Motor Vehicle-Related Air Toxics Study
&EPA
United States
Environmental Protection
Agency

-------
               Motor Vehicle-Related Air  Toxics  Study
                                   Technical Support Branch
                              Emission Planning and Strategies Division
                                   Office of Mobile Sources
                                   Office of Air and Radiation
                               U.S. Environmental Protection Agency
&EPA
United States
Environmental Protection
Agency
EPA-420-R-93-005
April 1993

-------
                                                        EPA-420-R-93-005
                                                            April 1993
                         Acknowledgements
     The  primary  EPA contributors  to  this  study were  Pamela
Brodowicz, Penny Carey,  Richard Cook,  and Joseph Somers.  Gertrude
Venlet and Betty Measley  provided  clerical  support.   An EPA work
group was formed to provide review, comments, and necessary input
from other parts of EPA.   Members of  this work group are Dwight R.
Atkinson, Joseph  J.  Bufalini,  James W.  Caldwell,  Ila  L.  Cote,
Stanley B. Durkee,  Robert  B.  Faoro, Robert Fegley, John T. Hannon,
Oscar Hernandez, Alan H. Huber, Gary L.  Kimmel, Kenneth T. Knapp,
Thomas F. Lahre, Mike Lidgard, Chris Lindhjem,  Thomas R. McCurdy,
William  E.  Pepelko,  Steven  A.  Shedd,  and  Chon  R.  Shoaf.   The
contributions of these individuals are gratefully acknowledged.

     This study incorporates material and  information  from four
work assignments, three initiated specifically to provide  input for
this  study.     Systems  Applications  International prepared  two
reports titled "Atmospheric Transformation of Air Toxics: Benzene,
1,3-Butadiene,  and  Formaldehyde,"  under  the direction of Mary P.
Ligocki,   and  "Atmospheric   Transformation   of  Air   Toxics:
Acetaldehyde  and Polycyclic  Organic Matter,"  prepared  for EPA's
Office of Policy Planning and Evaluation by Mary P.  Ligocki and
Gary Whitten.  Also, Clement International Corporation prepared a
report titled "Motor Vehicle  Air Toxics Health Information," under
the direction of Sharon Siegel.   Finally, Ted  Johnson,  Roy Paul,
and Jim  Capel  of  International Technology  Air  Quality Services
prepared  a  report  titled  "Application of the  Hazardous  Air
Pollutant Exposure Model  (HAPEM) to Mobile Source Pollutants."

     Other individuals who contributed significantly to this study,
providing technical  expertise and guidance, include  Jeff Alson,
David  Brzezinski,  Gary Dolce,  Alva  Edwards,  Carl  Fulper,  Karl
Hellman, Greg Janssen, Karen Levy, Phil Lorang,  Steve Mayotte, Carl
Mazza, Will  Smith,  Todd Ramsden, Rick Rykowski,  and Mark Wolcott of
EPA,  Robert  Kelty   and   Tom  McDonnell  of  Computer  Sciences
Corporation, and Edward Sienicki of Navistar Corporation.  Comments
on  preliminary drafts  of various  portions of  the study  were
provided  by Larry  Claxton,  Larry Cupitt,   Linda  Birnbaum,  Mike
Davis, and Vanessa  Vu of EPA,  and K.  D. Drachand and Joe DeVita of
the California Air  Resources Board.  Additional  EPA comments on the
public review draft  were  provided by Aparna M.  Koppikar,  Cheryl
Siegel  Scott,  and  Hugh  McKinnon.    The  following organizations
provided comments on the public review draft:  American Automobile
Manufacturers   Association,    American    Petroleum   Institute,
Association   of  International  Automobile   Manufacturers,   Arco
Chemical  Company,   California  Air  Resources  Board,  California
Environmental Protection Agency, Chemical  Manufacturers Association
(CMA),  Engine  Manufacturers  Association,   Ford  Motor  Company,
General Motors Corporation, Health Effects Institute, Konheim and
Ketcham, Northeast  States  for Coordinated Air Use Management, and
Zephyr Consulting.

     The  authors  of  this study  wish to express  their sincere
appreciation for the efforts of all participants.

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
EXECUTIVE SUMMARY

1 . 0   INTRODUCTION                                      1-1
      1.1  Background                                    1-1
      1.2  Congressional Mandate                         1-2
      1.3  Scope of Study                                1-2
      1.4  Participation by Other EPA Offices  and       1-4
          the Public
      1.5  References for Chapter 1                      1-7

2.0   SCENARIOS  STUDIED                                 2-1
      2.1  Baseline                                      2-1
      2.2  Additional Control Scenarios                  2-2
          2.2.1     Expanded Use of Reformulated       2-2
                    Gasoline
          2.2.2     Expanded Adoption of                2-2
                    California Motor Vehicle
                    Standards

3.0   EMISSION FACTOR METHODOLOGY                       3-1
      3.1  Methodology  for Benzene, Formaldehyde,        3-1
          1, 3 -Butadiene, and Acetaldehyde
          3.1.1     Approach                            3-1
          3.1.2     Assumptions                         3-1
          3.1.3     Emission Factor Requirements       3-3
                3.1.3.1  Scenario Components            3-3
                3.1.3.2  Percent of Nationwide          3-11
                         Fuel Use by Component  for
                         Each Scenario
                3.1.3.3  Emission Fractions             3-17
                         Associated with
                         Components
                3.1.3.4  I/M Programs Associated       3-17
                         with Components
                3.1.3.5  Estimating Risk Under          3-18
                         Different Scenarios
          3.1.4     MOBTOX Emissions Model  Inputs       3-18
                3.1.4.1  HC Exhaust Reductions  for     3-18
                         Gasoline Oxygenated
                         Blends
                3.1.4.2  California LEV Standards       2-22
                3.1.4.3  Toxic Exhaust Fractions       3-23
                3.1.4.4  Other Inputs                   3~27
      3.2  Methodology  for Diesel Particulate            3-27
          Matter
      3.3  Methodology  for Gasoline Particulate          3-28
          Matter
      3.4  References for Chapter 3                      3-29
4.0   EXPOSURE METHODOLOGY

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
      4.1  Annual Average Population Exposure            4-1
          Estimation
          4.1.1     The NAAQS Exposure Model  (NEM)      4-2
          4.1.2     Use Of HAPEM-MS Model               4-4
          4.1.3     Use of Ambient Monitoring Data      4-8
          4.1.4     Procedure for Calculating           4-13
                    Cancer Incidences or Deaths
      4.2  Short-Term Microenvironment Exposures         4-13
      4.3  References for Chapter 4                      4-15

5.0   BENZENE                                           5-1
      5.1  Chemical and Physical Properties              5-1
      5.2  Formation and Control Technology              5-2
      5.3  Emissions                                     5-3
          5.3.1     Emission Fractions Used in the      5-3
                    MOBTOX Emissions Model
                5.3.1.1  Benzene Exhaust Emission       5-4
                         Fractions
                5.3.1.2  Benzene Diurnal and Hot        5-4
                         Soak Evaporative Emission
                         Fractions
                5.3.1.3  Benzene Running, Resting,      5-6
                         and Refueling Loss
                         Evaporative Emission
                         Fractions
          5.3.2     Emission Factors for Baseline       5-6
                    and Control Scenarios
          5.3.3     Nationwide Mobile Source            5-8
                    Benzene Emissions
          5.3.4     Other Sources of Benzene            5-8
      5.4  Atmospheric Reactivity and Residence          5-10
          Times
          5.4.1     Atmospheric Transformation          5-11
                    Processes
          5.4.2     Gas Phase Chemistry of Benzene      5-11
                5.4.2.1  Gas Phase Reactions            5-12
                5.4.2.2  Reaction Products              5-12
          5.4.3     Aqueous Phase Chemistry of          5-12
                    Benzene
          5.4.4     Atmospheric Residence Times         5-13
                5.4.4.1  Definition and                 5-13
                         Limitations
                5.4.4.2  Chemical and Physical          5-13
                         Processes
                5.4.4.3  Generation of Input            5-15
                         Values
                5.4.4.4  Benzene Residence Times        5-15
      5.4.5      Limited Urban Airshed Modeling of       5-17
                Air Toxics
                5.4.5.1  General Results from the       5-19
                         UAM Simulations

                                ii

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
                                                       Page

                5.4.5.2  UAM Results for Benzene       5-21
      5.5  Exposure Estimation                          5-23
          5.5.1     Annual Average Exposure Using      5-23
                    HAPEM-MS
          5.5.2     Comparison of HAPEM-MS             5-25
                    Exposures to Ambient
                    Monitoring Data
          5.5.3     Short-Term Microenvironment        5-29
                    Exposures
      5.6  Carcinogenicity of Benzene and Unit Risk     5-32
          Estimates
          5.6.1     Most Recent EPA Assessment         5-32
                5.6.1.1  Description of Available      5-33
                         Carcinogenicity Data
                5.6.1.2  Weight-Of-Evidence            5-36
                         Judgment of Data and EPA
                         Classification
                5.6.1.3  Data Sets Used for Unit       5-37
                         Risk Estimate
                5.6.1.4  Dose-Response Model Used      5-37
                5.6.1.5  Unit Risk Estimates           5-39
          5.6.2     Other Views and Unit Risk          5-39
                    Estimates
          5.6.3     Recent and Ongoing Research        5-45
                5.6.3.1  Genotoxicity                  5-45
                5.6.3.2  Pharmacokinetics              5-47
                5.6.3.3  Carcinogenicity - Animal      5-49
                         Studies
                5.6.3.4  Carcinogenicity -             5-51
                         Epidemiological Studies
      5.7  Carcinogenic Risk for Baseline and           5-54
          Control Scenarios
      5.8  Non-Carcinogenic Effects of Inhalation       5-54
          Exposure to Benzene
      5.9  References for Chapter 5                     5-59

6.0   FORMALDEHYDE                                     6-1
      6.1  Chemical and Physical Properties             6-1
      6.2  Formation and Control Technology             6-2
      6.3  Emissions                                    6-2
          6.3.1     Emission Fractions Used in the     6-2
                    MOBTOX Emissions Model

          6.3.2     Emission Factors for Baseline      6-4
                    and Control Scenarios
          6.3.3     Nationwide Mobile Source           6-4
                    Formaldehyde Emissions
          6.3.4     Other Sources of Formaldehyde      6-4
      6.4  Atmospheric Reactivity and Residence         6-7
          Times
          6.4.1     Gas Phase Chemistry of             6-7

                               iii

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
                    Formaldehyde
                6.4.1.1  Formation
                6.4.1.2  Gas Phase Reactions
                6.4.1.3  Reaction Products
          6.4.2     Aqueous Phase Chemistry of
                    Formaldehyde                        6-10
          6.4.3     Formaldehyde Residence Times        6-12
          6.4.4     Limited Urban Airshed Modeling
                    Results for Formaldehyde            6-15
      6.5  Exposure Estimation                           6-15
          6.5.1     Annual Average Exposures Using
                    HAPEM-MS                            6-15
          6.5.2     Comparison of HAPEM-MS
                    Exposures to Ambient
                    Monitoring Data                     6-22
          6.5.3     Short-Term Microenvironment
                    Exposures                           6-24
      6.6  Carcinogenicity of Formaldehyde and Unit
          Risk Estimates                                6-24
          6.6.1     Most Recent EPA Assessment          6-25
                6.6.1.1  Description of Available
                         Carcinogenicity Data           6-29
                6.6.1.2  Weight-Of-Evidence
                         Judgment of Data and EPA
                         Classification                 6-30
                6.6.1.3  Data Sets Used for Unit
                         Risk Estimate                  6-30
                6.6.1.4  Dose-Response Model Used       6-30
                6.6.1.5  Unit Risk Estimates            6-32
          6.6.2     Other Views and Unit Risk
                    Estimates                           6-42
          6.6.3     Recent and Ongoing Research         6-42
                6.6.3.1  Genotoxicity                   6-42
                6.6.3.2  Pharmacokinetics               6-42
                6.6.3.3  Carcinogenicity - Animal
                         Studies                        S-45
                6.6.3.4  Carcinogenicity -
                         Epidemiological Studies        6-48
      6.7  Carcinogenic Risk for Baseline and
          Control Scenarios                             6-50
      6.8  Non-Carcinogenic Effects of Inhalation
          Exposure to Formaldehyde                      6-54
      6.9  References for Chapter 6
                                                        7-1
7.0   1,3-BUTADIENE                                     7-1
      7.1  Chemical and Physical Properties              7"1
      7.2  Formation and Control Technology              7~2
      7.3  Emissions                                     7~2
          7.3.1     Emission Fractions Used in the
                    MOBTOX Emissions Model              7"4
          7.3.2     Emission Factors for Baseline

                                iv

-------
                                                        EPA-420-R-93-005
                                                            April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
                    and Control Scenarios               7-4
          7.3.3     Nationwide Mobile Source
                    1,3-Butadiene Emissions             7-4
          7.3.4     Other Sources of 1,3-Butadiene      7-7
      7.4  Atmospheric Reactivity and Residence
          Times                                         7-7
          7.4.1     Gas Phase Chemistry of
                    1,3-Butadiene                       7-8
                7.4.1.1  Gas Phase Reactions            7-8
                7.4.1.2  Reaction Products              7-8
          7.4.2     Aqueous Phase Chemistry of
                    1,3-Butadiene                       7-8
          7.4.3     1,3-Butadiene Residence Times       7-11
          7.4.4     Limited Urban Airshed Modeling
                    Results for 1,3-Butadiene           7-12
      7.5  Exposure Estimation                           7-12
          7.5.1     Annual Average Exposures Using
                    HAPEM-MS                            7-14
          7.5.2     Comparison of HAPEM-MS
                    Exposures to Ambient
                    Monitoring Data                     7-18
          7.5.3     Short-Term Microenvironment
                    Exposures                           7-20
      7.6  Carcinogenicity of 1,3-Butadiene and
          Unit Risk Estimates                           7-20
          7.6.1     Most Recent EPA Assessment          7-21
                7.6.1.1  Description of Available
                         Carcinogenicity Data           7-24
                7.6.1.2  Weight-Of-Evidence
                         Judgment of Data and EPA
                         Classification                 7-25
                7.6.1.3  Data Sets Used for Unit
                         Risk Estimate                  7-25
                7.6.1.4  Dose-Response Model Used       7-25
                7.6.1.5  Unit Risk Estimates            7-27
          7.6.2     Other Views and Unit Risk
                    Estimates                           7-34
          7.6.3     Recent and Ongoing Research         7-34
                7.6.3.1  Genotoxicity                   7~36
                7.6.3.2  Pharmacokinetics               7-37
                7.6.3.3  Carcinogenicity - Animal
                         Studies                        7~39
                7.6.3.4  Carcinogenicity -
                         Epidemiological Studies        7-41
      7.7  Carcinogenic Risk for Baseline and
          Control Scenarios                             7-43
      7.8  Non-Carcinogenic Effects of Inhalation
          Exposure to 1,3-Butadiene                     7-46
      7.9  References for Chapter 7
                                                        8-1
8 . 0   ACETALDEHYDE                                      8"I

                                v

-------
                                                   EPA-420-R-93-005
                                                       April 1993
                   TABLE  OF CONTENTS
    (Note: page numbers may be slightly off in electronic PDF version)
8.1 Chemical  and  Physical  Properties             8-2
8.2 Formation and Control  Technology             8-2
8.3 Emissions                                    8-2
    8.3.1      Emission  Fractions Used in the
               MOBTOX  Emissions  Model             8-3
    8.3.2      Emission  Factors  for Baseline
               and Control  Scenarios              8-3
    8.3.3      Nationwide Mobile Source
               Acetaldehyde Emissions             8-3
    8.3.4      Other Sources of  Acetaldehyde      8-6
8.4 Atmospheric Reactivity and  Residence
    Times                                         8-6
    8.4.1      Gas Phase Chemistry of
               Acetaldehyde                       8-7
          8.4.1.1  Formation                     8-7
          8.4.1.2  Gas  Phase Reactions           8-8
          8.4.1.3  Reaction Products             8-8
    8.4.2      Aqueous Phase Chemistry of
               Acetaldehyde                       8-8
    8.4.3      Acetaldehyde Residence Times       8-11
    8.4.4      Limited Urban Airshed Modeling
               Results for  Acetaldehyde           8-13
8.5 Exposure  Estimation                          8-13
    8.5.1      Annual  Average Exposures Using
               HAPEM-MS                            8-13
    8.5.2      Comparison of HAPEM-MS
               Exposures to Ambient
               Monitoring Data                    8-16
    8.5.3      Short-Term Microenvironment
               Exposures                          8-19
8.6 Carcinogenicity of  Acetaldehyde and Unit
    Risk Estimates                               8-19
    8.6.1      Most Recent  EPA Assessment
                                                  8-19
          8.6.1.1  Description  of Available
                   Carcinogenicity Data          8-24
          8.6.1.2  Weight-Of-Evidence
                   Judgment of  Data and EPA
                   Classification                8-25
          8.6.1.3  Data Sets Used for Unit
                   Risk Estimate                 8-25
          8.6.1.4  Dose-Response Model Used      8-25
          8.6.1.5  Unit Risk Estimates           8-25
    8.6.2      Other Views  and Unit Risk
               Estimates                          8"26
    8.6.3      Recent  and Ongoing Research        8-26
          8.6.3.1  Genotoxicity                  8"27
          8.6.3.2  Metabolism and
                   Pharmacokinetics              8-28
8.7 Carcinogenic  Risk for  Baseline and
    Control Scenarios                            8-30

                          vi

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
      8.8  Non-Carcinogenic Effects of  Inhalation        8-30
          Exposure to Acetaldehyde                      8-32
          8.8.1     Toxicity
          8.8.2     Reference Concentration  for
                    Chronic Inhalation Exposure         8-34
                    (RfC)
          8.8.3     Reproductive and Developmental      8-37
                    Effects
      8.9  References for Chapter 8                      9-1
                                                        9-1
9.0   DIESEL  PARTICULATE MATTER                         9-1
      9.1  Chemical and Physical Properties              9-3
      9.2  Formation and Control Technology              9-3
      9.3  Emissions
          9.3.1     Diesel Particulate Matter           9-3
                    Emission Standards                 9-5
          9.3.2     Methodology
                9.3.2.1  Calculation of Urban
                         Diesel Vehicle Miles           9-6
                         Travelled
                9.3.2.2  Calculation of Diesel
                         Particulate Matter             9-8
                         Emission Rate
                9.3.2.3  Calculation of Urban
                         Diesel Particulate  Matter      9-8
                         Emissions
                9.3.2.4  Calculation of the  Urban
                         Diesel Particulate  Matter
                         National Fleet Average
                         Emission Factor                9-9

          9.3.3     Nationwide Diesel  Particulate       9-10
                    Matter Emissions
      9.4  Atmospheric Reactivity and Residence
          Times of Particulate Phase Polycyclic         9-10
          Organic Matter (POM)                          9-11
          9.4.1     Particulate Phase  Chemistry         9-11
          9.4.2     Aqueous Phase Chemistry             9-12
          9.4.3     Reaction Products
          9.4.4     Polycyclic Organic Matter           9-13
                    Residence Times                     9-15
                9.4.4.1  Pyrene                         9-15
                9.4.4.2  Benzo[alpyrene                 9-15
                9.4.4.3  Other POM Species              9-17
                9.4.4.4  POM as a Class                 9-2
          9.4.5     Urban Airshed Modeling of POM       9~20
      9.5  Exposure Estimation
          9.5.1     Annual Average Exposures Using      9-20
                    HAPEM-MS
          9.5.2     Comparison of HAPEM-MS
                    Exposures to Ambient                9-20

                               vii

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
                    Monitoring Data                     9-20
      9.6  Carcinogenicity of Diesel Particulate         9-21
          Matter and Unit Risk Estimates
          9.6.1     Most Recent EPA Assessment          9-29
                9.6.1.1  Description of Available
                         Carcinogenicity Data
                9.6.1.2  Weight-Of-Evidence             9-30
                         Judgment of Data and EPA
                         Classification                 9-30
                9.6.1.3  Data Sets Used for Unit        9-30
                         Risk Estimate                  9-34
                9.6.1.4  Dose-Response Model Used
                9.6.1.5  Unit Risk Estimate             9-39
          9.6.2     Other Views and Unit Risk           9-39
                    Estimates
          9.6.3     Recent and Ongoing Research         9-41
                9.6.3.1  Metabolism and
                         Pharmacokinetics               9-43
                9.6.3.2  Carcinogenicity - Animal       9-44
                         Studies
      9.7  Carcinogenic Risk                             9-49
      9.8  Non-Carcinogenic Effects of Inhalation
          Exposure to Diesel Particulate Matter         10-1
      9.9  References for Chapter 9                      10-1
                                                        10-1
10.0  GASOLINE  PARTICULATE  MATTER                       10-2
      10.1  Chemical  and Physical  Properties             10-2
      10.2  Formation and Control  Technology
      10.3  Emissions                                    10-2
          10.3.1    Emission Factors for Baseline
                    Scenarios                           10-2
      10.4  Atmospheric  Reactivity and Residence
           Times
          10.4.1    Urban Airshed Modeling of           10-3
                    Reformulated Gasoline Impact        10-4
                    on Ambient POM
      10.5  Exposure  Estimation                          10-4
      10.6  Carcinogenicity  of Gasoline Particulate      10-4
           Matter and Unit  Risk Estimates
          10.6.1    Most Recent EPA Assessment          10-4
                10.6.1.1 Description of Available
                         Carcinogenicity Data
                10.6.1.2 Weight-Of-Evidence             10~5
                         Judgment of Data and EPA
                         Classification                 10-5
                10.6.1.3 Data Sets Used for Unit        1Q-5
                         Risk Estimate                  10-6
                10.6.1.4 Dose-Response Model Used
                10.6.1.5 Unit Risk Estimates            10~6
          10.6.2    Other Views and Unit Risk           10"6
                    Estimates                           10~8

                               viii

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
          10.6.3    Recent and Ongoing Research         10-9
      10.7  Pro  Forma Carcinogenic Risk
      10.8  Non-Carcinogenic Effects of Inhalation       11-1
           Exposure to Gasoline Particulate Matter      11-1
      10.9  References for Chapter 10                    11-2
                                                        11-2
11.0  GASOLINE  VAPORS
      11.1  Chemical and Physical Properties             11-2
      11.2  Exposure Estimation                          11-2
      11.3  Carcinogenicity of Gasoline Vapors and
           Unit Risk Estimates                          11-7
          11.3.1    Most Recent EPA Assessment
                11.3.1.1 Description of Available
                         Carcinogenicity Data           11-7
                11.3.1.2 Weight-Of-Evidence
                         Judgment  of Data  and EPA       11-7
                         Classification                 11-7
                11.3.1.3 Data Sets Used for  Unit        11-8
                         Risk Estimate
                11.3.1.4 Dose-Response Model Used       11-12
                11.3.1.5 Unit Risk Estimates
          11.3.2    Other Views and Unit Risk           11-12
                    Estimates
          11.3.3    Recent and Ongoing Research

          11.3.3.1  Alpha2u-Globulin: Association        11-13
                    with Chemically Induced  Renal       11-14
                    Toxicity and Neoplasia in the
                    Male Rat                            11-15
                11.3.3.2 Genotoxicity
                11.3.3.3 Metabolism and                 11-16
                         Pharmacokinetics
                11.3.3.4 Carcinogenicity --  Animal      11-16
                         Studies                        11-18
                11.3.3.5 Carcinogenicity --
                         Epidemiological Studies        11-19
      11.4  Carcinogenic Risk
      11.5  Non-Carcinogenic Effects of Inhalation       12-1
           Exposure to Gasoline Vapors                  12-1
      11.6  References for Chapter 11                    12-1

12.0  EPA'S INTEGRATED AIR CANCER PROJECT               12-3
      12.1  Background
      12.2  Methodology for Mutagenicity                 12-3
           Apportionment                                12-5
      12.3  Apportionment of Mutagenicity from           12-5
      Field Measurement Programs                        12-6
          12.3.1    Raleigh, North Carolina             12"6
          12.3.2    Albuquerque, New Mexico             12-7
          12.3.3    Boise, Idaho                        12~8
      12.4  Other IACP Studies                           12~8

                                ix

-------
                                                        EPA-420-R-93-005
                                                             April 1993
                        TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
          12.4.1    Human Cancer Risk Estimates         12-9
          12.4.2    Human Exposure                      12-10
          12.4.3    Atmospheric Transformation
      12.5  Roanoke Field Study                          13-1
      12.6  Implications                                 13-2
      12.7  References  for Chapter 12                    13-2

13.0  TOXICS  ASPECTS OF ALTERNATIVE FUELS               13-3
      13.1  Methanol
          13.1.1    Health Effects  of Toxic             13-4
                    Emissions from  Methanol Use
                13.1.1.1 Effects of Chronic and
                         Acute Exposures  -- Humans      13-6
                13.1.1.2 Effects of Chronic and         13-7
                         Acute Exposures  --Animal
                         Studies                        13-9
                13.1.1.3 Health Based Criteria          13-10
          13.1.2    Effects of Methanol Use on Air      13-11
                    Toxic Levels                        13-12
      13.2  Ethanol
      13.3  Compressed  Natural  Gas                       14-1
      13.4  Liquid Propane Gas
      13.5  References  for Chapter 13                    15-1
                                                        15-1
14.0  NONROAD MOBILE SOURCES                            15-1
                                                        15-3
15.0  INITIAL COST CONSIDERATIONS                       15-3
      15.1  Costs of Various  Regulatory Programs         15-3
          15.1.1    Tier 1 Standards
          15.1.2    California Standards                15-4
          15.1.3    Reformulated Gasoline  Program
          15.1.4    Inspection/Maintenance (I/M)        15-5
                    Programs                            15-5
          15.1.5    Winter Oxygenated Fuels             15-5
                    Program
          15.1.6    Diesel Particulate Standards        15-6
          15.1.7    Diesel Fuel Sulfur Regulation       15-7
      15.2  Qualitative Discussion of Toxics             15-7
           Benefits                                     15-7
          15.2.1    Tier 1 Standards
          15.2.2    California Standards                15-8
          15.2.3    Reformulated Gasoline  Program
          15.2.4    Inspection/Maintenance (I/M)        15-8
                    Programs
          15.2.5    Winter Oxygenated Gasoline          15-9
                    Program
          15.2.6    Diesel Particulate Standards        16-1
                    and Fuel Sulfur Regulation
      15.3  References  for Chapter 15                    16-1
                            y                          16-3
16.0  MOTOR VEHICLE TOXICS IN  TITLE III AND             16~3

                                x

-------
                                                         EPA-420-R-93-005
                                                              April 1993
                         TABLE OF CONTENTS
          (Note: page numbers may be slightly off in electronic PDF version)
                                                        Page

      METALLIC POLLUTANTS                               16-10
      16.1 Dioxins
      16.2 MTBE
      16.2 N-Nitrosodimethylamine
      16.3 References for Chapter  16

APPENDICES
A     EPA Work Group Members
B     Emission Factor Data for Benzene,
      Formaldehyde, 1,3-Butadiene, and
      Acetaldehyde
C     Ambient Monitoring Data  for  Benzene,
      Formaldehyde, 1,3-Butadiene, and
      Acetaldehyde
D     Time Series Plots for Benzene,  Formaldehyde,
      1,3-Butadiene, and Acetaldehyde
E     Benzene Unit Risk Estimates  Based on 21
      Models

F     Lay Description of the Linearized Multistage
      Model
G     Diesel Particulate Emission  Factor Inputs
H     Unleaded Gasoline Particulate Emission
      Fractions
I     Summary of Comments on Public Review Draft
      of Motor Vehicle-Related Air Toxics Study
                                XI

-------
                                                        EPA-420-R-93-005
                                                            April 1993
                        EXECUTIVE SUMMARY
     Section 202(1)(1) of the Clean Air Act  (CAA),  as amended
(Section 206 of the Clean Air Act Amendments  (CAAA) of 1990 added
paragraph (1) to Section 202 of the CAA),  directs EPA to complete
a study by May 15, 1992 of the need for, and feasibility of,
controlling emissions of toxic air pollutants which are
unregulated under the Act and associated with motor vehicles and
motor vehicle fuels.   In addition, the study is to consider the
means and measures for such controls.  The required study is to
focus on those categories of emissions that pose the greatest
risk to human health or about which significant uncertainties
remain, including emissions of benzene, formaldehyde, and
1,3-butadiene.  This study has been prepared in response to
Section 202 (1) (1)  .

     Motor vehicle emissions are extremely complex.  Hundreds of
compounds have been identified.  For this study, specific
pollutants or pollutant categories which are discussed include
benzene, formaldehyde, 1,3-butadiene, acetaldehyde, diesel
particulate matter, gasoline particulate matter, and gasoline
vapors, all of which have been considered in previous analyses of
air toxics,  as well as selected metals and motor vehicle-related
pollutants identified in Section 112(b) of the Clean Air Act.

     The focus of the study is on carcinogenic risk.  The
discussion of non-carcinogenic effects is less quantitative due
to the lack of sufficient health data.  Nevertheless,
noncarcinogenic effects should not be viewed as less important.
Noncancer effects associated with exposures to the pollutants
discussed in this study await assessment.

     There are a number of major limitations and uncertainties
which need to be considered carefully when reviewing the results
of this study.  In the interest of readability, the contents of
this study are discussed first, then the limitations and
uncertainties presented.

     There are chapters devoted to each individual pollutant or
pollutant category.  Topics covered for each pollutant/pollutant
category include chemical and physical properties,  formation and
control technology, emissions  (including other emission sources),
atmospheric reactivity and residence times, exposure estimation,
EPA's carcinogenicity assessment, other views of carcinogenicity
assessment,  recent and ongoing research, carcinogenic risk, and
non-cancer health effects.  There is also a chapter which
describes EPA's Integrated Air Cancer Project, aimed at
identifying the major carcinogenic chemicals emitted into the
air, and the sources of these chemicals.  A chapter is also
included which describes qualitative changes in toxic pollutant
levels with the use of alternative clean fuels such as methanol,
ethanol, compressed natural gas, and liquid propane gas.  Another
brief chapter discusses toxic emissions from nonroad mobile
sources.  In addition, a chapter discusses the costs of various


                               ES-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993

existing regulatory programs and a qualitative discussion of the
toxics benefits of these programs.

     This study attempts to summarize what is known about motor
vehicle-related air toxics and to present all significant
scientific opinion on each issue.  Based on information presented
in this study and other relevant information, EPA is to
promulgate (and from time to time revise) regulations by May 15,
1995 that contain reasonable requirements to control hazardous
air pollutants from motor vehicles and motor vehicle fuels.  The
regulations,  at a minimum, apply to emissions of benzene and
formaldehyde.  This study does not address whether to promulgate
standards or what standards should be promulgated, since those
issues will be addressed in the rulemaking activity.

     Briefly, cancer risk estimates were obtained in the
following manner.  First, emission factors in units of gram/mile
were estimated as a function of vehicle technology and fuel
composition.   These emission factors were then used in a model to
calculate annual average exposures.  The annual nationwide
exposures were compared to the range of ambient data, and where
necessary, adjustments were applied such that modeled data
matched the upper end of the ambient range.  Then, the adjusted
exposures were multiplied by the population of interest and the
EPA unit risk factor to calculate lifetime cancer incidence or,
for benzene and diesel particulate matter, cancer deaths.  The
unit risk factor is the excess individual lifetime risk due to
continuous lifetime exposure to one unit  (in this case, ug/m3)  of
carcinogen concentration.  To calculate annual cancer incidence
(or deaths),  the lifetime cancer incidence (or deaths) was
divided by 70, the average years per lifetime.

     Cancer risk estimates for benzene, diesel particulate
matter, formaldehyde,  1,3-butadiene, and acetaldehyde are
provided for the following years:  1990, 1995, 2000, and 2010.
The following scenarios are examined:

     1)   a base control scenario, which takes into account
          implementation of the motor vehicle-related Clean Air
          Act requirements,

     2)   a scenario involving expanded use of reformulated
          gasoline, and

     3)   a scenario involving expanded adoption of California
          motor vehicle emission standards.

     The expanded control scenarios are not intended to be
predictive, but instead are intended to encompass a wide range of
possibilities.  Base control scenarios for the years examined
take into account implementation of the motor vehicle-related CAA
requirements, but assume no expanded adoption of CAA programs or
California standards.   The expanded use of reformulated fuel
scenario is considered for the years 1995, 2000, and 2010.  In
this scenario, all ozone nonattainment areas opt into the federal


                               ES-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

reformulated gasoline program.  The expanded adoption of
California standards scenario is considered for the years 2000
and 2010.  California emission standards are similar to federal
motor vehicle-related standards in 1995; thus, this scenario is
not considered for that year.   However, California motor vehicle
emission standards become increasingly more stringent with time,
so that in 2000 and 2010, they are markedly lower than federal
standards.  In this scenario, Northeast states and states with
ozone nonattainment areas categorized as extreme, severe, or
serious adopt California emission standards.  This scenario also
assumes expanded use of reformulated gasoline, as described in
the previous scenario.  Federal Tier II standards were not
evaluated in this study.

     Table ES-1 summarizes the emission factors, annual average
exposure estimates, nationwide cancer incidence  (or deaths),  and
nationwide annual individual risks for all scenarios/years.  The
limitations and uncertainties listed in the footnotes to this
table and discussed at the end of the executive summary should be
considered when reviewing these numbers.  For the base control
scenarios, the cancer incidences or deaths decrease from 1990 to
1995 and from 1995 to 2000.  From 2000 to 2010, the cancer
incidences or deaths  increase for 1,3-butadiene, formaldehyde,
and acetaldehyde.  For these toxics, even though the fleet
average emission factors in gram/mile continue to decrease from
2000 to 2010, the projected increase in vehicle miles travelled
(and population to a lesser extent)  more than offsets this
decrease.  For benzene, cancer deaths remain unchanged from 2000
to 2010, whereas for diesel particulate, cancer deaths decrease.
It should be noted that, due to uncertainties associated with the
additivity of cancer risk associated with the toxics, total
cancer risk for all toxics for a given scenario/year are not
presented in Table ES-1.

     The expanded use of reformulated gasoline and expanded
adoption of California motor vehicle emission standards scenarios
result in lower cancer deaths or incidences for benzene and 1,3-
butadiene relative to their base control scenarios.  Cancer
incidences due to formaldehyde increase slightly, but are more
than offset by the benzene and 1,3-butadiene decreases.

     Oxygenated fuels provide overall health benefits because
they significantly reduce winter CO in areas which exceed CO
ambient air quality standards.  Increased use of oxygenated fuels
may result in small increases in ambient aldehyde levels and may
increase intermittent exposures to concentrations higher than
ambient levels.  However, the use of oxygenated fuels also
results in
                               ES-3

-------
  Table ES-1.   Summary of Estimates of Emission Factors,  Annual  Average  Exposure,
 Annual  Cancer  Deaths  or  Incidences, and Nationwide Annual Individual Risk for All

Pollutant
BENZENE
1990
Base
Control
1995
Base
Control
Expanded
Reform.
Gasoline
Use
2000
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
2010
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
c,d
Estimated Cancer Deaths with Estimates of Exposure Calculated in this Study
Estimated cancer deaths are based on the EPA 1985 unit risk of 8.3xlO"6 per ug/m3, determined using human data.
b
EF (g/mi)
c
Exposure
(ug/m3)
d
Cancer Deaths
Average of e
Individual Risk

Range of Exposure
(ug/m3)
Range of Cancer
Deaths

Exposure (ug/m3)
Cancer Deaths
(Clement, 1988)
Cancer Deaths
(GARB, 1984)
0.0882
2.36
70
2.8xlO"7
Estimated
1.37-
3.98
41-118
0.0472
1.40
43
1.7xlO"7
0.0413
1.20
37
1.4xlO"7
0.0351
1.10
35
1.3xlO"7
0.0301
0.98
31
1.2xlO"7
Cancer Deaths with Alternative Estimates
0.81-
2.36
25-72
0.70-
2.02
22-62
0.64-
1.86
21-59
0.57-
1.65
18-52
0.0305
0.98
31
1.2xlO"7
0.0285
1.05
35
1.2xlO"7
0.0248
0.93
31
l.lxlO"7
0.0228
0.84
28
9.9xlO"8
of Exposure
0.57-
1.65
18-52
0.61-
1.77
18-53
0.54-
1.57
18-52
0.49-1.42
16-47
Estimated Cancer Deaths with Clement Associates, 1988 Unit Risk (4.3xl(T8 per ug/m3) or
GARB, 1984 Unit Risk (5.2xl(T5 per ug/m3)  These are not directly comparable to the
official EPA unit risk estimates.9
2.36
< 1
438
1 .40
< 1
269
1.20
< 1
232
1.10
< 1
219
0.98
< 1
194
0.98
< 1
194
1.05
< 1
219
0.93
< 1
194
0.84
< 1
175
Please refer to footnotes on page ES-9.
                                             ES-4

-------
                                        Table ES-1  Continued.
                                                                                          EPA-420-R-93-005
                                                                                              April 1993

Pollutant
FORMALDEHYDE
1990
Base
Control
1995
Base
Control
Expanded
Reform.
Gasoline
Use
2000
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
2010
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
c,d
Estimated Cancer Incidences with Estimates of Exposure Calculated in this Study
Estimated cancer incidences are based on the EPA 1987 upper bound unit risk of 1.3xlO"5 per ug/m3, determined
using animal data.
b
EF (g/mi)
c
Exposure
(ug/m3)
d
Cancer Cases
Average of e
Individual Risk

Range of Exposure
(ug/m3)
Range of Cancer
Cases

Exposure (ug/m3)
h
Cancer Cases
(EPA, 1991)
Cancer Cases
(EPA, 1987)
0.0412
0.95
44
l.SxlO"7
0.0234
0.58
28
l.lxlO"7
0.0251
0.62
30
1.2xlO"7
0.0162
0.42
21
7.8xlO"8
0.0166
0.44
22
8.2xlO"8
0.0168
0.44
22
8.2xlO"8
0.0140
0.42
22
7.8xlO"8
0.0143
0.46
24
8.5xlO"8
0.0138
0.42
22
7.8xlO"8
Estimated Cancer Incidences with Alternative Estimates of Exposure
0.95-
2.87
44-133
0.58-
1.75
28-85
0.62-
1.87
30-91
0.42-
1.27
21-63
0.44-
1.33
22-67
Estimated Cancer Incidences with EPA, 1991 Draft
Ug/m3) or EPA, 1987 Upper Bound Unit Risk (1.3x10
estimate is not an official EPA estimate.9
0.95
2
44
0.58
1
28
0.62
1
30
0.42
1
21
0 .44
1
22
0.44-
1.33
22-67
0.42-
1.27
22-67
0.46-
1.39
24-73
0.42-
1.27
22-67
Upper Bound Unit Risk (6.0xlO~7 per
5 per ug/m3)  The draft EPA, 1991
0 .44
1
22
0.42
1
22
0.46
1
24
0.42
1
22
Please  refer to  footnotes  on page ES-9.
                                                 ES-5

-------
                                        Table ES-1 Continued.
                                                                                          EPA-420-R-93-005
                                                                                              April 1993

Pollutant
1,3 -BUTADIENE
1990
Base
Control
1995
Base
Control
Expanded
Reform.
Gasoline
Use
2000
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
2010
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
c,d
Estimated Cancer Incidences with Estimates of Exposure Calculated in this Study
Estimated cancer incidences are based on the EPA 1985 upper bound unit risk of 2.8xlO"4 per ug/m3, determined
using animal data.
b
EF (g/mi)
c
Exposure
(ug/m3)
d
Cancer Cases
Average of e
Individual Risk

Range of Exposure
(ug/m3)
Range of Cancer
Cases

Exposure (ug/m3)
Cancer Cases
(Hattis and Watson
1987)
Cancer Cases
(ICE, 1986)
0.0156
0.30
304
1.2xlO"6
0.0094
0.20
209
S.lxlO"7
0.0093
0.20
207
S.OxlO"7
0.0071
0.16
176
6.6xlO"7
0.0069
0.16
171
6.4xlO"7
0.0069
0.16
172
6.4xlO"7
0.0067
0.18
204
7.2xlO"7
0.0064
0.17
194
6.9xlO"7
0.0062
0.16
186
6.6xlO"7
Estimated Cancer Incidences with Alternative Estimates of Exposure
0.07-
0.56
70-560
0.05-
0.37
48-385
0.05-
0.37
48-381
0.04-
0.30
41-324
0.04-
0.30
39-315
0.04-
0.30
40-317
0.04-
0.34
47-376
0.04-
0.32
45-357
0.04-
0.30
43-343
Estimated Cancer Incidences with Hattis and Watson, 1987 Upper Bound Unit Risk (l.lxlO~7
per ug/m3) or ICE, 1986 Upper Bound Unit Risk (3.4xlO~3 per ug/m3) . These are not directly
comparable to the official EPA unit risk estimate.9
0.30
< 1
3691
0.20
< 1
2538
0.20
< 1
2514
0.16
< 1
2137
0.16
< 1
2076
0.16
< 1
2089
0.18
< 1
2477
0.17
< 1
2356
0.16
< 1
2259
Please  refer to  footnotes  on page ES-9.
                                                 ES-6

-------
                                        Table ES-1  Continued.
                                                                                          EPA-420-R-93-005
                                                                                               April 1993

Pollutant
ACETALDEHYDE
1990
Base
Control
1995
Base
Control
Expanded
Reform.
Gasoline
Use
2000
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
2010
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
c,d
Estimated Cancer Incidences with Estimates of Exposure Calculated in this Study.
Estimated cancer incidences are based on the EPA 1987 upper bound unit risk of 2.2xlO"6 per ug/m3, determined
using animal data.
b
EF (g/mi)
c
Exposure
(ug/m3)
d
Cancer Cases
Average of e
Individual Risk

Range of Exposure
(ug/m3)
Range of Cancer
Cases

Exposure (ug/m3)
Cancer Cases
(EPA, 1987)
Cancer Cases
(GARB, 1992)
0.0119
0.67
5.3
2.0xlO"8
0.0071
0.44
3.6
1.4xlO"8
0.0071
0.44
3.6
1.4xlO"8
0.0051
0.33
2.8
l.OxlO"8
0.0051
0.33
2.8
l.OxlO"8
0.0052
0.33
2.8
l.OxlO"8
0.0045
0.34
3.0
l.lxlO"8
0.0044
0.34
3.0
l.lxlO"8
0.0041
0.31
2.8
9.9xlO"9
Estimated Cancer Incidences with Alternative Estimates of Exposure
0.67-
1.71
5.3-
13.4
0.44-
1.12
3.6-
9.1
0.44-
1.12
3.6-
9.1
0.33-
0.84
2.8-
7.1
0.33-
0.84
2.8-
7.1
0.33-
0.84
2.8-
7.1
0.34-
0.87
3.0-
7.6
0.34-
0.87
3.0-
7.6
0.31-
0.79
2.8-
7.1
Estimated Cancer Incidences with EPA 1987 Upper Bound Unit Risk (2.2xl(T6 per }ig/m3) or
GARB, 1992 Upper Bound Unit Risk (2.7xl(T6 per ug/m3)  The GARB, 1992 estimate is not
directly comparable to the official EPA estimate.9
0.67
5.3
6.5
0 .44
3.6
4 .4
0 .44
3.6
4 .4
0.33
2.8
3.4
0.33
2.8
3.4
0.33
2.8
3.4
0.34
3.0
3.7
0.34
3.0
3.7
0.31
2.8
3.4
Footnotes can be  found on page ES-9.
                                                 ES-7

-------
                                       Table  ES-1 Continued.
                                                                                         EPA-420-R-93-005
                                                                                              April 1993

Pollutant
DIESEL PARTICULATE
MATTER
1990
Base
Control
1995
Base
Control
Expanded
Reform.
Gasoline
Use
2000
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
2010
Base
Control
Expanded
Reform.
Gasoline
Use
Expanded
Adoption
Calif.
Stds.
c,d
Estimated Cancer Deaths with Estimates of Exposure Calculated in this Study
Estimated cancer deaths are based on the EPA 1991 draft upper bound unit risk of 1.7xlO"5 per ug/m3, determined
using animal data. This unit risk has not been peer reviewed and is subject to change.
b
EF (g/mi)
c
Exposure
(ug/m3)
d
Cancer Deaths
Average of e
Individual Risk

Exposure (ug/m3)
Cancer Deaths
(Albert and Chen,
1986)
Cancer Deaths
(Harris, 1983)
0.0669
1.80
109
4.4xlO"7
0.0356
1.05
66
2.5xlO"7
_
-
66
2.5xlO"7
0.0188
0.60
39
1.4xlO"7
_
-
39
1.4xlO"7
_
-
39
1.4xlO"7
0.0105
0.39
27
9.6xlO"8
_
-
27
9.6xlO"8
_
-
27
9.6xlO"8
Estimated Cancer Deaths with Albert and Chen, 1986 Upper Bound Unit Risk (1.2xl(T5 per
ug/m3) or Harris, 1983 Upper Bound Unit Risk (4.1x10" per ug/m3) These are not directly
comparable to the draft EPA unit risk estimate.9
1.80
77
26,346
1.05
47
15,967
_
47
15,967
0.60
28
9409
_
28
9409
_
28
9409
0.39
19
6443
_
19
6443
_
19
6443
Footnotes can be  found on the  following page.
                                                ES-8

-------
                                   Footnotes to Table ES-1.                        EPA-420-R-93-oos
                                                                                       April 1993

aThere  are  many inherent  uncertainties in the emission estimates,  exposure,  and dose-response
information that need to be considered when reviewing these results.  These uncertainties are
discussed at the end of the executive summary and in the individual chapters.  Point estimates
are presented due to the difficulty in reporting a range that would accurately bound the
estimates.   The true risk could be as low as zero or even fall above the point estimates in
this table.

bA modified version of  the  MOBILE4.1  emission model,  designated MOBTOX,  was  used to develop
the nationwide emission factors.  The emission factors are roughly 25-40% lower than those
that would be obtained using the current version, MOBILE5a.  The resulting annual average
exposure estimates should not change appreciably, however, since the conversion from g/mile to
ug/m3  is based on CO as a surrogate.   The CO emission factors  with MOBILE5a  relative to
MOBILE4.1 increase roughly in proportion to the toxic emission factors.

Exposures  given are nationwide  annual average estimates.   The HAPEM-MS model was used to
calculate exposures.  Then for each pollutant, the HAPEM-MS derived exposures for 1990 were
compared with the range of available ambient monitoring data  (with adjustments applied to
account for such factors as lower exposure from time spent indoors).  Where the HAPEM-MS
exposures fell outside the range of ambient monitoring data, an adjustment,  based on comparing
the modeled versus ambient data, was applied to the modeled data to match the upper end of the
range.   This adjustment was then applied to the HAPEM-MS derived exposures for all years.  For
1,3-butadiene, the range of ambient data varied by over a factor of four; consequently,
estimates of cancer incidence given here are roughly four times higher than those that would
be calculated using the lower bound.

dThe cancer risk estimates  are based  on plausible upper bound  estimates of unit risk (in
accordance with procedures referenced in the Risk Assessment Guidelines of 1986), except for
benzene.   This is because an established procedure does not yet exist for making "most
likely" or "best" estimates of risk.   The unit risk for benzene is based on human data.  The
cancer risk estimates are meant to be used in a relative sense to compare risks among
pollutants and scenarios, and to assess trends.  They are not meant to represent actual risk.

Estimated  annual individual  risk is  the cancer risk divided by the U.S.  population for the
year of interest.  Since results are presented as national annual averages,  changes in cancer
incidences or deaths presented for the expanded control scenarios do not necessarily represent
changes that would occur in specific areas where the strategies are implemented, such as the
Northeast.
                                             ES-9

-------
fThe  range  of  nationwide annual average exposures is obtained using the results of Era\biacnR-93-oo5
ambient monitoring studies.  The lower end of the range is the lowest annual average  st^sRfly993
result, with an adjustment of 0.89 based on HAPEM-MS to account for nationwide exposure  (i.e.,
incorporating estimated rural exposure), an adjustment applied to account for the motor
vehicle fraction, and an adjustment of 0.622 to account for integrated exposure  (i.e., time
spent indoors at home, indoors at work, outdoors, and in motor vehicles).  The upper  end of
the range is the highest annual average study result, with the nationwide and integrated
exposure adjustments, but without the motor vehicle fraction adjustment.  The motor vehicle
adjustment is removed for the upper end since the relative contributions of motor vehicle and
non-motor vehicle sources are not clear, especially for the nonroad contribution.  The
contribution of motor vehicles is likely to vary significantly from location to  location and
for pollutant to pollutant.

Alternative unit risks  were derived using different sets  of data,  models,  assumptions and
other parameters.  Thus, they are not directly comparable.

hln the 1991 draft EPA formaldehyde  risk assessment,  EPA's Office of Pollution Prevention and
Toxics presented several estimates of risk, the lowest of which is based on DPX  formation in
monkeys and is used in this table.  Each estimate embodies a different set of uncertainties.
Comments by the Science Advisory Board to OPPT strongly recommended that a rigorous discussion
of these uncertainties and how they impact on the confidence for making human risk inferences
be undertaken.  This document remains in draft and the risk estimates have not been adopted by
the agency.  EPA's official unit risk remains the unit risk estimate from EPA, 1987.
                                             ES-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993
reductions of other toxic compounds, like benzene and other
aromatic compounds, that would offset the potential impact from
increased aldehyde emissions.  Uncertainties still remain
regarding health effects from exposure to oxygenated fuels.  Work
is in progress by EPA and others to address this issue.

     Alternative cancer risk estimates are also presented in
Table ES-1 to illustrate the effect of alternative annual average
exposure estimates and unit risk estimates.  The alternative
estimates are not documented in the individual chapters, although
the information used to develop these estimates is extensively
documented.

     First, cancer incidences for 1,3-butadiene, formaldehyde,
and acetaldehyde or cancer deaths for benzene and diesel
particulate matter were adjusted based on a range of annual
average exposures.  The range of nationwide annual average
exposures is obtained using the results of urban ambient
monitoring studies.  The lower end of the range is the lowest
annual average study result, with an adjustment of 0.89 based on
HAPEM-MS to account for nationwide exposure (i.e., incorporating
estimated rural exposure),  an adjustment applied to account for
the motor vehicle fraction, and an adjustment of 0.622 to account
for integrated exposure  (i.e., time spent indoors at home,
indoors at work, outdoors,  and in motor vehicles).  The upper end
of the range is the highest annual average study result, with the
nationwide and integrated exposure adjustments, but without the
motor vehicle fraction adjustment.  The motor vehicle adjustment
is removed for the upper end since the relative contributions of
motor vehicle and non-motor vehicle sources are not clear,
especially for the nonroad contribution.  The contribution of
motor vehicles is likely to vary significantly from location to
location and for pollutant to pollutant.  Annual average
exposures for each toxic from various studies are given in the
individual chapters for each toxic.

     Also, alternative estimates of cancer risks are provided
using the single estimate of exposure from this study, but using
alternative unit risk estimates either from non-EPA organizations
or unapproved EPA estimates.  Both the lowest and highest
alternative unit risk estimates reported in this study were used
to calculate the cancer risks.

     Following is a synopsis of each chapter,  beginning with
Chapter 3.

Emission Factor Methodology

     For benzene, formaldehyde, acetaldehyde,  and 1,3-butadiene,
available vehicle emissions data are used to estimate toxic
emissions as fractions of total organic gases  (TOG).  TOG
includes all hydrocarbons as well as aldehydes, alcohols, and

                              ES-11

-------
                                                        EPA-420-R-93-005
                                                            April 1993
other oxygenated compounds.  These fractions are then applied to
an updated version of MOBILE4.1, designated MOBTOX, developed
specifically to calculate in-use toxic grams per mile emission
factors.  (MOBTOX TOG and toxic estimates are about 25-40% lower
than those that would be obtained using the recently released
current version of the mobile model, MOBILE5a.  As discussed
later,  the overall cancer risks would not change appreciably.)
This approach was used because virtually all the available
emission data are from low mileage, well-maintained vehicles.  To
simply use the g/mile data from these studies directly would
likely result in a large underestimation of true emissions.
Also, available data suggest relatively constant fractions
(toxics/TOG)  independent of TOG emission level.

     For diesel particulate matter, recent analyses performed by
Navistar Corporation were used to predict total grams of urban
diesel particulate matter, as well as national fleet average
emission factors, for base control scenarios in the years 1990,
1995, 2000,  and 2010.  NavistarTs analyses generally agree with
previous but far less comprehensive EPA analyses.  These
predictions utilize the most recent inputs available; thus, the
particulate emission factors derived by Navistar were used with
only minor adjustments to develop diesel particulate matter risk
estimates.  Later, EPA may develop particulate emission factors
to use in developing risk estimates independently.

     For gasoline particulate matter, the available emission data
were reviewed.  The limited data appear to indicate a correlation
between exhaust HC and gasoline particulate matter emissions.
Gasoline particulate matter was thus estimated to be 1.1% of
exhaust HC.   It should be noted, however, that this is extremely
uncertain and subject to change.  This percentage was then used
in the MOBTOX model to calculate in-use g/mile emission factors
for gasoline particulate matter.

Exposure Methodology

     Annual average exposures to toxic air pollutants from motor
vehicles were estimated using a model referred to as the
Hazardous Air Pollutant Exposure Model for Mobile Sources, or
HAPEM-MS, developed by International Technology under an EPA
contract.  The annual average exposures estimated by HAPEM-MS
represent the 50th percentiles of the population distributions of
exposure, i.e., half the population will be above and half below
these values.  HAPEM-MS accounts for time spent indoors and in
various microenvironments.  It uses carbon monoxide  (CO) as a
surrogate for motor vehicle emissions, since the vast majority of
CO comes from motor vehicles.   HAPEM-MS calculates urban and
rural annual average exposure to CO for the year 1988, using data
from fixed site monitors, personal monitoring studies and
personal activity studies.  Fixed site monitor values were
adjusted using microenvironmental CO measurements from personal

                              ES-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993
exposure monitors.  The MOBILE4.1 emissions model was used to
estimate the corresponding CO emission factor  (g/mile) for 1988.
The urban and rural concentrations predicted by HAPEM-MS for 1988
were divided by the 1988 MOBILE4.1 emission factor to get g/mile
to ug/m3 conversion factors for urban and rural areas.   To obtain
exposure estimates for the toxic of interest, these conversion
factors were simply multiplied by the emission factor for the
toxic of interest. An additional adjustment factor was applied to
account for the increase in vehicle miles travelled (VMT) in
excess of the population increase for the year of interest
relative to 1988, since HAPEM-MS does not account for changes in
VMT.

     The premise of the HAPEM-MS model is that the dispersion and
atmospheric chemistry of the toxic of interest is similar to CO.
This premise will not be valid for the more reactive pollutants
such as 1,3-butadiene, in part because such pollutants typically
have significant indoor sinks relative to non-reactive compounds
such as CO.

     Also,  the reliability of the present methodology depends on
the representativeness of the population by 6 cohorts which are
exposed to concentrations within 5 microenvironments.   Based on
the study of available exposure measurements, the upper 10th
percentile of the population exposures is believed to be
underestimated.  The present use of annual average concentrations
to determine cancer risk assumes that the dose-response
relationship is linear.  Improved methodology must be developed
before a non-linear dose-response relationship could be used.
Also, assessing chronic non-cancer effects will require
consideration of a distribution of annual exposures (e.g., the
90th percentile) and not simply the annual mean average.

     If MOBILE5a CO emission factors were used in estimating
g/mile to ug/m3 conversion factors,  the  factors would  be 30-35%
lower.  However, as discussed earlier, the toxic emission factors
using MOBILE5a would be 25-40% higher; thus, the overall cancer
risk estimate would not change appreciably.

     To check the reasonableness of the HAPEM-MS modeling
results, the urban HAPEM-MS concentrations for 1990 were compared
to urban ambient monitoring data for recent years.   Monitoring
data from the EPA Aerometric Information Retrieval System  (AIRS),
the Urban Air Toxic Monitoring Program (UATMP), and the National
Ambient Volatile Organic Compounds  (NAVOC) Data Base were used.
The monitoring data used in this study are annual average
exposures  (arithmetic means) for each database and year.  In
order to directly compare the ambient and modeled concentrations,
the ambient data were adjusted in two ways.  First, the ambient
monitoring data were adjusted to represent the amount that is
attributed to motor vehicles, using emissions inventory
apportionment.  Second, the estimated ambient motor vehicle level

                              ES-13

-------
                                                        EPA-420-R-93-005
                                                            April 1993
was adjusted to account for integrated exposure, i.e., time spent
indoors at home, indoors at work, outdoors, and in motor
vehicles.  The latter 'integrated' adjustment factor was
estimated, based on CO exposure, to be 0.622.  The following
sections on specific air toxics compare the HAPEM-MS modeling
results to the ambient data, using these adjustments.

     Short-term, high level microenvironment exposures are also
addressed and compared to exposures for which non-carcinogenic
health effects have been observed.  For many individuals, the
greatest source of microenvironmental exposure is the personal
garage.  EPA's model for personal garage exposure is presently
being reevaluated; thus, microenvironment exposure in the
following sections focus on available studies where toxics
concentrations have been measured in-transit and in other
microenvironments where elevated levels would be expected.  The
inhalation Reference Concentration (RfC)  methodology provides a
tool making chronic noncancer assessments.  The study reports
RfCs for two pollutants; diesel particulate matter and
acetaldehyde.  New methodology must be developed before risks to
acute exposures can be assessed.

Benzene

     Benzene is a clear, colorless, aromatic hydrocarbon which is
both volatile and flammable.  Benzene is present in both exhaust
and evaporative emissions.  The TOG percentage of benzene in the
exhaust varies depending on control technology and fuel
composition but is generally about 3 to 5%.  The TOG percentage
of benzene in the evaporative emissions also depends on control
technology (e.g., whether the vehicle has fuel injection or a
carburetor)  and fuel composition  (e.g., benzene level and RVP)
and is generally about 1%.  Control techniques are available and
in use for both evaporative and exhaust emissions of benzene.

     Motor vehicles account for approximately 60% of the total
benzene emissions, with the remainder attributed to nonroad
mobile sources  (25%) and stationary sources  (15%).   Many of the
stationary sources are industries producing benzene, sometimes as
a side product, and those industries that use benzene to produce
other chemicals.

     EPA's Total Exposure Assessment Methodology (TEAM) Study
identified the major sources of exposure to benzene for much of
the U.S. population.  The most important source of benzene
exposure is active smoking of tobacco, accounting for roughly
half of the total population exposure to benzene, which is over
and above that from motor vehicles.  Outdoor concentrations of
benzene, due mainly to motor vehicles, account for roughly one-
quarter of the total.  Benzene is the only motor vehicle-related
toxic for which such information exists.
                              ES-14

-------
                                                          EPA-420-R-93-005
                                                              April 1993
     Benzene  is  quite stable in the atmosphere.   The only benzene
reaction which is  important in the lower atmosphere is the
reaction with OH radicals.   Yet even this  reaction is relatively
slow.  The
                               ES-15

-------
                                                        EPA-420-R-93-005
                                                            April 1993
products of this reaction are primarily phenols and aldehydes,
which react quickly and also are removed by incorporation into
rain.  Benzene itself will not be incorporated into clouds or
rain to any large degree because of its low solubility.  Benzene
is not produced by atmospheric reactions.

     Atmospheric residence times for benzene were calculated for
four cities and two seasons.  In the summertime, the daytime
residence times under clear-sky conditions are calculated to be
1-2 days.  Under these conditions,  benzene can be transported far
from source regions.  At night,  benzene can be considered
essentially inert.  Winter residence times in most cases are
greater than summer residence times by roughly a factor of ten.
The presence of cloud cover slows down photochemistry and
increases the residence time for all species.

     Urban Airshed Model simulations for a hypothetical day in
the summer of 1990 in St. Louis demonstrated the role of
atmospheric transformation in determining ambient concentrations
of benzene.  In the case of benzene, atmospheric transformation
was shown to have only a minor effect on ambient concentrations
during afternoon hours, and virtually no effect during other
times of day.  Simulations in the Baltimore-Washington area
indicated that the motor vehicle-related concentration of ambient
benzene would be higher in winter,  due to less atmospheric
transformation.  Simulations in Baltimore-Washington predicted
significant decreases in ambient levels of benzene with use of
reformulated gasoline, on the order of 7 percent.  However,
simulations for the summer Houston episode predicted little
effect on maximum daily average concentration of benzene with use
of reformulated gasoline at the site of maximum concentration.

     The annual average ambient level of benzene ranges from 4.13
to 7.18 ug/m3,  based on urban air monitoring  data.   Applying the
motor vehicle adjustment factor of 0.60 and the integrated
adjustment factor of 0.622, the integrated motor vehicle exposure
is estimated to range from 1.54 to 2.68 ug/m3.   Since  the  HAPEM-
MS 1990 base control number matches the upper end of the range,
the HAPEM-MS 1990 base control level of 2.67 ug/m3  will be used
to estimate cancer deaths.  As a result, the HAPEM-MS exposures
were used as a reasonable estimate of the annual motor vehicle
exposure level of benzene for all scenarios and years.

     Based on the available exposure data, maximum
microenvironment exposure levels to benzene range from 40 ug/m3
from in-vehicle exposure to 288 ug/m3  from exposure during
refueling.  However, information on health effects from short-
term acute exposure to benzene is limited; thus, the impact of
such microenvironmental exposure is difficult to assess.

     Long-term exposure to high levels of benzene in air has been
shown to cause cancer of the tissues that form white blood cells

                              ES-16

-------
                                                        EPA-420-R-93-005
                                                            April 1993
(leukemia), based on epidemiology studies with workers.
Leukemias and lymphomas, as well as other tumor types, have been
observed in experimental animals that have been exposed to
benzene by inhalation or oral administration.  Exposure to
benzene has also been linked with genetic changes in humans and
animals.  Based on this evidence, EPA has concluded that benzene
is a Group A, known human carcinogen.  The International Agency
for Research on Cancer  (IARC) has also classified benzene as a
human carcinogen.  EPA calculated a cancer unit risk factor for
benzene of 8. 3xlO"6 (ug/m3)-1 based on the results of three
epidemiological studies in benzene-exposed workers in which an
increase of death due to nonlymphocytic leukemia was observed.
EPA's Office of Research and Development has just recently
started the process to review and update the benzene risk
assessment.

     Since the benzene cancer risk assessment was conducted by
EPA in 1985, several new epidemiological studies have been
published.  Generally, these studies are updates of the studies
considered by EPA.  The updated studies provide continued
evidence of the carcinogenicity of benzene in humans, and
incorporation of increased study population sizes and improved
exposure analyses in these studies may strengthen the current
cancer risk assessment for benzene.  New animal studies provide
additional support for the carcinogenicity of benzene in animals
by both the oral and inhalation routes and provide the first
animal model for the type of cancer identified most closely with
occupational exposure, acute myelogenous leukemia.

     Recent research has also been conducted on the
pharmacokinetics of benzene.  These studies demonstrate that
species differ with respect to their ability to metabolize
benzene.  These differences may be important when choosing an
animal model for human exposure and when extrapolating high dose
exposures in animals to the low levels of exposure typically
encountered in occupational situations.  The recent development
of a physiologically-based pharmacokinetic model for benzene
should help in performing interspecies and route-to-route
extrapolations of cancer data.  New information on the ability of
benzene to alter the genetic material provides additional support
for the occurrence of this effect with benzene and its
metabolites.  Furthermore, the occurrence of certain chromosomal
aberrations in individuals with known exposure to benzene may
serve as a marker for those at risk for contracting leukemia.

     Alternate views and/or risk assessments generally concur
with EPA's choice of epidemiological data upon which to base the
cancer risk estimate, but differ with respect to the mathematical
models and assumptions used to derive the risk estimate and the
specific tumor incidence and/or exposure data to use.  The CARB
risk estimate is actually a range, with the number calculated by
EPA serving as the lower bound of cancer risk and a more

                              ES-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993
conservative (i.e., higher) number, based on animal data, serving
as the upper bound of cancer risk.  The Clement Associates risk
estimate (conducted for API) is also expressed as a range with
the lower bound two orders of magnitude lower than the unit risk
factor calculated by EPA; the upper bound is still approximately
eight times lower than the EPA unit risk.

     Please note that, unlike the other pollutants addressed in
this study, the cancer unit risk estimate for benzene is based on
human data.  Cancer numbers are expressed as cancer deaths.  The
estimate of cancer deaths may underestimate cancer incidence
associated with benzene,  since survivorship rates are not
included in the supporting studies.  The 1990 base control
scenario estimates the total annual average cancer deaths to be
70 deaths  (59 urban,  11 rural).   When comparing annual cancer
deaths for the base control scenarios relative to 1990, there is
a 39% reduction in 1995,  a 50% reduction in 2000, and a 50%
reduction in 2010.  The reduction in per vehicle emissions is
considerably higher,  particularly in the later years.  The
projected increase in both population and vehicle miles traveled
(VMT) from 2000 to 2010 appears to offset the gains in emissions
reduction achieved through fuel and vehicle modifications.

     The base control and expanded use scenarios within each year
can be directly compared since the same VMT and populations are
applied to both.  In 1995, expanding the reformulated gasoline
program reduces the cancer deaths by another 8% from the 1990
base control.  The expanded use of reformulated fuels and the
expanded adoption of the California program in the year 2000
produces another 6% reduction in cancer deaths, for both
scenarios,  when compared to 1990.  Expanded reformulated gasoline
use in 2010 reduces the cancer deaths by 6% relative to 1990 and
by approximately 10% for the expanded adoption of California
standards scenario.  Like the base case comparison, the cancer
deaths for the control scenarios are similar for 2000 and 2010
despite continued emissions reduction, due to the projected
population and VMT increase.

     A number of adverse noncancer health effects have also been
associated with exposure to benzene.  Benzene is known to cause
disorders of the blood.  People with long-term exposure to
benzene at levels that generally exceed 50 ppm (162,500 ug/m3)
may experience harmful effects on the blood-forming tissues,
especially the bone marrow.  These effects can disrupt normal
blood production and cause a decrease in important blood
components, such as red blood cells and blood platelets, leading
to anemia and a reduced ability to clot.  Exposure to benzene at
comparable or even lower levels can be harmful to the immune
system, increasing the chance for infection and perhaps lowering
the body's defense against tumors by altering the number and
function of the body's white blood cells.   In studies using
animals, inhalation exposure to benzene may also indicate that it

                              ES-18

-------
                                                        EPA-420-R-93-005
                                                            April 1993
is a developmental and reproductive toxicant.  Studies with
pregnant animals show that breathing 10-300 ppm  (32,500-975,000
ug/m3)  of  benzene  has  adverse effects  on the developing fetus,
including low birth weight, delayed bone formation, and bone
marrow damage.

Formaldehyde

     Formaldehyde is a colorless gas at normal temperatures and
is the simplest member of the family of aldehydes.  Formaldehyde
gas is soluble in water, alcohols, and other polar solvents.
Formaldehyde is the most prevalent aldehyde in motor vehicle
exhaust and is formed from incomplete combustion of the fuel.
Formaldehyde is emitted in the exhaust of both gasoline and
diesel-fueled vehicles.  It is not a component of evaporative
emissions.  Use of a catalyst has been found to be effective for
controlling formaldehyde emissions.  The TOG percentage of
formaldehyde in motor vehicle exhaust varies from roughly 1 to 4
percent depending on control technology and fuel composition.

     The motor vehicle contribution to ambient formaldehyde
levels contains both primary  (i.e., direct emissions) and
secondary formaldehyde  (i.e., formed from photooxidation of
volatile organic compounds, or VOCs).   It appears that roughly
33% of formaldehyde in the ambient air may be attributable to
motor vehicles.  This was calculated based on the results of
various studies using the following apportionment:  30% primary
formaldehyde in the ambient air of which 28% is from motor
vehicles and 70% secondary formaldehyde in the ambient air of
which 35% is due to motor vehicles.  Formaldehyde is produced in
the U.S. by 13 chemical companies in 46 locations encompassing 18
states and it is used in the manufacture of four major types of
resins.  In addition,  formaldehyde is produced as a by-product in
the following types of processes:  combustion (mobile,
stationary, and natural sources), petroleum refinery catalytic
cracking and coking, phthalic anhydride production, asphaltic
concrete production, and atmospheric photooxidation of unburned
hydrocarbons.

     Formaldehyde exhibits extremely complex atmospheric
behavior.   It is present in emissions but is also formed by the
atmospheric oxidation of virtually all organic species.  It is
ubiquitous in the atmosphere because it is formed in the
atmospheric oxidations of methane and biogenic hydrocarbons.
Formaldehyde is photolyzed readily, and its photolysis is an
important source of photochemical radicals in urban areas.  It is
also destroyed by reaction with OH.  An important carbon-
containing product of all gas-phase formaldehyde reactions is
carbon monoxide.  Because formaldehyde is often the dominant
source of radicals in urban atmospheres, formaldehyde
concentrations have a feedback effect on the chemical residence
time of other atmospheric species.  Formaldehyde is highly water

                              ES-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993
soluble and participates in a complex set of chemical reactions
within clouds.  The product of the aqueous-phase oxidation of
formaldehyde is formic acid.

     Atmospheric residence times for formaldehyde were calculated
for four U.S. cities and two seasons.  In the summertime, the
daytime residence times under clear-sky conditions are calculated
to be 2-4 hours for formaldehyde.  Winter residence times in most
cases are greater than summer residence times by roughly a factor
of ten.  The presence of cloud cover slows down photochemistry
and increases the residence time for all species, although the
increase for formaldehyde is partially offset by its rapid
in-cloud destruction due to its high water solubility.  The
physical removal processes of wet and dry deposition are
important for formaldehyde, especially under wintertime
conditions.  Scavenging by falling raindrops will result in
formaldehyde residence times of an hour or less in colder
seasons.

     Urban Airshed Model simulations for a hypothetical day in
the summer of 1990 in St. Louis demonstrated the role of
atmospheric transformation in determining ambient concentrations
of formaldehyde.  The UAM simulation showed that simulated
formaldehyde concentrations were about twice as high as they
would be in the absence of photochemical reactions, indicating
that formaldehyde is formed more rapidly than it is destroyed in
urban areas in the summertime.  The simulation demonstrated that
the component of the concentration due to primary emissions is
small relative to the component due to secondary formation in the
atmosphere.  Simulations for the summer Baltimore-Washington area
episode resulted in both increases and decreases in ambient
formaldehyde with use of federal reformulated gasoline, with
increases due to increased primary formaldehyde in near-source
areas,  and decreases due to decreased secondary formaldehyde in
downwind areas.  Use of California reformulated gasoline resulted
in a decrease in secondary formaldehyde nearly three times as
large as in federal reformulated gasoline scenarios, with similar
primary formaldehyde increases.  Simulations for the winter
Baltimore-Washington area episode resulted in slight increases in
ambient levels of formaldehyde with the use of federal
reformulated gasoline, on the order of 1-2 percent, with a
primary formaldehyde increase and a secondary formaldehyde
decrease.  Simulations for the summer Houston episode predicted
slight increases in the simulated daily average concentration
throughout most of the domain with use of federal reformulated
gasoline.

     The annual average ambient level of formaldehyde will be
taken from the 1990 UATMP data since it is the only program that
accounted for the interference of ozone in the measurement
method.  The resulting 1990 UATMP level is 1.71 ug/m3.   Applying
the motor vehicle adjustment factor of 0.33 and the integrated

                              ES-20

-------
                                                        EPA-420-R-93-005
                                                            April 1993
adjustment factor of 0.622, the integrated motor vehicle exposure
is estimated to be 1.06 ug/m3.   The HAPEM-MS 1990 base control
exposure level of 1.25 ug/m3 must  be multiplied by a factor of
0.848 to agree with the ambient data.  All HAPEM-MS derived
exposure levels will have this factor applied.

     Any formaldehyde exposures projected by HAPEM-MS itself
should be viewed with caution.  The adjusted HAPEM-MS exposure
estimates attempt to account for both primary and secondary
formaldehyde; however, these estimates are based only on changes
in primary emissions of formaldehyde.  The reactivity of motor
vehicle VOC emissions is likely to change with technology and
fuel changes.  Changes in the reactivity of these emissions,
which would result in changes to secondary formaldehyde levels,
cannot be accounted for by HAPEM-MS.

     Based on available exposure data, maximum microenvironment
exposure levels range from 4.9 ug/m3 from exhaust exposure at a
service station to 41.8 ug/m3  from parking garage exposure.
Formaldehyde is a known human irritant for the eyes, nose, and
upper respiratory system at acute exposure levels as low as 62
ug/m3,  though levels  below this  are not  necessarily free from
risk.  Studies in experimental animals provide sufficient
evidence that long-term inhalation exposure to formaldehyde
causes an increase in the incidence of squamous cell carcinomas
of the nasal cavity.   Epidemiological exposure studies suggest
that long-term inhalation of formaldehyde may be associated with
tumors of the nasopharyngeal cavity, nasal cavity, and sinus.
Based on this information, EPA has classified formaldehyde as a
Group Bl, probable human carcinogen.  IARC concurs that
formaldehyde is probably carcinogenic to humans.  EPA calculated
the present, and still official, cancer unit risk factor of
1.3xlO"5  (ug/m3)"1 for formaldehyde based on the results of a study
in rats in which an increase in the incidence of nasal tumors was
observed.  In a 1990 update of this 1987 cancer risk assessment
(still in draft),  EPA modified the cancer risk estimate to 6xlO"7
(ug/m3)"1 by incorporating recent data on the quantification of
DNA-protein cross-links  (DPX)  caused by formaldehyde in monkey
nasal tissue.  The binding of DNA to protein to which
formaldehyde is bound, forming a separate entity that can be
quantified, is considered a more accurate way to measure the
amount of formaldehyde that is present inside a tissue.  Cancer
incidence estimates in this study use the 1987 unit risk factor,
since the updated one is still not an official estimate and may
change.

     Please note that the cancer unit risk estimate for
formaldehyde is based on animal data and is considered an upper
bound estimate for human risk.  True human cancer risk may be as
low as zero.
                              ES-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Several studies in experimental animals have been published
since EPA conducted the cancer risk assessment for formaldehyde
in 1987.  These studies confirm the previous findings of an
increased incidence of squamous cell carcinomas of the nasal
cavity in rats exposed by inhalation.  In addition, the
distribution of nasal tumors in rats has been better defined; the
findings suggest that not only regional exposure but also local
tissue susceptibility may be important for the distribution of
formaldehyde-induced tumors.  Recent epidemiological studies
provide additional evidence that "modest" increases in
nasopharyngeal and nasal cavity and sinus cancer risks, and
possibly in lung cancer risks,  have been observed among various
occupational subgroups.  However, the evidence for an association
between lung cancer and occupational formaldehyde is tenuous, and
collectively,  the recent studies do not conclusively demonstrate
a causal relationship between cancer and exposure to formaldehyde
in humans.

     Recent work on the pharmacokinetics of formaldehyde has
focused on the validation of measurement of DNA-protein adducts,
or cross-links (DPX) as internal dosimeters of formaldehyde
exposure (as discussed above).   An internal dosimeter for
formaldehyde exposure is desirable because the inhaled
concentration of formaldehyde may not reflect actual tissue
exposure levels.   The difference in inhaled concentration and
actual tissue exposure level is due to the action of multiple
defense mechanisms that act to limit the amount of formaldehyde
that reaches cellular DNA.  These studies have provided more
accurate data with which to quantify the level of formaldehyde in
the cell.

     Alternate views and risk assessments have been published for
formaldehyde which all use the same rat data, but differ with
respect to the mathematical models and assumptions used to
extrapolate from animals to humans and the methods used to
estimate internal formaldehyde dose.  When using only the rat
data, the 1992 CARB unit risk factor delineates the lower bound
of risk factors,  approximately 50 percent lower than the present
EPA factor, whereas, OSHA's unit risk factor, as the upper bound,
is over three orders of magnitude greater than the EPA's.

     The 1990 base control scenario estimates the total annual
cancer incidence to be 44 cancer cases (37 urban, 7 rural).  When
comparing cancer incidence for the base control scenarios
relative to 1990, there is a 36% reduction in 1995, a 52%
reduction in 2000, and a 50% reduction in 2010.  The reduction in
per vehicle emissions is considerably higher, particularly in the
out years.   The projected increase in both population and vehicle
miles traveled (VMT) from 2000 to 2010 appears to offset the
gains in emissions achieved through fuel and vehicle
modifications.
                              ES-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     The expanded use scenarios provide either no decrease or a
slight increase in the cancer cases.  This is generally due to
the fact that increased use of oxygenates in gasoline will
increase direct formaldehyde emissions.

     Noncancer adverse health effects associated with exposure to
formaldehyde in humans include irritation of the eyes and nose
(0.1-1.0 ppm or 123-1230 ug/m3),  throat (0.05-2.0 ppm or 62-2,460
ug/m3) ,  and lower airway at low levels (5.0-30  ppm or 6,150-
36,900 ug/m3).   There is also suggestive,  but not conclusive,
evidence in humans that formaldehyde can affect immune function.
Adverse effects on the liver and kidney have also been noted in
experimental animals exposed to higher levels of formaldehyde.

1,3-Butadiene

     1,3-Butadiene is a colorless, flammable gas at room
temperature, is insoluble in water, and its two conjugated double
bonds make it highly reactive.  1,3-Butadiene is formed in
vehicle exhaust by the incomplete combustion of the fuel and is
assumed not to be present in vehicle evaporative and refueling
emissions.  1,3-Butadiene emissions appear to increase roughly in
proportion to exhaust hydrocarbon emissions.  Since hydrocarbons
are decreased by the use of a catalyst on a motor vehicle, 1,3-
butadiene emissions are expected to decrease proportionally.  The
TOG percentage of 1,3-butadiene in motor vehicle exhaust varies
from roughly 0.4 to 1.0 percent depending on control technology
and fuel composition.
     Current EPA estimates indicate that mobile sources account
for approximately 94% of the total 1,3-butadiene emissions.  The
remaining 1,3-butadiene emissions  (6%) come from stationary
sources mainly related to industries producing 1,3-butadiene and
those industries that use 1,3-butadiene to produce other
compounds.  Approximately 59% of the mobile source 1,3-butadiene
emissions  (56% of total 1,3-butadiene emissions) can be
attributed to onroad motor vehicles, with the remainder
attributed to nonroad mobile sources.
     1,3-Butadiene is transformed rapidly in the atmosphere.
There are three chemical reactions of 1,3-butadiene which are
important in the ambient atmosphere:  reaction with hydroxyl
radical (OH),  reaction with ozone  (03) ,  and reaction with
nitrogen trioxide radical  (N03) .   All  three of  these reactions
are relatively rapid, and all produce formaldehyde and acrolein,
species which are themselves toxic and/or irritants.  The
oxidation of 1,3-butadiene by N03 produces organic nitrates as
well.  Incorporation of 1,3-butadiene into clouds and rain will
not be an important process due to the low solubility of
1,3-butadiene.   1,3-Butadiene is probably not produced by atmo-
spheric reactions.

     Atmospheric residence times were calculated for 1,3-
butadiene for four U.S. cities and two seasons.  In the

                              ES-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993
summertime, the daytime residence times under clear-sky
conditions are calculated to be one hour or less for
1,3-butadiene.  Under these conditions, 1,3-butadiene will
generally be present in high concentrations only near source
regions.  At night, the residence times for 1,3-butadiene remain
short under conditions conducive to the formation of N03 (high
03,  high N02, low NO), but increase dramatically under low N03
conditions.  Winter residence times in most cases are greater
than summer residence times by roughly a factor of ten.  The
residence time of 1,3-butadiene can exceed one day in the winter-
time, especially if clouds are present.  The presence of cloud
cover slows down photochemistry and increases the residence time.

     Urban Airshed Model simulations for a hypothetical day in
the summer of 1990 in St. Louis demonstrated the role of
atmospheric
transformation in determining ambient concentrations of
1,3-butadiene.  The afternoon concentration of 1,3-butadiene was
reduced by 90 percent due to atmospheric reactions.  Simulations
for the summer Baltimore-Washington area episode resulted in
little change in ambient concentrations of 1,3-butadiene with the
use of federal reformulated gasoline.  Use of California
reformulated gasoline also had little impact on ambient
concentrations of 1,3-butadiene.   Reformulated gasoline use had
very little effect on winter 1,3-butadiene ambient
concentrations.  Simulations for the summer Houston episode also
predicted little effect on maximum daily average concentration of
1,3-butadiene with reformulated gasoline.

     The annual average ambient level of 1,3-butadiene ranges
from 0.12 to 0.56 ug/m3.   Applying  the  motor vehicle adjustment
factor of 0.56 and the integrated adjustment factor of 0.622, the
integrated motor vehicle exposure is estimated to range from 0.08
to 0.35 ug/m3.   The HAPEM-MS  1990 base  control  level of  0.48
ug/m3 lies  above this  range.   The HAPEM-MS 1990  base control
level must be multiplied by a factor of 0.729 to agree with the
upper end of the ambient data.  All the HAPEM-MS derived exposure
levels have this factor applied.

     Based on a single study, in-vehicle exposure to 1,3-
butadiene was found to average 3.0 ug/m3.   Since data on non-
cancer health effects of acute 1,3-butadiene exposure are very
limited, the impact of microenvironmental exposure is difficult
to assess.

     Long-term inhalation exposure to 1,3-butadiene has been
shown to cause tumors in several organs in experimental animals.
Studies in humans exposed to 1,3-butadiene suggest that this
chemical may cause cancer.  These epidemiological studies of
occupationally exposed workers are inconclusive with respect to
the carcinogenicity of 1,3-butadiene in humans,  however, because
of a lack of adequate exposure information and concurrent

                              ES-24

-------
                                                        EPA-420-R-93-005
                                                             April 1993
exposure to other potentially carcinogenic  substances.   Based on
the inadequate human evidence and sufficient  animal  evidence,  EPA
has concluded that 1,3-butadiene is a Group B2, probable human
carcinogen.  IARC has classified 1,3-butadiene  as  a  Group 2A,
probable human carcinogen.  EPA calculated  a  cancer  unit risk
factor of 2.8xlO"4  (ug/m3)"1 for 1,3-butadiene  based on  the results
of a study in mice in which an increase  in  the  incidence of
tumors in the lung and blood vessels of  the heart, as  well as
lymphomas were observed.  A special factor  was  incorporated  into
these calculations to account for the actual  amount  of 1,3-
butadiene that is absorbed following inhalation.   EPA's Office of
Research and Development has just recently  started the process of
updating the 1,3-butadiene risk assessment.
                              ES-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Please note that the cancer unit risk estimate for 1,3-
butadiene is based on animal data and is considered an upper
bound estimate for human risk.  True human cancer risk may be as
low as zero.

     Since EPA conducted its cancer risk assessment for 1,3-
butadiene in 1985, several updates of the epidemiology studies
considered by EPA and one new study in humans have been
published.  These studies collectively show positive, though
limited evidence that 1,3-butadiene may be carcinogenic in
humans.  A new inhalation study was conducted in mice because the
study used by EPA in 1987 was limited due to high mortality
occurring early in the study.  The new study demonstrates the
occurrence of cancer in mice at additional sites at lower
concentrations of 1,3-butadiene than those used to derive the
cancer unit risk factor.

     Studies in animals also indicate that 1,3-butadiene can
alter the genetic material.  Recent studies on the genotoxic
potential of 1,3-butadiene confirm the ability of 1,3-butadiene
to cause these effects.  Recent studies on the fate of 1,3-
butadiene in the body have focused on the mechanism behind the
differences in carcinogenic responses seen between species.
Recent pharmacokinetic research has found marked differences
among mice, rats, and human tissue preparations in their ability
to metabolize 1,3-butadiene and its metabolites.  The results
suggest that the effective internal dose of DNA-reactive
metabolites may be less in humans than in mice for a given level
of exposure.

     Alternate views and/or risk assessments that have been
published for 1,3-butadiene differ with respect to the
mathematical models and assumptions used to extrapolate from
animals to humans, the methods used to estimate internal 1,3-
butadiene dose, and the specific tumor incidence data to use.
The cancer unit risks range from the one calculated by EPA based
on pooled female mouse tumors which represents the upper bound of
unit risk estimates,  to the unit risk calculated by Hattis and
Watson, 1987, based on total tumors in male rats, which is
approximately 2500 times lower than the EPA estimate.

     The 1990 base control scenario estimates the total annual
cancer incidence to be 304 cancer cases (258 urban, 46 rural).
When comparing cancer incidence for the base control scenarios
relative to 1990, there is a 31% reduction in 1995, a 42%
reduction in 2000, and a 33% reduction in 2010, which is actually
an increase when compared to 2000.  The reduction in per vehicle
emissions is considerably higher, particularly in the later
years.  The projected increase in both population and vehicle
miles traveled  (VMT)  from 2000 to 2010 appears to offset the
gains in emissions achieved through fuel and vehicle
                              ES-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993
modifications.  The expanded use scenarios provide little
additional reduction in the cancer cases.

     Exposure to 1,3-butadiene is also associated with adverse
noncancer health effects.  Exposure to high levels (on the order
of hundreds to thousands ppm) of this chemical for short periods
of time can cause irritation of the eyes, nose, and throat, and
exposure to very high levels can cause effects on the brain
leading to respiratory paralysis and death.  Studies of rubber
industry workers who are chronically exposed to 1,3-butadiene
suggest other possible harmful effects including heart disease,
blood disease, and lung disease.  Studies in animals indicate
that 1,3-butadiene at exposure levels of greater than 1,000 ppm
(2.2xl06 ug/m3) may adversely affect the blood-forming organs.
Reproductive and developmental toxicity has also been
demonstrated in experimental animals exposed to 1,3-butadiene at
levels greater than 1,000 ppm.

Acetaldehyde

     Acetaldehyde is a saturated aldehyde that is a colorless
liquid and volatile at room temperature.  Both the liquid and the
vapors are highly flammable.  Acetaldehyde as a liquid is lighter
than water, and the vapors are heavier than air.  It is soluble
in water.  Acetaldehyde is found in motor vehicle exhaust and is
formed as a result of incomplete combustion of the fuel.
Acetaldehyde is emitted in the exhaust of both gasoline and
diesel-fueled vehicles.  It is not a component of evaporative
emissions.  Use of a catalyst has been found to be effective for
controlling formaldehyde and other aldehyde emissions.
Acetaldehyde emissions are presumed to be controlled to roughly
the same extent as total hydrocarbon emissions with a catalyst.
The TOG percentage of acetaldehyde in motor vehicle exhaust
varies from roughly 0.4 to 1.0 percent depending on control
technology and fuel composition.

     The motor vehicle contribution to ambient acetaldehyde
levels contains both primary and secondary acetaldehyde.  Data
from emission inventories and atmospheric modeling indicate that
roughly 39% of ambient acetaldehyde levels may be attributable to
motor vehicles.  Acetaldehyde is ubiquitous in the environment
and is naturally released.  It is a metabolic intermediate of
higher plant respiration and alcohol fermentation.  It is also
found in many flowers, herbs, and fruits and could be available
for release to the ambient air.  Acetaldehyde is also produced
from aliphatic and aromatic hydrocarbon photooxidation reactions.
Acetaldehyde is formed as a product of incomplete wood combustion
in residential fireplaces and woodstoves and is released into the
atmosphere by the coffee roasting process.  Together these two
processes accounted for 78% of the national primary acetaldehyde
emissions.  Manufacturing plants that produce acetaldehyde also
                              ES-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993
emit acetaldehyde, as do manufacturing plants that produce
ethanol, phenol, acrylonitrile, and acetone.

     The atmospheric chemistry of acetaldehyde is similar in many
respects to that of formaldehyde.  Like formaldehyde, it can be
both produced and destroyed by atmospheric chemical
transformation.  However, there are important differences between
the two.  Acetaldehyde photolyzes, but much more slowly than
formaldehyde.  Acetaldehyde reacts with OH and N03  radicals,  and
produces formaldehyde and peroxyacetyl nitrate (PAN) as reaction
products.  Acetaldehyde is also significantly less water soluble
than formaldehyde.

     Atmospheric residence times for acetaldehyde were calculated
for four U.S. cities and two seasons.  In the summertime, the
daytime residence times under clear-sky conditions are calculated
to be 5 hours or less for acetaldehyde.  At night,  the calculated
residence time of acetaldehyde ranges from 18 hours for Los
Angeles to 7 days for St. Louis.  Under cloudy-sky conditions,
residence times increased.  The resulting climatological average
residence times for July were 6 to 11 hours for acetaldehyde.  In
the wintertime, calculated daytime, clear-sky residence times
were longer, in the range of 20 to 60 hours for acetaldehyde, and
relatively inert at night.  The resulting climatological average
residence times for January were 3 to 8 days.

     Urban Airshed Modeling simulations for a summer day in 1990
in St. Louis demonstrated the role of atmospheric transformation
in determining concentrations of ALD2  (an aldehyde surrogate
species composed of acetaldehyde, higher aldehydes, and lower
reactivity olefins with internal double bonds).  In near-source
areas of the modeling domain, ALD2 behaved as a primary species,
with concentration peaks in the early morning and early evening.
In downwind areas, however, ALD2 behaved as a secondary species,
with concentration peaks in the midafternoon.  The simulation
suggested that motor vehicles may be a more important contributor
to ambient acetaldehyde than they are to formaldehyde levels.

     For Baltimore-Washington and Houston area simulations,
primary and secondary acetaldehyde were modeled explicitly.
Simulations for the summer Baltimore-Washington area episode
resulted in decreases in ambient acetaldehyde with the use of
reformulated gasoline, with little change in primary acetaldehyde
and decreased secondary acetaldehyde throughout the domain.  Use
of California reformulated gasoline resulted in a decrease in
secondary acetaldehyde roughly twice as large as in federal
reformulated gasoline scenarios.  In winter, motor vehicle-
related acetaldehyde emissions were about the same with
reformulated gasoline use.  Simulations for the summer Houston
episode predicted slight decreases in simulated daily average
concentration of acetaldehyde throughout most of the domain with
use of reformulated gasoline.

                              ES-28

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     The annual average ambient level of acetaldehyde is based on
only the 1990 UATMP data due to a potential measurement method
ozone interference problem with the other ambient databases.  The
1990 UATMP annual average exposure of 3.10 ug/m3 will be used for
the comparison to HAPEM-MS.   Applying the motor vehicle
adjustment factor of 0.39 and the integrated adjustment factor of
0.622,  the integrated motor vehicle exposure is estimated to be
0.75 ug/m3.   When compared to the  HAPEM-MS  1990 base control
level of 0.36 ug/m3,  the 1990 UATMP  adjusted ambient level  is
observed to be approximately two times greater than the HAPEM-MS
base control level.  The HAPEM-MS 1990 base control exposure
level of 0.36 ug/m3 must be  increased by a  factor of 2.09,  to
0.75 ug/m3  to agree with the ambient data.   The HAPEM-MS derived
exposure levels have this factor applied.

     Any acetaldehyde exposures projected by HAPEM-MS itself
should be viewed with caution.  The adjusted HAPEM-MS exposure
estimates attempt to account for both primary and secondary
acetaldehyde; however,  these estimates are based only on changes
in primary emissions of acetaldehyde.  However, the reactivity of
motor vehicle VOC emissions is likely to change with technology
and fuel changes.  Changes in the reactivity of these emissions,
which would result in changes to secondary acetaldehyde levels,
cannot be accounted for by HAPEM-MS.

     There is sufficient evidence that acetaldehyde produces
cytogenic damage in cultured mammalian cells.  Although there are
only three studies in whole animals, they suggest that
acetaldehyde produces similar effects in vivo.  Thus, the
available evidence indicates that acetaldehyde is mutagenic and
may pose a risk for somatic cells (all body cells excluding the
reproductive cells).  Current knowledge, however, is inadequate
with regard to germ cell  (reproductive cell) mutagenicity because
the available information is insufficient to support any
conclusions about the ability of acetaldehyde to reach mammalian
gonads and produce heritable genetic damage.

     Studies in experimental animals provide sufficient evidence
that long-term inhalation exposure to acetaldehyde causes an
increase in the incidence of squamous cell carcinomas of the
nasal cavity.  In one epidemiological study, with occupationally
exposed workers, the evidence was inadequate to suggest that
long-term inhalation of acetaldehyde may be associated with an
increase in total cancers.  Based on this information, EPA has
classified acetaldehyde as a Group B2, probable human carcinogen.
IARC has classified acetaldehyde as a Group 2B, possible human
carcinogen. EPA calculated the cancer unit risk factor of 2.2x10"
6  (ug/m3)"1  for acetaldehyde  based  on the  results  of  the  two
studies in rats in which an increase in the incidence of nasal
tumors was observed.
                              ES-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Please note that the cancer unit risk estimate for
acetaldehyde is based on animal data and is considered an upper
bound estimate for human risk.  True human cancer risk may be as
low as zero.

     An alternate view and/or risk assessment has been published
by CARB as a preliminary draft for acetaldehyde and differs with
respect to the mathematical model and assumptions used to
extrapolate from animals to humans.  CARB, like EPA, has
concluded that acetaldehyde is a probable human carcinogen.  The
UCL for unit risk for lifetime exposure calculated by CARB is
4.8xlO"6 ppb"1  (2.7xlO~6 [ug/m3]"1).   CARB  also  calculated  a range
of UCL for unit risks.  This range is 9.7xlO"7 ppb"1  for  female
rats without a scaling factor to 2.7xlO"5 ppb"1  for male  rats  with
a contact area correction (1.19xlO~6 to  3.32xlO"5  [ug/m3]"1).

     Since the acetaldehyde cancer risk assessment was conducted
by EPA in 1987, little new research in whole animals and
epidemiological studies have been accomplished.

     The 1990 base control scenario estimates the total annual
cancer incidence to be 5.3 cancer cases  (4.5 urban,  0.8 rural).
Cancer cases are presented here to one decimal place due to the
small numbers involved.  When compared to the 1990 base control,
the cancer incidence decreases by 32% in 1995, 47% in 2000, and
43% in 2010, which is actually an increase when compared to 2000.
The reductions are basically due to the tighter tailpipe
standards specified by the Tier 1 standards.  In contrast, when
compared to the 1990 base control, the emission factors decrease
32% in 1995, 57% in 2000 and 62% in 2010.  The difference
observed between the emission factor and cancer case reductions,
and the increases observed in 2010, is due to the expected
increase in population and VMT, which appear to offset the
emission gains achieved through fuel and vehicle modifications.

     The expanded use of reformulated gasoline and the expansion
of the California standards provide no significant decrease in
the cancer cases and, in several scenarios, the cancer cases
increase.

     The new genotoxicity studies, which utilize lower
concentrations of acetaldehyde, have not produced chromosomal
aberration and/or cellular mutations.

     Non-cancer effects in studies with rats and mice showed
acetaldehyde to be moderately toxic by the inhalation route,
oral, and intravenous routes.  Acetaldehyde is a sensory irritant
that causes a depressed respiration rate in mice.  In rats,
acetaldehyde increased blood pressure and heart rate after
exposure by inhalation.  The primary acute effect of human
exposure to acetaldehyde vapors is irritation of the eyes, skin,
and respiratory tract  (135 ppm for 30 minutes).  At low levels of

                              ES-30

-------
                                                        EPA-420-R-93-005
                                                            April 1993
exposure (concentrations up to 100 ppm in air) ,  inhaled
acetaldehyde is rapidly absorbed and metabolized.  At high
concentrations (>100 to 200 ppm),  irritation and ciliastatic
effects can occur, which could facilitate the uptake of other
contaminants.  Clinical effects include reddening of the skin,
coughing, swelling of the pulmonary tissue, and localized tissue
death.  Respiratory paralysis and death have occurred at
extremely high concentrations.  It has been suggested that
voluntary inhalation of toxic levels of acetaldehyde would be
prevented by its irritant properties, since irritation occurs at
levels below 200 ppm (360,000 ug/m3).

     Acetaldehyde is only one of two air toxics in this study
with a reference concentration for chronic inhalation exposure
(RfC) .  This RfC was recently determined to be 9xlO"3 mg/m3 (9.0
ug/m3 or 5xlO"3 ppm) .   An RfC is an estimate of the continuous
exposure to the human population that is likely to be without
deleterious effects during a lifetime.  As such, it is useful in
evaluating non-cancer effects.  The RfC was determined based on
studies done with male rats, which indicated a NOAEL (no-
observed-adverse-effect-level) of 150 ppm.

     Based on a single study, the in-vehicle exposure level of
acetaldehyde was found to average 13.7 ug/m3  (7.6xlO~3 ppm).   The
average in-vehicle exposure level from the above study is higher
than EPA's RfC.  However, the RfC is based on continuous exposure
whereas the level observed in the study is short-term in
duration.

     The research into reproductive and developmental effects of
acetaldehyde is based on intraperitoneal injection, intravenous,
or oral administration of acetaldehyde to rats and mice, and also
in vitro studies.  However, little or no research into effects of
inhalation of acetaldehyde on reproductive and development
effects was found.  The in vivo and in vitro studies provide
evidence to support the fact that acetaldehyde may be the
causative factor in birth defects observed in fetal alcohol
syndrome.

Diesel Particulate Matter

     Diesel exhaust particulate matter consists of a solid core
composed mainly of carbon, a soluble organic fraction,  sulfates,
and trace elements.  Light-duty diesel engines emit from 30 to
100 times more particles than comparable catalyst-equipped
gasoline vehicles.  Diesel particulate matter is mainly
attributable to the incomplete combustion of fuel hydrocarbons.
Lubricating oil also contributes significantly to diesel
particulate matter. Some may be due to other fuel components as
well.  The particles may also become coated with adsorbed and
condensed high molecular weight organic compounds.
                              ES-31

-------
                                                        EPA-420-R-93-005
                                                             April 1993
     The control of diesel emissions  can  take  three  forms.   The
first is controlling emissions before they  are formed with  engine
modifications  (such as altered combustion chamber shape,  modified
injection systems, or improved engine manufacturer specifications
and engine seals to reduce the contribution of lubricating  oil).
Such modifications are in various  stages  of development.  A
second way to control emissions  is  to add aftertreatment
technologies to the exhaust system.   A third way to  control
emissions is by reformulation of diesel fuel.
                              ES-32

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Diesel particulate matter itself has not been explicitly
modeled to determine its atmospheric transformation and residence
times.  Residence time calculations have been done with
hypothetical non-reactive particulate-phase polycyclic organic
matter (POM) for four U.S. cities and two seasons.  The residence
time calculated for this hypothetical particle under clear-sky
summer conditions was 60 hours.  In the winter, the residence
time increases to 120 hours.  Under rainy conditions, residence
times decreased dramatically for all POM that are particle based
ranging from 0.5 to 4 hours.  A climatological average of 12 to
70 hours was determined for the non-reactive particulate-phase
POM.

     The explicit Urban Airshed Modeling of the non-reactive
particulate-phase POM is difficult to achieve due to the inherent
complexity of diesel emissions itself.  Major consideration needs
to be given to the relative abundance of the various POM species
in the atmosphere, the availability of emissions data, and
determining an area's specific area, mobile, and point sources.
Due to these many considerations and parameters, and the absence
of software to implement these factors, Urban Airshed Modeling
was not done for diesel particulate matter in St. Louis.
However,  POM was treated explicitly in the Baltimore-Washington
and Houston area studies.

     To obtain urban and rural annual average exposures, urban
diesel particulate matter national fleet average emission factors
were first multiplied by the urban and rural g/mile to ug/m3
conversion factors obtained from HAPEM-MS for 1988.  This
provides an estimate of urban and rural exposure relative to the
number of vehicle miles travelled (VMT) in 1988.  To obtain
exposure estimates for the years of interest, these values were
then multiplied by incremental adjustments to allow for the VMT
increase in excess of the population increase for the year of
interest.  Resulting nationwide annual average exposures range
from 1.80 to 0.39 ug/m3, for the  period 1990 to 2010.   HAPEM-MS
exposure estimates compare well to adjusted ambient data;
therefore, no further adjustment was made to the modeled data.

     Studies in experimental animals provide sufficient evidence
that long-term inhalation exposure to high levels of diesel
exhaust causes an increase in the induction of lung tumors in two
strains of rats and two strains of mice.  In two key
epidemiological studies on railroad workers occupationally
exposed to diesel exhaust, it was observed that long-term
inhalation of diesel exhaust produced an excess risk of lung
cancer.  Collectively, the epidemiological studies show a
positive, though limited,  association between diesel exhaust
exposure and lung cancer.

     Recently published, or soon to be completed studies have
concentrated on the hypothesis that the carbon core of diesel

                              ES-33

-------
                                                        EPA-420-R-93-005
                                                            April 1993
particulate matter is the causative agent in the genesis of lung
cancer.  By exposing rats to carbon black and diesel soot and
comparing the results to diesel exhaust itself, the tumor
response to diesel exhaust and carbon black is qualitatively
similar.  Also,  as a result of extensive studies, the direct-
acting mutagenic activity of both particle and gaseous fractions
of diesel exhaust has been shown.  Based on the above
information, EPA has classified diesel exhaust as a Group Bl,
probable human carcinogen.  IARC concurs that diesel exhaust is
probably carcinogenic to humans.  EPA calculated a cancer unit
risk factor for diesel exhaust based only on exposure to the
carbon core of the particle from three rat inhalation studies.
The unit risk (though still draft and subject to change) of
l.VxlO"5  (ug/m3)"1 was determined from a geometric mean of the unit
risks from these three studies.

     An attempt was made by EPA to develop a unit risk estimate
for lung cancer based on human epidemiological data.  Using these
data, EPA carried out more than 50 analyses of the relationship
between diesel exhaust exposure and tumor incidence.  None of
these analyses demonstrated a pattern that was consistent with an
association between diesel exhaust exposure and lung cancer.  The
inability to obtain an adequate dose response was attributed to
the limitations regarding exposure estimates for the various job
categories, coupled with the small increases in lung cancer
mortality.  Consequently, it was concluded that the data are
inadequate for quantitative risk assessment, based on human
epidemiological data.

     An understanding of the pharmacokinetics associated with
pulmonary deposition of diesel exhaust particles and their
adsorbed organics is critical in understanding the carcinogenic
potential of diesel engine emissions.  The pulmonary clearance of
diesel exhaust particles has multiple phases and involves several
processes including a relatively rapid transport system and slow
macrophage-mediated processes.  The observed dose-dependent
increase in the particle burden of the lungs is due, in part, to
an overloading of alveolar macrophage function.  The resulting
increase in particle retention has been shown to increase the
bioavailability of particle adsorbed mutagenic and carcinogenic
components such as benzo[a]pyrene and 1-nitropyrene.
Experimental data also indicate the ability of the alveolar
macrophage to metabolize and solubilize the particle-adsorbed
components.  Although macromolecular binding of diesel exhaust
particle-derived POM and the formation of DNA adducts following
exposure to diesel exhaust have been reported, a quantitative
relationship between these and increased carcinogenicity is not
available.

     Alternate views and/or risk assessments based on rat data
generally concur with EPA's unit risk estimate, but differ with
respect to the mathematical models and assumptions used to derive

                              ES-34

-------
                                                        EPA-420-R-93-005
                                                            April 1993
the risk estimate.  The lower bound of other risk estimates is
approximately 1.5 times lower than the EPA draft unit risk,
whereas, the upper bound is approximately 5 times higher than the
EPA unit risk.  By using the comparative potency method, all the
risk estimates determined  (except one) fall in the range
presented by the rat data.

       The 1990 base control scenario estimates the total annual
cancer deaths to be 109 (92 urban, 17 rural).   When comparing the
annual cancer deaths for the base control scenarios relative to
1990,  there is a 39% reduction in 1995, a 64% reduction in 2000,
and a 75% reduction in 2010.  The reduction in the emission
factor is considerably higher, particularly in later years.  In
this case, the projected increase in both population and vehicle
miles traveled (VMT) from 2000 to 2010 does not completely offset
the gains in emissions achieved through fuel and engine
modifications.

     A number of adverse noncancer health effects have also been
associated with exposure to acute, subchronic, and chronic diesel
exhaust at levels found in the ambient air.  Most of the effects
observed through acute and subchronic exposure are respiratory
tract irritation and diminished resistance to infection.
Increased cough and phlegm and slight impairments in lung
function have also been documented.  Animal data indicate that
chronic respiratory diseases can result from long-term  (chronic)
exposure to diesel exhaust.  It appears that normal, healthy
adults are not at high risk to serious noncancer effects of
diesel exhaust at levels found in the ambient air.  The data base
is inadequate to form conclusions about sensitive subpopulations.

     The reference concentration for chronic inhalation exposure
(RfC)  for diesel particulate matter has only recently been
established.  This RfC was determined to be S.OxlO"3 mg/m3.   As
previously mentioned,  an RfC is an estimate of the continuous
exposure to the human population that is likely to be without
deleterious effects during a lifetime.  As such, it is useful in
evaluating non-cancer effects.  The RfC for diesel particulate
matter was estimated based on studies with rats exposed to
particulate matter from light-duty and heavy-duty diesel
vehicles, with an NOAEL of 0.46 mg/m3 (0.26 ppm).   Details  on the
derivation of this RfC can be found in Chapter 9.

     Recent epidemiological studies seem to indicate that PM10
(particulate matter less than 10 microns in diameter)  might
influence daily mortality rates at concentrations lower than the
ranges encountered in the earlier studies.  In particular,
several studies that examined PM10 pollution found that  the
relative risk of daily mortality increases in a generally linear
fashion with increasing concentrations of PM10.  In  some cities,
the association was seen between PM10 and mortality  even when
particle levels never violate the current standard.  These recent

                              ES-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993
studies emphasize the lack of an apparent threshold, and indicate
that PM10 may be influencing mortality even at levels well below
the current standard of 150 ug/m3.

Gasoline Particulate Matter

     Gasoline exhaust particulate matter consists of a solid core
probably composed mainly of carbon, a soluble organic fraction,
sulfates, and trace elements.  The remaining chemical and
physical properties of gasoline particulate matter are very
similar to those of diesel particulate matter.  Gasoline
particulate matter is formed as a result of incomplete combustion
of gasoline.  Lubricating oil and other fuel hydrocarbons may
also contribute.  The sulfate particles are mostly emitted from
catalyst equipped vehicles using unleaded gasoline.  At present,
there are no motor vehicle standards being implemented for
gasoline particulate matter,  though new standards that take
effect in 1994 will limit particulate matter to 0.08 g/mile for
all light-duty engines.

     Gasoline particulate matter has not been explicitly modeled
to determine its atmospheric transformation and residence times.
Residence time calculation for gasoline particulate matter would
be expected to be similar to the non-reactive particulate-phase
POM that was described under diesel particulate matter.

     Simulations for the summer Baltimore-Washington area episode
resulted in slight decreases in POM with the use of federal
reformulated gasoline.  California reformulated gasoline resulted
in larger POM decreases than federal reformulated gasoline,
because of reductions in the T90 distillation point of the fuel.
Motor vehicle-related POM concentrations with federal
reformulated gasoline use decreased more in winter than in
summer.  Simulations for the summer Houston episode predicted
larger decreases than in the Baltimore-Washington area with the
use of reformulated gasoline.

     Because gasoline particulate matter is emitted at such low
levels, it is difficult to measure accurately.  The available
emissions data are limited and scattered.  Furthermore, all the
available data,  with the exception of one study, apply to 1986
and prior model year vehicles.  Since this study is meant to
provide a prospective look at emissions, data from the only study
which includes post-1986 model year vehicles was used solely.
Data from the other studies were used as support.  Data from this
study indicate that gasoline particulate matter is roughly 1.1%
of exhaust hydrocarbons.  This percentage was used as input to
MOBTOX and applied to all gasoline vehicle categories.

     At this time, there exists no official EPA document
detailing the carcinogenicity evidence relating to gasoline
particulate matter.  Much of the information is found in several

                              ES-36

-------
                                                        EPA-420-R-93-005
                                                            April 1993
sources, some relating to particles in general and others
focusing on the organic compounds associated with gasoline
particulate matter.

     The information on the actual carcinogenicity of gasoline
particulate matter is based mainly on in vitro and in vivo
bioassays.  This information is based on gasoline particulate
matter collected from two vehicles, one using leaded fuel and the
other using unleaded fuel.  The organic material was extracted
from the particles and used in the bioassays.  In the four in
vitro bioassays conducted to determine DNA damage (recombination,
chromatid exchanges,  unscheduled DNA repair, and sister chromatid
exchanges),  the gasoline particulate organics did produce DNA
strand breaks and sister chromatid exchanges.  There was no
evidence to support chromosomal aberrations in any of the related
studies.

     In the in vivo bioassays, the organics extracted from the
gasoline particles were able to transform embryonic cells into
malignant cells.  The most critical of the in vivo bioassays,
skin tumor initiation in mice, produced both benign and malignant
tumors.  This assay is critical because of the fact that it is
used to determine a unit risk for gasoline particulate matter
using the comparative potency method.

     At the present time, there is only a unit risk based on the
comparative potency method (no human data) and an EPA
classification does not exist.  The comparative potency method
uses epidemiological data from coke oven emissions,  roofing tar
emissions, and cigarette smoke and develops a correlation with
the gasoline particulate organics based on the relative potencies
in the mouse skin tumor initiation assay.  This process then
determines the unit risk.  For the automobile with a catalyst
using unleaded fuel,  the unit risks are 1.2xlO"4(ug organic
matter/m3)"1  and 5.1xlO"5(ug particulate matter/m3)-1.   For the
automobile without a catalyst using leaded fuel, the unit risk is
1.6xlO"5(ug particulate matter/m3)"1.  IARC has no potency for
gasoline engine
exhaust but has classified gasoline engine exhaust as a Group 2B
carcinogen,  i.e., possibly carcinogenic to humans.

     Although gasoline engine emission particulate matter is
similar to diesel exhaust in terms of chemical and most physical
properties,  the cancer unit risk estimate for gasoline engine
exhaust is based on the comparative potency method rather than
particles, for a number of reasons.  The comparative potency
method is believed, at present, to be the most logical approach
for estimating cancer risk from gasoline engine exhaust because,
first, the EPA's particle based unit risk estimate is not an
official estimate and is subject to change.  Also, while the
composition of gasoline exhaust particulate matter may be similar
to that of diesel exhaust, the particles are considerably

                              ES-37

-------
                                                        EPA-420-R-93-005
                                                            April 1993
smaller.  Cancer potency may therefore differ from diesel exhaust
because of greater particle surface area per unit volume and
because of altered deposition patterns.  Finally, since no
chronic inhalation bioassays have been carried out on gasoline
engine emissions, a particle based cancer risk estimate, using
the same methodology as for diesel would contain a considerable
degree of uncertainty.

     The cancer incidences calculated below are based on
extremely uncertain emissions data, exposure estimations, and an
unofficial EPA unit risk estimate.  The unit risk estimate, as
mentioned above, is based on the mutagenicity of the extractable
organics from the particles in the comparative potency method
using only the emissions from one unleaded gasoline vehicle.  Due
to these factors, the cancer incidences discussed below should be
considered pro forma and will not be presented in the executive
summary table which details cancer incidences/deaths due to motor
vehicles.

     For estimating annual pro forma cancer incidence, the
gasoline unit risk for catalyst vehicles based on the comparative
potency method was used.  It should be pointed out that the unit
risk is expressed in terms of whole particles, although potency
is estimated based on the organic fraction.  Nationwide annual
average exposures for the 1990, 1995, 2000, and 2010 base control
scenarios, estimated using the HAPEM-MS model, were 0.51, 0.29,
0.20, and 0.17 ug/m3,  respectively.

     The 1990 base control scenario estimates the total annual
average pro forma cancer incidence to be 93 cancer cases (79
urban, 14 rural).  When comparing pro forma cancer incidence for
the base control scenarios relative to 1990, there is a 42%
reduction in cancer incidence in 1995, a 58% reduction in 2000,
and a 63% reduction in 2010.  The reduction in per vehicle
emissions are higher, particularly in later years.  The projected
increase in both population and vehicle miles traveled  (VMT) from
2000 to 2010 appears to offset some of the gains in emissions
achieved through fuel and vehicle modifications.

     No studies exist that specifically address noncancer effects
of gasoline particulate matter.  The studies relating noncancer
effects to PM10 levels in general are applicable to both diesel
and gasoline particulate matter.

Gasoline Vapors

     Gasoline exists in two phases, liquid and vapor, with the
hydrocarbon compositions being different.   Gasoline vapors
consist mainly of short-chained and iso-alkanes  (84 to 93
percent),  alkenes (2 to 6 percent), and aromatics (1 to 5
percent).   In contrast,  liquid gasoline consists principally of
                              ES-38

-------
                                                        EPA-420-R-93-005
                                                            April 1993
66 to 69 percent paraffins  (alkanes) ,  24 to 27 percent aromatics,
and 6 to 8 percent olefins  (alkenes) .

     The major sources of exposure to gasoline vapors are from
service station operations and as a result of gasoline leakage
from underground storage tanks.  The principal exposure pathways
are from the ambient air, gasoline migration into the basements
of homes,  and the ingestion of gasoline contaminated groundwater.
The populations that receive the greatest exposure in the chain
of fuel handling are refinery workers, bulk fuel truck drivers,
service station attendants, self-service customers, and residents
of neighborhoods close to refineries,  bulk storage terminals, and
service stations.

     Studies in experimental animals provide sufficient evidence
that long-term inhalation exposure to wholly vaporized gasoline
induced a significant increase in renal carcinomas in the kidney
cortex of male rats and also a significant increase in liver
carcinomas in female mice.  Female rats and male mice had no
significant treatment related induction of tumors at any organ
site.  The incidence of renal carcinomas was significantly
increased only at the highest dose tested.  Epidemiological
studies in occupationally exposed workers suggest that long-term
inhalation of gasoline vapors may be associated with certain
types of cancer.  However, the epidemiologic evidence for
evaluating gasoline as a potential carcinogen is considered
inadequate.  Mutational bioassays performed in vivo in animals
and epidemiological studies provided negative or inconclusive
results on the mutagenicity of gasoline vapors.  Based on this
information, EPA has classified gasoline vapors as a Group B2,
probable human carcinogen.  EPA calculated a range of unit risk
factors of 2.1xlO"3 to 3.5xlO"3  (ppm) -1  for gasoline  vapors  based
on the results of a study indicating an increase in the incidence
of kidney tumors in male rats exposed to wholly vaporized
gasoline.

     Several studies in experimental animals have been published
since EPA conducted the cancer risk assessment for gasoline
vapors in 1985.  These studies confirm the previous findings of
an increased incidence of kidney tumors in male rats exposed by
inhalation to whole gasoline vapor.  Several studies tested only
the lighter hydrocarbons, which would be more characteristic of
the major fraction of gasoline vapor,  and found no evidence of
nephrotoxicity in rats.  Recent epidemiological studies do not
provide supportive evidence of a causal relationship between
cancer and exposure to gasoline vapors in humans.  Recent
genotoxicity assays generally do not support the concept of the
mutagenicity of gasoline vapors.

     Much, but not all, of the pharmacokinetic data that have
been generated since the publication of the 1985 EPA risk
assessment has been devoted to trying to determine the mechanism

                              ES-39

-------
                                                        EPA-420-R-93-005
                                                            April 1993
involved in the development of the chemically-induced kidney
tumors observed in the male rat.  A recent EPA report, Alpha2u-
globulin:  Association with Chemically Induced Renal Toxicity and
Neoplasia in the Male Rat, provided Agency-wide guidelines for
evaluating renal tumors in the male rat.  When evaluating a
possible nephrotoxic chemical, if the nephrotoxicity involves the
accumulation of the protein alpha2u-globulin in the kidney, then
the tumor incidence should not be used, since this series of
events is specific to the male rat.  This EPA policy is an
important change in EPA's general approach to cancer risk
assessment and may affect the current EPA position on gasoline
vapor carcinogenicity.

     Alternate views and/or risk estimates have been published
for gasoline vapors since the EPA risk assessment in 1985.  In a
series of studies and/or evaluations, it has been found that the
lighter hydrocarbons were not nephrotoxic, the epidemiological
evidence is weak, and there was no proof of an association
between exposure to petroleum vapors and increase in kidney
cancer.  NESCAUM (Northeast States for Coordinated Air Use
Management) determined individual lifetime cancer risks
associated with exposure to unleaded gasoline, ranging from
l.lxlO"5 to 6.3xlO"3  risk/person/lifetime.

     The baseline average annual cancer incidence from exposure
to gasoline vapor was conducted by EPA in a 1987 draft regulatory
impact analysis.  The gasoline vapor risk values determined in
this document use the EPA unit risk for wholly vaporized
gasoline.  The values, presented as the average annual values for
the study period of 1988 to 2020, range from a low of 1.3 cancer
cases from exposure at bulk plants to a high of 51 cancer cases
due to the exposure of the public at service stations.

     EPA has not initiated any specific effort to re-examine the
weight-of-evidence for gasoline vapors based on the new tumor
evaluation criteria.  It may seem timely to review the data for
gasoline because of the new criteria.  However, re-examination
would not be limited to evaluating the kidney tumor position.
EPA would also consider other newly available data relevant to
the overall framework of weight-of-evidence evaluation including
epidemiological data, toxicology data on non-cancer endpoints,
mechanism of action, information for complex mixtures, and
chemical specific information on gasoline components.  It is
possible that the resulting classification could be lower,
higher, or unchanged, based on this comprehensive review.

     When considering the other views and the recent and ongoing
research it is reasonable to assume that the values mentioned
above are conservative and more highly uncertain than the risk
estimates for the other pollutants examined in this study.  Due
to this fact, these values are considered pro forma and will not
                              ES-40

-------
                                                        EPA-420-R-93-005
                                                            April 1993
be presented in the executive summary table which details cancer
incidences/deaths due to motor vehicles.

EPA's Integrated Air Cancer Project

     The Integrated Air Cancer Project  (IACP) is an EPA
interdisciplinary research program aimed at identifying the major
carcinogenic chemicals emitted into the air, the specific sources
of these chemicals and the impact on humans of exposure to
ambient concentrations of these chemicals.  The IACP research
strategy was designed to focus on products of incomplete
combustion  (PICs).   PICs include polycyclic organic matter  (POM),
primarily absorbed to respirable particles.  This POM comprises
most of the human cancer risk of PICs.

     The IACP has primarily taken the approach of measuring the
mutagenicity of ambient air samples and apportioning this
mutagenicity to sources.  The IACP has looked at apportionment in
Raleigh, North Carolina; Albuquerque, New Mexico; and Boise,
Idaho.  In Boise, the IACP has also assessed exposure from
airborne carcinogens based on ambient measurements and human
time-activity profiles, analyzed the role of atmospheric
transformation on mutagenicity, and estimated human cancer risk
using the comparative potency method.  A field study has also
been conducted in Roanoke, Virginia, but to date, little analysis
has been done.

     Mutagenicity studies focused on extractable organic material
(EOM) obtained from samples.  EOM is basically the amount of
particulate organic material that can be extracted from ambient
air samples collected on filters using methylene chloride.  Some
mutagenicity studies were also done on semivolatile organic
compounds (SVOCs),  extracted from ambient air samples using an
absorbent known as XAD-2.  In addition, volatile organic
compounds (VOCs)  were collected in canisters, and in the Boise
study, mutagenicity was measured before and after irradiation to
determine the effects of atmospheric transformation.

     For EOM, the IACP approach involves collection of ambient
air samples on filters and extraction of organic material.  Then,
detailed chemical characterization is done using gas
chromatography and other techniques.  Next, mutagenicity is
determined using the Salmonella mutagenicity assay, and
apportioned using the receptor model approach, involving the use
of chemical tracers to identify
sources.  The procedure for measuring mutagenicity in SVOCs and
VOCs varies somewhat, due to the different collecting techniques.

     Human exposure estimates from the Boise study indicate that
mobile sources account for about 27% of the annual EOM exposure.
Furthermore, the mutagenic potency of EOM from mobile sources was
roughly three times higher than for woodsmoke, and the lifetime

                              ES-41

-------
                                                        EPA-420-R-93-005
                                                            April 1993
unit risk for mobile sources, based on the comparative potency
method, was roughly two and a half times higher than for
woodsmoke. Thus, mobile sources account for 56% of the
mutagenicity of EOM in Boise, as well as 20% of the mutagenicity
in Raleigh and 36% in Albuquerque.  In larger cities, where
mobile sources would be expected to contribute a greater
proportion of the ambient EOM, this contribution to mutagenicity
would be even higher.  Finally, atmospheric transformation may
greatly exacerbate the risk from mobile sources, since the
contribution of VOCs to mutagenicity of ambient samples increases
dramatically following irradiation in a smog chamber.

Toxics Aspects of Alternative Fuels

     As a result of the centrally fueled clean fuel fleet
program, the new California standards, and the Comprehensive
National Energy Policy Act of 1992, more alternatively fueled
vehicles could possibly be added to the fleet over the next two
decades.  It is likely that most of these alternatively fueled
vehicles would run on high level methanol/gasoline blends, neat
methanol  (M100), high level ethanol/gasoline blends, neat ethanol
(E100), compressed natural gas (CNG),  or liquid propane gas (LPG)
with a small number of electric vehicles produced to meet
California's zero emission vehicle (ZEV) requirement.  Thus, the
potential cancer reduction benefits resulting from the combustion
of these alternative fuels should be addressed.  Although engine
technology for these fuels is still being developed, potential
cancer reduction benefits can be projected with reasonable
confidence based on available data.

     Use of M100 in motor vehicles will result in substantial
reductions (i.e., 97% or greater) or elimination of benzene, 1,3-
butadiene, acetaldehyde,  gasoline refueling vapors, and
particulate matter.  However, tailpipe emissions of formaldehyde
(i.e., primary formaldehyde)  will go up by about 200% for
optimized vehicles, although no formaldehyde would be associated
with evaporative emissions.  Conversely, the use of methanol,
with its lower hydrocarbon emissions,  will result in decreased
levels of secondary formaldehyde resulting from exhaust
emissions, which is formed in the ambient air from photochemical
oxidation of hydrocarbons.  In fact,  when improvement in methanol
engine and emission control technology are considered along with
secondary formaldehyde emissions reductions, no substantial
increase in overall mass of formaldehyde emissions with use of
M100 in dedicated vehicles is projected.  However, exposure from
primary emissions of formaldehyde would likely be greater than
for secondary formaldehyde.

     For vehicles fueled with 85% methanol, significant
reductions are also expected, although these reductions are less
than that for M100 vehicles.   It should be noted that primary
                              ES-42

-------
                                                        EPA-420-R-93-005
                                                            April 1993
formaldehyde emissions are much higher than those of a dedicated
methanol vehicle.

     A large percentage of total exhaust and evaporative organic
emissions from motor vehicles running on either M100 or methanol
blends is methanol itself.  There is uncertainty as to whether
exposures to methanol vapors that may be encountered can result
in negative health effects.  EPA will assess the situation as new
information is developed.

     Like methanol, use of ethanol as a clean fuel would result
in substantial reductions in air toxics emissions.  Emissions
data for higher level ethanol blends and E100 vehicles are sparse
though.  It is likely that substantial reductions in benzene,
1,3-butadiene, refueling vapors, and particulate matter would
occur, while formaldehyde would be emitted at levels similar to
gasoline vehicles.  Acetaldehyde emissions, on the other hand,
would increase substantially.  Since the acetaldehyde cancer
potency  (2.2 x 10"6 unit risk) is much lower than the 1,3-
butadiene potency  (2.8 x 10"4 unit risk), any increase in cancer
incidence due to acetaldehyde would be greatly offset by the
large decrease in cancer incidence due to 1,3-butadiene exposure.
It should be noted, however, that acetaldehyde is an irritant and
may have some chronic and acute respiratory effects.  Thus, non-
carcinogenic health effects of increased acetaldehyde exposure
due to ethanol combustion may be a concern  (to a lesser extent,
this would be a concern with methanol combustion as well).

     CNG use would also yield substantial air toxics benefits.
Since use of CNG as a fuel requires a closed delivery system,
evaporative emissions from a dedicated CNG vehicle are assumed to
be zero.  Also, CNG contains no benzene, so refueling and running
losses of this toxic would also be zero.  Moreover, exhaust
emissions of benzene and 1,3-butadiene are very low.
Formaldehyde and acetaldehyde exhaust emissions are roughly the
same as for gasoline.

     LPG is another possible alternative fuel for motor vehicles.
LPG would be expected to have very little evaporative emissions.
LPG has very low 1,3-butadiene and benzene emissions, but
aldehyde emissions increase substantially,  as with alcohol fuels.
However, these higher aldehyde emissions would likely be reduced
with a catalyst specifically designed for an LPG vehicle.

Nonroad Mobile Sources

     The terms "nonroad engines" and "nonroad vehicles" cover a
diverse collection of equipment ranging from small equipment like
lawn mowers and chain saws, to recreational equipment,  farm
equipment,  and construction machinery.  Nonroad engines are not
presently regulated for emissions, and very few nonroad engines
currently use emission control technology.   Because of the

                              ES-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993
diversity of nonroad equipment, characterization of the emissions
from nonroad engines is a complex task.  As a group, nonroad
engines represent the last uncontrolled mobile source.  The
limited availability of toxic emission data for nonroad sources
makes it difficult to quantify precisely the contribution to
ambient air toxic levels from nonroad sources.  Many toxics such
as benzene, 1,3-butadiene, aldehydes, and gasoline vapors are
included in the broad category of pollutants referred to as VOCs.
Measures to control VOC emissions should reduce emissions of
these air toxics.  However, the magnitude of reduction will
depend on whether the control technology reduces the individual
toxics in the same proportion that total VOCs are reduced.  Since
nonroad vehicles have significant VOC impacts, they are expected
to have significant toxics impacts as well.  While Section 202(1)
of the Act addresses toxic air pollutants associated with motor
vehicles and motor vehicle fuels, EPA included nonroad engines
and vehicle in this study for purpose of completeness.
                              ES-44

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Approximately 30% of mobile source benzene emissions, or 25%
of total benzene emissions, is attributable to nonroad sources.
An estimated 13% of total formaldehyde is attributable to nonroad
sources, and an estimated 5% of total particulate matter is from
nonroad sources.  Approximately 41% of mobile source 1,3-
butadiene emissions, or about 39% of total 1,3-butadiene
emissions, is attributable to nonroad sources.  Neither this
study nor EPA's 1991 Nonroad Engine and Vehicle Emission Study
provides an estimate of the nonroad contribution to total
acetaldehyde emissions.

Initial Cost Considerations

     EPA has not done an independent evaluation of cost
considerations associated with controlling toxic emissions.
Instead, this study summarizes available cost information for
various regulatory programs which may result in reductions of
motor vehicle-related air toxics.  Cost information will be
addressed more fully in any subsequent regulatory activity.

     The estimate for the dollar cost/ton of volatile organic
compounds (VOC) reduction as it relates to the Tier 1 Standards
ranges from $3700 to $6018/ton.  For the reformulated fuel
program, the estimated nationwide summertime cost per ton of VOC
reduced ranges from $1500 to $3700.  The estimated costs for I/M
programs, based on the cost of VOC reduction per ton accounting
for NOX and  CO benefits,  can range  from $461  to  $4518.  EPA has
not done a cost-effectiveness analysis of the California LEV
Program and has not presented information on the cost per ton of
VOC or toxics reductions.  The report, however,  provides
information for the readers' benefit that was presented to EPA by
various parties as part of California's request for a waiver of
federal preemption, pursuant to Section 209(b) of the Clean Air
Act,  for the California low-emission vehicle standards and
vehicle test procedures.

     EPA's recent diesel particulate matter control regulations
focus to a large extent on diesel fuel desulfurization (although
the diesel particulate matter bus program called for in the 1990
Clean Air Act Amendments is also an important program).  The
diesel fuel sulfur regulation was developed to reduce the amount
of diesel particulate matter emitted by heavy-duty diesel
engines.  The costs are expressed as cost per ton of particles
reduced and were estimated using a calendar-year approach
discounted over a 33-year period (1994-2025).   The estimated cost
assuming no engine wear credits is $2826 to $6773/ton.

     The reduction in vehicle emissions basically takes two
forms, exhaust and evaporative, and the regulatory programs
discussed above address either one or both of these emissions.
The four toxic pollutants addressed most often,  benzene,  1,3-
butadiene, formaldehyde, and acetaldehyde, are all produced in

                              ES-45

-------
                                                        EPA-420-R-93-005
                                                            April 1993
the combustion process and emitted to the environment via the
tailpipe.  This is also true for diesel particulate matter.  Only
benzene contributes to the ambient level through evaporative
emissions due to its presence in gasoline.  Thus, those
regulatory programs that are most effective in reducing exhaust
emissions will be the most successful in reducing the greatest
number and mass of air toxics.  This is generally true assuming
that gasoline is used, but the emissions do change as the fuels
are modified.  With many of the new fuels there will be an
immediate effect on many toxic emissions  (some reduced, some
increased) since these programs affect all vehicles
simultaneously.  The exhaust emission standards will only affect
vehicles from a particular model year onward and total effects
will not be seen until there is a complete fleet turnover.

Motor Vehicle Toxics in Section 112(b) of the CAA and Metallic
Pollutants

     The list of 189 compounds in Section 112(b) of the Clean Air
Act (as amended in 1990) were reviewed to identify those
compounds (29 in all) that are either known or, based on their
structure, have the potential to be emitted from motor vehicles.
MTBE (methyl-t-butyl ether) is one of these compounds; there are
a large number of programs underway to obtain health data on
MTBE.   Another compound in this list that may be emitted from
mobile sources is 2,3,7,8-tetrachlorodibenzo-p-dioxin.  The six
metals chosen are all potential fuel additives.  Various health-
based criteria (e.g., threshold limit value  [TLV], reference dose
[RfD],  reference concentration [RfC])  have been developed for
many of these compounds.  RfCs or RfDs, as determined by EPA, do
not exist for fifteen of these compounds and three of the metals.
This is based on the fact that EPA considers the health
information inadequate or insufficient to develop the RfC or RfD
that is needed.  The Occupational Safety and Health Association
(OSHA)  and the American Conference of Governmental Industrial
Hygienists (ACGIH) have
established threshold limit values (TLV),  and/or short-term
exposure limits (STEL) for many of the compounds where EPA has
yet to determine or verify a value.

HEI Air Toxics Workshop

     In December of 1992, the Health Effects Institute conducted
a Mobile Air Toxics Workshop to identify priorities for research
that would reduce uncertainties in risk assessments for five
compounds.  These compounds are benzene, aldehydes, 1,3-
butadiene, methanol,  and POM.  Also,  six cross-cutting areas were
identified from the various individual compound sessions.  These
areas are dosimetry,  high-to-low dose extrapolation,
epidemiology, exposure assessment, molecular biological
approaches,  and neurotoxic, reproductive,  and developmental
                              ES-46

-------
                                                        EPA-420-R-93-005
                                                            April 1993
effects.  The final report should be available in the spring of
1993.

Limitations

     This section summarizes the major limitations of analyses
done in this study.  These limitations need to be considered when
reviewing the results of this study.

     Point estimates of risk are presented due to the difficulty
in reporting a range that would accurately bound the estimates.
The true risk could be as low as zero or fall above the point
estimates given in Table ES-1.  Thus, the cancer risk estimates
are not meant to be representative of actual risk.  Instead, they
are meant to be used in a relative sense to compare risks among
pollutants and  scenarios, and to assess trends.  However, the
degree of uncertainty in potency, emission and exposure estimates
is not the same for each pollutant.  A formal uncertainty
analysis would be needed to quantify the certainty of risk
associated with exposure to each pollutant.

     For all pollutants except benzene, the cancer risk estimates
are based on upper bound estimates of unit risk, determined using
animal data.  Uncertainties exist with regard to animal-to-human
and exposure-to-dose extrapolations.  Also, different
interpretations of the same health data and/or use of different
models often result in wide ranges in unit risk factors.  There
appears to be a need for more pharmacokinetic data.  Recent
pharmacokinetic research for benzene, formaldehyde, and
1,3-butadiene has been conducted and summarized in this study;
however, these data are not reflected in the risk estimates.  EPA
is currently reevaluating the health data for formaldehyde,
1,3-butadiene, and benzene.  An EPA risk assessment for diesel
particulate matter is also in progress.

     While many of the uncertainties associated with this study
are likely to result in overestimates of risk, a number of
uncertainties could result in underestimates.  The risk
assessments in this study are limited to certain components of
the mixture of chemicals in the atmosphere to which individuals
are exposed.  Risks from mixtures of chemicals in motor vehicle
emissions and mixtures resulting from the combination of
emissions from motor vehicles with emissions from other sources
or atmospheric transformation products are largely
uncharacterized.  In addition, the role of atmospheric
transformation in affecting the mutagenicity and carcinogenicity
of motor vehicle emissions is uncertain.  Atmospheric
transformation products could be important (e.g., peroxyacetyl
nitrate, or PAN, acrolein, and secondary formaldehyde),
especially since available smog chamber data suggest that
atmospheric transformation creates significantly increased
mutagenic activity.

                              ES-47

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     The discussion of non-carcinogenic effects is less
quantitative than the discussion of carcinogenic effects due to
the lack of available health data.  No attempt has been made to
synthesize and analyze the data encompassed.  Also, no attempt
was made to accord more importance to one type of noncancer
effect over another.  The objective was to research all existing
data,  describe the noncancer effects observed, and refrain from
any subjective analysis of the data.  Noncancer effects
associated with exposures to the pollutants discussed in this
study will be important to assess.

     Toxic emissions data are limited, particularly for
oxygenated fuels.  Furthermore, most data are only available for
low mileage and/or properly maintained vehicles.   In order to
estimate likely real world emissions, the available emissions
data for all the toxics except diesel particulate matter were
expressed as a fraction of total organic gases and used in a
special version of EPA's MOBILE4.1 model, called MOBTOX, to
calculate toxic emission factors.  The resulting toxic emission
estimates are thus derived rather than taken directly from
available data.  In addition, many limitations are inherent in
MOBTOX and the MOBILE4.1 model on which it is based.

     With a prospective study like this, many uncertainties are
involved with making projections.  For example, the catalyst and
fuel technology mixes in the future are only projections.  Also,
the composition of reformulated and winter oxygenated fuels and
the effect of these fuels on emissions are estimated.  The study
assumed MTBE fuel use in areas participating in the reformulated
gasoline program and oxygenated gasoline CO program, but similar
toxics benefits are expected with ethanol use.  Also, this study
is not intended to provide a comparison of different reformulated
gasoline blends.

     It should be emphasized that the expanded control scenarios
included in this study are not intended to be predictive, but are
instead intended to encompass a wide range of possibilities.
Assumptions included in the scenarios, such as types of I/M
programs, percent hydrocarbon reductions associated with
oxygenated fuel use, properties of reformulated fuels, and
estimates of fuel use under different scenarios were made using
the best available assumptions at the time the analyses were
done.   The effects of these assumptions are likely to be
significant.  Since results are presented as national annual
averages, changes in cancer incidences or deaths presented for
the expanded control scenarios do not necessarily represent
changes that would occur in specific areas where the strategies
are implemented, such as the Northeast.  Area specific analyses
would be valuable, but are beyond the scope of this study.  In
addition, the expanded control scenarios did not assess all
viable national strategies for controlling air toxics from motor
                              ES-48

-------
                                                        EPA-420-R-93-005
                                                            April 1993
vehicles.  It would be useful to evaluate the benefits of
transportational control measures, for example.

     Estimation of exposure is somewhat uncertain.  The model
used in this study for estimating annual average exposure is
based on carbon monoxide (CO) as a surrogate for motor vehicle
emissions.  This approach is particularly uncertain for the more
reactive toxics such as 1,3-butadiene.  Another limitation of the
exposure estimation is that the model uses CO NAAQS fixed site
monitoring data; however, the purpose of siting fixed site
monitoring stations is not to adequately measure ambient levels
of CO but to locate exceedances of the CO standard.  As pointed
out by several commentors,  data from fixed site monitor locations
are not likely to be adequate measures of ambient outdoor CO
concentration in the community as a whole.  As a result, the
monitor values were adjusted based on personal monitoring data
obtained from one city (Denver) over a four month period during
the winter of 1982-1983.   There is uncertainty as to whether the
resulting estimates are applicable to other areas and other
seasons.  The same general comment also applies to the activity
pattern data, which were collected in a single city (Cincinnati).
Also, the fixed site monitoring data were not adjusted to account
for non-motor vehicle sources of CO, since motor vehicles are
thought to be the predominant source of CO in urban areas.  This
assumption will serve to overestimate motor vehicle exposure.  On
the other hand, the cohort classification scheme in the model was
not intended to account for groups of people who are both highly
exposed and few in number (e.g. toll booth attendants).   This may
underestimate the highest exposure actually experienced by the
residents of the associated study area.  Finally, CO data from
only two rural areas were used to extrapolate to all rural areas
in the U.S.  There is uncertainty regarding the
representativeness of these two areas.

     In all cases, the HAPEM-MS derived exposures were compared
to ambient monitoring data,  and adjustments made to the modeled
exposures to better align them with the ambient data.   However,
there is also uncertainty associated with the ambient databases.
The sites chosen may not be representative of nationwide
exposure.  Also, for 1,3-butadiene in particular, there was a
wide range of ambient values, spanning over a factor of four.

     EPA's Total Exposure Assessment Methodology  (TEAM)  study
identified the major sources of exposure to benzene for much of
the U.S. population as well as the contributions of these sources
to personal exposure.  The most important source of benzene
exposure is active smoking of tobacco versus vehicle exposure.
Benzene is the only motor vehicle toxic for which such integrated
exposure information is available.  Some rough estimates have
been made on formaldehyde exposure suggesting most formaldehyde
exposure occurs indoors due to a large extent from the release of
                              ES-49

-------
                                                         EPA-420-R-93-005
                                                             April 1993
formaldehyde from consumer products  (e.g.,  particle board,
carpeting, etc.).

     Clearly, many limitations  are  inherent in the analyses used
in this study to assess  the  health  risk from motor vehicle air
toxics.  The EPA welcomes comments  on how to reduce these
limitations.  Moreover,  the  EPA recognizes a need to explicitly
address uncertainties.   Future  research is necessary before
critical areas
of uncertainty can be explicitly addressed.  EPA will consider the
comments received on this study to assist in prioritizing future
research planning.

Summary of Comments on Public Review Draft of Motor Vehicle-
Related Air Toxics Study

     Appendix I contains a summary of comments provided on the
public review draft of the Motor Vehicle-Related Air Toxics Study.
Many of these comments have been incorporated into the final
version of the study.  The remaining comments will be considered
by EPA during the subsequent regulatory decision making process.
Commentors on the public review draft were:  the American
Automobile Manufacturers Association  (in conjunction with the
American Petroleum Institute, the Engine Manufacturers Association
and the Association of International Automobile Manufacturers),
the American Petroleum Institute, Arco Chemical Company,  the
California Air Resources Board,  the California Environmental
Protection Agency,  the Chemical Manufacturers Association, Ford
Motor Company,  General Motors Corporation, the Health Effects
Institute, Konheim and Ketcham,  the Northeast States for
Coordinated Air Use Management,  and Zephyr Consulting.

     A number of commentors stated that the study needed to deal
with uncertainties more explicitly.   Several commentors also
pointed out the need to update EPA risk assessments for
formaldehyde, acetaldehyde,  and 1,3-butadiene.  In addition,
several commentors stated that EPA should treat diesel particulate
matter carcinogenesis as a threshold phenomenon.  A number of
comments pertained to assumptions in the HAPEM-MS exposure model.
One major comment on HAPEM-MS was that fixed site monitors are not
randomly chosen,  but placed in locations where high CO levels are
expected.   Thus,  an adjustment factor should be applied to CO
monitor readings to make them more representative of actual
exposure levels.   Another major comment was that the effect of
uncertainty in exposure predictions introduced through differences
between the diurnal profiles of reactive air toxics and CO should
be characterized.  Commentors also pointed out that EPA did not
adequately account for the nonroad contribution to mobile source
toxic emissions,  particularly for benzene and 1,3-butadiene.
Finally, two commentors expressed concern that EPA did not
adequately address the issue of motor vehicles  (especially
diesels) as a potential source of 2,3,7,8-tetrachlorodibenzo-p-
dioxin emissions.

                              ES-50

-------
                                                         EPA-420-R-93-005
                                                              April 1993
     As noted earlier, this  study  attempts  to summarize what is
currently known about motor  vehicle-related air toxics and to
present all significant scientific  opinion  on each issue.   This
study provides an important  foundation  for  any future regulatory
decision making in this area,  including decisions under Section
202 (1)(2) of the Act.  While  this  study does not resolve  the
various issues discussed herein and in  the  public comments,  EPA
will continue to explore and address these  in the context  of such
future regulatory decision making.
                               ES-51

-------
                                                        EPA-420-R-93-005
                                                            April 1993

1.0  INTRODUCTION

1.1  Background

     The U.S. Environmental Protection Agency  (EPA) initially
conducted a broad "scoping" study, with the goal of gaining a
better understanding of the size and causes of the health
problems caused by outdoor exposure to air toxics  (Haemisegger et
al.,  1985).   This study is widely referred to as the Six-Month
Study since it was meant to be conducted in a six month time
period.  The Six-Month Study contains quantitative estimates of
the cancer risks posed by selected air pollutants and their
sources.  The estimates of upper bound cancer incidence ranged
from 1300 to 1700 cases annually nationwide for all pollutants
combined.  The results further indicate that mobile sources may
be  responsible for a large portion (i.e., up to 60 percent) of
the aggregate cancer incidence.

     Based on the results of the Six-Month Study, EPA's Office of
Mobile Sources conducted a study that focused on cancer risks
posed by air toxics emissions from motor vehicles  (Carey, 1987;
Carey and Somers, 1988; Adler and Carey, 1989).  The nationwide
aggregate upper bound risk in 1986 was estimated to range from
586 to 1650 cancer incidences and dropped roughly 30 percent by
1995.  Reasons for the projected decrease in risk in 1995
include: 1)  the more stringent diesel particulate standards for
both light- and heavy-duty vehicles,  and 2) the increasing use of
3-way catalyst-equipped vehicles coupled with the phase out of
non-catalyst-equipped vehicles.  The aggregate risk in 2005 was
similar to that in 1995.  Even though emissions per vehicle mile
were predicted to decrease in 2005 relative to 1995, this
appeared to be offset by increases in vehicle miles travelled and
population from 1995 to 2005.

     EPA's Office of Air Quality Planning and Standards sponsored
a study to define the multi-source, multi-pollutant nature of the
urban air toxics problem (i.e., cancer risk) in five different
areas of the U.S., to determine what reduction is likely to occur
as  a result of ongoing regulatory activities, and to investigate
what further reductions might be possible with additional
controls.  The study is commonly referred to as the 5 City Study.
The 5 City Study was conducted in two phases, the base year
analysis for 1980 (EPA, 1989) and the projection analysis for
1995 (Pechan, 1990).  Motor vehicles were found to be responsible
for 53 percent of the average 5 city aggregate cancer incidence
in  1980 and 31 to 54 percent in 1995, depending on the control
scenario.

     EPA's Office of Air Quality Planning and Standards also
sponsored an analysis of cancer risks in the U.S. from outdoor
exposures to air toxic pollutants  (EPA, 1990).  The purpose of
this study was to update the 1985 Six-Month Study.  Based on the
pollutants and source categories examined, total upper bound
excess cancer cases were estimated to be between 1,700 and 2,700
                               1-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993

per year nationwide.  In this study, motor vehicles accounted for
almost 60 percent of total cancer incidence.

     Collectively, the results of these studies indicate that
motor vehicles could be a significant contributor to excess
cancer incidence from outdoor exposure to air toxic emissions.

1.2  Congressional Mandate

     Section 202(1)(1) of the Clean Air Act  (CAA) as amended in
1990 directs EPA to complete a study of the need for, and
feasibility of, controlling emissions of toxic air pollutants
which are unregulated under the Act and associated with motor
vehicles and motor vehicle fuels.  The study shall also address
the means and measures for such controls.  The study shall focus
on those categories of emissions that pose the greatest risk to
human health or about which significant uncertainties remain,
including emissions of benzene, formaldehyde, and 1,3-butadiene.
The proposed study shall be available for public review and
comment and shall include a summary of all comments.   The study
was due May 15, 1992.

     Pursuant to Section 202(1)(2), by May 15, 1995 EPA shall,
based on the study, promulgate (and from time to time revise)
regulations containing reasonable requirements to control
hazardous air pollutants from motor vehicles and motor vehicle
fuels.  The regulations shall contain standards for such fuels or
vehicles, or both, which EPA determines reflect the greatest
degree of emissions reduction achievable through the application
of technology which will be available, taking into consideration
the standards established under section 202(a),  the availability
and costs of the technology, and noise, energy,  and safety
factors, and lead time.  Such regulations shall not be
inconsistent with the standards under section 202(a).  The
regulations shall, at a minimum,  apply to emissions of benzene
and formaldehyde.

     This study is issued pursuant to Section 202(1)(1).  A
Federal Register notice announcing availability of the public
review draft of this study was published on January 13, 1993  (FR
58(8):4165).   The deadline for comments on the public review
draft was March 1, 1993.

1.3  Scope of Study

     The purpose of this study is to focus on air toxics
emissions from motor vehicles and their fuels.  Specific
pollutants or pollutant categories which will be discussed
include benzene, formaldehyde, 1,3-butadiene, acetaldehyde,
diesel particulate, gasoline particulate, gasoline vapors as well
as selected metals and motor vehicle-related pollutants
identified in Section 112(b) of the Clean Air Act as amended in
1990.   The focus of the study is on carcinogenic risk.  The study
also discusses non-cancer effects for these and other pollutants.
                               1-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

The discussion of non-carcinogenic effects is less quantitative
due to the lack of sufficient health data.

     Two general, but important, overall guidance documents, the
Habicht memo on risk characterization  (EPA, 1992a) and the new
exposure guidelines  (EPA, 1992b) were used in this study.

     Cancer incidence estimates for formaldehyde, 1,3-butadiene,
acetaldehyde,  and gasoline particulate matter, and cancer death
estimates for benzene and diesel particulate matter are provided
for the following calendar years:  1990, 1995, 2000, and 2010.
The following scenarios are examined:

     1)   a base control scenario, which takes into account
          implementation of the motor vehicle-related Clean Air
          Act requirements,

     2)   a scenario involving expanded use of reformulated
          gasoline,

     3)   a scenario involving expanded adoption of California
          standards.

          The scenarios are described in more detail in Chapter
2 .

     With respect to benzene, formaldehyde, 1,3-butadiene,
acetaldehyde,  diesel particulate, gasoline particulate, and
gasoline vapors, the study discusses the chemical and physical
properties of the pollutant, formation and control technology,
emissions (including other emission sources),  atmospheric
reactivity and residence times, exposure estimation, EPA's
carcinogenicity assessment, other views of carcinogenicity
assessment,  recent and ongoing research, carcinogenic risk, and
non-carcinogenic effects from inhalation exposure.  The study
also describes the qualitative change in toxic pollutant levels
with the use of alternative clean fuels, along with a summary of
toxic emissions from nonroad mobile sources.  Finally, the study
discusses the costs of various existing regulatory control
programs and provides a qualitative discussion of the toxics
benefits of these programs.

     The study attempts to summarize what is known and all
significant scientific opinion on each issue.   It will serve as a
background and status report, to be updated during the subsequent
regulatory decision making process.  This study does not include
a decision on whether and what standards to promulgate.
                               1-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993
1.4  Participation by Other EPA Offices and the Public

     An informal EPA work group was formed to provide review and
comment on plans, inputs, and drafts of the study.  The following
EPA offices were represented on the work group:

     Office of Air and Radiation
     Office of Air Quality Planning and Standards
     Office of Mobile Sources
     Office of Policy Planning and Evaluation
     Office of Research and Development
     Office of General Counsel
     Office of Pesticides and Toxic Substances

     A complete list of the work group members is included in
Appendix A.  Comments made by work group members on both the
public review draft and a previous draft of this study have been
incorporated.

     Also, a briefing was conducted on March 25, 1991 with
representatives from the automobile and oil industries to
describe plans and obtain input on the direction of the study.  A
similar briefing was also held on August 8, 1991 with the
Environmental Risk Assessment Committee of the Motor Vehicle
Manufacturers Association (MVMA).   In addition, on April 18,
1991, letters providing the status of the study and an offer to
hold a briefing on our plans for this study were sent to various
other organizations thought to have an interest in the study.
These organizations include the following:

     Oxygenated Fuels Association
     Environmental Defense Fund
     Health Effects Institute
     STAPPA/ALAPCO
     NESCAUM
     Natural Resources Defense Council
     California Air Resources Board
     Information Resources,  Inc.
     Citizen Action

No specific requests were received for briefings or additional
information; however, the California Air Resources Board provided
extensive 1,3-butadiene emission data which are used in this
study.

     This study incorporates material and information from four
reports, three resulting from work assignments initiated
specifically to provide input for this study.  One summarizes the
available information on the health effects of benzene,
1,3-butadiene, formaldehyde, the motor vehicle toxics in Title
III of the Clean Air Act Amendments, and several metallic
compounds  (Clement, 1991).   The second report summarizes current
understanding of the atmospheric behavior of benzene,
1,3-butadiene, and formaldehyde from an air quality standpoint,


                               1-4

-------
                                                        EPA-420-R-93-005
                                                            April 1993

including atmospheric formation and destruction reactions, major
physical and chemical atmospheric removal processes, and
simulated concentrations of these toxics in an urban area
(Ligocki et al.,  1991).  These first two reports were sent in
October, 1991 to the American Petroleum Institute, Ford Motor
Company, the Engine Manufacturers Association, General Motors
Research Laboratory, the MVMA Environmental Risk Assessment
Committee,  and the other organizations listed above, requesting
comments.  Comments were received from the American Petroleum
Institute.   API's comments on the contractor reports are
reflected in this study.  A third report summarized current
understanding of the atmospheric behavior of acetaldehyde and
polycyclic organic matter, and was prepared for EPA's Office of
Policy Planning and Evaluation (Ligocki and Whitten, 1991).  The
fourth report presents a modification of the Hazardous Air
Pollution Model (HAPEM) for mobile sources, referred to as HAPEM-
MS, used to predict annual average exposures to toxic air
pollutants dispersing from mobile sources  (Johnson, et al.,
1992) .

     On March 25,  1992, EPA mailed copies of large documents on
the following three subjects to about 100 people on a public
distribution list (including the organizations mentioned
previously) requesting comments:

     Toxic emission factors and control scenarios

     Exhaust hydrocarbon emission benefits with oxygenated fuels

     The HAPEM-MS model

Comments on the toxic emission factors were received from the
California Air Resources Board and these comments were
incorporated in this draft of the report.  Also, a briefing on
this material was given to the Coordinating Research Council
Auto/Oil air toxics project group on April 30, 1992.  The major
comments received dealt with the uncertainties and inadequacies
of the EPA carcinogenic potencies.  Moreover, API presented an
analysis of the HAPEM-MS model at the June 10-11, 1992 Workshop
on Research Status on Emissions,  Models, and Exposure Assessment
at Research Triangle Park, North Carolina.  API's major
criticisms dealt with uncertainties in CO measurement, its
apportionment to sources, and the validity of assuming constant
pollutant/CO ratios.  Comments were also recently received from
the American Automobile Manufacturers Association and the Engine
Manufacturers Association on the above mentioned documents.
These comments are contained in four separate contractor reports,
Environ  (1992a,b), Ligocki (1992), and Whitten  (1992).  EPA
responded to these comments and incorporated many into the final
study.
                               1-5

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     In addition, EPA opened a Public Docket  (Air Docket A-91-19)
titled, "Availability of Information on the Mobile Source-Related
Air Toxics Study Required by Section 206 of Title II of the 1990
Clean Air Act Amendments" to include information related to this
study with the emphasis on material received from the public.

     After release of the public review draft, the American
Automobile Manufacturers Association requested a meeting with EPA
to discuss comments on the study.  This meeting was held in
Detroit, Michigan on February 10, 1993.  Representatives from
Ford Motor Company, General Motors Corporation, the Engine
Manufacturers Association, the Association of International
Automobile Manufacturers, Chrysler Corporation, the Health
Effects Institute, Environ Corporation, and Caterpillar
Corporation also attended.

     Public Comments received on the public review draft were
reviewed and incorporated as appropriate in the final version.  A
complete list of commentors and a summary of the comments are
included in Appendix I.


1.5  References for Chapter 1

Adler,  J.M. and P.M. Carey. 1989. Air Toxics Emissions and Health
Risks from Mobile Sources.  Ann Arbor, Michigan:  Environmental
Protection Agency, Office of Mobile Sources.  Prepared for the
Air and Waste Management Association.  AWMA Paper 89-34A.6.

Carey,  P.M. 1987.  Air Toxics from Motor Vehicles.  Ann Arbor,
Michigan:  Environmental Protection Agency, Office of Mobile
Sources.  Publication no. EPA-AA-TSS-PA-86-5.

Carey,  P.M. and J.H. Somers.  1988. Air Toxics Emissions from
Motor Vehicles.  Ann Arbor, Michigan:  Environmental Protection
Agency, Office of Mobile Sources.  Prepared for the Air and Waste
Management Association.   AWMA Paper 88-128.1.

Clement International Corporation. 1991.  Motor Vehicle Air
Toxics Health Information.  Prepared for Environmental Protection
Agency, Office of Mobile Sources.  September,  1991.

E.H. Pechan & Associates, Inc. 1990. Analysis of Air Toxics
Emissions, Exposures, Cancer Risks and Controllability in Five
Urban Areas, Volume II.   Prepared for Environmental Protection
Agency, Office of Air Quality Planning and Standards.  April,
1990.

Environ.  1992a.  Evaluation of Procedures Used by EPA to Assess
Potential Health Risks of Vehicle Exhaust Particulates.   May,
1992.

Environ.  1992b.  Review of U.S. EPA's Application of the HAPEM
Exposure Model to Mobile Source Pollutants.  July 15, 1992.


                               1-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Environmental Protection Agency. 1989. Analysis of Air Toxics
Emissions, Exposures, Cancer Risks and Controllability in Five
Urban Areas, Volume I, Base Year Analysis and Results.  Research
Triangle Park, North Carolina: Office of Air Quality Planning and
Standards.  Publication no. EPA-450/2-89-012a.  July, 1989.

Environmental Protection Agency. 1990. Cancer Risk from Outdoor
Exposure to Air Toxics. Research Triangle Park, North Carolina:
Office of Air Quality Planning and Standards.  Publication no.
EPA-450/l-90-004a.

Environmental Protection Agency. 1992a.  Guidance on Risk
Characterization for Risk Managers and Risk Assessors.  Memo from
F. Henry Habicht II to the Assistant and Regional Administrators.
February 26, 1992.

Environmental Protection Agency. 1992b.  EPA Guidelines for
Exposure Assessment; Notice.  FR Vol. 57, No. 104.  May 29, 1992.
pp. 22888-22938.


Haemisegger, E., A. Jones, B. Steigerwald, and V. Thomson. 1985.
The Air Toxics Problem in the United States: An Analysis of
Cancer Risks for Selected Pollutants.  Environmental Protection
Agency, Office of Air and Radiation, and Office of Policy,
Planning and Evaluation.  May, 1985.

Johnson, T.R., R.A. Paul, and J.E. Capel.  1992.  Draft Report:
Application of the Hazardous Air Pollutant Exposure Model  (HAPEM)
to Mobile Source Pollutants.  Durham, North Carolina.
International Technology Corporation.

Ligocki, M.P.  1992.  Review of EPA Memorandum on Calculation of
Toxic Emission Mass Fractions.  Systems Applications
International,  June 23, 1992.  SYSAPP-92/075.

Ligocki, M.P., G.Z. Whitten, R.R. Schulhof, M.C. Causley, and
G.M. Smylie. 1991.  Atmospheric Transformation of Air Toxics:
Benzene, 1,3-Butadiene, and Formaldehyde.  Systems Applications
International, San Rafael, California  (SYSAPP-91/106).

Ligocki, M.P., and G.Z. Whitten.  1991.  Atmospheric
Transformation of Air Toxics:  Acetaldehyde and Polycyclic
Organic Matter.  Systems Applications International, San Rafael,
California  (SYSAPP-91/113).

Whitten, G.Z.  1992.  Review of EPA Memorandum "Interim Analysis
of HC Reduction With the Use of Oxygenated Fuel Blends With 3-Way
Catalyst Vehicles".  Systems Applications International, San
Rafael, California  (SYSAPP-92/082).
                               1-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993
2.0  SCENARIOS STUDIED

     As mentioned in Chapter 1, cancer incidence resulting from
exposure to benzene, formaldehyde, 1,3-butadiene, acetaldehyde,
diesel particulate and gasoline particulate was estimated for
several possible control scenarios in the years 1990, 1995, 2000,
and 2010.  The scenarios examined in this study include a base
control scenario, which takes into account implementation of
requirements in the CAAA of 1990, a scenario involving expanded
use of reformulated gasoline, and a scenario involving expanded
adoption of California motor vehicle standards.  These scenarios
were chosen to compare the possible effects different control
programs could have, and do not necessarily represent EPA's
expectations for the scope of possible expanded implementation
for these control programs.  In addition, the scenarios are not
intended to indicate effects in specific areas where the
strategies are implemented, such as the Northeast.  Area specific
analyses would be valuable, but are beyond the scope of this
study.  Although diesel particulate emissions were examined for
1990,  1995, 2000, and 2010, individual scenarios were not studied
for this toxic, since expanded use of reformulated gasoline and
the expanded adoption of California standards would not affect
diesel particulate.

     The use of alternative clean fuels, such as 85-100%
methanol, 85-100% ethanol, and compressed natural gas, was not
considered as part of any of these scenarios, since it is likely
to comprise only a small fraction of total nationwide fuel use
under current legislation  (primarily as part of California's low
emission vehicle program and the federal centrally fueled clean
fuel fleet program).  However,  the use of alternative fuels could
yield substantial toxics benefits, and their potential role in
reducing motor vehicle-related air toxics will be discussed in
Chapter 13.

2.1 Baseline

     Base control scenarios for the years examined take into
account implementation of the motor vehicle-related CAAA
requirements, but assume no expanded adoption of CAAA programs or
California standards, and no expanded use of gasohol beyond 1990
levels.

     The 1990 base control scenario includes no new CAAA
programs, since none were in place at this time.  The 1995 base
control scenario, however, includes Phase 1 of the federal
reformulated gasoline program  (coverage limited to the nine major
metropolitan areas mandated by Section 211 (k) of the Act), Phase
1 of the California reformulated gasoline program, and the
oxygenated fuels CO program.  These programs are described in
greater detail in Section 3.1.3.  The 2000 and 2010 base control
scenarios vary from the 1995 base control scenario in that Phase

                               2-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993
2 federal and California reformulated gasoline will be in use,
rather than Phase 1.  Federal and California Phase 2 reformulated
gasolines differ from Phase 1 fuels primarily in that they have
lower RVP standards.

2.2  Additional Control Scenarios

2.2.1  Expanded Use of Reformulated Gasoline

     This scenario is considered for the years 1995, 2000, and
2010.  In this scenario, all ozone nonattainment areas opt into
the federal reformulated gasoline program.  In Section 211(k) of
the Act,  ozone nonattainment areas are given the option of
participating in the program.  In addition, all Northeast states
have expressed intent to opt into the federal reformulated
gasoline program; thus, they will be considered participants in
this program under the expanded use of reformulated gasoline
scenario.  In 1995, Phase 1 federal and California reformulated
gasoline will be in use, while in 2000 and 2010, Phase 2 federal
and California reformulated gasoline will be in use.

2.2.2  Expanded Adoption of California Motor Vehicle Standards

     This scenario is considered for the years 2000 and 2010.
California emission standards are similar to federal standards in
1995; thus, this scenario is not considered for that year.
However,  California standards become increasingly more stringent
with time, so that in 2000 and 2010, they are markedly lower than
federal standards.

     In this scenario, all Northeast states and states with ozone
nonattainment areas categorized as extreme, severe, or serious
adopt California motor vehicle emission standards.  This scenario
also assumes expanded use of reformulated gasoline, as described
in the previous section.

     California's new emission standards also involve the use of
reactivity adjustment factors which normalize the mass of non-
methane organic gas (NMOG) emissions from various fuels (such as
reformulated gasoline, methanol, ethanol, and compressed natural
gas) according to their ozone-forming potential.  Furthermore,
California certifies vehicles in several different categories
according to their ozone-forming potential, and any combination
of vehicles and fuels in these categories can be used to meet
standards.  For the sake of simplicity, it was assumed that the
standards would be met using gasoline.  More information on the
California standards can be found in Section 3.1.3.1.
                               2-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993
3.0  EMISSION FACTOR METHODOLOGY
3.1  Methodology for Benzene, Formaldehyde, 1,3-Butadiene, and
Acetaldehyde

3.1.1  Approach

     In order to obtain risk estimates, emission factors must be
calculated.  With the approach used for this report, available
vehicle emissions data are used to estimate toxic emissions as
fractions of total organic gases  (TOG).  TOG includes all
hydrocarbons as well as aldehydes, alcohols, and other oxygenated
compounds.  These fractions are then applied to an updated
version of MOBILE4.1, designated MOBTOX, developed specifically
to calculate toxic grams per mile emission factors.  This same
basic approach was used in previous EPA papers (Carey, 1987;
Carey and Somers, 1988; Adler and Carey, 1989), where emission
fractions for air toxics were applied to MOBILE4 THC output.

     MOBTOX calculates in-use g/mile toxic emission factors.
This approach was used because virtually all the available
emission data are from low mileage, well-maintained vehicles.  To
simply use the g/mile data from these studies directly would
likely result in an underestimation of true emissions.

     The approach outlined in this section will be used for
benzene, 1,3-butadiene, formaldehyde, and acetaldehyde.  In order
to estimate these emission factors, mass fractions of exhaust TOG
emissions and evaporative  emissions  (for benzene) must be
obtained for these toxics from actual data, to input into MOBTOX.
These fractions must be calculated for various motor vehicle
classes, catalyst types, fuel systems, and fuel blends.  Separate
sets of fractions resulting from implementation of different
regulations must also be calculated.  Section 3.1 describes the
methodology for obtaining mass fractions for the non-particulate
air toxics and for developing MOBTOX inputs.  It should be noted
that all mass fractions are expressed as fractions of TOG.

3.1.2  Assumptions

     A number of important assumptions were made in the approach
outlined in this section.  Several of these assumptions were:

     1)   Increase in air toxics due to vehicle deterioration
          with increased mileage is proportional to increase in
          TOG.
     2)   Toxics fractions remain constant with ambient
          temperature changes.
     3)   The fractions are adequate to use for the excess
          hydrocarbons that come from malfunction and
          tampering/misfueling.
                               3-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     These assumptions can be addressed by looking at high
mileage data, temperature data, malfunction data, and misfueling
data.  First, Carey  (1987) analyzed formaldehyde and benzene data
from the 46 car study  (Sigsby et al.,  1987), and found very
little difference in fractions of these compounds among vehicles
with high and low hydrocarbon emissions.  Also, an earlier study
(Smith and Carey, 1982) shows high mileage cars control
formaldehyde roughly to the same extent as total hydrocarbons.
Similarly, a General Motors study (Dasch and Williams, 1991)
showed no significant increase in benzene fractions with mileage.
Furthermore, the emission fractions calculated from low-mileage
vehicles in the current analysis are similar to the in-use
fractions in the General Motors study.  Thus, it is reasonable to
assume that these two compounds increase proportionally to TOG.
Finally, results from a recent Auto/Oil analysis (Auto/Oil, 1993)
indicated that fuel effects on toxic emissions were similar in
normal and high emitting vehicles.  Furthermore , the toxic
fractions were similar for normal and high emitting vehicles.
This analysis included the   toxics formaldehyde, acetaldehyde,
benzene, and 1,3-butadiene.

     Stump et al. (1989, 1990, unpublished) looked at the effects
of ambient temperature on exhaust toxics.  Stump et al. (1989,
1990),  in their low temperature study (20F to 70F range),  found
a slight increase with temperature reduction of formaldehyde
emissions, but overall, the composition of hydrocarbon emissions
did not vary appreciably with temperature.  In the high
temperature study (Stump et al.,  unpublished, 75F to 105F
range),  exhaust and evaporative emissions were analyzed.
Formaldehyde exhaust emissions increased slightly in PFI vehicles
with increased temperature,  but decreased slightly for the one
carbureted vehicle studied.   There was no appreciable change for
other aldehydes.  Moreover,  the authors state that tailpipe
emissions for benzene and 1,3-butadiene in general followed total
hydrocarbon levels.   For diurnal evaporative emissions, aromatics
fractions as a whole were measured.   It is expected that benzene
fractions would track the aromatics trend.  Aromatics fractions
went down with temperature,  for both the fuel injected vehicles
and the carbureted vehicle.   Hot soak fractions of aromatics in
fuel injected vehicles went up when going from 75 to 90F,  but
down when going from 90 to 105F.   For the carbureted vehicle,
however, aromatics went down when going from 75 to 90F,  but up
when going from 90 to 105F.   Also,  a separate analysis has been
performed (EPA, 1992a)  in which a number of MOBILE4.1 runs were
done at four different temperatures in 1990 and 2000.  The
results indicated that the ratio of hydrocarbon and carbon
monoxide from one temperature to another is relatively constant
in 1990 and 2000.  Based on the results of these studies,  a broad
generalization was made that emission fractions would not change
as a function of temperature.

     Carey  (1987) also analyzed available malfunction and
misfueling exhaust data for aldehydes and benzene.  For
aldehydes, Carey reviewed misfueling data available for a single
vehicle (Nebel, 1981) and found only a slight increase in


                               3-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

percentages of aliphatic aldehydes, which should be an indicator
of formaldehyde and acetaldehyde emissions.  In addition, an
analysis of malfunction studies  (Urban, 1980a, 1980b, 1980c,
1981; Urban and Garbe, 1979, 1980) indicated roughly similar
formaldehyde percentages with and without several malfunctions
for vehicles with no catalyst, but small decreases in
formaldehyde percentages with malfunctions in catalyst equipped
vehicles.  From this review, Carey (1987) concluded that,
overall, formaldehyde percentages were relatively stable under
malfunction and misfueling conditions.

     For benzene, Carey (1987) analyzed data from the same
malfunction studies analyzed for formaldehyde.  A 12 percent
misfire mode decreased benzene exhaust percentages appreciably,
while a rich best idle mode increased benzene exhaust percentages
appreciably.  Since these two malfunctions were offsetting, and
other malfunctions had lesser effects, no adjustments were made
to benzene fractions for malfunctioning.  No misfueling studies
were available for benzene; thus, we assumed no misfueling
effects on benzene fractions.

     No malfunction or misfueling data were available for 1,3-
butadiene; however, the CARB data used to determine 1,3-butadiene
fractions were based on in-use vehicles, tested as received, with
the same fuel as received.  Thus, there was no need to address
the effects of malfunction or misfueling on emission fractions
for 1,3-butadiene.

3.1.3  Emission Factor Requirements

3.1.3.1  Scenario Components

     Before developing exhaust and evaporative mass fractions to
use in determining emission factors,  it is necessary to consider
the various scenarios to be included in the report.  The
scenarios, which are described in Chapter 2, include:

  1)  a base control scenario, which takes into account
     implementation of the motor vehicle-related Clean Air Act
     requirements,
  2)  a scenario involving expanded use of reformulated gasoline,
     and
  3)  a scenario involving expanded adoption of California
     standards.

     The effects of the different scenarios on overall emissions
will be considered for the following years:  1990, 1995, 2000,
and 2010.  It will not be possible to simply run MOBTOX once for
each scenario/calendar year.  This is because various areas of
the country have different fuel and/or emission standard
requirements, as well as different I/M programs.  From an
examination of the Clean Air Act requirements, the scenarios to
be considered, and the California program, nine different
fuel/emission standard combinations were identified.  These
                               3-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993

fuel/emission standard combinations will be referred to as
components.

     This section focuses on the fuel specifications, emission
standards, and calendar years applicable for the nine components.
A list of components and the fuel specifications assumed for
these components is given in Table 3-1.  For federal reformulated
gasoline, fuel parameters are not certain at this time,
particularly for Phase 2 gasoline.  However, the fuel parameters
used in this report for Phase 1 and 2 meet the toxic performance
requirements required in the Clean Air Act.  Section 3.1.3.2
provides more information on the scenarios, including which
components are considered for each scenario, their relative
weighting by fuel use, and the specific areas/cities covered
under each scenario component.

1)   Baseline Gasoline Use (Federal Emission Standards) -- Covers
     areas of the country using a typical 1990+ baseline
     gasoline.  Baseline gasoline for 1990 and subsequent years
     was assumed to contain 1.53% benzene, 32% aromatics, and 0%
     oxygen, at 8.7 psi Reid Vapor Pressure (RVP).   These levels
     are given as summertime baseline gasoline specifications for
     the reformulated gasoline program in Section 211(k) of the
     CAA.  According to the national fuel survey (MVMA, 1990),
     regular unleaded gasoline in summer 1990 contained 1.46%
     benzene, 27.8% aromatics,  and 0% oxygen,  at 8.6 psi RVP.
     These specifications are similar to those given in Section
     211(k).

     The federal THC/NMHC 50,000 mile emission certification
     standards for light duty vehicles (< 3750 Ibs.)  are of
     interest for this analysis.  The THC standard is currently
     0.41 gram per mile.  The Tier 1 tailpipe standard of 0.25
     gram per mile for NMHC will be phased in beginning in 1994.
     A Tier 2 tailpipe NMHC standard of 0.125 gram per mile
     beginning in 2004 is contingent on determination of cost-
     effectiveness and feasibility by EPA.  For this analysis,  it
     is assumed that Tier 2 will not be implemented.

     For the sake of simplification, California is included in
     this component for 1990 since the current California exhaust
     NMOG 50,000 mile certification standard of  .390 grams per
     mile NMOG is similar to the current federal THC standard of
     0.41 grams per mile.  This component is considered for all
     the calendar years of interest.

2)   Baseline Fuel Use  (California Emission Standards)  -- Under
     an expanded scenario, all states with extreme,  severe or
     serious ozone nonattainment areas adopt California emission
     standards.  Also under this expanded scenario,  all Northeast
     states adopt California standards. This scenario may result
     in attainment
                               3-4

-------
                                                     EPA-420-R-93-005
                                                         April 1993
Table 3-1.  Fuel Specifications for the Various Components.

Components
Baseline Gasoline Use
Federal Standards
Baseline Gasoline Use
California Standards
Federal/Calif. Reform. Gasoline
Use
Federal Phase 1 (1995-1999)
Calif. Phase 1 (1992-1995)
Federal/Calif. Standards
Federal Reform. Gasoline Use
Phase 2 (2000+)
Federal Standards
Federal Reform. Gasoline Use
Phase 2
Calif. Standards
Winter Oxygenated Gasoline Use
Federal/Calif. Standards (1995)
Federal Standards (2000, 2010)
Winter Oxygenated Gasoline Use
Calif. Standards (2000, 2010)
California Only
Calif. Reform. Gasoline Use
Phase 2 (1996+)
Calif. Standards
Gasohol Fuel Use
Federal Standards
Fuel Specifications
Benzene
(% Vol.)
1.53
1.53
1.0
1.0
1.0
1.05
1.05
1.0
1 .4
Aromatics
(% Vol.)
32
32
25
25
25
22
22
25
28.8
Oxygen
(% Wt.)
0
0
2.0
2.0
2.0
2.7
2.7
2.0
3.5
RVP
(psi)
8.7
8.7
8.1
7.8
7.8
8.7
8.7
7.0
9.7
  areas in those states having baseline fuel use  in
  conjunction with California emission standards.  In  1995,
  this has little effect on emission factors,  since  federal
  and California light duty vehicle  (< 3750 Ibs.) exhaust
  emission standards are similar  (0.250 g/mile NMHC  under
  federal regulations; 0.231 g/mile NMOG under California
  regulations).   Thus there is no need to distinguish  between
  the two sets of standards.  In 2000 and 2010, however,
  federal and California standards are markedly different,
  with the federal standard remaining at 0.250 g/mile  NMHC
  (under the assumption that Tier 2 is not implemented), while
  the California standard is 0.073 g/mile NMOG for 2000  and
  0.062 g/mi NMOG for 2010.  For these years,  then,  fuel use
  for attainment areas using baseline fuel with California
                            3-5

-------
                                                   EPA-420-R-93-005
                                                       April 1993

standards must be treated separately from baseline fuel use
with federal emission standards.

Federal Reformulated Gasoline Program, Phase 1  (Federal
Emission Standards)  and California Reformulated Gasoline
Program, Phase 1 (California Emission Standards) -- This
component covers regions participating in Phase 1 of the
federal reformulated gasoline program, from 1995 through
1999, under federal emission standards.  It also covers
Northeast states participating in the federal reformulated
gasoline program and also opting into the program for
California emission standards in 1995, as well as California
under the California Phase 1 reformulated gasoline program
(1992 - 1995)  with California emission standards.  Due to
the timing of the Phase 1 requirements, this component is
considered only for the calendar year 1995.

Phase 1 federal reformulated gasoline must contain at least
2.0% oxygen, and must not result in a NOX increase.
Reduction of both ozone forming VOCs and air toxics must be
least 15%, relative to emission levels from 1990 model year
vehicles with a baseline gasoline.  The required 15% minimum
toxics reduction for reformulated gasoline is measured on a
mass basis for 5 specific pollutants -- benzene,
formaldehyde,  acetaldehyde, 1,3-butadiene, and POM.  The
toxics requirement is year-round while the VOC requirement
applies during the summer months.  Reformulated gasoline
fuel specifications of 2.0% oxygen, 1.0% benzene, 25%
aromatics and 8.1 psi RVP  (for ASTM Class C areas)  were
assumed for CY 1995 - 1999.   The oxygen and benzene
specifications are minimum or maximum requirements specified
in Section 211(k) of the Act.  The RVP level is an estimate
for Class C areas.

Based on EPA's proposed regulations for reformulated
gasoline, EPA assumed maximum RVP levels for the high ozone
season  (June 1 through September 15) of no more than 7.2 psi
in Class B areas (in Southern states)  and 8.1 psi in Class C
areas (in Northern states).  However,  a recent EPA proposal
seeks comment on a decision by former President Bush to
effectively grant gasohol a 1 psi RVP waiver for up to 30%
of the total reformulated gasoline market in the Northern
cities.  The increase in VOC emissions from the higher RVP
would be compensated for through a requirement that the
volatility of reformulated gasoline blendstock in these
cities be reduced by 0.3 psi to 7.8 psi.  A similar
provision would be made available for Southern cities to opt
into, except that gasohol would effectively receive a 1 psi
waiver for up to 20% of the total reformulated gasoline
market, requiring the use of 7.0 psi RVP blendstock
gasoline.  Details of this waiver are presented in a recent
proposed rule for standards for reformulated gasoline (EPA,
1993) .
                          3-6

-------
                                                   EPA-420-R-93-005
                                                       April 1993

California Phase 1 reformulated gasoline in CY 1992 - 1995
has similar specifications.  Thus for 1995, there is no need
to distinguish between federal and California Phase 1
reformulated gasoline.

There are a number of areas where California has more
stringent standards or special programs not implemented in
the rest of the country.  For instance, the CAAA establish
provisions for a California clean car pilot program,
applying to a limited number of cars starting in 1996.  The
pilot program is not considered in this report.  Also under
the CAAA, California is permitted to develop its own, more
stringent vehicle control program.

The California Air Resources Board (CARB) has adopted
regulations establishing increasingly stringent vehicle
certification standards beginning in 1994  (CARB, 1990).
Requirements for non-methane organic gas (NMOG, which is TOG
less methane)  begin at 0.250 grams per mile for light duty
vehicle  (< 3750 Ibs.)  exhaust at 50,000 miles in 1994 and
are progressively reduced to 0.062 grams per mile in 2003
(with a requirement of 0.231 grams per mile in 1995).
CARB's new standards involve the use of reactivity
adjustment factors which normalize the mass of NMOG
emissions from various fuels according to their ozone-
forming potential.  CARB certifies vehicles in several
categories based on the ozone-forming potential of their
emissions.  These categories are:  Transitional-Low Emission
Vehicles  (TLEVs),  Low Emission Vehicles  (LEVs), Ultra-Low
Emission Vehicles (ULEVs),  and Zero Emission Vehicles
(ZEVs).   Under the 1994 standards, any combination of TLEVs,
LEVs, ULEVs, ZEVs and 1993 conventional vehicles can be used
to meet fleet average requirements.  The 50,000 mile exhaust
emission certification standards for the light duty vehicle
(< 3750 Ibs.)  categories are described in Table 3-2.

Although over time California emission standards are more
stringent than federal standards, they are similar for 1995.
Since California and federal Phase 1 reformulated gasoline
specifications are also similar in 1995, all areas with
combinations of federal and California Phase 1 reformulated
gasoline and federal and California emission standards can
be considered as one component.  This includes many
Northeast states which are considering participating in the
federal reformulated gasoline program and also opting into
the program for California standards.  In these states,
vehicles will be certified on California gasoline and will
                          3-7

-------
                                                       EPA-420-R-93-005
                                                           April 1993
Table 3-2.  California Low Emission Vehicle 50,000 Mile Exhaust
   Emission Certification Standards for Light Duty Vehicles
                         (<  3750  Ibs.).
Vehicle Category3
Current
1993
TLEV
LEV
ULEV
ZEV
Grams/Mile by Pollutant
NMOG1
0.390
0.250
0.125
0.075
0.040
0.000
NOV
0.4
0.4
0.4
0.2
0.2
0.0
CO
7.0
3.4
3.4
3.4
1.7
0.0
HCHO
none
0.0152
0.015
0.015
0.008
0.000
    1NMHC for current and 1993  standards,  NMOG with reactivity
    adjustment for others.
    2Methanol-fueled vehicles only.
    3Emission levels in this table do not  include stationary
    source emissions related to fuel generation, including
    generation of electricity for ZEVs.
    have to meet California standards, but for purposes of this
    study are presumed to be running on federal reformulated
    gasoline in-use.

    Federal Reformulated Gasoline Program, Phase 2  (Federal
    Emission Standards) -- This component covers regions
    participating in Phase 2 of the federal reformulated
    gasoline program, under federal emission standards.

    Beginning in the year 2000, under Phase 2 of the
    reformulated gasoline program, ozone forming VOC and toxics
    reductions must be at least 25%, or 20% if the  25% reduction
    is judged to be unfeasible.  Once again, the toxics
    requirement is year-round while the VOC requirement applies
    during the summer months.

    It is assumed that Phase 2 federal reformulated gasoline
    will have similar benzene and oxygen requirements as Phase 1
    gasoline.  For purposes of this study an RVP of 7.8 psi is
    assumed  (for ASTM Class C areas),  slightly lower than the
    8.1 psi assumed for Phase 1.  The component is  considered
    for calendar years 2000 and 2010.
                              3-1

-------
                                                   EPA-420-R-93-005
                                                       April 1993

Federal Reformulated Gasoline Program, Phase 2  (California
Emission Standards) -- This component covers non-California
regions participating in Phase 2 of the federal reformulated
gasoline program, under California motor vehicle emission
standards.

Regions participating in Phase 2 of the federal reformulated
gasoline program under California standards cannot be
considered with Phase 2 of the California program, because
California Phase 2 gasoline has a much lower RVP
requirement.  This component is only applicable for the
scenario involving expanded adoption of California standards
for calendar years 2000 and 2010.

Oxygenated Fuels CO Program, (Federal and California
Emission Standards, 1995; Federal Emission Standards, 2000,
2010)  -- This component covers regions participating in the
seasonal oxygenated gasoline CO program, beginning November
1, 1992, while complying with federal or California motor
vehicle emission standards in 1995 scenarios.  It also
covers regions complying with federal emission standards in
scenarios for the years 2000 and 2010.  Regions with
California and federal standards are considered as one
component in 1995 because of the similar federal and
California emission standards during this year.

Section 211(m) of the Act specifies a minimum 2.7% oxygen
level for gasoline in this program.  Winter oxygenated
gasoline, used in the oxygenated fuels CO program, was
assumed to be 2.7% oxygen (15% MTBE),  22% aromatics, 1.05%
benzene and 8.7 psi RVP.  The estimate of 22% aromatics was
chosen after examining fuel specifications of 15% MTBE
blends used in various test programs.   Aromatic levels in
the 22% range were fairly consistent across these studies.
The percent reduction in aromatics from the baseline level
of 32% to 22% was then applied to the baseline benzene level
of 1.53% to obtain the estimate of 1.05% benzene.  8.7 psi
RVP was chosen arbitrarily.   It is likely winter fuel would
have a higher RVP, but changing RVP would have a minor
effect on the exhaust fractions calculated.

Some regions participating in this program will also be
participating in the federal reformulated gasoline program
or the California reformulated gasoline program.  In regions
participating in two programs,  fuel requirements for both
the winter oxygenated and Phase 1 or Phase 2 federal or
California reformulated gasoline (depending on the year)
will have to be met during winter.   The primary differences
between these fuels which may affect toxics emission
fractions are RVP and oxygen content.   Since RVP is not a
significant factor during winter months, and VOC control for
reformulated gasoline is limited to the summer months, it
was assumed for modeling purposes that winter oxygenated
gasoline would be used in all of these regions during the
winter months.  In the modeling, use of winter oxygenated


                          3-9

-------
                                                   EPA-420-R-93-005
                                                       April 1993

gasoline still meets the toxics reduction requirements of
the federal reformulated gasoline program.  This component
is considered for calendar years 1995, 2000, and 2010.

It should be noted that we assumed the oxygenated gasoline
CO program would utilize a 2.7% oxygenate MTBE blend.  Other
oxygenated blends with ethanol  (at the 2.7 oxygen level)
will also be used.  However, similar toxics benefits are
expected with the use of gasohol in reformulated areas.

Oxygenated Gasoline CO Program  (California Emission
Standards,  2000, 2010)  -- This component covers regions
participating in the oxygenated gasoline CO program, while
complying with California emission standards for the years
2000 and 2010.  These regions will be found in California
and, under an expanded scenario, in states with extreme,
severe, and serious ozone nonattainment areas adopting
California standards, and also Northeast states adopting
California standards.  Once again, some regions may also be
participating in the federal reformulated gasoline program.
These regions will have to meet fuel requirements for both
the winter oxygenated and Phase 2 federal reformulated
gasoline during the winter.  It is assumed that regions in
California participating in this program will have to meet
requirements for winter oxygenated and California Phase 2
gasoline.  As with the previous component, it was assumed
for modeling purposes that winter oxygenated gasoline would
be used in winter for all regions considered as part of this
component.   This component is considered for calendar years
2000 and 2010.

California Reformulated Gasoline, Phase 2 (California
Emission Standards) -- This component covers California
under Phase 2 California reformulated gasoline requirements
(1996+),  under California emission standards.  Phase 2
California reformulated gasoline includes maximum limits of
1.0% benzene, 25% aromatics, 2.0% oxygen and 7.0 psi for
each gallon refined  (Refiners can choose instead to average
production over 90 days, meeting lower averaged limits for
benzene and aromatics of 22 and 0.80 percent, respectively).
This component is considered for calendar years 2000 and
2010.

Ethanol Fuel Use  (Federal Emission Standards) -- This
component is based on vehicle consumption of 10% ethanol in
gasoline (or gasohol).   It is considered for all the
calendar years.

The composition of gasohol is assumed to be 1.4% benzene,
28.8% aromatics, and 9.7 psi RVP.  The composition was
estimated by assuming a 10% reduction of benzene and
aromatics,  and an increase of 1 psi in RVP from dilution of
gasoline with 10% denatured ethanol, applied to the baseline
gasoline specifications.  This composition is similar to the
composition of the 10% ethanol blends used in the Auto/ Oil


                          3-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     study (1991) which had benzene levels ranging from 1.4 to
     1.5%, aromatics ranging from 18 to 29%, and RVP ranging from
     9.0 to 9.6 psi.  The composition of the 10% ethanol blend
     used in another recent study used as a data source in this
     report (Warner-Selph and Smith, 1991) was also similar, with
     1.35% benzene, 22.8% aromatics, and an RVP of 10.15 psi.

     Although the CAA establishes provisions for a California
pilot program and a clean fuel fleet program for centrally fueled
fleets, we will not consider scenarios specifically involving
components for these programs.  As mentioned above, California is
establishing its own standards which could effectively supplant
the standards specified by the pilot program in California.
Since the centrally fueled clean fuel fleet program covers a
small number of vehicles  (30% of new fleet purchases in 26
metropolitan areas, starting in 1998, for fleets with central
refueling),  it was deemed unnecessary to include a component for
this program in this report.  The toxics benefits associated with
using 85-100% methanol, 85-100% ethanol, and compressed natural
gas as alternative fuels  (EPA, 1989a, 1990a, 1990b, and 1990c)
are qualitatively discussed in Chapter 13.

3.1.3.2  Percent of Nationwide Fuel Use by Component for Each
Scenario

     Table 3-3 consists of a matrix allocating nationwide fuel
use in 1990,  1995, 2000, and 2010 for the various components of
each scenario.  Descriptions of the three scenarios listed
earlier for each calendar year are given below.  These include
descriptions of how fuel use percentages in a given year were
determined for each component in a scenario.  Assumptions made in
determining these
fuel use percentages are also discussed.  Also included are the
specific areas/cities covered under each scenario component.

1)    1990 Base Control -- Since no new CAA programs were in
     effect in 1990, this scenario includes only two components -
     - one for baseline gasoline use, and one for gasohol fuel
     use.  An estimate of 6% gasohol fuel use for 1990 was
     obtained from data compiled by the U.S. Department of
     Transportation (1991).   These data were based on gross
     gallons of gasohol reported by wholesale distributors to
     State motor fuel tax agencies, and include highway use,
     nonhighway use, and losses.  The remainder of fuel use in
     this scenario  (94%) was assumed to be baseline gasoline use.
                               3-11

-------
Table 3-3.   Nationwide Fuel Use for  the Various  Components  Under Different  Scenari
                                                                                            EPA-420-R-93-005
                                                                                                   iril 1993
                                    Percent of Total Nationwide Fuel Use*


Components /Scenarios


Baseline Gasoline Use
Federal Standards
Baseline Gasoline Use
California Standards
Federal/Calif. Reform. Gasoline
Use
Federal Phase 1 (1995-1999)
Calif. Phase 1 (1992-1995)
Federal/Calif. Standards
Federal Reform. Gasoline Use
Phase 2 (2000+)
Federal Standards
Federal Reform. Gasoline Use
Phase 2
Calif. Standards
Winter Oxygenated Gasoline Use
Federal/Calif. Standards (1995)
Federal Standards (2000, 2010)
Winter Oxygenated Gasoline Use
Calif. Standards (2000, 2010)
California Only
Calif. Reform. Gasoline Use
Phase 2 (1996+)
Calif. Standards
Gasohol Fuel Use
Federal Standards
1990
Base
Control


94

0

0




0


0


0


0

0



6

1995
Base
Control


59

0

18




0


0


17


0

0



6

Expanded
Reform.
Gasoline
Use
27

0

50




0


0


17


0

0



6

2000, 2010
Base
Control


59

0

0




10


0


12


5

8



6

Expanded
Reform.
Gasoline
Use
27

0

0




42


0


12


5

8



6

Expanded
Adoption
Calif.
Standards
22

4

0




13


30


3


14

8



6

         *Each vertical column totals 100 percent.
                                                3-12

-------
                                                   EPA-420-R-93-005
                                                       April 1993


1995 Base Control -- This scenario includes gasoline use
under Phase 1 of the federal reformulated gasoline program
and the California program, the oxygenated gasoline CO
program, and gasohol fuel use.  Fuel use under Phase 1 of
the federal reformulated gasoline program and Phase 1 of the
California reformulated gasoline program combined was
estimated to be 18%.  The base control scenario assumes only
the 9 extreme/severe ozone nonattainment areas participate
in the federal reformulated gasoline program.  These areas
are:

     1) New York
     2) Philadelphia
     3) Hartford, Connecticut
     4) Los Angeles
     5) Baltimore
     6) San Diego
     7) Chicago
     8) Milwaukee
     9) Houston

Gasoline use data for these nine areas were obtained from
the Standards Development and Support Division (RDSD), in
EPA's Office of Mobile Sources, and were used by RDSD to
calculate fuel consumption figures contained in the draft
regulatory impact analysis for reformulated gasoline and
anti-dumping regulations (EPA, 1991a).  In 1990,  these 9
areas were responsible for 22.2% of the annual fuel use in
the United States.  Fuel use percentages for the two
extreme/severe ozone nonattainment areas located in
California (6.7%) were subtracted from this 22.2% since
California was assumed to have its own reformulated fuel
program statewide.  Fuel use for the extreme/severe ozone
nonattainment areas outside California was adjusted to
account for an estimated 15% "spillover" of reformulated
gasoline into uncovered areas.  This 15% estimate was
obtained from RDSD's draft regulatory impact analysis cited
above.  Then, fuel use in California (12.0%) was added to
the total.  The fuel use estimate for California
reformulated fuel under the California program was based on
the reported gasoline consumption for California in 1990
(12.0% of fuel used), obtained from data compiled by the
U.S. Department of Transportation (1990).  This estimate was
adjusted for the projected population increase in California
between 1990 and 1995 (about 9%; Wetrogan, 1990).  It was
assumed the increase in fuel consumption would be
proportional to the increase in population.  Winter
oxygenated gasoline use for areas participating in the
federal and California reformulated gasoline programs
(11.9%) was also subtracted from the total, thus giving the
estimate of 18% of nationwide gasoline use for this
component.

-------
                                                   EPA-420-R-93-005
                                                       April 1993

Winter oxygenated gasoline use was estimated using data
provided by EPA's Field Operations and Support Division
(FOSD) in the Office of Mobile Sources.  FOSD provided
percent gasoline use data for each of the 39 regions in the
oxygenated fuels CO program.  For the purposes of this
report, it was assumed that these same 39 areas would have
the winter oxygenate program in place for scenarios in 1995,
2000, and 2010.  These fuel use percentages were for the
entire year,  so assuming that fuel use was constant through
the entire year (which is admittedly an approximation since
fuel usage is greater in the summer versus winter months),
the percentages were multiplied by the fraction of the year
each region was expected to be in the program.  All regions
were assumed to have four month programs, with the following
exceptions:  Las Vegas and Phoenix with 5 month programs,
Los Angeles and Spokane with 6 month programs, and New York
with a 12 month program.  Winter oxygenated fuel use was
estimated to be 17% for 1995.  This fuel use estimate
includes an adjustment to account for 15% spillover.

Gasohol fuel use was assumed to remain constant at six
percent, relative to 1990, for this and all scenarios.

1995 Expanded Use of Reformulated Gasoline --In Section
211(k) of the Act, any ozone nonattainment area may opt into
the federal reformulated gasoline program.  In this
scenario, all ozone nonattainment areas are considered to
opt into the program.  At the time this analysis was done,
all Northeast states except Delaware and Vermont had opted
into the federal reformulated gasoline program and were thus
included.  These states include the following:
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
Maine
New Hampshire
Massachusetts
Rhode Island
New York
New Jersey
Pennsylvania
Connecticut
Maryland
Virginia
Washington, D.C.
Delaware and Texas have since opted into the program.
Northeast states and serious and above ozone nonattainment
areas may also adopt California emission standards, but this
will have no effect on the fuel use weightings for this
scenario, because of the similarity between federal and
California emission standards in 1995.
                          3-14

-------
                                                   EPA-420-R-93-005
                                                       April 1993

Gasoline use under Phase 1 of the federal reformulated
gasoline program and Phase 1 of the California reformulated
gasoline program combined was estimated to be 50% for this
scenario.  Phase 1 federal reformulated gasoline use in this
scenario was based on data from SDSB's draft regulatory
impact analysis for reformulated gasoline and anti-dumping
regulations (EPA,  1991a).   1990 fuel use percentages for
regions in California and the Northeast, calculated to be
29.6%, were subtracted from the total fuel use in all ozone
nonattainment areas (53.8%).  (To simplify the analysis,
individual nonattainment areas in the Northeast were not
considered and it was assumed the entire state received
reformulated gasoline.  While only those ozone nonattainment
areas included in the state governor's opt-in request are
technically included in the federal reformulated program,
these typically covered the major metropolitan areas of the
state.  In combination with the fungible gasoline
distribution system serving the Northeast, this should mean
that reformulated gasoline will be distributed throughout
the entire Northeast.   Similarly, individual nonattainment
areas in California were not considered since the entire
state of California was assumed to have California
reformulated gasoline).  The resultant percentage  (24.2%)
was increased by 15% to account for spillover.  Then
projected statewide fuel use percentages for all opt in
states and California were added (34.4%).  These projected
percentages were obtained by taking fuel consumption
estimates from the U.S. Department of Transportation (1990).
These estimates were adjusted for the projected population
increases in these states between 1990 and 1995 using
Department of Commerce data (Wetrogan, 1990).   It was
assumed increases in fuel consumption would be proportional
to increases in population.  Finally,  winter oxygenate
gasoline use in all regions and states participating in the
federal reformulated gasoline program and California (12.4%)
was subtracted from the total, resulting in a total fuel use
estimate for this component of 50%.  Fuel use for other
components (except for a reduction in baseline fuel use to
27%) remained the same as in the base control scenario.

2000, 2010 Base Control -- This scenario differs from the
1990 base control scenario in that Phase 2 federal and
California reformulated gasoline, rather than Phase 1,  will
be in use.  Phase 2 federal reformulated gasoline was
assumed for purposes of this study to have an RVP of 7.8
psi, while Phase 2 California reformulated gasoline has an
RVP of 7.0 psi.  Moreover, as previously discussed, federal
and California emission standards will be much different in
these years.   Thus, California reformulated fuel use in
California (8%) and winter oxygenated fuel use in California
(5%) were treated as separate components. Otherwise, it was
assumed fuel use in different programs will remain the same
(implying there will be no population shifts among regions).
This assumption was made because of the difficulty in
accurately projecting population changes in various regions
within states.

                          3-15

-------
                                                   EPA-420-R-93-005
                                                       April 1993

2000, 2010 Expanded Use of Reformulated Gasoline -- Once
again, this scenario differs from the 1995 expanded
reformulated gasoline use scenario, in that Phase 2 federal
and California reformulated gasoline will be in use, rather
than Phase 1,  and federal and California emission standards
will be markedly different in those years.

2000, 2010 Expanded Adoption of California Standards --
Under this scenario, all Northeast states and states with
ozone nonattainment areas categorized as extreme, severe, or
serious adopt California emission standards.  This scenario
assumes expanded use of reformulated gasoline also.

Phase 2 federal reformulated gasoline use under California
emission standards was estimated to be 30%.   First, fuel
use in all extreme, serious, and severe ozone nonattainment
areas was estimated at 29.3%, based on data from SDSB's
draft regulatory impact analysis for reformulated gasoline
and anti-dumping regulations (EPA, 1991a) .  Fuel use in
regions in California and Northeast states  (20.7%)  was
subtracted from this total.  Fuel use in all moderate and
marginal ozone nonattainment areas in all other states with
California standards was then added.  The resultant 12.8%
was adjusted for 15% spillover.  Then, projected statewide
fuel percentages for all Northeast states included in the
expanded reformulated fuel use scenario were added  (22.2%).
Finally, winter oxygenated fuel use in all extreme, severe,
and serious ozone nonattainment areas also classified as CO
nonattainment areas (7.24%) was subtracted from the total,
resulting in the total fuel use estimate for this component
of 30%.

Phase 2 federal reformulated gasoline use under federal
emission standards was estimated to be 13%.  First, fuel use
in all moderate and marginal ozone nonattainment areas was
estimated at 24.4%, based on data from SDSB's regulatory
impact analysis cited above.  Then fuel use in moderate and
marginal nonattainment areas with California emission
standards (12.1%) was subtracted from this total, and the
remaining 12.3% adjusted for 15% spillover.  Finally, winter
oxygenate fuel use in moderate and marginal ozone
nonattainment areas with federal emission standards  (0.8%)
was subtracted, resulting in a 13% estimate for this
component.

In this scenario, states with extreme, serious and above
ozone nonattainment areas adopting California standards may
have baseline fuel use under California standards outside
the nonattainment areas.  Fuel use for this component was
estimated by subtracting fuel use in federal reformulated
fuel areas with California standards  (37.0%) from fuel use
in all states with California standards, exclusive of
California (40.8%).  If there were any areas with winter
oxygenated fuel use under California standards which were in
ozone attainment, fuel use in these areas would also have to
                          3-16

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     be subtracted.  However, no such areas exist.  Thus total
     fuel use for this component is about 4%.

     Winter oxygenated fuel use under California emission
     standards was estimated to be 14%, while winter oxygenated
     fuel use under federal emission standards was estimated to
     be 3%.  The fuel use estimate for California reformulated
     fuel remained the same as under the expanded reformulated
     fuel use scenario for 2000 and 2010.  The remainder of fuel
     use was assigned to baseline fuel use under federal emission
     standards.

3.1.3.3  Emission Fractions Associated with Components

     After determining the nine components to be included for
each calendar year scenario, emission fractions for the various
fuels considered in these components were estimated.  For
baseline fuel use, emission fractions were calculated for
gasoline and diesel fuel.  As will be seen later, it was
relatively easy to calculate the diesel numbers.  For the
components with federal and California reformulated fuel use,
emission fractions were determined for 11% MTBE blends  (2%
oxygen).   For the gasohol component, emission fractions for 10%
ethanol were determined.  For the oxygenated fuels CO program,
emission fractions for 15% MTBE blends (2.7% oxygen) were
determined.  For the components with California emission
standards, the same emission fractions for Phase 1 federal and
California reformulated fuels were used,  since the fuel
characteristics are similar.  One difference is in RVP, which is
assumed to be 8.1 psi for Phase 1 federal and Phase 1 California
reformulated fuel, but 7.0 psi beginning in 1996 for Phase 2
California fuel.  This results in different benzene evaporative
emission fractions for the two components.  Also, Phase 1 federal
reformulated gasoline is assumed to have a higher RVP than Phase
2 (8.1 versus 7.8) resulting in slightly different benzene
evaporative emission fractions.

3.1.3.4  I/M Programs Associated with Components

     The CAA requires that all ozone and carbon monoxide
nonattainment areas must implement some kind of vehicle
Inspection and Maintenance  (I/M) program.  Depending on the
severity of the nonattainment problem, these areas will have to
implement either a basic I/M program  (required in areas with
moderate ozone nonattainment, and in marginal areas with existing
I/M programs) or an enhanced program  (required in most serious,
severe, and extreme ozone areas, as well as most carbon monoxide
areas registering greater than 12.7 ppm and larger metropolitan
statistical areas in the Northeast Ozone Transport Region) (EPA,
1992b).  The enhanced I/M program used in modeling includes
annual centralized testing of light duty vehicles and trucks, an
IM240 test, antitampering tests and functional tests of the
evaporative emission control system, including pressure and purge
testing.   The basic I/M program used in modeling was the ideal
minimum I/M program recommended by the Agency.
                               3-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The choice of I/M program to input into MOBTOX when modeling
components affects the resultant toxics emission factors.  In
fact, components have a mixture of no, basic and enhanced
programs in different areas.  To account for this, separate
MOBTOX runs for each type of I/M program were done for a
component, and the resultant emission factors weighted according
to the frequency of the I/M program within that component, to
obtain an I/M weighted emission factor.

     The I/M program weightings for each component were
calculated by comparing a EPSD database compiled by EPA's
Emission Planning and Strategies Division listing metropolitan
statistical areas with their current and expected future I/M
programs to the specific areas/cities covered under each scenario
component described above, and estimating the percentage of
individual scenario components covered by each type of I/M
program.  The breakdown of I/M programs expected in various areas
has changed slightly since this database was compiled.  The
weightings for each component are given in Table 3-4.

3.1.3.5  Estimating Risk Under Different Scenarios

     To estimate air toxics risk estimates under different
scenarios, I/M weighted emission factors for each component of a
scenario were weighted by the percent of total fuel use for the
component on a calendar year basis,  to obtain overall emission
factors for each scenario.  These emission factors for each
scenario were then multiplied by urban and rural g/mile to ug/m3
conversion factors, obtained from the Hazardous Air Pollutant
Exposure Model for Mobile Sources (HAPEM-MS; Johnson et al.,
1992), to obtain urban and rural annual average exposures.  These
urban and rural annual average exposures were then applied to the
equation described in Section 4.1 to calculate urban and rural
cancer cases in a given year for the air toxic of interest.

3.1.4  MOBTOX Emissions Model Inputs

3.1.4.1  HC Exhaust Reductions for Gasoline Oxygenated Blends


     MOBTOX also requires a single input for TOG exhaust
reduction for gasoline oxygenated blends.  MOBTOX already
calculates changes in evaporative emissions with gasoline in the
same fashion that MOBILE4.1 does.  MOBILE4.1 does this
calculation for evaporative emissions solely as a function of
RVP.
                               3-18

-------
                                                                                            EPA-420-R-93-005
                                                                                                 April 1993
Table 3-4.   I/M Program  Weightings for the  Various  Components Under Different Scenarios.
                          Percent of Total Fuel Use Within Components for Each I/M Program

Components /Scenarios
Baseline Gasoline Use
Federal Standards
Baseline Gasoline Use
California Standards
Federal/Calif. Reform. Gasoline
Use
Federal Phase 1 (1995-1999)
Calif. Phase 1 (1992-1995)
Federal/Calif. Standards
Federal Reform. Gasoline Use
Phase 2 (2000+)
Federal Standards
Federal Reform. Gasoline Use
Phase 2
Calif. Standards
Winter Oxygenated Gasoline Use
Federal/Calif. Standards (1995)
Federal Standards (2000, 2010)
Winter Oxygenated Gasoline Use
Calif. Standards (2000, 2010)
California Only
Calif. Reform. Gasoline Use
Phase 2 (1996+)
Calif. Standards
Gasohol Fuel Use
Federal Standards

I/M
Program
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
None
Basic
Enhanced
1990
Base
Control
32
68
0







32
68
0
1995
Base
Control
37
39
24

0
15
85


2
17
81


37
39
24
Expanded
Reform.
Gasoline
Use
49
35
16

17
35
48


2
17
81


49
35
16
2000, 2010
Base
Control
37
39
24


100

2
16
82
0
22
78
0
33
67
37
39
24
Expanded
Reform.
Gasoline
Use
48
36
16


20
34
46

2
16
82
0
22
78
0
33
67
48
36
16
Expanded
Adoption
Calif.
Standards
88
12
0
85
15
0

25
75
0
19
18
63
9
63
28
0
7
93
0
33
67
88
12
0
                                                 3-19

-------
                                                         EPA-420-R-93-005
                                                             April 1993

     However, changes  in  exhaust  hydrocarbons for gasoline
oxygenated blends were not  included in MOBILE4.1 even though
changes in exhaust CO were  included.   The changes in exhaust CO
were based on an analysis of  the  EPA emission factor data base
(EPA, 1991b).  A similar  analysis has since been done for exhaust
TOG emissions for gasoline  oxygenated blends using the emission
factor data  (EPA, 1992c).   Also,  an analysis using the emission
factor data was done for  Phase  1  and Phase 2 reformulated
gasoline for exhaust NMHC (EPA,  1992d);  similar reduction would
be found for TOG.  These  analyses were done for both normal and
high emitting vehicles since  the  two classes of vehicles have
different emission benefits (higher emitting vehicles achieve a
greater benefit with the  use  of  reformulated gasoline).

     The reformulated gasoline  analysis shows a 9.4% exhaust NMHC
reduction for a Phase  1 reformulated gasoline (with 2.0% oxygen
content) which will be assumed  to be the same regardless of the
type of I/M program used  (none,  basic,  enhanced)1.   The  remaining
reduction required for the  minimum 15% total vehicle emission
reduction comes from reduced  evaporative emissions from lower RVP
in the reformulated gasoline  --  8.1 psi for Class C areas
compared to an 8.7 psi baseline  value.   The MOBTOX runs were done
assuming temperature ranges (68-84F)  and RVPs for Class C areas.
For the purposes of this  report,  where benzene is the only toxic
component of evaporative  emissions and the evaporative benzene
contribution is small compared  to the exhaust benzene, it is
assumed that the same proportional reductions are obtained for
Class A and B areas as for  Class  C.  A somewhat similar
assumption is being used  for  temperature with the summertime
Class C type temperatures assumed to be representative of the
country as a whole for establishing ratios of vehicle toxic
emissions for the different components of the scenarios for the
years examined  (EPA, 1992a).

     This analysis also shows that the Phase 2 exhaust NMHC
reduction depends on the  stringency of the I/M program.   For
either no I/M or a basic  I/M, the exhaust reduction is 10.2%
NMHC.  Again, the remaining vehicle emission reductions come
about from reduced evaporative  emissions due to lower gasoline
RVP; a 7.8 psi RVP is assumed for Class C areas.  An enhanced I/M
program (which catches vehicles  with high evaporative emissions,
resulting in necessary repairs  and a lowering of these emissions)
increases the need for greater  exhaust emission reductions to
meet the minimum 20-25% total emission reduction.  A 14.4%
exhaust NMHC benefit is projected for Phase 2 fuel with an
enhanced I/M program.  For  the  purposes of this report,  a single
emission reduction of 22.5% (the  average of the 20% and 25%
numbers) is being used.
      This exhaust NMHC reduction (and NMHC reductions given in the following
paragraphs) was calculated relative to baseline fuel, rather than indolene.
Because of limitations in the MOBTOX model, hydrocarbon emission levels for
indolene and baseline fuel were assumed to be comparable.

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Also, emission reduction benefits have to be assigned to
Phase 2 California reformulated gasoline.  Based on an EPA
analysis of Arco data (DeJovine et al.,  1991), an initial number
to use is a 23% exhaust reduction benefit  (EPA, 1992e).

     For the winter oxygenate program,  TOG reductions are based
on a gasoline with 2.7% oxygen.  These numbers come from an
analysis of the EPA emission factor program (EPA, 1992c) and by
extrapolating the Phase 1 reformulated gasoline analysis from 2.0
to 2.7% oxygen content.   This results in a 12.7% exhaust TOG
benefit for the winter oxygenate program.

     Finally, an exhaust benefit is needed for use of gasohol.
The recent EPA analysis (EPA, 1992c) shows approximately a 15%
TOG exhaust benefit from use of ethanol.   This benefit can be
calculated by assuming that the emission factor data represent a
typical in-use spectrum of vehicles so that all the data can be
averaged.  The same number is obtained if the benefits for
normal, high, and very high emitters are taken and applied to the
proportion of these vehicles for the 1990 in-use fleet.   However,
this analysis shows a lower benefit for 10% ethanol relative to
15% MTBE for PFI normal and high emitters, but a higher ethanol
benefit relative to MTBE for PFI very high emitters.  In 2000 and
2010, the relative number of very high emitters is expected to be
lower.  Also, this analysis shows a higher ethanol benefit for
carbureted than fuel injected vehicles,  and carbureted vehicles
are likely to represent a very small portion of the fleet in 2000
and 2010.  Thus, the 15% TOG exhaust benefit from use of ethanol
might be an overestimate for these years.  In these later years,
an EPA estimate of a 9.6% NMHC exhaust benefit from use of 10%
ethanol  (1992f.) , calculated for 1990 technology type vehicles
with 1990 sales weightings for each technology type, might be
more appropriate.  Consideration is being given to modification
of MOBTOX in later years to reflect this difference.

     The benefits for the winter oxygenate and gasohol components
are assumed to be constant with calendar year, unlike the
reformulated gasoline benefits, which increase in 2000 versus
1995.  The CAAA specify increased benefits for reformulated
gasoline in 2000.  It is expected that fuel parameters such as
lower sulfur levels and changes in distillation characteristics
will give the increased benefit; it is also expected that these
parameters will not change in areas of the country where non-
reformulated gasoline is being used.  The reformulated gasoline
proposed rulemaking (EPA,  1991c) prohibits gasoline in the non-
reformulated areas from deteriorating as the oil companies
produce reformulated gasoline.

     In all these analyses, the benefits derived from 3-way
catalyst vehicles are being applied to the in-use fleet rather
than using separate benefits for 3-way catalyst, oxidation
catalyst, and non-catalyst vehicles.  Doing a separate weighting
makes little difference.  First, previous EPA guidance  (EPA,
1988) shows oxidation catalyst equipped vehicles obtain 14.5% and
12% exhaust emission benefits with 3.5% and 2.7% oxygen blends
(gasohol and MTBE/gasoline).   These numbers are remarkably close

                               3-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993

to the 15% and 12.7% benefits for the 3-way catalyst fleet.  The
only years where oxidation catalyst vehicles would have any
noticeable impact is the 1990 and 1995 runs; no effect would be
seen for the 2000 and 2010 projections.  Also, it is assumed that
the benefits for the 3-way catalyst vehicles would be the same as
for the 3-way plus oxidation catalyst vehicles that will be used
more in the future to meet stricter exhaust emission standards.
Since cars meeting future exhaust emission standards may have
less open-loop operation where electronic feedback does not
control exhaust TOG as much,  the benefits may be slightly lower
for the newer cars.  However, limited or no data are available
with which to make projections for benefits for future cars.
Thus, the same benefits are being assumed for cars with 3-way
catalysts and 3-way plus oxidation catalysts.

3.1.4.2  California LEV Standards

     As mentioned previously, California has separate 50,000 mile
exhaust emission certification standards for TLEVs, LEVs, ULEVs,
and ZEVs, beginning in 1994.   MOBTOX only accounted for these
separate categories in vehicle classes less than or equal to 8500
Ibs.   Also, MOBTOX did not account for intermediate compliance
standards.  Table 3-5 lists the 50,000 mile exhaust emission
certification standards, zero mile emission levels, 50,000 mile
deterioration rates, and 100,000 mile deterioration rates used in
MOBTOX for California vehicle emission categories with test
weights less than or equal to 8500 Ibs.

     When California standards are combined with what EPA defines
as an "appropriate" I/M program, greater emission reductions
would be expected than with no I/M, basic I/M, or even enhanced
I/M.   Thus, lower deterioration rates would be used in modeling
with California LEV standards than with federal standards.  (EPA
defines appropriate I/M as an I/M program that would ensure
vehicles will meet California LEV standards in use.)  However,
since this analysis did not assume all areas with California LEV
standards would have appropriate I/M, lower deterioration rates
were not used.  Thus, emission factors for components with
California emission standards are higher than they would be if
areas adopting these standards also adopted appropriate I/M
concurrently.

     Vehicles are classified in Table 3-5 by California emission
categories within Federal weight categories, rather than the
comparable California weight categories.  These categories
include passenger cars, or light duty gasoline vehicles  (LDGVs),
and four categories of light duty gasoline trucks  (LDGTs la, Ib,
2a, 2b).   ZEVs are not included in the table, since values in all
categories are zero.

     Table 3-6 lists the phase-in schedule used in MOBTOX for
TLEVs, LEVs, ULEVs, and ZEVs.  Although California has separate
phase-in schedules for light duty and medium duty vehicles, both
based on market shares, two phase-in schedules could not be
incorporated into MOBTOX due to limitations of the model.
Instead,  a combined phase-in schedule was input into the model,

                               3-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993

with fleet percentages weighted according to model year market
share projections for light and medium duty vehicles.  Any error
introduced into the model by combining phase in schedules would
be minor, since the market share of medium duty vehicles is small
relative to light duty vehicles.

3.1.4.3  Toxic Exhaust Fractions

     Emission fractions were disaggregated by vehicle class and
catalyst type for exhaust emissions, and fuel system for
evaporat ive emi s s ions.

     The following vehicle classes were included in the
calculations:  LDGVs, LDGTs, heavy-duty gasoline vehicles
(HDGVs),  light duty diesel vehicles  (LDDVs),  light duty diesel
trucks (LDDTs) and heavy duty diesel vehicles (HDDVs).  These
vehicle classes are consistent with those in MOBTOX.  LDGTs and
LDDTs were assumed to have the same mass fractions as LDGVs and
LDDVs, respectively.  For LDGV/LDGT exhaust emissions, fractions
were disaggregated by four catalyst types -- non-catalyst,
oxidation catalyst, three-way catalyst, and three-way plus
oxidation catalyst.  For LDGV/LDGT evaporative emissions,
fractions were disaggregated by fuel system -- either carbureted
or fuel injection  (PFI and TBI were considered to be the same so
we simply pooled all the fuel injection data).  HDGVs were
assumed to have either no catalyst or a three way catalyst with a
carbureted fuel system.  Calculations were done for vehicles
running on non-oxygenated gasoline, 10% ethanol, 5.5% MTBE, 9.0%
MTBE, 12.5% MTBE, 15% MTBE, and 16.4% MTBE.  Fuels with these
MTBE levels were used in major test programs.

     All exhaust mass fractions were calculated as fractions of
total organic gases  (TOG), on a vehicle by vehicle basis.  TOG
includes methane, ethane, and all oxygenated hydrocarbons, such
as aldehydes, and also alcohols and ethers when oxygenated blends
are used.  Mass of total hydrocarbons  (THC),  as determined by the
flame ionization detector  (FID), was multiplied by a THC to TOG
composite correction factor (CCF).   A recent EPA analysis  (1991d)
described the procedure for generating THC to TOG correction
factors for various vehicle class/catalyst combinations running
on gasoline or diesel.   These are the same correction factors
used in MOBILE4.1 and MOBTOX.   Although actual TOG values exist
for much of the data, this approach of calculating TOG using a
correction factor was used so that the emission fractions derived
are consistent with the TOG values contained in MOBILE4.1/MOBTOX.
A summary of CCFs for
                               3-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Table 3-5.   Zero Mile Levels and Deterioration Rates Based  on
 California  Exhaust Emission Certification Standards for Low
                      Emission Vehicles.
Vehicle Emission
Category
LDGV
TLEV
LDGV
LEV
LDGV
ULEV
LDGTla (< 3750 Ibs . )
TLEV
LDGTla
LEV
LDGTla
ULEV
LDGTlb (3751-5750 Ibs . )
TLEV
LDGTlb
LEV
LDGTlb
ULEV
LDGT2a (3751-3500)
(California medium duty)
TLEV
LDGT2a
LEV
LDGT2a
ULEV
LDGT2b (5751-8500)
TLEV
LDGT2b
LEV
LDGT2b
ULEV
50,000 Mile
Exhaust
Emission
Standard
0.125
0.075
0.040
0.125
0.075
0.040
0.160
0.100
0.050
0.500
0.160
0.100
0.500
0.195
0.117
Zero Mile
Level
0.1001
0.0600
0.0320
0.1001
0.0600
0.0320
0.1281
0.0800
0.0400
0.4002
0.1281
0.0800
0.4002
0.1561
0.0937
50,000 Mile
Deterioration
Rate
0.0518
0.0518
0.0518
0.0518
0.0518
0.0518
0.0518
0.0518
0.0518
0.0768
0.0518
0.0518
0.0768
0.0518
0.0518
100,000 Mile
Deterioration
Rate
0.0748
0.0748
0.0748
0.0748
0.0748
0.0748
0.0748
0.0748
0.0748
0.0768
0.0748
0.0748
0.0768
0.0748
0.0748
                              3-24

-------
                                                        EPA-420-R-93-005
                                                            April 1993
  Table  3-6.
Market Share Fractions for California Low Emission
          Vehicle  Categories.
Year
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003 +
Federal
Standard
0.91
0.86
0.82
0.75
0.50
0.25
0.02
0.00
0.00
0.00
TLEV
0.09
0.14
0.18
0.00
0.00
0.00
0.00
0.00
0.00
0.00
LEV
0.00
0.00
0.00
0.23
0.46
0.71
0.94
0.90
0.85
0.76
ULEV
0.00
0.00
0.00
0.02
0.02
0.02
0.02
0.05
0.10
0.15
ZEV
0.00
0.00
0.00
0.00
0.00
0.02
0.02
0.05
0.05
0.09
gasoline and diesel fueled vehicles is included in Table  3-7.   It
should be noted that CARB  uses THC to TOG CCFs which are higher
than EPA's.  Whereas THC as measured by FID assumes a C:H ratio
of  1:1.85 for every exhaust HC compound, CARB corrects this  FID
calculation for the true mix of C:H ratios to more accurately
report true mass.  Eventually, EPA may adopt this approach.
Also,  in making its adjustments, CARB inaccurately assumes  that
all oxygenated  compounds  (e.g. aldehydes) are not measured by
the FID.

     When estimating TOG for vehicles using MTBE fuel blends,
another adjustment factor had to be introduced to account for the
difference in emissions when a car runs on an MTBE blend  rather
than standard gasoline.  A relatively recent EPA analysis
(1989b), calculated relative adjustment factors for 0, 11 and 15%
MTBE to account for this difference.  The adjustment factors  are
as follows:

     1)  1.00 for 0% MTBE
     2)  1.0144 for 11% MTBE
     3)  1.0197 for 15% MTBE

Unlike the factors for gasoline vehicles, these correction
factors are not technology specific.  There is a linear
relationship between these adjustment factors and MTBE content;
                               3-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993
thus, a regression equation could be generated and adjustment
factors then calculated for various MTBE levels.  The analysis
also included a relative adjustment factor for 10% ethanol
(1.0232).   For a vehicle class/catalyst combination, THC as
measured by FID was first multiplied by the THC to TOG CCF, then
by this relative oxygenate adjustment factor to account for MTBE
or ethanol content.
       Table  3-7.   THC  to  TOG Composite Correction Factors
Vehicle Class
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
LDDV
HDDV
HDGV
HDGV
Catalyst
Technology
none
3 -way + ox
3 -way
ox


none
3 -way
Adjustment
Factor
1.0333
1.0175
1.0125
1.0170
1.0490
1.0342
1.0358
1.0178
     Sources for the data used to determine emission fractions
are summarized in Appendix Bl.   Appendix B2 contains a series of
spreadsheets listing, on a vehicle by vehicle basis, exhaust
emissions (and for benzene, evaporative emissions also) in
mg/mile for formaldehyde, acetaldehyde, 1,3-butadiene, and
benzene, TOG,  and resultant fractions of TOG.    For exhaust
emissions, vehicles were sorted by class, catalyst type, and
fuel,  as listed above.  For evaporative emissions, vehicles were
sorted by class,  fuel system and fuel.  Averages were calculated
for each fuel type within a vehicle class/catalyst or vehicle
class/fuel system category.  Appendix B3 contains summary
spreadsheets listing averages for the various categories.

     Because of a surfeit of extensive data on a reasonably large
number of vehicles for LDGVs and LDGTs with three-way catalysts,
only data from three recent Arco studies and the Auto/Oil Program
were used.  RDSD, in an early Notice of Proposed Rulemaking on
reformulated gasoline standards  (EPA, 1991c), limited its
analyses to Auto/Oil data  (Auto/Oil, 1990); specifically, current
1989-90 type vehicles with three way catalysts, running on an
"industry average" fuel, designated fuel A.  Although vehicles
were tested on a number of other non-oxygenated blends, for three
                               3-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993
way catalysts this analysis likewise only used fuel A data for
current vehicles.  Fuel A matches the baseline fuel
specifications in Section 211 of the Act, and this fuel/vehicle
technology combination is expected to be the most representative
for this analysis.  Arco has recently released three studies
(Boekhaus et al., 1991a and 1991b, DeJovine et al.,  1991), which
include a large amount of data on oxygenated fuel blends; thus,
it was useful to add these data to the Auto/Oil data.

     For other vehicle class/catalyst categories, all available
study data (even from programs where only 1 or 2 cars were
tested) were used because a very limited amount was available.

3.1.4.4  Other Inputs

     MOBTOX runs assumed regions modeled were low altitude
regions.  Also, the average speed assumed in MOBTOX runs was 19.6
miles per hour.  This is the average speed in the FTP test.  An
average daily temperature of 75F was assumed.   As mentioned in
Section 3.1.4.1,   the minimum temperature assumed was 68F and
the maximum temperature was 84F.  These represent the average
temperature and temperature ranges, respectively, in the FTP.

     Like MOBILE4.1 (EPA, 1991e), MOBTOX requires the user to
input certain assumptions about operating mode.  The federal FTP
has three distinct vehicle operating modes: cold start,
stabilized, and hot start.  The percentage of time vehicles spend
in each mode affects emissions (e.g., emissions are higher in
cold start mode).  MOBTOX requires the percentage of time spent
in cold start mode by non-catalyst vehicles, the percentage of
time spent in hot start mode by catalyst equipped vehicles, and
the percentage of time spent in cold start mode by catalyst
equipped vehicles.  The inputs in all runs for these three
variables were 20.6, 27.3, and 20.6, respectively.  The values
used for these three variables correspond to the conditions of
the FTP.

3.2  Methodology for Diesel Particulate Matter

     The Environmental Protection Agency prepared an estimate of
diesel particulate emissions in 1983 (EPA, 1983) .  In the 1983
analysis, EPA assessed the impact of "base" and  "relaxed"
scenarios on diesel particulate emissions in 1995, relative to
those in 1980 and 1986.  The base scenario assumed particulate
standards would be 0.20 g/mi, 0.26 g/mi, and 0.25 g/BHP-hr for
LDDVs, LDDTs, and heavy-duty diesel engines (HDDEs),
respectively.  The relaxed scenario assumed standards of 0.60
g/mi for LDDVs and LDDTs and 0.60 g/BHP-hr for HDDEs.

     In 1986, the Motor Vehicle Manufacturers Association and
Engine Manufacturers Association published an analysis of EPA's
diesel particulate matter study  (MVMA and EMA,  1986).  While MVMA

                               3-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993
and EMA generally agreed with EPA's methodology for estimating
diesel particulate emissions, they felt that many of the inputs
EPA used were outdated, and consequently, the contribution of
diesel engines to particulate levels was overstated.  MVMA and
EMA thus estimated particulate emissions using inputs which they
felt were more realistic.

     Another analysis of diesel particulate emissions was done by
EPA in 1987  (Carey, 1987), as part of an air toxics report.  This
analysis assumed that particulate standards in 1987 and later
years would be 0.20 g/mi and 0.26 g/mi for LDDVs and LDDTs,
respectively.  It also assumed a HDDE standard of 0.60 g/BHP-hr
for 1988-1990, 0.25 g/BHP-hr for 1991-1993 (except for buses, at
0.10 g/BHP-hr), and 0.10 g/BHP-hr for 1994 and later.

     Recently, Sienicki and Mago (1991) updated spreadsheets from
the 1986 MVMA and EMA analysis, and used these updated
spreadsheets to predict the total metric tons of diesel
particulate matter and concentration in urban areas from on-
highway vehicle fleets for the target years of 1995 and 2015.
Their analysis included more stringent standards for 1995 and
later years, set by EPA, rather than those assumed in the 1983
EPA diesel particulate matter study.
     Sienicki  (1992a, 1992b) has also used updated analyses to
predict total grams of urban diesel particulate matter, as well
as national fleet average emission factors, for the years 1990,
1995, 2000, and 2010.  These predictions utilize the most recent
inputs available; thus, the particulate emission factors derived
by Sienicki were used with only minor adjustments to develop
diesel particulate matter risk estimates for the air toxics
report.  Later, EPA may develop particulate emission factors to
use in developing risk estimates independently.

     A detailed discussion of the methodology is contained in
section 9.3.

3.3  Methodology for Gasoline Particulate Matter

     Historically, gasoline particulate matter has been difficult
to measure accurately due to the extremely low levels in exhaust.
As a result, emission data for gasoline particulate matter are
sparse.  For this report, the available emission data were
reviewed.  The limited data appear to indicate a correlation
between exhaust HC and gasoline particulate emissions.  Gasoline
particulate matter was thus estimated to be 1.1% of exhaust HC.
This percentage was then used in the MOBTOX model to calculate
in-use g/mile emission factors for gasoline particulate matter.
An alternative approach was to assign a single g/mile value for
gasoline particulate matter, based on the emission data.
Unfortunately, this alternative approach would not allow expected
changes to gasoline particulate emissions with either time or
with changes to fuels and/or vehicle standards.

                               3-28

-------
                                                          EPA-420-R-93-005
                                                               April 1993
     A detailed discussion of  the  available emission  data and the
derivation  of  the exhaust HC percentage is contained  in section
10.3.
                                3-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993
3.4  References for Chapter 3.0

Adler, J. M. and P. M. Carey.  1989.  Air Toxics Emissions and
Health Risks from Mobile Sources.  Ann Arbor, Michigan:
Environmental Protection Agency, Office of Mobile Sources.
Prepared for the Air and Waste Management Association.  AWMA
Paper 89-34A.6.

Auto/Oil Air Quality Improvement Research Program.  1990.  Phase
1 Working Data Set  (published in electronic form).  Prepared by
Systems Applications International, San Rafael, CA.

Auto/Oil Air Quality Improvement Research Program.  1991.
Technical Bulletin No. 6: Emission Results of Oxygenated Gasoline
and Changes in RVP.

Auto/Oil Air Quality Improvement Research Program.  1993.
Technical Bulletin No. 11.  A Study of Fuel Effects on Emissions
from High Emitting Vehicles.

Boekhaus, K. L., L. K. Cohu, L. A. Rapp and J. S. Segal.  1991a.
Clean Fuels Report 91-02:  Impact of EC-1 Reformulated Gasoline
Emissions and Their Reactivity on Five 1989 Cars.  Arco Products
Co., Anaheim, California.

Boekhaus, K. L., J. M. DeJovine, D. A. Paulsen, L. A. Rapp, J.S.
Segal and D. J. Townsend.  1991b.  Clean Fuels Report 91-03:
Fleet Test Emissions Data -- EC-Premium Emission Control
Gasoline.  Arco Products Co., Anaheim, California.

California Air Resources Board.  1990.  Proposed Regulations for
Low-Emission Vehicles and Clean Fuels: Staff Report.  Sacremento,
California: Mobile Source Division, Stationary Source Division.
Release Date: August 13, 1990.

Carey, P. M.  1987.  Air Toxics from Motor Vehicles.  Ann Arbor,
Michigan: U.S. Environmental Protection Agency, Office of Mobile
Sources.  Publication no. EPA-AA-TSS-PA-86-5.

Carey, P. M., and J. H. Somers.  1988.  Air Toxics Emissions from
Motor Vehicles.  Ann Arbor,  Michigan: Environmental Protection
Agency, Office of Mobile Sources.  Prepared for the Air and Waste
Management Association.  AWMA Paper 88-128.1.

Dasch, J. M., and R. L. Williams.  1991.  Benzene exhaust
emissions from in-use General Motors vehicles.  Environ. Sci.
Technol. 25:853-862.

DeJovine, J. M., K. J. McHugh, D. A. Paulsen, L. A. Rapp, J. S.
Segal, B. K. Sullivan, D. J. Townsend.  1991.  Clean Fuels Report
                               3-30

-------
                                                        EPA-420-R-93-005
                                                            April 1993
91-06:  EC-X Reformulated Gasoline Test Program Emissions Data.
Arco Products Co., Anaheim, California.

Environmental Protection Agency.  1983.  Diesel Particulate
Study.  Ann Arbor, Michigan: Office of Mobile Sources.  October,
1983.

Environmental Protection Agency.  1988.  Guidance on estimating
motor vehicle emission reductions from the use of alternative
fuels and fuel blends.  Ann Arbor, Michigan: Office of Mobile
Sources.  Publication no. EPA-AA-TSS-87-4.

Environmental Protection Agency.  1989a.  Analysis of the
Economic and Environmental Effects of Methanol as an Automotive
Fuel.  Ann Arbor, Michigan: Office of Mobile Sources.  Special
Report, September, 1989.

Environmental Protection Agency.  1989b.  Oxygenated fuel exhaust
HC, Evap, and MPG corrections.  Memo from Craig Harvey to Phil
Lorang, April 28, 1989.

Environmental Protection Agency.  1990a.  Analysis of the
Economic and Environmental Effects of Ethanol as an Automotive
Fuel.  Ann Arbor, Michigan: Office of Mobile Sources.  Special
Report, April, 1990.

Environmental Protection Agency.  1990b.  Analysis of the
Economic and Environmental Effects of Compressed Natural Gas as a
Vehicle Fuel; Volume I: Passenger Cars and Light Trucks.  Special
Report, April, 1990.

Environmental Protection Agency.  1990c.  Analysis of the
Economic and Environmental Effects of Compressed Natural Gas as a
Vehicle Fuel; Volume II: Heavy-Duty Vehicles.  Special Report,
April, 1990.


Environmental Protection Agency.  1991a.  Draft Regulatory Impact
Analysis: Reformulated Gasoline and Anti-Dumping Regulations.
Ann Arbor, Michigan: Office of Mobile Sources.

Environmental Protection Agency.  1991b.  CO reduction with the
use of oxygenated blends.  Memo from Greg Janssen to Phil Lorang,
July 31, 1991.

Environmental Protection Agency.  1991c.  Regulation of Fuel and
Fuel Additives: Standards for Reformulated Gasoline; Proposed
Rule.  Federal Register 56(131): 31176-31263.

Environmental Protection Agency.  1991d.  Correction factors to
convert THC to TOG.  Memo from Greg Janssen to Phil Lorang,
September 3, 1991.

                               3-31

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Environmental Protection Agency.  1991e.  User's Guide to
MOBILE4.1 (Mobile Source Emission Factor Model).  Ann Arbor,
Michigan:  Office of Mobile Sources.  July, 1991.

Environmental Protection Agency.  1992a.  Ratios of emission
factors at different temperatures for 1990 and 2000.  Memo from
Greg Janssen to Phil Lorang, February 5, 1992.

Environmental Protection Agency.  1992b.  Inspection/Maintenance
Program Requirements; Final Rule.  Federal Register 57(215):
52950-53014.

Environmental Protection Agency.  1992c.  Interim analysis of HC
reduction with the use of oxygenated fuel blends with 3-way
catalyst vehicles.  Memo from Greg Janssen to Phil Lorang,
January 23,  1992.

Environmental Protection Agency.  1992d.  Emission reduction
estimates of phase II reformulated gasoline.  Memo from Chris E.
Lindhjem to Phil Lorang, January 9, 1992.

Environmental Protection Agency.  1992e.  Emission levels of
target fuels.  Note from Chris E. Lindhjem to Gary Dolce,
February 10, 1992.

Environmental Protection Agency.  1992f..  Effect of Oxygenate on
Emissions.  Memo from Chris E. Lindhjem to Richard Rykowski,
January 7, 1992.

Environmental Protection Agency.  1993.  Regulation of Fuel and
Fuel Additives: Standards for Reformulated Gasoline; Proposed
Rule.  Federal Register 58(37): 11722-11763.

Johnson, T.  R., R. A. Paul, and J. E. Capel.  1992.  Application
of the Hazardous Air Pollutant Exposure Model  (HAPEM) to Mobile
Source Pollutants.  Durham, North Carolina: International
Technology Corporation, August 1992.

Motor Vehicle Manufacturers Association and Engine Manufacturers
Association.  1986.  Analysis of the Environmental Protection
Agency's Diesel Particulate Study and a Diesel Particulate
Emissions Projection.  September 22, 1986.

Motor Vehicle Manufacturers Association.  1990.  National Fuel
Surveys: Gasoline and Diesel Fuel.  Summer 1990.

Nebel, G. J.  1981.  The effect of misfueling on aldehyde and
other auto exhaust emissions.  APCA Journal 31: 877-879.

Sienicki, E. J. and R. S. Mago.  1991.  Reevaluation of Diesel
Engine Particulate Emission Inventories.  Presented at Air and
                               3-32

-------
                                                            EPA-420-R-93-005
                                                                April 1993
Waste Management Association Conference,  October  16-18,  1991,
Detroit,  Michigan.
                                 3-33

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Sienicki, E. J.  1992a.  Letter from E. J. Sienicki, Navistar
International Transportation Corporation, to Phil Lorang,
Environmental Protection Agency, January 6, 1992.

Sienicki, E. J.  1992b.  Letter from E. J. Sienicki, Navistar
International Transportation Corporation, to Phil Lorang,
Environmental Protection Agency, April 23, 1992.

Sigsby, J. E., S. Tejeda, W. Ray, J. M. Lang, and J. W. Duncan.
1987.  Volatile organic compound emissions from 46 in-use
passenger cars.  Environ. Sci. Technol. 21:466-475.

Smith, L. R. and P. M. Carey.  1982.  Characterization of exhaust
emissions from high mileage catalyst-equipped automobiles.  SAE
Paper 820783.

Stump, F. D., S. Tejada, W. Ray, D. Dropkin, F. Black, R. Snow,
W. Crews, P. Siudak, C. 0. Davis, L. Baker and N. Perry.  1989.
The influence of ambient temperature on tailpipe emissions from
1984 to 1987 model year light-duty gasoline vehicles.
Atmospheric Environment 23: 307-320.

Stump, F. D., S. Tejeda, W. Ray, D. Dropkin, F. Black, R. Snow,
W. Crews, P. Siudak, C. 0. Davis and P. Carter.  1990.  The
influence of ambient temperature on tailpipe emissions from 1985-
1987 model year light-duty gasoline vehicles -- II.  Atmospheric
Environment 24A: 2105-2112.

Stump, F. D., K. T. Knapp, W. D. Ray, R. Snow and C. Burton.  The
composition of motor vehicle organic emissions under elevated
temperature summer driving conditions  (75 to 105F) .
Unpublished.

U.S. Department of Transportation.  1990.  Highway Statistics
1990.  Washington, D.C.: Federal Highway Administration.

U.S. Department of Transportation.  1991.  Monthly Motor Fuel
Reported by States, April 1991.  Washington D.C.: Federal Highway
Administration.  Publication no. FHWA-PL-91-011.

Urban, C. M.  1980a.  Regulated and Unregulated Exhaust Emissions
from Malfunctioning Non-Catalyst and Oxidation Catalyst Gasoline
Automobiles.  Ann Arbor, Michigan: U.S. Environmental Protection
Agency, Office of Mobile Sources.  Publication no. EPA-460/3-80-
003.

Urban, C. M.  1980b.  Regulated and Unregulated Exhaust Emissions
from Malfunctioning Three-Way Catalyst Gasoline Automobiles.  Ann
Arbor, Michigan: U.S. Environmental Protection Agency, Office of
Mobile Sources.  Publication no. EPA-460/3-80-004.
                               3-34

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Urban, C. M.  1980c.  Regulated and Unregulated Exhaust Emissions
from a Malfunctioning Three-Way Catalyst Gasoline Automobile.
Ann Arbor, Michigan: U.S. Environmental Protection Agency, Office
of Mobile Sources.  Publication no. EPA-460/3-80-005.

Urban, C. M.  1981.  Unregulated Exhaust Emissions from Non-
Catalyst Baseline Cars Under Malfunction Conditions.  Ann Arbor,
Michigan: U.S. Environmental Protection Agency, Office of Mobile
Sources.  Publication no. EPA-460/3-81-020.

Urban, C. M., and R. J. Garbe.  1979.  Regulated and unregulated
exhaust emissions from malfunctioning automobiles.  SAE Paper
790696.

Urban, C. M., and R. J. Garbe.  1980.  Exhaust emissions from
malfunctioning three-way catalyst-equipped automobiles.  SAE
Paper 800511.

Warner-Selph, M. A., and L. R. Smith.  1991.  Assessment of
Unregulated Emissions from Gasoline Oxygenated Blends.  Ann
Arbor, Michigan: U.S. Environmental Protection Agency, Office of
Mobile Sources.  Publication no. EPA-460/3-91-002.

Wetrogan, S. I.  1990.  Projections of the Populations of States
by Ages, Sex, and Race: 1989 to 2010.  Washington, B.C.: U.S.
Department of Commerce, Bureau of the Census.  Series P-25,
publication no. 1053.
                               3-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993
4.0  EXPOSURE METHODOLOGY
     This chapter describes the methodology used to project
exposure to motor vehicle air toxics.  Exposure estimates have to
be made for two types of situations.  The first is an overall
annual exposure estimate which can be used for carcinogenic risk
assessments in the linear no-threshold model used by EPA to
predict cancer impact.  In this model, the lifetime or annual
cancer impact is the product of the lifetime or annual average
exposure level times the potency of the substance.  The second
exposure estimate needed is a localized exposure estimate for
specific microenvironments highly impacted by mobile source
emissions.  Such microenvironments include urban street canyons,
congested freeways, large commercial parking garages where many
vehicles exit at once such as after a sporting event, residential
garages attached to homes, and even roadway tunnels.  The concern
with exposure in these microenvironments generally is acute
non-cancer effects.

     These exposures can be estimated two ways.  The first is use
of models to predict either annual exposure or exposure in
certain microenvironments.  The second is using ambient data.
Ambient data can be used to estimate annual average exposures or
even localized exposure in microenvironments depending on monitor
location and averaging time for exposure.  However, few monitors
are designed to collect short term averages of motor vehicle
emissions in microenvironment areas where the highest exposures
would be expected  (such as residential garages).

4.1  Annual Average Population Exposure Estimation

     EPA work on developing models has emphasized those that
predict annual average exposure.  The models predicting annual
average exposure assume a person's actual exposure can be
predicted by levels measured at the monitors set up to measure
compliance with the National Ambient Air Quality Standards.
Several years ago, EPA conducted a number of studies measuring
human exposure to carbon monoxide in Washington B.C. and Denver
during the winter of 1982-83 after some initial work was done in
Stamford, Connecticut (Akland et al., 1985; Johnson, 1984;
Clayton et al.,  1985; Hartwell et al., 1984; Settergren et al.,
1984, Rumba, 1981).  In these studies, individuals carried carbon
monoxide monitors as they went about their day to day activities.
The individuals recorded their activities in a diary and the
personal CO monitor recorded the CO level during each activity.
When a person changed activities (as defined by guidelines given
to the person carrying the monitor), the person reset the monitor
so each monitor reading was associated with only one activity.
The measurements were taken over approximately 100 days.  On each
day, about 10 different individuals were selected to use the
monitors.  These studies were used to determine the relationship
of personal exposure levels to those found at the NAAQS monitors.
This work showed very good correlation between the monitor values
and ambient exposure for all groups except the top 10% of the
                               4-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993

exposed individuals which had greater exposure than would be
predicted by the NAAQS monitors.

     The EPA Office of Mobile Sources has adapted two models
developed by the EPA Office of Air Quality Planning and Standards
to predict annual average exposure to motor vehicle carbon
monoxide as a function of emission rate.  The first is the NAAQS
(National Ambient Air Quality Standard) Exposure Model or NEM
which was originally developed to predict exposure to carbon
monoxide.  The second is the Hazardous Air Pollution Exposure
Model (HAPEM),  which was originally developed to predict exposure
to air toxics generally from specific point sources.  A version
of HAPEM adopted for mobile sources, called HAPEM-MS (Johnson et
al.,  1992), is used in this study.

4.1.1  The NAAQS Exposure Model (NEM)

     In order to understand HAPEM-MS, it also helps to understand
the NAAQS Exposure Model (NEM).   The NEM has been used by EPA in
the past to estimate nationwide annual person-hours of exposure
to specific levels for any mobile-source pollutant of interest.
The model relies on an activity pattern model that simulates a
set of population groups called cohorts as they go about their
day-to-day activities.  Each of these cohorts is assigned to a
specific location type during each hour of the day.  Each of
several specific location types in the urban area is assigned a
particular ambient pollutant concentration based on fixed site
monitor data.  The model computes the hourly exposures for each
cohort and then sums up these values over the desired averaging
time to arrive at average population exposure and exposure
distributions.   Annual average exposures are theoretically
possible since a full year's data from fixed site monitors is an
input to the model (Johnson and Paul, 1982) .

     Southwest Research Institute, under EPA contract,   modified
the NEM so that it would determine exposures specifically from
mobile source pollutants (Ingalls, 1985).   The CO NEM was
selected since outdoor CO is largely a mobile source pollutant,
especially in urban areas where about 80% of CO comes from motor
vehicles.  Since the CO monitoring data, on which the CO NEM was
based, can be assumed to be related to mobile source emissions in
g/mile,  exposure to other mobile source pollutants can be
estimated from this model,  based on relative concentrations of
these pollutants to total emissions.  It is important to note,
however, that CO is relatively non-reactive photochemically.
Thus, non-reactive substances are modeled more accurately.

     The CO NEM divides all non-rural areas into the following
six neighborhoods:
                               4-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

          Urban residential
          Urban commercial
          Urban industrial
          Suburban residential
          Suburban commercial
          Suburban industrial

The neighborhoods were chosen to match the neighborhood
descriptions used in identifying EPA ambient monitor sites.  In
turn, each neighborhood is divided into the following six
microenvironments:

          Indoors,  work or school
          Indoors,  home or other
          Inside a transport vehicle
          Roadside
          Outdoors
          Kitchen

Each person in a city was assigned to a neighborhood type and to
a microenvironment within that neighborhood for each hour of the
day.  The population was divided into 12 age-occupation groups
with each of these groups being divided into subgroups; each
group and subgroup were assigned to a particular neighborhood
type and microenvironment depending on activity patterns.

     Also, time spent in the following three microenvironments
heavily impacted by motor vehicles was specifically accounted
for:

          Street canyons
          Tunnels
          Parking garages

     A total of 99 of the 346 monitors used in the 116 largest
urban areas in 1981 were used to assign ambient CO levels for
this model.  The monitors selected had to meet certain criteria.
One was that sufficient hourly data had to be available to
calculate an annual average level.  Also, the monitor could not
be in areas such as street canyons that would be highly impacted
by mobile source emissions.  Street canyons were represented
separately by another set of 23 monitors that are located near
street canyons.  Moreover, four different microenvironment
scaling factors were used  (as appropriate) to adjust the ambient
monitoring data to better represent CO levels in the locations
used in the model (Johnson and Paul, 1982):

     Microenvironment         Ambient CO Scaling Factor

          Indoors                  0.85
          Transport vehicle        2.10
          Roadside                 1.20
          Outdoors                 0.95
                               4-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     CO emission factors from MOBILES  have been used for the
1981 calendar year as inputs for each of the six neighborhoods
and the three microenvironments specifically impacted by mobile
sources.   These emission factors are generally based on the FTP.
The user assigns as input the emission factor for the compound of
interest in the year of interest.  In effect, the model takes the
ratio of the 1981 CO emission factor to the input emission factor
for the compound of interest and calculates the exposure based on
the ratio of the emission factors.   The output of the model is a
listing of person hours exposure in the urban areas in the
country as a whole for specific concentration levels.

     Rural exposure levels,  which are always much lower than
urban levels (with the exception of Class 8 heavy duty diesel
trucks which are operated mostly on interstate highways from city
to city versus in urban areas themselves) can be calculated
assuming exposures no greater than 2 ppm for CO.

     The NEM has an input for increased population in future
years and thus accounts for the greater number of people exposed.
However, it does not account for increases in the number of
vehicles (i.e., increases in vehicle miles traveled)  which is
handled separately from the model outputs.

4.1.2     Use of HAPEM-MS Model

     The EPA Office of Mobile Sources decided to update the
exposure model to incorporate some of the data available from the
Denver CO personal monitoring studies as well as some updated
personal activity data obtained by EPA in Cincinnati.  Also, a
model that would predict the actual annual average exposure  (and
number of people exposed to different annual averages) would have
more long term applicability than the modified NEM mentioned
above, which predicts only the number of person hours at specific
levels giving no specific annual average exposure levels.
Knowing the distributions of annual average exposure levels is
useful in determining whether there are large numbers of people
exposed to higher annual average levels balanced by a large
number of people at lower levels versus having the distribution
closely grouped around the overall annual average as a whole.
Such information can also be useful in evaluating carcinogenic
impacts from non-linear models versus the linear no-threshold
model used by EPA.

     The EPA Office of Air Quality Planning and Standards, in
conjunction with its contractor  (International Technology) that
developed the NEM, developed another exposure model,  the
Hazardous Air Pollution Exposure Model or HAPEM (Johnson, et al. ,
1991).  This model is generally used to predict annual average
exposures to toxic air pollutants dispersing from stationary
sources.  However, for this study, it was modified to predict
annual average exposures to toxic air pollutants from motor
vehicles.  The modified model is named the Hazardous Air
Pollution Exposure Model - Mobile Sources or HAPEM-MS (Johnson et
                               4-4

-------
                                                        EPA-420-R-93-005
                                                            April 1993

al.,  1992).   Like NEM, HAPEM-MS is based on the assumption that
CO can be used as a surrogate for motor vehicle exposure.

     The first step in adapting this model is to select
representative urban and rural areas for exposure estimates.  The
following 11 model urban areas were selected:

          Boston
          Denver
          Houston
          Los Angeles
          Minneapolis/St.  Paul
          New York City
          Philadelphia
          Phoenix
          St. Louis
          Spokane
          Washington B.C.

Paducah, Kentucky and Farmington, New Mexico were selected as
rural areas with sufficient CO monitoring data.

     Each urban area was then divided into exposure districts
generally based on locations of the CO NAAQS monitors so that the
number of exposure districts in the 11 urban areas would equal
the number of CO monitors for which annual average data exist for
the base year of the modified model (1988).   The population was
divided into the following demographic groups:

          Children, 0 to 5 years old
          Children, 0 to 13 years old
          Children, 14 to 18 years old
          Workers with low probability of outdoor work
          Workers with moderate probability of outdoor work
          Workers with high probability of outdoor work
          Nonworking adults under 35 years old
          Nonworking adults 35-54 years old
          Nonworking adults 55+ years old

Each demographic group was further subdivided into cohorts such
that each cohort represented a distinct combination of home and
work locations.  The fraction of time spent by each cohort in
each exposure district and microenvironment within the exposure
district was calculated based on a detailed activity pattern
study conducted in Cincinnati in which over 900 subjects
completed detailed three-day diaries.   The data were adjusted
based on season, day type (weekday or weekend), ambient
temperature, and other factors (Johnson, 1990).  All of the
nonworking cohorts were assumed to spend all of their time in the
residential exposure district.  The working cohorts were assumed
to spend their working time in
                               4-5

-------
                                                        EPA-420-R-93-005
                                                            April 1993

specific fractions of each exposure district and commuting times
were specifically considered  (Johnson, et al.,  1991).

     The model uses CO NAAQS fixed site monitoring data; however,
the purpose of siting fixed site monitoring  stations is not to
adequately measure ambient levels of CO but  to locate exceedances
of the CO standard.  As pointed out by several commentors, data
from fixed site monitor locations are not likely to be adequate
measures of ambient outdoor CO concentration in the community as
a whole.  As a result, the monitor values were adjusted based on
personal monitoring data obtained in Denver.  The personal
exposure monitor CO concentrations associated with a particular
microenvironment were regressed against simultaneous CO
concentrations reported by fixed site monitors to obtain
adjustment factors for each microenvironment.

     The following five microenvironments and factors with which
to adjust the NAAQS CO monitor value were incorporated into this
model:

     Microenvironment                             Factor

     Indoors - residence                          0.495
     Indoors - other locations (e.g., office)     0.619
     Outdoors - near road                         1.001
     Outdoors - other locations                   0.758
     Inside motor vehicle                         1.554

     A total of 323 urban areas with a population ranging from
58,000 to 8,600,000 were modeled by grouping each of these areas
with one of the 11 model urban areas.  These 323 areas were
qualitatively grouped with the above 11 based primarily on
geographical proximity but also factors such as estimated traffic
density and vehicle types used.  Thus, not many areas are grouped
with New York City since Manhattan and other parts of New York
City have relatively unique traffic density  and vehicle types
used compared even to other large Northeastern urban areas such
as Philadelphia, Boston, and Washington B.C.

     CO exposures for areas grouped with the above 11 modeled
areas are adjusted based on annual average CO levels in 1988 for
the urban area of interest versus the model  area with which it is
being grouped.  For the few areas where average annual CO levels
are not available, the CO levels were estimated to be the median
of those for the other areas grouped with the same model urban
area.  The combined population of the urban  areas (334 cities
total)  was 189,000,000.

     All rural type areas were grouped with  one of two model
rural areas (Paducah, Kentucky and Farmington,  New Mexico).
Exposure in these areas was also estimated.  The rural population
totaled 57,000,000.

     Annual average urban and rural CO exposures in 1988, as
predicted by HAPEM-MS, are 842 and 470 //g/m3, respectively.   The


                               4-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993

1988 fleet average carbon monoxide emission  factor  is  estimated
to be 29.6 g/mile using MOBILE4.1.  In MOBILE4.1  runs,  all  areas
were assumed to be Class C.  The minimum temperature was  assumed
to be 68F and the maximum temperature was 84F.  Gasoline was
assumed to have an RVP of 10.5 psi.  32% of  the country was
assumed to have no I/M and 68% was assumed to have  basic  I/M.

     The concentrations predicted by HAPEM-MS for 1988  were
divided by the 1988 MOBILE4.1 emission factor to  get the  g/mile
to //g/m3  conversion factors  shown below for both urban and rural
areas.

               CONVurban =28.4  (//g/m3)/(g/mile)

               CONVrural = 15.9  (//g/m3)/(g/mile)

     MOBILESa, an update of MOBILE4.1,  has been prepared  for
release since this analysis was done.  If MOBILE5a  CO  emission
factors were used in estimating the g/mile to  //g/m3 conversion
factors,  the factors would be roughly 30-35% lower.  However,  it
should be noted that the toxic emission factors using  MOBILESa
would be roughly 25-40% higher; thus, the overall cancer  risk
would not change appreciably.

     To obtain exposure estimates for the scenario  of  interest,
these conversion factors are multiplied by the emission factor
for the scenario of interest.  An additional adjustment factor is
applied to account for the increase in vehicle miles travelled
(VMT) in excess of the population increase for the  year of
interest relative to 1988 (EPA, 1992; Wetrogan, 1990).  These
adjustment factors are given below:

               ADJ1990 = 1.031
               ADJ1995 = 1.123
               ADJ2000 = 1.218
               ADJ,nin = 1.412
                  '2010
This additional factor is applied because HAPEM-MS does  not
account for changes in VMT.

     There are a number of limitations inherent  in HAPEM-MS  that
should be taken into account when reviewing the  results.   First,
the fixed site monitoring data were not adjusted to  account  for
non-motor vehicle sources of CO, since motor vehicles  are  thought
to be the predominant source of CO in urban areas.   This would
serve to overestimate the motor vehicle exposure estimates.   The
microenvironment factors built into the model attempt  to account
for other sources of CO to some extent by using  subjects that
were nonsmokers and using indoor CO levels only  in homes with no
CO sources (e.g., gas stove, smokers).

     Also, the reliability of the present methodology  depends on
the representativeness of the population by 6 cohorts  which  are
exposed to concentrations within 5 microenvironments.  Based on
the study of available exposure measurements, the upper  10


                               4-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993

percentile of the population exposures  (e.g. tollbooth
attendants) is believed to be underestimated.  The present use of
annual age concentrations to determine cancer risk assumes that
the dose-response relationship is linear.  Improved methodology
must be developed before a non-linear dose-response relationship
could be used.  In addition, assessing chronic non-cancer effects
will require consideration of a distribution of annual exposures
(e.g., the 90th percentile)  and not simply the annual mean
average.

     The microenvironment factors were estimated using data
obtained from one city  (Denver) over a four month period during
the winter of 1982-1983.  There is uncertainty as to whether the
resulting estimates are applicable to other areas and other
seasons.   The same general comment also applies to the activity
pattern data, which were collected in a single city (Cincinnati).

     CO data from only two rural areas were used to extrapolate
to all rural areas in the U.S.  There is uncertainty regarding
the representativeness of these two areas.

     Finally, there are uncertainties regarding the use of CO as
a surrogate for motor vehicle toxic emissions.  The
microenvironment factors may vary by pollutant.  In addition,
HAPEM-MS relies on the assumption that the ratio of emission
factors for CO and the toxic of interest remains constant for the
entire U.S.  Any variation in these ratios between or within
cities is not accounted for in HAPEM-MS.  Also, the model assumes
that the rates of release and chemical transformation for the
toxic of interest is similar to CO.  This will not be valid for
the more reactive pollutants such as 1,3-butadiene.   This is
addressed in more detail in the individual pollutant chapters.

4.1.3  Use of Ambient Monitoring Data

     Urban ambient monitoring data will be used to check the
reasonableness of the HAPEM-MS modeling results.  Several EPA
data bases exist which contain the results of various air toxics
monitoring programs.   These programs have set up monitoring
devices which are used to collect air samples all over the United
States over a period of months or years.  Scientists at EPA and
elsewhere analyze these samples to determine the total mass and
identity of various volatile organic compounds  (VOCs)  collected.
These VOCs include the toxics benzene, 1,3-butadiene,
formaldehyde, and acetaldehyde.

     One of these programs is the Aerometric Information
Retrieval System (AIRS), which became operational in 1987 and
utilizes a network of monitoring stations called the State and
Local Air Monitoring System (SLAMS) (EPA, 1989a).  This network
consists of monitoring stations set up by every state in
accordance with regulations promulgated in response to
requirements of the Clean Air Act.  The Office of Air Quality
Planning and Standards  (OAQPS) administers the AIRS program using
its computer facilities at Research Triangle Park, North


                               4-8

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Carolina.  OAQPS also established another network of monitoring
stations called the National Air Monitoring System  (NAMS).  The
NAMS network is part of the larger SLAMS network but must meet
more stringent monitor location, equipment, and quality
standards.

     The AIRS program allows state and local agencies to  submit
local air pollution data and also have access to national air
pollution data  (EPA, 1989a).  EPA uses data from AIRS in  order to
monitor the states' progress in attaining air quality standards
for ozone,  carbon monoxide, nitrogen oxides, sulfur oxides, and
lead through the use of State Implementation Plans  (SIPS) .  In
addition to containing information about each monitoring  site,
including the geographic location of the site and who operates
it, the AIRS program also contains extensive information  on the
ambient levels of many toxic compounds.  These include compounds
specifically discussed in this report:  benzene, 1,3-butadiene,
formaldehyde, and acetaldehyde.    The AIRS database catalogues
ambient air pollution data from 18 to 55 monitors in 15 to 23
urban areas, depending on the pollutant.    These monitors
collect a 24 hour sample every 12 days.  However, in some cases
not every target compound was detected in every sample.   The
samples in which this occurred for the compounds specifically
mentioned above were included as half the minimum detection limit
in the averaging of the data for this report.

     The AIRS database also contains data from the Toxic  Air
Monitoring System  (TAMS) (Evans, 1990; EPA, 1987, 1988).  The
TAMS network operated on a routine basis between 1985 and 1989.
By 1989, this network included 10 monitoring sites in the
metropolitan areas of Boston, Chicago, Houston, and
Seattle/Tacoma.  Working with state and local agencies and
receiving guidance from OAQPS, EPA's Atmospheric Research and
Exposure Assessment Laboratory  (AREAL) in Research Triangle Park,
North Carolina, administered the TAMS program.  The objectives of
this program included evaluating methods of sample collection and
analysis specifically for toxic air pollutants, beginning to
characterize ambient concentrations in selected urban
atmospheres, comparing concentration profiles among
and within urban areas, establishing baseline levels for  trend
assessments, and transferring monitoring technology and results
to EPA regional offices as well as to state and local agencies.
The TAMS program focused on attempting to monitor 96 volatile
organic compounds, including benzene and formaldehyde.
Monitoring devices collected a 24 hour sample every 12 days.
Data listed and used to calculate average concentrations  of
benzene and formaldehyde were collected between 1987 and  1991.
The minimum detection limit used in the collection of data was
0.1 ppb.  If a compound was not detected in a sample, then the
TAMS staff assigned one half the detection limit (0.05 ppb) as
the amount of the compound detected.

     Another air monitoring program is the Urban Air Toxic
Monitoring Program  (UATMP), which the EPA developed in 1987 to
assist state and local agencies in determining the nature and


                               4-9

-------
                                                        EPA-420-R-93-005
                                                            April 1993

extent of urban air toxic pollution  (McAlister et al.,  1989,
1990, 1991; Wijnberg and Faoro, 1989).   Data from the UATMP is
also used in air toxic risk assessment models  (EPA 1989b,c; EPA
1990 a,b).   In 1989, the UATMP had 14 monitors in 12 urban areas.
These urban areas included Camden, New Jersey; Washington, B.C.;
Miami, Pensacola, and Ft. Lauderdale, Florida; Chicago and
Sauget,  Illinois; Dallas and Houston, Texas; Baton Rouge,
Louisiana;  Wichita,  Kansas; and St. Louis, Missouri.  In 1990,
the UATMP had 12 monitors in 11 urban areas, of which 9 also
participated in the 1989 monitoring program.  These  9 urban areas
are Camden, New Jersey; Washington, B.C.; Pensacola, Florida;
Chicago and Sauget,  Illinois; Houston,  Texas; Baton  Rouge,
Louisiana;  and Wichita, Kansas.  Urban monitors added included
Orlando,  Florida; Toledo, Ohio; and Port Neches, Texas.

     In 1989 and 1990, the UATMP network simultaneously monitored
37 non-methane organic compounds, selected metals, benzo(a)pyrene
(1989 only),  formaldehyde, acetaldehyde, and acetone for a 24
hour period once every 12 days.  The UATMP database  lists the
data collected from the monitoring network using two methods.  In
the first method, only the concentrations above the  detection
limit of the compound are included in the data.  In  the second
method,  if the concentration of a compound is zero or below the
detection limit, then one half of the compound's detection limit
is incorporated into the data.  The second method was used
because it seemed more accurate and allowed a greater number of
samples to be averaged.  Data collected in 1989 and  1990 were
studied for this report.

     The 1990 UATMP ambient monitoring data presented two unique
situations.  The first of these was the inclusion of Port Neches,
Texas in the sampling program.  This urban area does not affect
the overall average for benzene, formaldehyde, or acetaldehyde,
but the effects are significant for 1,3-butadiene.   Port Neches,
Texas does possess areas with high point source concentrations
and, coupled with the fact that the location of the  monitor is
difficult to ascertain in relation to the point sources, the
decision was made to exclude the 28 samples from Port Neches from
the final average ppb for the entire program.  This  changes the
ambient mean level from 1.02 ppb to 0.14 ppb.

      The second situation involves the problem of previous ozone
interference when testing the carbonyl samples.  Beginning with
the 1990 UATMP program, ozone was removed from ambient air
through the use of an ozone denuder.  This ozone denuder was
added to the sampling system after the heated sample probe to
eliminate ozone, which is an interferant with the material used
to trap the carbonyls in the sampling cartridge.  The use of an
ozone denuder in the sampling system results in higher and
presumably more accurate reported formaldehyde concentrations;
hence, only 1990 UATMP carbonyl data will be used to determine
ambient levels of formaldehyde and acetaldehyde.

     The National Ambient Volatile Organic Compounds (NAVOC) Data
Base contains approximately 175,000 records on the observed


                               4-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993

concentrations of 320 VOCs observed in one hour air samples taken
every 24 hours between 1970 and 1987  (Shah et al., 1988; Hunt et
al.,  1988).   However, only the most current NAVOC data, taken
during 1987, is used in this report.  In addition, samples which
had zero concentrations of the four compounds discussed in this
section were included in averaging the data for this report.
These air samples were collected using indoor and outdoor
monitoring devices.  Personal monitors were also used.  The types
of locations of outdoor monitoring sites included remote, rural,
suburban, and urban areas, as well as near specific point sources
of VOCs.  Indoor monitoring sites consisted of non-industrial
workplaces and residential environments.  Personal monitors are
also included in the indoor category.  This database was an
interim precursor to the air toxics portion of  (AIRS).  For this
report, only the outdoor urban data were used.

     Table 4-1 summarizes the average concentrations  (in ppb) of
benzene, 1,3-butadiene, formaldehyde, and acetaldehyde found at
the monitoring sites of each air monitoring program.  The table
also shows the total number of observations for each average and
the number of sites which monitored the compounds in each
program.  For AIRS, the average concentrations of the four
compounds are listed separately for 1987 through 1989.  It should
be noted that methods of averaging the data are not consistent
between air monitoring databases.  Also, in the NAVOC monitoring
network, samples were taken for one hour in a 24 hour period
while the other monitoring networks collected a 24 hour air
sample every 12 days.
                               4-11

-------
                                                                EPA-420-R-93-005
                                                                     April 1993
Table 4-1.
Summary of  Air Monitoring Program Results For
Benzene, 1,3-Butadiene,  Formaldehyde,  and
Acetaldehyde

AIRS
1987 Level
(ppb)
# Obs.
# Site
1988 Level
(ppb)
# Obs.
# Sites
1989 Level
(ppb)
# Obs.
# Sites
1990 Level
(ppb)
# Obs.
# Sites
1991 Level
(ppb)
# Obs.
# Sites
UATMP
1989 Level
(ppb)
# Obs.
# Sites
1990 Level
(ppb)
# Obs.
# Sites
TAMS
1987-89 Level
(ppb)
# Obs.
# Sites
NAVOC
1987 Level
(ppb)
# Obs.
# Sites
Benzene
2.13
422
23
1.27
560
36
1.28
373
13
	
	
1.96
397
14
1 .47
349
12
1.31
439
10
2.21
564
31
1,3 -Butadiene
	
0.46a
12
2
	
0.21a
97
6
0.10
117
6
0.21
390
13
0.14b
321
11
	
0.34
9
6
Formaldehyde
2.79
100
14
2.65
293
16
	
	
	
2.12
418
14
4.21C
356
12
1.75
362
10
3.25
36
1
Acetaldehyde
1.34
82
13
1.63
253
16
	
	
	
1.36
418
14
1.72C
356
12
	
	
aAverage ppb from all  four quarter data sites,  excluding Houston,  Texas.
bAverage ppb from all  sites, excluding Port Neches, Texas.
"Average ppb from all  sites.  All samples had an ozone denuder added; hence,
only these  ambient levels will  be used later in the report, since they
accounted for ozone interference.
                                   4-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993

4.1.4  Procedure for Calculating Cancer Incidences or Deaths

     Urban and rural cancer incidences  (for 1,3-butadiene,
acetaldehyde, formaldehyde) or deaths  (for benzene and diesel
particulate matter) were calculated for each scenario using the
following equation:

                    EXP x UR x POP H- 70 = CAN

where:

     EXP =     HAPEM-MS derived urban or rural annual average
               exposure,//g/m3,  adjusted to account for the
               increase in vehicle miles travelled  (VMT)  in
               excess of the population increase  for the  year of
               interest relative to 1988, as described in Section
               4.1.2 above

     UR =      EPA unit risk in cancer cases or deaths per person
               exposed in a lifetime to 1 //g/m3

     POP =     urban or rural U.S. population for the year of
               interest

                         Urban          Rural
               1990      187,418,000    62,473,000
               1995      194,715,000    64,905,000
               2000      200,811,000    66,937,000
               2010      211,542,000    70,514,000

               The population estimates were obtained from
               Wetrogan, 1990.

     70 =      years per lifetime

     CAN =     annual cancer incidences or deaths

Urban and rural cancer incidences or deaths were  added to obtain
total cancer incidences or deaths.  In some cases, the 1990
HAPEM-MS derived exposures were adjusted to better agree  with the
ambient data.  If an adjustment factor was deemed necessary, it
was applied to the HAPEM-MS derived exposures for all years.
This is discussed in more detail in the individual pollutant
chapters.

4.2  Short-Term Microenvironment Exposures

While carcinogenic effects are assumed to have no threshold and
are linearly related to exposure levels  (even at very low
exposure levels),  non-carcinogenic effects are assumed to have a
threshold.  At low enough levels, there would be no adverse
effect as would be found at higher levels; thus,  the concern is
with higher level exposures to these pollutants unless the
threshold is low enough to encompass even the low exposure
                               4-13

-------
                                                        EPA-420-R-93-005
                                                            April 1993

levels.  The higher level short term exposures occur in
microenvironments heavily impacted by motor vehicles.

     Particular attention needs to be given to the human
exposures in microenvironments such as personal garages, public
parking garages, in vehicles during transit, and other situations
where there is relatively little dispersion of emissions.
Maximum exposures are projected in personal garages, based on
modeling data.  The personal garage scenario was evaluated in the
development of the standards for emissions from methanol-fueled
motor vehicles  (EPA,  1989d).  It was determined in that analysis
that validation data for the personal garage were not available,
so that the accuracy of the model could not be determined.  The
number of uncertainties uncovered in this rulemaking demonstrated
that more investigation into cold idle emissions and exposure
modeling is necessary before accurate conclusions can be drawn
regarding public health risk in the personal garage.  EPA's
Office of Research and Development  (ORD) is presently re-
evaluating the personal garage model.  The determination of the
health risk in microenvironments in general is also complicated
by the fact that health information for non-cancer effects is
limited and no RfCs have been developed by EPA for many of the
compounds of concern.

     The exposure to air toxics in microenvironments will be
evaluated by presenting data from studies that have measured
toxics concentrations for people in-transit and in various other
microenvironments where elevated levels are expected.  New
methodology must be developed before risks to acute exposures can
be assessed.
                               4-14

-------
                                                        EPA-420-R-93-005
                                                            April 1993
4.3  References for Chapter 4
Akland, Gerald G., T.D. Hartwell, T.R. Johnson, and R.W.
Whitmore.  1985.  Measuring human exposure to carbon monoxide in
Washington B.C. and Denver, Colorado during the winter of
1982-83.  Environ. Sci. and Technol.,  19: 911-918.

Clayton, C.A., S.B. White, and S.K. Settergren.  1985.  Carbon
monoxide exposure of residents of Washington B.C.: comparative
analysis.  EPA contract report.

EPA.  1987, 1988.  Toxic air monitoring system  (TAMS) status
report.  Research Triangle Park, N.C.: Office of Research and
Development,  Atmospheric Research and Exposure Assessment
Laboratory.

EPA.  1989a.   AIRS user's guide volumes I-VII.  Research Triangle
Park, N.C.: Office of Air Quality Planning and Standards.  June
1989.

EPA. 1989b.  1989 Urban Air Toxics Monitoring Program.  Office of
Air Quality Planning and Standards, Research Triangle Park, N.C.
EPA Report No. EPA-450/4-91-001.

EPA. 1989c.  1989 Urban Air Toxics Monitoring Program Aldehdye
Results.  Office of Air Quality Planning and Standards, Research
Triangle Park, N.C.  EPA Report No. EPA-450/4-91-006.

EPA.  1989d.   Standards for Emissions from Methanol-Fueled Motor
Vehicles and Motor Vehicle Engines; Final Rule.  40CFR Part 86,
Volume 54, No. 68. 14426-14613.

EPA. 1990a.  1989 Urban Air Toxics Monitoring Program.  Office of
Air Quality Planning and Standards, Research Triangle Park, N.C.
EPA Report No. EPA-450/4-91-024.

EPA. 1990b.  1990 Urban Air Toxics Monitoring Program Carbonyl
Results.  Office of Air Quality Planning and Standards, Research
Triangle Park, N.C.  EPA Report No. EPA-450/4-91-025.

EPA.  1992.  MOBILE4.1 Fuel Consumption Model.  Ann Arbor,
Michigan: Office of Mobile Sources.

Evans, Gary,  F.  1990.  Final report on the operation and
findings of the toxics air monitoring system  (TAMS).  Research
Triangle Park, N.C.: Office of Research and Development,
Atmospheric Research and Exposure Assessment Laboratory.

Hartwell, T.D., C.A. Clayton, R.M. Michie, R.W. Whitmore, H.S.
Zelon, S.M. Jones, and D.A. Whitehurst.  1984.  Study of carbon
monoxide exposure of residents of Washington D.C. and Denver,
Colorado.  Part I.  EPA contract report.

Hunt, William, F., Jr., Robert Faoro,  A.B. Hudischewskyj, and
A.K. Pollack.  1988.  The EPA interim database for air toxics


                               4-15

-------
                                                        EPA-420-R-93-005
                                                            April 1993

volatile organic chemicals.  Research Triangle Park, N.C.:
Systems Applications, Inc. and the U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards.
Publication no. EPA-450/4-88-014.

Ingalls, Melvin N.  1985.  Improved mobile source exposure
estimation.  Environmental Protection Agency.  Publication no.
EPA-460/3-85-002.

Johnson, Ted.  1984.  A study of personal exposure to carbon
monoxide in Denver, Colorado.  Rsesearch Triangle Park, North
Carolina: Environmental Protection Agency.  Publication no. EPA-
450/5-83-004.

Johnson, T.R.  1990.  Estimation of ozone exposure in Houston
using a probabilistic version of NEM.  Paper 90-150.  Air and
Waste Management Association Meeting.

Johnson, Ted R., R.A. Paul, J.E. Capel, and W. Ollison.  1991.
Estimation of incremental benzene exposure associated with
several bulk gasoline storage facilities in North Carolina, 1st
Annual meeting of the Internatinal Society of Exposure Analysis,
Atlanta, Georgia.

Johnson, Ted, and R.A. Paul.  1982.  The NAAQS exposure model
(NEM) applied to carbon monoxide.  Draft final contract reports
by PEDCo Environmental for EPA.

Johnson, T.R., J.E. Capel, and D.M. Byrne.  1991.  The estimation
of commuting patterns in applicatins of the hazardous air
pollutant expsoure model.  AWMA Paper no. 91-172.6.  Presented at
Air and Waste Management Association Meeting, Vancouver, Canada.

Johnson, T.R., R.A. Paul, and J.E. Capel.  1992.  Application of
the Hazardous Air Pollutant Exposure Model (HAPEM) to Mobile
Source Pollutants.  International Technology Corporation, Durham,
North Carolina.

McAlister,  Robert, A., Wendy H. Moore, Joann Rice, Dave-Paul
Dayton, Robert F. Jongleux, Phyllis L. O'Hara, Raymond G.
Merrill, Jr., and Joan T. Bursey.  1989.  Nonmethane organic
compound monitoring program final report volume II: urban air
toxics monitoring program, 1988.  Research Triangle Park, N.C.:
U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards.  Publication no.: EPA-450/4-89-005.
                               4-16

-------
                                                        EPA-420-R-93-005
                                                            April 1993
McAlister, Robert, A., Wendy H. Moore, Joann Rice, Emily Bowles,
Dave-Paul Dayton, Robert F. Jongleux, Raymond G. Merrill, Jr.,
and Joan T. Bursey.  1990.  Urban air toxics monitoring program,
1989.  Research Triangle Park, N.C.: U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards.
Publication no. EPA-450/4-91-001.

McAlister, Robert, A., David L. Epperson, and Robert F. Jongleux.
1991.  Urban air toxics monitoring program aldehyde results,
1989.  Research Triangle Park, N.C.: Radian Corp. and the U.S.
Environmental Protection Agency, Office of Air Quality Planning
and Standards.  Publication no. EPA-450/4-91-006.

Rumba, R.G.  1981.  Monitoring human exposure to carbon monoxide
in Stamford, Connecticut.  EPA report.

Settergren, S.K., T.D. Hartwell, and C.A. Clayton.  1984.  Study
of carbon monoxide exposures of residents of Washington D.C.  --
additional analyses.  EPA contract report.

Shah, Jitendra, J. and Emily K. Heyerdahl.  1988.  National
ambient volatile organic compounds data base update  (project
report).   U.S. Environmental Protection Agency, Atmospheric
Sciences Research Laboratory, Research Triangle Park, N.C.
EPA/600/3-88/010(a).

Shah, Jitendra, J. and Emily K. Heyerdahl.  National ambient
volatile organic compounds data base update (project summary).
U.S. Environmental Protection Agency, Atmospheric Sciences
Research Laboratory, Research Triangle Park, N.C.  EPA/600/S3-
88/010.

Wetrogan, S.I.  1990.  Projections of the Population of States by
Age, Sex, and Race:  1989 to 2010.  Washington, D.C.:  U.S.
Department of Commerce, Bureau of the Census.  Series P-25,
publication no. 1053.

Wijnberg, Louis and Robert Faoro.  1989.  Urban air toxics
monitoring program -- results of aldehyde monitoring fiscal year
1988.  PEI Associates and the U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards.  Research
Triangle Park, N.C.  September 1989.
                               4-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993

5.0 BENZENE

5.1 Chemical and Physical Properties  (EPA, 1988)

     Benzene is a clear, colorless, aromatic hydrocarbon which
has a characteristic sickly, sweet odor.  It is both volatile and
flammable.  Selected chemical and physical properties of benzene
are presented in Table 5-1.

     Benzene contains 92.3 percent carbon and 7.7 percent
hydrogen with the chemical formula C6H6.  The benzene molecule is
represented by a hexagon formed by the six sets of carbon and
hydrogen atoms bonded together with alternating single and double
bonds.  The benzene molecule is the cornerstone for aromatic
compounds, most of which contain one or more benzene rings.

     Benzene is nonpolar, meaning it carries no major area of
charge in any portion of the molecule and no net electrical
charge considering the molecule as a whole.  It is relatively
soluble in water and is capable of mixing with polar solvents
(solvents which carry major portions of opposing charges within
the molecule)  such as chloroform, acetone, alcohol, and carbon
tetrachloride without separating into two phases.

     Benzene is a highly stable aromatic hydrocarbon, but it does
react with other compounds primarily by substitution of a
hydrogen atom.  Some reactions occur which can rupture or cleave
the molecule.
Table 5-1.  Chemical and Physical Properties of Benzene.
Property
Molecular weight
Melting point
Boiling point
Density at 20C (68F)
Vapor Pressure at 25C (77F)
Flash point (closed cup)
Solubility in water at 25C
Conversions at 25C
Value
78.11 g/mole
5.5C (41.9F)
80.1C (176. 2F)
0.879 g/ml
0.13 atm.
-11.1C (12.02F)
1.8 g/L
1 ppm = 3.25 mg/m3
1 mg/liter = 313 ppm
                               5-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.2 Formation and Control Technology
     Benzene is present in both exhaust and evaporative
emissions.  Data show the benzene level of gasoline to be about
1.5%, with diesel fuel containing relatively insignificant levels
of benzene.  Some exhaust benzene is unburned fuel benzene.  Some
work indicates that non-benzene aromatics in the fuels can cause
about 70 to 80% of the exhaust benzene formed.  Some benzene also
forms from engine combustion of non-aromatic fuel hydrocarbons.
The fraction of benzene in the exhaust varies depending on
control technology and fuel composition but is generally about 3
to 5%.  The fraction of benzene in the evaporative emissions also
depends on control technology  (e.g., whether the vehicle has fuel
injection or a carburetor) and fuel composition (e.g., benzene
level and RVP) and is generally about 1%.  These data also show
that diesel vehicles account for only about 3% of the total
mobile source benzene emitted  (Carey, 1987) .

     Control techniques are available and in use for both
evaporative and exhaust emissions of benzene.  For example,
positive crankcase ventilation (PCV) and evaporative controls
reduce evaporative emissions of benzene.  Fuel evaporative
controls were installed on all 1971 light-duty gasoline vehicles.
An absorption/regeneration system, one of the most common
evaporative control techniques, is a canister of activated carbon
that traps vapors such as benzene.  The vapors are ultimately fed
back to the combustion chamber.  Catalysts on automobiles have
been effective in reducing benzene exhaust emissions.  The amount
of reduction achieved is dependent on the type of catalyst
technology used and the drive cycle of the vehicle (EPA, 1988).
It is also dependent on the exhaust hydrocarbon standard to which
the vehicle has been certified.

     Section 202(a)(6) of the Act states that the EPA shall
promulgate standards for control of refueling emissions, after
consultation with the Department of Transportation.  EPA decided
not to promulgate such standards in March of 1992 after questions
were raised by the National Highway Traffic Safety Administration
on the safety of the onboard carbon canisters.  This decision was
also based on information concerning the effectiveness of this
technology to combat ozone.  The EPA then issued guidance for
vapor recovery technology, known as Stage 2,  to be installed on
gasoline pumps (EPA,  1992a).   On January 22,  1993 a Federal
appellate court directed EPA to promulgate standards requiring
automakers to control refueling emissions for new cars and light-
duty trucks.

5.3  Emissions

5.3.1  Emission Fractions Used in the MOBTOX Emissions Model

     Benzene fractions were determined using a series of
equations relating fuel properties to THC percent benzene in
exhaust and evaporative emissions rather than the actual vehicle
data in Appendix B2.   However, actual vehicle data were used to


                               5-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

corroborate the accuracy of these equations.  Please refer to
Appendix B2 for the emission fractions used in this section.

5.3.1.1  Benzene Exhaust Emission Fractions

     For benzene exhaust from gasoline vehicles, separate
equations were used for three-way catalysts, three-way plus
oxidation catalysts, and other catalyst types.  For vehicles with
a three-way catalyst, running on baseline gasoline, the following
equation was used:

3-way Bz%THC = 1.077 + 0.7732*(volume % benzene)
              + 0.0987*(volume % aromatics - volume % benzene).

This equation was obtained by the EPA Regulatory Development and
Support Division  (RDSD) from work done by Chevron Oil Company
(Chevron 1991).   An analogous equation for NMHC is being used by
RDSD in the Supplemental NPRM, on regulation of fuels and fuel
additives in reformulated and conventional gasoline (EPA, 1991a).
For vehicles with a three-way plus oxidation catalyst, running on
baseline gasoline, the equation used was:

3-way + ox Bz%THC = 0.6796*(volume % benzene)
                    + 0.0681*(volume % aromatics) - 0.3468.

This equation was obtained from the draft Regulatory Impact
Analysis for RVP regulations  (EPA, 1987a).  For vehicles with no
catalyst or an oxidation catalyst, the equation used was:

other Bz%THC = 0.8551*(volume % benzene)
               + 0.12198*(volume % aromatics) - 1.1626.

This equation was also given in the draft Regulatory Impact
Analysis for RVP regulations.  The same benzene fractions were
used for HDGVs.   Benzene fractions for LDDVs, LDDTs, and HDDVs
were based on the benzene fractions of THC used in the 1987 EPA
motor vehicle air toxics report (0.0240 for LDDVs and LDDTs;
0.0110 for HDDVs) (Carey,  1987).  These were then adjusted to
give benzene fractions of TOG using the TOG/THC ratios given in
Table 3-7.

     Next, it was necessary to determine whether an adjustment
factor should be applied to the gasoline vehicle equations for
MTBE and ethanol blends.   To calculate an appropriate adjustment
factor, percent exhaust benzene for individual vehicles in
various studies was compared for baseline and oxygenated blends
(Appendix B4).  The comparison between fuels was done on a
vehicle by vehicle basis because of the large amount of
individual variation in emissions among vehicles.  If data for
different vehicles running on a fuel type are pooled and then
compared, it is difficult to isolate trends probably due to car
to car variations.  Also,  if data for different MTBE or ethanol
blends (with the different aromatic, olefin content, etc.) are
pooled, fuel effects may also make comparison difficult.  This
comparison was performed for 15% MTBE and 10% ethanol.  Then, an


                               5-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993

average percent change  (expressed as a fraction) was calculated
for each catalyst type.  This average percent change was added to
1, representing the baseline emissions with gasoline, and the
equations were then multiplied by the resultant factor.  Since
the average percent change was calculated for 15% MTBE, for
blends with other MTBE levels the average percent change was
multiplied by a ratio of percent MTBE to 15.  Actual benzene TOG
fractions (from Appendix B2) were compared to predicted benzene
THC,  with and without the adjustment factor  (Appendix B5).  No
significant difference was observed in the accuracy of the
equations, with and without the adjustment factor, with both
typically predicting TOG benzene levels within +/- 20%.  Based on
these comparisons, the THC equations without adjustment factors
were used to determine benzene percent TOG fractions for MTBE and
ethanol blends, since these seemed to be just as accurate.

     Once the appropriate equations for benzene were chosen, the
fuel properties (% aromatics, benzene, and oxygen) to use with
the equations were then determined.  The resultant emission
fractions are contained in Appendix B6.

       For reformulated gasoline in CY 2000+, the fraction of
exhaust benzene (and the other toxics mentioned in CAAA Section
219)  is assumed to remain the same relative to CY 1995-1999.
However, the mass of TOG will be reduced as required by the CAAA.
As a result, the mass of benzene is assumed to be reduced
proportionately to TOG for exhaust.

     As mentioned earlier, under the California standards, fuel
characteristics for oxygenates are similar to those under the
reformulated gasoline regulations.  However, under Phase 2 of
CARB's reformulated fuel regulations, which go into effect in
1996, RVP will be limited to 7.0 psi.  Since RVP has little
effect on benzene exhaust fractions, it was assumed that benzene
exhaust fractions under the California standards are the same as
under reformulated gasoline regulations.

5.3.1.2  Benzene Diurnal and Hot Soak Evaporative Emission
Fractions

     For benzene evaporative emissions from gasoline vehicles,
two equations were used to determine fractions -- one for diurnal
emissions, and one for hot soak emissions.  The equation used for
diurnal emissions from vehicles running on gasoline MTBE blends
was:

Diurnal Benzene = [(1.3758 - (0.0579*(weight % oxygen/2.0)
                  - 0.080274*RVP)]*(volume % benzene).

The equation used for hot soak emissions from vehicles running on
MTBE fuel was:

Hot Soak Benzene = [(1.4448 - (0.0684*(weight % oxygen/2.0)
                   - 0.080274*RVP)]*(volume % benzene).
                               5-4

-------
                                                        EPA-420-R-93-005
                                                            April 1993

To calculate diurnal and hot soak emissions from vehicles running
on gasohol, the oxygen term  (which was developed specifically for
MTBE) was eliminated.  The oxygen term used for MTBE fuel
accounts for test data which have shown that the presence of MTBE
tends to reduce benzene's evaporative and running loss benzene
emissions.  However, test data with ethanol have not shown such
an effect on benzene emissions separate from its effect on
overall evaporative VOC emissions.  Thus, the diurnal and hot
soak equations for gasohol (and gasoline) are:

Diurnal Benzene =  [1.3758 -  (0.080274*RVP)]*(volume % benzene)

Hot Soak Benzene =  [1.4448 - (0.080274*RVP)]*(volume % benzene).

For both MTBE and gasohol, these equations were derived from GM's
tank vapor emissions model (1991)  for representative tank
temperatures, and were used in RDSD's reformulated gasoline NPRM,
(EPA, 1991a), and in the supplemental NPRM (EPA, 1992b).  The
supplemental NPRM states that this model was derived for vehicles
typical of in-use emissions rather than vehicles meeting the
emission standards.  Once again,  the same emission fractions were
used for HDGVs, LDGVs, and LDGTs.   Evaporative emissions from
LDDVs,  LDDTs, and HDDVs were assumed to be negligible.

     The accuracy of these equations was tested in predicting
evaporative benzene levels from fuel properties in baseline
gasoline, MTBE blends, and gasohol by comparing predicted benzene
levels to benzene levels from actual vehicle data (Appendix B5).
The equations underpredicted evaporative benzene emissions
significantly  (e.g., % predicted versus % observed)  for vehicles
with carburetors, and even more significantly for fuel injected
vehicles.  This may be because the model that the equations were
based on was derived for "typical in-use" vehicles,  and almost
all the vehicles in the database were vehicles with lower
evaporative emissions.  The equations were used in these
analyses, in order to be consistent with the reformulated fuels
NPRM.  In any case, evaporative benzene emissions are less than
20% of total vehicle benzene emissions so this underprediction is
not serious.

     Diurnal and hot soak benzene emission fractions for various
programs included in modeling components are included in Appendix
B6.  It was also assumed that the fraction of benzene in overall
evaporative emissions remains the same, regardless of
temperature, since all MOBTOX runs were done at a single
temperature range  (68-84).   Benzene evaporative emissions are
small compared to exhaust benzene so using a single temperature
range versus explicitly setting evaporative emissions of benzene
equal to zero in winter months is probably justified.  Higher
benzene exhaust emissions in winter months are not being
considered, so these approximations may cancel one another.

     For exhaust benzene emissions, RVP was not part of the
equations used to predict emission fractions.   RVP does affect
evaporative emission fractions,  however.  For example, an RVP of


                               5-5

-------
                                                        EPA-420-R-93-005
                                                            April 1993

8.1 was assumed for federal reformulated fuels in CY 1995-1999
for Class C areas, but an RVP of 7.8 in CY 2000+.  This results
in slightly higher diurnal and hot soak benzene  fractions for CY
2000+ compared to 1995-1999.  The overall mass of evaporative
benzene decreases, however, because the reduction in overall
evaporative THC is greater at lower RVPs.  Also, for California
standards, the benzene exhaust fractions are assumed to be the
same as those for EPA 1995-1999 reformulated gasoline standards.
For the 1995 scenarios, the diurnal and hot soak benzene
fractions came from EPA's reformulated gasoline  regulations.
However, since CARB's Phase II reformulated fuel regulations,
taking effect in 1996, specify an RVP of 7.0, scenarios for 2000
and 2010 used different benzene diurnal and hot  soak emission
fractions, calculated using the different RVP value.

5.3.1.3  Benzene Running, Resting, and Refueling Loss Evaporative
Emission Fractions

     Running loss evaporative emission fractions for benzene were
assumed to be the same as for hot soak.  Resting loss emission
fractions were assumed to be the same as for diurnal.  Refueling
loss benzene fractions were set at 0.01, following the VOC/PM
Speciation Data System (EPA, 1990a).

5.3.2  Emission Factors for Baseline and Control Scenarios

     The fleet average benzene emission factors  as determined by
the MOBTOX emissions model are presented in Table 5-2.  When
comparing the base control scenarios relative to 1990, the
emission factor is reduced by 46% in 1995, by 60% in 2000, and by
68% in 2010.  The expansion of reformulated fuel use in 1995
reduces the emission factor by another 7% relative to 1990.  In
2000, the expanded control scenarios reduce the  emission factor
by another 6 to 9%, and in 2010, by another 4 to 6%, relative to
1990.
                               5-6

-------
                                                             EPA-420-R-93-005
                                                                  April 1993
Table  5-2.   Annual  Emission  Factor Projections  for Benzene.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0882
0.0472
0.0413
0.0351
0.0301
0.0305
0.0285
0.0248
0.0228
Percent
Reduction
from 1990
-
46
53
60
66
65
68
72
74
                                  5-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.3.3  Nationwide Motor Vehicle Benzene Emissions
     The nationwide benzene metric tons are presented in Table 5-
3.   Total metric tons are determined by multiplying the emission
factor from Table 5-2 (g/mile) by the VMT determined for the
particular year.  The VMT, in billion miles, was determined to be
1793.07 for 1990, 2029.74 for 1995, 2269.25 for 2000, and 2771.30
for 2010.  When comparing the base control scenarios relative to
1990,  the metric tons are reduced by 39% in 1995, by 50% in 2000,
and remains constant at 50% in 2010.

5.3.4 Other Sources of Benzene

     Mobile sources account for approximately 85% of the total
benzene emissions.  Of the mobile source contribution, the
majority comes from the exhaust.  The remaining benzene emissions
(15%)  come from stationary sources.  Many of these are related to
industries producing benzene, sometimes as a side product, and
those industries that use benzene to produce other chemicals.
Coke ovens are responsible for 10% of the 15% with the other 5%
attributable to all other stationary sources (Carey, 1987).

     Approximately 70% of mobile source benzene emissions (60% of
total benzene emissions)  can be attributed to onroad motor
vehicles, with the remainder attributed to nonroad mobile
sources.  This figure is based on a number of crude estimates and
assumptions.  First, it was estimated that 25% of total VOC
emissions are from onroad vehicles, and 10% are from nonroad
sources  (based on a range of 7-13%).  These estimates were
obtained from EPA's Nonroad Engine and Vehicle Emissions Study
(NEVES)  (EPA, 1991b).  Thus, about 70% of mobile source VOC is
attributable to onroad vehicles.  This VOC split was adjusted by
onroad and nonroad benzene fractions (described below) to come up
with the estimate of 70% of mobile source benzene from on-road
vehicles.

     For nonroad vehicles, benzene was estimated to be about 3.0%
of exhaust hydrocarbon emissions and 1.7% of evaporative
hydrocarbon emissions, based on the NEVES report (EPA, 1991b).
The 1.7% evaporative emissions estimate is actually an estimate
for refueling emissions of nonroad gasoline engines.  Since no
estimate existed for benzene evaporative emissions, it was
assumed that percent benzene evaporative emissions was the same
as refueling.  The split between exhaust and evaporative benzene
emissions was assumed to be 80% exhaust to 20% evaporative.
Thus,  the overall benzene fraction of nonroad hydrocarbon
emissions was estimated to be 2.74%.

     For onroad vehicles, benzene was estimated to be 3.89% of
exhaust hydrocarbon and 1.04% of evaporative hydrocarbon
emissions.  The exhaust fraction is a 1990 fleet average toxic
fraction, with fractions in Appendix B2 weighted using 1990 VMT
fractions.  The evaporative fraction is the benzene fraction
given in Appendix B6
                               5-1

-------
                                                             EPA-420-R-93-005
                                                                  April 1993
Table  5-3.  Nationwide Metric Tons Projection  for Benzene.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0882
0.0472
0.0413
0.0351
0.0301
0.0305
0.0285
0.0248
0.0228
Metric
Tons
158,149
95, 804
83,828
79,651
68,304
69,212
78, 982
68,728
63,186
                                  5-9

-------
                                                        EPA-420-R-93-005
                                                            April 1993

for gasoline-fueled vehicles.  The split between exhaust and
evaporative hydrocarbon emissions was estimated to be 60% exhaust
to 40% evaporative.  Thus, the overall benzene fraction for
onroad hydrocarbon emissions was 2.74%.  If the VOC split is
adjusted by these benzene fractions for onroad and nonroad
emissions, 70% of benzene from mobile sources is estimated to
come from on road vehicles.

     Data from EPA's Total Exposure Assessment Methodology (TEAM)
Study identified the major sources of exposure to benzene for
much of the U.S. population.  The TEAM study is described in
detail in a four-volume EPA publication (EPA, 1987b).   The study
measured 24-hour personal exposures in air and drinking water for
20 to 25 target volatile compounds for a selected group of
subjects from six cities.  Subjects were selected according to
census information, socioeconomic factors,  and their proximity to
potential industrial and mobile sources.  Large numbers of homes
were visited by trained interviewers to collect information on
age, sex, occupation,  smoking status, and other factors for each
person in the household.  A total of 700 subjects representing
more than 800,000 residents of the various cities were sampled.

     The final results of TEAM total benzene exposure (Wallace,
1989),  show the most important source of benzene exposure is
active smoking of tobacco.  Smoking accounts for about half of
the total population exposure to benzene.   Personal exposures due
to riding in automobiles, passive smoking,  and exposure to
consumer products account for roughly one-quarter of the total
exposure, with outdoor concentrations of benzene, due mainly to
vehicle exhaust, accounting for the remaining portion.
Occupational exposures, pumping gasoline,  living near chemical
plants or petroleum refining operations, food, water,  and
beverages appear to account for no more than a few percent of
total nationwide exposure to benzene.


5.4  Atmospheric Reactivity and Residence Times

     Laboratory evaluations indicate that benzene is minimally
reactive in the atmosphere,  compared to the reactivity of other
hydrocarbons.  This then gives benzene long-term stability in the
atmosphere.  Oxidation of benzene will occur only under extreme
conditions, involving a catalyst or elevated temperature or
pressure.  Photolysis is possible only in the presence of
sensitizers and is dependent on wavelength absorption.

     The information that follows on transformation and residence
times has been largely excerpted from a report produced by
Systems Applications International for the EPA (Ligocki et al.,
1991) .
                               5-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.4.1 Atmospheric Transformation Processes

     A variety of atmospheric transformation processes of
importance to air toxics can occur in urban atmospheres.  Species
can be destroyed by reaction with atmospheric oxidants, or by
photolysis.  The oxidant of most importance on a global scale is
the hydroxyl radical  (OH),  which is produced photolytically
everywhere in the atmosphere and reacts with nearly every organic
substance.  In urban atmospheres, ozone  (03)  can also be an
important oxidant.  At night, OH concentrations drop off
significantly because little OH is produced in the absence of
sunlight, but concentrations of the nitrate radical  (N03)  can
increase to fairly high levels when high concentrations of
nitrogen oxides  (NOX)  are present.   Other atmospheric oxidants
are the hydroperoxyl radical (H02) ,  the oxygen atom,  and the
chlorine atom (Cl),  which may be important under some
circumstances.  A few atmospheric species react directly with
nitrogen dioxide  (N02) .

     Photolysis refers to decomposition following absorption of
ultraviolet radiation.  While reaction with oxidants is common to
virtually all organic molecules, photolysis usually involves
oxygenated intermediates containing the carbonyl (C=0) bond, such
as formaldehyde and acetaldehyde. (Whitten, 1983).

     Many atmospheric species react rapidly in the aqueous phase
of clouds, fogs, and aqueous aerosols.  For highly soluble and
highly reactive species, this can be a major atmospheric
transformation pathway.

     Atmospheric transformation can also include the condensation
of gaseous species onto atmospheric aerosols.  This process is a
function of the vapor pressure of the species, the amount of
aerosol present in the atmosphere, and the temperature.  Although
benzene, 1,3-butadiene, formaldehyde, and acetaldehyde exhibit
sufficiently high vapor pressures that they will not condense
onto aerosols to any significant degree, this process can be of
major importance for other types of air toxics such as polycyclic
organic matter associated with diesel and gasoline particulate.

5.4.2 Gas Phase Chemistry of Benzene

      The aromatic ring structure of benzene is extremely stable
and resistant to chemical attack.  Therefore, of all the toxic
species to be addressed in this report, benzene is the least
reactive in the atmosphere.  Not only does benzene oxidize
slowly, but one of its key oxidation products, phenol, suppresses
ozone formation under N0x-limited conditions  because it acts as a
free radical scavenger.
                               5-11

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.4.2.1 Gas Phase Reactions

     The only benzene reaction which is important in the lower
atmosphere is reaction with the OH radical.  Yet even this
reaction is relatively slow.  The reaction proceeds by OH
addition, forming a complex which can decay back to the original
reactants.  At relevant tropospheric temperatures, this decay
rate is negligible.  The temperature dependence of this reaction
is not well known.  Benzene reacts more slowly with OH radicals
than do most other aromatic species.  Toluene and m-xylene react
five times and 19 times as fast as benzene, respectively
(Atkinson, 1990).

     The reactions of benzene with oxygen atoms, ozone  (03),  and
nitrate  (N03)  have been measured.   Since the rate of these
reactions are slower than rate of reaction of benzene with OH,
and/or their concentrations in the atmosphere are generally much
lower than OH concentrations, these reactions are not important
in the atmospheric transformation of benzene.

     Reactions with Cl atoms are known to be important in the
stratosphere, where they are associated with the ozone depletion
cycle.  However,  Cl concentrations in the troposphere are low,
roughly three orders of magnitude smaller than OH concentrations
(Singh and Kasting, 1988).   Since the reaction rate is only a
factor of ten larger than the OH rate, this reaction is not
important in the lower atmosphere.

5.4.2.2 Reaction Products

     The observed stable products from the atmospheric oxidation
of benzene are phenols  (phenol and nitrophenol), and aldehydes
(mainly glyoxal  [CHO]2)  with reported yields of 24 percent for
phenol (Atkinson et al., 1989) and 21 percent for glyoxal  (Tuazon
et al., 1986).  Nitrophenol yields of 3 percent at low NOX
concentrations have been reported  (Atkinson, 1990).  Thus, the
known products do not completely account for all the mass
reacted.   Phenol is highly reactive under smog conditions and
will react rapidly with OH radicals during the daytime and with
N03  radicals  at nighttime.   Glyoxal is also highly reactive,  with
a chemistry similar to that of formaldehyde.  Both phenol and
glyoxal,  besides being highly reactive, are also highly
water-soluble, and will be removed rapidly by incorporation into
clouds or rain.

5.4.3 Aqueous Phase Chemistry of Benzene

     Benzene reacts rapidly in aqueous solution with the OH
radical and the sulfate radical (S04~) ,  forming products that  are
removed by their incorporation into rain.  Despite the rapid
reaction of benzene in aqueous solution, its low solubility
limits the importance of aqueous-phase processes for this
compound and it will not be incorporated into clouds or rain to
any large degree.


                               5-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993

5.4.4 Atmospheric Residence Times

5.4.4.1 Definition and Limitations

     In assessing the potential impact of emissions of toxic
species into the atmosphere, it is important to have some measure
of their atmospheric persistence.  Species which persist for long
periods of time can accumulate to high concentrations during
stagnation periods and can be transported further from their
sources than species which are destroyed rapidly.  Common
measures of atmospheric persistence are the residence time, or
lifetime (T), and the half-life, both of which are measures of
the time required for a fixed concentration of a species to decay
to a certain percentage of its initial concentration.  The
residence time and the half-life are times at which the
concentration has been reduced to 37% and 50% of its original
value,  respectively.  The atmospheric residence time is thus a
mathematical formulation which provides a common ground for
comparison of the persistence of different chemical species.

     One limitation of residence time calculations is that they
cannot be used to predict ambient concentrations of toxic
species.  Concentrations are determined by atmospheric dispersion
characteristics combined with emissions patterns, formation, and
removal rates.  In urban areas, the effective residence time of
toxic species in the atmosphere may be determined by the time
required to transport emissions out of the air basin, rather than
the time required for their chemical or physical removal within
the air basin.  Also, residence time calculations do not
incorporate chemical production rates for secondary species.
Thus, residence time calculations may indicate that a species
such as formaldehyde is removed rapidly during the daytime, when
actually formaldehyde is being produced more rapidly than it is
being removed.  Finally, residence time calculations consider
atmospheric reactions as destruction processes and do not
consider the possible transformation of toxic species into
equally toxic products.

     Despite these limitations, atmospheric residence time
calculations can be valuable when viewed in context with these
other issues.

5.4.4.2 Chemical and Physical Processes

     A variety of chemical and physical processes must be taken
into consideration when determining the residence time of a
compound.  Chemical processes include gas-phase chemical
reactions,  photolysis, and in-cloud chemical destruction.
Physical processes include wet and dry deposition.  With regard
to gas-phase chemical reactions, typical atmospheric oxidant
concentrations are required for residence time calculations.
Concentrations of OH radicals are of particular importance, since
chemical residence times for many atmospheric species are
determined by their rate of reaction with the OH radical.  At
night,  photolysis is absent and OH radical concentrations are


                               5-13

-------
                                                        EPA-420-R-93-005
                                                            April 1993

very low.  Other chemical reactions, such as reaction with N03
radical or 03,  may be important at night.

     For species which photolyze, photolysis can compete with the
OH reaction as the dominant daytime removal mechanism.
Photolysis rates depend only upon the amount of ultraviolet  (UV)
radiation reaching the lower troposphere,  and thus can be
determined on the basis of latitude, altitude, and time of year.

     Cloud cover is often neglected in atmospheric residence time
calculations.  Yet, many areas of the United States experience a
significant degree of cloud cover throughout much of the year.
Cloud cover affects the residence time of atmospheric pollutants
in two major ways.  First, clouds attenuate the solar UV
radiation at ground level, slowing photolysis rates and
decreasing radical concentrations.  Second, clouds are themselves
a reactive medium in which chemical transformation will take
place.  Therefore, the presence of clouds may increase or
decrease the atmospheric residence time of specific pollutants.

     The physical processes of wet and dry deposition can also be
significant removal routes for some atmospheric pollutants.  Wet
deposition refers to the capture and removal of species by
hydrometers including rain, snow, hail, etc.  Dry deposition
refers to the loss of atmospheric species to surfaces by
diffusion, sedimentation, impaction, etc.   The atmospheric
residence time due to physical processes depends upon whether the
species is present in the atmosphere only as a vapor, or
partially adsorbed to particles.  This partitioning is determined
by the vapor pressure of the species.   For calculation purposes,
all precipitation was assumed to be in the form of rain, since
partitioning of organic compounds from the atmosphere to snow or
other forms of frozen precipitation is less well understood.

     The rate of dry deposition of volatile organic compounds is
highly uncertain.  A method proposed for incorporation into
regional air quality models was used to calculate dry deposition
rates, although its validity has not been demonstrated for
organic species.

     For species which are present in the atmosphere as gases or
vapors, deposition processes may be reversible.  For instance,
volatile compounds present in rain which falls on a surface such
as a street or sidewalk and subsequently evaporates will return
to the atmosphere.  It has been proposed that formaldehyde
rapidly deposits to dew-covered surfaces overnight and in the
early morning, and then is released when the dew evaporates at
mid-morning  (Ireson et al., 1990).  To the extent possible, these
types of reversible processes should not be considered in
atmospheric residence time calculations.

5.4.4.3 Generation of Input Values
                               5-14

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The oxidant concentrations required for the residence time
calculations were obtained from trajectory model simulations for
the four cities, Los Angeles, St. Louis, New York, and Atlanta.

     These locations were chosen to represent a variety of
regions within the United States, and were also chosen because
summer model input data were available for these cities.  The
simulations were conducted using the Ozone Isopleth Plotting
Model, Version 4 with Carbon Bond Mechanism IV  (OZIPM-4)  (Hogo
and Gery, 1988).  This is a model which is used routinely to
predict ozone formation as a function of VOC and NOX emissions;
however, as an intermediate step, it calculates radical
concentration such as OH.

       Simulations began at 9 a.m. and continued through 4 a.m.
the following day.  The calculations were conducted for daytime
and nighttime,  and then weighted by the length of the day and
night to obtain 24-hour averages.  Because these simulations were
for severe ozone episodes, the oxidant concentrations generated
may be somewhat larger than seasonal average values.

     For each city,  calculations were conducted for both the
summer  (July) and winter  (January) seasons.  For each season,
residence time calculations were also conducted for clear-sky and
cloudy conditions.

     The winter simulations used the same summer input files
except for the following:  (1)  the time zone was increased 1 hour
to convert to standard time,  (2) the temperatures were changed to
start at the average winter low and smoothly reach the average
winter high at about 1400 hours,  (3) the date was set to 15
January, and (4) the mixing height maximum was adjusted downward.
Each of the residence time calculations was conducted for clear-
sky conditions and cloudy conditions.  Cloudy conditions take
into account the UV transmission factor, the in-cloud OH
concentration,  the gas-phase oxidant concentrations, and the
cloud liquid water content.

       The residence times are most useful for comparison
purposes rather than as absolute numbers, because of the
necessary assumptions and simplifications which went into the
calculations.  More details regarding the model input files and
parameters used in calculating residence times, such as oxidant
concentrations and rates of reaction, are given in Ligocki et
al., 1991.

5.4.4.4 Benzene Residence Times

     Residence times for benzene were calculated by considering
gas phase chemical reactions with OH and N03,  in-cloud chemical
reaction with OH, and wet and dry deposition.  The results of the
residence time calculation for benzene are presented in Table 5-
4.
                               5-15

-------
EPA-420-R-93-005
April 1993
kBLE 5-4. Atmospheric residence time calculation for benzene. All times are in hours
unless otherwise noted.
Los Angeles St. Louis Atlanta New
July
Clear sky - day 40
Clear sky - night 3000
Clear sky - avg 70
Cloudy - day 80
Cloudy - night 800
Cloudy - avg 120
Monthly 80
Climatological
Average
Jan July
300 30
14000 4000
700 50
(30
d)
600 60
7000 900
1300 90
(56 d)
900 70
(37 d)
Jan July
500 30
18000 3000
1100 50
(46 d)
800 50
8000 300
1800 80
(75 d)
1500 60
(62 d)
Jan July
500 50
14000 4000
1100 90
(45 d)
800 100
7000 1500
1700 150
(71 d) (6 d)
1400 110
(58 d)
York
Jan
900
18000
2200
(92 d)
1600
12000
3600
(150
d)
2900
(120
d)
5-16

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Calculated residence times ranged from 2 days under summer,
clear-sky conditions, to several months under winter, cloudy-sky
conditions.  These values can be compared to estimated benzene
half-lives of 4 days under summer, urban conditions  (CARB, 1984)
and 6 days under summer conditions at 60N latitude (Nielsen et
al.,  1983).

     The main atmospheric destruction pathway for benzene is the
reaction with OH radical.  Even at night, the residence time of
benzene was found to be determined primarily by the reaction with
OH, with a slight contribution from in-cloud destruction.  The
reaction with N03  was found to be  unimportant for benzene.

     As discussed above, estimates of residence times due to dry
deposition should be regarded as highly uncertain.  The residence
times of benzene due to dry deposition are estimated to be on the
order of 20 days for summer, daytime conditions and one year or
more for all other conditions.

     In-cloud chemical destruction and wet deposition will not be
rapid removal processes for benzene.  The residence times due to
in-cloud chemistry ranged from 11 days in the summer to over 2
years in the winter.  The calculated residence times due to wet
removal ranged from 3 years in the winter to 10 years in the
summer.

     Residence times for different cities within a given season
varied by factors of 2-3.  A much larger effect was predicted for
the difference between summer and winter conditions at all sites,
with winter residence times 10-30 times greater than summer
residence times.

     The major uncertainties in these calculations for benzene
are the OH radical concentrations, which vary from day to day by
roughly a factor of two.  The uncertainty in the OH rate constant
is much smaller than this  (about 20 percent).  Although the
uncertainty in the deposition velocity is much larger than a
factor
of two, it does not have a large effect on the overall
uncertainty because dry deposition is only of minor importance as
a removal mechanism.

     These results suggest that, on an urban scale, atmospheric
transformation of benzene would not be expected to be a
significant determinant of ambient benzene concentrations.  Under
all conditions examined, the calculated residence time of benzene
was greater than one day.  Therefore, significant day-to-day
carryover of benzene concentrations would be expected.

5.4.5 Limited Urban Airshed Modeling of Air Toxics

     Much of the information below on the Urban Airshed Model and
the benzene results are excerpted from reports conducted for two
EPA offices (Office of Mobile Sources and Office of Policy,
Planning, and Evaluation) by Systems Applications International


                               5-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993

(SAI)  (Ligocki et al.,  1991, Ligocki and Whitten, 1991, Ligocki
et al. ,  1992).   The modified version of the UAM used in these
reports, with explicit treatment of several toxics, will be
referred to as UAM-Tox.  UAM-Tox in Ligocki et al.  (1991) and
Ligocki and Whitten (1991) which was used to model St. Louis, did
not include explicit chemistry for acetaldehyde and POM.  UAM-Tox
in Ligocki et al.  (1992)  which was used to model the Baltimore-
Washington area and Houston, does treat these toxics explicitly,
however.  Details of inputs and modifications for the UAM are
presented in the above references.  The treatment of each toxic
in UAM-Tox is discussed in the results section for each toxic.

     The Urban Airshed Model (UAM) is a three-dimensional grid
model designed to simulate all important physical and chemical
processes which occur in the atmosphere.  In a grid model, the
region of interest  (domain) is divided into grid cells which are
equally spaced in the horizontal directions, and may have varying
heights depending upon the atmospheric mixed-layer height.
Within each grid cell,  concentrations are assumed to be uniform,
and any emissions which are injected into that cell will
instantaneously spread throughout the cell.  The model
incorporates mathematical representations of the processes of
transport, diffusion,  chemical reaction, and deposition.  Based
upon inputs such as emissions,  winds, mixing heights, initial
concentrations of each species, and concentrations of each
species on the boundaries of the domain, the model computes
concentrations for each species for each grid cell for each hour
of the simulation.

     The UAM has been used primarily for the simulation of ozone
and the development of control strategies for ozone precursors.
It has been evaluated in terms of its ability to predict
concentrations of ozone and a few other species such as NOX and
peroxyacetyl nitrate (PAN).  The UAM has not been evaluated for
the prediction of concentrations of air toxics, and such an
evaluation was beyond the scope of the study summarized here.
Until such an evaluation is conducted, the model results are most
useful for the comparisons they provide of the importance of
atmospheric transformation.

      To illustrate the effects of atmospheric persistence and
transformation on ambient concentrations in an urban area, an
initial urban airshed modeling study of benzene, 1,3-butadiene,
formaldehyde, and acetaldehyde was conducted for a hypothetical
day in the summer of 1990 in the St. Louis area  (Ligocki et al.,
1991;  Ligocki and Whitten, 1991).   A summer day was selected in
order to maximize the potential effects of atmospheric
transformation.  The St.  Louis urban area was selected primarily
because the necessary model inputs were readily available;
however, St. Louis is also of interest because relatively high
benzene concentrations have been measured there  (McAllister et
al.,  1990).   Only one city was modeled due to resource
constraints.  Understanding how the calculated results may vary
in different cities with different emissions and air quality
patterns would help address some of the uncertainty in the


                               5-18

-------
                                                        EPA-420-R-93-005
                                                            April 1993

results.  Subsequently, additional urban airshed modeling was
done for multi-day episodes in the Baltimore-Washington area and
Houston, as part of another study  (Ligocki et al.,  1992).  Both
of these areas are severe ozone nonattainment areas, and will
participate in the federal reformulated gasoline program.
Modeling was conducted for hypothetical episodes in 1995 and
1999, and took into account provisions of the CAA.   Since toxics
provisions of the reformulated gasoline program are year round, a
winter episode was simulated for Baltimore.  The Baltimore and
Houston areas represent opposite ends of the spectrum in terms of
expected air quality benefits of reformulated gasoline.

     The St. Louis episode selected for the initial study was an
historical episode from July 13, 1976.  The meteorological and
air quality inputs for that episode were originally developed for
the EPA as part of the St. Louis Ozone Modeling Project  (Schere
and Sheffler, 1982;  Cole et al.,  1983).  This episode also was
modeled by SAI as part of the EPA Five Cities Study (Morris et
al.,  1989).  Levels of air pollutants have declined significantly
in most cities over the past 15 years.  Although the available
inputs for this simulation were for a 1976 episode, it was judged
to be more useful to conduct the simulation for current
conditions.  Therefore, the emission inventory was updated to a
summer weekday in 1990.  The episode represents a hypothetical
day in 1990 in which the dispersion characteristics correspond to
an actual day in 1976.  Details of other inputs and modifications
for the UAM are presented in detail in Ligocki et al.   (1991) and
Ligocki and Whitten (1991).   The treatment of each toxic in the
UAM is discussed in the results sections for each toxic.

     For modeling in the Baltimore-Washington area, the episode
selected was an historical episode from July 5-7,  1988.  The July
5-7 episode is part of a larger, regional-scale ozone episode
that has been modeled with the Regional Oxidant Model  (Possiel et
al.,  1990).  A number of simulations were conducted for this
episode in the base year of 1988,  1995, and 1999.   Base, federal
reformulated gasoline, California phase 2 reformulated gasoline
and reduced motor vehicle NOX simulations  were  conducted.
Simulations were also done for both summer and winter, and with
motor vehicles removed.  For modeling in the Houston area, the
episode selected was an historical episode from September 3-5,
1987.  Simulations for summer were conducted for this episode in
the base year of 1987, and for base case,  reformulated gasoline,
and no motor vehicle scenarios in 1995.

5.4.5.1 General Results From the UAM Simulations

     Two base-case UAM simulations were conducted for the St.
Louis study.  The simulations used identical input parameters,
but in one of them all chemistry was "turned off"  assuming the
toxic species of concern to be inert.  The second simulation
assumes all "chemistry on",  referred to as the reactive
simulation.  The UAM simulations began at 1 a.m. local daylight
time and ran through 11 p.m.
                               5-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Results are presented as time-series plots of concentration
at a specific grid cell.  The time-series plots are presented in
Appendix D and include predicted total concentrations of each
toxic from both the reactive and inert simulations, and also
include concentrations of the mobile- and stationary-source
components from the reactive simulation.  All values presented in
the time series plots are hourly averages.

     The simulations indicated that summertime concentrations of
primary toxic species derived from mobile sources will be
greatest during morning commute hours, when emissions are
maximized, atmospheric dispersion is poor, and photochemistry is
slow.  The afternoon commute hours are less likely to produce
peaks in mobile-source toxics in the summertime because they
occur while mixing heights are higher and photochemistry is at
its peak.

     The Baltimore-Washington area and Houston simulations also
indicated that concentrations of primary toxics species will be
greatest during morning commute hours.

     Federal reformulated gasoline simulations for 1995 and 1999
in the Baltimore-Washington area indicated a decrease in peak
ozone of 0.2 pphm in 1995 (1.1% of total) and 0.15 pphm in 1999
(0.85% of total).   This decrease corresponded to 20% of the peak
ozone attributed to motor vehicles.  For Houston, federal
reformulated gasoline usage produced smaller ozone benefits, with
a decrease in peak ozone of 0.013 pphm in 1995 (0.04% of total).
This decrease corresponds to only 2% of the peak ozone
attributable to motor vehicles.  In both the Baltimore-Washington
area and Houston,  use of federal reformulated gasoline resulted
in reductions of ambient benzene, acetaldehyde, and POM
concentrations.  For butadiene, there was virtually no effect on
ambient concentrations.  For formaldehyde, there were both
decreases and increases, depending on the simulation.

     The combination of the UAM results with results from the
residence time calculations provides an estimate of the
differences in concentrations which might be expected under
wintertime conditions.  Differences in emission rates and
atmospheric dispersion parameters will also be important factors
in determining wintertime concentrations.  A comparison of summer
and winter simulations in Baltimore indicated that, although
benzene emissions from motor vehicles were lower in winter than
in summer, motor vehicle-related concentrations of benzene were
higher.  Even so,  the motor vehicle fraction of the simulated
concentrations was
                               5-20

-------
                                                        EPA-420-R-93-005
                                                            April 1993

similar in winter, due to an increase in stationary source
concentrations.

5.4.5.2 UAM Results for Benzene

     Benzene was treated explicitly in the UAM-Tox.  Mobile and
stationary emissions of benzene were tagged separately and
carried through simulations separately in the model.  The gas
phase reactions discussed previously were added to the chemistry
subroutines.  Because the focus of the study was on destruction
of the toxic species rather than on the subsequent chemistry of
their reaction products, no products were included in the UAM
modifications for benzene.

St. Louis Simulation

     A time-series plot of predicted benzene concentrations in
St. Louis at the grid cell with the largest mobile-source benzene
concentration is presented in Figure D-l of Appendix D.  At the
time of the mobile-source benzene concentration peak, mobile-
source benzene contributed roughly half of the total benzene
concentration of 0.54 ppb.  As the day progressed, the mobile-
source benzene concentration decreased, while the total benzene
increased to a peak of 0.7 ppb at 11 a.m.  There was no evidence
of a peak in the mobile-source concentration during the afternoon
commute,  probably due to the fact that the mixing height during
the afternoon commute was still roughly 1500 m, compared to 400 m
in the morning.   Thus all emissions would be diluted into a much
larger air volume in the afternoon.

     The low reactivity of benzene is apparent from the
comparison of the "inert" benzene and total benzene curves in
Figure D-l.  There is no difference between the two curves until
mid-morning, and even in the mid-afternoon the difference between
the two curves is less than 0.1 ppb.  Thus, atmospheric
transformation was shown to have only a minor effect on ambient
concentrations during afternoon hours, and virtually no effect
during other times of day.  This illustrates the conclusion drawn
from the residence time calculations, that atmospheric chemical
transformation of benzene in a urban environment is less
important than location of sources and atmospheric dispersion
characteristics in the assessment of benzene concentrations.
Little seasonal effect would be expected for benzene.

     The benzene concentration at the end of the simulation was
0.7 ppb (Figure D-l).  Because benzene is not destroyed
chemically at night (Table 5-2),  in the absence of strong winds
this concentration would be expected to persist into the
following day.  Therefore, the initial concentration of benzene
of 0.1 ppb used for this simulation is likely to be too low.
Future benzene simulations should be conducted for multiple days
in order to
                               5-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993

quantify the importance of day-to-day carryover of benzene
concentrations.

     The effect of initial concentration assumptions for benzene
was examined in a sensitivity study in which the concentration
fields from the end of the base-case simulation were used as the
initial concentrations.  This has the effect of increasing the
initial concentrations of benzene.  The peak concentrations
within the city do not increase substantially from their base-
case values.  The afternoon maximum concentration only increases
by 0.1 ppb.  This result indicates that the meteorology of the
simulated episode was such that concentrations were dominated by
local  emissions.  For other episodes and other locations, more
stagnant conditions might exist, and the importance of the
initial concentrations might be greater.

     When a comparison of simulated concentrations of benzene is
made with ambient measured concentrations,  the simulated
concentrations were much lower than typical measured concentra-
tions.  This discrepancy may be due to uncertainties in the
emission inventory for benzene.  Another possibility is that the
ambient monitors were located in areas not represented well in
the UAM.  The American Petroleum Institute (API)  has stated that
these  differences may also be due to the fact that the UAM is not
able to predict the concentrations and residence times of
reactive air toxics well, and concentrations of the more reactive
compounds show better agreement due to compensating errors in the
model  (API, 1991).  For a full accounting of API's analysis
please consult API, 1991.

Houston and Baltimore-Washington Area Simulations

     Simulations for the summer Baltimore-Washington area episode
resulted in significant decreases in ambient levels of benzene
with use of federal reformulated gasoline,  amounting to as much
as 12  percent of ambient benzene concentrations.   Use of
California reformulated gasoline resulted in slightly larger
decreases in ambient benzene.  Maximum daily average benzene
concentration for the 1988 base scenario was 2.2 ppb.  Motor-
vehicle related benzene accounted for about 58% of total benzene
emissions.  This agrees with the 60% estimate obtained in Section
5.3.4  for motor vehicles.
     The summer Baltimore-Washington area simulations do not
significantly underpredict benzene like the St. Louis simulation.
In fact, simulated benzene concentrations were in good agreement
with the average measured values from the UATMP data base.
Ligocki et al.  attribute this to an effort made to improve the
emission mass fractions in the motor vehicle, area, and point
source speciation profiles.

     In the winter 1988 base scenario, the maximum daily average
benzene concentration was 3.6 ppb, about 40 percent higher than
in summer.  Motor-vehicle related benzene accounted for about 37%
of total benzene emissions.  Simulations for the winter
Baltimore-Washington area episode resulted in significant


                               5-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993

decreases in ambient levels of benzene with use of reformulated
gasoline, on the order of 7 percent.  Motor vehicle benzene
emissions were about 30 percent lower with reformulated gasoline
use, and comprised a smaller fraction of total benzene emissions.
However, the motor vehicle-related concentration of ambient
benzene would be higher in winter, due to less atmospheric
transformation.  Comparison of simulated concentrations with
measured concentrations indicate that the model may underpredict
winter benzene concentrations.

     For the summer 1987 base scenario in Houston, the maximum
daily average benzene concentration was 41.4 ppb.  Motor-vehicle
related benzene accounted for about 21% of total benzene
emissions.  The maximum motor vehicle contribution to ambient
benzene was 25%, based on the 1995 no motor vehicle scenario.
Thus, motor vehicle-related benzene contributed less to overall
ambient benzene in Houston than in Baltimore.   Simulations for
the summer Houston episode predicted little effect on maximum
daily average concentration of benzene with use of reformulated
gasoline at the site of maximum concentration, since in Houston
maximum daily average concentrations are primarily influenced by
point sources due to many large industrial facilities.  However,
for the entire Houston modeling domain, the maximum decrease in
daily average concentration was about 8 percent.  Comparison of
simulated concentrations with measured concentrations suggest the
model accurately predicts benzene concentrations.

5.5 Exposure Estimation

5.5.1  Annual Average Exposure Using HAPEM-MS

     The data presented in Table 5-5 represent the results
determined by the HAPEM-MS modeling that was described previously
in Section 4.1.1.  These numbers have been adjusted to represent
the increase in VMT expected in future years.

     The HAPEM-MS exposure estimates in Table 5-5 represent the
50th percentiles of the population distributions of exposure,
i.e., half the population will be above and half below these
values.  High end exposures can also be estimated by using the
95th percentile of the distributions.  According to the HAPEM-MS
sample output for benzene, the 95th percentile is 1.8 times
higher than the 50th percentile for urban areas, and 1.2 times
high for rural areas.   Applying these factors to the exposure
estimates in Table 5.5, the 95th percentiles for urban areas
range from 1.69 ug/m3  for the  2010 expanded adoption of  the
California standards scenario to 4.81 ug/m3 for the  1990 base
control scenario.  The 95th percentiles for rural areas range
from 0.61 to 1.74 ug/m3,  respectively.
                               5-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Table  5-5.  Annual Average HAPEM-MS Exposure  Projections  for
            Benzene.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Exposure
(ug/m3)
Urban
2.67
1.56
1.37
1.25
1.08
1.10
1.18
1.04
0.94
Rural
1.45
0.84
0.74
0.68
0.58
0.59
0.64
0.56
0.51
Nationwide
2.36
1.40
1.20
1.10
0.98
0.98
1.05
0.93
0.84
                             5-24

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.5.2
Data
Comparison of HAPEM-MS Exposures to Ambient Monitoring
     As stated in section 4.1.2, four national air monitoring
programs/databases contain data on benzene.  The Aerometric
Information Retrieval System  (AIRS), the Toxic Air Monitoring
System (TAMS),  the Urban Air Toxic Monitoring Program  (UATMP),
and
the National Ambient Volatile Organic Compounds Data Base  (NAVOC)
all have a significant amount of data for benzene.  The urban
exposure data for benzene from all four databases is summarized
in Table 5-6.  The AIRS data base contains data on benzene from
1987 to 1989  (EPA, 1989).   The location and number of the sites
varies between years.  Referring back to Table 4-2 in Section
4.1.2 and to Table C-l in Appendix C, 23 sites monitored benzene
in 1987,  36 in 1988, and 13 in 1989.  The cities where monitoring
sites are located are listed below.
     Birmingham, AL
     Oakland, CA
     Fresno, CA
     Bakersfield, CA
     Los Angeles, CA
     Merced, CA
     Riverside, CA
     Sacramento, CA
     San Bernadino, CA
     San Diego, CA
     San Francisco, CA
     Stockton, CA
     Santa Barbara, CA
     San Jose, CA
     Modesto, CA
     Oxnard, CA
     Miami, FL
     Jacksonville, FL
                                 St. Louis, MO
                                 Louisville, KY
                                 Atlanta, GA
                                 Chicago, IL
                                 Baton Rouge, LA
                                 Lowell, MA
                                 Boston, MA
                                 Detroit, MI
                                 Port Huron, MI
                                 Dearborn, MI
                                 Lansing/E. Lansing, MI
                                 New York, NY
                                 Cleveland, OH
                                 Dallas, TX
                                 Houston, TX
                                 Deer Park, TX
                                 Burlington, VT
                                 Tacoma, WA
The average level of benzene  (averaged equally by the number of
sites) was 6.92 ug/m3 (2.13 ppb)  in 1987,  4.13 ug/m3  (1.27  ppb)
in 1988, and 4.16 ug/m3  (1.28 ppb)  in 1989.   Because the number
of sites differs from year to year and the number of samples
taken at the various sites varies greatly, it is misleading to
directly compare these numbers.  However, these numbers do
provide a general idea of the amount of benzene being emitted.

     Looking at the data on an individual site basis, St.  Louis
had the highest level of benzene, 31.0 ug/m3 (9.54 ppb)  in 1987
at an industrial suburban site.  However, only 5 samples were
collected at that site in 1987.  The lowest level of benzene was
found in Boston, 2.50 ug/m3 (0.77 ppb)  in 1987 at an industrial
urban site in the downtown area; however, only 4 samples were
collected.  In 1988, a commercial urban downtown site in
Cleveland had the highest local average of all the sites
monitoring benzene, 11.25 ug/m3 (3.46 ppb)  with 4 samples
collected.  Two commercial suburban sites
                               5-25

-------
                                                         EPA-420-R-93-005
                                                             April 1993
Table 5-6.  Air Monitoring Results  for Benzene.
Program
AIRS
UATMP
TAMS
NAVOC
Years
1989
1988
1987
1989
1990
1987-89
1987
Ambient Data3
ug/m3
4.16
4.13
6.92
6.37
4.78
4.26
7.18
Estimated
Motor Vehicle
Contribution1"
ug/m3
2.50
2 .48
4.15
3.82
2 .87
2.55
4.31
aCaution should be taken in comparing these numbers.   The  methods
of averaging the data are not  consistent  between air monitoring
databases and the sampling methodology  is also inconsistent.

bThe  ambient data are adjusted to represent the  motor  vehicle
contribution to the ambient concentration,  which for benzene is
estimated to be 60%, based on  emissions inventory apportionment.
                               5-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993

in Jacksonville, Florida had the lowest sample levels of benzene,
both 1.82 ug/m3 (0.56 ppb)  with 17  and 5 samples collected.  A
residential suburban site in Houston had the highest average
levels of benzene, 6.34 ug/m3 (1.95 ppb)  in 1989 with 20 samples
collected.  Also in 1989, Lowell, Massachusetts had the lowest
average benzene level at a residential suburban site, 2.28 ug/m3
(0.70 ppb), with 17 samples collected.

     Referring to Table 4.2 and Table C-2, ten sites in the Toxic
Air Monitoring System (TAMS) collected samples of benzene between
1987 and 1989.  Boston,  Chicago, and Houston each had three sites
and Seattle/Tacoma had one.  Averaged together, these sites had  a
benzene level of 4.26 ug/m3 (1.31 ppb).   The highest local
average  level of benzene was at an urban  industrial area  in
Houston, 6.66 ug/m3 (2.05 ppb).   The lowest average local level
of benzene was found at an industrial area in Tacoma, 2.02 ug/m3
(0.62 ppb).  It should be noted that Tacoma was added as a site
in TAMS later than the other sites.  Therefore, data were
collected for benzene starting in 1988 instead of 1987.  As
stated in Section 4.1.2, TAMS is a subset  of AIRS and so it has  a
limited number of sites.

     In the 1989 Urban Air Toxics Monitoring Program  (UATMP),  397
measurements of benzene were taken at 14 sites.  These sites were
in the cities listed below.

          Baton Rouge, LA               Chicago, IL
          Camden, NJ                    Dallas, TX
          Fort Lauderdale, FL           Houston, TX
          Miami, FL                     Pensacola, FL
          St. Louis, MO                 New Sauget, IL
          Washington, B.C.              Wichita, KS

The highest average was 12.9 ug/m3  (3.97 ppb)  at an urban
commercial site in downtown St. Louis, Missouri.  Thirty samples
were collected at this site.  The lowest average was 1.95 ug/m3
(0.60 ppb) at a suburban industrial site in Pensacola, Florida.
Only seven samples were collected at this  site.  The next  lowest
average was 2.99 ug/m3 (0.92 ppb)  at a urban commercial site in
Dallas, Texas.  Twenty-five samples were collected at this site,
providing a statistically more valid average.  The overall
average of the averages for each site was  6.37 ug/m3 (1.96 ppb).
For more detailed information on UATMP, please refer to Table  C-
3.

     In the 1990 Urban Air Toxics Monitoring Program  (UATMP),  349
measurements of benzene were taken at 12 sites.  These sites were
in the cities listed below.
                               5-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993

          Baton Rouge, LA               Chicago,  IL
          Camden, NJ                    Houston,  TX
          Orlando, FL                   Pensacola, FL
          Port Neches, TX               Sauget,  IL
          Toledo, OH                    Washington, B.C.
          Wichita, KS

The highest average was 8.74 ug/m3 (2.69 ppb)  at an suburban
residential site in Houston, Texas.  Twenty-eight samples were
collected at this site.  The lowest average was  2.73 ug/m3 (0.84
ppb) at a suburban residential site in Toledo, Ohio.  Twenty-one
samples were collected at this site.  The overall average of  the
averages for each site was 4.78 ug/m3 (1.47 ppb)  .

     In the National Ambient Volatile Organic Compounds  (NAVOC)
program, 564 measurements of benzene were taken  in 31 cities.
These cities are listed below.

          Bakersfield, CA               Chula Vista, CA
          Citrus Heights, CA            Concord,  CA
          El Cajon, CA                  El Monte, CA
          Fremont, CA                   Fresno,  CA
          Long Beach, CA                Los Angeles, CA
          Merced, CA                    Modesto,  CA
          Richmond, CA                  Rubidoux, CA
          San Francisco, CA             San Jose, CA
          Santa Barbara, CA             Simi Valley, CA
          Stockton, CA                  Upland,  CA
          Philadelphia, PA              San Leandro, CA
          Livermore, CA                 San Rafael, CA
          Napa, CA                      Vallejo,  CA
          Santa Rosa, CA                Redwood  City, CA
          Mountain View, CA             Oakland,  CA
          Baton Rouge, LA

The highest measurement was 11.7 ug/m3 (3.60 ppb), which was
found at an urban site in San Francisco.  However, this was only
one sample instead of an average.  The highest average was 11.47
ug/m3  (3.53  ppb),  which was  found at an urban site in Fresno and
consisted of 11 samples.  The lowest measurement of benzene was
2.60 ug/m3  (0.80  ppb),  which was  found at an urban site in
Oakland.  Once again, this was only one measurement instead of an
average of multiple measurements.  The lowest average was 3.51
ug/m3  (1.08  ppb),  which was  found at an urban site in Livermore,
California,  and consisted of 8 samples.  The overall average  of
the averages from the 31 cities was 7.18 ug/m3 (2.21 ppb).   For
more detailed data, please refer to Table C-4.

     The premise of the HAPEM-MS model is that the dispersion and
atmospheric chemistry of benzene is similar to CO.  The average
atmospheric lifetime of CO ranges from one to four months  (EPA,
1990b).   Since both benzene and CO have long atmospheric
lifetimes,  the HAPEM-MS model should be a reliable indicator  of
benzene exposure from motor vehicles.  To test the reasonableness
of the HAPEM-MS modeling results, the HAPEM-MS results for 1990


                               5-28

-------
                                                        EPA-420-R-93-005
                                                            April 1993

are compared to ambient monitoring results for recent years.
Before comparing the HAPEM-MS results to the ambient data, the
ambient monitoring data should be adjusted in two ways.  First,
the ambient monitoring data should be adjusted to represent the
amount that is attributed to motor vehicles.  The data derived
from emission inventories estimate that 60% of the ambient
benzene can be apportioned to motor vehicles.  The numbers in the
second column of Table 5.6 are 60% of the ambient levels and thus
represent estimated ambient motor vehicle levels.

     Second, the estimated ambient motor vehicle level should be
adjusted to account for integrated exposure, i.e., time spent
indoors and in various microenvironments.   Pezda et al. (1991)
refer to data collected in California (Robinson et al., 1989),
which indicate that people spend 5.9% of their time outdoors,
61.9% indoors at home, 24.6% at work, and 7.6% during some form
of transportation  (car, bus, train, etc.).  Using these activity
patterns, it is next necessary to estimate how much of the
ambient mobile source level people in these microenvironments are
exposed to.  HAPEM-MS provides the following microenvironment
factors:  indoors residence - 0.495; indoors other - 0.619;
outdoors - 0.758; and inside motor vehicle - 1.554.  These
microenvironment factors are based on correlations between CO
measured by personal exposure monitors and CO measured by fixed
site monitors located within 10 km.

     Combining the activity patterns and microenvironment
factors, an adjustment factor to the ambient motor vehicle level
to account for integrated motor vehicle exposure is calculated as
shown below:

[(0.059) (0.758)+ (0.619) (0 . 495) + (0.246) (0 . 619) + (0.076) (1.554)]
     = 0.622
     The ambient motor vehicle level ranges from 2.48 to 4.31
ug/m3.   Applying the factor of 0.622 to this range,  the
integrated motor vehicle exposure is estimated to range from 1.54
to 2.68 ug/m3.   Since the unit risk estimate for benzene is an
upper bound estimate, and the HAPEM-MS 1990 base control number
matches the upper end of the range, the HAPEM-MS 1990 base
control level of 2.67 ug/m3 will  be used to estimate cancer
deaths.  See Section 5.3.3 for the discussion of total integrated
benzene exposure in the TEAM study  (EPA, 1987; Wallace, 1989).

5.5.3  Short-Term Microenvironment Exposures

     The primary emphasis for benzene and other exposures in
microenvironments are relatively localized scenarios which are
highly impacted by motor vehicle emissions.  These
microenvironments include in-vehicle exposure, parking garage
exposure,  and exposure to vehicle refueling emissions.  The
information contained in Table 5-7 is excerpted from four studies
that have measured microenvironment exposures to benzene.  These
four studies are the EPA's Total Exposure Assessment Methodology


                               5-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993

(TEAM) Study (EPA, 1987b),  Commuter's Exposure to Volatile
Organic Compounds, Ozone, Carbon Monoxide, and Nitrogen Dioxide
(Chan et al.,  1989),  In-Vehicle Air Toxics Characterization Study
in the South Coast Air Basin  (Shikiya et al.,  1989), and Air
Toxics Microenvironment Exposure and Monitoring Study  (Wilson et
al.,  1991).  See the information in Section 4.2 for more details
about the methodology.

     The TEAM Study (EPA, 1987b; Wallace, 1989) was planned in
1979 and completed in 1985.  The goals of this study were:  1) to
develop methods to measure individual total exposure (exposure
through air,  food and water)  and resulting body burden to toxic
and carcinogenic chemicals, and 2)  to apply these methods with a
probability-based sampling framework to estimate the exposures
and body burdens of urban populations in several U.S. cities.
This was achieved through the use of small personal samplers, a
specially designed spirometer (used to measure the chemicals in
exhaled breath),  and a survey designed to insure the inclusion of
potentially highly exposed groups.

     The study, Commuter's Exposure to Volatile Organic
Compounds,  Ozone, Carbon Monoxide,  and Nitrogen Dioxide  (Chan et
al.,  1989), focused on the driver's exposure to VOC's in the
Raleigh, NC area.  The primary objective of this study was to
measure driver's exposure to all possible VOC and some combustion
gases during one rush-hour driving period (18 sampling days, two
trips per day).  Factors that could influence driver's exposure,
such as different roadways, car models, vehicle ventilation modes
and times of driving were also tested.  Car exterior samples were
also collected from the exterior of the moving vehicles by
setting sampling probes on the middle of the car roof.   Another
objective was to find the relationships between fixed-site
measurements and drivers' exposure (one fixed-site monitor
matched per trip).  Lastly, the pedestrian's exposure to VOC in
urban walking was evaluated with six walking samples.

     The study by the South Coast Air Quality Management District
(SCAQMD),  In-Vehicle Air Toxics Characterization Study in the
South Coast Air Basin (Shikiya et al., 1989),  was conducted to
refine the assessment of health risk due to exposure to toxic air
pollutants.  This study examines the relative contribution of in-
vehicle exposure to airborne toxics to an individual's total
exposure by measuring concentrations within vehicle interiors
during home-to-work commutes.  Other objectives of this study
were to develop statistical and concentration measurement methods
for a vehicular survey and to identify measures which might
reduce commuters' exposure to toxic air pollutants.  Vehicles of
home-to-work
                               5-30

-------
                                                                                  EPA-420-R-93-005
                                                                                      April 1993
Table 5-7.  Microenvironment  Exposure to Benzene (jig/m3) .
Scenarios
TEAM Study
(EPA, 1987b)
Raleigh, NC
Study0
(Chan et al . ,
1989)
SCAQMD Studyd
(Shikiya et
al., 1989)
SCAQMD Studye
(Wilson et
al., 1991)
In-Vehicle
Mean

10.9
42.5

Max.
40-60a
42 .8
267.1

Service Station
Mean




Max.
3000b


288
Parking Garage
Mean




Max.



67.1
Office Building
Mean




Max.



16.0
""Maximum benzene concentrations could not be  reliably determined because exposures were
 averaged over a 12 hour period;  however,  maximum concentrations of 3 to 4 times  normal
 exposures were calculated.
bThis concentration was estimated, rather than measured directly.
CA one-hour measurement was taken for each experimental trip.
dThe estimated sampling time period was 1.5 hours/round-trip.
eThe measurements from this study are five minutes  levels.
                                            5-31

-------
                                                        EPA-420-R-93-005
                                                            April 1993

commuters from a non-industrial park were sampled for in-vehicle
concentrations of 14 toxic air pollutants, carbon monoxide, and
lead.

     The second study by SCAQMD, Air Toxics Microenvironment
Exposure and Monitoring Study  (Wilson et al.,  1991), attempted to
monitor exposures to motor vehicle emissions in microenvironments
other than in-vehicle.  The study randomly sampled 100 self-
service filling stations and took samples at 10 parking garages
and 10 offices nears the garages in Los Angeles, Orange,
Riverside,  and San Bernadino Counties of Southern California.
The study took five-minute samples of 13 motor vehicle air
pollutants in each microenvironment and in the ambient
environment.

     Maximum microenvironment exposure levels of benzene related
to motor vehicles were determined in these studies to range from
40 ug/m3  from in-vehicle exposure to 288  ug/m3 from exposure
during refueling.  This compares to ambient levels of 4.13 to
7.18 ug/m3  determined through air monitoring  studies and
presented in Table 5-6.  Since for the majority of the population
these are short-term acute exposures to benzene, the concern
would be with non-cancer effects.  Health information for
non-cancer effects is limited and no RfC has been developed by
EPA.  Several studies recently conducted in rats and mice have
observed depressed cell proliferation in specific bone marrow
cells at short-term exposures of 3.2xl04  ug/m3 benzene.  Please
see Section 5.8 for more information on non-cancer effects.

     Due to more stringent fuel and vehicle regulations, short-
term exposure to benzene in these microenvironments is expected
to decrease in future years.

5.6  Carcinogenicity of Benzene and Unit Risk Estimates

5.6.1  Most Recent EPA Assessment

     The information presented in Section 5.6.1 has been
abstracted from EPA's Interim Quantitative Cancer Unit Risk
Estimates Due to Inhalation of Benzene (EPA,  1985), EPA's
Integrated Risk Information System  (EPA,  1992c), and the Motor
Vehicle Air Toxics Health Information (Clement,  1991).  The
carcinogenicity risk assessment for benzene was last updated on
IRIS in January 1992, and contains data published through 1987.
However,  it is essentially unchanged from the risk assessment
published in 1985.  EPA's Office of Research and Development has
just recently started the process to review the benzene risk
assessment.  Data published since the 1985 risk assessment for
benzene is summarized in Section 5.6.3.
                               5-32

-------
                                                           EPA-420-R-93-005
                                                                April 1993

5.6.1.1    Description of Available Carcinogenicity Data

Genotoxicity

      Benzene has been  found to induce chromosomal aberrations
(i.e., abnormalities in the  chromosomes)  in bone marrow cells
from rabbits (Kissling  and Speck 1973), mice (Meyne and Legator
1980), and rats  (Anderson and Richardson  1979).   Several
investigators have reported  positive results for benzene  in  mouse
micronucleus assays  (Meyne and Legator 1980).   The micronucleus
assay is  a laboratory method in which blood cells are examined to
determine if broken chromosomes have formed small extra nuclei in
the cytoplasm of the cell.   Benzene was not mutagenic  (i.e., did
not cause changes in the genetic material)  in several bacterial
and yeast systems (e.g., Crebelli et al.  1986;  De Flora et al.
1984; Glatt et al.  1989; Lee et al.  1988;  Tanooka 1977),  in  the
sex-linked recessive lethal  mutation assay with Drosophila
melanogaster (fruit fly)  (Kale and Baum 1983)  or in the mouse
lymphoma  cell forward mutation assay  (Oberly et al.  1984).

Animal Studies

     Exposure of rodents to  benzene either by gavage  (compound is
administered directly into the stomach by means of a stomach tube
inserted  down the throat) or inhalation has resulted in tumor
formation.   Maltoni and Scarnato (1979) and Maltoni et al.  (1983)
administered 0, 50,  250, and 500 mg/kg benzene by gavage  to
Sprague-Dawley1 rats  (30-40/sex/dose)  for life.   Rats
demonstrated dose-related increases in the incidence of mammary
tumors  (females), Zymbal gland carcinomas (a malignant tumor of a
gland that surrounds the ear canal in rats that secretes  an  oily
substance),  oral cavity carcinomas,  and leukemias/lymphomas  in
both sexes.   Leukemia is an  acute or chronic disease that is
characterized by unrestrained growth of leukocytes (white blood
cells) and their precursors  in the tissues2.  Lymphoma is a
general
      The names and/or numbers preceding rats or mice throughout this document
denote specific laboratory strains.

      Leukemia may be divided into  granulocytic leukemias (which include
myelocytic, monocytic, and erythroblastic cell types) and lymphocytic leukemias.
Both granulocytic and lymphocytic leukemia may, in turn, be separated into acute
and chronic forms.  In acute myeloid leukemia  (AML) there is diminished
production of normal erythrocytes, granulocytes, and platelets which leads to
death by anemia, infection, or hemorrhage. These events can be rapid.  In
chronic myeloid leukemia  (CML) the leukemic cells retain the ability to
differentiate (i.e., be responsive to stimulatory factors) and perform function;
later there is a loss of the ability to respond.

                                5-33

-------
                                                        EPA-420-R-93-005
                                                             April 1993

term for inappropriate growth  of  new tissue or neoplastic3  growth
in the lymphatic system.

     In a National Toxicology  Program (NTP 1986)  study,  benzene
was administered by gavage  to  Fischer-344/N rats  (50/sex/dose)  at
doses of 0, 50, 100, or 200 mg/kg and to B6C3F-L mice
(50/sex/dose) at doses of 0, 25,  50,  or 100 mg/kg .   The animals
were treated 5 times/week for  103 weeks.   There were significant
increases in the incidence  of  various neoplastic  growths in both
sexes of both rats and mice.   Both species had an increased
incidence of carcinomas of  the Zymbal gland.   Male and female
rats had oral cavity tumors, and  males showed an  increased
incidence of skin tumors.   Males  were observed to have tumors of
the Harderian  (a gland located within the eye of  the rat)  and
preputial gland  (a small gland located near the head of the penis
that secretes an odiferous  discharge important to mating),  and
females had tumors of the mammary gland and ovaries.   In general,
the increased incidence was dose-related.

     Inhalation exposure of male  C57B1 mice to 300 ppm benzene on
a workday schedule  (6 hours/day,  5 days/week)  for 488 days
resulted in slight increases in the incidence of  hematopoietic
neoplasms  (Snyder et al. 1989).   However,  there was no increase
in tumor incidence in male  AKR mice exposed to 100 ppm or male
CD-I mice exposed to 100 or 300 ppm benzene.   Likewise,  male
Sprague-Dawley rats exposed by inhalation to 300  ppm benzene were
not observed to have an increased incidence of neoplasia.

     Maltoni et al.  (1983)  treated male and female Sprague-Dawley
rats in the following manner.   Starting at 13 weeks of age, rats
were exposed to 200 ppm benzene by inhalation 4 hours/day,  5
days/week for 7 weeks.  Animals were then exposed to the same
concentration for 7 hours/day,  5  days/week for 12 weeks,  and
finally 300 ppm 7 hours/day, 5 days/week for 85 weeks.  A time-
weighted average  (TWA) of 241  ppm for an 8 hours/day, 5 days/week
exposure was calculated.  In this study,  a statistically
significant increase in the incidence of liver tumors (hepatomas)
and carcinomas of the Zymbal gland was found.

     Goldstein et al.  (1980) conducted studies in male Sprague-
Dawley rats exposed to 0 ppm  (67/group),  75 ppm (40/group), or
225 ppm (45/group) benzene  by  inhalation for an unreported period
of time.  In this study, one animal contracted leukemia in the 75
ppm concentration group.  In addition,  AKR,  C57BL,  and CD-I mice
were exposed to 0 ppm  (210/group),  75 ppm (50/group), or 225 ppm
(160/group) benzene again for  an  unreported period of time.
After this treatment, two animals in the high-concentration
exposure group developed leukemia.

Human data
      Neoplastic growth is characterized by new and abnormal formation of
tissue,  usually as a tumor.  By custom,  this refers to the pathological process
in tumor formation, i.e., cancer.

                               5-34

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Rinsky et al.  (1981) followed 748 Pliofilm8 (a film made of
rubber hydrochloride) workers  (all white males) exposed to
benzene at levels that averaged from 10-100 ppm over an 8-hour
workday (8-hour time-weighted average, TWA) for at least 24 years
(17,020 person-years, an expression of cumulative dose).  Seven
deaths resulted from leukemia in this group.  This increased
incidence was statistically significant with a standard mortality
ratio  (SMR)  of 560.  The standard mortality ratio is the number
of deaths, either total or cause-specific, in a given group
expressed as a percentage of the number of deaths that would have
been expected in that group if they were the same as the age-and-
sex-specific rates in the general population.  For the 5 leukemia
deaths that occurred among workers with more than 5 years of
exposure,  the SMR was 2,100.  Exposures were described as less
than the recommended standards (25 ppm) for the time period of
1941-1969.  A computer tape containing follow-up information for
the Rinsky population through the year 1978 was used in addition
to the original Rinsky et al.   (1981)  data to develop unit risk
estimates.  No effort was made to correct for smoking or other
potential confounding exposures.

     Ott et al. (1978) studied 594 white male workers
occupationally exposed to benzene in a chemical manufacturing
facility at concentrations ranging from 2 to 25 ppm  (8-hour TWA).
This group was followed for at least 23 years in a retrospective
cohort mortality study.  A retrospective cohort is a group of
people, defined by arbitrary criteria as alike in some way, some
of whom are known to have experienced particular exposures as
well as particular health effects at some time prior to the start
of the investigation.  Although three leukemia deaths were
observed,  the increase was not statistically significant when
compared to an unexposed population.

     Wong et al.  (1983) studied 4,062 male  (both white and
nonwhite)  chemical workers who had been exposed to benzene for at
least 6 months between 1946-1975.   The study population was drawn
from seven chemical plants, and jobs were categorized with
respect to peak exposure.  Subjects with at least 3 days/week
exposure  (3,036 individuals) were further categorized on the
basis of an 8-hour TWA and were compared to controls who held
jobs at the same plants for at least 6 months without exposure to
benzene.  The range of exposures experienced by these workers was
<1 ppm to >50 ppm  (8-hour TWAs).   Statistically significant dose-
dependent increases in the incidence of leukemia, lymphatic, and
hematopoietic cancer  (i.e., cancers of the blood forming organs)
were found when the data were analyzed in terms of cumulative
exposure  (i.e., exposure level multiplied by duration of
exposure).  The incidence of leukemia was responsible for a
majority of the increase, due largely to the fact that the
incidence of mortality due to neoplasia in unexposed subjects was
lower than expected.

     Aksoy et al.   (1974)  reported effects of benzene exposure
among 28,500 Turkish workers employed in the shoe industry who
used benzene-containing adhesives.  Mean duration of employment


                               5-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993

was 9.7 years (1-15 year range)  and mean age was 34.2 years.
Peak exposure was reported to be 210-650 ppm.  Twenty-six cases
of leukemia and a total of 34 leukemias or preleukemias  (blood
conditions that are thought to precede the onset of leukemia)
were observed, corresponding to an incidence of 13/100,000  (by
comparison to 6/100,000 for the general population).  This
represents a statistically significant increase in the incidence
of leukemia among the shoe workers.  The possibility of
concomitant exposure to other agents was not discussed.  A
follow-up paper (Aksoy 1980)  reported eight additional cases of
leukemia as well as evidence suggestive of increases in other
malignancies in exposed workers.

     The leukemogenic (i.e.,  the ability to induce leukemia)
effects of benzene exposure were studied in 748 white males
employed from 1940-1949 in the manufacturing of rubber products
in a retrospective cohort mortality study  (Infante et al.
1977a,b).   Statistics were obtained through 1975.  A
statistically significant increase in the incidence of leukemia
was found by comparison to the general U.S. population.  The
worker exposures to benzene were between 100 ppm and 10 ppm
during the years 1941-1945.  There was no evidence of solvent
exposure other than benzene.

     There are many other epidemiologic and case studies that
report increased incidence or a causal relationship between
leukemia and benzene exposure.  In addition, numerous
investigators have found significant increases in chromosomal
aberrations of bone marrow cells and peripheral lymphocytes from
workers with exposure to benzene (IARC 1982).

5.6.1.2  Weight-of-Evidence Judgment of Data and EPA
Classification

     The weight-of-evidence indicates that benzene is a Group A,
known human carcinogen.  This is based on sufficient human
epidemiologic evidence (Rinsky et al.  1981; Ott et al. 1978; Wong
et al.  1983) demonstrating an increased incidence of
nonlymphocytic leukemia from occupational inhalation exposure, in
addition to supporting animal evidence  (Goldstein 1980; NTP 1986;
Maltoni et al.,  1983) in which there was an increased incidence
of neoplasia in rats and mice exposed by inhalation and gavage.
                               5-36

-------
                                                         EPA-420-R-93-005
                                                              April 1993
5.6.1.3   Data  Sets  Used For Unit Risk Estimate

     The data sets used to estimate the unit risk4 for benzene
were obtained from a reorganization of the 1981 Rinsky  et  al.
data  (followup  from  1940 to 1978),  Wong et al.  (1983),  and Ott  et
al.   (1978).  These three studies are summarized in Table 5-8.

     The Rinsky data used were from an updated tape  that reports
one more case of  leukemia than was published in 1981.   It  should
be noted that a recently published paper  (Rinsky et  al. 1987)
reported 2 additional cases of leukemia from the study  population
but was not used  for the current risk estimate.  Updates of other
cohorts used in the  current EPA assessment are discussed in
Section 5.6.3.  Generally,  the updates report increased cohort
size, improved  exposure analyses, and/or alternative methods  to
analyze the cancer incidence data.

     Although the data from Aksoy et al.   (1974) and  Aksoy  (1978)
indicated an association of leukemia with benzene, it was  decided
by EPA that the exposure information was so imprecise that it was
not suitable for  quantitative estimation.

     Selection  of the models used in the EPA estimate of unit
risk was "a matter of judgement."  The estimates were based on
the most extensive and inclusive body of data available that  is
of acceptable quality,  so all three epidemiologic studies  were
used.  The choice of the studies in which the species  (human) and
route of exposure (inhalation)  most closely corresponded to the
environmentally exposed population were given the most  weight.
Animal studies  were  merely used for confirmation purposes.

5.6.1.4 Dose-Response Model Used

     The unit lifetime risk estimate was determined  by  using  the
relative risk model  and the absolute risk model with three
different measures of dose (total of six models) to  develop 21
maximum likelihood estimates (MLEs).   These 21 MLEs  were then
used to calculate a  geometric mean to determine the  unit risk
estimate.
     4
      Under an assumption of low-dose linearity, the unit cancer risk is the
excess lifetime risk due to continuous constant lifetime exposure to one unit of
carcinogen concentration.  Typical exposure units include ppm or ppb in food or
water, mg/kg/day by ingestion,  or ppm or ug/m3  in air (EPA 1986b).

                               5-37

-------
                                                                                    EPA-420-R-93-005
                                                                                         April 1993
Table 5-8.   Summary of Data Sets Used to Calculate the Unit  Risk Factor For Benzene3
Study
Rinsky et
al. (1981)
Ott et al.
(1978)
Wong et al .
(1983)
Population
Studied/Years of
Follow-up
748 Pliofilm
workers (all
white males) /
38 years
594 chemical
workers (white
and nonwhite) /
at least 23
years
44,062 male
chemical workers
(white and
nonwhite) /
29 years
Duration of
Exposure
At least 24
years
Not
specified
At least 6
months
Exposure
Level (s)
10-100 ppm
(8 -hour
TWA)
2-25 ppm
(8 -hour
TWA)
50
ppm
Effect(s)
Statistically
significant increased
incidence of leukemia
Increased incidence of
leukemia (not
statistically
significant)
Statistically
significant increased
incidence of leukemia,
lymphatic, and
hematopoietic cancer
"Various subsets of  these  studies were used  to  develop the 21 unit  risk factors.
                                             5-38

-------
                                                           EPA-420-R-93-005
                                                                April 1993
5.6.1.5  Unit Risk Estimates5
     As  stated above, six  models and various  combinations  of
epidemiological data sets  were used to derive a total of 21
MLEs,with  their 95% statistical confidence  bounds.  Because  the
EPA had  no basis for choosing one model over  another, the
geometric  mean of these  21 unit risk estimates was taken to
obtain a pooled model average estimate, resulting in a maximum
likelihood estimate  (MLE)  unit risk of 2.7xlO"2 for leukemia  due
to a lifetime exposure of  1 ppm benzene in  the air (8.3xlO~6
[ug/m3]-1) .

     The actual 95% upper  bound (UCL) was calculated for each MLE
derived  with each of the 21 models.  These  data are presented in
Tables E-l through E-6 in  Appendix E, which were reproduced  from
(EPA, 1985).   A geometric  mean of the UCLs  was not calculated.

5.6.2  Other Views and Risk Estimates

     This  section presents alternative views  and/or risk
assessments for benzene.   These alternative risk assessments  are
summarized in Table 5-9.

International Agency for Research on Cancer (IARC)

     IARC  has classified benzene as a Group 1 carcinogen.  A
Group 1  carcinogen is defined as an agent that is carcinogenic to
humans.  This classification is based on sufficient evidence  for
carcinogenicity in humans  (IARC, 1987).  IARC (1987)  based this
conclusion on the fact that numerous case reports and follow-up
studies  have suggested a relationship between exposure to  benzene
and the  occurrence of various types of leukemia.

     In  addition, IARC  (1987)  considers the evidence for
carcinogenicity to animals to be sufficient.   No unit risk was
determined by IARC for benzene.

California Air Resources Board  (CARE)

     The California Department of Health Services (DHS, 1984)
(which provides technical  support to CARB)  has also determined
that there is sufficient evidence to consider benzene a human
      For any dose-response model,  one typically obtains risk  estimates for
various dose levels.  It is possible to obtain maximum-likelihood estimates
(MLEs) and upper  confidence limits (UCLs)  for those risks.  The MLEs represent
the best description of the observed data  that can be obtained  for any given
dose-response model.  However, because there are many sources of error that
affect the observation of responses  (including,  but not  limited to,  random error)
it is often desirable to determine upper bounds on the risks.  The UCLs are
statistical estimates of those upper bounds; they determine the highest levels of
risks associated  with specific dose  levels that are consistent  with the observed
responses, the dose-response model,  and the level of certainty  required by the
user.

                                5-39

-------
Table 5-9.   Comparison of  Benzene Inhalation  Unit  Risk  Estimates.
                                                                                                  EPA-420-R-93-005
                                                                                                       April 1993
Source




EPA (1985)

IARC (1987)

CARB (DHS 1984)






CARB (CAPCOA
1991)
Clement (1988)


Tumor Type




Nonlymphocytic leukemia
in occupational studies
Leukemia in
occupational studies
Leukemia in
occupational studies
(for lower bound on
risk) and preputial
gland tumors in mice
and rats (for upper
bound on risk)
	

Leukemia in
occupational studies

Classification




Group Aa

Group ld

Human
Carcinogen





	

h


Cancer
Unit Risk
Estimate
(ug/m3)-1

6.7xlO"6 -
l.SxlQ-4 b
e

5.2xlO"5 f






2.9xlO"5
"best value"
i


Cancer
Unit Risk
Estimate
(ug/m3)-1
MLE
8.3xlO"6 c

e

8.3xlO"6 g








4.3xlO"8
-
l.lxlO"6 D
aGroup A = Human Carcinogen
bRange of 21 95% UCLs.  Geometric mean  not calculated
cGeometric mean of 21 MLEs
dGroup 1 = Human Carcinogen
eIARC did not conduct a quantitative  risk assessment
Lower bound of cancer risk
9Upper bound of cancer risk
hClement did not classify benzene
'"Clement did not calculate UCLs
jRange of MLEs calculated using different assumptions and data
                                                    5-40

-------
                                                        EPA-420-R-93-005
                                                             April 1993

carcinogen.  CARB performed a risk assessment of benzene  that  was
very similar to EPA's risk assessment.  DHS, like  EPA,  assumed
that there is no threshold for benzene-induced  carcinogenicity,
and that the multistage theory most appropriately  describes  the
phenomenon of benzene-induced carcinogenesis.

     The CARB potency factor for benzene  is actually  a  range of
potency factors.  For the lower end of the range,  DHS calculated
a MLE potency estimate for benzene, like  EPA had originally  done,
based on the reevaluation of three epidemiological studies using
the linearized multistage model and cumulative  exposure averaged
over the individual's lifetime  (see Appendix F  for a  lay
description of this model).  However, whereas EPA  calculated a
geometric mean of each of the three study's estimated slopes to
obtain one slope factor, DHS used combined input data (i.e.,
background rate of leukemia, relative risk, and lifetime  average
exposure level) from the three studies to calculate one slope
factor.  As a result, the MLE slope factor calculated by  EPA
yielded an increase in risk of 2.7xlO"2 due to a continuous
lifetime exposure of 1 ppm benzene in the air  (8.3xlO~6  [ug/m3] -1)
whereas the DHS MLE slope factor corresponds to a  lifetime risk
of 4.8xlO"2  (ppm)"1 (1.5xlO"5  [ug/m3]"1).  CARB chose  to  use  EPA's
MLE value of 2.7xlO"7 ppm"1 as  the  lower bound for  the benzene
cancer potency factor range.

     For the upper end of the range, a 95% UCL  was calculated
based on the most sensitive site in rats  and mice,  the  preputial
gland,  using the data from the NTP  (1983) study.   This  data  set
yielded a risk of l.VxlO"1 per ppm  in air  (5.2xlO~5   [ug/m3]"1),
which is 3.5-7 times as great as the risk estimated from  human
mortality data.  Thus, the CARB potency estimate for  benzene
ranges from 2.7xlO"2 to 1.7X10"1  ppm"1 (8.3xlO~6 to 5.2xlO"5
[ug/m3] ) .

     CARB has also used what is termed as a  "best  value"  for the
benzene estimate provided by the California Department  of Health
Services in conjunction with the California Air Pollution Control
Officers Association  (CAPCOA, 1991).  The CARB  "best  value"  for
benzene is 2.9xlO"5  (ug/m3)"1  (9.4xlO~2 ppm"1) which  falls  within a
range of unit risk factors,  0.75xlO"5 to 5.3xlO"5 (ug/m3)"1,
(2.4xlO~2 to 1.7X10"1 ppm"1)  recommended by DHS for health  effects
assessments which were prepared for the State's Toxic Air
Contamination Program.

Motor Vehicle Manufacturer's Association  (MVMA)

     MVMA contracted with Environ  (Environ 1987) to evaluate the
risk assessment issues in EPA's technical report "Air Toxics
Emissions from Motor Vehicles"  (Carey 1987).  It is important  to
note that the Environ document is not actually  a risk assessment
of benzene; rather,  it is a critique of EPA's risk assessment  of
motor vehicle air toxics.

     In a discussion of the possible impact of  alternative
approaches, Environ (1987) mentioned a risk assessment  it
performed for the Western Oil and Gas Association,  in which  it


                               5-41

-------
                                                        EPA-420-R-93-005
                                                            April 1993

calculated an estimate of benzene risk that was approximately
one-fourth of that developed by EPA.  This estimate was based
upon the Rinsky et al.  (1987) cohort, and assumptions about their
exposure as developed by Crump and Allen  (1984).  No unit risk
was determined in this study by Environ for the MVMA.

American Petroleum Institute

      The American Petroleum Institute (API) states that the Yin
et al.   (1987) study should not be used in evaluating human health
risks associated with benzene exposure because of the technical
limitations of the study.  Technical problems cited by API
include exposure to mixtures of chemicals, retrospective benzene
exposure measurements,  lack of information on other exposures,
and no confounding factors (such as smoking) were taken into
consideration.  API also states that there are problems in the
comparison of the chosen exposed and non-exposed cohorts.

     API also does not support the conclusion of the most recent
data that benzene could be a developmental and reproductive
toxicant.  API states that its preliminary analysis of these
studies indicates that there is a lack of quality data necessary
to support this conclusion.  API specifically cites the study by
Savitz et al. (1989)  which reports an elevated risk of still
births when fathers were exposed to benzene and specifically
finds fault with the exposure methodology.  In this study though,
Savitz et al. (1989)  do discuss the limitations and suggest
further evaluation while assuming no definitive relationships.

     API has also taken the results of the Lange et al.  (1973b)
immunological effects study and performed its own analysis.
Lange et al.  (1973b)  indicated a relationship between benzene and
an allergic blood disease.  API claims that the data in Lange et
al.  (1973b) indicate non-significant differences between exposed
and non-exposed groups.

Clement Associates, Inc.

     Under the sponsorship of the American Petroleum Institute,
the Chemical Manufacturers Association, and the Western Oil and
Gas Association, Clement Associates, Inc. (Clement 1988)
performed a quantitative re-evaluation of the human leukemia risk
associated with inhalation exposure to benzene.  Clement's risk
assessment differed from EPA's in the following ways:

     1)   The analysis of risk is based solely on the Rinsky
          epidemiology cohort, rather than Rinsky, Ott, and Wong.
          The justification for this was that the Rinsky study
          was the only one that was not confounded by exposure to
          other chemicals and had an observed statistically
          significant dose relationship.   Furthermore, the
          Clement reanalysis made use of three more years of
          follow-up data on this cohort  (Rinsky et al. 1987), and
          corrected several job code errors that existed in the
          data used in 1985.
                               5-42

-------
                                                             EPA-420-R-93-005
                                                                 April 1993

           The absolute risk  model, rather than a  combination  of
           the absolute and relative risk models,  was used in  the
           Clement  reanalysis.

           The Clement reanalysis selected the weighted cumulative
           form as  the most realistic biological latency
           distribution, rather than both a cumulative dose and a
           weighted cumulative  dose weighting function.

           The Clement reanalysis uses  the exact time-dependent
           exposure for each  of the 1,740 individuals in the
           cohort as the critical information necessary to
           calculate the transition rate  per unit  of  exposure
           parameter (i.e., the probability that one  unit of
           exposure will result in the  biological  event that leads
           to leukemia).  EPA calculated  transition  rate by
           relying  on an aggregation of data that  grouped all
           person-years observed in the epidemiology  studies into
           six exposure intervals rather  than treating each
           exposed  individual separately.

           The Clement reanalysis adopts  a different  definition of
           the types of diseases (i.e., acute leukemia and
           myelodysplastic syndrome/chronic "aplastic" anemia)
           associated with benzene exposure and adjusts background
           rates accordingly.

           The Clement reanalysis adds  a  quadratic model based on
           the hypothesis that  two hits (i.e., two molecules of a
           benzene  metabolite)  are required to induce the
           biological event that leads  to leukemia and calculates
           a linear quadratic model6 as an  upper bound on the two-
           hit model. The results of these differences in approach
           are summarized in  Table E-7  in Appendix E.
      Dose-response functions are often referred to based on the mathematical
equations that define them.  Many of the equations that are used are polynomials,
which can be expressed in the general form

           a0 + ax*d +  a2*d2 + ...+ ak*dk.

Each of the groups of symbols between the plus signs is referred to as a term of
the polynomial.  Polynomials are referred to by their degree, which is the
highest power to which dose, d, is raised.  The equation shown above has degree
k.  Multistage models that have been applied to cancer risk assessment are based
on such polynomial expressions.

     A linear-quadratic model is a polynomial-based model that has degree  2.
That is,  the polynomial on which a linear-quadratic model is based includes a
term for the background (a0) ,  a term that is linear in  dose (dose  raised to the
power 1,  which is often represented with no exponent, i.e., d1 is  the  same as d) ,
and a term with dose raised to the second power (a2*d2) .

     A pure quadratic model  is similar  to the linear-quadratic model in that  it
has degree 2.  However,  the pure quadratic model dose not have a linear term; it
includes only the background term (a0)  and a term  with  dose raised to  the second
power.  One can think of the pure quadratic model as a submodel of the linear-
quadratic model,  where the coefficient for the linear term (ax)  has been set to
zero.
                                 5-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993

EPA Rebuttal to Clement (API) Risk Assessment of Benzene

     EPA (Chen et al.  1989) raised several issues in response to
the Clement (API) risk assessment of benzene.  These issues are
summarized below:

     1)    Use of Rinsky's Cohort as the Sole Data Base.  Chen et
          al.  stated that the Rinsky study lacks adequate
          exposure information during the early but critical
          years of employment of the cases.  Also, none of the
          three epidemiological studies used by EPA is considered
          to be superior to any other.

          Furthermore, it was stated that Clement gave "an
          incomplete picture of other studies and therefore
          reduced their usefulness by leaving out important
          details about those studies that do not support the use
          of the Rinsky study as the sole data source."
          Therefore, Chen et al.  (1989) does not agree with
          choosing the Rinsky cohort as the sole data base for
          the benzene risk assessment.

     2)    Differences Between Clement  (API) and Rinsky himself in
                    the Use of Rinsky Data Tapes.  Chen et al.
                    (1989)  stated that there appears to be some
                    differences between the Rinsky data tapes
                    used by API and the Rinsky data tapes used by
                    Rinsky himself in his 1987 published paper.

     3)    Only Certain Types of Leukemia are Induced by Benzene.
          Chen et al.  (1989) did not agree that acute myelogenous
          leukemia and aplastic anemia were the only disease end
          points associated with benzene exposure.  They stated
          that there is also evidence linking acute and chronic
          forms of lymphocytic leukemia as well as acute
          nonlymphatic leukemia and multiple myeloma to benzene
          exposure,  and that these should be included in a risk
          assessment of benzene.
                               5-44

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     4)   Blood Counts and the Crump-Allen Exposure Estimate.
          Chen et al.   (1989)  stated that the evidence provided by
          Clement to justify the use of the Crump and Allen
          (1984)  exposure estimate is disputed by Rinsky, and
          that both the Rinsky and Crump and Allen exposure
          estimates should be considered.  Clement stated that
          the Crump and Allen exposure estimate was superior to
          Rinsky's because higher blood counts are correlated
          with lower exposure estimates, while no correlation was
          found using the Rinsky estimate.

     5)   Benzene has a Non-Linear and Threshold Dose Effect.
          Clement stated that the Rinsky study showed a strong
          non-linearity of leukemia mortality rate with dose
          using either the Rinsky or the Crump and Allen exposure
          estimates.  EPA's view is that linear low-dose
          extrapolation is preferred, unless low dose data and/or
          mechanism/metabolism knowledge show otherwise.

     6)   Clement's (API's) Model is Superior to EPA's 1985
          Model.   As discussed above, Clement stated that their
          model represents an improvement over EPA's 1985 model
          because it incorporates latency period data and
          individual exposure information.  Chen et al.  (1989)
          maintains that the way the latency is incorporated in
          the model is not appropriate, and that the equation
          used by Clement to estimate benzene-induced age-
          specific cancer rate is not accurate (see item 7
          below).

     7)   Problems in the Clement (API) Procedures for Risk
          Calculation.  Chen et al.   (1989) stated that the way
          the latency is incorporated by API into the Moolgavkar
          (MVK) model is both mathematically and biologically
          inappropriate because it assumes that one and only one
          single tumor cell will eventually lead to leukemia
          death.   Furthermore, it is stated by Chen et al.  (1989)
          that the argument provided in the Clement assessment to
          support the use of an absolute risk model over a
          relative risk model is not convincing.

5.6.3  Recent and Ongoing Research

5.6.3.1 Genotoxicity

     Benzene and its metabolites have been shown to cause
clastogenic effects (damages or breaks of the genetic material
that can be observed at the chromosome level) such as sister
chromatid exchange  (SCE),  micronuclei, and chromosomal
aberrations in both in vivo and in vitro systems in both humans
and animals.  However, studies attempting to show mutagenic
activity of benzene have generally been negative (Shahin and
Fournier 1978; Lebowitz et al. 1979; Bartsch et al. 1980;
Nestmann et al. 1980;  Shimizu et al. 1983; Nylander et al. 1978).
Recent work by Glatt et al. (1989) has shown, using a closed
desiccator system, that benzene is mutagenic in Salmonella


                               5-45

-------
                                                        EPA-420-R-93-005
                                                            April 1993

typhimurium (a type of bacteria) strain TA1535 in the presence of
activated microsomal enzymes.  These results suggest that only
the metabolites of benzene are mutagenic.

     Conflicting results have been obtained regarding the ability
of benzene to form DNA adducts in vivo.  Although DNA adducts
have been demonstrated in in vitro experiments with a variety of
benzene metabolites, Reddy et al. (1989) have not observed DNA
adducts in samples of liver, kidney, bone marrow, and mammary
gland obtained from Sprague-Dawley rats following oral
administration of benzene (500 mg/kg/day, 5 days/week, for up to
10 weeks).   The only potential adducts identified were observed
in isolated rat Zymbal glands.   In contrast, Snyder et al.  (1989)
observed a peak upon HPLC analysis of bone marrow DNA from rats
treated with 1 mL/kg of benzene, 1 time per day for 4 days, with
a longer retention time than any of the deoxynucleotide
standards,  suggesting covalent binding of benzene/benzene
metabolite with the DNA.

     Recent research has also examined the genotoxicity  (i.e.,
the ability to damage the chromosomes at the DNA level) of the
recently identified benzene metabolite, trans, trarzs-muconaldehyde
(Latriano et al.  1986).  Trans, trarzs-muconaldehyde has been
demonstrated to form stable DNA adducts when reacted with
deoxyguanosine monophosphate  (Latriano et al. 1989).
Deoxyguanosine monophosphate is a nucleic acid that is one of the
building blocks of DNA.  Also, it has been shown to be strongly
mutagenic in Chinese hamster V79 cells and weakly mutagenic in
bacterial systems (Glatt and Witz 1990) .  When administered to
mice, trans, trarzs-muconaldehyde  increased sister chromatid
exchanges (Witz et al. 1990).

     Recent work has also demonstrated that 1,4-benzoquinone and
1,2,4-benzenetriol,  which are metabolites of benzene, inhibit DNA
synthesis in a cell-free DNA synthetic system (Lee et al. 1989) .
The inhibitory effect was concluded to be due to inhibition of
polymerase a by these metabolites.

     Chromosomal aberrations occurring in humans with leukemia
thought to be associated with exposure to benzene have been
reported.   A recent letter to The Lancet by Lumley et al.  (1990)
described the case of a 58-year old heavy-goods-vehicle driver
who had heavy exposure to gasoline  (and thus, benzene) who
developed thrombocytopenia,  neutropenia, and acute myeloid
leukemia.   He was found to have multiple chromosomal
abnormalities including deletion of the long arm of chromosome 5,
which the authors describe as a cytogenetic hallmark of secondary
leukemia.   The authors cite this example of non-random
chromosomal changes in individuals with known benzene exposure as
useful for early detection of those at risk for developing
leukemia.
                               5-46

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     In a recent study, Irons et al.   (1992) tested the effects of
benzene metabolites on the growth of myeloid progenitor cells
(bone marrow cells that are the precursor to granulocytes and
macrophages).   The benzene metabolite, hydroquinone,  has been
shown in previous studies to cause malignant transformations such
as inhibition of microtubule assembly  (essential for cell
division) and nondisjunctional events  (a loss of all or part of
chromosomes 5 and 7)  in these progenitor cells.  Irons et al.
(1992)  hypothesized that if agents with leukemogenic potential
(such as hydroquinone)  have the ability to produce alterations in
the regulation of these cells ( i.e., increased growth), the
absolute number of dividing progenitor cells would be increased.
The increased size of this dividing cell population would serve
to increase the probability of malignant transformations
occurring.  Actively dividing cells are also more susceptible to
transformations due to their nature.

     In vitro pretreatment of murine  (mouse) bone marrow cells
with  hydroquinone and the stimulating factor that is required
for their differentiation and replication results in an
enhancement of granulocyte/macrophage colony formation.  The
magnitude of hydroquinone-enhanced colony formation equals or
exceeds that described for the synergistic action of other
compounds known to cause cell differentiation with the
stimulating factor.  The potential of hydroquinone to alter
growth response and induce differentiation in a myeloid (bone
marrow) progenitor cell population may be important in the
pathogenesis of acute myelogenous leukemia secondary to benzene
exposure.  Benzene leukemogenesis may result from the dual
ability of its metabolites to promote progenitor cell growth and
differentiation and also induce cytogenetic changes in
replicating cells.  If other leukemogenic agents act similarly,
alterations in myeloid progenitor cell differentiation may be
important in the pathogenesis of secondary acute myelogenous
leukemia in general.

     These new studies provide additional support for the
clastogenic ability of benzene metabolites and provide new
evidence for the potential mutagenic activity of some of these
metabolites.  Furthermore, the occurrence of certain chromosomal
aberrations in individuals with known exposure to benzene may
serve as a marker for those at risk for contracting leukemia.

5.6.3.2  Pharmacokinetics

     The tumor diversity observed in different strains and
species of rodents has been proposed to be due to the production
of a number of potentially carcinogenic metabolites of benzene
that may act singly or in combination  (although the specific
"active" metabolites have not yet been identified)  (Huff et al.
1989).

     A number of recent studies have examined the effects of
dose,  dose rate, route of administration, and species on benzene
metabolism.  For example, Sabourin et al.  (1987) demonstrated
that administration of bolus doses > 50 mg/kg by oral or


                               5-47

-------
                                                        EPA-420-R-93-005
                                                            April 1993

intraperitoneal injection to rats and mice exceeded the metabolic
capacity of these rodents and resulted in a portion of the dose
being exhaled as benzene and a decrease in conversion to reactive
benzene metabolites.  As the dose was increased above 50 mg/kg,
proportionately more was exhaled and less converted to benzene
metabolites.  With inhalation exposures, a similar phenomenon was
observed in rats and mice.  However, mice had a more rapid
metabolic rate than rats, resulting in higher metabolite
production in mice after an extended inhalation exposure (i.e., 6
hr).   In mice the toxic pathways became saturated, whereas, in
rats there was a relative increase in nontoxic pathways as the
dose increased.
     In addition to the higher metabolic rate in mice than in
rats, Sabourin et al.   (1988) demonstrated that mice and rats use
the various metabolic pathways for benzene to differing degrees.
In mice, detoxification also predominated, but substantial
conversion to toxic metabolites was apparent.  In rats, no
saturation of toxic pathways was evident with increasing dose
rate; however, increases in the mouse exposure level resulted in
a shift from toxic metabolic pathways to detoxification pathways
(Sabourin et al.  1988).  These results indicate that the net
result of exposures to high concentrations (200 ppm by the
inhalation route) is to decrease the proportion of toxic
metabolites formed relative to the dose administered in both mice
and rats, with low level, long duration exposures of mice
producing more toxic metabolites.

     Age-related differences in benzene pharmacokinetics also
appear to occur.   McMahon and Birnbaum  (1991) found that the
disposition of benzene differed between 3- and 18-month-old male
B6C3F1 mice administered as a single oral dose of either 10 or
200 mg/kg benzene.  These differences include increased urinary
and fecal elimination, increased expiration of benzene derived
C02,  and an effect on  the metabolism of  benzene  to specific
metabolites.  While these differences may be due to the
physiological effects of aging rather than direct age-related
differences in the metabolism of benzene, these results have
important implications with respect to the extrapolation of data
obtained in young or old animals to young or old humans.

     Using information about the relationship of the exposure
conditions with the internal dose of various metabolites,
computer simulations can be generated to estimate metabolite
concentrations after differing exposure regimens  (Medinsky et al.
1989) and derive internal doses that may be correlated with
observed carcinogenic responses  (Bailer and Hoel 1989) for risk
assessment.  Recently, Bois and Spear (1991)  attempted to
correlate circulating levels of phenol and hydroquinone with the
onset of cancer in rats and mice using a computer model of
benzene metabolism.  No correlation was observed, indicating that
other metabolites or combinations of
                               5-48

-------
                                                        EPA-420-R-93-005
                                                            April 1993

metabolites may be important in the initiation of cancer
following benzene exposure.

     Two benzene metabolites that have been shown to interact
metabolically are phenol and hydroquinone to produce 1,4-
benzoquinone in vitro (Eastmond et al.  1987).   The observation
that phenol and hydroquinone, when administered together in mice,
produced a much greater decrease in bone marrow cellularity than
did administration of either metabolite alone, suggests that a
similar enhancement of the formation of the myelotoxic
metabolite, 1,4-benzoquinone, may also occur in vivo.

     A physiologically based pharmacokinetic model (PBPK) model
for benzene has been developed by Travis et al.  (1990).  PBPK
models are designed to allow more accurate prediction of actual
internal doses across species.  This particular model was
developed to quantitatively predict the fate of benzene in mice,
rats, and humans following several routes of exposure.  One of
the advantages to having a highly predictive PBPK model for
benzene is that exposure data from benzene-induced cancers in
animals may be directly compared to exposure data from human
epidemiological studies in terms of metabolized dose, and
therefore, route-to-route extrapolations can be made with a
higher degree of confidence.

     These studies demonstrate that species differ with respect
to their ability to metabolize benzene.  These differences may be
important when choosing an animal model for human exposures and
when extrapolating high dose exposures in animals to the low
levels of exposure typically encountered in occupational
situations.  The development of a PBPK model for benzene should
help in performing interspecies and route-to-route extrapolations
of cancer data.  Furthermore, metabolite interactions should be
considered in developing PBPK models.

5.6.3.3  Carcinogenicity - Animal Studies

     Recent studies examining the carcinogenicity of benzene in
rodents have demonstrated that benzene is a potent carcinogen in
a number of organs in a variety of species and strains of mice
and rats, whether administered orally or by inhalation.  In a
recent NTP bioassay (Huff 1986),  administration of benzene by
gavage produced a variety of types of tumors in male and female
F344/N rats and B6C3F1 mice.  Male rats were administered 0, 50,
100, or 200 mg/kg benzene and female rats and male and female
mice were administered 0, 25, 50, 100 mg/kg in corn oil by gavage
for 103 weeks.   Female rats administered benzene at 25 mg/kg and
above caused significantly increased incidences of Zymbal gland
carcinoma and at 50 mg/kg and above, squamous cell carcinomas and
papillomas of the oral cavity. In male rats, at 100 mg/kg and
above, Zymbal gland carcinomas, squamous cell carcinomas
(malignant tumors of the skin), and papillomas (benign tumors) of
the oral cavity were
                               5-49

-------
                                                        EPA-420-R-93-005
                                                            April 1993

significantly increased.  Also in male rats, skin papillomas were
increased at 200 mg/kg and above.

     In mice, significantly increased incidences of Zymbal gland
carcinomas, malignant lymphomas,  and alveolar/bronchiolar
carcinomas at 50 mg/kg and above were observed.  Harderian gland
adenomas increased at 25 mg/kg and above, and squamous cell
carcinomas of the preputial gland increases at 50 mg/kg and above
were observed in male mice.  At 25 mg/kg and above, malignant
lymphomas increased and ovarian granulosa cell tumors  (tumor of
the ovary), carcinomas of the mammary gland, and
alveolar/bronchiolar carcinomas increased at 50 mg/kg and above
in female mice.  Zymbal gland carcinomas at 100 mg/kg were also
observed in females.  Alveolar tumors are located in the deepest
part of the lung in the tissue where air exchange with blood
takes place.  Bronchiolar tumors are located in the bronchial
tubes in the lungs.  In general,  mice were more sensitive to the
carcinogenic effects of benzene than were rats.

     Similar results were presented by Maltoni et al.   (1989).
When administered benzene  (0, 50, 250 mg/kg or 0, 500 mg/kg) by
gavage in olive oil for 104 weeks, benzene-exposed Sprague-Dawley
rats had increased incidences of tumors of the Zymbal gland, oral
cavity, nasal cavity, skin, forestomach, liver angiosarcomas
(malignant tumors in blood vessels in the liver) and marginal
increases in carcinomas of the mammary glands, hepatomas  (liver
tumors), and leukemias.  Wistar rats administered 0 or 500 mg/kg
by gavage in olive oil for 104 weeks had increased incidences of
carcinomas of the Zymbal gland, oral cavity, and nasal cavity in
benzene-exposed animals.  Swiss mice administered 0 or 500 mg/kg
by gavage in olive oil for 78 weeks had increased incidences of
carcinomas of the mammary gland,  lung tumors, and carcinomas of
the Zymbal glands in those mice exposed to benzene.  RF/J mice
administered 0 or 500 mg/kg by gavage in olive oil for 52 weeks
had increased incidences of mammary carcinomas, lung tumors, and
leukemias in those mice exposed to benzene.  When adult Sprague-
Dawley rats inhaled either 0 or 200 ppm 4 hr/day, 5 days/week for
7 weeks followed by 200 ppm 7 hr/day, 5 days/week for 12 weeks
and then 300 ppm 7 hr/day, 5 days/week for 85 weeks, an increased
incidence of carcinomas of the Zymbal glands and oral cavity were
observed with marginal increases in carcinomas of the nasal
cavity, mammary glands, and hepatomas in benzene-exposed rats.
Slightly greater numbers of tumors were observed when inhalation
exposure at the above concentrations began on day 12 of
gestation.

     Recently, an increased incidence of myelogenous leukemias
(the type of cancer associated with benzene exposure in humans)
was reported in mice exposed to benzene by inhalation  (Cronkite
et al.  1989).  CBA/Ca mice were used in this study.  These mice
come from the same stock as a strain (CBA/H) known to have a low
incidence of acute myeloblastic leukemia (a type of myelogenous
leukemia), but which respond to ionizing radiation with a high
incidence of these tumors.  Inhalation of 300 ppm, 6 hr/day, 5
days/week, for 16 weeks significantly decreased survival and
increased the incidence of myelogenous neoplasms in male and


                               5-50

-------
                                                         EPA-420-R-93-005
                                                             April 1993

female CBA/Ca mice.  Also,  an increased incidence of neoplasms
other than hepatic and  hematopoietic cancers such as squamous
cell carcinoma, mammary adenocarcinoma (tumors of the mammary
gland),  Zymbal and Harderian gland tumors,  and papillary
adenocarcinomas of the  lung was  observed in these mice.  These
tumors (myelogenous neoplasms and other neoplasms) were observed
earlier in benzene-treated  animals than in the controls.  Hepatic
neoplasms also appeared sooner in the benzene-exposed mice;
however,  both hepatic and lymphomatous neoplasms were
significantly decreased in  benzene-treated mice.  At lower
concentrations  (100 ppm), an increased incidence of tumors other
than hematopoietic and  hepatic neoplasms was also observed
although no significant increase in myelogenous neoplasms was
seen.  Preliminary results  indicated that exposure of these mice
to much higher concentrations of benzene (3,000 ppm for 8 days)
did not produce similar increases in mortality or cancer
incidence.  The absence of  neoplastic effects at this high dose
is consistent with the  much lower hematotoxicity  (blood disease)
observed with exposure  to 3,000  ppm for 2 days as compared with
exposure to 316 ppm for 19  days  (exposures designed to yield
similar total doses of  benzene).

     These new studies  provide additional support for the
carcinogenicity of benzene  in animals by both the oral and
inhalation routes and provide the first animal model for the type
of neoplasm identified  most closely with occupational exposure,
acute myelogenous leukemia.   Benzene has been shown to be
carcinogenic in both sexes,  at multiple sites, in several strains
of rats and mice.

5.6.3.4  Carcinogenicity -  Epidemiological Studies

     Several studies have become available since the 1985 EPA
carcinogenicity assessment,  and have not been considered in the
derivation of the cancer potency factor for benzene.  The study
by Rinsky et al.  (1981)  has been updated through December 31,
1981  (Rinsky et al. 1987).   An additional two deaths attributable
to leukemia were included in the update,  bringing the total
number of leukemia deaths to nine.  The standardized mortality
ratio (SMR) for leukemia calculated in the update was 337  (95%
confidence interval = 154-641).   Also, a significant increase in
multiple myeloma7 was observed in  the  updated cohort (SMR=409,
95% confidence interval=110-1047).  Latency for the leukemia
deaths ranged from 5-30 years with seven of the nine deaths
occurring with a latency of <20  years.  In contrast, latency for
all of the cases of multiple myeloma was >20 years.  A matched
case-control analysis was also performed using conditional
logistic regression analysis.   Conditional logistic regression is
used in a case-control  study when the cases  (i.e., exposed
individuals) and controls  (i.e.,  non-exposed individuals) have
      Myeloma is a tumor originating in the cells of the blood-forming portion
of bone marrow.  Multiple myeloma is a type of cancer characterized by the
infiltration of bone and bone marrow with myeloma cells that form multiple tumor
masses. This disease is usually progressive and fatal, and is accompanied by
anemia, renal lesions,  and high globulin levels in the blood.

                               5-51

-------
                                                        EPA-420-R-93-005
                                                            April 1993

been matched (i.e., matched pairs).   It then can provide an
unbiased estimate on a number of factors of the relative risk.
Although the information reported in the Rinsky et al.  (1987)
update do not qualitatively change the current EPA risk
assessment for benzene (i.e., they support the conclusion that
benzene exposure is associated with an increased incidence of
leukemia),  the analytical methods used in this update, the
improved exposure data, and the larger cohort size may impact the
quantitative assessment of cancer risk based on this cohort.

     The study by Ott et al.   (1978)  has also been updated (Bond
et al.  1986)  expanding the cohort size from 594 to 956 and
increasing the period of observation to 1940-1982.  An additional
death was reported in this update, bringing the total to four
leukemia deaths  (all of the myelogenous type).   Myelogenous
leukemias are diseases where there is unrestrained growth of
myelocytes, which are large cells found in the bone marrow that
develop into white blood cells.  Although the SMR for leukemia
deaths was not significantly elevated, the mortality due to acute
myelogenous leukemia was significantly increased.  However,  a
positive dose-relationship between benzene exposure and leukemia
was not observed.

     A conditional logistic regression case-control analysis of
the Ott et al.  (1978) and Bond et al.   (1986) studies was
performed by the American Petroleum Institute (API) similar to
that described in the Rinsky et al.  (1987) update.  However, API
failed to observe a statistically significant relationship
between increasing cumulative benzene exposure and increased risk
of leukemia (Peterson 1986).

     The study by Wong et al.  (1983) has also been updated  (Wong
1987).   Two additional leukemias have been added to the cohort,
bringing the total to seven.   When compared to workers with no
occupational exposure to benzene, those with at least 720 ppm-
months of exposure to benzene had a relative risk of 3.93 for
lymphatic and hematopoietic cancer.   Workers with <180, 180-719,
and >720 ppm-months of exposure had a borderline significantly
increased incidence of non-Hodgkin's lymphopoietic cancer with
increased exposure.

     A new retrospective mortality study was published by Yin et
al.   (1989)  of 28,460 benzene-exposed workers from 83 factories in
China.   Mortality of workers with at least six months of exposure
to benzene between January 1, 1972 and December 31, 1981 was
compared with mortality of a similar number of workers from these
factories who had not been exposed to benzene.   Significantly
increased SMRs for leukemia  (SMR=5.74) and lung cancer  (SMR=2.31)
were observed among exposed males and an increased SMR for
leukemia was observed among exposed females.  A higher proportion
of acute nonlymphocytic leukemias were observed and a lower
proportion of acute lymphocytic leukemias were seen than in the
general population.  The risk of leukemia increased with exposure
duration up to 15 years and then declined with additional years
of exposure.   Cumulative exposure estimates were also performed
in this study although measurements of ambient benzene levels


                               5-52

-------
                                                        EPA-420-R-93-005
                                                            April 1993

were not complete for all of the subjects.  The cumulative
exposure estimates supported the findings by Rinsky et al.   (1987)
that leukemia was, in many cases, seen in workers with continuous
low dose exposure less than 400 ppm-yr of exposure.  Smoking
histories were determined in this study and the results
demonstrated that smoking had no effect on leukemia mortality.
Smoking increased the mortality due to lung cancer, but
significantly greater lung cancer mortality was observed in
exposed nonsmokers than in nonexposed nonsmokers, suggesting that
lung cancer may also be associated with benzene exposure.

     The updated studies provide continued evidence of the
carcinogenicity of benzene in humans, and incorporation of
increased cohort sizes and improved exposure analyses in these
studies may strengthen the current cancer risk assessment for
benzene.  Furthermore, the observation of significantly increased
lung cancer, as well as increased acute myelogenous leukemia, in
the new study by Yin et al.  (1989), suggests that benzene might
be a multisite carcinogen in humans, as has been indicated in
animal studies.

     Morris and Seifter (1992)  hypothesize that the increase in
breast cancer incidence observed in urban areas may be due to the
increased exposure to aromatic hydrocarbons found in urban
pollution.  Aromatic hydrocarbons are capable of inducing breast
cancer in animals and benzene is a known cause of leukemia in
humans.

     Most aromatic hydrocarbons and benzene are readily soluble
in fatty tissue (e.g., breast tissue) where they are stored,
concentrated, and metabolized in the breast tissue to
carcinogenic compounds.  Some of these aromatic hydrocarbons
produce electron seeking metabolites which can adduct to the DNA,
causing mis-replication which can lead to tumor production.
Other metabolites can function in the role of tumor promoter by
producing an oxidant through their metabolic detoxification
pathways.  These oxidants, oxygen free radicals  (activated forms
of oxygen),  consume glutathione  (an anti-oxidant in the cells)
that would otherwise protect against tumor promotion.  Some
aromatic hydrocarbons can react with cell membrane receptor sites
causing oxygen free radicals to peroxidant the polyunsaturated
lipids of the cell membrane.  These lipid peroxidases and their
degradation by-products cause chromosomal breaks in the related
cell and also in remote tissues.  The consequence of long term
hydrocarbon exposure is the possibility of an increased pro-
oxidant state which destabilizes DNA, causes chromosomal breaks,
and allows for initiation and promotion of breast cancer.

     In urban communities, there is increased exposure to
hydrocarbons due to the use of fossil fuels and, concurrently,
there is an increased personal exposure to hydrocarbons.  It is
the authors' contention that this low dose, long term exposure to
many mammary specific hydrocarbon carcinogens and to the
promotional effects of perhaps hundreds of other carcinogenic and
non-carcinogenic hydrocarbon metabolites accounts for the urban
factor in breast cancer.


                               5-53

-------
                                                        EPA-420-R-93-005
                                                            April 1993

5.7  Carcinogenic Risk for Baseline and Control Scenarios

     Since the benzene unit risk estimate is based on human
epidemiology of death data, cancer numbers should be expressed as
cancer deaths.  The estimate of cancer deaths may underestimate
cancer incidence associated with benzene, since survivorship
rates were not included in the supporting studies.  Table 5-10
summarizes the maximum likelihood estimates of annual cancer
deaths for all scenarios.  When comparing cancer deaths for the
base control scenarios relative to 1990, there is a 39% reduction
in 1995,  a 50% reduction in 2000, and remains constant at a 50%
reduction in 2010.  The reduction in emissions is considerably
higher, particularly in the out years.  The projected increase in
both population and vehicle miles traveled (VMT)  from 2000 to
2010 appears to offset the gains in emissions achieved through
fuel and vehicle modifications.

     The base control and expanded use scenarios within each year
can be directly compared since the same VMT and populations are
applied to both.  In 1995, expanding the reformulated fuels
program reduces the number of cancer deaths by another 8% from
the 1990 base control.  The expanded use of reformulated fuels
and the California program in the year 2000 produces another 6%
reduction in cancer cases, for both scenarios, when compared to
1990.  Expanded reformulated fuel use in 2010 reduces deaths due
to cancer  by 6% relative to 1990 and by approximately 10% for
the expanded California standards scenario.  Like the base case
comparison,  the cancer cases for the control scenarios are
similar for 2000 and 2010 despite continued emission reduction,
due to the projected population and VMT increase.

5.8  Non-Carcinogenic Effects of Inhalation Exposure to Benzene

     EPA has no inhalation reference concentration for the
noncancer effects of benzene that can be used as a basis for risk
assessment.   Benzene's carcinogenic effects serve as the basis
for the benzene risk assessment.  Since the focus of this report
is on the carcinogenic potential of the various compounds, the
noncancer
                               5-54

-------
                                                                                  EPA-420-R-93-005
                                                                                       April 1993
Table 5-10.  Annual Cancer  Death Projections for Benzene.
                                                          a,b
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0882
0.0472
0.0413
0.0351
0.0301
0.0305
0.0285
0.0248
0.0228
Urban
Cancer
Deaths
59
36
31
30
26
26
30
26
24
Rural
Cancer
Deaths
11
7
6
5
5
5
5
5
4
Total
Cancer
Deaths
70
43
37
35
31
31
35
31
28
Percent Reduction
from 1990
EF
-
46
53
60
66
65
68
72
74
Cancer
-
39
47
50
56
56
50
56
60
""Projections have inherent uncertainties  in  emission estimates,  dose-response, and
exposure.
bThe unit risk estimate for benzene  is based on human data.   Benzene is classified by  EPA
as a Group A, known human  carcinogen based on sufficient human epidemiologic  evidence in
addition to supporting  animal  evidence.
                                            5-55

-------
                                                           EPA-420-R-93-005
                                                               April 1993

information will be dealt with in a more cursory fashion.  No
attempt has been made to synthesize and analyze  the data
encompassed below.  Also, no  attempt has been made to accord more
importance  to one type of noncancer effect over  another.  The
objective is to research all  existing data, describe the
noncancer effects observed, and refrain from any subjective
analysis of the data.

     The respiratory route  is the major source of human exposure
to benzene,  and much of this  exposure is by way  of gasoline
vapors and  automotive emissions (EPA, 1980) .  Individuals
employed in industries that use or make benzene  or benzene-
containing  products may be  exposed to the highest concentrations
of benzene,  primarily by inhalation.  In 1987, OSHA estimated
that approximately 238,000  workers were exposed  to benzene in
seven major industry sectors,  including petrochemical plants,
petroleum refineries, coke  and chemicals, tire manufacturers,
bulk terminals,  bulk plants,  and transportation  via tank trucks
(OSHA 1987).   The toxic effects of benzene in humans and other
animals following inhalation  exposure include central nervous
system  (CNS),  hematological,  and immunological effects.  In
humans, acute exposure to 20,000 ppm is usually  fatal within 5-10
minutes  (Gerarde 1960).  Death is preceded by CNS effects such as
drowsiness,  headache, nausea,  staggering gait, delirium, vertigo,
tremors, convulsions, and unconsciousness  (Cronin 1924; Gerarde
1960; Browning 1965).  In humans,  death has been tentatively
attributed  to asphyxiation, respiratory arrest,  CNS depression,
or cardiac  arrhythmia (Winek  and Collum 1971).   Organ hemorrhage
was also reported.  An inhalation LC50  (the  concentration that  is
lethal to half of the animals exposed by the inhalation route)
value for rats was calculated as 13,700 ppm for  a 4-hour exposure
(Drew and Fouts 1974).

     Benzene induces hematological effects in humans and animals.
Early stages of benzene toxicity may be characterized by
deficiencies in specific blood elements, resulting in anemia  (a
reduction in the number of  red blood cells), leukopenia  (a
reduction in the number of  white blood cells) , or
thrombocytopenia  (a reduction in the number of blood platelets).
Chronic inhalation exposure to benzene in humans results in
pancytopenia,  a condition characterized by decreased numbers of
circulating erythrocytes  (red blood cells), leukocytes  (white
blood cells),  and thrombocytes (blood platelets)  (Aksoy and Erdem
1978; Aksoy et al. 1971)8.   Individuals that develop  pancytopenia
      Pancytopenia is the reduction in the number of  all three major types of
blood cells (erythrocytes, or red blood cells,  thrombocytes, or platelets, and
leukocytes, or white blood cells).  In adults,  all three major types of blood
cells are produced in the red bone marrow of the vertebra,  sternum, ribs, and
pelvis.  The red bone marrow contains immature  cells,  known as multipotent
myeloid stem cells, that later differentiate into the  various mature blood cells.
Pancytopenia results from a reduction in the ability of the red bone marrow to
produce adequate numbers of these mature blood  cells.  Aplastic anemia is a more
severe blood disease and occurs when the bone marrow ceases to function, i.e.,
these stem cells never reach maturity.  The depression in bone marrow function
occurs in two stages - hyperplasia, or increased synthesis of blood cell
elements, followed by hypoplasia, or decreased  synthesis.  As the disease
progresses, the bone marrow decreases functioning.  This myeloplastic dysplasia

                                5-56

-------
                                                        EPA-420-R-93-005
                                                            April 1993

and have continued exposure to benzene may develop aplastic
anemia  (pancytopenia associated with fatty replacement of
functional bone marrow),  whereas others exhibit both pancytopenia
and bone marrow hyperplasia, a condition that may indicate a
preleukemic state  (Aksoy et al. 1974; Aksoy and
Erdem 1978).   Similar hematological effects have been reproduced
in animals.

     Symptoms of immunotoxicity have been reported in workers
chronically exposed to benzene at concentrations that ranged from
3.44-53.21 ppm for 1-21 years.  Alterations in serum levels of
immunoglobulin (proteins in the blood that are capable of acting
as antibodies) and complement  (a series of enzymatic proteins  in
normal serum that, in the presence of a specific stimulus,
destroy bacteria and other cells) and indications of benzene-
induced autoimmunity and allergy have been observed in benzene-
exposed workers whose exposure has been intermediate or chronic
(Lange et al.  1973a, 1973b).  Eosinophilia, an indication of an
allergic response, has been noted in Turkish workers  (Aksoy et
al.  1971).  Evidence of a positive leukocyte autoagglutinnin
test,  associated with decreased granulocyte levels, was
suggestive of allergic blood dyscrasia  (disease)  (Lange et al.
1973b).   The autoagglutinnin test measures the clumping of one's
own blood cells.   A positive response indicates that one's own
blood cells stimulate an allergic response in the body.  In
animals, lymphopenia appears to be the most consistent response
to subchronic benzene exposure, and may be seen at exposures as
low as 25 ppm (Cronkite et al. 1989).  A dose-response study of
short-term inhalation exposure to benzene in mice at levels of
10-30 ppm showed significantly depressed proliferative responses
of bone-marrow-derived B cells and splenic T cells in mice  (Rozen
et al. 1984).   Mice with Listeria monocytogenes  (a form of
bacteria) exposed to intermittent benzene concentrations of 300
ppm resulted in delayed cell-mediated immunity, causing increased
bacterial numbers  (730% of controls) on day 4  (Rosenthal and
Snyder 1985).

     The available human data on developmental effects of benzene
are inconclusive.  Savitz et al. (1989) conducted an
epidemiological study aimed at assessing the effect of parents'
occupational exposures on risk of stillbirth, preterm delivery,
and small-for-gestational age infants.  They used data from
National Natality and Fetal Mortality surveys on the probability
samples of live births and fetal deaths that occurred in the US
in 1980 among married women.  Savitz et al.  (1989) found that  a
high maternal linkage to benzene was predictive of stillbirth
risk.   Another significant association was found for paternal
exposure to lead and risk of small-for-gestational age.  Despite
the limitations inherent in the study design  (i.e., lack of
exposure data, small size of exposed populations, and possible
confounding factors not accounted for), these results suggest
that occupational exposure to benzene may be associated with
adverse developmental and reproductive outcomes.
without acute leukemia is known as preleukemia.  The aplastic anemia can progress
to AML.

                               5-57

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Several animal studies, involving acute inhalation exposure
during pregnancy, have shown that exposure to benzene decreased
body weight and increased skeletal variants such as missing
sternebrae and extra ribs (Murray et al. 1979; Kimmel and Wilson
1973).   Alterations in hematopoiesis (growth and development of
blood elements) have been observed in the fetuses and offspring
of pregnant mice exposed to benzene  (Keller and Snyder 1986).
Two recent reports have described adverse immunological and
hematopoietic effects associated with in utero exposure to
benzene. One of these studies, Wierda et al.  (1989), produced
results that suggest that in utero exposure of benzene may
adversely alter B cell development and responsiveness, and thus,
compromise the immune system after birth.

     Benzene may impair fertility by causing ovarian atrophy
among women occupationally exposed to high levels  (levels not
specified) of benzene (Vara and Kinnunen 1946).  In mice,
histopathological changes were observed in ovaries  (bilateral
cysts)  and testes (atrophy/degeneration, decrease in spermatozoa,
moderate increase in abnormal sperm forms) following exposure to
300 ppm benzene for 13 weeks  (Ward et al. 1985).  No studies were
located regarding respiratory, hepatic, or renal effects in
humans or animals after inhalation exposure to benzene.

     Corti and Snyder (1990) reported in a recent abstract that
inhalation exposure of female Swiss Webster mice to 10 ppm
benzene on gestation days 6-15 resulted in a reduction in the
number of erythrocyte progenitor cell colonies in bone marrow
cell cultures from female offspring 6 weeks after birth.  There
was no effect apparent in the male offspring.  These results
suggest that in utero exposure to benzene may adversely affect
normal hematopoietic development.

     The inhalation reference concentration  (RfC) for benzene is
currently under review by the EPA RfD/RfC Workgroup  (EPA, 1992c).
The oral reference dose (RfD) for benzene will be reviewed by the
EPA RfD/RfC Workgroup (EPA,  1992c).
                               5-58

-------
                                                        EPA-420-R-93-005
                                                            April 1993
5.10  References for Chapter 5
Aksoy, M., K. Dincol, T. Akgun, S. Erdem, and G. Dincol.  1971.
Haematological effects of chronic benzene poisoning in  217
workers.  Br. J. Ind. Med. 28:296-302.

Aksoy, M., S. Erdem, and G. Dincol.  1974.  Leukemia in shoe-
workers exposed chronically to benzene.  Blood 44:837.

Aksoy, M.  1978.  Benzene and leukemia.  Lancet 1: 441.

Aksoy, M., and S. Erdem.  1978.  Follow-up study on the mortality
and the development of leukemia in 44 pancytopenic patients with
chronic benzene exposure.  Blood 52:285-292.

Aksoy, M.  1980.  Different types of malignancies due to
occupational exposure to benzene:  A review of recent
observations in Turkey.  Environ. Res.  23: 181.

American Petroleum Institute.  1991.  Letter from Terry F. Yosie
of API to Phil Lorang of EPA.  December 9, 1991.

Anderson, D. and C.R. Richardson.  1979.  Chromosome gaps are
associated with chemical mutagenesis  (abstract No. Ec-9).
Environ. Mutat.   1: 179.

Atkinson, R.  1990.  Gas-phase tropospheric chemistry of organic
compounds:  a review.  Atmos. Environ., 24A:1-41.

Atkinson, R., S. M. Aschmann, J. Arey, and W.P.L. Carter.  1989.
Formation of ring-retaining products from the OH radical-
initiated reactions of benzene and toluene.  Int. J. Chem.
Kinet., 21:801-827.

Bailer, A.J. and D.G. Hoel.  1989.  Metabolite-based internal
doses used in a risk assessment of benzene.  Environ. Health
Perspect. 82:177-184.

Bartsch, H., C.  Malaveille, A.M. Camus, G. Martel-Planche, G.
Brun, A. Hautfeuille, N. Sabadie, A. Barbin, T. Kuroki, C.
Drevon, C. Piccoli, and R. Montesano.  1980.  Validation and
comparative studies on 180 chemicals with S. typhimurium strains
and V79 Chinese hamster cells in the presence of various
metabolizing systems.  Mutat. Res. 76:1-50.

Bois, F. and R.  Spear.  1991.  Modelling benzene exposure.
Health and Environ. Digest 5:5.

Bond, G.G., E.A. McLaren, C. Baldwin, and R.R. Cook.  1986.  An
update of mortality among chemical workers exposed to benzene.
Br. J. Ind. Med. 43:685-691.

Browning, E.  1965.  Toxicity and metabolism of organic solvents.
Elsevier, Amsterdam.
                               5-59

-------
                                                        EPA-420-R-93-005
                                                            April 1993

California Air Pollution Control Officers Association  (CAPCOA).
1991.  Air Toxics "Hot Spots" Program:  Risk Assessment
Guidelines, January 1991.

California Air Resources Board  (CARB).  1984.  Report  to the
Scientific Review Panel on Benzene.

Carey, P.M.  1987.  Air toxics emissions from motor vehicles.
Ann Arbor, MI:  U.S. Environmental Protection Agency,  Office of
Mobile Sources, EPA Report no. EPA-AA-TSS-PA-86-5.

Chan, C.C., H. Ozkaynak, J.D. Spengler, L. Sheldon, W. Nelson,
and L. Wallace.  1989.  Commuter's exposure to volatile organic
compounds, ozone, carbon monoxide, and nitrogen dioxide.
Prepared for the Air and Waste Management Association.  AWMA
Paper 89-34A.4.

Chen, C., D. Bayliss, and A. Chiu.  1989.  Benzene public
comments.  Memorandum to Jack R. Farmer.  May 5,  1989.

Chevron Oil Company.  1991.  Communication to EPA summarizing the
following studies: "Study to Determine the fate of Benzene
Precursors in Gasoline", NIPER  (under CARB Agreement 150128-32),
1988; "Exhaust Benzene Emissions from Late-Model Vehicles", API
Publication No. 841-44700, 10/88;  "Vehicle Evaporative and
Exhaust Emissions as Influenced by Benzene Content of  Gasoline",
NIPER (Under CRC CAPE-35-83 and U.S. DOE), 4/86.

Clement Associates,  Inc.  1988.  Quantitative re-evaluation of
the human leukemia risk associated with inhalation exposure to
benzene.  Supported by American Petroleum Institute, Chemical
Manufacturers Association, Inc., and Western Oil and Gas
Association.  Final Report.  October 1988.

Clement Associates,  Inc.  1991.  Motor vehicle air toxics health
information.  For U.S. EPA Office of Mobile Sources, Ann Arbor,
MI.  September 1991.

Cole, H. S., D. E. Lapland, G. K. Moss, and C. F. Newberry.
1983.  The St. Louis Ozone Modeling Project.  U. S. Environmental
Protection Agency (EPA-450/4-83-021).

Corti, M. and C. Snyder.  1990.  Long-term hemopoietic effects
caused by in utero exposure to 10 ppm inhaled benzene  and 5%
ingested ethanol.  Toxicologist 10:58.
Crebelli, R., D. Bellincampi, G. Conti, L. Conti, G. Morpurgo,
and A. Carere.  1986.  A comparative study on selective chemical
carcinogens for chromosomal segregation, mitotic crossing-over
and forward mutation induction in Aspergillus nidulans.  Mutat.
Res. 172:139-149.
                               5-60

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Cronin, H.J.  1924.  Benzol poisoning in the rubber industry.
Boston Medical and Science Journal 191:1164-1166.

Cronkite, E.P.,  R.T. Drew, T. Inoue, Y. Hirabayashi, and J.E.
Bullis.  1989.  Hematotoxicity and carcinogenicity of inhaled
benzene.  Environ. Health Perspect.  82:97-108.

Crump K.S. and B.C. Allen.  1984.  Quantitative estimates of risk
of leukemia from occupational exposure to benzene.  Prepared for
the Occupational Safety and Health Administration.

De Flora, S.,  P. Zanacchi, A. Camoirano, C. Bennicelli, and G.S.
Badolati.  1984.  Genotoxic activity and potency for 135
compounds in the Ames reversion test and in a bacterial DNA-
repair test.  Mutat. Res. 133: 161-198.

DHS.   1984.  California Department of Health Services.  Part B.
Health Effects of Benzene.  Epidemiological Studies Section.
November 1984.

Drew, R.T. and J.R. Fouts.  1974.  The lack of effects of
pretreatment with phenobarbital and chlorpromazine on the acute
toxicity of benzene in rats.  Toxicol.  Appl.  Pharmacol.
27:183-193.
Eastmond, D.A.,  M.T. Smith, and R.D. Irons.  1987.  An
interaction of benzene metabolites reproduces the myelotoxicity
observed with benzene exposure.  Toxicol. Appl. Pharmacol. 91:85-
95.

Environ Corporation.  1987.  Risk assessment issues in EPA's
technical report "Air Toxics Emissions from Motor Vehicles".
Prepared for the Motor Vehicle Manufacturers Association.

EPA.   1980.  Ambient water quality criteria document for benzene.
Prepared by the Office of Health and Environmental Assessment,
Environmental Criteria and Assessment Office  (Cincinnati, OH) and
Carcinogen Assessment Group  (Washington, DC),  and the
Environmental Research Labs  (Corvalis,  OR; Duluth, MN; Gulf
Breeze, FL) for the Office of Water Regulations and Standards,
Washington, DC.   EPA 440/5-80-018 .

EPA.   1985.  Interim quantitative cancer unit risk estimates due
to inhalation of benzene.  Prepared by the Office of Health and
Environmental Assessment, Carcinogen Assessment Group,
Washington, DC for the Office of Air Quality Planning and
Standards, Washington, DC.

EPA.   1986a.  Mobile source benzene emissions and a preliminary
estimate of their health impacts.  Memo from Charles L. Gray, Jr.
to Richard D.  Wilson, May 15, 1986.

EPA.   1986b.  The Risk Assessment Guidelines of 1986.  U.S.
Environmental Protection Agency, Office of Health and
Environmental Assessment, Office of Research and Development.
Washington, DC.   EPA/600/8-87/045.
                               5-61

-------
                                                        EPA-420-R-93-005
                                                            April 1993

EPA.  1987a.  Draft Regulatory Impact Analysis: Control of
Gasoline Volatility and Evaporative Hydrocarbon Emissions from
New Motor Vehicles.  Ann Arbor, Michigan: Office of Mobile
Sources.  July, 1987.

EPA. 1987b.  The Total Exposure Assessment Methodology  (TEAM)
Study:  Summary and Analysis:  Volume I.  Office of Research and
Development, Washington, D.C.  June 1987.  EPA Report No.
EPA/600/6-87/002a.

EPA.  1988.  Locating and estimating air emissions from sources
of benzene.  Office of Air Quality Planning and Standards,
Research Triangle Park, NC:   EPA-450/4-84-007q.

EPA.  1989.  AIRS user's guide volumes I-VII.  U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards.
Research Triangle Park, N.C.

EPA.  1990a.  VOC/PM Speciation Data System, version 1.32a
(published in electronic form).  Research Triangle Park, North
Carolina: Office of Air Quality Planning and Standards.

EPA.  1990b.  Air quality criteria for carbon monoxide.  Office
of Health and Environmental Assessment.  EPA/600/8-90/045A.
External review draft.  March 1990.

EPA.  1991a.  Regulation of Fuel and Fuel Additives: Standards
for Reformulated Gasoline; Proposed Rule.  Federal Register
56(131): 31176-31263.

EPA.  1991b.  Nonroad Engine and Vehicle Emission Study.  Office
of Air and Radiation, Office of Mobile Sources, Ann Arbor, MI.
November 1991.  EPA Report No. 21A-2001.

EPA. 1992a.  Control of Air Pollution From New Motor Vehicles and
New Motor Vehicle Engines; Refueling Emission Regulations for
Gasoline-Fueled Light-Duty Vehicles and Trucks and Heavy-Duty
Vehicles; Proposed Rule.  Federal Register 57 (73) :13220-13231.

EPA.  1992b.  Regulation of Fuel and Fuel Additives: Standards
for Reformulated and Conventional Gasoline; Proposed Rule.
Federal Register 57 (74) :13416-13495 .

EPA.  1992c.  Integrated Risk Information System.  U.S.
Environmental Protection Agency.  Office of Health and
Environmental Assessment, Environmental Criteria and Assessment
Office,  Cincinnati, OH.

General Motors Corporation.   1991.  Communication from S. R.
Reddy to Chris Lindjhem, April 16, 1991.

Gerarde, H.W.  1960.  Toxicology and biochemistry of aromatic
hydrocarbons.  London, England:  Elsevier Publishing Co., 44-46.

Glatt, H. and G. Witz.  1990.  Studies on the induction of gene
mutations in bacterial and mammalian cells by the ring-opened


                               5-62

-------
                                                        EPA-420-R-93-005
                                                            April 1993

benzene metabolites trans, trarzs-muconaldehyde and trans, trans-
muconic acid.  Mutagenesis 5:263-266.

Glatt, H., R. Padykula, G.A. Berchtold, G. Ludewig, K.L. Platt,
J. Klein, and F. Oesch.  1989.  Multiple activation pathways of
benzene leading to products with varying genotoxic
characteristics.  Environ. Health Perspect. 82:81-89.

Goldstein B.D., C.A. Snyder, S. Laskin, I. Bromberg, R.E. Albert,
and N. Nelson.  1980.  Myelogenous leukemia in rodents inhaling
benzene.  Unpublished data.   (Cited in EPA, 1985)

Hogo, H. and M. W. Gery.  1988.  "User's Guide for Executing
OZIPM-4 with CBM-IV or Optional Mechanisms."  U. S. Environmental
Protection Agency.   (EPA-600/8-88/082).

Huff, J.E.  1986.  Toxicology and carcinogenesis studies of
benzene  (CAS No. 71-43-2) in F344/N rats and B6C3F1 mice  (gavage
studies).  Technical report no. 289.  DHHS, National Toxicology
Program/National Institute of Environmental Health Sciences,
Research Triangle Park, NC.

Huff, J.E., J.K. Haseman, D.M. DeMarini, S. Eustis, R.R.
Maronpot, A.C. Peters, R.L. Persing, C.E. Chrisp, and A.C.
Jacobs.  1989.  Multiple-site carcinogenicity of benzene in
Fischer 344 rats and B6C3F1 mice.  Environ. Health Perspect.
82:125-163.

IARC.  1982.  IARC monographs on the evaluation of carcinogenic
risk of chemicals to humans.  Volume 29.  Some industrial
chemicals and dyestuffs.  International Agency for Research on
Cancer.  World Health Organization, Lyon, France.  p. 345-389.

IARC.  1987.  IARC monographs on the evaluation of carcinogenic
risk of chemicals to humans.  Supplement 7.  Overall evaluations
of carcinogenicity:  An updating of IARC monographs volumes 1 to
42.  International Agency for Research on Cancer.  World Health
Organization, Lyon, France.  p. 120-122.


Infante, P.F., R.A. Rinsky, J.K. Wagoner and R.J. Young.  1977a.
Benzene and leukemia.  The Lancet  2(8043): 867-869.

Infante, P.F., R.A. Rinsky, J.K. Wagoner and R.J. Young.  1977b.
Leukemia in benzene workers.  Lancet  19: 76-78.

Ireson, R. G., E. L. Carr, J. L. Fieber, J. G. Heicken, T. C.
Myers, G. M. Wilson, and G. Z. Whitten.  1990.  "Urban
formaldehyde and methanol concentrations for alternative methanol
vehicle scenarios."  Systems Applications International, San
Rafael, California  (SYSAPP-90/104).

Irons, R.D., W.S. Stillman, D.B. Colagiovanni, and V.A. Henry.
1992.  Synergistic action of the benzene metabolite hydroquinone
on myelopoietic stimulating activity of granulocyte/macrophage
                               5-63

-------
                                                        EPA-420-R-93-005
                                                            April 1993

colony-stimulating factor in vitro.  Proc. Natl. Acad. Sci.
89:3691-3695.

Kale, P.G. and J.W. Baum.  1983.  Genetic effects of benzene in
Drosophila melanogaster males.  Environ. Mutagen. 5:223-226.

Keller, K.A. and C.A. Snyder.  1986.  Mice exposed in utero to
low concentrations of benzene exhibit enduring changes in their
colony forming hematopoietic cells.  Toxicology 42:171-181.

Kimmel, C.A. and J.G. Wilson.  1973.  Skeletal deviations in
rats:  Malformations or variations?  Teratology 8:309-316.

Kissling, M. and B. Speck.  1973.  Chromosome aberrations in
experimental benzene intoxication.  HELV.  Med. Acta.  36: 59-66.

Lange, A., R. Smolik, W. Zatonski, and H. Glazman.  1973a.
Leukocyte agglutinins in workers exposed to benzene, toluene, and
xylene.  Int. Arch. Arbeitsmed. 31:45-40.

Lange, A., R. Smolik, W. Zatonski, and J. Szymanska.  1973b.
Serum immunoglobulin levels in workers exposed to benzene,
toluene,  and xylene.  Int. Arch. Arbeitsmed. 31:37-44.

Latriano, L., B.D. Goldstein, and G. Witz G.  1986.  Formation of
muconaldehyde, an open-ring metabolite of benzene, in mouse liver
microsomes.  An additional pathway for toxic metabolites.  Proc.
Natl. Acad. Sci (U.S.)  83:8356-8360.

Latriano, L., G. Witz,  B.D. Goldstein, and A.M. Jeffrey.  1989.
Chromatographic and Spectrophotometric characterization of
adducts formed during the reaction of trans, trarzs-muconaldehyde
with 14C-deoxyguanosine 5'-phosphate.  Environ. Health Perspect.
82 :249-251.

Lebowitz, H., D. Brusick, D. Matheson, D.R. Jagannath, M. Reed,
S. Goode, and G. Roy.  1979.  Commonly used fuels and solvents
evaluated in a battery of short-term bioassays.  Environ.
Mutagen.  1:172-173.

Lee, E.W., C.D. Garner, and J.T. Johnson.  1988.  A proposed role
played by benzene itself in the induction of acute cytopenia:
Inhibition of DNA synthesis.  Res. Commun. Chem. Pathol.
Pharmacol.  60: 27-46.

Lee, E.W., J.T. Johnson, and C.D. Garner.  1989.  Inhibitory
effect of benzene metabolites on nuclear DNA synthesis in bone
marrow cells.  J.  Toxicol. Environ. Health 26:277-292.

Ligocki,  M. P., and G.  Z. Whitten.  1991.  Atmospheric
transformation of air toxics: Acetaldehyde and Polycyclic Organic
Matter.  Systems Applications International, San Rafael,
California  (SYSAPP-91/113).

Ligocki,  M.P., G.Z. Whitten, R.R. Schulhof, M.C. Causley, and
G.M. Smylie.  1991.  Atmospheric transformation of air toxics:


                               5-64

-------
                                                        EPA-420-R-93-005
                                                            April 1993

benzene, 1,3-butadiene, and formaldehyde.  Systems Applications
International, San Rafael, California  (SYSAPP-91/106).

Ligocki, M.P., R.R. Schulhof, R.E. Jackson, M.M. Jimenez, G. Z.
Whitten, G.M. Wilson, T.C. Meyers, and J.L. Fieber.   1992.
Modeling the effects of reformulated gasolines on ozone and
toxics concenetrations in the Baltimore and Houston areas.
Systems Applications International, San Rafael, California
(SYSAPP-92/127).

Lumley, M., H. Barker, and J.A. Murray.  1990.  Benzene in
petrol.  Lancet 336:1318-1319.

Maltoni, C. and C. Scarnato.  1979.  First experimental
demonstration of the carcinogenic effects of benzene.  Long-term
bioassays on Sprague-Dawley Rats by oral administration.  Med.
Lav. 70: 352-357.

Maltoni, C., B. Conti and G. Cotti.  1983.  Benzene:  A
multipotential carcinogen.  Results of long-term bioassays
performed at the Bologna Institute of Oncology.  Am.  J. Ind. Med.
4:589-630.

Maltoni, C., A. Ciliberti, G. Cotti, B. Conti, and F. Belpoggi.
1989.  Benzene, an experimental multipotential carcinogen:
Results of the long-term bioassays performed at the Bologna
Institute of Oncology.  Environ. Health Perspect. 82:109-124.

McAllister, R. A., W. H. Moore, J. Rice, E. Bowles, D. P. Dayton,
R. F. Jongleux, R. G. Merrill, Jr., and J. T. Bursey.  1990.
Urban Air Toxics Monitoring Program.  U. S. Environmental
Protection Agency  (EPA-450/4-91-001).

McMahon, T.F. and L.S. Birnbaum.  1991.  Age-related  changes in
disposition and metabolism of benzene in male C57BL/6N mice.  In
press.

Medinsky, M.A., P.J. Sabourin, R.F. Henderson, G. Lucier, and
L.S. Birnbaum.  1989.  Differences in the pathways for metabolism
of benzene in rats and mice simulated by a physiological model.
Environ. Health Perspect. 82:43-49.

Meyne,  J. and M.S. Legator.  1980.  Sex-related differences in
cytogenic effects of benzene in the bone marrow of Swiss mice.
Environ. Mutat.  2:43-50.

Morris, J.J. and E. Seifter.  1992.  The role of aromatic
hydrocarbons in the genesis of breast cancer.  Medical Hypotheses
38:177-184.

Morris, R.  E., T.  C. Myers, H. Hogo, L. R. Chinkin, L. A.
Gardner, R. G. Johnson.  1989.  A low-cost application of the
Urban Airshed Model to the New York metropolitan area and the
city of St. Louis  (Five Cities UAM Study Phase I).  Systems
Applications International, San Rafael, California  (SYSAPP-
89/070) .


                               5-65

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Murray, F.J., J.A. John, L.W. Rampy, R.A. Kuna, and B.A. Schwetz.
1979.  Embryotoxicity of inhaled benzene in mice and rabbits.
Am. Ind. Hyg. Ass. J. 40:993-998.

Nestmann, E.R., E.G. Lee, T.I. Matula, G.R. Douglas, and J.C.
Mueller.  1980.  Mutagenicity of constituents identified in pulp
and paper mill effluents using the Salmonella/mammalian-microsome
assay.  Mutat. Res. 79:203-212.

Nielsen, T., T. Ramdahl, and A. Bj0rseth.  1983.  The fate of
airborne polycyclic organic matter.  Environ. Health Perspect.,
47:103-114.

NTP.  1983.  NTP technical report on the toxicology and
carcinogenicity studies of benzene  (CAS No. 71-43-2) in F344/N
rats and B6C3F-L mice (gavage  studies) ,  NIH Publication No.  84-
2545.  Research Triangle Park, NC.  Board draft, July 1984.

NTP.  1986.  Toxicology and carcinogenesis studies of benzene
(CAS No. 71-43-2)  in F344/N rats and B6C3F mice (gavage studies).
NTP Technical Report Series No. 289.  NIH Publication No. 86-
2545.

Nylander, P.O., H. Olofsson,  B. Rasmuson, and H. Svahlin.  1978.
Mutagenic effects of petrol in Drosophila melanogaster.  I.
Effects of benzene and 1,2-dichloroethane.  Mutat. Res. 57:163-
167.

Oberly, T.J., B.J. Bewsey, and G.S. Probst.  1984.  An evaluation
of the L5178Y TK+/- mouse lymphoma forward mutation assay using
42 chemicals.  Mutat. Res. 125:291-306.

OSHA.  1987.  Occupational exposure to benzene.  Final Rule.
U.S. Department of Labor, Occupational Safety and Health
Administration.  Federal Register 52: 34460-34578.

Ott, M.G., J.C. Townsend, W.A. Fishbeck, and R.A. Langner.  1978.
Mortality among individuals occupationally exposed to benzene.
Arch. Environ. Health 33:3-10.

Peterson, H.D.  1986.  Case-control analysis of Ott et al. cohort
(letter).  OSHA Docket H-059C, Exhibit 247, Tab D, Attachment 2.

Pezda, S.A., J.J. Vostal, and H.J. Wimette. 1991. Current and
Future Research Needs in Health Effects of Air Toxics - An
Automotive Industry Perspective.  Presented at the Air and Waste
Management Association Conference, October 16-18, 1991, Detroit,
Michigan.

Possiel, N.C., D.C. Doll, K.A. Baugues, E.W. Baldridge, and R.A.
Wayland.  1990.  Impact of regional control strategies on ozone
in the Northeastern United States.  Prepared for the Air and
Waste Management Association.  AWMA Paper 90-93.3.
                               5-66

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Reddy, M.V., G.R. Blackburn, C.A. Schreiner, M.A. Mehlman, and
C.R. Mackerer.  1989.  32P analysis of DNA adducts in  tissues  of
benzene-treated rats.  Environ. Health Perspect. 82:253-257.

Rinsky, R.A., A.B. Smith, R. Hornung, T.G. Filloon,  R.J. Young,
A.H. Okun, and P.J. Landrigan.  1987.  Benzene and leukemia:  an
epidemiologic risk assessment.  N. Eng. J. Med. 316:1044-1050.

Rinsky R.A., R.J. Young and A.B. Smith.  1981.  Leukemia in
benzene workers.  Am. J. Ind. Med.  2: 217-245.

Robinson, J.P, J.A. Wiley, T. Piazza, and K. Garett.  1989.
Activity Patterns of California Residents and Their Implications
for Potential Exposure to Pollution.  Draft final report.
CARB-46-177-33.   California Air Resources Board.  Sacramento, CA.

Rosenthal, G.J.  and C.A. Snyder.  1985.  Modulation of the immune
response to Listeria monocytogenes by benzene inhalation
[Abstract].  Toxicol. Appl.  Pharmacol. 80:502-510.

Rozen, M.G., C.A. Snyder, and R.E. Albert.  1984.  Depression in
B- and T-lymphocyte mitogen-induced blastogenesis in  mice exposed
to low concentrations of benzene.  Toxicol. Lett. 20:343-349.

Sabourin, P.J.,  B.T. Chen, G. Lucier, L.S. Birnbaum,  E. Fisher,
and R.F. Henderson.  1987.  Effect of dose on the absorption  and
excretion of  [14C] benzene administered orally or by inhalation in
rats and mice.  Toxicol. Appl. Pharmacol. 87:325-336.

Sabourin, P.J.,  W.E. Bechtold, L.S. Birnbaum, G.W. Lucier, and
R.F. Henderson.   1988.  Difference in the metabolism  of inhaled
3H-benzene by F344/N rats and B6C3F1  mice.   Toxicol.  Appl.
Pharmacol. 94:128-140.

Savitz, D.A., E.A. Whelan, and R.C. Kleckner.  1989.  Effect  of
parents' occupational exposures on risk of stillbirth, preterm
delivery, and small-for-gestational-age infants.  American
Journal of Epidemiology 129:1201-1218.

Schere, K. L., and J. H. Sheffler.  1982.  Final Evaluation of
Urban-Scale Photochemical Air Quality Simulation Models.
Environmental Sciences Research Laboratory, U. S. Environmental
Protection Agency  (EPA-600/3-82-094).

Shahin, M.M. and F. Fournier.  1978.  Suppression of  mutation
induction and failure to detect mutagenic activity with Athabasca
tar sand fractions.  Mutat.  Res. 58:29-34.

Shikiya, D.C., C.S. Liu, M.I. Kahn, J. Juarros, and W.
Barcikowski.  1989.  In-vehicle air toxics characterization study
in the south coast air basin.  South Coast Air Quality Management
District, El Monte, CA.  May, 1989.

Shimizu, M., Y.  Yasui, and N. Matsumoto.  1983.  Structural
specificity of aromatic compounds with special reference to
                               5-67

-------
                                                        EPA-420-R-93-005
                                                            April 1993

mutagenic activity in Salmonella typhimurium: a series of chloro-
or fluoro-nitrobenzene derivatives.  Mutat. Res. 116:217-238.

Singh, H. B. and J. F. Kasting.  1988.  Chlorine-hydrocarbon
photochemistry in the marine troposphere and lower stratosphere.
J. Atmos. Chem.,  7:261-285.

Snyder, R., E. Dimitriadis, R. Guy, P. Hu, K. Cooper, H. Bauer,
G. Witz, and B.D. Goldstein.  1989.  Studies on the mechanism of
benzene toxicity.  Environ. Health Perspect. 82:31-35.

Tanooka, H.  1977.  Development and application of Bacillus
subtilis test systems for mutagens, involving DNA-repair
deficiency and suppressible auxotrophic mutations.  Mutat. Res.
42: 19-32.

Travis, C.C., J.L. Quillen, and A.D. Arms.  1990.
Pharmacokinetics of benzene.  Toxicol. Appl. Pharmacol. 102:400-
420.

Tuazon, E. C., H. MacLeod, R. Atkinson, and W.P.L. Carter.  1986.
a-Dicarbonyl yields from the N0x-air photooxidations  of  a series
of aromatic hydrocarbons in air.  Environ. Sci. Technol., 20:383-
387.

Vara,  P. and 0. Kinnunen.  1946.  Benzene toxicity as a
gynecologic problem.  Acta. Obstet. Gynecol. Scand. 26:433-452.

Wallace, L.A.  1989.  Major sources of benzene exposure.
Environ. Health Perspect. 82:165-169.

Ward,  C.O., R.A.  Kuna, N.K. Snyder, et al.  1985.  Subchronic
inhalation toxicity of benzene in rats and mice.  Am. J. Ind.
Med. 7:457-473.

Whitten, G. Z.  1983.  The chemistry of smog formation:  A review
of current knowledge.  Environ. International, 9:447-463.

Wierda, D., A.G.  King, R.W. Luebke, M.J. Reasor, and R.J.
Smialowicz.  1989.  Perinatal immunotoxicity of benzene toward
mouse B cell development.  Journal of the American College of
Toxicology 8:981-996.

Wilson, A. et al.  1991.  Air toxics microenvironment exposure
and monitoring study.  South Coast Air Quality Management
District, El Monte, CA.

Winek, C.L. and W.D. Collom.  1971.  Benzene and toluene
fatalities.  J. Occup. Med. 13:259-261.

Witz,  G., W. Maniara, and V. Mylavarapu.  1990.  Comparative
metabolism of benzene and trans, trarzs-muconaldehyde to
trans, trarzs-muconic acid in DBA/2N and C57BL/6 mice.  Biochem.
Pharmacol. 40:1275-1280.
                               5-68

-------
                                                        EPA-420-R-93-005
                                                             April 1993

Wong, 0.  1987.  An industry-wide mortality study of chemical
workers occupationally exposed  to benzene.   II.  Dose-response
analyses.  Br. J. Ind. Med. 44:382-395.

Wong, 0., R.W. Morgan and M.D.  Whorton.   1983.   Comments  on the
NIOSH study of leukemia in benzene workers.   Technical report
submitted to Gulf Canada, Ltd., by Environmental Health
Associates.

Yin, S-N., G-L. Li, F-D. Tain,  Z-I.  Fu,  C.  Jin,  Y-J. Chen,  S-J.
Luo, P-Z. Ye, J-Z. Zhang, G-C.  Wang,  X-C.  Zhan,  H-N. Wu,  and Q-C.
Zhong.  1989.  A retrospective  cohort study of  leukemia and other
cancers in benzene workers.  Environ.  Health Perspect. 82:207-
213.
                               5-69

-------
                                                        EPA-420-R-93-005
                                                            April 1993

6.0  FORMALDEHYDE

6.1 Chemical and Physical Properties  (EPA, 1991a,  1992)

     Formaldehyde is a colorless gas at normal temperatures with
a pungent, irritating odor.  It is the simplest member of  the
family of aldehydes and has the chemical  formula HCHO.
Formaldehyde gas is soluble in water, alcohols, and other  polar
solvents.  The chemical and physical properties of pure
formaldehyde are presented in Table 6-1.

     In the presence of air and moisture  at room temperature,
formaldehyde readily polymerizes to a solid mixture known  as
paraformaldehyde.   Another common form of formaldehyde is  its
cyclic trimer  (three formaldehyde molecules forming a ring) known
as trioxane (C3H603) .   In aqueous solutions,  formaldehyde reacts
with water to form methylene glycol.

     Pure, dry formaldehyde gas is stable from 25-100C  (77-
212F)  and decomposes very slowly up to 300C (572F).
Polymerization takes place slowly below room temperature but is
accelerated by the presence of impurities.  Decomposition  of
formaldehyde produces carbon monoxide and hydrogen gas.  When
catalyzed by certain metals (platinum, copper, or  chromia  and
alumina), formaldehyde decomposition can produce methanol, methyl
formate, formic acid, carbon dioxide, and methane.
Table 6-1.  Chemical and Physical Properties of Pure
Formaldehyde.
Properties
Molecular weight
Melting point
Boiling point at 1 atm.
Density at -20C (-4F)
Vapor pressure at -19.5C
Flash point
Solubility in water at 25C
Conversions
Values
30.03 g/mole
-92.0C (-133.4F)a
-19.5C (-3.1F)
0.8153 g/ml
1 atm.
60C (140F) at a 40% solution
very soluble (up to 55
0, \
"0 1
1 ppm = 1.23 mg/m3
                               6-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993
6.2 Formation and Control Technology
     Formaldehyde is the most prevalent aldehyde in vehicle
exhaust and is formed from incomplete combustion of the fuel.
Formaldehyde is emitted in the exhaust of both gasoline and
diesel-fueled vehicles.  It is not a component of evaporative
emissions.

     Use of a catalyst has been found to be effective for
controlling formaldehyde emissions.  Formaldehyde emissions are
controlled to roughly the same extent as total hydrocarbon
emissions with a catalyst (Carey, 1987).


6.3  Emissions

6.3.1  Emission Fractions Used in the MOBTOX Emissions Model

     Emission fractions for formaldehyde were developed using
vehicle emission test data from various programs (Appendix B2).
Formaldehyde emission fractions for different components included
in the scenarios are included in Appendix B6.

     The formaldehyde TOG emission fraction for LDGVs/LDGTs with
three-way catalysts, running on baseline fuel, was based on data
from 38 vehicles tested in four studies (Boekhaus et al., 1991a,
1991b, DeJovine et al., 1991, and Auto/Oil, 1990).   The TOG
fraction for LDGVs/LDGTs with three-way plus oxidation catalysts,
running on baseline fuel, was based on data from 25 vehicles
tested in eight studies  (Urban, 1980a, 1980b, Sigsby et al.,
1987, Stump et al.,  1989, 1990, unpublished, Warner-Selph and
DeVita, 1989, Boekhaus et al.,  1991b, Auto/Oil, 1990).  The TOG
fraction for LDGVs/LDGTs with oxidation catalysts,  running on
baseline fuel, was based on data from 41 vehicles tested in eight
studies (Urban, 1980a,  Springer, 1979, Sigsby et al., 1987,
Smith, 1981, Stump et al.,  1989, 1990, Auto/Oil, 1990, Boekhaus
et al., 1991a, Warner-Selph and Smith, 1991).  The TOG fraction
for LDGVs/LDGTs with no catalysts, running on baseline fuel, was
based on data from 11 vehicles tested in four studies (Urban,
1981, Urban 1980a,  Sigsby et al., 1987, and Warner-Selph and
Smith, 1991).  The LDDV fraction was based on data from 7
vehicles tested in two studies  (Springer,  1977 and Springer,
1979).  The HDDV and HDGV non-catalyst fractions were based on
13-mode data from two engines and one engine, respectively,
tested in one study (Springer,  1979).  To estimate the three-way
fraction for HDGVs,  the non-catalyst to three-way fraction for
LDGVs/LDGTs was applied to the HDGV non-catalyst fraction.

     To calculate TOG fractions for vehicles running on MTBE
blends and 10% ethanol, adjustment factors were applied to the
baseline emission fractions for each vehicle class/catalyst
combination based on average percent change.  The average percent
change
                               6-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

numbers for vehicle class/catalyst combinations are contained in
Appendix B4.

     It should be noted that percent change was calculated on a
vehicle by vehicle basis and the average of these percent changes
was then calculated for each vehicle class/catalyst combination.
When a draft memo was distributed by EPA describing the
methodology used to calculate emission fractions for this report
(EPA, 1992a),  a comment was made in a review prepared by Systems
Applications International for the Motor Vehicle Manufacturers
Association (MVMA) (Ligocki, 1992) questioning this averaging
approach.  The review pointed out that if a car had low total
mass emissions, but a large change in percent of a toxic, this
could result in an overestimate of the effect of this car on the
toxic level in the fleet, and an overestimate of the toxic level
in reformulated fuel relative to baseline.  However, the
potential source of error resulting from this averaging technique
is diminished by a number of factors, including the fact that
data from cars exhibiting unreasonably large changes in toxic
levels were discarded.  Also, the potential source of error would
be expected to affect only formaldehyde and acetaldehyde, since
benzene fractions were calculated from equations, and oxygenate
level has little effect on 1,3-butadiene.  In any case, the MVMA
approach is not appreciably more accurate than the EPA approach
in predicting actual toxic fractions.

     The 15% MTBE and 10% ethanol adjustment factors for
LDGVs/LDGTs with various catalyst technologies are summarized in
Table 6-2.  Note that use of oxygenated fuels increases
formaldehyde emissions for all catalyst technologies.  These 15%
MTBE numbers were estimated using data from Auto/Oil (1991) and
DeJovine et al. (1991) for LDGVs/LDGTs with three-way catalysts,
Auto/Oil  (1991) for LDGVs/LDGTs with three-way plus oxidation and
oxidation catalysts,  and Warner-Selph and Smith  (1991)  for
vehicles
Table 6-2.
15% MTBE and 10% Ethanol Emission Fraction Adjustment
        Factors for Formaldehyde.
Vehicle
Class
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
Catalyst
Technology
3 -way
3 -way + ox
oxidation
non-cat
15% MTBE
Adjustment
Factor
1.6746
1.2672
2 .0244
1.5256
10%
Ethanol
Adjustment
Factor
1.4758
1.2288
1.2400
1.1034
with no catalysts.  The 10% ethanol numbers were estimated using
data from Auto/Oil  (1991),  Warner-Selph and Smith  (1991) and the
Colorado Department of Health  (1987) for LDGVs/LDGTs with three-
                               6-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993

way catalysts, the Colorado Department of Health  (1987) for
LDGVs/LDGTs with three-way plus oxidation catalysts, and Warner-
Selph and Smith (1991) and the Colorado Department of Health
(1987)  for LDGVs/LDGTs with oxidation catalysts or no catalysts.

     Since the average percent change was calculated for 15% MTBE
(2.7% weight percent oxygen), and 11.0% MTBE  (2.0% oxygen) was
assumed for reformulated fuel and California standards
components, average percent changes in the formaldehyde TOG
fraction from 0 to 15% MTBE  were multiplied by 2.0/2.7, the
ratio of oxygen contents by weight.  For HDGVs with three-way
catalysts and with no catalysts, the same 15% MTBE and 10%
ethanol adjustment factors were assumed as for LDGVs/LDGTs with
the same catalyst technologies.

6.3.2  Emission Factors for Baseline and Control Scenarios

     The fleet average formaldehyde emission factors as
determined by the MOBTOX emissions model are presented in Table
6-3.  When comparing the base control scenarios relative to
1990,the emission factor is reduced by 43% in 1995, by 61% in
2000, and by 66% in 2010.  The expansion of reformulated fuel use
in 1995 actually increases the emission factor, resulting in a
39% reduction relative to 1990.  In 2000, the expanded control
scenarios increase the emission factor slightly, when compared to
the 2000 base control.  In 2010, there is similarly little or no
change from the 2010 base control for the expanded control
scenarios.

6.3.3  Nationwide Motor Vehicle Formaldehyde Emissions

     The nationwide formaldehyde metric tons are presented in
Table 6-4.   Total metric tons are determined by multiplying the
emission factor (g/mile) by the VMT determined for the particular
year.  The VMT, in billion miles, was determined to be 1793.07
for 1990, 2029.74 for 1995, 2269.25 for 2000, and 2771.30 for
2010.  When comparing the base control scenarios relative to
1990, the metric tons are reduced by 36% in 1995 and by 45% in
2000.  Even though the emission factor continues to decrease from
2000 to 2010, this is more than offset by the large increase in
VMT.  As a result, metric tons in 2010 actually increase relative
to 2000.

6.3.4 Other Sources of Formaldehyde

     The onroad motor vehicle contribution to ambient
formaldehyde levels contains both direct (primary) and secondary
formaldehyde formed from photooxidation of VOC.  It appears that
roughly 33% of formaldehyde emissions may be attributable to
motor vehicles.  Section 6.5.2 contains a complete explanation of
how this number is determined.
                               6-4

-------
                                                            EPA-420-R-93-005
                                                                 April 1993
Table  6-3.   Annual  Emission Factor Projections for  Formaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0412
0.0234
0.0251
0.0162
0.0166
0.0168
0.0140
0.0143
0.0138
Percent
Reduction
from 1990
-
43
39
61
60
59
66
65
67
                                  6-5

-------
                                                             EPA-420-R-93-005
                                                                 April 1993
Table  6-4.   Nationwide Metric  Tons Projection for Formaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0412
0.0234
0.0251
0.0162
0.0166
0.0168
0.0140
0.0143
0.0138
Metric
Tons
73,874
47,496
50, 946
36,762
37,670
38,123
38,798
39,630
38,244
                                  6-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Formaldehyde is produced in the U.S. by 13 chemical
companies in 46 locations encompassing 18 states  (EPA, 1991a).
Formaldehyde is used in the manufacture of four major types of
resins: urea-formaldehyde, melamine-formaldehyde, phenol-
formaldehyde,  and polyacetal resins.  These resins are used in a
wide variety of products, such as plywood, particle board, and
counter tops.   Formaldehyde is also used as a raw material in
several synthetic organic chemical production processes, in the
production of solid urea  (used as a fertilizer, a protein
supplement for animal feed, and in plastics),  and in the
production of ureaform fertilizers.

     In addition, formaldehyde is produced as a by-product in the
following types of processes:  combustion (mobile, stationary,
and natural sources),  petroleum refinery catalytic cracking and
coking, phthalic anhydride production, asphaltic concrete
production, and atmospheric photooxidation of unburned
hydrocarbons.

     In an attempt to determine the effects of actual
formaldehyde exposure,  an analysis was conducted by the EPA
Office of Mobile Sources  (EPA, 1987a)  to determine the cancer
risk attributable to indoor and outdoor sources of formaldehyde.
The analysis consisted of three parts: (1) estimation of the U.S.
population distribution and amount of time spent in each of
several environments,  (2) estimation of the formaldehyde
concentrations in the various environments,  and  (3) estimation of
unit risks.

     This analysis determined that the largest single source of
risk is the home environment, which accounted for 60 percent of
the most likely number of malignant and benign tumors.  The
uncertainty in the formaldehyde concentrations experienced in
this environment and the entire analysis is high. This is not an
unexpected result as the nation, on average, spends nearly two-
thirds of its time in non-mobile homes.  The high exposure
scenarios  (mobile homes, high office exposures, high industrial
exposures) which had high concentrations, accounted for just
slightly more than ten percent of all tumors due to the small
population involved.  Mobile sources were calculated to only
account for two to six percent of the total risk.


6.4 Atmospheric Reactivity and Residence Times

6.4.1 Gas-Phase Chemistry of Formaldehyde

     As a result of its structure, formaldehyde has a high degree
of chemical reactivity and good thermal stability.  Formaldehyde
is thus capable of undergoing a wide variety of chemical
reactions.  The major mechanisms of destruction in the atmosphere
are reaction with hydroxyl radicals and photolysis.
                               6-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Formaldehyde is present in emissions but is also formed by
the atmospheric oxidation of virtually all reactive organic
species.  As a result, it is ubiquitous in the atmosphere.

     The processes involved in transformation and residence times
were previously discussed in Section 5.4 with the same
information concerning benzene.  For a more detailed explanation
of the various parameters involved in these processes, please
refer to Section 5.4.  The information that follows on
transformation and residence times has been mainly excerpted from
a report produced by Systems Applications International for the
EPA (Ligocki et al. ,  1991).

     Since formaldehyde is formed by the oxidation of methane and
biogenic hydrocarbons, it is ubiquitous in the atmosphere.  The
chemical system of NO and formaldehyde is the minimum system
needed to generate urban-like photochemical ozone in air.  This
property has lead to the use of formaldehyde/NO smog chamber
experiments for testing the inorganic reactions needed in smog
mechanisms.  On a per-carbon basis, formaldehyde has also been
identified as the most important smog precursor in urban
atmospheres (Smylie et al., 1990).  Furthermore, formaldehyde is
perhaps the most common secondary product from the atmospheric
oxidation of all organic compounds.

6.4.1.1 Formation

     Formaldehyde is formed from the atmospheric oxidation of
many types of natural and anthropogenic (human produced) organic
compounds.  In remote areas, the slow oxidation of methane and
the rapid oxidation of biogenic hydrocarbons such as isoprene
produces a background concentration of about 0.6 ppb of
formaldehyde during daylight hours (NRC, 1981).  In urban areas,
the oxidation of olef ins such as ethene (C2H4) and propene  (C3H6) ,
and aromatics, such as toluene and xylene, produce formaldehyde.
Dodge (1990) showed that the most important precursors for
formaldehyde production are ethene, olefins, and higher
aldehydes.  Production of formaldehyde in the reaction of ethene
with OH is particularly efficient because each mole of ethene
reacts to produce 1.56 moles of formaldehyde.  The atmospheric
oxidation of methanol also produces formaldehyde.

6.4.1.2 Gas Phase Reactions

     The reactions of formaldehyde with the OH radicals are
responsible for a part of the destruction of formaldehyde in the
atmosphere, while the reactions with H02,  oxygen atoms,  03, and
Cl are not important in the ambient atmosphere.

     An important destruction and radical production pathway is
found in the photolysis of formaldehyde in the atmosphere.  Three
factors determine the rate of photolysis of a chemical species in
the atmosphere (Jeffries and Sexton,  1987) .  The first factor is
the amount of sunlight of a particular wavelength passing through
the atmosphere at a given time.  The second factor is the ability


                               6-8

-------
                                                        EPA-420-R-93-005
                                                            April 1993

of the chemical to absorb radiation.  The third factor is the
tendency of the molecule to form a particular set of products
after it has absorbed a photon.  The product of these three
factors, integrated over the range of wavelengths of light
present in the atmosphere, determines the photolysis rate for a
given reaction.

     A key property of formaldehyde photochemistry is its
photolysis to form radical products.  Under many conditions, the
radicals from formaldehyde photolysis are the most important net
source of smog generation.  In addition, these radicals determine
the chemical residence time of other toxic species.  Formaldehyde
absorbs UV radiation from below 290 nm to about 340 nm.  Two
pathways of photolysis are widely recognized:  one pathway
produces two relatively stable products, molecular hydrogen  (H2)
and carbon monoxide (CO),  whereas the other pathway produces two
radicals, the formyl radical (HCO) and a hydrogen atom  (H).  Both
of these radicals react quickly with atmospheric oxygen  (02)  to
give hydroperoxyl radicals  (H02)  and CO.

6.4.1.3 Reaction Products

     The oxidation of formaldehyde by OH proceeds primarily by
H-atom abstraction, forming an HCO radical which rapidly reacts
with atmospheric 02 to form CO  and H02 radicals.  Production  of
formic acid (HCOOH) in the HCHO + OH reaction has been measured
and found to account for only 2 percent of the product yield
(Yetter et al. , 1989)  .  The HCHO + H02 reaction does produce
formic acid; however,  the rapid back-reaction precludes this from
being a major formaldehyde transformation pathway.  Therefore,
the dominant carbon-containing product from all atmospheric
formaldehyde reactions, including both photolysis pathways,  is
carbon monoxide.

6.4.2 Aqueous Phase Chemistry of Formaldehyde

     In contrast to benzene and 1,3-butadiene, formaldehyde  is
quite soluble in water because it rapidly hydrates in solution to
form a glycol  (CH2(OH)2).  Formaldehyde is readily incorporated
into clouds and rain,  and is an important species in cloud
chemistry.  The product of the aqueous-phase oxidation of
formaldehyde is formic acid.

     Formaldehyde is also interesting because of its
participation in sulfur chemistry within clouds.  Aqueous
formaldehyde reacts with aqueous S02 (S(IV))  to form the stable
adduct hydroxymethanesulfonate (HMS)  (Hunger et al., 1984).  This
reaction has been proposed to stabilize aqueous S(IV) against
oxidation to sulfate  (McArdle and Hoffmann,  1983).
                               6-9

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Formaldehyde is formed in the aqueous phase by the oxidation
of methanol (Jacob, 1986),  and by the oxidation of HMS  (Martin et
al.,  1989).  However, the rate of in-cloud formation of
formaldehyde is negligible relative to the rate of gas-phase
formation.

6.4.3 Formaldehyde Residence Times

     Residence times for formaldehyde were calculated by
considering gas-phase chemical reactions with OH, N03,  and H02,
photolysis, in-cloud chemical reaction with OH, and wet and dry
deposition.  The reaction of aqueous formaldehyde with aqueous
S02 was  not  considered.   Although this  reaction is fast and may
be important to cloud chemistry as a whole, it does not destroy
formaldehyde but merely binds it up as an adduct.

     The results of the residence time calculation for
formaldehyde are presented in Table 6-5.  During the daytime,
under clear-sky conditions, the residence time of formaldehyde is
determined roughly equally by its photolysis and reaction with
OH,  leading to calculated residence times on the order of a few
hours under summer, daytime, clear-sky conditions.  The summer,
daytime residence times for formaldehyde presented in Table 6-5
are comparable to a half-life of 2.6 h  (equal to a residence time
of 3.8 h)  previously estimated for formaldehyde under polluted
urban conditions (NRC, 1981).  The residence time of formaldehyde
in the atmosphere has also been estimated by EPA to range from
0.1 to 1.2 days (Cupitt, 1980), in good agreement with the values
presented in Table 6-5.

     In the presence of clouds, approximately 10 to 30 percent of
the daytime chemical destruction of formaldehyde and 20 to 90
percent of the nighttime chemical destruction of formaldehyde was
estimated to occur in clouds.  The presence of clouds would also
be expected to decrease the formation rate of formaldehyde; thus,
cloud cover may actually decrease formaldehyde concentrations
despite the predicted increase in residence time.

     At night, formaldehyde is destroyed slowly because of its
relatively slow rate of reaction with N03.   The reaction of
formaldehyde with H02 may be important  at  night under low N03
conditions, because the concentration of H02 radicals does not
decrease at night as rapidly as does OH.  However, since this
reaction is reversible, the calculated residence time will be an
upper bound.  For the cases in which this reaction might be
important, the residence times calculated with and without the
H02 reaction are presented  in Table  6-5 as a range of possible
residence times.

     Dry deposition may also be important as a removal mechanism
for formaldehyde.   Residence times due to dry deposition were
estimated to range from 90 h under winter, nighttime conditions
to
                               6-10

-------
EPA-420-R-93-005
April 1993
TABLE 6-5. Atmospheric residence time calculation for formaldehyde. All times are in
hours unless otherwise noted.
Los Angeles
July
Clear sky - day 3
Clear sky - night 20-60*
Clear sky - avg 4
Cloudy - day 5
Cloudy - night 14-30*
Cloudy - avg 7
Rainy - day --**
Rainy - night --**
Rainy - avg --**
Monthly 5
Climatological
Average
Jan
10
90
20
20
70
30
3
1.4
2
18
St. Louis
July
2
30-250*
3-4*
4
14-30*
6
3
3
3
4
Jan
13
90
30
20
70
40
0.8
0.3
0.4
18
Atlanta New York
July
2
20-70*
4
3
6-8*
4
2
3
2
4
Jan July
10 3
80 20-110*
20 5
19 6
70 18-50*
30 9
1.6 3
0.7 3
0.9 3
14 7
Jan
17
90
40
30
80
50
0
0
0
17






.8
.5
.6

*Range  of  values  obtained with and without HCHO + H02 reaction (see text).
**Not calculated  since July  rainfall  is  zero  for Los Angeles (Table 2-1).
                                            6-11

-------
                                                        EPA-420-R-93-005
                                                            April 1993
800 h for summer, nighttime conditions.  For the cases considered
here, dry deposition was a minor removal mechanism except under
winter,  nighttime conditions.  However, the deposition rate of
formaldehyde to water surfaces is much greater than the
deposition rate used in this calculation, and may be important to
consider for urban areas located near oceans or major lakes and
rivers.

     Under wintertime conditions, the photolysis rate is not
decreased by as large a factor as the OH radical concentration.
Therefore,  in the absence of precipitation, photolysis determines
the winter,  daytime formaldehyde residence time.  Wet deposition,
particularly under wintertime conditions, is an extremely
effective removal mechanism for formaldehyde.  Residence times
for formaldehyde during winter rainy conditions range from
fractions of an hour in colder climates to a few hours in warmer
climates.  Wet deposition accounts for roughly half of the
monthly average removal of formaldehyde during the wintertime.
It should be emphasized that this calculation assumes that the
partitioning of formaldehyde in rain holds for all forms of
precipitation.  For colder climates where January precipitation
is primarily in the form of snow, this assumption may not be
appropriate.

     As with benzene and 1,3-butadiene, the differences in
formaldehyde residence time between cities within a season were
not as large as the difference between seasons.  The summer
residence times are short in most cases, whereas the winter
residence times are greater than one day in most cases.  Thus,
formaldehyde as well as 1,3-butadiene must be considered to be
persistent in wintertime.  Unlike the other two species, however,
the effect of this longer winter residence time is difficult to
assess for formaldehyde because of the importance of secondary
formation.   Rates of formation of formaldehyde will be roughly an
order of magnitude slower in the wintertime.  Thus, it is
difficult to predict whether ambient concentrations of
formaldehyde will increase or decrease in winter.

     The major uncertainties in the residence time calculation
for formaldehyde include the factor-of-two uncertainty in the OH
radical concentration and the uncertainties in the deposition
velocity.  The uncertainty in the photolysis rate has only a
minor effect on the overall uncertainty.  The uncertainties
associated with the N03  concentration and the N03 rate constant
are less important for formaldehyde than for I,3-butadiene
because the N03  reaction with formaldehyde is much  slower than
the corresponding I,3-butadiene reaction.

6.4.4 Limited Urban Airshed Modeling Results for Formaldehyde

     The Urban Airshed Model (UAM) has been previously discussed
in Section 5.4.   Please refer to this section for details about

                               6-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993
the model, its inputs, and modifications.  Much of the
information below has been excerpted from reports conducted for
EPA by Systems Applications International (SAI) (Ligocki et al.,
1991, 1992) .

     Formaldehyde is an existing UAM species.  The simulations
included three formaldehyde species; one each for mobile and
stationary-source primary formaldehyde and one for secondary
formaldehyde.  Secondary formaldehyde is that produced by
atmospheric reactions.  The full radical and product chemistry of
formaldehyde was retained,  with the only change being that all
formaldehyde production was assigned to the secondary species
"FORM".   Since formaldehyde is a product of the photooxidation of
virtually all atmospheric organic compounds, it was not possible
within the scope of this study to track secondary formaldehyde
formed from mobile-source precursors.

St. Louis

     A time series plot of formaldehyde concentrations in the St.
Louis urban area is presented in Figure D-2 in Appendix D.
Mobile-source and stationary-source primary formaldehyde species
concentrations remain below 1 ppb throughout the simulation,
whereas  secondary formaldehyde increases to more than 5 ppb in
the afternoon.  The UAM simulation showed that formaldehyde
concentrations were about twice as high in the simulation with
chemistry as they were in the inert simulation, indicating that
formaldehyde is formed more rapidly than it is destroyed in urban
areas in the summertime.  The concentration of formaldehyde would
be expected to decrease in the wintertime due to a decrease in
photolysis activity on formaldehyde precursors.

     The contribution of mobile-source precursors to the
secondary formaldehyde concentrations can be estimated by
examining the mobile vs. stationary emissions of formaldehyde
precursors.  For formaldehyde, the simulation demonstrated that
the component of the concentration due to primary formaldehyde
emissions is small  (20 percent)  relative to the component due to
secondary formation in the atmosphere.  The fraction of this
secondary formaldehyde which formed from mobile-source precursors
is not known, but based on emissions of important formaldehyde
precursors, it appears to be 25-50 percent.

     The comparison of simulated concentrations with ambient
measured concentrations showed good agreement for formaldehyde.

     The formaldehyde photolysis rates used in the UAM for this
study were the higher (and currently accepted) values rather than
those used in the Carbon Bond Mechanism-IV  (CBM-IV).   Besides the
effect which changing the photolysis rate would have on
formaldehyde concentrations, there is the potential for secondary
effects  on other species concentrations, such as 1,3-butadiene,

                              6-13

-------
                                                        EPA-420-R-93-005
                                                            April 1993
because the formaldehyde photolysis is a source of radicals and,
ultimately, ozone.  A sensitivity study was conducted in which
the formaldehyde photolysis rate was increased by an additional
30%. The results from this simulation  (with the base-case initial
and boundary concentrations)  showed, as expected, the higher
photolysis rate caused a decrease in the predicted formaldehyde
concentrations during the afternoon.  This decrease was roughly
10% of the formaldehyde concentration.  Because the higher
formaldehyde photolysis rate caused increased production of
reactive radicals, the 1,3-butadiene concentration decreased by
about 3% in the mid-afternoon in this simulation as compared to
the base case.

Houston and Baltimore-Washington Area Simulations

     Simulations for the summer Baltimore-Washington area episode
(Ligocki et al.,  1992) resulted in both increases and decreases
in ambient formaldehyde with use of federal reformulated
gasoline,  with increases due to increased primary formaldehyde in
near-source areas, and decreases due to decreased secondary
formaldehyde in downwind areas.  Overall, the increases and
decreases in simulated ambient formaldehyde concentration
approximately cancel out.  Use of California reformulated
gasoline resulted in a decrease in secondary formaldehyde nearly
three times as large as in federal reformulated gasoline
scenarios, with similar primary formaldehyde increases.  Maximum
daily average formaldehyde concentration for the 1988 base
scenario was 9.3 ppb.   Motor vehicle-related formaldehyde
accounted for about 35% of total formaldehyde emissions.  Motor
vehicle-related formaldehyde also accounted for about 10% of
total simulated ambient formaldehyde on day 2 and 15% on day 3,
based on the 1995 no motor vehicle scenario.  75 to 80 percent of
this formaldehyde was secondary.

     Summer Baltimore-Washington area simulations were in fairly
good agreement with UATMP data for formaldehyde in the Baltimore
part of the domain, but UAM-Tox overpredicted formaldehyde in the
Washington part of the domain  (although the overprediction was
lower than for UAM).

     In the winter 1988 base scenario, the maximum daily average
formaldehyde concentration was 10.2 ppb, slightly higher than in
summer.  However, simulated concentrations throughout most of the
domain were lower.  Simulations for the winter Baltimore-
Washington area episode resulted in slight increases in ambient
levels of formaldehyde with the use of reformulated gasoline, on
the order of 1-2 percent, with a primary formaldehyde increase
and a secondary formaldehyde decrease.  Motor vehicle-related
formaldehyde emissions accounted for about 43% of total
formaldehyde emissions.  Motor vehicle primary formaldehyde
emissions were about 30 percent higher with reformulated gasoline
use.  Motor-vehicle related formaldehyde accounted for about 12%

                               6-14

-------
                                                        EPA-420-R-93-005
                                                            April 1993
of the maximum simulated concentration, based on the 1995 no
motor vehicle scenario.  Comparison of simulated concentrations
with measured concentrations in the Washington part of the
modeling domain indicate that the model may underpredict winter
formaldehyde concentrations.

     For the summer 1987 base scenario in Houston, the maximum
daily average formaldehyde concentration was 23.4 ppb.   Motor
vehicle-related formaldehyde accounted for about 19% of total
formaldehyde emissions in the 1987 base scenario, and 6% of the
maximum simulated concentration, based on the 1995 no motor
vehicle scenario.  Simulations for the summer Houston episode
predicted slight increases in the simulated daily average
concentration throughout most of the domain with use of
reformulated gasoline.  Comparison of simulated concentrations
with measured concentrations suggest the model may overpredict
formaldehyde concentrations in Houston.

6.5  Exposure Estimation

6.5.1  Annual Average Exposure Using HAPEM-MS

     The data presented in Table 6-6 represent the results
determined by the HAPEM-MS modeling that was described previously
in Section 4.1.1.  These numbers have been adjusted to represent
the increase in VMT expected in future years.

     The HAPEM-MS exposure estimates in Table 6-6 represent the
50th percentiles of the population distributions of exposure,
i.e., half the population will be above and half below these
values.  High end exposures can also be estimated by using the
95th percentile of the distributions.  According to the HAPEM-MS
sample output for benzene, the 95th percentile is 1.8 times
higher than the 50th percentile for urban areas, and 1.2 times
high for rural areas.   Applying these factors to the exposure
estimates in Table 6-6, the 95th percentiles for urban areas
range from 1.03 ug/m3  for  the  2010  expanded California  standards
scenario to 2.25 ug/m3 for the 1990  base  control scenario.   The
95th percentiles for rural areas range from 0.37 to 0.82 ug/m3,
respectively.

6.5.2  Comparison of HAPEM-MS Exposures to Ambient Monitoring
Data

     As stated in section 4.1.2, four national air monitoring
programs/databases contain data on formaldehyde.  The Aerometric
Information Retrieval System  (AIRS), the Toxic Air Monitoring
System (TAMS),  the Urban Air Toxic Monitoring Program (UATMP),
and the National Ambient Volatile Organic Compounds Data Base
(NAVOC) all have a significant amount of data for formaldehyde.
The urban exposure data for formaldehyde from all four databases
are summarized in Table 6-7.

                               6-15

-------
                                                         EPA-420-R-93-005
                                                              April 1993
     The AIRS data  base contains data on formaldehyde  for 1987
and 1988  (AIRS User's  Guide Volume I-VII, 1989).   The  location
and number of the sites varies between the two years.   Referring
back to the summary table in section 4.1.2, 14 sites monitored
                                6-16

-------
                                                            EPA-420-R-93-005
                                                                April 1993
Table  6-6.
Annual  Average HAPEM-MS Exposure  Projections  for
Formaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Urban Exposure
ug/m3
1.25
0.78
0.83
0.58
0.60
0.60
0.58
0.59
0.57
Rural Exposure
ug/m3
0.68
0.42
0.45
0.31
0.32
0.33
0.31
0.32
0.31
                                 6-17

-------
                                                        EPA-420-R-93-005
                                                             April 1993
Table 6-7.  Air Monitoring Results  for  Formaldehyde.
Program
AIRS
UATMP
TAMS
NAVOC
Years
1988
1987
1989
1990
1987-89
1987
Ambient Data3
ug/m3
3.26
3.43
2.61
5.18C
2.15
4.00
Estimated
Motor Vehicle
Contribution1"
ug/m3
1.08
1.13
0.86
1.71
0.71
1.32
aCaution should be taken in comparing these numbers.  The methods
of averaging the data are not consistent  between air monitoring
databases.  The sampling methodology  is also  inconsistent.

bThe  ambient data are adjusted to represent the motor vehicle
contribution to the ambient concentration,  which for formaldehyde
is estimated to be 33%, based on emissions  inventory
apportionment and modeling.

The  1990 UATMP is the only program which accounted  for ozone
interference in the measurement method.
                               6-18

-------
                                                        EPA-420-R-93-005
                                                            April 1993
formaldehyde in 1987 and 16 sites monitored it in 1988.  All the
cities where the monitoring sites were located are listed below.

          Birmingham, AL                Miami, FL
          Jacksonville, FL              Atlanta, GA
          Chicago, IL                   St. Louis, MO
          Louisville, KY                Baton Rouge, LA
          Dearborn, MI                  Detroit, MI
          Port Huron, MI                Lansing/E. Lansing, MI
          Cleveland, OH                 Dallas, TX
          Houston, TX                   Burlington, VT

The average level of formaldehyde for 1987  (averaged equally by
the number of sites) was 2.79 ppb.  In 1988, the average was 2.65
ppb.
Because the number of sites differs from year to year and the
number of samples taken at the various sites varies greatly, it
is misleading to make direct comparisons between these two
numbers.  However, these numbers do provide a general idea of the
average amount of formaldehyde being emitted in a year.

     Looking at the AIRS data on a site by site basis for 1987,
Cleveland, Ohio had the highest average level of formaldehyde
among the 14 sites sampled (4.72 ppb) at a site located in a
central, urban, commercial area.  Six samples were taken at this
site.  Miami, Florida had the lowest average level of
formaldehyde in 1987 (1.43 ppb) with six samples taken in an
urban commercial area.   In 1988, Louisville, Kentucky had the
highest average reading of formaldehyde (5.03 ppb) with 20
samples taken at a downtown urban commercial area.  Port Huron,
Michigan had the lowest average reading (1.20 ppb) with 19
samples taken in a suburban residential area.

     Referring to the table in section 4.1.2., ten sites in the
Toxics Air Monitoring System (TAMS) monitored formaldehyde in the
following 4 cities.

                         Boston (3 sites)
                         Houston  (3 sites)
                         Chicago  (3 sites)
                         Seattle/Tacoma (1 site)


The period of time took place in various time periods between
1987 and 1989.  The overall average for the 10 sites was 1.75
ppb.  Because of the varying time intervals, it may be misleading
to make direct comparisons between the four cities involved, but
the measurements do give a general indication of the amount of
formaldehyde being emitted.

     One of the three TAMS sites in Chicago recorded the highest
amount of formaldehyde  (2.27 ppb)   Chicago also had the highest

                               6-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993
measurement of formaldehyde averaged from all three of the  sites
located there  (2.13 ppb).  One of the sites in Houston had  the
lowest level of formaldehyde  (1.23 ppb).  Although it only  had
one site, Seattle/Tacoma had the lowest overall average of
formaldehyde (1.54 ppb).  Boston had the next lowest average with
three sites (1.56 ppb).

     The Urban Air Toxics Monitoring Program  (UATMP) monitored
the twelve cities listed below.

          Baton Rouge, LA               Miami, FL
          Chicago, IL                   Pensacola, FL
          Camden, NJ                    St. Louis, MO
          Dallas, TX                    Sauget, IL
          Ft.  Lauderdale, FL            Washington, B.C.
          Houston, TX                   Wichita, KS

Washington, B.C. and Wichita, Kansas each had two monitoring
sites, while the other 10 cities each had one monitoring  site.
At least 28 samples were collected at each site, except for
Pensacola, Florida (7 samples).  The comparatively larger number
of samples taken in UATMP makes the data more reliable.   The
overall average formaldehyde level for all the samples was  2.13
ppb.

     Averaged together, the two sites in Washington, B.C  had the
highest level of Formaldehyde  ((3.77 + 3.09)/2 = 3.43 ppb).
Twenty-eight samples were collected at one site and thirty  were
collected at the other site.  Also averaged together, the two
sites in Wichita, Kansas had the lowest level of formaldehyde
((1.46 + 1.40)/2 = 1.43 ppb).

     In the 1990 Urban Air Toxics Monitoring Program  (UATMP), 354
measurements of formaldehyde were taken at 12 sites.  These sites
were in the cities listed below.

          Baton Rouge, LA               Chicago, IL
          Camden, NJ                    Houston, TX
          Orlando, FL                   Pensacola, FL
          Port Neches, TX               Sauget, IL
          Toledo, OH                    Washington, B.C.
          Wichita, KS

The highest average was 6.44 ug/m3 (7.92 ppb)  at an urban
commercial site in Washington, B.C..  Thirty samples were
collected at this site.  The lowest average was 1.83 ug/m3  (1.49
ppb) at a suburban residential site in Houston, Texas.  Twenty-
six samples were collected at this site.  The overall average of
the averages for each site was 5.18 ug/m3 (4.21 ppb).   Ozone was
removed from the ambient air collected in this program through
the use of an ozone denuder.  The use of an ozone denuder in the
sampling system resulted in higher, but more accurate, reported

                               6-20

-------
                                                        EPA-420-R-93-005
                                                            April 1993
formaldehyde concentrations.  Only the 1990 UATMP data will be
used for the comparisons in this study.

     The National Ambient Volatile Organic Compound  (NAVOC)
program only had one monitoring site in Philadelphia,
Pennsylvania for formaldehyde.  Thirty-six samples were collected
and averaged equally.  This resulted in an average of 3.25 ppb of
formaldehyde.

     HAPEM-MS assumes that the dispersion and atmospheric
chemistry of formaldehyde is similar to CO.  This assumption is
not valid for a reactive compound like formaldehyde, which is
both destroyed and formed in the atmosphere.  For formaldehyde,
HAPEM-MS would overestimate the primary (i.e., directly emitted)
concentration in the atmosphere, since formaldehyde is more
reactive than CO.  On the other hand, HAPEM-MS would not account
for, and thus underestimate, the secondary  (i.e., atmospherically
formed) formaldehyde, since HAPEM-MS does not account for
atmospheric transformation.  Since these two factors offset one
another to some extent,  it is possible that the HAPEM-MS results
could still provide a reasonable estimate of the formaldehyde
exposure from motor vehicles.  Also, HAPEM-MS offers the
advantage of being able to project future formaldehyde levels,
based on emission data.

     To test the reasonableness of using the HAPEM-MS modeling
results, the HAPEM-MS results for 1990 are compared to ambient
monitoring results for recent years.  In order to make this
comparison, the motor vehicle contribution to total ambient
formaldehyde needs to be estimated.  This requires first
estimating the fractions of total ambient formaldehyde due to
primary and secondary formaldehyde, and then estimating the motor
vehicle contribution to primary and secondary formaldehyde.

     The Five-City Study (EPA, 1989) and the UAM-Tox atmospheric
modeling studies conducted by SAI  (Ligocki et al.,  1991, 1992)
attempted to apportion the formaldehyde in the atmosphere into
primary and secondary contributions.  The Five City Study
estimated that primary formaldehyde emissions account for about
40% of the total ambient formaldehyde.  The UAM-Tox modeling
studies determined that the concentration due to primary
emissions is small, about 20%, relative to secondary formation of
formaldehyde.   The mid-point of these studies, 30%, was chosen to
represent the contribution of primary formaldehyde emissions.
Therefore, 70% was chosen to represent the contribution of
secondary formaldehyde.

     These studies also attempted to apportion a fraction of
secondary formaldehyde formation in the atmosphere to motor
vehicles.   The Five City Study determined that motor vehicles are
responsible for 35% of total VOC which contributes to secondary
formaldehyde production.  The St. Louis modeling study  (Ligocki

                               6-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993
et al.,  1991) stated that the fraction of secondary formaldehyde
formed from motor vehicle precursors is not known, but based on
emissions of important formaldehyde precursors, it appears to be
25 to 50%.  Based on this information, it was then estimated that
motor vehicles account for approximately 35% of the secondary
formaldehyde.

     The 1987 EPA Air Toxics Report (Carey, 1987) attributed
approximately 28% of the primary formaldehyde emissions to motor
vehicles.  This percentage is based on 1985 emissions data.

     By using the numbers described above, the portion of
formaldehyde in the ambient air that is attributable to motor
vehicles was determined to be 33%.  The fractions are:  30%
primary formaldehyde in the ambient air of which 28% is from
motor vehicles and 70% secondary formaldehyde in the ambient air
of which 35% is due to motor vehicles.  The calculation is as
follows:
     .30(.28) + .70(.35) = 33% of total ambient formaldehyde from
     (primary) (secondary)    motor vehicles

or:  8.4% + 24.5% = 33%
     This estimate is higher than the estimates in the Houston
and Baltimore-Washington Area UAM-Tox simulations  (Ligocki et
al.,  1992).  In Baltimore-Washington, motor vehicle-related
formaldehyde accounted for about 10% of total simulated ambient
formaldehyde on day 2 and 15% on day 3, while in Houston, motor
vehicle- related formaldehyde accounted for about 19% of total
simulated ambient formaldehyde.  The estimate of 33% will be used
in this study to represent the nationwide average percentage of
ambient formaldehyde attributable to motor vehicles, while
acknowledging the apparent area-to-area variations and the
possibility that this may overestimate the motor vehicle
contribution.  Using this estimate, two approaches are used to
compare the HAPEM-MS and the air monitoring results.  The first
approach attempted to adjust the HAPEM-MS number upward to
account for secondary formaldehyde.  If it is assumed that motor
vehicles contribute 33% to ambient formaldehyde, 8.4% primary and
24.5% secondary (as determined in the equation above), then the
ratio of secondary to primary is 2.92:1.  If the primary
formaldehyde is 1.25 ug/m3  (from HAPEM-MS)  then the secondary is
2.92x1.25 or 3.65 ug/m3.  When added to the primary formaldehyde
result of 1.25 ug/m3,  the total is  4.90 ug/m3 formaldehyde
attributable to motor vehicles.  This is inconsistent with the
ambient monitoring data presented in Table 6-7 and thus was not
used.
                               6-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     The second approach was to adjust the ambient air monitoring
data to estimate the motor vehicle portion.  This method applied
the 33% formaldehyde from motor vehicles to each of the air
monitoring results.  This is presented as part of Table 6-7.
Since the only program that accounted for the interference of
ozone was the 1990 UATMP, only that ambient data will be used for
this comparison.  The resulting 1990 UATMP level is 1.71 ug/m3.
When the adjustment factor of 0.622 for the ambient motor vehicle
levels, that was determined in Section 5.5.2 is applied, this
exposure level becomes 1.06 ug/m3.   The  HAPEM-MS 1990  base
control exposure level of 1.25 ug/m3 must  be multiplied by  a
factor of 0.848 to reduce it to 1.06 ug/m3 to agree with the
ambient data.  All analysis based on the HAPEM-MS ambient motor
vehicle levels will have this factor applied.  Adjusted urban,
rural, and nationwide exposures are found in Table 6-8.

     Any formaldehyde exposures projected by HAPEM-MS itself
should be viewed with caution.  The adjusted HAPEM-MS exposure
estimates attempt to account for both primary and secondary
formaldehyde; however, these estimates are based only on changes
in primary emissions of formaldehyde.  The reactivity of motor
vehicle VOC emissions is likely to change with technology and
fuel changes.  Changes in the reactivity of these emissions,
which would result in changes to secondary formaldehyde levels,
cannot be accounted for by HAPEM-MS.

6.5.3  Short-Term Micrenvironment Exposures

     The primary emphasis for formaldehyde exposure will be
exposure in microenvironments that are enclosed, increasing the
exposure to tailpipe emissions.  These microenvironments include
in-vehicle and parking garage exposure.   The information
contained in Table 6-9 is excerpted from two studies that have
measured microenvironment exposures to formaldehyde.  These two
studies are the In-Vehicle Air Toxics Characterization Study in
the South Coast Air Basin (Shikiya et al., 1989) and Air Toxics
Microenvironment Exposure and Monitoring Study  (Wilson et al.,
1991).  See the information in Section 4.2 for more details about
the methodology, and Section 5.5.3 for a description of the
studies.

     Maximum microenvironment exposure levels of formaldehyde
related to motor vehicles were determined in these studies to
range from 4.9 ug/m3  from exhaust  exposure at a service station
to 41.8 ug/m3 from parking garage  exposure.   This compares  to
ambient levels of 2.15 to 4.0 ug/m3 determined through air
monitoring studies and presented in Table  6-7.  Since for the
majority of the population these are short-term acute exposures
to formaldehyde, the concern would be with non-cancer effects.
No RfC has been developed by EPA though formaldehyde is a known
irritant for the eyes, nose, and upper respiratory system at
levels of 123 ug/m3,  and become widespread at concentrations near

                               6-23

-------
                                                         EPA-420-R-93-005
                                                             April 1993
3.7xl03 ug/m3  in humans.  Exposures greater than 3.7xl03 ug/m3  are
generally intolerable  for more  than short periods.   Sensitive
humans may detect effects at  lower concentrations.   Please see
Section 6.8 for more information  on non-cancer effects.

     Due to more stringent  fuel and vehicle regulations, short-
term exposure to formaldehyde in  these microenvironments is
expected to decrease in  future  years.
                               6-24

-------
                                                           EPA-420-R-93-005
                                                                April 1993
Table 6-8.
Adjusted Annual Average  HAPEM-MS Exposure
     Projections for Formaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Exposure
(ug/m3)
Urban
1.06
0.66
0.70
0.49
0.51
0.51
0.49
0.50
0.49
Rural
0.57
0.35
0.38
0.27
0.27
0.28
0.27
0.27
0.26
Nationwide
0.95
0.58
0.62
0.42
0.44
0.44
0.42
0.46
0.42
                                 6-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Table 6-9.  Microenvironment Exposure to Formaldehyde  (iig/m3) .
Scenarios
SCAQMD Study3
(Shikiya et
al., 1989)
SCAQMD Studyb
(Wilson et
al., 1991)
In-Vehicle
Mean
15.4

Max.
35.4

Service
Station
Mean


Max.

4.9
Parking
Garage
Mean


Max.

41.8
Office
Building
Mean


Max.

44.2
aThe  estimated sampling time period was 1.5 hours/round-trip.
bThe  measurements from this study are five minute levels.
6.6  Carcinogenicity of Formaldehyde and Unit Risk Estimates

6.6.1  Most Recent EPA Assessment

     The information presented in Section 6.6.1 was obtained  from
EPA's Assessment of Health Risks to Garment Workers  (EPA, 1987a),
the Integrated Risk Information System  (IRIS)  (EPA, 1992b), the
Motor Vehicle Air Toxics Health Information  (Clement, 1991),  as
well as the primary sources cited in these documents.  The
carcinogenicity risk assessment for formaldehyde was last updated
on IRIS in January 1992, and contains data published through
1987.  The 1991 version of the formaldehyde risk assessment on
IRIS does not contain any information that is not included in the
1987 risk assessment.  The Office of Toxic Substances  (OTS)
prepared a formaldehyde risk assessment update in September 1990
(EPA 1990a, external review draft).  This document is not yet
final,  and thus does not yet represent official Agency policy
with regard to the risk assessment of formaldehyde.  EPA's
Science Advisory Board has reviewed this document, and has
requested additional analyses.  Therefore, the results presented
in the OTS assessment are likely to change.  Nevertheless, new
issues discussed in this 1990 risk assessment update will be
summarized in Section 6.6.2.  Section 6.6.3 summarizes recent and
ongoing research not included in the 1987 EPA risk assessment for
formaldehyde.  Some of this recent and ongoing research  is
discussed in the 1990 risk assessment, but is not yet part of the
official Agency risk assessment for formaldehyde.
                               6-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993
6.6.1.1  Description Of Available Carcinogenicity Data

Genotoxicity

     Mutagenic activity of formaldehyde has been demonstrated in
viruses, Escherichia coli, Pseudomonas fluorescens, Salmonella
typhimurium (all three are bacteria), and certain strains of
yeast, fungi,  Drosopohila  (fruit fly), grasshopper, and mammalian
cells  (Ulsamer et al.,  1984).  Formaldehyde has been shown to
cause gene mutations,  single strand breaks in DNA, DNA-protein
crosslinks, sister chromatid exchanges, and chromosomal
aberrations.  Formaldehyde produces in vitro transformation in
BALB/c 3T3 mouse cells, BHK21 hamster cells and C3H-10TI/2 mouse
cells, enhances the transformation of Syrian hamster embryo cell
by SA7 adenovirus, and inhibits DNA repair (Consensus Workshop on
Formaldehyde,  1984).

Animal Data

     According to EPA  (1987b),  the principal studies indicating
formaldehyde may cause cancer in animals are Kerns et al.  (1983)
(the Chemical Industry Institute of Toxicology  [CUT] study),
Albert et al.  (1982) (the NYU study), Sellakumar et al.  (1985),
and Tobe et al.  (1985).  The carcinoma response in animals was
similar for the four studies but the benign tumor response
differed among the studies.  EPA (1987b) concluded that there was
"sufficient" evidence of carcinogenicity of formaldehyde in
animals by the inhalation route based on increased incidence of a
rare type of malignant cancer (i.e., squamous cell carcinoma) in
rats and mice and in both sexes.

     In the CUT study, Fischer 344 rats and B6C3F1 mice
(120/sex/group)  inhaled 0, 2, 5.6 or 14.3 ppm formaldehyde,
6 hours/day, 5 days/week, for 24 months followed by 6 months of
recovery  (Kerns et al., 1983).   Animals were sacrificed at
6 months, 12 months, and 18 months, while at 24 and 27 months,
the number of animals sacrificed was unclear from the report.
The study was terminated at 30 months.  For the rats, mortality
was significantly increased in the 14.3 ppm exposed animals after
12 months and in the 5.6 ppm exposed males after 17 months.  At
the end of the study period, squamous cell carcinomas in nasal
cavities were reported in 51 of 117 male rats and 52 of
115 female rats at the high dose and in 1 of 119 male rats and 1
of 116 female rats at the intermediate dose.  No tumors were
observed at 0 or 2 ppm formaldehyde exposure.  Polypoid adenomas
(benign tumors)  of the nasal mucosa were also seen in rats at all
doses in a significant negative dose-related trend, although the
incidence was significant only at 2 ppm formaldehyde.  In the
B6C3F1 mice, squamous cell carcinomas were observed in only two
of the males exposed to 14.3 ppm formaldehyde.  Although this
increase was not significant, the occurrence of this carcinoma

                               6-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993
type in mice was considered to be formaldehyde-related because
this cancer is rare in mice.

     In the Tobe study, male Fischer 344 rats who were exposed to
0 (methanol only),  0.3, 2, 3.3 or 15 ppm formaldehyde in aqueous
solution methanol,  6 hours/day, 5 days/week for 28 months  (Tobe
et al.,  1985).   At the end of the 15-month exposure period,
mortality in the high dose group was 88% while mortality was 60%
in controls.  Mortality was 32% in the low-dose group at 28
months.   Squamous cell carcinomas occurred in 14 of 27 high-dose
rats surviving past 12 months.  No polypoid adenomas were
observed but benign nasal papillomas were evident in
formaldehyde-exposed animals.

     Moreover,  in the NYU study, after Sprague-Dawley rats
inhaled 0 (air) or 14.2 ppm formaldehyde, 6 hours/day, 5
days/week, for a lifetime, there was a statistically significant
elevation of the squamous cell carcinoma in 38 of 100 rats
(Albert et al., 1982).  Papilloma or polyps were detected in 10
of 100 exposed rats.  The study was limited because only one
exposure level was tested.

     Sellakumar et al. (1985) exposed male Sprague-Dawley rats, 6
hours/day, 5 days/week for lifetime to 10 ppm HC1 and to 14 ppm
formaldehyde.   The HC1 and formaldehyde were administered
simultaneously and separately, with an equal number of rats
receiving an air control.  HC1 was administered to determine if
tumor response was enhanced by an additional irritant effect or
by the combining of formaldehyde and HC1 to form bis-
(chloromethyl)ether (BCME).    Groups receiving formaldehyde alone
or with HC1 showed an increase in nasal squamous cell carcinomas;
those without formaldehyde were free of carcinomas and other
tumors,  although rhinitis and hyperplasia were of comparable
incidence.

     Two other chronic inhalation studies examined the
carcinogenicity of formaldehyde in upper and lower airways
(Dalbey et al., 1982;  Horton et al.,  1963).  However, nasal
tissues were not systematically examined which limited the
usefulness of these studies.  Horton et al. (1963) exposed C3H
mice to coal tar aerosol and/or formaldehyde at 40, 80, and 160
ppm for 1 hour/day, 3 days/week, for 35 weeks (4 weeks for the
160 ppm group).  The study was limited because of insufficient
animals surviving the first year, individual exposures were
short,  and complete histopathology was not reported.  Dalbey
(1982)  exposed male Syrian golden hamsters (88-132/group)  to 10
ppm formaldehyde, 5 days/week, for a lifetime.  Results showed
that there was no evidence of carcinogenic activity following
exposure to 10 ppm formaldehyde in animals although survival was
reduced relative to controls.  EPA (1987a) found that the
pathology evaluation in the study was less rigorous compared to
                               6-28

-------
                                                        EPA-420-R-93-005
                                                            April 1993
the Kerns et al.  (1983) study and the study was limited because
only one dose was tested.


     Rusch et al.  (1983) carried out a 6-month toxicity study in
6 male cynomolgus monkeys, 40 F344 rats, and 20 Syrian golden
hamsters with 22 hours/week exposure to three levels of
formaldehyde with corresponding controls.  The highest dose
tested was 2.95 ppm.  The short duration of the assay, the small
sample sizes, and,  possibly, the low concentrations tested,
limited the sensitivity of the assay to detect tumors.  In the
highest dose group in both rats and monkeys, incidences of
squamous metaplasia/hyperplasia of the nasal turbinates were
significantly elevated.

     Furthermore,  several recently published drinking water
studies provide additional suggestive evidence that formaldehyde
is carcinogenic following oral exposure as well.  The tumor-
promoting potential of formaldehyde in mouse skin, rat trachea,
and rat stomach has been also recently been demonstrated.  The
recent carcinogenicity studies referred to above are summarized
in Section 6.6.3.3.

Human Data

     EPA reviewed only cohort or case-control studies because
these studies yielded the best qualitative information for
evaluating causality.  There was a total of 28 studies but many
of them had limitations that could potentially influence the
conclusions, and therefore will not be addressed in this section.
Of these studies,  11 were of chemical or industrial workers and 7
were of medically-related professionals  (e.g., morticians,
pathologists).   The other 10 were case-control studies examining
workers with sinonasal cavity and pharyngeal cancers.  Only six
studies had enough data to evaluate exposure-response effects;
these are the studies that will be reviewed in this section.  Of
these six, two cohort studies (Blair et al., 1986, 1987; Stayner
et al.,  1988) and one case-control study (Vaughan et al., 1986)
were well conducted and specifically designed to detect small to
moderate formaldehyde-associated human risks.  These three
studies were discussed in the IRIS cover sheet. According to EPA,
weaknesses inherent in the human studies in general included: 1)
inference of formaldehyde levels from industrial hygiene data,
2) concurrent exposures to other chemicals which prevented
determination of specific exposure levels,  3) small sample size
for cohorts, 4) small number of site-specific deaths, and  5)
insufficient follow-up.

     Blair et al.  (1986) conducted the largest occupational
exposure study that has been published to date (see Section
6.6.3.4 for more details on the followup).   They reported a
significant increase in lung and nasopharyngeal cancer in a

                               6-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993
cohort study at 10 industrial sites.  The authors concluded that
there was little evidence showing an association between lung
cancer and formaldehyde exposure because the risk did not
increase with exposure intensity or cumulative exposure.  The
observation of nasopharyngeal tumors support similar findings in
animals.  EPA considered the lung and nasopharyngeal cancer
mortality data "meaningful" despite the lack of significant
trends.  EPA also believed that misclassification of exposure and
categorization of deaths into 4 exposure levels (although not
specified by EPA 1987a) may account for the lack of a dose-
response relationship.

     A cohort study by Stayner et al.  (1985) found buccal cavity
tumors in formaldehyde-exposed garment workers.  The SMR was
highest in workers with the longest duration of employment
(exposure)  and follow-up period (latency).   There were no other
details reported.

     A significant association was reported between
nasopharyngeal cancer and people living 10 years or more in a
"mobile home" built in the 1950's to 1970's (Vaughan et al.,
1986).   The walls and flooring in mobile homes are generally made
out of plywood or some sort of wood composite material that
contains urea-formaldehyde resins or adhesives.  Exposure to
formaldehyde in residents of mobile homes occurs when the
formaldehyde in these resins and adhesives offgas as the material
ages.

     The studies by Olsen et al.  (1984), Hayes et al.  (1986), and
Hardell et al. (1982)  reported significant excesses of sinonasal
cancer in individuals exposed to both formaldehyde and wood-dust.
However, only the first two studies controlled for wood-dust
exposure.  The detection limits in both studies exceeded
corresponding expected excesses in the incidence of sinonasal
tumor and,  therefore,  no significant excesses were likely to have
been observed (EPA, 1987a).  Acheson et al.  (1984) compared
excess mortalities due to lung cancer in one of six formaldehyde
resin producing plants in England.   Only borderline significance
was observed.  The authors concluded that the increases in
mortality from lung cancer were not related to exposure since the
elevation was not statistically significant when compared with
local lung cancer rates.  However,  EPA  (1987b) believed that the
risk was sufficiently increased to enable this study to be used
for corroboration.  Other studies,  Pattanen et al. (1985),
Bertazzni et al.   (1986), and Blair et al. (1986, 1987) also
indicated that lung cancer also may be associated with
occupational exposure to formaldehyde.  The risk associated with
sinonasal cancer appeared to be specific for the histologic type,
squamous cell carcinoma.  The relative risks observed for upper
respiratory tract cancers in all the reviewed studies ranged from
just above 1.0 (a risk of 1.0 implies no association between
exposure and disease)  to 3.0, depending on the site.

                               6-30

-------
                                                         EPA-420-R-93-005
                                                              April 1993
     There were  19  studies that indicated the possibility  that
observed leukemia and  neoplasms of the brain and colon may be
associated with  formaldehyde exposure; however, the biological
support for these findings has not yet been demonstrated.
                               6-31

-------
                                                        EPA-420-R-93-005
                                                            April 1993
6.6.1.2 Weight-of-Evidence Judgment of Data and EPA
Classification

     EPA has classified formaldehyde as a Group Bl, probable
human carcinogen under its Guidelines for Carcinogen Risk
Assessment.  This is based on limited epidemiological evidence
and sufficient evidence of carcinogenicity in animal studies.  In
addition, this evidence is supported by mutagenic activity in
various in vitro test systems.

     The CUT inhalation study  (Kerns et al. , 1983) in rats is
considered the primary study for estimating unit risk.  The study
was well designed,  well conducted, multiple doses  (4 exposure
levels) were included, and sufficient animals were tested.  The
malignant tumor data  (i.e., squamous cell carcinoma in nasal
cavity) in the Kerns et al. (1983) study were used for estimating
risk since the response in treated rats was definite and
unequivocal in both males and females and there was an increasing
dose-related trend.  Furthermore, similar malignant tumor types
were evident in all rat and mouse inhalation studies with
formaldehyde exposure.  EPA also believes that the appearance of
benign tumors in the Kerns et al. (1983) study contributes to the
qualitative weight-of-evidence that formaldehyde may pose a
carcinogenic hazard.

     The other animal inhalation studies had limitations that
prevented their use for quantitative risk assessment.  Data from
Sellakumar et al.   (1985), the NYU (Albert et al. 1982) and Tobe
et al.  (1985) studies were also considered by EPA for unit risk
estimates.  The Tobe et al. (1985) study gave supportive evidence
in the same strain of rats but was not used for primary risk
estimation because a tumor response was observed only at the high
dose.  The Albert et al.  (1982) study was considered less
appropriate because it contained only one nonzero exposure
concentration.  The Kerns et al.  (1983) study also suggests
carcinogenicity at the high dose but the response was limited or
not significant.

     A degree of uncertainty was due to the different responses
of animals to formaldehyde exposure.  Only the rats showed
statistically significant numbers of neoplasms.  The mice only
had two carcinomas  (Tobe et al., 1985), but the response was
complicated by the fact that mice were able to reduce their
breathing rate to a greater extent than rats.  According to EPA,
the mice with tumors received 14.3 ppm formaldehyde, a dose that
approximates that which rats received at 5.6 ppm.  Thus, on a
"dose"  received basis, the rats and mice may be equally sensitive
to formaldehyde.
                               6-32

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     The epidemiological data had limited exposure information,
insufficient sample size, and concurrent exposures for risk
estimate determination.

6.6.1.3  Data Sets Used For Unit Risk Estimates

     The consequences of inhalation exposure to formaldehyde has
been studied in rats, mice, hamsters, and monkeys.  The principle
evidence comes from positive studies in both sexes of two strains
of rats (Kerns et al.,  1983; Albert et al.,  1982; Tobe et al.,
1985) and males of one strain of mice (Kerns et al., 1983), all
showing squamous cell carcinomas.  The primary data set is the
squamous cell carcinomas of the nasal turbinates in Fischer 344
rats from the CUT study (Kerns et al. ,  1983).  This data set
used to calculate the cancer risk estimate for formaldehyde is
summarized in Table 6-10.

     Three epidemiological studies are also used as supporting
evidence.   Two cohort studies (Blair et al.,  1986, 1987; Stayner
et al., 1988) and one case-control study (Vaughan et al.,
1986a,b) were well-conducted and specifically designed to detect
small to moderate increases in formaldehyde-associated human
risks.   These were discussed previously in Section 6.6.1.Ib.
Primates and rats have been shown to respond similarly to
formaldehyde exposure with
respect to the development of nasal tumors.   In any case, ppm is
considered equivalent across species, so a species scaling factor
was needed.

6.6.1.4  Dose-Response Model Used

     Since low level exposure can not be measured in animal or
human studies, several models are possible for low-dose
extrapolation.  Data were inconsistent regarding a linear or
nonlinear relationship between formaldehyde exposure and
carcinogenicity.  Because of the absence of biological evidence
on the mechanism of action for formaldehyde,  the linearized
multistage procedure was chosen as the default model as specified
by EPA guidelines (see Appendix F for a complete explanation),
although various other models were presented for comparative
purposes.   They found that only the one-hit model produced higher
risk estimates  (about 10-fold higher).

6.6.1.5  Unit Risk Estimates

     The inhalation unit cancer risk is 1.3xlO"5  (ug/m3)'1 or
1.6xlO"2 (ppm)"1  based on  squamous cell carcinoma  in  F344  rats.
The unit risk should not be used if the air concentration exceeds
800 ug/m3.   The  major contributor to the uncertainty in the risk
estimate using the multistage model is due to the steep dose
response observed in the CUT study.  There was a 50-fold
increase in the number of tumors compared to a 2.5-fold increase

                              6-33

-------
                                                          EPA-420-R-93-005
                                                              April 1993
in the dose level;  0  tumors at 2 ppm,  2  at  5.6 ppm,  and 103 at
14.5 ppm.  Other  uncertainties are the marked nonlinearity of the
response and  the  different responses observed in the tested
animals.
                                6-34

-------
                                                                                   EPA-420-R-93-005
                                                                                       April 1993
Table 6-10. Summary  of  Data Set Used to Calculate Unit Risk Estimate for Formaldehyde.
Source
Kerns et
al.
(1983)a



Kerns et
al.
(1983)b



Test
Animal
F344
rats,
male and
female,
combined



F344
rats,
male and
female,
combined



Tumor
Type
Squamous
cell
carcinoma



Squamous
cell
carcinoma



Administered
Dose
(ppm)
0
2
5.6
14.3
0
2
5.6
14.3
Human
Equivalent
Dose
(mg/kg/day)
0
2
5.6
14.3
Oc
15.3
70.8
318
Tumor
Incidence
0/156
0/159
2/153
94/140
0/156
0/159
2/153
94/140
aData set used in EPA 1987b cancer  risk  assessment
bData set used in EPA 1990 cancer risk assessment
Delivered dose expressed as pmol/mg/DNA/day calculated using rat DPX data and
for average daily dose
adjusted
                                            6-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     There is a wide range between MLE and upper bound estimates
of risk (ranging from 1-5 orders of magnitude) at different
exposure levels showing the statistical uncertainty of the
estimates that were generated from highly non-linear data.  For
example, using the 1987 unit risk, at an exposure level of 3 ppm
(3,685 ug/m3)  for  36  hours/week for 40 years (typical
occupational exposure conditions), the upper bound estimate of
lifetime cancer risk is 6xlO"3 and  the maximum  likelihood
estimate of risk is 6xlO"4, whereas a  10-year exposure to
0.07 ppm (86 ug/m3)  formaldehyde (believed to be the
home/environment background upper limit in conventional homes),
the upperbound estimate of risk is l.OxlO"4  and the maximum
likelihood estimate of risk is 6.0X10"11.   However,  the predictive
power of the model is not significantly disturbed by slight
perturbations of the data.

6.6.2  Other Views and Unit Risk Estimates

     This section presents alternate views and/or risk
assessments for formaldehyde.  These are summarized in Table 6-
11.

Office of Toxic Substances 1991 Formaldehyde Draft Report

     The OTS  (EPA, 1991b) risk assessment for  formaldehyde
concluded that recent animal studies confirm the previous
findings of an increased incidence of squamous cell carcinomas of
the nasal cavity in rats exposed by inhalation and a steep dose-
response curve.  In addition, the distribution of nasal tumors in
rats has been better defined; the findings suggest that not only
regional exposure but also local tissue susceptibility may be
important for the distribution of formaldehyde-induced neoplasms.
Many of the recent studies used in EPA (1991b) are discussed in
Section 6.6.3.

     In the OTS risk assessment update concerning the
epidemiological data, it was concluded that when the risk
assessment is examined in context of the previously reviewed
studies, the human studies released since 1987 support the
conclusions drawn by EPA in its 1987 document  and do not alter
the evaluation that  'limited' evidence exists  for an association
between formaldehyde and human cancer.  Collectively,  however,
the data do not conclusively demonstrate a causal relationship.

     The 1991 update goes on to describe that  recent
epidemiological studies provide additional evidence that "modest"
increases in nasopharyngeal and nasal cavity and sinus cancer
risks, and possibly in lung cancer risks, have been observed
among various occupational subgroups.  However, the evidence for
an association between lung cancer and occupational formaldehyde
is tenuous.  The recent epidemiological studies referred to above
are summarized in Section 6.6.3.4.

                               6-36

-------
                                                        EPA-420-R-93-005
                                                             April 1993
     The OTS update also concurred with  the weight-of-evidence
evaluation presented in the 1987 risk assessment.   Based  on
Table 6-11.
Comparison of Formaldehyde Inhalation Unit Risk
Estimates.
Source
EPA (1987)
EPA (1991b)

IARC (1987)
CARB (1992b)
OSHA (1987)
Classification
Group Blb
Group Bl

Group 2Ae
Probable Human
Carcinogen
Potential Occupational
Carcinogen
Cancer Unit Risk
Estimate
ug/m3)-1
Upper Bound3
1.3xlO"5
6xlO'7 c
8xlQ-6 d
f
6.0xlO"6
2 .64xlO"2 g
aMLEs  have not been presented because EPA does not generally
compare MLEs based on animal data because of  the  high  variability
associated with these numbers.  Therefore,  they are  of little
value.
bGroup Bl  = Probable Human Carcinogen
""Calculated using monkey DPX data
Calculated using rat DPX data
eGroup 2A  = Probable Human Carcinogen
fIARC  did  not conduct a quantitative risk assessment
gUpper bound estimates calculated for risk to 100,000 workers
exposed to 1 ppm for 45 years.  It  should be  noted that OSHA used
the maximum likelihood estimate, and not the  upper bound,  for
regulatory purposes.
                               6-37

-------
                                                        EPA-420-R-93-005
                                                            April 1993
sufficient animal evidence  (mainly nasal cancers in rats and
mice),  limited human evidence associating nasal and
nasopharyngeal cancer with formaldehyde exposure, and other key
evidence including structure-activity considerations, the known
genotoxic activity of formaldehyde, and the ability of
formaldehyde to injure cells and affect cell division, OTS
concurred that formaldehyde should be classified as a probable
human carcinogen  (Group Bl).

     The OTS updated risk assessment calculated new cancer unit
risks for formaldehyde.  However, the data set used as the basis
for the unit risks is the same as that used by EPA in 1987;
squamous cell carcinomas of the nasal turbinates in Fischer 344
rats from the CUT study (Kerns et al. ,  1983) .   The data set used
to calculate the cancer risk estimate for formaldehyde is
summarized in Table 6-9.

     In the OTS update, EPA chose to continue with the linearized
multistage model to calculate the new unit risk estimates because
there is insufficient evidence, especially with respect to
mechanism, to warrant a departure from this model.

     These unit risk estimates have been modified in the 1991 OTS
update to reflect new information regarding dose-rate effects and
the use of DNA binding data as an intracellular dosimeter for
formaldehyde.

     Since the 1987 risk assessment, data have become available
regarding nasal DNA binding of formaldehyde in the form of DNA-
protein cross-links (DPX),  and the quantitation of these DPX
levels (see Section 6.6.3.2 for a discussion of DPX).  OTS
concluded that these new data support the use of DPX as an
internal measure of formaldehyde dose,  and has used the data of
Casanova et al. (1989) in F344 rats and Heck et al.   (1989) in
Rhesus monkeys to calculate internal formaldehyde doses to be
used in the revised risk assessment.  Specifically,  the rat DPX
data were input into the linearized multistage model, and the
risk to humans was then calculated by applying monkey DPX data to
the resulting equation because it was believed that the actual
risk to humans lies somewhere between the risk estimates derived
using only the rat or the monkey DPX data.  The modified unit
risk estimates also used a different method to adjust the
calculated delivered doses to average daily doses to be input
into the linearized multistage model.  Generally, experimental
exposure rates are multiplied by a factor of 5/7 and 6/24  (to
reflect the fact that exposure only occurred for 6 hours/day,
5/days/week) to convert to continuous lifetime daily average
exposure.  In the case of formaldehyde,  OTS considered this to be
inappropriate.  There is evidence to support the hypothesis that
dose rate (or the concentration of formaldehyde reaching the
target tissue) is more critical in determining the severity of
the toxic effects, such as cell proliferation and histological

                               6-38

-------
                                                        EPA-420-R-93-005
                                                            April 1993
changes, than average exposure concentration.  Therefore,  OTS
adjusted the DPX-derived dose levels by a factor of only  5/7
because this approach acknowledges the importance of a possible
dose-rate effect by not averaging the exposure and expected DPX
over the course of 24 hours.  Tumor incidences were not induced
by a continuous exposure regiment, and these tumor incidences
should be linked as nearly as possible with the exposure  levels
which caused them.

     Based on the results of many pharmacokinetics studies, EPA
has concluded that most of the objections expressed in the expert
panel's report have been adequately addressed and that the use of
DPX as the surrogate dose for risk estimates appears appropriate
with the following reservations.  The DPX data were obtained
following a single exposure to formaldehyde  (acute), whereas the
carcinogenic bioassay was a chronic  (2-year) study.  The
different exposure conditions may have little effect on DPX
yields at low concentrations, where the normal morphology of the
nose is unaltered by exposure, but it may have a major effect at
high concentrations.  This is due to the proliferation of the
squamous  cells which may have very different metabolic
abilities, formaldehyde uptake, and detoxifying mechanisms than
the epithelial cell examined in the DPX experiments.  Indeed,
recent studies at CUT on the formation of DPX in rats exposed
subchronically to 15 ppm of formaldehyde indicate that such
effects can occur (Casanova and Heck, 1991).

     Another reservation regarding the use of DPX is that the
role of DPX, if any, in the induction of nasal cancer is  not
completely understood.  This problem is relatively insignificant
if DPX are used only as a dosimeter, and the linearized
multistage model is used to estimate risk.  However, it could
become more important if the DPX were given a specific
mechanistic role in a biologically-based model.

     A final reservation is that the current DPX data should not
be used to make assumptions about species differences in
sensitivity (response) since the necessary mechanistic
information is lacking.

     The resulting modified unit risk factors  (UCL) calculated by
OTS are 6xlO"7  (ug/m3)"1 (7xlO~4  [ppm]"1) using the  monkey-based DPX
data and 8xlO"6  (ug/m3)"1 (IxlO"2  [ppm]"1) using the rat-based DPX
data.  The MLE' s are S.lxlO"8  (ug/m3)"1 (IxlO"4  [ppm]"1)  and  S.lxlO"6
(ug/m3)"1 (IxlO"2  [ppm]"1), respectively.  The unit  risk estimate
calculated using the rat DPX data is lower than that calculated
in EPA  (1987a) presented above by an approximately 25-fold
difference, 4-fold of which is due to the difference in
continuous exposure correction used and 6-fold of which is due to
the use of the rat
                               6-39

-------
                                                         EPA-420-R-93-005
                                                             April 1993
DPX data.  The risk  estimate calculated using the monkey DPX data
is approximately  9 times  lower than the 1987 risk estimate.1

International Agency for  Research on Cancer (IARC)

     IARC has classified  formaldehyde as a Group 2A carcinogen.
A Group 2A carcinogen is  defined as an agent that is probably
carcinogenic to humans.   This classification is based on limited
evidence for carcinogenicity in humans and sufficient evidence
for carcinogenicity  in animals (IARC 1982, 1987).

     IARC reviewed the available human data and concluded that
these studies do  provide  some evidence that occupational exposure
to formaldehyde is associated with an excess of various forms of
cancer.  Cancers  that occurred in excess in more than one study
are:  Hodgkin's disease,  leukemia,  and cancer of the buccal
cavity and pharynx  (particularly nasopharynx),  lung, nose,
prostate, bladder, brain,  colon,  skin, and kidney (IARC 1982b,
1987c).  However, in many of these studies, actual exposure to
formaldehyde is unknown.

     IARC concluded  that  the available animal data provide
sufficient evidence  of the carcinogenicity of formaldehyde.
These data consist of inhalation studies in one strain of mice
and two strains of rats.   No unit risk estimate was determined by
IARC.

Motor Vehicle Manufacturer's Association  (MVMA)

     MVMA contracted with Environ Corporation to 1)  describe the
means for conducting an assessment that incorporates all
scientific information pertinent to the question of risk for
formaldehyde  (Environ,  1986); and 2)  to evaluate the risk
assessment issues in EPA's technical report "Air Toxics Emissions
from Motor Vehicles"  (Environ,  1987).   It is important to note
that neither document is  actually a risk assessment of
formaldehyde, i.e.,  no alternative unit risk estimates were
developed; rather, they are critiques of existing risk
assessments  (the  1986 document critiques OSHA's 1985 risk
assessment of formaldehyde [50 FR 50412-40499] , and the 1987
document critiques EPA's  risk assessment of motor vehicle air
toxics (Carey, 1987)  and  descriptions of elements that should be
considered in a comprehensive risk assessment of formaldehyde.

     Environ  (1986)  points out that any risk assessment of
formaldehyde must take into consideration the following issues:
    1The MLEs calculated in the 1991 updated risk assessment have been presented
here, but it is explicitly stated in this document that EPA does not generally
compare MLEs based on animal data because of the high variability associated with
these numbers.  Therefore, they are of little value.

                               6-40

-------
                                                         EPA-420-R-93-005
                                                              April 1993
      (a) the mechanism of action of formaldehyde
      (b) the relationship between the magnitude and duration of
     exposure  to  formaldehyde and the 'target-site  dose'  of the
     proximate carcinogen
      (c) the relative importance of the genotoxic activity of
     formaldehyde compared to its other biological  effects in
     determining  risk
      (d) the shape of the dose-response curve at dose  levels
     below the experimental range
      (e) the relationship between risk in rodents and  risk in
     humans.

     With regard  to mechanism of action and its role in risk
assessment, Environ (1986,  1987) describes several  mechanisms
that have been proposed to account for the carcinogenic effects
of formaldehyde.   These include:

      (1) Chemistry and metabolism.  This takes into account the
     nonlinear relationship between the concentration  of
     formaldehyde in the air and the level of DNA adducts to
     establish a  relationship between dose and response.

      (2)  Physiological effects.  Animals exposed to high levels
     of formaldehyde reduce their rate and depth of breathing,
     thus resulting in a reduction of inhaled dose.  In addition,
     formaldehyde reduces the protective flow of mucus over the
     surface of the nasal passages (nasal epithelium),  thus
     resulting in the slower removal of dissolved proportions of
     inhaled formaldehyde.   This information should be used to
     adjust the dose used in modeling the dose-response
     relationship to more accurately reflect the target-site
     dose.2
      (3)  Effect  on proliferation of respiratory epithelium.
     High concentrations of formaldehyde have been  shown to cause
     cellular  degeneration and abnormal stimulation of cell
     replication  that attempts to replace the dead  cells
      (regenerative hyperplasia)  of the nasal epithelium.   This
     effect may contribute to the carcinogenic action  of
     formaldehyde.   This proliferation does not occur  at exposure
     levels below 6 ppm,  therefore, this mechanism  would not
     likely contribute to carcinogenesis at low levels.

     Environ agrees with EPA and OSHA that the data set that
provides the best estimate of the relationship between dose and
response for formaldehyde is the CUT rat study  (Kerns et al. ,
    2OTS, in their updated risk assessment for formaldehyde (EPA 1990),
acknowledges that these two mechanisms  (reduced respiration rate and mucociliary
clearance), may alter the  dose of formaldehyde that reaches the target tissue.
However,  they concluded that not enough is known to quantitate the amount that
these two mechanisms may alter the actual delivered dose of formaldehyde.

                               6-41

-------
                                                        EPA-420-R-93-005
                                                            April 1993
1983).   However, Environ  (1986) cautions that all data sets
should be used to provide a range of risk estimates to provide a
better indication of the uncertainty of the estimates.

     Environ (1986, 1987) contends that EPA's and OSHA's use of
the linearized multistage model for low dose extrapolation
overestimates the carcinogenic risk of formaldehyde.  They
suggest that there is some evidence to indicate that formaldehyde
may be a threshold carcinogen.  This, together with the fact that
the dose-response data for formaldehyde are not linear, led
Environ to conclude that linear extrapolation of responses that
occur following exposure to high doses to predict responses at
low doses may not be entirely valid.  They suggest that if a
model that better fits the data is used, for example, a non-
linearized multistage model (i.e., a five-stage or six-stage
model)  or a Weibull model, then the predicted risks at low dose
levels are orders-of-magnitude lower than those predicted using
the linearized multistage model.

California Air Resources Board  (CARE)

     CARB (1992b),  like EPA and IARC, has concluded that
formaldehyde is a probable human carcinogen.  CARB  (1992b) has
performed an assessment of the carcinogenic risk of formaldehyde
using the CUT rat data  (Kerns et al. ,  1983) in the linearized
multistage model.  However, their assessment differs from EPA
(1987a) in the following two ways:

     (1)   The present approach uses the rate of binding of
          formaldehyde to DNA in the nasal lining of the rat, in
          order to characterize the dose rate.  The EPA in its
          1987 risk assessment decided to use administered dose
          (inhalation exposure) rather than estimated tissue dose
          for risk estimation purposes because their reviewers
          did not consider the tissue data then available for
          their assessment to be adequate.

       (2) The present approach uses three different scaling
          factors to extrapolate the equivalent dose rate from
          rats to humans.  EPA  (1987a)  did not specifically
          discuss the issue of scaling to extrapolate from
          rodents to humans for formaldehyde.

     The UCL for unit risk for lifetime exposure calculated by
CARB (1992b) using the methods and assumptions described above is
7.0xlO"3 ppm"1  (6.0xlO"6  [ug/m3] -1) .  The  two differences  in
methodology (i.e.,  target-site dose and scaling factor) result in
a doubling of the upper confidence limit (UCL) on the unit risk
calculated by EPA.   CARB  (1992b) did not calculate MLEs for
formaldehyde.   CARB also calculated a range of UCL for unit risks
based on the three scaling factors and two measures of exposure
                               6-42

-------
                                                        EPA-420-R-93-005
                                                            April 1993
to formaldehyde by inhalation.  This range is 0.3xlO"3  ppm"1 to
40xlO"3 ppm"1  (0.25xlO"6 to 33xlO"6  [ug/m3]-1) .

Occupational Safety and Health Administration  (OSHA)

     OSHA published a final rule  for occupational exposure  to
formaldehyde in 1987, in which they concluded that  formaldehyde
should be regarded as a "potential occupational carcinogen"
(OSHA, 1987).   The 1987 final rule differs only slightly  from  the
1985 proposed rule mentioned above.  With regard to the adequacy
of the available human data, in the 1985 proposed rule, OSHA has
not relied on the epidemiologic results to assess risk of
lifetime exposure of workers to formaldehyde.  However, in  1987,
OSHA stated that the evidence regarding human risk  of  exposure to
formaldehyde has become substantial.

     OSHA has also selected the CUT rat study  (Kerns  et  al. ,
1983)  as the basis for its risk assessment for formaldehyde.
OSHA,  like EPA and CARB, selected the linearized multistage model
to calculate lifetime risk of exposure to formaldehyde.   Unlike
CARB,  OSHA chose not to use a scaling factor and also  chose not
to use the pharmacokinetic model  relating DNA-formaldehyde
adducts to external exposure dose to estimate target-site dose.
OSHA  (1987)  concluded that the pharmacokinetic model,  as  it
presently exists, is greatly limited by the scarcity of data
identifying DNA protein- formaldehyde cross-links,  and it cannot
be presumed to predict overall human cancer risk resulting  from
long-term repeated exposures to formaldehyde.

     As a result of the data set  and low-dose extrapolation model
chosen, and the assumptions made with regard to scaling and
correct estimation of dose, OSHA  (1987) calculated  the following
lifetime risk of cancer per 100,000 workers:


Exposure level (ppm)     Maximum  Likelihood  Upper  Confidence
                         Estimate  (MLE)      Limit  (UCL)

3                           71                      834
2                           11.4                    534
1                            0.6                    264
0.5                          0.03                   132
0.1                          0.001                   26
Universities Associated for Research and Education  in  Pathology
(UAREP)

   This panel  reviewed  the  same  body of  literature  (UAREP,  1988)
as IARC (1987c) and EPA (1987) using a metanalysis  approach.  The
UAREP panel commented only on the determination of  causality.
Unlike the IARC and EPA, the UAREP panel did not attempt  to

                               6-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993
categorize the epidemiological evidence other than whether
causality could be established.  The panel concluded that a
causal relationship has not been established for cancer at any
site.  In addition, the panel noted that if such a causal
relationship exists, the excess risk must be small.  The panel
noted elevated risks in nasopharyngeal cancer with formaldehyde
exposure in several studies, and concluded that the evidence for
causality was weak.  With respect to observed excesses in nasal
cavity and sinus cancers and any formaldehyde exposures, several
studies suggest an approximate doubling of the risk, while other
studies could not exclude an elevation of the size.  Overall, the
panel concluded that the presence or absence of an association
could not be firmly established.  With respect to lung cancer,
the panel thought the evidence was not consistent and did not
indicate a causal association with formaldehyde exposure.

   For sites which  are not  directly in contact with
formaldehyde, the panel stated that the rapid metabolism of
formaldehyde makes it unlikely that formaldehyde is the agent
responsible for increased brain tumors observed in the group that
used formalin.   For the excesses in leukemia observed in several
studies of anatomists,  embalmers, and pathologists, the panel
concluded that socioeconomic factors influencing diagnosis may
explain the elevations observed in these groups.

"Epidemiological Evidence on the Relationship Between
Formaldehyde Exposure and Cancer" (Blair et al.  1990b)

   Blair et  al.  (1990b) performed a metanalysis on essentially
the same body of literature as reviewed by IARC (1987c), UAREP
(1988),  and EPA (1987)  with the addition of more recent findings,
either published or in press.  From this analysis, the authors
found excesses in deaths due to cancers of the nasal cavities,
nasopharynx, lung, and brain, and due to leukemia.  The
investigators believed that a causal role for formaldehyde was
most probable for cancers of the nasopharynx and,  to a lesser
extent,  the nasal cavities.  Blair et al.  (1990b)  derived their
support for the conclusion from statistically significant
increases in nasal cavity cancer risk, from the apparent
specificity of the association with squamous cell carcinoma, and
from histological changes in the nasal mucosa seen in industrial
studies which correspond to those observed in the rat.

   The investigators further concluded that the excesses in  lung
cancer were difficult to interpret due to inconsistencies among
studies and lack of trends with either level or duration of
exposure.  In addition, the excesses of leukemia,  brain, and
colon cancer observed among professionals were most likely not
related to formaldehyde since similar excesses were not observed
among the industrial workers.
                               6-44

-------
                                                          EPA-420-R-93-005
                                                              April 1993
"Quantitative  Cancer Risk Estimation  for Formaldehyde"  (Starr,
1990)

    Starr (1990)  calculated  cancer risks based on the  DPX (DNA
protein  cross-link)  experiments of  Casanova et al.   (1989) in
rats, and of Heck et al.  (1989) in  monkeys.  Using the  linear
multistage model,  Starr fit a  "three-stage" model using  rat DPX
levels interpolated from the DPX  experiment to correspond to  the
bioassay exposures of Kerns et al.  (1983).   Predicted risks
corresponding  to 0.1, 0.5,  and 1.0  ppm formaldehyde in  air, based
on the DNA-binding data for both  rats and monkeys are reproduced
in the table above.   Starr also did not address the non-nasal DPX
observed in monkeys in making his calculations.   Starr  concluded
that point estimates of human risk  (also called maximum
likelihood estimates, or MLEs) based  on DPX in monkeys were lower
than those based on airborne concentrations to rats (the basis of
EPA's 1987 unit  risk),  by as much as  1,500,000-fold.

Comparison of  Risk Estimates form Starr (1990),  Upper Bounds  and
Point (Maximum Likelihood)  Estimates3
Air Upper Bound and (MLE) Estimates
Cone . Several Formaldehyde Exposure
(ppm) Rat/1983b Rat/1989c
0.1 2 E-4e (3 E-7) 7 E-5 (2 E-9)
0.5 8 E-4 (3 E-5) 4 E-4 (3 E-7)
1.0 2 E-3 (3 E-4) 1 E-3 (6 E-6)
of Risk From
Measures
Monkev/1989d
8 E-6 (2 E-12)
4 E-5 (3 E-10)
1 E-4 (1 E-8)
    Continuous lifetime average exposure adjustment not used.
    Kern et al.  (1983) exposure concentrations  (ppm).
    DNA-protein cross  links  (pmol/mg DNA) from  Casanova et al.  (1989).
    Using the 1989 rat DNA=binding data for the dose-response relationship, and
    the Heck et al (1989)  DNA-protein cross-links for delivered dose at 0.1, 0.5,
    and 1.0 ppm.
    2 E-4 = 2 x 10~4
    Starr calculated a MLE  risk of 3 x 10"7 based  on the air
concentration  of  0.1 ppm administered to rats,  while the
corresponding  MLE human risk based  on monkey DPX was 2 x 10"12.
The differences between upper bounds  on risk were less dramatic,
the largest  difference being 25-fold  between an upper bound  rat
dosimetry-based risk of 2 x 10"4 and upper bound monkey
dosimetry-based risk of 8 x 10"6 at  an air concentration of  0.1
ppm.
                                6-45

-------
                                                        EPA-420-R-93-005
                                                            April 1993
6.6.3  Recent and Ongoing Research

6.6.3.1  Genotoxicity

   Recent  studies on the genotoxicity  of  formaldehyde  have
demonstrated the covalent binding and  induction of DNA strand
breaks, induction of chromosomal aberrations in vitro, co-
mutagenesis of formaldehyde and x-rays in Drosophila,
cytotoxicity and mutagenicity in human lymphocytes and Salmonella
and sister chromatid exchange in anatomy students exposed to
embalming solution  (Bogdanffy et al.,   1987; Casanova and Heck,
1987; Casanova et al.,  1989; Craft et  al., 1987;  Crosby, 1988;
Dresp and Bauchinger,  1988; Dowd et al.,  1986; Ecken and Sobels,
1986; Liber et al.,  1989; Heck and Casanova, 1987; Heck et al.,
1989; Schmid et al., 1986;  Snyder and Van Houten, 1986; Yager  et
al.,  1986).  These  studies have been reviewed by EPA  (1990a) and
it was concluded that they added nothing new or substantially
different to what was written in EPA (1987a) regarding the
genotoxic effects of formaldehyde.

6.6.3.2  Pharmacokinetics

   Recent  work on the pharmacokinetics of  formaldehyde has
focused on the validation of measurement of DNA-protein adducts
(DPX) as internal dosimeters of formaldehyde exposure.  In other
words, the binding  of DNA to protein to which formaldehyde is
bound to form a separate entity that can be quantified may serve
as a means to measure the amount of formaldehyde that  is present
inside a tissue.   An internal dosimeter for formaldehyde exposure
is desirable because the inhaled concentration of formaldehyde
may not reflect actual tissue exposure levels.  The difference in
inhaled concentration and actual tissue exposure level is due  to
the action of multiple defense mechanisms  (such as the protection
of underlying cells by the mucociliary apparatus) that act to
limit the amount of formaldehyde that  reaches cellular DNA.  At
issue is the rebuttal by EPA and the Science Advisory Review
Board  (summarized in EPA 1987a) of the assertion that DPX
measurements could  be used in quantitative cancer risk
assessments as an indication of intracellular dose (Starr and
Buck, 1984).   The rebuttal was based on EPA and the Science
Advisory Review Board's belief that inadequate evidence was
presented demonstrating that the method used to measure DPX was
valid, the measurement of DPX as an intracellular dosimetric
marker was adequate, and the results obtained in acute studies
that measured DPX could be extrapolated to the chronic exposure
situation.

6.6.3.3  Carcinogenicity - Animal Studies

   Recent  studies examining the carcinogenicity  of formaldehyde
in animals have further studied the characteristics of nasal

                               6-46

-------
                                                        EPA-420-R-93-005
                                                            April 1993
tumor induction as a result of inhalation of formaldehyde.
Morgan et al.  (1986) mapped the specific location of the nasal
squamous cell carcinomas that were observed in rats in the study
by Kerns et al.  (1983).   The authors proposed that in addition to
regional exposure, local tissue susceptibility may be an
important determinant for distribution of formaldehyde-induced
neoplasms.

   In the CUT  study  (Kerns et al. ,  1983), nasal  tumors were
induced by formaldehyde at concentrations that also induced
severe degenerative, hyperplastic, and metaplastic changes in the
nasal epithelium,  suggesting that cytotoxicity and/or increased
cell proliferation may have had a role in tumor induction.
Recent studies by Woutersen et al. (1989) and Feron et al. (1988)
and an ongoing study by Monticello and Morgan (1990) support the
association between cytotoxicity and cell proliferation and tumor
induction at exposures of 10 to 20 ppm, 6 hours/day, 5 days/week
at exposures ranging from 4 weeks to 28 months.  The study by
Woutersen et al.  (1989)  more directly examined the effect of
tissue damage on the tumorigenic response of formaldehyde.  These
authors found that external sources of damage to the nasal
epithelium could enhance the tumorigenic response of Wistar rats
to formaldehyde.

   One explanation  for  the increase  in tumor induction in areas
of tissue damage proposes that nasal defense mechanisms may be
irreparably damaged in such areas.  For example, the mucociliary
apparatus has been proposed to trap and remove formaldehyde in
the mucus layer before it has a chance to reach underlying cells
(Zwart et al.,  1988).   The tissue damage may prevent adequate
functioning of the mucociliary apparatus.  Both in vitro and in
vivo studies (Morgan et al.,  1983, 1986)  have shown that there is
a clear dose-dependent effect of formaldehyde on the mucociliary
apparatus of the rats.   In addition to tissue damage, exposure to
high concentrations of formaldehyde has been suggested to
interfere with the protective function of mucus.  A recent study
by Bogdanffy et al.  (1987) examined  [14C]-formaldehyde binding to
nasal mucus from rats and a human volunteer and found the
formaldehyde bound to albumin within the mucus.   These authors
postulated that formaldehyde binding to mucus may alter the
physical characteristics of mucus and lead to mucostasis.  This
would allow formaldehyde to penetrate to the submucosal cell
layer.  In humans, nasal mucociliary function was inhibited by
exposure to 0.3 ppm formaldehyde for 1 to 5 hours (Anderson and
Molhave,  1983).   It is also known that formaldehyde at levels
below 1 ppm can be detected in the olfactory region of the human
nose, indicating that formaldehyde is not completely removed by
the mucus layer, even at low concentrations.

   Another explanation  for the increase  in tumor  induction in
areas of tissue damage is that cytotoxicity may increase cell
proliferation thereby increasing the amount of single-stranded

                               6-47

-------
                                                        EPA-420-R-93-005
                                                            April 1993
DNA available for damage by formaldehyde.  Alternatively,
cytotoxicity may in some way promote the carcinogenic response in
formaldehyde-initiated cells.

   Cell proliferation  in response  to formaldehyde  has been
observed in human tissues and the monkey, as well as in the rat.
In studies by Klein-Szanto et al.  (1989) and Ura et al.  (1989)
human tracheobronchial epithelia were transplanted into
deepithelialized rat tracheas.  A concentration-dependent
proliferative response similar to that observed in the rat was
observed when the tracheas were exposed in vivo to devices that
slowly released formaldehyde.  Exposure of monkeys to 0 or 6 ppm
of formaldehyde for 1 or 6 weeks resulted in an 18-fold increase
in cell proliferation in formaldehyde exposed animals (Monticello
et al. ,  1989) .

   Although  increased  cell proliferation has been  observed  in a
number of studies in which nasal tumors have been induced,
stimulation of cell proliferation does not appear to be
sufficient to cause tumors.  For example, in the study by
Monticello and Morgan  (1990), although proliferation and
inflammation were observed at the same doses at which
carcinogenicity was observed, proliferation and inflammation were
not observed only at those sites at which tumors developed.
Also, Zwart et al.   (1988)  found that exposure of formaldehyde
produced patterns of cell proliferation that were not consistent
with carcinogenic patterns.  For example, after 3 days of
exposure to 3 ppm,  increases in cell proliferation were observed
in regions with a high tumorigenic response; but, after 13 weeks
the proliferation in these areas was slightly less than in
controls.  These acute (Swenberg et al., 1983; Zwart et al.,
1988) and chronic (Monticello and Morgan, 1990) studies have
demonstrated that there is a correlation between cytotoxicity and
cell proliferation induced by formaldehyde in the rat nasal
epithelium and that the cell proliferation rate is concentration-
dependent .

   The role  of concentration versus total dose  (i.e., the total
dose that an animal receives is the exposure concentration
multiplied by the duration of exposure)  in the response of
respiratory tissue to formaldehyde was examined in two studies by
Wilmer et al. (1987, 1989).  In both studies the Wistar rats were
exposed to formaldehyde on a continuous and intermittent basis
and the response appeared to be more dependent on concentration
than on total dose.

   A  number  of recent  studies have also examined the
carcinogenic potential of formaldehyde by the oral route.
Exposure of rats to 0.2% formaldehyde (0.001 to 0.25%) in the
drinking water produced squamous cell papillomas in the
forestomach  (Takahashi et al., 1986)  and an increase in gastric
neoplasms (squamous cell carcinomas,  adenocarcinomas, and

                               6-48

-------
                                                        EPA-420-R-93-005
                                                            April 1993
leiomyosarcomas  [a tumor of the smooth muscle tissue])   (Soffritti
et al.,  1989).   A slight increase in leukemia was also observed
in treated animals,  but the significance of this finding was not
addressed.  In contrast, no increase in tumors was observed in
Wistar rats exposed to formaldehyde  (0.002 to 0.5%) in the
drinking water (Til et al.,  1989; Tobe et al.,  1989) but
hyperplasia and inflammation of the forestomach and glandular
stomach were reported.  These results provide suggestive evidence
of carcinogenicity of formaldehyde by the oral route.

   A number  of recent studies have also examined the tumor
promotion potential of formaldehyde.   Using rat tracheal
explants,  Cosma and Marchok (1987) examined the effects of
formaldehyde, benzo[a]pyrene,  and the combination of these agents
on the induction of carcinogenesis.   Tracheal explants are
tracheal cells taken from a rat and grown in tissue culture
outside of the animal.  Carcinogenicity was quantified as the
number of growth altered populations observed per tracheal
explant.  Formaldehyde treatment  (0.2%) twice weekly for 4.5
months by itself produced only 0.25 altered populations per
explant, benzo[a]pyrene produced 2.37 altered populations per
explant, and pretreatment with benzo[a]pyrene followed by
formaldehyde treatment produced 7.83 populations per explant,
indicating the tumor promotion potential of formaldehyde.  Also,
in a skin painting experiment by Iversen  (1986),  hr/hr Oslo
strain mice were treated with 51.2 ug of the tumor initiator
dimethylbenz[a]anthracene (DMBA).  Nine days later, a group of
these mice was treated with 200 ul of 10% formaldehyde twice a
week for 60 weeks.  Although the incidence of tumors was similar
in DMBA treated animals both with and without formaldehyde
treatment (approximately 38%),  the time of appearance of the
tumors was significantly reduced in those mice treated with both
the formaldehyde and DMBA.   However,  a later experiment using
SENCAR mice  (bred for maximal sensitivity to carcinogens) found
no change in tumor induction when mice that had been pretreated
with 51.2 ug of DMBA were treated twice weekly with 4%
formaldehyde (Iversen, 1988).   Thus,  in some tissues formaldehyde
may have tumor promoting potential.

6.6.3.4  Carcinogenicity -  Epidemiological Studies

   Since  the 1987 EPA carcinogenicity  assessment,  a  limited
number of new epidemiologic studies and reanalyses of previous
studies have been published.  Many of the reanalyses have
examined the results of the largest study that has been published
to date (Blair et al., 1986, 1987).   This study examined the
mortality experience of 26,561 workers employed in a total of 10
plants known to use formaldehyde.  The estimated 8-hour time-
weighted-average exposure to formaldehyde fell into five
categories:  trace, <0.1 ppm, 0.1-0.5 ppm,  0.5-<2.0 ppm, and >2.0
ppm based on job category.   Blair et al.  (1986, 1987) reported
that workers exposed to >0.1 ppm formaldehyde had an elevated

                               6-49

-------
                                                         EPA-420-R-93-005
                                                              April 1993
rate of lung  cancer;  but,  that no increase  in  lung  cancer
incidence could  be  correlated with increases in  exposure.  Also,
these authors reported that workers with exposure to both
formaldehyde  and particulates had a dose-related elevated rate of
nasopharyngeal cancer.

    The  increase  in nasopharyngeal cancer was  reanalyzed by
Collins et al.  (1988)  and was reported to be confined to only one
of the 10 plants studied.   Also, the dose-related increase in
nasopharyngeal cancer originally reported by Blair  et al. (1986,
1987) was not seen  if only those workers with  simultaneous
exposure to particulates and formaldehyde were considered.  Blair
et al.  (1986,  1987)  had grouped exposure to formaldehyde and
particulates  irrespective of whether the exposures  had occurred
simultaneously.   Although Collins et al. (1988)  indicated that
these data showed a lack of an association  between  formaldehyde
exposure and  increased incidence of nasopharyngeal  cancer, EPA
(1990a) reevaluated the data using a Poisson trend  statistic and
found a significant trend for increased nasopharyngeal cancer
with increasing  formaldehyde exposure.

    Robins  et  al. (1988) reanalyzed the lung cancer  data using a
method developed to correct for the existence  of a  healthy worker
effect.  These authors confirmed the lack of an  association
between lung  cancer and increased formaldehyde exposure.
Sterling and  Weinkam (1988 1989) also reanalyzed the lung cancer
data from Blair  et  al.  (1986 1987)  using methods that would
reduce the influence of a healthy worker effect.  This included
using a time-integrated exposure score and  comparison of internal
high and low  exposure groups.  The report published in 1988
contained calculation errors and was amended in  1989.  The 1989
paper reported a significant increase in the odds ratio (OR)3  for
lung cancer in those over age 40 (40-55 yr, OR = 11.10,  95% CI4  =
6.45 to 1926;  55+ yr,  OR = 67.44, 95% CI =  35.59 to 127.59).  A
significant increase in the odds ratio for  lung  cancer for hourly
workers was also observed (OR = 1.61, 95% CI = 1.12 to 2.31).
Also, a significant trend for increased lung cancer incidence was
reported with increased cumulative exposure, although none of the
cumulative exposure levels was associated with a significant
increase in lung cancer incidence (<0.1 ppm-yr,  OR  = 1.0; 0.1-0.5
ppm-yr, OR =  1.21,  95% CI = 0.84 to 1,74; 0.5-2  ppm-yr,  OR =
1.19, 95% CI  = 0.78 to 1.83; 2+ ppm-yr, OR  = 1.56,  95% CI = 0.95
to 2.56).
     The odds ratio (OR) is an estimate of the relative risk (RR).   It is a
measure of association between the characteristic and disease in a case-control
study.  Relative risks (i.e., odds ratios)  that are >1 imply an association
between the characteristic and the disease.

    4The confidence interval (CI)  is the investigator's assurance that the
sample selected is one of 95% (for a 95% confidence interval) of all samples  that
will provide a correct statement based on the interval.

                               6-50

-------
                                                        EPA-420-R-93-005
                                                            April 1993
   Blair et al.  (1990) disputed  the  results  reported by  Sterling
and Weinkam (1989) based on the observation that Sterling and
Weinkam had grouped all respiratory cancer deaths rather than
examining only lung cancer deaths.  However, reanalysis using
only lung cancer deaths lowered the calculated risks,  but did not
affect the overall conclusions of Sterling and Weinkam. Further
analysis of the association between lung cancer and formaldehyde
exposure by Blair et al.  (1990) revealed that lung cancer
mortality was elevated in workers with formaldehyde and
particulate exposure from the production of resin and molding
compounds and that exposure to melamine, urea, phenol,  or wood
dust in these operations may have accounted for the increases in
lung cancer that were attributed to formaldehyde exposure.

   Two new case control studies  reported the  cancer mortality of
persons occupationally exposed to formaldehyde.  In the study by
Gerin et al.  (1989), an elevated odds ratio of 2.3 (95% CI = 0.9
to 6.0)  was determined for persons with adenocarcinoma of the
lung and long-duration, high exposure to formaldehyde.   The odds
ratio appeared to increase between those with long-duration low
level exposure (OR = 0.8,  95% CI = 0.3 to 1.3), those with long-
duration,  medium level exposure  (OR = 0.8,  95% CI = 0.4 to 1.6),
and those with long-duration high level exposure,  but this was
not statistically analyzed.  In the study by Roush et al. (1987),
an odds ratio of 2.3 (95% CI = 0.9 to 6.0)  was determined for
persons with nasopharyngeal cancer and occupational exposure to
formaldehyde at high levels 20 years prior to death.   The odds
ratio was statistically significant for those persons over 68
years of age (OR = 4.0, 95% CI = 1.3 to 12.0).  No such increase
was observed for persons with sinonasal cancer and occupational
exposure to high levels of formaldehyde 20 years prior to death
(OR = 1.5,  95% CI = 0.6 to 3.1).

   Another case-control study  (Partanen et  al., 1990)  examined
possible associations between formaldehyde and respiratory cancer
of 136 respiratory cancers among 7307 male Finnish woodworkers.
These men were employed in jobs in particleboard,  plywood,
construction carpentry, furniture manufacturing, and glue
manufacturing plants, and in sawmills.  After accounting for a
minimum latency period of 10 years,  smoking, and vital status at
the time of data collection, an elevated odds ratio for
respiratory cancer  (OR = 1.4, 90% CI = 0.4 to 4.1) was found with
exposure to cumulative formaldehyde  (either dustborne or as gas)
(>3 ppm-months).   When further analyzing upper respiratory cancer
and lung cancer separately, the odds ration for upper respiratory
cancer becomes OR = 2.4 and that of lung cancer OR = 0.9.
Partanen et al. (1990)  believed these results are compatible
either with chance or with a weak elevated risk mainly due to
cancers of the upper respiratory organs.

   Hayes et al.  (1990) conducted  a proportional mortality ratio
(PMR) study of embalmers and funeral directors in the U.S.

                               6-51

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Statistically significantly elevated proportions of deaths were
found from a variety of causes, specifically cancers of the
nasalpharyngeal, colon and lymphatic and hematopoietic systems.
There was also a significant increase in ischemic heart disease.
The authors believed the apparent elevated proportions of death
due to nasalpharyngeal cancer and leukemia were consistent with
previous observations in formaldehyde-exposed industrial cohorts
and other studies of professionals.

   Other related studies examining  cancer mortality among
workers exposed to formaldehyde include a population-based case
control study by Linos et al.   (1990) that observed an increase in
follicular non-Hodgkin's lymphoma and acute myeloid leukemia
among embalmers and funeral directors.  Also, Malker et al.
(1990) found a significant increase in nasopharyngeal cancer in
workers in fiberboard plants and among book binders (both are
subject to formaldehyde exposure).  A study of 9.365 leather
tannery workers reported 1 death due to squamous cell carcinoma
of the nasal cavity (0.4 expected) and attributed the death to 18
years of exposure to a variety of chemicals, including chrome and
formaldehyde (Stern et al.  1987).

   Histochemical analyses of biopsies taken  from nasal tissues
of workers exposed to formaldehyde revealed precancerous lesions.
Holmstrom et al. (1989) observed significant changes in the
middle turbinate of workers exposed to well-defined levels of
formaldehyde.  However, similar changes were not observed in
nasal tissues of workers exposed to formaldehyde and wood dust.
Boysen et al. (1990) also found a significant increase in the
degree of metaplasia in the nasal cavity of workers exposed to
formaldehyde.

   These new studies support the  previous conclusion by EPA
(1987a) that limited evidence of an association between
formaldehyde exposure and nasopharyngeal and, possibly lung,
cancer in humans exists.  No definitive causal relationships are
demonstrated in the new studies.

6.7  Carcinogenic Risk for Baseline and Control Scenarios

   Table 6-12 summarizes the annual cancer  incidences for all
the scenarios.   When comparing cancer incidence for the base
control scenarios relative to 1990, there is a 36% reduction in
1995, a 52% reduction in 2000,  and 50% in 2010, which is actually
an increase when compared to 2000.  The reduction in emissions
are considerably higher, particularly in the out years.   The
projected increase in both population and vehicle miles traveled
(VMT) from 2000 to 2010 appears to offset the gains in emissions
achieved through fuel and vehicles modifications.
                               6-52

-------
                                                        EPA-420-R-93-005
                                                            April 1993
   From Table  6-12  it  can  also  be  observed that the expanded use
scenarios provide no decrease in the cancer cases and,  in one
scenario,  the cancer cases increase slightly.   This  is  generally
due to the fact that increased use of oxygenates in  gasoline will
increase formaldehyde emissions.  The HAPEM-MS  exposure model
estimates exposure based on direct emissions of formaldehyde.  As
discussed in Section 6.5.2, however, the use of oxygenates  in
gasoline is expected to change the reactivity of the emissions.
It is probable that secondary (i.e., atmospherically formed)
formaldehyde could be reduced with the use of oxygenates.   As  a
result, the cancer risk estimates given in Table 6-11 should be
considered conservative estimates.

   Please note that the cancer  unit risk  estimate  for
formaldehyde is based on animal data and is considered  an upper
bound estimate for human risk.  True human cancer risk  may  be  as
low as zero.
                               6-53

-------
                                                                                  EPA-420-R-93-005
                                                                                       April 1993
Table 6-12.  Annual Cancer  Incidence Projections for Formaldehyde.
                                                                   a,b
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0412
0.0234
0.0251
0.0162
0.0166
0.0168
0.0140
0.0143
0.0138
Urban
Cancer
Cases
37
24
25
18
19
19
19
20
19
Rural
Cancer
Cases
7
4
5
3
3
3
3
4
3
Total
Cancer
Cases
44
28
30
21
22
22
22
24
22
Percent Reduction
from 1990
EF
-
43
39
61
60
59
66
65
67
Cancer
-
36
32
52
50
50
50
45
50
""Projections have inherent uncertainties in emission  estimates,  dose-response, and
exposure.
bCancer incidence estimates are based on upper bound  estimates of unit risk,  determined
from animal studies.  EPA  has  classified formaldehyde as a Group Bl, probable human
carcinogen based on limited epidemiological  evidence and sufficient evidence  in  animal
studies.
                                            6-54

-------
                                                        EPA-420-R-93-005
                                                            April 1993

6.8 Non-carcinogenic Effects of Inhalation Exposure to
Formaldehyde

   Since the  focus of  this  report  is  on  the  carcinogenic
potential of the various compounds, the noncancer information
will be dealt with in a more cursory  fashion.  No attempt has
been made to synthesize and analyze the data encompassed below.
Also, no attempt has been made to accord more importance to one
type of noncancer effect over another.  The objective is to
research all existing data, describe  the noncancer effects
observed, and refrain from any subjective analysis of the data.

   Irritation of the eyes  (lacrimation and increased  blinking)
and mucous membranes is the principal effect of exposure to low
concentrations  (0.05-2.0 ppm) of formaldehyde observed in humans
(NRC, 1981).  Other human upper respiratory effects associated
with acute formaldehyde exposure include a dry or sore throat,
and a tingling sensation of the nose. These effects are
frequently seen following exposure to 1-11 ppm  (NRC,  1981) .
Sensitive humans may detect effects at lower concentrations
(CARB, 1991b). Tolerance to eye and upper airway irritation may
develop after 1-2 hours exposure, but symptoms may return if
exposure is resumed following an interruption (NRC, 1981).  Nasal
mucocillary clearance system effects  (loss of cilia,  keratosis,
mild dysplasia)  have been reported in humans at concentrations of
0.1 ppm  (Edling et al., 1985), and following chronic  exposure to
undetermined concentrations  (NRC, 1981).   Forty percent of
formaldehyde-producing factory workers reported nasal symptoms
such as rhinitis, nasal obstruction,  and nasal discharge
following chronic exposure  (Wilhelmsson and Holmstrom, 1987).  In
persons with bronchial asthma, the upper respiratory  irritation
caused by formaldehyde can precipitate an acute asthmatic attack,
sometimes at concentrations below 5 ppm  (Burge et al., 1985);
formaldehyde exposure may also cause bronchial asthma-like
symptoms in nonasthmatics  (Hendrick et al.,  1982; Nordman et al.,
1985).  However, it is unclear whether asthmatics are more
sensitive than nonasthmatics to formaldehyde's effects  (EPA,
1990a).   Lower airway irritation, characterized by cough,
wheezing, and chest tightness, has been reported often in people
chronically exposed to 5-30 ppm formaldehyde, and has been
observed in concentrations below 1 ppm (EPA,  1987a).  However,
acute exposure did not cause lower airway symptoms in medical
students in an anatomy laboratory  (Uba et al.,  1989).  Neither
lower airway irritation (cough, chest symptoms,  and dyspnea
[labored or difficult breathing]) nor decrements in pulmonary
functioning were more frequently reported among asthmatics than
among nonasthmatics (Uba et al.,  1989).   Formaldehyde
concentrations exceeding 50 ppm may cause severe lower
respiratory tract reactions, in which not only the airways, but
also the alveolar tissue is involved.  This acute injury includes
pneumonia,  bronchial inflammation, pulmonary edema, and, at
concentrations exceeding 100 ppm, death may occur in  sensitive
individuals (NRC, 1981).  Pulmonary effects,  as measured by
pulmonary function tests,  have not been reported consistently
across studies.   Overall,  chronic decrements in lung  function do
not appear to be associated with formaldehyde exposure  (Witek et
al.,  1987;  Sauder et al.,  1987),  although small transient

                               6-55

-------
                                                        EPA-420-R-93-005
                                                            April 1993

decreases have been noted  (Sauder et al.,  1986; Horvath et al.,
1988) .

   Immune stimulation may  occur  following formaldehyde  exposure,
although conclusive evidence is not available.  Patterson et al.
(1986)  demonstrated the presence of IgE antibodies against
formaldehyde-human serum albumin conjugates and human serum
albumin  (HSA).   IgE (immunoglobulin gamma E) is a protein
antibody produced by cells of the lining of the respiratory and
intestinal tract. It appears that formaldehyde is capable of
inducing respiratory tract allergy, but data are lacking on
induction concentrations (Burge et al.,  1985; Nordman et al.,
1985).   Central nervous system effects  such as dizziness, apathy,
inability to concentrate, and sleep disturbances have been
reported in a variety of studies following inhalation exposure in
humans (EPA, 1987a).   However, in general,  formaldehyde's effect
on the CNS is not clearly defined  (Consensus Workshop, 1984).

   With  regard to the developmental toxicity of  formaldehyde,
menstrual disorders were reported among 47.5% of women
occupationally exposed to formaldehyde vapors from urea-
formaldehyde resins,  with dysmenorrhea  (pain in association with
menstruation)  being the most common disorder (Shumilina, 1975).
There have been several animal inhalation studies conducted to
assess the developmental toxicity of formaldehyde.  The only
exposure-related effect noted in a study conducted by Martin
(1990)  observed a decrease in maternal body weight gain at the
high-exposure level but no adverse effects on reproductive
outcome or the fetuses that could be attributed to treatment were
noted.   In another study conducted by Sallenfait et al.  (1989),
reduced fetal weight was noted following exposure of pregnant
Sprague Dawley rats to 20 or 40 ppm formaldehyde on gestations
days 6-20.  No effects on embryonic or  fetal lethality, or in the
external, visceral,  or skeletal appearance of the fetuses were
noted.   In Ulsamer et al. 1984, other effects such as increased
duration of gestation and body weight of offspring, microscopic
changes in the liver,  kidneys, and other organs of fetuses from
exposed dams,  and decreased levels of nucleic acid in the testes
of exposed males have been reported.

   Acute and subchronic  inhalation  exposure of various
laboratory animals to low  (<1 ppm) or moderate (10-50 ppm)
concentrations of formaldehyde vapor is known to cause increased
airway resistance,  decreased sensitivity of the nasopalatine
nerve (a nerve that innervates both the nose and the palate),
irritation of the eyes and respiratory  system, and changes in the
hypothalamus (a part of the brain that  is important in
controlling certain metabolic activities such as maintenance of
water balance,  sugar and fat metabolism regulation of body
temperature and secretion of hormones).   Exposure to high
concentrations (>100 ppm) of formaldehyde vapor can cause
salivation,  acute dyspnea,  vomiting, cramps, and death  (CIR,
1984).   Subchronic and chronic inhalation exposure in
experimental animals has resulted in a variety of nasal cavity
lesions,  including dysplasia  (abnormal development of tissue) and
squamous metaplasia (conversion of one kind of tissue into a form
that is not normal for that tissue) of  respiratory epithelium,

                               6-56

-------
                                                        EPA-420-R-93-005
                                                            April 1993

purulent or seropurulent rhinitis  (an inflammation of the nasal
tissue characterized by discharges that contain pus or serum and
pus),  interstitial inflammation of the lungs  (CIR, 1984), reduced
weight gain, reduced liver weights, and lesions of the kidney,
liver, cerebral cortex, and respiratory tract  (EPA, 1985c).
Rusch et al.,   (1983)  determined a NOAEL for squamous metaplasia
and rhinitis of 1.0 ppm for rats and monkeys, although the study
duration was not specified.  Effects on the liver  (decreased
liver weights, histological changes) and kidney  (vasodilation in
a part of the renal cortex that is near the renal medulla)
effects were also seen in animals following subchronic inhalation
exposure (Rusch et al., 1983; Feldman and Bonashevskaya, 1971).
In animals, formaldehyde has been shown to affect the firing rate
of certain nerves in the nasal sensory system  (EPA, 1987a).  At
high concentrations,  formaldehyde has been reported to cause
cerebral acid proteinase activity in rats in one study and
decrease in cerebral RNA concentration, together with decreases
in the succinate dehydrogenase and acid proteinase activities, in
another  (Consensus Workshop, 1984) .

   A  range  of  predicted responses  for upper  respiratory  and  eye
irritation risk, for a given formaldehyde concentration  is
obtained when seven studies are examined comparatively  (Bender et
al., 1983;  Hanrahan et al., 1984; Horvath et al., 1988;  Kulle,
1985;  Liu et al.,  in press; Anderson and Molhave, 1984;  Ritchie
and Lehnen, 1987).  Caution must be taken in inferring the
results in EPA  (1987) and in this data to the general population.
Limitations in these studies at the present prevent the  inference
of eye and upper respiratory risks.  None of the studies reviewed
in this document and in EPA  (1987) provide adequate data to
precisely quantify general population risks for eye and  upper
respiratory effect associated with a specific formaldehyde
concentration.  Nevertheless, these studies document eye and
upper respiratory tract effects at levels previously identified ,
0.1 ppm to 3.0 ppm.  Even though the prevalence of exposure can
not be precisely estimated for a given formaldehyde
concentration, these studies support the conclusion that the
number of individuals responding in a population will increase
with increasing formaldehyde concentration.

   An inhalation reference  concentration  (RfC) is  not  available
for formaldehyde at this time.  EPA (1992b) has derived  an oral
reference dose  (RfD)  of 2X10"1 mg/kg/day, based on  reduced weight
gain and histopathology in rats following a 2-year drinking water
study (Til et al., 1989).   An uncertainty factor of 100  and a no-
observed-adverse-effect level (NOAEL)  of 15 mg/kg/day in male
rats were used to derive the RfD.

   A  recent study by Krzyzanowski  et al.  (1990)  analyzed the
relation of chronic respiratory symptoms and pulmonary function
to indoor formaldehyde exposure in a sample of children  and
adults in Tucson,  Arizona.   The average concentration of
formaldehyde,  measured in 202 households,  was 26 ppb (32 ug/m3) .
In only a few cases did the formaldehyde exceed 90 ppb  (111
ug/m3) ,  with a maximum  value of  140 ppb (172  ug/m3) .  The data
were collected from 298 children and 613 adults.
                               6-57

-------
                                                        EPA-420-R-93-005
                                                             April 1993

    In children,  the  prevalence  rates  of chronic respiratory
symptoms were not related to the formaldehyde exposure
(considered in three categories:  below 40 ppb,  41-60 ppb,  and
over 60 ppb).   However, the diseases  diagnosed  by  a  doctor,
asthma and chronic bronchitis in children 6-15  years of  age,  were
more prevalent in  houses with  formaldehyde  levels of 60-120 ppb
(74-148 ug/m3)  than in those children less exposed.  This is
especially evident in children  also exposed  to  environmental
tobacco smoke.  The effects in  asthmatic children  exposed to
formaldehyde below 50 ppb  (62 ug/m3)  were greater  than in healthy
ones.  The effects in adults were less  evident:  decrements in
expiratory flow rates due to formaldehyde over  40  ppb  (49 ug/m3)
were seen only in the morning,  and mainly in smokers.  This
childhood data are considered signs of  developmental toxicity
since the definition of developmental    toxicity  includes
children up to the time of puberty.
                               6-58

-------
                                                        EPA-420-R-93-005
                                                            April 1993
6.9  References for Chapter 6
Acheson, E.D., M.J. Gardner, B. Pannett, H. Barnes, C. Osmond,
and C. Taylor.  1984.  The Lancet 1:611.

Albert, R.E., A.R. Sellakumar, S. Laskin, M. Kuschner, N. Nelson,
and D.A. Snyder.  1982.  Gaseous formaldehyde and hydrogen
chloride induction of nasal cancer in the rat.  J. Natl. Cancer
Inst. 68:597-603.

Anderson, I. and L. Molhave.  1984.  Controlled human studies
with formaldehyde.  In:  Gibson, J.E. (ed.) "Formaldehyde
Toxicity",  Hemisphere Publishing Corp.,  New York. pp. 154-165.

Auto/Oil Air Quality Improvement Research Program.  1990.  Phase
1 Working Data Set (published in electronic form).  Prepared by
Systems Applications International, San Rafael, CA.

Auto/Oil Air Quality Improvement Research Program.  1991.
Technical Bulletin No. 6: Emission Results of Oxygenated Gasoline
and Changes in RVP.

Bender, J.R., L.S. Mullin, G.J. Graepel, and W.E. Wilson.  1983.
Eye irritation response of humans to formaldehyde.  Am. Ind. Hyg.
ASSOC. J. 44:  463-465.

Bertazzi, P.A., C. Zochetti, A. Pesatori, L. Radice, and T. Vai.
1986.  Exposure to formaldehyde and cancer mortality in a cohort
of workers producing resins.  Scand. J.  Work Environ. Health 12:
461-468.

Blair, A.,  P. Stewart, M. O'Berg, W. Gaffey, J. Walrath, J. Ward,
R. Bales, S. Kaplan,  and D. Cubit.  1986.  Mortality among
industrial workers exposed to formaldehyde.  JNCI 76:1071-1084.

Blair, A.,  P. Stewart, R.N. Hoover, J.F. Fraumeni, J. Walrath, M.
O'Berg, and W. Gaffey.  1987.  Cancers of the nasopharynx and
oropharynx and formaldehyde exposure.  JNCI 78:191-192.

Blair, A.,  P.A. Stewart, and R.N. Hoover.  1990a.  Mortality from
lung cancer among workers employed in formaldehyde industries.
Am. J. Indust. Med. 17:683-699.

Blair, A.,  R. Saracci, P.A. Stewart, R.B. Hayes, and C. Shy.
1990b.  Epidemiologic evidence on the relationship between
formaldehyde exposure and cancer.  Scand. J. Work Environ. Health
16:  381-393.

Boekhaus, K. L., L. K. Cohu, L. A. Rapp and J. S. Segal.  1991a.
Clean Fuels Report 91-02:  Impact of EC-1 Reformulated Gasoline
Emissions and Their Reactivity on Five 1989 Cars.  Arco Products
Co., Anaheim, California.

Boekhaus, K. L., J. M. DeJovine, D. A. Paulsen, L. A. Rapp, J.S.
Segal and D. J. Townsend.  1991b.  Clean Fuels Report 91-03:
Fleet Test Emissions Data -- EC-Premium Emission Control
Gasoline.  Arco Products Co., Anaheim, California.

                               6-59

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Bogdanffy, M.S., P.H. Morgan, T.B. Starr, and E.T. Morgan.  1987.
Binding of formaldehyde to human and rat nasal mucus and bovine
serum albumin.  Toxicol.  Lett. 38:145-154.

Boysen, M., E. Zadig, V.  Digernes, V. Abeler, A. and Reith.
1990.  Nasal mucosa in workers exposed to formaldehyde: a pilot
study.  Br. J. Indust.  Med. 47:116-121.

Burge, P.S., M.G. Harries, W.K. Lam, I.M. O'Brien, and P.A.
Patchett. 1985.  Occupational asthma due to formaldehyde.  Thorax
40:225-260.

CARB.  1992a.  Proposed identification of formaldehyde as a toxic
air contaminant.  Part A  Exposure assessment.  California Air
Resources Board, Stationary Source Division.  January, 1992.

CARB.  1992b.  Proposed identification of formaldehyde as a toxic
air contaminant.  Part B Health assessment.  California Air
Resources Board, Stationary Source Division.  January, 1992.

Carey, P.M.  1987.  Air toxics emissions from motor vehicles.
Ann Arbor, MI:  U.S. Environmental Protection Agency, Office of
Mobile Sources, EPA Report no. EPA-AA-TSS-PA-86-5.

Casanova, M., D.F. Deyo,  and Hd'A. Heck.  1989.  Covalent binding
of inhaled formaldehyde to DNA in the nasal mucosa of Fischer-344
rats:  Analysis of formaldehyde and DNA by high-performance
liquid chromatography and provisional pharmacokinetic
interpretation.  Fund.  Appl.  Toxicol. 12:397-417.

Casanova, M. and Hd'A.  Heck.   1987.  Further studies of the
metabolic incorporation and covalent binding of inhaled 3H and
14C formaldehyde in Fischer-344 rats: Effects of glutathione
depletion.  Toxicol. Appl. Pharmacol. 89:105-121.

Casanova, M. and Hd'A.  Heck.   1991.  The impact of DNA-protein
cross-linking studies on quantitative risk assessments of
formaldehyde.  CUT Activities, Vol. 11, No. 8.

Casanova-Schmitz, M., T.B. Starr, and Hd'A. Heck.  1984.
Differentiation between metabolic incorporation and covalent
binding in the labeling of macromolecules in the rat nasal mucosa
and bone marrow by inhaled 14C and 3H formaldehyde.  Toxicol.
Appl. Pharmacol. 76:26-44.
                               6-60

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Casanova, M., K.T. Morgan, W.H. Steinhagen, J.I. Everitt, J.A.
Popp, and Hd'A. Heck.  1991.  Covalent binding of inhaled
formaldehyde to DNA in the respiratory tract of Rhesus monkeys :
Pharmacokinetics, rat-to-monkey interspecies scaling, and
extrapolation to man.  Fundam. Appl . Toxicol .  (in press).

CIR.  1984.  Cosmetic ingredient review.  Final report on the
safety assessment of formaldehyde.  J. Am. Coll. Toxicol. 3:157-
184.

Clement International Corporation.  1991.  Motor vehicle air
toxics health information.  For U.S. EPA Office of Mobile
Sources, Ann Arbor, MI:  September  1991.

Collins, J.J., J.C. Caporossi, and  H.M.D. Utidjian.  1988.
Formaldehyde exposure and nasopharyngeal cancer : re-examination of
the National Cancer Institute study and an update of one plant.
JNCI 80:376-377.

Colorado Department of Health.  1987.  Unpublished data from a
motor vehicle emissions toxics study of regulated and non-
regulated pollutants.  Aurora Emission Technical Center, Aurora,
Colorado.

Consensus Workshop on Formaldehyde.  1984.  Final report:
Deliberations of the consensus workshop on formaldehyde, October
3-6, 1983, Little Rock, AK.

Cosma, G.N. and A.C. Marchok.  1987.  The induction of growth
altered cell population (tumor-initiation sites) in rat tracheal
implants exposed to benzo (a) pyrene  and formaldehyde.
Carcinogenesis 8:1951-1953.

Craft, T.R., E. Bermudez,  and T.R.  Skopek.  1987.  Formaldehyde
mutagenesis and formation if DNA-protein crosslinks in human
lymphoblasts in vitro.  Mutat . Res. 176:147-155.

Crosby, R.M., K.K. Richardson, T.R. Craft, K.B. Benforado, H.L.
Liber, and T.R. Skopek.  1988.  Environ. Mol . Mutagenesis 12:155-
166.
Cupitt,  L. T.  1980.
the air environment."
600/3-80-084) .
                      "Fate of toxic and hazardous materials in
                       U.S. Environmental Protection Agency  (EPA-
Dalbey, W.E.  1982.  Formaldehyde and tumors in hamster
respiratory tract.  Toxicology 24:9-14.

DeJovine, J. M., K. J. McHugh, D. A. Paul sen, L. A. Rapp, J. S.
Segal, B. K. Sullivan, D. J. Townsend.  1991.  Clean Fuels Report
91-06:  EC-X Reformulated Gasoline Test Program Emissions Data.
Arco Products Co., Anaheim, California.

Dodge, M. C.  1990.  Formaldehyde production in photochemical
smog as predicted by three state-of -the-science chemical oxidant
mechanisms.  J. Geophys .  Res., 95:3635-3648.
                               6-61

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Dowd, M.A., M.E. Gaulden, B.L. Proctor, and G.B. Seiber.  1986.
Formaldehyde-induced acentric chromosome fragments and chromosome
stickiness in Chortophaga neuroblasts.  Environ. Mutagenesis
8:401-411.

Dresp, J. and M. Bauchinger.  1988.  Direct analysis of the
clastogenic effect of formaldehyde in unstimulated human
lymphocytes by means of the premature chromosome condensation
technique.  Mutat.  Res. 204:349-352.

Ecken, J.C.J. and F.H. Sobels.  1986.  The effect of X-
irradiation and formaldehyde treatment on spermatogonia on the
reversion of an unstable, P-element insertion mutation in
Drosophila melanogaster.  Mutat. Res. 1765:61-65.

Edling, C., L. Odkvist, H. Hellquist.  1985.  Formaldehyde and
the nasal mucosa.  Br. J. Ind. Med. 42:570-571.

Environ Corporation.  1986.  Improving the assessment of risk of
exposure to formaldehyde.  Prepared for the Motor Vehicle
Manufacturers Association.

Environ Corporation.  1987.  Risk assessment issues in EPA's
technical report "Air Toxics Emissions from Motor Vehicles".
Prepared for the Motor Vehicle Manufacturer Association.

EPA.  1985.  Health and Environmental Effects Profile for
Formaldehyde.  Cincinnati OH:  U.S. Environmental Protection
Agency, Environmental Criteria and Assessment Office.  October
1985.  ECAO-CIN-P142.  Final Draft.

EPA.  1987a.  Cancer Risk Due to Indoor and Outdoor Sources of
Formaldehyde.  Memo from Charles L. Gray to Richard D. Wilson.
July 16, 1987.

EPA.  1987b.  Assessment of health risks to garment workers and
certain home residents from exposure to formaldehyde.  April
1987.  Office of Pesticides and Toxic Substances.

EPA.  1991a.  Locating and estimating air emissions from sources
of formaldehyde  (revised).  Office of Air Quality Planning and
Standards, Research Triangle Park, NC.  Report no. EPA-450/4-91-
012.c

EPA.  1991b.  Formaldehyde risk assessment update.  June 11,
1991.  Office of Toxic Substances, U.S. Environmental Protection
Agency, Washington, DC.  External review draft, June 11, 1991.


EPA. 1992a.  Calculation of exhaust and evaporative toxic
emissions mass fractions for the motor vehicle related air toxics
report.  Draft memo from Richard Cook to Phil Lorang.  March 9,
1992.

EPA.  1992b.  Integrated Risk Information System.  U.S.
Environmental Protection Agency.  Office of Health and
                               6-62

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Environmental Assessment, Environmental Criteria and Assessment
Office, Cincinnati, OH.

Feldman, Y.G. and T.I. Bonashevskaya.   1971.  On the effects of
low concentrations of formaldehyde.  Hyg. Sanit. 36:174-180.

Feron, V.J., J.P. Bruijntjes, P.A. Woutersen, H.R. Immell, and
L.M. Appelman.  1988.  Nasal tumors in rats after short-term
exposure to a cytotoxic concentration of formaldehyde.  Cancer
Lett. 39:101-111.

Gerin, M.,  J. Siemiatycki, L. Nadon, R. Dewar, and D. Krewski.
1989.  Cancer risks due to occupational exposure to formaldehyde:
Results of a multi-site case-control study in Montreal.   Int. J.
Cancer 44:53-58.

Hanrahan, L.P., K.A. Dally, H.A. Anderson, M.S. Kanarak,  and J.
Rankin.  1984.  Formaldehyde vapor in mobile homes:  A cross
sectional survey of concentrations and irritant effects.  Am. J.
Public Health 74:  1026-1027.

Hardell, L., B. Johansson, and D. Axelson.  1982.  Amer.  J. Ind.
Med.  3: 247-257.

Hayes, R.B., A. Blair, P.A. Stewart, R.F. Herrick, and H. Mahar.
1990.  Mortality of US embalmers and funeral directors.   Am. J.
Ind. Med. 18:  641-652.

Hayes, R.B., J.W. Raatgever, and M. Gerin.  1986.  Int. J. Cancer
37: 487-492.

Heck, Hd'A. and M. Casanova.  1987.  Isotope effects and  their
implications for the covalent binding of inhales 3H and 14C
formaldehyde in the rats nasal mucosa.  Toxicol. Appl. Pharmacol.
89:122-134.

Heck, Hd'A., M. Casanova, W.H. Steinhagen, J.I. Everitt,  E.T.
Morgan, and J.A. Popp.  1989.  Formaldehyde toxicity.
DNA_protein crosslinking studies in rats and nonhuman primates.
In: Feron VJ and Bosland MC  (eds.), "Nasal Carcinogenesis in
Rodents:  Relevance to Human Health Risk", Pudoc Wagenigen, The
Netherlands, pp. 159-164.

Hendrick, D.J., R.J. Rando, D.J. Lane, and M.J. Morris.   1982.
Formaldehyde asthma: Challenge exposure levels and fate after
five years.  J. Occup. Med. 24:893-897.


Holmstrom,  M., B. Wilhelmsson, H. Hellquist, and G. Rosen.  1989.
Histological changes in the nasal mucosa in persons
occupationally exposed to formaldehyde alone and in combination
with wood dust.  Acta. Otolaryngol. (Stockh) 107:102-129.

Horton, A.W., R. Tye, and K.L. Stemmer.  1963.  Experimental
carcinogenesis of the lung.  Inhalation of gaseous formaldehyde
or an aerosol of coal tar by CeH mice.  J. Natl. Cancer Inst. 30:
31-43.

                               6-63

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Horvath, E.P., H. Anderson W.E. Pierce, L. Hanrahan, and J.D.
Wendlick.  1988.  Effects of formaldehyde on the mucous membranes
and lungs.  J. Am. Med. Assoc. 259:  701-707.

IARC.  1982.  IARC Monographs on the Evaluation of Carcinogenic
Risk of Chemicals to Humans.  Volume 29.  Some Industrial
Chemicals and Dyestuffs.  International Agency for Research on
Cancer.  World Health Organization, Lyon, France.  p. 345-389.

IARC.  1987.  IARC Monographs on the Evaluation of Carcinogenic
Risk of Chemicals to Humans.  Supplement 7.  Overall Evaluations
of Carcinogenicity:   An updating of IARC monographs volumes 1 to
42.  International Agency for Research on Cancer.  World Health
Organization, Lyon,  France.  p. 211-216.

Iversen, O.H.  1986.  Formaldehyde and skin carcinogenesis.
Environ. Internat. 12:541-544.

Iversen, O.H.  1988.  Formaldehyde and skin tumorigenesis  in
SENCAR mice.  Environ. Internat. 14:23-27.

Jacob,  D. J.  1986.   Chemistry of OH in remote clouds and  its
role in the production of formic acid and peroxymonosulfate.  J.
Geophys. Res., 91(D9)  :9807-9826 .

Jeffries, H. E. and K. G. Sexton.  1987.  "Technical Discussion
Related to the Choice of Photolytic Rates for Carbon Bond
Mechanisms in OZIPM4/EKMA."  U. S. Environmental Protection
Agency  (EPA-450/4-87-003).

Kerns,  W.D., K.L. Pavkov, D.J. Donofrio, E.J. Gralla, and  J.A.
Swenberg.  1983.  Carcinogenicity of formaldehyde in rats  and
mice after long-term inhalation exposure.  Cancer Res. 43:4382-
4392.

Klein-Szanto, A.J.P.,  H. Ura, and J. Resau.  1989.  Formaldehyde
induced lesions of xenotransplanted human nasal respiratory
epithelium.  Toxicol.  Pathol. 17:33-37.

Krzyzanowski, M., J.J. Quackenboss, and M.D. Lebowitz.  1990.
Chronic respiratory effects of indoor formaldehyde exposure.
Environ. Res.  52:117-125.

Kulle,  T.J.  Letter to C.S. Scott, U.S. EPA, August 28, 1985.

Liber,  H.L., K. Benforado, R.M. Crosby, D. Simpson, and T.R.
Skopek.  1989.  Formaldehyde-induced and spontaneous alterations
in human hprt DNA sequence and mRNA expression.  Mutat. Res.
226:31-37.

Ligocki, M.P.  1992.  Review of EPA memorandum on calculation of
toxic emission mass fractions.  Prepared for Motor Vehicle
Manufacturers Association.  Systems Applications International,
San Rafael, California  (SYSAPP-92/075).

Ligocki, M.P., G.Z.  Whitten, R.R. Schulhof, M.C. Causley,  and
G.M. Smylie.  1991.   Atmospheric transformation of air toxics:

                               6-64

-------
                                                        EPA-420-R-93-005
                                                            April 1993

benzene, 1,3-butadiene, and formaldehyde.  Systems Applications
International, San Rafael, California  (SYSAPP-91/106).

Ligocki, M.P., R.R. Schulhof, R.E. Jackson, M.M. Jimenez, G. Z.
Whitten, G.M. Wilson, T.C. Meyers, and J.L. Fieber.   1992.
Modeling the effects of reformulated gasolines on ozone and
toxics concenetrations in the Baltimore and Houston areas.
Systems Applications International, San Rafael, California
(SYSAPP-92/127).

Linos, A.,  A. Blair, K.P. Cantor, L. Burmeister, S. VanLier, R.W.
Gibson, L.  Schuman, and G. Everett.  1990.  Leukemia  and non-
Hodgkin's lymphoma among embalmers and funeral directors.  JNCI
82:66.

Liu, K.S.,  F.Y. Huang, S.B. Hayward, J. Wesolowski, and K.
Sexton.  Irritant effects of formaldehyde exposure in mobile
homes.  Environ.  Hlth. Perspec.   (in press).

Malker, H.S.R., J.K. McLaughlin, J.A. Weiner, D.T. Silverman,
W.J. Blot,  J.L.E. Ericsson, and J.F. Fraumeni.  1990.
Occupational risk factors for nasopharyngeal cancer in Sweden.
Br. J. Indust. Med. 47:213-214.

Martin, L.R., M.P. Easton, J.W.  Foster, and M.W. Hill.  1989.
Oxidation of hydroxymethanesulfonic acid by Fenton's  reagent.
Atmos. Environ.,  23:563-568.

Martin, W.J.  1990.  A teratology study of inhaled formaldehyde
in the rat.  Reproductive Toxicology 4:237-239.

McArdle, J. V., and M. R. Hoffmann.  1983.  Kinetics  and
mechanism of the oxidation of aquated sulfur dioxide  by hydrogen
peroxide at low pH.  J. Phys. Chem., 87:5425-5429.

Monticello, T.N., E.T. Morgan, J.I. Everitt, and J.A. Popp.
1989.  Effects of formaldehyde gas on the respiratory tract of
Rhesus monkeys.  Am. J. Pathol.  134:515-527.

Monticello, T.N.  and K.T. Morgan.  1990.  Correlation of cell
proliferation and inflammation with nasal tumors in F-344 rats
following chronic formaldehyde exposure.  Am. Assoc.  Cancer Res.
31:139.

Morgan, E.T., X-Z. Jiang, T.B. Starr, and W.D. Kerns.  1986.
More precise localization of nasal tumors associated  with chronic
exposure of F-344 rats to formaldehyde gas.  Toxicol. Appl.
Pharmacol.  82:264-271.

Munger, J.W., D.J. Jacob, and M.R. Hoffmann.  1984.   The
occurrence of bisulfite-aldehyde addition products in fog- and
cloudwater.  J. Atmos. Chem., 1:335-350.

Nordman, H., H. Keskinen, and M. Tuppurainen.  1985.
Formaldehyde asthma - rare or overlooked?  J. Allergy Clin.
Immunol. 75:91-99.
                               6-65

-------
                                                        EPA-420-R-93-005
                                                            April 1993

National Research Council  (NRC).  1981.  Formaldehyde and other
aldehydes.  Washington, DC:  National Research Council, Committee
on Aldehydes, Board on Toxicology and Environmental Health
Hazards.  National Academy Press.

Olsen, J.H., S.P. Jensen, M. Hink, K. Faurbo, N. Breum, and 0.
Jensen.  1984.  Occupational formaldehyde exposure and increased
nasal cancer risk in man.  Int. J. Cancer 34: 639-644.

OSHA.  1987.  Occupational exposure to formaldehyde.  Final Rule.
U.S. Department of Labor, Occupational Safety and Health
Administration.  Fed. Reg. 50: 46168-46312.

Partanen, T., T. Kauppinen, S. Hernberg, J. Nickels, R.
Luukkoner, T. Hakulinen, and M.A. Pukkala.  1990.  Formaldehyde
exposure and respiratory cancer among woodworkers - an update.
Scan. J. Work Environ. Health 16: 394-400.

Partanen, T., T. Kauppinen, M. Nurminen, J. Nickels, S. Hernberg,
T. Hakulinen, E. Pukkala, and E. Savonen.  1985.  Formaldehyde
exposure and respiratory and related cancers:  A case-referent
study among Finnish workers.  Scand. J. Work Environ. Health 11:
409-411.

Patterson, R., V. Pateras, L.C. Grammer, et al.  1986.  Human
antibodies against formaldehyde human serum albumin conjugates or
human serum albumin in individuals exposed to formaldehyde.  Int.
Archs. Allergy Appl.  Immun. 79:53-59.

Perry, R.H. and Chilton, C.H.  1973.  Chemical Engineer's
Handbook, Fifth Edition.  McGraw-Hill Book Company.

Ritchie, I.N. and R.G. Lehnen.  1987.  Formaldehyde-related
health complaints of residents living in mobile and conventional
homes.  Am. J. Public Health 77:  323-328.

Robins, J., N. Pambrun, C. Chute, and D. Blevins.  1988.
Estimating the effect of formaldehyde exposure on lung cancer and
non-malignant respiratory disease (NNRD) mortality using a new
method to control for the healthy worker effect.  In: Progress in
Occupational Methodology.  Hogstedt C and Reuterwall C.  (eds.).
Excerpta Medica, Amsterdam, pp. 75-78.

Roush, G., J. Walrath, L. Sayner, S. Kaplan, J.T. Flannery, and
A. Blair.  1987.  Nasopharyngeal cancer, sinonasal cancer, and
occupations related to formaldehyde: A case-control study.  JNCI
79:1221-1224.

Rusch, G.M., J.J. Clary, W.E. Rinehart, and H.F. Bolte.  1983.  A
26-week inhalation toxicity study with formaldehyde in the
monkey, rat, and hamster.  Toxicol.  Appl. Pharmacol.  68:  329-
343.

Saillenfait, A.M, P.  Bonnet, and J.  De Ceaurriz.  1989.  The
effects of maternally inhaled formaldehyde on embryonal and
foetal development in rats.  Fd. Chem. Toxic. 27:545-548.
                               6-66

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Sauder, L.R., M.D. Chatham, D.J. Green, and T.J. Kulle.  1986.
Acute pulmonary response to formaldehyde exposure in healthy
nonsmokers.   J. Occup. Med. 28:420-424.

Sauder, L.R., D.J. Green, N.D. Chatham, et al.  1987.  Acute
pulmonary response of asthmatics to 3.0 ppm formaldehyde.
Toxicol. Industr. Health 3:569-578.

Schmidt, U.  and E. Loeser.  1986.  Epoxidation of 1,3-butadiene
in liver and lung tissue of mouse, rat, monkey and man.  Adv.
Exp. Med. Biol. 197:951-957.

Sellakumar,  A.R., C.A. Snyder, J.J. Solomon, and R.E. Albert.
1985.  Carcinogenicity of formaldehyde and hydrogen chloride in
rats.  Toxicol. Appl.  Pharmacol. 81:  401-406.

Shikiya, B.C., C.S. Liu, M.I. Kahn, J. Juarros, and W.
Barcikowski.  1989.  In-vehicle air toxics characterization study
in the south coast air basin.  South Coast Air Quality Management
District, El Monte, CA.  May, 1989.

Shumilina, A.V.  1975.  Menstrual and child-bearing functions of
female workers occupationally exposed to the effects of
formaldehyde.  Gig. Truda. I. Professional Nye. Zabolevaniya
12:18-21.

Sigsby, J. E., S. Tejeda, W. Ray, J. M. Lang, and J.  W. Duncan.
1987.  Volatile organic compound emissions from 46 in-use
passenger cars.  Environ. Sci. Technol. 21:466-475.

Smith, L. R.  1981.  Characterization of Exhaust Emissions from
High Mileage Catalyst-Equipped Automobiles.  Ann Arbor, Michigan:
U.S. Environmental Protection Agency, Office of Mobile Sources.
Publication no. EPA-460/3-81-024.

Snyder, R.D. and B. Van Houten.  1986.  Genotoxicity of
formaldehyde and an evaluation of its effect on the DNA repair
process in human diploid fibroblasts.  Mutat. Res. 165:21-30.

Soffritti, M., C. Maltoni, F. Maffei, and R. Biagi.  1989.
Formaldehyde: An experimental multipotent carcinogen.  Toxicol.
Indust. Health 5:699-730.

Springer, K. J.  1977.  Investigation of Diesel-Powered Vehicle
Emissions VII.  Ann Arbor, Michigan: U.S. Environmental Agency,
Office of Mobile Sources.  Publication no. EPA-460/3-76-034.

Springer, K. J.  1979.  Characterization of Sulfates, Odor,
Smoke, POM and Particulates from Light and Heavy-Duty Engines --
Part IX.  Ann Arbor, Michigan: U.S. Environmental Protection
Agency, Office of Mobile Sources. Publication no. EPA-460/3-79-
007.

Starr T.B.  1990.  Quantitative cancer risk estimation for
formaldehyde.  Risk Anal. 10:  85-91.
                               6-67

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Stayner, L.,  A.B. Smith, G. Reeve, et al.  1985.  Proportionate
mortality study of workers exposed to formaldehyde in the garment
industry.  Am. J. Ind. Med. 7:229-240.

Stayner, L.  T., L. Elliot, L. Blade, R. Keenlyside, and W.
Halperin.  1988.  A retrospective cohort mortality study of
workers exposed to formaldehyde in the garment industry.  Am. J.
Ind. Med. 13:667-682.

Sterling, T.D. and J.J. Weinkam.  1989.  Reanalysis of lung
cancer mortality in a National Cancer Institute study on
mortality among industrial workers exposed to formaldehyde.
Additional discussion.  J. Occup. Med. 31:881-883.

Stern, F.B.,  I.I. Beaumont, W.E. Halperin, L.I. Nurthy, B.W.
Hills, and J.M. Fajen.  1987.  Mortality of chrome leather
tannery workers and chemical exposures and tanneries.  Scand. J.
Work Environ. Health 13:108-117.

Stump, F. D., S. Tejada, W. Ray, D. Dropkin, F. Black, R. Snow,
W. Crews, P.  Siudak, C. 0. Davis, L. Baker and N. Perry.  1989.
The influence of ambient temperature on tailpipe emissions from
1984 to 1987 model year light-duty gasoline vehicles.
Atmospheric Environment 23: 307-320.

Stump, F. D., S. Tejeda, W. Ray, D. Dropkin, F. Black, R. Snow,
W. Crews, P.  Siudak, C. 0. Davis and P. Carter.  1990.  The
influence of ambient temperature on tailpipe emissions from 1985-
1987 model year light-duty gasoline vehicles -- II.  Atmospheric
Environment 24A: 2105-2112.

Stump, F. D., K. T. Knapp, W. D. Ray, R. Snow and C. Burton.  The
composition of motor vehicle organic emissions under elevated
temperature summer driving conditions  (75 to 105F).
Unpublished.

Swenberg, J.A., C.S. Barrow, C.J. Boreikol, Hd'A Heck, R.J.
Levine, E.T.  Morgan, and T.B. Starr.  1983.  Non-linear
biological responses to formaldehyde and their implications for
carcinogenic risk assessment.  Carcinogenesis 4:  945-952.

Smylie, G.M., G.Z. Whitten, R.E. Morris, J.L. Fieber, and K.
O'Connor.  1990.  "Technical analysis of the ARE staff proposal
for hydrocarbon reactivity adjustment factors for clean fuel/low
emission vehicle regulations."  Systems Applications
International, San Rafael, California  (SYSAPP-90/092).

Takahashi, M., R. Hasegawa, F. Furukawa, K. Toyoda, H. Sato, and
Y. Hayashi.   1986.  Effects of ethanol, potassium metabisulfite,
formaldehyde, and hydrogen peroxide on gastric carcinogenesis in
rats after initiation with N-methyl-N-nitrosoguanidine.  Jpn. J.
Cancer Res.   (Gann) 77:118-124.

Til, H.P., P.A. Woutersen, V.J. Feron, V.H.M. Hollanders, and
H.E. Falke.   1989.  Two-year drinking-water study of formaldehyde
in rats.  Fd. Chem. Toxicol. 27:77-87.
                               6-68

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Tobe, M., T. Kaneko, Y. Uchida, et al.  1985.  Studies of the
inhalation toxicity of formaldehyde.  National Sanitary and
Medical Laboratory Service  (Japan).  pp 1294.

Tobe, M., K. Naito, and Y. Kurokawa.  1989.  Chronic toxicity
study on formaldehyde administered orally to rats.  Toxicology
56:79-86.

Uba, G., D. Pachorek, J. Bernstein, et al.  1989.  Prospective
study of respiratory effects of formaldehyde among healthy and
asthmatic medical students.  Am. Jour. Ind. Med. 15:91-101.

Ulsamer, A.G., K.C. Gupta, M.S. Cohen, and P.W. Preuss.  1982.
Formaldehyde in indoor air:  Toxicity and risk.  In:  Proceed.
45th(2) Ann. Meet. Air Pollut. Control Assoc. 16 p.

Ulsamer, A.G., J.R. Beall, H.K. Kang, et al. 1984.  Overview of
health effects of formaldehyde.  In:  Saxsena J.  (ed.), Hazard
Assessment of chemicals -- Current Developments.  NY:  Academic
Press, Inc. 3:337-400.

Universities Associated for Research and Education in Pathology,
Inc. 1988.  Epidemiology of chronic occupational exposure to
formaldehyde:  report of the ad hoc panel on health aspects of
formaldehyde.  Toxicol. Indus. Health 4:  77-90.

Ura. H., P. Nowak, S. Litwin, P. Watts, R.D. Bonfil, and A.J.P.
Klein-Szanto.  1989.  Effects of formaldehyde on normal
xenotransplanted human tracheobronchial epithelium.  Am. J.
Pathol. 134:99-106.

Urban, C. M.  1980a.  Regulated and Unregulated Exhaust Emissions
from Malfunctioning Three-Way Catalyst Gasoline Automobiles.  Ann
Arbor, Michigan: U.S. Environmental Protection Agency, Office of
Mobile Sources.  Publication no. EPA-460/3-80-004.

Urban, C. M.  1980b.  Regulated and Unregulated Exhaust Emissions
from a Malfunctioning Three-Way Catalyst Gasoline Automobile.
Ann Arbor, Michigan: U.S. Environmental Protection Agency, Office
of Mobile Sources.  Publication no. EPA-460/3-8-005.

Urban, C. M.  1980c.  Regulated and Unregulated Exhaust Emissions
from Malfunctioning Non-catalyst and Oxidation Catalyst Gasoline
Automobiles.  Ann Arbor, Michigan: U.S. Environmental Protection
Agency, Office of Mobile Sources.  Publication no. EPA-460/3-80-
003.

Urban, C. M.  1981.  Unregulated Exhaust Emissions from Non-
Catalyst Baseline Cars Under Malfunction Conditions.  Ann Arbor,
Michigan: U.S. Environmental Protection Agency, Office of Mobile
Sources.  Publication no. EPA-460/3-81-020.

Vaughan, T.L., C. Strader, S. Davis, and J.R. Daling.  1986a.
Formaldehyde and cancers of the pharynx, sinus, and nasal cavity:
I. Occupational exposures.  Int. J. Cancer.  38:  677-683.
                               6-69

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Vaughan, T.L., C. Strader, S. Davis, and J.R. Baling.  1986b.
Formaldehyde and cancers of the pharynx, sinus, and nasal cavity.
II. Residential exposures.  Int. J. Cancer.  38:  685-688.

Warner-Selph, M. A., and L. R. Smith.  1991.  Assessment of
Unregulated Emissions from Gasoline Oxygenated Blends.  Ann
Arbor, Michigan: U.S. Environmental Protection Agency, Office of
Mobile Sources.  Publication no. EPA-460/3-91-002.

Warner-Selph, M. A., and Joseph DeVita.  1989.  Measurements of
toxic exhaust emissions from gasoline-powered light duty
vehicles.  SAE Paper 892075.

Wilmer, J.W., P.A. Woutersen, L.M. Appelman, W.R. Leeman, and
V.J. Feron.  1989.  Subchronic  (13-week) inhalation toxicity
study of formaldehyde in male rats: 8-hour intermittent versus 8-
hour continuous exposure.  Toxicol. Lett. 47:287-293.

Wilmer, J.W., P.A. Woutersen, L.M. Appelman, W.R. Leeman, and
V.J. Feron.  1987.  Subacute  (4-week) inhalation  toxicity study
of formaldehyde in male rats: 8-hour intermittent versus 8-hour
continuous exposure.  J. Appl. Toxicol. 7:15-16.

Wilson, A. et al.  1991.  Air toxics microenvironment exposure
and monitoring study.  South Coast Air Quality Management
District, El Monte, CA.

Witek, T.J., E.N. Schachter, T. Tosun, et al.  1987.  An
evaluation of respiratory effects  following exposure to 2.0 ppm
formaldehyde in asthmatics:  Lung  function, symptoms, and airway
reactivity.  Arch. Environ. Health 42:230-237.

Woutersen, P.A., A. van Garderen-Hoetmer, J.P. Bruijntjes, A.
Zwart, and V.J. Feron.  1989.  Nasal tumors in rats after severe
injury to the nasal mucosa and prolonged exposure to 10 ppm
formaldehyde.  J. Appl. Toxicol. 9:39-46.

Yager, J.W., K.L. Cohn, R.C. Spear, J.N. Fisher,  and L. Morse.
1986.  Sister-chromatid exchanges  in lymphocytes  of anatomy
students exposed to formaldehyde-embalming solution.  Mutat. Res.
17:135-139.

Yetter, R.A., H. Rabitz, F.L. Dreyer, R.G. Maki,  and R.B. Klemm.
1989.  Evaluation of the rate constant for the reaction OH +
H2CO:   Application of modeling and sensitivity analysis
techniques for determination of the product branching ratio.  J.
Chem. Phys., 91:4088-4097.

Zwart, A., P.A. Woutersen, L.M. Appelman, J.W. Wilmer, and B.J.
Spit.  1988.  Cytotoxic and adaptive effects in rats nasal
epithelium after 3-day and 13-week exposure to low concentrations
of formaldehyde vapour.  Toxicology 51:87-99.
                               6-70

-------
                                                        EPA-420-R-93-005
                                                            April 1993
7.0  1,3-BUTADIENE

7.1 Chemical and Physical Properties  (EPA, 1989;  1992)

     1,3-Butadiene is a colorless, flammable gas  at room
temperature with a pungent, aromatic odor, and a  chemical  formula
C4H6  (CH2:CHCH:CH2) .   Table  7-1  summarizes 1, 3-butadiene ' s
chemical and physical properties.  1,3-Butadiene  is insoluble  in
water,  slightly soluble in methanol and  ethanol,  and  soluble in
organic solvents such as benzene and ether.  1,3-Butadiene is
also structurally related to known carcinogens.

     Because of its reactivity, 1,3-butadiene is  estimated to
have a short atmospheric lifetime.  The  actual lifetime depends
upon the conditions at the time of release.  The  primary removal
mechanisms are through chemical reactions  with hydroxyl radicals
and ozone.  Therefore, factors influencing 1,3-butadiene's
atmospheric lifetime, such as the time of  day, sunlight
intensity, temperature, etc., also include those  affecting the
availability of hydroxyl radicals and ozone.
Table 7-1.  Chemical and Physical Properties of 1,3-Butadiene.
Properties
Molecular weight
Melting point
Boiling point
Density at 20C (68F)
Vapor pressure at 20C
Flash point
Solubility in water at 20C
Conversions at 25C (77F)
Values
54.09 g/mole
-108.91C (-164.04F)
-4.41C (24.06F)
0.6211 g/cm3
1.2 atm.
-105. 0C (-157. 0F)
0.735 g/L
1 ppm (by volume) = 2.21 mg/m3
1 mg/m3 = 0.45 ppm (by volume)
7.2 Formation and Control Technology

     1,3-Butadiene is formed in vehicle exhaust by the  incomplete
combustion of the fuel and is assumed not to be present  in
vehicle evaporative and refueling emissions.  As a rule,  refiners
try to minimize the level of 1,3-butadiene in gasoline  and diesel
fuel because it tends to readily form a varnish which can be
harmful to engines (EPA, 1989).  Therefore, the majority of
gasoline and diesel fuel should have no significant  1,3-butadiene
content.
                               7-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     1,3-Butadiene emissions appear to increase roughly in
proportion to hydrocarbon emissions.  Since hydrocarbons are
decreased by the use of a catalyst on a motor vehicle, 1,3-
butadiene emissions are expected to decrease proportionally.


7.3  Emissions

7.3.1  Emission Fractions Used in the MOBTOX Emissions Model

     Actual vehicle emissions were used to develop 1,3-butadiene
emission fractions.  Because 1,3-butadiene decays rapidly in the
Tedlar bags used to collect emissions samples, exhaust speciation
analyses often underestimate 1,3-butadiene emissions.  This is
especially true of older studies.  Thus, although 1,3-butadiene
emissions data from many studies were available (Appendix B2), it
was decided to use data from one study with a very large number
of vehicles,  recently conducted by CARB (1991), in which
deterioration of 1,3-butadiene was strictly controlled,  and
emission fractions were adjusted to account for time lag between
sample collection and sample analysis.  This study also tested
more typical in-use vehicles, rather than low mileage vehicles as
in other studies.  1,3-Butadiene emission fractions for different
programs included in modeling components are included in Appendix
B6.  For vehicles with three-way catalysts, 1,3-butadiene
emission fractions from the Auto/Oil study were somewhat lower
than emission fractions from the CARB data (overall, about 15%).
This may indicate that later model vehicles,  with more efficient
catalysts,  have lower 1,3-butadiene fractions.  Thus, with fleet
turnover, 1,3-butadiene fractions in motor vehicle emissions may
drop.  Also,  it should be noted that most 1,3-butadiene emissions
occur during cold starts, and use of heated catalysts in future
years will reduce these cold start emissions  (see Ford Motor
Company comments in Appendix I).

     CARB measured 1,3-butadiene mass emissions for 55
LDGVs/LDGTs with three-way catalysts or three-way plus oxidation
catalysts,  7 LDGVs/LDGTs with oxidation catalysts, 16 LDGVs/LDGTs
with no catalysts, 2 LDDVs, and 1 HDDV.  CARB then calculated THC
fractions and converted these fractions to TOG fractions using
conversion factors.  CARB's THC fraction was converted to a TOG
fraction using the conversion factors in Table 3-7, rather than
the CARB conversion factors.  The resultant TOG fractions for
vehicles running on baseline fuel are listed in Appendix B5.
CARB calculated an average emission fraction for three-way
catalyst and three-way plus oxidation catalyst LDGV/LDGT
combined.  Since only 7 of the 55 vehicles from the combined
category had three-way plus oxidation catalysts,  the average
emission fraction was applied to the three-way catalyst category.
For LDGVs/LDGTs with three-way plus oxidation catalysts, it was
assumed that TOG fractions for this category would be the same as
for oxidation catalysts.  CARB also did not measure 1,3-butadiene
emissions for HDGVs.  It was assumed the fraction for HDGVs with
                               7-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993
three-way catalysts was the same as for LDGVs with three-way
catalysts, and that the fraction for HDGVs with no catalysts was
the same as for LDGVs with no catalysts.

     No 1,3-butadiene data were available for oxygenated fuels
from the CARB study.  To calculate TOG fractions for vehicles
running on MTBE blends and 10% ethanol, adjustment factors were
applied to the baseline emission fractions for each vehicle
class/catalyst combination.  To calculate an appropriate
adjustment factor, percent of 1,3-butadiene in exhaust was
compared for baseline and oxygenated blends (Appendix B4).  This
comparison was performed for 15% MTBE and 10% ethanol.  The
average percent change (expressed as a fraction) was added to 1,
representing baseline emissions with gasoline, and the baseline
1,3-butadiene fractions then multiplied by the resultant factor.
The 15% MTBE and 10% ethanol adjustment factors for LDGVs/LDGTs
with various catalyst technologies are summarized in Table 7-2.
The 15% MTBE numbers were estimated using data from Auto/Oil
(1990)  and DeJovine et al.  (1991) for LDGVs/LDGTs with three-way
catalysts, Auto/Oil (1991)  for LDGVs/LDGTs with three-way plus
oxidation and oxidation catalysts, and Warner-Selph and Smith
(1991)  for LDGVs/LDGTs with no catalysts.  The 10% ethanol
numbers were estimated using data from Auto/Oil (1990) and
Warner-Selph and Smith (1991) for LDGVs/LDGTs with three-way
catalysts, and Warner-Selph and Smith  (1991) for LDGVs and LDGTs
with oxidation catalysts or no catalysts.  Due to a lack of data,
the adjustment factor for LDGVs/LDGTs with three-way plus
oxidation catalysts was assumed to be equal to the one for
LDGVs/LDGTs with oxidation catalysts.
Table 7-2.  15% MTBE and 10% Ethanol Emission Fraction Adjustment
                    Factors  for 1,3-Butadiene.
Vehicle
Class
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
Catalyst
Technology
3 -way
3 -way + ox
oxidation
non-cat
15% MTBE
Adjustment
Factor
0.9798
0.9873
1.1790
1.2382
10%
Ethanol
Adjustment
Factor
0.8812
0.9375
0.9375
1.1233
     Since the average percent change was calculated for 15% MTBE
(2.7% weight percent oxygen), and 11.0% MTBE  (2.0% oxygen) was
assumed for reformulated fuel and California standards
components, average percent changes in the 1,3-butadiene TOG
fraction from 0 to 15% MTBE  were multiplied by 2.0/2.7, the
                               7-3

-------
                                                        EPA-420-R-93-005
                                                            April 1993
ratio of oxygen contents by weight for reformulated gasoline and
15% MTBE.  For HDGVs with three-way catalysts and with no
catalysts, the same 15% MTBE and 10% ethanol adjustment factors
were assumed as for LDGVs/LDGTs with the same catalyst
technologies.

7.3.2  Emission Factors for Baseline and Control Scenarios

     The fleet average 1,3-butadiene emission factors as
determined by the MOBTOX emissions model are presented in Table
7-3.  When comparing the base control scenarios relative to 1990,
the emission factor is reduced by 40% in 1995, by 54% in 2000,
and by 57% in 2010.  The expansion of reformulated fuel use in
1995 provides no net reduction in the emission factor, relative
to 1990.  In 2000 and in 2010, the expansion of reformulated fuel
usage and the California standards reduces the emission factor by
another 2 to 3% over the base control from the respective year.

7.3.3  Nationwide Motor Vehicle 1,3-Butadiene Emissions

     The nationwide 1,3-butadiene metric tons are presented in
Table 7-4.  Total metric tons are determined by multiplying the
emission factor (g/mile)  by the VMT determined for the particular
year.  The VMT, in billion miles, was determined to be 1793.07
for 1990, 2029.74 for 1995,  2269.25 for 2000, and 2771.30 for
2010.  When comparing the base control scenarios relative to
1990, the metric tons are reduced by 32% in 1995 and by 42% in
2000.  Even though the emission factor continues to decrease from
2000 to 2010,  this is more than offset by the large increase in
VMT.  As a result, metric tons in 2010 actually increase relative
to 2000.

7.3.4 Other Sources of 1,3-Butadiene

     Mobile sources account for approximately 94% of the total
1,3-butadiene emissions (EPA, 1989).   The remaining 1,3-
butadiene emissions (6%)  come from stationary sources related to
industries producing 1,3-butadiene and those industries that use
1,3-butadiene to produce other compounds.

     Of the 6% attributable to stationary 1,3-butadiene sources,
73.8% is produced by the styrene-butadiene copolymer  (rubber and
latex)  industry.  Another 25.8% of the 1,3-butadiene emissions
are produced by the industries manufacturing polybutadiene,
neoprene rubber, acrylonitrile-butadiene-styrene resin, nitrile
rubber, and adiponitrile,  the raw material for nylon 6,6
production.  There are also miscellaneous producers/users that
account for only a small percentage of the total stationary
source emissions.   The final 0.4% is produced by the
manufacturing of 1,3-butadiene itself.  Of the 11 producers of
1,3-butadiene, 9 are located in Texas, and 2 in Louisiana  (EPA,
1989).
                               7-4

-------
                                                            EPA-420-R-93-005
                                                                 April 1993
Table  7-3.   Annual  Emission Factor Projections for  1,3-Butadiene,
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0156
0.0094
0.0093
0.0071
0.0069
0.0069
0.0067
0.0064
0.0062
Percent
Reduction
from 1990
-
40
40
54
56
56
57
59
60
                                  7-5

-------
                                                             EPA-420-R-93-005
                                                                 April 1993
Table  7-4.   Nationwide Metric  Tons Projection for  1,3-Butadiene,
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0156
0.0094
0.0093
0.0071
0.0069
0.0069
0.0067
0.0064
0.0062
Metric
Tons
27, 972
19,080
18, 877
16,112
15,658
15,658
18,568
17,736
17,182
                                  7-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Approximately 59% of the mobile source 1,3-butadiene
emissions (56% of total 1,3-butadiene emissions)  can be
attributed to onroad motor vehicles, with the remainder
attributed to nonroad mobile sources.  This figure is based on
the average of an EPA estimate and a California Air Resources
Board estimate (GARB, 1991).

     Analysis of EPA data indicated that about 46% of mobile
source 1,3-butadiene emissions (43% of total 1,3-butadiene
emissions)  can be attributed to onroad motor vehicles, with the
remainder attributable to nonroad mobile sources.  This figure is
based on a number of crude estimates and assumptions.  First, it
was estimated that about 70% of mobile source VOCs are
attributable to onroad vehicles (Section 5.3.4).   This VOC split
was adjusted by onroad and nonroad 1,3-butadiene fractions to
come up with the estimate of 46% of mobile source 1,3-butadiene
from on-road motor vehicles.   For onroad vehicles, 1,3-butadiene
was estimated to be 0.61% of exhaust.  This is a 1990 fleet
average toxic fraction, with fractions in Appendix B2, weighted
using 1990 VMT fractions.  For nonroad vehicles,  1,3-butadiene
was estimated to be 1.3% of exhaust, based on the NEVES report
(EPA, 1991) .

     The CARB study cited above indicated that, in California, of
about 96% of 1,3-butadiene estimated to come from mobile sources,
71% could be attributed to onroad motor vehicles, and the
remainder to other mobile sources.  Thus, by averaging the EPA
estimate and the CARB estimate, an estimated contribution of 59%
of mobile source 1,3-butadiene emissions from onroad motor
vehicles was derived.

7.4 Atmospheric Reactivity and Residence Times

7.4.1 Gas Phase Chemistry of 1,3-Butadiene

     The processes involved in transformation and residence times
were previously discussed in Section 5.4.  For a more detailed
explanation of the various parameters involved in these processes
please refer to Section 5.4.   The information that follows on
transformation and residence times has been mainly excerpted from
a report produced by Systems Applications International for the
EPA  (Ligocki et al.,  1991).

     The structure of 1,3-butadiene  (C4H6) is a straight-chain
molecule with two conjugated double bonds (H2C=CH-CH=CH2) .
Species
containing double bonds are referred to as "alkenes" or
"olefins".   These double bonds represent extremely active sites
for atmospheric oxidation.   In contrast to the slow rate of
reaction of benzene in the atmosphere, 1,3-butadiene reacts quite
rapidly with the hydroxyl radical, ozone and nitrate radical.
Furthermore,  the oxidation of 1,3-butadiene produces two species
which are themselves toxic,  formaldehyde and acrolein (C3H40).


                               7-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Acrolein and
its oxidation products are powerful lacrimators  (compounds which
cause eye irritation).  Concern for the high atmospheric
reactivity of compounds such as 1,3-butadiene and their
undesirable products  such as acrolein lead to early regulations
limiting the olefin content of gasoline  (e.g., Rule 66 by the Los
Angeles Air Pollution Control District in 1966) .

7.4.1.1 Gas Phase Reactions

     There are three  chemical reactions of 1,3-butadiene which
are important in the  ambient atmosphere: reaction with OH,
reaction with 03,  and reaction with N03.  All of  these reactions
are rapid, indicating that 1,3-butadiene will be transformed
rapidly in the atmosphere.  The reaction of 1,3-butadiene with
the oxygen radical and with N02 do  occur in the atmosphere,  but
because their concentrations are much lower than OH and 03,  these
reactions are not important in the ambient atmosphere.

7.4.1.2 Reaction Products

     The atmospheric  oxidation of 1,3-butadiene by OH proceeds
primarily by addition across the double bonds.  Olefins generally
react to form two aldehyde products, one from each side of the
original double bond.  Therefore, the major products from
1,3-butadiene are formaldehyde  (from the terminal carbon) and
acrolein  (from the internal carbon).  These products would be
expected, at least to some extent,  from any of the atmospheric
reactions of this diolefin, although the mechanism and products
from the 03  reaction are  not completely understood.   An exception
is the reaction of 1,3-butadiene with N03.   This  reaction
apparently proceeds primarily by addition, producing
approximately 60% nitrates, with the yield of formaldehyde and
acrolein only 12% (Barnes et al., 1990).

     Formaldehyde has many sources in the atmosphere.  The
production of formaldehyde from the oxidation of 1,3-butadiene
would not be expected to be a significant portion of the total
secondary formaldehyde production.   However, 1,3-butadiene can be
considered to be the  major precursor species for atmospheric
acrolein production.

7.4.2 Aqueous Phase Chemistry of 1,3-Butadiene

     The aqueous solubility of 1,3-butadiene is very low, an
order of magnitude smaller than that of benzene.  Incorporation
of 1,3-butadiene into clouds and rain will not be an important
process due to the low solubility of 1,3-butadiene despite the
relatively rapid reaction with OH radical.

7.4.3 1,3-Butadiene Residence Times

     Residence times  for 1,3-butadiene were calculated by


                               7-8

-------
                                                        EPA-420-R-93-005
                                                            April 1993
considering only gas-phase chemical reactions with OH, N03,  and
03.   Due to the low solubility of 1,3-butadiene,  wet deposition
and in-cloud chemical destruction are not important processes for
1,3-butadiene, and were not considered in the calculations.  The
importance of dry deposition for 1,3-butadiene depends upon the
value assumed for the reactivity parameter.  For these
calculations, 1,3-butadiene was assumed to be unreactive on
surfaces; if this assumption is valid then its deposition
velocity will be negligibly small.

     The results of the residence time calculations for
1,3-butadiene are presented in Table 7-5.  During the daytime,
the residence time of 1,3-butadiene is determined primarily by
its reaction with OH radical, with a small contribution from the
reaction with 03.   The  residence time  of  1, 3-butadiene under
summer, daytime, clear-sky conditions is one hour or less for all
four cities.  The residence time of 1,3-butadiene has previously
been estimated at 4 hours under clean, background atmospheric
conditions  (Cupitt, 1987).   The shorter residence times estimated
here are a result of the higher oxidant concentrations predicted
for urban areas.

     At night, the rapid reaction of 1,3-butadiene with N03,  and
to a lesser extent 03,  leads  to residence times  of  0.5-6  h under
clear-sky conditions.  In fact, for Los Angeles, the summer,
clear-sky residence time for 1,3-butadiene is estimated to be
shorter at night than it is during the daytime.   These relatively
short residence times for 1,3-butadiene even at night indicate
that there is very little possibility of carryover of
1,3-butadiene concentrations from one day to the next during the
summertime under clear-sky conditions.  Under cloudy-sky
conditions, summer nighttime residence times were estimated to be
6-15 h.  These are long enough to allow for the possibility of
day-to-day carryover.

     Daytime residence times for different cities within a given
season varied by factors of 2-3, whereas nighttime residence
times varied by much larger factors.  As with benzene, the
difference
between summer and winter conditions was large at all sites, with
winter residence times 10-30 times greater than summer residence
times.

     Under wintertime conditions, the residence time of
1,3-butadiene was estimated to be in the range of 12-2000 h.
Although daytime residence times during the winter are still
relatively short, residence times at night can be very long
because of the extremely low N03 concentrations.   Behavior of
1,3-butadiene is therefore very different in the winter than it
is during the summer.  Significant day-to-day carryover of
1,3-butadiene concentrations is possible in the winter,
particularly under cloudy-sky conditions.
                               7-9

-------
                                                          EPA-420-R-93-005
                                                              April 1993
     The major  uncertainties in the  residence time calculations
for 1,3-butadiene are most likely the  nighttime N03
concentrations  and
the N03 reaction  rate.   Concentrations of N03 are highly
variable,
                                7-10

-------



TABLE 7-5. Atmospheric residence time calculation
hours unless otherwise noted.
Los Angeles St. Louis

Clear sky - day
Clear sky - night
Clear sky - avg
Cloudy - day
Cloudy - night
Cloudy - avg
July
0.8
0.4
0.6
1.7
6
2
Jan July Jan
5 0.5 7
16 6 200
82 17
10 1.2 16
90 15 400
20 2 40
EPA-420-R-93-005
April 1993
for 1,3 -butadiene. All times are in
Atlanta New York
July Jan July Jan
0.6 7 1.0 14
0.6 5 1.1 1600
0.6 6 1.0 40
1.2 16 2 30
7 90 11 2000
1.8 30 3 80
Monthly
Climatological
Average
0.8
11
30
1.0
12
1.7
50
                                         7-11

-------
                                                        EPA-420-R-93-005
                                                            April 1993
often peaking shortly after sunset and then decreasing rapidly by
midnight (Platt et al.,  1980).  In addition, the N03 reaction
rate is only known to within a factor of two.  Therefore,
although the daytime residence times are accurate to about a
factor of two, nighttime residence times are certain only to
within an order of magnitude.

7.4.4 Limited Urban Airshed Modeling Results for 1,3-Butadiene

     The Urban Airshed Model  (UAM) has been previously discussed
in Section 5.4.  Please refer to this section for details about
the model,  its inputs,  and modifications.  Much of the
information below has been excerpted from reports conducted for
EPA by Systems Applications International (SAI) (Ligocki et al.,
1991, 1992) .

     1, 3-Butadiene was treated explicitly in the UAM-Tox.  Mobile
and stationary emissions of 1,3-butadiene were tagged separately
and carried through the simulation as distinct species.  The gas
phase reactions discussed previously were added to the chemistry
subroutines.   Because the focus of the study was on destruction
of the toxic species rather than on the subsequent chemistry of
their reaction products, no products were included in the UAM
modifications for 1,3-butadiene.

St. Louis Simulation

     In the St. Louis simulation, the high reactivity of
1,3-butadiene is demonstrated by the large deviation of the
reactive and inert 1,3-butadiene curves in Figure D-3 located in
Appendix D.   By mid-afternoon, the concentration predicted in the
absence of chemistry would be 0.4 ppb, versus 0.05 ppb for the
simulation which included chemistry.  Thus,  the afternoon
concentration of 1,3-butadiene was reduced by 90 percent due to
atmospheric reactions.   The two curves approach each other again
after sunset, when the 1,3-butadiene concentration reached its
highest value of the simulation.   These results suggest that
human exposure to 1,3-butadiene during the summertime will be
limited to areas near sources.  According to Table 7-5,
1,3-butadiene is destroyed relatively rapidly even at night, so
little or no carryover of 1,3-butadiene concentration to the
following day would be expected.   The concentration of
1,3-butadiene would be expected to be greater in the wintertime
due to the less active photochemistry.

     The comparison of simulated concentrations with ambient
measured concentrations showed good agreement for 1,3-butadiene.

Baltimore-Washington and Houston Area Simulations

     Simulations for the summer Baltimore-Washington area episode
(Ligocki et al.,  1992)  resulted in little change in ambient
concentrations of 1,3-butadiene with the use of federal


                               7-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993
reformulated gasoline.  Use of California reformulated gasoline
also had little impact on ambient concentrations of 1,3-
butadiene.   Maximum daily average 1,3-butadiene concentration for
the 1988 base scenario was 0.95 ppb.  Motor vehicle-related 1,3-
butadiene emissions accounted for about 23% of total 1,3-
butadiene emissions, based on the 1995 no motor vehicle scenario.
This motor vehicle emission estimate is lower than the 56%
estimate obtained in Section 7.3.4 for motor vehicles.  The
Ligocki et al.  (1992)  study suggests that a reason for this
discrepancy could be the inclusion of more types of area source
toxic emissions in the UAM-Tox inventory than had been considered
previously.  Also, the nonroad contribution to mobile source 1,3-
butadiene could be underestimated.  Summer Baltimore-Washington
area simulations were in very good agreement with UATMP data
throughout the domain.

     In the winter 1988 base scenario, the maximum daily average
1,3-butadiene concentration was 2.57 ppb, about 3 times higher
than in summer.  A major reason for this is slower reactive decay
of 1,3-butadiene in winter.  Motor vehicle-related 1,3-butadiene
accounted for 29% of total 1,3-butadiene emissions.  Reformulated
gasoline use had very little effect on winter 1,3-butadiene
ambient concentrations.

     For the summer 1987 base scenario in Houston,  the maximum
daily average 1,3-butadiene concentration was 33.2 ppb.  This
high level was due to the single largest point source of 1,3-
butadiene emissions in the United States.  However, the model may
have significantly overestimated the magnitude of this
concentration.   Throughout the rest of the modeling domain,
concentrations were around 2 ppb.  Motor vehicle-related 1,3-
butadiene accounted for 16% of total 1,3-butadiene emissions,
based on the 1995 no motor vehicle scenario.  Motor vehicle-
related 1,3-butadiene contributed less to overall 1,3-butadiene
in Houston than in the Baltimore-Washington area, due to the
large impact of point sources in Houston.  Simulations for the
summer Houston episode predicted little effect on maximum daily
average concentration of 1,3-butadiene with reformulated
gasoline.


7.5  Exposure Estimation

7.5.1  Annual Average Exposure Using HAPEM-MS

     The data presented in Table 7-6 represent the results
determined by the HAPEM-MS modeling that was described previously
in Section 4.1.1.  These numbers have been adjusted to represent
the increase in VMT expected in future years.

     The HAPEM-MS exposure estimates in Table 7-6 represent the
50th percentiles of the population distributions of exposure,
i.e.,  half the population will be above and half below these


                               7-13

-------
                                                             EPA-420-R-93-005
                                                                  April 1993
values.   High end exposures  can also be  estimated  by using  the
95th
                                 7-14

-------
                                                            EPA-420-R-93-005
                                                                April 1993
Table 7-6.
Annual  Average HAPEM-MS Exposure  Projections  for
1,3-Butadiene.
Year -Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Exposure
(uq/m3)
Urban
0.48
0.31
0.31
0.26
0.25
0.25
0.28
0.27
0.26
Rural
0.26
0.17
0.17
0.14
0.13
0.13
0.15
0.14
0.14
Nationwide
0.42
0.28
0.27
0.23
0.22
0.22
0.25
0.24
0.23
                                 7-15

-------
                                                        EPA-420-R-93-005
                                                            April 1993
percentile of the distributions.  According to the HAPEM-MS
sample output for benzene, the 95th percentile is 1.8 times
higher than the 50th percentile for urban areas, and 1.2 times
high for rural areas.   Applying these factors to the exposure
estimates in Table 7-6, the 95th percentiles for urban areas
range from 0.45 ug/m3  for the  2000 expanded reformulated gasoline
and California standards use scenarios to 0.86 ug/m3 for the 1990
base control scenario.  The 95th percentiles for rural areas
range from 0.16 to 0.31 ug/m3,  respectively.

7.5.2  Comparison of HAPEM-MS Exposures to Ambient Monitoring
Data

     As stated in section 4.1.2, four national air monitoring
programs/databases contain data on air toxics and the data for
1,3-butadiene is found in only three.  The Aerometric Information
Retrieval System (AIRS),  the Urban Air Toxic Monitoring Program
(UATMP),  and the National Ambient Volatile Organic Compounds Data
Base (NAVOC) all have data for 1,3-butadiene.  The urban exposure
data for 1,3-butadiene from the three databases are found in
Appendix C and summarized in Table 7-7.

     In the 1988 Aerometric Information Retrieval System  (AIRS),
18  measurements of 1,3-butadiene were taken at 3 sites.  These
sites were in the cities listed below.

          Louisville,  KY                Houston, TX
          Burlington,  VT

The highest average was 2.45 ug/m3 (1.11  ppb)  at an suburban
residential site in Houston, Texas.  Six samples were collected
at this site.  Houston, Texas does possess areas with high point
source concentrations and, coupled with the fact that the
location of the monitor is difficult to ascertain in relation to
the point sources,  the decision was made to exclude the 6 samples
from Houston from the final average ppb for the entire program.
The second highest average was 1.04 ug/m3 (0.47  ppb)  at a urban
commercial site in Burlington, Vermont.  Six samples were
collected at this site.  The lowest average was 0.97 ug/m3 (0.44
ppb)  at a urban industrial site in Louisville, Kentucky.  Six
samples were also collected at this site.  The overall average of
the averages for each site was 1.48 ug/m3 (0.67  ppb).   The
removal of the 6 Houston samples changes the ambient mean level
from 1.48 ug/m3  (0.67  ppb)  to  1.01 ug/m3  (0.46 ppb).

     In the 1990 Aerometric Information Retrieval System  (AIRS),
101  measurements of 1,3-butadiene were taken at 7 sites.  These
sites were in the cities listed below.

          Detroit,  MI                   Houston, TX
          Arlington County, VA          Henrico County, VA
          Hampton,  VA                   Hopewell, VA
          Roanoke,  VA


                               7-16

-------
                                                         EPA-420-R-93-005
                                                             April 1993
Table 7-7.  Air Monitoring Results  for 1,3-Butadiene.
Program
AIRS
UATMP
NAVOC
Years
1988
1990
1991
1989
1990
1987
Ambient Data3
ug/m3
1.01
0.47
0.22
0.46
0.31
0.75
Estimated
Motor Vehicle
Contribution1"
ug/m3
0.56
0.26
0.12
0.26
0.17
0.42
aCaution should be taken in comparing these numbers.  The  methods
of averaging the data are not  consistent  between air monitoring
databases.  The sampling methodology is also  inconsistent.

bThe  ambient data are adjusted to represent the  motor vehicle
contribution to the ambient concentration,  which for 1,3-
butadiene is estimated  to be 56%, based on emissions inventory
apportionment.
                               7-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993
The highest average was 1.58 ug/m3 (0.72 ppb)  at an suburban
residential site in Houston, Texas.  Four samples were collected
at this site.  Due to the reasons cited above,  the  four  samples
from Houston, Texas were excluded from the final average ppb  for
the entire program.  The second highest average was 0.73 ug/m3
(0.33 ppb) at an urban commercial site in Detroit,  Michigan.
Nineteen samples were collected at this site.   The  lowest  average
was 0.29 ug/m3 (0.13  ppb)  at a suburban residential site in
Hopewell,  Virginia.  Sixteen samples were also  collected at this
site.  The overall average of the averages for  each site (minus
Houston, Texas) was 0.47 ug/m3 (0.21  ppb).

     In the 1991 Aerometric Information Retrieval System (AIRS),
117  measurements of 1,3-butadiene were taken at 6  sites.  These
sites were in the cities listed below.
          Detroit, MI
          Henrico County, VA
          Hopewell, VA
Arlington County, VA
Hampton, VA
Roanoke, VA
The highest average was 0.27 ug/m3 (0.12 ppb)  at suburban
residential sites in Henrico County and Roanoke, Virginia.
Twenty-one and fourteen samples were collected, respectively,  at
each site.  The lowest average was 0.13 ug/m3 (0.06 ppb) at a
suburban residential site in Hopewell, Virginia.   Sixteen  samples
were collected at this site.  The overall average  of  the averages
for each site was 0.22 ug/m3 (0.10 ppb) .

     In the 1989 Urban Air Toxics Monitoring Program  (UATMP),  160
measurements of 1,3-butadiene were taken at 14  sites.   These
sites were in the cities listed below.
          Baton Rouge, LA
          Camden, NJ
          Fort Lauderdale, FL
          Miami, FL
          St. Louis, MO
          Washington, D.C.
Chicago, IL
Dallas, TX
Houston, TX
Pensacola,  FL
New Sauget, IL
Wichita, KS
The highest average was 1.33 ug/m3 (0.60 ppb)  at a suburban
residential site in Houston, Texas.  Thirty-four  samples were
collected at this site.  The lowest average was 0.11 ug/m3 (0.05
ppb) at a suburban industrial site in Pensacola,  Florida.  Only
seven samples were collected at this site.  The overall average
of the averages for each site was 0.46 ug/m3 (0.21 ppb).

     In the 1990 Urban Air Toxics Monitoring Program  (UATMP),  349
measurements of 1,3-butadiene were taken at 12 sites.  1,3-
Butadiene was identified in 106 of the samples.   These sites were
in the cities listed below.
                               7-18

-------
                                                        EPA-420-R-93-005
                                                            April 1993
          Baton Rouge, LA
          Camden, NJ
          Orlando, FL
          Port Neches, TX
          Toledo, OH
          Wichita, KS
Chicago, IL
Houston, TX
Pensacola,  FL
Sauget, IL
Washington, B.C.
The highest average was 24.51 ug/m3 (11.09 ppb)  at a suburban
residential site in Port Neches, Texas.  Twenty-eight samples
were collected at this site.  Port Neches, Texas does possess
areas with high point source concentrations and, coupled with the
fact that the location of the monitor is difficult to ascertain
in relation to the point sources, the decision was made to
exclude the 28 samples from Port Neches from the final average
ppb for the entire program.  The second highest average was  1.04
ug/m3  (0.47  ppb)  at  a suburban residential site in Houston,
Texas.  Twenty-eight samples were collected at this site. The
lowest average was 2.73 ug/m3  (0.06 ppb)  that was measured at
five different sites.  The overall average of the averages for
each site was 2.25 ug/m3  (1.02 ppb).   The removal of the 28 Port
Neches samples changes the ambient mean level from 2.25 ug/m3
(1.02 ppb) to 0.31 ug/m3  (0.14 ppb).

     In the 1987 National Ambient Volatile Organic Compound
Database  (NAVOC), 9 measurements of 1,3-butadiene were taken at 6
sites.  These sites were in the cities listed below.
          Bakersfield, CA
          Fremont,  CA
          San Jose, CA
Concord, CA
Richmond, CA
Stockton, CA
The highest average was 1.33 ug/m3 (0.60 ppb)  at an urban site in
Fremont, California.  Two samples were used for the average at
this site.  The lowest average was 0.55 ug/m3  (0.25 ppb)  also at
an urban site in San Jose, California.  Two samples were also
used for the average for this site.  The overall average of the
averages for each site was 0.75 ug/m3 (0.34 ppb) .

     HAPEM-MS assumes that the dispersion and atmospheric
chemistry of 1,3-butadiene is similar to CO.  This assumption
would appear not to be valid for a reactive compound like 1,3-
butadiene, which is transformed in the atmosphere.  To test the
reasonableness of the HAPEM-MS modeling results, the HAPEM-MS
results for 1990 are compared to ambient monitoring results for
recent years.  Before comparing the HAPEM-MS results to the
ambient data, the ambient monitoring data must be adjusted to
represent the amount that is attributed to mobile sources.  The
data derived from emission inventories estimate that 56% of the
ambient 1,3-butadiene can be apportioned to motor vehicles.  The
numbers in the second column of Table 7.7 are 56% of the ambient
levels and thus represent estimated motor vehicle levels.

     The motor vehicle apportionment of the ambient monitoring
                               7-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993
data, presented in Table 7-7, ranges from 0.12 to 0.56 ug/m3.
When the adjustment factor of 0.622 that was determined in
Section 5.5.2 is applied, this range becomes 0.08 to 0.35 ug/m3.
The HAPEM-MS 1990 base control level of 0.48 ug/m3 lies above
this range.  Since the unit risk estimate for 1,3-butadiene is an
upper bound estimate, the upper end of the ambient range is used
to calculate cancer incidences.  The HAPEM-MS 1990 base control
level must be multiplied by a factor of 0.73 to agree with the
upper end of the ambient data.  All analysis based on the HAPEM-
MS ambient motor vehicle levels will have this factor applied.
Adjusted urban, rural, and nationwide exposures are found in
Table 7-8.

     In an ambient monitoring study conducted by  the California
Air Resources Board  (CARB, 1992a) 20 monitoring sites were
established throughout the State of California to assess 1,3-
butadiene levels.  The range of the averages of the six basins
detailed in this study was 0.49 to 0.93 ug/m3 (0.22  to 0.42  ppb).
When this range is adjusted for the motor vehicle contribution
and integrated exposure it becomes 0.17 to 0.32 ug/m3  (0.08  to
0.15 ppb).  The upper end of this range compares  favorably with
the adjusted HAPEM-MS exposure number.

     The degree of confidence in the air monitoring programs,
especially the Urban Air Toxics Monitoring Program  (UATMP),
appears to be high.  The UATMP analyzed 1,3-butadiene using gas
chromatography/ multiple detector (GC/MD).  The GC/MD compound
identifications were confirmed by analyzing about 15% of the 1989
UATMP samples by gas chromatography/mass spectrometer  (GC/MS).
The GC/MS samples confirmed 94.1% of all the compound
identifications resulting from the initial analysis.  UATMP also
determined the level of confidence in its 1,3-butadiene
identification analysis.  The precision  (percent  coefficient of
variation, % CV) was calculated for the compound  response ratio
in the sample and the compound response ratio in  the standard.
It was determined that approximately 22% of the samples were
within +20% CV, while 60% were below the 0.10 ppbv detection
limit (EPA, 1990) .

     As demonstrated in the section above, it is  very clear that
there is a need for better ambient data and exposure methodology
for all the pollutants examined in this study.  An individual's
annual exposure could be very different then the  one number
presented in this study due to geographic and temporal variation
inherent in exposures.  Actual exposure estimates need to take
this into account.

7.5.3  Short-Term Micrenvironment Exposures

     The primary emphasis for 1,3-butadiene exposure will be
exposure in microenvironments that are enclosed,   increasing the
exposure to tailpipe emissions.  These microenvironments include
                               7-20

-------
                                                           EPA-420-R-93-005
                                                                April 1993
Table 7-8.
Adjusted Annual Average  HAPEM-MS Exposure
     Projections for  1,3-Butadiene.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Exposure
(ug/m3)
Urban
0.35
0.23
0.23
0.19
0.18
0.18
0.20
0.20
0.19
Rural
0.19
0.12
0.12
0.10
0.09
0.09
0.11
0.10
0.10
Nationwide
0.30
0.20
0.20
0.16
0.16
0.16
0.18
0.17
0.16
                                 7-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993
in-vehicle and parking garage exposure, though, actual exposure
information is only available for in-vehicle exposure.  This
information is taken from the Commuter's Exposure to Volatile
Organic Compounds, Ozone, Carbon Monoxide, and Nitrogen Dioxide
(Chan et al.,  1989), which focused on the driver's exposure to
VOC's in the Raleigh, NC area.  See the information in Section
4.2 for more details about the methodology, and Section 5.5.3 for
a description of the study.

     The in-vehicle exposure level of 1,3-butadiene was
determined in this study to have a mean of 3.0 ug/m3  and a
maximum measured level of 17.2 ug/m3.   Exterior to the vehicle,
the mean was determined to also be 3.0 ug/m3  with a maximum level
of 6.9 ug/m3.   This compares  to  ambient levels  of 0.31 to 1.48
ug/m3  determined through air  monitoring studies and presented in
Table 7-7.  Since for the majority of the population these are
short-term acute exposures to 1,3-butadiene,  the concern would be
with non-cancer effects.  Health information for non-cancer
effects is very limited and no RfC has been developed by EPA.
Inhalation of 1,3-butadiene is mildly toxic in humans at low
concentrations  (data on actual levels are not conclusive) and may
result in a feeling of lethargy and drowsiness.  At very high
concentrations, 1,3-butadiene causes narcosis leading to
respiratory paralysis and death.  Please see Section 7.8 for more
information on non-cancer effects.

     Due to more stringent fuel and vehicle regulations, short-
term exposure to 1,3-butadiene in microenvironments is expected
to decrease in future years.


7.6  Carcinogenicity of 1,3-Butadiene and Unit Risk Estimates

7.6.1     Most Recent EPA Assessment

     The information presented in Section 7.6.1 was obtained from
the EPA document Mutagenicity and Carcinogenicity Assessment of
 1,3-Butadiene  (EPA, 1985), EPA's Integrated Risk Information
System (IRIS)   (EPA, 1992), the Motor Vehicle Air Toxics Health
Information (Clement, 1991),  as well as the primary sources cited
in these documents.  The Carcinogenicity risk assessment for 1,3-
butadiene was last updated on IRIS in January 1992, and contains
data published through 1991.   However, with the exception of a
change in absorption factor  (used to calculate the target dose)
based on new pharmacokinetic data, the 1992 version of the 1,3-
butadiene risk assessment on IRIS is based on the same study as
the 1985 risk assessment.  EPA's Office of Research and
Development has just recently started the process to review the
1,3-butadiene risk assessment.  Section 7.6.3 summarizes recent
and ongoing research not included in the 1985 EPA risk assessment
for 1,3-butadiene.
                               7-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993
7.6.1.1   Description of Available Carcinogenicity Data

Genotoxicity

     Three studies have shown 1,3-butadiene to be mutagenic for
Salmonella typhimurium upon addition of mammalian hepatic  (liver)
homogenates for metabolism  (de Meester et al., 1978, 1980;
Poncelet et al.,  1980).   The weight of evidence available
suggests that 1,3-butadiene is a promutagen in bacteria; its
mutagenicity depends on metabolic activation by hepatic
homogenates prepared from chemically induced animals.  No whole-
animal mutagenicity studies have been reported.

     Pharmacokinetic and various types of toxicity studies
indicate that the carcinogenic effects of 1,3-butadiene can be
attributed to the metabolites 3,4-epoxybutene and/or 1,2,3,4-
diepoxybutane.  These metabolites, which are potent alkylating
agents  (chemically react with DNA),  have been shown to be
mutagenic and carcinogenic  (Lawley and Brookes, 1967; Ehrenberg
and Hussain,  1981).   The metabolite, 3,4-epoxybutene, is a
direct-acting mutagen in bacteria, and induces sister chromatid
exchanges and chromosomal aberrations in mice  (de Meester et al.,
1978; Voogd et al.  1981, Hemminki et al., 1980).

     1,2:3,4-Diepoxybutadiene is a bifunctional alkylating agent,
and as such it can form cross-links between two strands of DNA.
It is mutagenic in bacteria (Voogd et al., 1981, Wade et al.,
1979), fungi (Olszewska and Kilbey,  1975; Luker and Kilbey,
1982), and the germ cells of Drosophila (Sankaranarayanan, 1983;
Sankaranarayanan et al., 1983).  It also induces DNA damage in
cultured hamster cells and in mice  (Perry and Evans, 1975; Conner
et al.,  1983), is clastogenic in fungi and cultured rat cells
(Zaborowski et al.,  1983; Dean and Hodson-Walker, 1979), and
produces chromosome damage/breakage in Drosophila germ cells
(Zimmering,  1983).

     Under certain conditions, such as during rubber curing, 1,3-
butadiene can dimerize  (two molecules bonding together).  The
dimer was not mutagenic in the Salmonella preincubation assay in
the presence of liver homogenates from chemically induced rats or
hamsters (NTP, 1985).   In contrast,  the metabolites of the dimer
are mutagenic or clastogenic in various in vitro bacterial and
animal cell systems as a base-pair substitution mutagen  (Murray
and Cummins,  1979;  Simmon and Baden, 1980; Truchi et al., 1981;
Voogd et al.,  1981).  Therefore, the evidence indicates that 3,4-
epoxybutene,  diepoxybutane, and other mono- and diepoxide
metabolites are mutagens/clastogens in microbes and animals.

Animal Data

     In a chronic study conducted by the National Toxicology
Program (NTP,  1984), B6C3F1 mice  (50/sex/group) were exposed via
inhalation to 0,  625,  and 1,250 ppm 1,3-butadiene, 6 hours/day, 5


                               7-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993
days/week.  Because of excessive deaths, primarily due to
lymphoma, among treated animals, mice were sacrificed after 60-61
weeks instead of the planned lifetime exposure  (2 years).
Histopathologic examination revealed an increased frequency of
primary tumors at both exposure levels.  These tumors included
hemangiosarcomas (malignant tumors in the blood vessels) that
were found primarily in the heart, alveolar/bronchiolar adenomas
and carcinomas, and lymphomas.   The incidence of acinar cell
carcinoma (mammary) and granulosa cell tumor  (layer cells located
in the ovary) or ovarian carcinomas were increased in females in
the high-dose group.  According to EPA  (1985), NTP conducted an
audit of the study and reported that some genetic variation was
observed during 1981 in the male C3H parents of the mice used in
this study.   The effect of genetic nonuniformity in the hybrid
mice on the study was unknown,  but the results were considered
valid because of the use of matched concurrent controls.  More
recent work by NTP is discussed in Section 7.6.3.

     In a two-year study conducted by Hazleton Laboratories
Europe,  Ltd. (1981) (later published as Owen et al.,  1987, see
Section 7.6.3.2) Charles River CD rats  (110/sex/group) inhaled 0,
1,000, and 8,000 ppm 1,3-butadiene, 6 hours/day, 5 days/week, for
111 weeks (males) and 105 weeks (females).   The authors reported
significant increases in both common and uncommon tumors.  There
was an increase in multiple mammary gland tumors in females for
all treatment groups,  thyroid follicular adenoma and carcinoma in
the high-dose females, and Leydig cell  (cells located in the
testicles believed to be responsible for secreting testosterone)
adenoma and carcinoma in the high-dose males.  The report did not
include detailed histopathological evaluations and did not
perform independent data quality evaluation  (EPA, 1985).
Therefore, uncertainty about the number of tissues examined
limited the usefulness of animal-to-human extrapolation.  In
addition, the higher butadiene dimer (the combination of two
butadiene molecules) content of the material for rats might
contribute to the difference in the effective dose,  although its
effect on the study results is unknown  (EPA, 1985).   The
incidence of tumors in hormonal-dependent tissues was greater in
rats, although it may have been masked by early deaths in mice.

Human Data

     There were several epidemiological studies evaluating
mortality due to cancer in workers exposed to 1,3-butadiene.
These study results were inconsistent and limited because of
concurrent occupational exposures to other contaminants, usually
styrene  (potential carcinogen and leukemogen), and the lack of
adequate exposure data on 1,3-butadiene concentrations.

     Excess mortalities were reported in 6,678 male workers in a
rubber tire manufacturing plant in Akron,  Ohio  (McMichael et al.,
1974; 1976).  During a 9-year follow-up period from 1964-1972,
statistically significant increases in deaths were due to stomach


                               7-24

-------
                                                        EPA-420-R-93-005
                                                            April 1993
and prostate cancer, lymphosarcoma, and leukemia, as well as
diabetes mellitus and arteriosclerosis.  The 1968 U.S. male
population was used as the standard population.  An age-
standardized risk ratio of 6.2 for lymphatic and hematopoietic
cancer was calculated for workers with at least 5 years of
exposure.  To further evaluate the cancer-specific deaths,
McMichael et al.  (1976)  conducted a case-control study which
indicated that certain cancers were significantly elevated in
certain job classifications and with at least 5 years of
exposure.  Levels of exposure to 1,3-butadiene were not
quantified.

     In a historic prospective cohort study, increased incidence
of lymphatic and hematopoietic cancers were reported in 8,938
males in the rubber manufacturing plant during 1964-1973
(Andjelkovich et al.,  1976).   Data were collected from company
records, life insurance death claims, and bureaus of vital
statistics.  The increased mortality ratios were evaluated in
relation to work areas by using the entire cohort as a reference
group.  EPA (1985)  concluded that the study was limited because
of the uncertainty regarding the duration that a subject worked
in a specific job department (i.e., estimated duration ranged
from 10% to 100% of employment)  and the use of 1968 mortality
data which may have underestimated expected deaths.  Furthermore,
the levels of exposure were not quantified.

     Checkoway and Williams (1982)  evaluated the same group of
rubber manufacturing workers as the McMichael et al.   (1976) case-
control study.  The objective of this study was to quantify
exposure and to relate it to hematologic measurements.  Air
sampling of the plant and blood samples from the subjects were
taken in May 1979.   Time-weighted averages of 20.03 and 13.67 ppm
were determined for 1,3-butadiene and styrene, respectively.  No
association was found between hematologic values and 1,3-
butadiene exposure.  Because the study was cross-sectional,
excess cancer risk was not expected to be identified since
subjects who may have developed cancers and left the job force
were not available for evaluation.   The study was also limited
because air sampling could not be used to generalize past
exposure levels.   Furthermore,  concurrent exposure to more than
one potentially toxic chemical renders it impossible to associate
any adverse health effects that may be seen with exposure to a
particular chemical.

     A retrospective cohort mortality study was conducted on two
rubber plants in Texas by Meinhardt et al.  (1982).   The time-
weighted average exposures of butadiene were 1.24 and 13.5 ppm
for the two plants.  Subjects were also concurrently exposed to
styrene and benzene.  Deaths due to lymphatic and hematopoietic
cancers and lymphatic leukemia were exhibited, although they were
not statistically significant.   Results showed borderline
significance for the subcohort employed during the batch process
of the
                               7-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993
production process.  This finding may have been biased by
uncertainty in the number of deaths and/or factors in choosing
the study group (EPA, 1985).

     There were no excess mortalities in a retrospective cohort
mortality study involving 8 styrene-butadiene rubber plants
(Matanoski et al.,  1982).  None of the SMRs were statistically
significant.  According to EPA, information on the employees was
gathered from company records; however, this study evaluated less
than 50% of the population of the 8 plants and may have
underestimated the number of deaths.

7.6.1.2  Weight-of-Evidence Judgment of Data and EPA
Classification

     1,3-Butadiene is classified by EPA as a Group B2, probable
human carcinogen using EPA's Proposed Guidelines for Carcinogen
Risk Assessment (EPA, 1984).   This classification was based on
sufficient evidence from two species of rodents and inadequate
epidemiologic evidence, as described in Section 7.6.1.1 above.

     The mouse inhalation study by NTP (1984) was considered the
primary study for calculating the cancer risk estimate of 1,3-
butadiene (EPA, 1985).   It was the most appropriate choice
because the study was well-conducted and tumors were observed in
animals of both treatment groups.  The rat bioassay (Hazleton
Laboratories, 1981) had deficiencies that limited its use as the
primary data set for animal-to-human extrapolation.  According to
EPA, the quality of the study and its results have not been peer-
reviewed or published  (this study has since been published as
Owen et al., 1987,  see Section 7.6.3.2),  the histopathology
report was not available, and the calculated slope factors for
male and female rats were limited from a modeling standpoint
since it had only one effective dose.  In spite of the fact that
this study has been published, EPA still considers it inadequate
for risk assessment because of reporting problems, and because
the pharmacokinetic analysis in Owens et al.  (1987) is considered
by EPA to indicate that the effective doses were the same for
both treatment groups.

     The human studies were not used for determining unit risk
because there were inadequate data on the carcinogenicity of 1,3-
butadiene, a lack of exposure information, and concurrent
exposures to several other possible carcinogens (i.e., styrene)
to the workers.  However, EPA did conduct quantitative estimates
based on mouse-to-man extrapolation to predict human responses in
several epidemiologic studies.  Comparisons were hampered by
scarcity of information concerning actual exposures, age
distributions, and work histories.  Considering the uncertainties
in the human exposure data,  the estimate based on animal
extrapolation is consistent and the best that can be achieved.
                               7-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993
7.6.1.3   Data Sets Used For Unit Risk Estimates

     A summary of the data set used to calculate the cancer unit
risk estimate for 1,3-butadiene  (EPA, 1985) is presented  in Table
7-9.  EPA used both male and female mouse data sets from  the NTP
(1984)  study in determining the unit cancer risk estimates.
Animals with at least one of the statistically significant
increased tumors or tumors considered unusual were included in
the data set.  The data set had an adequate number of animals per
treatment group and the estimates across species for females were
relatively close which supports the confidence of the slope
factor.
However, only high doses were tested so the true shape of the
dose-response curve at low environmental levels is not known.

7.6.1.4   Dose-Response Model Used

     The low-dose linear multistage extrapolation model was used
for calculating the unit risk estimate  (see Appendix F for a
description of the linearized multistage model), although
alternative models were discussed but found inappropriate by EPA.
This model gave a conservative estimate while the other models
result in a lower risk estimate.

     Because 1,3-butadiene is considered a partially soluble
vapor,  the average dose/day is proportional to the 02 consumption
and is proportional to two-thirds of the weight and also  to gas
solubility in body fluids  (expressed as absorption coefficient).
All three factors listed above must be utilized when determining
average dose/day.  In the absence of experimental information,
the absorption fraction is assumed to be the same for all
species.  In
order to convert to internal dose in animals, EPA used the
absorption study by Bond et al.  (1986) which reported 20%
absorption of 1,3-butadiene following inhalation exposure in rats
and mice.

     Because mice were exposed to 1,3-butadiene for a less-than-
lifetime duration, an adjustment was made for extrapolation from
the 60-61  weeks in the NTP mouse study to two full years
(lifetime exposure).

7.6.1.5   Unit Risk Estimates (UCL and MLE)

     The upper-limit unit risks were 3.4X10"1 ppm"1  (3.8x10"
4 [ug/m3]"1)  for  male mice  and  1.9X10"1 ppm"1 (2.1xlO~4  [ug/m3] -1) for
female mice using a 20% absorption rate at low exposures.  The
geometric mean of unit risks was 2.5X10"1 ppm"1  (2.8xlO~4 [ug/m3]"1)
for the two mouse unit risks (EPA, 1992).  Calculating geometric
means of several unit risk estimates is standard EPA policy to
derive a single unit risk estimate.  However, the unit risk
should not be used if air concentrations exceed 16 ug/m3.   The
maximum likelihood estimate
                               7-27

-------
                                                             EPA-420-R-93-005
                                                                  April 1993
(MLE)  of unit risk based on  several tumor types observed in  male
                                 7-28

-------
                                                                                       EPA-420-R-93-005
                                                                                           April 1993
Table 7-9.
Summary  of  Data Set Used to Calculate Unit Risk  Estimate for  1,3-Butadiene,
Source
NTP
(1984)


Test Animal
B6C3F1 mice


Tumor Type
Hemangiosarcomas
of the heart,
lymphomas, and
alveolar/
bronchiolar
adenomas/
carcinomas


Administered
Dose
(ppm)
0
625
1250
Internal
Dose
(mg/kg/day)
0
18.4
27.8
Tumor
Incidence
2/50 (male)
4/48 (female)
43/49 (male)
31/48 (female)
40/45 (male)
45/49 (female)
                                              7-29

-------
                                                        EPA-420-R-93-005
                                                             April 1993
and female mouse NTP data for a 1 ppm continuous  lifetime
exposure in the air is 2.5xlO"2  (2.8xlO~5 [ug/m3] ~1) .

     The unit risk estimates in EPA  (1985) were different  from
those reported in EPA  (1992) because the values were  calculated
using the absorption data from the 1985 NTP  absorption  study
which reported an absorption rate of 54% in  mice  and  rats.
Therefore, according to EPA  (1985), calculations  from the  NTP
(1985) study resulted in an inhalation unit  risk  estimate  of
9-lxlO"1  (ppm)"1  (l.OxlO"3 [ug/m3]"1)  for  males  and 4.5X10"1 (ppm)"1
(S.OxlO"4  [ug/m3]"1) for females with a geometric mean  of 6.4X10"1
(ppm)"1  (7.2xlO~4  [ug/m3]"1) .   In this  case,  the unit  risk was not
to be used if the air concentration exceeded 40 ug/m3 since  the
slope factor may differ from that stated at  higher
concentrations.

7.6.2     Other Views and Risk Estimates

     This section presents alternate views and/or risk
assessments for 1,3-butadiene.  These alternate risk  assessments
are summarized in Table 7-10.  All alternate risk assessments are
expressed as UCLs; no MLEs are presented.

International Agency for Research on Cancer  (IARC)

     IARC has classified 1,3-butadiene as a  Group 2A  carcinogen.
A Group 2A carcinogen is defined as an agent that is  probably
carcinogenic to humans.  This classification is based on limited
evidence for carcinogenicity in humans and sufficient evidence
for carcinogenicity in animals (IARC, 1992) .

     IARC reviewed the available human data  and concluded  that
these studies do provide some evidence that  occupational exposure
to 1,3-butadiene is associated with an excess of  leukemias and
lymphomas.  However, these data are considered by IARC  to  be
limited because concomitant exposure to other potentially
carcinogenic agents (e.g., styrene and benzene) preclude any
definitive causative link to be drawn between exposure  to
1,3-butadiene and cancer.

     IARC concluded that the available animal data  provide
sufficient evidence of the carcinogenicity of 1,3-butadiene.
These data consist of inhalation studies in  mice  and  rats
conducted by NTP  (1984) and Hazleton Laboratories Europe,  Ltd.
(Owen et al.,  1987).  Details of these studies were mentioned
previously in Section 7.6.1.1.  No unit risk was  determined by
IARC.

California Air Resources Board (CARE)

     CARB (1992a,b) has performed an assessment of  the
carcinogenic risk of 1,3-butadiene using both the mouse (NTP,
1984; Melnick et al.,  1990) and rat  (Owen et al., 1987)  data in


                               7-30

-------
                                                            EPA-420-R-93-005
                                                                April 1993
the linearized multistage model.  As  EPA did, total  significant
tumor incidences
                                 7-31

-------
                                                                                    EPA-420-R-93-005
                                                                                         April 1993
Table 7-10.
Comparison  of  1,3-Butadiene Unit  Risk Estimates and Calculated Doses for an
Extra Lifetime Cancer Risk of  IxlO'6.
Source
OSHA (1990)b


EPA (1985)


ICF (1986)

Turnbull et al .
(1990)
(Environ)

EPA (1992)
Tumor Types
Pooled female mouse
tumors, multiple types0
Pooled female mouse
hemangiosarcomas
Pooled female rat
tumorsd
Pooled male and female
mouse tumors
Pooled male rat tumors
Pooled female rat
tumors
Male mouse lymphomas
Female mouse liver
tumors
Pooled male rat tumors9
Pooled female rat
tumorsh
Pooled female mouse
data
Classification
Human
Carcinogen


Group B2e


_

-

Group B2
Cancer
Unit Risk
Estimate
(ug/m3)-1
5.8xlO"6
2.7xlO"6
7.5xlO"6
7.2xlO"4f
4.7xlO"4
6.2xlO"4
3.4xlO"3
2.6xlO"4
5.9xlO"7
5.1xlO"7
2.8xlO"4
Dose (ug/m3)
For a
Cancer Risk
of IxlO"6
1.7X10"1
3.7X10"1
l.BxlO"1
1.4xlO"3
2. IxlO"3
1.6xlO"3
2.9xlO"4
3.8xlO"3
1.7xlO+0
2.0xlO+0
3.5xlO"3
Table 7-10.  Continued.
                                             7-32

-------
                                                                                                     EPA-420-R-93-005
                                                                                                           April 1993
Source
Hattis and
Watson (1987)



IARC (1992)
CARB (1992b)
CARB (1992b)
Tumor Type
Male rat -total tumors
Female rat -total tumors
Male mice-total tumors
Female mice-total
tumors
Multiple tumors
Total rat tumors (less
mammary f ibroadenomas
and uterine tumors)
Total mouse tumors
Classification
-



Group 2A1
-
-
Cancer
Unit Risk
Estimate
(uq/m3)-1
l.lxlO"7
1.3xlO"6
2.3xlO"5
l.VxlO"5
_
4.4xlO"6
l.VxlO"4
Dose (ug/m3)
For a
Cancer Risk
of IxlO"6
9.3xlO+0
7.6x10"'
4.5xlO"2
6.0xlO"2
_
2.3X10"1
6.0xlO"3
aMLEs  are  not  presented because they were not always calculated by the various organizations.   Furthermore, EPA
dose not generally compare MLEs based on animal data because of the high variability associated with  these
numbers.  Therefore, they are of little value.
bSource:   Grossman and Martonik,  1990.   Based on estimates of extra risk per 10,000 for a lifetime occupational
exposure.   The following assumptions were made:  absorption at low doses is 54%, adult body  weight  is 70  kg,
adult breathing rate is 10 m3/8-hour day,  exposure is for 250 days/year for 45 years of a 74 year lifetime.
Incidence of  lymphoma excluded from pooled tumor incidence.
dHigh-dose group  dropped from the analysis.
eGroup B2  = Probable Human Carcinogen
UCLs  is the geometric mean of the UCLs estimated from the male mouse data and the female mouse data.
Incidence of  Zymbal gland carcinoma excluded from pooled tumor incidence.
hlncidence of  mammary fibroadenoma excluded from pooled tumor incidence.
       2A  = Probable human carcinogen.
                                                      7-33

-------
                                                        EPA-420-R-93-005
                                                            April 1993
for each species and sex were used instead of individual site-
specific tumor incidences because CARB believed that tumors which
rapidly resulted in animal mortality may have masked the
development of tumors at other, possibly more sensitive sites.

     CARB concluded that, for use in risk assessment, the quality
of the Melnick et al.  (1990) is superior to that of the rat data.
The primary reasons for this conclusion are: 1) the use of lower,
more relevant dose levels in the Melnick et al.  (1990) study;
2)the use of five dose levels in the Melnick et al.  (1990) study,
compared to two in the rat study; 3) the presence of two mouse
studies; 4) the fact that the rat study has not been replicated;
5) the consistency in sites of carcinogenicity between the two
mouse
studies; 6) the greater detail in the available mouse data which
allows in-depth analysis; and 7) suggestions from limited
epidemiological observations that 1,3-butadiene exposure may be
associated in humans with lymphatic and hematopoietic cancers,
effects that were seen in mice.  The continuous internal dose
(i.e., the dose of butadiene that is retained in the animal) was
considered by CARB to be the best estimate of delivered dose
(i.e., the dose of butadiene that is actually available at target
tissue sites) available.  The continuous internal doses were
derived from the applied external doses using the data of Bond et
al. (1986)  .

     CARB calculated the theoretical human risk associated with a
continuous lifetime exposure to butadiene  (qx*,  95% UCL) as  1.7 x
10"4  [ug/m3] "1 based on the mouse inhalation study of Melnick et
al. (1990).  The risk based on the mouse data is comparable to
EPA's current unit risk of 2.8xlO"4  [ug/m3]"1, which is a geometric
mean of the unit risks derived from the male and female mouse
data sets  (EPA, 1992,  see Section 7.6.1.5).

     CARB also fit the data to various other models.  They
concluded that the data gave better fits to the linearized
multistage model or the GLOBAL 86 version of the linearized

multistage model than to the other models.  They concluded that
the mouse provides the best estimate for the upper bound for
plausible excess cancer risk to humans.

     Based on the findings of 1,3-butadiene-induced
carcinogenicity and the results of the risk assessment, CARB
finds that, at ambient concentrations, 1,3-butadiene is an air
pollutant which may cause or contribute to an increase in
mortality or an increase in serious illness, or which may pose a
present or potential hazard to human health.

Occupational Safety and Health Administration  (OSHA)

     OSHA contracted with ICF/Clement  (ICF/Clement, 1986) to
conduct a risk assessment on 1,3-butadiene to be used as rule-


                               7-34

-------
                                                        EPA-420-R-93-005
                                                            April 1993
making support in setting a revised occupational exposure
standard for this chemical.  ICF/Clement's risk assessment
differed from EPA (1985) only by the data set used, and the final
adjusted doses used in the risk assessment.  ICF/Clement
expressed dose in parts per million, employed the linearized
multistage model for low-dose extrapolation, adjusted for less-
than-lifetime exposure, and adjusted the experimental dose for
absorption.  This adjustment differed from that used by EPA
(1985) and was done using a line generated by plotting a log-log
scale based on data reported in EPA (1985) that indicated
retention of 1,3-butadiene is inversely related to dose.  As in
the EPA (1985) risk assessment, the data from the NTP (1984)
mouse bioassay were used to calculate the unit risk for 1,3-
butadiene.   However, EPA used pooled tumor incidence data for
mice, whereas ICF/Clement used site-specific individual tumor
data.  ICF/Clement claimed that the multistage model is based on
the observation that cancer is a progressive disease that
develops in stages.   Data are available suggesting that the
number of stages or the stage at which a particular carcinogen
acts may vary among different organ systems in the body.
Therefore,  ICF/Clement concluded that the use of pooled tumor
data is not well-justified on theoretical grounds, even though
EPA felt differently when it developed a potency factor for 1,3-
butadiene.   The results of these analyses for the worst case
(male mouse lymphoma)  and best case (female mouse liver tumors),
as compared to those calculated by EPA are presented in Table 7-
10.

     OSHA also conducted its own risk assessment of 1,3-butadiene
(Grossman and Martonik, 1990).   In this risk assessment,
experimental dose was measured in milligrams per kilogram per day
and adjusted for absorption  (method not specified).  The risks
were derived using both pooled tumor and site-specific tumor
incidence data for both mice (NTP, 1984) and rats  (Owen et al.,
1987), using the multistage model.  The results of these
analyses,  as compared to those performed by EPA (1985),
ICF/Clement (1986),  CARB (1991),  and Environ (see section below)
are presented in Table 7.10.

Chemical Manufacturer's Association (CMA)

     CMA contracted with Environ Corporation to perform an
independent assessment of the potential risk to workers from
exposure to 1,3-butadiene  (Turnbull et al., 1990).  Environ's
risk assessment departed from that of EPA  (EPA, 1985) with regard
to the data set used and the low dose extrapolation models
employed.   However,  like EPA,  (1985),  they measured dose in
milligrams per kilogram per day and adjusted the experimental
dose of 1,3-butadiene for retained dose, assuming an absorption
of 54%, regardless of dose.  EPA  (1992) has since revised their
calculation of the unit risk by assuming a 20% absorption rate at
low exposures as per Cote and Bayard (1990) (see discussion in
Section 7.6.1.5).  Environ disagreed with the choice of the mouse


                               7-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993
data from the NTP (1984) study as the basis for the unit risk for
a number of reasons (Environ, 1987) .

     In an attempt to address some of the perceived uncertainties
with the NTP (1984)  mouse data listed in Environ 1987, Environ
employed the following procedures.  Separate extrapolations were
conducted based on tumor-bearing animals having any of the tumors
that showed a significant increase in incidence in one or both of
the treated groups,  and on the same animals except those that
developed lymphoma.   To account for the less-than-lifetime
exposure duration in the mouse study, the Hartley-Sielken general
product model was used.  The results of these calculations are
summarized in Table 7-10.

     Environ also calculated unit risks on the rat data from the
Hazleton Laboratories Europe, Ltd. study (Owen et al., 1987)
using three different low-dose extrapolation models.  However,
the tumor incidences from this study that were used by Environ
differed from those used by EPA for unexplained reasons.  The
results of the unit risk estimates resulting from the use of the
multistage model only are presented in Table 7-10.

     The results of these analyses generally predicted lower
risks than those predicted by EPA, and indicate that mice appear
to be at a greater risk  (by a factor of 5-fold to 40-fold) than
rats.  Environ noted that some of this species difference (3-fold
to 5-fold) may be due to differences in metabolism, and that mice
metabolize 1,3-butadiene to the carcinogenic epoxide derivatives
at a higher rate than rats (though not mentioned by Environ in
its analysis, this species difference could reflect decreased
elimination of reactive intermediates in mice as compared to
rats).   In addition, Environ predicted the lifetime risk to
humans using several exposure levels based on all of the risk
estimates derived above.  This risk was then used to calculate
the expected number of extra deaths from lymphopoietic cancer and
compared them to the actual number of deaths observed in the
Matanoski et al. (1982) cohort.  This exercise led to the
conclusion that these risks were inconsistent with the
observations made in occupational studies,  i.e., the animal risks
(particularly those based on the mouse data) overpredicted the
risk to humans.  However, it should be noted that the revised
risk estimate for butadiene cited in EPA (1992) that incorporates
a new absorption factor results in the prediction of 40% less
excess cancer cases (Cote and Bayard, 1990).  Therefore, the
conclusions of Turnbull et al.  (1990) with regard to
overpredicting the risk to humans exposed to butadiene may no
longer be valid.

Hattis and Wasson (1987)

     Hattis and Wasson  (1987) developed a
pharmacokinetic/mechanism-based model for butadiene in an attempt
to further refine the estimate of the "effective" dose for 1,3-


                               7-36

-------
                                                         EPA-420-R-93-005
                                                              April 1993
butadiene to be used in risk assessment.  This model  estimates
the effective dose  "as  the total amount of butadiene  that
eventually undergoes at least the first step of metabolic
activation to 3,4-epoxy-l-butene".   Development of  this  model
required an estimation  of the octanol/water partition coefficient
from chemical structural information and a water/air  partition
coefficient from  aqueous solubility information as  a  function of
temperature so  that  tissue/blood and blood/air partition
coefficients could be estimated.  Contrary to previous
assumptions that  butadiene was metabolized only in  the liver, the
model was structured to allow butadiene metabolism  throughout the
"vessel-rich group"  (kidneys, viscera, and brain) in  addition to
the liver.  Finally,  maximal metabolic rates for humans  were
scaled using general metabolic rates -- (body weight)'75.  Using
their model, Hattis  and Wasson  (1987)  calculated rodent
metabolized doses that  were 2-4.5 times the absorbed  doses  used
in earlier  (i.e., EPA [1985]  and Environ  [1987]) risk
assessments.  This larger metabolized dose effectively reduces
the apparent carcinogenic potency of butadiene as compared  to the
earlier risk assessments.  Another factor that reduces the
carcinogenic risk of butadiene is the fact that this  model
predicts that net absorption will represent only about 11-15% of
the butadiene reaching  the alveoli over an 8-hour period, and
only 8-10.5% of total inhaled butadiene.

     Risk assessments conducted prior to Hattis and Wasson's work
assumed 50% of  total inhaled butadiene would be absorbed by
humans at low doses.   (However, in the most current EPA
assessment  [1992], an absorption factor of 20% is used,  see
discussions above).   This difference results in a further
reduction of human delivered dose,  and therefore, risk.   Hattis
and Wasson  (1987) also  differed from the EPA  (1985) and  Environ
(1987) risk assessments in the manner in which they treated tumor
incidence.  Rather than add up the tumor-bearing animals at all
sites with statistically significant tumor increases  before
calculating risk, they  calculated separate risks from each
individual site and  then added up the overall expected risks from
all of the sites  at  the end.1  The  effect  of such an approach is
most likely to  overestimate the risk.   The UCLs calculated  using
the effective doses  estimated with their model as compared  to the
UCLs calculated by EPA,  CARB, ICF/Clement, OHSA, and  Environ are
summarized in Table  7-10.

National Institute for  Occupational Safety and Health (NIOSH)

     NIOSH  (Dankovic et al.,  1991)  developed a quantitative risk
      While it is appropriate and acceptable to consider each tumor separately,
model each response, and then combine the probabilities at the end to arrive at
an overall expected risk,  it is not correct to simply add the individual tumor
risks.  It is correct to add the individual risks together and then subtract out
the product of the risks to arrive at the overall risk, and this is most likely
what Hattis and Wasson (1987) actually did, as evidenced by the numbers presented
in their table.

                               7-37

-------
                                                        EPA-420-R-93-005
                                                            April 1993
assessment of 1,3-butadiene based on the NTP bioassay in E6C3F1
mice of Melnick et al.   (1990) .   The risk assessment utilized the
data from the published report as well as the data on the time of
death and tumor status of each individual mouse in the study.
The NIOSH study also chose to use exposure concentration instead
of internal dose and the Weibull time-to-tumor model to determine
their risk estimate.

     Excess risk estimates were derived from fitting the one-
stage,  two-stage, and three-stage Weibull time-to-tumor models to
the seven individual tumor types observed in the male mice, and
to the nine individual tumor types observed in the female mice.
Overall,  the estimate yielding the largest extrapolated human
risks at low exposure concentrations, that is, the most sensitive
site, was the female mouse lung.  Based on this site, the
projected excess risk for a person occupationally exposed to 2
ppm 1,3-butadiene, for an entire working lifetime, is estimated
to be 597 cases of cancer per 10,000 (5.97xlO~2), or
approximately 6 per 100.

     Caution must be taken in comparing this number to the
previously stated risk estimates summarized in Table 7-10.  The
risk estimates in Table 7-10 are based on a 70 year lifetime
exposure to 1 ug/m3  of  1,3-butadiene whereas,  the  NIOSH risk
estimate is based on a working lifetime exposure of 2 ppm
(4.42xl03 ug/m3) 1,3-butadiene.

7.6.3     Recent and Ongoing Research

7.6.3.1   Genotoxicity

     Two new studies were published after the EPA assessment that
supported the observation that metabolites of 1,3-butadiene are
genotoxic (Gervasi et al.,  1985; Sharief et al., 1986).  The
study by Gervasi et al. (1985)  observed that 1,2:3,4-
diepoxybutane is a potent mutagen in the S. typhimurium mammalian
microsome assay.  Gervasi et al. (1985) also demonstrated that
the potency of 1,2:3,4-diepoxybutane in this assay correlated
well with the alkylating ability of this compound using
nicotinamide as a substrate.  Sharief et al.   (1986) examined the
in vivo genotoxicity of another 1,3-butadiene metabolite, 1,2-
epoxybutene-3.   A single intraperitoneal injection of 1,2-
epoxybutene-3,  at doses as low as 25 mg/kg, produced a
significantly increased frequency of sister chromatid exchange
(SCE) and chromosomal aberrations in bone marrow cells of C57B1/6
mice.

     At the time of the EPA assessment (EPA,  1985), no in vivo
studies of the genotoxicity of 1,3-butadiene were available for
review.  A number of inhalation studies have since been completed
that examine the genotoxic effects of 1,3-butadiene exposure.
For example, exposure of rats and mice to concentrations of 1,3-
butadiene ranging from 10 to 10,000 ppm for 6 hr/day for 2 days


                               7-38

-------
                                                          EPA-420-R-93-005
                                                              April 1993
produced no  increase in the frequency of micronucleus (MN)
induction or SCE  in bone marrow of Sprague-Dawley rats,  but
significantly increased the frequency of MN and SCE in bone
marrow of B6C3F1  mice at doses as low as 100 ppm (Choy et al.,
1986; Cunningham  et al.,  1986) .2   No increase in MN or SCE was
observed in  the B6C3F1 mice at 50 ppm.  Exposure of B6C3F1 mice
to concentrations of 1,3-butadiene ranging  from 6.25  to 625 ppm
for a somewhat longer period  (6 hr/day, 5 days/week,  for 2 weeks)
revealed significant increases in SCE at 6.25 ppm,  MN at 62.5
ppm, and chromosomal aberrations at 625 ppm in  bone marrow  (Tice
et al., 1987).  Chromosomal aberrations were predominantly
chromatid-type breaks and exchanges.  Increases in MN in
peripheral blood  were observed at doses of  1,3-butadiene as low
as 6.25 ppm  following longer-term exposure  (for 6 hr/day, 5
days/week, for 13 weeks)  (Jauhar et al., 1988).   The  potent
genotoxicity of 1,3-butadiene in the mouse  compared with the
absence or low level of such effects in rats are consistent with
the relative carcinogenic effects in these  two  species.

     The strain specificity of the genotoxic effects  in the mouse
was tested by comparing chromosomal damage  in B6C3F1  mice with
that seen in NIH  Swiss mice (Irons et al. 1987a) .   After a single
6-hour exposure to 1,250 ppm 1,3-butadiene,  a high frequency of
chromosomal  aberrations,  chromatid breaks,  and  chromatid and
isochromatid gaps were seen in both strains of  mice.3  The NIH
Swiss mouse  does  not possess murine leukemia virus, indicating
that the genotoxicity of 1,3-butadiene  is not dependent on the
presence of  this  virus.   However, the virus may play a role in
the expression of murine leukemogenesis in  the  B6C3F1 strain.

     Exposure of  B6C3F1 mice and Wistar rats to (14C)-1,3-
butadiene  (approximately 700 ppm for 4-7 hours)  resulted in
covalent binding  of the radioactivity to liver  nucleoproteins  (a
combination  of a  nucleic acid and a protein that is found in cell
nuclei) and  DNA in both species  (Kreiling et al.,  1986a).  The
alkylation of nucleoproteins was approximately  twice  as high in
mice as in rats.   The degree of alkylation  was  proportional to
the different rates of metabolism of 1,3-butadiene in these two
species.  In contrast, the incorporation of radioactivity into
DNA was approximately equal in both mice and rats.   It is unclear
to what extent the incorporation of radioactivity in DNA
represented  alkylation of nucleosides or metabolic incorporation
into nucleosides.  However, an alkylation product of  guanine, 7-
      Micronuclei are formed after cell division when pieces of chromosomes do
not get included within either nucleus of the newly formed cells.  Sister
chromatid exchange occurs when pieces of DNA break off and reattach to another
piece of DNA.  An increased frequency of micronuclei or SCE indicates chromosome
breakage.

      A chromatid gap is a short missing region of DNA in one strand of a
dividing chromosome.  An isochromatid gap is a short missing region of DNA in one
strand of an abnormally dividing  (i.e., two strands break instead of each of the
strands separating intact) chromosome.

                               7-39

-------
                                                        EPA-420-R-93-005
                                                            April 1993
(l-hydroxy-3-buten-2-yl) guanine was identified in mouse liver
DNA after inhalation exposure to 1,3-butadiene  (Laib and
Kreiling, 1987) .

     A dominant lethal study in CD-I mice was performed following
inhalation exposure of males to concentrations of 1,3-butadiene
ranging from 200 to 5,000 ppm for 6 hr/day for 5 days  (Hackett et
al.,  1988b).   A significant increase in intrauterine deaths was
observed in females bred with males exposed to 1,000 ppm but not
in females bred with males exposed to 5,000 ppm.

     Cytogenetic monitoring of 1,3-butadiene rubber workers was
reported in an abstract by Zhou et al.   (1986).  No significant
increase in chromosomal aberrations or SCE in peripheral
lymphocytes was observed when a group of 30 styrene-butadiene
workers were compared with matched controls.  Sex, age, and
smoking status were considered in the analysis.  However, 1,3-
butadiene exposure levels were not measured and workers may have
been exposed to toluene.

7.6.3.2   Pharmacokinetics

     1,3-Butadiene is a carcinogen in both rats and mice, with
mice being substantially more sensitive than rats (Csanady and
Bond, 1991a).   In the development of the pharmacokinetic model by
CUT, both in vitro and in vivo studies have demonstrated that
1,3-butadiene is metabolized by cytochrome P-450 to l,2-epoxy-3-
butene (butadiene monoepoxide, BMO).  Further metabolic activity
may transform BMO to two other metabolites, 1,2-epoxy-3,4-
butanediol and diepozybutane  (DEB).  All three epoxides can
potentially interact with DNA (Bryant and Osterman-Golkar, 1991).
Not  all of the metabolites of 1,3-butadiene have been  identified
yet,  and those that have been identified, there has been limited
pharmacokinetic testing.

     Differences in the pharmacokinetics of 1,3-butadiene in mice
and rats have been more closely examined in recent studies in an
effort to explain the differences in the carcinogenic  potency of
1,3-butadiene in these two species.

     Many studies (Bond et al.,  1986, 1987, 1988; Deutschmann and
Laib, 1989;  Kreiling et al.,  1986b, 1987, 1988; Schmidt and
Loeser, 1985,  1986;  Jelitto et al., 1989) suggest that
differences in species carcinogenicity susceptibility  may be
related to differences in 1,3-butadiene metabolism.   When
compared to the rat, mice have both a higher rate of 1,2-
epoxybutene-3 synthesis and presence of DNA adducts,  as well as a
limited ability to detoxify this metabolite.

     A physiologically-based, pharmacokinetics model for 1,3-
butadiene exposure in rats and mice, based on the conversion of
1,3-butadiene to 1,2-epoxybutene-3, was developed by Hattis and
Wasson (1987)  utilizing blood butadiene concentrations as


                              7-40

-------
                                                        EPA-420-R-93-005
                                                            April 1993
described by Bond et al.   (1986) and metabolic rates as described
by Kreiling et al.   (1986b).   For this model, blood/air and
tissue/air partition coefficients were estimated from structural
and solubility information.   According to this model, however,
differences in pharmacokinetics failed to account for the
differences in carcinogenicity of 1,3-butadiene in these two
species.

     Recent data from Csanady and Bond (1991b) indicate that the
maximum rates of 1,3-butadiene metabolism in liver microsome
isolated from humans,  B6C3F-L mice,  and  Sprague-Dawley rats  (all
male)  are in the ratio of approximately 3:3:1.  However, the
investigators found that the key ratio that governs metabolic
rates at low concentrations was 5-fold greater in mouse liver
than in human liver.  This means that the mouse produces more  in
the way of potential adduct forming metabolite (BMO), and thus is
considered a more susceptible species.   Csanady and Bond (1991b)
also report that B6C3F-L mouse lung  microsomes  are  much more
active than human lung microsomes in metabolizing 1,3-butadiene
to BMO, with the key metabolic ratio in the mouse lung being
approximately 6-fold greater than in the human lung.

     In a recent presentation by CUT  (Recio et al. , 1991)  the in
vivo mutagenicity of 1,3-butadiene was assessed in lung, liver,
and bone marrow using a transgenic mutagenicity assay.  It was
found that the overall activation  (by oxidation)  of 1,3-butadiene
to BMO was significantly higher for mice, especially in the lung.
The detoxification of BMO (by hydrolysis) is slower in the mouse
than in the rat or the human; thus, this correlates with the
higher carcinogenicity sensitivity of mice than rats to 1,3-
butadiene.

7.6.3.3   Carcinogenicity - Animal Studies

     Additional information regarding the carcinogenicity of 1,3-
butadiene in animals has been presented since the EPA
mutagenicity and carcinogenicity assessment of 1,3-butadiene was
performed in 1985.   For example, another long-term inhalation
study of 1,3-butadiene in B6C3F1 mice was initiated.  The need
for another mouse carcinogenicity study arose because the study
that was originally evaluated in the EPA assessment demonstrated
a strong multiple-organ carcinogenic response to 1,3-butadiene at
exposure concentrations of 625 and 1,250 ppm  (Huff et al.,  1985),
but clear dose-response relationships were not established and
the study was terminated after 60 weeks of exposure because of
reduced survival due to fatal tumors.  Therefore,  a study that
examined lower exposure concentrations (6.25-625 ppm) was
initiated.   Preliminary results from that study through week 65
were reported  (Melnick et al., 1988, 1989a, 1989b).  After 65
weeks of exposure,  73/90 males and 80/90 females exposed to 625
ppm have died.  The primary lesion observed in these animals was
lymphocytic lymphoma.   This was more prevalent in the males than
in females.   Other types of cancer observed in the high-dose


                               7-41

-------
                                                        EPA-420-R-93-005
                                                            April 1993
animals included hemangiosarcoma of the heart, squamous cell
neoplasms in the forestomach, alveolar-bronchiolar neoplasms,
adenoma of the Harderian gland, mammary gland adenocarcinoma, and
granulosa cell neoplasms of the ovary.  Elevated incidences of
these neoplasms were also seen at 200 ppm 1,3-butadiene.
Alveolar-bronchiolar neoplasms of the lung in females were
increased above the incidence in controls at all concentrations
of 1,3-butadiene tested.

     As part of this study, three groups of animals were also
exposed for limited periods of time to study the relationship
between exposure levels and duration of exposures on butadiene-
induced carcinogenicity.  The groups consisted of male mice
exposed to 625 ppm for either 13 weeks or 26 weeks and mice
exposed to 312 ppm for 52 weeks.  The animals were then held
until 65 weeks from the start of exposure.  By week 65 of the
study, the incidence of lymphocytic lymphoma in animals exposed
to 625 ppm for 26 weeks (60%) was twice that observed in animals
exposed to 625 ppm for 13 weeks (30%), but was much greater than
the incidence in animals exposed to 312 ppm for 52 weeks (6%).
Thus,  the multiple of the exposure concentration times the
exposure duration did not predict the incidence of lymphocytic
lymphoma in these mice.  However,  this study revealed that the
early incidence of fatal lymphocytic lymphoma in the high-dose
animals appeared to limit the expression of tumors at other
sites.  Substantially higher levels of some tumor types were
observed in the dose group with low levels of lymphatic lymphoma
than in the dose group with high levels of lymphatic lymphoma.
For example, a much higher incidence of hemangiosarcoma of the
heart was observed in mice exposed to 312 ppm for 52 weeks (30%)
than in mice exposed to 625 ppm for 26 weeks  (10%).   Other tumor
types observed at a higher incidence in mice that survived 45-65
weeks that were not observed in the NTP study because of early
deaths due to lymphatic lymphoma include squamous cell neoplasms
of the forestomach, alveolar-bronchiolar neoplasms,  Harderian
gland adenomas, adenocarcinomas of the mammary gland, granulosa
cell neoplasms of the ovary, and hepatocellular neoplasms.   This
study is significant in that it demonstrated that:   (1) exposure
to lower levels of 1,3-butadiene than those used in the study
that served as the basis for the EPA risk assessment allows the
expression of neoplasms at other sites because of a lower number
of early mortalities;  (2)  a clearer dose response relationship
for 1,3-butadiene-induced lymphocytic lymphomas was obtained
using the lower exposure levels because of increased survival;
and,  (3) the multiple of exposure duration and concentration does
not predict the incidence of lymphocytic lymphomas.   These
findings are relevant to the current EPA assessment of 1,3-
butadiene carcinogenicity because they demonstrate that the
induction of neoplasms in mice at multiple sites  (i.e., some that
were not considered in the current assessment) occurs at lower
concentration levels than those used to derive the cancer potency
factor.
                               7-42

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     A study characterizing the lymphomas observed in B6C3F1 mice
exposed to 1,250 ppm 1,3-butadiene for 28 to 45 weeks reported
that the lymphomas consisted of well-differentiated lymphoblasts
of T-cell origin  (i.e.,  cells that develop into lymphocytes that
subsequently migrate to the thymus gland) with variable but
elevated levels of murine leukemic virus antigens  (virus proteins
found in mice that cause leukemia) (Irons et al.,  1986c).  In
order to test the role of the endogenous retrovirus (i.e., a type
of virus that is known to cause cancer), murine leukemic virus,
in the development of the thymic lymphoma/leukemia in B6C3F1
mice, NIH Swiss mice (which do not express the retrovirus) were
exposed to 1,3-butadiene under identical conditions as the B6C3F1
mice (Irons et al.,  1989).  This study revealed that B6C3F1 and
NIH Swiss mice exposed to 1,250 ppm 6 hours/day,  5 days/week, for
1 year had similar increases in chromosomal aberrations and
micronuclei in bone marrow and micronuclei in the peripheral
blood,  but NIH Swiss mice had a much lower incidence of lymphoma
(14%) than did the B6C3F1 mice (57%).  The tumors in both strains
were morphologically similar, but the lymphoblasts in NIH Swiss
mice did not have surface antigens for the murine leukemic virus.
These results demonstrate that expression of the retrovirus is
not entirely responsible for the incidence of lymphoma.  However,
the murine leukemic virus may influence the incidence of the
lymphoma in B6C3F1 mice.

     The carcinogenicity study in rats that was reviewed in the
EPA carcinogenicity assessment but was available only as an
unpublished report from Hazleton, has been published  (Owen et
al.,  1987).  This report  (Owen et al.,  1987) contains the same
information as the Hazleton report that is summarized in the EPA
carcinogenicity assessment.  In summary, the data suggested
treatment related increases in mammary gland, thyroid, and
testicular tumors.  The authors proposed that the carcinogenic
effect was likely an indirect effect mediated through the
endocrine system rather than through the production of reactive
intermediates.

7.6.3.4   Carcinogenicity - Epidemiological Studies

     The results of another epidemiologic study of 1,3-butadiene-
exposed workers have been reported since the original EPA
assessment (EPA, 1985).   This study examined the mortality of
2,586 workers employed for at least 6 months between 1943 and
1979 at a 1,3-butadiene manufacturing facility (Downs et al.,
1987).   Data regarding exposure levels were not available, but
workers were divided according to 4 qualitative exposure
categories based on employment records.   The categories of
exposure were:  low exposure, routine exposure (included process
workers),  nonroutine exposure (intermittent exposure;  maintenance
workers),  and unknown exposures.   The overall mortality of the
workers was significantly below the U.S. national average
(standardized mortality ratio (SMR) = 80).   However, the SMR for
lympho- and reticulo-sarcoma of the whole cohort was


                               7-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993
significantly greater than the U.S. national average  (SMR = 235;
95% CI = 101 to 463).   Lympho- and reticulosarcoma are malignant
diseases of the lymphatic system and the reticuloendothelial
system, respectively.   The reticuloendothelial system are cells
scattered throughout the body that have the power to ingest
particulate matter.  When calculated by exposure category, the
routine exposure group had significant increases in
lymphohematopoietic and kidney cancer and the nonroutine exposure
group had a significantly increased rate of leukemia when
compared to the U.S. national average.  These rates of cancer
were elevated when compared to the local cohort, but were not
statistically significant.  Limitations of this study included an
unreliable designation of race, lack of worker histories, and the
observations that nearly half of the cohort worked at the
facility for less than 5 years and that many workers had spent
time working at neighboring styrene-butadiene rubber plants.

     An update of the Downs et al.  (1987) study and updates of
two studies originally contained in the EPA 1985 assessment have
also been published.  In the update of the study by Downs et al.
(1987), the workers' mortality experience through 1985 was
examined (Divine, 1990).  One additional death from lymphosarcoma
had occurred since the previous analysis by Downs et al.   (1987).
Findings were similar to those reported in the 1987 study.
Excess mortality due to lymphatic and hematopoietic cancers were
seen primarily in those occupational categories with the greatest
known exposure to 1,3-butadiene, in those with less than 10 years
of employment, and in those employed during World War II.

     In the update of the study by Matanoski and Schwartz (1987) ,
followup of workers was improved and extended through 1982
(Matanoski et al.,  1990).   The cohort in the update was
restricted to 12,100 workers by limiting employees from the one
Canadian plant to those who had worked 10 years or more or who
had reached age 45 during employment.  Overall mortality in these
workers was less than the U.S. national average (SMR = 81).   The
only significant increase in mortality observed in the workers
was in arteriosclerotic heart disease among black employees
compared to the U.S. national average (SMR = 1.48, 95% CI: 1.23-
1.76).  The workers were subdivided according to the job held the
longest.  The categories of employment included production,
utilities,  maintenance,  and a combination of all others.
Significant increases in mortality were observed among the
production workers.  Combined race data showed a significant
increase in other lymphatic malignancies (SMR = 2.60, 95% CI:
1.19-4.94)  and blacks had significant increases in all
lymphopoietic cancers (SMR = 5.07,  95% CI:  1.87-11.07) and in
leukemia (SMR = 6.56,  95% CI: 1.35-19.06).   Whites had elevated
mortality due to lymphatic (SMR = 2.30,  CI:  0.92-4.73) and
hematopoietic (SMR = 1.10, CI: 0.58-1.87) malignancies, but the
increase was not statistically significant.   This study is
somewhat limited in that the race designation of approximately
15% of the workers was unknown but was assumed to be white.
                               7-44

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     A nested case-control study of the lymphopoietic neoplasms
from the cohort of butadiene workers studied by Matanoski and
Schwartz (1987) was conducted by Santos-Burgoa (1988).   The 59
cases of lymphopoietic neoplasms were matched to controls based
on plant, age, hire date, duration of work, and survival to the
death date of the case.  Exposures to 1,3-butadiene for the cases
and controls were estimated through a ranked job exposure matrix
which was multiplied by the duration of exposure to yield
estimates of cumulative exposure.  In a matched analysis based on
a categorization of exposure above and below the geometric mean,
a statistically significant odds ratio of approximately 2.0 (OR =
2.0) for all lymphopoietic neoplasms and OR = 9.0 for leukemia
was calculated for butadiene exposure.  Both types of analysis
that were performed in this study showed a significant trend for
leukemia with cumulative butadiene exposure, but not for all
lymphopoietic neoplasms.

     An occupational epidemiological pilot study was conducted
(Ward et al.,  1992) to evaluate the effects of 1,3-butadiene
exposure on the frequencies of lymphocytes containing mutations
at the hypoxanthine guanine phophoribosyl transferase (hprt)
locus in workers in a 1,3-butadiene production plant.  Seven
workers from areas of the plant where the highest exposures to
1,3-butadiene occur were compared to four workers from plant
areas where 1,3-butadiene exposures were low.  In addition, four
workers from the investigating laboratory were also studied as
outside controls.  All the subjects were non-smokers.  An air
sampling survey indicated that average 1,3-butadiene levels in
the high exposures area were about 3.57.5 ppm while they were
0.030.03 in the low exposure area.  The low-exposed controls and
the outside controls mean variant frequencies, 1.19 and 1.03
respectively,  were not significantly different,  but the mean
frequency of mutant lymphocytes in the seven exposed subjects
(4.09)  was significantly higher when compared to the means of the
eight controls.  The observation of an elevated mean in the
exposed subjects indicates that exposures occurring in areas
where higher concentrations of 1,3-butadiene have been documented
were sufficient to induce higher frequencies of somatic cell
mutants.  Additional studies are being conducted to confirm the
effects that have been observed.
7.7  Carcinogenic Risk for Baseline and Control Scenarios

     Table 7-11, summarizes the annual cancer incidences for all
the scenarios.  When comparing cancer incidence for the base
control scenarios relative to 1990, there is a 31% reduction in
1995,  a 42% reduction in 2000, and a 33% reduction in 2010 which
is actually an increase when compared to 2000.  The reduction in
emissions are considerably higher, particularly in the out years.
The projected increase in both population and vehicle miles
traveled  (VMT) from 2000 to 2010 appears to offset the gains in
emissions achieved through fuel and vehicles modifications.


                               7-45

-------
                                                          EPA-420-R-93-005
                                                               April 1993
     From  Table 7-11 it can  also be observed that  the expanded
use scenarios  provide little additional reduction  in the cancer
cases.
                                7-46

-------
                                                                                  EPA-420-R-93-005
                                                                                      April 1993
Table 7-11.  Annual Cancer  Incidence Projections for 1,3-Butadiene.
                                                                     a,b
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0156
0.0094
0.0093
0.0071
0.0069
0.0069
0.0067
0.0064
0.0062
Urban
Cancer
Cases
258
177
175
149
145
146
173
164
158
Rural
Cancer
Cases
46
32
32
27
26
26
31
30
28
Total
Cancer
Cases
304
209
207
176
171
172
204
194
186
Percent Reduction
from 1990
EF
-
40
40
54
56
56
57
59
60
Cancer
-
31
32
42
44
43
33
36
39
""Projections have inherent uncertainties in emission  estimates,  dose-response, and
exposure.
bCancer incidence estimates are based on upper bound  estimates of unit risk, determined
from animal studies.   1,3-Butadiene is  classified by EPA as a Group B2, probable  human
carcinogen based on sufficient  evidence in two rodent studies and inadequate  epidemiologic
evidence.
                                            7-47

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Please note that the cancer unit risk estimate for 1,3-
butadiene is based on animal data and is considered an upper
bound estimate for human risk.  True human cancer risk may be as
low as zero.
7.8  Non-Carcinogenic Effects of Inhalation Exposure to 1,3-
Butadiene

     Since the focus of this report is on the carcinogenic
potential of the various compounds, the noncancer information
will be dealt with in a more cursory fashion.  No attempt has
been made to synthesize and analyze the data encompassed below.
Also, no attempt has been made to accord more importance to one
type of noncancer effect over another.  The objective is to
research all existing data, describe the noncancer effects
observed, and refrain from any subjective analysis of the data.

     1,3-Butadiene is used primarily as a monomer in the
production of rubber and plastics  (Chemical and Engineering News,
1986).  It is also found in automobile exhaust  (CARB, 1991).
Although no human data on the metabolism of 1,3-butadiene exist,
animal studies indicate that this chemical is rapidly absorbed
following inhalation (Hattis and Wasson, 1987).  Inhalation of
1,3-butadiene
is mildly toxic in humans at low concentrations  (not otherwise
specified) and may result in a feeling of lethargy and
drowsiness.  At very high concentrations, 1,3-butadiene causes
narcosis leading to respiratory paralysis and death.  The first
signs of toxicity observed in humans are central nervous system
symptoms including blurred vision, nausea, paresthesia  (a sense
of numbness, prickling, or tingling), and dryness of the mouth,
throat,  and nose, followed by fatigue, headache, vertigo,
decreased blood
pressure and pulse rate, and unconsciousness  (Sandmeyer, 1981).
Retrospective epidemiological studies indicate the possibility of
higher than normal mortality rates from cancer and certain
cardiovascular diseases, mainly chronic rheumatic and
arteriosclerotic heart diseases, among middle-aged rubber workers
(McMichael et al.,  1974, 1976).   Workers exposed to unknown
concentrations of 1,3-butadiene during the manufacture of rubber
complained of irritation of the eyes, nasal passages, throat, and
lungs (Wilson, 1944).  An increased rate of emphysema among
rubber workers was reported by McMichael et al.  (1976).  No human
studies on the renal, hepatic, or immunological effects of
inhaled 1,3-butadiene were located in the available literature.

     An LC50 of 129,000 ppm in rats after 4 hours of exposure  and
an LC50 of 122,000 ppm in mice after 2 hours of exposure were
reported  (Shugaev,  1969), indicating that 1,3-butadiene is only
mildly acutely toxic.  After chronic exposure to 1,250 ppm 1,3-
butadiene, mice exhibited respiratory changes such as chronic
inflammation of the nasal cavity, fibrosis, cartilaginous


                              7-48

-------
                                                        EPA-420-R-93-005
                                                            April 1993
metaplasia, osseous metaplasia, and atrophy of the sensory
epithelium (NTP, 1984).   No histopathological cardiovascular
lesions were found in mice following subchronic exposure  (Crouch
et al.,  1979) or rats (Owen et al.,  1987) following chronic
exposure to 1,3-butadiene; however,  NTP  (1984) observed
endothelial hyperplasia in the hearts of mice after 61 weeks of
exposure.  In a chronic study, high incidences of liver necrosis
and epithelial hyperplasia in the forestomach of mice were found
at 625 ppm (LOAEL)  (NTP, 1984), but no nonneoplastic
gastrointestinal lesions were found in rats exposed chronically
(Owen et al., 1987) or mice exposed subchronically (NTP,  1984).
Macrocytic-megaloblastic anemia was observed in mice exposed to
1,250 ppm butadiene for 6-24 weeks  (Irons et al.,  1986a,  1986b).
Bone marrow damage was expressed as reduced numbers of red blood
cells,  decreased hemoglobin concentration and hematocrit, and
increased mean corpuscular volume of circulating erythrocytes.
Decreases in red blood cell counts and hemoglobin concentrations
were reported in male mice after an intermediate duration
exposure of at least 62.5 ppm  (Melnick et al., 1989b).  However,
other studies found no hematological effects in animals following
subchronic and chronic exposure to high exposure concentrations
of 1,3-butadiene (Carpenter et al.,  1944; Crouch et al.,  1979;
Owen et al.,  1987).
     1,3-Butadiene appears to be a developmental toxicant.  When
exposed to concentrations up to 8,000 ppm of 1,3-butadiene during
gestation days 6-15,  depressed body weight gain among dams was
observed at all concentrations, and fetal growth was
significantly decreased in the 8,000 ppm group.   Major skeletal
abnormalities (wavy ribs, irregular rib ossification) were
observed in the 1,000 and 8,000 ppm groups  (Irvine, 1981).  In
studies conducted by NTP  (Morrissey et al.,  1990), pregnant
Sprague Dawley rats exposed to 1,000 ppm 1,3-butadiene by
inhalation on gestation days 6-15 exhibited depressed body weight
gain, but there was no evidence of developmental toxicity in
their offspring.  In contrast, male and female fetuses of mice
similarly exposed exhibited reduced weight at levels of 40 ppm
and higher, and 200 ppm and higher,  respectively.

     Melnick et al. (1990) reported that testicular atrophy was
observed in male B6C3F1 mice exposed to 625 ppm 1,3-butadiene for
65 weeks, and ovarian atrophy was observed in female B6C3F1 mice
exposed to >20 ppm for 65 weeks.  A concentration-related
increase in the incidence of sperm-head abnormalities occurred in
mice after exposure to 1,000 and 5,000 ppm of 1,3-butadiene for 6
hours/day for 5 days  (Hackett et al.,  1988a).  Dominant lethality
(i.e.,  a gene mutation that must only occur in one copy of the
gene to result in death of the offspring) in mice was also
observed during the first 2 postexposure weeks after the males
were exposed to 200,  1,000 or 5,000 ppm  (Hackett et al.,  1988b),
suggesting that more mature cells (spermatozoa and spermatids)
may be altered by 1,3-butadiene exposure.
                               7-49

-------
                                                        EPA-420-R-93-005
                                                             April 1993
     CARB used the two-year inhalation  studies  with  mice  (Huff et
al., 1985; Melnick et al., 1988, 1989a,  1989b;  Miller,  1989)
exposed to 0, 6.25, 20,  62.5, 200, and  625 ppm  1,3-butadiene  to
establish a LOAEL.  These studies were  designed as cancer
bioassays.  Gonadal atrophy was observed at  a high incidence  in
exposed animals of both  sexes at levels of 200  ppm and  above,  but
not in any of the control animals.   In  the later  study, using the
entire dose range, levels of 6.25 ppm and higher  also produced
gonadal atrophy in females.  Thus, a NOAEL was  not established in
these studies, but a LOAEL of 6.25 ppm  was observed.  In
contrast, the Hazelton rat bioassay  (Owen et al.,  1987) did not
report any reproductive  effects even at 8000 ppm  level.

     Neither an inhalation reference concentration  (RfC)  nor  an
oral reference dose (RfD) is available  for 1,3-butadiene  at this
time.
                               7-50

-------
                                                        EPA-420-R-93-005
                                                            April 1993
7.9  References for Chapter 7

Andjelkovich, D., J. Taulbee, and M. Symons.  1976.  Mortality
experience of a cohort of rubber workers, 1964-1973.  J. Occup.
Med.  18:387-394.
Auto/Oil Air Quality Improvement Research Program.
1 Working Data Set  (published in electronic form).
Systems Applications International, San Rafael, CA.
1990.  Phase
Prepared by
Auto/Oil Air Quality Improvement Research Program.  1991.
Technical Bulletin No. 6: Emission Results of Oxygenated Gasoline
and Changes in RVP.

Barnes, I., V. Bastian, K. H. Becker, and Z. Tong.  1990.
Kinetics and products of the reactions of N03 with monoalkenes,
dialkenes, and monoterpenes.   J. Phys. Chem., 94:2413-2419.

Bond, J.A., A.R. Dahl, R.F. Henderson, G.S. Butcher, J.L.
Mauderly, and L.S. Birnbaum.   1986.  Species differences in the
distribution of inhaled butadiene.  Toxicol. Appl. Pharmacol.
84:617-627.

Bond, J.A., A.R. Dahl, R.F. Henderson, and L.S. Birnbaum.  1987.
Species differences in the distribution of inhaled butadiene in
tissues.  Am. Ind. Hyg. Assoc. J. 48:857-862.

Bond, J.A., O.S. Martin, L.S. Birnbaum, A.R. Dahl, R.L. Melnick,
and R.F. Henderson.   1988.  Metabolism of 1,3-butadiene by lung
and liver microsomes of rats and mice repeatedly exposed by
inhalation to 1,3-butadiene.   Toxicol. Lett. 44:143-151.

Bryant, M.S. and S.M. Osterman-Golkar.  1991.  Hemoglobin adducts
as dosimeters of exposure to DNA-reactive chemicals.  CUT
Activities, 11(10) .

CARB.  1991.  Butadiene Emission Factors.  memo from K. D.
Drachand to Terry McGuire and Peter Venturini, July 17, 1991.

CARB.  1992a.  Proposed identification of 1,3-butadiene as a
toxic air contaminant.  Part A Exposure assessment.  California
Air Resources Board, Stationary Source Division.  May 1992.

CARB.  1992b.  Proposed identification of 1,3-butadiene as a
toxic air contaminant.  Part B Health assessment.  California Air
Resources Board, Stationary Source Division.  May 1992.

Carpenter, C.P., C.B. Shaffer, C.S. Weil, and H.F. Smyth, Jr.
1944.  Studies on the inhalation of 1,3-butadiene; with a
comparison of its narcotic effect with benzol, toluol, and
styrene, and a note on the elimination of styrene by the human.
J. Ind. Hyg. Toxicol. 26:69-78.
                               7-51

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Chan, C.C., H. Ozkaynak, J.D. Spengler, L. Sheldon, W. Nelson,
and L. Wallace.  1989.  Commuter's exposure to volatile organic
compounds, ozone, carbon monoxide, and nitrogen dioxide.
Prepared for the Air and Waste Management Association.  AWMA
Paper 89-34A.4.

Checkoway, H. and T.M. Williams.  1982.  A hematology survey of
workers at a styrene-butadiene synthetic rubber manufacturing
plant.  Am. Ind. Hyg.  Assoc. J. 43:164-169.

Chemical and Engineering News.  1986.  Key Chemicals Profile
Butadiene.  June 9, 15.

Choy, W.N., D.A. Vlachos, M.J. Cunningham, G.T. Arce, and A.M.
Sarrif.  1986.  Genotoxicity of 1,3-butadiene.  Induction of bone
marrow micronuclei in B6C3F1 mice and Sprague-Dawley rats in
vivo.  Environ. Mutagen. 8:18.

Clement International Corporation.  1991.  Motor vehicle air
toxics health information.  For U.S. EPA Office of Mobile
Sources, Ann Arbor, MI:  September 1991.

Conner, M., J. Lou, and 0. Gutierrez de Gotera.  1983.  Induction
and rapid repair of sister-chromatid exchanges in multiple murine
tissues in vitro by diepoxybutane.  Mutat. Res. 108:  251-263.

Cote, I.L. and S.P. Bayard.  1990.  Cancer risk assessment of
1,3-butadiene.  Environ. Health Perspect.  86:149-153.

Crouch, C.N., D.H. Pullinger, and I.F. Gaunt.  1979.  Inhalation
toxicity studies with 1,3-butadiene:  2. 3 month toxicity study
in rats.  Am. Ind. Hyg. Assoc. J. 40:796-802.

Csanady, G.A. and J.A. Bond.  1991a.  Species differences in the
biotransformation of 1,3-butadiene to DNA-reactive epoxides:
role in cancer risk assessment.  CUT Activities 11(2) :  1-8.

Csanady, G.A. and J.A. Bond.  1991b.  Species and organ
differences in the metabolic activation of 1,3-butadiene.
Toxicologist 11,47.

Cunningham, M.J., W.N. Choy, G.T. Arce, L.B. Rickard, D.A.
Vlachos, L.A. Kinney,  and A.M. Sarrif.  1986.  In vivo sister
chromatid exchange and micronucleus induction studies with 1,3-
butadiene in B6C3F1 mice and Sprague-Dawley rats.  Mutagenesis
1:449-452.

Cupitt, L. T.  1987.  "Atmospheric persistence of eight air
toxics."  U.S. Environmental Protection Agency (EPA-600/3-
87/004).
                               7-52

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Dankovic, D.A., R.J. Smith, J. Seltzer, A.J. Bailer, and L.T.
Stayner.  1991.  A Quantitative Assessment of the Risk of Cancer
Associated with Exposure to 1,3-Butadiene, Based on a Low Dose
Inhlation Study in B6C3F Mice.  Washington, DC:  U.S. Department
of Health and Human Services, Public Health Service, Centers for
Disease Control, National Institute for Occupational Safety and
Health.  September 27, 1991, Docket #H-041.

Dean, B.J.,  and G. Hodson-Walker.   1979.  An in vitro chromosome
assay using cultured rat-liver cells.  Mutat.  Res. 64:  329-337.

DeJovine, J. M., K. J. McHugh, D.  A. Paulsen,  L. A. Rapp, J. S.
Segal, B. K. Sullivan, D. J. Townsend.  1991.   Clean Fuels Report
91-06:  EC-X Reformulated Gasoline Test Program Emissions Data.
Arco Products Co., Anaheim, California.

de Meester,  C., F. Poncelet, F. Roberfroid and M. Mercier.  1978.
Mutagenicity of butadiene and butadiene monoxide.  Biochem.
Biophys. Res. Commun. 80:  298-305.

de Meester,  C., F. Poncelet, F. Roberfroid and M. Mercier.  1980.
The mutagenicity of butadiene towards Salmonella typhimurium.
Toxicol. Lett. 6:  125-130.

Deutschman,  S., R.J. Laib.  1989.   Concentration-dependent
depletion of non-protein sulfhydryl  (NPSH) content in lung, heart
and liver tissue of rats and mice after acute inhalation exposure
to butadiene.  Toxicol. Lett. 45:175-183.

Divine, B.J.  1990.  An update on mortality among workers at a
butadiene facility - preliminary results.  Environ. Health
Perspect. 86:119-128.

Downs, T.D., M.M. Crane, K.W. Kim.  1987.  Mortality among
workers at a butadiene facility.  Am. J. Ind.  Med. 12:311-330.

Ehrenberg, L., and S. Hussain.  1981.  Genotoxicity of some
important epoxides.  Mutat. Res. 86:  1-113.

Environ Corporation.  1987.  Risk assessment issues in EPA's
technical report "Air Toxics Emissions from Motor Vehicles".
Prepared for the Motor Vehicle Manufacturer Association.

EPA.  1984.   Proposed guidelines for carcinogen risk assessment.
Federal Register  49(227): 46294-46301.

EPA.  1985.   Mutagenicity and carcinogenicity assessment of 1,3-
butadiene.  U.S. Environmental Protection Agency: Washington, DC:
Office of Health and Environmental Assessment.  Publication No.
EPA/600/8-85/004F.
                               7-53

-------
                                                        EPA-420-R-93-005
                                                            April 1993
EPA.  1989.  Locating and estimating air emissions from sources
of 1,3-butadiene.   Office of Air Quality Planning and Standards,
Research Triangle Park, NC.   Report no. EPA-450/2-89-021.

EPA.  1990.  1989 Urban Air Toxics Monitoring Program.  Office of
Air Quality Planning and Standards, Research Triangle Park, NC.
Report no. EPA-450/4-91-001.

EPA.  1991.  Nonroad Engine and Vehicle Emission Study.  Office
of Air and Radiation, Office of Mobile Sources, Ann Arbor, MI.
November 1991.  EPA Report No. 21A-2001.

EPA.  1992.  Integrated Risk Information System.  U.S.
Environmental Protection Agency.  Office of Health and
Environmental Assessment, Environmental Criteria and Assessment
Office,  Cincinnati, OH.

Gervasi, P.G., L.  Citti, M.  Del Monte, V. Longo, and D. Benetti.
1985.   Mutagenicity and chemical reactivity of epoxidic
intermediates of isoprene metabolism and other structurally
related compounds.  Mutat.  Res. 156:77-82.

Grossman, E.A. and J. Martonik.  1990.  OSHA's approach to risk
assessment for setting a revised occupational exposure standard
for 1,3-butadiene.  Environ.  Health Perspect.  86:155-158.

Hackett, P.L., B.J. McClanahan, M.G. Brown, R.L. Buschbom, and
M.L. Clark.  1988a.  Sperm-head morphology study in B6C3F1 mice
following inhalation exposure to 1,3-butadiene:  final technical
report.   Report to National Institute of Environmental Health
Sciences, National Toxicology Program by Pacific Northwest
Laboratory, Richland, WA.  PNL-6459; DE88008620.

Hackett, P.L., T.J. Mast, M.G. Brown, M.L. Clark, and J.J.,
Evanoff.  1988b.  Dominant lethal study in CD-I mice following
inhalation exposure to 1,3-butadiene: final technical report.
ISS PNL-6545; DE88010185.

Hattis.  D. and J.  Wasson.  1987.  Pharmacokinetic/mechanism-based
analysis of the carcinogenic risk of butadiene.  Report to
National Institute for Occupational Safety and Health, Rockville,
MD, by Massachusetts Institute of Technology, Cambridge, MA.
NTIS PB88-202817.

Hazleton Laboratories Europe, Ltd.  (HLE).  1981.  The toxicity
and carcinogenicity of butadiene gas administered to rats by
inhalation for approximately 24 months.  Prepared for the
International Institute of Synthetic Rubber Producers, New York,
NY.  Unpublished.
Hemminki, K.,  E. Franssila, and H. Vianio.  1980.  Spontaneous


                               7-54

-------
                                                        EPA-420-R-93-005
                                                            April 1993
abortions among female chemical workers in Finland.  Int. Arch.
Occup. Environ. Health 45:123-126.

Huff, J.E., R.L. Melnick, H.A. Solleveld, J.K. Haseman, M.
Powers, and R.A. Miller.  1985.  Multiple organ carcinogenicity
of 1,3-butadiene and B6C3F1 mice after 60 weeks of inhalation
exposure.  Science 227:  548-549.

IARC.  1986.  IARC Monographs on the Evaluation of Carcinogenic
Risk of Chemicals to Humans.  Volume 39.  1,3-Butadiene.
International Agency for Research on Cancer.   World Health
Organization, Lyon, France.  p. 156-179.

IARC.  1987.  IARC Monographs on the Evaluation of Carcinogenic
Risk of Chemicals to Humans.  Supplement 7.  Overall Evaluations
of Carcinogenicity:  An updating of IARC monographs volumes 1 to
42.  International Agency for Research on Cancer.  World Health
Organization, Lyon, France.  p. 136-137.

IARC.  1992.  IARC Monographs on the Evaluation of Carcinogenic
Risk to Humans.  Volume 54.  Occupational Exposures to Mists and
Vapours from Strong Inorganic Acids; and Other Industrial
Chemicals.  International Agency for Research on Cancer.  World
Health Organization, Lyon, France.  p. 237-286.

ICF/Clement.  1986.  Rulemaking support for 1,3-butadiene.  Draft
Final Report:  Characterization of risks associated with
occupational exposure to 1,3-butadiene.   Prepared for the Office
of Health Standards, Occupational Safety and Health
Administration, U.S. Department of Labor.  March 10, 1986.

Irons, R.D., C.N. Smith, W.S. Stillman,  R.S.  Shah, W.J.
Steinhagen, and L.J. Leiderman.  1986a.   Macrocytic-megaloblastic
anemia in male B6C3F1 mice following chronic exposure to 1,3-
butadiene.  Toxicol. Appl. Pharmacol.  83:95-100.

Irons, R.D., C.N. Smith, W.S. Stillman,  Shah, W.J. Steinhagen,
and L.J. Leiderman.  1986b.  Macrocytic-megaloblastic anemia in
male NIH Swiss mice following chronic exposure to 1,3-butadiene.
Toxicol. Appl. Pharmacol. 85:450-455.

Irons, R.D., W.S. Stillman, R.S. Shah, M.S. Morris, and M.
Higuchi.  1986c.  Phenotypic characterization of 1,3-butadiene-
induced thymic lymphoma in male B6C3F1 mice.   The Toxicologist
6:27.

Irons, R.D., M. Oshimura, and J.C. Barrett.  1987a.  Chromosome
aberrations in mouse bone marrow cells following in vivo exposure
to 1,3-butadiene.  Carcinogenesis 8:1711-1714.
Irons, R.D., W.S. Stillman, and M.W. Cloyd.  1987b.  Selective


                               7-55

-------
                                                        EPA-420-R-93-005
                                                            April 1993
activation of endogenous esotropic retrovirus in hematopoietic
tissues of B6C3F1 mice during the preleukemic phase of 1,3-
butadiene exposures.  Virology 161:457-462.

Irons, R.D., H.P. Cathro, W.S. Stillman, W.H. Steinhagen, and
R.S. Shah.  1989.  Susceptibility to 1,3-butadiene-induced
leukemogenesis correlates with endogenous esotropic retroviral
background in the mouse.  Toxicol.  Appl. Pharmacol. 101:170-176.

Irvine, L.F.H.  1981.  1,3-Butadiene:   Inhalation  teratogenicity
study in the rat:  final report.  Harrogate, England:  Hazleton
Laboratories Europe Ltd.  OTS Submission, Microfiche No. 050 545.

Jauhar, P.P., P.R. Henika, J.T. MacGregor, C.M. Wehr, M.D.
Shelby, S.A. Murphy, and B.H. Margolin.   1988.  1,3-Butadiene:
induction of micronucleated erythrocytes in the peripheral blood
of B6C3F1 mice exposed by inhalation for 13 weeks.  Mutat. Res.
209:171-176.

Jelitto, B., R.R. Vangala, and R.J. Laib.  1989.   Species-
differences in DNA damage by butadiene:   role of diepoxybutane.
Arch. Toxicol. 13:  246-249.

Kreiling, R., R.J. Laib, and H.M. Bolt.   1986a.  Alkylation of
nuclear proteins and DNA after exposure of rats and mice to  [1,4-
14C]1,3-butadiene.  Toxicol. Lett.  30:131-136.

Kreiling, R., R.J. Laib, J.G. Filser,  and H.M. Bolt.  1986b.
Species differences in butadiene metabolism between mice and rats
evaluated by inhalation pharmacokinetics.  Arch. Toxicol. 58:235-
238.

Kreiling, R., R.J. Laib, J.G. Filser,  and H.M. Bolt.  1987.
Inhalation pharmacokinetics of 1,2-epoxybutene-3 reveal species
differences between rats and mice sensitive to butadiene-induced
carcinogenesis.   Arch. Toxicol. 61:7-11.

Kreiling, R., R.J. Laib, and H.M. Bolt.   1988.  Depletion of
hepatic nonprotein sulfhydryl content during exposure of rats and
mice to butadiene.  Toxicol. Lett.  41:209-214.

Laib, R.J. and R. Kreiling.  1987.   Evaluation of  the reactive
principles responsible for genotoxicity as a prerequisite for
carcinogenic exposure monitoring of the halogenated ethylenes and
butadiene.  Clin. Toxicol. 23:1172.

Lawley, P.O. and P. Brookes.  1967.  Interstrand cross-linking of
DNA by difunctional alkylating agents.  J. Mol. Biol. 25:  143-
160.
                               7-56

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Lemen, R.A., T.J. Meinhardt, M.S. Crandall, J.M. Fagan, and D.P.
Brown.  1990.  Environmental epidemiologic investigations in the
styrene-butadiene rubber production industry.  Environ. Health
Perspect. 86:103-106.

Ligocki, M.P., G.Z. Whitten, R.R. Schulhof, M.C. Causley, and
G.M. Smylie.  1991.  Atmospheric transformation of air toxics:
benzene, 1,3-butadiene, and formaldehyde.  Systems Applications
International, San Rafael, California  (SYSAPP-91/106).

Ligocki, M.P., R.R. Schulhof, R.E. Jackson, M.M. Jimenez, G. Z.
Whitten, G.M. Wilson, T.C. Meyers, and J.L. Fieber.   1992.
Modeling the effects of reformulated gasolines on ozone and
toxics concenetrations in the Baltimore and Houston areas.
Systems Applications International, San Rafael, California
(SYSAPP-92/127).

Luker, M.A.  and B.J. Kilbey.  1982.  A simplified method for the
simultaneous detection of intragenic and intergenic mutations
(deletions)  in Neurospora crassa.  Mutat. Res. 92:  63-68.

Matanoski,  G.M.,  L. Schwartz, J. Sperrazza, and J. Tonascia.
1982.  Mortality of workers in the styrene-butadiene  rubber
polymer manufacturing industry.  Johns Hopkins University School
of Hygiene and Public Health, Baltimore, MD.  Unpublished.

Matanowski,  G.M.  and J. Schwartz.  1987.  Mortality of worker in
styrene-butadiene polymer production.  J. Occup. Med. 29:675-680.

Matanoski,  G.M.,  C. Santos-Burgoa, S.L. Zeger, and L. Schwartz.
1990.  Mortality of a cohort of workers in the styrene-butadiene
polymer manufacturing industry  (1943-1982).  Environ. Health
Perspect. 86:107-117.

McMichael,  A.J.,  R. Spirtas, L.L. Kupper.  1974.  An
epidemiologic study of mortality within a cohort of rubber
workers, 1964-1972.  J. Occup. Med. 16:458-464.

McMichael,  A.J.,  R. Spirtas, J.F. Gamble, and P.M. Tousey.  1976.
Mortality among rubber workers.  Relationship to specific jobs.
J. Occup. Med. 18:178-185.

Meinhardt,  T.J.,  R.A. Lemen, M.S. Crandall, and R.J.  Young.
1982.  Environmental epidemiologic investigation of the styrene-
butadiene rubber industry.  Scand. J. Work Environ. Health 8:250-
259.

Melnick, R.L., J.E. Huff, J.K. Haseman, and E.E. McConnell.
1988.  Chronic toxicity results and ongoing studies of 1,3-
butadiene by the National Toxicology Program.  Ann. New York
Acad. Sci.  534:649-662.
                               7-57

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Melnick, R.L., J. Huff, and R. Miller.  1989a.  Toxicology and
carcinogenicity of 1,3-butadiene.   In:  Mohn U, ed.  Assessment
of inhalation hazards:  Integration and extrapolation using
diverse data.  New York, NY:  Springer-Verlag, 177-188.

Melnick, R.L., J.H. Roycroft, B. Chou, and R. Miller.  1989b.
Carcinogenicity of 1,3-butadiene in mice at low inhalation
exposure concentrations.  Proc.  Am. Ass. Cancer Res. 143.

Melnick, R.L., J.E. Juff,  J.H. Roycroft, B.J. Chou, and R.A.
Miller.  1990.  Inhalation toxicology and carcinogenicity of 1,3-
butadiene in B6C3F1 mice following 65 weeks of exposure.
Environ. Health Perspect.   86:27-36.

Miller, R.A.  1989.  Neoplastic lesions induced by 1,3-butadiene
in B6C3F1 mice.  Assessment of inhalation hazards:  Integration
and extrapolation of diverse data.  Conference at Hannover
Medical School, February 16-24,  1989.  Hannover, Federal Republic
of Germany.

Morrissey,  R.E., B.A.  Schwetz, P.L. Hacket, M.R. Sikov, B.D.
Hardin, B.J. McClanahan, J.R. Decker, and T.J. Mast.  1990.
Overview of reproductive and developmental toxicity studies of
1,3-butadiene in rodents.   Environ. Health Perspect. 86:79-84.

Murray, M.P. and J.E.  Cummins.  1979.  Mutagenic activity of
epoxy embedding reagents employed in electron microscopy.
Environ. Mutagen. 1:  307-313.

NTP.  1984.  NTP technical report on the toxicology and
carcinogenesis studies of 1,3-butadiene in B6C3F1 mice.  National
Toxicology Program.  Research Triangle Park, NC.  NTP TR 288.
NTP-83-071.  NIH Pub.  No.  84-2544.

NTP. 1985.   Draft report on the toxicology and carcinogenesis
studies of 4-vinylcyclohexane in F433/N rats and B6C3F1 mice.
NIH Publication NO. 85-2559.

Olszewska,  E. and B.J. Kilbey.  1975.  The mutagenic activity of
diepoxybutane in yeast.  Mutat.  Res. 33:  383-390.

Owen,  P.E., J.R. Glaister, I.F.  Gaunt, and D.H. Pullinger.  1987.
Inhalation toxicity studies with 1,3-butadiene.  3.  2-year
toxicity/carcinogenicity study in rats.  Am. Ind. Hyg. Assoc. J.
48:407-413 .

Perry,  P. and H.J. Evans.   1975.  Cytological detection of
mutagen-carcinogen exposure by sister chromatid exchange.
Nature. 258:  121-125.
Platt, U., D. Perner, A. M. Winer, G. W. Harris, and J. N. Pitts,


                               7-58

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Jr.  1980.  Detection of N03 in the polluted troposphere by
differential optical absorption.  Geophys. Res. Lett.,  7:89-92.

Poncelet, F. C. de meester, M. Duverger-van Bogaert, M. Lambotte-
Vandepaer, M. Roberfroid, and M. Mercier.  1980.  Influence of
experimental factors on the mutagenicity of vinylic monomers.
Arch. Toxicol.  Suppl.  4:  63-66.

Recio,  L., 0. Moss, J. Bond, and G.A. Csanady.  1991.
Biotransfromation of butadiene by hepatic and pulmonary tissues
from mice, rats, and humans:  relationship to butadiene
carcinogenicity and in vivo mutagenicity.  Proceedings  from the
Air and Waste Management Association's specialty conference "Air
Toxics Pollutants from Mobile Sources:  Emissions and Health
Effects."  October 16-18, 1991.

Sandmeyer, E.E.  1981.  Aliphalic hydrocarbons:  Butadiene.  In:
Clayton GD, Clayton FE, eds.  Patty's industrial hygiene and
toxicology, 3rd revised ed.  New York, NY:  John Wiley  and Sons.

Sankaranarayanan, K.  1983.  The effects of butylated
hydroxytoluene on radiation and chemically-induced genetic damage
in Drosophila melanogaster.  Mutat. Res. 108:  203-223.

Sankaranarayanan, K.,  W. Ferro, and J.A. Ziljlstra.  1983.
Studies on mutagen-sensitive strains on Drosophila melanogaster.
III. A comparison of the mutagenic sensitivities of the ebony  (UV
and X-ray sensitive) and Canton-S  (wild-type) strains to MMS,
ENU, DEB, and 2,3,6-Cl3-PDMT.   Mutat.  Res.  100:   59-70.

Santos-Burgoa,  C.  1988.  Dissertation submitted to the Johns
Hopkins University, Baltimore, Maryland, August 1988.

Schmidt, U. and E.  Loeser.  1985.  Species differences  in the
formation of butadiene monoxide from 1,3-butadiene.  Arch.
Toxicol. 57:222-225.

Schmidt, U. and E.  Loeser.  1986.  Epoxidation of 1,3-butadiene
in liver and lung tissue of mouse, rat,  monkey and man.  Adv.
Exp. Med. Biol. 197:951-957.

Sharief, Y., A.M. Brown, L.C. Backer, J.A. Campbell, B.
Westbrook-Collins,  A.G. Stead, and J.W.  Allen.  1986.   Sister
chromatid exchange and chromosome aberration analyses in mice
after in vivo exposure to acrylonitrile, styrene, or butadiene
monoxide.  Environ. Mutagen. 8:439-448.

Shugaev, B.B.  1969.  Concentrations of hydrocarbons in tissues
as a measure of toxicity.  Arch. Environ. Health 18:878-882.


Simmon, V.F. and J.M.  Baden.  1980.  Mutagenic activity of vinyl
compounds and derived epoxides.  Mutat.  Res. 78:  227-231.


                               7-59

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Tice, R.R., R. Boucher, C.A. Luke, and M.D. Shelby.  1987.
Comparative cytogenetic analysis of bone marrow damage induced in
male B6C3F1 mice by multiple exposure to gaseous 1,3-butadiene.
Environ. Mutagen. 9:  235-250.

Truchi, G., S. Bonatti, L. Citti, P.G. Gervasi, and A.
Abbondandolo.   1981.  Alkylating properties and genetic activity
of 4-vinylcyclohexane metabolites and structurally related
epoxides.  Mutat. Res. 83:  419-430.

Turnbull, D.,  J.V. Rodricks, S.M. Brett.  1990.  Assessment of
the potential risk to workers from exposure to 1,3-butadiene.
Environ. Health. Perspect.  86: 159-171.

Voogd, C.E., J.J. van de Stel, and J.A. Jacobs.  1981.  The
mutagenic action of aliphatic epoxides.  Mutat. Res. 89:  269-
282.
Wade, M.J., J.W. Moyer, and C.H. Hine.
of the series of epoxides.  Mutat. Res.
1979.   Mutagenic action
66:   367-371.
Ward, J.B. Jr., M.M. Ammenheuser, E.B. Whorton, Jr., and M.S.
Legator.  1992.  hprt Mutant Lymphocyte Frequencies in Workers at
a Butadiene Production Plant.  Abstract presented at the Fourth
European International Society for the Study of Xenobiotics
(ISSX),   Bologna, Italy, July 3-6, 1992.

Warner-Selph, M. A., and L. R. Smith.  1991.  Assessment of
Unregulated Emissions from Gasoline Oxygenated Blends.  Ann
Arbor, Michigan: U.S. Environmental Protection Agency, Office of
Mobile Sources.  Publication no. EPA-460/3-91-002.

Wilson,  R.H.  1944.  Health hazards encountered in the
manufacture of synthetic rubber.  J. Am. Med. Assoc. 124:701-703.

Zaborowski, D., Z. Swietlinska, and J. Zuk.  1983.  Induction of
mitotic recombination by UV and diepoxybutane and its enhancement
by hydroxyurea in Saccharomyces cerevisiae.  Mutat. Res. 120:
21-26.

Zhou, X.T., L.R. Li, M.Y. Cui, et al.  1986.  Cytogenetic
monitoring of petrochemical workers  [Abstract].  Environ.
Mutagen. 8:96.

Zimmering, S.  1983.  The mei-ga test for chromosome loss in
Drosophila:  a review of assays of 21 chemicals for chromosome
breakage.  Environ. Mutagen. 5:  907-921.
                               7-60

-------
                                                        EPA-420-R-93-005
                                                            April 1993

8.0  ACETALDEHYDE

8.1 Chemical and Physical Properties

     The information below is excerpted from the EPA health
assessment draft document (EPA, 1987) and Perry and Chilton,  1973.

     Acetaldehyde is a saturated aldehyde with a pungent and
suffocating odor, but at more dilute concentrations the odor  is
fruity and pleasant.  It has the chemical formula CH3CHO.   It is a
colorless liquid, volatile at room temperature, and both the  liquid
and the vapors are highly flammable.  Acetaldehyde as a liquid is
lighter than water, and the vapors are heavier than air.   It  is
soluble in water, alcohol, ether, acetone, and benzene.  The
chemical and physical properties are listed in Table 8-1.

     As the vapor pressure of acetaldehyde is very high and it is
soluble in water, the most important environmental behavior will be
in air and water.  This is due to vaporization from the soil  (and
other sources) into the air and leaching from soil into the water.
Acetaldehyde may remain bound in the soil because of its high
reactivity, but it is also readily metabolized by soil
microorganisms.

     Acetaldehyde is a component of photochemical smog, and as such
its movement within the atmosphere corresponds to that of  the smog
front.  The high solubility of acetaldehyde in water increases the
likelihood of its being leached into the soil.

     In the atmosphere, acetaldehyde would be degraded through
photooxidation and oxidation by the hydroxyl radical.  The main
product of photooxidation in the presence of NOX is peroxyacetyl
nitrate.

Table 8-1.  Chemical and Physical Properties of Acetaldehyde.
Properties
Molecular weight
Melting point
Boiling point
Density at 18C (64.4F)
Vapor pressure at 20C (68F)
Flash point (closed cup)
Solubility in water at 25C
Conversion at 25C (77F)
and 760 mm Hg
Value
44.06 g/mole
-123. 5C (-190. 3F)
20.16C (68.3F)
0.783 g/ml
0.97 atm.
-38.0C (-36.4F)
infinite
1 ppm = 1.8 mg/m3
                                 1-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993
8.2 Formation and Control Technology
     Acetaldehyde is another aldehyde which is found in vehicle
exhaust and is formed as a result of incomplete combustion of the
fuel.  Acetaldehyde is emitted in the exhaust of both gasoline and
diesel-fueled vehicles.  It is not a component of evaporative
emissions.

     Use of a catalyst has been found to be effective for
controlling formaldehyde and other aldehyde emissions.
Acetaldehyde emissions are presumed to be controlled to roughly the
same extent as total hydrocarbon emissions with a catalyst.


8.3  Emissions

8.3.1  Emission Fractions Used in the MOBTOX Emissions Model

     Like 1,3-butadiene and formaldehyde, emission fractions for
acetaldehyde were developed using vehicle emissions data  (Appendix
B2).   Acetaldehyde emission fractions for different components
included in the scenarios are included in Appendix B6.  Emission
fractions for the various vehicle class/catalyst technology groups
were based on the same number of cars and studies as the
formaldehyde emission fractions.

     To calculate TOG fractions for vehicles running on MTBE blends
and 10% ethanol,  adjustment factors were applied to the baseline
emission fractions for each vehicle class/catalyst combination, in
the same manner as was done for 1,3-butadiene and formaldehyde.
The average percent change numbers for vehicle class/catalyst
combinations by study are contained in Appendix B4.   The 15% MTBE
and 10% ethanol adjustment factors for LDGVs/LDGTs with various
catalyst technologies are summarized in Table 8-2.  Note that use
of oxygenated fuels increases acetaldehyde emissions for all
catalyst technologies, and that acetaldehyde increases more than
200% for all catalyst technologies with 10% ethanol use.

     These 15% MTBE and 10% ethanol numbers were estimated using
data from the same studies as formaldehyde.  Once again, since the
average percent change was calculated for 15% MTBE (2.7% weight
percent oxygen),  and 11.0% MTBE (2.0% oxygen) was assumed for
reformulated fuel and California standards components, average
percent changes in the formaldehyde TOG fraction from 0 to 15% MTBE
were multiplied by 2.0/2.7.  For HDGVs with three-way catalysts and
with no catalysts, we assumed the same 15% MTBE and 10% ethanol
adjustment factors as for LDGVs/LDGTs with the same catalyst
technologies.
 Table 8-2.  15% MTBE and 10% Ethanol Emission Fraction Adjustment
                     Factors for Acetaldehyde.
                                 1-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Vehicle
Class
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
LDGV/LDGT
Catalyst
Technology
3 -way
3 -way + ox
oxidation
non-cat
15% MTBE
Adjustment
Factor
1.0826
1.0136
1.2114
1.4377
10% EtOH
Adjustment
Factor
2.1369
2.2453
2.9609
2 .1445
8.3.2  Emission Factors for Baseline and Control Scenarios

     The fleet average acetaldehyde emission factors as determined
by the MOBTOX emissions model are presented in Table 8-3.  When
comparing the base control scenarios relative to 1990, the emission
factor is reduced by 40% in 1995, by 57% in 2000, and by 62% in
2010.  The expansion of reformulated fuel use in 1995 actually has
no net impact on the emission factor.  In 2000, the expansion of
reformulated fuel usage also has no net impact on the emission
factor, whereas the expanded California standard scenario increases
the emission factor by 1%, relative to 1990.  In 2010, there is a
decrease from the 2010 base control for the reformulated fuels
scenario of 1% and the California standards scenario of 4%.

8.3.3  Nationwide Motor Vehicle Acetaldehyde Emissions

     The nationwide acetaldehyde metric tons are presented in Table
8-4.  Total metric tons are determined by multiplying the emission
factor (g/mile) by the VMT determined for the particular year.  The
VMT, in billion miles, was determined to be 1793.07 for 1990,
2029.74 for 1995, 2269.25 for 2000, and 2771.30 for 2010.  When
comparing the base control scenarios relative to 1990, the metric
tons are reduced by 32% in 1995, by 46% in 2000, and by 42% in
2010, which is actually an increase when compared to 2000.

8.3.4  Other Sources of Acetaldehyde

     The motor vehicle contribution to ambient acetaldehyde levels
contains both direct  (primary) and secondary acetaldehyde formed
from photooxidation of VOC, though the rate of photooxidation is
much less than that of formaldehyde.  It appears that roughly 39%
of acetaldehyde emissions may be attributable to motor vehicles.
Section 8.5.2 contains a complete explanation of how this number is
determined.
                                 1-3

-------
                                                             EPA-420-R-93-005
                                                                 April 1993

Table  8-3.   Annual  Emission  Factor Projections for  Acetaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0119
0.0071
0.0071
0.0051
0.0051
0.0052
0.0045
0.0044
0.0041
Percent
Reduction
from 1990
-
40
40
57
57
56
62
63
66

-------
                                                             EPA-420-R-93-005
                                                                  April 1993

Table  8-4.   Nationwide Metric Tons Projection for  Acetaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0119
0.0071
0.0071
0.0051
0.0051
0.0052
0.0045
0.0044
0.0041
Metric
Tons
21,338
14,411
14,411
11,573
11,573
11,800
12,471
12, 194
11,362
                                    1-5

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Acetaldehyde is ubiquitous in the environment and is naturally
released.  It is a metabolic intermediate of higher plant
respiration and alcohol fermentation. It is also found in many
flowers, herbs,  and fruits and could be available for release to
the ambient air.  Acetaldehyde is also produced from aliphatic and
aromatic hydrocarbon photooxidation reactions.

     Acetaldehyde is formed as a product of incomplete wood
combustion in residential fireplaces and woodstoves and is released
into the atmosphere by the coffee roasting process.  Together these
two processes accounted for 78% of the national acetaldehyde
emissions  (Eimitus et al.,  1978).  Acetaldehyde is also released
through the burning of tobacco (Braven et al., 1967), the
combustion of organic fuels, coal refining, and coal waste
processing (Versar Inc., 1975), and also as a product of plastics
combustion (Boettner et al. , 1973).

     Manufacturing plants that produce acetaldehyde also emit
acetaldehyde, as do manufacturing plants that produce ethanol,
phenol, acrylonitrile, and acetone (Eimitus et al., 1978;
Mannsville Chemical Products Corp., 1984; Delaney and Hughs, 1979).
Chemical processes that involve acetaldehyde as an intermediate
also emit acetaldehyde.  This includes the production of peracetic
acid, pentaerythritol, pyridine,  terephthalic acid, 1,3-butylene
glycol, and crotonaldehyde.


8.4 Atmospheric Reactivity and Residence Times

     The processes involved in transformation and residence times
were previously discussed in Section 5.4.  For a more detailed
explanation of the various parameters involved in these processes
please refer to Section 5.4.  The information that follows on
transformation and residence times has been mainly excerpted from a
report produced by Systems Applications International for the EPA
(Ligocki and Whitten, 1991).

8.4.1 Gas Phase Chemistry of Acetaldehyde

     The atmospheric transformation chemistry of acetaldehyde
(CH3CHO)  is similar in many  respects  to  that of  formaldehyde.   Like
formaldehyde, it can be both produced and destroyed by atmospheric
chemical transformation.  The reaction rate of acetaldehyde with OH
is in fact about the same as formaldehyde.  However, there are
important differences between the two.  Acetaldehyde photolyses,
but much more slowly than formaldehyde.   Whereas formaldehyde
produces CO upon reaction or photolysis, acetaldehyde produces
organic radicals that ultimately form peroxyacetyl nitrate  (PAN)
and formaldehyde.
8.4.1.1 Formation

     Acetaldehyde is formed from the atmospheric oxidation of many
types of organic compounds.  Unlike formaldehyde, acetaldehyde is


                                8-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993

not produced in the atmospheric oxidations of methane and isoprene,
but may be produced in the atmospheric oxidation of other naturally
occurring organic compounds such as terpenes.  In urban areas, the
oxidation of olefins such as propene  (C3H6),  and paraffins such as
propane (C3H8) and ethanol  (C2H5OH)  produces acetaldehyde.

     Paraffins  (also termed alkanes and saturated hydrocarbons) are
organic compounds containing only  single-bonded carbon.  Paraffins
are generally present in urban atmospheres in high concentrations,
but react relatively slowly.  The  pathways by which paraffins are
converted to aldehydes such as formaldehyde  and acetaldehyde have
been summarized by the National Research Council  (NRC, 1981).
Briefly, the process is initiated  by the reaction of a paraffin
(such as propane) with OH.  This reaction proceeds forming an
organic radical that rapidly reacts with atmospheric 02 to form an
organic peroxy radical (often represented as R02) .   In urban
atmospheres, these R02  radicals  typically react with NO,  forming
N02 and fueling  the  photochemical  ozone production cycle.   The
organic intermediate formed in these reactions rapidly produces
aldehydes.  The specific aldehydes formed in a given reaction
depend upon the initial chain length of the paraffin and the
position along the chain at which  the initial OH attack occurred.
It can easily be seen that a whole family of aldehydes could be
produced in varying yields in the  oxidation of a single compound.

     Olefins (also termed alkenes  and unsaturated hydrocarbons) are
species containing one or more double bonds.  Both OH and 03 react
rapidly with olefins by addition to these reactive double bonds,
again forming radical intermediates that decay through a variety of
pathways to form aldehydes.  In these cases, the particular
aldehyde produced will depend upon the location of the double bond.

8.4.1.2 Gas Phase Reactions

     Acetaldehyde reacts more rapidly than formaldehyde with the OH
and N03 radicals.  Acetaldehyde  reactions with the H02, oxygen
atoms,  03,  and  Cl radicals are  not important to the atmospheric
chemistry of acetaldehyde due to low concentrations in the
atmosphere and/or low to negligible reaction rates.

     Acetaldehyde absorbs ultraviolet  (UV) radiation from
wavelengths below 290 nanometers  (nm) to about 345 nm.  Although
there are three possible pathways  for acetaldehyde photolysis, only
one is important at wavelengths >290 nm.

     The resulting photolysis rate is less than 10 percent of the
formaldehyde photolysis rate.  Therefore, photolysis is a
relatively minor atmospheric transformation pathway for
acetaldehyde.

8.4.1.3 Reaction Products

     The oxidation of acetaldehyde by OH, oxygen atoms, and N03
radicals form a CH3CO radical that rapidly reacts with atmospheric
02 to  form the  peroxyacetyl radical,  CH3C(0)00.  This radical can
then react with atmospheric NO and N02.   The reaction with N02


                                8-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993

produces peroxyacetyl nitrate  (PAN),  whereas the reaction with NO
ultimately produces formaldehyde.  Minor products of the
peroxyacetyl radical reactions are peroxyacetic acid and acetic
acid.  Although acetaldehyde is a PAN precursor, methylglyoxal and
other species derived from the oxidation of aromatic compounds are
more important PAN precursors in urban atmospheres than
acetaldehyde.  The photolysis of acetaldehyde produces the CH302
radical, which reacts with NO to form formaldehyde.  Thus, the
major acetaldehyde decomposition products are formaldehyde and PAN,
both of which are of concern as toxic and/or irritant species.
However, in neither case is acetaldehyde a dominant source of these
species.

8.4.2 Aqueous Phase Chemistry of Acetaldehyde

     Acetaldehyde is slightly soluble, and will be incorporated
into clouds and rain, but to a much lesser degree than
formaldehyde.  The rate of the acetaldehyde-OH reaction is roughly
two-thirds of the aqueous formaldehyde reaction rate.  The product
of the aqueous phase oxidation of acetaldehyde is expected to be
acetic acid  (Jacob et al.,  1989).

     Acetaldehyde, like formaldehyde, can participate in sulfur
chemistry within clouds.  Aqueous acetaldehyde combines with
aqueous S02 to form the  stable  adduct 1-hydroxy-l-ethanesulfonate
(HES) (Olson and Hoffmann,  1989).  However, this species does not
appear to be of major importance in cloud chemistry.

8.4.3 Acetaldehyde Residence Times

     Residence times for acetaldehyde were calculated by
considering gas-phase chemical reactions with OH and N03
photolysis, in-cloud chemical reaction with OH, and wet and dry
deposition.  The results of the residence time calculation for
acetaldehyde are presented in Table 8-5.  During the daytime, under
clear-sky conditions, the residence time of acetaldehyde is
determined primarily by its reaction with OH, with photolysis
accounting for only 2 to 5 percent of the removal.  Calculated
residence times under these conditions were on the order of a few
hours.  The National Research Council (NRC, 1981) did not estimate
an atmospheric residence time for acetaldehyde, but stated that it
would be comparable to the half-life of formaldehyde (2.6 hours,
corresponding to a residence time of 3.8 hours).  The residence

-------


TABLE 8-5. Atmospheric residence time calculation
noted.


Clear sky - day
Clear sky - night
Clear sky - avg
Cloudy - day
Cloudy - night
Cloudy - avg
Rainy - day
Rainy - night
Rainy - avg
Monthly Climatological
Average
Los Angeles
July Jan
4 20
18 700
6 50
8 50
150 1800
14 130
50
400
110
7 70

for acetaldehyde.
St. Louis
July Jan
3 30
170 3000
4 80
6 80
300 3000
9 190
6 60
300 150
9 90
6 130

All times are in hours
Atlanta
July Jan
3 30
21 300
4 70
6 80
150 2000
10 180
6 70
130 200
10 120
6 110

unless
EPA-420-R-93-005
April 1993
otherwise
New York
July
5
40
7
11
300
17
11
200
17
11
Jan
60
3000
160
140
3000
400
100
200
140
200
*Not  calculated  since  July  rainfall  is  zero  for Los Angeles.
                                                        1-9

-------
                                                        EPA-420-R-93-005
                                                            April 1993

times presented in Table 8-5 are somewhat longer that those
calculated for formaldehyde (Ligocki et al.,  1991) because of the
slower photolysis rate for acetaldehyde.

     In contrast to the situation for formaldehyde, neither in-
cloud oxidation nor wet deposition is important for acetaldehyde.
In-cloud oxidation accounted for only 1 percent or less of the
atmospheric removal of acetaldehyde, compared to 10 to 25 percent
of the daytime chemical destruction of formaldehyde and 20 to 90
percent of the nighttime chemical destruction of formaldehyde.  The
presence of clouds would also be expected to decrease the formation
rate of acetaldehyde; thus, cloud cover may actually decrease
acetaldehyde concentrations despite the predicted increase in
residence time.

     At night, for Los Angeles, Atlanta,  and New York, the reaction
of acetaldehyde with N03  leads  to residence times  on the order of
tens of hours during the summertime.  However, because of the low
N03  concentration predicted for St.  Louis, the loss of acetaldehyde
by reaction with N03  is only comparable to the loss by reaction
with OH, and neither is rapid.

     Dry deposition is not an important removal mechanism for
acetaldehyde.  Residence times due to dry deposition were estimated
to range from 20 days under summer,  daytime conditions to over a
year for the other conditions.   For the cases considered here, dry
deposition was a minor removal mechanism except under winter,
nighttime conditions.  Under these conditions, dry deposition is
slow, but all other processes are slower.

     The differences in acetaldehyde residence time among cities
within a season were not as large as the difference between
seasons.  The calculated summer residence times are short in most
cases, whereas the winter residence times are on the order of days.
Thus, acetaldehyde must be considered to be persistent in
wintertime.  Like formaldehyde, however,  the effect of this longer
winter residence time is difficult to assess for acetaldehyde
because of the importance of secondary formation.   Rates of
formation of acetaldehyde will be roughly an order of magnitude
slower in the wintertime.  Thus, it is difficult to predict whether
ambient concentrations of acetaldehyde will increase or decrease in
winter.

     The major uncertainty in the residence time calculation for
acetaldehyde is the OH radical concentration, which varies from day
to day by roughly a factor of two.  The uncertainty in the OH rate
constant is much smaller than this  (about 13  percent).  The
uncertainties associated with the photolysis rate, N03
concentrations, the rate constant, and dry deposition velocity are
of minor importance for acetaldehyde because these processes are
relatively slow.


8.4.4 Limited Urban Airshed Modeling Results for Acetaldehyde
                                8-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The Urban Airshed Model  (UAM) has been previously discussed in
Section 5.4.  Please refer to this section for details about the
model, its inputs, and modifications.  Much of the information
below has been excerpted from reports conducted for EPA by Systems
Applications International (SAI)  (Ligocki et al.,  1992, Ligocki and
Whitten, 1991) .

St. Louis Study

     The Carbon Bond Mechanism-IV chemical mechanism in the UAM
uses the "lumped structure approach".  In this approach individual
chemical species are broken up into reactive units based on the
type of bonds and functional groups present in the molecule.  In
this model, the species ALD2 represents acetaldehyde, the aldehyde
functional group of higher aldehydes, and olefins containing
internal double bonds which react rapidly in the atmosphere to
produce aldehydes.  These are the primary ALD2 aldehydes.

     Secondary ALD2 is produced through the reactions of paraffins
(hydrocarbons with single carbon bonds),  olefins (hydrocarbons with
double carbon bonds), and other species.   A large number of
aldehydes of varying size can be produced by the oxidation of a
single hydrocarbon.

     The magnitude of the changes required to model acetaldehyde
explicitly, specifically secondary acetaldehyde, placed this beyond
the scope of the St. Louis study.  Instead, the results of the St.
Louis air toxics simulations presented previously (Ligocki et al.,
1991) were re-examined in terms of the ALD2 concentrations.

     Results are presented as time-series plots of predicted hourly
average ALD2 concentrations and include curves from both the
reactive and inert simulations.  The results from the base-case
simulations are shown in Figure D-4 and Figure D-5 in Appendix D
for two representative urban grid cells.   The grid cell represented
in Figure D-4 is located near the area of maximum mobile-source
emissions, and thus represents the area with maximum primary ALD2
impact.  The grid cell represented in Figure D-5 is located 8 km
downwind, and represents an area where secondary ALD2 production is
maximized.

     In near-source areas of the modeling domain,  ALD2 behaved as a
primary species, with concentration peaks in the early morning and
early evening (Figure D-4).   In downwind areas, however, ALD2
behaved as a secondary species, with concentration peaks in the
midafternoon (Figure D-5).  The simulation also suggested that
motor vehicles may be a more important contributor to ambient
acetaldehyde levels than they are to formaldehyde levels.

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The dominance of primary ALD2 shown in Figure D-4 for the
morning commute hours, combined with the 68 percent contribution of
motor vehicle of ALD2, suggest that a large fraction of the
simulated ALD2 is attributable to motor vehicles.  However, a
smaller fraction of the secondary ALD2 is attributable to motor
vehicles,  because the mobile contribution to the major ALD2
precursor emissions is smaller, particularly in the afternoon when
the main ALD2 secondary production occurs.

     Simulated ALD2 concentrations were three times as high as
measured acetaldehyde concentrations.  Because simulated
formaldehyde concentrations agreed well with measured
concentrations, it is likely that this discrepancy for acetaldehyde
is due to the inclusion of higher aldehydes in the ALD2 composite
species.  If this is the case, urban ambient concentrations of
higher aldehydes may be comparable to those of formaldehyde and
acetaldehyde.

     The results from the day-to-day carryover sensitivity
simulations, with the exception of the first few hours of the
simulation,  are comparable to the base-case results.  The peak
concentrations were not affected by the change in initial
concentrations.

Baltimore-Washington and Houston Area Simulations

     For the Baltimore-Washington and Houston area simulations,
primary and secondary acetaldehyde were modeled explicitly.  The
modifications made to UAM to model this species explicitly are
described in Ligocki et al.  (1992).

     Simulations for the summer Baltimore-Washington area episode
resulted in decreases in ambient acetaldehyde with the use of
reformulated gasoline, with little change in primary acetaldehyde
and decreased secondary acetaldehyde throughout the domain.  Use of
California reformulated gasoline resulted in a decrease in
secondary acetaldehyde roughly twice as large as in federal
reformulated gasoline scenarios.  Maximum daily average
acetaldehyde for the 1988 base scenario was 6.1 ppb.  Motor
vehicle-related acetaldehyde accounted for about 36% of total
acetaldehyde emissions, based on the 1995 no motor vehicle
scenario.   Motor vehicle-related acetaldehyde also accounted for
about 15% of total simulated ambient acetaldehyde.  90 to 95% of
this acetaldehyde was secondary.

     Summer Baltimore-Washington area simulations appear to
somewhat overpredict the measured data.  Since most of the
simulated acetaldehyde is secondary, the concentrations are very

-------
                                                        EPA-420-R-93-005
                                                            April 1993

sensitive to the product distribution between acetaldehyde and
other higher aldehydes in the chemical mechanism.

     In the winter 1988 base scenario, the maximum daily average
acetaldehyde concentration was 5.2 ppb, slightly lower than in
summer.  Simulations for the winter Baltimore-Washington area
episode resulted in very small decreases in primary and secondary
acetaldehyde.  Motor vehicle-related acetaldehyde emissions were
about the same with reformulated gasoline use.  Motor vehicle-
related acetaldehyde accounted for about 13% of the maximum
simulated concentration, based on the 1995 no motor vehicle
scenario.

     For the summer 1987 base scenario in Houston, the maximum
daily average acetaldehyde concentration was 18.2 ppb.  Motor
vehicle-related acetaldehyde accounted for about 13% of total
acetaldehyde emissions and 18% of the maximum simulated
concentration for the 1987 base scenario.  Simulations for the
summer Houston episode predicted slight decreases in simulated
daily average concentration throughout most of the domain with use
of reformulated gasoline.  Simulated concentrations of acetaldehyde
were in good agreement with measured concentrations.

8.5  Exposure Estimation

8.5.1  Annual Average Exposures Using HAPEM-MS

     The data presented in Table 8-6 represent the results
determined by the HAPEM-MS modeling that was described previously
in Section 4.1.1.  These numbers have been adjusted to represent
the increase in VMT expected in future years.

     The HAPEM-MS exposure estimates in Table 8-6 represent the
50th percentiles of the population distributions of exposure, i.e.,
half the population will be above and half below these values.
High end exposures can also be estimated by using the 95th
percentile of the distributions.  According to the HAPEM-MS sample
output for benzene, the 95th percentile is 1.8 times higher than
the 50th percentile for urban areas, and 1.2 times high for rural
areas.  Applying these factors to the exposure estimates in Table
8-6, the 95th percentiles for urban areas range from 0.32 ug/m3 for
the 2000 expanded reformulated fuel use and the 2010 expansion of
the California standards scenarios, to 0.65 ug/m3 for the 1990 base
control scenario.  The 95th percentiles for rural areas range from
0.11 to 0.24 ug/m3,  respectively.

8.5.2  Comparison of HAPEM-MS Exposures to Ambient Monitoring Data

     As stated in section 4.1.2, four national air monitoring
programs/databases contain data on air toxics and the data for
                                1-13

-------
Table  8-6.
                                            EPA-420-R-93-005
                                                April 1993

Annual  Average HAPEM-MS Exposure  Projections for
Acetaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Urban Exposure
ug/m3
0.36
0.24
0.24
0.19
0.18
0.19
0.19
0.19
0.18
Rural Exposure
ug/m3
0.20
0.13
0.13
0.10
0.10
0.10
0.10
0.10
0.09

-------
                                                        EPA-420-R-93-005
                                                            April 1993

acetaldehyde is found in only two.  The Aerometric Information
Retrieval System (AIRS),  and the Urban Air Toxic Monitoring Program
(UATMP) have data for acetaldehyde.  The urban exposure data for
acetaldehyde from the two databases are summarized in Table 8-7.

     In the 1990 Urban Air Toxics Monitoring Program  (UATMP), 332
measurements of acetaldehyde were taken at 12 sites.  These sites
were in the cities listed below.

          Baton Rouge, LA               Chicago, IL
          Camden, NJ                    Houston, TX
          Orlando,  FL                   Pensacola, FL
          Port Neches, TX               Sauget, IL
          Toledo, OH                    Washington, B.C.
          Wichita,  KS

The highest average was 4.48 ug/m3 (2.49  ppb)  at an urban
commercial site in Baton Rouge, Louisiana.  Twenty-two samples were
collected at this site.   The lowest average was 1.34 ug/m3 (0.75
ppb) at a suburban residential site in Houston, Texas.  Twenty-
three samples were collected at this site.  The overall average of
the averages for each site was 3.10 ug/m3 (1.72 ppb).   Ozone was
removed from the ambient air collected in this program through the
use of an ozone denuder.   The use of an ozone denuder in the
sampling system resulted in higher, but more accurate, reported
acetaldehyde concentrations.  Only the 1990 UATMP data will be used
for the comparisons in this study.

     HAPEM-MS assumes that the dispersion and atmospheric chemistry
of acetaldehyde is similar to CO.  This assumption would appear to
be somewhat valid for acetaldehyde since it is less reactive than
formaldehyde, but acetaldehyde is transformed in the atmosphere to
some extent.  To test the reasonableness of the HAPEM-MS modeling
results, the HAPEM-MS results for 1990 are compared to ambient
monitoring results for recent years.  Before comparing the HAPEM-MS
results to the ambient data, the ambient monitoring data must be
adjusted to represent the amount that is attributed to motor
vehicles.  The data derived from emission inventories and
atmospheric modeling conducted by SAI for St. Louis (Ligocki and
Whitten, 1991) estimate that 39% of the ambient acetaldehyde can be
apportioned to motor vehicles.  This number actually represents
acetaldehyde and higher aldehydes.

     This estimate is higher than the estimates in the Houston and
Baltimore-Washington Area UAM-Tox simulations  (Ligocki et al.,
1992).   In these studies, acetaldehyde was modeled explicitly;
thus, the estimates do not represent both acetaldehyde and higher
aldehydes.  In Baltimore-Washington, motor vehicle-related
acetaldehyde accounted for about 15% of total simulated ambient
acetaldehyde in summer,  while in Houston, motor vehicle-related
acetaldehyde accounted for about 18% of the maximum simulated
concentration.

     The estimate of 39% will be used in this study to represent
the nationwide average percentage of ambient acetaldehyde
attributable to motor vehicles, while acknowledging the apparent


                                8-15

-------
                                                        EPA-420-R-93-005
                                                            April 1993

area-to-area variations and the possibility that this may
overestimate the motor vehicle contribution to ambient
acetaldehyde, possibly in part because the estimate actually
represents both acetaldehyde and higher aldehydes.  The numbers in
the second column of Table 8-7 below are 39% of the ambient levels
and thus represent estimated motor vehicle levels.

     The motor vehicle apportionment of the ambient monitoring data
ranges from 0.94 to 1.21 ug/m3.   When the  adjustment factor of
0.622 for the ambient mobile source levels, that was determined in
Section 5.5.2 is applied, this range becomes 0.58 to 0.75 ug/m3.
Due to a potential ozone interfernece problem with the ambient data
other than the 1990 UATMP, only the 1990 UATMP adjustment estimate,
0.75 ug/m3,  will  be used for the  comparison to HAPEM.   When
compared to the HAPEM-MS 1990 base control level of 0.36 ug/m3,  the
1990 UATMP adjusted ambient monitoring data is observed to be
approximately two times greater then the HAPEM-MS base control
level.  The fact that modeled levels are 62% lower than monitored
data is consistent with secondary-formed acetaldehyde.  The HAPEM-
MS 1990 base control exposure level of 0.36 ug/m3  must be increased
by a factor of 2.09, to 0.75 ug/m3 to agree with the ambient data.
All analysis based on the HAPEM-MS ambient mobile source levels
will have this factor applied.  Adjusted urban, rural, and
nationwide exposures are found in Table 8-8.

     Any acetaldehyde exposures projected by HAPEM-MS itself should
be viewed with caution.  The adjusted HAPEM-MS exposure estimates
attempt to account for both primary and secondary acetaldehyde;
however, these estimates are based only on changes in primary
emissions of acetaldehyde.  The reactivity of motor vehicle VOC
emissions is likely to change with technology and fuel changes.
Changes in the reactivity of these emissions, which would result in
changes to secondary acetaldehyde levels,  cannot be accounted for
by HAPEM-MS.

8.5.3 Short-Term Micrenvironment Exposures

     The primary emphasis for acetaldehyde exposure will be
exposure in microenvironments that are enclosed, increasing the
exposure to tailpipe emissions.  These microenvironments include
in-vehicle and parking garage exposure, though, actual exposure
information is only available for in-vehicle exposure.  This
information is taken from the In-Vehicle Air Toxics
Characterization Study in the South Coast Air Basin (Shikiya et
al.,  1989), which focused on the driver's exposure to VOC's in the
southern California area.  See the information in Section 4.2 for
more details about the methodology, and Section 5.5.3 for a
description of the study.
                                1-16

-------
                                                         EPA-420-R-93-005
                                                             April 1993
Table 8-7.  Air Monitoring Results  for Acetaldehyde.
Program
AIRS
UATMP
Years
1988
1987
1989
1990
Ambient Data3
ug/m3
2.93
2 .41
2.45
3.10
Estimated
Mobile Source
Contribution1"
ug/m3
1.14
0.94
0.96
1.21
aCaution should be taken in comparing these numbers.  The  methods
of averaging the data are not  consistent  between air monitoring
databases.  The sampling methodology is also  inconsistent.

bThe  ambient data are adjusted to represent the motor vehicle
contribution to the ambient concentration,  which for acetaldehyde
is estimated to be 39%, based  on  emissions  inventory apportionment
and modeling.

-------
                                                           EPA-420-R-93-005
                                                                April 1993
Table 8-8.
Adjusted Annual Average  HAPEM-MS Exposure
     Projections for Acetaldehyde.
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Exposure
(ug/m3)
Urban
0.75
0.49
0.49
0.38
0.38
0.39
0.39
0.38
0.36
Rural
0.41
0.35
0.35
0.21
0.20
0.21
0.21
0.21
0.19
Nationwide
0.67
0.44
0.44
0.33
0.33
0.33
0.34
0.34
0.31
                                  8-18

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The in-vehicle exposure level of acetaldehyde was determined
in this study to have a mean of 13.7 ug/m3 and a maximum measured
level of 66.7 ug/m3.   This compares to ambient levels of 2.41 to
3.10 ug/m3  determined through air monitoring studies and presented
in Table 8-7.  Since for the majority of the population these are
short-term acute exposures to acetaldehyde, the concern would be
with non-cancer effects.  The primary acute effect of exposure to
acetaldehyde vapors is irritation of the eyes, skin, and
respiratory tract.  At high concentrations, irritation and
ciliastatic effects can occur.  Clinical effects include erythema,
coughing,  pulmonary edema, and necrosis.  It has been suggested
that voluntary inhalation of toxic levels of acetaldehyde would be
prevented by its irritant properties, since irritation occurs at
levels below 200 ppm  (3.6xl05 ug/m3).  Please  see Section 8.8  for
more information on non-cancer effects.

     A RfC of 9.0 ug/m3  per day over a lifetime has been developed
by EPA.  An RfC is an estimate of the continuous exposure to the
human population that is likely to be without deleterious effects
during a lifetime.  The mean and maximum levels, 13.7 ug/m3  and
66.7 ug/m3  respectively,  observed in Shikiya et al.,  (1989)  are
higher than RfC developed by EPA.

     Due to more stringent fuel and vehicle regulations, short-term
exposure to acetaldehyde in microenvironments is expected to
decrease in future years.

8.6 Carcinogenicity of Acetaldehyde and Unit Risk Estimates

8.6.1 Most Recent EPA Assessment

     An external review draft document entitled Health Assessment
Document for Acetaldehyde  (EPA, 1987) has been prepared.  Much of
the information contained in this section has been taken from this
document and the most recent IRIS summary  (EPA, 1992).

8.6.1.1 Description of Available Carcinogenicity Data

     The majority of information that exists to evaluate the
Carcinogenicity of acetaldehyde emissions relies on mutagenicity
studies and a few animal studies.

Genotoxicity

     Acetaldehyde has been shown in studies by several different
laboratories to induce sister chromatid exchange (SCE) in cultured
mammalian cells, e.g., Chinese hamster cells  (Obe and Ristow, 1977;
Obe and Beer, 1979; de Raat et al., 1983) and human peripheral
lymphocytes  (Ristow and Obe, 1979; Jansson, 1982; Bohlke et al.,
1983; Norrpa et al.,  1985; Obe et al., 1986) in a dose-related
manner.  A study by He and Lambert  (1985) provided evidence that
SCE-inducing lesions may be persistent for several cell
generations.  The induction of SCEs by acetaldehyde has also been
detected in the bone marrow cells of whole mammals, namely mice and
Chinese hamsters  (Obe et al., 1979; Korte and Obe,  1981).  In
addition to acetaldehyde's ability to induce SCEs,  it has been


                                8-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993

shown to be a clastogen in mammalian cell cultures  (Bird et al.,
1982) and plants (Rieger and Michaelis, I960).  Acetaldehyde has
produced chromosomal aberrations  (micronuclei, breaks, gaps, and
exchange-type aberrations) also in a dose-related manner  (Bird  et
al.,  1982; Bohlke et al.,  1983) .   In a study by Eker and Sanner
(1986),  acetaldehyde and formaldehyde were both able to initiate
cell transformation, though formaldehyde was 100 times more potent.
In Drosophila, chromosomal effects (i.e., reciprocal
translocations)  were not found after acetaldehyde treatment
(Woodruff et al.,  1985).   The clastogenicity of acetaldehyde in
whole mammals has not been sufficiently evaluated.  In the one
study that was available,  female rats were intra-amniotically
injected on the 13th day of gestation, and the treated embryos  had
high frequencies of chromosomal gaps and breaks (Barilyak and
Kozachuk, 1983).

     Although acetaldehyde did not produce chromosomal
translocations in Drosophila, it was found to induce gene mutations
(sex-linked recessive lethals) at the same concentration when
administered by injection (Woodruff et al.,  1985).  Positive
results for gene mutations were reported in the nematode,
Caenorhabitis (Greenwald and Horvitz, 1980),  and an equivocal
result was obtained for mitochondrial mutations in yeast  (Bandas,
1982).   Salmonella testing in numerous strains has been reported as
negative  (Commoner, 1976;  Laumbach et al.,  1976; Pool and Wiesler,
1981; Marnett et al.,  1985;  Mortelmans et al., 1986).  In two
studies utilizing Escherichia coli to detect a mutagenic effect,
one yielded positive results  (Veghelyi et al., 1978) and the other
study negative results (Hemminki et al.,  1980).  There were no
available data on the ability of acetaldehyde to produce gene
mutations in cultured mammalian cells.

     Acetaldehyde has not been shown to cause DNA strand breaks in
mammalian cells in vitro  (Sina et al., 1983;  Saladino et al., 1985;
Lambert et al.,  1985).  However,  if acetaldehyde produces SCEs  and
chromosomal aberrations by DNA-DNA or DNA-protein cross-linking, it
may not necessarily produce DNA strand breaks (Bradley et al.,
1979).   Acetaldehyde has been shown to produce crosslinks between
protein and DNA in the nasal respiratory mucosa of rats  (Lam et
al. ,  1986) .

     In conclusion, there is sufficient evidence that acetaldehyde
produces cytogenic damage in cultured mammalian cells.  Although
there are only three studies in whole animals, they suggest that
acetaldehyde produces similar effects in vivo.  Acetaldehyde
produced gene mutations in Drosophila but not in Salmonella; no
studies were found for cultured mammalian cells.  Thus, the
available evidence indicates that acetaldehyde is mutagenic and may
pose a risk for somatic cells.  On the other hand, it has been
suggested that acetaldehyde may be capable of deactivating free
cysteine in bronchial epithelial cells, thereby suppressing the
"thiol defense" of the epithelium against the attack of mutagens
and carcinogens (Braven et al. 1967;  Fenner and Braven, 1968).
Current knowledge,  however,  is inadequate with regard to germ cells
(reproductive cells) mutagenicity because the available information
is insufficient to support any conclusions about the ability of


                               8-20

-------
                                                        EPA-420-R-93-005
                                                            April 1993

acetaldehyde to reach mammalian gonads and produce heritable
genetic damage.

Animal Data

     Acetaldehyde has been tested for carcinogenicity in hamsters
by intratracheal instillation and inhalation and in rats by
subcutaneous injection and inhalation.  In the
inhalation/instillation study of hamsters  (Feron, 1979), two
testing protocols were used.  In part one, male hamsters were
exposed to 0 or 1500 ppm acetaldehyde by inhalation 7 hours/day, 5
days/week, for 52 weeks.  These animals were also exposed
intratracheally to benzo[a]pyrene (BaP) for a total concentration
at the end of 52 weeks ranging from 0 to 52 mg.   Exposure to
acetaldehyde by inhalation and intratracheal BaP induced
inflammatory changes, hyperplasia and metaplasia of the nasal,
laryngeal, and tracheal epithelium,  and tumors of the nose and
larynx.  Acetaldehyde enhanced the development of BaP-initiated
tracheobronchial carcinoma yielding twice the incidence of squamous
cell carcinoma compared with the same dose of BaP alone.  No
neoplastic effects due to acetaldehyde alone were found.

     In the second part of Feron (1979),  male and female hamsters
were intratracheally instilled with 4 or 8 uL acetaldehyde, BaP,
BaP and 4 uL acetaldehyde, diethylnitrosamine (DENA, a  tumor
promotor), or DENA and 4 uL acetaldehyde.  Acetaldehyde alone
produced no tumors in the larynx, trachea, or bronchi.  However,
large numbers of tracheal papillomas and lung adenomas  were found
in groups treated with acetaldehyde plus BaP or DENA.   There was no
evidence of acetaldehyde enhancing the development of DENA-
initiated respiratory tract tumors.


     In an extension of the above study  (Feron et al.,  1982), male
and female hamsters were exposed to a high concentration of
acetaldehyde vapor alone or simultaneously with either  BaP or DENA.
The animals were exposed 7 hours/day, 5 days/week, for  52 weeks to
a time weighted average concentration of 2028 ppm.  Tumors were
slightly increased in the nose and significantly increased in the
larynx of animals exposed to acetaldehyde vapor alone,  but no
tracheal tumors were observed.  The incidence of carcinomas in the
trachea and bronchi were significantly higher in hamsters exposed
to acetaldehyde and treated with high doses of BaP than in hamsters
treated with the same dose of BaP but exposed to air.   There was no
evidence that acetaldehyde exposure increased the incidence or
affected the type of DENA-induced tumors in any part of the
respiratory tract.

     Watanabe and Sugimoto  (1956) reported spindle-cell sarcoma in
20% to 25% of the rats tested at the site of repeated acetaldehyde
injection.  No conclusion can be drawn from this study  because
neither the total doses of acetaldehyde nor the tumor incidence in
controls could be determined from available data.

     The carcinogenicity of acetaldehyde was studied in albino SPF
Wistar rats (Woutersen and Appelman, 1984; Woutersen et al.,  1985,


                                8-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993

1986).   The studies are summarized in Table 8-9.  The animals were
exposed by inhalation to atmospheres containing 0, 750, 1500, or
3000 ppm acetaldehyde for 6 hours/day, 5 days/week for 27 months.
The concentration in the highest dose group was gradually reduced
from 3000 to 1000 ppm because of severe growth retardation,
occasional loss of body weight, and early mortality in this group.
Interim sacrifices were carried out at 13, 26, and 52 weeks.  One
tumor was observed in the 52 week sacrifice group and none at
earlier times.  Exposure to acetaldehyde increases the incidence of
tumors in an exposure-related manner in both male and female rats.
In addition, there were exposure-related increased in the incidence
of multiple respiratory tract tumors.  Adenocarcinomas were
increased significantly in both male and female rats at all
exposure levels, whereas squamous cell carcinoma were increased
significantly in male rats at middle and high doses and in the
female rats only at the high dose.  The squamous cell carcinomas
incidence showed a clear dose-response relationship.  The incidence
of adenocarcinomas was highest in the mid-exposure group  (1500 ppm)
in both male and female rats, but this was probably due to the high
mortality and competing squamous cell carcinomas at the highest
exposure level.  In the low-exposure group, the adenocarcinoma
incidence was higher in males than in females.

     In a concurrent study, referred to as the "recovery study", 30
animals of each sex were exposed to the same concentrations of
acetaldehyde for 52 weeks followed by a recovery period of 26 weeks
or 52 weeks.  Significant increases in nasal tumors were observed
in male and female rats, including adenocarcinomas and squamous
cell carcinomas, in both recovery groups.  These findings indicate
that after 52 weeks of exposure to acetaldehyde, proliferative
epithelial lesions of the nose may develop into tumors even without
continued exposure.
                                !-22

-------
                                                                                                    EPA-420-R-93-005
                                                                                                         April 1993
Table  8-9.   Animal Data  Used for EPA's Unit  Risk  Estimates.
  REPORT
                         ANIMAL
                EXPOSURE
                CONCENTRATION
                   LENGTH  OF
                   EXPOSURE
                   MAJOR  RESULTS
 Woutersen and
 Appelman  (1984) also
 as Wouterson et al.,
  (1986)

 Woutersen et al.
  (1985)

 "lifetime study"
albino SPF
Wistar rats
male and
female
0, 750,  1500,
and 3000 ppm
(lowered to 1000
ppm)  of
acetaldehyde
6h/d,  5d/wk,  for
27 months
1.   Acetaldehyde vapor exposure
caused two types of tumors in
the nasal tract of rats in an
exposure related manner.   The
four exposure levels gave tumor
incidences of 1, 21, 52,  and 51%
respectively.

2.   Degeneration of nasal tissue
was observed at all dose  levels.

3.   Exposure appears not  to
affect any other organ directly
except for lesions of the larynx
and, to a minor degree, the
trachea.

4.   Animals in the high exposure
group (3000 ppm) suffered severe
growth retardation, respiratory
distress, and high early
mortality.
 Woutersen and
 Appelman  (1984)

 recovery subgroup of
 original study
albino SPF
Wistar rats
male and
female
same exposures
as used above

subjected to a
26 or a 52 week
recovery period
same as used
above but for 52
weeks
With respect to recovery,  during
the first 26 weeks the nasal
tumor rates and death rates were
essentially the same as the
lifetime exposure group.   From
26-52 weeks both low- and mid-
exposure recovery groups  had
significantly decreased nasal
tumor rates.  This indicates
that  nasal lesions may still
develop into tumors even  after
exposure stops, and that  the
nasal tissue may also be  able to
repair some damage.	
                                                       8-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993


Human Data

     The only epidemiological study involving acetaldehyde exposure
showed an increased crude incidence rate of total cancer in
acetaldehyde production workers as compared to the general
population  (Bittersohl, 1974).  The study was performed on workers
from an aldol and aliphatic acetaldehyde factory.  The study showed
a five times higher cancer rate than that of the general
population.  An incidence rate of 6000/100,000 population for total
cancer was calculated for this study, which contrasts with an
incidence rate of 1200/100,000 for the general population of
Germany during the same period.  Because the incidence rate was not
age adjusted, and because this study has several other major
methodological limitations (concurrent exposure to cigarette smoke
and other chemicals,  short duration, small number of subjects, and
lack of information on subject selection, age, and sex
distribution),  the evidence is considered inadequate for the
carcinogenicity of acetaldehyde in humans.

8.6.1.2 Weight-of-Evidence Judgement of Data and EPA Classification

     The data used for the quantitative estimates for acetaldehyde
are limited to the Woutersen and Appelman (1984) and the Woutersen
et al., (1985)  rat inhalation studies  (summarized in Table 8-9)
showing an exposure related increase in nasal tumors in Wistar rats
and supported by positive results for mutagenicity.   This evidence
for carcinogenicity of acetaldehyde in animals is considered to be
sufficient based on the U.S.  EPA cancer assessment guidelines.
Neither of the hamster studies (Feron, 1979; Feron et al.,  1982)
are considered satisfactory based on the fact that one was an
intratracheal instillation study and the other was an inhalation
study which had very high exposure levels of acetaldehyde and only
one exposure group.

     The only epidemiological study, Bittersohl  (1974),  showed an
increase in crude incidence rate of total cancer in acetaldehyde
production workers as compared with the general population.
Because the incidence rate was not age adjusted, and because this
study has several other major methodological limitations,  the
evidence is considered inadequate for the carcinogenicity of
acetaldehyde in humans.

     On the basis of inadequate evidence for carcinogenicity of
acetaldehyde emissions in humans, and relying totally on the
sufficient evidence from animals and mutagenicity, acetaldehyde
emissions are considered to best fit the weight-of-evidence
category B2.  This classifies acetaldehyde as a probable human
carcinogen.
                                !-24

-------
                                                        EPA-420-R-93-005
                                                            April 1993


8.6.1.3 Data Sets Used for Unit Risk Estimates

     To actually determine the unit risk of acetaldehyde, only two
of the animal studies are selected for risk calculations because
they are inhalation studies that involve more than one exposure
group.  The two rat studies used are Woutersen and Appelman  (1984)
(also known as Wouterson et al.,  1986), and Woutersen et al.
(1985).  These studies are summarized in  Table 8-9.

8.6.1.4 Dose-Response Model Used

     The linearized multistage model is used to calculate unit risk
estimates using various exposure inputs.  All unit risk estimates
that currently exist for acetaldehyde are based exclusively on
animal data.

8.6.1.5 Unit Risk Estimates

     The upper-limit unit risk estimate for acetaldehyde is 2.2x
10"6 (ug/m3)-1, derived from the male rat tumor data.  Corresponding
maximum likelihood estimates  (MLE's) were not given.

8.6.2 Other Views and Risk Estimates

          This section presents alternate views and/or risk
assessments for acetaldehyde.

International Agency for Research on Cancer  (IARC)

     IARC has classified acetaldehyde as a Group 2B carcinogen.  A
Group 2B carcinogen is defined as an agent that is possibly
carcinogenic to humans.  This classification is based on inadequate
evidence for carcinogenicity in humans and sufficient evidence for
carcinogenicity in animals (IARC, 1987) .


California Air Resources Board (CARE)

     CARB (1992b),  like EPA and IARC, has concluded that
acetaldehyde is a probable human carcinogen.  CARB  (1992b) has
performed an assessment of the carcinogenic risk of acetaldehyde
using the Wouterson et al.,   (1986) rat nasal carcinoma data
(discussed previously in Section 8.6.1) in the linearized time-
dependent multistage model.   However, their assessment differs from
EPA  (1987)  in the following ways:


      (1)  The EPA (1987) risk assessment considered all 55 animals
          in the experimental groups to be at risk, whereas CARB
          used only the 49-53 animals of each group that were
          examined for nasal changes.

      (2)  CARB used only the male rat data from Wouterson et al.,
          (1986)  whereas EPA used both the male and the female
          data.  CARB stated that the male rat is more sensitive to


                               8-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993

          tumor induction by acetaldehyde than the  female  rat  and
          this is the proper sex to select based on CARB procedures
          for cancer risk assessment.


     (3)  The  EPA  (1987) risk assessment combined  two experiments
          by Wouterson et al., the lifetime exposure experiment  and
          an experiment in which one year of exposure was  followed
          by one year of recovery.  CARB , however, used only  the
          lifetime exposure experiment.

     (4)  EPA (1987) used two versions of the linearized multistage
          model: the standard version and the time-to-tumor
          version.  CARB used only the standard version citing that
          the information to adequately use the time-to-tumor
          version was not available in the experimental data and
          thus should not be used.

      (5) The CARB approach uses three different scaling factors to
          extrapolate the equivalent dose rate from rats to humans.
          EPA (1987) did not specifically discuss the issue of
          scaling to extrapolate from rodents to humans for
          formaldehyde.

     The UCL for unit risk for lifetime exposure calculated by CARB
(1992b)  using the methods and assumptions described above  is
4.8xlO"6 ppb"1  (2.7xlO~6  [ug/m3]"1).   CARB  also  calculated  a  range of
UCL for unit risks.  This range is 9.7xlO"7 ppb"1  for female rats
without a scaling factor to 2.7xlO"5 ppb"1  for male  rats  with a
contact area correction  (1.19xlO~6  to 3.32xlO"5 [ug/m3]"1).

8.6.3 Recent and Ongoing Research

8.6.3.1  Genotoxicity

     Dulout and Furnus (1988) determined that the most notable
cytogenetic effect of acetaldehyde in cultured Chinese hamster
ovary (CHO) cells was aneuploidy  (the chromosome number is not an
exact multiple of the haploid number) and not chromosomal  breakage.
Acetaldehyde added for 24 hours to cultures at concentrations  of
0.002%,  0.004%,  and 0.006% produced an increased frequency of
aneuploidy as compared to controls.  The aneuploidy was observed at
all doses tested, whereas chromosomal aberrations and sister
chromosomal exchanges only occurred at the two highest levels.

     The effect of acetaldehyde on the frequency of meiotic
micronuclei in groups of four hybrid male mice was  assessed 13 days
after a single intraperitoneal injection of 0, 125, 250, 375,  of
500 mg/kg acetaldehyde in saline solution.  No significant
increases in the frequency of micronuclei were observed (Lahdetie,
1988).   The alkaline dilution technique was used by Garberg, et  al.
(1988)  to determine whether the DNA of mouse lymphoma cells exposed
to acetaldehyde contain single-strand breaks.  Single -strand
breaks were not detected in this cell type or in rat hepatocytes,
human lymphocytes, and bronchial epithelial cells studied
                                8-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993

previously (Sina et al.,  1983, Lambert et al.,  1985; Saladino et
al. ,  1985) .

8.6.3.2 Metabolism and Pharmacokinetics

     The following has been excerpted from EPA (1987).   The
extensive references have been omitted to facilitate the
comprehension of this section.  For the complete list of
references,  please consult Chapter 4 of EPA (1987) .  Other studies
that have been published since the issuance of the 1987 draft
document support the 1987 position summarized below.

     The principal routes of entry of acetaldehyde into the body
are by gastrointestinal and inhalation absorption.  Acetaldehyde,
whether from exogenous (from outside the body)  sources or generated
from ethanol metabolism,  is known to be very rapidly and
extensively metabolized oxidatively in mammalian systems to a
normal endogenous  (inside the organism) metabolite, acetate,
primarily by aldehyde dehydrogenases (specific enzymes) widely
distributed in body tissue.  Acetate enters the metabolic pool of
intermediary metabolism and is used in cellular energy production
(end products C02 and water)  or in synthesis of  cell constituents.
In contrast to the situation for acetaldehyde generated from
ethanol metabolism, there are few studies of the kinetics of
acetaldehyde of exogenous origin, i.e., from environmental exposure
or experimental dosing.   It is known, however,  that all mammalian
species have a high capacity to rapidly and virtually completely
metabolize acetaldehyde by most tissues in the body, including the
gastrointestinal mucosa and respiratory mucosa and lungs.  Although
hepatic (liver) capacity is the highest after oral or inhalation
administration, experimental evidence indicates that a substantial
first-pass metabolism in the liver or respiratory organs
effectively limits acetaldehyde access to the systemic circulation.
However, adequate studies have not been conducted to establish
dose-metabolism relationships, or dose-blood concentration
relationships.

     Acetaldehyde readily crosses body compartmental membranes into
virtually all body tissues, including the fetus,  after
administration or endogenous generation.  Animal experiments have
demonstrated a rapid exponential disappearance from circulating
blood, consistent with first-order kinetics,  with a short half-time
of elimination of less than 15 minutes.  Since less than 5 percent
escapes unchanged in exhaled breath, and acetaldehyde is not known
to be excreted into the urine, the elimination from the body is
essentially by metabolism.

     Acetaldehyde is a highly reactive compound and at high
concentrations episode,  for example, at the respiratory mucosa with
inhalation exposure, it readily forms adducts nonenzymatically with
membranal and intracellular macromolecules.   Stable and reversible
adduct formation including cross-linking have been demonstrated
with proteins,  nucleic acids  (including DNA),  and phospholipids.
Moreover,  even at physiological levels  (10 to 150 umol/L blood),
acetaldehyde has been found to form adducts with cellular
macromolecules.  From these observations, it has been considered


                                8-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993

that acetaldehyde-adduct formation may play a role in the organ and
cellular injury associated with acetaldehyde toxicities, and in the
potential promoter or carcinogenic effect assigned to this
compound.  Acetaldehyde also readily reacts nonenzymatically with
cysteine and glutathione (proteins with sulfur groups [thiols]
attached) to form stable and reversible adducts, respectively.
Hence,  acetaldehyde may be an effective depleter of these important
cellular nonprotein thiols, which represent a thiol defense against
the attack of toxic aldehydes and other mutagens and carcinogens.


8.7  Carcinogenic Risk for Baseline and Control Scenarios

     Table 8-10 summarizes the annual cancer incidences for all the
scenarios.  These numbers are presented as decimals due to the fact
that the cancer cases are low enough that rounding the decimal up
or down would significantly affect the total number.   The cancer
cases do decrease slightly from a comparison drawn between base
control scenarios.  When compared to the 1990 base control, the
cancer incidence decreases by 32% in 1995, 47% in 2000,  and 43% in
2010, which is actually an increase when compared to 2000.  The
reductions are basically due to the tighter tailpipe standards
specified by the Tier 1 standards.  In contrast, when compared to
the 1990 base control, the emission factors decrease 24% in 1995,
45% in 2000 and 65% in 2010.  The difference observed between the
emission factor and cancer case reductions, and the increases
observed in 2010, is due to the expected increase in population and
VMT, which appear to offset the emission gains achieved through
fuel and vehicle modifications.

     From Table 8-10 it can also be observed that the expanded use
of reformulated fuel and the expansion of the California standards
provide no significant decrease in the cancer cases and, in several
scenarios, the cancer cases increase.  As mentioned in previous
sections, the exposure estimates are based on changes in direct
emissions of acetaldehyde.   Changes in reactivity of the emissions,
which would result in changes to secondary acetaldehyde, are not
accounted for.  Since it is probable that secondary acetaldehyde
could be reduced with the use of oxygenates, the cancer risk
estimates given in Table 8-10 should be considered conservative
estimates.
                                8-28

-------
                                                                                  EPA-420-R-93-005
                                                                                      April 1993
Table 8-10.  Annual Cancer  Incidence  Projections for Acetaldehyde.
                                                                   a,b
Year- Scenario
1990 Base Control
1995 Base Control
1995 Expanded Reformulated
Fuel Use
2000 Base Control
2000 Expanded Reformulated
Fuel Use
2000 Expanded Adoption of
California Standards
2010 Base Control
2010 Expanded Reformulated
Fuel Use
2010 Expanded Adoption of
California Standards
Emission
Factor
g/mile
0.0119
0.0071
0.0071
0.0051
0.0051
0.0052
0.0045
0.0044
0.0041
Urban
Cancer
Cases
4.5
3.0
3.0
2.4
2.4
2.4
2.6
2.6
2 .4
Rural
Cancer
Cases
0.8
0.6
0.6
0.4
0.4
0.4
0.4
0.4
0.4
Total
Cancer
Cases
5.3
3.6
3.6
2.8
2.8
2.8
3.0
3.0
2.8
Percent Reduction
from 1990
EF
-
40
40
57
57
56
62
63
66
Cancer
-
32
32
47
47
47
43
43
47
""Projections have inherent uncertainties in emission  estimates,  dose response,  and exposure.
bCancer incidence estimates are based on upper bound  estimates  of unit risk,  determined from
animal studies.  Acetaldehyde  is  classified by EPA as a category B2, probable human  carcinogen,
based on insufficient evidence in humans  and sufficient evidence in animals and  in mutagenicity
bioassays.
                                              8-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Please note that the cancer unit risk estimate for
acetaldehyde is based on animal data and is considered an upper
bound estimate for human risk.  True human cancer risk may be as
low as zero.

8.8 Non-Carcinogenic Effects of Inhalation Exposure to
Acetaldehyde

     Since the focus of this report is on the carcinogenic
potential of the various compounds, the noncancer information
will be dealt with in a more cursory fashion.  No attempt has
been made to synthesize and analyze the data encompassed below.
Also, no attempt has been made to accord more importance to one
type of noncancer effect over another.  The objective is to
research all existing data, describe the noncancer effects
observed, and refrain from any subjective analysis of the data.

8.8.1  Toxicity

     The results of eight acute toxicity studies in mammals, by
inhalation in rats (Skog, 1950; Lewis and Tatkin, 1983) and mice
(Kane et al.,  1980; Barrow, 1982), by the oral route in rats
(Windholz et al.,  1983; Lewis and Tatkin, 1983; Omel'yanets et
al.,  1978) and mice (National Research Council, 1977) along with
intravenous instillation in guinea pigs  (Mohan et al., 1981) and
subcutaneous injection in rats and mice  (Skog, 1950) all show
LD50 effects.   The  acute oral LD50 of acetaldehyde ranged from 1232
mg/kg to 5300 mg/kg.   The LD50 for  subcutaneous injection ranged
from 560 mg/kg to 640 mg/kg.  The acute inhalation LC50 was
20,000 ppm in rats exposed to acetaldehyde for 30 minutes.  In
one study (Lewis and Tatkin, 1983), 4000 ppm for 4 hours killed
some exposed rats.  The following section will discuss some of
these acute toxicity studies in more detail.

     Studies with rats and mice showed acetaldehyde to be
moderately toxic by the inhalation route, oral, and intravenous
routes.  Acetaldehyde is a sensory irritant that causes a
depressed respiration rate in mice (Kane et al., 1980; Barrow,
1982).  This yielded RD50's  (the concentration  that produces a
50% decrease in respiratory rate)  of 4946 ppm and 2845 ppm,
respectively.   The current TLV for acetaldehyde is 100 ppm
(American Conference of Governmental Industrial Hygienists,
1980), and is between 0.1 and 0.01 times the cited RD50 values.
In rats, acetaldehyde increased blood pressure and heart rate
after exposure by inhalation  (Egle, 1972) and intravenous
injection (Mohen et al.,  1981; Egle et al. , 1973).

     Three subchronic inhalation studies in rats and one in
hamsters have been conducted.  Appelman et al. (1982) used the
exposure information discussed below in the RfC section.  Rats
exposed to the highest concentration exhibited severe dyspnea and
marked excitation during the first 30 minutes of exposure.  Rats
at the highest exposure also exhibited decreased body weight,
lymphocytes, and liver weights when compared to controls.  The
neutrophil counts and lung weights were increased.
Histopathological alterations of the respiratory system were seen


                               8-30

-------
                                                        EPA-420-R-93-005
                                                            April 1993

at all dose levels, with the nose being the most severe.  The
study by Appelman et al.  (1986)  (exposure information is detailed
in RfC section) was also observed for non-cancer effects.
Uninterrupted exposure to 500 ppm did not produce any changes in
condition, behavior, or body weight of the rats; however, rats
exposed to 500 ppm with a peak exposure of 3000 ppm exhibited
irritation, as indicated by eye blinking, excessive running, and
nose twitching.  Mean body weights in the latter groups were
significantly lower than in controls.  In addition, a reduced
phagocytotic index was significantly decreased at the highest
dose.

     The effect of acetaldehyde on pulmonary mechanics was
studied following exposure of groups of Wistar rats to
acetaldehyde vapors at concentrations of 0 or 243 ppm (0 or 105.3
mg/m3) ,  8  hours/day,  5  days/week for 5  weeks.  (Hilaro et al.,
1985).  A significant increase in respiratory frequency,
functional residual capacity, residual volume, and total lung
capacity was noted.  The subchronic inhalation study in hamsters
(Kruysse et al.,  1975)  is discussed below in the RfC section.

     One subchronic investigation of the effects of acetaldehyde,
on the phospholipid composition of pulmonary surfactant, was
found in the literature (Prasanna et al., 1981).  Pulmonary
surfactant is a lipoprotein complex with a high phospholipid
content which prevents alveolar collapse during expiration by
maintaining the stability and physical elasticity of the alveolar
walls, and by reducing the surface tension of the fluid lining
the alveoli.  Acetaldehyde injected intraperitoneally to rats at
200 mg/kg significantly reduced the phospholipid concentration of
pulmonary surfactant.  In two subchronic intravenous studies, one
in guinea pigs (Mohen et al., 1981), and the other in rats  (Egle
et al.,  1973), a dosage of 20 mg/kg acetaldehyde or lower caused
an immediate and significant increase in blood pressure.

     In a chronic inhalation study  (Feron, 1979), acetaldehyde
vapor at 1500 ppm for 52 weeks produced systemic effects in the
hamster:  growth retardation, slight anemia, increased UGOT
(urinary glutamic-oxaloacetic transaminase)  activity, increased
urine protein content,  increased kidney weights, and
histopathological changes in the nasal mucosa and trachea.  In a
separate study (Feron,  1979), intratracheal instillation of
acetaldehyde  (2 to 4 percent) to hamsters weekly or biweekly for
up to 52 weeks caused severe hyperplastic and inflammatory
changes in the bronchioalveolar region of the respiratory tract.
In Feron et al. (1982), male and female hamsters exposed to
levels of acetaldehyde vapor decreasing from 2500 ppm to 1650 ppm
over 52 weeks had lower body weights than controls and distinct
histopathological changes in the nose,  trachea, and larynx.

     Humans are frequently exposed to acetaldehyde from cigarette
smoke, vehicle exhaust fumes, or other sources.  Metabolism of
ethanol would be the major source of acetaldehyde among consumers
of alcoholic beverages.
                               1-31

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     The primary acute effect of exposure to acetaldehyde vapors
is irritation of the eyes, skin, and respiratory tract  (Sim and
Pattle, 1957).   At high concentrations, irritation and
ciliastatic effects can occur, which could facilitate the uptake
of other contaminants  (NRC, 1981).   Clinical effects include
erythema,  coughing, pulmonary edema, and necrosis  (Dreisbach,
1980).   Respiratory paralysis and death have occurred at
extremely high concentrations.  It has been suggested that
voluntary inhalation of toxic levels of acetaldehyde would be
prevented by its irritant properties, since irritation occurs at
levels below 200 ppm (Sittig, 1979).  It was concluded by the
Committee of Aldehydes of the National Research Council  (1981)
that direct pulmonary sensitization to aldehyde vapors appears to
be relatively rare and asthma-like symptoms are rarely caused by
the inhalation of aldehydes.

     The main route of occupational exposure is by inhalation of
acetaldehyde vapor.  The allowable federal time weighted average
(TWA) is 200 ppm (360 mg/m3)  for eight  hours  per day,  five days
per week.   The American Conference of Governmental and Industrial
Hygienists  (1985) recommends a threshold limit value (TLV) of 100
ppm  (180 mg/m3)  for eight  hours  per day,  five days  per  week.

8.8.2 Reference Concentration for Chronic Inhalation Exposure
(RfC)

     At the present time,  the reference dose for chronic oral
exposure (RfD)  assessment is not available but the reference
concentration for chronic inhalation exposure  (RfC) has recently
been completed  (EPA, 1992).  An RfC is an estimate of the
continuous exposure to the human population that is likely to be
without deleterious effects during a lifetime.

      Two short term studies conducted by the same research group
are the principal studies used.   While these studies are short
term in duration, together they establish a concentration-
response for lesions after only 4 weeks of exposure.

     Appelman et al. (1986) conducted two inhalation studies on
male Wistar rats exposing them 6 hrs/day, 5 days/week for 4 weeks
to 0, 150,  and 500 ppm (0, 273,  and 910 mg/m3,  respectively.   One
group was exposed without interruption, a second group was
interrupted for 1.5 hours halfway through the exposure, and a
third group was interrupted as described with a peak exposure
imposed four times in a three hour period (concentration at peak
was six times the basic concentration).  Degeneration of the
olfactory epithelium was observed in rats exposed to 500 ppm.
Interruption of the exposure or interruption combined with peak
exposure did not visibly influence this adverse effect.  No
compound-related effects were observed in rats interruptedly or
uninterruptedly exposed to 150 ppm during the 4 week exposure
period; therefore,  the NOAEL is 150 ppm  (no-observed-adverse-
effect level).

     Appelman et al, (1982) exposed Wistar rats for 6 hour.day, 5
days/week for 4 weeks to 0, 400, 1000,  2200,  or 5000 ppm


                              8-32

-------
                                                        EPA-420-R-93-005
                                                            April 1993

acetaldehyde (0, 728, 1820, 4004, and 9100 mg/m3) .   The nasal
cavity was most severely affected and exhibited a concentration-
response relationship.  At all levels of acetaldehyde exposure in
this experiment, there was found nasal olfactory degeneration
that increased in severity as the concentration increased.  Also
as the concentration increased, the laryngeal and tracheal
epithelium became involved (1000 to 5000 ppm).  Based on the
degenerative changes observed in the olfactory epithelium, the
400 ppm level is designated as a LOAEL  (lowest-observed-adverse-
effect level).

     There are also three additional studies used to support the
inhalation RfC.  Woutersen et al. (1986), which was discussed
previously, exposed rats to 0, 750,  1500, and 3000/1000 ppm
acetaldehyde vapor.  The only exposure related histopathology
occurred in the respiratory system and showed a concentration-
response relationship.  The most severe abnormalities were found
in the nasal cavity.  Basal cell hyperplasia of the olfactory
epithelium was seen in the low- and mid-concentration rats.  The
decrease in these changes in the olfactory epithelium was
attributed to the incidence of adenocarcinomas at the higher
levels.  The lowest exposure concentration, 750 ppm, is clearly a
LOEAL based on the above changes in the olfactory epithelium.

     Woutersen and Feron (1987) conducted an inhalation study in
which Wistar rats were exposed to 0, 750, 1500, or 3000/1500 ppm
acetaldehyde (0, 1365, 2730,  5460/2730 mg/m3,  respectively)  for 6
hours/day, 5 days/week for 52 weeks with a 26- or 52-week
recovery period.  Degeneration of the olfactory epithelium was
similar in rats terminated after 26 weeks of recovery and rats
killed immediately after exposure termination.  Histopathological
changes found in the respiratory epithelium were comparable with,
but less severe than, those observed immediately after exposure
termination.  After 52 weeks of recovery, the degeneration of the
olfactory epithelium was still visible to a slight degree in
animals from all exposure groups.  The data suggest that there is
incomplete recovery of olfactory and respiratory epithelium
changes induced at all exposure concentrations for periods as
long as 52 weeks after exposure termination.

     Kruysse et al.  (1975)  conducted a 90-day inhalation study in
hamsters.   The hamsters were exposed to acetaldehyde vapor at
concentrations of 0, 390, 1340, or 4560 ppm  (0, 127, 435.3, or
1482 mg/m3),  for 6  hours/day,  5 days/week for 90  days.   In this
study, as in the previous studies, the histopathological changes
attributable to exposure were observed only in the respiratory
tract.  At the 390 ppm concentration, with one exception, no
adverse effects were observed.  The 390 ppm concentration was
identified by the authors as a NOAEL.


     The final RfC was calculated using the NOAEL from Appelman
et al. (1986) of 273 mg/m3, an uncertainty factor (UF)  of 1000,
and a modifying factor (MF) of 1.  The UF of 1000 was obtained by
assigning an uncertainty factor of 10 to account for sensitive
human populations,  another factor of 10 for both uncertainty in


                               8-33

-------
                                                        EPA-420-R-93-005
                                                            April 1993

the interspecies extrapolation using dosimetric adjustments and
to account for the incompleteness of the data base, and a third
factor of 10 to account for subchronic to chronic extrapolation.
The MF of 1 is the default and is based upon an assessment of the
scientific uncertainties of the toxicological data base not
treated with the UF.  The final number arrived at for the RfC is
9xlO"3 mg/m3 per day over a lifetime.

8.8.3  Reproductive and Developmental Effects

     No inhalation studies for reproductive or developmental
effects have been performed.  In all the in vivo studies cited
below, acetaldehyde is administered by the oral, intravenous, or
intraperitoneal route.

     Ali and Persaud  (1988) studied the role of acetaldehyde in
the pathogenesis of ethanol-induced developmental effects.
Sprague-Dawley rats received intraperitoneal injections of a 1%
solution of acetaldehyde at a dose of 100 mg/kg/day from days 9
through 12 of gestation.  On day 12, the embryos were recovered
and examined for
morphological abnormalities and crown-rump and head length.
Acetaldehyde produced a significant reduction in head length, but
had no significant effect on morphological abnormalities or
crown-rump length.  The reduction in head length was considered
to be important, since it may be a causative factor in the
microencephaly and CNS dysfunction found in fetal alcohol
syndrome.

     Kalmus and Buckenmaier (1989)  investigated the effects of
acetaldehyde on cultured preimplantation 2-cell stage mouse
embryos in vitro.  Embryos were exposed to 0, 5, 10, 200, or 500
mg acetaldehyde/100 ml culture medium for 60 minutes.  Embryo
growth was evaluated at a time period corresponding to an embryo
age of 105 hours.  No effects were observed at 5 and 10 mg/lOOml;
exposures to 50mg/100ml and higher were lethal.  The results
indicate that the 2-cell stage embryos are highly resistant to
high in vitro dosages of acetaldehyde; however, the reason for
the apparent resistance is not known.

     Zorzano and Herrera  (1989) studied the pattern of
acetaldehyde appearance in maternal and fetal blood, maternal and
fetal liver and placenta after oral ethanol administration or
intravenous acetaldehyde administration (lOmg/kg) to pregnant
Wistar rats.  The study demonstrated that acetaldehyde was able
to cross the placental barrier at high concentrations; maternal
blood concentration had to be greater that 80 uM.  The fetal
oxidation capacity in liver and placenta was shown to be lower
than that of the maternal liver.  A threshold above which the
removal capacity of acetaldehyde metabolism by the fetoplacental
unit would be surpassed was estimated to be 80 uM  (maternal blood
concentration) in the 21-day pregnant rat and possibly lower at
early pregnancy when aldehyde dehydrogenase is absent from fetal
liver.
                               1-34

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Lahdetie (1988) is the only study available on the in vivo
effects of acetaldehyde on the male reproductive system.  Groups
of hybrid male mice were given intraperitoneal injections of
saline solution 0, 62.5, 125, or 250 mg acetaldehyde/kg daily for
5 days.  No significant effects on sperm were seen for sperm
count, sperm morphology, testes weight, or seminal vesicle weight
when compared with controls.  The authors speculated that, since
no significant effects on sperm were seen, although acetaldehyde
had been shown to produce mutagenic effects in somatic cells,
germ cells were either less sensitive to the genotoxic effects or
the acetaldehyde concentrations was too low because of its
binding to erythrocytes and limited passage through the blood-
testes barrier.

     In female,  pregnant rats, across several studies  (Sreenathan
et al., 1982, 1984a,b; Sreenathan and Padmanabhan, 1984;
Padmanabhan et al.,  1983; Checiu et al, 1984; Barilyak and
Kozachuk,  1983;  Dreosti et al.,  1981), many of the same effects
were observed.  These include increased fetal resorption,
increases in litter malformations, retardation in fetal growth,
decreased placental weight, increased placental lesions,
decreases in skeletal formation, delayed segmentation and
differentiation of the embryo, increased cell fragmentation of
the embryo, increased chromosomal abnormalities, and interference
in thymidine incorporation into the DNA of the brain and liver.

     In female,  pregnant mice (Blakley and Scott, 1984a,b;
Bannigan and Burke,  1982; Webster et al.,  1983; 0'Shea and
Kaufman,  1979, 1981),  there are several studies used to
demonstrate reproductive and developmental effects, with most
results leading to much uncertainty or doubt as to their
advantage to understanding these effects.   Several studies found
no effects, whereas, some found an increase in fetal resorptions,
fetal growth retardation, and increased number of fetuses with
malformations.  Most of these malformations were neural tube
defects.

     There are additional data that support the hypothesis that
acetaldehyde interferes with placental function.  In a series of
studies (Henderson et al.,  1981, 1982; Asai et al., 1985; Fisher
et al., 1981a,b; 1984), the ability of acetaldehyde to interfere
with amino acid uptake across the placenta was demonstrated.
This demonstrates that this disruption in placental function may
create a state of fetal malnutrition that is independent of
maternal nutritional status.  Such a state may be a factor in
intrauterine growth retardation.  These studies must be
interpreted with some caution.  These studies examined the status
of term placentas and it remains to be determined what
relationship this has to preplacental structures.

     There are also several studies (Thompson and Folb, 1982;
Higuchi and Matsumoto, 1984; Campbell and Fantel, 1983; Popov et
al.,  1982; Prescott, 1985)  that have examined the direct
embryotoxic properties of acetaldehyde utilizing whole embryo
cultures  (rat and mouse).  The majority of these data demonstrate
                               1-35

-------
                                                        EPA-420-R-93-005
                                                            April 1993

that acetaldehyde can produce growth retardation and
malformations in vitro.

      The primary support for acetaldehyde-induced reproductive
dysfunction is derived from in vitro studies examining the
influence of acetaldehyde on androgen  (male hormone) production.
The majority of these studies  ( Cobb et al., 1978, 1980; Boyden
et al.,  1981; Badr et al.,  1977; Cicero et al. , 1980a,b; Santucci
et al.,  1983; Johnson et al.,  1981; Cicero and Bell, 1980) have
demonstrated that acetaldehyde significantly depresses HCG-(human
chorionic gonadotrophin) stimulated testosterone production;
however, the exact mechanism is unknown.  This effect has been
reported in a number of  species, including mice, rats, and dogs.

     Only one study has  examined the reproductive effects of
acetaldehyde aside from  endocrine influences.  Anderson et al.
(1982)  assessed the effects of acetaldehyde on sperm
capacitation.  These authors demonstrated that acetaldehyde did
not alter the in vitro fertilizing capacity of mouse spermatozoa,
though the relevance of  this culture system to in vivo
fertilization is unclear.

     In vitro data strongly suggest the possibility of male
reproductive toxicity and support the need for such data to be
generated in in vivo systems.
                               8-36

-------
                                                        EPA-420-R-93-005
                                                            April 1993
8.10  References for Chapter 8
All, F. and T. Persaud.  1988.  Mechanisms of fetal alcohol
effects:  role of acetaldehyde.   Exp. Pathol. 33:  17-21.

American Conference of Governmental Industrial Hygienists.  1980.
Documentation of the threshold limit values.  Fourth edition:
1980.  Cincinnati, OH:  American Conference of Governmental
Industrial Hygienists; pp. 3-4.

American Conference of Governmental Industrial Hygienists.  1985.
TLVs: threshold limit values and biological exposure indices for
1985-86.  Cincinnati, OH:  American Conference of Governmental
Industrial Hygienists; pp. 3-4.

Anderson, R.A. J.M. Reddy, C. Joyce, B.R. Willis, H. Van der
Vend, and L.J. Zaneveld.  1982.   Inhibition of mouse sperm
capacitation by ethanol.  Biol.  Reprod. 27:  833-840.

Appelman, L.M., R.A. Woutersen,  and V.J. Feron.  1982.
Inhalation toxicity of acetaldehyde in rats.  I.  Acute and
subacute studies.  Toxicology.  23:  293-307.

Appelman, L.M., R.A. Woutersen,  V.J. Feron R.N. Hooftman, and
W.R.F. Notten.  1986.  Effect of variable versus fixed exposure
levels on the toxicity of acetaldehyde in rats.  J. Appl.
Toxicol. 6(5):  331-336.

Asai, M., 0. Narita, and S. Kashiwamatu.  1985.  Effect of
acetaldehyde and/or ethanol on neutral amino acid transport
systems in microvillous brush border membrane vesicles prepared
from human placenta.  Experientia 41:  1566-1568.

Badr, F.M.,  A. Bartke, S. Dalterio, and W. Bulger.  1977.
Suppression of testosterone production by ethyl alcohol.
Possible mode of action.  Steroids 30:  647-655.

Bandas, E.L.  1982.  Study of the role of metabolites and
impurities in the mutagenic action of ethanol on yeast
mitochondria.  Genetika 18:  1056-1061.

Bannigan, J., and P. Burke.  1982.  Ethanol teratogenicity in
mice:  a light microscope study.  Teratol. 26:  247-254.

Barilyak, I.R. and S.Y. Kozachuk.  1983.  Embryotoxic and
mutagenic activity of ethanol after intra-amniotic injection.
Tsitologiya i Genetika 17:  57-60.

Barrow, C.S.  1982.  Sensory irritation of inhaled aldehydes:
structure activity studies.  CUT 2, no. 9.


Bird, R.P.,  H.H. Draper, and P.K. Basrur.  1982.  Effect of
malonaldehyde and acetaldehyde on cultured mammalian cells;
production of micronuclei and chromosomal aberrations.  Mutat.
Res. 101:  237-246.
                               1-37

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Bittersohl, G.  1974.  Epidemiologic investigations on cancer
incidence in workers contacted by acetaldol and other aliphatic
aldehydes.  Arch. Geschwulstforsch.  43:  172-176.

Blakley, P.M. and W.J. Scott.  1984a.  Determination of the
proximate teratogen of the mouse fetal alcohol syndrome.  1.
Teratogenicity of ethanol and acetaldehyde.  Tox. Appl.
Pharmacol. 72:  355-363.

Blakley, P.M. and W.J. Scott.  1984b.  Determination of the
proximate teratogen of the mouse fetal alcohol syndrome.  2.
Pharmacokinetics of the placental transfer of ethanol and
acetaldehyde.  Tox. Appl. Pharmacol. 72:  364-371.

Boettner, E.A., G.L. Ball, and B. Weise.  1973.  Combustion
products from the incineration of plastics.  Ann Arbor, MI:
University of Michigan, School of Public Health.  Available from:
NTIS, Springfield, VA; PB-222001/0.

Bohlke, J.U., S. Singh, and H.W. Goedde.  1983.  Cytogenetic
effects of acetaldehyde in lymphocytes of Germans and Japanese:
SCE, clastogenic activity, and cell cycle delay.  Hum. Genet. 63:
285-289.

Boyden, T.W., M.A. Silvery, and R.W. Pamenter.  1981.
Acetaldehyde acutely impairs canine testicular testosterone
secretion.  Eur. J. Pharmacol. 70:  571-576.

Bradley, M., I.C. Hsu, and C.C. Harris.  1979.  Relationships
between sister chromatid exchange and mutagenicity, toxicity, and
DNA damage.  Nature 282:  318-320.

Braven, J., G.J. Bonker, M.L. Fenner, and B.L. Tonge.  1967.  The
mechanism of carcinogenesis by tobacco smoke.  Some experimental
observations and a hypothesis.  Br. J. Cancer 21:  623-633.

Campbell, M.A., and A.G. Fantel.  1983.  Teratogenicity of
acetaldehyde in vitro:  relevance to fetal alcohol syndrome.
Life Sci. 32:  2641-2647.

CARB.  1992a.  Preliminary Draft:  Proposed identification of
acetaldehyde as a toxic air contaminant.  Part A  Exposure
assessment.  California Air Resources Board, Stationary Source
Division.  August, 1992.
CARB.  1992b.  Preliminary Draft:  Proposed identification of
acetaldehyde as a toxic air contaminant.  Part B Health
assessment.  California Air Resources Board, Stationary Source
Division.  August, 1992.

Checiu, M. S. Sandor, and Z. Garban.  1984.  The effect of
ethanol on early development in mice and rats.  VI.  In vitro
effect of acetaldehyde upon preimplantation stages in the rat.
Morphol.  Embryol.  (Bucur) (Romania) 30:  175-184.


                               8-38

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Cicero, T.J., and R.D. Bell.  1980.  Effects of ethanol and
acetaldehyde on the biosynthesis of testosterone in the rodent
testes.  Biochem. Biophys. Res. Commun. 94:  814-819.

Cicero, T.J., R.D. Bell, and T.M. Badger.  1980a.  Acetaldehyde
directly inhibits the conversion of androstenedione to
testosterone in the testes.  Adv. Exp. Med. Biol. 132:  211-217.

Cicero, T.J., R.D. Bell, E.R. Meyer, and T.M. Badger.  1980b.
Ethanol and acetaldehyde directly inhibit testicular
steroidogenesis.  J. Pharmacol. Exp. Therap. 213:  228-233.

Cobb, C.F., M.F. Ennis, F. Michael, D.H. Van Thiel, J.S. Gavaler,
and R. Lester.  1978.  Acetaldehyde and ethanol are direct
testicular toxins.  Surg.  Forum 29:  641-644.

Cobb, C.F., M.F. Ennis, D.H. Van Thiel, J.S. Gavaler, and R.
Lester.  1980.  Isolated testes perfusion:  a method using a
cell- and protein-free perfusate useful for the evaluation of
potential drug injury.  Metabolism 29:71-79.

Commoner, B.  1976.  Reliability of bacterial mutagenesis
techniques to distinguish carcinogenic chemicals.  EPA-600/1-76-
022.  U.S. Environmental Protection Agency.

Delaney J.L. and T.W. Hughs.  1979.  Source assessment:
manufacture of acetone and phenol from cumene.   Research Triangle
Park, NC:  U.S. Environmental Protection Agency, Industrial
Environmental Research Laboratory:  EPA Report No.  EPA-600/2-79-
019d.

de Raat, W.K., P.B. Davis, and G.L. Bakker.  1983.  Induction of
sister-chromatid exchanges by alcohol and alcoholic beverages
after metabolic activation by rat-liver homogenate.  Mutat. Res.
167:  89-105.

Dreisbach, R.H.  1980.  Acetaldehyde, metaldehyde, and
paraldehyde.  In:  Handbook of poisoning:  prevention, diagnosis,
and treatment.  Los Angeles, CA:  Lange Medical Publications; pp.
177-180.
Dreosti, I.E., F.J. Ballard, G.B. Belling, I.R. Record, S.J.
Manuel, and B.S. Hetzel.  1981.  The effect of ethanol and
acetaldehyde an DNA synthesis in growing cells and on fetal
development in the rat.  Alcohol:  Clin. Exp. Res. 5:  357-362.

Dulout F. and T. Furnus.  1988.  Acetaldehyde induced aneuploidy
in cultured Chinese hamster cells.  Mutagenesis 3:  207-211.

Egle, J.L., Jr.  1972.  Effects of inhaled acetaldehyde and
propionaldehyde on blood pressure and heart rate.  Toxicol, Appl.
Pharmacol. 23:  131-135.
                               8-39

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Egle, J.L., P.M. Hudgins, and F.M. Lai.  1973.  Cardiovascular
effects of intravenous acetaldehyde and propionaldehyde in the
anesthetized rat.  Toxicol.  Appl. Pharmacol. 24:  636-644.

Eimutis, E.G., R.P. Quill, and G.M. Rinaldi.  1978.  Source
assessment:  noncriteria pollutant emissions.  Research Triangle
Park, NC:   U.S. Environmental Protection Agency, Industrial
Environmental Research Laboratory; EPA Report No. EPA-600/2-78-
004t.

Eker, P. and T. Sanner.  1986.   Initiation of in vitro cell
transformation by formaldehyde and acetaldehyde as measured by
attachment-independent survival  of cells in aggregates.  Eur. J.
Cancer Clin. Oncol. 22(6):  671-676.

EPA.  1987.  Health assessment document for acetaldehyde.  Office
of Research and Development, Office of Health and Environmental
Assessment, Environmental Criteria and Assessment Office,
Research Triangle Park, NC.   EPA-600/8-86/015A.  (External Review
Draft) .

EPA.  1992.  Integrated Risk Information System  (IRIS).  Office
of Health and Environmental Assessment, Environmental Criteria
and Assessment Office, Cincinnati, OH.

Fenner,  M.L. and J. Braven.   1968.  The mechanism of
carcinogenesis by tobacco smoke.  Further experimental evidence
and a prediction from the thiol-defense hypothesis.  Br. J.
Cancer 22:   474-479.

Feron, V.J.  1979.  Effects of exposure to acetaldehyde in Syrian
hamsters simultaneously treated  with benzo(a)pyrene or
diethylnitrosamine.  Prog. Exp.  Tumor Res. 24:  162-176.

Feron, V.J., A. Kruysse, and R.A. Woutersen.  1982.  Respiratory
tract tumors in hamsters exposed to acetaldehyde vapour alone or
simultaneously to benzo(a)pyrene or diethylnitrosamine.  Eur. J.
Cancer Clin. Oncol. 18:  13-31.

Fisher,  S.E., M. Atkinson, I. Holzman, R. David, and D.H. Van
Thiel.  1981a.  Effects of ethanol upon placental uptake of amino
acids.  Prog. Biochem. Pharmacol. 18:  216-223.

Fisher,  S.E., M. Atkinson, D.H.  Van Thiel, E. Rosenblum, R.
David, and I. Holzman.  1981b.   Selective fetal malnutrition:
effects of ethanol and acetaldehyde upon in vitro uptake of alpha
amino isobutyric acid by human placenta.  Life Sci. 29:  1283-
1288.

Fisher,  S.E., M. Atkinson, and D.H. Van Thiel.  1984.  Selective
fetal malnutrition:  the effect  of nicotine, ethanol, and
acetaldehyde upon in vitro uptake of a alpha-aminoisobutyric acid
by human placenta.  Dev. Pharmacol. Ther. 7:  229-238.

Garberg, P., E-L Akerblom, and G. Bolcsfoldi.  1988.  Evaluation
of a genotoxicity test measuring DNA strand breaks in mouse


                               8-40

-------
                                                        EPA-420-R-93-005
                                                            April 1993

lymphoma cells by alkaline unwinding and hydroxyapatite elution.
Mutat. Res 203:  155-176.

Greenwald, I.S., and H.R. Horvitz.  1980.  unc-93  (elSOO):  a
behavioral mutant of Caenorhabditis elegans that defines  a gene
with a wild-type null phenotype.  Genetics 96:  147-164.

He, S.M. and B. Lambert.  1985.   Induction and persistence of
SCE-inducing damage in human lymphocytes exposed to vinyl acetate
and acetaldehyde in vitro.  Mutat.  Res. 158:  201-208.

Hemminki, K.,  K. Falck, H. Vainio.  1980.  Comparison of
alkylation rates and mutagenicity of directly acting industrial
and laboratory chemicals.  Arch. Toxicol. 46:  277-285.

Henderson G.I., D. Turner, R.V. Patwardhan, L. Lumeng, A.M.
Hoyumpa, and S. Schenker.  1981.  Inhibition of placental valine
uptake after acute and chronic maternal ethanol consumption.  J.
Pharmacol. Exper. Ther. 216:  465-472.

Henderson G.I., R.V. Patwardhan, S. McLeroy, and S. Schenker.
1982.  Inhibition of placental amino acid uptake in rats
following acute and chronic ethanol exposure.  Alcoholism 6:
495-500.

Higuchi, Y., and N. Matsumoto.  1984.  Embryotoxicity of  ethanol
and acetaldehyde: direct effects on mouse embryo in vitro.
Congen. Anom.   (Senten IJO) 24:  9-28.

Hilaro et al.   1985.

IARC.  1987.  IARC Monographs on the Evaluation of Carcinogenic
Risk of Chemicals to Humans.  Supplement 7.  Overall Evaluations
of Carcinogenicity:  An updating of IARC monographs volumes 1 to
42.  International Agency for Research on Cancer.  World  Health
Organization,  Lyon, France.

Jacob, D. J.,  E. W. Gottlieb, and M. J. Prather.  1989.
Chemistry of a polluted boundary layer.  J. Geophys. Res.,
94:12975-13002.

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Jansson, T.  1982.  The frequency of sister chromatid exchanges
in human lymphocytes treated with ethanol and acetaldehyde.
Hereditas 97:  301-303.

Johnson, D.E.,  Y.-B. Chiao, J.S. Gavaler, and D.H. Van Thiel.
1981.  Inhibition of testosterone synthesis by ethanol and
acetaldehyde.  Biochem. Pharmacol.  36:  1827-1831.

Kalmus G. and C. Buckenmaier.  1989.  Effects of ethanol and
acetaldehyde on cultured pre-implantation mouse embryos.
Experientia 45:  484-487.

Kane, L.E., R.  Dombroske, and Y. Alarie.  1980.  Evaluation of
sensory irritation from some common industrial solvents.  Am.
Ind. Hyg. Assoc. J. 41:  451-455.

Korte, A. and G. Obe.   1981.  Influence of chronic ethanol uptake
and acute acetaldehyde treatment on the chromosomes of bone-
marrow cells and peripheral lymphocytes of Chinese hamsters.
Mutat. Res. 88:  389-395.

Kruysse, A., V.J. Feron, and H.P. Til.  1975.  Repeated exposure
to acetaldehyde Vapor.  Arch. Environ. Health.  30:  449-452.

Lahdetie, J.  1988.  Effects of vinyl acetate and acetaldehyde on
sperm morphology and meiotic micronuclei in mice.  Mutat. Res.
202:  171-178.

Lam, C.W.,  M. Casanova, and H. D'A. Heck.  1986.  Decreased
extractability of DNA from proteins in the rat nasal mucosa after
acetaldehyde exposure.  Fund. Appl. Toxicol.  6:  541-550.

Lambert, B., Y. Chen,  S.M. He, and M. Sten.  1985.  DNA cross-
links in human leucocytes treated with vinyl acetate and
acetaldehyde in vitro.  Mutat. Res. 146:  301-303.

Laumbach, A.D., S. Lee, J. Wong, U.N. Streips.  1976.  Studies on
the mutagenicity of vinyl chloride metabolites and related
chemicals.   Proceedings of the 3rd International Symposium on the
Prevention and Detection of Cancer, Vol. 1, pp. 155-169.

Lewis. R.J., and R.L.  Tatkin.  1983.  Registry of toxic effects
of chemical substances.  1981-1982 ed. Cincinnati, OH:  U.S.
Department of Health and Human Services, National Institute for
Occupational Safety and Health; p.  156; DHHS (NIOSH) publ. no.
83-107.

Ligocki, M.P.,  R.R. Schulhof, R.E.  Jackson, M.M. Jimenez, G. Z.
Whitten, G.M. Wilson,  T.C. Meyers,  and J.L. Fieber.  1992.
Modeling the effects of reformulated gasolines on ozone and
toxics concentrations in the Baltimore and Houston areas.
Systems Applications International, San Rafael, California
(SYSAPP-92/127).

Ligocki, M.P.,  G.Z. Whitten, R.R. Schulhof, M.C. Causley, and
G.M. Smylie.  1991.  Atmospheric transformation of air toxics:


                               8-42

-------
                                                        EPA-420-R-93-005
                                                            April 1993

benzene, 1,3-butadiene, and formaldehyde.  Systems Applications
International, San Rafael, California  (SYSAPP-91/106).

Ligocki, M.P. and G.Z. Whitten.  1991.  Atmospheric
transformation of air toxics:  acetaldehyde and polycyclic
organic matter.  Systems Applications  International, San Rafael,
California (SYSAPP-91/113).

Mannsville Chemical Products Corporation.  1984.  Chemical
products synopsis.  Acetic acid.  Cortland, NY:  Mannsville
Chemical Products Corporation.

Marnett, L.J., H.K. Kurd, M.C. Hollstein, D.E. Levin, H.
Esterbauer, and B.N. Ames.  1985.  Naturally occurring carbonyl
compounds are mutagenic in Salmonella  tester strain TA104.
Mutat. Res. 148:  381-382.

Mohan, M., U. Rai, L.P. Reddy, C.V. Prasanna, and S.
Ramakrishnan.  1981.  Cardiovascular effects of acetaldehyde in
guinea pigs.   Indian J. Physiol. Pharmacol. 25:  246.

Mortelmans, K. S. Haworth, T. Lawlor,  W. Speck, B. Tainer, and E.
Zeiger.  1986.  Salmonella mutagenicity tests:  II.  Results from
testing of 270 chemicals.  Environ. Mutagen.  8:  1-39.

Norppa, H., F. Tursi, P. Pfaffli, J. Maki-Paakkanen, and H.
Jarventaus.  1985.  Chromosome damage  induced by vinyl acetate
through in vitro formation of acetaldehyde in human lymphocytes
and Chinese hamster ovary cells.  Cancer Res. 45:  4816-4821.

National Research Council.  1977.  Drinking water and health:
part II, chapters 6 and 7.  Washington, B.C.:  National Academy
of Sciences;  pp. VI-219-VI-220. Available from:  NTIS,
Springfield,  VA; PB-270423.

National Research Council.  1981.  Formaldehyde and other
aldehydes.  Washington, B.C.:  National Academy Press.

Obe, G. and B. Beer. 1979.  Mutagenicity of aldehydes.  Brug and
Alcohol Bependence 4:  91-94.

Obe, G. and H. Ristow.  1977.  Acetaldehyde, but no ethanol,
induces sister chromatid exchanges in  Chinese hamster cells in
vitro.  Mutat. Res. 56:  211-213.

Obe, G., R. Jonas, and S. Schmidt.  1986.  Metabolism of ethanol
in vitro produces a compound which induces sister-chromatid
exchanges in human peripheral lymphocytes in vitro:
acetaldehyde, not ethanol, is mutagenic.  Mutat. Res. 174:  47-
51.

Obe, G., A.T. Natarajan, M.  Meyers, and A. Ben Hertog.  1979.
Induction of chromosomal aberrations in peripheral lymphocytes of
human blood in vitro, and of SCEs in bone-marrow cells of mice in
vitro by ethanol and its metabolite acetaldehyde.  Mutat. Res.
68:  291-294.


                               8-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Olson, T.M. and M.R. Hoffmann, 1989.  Hydroxyalkylsulfonate
formation:  its role as a S(IV) reservoir in atmospheric water
droplets.  Atmos.  Environ., 23:  985-997.

Omel'yantes, N.I., N.V. Mironets, N.V. Martyshchenko, I.A.
Gubareva, L.F. Piven, and S.N. Starchenko.  1978.  Experimental
substantiation of the maximum permissible concentrations of
acetone and acetaldehyde in reclaimed potable water.  Biol.
Aviakosm Med. 1978:  67.

O'Shea, K.S. and M.H. Kaufman.  1979.  Teratogenic effects of
acetaldehyde:  implications for the study of fetal alcohol
syndrome.  J. Anat. 128:  65-76.

O'Shea, K.S. and M.H. Kaufman.  1981.  Effects of acetaldehyde on
neuroepithelium of early mouse embryos.  J. Anat. 132:  107-118.

Padmanabhan, R., R.N. Sreenathan, and S. Singh.  1983.  Studies
of the lethal and teratogenic effects of acetaldehyde in the rat.
Conge. Anom.  (Senten IJO) 23: 13-23.

Perry, R.H. and Chilton, C.H.  1973.  Chemical Engineer's
Handbook, Fifth Edition.  McGraw-Hill Book Company.

Pool, B.L., and M. Wiessler.  1981.  Investigations on the
mutagenicity of primary and secondary a-acetoxynitrosamines with
Salmonella typhimurium:  activation and deactivation of
structurally related compounds by S-9.  Carcinogenesis 2:  991-
997.

Popov, V.B., B.L.  Vaisman, V.P. Puchkov, and T.V. Ignat'eva.
1982.  Toxic effects of ethanol and its biotransformation
products on post-implantation embryos in culture.  Bull. Exp.
Biol. Med.  (Rus.)  92:  1707-1710.

Prasanna, C.V., S. Ramakrishnan, B. Krishnan, and A. Srinivasa
Rao.  1981.  Effect of acetaldehyde on lung surfactant.  Indian
J. Exp. Biol. 19:   580-581.

Prescott, P.K.  1985.  The effect of acetaldehyde and 2,3-
butanediol on rat embryos developing in vitro.  Biochem.
Pharmacol. 34:  529-532.

Rieger, R., and A. Michaelis.  1960.  Chromosomenaberrationen
nach Einwirkung von Acetaldehyd auf Primawurzeln von Vicia faba.
Biol. Zbl. 79:  1-5.

Ristow, H. and G.  Obe.   1979.  Acetaldehyde induces cross-links
in DNA and causes sister-chromatid exchanges in human cells.
Mutat. Res. 58:  115-119.

Saladino, A.J., J.C. Willey, J.F. Lechner, R.C. Grafstrom, M.
LaVeck, and C.C. Harris  1985.  Effects of formaldehyde,
acetaldehyde, benzoyl peroxide, and hydrogen peroxide on cultured
normal human bronchial epithelial cells.  Cancer Res. 45:  2522-
2526.


                               8-44

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Santucci, L.,  T.J. Graham, and D.H. Van Thiel.  1983.  Inhibition
of testosterone production by rat Leydig cells with ethanol and
acetaldehyde:  prevention of ethanol toxicity with 4-
methylpyrazole.   Alcoholism 7:  135-139.

Shikiya, B.C., C.S. Liu, M.I. Kahn, J. Juarros, and W.
Barcikowski.   1989.  In-vehicle air toxics characterization study
in the south coast air basin.  South Coast Air Quality Management
District, El Monte, CA.  May, 1989.
Sim, V.M., and R.E. Prattle.
irritants on human subjects.
1913.
                   1957.  Effect of possible smog
                   JAMA J. Am. Med. ASSOC. 165:  1908-
Sina, J.F., C.L. Bean, G.R. Dysart, V.I. Taylor, and M.O.
Bradley.  1983.  Evaluation of the alkaline elution/rat
hepatocyte assay as a predictor of carcinogenic/mutagenic
potential.  Mutat.  Res. 113:  357-391.
Sittig, M.
chemicals.
1979.   Haxardous and toxic effects of industrial
 Park Ridge,  NJ:  Noyes Data Corporation; pp. 1-2.
Skog, E.  1950.  A toxicological investigation of lower aliphatic
aldehydes.  I.  toxicity of formaldehyde, acetaldehyde,
propionaldehyde butyraldehyde; as well as acrolein and
crotonaldehyde.  Acta Pharmacol. Toxicol. 6:  299-318.

Sreenathan,  R.N. and R. Padmanabhan.  1984.  Structural changes
of placenta following maternal administration of acetaldehyde in
the rat.  Z. Mikrosk. Anat.  Forsch.  (E.Germ.) 98:  597-600.
Sreenathan, R.N., R. Padmanabhan, and S. Singh.
Teratogenic effects of acetaldehyde in the rat.
Depend. 9:  339-350.
                                      1982.
                                      Drug Alcohol
Sreenathan, R.N., S. Singh, and R. Padmanabhan.  1984a.
Implications of the placenta in the acetaldehyde-induced intra-
uterine growth retardation.  Drug Alcohol Depend. 13:  199-204.

Sreenathan, R.N., S. Singh, and R. Padmanabhan.  1984b.  Effects
of acetaldehyde on skeletogenesis in the rat.  Drug Alcohol
Depend. 14:  165-174.

Thompson,  P.A.C., and P.I. Folb.   1982.  An in vitro model of
alcohol and acetaldehyde teratogenicity.   J. Appl. Toxicol. 2:
190-195.

Veghelyi,  P.V., M. Osztovics, G.  Kardos,  L. Leisztner, E.
Szaszovszky, S. Igali, and J. Imrei.   1978.  The fetal alcohol
syndrome:   symptoms and pathogenesis.   Acta Paediatria Academiae
Scientiarum Hungaricae 19:  171-189.

Versar Inc.  1975.  Identification of Organic Compounds in
Effluents From Industrial Sources.
                               1-45

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Watanabe, F. and S. Sugimoto.  1956.  Study on the
carcinogenicity of aldehyde, 3rd report:  four cases of sarcomas
of rats appearing in the areas of repeated subcutaneous
injections of acetaldehyde.   Gann 47:  599-601.

Webster, W.S., D.A. Walsh, S.E. McEwen, and A.H. Lipson.  1983.
Some teratogenic properties of ethanol and acetaldehyde in
C57BL/6J mice; implications for the study of fetal alcohol
syndrome.  Teratol. 27:  231-243.

Wilson, A. et al.   1991.  Air toxics microenvironment exposure
and monitoring study.  South Coast Air Quality Management
District, El Monte, CA.

Windholz, M., S. Budavari, R.F. Blumetti, and E.S. Otterbein,
eds.   1983.  Acetaldehyde.  In:  The Merck Index:  an
encyclopedia of chemicals, drugs, and biologicals.  10th ed.
Rahway, NJ:  Merck and Co.,  Inc.; p. 30.

Woodruff, R.C., J.M. Mason,  R. Valencia, and S Zimering.  1985.
Chemical mutagenesis testing in Drosophila.  V.  Results of 53
coded compounds tested for the National Toxicology Program.
Environ. Mutagen.  7:  677-702.

Woutersen, R.A., and Appelman L.M.  1984.  Lifespan inhalation
carcinogenicity study of acetaldehyde in rats.  III. Recovery
after 52 weeks of exposure.   Report No. V84.288/190172.  CIVO-
Institutes TNO, The Netherlands.

Woutersen, R.A. Van Garderen-Hoetmer, A., and Appelman, L.M.
1985.  Lifespan (27 months)  inflation carcinogenicity study of
acetaldehyde in rats.  Report No. V85.145/190172.  CIVO-
Institutes TNO, The Netherlands.

Woutersen, R.A., L.M. Appelman, A. Van Garderen-Hoetmer, and V.J.
Feron.  1986.  Inhalation toxicity of Acetaldehyde in rats.  III.
Carcinogenicity study.  Toxicology.  41:  213-231.
                               8-46

-------
                                                         EPA-420-R-93-005
                                                             April 1993


Woutersen, R.A., and V.J.  Feron.   1987.   Inhalation toxicity of
acetaldehyde in rats.   IV.   Progression and regression of nasal
lesions after discontinuation of  exposure.   Toxicology.  47:
295-305.

Zorzano, A. and E, Herrera.   1989.   Disposition of ethanol and
acetaldehyde in late pregnant rats  and their fetuses.  Pediat.
Res. 25:  102-106.
                                !-47

-------
                                                        EPA-420-R-93-005
                                                            April 1993

9.0  DIESEL PARTICULATE MATTER

9.1  Chemical and Physical Properties

     Diesel exhaust particulate matter consists of a solid core
composed mainly of carbon, a soluble organic fraction, sulfates,
and trace elements.  When comparing the size distribution of
diesel particles to gasoline particles the majority of the diesel
particles range from 0.1 to 1.0 urn with a peak at around 0.15 urn,
while the gasoline particles range from 0.01 to 0.1 urn with the
peak at around 0.02 (NRC, 1982).   When a particle is less than 1
micron (urn) in diameter it is small enough to be inhaled deeply
into the lungs.  Although the gasoline particles are smaller, the
light-duty diesel engines emit from 30 to 100 times more
particles than comparable catalyst-equipped gasoline vehicles
(NRC, 1982).   At temperatures above 500C the particles
themselves are actually solid chain aggregates of carbon-hydrogen
spheres with diameters ranging from 100 to 800 angstroms (A).
These are mainly attributed to the incomplete combustion of fuel
hydrocarbons, though some may be due to engine oil or other fuel
components.  Photomicrographs show that diesel particles have a
very light, fluffy structure, with a density of about 0.07 g/m3
(NRC, 1982) .

     At temperatures below 500C,  the particles become coated
with adsorbed and condensed high molecular weight organic
compounds.  Typically, about 25 percent of the particle consists
of extractable organics, although different vehicles may have
extractable fractions of 5-90 percent, depending on operating
conditions.  These compounds include open-chain hydrocarbons of
14-35 carbon atoms, alkyl-substituted benzenes, and derivatives
of the polycyclic aromatic hydrocarbons (PAH), such as ketones,
carboxyaldehydes, acid anhydrides, hydroxy compounds, quinones,
nitrates, and carboxylic acids (Johnson, 1988) .  There are also
heterocyclic compounds containing sulfur,  nitrogen, and oxygen
atoms within the aromatic ring.  Inorganic compounds also are
present and include sulfur dioxide, nitrogen dioxide, and
sulfuric acid  (NRC, 1982).

     To best describe the diesel particle content adequately, the
temperature at the time of the sample collection and the means by
which that temperature was reached must be specified.  Diesel
particulate matter is generally defined as any material that is
collected, at a temperature of 52C or less,  on a filtering
medium after dilution of the raw exhaust gases (NRC, 1982).
Water that condenses on the filter is not considered to be diesel
particulate matter.


9.2 Formation and Control Technology

     The chemical mechanism which accounts for carbon formation
by diesel engines is not completely established (EPA, 1990b); the
major weight of scientific opinion seems to support some role for
intermediate formation of polycyclic aromatic matter  (POM)  in the
process.   Carbon is a stable combustion product normally of rich

                               9-1

-------
                                                        EPA-420-R-93-005
                                                            April 1993

flames; carbon formation normally takes place over a rather
narrow temperature range.  A significant fraction of the diesel
particulate matter consists of oil-derived hydrocarbons and
related solid matter.  The formation of the carbon particle is
thought to involve polymerization of gaseous intermediates at the
surface of the smaller particles.  Growth and agglomeration of
the carbon particles are probably gas-to-particle processes.  The
POM's that are produced in the combustion process are adsorbed
onto the surface of the carbon particle.  Several of these, such
as benzo[a]pyrene  (B[a]P) and 1-nitropyrene, are known or
potential human carcinogens.  Recent data have indicated that the
particles themselves may have intrinsic toxic and carcinogenic
properties.

     Studies of diesel particle composition have produced some
information about the fate of fuel sulfur (EPA, 1990b).   Sulfate
has been found to be a significant component of diesel particles.
Generally, the sulfate found in particles accounts for only about
2% of the fuel sulfur, the balance being emitted as sulfur
dioxide.  At present, no means of reducing sulfate formation is
available other than reducing the sulfur concentration of diesel
fuel.

     EPA's Five City Study  (EPA, 1989) determined that POM
contributed to 27% of the average excess aggregate cancer
incidence in the five cities.  Of this 27%,  diesel particulate
matter was the major contributor, accounting for 45% of the total
POM.

     The control of diesel emissions can take two forms.  The
first is controlling emissions before they are formed with either
engine modifications or aftertreatment systems to the exhaust
system.  Each of these takes many forms and are in various stages
of development (EPA, 1990b).

     One way to modify the engine is by refining the combustion
process and many different modifications are in use.  Many new
diesel engines being made today are going from the indirect to
the direct injection engine.  These direct injection engines are
low-emitting and fuel efficient.  Also being considered are
changes to the combustion chamber design to decrease emissions.

     At this time, most diesel engines still rely on mechanical
engine control systems.  On newer engines, there is now expanded
use of computerized electronic control systems that increase the
potential flexibility in controlling emissions.

     Another technology used to control emissions is the combined
technology of turbocharging and intercooling.   Most heavy duty
diesel engines have them and were required for virtually all
engines in 1991.   The turbocharger increases the air mass in the
cylinder and the intercooler reduces the temperature.  This
system is successful in reducing both NOX and  diesel particle
emissions as well as increasing fuel economy and power output.
Also being considered for use on some heavy duty engines is
                               9-2

-------
                                                        EPA-420-R-93-005
                                                            April 1993

intake manifold tuning.  At this time, it is being used on high-
performance cars to enhance the airflow.

     Control of the lubricating oil is important to diesel
particulate matter reduction since 10 to 50% of the particulates
being formed are from engine oil.   Oil consumption can be reduced
primarily by improved engine manufacturing specifications and
engine seals.

     A second way to control emissions is to add aftertreatment
technologies to the exhaust system.  A trap oxidizer is being
considered.  This is located in the exhaust system to trap the
particulate matter and provide some means of cleaning the filter
by burning the collected particulate matter.  Passive system
traps, traps that attain the proper conditions for regeneration
during normal operation, require the use of a catalyst in most
cases.  Some catalysts being considered are platinum, palladium,
rhodium,  silver, vanadium, and copper.  Cerium has also been
considered as a fuel additive to be used with the catalyst.
There is still development to be done in this area.

     Catalytic converters are another technology being evaluated
along with fuel modifications.  By reducing the sulfur  (now at a
maximum of 0.05% by weight) and aromatic hydrocarbon content,
emissions of diesel particles and POM can be reduced.  Fuel
additives are also being tested.

     Alternative fuels are being researched for use in diesel
engines.   The fuels being tested at this time are natural gas,
methanol, and liquified petroleum gas.


9.3  Emissions

9.3.1  Diesel Particulate Matter Emission Standards

     Diesel particulate matter emission standards for light duty
vehicles (LDVs), light duty trucks (LDTs),  and HDDEs are
summarized in Table 9-1.  The LDV and LDT categories include both
gasoline and diesel powered vehicles.

9.3.2  Methodology

     As mentioned in section 3.2,  the urban diesel particulate
matter national fleet average emission factors derived by
Sienicki (1992a, 1992b) are used for this analysis.  The general
methodology used by Sienicki to calculate urban diesel
particulate matter is summarized in the appendix of MVMA and EMA
(1986).   All input data
                               9-3

-------
                                                                                EPA-420-R-93-005
                                                                                      April 1993
    Table 9-1.    Diesel  Particulate Matter Emission Standards.
Year
1982-
1986
1987
1988-
1990
1991
1992
1993
1994
1995 +
LDVa
(gpm)
0.60
0.20
0.20
0.20
0.20
0.20
0.08/
0.10d
0.08/
0.10d
LDTlbc
(gpm)
0.60
0.26
0.26
0.26
0.26
0.26
0.26
0.08/
0.10d
LDT2bc
(gpm)
0.60
0.26
0.26
0.26
0.26
0.26
0.26
0.08/
0.10d
LDT3bc
(gpm)
0.60
0.26
0.26
0.26
0.26
0.26
0.10
0.10
LDT4bc
(gpm)
0.60
0.26
0.26
0.26
0.26
0.26
0.12
0.12
HDDE
Urban bus
(g/bhp-
hr)
None
None
0.60
0.25
0.25
0.10
0.05
0.05
HDDE
Other
diesels
(g/bhp-
hr)
None
None
0.60
0.25
0.25
0.25
0.10
0.10
a!994  standards are phased  in over three years:  40% MY 1994,

b!995  standards are phased  in over three years:  40% MY 1995,
MY 1995,  100% MY 1996 and after.

MY 1996,  100% MY 1997 and after.
Light light-duty trucks  consist of weight categories LDT1 and LDT2,  and are less than or equal
to 6000  Ibs. gross vehicle weight  rating  (GVWR).   Heavy light-duty trucks consist of weight
categories LDT3 and LDT4, and are  greater than 6000 Ibs GVWR.  LDT1 = light light-duty trucks up
to 3750  Ibs  loaded vehicle weight  (LVW).  LDT2 =  light  light-duty trucks greater than 3750 Ib
LVW.   LDT3 = heavy light-duty trucks up through 5750 Ibs adjusted loaded vehicle weight  (ALVW).
LDT4  = heavy light-duty trucks greater than 5750  Ibs ALVW.   LVW = curb weight  (nominal vehicle
weight)  plus 300 Ibs.  ALVW = numerical average of curb weight and GVWR.

aThe  first number is  the  5  year/50,000 mile standard; the second number  is  the 10 year/100,000
mile  standard.
                                             9-4

-------
                                                        EPA-420-R-93-005
                                                            April 1993

and values calculated in each step of Sienicki's analysis  are
contained in Appendix G.

     Urban diesel particulate matter emissions can  simply  be
considered as the product of urban diesel vehicle miles  travelled
and the diesel particle emission rate:

                       (1)   DP = UVMTd  x ER

where:         DP = Urban Diesel Particulate Matter Emissions  (g)
               UVMTd = Urban Diesel Vehicle  Miles Travelled (mi)
               ER = Diesel Particulate  Matter Emission Rate
(g/mi).

DP is calculated separately for each vehicle class  by model year
for the 20 most recent model years; these values are then  added
to obtain overall urban diesel particulate matter for the  year of
interest.  However, to calculate UVMTd and ER for individual
classes in a model year, a series of steps,  which will be
detailed in the following sections, must first be employed.

     An overall national fleet emission factor  (EF) in grams per
mile can be calculated by dividing the  total DP, after applying a
freeway road use adjustment  (described  later), by total  UVMT for
both gas and diesel vehicles:

2)  EF = Total DP/Total UVMT

9.3.2.1  Calculation of Urban Diesel Vehicle Miles  Travelled

UVMTd is  determined  as  the product  of  fleet  VMT,  diesel mile
fraction (DMF),  and the diesel urban fraction (DUF):

                   (3) UVMTd  =  VMT  x DMF x DUF.

The DMF  is  the  ratio of diesel miles  travelled divided by total
miles  travelled.     It  can  be  calculated   using   the   following
equation:

            (4) DMF = DSF/{(DSF +  (1  -  DSF) (VMTg/VMTd) }

where:         DSF = Diesel Sales Fraction
               VMTg = gasoline  annual  vehicle miles travelled
               VMTd = diesel  annual vehicle  miles travelled.

DSF is obtained by dividing diesel market shares by 100  (listed
for each vehicle class by model year in Appendix G).  Sienicki
based the diesel market shares he used  on industry  opinion.  The
industry opinion he used predicts lower LDDV and LDDT sales than
MOBILE4  (EPA, 1988).  Sienicki used several sources to obtain  his
gasoline and diesel annual VMT rates.   For all vehicle classes
except buses, two sources were used --  the MVMA and EMA  analysis
(MVMA and EMA, 1986) and MOBILE4 (EPA,  1991a).  The annual VMT
rates for buses were based on data from the American Public
Transport Association  (APTA, 1990)  .
                               9-5

-------
                                                        EPA-420-R-93-005
                                                            April 1993
     Sienicki calculated fleet VMT by model year in each class
(VMTy)  as  the product of the VMT fraction for each model year in
a given class (VMTfy)  and the total VMT for each class  (TVMT) :

                      (5)  VMTy = VMTfy x TVMT

VMTfy is calculated by multiplying vehicle sales per vehicle
class for each model year  (VEHy)  by annual VMT for gas and diesel
vehicles  (VMTy)  and the survival rate (SRy) ,  then dividing  the
product by the sum of products for all model years:


                              VEH x  VMT  x SR
                 6)   VMTf,, =
                            20

                               (VEHy x VMTy x SRy)
                            y=l
Vehicle sales per vehicle class were estimated by first
establishing a base 20 year sales fleet from historical data, up
to the year 1990.  The sales fleet from the MVMA and EMA analysis
(MVMA and EMA, 1986) was updated using data mostly  from MVMA
(1991).   For estimation of diesel particulate matter in future
years (1995, 2000, 2010), Sienicki had to estimate  growth  in  the
size of the sales fleet for each class  (and hence,  growth  in
VMT).   This was done using fuel usage growth in the
transportation sector as a surrogate.  Fuel usage predictions
were obtained from the Department of Energy  (DOE, 1991).
Survival rates for all vehicles except buses were obtained from
MVMA and EMA  (1986).  Estimation of survival rates  for buses  are
described in Sienicki and Mago  (1991).  TVMT is the product of
vehicle sales per vehicle class for a 20 year period ending in
the target year, annual VMT, and annual SR.  Mathematically,  it
can essentially be expressed as the denominator in  Equation 6.

     Sienicki's final step in determining UVMT for  each class in
a given model year was to multiply VMT and DMF by the diesel
urban fraction  (DUF),  the fraction of miles travelled in urban
areas by diesel vehicles in a vehicle class.  Sienicki used the
same DUFs found in UMTRI (1988).

9.3.2.2  Calculation of Diesel Particulate Matter Emission Rate

     ERs for all classes prior to 1987 were obtained from  MVMA
and EMA (1986).  HDDE particulate matter emission rates for 1988
through 1991 were based on mean values for each class from
federal certification test results.  Model years 1992 and  1993
were assumed to be the same as 1991, and for 1994 and later
years, emission rates were set at the design target for the 0.10
g/bhp-hr standard at 0.084 g/bhp-hr.  LDDV and LDDT rates  were
set to EPA emission standards  (the 10 year/100,000  mile standard
for 1994 and later years).   Bus emission rates were assumed to be
the same as those for vehicles in class VIIIB until 1993,  when
                               9-6

-------
                                                        EPA-420-R-93-005
                                                            April 1993

they were set to a design target of 0.084 g/bhp-hr.  For 1994 and
later, they were assumed to be 0.06 g/bhp-hr.

     Conversion factors  (CFs) were used to convert heavy duty
emission rates from g/bhp-hr to grams per mile.  CFs describe the
average work per mile required for each vehicle class, and can be
calculated as the ratio of fuel density to the product of brake
specific fuel consumption (BSFC) and fuel economy  (MPG)  (EPA,
1988; MVMA,  1983):

                6)   CF = Fuel Density/(BSFC x MPG)

Conversion factors predicted for future years must take into
account any improvements in vehicle efficiency.  Although EPA's
MOBILE4 emissions model assumes no improvement in heavy duty
vehicle efficiency after 1986  (EPA, 1988),  Sienicki claims that a
number of factors,  such as increased market penetration of radial
tires and aerodynamic bodies, higher efficiency radial tires and
electronic emission control will continue to improve vehicle
efficiency.   He developed efficiency improvement factors based on
fuel economy improvement and market penetration analysis prepared
by industry market analysts for each vehicle class.  CFs for
future years were then calculated using previous years' CFs
divided by one plus the efficiency increase expected:

          7)  CFy+1 =  CFy/(l+ % efficiency increase/100)

EPA recently developed new conversion factors for heavy duty bus
engines (Kitchen and Damico, 1992) to more accurately  reflect the
effect that different types of bus operations have on  relative
levels of emissions of specific pollutants.  These new heavy duty
bus conversion factors are not used in this analysis.

     Sienicki also adjusted gram per mile emission rates for the
use of low sulfur fuel in 1991 and later years to obtain final
gram per mile emission rates for the 20 most recent model years
starting with the model year of interest.  Sienicki assumed a
0.025 g/bhp-hr reduction in particulate matter resulting from a
0.10 weight percent change in fuel sulfur,  based on results from
a recent study which addressed the effects of fuel composition on
diesel exhaust  (Ullman, 1989).  An earlier study (Ingham and
Warden, 1987) predicted particle reductions from fuel  sulfur that
were in the same range, although slightly lower.  Sienicki
assumed an average fuel sulfur level of 0.25 weight percent for
1990 and earlier years, reduced to a standard of 0.10 weight
percent in 1991-1993, and a standard of 0.05 weight percent in
1994 and later years  (EPA, 1990a).

9.3.2.3  Calculation of Urban Diesel Particulate Matter Emissions

     Once UMVT and ER have been calculated, DP for each model
year in a class can then be calculated using Equation  1.  The sum
of DPs for the 20 most recent model years in all classes can be
combined to yield a total DP estimate for the year of  interest.
After calculating a total DP for each year of interest, Sienicki
then applied a final adjustment to his total DP estimates to

                               9-7

-------
                                                        EPA-420-R-93-005
                                                            April 1993

account for freeway road use.  While the heavy duty transient
emission test cycle assumes 25% freeway operation, a recent
University of Michigan study  (1988) estimated that class VIIIB
vehicles accumulated 73% of their VMT on freeways when in large
urban areas.  Sienicki adjusted for this discrepancy using the
following equation:

   8)   DPAdj = DP x [(I  -  FMF)  x (4/3  -  1/3 x FFR)  + FMF x FFR]

where:         DPAdj = adjusted diesel particulate mass
               FMF = freeway mileage factor
               FFR = freeway factor ratio

FMF values were reported in the recent University of Michigan
study  (1988), and FFR is the ratio of freeway and non-freeway
emission rates in grams per mile.  A detailed explanation of this
equation, and the derivation of terms in this equation can be
found in Sienicki and Mago  (1991) .

9.3.2.4  Calculation of the Urban Diesel Particulate Matter
National Fleet Average Emission Factor

     The urban diesel particulate matter national fleet average
EF can be calculated using Equation 2.  Total UVMT in Equation 2
is the sum of UVMTd and  gasoline urban vehicle miles travelled
(UVMTg) :

                  9)  Total UVMT = UVMTg +  UVMTd

UVMTg is  calculated using  the following equation:

                10)  UVMTg  =  VMT x  (1  -  DMF)  x GUF

where:         GUF = gasoline urban fraction

Sienicki used the same GUFs found in MVMA  and EMA  (1986).  Urban
diesel particulate matter national fleet average EFs for 1990,
1995, 2000, and 2010 are 0.0573, 0.0305, 0.0160, and 0.009
g/mile, respectively.

     Sienicki's fleet average EFs can be compared to EFs derived
from information in a previous EPA air toxics report  (Carey,
1987).  Projected 1995 national fleet average EFs assuming low
and high diesel sales can be estimated using vehicle class EFs
and urban VMT fractions from this EPA report.  The low diesel
sales scenario in the 1987 EPA report assumed that diesel sales
remained constant at mid-1980's levels, while the high sales
scenario assumed an increase consistent with EPA projections at
that time.  The low diesel urban sales EF  was 0.0359 g/mi, and
the high diesel urban sales EF was 0.0507  g/mi.  The lower EF for
1995 predicted by Sienicki  (0.0305 g/mi) is partly due to
development of stricter standards than predicted in 1987, and
also to such factors as even lower light duty diesel vehicle
market shares than in either of the 1987 EPA report scenarios,
Sienicki1s low sulfur fuel and freeway road use adjustments, and
the smaller g/bhp-hr to g/mi conversion factors he used.


                               9-8

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     For this report, Sienicki's fleet average EFs without the
freeway road use adjustments will be used for the risk estimates.
This is consistent with past EPA practice.  However, further
investigation of the use of such freeway road use adjustments is
warranted.  Resulting urban diesel particulate matter national
fleet average EFs for 1990, 1995, 2000, and 2010 are summarized
in Table 9-2 .
    Table  9-2.
Urban Diesel Particulate Matter National Fleet
           Average EFs.
Year
1990
1995
2000
2010
EF (g/mi)
0.0669
0.0356
0.0188
0.0105
9.3.3  Nationwide Diesel Particulate Matter Emissions

     Sienicki's urban diesel particulate matter fleet average EFs
are based on the mix of vehicle classes expected in an urban
area.  Urban and rural VMT fractions differ, particularly for
some of the heavy duty vehicle classes where more rural use
occurs.  Since these heavy duty subclasses are responsible for
the majority of diesel particulate matter emissions, it would not
be appropriate to use urban fleet average EFs to calculate
nationwide diesel particulate matter emissions.  Instead, the EFs
by vehicle class calculated by Sienicki (without the freeway road
use adjustment) were combined using the nationwide VMT splits
from the MOBILE4.1 fuel consumption model (EPA, 1992) to estimate
nationwide fleet average EFs.  Using this approach, the
nationwide fleet average EFs for 1990, 1995, 2000, and 2010 are
0.0910, 0.0523, 0.0291, and 0.0178 g/mi respectively.

     These nationwide fleet average EFs were then multiplied by
total nationwide fleet VMT obtained from the MOBILE4.1 fuel
consumption model to estimate nationwide diesel particulate
matter emissions.  No recent rural diesel particulate matter
fleet average EFs were available; thus, separate rural and urban
diesel particulate matter emission levels could not be estimated.
Nationwide diesel particulate matter emission estimates are
listed in Table 9-3.  The 1990 nationwide diesel particulate
matter emission estimate of 163,118 metric tons compares with a
higher 1990 estimate of 384,000 metric tons for diesel vehicles
in a recent EPA report on air pollutant emission estimates (EPA,
1991b).

Table 9-3.  Nationwide Diesel Particulate Matter Emissions.
                               9-9

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Year




1990
1995
2000
2010
Total
Nationwide
Fleet VMT (mi)


1793.07 x 109
2029.74 x 109
2269.25 x 109
2771.30 x 109
Nationwide
Diesel
Particulate
Matter
(metric tons)
163,118
106,080
66,076
49,441
9.4  Atmospheric Reactivity and Residence Times of Particulate
     Phase Polycyclic Organic Matter  (POM)

     POM species can exist in both the gas and particulate phases
in the atmosphere.  The distribution between the two phases is
determined by the vapor pressure of the species, the ambient
temperature, and the amount of airborne particulate matter
present.  Cold temperatures and higher aerosol concentrations
lead to greater association of POM with particles.  The focus of
this section is on particulate phase POM, since most of the POM
emitted by motor vehicles is in this form.  The information that
follows on transformation and residence times has been mainly
excerpted from a report produced by Systems Application
International for the EPA (Ligocki and Whitten, 1991) .

9.4.1  Particulate-Phase Chemistry

      The determination of rate constants for POM that are
adsorbed to particles is difficult, because these rates are
strongly influenced by the characteristics of the surface to
which the POM are adsorbed.   Thus, the observations reported in
the literature regarding the reactivity of adsorbed POM tend to
appear contradictory.  Early studies and extrapolation from
reactivity studies of POM in organic solution suggested that POM
compounds react rapidly on surfaces (NAS, 1972).  Later work
demonstrated that, although POM present on substrates such as
silica and alumina photolyze rapidly,  POM present on coal fly ash
and carbon black were resistant to photochemical degradation
(Korfmacher et al.,  1980, 1981; Behymer and Kites, 1985).
Significant differences in photochemical degradation rates have
been reported between two different fly ashes  (Dlugi and Giisten,
1983) .

     Nonetheless, some POM may be capable of being oxidized in
the particulate phase.  Fox and Olive (1979) reported that 90
percent of anthracene present on atmospheric particulate matter
disappeared in four days when exposed to sunlight, whereas only a
small fraction of the anthracene which was exposed to ambient air
but shielded from light disappeared.  The conversion of POM
present on diesel particulate matter and exposed to ozone has
been reported (Van Vaeck and Van Cauwenberghe, 1984) and appears
                               9-10

-------
                                                        EPA-420-R-93-005
                                                            April 1993

to be an important removal pathway for some POM.  However,
Grosjean and co-workers  (1983) found no degradation of
benzo[a]pyrene and perylene adsorbed onto a variety of
substrates, including diesel soot, over a three hour exposure  to
100 ppb ozone.  The conversion of POM present in  soot and  exposed
to NOX has  also been reported (Butler and Crossley,  1981) .
Unfortunately, the species responsible for this observed
oxidation was not determined.  Even though POM react readily with
N205 in  the gaseous phase, the reaction of N205 with  adsorbed POM
is significantly lower.

9.4.2  Aqueous Phase Chemistry

     POM are slightly soluble in water, and will  be incorporated
to some degree into clouds and rain.  For species which  are
associated with particles, the water-affinity of  the particle
surface will determine the degree to which they will be
incorporated into clouds and/or rain.  Polycyclic ketones  and
quinones are much more soluble in water than the  parent  POM and
will be incorporated into clouds and rain to a much greater
degree.

9.4.3  Reaction Products

     Most POM reactions proceed by addition, forming polycyclic
aromatic ketones, quinones,  epoxides, and nitro compounds.

     Much of the focus on POM oxidation products  has centered  on
the nitro-POM, since several of these compounds,  such as the
dinitropyrenes, are known to be extremely potent  mutagens.
Although the yields of these species are generally not large,
they may still account for a significant fraction of the observed
                               9-11

-------
                                                        EPA-420-R-93-005
                                                            April 1993

mutagenicity of ambient POM.  In diesel exhaust particulate
matter, 3-nitrofluoranthene was the major constituent.

9.4.4  Polycyclic Organic Matter Residence Times

     For particulate species, the rate of removal by wet and dry
deposition will depend upon the particle size distribution.
Large particles are removed rapidly from the atmosphere by
sedimentation and impaction.  Smaller particles do not contain
sufficient mass to sediment or impact, but diffuse much more
rapidly than do large particles.  As a result, removal rates of
atmospheric particles are governed by the competition between
these two types of processes, and generally reach a minimum
somewhere in the range 0.1 to 1.0 micrometer  (urn).  This size
range is often referred to as the accumulation mode, because
particles in this size range tend to persist, and hence
accumulate.  The National Ambient Air Quality Standard for PM10
is based on the particulate matter less than 10 urn in diameter.

     Particle size distributions for a few POM have been
reported.  Evidence suggests that the larger, less volatile POM
tend to be present on smaller particles than the smaller, more
volatile POM (Pistikopoulos et al.,  1990), but that all POM are
primarily associated with submicron particles.  Van Vaeck and Van
Cauwenberghe (1978)  measured particle size distributions for a
set of POM finding that 90% of the 4-ring POM and 91% of the 5-
ring POM are associated with particles <1.5 urn in diameter.
These data are comprehensive, and were used in calculating the
wet and dry deposition of POM.

     Residence times are presented for two individual POM
species:  an intermediate POM (pyrene) and a particulate-phase
POM (benzo[a]pyrene).   These examples provide comparisons of the
importance of chemical transformation to other species versus
physical removal, and differences between POM species of varying
size.   It should be noted, however,  that in some cases it is not
appropriate to view atmospheric reactions as destruction pathways
for toxic species, because the products formed from its
destruction may be equally toxic, or even more toxic.  Most POM
reactions, for instance, proceed by addition, forming polycyclic
aromatic ketones, quinones, epoxides, and nitro compounds.

     Residence times for POM as a class are also presented.  For
POM as a class, however, atmospheric residence times are
determined by physical processes only.  Chemical reactions may
transform individual compounds,  but available evidence suggests
that the products of this transformation are also POM species.
Therefore, the residence time of POM as a class may be determined
by wet and dry deposition only.
                               9-12

-------
                                                        EPA-420-R-93-005
                                                            April 1993


9.4.4.1  Pyrene

     Pyrene has a vapor pressure that falls within the range
where either gas phase or particulate phase processes might
dominate depending upon ambient conditions.  Under wintertime
conditions, and/or high particle loading conditions, a majority
of pyrene concentration may be associated with particles.
Therefore,  its atmospheric residence time is determined by both
gas-phase and particulate-phase processes.

     Residence times for pyrene were calculated by considering
gas-phase chemical reactions with OH and N205, particulate phase
chemical reaction with 03,  aqueous  phase chemical  reactions  with
OH and 03,  and wet and dry deposition.   The calculated residence
times for pyrene are presented in Table 9-4.  Because of the
similarities between the chemical reactivity and physical
properties of pyrene and fluoranthene,  the residence times
presented in Table 9.1 also can be considered to apply to
fluoranthene.

     The calculated residence times for fluoranthene and pyrene
range from 0.8 to 1.6 hours under summer, daytime, clear-sky
conditions.  These residence times are roughly half as long as
those calculated for naphthalene, a POM present virtually
exclusively in the gas phase.  As with naphthalene, gas phase
reaction with OH is the most important atmospheric removal
pathway.  However, for fluoranthene and pyrene, particulate-phase
processes including reaction with 03  and wet and dry deposition
are also significant.

     Under cloudy conditions, in-cloud chemical destruction
accounts for 10 to 30 percent of pyrene removal at night in the
summer.  In the daytime and in the winter, in-cloud processes are
less important.  Both the OH and 03 oxidations  contribute to the
aqueous reactivity, with the OH pathway more important in the
summertime, and both pathways important in the winter.

     Wet deposition is very rapid for particulate-phase
fluoranthene/pyrene.  Particle scavenging leads to residence
times on the order of 2 to 20 hours in the wintertime.  Dry
deposition is a major removal mechanism for fluoranthene/pyrene
at night,especially in winter and under cloudy-sky conditions.
Dry deposition of particulate-phase fluoranthene/pyrene is more
efficient than that of gas-phase fluoranthene/pyrene.

     Major uncertainties in the estimate of residence times for
fluoranthene/pyrene include the order-of-magnitude uncertainty in
the particle scavenging rate and the rate of reaction of the
particulate-phase species with ozone. Also significant is the
factor-of-two uncertainty in the OH radical concentration.
                               9-13

-------
                                                                                                      EPA-420-R-93-005
                                                                                                           April 1993
TABLE 9-4.  Atmospheric residence  time calculation  for fluoranthene/pyrene.   All times are in hours unless
otherwise noted.
                                 Los Angeles

                               July       Jan
   St.  Louis

July         Jan
   Atlanta

July      Jan
   New York


July      Jan
Clear sky - day 1.2
Clear sky - night 60
Clear sky - avg 2
Cloudy - day 3
Cloudy - night 50
Cloudy - avg 5
Rainy - day --*
Rainy - night --*
Rainy - avg --*
Monthly Climatological
9 0.8 18 0.8 14 1.6
80 80 90 70 60 70
18 1.3 40 1.3 30 2
18 1.8 30 1.8 30 3
110 70 80 50 80 80
40 3 50 3 50 5
2-9" 1.5-1.8" 1.0-6" 1.7 2-9" 3
1.2-8" 4-18" 0.7-5" 6-17" 1-7" 4-20"
1.5-8" 2-3" 0.8-5" 2-3" 1-8** 3-5**

30
130
60
50
130
80
0.9-6"
0.7-5"
0.7-5"

   Average
                                             20
            30-40
         18-30
          20-60
*Not calculated since July rainfall is zero for Los Angeles.
"Range of values  calculated  using  high and low estimates for particle scavenging.
                                                       9-14

-------
                                                        EPA-420-R-93-005
                                                            April 1993
9.4.4.2  Benzo[a]pyrene  (B[a]P)
     Benzo[a]pyrene is present in the particulate phase under
most conditions.  Therefore, its atmospheric residence time is
primarily determined by particulate-phase processes.  Residence
times for benzo[a]pyrene were calculated by considering gas-phase
chemical reaction with OH, particulate phase chemical reaction
with 03,  aqueous phase chemical  reactions with OH and 03,  and wet
and dry deposition.  The calculated residence times  for
benzo[a]pyrene are presented in Table 9-5.  By comparison to
fluoranthene/pyrene, the calculated summertime residence times
are longer for benzo[a]pyrene.   This is due to the greater
association with particles of benzo[a]pyrene.  Interestingly,
however,  the calculated wintertime residence times for
benzo[a]pyrene are shorter than fluoranthene/pyrene.  In fact the
difference in residence time between summer and winter is only a
factor of two to three for benzo[a]pyrene.

     The reaction of 03 with particulate-phase benzo[a]pyrene is
predicted to be the dominant removal mechanism for benzo[a]pyrene
under most conditions.  The reaction of gas-phase benzo[a]pyrene
with OH is also expected to be significant in the summertime,
despite the relatively small fraction of benzo[a]pyrene present
in the gas phase.

     Unlike the gaseous POM, wet and dry deposition  are
significant
atmospheric removal mechanisms for benzo[a]pyrene .  Wet
deposition leads to atmospheric residence times of 0.5 to 3 hours
on rainy days and contributes significantly to monthly
climatological average residence time.   Dry deposition is less
important, but still contributes up to 30 percent of the removal.

     Major uncertainties in the estimate of residence times for
benzo[a]pyrene include the order of magnitude uncertainties in
the rate of reaction of the particulate-phase species with ozone,
and the particle scavenging rate.   The calculated residence times
for benzo[a]pyrene are, therefore, significantly more uncertain
than those calculated for fluoranthene/pyrene.

9.4.4.3  Other POM Species

     Among the particulate-phase POM, the residence  times
calculated for benzo[a]pyrene are probably valid for other
reactive species such as perylene.  However, more stable POM,
such as the benzofluoranthenes,  benzo[e]pyrene, and  coronene, may
be removed primarily by physical processes, and would have
residence times up to ten times longer than that calculated for
benzol[a]pyrene.

9.4.4.4  POM as a Class

     The summertime residence times for pyrene and B[a]P suggest
that POM are transformed relatively rapidly in the summertime.
For
                               9-15

-------
                                                                                                      EPA-420-R-93-005
                                                                                                            April 1993

TABLE 9-5.  Atmospheric residence  time  calculation for benzo[a]pyrene.  All times are in hours unless  otherwise
noted.
                                Los Angeles

                              July       Jan
                          St.  Louis

                       July         Jan
                                    Atlanta

                                July       Jan
                                          New York

                                      July       Jan
  Clear sky - day

  Clear sky - night

  Clear sky - avg
 4

11


 5
13

30

19
 3

11


 5
20


40


30
 4


11


 5
15


20


17
 5


11


 6
30


90


50
  Cloudy - day

  Cloudy - night

  Cloudy - avg
 6

11


 7
19

30

20
 6


11


 7
30


40


40
 6


11


 7
18


20


20
 7


11
50


90


70
  Rainy - day

  Rainy - night

  Rainy - avg


  Monthly Climatological
  Average
           0.5-3

           0.5-3*

           0.5-3*


           10-18*
            0.5-3


            0.5-3*


            0.5-3*



              4-6*
         0.5-4


         0.5-4*


         0.5-4*



         20-30*
       0.5-3


       0.5-3*


       0.5-3*


         5-6*
         0.5-3


         0.5-3*


         0.5-3*



          8-17*
       0.5-3


       0.5-3*


       0.5-3*



         5-7*
          0.5-4


          0.5-4*


          0.5-4*



          16-40*
*Not calculated since July rainfall is zero for Los Angeles.
"Range of values  calculated  using high and low estimates for particle scavenging.
                                                       9-16

-------
                                                        EPA-420-R-93-005
                                                            April 1993

the case of species such as fluoranthene and pyrene, their
oxidation products will condense onto atmospheric particles.  At
that point, they may be relatively stable against further
oxidation, and may persist until removed by wet and dry
deposition.

     The atmospheric residence time of a generic non-reactive
particulate-phase POM that is removed only by physical processes
(i.e., wet and dry deposition) is presented in Table 9-6.
Because the algorithms used to calculate the dry deposition
velocities and wet deposition rates did not contain any city-
specific information, the calculated clear-sky and rainy
residence times are the same for all cities.  The differences in
the monthly climatological average residence times reflect only
the differences in monthly rainfall among the cities.
Atmospheric residence times range from less than a day to three
days in both summer and winter.

9.4.5 Urban Airshed Modeling of POM

     The explicit modeling of POM is difficult to achieve due to
the inherent complexity of POM itself.  Major consideration needs
to be given to the relative abundance of the various POM species
in the atmosphere, the availability of emissions data, and
determining an area's specific area, mobile, and point sources.

     Since POM basically consists of three distinct species
categories, all three would have to be taken into consideration.
These are the naphthalenes, which are an order of magnitude
higher than the concentrations of any of the other POM  (thought
not among the more toxic constituents of POM); the other gas-
phase POM concentrations that are much greater than the
particulate-phase concentrations; and the particulate phase
itself.  Each of these species has its own transformation and
reactivity parameters that need to be taken into consideration.

     Due to these many considerations and parameters, and the
absence of software to implement these factors, the Urban Airshed
Modeling of POM was not accomplished in the St. Louis study
(Ligocki and Whitten, 1991) .

     However, POM was treated explicitly in the Baltimore-
Washington and Houston area studies (Ligocki et al., 1992).  POM
was assigned to three species categories in the UAM-Tox  (as
described above),  based upon molecular weight  (MW):

     NAPH      MW < 160
     POM1      160 < MW < 220
     POM2      220 < MW

The species NAPH consists largely of naphthalene and substituted
naphthalenes, which account for the bulk of the POM mass.  NAPH
reacts rapidly with OH and slowly with N205.  POM1 and POM2 are
represented as nonreactive.  Additional information on the
modifications made to UAM to model POM explicitly are described
in the reference cited above.

                               9-17

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Simulations for the summer Baltimore-Washington area episode
resulted in slight decreases in POM with the use of federal
reformulated gasoline.  California reformulated gasoline resulted
in larger POM decreases than federal reformulated gasoline,
because of reductions in the T90 distillation point of the  fuel.
The maximum daily average POM for the 1988 base scenario was 6.8
ug/m3.   Simulated daily average  POM concentrations  were  much
lower in the Washington area (0.5-1.0 ug/m3)  than in  the
Baltimore area  (1-6.8 ug/m3) .  Motor vehicle-related  NAPH
accounted for about 15% of total NAPH emissions, motor vehicle-
related POM1 accounted for about 43% of total POM1 emissions, and
motor vehicle-related POM2 accounted for about 35% of total POM2
emissions.  Furthermore, motor vehicle-related POM accounted for
about 15% of the total simulated POM concentration, based on the
1995 no motor vehicle scenario.

     Since no data were available on measured POM concentrations
in the Baltimore-Washington area,  simulated concentrations were
compared to measured concentrations from other cities.
Concentrations of POM in Washington were in line with
concentrations in other cities,  but concentrations in Baltimore
appear to be overpredicted.

     In the winter 1988 base scenario, the maximum daily average
POM concentration was 4.4 ug/m3, lower than in summer.   NAPH
emissions decreased because they were primarily influenced by
evaporative emissions from asphalt paving.  Emissions of POM1 and
POM2,  the larger POM components, increased significantly in
winter because of residential wood combustion.  Motor vehicle-
related POM concentrations with federal reformulated gasoline use
decreased more in winter than in summer, ranging from 4 to 8
percent.  Motor vehicle-related POM accounted for about 10% of
the maximum simulated concentration, based on the 1995 no motor
vehicle scenario.

     For the summer 1987 base scenario in Houston,  the maximum
daily average POM concentration was 3.4 ug/m3.   Motor vehicle-
related NAPH accounted for about 17% of total NAPH emissions,
motor vehicle-related POM1 accounted for about 24% of total POM1
emissions, and motor vehicle-related POM2 accounted for about 19%
of total POM2 emissions.  Furthermore, motor vehicle-related POM
accounted for about 18% of the maximum simulated concentration,
based on the 1995 no motor vehicle scenario.  Simulations for the
summer Houston episode predicted larger decreases than in the
Baltimore-Washington area with the use of reformulated gasoline.
Simulated concentrations of POM were in good agreement with
concentrations measured in other cities.
                               9-18

-------
ABLE 9-6. Atmospheric residence time calculation for a generic particulate-phase POM which
hysical processes only. All times are in hours unless otherwise noted.
Los Angeles St. Louis Atlanta
July Jan July Jan July Jan
Clear sky - day 60 120 60 120 60 120
Clear sky - night 90 90 90 90 90 90
Clear sky - avg 70 100 70 100 70 100
Rainy - day --* 0.5-4" 0.5-4" 0.5-4" 0.5-4,, 0.5-4"
Rainy - night --* 0.5-4" 0.5-4" 0.5-4" 0.5-4" 0.5-4"
Rainy - avg --* 0.5-4" 0.5-4" 0.5-4" 0.5-4" 0.5-4"
Monthly Climatological
Average 70 16-60" 15-50" 30-80" 12-40 13-50**
EPA-420-R-93-005
April 1993
is removed by
New York
July Jan
60 120
90 90
70 100
0.5-4" 0.5-4"
0.5-4" 0.5-4"
0.5-4" 0.5-4"
15-50" 18-60"
*Not  calculated since July rainfall is zero for Los Angeles.
"Range of values calculated using high and low estimates  for particle  scavenging.
                                                     9-19

-------
                                                        EPA-420-R-93-005
                                                            April 1993

9.5  Exposure Estimation

9.5.1  Annual Average Exposures Using HAPEM-MS

     To obtain urban and rural annual average exposures, urban
diesel particulate matter national fleet average emission factors
in Table 9-2 were first multiplied by the urban and rural g/mile
to ug/m3  conversion factors  obtained from HAPEM-MS for 1988
(Johnson et al.,  1992):

               CONVurban =29.4  (ug/m3)/(g/mile)
               CONVrural = 15.9  (ug/m3)/(g/mile)

This provides an estimate of urban and rural exposure relative to
the number  of vehicle  miles travelled  (VMT) in  1988.   To obtain
exposure estimates for the years of interest, these values were
then multiplied  by incremental adjustments  to allow  for the VMT
increase in excess of the population increase for the year of
interest.   The adjustment factors  used for 1990,  1995,  2000, and
2010 are 1.031,  1.123,  I.218, and 1.412, respectively.  Resulting
urban and rural  annual  average exposures for 1990, 1995, 2000, and
2010 are given in Table 9-7.

9.5.2  Comparison of HAPEM-MS to Ambient Monitoring Data

     Using  ambient  monitoring  data  (EPA,  1991b,   1991c),  the
concentration of diesel particulate matter in ambient air samples
can be estimated.  For 1990, the national average total suspended
particle concentration  is estimated  to be  about  48  ug/m3  (EPA,
1991c).   This can be multiplied by percent contribution of diesel
particulate matter to TSP, which is calculated to be 5.12%.  This
percentage  was   obtained by  dividing  an  estimate  for  diesel
emissions of 384,000 metric  tons (EPA,  1991b) by a TSP estimate of
7.5 x 10s metric tons (EPA, 1991c) .   The resultant concentration of
diesel particulate matter obtained by multiplying 48 ug/m3 by 5.12%
is  2.46  ug/m3.    This  number  was  then adjusted  for integrated
exposure, resulting in integrated exposure estimate of 1.52  ug/m3.
The HAPEM-MS 1990  urban  diesel particulate matter annual average
exposure of 2.03 ug/m3  is about 134% of this value.  The HAPEM-MS
1990 rural  diesel  particulate matter  annual average  exposure of
1.10 ug/m3  is  about 72%  of this value.

9.6  Carcinogenicity of Diesel Particulate  Matter  and Unit Risk
Estimates

9.6.1 Most Recent EPA Assessment

     A draft  health assessment document for diesel emissions has
been prepared (EPA, 1990b).   Much of the information contained in
this section has been taken  from this  document.  An update of this
document is expected shortly.
                               9-20

-------
                                                        EPA-420-R-93-005
                                                            April 1993
 Table 9-7.  Diesel Particulate Matter Annual Average Exposures
Year
1990
1995
2000
2010
Exposure
(ug/m3)
Urban
2.03
1.18
0.67
0.44
Rural
1.10
0.64
0.36
0.24
Nationwide
1.80
1.05
0.60
0.39
9.6.1.1  Description of Available Carcinogenicity Data

     To evaluate the carcinogenicity of diesel engine particulate
emissions, controlled animal and mutagenicity studies were
conducted as well as studies of populations occupationally
exposed to diesel exhaust.  The following paragraphs contain a
brief summary of the  EPA evaluation of these studies; the EPA
draft document discusses these studies in more detail  (EPA,
1990b).

Genotoxicity

     Extensive Ames test studies with Salmonella have
unequivocally demonstrated direct-acting mutagenic activity in
both the particle and gaseous fractions of diesel exhaust
(Huisingh et al.,  1978; Siak et al.,  1981; Claxton, 1981, 1983;
Claxton and Kohan, 1981; Dukovich et al.,  1981; Lewtas, 1983;
Brooks et al.,  1984; Matsushita et al.,  1986).  The induction of
gene mutations has been reported in several in vitro mammalian
cell lines after exposure to extracts of diesel particulate
matter (Casto et al.,  1981; Chescheir et al.,  1981; Curren et
al., 1981; Liber et al., 1981; Mitchell et  al., 1981; Barfnecht
et al., 1982;  Li and Royer, 1982; Brooks et al., 1984; Morimoto
et al., 1986).   Dilutions of whole diesel exhaust did not induce
sex-linked recessive lethals in Drosophila (Schuler and Niemeier,
1981) or specific-locus mutations in male mouse germ  (sperm)
cells (Russell et al.,  1980).

     Structural chromosome aberrations and sister chromatid
exchanges (SCE) in mammalian cells have been induced by particles
and direct diesel exhaust  (Guerrero et al.,  1981; Mitchell et
al., 1981; Lewtas, 1983: Morimoto et al.,  1986; Pereira et al. ,
1982; Tucker et al., 1986).  Sister chromatid exchanges, but not
chromosomal aberrations, were observed in Chinese hamster cells
upon exposure to particle extracts (Brooks et al., 1984).  Whole
exhaust induced micronuclei, but not SCE or structural
aberrations were found in bone marrow of male Chinese hamsters
exposed to whole diesel emissions for 6 months.  In shorter
exposure  (7 weeks), neither micronuclei nor structural
aberrations were increased in bone marrow of female Swiss mice
                               9-21

-------
                                                        EPA-420-R-93-005
                                                            April 1993

(Pereira et al.,  1981a).  Likewise whole diesel exhaust did not
induce dominant lethal or heritable translocations in male mice
exposed for 7.5 and 4.5 weeks, respectively  (Russell et al.,
1980) .

      Analysis of caudal sperm for sperm head abnormalities was
conducted  (Pereira et al.,  1981b) after exposure to diesel
exhaust particles and it was found that the exposed incidence of
abnormalities was not above the control levels.   Conversely,
male Chinese hamsters exposed to diesel particulate matter
(Pereira et al.,  1981c) exhibited almost a threefold increase in
sperm head abnormalities.

Animal Studies

     As early as 1955, there was evidence  (Kotin et al., 1955)
for tumorigenicity and carcinogenicity of acetone extracts of
diesel exhaust in skin tumorigenesis tests.  Also data  suggested
a difference in response depending on the engine operating mode.
Until the mid 1980's, no chronic studies assessing inhalation of
diesel exhaust,  the relevant mode for human exposure, had been
reported.  This is,  however, the route of exposure which was used
in the most extensive, recent studies.  Studies employing rats
and an adequate experimental design were nearly all positive in
demonstrating diesel exhaust-induced increases in tumorigenicity.
The 9.5 percent increase in tumor incidence for female  Wistar
rats reported by Heinrich et al. (1986a) is supported by the
report by Mauderly et al.  (1987), which showed a 3.6 percent and
12.8 percent increase in tumor incidence for F344 rats  following
chronic exposure to diesel exhaust at particle concentrations of
3.5 and 7.0 mg/m3, respectively.  However,  only one of the
squamous cell tumors reported by Heinrich et al.  (1986a) was
classified as a carcinoma.   In the Mauderly et al.  (1987) study,
the carcinoma incidence was 0.9, 1.3, 0.5, and 7.5 percent for
the control, low, medium,  and high exposure groups, respectively.

     The inhalation studies by Wong et al.  (1986) and Bond et al.
(1990)  affirm observations of the potential carcinogenicity of
diesel exhaust by providing evidence for DNA damage in  rats.
Similarly, Iwai et al.  (1986)  demonstrated diesel exhaust-induced
tumorigenicity in rats exposed to an exhaust particle
concentration of 4.9 mg/m3,  although the sample size was small.
This study also reported development of a splenic lymphoma, which
represents the only nonpulmonary tumor resulting from inhalation
exposure to diesel exhaust.   The long-term inhalation study by
Ishinishi et al.  (1986) showed a greater incidence of carcinomas
(6.5 percent) in rats following 30-month exposure to diesel
exhaust at 4 mg/m3,  but not  at lower (0.4,  1.0,  or 2.0 mg/m3)
exposure levels.   However,  Brightwell et al.  (1986) demonstrated
a dose-dependent increase in tumor incidence for male and female
F344 rats exposed to filtered, but no unfiltered diesel exhaust
(five 16 hour periods per week), at concentrations as low as 2.2
mg/m3 and also at 6.6 mg/m3.   Filtered and unfiltered exhaust are
used to discriminate between the gaseous and particle effects.
This study indicated that,  for unfiltered exhaust, the  tumor
incidence was higher for female rats  (0 percent, 15 percent, or

                               9-22

-------
                                                        EPA-420-R-93-005
                                                            April 1993

54 percent at 0.0, 2.2, or 6.6 mg/m3)  than for male rats (1
percent, 4 percent, or 23 percent for 0.0, 2.2, or 6.6 mg/m3).
The filtered exhaust showed no increase in tumors when compared
to controls.  Thus, these studies demonstrated carcinogenic
effects in rats at exposure levels ranging from 2.2 to 7.0 mg/m3.

     The inhalation of whole diesel exhaust by NMRI mice
(Heinrich et al.,  1986a,b; Stober, 1986), Sencar mice  (Pepelko
and Peirano, 1983), and in rats  (Takaki et al., 1989) also
provided evidence of carcinogenicity.  In Orthoefer et al.
(1981),  the exposure of Strain A mice to irradiated diesel
exhaust (to simulate sunlight exposure and resulting reactions)
did not produce any significant signs of gross toxicity or affect
the growth rates of the mice.  Exposures ranged from 20 hr/day, 7
days\week, for 7 weeks, diluted 1:13 in one experiement to an 8
week inhalation at 6 mg/m3 in another.   The mice were then held
for an additional 26 weeks in clean air after cessation of
exposure.   Exposures to either irradiated or nonirradiated
exhaust did not result in significantly increased lung tumor
incidences compared with controls.  Due to short exposures
selected for these studies they are considered to be screening
tests.  The short exposure and holding periods prior to sacrifice
are based upon the rapid increase in tumor rates in positive
tests.  The observed increase in mutagenicity of irrradiated
exhaust observed in chronic bioassys is discussed in Chapter 12,
Section 12.4.3.

     Both the Heinrich et al. (1986a) and Brightwell et al.
(1986) studies provide negative results for tumorigenicity of
diesel exhaust in hamsters, a species known for its resistance to
tumor induction.   Negative results were also presented by several
other investigators (Takemoto et al., 1986; Schreck et al. , 1982;
Barnhart et al.,  1981; Karagianes et al., 1981), but these
studies tended to employ inadequate exposure durations, low
exposure concentrations, or inadequate animal numbers per group.
A negative study reported by Kaplan et al.  (1982) contained a
high incidence of tumors in the control group. Similarly, the
studies using monkeys  (Lewis et al., 1986) and cats  (Pepelko and
Peirano, 1986) were of inadequate duration  (2 years) for these
longer-lived species.

     Alternate exposure routes including dermal exposure, skin
painting,  and subcutaneous injection provided additional evidence
for tumorigenic effects of diesel exhaust.  Evidence for
tumorigenicity was demonstrated by Kotin et al.  (1955) for mice
to which an acetone extract of diesel exhaust particles was
applied dermally.   Nesnow et al.   (1982) also showed that extracts
from some diesel engines were potentially tumorigenic following
dermal application to rodents.  A significant increase in the
incidence of subcutaneous tumors in female C57B1 mice was
reported by Kunitake et al.  (1988) for subcutaneous
administration of light-duty diesel exhaust tar extract at doses
of 500 mg/kg.  Doses at or below 200 mg/kg, however, were
negative.   Takemoto et al. (1988) provided additional data for
this study and reported an increased tumor incidence in the mice
following injection of light-duty engine exhaust extract at doses

                               9-23

-------
                                                        EPA-420-R-93-005
                                                            April 1993

of 100 and 500 mg/kg.  Negative results were reported by Depass
et al. (1982) for skin-painting studies using mice and acetone
extracts of diesel exhaust particle suspensions.  However, in
this study the exhaust particles were collected at temperatures
of 100C,  a temperature that would minimize the condensation of
vapor-phase organics and, therefore, reduce the availability of
potentially carcinogenic compounds that might normally be present
on diesel exhaust particles.  Intraperitoneal injection studies
using Strain A mice were generally negative.

     Diesel exhaust is composed of gaseous and particle phases
and is known to be a complex mixture containing verified and
potential carcinogens.  Nevertheless, because of the negative
results via inhalation, the fact that most POM is adsorbed onto
particles, and because the potentially carcinogenic agents
present in the gaseous phase, i.e., benzene, formaldehyde, are
not known to induce lung tumors, it is unlikely that this
component contributes to the tumorigenic responses.   A study by
Grimmer et al. (1987) demonstrated that a whole exhaust
condensate fraction containing polycyclic aromatic hydrocarbons
(PAH) with 4 to 7 rings produced a high tumor incidence when
implanted into rat lungs.  It was also noted that this fraction
represented only 0.8 percent of the total weight of the exhaust
condensates, and that some tumorigenicity was also associated
with nitroaromatic fractions.  The PAH fraction produced a tumor
incidence similar to that of a low concentration of
benzo[a]pyrene (BaP).

       Several of the previously discussed studies indicated that
only the whole (unfiltered) diesel exhaust is tumorigenic or
carcinogenic and that these properties are eliminated or greatly
minimized in filtered diesel exhaust exposure.  Inhalation
experiments using tumor initiators  (Brightwell et al.,  1986;
Heinrich et al.,  1986a; Takemoto et al.,  1986) did not provide
conclusive results regarding the carcinogenic potential of
filtered vs. whole diesel exhaust.  Although the tumorigenicity
of the gaseous fraction is presently unresolved and most
experiments using filtered exhaust were negative, most of these
experiments did not provide definitive evidence that a maximum
tolerated dose was achieved.  The carbon core of the exhaust
particle has been determined to have carcinogenic potential.  The
fact that allegedly inert, insoluble biochemically,
"noncarcinogenic" particles such as titanium dioxide (Lee et al.,
1986) or instillation of activated carbon  (Kawabata et al.,  1986)
have been shown to induce lung cancer at very high concentrations
is of concern in this respect.  Studies currently in progress,
indicating that carbon black, containing essentially no organics,
was as effective as diesel exhaust in lung cancer induction,  (see
Section 9.6.3.1)  supports the approach that the carbon core plays
a major role in the pulmonary carcinogenicity of diesel exhaust
in rats.

     Although uncertainties exist regarding the tumorigenic
potential of the gaseous component and the carbon core component
of diesel exhaust, it is clear that diesel exhaust is
carcinogenic in animals inducing pulmonary tumors.  This

                               9-24

-------
                                                        EPA-420-R-93-005
                                                            April 1993

contention is supported by positive results in numerous,
independent studies in male and females of at least two species
and by several routes of administration, including inhalation,
intratracheal administration, skin painting, and subcutaneous
injection.

Human Data

     Certain extracts of diesel exhaust also have been
demonstrated to be mutagenic and carcinogenic in humans.  Since
large working populations are currently exposed to diesel
exhaust,  and since nonoccupational exposures currently are of
concern as well,  the possibility that exposure to this complex
mixture may be carcinogenic to humans has become an important
public health issue.

     A major difficulty with the occupational studies considered
here was the measurement of the actual diesel exhaust exposure.
Most studies compared men in job categories with presumably some
exposure to diesel exhaust with either standard populations
(presumably no exposure to diesel exhaust) or with men in other
job categories from industries with little or no potential for
diesel exhaust exposure.  A few studies have included
measurements of diesel fumes, but there is no standard method for
the measurement.   No attempt is made to correlate these exposures
with the cancers observed in any of these studies, nor is it
clear exactly which diesel particulate matter should be measured
to assess the occupational exposure to diesel exhaust.  The
occupations involving potential exposure to diesel exhaust are
miners, truck drivers, transportation workers, railroad workers,
and heavy equipment operators.

     The seven cohort studies reviewed by EPA (1990b) have mainly
demonstrated an increase of lung cancer.  The three cohort
studies of bus company workers by Waller  (1981), Rushton et al.
(1983), and Edling et al.  (1987) failed to demonstrate any
statistically significant excess risk of lung cancer, but these
studies have certain methodological problems such as small sample
sizes, short follow-up periods, lack of information on
confounding variables, and lack of analysis by duration of
exposure or latency that preclude their use in determining the
carcinogenicity of diesel exhaust.  Although the Waller (1981)
study had a 25-year follow-up period, the cohort was restricted
to only employees  (ages 45 to 64) currently in service.
Employees who left the job earlier, as well as those who were
still employed after age 64 and who may have died from cancer,
were excluded.

     Wong et al.  (1985) conducted a mortality study of heavy
equipment operators that demonstrated a significant increased
risk of liver cancer in the total cohort and in various
subcohorts.  The same analysis also showed statistically
significant deficits in cancers of the large intestine and
rectum.  Metastasis from the cancers of the large intestine and
rectum to the liver probably were misclassifled as primary liver
cancer which lead to an observed excess risk.  This study did

                               9-25

-------
                                                        EPA-420-R-93-005
                                                            April 1993

demonstrate a nonsignificant positive trend for cancer of the
lung with length of membership and latency.  Individuals without
work histories who started work prior to 1967 when records were
not kept may have been the ones who were in the same job for the
longest period of time.  The workers without job histories
included those who had the same job before and after 1967 and
thus may have worked about 12 to 14 years longer; these workers
exhibited significant excess risks of lung cancer and stomach
cancer.  If this assumption about their jobs is correct, then
these site-specific causes may be linked to diesel exhaust
exposure.  However, this study has quite a few methodological
limitations such as the absence of detailed work histories for 30
percent of the cohort and the availability of only partial work
histories for the remaining 70 percent;  thus,  jobs were
classified and ranked according to presumed diesel exposure.
Information is lacking regarding duration of employment in the
job categories (used for surrogate of exposure),  and other
confounding factors (alcohol consumption, cigarette smoking,
etc.).

     A two-year mortality analysis of the American Cancer
Society's prospective study by Boffetta et al.  (1988), after
controlling for age and smoking, demonstrated an excess risk of
lung cancer in certain occupations with potential exposure to
diesel exhaust (railroad workers, heavy equipment operators,
truck drivers, and miners).   These excesses were statistically
significant among miners  (RR =2.67, 95 percent CI = 1.63 to
4.37)  and heavy equipment operators  (RR =2.6,  95 percent CI =
1.12 to 6.06).  The elevated risks were nonsignificant in
railroad workers (RR = 1.59) and truck drivers (RR = 1.24).   RR
(OR) and CI are defined in Section 6.6.3.4.  A dose response was
also observed for the truck drivers.  With the exception of
miners, exposure to diesel exhaust occurred in the three other
occupations showing an increase in the risk of lung cancer.   This
study exhibited two methodological limitations.  These include,
the lack of representiveness of the study population composed of
volunteers only and the questionable reliability of exposure data
based on self-administered questionnaires which were not
validated. Despite these limitations this study is suggestive of
a causal association between exposure to diesel exhaust and
excess risk of lung cancer.

     There were two mortality studies conducted on railroad
workers by Howe et al. (1983) in Canada and Garshick et al.
(1988)  in the United States.  The Canadian study found relative
risks of 1.2 and 1.35 among "possibly" and "probably" exposed
groups, respectively.   The trend test showed a highly significant
dose response relationship with exposure to diesel exhaust and
the risk of lung cancer.   The main limitation of the study was
the inability to separate overlapping exposures of coal dust and
diesel fumes.   Information on jobs was available at retirement
only.   There was also insufficient detail on the classification
of jobs by diesel exhaust exposure.  The exposures could have
been noncurrent,  but since the data are lacking,  it is possible
that observed excess could be due to the effect of both coal dust
and diesel fumes and not due to just one or the other.  However,

                               9-26

-------
                                                        EPA-420-R-93-005
                                                            April 1993

it should be noted that, so far, coal dust has not been
demonstrated to be a pulmonary carcinogen in studies on coal
miners.  But lack of data on confounders such as asbestos and
smoking makes interpretation of this study difficult.  The
findings of this study are, at best, suggestive of diesel exhaust
being a lung carcinogen.

     The most definitive evidence for linking diesel exhaust
exposure to lung cancer comes from a railroad worker study
conducted in the United States  (Garshick et al.,  1988) which was
funded by EPA.  Relative risks of 1.57  (95 percent CI = 1.19 to
2.06) and 1.34  (95 percent CI = 1.02 to 1.76) were found for ages
40 to 44 and 45 to 49, respectively, after the exclusion of
workers exposed to asbestos.  This study also found that risk of
lung cancer increased with increasing duration of employment.
This large cohort study with lengthy follow up and adequate
analysis, including dose response (based on duration of
employment as a surrogate) as well as adjustment for other
confounding factors such as asbestos and smoking, makes the
observed association between increased lung cancer and exposure
to diesel exhaust more meaningful.

     Among the seven lung cancer case-control studies reviewed in
EPA  (1990b),  the study by Lerchen et al. (1987)  was the only one
that did not find increased risk of lung cancer,  after adjusting
for age and smoking, for diesel fume exposure.  The major
limitation of this study was lack of adequate exposure data
derived from the job titles obtained from occupational histories.
Next of kin provided the occupational histories for 50 percent of
the cases which were not validated.   The power of the study was
small  (analysis done on males only,  333 cases).   On the other
hand, statistically nonsignificant excess risks were observed for
diesel exhaust exposure by Williams et al.  (1977) in railroad
workers  (OR = 1.4) and truck drivers (OR = 1.34), by Hall and
Wynder (1984) for workers who were exposed to diesel exhaust
versus workers who were not (OR = 1.4 and 1.7 with two different
criteria),  and by Damber and Larsson (1987) in professional
drivers  (OR = 1.2).   These rates adjusted for age and smoking.
Both Williams et al.  (1977) and Hall and Wynder  (1984) had high
nonparticipation rates of 47 percent and 36 percent,
respectively.  In addition, the self-reported exposures used in
the study by Hall and Wynder (1984)  were not validated.  This
study also had low power to detect excess risk of lung cancer for
specific occupations.

     The study by Benhamou et al. (1988), after adjusting for
smoking,  found significantly increased risks of lung cancer among
French motor vehicle drivers (RR = 1.42) and transport equipment
operators  (RR = 1.35).  The main limitation of the study was the
inability to separate the exposures to diesel exhaust from those
of gasoline exhausts since both motor vehicle drivers and
transport equipment operators probably were exposed to the
exhausts of both types of vehicles.   Hayes et al. (1989) combined
data from three studies (conducted in three different states) to
increase the power to detect an association of lung cancer with
different occupations that had high potential for exposure to

                               9-27

-------
                                                        EPA-420-R-93-005
                                                            April 1993

diesel exhaust.   They found that truck drivers employed for more
than 10 years had a significantly increased risk of lung cancer
(OR = 1.5, 95 percent CI = 1.1 to 1.9).   This study also found a
significant trend of increasing risk of lung cancer with
increasing duration of employment among truck drivers.  These
relative odds were computed by adjusting for birth cohort,
smoking, and state of residence.  The main limitation of this
study is again the mixed exposures to diesel and gasoline
exhausts, since information on type of engine was lacking.
Potential bias may have been introduced since the way in which
the cause of death was ascertained for the selection of cases
varied in the three studies.  The methods used in these studies
to classify the occupational categories are different, hence
probably leading to incompatibility of occupational categories.

     In a case-control study by Steenland et al.  (1990) involving
truck drivers with at least 35 years of experience, the relative
odds ratio was 1.89.  This study also showed a dose-response
trend with the risk of lung cancer increasing with increasing
years of exposure when employment after 1959 was considered.  The
limitations of this study include possible misclassifications of
exposure and smoking, lack of levels of diesel exposure, smaller
exposed population, and insufficient latency period.

     The most convincing comes from the Garshick et al. (1987)
case-control study among railroad workers.  After adjustment for
asbestos and smoking, the relative odds for continuous exposure
were 1.39 (95 percent CI = 1.05 to 1.83).  Among the younger
workers with longer diesel exhaust exposure, the risk of lung
cancer increased with the duration of exposure after adjusting
for asbestos and smoking.  Even after the exclusion of recent
diesel exposure (5 years before death),  relative odds increased
to 1.43  (95 percent CI = 1.06 to 1.94).   This study appears to be
a well conducted and well analyzed case-control study with
reasonably good power.  Potential confounders were controlled
adequately,  and interactions between diesel exhaust and other
lung cancer risk factors were tested.

     Of the seven bladder cancer case-control studies, four
studies found increased risk in occupations with a high potential
diesel exhaust exposure.  A significantly increased risk of
bladder cancer was found in Canadian railroad workers  (RR = 9.0,
95 percent CI = 1.2 to 349.5; in Howe et al.,  1980) truck drivers
(OR = 2.9, Hoar and Hoover et al.,  1985) and in Argentinean truck
and railroad drivers  (RR = 4.32; Iscovich et al.,  1987).
Significantly increased risks were observed with increasing
duration of employment of >20 years in truck drivers  (OR = 12)
and railroad industry workers (OR = 2.21; Steenland and Burnett,
1987).  No significant increased risk was found for any diesel-
related occupations in studies by Wynder et al. (1985), Iyer et
al.   (1990),  and Steinbeck et al. (1990).  All these studies had
several limitations including inadequate characterization of
diesel exhaust exposure, lack of validation of surrogate measures
of exposure, and presence of other confounding factors  (urinary
retention, concentrated smoke within the truck cab, etc.); most
                               9-28

-------
                                                        EPA-420-R-93-005
                                                            April 1993

of them had small sample sizes, and none presented any latency
analysis.

     In summary, in regard to lung cancer which is the endpoint
used for the EPA unit risk, an excess risk of lung cancer was
observed in three out of seven cohort studies and six out of
seven case-control studies.  Of these studies, two cohort and two
case-control studies observed a dose-response relationship using
duration of employment as a surrogate for dose.  However, because
of the lack of actual data on exposure to diesel exhaust in these
studies and other methodologic limitations such as lack of
latency analysis, the evidence of carcinogenicity in humans is
considered to be limited for diesel exhaust exposure.

9.6.1.2 Weight-of-Evidence Judgement of Data and EPA
Classification

     Based upon the inductions of lung tumors in the three F344
rat studies, as well as the other research mentioned above and
supported by positive results for mutagenicity, the evidence for
carcinogenicity of diesel exhaust in animals is considered to be
sufficient based on U.S. EPA cancer assessment guidelines.

     Collectively, the epidemiological studies show a positive
association between diesel exhaust exposure and lung cancer.
However, because of the uncertainties due to limited exposure
data and low relative risk ratios in these populations, the
evidence for carcinogenicity of diesel engine emissions in humans
is considered to be limited.  This means that a causal
interpretation is credible, but alternative explanations such  as
chance, bias, or confounding factors cannot be ruled out.

     On the basis of limited evidence for carcinogenicity of
diesel engine emissions in humans, supported by sufficient
evidence in animals and positive mutagenicity data, diesel engine
emissions are considered to best fit the weight-of-evidence
category Bl.  Agents
                               9-29

-------
                                                        EPA-420-R-93-005
                                                            April 1993

classified into this category are considered to be probable human
carcinogens.

9.6.1.3  Data Sets Used for Unit Risk Estimates

     The most critical of the above mentioned animal studies are
those that involve a chronic inhalation exposure of diesel
particulate matter.  To actually determine the unit risk of this
particle, only three of the rat inhalation studies are selected
for risk calculations because each study consists of multiple
exposure groups and thus is more appropriate for risk
calculations.  The three studies used are Mauderly et al.   (1987),
Ishinishi et al.   (1986), and Brightwell et al.  (1986).  These
studies are summarized in Table 9-8.

     The EPA also attempted to use two epidemiological studies,
the Garshick et al. studies published in 1987 and 1988 to develop
a unit risk estimate. These are summarized in Table 9-9.  Though
a relationship exists between diesel exhaust exposure and the
incidence of lung cancer, both of these studies have strengths
and weaknesses.  These studies give a large sample size in a
relatively stable workforce, and also take into account the
confounding factors of smoking and asbestos exposure.  The major
weaknesses at this time are the limited qualitative and
quantitative data on the exposure of these individuals and a
short follow-up period.  The number of years of exposure to
diesel exhaust was used as a substitute for an actual dose so it
is difficult to accurately assess the amount of diesel exhaust
they were exposed to.

9.6.1.4 Dose-Response Model Used

     The linearized multistage model is used to calculate unit
risk estimates using various dose equivalence assumptions.  All
unit risk estimates that currently exist for diesel particulate
matter are based exclusively on animal data.

9.6.1.5 Unit Risk Estimates

     The approach that has been adopted by EPA in determining the
unit risk from diesel particulate matter is the one that
attributes the carcinogenicity to that of the particle itself
rather than the
organics.  The methodology used to develop the most recent EPA
quantitative risk estimate differs from other chronic bioassay
based estimates in several ways.  Unlike earlier estimates, the
present one uses a sophisticated dosimetry model to extrapolate
lung burdens of particulate matter from animal exposures to
humans.  This model accounts for species differences in
deposition efficiency, respiration rates, normal particle
clearance rates,  particle transport to lung associated lymph
nodes, and effect of particle overload upon clearance rates.
                               9-30

-------
                                                                                          EPA-420-R-93-005
                                                                                               April 1993
TABLE 9-8.  Animal Data Used for EPA's Unit Risk Estimates.
REPORT




Mauderly,
et al. (1987)

(Lovelace)





Ishinishi,
et al. (1986)

(JARI)












Brightwell,
et al. (1986)

(Battelle -Geneva)






ANIMAL




F344/Crl rats,
male and
female






F344/Jcl rats,
male and
female













F344 rats,
male and
female

pretreated 3d
prior to
exposure with
tumor promoter


PARTICLE
EXPOSURE
CONCENTRATION
AND TIME OF
EXPOSURE
0.35, 3.5, and
7 . 0 mg/m3

7h/d, 5d/week,
for 30 mo.




0, 0.1, 0.4,
1.0, or 2.0
mg/m3 from light
duty engine

0, 0.4, 2.0, or
4 . 0 mg/m3 from
heavy duty
engine


16h/d, 6d/week,
for 30 mo.



0.7, 2.2, or 6.6
mg/m3, filtered
and unfiltered


5-16h periods/wk
over 2 yrs.



ENGINE TYPE
AND CYCLE



1980 model
5.7L
Oldsmobile V8
run at FTP
hot start
certification
cycle


light duty
engines were
1.8L-4
cylinder,
swirl chamber
operated at
1200 rpm

heavy duty
engines were
11L-6
cylinder,
direct
injection
operated at
1700 rpm
1 . 5L engine
(no
manufacturer
given) using
U.S. 72 (FTP)
cycle




MAJOR RESULTS




Control and low level exposure shows
no statistical increase in lung tumors
(0.9% and 1.3% increase in tumors
respectively)

Medium and high level exposure shows
that the lung tumor incidence was
statistically higher (3.6% and 12.8%
increase in tumors respectively)
Light duty engine exposure:
carcinomas were dose-dependent with
the highest incidence in the 1.0 mg/m3
exposure (4.1% increase), there were
no significant changes between groups

Heavy duty engine exposure:
carcinomas were dose-dependent with
the highest incidence in the 4 . 0 mg/m3
exposure (6.5% increase), significant
changes were found between the 0 and
4 . 0 mg/m3 exposure groups




No significant increase was found in
either group at low or control
exposure

Medium, unfiltered exposure 4-15%
increase, and high, unfiltered
exposure 23-54% increase in lung
tumors
Dose-dependent, promoter had no
statistical effect
Table  9-9.  Epidemiological Data Used for EPA's Unit Risk  Estimates.
 STUDY
TYPE AND SUBJECTS
EFFECTIVE
START
PARAMETERS
OBSERVED
MAJOR  RESULTS
                                                9-31

-------
                                                                                 EPA-420-R-93-005
                                                                                     April 1993
 Garshick,
 et al.
  (1987)
Case-Control
study on male
railroad workers

each cancer death
(case)  was
matched to two
randomly selected
deceased workers
(within 3Id of
death)  with no
evidence of
cancer (control)
1959 (95% of
locomotives were
diesel)
Death from lung
cancer between
March 1, 1981
through February
28, 1982
Relative odds for
lung cancer is 1.5
for the highest
exposure category
(low odds but 95%
confidence interval
is narrow) and it
has been adjusted
for smoking and
asbestos exposure.

This study supports
the hypothesis that
occupational
exposure to diesel
exhaust increases
lung cancer.	
 Garshick,
 et al.
  (1988)
Cohort study on
male railroad
workers

(a group of
individuals
having a
statistical
factor in common
in an
epidemiological
study  [i.e.
diesel exposure])
                                same as above
                  Deaths due to
                  lung cancer from
                  1959 to 1980
                  Relative risk for
                  lung cancer is 1.5
                  (modest risk) and it
                  has been adjusted
                  for smoking and
                  asbestos exposure.

                  This study shows a
                  positive association
                  between occupational
                  diesel exhaust
                  exposure and a
                  modest increase in
                  lung cancer	
     A important feature of the dosimetry model is that it accounts for high dose
inhibition of particle clearance.  If this adjustment is not made, lung burden of
particulate matter will be overestimated during extrapolation to  low doses with an
accompanying overestimation of cancer potency.  Since most of the organics desorb from  the
particle surface even with normal clearance rates, inhibition of  particle clearance will
affect the concentration of organics only slightly.  Cancer potency estimates may
                                           9-32

-------
                                                        EPA-420-R-93-005
                                                            April 1993

therefore differ depending on whether they are based upon lung
burden of particles or organics.

     There are two reasons for using a particle based risk
assessment.  First of all, the concentration of PAH's on the
particles is quite small.  The quantity present at the exposure
levels used in the chronic bioassays is unlikely to be great
enough to produce the tumor response seen.  B[a]P is present at a
concentration of about 1 ug/gm particulate matter.  Secondly,
insoluble biochemically inert particles such as titanium dioxide
(Lee et al.,  1986) or activated carbon (Kawabata et al., 1986)
can induce lung cancer at very high concentrations.  Even more
significant were the findings of Mauderly et al.  (1991) and
Heinrich et al.  (1991) (see Section 9.6.3.2) that carbon black,
which is very similar to the carbon core of the diesel particle,
but contains essentially no adsorbed organics, induced lung
cancer at the same concentrations as diesel exhaust.  Additional
support for the predominance of the particle effects was also
contained in the report by Heinrich et al.  (1991).  In this
report, pyrolyzed pitch condensate, which does not have an
insoluble particle core,  but contains about 1000 fold greater
concentration of PAHs than diesel particles, is not that much
more potent than diesel exhaust in the induction of lung cancer.

     While this method is an improvement over previous ones, an
important uncertainty remains.  Particles deposited in the
alveolar regions are ingested by macrophages, which are then
induced to secrete a variety of cytokines, oxidants, and
proteolytic enzymes.  Some combination of these are thought to
act upon adjacent alveolar cells to induce tumor formation.  It
is uncertain if very low macrophage particle burdens will induce
release of these factors, or if there is a threshold for their
effects.  Use of a linearized multistage model to extrapolate to
low doses could result in an overestimate of risk.  Data,
however, are presently inadequate to prove or disprove this
possibility.   Thus, EPA still employs the conservative linearized
low-dose extrapolation model.

     The unit risks based on the long term rat inhalation studies
of Mauderly et al., (1987), Ishinishi et al.,  (1986), and
Brightwell et al.,  (1986) were calculated by EPA using the carbon
core only.  The availability of preliminary data from studies
discussed in Section 9.6.3.2 conducted on animals exposed to
carbon black, though not used in the risk calculations, did
influence the methodology.  A geometric mean of the three unit
risks was then

determined to be l.VxlO"5  (ug/m3)"1.  This unit risk was presented
at the Air and Waste Management Association meeting in October,
1991 (Pepelko and Ris, 1991d).   This unit risk is also presented
in the latest EPA diesel draft document (unpublished).  It has
yet to undergo Science Advisory Board (SAB) review and thus is
subject to change.  EPA is also in the process of developing a
unit risk estimate that attempts to account for both particle and
organic effects.
                               9-33

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Pepelko and Ris  (1991d) also discussed the attempt to
develop a unit risk estimate for lung cancer based on human
epidemiological data using Garshick et al.,  (1988).  Using data
from this study, the EPA carried out more than 50 analyses of the
relationship between diesel exhaust exposure and tumor incidence.
None of these analyses demonstrated a pattern that was consistent
with an association between diesel exhaust exposure and lung
cancer.  The inability to obtain an adequate dose response was
attributed to the limitations regarding exposure estimates for
the various job categories, coupled with the small increases in
lung cancer mortality.  Consequently, it was concluded that the
data are inadequate for quantitative risk assessment.

9.6.2 Other Views and Unit Risk Estimates

     This section presents alternative views and/or risk
assessments for diesel exhaust particulate matter.  These
alternative risk assessments are summarized in Table 9-10.

Comparative Potency Method

     The comparative potency method is a method developed by EPA
to predict human cancer risk from mutagenicity and animal
bioassay data.  The comparative potency method was developed
because of a lack of chronic animal bioassays and a need to
develop a potency estimate in the early 1980's.  It has been
applied to the polycyclic organic matter  (POM) from selected
emission sources, including diesel vehicles (Albert et al., 1983;
Lades,  1991).   POM is a general term referring to a complex
mixture of polycyclic aromatic compounds generally associated
with the particles or soot of emissions, and derived from the
combustion of fossil fuels, vegetative matter, and synthetic
chemicals.

     In this comparative potency method, the risk of diesel
particulate matter is estimated by comparison of diesel
particulate matter bioassay potencies to the bioassay potencies
of known human carcinogens  (coke oven, roofing tar, cigarette
smoke)  according to the following equation:
                               9-34

-------
                                                               EPA-420-R-93-005
                                                                    April 1993

Table 9-10. Comparison  of Diesel  Exhaust Particulate  Matter
                 Inhalation Unit Risk Estimates.
Source
Albert et al . (1983)
Lewtas (1991)
Albert et al . (1983)
Lewtas (1991)
Harris (1983)
Cuddihy et al .
(1984)
Albert and Chen
(1986)
Pott and Heinrich
(1987)
McClellan et al .
(1989)
Smith and Stayner
(1990)
Harris (1983)
McClellan et al .
(1989)
Method
Comparative potency
method using extracted
organics from one light-
duty (LD) diesel engine
same as above, using an
average of three LD
engines
Comparative potency
methodb
Comparative potency
methodb'c
Multistage model, lung
cancer in ratsd
Straight line
extrapolation, lung
cancer in ratse
Logistic regression,
lung cancer in ratsf
Time -to -tumor model,
lung cancer in ratsd
Epidemiological
analy s i s , London
Transport Study (Waller,
1981)
Epidemiological
analysis, Railroad
workers (Garshick et al .
1987)
Cancer Unit
Risk Estimate
(ug/m3)-1
Upper Bounda
3.5xlO"5
2 .6xlO"5
2 .9xlO"4
7.0xlO"5
1.2xlO"5
6.0-12 .OxlO"5
S.OxlO"5
1.5-3. OxlO"5
4.1xlO"3
0.6-2. OxlO"3
Estimated upper bound of  lifetime risk of continuous  exposure to 1 ug/m3 diesel
exhaust particulate matter.
bUsed data from studies by Albert et al.   (1983).
cUsed data from studies by Harris  (1983).
dUsed data from studies by Mauderly et al. (1987).
eUsed data from studies by Brightwell et al.  (1986)
Mauderly et  al.  (1987).
Used data from studies by Brightwell et al.  (1986),  Ishinishi et al.  (1986),
Iwai et al.  (1986), and Mauderly et al.  (1987).
                                               Heinrich et  al.  (1986a),  and
                                 9-35

-------
                                                        EPA-420-R-93-005
                                                             April 1993
  unit risk (diesel) = unit risk (known care.
                                         bioassay potency (diesel]
                                       bioassay potency (known care.)
The term in brackets is the ratio of  the  slopes  of  the dose
responses from the same bioassay, and is  referred to as the
relative potency.  The underlying assumption  of  the comparative
potency method is that the relative potency is constant across
different bioassay systems.  The equation above  was applied using
the extract of a light duty Nissan engine in  the mouse skin tumor
initiation bioassay to estimate the risk  of diesel  particulate
matter.  The mouse skin tumor initiation  bioassay was chosen
because the relative potencies of the known human carcinogens
obtained with this bioassay correlated well with the relative
potencies obtained with the human data.   The  mouse  skin tumor
test was also used because it gave a  strong dose-response in the
Nissan engine.

     Extracts from particle samples from  three light-duty diesel
vehicles and one heavy-duty diesel engine were used.   The unit
risk estimates for two other light duty engines  and a heavy duty
engine were derived by comparing their potencies with that of the
Nissan engine using three short-term  tests.   The average lifetime
risk from the three light-duty diesel samples across the three
comparative human carcinogens was 2.3xlO~4(ug  organic matter/m3)"1
or 2.6xlO~5(ug particles/m3)"1.  The lifetime cancer  risk/ug
particles/m3  ranged from l.SxlO"6 for  the  heavy duty engine,  to
3.5xlO"5 for the most potent light duty diesel engine  (Albert et
al., 1983; Lewtas, 1991).

     The comparative potency method predicted a  human lung cancer
unit risk very similar to the unit risk estimate for diesel
particulate matter that has been recently extrapolated from three
rodent inhalation studies.  The lifetime  unit risk  for the rodent
studies is the same one cited earlier, 1. VxlO"5 (ug/m3) "1.   This
compares to 2 . 6xlO"5 (ug/m3) "1 by the comparative potency method.
This demonstrates that these two independent  approaches to cancer
risk from diesel emissions result in  very similar cancer unit
risk estimates.

     Harris (1983) developed comparative  potency estimates for
the same four engines used by Albert  et al.,  (1983)  but used only
two epidemiological based potency estimates,  those  for coke ovens
emissions and for roofing tar.  Harris (1983) also  used
preliminary data for three of the same assays as did Albert et
al., (1983).   After making adjustments to adjust for lifetime
exposure the Harris  (1983) overall estimated  unit risk value was
2.9xlO"4  (ug/m3)"1 for the three light-duty engines.
                               9-36

-------
                                                        EPA-420-R-93-005
                                                             April 1993

     Cuddihy et al.  (1984) reported a unit  risk  of  about  V.OxlO"5
(ug/m3)"1 using a comparative potency method similar to those
reported in the preceding paragraphs.  The  data  base  was  similar
to that used by Albert et al.  (1983) and Harris  (1983) .

     The comparative potency method suffers from two  major
uncertainties.  The first being that mutagenicity is  not  a
reliable predictor of carcinogenicity.  Secondly, the relative
cancer  potency of diesel to the other agents used  may be much
different than relative potency in short-term.

Alternate Risk Estimates Derived From Rat Data

     With the availability of  chronic cancer bioassays, more
recent assessments were based  on lung tumor induction in  rats.
Albert and Chen (1986) reported a risk estimate  based upon the
chronic rat bioassay conducted by Mauderly  et al.  (1987).   Using
a multistage model and assuming equivalent  deposition efficiency
in humans and rats, they derived an upper bound  for a lifetime
risk of 1.2xlO"5  (ug/m3)"1.  Pott and Heinrich  (1987) used  a linear
extrapolation, including data  reported by Brightwell  et al.
(1986), Heinrich et al.  (1986a), and Mauderly et al.  (1987).
They reported risk estimates of 6.0xlO"5 to  12.0xlO"5  (ug/m3)"1.
Most recently, Smith and Stayner  (1990), using a time-to-tumor
model based on the data of Mauderly et al.  (1987),  derived and
upper bound of 1.5xlO"5 to 3.0xlO"5  (ug/m3)"1.   In  McClellan et al.
(1989) a logistic regression was used with  the data from
Brightwell et al.  (1986), Ishinishi et al.  (1986),  Iwai et al.
(1986), and Mauderly et al.  (1987) to derived a  unit  risk
estimate of S.OxlO"5  (ug/m3)"1.

Alternate Risk Estimates Derived From Epidemiological Data

     Harris (1983) also assessed the risk of exposure to  diesel
engine emissions using data from the London Transport Worker
Study by Waller (1981).  Five  groups of employees from the study
were used (one high exposure,  two intermediate exposure,  and two
with no exposure.  Harris (1983) compared the exposed groups with
internal controls.  He merged  the three exposed  groups and
compared them with the two groups considered to  be  unexposed.   An
adjustment was made for the greatest exposure groups.  Using this
method, the relative risk of the exposed groups  was greater than
1, but was statistically significant for only the highest
exposure groups from 1959 to 1960.

     Harris (1983) identified  a variety of  uncertainties  in the
assessment.   Taking the uncertainties into  account, he derived a
maximum likelihood estimate of l.OxlO"3  (ug/m3)"1  and a upper bound
of 4.1xlO"3  (ug/m3)"1.

     McClellan et al. (1989) developed risk estimates based on
the Garshick et al.  (1987) study in which lung cancer in  railroad
workers was evaluated.  Using  a proportional risk model the
values
of a lifetime risk of exposure to diesel exhaust ranged from
0.6xlO"3 to 2.0xlO"3  (ug/m3)"1.

                               9-37

-------
                                                        EPA-420-R-93-005
                                                            April 1993
Environ
     The Environ report  (Environ, 1987),  prepared for the Motor
Vehicle Manufacturers Association, was in response to the EPA
report on air toxics (Carey, 1987).   In its report, EPA uses a
range of potency estimates, 2.0 x 10"5  to 1.0 x 10"4 (ug/m3) "\  for
estimating the lung cancer risk from diesel exhaust.  The range
of potencies all involved use of the comparative potency method,
described above.  This Environ  (1987)  report as well as Carey
(1987) were completed while animal inhalation studies were still
in progress, so neither report evaluates more recent data.

     Environ, for several reasons, questions the validity of the
comparative potency approach.  Although there is some evidence
that activity in short-term genotoxicity tests may be indicative
of carcinogenic potential, Environ cites studies that show the
correlation between genotoxicity and carcinogenicity was just
60%.  Also, quantitative correlations  have not been established
for any individual carcinogens let alone a complex mixture such
as diesel particles.  Environ believes this renders the procedure
scientifically unsound.

     Environ raises further questions  regarding the validity of
the procedure by challenging the fact  that it is based solely on
extracts of diesel particles and ignores substances in the
emissions that are not associated with the particle or are not
extractable.  Environ also states that organics adsorbed onto the
particle may not be bioavailable so using extract overestimates
the potency.

     Environ states that these factors may increase the
uncertainty associated with the Carey  (1987) risk estimates for
diesel emissions.  They then offer no  new or additional data
analysis to support their claims or a  unit risk estimate of their
own.

International Agency for Research on Cancer (IARC)

     IARC  (IARC, 1989), does not estimate potencies for
carcinogens, but has classified diesel engine exhaust into cancer
weight-of-evidence Category 2A.  Agents classified into Category
2A are considered to be probable human carcinogens.  This
classification is based on limited evidence for carcinogenicity
in humans.  This is supported by sufficient evidence for
carcinogenicity in animals with whole  diesel engine exhaust and
in animals with extracts of the diesel engine exhaust particles.
IARC considers the evidence for the carcinogenicity in animals of
the gas-phase of diesel engine exhaust (with particles removed)
to be inadequate.  EPA did not develop separate weight-of-
evidence evaluations for the gaseous and particle phases of
diesel exhaust.

National Institute for Occupational Health and Safety (NIOSH)

     In 1986, NIOSH evaluated the health effects of diesel
exhaust in Evaluation of the Potential Health Effects of

                               9-38

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Occupational Exposure to Diesel Exhaust in Underground Coal
Mines.   This document describes the short-term effects as well as
stating that there is a causal association between exposure to
whole diesel exhaust and cancer.

     In a later publication, NIOSH  (1988),  the recent animal
studies in rats and mice discussed previously in Section 9.6.1
(Brightwell et al.  1986; Heinrich et al.  1986; Ishinishi et al.
1986; Iwai et al. 1986; Mauderly et al. 1987) are used to confirm
an association between the induction of cancer and exposure to
whole exhaust.  The lung is the primary site identified with
carcinogenic or tumorigenic responses following inhalation
exposures.  Limited epidemiological evidence  (Edling et al. 1987;
Garshick et al.  1987, 1988) suggests an association between
occupational exposure to diesel engine emissions and lung cancer.
The consistency of these toxicologic and epidemiologic findings
suggests that a potential occupational carcinogenic hazard exists
in human exposure to diesel exhaust.

9.6.3  Recent and Ongoing Research

9.6.3.1 Metabolism and Pharmacokinetics

     Much of the information in this section is summarized from
the more detailed report, the Draft Health Assessment Document
for Diesel Emissions (EPA 1990b).   To examine this information
further please refer to the document mentioned above.

     Several studies affirm the bioavailability from inhaled
diesel  exhaust particles of compounds such as B[a]P and 1-
nitropyrene (1-NP)  which are known to be carcinogenic or
mutagenic.  Biotransformation of B[a]P, 1-NP, and some of the
dinitropyrenes to reactive intermediates following inhalation of
diesel  exhaust particles has been verified.  Furthermore, several
reports have provided data indicating the formation of DNA
adducts, considered an underlying mechanism of carcinogenicity,
following administration of these compounds.  The development of
lung tumors in experimental laboratory animals following chronic
exposures to particulate diesel exhaust occurs under conditions
in which alveolar macrophage-mediated particle clearance from the
lung is compromised.  Although tumors have also been found to
develop with other types of particles  (e.g., titanium oxide) when
this clearance mechanism is diminished, tumors developing in the
lungs of diesel emissions-exposed rats with smaller lung mass or
comparatively less volume burden of diesel particles suggest that
the carcinogenic response is not exclusively related to an
overabundance of the particles in the lungs per se.  Therefore,
the organic components on diesel particles may be importantly
involved in the development


of lung tumors.   The lung's pulmonary macrophages, which
phagocytize deposited diesel particles, probably participate in
the gradual in situ extraction and metabolism of procarcinogens
associated with the diesel particles.  Additionally, the normal
tumoricidal activities of the pulmonary macrophages may be

                               9-39

-------
                                                        EPA-420-R-93-005
                                                            April 1993

compromised upon interaction with excessive numbers of diesel
particles, and diesel particle-macrophage interactions could lead
to the generation of reactive oxygen species that have been shown
to be at least mutagenic.   Alternatively, there is evidence that
particles with a very large surface area/unit volume, such as
diesel particles or carbon black, can stimulate production of
harmful products by macrophages at much lower lung burdens,
perhaps even at lung burdens insufficient to inhibit clearance.
Processes and potential mechanisms discussed herein have largely
been derived from animal data, and further research is required
to determine how the activities of human pulmonary macrophages in
response to particulate diesel exhaust compare with pulmonary
macrophages from experimental animals.  Most importantly, valid
dosimetry for the human condition will require the elucidation of
the underlying mechanisms involved in the development of lung
tumors following chronic exposure to whole diesel exhaust.

     An understanding of the pharmacokinetics associated with
pulmonary deposition of diesel exhaust particles and their
adsorbed organics is critical in understanding the carcinogenic
potential of diesel engine emissions.  The pulmonary clearance of
diesel exhaust particles is multiphasic and involves several
processes including a relatively rapid mucociliary transport and
slower macrophage-mediated processes.  The observed dose-
dependent increase in the particle burden of the lungs is due, in
part, to an overloading of alveolar macrophage function.  The
resulting increase in particle retention has been shown to
increase the bioavailability of particle adsorbed mutagenic and
carcinogenic components such as B[a]P and 1-NP.  Experimental
data also indicate alveolar macrophage-mediated metabolism and
phagolysosomal solubilization of particle-adsorbed components.
Although macromolecular binding of diesel exhaust particle-
derived PAH and the formation of DNA adducts following exposure
to diesel exhaust have been reported, a quantitative relationship
between these and increased carcinogenicity is not available.

     In addition to the aforementioned points, one must also
consider the fact that other compounds (e.g., gas-phase chemical
irritants) may alter respiratory rate and, therefore the actual
inhaled dose of potentially toxic components.  Moreover, a better
knowledge of particle dissolution rate and particle removal rate
is necessary for more accurately assessing bioavailability of
potentially carcinogenic components of diesel exhaust.
                               9-40

-------
                                                        EPA-420-R-93-005
                                                            April 1993
9.6.3.2 Carcinogenicity - Animal Studies
     In a recent study by the Fraunhofer Institute  (Heinrich et
al.  1991) female Wistar rats were exposed to carbon black  (CB),
tar/pitch condensate (yielding PAH in the form of benzo(a)pyrene
[BaP]),  and mixtures of these compounds to assess their
carcinogenic effect.  Exposure time for all groups was 17
hrs/day, 5 days/week, for 10 and 20 month periods.  A 17%
increase in lung tumors incidences was reported following
exposure of rats to carbon black particles for 10 months at a
concentration of 6 mg/m3,  then clean air for the  remainder of
their lifetime.  This is very close to the TLV value.  The
tar/pitch condensates (BaP)  at 20 or 50 ug/m3 gave increases in
lung tumors of 4% and 39% at 10 months and 33% and 97% at 20
months,  respectively.  The mixture  (2 or 6 mg/m3  CB and 50 ug/m3
BaP) test results were only for the 10 month period, and the
increases ranged from 72 to 89%.  In the abstract submitted, the
authors claim carbon black in this experiment induced almost the
same lung tumor rates in rats as diesel soot did in Heinrich et
al.  (1986a).   No inferences were made concerning the greater
increases in tumor incidences with pitch plus carbon black than
with carbon black alone.  This study shows that particles alone,
devoid of organics are capable of inducing lung cancer.  The
study also shows that the effects of particles are enhanced by
the presence of BaP.  It should be noted, however, BaP
concentrations in this study were much greater than those present
on diesel particles.

     In a study by Pott et al.  (1991), also from the Fraunhofer
Institute,  lung tumors were observed in female Wistar rats
intratracheally instilled with various non-fibrous and fibrous
dusts.   The percent of rats with primary lung tumors ranged from
a  60 to 66% increase following exposure to 30 to 60 mg diesel
soot (two types) or carbon black.  The rats exposed to 45 mg of
one type of diesel soot and those exposed to 45 mg of carbon
black gave identical rates of 65%.  This appears to support the
contention that the carbon particle itself is the carcinogen.
The abstract does not detail the differences in the two diesel
soot types tested, nor does it provide data regarding the amounts
of types of particle bound organics.

     In a recent and as yet uncompleted Health Effects Institute
study by Mauderly et al.  (1991) of the Inhalation Toxicology
Research Institute  (ITRI), there was a direct comparison of
carbon black particles and diesel exhaust.  The study exposed
F344/N rats 16 hours/day, 5 days/week for 24 months to carbon
black (2.5 or 6.6 mg/m3)  or  diesel exhaust (2.4 or 6.4  mg/m3).
They were sacrificed at 3, 6, 12, 18 and 23 months with remaining
rats held for post-exposure observation.  The results at this
time are interim, but the responses to diesel exhaust and carbon
black were qualitatively similar.  Diesel exhaust caused a
greater response than carbon black in lung weight  (increased),
lung burden of retained particles, and lung inflammation and
cytotoxicity.  Diesel exhaust and carbon black caused
approximately similar responses in body weight (decreased), lymph
node burden of retained particles and mortality.   The numbers of

                               9-41

-------
                                                        EPA-420-R-93-005
                                                            April 1993

lung tumors observed grossly at necropsy were nearly identical
for diesel exhaust and carbon black.  Observations to date do not
suggest that there is a difference between diesel exhaust and
carbon black in lung tumor type, multiplicity or growth in nude
mice.  Mauderly et al.  (1991), in the interim, concludes that the
information at this time suggest that soot-associated organic
compounds do not play a significant role in the pulmonary
carcinogenicity of diesel exhaust in rats.

     Dr. Werner Stober of the Fraunhofer Institute of Technology
and Aerosol Research, whose research is funded in part by the
German automobile manufacturers, states his position in several
papers.  This position is outlined in Stober  (1987, 1989) and
more recently by Stober (1991),  in which the present
epidemiological data, comparative potency method, and animal
inhalation studies are evaluated.

     A search of the scientific literature for data on the health
effects (especially carcinogenic risk)  of inhaled diesel
emissions was performed.   Dr. Stober states that this search
provides some very weak and disputed epidemiological evidence
suggesting that, at certain occupational exposures, there may
have been a health hazard for certain workers.  The major
confounding factor in all of these investigations is the
influence of the cancer statistics of cigarette smoking.  He
states that it is most likely the residual effects which are
attributed to occupational diesel exposure are due to surrogate
and incomplete information about the smoking habits of the
cohorts.  He finds it interesting that the supporters of
occupational risk from diesel emissions do not propose a risk to
the general populations at the present level of diesel emissions.

     Stober (1989) also states that the labeling of diesel
exhaust as a potential or probable carcinogen by Germany, the
World Health Organization (IARC), and the U.S. EPA were made
without any reference to the evaluation of risk.  He states that
the risk determined by the epidemiological studies should not be
used for the general public.  The risk for the public at large,
at present levels of diesel emissions,  is actually non-existent,
according to Dr. Stober.   He compares the lifetime carcinogenic
risk from 1 ug/m3  diesel  emissions  to the  risk of being struck by
lightning in Germany (a lifetime risk of 2:100,000 or 2.0 x 10"
5) .

     Stober (1989) does concede that if a genotoxic mechanism can
be shown to play a significant role in experimental tumor
induction, then a small residual risk may be assumed to have been
obscured by the uncertainties of past epidemiological studies.
In that case,  proper development and implementation of target
control strategies is advisable to lower the cancer risk.
                               9-42

-------
                                                        EPA-420-R-93-005
                                                            April 1993

     Stober (1989) states further that he does not imply that it
is unnecessary to regulate diesel exhaust emissions today.  The
future growth of unregulated diesel-powered vehicles would
degrade the present particle levels, and the general public will
resent this deterioration.  But there is only a very low
probability, if any, that this issue involves more than an
insignificantly small residual risk of a health effect.

     In Dr. Stober's Air and Waste Management Association
presentation in October, 1991  (Stober 1991),  he mentions several
points regarding the limitations of the rat studies.  At present,
the rat data suggest that a threshold may exist for the exposure
to diesel particulate matter and the appearance of tumors.  He
goes on to further state that the particle overload at the two
highest levels where the tumors occurred is also an issue.  Using
the preliminary data from recent ITRI and Fraunhofer research
(see Section 9.6.3 for details), he states that it shows
qualitatively similar results or more pronounced responses from
carbon black than diesel particulate matter.   He proposes a
possible epigenetic mechanism of tumor induction (i.e., the
tumors have not been caused by a metabolic degradation of organic
matter associated with the particles, but by the particle
deposits itself).

9.7  Carcinogenic Risk

     Urban and rural diesel particulate matter carcinogenic
risks, expressed as annual cancer deaths, were calculated
following the methodology discussed in Section 4.1, based on the
HAPEM-MS exposure estimates from Section 9.5.1 and the EPA unit
risk estimate given in Section 9.6.1.5.  The resultant urban,
rural, and total cancer deaths are given in Table 9-11.  These
cancer incidences are upper bound estimates and the risk may be
less,  but is unlikely to be more.

     Table 9-11.  Diesel Particulate Matter Cancer Deaths.a'b
Year
1990
1995
2000
2010
Urban
92
56
33
23
Rural
17
10
6
4
Total
109
66
39
27
     ""Projections  have inherent uncertainties in emission
     estimates, dose-response, and exposure.
     bCancer deaths  are based on the EPA 1991 draft unit risk,
     determined using animal data.  This unit risk has not been
     peer reviewed and is subject to change.
                               9-43

-------
                                                        EPA-420-R-93-005
                                                            April 1993


9.8  Non-carcinogenic Effects of Inhalation Exposure to Diesel
     Particulate Matter

     Since the focus of this report is on the carcinogenic
potential of the various compounds, the noncancer information
will be dealt with in a more cursory fashion.  No attempt has
been made to synthesize and analyze the data encompassed below.
Also, no attempt has been made to accord more importance to one
type of noncancer effect over another.  The objective is to
research all existing data, describe the noncancer effects
observed, and refrain from any subjective analysis of the data.

Diesel Particulate Matter

     The symptoms of acute (short) exposure to high levels  (i.e.
above ambient)  of diesel exhaust have been detailed through the
study of occupationally exposed workers.  These workers include
underground miners, bus garage workers, dock workers, and
locomotive repairmen.  The symptoms may be manifested as one or
more of the following:  mucous membrane and eye irritation,
headache, light-headedness, nausea, vomiting, heartburn,
weakness, numbness and tingling in extremities, chest tightness,
and wheezing.  The odors associated with diesel exhaust emissions
also cause some effects, such as nausea, headache, and loss of
appetite.

     Even though this appears to be a formidable list of
symptoms, the effects of a short-term diesel exhaust exposure are
dissipated as soon as the exposure stops or the subject leaves
the area.  Any of the changes in respiratory symptoms and
pulmonary function over the course of a workshift were generally
found to be minimal.

     The chronic (long-term)  exposure to diesel exhaust emissions
have also been followed in occupationally exposed workers, but
the data are insufficient to make a correlation between the
effects and the exposure experienced.  Most of the chronic
exposure data are derived from the use of animal studies.

     Many of the changes observed in rats and other small animals
exposed to diesel exhaust affect the cellular and structural
make-up of the lung.   These effects include:  accumulation of
particles in the lungs, increased lung weight, tissue
inflammation, increased macrophages and leukocytes  (white blood
cells),  macrophage aggregation, hyperplasia  (excess cell
formation) in the alveolar and bronchiolar epithelium (surface
cell layer),  and a thickening of alveolar septa (partitions).  In
some studies, a reduction in the growth of animals is also
observed at 2 mg/m3 for 16h/day,  and alterations  of  pulmonary
function parameters were seen at 2 to 6 mg/m3.

     All of these changes appear to be dependent on the
concentration of the exhaust particulate matter,  the pulmonary
deposition of the particle, and the ability of the lung to clear
the particulate matter.

                               9-44

-------
                                                        EPA-420-R-93-005
                                                            April 1993

Reference Concentration for Chronic Inhalation Exposure  (RfC)

     The reference concentration for chronic inhalation exposure
(RfC) for diesel particulate matter has recently been established
(EPA, 1993) .   This RfC was determined to be 5.0xlO"3 mg/m3 per
day, over a lifetime.  An RfC is an estimate of the continuous
exposure to the human population that is likely to be without
deleterious effects during a lifetime.  As such, it is useful in
evaluating non-cancer effects.

     The two critical studies used in determining the diesel
particulate RfC are chronic rat inhalation studies by Mauderly et
al.   (1998)  and Ishinishi et al.  (1988).  These two studies
observed various non-cancer endpoints and at various time points.

     In Mauderly et al.  (1988),  rats and mice were exposed to
target diesel particulate matter concentrations of 0, 0.35, 3.5,
or 7.0 mg/m3  for 7  hours/day,  5  days/week for up to 30  months for
the rats or 24 months for the mice.  Endpoints examined in this
study include carcinogenicity, respiratory tract histopathology
and morphometric analysis, particle clearance, lung burden of
diesel particulate matter, pulmonary function testing,  lung
biochemistry, lung lavage biochemistry and cytology, immune
function, and lung cell labeling index.  Aggregates of particle-
laden macrophages were seen after 6 months in rats exposed to 7.0
mg/m3 target  concentrations  and,  after 1  year of exposure,
histological changes were seen including focal areas of
epithelial metaplasia.  Fibrosis and metaplasia increased w