PM and on
      environmental co-factors

      Possible. Adverse health effects
      from ambient PM are different in
      sites where ambient PM has
      different co-factors with same PM
      range.

      Probable.  Adverse health effects
      from ambient PM are different in
      sites where PM has different
      physical properties, composition, or
      co-factors, even in the same
      ambient PM size range	
 1      • New studies suggest that infants and children may represent an additional subgroup at special
 2       risk for ambient PM exposure effects.  The new results most clearly indicate that children

 3       appear to be susceptible to respiratory effects associated with ambient PM exposures, including
 4       exacerbation of asthma and respiratory symptoms in school-age children. A few studies also
 5       report ambient PM to be associated with intrauterine growth reduction and low birth weight
 6       (known infant health risk factors) and excess infant mortality. However, some studies found

 7       that PM is not as good a predictor of these endpoints as other pollutants (e.g., CO), and no
 8       toxicologic evidence has yet been advanced to support biological plausibility of such effects
 9       due to ambient PM  or to identify pathophysiologic mechanisms involved.
10      • Cardiovascular causes of death and hospitalization in older adults may also be a significant

11       component of PM-attributable mortality, as well as respiratory causes
12      • PM health effects have been reported to be associated with several different ambient PM size

13       fractions (ultrafine,  fine, coarse), but some health effects may be absent from some ambient PM
14       mass fractions under some circumstances.
15      • PM health effects may occur at different time scales for exposure to PM10 or PM25, from (a)
16       short -term responses to daily exposure through (b) larger excess mortality associated with
        October 1999
            1-12
DRAFT-DO NOT QUOTE OR CITE

-------
 1       medium term exposures (15 to 120 day averages) to (c) excess morbidity or mortality
 2       associated with long-term (multi-year) exposures.
 3      • Adverse health effects attributable to PM2 5 in short-term exposure studies are at times seriously
 4       confounded by exposure to gaseous co-pollutants making it difficult to estimate, quantitatively
 5       that portion of the risk attributable to: PM acting alone; PM acting in combination with
 6       gaseous co-pollutants; the gaseous co-pollutants per se; or the overall ambient pollutant mix.
 7
 8      1.4.2 Toxicologic Responses to PM in Animals and Humans
 9           Data on the toxicology of PM are derived from controlled inhalation exposure studies of
10      humans and animals, intratracheal instillation studies in humans and animals, occupational
11      studies, and ex vivo studies of human and animal cells grown in culture. The human or animal
12      populations (cells) studied vary by age, health status, or other host factors. Exposures vary by
13      duration, mass or number concentration, chemical composition and size of the PM in addition to
14      other exposure variables (e.g., temperature, humidity, activity levels, etc.). Responses to PM in
15      the respiratory tract are dependent on the physiological status of the host as well as on
16      translocation of PM or PM constituents to other sites. Ex vivo studies provide important
17      additional information regarding the mechanism of action of PM or PM constituents on cells or
18      cellular components.
19           Responses to acidic aerosols (sulfuric acid, sulfates, nitrates) have been comprehensively
20      reviewed in previous documents. Much of the newly available research focuses on
21      combustion-related PM such as concentrated ambient particles (CAPs) from urban air or various
22      forms of fly ash PM.
23      • Acute exposures to metal particles can cause inflammatory responses in the respiratory tract of
24       humans and animals. The effective exposure levels are typically much higher than ambient air
25       metal concentrations in the U.S. atmosphere. Endotoxin, a lipopolysacharide associated with
26       bacteria, causes inflammation in human occupational exposures at concentrations that are also
27       much higher than in the ambient air.
28      • Combustion-related particles (fly ash and urban air particles) cause a spectrum of responses in
29       the airways of animals. These include inflammation, cellular injury, and increased
30       permeability.  Metal components (V, Cu, Zn, Fe, Ni) of combustion particles have been

        October 1999                              1-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1       implicated in the responses, possibly related to oxidant production and release of intercellular
 2       signaling molecules (cytokines).
 3      • Cells primed by inflammatory mediators show increased cytokine responses to PM.
 4       Combustion-related particles may cause increased oxidant production in in vitro systems,
 5       presumably related to metal components of particles. Combustion-related particles cause
 6       damage to cells in vitro. Responses include impaired macrophage phagocytosis and altered
 7       permeability.
 8
 9      Susceptibility
10      • Chemically or pharmacologically treated rat "models of cardiorespiratory disease" as well as
11       older rats  demonstrate increased pulmonary responses to combustion-related particles.
12      • Diesel particulate matter and oil fly ash (ROFA) may augment responses to antigens in allergic
13       animals or humans.  These studies provide a plausible mechanism for an association between
14       combustion PM exposure and exacerbation of asthma.
15      • Inhaled or instilled particles can have systemic effects, especially on the cardiovascular system
16       which in certain circumstances can be lethal.
17
18      Mixtures
19      • Mixtures of ozone and PM (urban PM, sulfate aerosols, ultrafine carbon) may cause enhanced
20       effects on lung cells, increased inflammation, and decrements in human lung function.
21
22      Mechanisms
23      • A number of studies indicate that reactive oxidant species (ROS) play a role in PM-induced
24       responses. Catalysis of ROS is likely related to soluble metals in PM.
25      • ROFA and urban PM can induce apoptosis  (programmed cell death) of human alveolar
26       macrophages.
27      • Studies on ultrafine  compared to fine particles indicate a greater response to ultrafines in terms
28       of airway  inflammation, an effect that appears to be related to their greater surface area.
29
        October 1999                              1-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.4.3 Population Groups at Risk
 2           Susceptibility can be affected by factors which influence dosimetry or the response of
 3      tissues to particle burdens. Host factors that may increase the susceptibility to PM include both
 4      changes in physiologic factors affecting respiratory tract deposition and pathophysiologic factors
 5      affecting response.
 6      • Susceptible groups most clearly at special risk for PM effects include the elderly and those with
 7       cardiopulmonary disease, based on available epidemiology findings.
 8      • Epidemiology findings indicate that risk of mortality and morbidity due to lower respiratory
 9       disease (e.g. pneumonia) is increased by ambient PM exposure. This may be due to
10       exacerbation, by PM,  of already existing respiratory disease. PM may also increase
11       susceptibility to infectious disease by decreasing clearance, impairing macrophage function, or
12       through other specific and nonspecific effects on the immune system.  The epidemiologic
13       findings also indicate  that individuals with preexisting infectious respiratory disease (e.g.,
14       pneumonia) are at  increased risk for PM effects.
15      • Epidemiologic findings indicate that ambient PM exposures are also associated with increased
16       risk for mortality and hospitalization due to  cardiovascular causes.  Cardiac arrhythmia has
17       been hypothesized as being involved in mortality due to acute PM exposure.
18      • Studies of infants and children indicate that they are a potentially susceptible population.
19       Panel studies on asthma and other respiratory conditions show exacerbation by PM exposure.
20       Children are susceptible to respiratory effects associated with PM exposure from pre-natal and
21       post-natal effects through exacerbation of asthma and respiratory symptoms in school age
22       children.
23
24
25      1.5 ENVIRONMENTAL EFFECTS
26      1.5.1 Vegetation and Ecosystem
27      • Human existence on this planet depends on  the life-support service that ecosystems provide.
28       Ecosystem structure and function play an essential role in providing two types of benefits to
29       society. From the  structural aspects (biodiversity, abundance, mass and arrangement of
30       species), an ecosystem provides: (1) products with market value such  as fish, minerals, forage,

        October 1999                              1-15        DRAFT-DO NOT QUOTE OR CITE

-------
 1       forest products, biomass fuels, natural fiber, and many pharmaceuticals, and the genetic
 2       resources of valuable species (e.g., plants for crops and timber and animals for domestication);
 3       and (2) the use and appreciation of ecosystem for recreation, aesthetic enjoyment, and study.
 4      • Ecosystems maintain clean water, pure air, a green earth, and a balance of creatures, the
 5       functions that enable humans to survive. The benefits they impart include absorption and
 6       breakdown of pollutants, cycling of nutrients, binding of soil, degradation of organic waste,
 7       maintenance of a balance of gases in the air, regulation of radiation balance, climate, and the
 8       fixation of solar energy.
 9      • Concern has risen in recent years concerning the integrity of ecosystems because there are few
10       ecosystems on planet earth today that are not influenced by humans. For this reason,
11       understanding the effects of PM deposition and its impact on vegetation and ecosystems is of
12       prime importance.
13      • The criteria pollutant presently defined as PM10 has no particular relevance to particulate effects
14       on vegetation for which chemical composition is more relevant than mass.  The PM whose
15       effects on vegetation are considered in this chapter is not a single pollutant, but a
16       heterogeneous mixture of particles of differing in size, origin, and chemical constituents.
17       Exposure to a given mass concentration of PM10 may, depending on the particular mix of
18       deposited particles, lead to widely differing phytotoxic responses.  This variable has not been
19       adequately characterized.
20      • Atmospheric deposition of particles to ecosystems takes place via both wet and dry processes
21       through the three major routes indicated below:
22       (1) Precipitation scavenging in which particles are deposited in rain and  snow
23       (2) Fog, cloud-water, and mist interception
24       (3) Dry deposition, a much slower, yet more continuous removal to surfaces.
25      • Deposition of heavy metal particles to ecosystems occurs by wet and dry processes. Dry
26       deposition is considered more effective  for coarse particles of natural origin and elements such
27       as iron and manganese, whereas wet deposition generally is more effective for fine particles of
28       atmospheric origin and elements such as cadmium, chromium, lead, nickel, and vanadium.
29      • The actual importance of wet versus dry deposition, however, is highly variable, depending on
30       the type of ecosystem, location and elevation. The range of particle sizes, the variety of
31       chemical constituents in airborne PM, and the diversity of canopy surfaces, have slowed
        October 1999                              1-16        DRAFT-DO NOT QUOTE OR CITE

-------
 1       progress in both prediction and measurement of dry particulate deposition. Wet deposition
 2       generally is confounded by fewer factors and has been easier to quantify.
 3      • Emphasis in this chapter has been placed on the effects of PM on natural plants and terrestrial
 4       ecosystems. Neither nitrate or sulfate deposition on crops is discussed as they are frequently
 5       added in fertilizers.  Except for nitrogen and sulfur-containing compounds and their effects in
 6       acidic precipitation, information concerning the effects of PM on crops is not readily available.
 7      • Particulate matter when transferred from the atmosphere to plant surfaces may cause direct
 8       effects when they (1) reside on the leaf, twig or bark surface for an extended period; (2) are
 9       taken up through the leaf surface; or indirect effects when (3) removed from the plant via
10       suspension to  the atmosphere, washing by rainfall, or by litter-fall with  subsequent transfer to
11       the soil.
12      • Deposition of PM on above-ground plant parts can have either a physical and or chemical
13       impact, or both. The effects of "inert" PM are mainly physical, while the effects of toxic
14       particles are both chemical and physical. The effects of dust deposited on plant surfaces or on
15       soil are more likely to be associated with their chemistry than with the mass deposited particles
16       and are usually of more importance than any physical effects.  Studies of the direct effects of
17       chemical additions to foliage in particulate deposition have found little or no effects of PM on
18       foliar processes unless exposure levels were significantly higher than would typically be
19       experienced in the ambient environment.
20      • The majority of the easily identifiable direct and indirect effects, other than climate, occur in
21       severely polluted areas around heavily industrialized point sources such as limestone quarries,
22       cement kilns,  iron, lead, and various smelting factories. Indirect impacts are usually the most
23       significant because they can alter nutrient cycling in the soil and inhibit plant uptake of
24       nutrients.
25      • Most PM deposited on vegetation eventually enters the soil environment which is one of the
26       most dynamic sites of biological interaction in nature. The impact of particulate deposition on
27       plants results from changes  in the soil environment and plant nutrient uptake. These changes
28       determine plant and ecosystem response.
29      • Bacteria and fungi in the soil have an important role in plant nutrition. Bacteria are essential
30       components of the nitrogen and sulfur  cycles that make these  elements available for plant
31       uptake. Fungi form mycorrhizae, a mutualistic, symbiotic relationship, that is integral in the
        October 1999                               1-17        DRAFT-DO NOT QUOTE OR CITE

-------
 1       uptake of mineral nutrients. The impact of nitrates, sulfates and metals in PM is determined by
 2       their affect on the growth and functions of the bacteria and fungi involved in making nutrients
 3       available for plant uptake.
 4      • The major impact of PM on ecosystems is indirect and occurs in the soil through the deposition
 5       of nitrates and sulfates and the acidifying effects of the H+ ion associated with these
 6       compounds in wet and dry deposition.
 7      •  Intensive research over nearly a decade indicates that although the soils of most North
 8       American forests are nitrogen limited, there are forests that exhibit symptoms of excess
 9       nitrogen. Nitrogen saturation results in a progressive syndrome of concurrent responses to
10       long-term, chronic nitrogen deposition. As nitrogen reaches saturation in temperate-zone
11       forests, there is decrease in nitrogen mineralization and an increase in the trends of foliar
12       Mg:N and Ca:Al ratios.  Preliminary evidence suggests some forests may decline in
13       productivity and experience greater mortality as a result of chronic nitrogen deposition.
14      • Increases in soil nitrogen plays a selective role.  Plant succession patterns and biodiversity in
15       some ecosystems are significantly affected by chronic nitrogen additions.  Long-term nitrogen
16       fertilization studies in both New England and Europe suggest that some forests receiving
17       chronic inputs of nitrogen may decline in productivity and experience greater mortality.
18       Studies also suggest that declining coniferous forest stands with slow nitrogen cycling may be
19       replaced by deciduous fast-growing forests that cycle nitrogen rapidly. Excess nitrogen inputs
20       to unmanaged heathlands in the Netherlands has resulted in nitrophilous grass species replacing
21       slower growing heath species. Over the past several decades the composition of plants in the
22       forest herb layers had been shifting toward species commonly found on nitrogen-rich areas.
23       It also was observed that the fruiting bodies of mycorrhizal fungi had decreased in number.
24      • The effects of excessive  deposition of nitrogen, particularly NH3 and NH4+ Deposition have
25       lead to changes in Dutch heathlands via:  (1) acidification of the soil and the loss of cations  of
26       K+, Ca 2+ and Mg 2+; and (2) nitrogen enrichment which results in increased plant growth rates
27       and altered competitive relationships. Alteration of any of a number of parameters
28       (e-g-> increased nitrogen) can alter ecosystem structure and function.
29      • There is a major concern that soil acidification will result in nutrient deficiency.  Growth of tree
30       species can be affected when high aluminum to nutrient ratios limit uptake of calcium and
31       magnesium.  Calcium is  essential in the formation of wood and the maintenance of cells, the
        October 1999                               1-18         DRAFT-DO NOT QUOTE OR CITE

-------
 1       primary plant tissues necessary for tree growth. Calcium must be dissolved in the soil water to
 2       be taken up by plants. A major concern is that soil acidity will lead to nutrient deficiency.
 3       Acid deposition can increase the aluminum concentrations in soil water by lowering the pH in
 4       aluminum-rich soils through dissolution and ion-exchange processes. Aluminum in soil can be
 5       taken up by roots more readily than calcium because of its greater affinity for negatively
 6       charged surfaces. Tree species can be adversely affected if high Ca/Al ratios impair Ca and
 7       Mg uptake.
 8
 9      1.5.2 Particulate Matter-Related Effects on Materials
10           The effects of particulate matter and SO2 on materials are related to both aesthetic appeal
11      and physical damage. Studies have demonstrated particles, primarily consisting of carbonaceous
12      compounds, cause soiling of commonly used building materials and culturally important items
13      such as statutes and works of art. Physical damage from the dry deposition of SO2, particles and
14      the absorption or adsorption of corrosive agents on deposited particles can also result in the
15      acceleration of the weathering of manmade building and naturally occurring cultural materials.
16      • The natural process of metal corrosion from exposure to environmental elements (wind,
17       moisture,  sun, temperature fluctuations, etc.) is enhanced by exposure to anthropogenic
18       pollutants, in particular SO2.
19      • Dry deposition of SO2 enhances the effects of environmental elements on calcereous stones
20       (limestone, marble, and carbonated cemented) by converting the calcium carbonate (calcite) in
21       the stone to calcium sulphate dihydrate (gypsum). The rate of deterioration is determined by
22       the SO2 concentration, the stone's permeability and moisture content, and the deposition rate.
23      • Sulfur dioxide limits the life expectancy of paints by causing discoloration, loss of gloss, and
24       loss of thickness of the paint film layer.
25      • A significant detrimental effect of particulate pollution is the soiling of painted surfaces and
26       other building materials. Soiling is a degradation process requiring remediation by cleaning or
27       washing, and depending on the soiled surface, repainting.  Soiling decreases the reflectance of a
28       material and reduces the transmission of light through transparent materials. Soiling may
29       reduce the life usefulness of the material soiled.
30

        October 1999                              1-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1      1.5.3  Particulate Matter-Related Effects on Visibility
 2           Visibility is the degree to which the atmosphere is transparent to visible light and the clarity
 3      (transparency) and the color fidelity of the atmosphere.  The farthest distance at which a large
 4      black object can be distinguished against the horizontal sky is the visual range. For regulatory
 5      purposes, visibility impairment, any humanly perceptible change in visibility (light extinction,
 6      visual range, contrast, coloration), is classified into two principal forms: "reasonably
 7      attributable" impairment, attributable to a single source/small group of sources, and regional
 8      haze, any perceivable change in visibility caused by a combination of many sources over a wide
 9      geographical area.
10      • Visibility is measured by human observation, light scattering by particles, the light
11       extinction-coefficient (the sum of the light scattering coefficient and light absorption
12       coefficient) and parameters related to the light-extinction coefficient (visual range and deciview
13       scale), the light scattering coefficient, and fine particulate matter concentrations.
14      • Light scattering by gases is the major component of light extinction. Light absorption by gases
15       is almost entirely due to NO2, and is typically significant only near NO2 sources. Light
16       absorption by particles is primarily caused by elemental carbon.
17      • Visibility impairment or haziness is often associated with fine mass concentrations. Visibility
18       impairment or haziness is greatest in the eastern United States and southern California.
19       Haziness in the southeastern United States, caused by increased atmospheric sulfate, has
20       increased by approximately 80% since  the 1950s and is greatest in the summer months,
21       followed by the spring and fall, and winter.
22      • Visibility impairment in southern California is primarily caused by light extinction by nitrates.
23       Nitrates  contribute about 40% to the total light extinction in southern California. Nitrates
24       account for 10 to 20% of the total extinction in other areas of the United States.
25      • Some of the visibility impairment in northern California and Nevada, including Oregon,
26       southern Idaho and western Wyoming,  results from coarse mass and soil, primarily considered
27       natural extinction. In some areas of the United States, extinction from coarse mass is almost
28       negligible because the overall extinction is so high.
        October 1999                               1 -20        DRAFT-DO NOT QUOTE OR CITE

-------
 1      • High dust concentrations transported from southern California and the subtropics have
 2       contributed to regional haze in the Grand Canyon and other class I areas in the southwestern
 3       United States.
 4      • Organics are the second largest contributors to light extinction in most areas in the United
 5       States. Organic carbon is the greatest cause of light extinction in the Pacific Northwest,
 6       Oregon, Idaho, and Montana, accounting for 40 to 45% of the total extinction. Organic carbon
 7       contributes between 15 to 20% to the total extinction in most of the western United States and
 8       20 to 30% in the remaining areas of the United States.
 9      • Light absorption by carbon is relatively insignificant but is highest in the Pacific Northwest
10       (up to 15%) and in the eastern United States (3%).
11
12      1.5.4 Environmental and Economic Impacts of PM
13      • The four important categories of environmental costs and benefits of PM are:  agriculture and
14       forestry, cleaning and materials damage, visibility, and ecosystem functions. EPA has
15       developed and applied cost methodologies to these cost categories in great detail, and its
16       procedures have passed a number of scientific reviews.
17      • Any given level of particulate matter will be associated with resulting environmental effects
18       that potentially have economic significance. Examples include the level of crop damage or
19       visibility impairment that result from specified levels of PM.  Defining the welfare effects of
20       PM changes requires that baseline levels of effects be defined.
21      • Estimating benefits for visibility and for ecosystem services is a more difficult and less precise
22       exercise because the effects are not valued in markets.
23      • Once endpoints reflecting physical and biological outcomes have been defined, several
24       economic methods may be used to estimate economic damages. Some of the results of existing
25       research were summarized for the major categories of endpoints.
26      • The measured economic costs of PM are particularly significant for reduced visibility, both in
27       residential areas and in recreational areas with special value (e.g., the National Parks).
28      • It is possible that the  costs imposed on ecosystems are significant as well.  Making progress on
29       measuring these ecosystem costs depends on improvements in linking environmental  endpoints
        October 1999                              1 -21         DRAFT-DO NOT QUOTE OR CITE

-------
1       to PM levels, and then on using these endpoints as a basis for improved techniques to elicit
2       willingness to pay for changes in ecosystem quality.
      October 1999                             1 -22        DRAFT-DO NOT QUOTE OR CITE

-------
 i                                  2.  INTRODUCTION
 2
 3
 4           This document is an update of "Air Quality Criteria for Particulate Matter" published by the
 5      United States Environmental Protection Agency (EPA) in 1996, and it will serve as the basis for
 6      Congressionally-mandated periodic review of the National Ambient Air Quality Standards
 7      (NAAQS) for particulate matter (PM). The present document critically assesses the latest
 8      scientific information relative to determining the health and welfare effects associated with
 9      exposure to various concentrations of PM in ambient air.  The document is not intended as  a
10      complete and detailed literature review, but rather focuses on thorough evaluation of that
11      information most relevant to PM NAAQS criteria development, based on pertinent literature
12      available through mid-1999. This introductory chapter presents a brief summary of the history of
13      the PM NAAQS, provides an overview of issues addressed and procedures utilized in the
14      preparation of the present document, and provides orientation to the general organizational
15      structure of this document.
16
17
18      2.1 LEGISLATIVE REQUIREMENTS
19           Sections 108 and 109 of the U.S. Clean Air Act (CAA) (U.S. Code, 1991) govern the
20      establishment, review, and revision of National Ambient Air Quality Standards (NAAQS).
21      Section 108 directs the Administrator of the U.S. Environmental Protection Agency (EPA)  to list
22      pollutants that may reasonably be anticipated to endanger public health or welfare and to issue air
23      quality criteria for them.  The air quality criteria are to reflect the latest scientific information
24      useful in indicating the kind and extent of all exposure-related effects on public health and
25      welfare that may be expected from the presence of the pollutant in ambient air.
26           Section 109(a, b) directs the Administrator of EPA to propose and promulgate "primary"
27      and "secondary" NAAQS for pollutants identified under Section 108. Section 109(b)(l) defines
28      a primary standard as a level of air quality, the attainment and maintenance of which, in the
29      judgment of the Administrator, based on the criteria and allowing for an adequate margin of
30      safety, is requisite to protect the public health. Under Section 109(b) of the CAA, the

        October 1999                               2-1         DRAFT-DO NOT QUOTE  OR CITE

-------
 1     Administrator must consider available information to set secondary NAAQS that are based on
 2     the criteria and are requisite to protect the public welfare from any known or anticipated adverse
 3     effects associated with the presence of such pollutants. Welfare effects include effects on
 4     vegetation, crops, soils, water, animals, manufactured materials, weather, visibility, and climate,
 5     as well as damage to and deterioration of property, hazards to transportation, and effects on
 6     economic value and personal comfort and well-being. Section  109(d) of the CAA requires
 7     periodic review and, if appropriate, revision of existing criteria and standards. Also, an
 8     independent committee of non-EPA experts, the Clean Air Scientific Advisory Committee
 9     (CASAC), is to provide the EPA Administrator advice and recommendations regarding the
10     scientific soundness and appropriateness of criteria and NAAQS for PM and other "criteria air
11     pollutants."
12
13
14     2.2 HISTORY OF PREVIOUS PM CRITERIA/STANDARDS REVIEWS
15           On April 30, 1971 (Federal Register, 1971), EPA promulgated the original primary and
16     secondary NAAQS for particulate matter (PM) under Section 109 of the CAA. The reference
17     method for measuring attainment of these standards was the "high-volume" sampler (Code of
18     Federal Regulations, 1977), which collects PM up to a nominal size of 25 to 45 micrometers
19     (/^m), i.e., so-called "total suspended particulate" or "TSP". Thus, TSP was the original indicator
20     for U.S. PM NAAQS. The primary standards for PM (measured as TSP) were 260 //g/m3 (24-h
21     average) not to be exceeded more than once per year, and 75 //g/m3 (annual geometric mean).
22     The secondary standard (measured as TSP) was 150 //g/m3 (24-h average) not to be exceeded
23     more than once per year.
24           The next review of PM air quality criteria and standards was completed in July 1987 with
25     notice of a final decision to revise the existing standards published in the Federal Register
26     (Federal Register, 1987).  In that decision, EPA changed the indicator for PM from TSP to PM10.
27     PM10 refers to particles with an upper 50% cut point of 10 //m aerodynamic diameter. Identical
28     primary and secondary PM10 standards were set for two averaging times:  150 //g/m3 (24-h ave.)
29     with no more than one expected exceedance per year; and 50 //g/m3 (expected annual arithmetic
30     mean) averaged over 3 years.
31
       October 1999                             2-2        DRAFT-DO NOT QUOTE OR CITE

-------
 1     2.2.1  The 1997 PM NAAQS Revision
 2          The EPA initiated the last previous review of the air quality criteria and standards for PM
 3     in April 1994 by announcing its intention to develop revised Air Quality Criteria for Particulate
 4     Matter (henceforth, the "PM Air Quality Criteria Document" or PM AQCD). Thereafter, the
 5     EPA presented its plans for review of the criteria and standards for PM under a highly
 6     accelerated, court-ordered schedule at a public meeting of the CAS AC in December 1994.  A
 7     court order entered in American Lung Association v. Browner, CIV-93-643-TUC-ACM (U.S.
 8     District Court of Arizona, 1994), as subsequently modified, required publication of EPA's final
 9     decision on the review of the PM NAAQS by July 19, 1997.
10          Several workshops were held by EPA's National Center for Environmental Assessment's
11     RTP Division (NCEA-RTP) in November 1994 and January 1995 to discuss important new
12     health effects information useful in preparing initial PM AQCD draft materials. External review
13     drafts of the PM AQCD were then made available for public comment and were reviewed by
14     CAS AC at public meetings held in August 1995, December 1995, and February 1996. The
15     CASAC came to closure in its review of the PM AQCD, advising the EPA Administrator in a
16     March 15, 1996 closure letter (Wolff, 1996) that "although our understanding of the health
17     effects of PM is far from complete, a revised Criteria Document which incorporates the Panel's
18     latest comments will provide an adequate review of the available scientific data and relevant
19     studies of PM." Revisions in response to public and CASAC comments were incorporated as
20     appropriate in the final 1996 PM AQCD (U.S. Environmental Protection Agency, 1996a). A PM
21     Staff Paper (SP), prepared by U.S. EPA's Office of Air Quality Planning and Standards
22     (OAQPS) and drawing upon the 1996 PM AQCD and other exposure and risk assessments to
23     pose options for PM NAAQS decisions, also underwent similar CASAC review and public
24     comment, with consequent revision to its July  1996 final form (U.S. Environmental Protection
25     Agency, 1996b).
26          The SP  analyses served as key inputs to subsequently published proposal for revision of the
27     PM NAAQS. Taking into account information and assessments presented in the PM AQCD and
28     the Staff Paper, advice and recommendations of CASAC, and public comments received on the
29     proposal, the EPA Administrator revised the PM NAAQS by adding new PM2 5 standards and by
30     revising the form of the 24-h PM10 standard. Specifically, in July 1997, the Administrator made
31     the following revisions to the PM NAAQS:
       October 1999                              2-3        DRAFT-DO NOT QUOTE OR CITE

-------
 1      (1)  The suite of PM standards was revised to include an annual primary PM25 standard and a
 2          24-h PM2 5 standard.
 3      (2)  The annual PM2 5 standard is met when the 3-year average of the annual arithmetic mean
 4          PM2 5 concentrations, from single or multiple community-oriented monitors is less than or
 5          equal to 15 //g/m3, with fractional parts of 0.05 or greater rounding up.
 6      (3)  The 24-h PM25 standard is met when the 3-year average of the 98th percentile of 24-h PM25
 7          concentrations at each population-oriented monitor within an area is less than or equal to
 8          65 //g/m3, with fractional parts of 0.5 or greater rounding up.
 9      (4)  The form of the 24-h PM10  (150 //g/m3) standard was revised to be based on the 3-year
10          average of the 99th percentile of 24-h PM10 concentrations at each monitor within an area.
11      (5)  In addition, the Administrator retained the annual PM10 standard at the level of 50 //g/m3,
12          which is met when the 3-year average of the annual arithmetic mean PM10 concentrations at
13          each monitor within an area is less than or equal to 50 //g/m3, with fractional parts of 0.5 or
14          greater rounding up.
15           The principal focus of the last review of the air quality criteria and standards for PM was on
16      recent epidemiological evidence reporting associations between ambient concentrations of PM
17      and a range of serious health effects.  Particular attention was given to several size-specific
18      classes of particles, including PM10 and the principal fractions of PM10, referred to as the fine
19      (PM2 5) and coarse (PM10_2 5) fractions. PM2 5 refers to particles with an upper 50% cutpoint of
20      2.5 //m aerodynamic diameter.  PM10_2 5 refers to those particles with an upper 50% cutpoint of
21      10 //m and a lower 50% cut point of 2.5 //m aerodynamic diameter. In other words, the coarse
22      fraction (PM10_25) refers to the inhalable particles that remain if fine (PM25) particles are removed
23      from a sample of PM10 particles. As discussed in the  1996 PM AQCD, fine and coarse fraction
24      particles can be differentiated by their sources and formation processes and their chemical and
25      physical properties, including behavior in the atmosphere. Detailed discussions  of atmospheric
26      formation, ambient concentrations, and health and welfare effects of PM, as well as quantitative
27      estimates of human health risks associated with exposure to PM, can be found in the 1996 PM
28      AQCD and in the 1996 OAQPS Staff Paper (U.S. Environmental Protection Agency,  1996b).
29
30
31

        October 1999                              2-4         DRAFT-DO NOT QUOTE OR CITE

-------
 1      2.2.2 Presidential Memorandum: Next Particulate Matter Review
 2            and Research
 3           On July 18,1997, the EPA published a final rule revising the NAAQS for PM (Federal
 4      Register, 1997a), and, on the same day, a final rule revising the Ozone NAAQS (Federal
 5      Register, 1997b). A Presidential Memorandum (Federal Register, 1997c) was also published
 6      outlining the Administration's goals for implementing the revised PM and Ozone NAAQS.  The
 7      Memorandum directed EPA to provide to CASAC within 90 days and to publish a notice
 8      outlining its schedule for the next periodic review of PM and to complete the next review,
 9      including review by CASAC, within 5 years after issuance of the revised standards (i.e., by July
10      2002). Such a schedule would ensure that EPA's review of emerging scientific information,
11      which forms the criteria upon which the standards are based, and of the standards themselves will
12      have been completed prior to any areas being  designated as "nonattainment" under the newly
13      established standards for fine particles (i.e., PM25 standards) and prior to the imposition of any
14      new controls related to the revised standards.  The Presidential Memorandum also directed EPA
15      and other relevant Federal  agencies to develop and implement a greatly expanded, coordinated
16      research plan. These PM research plans are outlined in the following section. To facilitate
17      timely scientific research within this review period, EPA initiated certain activities immediately,
18      as noted below in the discussion of the PM Research Program.
19
20      PM Research Program
21           The EPA has broadened its ongoing PM research activities by developing, in partnership
22      with other Federal agencies, a coordinated interagency PM research program.  This interagency
23      program will contribute to  expanding scientific knowledge of PM health effects, as well as the
24      development of improved monitoring methods and cost-effective mitigation strategies.  The
25      interagency effort is also promoting further coordination with other research organizations
26      including state-, university-, and industry-sponsored research groups. Beginning in the fall of
27      1997, public participation has been and continues to be encouraged through workshops and
28      review of program documentation.  Workshops and the availability of relevant documentation are
29      being announced in  the Federal Register.
30           To aid identification of needed research efforts, EPA published a particulate matter health
31      risk research needs document (U.S. Environmental Protection Agency, 1998a). That document

        October 1999                             2-5         DRAFT-DO NOT QUOTE OR CITE

-------
 1      identifies research needed to improve scientific information supporting future health risk
 2      assessment and review of the PM NAAQS. The document aimed to provide a foundation for PM
 3      research coordination among Federal agencies and other research organizations and served as one
 4      useful input to National Research Council PM research deliberations. In January 1998, the
 5      National Research Council (NRC) established its Committee on Research Priorities for Airborne
 6      Particulate Matter in response to a request from Congress in the Fiscal 1998 appropriation to
 7      EPA.  This Committee is charged to identify the most important research priorities relevant to
 8      setting particulate matter standards, to develop a conceptual plan for particulate matter research,
 9      and to monitor research progress toward improved understanding of the relationship between
10      particulate matter and public health. The Committee issued its first report in early  1998
11      (National Research Council, 1998) and a second one recently in 1999 (National Research
12      Council, 1999).
13           The EPA's research program includes studies to improve understanding of the formation
14      and composition of fine PM, the characteristics or components of PM that are responsible for its
15      health effects, the mechanisms by which these effects are produced,  and improved measurements
16      and estimation of population exposures to PM. Specific EPA research efforts include controlled
17      human exposure studies, in vivo and in vitro toxicology, epidemiology, atmospheric sciences
18      including monitoring and modeling studies, development of data on  emissions of fine particles
19      from stationary and mobile sources, and identification and evaluation of risk management
20      options.  The results from these efforts, as well as related efforts by other Federal agencies and
21      the general scientific community, will advance the scientific and technical bases for future
22      decisions on the PM NAAQS  and  for the implementation of PM monitoring and control efforts.
23
24
25      2.3 CURRENT PM CRITERIA AND NAAQS REVIEW
26      2.3.1 Criteria Review Plans and Schedule
27           The EPA's plans for this current review of PM criteria are outlined in Table 2-1 below,
28      together with target dates for key milestones. As with all NAAQS reviews, the purpose is to
29      update the criteria and to determine whether it is appropriate to revise the standards in light of
30      new scientific and technical information. Although the EPA concluded in its recent final rule on

        October 1999                             2-6         DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 2-1. SCHEDULE FOR DEVELOPMENT OF THE CURRENT REVISED
          PARTICULATE MATTER AIR QUALITY CRITERIA DOCUMENT (PM AQCD)
       Major Milestones
                     Target Dates
       PM NAAQS Review Plan to CASAC
       Prepare AQCD Development Plan
       Begin Literature Search
       Federal Register Call for Information/Sources Sought
       CASAC Meeting on AQCD Development Plan
       Prepare Workshop Drafts of Chapters
       Peer Review Workshop
       Prepare External Review Draft AQCD
       First External Review Draft to CASAC
       Public Comment Period on Draft AQCD
       CASAC Meeting on Draft AQCD
       Prepare Revised Draft AQCD
       Second External Review Draft to CASAC
       Public Comment Period on Second Draft
       CASAC Meeting on Second Draft
       Complete Final PM AQCD	
           October 1997
           November 1997 to January 1998
           February 1998
           April 1998
           May 1998
           May to December 1998
           April 1999
           March to September 1999
           October 1999
           October 1999 to January 2000
           December 1999
           January to June 2000
           June 2000
           July - August 2000
           September 2000
           December 2000
       Source: Modified from Federal Register (1997c).
1     the PM standards (Federal Register, 1997a) that the current scientific knowledge provides a
2     strong basis for the revised PM standards, including the establishment of PM2 5 standards, there
3     remain scientific uncertainties associated with the health effects of PM and with the means of
4     reducing such effects. Recognizing the importance of developing a better understanding of the
5     effects of fine particles on human health, including their causes and mechanisms, as well as the
6     species and sources of PM25, the EPA will continue to sponsor research to address these
7     uncertainties even as this criteria review progresses as per Table 2-1.
      October 1999
2-7
DRAFT-DO NOT QUOTE OR CITE

-------
 1          As with other NAAQS reviews, a rigorous assessment of relevant scientific information is
 2     to be presented in this updated, revised PM AQCD being prepared by EPA's NCEA-RTP
 3     Division.  The development of the document has and will continue to involve substantial external
 4     peer review through public workshops involving the general aerosol scientific community,
 5     through iterative reviews of successive drafts by CAS AC, and through comments from the
 6     public.  The final document will reflect input received through these reviews and will serve to
 7     evaluate and integrate the latest available scientific information to ensure that the review of the
 8     PM standards is based on sound science. The schedule for this review will allow for
 9     consideration of relevant new peer-reviewed scientific studies published or accepted for
10     publication from early 1996 (when the previous PM AQCD was completed) through mid-2000.
11          After the December 1999 CAS AC meeting noted above in Table 2-1, EPA's Office of Air
12     Quality Planning and Standards (OAQPS) will also start to prepare a Staff Paper (SP) for the
13     Administrator, drawing on information in the newly revised PM AQCD. The SP will evaluate
14     the policy implications of the key studies and scientific information contained in the AQCD and
15     identify critical elements that EPA staff believes should be considered in reviewing the
16     standards. The SP is intended to bridge the gap between the scientific review in the AQCD and
17     the public health and welfare policy judgments required of the Administrator in reviewing the
18     PM NAAQS. For that purpose, the SP will present technical analyses, including air quality
19     analyses and a quantitative health risk assessment, and other factors relevant to the evaluation of
20     the PM NAAQS, as well as staff conclusions and recommendations of options for the EPA
21     Administrator's consideration. The SP will also be reviewed by CAS AC and the public, and the
22     final SP will reflect the input received through these reviews.
23          Following completion of the final SP, the Administrator will then announce in the Federal
24     Register proposals for retaining or revising the current PM NAAQS, and opportunities will be
25     provided for public comment and CASAC review of those proposals.  Taking into account public
26     comments and CASAC recommendations, final decisions regarding the current PM NAAQS
27     review are to be promulgated by July 2002.
28
29     2.3.2  Methods and Procedures for Document Preparation
30          The procedures followed for developing this revised PM AQCD build on the knowledge
31     and methods derived from the  last PM, ozone, and CO AQCD preparation efforts. Briefly, the
       October 1999                             2-8        DRAFT-DO NOT QUOTE OR CITE

-------
 1      respective responsibilities for production of the present PM AQCD are as follows.  An NCEA-
 2      RTF team has been appointed to be responsible for developing and implementing the project plan
 3      for preparation of the PM AQCD and inputs from individuals in other EPA program and policy
 4      offices identified as part of the EPA PM Work Group have been obtained. The resulting project
 5      plan was then discussed with CAS AC (May 1998) and appropriately revised. An ongoing
 6      literature search has continued to be conducted to identify, to the extent possible, all PM
 7      literature published since early 1996. Additionally, EPA published (1) a request for information
 8      in the Federal Register asking for recently available research information on PM that may not yet
 9      be published and (2) a request for individuals with the appropriate type and level of expertise to
10      contribute to the writing of PM AQCD materials to identify themselves (U.S. Environmental
11      Protection Agency, 1998b).  Specific authors of chapters or sections of the proposed document
12      were selected on the basis of their expertise on the subject areas and their familiarity with the
13      relevant literature, and these include both EPA and non-EPA scientific experts. The project team
14      defined critical issues and topics to be addressed by the authors and provided direction in order to
15      emphasize evaluation of those studies most clearly identified as important for standard setting.
16          The main focus of this revised criteria document is the evaluation and interpretation of air
17      quality data and health and welfare effects information newly published since that assessed in the
18      1996 PM AQCD and likely to be useful in deriving criteria for PM NAAQS.  Draft AQCD
19      chapters were evaluated via expert workshops and/or expert written peer reviews, which focused
20      on the selection of pertinent studies included in the chapters, the potential need for additional
21      information to be added to the chapters, and the quality of the summarization and interpretation
22      of the literature. The authors of the draft chapters then revised them on the basis of the workshop
23      and/or written expert review recommendations. These and other integrative summary materials
24      have been incorporated into this First External Review Draft of the PM AQCD now being
25      released for public comment and CASAC review.  Necessary revisions will be made on the basis
26      of the public comments, CASAC recommendations, and newly emerging research results before
27      a further Second External Review will be released in mid-2000 for public comment and CASAC
28      review (Fall 2000). The final version of the newly revised PM AQCD will incorporate changes
29      made in response to public comments and CASAC review of that Second External Review Draft.
30
        October 1999                              2-9         DRAFT-DO NOT QUOTE OR CITE

-------
 1           New research results are being incorporated into this document as they become available.
 2      In order to foster presentation and publication of any new research findings not published during
 3      1999, NCEA-RTP is working with the Air and Waste Management Association (AWMA) to
 4      hold an International Speciality Conference, entitled PM 2000: Particulate Matter and Health, in
 5      late January 2000, in Charleston, SC. The conference is being co-sponsored in cooperation with
 6      one or more government agencies or private organizations that also fund PM research. Topics to
 7      be covered will include new research results concerning the latest advances in PM atmospheric
 8      sciences (e.g., PM formation, transport, transformation); PM exposure; PM dosimetry and
 9      extrapolation modeling; PM toxicology (e.g., mechanisms, laboratory animal models, human
10      clinical responses); and PM epidemiology. The main purpose of the conference will be to
11      facilitate having the latest scientific information available for incorporation into the final draft of
12      the revised U.S. EPA PM Air Quality Criteria Document (AQCD) in time for the anticipated
13      final CAS AC review of the draft PM AQCD projected for Fall 2000.  Arrangements will be
14      made for PM 2000 presenters to submit written manuscripts at the conference and to have
15      professional societies/journals prepared to expedite processing of the submitted papers through
16      their peer-review processes, so that decisions on acceptance for publication can be made by
17      April/May 2000. Given that extensive additional information is expected to be published during
18      the next 6 to 9 mo as a result of the PM 2000 conference and elsewhere, the evaluations and
19      findings set forth in this draft of the revised PM AQCD, overall, must be considered only
20      provisional at this time.
21
22      2.3.3 Approach
23           The approach to organization and content of this revised PM AQCD is somewhat different
24      from those used for previous criteria documents. Since the recent document (U.S. Environmental
25      Protection Agency, 1996a) provides an extensive discussion of most topic areas, this new
26      document focuses more specifically on critical issues that have been identified as areas needed to
27      improve the scientific basis (criteria) for PM NAAQS, particularly for those areas in which the
28      information data base has continued to evolve rapidly.
29           Detailed review of key new research was undertaken as an initial step.  However, instead of
30      presenting a comprehensive review of all the literature, emphasis in this revised AQCD is placed
31      on (1) the concise summary of key findings derived from previous PM criteria reviews and
        October 1999                              2-10       DRAFT-DO NOT QUOTE OR CITE

-------
 1      (2) evaluation of the most pertinent new key information, with greater emphasis on more
 2      interpretive assessment. This approach reflects recommendations made by CAS AC.
 3           An initial step was to review the available scientific literature and to focus on the selection
 4      of pertinent issues to include in the document as the basis for the development of PM NAAQS
 5      criteria. Preliminary issues were identified by the NCEA Project Team and through input from
 6      other EPA program and policy offices. Development of issue topics started with the last AQCD,
 7      staff paper, CAS AC and public reviews, EPA's PM Research Needs Document, and from the
 8      standard promulgation process.  Further identification and clarification of issues resulted from
 9      the National Research Council (NRC) review and report on PM research priorities.  The CASAC
10      review of the PM AQCD Development Plan and public comments on draft AQCD materials at
11      various stages of their development has also played an important role in issue identification.
12           To aid in development of a concise document, compilation of summary tables of the
13      relevant published literature and selective discussion of the literature was undertaken.  Building
14      on the previous PM AQCD, most of the scientific information selected for review and discussion
15      in the text was from the literature published since completion of the previous PM AQCD (U.S.
16      Environmental Protection Agency, 1996a).
17
18      2.3.4 Key Issues of Concern
19           Summarized below are several broad topics related to the main issues of concern to be
20      addressed by this revised PM AQCD. The document reviews and assesses available data bearing
21      on each of the issues identified below.
22      1. Causality.  Evaluation of the evidence for or against a causal relationship between health
23        outcomes and ambient PM and/or specific physical-chemical components.
24        •  Specific components of interest include size classes such as PM10, PM10_25, PM25, and
25          ultrafine  particles.  Chemical components include transition metals, acidity, sulphates,
26          nitrates, and organics.
27        • Expand review of foundations of causal inference for associated PM air pollution health
28          effects.
29        • Access new long-term PM  exposure and health data to broaden interpretation of long-term
30          exposure findings.

        October 1999                              2-11        DRAFT-DO NOT QUOTE OR CITE

-------
 1         •  Review data exploring potential mechanisms of response to PM physical-chemical
 2           characteristics, response pathway, and exposure-dose-response relationships (laboratory and
 3           clinical research).
 4      2.  Uncertainties. In carrying out overall assessment, address the following types of uncertainty:
 5         •  Uncertainties between stationary PM monitoring instruments and personal exposure to PM
 6           of ambient origin, especially for susceptible groups and their related activity patterns.
 7           Specific topics include: measurement error in outdoor monitors themselves; use of central
 8           monitors for estimates of community concentrations; and the use of community
 9           concentrations as a surrogate for personal exposure to particles of ambient origin.
10         •  Uncertainties related to particulate matter size fraction, particle number, surface area, and
11           content of semi-volatile components.
12         •  Uncertainties about the effects of long term PM exposure, such as life shortening,  and
13           development and progression of disease.
14         •  Uncertainties due to coexposure to other pollutants such as O3, SO2, and CO and
15           meterological factors.
16         •  Uncertainties due to confounding in epidemiologic studies (e.g., economic factors,
17           demographic and lifestyle attributes, genetic susceptibility factors, occupational exposure,
18           and medical care).
19         •  Uncertainty about shape of concentration-response (CR) relationships and associated
20           community risks (linear and threshold models for CR).
21         •  Uncertainty about methods for synthesis of health  outcome studies and evaluation of
22           sensitivity and confounding aspects including but not limited to meta-analysis.
23      3.  Biological Mechanisms of Action. Evaluate data examining mechanisms for health outcomes
24         ofPM. Mechanistic information aids judgement about causality
25         •  New studies have  examined mechanisms of action of PM  constituents including transition
26           metals, airborne allergens, and the generation of reactive oxygen species. Different cell
27           types have differing responses to PM components.
28         •  Newly published studies have also identified potential mechanisms for the production of
29           cardiac arrhythmias by PM constituents, especially in animal models of disease and suggest
30           that particular attention should be accorded to PM metal constituents.


        October 1999                              2-12        DRAFT-DO NOT QUOTE OR CITE

-------
 1         • Although many new animal toxicology studies involve instillation of previously collected
 2          particles and this technique is appropriate to study mechanisms of action, extrapolation to
 3          human equivalent exposure/doses is uncertain.
 4         • Ongoing work on the effects of lung inflammation and PM phagocytosis on subsequent
 5          systemic effects,  especially cardiac or vascular effects, is needed to provide further
 6          information on the relationship between inhaled pollutants and cardiac events.
 7         • Interpretation of concentrated ambient particles (CAPs) studies.  Newly available
 8          information will be examined from toxicology studies using devices which concentrate
 9          (to variable extents) ambient PM to determine PM concentration-response relationships.
10          Again, difficulties are encountered regarding extrapolation to comparable human exposures
11          to ambient PM levels.
12      4.  Susceptible Populations.  Examine health outcome data to determine specific risk groups that
13         are more susceptible than normal healthy adults to adverse effects from PM exposure.
14         • Preexisting respiratory or cardiovascular disease in conjunction with advanced age, appear
15          to be important factors in PM mortality susceptibility.
16         • For morbidity health endpoints, children and asthmatics may potentially display increased
17          sensitivity to PM exposure. Data will be examined for coherence.
18         • Patterns of respiratory tract deposition, clearance, and retention in susceptible populations
19          have been recently studied and provide evidence of increased deposition associated with
20          lung disease.
21         • Animal models of lung disease exposed to PM constituents suggest a role for PM in cardiac
22          death.
23      5.  Environmental Effects. Evaluation of several types of PM welfare effects:
24         • Vegetation and ecosystem effects.
25         • Visibility effects.
26         • Materials damage.
27         • Role of PM in atmospheric radiative transfer and potential consequences for penetration of
28          biologically harmful UVB to the earth surface and for climate  change.
29      6.  Background Information Topics Useful in Evaluating Health Risks. Topics include:
30         • New monitoring methods, especially methods used in epidemiology studies
31         • Indicator topics such as PM2 5 versus  PMj 0. ultrafine;  and PM2 5 versus PM10_2 5

        October 1999                              2-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1         • New data patterns of daily and annual concentrations for PM25, PM10_25, and PM10
 2
 3
 4      2.4 DOCUMENT CONTENT AND ORGANIZATION
 5           The present document critically reviews and assesses relevant scientific literature on PM
 6      through mid-1999. The material selected for review and comment in the text generally comes
 7      from the more recent literature published since early 1996, with emphasis on studies conducted at
 8      or near PM pollutant concentrations found in ambient air. Literature discussed in detail in the
 9      previous 1996 EPA PM AQCD (U.S. Environmental Protection Agency, 1996a) is generally not
10      discussed in depth in this document.  However, some limited treatment is included of the earlier
11      studies judged to be potentially useful in deriving PMNAAQS.  Key literature is discussed in
12      the text and presented in tables as well. Reports of lesser importance for the purposes of this
13      document are typically only summarized in tables.
14           The primary emphasis is on consideration of published material that has undergone
15      scientific peer review. However, in the interest of admitting new and important information
16      expected to become available shortly, some material not yet fully published in the open literature
17      but meeting other standards of scientific reporting (i.e., peer review and quality assurance) are
18      now provisionally included.  As noted earlier, emphasis has  been placed on studies in the range
19      of current ambient levels. However, studies examining effects of higher concentrations have
20      been included if they contain unique data, documentation of a previously unreported effect or
21      mechanism. In reviewing and summarizing the literature, an attempt is made to present
22      alternative points of view where scientific controversy exists.
23           The present document consists of 9 chapters. The Executive Summary for the entire
24      document is contained in Chapter 1, followed by this general introduction in Chapter 2.
25      Chapters 3 through 5 provide background information on physical and chemical properties of PM
26      and related compounds; sources and emissions; atmospheric transport, transformation, and fate
27      of PM; methods for the collection and measurement of PM;  and ambient air concentrations and
28      factors affecting exposure of the general population. Chapters 6  through 8 evaluate information
29      concerning the health effects of PM. Chapter 6 discusses epidemiological studies, and Chapter 7
30      discusses dosimetry of inhaled particles in the respiratory tract information on the toxicology of
31      specific types of PM constituents, including laboratory animal studies and controlled human
        October 1999                              2-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1      exposure studies.  Chapter 8 integrates key information on exposure, dosimetry, and critical
 2      health risk issues derived from studies reviewed in the prior chapters. Lastly, Chapter 9
 3      describes PM environmental effects on:  vegetation and ecosystems; visibility; manmade
 4      materials; and climate, as well as economic impacts of such welfare effects.
 5           Neither control techniques nor control strategies for the abatement of PM are discussed in
 6      this document, although some topics covered may be incidentally relevant to abatement
 7      strategies. Technologies for controlling PM emissions are discussed in other documents issued
 8      by EPA's Office of Air Quality Planning and Standards (OAQPS).  Likewise, issues germane to
 9      the scientific basis for control strategies, but not pertinent to the development of NAAQS criteria,
10      are addressed in numerous other documents issued by OAQPS.
11           In addition, certain issues of direct relevance to standard setting are not explicitly addressed
12      in this document, but are instead analyzed in documentation prepared by OAQPS as part of its
13      regulatory analyses materials.  Such analyses include (1) delineation of particular adverse effects
14      that the primary and secondary NAAQS are intended to protect against, (2) exposure analyses
15      and assessment of consequent risk, and (3) discussion of factors to be considered in determining
16      an adequate margin of safety.  Key points and conclusions from such analyses will be
17      summarized in the PM Staff Paper to be prepared by OAQPS and reviewed by CASAC.
18      Although scientific data contribute significantly to decisions regarding the above issues, their
19      resolution cannot be achieved solely on the basis of experimentally acquired information.  Final
20      decisions on items (1) and (3) are made by the Administrator, as mandated by the Clean Air Act.
21           A fourth issue directly pertinent to standard setting is identification of populations at risk,
22      which is basically a selection by EPA  of the subpopulation(s) to be protected by the promulgation
23      of a given standard.  This issue is addressed only partially in this document. For example,
24      information is presented on factors, such as preexisting disease, that may biologically predispose
25      individuals and subpopulations to adverse effects from exposures to PM. The characterization of
26      population risk, however, requires information above and beyond data on biological
27      predisposition, such as information on estimated exposure, activity patterns, and personal habits.
28      Such information is typically addressed in the Staff Paper developed by OAQPS.
29
        October 1999                              2-15        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES

  2       Code of Federal Regulations. (1977) Appendix B—Reference method for the determination of suspended
  3             particulate matter in the atmosphere (high-volume method). C. F. R. 40: §50.
  4       Federal Register. (1971) National primary and secondary ambient air quality standards. F. R. (April 30)
  5             36:8186-8201.
  6       Federal Register. (1987) Revisions to the national ambient air quality standards for particulate matter. F. R. (July 1)
  7             52:24,634-24,669.
  8       Federal Register. (1997a) National ambient air quality standards for particulate matter; final rule. F. R. (July 18)
  9             62:38,652-38,752.
10       Federal Register. (1997b) National ambient air quality standards for ozone; final rule. F. R. (July 18)
11             62:38,856-38,896.
12       Federal Register. (1997c) Implementation of revised air quality standards for ozone and particulate matter. F. R.
13             (July 18) 62: 38,421-38,422.
14       National Research Council. (1998) Research priorities for airborne particulate matter. I. Immediate priorities and a
15             long-range research portfolio. Washington, DC: National Academy Press.
16       National Research Council. (1999) Research priorities for airborne particulate matter. II. Evaluating research
17             progress and updating the portfolio. Washington, DC: National Academy Press.
18       U.S. Code. (1991) Clean Air Act, §108, air quality criteria and control techniques, §109, national ambient air
19             quality standards. U. S. C. 42:  §§7408-7409.
20       U.S. District Court of Arizona. (1994) American Lung Association v. Browner. 884 F.Supp.  345
21             (CIV-93-643-TUC-ACM).
22       U.S. Environmental Protection Agency. (1996a) Air quality criteria for particulate matter. Research Triangle Park,
23             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
24             Available from: NTIS, Springfield, VA; PB96-168224.
25       U.S. Environmental Protection Agency. (1996b) Review of the national ambient air quality standards for particulate
26             matter: policy assessment of scientific and technical information. OAQPS staff paper. Research Triangle
27             Park, NC: Office of Air Quality Planning and Standards; report no. EPA/452/R-96-013. Available from:
28             NTIS, Springfield, VA; PB97-115406REB.
29       U.S. Environmental Protection Agency. (1998a) Particulate matter research needs for human health risk assessment
30             to support future reviews of the national ambient air quality standards for particulate matter. Research
31             Triangle Park, NC: National Center for Environmental Assessment; report no. EPA/600/R-97/132F.
32       U.S. Environmental Protection Agency. (1998b) Review of national ambient air quality standards for particulate
33             matter. Commer. Bus. Daily: February 19. Available online: cbdnet.access.gpo.gov.
34       Wolff, G. T. (1996) Closure by the Clean Air Scientific Advisory Committee (CASAC) on the draft air quality
35             criteria for particulate matter [letter to Carol M. Browner, Administrator, U.S. EPA]. Washington, DC:
36             U.S. Environmental Protection Agency, Clean Air Scientific Advisory Committee.;
37             EPA-SAB-CASAC-LTR-96-005; March 15.
38
         October 1999                                   2-16          DRAFT-DO NOT QUOTE OR CITE

-------
 i          3.  PHYSICS, CHEMISTRY, AND MEASUREMENT
 2                         OF  PARTICULATE MATTER
 3
 4
 5          An extensive review of the physics and chemistry of particulate matter (PM) was included
 6     in Chapter 3 of the 1996 Air Quality Criteria for Particulate Matter (AQC PM 96) (U.S.
 7     Environmental Protection Agency, 1996). Section 3.1 of this new version of the Air Quality
 8     Criteria for Particulate Matter provides background information on the physics and chemistry of
 9     atmospheric particles that may be useful in reading subsequent sections and chapters.
10     New information needed to understand risk assessment will be discussed. Emphasis will be
11     placed on the differences between fine and coarse particles and the differences between the
12     nuclei mode and the accumulation mode within fine particles.
13          Chapter 4 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996) contained a
14     review of the state-of-the-art of PM measurement technology. Since that time there has been
15     considerable progress in understanding problems and errors in the measurement of PM mass,
16     chemical composition, and physical parameters. There has also been some progress in
17     developing new and improved measurement techniques.  Therefore, a more extensive survey on
18     measurement errors and newly developed measurement techniques is included in Section 3.2.
19     For more detail and older references the reader is referred to Chapter 3 and 4 of the AQC PM 96
20     (U.S. Environmental Protection Agency, 1996).
21
22
23     3.1  PHYSICS AND CHEMISTRY OF PARTICULATE  MATTER
24     3.1.1  Definitions
25          Atmospheric particles originate from a variety of sources and possess a range of
26     morphological, chemical, physical, and thermodynamic properties. Examples include
27     combustion-generated particles such as diesel soot or fly ash, photochemically produced particles
28     such as those found in urban haze, salt particles formed from sea spray, and soil-like particles
29     from resuspended dust.  Some particles are liquid, some are solid; others contain a solid core
30     surrounded by liquid. Atmospheric particles contain inorganic ions, metallic compounds,

       October 1999                             3-1         DRAFT-DO NOT QUOTE OR CITE

-------
 1      elemental carbon, organic compounds, and crustal compounds.  Some atmospheric particles are
 2      hygroscopic and contain particle-bound water.  The organic fraction is especially complex,
 3      containing hundreds of organic compounds.
 4           The composition and behavior of airborne particles are fundamentally linked with those of
 5      the surrounding gas. Aerosol is defined as a suspension of solid or liquid particles in air and
 6      includes both the particles and all vapor or gas phase components of air.  However, the term
 7      aerosol is often used to refer to the suspended particles only. Particulate is an adjective and
 8      should only be used as a modifier, as in particulate matter.
 9           A complete description of the atmospheric aerosol would include an accounting of the
10      chemical composition, morphology, and size of each particle and the relative abundance of each
11      particle type as a function of particle size (Friedlander, 1970). However, most often the physical
12      and chemical characteristics of particles are measured separately. Number size distributions are
13      often determined by physical means, such as electrical mobility or light-scattering of suspended
14      particles. Chemical composition usually is determined by analysis  of collected samples although
15      sulfate can be measured in-situ. The mass and average chemical composition of particles,
16      segregated by aerodynamic diameter, by cyclones or impactors,  can also be determined,
17      However, recent developments in single particle analysis techniques, by electron microscopy
18      with x-ray analysis  of particles collected on a substrate or by mass spectroscopy of suspended
19      particles passing through a sensing volume, provide elemental composition of individual
20      particles by particle size and thus are bringing the description envisioned by Friedlander closer to
21      reality.
22
23      3.1.2 Physical Properties and Processes
24      3.1.2.1  Definitions of Particle Diameter
25           The diameter  of a spherical particle may be determined geometrically, from optical or
26      electron microscopy; by light scattering and Mie theory; or by its behavior, such as its electrical
27      mobility or its aerodynamic behavior.  However, atmospheric particles often are not spherical.
28      Therefore, their diameters are often described by an "equivalent" diameter, i.e., that of a unit
29      density sphere which would have the same physical behavior. The aerodynamic diameter is
30      important for particle transport, collection, and respiratory tract deposition. The aerodynamic
31      diameter, Da, depends on particle density and is defined as the diameter of a spherical particle
        October 1999                              3-2         DRAFT-DO NOT QUOTE OR CITE

-------
 1      with equal settling velocity but a material density of 1 g/cm3. Particles with the same physical
 2      size and shape but different densities will have different aerodynamic diameters. Detailed
 3      definitions of the various sizes and their relationships are given in standard aerosol textbooks,
 4      e.g., Friedlander (1977), Reist (1984), and Seinfeld and Pandis (1998).
 5
 6      3.1.2.2  Aerosol Size Distributions
 7           Particle size, as indexed by one of the "equivalent" diameters, is an important parameter in
 8      determining the properties, effects and fate of atmospheric particles. The atmospheric deposition
 9      rates of particles, and therefore, their residence times in the atmosphere, are a strong function of
10      aerodynamic diameter.  The aerodynamic diameter also influences deposition patterns of
11      particles within the lung. Light scattering is strongly dependent on the optical particle size.
12      Particle size distributions, therefore, have a strong influence on atmospheric visibility and,
13      through their effect on radiative balance, on  climate.  Size  distribution studies using impactors
14      give direct measurements of the aerodynamic diameter. The diameter of atmospheric particles
15      range from 1 nanometer to 100 // meters, 5 orders of magnitude.  A variety of different
16      instruments, measuring a variety of equivalent diameters, are required to cover this range.
17           Older particle counting studies used optical particle counters to cover the range of 0.3 to
18      30 //m diameter. Diameters of particles below 0.5 //m were measured as mobility diameters.
19      The particle diameters used in size distribution graphs from these studies are usually given as
20      geometric diameters rather than aerodynamic diameters. In recent years, aerodynamic particle
21      sizers, which give a direct measurement of the aerodynamic diameter in the range of
22      approximately 0.7 to  10 //m diameter, have been used with electrical mobility analyzers, which
23      measure the mobility diameter from approximately 0.5 //m to very small particles of the order of
24      .005 //m, to cover the range of regulatory interest. Unfortunately, there is no agreed-upon
25      technique for combining the various equivalent diameters.  Some workers use various
26      assumptions to combine the various measurements into one presentation; others report each
27      instrument separately. Therefore, the user of size distribution data must be careful to determine
28      exactly which equivalent diameter is reported.
29           Aerodynamic diameter is the most widely used equivalent diameter. Therefore, in future
30      discussions, particle diameters, unless otherwise indicated, will be understood to refer to the
31      aerodynamic diameter.

        October 1999                               3-3         DRAFT-DO NOT QUOTE OR CITE

-------
 1      Particle Size Distribution Functions
 2           The distribution of particles with respect to size is an important physical parameter
 3      governing their behavior.  Because atmospheric particles cover several orders of magnitude in
 4      particle size, size distributions are often expressed in terms of the logarithm of the particle
 5      diameter, on the X-axis, and the measured differential concentration on the Y-axis:
 6      AN/A(logDp) = the number of particles per cm3 of air having diameters in the size range from log
 7      Dp to log(Dp + ADp). It is not proper formally to take the logarithm of a dimensional quantity.
 8      However, one can think of the distribution as a function of log(Dp/Dp0), where the reference
 9      diameter Dp0 = 1 //m is not explicitly stated. If AN/A(logDp) is plotted on a linear scale, the
10      number of particles between Dp and Dp + ADp is proportional to the area under the curve of
11      AN/A(logDp) versus logDp. Similar considerations apply to distributions of surface, volume, and
12      mass. It has been found that atmospheric aerosol size distributions frequently may be
13      approximated by a sum of log-normal distributions corresponding to the various modes or
14      fractions. When approximated by a function, the distributions are usually given as dN/d(log Dp)
15      rather than AN/A(log Dp).
16
17      Atmospheric Aerosol Size Distributions
18           Averaged atmospheric size distributions are shown in Figures 3-1,2,  and 3 (Whitby, 1978;
19      Whitby and Sverdrup, 1980). Figure 3-1 describes the number of particles  as a function of
20      particle diameter for rural, urban-influenced rural, urban, and freeway-influenced urban aerosols.
21      For some of the  same data, the particle volume distribution is shown in Figure 3-2. Figure 3-3
22      show the number, surface, and volume distribution for the grand average continental size
23      distribution. Note that the particle diameter is always shown on a logarithmic scale. The particle
24      number is frequently shown on a logarithmic scale in order to display the wide range in number
25      concentration for different particle sizes and different sites.  Volume and surface area, and
26      sometimes number, are shown on an arithmetic scale with the distributions plotted such that the
27      volume, surface area, or number of particles in any specified size range is proportional to the
28      corresponding area under the curve. These distributions show that most of the particles are quite
29      small, below 0.1 //m, while most of the particle volume (and therefore most of the mass) is found
30      in particles > 0.1 //m. The surface area peaks around 0.1 //m.
31

        October 1999                              3-4        DRAFT-DO NOT QUOTE OR CITE

-------
           1,000,000 -

              10,000 -
          o:       1 -
         O,
         ^3)
         |     0.01 H
         z
         T3
              0.0001 -

            0.000001 -
• Clean Rural
 Urban Influenced Rural
                    	Average Urban
                    	Urban + Freeway
                                                      200,000
                  0.01    0.1    1     10    100
                    Particle Diameter, Dp (u.m)
                                                   O
                          C)
                          T3
                                                      150,000
                             100,000 -
                              50,000 -
                                   0.01       0.1         1         10
                                     Particle Diameter, Dp (pm)
      Figure 3-1.  Number of particles as a function of particle diameter: (a) number
                   concentrations are shown on a logarithmic scale to display the wide range by
                   site and size; (b) number concentrations for the average urban distribution
                   are shown on a linear scale for which the area under any part of the curve is
                   proportional to particle number in that size range.
      Source: Whitby and Sverdrup (1980).
1          An important feature of the mass or volume size distributions of atmospheric aerosol is
2     their multimodal nature. Volume distributions, measured in ambient air in the United States, are
3     almost always found to be bimodal, with a minimum between 1.0 and 3 //m. The distribution of
4     particles that are mostly larger than the minimum is termed "coarse". The distribution of
5     particles that are mostly smaller than the minimum is termed "fine." Whitby and Sverdrup
6     (1980); Whitby (1978); and Willeke and Whitby (1975) identified three modes:  nuclei,
7     accumulation, and coarse.  The three modes are most apparent in the freeway-influenced size
8     distribution of Figure 3-2b and in the surface area distribution of Figure 3-3b.  However, the
9     nuclei mode, corresponding to particles below about 0.1 //m, may not be noticeable in volume or
      October 1999
                        3-5
DRAFT-DO NOT QUOTE OR CITE

-------
/u -

65 -

60 -
55 -
50 -
_ 45 -
; 40 -
^35 -
i,30 -
7)
| 25 -
3 20 -
15 -
10 -
5 -



• \
• \ Rural

.' * 	 South-Central
r • New Mexico







/ v' A\
/ / \\
J'"'~\\ \
/ _.- --^=^: 	 ^s ^'~'~--..^^
      70
      65  -
      60  -
      55  -
      50  -
   «  45  -
   f 40  -
   ^35  -
   3
   9.30  -
    Ol
   =§ 25  "
   ^ 20  -
      15  -
      10  -
       5  -
       0
                                                                                Average Urban
                                                                                Urban + Freeway
              0.01      0.1       1        10
                       Particle Diameter, Dp (|jm)
100     0.01      0.1       1       10       100
                  Particle Diameter, Dp (|jm)
      Figure 3-2.  Particle volume distribution as a function of particle diameter: (a) for the
                   averaged rural and urban-influenced rural number distributions shown in
                   Figure 3-1 and a distribution from south central New Mexico, and (b) for the
                   averaged urban and freeway-influenced urban number distributions shown in
                   Figure 3-1.
      Source: Whitby and Sverdrup (1980) and Kim et al. (1993).
1     mass distributions. The middle mode, from 0.1 to 1 or 2 //m, is the accumulation mode. Fine
2     particles include both the accumulation and the nuclei modes.  The third mode, containing
3     particles larger than 1 or 2 //m, is known as the coarse particle mode. The number concentrations
4     of coarse particles are usually too small to see in arithmetic plots (Figures 3-lb and 3-3a) but can
5     be seen in a logarithmic plot (Figure 3-lb). Whitby and Sverdrup (1980) observed that rural
6     aerosols, not influenced by sources, have a small accumulation mode and no observable nuclei
7     mode. For urban aerosols, the accumulation  and coarse particle modes are comparable in
8     volume. The nuclei mode is small in volume but it dominates the number distributions of urban
9     aerosols. Whitby's conclusions were based on extensive studies of size distributions in a number
      October 1999
3-6
DRAFT-DO NOT QUOTE OR CITE

-------
                                   Nn = 7.7x 10
                                 DGNn = 0.013
                                                          Nc = 4.2
                                                       DGNC = 0.97
    CO
              30 H
       
           o
              10H
                          Vn = 0.33
                       DGVn = 0.031
                 0.001
0.01
0.1
1.0
 i
10
                                             Dr
100
Figure 3-3.  Distribution of coarse (c), accumulation (a), and nuclei or ultrafine (n), mode
           particles by three characteristics, volume (V), surface area (S), and number
           (N) for the grand average continental size distribution. DGV = geometric
           mean diameter by volume; DGS = geometric mean diameter by surface area;
           DGN = geometric mean diameter by number; Dp = geometric diameter.

Source: Whitby (1978).
October 1999
           3-7
          DRAFT-DO NOT QUOTE OR CITE

-------
 1      of western and midwestern locations during the 1970's (Whitby, 1978; Whitby and Sverdrup,
 2      1980).  No size-distribution studies of similar scope have been published since then.  Newer
 3      results from particle counting techniques and impactor, size-segregation studies, including data
 4      from Europe (U.S. Environmental Protection Agency, 1996) and Australia (Keywood et al.,
 5      1999), show similar results.
 6
 7      Definitions of Particle Size Fractions
 8           In the preceding discussion several subdivisions of the aerosol size distribution were
 9      identified. The aerosol community uses four different approaches or conventions in the
10      classification of particles by size:  (1) modes, based on the observed size distributions and
11      formation mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling
12      device; (3) occupational sizes, based on the entrance into various compartments of the respiratory
13      system; and (4) legally-specified, regulatory sizes for air quality standards.
14
15           Modal. The modal classification, first proposed by Whitby (1978), is shown in Figure 3-3.
16      The nuclei mode can be seen clearly in the volume distribution only in traffic or near traffic or
17      other sources of nuclei mode particles (Figure 3-4). The observed modal structure is frequently
18      approximated by several log-normal distributions. Definitions of terms used to describe size
19      distributions in modal terms are given below.
20
21           Coarse Mode: The distribution of particles with diameters mostly greater than the
22           minimum in the particle mass  distribution, which generally occurs between 1 and 3 //m.
23           These particles are usually mechanically generated.
24
25           Fine Mode: The distribution of particles with diameters mostly smaller than the minimum
26           in the particle mass distribution, which generally occurs between 1 //m and 3 //m.  These
27           particles are generated in combustion or formed from gases. The fine mode includes the
28           accumulation mode and the nuclei mode.
29
30           Nuclei Mode: That portion of the fine particle fraction with diameters below about 0.1 //m.
31           The nuclei mode can be observed as a separate mode in mass or volume distributions only
32           in clean or remote areas or near sources of new particle formation by nucleation.
33
34           Toxicologists use ultrafine to refer to particles, generated in the laboratory, which are in the
35           nuclei-mode size range. Aerosol physicists and material scientists tend to use nanoparticles
36           to refer to particles in this size range generated in the laboratory.

        October 1999                               3-8         DRAFT-DO NOT QUOTE OR CITE

-------
        C?4
         O)

            3
            2

            1
                                                                Mechanically
                                                                  Generated
              0.002
                  0.01
                      Nuclei Mode
 0.1              1
Geometric Diameter,  Dp, |jm
Accumulation Mode
10
                                                            Coarse Mode
100
                            Fine-Mode Particles
                                                       Coarse-Mode Particles
      Figure 3-4.  Volume size distribution, measured in traffic, showing fine-mode and
                  coarse-mode particles and the nuclei and accumulation modes within the
                  fine particle mode. DGV (geometric mean diameter by volume, equivalent to
                  volume median diameter) and og (geometric standard deviation) are shown for
                  each mode. Also shown are transformation and growth mechanisms (e.g.,
                  nucleation, condensation, and coagulation).

      Source:  Adapted from Wilson and Suh (1997).
1
2
3
4
5
6
7
     Accumulation Mode: That portion of the fine particle fraction with diameters above about
     0.1 //m. Accumulation-mode particles normally do not grow into the coarse mode.
     Nuclei-mode particles grow by coagulation (two particles combining to form one) or by
     condensation (low-equilibrium vapor pressure gas molecules condensing on a particle) and
     "accumulate" in this size range.

     Over the years, the terms fine and coarse, as applied to particle sizes, have lost the precise

meaning given in Whitby's (1978) definition. In any given article, therefore, the meaning of fine

and coarse, unless defined, must be inferred from the author's usage.  In particular, PM2 5 and
      October 1999
                                         3-9
                    DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
4
5
fine mode particles are not equivalent. In this document, the term mode is used with fine and
coarse when it is desired to specify the distribution of fine-mode particles or coarse-mode

particles as shown in Figures 3-4 and 3-5.
           70
       E
       E
 Q
 O)

 
-------
 1           Sampler Cut Point. Another set of definitions of particle size fractions arises from
 2      considerations of size-selective sampling.  Size-selective sampling refers to the collection of
 3      particles below or within a specified aerodynamic size range, usually defined by the upper 50%
 4      cut point size, and has arisen in an effort to measure particle size fractions with some special
 5      significance, e.g., health, visibility, source  apportionment, etc.  Dichotomous samplers split the
 6      particles into smaller and larger fractions, which may be collected on separate filters.  Cascade
 7      impactors use multiple size cuts to obtain a distribution of size cuts for mass or chemical
 8      composition measurements. One-filter samplers with a variety of upper size cuts have also been
 9      used.
10
11           Occupational Health Size Cuts.  The occupational health community has defined size
12      fractions for use in the protection of human health.  This  convention classifies particles into
13      inhalable, thoracic, and respirable particles according to their upper size cuts.  However, these
14      size fractions may also be characterized in  terms of their  entrance into various compartments of
15      the respiratory system. Thus, inhalable particles enter the respiratory tract, including the head
16      airways. Thoracic particles travel past the  larynx and reach the lung airways and the gas-
17      exchange regions of the lung.  Respirable particles are a subset of thoracic particles which are
18      more likely to reach  the gas-exchange region of the lung.  In the past exact definitions of these
19      terms have varied among organizations. As of 1993 a unified set of definitions was adopted by
20      the American Conference of Governmental Industrial Hygienists (ACGIH) (1994), the
21      International Standards Organization (ISO), and the European Standardization Committee
22      (CEN).  The curves which define inhalable (IPM), thoracic (TPM), and respirable (RPM)
23      particulate matter are shown in Figure 3-6.
24
25           Regulatory Size Cuts. In 1987, the NAAQS for PM were revised to use PM10, rather than
26      TSP, as the  indicator for the NAAQS for PM (Federal Register, 1987). The use of PM10 as an
27      indicator is  an example of size-selective sampling based on a regulatory size cut (Federal
28      Register, 1987). The selection of PM10 as an indicator was based on health considerations and
29      was intended to focus regulatory concern on those particles small enough to enter the thoracic
30      region. The PM25 standard, set in 1997, is also an example of size-selective sampling based on a
31      regulatory size cut (Federal Register,  1997). The PM2 5 standard was based primarily on

        October 1999                              3-11        DRAFT-DO NOT QUOTE OR CITE

-------
                                                                           10
                                                                    • IPM
                                                                    • TPM
                                                                    O RPM
                                                                    v PM
                              2.5     4             10    20           50
                                    Aerodynamic Diameter (|jm)
                                 100
      Figure 3-6. Specified particle penetration (size-cut curves) through an ideal (no-particle-
                 loss) inlet for five different size-selective sampling criteria. PM 10 is defined in
                 the Code of Federal Regulations (1991). PM25 is also defined in the Federal
                 Register (1997).  Size-cut curves for inhalable particulate matter (IPM),
                 thoracic particulate matter (TPM) and respirable particulate matter (RPM)
                 size cuts are computed from definitions given by American Conference of
                 Governmental and Industrial Hygienists (1994).
1     epidemiological studies using concentrations measured with PM2 5 samplers as an exposure

2     index.  However, the PM25 sampler was not designed to collect respirable particles. It was

3     designed to collect fine-mode particles because of their different sources (Whitby et al., 1974).

4     Thus, the PM25 standard will increase regulatory concern with the sources of fine-mode particles.

5          Prior to 1997, the indicator for the NAAQS for PM was total suspended particulate matter

6     (TSP).  TSP is defined by the design of the High Volume Sampler (hivol) which collects all of

7     the fine particles but only part of the coarse particles. The upper cut off size of the hivol depends
      October 1999
3-12
DRAFT-DO NOT QUOTE OR CITE

-------
 1      on the wind speed and direction, and may vary from 25 to 40 //m. Heroic measures, such as
 2      those undertaken with the Wide Range Aerosol Classifier (WRAC), are required to collect the
 3      entire coarse mode (Lundgren and Burton, 1995).
 4           An idealized distribution showing the normally observed division of ambient aerosols into
 5      fine-mode particles and coarse-mode particles, and the size fractions collected by the WRAC,
 6      TSP, PM10, PM25 and PM(10_25) samplers, is shown in Figure 3-5.  PM10 samplers, as defined in
 7      Appendix J to 40 Code of Federal Regulations (CFR) Part 50 (Code of Federal Regulations,
 8      199 la, Federal Register, 1987), collect all of the fine particles and part of the coarse particles.
 9      The upper cut point is defined as having a 50% collection efficiency at 10±0.5 //m aerodynamic
10      diameter. The slope of the collection efficiency curve is defined in amendments to 40 CFR,
11      Part 53, (Code of Federal Regulations, 1991b). An example of a PM10 size-cut curve is shown in
12      Figure 3-6.
13           An example of a PM25 size-cut curve is also shown in Figure 3-6. The PM25 size-cut
14      curve, however, is defined by the design of the Federal Reference Method Sampler. The basic
15      design of the FRM is given in the Federal Register (1997, 1998) and as 40 CFR Part 50,
16      Appendix L in the Code of Federal Regulations (Code of Federal Regulations, 1997a).
17      Additional performance  specifications are given in 40 CFR Parts 53 and 58  (Code of Federal
18      Regulations, 1997b). Each actual PM25 reference method, as represented by a specific sampler
19      design and associated manual  operational procedures, must be designated as a reference method
20      under Part 53 (see section 1.2  of Appendix L).  Thus there may be many somewhat different
21      PM25 FRMs (currently, 6 have been designated).
22           Papers discussing PM10 or PM2 5 frequently insert an explanation such as PMX (particles less
23      than x //m diameter) or PMX (nominally, particles with aerodynamic diameter x are collected and not all particles 
-------
       O
      '-i—>
       03
      -I—<
       0)
100
 90-
 80-
 70-
 60-
 50-
      S.    40H
            30-
            20-
            10-
             0-
       Kimoto cyclonic inlet
       	Manufacturer
          •  Tsai and Cheng (1996)
                                                               Wedding Cyclonic Inlet
                                                                  O  U=2 km/h
                                                                  D  U=8 km/h
                                                                  A  U=24 km/h
                                                                 Wedding and Weigand
                                                                        (1993)
               1
      Figure 3-7.
                                                8   10
                     20
30
                         Aerodynamic diameter, |jm
                                                           Source: Tsai and Cheng (1996)
       Comparison of penetration curves for two PM10 beta gauge samplers using
       cyclone inlets. The Wedding PM10 sampler uses the U.S. EPA definition of
       PMX as x = 50% cut point.  The Kimoto PM10 defines PMX as x = the 100% cut
       point (or zero penetration).
      Source: Tsai and Cheng (1996).
1          In an analysis reported in 1979, EPA scientists endorsed the need to measure fine and
2     coarse particles separately (Miller et al., 1979). Based on the availability of a dichotomous
3     sampler with a separation size of 2.5 //m, they recommended 2.5 //m as the cut point between
4     fine and coarse particles. Because of the wide use of this cut point, the PM2 5 fraction is
5     frequently referred to as "fine" particles. However, while the PM25 sample contains all of the
6     fine particles it may, especially in dry areas or during dry conditions, collect a small fraction of
7     the coarse particles.  A PM10-PM25 size fraction may be obtained from a dichotomous sampler or
8     by subtracting the mass collected by a PM2 5 sampler from the mass collected by a PM10 sampler.
      October 1999
                                   3-14
DRAFT-DO NOT QUOTE OR CITE

-------
 1      The resulting PM10-PM25 mass, or PM(10_25), is sometimes called "coarse" particles.  However,
 2      it would be more correct to call PM2 5 an indicator of fine-mode particles (since it contains some
 3      coarse-mode particles), PM(10_25) an indicator of the thoracic component of coarse-mode particles
 4      (since it excludes some coarse-mode particles below 2.5 //m and above 10 //m).  It would be
 5      appropriate to call PM10 an indicator of thoracic particles. PM10 and thoracic PM, as shown in
 6      Figure 3-6, have the same 50% cut point. However, the thoracic cut is not as sharp as the PM10
 7      cut so thoracic PM contains some particles between 10 and 30 //m diameter that are excluded
 8      from PM10.
 9
10      3.1.2.3  Nuclei-Mode Particles
11           Some further discussion of nuclei-mode particles is justified because of their possible
12      health importance.  The current PM25 standard is based largely on the statistical association of
13      health outcomes with particle mass.  A recent epidemiologic study in Erfurt, Germany (Peters
14      et al., 1997) found a better statistical association of asthma with particle number than with PM25.
15      Toxicologic studies (Oberdorster et al., 1992, 1994; Li et al., 1997; Lison et al., 1997) suggest
16      that nuclei-size particles (diameter 
-------
 1      growth of fine particles may be found in the AQC PM 1966 (U.S. Environmental Protection
 2      Agency, 1966).
 3
 4      Equilibrium Vapor Pressures
 5           An important parameter in particle nucleation and in particle growth by condensation is the
 6      saturation ratio S, defined as the ratio of the partial pressure of a species, p, to its equilibrium
 7      vapor pressure above a flat surface, p0: S = p/p0. For either condensation or nucleation to occur,
 8      the species vapor pressure must exceed its equilibrium vapor pressure. For particles, the
 9      equilibrium vapor pressure is not the same as p0. Two effects are important: (1) the Kelvin
10      effect, which is an increase in the equilibrium vapor pressure above the surface due to its
11      curvature; thus very small particles have higher vapor pressures and will not be stable to
12      evaporation until they attain a critical size and (2) the solute effect, which is a decrease  in the
13      equilibrium vapor pressure of the liquid due to the presence of other compounds in solution.
14           For an aqueous  solution of a nonvolatile salt, the presence of the salt decreases the
15      equilibrium vapor pressure of the water over the droplet. This effect is in the opposite direction
16      of the Kelvin effect, which increases the equilibrium vapor pressure above a droplet because of
17      its curvature.  The existence of an aqueous solution will also influence the vapor pressure of
18      water-soluble species. The vapor pressure behavior of mixtures of several liquids or of liquids
19      containing several solutes is complex.
20
21      New Particle Formation
22           When the vapor concentration of a species exceeds its equilibrium concentration (expressed
23      as its equilibrium vapor pressure), it is considered condensable.  Condensable species can either
24      condense on the surface of existing particles or can form new particles.  The relative importance
25      of nucleation versus condensation depends on the rate of formation of the condensable species
26      and on the surface or  cross-sectional  area of existing particles (McMurry and Friedlander, 1979).
27      In ambient urban environments, the available particle surface area is sufficient to rapidly
28      scavenge the newly formed condensable species.  Formation of new particles (nuclei mode) is
29      usually not important except near sources of condensable species.  Wilson et al. (1977)  report
30      observations of the nuclei mode in traffic. New particle formation can also be observed in
31      cleaner, remote regions. Bursts of new particle formation in the atmosphere under clean

        October 1999                              3-16        DRAFT-DO NOT QUOTE OR CITE

-------
 1      conditions correspond to low aerosol surface area concentrations (Covert et al., 1992). High
 2      concentrations of nuclei mode particles can occur in regions corresponding to low particle mass
 3      concentrations, indicating that new particle formation is inversely related to the available aerosol
 4      surface area (Clarke, 1992).
 5
 6      Sources of Nuclei Mode Particles
 7           Nuclei mode particles are the result of nucleation of gas phase species to form condensed
 8      phase species with very low equilibrium  vapor pressure. In the atmosphere there are four major
 9      classes of sources which yield particulate matter with equilibrium vapor pressures low enough to
10      form nuclei mode particles:
11           (1)  Particles containing heavy metals. Nuclei mode particles of metal oxides or other
12           metal compounds are generated when metallic impurities in coal or oil are vaporized during
13           combustion and the vapor undergoes nucleation.  Metallic ultrafine particles may also
14           formed from metals in lubricating oil or fuel additives that are vaporized during
15           combustion of gasoline or diesel fuels. Nuclei-mode metallic particles were discussed in
16           6.9 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996)
17           (2)  Elemental carbon or soot, Ce.  Ce particles are formed primarily by condensation of C2
18           molecules generated during the combustion process. Because Ce has a very low
19           equilibrium vapor pressure, ultrafine Ce particles can nucleate even at high temperatures
20           (Kittelson, 1998; Morawska et al.,  1998a).
21           (3)  Sulfates. Sulfuric acid (H2SO4), or its neutralization products with ammonia (NH3),
22           ammonium sulfate ((NH4)2SO4) or  ammonium acid  sulfate (NH4HSO4), are generated in the
23           atmosphere by conversion of sulfur dioxide (SO2) to H2SO4.  As H2SO4 is formed, it can
24           either nucleate to form new ultrafine particles or it can condense on existing nuclei mode or
25           accumulation mode particles. (Clark and Whitby, 1975; Whitby, 1978).
26           (4)  Organic carbon. Recent smog chamber studies and indoor experiments show that
27           atmospheric oxidation of certain organic compounds found in the atmosphere can produce
28           highly oxidized organic compounds with an equilibrium vapor pressure sufficiently low to
29           result in nucleation (Kamens et al., 1999; Weschler and Shields, 1999).
30
        October 1999                             3-17        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Concentration of Nuclei-Mode Particles: A Balance Between Formation and Removal
 2           Nuclei-mode particles may be removed by dry deposition or by growth into the
 3      accumulation mode.  This growth takes place as other low vapor pressure material condenses on
 4      the particles or as nuclei-mode particles coagulate with themselves or with accumulation mode
 5      particles.  Since the rate of coagulation would vary with the concentration of accumulation-mode
 6      particles, it might be expected that the concentration of nuclei-mode particles would increase
 7      with a decrease in accumulation-mode mass. On the other hand, the concentration of particles
 8      would be expected to decrease with a decrease in the rate of generation of particles by reduction
 9      in emissions of metal and carbon particles or a decrease in the rate of generation of H2SO4 or
10      condensable organic vapor. The rate of generation of H2SO4 depends on the concentration of
11      SO2 and OH, which is generated primarily by the photolysis of O3. Thus, the reductions in SO2
12      and O3 that are expected to form a major basis  for attaining PM25 and O3 standards and
13      implementation of Title II and Title IV Clean Air Act programs should lead to a decrease in the
14      rate of generation of H2SO4 and condensable organic vapor and a decrease in the concentration of
15      nuclei-mode particles.  These processes can be modeled using a general dynamic equation for
16      particle size distribution (Friedlander, 1977) or by aerosol dynamics modules in newer air quality
17      models (Binkowski and Shanker, 1995; Binkowski and Ching, 1996).
18           Since preliminary studies of the effects of ultrafine particles suggest the potential for
19      enhanced toxicity of this size range, further research in this area is important.  It is possible that
20      freshly generated ultrafine particles relatively near significant sources could present an additional
21      risk to health, above  those associated with particle mass. It will, therefore, be important to
22      monitor particle number and surface as well as mass to further delineate the relative effectiveness
23      of strategies for reducing particle mass, surface, and number.
24
25      3.1.3 Chemistry of Atmospheric PM
26           The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
27      ions; particle-bound water; elemental carbon; a great variety of organic compounds; and crustal
28      material. Atmospheric PM also contains a large number of elements in various compounds and
29      concentrations. More information, references, and the composition of PM, measured in  a large
30      number of studies in the U.S., may be found in AQC PM 96 (U.S. Environmental Protection

        October 1999                             3-18        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Agency, 1996). The composition and concentrations of PM are discussed in Chapter 4 of this
 2      document.
 3
 4      3.1.3.1   Chemical Composition and Its Dependence on Particle Size
 5           Studies conducted in most parts of the U.S. indicate that sulfate, ammonium, and hydrogen
 6      ions; elemental carbon, secondary organic compounds and some primary organic compounds;
 7      and certain transition metals are found predominantly in the fine particle mode. Crustal materials
 8      such as calcium, aluminum, silicon, magnesium, and iron are found predominately in the coarse
 9      particles. Some organic materials such as pollen, spores, and plant and animal debris are also
10      found predominantly in the coarse mode. Some components such as potassium and nitrate may
11      be found in both the fine and coarse particle modes but from different sources or mechanisms.
12      Potassium in coarse particles comes from soil. Potassium is also found in fine particles in
13      emissions from burning wood or cooking meat. Nitrate in fine particles comes primarily from
14      the reaction of gas-phase nitric acid with gas-phase ammonia to form particulate ammonium
15      nitrate.  Nitrate in coarse particles comes primarily from the reaction of gas-phase nitric acid with
16      pre-existing coarse particles.
17
18      3.1.3.2   Primary and Secondary Particulate Matter
19           Particulate material can be primary or secondary. PM is called primary if it is in the same
20      chemical form in which it was emitted into the atmosphere. PM is called secondary if it is
21      formed by chemical reactions in the atmosphere. Primary coarse particles are usually formed by
22      mechanical processes. This includes material emitted in particulate form such as wind-blown
23      dust, sea salt, road dust, and combustion-generated particles such as fly ash and soot. Primary
24      fine particles are emitted from sources, either directly as particles or as vapors which rapidly
25      condense to form particles. This includes soot from diesel engines as well as compounds of As,
26      Se, Zn,  etc., condensed from vapor formed during combustion or smelting. The concentration of
27      primary particles depends on their emission rate, transport and dispersion, and removal rate from
28      the atmosphere.
29           Secondary PM is formed by chemical reactions of free, adsorbed, or dissolved gases. Most
30      secondary fine PM is formed from condensable vapors generated by chemical reactions of
31      gas-phase precursors.  Secondary formation processes can result in either the  formation of new

        October 1999                             3-19        DRAFT-DO NOT QUOTE OR CITE

-------
 1      particles or the addition of particulate material to preexisting particles. Most of the sulfate and
 2      nitrate and a portion of the organic compounds in atmospheric particles are formed by chemical
 3      reactions in the atmosphere. Secondary aerosol formation depends on numerous factors
 4      including the concentrations of precursors; the concentrations of other gaseous reactive species
 5      such as ozone, hydroxyl radical, or hydrogen peroxide; atmospheric conditions including solar
 6      radiation and relative humidity; and the interactions of precursors and preexisting particles within
 7      cloud or fog droplets or on or in the liquid film on solid particles.  As a result, it is considerably
 8      more difficult to relate ambient concentrations of secondary species to sources of precursor
 9      emissions than it is to identify the sources of primary particles.
10
11      Formation ofSulfates and Nitrates
12          A substantial fraction of the fine particle mass, especially during the warmer months of the
13      year, is secondary sulfate and nitrate, formed as a result of atmospheric reactions. Such reactions
14      involve the gas phase  conversion of SO2 to H2SO4 by OH radicals  and aqueous-phase reactions of
15      SO2 with H2O2, O3, or O2 (catalyzed by Fe and Mn). These heterogeneous reactions may occur in
16      cloud and fog droplets or in films on atmospheric particles.  The NO2 portion of NOX can be
17      converted to HNO3 by reaction with OH radicals during the day. At night, NOX is also oxidized
18      to nitric acid by a sequence of reactions initiated by O3, that include nitrate radicals (NO3) and
19      dinitrogenpentoxide (N2O5). Both H2SO4 and HNO3 react with atmospheric ammonia (NH3).
20      Gaseous NH3 reacts with gaseous HNO3 to form particulate NH4NO3. Gaseous NH3 reacts with
21      H2SO4 to form acidic HSO4 (in NH4 HSO4) as well as in SO4 in (NH4)2SO4. In addition,  acid
22      gases such as SO2 and HNO3 may react with coarse particles  to form coarse secondary PM
23      containing sulfate and nitrate. Examples include reactions with basic compounds resulting in
24      neutralization, e.g., CaCo3 + 2NHO3 -> Ca (NO3)2 + H2CO31, or with salts of volatile acids
25      resulting in release of the volatile acid, e.g., SO2 + 2NaCl + H2O - Na2SO3 + 2HC11.
26           Chemical reactions of SO2 and NOX within plumes are an important source of H+, SO4 and
27      NO3. These conversions can occur by gas-phase and aqueous-phase mechanisms. In power-
28      plant or smelter plumes containing SO2 and NOX, the gas-phase chemistry depends on plume
29      dilution, sunlight and volatile organic compounds, either in the plume or in the ambient air
30      mixing into and diluting the plume.  For the conversion of SO2 to H2SO4, the gas-phase rate in
31      such plumes during summer midday conditions in the eastern United States typically varies

        October 1999                               3-20        DRAFT-DO NOT QUOTE OR CITE

-------
 1     between 1 and 3% h"1 but in the cleaner western United States rarely exceeds 1% h"1. For the
 2     conversion of NOX to HNO3, the gas-phase rates appear to be approximately three times faster
 3     than the SO2 conversion rates. Winter rates for SO2 conversion are approximately an order of
 4     magnitude lower than summer rates.
 5           The contribution of aqueous-phase chemistry to particle formation in point-source plumes
 6     is highly variable, depending on the availability of the aqueous phase (wetted aerosols, clouds,
 7     fog, and light rain) and the photochemically generated gas-phase oxidizing agents, especially
 8     H2O2 for SO2 chemistry. The in-cloud conversion rates of SO2 to SO4  can be several times
 9     larger than the gas-phase rates given above. Overall, it appears that SO2 oxidation rates to SO4
10     by gas-phase and aqueous-phase mechanisms may be comparable in summer, but aqueous phase
11     chemistry may dominate in winter.
12           In the western United States, markedly higher SO2 conversion rates have been reported in
13     smelter plumes than in power plant plumes. The conversion is predominantly by a gas-phase
14     mechanism. This result is  attributed to the lower NOX in smelter plumes. In power plant plumes
15     NO2 depletes OH and competes with SO2 for OH.
16           In urban plumes, the  upper limit for the gas-phase  SO2 conversion rate appears to be about
17     5% h"1 under the more polluted conditions. For NO2, the rates appear to be approximately three
18     times faster than the SO2 conversion rates. Conversion rates of SO2 and NOX in background air
19     are comparable to the peak rates in diluted plumes.  Neutralization of H2SO4 formed by SO2
20     conversion increases with plume age and background NH3 concentration. If the NH3
21     concentrations are more than sufficient to neutralize H2SO4  to (NH4)2SO4, the HNO3 formed from
22     NOX conversions may be converted to NH4NO3.
23
24     Formation of Secondary Organic PM
25           Atmospheric reactions, involving volatile organic compounds such as alkenes, aromatics,
26     and terpenes (or any reactive organic gas which contains at least seven carbon atoms), yield
27     organic compounds  with low ambient temperature vapor pressures which nucleate or condense
28     on existing particles to form secondary organic PM.  While  the mechanisms and pathways for
29     forming inorganic secondary particulate matter are fairly well known, those for forming
30     secondary organic PM are not as well understood.  Ozone and the hydroxyl radical are thought to
31     be the major initiating reactants. However, HO2 and NO3 radicals may also initiate reactions and

       October 1999                              3-21        DRAFT-DO NOT QUOTE OR CITE

-------
 1      organic radicals may be nitrated by HNO2, HNO3, or NO2.  Understanding the mechanisms of
 2      formation of secondary organic PM is important because secondary organic PM can contribute in
 3      a significant way to ambient PM levels, especially during photochemical smog episodes. Studies
 4      of the production of secondary organic PM in ambient air have focused on the Los Angeles
 5      Basin. Turpin and Huntzicker (1991) and Turpin et al. (1991) provided strong evidence that
 6      secondary PM formation occurs during periods of photochemical ozone formation in Los
 7      Angeles and that as much as 70% of the organic carbon in ambient PM was secondary in origin
 8      during a smog episode in 1987. Schauer et al. (1996) estimated that 20 to 30% of the total
 9      organic carbon PM in the <2.1 //m size range in the Los Angeles airshed is secondary in origin
10      on an annually averaged basis.
11           Pandis et al. (1992) identified three mechanisms for formation of secondary organic PM:
12      (1) condensation of oxidized end-products of photochemical reactions (e.g., ketones, aldehydes,
13      organic acids, and hydroperoxides); (2) adsorption of organic gases onto existing solid particles
14      (e.g., polycyclic aromatic hydrocarbons); and (3) dissolution of soluble gases which can undergo
15      reactions in particles (e.g., aldehydes). The first and third mechanisms are expected to be of
16      major importance during the summertime when photochemistry is at its peak. The second
17      pathway can be driven by diurnal and seasonal temperature and humidity variations at any time
18      of the year.  With regard to the first mechanism, Odum et al. (1996) suggested that the products
19      produced by the photochemical oxidation of reactive organic gases are semivolatile and can
20      partition themselves onto existing organic carbon at concentrations below their saturation
21      concentrations. Thus, the yield of secondary organic PM depends not only on the identity of the
22      precursor organic gas but also on the ambient levels of organic carbon capable of absorbing the
23      oxidation product.
24           Haagen-Smit (1952) first demonstrated that hydrocarbons irradiated in the presence of NOX
25      produce light scattering aerosols. The aerosol forming potentials of a wide variety of individual
26      anthropogenic and biogenic hydrocarbons were compiled by Pandis et al. (1992) based mainly on
27      estimates made by Grosjean and Seinfeld (1989) and data from Pandis et al. (1991) for p-pinene
28      and Izumi and Fukuyama (1990) for aromatic hydrocarbons. Zhang et al. (1992) examined the
29      oxidation of cc-pinene. Pandis et al. (1991) found no aerosol products formed in the
30      photochemical oxidation of isoprene,  although they and Zhang et al. (1992) found that the
31      addition of isoprene to reaction mixtures increased the reactivity of the systems studied. Further

        October 1999                             3-22        DRAFT-DO NOT QUOTE OR CITE

-------
 1      details about the oxidation mechanisms and secondary organic PM yields from various reactive
 2      organic gases given in the above studies and estimates of the production rate of secondary
 3      organic PM in the Los Angeles airshed are provided in the previous AQCD for PM (U.S.
 4      Environmental Protection Agency, 1996).
 5           More recently Odum et al. (1997a,b) have found that the aerosol formation potential of
 6      whole gasoline vapor can be accounted for solely by summing the contributions of the individual
 7      aromatic compounds in the fuel. In general, data for yields for secondary organic PM formation
 8      can be broken into two distinct categories. The oxidation of aromatic compounds containing
 9      zero or one methyl and ethyl groups (i.e., toluene, ethylbenzene, and ethyltoluenes) and
10      n-propylbenzene produce higher yields of secondary organic PM than did the oxidation of
11      aromatic compounds containing two or more methyl groups (i.e., xylenes, di-, tri-, and
12      tetra-methylbenzenes). Yields in the first group ranged from about 7 to 10% and in the second
13      group were generally between 3 and 4% within a range of existing organic carbon levels between
14      13 and 100 //g/m"3.  This grouping is consistent with those found by Izumi and Fukuyama (1990).
15      Reasons for the differences in secondary organic PM yields found between the two classes of
16      compounds are not clear.
17           Hoffmann et al. (1997) using the same framework described above found secondary
18      organic PM yields  of ~5%  for open-chain biogenic hydrocarbons such as ocimene and linalool,
19      5 to 25% for monounsaturated cyclic monoterpenes such as cc-pinene, d-3 carene and terpinene-
20      4-ol,  and ~40% for a cyclic monoterpene with two double bonds such as d-limonene. Secondary
21      organic PM yields  of close to 100% were observed during the photochemical oxidation of one
22      sesquiterpene, trans-caryophyllene.  These results were all obtained for initial hydrocarbon
23      mixing ratios of 100 ppb.
24           Kamens et al. (1999) observed secondary organic PM yields of 20-40% for cc-pinene.
25      Using information on the composition of secondary PM formed from cc-pinene (Jang and
26      Kamens, 1999), they were  able to calculate formation rates with a kinetic model including
27      formation mechanisms for O3 + cc-pinene reaction products.  Griffin  et al. (1999) introduced the
28      concept of incremental aerosol reactivity, the change in the secondary organic aerosol mass
29      produced (in //g/m3) per unit change of parent organic reacted (in ppb), as a measure of the
30      aerosol-forming  capability of a given parent organic compound in a prescribed mixture of other
31      organic compounds. They measured the incremental aerosol reactivity for a number of aromatic

        October 1999                              3-23         DRAFT-DO NOT QUOTE OR CITE

-------
 1      and biogenic compounds for four initial mixtures.  Incremental aerosol reactivity ranged from
 2      0.133 to 10.352 //gm"3 ppb"1 and varied by almost a factor of two depending on the initial
 3      mixture.
 4           Kao and Friedlander (1995) examined the statistical properties of a number of PM
 5      components in the South Coast Air Basin.  They found that the concentrations of non-reactive,
 6      primary components of PM10 have approximately log normal frequency distributions and
 7      constant values of geometric standard deviations (GSDs) regardless of source type and location
 8      within their study area. However, aerosol constituents of secondary origin (e.g., SO4=, NH4+, and
 9      NO3") were found to have much higher GSD's.  Surprisingly, the GSD's of organic (1.87) and
10      elemental (1.74) carbon were both found to be within  lo (0.14) of the mean GSD (1.85) for
11      non-reactive primary species, compared to  GSD's of 2.1 for sulfate, 3.5 for nitrate, and  2.6 for
12      ammonium. These results suggest that most of the organic carbon seen in ambient samples is of
13      primary origin. Pinto et al. (1995) found similar results for data obtained during the summer of
14      1994.  Further studies are needed to determine if these relations are valid at other locations and to
15      determine to what extent the results might be influenced by the evaporation of volatile
16      constituents after sampling.
17           It must be emphasized that the inferences drawn from field studies in the Los Angeles
18      Basin are unique to that area and cannot be extrapolated to other areas of the country.
19      In addition, there is a high degree of uncertainty associated with all aspects of the calculation of
20      secondary organic PM concentrations which is compounded by the volatilization of organic
21      carbon from filter substrates during and after sampling as well as potential positive artifact
22      formation from the  absorption  of gaseous hydrocarbon on quartz filters. Significant uncertainties
23      always arise in the interpretation of smog chamber data because of wall reactions. Limitations
24      also exist in extrapolating the results of smog chamber studies to ambient conditions found in
25      urban airsheds and forest canopies. Concentrations of terpenes and NOX are much lower in forest
26      canopies (Altshuller, 1983) than the levels  commonly used in smog chamber studies.  The
27      identification of aerosol products of terpene oxidation has not been a specific aim of field studies,
28      making it difficult to judge the results of model calculations of secondary organic PM formation.
29      Uncertainties also arise because of the methods used to measure biogenic hydrocarbon emissions.
30      Khalil and Rasmussen (1992) found much lower ratios of terpenes to other hydrocarbons (e.g.,
31      isoprene) in forest air than were expected, based on their relative emissions strengths  and rate

        October 1999                             3-24       DRAFT-DO NOT QUOTE OR CITE

-------
 1      coefficients for reaction with OH radicals and O3.  They offered two explanations, either the
 2      terpenes were being removed rapidly by some heterogeneous process or emissions were
 3      enhanced artificially by feedbacks caused by the bag enclosures they used.  If the former
 4      consideration is correct, then the production of aerosol carbon from terpene emissions could be
 5      substantial; if the latter is correct, then terpene emissions could have been overestimated by the
 6      techniques used.
 7
 8      3.1.3.3  Particle-Vapor Partitioning
 9           Several atmospheric aerosol species, such as ammonium nitrate and certain organic
10      compounds, are semivolatile and are found in both gas and particle phases. A variety of
11      thermodynamic models have been developed to predict the temperature and relative humidity
12      dependence of the ammonium nitrate equilibria with gaseous nitric acid and ammonia.  However,
13      under some atmospheric conditions, such as cool, cold, or very clean air, the relative
14      concentrations of the gas and solid phases are not accurately predicted by equilibrium
15      considerations alone, and transport kinetics can be important. The gas-particle distribution of
16      semivolatile organic compounds depends on the equilibrium vapor pressure of the compound,
17      total particle surface area,  particle composition, atmospheric temperature, and relative humidity.
18      Although it is generally assumed that the gas-particle partitioning of semivolatile organics is in
19      equilibrium in the atmosphere, neither the equilibria nor the kinetics of redistribution are well
20      understood. Diurnal temperature fluctuations, which cause gas-particle partitioning to be
21      dynamic on a time scale of a few hours, can cause semivolatile compounds to evaporate during
22      the sampling process. The pressure  drop across the filter can also contribute to loss of
23      semivolatile compounds.  The dynamic changes in gas-particle partitioning, caused by changes in
24      temperature, pressure and  gas-phase concentration, both in the atmosphere and after collection,
25      cause serious sampling problems which are discussed in Section 3.2.3.
26
27      Equilibria with Water Vapor
28           As a result of the equilibrium of water vapor with liquid water in hygroscopic particles,
29      many ambient particles  contain liquid water (particle-bound water). Unless removed, this
30      particle-bound water will be measured as a component of the particle mass. Particle-bound water
31      is important in that it influences the  size of the particles and in turn their aerodynamic properties

        October 1999                              3-25       DRAFT-DO NOT QUOTE OR CITE

-------
 1      (important for deposition to surfaces, to airways following inhalation, and in sampling
 2      instrumentation) and their light scattering properties. The aqueous solution provides a medium
 3      for reactions of dissolved gases, including reactions that do not take place in the gas phase. The
 4      aqueous solutions may also act as a carrier to convey soluble toxic species to the gas-exchange
 5      regions of the respiratory system, including species that would be removed by deposition in the
 6      upper airways if the gas phase (Friedlander and Yeh, 1998; Kao and Friedlander, 1995; Wilson,
 7      1995). An extensive review of this equilibrium as it pertains to ambient aerosols was given in
 8      Chapter 3 of the AQC PM 96 (U.S. Environmental Protection, Agency, 1996).
 9           Briefly the interaction of particles with water vapor may be described as follows.
10      As relative humidity increases, crystalline soluble salts in aerosol particles, such as (NH4) 2SO4,
11      NH4HSO4, or NH4NO3, undergo a phase transition to become aqueous solution aerosols.
12      According to the phase rule, for particles consisting of a single component, this phase transition
13      is abrupt, taking place at a relative humidity that corresponds to the vapor pressure of water
14      above the saturated solution (the deliquescence point).  With further increase in relative
15      humidity, the particle adds water (and the concentration of the solute decreases) so that the vapor
16      pressure of the solution is maintained equal to that of the surrounding relative humidity; thus the
17      particle tends to follow the equilibrium growth curve. As relative humidity decreases, the
18      particle follows the equilibrium curve to the deliquescence point.  However, rather than
19      crystallizing at the deliquescence relative humidity, the particle remains a solution in a
20      supersaturated solution to considerably lower relative humidities. Ultimately the particle
21      abruptly loses its water vapor (efflorescence), returning typically to the initial, stable crystalline
22      form.
23           For particles consisting of more than one component, the solid to liquid transition will take
24      place over a range of relative humidities, with an abrupt onset at the lowest deliquescence point
25      of the several components, and with subsequent growth as crystalline material in the particle
26      dissolves according to the phase diagram for the particular multicomponent system. Under such
27      circumstances a single particle may undergo several more or less abrupt phase transitions until
28      the soluble material is fully dissolved.  At decreasing relative humidity such particles  tend to
29      remain in solution to relative humidities well below the several deliquescence points.  In the case
30      of the sulfuric acid-ammonium sulfate-water system the phase diagram is fairly completely
31      worked out.  Mixed anion systems containing nitrate are more difficult due to the equilibrium

        October 1999                               3-26        DRAFT-DO NOT QUOTE OR CITE

-------
 1      between particulate NH4NO3 and gaseous NH3 and HNO3. For particles of composition
 2      intermediate between NH4HSO4 and (NH4)2SO4 this transition occurs in the range from 40% to
 3      below 10%, indicating that for certain compositions the solution cannot be dried in the
 4      atmosphere. At low relative humidities, particles of this composition would likely be present in
 5      the atmosphere as supersaturated solution droplets (liquid particles) rather than as solid particles,
 6      thus they would exhibit hygroscopic rather than deliquescent behavior during relative humidity
 7      cycles.
 8           Other pure compounds, such as sulfuric acid (H2SO4), are hygroscopic, i.e. they form water
 9      solutions at any relative humidity and maintain a solution vapor pressure over the entire range of
10      relative humidity. Soluble organic compounds may also contribute to the hygroscopicity of the
11      atmospheric aerosol (Saxena et al., 1995;  Saxena and Hildeman, 1996), but the equilibria
12      involving organic compounds and water vapor, and especially for mixtures of salts, organic
13      compounds and water, are not so well understood. These equilibrium processes may cause an
14      ambient particle to significantly increase its diameter at relative humidities above about 40%
15      (Figure 3-8). A particle can grow to 5 times its dry diameter as the RH approaches 100%
16      (Figure 3-9). The Federal Reference  Methods, for filter measurements of PM2 5 and PM10 mass,
17      require, after collection, equilibration at a specified, low relative humidity (-40% RH) to remove
18      particle-bound water (see 3.2 for details and references).  Otherwise, particle mass would be a
19      function of relative humidity and, at higher relative humidities, the particle mass would be
20      largely particle-bound water.  Continuous monitoring techniques must remove particle-bound
21      water before measurement, either by heating or dehumidification.  Semivolatile material may be
22      lost during sampling, lost during equilibration, and is certainly lost when the collected sample is
23      heated above ambient. In addition to problems due to the loss of semivolatile species, recent
24      studies have shown that significant amounts of particle-bound water are retained in particles
25      collected on impaction surfaces even after equilibration and that the amount of retained particle-
26      bound water increases with relative humidity during  collection (Hitzenberger et al., 1997). Large
27      increases in mass with increasing relative humidity were observed for the accumulation mode.
28      The change in particle size with relative humidity also means that particle measurements such as
29      surface area or volume, or composition as a function of size, must all be made at the same RH if
30      the results are to be comparable. These problems are addressed in more detail in Section 3.2 on
31      Measurement of Particulate Matter.

        October 1999                             3-27        DRAFT-DO NOT QUOTE OR CITE

-------
      2.0
   O
   Q.
  Q
       1.5-
       1.0
        0
H
                                         8
                                                                   - 7
                                                                   - 6
                                                                   h 5
                          (NH4)2 S04
                                               O
                                               Q.
                                                                           Q_
                                                                   h 3
                                                                   - 2
                                       - 1
                                         0
                   30
       50
       RH, %
  70
90
Figure 3-8.  Particle growth curves showing fully reversible hygroscopic growth of
           sulfuric acid (H2SO4) particles, deliquescent growth of ammonium sulfate
           [(NH4)2 SO4] particles at about 80% relative humidity (RH), hygroscopic
           growth of ammonium sulfate solution droplets at RH greater than 80%, and
           hysteresis (the droplet remains supersaturated as the RH decreases below
           80%) until the crystallization point is reached.

Source: National Research Council (1993) adapted from Tang (1980).
October 1999
          3-28
DRAFT-DO NOT QUOTE OR CITE

-------
 o
Q
Q
o"
"on
          5-
          3
      o
      5
          2  -
              0
                             T
                                       T
                   	 Theoretical Prediction at 22°c
                   ooooo Experimental Measurements
                                                           RH=99.8%
                                                           RH=98%
                                                           _^_i^^^^^^—
                                                           RH=96%
                                                                               216
                                                                        -125
                                                                             -64
                     -27
                                                                        -8
                                                                                0
                       50             100             150
                   NH4 HSO4  Dry Particle Diameter (nm)
                   200
                                                                                      o
      Figure 3-9.  Theoretical predictions and experimental measurements of growth of
                 NH4HSO4 and particles at relative humidity between 95 and 100%.
      Source: Lietal. (1992).
1     3.1.3.4 Removal Processes
2          The lifetimes of particles vary with size.  Coarse particles can settle rapidly from the
3     atmosphere within hours, and normally travel only short distances.  However, when mixed high
4     into the atmosphere, as in dust storms, the smaller-sized coarse-mode particles may have longer
5     lives and travel distances. Nuclei mode particles rapidly grow into the accumulation mode.
6     However, the accumulation mode does not grow into the coarse mode. Accumulation-mode fine
7     particles are kept suspended by normal air motions and have very low deposition rates to
8     surfaces.  They can be transported thousands of km and remain in the atmosphere for a number of
      October 1999
                                       3-29
DRAFT-DO NOT QUOTE OR CITE

-------
 1      days.  Coarse mode particles of less than ~10 //m diameter as well as accumulation-mode and
 2      nuclei-mode (or ultrafine) particles all have the ability to penetrate deep into the lungs and be
 3      removed by deposition in the lungs. Dry deposition rates are expressed in terms of a deposition
 4      velocity which varies as the particle size, reaching a minimum between 0.1 and 1.0 //m
 5      aerodynamic diameter. Accumulation-mode particles are removed from the  atmosphere
 6      primarily by cloud processes. Fine particles, especially particles with a hygroscopic component,
 7      grow as the relative humidity increases, serve as cloud condensation nuclei, and grow into cloud
 8      droplets. If the cloud droplets grow large enough to form rain, the particles are removed in the
 9      rain. Falling rain drops impact coarse particles and remove them. Ultrafine  or nuclei mode
10      particles are small enough to diffuse to the falling drop and be removed.  Falling rain drops,
11      however, are not effective in removing accumulation-mode particles.
12
13      3.1.3.5  Particulate Matter and Acid Deposition
14           The EPA is required by law to set primary standards to product human health and
15      secondary standards to mitigate welfare effects. The role of particles in reducing visibility and
16      affecting radiative balance through scattering and absorption of light is evident as are the effects
17      of particles in soiling and damaging materials.  Visibility effects are addressed through regional
18      haze regulations. The direct effects of particles in scattering and absorbing light and the indirect
19      effects of particles on clouds are being addressed in climate change programs in several
20      government agencies with the lead role assigned to the Department of Energy. These welfare
21      effects are discussed briefly in Chapter 9.
22           Concerns over the possible ecological effects of acid deposition in the United States led to
23      the creation of a major research program in 1980 under the new National Acid Precipitation
24      Assessment Program (NAPAP). However, the role of PM in acid deposition has not always been
25      recognized. Acid deposition and PM are intimately related, however, first because particles
26      contribute significantly to the acidification of rain and secondly because the gas phase  species
27      that lead to dry deposition of acidity are also precursors of particles. Therefore, reductions in
28      SO2 and NOX emissions will decrease both acid deposition and PM concentrations.
29           Sulfate, nitrate, and some partially oxidized organic compounds are hygroscopic and act as
30      nuclei for the formation of cloud droplets.  These droplets provide chemical reactors in which
31      (even slightly) soluble gases can dissolve and react. Thus SO2 can dissolve in cloud droplets and

        October 1999                              3-30        DRAFT-DO NOT QUOTE OR CITE

-------
 1      be oxidized to sulfuric acid by dissolved ozone or hydrogen peroxide.  These reactions do not
 2      take place in the gas phase but only in solution in water. Sulfur dioxide may also be oxidized
 3      more slowly by dissolved oxygen if metal catalysts such as iron or manganese are present in
 4      solution. If the  droplets evaporate, larger particles are left behind. If the droplets grow large
 5      enough to fall out they will fall as rain and the particles will be removed from the atmospheric
 6      with potential effects on the materials, plants, or soil on which the rain falls.  (Similar
 7      considerations apply to dew.) Atmospheric particles which nucleate cloud droplets may also
 8      contain other soluble or non-soluble materials such as metal salts and PNA organic compounds
 9      which may add  to the toxicity of the rain.  Thus the adverse effects of acid deposition on soils,
10      plants, and trees as well as lakes, streams and fish must be taken into account in setting
11      secondary PM standards.  These effects are discussed in Chapter 9.
12           Sulfuric acid, ammonium nitrate, and organic particles are also deposited on surfaces by dry
13      deposition. The utilization of nitrate by plants leads to the production of acidity.  Therefore, dry
14      deposition of particles can also contribute to the  ecological damages caused by acid deposition.
15
16      3.1.3.6  Particles as Carriers of Toxic Species
17           Wilson (1995) has suggested that particles could carry toxic species into the deep lung.
18      "It is possible that water-soluble gases, which would be removed by deposition to wet surfaces in
19      the upper respiratory system during inhalation, could dissolve in particle-bound water and be
20      carried with the particles into the deep lung. Water-soluble gases in polluted air include oxidants
21      such as O3, H2O2, and organic peroxides; acid gases such as SO2, HC1, HNO3, HONO, and formic
22      acid; and polar organic species such as formaldehyde." Friedlander and Yeh (1998) have
23      discussed this possibility for peroxides.  Wexler and Sarangapani (1998), quoted below, have
24      discussed this process and investigated it quantitatively.
25                 Air pollutants are deposited in the human airway via two pathways—particle
26           deposition and vapor deposition.  In the absence of particles, vapors deposit at
27           different locations in the lung depending on their solubility in mucus, which is over
28           99% water. High-solubility compounds, such as nitric acid or hydrogen peroxide,
29           are rapidly removed in the upper airways while low-solubility compounds, such as
30           oxygen or ozone, are less well removed and so penetrate to the alveoli.  Pollutant
31           deposition in the upper airways is less harmful than in the lower airways because

        October 1999                              3-31        DRAFT-DO NOT QUOTE OR CITE

-------
 1           upper airways clearance is more rapid and the epithelium is protected by a mucus
 2           layer. As a result, low-solubility pollutants, such as ozone, may harm the alveoli,
 3           while high-solubility pollutants, such as nitric acid, do not reach these tissues.
 4                 In the presence of aerosol particles, this scenario changes. Undermost
 5           ambient conditions, aerosol particles contain some liquid water so that soluble
 6           compounds are partitioned between the gas phase and the aerosol liquid-water phase.
 7           The degree of deposition via the gas compared to that via the particles is a function of
 8           a number of factors including the solubility of the compound and the liquid-water
 9           content of the aerosol. Since highly soluble compounds deposit in the upper airways,
10           particles may provide a vector for deposition of these compounds in the lower
11           airways. Lower-solubility compounds may persist in the vapor phase and so may
12           deposit in lower airway segments.
13                 As aerosols are inhaled, soluble vapors deposit on the mucus, disrupting the
14           gas-particle equilibrium, and the compound begins to evaporate from the aerosol
15           particles. If the evaporation is rapid, the pattern of deposition of the compound
16           will not be influenced by the presence of the particles, i.e., the deposition pattern is
17           essentially that of the vapor alone. If the evaporation is very slow compared to a
18           breathing cycle, a significant amount of the compound will remain in the particle
19           phase and the pattern of deposition may be shifted toward the pulmonary region by
20           the particles.
21           Wexler and Sarangapani (1998) calculated the concentrations of a water-soluble gas,
22      relative to its initial concentration, as a function of airway generation (a measure of penetration
23      into the lung) for a range of Henry's law coefficients (a measure of solubility) for 0.1, 0.3, and
24      1 //m diameter particles,  assuming that there is no resistence to evaporation of the gas from the
25      particle-bound water. Wexler and Sarangapani conclude that, "Particles do not increase vapor
26      deposition in human airways".  This conclusion is based largely on the fact that only a very small
27      fraction  of the gas is dissolved in the particle-bound water for normal relative humidities.
28      However, their calculations do show that soluble gases are carried to higher generation airways
29      (deeper into the lung) in the presence of particles than in the absence of particles.
30           Underhill (1999) has pointed out that species such as SO2 and formaldehyde either react to
31      form different species (SO2 ^ H+ + HSO7 ) or hydrates (HCHO + H2O ^ CH2 (OH)2}. These

        October 1999                              3-32        DRAFT-DO NOT QUOTE OR CITE

-------
 1      reactions reduce the concentration of the dissolved gas-phase species and provide a kinetic
 2      resistence to the evaporation of the dissolved gas. Thus, the concentration of the dissolved
 3      species may be greater than that predicted by the equilibrium calculations of Wexler and
 4      Sarangapani. Underhill further points out that Amdur (1957, 1960, 1966, 1969), Amdur and
 5      Mead (1958), and Amdur and Underhill (1968), in studies of the effects of particles and gases on
 6      pulmonary resistence in guinea pigs, observed synergistic effects between NaCl particles and SO2
 7      and HCHO, but not between NaCl and formic acid. The reaction rates for converting the reacted
 8      species back to the gas-phase compound is fast enough to maintain equilibrium conditions for
 9      formic acid, but not for SO2 and HCHO.
10           It is also possible that toxic gases could be absorbed on solid particles thus be carried into
11      the lungs. The gases might be desorbed in the higher temperature and relative humidity of the
12      lung. If the toxic gas remains absorbed on the particle, it could be brought into direct contact
13      with cells in the respiratory system.  The thermal release of NO from ambient air and diesel
14      particles has been studied (Ball et al., 1999) as well as the absorption of NO2 on carbon particles
15      (Kalberer et al., 1999). The partitioning of semivolatile organic compounds between the gas-
16      phase and the liquid organic layer of an atmospheric particle has been investigated and modeled
17      (e.g., Pankow et al., 1993; Pankow, 1994a,b; Jang et al., 1997). Less information is available on
18      the kinetics of the partitioning process. However, Kamens and Coe (1997) have measured the
19      rate of evaporation of polycyclic aromatic hydrocarbons (PAHs) from fresh diesel soot particles.
20      For some PAHs the evaporation rates were slow enough (order of seconds) to allow particles to
21      carry PAHs into higher generation airways. Kamens et al. (1999) also report calculations of
22      desorption rates for certain products of the O3 + cc-pinene reaction from the secondary particles
23      formed in the reaction. Muzyka et al. (1998) report that particulate matter from diesel exhaust
24      contains absorbed benzene as well as polycyclic aromatic hydrocarbons (PAH) and nitro-PAH.
25      Recent studies also suggest that a variety of allergens may be absorbed on atmospheric particles
26      and carried into the lung and contribute to aggregation of allergy and asthma (Schappi et al.,
27      1999; Ormstad et al., 1998).
28
29      3.1.3.7   Separation of Fine and Coarse Particles
30           The many reasons for wanting to collect fine and coarse particles separately and
31      considerations as to the appropriate cutpoint for separating fine and coarse particles were

        October 1999                              3-33         DRAFT-DO NOT QUOTE OR CITE

-------
 1      discussed in Chapter 3 of the AQC PM 96 (U.S. Environmental Protection Agency, 1996).
 2      A review of atmospheric particle-size-distribution data did not provide a clear or obvious
 3      rationale for selection of an appropriate cutpoint. Depending on conditions, a significant amount
 4      of either fine or coarse mode material may be found in the intermodal region between 1.0 and
 5      3 //m.  However, the analysis of the existing data did demonstrate the important role of relative
 6      humidity in influencing the size of the fine particle mode and indicated that significant fine mode
 7      material is found above 1.0 //m only during periods of very high relative humidity.
 8          Thus, a PM2 5 sample will contain most of the fine mode material, except during periods of
 9      RH near 100 %.  However, especially in conditions of low RH, it may contain 5 to 20 % of the
10      coarse mode material below 10 //m in diameter. A PMLO sample will prevent misclassification of
11      coarse mode material as fine but under high RH conditions could result in some of the fine mode
12      material being misclassified as coarse. A reduction in RH, either intentionally or inadvertently,
13      will reduce the size of the fine mode. A sufficient reduction in RH will yield a dry fine particle
14      mode with very little material above 1.0 //m.  Studies of the changes in particle size with changes
15      in relative humidity suggest that only a small amount of accumulation mode particles will be
16      above  1 //m in diameter at RHs below 60% but a substantial fraction will grow above 1 //m for
17      RH above 80% (Hitzenberger et al, 1997; McMurry and Stolzenberg, 1989; U.S. Environmental
18      Protection Agency, 1996). As discussed in Section 3.2, some new techniques have been
19      developed for both integrated and continuous measurement of fine particulate matter minus
20      particle-bound water, but including semivolatile nitrate and organic compounds.  These
21      techniques require reduction of RH prior to collection. With such techniques PMj 0 would be an
22      appropriate cut-point.
23
24      Summary
25          The physical and chemical properties of ultrafine mode, accumulation mode, and coarse
26      mode particles are summarized in Table 3-1.
27
28
29      3.2  MEASUREMENT OF PARTICULATE MATTER
30          The 1996 Air Quality Criteria Document for Particulate Matter (AQC PM 96) (U.S.
31      Environmental Protection Agency, 1996) summarized sampling and analytical techniques for
        October 1999                             3-34        DRAFT-DO NOT QUOTE OR CITE

-------
                  TABLE 3-1. COMPARISON OF AMBIENT PARTICLES
       FINE MODE (Nuclei Mode Plus Accumulation Mode) AND COARSE MODE
                                      Fine
                                                              Coarse
                        Nuclei
                           Accumulation
 Formed from:
 Formed by:
         Combustion, high temperature
       processes and atmospheric reactions
 Composed of:
 Solubility:
 Sources:
 Atmospheric
 half-life:

 Removal
 Processes:
Nucleation
Condensation
Coagulation
Sulfates
Elemental carbon
Metal compounds
Organic compounds
with very low,
saturation vapor
pressure at ambient
temperature
Probably less
soluble than
accumulation mode

Combustion
Atmospheric
transformation of
SO2 and some
organic compounds
High temperature
processes
Minutes to hours
Grows into
accumulation mode
 Travel distance:   
-------
 1     particulate matter (PM) and acid deposition that had appeared in the literature since the 1982 Air
 2     Quality Criteria for Particulate Matter (U.S. Environmental Protection Agency, 1982). Two
 3     other excellent reviews have been published in recent years by Chow (1995) and McMurry
 4     (1999). This section will concentrate on problems in measuring PM, new techniques that attempt
 5     to alleviate these problems or measure problem species, the current EPA monitoring program
 6     (including measurements with Federal Reference Methods, speciation monitors, and continuous
 7     monitors), and the importance of intercomparison studies in the absence of any reference
 8     standard for suspended atmospheric particles.
 9
10     3.2.1 Problems in Measuring Particulate Matter
11           The decision by the US EPA to revise the PM standards by adding daily and yearly
12     standards for PM2 5 has led to a renewed interest in the measurement of atmospheric particles and
13     also to a better understanding of the problems in developing precise and accurate measurements
14     of particles. Unfortunately, it is very difficult to measure and characterize particles suspended in
15     the atmosphere.
16           The US Federal Reference Methods (FRM) for PM25 and PM10 provide relatively precise
17     (±10 %) methods for determining the mass of material remaining on a Teflon filter after
18     equilibration.  However, numerous uncertainties remain as to the relationship between the mass
19     and composition of material remaining on the filter, as measured by the FRMs, and the mass and
20     composition of material that exists in the atmosphere as suspended PM. The goal of a PM
21     indicator might be to accurately measure what exists as a particle in the atmosphere. However,
22     this is not currently possible, in part because of the difficulty of creating a reference standard for
23     particles suspended in the atmosphere.  As a result, EPA defines accuracy for PM measurements
24     in terms of agreement  of a candidate sampler with a reference sampler. Therefore,
25     intercomparisons of samplers become very important in determining how well various samplers
26     agree and how various design choices influence what is actually measured.
27           There are five general areas where choices must be made in designing an aerosol indicator.
28     These include treatment of semivolatile components; selection of an upper cut point; separation
29     of fine-mode and coarse-mode PM; treatment of pressure, temperature, and relative humidity;
30     and how to assess the reliability of the measurement technique. In many cases choices have been
31     made by default and with inadequate understanding of the consequences. As a result,
       October 1999                             3-36        DRAFT-DO NOT QUOTE OR CITE

-------
 1     measurement methods developed by different organizations may give different results when
 2     sampling the same atmosphere, even though the techniques appear to be identical.
 3
 4     3.2.1.1   Treatment of Semivolatile Components of PM
 5           Current filtration-based mass measurements lead to significant evaporative losses, during
 6     and possibly after collection, of a variety of semivolatile components, i.e., species which exist in
 7     the atmosphere in dynamic equilibrium between the condensed phase and gas phase. Important
 8     examples include ammonium nitrate, semivolatile organic compounds, and particle-bound water.
 9     This problem is illustrated in Figure 3-10.
10           Possible approaches that have been used to address the problem of potentially lost
11     semivolatile components include the following which will be discussed in more detail in
12     subsequent sections:
13     1.  Collect/measure all components present in the atmosphere in the condensed phase except
14         particle-bound water. (Examples: Brigham Young absorptive sampler, Harvard pressure
15         drop monitor. Both require pre-concentration of the accumulation mode and reduction of
16         ambient humidity.)
17     2.  Stabilize PM at a specified temperature high enough to remove all particle-bound water.
18         This results in loss of most of the semivolatile PM. (Examples: TEOM operated at 50°C,
19         beta gauge with heated inlet.)
20     3.  Equilibrate collected material at fixed, near-room temperature and low relative humidity to
21         remove particle-bound water. Accept loss of an unknown but possibly significant fraction of
22         semivolatile PM. (Example: US Federal Reference Method and most filter-weighing
23         techniques. There is also information to suggest that not all particle-bound water is removed
24         by the equilibration process.)
25           The semivolatile artifact is composition dependent and has been shown to be significant in
26     air sheds with high nitrate, wood smoke or secondary organic aerosols.
27
28     3.2.1.2   Upper Cut Point
29           A technique must be used that gives  an upper cut-point, and its standard deviation, that is
30     independent of wind speed and direction (the classical high volume sampler head was
31     unsatisfactory because of radial asymmetry).

       October 1999                              3-37       DRAFT-DO NOT QUOTE OR CITE

-------
       Should be
        retained
                                                      Particle-bound water
                                                       should be removed
                           elemental
                         Mineral/Metal
                                                      1.0
                           Aerodynamic Diameter, |jm
                           2.5
           Semivolatile components subject to evaporation during or after sampling
      Figure 3-10.  Schematic showing major nonvolatile and semivolatile components of PM 2_5.
                 Semivolatile components are subject to partial to complete loss during
                 equilibration or heating. The optimal technique was remove all
                 particle-bound water but no ammonium nitrate or semivolatile organic PM.
1         A separation which simulates the removal of particles by the upper part of the human

2    respiratory system would appear to be a good choice, i.e., measure what gets into the lungs. The

3    ACGIH-ISO-CEN penetration curve for thoracic particles, with a 50% cut-point at 10 //m
     October 1999
3-38
DRAFT-DO NOT QUOTE OR CITE

-------
 1      aerodynamic diameter (AD), would be an appropriate choice. (Thoracic particles are able to pass
 2      the larynx and penetrate into the bronchial and alveolar regions of the lung.) Some countries,
 3      however, use PM10 to refer not to samplers with a 50% cut at 10 //m AD but samplers with 100%
 4      rejection of all particles greater than 10 //m AD. Such samplers miss too much of the thoracic
 5      PM.  The US PM10 separation curve, while sharper than the thoracic curve, is probably
 6      satisfactory both for regulatory and health risk monitoring.  It has the advantage of reducing the
 7      problem of maintaining the finite collection efficiency specified by the thoracic penetration curve
 8      for particles larger than 10 //m AD. (See Figure 3-6 and Section 3.1.2.2.)
 9
10      3.2.1.3  Separation of Fine-Mode and  Coarse-Mode PM
11           Fine-mode and coarse-mode particles differ not only in size and morphology (e.g., smooth
12      droplets vs rough solid particles) but also in formation mechanisms; sources; and chemical,
13      physical, and biological properties. They also differ in terms of dosimetry (deposition in the
14      respiratory system), toxicity, and health effects as observed by epidemiologic studies.
15           At high relative humidity, such as that found in fog and clouds, hygroscopic fine-mode
16      particles will increase in size  due to accretion of particle-bound water.  Some, originally
17      sub-micrometer, fine-mode PM may be found with an AD above 1 //m. At very low relative
18      humidity, coarse-mode particles may be  fragmented into smaller sizes and small amounts of
19      coarse-mode PM may be found with an AD below 1 //m. It is desirable to separate fine-mode
20      PM and coarse-mode PM as cleanly as possible in order to properly allocate health effects to
21      either fine-mode PM or coarse-mode PM and to correctly determine sources by factor analysis
22      and/or chemical mass balance. For example sulfate in the fine-mode is associated with hydrogen
23      and/or ammonium ions; sulfate in the coarse mode is associated with basic metal ions.  The
24      sources are different and the health effects are likely to be different. Transition metals in the
25      coarse mode are likely to be associated with soil and tend to be less soluble than transition metals
26      in the fine mode which may be found in  fresh combustion particles.
27           The current practice of separating fine-mode and coarse-mode particles at 2.5 //m AD,
28      while satisfactory for a health-based standard, does not provide an adequate separation for
29      epidemiologic studies, especially in areas where winds cause high concentrations of wind blown
30      soil, or for the determination of source categories to guide control strategy. A possible approach,
31      which would provide much better separation  of fine-mode PM and coarse-mode PM, would be to

        October 1999                              3-39         DRAFT-DO NOT QUOTE OR CITE

-------
 1      dehumidify the air stream to some fixed humidity that would remove all or most particle-bound
 2      water without evaporating semivolatile components and make the cut near 1 //m AD. (See
 3      Section 3.1.3.7.)
 4
 5      3.2.1.4  Treatment of Pressure, Temperature, and Relative Humidity
 6          There are a variety of techniques for defining (or ignoring) the pressure, temperature, and
 7      relative humidity during or after sampling.
 8      Temperature  and Pressure:
 9        a. Sample volume based on mass or volumetric flow corrected to standard temperature and
10           pressure (273 °K and 1 atm.) (former EPA technique for PM10).
11        b. Sample volume based on volumetric flow at ambient conditions of temperature and
12           pressure (current EPA technique for PM2 5 and PM10).
13      Temperature  During Collection:
14        a. Heat enough to remove all particle-bound water (i.e., TEOM at 50 °C).
15        b. Heat several degrees to prevent condensation of water in sampling system.
16        c. Try to maintain sampler near ambient temperature.
17        d. Maintain sampler at constant temperature inside heated/air conditioned shelter.
18      Temperature  After Collection:
19        a. No control
20        b. Constant Temperature (room temperature)
21        c. Store at cool temperature (4 °C)
22      Relative Humidity: Changes in relative humidity cause changes in particle size of hygroscopic or
23      deliquescent particles. Changing relative humidity by adding or removing water vapor affects
24      measurements of:
25        a. Particle number, particle surface area and particle size distribution
26        b. Amount of overlap of fine-mode and coarse-mode particles
27        Changing  relative humidity by intentional or inadvertent changes in temperature affects above
28      measurements plus:
29        c. Amount of loss of ammonium nitrate and semivolatile organic compounds.
30          Studies  of relationships between personal/indoor/outdoor measurements present special
31      problems.  Indoor environments are typically dryer than outdoors and may be warmer or, if

        October 1999                            3-40        DRAFT-DO NOT QUOTE OR CITE

-------
 1      air-conditioned, cooler. These differences may change particle size and the amount of
 2      volatilization of semivolatile components. Such changes between indoors and outdoors will
 3      complicate the comparison of indoor to outdoor concentrations, the modeling of personal
 4      exposure to all particles, and the disaggregation of personal exposure into exposure to particles
 5      of ambient origin and exposure to particles of indoor origin.
 6
 7      3.2.1.5  No Way to Determine Accuracy
 8           Precision and accuracy are normally used to describe the quality of a measurement.
 9      Precision is typically determined by comparison of collocated samplers or through replicate
10      analyses, while accuracy is determined through the use of traceable calibration standards.
11      Unfortunately, no standard reference material for suspended PM exists. Therefore, it is not
12      possible to establish the accuracy of a PM monitoring technique. Intercomparison studies, to
13      establish the precision of identical monitors and the extent of agreement between different types
14      of monitors are essential for establishing the reliability of PM measurements. Intercomparison
15      studies have contributed greatly to our understanding of the problems in PM measurement. Such
16      studies will be discussed as they apply to  specific measurement problems, monitoring
17      instruments, or analytical techniques. Measurement errors of concern in PM10 sampling that arise
18      due to uncertainty tolerances in cutpoint; particle bounce and re-entrainment; impactor surface
19      overloading; and losses to sampler internal surfaces were discussed in detail in the AQC PM 96
20      (U.S. Environmental Protection Agency, 1996). Measurement errors of concern in PM25
21      sampling arise because of our inability to  assess accuracy in an absolute sense due to a lack of
22      primary calibration standards, because of the use of an operational  definition of PM25 as a
23      surrogate for fine particles, and because of problems associated with trying to measure the mass
24      of particles as they exist in the air rather than what remains after collection on a filter.
25           Because of the difficulties associated with determining the accuracy of PM measurements,
26      the U.S. Environmental Protection Agency has sought to  make FRM measurements equivalent by
27      specifying operating conditions and, in the case of PM25 samplers,  by specifying details of the
28      sampler design. Thus, both the PM10 as well as PM2 5 standards are defined with consistency of
29      measurement technique, rather than accuracy of the true mass concentration measurement, in
30      mind (McMurry, 1999). It is acknowledged in the Federal Register (1997) that "because the size
31      and volatility of the particles making up ambient particulate matter vary over a wide range and

        October 1999                              3-41         DRAFT-DO NOT QUOTE OR CITE

-------
 1      the mass concentration of particles varies with particle size, it is difficult to define the accuracy
 2      of PM25 measurements in an absolute sense..."  Thus, accuracy is defined as the degree of
 3      agreement between a subject field PM2 5 sampler and a collocated PM2 5 reference method audit
 4      sampler (McMurry, 1999).  The Federal Reference Method (FRM) for PM2 5 will be discussed in
 5      Section 3.2.3.3.  The accurate measurement of the mass of PM suspended in the atmosphere was
 6      of concern in PM10 sampling and is of even greater concern for  PM25.  As mentioned earlier,
 7      volatilization losses, during sampling or post-sampling handling, of some organics as well as
 8      ammonium nitrate can lead to significant underestimation of the true fine particulate mass
 9      concentration in some locations.  Sources of error in the measurement of true PM25 mass also
10      arise due to adsorption of organic vapors onto collected PM, filter media, or other sampler
11      surfaces; neutralization of acid or basic vapors on either filter media or collected PM; and the
12      role of particle-bound water in PM sampling.
13          The lack of a standard reference material or a primary calibration standard for PM
14      suspended in the atmosphere has prevented any  evaluation of accuracy. In the last 25 years, there
15      have been advancements in the generation of monodisperse aerosols and classification, as well as
16      in the development of electron microscopy and imaging analysis, that have contributed to the
17      advancement in  aerosol calibration (Chen, 1993). Still, one of the limitations in PM sampling
18      and analysis remains the lack of primary calibration standards for evaluating analytical methods
19      and for intercomparing laboratories. Klouda et al. (1996) examined the possibility of
20      resuspending the NIST Standard Reference Material 1649 (Urban Dust) in air for collection on
21      up to 320 filters  simultaneously, using SRI, International's dust generation and collection system,
22      however little additional work in this area has been reported.
23          Methods validation was discussed in the previous AQC PM 96 (U.S. Environmental
24      Protection Agency, 1996), and the usefulness of intercomparisons and "internal redundancy" was
25      emphasized.  For example, a number of internal consistency checks are applied to the IMPROVE
26      network (Malm et al., 1994).  These include mass balances, sulfur measurements by both proton
27      induced x-ray emission (PIXE) and ion chromatography (1C); and comparison of organic matter
28      by combustion and by proton elastic scattering analysis (PESA) analysis of hydrogen. Mass
29      balances compare the gravimetrically determined mass with the mass calculated from the sum of
30      the major chemical components, i.e. crustal elements plus associated oxygen, organic carbon,
31      elemental carbon, sulfate, nitrate, ammonium, and hydrogen ions. Mass balances are useful

        October 1999                             3-42        DRAFT-DO NOT QUOTE OR CITE

-------
 1      validation techniques, however, they do not check for, or account for, artifacts associated with
 2      the absorption of gases during sampling, or the loss of semi-volatile material during sampling.
 3      The mass balance check may appear reasonable even if such artifacts are present, since only the
 4      material collected on the filter is included in the balance.
 5
 6      3.2.2 Why Measure Particles
 7      3.2.2.1  Attainment of a Standard
 8           A critical need for particle measurements is to determine if a location is in compliance with
 9      an existing standard and to determine if trends show improvements in air quality. For this
10      purpose, precision of the measurement by the variety of indicators in use is the most important
11      consideration.  Therefore, intercomparisons of various potential indicators, under a variety of
12      atmospheric and air quality conditions are essential. PM standards are currently based on
13      24-hour measurements, with measurements every sixth day being used to estimate an annual
14      average.
15
16      3.2.2.2  Implementation of a Standard
17           In order to develop State Implementation Plans, to reduce pollution in order to attain a
18      standard, local  agencies and national research organizations need measurements to identify
19      source categories and to develop and validate air quality models.  For these purposes PM
20      parameters other than mass, such as chemical composition and size distribution, must be
21      measured.  Also measurements are needed with shorter time resolution in order to match changes
22      in pollution associated with diurnal changes in the boundary layer.
23
24      3.2.2.3  Determination of Health Effects
25           PM measurements are needed to determine exposure for use in epidemiological studies, to
26      assess exposure for risk assessment and to determine components of PM to guide planning and
27      interpretation of toxicologic experiments. For these purposes size and chemical composition
28      may be needed. For exposure assessment, different measurement time intervals may be needed.
29      For acute epidemiology, one-hour or continuous measurements may be needed as well as 24-hour
30      measurements. For chronic epidemiologic studies, measurements which integrate over a week to
31      a month may be more cost effective. For dosimetric studies and  modeling, information will be
        October 1999                             3-43        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      needed on the particle size distribution and on the behavior of particles as the relative humidity
 2      and temperature are increased to those in the respiratory system.
 3
 4      3.2.2.4  Determination of Ecological Effects
 5           Measurement of particles, and of the chemical components of particulate matter in rain, fog
 6      and dew, are needed to understand the contributions of PM to soiling of surfaces and damage to
 7      materials and to understand the wet and dry deposition of acidity and toxic substances to surface
 8      water, soil and plants.  Some differentiation into particle size is needed to determine dry
 9      deposition.  Information on chemical composition is also needed to understand materials damage
10      and ecological damage.
11
12      3.2.2.5  Determination of Radiative Effects
13           Particles reduce visibility by scattering and absorbing light. They also have a direct effect
14      on the climate by scattering visible and ultraviolet light back into space and indirectly, as cloud
15      condensation nuclei, by changing the albedo and stability of clouds.  For understanding these
16      effects information is needed on refractive index (including ratio of scattering to absorption), size
17      distribution, and change in particle size with change in relative humidity.
18
19      3.2.2.6  PM Components/Parameters Which Need To Be Measured
20           The large variety of components of PM or PM parameters that need to be measured for
21      various purposes are summarized in Table 3-2.
22
23      3.2.3 Problems Associated with Semivolatile Particulate Matter
24           It is becoming increasingly apparent that the semivolatile component of PM may
25      significantly impact the quality of the measurement, and can lead to both positive and negative
26      sampling artifacts. Losses of semivolatile species, like ammonium nitrate and many organic
27      species, may occur during sampling, due to changes in temperature, relative humidity, or
28      composition of the aerosol, or due to pressure drop across the filter (McMurry, 1999).
29      Semivolatile species may adsorb onto, or react with, filter media and/or collected PM, leading to
30      a positive sampling artifact. Tsai and Huang (1995) observed positive sulfate and nitrate artifacts
31      on high volume PM10 quartz filters and attributed the artifacts to interactions between acidic
        October 1999                              3-44        DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 3-2.  PM COMPONENTS/PARAMETERS OF INTEREST FOR HEALTH,
              ECOLOGICAL, OR RADIATIVE EFFECTS; FOR SOURCE CATEGORY
          APPORTIONMENT STUDIES; OR FOR AIR QUALITY VALIDATION STUDIES
        Particle number
        Particle surface area
        Particle size distribution
        PM mass (fine-mode {PM: „} and coarse-mode {PM 1(M} mass as well as PM2 5 and PM10; nonvolatile mass,
        Federal Reference mass, and mass including semivolatile components such as ammonium nitrate and semivolatile
        organic compounds but not particle-bound water)
        Ions (sulfate, nitrate and ammonium)
        Strong acidity (H+)
        Elemental carbon
        Organic carbon (total, nonvolatile and semivolatile; functional groups and individual species)
        Transition metals (water soluble, bioavailable, oxidant generation)
        Specific toxic elements
        Crustal elements
        Bioaerosols
        Particle refractive index (real and imaginary)
        Particle density
        Particle size change with changes in relative humidity	
1      gases SO2, HONO and HNO3 and both the filter media (either glass fiber or quartz) and the
2      coarse particles collected on the filter.  Volatilization losses have also been reported to occur
3      during sample transport and storage (Chow, 1995).  Evaporative losses of particulate nitrates
4      have been investigated in laboratory and field experiments (e.g., Wang and John, 1988), and in
5      theoretical studies (Zhang and McMurry, 1992). It has been known for some time that
6      volatilization losses of SVOC can be significant (e.g., Eatough et al., 1993).
7           The theory describing phase equilibria of SVOC continues to be developed.  Liang et al.
8      (1997), Jang et al. (1997), and Strommen and Kamens (1997) modeled the gas/particle
9      partitioning of SVOC on inorganic, organic, and ambient smog aerosols.
       October 1999                               3-45        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Adsorption of organic vapors onto quartz filters has also been recognized as a source of
 2      positive sampling error. Although original experiments investigating this sampling artifact were
 3      typically carried out utilizing two quartz fiber filters deployed in series, the second quartz filter
 4      can indicate both gaseous VOC adsorbed on both filters (positive) artifact and SVOC evaporated
 5      from particles on the first filter and subsequently adsorbed on the second filter (negative artifact).
 6      Unless the individual compounds are identified, the investigator does not know what to do with
 7      the loading value on the second filter (i.e., to add or subtract from the first filter loading value).
 8      The developing state of the art in which diffusion denuder technology is being applied to SVOC
 9      sampling (e.g., Eatough et al.,  1993; Gundel et  al., 1995a), as well as for sampling of gas and
10      particulate phase organic acids (Lawrence and Koutrakis, 1996a,b), holds promise for improving
11      our understanding of SVOC sampling artifacts.
12           Finally, Eatough et al. (1999) have reported on a batch sampler (the Particle Concentrator -
13      Brigham Young University Organic Sampling System, or PC-BOSS) and a continuous sampler
14      (Real-Time Air Monitoring System or RAMS)  which attempt to correct simultaneously for
15      volatilization losses of both nitrate and SVOC.  These samplers will be discussed in more detail
16      in Section 3.2.3.2.
17
18      3.2.3.1  Particulate Nitrates
19           It is well known that volatilization losses  of particulate nitrates (e.g., Zhang and McMurry
20      [1992]; see also Hering and Cass [1999], and references therein) will occur during sampling on
21      Teflon filters. The impact on the accuracy of atmospheric particulate measurements from these
22      volatilization losses will be even more significant for PM2 5 than they are for PM10.  The FRM for
23      PM25 will suffer loss of nitrates, similar to the losses experienced with other simple filter
24      collection systems. Sampling artifacts due to the loss of particulate nitrates will represent a
25      significant problem in areas that experience high amounts of nitrogen species, like southern
26      California.
27           Hering and Cass (1999) examined the errors in PM25 mass measurements due to
28      volatilization of particulate nitrate by looking at data from two field measurement campaigns
29      conducted in southern California - the Southern California Air Quality Study (SCAQS, Lawson,
30      1990), and the 1986 CalTech study (Solomon et al., 1992).  In both these studies, side-by-side
31      sampling of PM2 5 was conducted. One sampler collected particles directly onto a Teflon filter.

        October 1999                             3-46        DRAFT-DO NOT QUOTE OR CITE

-------
1      The second sampler collected particles onto a nylon filter, following a denuder to remove
2      gaseous nitric acid. In both studies, the denuder consisted of MgO coated glass tubes (Appel
3      et al., 1981). Fine particulate nitrate collected on the Teflon filter was compared to fine
4      particulate nitrate collected on the denuded nylon filter. In both studies, the PM2 5 mass lost due
5      to volatilization of ammonium nitrate represented a significant fraction of the total PM25 mass,
6      and the fraction lost was higher during summer than during fall (17% versus 9% during the
7      SCAQS study, and 21% versus 13% during the CalTech study) (Figure 3-11). Hering and Cass
8      (1999) found that nitrate obtained from Teflon filter samples was on average 28% lower than that
9      obtained by denuded nylon filters.
           80%
        o
          60% -
          40% -
          20% -
           0%
                                SCAQS Data Set
                               o Summer Measurements
                               • Fall Measurements
                                                        80%
 50     100    150    200
PM25 Gravimetric Mass (ng/m3)
                                              250
                                                      .i 60% -
                                                        40% -
                                                        20% -
                                                                        100
                                                            Caltech Data Set
                                                            o April - September
                                                            • October - March
                                                                               150
                                                                                      200
                                                                                            250
                                                                    PM25 Gravimetric Mass (ng/m
       Figure 3-11.  Amount of ammonium nitrate volatilized from Teflon filters, expressed as a
                     percentage of the measured PM25 mass, for the SCAQS and CalTech studies,
                     for spring and fall sampling periods.
       Source: Herring and Cass (1999).
1           Hering and Cass (1999) also analyzed these data by extending the evaporative model
2      developed by Zhang and McMurry (1987).  The extended model utilized by Hering and Cass
3      (1999) takes into account dissociation of collected particulate ammonium nitrate on Teflon filters
4      into nitric acid and ammonia, via three mechanisms: scrubbing of nitric acid and ammonia in the
       October 1999
                             3-47
DRAFT-DO NOT QUOTE OR CITE

-------
 1      sampler inlet (John et al. [1988] showed that clean PM10 inlet surfaces serve as an effective
 2      denuder for nitric acid), heating of the filter substrate above ambient temperature by sampling,
 3      and pressure drop across the Teflon filter. For the sampling systems modeled, the flow-induced
 4      pressure drop was measured to be less than 0.02 atmospheres, and the corresponding change in
 5      vapor pressure was 2%, so losses driven by pressure drop were not considered to be significant in
 6      this work.  Losses from Teflon filters were found to be higher during the summer compared to
 7      the winter, and higher during the day compared to night, and were reasonably consistent with
 8      modeled predictions.
 9           Finally, during the SCAQS study, particulate samples were also collected using a Berner
10      impactor and greased Tedlar substrates, in size ranges from 0.05 to 10 mm in aerodynamic
11      diameter. The Berner impactor PM2 5 nitrate values were much closer to those from the denuded
12      nylon filter than those from the Teflon filter, with the impactor nitrate being approximately 2%
13      lower than the nylon filter nitrate for the fall measurements, and approximately 7% lower during
14      the summer measurements. When the impactor collection was compared to the Teflon filter
15      collection for a nonvolatile species (sulfate), the results were in agreement.
16           It should be noted that during these intercomparison studies, filters or collection surfaces
17      were removed immediately after sampling and placed into vials containing a basic extraction
18      solution. Therefore, losses that might occur during handling, storage, and equilibration of filters
19      or impaction surfaces were avoided.  The  loss of nitrate observed from Teflon filters and
20      impaction surfaces in this study, therefore, is a lower limit compared to losses that might occur
21      during the normal processes involved in equilibration and weighing of filters and impaction
22      surfaces.
23           In atmospheres with high sulfate and low  ammonia, the PM tends to be acidic (NH4HSO4
24      or H2SO4) and nitric acid remains as a gas. In atmospheres with lower sulfate and higher
25      ammonia, there may be sufficient ammonia  to fully neutralize the H2SO4 and also react with
26      FfNO3 to form NH4NO3 particles.  In the U.S., therefore, loss of nitrate will be a bigger problem
27      in the western U.S.  than in the eastern U.S.  However, as SO2 emissions are reduced in the
28      eastern U.S., nitrate may become a larger  fraction of the suspended PM.
29
        October 1999                              3-48         DRAFT-DO NOT QUOTE OR CITE

-------
 1      3.2.3.2  Semivolatile Organic Compounds
 2           Although there is less information on losses of semivolatile organic compounds (SVOC), it
 3      is known that these species can similarly be lost from Teflon filters due to volatilization, and can
 4      also cause the PM2 5 mass to be significantly underestimated. Like particulate nitrates, the FRM
 5      for PM2 5 will suffer loss of SVOC, similar to the losses experienced with other simple filter
 6      collection systems. It has been shown that attempting to correct for the losses of SVOC during
 7      sampling by deploying a second quartz filter directly behind either a quartz or Teflon filter can
 8      significantly underestimate the volatilization losses (e.g., Eatough et al., 1993). Using their
 9      multichannel diffusion denuder sampling system (BOSS), Eatough et al. (1995) reported that, for
10      samples collected at the South Coast Air Quality Management District sampling site at Azusa,
11      CA, changes in the phase distribution of SVOC could result in a loss on average of 35% of the
12      particulate organic material; the fraction of the total fine particulate matter that this loss
13      represents was not given. At present, there are limited data available specifically on the fraction
14      of PM25 mass lost during sampling onto Teflon filters due to volatilization of organic species,
15      and even less on the regional differences in the effects of volatilization losses of SVOC.  Cui
16      et al. (1998) found that losses of SVOC from particles in the Los Angeles Basin during the
17      summer were greater during the day than at night. Cui et al. (1998) determined that on average,
18      42 and 62% of the particulate organic material was semivolatile organic compounds lost from
19      particles during sampling for daytime and nighttime samples, respectively.
20           In addition to their contribution to suspended PM mass, SVOC are also of interest because
21      of their possible health effects. SVOC include products of incomplete combustion such as, for
22      example, polycyclic aromatic hydrocarbons (PAHs) and polycyclic organic matter, which has
23      been identified as a hazardous air pollutant. PAHs have also been suggested as alternative
24      particulate tracers for automobile emissions, since the phase-out of organo-lead additives to
25      gasoline means that lead is no longer a good tracer for automobiles (Venkataraman et al., 1994).
26      PAHs are also emitted during biomass burning, including burning of cereal crop residues and
27      wood fuels  (Jenkins et al., 1996; Roberts and Corkill, 1998). The semivolatile PAHs are also of
28      interest in ambient PM studies, because of their potential for causing both positive and negative
29      sampling artifacts if not properly accounted for. Several investigators have observed that
30      collection of particles on a filter can result in underestimation of particulate organic compounds
31      due to losses of semivolatile organic material during sample collection (negative sampling

        October 1999                              3-49        DRAFT-DO NOT QUOTE OR CITE

-------
 1      artifact) (Eatough et al, 1993; Tang et al, 1994; Eatough et al, 1995; Gundel et al, 1995a; Finn
 2      et al., 1999). Positive sampling artifacts can also occur due to the adsorption of gases onto the
 3      filter materials (e.g., Gundel et al., 1995a). There appears to be a larger positive artifact due to
 4      adsorption of organic vapor onto quartz fiber filters than to Teflon filters (Turpin et al., 1994;
 5      Chow et al., 1994, 1996; Eatough et al., 1996; Finn et al., 1999).
 6           Phase distribution of semivolatile organic species has been the subject of several studies
 7      that have employed denuder technology (see Gundel et al., 1995a; Gundel and Lane,  1999) to
 8      directly determine the phase distributions while avoiding some of the positive and negative
 9      sampling artifacts associated with using backup quartz filters. For measuring particulate phase
10      organic compounds, the denuder-based sampling system is a definite improvement over the
11      filter/adsorbent collection method (Turpin et al., 1993). Some researchers have reported that
12      denuder coatings themselves can introduce contamination (Mukerjee et al., 1997), or the
13      adsorbed species may be difficult to remove from the coating (Eatough et al., 1993).
14      In calculating the overall phase distributions of SVOC PAH from a denuder system, the
15      collection efficiency for each compound is needed.  The efficiency of silicone-grease-coated
16      denuders for the collection of polynuclear aromatic hydrocarbons was examined by Coutant et al.
17      (1992), who examined the effects of uncertainties in the diffusion coefficients, and in the
18      collisional reaction efficiencies, on the overall phase distributions of SVOC PAH calculated
19      using denuder technology. In their study, they used a single stage, silicone-grease-coated
20      aluminum annular denuder, with a filter holder mounted ahead of the denuder, and an XAD trap
21      deployed downstream of the denuder.  In a series of laboratory experiments, they spiked the filter
22      with a mixture of perdeuterated PAH, then swept the system with ultra-high purity air for several
23      hours, and then analyzed the filter and the XAD.  They found that the effects of these
24      uncertainties, introduced by using a single compound as a surrogate PAH (in their case,
25      naphthalene) for validation of the denuder collection efficiency, are less significant than normal
26      variations due to  sampling and analytical effects. Results on field studies using their sampling
27      system  have not been published.
28           Losses of the SVOC fraction of particulate organic matter occurring during sampling were
29      investigated by Eatough et al. (1995), who found that on average losses of 35% of the POM in
30      samples collected at a site in southern California resulted. In this study, the Brigham Young
31      University Organic Sampling System (BOSS) (Eatough et al., 1993) was used for determining

        October 1999                             3-50       DRAFT-DO NOT QUOTE OR CITE

-------
 1      POM composition, and a high-volume version (BIG BOSS) (flowrate 200 Lpm) was utilized for
 2      determining the particulate size distribution and the chemical composition of SVOC in fine
 3      particles.  The BOSS, a multi-channel diffusion denuder sampling system, consists of two
 4      separate samplers (each operating at 35 Lpm). The first sampler consists of a multi-parallel plate
 5      diffusion denuder with charcoal-impregnated filter papers as the collection surfaces, followed by
 6      a two-stage quartz filter pack, followed by a two-stage charcoal-impregnated filter pack.  The
 7      second sampler operating in parallel with the first consists of a two-stage quartz filter pack,
 8      followed by the parallel plate denuder, followed by the two-stage charcoal-impregnated filter
 9      pack.  The filter samples collected by the BOSS sampler were analyzed by temperature-
10      programmed volatilization analysis. Eatough et al. (1995) also operated a two-stage quartz filter
11      pack alongside the BOSS sampler.  The BIG BOSS system (Tang et al., 1994) consists of
12      4 systems (each with a flowrate of 200 Lpm).  Particle size cuts of 2.5, 0.8, and 0.4 mm are
13      achieved by virtual impaction, and the sample subsequently flows through a denuder, then is
14      split, with the major flow (150 Lpm) flowing through a quartz filter followed by an XAD-II bed.
15      The minor flow is sampled through  a quartz filter backed by a charcoal-impregnated filter paper.
16      The samples derived from the major flow (quartz filters and XAD-II traps) were extracted with
17      organic solvents and analyzed by gas chromatography and GC-mass spectroscopy. The organic
18      material lost from the particles was  found to represent all classes  of organic compounds.
19          Eatough et al. (1996) operated the BOSS sampler for a year at the IMPROVE site at
20      Canyonlands National Park, Utah, alongside the IMPROVE monitor and alongside a separate
21      sampler consisting of a two-stage quartz filter pack. They found that concentrations of
22      particulate carbon determined from  the quartz filter pack  sampling system were low on average
23      by 39% due to volatilization losses of SVOC from the quartz filters.
24          Six-hour daytime samples and 9-hour nighttime samples were collected with the BIG
25      BOSS sampler in the LA Basin in September of 1994 (Cui et al.,  1998). During this study, Cui
26      et al. (1998) determined that an average of 42  and 62% of organic PM was SVOC lost from the
27      particles during daytime and nighttime sampling, respectively. The negative sampling artifact
28      associated with SVOC losses was an order of magnitude larger than the positive quartz filter
29      artifact that results due to adsorption of gas phase organic material.
30          The BIG BOSS sampler developed by Eatough and colleagues (Eatough et al., 1999) has
31      been used to determine the total carbonaceous material collected by the quartz filters and the

        October 1999                              3-51        DRAFT-DO NOT QUOTE OR CITE

-------
 1      charcoal impregnated filter (GIF) by temperature programmed volatilization (TPV) and for GC
 2      analyses of extractable material from the XAD-II sorbent beds.  In TPV, a portion of the sample
 3      is continuously heated from ambient temperature to 800 C in a mixed nitrogen/oxygen air stream,
 4      and the volatilized gases are converted to CO2 in a catalytic furnace and detected as CO2.  The
 5      TPV analysis gives a measure of the EC and OC. To thermally desorb the GIF, a pure nitrogen
 6      stream is used as the carrier gas, to avoid oxidative decomposition of the GIF substrate, as well as
 7      the collected organic material.  To determine organic compounds collected on the XAD-II
 8      substrate, the sorbent is solvent extracted with dichloromethane and analyzed via GC and
 9      GC-MS. In this way, aromatic compounds, paraffins, organic acids, and esters have been
10      detected (quantities were not reported) in samples collected at Azusa (Los Angeles Basin, CA)
11      and in Philadelphia (Eatough et al.,  1995).
12          Ding et al. (1998a) developed a method for the determination of total n-nitroso compounds
13      in air samples, and used the method to examine organic compounds formed from NOX chemistry
14      in Provo, UT (Ding et al., 1998b). In their method, n-nitroso compounds are selectively
15      decomposed to yield nitric oxide, which is then detected using chemiluminescence.  From the
16      samples from Provo, UT, they found that the majority of the n-nitroso and nitrite organic
17      compounds that were present in fine particulate matter were semivolatile organic compounds that
18      could be evaporated from the particles during sampling. They found particulate n-nitroso
19      compound concentrations ranging between <1 and 3 nmoles/m3, and gas-phase n-nitroso
20      compound concentrations in the same range.  Particulate organic nitrite concentrations were
21      found in the range of <1 to ~5 nmoles/m3, and gas-phase concentrations as high as 10 nmoles/m3
22      were found.
23          Turpin et al. (1993) developed a sampling system based upon a diffusion separator, which
24      corrects for the loss of semivolatile organic compounds during sampling by removal of most of
25      the gas phase material from the particles in a diffusion separator sampling system. Unlike the
26      previously mentioned systems, wherein the particulate phase is measured directly, in the system
27      of Turpin et al., the gas-phase is measured directly. In their laminar flow system, ambient,
28      particle-laden air enters the sampler as an annular flow. Clean, particle-free air is pushed through
29      the core inlet of the separator.  The clean air and ambient aerosol join downstream of the core
30      inlet section, and flow parallel to each other through the diffusion zone. Because of the much
31      higher diffusivities for gases compared to particles, the SVOC in the ambient air diffuse to the

        October 1999                              3-52        DRAFT-DO NOT QUOTE OR CITE

-------
 1      clean, core flow. The aerosol exits the separator in the annular flow, and the core flow exiting
 2      the separator now contains a known fraction of the ambient SVOC. Downstream of the diffusion
 3      separator, the core exit flow goes into a PUF plug, where the SVOC is collected. The adsorbed
 4      gas phase on the PUF plug is extracted with supercritical fluid CO2, and analyzed by gas
 5      chromatography/mass-selective detection (GC/MSD).  The gas-phase SVOC is thus determined.
 6      Ultimately, to determine particulate phase SVOC concentrations, the total compound
 7      concentration will also be measured, and the particulate phase obtained by difference. The
 8      system has been evaluated for the collection of PAH.
 9           Gundel et al. (1995a) recently developed a technique for the direct determination of phase
10      distributions of semivolatile polcyclic aromatic hydrocarbons, rather than phase distribution
11      determination using difference method, using  annular denuder technology. That technique has
12      been applied to characterizing PAHs in environmental tobacco smoke (Gundel et al., 1995b), and
13      to ambient air sampling (Lane and Gundel,  1996). The method developed by Gundel et al.
14      (1995a), called the integrated organic vapor/particle sampler (IOVPS), uses a cyclone inlet with a
15      D50 cutpoint of 2.5 mm.  The airstream then goes through two or three sandblasted glass annular
16      denuders that are coated with ground adsorbent resin material (XAD-4 was initially examined)
17      that traps vapor-phase organics.  The airstream subsequently passes through a filter, followed by
18      a backup denuder.
19           The IOVPS, which operates at 10 Lpm, was tested for sampling semivolatile PAH in
20      laboratory indoor air, and environmental tobacco smoke (ETS). After exposure, the denuders,
21      filters, and sorbent traps were extracted with cyclohexane (Gundel et al.,  1995a) and analyzed for
22      PAHs from naphthalene to chrysene using dual-fluorescence detection (Mahanama et al., 1994).
23      Recoveries from both denuders and filters were approximately 70% for 30 samples. Detection
24      limits (lower limits of detection, defined as  3 times the standard deviation of the blanks) for gas
25      phase SVOC PAHs ranged from 0.06 ng for anthracene to 19 ng for 2-methylnaphthalene. The
26      95% confidence interval for reproduction of an internal standard concentration was 6.5% of the
27      mean value. Relative precision as determined either from a propagation of errors analysis, or
28      from the 95% confidence interval from replicate analyses of standard reference material SRM
29      1649 (urban dust/organics) was 12% on  average, and ranged from 8% for naphthalene to 22% for
30      fluorene. Sources of error included sampling  flow rate, internal standard concentration, and
31      co-eluting peaks.

        October 1999                              3-53        DRAFT-DO  NOT QUOTE OR CITE

-------
 1           Gas-phase PAH concentrations in indoor laboratory air ranged from 0.4 (0.1 - 0.6) ng/m3
 2      for anthracene to 338 (162 - 545) ng/m3 for naphthalene.  In ETS, gas-phase PAHs ranged from
 3      0.4 (0.2 -1.1) ng/m3 for benz(a)anthracene to 1099 (784 - 1690) ng/m3 for naphthalene. In ETS,
 4      the fraction of PAHs in the particle phase was less than 0.11 for most, except for fluoranthene
 5      (38% in particle phase); pyrene (18%); benz(a)anthracene (99%); and chrysene (97%).
 6           Particulate phase PAHs have been measured by Gundel and Lane (1999) using sorbent-
 7      coated diffusion denuders (IOVPS). A series of 6 IOVPS were used sequentially over a 24-hr
 8      period to measure PAH phase distributions near a roadway in Berkeley, CA. Particulate fractions
 9      of PAH varied from 7 to 91% and were higher during colder periods. Gundel and Lane also
10      reported that roughly two third of particulate PAH fluoranthene, pyrene, benz [a] anthracene and
11      chrysene were found on the post-filter denuders, so that it is likely that considerable desorption
12      from the collected particles took place. The IOVPS has also been used by Kamens and
13      associates to study semivolatile PAH and nitro-PAH, and dioxins, primarily in smog chamber
14      studies. Fan et al. (1996a) found that particulate phase nitro-PAH decayed quickly in sunlight,
15      and that degradation by ozone was important at night (Fan et al., 1996b).  In a study of the
16      atmospheric behavior of dioxins, Penisse and Kamens (1996) found that, under high particle
17      loadings (TSP concentrations ranging from 1 to 7 mg/m"3), the tetra and pentachlorinated dioxins
18      and furans partitioned into the gas phase.
19           Solid adsorbent-based denuder systems have been investigated by other researchers, as
20      well. Bertoni et al.  (1984) described the development of a charcoal-based denuder system, for
21      the collection of organic vapors. Risse et al. (1996) developed a diffusion denuder system to
22      sample aromatic hydrocarbons. In their system, denuder tubes with charcoal coating and
23      charcoal paper precede a filter pack for particulate collection, and an adsorption tube to  capture
24      particle blow-off from the filter sample. Breakthrough curves for benzene, toluene, ortho-xylene,
25      and meta-xylene were developed for denuder tubes of length 60, 90, and 120 cm. The effects of
26      relative humidity on the adsorption capacities of the denuder system were examined, and it was
27      found that the capacity of the charcoal was not significantly impacted by increases in relative
28      humidity. The feasibility of outdoor air sampling with the system was demonstrated. Risse et al.
29      (1996) developed a diffusion denuder system for sampling aromatic hydrocarbons, in which
30      denuder tubes were coated with charcoal.
        October 1999                             3-54        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Krieger and Kites (1992) designed a diffusion denuder system that uses capillary gas
 2      chromatographic columns as the tubes of SVOC collection.  The denuder was followed by a filter
 3      to collect particles, which in turn was followed by a polyurethane foam (PUF) plug to collect
 4      organic material volatilizing off the filter.  Denuder samples were analyzed by liquid solvent
 5      extraction (CH2C12) followed by GC-MS analysis.  The PUF plugs and filters were extracted with
 6      supercritical fluid extraction using supercritical N2O.  Using this system, an indoor air sample
 7      was found to contain primarily chlorinated biphenyls, ranging from trichlorobiphenyls (vapor
 8      pressures 10~3 - 10"4 Torr at 25 °C) to octachlorobiphenyls (10~6 - 10~7 Torr), which demonstrated
 9      that the sampler collects compounds with a wide range of volatility. They also found that on-line
10      desorption is successful in maintaining good chromatographic peak shape and resolution. The
11      entire method, from sample collection to the end of the chromatographic separation, took
12      2 hours.
13           Organic acids in both the vapor and particulate phases may be important contributors to
14      ambient acidity, as well as representing an important fraction of organic particulate matter.
15      Lawrence and Koutrakis (1996a,b) used a modified Harvard/EPA annular denuder system
16      (HEADS) to sample both gas and particulate phase organic acids in Philadelphia, PA in the
17      summer of 1992. The HEADS sampler inlet had a 2.1 mm cutpoint impactor (at 10 Lpm),
18      followed by two denuder tubes, and finally a filter pack with a Teflon filter. The first denuder
19      tube was coated with KOH to trap gas phase organic acids. The second denuder tube was coated
20      with citric acid to remove ammonia and thus to  avoid neutralizing particle phase acids collected
21      on the filter. The KOH-coated denuder tube was reported to collect gas phase formic and acetic
22      acids at better than 98.5% efficiency, and with precisions of 5% or better (Lawrence and
23      Koutrakis, 1994). It was noted that for future field measurements of particulate organic acids,
24      a Na2CO3-coated filter should be deployed downstream of the Teflon filter to trap organic acids
25      that may evaporate from the Teflon filter during sampling.
26
27      3.2.3.3  Use of Denuder Systems To Measure Semivolatile Compounds
28           Much progress has been made to date in the design of diffusion denuder systems for the
29      measurement and characterization of both the particulate and gaseous phases of semivolatile
30      compounds. Some of the recent research has focused upon reduction in the size of the denuder,
31      optimization of the residence time in the denuder, understanding the effect of diffusion denuders

        October 1999                              3-55        DRAFT-DO NOT QUOTE OR CITE

-------
 1      on the positive quartz filter artifact, identifying changes in chemical composition that occur
 2      during sampling, determining the effects of changes in temperature and relative humidity, and
 3      identifying possible loses by absorption in impactor coatings.
 4
 5      Reducing the Size ofDenuders
 6           Diffusion denuder systems have been used in a number of studies aimed at quantifying both
 7      gaseous and particulate pollution.  The typical denuder configuration is an annular diffusion
 8      denuder tube of significant length (e.g., 26.5 cm for 10 Lpm, Koutrakis et al., 1988).  A more
 9      compact design based upon a honeycomb configuration was shown to significantly increase the
10      capacity (Koutrakis et al., 1993).  However, in intercomparisons with an annular denuder/filter
11      pack system (Koutrakis et al., 1988), significant losses of ammonia and nitric acid were observed
12      for the honeycomb configuration, and attributed to the large inlet surface area and long sample
13      residence time of the honeycomb design, relative to the annular denuder system.  Sioutas et al.
14      (1996a) subsequently designed a modified glass honeycomb denuder/filter pack sampler (HDS)
15      with an inlet that minimizes vapor losses on the inlet surfaces.  The modified HDS has reduced
16      inlet surfaces and decreased residence time of sampled gases (NH3 and HNO3) compared to its
17      predecessor (Sioutas et al., 1994).  Sioutas et al. (1996b) tested various inlet materials (glass,
18      PFA, and PTFE) in laboratory tests and found that a PTFE Teflon coated inlet minimized loss of
19      sampled gases (1 - 8% losses of HNO3 observed, and -4 - 2 % losses of NH3 observed).  The
20      highest inlet losses were observed for HNO3 lost to PFA surfaces (14 - 25%).  The modified
21      HDS was tested in laboratory and field tests and found to agree within  10% with the annular
22      denuder system.
23
24      Residence Time in the Denuder
25           The efficiency of a diffusion denuder sampler for the removal of gas phase material can be
26      improved by increasing the residence time of the sampled aerosol in the denuder. However, the
27      residence time can only be increased within certain limits. Since the diffusion denuder reduces
28      the concentration of gas phase semivolatile organic material, semivolatile organic matter present
29      in the particles passing through the denuder will be in a thermodynamically unstable environment
30      and  will tend to outgas SVOC during passage through the denuder. The residence time of the
31      aerosol in the denuder, therefore, should be short enough to prevent significant loss of particulate

        October  1999                             3-56        DRAFT-DO NOT QUOTE OR CITE

-------
 1      phase SVOC to the denuder. Various studies have suggested that the residence time in the
 2      denuder should be less than about 2 seconds (Gundel and Lane, 1999; Kamens and Coe, 1997;
 3      Kamens et al., 1995). The residence times in the various denuder designs described by Gundel
 4      and Lane (1999) are from 1.5 to 0.2 seconds.
 5
 6      Effect of Diffusion Denuders on the Positive Quartz Filter Artifact
 7           Denuder systems may be useful to reduce artifact reactions between gases and either the
 8      filter substrate or collected particles (Durham et al., 1978).  Annular denuder systems, coupled
 9      with filter packs employing both Teflon and nylon filters, have been used to study ammonium
10      nitrate, nitric acid, and ammonia in the vapor and aerosol phases (e.g., Benner et al., 1991).
11      To account for the volatilization losses of semi-volatile organic compounds, Turpin et al. (1994)
12      recommended that a quartz filter be placed behind a Teflon filter in a parallel sampler. Addition
13      of a vapor trap (e.g., polyurethane foam plug) downstream of the filter was  also suggested as a
14      method to collect semi-volatile organic compounds. However, it was noted that these methods
15      (addition of some type of trap behind the Teflon filter) collected both vapor phase organics as
16      well as "blow-off from the Teflon filter i.e., material vaporized from particles collected on
17      Teflon  filter (Van Vaeck et al., 1984).  At the time of the previous AQC PM 96 (U.S.
18      Environmental Protection Agency, 1996), some investigators were beginning to examine the
19      phase partitioning of SVOC and accounting for the vapor phase SVOC, the particulate organic
20      matter captured on the filter, and any SVOC that was subsequently lost from the filter by
21      volatilization. Eatough et al. (1993) measured both adsorption and volatilization artifacts by
22      using a sampling train that consisted of a diffusion denuder followed by a filter pack followed by
23      a sorbent bed. The effects of face velocity and pressure drop across the filter were noted, and
24      sampling systems optimizing these parameters to minimize artifacts were discussed.  Gundel
25      et al. (1995a,b) have also used sorbent-coated diffusion denuders to examine the phase
26      distributions of semi-volatile organic compounds.
27           The adsorption of organic compounds by a second quartz filter has been shown to be
28      reduced, but not eliminated, in samples collected in the Los Angeles Basin, if a multi-channel
29      diffusion denuder with quartz filter material as the denuder collection surface preceded the quartz
30      filters (Fitz, 1990).  This artifact can be further reduced by the use of activated charcoal at the
31      denuder surface and use of a particle concentrator to reduce the amount of gas phase organic

        October 1999                              3-57        DRAFT-DO NOT QUOTE OR CITE

-------
 1      compounds relative to condensed phase organic compounds (Cui et al., 1998, 1997; Eatough,
 2      1999). Recent experiments (Cotham and Bidleman, 1992; Cui et al., 1998; Eatough et al., 1995,
 3      1996) have shown that the quartz filter artifact can result both from the collection of gas phase
 4      organic compounds and from the collection of semivolatile organic compounds lost from
 5      particles during sampling.  Thus, results available to date suggest that both a "positive" and a
 6      "negative" artifact can be present in the determination of particulate phase organic compounds,
 7      using two tandem quartz filters.
 8
 9      Changes in Chemical Composition During Sampling
10           The use of sampling systems designed to correctly identify the atmospheric gas and
11      particulate phase distributions of collected organic material has been outlined above.
12      An additional sampling artifact which has received little consideration in the collection of
13      atmospheric sampling is the potential alteration of organic compounds as a result of the sampling
14      process.  These alterations appear to result from the movement of ambient air containing
15      oxidants and other reactive compounds past the collected particles. The addition of NO2
16      (
-------
 1      1990).  The BOSS denuder should be effective in eliminating most of the chemical
 2      transformation artifacts since reactive gases are removed by the charcoal denuder which proceeds
 3      the particle collection filter. The implication from these studies is that, if the organic material
 4      lost from particles is important to respiratory health problems, then the traditional PM
 5      measurement techniques based upon integrated filter sampling followed by equilibration and
 6      gravimetric analysis may significantly underestimate the concentrations of POM.  Moreover, it
 7      appears that the negative sampling artifact associated with volatilization losses from quartz filters
 8      is usually much greater than the positive sampling artifact that would be associated with
 9      adsorption of SVOC vapors onto quartz fiber filters. Because of the redistribution of SVOC
10      between phases and the potential for chemical changes that occur during sampling, it may be
11      necessary to use diffusion denuders or diffusion separation systems like those described in this
12      section to adequately characterize particulate organic matter in ambient air.
13
14      Temperature and Relative Humidity Effects
15           The problems of sampling artifacts associated with SVOC adsorption and evaporation are
16      compounded by temperature and relative humidity effects (Pankow and Bidleman, 1991; Pankow
17      et al., 1993; Falconer et al., 1995; Goss and Eisenreich, 1997).  Effects of temperature on the
18      partitioning of PAH was examined by Yamasaki et al. (1982), who found that the partition
19      coefficient (PAHvapo/PAHpart) was inversely related to temperature and could be described using
20      the Langmuir adsorption concept. The dissociation of ammonium nitrate aerosol is also a
21      function of temperature. Bunz et al. (1996) examined the dissociation and subsequent
22      redistribution of NH4NO3 within a bimodal distribution, using a 9-stage low pressure Berner
23      impactor followed by analysis by ion chromatography and found a strong temperature
24      dependency on the redistribution. Bunz et al. found that at lower temperatures (below 10 °C),
25      there was little change in the aerosol size distribution. At temperatures between 25 and 45 °C,
26      however, the lifetime of NH4NO3 particles decreases by more than a factor of 10, and size
27      redistribution, as measured by average ending particle diameter, increased more for higher
28      temperatures than for lower temperatures.
29           The effects of relative humidity on the sorption of SVOC on particles are not well
30      understood. In a series of laboratory experiments, Goss and Eisenreich (1997) examined the
31      sorption of both polar (hydrocarbons and chlorinated hydrocarbons)  and non-polar (ethyl ether

        October 1999                              3-59       DRAFT-DO NOT QUOTE OR CITE

-------
 1      and acetone) volatile organic compounds (VOC) onto combustion soot particles as a function of
 2      temperature and relative humidity. The soot particles used in their experiments were collected
 3      from oil furnaces and contained 60% (w/w) iron sulfate (water-soluble fraction) and 9% (w/w)
 4      elemental and organic carbon.  The carbon and sulfate contents of their particulate matter are
 5      comparable to the chemical composition of ambient fine particles.  They found that, for all
 6      compounds, the sorption of VOC onto soot particles decreased with increasing relative humidity
 7      over the range of 10 to 95%. They also observed hysteresis in the relative humidity dependency,
 8      with sorption coefficients at a given relative humidity higher when the RH is being increased
 9      than when the RH is being decreased.  The sorption coefficients were fit with an exponential
10      function to the RH so that the slope of the regression line would provide a measure of the
11      influence of relative humidity. Based upon the magnitude of the slope, they concluded that the
12      RH-dependency of sorption was stronger for water-soluble organic compounds. In another study
13      by Jang and Kamens (1998), humidity effects on gas-particle partitioning of SVOC  were
14      examined using outdoor environmental chambers and the experimentally determined partitioning
15      coefficients were compared to theoretical values. They examined the partitioning of SVOC onto
16      wood soot, diesel soot, and secondary aerosols and concluded that "the humidity effect on
17      partitioning was most significant for hydrophobic compounds adsorbing onto polar aerosols."
18      Although these two studies seem to be contradictory, upon closer examination, it is  difficult to
19      compare the two studies for several reasons. The experiments conducted by Jang and Kamens
20      (1998) were conducted in outdoor chambers at ambient temperatures and humidities. Their
21      model was for absorptive partitioning of SVOC on liquid-like atmospheric particulate matter.
22      In contrast, the results of Goss and Eisenreich (1997) were obtained from a gas chromatographic
23      system operated at 70 °C higher than ambient conditions.  The model of Goss and Eisenreich
24      (1997) was for adsorptive partitioning  of VOC on solid-like atmospheric particulate matter.
25      In the study of Jang and  Kamens (1998), calculated theoretical values  for water activity
26      coefficients for diesel soot were based upon an inorganic salt content of 1 to 2%, while the
27      combustion particles studied by Goss and Eisenreich (1997) contained 60% water-soluble,
28      inorganic salt content. Jang and Kamens (1998) obtained their diesel soot from their outdoor
29      chamber, extracted it with organic solvent (mixtures of hexane and methylene chloride), and
30      measured the organic fraction. The resulting salt content of 2% of the particulate matter studied


        October 1999                              3-60        DRAFT-DO NOT QUOTE OR CITE

-------
 1      in Jang and Kamens (1998) is enough to affect water uptake but presumably not to affect the
 2      sorption partitioning of organics.
 3
 4      Adsorption by Impactor Coatings
 5           There are other sources of error inherent in some of the currently acceptable practices that
 6      could potentially affect particulate mass concentration measurements and that will surely become
 7      even more important as more emphasis in particulate sampling is placed upon chemical
 8      speciation.  Allen et al. (1999) reported that the practice of greasing impaction substrates may
 9      introduce an artifact from the absorption of semivolatile species from the gas phase by the grease,
10      which could artificially increase the amount of PAHs and other organic compounds attributed to
11      the aerosol.  Allen et al. (1999) offer several criteria to ensure that this absorption artifact is
12      negligible, including selecting impaction oils in which analytes of interest are negligibly soluble;
13      and ensuring that species do not have time to equilibrate between the vapor and oil phases
14      (criterion is met for nonvolatile species).  They recommend using oiled impaction substrates only
15      if the absorption artifact is negligible as determined from these criteria.  Similarly, the use of
16      coating solutions, especially those that contain organic liquids, in denuder systems may lead to
17      adsorption of volatilized coating solution components on the downstream filters (Finn et al.,
18      1999).
19
20      3.2.3.4  Particle-Bound Water
21           It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
22      compounds. However, for many measurement of suspended particle mass, it is desirable to
23      remove the particle-bound water before determining the mass. In other situations it may be
24      important to know how much of the suspended particle's mass or volume is due to particle-
25      bound water. The water content of PM is significant and highly variable. Moreover, there is
26      significant hysteresis in the water adsorption-desorption pathways (Seinfeld and Pandis, 1998),
27      further complicating the mass measurement. Figures 3-8 and 3-9 show the change in diameter of
28      sulfate particles as a function of relative humidity.  Figure 3-8 shows the difference between
29      deliquescence and crystallization points.
30           Pilinis et al. (1989) calculated the water content of atmospheric particulate matter above
31      and below the deliquescent point. They predicted that aerosol water content is strongly

        October 1999                              3-61        DRAFT-DO NOT QUOTE OR CITE

-------
 1      dependent upon composition, and concluded from their calculations that liquid water could
 2      represent a significant mass fraction of aerosol concentration at relative humidities above 60%.
 3      Since then, a few researchers have attempted to measure the water content of atmospheric
 4      aerosol.  Most techniques have focused on tracking the particle mass as the relative humidity is
 5      changed, and are  still in the development phase. There have been only a few demonstrations
 6      using actual ambient aerosol, to date. Of interest, in particular, is the development of the
 7      Tandem Differential Mobility Analyzer (TDM A) and its applications in investigations of the
 8      effects of relative humidity on particle growth.
 9           Lee et al. (1997) examined the influence of relative humidity on the size of atmospheric
10      aerosol using a TDMA coupled with a scanning mobility particle sizer (SMPS). They reported
11      that the use of the TDMA/SMPS system allowed for the abrupt size changes of aerosols at the
12      deliquescence point to be precisely observed. They also reported that, at relative humidities
13      between 81 and 89%, the water content of ammonium sulfate aerosols (by mass) was 47  to 66%.
14           Andrews and Larson (1993) investigated the interactions of single aerosol particles coated
15      with an organic film with a humid environment. Using an electrodynamic balance, they
16      conducted laboratory experiments in which sodium chloride and carbon black particles were
17      coated with individual organic surfactants, intended to simulate the surface-active, organic films
18      that many atmospheric aerosol particles may exhibit,  and their water sorption curves examined.
19      Their results showed that when ordinarily hydrophobic carbon black particles were coated with
20      an organic surfactant, they sorbed significant amounts of water (20 - 40% of the dry mass of the
21      particle).
22           Liang and Chan (1997) developed a fast technique using the electrodynamic balance to
23      measure the water activity of atmospheric aerosols. In their technique, the mass of a levitated
24      particle is determined as the  particle either evaporates or grows in response to a step change in
25      the relative humidity. Their  technique was demonstrated using laboratory experiments with
26      NaCl, (NH4)2SO4, NaNO3, and (NH4)2SO4/NH4NO3 solutions. They conclude that one of the
27      advantages of their fast method is the ability to measure the water activity of aerosols containing
28      volatile solutes such as ammonium chloride and some organics.
29           Mclnnes et  al. (1996) measured aerosol mass concentration, ionic composition, and
30      associated water mass of marine aerosol over the remote Pacific Ocean. The mass of particle-
31      bound water was  determined by taking the difference between the mass obtained at 48%  RH and

        October 1999                             3-62        DRAFT-DO NOT QUOTE OR CITE

-------
 1      at 19% RH, assuming the aerosol particles were dry at 19% RH. Based upon a comparison of the
 2      remote Pacific aerosol to aerosol collected at a site at the marine/continental interface of the
 3      Washington coast, the amount of water associated with the aerosol was observed to be a function
 4      of the ammonium to sulfate ratio.  They found that the amount of water associated with the
 5      submicrometer aerosol comprised 29% of the total aerosol mass collected at 47% RH, and 9% of
 6      the total mass at 35% RH.
 7           Ohta et al.  (1998) characterized the chemical composition of atmospheric fine particles
 8      (D50 = 2 mm) in Sapporo, Japan, and as part of their measurements, determined the water
 9      content using the Karl Fischer method (Meyer and Boyd, 1959). After exposing a Teflon filter, a
10      portion of the filter was equilibrated at 30% RH for 24 hours. Then the filter piece was placed in
11      a water evaporator heated at 150 °C, vaporizing the particle-bound water.  The vapor evolved
12      was analyzed for water in an aqua-counter where it was titrated coulometrically in Karl Fischer
13      reagent solution  (containing iodine, sulfur, and methanol). The accuracy of the aqua-counter is
14      ±1 mg. Using this technique,  they determined that the water content of the particles ranged from
15      0.4 to 3.2% of the total particulate mass (at RH < 30%).  This represents a smaller portion of
16      water compared  to their previous reported values (Ohta and Okita,  1990) which were determined
17      by calculation at RH of 50%.
18           Speer et al. (1997) developed an aerosol liquid water content analyzer (LWCA), in which
19      aerosol samples  are collected  on PTFE filters, and then placed in a closed chamber in which the
20      relative humidity is closely controlled. The aerosol mass is monitored using a beta-gauge, as the
21      relative humidity is increased  from low RH to high RH, and then as the RH is decreased again.
22      They demonstrated the LWCA on generated aerosol and on an ambient PM2 5 sample collected in
23      RTF, NC.  The ambient aerosol sample was also analyzed for chemical constituents.  It is
24      interesting to note that while their laboratory-generated (NH4)2SO4 aerosol demonstrated a sharp
25      deliquescent point, their atmospheric aerosol, which was essentially (NH4)2SO4, did not show a
26      sharp deliquescent point.
27           Hygroscopic properties of aerosols have been studied from the viewpoint of their ability to
28      act as condensation nuclei.  The hygroscopic properties of fresh and aged carbon and diesel soot
29      particles were examined by Weingartner et al. (1997) who found that fresh, submicron-size
30      particles tended  to shrink with increasing relative humidity, due to  a crystalline structural change.
31      Lammel and Novakov (1995)  found, in laboratory studies, that the  hygroscopicity of soot

        October 1999                             3-63        DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
4
5
6
7
8
9


















particles could be increased by chemical modification, and that the cloud condensation
nucleation characteristics of diesel soot were similar to those of wood smoke aerosol.
The results of several of the above studies, in which aerosol water content as a function of
relative humidity was determined, are summarized in Figure 3-12. In this figure, the results of
Lee et al. (1997), Mclnnes et al. (1996), and Ohta et al. (1998) are included. Relative humidity
ranged from 9% (at which the aerosol water content was assumed to be zero, Mclnnes et al.,
1996) to 89%, at which the aerosol water content was determined to be 66% by mass (Lee et al.,
1997).

mn
t_ I UU •
CD
-i— •
<> 90-
T3
§ 80-
CD
i 70-
CD
O
•-E 60-
CD
Q_
0 50-
CD
= 40-
1 30-
3 20-
s. 1°-
° n.

• Mclnnes etal., 1996
• Lee etal., 1997
A Ohta etal., 1998


f







•

A
	 m, ........
10      20      30      40      50      60      70
                      Relative Humidity, %
                                                                           80
                                   90
100
      Figure 3-12.  Aerosol water content expressed as a mass percentage, as a function of
                    relative humidity.

      Source: Mclnnes et al. (1996); Lee et al. (1997); and Ohta et al. (1998).
1           The effects of relative humidity on particle growth were also examined in several studies.
2     Fang et al. (1991) investigated the effects of flow-induced relative humidity (RH) changes on
      October 1999
3-64
                                       DRAFT-DO NOT QUOTE OR CITE

-------
 1      particle cut sizes for aqueous sulfuric acid particles in a multi-nozzle micro-orifice uniform
 2      deposit impactor (MOUDI).  Laboratory experiments were conducted in which polydisperse
 3      sulfuric acid aerosols were generated and the RH was adjusted. The aerosols were analyzed by a
 4      differential mobility analyzer. Fang et al. (1991) observed that for inlet RH less than 80%, the
 5      cut sizes for the sulfuric acid aerosols were within 5% of that for nonhygroscopic particles except
 6      at the stage for which the cut size was 0.047 mm, where the cut size was 10.7% larger than the
 7      nonhygroscopic particle cut size. They concluded that flow-induced RH changes would have
 8      only a modest effect on MOUDI cut sizes at RH < 80%.
 9          Hitzenberger et al. (1997)  collected atmospheric aerosol in the size range of 0.06 - 15 mm
10      in Vienna, Austria using a 9-stage cascade impactor and measured the humidity-dependent water
11      uptake when the individual impaction foils were exposed to high RH. They observed particle
12      growth with varying growth patterns. Calculated extinction coefficients and single scattering
13      albedo increased with humidity.
14          Hygroscopic properties, along with mixing characteristics, of submicrometer particles
15      sampled in Los Angeles, CA during the summer of 1987 SCAQS study and at the Grand Canyon,
16      AZ during the 1990 Navajo Generating Station Visibility Study were reported by Zhang et al.
17      (1993). They used a tandem differential mobility analyzer (TDMA, McMurry and Stolzenburg,
18      1989)  to measure the hygroscopic properties for particles in the 0.05 to 0.5 mm range. In their
19      experimental technique, monodisperse particles of a known size are selected from the
20      atmospheric aerosol with the first DMA. Then, the relative humidity of the monodisperse
21      aerosol is adjusted and the new particle size distribution is measured with the second DMA.
22      At both sites, they observed that monodisperse particles could be classified according to "more"
23      hygroscopic and "less" hygroscopic. Aerosol behavior observed at the two sites differed
24      markedly.  The "less" hygroscopic particles sampled in Los Angeles did not grow to within the
25      experimental uncertainty (±2%) when the RH was increased to 90%, whereas at the Grand
26      Canyon, the growth of the "less" hygroscopic particles varied from day to day, but ranged from
27      near 0% to 40% when the RH was increased to 90%.  The growth of the "more" hygroscopic
28      particles in Los Angeles, CA was dependent upon particles size (15% at 0.05 //m to 60% at
29      0.5 // m) whereas at the Grand Canyon, the "more" hygroscopic particles grew by about 50%,
30      with the growth not varying significantly with particle size. By comparison of the TDMA data to
31      impactor data, Zhang et al. (1993) surmised that the more hygroscopic particles contained more

        October 1999                             3-65        DRAFT-DO NOT QUOTE OR CITE

-------
 1      sulfates and nitrates, while the less hygroscopic particles contained more carbon and crustal
 2      components.
 3           Although most of the work to date on the hygroscopic properties of atmospheric aerosols
 4      has focused on the inorganic fraction, the determination of the contribution of particle-bound
 5      water to atmospheric particulate mass is greatly complicated by the presence of organics.  The
 6      effects of RH on adsorption of semivolatile organic compounds is discussed elsewhere in this
 7      chapter. Saxena et al. (1995) observed that particulate organic  compounds can also affect the
 8      hygroscopic behavior of atmospheric particles. They idealized the organic component of aerosol
 9      as containing a hydrophobic fraction (high-molecular weight alkanes, alkanoic acids, alkenoic
10      acids, aldehydes, and ketones) and a hydrophilic fraction (e.g., lower-molecular weight
11      carboxylic acids, dicarboxylic acids, alcohols, aldehydes, etc.) that would be likely to absorb
12      water. They then analyzed data from a tandem differential mobility analyzer in conjunction with
13      particle composition observations from an urban site (Claremont, CA) and from a non-urban site
14      (Grand Canyon, AZ) to test the hypothesis that, by adding particulate organics to an inorganic
15      aerosol, the amount of water absorbed would be affected, and the effect could be positive or
16      negative, depending on the nature of the organics added. They further presumed that the
17      particulate organic matter in nonurban areas would be predominantly secondary and thus
18      hydrophilic, compared to the urban aerosol that was presumed to be derived from primary
19      emissions and thus hydrophobic in nature.  Their observations were consistent with their
20      hypothesis, in that at the Grand Canyon, the presence of organics tended to increase the water
21      uptake by aerosols, whereas  at the Los Angeles site, the presence of organics tended to decrease
22      water uptake.
23           Non-equilibrium issues may become of important for the  TDMA,  as well as for other
24      methods of measuring water content. While approach to equilibrium when the RH is increased is
25      expected to be rapid for pure salts, it may be much slower for aerosols containing a complex mix
26      of components (Saxena et al., 1995). For example, if an aerosol contains an organic film or
27      coating, that  film may impede the transport of water across the  particle surface,  thus increasing
28      the time required for equilibrium (Saxena et al., 1995). Insufficient time to achieve equilibrium
29      in the TDMA could result in underestimation of the water content.
30
        October 1999                             3-66        DRAFT-DO NOT QUOTE OR CITE

-------
 1      3.2.4 EPA Monitoring Programs
 2      3.2.4.1   The Federal Reference Methods (FRM) for Equilibrated Mass
 3           Federal Reference Methods (FRM) have been specified for measuring PM10 (Federal
 4      Register, 1987) and for measuring PM25 (Federal Register, 1997).  The FRM for PM10 has been
 5      discussed in previous AQC for PM and will only be briefly reviewed. The PM10 FRM defines
 6      performance specifications for samplers in which particles are inertially separated with a
 7      penetration efficiency of 50% at an aerodynamic diameter of 10 ± 0.5 //m. The collection
 8      efficiency increases to ~100% for smaller particles and drops to ~0% for larger particles.
 9      Particles are collected on filters, and mass concentrations are determined gravimetrically.
10      Sample volumes are adjusted to standard conditions (1 atm and 25 °C).  Measurement precision
11      for 24-hr samples must be ± 5 ug/m3 for PM10 concentrations below 80 ug/m3, and 7% above this
12      (McMurry, 1999).
13           As opposed to the performance-based FRM standard for PM10, the new FRM for PM25
14      specifies certain details of the sampler design, as well as of sample handling and analysis, while
15      other aspects have performance specifications.  The PM25 FRM sampler consists of a PM10 inlet,
16      an oil-soaked impaction substrate to remove particles larger than 2.5 mm, and a 47-mm
17      polytetrafiuoroethylene (PTFE) filter with a particle collection efficiency greater than 99.7%.
18      The sample duration is  24 hours, during which the sampler temperature is not to exceed ambient
19      temperatures by more than 5 °C. Filters are weighed before and after sampling at relative
20      humidities in the range of 30 - 40%, but controlled to within ± 5%. For sampling conducted at
21      ambient relative humidity less than 30%, mass measurements at relative humidities down to 20%
22      are permissible (McMurry, 1999).
23           The FRM also allows for Class I, II, and III equivalent methods for PM2 5. Class I
24      equivalent methods use samplers with relatively small deviations from the sampler described in
25      the FRM. Class II equivalent methods include "all other PM2 5 methods that are based upon
26      24-hr integrated filter samplers that are subjected to subsequent moisture equilibration and
27      gravimetric mass  analysis." Class III equivalent methods include filter-based methods having
28      other than a 24-hr collection interval or non-filter-based methods such as beta attenuation,
29      harmonic oscillating elements, and nephelometry (McMurry, 1999).
30           The strength of the PM25 FRM is that specification of all details of the sampler design
31      ensures that measurements at all locations should be comparable. However, the FRM requires
        October 1999                             3-67        DRAFT-DO NOT QUOTE OR CITE

-------
 1      maintenance because of the oil-soaked impaction substrate that could otherwise become loaded
 2      with coarse particles. Failure to do so could lead to coarse particle bounce, thus artificially
 3      increasing the fine particle concentrations. Moreover, the specification of a PM10 inlet requires
 4      the oil-soaked impaction substrate to collect all particles between 2.5 and 10 mm - if an inlet
 5      with a smaller cutpoint diameter were specified, coarse particle bounce could potentially be
 6      reduced, and perhaps the maintenance frequency could be reduced (McMurry, 1999).
 7           Since the implementation of the PM10 standard in  1987 (Federal Register, 1987)
 8      considerable information has accumulated on the factors that affect the quality of the data
 9      gathered from the EPA reference method for PM10. These include inlet losses of coarse fraction
10      particles (e.g., Anand et al., 1992); biases in concentrations due to differences between samplers
11      in large particle cutpoints that are within the EPA's specified acceptable tolerances (Ranade
12      et al., 1990); and particle bounce tolerances and re-entrainment leading to as much as 30% errors
13      (Wang and John, 1988).  The sampling issues associated with cutpoint tolerances are predictable,
14      and the particle bounce and re-entrainment problems have since been dealt with voluntarily by
15      the manufacturers by recommending operational procedures including oiling of impact surfaces
16      and regular cleaning. The AQC PM 96 (U.S. Environmental Protection Agency, 1996)
17      concluded that the PM10 sampling systems can be designed such that concentration measurements
18      are within ±10% of the true concentrations. For PM25, cutpoint tolerances are not expected to
19      affect the mass concentration as much as for PM10, since the 2.5 mm cutpoint generally occurs
20      near a minimum in the mass distribution (e.g., Figure 3-5).
21           The PM25 mass concentration will be affected, on the other hand, by other sampling issues
22      mentioned but not discussed extensively in the previous AQC PM 96 (U.S. Environmental
23      Protection Agency, 1996).  These included gas/particle and particle/substrate interactions for
24      sulfates and nitrates  (e.g., Appel et al., 1984); volatilization losses of nitrates (Zhang and
25      McMurry, 1992); semivolatile organic compound (SVOC) losses  (e.g., Eatough et al., 1993); and
26      relative humidity effects (e.g., Keeler et al., 1988).  Due to conversion of SO2 and nitrogen
27      oxides to particulate sulfates and nitrates, respectively, on glass fiber filters, according to the
28      previous AQC  PM 96 (U.S. Environmental Protection Agency, 1996), TSP concentration
29      measurements could be inflated by as much as 10 to 20 mg/m3. Losses of particulate nitrates,
30      chlorides,  and/or ammonium from quartz fiber filters were noted during storage or during
31      sampling by several  researchers (e.g., Witz et al., 1990). Although losses of fine particle nitrates

        October 1999                              3-68        DRAFT-DO NOT QUOTE OR CITE

-------
 1      from Teflon filters were reported by several investigators, there was some evidence that the
 2      nitrate artifacts were minor except in California (Malm et al., 1994), and unquantifiable with the
 3      current theory. Similarly, significant losses of particulate organic compounds on quartz filters
 4      were observed by Eatough et al. (1993).  Adsorption of organic vapors onto the filter or collected
 5      particulate matter was also observed to cause a positive sampling artifact during the
 6      Carbonaceous Species Methods Intercomparison Study (Hering et al., 1990), so that in regions
 7      where a significant fraction of the ambient PM mass is organic, there may be significant positive
 8      or negative errors in the mass concentration measurement.
 9
10      3.2.4.2  Speciation Monitoring
11           In addition to FRM sampling to determine compliance with PM standards, EPA requires
12      States to conduct speciation sampling primarily to determine source categories and trends. The
13      current samplers include three filters:  Teflon for equilibrated mass and elemental analysis, a
14      Nylon filter with a nitric acid denuder to collect nitrate, and a quartz fiber filter for elemental and
15      organic carbon (but without any correction for positive or negative artifacts due to adsorption of
16      organic gases or the quartz filters or evaporation of semivolatile organics from the collected
17      particles.
18
19      Measurements for Source Category Apportionment
20           Chemical analyses from the speciation network will be used for source  category
21      apportionment via receptor modeling of PM. There are two  major approaches to receptor
22      modeling: the chemical mass balance (CMB) receptor modeling approach, and statistically based
23      approaches.  The CMB approach requires chemical characterization of all relevant sources and
24      receptor sample characterization should be performed using  the same analyses.  A considerable
25      amount of receptor modeling work has been conducted with CMB and using elemental analyses
26      coupled with OC/EC and some ionic species (e.g., Watson et al., 1994;  Hidy and Venkataraman,
27      1996; McLaren and Singleton, 1996; Vega et al., 1997).   Recent developments in receptor
28      modeling include using organic analyses for tracers of specific sources (Benner et al., 1995), and
29      very detailed organic analyses for source fingerprinting (Rogge et al., 1991, 1993a,b,c,d, 1994,
30      1997a,b, 1998) and chemical mass balance receptor modeling (Schauer et al., 1996).  Further


        October 1999                              3-69         DRAFT-DO NOT QUOTE OR CITE

-------
 1      detail on the organic analyses for these studies is beyond the scope of this chapter and will not be
 2      discussed further, here.
 3           Statistical models based upon factor analysis or principal component analysis have the
 4      advantage of not requiring detailed source characterization but the drawback is that they require a
 5      large data set of receptor sample analyses. These statistically based models have an additional
 6      benefit in that they can also use other parameters such as meteorology.  For a detailed review of
 7      factor analysis and PC A, see Henry et al. (1984). In PC A, many intercorrelated variables within
 8      a large  data set are sorted into a smaller number of independent components, or factors, that
 9      account for the variability in the data set. Veltkamp et al. (1996) reported on a principal
10      component analysis (PCA) study conducted at Niwot Ridge, Colorado, during which organic
11      constituents of atmospheric aerosols were measured, along with physical and meteorological
12      data. Organic compounds were thermally desorbed from the aerosol particles at 250 C in a pure
13      helium atmosphere, separated by gas chromatography, and identified by mass spectrometry.
14      Veltkamp et al. (1996) did not report desorption recoveries or calibration procedures. For each
15      of 48 samples, 31 variables, including 18 particulate organic compounds, along with 11 organic
16      and inorganic vapor species (e.g., NO, NO2, FfNO3, HONO,  PAN, H2O2, etc.) wind direction and
17      time of day were used as variables in a principle component analysis. Several factors were
18      identified that served to distinguish various sources, and included gas phase internal combustion
19      products, particulate phase, oxygenated biogenic hydrocarbons, high molecular weight n-alkanes,
20      particulate phase anthropogenic products, and particulate phase biogenic aldehydes.
21           Pinto  et al.  (1998) also used a combination of PM25 chemical speciation and ambient
22      monitoring data in a receptor modeling calculation to determine the relative sources of particulate
23      pollution in an industrial area in the northern Bohemia region of the Czech Republic.  During
24      that study, a severe air pollution episode occurred in 1993 during which smoke and SO2
25      concentrations were 1800 and  1600 //g/m3, respectively.
26           In addition to chemical speciation for factor analysis and source apportionment, Norris
27      et al. (1999) showed that meteorological indices could prove useful in identifying sources of
28      particulate matter that are responsible for observed health effects (specifically asthma) associated
29      with exposure to particulate matter. They examined meteorology associated with elevated
30      pollution events in Spokane and in Seattle, WA, and identified a "stagnation index" that was
31      associated with low wind speeds and increases in concentrations of combustion-related

        October 1999                              3-70        DRAFT-DO NOT QUOTE OR CITE

-------
 1      pollutants. Their factor analysis also identified a meteorological index (low relative humidity
 2      and high temperatures) that was associated with increases in soil-derived particulate matter, as
 3      well as a third factor (low temperatures and high relative humidity) that was associated with
 4      increasing concentrations of particulate sulfate and nitrate species (Norris, 1998).
 5           Ondov (1996) discussed the feasibility of using sensitive isotopic and elemental tracer
 6      materials to determine the contributions of petroleum-fueled sources of PM10 in the San Joaquin
 7      Valley, CA.  Costs of these experiments are affected not only by the tracer materials cost, but
 8      also by the sensitivities of the analytical methods for each, as well as the background levels.
 9      Suarez et al. (1996) used iridium tracer to tag emissions from diesel-burning sanitation trucks in
10      Baltimore, MD, and determined the size distribution of soot from the trucks.
11
12      Elemental Analyses
13           X-ray emission, either stimulated by X-rays (X-ray fluorescence, XRF) or by proton beams
14      (Proton Induced X-ray Emission, PIXE) are standard techniques for non-destructive analysis of
15      certain elements. Some newer techniques with some advantages have become available in recent
16      years.
17           Instrumental Neutron Activation Analysis (INAA).  Instrumental  neutron activation
18      analysis (INAA) was mentioned briefly in the previous AQC PM 96 and will be expanded upon
19      here. INAA has been used to examine the chemical composition of atmospheric aerosols in
20      several studies, either as the only method of analysis, or in addition to XRF (e.g., Yatin et al.,
21      1994; Gallorini, 1995).  INAA has the advantage of having a higher sensitivity for many trace
22      species, and it is particularly useful in analyzing for many trace metals. Landsberger and Wu
23      (1993) analyzed air samples collected near Lake Ontario for Sb, As, Cd,  In, I, Mo, Si, and V,
24      using INAA. They demonstrated that using INAA in conjunction with epithermal neutrons and
25      Compton suppression produces very precise values with relatively low detection limits.
26           Enriched rare-earth isotopes have been analyzed via INAA and used to trace sources of
27      particulate matter from a coal-fired power plant (Ondov et al., 1992); from various sources in the
28      San Joaquin Valley (Ondov, 1996);  from intentially tagged (iridium) diesel emissions from
29      sanitation trucks (Suarez et al.,  1996; Wu et al.,  1998) and from iridium-tagged emissions from
30      school buses (Wu et al., 1998).


        October 1999                              3-71        DRAFT-DO NOT QUOTE OR CITE

-------
 1           An intercomparison was conducted in which 18 pairs of filters were sent to participants in
 2     the Coordinated Research Program (CRP) on Applied Research on Waste Using Nuclear Related
 3     Analytical Techniques (Landsberger et al., 1997). As part of that study, participants used PIXE,
 4     INAA, XRF, or AAS to analyze the samples. Many of the results for XRF and PIXE in the coarse
 5     fraction were observed to be biased low compared to INAA. The authors speculated that there is
 6     a systematic error due to self-attenuation of the X-rays due to particle size effect.
 7           Atomic Absorption Spectrophotometry (AAS). Atomic Absorption Spectrophotometry
 8     (AAS) was used to characterize the atmospheric deposition of trace elements Zn, Ni, Cr, Cd, Pb,
 9     and Hg, to the Rouge River watershed by particulate deposition (Pirrone and Keeler, 1996).  The
10     modeled deposition rates were compared to annual emissions of trace elements that were
11     estimated from the emissions inventory for coal and oil combustion utilities, iron-steel
12     manufacturing, metal production, cement manufacturing, and solid waste and sewage sludge
13     incinerators.  They found generally good agreement between the trend observed in atmospheric
14     inputs to the river (dry + wet deposition) and annual emissions  of trace elements, with
15     r2 ~ 0.84 - 0.98.  Both atmospheric inputs and emissions were found to have followed downward
16     trends for Pb.  For the period of 1987 to 1988, steady increases  were observed for Cd  (major
17     sources are municipal solid waste incineration, coal combustion, sludge incineration, and iron
18     and steel manufacturing); Cr and Ni (major sources are iron and steel production, and coal
19     combustion); and Hg (major sources are coal, the contribution from which had decreased from
20     53 to 45%, and municipal, solid waste, and medical waste incineration, the contribution from
21     which has increased).
22           Inductively Coupled Plasma - Mass Spectroscopy (ICP-MS). Keeler and Pirrone (1996)
23     also used inductively coupled plasma - mass spectroscopy (ICP-MS) to determine trace elements
24     Cd, Mn, V, As, Se, and Pb in atmospheric particulate fine (PM25) and total suspended particulate
25     samples collected in two Detroit sites. The results were then similarly used in a deposition
26     model to estimate the dry deposition flux of trace elements to Lake Erie.
27
28     Elemental/Organic Carbon, Soot, or Particulate Organic Matter
29           Total carbon in aerosol particles (TC) can be expressed as the sum of organic carbon (OC),
30     elemental carbon (EC), and carbonate carbon (CC), with the contribution of CC to TC usually on
31     the order of 5% or less, for particulate samples collected in urban areas (Appel, 1993). The AQC

       October 1999                             3-72        DRAFT-DO NOT QUOTE OR CITE

-------
 1      PM 96 (U.S. Environmental Protection Agency, 1996) listed several filter-based, thermal
 2      methods for measuring OC and EC, and described the thermal/optical reflectance (TOR) method,
 3      which was noted, along with thermal manganese oxidation, to be one of the most commonly
 4      applied methods in the U.S. at the time. In thermal separation methods, thermally evolved
 5      OC and EC are oxidized to CO2 and quantified either by nondispersive infrared detection or
 6      electrochemically, or the CO2 can be reduced to CH4 and quantified via flame ionization
 7      detection (FID). The various methods give similar results for TC, but not for EC or OC. In a
 8      methods comparison study (Countess, 1990), it was shown that it is necessary to minimize or
 9      correct for pyrolytically generated EC ("char"), and that CC found in wood smoke and
10      automobile exhaust samples may interfere with some of the thermal methods. Recently,
11      Lavanchy et al. (1999) reported on a study in which the operation of a catalytic oxidation system
12      was modified in an attempt to minimize pyrolysis of OC, and at the same time minimize the
13      oxidation of CaCO3.  In the thermal apparatus used by Lavanchy et al. (1999), a filter sample is
14      placed in a moveable sample boat and, prior to insertion into the 340 °C oven, the sample is flash
15      heated in the 650 °C oven for one minute, in order to minimize charring.  It then is inserted into
16      the first stage of a two-stage oven, in which OC is oxidized to CO2 at 340 °C in the presence of
17      O2 for 42 minutes.  The filter is then moved into the second stage oven, in which EC is oxidized
18      at 650 °C, for 32 minutes. This temperature is reported to be sufficient to completely oxidize
19      EC, but with only about 1% of the CaCO3 being oxidized (Lavanchy et al., 1999; Petzold et al.,
20      1997). In order to test for charring, they challenged their system with atmospheric samples for
21      which duplicates were analyzed via the German reference method for measuring OC and EC in
22      atmospheric samples (Petzold and Niessner, 1995), in which a solvent extraction is used to
23      remove organics before combustion.  Lavanchy et al. (1999) reported a high correlation
24      (R2 = 0.97) between their thermal oxidation method and the German method VDI. They also
25      reported detection limits of 1.3 //g for EC and 1.8 //g for OC.
26           The thermal/optical transmission method (TOT) was mentioned briefly in the AQC PM 96
27      (U.S. Environmental Protection Agency, 1996), as being similar to the TOR with the exception
28      that light transmission rather than reflectance is monitored on the filter throughout the analysis.
29      The National Institute for Occupational Safety and Health (NIOSH) Method 5040 for monitoring
30      elemental carbon as a marker for particulate diesel exhaust is based upon a TOT method analyzer
31      (Birch and Gary, 1996) and has gained significantly in popularity since then, and so will be

        October 1999                              3-73         DRAFT-DO NOT QUOTE OR CITE

-------
 1      described in more detail, here.  The OC/EC method described by Birch and Gary (1996) is a TOT
 2      method similar to the TOR method described in the AQC PM 96 (U.S. Environmental Protection
 3      Agency, 1996) in that temperature and atmospheric control are used to accomplish carbon
 4      speciation; several temperature steps are utilized; carbon evolved is converted to methane and
 5      quantified with FID; and light transmission, rather than reflectance, off the filter is measured
 6      during the analysis. In thermal/optical methods, the optical feature allows for correction for
 7      pyrolytic char generated during the analysis. The OC/EC method of Birch and Gary (1996)
 8      consists of a two stage process, with the first stage being conducted in a pure helium atmosphere,
 9      and the second stage conducted in a 10% oxygen-helium mix.  The temperature is ramped to
10      about 820 °C in the helium phase, during which organic and carbonate carbon are volatilized
11      from the filter. In the second stage, the oven temperature is reduced, and then raised to about
12      860°C. During the second stage, pyrolysis correction and EC measurement  are made.
13      Figure 3-13, an example of a thermogram, shows temperature, transmittance, and FID response
14      traces. Peaks are evident that correspond to OC, CC, EC, and pyrolitic carbon (PC). As can be
15      seen in this figure, the high temperature in the first stage allows for decomposition of CC. The
16      ability to quantify PC is particularly important in high OC/EC regions (like wood smoke
17      -impacted airsheds), allowing for the volatilization of any remaining complex organic
18      compounds so they are not apportioned to the EC phase.
19          Black carbon (BC) is also used, in addition to the thermal and thermal/optical methods, for
20      determining EC as a measure of soot (Penner and Novakov, 1996). Both EC and BC define a
21      similar fraction of aerosol, but EC is determined based upon thermal properties, while BC is
22      based upon light-absorption properties. Optical methods for determining BC tend to suffer from
23      calibration problems (Hitzenberger et al., 1996). Lavanchy et al. (1999) compared their EC
24      concentrations as determined from their catalytic thermal oxidation method to BC concentrations
25      determined using an aethalometer operated at the same site, and found that the instrumental
26      calibration factor provided by the manufacturer was on the order of two times the calibration
27      factor they determined (9.3 ± 0.4 m2 g"1).  It is possible to calculate a theoretical specific
28      absorption coefficient (BJ from Mie theory given a known size distribution  and refractive index,
29      and typically BC aerosols have values of Ba between 3  and 17m2 g"1 (Hintzenberger et al. [1996]
30      and references therein). The absorption coefficient, (BJ, is defined as absorption per mass
31      concentration and can be calculated given the sample filter area, the total deposited mass, and

        October 1999                              3-74        DRAFT-DO NOT QUOTE OR CITE

-------
                                                             DC . EC spirt
                                            Time/mi ft
      Figure 3-13.  This thermogram, for a sample containing rock dust (a carbonate source)
                    and diesel exhaust, shows three traces that correspond to temperature, filter
                    transmittance, and FID detector response. Peaks correspond to organic
                    (OC), carbonate (CC), pyrolytic (PC), and elemental (EC) carbon.  The final
                    peak is a methane calibration peak.

      Source: Birch and Gary (1996).
1

2

3

4

5

6
absorption signals for both the loaded and unloaded filters. Often, when no direct measurements

are available, values of Ba on the order of 10 m2 g"1 have been used (Hintzenberger et al. (1996),

and references therein).  European countries are trying to set air pollution standards that target

diesel vehicles, one of the principal sources of BC in urban areas (Hintzenberger et al. (1996),

and references therein) and so it is essential that accurate values for Ba are available.

Hintzenberger et al. (1996) investigated the feasibility of using an integrating sphere photometer
      October 1999
                                         3-75
DRAFT-DO NOT QUOTE OR CITE

-------
 1      as an adequate measurement system for the BC content and the absorption coefficient. Based
 2      upon samples collected during a 10-day period in May 1994, they determined that the usually
 3      assumed value of 10 m2 g"1 was also applicable to aerosol BC occurring in Vienna.
 4           In 1986, the Carbonaceous Species Methods Comparison Study (CSMCS) was conducted
 5      in Los Angeles, CA, during which a number of methods for the measurement of this species were
 6      intercompared.  The CSMCS was mentioned in the previous AQC PM 96 (U.S. Environmental
 7      Protection Agency, 1996), however it is interesting to add that Hansen and McMurry (1990)
 8      specifically compared two very dissimilar methods for aerosol elemental carbon - collection of
 9      impactor samples backed by a quartz fiber afterfilter, followed by EC analysis by oxidation in
10      helium over a MnO2 catalyst, and real-time measurements using an aethalometer (an optical
11      absorption technique) - and found good agreement between these two, very different methods.
12      The comparisons between organic carbon measurements exhibited considerably less agreement.
13           Hitzenberger et al. (1999) recently reported on a study in which the integrating sphere
14      method was compared to an aethalometer (Hansen et al., 1984), the thermal method of Cachier
15      et al. (1989), and the thermal/optical method of Birch and Gary (1996). The absorption
16      coefficients that were obtained from both the integrating sphere and the aethalometer were
17      comparable. The BC mass concentration obtained from the aethalometer were 23% of those
18      obtained from the integrating sphere. Compared to the thermal method, the integrating sphere
19      overestimated the BC mass concentrations by 21%.  Compared to the thermal/optical method, the
20      integrating sphere was within 5% of the 1:1 line, however the data were not so well-correlated.
21           Turpin et al. (1990) reported on an in-situ, time-resolved analyzer for particulate organic
22      and elemental carbon that could operate on a time cycle as short as 90 minutes. The analyzer is
23      comprised of a filter-based sampling section and a thermal-optical carbon detector. Adsorbed
24      organic material is thermally desorbed from the filter at 650 °C and oxidized at 1000°C over a
25      MnO2 catalyst bed. The evolved CO2 is converted to methane over a nickel catalyst, and the
26      methane is measured in a flame ionization detector.  Then the elemental carbon is oxidized in a
27      98% He-2% O2 atmosphere, at 350°C. Correction is made for pyrolytic conversion of some of
28      the organic particulate matter. The instrument was operated with a 2 h time resolution during the
29      Southern California Air Quality Study (SCAQS) in 1987 (Turpin and Huntzicker, 1991), as well
30      as during the Carbonaceous Species Methods Comparison Study (CSMCS), in 1986.


        October  1999                              3-76        DRAFT-DO NOT QUOTE OR  CITE

-------
 1          In summary, the state of the art for soot measurements continues to develop, and although
 2     advances are being made, the definitions of EC and BC continue to be operational and
 3     determined by the method employed.
 4
 5     3.2.5  Continuous Monitoring
 6          The U.S. EPA expects that 100 local agency monitoring sites throughout the States will
 7     operate continuous PM monitors. However, EPA has not yet provided any guidance regarding
 8     what continuous monitoring technique to use.  All currently available continuous measurements
 9     of suspended particle mass share the problem of dealing with semivolatile PM components.
10     In order not to include particle-bound water as part of the mass, the particle bound water must be
11     removed by heating or dehumidification. However, heating also causes loss of ammonium
12     nitrate and semivolatile organic components. A variety of potential candidates for continuous
13     measurement of mass or  chemical components will be discussed in this section.
14
15     3.2.5.1  TEOM
16          The advantages of continuous PM monitoring, and the designation of the Tapered Element
17     Oscillating Microbalance (TEOM) as an equivalent method for PM10, has led to the deployment
18     of the TEOM at a number of air monitoring sites.  The TEOM  is also being used to measure
19     PM25.  The TEOM differs philosophically from the federal reference methods for particulate
20     mass in that it does not require equilibration of the samples at a specified temperature and
21     relative humidity. Moreover, the TEOM samples at a constant temperature, typically heated to
22     some temperature higher than the ambient temperature (Meyer et al.,  1995; Meyer and
23     Rupprecht, 1996), whereas the federal reference methods sample at the ambient temperature.
24     Thus, the TEOM may not provide data equivalent to the FRM due to losses of volatile species.
25     Volatilization losses in the TEOM sampler can be reduced by operating the instrument heated to
26     30 °C rather than the 50 °C specified, during the cooler times of the year, and by using Nafion
27     dryers  on the inlet.
28          This philosophical difference in operation and the implications for fine particle
29     measurements were examined by researchers at CSIRO Atmospheric Research in Australia
30     (Ayers et al., 1999). That group compared 24-hr mean PM25 mass concentrations as determined
31     by a TEOM and by two manual, gravimetric samplers (a low-volume filter sampler and a
       October 1999                             3-77        DRAFT-DO NOT QUOTE OR CITE

-------
 1      MOUDI sampler) in four Australian cities, on 15 days in the winter half-year. The TEOM was
 2      operated at 50 °C at one location and at 35 °C at the other three locations.  A systematically low
 3      TEOM response in comparison to the integrated gravimetric methods was observed. In a
 4      comprehensive study, Allen et al. (1997) reported results in which TEOM data collected at
 5      10 urban sites in the U.S. and Mexico were compared with 24-hr integrated mass concentrations,
 6      for both PM10 and PM2 5. They collected  a large data set that included both winter and summer
 7      seasons.  Allen et al. (1997) concluded that, especially for urban areas, a significant portion of
 8      PM10 could be semivolatile compounds that could be lost from the heated filter in the TEOM
 9      thus leading to a systematic difference between the TEOM and the EPA FRM for PM10.
10      Moreover, they suggested that this difference is likely to be larger for PM25 than it is for PM10
11      (Allen etal,  1997).
12
13      3.2.5.2   RAMS
14           A Real-Time total Ambient Mass Sampler, RAMS, based on diffusion denuder and TEOM
15      monitor technology has been developed, validated, and field tested (Eatough et al., 1999; Obeidi
16      and Eatough, 1999) for the real-time determination of total fine particulate mass, including
17      semivolatile species.  The RAMS measures total mass of collected particles, including
18      semivolatile species with a TEOM monitor using a "sandwich filter". The "sandwich" contains a
19      Teflon coated particle collection filter followed by a charcoal impregnated filter (GIF) to collect
20      any semivolatile  species lost from the particles during sampling. Since the instrument measures
21      total mass collected by the "sandwich filter," all gas phase compounds that can be adsorbed by a
22      GIF must be removed from the sampling stream prior to the TEOM monitor. Laboratory and
23      field validation data indicate that the precision of fine PM mass determination is better than 10%.
24      The RAMS uses a Nafion dryer to remove particle bound water from the suspended particles and
25      a particle concentrator to reduce the amount of gas phase organics that must be removed by the
26      denuder. An example of data from the RAMS, the TEOM,  and the  PC-BOSS is shown in
27      Figure 3-14.
28
29      3.2.5.3   CAMM
30           Koutrakis and colleagues (Koutrakis et al., 1996; Wang, 1997) have developed the
31      Continuous Ambient Mass Monitor (CAMM), a technique for the continuous measurement of

        October  1999                             3-78         DRAFT-DO NOT QUOTE OR CITE

-------
                   PC-BOSS (Nonvolatile Material)
                 PC-BOSS (Lost From Particles)
                     TEOM
                      at35C
—B—  RAMS
       at35C
       	FRM PM25
             24 h average
          100
                                      Riverside,  CA
                                                         •Q
                                    &JL	5L_
              13 14  15 16 17  18 19  20 21  22 23  0  1   2  3   4  5  6   7  8   9
                                           Time of Day
      Figure 3-14.  Comparison of mass measurements with collocated RAMS (real-time data),
                  PC-BOSS (1-hour data), FRM PM25 sampler (24-hour data) and a
                  conventional TEOM monitor (real-time data).  The semivolatile fine
                  particulate matter is sampled with the RAMS and PC-BOSS, but not with
                  the TEOM monitor or the FRM PM25 single filter sampler. The PC-BOSS
                  provides information on both the nonvolatile component (NV) and the
                  semivolatile organic component (SVOC).

      Source: Eatoughetal. (1999).
1     ambient particulate matter mass concentration, based upon the measurement of pressure drop

2     increase with particle loading across a membrane filter. Recently, Sioutas et al. (1999) examined

3     the increase in pressure drop with increasing particle loading on Nuclepore filters. They tested

4     filters with two pore diameters (2 and 5 mm) and filter face velocities ranging from 4 to 52 cm/s,

5     and examined the effects of relative humidity in the range of 10 to 50%. They found that, for
      October 1999
      3-79
DRAFT-DO NOT QUOTE OR CITE

-------
 1      hygroscopic ammonium sulfate particles, the change in pressure drop per unit time and
 2      concentration was a strong function of relative humidity, decreasing with increasing relative
 3      humidity.  These results suggest that particulate concentration measurements like the method of
 4      Koutrakis et al. (1996) that use the pressure drop method may be subject to additional
 5      uncertainties if used in an environment where the ambient relative humidity cannot be accurately
 6      controlled.  The current version of the CAMM (Wang, 1997) uses a particle concentrator, a
 7      Nafion dryer,  and frequent changes of the position on the filter tape were the pressure drop
 8      measurement  is made to avoid artifacts due to semivolatile components.
 9
10      3.2.5.4  Light Scattering
11           The evaporation of ammonium nitrate aerosol in a heated nephelometer was examined by
12      Bergin et al. (1997). This is potentially of concern because the nephelometer operates in part by
13      heating the ambient aerosol to a low reference relative humidity of 40%, in order to measure the
14      light scattering intrinsic to the aerosol rather than including atmospheric relative humidity.
15      Bergin et al. conducted laboratory experiments at low relative humidity (~10%) and as a function
16      of temperature (300 - 320K), mean residence time in the nephelometer, and initial particle size
17      distribution. The evaporation of ammonium nitrate aerosol was also modeled, for comparison,
18      and was found to accurately describe the decrease in aerosol scattering coefficient as a function
19      of aerosol physical properties, and nephelometer operating conditions. Bergin et al. (1997)
20      determined an upper limit estimate of the decrease in the aerosol light scattering coefficient at
21      450 nm due to evaporation for typical  field conditions. The model estimates for their worst-case
22      scenario suggest that the decrease in the aerosol scattering coefficient could be roughly 40%.
23      Under most conditions, however, they estimate that the decrease in aerosol scattering coefficient
24      is generally expected to be less than 20%.
25           Morawska et al. (1996) examined the correlations between PM10, visibility, and submicron
26      concentration data in Brisbane, and concluded that the different principles  of operation for each
27      instrument and the different  aerosol characteristics measured by each technique make it difficult
28      to observe any relationships. Morawska et al. (1998b) reported on a long-term monitoring
29      program that included the criteria pollutants as well as light scattering, number/size distributions,
30      number concentrations, and  elemental analysis via inductively coupled plasma mass
31      spectrometry. Particle size classification was conducted using a TSI scanning mobility particle

        October 1999                              3-80        DRAFT-DO NOT QUOTE  OR CITE

-------
 1      sizer for the size range of 0.016 to 0.7 mm, and a TSI aerodynamic particle sizer for the size
 2      range of 0.7 to 30 mm. They reported correlation coefficients between the light-scattering
 3      coefficient and PM10, SMPS concentration, and APS concentration of 0.58, 0.38, and 0.37,
 4      respectively. They also reported a correlation coefficient between PM10 and the SMPS
 5      concentration of 0.25, which is consistent with the notion that PM10 mass measurements would
 6      provide less information about ultrafine particles.
 7
 8      3.2.5.5  Beta-Gauge Techniques
 9          The use of absorption of beta radiation as a indicator of particle mass has been used
10      effectively to measure the mass of equilibrated particulate matter collected on Teflon filters
11      (Jaklevic et al., 1981; Courtney et al., 1982). The technique has also been used to provide near
12      real-time measurements with time intervals on the order of an hour (Wedding and Weigand,
13      1993). However, real-time beta gauge monitors experience the same problems as other
14      continuous or near real-time particular matter mass monitoring techniques. Particle-bound water
15      must be removed to reduce the sensitivity of the indicated mass to relative humidity.  However,
16      the simplest technique, mild heating, will remove a portion of the ammonium nitrate and the
17      semivolatile organic compounds as well as the particle-bound water.
18          An intercomparison study of two beta gauges at three sites indicated that the Wedding beta
19      gauge and the Sierra Anderson SA 1200 PM10 samplers were highly correlated (r>0.97) (Tsai and
20      Cheng, 1996). The  Wedding beta gauge was not sensitive to relative humidity but was
21      approximately seven percent lower. This suggests that the mild heating in the beta gauge causes
22      losses comparable to those due to equilibration, although the differences could be due to slight
23      differences in the upper cut points. The Kimoto beta gauge, however, which was operated at
24      ambient temperature, was sensitive to relative humidity, yielding significantly higher mass
25      concentrations relative to the Sierra Anderson SA 1200 for RH>80% than for RH<80%, even
26      though the correlation with the SA 1200 was reasonable, r=0.94 for RH>80% and 0.83 for
27      RH<80%.
28
29      3.2.5.6  Measurements of Individual Particles
30          A new technique, aerosol time-of-flight mass spectroscopy (ATOFMS), has demonstrated
31      the ability for real-time measurement of correlated size and composition profiles of individual

        October 1999                             3-81        DRAFT-DO NOT QUOTE OR CITE

-------
1      atmospheric aerosol particles (Noble and Prather, 1996; Gard et al., 1997). Measurements are
2      made in-situ by combining a dual-laser aerodynamic particle sizing system to size and track
3      individual particles through the instrument and laser desorption/ionization time-of-flight mass
4      spectrometry to obtain correlated single particle composition data. By measuring both positive
5      and negative  ions from the same particle, information can be obtained about the chemical
6      composition, not just the elemental composition, of individual particles of know aerodynamic
7      diameter. This information is especially useful in determining sources of particles.  An example
8      of the type of information that can be determined is shown in Figure 3-15.
           o_
           eg

       Figure 3-15.  Size Distribution of particles divided by chemical classification into organic,
                    marine, and crustal.
1           Until recently, single particle ATOFMS systems have only been able to characterize
2     particles that are larger than approximately 0.2 to 0.3 mm in diameter. The work of Wexler and
3     colleagues (Carson et al., 1997; Ge et al., 1998) have developed a single particle, TOFMS
       October 1999
3-82
DRAFT-DO NOT QUOTE OR CITE

-------
 1      instrument that it is able to size, count and provide chemical composition on individual particles
 2      ranging in size from 10 nm to 2 //m.
 3
 4
 5      3.3  SUMMARY
 6           Atmospheric particles originate from a variety of sources and possess a range of
 7      morphological, chemical, physical, and thermodynamic properties. The composition and
 8      behavior of airborne particles are linked with those of the surrounding gas.  Aerosol is defined as
 9      a suspension of solid or liquid particles in air and includes both the particles and all vapor or gas
10      phase components of air. However, the term aerosol is often used to refer to the suspended
11      particles only. Particulate is an adjective and should only be used as a modifier, as in particulate
12      matter.
13           A complete description of the atmospheric aerosol would include an accounting of the
14      chemical composition, morphology, and size of each particle and the relative abundance of each
15      particle type as a function of particle size. Recent developments in single particle analysis
16      techniques are bringing such a description closer to reality.
17           The diameter of a spherical particle may be determined geometrically, from optical or
18      electron microscopy; by light scattering and Mie theory; or by its behavior,  such as its electrical
19      mobility or its aerodynamic behavior.  However, the various types of diameters may be different
20      and atmospheric particles often are not spherical. Therefore, particle diameters are described by
21      an "equivalent" diameter.  Aerodynamic diameter, i.e., the diameter of a unit density sphere
22      which would have the same physical behavior, is the most widely used equivalent diameter.
23      Therefore, in this document, particle diameters, unless otherwise indicated, refer to the
24      aerodynamic diameter.
25           Atmospheric size distributions show that most atmospheric particles are quite small, below
26      0.1 //m, while most of the particle volume (and therefore most of the mass) is found in particles
27      greater than 0.1 //m. The surface area peaks around 0.1 //m. An important feature of the mass  or
28      volume size distributions of atmospheric aerosols is their multimodal nature.  Volume
29      distributions, measured in ambient air in the United States, are almost always found to be
30      bimodal, with a minimum between 1.0 and 3.0 //m.  The distribution of particles that are mostly
31      larger than the minimum is termed the coarse mode.  The distribution of particles that are mostly
        October 1999                              3-83         DRAFT-DO NOT QUOTE OR CITE

-------
 1      smaller than the minimum is termed the fine mode. Fine-mode particles include both the
 2      accumulation mode and the nuclei mode.  Accumulation-mode particles are that portion of the
 3      fine particle fraction with diameters above about 0.1 //m. The nuclei mode, that portion of the
 4      fine particle fraction with diameters below about 0.1 //m, can be observed as a separate mode in
 5      mass or volume distributions only in clean or remote areas or near sources of new particle
 6      formation by nucleation. Toxicologists use ultrafine to refer to particles, generated in the
 7      laboratory, which are in the nuclei-mode size range. Aerosol physicists and material scientists
 8      tend to use nanoparticles to refer to particles in this size range generated in the laboratory.
 9           The aerosol community uses four different approaches or conventions in the classification
10      of particles  by size:  (1) modes, based on the observed size distributions and formation
11      mechanisms; (2) cut point, usually based on the 50% cut point of the specific sampling device,
12      i.e., the particle size at which 50% of the particles enter and 50% of the particles are rejected;
13      (3) occupational sizes, based on the entrance into various compartments of the respiratory
14      system; and (4) legally-specified, regulatory sizes for air quality standards.  Over the years, the
15      terms fine and coarse, as applied to particle sizes, have lost the original precise meaning of fine
16      mode and the coarse mode. In any given article, therefore, the meaning of fine and coarse, unless
17      defined, must be inferred from the author's usage.  In particular, PM25 and fine mode particles
18      are not equivalent.  In this document, the term mode is used with fine and coarse when it is
19      desired to specify the distribution of fine-mode particles or coarse-mode particles as shown in
20      Figures 3-4 and 3-5.
21           Several processes influence the formation and growth of particles. New particles may be
22      formed by nucleation from gas phase material. Particles may grow by condensation as gas phase
23      material condenses onto existing particles. Particles may also grow by coagulation as two
24      particles combine to form one. Gas phase material condenses preferential on smaller particles
25      and the rate constant for coagulation of two particles decreases as the particle size increases.
26      Therefore, nuclei mode particles grow into the accumulation mode but accumulation mode
27      particles do not grow into the coarse mode.
28           The major constituents of atmospheric PM are sulfate, nitrate, ammonium, and hydrogen
29      ions; particle-bound water; elemental carbon; a great variety of organic compounds; and crustal
30      material. Atmospheric PM contains a large number of elements in various compounds and
31      concentrations  and hundreds to thousands of specific organic compounds. Particulate material

        October 1999                              3-84       DRAFT-DO NOT QUOTE OR CITE

-------
 1      can be primary or secondary. PM is called primary if it is in the same chemical form in which it
 2      was emitted into the atmosphere.  PM is called secondary if it is formed by chemical reactions in
 3      the atmosphere. Primary coarse particles are usually formed by mechanical processes. Primary
 4      fine particles are emitted from sources,  either directly as particles or as vapors which rapidly
 5      condense to form particles.
 6           Most of the sulfate and nitrate and a portion of the organic compounds in atmospheric
 7      particles are secondary, i.e., they are formed by chemical reactions in the atmosphere. Secondary
 8      aerosol formation depends on numerous factors including the concentrations of precursors; the
 9      concentrations of other gaseous reactive species such as ozone, hydroxyl radical, or hydrogen
10      peroxide; atmospheric conditions including solar radiation and relative humidity; and the
11      interactions of precursors and preexisting particles within cloud or fog droplets or on or in the
12      liquid film on solid particles. As a result, it is considerably more difficult to relate ambient
13      concentrations of secondary species to sources of precursor emissions than it is to identify the
14      sources of primary particles.
15           The lifetimes  of particles vary with particle size.  Coarse particles can settle rapidly from
16      the atmosphere within minutes or hours, and normally travel only short distances. However,
17      when mixed high into the atmosphere, as in dust storms, the smaller-sized, coarse-mode particles
18      may have longer lives and travel distances. Nuclei mode particles rapidly grow into the
19      accumulation mode. However, the accumulation mode does not grow into the coarse mode.
20      Accumulation-mode fine particles are kept suspended by normal air motions and have very low
21      deposition rates to surfaces. They can be transported thousands of km and remain in the
22      atmosphere for a number of days. Accumulation-mode particles are removed from the
23      atmosphere primarily by cloud processes. Coarse mode particles of less than ~10 //m diameter as
24      well as accumulation-mode and nuclei-mode (or ultrafine) particles all have the ability to
25      penetrate deep into  the lungs and be removed by deposition in the lungs. Dry deposition rates are
26      expressed in terms of a deposition velocity which varies as the particle size, reaching a minimum
27      between 0.1 and 1.0 //m aerodynamic diameter.
28           The role of particles in reducing visibility and affecting radiative balance through scattering
29      and absorption of light is evident as are the effects of particles in soiling and damaging materials.
30      EPA addresses visibility effects through regional haze regulations.  The direct effects of particles


        October 1999                              3-85        DRAFT-DO NOT QUOTE OR CITE

-------
 1      in scattering and absorbing light and the indirect effects of particles on clouds are being
 2      addressed in climate change programs in several government agencies.
 3           The role of PM in acid deposition has not always been recognized.  Acid deposition and
 4      PM are intimately related, however, first because particles contribute significantly to the
 5      acidification of rain and secondly because the gas phase species that lead to dry deposition of
 6      acidity are also precursors of particles.  Therefore, reductions in SO2 and NOX emissions will
 7      decrease both acid deposition and PM concentrations. Sulfuric acid, ammonium nitrate, and
 8      organic particles are also deposited on surfaces by dry deposition.  The utilization of nitrate by
 9      plants leads to the production of acidity. Therefore, dry deposition of particles can also
10      contribute to the ecological damages caused by acid deposition.
11           It has been proposed that particles could act as carriers to transport toxic gases into the deep
12      lung.  Water-soluble gases, which would be removed by deposition to wet surfaces in the upper
13      respiratory system during inhalation, could dissolve in particle-bound water and be carried with
14      the particles into the deep lung.  Equilibrium calculations indicate that particles do not increase
15      vapor deposition in human airways. However, their calculations do show that soluble gases are
16      carried to higher generation airways (deeper into the lung) in the presence of particles than in the
17      absence of particles. In addition, species such as SO2 and formaldehyde react in water,  reducing
18      the concentration of the dissolved gas-phase species, and providing a kinetic resistence to the
19      evaporation of the dissolved gas. Thus, the concentration of the dissolved species may be greater
20      than that predicted by the equilibrium calculations. Toxic species, such as NO, NO2, benzene,
21      polycyclic aromatic hydrocarbons (PAH), nitro-PAH, and a variety of allergens may be absorbed
22      on solid particles and carried into the lungs.
23           The decision by the US EPA to revise the PM standards by adding daily and yearly
24      standards for PM2 5 has led to a renewed interest in the measurement of atmospheric particles and
25      also to a better understanding of the problems in developing precise and accurate measurements
26      of particles.  Unfortunately, it is very difficult to measure  and characterize particles suspended in
27      the atmosphere.
28           The US Federal Reference Methods (FRM) for PM25 and PM10 provide relatively precise
29      (±10 %) methods for determining the mass of material remaining on a Teflon filter after
30      equilibration. However, numerous uncertainties exist as to the relationship between the mass and
31      composition of material remaining on the filter, as measured by the FRMs, and the mass and

        October 1999                              3-86        DRAFT-DO  NOT QUOTE OR CITE

-------
 1      composition of material that exists in the atmosphere as suspended PM. It is currently not
 2      possible to accurately measure what exists as a particle in the atmosphere, in part because of the
 3      difficulty of creating a reference standard for particles suspended in the atmosphere. As a result,
 4      EPA defines accuracy for PM measurements in terms of agreement of a candidate sampler with a
 5      reference sampler. Therefore, intercomparisons of samplers become very important in
 6      determining how well various samplers agree and how various design choices influence what is
 7      actually measured. Reasons for measuring particles include:  attainment of a standard,
 8      implementation of a standard, determination of health effects, determination of ecological
 9      effects, and determination of radiative effects.
10           Current filtration-based mass measurements lead to  significant evaporative losses, during
11      and possibly after collection, of a variety of semivolatile components, i.e., species which exist in
12      the atmosphere in dynamic equilibrium between the condensed phase and gas phase. Important
13      examples include ammonium nitrate, semivolatile organic compounds, and particle-bound water.
14      Other areas where choices must be made in designing an aerosol indicator include selection of an
15      upper cut point; separation of fine-mode and coarse-mode PM; and treatment of pressure,
16      temperature, and relative humidity.
17           It is becoming increasingly apparent that the semivolatile component of PM may
18      significantly impact the quality of the measurement, and can lead to both positive and negative
19      sampling artifacts. Negative artifacts, due to loss of ammonium nitrate and semivolatile organic
20      compounds, may occur during  sampling, due to changes in temperature, relative humidity, or
21      composition of the aerosol, or due to pressure drop across the filter. Negative artifacts may also
22      occur during handling and storage due to evaporation. Positive artifacts occur when volatile
23      species adsorb onto, or react with, filter media and/or collected PM.
24           The loss of particulate  nitrate may be determined by comparing nitrate collected on a
25      Teflon filter to that collected on a nylon filter (which absorbs nitrate) preceded by a denuder to
26      remove nitric acid. In two studies, the PM2 5 mass lost due to volatilization of ammonium nitrate
27      was found to represent a significant fraction of the total PM2 5 mass (9% to 21%. The fraction
28      lost was higher during summer than during fall. The nitrate obtained from Teflon filter samples
29      was on average 28% lower than that obtained from denuded nylon filters. In these studies
30      samples were extracted immediately after sampling. Therefore, losses that might occur during
31      handling, storage, and equilibration were avoided.  Semivolatile organic compounds (SVOC) can

        October 1999                              3-87        DRAFT-DO NOT QUOTE OR CITE

-------
 1      similarly be lost from Teflon filters due to volatilization during or after collection. Such losses
 2      can cause the PM2 5 mass to be significantly underestimated. The FRM for PM2 5 will suffer loss
 3      of particulate nitrates and SVOC, similar to the losses experienced with other single filter
 4      collection systems.
 5           Much progress has been made to date in the design of diffusion denuder systems for the
 6      measurement and characterization of both the particulate and gaseous phases of semivolatile
 7      compounds.  Some of the recent research has focused upon reduction in the size of the denuder,
 8      optimization of the residence time in the denuder, understanding the effect of diffusion denuders
 9      on the positive quartz filter artifact, identifying changes in chemical composition that occur
10      during sampling, determining the effects due to changes in temperature and relative humidity,
11      and identifying possible loses by absorption in impactor coatings.
12           It is generally desirable to collect and measure ammonium nitrate and semivolatile organic
13      compounds.  However, for many measurement of suspended particle mass, it is desirable to
14      remove the particle-bound water before determining the mass.  In other situations it may be
15      important to know how much of the suspended particle's mass or volume is due to
16      particle-bound water. Calculation and measurements indicate that aerosol water content is
17      strongly dependent upon composition but that liquid water could represent a significant mass
18      fraction of aerosol concentration at relative humidities above 60%.
19           Federal Reference Methods (FRM) for equilibrated mass have been specified for PM10 and
20      PM25. In addition to FRM sampling to determine compliance with PM standards, EPA requires
21      States to conduct speciation  sampling primarily to determine source categories and trends.  The
22      current speciation samplers include three filters: Teflon for equilibrated mass and elemental
23      analysis, a Nylon filter with a nitric acid denuder to collect nitrate, and a quartz fiber filter for
24      elemental and organic carbon (but without any correction for positive or negative artifacts due to
25      adsorption of volatile organic compounds on the quartz filters or evaporation of semivolatile
26      organic compounds from the collected particles.
27           The U.S. EPA expects that 100 local agency monitoring sites throughout the States will
28      operate continuous  PM monitors. However, EPA has not yet provided any guidance regarding
29      appropriate continuous monitoring techniques. All currently available continuous measurements
30      of suspended particle mass share the problem of dealing with semivolatile PM components.
31      In order not to include particle-bound water as part of the mass, the particle bound water must be

        October 1999                              3-88         DRAFT-DO NOT QUOTE OR CITE

-------
 1     removed by heating or dehumidification. However, heating also causes loss of ammonium
 2     nitrate and semivolatile organic components. Potential candidates for continuous measurement
 3     of mass include the Tapered Element Oscillating Microbalance (TEOM), which determines
 4     non-volatile mass by measuring the change in frequency of a vibrating quartz tube with a filter on
 5     the end; the Real-Time total Ambient Mass Sampler, RAMS, based on diffusion denuder and
 6     TEOM monitor technology, which attempts to determine nonvolatile plus semivolatile mass by
 7     using a charcoal impregnated filter (GIF) to collect any semivolatile species lost from the
 8     particles during sampling (all gas phase compounds that can be adsorbed by a GIF must be
 9     removed from the sampling stream prior to the TEOM monitor); the Continuous Ambient Mass
10     Monitor (CAMM), a technique based upon the measurement of pressure drop increase with
11     particle loading across a membrane filter; and a variety of techniques based on light scattering or
12     measurement of particles mass on a filter by absorption of beta radiation.  In addition to
13     continuous mass measurement a number of techniques for continuous measurement of sulfate or
14     semicontinuous measurements of nitrate and elements have been demonstrated. More recently, a
15     new technique, aerosol time-of-fiight mass spectroscopy (ATOFMS), that has demonstrated the
16     ability for real-time measurement of correlated size and composition profiles of individual
17     atmospheric aerosol particles.
18
       October 1999                             3-89        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES
  2
  3       Allen, G.; Sioutas, C.; Koutrakis, P.; Reiss, R.; Lurmann, F. W.; Roberts, P. T. (1997) Evaluation of the TEOM
  4             method for measurement of ambient particulate mass in urban areas. J. Air Waste Manage. Assoc.
  5             47:682-689.
  6       Allen, J. O.; Paschkewitz, J. S.; Plummer, E. F.; Lafleur, A. L.; Sarofim, A. F.; Smith, K. A. (1999) Absorption of
  7             semi-volatile compounds in oiled impaction substrates: measurement of pyrene absorption. Aerosol Sci.
  8             Technol. 30: 16-29.
  9       Altshuller, A. P. (1983) Review: natural volatile organic substances and their effect on air quality in the United
10             States. Atmos. Environ. 17:2131-2165.
11       Amdur, M. O. (1957) The influence of aerosols upon the respiratory response of guinea pigs to sulfur dioxide.
12             Am. Ind. Hyg. Assoc. Q. 18: 149-155.
13       Amdur, M. O. (1960) The response of guinea pigs to inhalation of formaldehyde and formic acid alone and with a
14             sodium chloride aerosol. Int. J. Air Pollut. 3: 201-220.
15       Amdur, M. O. (1966) Respiratory absorption data and SO 2 dose-response curves. Arch. Environ. Health
16             12:729-732.
17       Amdur, M. O. (1969) Toxicologic appraisal of particulate matter,  oxides of sulfur, and sulfuric acid. J. Air Pollut.
18             Control Assoc. 19: 638-644.
19       Amdur, M. O.; Mead, J. (1958) Mechanics of respiration in unanesthetized guinea pigs. Am. J. Physiol.
20             192: 364-368.
21       Amdur, M. O.; Underbill, D. (1968) The effect of various aerosols on the response of guinea pigs to sulfur dioxide.
22             Arch. Environ. Health 16: 460-468.
23       American Conference of Governmental Industrial Hygienists. (1994) Appendix D: particle size-selective sampling
24             criteria for airborne particulate matter. In:  1994-1995 Threshold limit values for chemical substances and
25             physical agents and biological exposure indices. Cincinnati, OH: American Conference of Governmental
26             Industrial Hygienists; pp. 43-46.
27       Anand, N. K.; McFarland, A. R.; Kihm, K. D.; Wong, F. S. (1992) Optimization of aerosol penetration through
28             transport lines. Aerosol Sci. Technol. 16: 105-112.
29       Andrews, E.;  Larson, S. M. (1993) Effect of surfactant layers on the size changes of aerosol particles as a function
30             of relative humidity. Environ. Sci. Technol.  27: 857-865.
31       Appel, B. R. (1993) Atmospheric sample analysis and sampling artifacts. In: Willeke, K.; Baron, P. A., eds. Aerosol
32             measurement: principles, techniques, and applications. New York, NY: Van Nostrand Reinhold; pp. 233-259.
33       Appel, B. R.; Tokiwa, Y.; Haik, M. (1981) Sampling of nitrates in ambient air. Atmos. Environ. 15: 283-289.
34       Appel, B. R.; Tokiwa, Y.; Haik, M.; Kothny, E. L.  (1984) Artifact particulate sulfate  and nitrate formation on filter
35             media. Atmos. Environ. 18: 409-416.
36       Ayers, G. P.; Keywood, M. D.; Gras, J. L. (1999) TEOM vs. manual gravimetric methods for determination of
37             PM2.5 aerosol mass concentrations. Atmos. Environ. 33: 3717-3721.
38       Ball, J. C.; Hurley, M. D.; Straccia, A. M.; Gierczak, C. A. (1999) Thermal release of nitric oxide from ambient air
39             and diesel particles. Environ. Sci. Technol. 33:  1175-1178.
40       Benner, C. L.; Eatough, D. J.; Eatough, N. L.; Bhardwaja, P. (1991) Comparison of annular denuder and filter pack
41             collection of HNO3 (g), HNO2 (g), SO2 (g), and particulate-phase nitrate, nitrite and sulfate in the south-west
42             desert. Atmos. Environ. Part A 25: 1537-1545.
43       Benner, B. A., Jr.; Wise, S. A.; Currie, L. A.; Klouda,  G. A.; Klinedinst, D. B.; Zweidinger, R. B.; Stevens, R. K.;
44             Lewis, C. W. (1995) Distinguishing the contributions of residential wood combustion and mobile source
45             emissions using relative concentrations of dimethylphenanthrene isomers. Environ. Sci. Technol.
46             29:2382-2389.
47       Bergin, M. H.; Ogren, J. A.; Schwartz, S.  E.; Mclnnes, L. M. (1997) Evaporation of ammonium nitrate aerosol in a
48             heated nephelometer: implications for field measurements.  Environ. Sci. Technol. 31: 2878-2883.
49       Bertoni, G.; Febo, A.; Perrino, C.; Possanzini, M. (1984) Annular active diffuse sampler: a new devise for the
50             collection of organic vapours. Ann. Chim. (Rome) 74: 97-104.
51       Binkowski, F. S.; Ching, J. K.  S. (1996) Modeling fine particulate mass and visibility using the EPA regional
52             particulate model. In: Proceedings from the ninth joint conference on applications of air pollution
53             meteorology; Jan-Feb 1996; Atlanta, GA. Boston, MA: American Meteorological Society; pp. 565-569.
         October 1999                                   3-90         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Binkowski, F. S.; Shankar, U. (1995) The regional particulate matter model. I. Model description and preliminary
  2             results. J. Geophys. Res. [Atmos.] 100: 26,191-26,209.
  3       Birch, M. E.; Gary, R. A. (1996) Elemental carbon-based method for occupational monitoring of particulate diesel
  4             exhaust: methodology and exposure issues. Analyst 121: 1183-1190.
  5       Brorstrom, E.; Grennfelt, P.; Lindskog, A. (1983) The effect of nitrogen dioxide and ozone on the decomposition of
  6             particle-associated polycyclic aromatic hydrocarbons during sampling from the atmosphere. Atmos. Environ.
  7             17:601-605.
  8       Bunz, H.; Koyro, M.; Mohler, O.; Saathoff, H. (1996) Temperature dependence of the redistribution mechanisms
  9             for volatile components of atmospheric aerosol particles. J. Aerosol Sci. 27(suppl. 1):  S653-S654.
10       Cachier, H.; Bremond, M.-P.; Buat-Menard, P. (1989) Determination of soot carbon with a simple thermal method.
11             Tellus 416:379-390.
12       Carson, P. G.; Johnston, M. V.; Wexler, A. S. (1997) Laser desorption/ionization of ultra fine aerosol particles.
13             Rapid Commun. Mass Spectrom. 11: 993.
14       Chen, B. T. (1993) Instrument calibration. In: Willeke, K.; Baron, P. A., eds. Aerosol measurement: principles,
15             techniques, and applications. New York, NY: Van Nostrand Reinhold; pp. 493-520.
16       Chow, J. C. (1995) Measurement methods to determine compliance with ambient air quality standards for
17             suspended particles. J. Air Waste Manage. Assoc. 45: 320-382.
18       Chow, J. C.; Fujita, E.  M.; Watson, J. G.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994) Evaluation of filter-based
19             aerosol measurements during the 1987 Southern California Air Quality Study.  Environ. Monit. Assess.
20             30:49-80.
21       Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Countess, R. J. (1996) Sources and chemistry of PM 10 aerosol in
22             Santa Barbara County, CA. Atmos. Environ. 30: 1489-1499.
23       Clark, W. E.; Whitby, K. T. (1975) Measurements of aerosols produced by the photochemical oxidation of SO 2 in
24             air. J. Colloid Interface Sci. 51:477-490.
25       Clarke, A. D. (1992) Atmospheric nuclei in the remote free-troposphere. J. Atmos. Chem. 14: 479-488.
26       Code of Federal Regulations. (1991a) Appendix J to Part 50—reference method for the determination of particulate
27             matter as PM10 in the atmosphere. C. F.  R. 40: §50.
28       Code of Federal Regulations. (1991b) Ambient air monitoring reference and equivalent methods. C. F. R. 40: §53.
29       Code of Federal Regulations. (1997a) Appendix L to Part 50—reference method for the determination of fine
30             particulate matter as PM2 5 in the atmosphere. C. F. R. 40: §50.
31       Code of Federal Regulations. (1997b) Revised requirements for designation of reference and equivalent methods for
32             PM2 5 and ambient air quality surveillance for particulate matter. C. F. R.  40: §§53, 58.
33       Gotham, W. E.; Bidleman, T. F. (1992) Laboratory investigations of the partitioning of organochlorine compounds
34             between the gas phase and atmospheric aerosols on glass fiber filters. Environ. Sci. Technol. 26: 469-478.
35       Countess, R. J. (1990)  Interlaboratory analyses of carbonaceous aerosol samples. Aerosol Sci. Technol.
36             12:114-121.
37       Courtney, W. J.; Shaw, R. W.; Dzubay, T. G. (1982) Precision and accuracy of a  P gauge for aerosol mass
38             determinations.  Environ. Sci.  Technol. 16: 236-239.
39       Coutant, R. W.; Callahan, P. J.; Chuang, J. C.;  Lewis, R. G. (1992) Efficiency of silicone-grease-coated denuders
40             for collection of polynuclear aromatic hydrocarbons. Atmos. Environ. Part A 26: 2831-2834.
41       Covert, D. S.; Kapustin, V. N.; Quinn, P. K.; Bates, T. S. (1992) New particle formation in the marine boundary
42             layer. J. Geophys. Res. [Atmos.] 97: 20,581-20,589.
43       Cui, W.; Machir, J.; Lewis , L.; Eatough, D. J.; Eatough, N. L. (1997) Fine particulate organic material at Meadview
44             during the project MOHAVE summer intensive study. J. Air Waste Manage. Assoc. 47: 357-369.
45       Cui, W.; Eatough, D. J.; Eatough, N. L. (1998) Fine particulate organic material in the Los Angeles Basin -1:
46             assessment of the high-volume Brigham Young University organic sampling system, BIG BOSS. J. Air
47             Waste Manage.  Assoc. 48: 1024-1037.
48       Davis, C. S.; Fellin, P.; Otson, R. (1987) A review of sampling methods for polyaromatic hydrocarbons in air.
49             JAPCA 37: 1397-1408.
50       Ding, Y.; Cui, W.; Lee, M. L.; Eatough, D. J. (1998a) Fine particulate n-nitroso and nitrite organic compounds in
51             the atmosphere. Appl. Occup. Environ. Hyg. 13: 432-438.
52       Ding, Y.; Lee, M. L.; Eatough, D. J.  (1998b) The determination of total nitrite and n-nitroso compound in
53             atmospheric samples. J. Environ. Anal. Chem. 69: 243-255.
54       Durham, J. L.; Wilson, W. E.; Bailey, E. B. (1978) Application of an SO 2-denuder for continuous measurement of
55             sulfur in submicrometric aerosols. Atmos. Environ. 12: 883-886.


         October 1999                                   3-91         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Eatough, D. J. (1999) BOSS, the Brigham Young University Organic Sampling System: determination of participate
  2             carbonaceous material using diffision denuder sampling technology. In: Lane, D., ed. Gas/phase partition
  3             measurements of atmospheric organic compounds. New York, NY: Gordon and Breach Science Publishers;
  4             in press.
  5       Eatough, D. J.; Wadsworth, A.; Eatough, D. A.; Crawford, J. W.; Hansen, L. D.; Lewis, E. A. (1993)
  6             A multiple-system, multi-channel diffusion denuder sampler for the determination of fine-particulate organic
  7             material in the atmosphere. Atmos. Environ. Part A 27: 1213-1219.
  8       Eatough, D. J.; Tang, H.; Cui, W.; Machir, J. (1995) Determination of the size distribution and chemical
  9             composition of fine particulate semi-volatile organic material in urban environments using diffusion denuder
10             technology. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on particulate air pollution
11             and human mortality and morbidity, part II; January 1994; Irvine, CA. Inhalation Toxicol. 7: 691-710.
12       Eatough, D. J.; Eatough, D. A.; Lewis, L.; Lewis, E. A. (1996) Fine particulate chemical composition and light
13             extinction at Canyonlands National Park using organic particulate material concentrations obtained with a
14             multisystem, multichannel diffusion denuder sampler. J. Geophys. Res. [Atmos.] 101: 19,515-19,531.
15       Eatough, D. J.; Obeidi, F.; Pang, Y.; Ding, Y.; Eatough, N. L.; Wilson, W. E.  (1999) Integrated and real-time
16             diffusion denuder sampler for PM25. Atmos. Environ. 33: 2835-2844.
17       Falconer, R. L.; Bidleman, T. F.; Gotham, W. E. (1995) Preferential sorption of non- and
18             mono-ortho-polychlorinated biphenyls to urban aerosols. Environ. Sci. Technol. 29: 1666-1673.
19       Fan, Z.; Kamens, R.; Hu, J. X.; Zhang, J. B.; McDow, S. (1996a) Photostability of nitro-polycyclic aromatic
20             hydrocarbons on combustion soot particles in sunlight. Environ. Sci. Technol. 30: 1358-1364.
21       Fan, Z.; Kamens, R. M.; Zhang, J.; Hu, J. (1996b) Ozone—nitrogen dioxide—NPAH heterogeneous soot particle
22             reactions and modeling NPAH in the atmosphere. Environ. Sci. Technol. 30:  2821-2827.
23       Fang, C. P.; McMurry, P.  H.; Marple, V. A.; Rubow, K. L. (1991) Effect of flow-induced relative humidity changes
24             on size cuts for sulfuric acid droplets in the microorifice uniform deposit impactor (MOUDI). Aerosol Sci.
25             Technol. 14: 266-277.
26       Federal Register. (1987) Revisions to the national ambient air quality standards for particulate matter. F.  R.  (July 1)
27             52:24,634-24,669.
28       Federal Register. (1997) National ambient air quality standards for particulate matter; final rule. F. R. (July  18)
29             62:38,652-38,752.
30       Federal Register. (1998) National ambient air quality standards for particulate matter and revised requirements for
31             designation of reference and equivalent methods for PM 2 5 and ambient air quality surveillance for particulate
32             matter; correction. F. R. (February 17) 63: 7710-7718.
33       Finn, D.; Rumburg, B.;  Claiborn, C.; Bamesberger, W.; Norris, G.; Koenig, J.; Larson, T. (1999) Sampling artifacts
34             from the use of Teflon filters and denuder tubes during the measurement of airborne carbonaceous particulate
35             matter. J. Air Waste Manage. Assoc.: submitted.
36       Fitz, D. R. (1990) Reduction of the positive organic artifact on quartz filters. Aerosol. Sci. Technol. 12: 142-148.
37       Friedlander, S. K. (1970) The characterization of aerosols distributed with respect to size and chemical composition.
38             J. Aerosol Sci. 1:295-307.
39       Friedlander, S. K. (1977) Smoke, dust and haze: fundamentals of aerosol behavior. New York, NY: John Wiley &
40             Sons.
41       Friedlander, S. K.; Yeh. E. K. (1998) The submicron atmospheric aerosol as a carrier of reactive chemical species:
42             case of peroxides. Appl. Occup. Environ. Hyg. 13: 416-420.
43       Gallorini, M.  (1995) Trace element monitoring in atmospheric pollution processes by neutron activation analysis.
44             Microchem.  J. 51: 127-137.
45       Gard, E.; Mayer, J. E.; Morrical, B. D.; Dienes, T.; Fergenson, D. P.; Prather, K. A.  (1997) Real-time analysis of
46             individual atmospheric aerosol particles: design and performance of a portable ATOFMS. Anal. Chem.
47             69:4083-4091.
48       Ge, Z.; Wexler, A. S.; Johnston, M. V. (1998) Laser desorption/ionization of single ultrafine multicomponent
49             aerosols. Environ. Sci. Technol. 32: 3218-3223.
50       Goss, K.-U.; Eisenreich, S. J. (1997) Sorption of volatile organic compounds to particles from a combustion source
51             at different temperatures and relative humidities. Atmos. Environ. 31: 2827-2834.
52       Griffin, R. J.; Cocker, D. R., Ill; Seinfeld, J. H. (1999) Incremental aerosol reactivity: application to aromatic and
53             biogenic hydrocarbons. Environ.  Sci. Technol. 33: 2403-2408.
54       Grosjean, D.  (1983) Polycyclic aromatic hydrocarbons in Los Angeles air from samples collected on Teflon, glass
55             and quartz filters. Atmos. Environ. 17: 2565-2573.


         October 1999                                   3-92          DRAFT-DO NOT QUOTE OR CITE

-------
  1       Grosjean, D.; Parmar, S. S. (1990) Interferences from aldehydes and peroxyacetyl nitrate when sampling urban air
  2             organic acids on alkaline traps. Environ. Sci. Technol. 24: 1021-1026.
  3       Grosjean, D.; Seinfeld, J. H. (1989) Parameterization of the formation potential of secondary organic aerosols.
  4             Atmos. Environ. 23: 1733-1747.
  5       Gundel, L. A.; Lane, D. A. (1999) Sorbent-coated diffusion denuders for direct measurement of gas/particle
  6             partitioning by semi-volatile organic compounds. In: Lane, D. A., eds. Gas and Particle Phase Measurements
  7             of Atmospheric Organic Compounds. Newark, NJ: Gordon and Breach. (Advances in environmental process
  8             control technologies, v. 2).
  9       Gundel, L. A.; Lee, V. C.; Mahanama, K. R. R.; Stevens, R. K.; Daisey, J. M. (1995a) Direct determination of the
10             phase distributions of semi-volatile polycyclic aromatic hydrocarbons using annular denuders. Atmos.
11             Environ. 29: 1719-1733.
12       Gundel, L. A.; Mahanama, K. R. R.; Daisey, J. M. (1995b) Semivolatile and particulate polycyclic aromatic
13             hydrocarbons  in environmental tobacco smoke: cleanup, speciation, and emission factors. Environ. Sci.
14             Technol. 29: 1607-1614.
15       Haagen-Smit, A. J. (1952) Chemistry and physiology of Los Angeles smog. Ind. Eng. Chem. 44: 1342-1346.
16       Hansen, A. D. A.; McMurry, P. H. (1990) An intercomparison of measurements of aerosol elemental carbon during
17             the  1986 carbonaceous species method comparison study. J. Air Waste Manage. Assoc. 40: 894-895.
18       Hansen, A. D. A.; Rosen, H.; Novakov, T. (1984) The aethalometer - an instrument for the real-time measurement
19             of optical absorption by aerosol particles. Sci. Total Environ. 36: 191-196.
20       Henry, R. C.; Lewis,  C. W.; Hopke, P. K.; Williamson, H. J. (1984) Review of receptor model fundamentals.
21             Atmos. Environ. 18:1507-1515.
22       Hering, S.; Cass, G. (1999) The magnitude of bias in the measurement of PM 2 5 arising from volatilization of
23             particulate nitrate from Teflon filters. J. Air Waste Manage. Assoc. 49: 725-733.
24       Hering, S.  V.; Appel, B. R.; Cheng, W.; Salaymeh, F.; Cadle, S. H.; Mulawa, P. A.; Cahill, T. A.; Eldred, R. A.;
25             Surovik, M.; Fitz, D.; Howes, J. E.; Knapp, K. T.; Stockburger, L.; Turpin, B. J.; Huntzicker, J. J.; Zhang,
26             X.-Q.; McMurry, P. H. (1990) Comparison of sampling methods for carbonaceous aerosols in ambient air.
27             Aerosol Sci. Technol. 12: 200-213.
28       Hidy, G. M.; Venkataraman, C. (1996) The chemical mass balance method for estimating atmospheric particle
29             sources in southern California. Chem. Eng. Commun. 151: 187.
30       Hitzenberger, R.; Dusek, U.; Berner, A. (1996) Black carbon measurements using an integrating sphere. J. Geophys.
31             Res. [Atmos.] 101: 19,601-19,606.
32       Hitzenberger, R.; Berner, A.; Dusek, U.; Alabashi, R. (1997) Humidity-dependent growth of size-segregated aerosol
33             samples. Aerosol Sci. Technol. 27: 116-130.
34       Hitzenberger, R.; Jennings, S. G.; Larson, S. M.; Dillner, A.; Cachier, H.; Galambos, Z.; Roue, A.; Spain, T. G.
35             (1999) Intercomparison of measurement methods for black carbon aerosols. Atmos. Environ. 33: 2823-2833.
36       Hoffmann, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.; Seinfeld, J. H. (1997) Formation
37             of organic aerosols from the oxidation of biogenic hydrocarbons. J. Atmos. Chem. 26: 189-222.
38       Izumi, K.;  Fukuyama, T. (1990) Photochemical aerosol formation from aromatic hydrocarbons in the presence of
39             NOX. Atmos. Environ. Part A 24: 1433-1441.
40       Jaklevic, J. M.; Gatti, R. C.; Goulding, F. S.; Loo, B. W. (1981) A p-gauge method applied to aerosol samples.
41             Environ. Sci. Technol. 15: 680-686.
42       Jang, M.; Kamens, R. M. (1998) A thermodynamic approach for modeling partitioning of semivolatile organic
43             compounds on atmospheric particulate matter: humidity effects. Environ.  Sci. Technol. 32: 1237-1243.
44       Jang, M.; Kamens, R. M. (1999) Newly characterized products and composition of secondary aerosols from the
45             reaction of cc-pinene with ozone. Atmos. Environ. 33: 459-474.
46       Jang, M.; Kamens, R. M.; Leach, K. B.; Strommen, M. R. (1997) A thermodynamic approach using group
47             contribution methods to model the partitioning of semivolatile organic compounds on atmospheric particulate
48             matter. Environ. Sci. Technol. 31: 2805-2811.
49       Jenkins, B. M.; Jones, A. D.; Turn, S. Q.; Williams, R. B. (1996) Particle concentrations, gas-particle partitioning,
50             and species intercorrelations for polycyclic aromatic hydrocarbons (PAH) emitted during biomass burning.
51             Atmos. Environ. 30: 3825-3835.
52       John, W.; Wall, S. M.; Ondo, J. L. (1988) A new method for nitric acid and nitrate aerosol measurement using the
53             dichotomous sampler. Atmos. Environ. 22: 1627-1635.
54       Kalberer, M.; Ammann, M.; Gaggeler, H. W.; Baltensperger, U. (1999) Adsorption of NO 2 on carbon aerosol
55             particles in the low ppb range. Atmos. Environ. 33: 2815-2822.


         October  1999                                   3-93          DRAFT-DO NOT QUOTE OR  CITE

-------
  1       Kamens, R. M.; Coe, D. L. (1997) A large gas-phase stripping device to investigate rates of PAH evaporation from
  2             airborne diesel soot particles. Environ. Sci. Technol. 31: 1830-1833.
  3       Kamens, R.; Odum, J.; Fan, Z.-H. (1995) Some observations on times to equilibrium for semivolatile polycyclic
  4             aromatic hydrocarbons. Environ. Sci. Technol. 29: 43-50.
  5       Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of cc-pinene and ozone
  6             using a gas-phase kinetics-aerosol partitioning model. Environ. Sci. Technol. 33: 1430-1438.
  7       Kao, A. S.; Friedlander, S. K. (1995) Frequency distributions of PM 10 chemical components and their sources.
  8             Environ. Sci. Technol. 29: 19-28.
  9       Keeler, G. J.; Pirrone, N. (1996) Atmospheric transport and deposition of trace elements to Lake Erie from urban
10             areas. Water Sci. Technol. 33: 259-265.
11       Keeler, G. J.; Brachaczek, W. W.; Gorse, R. A., Jr.; Japar, S. M.; Pierson, W. R. (1988) Effect of ambient humidity
12             on dichotomous sampler coarse/fine ratios. Atmos. Environ. 22:  1715-1720.
13       Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
14             segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
15             Environ. 33: 2907-2913.
16       Khalil, M. A. K.; Rasmussen, R. A. (1992) Forest hydrocarbon emissions: relationships between fluxes and ambient
17             concentrations. J. Air Waste Manage. Assoc. 42: 810-813.
18       Kim, Y. J.; Boatman, J. F.; Gunter, R. L.; Wellman, D. L.; Wilkison, S. W. (1993) Vertical distribution of
19             atmospheric aerosol size distribution over south-central New Mexico. Atmos.  Environ. Part A
20             27: 1351-1362.
21       Kittelson, D. B. (1998) Engines and nanoparticles: a review. J. Aerosol Sci. 29: 575-588.
22       Klouda, G. A.; Klinedinst, D. B.; Steel, E. B.; Benner, B. A., Jr.; Parish, H.  J. (1996) Exploring a method to
23             produce an urban dust particle filter standard. J. Aerosol Sci. 27(suppl.  1): S351-S352.
24       Koutrakis, P.; Wolfson, J. M.; Spengler, J.  D. (1988) An improved method for measuring aerosol strong acidity:
25             results from a nine-month study in St. Louis,  Missouri and Kingston, Tennessee. Atmos. Environ.
26             22: 157-162.
27       Koutrakis, P.; Sioutas, C.; Ferguson, S. T.; Wolfson, J. M.; Mulik,  J. D.; Burton, R. M. (1993) Development and
28             evaluation of a glass honeycomb denuder/filter pack system to collect atmospheric gases and particles.
29             Environ. Sci. Technol. 27: 2497-2501.
30       Koutrakis, P.; Wang, P.-Y.; Wolfson, J. M.; Sioutas, C., inventors.  (1996) Method and apparatus to measure
31             particulate matter in gas. U.S. patent 5,571,945. November 5. Available:  www.patents.ibm.com.
32       Krieger, M. S.; Hites, R. A. (1992) Diffusion denuder for the collection of semivolatile organic compounds.
33             Environ. Sci. Technol. 26: 1551-1555.
34       Lammel, G.; Novakov, T.  (1995) Water nucleation properties of carbon black and diesel soot particles. Atmos.
35             Environ. 29:813-823.
36       Landsberger, S.; Wu, D. (1993) Improvement of analytical sensitivities for the determination of antimony, arsenic,
37             cadmium, indium, iodine, molybdenum, silicon and uranium in airborne particulate matter by epithermal
38             neutron activation analysis. J. Radioanal. Nucl. Chem. 167: 219-225.
39       Landsberger, S.; Wu, D.; Vermette, S. J.; Cizek, W. (1997) Intercomparison of IAEA airborne particulate matter
40             reference material.  J. Radioanal. Nucl. Chem. 215: 117-127.
41       Lane, D. A.; Gundel, L. A. (1996) Gas and particle sampling of airborne polycyclic aromatic compounds. In:
42             15th International symposium on polycyclic aromatic compounds: second biennial meeting of International
43             Society for Polycyclic Aromatic Compounds; September 1995; Belgirate, Italy. Polycyclic Aromat. Compd.
44             9:67-73.
45       Lavanchy, V. M. H.; Gaggeler, H. W.; Nyeki, S.; Baltensperger, U. (1999) Elemental carbon (EC) and black carbon
46             (BC) measurements with a thermal method and an aethalometer at the high-alpine research station
47             Jungfraujoch. Atmos. Environ. 33: 2759-2769.
48       Lawrence, J. E.; Koutrakis, P. (1994) Measurement of atmospheric formic and acetic acids: methods evaluation and
49             results from field studies. Environ. Sci. Technol. 28: 957-964.
50       Lawrence, J.; Koutrakis, P. (1996a) Measurement and speciation of gas and particulate phase organic acidity in an
51             urban environment: 1. analytical. J. Geophys. Res. [Atmos.] 101: 9159-9169.
52       Lawrence, J.; Koutrakis, P. (1996b) Measurement and speciation of gas and particulate phase organic acidity in an
53             urban environment: 2. speciation. J.  Geophys. Res. [Atmos.] 101: 9171-9184.
54       Lawson, D. R. (1990) The Southern California Air Quality Study. J. Air Waste Manage. Assoc. 40: 156-165.
         October 1999                                   3-94         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Lee, W.-M. G.; Shih, P.-M.; Wang, C.-S. (1997) The influence of relative humidity on the size of atmospheric
  2             aerosol. J. Environ. Sci. Health Part A: Environ. Sci. Eng. Toxic Hazard. Subst. Control 32: 1085-1097.
  3       Li, W.; Montassier, N.; Hopke, P. K. (1992) A system to measure the hygroscopicity of aerosol particles. Aerosol
  4             Sci. Technol. 17: 25-35.
  5       Li, X. Y.; Gilmour, P. S.; Donaldson, K.; MacNee, W. (1997) In vivo and in vitro proinflammatory effects of
  6             particulate air pollution (PM10). In: Driscoll, K. E.; Oberdorster, G., eds. Proceedings of the sixth
  7             international meeting on the toxicology of natural and man-made fibrous and non-fibrous particles;
  8             September 1996; Lake Placid, NY. Environ. Health Perspect. Suppl. 105(5): 1279-1283.
  9       Liang, Z.; Chan, C. K. (1997) A fast technique for measuring water activity of atmospheric aerosols. Aerosol Sci.
10             Technol. 26: 255-268.
11       Liang, C.; Pankow, J. F.; Odum, J. R.; Seinfeld, J. H. (1997) Gas/particle partitioning of semivolatile organic
12             compounds to model inorganic, organic, and ambient smog aerosols. Environ. Sci. Technol. 31: 3086-3092.
13       Lindskog, A.; Brorstrom-Lunden, E. (1987) Chemical transformation of PAH on airborne particles by exposure to
14             NO2 during sampling: a comparison between 2 filter media. Sci. Total Environ. 61: 51-57.
15       Lison, D.; Lardot, C.; Huaux, F.; Zanetti, G.; Fubini, B. (1997) Influence of particle surface area on the toxicity of
16             insoluble manganese dioxide dusts. Arch. Toxicol. 71: 725-729.
17       Lundgren, D. A.; Burton, R. M. (1995) Effect of particle size distribution on the cut point between fine and coarse
18             ambient mass fractions. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the colloquium on particulate air
19             pollution and human mortality and morbidity; January 1994; Irvine, CA. Inhalation Toxicol. 7: 131-148.
20       Mahanama, K. R. R.; Gundel, L. A.; Daisey, J. M. (1994) Selective fluorescence detection of polycyclic aromatic
21             hydrocarbons in environmental tobacco smoke and other airborne particles. Int. J. Environ. Anal. Chem.
22             56:289-309.
23       Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T. A. (1994) Spatial and seasonal trends in particle
24             concentration and optical extinction in the United States. J. Geophys. Res. [Atmos.] 99: 1347-1370.
25       Mclnnes, L. M.; Quinn, P. K.; Covert, D. S.; Anderson, T. L. (1996) Gravimetric analysis, ionic composition, and
26             associated water mass of the marine aerosol. Atmos. Environ. 30: 869-884.
27       McLaren, R.; Singleton, D. L. (1996) Real-world measurements of exhaust and evaporative emissions in the Cassair
28             Tunnel predicted by chemical mass balance modeling. Environ. Sci. Technol. 30: 3001.
29       McMurry, P. H. (1999) A review of atmospheric aerosol measurements. Prepared for the NARSTO assessment.
30       McMurry, P. H.; Friedlander, S. K. (1979) New particle formation in the presence of an aerosol.  Atmos. Environ.
31             13:1635-1651.
32       McMurry, P. H.; Stolzenburg, M.  R. (1989) On the sensitivity of particle size to relative humidity for Los Angeles
33             aerosols. Atmos. Environ. 23: 497-507.
34       Meyer, A. S., Jr.; Boyd, C.  M. (1959) Determination of water by titration with  coulometrically generated Karl
35             Fischer  reagent. Anal. Chem. 31:  215-219.
36       Meyer, M. B.;  Rupprecht, E. G. (1996) Particulate matter sampling methods: the importance of standardization.
37             J.  Aerosol. Sci. 27(suppl. 1):  S349-S350.
38       Meyer, M. B.;  Rupprecht, E.; Patashnick, H. (1995) Considerations for the sampling and measurement of ambient
39             particulate matter. In: Particulate matter: health and regulatory issues: proceedings of an international
40             specialty conference; April; Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association;
41             pp. 204-210. (A&WMA publication VIP-49).
42       Miller, F. J.; Gardner, D. E.; Graham, J.  A.; Lee, R. E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
43             considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc. 29: 610-615.
44       Morawska, L.; Thomas, S. B.; Bofinger,  N. D.; Neale, D.; Wainwright, D. (1996) Characterisation of airborne
45             particulate pollution. Presented at: 13th International clean air conference proceedings.
46       Morawska, L.; Bofinger, N. D.; Kocis, L.; Nwankwoala, A. (1998a) Submicrometer and supermicrometer particles
47             from diesel vehicle emissions. Environ. Sci. Technol. 32: 2033-2042.
48       Morawska, L.; Thomas, S.; Bofinger, N.; Wainwright, D.; Neale, D. (1998b) Comprehensive characterization of
49             aerosols in a subtropical urban atmosphere: particle size distribution and correlation with gaseous pollutants.
50             Atmos.  Environ. 32: 2467-2478.
51       Mukerjee, S.; Ellenson, W. D.; Lewis, R. G.; Stevens, R. K.; Somerville, M. C.; Shadwick, D. S.; Willis, R. D.
52             (1997) An environmental scoping study in the lower Rio Grande Valley of Texas—III. residential
53             micro environmental monitoring for air, house dust, and soil. Environ. Int. 23: 657-673.
54       Muzyka, V.; Veimer, S.; Shmidt, N. (1998) Particle-bound benzene from diesel engine exhaust. Scand.  J. Work
55             Environ. Health 24:  481-485.


         October 1999                                   3-95          DRAFT-DO NOT QUOTE OR CITE

-------
  1       National Research Council. (1993) Protecting visibility in national parks and wilderness areas. Washington, DC:
  2             National Academy Press. 3v.
  3       Noble, C. A.; Prather, K. A. (1996) Real-time measurement of correlated size and composition profiles of individual
  4             atmospheric aerosol particles. Environ. Sci. Technol. 30: 2667-2680.
  5       Norris, G. (1998) Air pollution and the exacerbation of asthma in an arid, western, US city [dissertation]. Seattle,
  6             WA: University of Washington. Available from: University Microfilms International, Ann Arbor, MI;
  7             publication no. 9836228.
  8       Norris, G.; Larson, T.; Koenig, J.; Claiborn, C.; Sheppard, L.; Finn, D. (1999) Asthma aggravation, combustion, and
  9             stagnant air. Thorax: accepted.
10       Obeidi, F. D.; Eatough, D. J. (1999) Continuous measurement of semi-volatile fine particulate mass in Provo, UT.
11             J. Aerosol Sci.: submitted.
12       Oberdorster, G.; Ferin, J.; Gelein, R.; Soderholm, S. C.; Finkelstein, J. (1992) Role of the alveolar macrophage in
13             lung injury: studies with ultrafine particles. Environ. Health Perspect. 97: 193-199.
14       Oberdorster, G.; Ferin, J.; Lehnert, B. E. (1994) Correlation between particle size,  in vivo particle persistence, and
15             lung injury. Environ. Health Perspect. 102(suppl. 5): 173-179.
16       Oberdorster, G.; Gelein, R. M.; Ferin, J.; Weiss, B. (1995) Association of particulate air pollution and acute
17             mortality: involvement of ultrafine particles? In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the
18             colloquium on particulate air pollution and human mortality and morbidity;  January 1994; Irvine, CA.
19             Inhalation Toxicol. 7:111-124.
20       Oberdorster, G.; Finkelstein, J. N.; Johnston, C.; Gelein, R.; Cox, C.; Baggs, R.; Elder, A. (1999) Ultrafine particles
21             as inducers of acute lung injury: mechanisms and correlation with age and disease. Cambridge, MA: Health
22             Effects Institute: submitted.
23       Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. (1996) Gas/particle partitioning
24             and secondary organic aerosol yields. Environ. Sci. Technol. 30: 2580-2585.
25       Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Forstner, H. J. L.; Flagan,  R. C.; Seinfeld, J. H. (1997a) Aromatics,
26             reformulated gasoline, and atmospheric organic aerosol formation. Environ. Sci. Technol. 31: 1890-1897.
27       Odum, J. R.; Jungkamp, T. P. W.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J.  H. (1997b) The atmospheric
28             aerosol-forming potential of whole gasoline vapor. Science (Washington, DC) 276: 96-99.
29       Ohta, S.; Okita, T. (1990) A chemical characterization of atmospheric aerosol in Sapporo. Atmos. Environ. Part A
30             24A: 815-822.
31       Ohta, S.; Hori, M.; Yamagata, S.; Murao, N. (1998) Chemical characterization of atmospheric fine particles in
32             Sapporo with determination of water content. Atmos. Environ. 32: 1021-1025.
33       Ondov, J. M. (1996) Particulate tracers for source attribution: potential for application to California's San Joaquin
34             Valley. J. Aerosol Sci. 27(suppl. 1): S687-S688.
35       Ondov, J. M.; Kelly, W. R.; Holland, J. Z.; Lin, Z. C.; Wight, S. A. (1992)  Tracing fly ash emitted from a coal-fired
36             power plant with enriched rare-earth isotopes: an urban scale test. Atmos. Environ. Part B 26B: 453-462.
37       Ormstad, H.; Johansen, B. V.; Gaarder, P. I. (1998) Airborne house dust particles and diesel exhaust particles as
38             allergen carriers. Clin. Exp. Allergy 28: 702-708.
39       Pandis, S. N.; Paulson, S. E.; Seinfeld, J. H.; Flagan, R. C. (1991) Aerosol  formation in the photooxidation of
40             isoprene and p-pinene. Atmos. Environ. Part A 25: 997-1008.
41       Pandis, S. N.; Harley, R. A.; Cass, G. R.; Seinfeld, J. H. (1992) Secondary organic aerosol formation and transport.
42             Atmos. Environ. Part A 26: 2269-2282.
43       Pankow, J. F. (1994a) An absorption model of gas/particle partitioning of organic compounds in the atmosphere.
44             Atmos. Environ. 28:  185-188.
45       Pankow, J. F. (1994b) An absorption model of the gas/aerosol partitioning  involved in the formation of secondary
46             organic aerosol. Atmos. Environ. 28: 189-193.
47       Pankow, J. F.; Bidleman, T. F.  (1991) Effects of temperature, TSP and per cent non-exchangeable material in
48             determining the gas-particle partitioning of organic compounds. Atmos. Environ. Part A 25: 2241-2249.
49       Pankow, J. F.; Storey, J. M. E.; Yamasaki, H.  (1993) Effects of relative humidity on gas/particle partitioning of
50             semivolatile organic compounds to urban particulate matter. Environ. Sci. Technol. 27: 2220-2226.
51       Parmar, S. S.; Grosjean, D. (1990) Laboratory tests of KI and alkaline annular denuders. Atmos. Environ. Part A
52             24:2695-2698.
53       Pellizzari, E. D.; Krost, K. J. (1984) Chemical transformations during ambient air sampling for organic vapors.
54             Anal. Chem. 56: 1813-1819.
         October 1999                                    3-96          DRAFT-DO NOT QUOTE OR CITE

-------
  1       Penner, J. E.; Novakov, K. (1996) Carbonaceous particles in the atmosphere: a historical perspective to the fifth
  2             international conference on carbonaceous particles in the atmosphere. J. Geophys. Res. [Atmos.]
  3             101: 19,373-19,378.
  4       Pennise, D. M.; Kamens, R. M. (1996) Atmospheric behavior of polychlorinated dibenzo- />-dioxins and
  5             dibenzofurans and the effect of combustion temperature. Environ. Sci. Technol. 30: 2832-2842.
  6       Peters, A.; Wichmann, H. E.; Tuch, T.; Heinrich, J.; Heyder, J. (1997) Respiratory effects are associated with the
  7             number of ultrafine particles. Am. J. Respir. Crit. Care Med. 155: 1376-1383.
  8       Petzold, A.; Niessner, R. (1995) Method comparison study on soot-selective techniques. Mikrochim. Acta
  9             117:215-237.
10       Petzold, A.; Kopp, C.; Niessner, R. (1997) The dependence of the specific attenuation cross-section on black carbon
11             mass fraction and particle size. Atmos. Environ. 31: 661-672.
12       Pilinis, C.; Seinfeld, J. H.; Grosjean, D. (1989) Water content of atmospheric aerosols. Atmos. Environ.
13             23: 1601-1606.
14       Pinto, J. P.; Pleasant, M.; Kellogg, R.; Torok, Sz.; Stiles, D.; Willis, R. D.; Koutrakis, P.; Allen, G. A. (1995)
15             Chemical characterization of the ambient aerosol in Philadelphia. Presented at: Particulate matter: health and
16             regulatory issues: an A&WMA international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA:
17             Air & Waste Management Association; submitted abstract.
18       Pinto, J. P.; Stevens, R. K.; Willis, R. D.; Kellogg, R.; Mamane, Y.; Novak, J.; Santroch, J.; Benes, I.; Leni cek, J.;
19             Bures, V. (1998) Czech air quality monitoring and receptor modeling study. Environ. Sci. Technol.
20             32: 843-854.
21       Pirrone, N.; Keeler, G. J. (1996) The Rouge River watershed pollution by trace elements: atmospheric depositions
22             and emission sources. Water Sci. Technol. 33: 267-275.
23       Ranade, M. B. (Arun); Woods, M. C.; Chen, F.-L.; Purdue,  L. J.; Rehme, K. A. (1990) Wind tunnel evaluation of
24             PM10 samplers. Aerosol Sci. Technol. 13: 54-71.
25       Reist, P. C. (1984) Introduction to aerosol science. New York, NY: Macmillan Publishing Company.
26       Risse, U.; Flammenkamp, E.; Kettrup, A. (1996) Determination of aromatic hydrocarbons in air using diffusion
27             denuders. Fresenius J. Anal. Chem. 356: 390-395.
28       Roberts, R.; Corkill, J. (1998) Grass seed field smoke and its impact on respiratory health. J. Environ. Health
29             60: 10.
30       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine organic
31             aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 25: 1112-1125.
32       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
33             aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
34             Technol. 27:636-651.
35       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
36             aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks. Environ.
37             Sci. Technol. 27: 1892-1904.
38       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine organic
39             aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
40             27:2700-2711.
41       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993d) Sources of fine organic
42             aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27: 2736-2744.
43       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Sources of fine organic
44             aerosol. 6. Cigarette smoke in the urban atmosphere.  Environ. Sci. Technol. 28: 1375-1388.
45       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997a) Sources of fine organic
46             aerosol. 7. Hot asphalt roofing tar pot fumes. Environ. Sci. Technol. 31: 2726-2730.
47       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997b) Sources of fine organic
48             aerosol. 8. Boilers burning no. 2 distillate fuel oil. Environ. Sci. Technol. 31: 2731-2737.
49       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1998) Sources of fine organic
50             aerosol. 9. Pine, oak,  and synthetic log combustion in residential fireplaces. Environ. Sci. Technol. 32: 13-22.
51       Saxena, P.; Hildemann, L. M. (1996) Water-soluble organics in atmospheric particles: a critical review of the
52             literature and applications of thermodynamics to identify candidate compounds. J. Atmos. Chem. 24: 57-109.
53       Saxena, P.; Hildemann, L. M.; McMurry, P. H.; Seinfeld, J. H. (1995) Organics alter hygroscopic behavior of
54             atmospheric particles. J. Geophys. Res. [Atmos.] 100: 18,755-18,770.
         October 1999                                   3-97         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Schappi, G. F.; Taylor, P. E.; Pain, M. C. F.; Cameron, P. A.; Dent, A. W.; Staff, I. A.; Suphioglu, C. (1999)
  2             Concentrations of major grass group 5 allergens in pollen grains and atmospheric particles: implications for
  3             hay fever and allergic asthma sufferers sensitized to grass pollen allergens. Clin. Exp. Allergy 29: 633-641.
  4       Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R. (1996) Source apportionment of
  5             airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
  6       Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change.
  7             New York, NY: John Wiley & Sons, Inc.
  8       Sioutas, C.; Koutrakis, P.; Wolfson, J. (1994) Particle losses in glass honeycomb denuder samplers. Aersol Sci.
  9             Technol. 21: 137.
10       Sioutas, C.; Koutrakis, P.; Wolfson, J. M.;  Azaroff, L. S.; Mulik, J. D. (1996a) Field evaluation of a glass
11             honeycomb denuder/filter pack system to collect atmospheric gases and particles. In: Gay, B.  W.; Fuerst,
12             R. G.; Jayanty, R. K. M., eds. Measurement of toxic and related air pollutants: proceedings of a conference;
13             May, 1994; Durham, NC. Research  Triangle Park, NC: U.S. Environmental Protection Agency, Office of
14             Research and Development; report no. EPA/600/R-94/136;  pp. 426-435.
15       Sioutas, C.; Wang, P. Y.; Ferguson, S. T.; Koutrakis, P.; Mulik, J.  D. (1996b) Laboratory and field evaluation of an
16             improved glass honeycomb denuder/filter pack sampler. Atmos. Environ. 30: 885-895.
17       Sioutas, C.; Koutrakis, P.; Wang, P.-Y.; Babich, P.; Wolfson, J. M. (1999) Experimental investigation of pressure
18             drop with particle loading in Nuclepore filters. Aerosol Sci. Technol. 30: 71-83.
19       Solomon, P. A.; Salmon, L. G.; Fall, T.; Cass, G. R. (1992) Spatial and temporal distribution of atmospheric nitric
20             acid and particulate nitrate concentrations in the Los Angeles area. Environ.  Sci. Technol. 26: 1596-1601.
21       Speer, R. E.; Barnes, H. M.; Brown, R. (1997) An instrument for measuring the liquid water content of aerosols.
22             Aerosol Sci. Technol. 27: 50-61.
23       Strommen, M. R.; Kamens, R. M. (1997) Development and application of a dual-impedance radial diffusion model
24             to simulate the partitioning of semivolatile organic compounds in combustion aerosols. Environ. Sci.
25             Technol. 31:2983-2990.
26       Suarez, A. E.; Caffrey, P. F.; Ondov,  J. M.; Thaung, K. C. (1996) Size distribution of Ir-tagged soot from diesel
27             sanitation trucks. J. Aerosol Sci. 27(suppl.  1): S697-S698.
28       Tang, I. N. (1980) Deliquescence properties and particle size change of hygroscopic aerosols. In: Willeke, K., ed.
29             Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann Arbor Science
30             Publishers, Inc.; pp. 153-167.
31       Tang, H.; Lewis, E. A.; Eatough, D. J.; Burton, R. M.; Farber, R. J. (1994) Determination of the particle size
32             distribution and chemical composition of semi-volatile organic compounds in atmospheric fine particles with
33             a diffusion denuder  sampling system. Atmos. Environ. 28: 939-947.
34       Tsai, C.-J.; Cheng, Y.-H. (1996) Comparison of two ambient beta gauge PM 10 samplers. J. Air Waste Manage.
35             Assoc. 46:  142-147.
36       Tsai, C.-J.; Huang, H.-Y. (1995) Atmospheric aerosol sampling by an annular denuder system and a high-volume
37             PM10 sampler. Environ. Int. 21: 283-291.
38       Turpin, B. J.; Huntzicker, J. J. (1991) Secondary formation of organic aerosol in the Los Angeles basin: a
39             descriptive analysis  of organic and elemental carbon concentrations. Atmos.  Environ. Part A 25: 207-215.
40       Turpin, B. J.; Gary, R. A.; Huntzicker, J. J. (1990) An in situ, time-resolved analyzer for aerosol organic and
41             elemental carbon. Aerosol Sci. Technol.  12: 161-171.
42       Turpin, B. J.; Huntzicker, J. J.; Larson, S. M.; Cass, G. R. (1991) Los Angeles summer midday particulate carbon:
43             primary and secondary aerosol. Environ. Sci. Technol. 25: 1788-1793.
44       Turpin, B. J.; Liu, S.-P.; Podolske, K. S.; Gomes, M. S. P.; Eisenreich, S. J.; McMurry, P. H. (1993) Design and
45             evaluation of a novel diffusion separator for measuring gas/particle distributions of semivolatile organic
46             compounds. Environ. Sci. Technol.  27: 2441-2449.
47       Turpin, B. J.; Huntzicker, J. J.; Hering, S. V. (1994) Investigation of organic aerosol sampling artifacts in the
48             Los Angeles basin. Atmos. Environ. 28: 3061-3071.
49       U.S. Environmental Protection Agency. (1982) Air quality criteria  for particulate matter and sulfur oxides. Research
50             Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
51             Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.
52       U.S. Environmental Protection Agency. (1996) Air quality criteria  for particulate matter.  Research Triangle Park,
53             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
54             Available from: NTIS, Springfield, VA; PB96-168224.
         October 1999                                   3-98         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Underbill, D. W. (1999) Aerosol synergism and hydrate formation - a possible connection. Inhalation Toxicol.:
  2             submitted.
  3       Van Vaeck, L.; Van Cauwenberghe, K.; Janssens, J. (1984) The gas-particle distribution of organic aerosol
  4             constituents: measurements of the volatilisation artefact in hi-vol cascade impactor sampling. Atmos.
  5             Environ.  18:417-430.
  6       Vega, E.; Garcia, L; Apam, D.; Ruiz, M.; Barbiaux, M. (1997) Application of a chemical mass balance receptor
  7             model to respirable particulater matter in Mexico City.  J. Air Waste Manage. Assoc. 47: 524.
  8       Veltkamp, P. R.; Hansen, K. J.; Barkley, R. M.; Sievers, R. E. (1996) Principal component analysis of summertime
  9             organic aerosols at Niwot Ridge, Colorado. J. Geophys. Res. [Atmos.]  101: 19,495-19,504.
10       Venkataraman, C.; Lyons, J. M.; Friedlander, S. K. (1994) Size distributions of polycyclic aromatic hydrocarbons
11             and elemental carbon. 1. Sampling, measurement methods, and source characterization. Environ. Sci.
12             Technol. 28: 555-562.
13       Wang, P. Y. (1997) Continuous aerosol mass measurement by flow obstruction.  Boston, MA: Harvard University,
14             School of Public Health.
15       Wang, H.-C.; John, W. (1988) Characteristics of the Berner impactor for sampling inorganic ions. Aerosol Sci.
16             Technol. 8: 157-172.
17       Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D.  R.; Ashbaugh, L. L. (1994)
18             Chemical mass balance source apportionment of PM 10 during the Southern California Air Quality Study.
19             Aerosol Sci. Technol. 21: 1-36.
20       Wedding, J. B.; Weigand, M. A.  (1993) An automatic particle sampler with beta gauging. Air Waste 43: 475-479.
21       Weingartner, E.; Burtscher, H.; Baltensperger, U. (1997) Hygroscopic properties of carbon and diesel soot particles.
22             Atmos. Environ. 31:2311 -2327.
23       Weschler, C. J.; Shields, H. C. (1999)  Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
24             Environ.  33:2301-2312.
25       Wexler, A. S.; Sarangapani, R. (1998) Particles do not increase vapor deposition in human airways. J. Aerosol Sci.
26             29: 197-204.
27       Whitby, K. T. (1978) The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135-159.
28       Whitby, K. T.; Sverdrup, G. M. (1980) California aerosols: their physical and chemical characteristics. In: Hidy,
29             G. M.; Mueller, P. K.; Grosjean, D.; Appel, B. R.; Wesolowski, J. J., eds. The character and origins of smog
30             aerosols: a digest of results from the California Aerosol Characterization  Experiment (ACHEX). New York,
31             NY: John Wiley & Sons, Inc.; pp. 477-517. (Advances in environmental science and technology: v. 9).
32       Whitby, K. T.; Charlson, R. E.; Wilson, W. E.;  Stevens, R. K. (1974) The size of suspended particle matter in air.
33             Science (Washington, DC) 183: 1098-1099.
34       Willeke, K.; Whitby, K. T. (1975) Atmosperic aerosols: size distribution interpretation. J. Air Pollut. Control Assoc.
35             25:529-534.
36       Williams, E. L., II; Grosjean, D.  (1990) Removal of atmospheric oxidants with annular denuders. Environ. Sci.
37             Technol. 24: 811-814.
38       Wilson, W. E. (1995) Aerosol exposure, physics, and chemistry. Inhalation Toxicol. 7: 769-772.
39       Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
40             epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
41       Wilson, W. E.; Spiller, L. L.; Ellestad, T. G.; Lamothe, P. J.; Dzubay, T. G.; Stevens, R. K.; Macias, E. S.; Fletcher,
42             R. A.; Husar, J. D.; Husar, R. B.; Whitby, K. T.; Kittelson, D. B.; Cantrell, B. K.  (1977) General Motors
43             sulfate dispersion experiment: summary of EPA measurements. J. Air Pollut. Control Assoc. 27: 46-51.
44       Witz, S.; Eden, R. W.; Wadley, M. W.; Dunwoody, C.; Papa, R.; Torre, K. J. (1990) Rapid loss of particulate
45             nitrate, chloride and ammonium on quartz fiber filters during storage. J. Air Waste Manage. Assoc.
46             40:53-61.
47       Wu, C. C.; Suarez, A. E.; Lin, Z.; Kidwell, C. B.; Borgoul, P.  V.; Caffrey, P. F.; Ondov, J. M.; Sattler, B.  (1998)
48             Application of an Ir tracer to determine soot exposure to students commuting to school on Baltimore public
49             buses. Atmos. Environ. 32: 1911-1919.
50       Yamasaki, H.; Kuwata, K.; Miyamoto, H. (1982) Effects of ambient temperature on aspects of airborne polycyclic
51             aromatic hydrocarbons. Environ. Sci. Technol. 16: 189-194.
52       Yatin, M.; Tuncel, S. G.; Tuncel, G.; Aras, N. K. (1994) Trace element composition of atmospheric aerosols in
53             Ankara, Turkey, determined by instrumental neutron activation analysis. J. Radioanal. Nucl. Chem.
54             181:401-411.
         October 1999                                   3-99         DRAFT-DO NOT QUOTE OR CITE

-------
1       Zhang, X. Q.; McMurry, P. H. (1987) Theoretical analysis of evaporative losses from impactor and filter deposits.
2            Atmos. Environ. 21: 1779-1789.
3       Zhang, X.; McMurry, P. H. (1992) Evaporative losses of fine particulate nitrates during sampling. Atmos. Environ.
4            Part A 26: 3305-3312.
5       Zhang, S.-H.; Shaw, M.; Seinfeld, J. H.; Flagan, R. C. (1992) Photochemical aerosol formation from  cc-pinene and
6            p-pinene. J. Geophys. Res. [Atmos.] 97: 20,717-20,729.
7       Zhang, X. Q.; McMurry, P. H.; Hering, S. V.; Casuccio, G. S. (1993) Mixing characteristics and water content of
8            submicron aerosols measured in Los Angeles and at the Grand Canyon.  Atmos. Environ. Part A
9            27: 1593-1607.
        October 1999                                 3-100         DRAFT-DO NOT QUOTE OR CITE

-------
 i        4.  CONCENTRATIONS, SOURCES, AND EMISSIONS
 2                       OF ATMOSPHERIC PARTICLES
 3
 4
 5     4.1  INTRODUCTION
 6          The present chapter incorporates material from Chapters 5 (Sources and Emissions of
 7     Atmospheric Particles) and Chapter 6 (Environmental Concentrations) of the previous document,
 8     Air Quality Criteria (CD96) for Particulate Matter (U.S. Environmental Protection Agency,
 9     1996) and presents updates to these materials where available.
10          Information on ambient concentrations of particles in various size ranges (PM10, PM25,
11     PM10_2 5) and their chemical composition, based on specific field studies, is presented in
12     Section 4.2. The results of field studies will be used to characterize the spatial and temporal
13     variability in PM and its components in selected urban areas in geographically diverse regions of
14     the United States as they become available.
15          Unlike gaseous criteria pollutants (SO2, NO2, CO, O3), which are well defined chemical
16     entities, atmospheric particulate matter (PM) is composed of a variety of particles differing in
17     size and chemical composition. Therefore, sources of each component of the atmospheric
18     aerosol must be considered in turn. Differences in the composition of particles emitted by
19     different sources also will lead to spatial and temporal heterogeneity in the composition of the
20     atmospheric aerosol. The nature of the sources and the composition of the emissions from these
21     sources are discussed in Section 4.3. Since PM is composed of both primary and secondary
22     constituents, emissions of both the primary components and the gaseous precursors of secondary
23     PM must be considered. Nationwide emissions estimates of primary PM and precursors to
24     secondary PM are discussed in Section 4.4. Estimates of contributions of various sources to
25     ambient PM levels given by source apportionment studies are presented in Section 4.5.
26     Information about the composition of emissions from various sources is given in Appendix A.
27
28
29
30

       October 1999                             4-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1     4.2  TRENDS AND PATTERNS IN AMBIENT PARTICULATE MATTER
 2          PM2 5 CONCENTRATIONS AND TRENDS
 3          A significant amount of data for characterizing PM10 mass concentrations and trends exists
 4     and that available up to about 1994 was presented in CD96. However, data sets for
 5     characterizing PM2 5 and PM(10_2 5) mass or trends are not as extensive. Results from the small
 6     number of recent studies in which daily mass and composition measurements are available for
 7     extended periods will be discussed in this section. Sources of data on PM25 (fine) and PM(10_25)
 8     (coarse), which were discussed in CD96, include EPA's Aerometric Information Retrieval
 9     System (AIRS) (Aerometric Information Retrieval System, 1995), IMPROVE (Eldred and Cahill,
10     1994; Cahill, 1996), The California Air Resources Board (CARB) (California Air Resources
11     Board, 1995), the Harvard Six-Cities Data Base (Spengler et al, 1986; Neas, 1996), and the
12     Harvard Philadelphia Data Base (Koutrakis, 1995). The Inhalable Particulate Network (IPN)
13     (Inhalable Particulate Network, 1985; Rodes and Evans, 1985) provided TSP, PM15 and PM25
14     data with only a small amount of PM10 data.
15          Summary tables giving the results of field studies which obtained data for the composition
16     of particles in the PM2 5, PM(10_2 5), or PM10 size ranges were presented in Appendix A
17     to Chapter 6 of CD96. The summary tables included data for mass, organic carbon, elemental
18     carbon, nitrate, sulfate, and trace elements. The results of sixty six studies were separated and
19     presented for the eastern, western and central United States. The  data for the broad
20     compositional categories given above from these studies are summarized in pie-chart form in
21     Figures 4-la, 4-lb, and 4-lc. The mean ratio of PM(10_25) to PM25 from these studies was  0.33 in
22     the East, 0.92 in the central United States, and 0.89 in the West.
23
24     PM10 Concentrations and Trends
25          Nationwide PM10 levels declined by 20% between 1988 and 1994 (U.S. Environmental
26     Protection Agency, 1996). The United States was divided into seven regions and on a regional
27     basis, reductions ranged from 17% in the Southeast to 33% in the Southwest. The estimated
28     ratio of PM25 to PM10 ranged from 0.38 to 0.70 in the seven regions considered. Darlington et al.
29     (1997) extended this analysis to include 1995.  Their analysis indicated a nationwide average
30     PM10 level of 34 //g/m3 in  1988 declining to 26 //g/m3 in 1995, corresponding to an overall
       October 1999                             4-2        DRAFT-DO NOT QUOTE OR CITE

-------
                       PM2.5 Components
                                         Minerals 4.3%
        Unknown 22.8%
          EC 3.9%
        OCx 1.4 20.9%

                    NO; 1.1%
                                                    so; 34.1%
                     (N 1-0*13%

Nitrate based on 3 studies
               Coarse PM(2.5-10) Components
    Unknown 41.5%
                                                    Minerals 51.8%
              (NhQ* 1.8%
                                       SO4" 4.9%
                Insufficient Nitrate, OC, and EC data available
Figure 4-1 a.  Major constituents of particles measured at sites in the eastern United States.
           (NH4+)* represents the concentration of NH4+ that would be required if all
           SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore, (NH4+)*
           represents an upper limit to the true concentration of NH4+.
Source: U.S. Environmental Protection Agency (1996).

October 1999                        4-3
                DRAFT-DO NOT QUOTE OR CITE

-------
                        PM2.5 Components
                    EC 9.0%
   Minerals 7.6%
     OC x 1.4 44.6%
                                                      so:  22.3%
                                                   (NH4+)*10.2%
                                              NO; 8.1%

                        Reconstructed sum = 124.8%
                Coarse PM(2.5-10) Components
     Unknown 33.0%
          so 3.1%
                                                     Minerals 62.8%
               Insufficient Nitrate, OC, and EC data available
 Figure 4-lb. Major constituents of particles measured at sites in the central United States
           (NH4+)* represents the concentration of NH4+ that would be required if all
           SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore, (NH4+)*
           represents an upper limit to the true concentration of NH4+.
 Source: U.S. Environmental Protection Agency (1996).

October 1999                        4-4
DRAFT-DO NOT QUOTE OR CITE

-------
                           PM2.5 Components
                 EC 14.7%
        OCx 1.4 38.9%
               Minerals 14.6%
                                                        so;  10.8%
                                                        (NhC)* 7.5%
                                                     NO3"15.7%
                            Reconstructed sum = 102.2%
                     Coarse PM(2.5-10) Components
          Unknown 27%
       (N HI)* 0.8%
         so; 3.1%
                                                      Minerals 69.9%
                     Insufficient Nitrate, OC, and EC data available
Figure 4-lc. Major constituents of particles measured at sites in the western United
           States. (NH4+)* represents the concentration of NH4+ that would be required
           if all SO4= were present as (NH4)2SO4 and all NO3 as NH4NO3. Therefore,
           (NH4+)* represents an upper limit to the true concentration of NH4+.

Source: U.S. Environmental Protection Agency (1996).
October 1999
4-5
DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
reduction of 24% or 3.4% per year. They also found that the 95th percentile value declined from
69 //g/m3 in 1988 to 52 //g/m3 in 1995.
     PM25 data have been collected continuously since 1994 as part of the children's health
study in twelve communities in southern California (Taylor et al., 1998). Data obtained at all
sites show decreases ranging from 2% at Santa Maria to 37% at San Dimas/ Glendora in PM2 5
from 1994 through 1998. These decrease were accompanied by decreases in major components
such as nitrate, sulfate, ammonium, and acids. However, undefined components showed a mixed
pattern of increases and decreases at the same sites. Information regarding PM2 5 levels mainly in
non-urban areas across the United  States is shown in Figure 4-2. Sufficient data are not yet
available to permit the calculation  of nationwide trends or average levels of PM25 or PM(10_2 5),
however some general conclusions can be reached. Darlington et al. (1997) proposed that since
the consistent reductions in PM10 levels were found in a wide variety of environments ranging
from urban to rural over large areas, that common factors or controls might be responsible and
that these factors affected fine particles more strongly than coarse particles because fine particles
can be transported over longer distances.
                                      - «»«J''
                              i ,-  *V "-"V. '  4
                             •' .'  , V  • •'(>,  »*
                            !>  I', •  '-"<<> '  V,«"
                                            «
       Figure 4-2.  Annual average PM25 concentration (1994-96).
       Source: CAPITA, 1999.
       (Available http://llcapita.wuste.edu/CAPITA/Capita Reports/Capita Activities 98-99/CAPITA Activities 98 99/)
        October 1999
                                           4-6
DRAFT-DO NOT QUOTE OR CITE

-------
 1           A few attempts to infer "background" levels of PM25 and PM10 have been made. The
 2      background levels most relevant to the present criteria document include (1) a "natural
 3      background" which excludes all anthropogenic sources anywhere in the world, and (2) a "natural
 4      and uncontrollable anthropogenic background" which includes anthropogenic sources outside of
 5      North America in addition to (1). Annual average natural background levels (1) of PM10 have
 6      been estimated to range from 4 to 8 //g/m3 in the western United States and 5 to 11 //g/m3 in the
 7      eastern United States. Corresponding PM25 levels have been estimated to range from 1 to
 8      4 //g/m3 in the western United States and from 2 to 5 //g/m3 in the eastern United States (U.S.
 9      Environmental Protection Agency, 1996). Twenty-four hour average natural background
10      concentrations may be substantially higher than the annual or seasonal average natural
11      background concentrations. Estimates of levels for background (2) are not yet available.
12           Data for characterizing the daily and seasonal variability of PM25 mass and composition
13      will be discussed in 4.2.1, the interrelations and correlations among the various PM components
14      and parameters will be discussed in 4.2.2, the spatial variability of various PM components will
15      be discussed in 4.2.3, new data on PM2 5 mass from EPA compliance network will be discussed
16      in 4.2.4, and trends  of PM2 5 mass in rural air will be discussed in 4.2.5.
17
18      4.2.1 Daily and Seasonal Variability
19           Information, useful for relating ambient concentrations to health effects, can be obtained by
20      analyzing long time series of concentration data at a single site. Collocated 24-hour  PM25 and
21      PM10 filter samples  were collected at a site in southwestern Philadelphia, PA from May 1992
22      through April 1995  (Koutrakis,  1995).  The PM2 5 filters were analyzed by X-ray fluorescence
23      (XRF) spectroscopy at EPA for the three years. This unique data set was collected on a nearly
24      daily basis, thereby allowing an assessment of day-to-day variability in aerosol concentrations
25      and relationships. EPA has also obtained similar data bases for at least one year from Phoenix,
26      AZ and Baltimore, MD.  PM2 5 and PM(10_2 5) mass data were also collected at a number of sites in
27      California (California Air Resources Board, 1995). The data collected at Philadelphia and at
28      Riverside-Rubidoux, CA were presented in CD96 and are shown here for completeness.
29      In addition, the compositional data obtained at the Philadelphia and Phoenix sites will be
30      presented in  Section 4.2.2 and spatial relations among PM  components in Philadelphia will be
31      presented in  Section 4.2.3.
        October 1999                              4-7         DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
4
5
6
7
     The data for Philadelphia are presented as box plots showing the lowest, lowest tenth
percentile, lowest quartile, median, highest quartile, highest tenth percentile, and highest PM2 5
values in Figure 4-3. The four three-month averaging periods shown (March-May, June-August,
September-November, December-February) correspond to the so-called climatological or
meteorological seasons.
70-
— 60~
a 5°-
o
o3 40-
8
§ 30-
o
20-
10-
n -











I
i





1
»
i










r
[
t
}
Philadelphia - PBY site
r PM2.5
(n = 1 024)



h
I
A /
j • 4
[ V ^



i
»
t
                         Mar- May
                                  June - Aug
Sep - Nov
Dec - Feb
      Figure 4-3.  Concentrations of PM2 5 measured at the PBY site in southwestern
                  Philadelphia.  The data show the lowest, lowest tenth percentile, lowest
                  quartile, median (black circles), highest quartile, highest tenth percentile, and
                  highest PM2 5 values.
1          Frequency distributions for PM2 5 are shown in Figure 4-4 for Philadelphia. Concentrations
2     predicted from the log-normal distribution, using mean values and geometric standard deviation
3     derived from the data, are also shown. Frequency distributions of particle concentrations at
4     several sites in the South Coast Air Basin (Kao and Friedlander, 1995) have also been shown to
5     be approximated reasonably well by log-normal distributions.
      October 1999
                                          4-8
   DRAFT-DO NOT QUOTE OR CITE

-------
                                              geometric mean = 15.2 |jg/m
                                                                og=  1.69
                         |LJ|LJ|LJ|LJ|LJ|LJ|LJ|
                            10    20     30    40     50      60
                                       Concentration (|jg/m3)
                              70     80
      Figure 4-4.  Frequency distribution of PM2 5 concentrations measured at the PBY site in
                  southwestern Philadelphia. Log-normal distribution fit to the data shown as
                  solid line.
1          In Philadelphia, the highest PM2 5 values were observed when winds were from the
2     southwest during sunny but hazy high pressure conditions. In contrast, the lowest values are
3     found after significant rainstorms during all seasons of the year.  Day-to-day concentration
4     differences in the data set are 6.8 ± 6.5 //g/m3 for PM2 5 and 8.6 ± 7.5 //g/m3 for PM10. Maximum
5     day-to-day concentration differences are 54.7 //g/m3 for PM2 5 and 50.4 //g/m3 for PM10.
6          Different conclusions could be drawn about data collected  elsewhere in the United States.
7     PM2 5 concentrations obtained in Phoenix, AZ are summarized in Figure 4-5 and frequency
8     distributions of PM25 concentrations obtained in Phoenix are shown in Figure 4-6. Day-to-day
9     concentration differences in this data set are 2.9 ±3.0 //g/m3  with a maximum day-to-day
      October 1999
4-9
DRAFT-DO NOT QUOTE OR CITE

-------
tu •

^ 30-
~B)
o 20-
2
c
Q)
c 10-
o
O
n -
Phoenix, AZ T















PM2.5


(n = 876)







y

















\\
_

T










1


0























T ?

                       Mar - May    June-Aug      Sep-Nov     Dec-Feb
      Figure 4-5.   Concentrations of PM2 5 measured at the EPA site in Phoenix, AZ. The data
                   show the lowest, lowest tenth percentile, lowest quartile, median (black
                   circles), highest quartile, highest tenth percentile, and highest PM2 5 values.
 1     concentration difference of 23 //g/m3. PM25 and PM(10_25) data were obtained at a number of sites
 2     in California on a sampling schedule of every six days with dichotomous samplers (California
 3     Air Resources Board, 1995).  Data for PM25 are summarized in box plot form in Figure 4-7.
 4     The frequency distribution of PM25 concentrations obtained at Riverside-Rubidoux from 1989
 5     to 1994 is shown in Figure 4-8. It can be seen that the data are not as well fit by a log normal
 6     distribution as can the data shown in Figures 4-4 and 4-6, mainly as the result of a significant
 7     number of days with PM25 >100 //g/m3 (Figure 4-7).
 8          An examination of the data from Philadelphia, PA, Phoenix, AZ, and Riverside, CA
 9     indicates that substantial differences exist in aerosol properties between widely separated
10     geographic regions. Fine mode particles make up most of the PM10 mass observed in
11     Philadelphia and appear to drive the daily and seasonal variability in PM10 concentrations there.
       October 1999
4-10
DRAFT-DO NOT QUOTE OR CITE

-------
            200
                                                                PM
                                                                   2.5
                                                   geometric mean = 10.5 |jg/m3
                                                              afl = 1.70
                                  10      15     20     25      30     35     40
                                     Concentration  (|jg/nn3)
      Figure 4-6.  Frequency distribution of PM2 5 concentrations measured at the EPA site in
                  Phoenix, AZ.
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
Coarse mode particles represent a larger fraction of PM10 mass in Phoenix and Riverside and
drive the seasonal variability in PM10 seen there. The ratio of PM25 to PM10 mass is much larger
in Philadelphia (0.72) than in either Phoenix (0.34) or Riverside (0.49).  Differences between
median and maximum concentrations in any size fraction are much larger at the Riverside site
than at either the Philadelphia or Phoenix sites. Many of these differences could reflect the more
sporadic nature of dust suspension at Riverside. These considerations demonstrate the hazards in
extrapolating conclusions about the nature of variability in aerosol  characteristics inferred at one
location to another.
       October 1999
                                        4-11
DRAFT-DO NOT QUOTE OR CITE

-------
                o
               O
140-
120-
100-
60-
40-
20-
o-
Fine
(n = 382)



•


r
•
•


ft 1 i
i ii
• T-T V
1 1 1
V I i Y
i i i i
Jan - Mar Apr - Jun Jul - Sept Oct - Dec
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr
        Figure 4-7.  Concentrations of PM25 measured at the Riverside-Rubidoux site. The data
                    show the lowest, lowest tenth percentile, lowest quartile, median (black
                    squares), highest quartile, highest tenth percentile, and highest PM2 5 values.
 1     4.2.2  Relations Between Mass and Chemical Component Concentrations
 2          Time series of elemental composition data for PM2 5 particles have been obtained at a few
 3     locations across the United States. The filter samples that were collected at the PBY site in
 4     southwestern Philadelphia and were used in the construction of Figures 4-3 and 4-4 were also
 5     analyzed by X-ray fluorescence. Concentrations of the trace elements and correlations between
 6     trace elements and the total mass of particles in the PM2 5 size range are shown in Table 4-1.
 7     Also shown in Table 4-1 are similar results obtained for filter samples collected in Phoenix, AZ.
 8     Filters from both monitoring studies were analyzed by the same X-ray spectrometer at the U.S.
 9     EPA facility in Research Triangle Park, NC. As can be seen from inspection of Table 4-1, the
10     analytical uncertainty (given in parentheses next to concentrations) as a fraction of the absolute
11     concentration is highly variable and it exceeds the concentration for a number of trace metals
12     whose absolute concentrations are low, while it is very small for abundant elements such as S.
13
       October 1999
4-12
DRAFT-DO NOT QUOTE OR CITE

-------
          1OO
           8O-
           6O-
           4O-
           2O-
                                                               PM
                                                                   2.5
                                                geometric mean = 26.6 |jg/m3
                                                            cjg = 2.10
                O      2O    4O     6O    8O    1OO    12O   14O    16O   18O   2OO
        Figure 4-8.  Frequency distribution of PM25 concentrations measured at the Riverside-
                    Rubidoux site.
 1           There are a number of distinct differences between the two data sets.  For instance, sulfate
 2     and associated cations and water appear to constitute a major fraction of the composition of the
 3     PM in the Philadelphia data set while they appear to constitute a much smaller fraction of the
 4     Phoenix data set. The highest PM2 5 values were observed in Philadelphia during episodes driven
 5     by high sulfate abundances and are due, at least partly, to higher sulfate concentrations.
 6     Correlation coefficients between SO^ and PM25 were 0.97 during the summer of 1993. Similar
 7     correlations between SO^  and PM2 5 were found at a site in northeastern Philadelphia (24 km
 8     distant from the site under discussion) during the summer of 1993.
 9           Concentrations of "crustal elements" (e.g., Al, Si, K, Ca, Ti, Mn, Fe) are higher relative to
10     PM25 mass in the Phoenix data set compared to the Philadelphia data set. Sulfur is very highly
11     correlated with PM2 5 (r = 0.92) in the Philadelphia data set while it is only weakly correlated
12     (r = 0.16) with PM25 in the Phoenix data set. Toxic trace metals e.g., Cr, Co, Ni, Cu, Zn, As, and
13     Pb are not well correlated ( 0.04 < r < 0.25) with PM25 in the Philadelphia data set while they are
14     more variably correlated (0.01 < r < 0.69) with PM25 in the Phoenix data set.  The uncertainty in
15     the concentration measurement most probably plays a role in determining a species' correlation
       October 1999
4-13
DRAFT-DO NOT QUOTE OR CITE

-------
  TABLE 4-1. CONCENTRATIONS OF PM25 (//g/m3) AND SELECTED ELEMENTS
 (ng/m3) IN THE PM25 SIZE RANGE AND CORRELATIONS BETWEEN ELEMENTS
    AND PM25 MASS. VALUES IN PARENTHESES REFER TO ANALYTICAL
       UNCERTAINTY IN X-RAY FLUORESCENCE DETERMINATIONS

PM25
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
As
Se
Br
Pb
Philadelphia, PA1
Concentration
17.0 ± 0.8 Mg/m3
4.0 (3 1.1) ng/m3
116(21.1)
8.6(10.3)
2100(143)
5.1 (3.4)
60.4 (4.7)
46.6 (4.2)
4.9(4.1)
8.8(1.8)
0.7 (0.7)
3.1(0.8)
109(10.5)
0.1 (1.4)
7.3 (1.4)
4.8(1.1)
36.9(3.7)
0.6(1.2)
1.5 (0.6)
5.0 (0.9)
17.6(2.5)

r
1
0.10
0.51
0.31
0.92
-0.01
0.50
0.39
0.44
0.37
0.15
0.39
0.50
0.04
0.22
0.25
0.21
0.18
0.63
0.11
0.19
Phoenix, AZ2
Concentration
9.4 ± 0.5 //g/m3
68.9 (27.2) ng/m3
209 (48.4)
7.6 (4.5)
408 (30.9)
11.4(2.4)
78.6 (8.2)
76.5 (9.7)
7.2(3.3)
0.7(1.0)
0.4 (0.4)
4.3 (0.6)
112(15.1)
-0.2 (0.8)
0.4 (0.4)
3.3 (0.7)
12.7(1.7)
1.3 (0.6)
0.3 (0.3)
3.1 (0.6)
4.5(1.0)

r
1
0.23
0.35
0.52
0.16
0.13
0.67
0.51
0.44
-0.28
0.41
0.64
0.80
-0.01
0.38
0.69
0.64
0.50
0.40
0.57
0.69
'n= 1105
2n = 643
October 1999
4-14
DRAFT-DO NOT QUOTE OR CITE

-------
 1      with PM2 5, especially when the analytical uncertainty is high relative to concentration, e.g., for
 2      trace metals such as Co.
 3
 4      4.2.3  Spatial Variability
 5           Three methods for comparing the chemical composition of aerosol databases obtained at
 6      different locations and times were discussed by Wongphatarakul et al. (1998). Log - log plots of
 7      chemical concentrations obtained at pairs of sampling sites accompanied by the coefficient of
 8      divergence (COD) were examined as a way to provide an easily visualized means of comparing
 9      two data sets1. Examples comparing downtown Los Angeles with Burbank and with Riverside
10      Rubidoux are shown in Figures 4-9 and 4-10, respectively. As the composition of two sampling
11      sites become more similar, the COD approaches zero, as their compositions diverge, the COD
12      approaches one. Cluster analyses based on the COD between individual data sets can be used to
13      determine the degree of similarity among a number of data sets. Correlation coefficients
14      calculated between components can be used to show the degree of similarity between pairs of
15      sampling sites.  In addition to  calculating correlation coefficients for total  mass or for individual
16      components, correlation coefficients for characterizing the spatial variation of the contributions
17      from given source types can also be calculated by averaging the correlation coefficients of the set
18      of chemical components that represent the source type.  The first two methods could be applied
19      either to aerosol data sets  collected at multiple sites within a given geographic region or to
20      aerosol data sets collected at widely different locations or times while the  third method is best
21      used to characterize sites within a particular  geographic region.
22           Correlation coefficients  showing the spatial relations among PM2 5 and contributions from
23      different source categories obtained at various sites in the South Coast Air Basin (SoCAB) Study
24      are shown in Table 4-2. In Wongphatarakul et al. (1998), crustal material (crust), motor vehicle
               The COD for two sampling sites is defined as follows:
                                                        j    X-:   2
        where x;j represents the average concentration for a chemical component i at site j, j and k represent two sampling
        sites and p is the number of chemical components.
        October 1999                              4-15        DRAFT-DO NOT QUOTE OR CITE

-------
       102
   <£ 10°
    c
    CO
    I  1C'1

    QQ
       io-2-
       10
          -3
                  COD=0.099
                                                  so,
                                                     2-
                                                             nknown
                                                               NO;
                                               Source: \Afongphatarakul (1997)
            10"
  \            I           I
1C'2         1C'1         10°

  Downtown  Los Angeles (|jg/m3)
                                  102
      *Jan. 2 - Dec. 28,1986 (63 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 |jm

      **Jan. 2 - Dec. 28,1986 (61 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 urn
Figure 4-9.  PM25 chemical components in downtown Los Angeles and Burbank (1986)*

           have similar characteristics.


Source: Wongphatarakul (1997).
October 1999
4-16
                         DRAFT-DO NOT QUOTE OR CITE

-------
    x
    13
    O
    T3
    J2
    13
    o:
       102
    E  10°-
       10"-
       10
          -3
                  COD=0.230
            10
              ,-3
                                                      Unknown
Gas Al
                                             Zn
                                               Source: Wongphatarakul(1997)
 T
10':
10'
               I

              10°
                       102
                          Downtown Los Angeles (pg/m3)
      *Jan. 2 - Dec. 28,1986 (63 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 urn

      **Jan. 2 - Dec. 28,1986 (60 data points), 24 hours sampling, sampling every 6 days, dp < 2.5 pm
Figure 4-10.  Concentrations of PM25 chemical components in Rubidoux and downtown

            Los Angeles (1986)*.  The diagram shows a significant spread in the

            concentrations for the two sites compared with downtown Los Angeles and

            Burbank (Figure 7).


Source: Wongphatarakul (1997).
October 1999
    4-17
DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 4-2. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
              PM2 5 MASS AND DIFFERENT SOURCES FOR PAIRS OF SAMPLING
                                     SITES IN SoCAB (1986)
R™ Rcmstal Rsec R™ Rresidmloll
Hawthorne and Rubidoux
Long Beach and Rubidoux
Anaheim and Rubidoux
Downtown Los Angeles and Rubidoux
Burbank and Rubidoux
Hawthorne and Anaheim
Long Beach and Anaheim
Burbank and Anaheim
Downtown Los Angeles and Anaheim
Downtown Los Angeles and Hawthorne
Burbank and Hawthorne
Long Beach and Burbank
Long Beach and Hawthorne
Downtown Long Angeles and Long Beach
Downtown Los Angeles and Burbank
-0.027
0.051
0.066
0.095
0.120
0.760
0.852
0.770
0.827
0.808
0.704
0.731
0.880
0.842
0.928


0.034
0.075
0.105
0.143
0.568
0.599
0.633
0.649
0.653
0.825


0.768
0.888
0.749
0.804
0.854
0.790
0.737
0.909
0.817
0.960


0.492
0.504
0.579
0.556
0.669
0.688
0.714
0.861
0.719
0.871


0.170
0.150
0.161
0.233
0.533
0.491
0.295
0.482
0.378
0.606
       Source: Wongphatarakul et al. (1998).
1     exhaust (mv), residual oil emissions (residual oil) and secondary PM (sec) were considered as
2     source categories. Al, Si, Fe and Ca were used as markers for crustal material (crustal). V and
3     Ni were used as markers for fuel oil combustion (residual oil). Pb, Br and Mn were used as
4     markers for motor vehicle exhaust (MV). NO3", NH4+ and SO4= represent secondary PM
5     components (sec). The average of the correlation coefficients of marker elements within each
6     source category are shown in Table 4-2. Total PM25 (tot) varies in a similar fashion among
7     Friedlander, 1995). Values of Rsec and Rmv are much higher than those for Rcmstal and
8     throughout the SoCAB suggesting a more uniform distribution of the contributions from
      October 1999
4-18
DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
secondary PM formation and automobiles than from crustal material and localized stationary
sources.
     In the Philadelphia area PM25 was found to be strongly correlated (r > 0.9) between seven
urban sites and one background site (Valley Forge, PA) during the summer of 1993 (Suh et al.,
1995). The same relations were also found during the summer of 1994 at four monitoring sites
as part of a separate study (Pinto et al., 1995). The results from these studies strongly suggest
that PM2 5 and SO^ concentrations are spatially uniform throughout the Philadelphia area, and
that variability in PM10 levels is caused largely by variability in PM2 5 (Wilson and Suh, 1997).
However, not enough data are available from regional sites to define the total areal extent of the
spatial homogeneity in PM2 5 and SO4= concentrations observed in the urban concentrations SO4=.
Correlation coefficients in Philadelphia air for PM2 5, crustal components (Al, Si, Ca, Fe), the
major secondary component (sulfate), organic carbon (OC) and elemental carbon (EC) are shown
in Table 4-2a. Because these data were obtained after Pb had been phased out of gasoline, a
motor vehicle contribution could not be extracted.
          TABLE 4-2a. CORRELATION COEFFICIENTS FOR SPATIAL VARIATION OF
             PM2 5 MASS AND DIFFERENT COMPONENTS FOR PAIRS OF SAMPLING
                                 SITES IN PHILADELPHIA (1994)

Castor Ave. and Roxboro
Castor Ave. and NE Airport
Castor Ave. and Broad St.
Roxboro and NE Airport
Roxboro and Broad St.
NE Airport and Broad St.
Rtot
0.92
0.93
0.93
0.98
0.95
0.95
15
^crustal
0.52
0.47
0.57
0.67
0.90
0.69
Rsec
0.98
0.99
0.99
0.98
0.98
0.99
ROC
0.88
0.88
0.85
0.83
0.86
0.84
Rec
0.84
0.77
0.89
0.82
0.79
0.63
        Source: Pinto etal. (1995).
       October 1999
                                         4-19
DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.2.4 Urban Concentrations and Patterns from the New PM2 5 Compliance
 2            Network
 3      4.2.5 Trends and Patterns
 4      4.2.5.1  Visual Range/Haziness
 5           Observations of visual range, obtained by the National Weather Service and available
 6      through the National Climatic Data Center of the National Oceanic and Atmospheric
 7      Administration, provide one of the few truly long-term, daily records of any parameter related to
 8      air pollution. After some manipulation, the visual range data can be used as an indicator of fine
 9      mode particle pollution. The data reduction process and analyses of resulting trends have been
10      reported by Husar et al. (1994), Husar and Wilson (1993), and Husar et al. (1981).
11           Visual range i.e., the maximum distance at which an observer can discern the outline of an
12      object, is an understandable and for many purposes an appropriate measure of the optical
13      environment. It has the disadvantage, however, of being related inversely to aerosol
14      concentration. It is usual, therefore, to convert visual range to a direct indicator of fine mode.
15      particle concentration.  The quantitative measure of haziness is the extinction coefficient, Bext,
16      defined as Bext=K/visual range, where K is the Koschmieder constant. The value of K is
17      determined both by the threshold sensitivity of the human eye and the initial contrast of the
18      visible object against the horizon sky. Husar et al. (1994) use K=1.9 in accordance with the data
19      given by Griffing (1980). The extinction coefficient is in units of km"1  and is proportional to the
20      concentration of light scattering and absorbing aerosols and gases. The radiative transfer
21      characteristics which determine the visual range depend on time of day. Only local noon
22      observations are used.
23
24      Haze Trend Summary
25           The U.S. haze patterns and trends from 1960 to 1992 were presented in the CD96  as
26      16 haze maps representing four time periods (5 year averages centered  at 1960, 1970, 1980, and
27      1990) and the four climatological seasons (Quarter 1 is for December of prior year, January and
28      February of indicated year, etc.) seasons.  The average haze patterns, centered on 5 year averages
29      for 1980 and 1990 are shown in Figure 4-11 and will be replaced by similar data for 1990 and
30      1995. Haze is indicated by the 75th percentile of the extinction coefficient which is calculated
31      from the visual range, corrected to 60% relative humidity by the Koschmeider relationship.
        October 1999                             4-20        DRAFT-DO NOT QUOTE OR CITE

-------
               1980

        *
        *
Figure 4-11.  Five-year average haze patterns (75th percentile of the extinction coefficient)
             centered on 1980 and 1990 (to be replaced with patterns for 1990 and 1995).
             The relationship between haze, expressed as extinction coefficient (corrected
             to 60% relative humidity) is 1.9/extinction coefficient in km"1 = visual range
             in km, i.e., the intensity of shading increases as visual range decreases:
             0.2 km-1 = 9.5 km, 0.17 = 11.2, 0.14 = 13.6, 0.11 = 17.3, 0.08 = 23.8, and
             0.05 = 38.0. Ql = Dec., Jan., Feb.; Q2 = March, April, May; Q3 = June,
             July, Aug.; Q4 = Sept., Oct., Nov.
October 1999
4-21
DRAFT-DO NOT QUOTE OR CITE

-------
 1     Regional Pattern
 2           Trends for specific regions in the eastern U.S. might be updated, from 1992 to 1998, and be
 3     presented in a manner similar to that shown in Figure 4-12 for 1940 to 1990 (CD96).  The trend
 4     graphs represent the 75th percentile of Bext for the stations located within the designated region.
 5     The trends are presented for Quarters 1 (winter) and 3 (summer) separately. The northwestern
 6     U.S. exhibits an increase of Quarter 3 haze between 1960 and 1970, and a steady decline between
 7     1973  (0.22) and 1992 (0.12). In the winter quarter the haziness has steadily declined from
 8     0.15 to 0.10 in the 30-year period.  The Mid-Atlantic region that includes the Virginias and
 9     Carolinas shows a strong summer increase between 1960 and 1973, followed by a decline.  The
10     winter haze was virtually unchanged over the 30-year period. The haziness over the Gulf states
11     increased between 1960 and 1970, and remained virtually unchanged since then. The central
12     Midwest including Missouri and Arkansas exhibit virtually no change during the winter season
13     and a slight increase in the summer (1960-1970).  The upper Midwest (Figure 4-12) shows an
14     opposing trend for summer and winter. While summer haze has increased, mostly 1960-1973,
15     the winter haze has declined.
16
17     4.2.5.2  Urban Trends
18           PM2 5 trends will be shown for those cities which have several years of PM2 5 data as they
19     become available. As an example, trend data from Stockton-Hazleton, CA are shown in
20     Figure 4-13.
21
22
23     4.3  SOURCES OF PRIMARY AND SECONDARY PARTICULATE
24           MATTER
25           Information about the nature and relative importance of sources of ambient PM is presented
26     in this section.  Table 4-3 summarize anthropogenic and natural sources for the major primary
27     and secondary aerosol constituents of fine and coarse particles. Major sources of each
28     constituent are shown in boldface type. Anthropogenic sources can be further divided into
29     stationary and mobile sources.  Stationary sources include fuel combustion for electrical utilities,
30     residential  space heating and industrial processes; construction and demolition; metals, minerals,
31     petrochemicals and wood products processing; mills and elevators used in agriculture; erosion

       October 1999                              4-22       DRAFT-DO NOT QUOTE OR CITE

-------
 o
 o
 r+
 O
 Tl
 H

 O
 O

 2
 O
 H

O
 c
 o
 H
 W

 O
 W
 O
 HH
 H
 W
^-v  0.26

 i

f=  0.22


,-x  0.18


j£J  0.14


    0.10
                          Upper Midwest
                                              Midwest
                                                               West Gulf
                                                               Southeast
                                                              Mid Atlantic
                                                               Northeast
      o>
     m
          0.06
                                               0.26


                                               0.22


                                               0.18


                                               0.14


                                               0.10
            1940 1950 1960 1970 1980 1990
                                    0.06
                                                               0.26


                                                               0.22


                                                               0.18


                                                               0.14


                                                               0.10
                                       1940 1950 1960 1970 1980 1990
                                                               0.06
                                                                                 0.26


                                                                                 0.22


                                                                                 0.18


                                                                                 0.14


                                                                                 0.10
                                                                  1940 1950 1960 1970 1980 1990
                                                                                          0.06
                                                                                 0.26


                                                                                 0.22


                                                                                 0.18


                                                                                 0.14


                                                                                 0.10
                                                                                             1940 1950 1960 1970 1980 1990
                                                                                                                     0.06
                                                                                 0.26


                                                                                 0.22


                                                                                 0.18


                                                                                 0.14


                                                                                 0.10
                                                                                                                        1940 1950 1960 1970 1980 1990
                                                                                                                                                0.06
                                                                                CD

                                                                                CD
                                                                                3
                                                                                O
                                                                                                                                                   1940 1950 1960 1970 1980 1990
     m
0.26


0.22


0.18


0.14


0.10
          0.06
0.26


0.22


0.18


0.14


0.10
            1940 1950 1960 1970 1980 1990
                                    0.06
0.26


0.22


0.18


0.14


0.10
                                       1940 1950 1960 1970 1980 1990
                                                               0.06
0.26


0.22


0.18


0.14


0.10
                                                                  1940 1950 1960 1970 1980 1990
                                                                                          0.06
                                                                                                                0.26


                                                                                                                0.22


                                                                                                                0.18


                                                                                                                0.14


                                                                                                                0.10
                                                                                             1940 1950 1960 1970 1980 1990
                                                                                                                     0.06
0.26


0.22


0.18


0.14


0.10
                                                                                                                        1940 1950 1960 1970 1980 19!
                                                                                                                                                0.06
                                                                                                                                                                          CD

                                                                                                                                                                          03

                                                                                                                                                                          a
                                                                                                                                                   1940 1950 1960 1970 1980 1990
Figure 4-12.  Secular haze trends (1940 to 1990) for six eastern U.S. regions, summer (Ql) and winter (Q2).

-------
o
o
r+
O
Tl
f
O
O
2
O
H
O
c
o
H
W
O
^
O
HH
H
W
                                Annual Arithmetic Mean (pg/m3 )
                 100

                  80

                  60

                  40

                  20

                   0
               	NAAQS
                                    Stockton-Hazelton, CA
                                                     100
                                      
                                                     80
                                                     70
                                                  eo  60
                                                  ~D> 50
                                                     40
                                                     30
                                                     20
                                                     10
                                                      0
                     89
                      90      91      92     93
                    • Total      	Coarse

                             90th Percentile (ug/nf )
                                 94      95
                                 •••• Fine
           180
           160
           140
           120
           100
           80
           60
           40
           20
             0
                                                                 (b)
              89
  90
• Total
91
   92      93
— Coarse
94      95
•••• Fine
                                                                                                   Every Sixth Day. 1991
                                                      1
                                                     o.g
                                                     0.8
                                                     0.7
                                                     °'4
                                                     0.3
                                                     0.2
                                                     0.1
                                               01/06/91  Io3/01/91 I  05/06/91 I  07/05/91  I 09/03/91 I  11/02/91 112/26/91
                                                    02/11/91  04/06/81  06/05/81    OB/10/81  10/03/91   12/08/81
                                                                     Date
                                                              0 COARSE  * FINE
                                                                                                PM2 5 as a Fraction of PM10
01/06/891 01/05/90 I 01/06/911 01/01/92loi/01/93l 01/02/941 01/04/95108/31/95
   07/03/89 07/04/90 07/05/91 07/05/92 07/06/93  07/01/94  07/Of —
                                                                                                                          7/02/95
                                                                                                   Date
Figure 4-13.  Trend data from Stockton-Hazelton from CARB:  (a) fine, coarse, and total means; (b) fine, coarse, and total 90th
               percentiles; (c) every-sixth-day fine and coarse mass for 1991; and (d) fine and coarse mass as a fraction of PM10.

-------
o
r+
O
&
CD
H^
>o
>o
>o
i ABLitL to. ^\jrv a 1 1 1 uiLr> i a \JF A i iviuarnjLKi^
rAKii^i^JLa Ar>u I±IJLIK IVIA^JUK auuK^
Jli»
Sources
Primary (PM <2.5 /^m) Primary (PM >2.5 /^m) Secondary PM Precursors (PM <2.5 /^m)
Aerosol
species Natural Anthropogenic Natural
SO4= Sea spray Fossil fuel combustion Sea spray
Sulfate
Anthropogenic Natural
— Oxidation of reduced sulfur
gases emitted by the oceans and
wetlands; and SO2 and H2S
emitted by volcanism and forest
fires
Anthropogenic
Oxidation of SO2 emitted
from fossil fuel
combustion1
O
         NO3-
         Nitrate
         Minerals
Erosion,
re-entrainment
         NH4+                —
         Ammonium


         Organic      Wild fires
         carbon (OC)
                 Motor vehicle exhaust2
Fugitive dust; paved,
unpaved roads;
agriculture and forestry


Motor vehicle exhaust2
                 Open burning, wood
                 burning, motor
                 vehicle exhaust,
                 cooking
                                                                            Oxidation of NOX produced by
                                                                            soils, forest fires, and lighting
                                                                                  Oxidation of NOX emitted
                                                                                  from fossil fuel
                                                                                  combustion; and in motor
                                                                                  vehicle exhaust
Erosion, re-entrainment
Fugitive dust; paved,
unpaved road dust,
agriculture and
forestry
                                                    Tire and asphalt wear,
                                                    paved road dust
                                                   Emissions of NH3 from wild
                                                   animals, undisturbed soil


                                                   Oxidation of hydrocarbons
                                                   emitted by vegetation,
                                                   (terpenes, waxes); wild fires
                                                        Emissions of NH3 from
                                                        animal husbandry,
                                                        sewage, fertilized land

                                                        Oxidation of hydrocarbons
                                                        emitted by motor vehicles,
                                                        open burning, wood
                                                        burning
Tl
H
O
O
2j
O
H
0
O
H
W
0
O
H
W
Elemental Wild fires Motor vehicle
carbon exhaust, wood
(EC) burning, cooking
Metals Volcanic Fossil fuel
activity combustion, smelting,
brake wear
Bioaerosols Viruses, —
bacteria
'Major source of each component shown in boldface type.
2Relatively minor source of substance, included only for the


Erosion, re-entrainment, — — —
organic debris
Plant, insect fragments, — — —
pollen, fungal spores,
bacterial agglomerates

sake of completeness.


-------
 1      from tilled lands; waste disposal and recycling; and fugitive dust from paved and unpaved roads.
 2      Mobile, or transportation related, sources include direct emissions of primary PM and secondary
 3      PM precursors from highway and off-highway vehicles and nonroad sources.  Also shown are
 4      sources for precursor gases whose oxidation forms secondary particulate matter.  A description of
 5      the atmospheric chemical processes producing secondary PM is given in Section 3.4.
 6           In general, the nature of sources of fine particulate matter is very different from that for
 7      coarse particulate matter. A large fraction of the mass in the fine size fraction is derived from
 8      material that has been formed during combustion (primary), has been volatilized in combustion
 9      chambers and then recondensed to form primary PM, or has been formed in the atmosphere from
10      precursor gases as secondary PM. Since precursor gases and fine particulate matter are capable
11      of traveling great distances, it is difficult to identify individual sources of constituents.  The
12      coarse PM constituents have shorter lifetimes in the atmosphere, so their impacts tend to be more
13      localized.  Only major sources for each constituent within each broad category shown at the top
14      of Table 4-3 are listed. Chemical characterizations of primary particulate emissions from a wide
15      variety of natural and anthropogenic sources as shown in Table 4-3 were given in Chapter 5 of
16      CD96.  Summary tables of the composition of source emissions presented in CD96 and updates
17      are given  in Appendix  A.  These profiles were based in large measure on the results of various
18      studies collecting source signatures for use in source apportionment studies.
19           Natural sources of primary PM include windblown dust from undisturbed land, sea spray,
20      and plant  and insect debris. The oxidation of a fraction of terpenes emitted by vegetation and
21      reduced sulfur species  from anaerobic environments leads to secondary PM formation.
22      Ammonium (NH4+) ions, which regulate the pH of particles, are derived from emissions of
23      ammonia (NH3) gas. Source categories for NH3 have been divided into emissions from
24      undisturbed soils (natural) and emissions which are related to human activities (e.g., fertilized
25      lands, domestic  and farm animal waste). It is difficult to describe emissions from biomass
26      burning as either natural or anthropogenic. Clearly, fuel wood burning is an anthropogenic
27      source of PM, whereas wildfires would be a natural source. Forest fires have been included as a
28      natural source, because of the lack of information on the amount of prescribed burning or
29      accidental fires caused by humans.  Similar considerations apply to the biogenic emissions of
30      trace metals which may be remobilized from anthropogenic inputs.


        October 1999                              4-26        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Receptor models are perhaps the primary means used to estimate source category
 2      contributions to particulate matter at individual monitoring sites.  Receptor models relate source
 3      category contributions to ambient concentrations based on composition analysis of ambient
 4      particulate and source emissions samples. In addition, receptor models have been developed for
 5      apportioning source categories of primary particulate matter and are not formulated to include the
 6      processes of secondary particulate matter formation.  However, hybrid receptor models which
 7      use elements from a chemical transport model have been used to apportion source categories of
 8      gaseous precursors to secondary particulate matter (Stevens and Lewis, 1984).
 9           A number of specialty conference proceedings, review articles, and books have been
10      published to provide greater detail about source category apportionment receptor models (Cooper
11      and Watson, 1980; Watson et al, 1981; Macias and Hopke, 1981; Dattner and  Hopke, 1982;
12      Pace, 1986; Watson et al., 1989; Gordon, 1980, 1988; Stevens and Pace, 1984; Hopke, 1985,
13      1991; Javitz et al., 1988).  Watson et al. (1994a) present data analysis plans which include
14      receptor models as an integral part of visibility and PM10 source apportionment and control
15      strategy development. A review of the various methods used to apportion PM  in ambient
16      samples among its source categories was given in Section 5.5.2 of CD96.  The collection of the
17      source category characterization profiles shown in Appendix 4A has been  motivated in many
18      cases by the need to use them in receptor modeling applications.
19
20      4.3.1  Source Contributions to Ambient PM
21           The results of several source apportionment studies will be discussed in this section to
22      provide an indication of different sources of particulate matter across the United States.  First,
23      results obtained by using the chemical mass balance (CMB) approach for estimating
24      contributions to PM2 5 from different source categories at monitoring sites  in the United States
25      will be  discussed and presented in Table 4-4. Results obtained at a number of monitoring sites in
26      the central and western United States by using the CMB model for PM10 are shown in Table 4-5.
27      The sampling  sites represent a variety of different source characteristics within different regions
28      of Arizona, California, Colorado, Idaho, Illinois, Nevada and Ohio. Several of these are
29      background sites, specifically Estrella Park, Gunnery Range, Pinnacle Peak, and Corona de
30      Tucson, AZ, and San Nicolas Island, CA. Definitions of source categories also vary from study

        October 1999                               4-27       DRAFT-DO NOT  QUOTE OR  CITE

-------
October 1999







-h
oo

o
H
O
0
2
o
H
O
O
H
W
O
O
H
W
TABLE 4-4. RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM2 5
% Contribution3
Gasoline Road Dust; Vegetation Secondary
Sampling Site Time Period Diesel Vehicles soil burning Sulfate
Pasadena, CA (Schauer etal., 1996) 1982 18.8 5.1 12.4 9.6 20.9
DowntownLA, CA (Schauer etal., 1996) 1982 35.7 6.5 11.1 5.8 20.3
WestLA,CA (Schauer etal., 1996) 1982 18.0 5.7 12.2 11.0 24.1
Rubidoux, CA (Schauer et al., 1996) 1982 12.8 0.7 13.1 1.2 13.8
Philadelphia, PA (Dzubay etal., 1988) Summer 1982 8.5e — 4.4 — 81.9f
Camden, NJ (Dzubay etal., 1988) Summer 1982 9.2e — 3.2 — 81.3f
Clarksboro, NJ (Dzubay etal., 1988) Summer 1982 5.8e — 2.7 — 84.6f
Grover City ILENP; (Glover etal., 1991) 1986 to 1987 — — 2.3 — 83.2f
Grover City, IL SSW; (Glover etal., 1991) 1986 to 1987 — — — — 59.0f
Grover City, IL WNW; (Glover etal., 1991) 1986 to 1987 2.4' — 5.1 — 88.5f
Grover City, IL NNW; (Glover etal., 1991) 1986 to 1987 — — 3.1 — 86.6f
Welby, CO (Lawson and Smith, 1998) Winter 1997 10 28 16 5 10"
Brighton, CO (Lawson and Smith, 1998) Winter 1997 10 26 11 2 15°
'secondary and other organic compounds fincluding associated cations and water
'secondary ammonium sincinerators
"meat cooking hoil fly ash
Vegetative detritus 'fluidized catalyst cracker
Value represents sum of diesel and gasoline vehicle exhaust -Kvind direction




Secondary Misc. Misc. Misc. Misc. Measured
Nitrate Source Source Source Source PM2S
1234 Concentration
7.4 5.3' 9.2b 8.5C l.ld
9.2 3.7' 9.2b 5.2C 0.6d
7.8 4.1' 9.4b 8.2C 1.6d
24.7 4.5' 12.1b 4.5C 0.5d
— 2.2s 1.9h 0.41 —
0.4 2.5s 2.5h 0.71 —
— 0.8s 1.5h 0.4s —
— 9.7k 3.01 1.2s —
— 11.6k 11.91 4.1s 4.6m
— 2.8k — — —
— 3.41 3.0" — —
25P 4, T
32P 2° T — —
klead smelter
'iron works
"copper smelter
"coal power plant
°as ammonium sulfate
pas ammonium nitrate




28.2
32.5
24.5
42.1
27.0
28.3
26.0












-------
o
o
r-K
o
VO
VO
VO
TABLE 4-5. RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM,
                                                                                               % Contribution3
-j^

vo
O
Tl
H
o
O
2
O
H
o
H
W
O
!*
O
I-H
H
W
                                               Primary
                                               Motor   Primary   Secondary  Secondary  Misc.  Misc.   Misc.   Misc.
                            Primary    Primary   Vehicle  Vegetative  Ammonium Ammonium  Source Source  Source  Source
                                                                                                                                                Measured
Sampling Site
Central Phoenix, AZ (Chow et al., 1991)
Craycroft, AZ (Chow et al., 1992a)
Hayden 1, AZ (Garfield) (Ryan et al., 1988)
Hayden 2, AZ (Jail) (Ryan et al., 1988)
Phoenix, AZ (Estrella Park) (Chow et al., 1991)
Phoenix, AZ (Gunnery Rg.) (Chow et al., 1991)
Phoenix, AX (Pinnacle Pk.) (Chow et al., 1991)
Rillito, AZ (Thanukos et al., 1992)
Scottsdale, AZ (Chow et al., 1991)
West Phoenix, AZ (Chow et al., 1991)
Bakersfield, CA (Magliano, 1988)
Bakerfield, CA (Chow et al., 1992b)
Crows Landing, CA (Chow et al., 1992b)
Fellows, CA (Chow et al., 1992b)
Fresno, CA (Magliano, 1988)
Fresno, CA (Chow et al., 1992b)
Indio, CA (Kim et al., 1992)
Kern Wildlife Refuge, CA (Chow et al., 1992b)
Long Beach, CA (Gray et al., 1988)
Long Beach, CA (Summer) (Watson et al., 1994a)
Long Beach, CA (Fall) (Watson et al., 1994a)
Riverside, CA (Chow et al., 1992c)
Rubidoux, CA (Gray et al., 1988)
Rubidoux, CA (Summer) (Watson et al., 1994a)
Rubidoux, CA (Fall) (Watson et al., 1994a)
Rubidoux, CA (Chow et al., 1992c)
San Nicolas Island, CA (Summer) (Watson et al.,
1994a)
Time Period
Winter 1989-1990
Winter 1989-1990
1986
1986
Winter 1989-1990
Winter 1989-1990
Winter 1989-1990
1988
Winter 1989-1990
Winter 1989-1990
1986
1988-1989
1988-1989
1988-1989
1986
1988-1989

1988-1989
1986
Summer 1987
Fall 1987
1988
1986
Summer 1987
Fall 1987
1988
Summer 1987
Geological
51.6
55.6
4.8
35.6
67.3
74.1
58.3
53.7
45.5
43.5
40.5
53.9
61.3
53.1
35.6
44.5
56.9
31.6
39.9
24.1
11.8
50.9
49.3
30.4
17.1
55.2
9.2
Construction
0.0
0.0
1.9b
6.8b
0.0
0.0
0.0
17.4b
0.0
0.0
4.4
2.0
0.0
2.6
1.5
0.0
5.2
4.2
0.0
0.0
0.0
0.0
4.6"
3.9
14.4
0.0
0.0
Exhaust
39.0
35.5
0.0
0.0
18.2
20.4
24.2
1.5f
34.5
36.2
8.1
9.7
4.2
3.8
8.3
9.5
7.6
4.6
9.8s
13.7
44.5
10.9
6.4s
15.1
27.1
11.7
5.2
Burning
3.6
0.0
0.0
0.0
1.6
0.0
8.3
0.0
13.5
14.5
14.21
8.2
6.5
6.2
19.11
7.1
12.2
8.4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Sulfate
0.3
3.0
3.8
6.8
2.9
3.7
7.5
0.0
1.1
0.6
8.3
6.9
5.3
9.3
3.7
5.0
6.2
6.9
15.4
23.6
4.0
7.5
7.3
8.3
1.9
6.1
21.3
Nitrate
4.4
2.6
0.0
0.0
0.0
0.0
0.0
0.0
6.5
4.5
0.0
16.0
12.4
13.7
0.0
14.5
7.1
3.1
17.7
1.7
24.1
33.4
24.4
23.9
28.2
24.9
2.9
1
0.0
5.1°
70.5°
47.5°
0.0
0.0
0.0
14.6s
0.0
0.0
0.7j
1.3"
1.0m
12.8m
0.2j
0.4m
0.3j
1.0m
0.2j
0.2j
O.Oi
0.5j
0.3j
0 (V
0 (V
0.&
o.tf
2
0.0
0.0
4.8d
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.9"
1.9"
2.6"
0.0
1.9"
1.7h
3.1°
3.9h
4.8h
2.8h
2.0h
1.1°
4.4h
1.0h
1.7h
24.7h
3
0.0
0.0
1.0e
l.T
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.8k
2.3k
2.6k
0.0
O.lk
0.0
1.5k
12.3k
0.0
0.0
1.7°
6.8k
0.0
0.0
6.6°
0.0
4
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Concentration
64.0
23.4
105.0
59.0
55.0
27.0
12.0
79.5
55.0
69.0
67.6
79.6
52.5
54.6
48.1
71.5
58.0
47.8
51.9
46.1
96.1
64.0
87.4
114.8
112.0
87.0
17.4

-------
o
o
r-K
o
VO

VO

VO
TABLE 4-5 (cont'd). RECEPTOR MODEL SOURCE CONTRIBUTIONS TO PM,
u>
o
O
Tl
H


O

O

2

O

H




Sampling Site Time Period
Stockton, CA (Chow et al., 1992b) 1989
Pocatello, ID (Houck et al., 1992) 1990
S. Chicago, IL (Hopke et al., 1988) 1986
S.E. Chicago, IL (Vermette et al., 1992) 1988
Reno, NV (Chow et al., 1988) 1986-1987
Sparks, NV (Chow et al., 1988) 1986-1987
Follansbee, WV (Skidmore et al., 1992) 1991
Mingo, OH (Skidmore et al., 1992) 1991
Steubenville, OH (Skidmore et al., 1992) 1991
"Smelter background aerosol.
bCement plant sources, including kiln stacks, gypsum pile, and kiln area.
"Copper ore.
dCopper tailings.
'Copper smelter building.
fHeavy-duty diesel exhaust emission.
background aerosol.
'Marine aerosol, road salt, and sea salt plus sodium nitrate.
'Motor vehicle exhaust from diesel and leaded gasoline.

Primary
Motor
Primary Primary Vehicle
Geological Construction Exhaust
55.1 0.8 8.3
8.3 7.5' 0.1
34.0 3.0 3.5
35.9¥ 0.0 2.2f
497 0.0 33.3
36.8 0.0 28.3
15.2 0.0 53.0
20.0 0.0 23.3
18.0 0.0 30.4
^Residual oil combustion.
kSecondary organic carbon.
'Biomass burning.
"Primary crude oil.
"NaCl + NaNO3.
"Lime.
FRoad sanding material.
'Asphalt industry.
%

Primary
Vegetative
Burning
7.7
0.0
0.0
0.0
6.3
32.7
0.0
6.8
1.7








Contribution3

Secondary Secondary Misc. Misc.
Ammonium Ammonium Source Source
Sulfate Nitrate 1 2
5.0 11.2 1.1" 2.9°
0.0 0.0 0.0 0.0
19.2s - 18.9' 2.7°
18.8 - 2.0' 0.7h
4.3 2.0 0.0 0.0
6.6 2.2 0.0 0.0
24.2 - 14.1' 0.0
25.0 - 5.7' 18.3X
30.4 - 8.3' 10.91
'Regional sulfate.
'Steel mills.
"Refuse incinerator.
"Local road dust, coal yard road dust,
"Incineration.
"Unexplained mass.




Misc. Misc. Measured
Source Source PMi0
3 4 Concentration
0.0k 0.0 62.4
84.1r 0.0 100.0
0.0 0.0 80.1
2.7W 18.8s 41.0
0.0 0.0 30.0
0.5k 0.0 41.0
0.0 0.0 66.0
0.0 0.0 60.0
0.0 0.0 46.0



steel haul road dust.




Thosphorus/phosphate industry.
o
H
W

O
O
I-H

H
W

-------
 1      to study.  The results of the PM10 source apportionment studies were given in CD96 and are
 2      presented here to allow easy comparison with results of PM25 source apportionment studies.
 3           There are several differences between the source categories shown at the tops of Tables 4-4
 4      and 4-5.  These differences reflect the nature of sources which are important for producing fine
 5      and coarse particulate matter shown in Table 4-3. They are also related to improvements in the
 6      ability to distinguish between sources of similar nature, e.g., diesel and gasoline vehicles, meat
 7      cooking and vegetation burning. It has only been recently that motor vehicle emissions can be
 8      broken down into contributions from diesel and gasoline vehicles through the use of organic
 9      tracers.  Meat cooking is also distinguished from vegetation burning in more recent studies,
10      although both are considered to be part of biomass burning.  Vegetation burning consists of
11      contributions from residential fuel wood burning, wild fires, prescribed burning and burning of
12      agricultural waste. Miscellaneous sources of fine particles include contributions from
13      combustion sources while miscellaneous sources of coarse particles consist of contributions from
14      soil and sea spray and industrial processing of geological material (e.g., cement manufacturing).
15      Although a large number of elements and chemical compounds are used to differentiate among
16      source categories, it can be seen from Tables 4-4 and 4-5 that only a relatively small number of
17      sources are needed to account for the mass of PM25 and PM10.
18           Secondary sulfate is the dominant component of PM25 samples collected in the studies of
19      Dzubay et al. (1988) and Glover et al. (1991). Both studies found that sulfate at their monitoring
20      site arose from regionally dispersed sources. Sulfate,  associated cations  and water also represent
21      the major components of PM25 found in monitoring studies in the eastern United States
22      (Figure 4-la). Motor vehicle emissions, arising mainly from diesels, are other major sources of
23      PM25. Contributions from road dust and soils are relatively minor, typically constituting less
24      than 10% of PM2 5 in the studies shown in Table 4-4.  The most notable difference in the relative
25      importance of major source categories of PM25 shown in Table 4-4 and PM10 shown in Table 4-5
26      involves crustal material, (e.g., soil, road dust) which represents about 40% on average of the
27      total mass of PM10 in the studies shown in Table 4-5.  The fraction is higher in locations located
28      away from specific sources such as sea spray or smelters.  Emissions of fugitive dust are
29      concentrated mainly in the PM(10_2 5) size range. The average fugitive dust source contribution is
30      highly variable among sampling sites within the same urban areas, as seen by differences
31      between the Central Phoenix (33 //g/m3) and Scottsdale (25 //g/m3) sites in Arizona, and it is also

        October 1999                              4-31         DRAFT-DO NOT QUOTE OR CITE

-------
 1      seasonally variable, as evidenced by the summer and fall contributions at Rubidoux, CA. The
 2      variability in fugitive dust loadings reflects the sporadic nature of its emissions and its short
 3      lifetime in the atmosphere.
 4           In Table 4-5, primary motor vehicle exhaust contributions account for up to 40% of average
 5      PM10 at many of the sampling sites. Vehicle exhaust contributions are also variable at different
 6      sites within the same study area. The mean value and the variability of motor vehicle exhaust
 7      contributions reflects the proximity of sampling sites to roadways and traffic conditions during
 8      the time of sampling. Many studies were conducted during the late 1980s, when a portion of the
 9      vehicle fleet still used leaded gasoline.  Pb and Br in motor vehicle emissions facilitated the
10      distinction of motor vehicle contributions from other sources.  Vehicles using leaded fuels have
11      higher emission rates than vehicles using unleaded fuels. Pb also poisons automobile exhaust
12      catalysts and produces adverse human health effects. As a result, Pb has been virtually
13      eliminated from vehicle fuels. However, recently organic species have replaced Pb as a source
14      marker for motor vehicle emissions (e.g., Rogge et al., 1993a).
15           Marine aerosol is found, as expected, at coastal sites such as Long Beach (average 3.8% of
16      total mass), and San Nicolas Island (25%). These contributions are relatively variable and are
17      larger at the more remote sites.  Individual values reflect proximity to  local sources. Of great
18      importance are the contributions from secondary ammonium sulfate in the eastern United States
19      and ammonium nitrate in the western United States.  These are especially noticeable at sites in
20      California's San Joaquin Valley (Bakersfield, Crows Landing, Fellows, Fresno, Kern Wildlife,
21      and Stockton) and in the Los Angeles area.
22           Samples selected for chemical analysis are often biased toward the highest PM10 mass
23      concentrations in the studies shown in Table 4-5, so average source contribution estimates are
24      probably not representative of annual averages.  Quoted uncertainties in the estimated
25      contributions of the individual sources shown in Table 4-4 range from 10 to 50%.  Uncertainties
26      of source contribution estimates are not usually reported with the average values summarized in
27      Table 4-5. Estimates of standard errors are calculated in source apportionment studies, and
28      typically range from 15 to 30% of the source contribution estimate.  They are much higher when
29      the chemical source profiles for different sources are highly uncertain  or too similar to
30      distinguish one source from another.
31

        October 1999                              4-32        DRAFT-DO NOT QUOTE OR CITE

-------
 1     4.4  EMISSIONS ESTIMATES AND THEIR UNCERTAINTIES
 2           In principle, source contributions to ambient PM could also be estimated on the basis of
 3     predictions made by chemistry-transport models (CTM) or even on the basis of emissions
 4     inventories alone. Uncertainties in emissions inventories have arguably been regarded as
 5     representing the largest source of uncertainty in CTMs (Calvert et al.,  1993). A number of
 6     factors limit the ability of an emissions inventory driven CTM to determine the effects of various
 7     sources on particle samples obtained at a particular location, apart from uncertainties in the
 8     inventories given above. Air pollution model predictions represent averages over the area of a
 9     grid cell, which in the case of the Urban Airshed Model typically has been 25 km2
10     (5 km x 5 km). The contributions of sources to pollutant concentrations at a monitoring site are
11     controlled strongly by local conditions which cannot be resolved by an Eulerian grid-cell model.
12     Examples would be the downward mixing of tall stack emissions and  deviations from the mean
13     flow caused by buildings. The impact of local sources at a particular point in the model domain
14     may not be predicted accurately, because their emissions would be smeared over the area of a
15     grid cell or if the local wind fields at the  sampling point deviated significantly from the mean
16     wind fields calculated by the model.
17
18     4.4.1 Emissions Estimates for Primary Particulate Matter and SO2, NOX, and
19            VOCs in the United States
20           Estimated emissions of primary PM25 from different sources in the United States are
21     summarized in Figure 4-14. The estimates are based on information presented in the U.S.  EPA
22     National Air Pollutant Emission Trends Report, 1997 (U.S. Environmental Protection Agency,
23     1998) to which the reader is referred for  detailed tables showing trends in PM25 emissions from a
24     number of source categories from!990 to 1997; descriptions of the methodology used in the
25     construction of these tables; and descriptions of the uncertainties involved in the estimation
26     process. This  document also provides information about emissions of PM10, sulfur dioxide
27     (SO2), nitrogen oxides (NOX ), volatile organic compounds (VOCs), and ammonia (NH3).
28           Estimated total nationwide emissions of primary PM25 were 7.5  Tg yr"1 in 1997. The
29     category of fossil fuel combustion referred to in Figure 4-14 includes fossil fuel burning by
30     electric utilities, industry, and residences. The industry category includes contributions from
31     metals processing, petroleum refining, agricultural products processing, mining, and the storage
       October 1999                             4-33        DRAFT-DO NOT QUOTE OR CITE

-------
                PM25 Total  Emissions  (1997) = 18 Tg  yr1
                          On-Road
                            2.5%
                   Incinerator
                      .07%
Industry
 6.3%
           Fossil Fuel
          Combustion
             5.8%
              Biomass
               Burning
               17.5%
                                 Non-Road
                                    4.9%
                                            Wind Erosion
                                                9.6%
                                                                      Agriculture
                                                                         11.1%
                                                                    Fugitive Dust
                                                                        41.7%
      Figure 4-14.  Nationwide emissions of PM2 5 from various source categories.


      Source: U.S. Environmental Protection Agency (1998).
1     and transport of industrial goods. Incineration refers to the burning of non-biomass waste by

2     residences and municipalities. The on-road vehicles category includes contributions from gas

3     and diesel powered vehicles. Non-road engines and vehicles include their emissions in

4     transportation, construction, and other commercial, industrial and recreational activities. Wind

5     erosion refers to the raising of mineral dust by the wind. The biomass burning category includes

6     contributions from residential wood burning, open burning of vegetation, and forest fires.

7     The agriculture category includes mainly emissions of soil dust related to the production of
      October 1999
                         4-34
DRAFT-DO NOT QUOTE OR CITE

-------
 1      agricultural crops and livestock. Fugitive dust refers mainly to mineral dust raised by on-road
 2      and non-road vehicles during their operation. As can be seen from inspection of Figure 4-14, the
 3      raising of mineral dust by wind erosion, agriculture, and as fugitive dust constitutes the largest
 4      primary source (62.4%) of PM25 on a nationwide basis. Biomass burning constitutes the second
 5      largest primary source (17.5%) of PM25. The gross composition of emissions from most of these
 6      categories is summarized in Table 4-3.  Total emissions of PM25 as well as contributions from
 7      individual source categories were relatively constant over the period from 1990 to 1997 (U.S.
 8      Environmental Protection Agency, 1998). While crustal dust constitute over 60% of the total
 9      PM25 inventory, they constitute less than about 10% of the source strengths inferred from the
10      receptor modeling studies shown in Table 4-4. However,  it should be remembered that
11      secondary components often represent the major fraction of ambient samples. Dust sources
12      constitute 88% of the annual average PM10 National Emissions Inventory (U.S. Environmental
13      Protection Agency,  1994), but they average more than 50% of the contribution to average PM10
14      concentrations in only about 40% of the entries shown in Table 4-5. The reasons for this
15      apparent discrepancy are not clear. In addition to errors in inventories or source apportionments,
16      weather-related factors (wind speed and ground wetness) and the dominance of local sources on
17      spatial scales too small to be captured in inventories may be involved. It should be remembered
18      that dust emissions are widely dispersed and highly sporadic. Dust particles also have short
19      atmospheric residence times and as a result their dominance in emissions inventories may not be
20      reflected in samples collected near specific sources.
21          The geographic distribution of primary PM25 emissions is shown in Figure 4-15 and the
22      distribution of primary PM10 emissions is shown in Figure 4-16. As may be expected, high
23      emissions of PM2 5 are centered around many large urban areas. Although mineral dust sources
24      account for most of the emissions, their contributions are distributed much more widely than are
25      those from combustion sources. The nature of the distribution of mineral dust sources is shown
26      more clearly in Figure 4-16 where high densities of PM10 emissions are also found in sparsely
27      populated areas.
28          Estimated contributions from individual sources to emissions of gaseous precursors to
29      secondary PM formation are summarized in Figure 4-16 for SO2, NOX , VOCs, and NH3.
30      Information about the yield of particulate matter formed during the oxidation of VOCs is given in
31      Section 3.4.

        October 1999                             4-35        DRAFT-DO NOT QUOTE OR CITE

-------
                     mi.)
            11 .11
            f-11
            4-f
                                          =^'
                                          I
                                        - ,1
      Figure 4-15.  Distribution of primary PM25 emissions across the United States.
      Source: U.S. Environmental Protection Agency (1998).
1           Although total emissions of gaseous precursors (SO2, NOX, VOC's, and NH3) are shown in
2      Figure 4-17, it should be remembered that these values cannot be directly translated into
3      production rates of particulate matter. Dry deposition and precipitation scavenging of some of
4      these gases can occur before they are converted to particulate matter in the atmosphere.
5      In addition, some fraction of these gases are transported outside of the domain of the continental
6      United States before being oxidized.  Likewise, emissions of these gases from areas outside the
7      United States can result in the transport of their oxidation products into the United States. While
8      the chemical oxidation of SO2 will lead quantitatively to the formation of SO4=, the yield of
9      particulate matter from the oxidation of VOCs will be much less because only a small fraction of
       October 1999
4-36
DRAFT-DO NOT QUOTE OR CITE

-------
                      rW,
             >n
             it -11
             7-11
             4-7
             0-4
       Figure 4-16.  Distribution of primary PM10 emissions across the United States.
       Source: U.S. Environmental Protection Agency (1998).
 1     VOC's react to form particles, and those that do have efficiencies less than 10% (c.f.
 2     Section 3.4).
 3           The values shown in this section are based on annually averaged quantities.  However,
 4     annual averages do not reflect the seasonality of a number of emissions categories. Residential
 5     wood burning in fireplaces and stoves, for example, is a seasonal practice which reaches its peak
 6     during cold weather. Cold weather also affects motor vehicle exhaust particulate emissions, both
 7     in terms of chemical composition and emission rates (e.g., Watson et al., 1990a; Huang et al.,
 8     1994). Planting, fertilizing, and harvesting are also seasonal activities. Forest fires occur mainly
 9     during the  local dry season and during periods of drought. Maximum dust production by wind
10     erosion in the United States occurs during the spring, while the minimum occurs during the
       October 1999
4-37
DRAFT-DO NOT QUOTE OR CITE

-------
                   SO2 Total Emissions (1997)

                           = "ISTgyr-1
                                  All Other
                      Non-Road Engines   7%
                       and Vehicles
                          5%
                 Metal Processing
                      3%
           Fuel Combustion - Other
                 4%
       Fuel Combustion - Industrial
              17%
                                                    Fuel Combustion -
                                                    Electrical Utilities
                                                                 NOX Total Emissions (1997)

                                                                         = 21 TgyH
                                                           Fuel Combustion-
                                                             Industrial
                                                               14%
                                                      Fuel Combustion -
                                                       Electric Utility
                                                          26%
                                                                               Fuel Combusion -
                                                                                  Other
                                                                                   5%
                             On Road Vehicles
                                 30%
                                                                                              Non-Road Engines
                                                                                                and Vehicles
                                                                                                   19%
                  VOCs Total Emissions (1997)
                           = 17Tgyr-1
                       All Other
                         19%
           Non-Road Engines
            and Vehicles
               13%
                                                                   NH3 Total Emissions (1997)

                                                                           = 2.9Tgyr1
                                                                              All Other
                                                                   On-Road Vehicles  4-2%
                                                                      7.6%
                                                                Waste Water
                                                                Treatment
                                                                  3.1%
                                                                Agricultural Chemical
                                                                   Manufacturini
                                                                     6.1%
                       On Road Vehicles
                           27%
       Figure 4-17.   Nationwide emissions of SO2, NOX, VOCs, and NH3 from various source
                       categories.

       Source: U.S. Environmental Protection Agency (1998).
1

2

3
summer (Gillette and Hanson, 1989).  Effects are being made to account for the seasonal

variations of emissions in the nationwide emissions inventories.
       October 1999
                                                4-38
DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.4.2  Uncertainties of Emissions Inventories
 2           As described in CD96, it is difficult to assign uncertainties quantitatively to entries in
 3      emissions inventories. Methods that can be used to verify or place constraints on emissions
 4      inventories are sparse. In general, the overall uncertainty in the emissions of a given pollutant
 5      includes contributions from all of the terms used to calculate emissions, i.e., activity rates,
 6      emissions factors, and control device efficiencies. Additional uncertainties arise during the
 7      compilation of an emissions inventory because of missing sources and computational errors.
 8      The variability of emissions can cause errors when annual average emissions are applied to
 9      applications involving shorter time scales.
10           Activity rates for well-defined point sources (e.g., power plants) should have the smallest
11      uncertainty associated with their use, since accurate production records need to be kept.  On the
12      other hand, activity rates for a number of areally dispersed fugitive sources are extremely
13      difficult to quantify. Emissions factors for easily measured fuel components which are released
14      quantitatively during combustion (e.g., CO2 and SO2) should be the most reliable. Emissions of
15      components formed during combustion are more difficult to characterize as the emissions rates
16      are dependent on factors specific to individual combustion units and on combustion stage (i.e.,
17      smoldering or active). Although the  AP-42 emissions factors (U.S. Environmental Protection
18      Agency, 1995) contain extensive information for a large number of source types, these data are
19      very limited in the number of sources sampled.  The efficiency of control devices is determined
20      by their age, their maintenance history, and operating conditions.  It is virtually impossible to
21      assign uncertainties in control device performance due to these factors.  It should be noted that
22      the largest uncertainties occur for those devices which have the highest efficiencies (>90%).
23      This occurs because the efficiencies are subtracted from one and small errors in assigning
24      efficiencies can lead to large errors in emissions.
25           Ideally an emissions inventory  should include all major sources of a given pollutant.  This
26      may be an easy task for major point sources, but becomes problematic for poorly characterized
27      area sources of both primary PM and precursors to secondary PM formation.  Further research is
28      needed to better characterize the sources of pollutants in order to  reduce this source of
29      uncertainty. Errors can arise from the misreporting of data, and arithmetic errors can occur in the
30      course of compiling entries from thousands of individual  sources. A quality assurance program
31      is required to check for outliers and arithmetic errors.
        October 1999                              4-39        DRAFT-DO NOT QUOTE OR CITE

-------
 1           Because of the variability in emissions rates, there can be errors in the application of
 2      inventories developed on an annually averaged basis (as are the inventories shown in
 3      Figures 4-14 to 4-17) to episodes occurring on much shorter time scales.  As an example, most
 4      modeling studies of air pollution episodes are carried out for periods of a few days.
 5           Uncertainties in annual emissions were estimated to range from 4 to 9% for SO2 and from
 6      6 to 11% for NOX in the 1985 NAPAP inventories for the United States (Placet et al, 1991).
 7      Uncertainties in these estimates increase as the emissions are disaggregated both spatially and
 8      temporally. The uncertainties quoted above are minimum estimates and refer only to random
 9      variability about the mean, assuming that the variability in emissions factors was adequately
10      characterized and that extrapolation of emissions factors to sources other than those for which
11      they were measured is valid. The  estimates do not consider the effects of weather or variations in
12      operating and maintenance procedures. Fugitive dust sources, as mentioned above, are extremely
13      difficult to quantify, and stated emission rates may represent only order-of-magnitude estimates.
14      As rough estimates, uncertainties in emissions estimates could be as low as 10% for the best
15      characterized source categories, while emissions figures for windblown dust should be regarded
16      as order-of-magnitude estimates.  Given (a) uncertainties in the deposition of SO2 and its
17      oxidation rate, (b)  the variability seen in OC and EC emissions from motor vehicles along with
18      the findings from past verification studies for NMHC and CO to NOX ratios, (c) ranges of values
19      found among independent estimates for emissions of individual species (NH3, OC), and (d) the
20      predominance of fugitive emissions, PM emissions rates should be regarded as
21      order-of-magnitude estimates.
22           There have been few field studies designed to  test emissions inventories observationally.
23      The most direct approach would be to obtain cross-sections of pollutants upwind and downwind
24      of major urban areas from aircraft. The computed mass flux through a cross section of the urban
25      plume can then be equated to emissions from the city chosen.  This approach has been attempted
26      on a few occasions. Results have  been ambiguous because of contributions from fugitive
27      sources, non-steady wind flows, and general logistic difficulties.
28
29
30
31

        October 1999                             4-40        DRAFT-DO NOT QUOTE OR CITE

-------
 1      4.5  LONG RANGE TRANSPORT OF PM FROM SOURCES OUTSIDE
 2           THE UNITED STATES
 3           Apart from sources within the continental United States, particulate matter can be brought
 4      in by long range transport from sources outside the United States, as evidenced by the transport
 5      of PM from uncontrolled biomass burning in Central America and southern Mexico which
 6      resulted in anomalously high PM levels observed in southern Texas and to a lesser extent,
 7      throughout the entire central and southeastern United States during the spring and early summer
 8      of 1998. Windblown dust from individual dust storms in the  Sahara desert has been observed in
 9      satellite images as plumes crossing the Atlantic Ocean and reaching the southeast coast of the
10      United States (e.g., Ott et al., 1991).  Dust transport from the  deserts of Asia across the Pacific
11      Ocean also occurs (Prospero, 1995).  Most dust storms in the deserts of China occur in the spring
12      after the snow has melted and before vegetation has grown following the passage of strong cold
13      fronts.  Strong winds and unstable conditions result in the rapid transport of dust into the middle
14      and upper troposphere (4-5 km altitude) where it is transported by strong westerly winds out over
15      the Pacific Ocean (Duce, 1995). Satellite images have been used to track the progress of a dust
16      cloud from the Gobi desert to the northwestern United States during the spring of 1998.
17           Satellite images obtained at visible wavelengths cannot track mineral dust across the
18      continents  because of a lack of contrast between the plume and the underlying surface.  Other
19      means must be used to track the spread of North African dust through the eastern United States.
20      Perry et al. (1997) used two criteria (PM2 5 soil concentration  > 3 //g m"3 and Al/Ca > 3.8) to
21      distinguish between soil of local origin from soil originating in North Africa in characterizing the
22      sources of PM in aerosol samples collected in the IMPROVE (Interagency Monitoring of
23      Protected Visual Environments) network. Their analysis indicates that incursions of Saharan
24      dust into the continental United States have occurred, on average, about three times per year from
25      1992 to 1995. These events have persisted for about ten days principally during the summer.
26      As can be expected, the frequency of dust events is highest in the southeastern United States and
27      about half are observed only within the state of Florida and these are associated with dense hazes
28      in Miami during the summer (Prospero et al., 1987) such that African dust is the dominant
29      aerosol constituent in south Florida during the summer (Prospero, 1999). The mass median
30      diameter of mineral dust over the oceans is typically between 2 and 3 //m (Duce, 1995).
31      North African dust has been tracked as far as Illinois (Gatz and Prospero, 1996) and to Maine

        October 1999                             4-41       DRAFT-DO NOT QUOTE OR CITE

-------
 1      (Perry et al., 1997). Larger scale events typically covered from 15 to 30% of the area of the
 2      continental United States and resulted in increases of PM25 levels of 8.7 ± 2.3 m m"3 throughout
 3      the affected areas with mean maximum dust contributions of 19.7 ± 8.4 mg m"3 during these
 4      events, and a peak contribution of 32 mg m"3 to 24-h average PM2 5 levels.
 5
 6
 7      4.6  SUMMARY AND CONCLUSIONS
 8           Ambient particulate matter contains both primary and secondary components.  The results
 9      of ambient monitoring studies and receptor modeling studies in the eastern United States indicate
10      that PM2 5 is dominated by secondary components.  Secondary constituents are smaller but still
11      important components of PM25 in the central and western United States.  Minerals constitute the
12      largest fraction of PM(10_25) throughout the United States.  Data collected in the Los Angeles
13      Basin and Philadelphia suggest that secondary PM components are more uniformly distributed
14      than are primary components. Compositional data obtained at multiple sites in other urban areas
15      are sparse.
16           Due to the complexity of the composition of ambient PM25 and PM(10_25), sources are best
17      discussed in terms of individual constituents of both primary and secondary PM2 5 and PM(10_2 5).
18      Each of these constituents can have anthropogenic and natural sources, as shown  in Table 4-3.
19      The distinction between natural  and anthropogenic sources is not always obvious. While
20      windblown dust might seem to be the result of natural processes, highest emission rates are
21      associated with agricultural activities in areas that are susceptible to periodic drought. Examples
22      include the dust bowl region of the midwestern United States and the Sahel of Africa. Most
23      forest fires in the United States may ultimately be of human origin, either through prescribed
24      burning or accident.
25           Emissions inventories are generally not the most appropriate way to apportion material in
26      ambient samples. Receptor modeling has proven to be an especially valuable  tool in this regard.
27      Compositional profiles developed for receptor modeling applications are perhaps the most
28      accessible and reliable means of characterizing the composition of emissions.  The results of
29      receptor modeling studies throughout the United States indicate that the  combustion of fossil and
30      biomass fuels is a major source of PM25. Fugitive dust, found mainly in the PM(10_25) range size,
31      represents the largest source of PM10 in many locations in the western United  States. Quoted
        October 1999                             4-42       DRAFT-DO NOT QUOTE OR CITE

-------
 1      uncertainties in source apportionments of constituents in ambient aerosol samples typically range
 2      from 10 to 50%. It is apparent that a relatively small number of source categories, compared to
 3      the total number of chemical species which are typically measured in ambient monitoring-source
 4      receptor model studies, are needed to account for most of the observed mass of PM in these
 5      studies. Again, it should be emphasized that, because of limitations in receptor modeling
 6      methods in treating secondary components, these efforts are more likely to be successful for
 7      primary components, although it should be mentioned that methods are being developed to
 8      apportion secondary constituents by source categories.
 9           Windblown dust from whatever source represents the largest single source of PM25 in the
10      U.S. emissions inventory.  Although dust emissions (62% of total U.S. PM25) are far in excess of
11      any other source of primary or secondary PM2 5 in any region of the country, measurements of
12      soil constituents in ambient samples suggest that the overall contribution from this source could
13      be less than 10%.  The reasons for this apparent discrepancy are not clear. In addition to errors in
14      inventories or source apportionments, weather-related factors (wind speed and ground wetness)
15      and the dominance of local sources on spatial scales too small to be captured in inventories may
16      be involved.  It should be remembered that dust emissions  are widely dispersed and highly
17      sporadic.  Dust particles also have short atmospheric residence times and, as a result, their
18      dominance in emissions inventories may not be reflected in samples collected away from the
19      sources of the dust.
20           As seen in Table 4-3, emissions of mineral dust, organic debris, and sea spray are
21      concentrated mainly in the coarse  fraction of PM10 ( > 2.5 //m aero. diam.). A small fraction of
22      this material  is in the PM25 size range ( < 2.5 //m aero. diam.).  Nevertheless, concentrations of
23      crustal material can be appreciable especially during dust events.  It should also be remembered
24      that virtually all of the Saharan dust reaching the United States is in the PM2 5 size range.
25      Emissions from combustion sources (mobile and stationary sources, biomass burning) are
26      predominantly in the PM2 5 size range.
27           Uncertainties in emissions inventories are difficult to quantify. They may be as low as 10%
28      for well-defined sources (e.g., for  SO2) and may range up to a factor of 10 or so for windblown
29      dust.  As a rule, total PM emissions rates should be regarded as order-of-magnitude estimates.
30      Because of the large uncertainty associated with emissions of suspended dust, trends of total


        October 1999                             4-43        DRAFT-DO NOT QUOTE OR CITE

-------
 1     PM10 emissions should be viewed with caution and emissions from specific source categories are
 2     best discussed on an individual basis.
 3           Although most emphasis in this chapter has been on sources within the United States,
 4     it should also be remembered that sources outside the United States also contribute to ambient
 5     PM levels that can, at times, result in exceedances of the ambient NAAQS for PM.  Saharan dust
 6     storms contribute routinely to PM loadings in areas east of the Mississippi River. The results of
 7     Perry et al. (1997) indicate that highest concentrations of mineral dust in the PM25 fraction are
 8     found in the eastern United States during the summer and not in arid areas of the western
 9     United States. Large scale dust storms in the deserts of central Asia have recently been found to
10     contribute to PM levels in the Northwest on an episodic basis. Uncontrolled biomass burning in
11     central America  and Mexico resulted in exceedances of the daily NAAQS  for PM in Texas.
12
       October 1999                             4-44        DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES
  2
  3       AIRS (1995).
  4       Cahill, T. A. (1996) [Data from IMPROVE data base.] Fort Collins, CO: Air Resource Specialists, Inc.
  5       California Air Resources Board. (1995) [Data base provided by John R. Holmes, Chief, Research Division.]
  6             Sacramento, CA: California Air Resources Board.
  7       Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality: some
  8             reflections on controlling vehicle emissions. Science (Washington, DC) 261: 37-45.
  9       CAPITA (1999) Available: http://capita.wustl.edu/CAPITA/.
10       Chow, J. C.; Watson, J. G.; Frazier, C. A.; Egami, R. T.; Goodrich, A.; Ralph, C. (1988) Spatial and temporal
11             source contributions to PM10 and PM2.5 in Reno, NV. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10:
12             implementation of standards, an APCA/EPA international specialty conference; February; San Francisco,
13             CA. Pittsburgh, PA: Air Pollution Control Association; pp. 438-457. (APCA transactions series TR-13).
14       Chow, J. C.; Watson, J. G.; Richards, L. W.; Haase, D. L.; McDade, C.; Dietrich, D. L.; Moon, D.; Sloane, C.
15             (1991) The 1989-90 Phoenix PM10 study, volume II: source apportionment final report. Phoenix, AZ:
16             Arizona Department of Environmental Quality; final report no. 8931.6F1; Table 3-1.
17       Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Frazier, C. A.; Hinsvark, B. A.; Prichett, L. C.; Neuroth, G. R.
18             (1992a) Wintertime PM 10 and PM2 5 chemical compositions and source contributions in Tucson, Arizona.
19             In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source controls, an
20             A&WMA/EPA international specialty conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air &
21             Waste Management Association; pp. 231-243. (A&WMA transactions series no. 22).
22       Chow, J. C.; Watson, J. G.; Lowenthal, D. H.; Solomon, P. A.; Magliano, K. L.; Ziman, S. D.; Richards, L. W.
23             (1992b) PM10 source apportionment in California's San Joaquin Valley. Atmos. Environ. Part A
24             26: 3335-3354.
25       Chow, J. C.; Liu, C. S.; Cassmassi, J.; Watson, J. G.; Lu, Z.; Pritchett, L. C. (1992c) A neighborhood-scale study of
26             PM10 source contributions in Rubidoux, California. Atmos. Environ. Part A 26: 693-706.
27       Cooper, J. A.; Watson, J. G., Jr. (1980) Receptor oriented methods of air particulate source apportionment. J. Air
28             Pollut. Control Assoc. 30: 1116-1125.
29       Darlington, T. L.; Kahlbaum, D. F.; Heuss, J. M.; Wolff, G. T. (1997) Analysis of PM10 trends in the United States
30             from 1988 through 1995. J. Air Waste Manage. Assoc. 47: 1070-1078.
31       Dattner, S. L.; Hopke, P. K. (1982) Receptor models applied to contemporary pollution problems: an introduction.
32             In: Frederick, E. R., ed. A specialty conference on: receptor models applied to contemporary pollution
33             problems; October; Danvers, MA. Pittsburgh, PA: Air Pollution Control Association; pp. 1-5. (APCA
34             specialty conference proceedings SP-48).
35       Duce, R. A. (1995) Sources, distributions, and fluxes of mineral aerosols and their relationship to  climate.
36             In: Charlson, R. J.; Heintzenberg, J., eds. Aerosol forcing of climate: report of the Dahlem  workshop on
37             aerosol forcing of climate; April 1994; Berlin, Federal Republic of Germany. Chichester, United Kingdom:
38             John Wiley & Sons, Ltd.; pp. 43-72.
39       Dzubay, T. G.; Stevens, R. K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E.; Courtney, W. J. (1988)  A composite
40             receptor method applied to Philadelphia aerosol. Environ. Sci. Technol. 22: 46-52.
41       Eldred, R. A.; Cahill, T. A. (1994) Trends in elemental concentrations of fine particles at remote sites in the United
42             States of America. Atmos. Environ. 28: 1009-1019.
43       Gatz, D. F.; Prospero, J. M. (1996) A large silicon-aluminum aerosol plume in central Illinois: North African Desert
44             dust? Atmos. Environ. 30: 3789-3799.
45       Gillette, D. A.; Hanson, K. J. (1989) Spatial and temporal variability of dust production caused by wind erosion in
46             the United States. J. Geophys. Res. [Atmos.] 94: 2197-2206.
47       Glover, D. M.; Hopke, P. K.; Vermette, S. J.; Landsberger, S.; DAuben, D. R. (1991) Source apportionment with
48             site specific source profiles. J. Air Waste Manage. Assoc. 41: 294-305.
49       Gordon, G. E. (1980) Receptor models. Environ. Sci. Technol. 14: 792-800.
50       Gordon, G. E. (1988) Receptor models. Environ. Sci. Technol. 22: 1132-1142.
51
         October 1999                                    4-45         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Gray, H. A.; Landry, B.; Liu, C. S.; Henry, R. C.; Cooper, J. A.; Sherman, J. R. (1988) Receptor modeling for
  2             PM10 source apportionment in the South Coast Air Basin of California. In: Mathai, C. V.; Stonefield, D. H.,
  3             eds. PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
  4             San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 399-418. (APCA transactions
  5             series TR-13).
  6       Griffing, G. W. (1980) Relations between the prevailing visibility, nephelometer scattering coefficient and
  7             sunphotometer turbidity coefficient. Atmos. Environ. 14: 577-584.
  8       Hopke, P. K. (1985) Receptor modeling in environmental chemistry. New York, NY: John Wiley & Sons.
  9       Hopke, P. K., ed. (1991) Receptor modeling for air quality management. Amsterdam, The Netherlands: Elsevier.
10             (Vandeginste, B. G. M.; Kvalheim, O. M., eds. Data handling in science and technology: v. 7).
11       Hopke, P. K.; Wlaschin, W.; Landsberger, S.; Sweet, C.; Vermette, S. J. (1988) The source apportionment of PM10
12             in south Chicago. In: Mathai, C. V.; Stonefield, D. H., eds. PM-10: implementation of standards, an
13             APCA/EPA international specialty conference; February; San Francisco, CA. Pittsburgh, PA: Air Pollution
14             Control Association; pp. 484-494. (APCA transactions series TR-13).
15       Houck, J. E.; Rau, J. A.; Body, S.; Chow, J. C. (1992) Source apportionment:  Pocatello, Idaho PM 10 nonattainment
16             area. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source controls, an
17             A&WMA/EPA international specialty conference, v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air &
18             Waste Management Association; pp. 219-230. (A&WMA transactions  series no. 22).
19       Huang, X.; Olmez, I.; Aras, N. K.; Gordon, G. E. (1994) Emissions of trace elements from motor vehicles:
20             potential marker  elements and source composition profile. Atmos. Environ. 28: 1385-1391.
21       Husar, R. B.; Wilson, W. E. (1993) Haze and sulfur emission trends in the eastern United States. Environ. Sci.
22             Technol. 27: 12-16.
23       Husar, R. B.; Holloway, J. M.; Patterson, D. E.; Wilson, W. E. (1981) Spatial  and temporal pattern of eastern U.S.
24             haziness: a summary. Atmos. Environ. 15: 1919-1928.
25       Husar, R. B.; Elkins, J. B.; Wilson, W. E. (1994) U.S. visibility trends, 1960-1992, regional and national. Presented
26             at: 87th annual meeting of the Air & Waste Management Association; June; Cincinnati, OH. Pittsburgh, PA:
27             Air & Waste Management Association; paper no. 94-MP3.05.
28       Inhalable Particulate Network. (1985) [Inhalable Particulate Network Database.] Research Triangle Park, NC: U.S.
29             Environmental Protection Agency, National Exposure Research Laboratory; Jose M. Sune, contact.
30       Javitz, H. S.; Watson, J. G.; Guertin, J. P.; Mueller, P. K. (1988) Results of a receptor modeling feasibility study.
31             JAPCA 38: 661-667.
32       Kim, B. M.; Zeldin, M.  D.; Liu, C. S. (1992) Source apportionment study for state implementation plan
33             development in the Coachella Valley. In: Chow,  J. C.; Ono, D. M., eds. PM 10 standards and nontraditional
34             particulate source controls: an A&WMA/EPA international specialty conference, v. II; January; Scottsdale,
35             AZ. Pittsburgh, PA: Air & Waste Management Association; pp. 979-991. (A&WMA transactions series
36             no. 22).
37       Koutrakis, P. (1995) [PM25 and PM10 data for Philadelphia, 1992-1995]. Boston, MA: Harvard School of Public
38             Health.
39       Lawson, D. R.; Smith, R. E. (1998) The northern front range air quality study: a report to the Governor and General
40             Assembly. Fort Collins, CO: Colorado State University;  December.
41       Macias, E. S.; Hopke, P. K., eds. (1981) Atmospheric aerosol: source/air quality relationships; based on a
42             symposium jointly sponsored by the Divisions of Nuclear Chemistry and Technology at the 180th national
43             meeting of the American Chemical Society; August 1980; Las Vegas, NV. Washington,  DC: American
44             Chemical Society. (Comstock, M. J., ed. ACS symposium series: 167).
45       Magliano, K. L. (1988)  Level 1 PM 10 assessment in a California air basin. In:  Mathai, C. V.; Stonefield, D. H., eds.
46             PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
47             San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 508-517. (APCA transactions
48             series TR-13).
49       Neas, L. M. (1996) [Dataprovided for 1986 from Harvard six cities database.] Boston, MA: Harvard School of
50             Public Health.
51       Ott, S.-T.; Ott, A.; Martin, D. W.; Young, J. A. (1991) Analysis of a trans-Atlantic Saharan dust outbreak based on
52             satellite and GATE data. Mon. Weather Rev. 119:1832-1850.
53       Pace, T. G., ed. (1986) Receptor methods for source apportionment: real world issues  and applications, an APCA
54             international specialty conference; March 1985; Williamsburg, VA. Pittsburgh, PA: Air Pollution Control
55             Association. (APCA transactions no. TR-5).


         October 1999                                    4-46         DRAFT-DO NOT  QUOTE OR CITE

-------
  1       Perry, K. D.; Cahill, T. A.; Eldred, R. A.; Dutcher, D. D.; Gill, T. E. (1997) Long-range transport of North African
  2             dust to the eastern United States. J. Geophys. Res. [Atmos] 102: 11,225-11,238.
  3       Pinto, J. P.; Pleasant, M.; Kellogg, R.; Torok, Sz.; Stiles, D.; Willis, R. D.; Koutrakis, P.; Allen, G. A. (1995)
  4             Chemical characterization of the ambient aerosol in Philadelphia. Presented at: Particulate matter: health and
  5             regulatory issues: an A&WMA international specialty conference; April; Pittsburgh, PA. Pittsburgh, PA:
  6             Air & Waste Management Association; submitted abstract.
  7       Placet, M.; Battye,  R. E.; Fehsenfeld, F. C.; Bassett, G. W. (1991) Emissions involved in acidic deposition
  8             processes. In: Irving, P. M., ed. Acidic deposition: state of science and technology, volume I: emissions,
  9             atmospheric processes, and deposition. Washington, DC: The U.S. National Acid Precipitation Assessment
10             Program. (State of science and technology report no. 1).
11       Prospero, J. M. (1995) The atmospheric transport of particles to the ocean. In: Ittekkot, V.; Honjo,  S.; Depetris,
12             P. J., eds. SCOPE report: particle flux in the ocean. New York, NY: John Wiley & Sons, Ltd.; in press.
13       Prospero, J. M. (1999) Long-range transport of mineral dust in the global atmosphere: impact of African dust on the
14             environment of the southeastern United States. Proceed. Natl. Acad. Sci. 96: 3396-3403.
15       Prospero, J. M.; Nees, R. T.; Uematsu, M. (1987) Deposition rate of particulate and dissolved aluminum derived
16             from Saharan dust in precipitation at Miami, Florida. J. Geophys. Res. [Atmos.] 92: 14,723-14,731.
17       Rodes, C. E.; Evans, E. G. (1985) Preliminary assessment of 10 pan particulate sampling at eight locations in the
18             United States. Atmos. Environ. 19: 293-303.
19       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
20             aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
21             Technol. 27:636-651.
22       Ryan, W. M.; Badgett-West, C. R.; Holtz, D. R.; Peters, T. A.; Cooper, J. A.; Ono, D. (1988) Reconciliation of
23             receptor and dispersion modeling impacts of PM 10 in Hayden, Arizona. In: Mathai, C. V.; Stonefield, D. H.,
24             eds.  PM-10: implementation of standards, an APCA/EPA international specialty conference; February;
25             San Francisco, CA. Pittsburgh, PA: Air Pollution Control Association; pp. 419-429. (APCA transactions
26             series TR-13).
27       Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass, G. R. (1996) Source apportionment of
28             airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
29       Skidmore, L. W.; Chow, J. C.; Tucker, T. T. (1992) PM 10 air quality assessment for the Jefferson County, Ohio air
30             quality control region. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional particulate source
31             controls, an  A&WMA/EPA international specialty conference, v. II; January; Scottsdale, AZ. Pittsburgh, PA:
32             Air and Waste Management Association; pp. 1016-1031. (A&WMA transactions series no.  22).
33       Spengler, J. D.; Briggs, S. L. K.; Ozkaynak, H. (1986) Relationship between TSP measurements and
34             size-fractionated particle mass measurements in six cities participating in the Harvard Air Pollution Health
35             Study. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning
36             and Standards; EPA cooperative agreement no. CR-812667-01-0; December 5.
37       Stevens and Lewis  (1984).
38       Stevens, R. K.; Pace, T. G. (1984) Overview of the mathematical and empirical receptor models workshop
39             (Quail Roost II). Atmos. Environ.  18: 1499-1506.
40       Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1995) Spatial variation in acidic sulfate and ammonia
41             concentrations within metropolitan Philadelphia.  J. Air Waste Manage. Assoc. 45:  442-452.
42       Thanukos, L. C.; Miller, T.; Mathai, C. V.; Reinholt, D.; Bennett, J. (1992) Intercomparison of PM 10 samplers and
43             source apportionment of ambient PM 10 concentrations at Rillito, Arizona. In: Chow, J. C.; Ono, D. M., eds.
44             PM10 standards and nontraditional particulate source controls, an A&WMA/EPA international specialty
45             conference,  v. I; January; Scottsdale, AZ. Pittsburgh, PA: Air & Waste Management Association; pp.
46             244-261. (A&WMA transactions series no. 22).
47       U.S. Environmental Protection Agency. (1994) National air quality and emissions trends report, 1993. Research
48             Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-94/026. Available
49             from: NTIS, Springfield, VA; PB95-159893.
50       U.S. Environmental Protection Agency. (1995) Compilation of air pollutant emission factors. Volume I: stationary
51             point and area sources. 5th ed. Research Triangle Park, NC: Office of Air Quality Planning and Standards;
52             report no. AP-42-ED-5. Available from: NTIS, Springfield, VA; PB95-196028.
53       U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research  Triangle Park,
54             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001 aF-cF. 3v.
55             Available from: NTIS, Springfield, VA; PB96-168224.


         October 1999                                    4-47         DRAFT-DO NOT QUOTE OR CITE

-------
  1       U.S. Environmental Protection Agency. (1998) National air quality and emissions trends report, 1997. Research
  2             Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA 454/R-98-016. Available:
  3             www.epa.gov/oar/aqtrnd97/.
  4       Vermette, S. J.; Williams, A. L.; Landsberger, S. (1992) PM 10 source apportionment using local surface dust
  5             profiles: examples from Chicago. In: Chow, J. C.; Ono, D. M., eds. PM 10 standards and nontraditional
  6             particulate source controls, an A&WMA/EPA international specialty conference, v. I; January; Scottsdale,
  7             AZ. Pittsburgh, PA: Air & Waste Management Association; pp. 262-278. (A&WMA transactions series
  8             no. 22).
  9       Watson, J. G.; Henry, R. C.; Cooper, J. A.; Macias, E. S. (1981) The state of the art of receptor models relating
10             ambient suspended particulate matter to sources. In: Macias, E. S.; Hopke, P. K., eds. Atmospheric aerosol:
11             source/air quality relationships; based on a symposium jointly sponsored by the Divisions of Nuclear
12             Chemistry and Technology and Environmental Chemistry at the 180th national meeting of the American
13             Chemical Society; August 1980; Las Vegas, NV. Washington, DC: American Chemical Society; pp. 89-106.
14             (Comstock, M. J., ed. ACS symposium series: 167).
15       Watson, J. G.; Chow, J. C.; Mathai, C. V. (1989) Receptor models in air resources management: a summary of the
16             APCA international specialty conference. JAPCA 39:  419-426.
17       Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Burns, S.; Ragazzi, R. A. (1990a) Composite  source
18             profiles for particulate motor vehicle exhaust source apportionment in Denver,  CO. In: Mathai, C. V., ed.
19             Visibility and fine particles: an A&WMA/EPA international specialty conference; October 1989; Estes Park,
20             CO. Pittsburgh, PA: Air & Waste Management Association; pp. 422-436. (A&WMA transactions series
21             no. TR-17).
22       Watson, J. G.; Chow, J. C.; Lowenthal, D. H.; Pritchett, L. C.; Frazier, C. A.; Neuroth, G. R.; Robbins, R. (1994a)
23             Differences in the carbon composition of source profiles for diesel- and gasoline-powered vehicles.
24             Atmos. Environ. 28: 2493-2505.
25       Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
26             epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
27       Wongphatarakul(1997).
28       Wongphatarakul, V.; Friedlander, S. K.; Pinto, J. P. (1998) A comparative study of PM 2 5 ambient aerosol  chemical
29             databases. Environ. Sci. Technol. 32: 3926-3934.
         October 1999                                   4-48         DRAFT-DO NOT QUOTE OR CITE

-------
 i                                     APPENDIX 4A
 2
 3                            Composition of PM Source Emissions
 4
 5
 6          This appendix includes figures and tables showing the elemental composition of emissions
 7      from various source categories discussed in Table 4A-3. The material is mainly derived from
 8      Chapter 5 of CD96, to which the reader is referred for further details. The primary emphasis in
 9      the figures and tables is on the source composition of PM2 5 particles.
10          The compositions of soils and average crustal material are shown in Table 4A-1 (adapted
11      from Warneck,  1988). Two entries are shown as representations of average crustal material.
12      Differences from the mean soil composition shown can result from local geology and climate
13      conditions. Major elements in both soil and crustal profiles are Si, Al, and Fe which are found in
14      the form of various minerals. In addition, organic matter constitutes a few percent, on average,
15      of soils.  In general, the soil  profile is similar to the crustal profiles, except for the depletion of
16      soluble elements such as Ca, Mg, Na, and K. It should be noted that the composition of soils
17      from specific locations can vary considerably from these global averages, especially for elements
18      like Ca, Mg, Na, and K.
19          Fugitive dust emissions arise from paved and unpaved roads, building construction and
20      demolition, parking lots, mining operations, storage piles, and agricultural tilling in addition to
21      wind erosion. Figure 4A-1 shows examples of size distributions in dust from paved and unpaved
22      roads, agricultural soil, sand and gravel, and alkaline  lake bed sediments which were measured in
23      a laboratory resuspension chamber as part of a study in California (Chow et al., 1994). This
24      figure shows substantial variation in particle size among some of these fugitive dust sources.
25      The PMj 0 abundance (6.9%) in the total suspended PM (TSP) from alkaline lake bed dust is
26      twice its abundance in paved and unpaved road dust.  Approximately 10% of the TSP is in the
27      PM25 fraction and approximately 50% of TSP is in the PM10 fraction.  The sand/gravel dust
28      sample shows that 65% of the mass is in particles larger than the PM10 fraction. ThePM25
29      fraction of TSP is approximately 30% to 40% higher  in alkaline lake beds and sand/gravel than
30      in the other soil types. The tests were performed after seiving and with a short (<1 min) waiting
31      period prior to sampling.  It  is expected that the fraction of PMj 0 and PM25 would increase with
        October 1999                             4A-1        DRAFT-DO NOT QUOTE OR CITE

-------
                TABLE 4A-1. AVERAGE ABUNDANCES OF MAJOR ELEMENTS IN
                                    SOIL AND CRUSTAL ROCK
Elemental Abundances (ppmw)
Element
Si
Al
Fe
Ca
Mg
Na
K
Ti
Mn
Cr
V
Co
Soil
(a)
330,000
71,300
38,000
13,700
6,300
6,300
13,600
4,600
850
200
100
8
Crustal Rock
(b)
277,200
81,300
50,000
36,300
20,900
28,300
25,900
4,400
950
100
135
25

(c)
311,000
77,400
34,300
25,700
33,000
31,900
29,500
4,400
670
48
98
12
        Source: (a) Vinogradov (1959); (b) Mason (1966); (c) Turekian (1971), Model A; as quoted in Warneck (1988).
 1     distance from a fugitive dust emitter as the larger particles deposit to the surface faster than do
 2     the smaller particles.
 3          The size distribution of samples of paved road dust obtained from a source characterization
 4     study in California is shown in Figure 4A-2.  As might be expected, most of the emissions are in
 5     the coarse size mode.  The chemical composition of paved road dust obtained in Denver, CO,
 6     during the winter of 1987-1988 is shown in Figure 4A-3. The chemical composition of paved
 7     road dust consists of a complex mixture of particulate matter from a wide variety of sources.
 8     Hopke et al. (1980) found that the inorganic composition of urban roadway dust in samples from
 9     Urbana, IL could be described in terms of contributions from natural soil, automobile exhaust,
10     rust, tire wear, and salt. Automobile contributions arose from exhaust emissions enriched in Pb;
       October 1999                             4A-2       DRAFT-DO NOT QUOTE OR CITE

-------
             100
              80
              60
            (0
              40
              20
  Paved
Road Dust
        Unpaved
       Road Dust
                                         Agricultural
                                            Soil
C3<1.0|jm
                                     2.5 |
                                                               Soil/Gravel

                                                               OTSP
Alkaline
Lake Bed
       Figure 4A-1.  Size distribution of particles generated in a laboratory resuspension
                     chamber.
       Source: Chow et al. (1994).
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from rust as Fe; tire wear particles enriched in Zn; brake linings enriched in Cr, Ba, and Mn; and
cement particles derived from roadways by abrasion. In addition to organic compounds from
combustion and secondary sources, road dust also contains biological material such as pollen and
fungal spores.
     The elemental composition of primary particulate matter emitted in the fine fraction from a
variety of power plants and industries in the  Philadelphia area is shown in Table 4A-2 as a
representative example of emissions from stationary fossil combustion sources (Olmez et al.,
1988). Entries for the coal fired power plant show that Si and Al followed by sulfate are the
major primary constituents produced by coal combustion, while fractional abundances of
elemental carbon were much lower and organic carbon species were not detected.  Sulfate is the
major particulate constituent released by the oil fired power plants examined in this study; and,
       October 1999
                                         4A-3
                                 DRAFT-DO NOT QUOTE OR CITE

-------
        100
         80
         60
       0.
       (0
       0>
       Q.
         40
         20
                   52.3%
                               92.8%
                               82.7%
                               (<2.5M)
                               81.6%
                                           95.8%
 93.1%
 (<2.5M
 92.4%
96.2%
(<10M)
92.3%
(<2.5M)
91.8%
99.2%


97.4%
(<2.5M)


87.4%
                                                                              34.9%
                                                                                10M)
            Road and   Agricultural   Residential     Diesel
            Soil Dust     Burning      Wood       Truck
                                   Combustion   Exhaust
                 Crude Oil  Construction
                Combustion    Dust
              Code:
 Figure 4A-2.   Size distribution of California source emissions, 1986.

 Source: Houcketal. (1989, 1990).
October 1999
4A-4
DRAFT-DO NOT QUOTE OR CITE

-------
          10'
          10
          10'
        8
       T5
        C
        D
       .a
          10
          10
          10
             2
             2
             -3

                                         Chemical Species
      Figure 4A-3. Chemical abundances for PM25 emissions from paved road dust in Denver,
                   CO. Solid bars represent fractional abundances, and the error bars
                   represent variability in species abundances. Error bars represent detection
                   limits when there are no solid bars.
      Source:  Watson and Chow (1994).
1
2
3
4
5
6
7
again, elemental and organic carbon are not among the major species emitted. Olmez et al.
(1988) also compared their results to a number of similar studies and concluded that their data
could have much wider applicability to receptor model studies in other areas with some of the
same source types. The high temperature of combustion in power plants results in the almost
complete oxidation of the carbon in the fuel to CO2 and very small amounts of CO. A number of
trace elements are greatly enriched over crustal abundances (in different fuels), such as Se in coal
      October 1999
                                        4A-5
DRAFT-DO NOT QUOTE OR CITE

-------
o
o_
o
o
l-l
TABLE 4A-2. COMPOSITION OF FINE PARTICLES RELEASED BY
SOURCES IN THE PHILADELPHIA AREA
Sneries
Fddvstnne Cnal-
Oil-Fired Power Plants
^ (Units) Fired Power Plant N
"vO








[V
^^
1
ON

O
£>
Tl
H
O
O
0
H
0
c
o
H
W
O

O
H
W

C-v (%)
C-e (%)
NH4 (%)
Na (%)
Al (%)
Si (%)
P (%)
S (%)
SO4 (%)
Cl (%)

K(%)
Ca (%)
Sc (ppm)
Ti (%)
V (ppm)

Cr (ppm)
Mn (ppm)
Fe (%)
Co (ppm)

Ni (ppm)




ND
0.89 ±0.12
1.89 ±0.19
0.31 ±0.03
14 ±2
21.8±1.6
0.62 ±0.10
3.4 ±0.6
11.9±1.2
0.022 ±0.11

1.20 ±0.09
1.4 ±0.5
42 ±2
1.1 ±0.2
550 ± 170

390 ±120
290 ±15
7.6 ±0.4
93 ±10

380 ±50





3
3
3
3
9
9
9
3
3

9
3
3
3
3

3
3
3
3

9



Eddystone

2.7 ± 1.2
7.7 ±1.5
3.5 ±1.6
3.0 ±0.8
0.45 ± 0.09
1.9 ±0.6
1.5 ±0.4
11±2
40 ±4
0.019 ±0.009

0.16 ±0.05
3.6±1.0
0.17 ±0.02
0.040 ± 0.044
11500 ±3000

235 ±10
380 ±40
1.6 ±0.2
790 ±150

15000 ±5000



N

3
3
3
3
3
9
9
9
3
2

9
3
3
9
3

3
3
3
3

9



Schuylkill

0.75 ± 0.63
0.22 ±0.1 7
3.7 ±1.7
3.3 ±0.8
0.94 ± 0.08
2.6 ±0.4
1.0 ±0.2
13± 1
45 ±7
ND

0.21 ±0.03
2.3 ±1.0
0.47 ± 0.02
0.12 ±0.02
20000 ± 3000

230 ±70
210±50
1.7 ±0.4
1100 ±200

19000 ±2000



N

4
4
4
3
3
11
11
11
4


11
3
3
11
3

3
3
3
3

11



Secondary
Al Plant

1.6±1.5
0.18 ±0.10
2.2 ± 0.9
16.3 ±0.8
1.74 ±0.09
3.1 ±2.2
0.45 ± 0.27
3±4
5.9 ±2
21±4

10.9 ± 1.5
0.12 ±0.09
0.092 ± 0.039
0.024 ± 0.003
36 ±7

410±20
120± 15
0.31 ±0.02
13±2

300 ±100



VARIOUS STATIONARY

N

2
2
2
1
1
2
2
2
2
1

2
2
1
2
1

1
1
1
1

2



Fluid Cat.
Cracker

ND
0.16 ±0.05
0.43 ± 0.22
0.38 ± 0.05
6.8 ±1.2
9.8 ±20.0
ND
4.2 ± 12.6
38 ±4
ND

0.031 ±0.005
0.030 ± 0.004
2.7 ± 0.4
0.38 ±0.1
250 ±70

59 ±8
14±3
0.20 ± 0.03
15±2

220 ±30




N


3
3
3
3
9

9
3


9
9
3
3
3

3
3
9
3

9



Municipal
Incinerator

0.57 ±0.26
3.5 ±0.2
0.36 ±0.07
6.6 ±3.5
0.25 ±0.10
1.7 ±0.3
0.63 ±0.12
2.9 ±0.8
6.8 ±2.3
29 ±5

7.6 ±2.3
0.23 ±0.10
0.11 ±0.02
0.030 ±0.015
8.6 ±5.3

99 ±31
165 ±40
0.22 ± 0.05
3.7 ±0.8

290 ± 40




N

4
4
4
3
3
10
10
10
4
3

10
10
1
10
2

3
3
3
3

10




-------
0
o
o
o*
fD
•O
•O
•O








£
1

a
§
H
a
0
o
H
O
c
o
H
W
O
o
HH
H
W
TABLE 4A-2 (cont'd).

Species
(Units)
Cu (ppm)
Zn (%)
As (ppm)
Se (ppm)
Br (ppm)
Rb (ppm)
Sr (ppm)
Zr (ppm)
Mo (ppm)
Ag (ppm)
Cd (ppm)
In (ppm)
Sn (ppm)
Sb (ppm)

Cs (ppm)

Ba (ppm)
La (ppm)

Ce (ppm)
Nd (ppm)
Sm (ppm)


Eddystone
y-1 1 T7:,-.-.,J
Coal-rired
Power Plant
290 ± 20
0.041 ± 0.005
640 ± 80
250 ±20
35 ±8
190 ±80
1290 ±60
490 ±190
170 ±60
ND
ND
0.71 ± 0.04
ND
(a)

9.2 ±0.9

ND
120 ±10

180±10
80 ±26
23 ±2


COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS STATIONARY
SOURCES IN THE PHILADELPHIA AREA
Oil-Fired Power Plants

N
9
3
3
3
3
1
9
9
2


2



2


3

2
3
3



Eddystone
980 ±320
1.3 ±0.3
33 ±6
26 ±9
90 ±60
ND
160 ±50
140 ±180
930 ±210
ND
ND
ND
320 ±230
370 ±410

ND

1960 ± 100
130±30

89 ±23
28 ±5
3.7 ±0.7



N
9
3
1
3
9

9
9
3



9
3



3
3

3
2
3



Schuylkill
1100 ±500
0.78 ±0.30
50 ±16
23 ±7
45 ±17
ND
280 ±70
100 ±120
1500 ±300
ND
ND
ND
200 ± 80
1020 ±90

ND

2000 ± 500
450 ±30

360 ± 20
230 ±20
20.5 ±1.5



N
11
3
3
3
11

11
11
3



11
3



3
3

3
3
3



Secondary
Al Plant
450 ±200
0.079 ± 0.006
15±6
66 ±3
630 ± 70
97 ±38
ND
ND
ND
ND
ND
ND
550 ±540
6100 ±300

ND

ND
19±2

ND
ND
ND




N Fluid Cat. Cracker N
2 14 ±8 9
1 0.0026 ± 0.0007 3
1 ND
1 15 ±1 3
2 5.6±1.8 9
1 ND
36 ±6 9
130 ±50 2
ND
ND
ND
ND
2 ND
1 7.7 ±1.5 3

ND

290 ±90 2
1 3300 ±500 3

2700 ± 400 3
1800 ±250 3
170 ±20 3



Municipal
Incinerator
1300 ±500
10.4 ±0.5
64 ±34
42 ± 16
2300 ± 800
230 ±50
87 ± 14
ND
240 ±130
71 ± 15
1200 ± 700
4.9 ±1.4
6700 ± 1900
1300± 1000

5.9 ±3.0

ND
1.1 ±0.5

ND
ND
ND




N
3
3
3
3
10
2
10

10
3
3
3
10
3

3


1







-------
                     TABLE 4A-2 (cont'd). COMPOSITION OF FINE PARTICLES RELEASED BY VARIOUS
o
r+
O
cr
fD
•O
•O
•O








^
i
oo

STATIONARY SOURCES IN THE PHILADELPHIA AREA
Species
(Units)
Eu (ppm)
Gd (ppm)
Tb (ppm)
Yb (ppm)
Lu (ppm)
Hf(ppm)
Ta (ppm)
W (ppm)
Au (ppm)
Pb (%)
Th (ppm)
% mass
Eddystone Coal-
Fired Power
Plant
5.1 ±0.5
ND
3.3 ±0.3
10.3 ±0.5
ND
5. 8 ±0.8
ND
20 ±8
ND
0.041 ± 0.004
24 ±2
24 ±2
Oil-Fired Power Plants
N
3

3
1

3

1

9
3
6
Eddystone N
ND
ND
ND
ND
ND
0.39 ±0.07 1
ND
60 ±5 2
0.054 ±0.017 2
1.8 ±0.6 9
1.9 ±0.5 2
93.5 ±2.5 6
Schuylkill
0.65 ± 0.23
ND
0.90 ± 0.29
ND
ND
ND
ND
ND
ND
1.0 ±0.2
ND
96 ±2

Secondary
N Al Plant
3 ND
ND
3 ND
ND
ND
ND
ND
ND
ND
11 0.081 ±0.014
ND
6 81 ±10
Fluid Cat.
N Cracker
4.9 ±0.7
71 ±10
8.9 ±1.3
3.7 ±0.4
0.59 ±0.17
0.99 ±0.08
0.56 ±0.10
ND
ND
2 0.0091 ± 0.0021
6.2 ± 0.7
2 97 ±2
N
3
3
3
3
3
3
3


9
3
7
Municipal
Incinerator
ND
ND
ND
ND
ND
ND
ND
ND
0.56 ±0.27
5. 8 ±1.2
ND
89 ±2
N








3
10

7
o
O
2
O
H
N = Number of samples.
ND = Not detected.
The "% mass" entries give the average percentage of the total emitted mass found in the fine fraction.
(a) Omitted because of sample contamination.

Source: Adapted from Olmez et al. (1988).
o
H
W
O
W
O
HH
H
W

-------
 1      and V and Ni in oil.  In fact, the higher V content of the fuel oil than in coal could help account
 2      for the higher sulfate seen in the profiles from the oil-fired power plant compared to the
 3      coal-fired power plant since V at combustion temperatures is known to catalyze the oxidation of
 4      reduced sulfur species. Although Table 4A-2 only gives values of the fine particle composition,
 5      measurements of coarse particle composition were also reported by Olmez et al. (1988) which
 6      were qualitatively similar.
 7           Apart from emissions in the combustion of fossil fuels, toxic trace elements are emitted as
 8      the result of various industrial processes such as steel and iron manufacturing and non-ferrous
 9      metal production (e.g., for Pb, Cu, Ni, Zn, and Cd). As may be expected, emissions factors for
10      the various trace elements are highly source-specific (Nriagu and Pacyna, 1988). Inspection of
11      Table 4A-3 reveals that the emissions from the catalytic cracker and the oil-fired power plant are
12      greatly enriched in rare-earth elements such as La compared to other sources.
13           Emissions from municipal waste incinerators are heavily enriched in Cl arising mainly from
14      the combustion of plastics and metals that form volatile chlorides. The metals can originate from
15      cans or other metallic objects  and some metals such as Zn and Cd are also additives in plastics or
16      rubber. Many elements  such as S, Cl, Zn, Br, Ag, Cd, Sn, In, and Sb are enormously enriched
17      compared to their crustal abundances. A comparison of the trace elemental composition of
18      incinerator emissions in Philadelphia, PA (shown in Table 4A-2) with the composition of
19      incinerator emissions in Washington B.C., and Chicago, IL, (Olmez et al., 1988) shows
20      agreement for most constituents to better than a factor of two.
21           The principal components emitted by diesel and gasoline fueled vehicles are organic carbon
22      (OC) and elemental carbon (EC) as shown in Table 4A-3. As can be seen, the variability among
23      entries for an individual fuel type is large and overlaps that found between different fuel types.
24      On average, the abundance  of elemental carbon is larger than that of organic carbon in the
25      exhaust of diesel vehicles, while organic carbon is the dominant species in the exhaust of
26      gasoline fueled vehicles. Per  vehicle, total carbon emissions from light and heavy duty diesel
27      vehicles can range from one to two orders of magnitude higher than those from gasoline vehicles.
28      There appears to be a tendency for emissions of elemental carbon to increase relative to
29      emissions of organic carbon for gasoline fueled vehicles as simulated driving conditions are
30      changed from a steady 55 km /hr to the various load conditions specified in the Federal Test
31

        October 1999                             4A-9       DRAFT-DO NOT QUOTE OR CITE

-------
               TABLE 4A-3.  FRACTIONAL ORGANIC AND ELEMENTAL CARBON
                          ABUNDANCES IN MOTOR VEHICLE EMISSIONS
Fuel Type
Diesel
Denver, COa
Los Angeles, CAa
Bakersfield, CAb
Phoenix, AZb
Unleaded gasoline
Denver, COa
Los Angeles, CA°
Los Angeles, CAa
Phoenix, AZb
Leaded gasoline
Denver, COa
Los Angeles, CA°
Los Angeles, CAa
Mixed (tunnel and roadside)
Denver, CO
Los Angeles, CAd
Phoenix, AZ
Organic Carbon

23 ± 8%
36 ± 3%
49 ±13%
40 ± 7%

76 ± 29%
93 ± 52%
49 ± 10%
30 ± 12%

67 ± 23%
52 ± 4%
31 ±20%

50 ± 24%
38 ± 6%
39 ± 19%
Elemental Carbon

74 ±21%
52 ± 5%
43 ± 8%
33 ± 8%

18 ± 11%
5 ± 7%
39 ± %
14 ± 8%

16 ±7%
13 ± 1%
15 ±2%

28 ± 19%
38 ± 5%
36 ± 11%
Ne

3
2
3
8

8
11
11
9

3
3
3


3

Sources

1,2
3,4,5,6
7
8

1,2
3,4,5,6
3,4,5,6
8

1,2
3,4,5,6
3,4,5,6

1,2
3
8
       Sources:  (1) Watson et al. (1990b), (2) Watson et al. (1990a), (3) Cooper et al. (1987), (4) NBA, Inc. (1990),
               (5) Peltier et al. (1990a), (6) Peltier et al. (1990b), and (7) Houck et al. (1989), cited in (8) Watson et al.
               (1994b).

       Notes: (a) Modified Federal Test Procedures followed in dynamometer tests; (b) Roof monitoring at
             inspection station; (c) 55 km/hr steady speed in dynamometer tests; (d) Rt. 1 tunnel at LA airport,
             (e) N = Number of samples.
1      Procedures (FTPs). Also shown are the results of sampling from mixed vehicle types along

2      roadsides and in tunnels.
       October 1999
4 A-10
DRAFT-DO NOT QUOTE OR CITE

-------
 1           The results shown in Table 4A-3 were obtained during the late 1980's, and, so, may not be
 2      entirely representative of current vehicles. Examples of data for the trace element composition of
 3      motor vehicle emissions obtained in Phoenix, AZ are shown in Table 4A-4. SO2 emissions are
 4      also shown in relation to the mass of fine particles emitted.  As can be seen, small quantities of
 5      soluble ions such as SO4= and NH4+ are emitted. The ammonium may be emitted as the result of
 6      an improperly functioning catalytic converter, or may simply be the result of contamination
 7      during sample handling and analysis. Four fractions are given for the organic carbon fraction and
 8      three for elemental carbon. These refer to abundances measured at different temperatures in a
 9      thermographic analysis.  Temperatures for OC1, OC2, OC3, and OC4 are 120 °C, 250 °C,
10      450 °C, and 550 °C, respectively; and, for EC1, EC2, EC3, they are 550 °C, 700 °C, and
11      800 °C, respectively, in He/2% O2. The abundances of trace elements are all quite low, with
12      most being less than 1%.  It is not clear what the source of the small amount of Pb seen in the
13      auto exhaust profile is. It is extremely difficult to find suitable tracers for automotive exhaust
14      since Pb has been removed from gasoline. However, it should also be remembered that
15      restrictions in the use of leaded gasoline have resulted in a dramatic lowering of ambient Pb
16      levels. Examples of data for the trace elemental composition of the emissions from a number of
17      vehicle classes obtained as part of the North Frontal Range Air Quality Study (NFRAQS) which
18      took place in December of 1997 in Colorado are shown in Table 4A-5. As can be seen from
19      Table 4A-5, emissions of total carbon (TC), which is equal to the sum of organic carbon (OC)
20      and elemental carbon (EC), from gasoline vehicles are highly variable. Emissions from
21      "smokers", i.e., light duty vehicles with visible smoke emitted from their tailpipes, are
22      comparable to those from diesel vehicles.  Thus, older poorly maintained gasoline vehicles could
23      be significant sources  of PM25 (Sagebiel et al.,  1997; Lawson and Smith, 1998) in addition to
24      being significant sources of gaseous pollutants (e.g., Calvert et al., 1993). Durbin et al. (1999)
25      point out that although "smokers" constitute only 1.1 to 1.7% of the light duty fleet in the South
26      Coast Air Quality Management District in California, they contribute roughly 20% of the total
27      PM emissions from the light duty fleet. In general, motor vehicles which are high emitters of
28      hydrocarbons and carbon monoxide will also tend to be high emitters of PM (Sagebiel et al.,
29      1997; Cadle et al., 1997). Particle emission rates are also correlated with vehicle acceleration
30      and emissions occur predominantly during periods of heavy acceleration even in newer vehicles
31      (Maricqetal, 1999).

        October  1999                            4A-11       DRAFT-DO NOT QUOTE OR CITE

-------
 TABLE 4A-4. PHOENIX PM2S MOTOR VEHICLE EMISSIONS PROFILES (% MASS)
Chemical Species
NO3-
S042'
NH4+
OC
OC1
OC2
OC3
OC4
EC
EC1
EC2
ECS
Al
Si
P
S
Cl
K
Ca
Ti
Cr
Mn
Fe
Cu
Zn
Sb
Ba
La
Pb
SO2a
Auto
3. 9 ±2.9
2.3 ±1.3
1.7±1.0
30.1 ±12.3
11.3±3.5
9.2 ±6.8
4.6 ±2.2
3. 5 ±1.5
13.5 ±8.0
11.7 ±7.2
3.1 ±1.6
0.15 ±0.30
0.41 ±0.20
1.64 ±0.88
0.11 ±0.07
1.01 ±0.48
0.34 ±0.32
0.25 ±0.14
0.71 ±0.41
0.07 ±0.13
0.02 ±0.01
0.10 ±0.04
0.68 ±0.42
0.07 ±0.06
0.27 ±0.22
0.02 ±0.13
0.06 ±0.40
0.15±0.51
0.16 ±0.07
32.8 ±13. 9
Diesel
0.31 ±0.40
2.4 ±1.0
0.87 ±0.13
40.1 ±6.6
21.0 ±6.3
9.1 ±1.9
5.9 ±1.3
4.0 ±1.5
32.9 ±8.0
4.4 ±1.3
27.9 ±5.6
0.69 ±0.82
0.17±0.12
0.46 ±0.18
0.06 ±0.06
1.24 ±0.28
0.03 ±0.06
0.04 ±0.03
0.16 ±0.06
0.00 ±0.15
0.00 ±0.01
0.01 ±0.01
0.16 ±0.07
0.01 ±0.01
0.07 ± 0.02
0.01 ±0.14
0.14 ±0.47
0.18 ±0.59
0.01 ±0.03
66.9 ±24.0
Source: Watson et al. (1994b).

Note: Elemental abundances <0.01% (V, Co, Ni, Ga, As, Se, Br, Rb, Sr. Y, Zr, Mo, Pd, Ag, Cd, In, Sn, Au, Hg, Tl,
     U) in XRF analyses excluded; OC = organic carbon; EC = elemental carbon.
aRelative to total PM25.
October 1999                             4A-12       DRAFT-DO NOT QUOTE OR CITE

-------
      TABLE 4A-5. EMISSION RATES (MG/MILE) FOR CONSTITUENTS OF
      PARTICIPATE MATTER FROM GASOLINE AND DIESEL VEHICLES
TC
OC
EC
NO3-
S04=
Na
Mg
Al
Si
P
S
Cl
K
Ca
Fe
Ni
Cu
Zn
Br
Ba
Pb

Low
9.07 ±0.75
6.35 ±0.54
2.72 ±0.52
0.039 ±0.027
0.158 ±0.036
0.060 ±0.063
0.036 ±0.022
0.083 ±0.016
0.066 ±0.008
0.035 ±0.004
0.085 ±0.006
0.024 ±0.012
0.010 ±0.009
0.060 ±0.010
0.143 ±0.004
0.001 ±0.004
0.002 ± 0.004
0.048 ±0.003
0.001 ±0.002
0.013 ±0.136
0.007 ± 0.006
Gasoline
Medium
41.30±1.68
26.02 ±1.31
15.28 ±0.99
0.057 ±0.028
0.518 ±0.043
0.023 ±0.111
0.068 ± 0.027
0.078 ±0.016
0.279 ±0.011
0.152 ±0.007
0.442 ± 0.009
0.038 ±0.012
0.019 ±0.009
0.212 ±0.011
0.756 ±0.005
0.005 ± 0.004
0.016 ±0.003
0.251 ±0.004
0.016 ±0.002
0.009 ±0.138
0.085 ±0.005
Vehicles
High
207.44 ± 7.29
95.25 ±4.28
112.19±5.82
0.141 ±0.031
0.651 ±0.052
0.052 ± 0.092
0.041 ±0.033
0.057 ±0.014
0.714 ±0.012
0.1 13 ±0.007
0.822 ±0.022
0.081 ±0.020
0.031 ±0.035
0.210 ±0.030
1.047 ±0.010
0.011 ±0.005
0.021 ±0.005
0.265 ± 0.023
0.079 ± 0.003
0.011 ±0.299
0.255 ±0.008

Smoker
456.38 ±16.80
350.24 ± 15.27
106.14 ±5.42
0.964 ±0.051
2.160 ±0.137
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
2.515±0.116
0.140±0.117
0.033 ±0.386
0.362 ±0.250
2.438 ±0.054
0.008 ±0.017
0.071 ±0.018
0.188 ±0.272
0.047 ±0.012
0.380 ±2.175
0.345 ±0.032
Diesel
LDDa
373.43 ±13.75
132.01 ±5. 82
241.42 ±12.11
1.474 ±0.071
2.902 ±0.165
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
0.000 ± 0.000
2.458 ±0.124
0.228 ±0.1 14
0.000 ± 0.426
0.150 ±0.304
0.515 ±0.057
0.014 ±0.018
0.024 ±0.021
0.000 ± 0.299
0.003 ±0.014
0.428 ±2.390
0.153 ±0.033
Vehicles
HDD"
1570.69 ±58.24
253.94 ±16.12
1316.75 ±55.33
1.833 ±1.285
3. 830 ±1.286
1.288 ±2.160
1.061 ±0.729
0.321 ±0.543
8.018 ±0.221
0.407 ±0.136
3.717±0.111
0.881 ±0.221
0.064 ±0.248
0.716 ±0.107
0.376 ±0.055
0.002 ±0.057
0.001 ±0.062
0.707 ±0.032
0.012 ±0.050
0.493 ±3. 108
0.008 ±0.154
"Light duty.
bHeavy duty.
 Source: Lawson and Smith (1998).
October 1999
4 A-13
DRAFT-DO NOT QUOTE OR CITE

-------
 1           In addition to fossil fuels, biomass in the form of wood may be burned in forest fires or as
 2      fuel for heating or cooking. At first glance these two broad categories might seem to serve to
 3      distinguish between natural and anthropogenic sources.  However, many forest fires result from
 4      human intervention, either deliberately through prescribed burning in forest management or
 5      accidentally through the improper disposal of flammable material or fugitive sparks
 6      (e.g., Andreae, 1991). On the other hand, human intervention also suppresses lightning triggered
 7      fires and can also lead to the buildup of combustible fuel on the forest floor. Not enough data are
 8      available to assess the effects of humans on forest fires, except for land clearing for agriculture.
 9      In contrast to the mobile and stationary sources discussed earlier, emissions from biomass
10      burning in woodstoves and forest fires are strongly seasonal and can be highly episodic within
11      their peak emissions seasons.  The burning of fuelwood is confined mainly to the winter months
12      and is acknowledged to be a major source of ambient air particulate matter in the northwestern
13      United States during the heating season.  Forest fires occur primarily during the driest seasons of
14      the year in different areas of the country and are especially prevalent during prolonged droughts.
15      PM produced by biomass burning outside the United States, e.g., in  central America during the
16      spring of 1988 can also strongly affect ambient air quality in the United States.
17           An example of the composition of fine particles (PM25) produced by woodstoves is shown
18      in Figure 4A-4. These data were obtained in Denver during the winter of 1987-1988 (Watson
19      and Chow, 1994). As was the case for motor vehicle emissions, organic and elemental carbon
20      are the major components of particulate emissions from wood burning. It should be remembered
21      that the relative amounts shown for organic carbon and elemental carbon vary with the type of
22      stove, the stage of combustion and the type and condition of the fuelwood. Fine particles are
23      dominant in studies of wood burning emissions. For instance, the mass median diameter of
24      wood-smoke particles was found to be about 0.17 //m in a study of the emissions from burning
25      hardwood, softwood and synthetic logs (Dasch, 1982).
26           Measurements of aerosol composition, size distributions, and aerosol emissions factors
27      have been made in biomass burning plumes either on towers (Susott et al., 1991) or aloft on fixed
28      wing aircraft (e.g., Radke et al., 1991) or on helicopters (e.g., Cofer  et al., 1988). As was found
29      for woodstove emissions, the composition of biomass burning emissions is strongly dependent
30      on the stage of combustion (i.e., flaming, smoldering, or mixed), and the type of vegetation (e.g.,
31      forest, grassland,  scrub).  Over 90% of the dry mass in particulate biomass burning emissions is

        October 1999                            4A-14      DRAFT-DO NOT QUOTE OR CITE

-------
                        10
                        10
                        10
                         ,-4
                                              Chemical Species

       Figure 4A-4.  Chemical abundances for PM2 5 emissions from wood burning in Denver,
                     CO.  Solid bars represent fractional abundances, and the error bars
                     represent variability in species abundances. Error bars represent detection
                     limits when there are no solid bars.
       Source:  Watson and Chow (1994).
 1     composed of organic carbon (Mazurek et al., 1991). Ratios of organic carbon to elemental
 2     carbon are highly variable ranging from 10:1 to 95:1, with the highest ratio found for smoldering
 3     conditions and the lowest for flaming conditions. Emissions factors for total particulate
 4     emissions increase by factors of two to four in going from flaming to smoldering stages in the
 5     individual fires studied by Susott et al. (1991).
 6          Particles in biomass burning plumes from a number of different fires were found to have
 7     three distinguishable size modes, namely a nucleation mode, an accumulation mode, and a coarse
 8     mode (Radke et al., 1991).  Based on an average of 81 samples, approximately 70% of the mass
 9     was found in particles < 3.5 //m in aerodynamic diameter. The fine particle composition was
10     found to be dominated by tarlike, condensed hydrocarbons and the particles were usually
       October 1999
4 A-15
DRAFT-DO NOT QUOTE OR CITE

-------
 1      spherical in shape. Additional information for the size distribution of particles produced by
 2      vegetation burning was shown in Figure 4A-2.
 3           An example of ambient data for the composition of PM25 collected at a tropical site that
 4      was heavily affected by biomass burning is shown in Table 4A-6.  The samples were collected
 5      during November of 1997 on the campus of Sriwijaya University, which is located in a rural
 6      setting on the island of Sumatra in Indonesia (Pinto et al., 1998). The site was subjected
 7      routinely to levels of PM25 well in excess of the U.S. NAAQS as a result of the Indonesian
 8      biomass fires from the summer of 1997 through the spring of 1988. As can be seen from a
 9      comparison of the data shown in Table 4A-6 with those shown in Figure 4A-4, there are a
10      number of similarities and differences (especially with regard to the heavy metal content) in the
11      abundances of many species. The abundances of some crustal elements (e.g., Si, Fe) are higher
12      in Table 4A-6 than in Figure 4A-4 perhaps reflecting additional contributions of entrained soil
13      dust.
14           Although sea-salt aerosol production is confined to salt water bodies, it is included here
15      because many marine aerosols can exert a strong influence on the composition of the ambient
16      aerosol in coastal areas.  In some respects, the production of sea-salt aerosols is like that of
17      windblown dust in that both are produced by wind agitation of the surface.  The difference
18      between the two categories arises because sea-salt particles are produced from the bursting of air
19      bubbles rising to the sea surface. Air bubbles are formed by the entrainment of air into the water
20      by breaking waves.  The surface energy of a collapsing bubble is converted to kinetic energy in
21      the form of a jet of water which can eject drops above the sea surface.  The mean diameter of the
22      jet drops is about 15% of the bubble diameter (Wu,  1979). Bubbles in breaking waves range in
23      size from a few //m to several mm in diameter. Field measurements by Johnson and Cooke
24      (1979) of bubble size spectra show maxima in diameters at around 100 //m, with the bubble size
25      distribution varying as (d/d0)"5 with d0 =  100 //m.
26           Since sea-salt particles receive water from the surface layer, which is enriched in organic
27      compounds, the aerosol drops are composed of this  organic material in addition to sea salt (about
28      3.5% by weight in sea water). Na+ (30.7%),Cr (55.0%), SO4= (7.7%), Mg2+ (3.6%), Ca2+ (1.2%),
29      K+ (1.1%), HCO3" (0.4%), and Br" (0.2%) are the major ionic species by mass in sea water
30      (Wilson, 1975). The composition of the marine aerosol also reflects the occurrence of


        October 1999                             4A-16      DRAFT-DO NOT QUOTE OR CITE

-------
               TABLE 4A-6.  MEAN AEROSOL COMPOSITION AT TROPICAL SITE
                 (SRIWIJAYA UNIVERSITY, SUMATRA, INDONESIA) AFFECTED
                          HEAVILY BY BIOMASS BURNING EMISSIONS1
Component
OC
EC
scv
Al
Si
Cl
K
Ca
Ti
V
Abundance (%)
76
1.2
11
bd2
9.3 x 1Q-2
4.4
0.7
4.5 x 1Q-2
4.2 x 1Q-3
bd2
Component
Cr
Mn
Fe
Ni
Cu
Zn
As
Se
Br
Pb
Abundance (%)
bd2
bd2
3.9 x 1Q-2
<3.8 x 1Q-5
4.8 x 1Q-4
3.1 x 1Q-3
6.4 x 1Q-4
2.8 x 1Q-4
3.6 x 1Q-2
3.1 x 1Q-3
         'The mean PM2 5 concentration during the sampling period (11/5-11/11/97) was 264 //g/m3.
         2beneath detection limit.
         Source: Pinto etal. (1998).
 1     displacement reactions which enrich sea-salt particles in SO4" and NO3" while depleting them of
 2     Cr and Br.
 3          Seasalt is concentrated in the coarse size mode with a mass median diameter of about 7 //m
 4     for samples collected in Florida, the Canary Islands and Barbados (Savoie and Prospero, 1982).
 5     The size distribution of sulfate is distinctly bimodal. Sulfate in the coarse mode is derived from
 6     sea water but sulfate in the submicron aerosol arises from the oxidation of dimethyl sulfide
 7     (CH3SCH3) or DMS. DMS is produced during the decomposition of marine micro-organisms.
 8     DMS is oxidized to MSA (methane sulfonic acid) a large fraction of which is oxidized to sulfate
 9     (e.g., Hertel etal., 1994).
10          Apart from sea spray, other natural sources of particles include the suspension of organic
11     debris  and volcanism.  Particles are released from plants in the form of seeds, pollen, spores, leaf
       October 1999
4 A-17
DRAFT-DO NOT QUOTE OR CITE

-------
 1      waxes and resins, ranging in size from 1 to 250 //m (Warneck, 1988).  Fungal spores and animal
 2      debris such as insect fragments are also to be found in ambient aerosol samples in this size range.
 3      While material from all the foregoing categories may exist as individual particles, bacteria are
 4      usually found attached to other dust particles (Warneck, 1988).  Smaller bioaerosol particles
 5      include viruses, individual bacteria, protozoa, and algae (Matthias-Maser and Jaenicke, 1994).
 6      In addition to natural sources, other sources of bioaerosol include industry (e.g., textile mills),
 7      agriculture, and municipal waste disposal (Spendlove, 1974). The size distribution of
 8      bioaerosols has not been as well characterized as it has for other categories.
 9           Trace metals are emitted to the atmosphere from a variety of sources such as  sea spray,
10      wind blown dust, volcanoes, wild fires and biotic sources (Nriagu, 1989). Biologically mediated
11      volatilization processes (e.g., biomethylation) are estimated to account for 30-50% of the
12      worldwide total Hg, As, and Se emitted annually, whereas other metals are derived principally
13      from pollens, spores, waxes, plant fragments, fungi, and algae.  It is not clear, however, how
14      much of the biomethylated species are remobilized from anthropogenic inputs.  Median ratios of
15      the natural contribution to globally averaged total sources for trace metals are estimated to be
16      0.39 (As), 0.15 (Cd), 0.59 (Cr), 0.44 (Cu), 0.41 (Hg), 0.35 (Ni), 0.04 (Pb), 0.41 (Sb), 0.58 (Se),
17      0.25 (V),  and 0.34 (Zn), suggesting a not insignificant natural source for many trace elements.
18      It should be noted though  that these estimates are based on emissions estimates which have
19      uncertainty ranges of an order of magnitude.
20           The discussion above focused mainly on the elemental composition of emissions. Carbon
21      was treated mainly as organic carbon and elemental carbon.  However, there are literally
22      hundreds  of organic compounds which have been quantified in ambient and source samples
23      which are lumped routinely into the category of organic carbon. These compounds, many of
24      which have been used in source apportionment studies as signature (e.g., Schauer et al., 1996),
25      consist of high molecular weight alkanes, hopanes and steranes, organic acids, aldehydes,
26      polycyclic aromatic hydrocarbons and steroids. Profiles of organic compounds from meat
27      cooking (Rogge et al., 1991); automobiles and heavy-duty diesel trucks (Rogge et al., 1993a);
28      road dust, tire debris and brake linings (Rogge et al., 1993b); vegetative detritus (Rogge et al.,
29      1993c); natural gas home  appliances (Rogge et al., 1993d); cigarette smoke  (Rogge et al., 1994);
30      hot asphalt roofing tar pots (Rogge et al., 1997a); distillate fuel oil burning (Rogge et al.,  1997b);
31      and pine,  oak, and synthetic log burning in residential fire places (Rogge et al.,  1998) have been

        October 1999                             4A-18       DRAFT-DO NOT QUOTE OR CITE

-------
1     obtained. Many individual compounds are present in concentrations much less than 1 ng/m3 in
2     ambient PM samples. They have been used in only a limited number of studies (mainly in
3     Los Angeles) by only a small number of groups.  Measurement methods need to be standardized
4     and made more cost-effective to take advantage of the opportunities they offer in studies
5     throughout the United States.
      October 1999                           4A-19      DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES
  2
  3       Andreae, M. O. (1991) Biomass burning: its history, use, and distribution and its impact on environmental quality
  4             and global climate. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric
  5             implications. Cambridge, MA: MIT Press; pp. 1-21.
  6       Cadle, S. H.; Mulawa, P. H.; Ball, J.; Donase, C.; Weibel, A.; Sagebiel, J. C.; Knapp, K. T.; Snow, R. (1997)
  7             Particulate emission rates from in-use high-emitting vehicles recruited in Orange county, California. Environ.
  8             Sci. Technol. 31:3405-3412.
  9       Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. (1993) Achieving acceptable air quality: some
10             reflections on controlling vehicle emissions. Science  (Washington, DC) 261: 37-45.
11       Chow, J. C.; Watson, J. G.; Houck, J. E.; Pritchett, L. C.; Rogers, C. F.; Frazier, C. A.; Egami, R. T.; Ball, B. M.
12             (1994) A laboratory resuspension chamber to measure fugitive dust size distributions and chemical
13             compositions. Atmos. Environ. 28: 3463-3481.
14       Cofer, W. R., Ill; Levine, J. S.; Sebacher, D. L; Winstead, E. L.; Riggin, P. J.; Brass, J. A.; Ambrosia, V. G.  (1988)
15             Particulate emissions from a mid-latitude prescribed chaparral fire. J. Geophys. Res. [Atmos.] 93: 5207-5212.
16       Cooper, J. A.; Redline, D. C.; Sherman, J. R.; Valdovinos, L. M.; Pollard, W. L.; Scavone, L. C.; Badgett-West, C.
17             (1987) PM10 source composition library for the South Coast Air Basin: volume I, source profile development
18             documentation final report. El Monte, CA: South Coast Air Quality Management District; July 15.
19       Dasch, J. M. (1982) Particulate and gaseous emissions from wood-burning fireplaces.  Environ. Sci. Technol.
20             16: 639-645.
21       Durbin, T. D.; Smith, M. R.; Norbeck, J. M.; Truex, T. J. (1999) Population density, particulate emission
22             characterization, and impact on the particulate inventory of smoking vehicles in the South Coast Air Quality
23             Management District. J. Air Waste Manage. Asssoc. 49: 28-38.
24       Hertel, O.; Christensen, J.; Hov, 0. (1994) Modelling of the end products of the chemical decomposition of DMS in
25             the marine boundary layer. Atmos. Environ. 28: 2431-2449.
26       Hopke, P. K.; Lamb, R. E.; Natusch, D. F. S.  (1980) Multielemental characterization of urban roadway dust.
27             Environ. Sci. Technol. 14: 164-172.
28       Houck, J. E.; Chow, J. C.; Watson, J. G.; Simons, C. A.; Pritchett, L. C.; Goulet, J. M.; Frazier, C. A. (1989)
29             Determination of particle size distribution and chemical composition of particulate matter from selected
30             sources in California: volume I and executive summary (final report). Sacramento, CA: California Air
31             Resources Board.
32       Houck, J. E.; Goulet, J. M.; Chow, J. C.; Watson, J. G.; Pritchett, L. C. (1990) Chemical characterization of
33             emission sources contributing to light extinction. In: Mathai, C. V., ed. Visibility and fine particles:
34             an A&WMA/EPA international specialty conference; October 1989; Estes Park, CO. Pittsburgh, PA:  Air &
35             Waste Management Association; pp. 437-446. (A&WMA transactions series no. TR-17).
36       Johnson, B. D.; Cooke, R. C. (1979) Bubble populations and spectra in coastal waters: a photographic approach.
37             J. Geophys. Res. C: Oceans Atmos. 84: 3761-3766.
38       Lawson, D. R.; Smith, R. E. (1998) The northern front range air quality study: a report to  the Governor and General
39             Assembly. Fort Collins, CO: Colorado State University; December.
40       Maricq, M. M.; Podsiadlik, D. H.; Chase, R. E. (1999) Gasoline vehicle particle size distributions: comparison of
41             steady state, FTP, and USOG measurements. Environ. Sci. Technol. 33: 2007-2015.
42       Mason, B. (1966) Principles of geochemistry. 3rd ed. New York, NY: John Wiley & Sons, Inc.
43       Matthias-Maser, S.; Jaenicke, R.  (1994) Examination of atmospheric bioaerosol particles  with radii > 0.2  pan.
44             J. Aerosol Sci. 25: 1605-1613.
45       Mazurek, M. A.; Cofer, W. R., Ill; Levine, J. S. (1991) Carbonaceous aerosols from prescribed burning of a  boreal
46             forest ecosystem. In: Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric
47             implications. Cambridge, MA: MIT Press; pp. 258-263.
48       NEA, Inc. (1990) Enhancement of the south coast air basin source profile library for chemical mass balance
49             receptor model applications: volume I, final report. Beaverton, OR: NEA.
50       Nriagu, J. O. (1989) A global assessment of natural sources of atmospheric trace metals. Nature (London)
51             338:47-49.
52       Nriagu, J. O.; Pacyna, J. M. (1988) Quantitative assessment of worldwide contamination of air, water and soils by
53             trace metals. Nature (London) 333: 134-139.
         October 1999                                  4A-20         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Olmez, I.; Sheffield, A. E.; Gordon, G. E.; Houck, J. E.; Pritchett, L. C.; Cooper, J. A.; Dzubay, T. G.; Bennett,
  2             R. L. (1988) Compositions of particles from selected sources in Philadelphia for receptor modeling
  3             applications. JAPCA 38: 1392-1402.
  4       Peltier, R. G.; Redline, D. C.; Cooper, J. A.; Aquilizan, A. V.; Miller, E. A.; White, J. L. (1990a) Enhancement of
  5             the south coast air basin source profile library for chemical mass balance receptor model applications.
  6             Volume II: source profile library supplement. Final report. Beaverton, OR: NEA, Inc.
  7       Peltier, R. G.; Redline, D. C.; Cooper, J. A.; Aquilizan, A. V.; Miller, E. A.; White, J. L. (1990b) Enhancement of
  8             the south coast air basin source profile library for chemical mass balance receptor model applications.
  9             Volume III: appendices. Final report. Beaverton, OR: NEA, Inc.
10       Pinto, J. P.; Grant, L. D.; Hartlage, T. A. (1998) Report on U.S. EPA air monitoring of haze from S.E. Asia biomass
11             fires. Research Triangle Park, NC: U.S. Environmental Protection Agency, National Center for
12             Environmental Assessment; report no. EPA/600/R-98/071.
13       Radke, L. F.; Hegg, D. A.; Hobbs, P. V.; Nance, J. D.; Lyons, J. H.; Laursen, K. K.; Weiss, R. E.; Riggan, P. J.;
14             Ward, D. E. (1991) Particulate and trace gas emissions from large biomass fires in North America. In:
15             Levine, J. S., ed. Global biomass burning: atmospheric, climatic, and biospheric implications. Cambridge,
16             MA: MIT Press; pp. 209-224.
17       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1991) Sources of fine organic
18             aerosol. 1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 25: 1112-1125.
19       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993a) Sources of fine organic
20             aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci.
21             Technol. 27:636-651.
22       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993b) Sources of fine organic
23             aerosol. 3. Road dust, tire debris, and organometallic brake lining dust: roads as sources and sinks. Environ.
24             Sci. Technol. 27: 1892-1904.
25       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993c) Sources of fine organic
26             aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants. Environ. Sci. Technol.
27             27:2700-2711.
28       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1993d) Sources of fine organic
29             aerosol. 5. Natural gas home appliances. Environ. Sci. Technol. 27: 2736-2744.
30       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1994) Sources of fine organic
31             aerosol. 6. Cigarette smoke in the urban atmosphere. Environ. Sci. Technol. 28: 1375-1388.
32       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997a) Sources of fine organic
33             aerosol. 7. Hot asphalt roofing tar pot fumes.  Environ. Sci. Technol. 31: 2726-2730.
34       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1997b) Sources of fine organic
35             aerosol. 8. Boilers burning no. 2 distillate fuel oil. Environ. Sci. Technol. 31: 2731-2737.
36       Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. (1998) Sources of fine organic
37             aerosol. 9. Pine, oak, and synthetic  log combustion in residential fireplaces. Environ. Sci. Technol. 32: 13-22.
38       Sagebiel, J. C.; Zielinska, B.; Walsh, P. A.; Chow, J. C.; Cadle, S. H.; Mulawa, P. A.; Knapp, K. T.; Zweidinger,
39             R. B.; Snow, R. (1997) PM-10 exhaust  samples collected during IM-240 dynamometer tests of in-service
40             vehicles in Nevada. Environ. Sci. Technol. 31: 75-83.
41       Savoie, D. L.; Prospero, J. M. (1982) Particle size distribution of nitrate and sulfate in the marine atmosphere.
42             Geophys. Res. Lett. 9:  1207-1210.
43       Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurik, M. A.; Cass,  G. R. (1996) Source apportionment of
44             airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30: 3837-3855.
45       Spendlove, J.  C. (1974) Industrial, agricultural, and municipal microbial aerosol problems. In: Current problems in
46             aerobiology: proceedings of the thirtieth general meeting of the Society for Industrial Microbiology; August
47             1973; Evanston, IL. Dev. Ind. Microbiol. 15:  20-27.
48       Susott, R. A.;  Ward, D. E.; Babbitt, R. E.; Latham, D. J. (1991) The measurement of trace emissions and
49             combustion characteristics for a mass fire. In: Levine,  J. S., ed. Global biomass burning: atmospheric,
50             climatic, and biospheric implications. Cambridge, MA: MIT Press; pp, 245-257.
51       Turekian, K. K. (1971) Elements, geochemical distribution of. In: Encyclopedia of science and technology, v. 4.
52             New York, NY: McGraw-Hill Book Company; pp. 627-630.
53       Vinogradov, A. P. (1959) The geochemistry of rare and dispersed chemical elements in soils. 2nd ed. New York,
54             NY: Consultants Bureau, Inc.
55       Warneck, P. (1988) Chemistry of the natural atmosphere. New York, NY: Academic Press, Inc.


         October 1999                                   4A-21         DRAFT-DO NOT QUOTE OR  CITE

-------
 1       Watson, J. G.; Chow, J. C. (1994) Clear sky visibility as a challenge for society. Annu. Rev. Energy Environ.
 2             19:241-266.
 3       Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Burns, S.; Ragazzi, R. A. (1990a) Composite source
 4             profiles for particulate motor vehicle exhaust source apportionment in Denver, CO. In: Mathai, C. V., ed.
 5             Visibility and fine particles: an A&WMA/EPA international specialty conference; October 1989; Estes Park,
 6             CO. Pittsburgh, PA: Air & Waste Management Association; pp. 422-436. (A&WMA transactions series
 7             no. TR-17).
 8       Watson, J. G.; Chow, J. C.; Pritchett, L. C.; Houck, J. A.; Ragazzi, R. A.; Burns, S. (1990b) Chemical source
 9             profiles for particulate motor vehicle exhaust under cold and high altitude operating conditions.  Sci. Total
10             Environ. 93: 183-190.
11       Watson, J. G.; Chow, J. C.; Blumenthal, D. L.; Lurmann, F. W.; Hackney, R. J.; Magliano, K. L.; Pederson, J. R.;
12             Neff, W. D.; Roth, P. M.; Solomon, P. A.; Thuillier, R. H.; Ziman, S. D. (1994b) Planning for data analysis.
13             In: Solomon, P. A.; Silver, T. A., eds. Planning and managing regional air quality modeling and measurement
14             studies: a perspective through the San Joaquin Valley air quality study and AUSPEX. Boca Raton, FL:
15             Lewis Publishers; pp.  335-349.
16       Wilson, T. R. S. (1975) Salinity and the major elements of sea water. In: Riley, J. P.; Skirrow, G., eds. Chemical
17             oceanography: v. 1. 2nd ed. London, United Kingdom: Academic Press Inc.; pp. 365-413.
18       Wu, J. (1979) Spray in the atmospheric surface layer: review and analysis of laboratory and oceanic results.
19             J. Geophys. Res. C: Oceans Atmos. 84:  1693-1704.
         October 1999                                  4A-22        DRAFT-DO NOT QUOTE OR CITE

-------
 i       5.  HUMAN EXPOSURE TO AMBIENT PARTICULATE
 2           MATTER:  RELATIONS TO CONCENTRATIONS
 3              OF AMBIENT AND NON-AMBIENT PM AND
 4                          OTHER AIR POLLUTANTS
 5
 6
 7     5.1   INTRODUCTION AND BASIC CONCEPTS
 8          This chapter examines ambient particulate matter (Plvf) air quality and that portion of
 9     ambient PM which penetrates into indoor microenvironments.  It also examines, to a lesser
10     extent, the contributions of sources of non-ambient PM to total PM exposure. This is to aid in
11     the interpretation of the acute and chronic epidemiology studies discussed in Chapter 6, in which
12     ambient PM concentrations are assumed to be an indicator, or a surrogate, for the mean
13     community exposure to PM of ambient origin, or an individual's exposure to ambient PM.
14     Thus, this chapter has three objectives:
15     (a) To provide a review of pertinent studies of personal exposures to total PM (ambient PM plus
16         non-ambient PM).
17     (b) To evaluate linkages of human exposure to PM of ambient origin estimated from
18         concentrations of PM measured at a fixed-site monitor located at some central location in a
19         community under study.
20     (c) To quantify the contribution of PM of ambient origin to total personal PM exposure.
21          The 1996 Particulate Matter Air Quality Criteria Document (PM AQCD) (U.S.
22     Environmental Protection Agency, 1996) thoroughly reviewed the PM exposure literature
23     through 1995 and early 1996. This chapter reviews the history of PM AQCD developments from
24     1969 to the present. It then thoroughly reviews key pre-1996 studies, the new PM exposure
25     literature  from 1995/6 through 1998 to date, and some literature accepted or submitted for
26     publication later in 1999.
27          The U.S. Environmental Protection Agency regulatory authority for PM extends to the
28     ambient air, defined in 40 CFR 50.1(e) as that portion of the atmosphere, external to buildings, to
             'in this chapter PM without a subscript refers to PM in general. PMX refers to the mass of PM collected by
       a monitor with a penetration fraction of 0.5 for particles with an aerodynamic diameter of X microns (XponAD).
       October 1999                            5-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1      which the general public has access (Code of Federal Regulations, 1994). Therefore, polluted
 2      air inside a building, or anywhere on private property owned or controlled by the source of
 3      pollution, is not regulated by the National Ambient Air Quality Standards (NAAQS). However,
 4      it is necessary to consider total personal exposure to ambient PM, both in regulated ambient air
 5      and within non-regulated indoor air. This is because ambient (outdoor) particles penetrate into
 6      non-ambient (indoor residential and occupational) micro environments where, on average, people
 7      spend 87% of their time (Klepeis et al., 1999).  Therefore, when people are indoors, they are
 8      exposed to  a mixture of PM of ambient origin and particles generated indoors from sources not
 9      regulated by EPA (e.g., cigarette smoke or an occupational activity).
10           Particulate matter represents a generic class of pollutants which requires a different
11      interpretation of exposure in contrast to that for the gaseous criteria pollutants, such as CO
12      (Mage, 1985). Whereas a molecule of CO emitted from a motor vehicle is indistinguishable
13      from a molecule of CO emitted from a cigarette, a 1-^m aerodynamic diameter (AD) particle
14      emitted from a motor vehicle and a l-^m AD particle emitted from a cigarette may have a
15      different shape, mass, chemical composition, solubility and toxicity (Siegmann et al., 1999).
16      In the atmosphere, a particle may be a single entity, or an agglomeration of particles, such as a
17      particle from motor vehicle exhaust bound to a particle from cigarette ash. Furthermore, indoor
18      sources of particles also produce a wide variety of particles of varying AD and composition that
19      people are exposed to, as shown in Figure 5-1 (Owen et al., 1992). Most of these particles can be
20      fragmented by mechanical activity and their detritus may exist at smaller ADs than shown.
21           A subject's personal exposure to PM is theoretically measured by sampling the
22      concentration of PM in the inhaled air entering the nose or mouth. The inlet  to a personal
23      monitor is normally placed at the outer limit of the breathing zone to avoid a negative sampling
24      bias that could be caused by dilution of the  collected air by the exhaled breath which is depleted
25      of PM. However, such placement does not allow sampling of directly inhaled cigarette smoke or
26      the inhaled air that passes through a dust mask.  Thus, personal monitoring of PM exposure of
               2In this chapter, the term "ambient air" means that portion of the ambient atmosphere that may be
        considered representative of a community and not unduly influenced by any specific identifiable source (e.g., not at
        the side of a highway, or next to a coke oven). "Outdoor air" is taken to mean that portion of the atmosphere,
        external to buildings, where a PM concentration measurement may be unduly influenced by an immediate source of
        PM (undiluted cigarette smoke, backyard barbecue, idling motor vehicle, etc.). Thus, all ambient air is outdoor air,
        but not all outdoor air is ambient air.
        October 1999                               5-2         DRAFT-DO NOT QUOTE OR CITE

-------
o
o
r-K
o
cr
o
l-l
^o
^o
'S*O
^-/l
o
Tl
H
O
0
0
H
0
0
H
W
O
O
HH
H
W
0
Plant
Animal
Mineral
Combustion
Home/
Personal
Care
Radioactive
Figure 5-1. Sh
Source: Owen et £
Particle Aerodynamic Diameter (|jm)
.01 0.1 1 10 100 1,0

— Corn starch
-_ Carbon
"^
Bacteriophage ^ 	

-., 	 Viruses 	 ^.

•••| • 	 |^ • Vnllen 	 L " 	


f ^ Starches ^
K ^ Milled Flour, Ground Corn ^ "" Grain Duste

Black - ^ *•
^ Channel Black ^. Textile Fibers
PuddinofMix ^Com Cob Chaff ^
Snuff 4 Cayenne Pepper ^
•^ 	 anu" 	 ™ 	 »-^ Sawdust ^ p
Droplet Nuclei1 	 ^ Bacteria
^ Gelatin ^
Spider Web ., Hair ^


^ » p,- • . . i n 9 ogray "nea mill-. rjust M|tes
•< >• Disintegrated Feces •<" Faces ^ Bone Dust * *»,
Asbestos _
^" Talo. ^ •—
^ Carbon Dust from Graphite > 	
^ Clay
^ Lead. Bromine ^

Metalurgical Dusts and Fumes
^ NI-UCI Fume
4 Sea Salt * ^
^ Tobacco Smoke
	 	 	 l»-Carbon Dust
^ ^
^Calcium, Zinc • Lead Dust ^

Special Use Insulation ^ Fiberglass " Glass Wool ^
Toxtiloa + 4 	 !*• Fertilizer, Ground Limestone f"

^^ Camant Dust "^.

"" ^ Burning Wood _ *""
^ Kosin smoKe ^
Coal Flue Gas

^ "Anti-Stick Sorav
Paint Pigments
^Alkali Fume ^ —
Face Powder: > 	 Talc 	
. Radon Progeny
1 Liquid droplets containing bacteria etc., sneezed, etc
2 Man-made mineral fibers
ses of various types of indoor particles.
il (1992).
— ^ Smouldering or Flaming Cooking Oil
^ Oil Smoke Fly Ash
^ Auto Emissions ^ 1
. . . _ > 	 < Ant iperspi rant

=— ^ Spray Pamt Spray Paint Dust
Freshener 	 +.
-, Humidifier £ 	 < Fabric Protector
•^ — ^ Nebulizer Drops fc
™ ^ ~
	 insecticide Dusts 	 £ 	 __». Emollients
	 > ^Clumps "" ' MgCO3
Pigment fc _ Binder
f " 4 ^Artificial Textile Fibers

	 k>


-------
 1     non-smokers not using a dust mask can be adequately characterized by a personal monitor worn
 2     by the subject with the sampling inlet close to, but not in, the breathing zone.
 3           The total PM exposure of an individual during a period of time is composed of exposure to
 4     many different particles from various sources while in different microenvironments. Duan
 5     (1982) defined a microenvironment as "a [portion]  of air space with homogeneous pollutant
 6     concentration"; it has also been defined (Mage, 1985) as a volume in space, during a specific
 7     time interval, during which the variance of concentration within the volume is significantly less
 8     than the variance between that microenvironment and its surrounding microenvironments. For
 9     example,  a kitchen with a wood stove can constitute a single microenvironment for total PM
10     when the  stove is off, and all people in the kitchen would have similar PM exposures. When the
11     stove is on, the kitchen could have a significant vertical PM concentration gradient, and a child
12     on the floor in a far corner and an adult standing at the stove could be exposed to significantly
13     different PM concentrations. When a concentration gradient exists, such as from smoke from a
14     lit cigarette, where the gradient goes from mg/rri to //g/m3, the former definition of a
15     microenvironment breaks down because there is no homogeneous concentration.  Alternatively,
16     the latter definition requires a rather impractical specification of a large number of transient
17     microenvironments.
18           In a given microenvironment, such as one in the kitchen example above, the particles may
19     originate from a wide variety of sources.  PM may be generated from within (e.g., the stove, deep
20     frying, burning toast), from without (ambient PM entering through an open window), from
21     another indoor microenvironment (cigarette smoke from the living room), or from a personal
22     activity that generates a heterogeneous mix of PM (e.g., sweeping the kitchen floor and
23     resuspending a mixture of PM from both indoor and outdoor sources that had settled out).
24           In general, as a function of space and time, people pass through a series of
25     microenvironments.  Thus their average total daily  exposure $ //g/m3) to PM can be expressed
26     as the sum of their exposures within the microenvironments they occupy. With appropriate
27     averaging over sets of four classes of microenvironments (e.g.,indoors, ambient-outdoors.
28     occupational, and in-traffic), the average total exposure of a non-smoker can be expressed as
29     follows (Mage, 1985):
30
31                       ir = fFt+Et+Et+EtVT                  f5n
                         ^  \^mlm + ^out1 out ^  r'occLocc ^ ^tra Ltra>'' L                  P1'

       October 1999                             5-4         DRAFT-DO NOT QUOTE OR CITE

-------
 1     where each value of E is the mean value of total PM concentration in the microenvironment class
 2     while the subject is in it, time (t) is the total time the subject is in that microenvironment during
 3     the study period, and T is equal to the sum of all times (usually 24-h). Note that ^ is not
 4     necessarily a 24-h time-weighted-average (TWA) of PM concentration in an indoor
 5     microenvironment. Thatcher and Layton (1995) report that "merely walking into a room
 6     increased the particle concentration by 100%". Consequently, an integrated measurement of air
 7     quality in an enclosed  space that includes time when it is unoccupied, or not occupied by a
 8     specific subject, may not be a valid measure that can be used to estimate the  exposure of that
 9     specific subject while in that microenvironment (Larssen et al., 1993).  A measured
10     microenvironmental concentration when the space is unoccupied will tend to be biased low as a
11     measure of the exposure within it during periods of occupancy.  For example it is incorrect, as it
12     is for NO2, to associate a PM exposure to a person, while cooking at a gas stove in a kitchen,
13     with a 24-h kitchen PM concentration measurement that is influenced by periods when the stove
14     was off (Smith et al., 1994).
15
16     5.1.1 The History of Understanding Human Exposure to Particulate Matter
17           The 1969 PM AQCD (National Air Pollution Control Administration, 1969) only discussed
18     ambient PM concentrations as indices of a general exposure to PM, to the extent that the
19     document index mentions neither exposure nor indoor.  The introduction stated "Air quality
20     criteria are descriptive; that is, they describe the effects that have been observed to occur when
21     the ambient level of a pollutant has been reached or exceeded specific figures for a specific time
22     period. . . .Epidemiologic studies  [are presented that] analyze the effects of pollution from
23     ambient exposure of groups of people living in the community." The indoor microenvironment
24     was considered protective and no  mention was made of sources of pollution indoors.  For
25     example, it was suggested that the decrease in mortality associated with reduction of PM
26     pollution in  London, after the 1952 Fog, may possibly have been related to the report that "a
27     great deal of publicity has been given to the harmful effects of smog, and more susceptible
28     individuals have been  encouraged to use masks and filters, and to stay indoors."
29           In the context of protection from ambient PM afforded by staying indoors, the document
30     caveats  as follows: "There has, however, been a minimum of attention paid to indoorand
31     domestic environments and their potential contribution. Measurement of such indoor exposures
       October 1999                              5-5         DRAFT-DO NOT QUOTE OR CITE

-------
 1     might be difficult, but omission of the information could well modify the appraisal of the
 2     importance of particulate pollution" [e.g., ambient PM as measured by a high volume sampler
 3     would now appear less potent because susceptible people would stay indoors on high pollution
 4     days].
 5           The 1982 PM AQCD (U.S. Environmental Protection Agency, 1982) reported on indoor
 6     PM concentrations as follows: "Yocom et al. (1971) studied TSP in public buildings, offices and
 7     homes using a scaled down version of the hi-vol sampler. . .Indoor levels were about half outdoor
 8     levels on the average. . .  Alzona et al. (1979) reported elemental analyses for calcium and iron,
 9     normally coarse-particle components, and for zinc, lead and bromine, components of fine
10     particles. . .  it appears that tracer components of coarse particles do not penetrate any of these
11     structures as readily as the fine components."
12           However, early personal PM exposure monitoring studies indicated that some people's
13     personal activities, along with PM generated by personal and indoor sources (e.g., cigarette
14     smoking), could lead to PM indoors and personal exposures to total PM that exceeded the
15     concentration of the PM found in the immediate outdoor air or in the local ambient air (Binder
16     et al., 1976;  Repace and  Lowrey,  1980; Spengler et al., 1980).  This was reported as follows:
17     "It is apparent that, in the absence of smoking, indoor and outdoor levels of fine particulate mass
18     are almost the same. However, smoking contributes very significantly to indoor level."
19           The section concluded:  "Therefore, fine particles readily penetrate buildings and occur
20     inside to about the same extent as outdoors.  Indoor activity  adds incrementally to outdoor  levels
21     and, frequently, somewhat higher levels of fine particles are observed indoors.  Smoking adds
22     very materially to indoor levels."
23           The 1982 document also summarized the situation as follows: "Because stationary
24     ambient-air  pollution monitors provide general statistics on composite population exposures, it
25     would be extremely difficult (if not impossible) to predict an individual's actual exposure to PM
26     on the basis of community air-monitoring data alone."
27     "Although outdoor concentrations of pollutants can be measured at particular sites, our highly
28     mobile population can be exposed to either higher or lower values than community monitors
29     show. Indoor particle levels can be high because of smoking, cleaning operations, or normal
30     activities. Exposures of individuals to PM can vary more than community monitors show."
31     (Volume H,  page 5-138)

       October 1999                              5-6        DRAFT-DO NOT QUOTE OR CITE

-------
 1           In the period between 1982 and 1996, many more studies on personal PM exposure and
 2      indoor PM were reported that documented the fact that, in most inhabited domestic
 3      environments, indoor PM concentrations, and personal PM exposures of the residents, were
 4      greater than the simultaneous ambient PM concentrations (e.g., Sexton et al., 1984; Spengler
 5      et al., 1985; Clayton et al., 1993). Clayton et al. (1993) referred to their finding of a 60% excess
 6      of daytime personal PM10 exposure in Riverside, CA, relative to a time-weighted-average of
 7      indoor and  outdoor PM10 concentrations, as a "personal cloud" which is discussed later in this
 8      chapter.
 9           In 1996 it was known, from personal monitoring and indoor monitoring, that some PM
10      constituents, such as sulfates, are almost always lower indoors than outdoors because of the
11      virtual absence of indoor sources and the presence of sinks for sulfates in indoor settings
12      (exceptions can occur if high sulfur coal or kerosene are used as fuel in a poorly vented stove or
13      space heater). However, this relationship does not hold for most other PM constituents, as the
14      indoor and personal monitoring data show both higher- and lower-than ambient PM
15      concentrations in indoor settings as a function of particle size and human activity patterns.
16           The largest coarse mode particles (>10//m AD), which are generally of non-anthropogenic
17      origin (e.g., wind blown dust), require turbulence to provide vertical velocity components greater
18      than their settling velocity to allow them to be lifted and remain suspended in the air
19      (Figure 5-1). Particles of ambient origin enter into  an indoor setting either by bulk flow, as
20      through an open window, in which all particles can enter at the inlet condition, or by pressure
21      driven drafts and diffusional flows through cracks and fissures in the barriers of the building
22      envelope when all windows are closed. In the latter mode of entry, velocities are relatively
23      lower, thereby increasing the settling out of the largest coarse particles (>25//m AD) in the
24      passage through the barriers (Larssen et al., 1993; Thatcher and Layton, 1995).
25           Indoor settings are usually quiescent (Matthews et al., 1989), unless fans  or Heating-
26      Ventilation-Air Conditioning (HVAC) systems are  in use. Ambient particles that enter indoors
27      quickly settle out by gravity or electrostatic forces, leading to familiar dust layers on horizontal
28      surfaces and vertical TV screens that require constant cleaning (Raunemaa et al., 1989; Kildeso
29      et al., 1999). However, human activity in indoor settings, such as smoking and cooking, does
30      generate fine particles (<2.5//m); cooking, dusting, vacuuming and general activity can generate


        October 1999                              5-7        DRAFT-DO NOT QUOTE OR CITE

-------
 1      coarser particles (>2.5 //m), or resuspend coarse particles that previously had settled out
 2      (Litzistorf et al, 1985; Thatcher and Layton, 1995; Abt et al, 1999a,b).
 3           The National Research Council (1991) summarized the changing concept of concern for
 4      computing a person's total exposure to air pollutants, such as PM, instead of using an ambient
 5      concentration as a surrogate, as follows:
 6      "Advances in indoor-air exposure studies have demonstrated the significant health effects from
 7      indoor emissions and exposures to contaminants that had been regulated only as outdoor
 8      pollutants. . .These demonstrations of high indoor contaminant levels showed the importance of
 9      accounting for incremental exposures from microenvironments when making risk assessments. . .
10      These data have indicated the potential importance of indoor sources of contaminants."
11           However, from efforts to help resolve the apparently paradoxical situation that health
12      effects were being associated with ambient PM concentrations, but personal exposures to total
13      PM were found to be uncorrelated with ambient PM concentrations, a new realization began to
14      emerge.  The early works of Janssen et al. (1995) at Wageningen University, and Tamura et al.
15      (1996a) at Tskuba University, revealed that longitudinal total PM exposures of an individual
16      were highly correlated with ambient PM  concentrations because the variance of non-ambient PM
17      exposures (same home and work place from day to day) seemed to be small compared to the
18      variance of ambient PM concentrations.  Retrospective analyses of other data (Lioy et al., 1990;
19      Clayton et al.,  1993) by Mage and Buckley (1995) also showed that the mean community
20      personal exposure on a given day was much more positively correlated with ambient PM
21      concentrations than the cross-sectional correlation of all the subjects' exposures taken
22      individually and analyzed together.
23           The non-peer-reviewed status of these publications (Tamura et al. [1996a] was in press at
24      the time) led to a very conservative treatment of these findings in the 1996 PM AQCD (U.S.
25      Environmental Protection Agency, 1996). Therefore, the 1996 PM AQCD, Chapter 7,
26      represented a cautious transition away from the  1982 PM AQCD, and, for the first time, it made
27      a distinction between exposure to PM of ambient origin and exposure to total PM of all origins.
28      Consequently, the exposure chapter gave equal emphasis to the presentation and discussion of
29      sources of PM in indoor domestic microenvironments and the infiltration of ambient PM into
30      these indoor microenvironments. However, in summary, several suppositions and conclusions
31      were cited, that, when woven together, could implicitly support the premise that exposures to PM

        October 1999                              5-8        DRAFT-DO NOT QUOTE OR CITE

-------
 1      of ambient origin are highly correlated with ambient PM concentrations and would make a
 2      correlation of ambient PM concentration with community health effects plausible. For example:
 3      "For the morbidity/mortality studies described in Chapter 12 that use SAM [stationary ambient
 4      monitoring] as the independent variable, that SAM can be interpreted to stand as a surrogate for
 5      the average community exposure to PM from sources that influence the SAM data." (Volume I,
 6      page 7-119).
 7      "(4) Personal exposures to outdoor-generated PM of any size fractions PM10 can be estimated
 8          from the fraction of time spent indoors and an estimate of the air exchange rate and
 9          deposition rate associated with that size fraction.
10      (5) The relationship between ambient concentration and [total] personal exposure [to PM] is
11          better for finer size fractions of ambient PM than for coarser PM. Higher correlations of
12          ambient concentrations and personal exposures have been found for fine PM constituents
13          (such as sulfates) without indoor sources.
14      (6) For a study population of nonsmokers for which there is a significant positive correlation
15          between personal exposures and ambient concentrations, the ambient concentration can
16          predict  the mean personal exposure with much less uncertainty than it can predict the
17          exposure of any given individual.
18      (22) Variations in personal exposure due to fluctuations produced by indoor sources of PM are
19          independent of the variations  in personal exposure produced by [fluctuations of] ambient
20          PM." (Volume I, pages 7-163 and 7-165)
21          Since 1996, the work of Janssen and colleagues has been peer reviewed and published, and
22      several other related articles discussing it have been published, or accepted for publication [Mage
23      et al., 1999; Wallace, 1999a] - and  are reported on later in this document.  The net result is that
24      the current literature appears to support the concepts that:
25      (1)  ambient PM concentration is an index of exposure to PM of ambient origin, and that this
26          index is  currently the most appropriate  quantity to relate to human health effects associated
27          with the ambient PM concentrations within communities in epidemiologic studies;
28      (2)  a measurements of an individual's personal exposure to total PM from all sources, indoors
29          and outdoors, ambient and non-ambient, is the most appropriate index to use to relate that
30          individual's health effects from that combined exposure to PM from all those different
31          categories of sources.

        October 1999                              5-9         DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.1.1.1  Caveat
 2           In virtually all the experimental studies of personal exposures to PM to be described in the
 3      rest of this document, with noteworthy exception of the U.S. Environmental Protection Agency
 4      Particle Total Exposure Assessment Methodology (PTEAM) study (Clayton et al., 1993), the
 5      Toronto, Ontario study of Pellizzari et al. (1999), and the Expolis exposure study (Jantunen et al.,
 6      1998), the studies were all conducted with subjects who were not chosen by a scientific
 7      probability-sampling schema (SPSS). Strictly speaking, without an SPSS, the results of all such
 8      studies apply only to the subjects sampled on the days that they were sampled, and no valid
 9      inference can be made to any other population or period of time.  Although such studies may
10      report significant differences, confidence intervals andp values, albeit they were peer reviewed,
11      perhaps not by survey statisticians, they have no statistical meaning.
12             "In many cases researchers are reluctant to face the problems that  may be
13             present in the survey. An 'ignorance is bliss'  attitude and gratuitous
14             assumptions are made about the quality of the data (the million nonrespondents
15             are adequately represented by the ten respondents).  For a one-time survey,
16             conventional wisdom often dictates using methods that are believed to give
17             good results for the funds available without adequate investigation of
18             alternatives." (Lessler and Kalsbeek, 1992).
19           This chapter reports experimental results of such non-SPSS studies for what they are,
20      without the authors' reports of statistical significance where such would not be valid, and it
21      integrates the content of all these studies with the abovecaveat in mind.
22
23      5.1.2 Exposure to PM of Ambient Origin and Total PM
24           Personal exposure to the PM of ambient origin is important for several reasons:
25      (1)  The U.S. EPA regulates PM emitted into the atmosphere from mobile sources and stationary
26          industrial or commercial sources, but it does not control PM  emissions in any private indoor
27          location.
28      (2)  The human body may react differently to PM of ambient origin and PM of non-ambient
29          origin, because such particle mixtures have different chemical composition.
30      (3)  Comparison of personal exposures to mixtures of PM of ambient origin and personal
31          exposures to mixtures of PM of non-ambient origin, if possible to  differentiate, may provide
        October 1999                             5-10        DRAFT-DO NOT QUOTE OR CITE

-------
 1          clues as to the difference in their acute toxicity on a unit size and mass basis (Siegmann
 2          etal, 1999).
 3      (4)  Often in epidemiologic studies, the lagged daily mean ambient PM concentration level in a
 4          community is used as a surrogate to characterize a subject's exposure to ambient PM.
 5          Personal exposure to PM of non-ambient origin might act as a confounder or an effect
 6          modifier in such epidemiologic studies, if people's exposures to non-ambient PM are
 7          correlated with their exposures to ambient PM.  This is true even if exposure to non-ambient
 8          PM may cause similar but independent health effects.
 9           In general one can think of an individual's total exposure to PM as arising from several
10      distinct categories:  1 - Ambient PM;  2 - Outdoor PM; 3- Indoor PM; 4- Personal activity PM;
11      5- Personal PM. Figure 5-2 shows how people can be exposed to combinations of these
12      categories while outdoors as well as indoors.  The following describes these categories in detail
13      to clarify their differences and definitions as used in this chapter:
14      (1)  Ambient PM: As defined in Section 5.1, PM of ambient origin is that PM that is formed in
15          the ambient atmosphere and emitted into it.  Ambient PM is well mixed in the outdoor air so
16          that all people in the community are  exposed to it, over time, at approximately the same
17          mean concentration. This is true more so for fine mode PM than for coarse mode PM.  The
18          major sources of primary and secondary ambient PM species are industry, traffic, commerce,
19          domestic emissions such as wood smoke, and natural wind blown dust or soil (see
20          Chapter 3).
21      (2)  Outdoor PM:  PM of outdoor origin differs conceptually from PM of ambient origin. It is
22          measured as the difference between the PM concentration at an outdoor location which is
23          not in the ambient atmosphere (i.e., on private property or by a road side) and the
24          simultaneous ambient PM concentration. Scaperdas and Colvile (1999) give an example  of
25          such non-representative air quality monitored outdoors 5-meters from an urban intersection.
26      (3)  PM emitted or formed indoors: PM is emitted at home as house dust resuspended by human
27          activity and cleaning procedures, environmental tobacco smoke (ETS),  cooking fumes, pets,
28          etc.  PM emitted indoors at work varies with type of occupation. Aerosol formation occurs
29          indoors from the dark (no sunlight) reaction of ozone with gaseous terpenes and other
30          hydrocarbon species such as cc-pinene and limonine, often found in household deodorizers.
31          Kamens et al. (1999) reported " Some of the products have subcooled liquid vapor pressures

        October 1999                             5-11        DRAFT-DO NOT QUOTE OR CITE

-------
              People Are Exposed to a Variety
               of Types of Particles Outdoors
                       Local
                     Outdoors,
Community
 Ambient,
                     Personal
                     Activity
                      pact
   Individual Only,
      pind
             People Are Exposed to a Variety of
                  Types of Particles Indoors
               Indoor/Ambient
              Interaction, Pa x
                 Personal
                 Activity
                  Pact
Indoor sources,
    Pi
   Outdoor Ambient
     Infiltrated
   > Indoors, Pa|
                         'S£ S S SS SS S 'SS /"^
                       /      I       \
      Individual Only,
          pind
 Figure 5-2.  Categories of particle exposure outdoors and indoors.

 Source: Wilson and Mage (1999).

October 1999                       5-12      DRAFT-DO NOT QUOTE OR CITE

-------
 1          which are low enough to initiate self-nucleation." Weschler and Shields (1999) report on a
 2          study of these dark ozone reactions, and report "The results demonstrate that ozone/terpene
 3          reactions can be a significant source of sub-micron particles in indoor settings", as high as
 4          95 //g/m3 under simulated conditions.  Indoor generated PM is considered to be well mixed
 5          in an indoor microenvironment.
 6      (4)  Personal activity PM: Personal activity sources can exist either  indoors or outdoors.  These
 7          are microscale PM generating activities that primarily influence the exposure of the person
 8          performing the activity, from either a PM generating activity (hobby or occupation) or a
 9          physical activity that brings the subject into the undiluted PM plume from a local source
10          (standing on a street corner or holding a lit cigarette between puffs). Thus personal activity
11          PM exposure is only measured by a personal monitor carried by the subject, because a
12          stationary monitor located nearby will not measure the high PM concentration generated by
13          that activity. The difference between a personal monitor measurement and an
14          area-representative measurement several meters away is sometimes called a "personal
15          cloud" (Wallace, 1999a).
16      (5)  Exposure to personal PM: This category pertains to all PM exposures that cannot be
17          measured by a personal exposure monitor (PEM). i) Actively smoking creates a stream of
18          smoke with a high PM concentration that is inhaled directly and it is not sampled by a
19          personal monitor carried by the subject.  This contribution to a daily exposure can be
20          estimated from the Federal Trade Commission (FTC) ratings of mg delivered divided by the
21          estimated total ventilation volume inhaled during a 24-h period by the smoker (Federal
22          Trade Commission, 1994). For a smoker who breathes at an average rate of 10 Lpm over a
23          day, each 1 mg of tar inhaled represents an exposure increment of 70//g/m3 to their daily
24          PM exposure as measured by a personal PM monitor,  ii)  Wearing a courtesy-mask for a
25          respiratory infection, or a dust-mask or respirator for an occupation or hobby, removes an
26          unknown fraction of the PM measured by a PEM in the inhaled  air.  The resulting
27          contribution to daily PM exposure can only be estimated from the efficiency of the filter
28          given by the manufacturer and the PEM data.
29           An important distinction that was developed in the 1996 PM AQCD between exposures to
30      ambient PM25 and non-ambient PM25 (categories 2 to 5 above) is the relative homogeneity of the
31      concentration of PM25 of ambient origin compared to the heterogeneity of exposures to PIV|5 of

        October 1999                              5-13        DRAFT-DO NOT QUOTE OR CITE

-------
 1     non-ambient origin. As discussed in the following section, virtually all people in a community
 2     are routinely exposed to a similar mixture of PMj 5 of ambient origin when in the ambient
 3     atmosphere, with more heterogeneous exposures to coarse mode PM and semivolatile PM
 4     constituents in the western U.S. than the eastern U.S. See the discussion in Chapter 4 of the
 5     spatial and temporal variability of chemical composition of ambient PM in rural and urban
 6     communities in different regions of the U.S.
 7
 8
 9     5.2   EXPOSURES TO PM CONCENTRATIONS IN THE COMMUNITY
10            AMBIENT ATMOSPHERE
11          This section addresses the question, "How well does a concentration of PM measured at a
12     community ambient monitoring station reflect the PM concentration in the outdoor air elsewhere
13     in that area at both the local and regional levels?"
14     Spatial variation ofPMat the local level.  Kotchmar et al. (1987) measured TSP, and fine
15     (PM25) and coarse (PM10 - PM25) using dichotomous samplers, in five cities across the U.S., at
16     Bakersfield, CA; Riverside, CA; Granite City, IL;  Owensboro, KY; Tampa, FL. In each city
17     three identical PM monitors of each type were sited in accord with EPA guidelines at separations
18     ranging from  1.6 km to 10 km. The monitors were run simultaneously for periods ranging from
19     two weeks to two months.  The means of the PM^ monitoring data within each city were found
20     to be highly uniform but the coarse PM  data had a major variability. The authors concluded
21     "Only fine particles were found to have  equivalent mean concentrations, suggesting that only one
22     monitoring site in each community is required to provide an adequate exposure estimate of the
23     outdoor component. However, variable concentrations of total inhalable and/or coarse particles
24     were found, which implies a requirement for multiple monitoring sites [for sampling ambient
25     coarse mode particles]".
26          Quackenboss et al. (1991) studied exposure to PM;0 and related health effects in Tuscon,
27     AZ. The ambient PM10 data measured by the Pima County Department of Environmental Quality
28     (PCDEQ), method unspecified, were compared to the outdoor PIV(0 measured at the homes of the
29     subjects in the study, using a Harvard indoor aerosol sampler (Marple et al., 1987). The
30     distances between the homes and the PCDEQ monitors ranged up to 20 km. The Harvard-
31     indoor-sampler data measured outdoors had a statistically significant slope of 0.63 ± 0.03 when

       October 1999                             5-14        DRAFT-DO NOT QUOTE OR CITE

-------
 1      fit through the origin [intercept = 0] versus the ambient PM10 values reported by PCDEQ. The
 2      authors do not report any collocated inter-comparison between these two different monitors to
 3      document whether the 0.63 factor was more due to differences in instrumental inlet penetration
 4      curves, than to higher PM at the ambient monitoring sites. The spread of the daily outdoor PMg
 5      about the ambient values was quite large (R2 = 0.185), but no data are provided to show whether
 6      this variance is predominantly in either the coarse mode or the fine mode of the PMg.
 7           Lillquist et al. (1998) measured outdoor PMj0 at three hospitals in Salt Lake City and
 8      compared the measurements from November 29, 1994 through April 29, 1995 with the
 9      simultaneous measurements at the Utah Department of Air Quality (DAQ) ambient monitoring
10      station nearby.  The three hospitals were 3.4, 5.6 and 12.5 km from the DAQ site, and the mean
11      PM10 values measured were quite different, leading to the conclusions that "Under conditions of
12      high atmospheric PM10 concentrations...Salt Lake City, UT requires more than one monitor.
13      When ambient PM10 concentration data are used as a predictor of individual exposure, more than
14      one centralized monitor is absolutely necessary."
15           These results for the Salt Lake City, UT area are in direct contrast to those reported by Pope
16      et al. (1999) at a similar scale of separation, in the same season and general area (Provo, UT in
17      the Utah Valley) one year later (November 18, 1995 - March 15, 1996). In this period the Utah
18      Valley PM10 data monitored at three sites (at separations of 4 to 12 km) were virtually identical,
19      with Pearson correlations of 0.92 and 0.96.  The greater degree of variability in the PMg of the
20      Salt Lake City, UT area, relative to the Provo, UT area, may be related to the higher presence of
21      wind-blown crustal material in the Salt Lake City area. Pope et al. (1999) reported that increased
22      health effects in the Utah Valley were associated with stagnation and thermal inversions leading
23      to a buildup of anthropogenic PMj0, whereas the similarly high concentrations of PM,0 created by
24      high winds picking up crustal materials were not associated with increased health effects in that
25      same community. Thus, this is an indication that it may be important to differentiate exposure to
26      PM of ambient anthropogenic origin from locally-variable high concentrations of wind blown
27      dust, as in  Salt Lake City, which may not be as important for health effect prediction as
28      anthropogenic ambient PM.
29           Buzorius et al. (1999) measured short-term aerosol number concentrations by using a
30      condensation particle counter at several locations in metropolitan Helsinki, Finland. They report
31      that number concentration can vary in magnitude with the local traffic intensity,  and that "during

        October 1999                             5-15        DRAFT-DO NOT QUOTE OR CITE

-------
 1      the working days concentration averages of 10 min - 1 h are good representatives of
 2      concentration variation in relatively large areas of the city." They conclude, "Therefore, by
 3      sampling at one point in the [urban] space one can describe changes in relatively large area of the
 4      city with correlation coefficient > 0.7".
 5           Dubowski et al.  (1999) point out that although the relationship of small variation of PM5
 6      mass concentration may hold for a community, there may be significant spatial variations of
 7      specific components of the total mass on a local scale.  An example is given of a study of
 8      concentrations of polycyclic aromatic hydrocarbons (PAH) at three indoor locations in a
 9      community; urban and semi-urban separated by 1.6 km and a suburban site located further away.
10      The authors found the geometric mean PAH concentrations at these three locations varied
11      respectively as 31 : 19:8 ng/m3, and suggest that the local variations in traffic density were
12      responsible for this gradient.  Note that these concentrations are 1000 times lower than the total
13      mass concentration, so that such a gradient of 0.03 to 0.01//g/m3 for these components would not
14      be noticeable for total PM25 mass measurements of order 25//g/m3.
15           Jedrychowski and Flak (1998) report on the spatial variation of suspended particulate
16      matter (SPM) measured as  "black smoke" (BS) in Cracow, Poland during the years 1991 - 1995.
17      The authors report that, for both winter and summer, the city could be divided into a central
18      "high pollution zone" and regions of lower pollution concentrations decreasing with distance
19      from the city center. The authors do not report the size fraction of the PM collected or the
20      calibration procedure  used to  convert the BS reading into SPM//g/m3. Thus, their results may
21      emphasize the effect of the black carbon content related to local traffic density variations which
22      could produce a gradient signal imposed on a more uniform background of well mixed emissions
23      ofPM.
24           Vakeva et al. (1999) measured the vertical gradient of submicron particles in an urban
25      street canyon of Lahti, Finland. They monitored number concentration by using  a TSI screen
26      diffusion battery and a condensation particle counter at 1.5 m and 25 m above the street at
27      rooftop level. "It was concluded that dilution and dispersion decreases the concentrations of
28      pollutants emitted at street level by a factor of roughly 5 between the two sampling heights."
29      The presence of such  a local vertical gradient of concentrations for people living in high-rise
30      buildings may need to be considered in studies  of exposures to PM of urban populations.


        October 1999                             5-16        DRAFT-DO NOT QUOTE OR CITE

-------
 1      Spatial variation of ambient PM on a regional scale.  Burton et al. (1996) report on the spatial
 2      variability of sulfates and PM in metropolitan Philadelphia, PA and their data, also discussed by
 3      Wilson and Suh (1997), show very little variance in spatial mass concentrations.  They showed
 4      that there was a gradient of ammonium sulfate across the city, with a maximum in the urban area
 5      center, indicative of ammonia generation by human activities, leading to a composition variation
 6      of that component of urban PM.
 7           Keywood et al. (1999) reported their analysis of PM variability measured in six Australian
 8      cities using a MOUDI sampler.  They reported that PM^ was more highly correlated with PM2 5
 9      than with coarse PM (PM10 - PM2 5), suggesting that "variability in PMj0 is dominated by
10      variability in PM25".  Although the authors found that mass of PM^ was highly correlated with
11      mass of PMj (r2 = 0.98), the mass of the ultrafines (nuclei mode), reported both as PIVj 15 and by
12      integration under a fitted curve, had weak correlations with PM.5 of r2 = 0.50 and 0.01,
13      respectively.  This suggests that although a single monitoring station may be adequate for
14      characterizing fine mode ambient PM in a community, the ultrafine mode mass of ambient PM,
15      like the coarse mode mass, may require additional monitoring.
16           Leaderer et al. (1999a) monitored 24-h PMj0, PM25 and sulfates during the summers of
17      1995 and 1996 at a regional site in Vinton, VA (6 km from Roanoke, VA).  One similar 24-h
18      measurement was made outdoors at residences in the surrounding area, at distances ranging from
19      1 km to > 175 km from the Vinton, VA site, at an average separation distance of 96 km.  The
20      authors reported significant correlations forPM2 5 and sulfates between the residential outdoor
21      values and those measured at Vinton, VA on the same day.  In addition, the mean values of the
22      regional site and residential sitePM25 and sulfates showed no significant differences in spite of
23      the large distance separations and mountainous terrain intervening in most directions. However,
24      for the concentrations of the coarse mode PM, estimated as PM0 - PM25, no significant
25      correlation among these sites was found (n = 30, r = -0.20).
26           Jantunen et al. (1998) and Koistinen et al. (1999) report on the protocol and quality
27      assurance procedures of the EXPOLIS exposure monitoring study that included measurements of
28      ambient and microenvironmental concentrations of PM,5, and personal exposures to PM^.
29      A planned article (Jantunen et al., 2000) is expected to report the final results of their exposure
30      analyses on these measurements. Their preliminary results show that: (1) in Basel and Helsinki,
31      a single ambient monitoring station was sufficient to characterize the ambient PM5

        October 1999                             5-17        DRAFT-DO NOT QUOTE OR CITE

-------
 1      concentration in each city - only a few simultaneous measurements of ambient PM5 were made
 2      in the four other EXPOLIS cities (Athens, Grenoble, Milan and Prague) so no other intra-city
 3      comparisons are available at this time; and (2) by using microenvironmental concentration data
 4      collected while the subjects were at home, at work, and outdoors, the time-weighted-averages of
 5      these data closely match the personal PM, 5 exposure data collected by the monitors carried by
 6      most of the subjects, with a few subjects, mostly smokers, being noticeable exceptions.
 7           In summary, a PM2 5 measurement at a properly cited ambient monitoring location can
 8      represent the mixed mean concentration of ambient PM, 5 that exists in the ambient atmosphere
 9      of that local area. Coarse mode PM is apparently influenced more by local sources than PM5 so
10      this finding of a single monitoring site as being sufficient for PIV|5 is not generally applicable to
11      the coarse fraction of PM.
12           White (1998) suggests that the higher random measurement error for the coarse PM
13      fraction compared to the error for the fine PM fraction may be responsible for a major portion of
14      the apparent greater spatial variability of coarse ambient PM concentration compared to fine
15      ambient PM concentration in a community (e.g., Burton et al., 1996; Leaderer et al., 1999a).
16      When PM25 and PM10 are collected independently, and the coarse fraction is obtained by
17      difference, as PM10 _2 5 = PM10 - PM25, then the expected variance in the coarse fraction is the
18      sum of the variances of the PM,0 and PM25 measurements. When a dichotomous sampler
19      collects PM25 and PM10_25 on two separate filters, the coarse fraction also is expected to have a
20      larger error than the fine fraction. There is a possible error due to loss of mass below the
21      cut-point size and a gain of mass above the cut-point size which is created by the asymmetry of
22      the product of the penetration times PM concentration about the cut-point size.  Because a
23      dichotomous PM sampler collects coarse mass using an upper and lower cut-point, it is expected
24      to have a larger variance than for the fine mass collected using the same lower cut-point.
25           Carrothers and Evans (1999) also discuss the effect of relative measurement errors (both
26      instrumental error and Berkson-type error) when analyzing the relative toxicity of ambient PM5
27      to that of the coarser mode (PMj0 - PM2 5 ).  They present a model that allows an estimate of the
28      relative bias in the regression coefficients of coarse and fine PM on daily mortality, and conclude
29      that "if one pollutant is truly more harmful than the other, then it must be measured more
30      precisely than the other, in order not to bias the ratio of the fine and coarse regression
31      coefficients". The authors note "the need for spatial variability data, i.e., several ambient

        October 1999                             5-18        DRAFT-DO NOT QUOTE OR CITE

-------
 1     monitors situated across a metropolitan area for a period of many months" so that "definitive
 2     conclusions can be made regarding the possibility of bias due to differences in measurement
 3     [Berkson-type] error among correlated pollutants".
 4           Wilson and Suh (1997) review the difference between the chemistry, sources and factors
 5     influencing exposures to generic fine and coarse ambient PM (see Chapter 3). They point out
 6     "The infiltration factor, which gives the  fraction of outdoor particles found indoors, ... is greater
 7     for the fine particles than for coarse particles, largely because of the lower indoor lifetimes of
 8     coarse particles relative to fine particles."  Wilson and Suh (1997) hypothesize that because the
 9     ambient monitoring data for PMj 5 are more constant across an urban area than the corresponding
10     concentration data for coarse mode PM, represented by (PM^ - PM25), and PM10 is more highly
11     correlated with PM25 than with (PM10 - PM25), the health effects associated with total PM,0 are
12     more likely to be related to the variations of the fine mode portion of the PIVf5 than the
13     variations of the (PM10 - PM25) measure of the coarse fraction.
14           The authors note that PM25 is most often a mixture of fine mode PM and coarse mode PM
15     because the lower tail of the coarse mode extends below 2.5//m AD, and since 2.5 //m is a 50%
16     cut point, not a 100% cut point, a PM^ sampler collects some PM > 2.5//m.  They conclude that
17     "Fine and coarse particles are separate classes of pollutants and should be measured separately in
18     research and epidemiologic studies. PM;0 and PM(10_25) are indicators or surrogates, but not
19     measurements, of fine and coarse particles." Janssen et al. (1999a) report "A method to estimate
20     the distributions of various fractions of PM,0in ambient air in the Netherlands". However, their
21     method estimates the PM25 fraction, but not the fraction of PMj0 that is exclusively in the fine
22     mode aerosol, as recommended by Wilson and Suh (1997).
23
24
25     5.3   EXPOSURES TO AMBIENT PM IN INDOOR
26            MICROENVIRONMENTS
27           Besides the exposure to ambient PM while outdoors, people are also exposed to PM of
28     ambient origin in the non-ambient  indoor-type microenvironments of the residence, workplace,
29     school, motor vehicle, etc. Ambient PM enters indoors from the  outdoors by both forced and
30     free convection. An equal quantity of indoor air must also leave through the same type of


       October 1999                            5-19        DRAFT-DO NOT QUOTE OR CITE

-------
1
2
3
passages to maintain the pressure equal between the indoors and outdoors. Figure 5-3 shows an
idealized indoor micro environment exchanging air with the ambient surroundings.
              v air flow out
                                                           Co  outdoor
                                                                 ambient PM
                                                                 concentration
                                                       airflow in
                                                                P fraction of PM
                                                                penetrating into
                                                                the indoor region
                      Volume V
                       emission rate =XM
                                 C  Indoor concentration
                                 of ambient PM

                                          I K deposition rate
                    Surface Area (A) with Mass (M) of Ambient PM on it
      Figure 5-3.  Two compartment model for PM deposition and resuspension by human
                 activity in a residential microenvironment.
1          As described by Alzona et al. (1979), Tung et al. (1999) and Kulmala et al. (1999) the
2     ambient PM intrusion process can be modeled for a well mixed indoor volume f7) by a mass
3     balance equation between start time t = 0, and stop time t = T, with initial condition C = C(0),
4     @ t = 0. Here C(0) is the concentration of PM that originated from the ambient air that is found
5     in the indoor air while the adjacent outdoor air has a PM concentration of Co = Co(0), @ t = 0:
                        VdC/dt=  vPCo -
                                                                           (5-2)
      October 1999
                                      5-20
DRAFT-DO NOT QUOTE OR CITE

-------
 1      where V= volume of the well mixed indoor air, m3;
 2           v =  volumetric air exchange rate between the indoor and outdoor condition, rrl/hr;
 3           P= fraction of ambient PM that is not deposited on the inlet surfaces during the entry into
 4               the indoor volume from the ambient condition, a function ofv,  0 <  P < 1;
 5           Co = concentration of ambient PM in the outdoor air that enters into indoor volumeF,
 6               //g / m3;
 7           C= concentration of the ambient PM in the indoor volume V, //g /m3;
 8           k=  deposition parameter for the ambient PM upon the interior surfaces within volumeF,
 9               1/hr; a function of the distribution of PM aerodynamic diameter (AD), presence of air
10               cleaners or in-line filters in air circulation systems, etc.
11           Qa = A discontinuous rate of resuspension of some of the ambient PM that had entered
12               volume Fat time t > 0 and had been deposited on interior surfaces,//g/hr; Qa > 0.
13           All the parameters, save for V, are assumed to be functions of time and PM AD.  For
14      example, the deposition parameter (fc) can increase as a step function if an air cleaner is turned
15      on.  Anuszewski et al. (1998) showed that in-line filters in heating systems may remove PM of
16      optical diameters less than 1 //m. The volumetric air exchange rate (v)  is always finite, but it
17      varies with wind speed, indoor to outdoor temperature difference, and  as windows  are opened or
18      closed. For particles larger than 1 //m AD, which can settle by gravity, the penetration factor/3 is
19      expected to  decrease with decreasing air exchange flow rate (v) because the time that is available,
20      for gravitational deposition in passage through cracks and fissures, increases as the flow rate
21      decreases. Suh et al. (1993) showed that operation of an air conditioning system leads to usage
22      of a lower air exchange rate, resulting in lower PM concentrations of ambient constituents, such
23      as sulfates.  The resuspension parameter (Q) can be zero when the volume Vis unoccupied, or
24      while occupied but people are sleeping or sedentary and Q > 0 when people are active within V.
25           It is important to note that the Koutrakis et al. (1992) reformulation of Equation 5-2,
26      presented as Equation (7-3) in U.S. Environmental Protection Agency (1996), was used to
27      compute average values of deposition rate (fc) independent of any resuspension of deposited
28      material.  This deposition parameter (fc) will vary widely for PM10 as its relative amounts of fine
29      mode and coarse mode PM vary. Therefore, the term Qs (emission factor of indoor sources,
30      in units of//g/hr) in Equation (7-3) op. cit., contains within it the resuspension of PM from


        October 1999                             5-21        DRAFT-DO NOT QUOTE OR CITE

-------
 1     ambient sources that was deposited on previous days or earlier on the very day of the
 2     measurement.
 3          If C(0) is of the order of Co(0) and we assume an average value for Q, we can estimate the
 4     average steady state value of C. This is done by setting dC/dt = 0 in Equation 5-2 and solving for
 5     C as follows:
 6
                                                       / (a+k\                      (5-3)
 7     where a =v/V, the number of air exchanges per hour.
 8           The outdoor PM (Co) at the location of the air inlet to indoor volumeFmay not be exactly
 9     equal to the ambient PM concentration (Ca) measured at the neighborhood air monitoring station
10     that can be several kilometers away. Let us assume that there are no major sources of PM
1 1     between the monitoring station where  Ca is measured and the location where Co is measured,
12     and that the surrounding urban area is  relatively homogeneous in terms of traffic, residential
13     communities and commercial activities including light industry. With this frame of reference, we
14     can model the measured outdoor concentration (Co) as equal to the measured ambient
15     monitoring station concentration (Ca)  plus a small random component ^) that has a mean of
16     order zero and a finite variance.
17           This parameter (e) covers the true spatial variation of ambient PM resulting from micro-
18     scale weather variations and local sources. For example, on some days there is a wind vector
19     component from the monitoring station to the modeled indoor location and on other days the
20     vector will be from the indoor location to the monitoring station. Sophisticated methods that
21     account for presence of sources and the topography between ambient monitoring stations are
22     available for estimating e by interpolating ambient pollutant concentrations between monitoring
23     stations (Beyea and Hatch, 1999). The parameter e also allows for random measurement errors
24     from the weighing of the filters and the measurement of the flow rate. This leads to the
25     relationship for the concentration of ambient PM in the indoor micro environment (C), in terms of
26     the measured ambient concentration (Ca) as:
27
                                                                    + k)                (5-4)
28

       October 1999                             5-22        DRAFT-DO NOT QUOTE OR CITE

-------
 1          Equation 5-4 indicates that an indoor microenvironment will tend to equilibrate with the
 2     ambient PM at a fractional value of the ambient PM concentration.
 3          Clayton et al. (1993), Ozkaynak et al. (1996a) and U.S. Environmental Protection Agency
 4     (1996) report the results of the Particle Total Exposure Assessment Methodology (PTEAM)
 5     Study carried out in Riverside, CA in 1990. 178 subjects carried personal PM0 monitors for one
 6     day each, while PM10 was being monitored in their home, outside their home (Co) and in their
 7     community (Ca).  At each home, an air exchange rate was measured using a continuously
 8     emitting source of a perfluorocarbon tracer (PFT).  Using the procedure of Koutrakis et al.
 9     (1992), Ozkaynak et al. (1996a) determined average values ofP and k for PM10 of 1 and 0.65/hr
10     respectively, and 1 and 0.39/hr for PM^.  For the 178 homes, 174 values of a were successfully
11     obtained during the daytime period from approximately 7am to 7pm. The histograms of the
12     values of P a I (a+k)  for the 174 monitored homes  are shown for PM,0 and PM2 5 as Figure 5-4.
13     These values represent an estimate of the average fraction of the outdoor PM that was found
14     inside the home.  Because of the positive resuspension term (Q) the actual values will be
15     expected to be higher than the values shown. However, there is minimal resuspension of
16     submicron particles and very little PM2 5 is resuspended (U.S. Environmental Protection Agency,
17     1996) so the reported values of k and Qis in the PTEAM study for PM25 are unlikely to have been
18     appreciably affected by ambient PM resuspension.  Exposure to the ambient accumulation mode
19     PM (-0.1 //m < AD < ~1 //m ) is inferred by the relations of exposure to sulfur and sulfates
20     which are predominantly in this size range (Leaderer et al., 1999a). PM in this size range has a
21     deposition parameter of order 0.2/hr (Ozkaynak et  al., 1996b).
22          In summary, the indoor microenvironment will have an appreciable amount of the ambient
23     PM25 equilibrated within it. In the case of the PTEAM homes during the fall season in
24     Riverside, CA, where temperatures were moderate and homes had an air exchange rate of
25     approximately a = 1/hr, the fraction of the ambient PMj0 to be found within the indoor residences
26     was approximately !/(!+ 0.65) -0.6.  For PM25 the fraction 17(1+0.39) ~ 0.7. For sulfate the
27     fraction 1/(1 + 0.16) = 0.85. The combined exposure to PM  of ambient origin both indoors and
28     outdoors is analyzed  in Section 5-7.
29
30


       October 1999                             5-23        DRAFT-DO NOT QUOTE OR CITE

-------
                                (a)
                                04   0 45  05  0 55  0 60  0 65 0 70  0.75  D 80 0.85  0 90  0 95
                                            Fraction of PM25 Found Indoors
                            "to 15'
                            £
                                (b)
                                0.15  0.20 0.25 0.30  0.35 0.40 0.45  0.50 0.55 0.60  0.65 0.70 0.75  0.80 0.8!
                                            Fraction of PM10 Found Indoors
       Figure 5-4.  Histograms of the estimated fractions of outdoor PM2 5 (a) and outdoor PM10
                   (b) found indoors during the PTEAM study in Riverside, CA (Computed from
                   data of Ozkaynak et al., 1996a).
1     5.4    EXPOSURES TO PM OF INDOOR ORIGIN
2          U.S. Environmental Protection Agency (1996), Wallace (1996), and Ott and Roberts (1998)
3     review and discuss the pre-1996 literature on PM as found indoors from indoor sources. Most
4     particles generated indoors by human activities have different chemical and physical properties
5     than those generated by anthropogenic ambient sources (Siegmann et al., 1999). In the U.S., in
6     general, combustion product PM from ambient sources is from the burning of fossil fuels (e.g.,
7     coal, gasoline, fuel oil) and wood, and combustion product PM from indoor sources is from
8     biomass burning (e.g., tobacco, wood, foods, etc.). However, some indoor sources of PM, such
9     as cigarette smoking, meat cooking and coal burning (in China), occur both indoors and outdoors
      October 1999
5-24
DRAFT-DO NOT QUOTE OR CITE

-------
 1      and may constitute an identifiable portion of the measured ambient PM by use of source
 2      apportionment techniques (Cha et al., 1996; Kleeman and Cass, 1998). During the PTEAM
 3      study (Ozkaynak et al., 1996a) some non-ambient particles that appeared on personal monitoring
 4      filters were identified as skin flakes, fibers and carpet debris. These PM species are considered
 5      as "inert or nuisance dusts" by the U.S. Department of Labor (Code of Federal Regulations,
 6      1998), which has established an 8-hour PM, 5 time-weighted-average (TWA) occupational
 7      standard of 5,000 //g/m3 for controlling them.  Some other non-inert PM species of indoor or
 8      occupational origin may be carcinogenic (e.g., cigarette tars, radon progeny) or produce chronic
 9      effects (e.g., silica, asbestos). However, there is no evidence that day-to-day fluctuations of
10      personal exposures to PM species such as these that are not known to produce acute effects at
11      lower concentrations, could cause the day-to-day fluctuations of the acute health effects
12      (mortality and morbidity) that are associated with the day-to-day fluctuations of ambient PM
13      concentrations in epidemiologic studies (Schwartz et al., 1999).
14          The major sources of indoor PM in the residence include cigarette smoking, cooking, and
15      unvented or poorly vented combustion devices such as stoves and kerosine heaters.  Human and
16      pet activities also lead to PM detritus production from track-in soil, fabrics, skin and hair, home
17      furnishings, etc., which can all be found in the ubiquitous house dust found on floors and in the
18      lint trapped by the filter of a clothes dryer. This generic house dust and lint is suspended in the
19      indoor air by air movements and, after deposition, it can be resuspended by cleaning activities of
20      sweeping, dusting and vacuuming. Other sources of PM in the home may arise from hobby
21      activity, and from para-occupational materials brought into the home by workers on their persons
22      (Sterling et al., 1995).  Biological aerosols commonly found indoors  are discussed in Chapter 7
23      of U.S. Environmental Protection Agency (1996).
24          Abt et al. (1999a) studied the PM size distribution and sources of PM in four non-smoking
25      households in the Boston metropolitan area, and confirm previous findings that the major indoor
26      emission  sources of PM are cooking, cleaning and human activity. They discuss  the size
27      characteristics of these ubiquitous sources and report "The size of the particles generated by these
28      activities  reflected their formation processes, with combustion processes  (oven cooking, toasting
29      and barbecuing) producing fine particles, and mechanical processes (sauteing, frying, cleaning,
30      and movement of people) creating coarse particles." The authors suggest that at air  exchange


        October 1999                             5-25        DRAFT-DO NOT QUOTE OR CITE

-------
 1     rates of less than 1/hr the indoor sources of PM predominate in the indoor micro environment and
 2     at air exchange rates above 2/hr, the outdoor sources of PM predominate.
 3          The smoking of cigarettes is the major contributor to PM concentrations from indoor
 4     sources in the homes where people smoke. Neas et al. (1994) report from the Harvard 6-City
 5     Study that the between 1983 and  1986, annual average PM, 5 was 3 l//g/m3 higher in the homes of
 6     consistent smokers than in the homes of consistent non-smokers.  An extensive investigation has
 7     been recently performed on environmental tobacco smoke (ETS) exposures of non-smokers in
 8     homes and workplaces in Europe by Phillips et al. (1994, 1996, 1997a,b, 1998a,b,c, 1999). For
 9     example, Phillips et al. (1999) sampled non-smokers homes and workplaces in Basel,
10     Switzerland and computed annual total ETS exposures as the mean "Potential inhaled quantity"
11     in mg of RSP.  They found that the median exposures of non-workers in smoking households
12     were 15% higher than in smoking households, and at the 90th percentile that smoking
13     households were 100% higher than the non-smoking households.
14          Jenkins et al. (1996a,b) made similar measurements in the U.S. and report that smoker's
15     homes in 1993 and 1993/1994 averaged 17 and 20 ug/m3 above the mean values in non-smoker's
16     homes, respectively. The higher effect of smoking (31 ug/ni) reported by Neas et al. (1994) in
17     1983 - 1986 may have been related to changes in smoking habits in the decade between these
18     studies, such as smoking reduced tar cigarettes and smoking less cigarettes in the home in the
19     more recent Jenkins  study.
20          Klepeis et al. (1996) measured PM, 5 due to environmental tobacco smoke (ETS) using a
21     TSI 8510 piezobalance. They measured air exchange rates in two airport glass-enclosed smoking
22     lounges and estimated a rate of emission of 1.43 mg/min per cigarette smoked by means of a box
23     model for a well-ventilated lounge volume. The PM, 5 coefficient of variation (o///) in the room
24     was 0.12 indicating that the lounge was indeed well-mixed. The authors concluded that personal
25     exposures to ETS can be satisfactorily modeled in such microenvironments. Such a model may
26     be useful for calculating the non-ETS PM found in indoor microenvironments where smokers are
27     present, which is often mostly of ambient origin.
28          For smoker's homes in Riverside, CA, the PTEAM study (Ozkaynak et al., 1996a) reported
29     that ETS constituted about 75% of the PM^ generated by indoor sources and 55% of the PM,0
30     generated by indoor sources, which corresponds to approximately 35% of the coarse PM


       October 1999                            5-26        DRAFT-DO NOT  QUOTE OR CITE

-------
 1      (PM10-PM2 5) generated by indoor sources. For those homes (smoking and non-smoking) in
 2      which cooking took place, cooking fumes were responsible for 66% of the PIV|5 indoor
 3      emissions.
 4           Occupational exposures to PM generated indoors are a major source of PM exposure for
 5      "blue collar" workers involved in the "dusty trades". OHSA standards for PM are both specific,
 6      for industries such as coal mining and textile manufacturing which have significantly different
 7      toxicities of their PM (e.g., coal dust vs cotton dust), and generic, for all "inert and nuisance
 8      dusts" not specifically mentioned (Code of Federal Regulations,  1998). An 8-h time-weighted-
 9      average (TWA) PM35 generic standard of 5 mg/m3 has been established, in part, to maintain
10      visibility in work places for personal safety, as well as for respiratory health protection.  It is
11      interesting to note that the American Conference of Governmental Industrial Hygienists (1996)
12      has established similar guidelines for generic PM, not specific to an industry ("containing no
13      asbestos and < 1% crystalline silica"), as "Particulates Not Otherwise Classified (PNOC)". Their
14      recommended 8-h TWA for such PNOC as "inhalable particulate" (PM^o) and "respirable
15      particulate" (PMO are 10 mg/m3 and 3 mg/m3, respectively.
16
17
18      5.5   PERSONAL EXPOSURES TO PM OF ONE'S OWN PERSONAL
19            ACTIVITIES
20           Personal activities, such as body motion, cigarette smoking, hobbies and occupational tasks
21      may generate a plume of particles that abruptly decreases in concentration with distance from the
22      person generating the particles.  This is especially important in certain occupational settings.
23      Average concentrations of inhalable dust [~PM50, American Conference of Governmental
24      Industrial Hygienists (1996)] over 50 mg/m3 have been measured by personal monitoring of
25      agricultural activities, with an average respirable fraction (PM;) of 4.5 mg/m3 (Nieuwenhuijsen
26      et al, 1999).
27           Teschke et al.  (1999) report on personal total PM exposure data (~PM50) from workers in
28      industry involving wood-production, wood-finishing and wood-construction, as collected by the
29      U.S. Occupational Safety and Health Administration Integrated Management Information System
30      (Stewart and Rice, 1990). The  data set consisted of 1632 observations over the period from 1979


        October 1999                            5-27       DRAFT-DO NOT QUOTE OR CITE

-------
 1     -1997. The arithmetic mean exposure was 7.93 mg/m3, the geometric mean was 1.86 mg/m3,
 2     12 values exceeded 100 mg/m3, and the maximum reported value was 604 mg/m3.
 3          During the time of the existence of the activity plume, a subject will be exposed to a much
 4     higher PM concentration than would be measured by a stationary PM monitor several meters
 5     away.  Mage and Ott (1996) analyzed this situation for a cigar being smoked in a large tavern.
 6     During the period while the cigar is burning (called thecc period) and during the period
 7     immediately after the cigar is extinguished, during which the plume is mixing into the rest of the
 8     indoor air (the P period), the person generating that plume will be exposed to a higher
 9     concentration of PM than an indoor monitor located in the same room.  Then,  after the plume is
10     mixed (the y period), the subject and a monitor anywhere  in that microenvironment will
11     experience the same concentration of that material.
12          The difference between the reading of a personal monitor and an indoor monitor during the
13     a and P periods represents the PM exposure due to a person's own personal activities.  This
14     incremental exposure above the surrounding micro environmental concentration is unique to the
15     subject, save for the case where someone else is immediately next to the monitored person (e.g.,
16     a helper holding parts together to be welded by a monitored welder).
17
18
19     5.6    PERSONAL PM EXPOSURE
20          This is PM exposure that occurs from direct inhalation of tobacco smoke by a smoker.
21     By placing the item (pipe, cigar, cigarette) to the mouth, the smoker inhales the concentrated
22     fumes directly into the lung and completely bypasses the inlet of a personal  PM monitor worn by
23     the smoker.  This exposure category is distinct from the personal smoking-activity exposure
24     described in the previous section, which can be captured by a personal monitor in the breathing
25     zone of the subject.
26          The magnitude of this source is appreciable and dominates all other categories of exposure
27     when computing the total exposure to PM of a smoker from all sources of ambient and
28     non-ambient PM. The nominal amount of PM (tars) delivered by each brand and type of
29     cigarette smoked are reported by the Federal Trade Commission (1994). The concentration
30     (mg/m3) delivered by each puff is not reported.  However, for a person breathing at an average of
31     10 Lpm over a day, or 14.4 m3/day, each 1 mg delivered adds approximately 70//g/m3 to the 24-h
       October 1999                             5-28       DRAFT-DO NOT  QUOTE OR CITE

-------
 1     average PM exposure of the smoker.  Consequently, a subject smoking 20 cigarettes (one pack)
 2     per day, rated by FTC at 10 mg tar each,  would have an average daily PM exposure of 14 mg/m.
 3          Siegmann et al. (1999) point out a major difference between cigarette smoke particles and
 4     other combustion source particles derived from candles, paper, and motor vehicles.  "The
 5     particles of cigarette smoke are much larger than the other ones, and grow with time to even
 6     larger ones" and "particles generated when a cigarette is smoked are known to contain liquid
 7     matter which will contract the particles to denser material due to the surface tension of the
 8     liquid". The latter effect would also increase the Stokes' Law settling velocity of the particle,
 9     which would increase the  effective aerodynamic diameter (AD) of the particle.
10
11
12     5.7    EXPOSURE  TO PM OF AMBIENT ORIGIN IN BOTH INDOOR
13            AND OUTDOOR MICROENVIRONMENTS
14          Let a subject living in the residence of volume V, at the location where the outdoor PM
15     concentration is Co, spend some fraction of time outdoors (y), and the remaining fraction of time
16     (1 - y) in the residence where  Equation 5-4 applies. For simplicity, let us also assume that while
17     outdoors the subject is close to home  and exposed to the local outdoor PM concentration Co
18     [Co = Ca + e, where Ca is the ambient monitoring station value ande is a random increment of
19     order zero with a finite variance]. The subject's total exposure to PM of ambient origin (Ea)
20     during a complete day will be as follows:

                 Ea - y(Ca + e) + (1 - y)[Ca Pa+ e Pa + Qa / V] I (a + k\              (5.5)

                               /3                                                       (5-6)

            where a = y + (1-y) P a I (a + k)
 1          Equation 5-6 means that a person's daily exposure to the ambient PM is proportional to the
 2     concentration Ca at the monitoring station (a Ca) plus a random variable (p). The variable P has
 3     a term with mean zero and finite variance representing the effect of the spatial variation in the

       October 1999                             5-29        DRAFT-DO NOT QUOTE OR CITE

-------
 1      ambient air (e), plus another positive term with a finite variance representing resuspension of that
 2      day's settled ambient PM by mechanical activity in volume V (walking on carpets, sitting on
 3      stuffed furniture, dusting or vacuuming, etc.).
 4           From Equation 5-6 we expect that someone living in a style such that they are not exposed
 5      to any appreciable additional occupational PM or indoor generated PM to have a measured
 6      personal exposure to total PM (E) that is highly correlated with the ambient PM concentration.
 7      However, should someone live in a style in which they are exposed to additional PM generated at
 8      work and at home, as from passive cigarette smoke, cooking, fireplace emissions, etc., then this
 9      increment to the exposure would appear mathematically as if it were a gross enlargement of the
10      random term p.  Because the strengths of these indoor sources are independent of both the
11      ambient concentration Ca and the outdoor concentration Co, the resulting correlation of total
12      personal exposure (to ambient PM plus non-ambient PM) with the ambient PM concentration
13      would be decreased, and approach zero. These are the relationships that have been observed in
14      the PM exposure literature, as discussed in the following section.
15           People are exposed to PM of ambient origin while indoors (C, defined as the concentration
16      of PM of ambient origin in the indoor microenvironment) and outdoors (Co, where Co^ Ca due
17      to possible spatial variation of PM from the monitoring station concentration Ca). Consequently
18      one must include the exposure to ambient PM indoors in any analysis of the total exposure to PM
19      of ambient origin.  The analysis below follows the formulation of Mage (1998) and Kulmala
20      etal. (1999).
21           During any given period T, the exposure to PM of ambient origin (Ea) is defined by
22      Equation 5-7.
                                            T            T
                               Ea  =  1 / T [  J SoiCo dt + J 8io C dt]                       (5-7)
                                            o            o
23
24      where 8oi = 1 if outdoors and = 0 if indoors; 8io = 1 if indoors and = 0 if outdoors.
25           Because of the current inability to monitor the quantity of ambient PM inside the indoor
26      locations where people spend their time, Equation 5-7 is often modified to estimate Ea by
27      assuming either 8io = 0 or C = Co at all times, resulting in Equation 5-8 as follows:


        October 1999                              5-30        DRAFT-DO NOT QUOTE OR CITE

-------
                                                 1
                                       Ea-1/TjCodt                               (5-8)
 1          Equations 5-7 and 5-8 are both attempts at addressing the question, "How much of the
 2     ambient PM that existed in the ambient atmosphere during the previous T = 24-h period are
 3     people exposed to during that period?" However, in practice, a calculation complication exists.
 4     A portion of the coarse mode ambient PM that enters the home and then is deposited in the home
 5     during the interval 0 < t < T is resuspended by human activity during the 24-h measurement
 6     period (Kildeso et al., 1999).  There is expected only minimal resuspension of fine PM directly,
 7     but any fine PM that settled out and adhered to a coarse particle could also be resuspended. It is
 8     difficult, if not impossible, to distinguish the resuspended ambient PM from the PM of
 9     non-ambient origin that is also resuspended by the same human activities.
10          Figure 5-3 shows the two compartment model that is used for this analysis.  The first
1 1     compartment is the indoor air of volume V and the second compartment is the surface area (A)
12     within volume V where ambient PM mass is deposited.  The other parameters are as defined
13     previously. The PTEAM study (Ozkaynak et al., 1996a,b) reports widely variable values olfr for
14     night and day conditions which may be influenced by diurnal patterns of resuspension.  The
15     fraction of the mass (M) of freshly deposited ambient PM on surfaced that is resuspended per
16     unit time is a function of human activity, such as walking on a surface and creating vibrations
17     and air currents sufficient to levitate a particle.  The differential equations that describe the daily
1 8     accumulation of ambient PM on interior surfaces (M) and the time variation of ambient PM in
19     volume V are as follows:
20
                        VdC/dt  =  PvCo-vC-kVC  +  7  M                 (5-9)
21
                                 dM/dt = kVC - 7  M                           (5-10)
22
23     where  y Mis the resuspension rate (mass/time)  of that day's previously deposited ambient PM
24     that is  assumed to be proportional to the mass $f) deposited on the surface, y ^0.
       October 1 999                             5-3 1        DRAFT-DO NOT QUOTE OR CITE

-------
 1            The solutions to equations 5-9 and 5-10 are coupled, and C and M must be solved for
 2      numerically using the randomly varying values of Co, P, k, v, and y due to weather and human
 3      activity in the indoor location.  The initial conditions for the integration are as follows:
 4      M = M(0)@t = 0;C = C(0)att = 0.
 5            At the start of the integration at t = 0, the ambient PM inside an indoor microenvironment
 6      (C) is some unknown fraction of the outdoor PM that developed from air exchange over the
 7      previous 24-hrs; if some of that material may already be deposited on the indoor surfaces then
 8      M(0) > 0. In the following analysis, the mass of ambient PM on the surfaces that had deposited
 9      from the previous weeks' depositions (since the last house cleaning) is assumed to be
10      uncorrelated with the concentration of ambient PM on the measurement day so it does not enter
11      into the equation.
12            Except for the trivial conditions where  y is equal to zero, the solutions to coupled
13      Equations 5-9 and 5-10 are beyond the scope of this chapter, because several parameters and
14      initial conditions  are unknown. For example, y is close to zero when the indoor occupants are all
15      asleep and/or sedentary; it is larger while the occupants are moving about; it is maximal while an
16      occupant is dusting, sweeping or vacuuming.  The value of the air exchange parameter v varies
17      with the wind speed and varying window and door openings, and k varies with the size
18      distribution of the mixture of PM in the ambient air that penetrates into the home. Consequently,
19      the approach chosen here is not to solve them simultaneously because of the unknown
20      parameters. Rather, Equation 5-9 is solved with y = 0, and the result is reported as an inequality
21      because the exposure to PM of ambient origin  with resuspension (y > 0) must be greater than the
22      exposure without resuspension (y = 0).
23            Table 5-1  summarizes some of the parameters necessary for creating a solution to
24      Equations 5-7 and 5-9.  The air exchange rate (a) is the ratio v I V,  and it was measured in the
25      PTEAM study by collection of a continually emitted tracer gas inside the subject homes.  These
26      data for a contain two important artifacts that lead to a negative bias (an underestimation) of the
27      average value of a.  First:  When the air exchange rate was too high, the collected tracer was
28      below the minimum detectable level (MDL) of the analytical procedure.  Ozkaynak et al.,
29      (1996a,b) used the MDL values of a as alternative default values for computing k and P in order
30      to maximize the amount of data available for the analysis using a non-linear optimization
31      procedure (Koutrakis et al., 1992).  Second: The concentration of tracer gas, and therefore the

        October 1999                              5-32        DRAFT-DO NOT QUOTE OR CITE

-------
                        TABLE 5-1.  SUMMARY OF THE MEAN VALUES OF
                            PM VARIABLES FROM THE PTEAM STUDY
Parameter
& (I/ hour)
a* (I/ hour)
J3**
Co (//g/m3)
Qother(mg/hr)***
PM25
7am-7pm
Day
0.27
1.144
1
48.9
1.46
PM25
7pm-7am
Night
0.39
0.98
0.89
50.5
0.784
PM25
7am-7am
Combined
0.39
0.97
1
49.7
1.08
PM10
7am- 7pm
Day
0.91
1.144
1
94.9
14.3
PM10
7pm-7am
Night
0.43
0.98
0.88
86.3
2.82
PM10
7am-7am
Combined
0.65
0.97
1
90.6
5.64
         Source of Data: Ozkaynak et al. (1996a).
         * Values of air exchange above the maximum level of detection (LOD) were assumed equal to the LOD.
         **P was constrained to the range 0
-------
 1         locations, or for all homes in a given location. For example, where it is hot during the day
 2         and cool at night, people with an HVAC system may keep windows closed during the day
 3         and open them at night to cool off.
 4      2.  Indoor emissions (including resuspended ambient PM) are much greater for coarse PM
 5         (PM10 - PM2 5) than for PM2 5, which is consistent with the literature findings that fine PM is
 6         held more tightly to surfaces than coarse PM. Resuspension of previously settled PM is
 7         expected to be greatest during the day when people are most active (Roorda-Knape et al.,
 8         1998).  For example, Larssen et al. (1993) report a study of three days in an unoccupied
 9         apartment in Oslo on a busy street, showing that the indoor fraction of ambient coarse PM in
10         their study is of order 0.2. In the PTEAM study the PM collected during the second 12-h day
11         period contained some of the PM deposited during the first 12-h night period that was
12         resuspended by daytime activity.  This increased the amount of the nighttime PM that the
13         subject was exposed to during the 24-h period, but the analysis treated it as a component of
14         Qother-
15      3.  There is an opposite and counter-intuitive relation with the& values. The computed k for
16         PM25 is higher at night than during the day (0.39/hour vs 0.27/hour), which is expected since
17         turbulence (which provides upward velocity components to counter Stoke's Law settling) is
18         less when homes are closed and people are resting and sleeping.  Turbulence reduces
19         gravitational settling but also decreases the boundary layer thickness through which PM must
20         diffuse to reach surfaces for deposition which may increase deposition rates for the ultra-fine
21         PM (< 1 //m AD). For PM10 the opposite variation of k is observed. The PM,0 k value
22         decreases from a day time value of 0.91/hour to a nighttime value of 0.43/hour which is
23         counter to the PM25 behavior, and also counter to the deposition increase from day to night
24         during passage of PM10 into the indoor environments.  There are several possible
25         explanations for these phenomena:
26      a)  There are experimental errors in all measured values, such as air exchange rate fy) and
27         concentrations (Co and C).  Such experimental errors can inadvertently cause artifacts to
28         appear like values ofP > 1 as cited earlier (Ozkaynak et al., 1996a). The air exchange rate
29         data set {a} contains two artifacts as previously discussed. The minimum detectable level
30         (MDL) of perfluorocarbon tracer (PFT) mass collected corresponds to a maximum detectable
31         air exchange rate because an increase in air exchange rate lowers the amount of tracer

        October 1999                              5-34        DRAFT-DO NOT QUOTE OR CITE

-------
 1         collected. The 
-------
 1         that indoor human activity at night is usually much less than daytime activity. In the absence
 2         of the influence of human activity the indoor value ofk only has a constant expectation if the
 3         relative size distribution, the proportion of PMj5 to (PM10 - PM25), is constant. In PTEAM
 4         the day time PM2 5 and (PM10 - PM2 5) were approximately equal and at night there was
 5         approximately 50% more PM25 than (PM10 - PM25) (Ozkaynak et al, 1996a).
 6      e)  Penetration (P) and decay rate (k) may both depend on air exchange rate (a) and, thus, may be
 7         correlated to each other.  This may lead to unstable or inaccurate analyses.
 8      f)  The results are real and there is some mechanism operating that is not completely understood
 9         causing the opposite behavior of the k values for fine PM (PM25) and coarse PM
10         (PM10-PM25).
11           An alternative procedure to estimate mean values, and to conserve the property that mean
12      24-h values must be approximately intermediate between their constituent 12-h day and 12-h
13      night values, would be to use only the complete data set that constitutes a nominal 24-h average
14      by discarding all daytime sets without a matching nighttime set andvice versa. Then the 24-h
15      mean value of indoor and outdoor concentrations could be used with the 24-h mean value of air
16      exchange and reported in two ways:  1) from averaging the mass of PFT collected using only
17      those data where it was greater than the MDL for both night and day; 2) from averaging the mass
18      of PFT collected using all data, but giving lower weights to the default values ofa computed
19      with the arbitrarily set MDL PFT values. The unknowns in the equation would be the
20      corresponding values off, k and emission strengths (e.g., Q,ther) averaged over 24-hrs. Such
21      procedures should then give mean values ofP, k, and emission strengths approximately
22      intermediate to their day and night values.
23           Abt et al. (1999b) also model their PM size and number density data (Abt et al., 1999a)
24      from four homes in the Boston area using the same Koutrakis et al. (1992) model described
25      above, as used in the PTEAM study. The authors define the "effective penetration efficiency of
26      outdoor air" for PM of various size ranges by the termP a /(a + k) which represents the fraction
27      of the outdoor PM found indoors at equilibrium. They report effective penetration efficiencies
28      ranged from 0.38 to 0.94 for 0.02 - 0.5//m particles with a maximum between 0.1 and 0.2
29      microns.
30           For 0.7 - 10 //m particles the efficiency values ranged from 0.53 to 0.12, decreasing with
31      increasing particle size.  Whereas the PTEAM study (Ozkaynak et al., 1996a) only reported mean

        October 1999                             5-36         DRAFT-DO NOT QUOTE OR CITE

-------
 1     values of penetration (P) and deposition (k) values for their 178 subjects, these Abt et al. (1999b)
 2     data indicate that "estimated decay rates varied considerably both within and between homes,
 3     with variability attributed to factors including differences in air flow rates, house volumes and
 4     surface materials." This suggests that the variances of the distributions off* a / (a + K) shown in
 5     Figure 5-4a and 5-4b, are underestimated because they were derived using the  two PTEAM mean
 6     values of k as a constant.
 7
 8     5.7.1  Estimation of the Daily Exposure to PM of Ambient Origin
 9          Let each person spend a fraction x of their time outdoors during the day and a fraction w
10     outdoors during the nighttime periods during which PM is monitored [y~ (x +  w)/2].  While
1 1     outdoors, people are exposed to 100% of the ambient concentration and while  indoors, they are
12     exposed to a lower fraction of the ambient PM as estimated by setting the derivative equal to zero
13     in Equation 5-2 that resulted in Equation 5-3.
14          The resulting equation for the fraction (z) of the daily ambient PM one is exposed to,
15     assuming 12-h daytime and  12-h nighttime sampling, is as follows:
16
         _ [x + (1 - x)/3 a I (a + k)](Co)day + [w + (1 - w)P a I (a + k)](Co) night + F(M)
        Z~                          (Co)day  + (Co)night                            (5"n)

17
18     where the function of deposited mass of ambient PM that is resuspended [F^/)] is > 0. Because
19     of the difficulty in computingMby solving Equation 5-1 1, as discussed previously, the
20     assumption that [F(M)J is equal to zero for all time  t > 0 allows a rewriting of Equation 5-11 as
21     an inequality:
22
        Z ~                         (Co) day + (Co) night                           (5"12)
23
24          The estimation of z can be made using parameters from the PTEAM data set shown in
25     Table 5-1 and the fractions of time spent outdoors (x and w) shown in Table 5-2.  In this analysis
       October 1999                             5-37        DRAFT-DO NOT QUOTE OR CITE

-------
            TABLE 5-2. AVERAGE FRACTIONS OF TIME SPENT OUTDOORS OR IN A
          VEHICLE IN THE PTEAM(DAY) STUDY AND NHAPS STUDY (NIGHT) [PTEAM
                              NIGHTTIME DATA NOT AVAILABLE]
         Location Outdoors          Daytime Fraction (8am-8pm)  Nighttime Fraction (8pm-8am)
         Outdoors at home location              0.040                      (See below)
         Outdoors at other location              0.087                  0.020 (Both locations)
         Outdoors in vehicle                    0.092                         0.032
         Total Outdoor Fraction	x= 0.219 (PTEAM)	w = 0.052 (NHAPS)
         * Sources of data: Ozkaynak et al. (1996a) and Klepeis et al. (1999).
 1     it is assumed that the ambient PM (away from traffic) is relatively uniform throughout the
 2     community. The air exchange rate while driving a closed vehicle is very large (> 13/hour at
 3     20 mph) so the subject in a vehicle would be exposed to-100% of the ambient PM measured at
 4     the central site plus the PM generated by the surrounding traffic (Ott et al., 1992; Park et al.,
 5     1998).
 6          A recent study (California Environmental Protection Agency,  1998) reports that PlVf 5 in
 7     motor vehicles is intermediate between the roadside concentration and the concentration
 8     measured immediately outside the vehicle. The inside/outside ratio for the vehicles in
 9     commuting traffic was approximately 2/3.  Note that the locally generated PM from the traffic
10     surrounding the vehicle does not directly influence the monitoring station measurement  (Ca) or
11     the outside air at the home (Co). Aim et al. (1999) monitored the particle count (by Climet-500
12     laser particle counter) in a commute vehicle in Kuopio, Finland and compared the results with
13     the background values, estimated as the mean count at the start and  finish of the trip at an
14     off-road location. The authors found that the excess  (vehicle - background) counts of fine PM
15     (optical equivalent diameter < 1 micron OD) increased as the average  vehicle and wind speeds
16     decreased, and that the excess counts of coarser PM (> 1 micron OD)  increased as wind speed
17     and vehicle speed increased.
18          Substituting the corresponding parameters from Tables 5-1 and 5-2 in Equation 5-12, the
19     mean values for the daily total fraction of the ambient PM that people are exposed to (z) are
20     estimated to be z >0.75 for PM2 5 and z > 0.64 for PM10. The bounds for the PTEAM study can
       October 1999                             5-38        DRAFT-DO NOT QUOTE OR CITE

-------
 1     be estimated from the daytime minimum (0) and maximum (0.976) values of x reported in
 2     PTEAM, and assuming that the nighttime minimum exposed subject is indoors (w = 0) and the
 3     maximum exposed subject at night is outdoors or in a vehicle for 4 of the 12 hours (w = 0.33).
 4     The estimates of the population range then becomes as follows:
 5                For PM25: z maximum  >0.88;  z mean >0.75; z minimum >0.72
 6                For PM10: z maximum >0.87;  z mean >0.64; z minimum >0.58
 7          The mean values, the caveats cited above not withstanding, are consistent with the PTEAM
 8     findings that approximately 3/4 of the indoor PM, 5 and 2/3 of the indoor PM10 were of ambient
 9     origin (Ozkaynak et al., 1996a). The minimum and maximum values are conservative because
10     identical daytime and nighttime values of Cout,a, k and P are implicitly assumed for all homes
11     in the study. A Monte Carlo analysis, allowing all parameters to vary randomly about their mean
12     values, would increase the estimated variance of z so the maximum would increase and the
13     minimum decrease.
14          Abbey et al.  (1999) evaluated the effect of time spent indoors as a surrogate for exposure to
15     PM from indoor sources.  They compared the relative risk (RR) of mortality by nonmalignant
16     respiratory disease with ambient PM;0 concentration, for different amounts of time spent
17     outdoors, in the 7th Day Adventist Health Study on Smog (ASHMOG). They assumed that the
18     indoor concentration of PM10 of ambient origin was 70% of the outdoor PM,0 concentration to
19     allow for a protective effect of staying indoors. They found that RR for a 50//g/m3 increase of
20     ambient PM10 increased when the number of hours spent outdoors increased (and the number of
21     hours indoors decreased) during the week, as shown in Table 5-3.  Thus, the authors reasoned
22     that increased exposure to indoor generated PM and decreased exposure to PM of ambient origin
23     did not appear to be a confounder in their analysis of the effects of ambient PM.  This increasing
24     RR trend with time outdoors is consistent with a higher exposure to particles of ambient origin
25     with more time outdoors.
26          Further in regard to time spent outdoors, Abbey et al. (1999) explain their observed gender
27     difference in health effects in which males appear to be more affected by ambient PM than
28     females, as being related to the males' greater percentage of time outdoors than females of
29     similar ages.  However, Brunekreef (1999) points out that this is not a likely explanation: "As the
30     authors (Abbey et al., 1999) note, fine particles readily penetrate indoors, and if it is the fine
31     particles that matter, small differences in time spent outdoors cannot matter all that much."

       October 1999                             5-39       DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 5-3. RELATIVE RISK (RR) OF NONMALIGNANT RESPIRATORY
        	MORTALITY FROM INCREASING TIME SPENT OUTDOORS	
        Hours outdoors per week    RR per increase of 50 //g/m3 PM10    95% Confidence Interval
        t<4                                   1.07                       0.85-1.34
        416                                 1.32                       1.02-1.68
        Source: Abbey et al. (1999).
 1          The presence of appreciable amounts of PM generated indoors would add to the mass of the
 2     ambient PM collected by a personal PM monitor and cause the high correlations of personal
 3     exposure to PM of ambient origin with ambient PM concentrations to degrade and approach zero
 4     (U.S. Environmental Protection Agency, 1996; Monn et al., 1997; Monn and Junker, 1999). For
 5     example, Monn et al. (1997), (with clarification by personal communication [Monn,  1999]),
 6     reported that human activity increased the median Indoor/Outdoor ratio by 20% for PM5 and
 7     50% for PM10. As a result, some authors (e.g., Gamble, 1998) have misinterpreted an absence of
 8     a significant correlation of ambient PM concentration with personal exposure tototal PM as
 9     implying an absence of a significant correlation between ambient PM concentration and personal
10     exposure to PM of ambient origin.  [See Kiinzli and Tager (1999) comments on Gamble (1998),
11     and the response by Gamble (1999).]
12          All people in a community, when outdoors, are exposed to a heterogeneous mixture of
13     ambient PM2 5 with small variations of composition and concentration from the ambient PlVf 5
14     measured at a central location. Therefore, no great ecological fallacy is produced and no large
15     Berkson-type error is involved, in the use of community ambient PM.5 concentration as a
16     surrogate for exposure to particles of ambient origin.  This is because the mean exposure of a
17     random sample of people in a community to PM, 5 of ambient origin, while both outdoors and
18     indoors (via PM25 infiltration from outdoors), is an excellent predictor for the simultaneous
19     exposure to PM25 of ambient origin of virtually everyone in that community - exceptions for
20     unusual circumstances such  as home use of high performance air cleaners not withstanding.
21     In stark contrast, the mean of the personal exposure to PM,5 or PM10 of non-ambient  origin

       October 1999                            5-40         DRAFT-DO NOT QUOTE OR CITE

-------
 1      (from residential and occupational emissions) of the same randomly chosen subjects would have
 2      virtually zero predictive power for estimating the exposure to PM^ or PM10 of non-ambient
 3      origin of any person in the community not residing in the home of a sampled person.
 4           Exposure to an arbitrary PM fraction (PM,,) from non-ambient sources is not of relevance to
 5      the question of personal exposure to PM,, of ambient origin, and an ambient PM^, concentration is
 6      not expected to be a surrogate for any arbitrary person's total exposure to P]V| as measured by a
 7      personal PMX monitor that collects PIV^ of both ambient and non-ambient origin. In the absence
 8      of appreciable indoor sources of PM^, a personal exposure to total PM^ while indoors is primarily
 9      to the PMX of ambient origin that infiltrates indoors. Under such conditions a measurement of
10      personal exposures to PIV^ that is mostly of ambient origin is expected to be highly correlated
11      with ambient PMX concentration as described in later sections.
12           People in a community are routinely exposed to widely different mixtures of PM of
13      non-ambient origin with a large variance of composition and concentration. This wide variance
14      arises from each individual's unique combination of occupation, personal habits (e.g., cigarette
15      smoking), social contacts (e.g., living with a smoker) and non-occupational/residential activities
16      (e.g., cooking, cleaning  and dusting). A subject's exposure to PM of ambient origin and its
17      chemical composition can be estimated from knowledge of the ambient PM concentration, its
18      composition, the air exchange rate between ambient and indoor locations where the subject
19      spends time, the deposition rate of the PM and the subject's time-activity patterns.  In contrast,
20      a subject's exposure to PM of non-ambient origin and its composition cannot be predicted
21      accurately from measurements of other peoples' exposures to non-ambient PM concentration and
22      its composition. Knowing the distributions of ambient PM concentrations (uninfluenced by an
23      immediate source such as a barbecue) in all other backyardsdoes allow the accurate prediction of
24      the ambient PM concentration in the subject's own backyard; knowing the distribution of
25      non-ambient PM concentrations in all other homes does not allow the accurate prediction of the
26      non-ambient PM concentration in the subject's own home.
27           When people are members of a cohort whose individual health outcomes are hypothesized
28      as due to ambient PM exposure (and other possibly co-occurring), and they are being tracked
29      from day to day, it is important to determine each individual's daily total personal exposure to
30      PM of ambient origin (Beyea and Hatch, 1999). However, additional information is needed for
31      such epidemiologic studies to be able to separate the total PM exposure, if measured, into its two

        October 1999                              5-41         DRAFT-DO NOT QUOTE OR CITE

-------
 1      general components of exposure to ambient PM and the exposure to non-ambient PM (e.g., from
 2      residential or occupational sources).  With this information it will be possible to improve the
 3      estimation of the health outcomes related to ambient PM exposure and separate them from those
 4      related to non-ambient PM exposure.
 5           This chapter examines ambient PM air quality and that portion of ambient PM which
 6      penetrates into indoor microenvironments.  It also examines, to a lesser extent, the contribution
 7      of sources of non-ambient PM to total PM exposure. This is to aid in interpretation of acute and
 8      chronic epidemiology studies assessed in Chapter 6, in which ambient PM concentrations are
 9      assumed to be an indicator or a surrogate for mean community exposure to PM of ambient origin
10      or an individual's exposure to ambient PM. Thus, this chapter has three objectives:
11      (a)  Provide a review of pertinent studies of personal exposures to total PM of various size
12          fractions.
13      (b)  Evaluate linkages of human exposures to PM of ambient origin estimated from
14          concentrations of PM measured at a fixed-site monitor located at some central  site in a
15          community under study.
16      (c)  Quantify the contributions of PM of ambient origin and non-ambient origin to total personal
17          PM exposure.
18           At the present time, little published data are available for time scales of less than 12-h to
19      compare short-term personal PM exposures, short-term peak ambient PM concentrations and
20      short-term indoor PM concentrations found in different locations within a community, all for the
21      same time interval. It has been hypothesized (Michaels, 1997, 1998) that some health effects
22      may be better correlated with short-term 1-h peak PM concentrations than the 12-h  or 24-h
23      concentration averages containing the peak value. This has been supported by Delfino et al.
24      (1998) with a finding of "1-h and 8-h maximum PM,0 having larger effects than the  24-h mean"
25      in relating asthma symptoms. Hourly PM data and even 1/2-h PM data (Keary et al., 1998)
26      obtained by use of the TEOM® sampler (see Chapter 4 and Ayers et al. [1999] for a  description
27      of loss of semivolatile components) are beginning to become available in the literature.  Such
28      data may allow for more study of health in relation to the lagged effect of peak ambient PM
29      concentrations.
30           The PM-mortality literature, to date, is based upon the general linearized model (GLM)
31      assumption that there is a virtually linear relationship between health effects and ambient PM

        October 1999                              5-42       DRAFT-DO NOT QUOTE OR CITE

-------
 1      exposure characterized by ambient PM concentration in the concentration range below the
 2      existing PM10 NAAQS (Liang and Zeger, 1986). The relative risk has generally been modeled as
 3      a proportional or percentage increase in non-trauma mortality per unit increase in ambient PM
 4      concentration, such as 0.1% per 1 //g/m3 of PM2 5.  For a strictly linear relation with PM of
 5      ambient origin, going from 10//g/m3 of PM25 to 110 //g/m3 of PM25 would increase the base
 6      mortality rate by 10%.  However, the actual 'compounding interest' effect of an increased rate of
 7      0.1% with each additional 1 //g/m3 of PM2 5 would only  increase the rate by (1.001}°° = 10.5%
 8      which closely approximates the linear increase. In the realistic range of ambient PM
 9      concentrations, the linear approximation is generally acceptable.
10           By the principal of superposition for a linear system, the health effect from exposure to a
11      mixture of ambient PM and non-ambient PM is the sum of the health effect of the ambient PM
12      exposure plus the health effect of the non-ambient PM exposure.  This assumed linear-relation
13      implies that the health effects of the PM of ambient  origin are independent of the health effects of
14      the PM of non-ambient origin in the current range of ambient and non-ambient PM
15      concentrations in modern society.  This  is complementary with the finding that sources of indoor
16      PM appear to operate independently of the ambient PM concentration.
17           Given this framework of linear analysis, if PM of non-ambient origin produced health
18      effects similar to those produced by ambient PM, the fluctuations in the health effects of PM
19      from non-ambient sources would be independent of the fluctuations in the health effects
20      produced by PM of ambient origin - and therefore would not act as a confounder.  Rather, they
21      would appear as a source of random error in the linear epidemiologic analyses of human health
22      and exposure to PM of ambient origin.  Until nonlinear effects of PM exposures are
23      demonstrated in clinical and epidemiologic studies, the  sources of non-ambient PM, and the
24      effects they produce on total personal exposure to PM, are treated as errors in the estimate of
25      human exposure to  PM of ambient origin. Thus PM of non-ambient origin (e.g., from personal
26      activity and occupational or residential sources) is considered, at the present time, to play a minor
27      role, if any, in the study of human health and its relationship to acute exposure to PM of ambient
28      origin in community time-series studies.
29
30
        October 1999                             5-43        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.8   AMBIENT PARTICULATE MATTER CONCENTRATION AS A
 2            SURROGATE FOR EXPOSURE TO PARTICULATE MATTER OF
 3            AMBIENT ORIGIN
 4          The health effects due to ambient PM may depend upon an individual's genetic makeup,
 5      lung anatomy, and previous health history, and the mass, size and composition of those inhaled
 6      particles that can be deposited within various regions of the respiratory tract.  The amount of this
 7      dose per unit lung area (mg/cnf-day) will depend on the concentration of ambient PM inhaled
 8      (e.g., the instantaneous personal exposure to ambient PM); the ventilation rate and respiration
 9      frequency (a function of physical activity and basal metabolism); and the fractional PM
10      deposition, which is a function of ventilation rate and respiration frequency, mode of breathing
11      (e.g., oral or nasal), and any alteration of normal pulmonary flows due to lung dysfunction. If all
12      people had identical ventilation rates (L/min - kg body weight), respiration frequency, and
13      deposition patterns, then the potential-dosage distribution (mg/crri-day) could be linearly scaled
14      to the personal ambient PM exposure distribution which would serve as a suitable primary
15      surrogate. The usage of ambient PM concentration in health studies as a surrogate for personal
16      exposure to PM of ambient origin, and thereby a secondary surrogate for the ambient PM dosage,
17      would be suitable if ambient PM concentration is linearly related to the personal exposure to PM
18      of ambient origin and the dose-response relationship is linear (Mage, 1983).
19          Adult ventilation rates are lowest (mean- 6 L/min) during the night while asleep, highest
20      (mean -12 L/min; peak ~ 60 L/min) during the day while awake (Adams, 1993), and in phase
21      with ambient PM concentration, which is also usually lower at night than during the day (Clayton
22      et al., 1993). Consequently, the product of the 24-h average ambient PM concentration, the 24-h
23      average ventilation rate, and the average deposition parameter for the average ventilation could
24      seriously under-predict the amount of ambient PM deposited in the respiratory tract (Mage,
25      1980).
26          In practice, when relating human health to ambient PM pollution variables (as in
27      Chapter 6), one is forced to use time-weighted-average (TWA) ambient PM concentration as a
28      surrogate for ambient PM exposure and ambient PM dosage because typically only fragmentary
29      data are available on personal exposures to PM of ambient origin in populations. Data are also
30      limited on ventilation rates as a function of basal metabolism and physical activities (Adams,
31      1993), and on the pulmonary deposition rates of particles people are inhaling. The size

        October  1999                             5-44        DRAFT-DO NOT QUOTE OR CITE

-------
 1      distribution of PM in the ambient air is usually unknown, and its pulmonary deposition is
 2      affected by unmeasured individual physiological parameters.
 3           In the sections that follow, the experimentally observed relationships between ambient PM
 4      concentration, indoor concentrations of ambient PM and non-ambient PM, total personal
 5      exposures to PM, and personal exposures to PM of ambient origin are discussed in detail. The
 6      following four caveats, developed in more detail in Chapter 7 of U.S. Environmental Protection
 7      Agency (1996), should be kept in mind:
 8      1. Ambient PM concentration (times volume inhaled) is a surrogate for the dosage of ambient
 9        PM inhaled and deposited in peoples' respiratory tracts.
10      2. The daily dosage of total ambient PM deposited per unit surface area of the sensitive
11        portion(s) of the respiratory tract is in turn a surrogate for the mass of the true (but unknown)
12        species and/or size fraction of the ambient PM that is the specific aetiologic toxic agent(s)
13        that act by a presently unknown mechanism.
14      3. Virtually all analyses and discussions of personal PM exposure presented here are based on
15        personal exposure to PM of self-declared non-smokers.
16      4. A total TWA personal exposure to total PM (ambient PM plus non-ambient PM) is expected
17        to be a poor surrogate for the personal exposure to PM of ambient origin.  This will be
18        particularly true for those people whose personal exposures to total PM are dominated by
19        residential and occupational indoor sources, or personal sources such as a hobby activity or
20        active smoking of tobacco.
21
22
23      5.9    CONCENTRATIONS OF AMBIENT PM FOUND INDOORS AND IN
24            OTHER NON-AMBIENT ENVIRONMENTS
25           In the absence of appreciable indoor sources, there is an excellent correlation between
26      ambient PM concentration (ambient PM is virtually all of ambient origin) and the concentration
27      of PM found indoors.  Figure 5-5 (Tamura et al., 1996a; U.S. Environmental Protection Agency,
28      1996) shows how the indoor PM,0 correlated with the outdoor PM,0 for a set of seven elderly
29      non-smoke exposed individuals living in traditional Japanese homes.  These Tokyo homes
30      (Itabashi ward), where people routinely took their shoes off prior to entering, hadtatami reed mat
31      or carpeting on tatami or wooden flooring, and had gas for cooking.  The study was designed to

        October 1999                              5-45        DRAFT-DO NOT QUOTE OR CITE

-------
110-
100-
E 80-
'§> 70-
tT 60-
8 50-
-o 40-
- 30-
20-
10-

E

m a
DO™
ffl
D On
mm
ffl° a ffl
Dft1^1 ffl Annual r = 0.922
™ffl Winter r = 0.920
Summer r = 0.961

                    40
                           80
                                  120
                                         160
                                                200
110-
100-
~ 90-
-S 80-
Cf) ^^
3; 70~
ri 60 —
o 50-
E 40-
30-
20-
10-

F


D Q
r*
B BO • °
n DB B
j^j1 Annual r = 0.897
If Winter r = 0.702
Summer r = 0.970
i i i i i i i i i
                    40
                           80
                                  120
                                         160
                                                200
110-
100-
"E so-
"ro 70-
— 60-
o 50-
o
? "-
— 30-
20-
10-
G
a
ffl m
D a
m OB D
gm

,rip |p ffl Annual r = 0.898
0 nyffl Winter r = 0.879
Summer r = 0.919
                                         160
                                                200
                          Outdoor (M9/m
                                                      120-
                                                      110-
                                                      100-
                                                      90-
                                                     r so-
   5 60-

   8 40-
   _E 30-
     20-
     10-
      0
                                                                             Annual   r = 0.983
                                                                             Winter   r = 0.980
                                                                             Summer  r = 0.982
                                                               40
                                                                             120
                                                                                    160
                                                               40
                                                                      80
                                                                             120
                                                                                    160
                                                               40
                                                                      80
                                                                             120
                                                                                    160
                                                                      Outdoor (jjg/m3
                                                                                           200
110-
100-
JT 90~
"°E 80-
D) 70-
•^- cn
b(J—
o 50-
E 40-
- 30-
20-
10-
B



B ffl
0° ^aSa°
a gg& Annual r= 0.929
^nSpB" Wnter r= 0.966
Summer r = 0.877
110-
100-
r~>~^
E 80-
ra 70-
•^ 60-
0 50-
? 40-
- 30-
20-
10-
0-
C

ffl
Qa_BBa
Cffl *
D qgff^ Annual r= 0.970
a Wnter r = 0.968
Summer r = 0.964
i i i i i i i i i
                                                                                           200
110-
100-
pT~ 90~
E 80-
ra 70-
— 60—
0 5Q_
o
-0 40-
- 30-
20-
10-

D
ffl ffl
m ffl
ffl
ffl ffl D
en a
tSb
_ ^§Q ffl
gji* Annual r= 0.838
D D Wnter r= 0.775
Summer r = 0.978
I I I I I I I I I
                                                                                           200
      Figure 5-5.  Individual indoor versus outdoor relationships of PM10 in Tokyo for the seven
                   subjects (A-G) reported on by Tamura et al. (1996a). El Winter, D Summer
1      monitor the exposures of people to ambient PM so the subjects were purposefully chosen in a

2      non-random manner to eliminate indoor combustion sources of PM. Consequently, these results

3      apply strictly to these seven people and cannot be used to infer a similar relationship in other
       October 1999
5-46
DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
Tokyo homes. Table 5-4 provides the correlation coefficients reported and the designations
A - G are the identifiers of the subjects with the indoor/outdoor relationships shown in
Figure 5-5.
            TABLE 5-4. SUMMARY OF CORRELATIONS BETWEEN PM in PERSONAL
                                                                        10
            EXPOSURES OF SEVEN TOKYO RESIDENTS AND THE PM in MEASURED
                                                                       10
         OUTDOORS UNDER THE EAVES OF THEIR HOMES, AND THE PM MEASURED
                          AT THE ITABASHI MONITORING STATION
Subject ID
A
B
C
D
E
F
G
A-G
Number of Samples
48-h PM10
9
9
11
9
10
7
9
64
Correlation between Correlation between Personal
Personal and Outdoor at and Itabashi Station (r)
home (r)
0.958
0.874
0.846
0.922
0.960
0.776
0.961
0.834
0.876
0.747
0.848
0.964
0.925
0.801
0.952
0.830
        Source: Tamura et al. (1996a).
 1          Tamura et al. (1996a) did not report duplicate measurements of the indoor/outdoor
 2     monitors so it is not possible to correct these data for the variance component due to
 3     experimental errors of filter weighing and flowrate measurement.  Such errors prevent two
 4     collocated measurements of PM from approaching a perfect correlation of r = 1, and are expected
 5     to decrease correlations, such as those shown on Figure 5-5.  These data show that in the absence
 6     of major sources of indoor PM,0 generation there is a high correlation relationship between
 7     indoor and outdoor PM10.
 8          Tamura et al. (1996b) performed another study in Osaka, Japan similar in design to
 9     their Tokyo study (Tamura et al., 1996a). The  authors measured indoor and outdoor PM
10     simultaneously at 26 homes (not chosen randomly) during the autumn seasons from 1990
       October 1999
                                        5-47
DRAFT-DO NOT QUOTE OR CITE

-------
 1     through 1995. A dichotomous sampler was used with cut points at 2 and 10//m AD, and a
 2     greased impactor plate to collect PM > 10//m AD.  There were 146 days when indoor PM and
 3     outdoor PM concentrations were successfully collected simultaneously in all three size ranges.
 4     The authors pooled all 146 pairs of observations and reported the group correlation coefficient
 5     between PM10 concentration measured immediately outdoors at the eaves of the home and the
 6     concentration of PM10 in the home. The regression line had a slope of 0.645, an intercept of
 7     9.5 //g/m3, and a correlation coefficient r = 0.865. It is interesting to note that when the authors
 8     pooled the data for only the 24 homes where 4 or more samples were collected (total number of
 9     data points not given) the correlation of r = 0.958 was almost identical to the summer correlation
10     (Tamura et al., 1996a) of r = 0.950. The  authors concluded "In the present study [Tamura et al.,
11     1996b], similar relationships [to Tamura et al., 1996a] were confirmed with 18 houses even
12     [though] the measurement period was limited in season and number of times [sampled].
13           Anuszewski et al. (1998) measured simultaneous indoor and outdoor hourly PM by light
14     scattering using a portable nephelometer for 18 days at nine homes of non-smokers in Seattle,
15     WA. Although light scattering does not provide a quantitative measure of concentration, the
16     light scattering occurs from particles of optical diameters (not aerodynamic diameters)
17     approximately centered about the wave length of the light source of the instrument. Therefore,
18     the particles that scatter light using this instrument are primarily in the accumulation mode
19     (0.2 < AD < 1 //m). An example was given in the article of how sweeping a patio (which raises
20     primarily coarse PM) next to a kitchen only caused 25% of the increase in light scattering
21     measured in the kitchen created by smoking a cigarette on the same patio.
22           The authors report that the mean indoor to outdoor ratio of hourly light scattering values
23     was 0.98, and that correlations of indoor  with outdoor were in the range 0.58 < Pv < 0.99. The
24     minimum of 0.58 occurred in a home with an electrostatic precipitator (Figure 5-6) and the
25     maximum of 0.99 occurred in a home with a standard in-line filter in the heating system
26     (Figure 5-7). Tung et al. (1999) discuss the PM removal by a standard in-line air filter in a
27     heating-ventilation-air conditioning (HVAC) system used in their study. "In the filtering process,
28     particles with diameters less than 0.1 //m were removed by diffusion removal mechanism.
29     Particles with the size greater than 1 //m were removed by interception and impaction."
30     Therefore, operation of the in-line filter in the house B-l may not have made any appreciable
31     impact on the indoor PM concentration in the range of PM optical diameters (0.2//m < AD < 1)

       October 1999                             5-48        DRAFT-DO NOT QUOTE OR  CITE

-------
0.7 -I
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
  0
                   House B-2
                                                                    R2 = 0.58
                                                                    Slope = 0.92
                                                                    Intercept = 0.003
                  0
              0.1
0.2
0.3
Out
0.4
0.5
0.6
       Figure 5-6.  Comparison of light scattering coefficient (km J) by PM indoors and outdoors
                   at a home with an electrostatic precipitator in operation.
       Source: Anuszewski et al. (1998).
 1     measured by the nephelometer. The absence of sources of PM in this size range in the home B-1
 2     would then explain the almost perfect correlation between the indoor and outdoor PM shown in
 3     Figure 5-7. They report no source of PM in House B-2 that can explain the higher values indoors
 4     compared with outdoors as opposed to House B-l where indoor and outdoor were virtually equal.
 5     House B-2 had "Some" activities in the studied room and "Medium" in/out traffic, whereas
 6     House B-l had "Many" activities in the studied room and "Light" in/out traffic.
 7          The average air exchange rates in both houses, B-l and B-2, were equal (1.7/hr), which is
 8     much higher than the average in the U.S. reported by Murray and Burmaster (1995).
 9          The electrostatic charging of PM in the size range monitored by the nephelometer in House
10     B-2, and that PM subsequent precipitation, could explain a portion of the decrease in the
11     indoor/outdoor PM ratio shown in Figure 5-6. The two homes with no in-line filters in the
12     heating system had the highest minimum slopes [1.0, 0.95] and the six homes with in-line filters
13     had intermediate minimum slopes [0.72 - 0.90].  The authors concluded that "For these nine
       October 1999
                                  5-49
                    DRAFT-DO NOT QUOTE OR CITE

-------
               0.6 -I
               0.5 -
               0.4 -
             = 0.3 -
               0.2 -
               0.1 -
                 0
House B-1
                              0.1
                      0.2
0.3
Out
                                                FT = 0.99
                                                Slope = 0.87
                                                Intercept = 0.004
0.4
0.5
0.6
        Figure 5-7.  Comparison of light scattering coefficients (km J) by PM indoor and outdoors
                    at a home with a standard in-line filter in the air recirculating system.
        Source:  Anuszewski et al. (1998).
 1      homes, the dominant source of indoor fine particles, as indicated by [light scattering coefficient],
 2      appears to be the outside air."
 3           Leaderer et al. (1999a) reported on an indoor/outdoor study of PIV(0, PM25 and related
 4      species (sulfate, nitrate, H+, NH4+) at 58 residences in southwest Virginia, measured during the
 5      summer periods from 1995 to 1997. The residents were all non-smoker families participating in
 6      an epidemiologic study of maternal and infant health with respect to indoor air pollution
 7      (Leaderer et al., 1999b). Forty nine of the 58 homes were air conditioned (A/C) and 21 of the
 8      homes used gas for cooking.3
 9           Table 5-5 summarizes these summer data collected in the area surrounding a central
10      monitoring site located at Vinton, VA (6 km east of Roanoke, VA). Because of happenstance,
              3 The residence locations were not chosen randomly to represent all residences in the area surrounding
        Vinton, VA so the standard statistical tests used by the authors may not apply to the sampled mean as representing
        the mean at any other locations in the southwest Virginia region.
        October 1999
                              5-50
         DRAFT-DO NOT QUOTE OR CITE

-------
       TABLE 5-5.  SUMMARY OF PM DATA DURING THE SUMMER IN AN
             INDOOR/OUTDOOR STUDY IN SOUTHWEST VIRGINIA
Pollutant
PM10 Cug/m3)



PM2 5 (,ug/m3)



Coarse Mode (//g/m3)



Sulfate (SO4=, nmol/m3)



Acidity (H+, nmol/m3)



Ammonium (NH4+, nmol/m3)



Nitrate (NO3", nmol/m3)



Sampling Site
Regional
Outside all homes
Inside A/C* homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
Regional
Outside all homes
Inside A/C homes
Inside non-A/C homes
n
47
43
49
8
50
43
49
9
47
42
48
8
45
42
47
9
47
45
49
9
43
45
49
9
42
42
49
9
Mean (Std Dev)
26.0(11.5)
28.0(17.7)
28.9(18.7)
33.3 (14.2)
20.2 ( 9.9)
21.8(14.8)
18.7(13.2)
21.1 ( 7.5)
6.3 ( 2.7)
7.7 ( 6.2)
10.4 ( 8.5)
11.4( 9.7)
88.4(51.6)
83.7(53.7)
47.8 (36.3)
63.0 (37.3)
41.0(28.5)
33.0 (36.9)
12.4(15.3)
16.7 ( 9.4)
124.6 (59.0)
129.4 (87.8)
78.3 (77.2)
96.7 (68.9)
10.2 ( 5.0)
8.0(5.4)
5.5 ( 8.9)
6.8 ( 4.6)
 * A/C = Air conditoned

 Source: Leaderer et al. (1999a).
October 1999
5-51
DRAFT-DO NOT QUOTE OR CITE

-------
 1      some data were missing so the total numbers of samples for each category are not equal. The
 2      main findings of the summer study were that -
 3      (1) there were no significant differences between the mean concentrations of all seven PM
 4         categories outside the homes and the same quantities at the regional site in Vinton, VA
 5         which was at an average distance of 96 km away from the residences.
 6      (2) the cross-sectional correlations of Vinton, VA ambient PM with the outdoor residential PM
 7         on paired days were r = 0.5 for fine mode (PMj5, n = 34) and r = -0.2 for the coarse mode
 8         (PM10 - PM2 5, n = 30).
 9      (3) the indoor mean PM concentrations of the six independently measured quantities listed in
10         Table 5-4 (PM coarse is not an independent measure) were all lower in the A/C homes than
11         in the non A/C homes, which is significant atp = (Yz)6 = 1/64.4
12           Leaderer et al.  (1999a) also reported  on PM sampling during the winter periods of 1995/96,
13      1996/97 and 1997/98, at 20 residences in Connecticut and 223 residences in the southwestern
14      and central Virginia areas, similar to the summer PM sampling study described above. However,
15      no central site PM data were collected during these winter periods in either state. The residents
16      were all non-smoker families participating in an epidemiologic study of respiratory health with
17      respect to indoor air pollution (Leaderer et al., 1999c). Kerosene heaters were used in 74 homes,
18      and 52 of the homes with no kerosene heaters had gas stoves.  Because the subjects were not
19      chosen randomly the home characteristics  and corresponding results of the study may not
20      necessarily apply to any other residences and locations within the sampled area.
21           One important difference between the reported results for the winter and summer periods is
22      that during the summer, fluctuations in ambient PM^ were more driven by the fluctuations of the
23      ambient fine mode PM mass and during the winter the fluctuations in outdoor PMg were more
24      driven by the fluctuations of the outdoor coarse mass.  Table 5-6 shows the regression
25      coefficients (R2) for these comparisons.
26           The authors assumed no generation of indoor sulfates for the homes without kerosene
27      heaters in operation, and no enhanced deposition of  sulfates for the homes with no A/C  during
              4 Because these are not random samples from a larger domain than N = 58, the measured means only
        represent the means at the 58 locations on the 58 days they were sampled, so there is a smaller sampling error than
        those predicted by the SD values in Table 5-4. The analytical measurement errors had an SD of 1.2 Mg/m3 for the
        PM2 5,1.1 Mg/m3 for the PM10, 2 nmol/m3 for the sulfate, and 1.3 nmol/m3 for the H+ ions.
        October 1999                              5-52        DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 5-6. SEASONAL VARIATION OF REGRESSION COEFFICIENTS FOR
                      PM10 WITH FINE AND COARSE MODE FRACTIONS
Season
Summer
Summer
Winter
Outdoor Location
Vinton, VA
Southwest VA
Southwest VA and CT
n
50
45
53
R2, PM10 vs Fine
0.96
0.21
0.07
R2, PM10 vs Coarse
0.40
0.12
0.90
        Source: Leaderer et al. (1999a)
 1     the summer. Then using the assumption that ambient sulfates characterized the fine mode
 2     component of ambient PM, and the measured ratios of sulfate indoors to outdoors, the authors
 3     estimated that -75% of the ambient fine mode PM was found indoors during the summer in the
 4     homes sampled in southwest Virginia, and-70% of the ambient fine mode PM was found
 5     indoors during the winter in the homes sampled in southwest Virginia and Connecticut.
 6
 7
 8     5.10  PERSONAL PM MONITORING STUDIES AND FACTORS THAT
 9           INFLUENCE THEIR ABILITY TO  ESTIMATE RELATIONSHIPS
10           TO EXPOSURE TO PM OF AMBIENT ORIGIN
11          The science aspects of personal exposure monitor (PEM) usage for monitoring exposure to
12     PM were reviewed in Chapter 7 of U.S. Environmental Protection Agency (1996). A PEM
13     strictly measures the total exposure of the person carrying it and it applies to no other person (see
14     the discussion of Figure 5-2 for possible exceptions from actively smoking and wearing of dust
15     masks).  A person stirs up PM instantaneously by the very act of arm waving (Bohne and Cohen,
16     1985; Cohen and Positano, 1986; Wallace et al., 1997) and Thatcher and Layton (1995)
17     demonstrated that the effect of simply walking into a room creates a rapid rise of suspended PM,
18     primarily >10 //m AD. These PM increases are created by the air currents of body movement,
19     particles and fibers dislodged from clothes, and the vibrations and mechanical action of stepping
20     or sitting on fabric surfaces.
21          Because of the PM gradients created by such movements and air currents, there is a
22     microscale variation of PM surrounding a subject and this may contribute to what is known as a
       October 1999                            5-53        DRAFT-DO NOT QUOTE OR CITE

-------
 1      'personal cloud'.  Ozkaynak et al. (1996a) defined 'personal cloud' as "a local increase in
 2      particles due to unknown sources", as evidenced by people in the PTEAM study who stayed
 3      home all day having higher personal PM exposure than the fixed indoor sampler recorded.  This
 4      personal cloud may lead to a much higher personal PEM concentration of a different chemical
 5      composition than that of the PM collected by an area monitor located several meters away from
 6      the subject or in another connecting room in the same indoor environment.  This phenomenon
 7      was recognized by Ogden et al. (1993) who noted that, when comparing static to personal
 8      sampling for occupational cotton dust exposure, "a mean background level of 0.5 mg/m would
 9      correspond to a median personal exposure of about 2.2 mg/mV'
10          McBride et al. (1999) documented the magnitude of the proximity effect by measuring, in a
11      room of a home, the microscale variations in concentrations surrounding local sources of sulfur
12      hexafluoride (SF6), CO, polyaromatic hydrocarbons, and PM generated by burning an incense
13      stick and by human activity of two-or-more people walking about the carpeted room [by Met-
14      One laser particle counter]. Ratios of particle counts at distances of 1.0 m and 5.7 m from the
15      sources of combustion and the center of activity were obtained at a constant height of 1.53 m
16      above floor level. When only walking was performed, the ratio of closer mean to further mean
17      was of order 1.5 for PM between 2.5 and 10 microns, and 1 for the PM less than 2.5 microns;
18      When only combustion occurred, and no one was in the room, the ratio was of order 3 for PM
19      less than 5 microns and more than 0.5 microns, and 1.5 for PM larger than 5 microns. They
20      conclude that the proximity effect may help explain the existence of the personal cloud denoted
21      by the difference between a personal monitor and an area monitor.
22          The 1996 PM AQCD Chapter 7 provided extensive coverage of many published studies,
23      available as of 1996, that compared measurements of personal exposures to total PM and its
24      constituents with the simultaneous PM and its constituents measured in the ambient air. It is
25      shown below that these studies can be divided into two mutually exclusive categories:
26          Type  1. Longitudinal or time-series studies:  A group of people are followed
27          simultaneously for a long enough time (to provide statistical power) to determine the
28          correlation in time of each person's personal exposure to PM with ambient PM
29          concentration.
30          Type 2. Cross-sectional studies: A group of people are followed sequentially for a period
31          of time, with each person's personal PM exposure sampled for only a few days during that
        October 1999                             5-54        DRAFT-DO NOT QUOTE OR CITE

-------
 1           period. A single correlation of the group's personal PM exposures with ambient PM
 2           concentration is computed from the subjects' pooled exposure data.
 3           Because of their importance, two key Type-1 studies (Lioy et al., 1990; Tamura et al.,
 4      1996a) that were discussed in U.S. Environmental Protection Agency (1996) are discussed again
 5      in this new light (Mage et al., 1999) along with Janssen et al. (1999b). These three studies are
 6      unique in that each subject in the study cohort was followed for virtually the same chronological
 7      period of time as all the other subjects.  This is in contrast to the Type-2 studies in the literature,
 8      such as PTEAM (Clayton et al., 1993), Spengler et al. (1985), and Pellizzari et al. (1999), in
 9      which different subjects were sampled on different days.
10           The important difference between these two study designs arises from the fact that each
11      subject may have major residential indoor sources of PM that are highly correlated from day to
12      day (the same stove and cooking fuel, the same rugs and furniture as reservoirs for resuspending
13      PM, the same co-occupants of the home with day-to-day similar smoking or non-smoking habits,
14      etc.) Given that these sources may provide a relatively constant increment of PM exposure to the
15      subjects from non-ambient sources, the day-to-day variation of the study subject's total exposure
16      may be driven by the variation of their exposures to PM of ambient origin. This coherence is lost
17      when different subjects, each with different indoor sources, are sampled on different days and
18      then grouped together for the data analysis, as described below.
19           As an example of such a Type-2 study, Pellizzari et al. (1999) monitored personal exposure
20      to PM2 5 of a random sample of several hundred people in Toronto, Canada, either once or twice
21      for 3-day  periods over the year from September, 1995 through August, 1996. The outdoor PlVf 5
22      was monitored simultaneously at the participants residences.  The mean outdoor PM5 during the
23      measurement periods was 24.3 ug/m3 and the mean personal exposure PM^ was 67.9 ug/m3.
24      The reported correlation between the logarithms of personal exposure and outdoor concentration
25      is 0.23, statistically significant at a level of 0.01. The authors concluded "These results, while
26      statistically significant, indicate that none of the outdoor concentration data types can adequately
27      predict personal exposure to particulate matter."
28           Lioy et al. (1990) measured 24-h personal PM;0 exposures of 14 non-smoking individuals
29      in Phillipsburg, NJ who were not otherwise exposed to cigarette smoke at home, for
30      14 consecutive days. Phillipsburg, NJ is a small city with a major industrial activity (cast iron
31      pipe production).  Lioy et al. installed four non-personal PIV(0 monitors at outdoor locations

        October 1999                             5-55        DRAFT-DO NOT QUOTE OR CITE

-------
 1      distributed throughout the city to monitor the ambient PM^.  These ambient PM10 monitors,
 2      included the Anderson Hi-Volume sampler and the 10 L/min Harvard impactor. Although they
 3      are designed to collect the same nominal PM,0 fraction as the personal PM10 monitors, different
 4      monitors have slightly different penetration curves and slightly different cut-points than the
 5      personal PM10 monitors. Consequently, it is  expected that this variation would increase the
 6      variance between the PM measured by a collocated personal PM0 monitor and a different type of
 7      ambient PM10 monitor.
 8          The matrices of daily personal PM exposure and ambient PM data were reported by U.S.
 9      Environmental Protection Agency (1996).  The matrices contain several estimated values
10      computed by the E-M algorithm to correct for missing data and three obvious outlier values, as
11      described in U.S. Environmental Protection Agency (1996) and by Mage et al. (1999).
12      Figure 5-8  is the average daily personal PM;0 exposure of the 14 subjects vs the average ambient
13      PM concentration from the four ambient monitors.  The regression PC value is 0.91, which
14      assumes that the mean of the ambient PM values is the true mean ambient PM average on a given
15      day. If the three personal exposure  outliers (453, 809, 971//g/m3) replace their E-M estimated
16      values, the regression would have a value of R2 = 0.34 instead of 0.91.  However, the average is
17      an estimate with sampling error, so  an orthogonal regression was also made to find the line that
18      minimizes  the sum of the squares of the prediction errors in both exposure  and ambient
19      concentration. With this framework, 98.4%of the variation of the mean exposure is predicted by
20      the variation of the mean of the ambient concentrations. U.S. Environmental Protection Agency
21      (1996) concluded that if additional subjects with similar non-smoke  exposure lifestyles had been
22      sampled on those 14 days in Phillipsburg, NJ and if additional air quality monitors had been
23      placed in the community to more precisely estimate the mean ambient PM  in the community,
24      then the correlation coefficient between mean personal PM exposure and mean ambient PM
25      would approach a limit of one with  increasing sample size.
26          Tamura et al. (1996a) measured personal PM;0 exposure of a set of seven elderly non-
27      smoke-exposed individuals living in traditional Japanese homes, and this study was discussed in
28      detail within U.S. Environmental Protection Agency (1996).  The study was designed to monitor
29      the exposures of people to ambient PM so they were purposefully chosen to minimize indoor
30      sources of PM. Figure  5-5 shows how well the indoor PIV(0 correlated with the outdoor PMj0 for
31      these seven homes, indicating the minimal effect of indoor sources of PM.

        October 1999                            5-56        DRAFT-DO NOT QUOTE OR CITE

-------
                        200-
                        150-
Q_
"CD
c
o
co
CD
Q.
CD
cn
CD
CD
>
                        100-
                         50-
                                                               y = 42 + 0.55x
                                        50           100          150
                                           Average Site PM10 (|jg/m3)
                                                        200
       Figure 5-8.  Plot of relationship between average personal PM10 exposure versus ambient
                   PM10 monitoring data from Phillipsburg, NJ, and regression line calculated by
                   U.S. EPA.
       Source: Lioy etal. (1990).
 1          Each subject carried the personal exposure monitor (PEM) for 48-hours, for up to
 2     11 periods total over four seasons of the year 1992. Some data were missing due to random
 3     equipment failure, and the collected data set was screened to remove any observations which
 4     were contaminated by identifiable indoor sources of PM, such as cigarette smoking of visitors, or
 5     operation of an indoor combustion source such as a mosquito coil or an incense burner. The
 6     ambient data, the remaining exposure data, and the E-M algorithm used to estimate missing
 7     exposure data were reported in U.S. Environmental Protection Agency (1996). The individual
 8     personal exposure correlations with the ambient PM ranged from 0.77 to 0.96. Figure 5-9 shows
 9     the completed means of the eleven 48-hr average values versus the simultaneous 48-hr local
10     Itabashi monitoring station values. The reported statistical analysis in U.S. Environmental
11
       October 1999
                           5-57
DRAFT-DO NOT QUOTE OR CITE

-------
              80
          "/—
           E  60
              40-
           CD
           Q_
           0)
           D)
           CD
              20-
               0
                  0
                                                                  y = 11 +0.47x
                           30              60              90
                                   Ambient PM10 (|jg/m3)
                         120
      Figure 5-9.  Plot of 48-h average personal PM10 exposure and ambient PM10 data from
                  Japan—linear regression.
      Source:  U.S. Environmental Protection Agency reanalyses of data from Tamura et al. (1996a).
1
2
3
4
5
6
7
Protection Agency (1996) led to the fitted line that has an intercept of 11.3//g/m3, a slope of
0.47, and a regression R2 value of 0.91.
     Tamura et al. (1996b) performed another study in Osaka, Japan similar in design to their
Tokyo study (Tamura et al., 1996a).  More than 24 subjects, all housewife non-smokers who
were not exposed to smokers living at home, were sampled in a non-random manner during the
autumn seasons between  1990 and 1995.  A dichotomous sampler was used with cut points at
2 and 10 //m AD, and a greased impactor plate to collect PM >  10//m AD. There were 77 days
when outdoor PM concentrations and personal exposures were  successfully collected
      October 1999
                                        5-58
DRAFT-DO NOT QUOTE OR CITE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
simultaneously in all three size ranges.  The authors pooled all 77 pairs of observations and
reported the group correlation coefficient between PM measured immediately outdoors at the
eaves of the home and the personal exposure to PM of the housewife resident.  The personal
versus outdoor correlation results were  for PM,, r = 0.741; for PM10, r = 0.672; for coarse mode
PM10_2 r = 0.343; and for the PM >  10//m AD, r = 0.05.  Because these data do not constitute a
random sample, no statistical inferences can be made to other households.
     Janssen (1998) reports the work of Janssen et al. (1995,  1997, 1998a,b, 1999b,c) on a study
of personal monitoring of children and adults in the Netherlands. Janssen (1998) found that in
longitudinal studies "Personal PMj0 exposures of both adults (aged 50 to 70 years) and
children(aged 10 to 12 years) were reasonably well correlated in time with ambient PMg
concentrations. Personal fine particle (FP) exposures were highly correlated with ambient FP
concentrations" as shown in (Table 5-7). These cohorts were  not random samples from a defined
population so these results apply strictly to only the adults and children sampled.
          TABLE 5-7. AVERAGE LEVELS OF PERSONAL EXPOSURES AND OUTDOOR
             CONCENTRATIONS AND THE CORRELATION ( r) BETWEEN THEM IN
                              LONGITUDINAL EXPOSURE STUDIES
Population
All subjects
Adults
Children
Children
Non-ETS
exposed
Adults
Children
Children
Size
fraction
-
PM10
PM10
PM25
-

PM10
PM10
PM25
Number of
subjects
-
37
45
13
-

23
25
9
Mean*
personal
Mg/rn3
-
62
105
28
-

51
89
23
Mean*
ambient
-
42
39
17
-

41
40
18
Median
individual
correlation r
-
0.50
0.63
0.86
-

0.71
0.73
0.92
Cross-sectional
correlation r**
-
0.34
0.28
0.41
-

0.50
0.49
0.84
        *Mean of individual averages.
       **Mean value. Estimated by randomly selecting one measurement per subject, 1000 times.
       Source of Data: Janssen (1998).
       October 1999
                                         5-59
DRAFT-DO NOT QUOTE OR CITE

-------
 1           These data show that by excluding subjects with environmental tobacco smoke (ETS)
 2      exposure, correlation coefficients increased in all cases. For PM. 5 the median correlation was
 3      r = 0.75 for all subjects, and r = 0.84 for all non-ETS exposed subjects. This indicates that, in
 4      these individual longitudinal time-series cases, the personal exposure to fine PM is highly
 5      correlated with the concentration of ambient fine PM. For comparison, Janssen (1998) obtained
 6      cross-sectional correlation coefficients by a Monte Carlo analysis.  The values shown in the last
 7      column of Table 5-7 were obtained for each cohort by taking one random sample from each of
 8      the individual subject data sets (with replacement), and repeating the procedure a total of 1000
 9      times.
10           Janssen et al. (1999b) report on a personal monitoring study of 13 non-randomly chosen
11      school children, ages 10 to  12 years, in the rural town of Wageningen, the Netherlands.  These
12      children were sampled on one school day per week for a total of eight weeks. Nine of these
13      children lived with parents and relatives who  did not smoke tobacco in any form, and their data
14      were chosen for analysis. Each of the nine children had five, six or seven 24-hour observations
15      of personal exposure to PM25 within the eight days monitored. The ambient PM, 5 was collected
16      at a central monitoring site on these same days for comparison using an identical monitor.
17      Because there were widely different numbers of children with data on any given day in this time
18      series (1 to 9), the missing data for each child was estimated using the E-M algorithm, and the
19      results are shown in Table 5-8 along with their mean exposure and corresponding ambient
20      concentration.
21           Figure 5-10 shows the relation of the estimated mean total PM of the children as a function
22      of the simultaneous ambient PM^ measured at the central station. The regression R2 of the PM25
23      daily mean exposures with the daily ambient PM is 97.8%, and the intercept is  12//g/m3.  This
24      analysis of children who were constrained by design to have the same time-activities at school
25      and at home, where there were no major indoor sources of PIV|5 that could vary from day to day,
26      shows that in the absence of large variability of non-ambient sources of PIV|5, variations in total
27      personal exposure to PM25 is virtually controlled by the variations of ambient PM, 5.
28           Janssen et al. (1998a) report on a study of non-randomly chosen adult's exposure to PMg in
29      the Netherlands during the two winter periods of 1994.  A total of 37 nonsmoking adults
30      (18o*, 19?), living with nonsmokers, average  age 64-y, were monitored with personal PM,,
31      monitors with simultaneous monitoring of the local ambient PM0.  One subset of 13 adults was

        October 1999                             5-60         DRAFT-DO NOT QUOTE OR CITE

-------
          TABLE 5-8. PERSONAL EXPOSURES TO PARTICULATE MATTER (PM 2 5) OF
         NON-SMOKE EXPOSED CHILDREN IN WAGENINGEN, NL, AND STATIONARY
           AMBIENT MONITORING DATA (SAM).  MISSING VALUES ESTIMATED (  )
Day

1
2
3
4
5
6
7
8
Person ID
01
19.8
29.5
15.3
(27.1)
(41.2)
29.3
34.1
(51.3)
03
(8.0)
15.9
11.2
(16.6)
20.2
26.0
29.3
53.7
05
12.5
17.1
28.3
(21.1)
15.8
30.8
35.9
(41.1)
08
13.5
15.0
22.9
20.5
19.6
21.0
26.7
(30.7)
09
14.6
20.7
(40.9)
(29.4)
19.5
24.4
30.7
(35.5)
11
17.2
17.0
(29.2)
(25.9)
19.7
28.7
42.1
(45.2)
13
19.9
13.7
11.9
(18.3)
15.0
22.2
29.4
(49.0)
14
18.3
17.1
(17.0)
(20.0)
15.7
26.7
29.0
34.8
15
13.1
(10.5)
11.2
(15.0)
13.4
(20.1)
26.7
40.7
Mean
Personal

15.21
17.39
20.88
21.54
20.01
25.47
31.54
42.44
SAM

6.24
6.61
12.15
12.51
15.70
20.56
31.67
45.24
       Source of Data: Janssen et al. (1999b).
 1     monitored in the period January through March 1994, and another subset of 24 adults was
 2     monitored in the period October - December, 1994.  Because there was no overlap of these two
 3     monitoring periods we analyzed each cohort separately.
 4          The data sets for the adults are shown in Tables 5-9 and 5-10, for the 13 and 24 adult
 5     cohorts, respectively.  In both cases the E-M algorithm could not be used to estimate the missing
 6     values because of the sparse coverage of these data.  The extensive missing data caused the
 7     estimation routines to converge for several values in the negative concentration domain, which
 8     has no physical meaning. The method chosen for analysis was to use a linear model that was
 9     fitted to these data which included terms for days and persons, but no interaction terms.
10          This model we chose is equivalent to running an unbalanced two-way analysis of variance
11     (ANOVA) on these data to estimate the marginal distribution of the means of the daily personal
12     exposure data.  Because we are only obtaining estimates  of the model parameters (not testing for
13     significance), the correlations between the people can be ignored (Winer, 1962). The last
14     columns in Tables 5-9 and 5-10 show the estimated  means that were used in the regression
15     analysis.
       October 1999                            5-61       DRAFT-DO NOT QUOTE OR CITE

-------
         CO
         c
         o
         (/)
         s_
         
-------
o
o
r-K
o
o
\Q
\Q
\Q









Lf\
i
S


o
Tl
H
O
O
o
H
O
c
o
H
W
0

HH
H
W





Day
102
103
202
203
702
902
903
1402
1403
1503
1701
2001
2102
2103
2203
2401
2701
2802
2803

2903

3003
3101
TABLE 5-9. PERSONAL PM10
SOURCE



Ambient 101 102 103 104
80.3
40.4
45.2 45.1
53.3 58.3
36.2 59.5 [68.4] 46.5 40.7
37.2
28.5 40.3 35.3
75.0 [66.1]
30.6 [95.6] 172.9
29.7 65.3
19.4 [70.9]
61.9
63.0 94.3 [95.2]
33.6 [76.8]
32.5 78.1 23.3
25.8 29.9
23.3 116.6 54.6
48.0 75.0 [108] 112.1
41.0 83.2 45.3

47.9

33.1 37.9
35.0 27.5
EXPOSURE 0"g/m3) OF 13 ADULTS IN AMSTERDAM,
OF DATA: JANSSEN ET AL. (1998A)
Person ID


105 107 109 110 111 113 114 116 118
143.4 71.8
94.7
56.2
50.5 125.9 87.1 58.8
172.9 87.5 [93.0]
43.9 75.1 32.8 79.1 38.6
39.7 [45.2] 33.1 57.4
[87.3] 47.0 65.2 88.7 72.9
42.1 40.4 70.4
[117] [68.4] 75.2 38.7
39.9
89.5 139.1 104.3 95.5
149.3 163.2 100.5 80.9
82.6 84.8 28.3 [70.3]
22.5 [58.6] 39.5
58.2 26.4 [99.9]
35.7 54.2 79.3 48.1
[80.0] 62.7 49.0 84.8
103.9 37.8 30.5 57.6 54.8

[69.7]

62.4
123.3 [70.7] 34.7 50.7
NL


Mean of
Personal
Raw Data
107.6
94.7
50.7
78.1
81.8
53.9
41.1
68.5
81.7
44.8
39.9
107.1
117.6
65.2
40.9
38.2
64.8
86.3
59

[69.7]

50.2
59.1
The data marked [ ], representing smoke exposure at home, were not used in the ANOVA. No estimate of the mean could be made for day 103



Mean of
Personal
Estimated*
90.6
-
52.4
70.2
66.4
60.8
55.2
81.5
88.8
58.1
40.7
77.8
124.7
48.1
65.2
41.1
56.7
87.6
58.0

-

68.9
69.4
because it had
only one datum point; person 102 was smoke exposed at home for all sampled days and was excluded from all analyses; on day 2903 there were no non-smoke
exposed data.

* - Estimated mean of all subjects' non-smoke PM exposure as computed by U.S. EPA via a 2-way ANOVA.

-------
TABLE 5-10. PERSONAL PM10 EXPOSURE fag/m3) OF 24 ADULTS IN AMSTERDAM, NL
o
o
•O
•O
•O








01
o^
Pv


o
£>
Tl
H
O
O
O
H
O
c
o
H
W
0
O
H
W

Day
\n
311
5U
6U
1\\
&U
1011
1212
1411
1512
1710
1711
1912
2010
2111
2212
2410
2411
2710
2811
3110

avjUKA^JL vjr jj/vi/v: jAmoojiim ILL p
Person ID
Ambient 301 302 303 306 307 308* 310 312 313 315 316 317 318 321 322
42.5 29.7 73.7 78.3 42.1
46.7 37 38.9 36.1 83.7
29 31.8 58.2 50
24.8 29.2 60.7 55.6
84.5 67 99.6 66.2 50.2 51.7 77.7
22.9 31.3 17.6 61.6 38.3 31 33.9
44.9 43.9 100.6 45.9 85.7 72 80.6
40.1 31.5 43.9 [41.6] 50.4 51.1 42.4 34.6 45.7
35.2 45.3 50.4 [23.3] 32.1 66.1
63.9 70.5 75.6 36.7 45.8 52.2
32.7 51.1 53.8 [92.2] 55.5 54.6
31 27.7 30.8 52.8 36 28.9
20.4 40.8 [90.9] 66.9 49.5 69.6 24 34 51.4
77.5 129.2 58.6 101.4 82.1 73.1
67.5 109.9 68.8 52.4
25.8
34 63.4 59.2
29.8 45.5 34.2 49.4 33.3 49.7 22.7 21.6
32.2 11.5 52.2 [35.8] 88.3 30.8 18.1
27.6 30.2 81.3 60.2 [71] 34.4 57.7
29.5 46.8 42.8 40.7 35.8 79.8

LLI. {lyyoA)

Mean of
Personal
325 327 331 333 334 336 337 338 339 Raw Data
71.7 57.4 61.7 86.0 109.8 67.8
73.9 56.8 86.2 58.9
31.3 40.3 45.0 42.8
48.5
114.4 55.8 75.6 59.6 71.8
44.0 26.5 59.8 68.1 41.2
50.3 [112.4] 68.4
44.2 75.4 50.7 47.0
[86.6] 34.1 61.6 63.4 50.4
69.9 75.7 65.7 85.8 64.2
53.8
69.4 40.9
71.6 42.7 84.9 53.5
80.1 81.8 86.6
87.1 48.3 40.9 68.9 51.8 75.6 67.1
71.5 50.6 92.7 71.6
70.0 35.6 56.6 61.3 34.0 54.3
104.3 52.2 45.9
32.6 46.6 40.0
51.5 51.1 57.1 52.9
31.0 42.8 45.7


Mean of
Personal
Estimated*
63.7
58.2
47.1
54.7
70.1
42.7
76.6
46.8
49.1
69.0
49.4
44.4
51.0
95.4
59.3
54.6
49.3
50.9
49.5
51.1
45.1

[ ] - smoker present at home
* - Estimated mean of all subjects' non-smoke PM exposure as computed by U.S. EPA via a 2 -way AN OVA.





-------
           120
     A
                                Ambient PM Concentration (|jg/m )
           140-
           120-
        E  100-
     B
        £   80-
to
o
Q.
UJ
60-
        CD   40-
        §
        0
        CL
            20-
                                                                     n = 13
                                              y = 36.4 +0.760 x
                                              R2 = 0.424
                     10      20     30     40     50     60     70     80      90
                                   Ambient PM Concentration (|jg/m3)
Figure 5-llA,B.  Completed mean personal PM10 exposures of two groups of adults (n = 24
                [A] and n = 13 [B]) not exposed to tobacco smoke at home in Amsterdam,
                NL, versus the simultaneous ambient PM10 measured in their community.
Source of data:  Janssenetal. (1998a)

October 1999
                           5-65
                                          DRAFT-DO NOT QUOTE OR CITE

-------
 1      ambient PM10 is reflected as an increase in the exposure to PM10 of the subjects in this study.
 2      These correlation coefficients for adults who were not smoke-exposed at home, are both higher
 3      than the estimated mean cross-sectional correlation of r = 0.50, shown in Table 5-7.
 4           The Janssen et al. (1999b) personal exposure intercept of 12//g/m3 for PM25 corresponds to
 5      the average portion of the daily PM25 exposure that is uncorrelated with the ambient PM25
 6      concentration, primarily while the subjects were indoors.  This quantity represents the
 7      combination of indoor emissions of PM25 and the children's 'personal cloud' of PM25 stirred up
 8      by their activities, and the net effect of experimental errors in ambient PM measurement. If the
 9      children's personal clouds for coarse PM were much greater than their personal clouds for fine
10      PM as in the PTEAM study (Ozkaynak et al., 1996a), and there was much more coarse PM
11      generated indoors than fine PM  (there was no cigarette smoke exposure at home or in school),
12      then their intercept for personal  PM10 exposure could be much greater and of the order of the
13      Lioy et al. (1990) intercept of 41 //g/m3 or the Janssen et al. (1999b) intercepts of 34 and
14      36 //g/m3  for PM10.  Values of this order are much larger than the Tamura et al. (1996a) intercept
15      value of 11 //g/m3 for PM10. This difference may be due in part to the relative cleanliness of the
16      Tokyo homes of the seven elderly Japanese subjects compared to the Phillipsburg, NJ and
17      Wageningen, Netherlands homes, and the lower physical  activity levels of these elderly people
18      compared to the activities of the monitored children and working adults in the Netherlands.
19           The results of these three regressions show that the variations in personal exposure for
20      these non-randomly chosen subjects who have similar life styles (habits and activities)  from day
21      to day, and who are not exposed to tobacco smoke on a routine basis, are driven by the variations
22      in the ambient PM concentration.  This finding, originally expounded by Janssen et al. (1995),
23      supports the plausibility of the use of fluctuations of ambient PM concentration as a surrogate for
24      fluctuations in human exposure  to PM of ambient origin for subjects residing in the community
25      surrounding an ambient monitoring station.  However, this plausibility is weakened by the
26      non-random choice of subjects and sampling periods which strictly limit the finding to the
27      subjects sampled on the days sampled.
28
29
        October 1999                             5-66        DRAFT-DO NOT QUOTE OR CITE

-------
 1      5.11  EXPOSURES TO AMBIENT PM25 OF PEOPLE BELIEVED TO BE
 2            SUSCEPTIBLE TO THE EFFECTS OF AMBIENT PM2 5
 3          The study of the historic air pollution episodes reveals that those people most subject to the
 4      mortality effects of exposure to ambient PM2 5 are the elderly with pre-existing cardio-pulmonary
 5      lesions (U.S. Environmental Protection Agency, 1996). Although it has been established in the
 6      previous section that, in general, peoples' exposures to PM of ambient origin are highly
 7      correlated with ambient PM in their community, prior to 1996, very few studies of the elderly
 8      and apparently susceptible people had been performed. For example, the Tamura et al. (1996a)
 9      study was of elderly people with no known cardio-pulmonary disease.
10          The question to be addressed here is 'Do elderly people with cardio-pulmonary disease
11      have the same relationship of their exposure to PM of ambient origin as do the members of the
12      general public who have been studied to date, as reported in the preceding section?' It may be
13      possible that the elderly or infirm modify their physical activities and living conditions in such a
14      manner as to influence the relationship of their exposures to ambient PM with the ambient PM
15      concentrations of their community.  If so, differences in exposure estimate errors between
16      susceptible-diseased groups and non-susceptible-healthy groups may result in differential
17      misclassification which has implications for the validity and interpretation of epidemiologic
18      studies (Armstrong et al.,  1992). The studies described in the following section address studies
19      of this apparently susceptible cohort.
20          Williams et al. (1999a) measured the personal exposures to PMj 5 of five elderly subjects
21      living at a retirement center in Baltimore, MD in January-February, 1997 for ten consecutive
22      days. Some of these subjects were under medical treatment for cardio-pulmonary conditions at
23      the time of this study, and they represent a non-random sample of elderly subjects living in an
24      environment of a retirement center. Each subject carried a personal PMj 5 monitor described by
25      Williams et  al. (1999b), and the ambient PM2 5 was monitored simultaneously outside the
26      retirement center by a dichotomous sampler.  Three subjects (1-3) lived apart in separate
27      efficiency apartments in a new unit and the two other subjects (4, 5) lived apart in an older unit,
28      in two separate smaller rooms, each with an attached bathroom.  The three newer apartments had
29      either a picture window or double windows and the two older single-rooms had only one
30      window. Thus, even though none of the residents reported opening windows during the study,
31      the newer units may have had a larger window-perimeter leakage area for air exchange directly

        October 1999                             5-67        DRAFT-DO NOT QUOTE OR CITE

-------
 1     with the outdoor air.  The new and old units also had different air handling systems in their two
 2     general living areas which were connected by staircases and elevators. The authors reported that
 3     there were no special air cleaners installed in either of these two air handling systems.
 4          The collected data are shown in Table 5-11 with the missing values estimated by the E-M
 5     algorithm (Mage et al., 1999) shown within parentheses ( ). Because of the different air
 6     handling systems for their living areas, the E-M algorithm was applied separately to these data
 7     for subjects 1-3, and then to these data for subjects 4 and 5.  These different air handling
 8     systems, and the difference in window leakage areas, may be a partial explanation of why
 9     subjects 1-3 had positive correlations of their exposures with the ambient PM ( r = 0.91, 0.12,
10     0.70 respectively),  and subjects 4 and 5 had negative correlations of their exposures with the
11     ambient PM ( r = -0.55, -0.05 respectively). The mean PMj 5 exposure vs ambient PM2 5 for all
12     five subjects is  shown as Figure 5-12, with an R2 of 27%. The range of nine observations of
13     ambient PM25 is fairly narrow (8 to 23 //g/m3) and inclusion of the last day of data with an
14     ambient PM2 5 value of 32 //g/m3 raises the correlation of their mean PM exposure with the
15     ambient PM from r = 0.269 to r = 0.522. It is expected that all these correlations would have
16     been higher than shown if both the ambient monitor and the personal monitor measured PM with
17     the same nominal size cut.
18          Bahadori  (1998) and Bahadori et al. (1999) report on a pilot study of the PM exposure of
19     ten non-randomly chosen Chronic Obstructive Pulmonary Disease (COPD) patients in Nashville,
20     TN, during the  summer of 1995. Each subject alternately carried a personal PM2 5 or PM10
21     monitor for a 12-h  daytime period (8 a.m. - 8 p.m.), for six consecutive days. These same
22     pollutants were monitored simultaneously indoors and outside their homes.  These homes were
23     all air conditioned with lower air exchange rates (mean a =  0.57/h), which may have contributed
24     to the finding that mean indoor PM2 5 was 66% of the mean ambient PM2 5. This can be
25     contrasted to the PTEAM study in Riverside, CA, where no air conditioners were in use and the
26     air exchange rate was much higher (mean a > 0.97/h). In the PTEAM study the mean indoor
27     PM25 was 98% of the mean ambient PM25 (Clayton et al., 1993).
28          Because each person carried a PM2 5 or a PM10 personal monitor for only three 12-h periods,
29     no Pearson correlations of personal PM vs ambient PM were reported for the individual subjects.
30     However, these data were combined into two groups of 30 and Bahadori et al. (1999) report the
31     correlations found between personal and outdoor concentrations when data were analyzed

       October 1999                             5-68        DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 5-11. PERSONAL EXPOSURES TO PM15 OF FIVE ELDERLY RESIDENTS
             OF A RETIREMENT CENTER IN BALTIMORE, MD, AS A FUNCTION OF
                             AMBIENT PM2 5 CONCENTRATION
Day
1
2
o
3
4
5
6
1
8
9
10
Correlation
Ambient
PM25
14.8
9.2
14.2
14.0
22.6
21.2
11.6
8.0
14.5
32.0

Person 1
Exposure
21
22
17
25
33
32
27
12
29
50
r = 0.907
Person 2
Exposure
31
16
35
20
38
29
33
35
45
(29.5)
r = 0.122
Person 3
Exposure
(32.0)
(38.7)
18
(39.5)
(39.6)
44
37
30
(35.2)
57
r = 0.702
Person 4
Exposure
(33.1)
39
40
18
32
26
51
39
38
(24.7)
r = -0.550
Person 5
Exposure
31
27
43
30
30
35
19
45
58
(34.2)
r = -0.050
Mean
Exposure
29.6
28.5
30.6
26.5
34.5
33.2
33.4
32.2
41.0
39.1
r = 0.522
        ( ) Missing data estimated by U.S. EPA using E-M algorithm (U.S. Environmental Protection Agency, 1996; Mage et al., 1999).
        Source of data: Williams et al. (1999a).
 1     cross-sectionally were r = 0.09 for PM2 5 and r = -0.08 for PM10. The authors suggest that
 2     "personal-outdoor correlations would be stronger if data were to be analyzed longitudinally.
 3     However, this analysis could not be performed due to insufficient data." That relationship, if
 4     true, would be consistent with that in U.S. Environmental Protection Agency (1996) where it was
 5     shown how combining personal exposures for an individual with those for other individuals,
 6     leads to a degradation of the correlation coefficient.
 7          Rojas-Bracho et al. (1999) report the results of a 1996/1997  study of the personal exposures
 8     to PM25 and PM10 of 18 COPD patients in the Boston, MA metropolitan area. The study design
 9     was based on the pilot study by Bahadori et al.  (1999) in Nashville, TN.  The subjects were also
10     not chosen as a random sample from the population of COPD patients in the Boston area, but
11     were recruited via local physicians, from COPD exercise groups, as well as by newspaper
12     advertisements. Therefore the results of the study strictly only apply to the 18 subjects on the
13     days sampled, and cannot be inferred to relate to Boston area COPD patients in general.
14
       October 1999                             5-69        DRAFT-DO NOT QUOTE OR CITE

-------
                                     Ambient PM25 Concentration (|jg/m3)
       Figure 5-12.  Mean personal exposures to PMj 5 of five elderly residents of a retirement
                    home in Baltimore, MD, as a function of ambient PM2 5 concentration.
                    Source of data: Williams et al. (1999a).
 1          Each subject carried a personal monitor that sampled both PM25 and PM10 for 12-hours per
 2     day for one, two or three consecutive 6-day periods between 8 a.m. and 8 p.m., while PM25 and
 3     PM10were also measured inside and outside their homes during the same time period.  Subjects
 4     were monitored for either summer only (2), winter only (2) or in both summer and winter(14).
 5     The regression coefficients (R2) between the daytime personal exposure and daytime outdoor PM
 6     measures for the 15 subjects with more than one week of sampling are shown in Table 5-12a.
 7     The 0.85 maximum value in the set is for the one subject who was only sampled in the summer
 8     period. As expected, the combining of summer and winter data reduced the correlation between
 9     personal and outdoor PM data because  air exchange rates in Boston are decreased in the winter to
10     save heat.  The article does not report the difference between summer and winter correlations for
11     the other 14 subjects.
       October 1999
5-70
DRAFT-DO NOT QUOTE OR CITE

-------
             TABLE 5-12A.  SUMMARY OF CORRELATIONS BETWEEN DAYTIME
       PERSONAL PM EXPOSURES AND DAYTIME OUTDOOR PM CONCENTRATIONS
            FOR 15 COPD PATIENTS DURING COMBINED SUMMER AND WINTER
                                   PERIODS IN BOSTON, MA
Person ID
1
2
3
4
5
6
9
10
11
12
13
14
16
17
18
Weeks sampled in
Summer (S) and
Winter (W)
S 1,W1
S2, Wl
S 1, Wl
S 1,W1
S 1,W1
S 1, Wl
S2, Wl
S2, Wl
S2, Wl
S 1, Wl
S2
S2, Wl
S1,W1
S2, Wl
S2, Wl
Number of
valid paired
observations
10
10
17
9
12
11
12
17
17
17
10
16
12
17
17
R2 PM25
Personal vs Outdoor
0.58
0.21
0.13
0.44
0.28
0.01
0.65
0.49
0.32
0.02
0.85 (maximum value)
0.83
0.52
0.01*
0.37
R2 PM10-PM25
Personal vs Outdoor
0.11*
0.10
0.06*
0.26*
0.02*
0.31
0.00
0.00
0.00*
0.09
0.00* (9 observations)
0.00
0.12
0.05*
0.39
       * Correlation coefficient (r) is negative.
       Source: Rojas-Bracho et al. (1999).
1          Table 5-12a shows that the personal exposures to PM2 5 were all positively correlated with
2     the ambient PM, except for subject 17 (the slope of the regression line is negative). However, for
3     the difference between PM10 and PM2 5, representing the coarse mode fraction between 2.5 and
4     10 microns AD (neglecting cutpoint imprecision), there were 7 negative correlations, one zero
5     correlation, and 7 positive correlations, which is an expected result if there is little or no
6     correlation, on average, between ambient coarse PM and personal exposure to coarse PM for
      October 1999
5-71
DRAFT-DO NOT QUOTE OR CITE

-------
 1      such people who spend the majority of their time indoors where coarse PM is generated by daily
 2      activities.
 3
 4
 5      5.12  PERSONAL EXPOSURE TO CONSTITUENTS OF PM:  SULFATES
 6            AND ACIDITY
 7          PM sulfates are species that exist in the ambient atmosphere from primary emissions of
 8      combustion products of fossil fuels containing sulfur, from formation by photochemical
 9      oxidation of gaseous sulfur species, and from non-anthropogenic sources (e.g. volcanic activity
10      and wind-blown soil). In the indoor environment, where ammonia is present, the only common
11      major source of sulfate may be resuspension by human activity of PM containing ammonium
12      sulfates and soil sulfates  that were tracked into the home. In some homes an unvented kerosine
13      heater using a high-sulfur fuel may be a major winter contributor (Leaderer et al., 1999a).
14          Trijonis et al. (1980) reported surface monitoring sulfate data from the U.S. EPA Regional
15      Air Pollution Study (RAPS) of the greater St. Louis, MO metropolitan area including portions of
16      Illinois.  The  authors reported that all site-to-site correlation coefficients of 24-h sulfate data were
17      greater than 0.84. Suh et al. (1993) reported a study of personal exposure to PM sulfate of
18      children living in State College, PA during a summer season.  Figure 5-13 shows the excellent
19      agreement between the personal exposure and ambient sulfate concentration. The regression
20      slope of 0.87 falls below the 1:1 line because sulfates, although primarily of < 1 //m AD, deposit
21      on inner home surfaces with a value of k of order 0.16/hr (Ozkaynak et al., 1996b).  This leads to
22      a value of [Pa / (a + k)] of slightly less than 1 in the summer when homes are not air conditioned
23      and sealed tightly (a »0 and P ~ 1).
24          If the home is sealed to conserve energy with air conditioning use, the air exchange
25      parameter a is decreased. If the return-air to the air conditioner is filtered, or passes through an
26      electrostatic precipitator, then the deposition parameter k is increased as some sulfates may be
27      removed. This  leads to a decrease in the parameter [Pa/(a + k)], as shown by the data on
28      Figure 5-13 where the open circles representing the air conditoned homes fall below the data for
29      the non air conditioned homes. The overall regression coefficient for these data from non air
30      conditioned homes is R2  = 0.92, indicating how well personal exposure to this material of almost
31      exclusively ambient origin correlates with ambient concentration.

        October  1999                            5-72        DRAFT-DO NOT QUOTE OR CITE

-------
                             600
                             500
                           0)
                           O 400
                           E
                           ^
                           
-------
 1           The effect of air conditioning may be important to consider when comparing responses to
 2      air pollution in cities with different percentages of air conditioned homes.  For example, if cities
 3      A and B have identical ambient air pollution concentrations and city A is completely air
 4      conditioned while city B has no need for air conditioning, the subjects in city A will be exposed
 5      to less ambient PM than the subjects in city B, and consequently, if the subjects in city A have
 6      less symptoms, the ambient PM in city A may appear to be less toxic.  See the discussion of the
 7      effect of assumed protection by going indoors when smog warnings are issued, from the 1969
 8      PM AQCD (National Air Pollution Control Association, 1969) mentioned in Section 5.1.1.
 9           Ozkaynak et al. (1996a,b) and Janssen et al. (1999c) report that in the U.S. PTEAM and
10      Netherlands studies respectively, that XRF analyses of indoor PM and the immediate outdoor
11      PM show that sulfur is the only element reported with virtually identical mass concentration in
12      both the indoor and outdoor air. Therefore, where there are no indoor sources of fine mode
13      sulfates, and no enhanced air cleaning systems operating, one may deduce the same relationship
14      between personal exposure to PM of ambient origin (C) and ambient concentrations (Co) in an
15      indoor microenvironment for other materials in the same aerodynamic size range as the fine
16      mode sulfates found in the atmosphere.
17           Larssen et al. (1993) report such a study of traffic generated PM in Oslo, Norway.  They
18      made measurements of indoor and outdoor fine fraction PM25 and coarse fraction (PM10 - PM25)
19      in a closed and uninhabited apartment facing  on a busy street.  Figure 5-14 presents three
20      examples of their daily data. In the absence of indoor sources, these data show that with low air
21      exchange rates  (not monitored), the PM equilibrium parameter [P a I (a + k)] is of order 5-10%
22      for the coarse PM fraction and about 80% for the fine PM fraction.
23           Tung et al. (1999) developed a methodology to measure the particulate penetration
24      coefficient (P) through a building shell by measuring simultaneously the air exchange rate (a),
25      and the indoor and outdoor PM concentrations as the indoor PM equilibrates with the outdoor
26      PM in the absence of any PM sources within the indoor microenvironment (microenvironment).
27      The PM deposition parameter (K) is then estimated by the measured decay rate of the PM
28      (e.g., a + K) minus the measured air exchange rate (a), and the penetration is computed as
29      P ~  C (a + k) I (Ca a).  The authors demonstrated their technique using a large enclosed office as
30      the simulated building microenvironment, while it was unoccupied at night with all mechanical
31      ventilation shut off. The office had two closed doors leading to a corridor with closed louvers

        October 1999                             5-74        DRAFT-DO NOT QUOTE OR CITE

-------
       250-1
        50-|
                 Coarse Fraction (2.5—10 |jm)
                     Outside
                     Inside
                   Day     Night     Day     Night     Day     Night
                 Fine Fraction (< 2.5 |jm)
                     Outside
                     Inside
                   Day     Night     Day     Night     Day     Night
Figure 5-14.  Results from simultaneous measurements of the indoor PM10 concentration
            and the immediate outdoor PM10 concentration of an uninhabited apartment
            in a building fronting to a busy street in Oslo, Norway.

Source: Larssen et al. (1993).
October 1999
5-75
DRAFT-DO NOT QUOTE OR CITE

-------
 1      above them, and the corridor PM simulated the ambient PM. In three tests of the procedure the
 2      penetration of PM10into the office had an average value of P ~ 0.8 corresponding to an average
 3      air exchange rate of a ~ 0.4/h.
 4           Ormstad et al. (1997) measured PM10 indoors and outdoors for 29 occupied homes in Oslo,
 5      Norway. Using a 60 Lpm open face sampler, pointing 45° downward, they found that the
 6      median indoor/outdoor mass ratio was 1.43. PM2 5 was not monitored, but the authors used
 7      transmission electron microscopy (TEM) to estimate that the vast majority of countable
 8      particles(> 90%) were under 2.5//m optical diameter. It is interesting to compare these data with
 9      the Larssen et al. (1993) data, cited just above. For a closed and unoccupied apartment in Oslo,
10      the indoor/outdoor ratio was of order 0.2. The difference between these indoor/outdoor ratios
11      (~ 1.4 vs -0.2) may represent the effects of indoor PM generated and resuspended by occupants,
12      and an increase in air exchange rate with door and window openings and closings by the
13      occupants.
14           Brauer et al. (1989) measured personal exposures and ambient concentrations of acidic
15      aerosols and gases using a personal annual denuder/filter pack sampling system.  They found that
16      personal exposures to aerosol strong acidity (H+) were slightly lower than concentrations
17      measured at a stationary site, and concentrations of sulfate and ammonium ions were similar to
18      those measured at the stationary site.
19           In the absence of personal activity to resuspend or generate PM, and with a very low air
20      exchange rate, the indoor PM fine fraction is virtually all of ambient origin and the PM coarse
21      fraction is minimal due to its higher settling values (k) and lower penetration values (P). The
22      caveat to this analogy is that while indoors, charged ambient particles and ultra-fine ambient PM
23      (< 0.1 //m AD) may have greater electrostatic deposition and diffusion deposition than the larger
24      sulfate particles (> 0.1 //m AD), which would inflate their deposition parameter k. Although the
25      correlation between the ambient concentrations of such particles and the related exposure
26      measurements may be much lower due to the fluctuations of the transient forces causing
27      deposition, it remains that any exposure to these  PM constituents which have no indoor sources,
28      will be entirely caused by the PM of ambient origin.
29
30


        October 1999                              5-76         DRAFT-DO NOT QUOTE OR CITE

-------
 1     5.13   CORRELATION OF AMBIENT PM AND PM EXPOSURE IN
 2            CROSS-SECTIONAL-TYPE STUDIES
 3          The results for personal sulfate exposure and longitudinal (Type-1) personal PM exposure
 4     studies stand in contrast to the lack of correlation found in cross-sectional studies (Type-2)
 5     described in Chapter 7 of U.S. Environmental Protection Agency (1996).  Typical of these
 6     studies is Spengler et al. (1985), which found virtually no cross-sectional correlation (r ~ 0)
 7     between personal exposure to PM3 5 and ambient PM3 5 concentration in the adjacent
 8     communities of Kingston and Harriman, in Eastern Tennessee. The reason for this difference
 9     can be visualized by comparison of the example data matrices for the Type-1 and Type-2 studies
10     shown in Tables 5-12 and 5-13:  The longitudinal Type-1 studies of Tamura et al. (1996a),
11     Janssen et al. (1999b) and Lioy et al. (1990) correspond to Table 5-12. However, the Spengler
12     et al. (1985) study and the PTEAM study (Clayton et al., 1993) were of Type-2, the class covered
13     by Table 5-13.
14
15
         TABLE 5-12.  EXAMPLE OF A COMPLETED MATRIX FOR A TYPE-1 ANALYSIS
                    OF PM EXPOSURE AND AMBIENT PM CONCENTRATION

Day 1
Day 2
Day 3
Day 4
Person 1
PM Exposure
X
X
*(X)
X
Person 2
PM Exposure
*(X)
X
X
X
Person 3
PM Exposure
X
*(X)
X
X
Person 4
PM Exposure
X
X
X
*(X)
Ambient PM
Concentration
X
X
X
X
        *(X) missing data filled in by the E-M algorithm from assumption of bivariate normal distributions of daily personal exposures and each
         individual's exposure time series.
 1          In the Spengler et al. (1985) study, subsets of people were sampled on subsets of days
 2     without any overlap. In the PTEAM study four different people were to be sampled on each
 3     consecutive day with no repeat sampling of any person, but some people's data were missed due
 4     to happenstance. Without any overlapping of sampling between all people one is unable to
 5

       October 1999                            5-77       DRAFT-DO NOT QUOTE OR CITE

-------
         TABLE 5-13. EXAMPLE OF AN INCOMPLETE MATRIX FOR TYPE-2 ANALYSIS
                    OF PM EXPOSURE AND AMBIENT PM CONCENTRATION

Day 1
Day 2
Day 3
Day 4
Person 1
PM Exposure
X
X
_*
.*
Person 2
PM Exposure
X
X
_*
.*
Person 3
PM Exposure
.*
_*
X
X
Person 4
PM Exposure
.*
_*
X
X
Ambient PM
Concentration
X
X
X
X
        *Missing data that cannot be estimated by E-M algorithm because there is no comparison data on how the
         personal PM exposures of Persons -3 and -4 relate to the simultaneous personal PM exposures of Persons
         - 1 and -2.
 1     estimate the missing data by the E-M algorithm in these two studies. For example, if person-3 in
 2     an extensive data set always has the mean personal exposure of person-1 and person-2 we can
 3     estimate person-3's exposure as the mean of the person-1 and person-2 exposures on the day
 4     when person-3 was not sampled. However, if person-3 was never sampled simultaneously with
 5     persons -1 and -2 this relationship cannot be discovered. Consequently, the E-M algorithm
 6     cannot be applied to complete the Type-2 matrices of the form shown as Table 5-13.
 7          In the PTEAM study all subjects were a set of randomly sampled residents of Riverside,
 8     CA aged 10 and above, who were self-declared nonsmokers. The daily PM exposure data from
 9     the subjects were averaged to create a time series of mean exposure values that were an estimate
10     of the mean personal exposure of the Riverside cohort on those days.  Most days had four
11     observations but some days had only two or three observations, so the means of the four values
12     are expected to have smaller confidence intervals than the means of the other values. Each
13     missing personal PM exposure datum was estimated using the E-M algorithm. This amounts to
14     finding a set of missing values that maintain an assumed bivariate normal correlation structure of
15     the time series of completed (sets of 4) mean PM exposure data and ambient PM data.
16          The resulting table  of average PM10 personal exposure and average ambient PM10
17     concentration of 43 days of data is as provided in Table 7-48 in U.S. Environmental Protection
18     Agency (1996). Figure 5-15 shows these data, and the regression parameters are given in
19     Table 5-14.

       October 1999                             5-78        DRAFT-DO NOT QUOTE OR CITE

-------
                  200-
                  150-
                                              • •
               CO
               c
               o
               cu
               Q.
                  100-
                  50-
                                50          100          150

                                      Ambient SAM |jg/m3
                            200
 Figure 5-15.  PTEAM mean 24-h PM10 data compared for personal PEM and SAM.


 Source: U.S. EPA reanalyses of data reported on by Ozkaynak et al. (1996a).
  TABLE 5-14.  RESULTS OF THE ANALYSIS OF THE PTEAM DAILY AVERAGE

       PERSONAL PM10 EXPOSURE DATA FROM RIVERSIDE, CALIFORNIA
Regression Model
Linear, normal error
Linear, lognormal error
Orthogonal
Linear adjusted for person error
Measures of Association
Correlation of averages
Fraction of variation explained by orthogonal regression
Slope
0.6174
0.6185
0.8071
0.9675



Intercept
59.7
57.1
44.2
31.0
Value
0.721
0.864
 Source: U.S. Environmental Protection Agency (1996) reanalysis of PTEAM data (Ozkaynak et al.,1996a,b).
October 1999
5-79
DRAFT-DO NOT QUOTE OR CITE

-------
 1           The regression R2 value for these averages is 0.52 which is much less than the regression
 2      R2 values for the longitudinal time-series data averages of order 0.9 corresponding to a completed
 3      matrix (Table 5-12). This, in effect, is caused by the wider variance in the increment of non-
 4      ambient PM added to the PTEAM subjects' exposure to ambient PM, produced by the day-to-day
 5      variation of the indoor PM sources and personal activities of the different people who were
 6      sampled for only a single 24-hr period. For example, on day 1, the four people chosen may not
 7      live with smokers and have personal PM exposures at or moderately above the ambient PM
 8      value. On day 2, if some of the new subjects live with smokers, their personal exposures may be
 9      well above the ambient PM value.
10           The coherence of the time-series can be reduced by treating all such individuals together
11      and the underlying high correlation of individuals with similar day-to-day exposure to PM of
12      non-ambient origin is lost, as was shown by Figure 7-26 of the previous PM AQCD (U.S.
13      Environmental Protection Agency, 1996). Janssen (1998) combined data from subjects with high
14      longitudinal correlations, by randomly sampling from each individual's data set and combining
15      them to create a cross-sectional data set.  In each case, as shown in Table 5-7, the correlation
16      coefficients were significantly reduced. It is this process of combining sets of people with
17      assumed similar exposure to ambient PM but widely different exposures to non-ambient PM on
18      different days that leads to the published studies that report little or no statistically significant
19      correlation between personal exposure to the total PM and the corresponding ambient PM
20      concentrations (e.g., Sexton et al, 1984; Spengler et al, 1985; Pellizzari et al, 1999).
21
22
23      5.14  EXPOSURE TO  AMBIENT GASEOUS  POLLUTANTS RELATED
24            TO AMBIENT PM
25           Several gaseous pollutants, such as SO2, NO (rapidly converted to NO2) and CO, are often
26      emitted at the same time as ambient PM is emitted from combustion sources. In addition,
27      photochemical oxidants, such as ozone (O3) are also formed  in the atmosphere by the same
28      processes that lead to formation of some of the photochemical aerosol species found in the fine-
29      mode PM fraction (Seinfeld and Pandis, 1998). Weather also introduces a correlation effect as a
30      stagnation inversion may cause all pollutant concentrations to rise, and a rain storm or frontal
31      passage may cause all pollutant concentrations to fall. Consequently, these gaseous pollutants

        October 1999                             5-80       DRAFT-DO NOT QUOTE OR CITE

-------
 1      may have ambient concentrations that are correlated with the concentrations of various fractions
 2      of the ambient PM mixture [ultra-fine PM, fine-mode PM, or coarse-mode PM less than 10//m
 3      AD].
 4           If these gaseous species are capable of producing the same effect as an ambient PM
 5      fraction, and their ambient concentrations are correlated with ambient PM concentration, it
 6      becomes difficult to separate out which factor, if any, is causing the observed effect (Morris and
 7      Naumova, 1998; Chen et al., 1999).  It is assumed that the generation of the primary pollutants
 8      CO, NO2 and SO2 in residential and occupational microenvironments is independent of ambient
 9      concentrations of those gases. However, there may be some locations where indoor emissions
10      may be correlated with ambient concentrations as in the winter heating season. Assuming a
11      linear system and no differential measurement errors, the health effects of indoor generated CO,
12      NO2 and SO2 would also be independent of, and uncorrelated with, the health effects of the
13      ambient PM.  If these assumptions are valid, the presence of indoor sources of these gases should
14      not confound the ambient PM mortality relationships reported in the literature.
15           An investigation (as yet unpublished) has been performed on the general correlation
16      behavior of ambient PM and the ambient gaseous criteria pollutants in the U.S. (Shadwick et al.,
17      1999). In order to examine the general geographic variability of the correlation coefficient, the
18      U.S. EPA public Aerometric Information Retrieval System (AIRS) data base for the years
19      1992 - 1996 was reviewed (http://www.epa.gov/airs/airs2.html). Shadwick et al. (1999) chose
20      the highest PM urban site and lowest rural PM site from each of the 50 states, Puerto Rico and
21      Washington B.C. irrespective of the years of available data.  The distributions of correlations
22      between ambient PM10 and the 24-h concentrations of CO, NO2, and SO2, and ozone (8-h and
23      1-h), for urban and rural sites respectively, are shown in Figures 5-16a,b to 5-20a,b for the
24      chosen urban and rural sites. For PM25 only 38 sites were  available for the entire U.S. so the
25      urban and rural stations  were combined for a joint analysis. The results for ambient PM^ and
26      the gaseous species CO, NO2, SO2 and ozone (8-h and 1-h), for combined urban and rural sites
27      are shown in Figures 5-16c to 5-20c. These data are summarized in Table 5-15.
28           To estimate the relationship between  exposure to ambient PM and exposures to the primary
29      gases of ambient origin, their hourly averages along with an air exchange rate estimate would be
30      required.  For the highly reactive gases with strong sinks on indoor surfaces (such as ozone


        October  1999                             5-81        DRAFT-DO NOT QUOTE OR CITE

-------
     A
9-
8-
7
0 Q_
£Z D
0) E_
3 5
5f 4-
0) H
LJ_ o
2-
1-
0-







Whitefish


i i i i







MN
1
|






























































Denver, CO
1
| |
i i i i i i
-07-06-05-04 -03-02-01 00 01 02 03 04 05 06 07 08 09
Urban Correlation Between PM10 and CO
15-
14-
13-
12-
11-
>, 10-
c 9-
0) 7-
5-
4-
3-
2-
1-










Columbia Falls, MN
1

1 1 1 1 1

| |


^_^^



	





—




































1
— 1



^^ Central Point, OR
|
ll— 1

-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rural Correlation Between PM10 and CO
10-
9-
8-
7
I1 «-
^ 5-
§- 4-
l_ o _
LJ_ °
2-
1-









^^^^^
_
Blue Island (Chicago), IL



i i i i i
i









— — 1 1 	





1 — 1

















Mammoth Lake, CA
1
rnr~i
i i i i i i i
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
                               Correlation Between PM,, and CO
Figure 5-16.  (a) Correlations of PM10-CO for highest urban PM10 site per state;
             (b) correlations of PM10-CO for lowest rural PM10 site per state; and
             (c) correlations of PM2 5-CO.

Source: Shadwick et al. (1999).
October 1999
5-82
DRAFT-DO NOT QUOTE OR CITE

-------
11-
10-
9-
>, 8-
0) 6-
Ag- 5-
lt 4-
3-
2-
1-







Philadelphia, PA
1














U 1 1 1 1 II 1 1 1 1
—




























Portland, OR






1











i i i i i
             -0.7  -0.6  -0.5  -0.4  -0.3 -0.2  -0.1  0.0  0.1  0.2  0.3  0.4 0.5   0.6 0.7  0.8  0.9

                             Urban Correlation Between PMin and NO,
       B
11-
10-
9-
8-
l°-
CD 6-
I 5-
LL 4-
3-
2-
1-
Christian County, KY
1


	
	


-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.


3 0
	 1 Salt Lake County, UT
, 	 , i
1
n
4 0.5 0.6 0.7 0.8 0.9
                               Rural Correlation Between PM10 and NO2
         0
         (D
11-
10-
9-
8_
7-
6-
5-
4-
3-
2-
1-







Columbia, SC
1
rnrn
^~











^



—




^_^_












Calexico, CA
|





-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Correlation Between PM25 and NO2
Figure 5-17. (a) Correlations of PM10-NO2 for highest urban PM10 site per state;
             (b) correlations of PM10-NO2 for lowest rural PM10 site per state; and
             (c) correlations of PM2 5-NO2.
Source: Shadwick et al. (1999).

October 1999
5-83
DRAFT-DO NOT QUOTE OR CITE

-------
15-
14-
13-
12-
11-
10-
9-
8-
6-
5-
4-
3-
2-
1-








Philadelphia, PA
1
n r~i
1 	 1


—

















—






l~~l — ~l
Sitka, AK
1
i—iH
                    -0.7 -0.6 -0.5 -0.4  -0.3  -0.2 -0.1 0.0  0.1  0.2 0.3  0.4  0.5  0.6  0.7  0.8  0.9

                                   Urban Correlation Between PM10 and S02
11-
10-
9-
B| e-

-------
15-
14-
13-
12-
11-
>, 10-
c 9-
n 13 8-
Ac^ 7-
iV" 6-
u_ u
5-
4-
3-

2-

1-














Salt Lake City, UT 1 	 1
1 1 1
1 II
n 1 1












— — |

















__





-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0






















































__

























Atlanta, GA
1
____










i i i i i iii
1 0.2 0.3 0.4 n 5 D R 07 n R 09
14 Urban Correlation Between PM10 and O3 Maximum 1-hr Average
13-
12-
11-
10-
O1 9-
g 8-
B §- 7-
CD 6-
5-
4-
3-
2-
1-








































Iberville Parish, LA






I

nn

-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0
1 1







—






	










	




	











	

















Kent County, Rl
i
i 	



i i i i i i
1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
            10
                       Rural Correlation Between PM10 and 03 Maximum 1-hr Average
Frequency
D_iK)W^UlC»-
1 1 1 1 1 1 1
Adams County, CO
i
nn


nf
n


n
	

Columbia, SC
1
             -0.7 -0.6 -0.5 -0.4  -0.3 -0.2 -0.1  0.0  0.1  0.2  0.3  0.4  0.5  0.6  0.7 0.8   0.9

                      Correlation Between PM25 and Maximum 031-hr Average
Figure 5-19.  (a) Correlations of PM10-O31-hr for highest urban PM10 site per state;
              (b) correlations of PM10-O31-hr for lowest rural PM10 site per state and
              (c) correlations of PM2 5-O31-hr.
Source: Shadwick et al. (1999).
 October 1999
5-85
DRAFT-DO NOT QUOTE OR CITE

-------
          A
            (D
15-
14-
13-
12-
11-
10-
9-
8-
6-
5-
4-
3-
2-
1-







^^-^^

Salt Lake City, UT
1



nH









n

—















^^


Providence, Rl
1

                -0.7  -0.6  -0.5  -0.4 -0.3 -0.2 -0.1 0.0 0.1   0.2 0.3   0.4 0.5  0.6  0.7  0.8  0.9

                       Urban Correlation Between PM10 and O3 Maximum 8-hr Average
               15-f
               14-
               13-
               12-
               11-
          B
            cr
            0)
7-
6-
5-
4-
3-
2-
1-
                                                                          Kent County, Rl
                -0.7  -0.6  -0.5  -0.4 -0.3 -0.2 -0.1  0.0  0.1  0.2  0.3  0.4  0.5  0.6  0.7  0.8  0.9
                        Rural Correlation Between  PM10 and O3 Maximum 8-hr Average
            0)
10-
9-
8-
7-
5-
4-
3-
2-
1-
Ad
r
amsCoun
~|
ty, CC

fin

nfln

n
—
Columbia, SC
n
                -0.7  -0.6  -0.5  -0.4 -0.3 -0.2 -0.1 0.0 0.1   0.2 0.3   0.4 0.5  0.6  0.7  0.8  0.9

                         Correlation Between PM25 and O3 Maximum 8-hr Average
Figure 5-20.  (a) Correlations of PM10-O3 8-hr for highest urban PM10 site per state;
              (b) correlations of PM10-O3 8-hr for lowest rural PM10 site per state; and
              (c) correlations of PM2 5-O3 8-hr.
Source: Shadwick et al. (1999).

October 1999
                              5-86
DRAFT-DO NOT QUOTE OR CITE

-------
              TABLE 5-15. AVERAGE ANNUAL CORRELATIONS BETWEEN PM AND
              CRITERIA GASEOUS POLLUTANTS IN THE UNITED STATES (Std. Dev.)
         Correlation PM vs Gas     vs CO      vs NO2      vs SO2      vs O3_lh      vs O3_8h
         Urban PM10             0.27(0.22)  0.41(0.21)   0.27(0.21)  0.31(0.25)   0.29(0.26)
         Rural PM10             0.09(0.18)  0.25(0.21)   0.10(0.13)  0.49(0.17)   0.46(0.18)
         Combined PM,,	0.44(0.29)  0.52(0.18)   0.20(0.20)  0.08(0.36)   0.03(0.38)
        Source of Data: Shadwick et al. (1999).
 1     which also oxidizes terpenes) their fractions of the ambient concentrations found indoors can be
 2     of order zero.  In the absence of hourly PM data and air exchange rate data the estimation of a
 3     correlation between the ambient PM exposure and the exposure to the gases of ambient origin
 4     cannot be made without introducing assumptions about the hourly variability of ambient PM and
 5     air exchange rate. However, it is evident that the correlation between the personal exposures to
 6     ambient PM and personal exposures to ambient  ozone, NO2 and SO2 must be closer to zero than
 7     the corresponding correlation of their respective atmospheric concentrations, because their
 8     different indoor deposition rates will create different and variable fractional exposures.
 9           Carbon monoxide (CO), however, is virtually non-reactive in indoor micro-environments
10     and it has no known surface sinks indoors.  On average, a subject will be exposed to virtually
11     100% of the ambient CO concentration. Consequently the correlation between exposure to CO
12     of ambient origin and exposure to PM of ambient origin will closely approach the correlation of
13     the ambient PM concentration and the ambient CO concentration. However, the presence of
14     indoor sources of CO, such as tobacco smoking, and outdoor in-vehicle exposures to traffic CO,
15     will make the correlation between personal exposure to total CO and ambient PM much lower, in
16     the same way that indoor sources of PM degrade the correlation of total PM exposure and
17     ambient PM concentration.
18           Table 5-15 shows that on an annual basis PM10 is generally correlated positively with
19     almost all gaseous criteria pollutants. The urban correlations are higher than the rural
20     correlations for the daily averages of primary pollutants (CO, NO2 and SO2) and lower for the
21     secondary pollutant ozone (1-h and  8-h). Examination of the corresponding histograms, with

       October 1999                              5-87        DRAFT-DO NOT QUOTE OR CITE

-------
 1      standard deviations of order 0.2, indicate that there is no general relationship; there are locations
 2      that can have no significant correlation (of order zero) and appreciable correlation (of order 0.5).
 3      Table 5-15 shows that PM25 is, on the average, most strongly correlated with CO and NO2,
 4      moderately correlated with SO2, and not at all correlated with ozone in the combined urban and
 5      rural locations we have analyzed.
 6           The correlations for PM25 with ozone (maximum 1-h and maximum 8-h) for both urban
 7      and rural sites combined has the largest variance of all the data sets. The mean correlation is
 8      close to zero; however, the standard deviation of approximately 0.4 resulted from the fact that
 9      some locations were highly positively correlated, while others had low negative correlations.
10      A possible explanation for a portion of this large variation in correlation between PM and ozone,
11      may arise from the difference in number of years of data at each site, and the influence of motor
12      vehicle traffic emissions of PM and NO. For example, Bernard et al. (1999) report that ozone
13      and NO2 in urban Montpellier, France had a highly negative correlation of r = -0.96,p < 0.01 in
14      the short time period studied (November 13 -20, 1995) as expected from the rapid gas  phase
15      reduction of ozone by NO. Consequently, if the U.S. monitoring stations for both PM and ozone
16      are close to local traffic emissions that have the same diurnal pattern, then the PM may be a
17      surrogate for accompanying NO emissions that would reduce the correlation of ozone with the
18      measured PM.
19           These annual correlations between PM and criteria gaseous pollutants are relevant with
20      respect to epidemiologic studies which analyze PM and health data over complete years of
21      record. U.S. Environmental Protection Agency (1996) and Chen et al. (1999) discuss the
22      variations in these gas-PM correlations between the summer and winter periods. Because of the
23      different seasonal emissions and weather patterns, summer corresponds to a period of ozone
24      maxima, and the primary combustion products (CO,  SO2) have their maxima during the winter,
25      which often leads to seasonal variations  of correlation.  Therefore the results cited above may not
26      be applicable to epidemiologic studies that analyze data by season, and a non-significant annual
27      correlation of order zero may mask the presence of significant  positive and negative correlations
28      in the different seasons of the year.
29           In summary it appears that there is no general rule that applies to PM and the gaseous
30      criteria pollutants. The possibility of confounding of a relation to ambient PM concentration by a


        October 1999                             5-88       DRAFT-DO NOT QUOTE OR CITE

-------
 1      gaseous ambient pollutant with a positive correlation in time to the ambient PM should be
 2      considered in epidemiologic studies.
 3
 4
 5      5.15  CONFOUNDING BY INDOOR PM
 6           It has been hypothesized that human exposure to indoor generated PM may be correlated
 7      with the ambient PM so that the health effects associated with the ambient PM concentration are
 8      in fact caused by the indoor generated PM and not the ambient PM (Crandall et al., 1996; Vedal,
 9      1997). "For example, if study subjects closed their windows on days with higher levels of
10      pollution, exposure to indoor pollutants might increase and actually be responsible for the
11      increase in adverse health outcomes, and therefore confound the particle and health association"
12      (Vedal, 1997).  However this scenario may not apply to the situations in these studies because the
13      ambient PM was virtually always below the PM NAAQS, and there were no episode warnings
14      issued to the public to stay indoors to avoid high ambient PM pollution.
15           Hoek et al. (1989) reported respirable PM data (96% efficient collection of PM with a 4//m
16      AD) indoors and outdoors before-during-after a moderate air pollution episode in the
17      Netherlands in  1985, for which no smog warnings were issued.  They found that indoor PM
18      increased and then fell similarly in homes with 0, 1 and 2 smokers, in phase with the increase in
19      ambient PM during the episode. Because Dutch homes with smokers have higher air exchange
20      rates than homes without  smokers (Lebret, 1985), Hoek et al. (1989) reasoned that "the increases
21      in indoor concentrations observed during the episode were largely due to penetration of outdoor
22      air, and not due to  decreased ventilation, leading to increased concentrations of pollutants
23      generated indoors, during the episode."
24           Mage et al. (1999) also tested this hypothesis of a decrease in air exchange rate leading to
25      higher indoor PM from indoor sources by using the PTEAM data set from Riverside,  CA
26      (Clayton et al.,  1993). When the PTEAM study was conducted in 1990, Riverside, CA was the
27      region of the country with the highest annual average concentration of ambient PM10.  The
28      PTEAM study utilized a constant emission source of perfluorocarbon tracer (PFT) and integrated
29      charcoal PFT collectors for estimating air exchange rates. The measurements of the collected
30      PFT were used to estimate the average number of air exchanges per hour (a) in each home during
31      the resident's PM sampling period.  Figure 5-21 shows that the measured air exchange rate had
        October 1999                              5-89        DRAFT-DO NOT QUOTE OR CITE

-------
           0)
           CD
           0)
           0)
           I
           o
           X
           LJJ
              0
                0
                                                                     R2 = 0.001
                       .
                               * *
100
200
300
400
500
600
                           Outdoor PM  Concentration at the Home
                                       10
      Figure 5-21.  Air-exchange rate versus outdoor PM10 concentration at the home in the
                   PTEAM study.
      Source: Mage et al. (1999).
1     no statistically significant relation to the ambient PM10 during the daytime (r = - 0.04).
2     Furthermore, each subject kept a time-activity diary so that the fraction of the time spent indoors
3     and outdoors has been computed for each person. Ozkaynak et al. (1996a) have determined the
4     deposition parameter (A=0.65/h) and the penetration parameter (P ~ 1) for PM10 from these very
5     same PTEAM data. Consequently we can estimate the exposure concentration of PM10 generated
6     by indoor sources and personal activities (Ei) as,
                              Ei = E - Co (a + ky) /(a + k)
                                                            (5-7)
      Where E is the measured total personal exposure; Co (a + ky)/(a + K) is the estimated personal
      exposure (Ea) to PM10 of ambient origin that also penetrates indoors (P ~ 1).
      October 1999
                    5-90
                     DRAFT-DO NOT QUOTE OR CITE

-------
 1           Figure 5-22 shows the quantity Ei for each subject vs the ambient concentration of PM10
 2      (Co) that was measured outside their home at the same time.  The regression has a slightly
 3      negative slope which is not significantly different from zero.  Note the negative values for Ei in
 4      Figure 5-22 may be caused by the exposure and concentration measurement errors, and errors in
 5      the estimated parameters P, a and k.  This suggests, at least for that Riverside cohort, that the
 6      personal exposure to a PM10 concentration of non-ambient origin does not appear to be positively
 7      correlated with the ambient PM10 concentration measured at that same time. We propose that
 8      this is the underlying general relationship between non-ambient generated PM and ambient PM
 9      concentration.
10           In order for there to be a positive correlation between exposure to indoor generated PM (Ei)
11      and ambient PM (Co or Ca), people would have to smoke more cigarettes, dust and vacuum
12      more, cook more, and perform more hobby-type activities indoors on days with higher ambient
13      PM concentrations than on days with lower ambient PM concentrations. We expect that, where
14      air quality standards are being met, people make their decisions to perform these personal
15      activities without any conscious or unconscious consideration of the ambient PM concentration
16      which is unknown to them (Hoek et al., 1989).  Thus, the amount of non-ambient PM that people
17      generate through their daily activities is expected to be independent of, and uncorrelated with, the
18      ambient PM concentration in their community.
19           Because the concentrations of PM due to such indoor sources were not observed to be
20      correlated with ambient PM concentration in either the Netherlands or Riverside, CA (the region
21      of the U.S. with the highest ambient PM concentration at the  time of the PTEAM study), it  is
22      reasoned that human exposure to non-ambient PM is likely to be independent of the
23      concentration of PM of ambient origin. Therefore it would be independent of the human
24      exposure to PM of ambient origin and not a confounder in the epidemiologic analysis. However,
25      these indoor generated PM species can be effects modifiers as discussed in the following
26      material.
27           Some PM species, and gaseous pollutants for which U.S. EPA has not established a
28      short-term National Ambient Air Quality Standard (NAAQS), may also play a role as acute effect
29      modifiers.  Such a condition may occur with long-term exposure to a pollutant with a chronic
30      effect without any apparent acute effects. That chronic exposure may make a subject more
31      susceptible to the acute effect of a given exposure to PM of ambient origin.  Chronic exposure to

        October 1999                             5-91         DRAFT-DO NOT QUOTE OR CITE

-------
             500
         o
         Q.
         X
         LU

         Q)
         "03
         E
             400-
             SOD-
         CD

         2
         
-------
 1      cigarette smoke, and chronic exposure to radon progeny and asbestos in PM of indoor origin,
 2      may cause malignant and non-malignant pulmonary disease which can make people more
 3      susceptible to the acute effects of ambient PM than otherwise healthy individuals (United
 4      Kingdom Ministry of Health, 1954). Therefore the daily concentration of such agents with
 5      chronic toxicity may be a surrogate for an underlying susceptibility to acute effects of ambient
 6      PM exposure in older subjects who lived in the monitored residence for a long period of time and
 7      who developed an underlying pulmonary insufficiency.
 8           There are also possible short-term effects of exposure to "nuisance and inert" PM,
 9      occupationally generated PM, and unmonitored ambient PM in the upper portion of the coarse
10      mode of PM  with an AD > 10 //m (see Figure 5-1).  PM species with biogenic or allergenic
11      properties could irritate and sensitize the lung to the ambient PM10 (see Section 7.2 of U.S.
12      Environmental Protection Agency, 1996) or increase the residence time of ambient PM10in the
13      lung. For example, if alveolar macrophages removed non-ambient particles and ambient
14      particles of AD > 10 //m, instead of deposited ambient PM10 species, the rate of pulmonary
15      clearance of the deposited ambient PM10 particles would be decreased.
16           The presence of effect-modification on mortality and morbidity is difficult to discern except
17      perhaps in animal testing where controlled dose-response studies can be used to pick up
18      modifications of responses to ambient PM exposures in the presence and  absence of candidate
19      agents. In epidemiologic studies, these effects would appear as statistically significant
20      interaction terms, but at the present time such interaction effects have not been reported (see
21      Chapter 12 of U.S. Environmental Protection Agency, 1996). The general-linearized-model
22      (GLM) described by Liang and Zeger (1986) has been used for most all the epidemiologic studies
23      discussed in Chapter 7 of this document. The GLM implies that the effect of exposure to
24      2 //g/m3 of ambient PM is twice the effect of exposure to 1 //g/m3 of the same mixture of ambient
25      PM, and the effect of an exposure to a mixture of 1 //g/m3 of ambient PM plus 1 //g/m3 of
26      non-ambient PM is equal to the effect of exposure to 1 //g/m3 of the same ambient PM mixture,
27      plus the effect of an exposure to 1 //g/m3 of the same non-ambient PM mixture.
28           Semi-volatile constituents of PM exist in the atmosphere in a dynamic equilibrium between
29      their vapor and condensed phases.  For example, water vapor in an air mass at a given relative
30      humidity (RH) is in dynamic equilibrium with the water content of the aerosol PM within that air
31      mass. After a particle is collected, fluctuations in relative humidity cause fluctuations in the

        October 1999                             5-93        DRAFT-DO NOT QUOTE OR CITE

-------
 1      amount of the water adsorbed on the particle, so consequently, the Federal Reference Method for
 2      Particulate Matter (40CFR50 Appendices B, L, M, J) calls for filters to be weighed, following a
 3      24-h dessication and equilibration period within a temperature and humidity controlled room, to
 4      reduce the water content of the collected particles.
 5           Other species, such as some inorganic nitrates and organic compounds, have high enough
 6      vapor pressures, in the range of ambient temperatures, that they, like water, can coexist in both
 7      the vapor and aerosol phase. Consequently, a variable fraction of such species may be lost from
 8      the collected mass on filters during equilibration.  Continuous monitors must also remove water
 9      associated with particles (particle-bound water). For example, the TEOM® monitor is operated
10      with the filter at 50°C to remove particle-bound water.  The TEOM® can record negative PM if
11      mass volatilizes from the filter collecting the PM, faster than new PM adheres to it.
12           Because ambient concentrations of these semi-volatile species are not currently measured,
13      and personal exposures to these constituents have not been reported in the literature, it is not
14      known whether these effervescent PM species could contribute to the health effects associated
15      with the non-volatile PM mass that is represented by the PM data in the epidemiology literature.
16      Although one might not expect the health effects associated with the non-volatile PM mass to be
17      caused by the unmeasured semi-volatile PM species, there may be a correlation between these
18      two masses so the captured non-volatile PM mass may in some aspects be a surrogate for the
19      mass of semi-volatile compounds that escaped from the collected PM.
20           In summary, this discussion supports the conclusion that the average concentrations of PM
21      measured at a community ambient monitoring station are reasonably good surrogate-estimates of
22      the average concentrations of PM of ambient origin to which people residing in that community
23      are exposed,  more so for fine mode PM than coarse mode PM. This  agreement between
24      exposure to ambient PM and ambient PM concentrations supports the plausibility of PM of
25      ambient origin, or a constituent thereof, being responsible for the fluctuations of health effects
26      that are correlated with fluctuations of ambient PM concentrations. The size distributions, the
27      chemical compositions,  and the related toxicities per unit mass of ambient PM may vary from
28      city to city in a season, and from season to season in a city.  In addition, different varieties of
29      climates between cities may influence the amount of time people spend outdoors and the air
30      exchange rates between their homes and the outdoor air, resulting in some cities having higher
31      percentages of the ambient PM penetrating into the residences than the others (Gamble, 1998).

        October 1999                             5-94       DRAFT-DO NOT QUOTE OR CITE

-------
 1     The resulting inter-urban differences of exposure to PM of ambient origin may explain a portion
 2     of the variance between the observed differences in mortality/morbidity per unit concentration of
 3     ambient PM that are reported in the literature and are discussed in Chapter 7 of this document.
 4
 5
 6     5.16  IMPLICATIONS OF THE AMBIENT PM EXPOSURE
 7            RELATIONSHIPS FOR EPIDEMIOLOGIC ANALYSIS AND
 8            A SUMMARY OF THE CHAPTER CONCLUSIONS
 9          The main conclusion of this chapter is that (a) exposures to PM of ambient origin are highly
10     correlated with ambient PM concentrations; and; (b) exposures to emissions of PM of
11     non-ambient origin (e.g., tobacco smoke, residential activity, occupational activity) have very
12     low correlations with ambient PM concentrations.  Therefore, it follows that the exposures to the
13     concentrations of PM produced by non-ambient sources (e.g., indoor and/or occupational) are
14     also uncorrelated with exposures to PM of ambient origin and the corresponding ambient PM
15     concentrations. Consequently the finding that personal exposures to total PM (both ambient
16     origin and non-ambient origin) are uncorrelated with ambient PM concentrations is not
17     important. This is because the acute health effects, if any,  caused by the exposure to
18     non-ambient PM will have an equally low correlation with the acute health effects created by the
19     personal exposure to PM of ambient origin.
20          As described in Section 5.14, ambient PM concentrations in the U.S. have a wide range of
21     correlations in time with the ambient gaseous criteria pollutants that may produce or influence
22     cardiac and pulmonary effects.  It appears that there is no a priori reason to neglect their ability
23     to confound the ambient PM vs health effect relationships. Therefore, these correlations need to
24     be evaluated at each location where an epidemiologic study is conducted  to determine whether
25     they are significantly different from zero. The chapter conclusions are as follows:
26     1.  Human exposure to PM of ambient origin for individuals in a community is often highly
27         correlated (R2 > 0.5) in time with concentrations of PM of ambient origin of the same size as
28         measured in that community.
29     2.  The longitudinal correlation coefficient for the ambient concentration of fine PM
30         (AD < 2.5 //m) with personal exposure to ambient fine PM is greater  than the corresponding
        October 1999                              5-95        DRAFT-DO NOT QUOTE OR CITE

-------
 1         correlation for the coarse fraction of ambient PM (2.5 //m < AD < 10 //m) as shown by
 2         studies of ambient sulfate concentrations and sulfate exposures.
 3      3.  People in a community surrounding an ambient monitoring station, over time, are exposed to
 4         relatively similar mixtures and concentrations of ambient PM2 5.
 5      4.  People in a community are exposed to widely different mixtures and concentrations of
 6         non-ambient PM due to the diversity of smoking habits, personal activities such as  hobbies,
 7         residential furnishings and appliances,  and varying occupations.
 8      5.  Exposures to PM of indoor origin appear to be uncorrelated with exposures to PM of ambient
 9         origin.
10      6.  The correlation of a single individual's sequence of daily personal exposures to total PM and
11         ambient PM concentrations will be greater than the correlation that would occur had a
12         different person been monitored on each of the same days [e.g., one person monitored
13         consecutively for n days vs sequentially monitoring n different people, each for one day, over
14         ndays].
15      7.  Ambient PM in the U.S. has average annual  correlations with the ambient gaseous pollutants
16         CO, ozone, NO2, and SO2 of order r = 0.25 with a standard deviation of order 0.25.
17      8.  Although exposures to PM from indoor sources and occupational activities may not be
18         correlated with ambient PM concentrations,  these non-ambient PM species may possibly act
19         as effect modifiers by making subjects  more or less susceptible to exposure to PM of ambient
20         origin.
21      9.  There are only limited data available, from non-probability samples, to evaluate how well the
22         exposures to PM of ambient origin for  susceptible subgroups correlates with the ambient PM
23         concentrations of similar AD size range as measured in their community.
24           In conclusion, day-to-day variation in the ambient concentration of fine particles is a good
25      surrogate for the day-to-day variation in the community average personal exposure to fine
26      particles of ambient origin.  This relationship is  not as clear for coarser particles (PM10  - PM2 5)
27      which do have more local sources. This supports the plausibility of ambient PM25 concentration
28      as a surrogate measure for personal exposure to  PM25 of ambient origin in time series studies,
29      and provides the linkage necessary to more completely evaluate the impact of ambient PM
30      regulation under Title I of the Clean Air Act Amendments.
31

        October 1999                               5-96       DRAFT-DO NOT QUOTE OR CITE

-------
  1       REFERENCES

  2       Abbey, D. E.; Nishino, N.; McDonnell, W. F.; Burchette, R. J.; Knutsen, S. F.; Beeson, W. L.; Yang, J. X. (1999)
  3             Long-term inhalable particles and other air pollutants related to mortality in nonsmokers. Am. J. Respir. Crit.
  4             Care Med. 159:373-382.
  5       Abt, E.; Suh, H. H.; Allen, G.; Koutrakis, P. (1999a) Characterization of indoor particle sources: a study conducted
  6             in the metropolitan Boston area. Environ. Health Perspect: submitted.
  7       Abt, E.; Suh, H. H.; Catalano, P.; Koutrakis, P. (1999b) The relative contribution of outdoor and indoor particle
  8             sources to indoor concentrations. Environ. Sci. Technol.: submitted.
  9       Adams, W. C. (1993) Measurement of breathing rate and volume in routinely performed daily activities [final
10             report]. Sacramento, CA: California Environmental Protection Agency, Air Resources Board; contract no.
11             A033-205.
12       Aim, S.; Jantunen, M. J.; Vartiainen, M. (1999) Urban commuter exposure to particle matter and carbon monoxide
13             inside an automobile. J. Exposure Anal. Environ. Epidemiol. 9: 237-244.
14       Alzona, J.; Cohen, B. L.; Rudolph, H.; Jow, H. N.; Frohliger, J. O. (1979) Indoor-outdoor relationships for airborne
15             particulate matter of outdoor origin. Atmos.  Environ. 13: 55-60.
16       American Conference of Governmental Industrial Hygienists. (1996) Threshold limit values for chemical
17             substances and physical agents: biological exposure indices. Cincinnati, OH: American Conference of
18             Governmental and Industrial Hygienists.
19       Anuszewski, J.; Larson, T. V.; Koenig, J. Q. (1998) Simultaneous indoor and outdoor particle light-scattering
20             measurements at nine homes using a portable nephelometer. J. Exposure Anal. Environ. Epidemiol. 8:
21             483-493.
22       Armstrong, B. K.; White, E.;  Saracci, R., eds. (1992) Principles of exposure measurement in epidemiology. New
23             York, NY: Oxford University Press. (Monographs in epidemiology and biostatistics: v. 21).
24       Ayers, G. P.; Keywood, M. D.; Gras, J. L. (1999) TEOM vs. manual gravimetric methods for determination of
25             PM2.5 aerosol mass concentrations. Atmos. Environ. 33: 3717-3721.
26       Bahadori, T. (1998) Human exposure assessment: relationship between outdoor, indoor, and personal measurements
27             [dissertation]. Boston, MA: Harvard University.
28       Bahadori, T.;  Suh, H. H.; Rojas-Bracho, L.; Koutrakis, P. (1999) Personal exposure to particulate matter of
29             individuals with chronic obstructive pulmonary disease (COPD). Submitted.
30       Bernard, N. L.; Gerber, M. J.; Astre, C. M.; Saintot, M. J. (1999) Ozone measurement with passive samplers:
31             validation and use for ozone pollution assessment in Montpellier, France. Environ. Sci. Technol. 33:
32             217-222.
33       Beyea, J.; Hatch, M. (1999) Geographic exposure modeling: a valuable extension of geographic information
34             systems for use in environmental epidemiology. Environ. Health Perspect. Suppl. 107(1): 181-190.
35       Binder,  R. E.; Mitchell, C. A.; Hosein, H. R.; Bouhuys, A. (1976) Importance  of the indoor environment in air
36             pollution exposure. Arch. Environ. Health 31: 277-279.
37       Bohne, J. E., Jr.; Cohen, B. S. (1985) Aerosol resuspension from fabric: implications for personal monitoring in the
38             beryllium industry. Am. Ind. Hyg. Assoc. J.  46: 73-79.
39       Brauer,  M.; Koutrakis, P.; Spengler, J. D. (1989) Personal exposures to acidic aerosols and gases. Environ. Sci.
40             Technol. 23: 1408-1412.
41       Brunekreef, B. (1999) All but quiet on the particulate front [editorial]. Am. J. Respir. Crit. Care Med. 159: 354-356.
42       Burton, R. M.; Suh, H. H.; Koutrakis, P. (1996) Spatial variation in particulate concentrations within metropolitan
43             Philadelphia. Environ. Sci. Technol. 30: 400-407.
44       Buzorius, G.; Hameri, K.; Pekkanen, J.; Kulmala, M. (1999) Spatial variation of aerosol number concentration in
45             Helsinki city. Atmos. Environ. 33: 553-565.
46       California Environmental Protection Agency. (1998) Measuring concentrations of selected air pollutants inside
47             California vehicles [final report]. Sacramento, CA: Air Resources Board, Research Division; contract no.
48             95-339
49       Carrothers, T. J.; Evans, J. S. (1999) Assessing the impact of differential measurement error on estimates of fine
50             particle mortality.  J. Air Waste Manage. Assoc.:  accepted.
51       Cha, Q.; Chen, Y.-Z.; Du, Y.-X.  (1996) Influence of indoor air pollution on quality of outdoor air in city of
52             Guangzhou, China. In: Indoor air '96: the 7th international conference on indoor air quality, volume I; July;
53             Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 1131-1135.
         October 1999                                    5-97          DRAFT-DO NOT QUOTE OR CITE

-------
  1       Chen, C.; Chock, D. P.; Winkler, S. L. (1999) A simulation study of confounding in generalized linear models for
  2             air pollution epidemiology. Environ. Health Perspect. 107: 217-222.
  3       Clayton, C. A.; Perritt, R. L.; Pellizzari, E. D.; Thomas, K. W.; Whitmore, R. W.; Wallace, L. A.; Ozkaynak, H.;
  4             Spengler, J. D. (1993) Particle total exposure assessment methodology (PTEAM) study: distributions of
  5             aerosol and elemental concentrations in personal, indoor, and outdoor air samples in a southern California
  6             community. J. Exposure Anal. Environ. Epidemiol. 3: 227-250.
  7       Code of Federal Regulations. (1994) National primary and secondary ambient air quality standards; definitions. C.
  8             F.R. 40: part 50, §50.1.
  9       Code of Federal Regulations. (1998) Occupational safety and health standards, subpart Z—toxic and hazardous
10             substances. Table Z-3, mineral dusts. C. F. R. 29: §1910.1000.
11       Cohen, B. S.; Positano, R. (1986) Resuspension of dust from work clothing as a source of inhalation exposure. Am.
12             Ind. Hyg. Assoc. J. 47: 255-258.
13       Crandall, R. W.; Rueter, F. H.; Steger, W. A. (1996) Clearing the air: EPA's self-assessment of clean-air policy.
14             Regulation 19(4): 35-46.
15       Delfino, R. J.; Zeiger, R. S.; Seltzer, J. M.; Street, D. H. (1998) Symptoms in pediatric asthmatics and air pollution:
16             differences in effects by symptom severity, anti-inflammatory medication use and particulate averaging time.
17             Environ. Health Perspect. 106: 751-761.
18       Duan, N. (1982) Models for human exposure to air pollution. Environ. Int. 8: 305-309.
19       Dubowsky, S. D.; Wallace, L.; Buckley, T. J. (1999) The contribution of traffic to indoor concentrations of
20             polycyclic aromatic hydrocarbons (PAH). J. Exposure Anal. Environ. Epidemiol.: accepted.
21       Federal Trade Commission. (1994) Tar, nicotine, and carbon monoxide of the smoke of 93 3 varieties of domestic
22             cigarettes. Washington, DC: Federal Trade Commission.
23       Gamble, J. F. (1998) PM2 5 and mortality in long-term prospective cohort studies: cause-effect or statistical
24             associations? Environ. Health Perspect. 106:  535-549.
25       Gamble, J. (1999) Reply to Kunzli and Tager regarding causality in PM 2 5 cohort studies [author's reply]. Environ.
26             Health Perspect. 107: A235-A236.
27       Hoek, G.; Brunekreef, B.; Hofschreuder, P. (1989) Indoor exposure to airborne particles and nitrogen dioxide
28             during an air pollution episode. JAPCA 39: 1348-1349.
29       Janssen, N. (1998) Personal exposure to airborne particles: validity of outdoor concentrations as a measure of
30             exposure in time series studies [dissertation]. Wageningen, Netherlands: Landbouwuniversiteit Wageningen.
31       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1995) A relation between personal and ambient PM 10.
32             Epidemiology 6(suppl.): S45.
33       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1997) Childhood exposure to PM 10: relation between
34             personal, classroom, and outdoor concentrations. Occup. Environ. Med. 54: 888-894.
35       Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H.; Mensink, L; Zuidhof, A. (1998a) Personal sampling of
36             particles in adults: relation among personal, indoor, and outdoor air concentrations. Am. J. Epidemiol.  147:
37             537-547.
38       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1998b) Personal sampling of airborne particles: method
39             performance and data quality. J. Exposure Anal. Environ. Epidemiol. 8: 37-49.
40       Janssen, L. H. J. M.; Buringh, E.; van der Meulen, A.; van den Hout, K.  D. (1999a) A method to estimate the
41             distribution of various fractions of PM10 in ambient air in the Netherlands. Atmos. Environ. 33: 3325-3334.
42       Janssen, N. A. H.; Hoek, G.; Harssema, H.; Brunekreef, B. (1999b) Personal exposure to fine particles in children
43             correlates closely with ambient fine particles. Arch. Environ. Health  54: 95-101.
44       Janssen, N. A. H.; Hoek, G.; Brunekreef, B.; Harssema, H. (1999c) Mass concentration and elemental composition
45             of PM10 in classrooms. Occup. Environ. Med. 56: 482-487.
46       Jantunen, M.  J.; Hanninen, O.; Katsouyanni, K.; Knoppel, H.; Kuenzli, N.; Lebret, E.; Maroni, M.;  Saarela, K.;
47             Sram, R.; Zmirou, D. (1998) Air pollution exposure in European cities: the " EXPOLIS" study. J. Exposure
48             Anal. Environ. Epidemiol. 8: 495-518.
49       Jantunen et al. (2000) [Manuscript in preparation.]
50       Jedrychowski, W.; Flak, E. (1998) Separate and combined effects of the outdoor and indoor air quality on chronic
51             respiratory symptoms adjusted for allergy among preadolescent children. Int. J. Occup. Med. Environ. Health
52             11:19-35.
53       Jenkins, R. A.; Palausky, A.; Counts, R. W.; Bayne, C. K.; Dindal, A. B.; Guerin, M. R. (1996a) Exposure to
54             environmental tobacco smoke in sixteen cities in the United States as determined by personal breathing zone
55             air sampling. J. Exposure Anal. Environ. Epidemiol. 6:  473-502.


         October  1999                                    5-98          DRAFT-DO NOT QUOTE OR CITE

-------
  1       Jenkins, R. A.; Palausky, M. A.; Counts, R. W.; Guerin, M. R.; Dindal, A. B.; Bayne, C. K. (1996b) Determination
  2             of personal exposure of non-smokers to environmental tobacco smoke in the United States. Lung Cancer
  3             14(suppl. 1): S195-S213.
  4       Kamens, R.; Jang, M.; Chien, C.-J.; Leach, K. (1999) Aerosol formation from the reaction of a-pinene and ozone
  5             using a gas-phase kinetics-aerosol partitioning model. Environ. Sci. Technol. 33: 1430-1438.
  6       Keary, J.; Jennings, S. G.; O'Connor, T. C.; McManus, B.; Lee, M. (1998) PM 10 concentration measurements in
  7             Dublin City. Environ. Monitor. Assess. 52: 3-18.
  8       Keywood, M. D.; Ayers, G. P.; Gras, J. L.; Gillett, R. W.; Cohen, D. D. (1999) Relationships between size
  9             segregated mass concentration data and ultrafine particle number concentrations in urban areas. Atmos.
10             Environ. 33:2907-2913.
11       Kildes0, J.; Vallarino, J.; Spengler, J. D.; Brightman, H. S.; Schneider, T. (1999) Dust build-up on surfaces in the
12             indoor environment. Atmos. Environ. 33: 699-707.
13       Kleeman, M. J.; Cass, G. R. (1998) Source contributions to the size and composition distribution of urban
14             particulate air pollution. Atmos. Environ. 32: 2803-2816.
15       Klepeis, N. E.; Ott, W. R.; Switzer, P. (1996) A multiple-smoker model for predicting indoor air quality in public
16             lounges. Environ. Sci. Technol. 30: 2813-2820.
17       Klepeis, N. E.; Nelson, W. C.; Tsang, A. M.;  Robinson, J. P. Hern, S. C.; Engelmann, W. H.; Behar, J. V. (1999)
18             The national human activity pattern survey (NHAPS): data collection methodology and selected results. J.
19             Exposure Anal. Environ. Epidemiol.: submitted.
20       Koistinen, K. J.; Kousa, A.; Tenhola, V.; Hanninen, O.; Jantunen, M.  J. (1999) Fine particle (PM 25) measurement
21             methodology, quality assurance procedures and pilot results of the EXPOLIS study. J. Air Waste Manage.
22             Assoc.: accepted.
23       Kotchmar, D. J.; McMullen, T.; Hasselblad, V. (1987) Adequacy of a single monitoring site for defining mean
24             outdoor concentrations of fine particles in a demarcated residential community. JAPCA 37: 3 77-3 81.
25       Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. (1992) Source apportionment of indoor aerosols in Suffolk and
26             Onondaga Counties, New York. Environ. Sci. Technol. 26: 521-527.
27       Kulmala, M.; Asmi, A.; Pirjola, L. (1999) Indoor air aerosol model: the effect of outdoor air, filtration and
28             ventilation on indoor concentrations. Atmos.  Environ. 33: 2133-2144.
29       Kiinzli, N.; Tager, I. B. (1999) Comments on "PM 2 5 and mortality in long-term prospective cohort studies:
30             cause-effect or statistical associations?" Environ. Health Perspect. 107:  A234-A235.
31       Larssen, S.; T0nnesen, D.; Clench-Aas, J.; Aarnes, M. J.; Arnesen, K. (1993) A model for car exhaust exposure
32             calculations to investigate health effects of air pollution. Sci. Total Environ. 134: 51-60.
33       Leaderer, B. P.; Naeher, L.; Jankun, T.; Balenger, K.; Holford, T. R.; Toth, C.; Sullivan, J.; Wolfson, J. M.;
34             Koutrakis, P. (1999a) Indoor, outdoor, and regional summer and winter concentrations of PM 10, PM25, SO42",
35             H+, NH4+, NO3", NH3, and nitrous acid in homes with and without kerosene space heaters. Environ. Health
36             Perspect. 107:223-231.
37       Leaderer, B. P.; Balenger, K.; Naeher, L.; Triche, E. W.; Hellenbrand, K.; Holford, T. R.; Zheng, H.; Bracken, M.
38             B.; Beckett, W. S. (1999b) Daily summer respiratory symptoms in infants associated with exposure to PM 10,
39             PM2.5, sulfate and ozone. Submitted.
40       Leaderer, B. P.; Balenger, K; Triche, E. W.; Hellenbrand, K.; Holford, T. R.;  Zheng, H.; Bracken, M. B.; Beckett,
41             W.  S. (1999c) Daily winter respiratory symptoms in infants associated with exposure to particulate and
42             sulfate air pollution. Submitted.
43       Lebret, E. (1985) Air pollution in Dutch homes: an exploratory study in environmental epidemiology. Wageningen,
44             The Netherlands: Department of Air Pollution, Department of Environmental and Tropical Health; report
45             R-138; report 1985-221.
46       Lessler, J.  T.; Kalsbeek, W. D. (1992) Nonsampling error in surveys. New York, NY: John Wiley & Sons,  Inc.;
47             p. 7.
48       Liang, K-Y.; Zeger, S. L. (1986) Longitudinal data analysis using generalized linear models. Biometrika 73: 13-22.
49       Lillquist, D. R.; Lee, J. S.; Ramsay, J. R.; Boucher, K. M.; Walton, Z. L.; Lyon, J. L. (1998) A comparison of
50             indoor/outdoor PM 10 concentrations measured at three hospitals and a centrally located monitor in Utah.
51             Appl. Occup. Environ. Hyg. 13: 409-415.
52       Lioy, P. J.; Waldman, J. M.; Buckley, T.; Butler, J.;  Pietarinen, C. (1990) The personal, indoor and outdoor
53             concentrations of PM-10 measured in an industrial community during the winter. Atmos. Environ. Part B 24:
54             57-66.
         October 1999                                   5-99         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Litzistorf, G.; Guillemin, M. P.; Buffat, P.; Iselin, F. (1985) Influence of human activity on the airborne fiber level
  2             in paraoccupational environments. J. Air Pollut. Control Assoc. 35: 836-837.
  3       Mage, D. T. (1980) The statistical form for the ambient particulate standard annual arithmetic mean vs annual
  4             geometric mean. J. Air Pollut. Control Assoc. 30: 796-798.
  5       Mage, D. T. (1983) Public health aspects of air quality surveillance. Public Health Rev. 11: 5-54.
  6       Mage, D. T. (1985) Concepts of human exposure assessment for airborne particulate matter. Environ. Int. 11:
  7             407-412.
  8       Mage, D. T. (1998) How much PM of ambient origin are people exposed to? In: Measurement of toxic and related
  9             air pollutants. Volume II: proceedings of a specialty conference; September; Gary, NC. Pittsburgh, PA: Air &
10             Waste Management Association; pp. 666-681. (A&WMA publication VIP-85).
11       Mage, D. T.; Buckley, T. J. (1995) The relationship between personal exposures and ambient concentrations of
12             particulate matter. Presented at: 88th annual meeting of the Air & Waste Management Association; June; San
13             Antonio, TX. Pittsburgh, PA: Air & Waste Management Association; paper no. 95-MP18.01.
14       Mage, D. T.; Ott, W. R. (1996) Accounting for nonuniform mixing and human exposure in indoor environments. In:
15             Tichenor, B. A., ed. Characterizing sources of indoor air pollution and related sink effects. Philadelphia, PA:
16             American Society for Testing and Materials; pp. 263-278.
17       Mage, D. T.; Wilson, W.; Hasselblad, V.; Grant, L. (1999) Assessment of human exposure to ambient particulate
18             matter. J. Air Waste Manage. Assoc.: in press.
19       Marple, V. A.; Rubow, K. L.; Turner, W.; Spengler, J. D. (1987) Low flow rate sharp cut impactors for indoor air
20             sampling: design and calibration. JAPCA 37: 1303-1307.
21       Matthews, T. G.; Thompson, C. V.; Wilson, D. L.; Hawthorne, A. R.; Mage, D. T. (1989) Air velocities inside
22             domestic environments: an important parameter in the study of indoor air quality and climate. Environ.  Int.
23             15: 545-550.
24       McBride, S. J.; Ferro, A.; Ott, W.; Switzer,  P.; Hildemann, L. M. (1999) Investigations of the proximity effect for
25             pollutants in the indoor environment. J. Exposure Anal. Environ. Epidemiol.: accepted.
26       Michaels, R. A. (1997) Particulate matter policy [letter]. Science (Washington, DC) 278: 1,696.
27       Michaels, R. A. (1998) Permissible daily airborne particle mass levels encompass brief excursions to the "London
28             Fog" range, which may contribute to daily mortality and morbidity in communities. Appl. Occup. Environ.
29             Hyg.  13:385-394.
30       Monn, C. (1999) [Personal communication with D. Mage concerning article appearing in Science of the Total
31             Environment 208: 15-21 (1997)]. Zurich, Switzerland: Institute for Hygiene und Applied Physiology; June 2.
32       Monn, C.; Junker, M. (1999) Indoor/outdoor ratios, deposition and resuspension of fine and coarse particles. In:
33             Indoor air '99; August; Edinburgh, United Kingdom.
34       Monn, C.; Fuchs, A.; Hogger, D.; Junker, M.; Kogelschatz, D.; Roth, N.; Wanner, H.-U. (1997) Particulate matter
35             less than 10 ^ (PM10) and fine particles less than 2.5 ^ (PM2 5): relationships between indoor, outdoor and
36             personal concentrations. Sci. Total Environ. 208: 15-21.
37       Morris, R. D.; Naumova, E. N. (1998) Carbon monoxide and hospital admissions for congestive heart failure:
38             evidence of an increased effect at low temperatures. Environ. Health Perspect.  106: 649-653.
39       Murray, D. M.; Burmaster, D. E. (1995) Residential air exchange rates in the United States: empirical and estimated
40             parametric distributions by season and climatic region. Risk Anal. 15: 459-465.
41       National Air Pollution Control Administration. (1969) Air quality criteria for particulate matter. Washington, DC:
42             U.S. Department of Health, Education, and Welfare, Public Health Service; NAPCA publication no. AP-49.
43             Available from: NTIS, Springfield, VA; PB-190251/BA.
44       National Research Council. (1991) Human exposure assessment for airborne pollutants: advances and opportunities.
45             Washington, DC: National Academy of Sciences.
46       Neas, L.  M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr.; Speizer, F. E. (1994) Concentration of
47             indoor particulate matter as a determinant of respiratory health in children. Am. J. Epidemiol. 139:
48             1088-1099.
49       Nieuwenhuijsen, M. J.; Noderer, K.  S.; Schenker, M. B.; Vallyathan, V.; Olenchock, S. (1999) Personal exposure to
50             dust, endotoxin and crystalline silica in California agriculture. Ann. Occup. Hyg. 43: 35-42.
51       Ogden, T. L.; Bartlett, I. W.; Purnell, C. J.; Wells, C. J.; Armitage, F.; Wolfson, H. (1993) Dust for cotton
52             manufacture: changing from static to personal sampling. Ann. Occup. Hyg. 37: 271-285.
53       Ormstad, H.; Gaarder, P. L; Johansen, B. V. (1997) Quantification and characterisation of suspended particulate
54             matter in indoor air. Sci. Total Environ. 193: 185-196.
55       Ott, W. R.; Roberts, J. W. (1998) Everyday exposure to toxic pollutants. Sci. Am. 278(2): 86-91.


         October 1999                                   5-100         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Ott, W.; Langan, L.; Switzer, P. (1992) A time series model for cigarette smoking activity patterns: model validation
  2             for carbon monoxide and respirable particles in a chamber and an automobile. In: Measuring, understanding,
  3             and predicting exposures in the 21st century: proceedings of the conference; November 1991; Atlanta, GA. J.
  4             Exposure Anal. Environ. Epidemiol. 2(suppl. 2): 175-200.
  5       Owen, M. K.; Ensor, D. S.; Sparks, L. E. (1992) Airborne particle sizes and sources found in indoor air. Atmos.
  6             Environ. Part A 26: 2149-2162.
  7       Ozkaynak, H.; Xue, J.; Weker, R.; Butler, D.; Koutrakis, P.; Spengler, J. (1996a) The particle TEAM (PTEAM)
  8             study: analysis of the data: final report, volume III. Research Triangle Park, NC: U.S. Environmental
  9             Protection Agency, Atmospheric Research and Exposure Assessment Laboratory; report no.
10             EPA/600/R-95/098. Available from: NTIS, Springfield, VA; PB97-102495.
11       Ozkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins, P. (1996b) Personal exposure to airborne
12             particles and metals: results from the particle TEAM study in Riverside, California. J. Exposure Anal.
13             Environ. Epidemiol. 6: 57-78.
14       Park, J.-H.; Spengler, J.  D.; Yoon, D.-W.; Dumyahn, T.; Lee, K.; Ozkaynak, H. (1998) Measurement of air
15             exchange rate of stationary vehicles and estimation of in-vehicle exposure. J. Exposure Anal. Environ.
16             Epidemiol. 8: 65-78.
17       Pellizzari, E. D.; Clayton, C. A.; Rodes, C. E.; Mason, R. E.; Piper, L. L.; Fort, B.; Pfeifer, G.; Lynam, D. (1999)
18             Particulate matter and manganese exposures in Toronto, Canada. Atmos. Environ. 33: 721-734.
19       Phillips, K.; Howard, D. A.; Browne, D.; Lewsley, J. M. (1994) Assessment of personal exposures to environmental
20             tobacco smoke in British nonsmokers. Environ. Int. 20: 693-712.
21       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G. (1996) Assessment of air quality in Stockholm by personal
22             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Scand. J.
23             Work Environ. Health 22(suppl. 1): 1-24.
24       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G.; Huici, A. (1997a) Assessment of air quality in Barcelona by
25             personal monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke.
26             Environ. Int. 23: 173-196.
27       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1997b) Assessment of air quality in Turin by personal
28             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
29             23:851-871.
30       Phillips, K.; Bentley, M. C.; Howard, D. A.; Alvan, G. (1998a) Assessment of air quality in Paris by personal
31             monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environ. Int.
32             24:405-425.
33       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1998b) Assessment of environmental tobacco smoke and
34             respirable suspended particle exposures for nonsmokers in Lisbon by personal monitoring. Environ. Int. 24:
35             301-324.
36       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1998c) Measured exposures by personal monitoring for
37             respirable suspended particles and environmental tobacco  smoke of housewives and office workers resident
38             in Bremen, Germany. Int. Arch.  Occup. Environ. Health 71: 201-212.
39       Phillips, K.; Howard, D. A.; Bentley, M. C.; Alvan, G. (1999) Assessment of environmental tobacco smoke and
40             respirable suspended particle exposures for nonsmokers in Basel by  personal monitoring. Atmos. Environ.
41             33: 1889-1904.
42       Pope, C. A.; Hill, R. W.; Villegas, G. M. (1999) Particulate air pollution and daily mortality on Utah's Wasatch
43             Front. Environ. Health Perspect: submitted.
44       Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment approaches to evaluate
45             respiratory health effects of particulate matter and nitrogen dioxide. J. Exposure Anal. Environ. Epidemiol. 1:
46             83-107.
47       Raunemaa, T.; Kulmala, M.; Saari, H.; Olin, M.; Kulmala, M. H. (1989) Indoor air aerosol model: transport indoors
48             and deposition of fine and coarse particles. Aerosol Sci. Technol. 11: 11-25.
49       Repace, J. L.; Lowrey, A. H. (1980) Indoor air pollution, tobacco smoke, and public health. Science (Washington,
50             DC) 208: 464-472.
51       Rojas-Bracho, L.; Suh, H. H.; Koutrakis, P. (1999) Relationship among personal, indoor, and outdoor fine and
52             coarse particle concentrations for individuals with COPD. Boston, MA: Harvard University, School of Public
53             Health.
54       Roorda-Knape, M. C.; Janssen, N. A. H.; De Hartog, J. J.; Van Vliet, P. H. N.; Harssema, H.; Brunekreef, B. (1998)
55             Air pollution from traffic in city districts near major motorways. Atmos. Environ. 32:  1921-1930.


         October 1999                                   5-101         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Scaperdas, A.; Colvile, R. N. (1999) Assessing the representativeness of monitoring data from an urban intersection
  2             site in central London, UK. Atmos. Environ. 33: 661-674.
  3       Schwartz, J.; Norris, G.; Larson, T.; Sheppard, L.; Claiborne, C.; Koenig, J. (1999) Episodes of high coarse particle
  4             concentrations are not associated with increased mortality. Environ. Health Perspect. 107: 339-342.
  5       Seinfeld, J. H.; Pandis, S. N. (1998) Atmospheric chemistry and physics: from air pollution to climate change. New
  6             York, NY: John Wiley & Sons, Inc.
  7       Sexton, K.; Spengler, J. D.; Treitman, R. D. (1984) Personal exposure to respirable particles: a case study in
  8             Waterbury, Vermont. Atmos. Environ. 18: 1385-1398.
  9       Shadwick, D.; Glen, G.; Smith, L. (1999) Correlations of particulate matter with gaseous pollutants. Research
10             Triangle Park, NC: U.S. Environmental Protection Agency; National Exposure Research Laboratory; Office
11             of Research and Development; October 1998.
12       Siegmann, K.; Scherrer, L.; Siegmann, H. C. (1999) Physical and chemical properties of airborne nanoscale
13             particles and how to measure the impact on human health. J.  Mol. Struct. (Theochem) 458: 191-201.
14       Smith, K. R.; Apte, M. G.; Ma, Y.; Wongsekiarttirat, W.; Kulkarni,  A. (1994) Air pollution and the energy ladder in
15             Asian cities. In: Tyler, S.; Sathaye, J.; Goldman, N., eds. Energy use in Asian Cities. Energy (Oxford)  19:
16             587-600.
17       Spengler, J. D.; Dockery, D. W.; Reed, M. P.; Tosteson, T.; Quinlan, P.  (1980) Personal exposures to respirable
18             particles. Presented at: 73rd annual meeting of the Air Pollution Control Association; June; Montreal, PQ,
19             Canada. Pittsburgh, PA: Air Pollution Control Association; paper no. 80-61.5b.
20       Spengler, J. D.; Treitman, R. D.; Tosteson, T. D.; Mage, D. T.; Soczek, M. L. (1985) Personal exposures to
21             respirable particulates and implications for air pollution epidemiology. Environ. Sci. Technol. 19: 700-707.
22       Sterling, T. D.; Rosenbaum, W. L.; Weinkam, J. J.  (1995) Concentration of indoor particulate matter as a
23             determinant of respiratory health in children [letter]. Am. J. Epidemiol. 141: 581-582.
24       Stewart, P. A.; Rice, C. (1990) A source of exposure data for occupational epidemiology studies. Appl. Occup.
25             Environ. Hyg. 5: 359-363.
26       Suh, H. H.; Koutrakis, P.; Spengler, J. D. (1993) Validation of personal  exposure models for sulfate and aerosol
27             strong acidity. J. Air Waste Manage. Assoc. 43:  845-850.
28       Tamura, K.; Ando, M.; Sagai, M.; Matsumoto, Y. (1996a) Estimation of levels of personal exposure to suspended
29             particulate matter and nitrogen dioxide in Tokyo. Environ. Sci. (Tokyo) 4: 37-51.
30       Tamura, K.; Ando, M.; Miyazaki, T.; Funasaka, K.; Kuroda, K. (1996b) Indoor concentrations and personal
31             exposure to particulate matter at roadside houses. In: Indoor air '96: the 7th international conference on
32             indoor air quality, volume I; July; Nagoya, Japan. Tokyo, Japan: Institute of Public Health; pp. 599-602.
33       Teschke, K.; Marion, S. A.; Vaughan, T. L.; Morgan, M. S.; Camp,  J. (1999) Exposures to wood dust in U.S.
34             industries and occupations, 1979 to 1997. Am. J. Ind. Med. 35: 581-589.
35       Thatcher, T. L.; Layton, D. W. (1995) Deposition, resuspension, and penetration of particles within a residence.
36             Atmos. Environ. 29: 1487-1497.
37       Trijonis, J.; Eldon, J.; Gins, J.; Berglund, G. (1980) Analysis of the  St. Louis RAMS ambient particulate data.
38             Volume I: final report. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air
39             Quality Planning and Standards; report no. EPA-450/4-80-006a. Available from: NTIS, Springfield, VA;
40             PB80-203359.
41       Tung, T. C. W.; Chao, C. Y. H.; Burnett, J. (1999) A methodology to investigate the particulate penetration
42             coefficient through building shell. Atmos. Environ. 33: 881-893.
43       U.S. Environmental Protection Agency. (1982) Air quality criteria for particulate matter and sulfur oxides. Research
44             Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
45             Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from:  NTIS, Springfield, VA; PB84-156777.
46       U.S. Environmental Protection Agency. (1996) Air quality criteria for particulate matter. Research Triangle Park,
47             NC: National Center for Environmental Assessment-RTF Office; report nos. EPA/600/P-95/001aF-cF. 3v.
48             Available from: NTIS, Springfield, VA; PB96-168224.
49       United Kingdom Ministry of Health. (1954) Mortality and morbidity during the London fog of December 1952.
50             London, United Kingdom: Her Majesty's Stationery Office. (Reports on public health and medical subjects
51             no. 95).
52       Vakeva, M.; Hameri, K.;  Kulmala,  M.; Lahdes, R.; Ruuskanen, J.; Laitinen, T. (1999) Street level versus rooftop
53             concentrations of submicron aerosol particles and gaseous pollutants in an urban street canyon. Atmos.
54             Environ. 33: 1385-1397.
55       Vedal, S. (1997)  Ambient particles and health: lines that divide. J. Air Waste Manage. Assoc. 47: 551-581.


         October 1999                                    5-102         DRAFT-DO NOT QUOTE OR CITE

-------
  1       Wallace, L. (1996) Indoor particles: a review. J. Air Waste Manage. Assoc. 46: 98-126.
  2       Wallace, L. (1999a) Correlations of personal exposure to particles with outdoor air measurements: a review of
  3             recent studies. Aerosol. Sci. Technol.: accepted.
  4       Wallace, L. (1999b) Penetration of PM [email to D. Mage]. Reston, VA: U.S. Environmental Protection Agency;
  5             September 1.
  6       Wallace, L.; Quackenboss, J.; Rodes, C. (1997) Continuous measurements of particles, PAH, and CO in an
  7             occupied townhouse in Reston, VA. In: Measurement of toxic and related air pollutants. Volume II:
  8             proceedings of a specialty conference; April-May; Research Triangle Park, NC. Pittsburgh, PA: Air & Waste
  9             Management Association, pp. 860-871. (A&WMA publication VIP-74).
10       Weschler, C. J.; Shields, H. C. (1999) Indoor ozone/terpene reactions as a source of indoor particles. Atmos.
11             Environ. 33:2301-2312.
12       White, W. H. (1998) Statistical considerations in the interpretation of size-resolved particulate mass data. J. Air
13             Waste Manage. Assoc. 48: 454-458.
14       Williams, R. W.; Creason,  J.; Shy, C.; Zweidinger, R.; Watts, R. R. (1999a) Indoor, outdoor and personal exposure
15             monitoring of particulate air pollution: the Baltimore elderly epidemiology-exposure pilot study. Submitted.
16       Williams, R. W.; Watts, R. R.; Stevens, R. K.; Stone, C. L.; Lewtas, J. (1999b) Evaluation of a personal air sampler
17             for twenty-four hour collection of fine particles and semivolatile organics. J. Exposure Anal. Environ.
18             Epidemiol. 9: 158-166.
19       Wilson, W. E.; Mage, D. T. (1999) Reinventing personal exposure to particulate matter. In: Third colloquium on
20             particulate air pollution and human health: abstracts; June; Durham, NC.
21       Wilson, W. E.; Suh, H. H. (1997) Fine particles and coarse particles: concentration relationships relevant to
22             epidemiologic studies. J. Air Waste Manage. Assoc. 47: 1238-1249.
23       Winer, B. J. (1962) Statistical principles in experimental design. New York, NY: McGraw-Hill Book Company; pp.
24             96-111.
25       Yocom, J. E.; Clink, W. L.; Cote, W. A. (1971) Indoor/outdoor air quality relationships. J. Air Pollut. Control
26             Assoc. 21:251-259.
         October 1999                                  5-103         DRAFT-DO NOT QUOTE OR CITE

-------
 i                                    APPENDIX 5-A
 2
 3                                        Nomenclature
 4
 5     a       air exchange rate between indoors and outdoors
 6     A      surface area for PM deposition in an indoor microenvironment
 7     AD     aerodynamic diameter (not to be confused with actual dimensions or optical diameter)
 8     C      indoor concentration of PM of ambient origin
 9     Ca     ambient concentration of PM (virtually all PM of ambient origin) at a monitoring station
10     Co     outdoor concentration of PM at a residence, not necessarily equal to Ca
11     E      total exposure to PM of ambient origin and PM of non-ambient origin.
12     Ea     exposure to PM of ambient origin
13     Ei      exposure to PM of non-ambient origin (PM generated indoors)
14     ETS    environmental tobacco smoke (PM)
15     F(M)   a function of ambient PM mass deposited indoors (M)
16     HVAC  Heating-Ventiliation-Air Conditioning
17     k       deposition rate of PM onto indoor surfaces, I/time
18     LOD   limit of detection
19     M      mass of PM of ambient origin deposited indoors
20     MDL   minimum detectable level
21     P      penetration factor for ambient PM from outdoors to an indoor microenvironment
22     PM^   particles collected by a monitor with a 50% collection efficiency at the given AD
23     Qa     rate of resuspension of ambient PM in an indoor microenvironment
24     Qother   PTEAM estimate of indoor PM emission rate from sources other than smoking and
25             cooking
26     r       Pearson correlation coefficient
27     R2     regression coefficient
28     SPSS   Scientific Probability Sampling Schema
29     T      time period of the study or the analysis
30     t       time, as a variable
31     v       alveolar ventilation rate

       October 1999                             5A-1        DRAFT-DO NOT QUOTE OR CITE

-------
 1     v      volumetric flow rate of air exchanged between indoors and outdoors
 2     V     volume of well mixed indoor microenvironment
 3     w     fraction of time spent outdoors at night in PTEAM (7pm - 7am)
 4     x      fraction of time spent outdoors during the day in PTEAM (7am - 7pm)
 5     y      fraction of time spent outdoors during a complete 24-h day
 6     z      fraction of ambient concentration (Ca) to which a person is exposed during a 24-hour
 7            period
 8     Y      resuspension rate of PM of ambient origin previously deposited in an indoor
 9            microenvironment
10     8io    delta function equal to  1 when subject is indoors and 0 when subject is outdoors, and8oi
11            is vice versa
12     e      spatial variation of ambient PM (Co - Ca)
13     a      arithmetic standard deviation
14     //      arithmetic mean
15
       October 1999                            5A-2        DRAFT-DO NOT QUOTE OR CITE

-------