-------
-------
VIP-34
Measurement of Toxic and Related Air Pollutants
Proceedings of the 1993 U.S. EPA/A&WMA
International Specialty Conference
Durham, North Carolina
May 1993
Report Number EPA/600/A93/024
Publication Policy
This publication contains technical papers published essentially as they were presented at a
recent A&WMA/U.S. EPA Specialty Conference. The papers have not been subject to the
editorial review procedures of the Association, and opinions expressed therein are not to be
interpreted as having the endorsement or support of the Association.
Notice
The information in this document has been funded wholly or in part by the United States
Environmental ProtectionAgencyunderacost-sharing agreement to Air &Waste Management
Association. It has been approved for publication as an EPA document. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
Copies of this book are available from the Air & Waste
Management Association for $90 (Association members
$60). For a complete publications listing, contact the
Order Fulfillment Clerk, A&WMA, One Gateway Center,
3rd Floor, Pittsburgh, PA 15222, (412) 232-3444, Fax (412)
232-3450.
Copyright 1993
Air & Waste Management Association
One Gateway Center, 3rd Floor
Pittsburgh, PA 15222
-------
Contents
Preface xv
Conference Committee xvi
Session 1
Radon Indoor Air Pollution
Evaluation of Radon Movement Through Soil and Foundation Substructures Marc 3
Y. Menetrez, Ronald B. Mosley, Richard Snoddy, et al.
Comparison of Measurement Techniques for Soil Permeability in EPA's Soil-Gas 9
Chamber Ronald B. Mosley, Marc Y. Menetrez, D. Bruce Harris, et al.
Radon in Florida Large Building Study Marc Y. Menetrez, Russell N. Kulp, 15
D. Bruce Harris, et al.
An Analytical Solution to Describe the Pressure/Flow Relationship in EPA's Soil- 21
Gas Chamber Ronald B. Mosley
Session 2
Air Pollution Dispersion Modeling
Uncertainties of Using Short-Term Ail Quality Concentrations to Estimate Annual 29
Average Concentrations Norman A. Huey and George J. Schewe
Further Development of Empirical Factors for Estimating Air Toxic Impacts George 35
J. Schewe and Norman A. Huey
Evaluating the Impact of Subsurface Contaminants on Indoor Air Quality Using 41
Estimates From an Advective-Diffusive Transport Model Wen-Whai Li and
Marybeth Long
Preliminary Investigation of Uncombusted Auto Fuel Vapor Dispersion Within a 52
Residential Garage Microenvironment Azzedine Lansari, John J. Stretcher, Alan
H. Huber, et al.
Modelling Ozone Deposition Onto Indoor Surfaces Richard Reiss, P. Barry Ryan, 58
Petros Koutrakis, et al.
Exposure Modeling of Acid Aerosols Michael P. Zelenka and Helen H. Suh 64
-------
Session 3
Measurement of Semi-Volatile Organic Compounds
The Distribution of Semi-Volatile Organic Compounds Between the Vapor and 73
Paniculate Phases: Diffusion Denuder Measurements DelbertJ.Eatough,Hongmao
Tang, Wenxuan Cui, et al.
Use of a High-Volume Small Surface Sampler (HVS3) for the Microbiological 82
Evaluation of Dust from Carpeted and Non-Carpeted Surfaces K. E. Leese,
R. M. Hall, E. C. Cole, et al.
Methods for Polycyclic Aromatic Hydrocarbons and Tobacco Smoke Markers in 88
House Dust: Laboratory and Field Evaluation Jane C. Chuang, Patrick J. Callahan,
Sydney M. Gordon, et al.
Sampling, Supercritical Fluid Extraction (SFE) and GC/MS Analysis of Indoor Air 94
Semi-Volatile Toxic Organics V. M. Kanagasabapathy, R. W. Bell, P. Yang, et al.
A Discussion of Volatile Organic Compound Sampling and Analysis Using a Mobile 100
GC/MS H. Yoest, T. Russell, M. Chu, et al.
Method 301 Field and Laboratory Validation of Target Semi-Volatile Organic 111
Compounds Using SW-846 Method 0010 Theresa Russell, Helen Yoest, Jesus
Peralta, et al.
Hydrocarbons in the Range of C10-C20 Emitted From Motor Vehicles; Diesel Versus 123
Spark Ignition B. Zielinska, J. Sagebiel, L. H. Sheetz, et al.
Modeling the Atmospheric Formation and Decay of Gas and Particle Bound Nitro 129
Polycyclic Aromatic Hydrocarbons Zhihua Fan, Richard M. Kamens, Danhua Chen,
etal.
Effect of Combustion Temperature on the Atmospheric Stability of Polybrominated 135
Dibenzo-p-Dioxins and Dibenzofurans Parag Birla and Richard M. Kamens
A Method for Studying Heterogeneous Photochemical Reactions of Polycyclic 141
Aromatic Hydrocarbons on Atmospheric Combustion Aerosols Elizabeth Ann
Hayes, Stephen R. McDow, and Richard M. Kamens
Mechanistic and Kinetic Studies of the Photodegradation of Benz(a)anthracene in the 147
Presence of Methoxyphenols Jay R. Odum, Steven R. McDow, and Richard
M. Kamens
-------
Session 4
Measurement and Monitoring of Toxics, O3 and PAN
Requirements for the Establishment of Enhanced Ozone Monitoring Networks Nash 155
O. Gerald, William F. Hunt, Jr., Geraldine Dorosz-Stargardt, et al.
Surveys of the 189 CAAA Hazardous Air Pollutants: I. Atmospheric Concentrations 161
in the U.S. Thomas J. Kelly, MukundRamamurthi, Alberts. Pollack, et al.
Surveys of the 189 CAAA Hazardous Air Pollutants: H. Atmospheric Lifetimes and 167
Transformation Products Thomas J. Kelly, Mukund Ramamurthi, Albert J. Pollack,
etal.
Session 5
Integrated Air Cancer Project
The Integrated Air Cancer Project: Overview of Roanoke Study and Comparison to 175
Boise Study J. Lewtas, D. B. Walsh, C. W. Lewis, et al.
A Comparison of Air Quality Measurements in Roanoke, VA, and Other Integrated 185
Air Cancer Project Monitoring Locations R. K. Stevens, A. J. Hoffman, J. D. Baugh,
etal.
Mutagenicity of Indoor and Outdoor Air in Boise, Idaho and Roanoke, Virginia 190
Debra Walsh, Sarah Warren, Roy Zweidinger, et al.
Radiocarbon Measurements of Extractable Organic Matter from the Integrated Air 197
Cancer Project Study in Roanoke, VA Donna B. Klinedinst, George A. Klouda,
Lloyd A. Currie, etal.
Source Apportionment of Fine Particle Organics and Mutagenicity in Wintertime 207
Roanoke Charles W. Lewis, Roy B. Zweidinger, Larry D. Claxton, et al.
Comparison of Residential Oil Furnace and Woodstove Emissions Robert 213
C. McCrillis, Randall R. Watts, and Roy B. Zweidinger
Session 6
Indoor Air Quality in Highly Confined Environments
Strategies Used to Manage Air Quality in Manned Spacecraft John T.James 221
Air Quality Monitoring in Spacecraft: Present and Future Thomas Limero, Hector 227
Leano, andJ. T. James
-------
Submarine Air Quality: Relationship to Human Body Burden Hugh J. O'Neill, 237
Sydney M. Gordon, Louise M. Brousek, et al.
Cabin Air Quality Aboard Commercial Airliners Niren L. Nagda, Michael 243
D. Koontz, and Roy C. Fortmann
Removal of Gaseous Indoor Air Contaminants by Commercial Air Filters P. R. 249
Nelson, R. B. Hege, F. W. Conrad, et al.
Session 7
Measurement Methods Development
Fast Analysis of C2-Cj2 Ambient Air Hydrocarbons Using a Multi-Column Gas 257
Chromatographic System Kochy Fung
Testing and Evaluation of 2 Prototype Devices for Direct Measurements of Air-Water 263
Transfer Processes Involving Toxic Chemicals W. Schroeder, D. Mackay,
A. Cassamalli, etal.
A Universal Air Preconcentrator for Automated Analysis of Ambient Air, Stack Gas, 269
Landfill Gas, and Automobile Exhaust Using GC/MS Methods Daniel B. Cardin
and John T. Deschenes
The Development and Evaluation of a Transportable Fast Gas Chromatograph for the 276
Monitoring of Organic Vapors in Air Jesus A. Gonzales, Steven P. Levine, and
Richard E. Berkley
A Cryogenless AutoGC System for Enhanced Ozone Monitoring Using a Simplified, 282
Single Detector Approach Daniel B. Cardin and John T. Deschenes
Photoionization Detection of Air Toxics with Microbore Chromatography Columns 289
Kenneth R. Carney, Aaron M. Mainga, and Edward B. Overton
Near Real-Time GC Analysis of Volatile Organic Compounds Using an On-line 296
Micro-trap Somenath Mitra, Hung Jen Lai, and Merrill Jackson
The Use of Analytical and Method Surrogates in GC/MD Analysis of Whole Air 301
Samples David A. Brymer, Larry D. Ogle, Christopher J. Jones, et al.
Moisture Management Techniques Applicable to Whole Air Samples Analyzed by 307
Method TO-14, H; GC/MS Considerations Larry D. Ogle, David A. Brymer,
Christopher J. Jones, et al.
-------
Determination of VOCs in Ambient Air at 0.1 PPBV for the Clean Air Status and 313
Trends Network (CASTNet) Michael G. Winslow, Dwight F. Roberts, and Michael
E. Keller
Development of Methods for Sampling and Analysis of Vent Stream Emissions From 319
Glycol Dehydration Units in the Natural Gas Industry Larry D. Ogle, Curtis
O. Rueter, Dirk L. Reif, et al.
Session 8
S S Canisters, Use and Techniques
Improved Sensitivity and Quality Assurance in Summa Canister Analysis Using EPA 327
Method TO14 Jon Wong, Daniel B. Cardin, and John T. Deschenes
A Simple, Accurate Procedure for Preparation of Analytical Standards for TO14 333
Instrument Calibration John T. Deschenes and Daniel B. Cardin
The Analysis of Canister Samples in Louisiana by Non-Cryogenic Concentration and 337
GC/MS Analysis James M. Hazlett, Peggy Hatch, and Charles K. Brown
Session 9
Quality Assurance
Changes to the EPA Quality System for the Collection and Evaluation of 345
Environmental Data Gary L. Johnson
The National Performance Audit Program (NPAP) Elizabeth T. Hunike and Joseph 351
B. Elkins, Jr.
Two New Gas Standards Programs at the National Institute of Standards and 357
Technology William J. Mitchell and Willie E. May
A Low Cost Procedure to Make Gaseous Pollutant Audit Materials William 363
J. Mitchell, Ellen W. Streib, Howard L. Crist, et al.
Round-Robin Analysis of Performance Evaluation Samples by Standard ARI700-88 370
Shirley J. Wasson and James B. Flanagan
EPA's QA Program on the Suppliers of Protocol Gases Avw P. Hines, William 378
/. Mitchell, Matthew Miller, et al.
One Size Does Not Fit All: A Panel Discussion on QA Approaches to Air Toxic 384
Issues Shrikant Kulkarni
-------
Session 10
Source Emissions and Measurements
Development of a Test Method for the Measurement of Gaseous Methanol Emissions 393
From Stationary Sources B. A. Pate, M. R. Peterson, R. K. M. Jayanty, et al.
Development and Validation of a Source Test Method for 2,4-Toluene Diisocyanate 399
F. W. Wilshire, J. E. Knoll, S. C. Foster, et al.
Simultaneous Supercritical Extraction of SemivolatUe and Volatile Organic 408
Compounds From XAD-2® Sorbent for Air Toxics and Stationary Source Emissions
Joette Steger and Steve Hoskinson
An Investigation of Process Mass Spectroscopy as a Continuous Emission Analyzer 414
for Stationary Sources Laura L. Kinner and Grant M. Plummer
Gas Phase Fl'lK Spectrometry as a Method of Measuring Hazardous Air Pollutant 423
Emissions Thomas J. Geyer, Grant M. Plummer, Thomas A. Dunder, et al.
Field Observations of Compliance Monitoring for Ethylene Oxide Emissions From 429
Hospital Sterilizers Kevin Mongar
Evaluation of Polynuclear Aromatic Hydrocarbons and Nitrogen Heterocycles in the 436
Stack Effluent of Asphalt Processing Plants Kim Hudak, Richard Pirolli, and Newt
Rowe
Session 11
Measurement of Aerosols
The Summer 1992 PM-10 Saturation Monitoring Study in the Ashland, KY Area Eric 445
S. Ringler, VanX. Shrieves, andNeilJ. Berg
Evaluation of a Real-Time Monitor for Particle-Bound PAH in Air Nancy 451
K. Wilson, Ruth K. Barbour, Robert M. Burton, et al.
Particle and Gas Transmission Characteristics of a VAPS System and Three Different 457
Inlets Dennis D. Lane, Stephen J. Randtke, Ray E. Carter, Jr., et al.
Impact of Changes in Sulfate Aerosol Loading on Greenhouse Warming V. K. 464
Saxena andJ. D. Grovenstein
Testing of a Triple-Path Denuder Designed for Quantitative Sulfate and Nitrate 470
Measurements Briant L. Davis, Yun Deng, Darcy J. Anderson, et al.
-------
Session 12
Risk Assessment
Assessment of Ambient-Air Quality at Landfills 8 and 10 of Wright-Patterson Air 479
Force Base, Ohio R. R. Mannar, J. R. Tucker, and Teresa S. Finke
The Use of Air Modeling in the Development of the Air Monitoring Program for the 488
Dubose Oil Company Superfund Site Remedial Action Thomas A. Peel, Scott
E. Rowden, and Steven Ratzlaff
Risk Assessment Methods for Exposure to Environmental Substances Found Indoors 494
TheodorD. Sterling, WilfL.Rosenbaum, Chris W. Collett, etal.
A Pilot Study to Assess Personal Exposure to Ozone in the Fraser Valley of British 501
Columbia Michael Brauer, Jeff Brook, and Mark Raizenne
Session 13
Mercury in the Environment
Mercury Determination in Environmental Materials: Methodology for Instrumental 509
Neutron Activation Analysis /. Olmez, M. Ames, andN. K. Aras
Determining the Wet Deposition of Mercury - A Comparison of Weekly, Biweekly, 515
and Monthly Collection of Precipitation Samples Ake Iverfeldt and John Munthe
Wet Deposition of Methyhnercury in Sweden John Munthe and Ake Iverfeldt 519
Session 14
Oxidants in the Atmosphere
Peroxyacetyl Nitrate Concentrations at Suburban and Downtown Locations in 525
Atlanta, GA Benjamin E. Hartsell, Viney P. Aneja, Daniel Grosjean, et at.
Trends Analysis of Ambient NO and NOy Concentrations in Raleigh, North Carolina 531
Deug-Soo Kim and Viney P. Aneja
Gaseous Hydrogen Peroxide Concentrations in the Central Piedmont Region of North 537
Carolina Mita Das and Viney P. Aneja
-------
Session 15
FTIR and Remote Sensing
Field Testing of Two VOC Emission Rate Estimation Methods Ray E. Carter, Jr., 545
Dennis D. Lane, Glen A Marotz, et al.
Ambient Air Monitoring Siting Criteria for Open Path Analyzers Measuring Nitrogen 551
Dioxide, Ozone, and Sulfur Dioxide Lee Ann B. Byrd
Monitoring for Fugitive Emissions at Superfund Sites During Remediation Activities 557
with an FTIR Remote Sensor Robert H. Kagann, Orman A. Simpson, and Robert
J. Kricks
A New Concept for Open Path Air Pollution Monitoring Lyle H. Taylor 563
Session 16
Measurement of VOCs
Recovery After Storage and Desorption Efficiencies for Volatile Organic Compounds 571
Spiked on Thermal Desorption Tubes Scott A. Hazard and Jamie L. Brown
Temporal and Spatial Variability of Toxic VOC Sources in Columbus, Ohio Mukund 579
Ramamurthi, Thomas J. Kelly, and Chester W. Spicer
A Method for Separating Volatile Organic Carbon From 0.1 m3 of Air to Identify 585
Sources of Ozone Precursors via Isotope (14C) Measurements George A. Klouda,
James E. Norris, Lloyd A. Currie, etal.
Monitoring Volatile Organic Compounds (VOCs) in the Green Bay Area Mark 604
K. Allen, Julian Chazin, and Jonnell Hecker
Comparison of NMOC Data Collected by Two Methods in Atlanta Jack H. Shreffler 610
VOCs in Mexico City Ambient Air Robert L. Seila, William A. Lonnemaii, Maria 616
Esther Ruiz Santoyo, etal.
Monitoring Studies of Ambient Level VOCs and Carbonyls in Phoenix, Arizona 622
Carmo Fernandes, James L. Guyton, Cheng Peter Lee, et al.
An Automated System for the On-Line Analysis of Ozone Precursors /. Seeley, 628
G. Broadway, A. Tipler, et al.
Field Monitoring of Ozone Precursors Using an Automated Gas Chromatographic 634
System Terri V. Brixen and John K. Stewart
-------
Performance Evaluation of the HP-5971A MSD for Analysis of VOCs in Air Alston 640
Sykes, Mitchell Howell, William Preston, et al.
Whole Air Analysis for TO-14 Low Level and High Level Samples Richard Jesser 646
Session 17
Acid Aerosols and Related Pollutants
Acid Aerosol Measurement Methods: A Summary of U.S. EPA Intercomparisons 651
T. G. Ellestad
Speciation and Determination of N-Nitrosodimethylamine and NOX Species in 657
Ambient Air by Surface Specific Preconcentration M. S. Thomson andR. S. Braman
Measurement of Ambient Particle Phase Organic Acidity Using an Annular Denuder- 663
Filter Pack System Joy Lawrence and Petros Koutrakis
Session 18
Indoor Air Quality
Ecology of Fungi in Buildings: Relationship to Indoor Air Quality Sidney A. Crow, 671
Donald G. Ahearn, Judith A. Noble, et al.
Session 19
Pesticides, Farm Exposure Studies
Mixer-Loader and Applicator Exposure of Nitrapyrin to Commercial Handlers and 679
Farmers Richard C. Honeycutt, Jim Vaccaro, and Terry Johnson
Analytical Method for the Screening of Pesticides and Polynuclear Aromatic 685
Hydrocarbons From Housedust Tapan K. Majumdar, David E. Camann, and Paul
W.Geno
A Pilot Study for Measuring Environmental Exposures From Agricultural 691
Applications of Pesticides: An Overview A. E. Bond, G. G. Aklund, R. G. Lewis, et
al.
Analytical Methods for Assessing the Exposure of Farmers and Their Families to 698
Pesticides Paul W. Geno, David E. Camann, Kevin Villalobos, et al.
NCI/EPA Agricultural Health Study (AHS): Development of the Biomarker 706
Questionnaire C. J. Nelson, Jacquelyn M. Clothier, Gerald Akland, et al.
-------
Measurements to Assess Exposure of the Farmer and Family to Agricultural 712
Pesticides David E. Camann, Paul W. Geno, H. Jac Harding, et al.
Collection and Analysis of Duplicate Diet Samples: A Pilot Study of Farmer 718
Exposure to Pesticides Kent W. Thomas, Linda S. Sheldon, Jeffrey T. Keever, et al.
Session 20
Acid Aerosols and Philadelphia Results
Exposures to Acid Aerosols and Gases in Schools and Youth Centers of Philadelphia 727
Jed M. Waldman, Chris S-K. Liang, Abdul Kitto, et al.
Spanning From Regional to Microenvironmental Scales of Exposures to Acid 733
Aerosols Chris SX. Liang and Jed M. Waldman
Session 21
Acid Aerosol Field Study Results
Measurement of Acidic Aerosol Species in Eastern Europe Michael Brauer, Thomas 741
5. Dumyahn, John D. Spengler, et al.
Acid Aerosol Measurements in Eastern Canada /. R. Brook and H. A. Wiebe 747
Development of a Low Cutpoint Virtual Impactor for Collection of Semi-Volatile 753
Organic Compounds Constantinos Sioutas, Petros Koutrakis, Steve Ferguson, et al.
Session 22
Personal Monitors
Evaluation of a Portable Multisorbent Tube Sampler for Monitoring Airborne 761
Organic Compounds Albert J. Pollack and Sydney M. Gordon
Session 23
General
Sensitive Real-Time Monitoring of NO, NO2 and Other Nitrocompounds Under 769
Atmospheric Conditions Josef B. Simeonsson, George W. Lemire, and Rosario
C. Sausa
Impact of the New CAA Regulations on the Design and Operation of Automated 775
VOC Sampling Instrumentation Joseph Krasnec and Ludovit Krasnec
-------
Pressure and Temperature Effects on Concentration of Gaseous Calibration Standards 781
James McAndrew, Keith Rogers, and Dmitry Znamensky
Session 24
General
Laboratory Evaluation of Gas Dilution Systems for Analyzer Calibration and 789
Calibration Gas Analysis Robert S. Wright and Robert W. Murdoch
A Dual Purge & Trap/Air Toxics Preconcentration System for 16-Position GC/MS 795
Analysis of Canisters, Waters, and Soils Leon Levan and Daniel B. Cardin
Air Emission Rate Measurements of Surface Impoundment Quiescent Water and 800
Sludge Surfaces William A. Butler
Characterization of Air Emission From the Simulated Open Combustion of Fiberglass 806
Materials Christopher C. Lutes, Jeffrey V. Ryan, and Paul M. Lemieux
Initial Implementation of the Photochemical Assessment Monitoring Stations 813
(PAMS) Network Geri Dorosz-Stargardt
Meteorological Considerations in Siting Monitors of Photochemical Pollutants 819
Shao-Hang Chu
Poster Session
Accumulation of Persistent Organic Compounds in Spruce Needles in Relation to 827
Concentrations in Air and Deposition Eva Brorstrom-Lunden and Anne Lindskog
Estimates of Pesticide Exposure From the Agricultural Health Study (AHS) Nicholas 836
/. Giardino, David E. Camann, Paul W. Geno, et al.
Sample Collection Methods to Assess Environmental Exposure to Agricultural 842
Pesticides H. J. Harding, P. M. Merritt, J. M. Clothier, et al.
Comparison of Transfer of Surface Chlorpyrifos Residues From Carpet by Three 848
Dislodgeable Residue Methods David E. Camann, H. Jac Harding, Susan
R.Agrawal, etal.
Applications of a Continuous Gas Chromatograph: Area Monitoring, Scrubbers, and 857
Stacks Robert C. Peterson
internal Standard Implementation in Air Monitoring Sharon P. Reiss and Wendy 863
L. Ballard
-------
Aspects of Data Management for an LDAR Program Ginger Darnell and Kelly Trilk 869
Gas and Particle Phase Measurements of Atmospheric Toxic Pollutants Douglas 874
A. Lane and N. Douglas Johnson
Evaluation of Polynuclear Aromatic Hydrocarbons and Nitrogen Heterocycles in the 880
Stack Effluent of Asphalt Processing Plants Kim Hudak, Richard Pirolli, and Newt
Rowe
Estimates of Pesticide Exposure From the Agricultural Health Study (AHS) Nicholas 886
/. Giardino, David E Camann, Paul W. Geno, et al.
Subject Index 893
Author Index 905
-------
Preface
The 1993 United States Environmental Protection Agency/Air and Waste Management
Association International Symposium, Measurement of Toxic and Related Air Pollutants was
held in Durham, North Carolina on May 3-7, 1993. This annual symposium is sponsored by the
United States Environmental Protection Agency (USEPA), Atmospheric Research and Exposure
Assessment Laboratory of Research Triangle Park, NC, and the Air and Waste Management
Association (A&WMA) of Pittsburgh, PA.
The four day technical program consisted of 190 oral papers presented in twenty four
separate sessions plus a poster session with twenty papers. Individual sessions concentrated on
recent advances in the measurement and monitoring of toxic and related air pollutants in the
ambient atmosphere, in the indoor air of homes and highly confined spaces, and in emissions
from stationary and mobile sources.
Course offered in conjunction with the symposium were taught by leaders in the field of air
pollution monitoring and focused on basic sampling and analytical methodology. Exhibits were
on display from sixty instrument and consulting services. The keynote address was presented by
Allen Klmek, Director of Air Management of the Department of Environmental Health and
Natural Resources of the state of North Carolina.
Measurement and monitoring research efforts are designed to anticipate potential
environmental problems. Research supports regulatory actions by developing an in-depth
understanding of the nature of processes that impact compliance with regulations and evaluates
the effectiveness of health and environmental protection through the monitoring of long-term
trends. EPA's Atmospheric Research and Exposure Assessment Laboratory is responsible for
research and development of new methods, techniques and systems for detection, identification
and characterization of pollutants in emission sources and in indoor and ambient environments.
The Laboratory has the responsibility of implementation of a national quality assurance program
for air pollutant measurement systems, and supplying technical support to Agency regulatory
programs on local, regional, and global scale.
The A&WMA provides a neutral forum where environmental professionals share technical
information about air pollution measurement and control. This year (1993) was the 13th
consecutive year of the symposium and the 8th year of its co-sponsorship with the A&WMA.
Bruce W. Gay Jr. (USEPA)
R.K.M. Jayanty (RTT)
-------
Conference Committees
Technical Program Committee
Cochairmen
Bruce W. Gay, Jr.
AREAL, U.S. EPA
R. K. M. Jayanty
Research Triangle Institute
Committee
Ron Mosley
AEERL, U.S. EPA
S.T.Rao
NYS Department of Environmental
Conservation
DelbertEatough
Brigham Young University
Joseph Sickles
U.S. EPA
Charles Lewis
AREAL, U.S. EPA
Roy Zweidinger
AREAL, U.S. EPA
John Munthe
Swedish Environmental Research
Institute
Viney P. Aneja
North Carolina Slate University
George Russwurm
ManTech Environmental
Technology, Inc.
William Herget
Radian Corp.
William McClenny
U.S. EPA
Petros Koutrakis
Harvard School of Public Health
Sydney Gordon
Battelle
Larry Ogle
Radian Corp.
ShriKulkami
Research Triangle Institute
Merrill Jackson
AREAL, U.S. EPA
Dennis Lane
University of Kansas
Lance Wallace
U.S.EPA
Dennis Naugle
Research Triangle Institute
Gerry Ackland
U.S. EPA
Robert G. Lewis
AREAL, U. S. EPA
James Mulik
U.S. EPA
Donald Scott
AREAL, U.S. EPA
William Gutknecht
Research Triangle Institute
Gary Hunt
ENSR Consulting and Engineering
-------
General Conference Cochairmen
Jay J. Messer Martin Rivers
AREAL, U.S. EPA Air & Waste Management
Association
Research Triangle Chapter
Mark Shanis Jim Flanagan
Chairman Treasurer
Joe Laznow Jim Southerland
Vice Chairman Membership
Anne A. Pope Anne A. Pope
Secretary Education
South Atlantic States Section
Cathy Taylor
Chairman
Robert Kaufman Douglas Pelton
Vice Chairman Treasurer
F. Vandiver Bradow Ronald L. Bradow
Secretary Education
Toxic Air Pollutants Division
Gary Hunt Dave Patrick
Chairman Vice Chairman
Ambient Monitoring Committee (EM-3)
R. K. M. Jayanty Paul A. Solomon
Chairman Second Vice Chairman
Fred Dowling Suresh Santanam
First Vice Chairman Secretary
Source Monitoring Committee (EM-4)
Mark S. Siegler
Chairman
James Jahnke J. Ron Jernigan
Vice Chairman Secretary
-------
Session 1
Radon Indoor Air Pollution
-------
Evaluation of Radon Movement Through Soil and Foundation Substructures
by: Marc Y. Menetrez and
Ronald B. Mosley
U.S. EPA
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Richard Snoddy,
Krish Ratanaphruks, and
Samuel A. Brubaker, Jr.
Acurex Environmental Corp.
P.O. Box 13109
Research Triangle Park, NC 27709
ABSTRACT
To study convective and diffusive soil gas movement through soil and foundation substructures, a chamber (2
x 2 x 4 m long) filled with soil having an elevated radium content will be described. The chamber is filled with
a soil of known characteristics such as radium and radon concentrations, packing density, and moisture retention
properties. A depressurized perforated tube located in the center of the soil chamber will draw radon-laden air
or a tracer gas through the soil and simulated building foundations under varying moisture conditions. Pressure
contours and radon concentrations will be measured using an array of pressure sensors and radon extraction
probes distributed within a two dimensional cross section extending the length of the chamber. Tracer gas
transport rates from 23 probe sites to the center tube will also be measured. Data generated will be used to
compute the soil permeability as a function of moisture in order to better understand radon transport in soils.
The project will yield valuable information about how radon moves through soil and enters homes and will
consolidate our understanding of other areas of research such as radon blocking and pressure and temperature
driving forces. The scope of this paper is to describe the chamber, materials, instrumentation, and
measurement methods utilized in this project.
INTRODUCTION
System Description
To evaluate the movement of radon through soil and foundation substructures, a pilot scale chamber was
designed to study the influences of several parameters (1,2). To better understand the influence of certain
entry routes on indoor radon and to design more effective countermeasures, a better understanding of the
transport mechanisms is required. Mathematical models describing radon transport and entry have been
developed; however, validation of the models is necessary through the simulation of transport processes under
controlled conditions (3). To simulate conditions for the movement of radon gas through soil, a research
chamber has been constructed containing 16 m3 of soil with high levels of naturally occurring radium (averaging
6 pCi Ran6/g soil) which is able to generate elevated levels of radon (approaching 2000 pCi Rn222/!). Pressure-
driven flow conditions will be monitored along a two dimensional plane intersecting the central length of the
chamber (see Figure 1). Pressure differential, radon concentration levels, and air drawn through the soil from
the surface will be measured. To simulate the flow of soil gas in the chamber a vacuum is created inside a
perforated pipe line (suction tube), located across the center of the chamber, and another pipe located across
the end at mid-depth of the chamber. These will provide the driving force for the convective flow of gas in
the soil. The data generated, analysis, and interpretation of results are not addressed and will be covered
separately.
-------
Chamber
The chamber is built from carbon steel sheet, 1.27 cm thick. The rectangular box measures approximately 4
x 2 x 2 m (LxWxH), with eight buttresses supporting the structure. The inside surface of the chamber and
measurement probes are coated with latex paint as a corrosion inhibitor. The chamber sides are bolted (with
1.25 cm bolts), and joints are sealed with silicon, allowing it to hold approximately 33,000 kg of soil and water
without leakage of water. Water used in the soil packing process and for its influence on radon emanation and
gas transport, is adjusted by adding or draining. Three drain openings at the base are connected to three sight
tubes and valves to allow for the adjustment of the water level inside the chamber. A spray bar sprinkler unit
mounted above the chamber can also be activated if an increase in soil moisture is desired. Soil moisture is
measured by use of a Troxler Sentry 200-AP moisture sensor which is described below.
Soil
The soil used in the chamber was obtained from a phosphate mining area of Florida, and transported to the
research facility at the Research Triangle Park, North Carolina. Eight samples were analyzed for radium
content by the University of Florida, yielding an average of 4.85 pCi Ra^/g. Eleven samples were also
analyzed by North Carolina State University, to determine soil properties, such as bulk density (averaging 1.636
g/cm3), saturated hydraulic conductivity (an average K = 31.698), and the soil content of components
(averaging sand 97.38 percent, silt 0.96 percent, and clay 1.66 percent). Moisture retention properties were
determined by equal weights of dry soil being packed at various densities and saturated by flooding, the
deviation in bulk density exhibited by soil samples was minimal (less than 3 percent), indicating that the sandy
soil would attain a similar density after saturation by flooding. The ability of this soil to attain near maximum
density after saturation allowed for the chamber-to be packed with careful attention given to areas surrounding
the sampling probes, to eliminate air pockets.
Soil Bed Construction for the Chamber
In excess of 20,000 kg of soil was used to fill the soil chamber, in incremental layers of 0.5 m, by the
following procedure. First the soil was passed through a 1.9 cm metal screen to isolate debris and other foreign
objects. The soil was evenly spread along the surface, and water was added to the sand until the water level
was approximately 3 to 4 cm above the soil and allowed to settle for 2 days. The water was drained, soil
density was measured (using the Troxler Nuclear Density Gauge, described below), and the next layer was
added. Attention was given to ensure that the buried probes did not retain air volumes between the probe
surface and the soil. Voids around probes could provide gases with a conduit for movement, which would
cause inaccurate measurements of flow through the soil bed.
Soil Moisture and Density
A nuclear density gauge was used to measure soil density during the soil bed preparation. An average of the
measurements yielded a soil density of 1682 kg/m3. Each layer of soil loaded into the chamber was measured
after flooding and draining: typically density values varied by plus or minus 5 percent. The moisture content
of the soil material was measured using a Troxler Sentry 200-AP responding to changes in the dielectric
constant of the material. The instrument utilizes a cylindrical probe that is lowered into a vertical access tube
made from polyvinyl chloride (PVC) pipe (5.8 cm ID). By lowering the probe to varying depths, along this
access tube, a vertical moisture profile of the soil bed can be established for the four locations of the access
tubes in the soil chamber. An example of the moisture profile with soil depth in the chamber, as measured in
four vertical PVC probes, is shown in Figure 2.
Instrumentation
The sand filled chamber was instrumented with 23 sampling probes, which can continuously measure
temperature and pressure and actively extract samples of soil gas by recirculating gas flow. During active
recirculation and temperature measurement, the soil gas flows out of the tip of the probe and is drawn into the
collar section. For pressure measurement, the pressure sensor is connected to the tip of the probe. Two
suction tubes are buried approximately 1 m down in the sand (vertically central) and are used to provide
depressurization to drive air flow through the soil. One suction tube is at the center of the box (horizontally
-------
central) and one is at the end of the box. These tubes are divided into three sections: left, right, and center.
Each section is attached to a valved vacuum pump, to provide the driving force for soil gas flow, and a
pressure-sensing line. The same constant negative pressure is maintained in all three sections to isolate the
center section from influences of edge effects arising from the finite dimensions of the chamber.
All three sections of the suction tube are automatically controlled to maintain the preset pressure by use of a
mass flow controller for each section, and valved lines to correct for pressure imbalances. A separate pressure
sensing system provides a safety valve to shut off the pressure source to prevent damage to the pressure
transmitters in the event of over pressurization. The barometric pressure of the facility housing the chamber
is measured with a Rosemont barometric pressure sensor, and sensors to indicate door openings of the test
facility are used for two doors leading to the chamber laboratory. The sensitivity of the differential pressure
cells used in all measurement probes requires a detailed accounting of all potential pressure influences.
Barometric pressure and temporary pressure fluctuation (such as doors opening and closing) have been shown
to influence chamber probe pressure measurements, and are considered when interpreting results.
The data collection system automatically records all pressure and temperature data from the 23 probes; the
suction tube temperature; left, right, and center pressures; barometric pressure; the flow through the mass flow
controllers; and the status (open/closed) of the front door to the test facility. It is necessary to monitor the
operation of the outside door because the pressure disturbances are registered by the pressure sensors. Each
of the 23 probes placed in the soil chamber is designed either to provide a point pressure measurement or to
sample the soil gas (e.g., for radon concentration) with minimum interference to soil gas flow pathways leading
to the suction tube. The spatial locations of the 23 probes are diagramed in Figure 3.
Chamber Operation
A vacuum pump pulls the soil gas from a large porous brass filter element near the end of the probe (diagramed
in Figure 4). A manually operated, four-way valve is used to either a) recirculate the soil gas directly back
into the box (to purge the lines) or b) direct the flow through a desiccating cylinder and scintillation cell (Lucas
Cell) before being returned to the box. The upper and lower plenum dividers in the desiccating canister have
been removed and the entire volume filled with desiccant (indicating anhydrous calcium sulfate) to reduce free
volume in the system. After recirculating soil gas for a minimum of 20 minutes, at a flow rate of 0.5 1/min,
soil gas flow is passed through the Lucas Cells for 4 min. Cells are then detached from the probe recirculation
system and analyzed for radon gas according to "Protocol For The Determination Of Indoor Radon
Concentration By Grab Sampling" (4). A series of experiments which were performed as part of this project
were successful in optimizing the radon gas sample collection technique. However, a discussion of these
experiments is beyond the scope of this paper.
An electrically operated, three-way valve isolates the differential pressure cell from the high pressure in the
flow circuit to prevent damage to the cell. During operation, a J-type thermocouple monitors the flow
temperature and is accurate to 0.25 °C. Differential pressure is measured with a Modus pressure transmitter.
This capacitance type cell has an output range of 0 to 50 Pa with the accuracy of 1 percent of the range, and
is selected to accommodate an expected differential pressure in the range of 15 Pa generated in the soil
chamber.
EXPERIMENTAL PLAN
Experiments have begun to measure changes in pressure, radon concentration, and temperature, when
convective driving forces are applied. A vacuum of approximately 80 Pa is applied to the centrally located
suction tube, and a flow rate of approximately 8 1/min is applied. Pressure and temperature are recorded over
time from the moment of the initiation of the pressure. Within seconds the pressure change is recorded
throughout the bed, from the probes closest to the suction tube to probes nearly 2 m from the suction tube along
the central longitudal plane. Radon gas grab samples are collected from each of the 23 probe locations, at a
rate of approximately 14 samples in an 8 hour period (1 working day). Changes in radon concentrations take
-------
10 to 20 days to reach a stable level. Tracer gas (SFj) times-of-flight are measured between the probes and
the suction tube. The probe is used as the introduction point by injecting 10 ml of 100 percent SF6 by syringe,
and analyzing the suction tube exhaust gases. The results characterize soil gas movement along the two-
dimensional vertical plane dissecting the length of the soil chamber.
The chamber contains a second suction tube located near the end wall, to study the same effects described above
on pressure and flow for distances up to 4 m between the suction tube and the most distance measurement
probe. A simulated foundation substructure and stem wall (containing measured openings in joints) can also
be incorporated into the chamber to study their influences on entry rates under defined conditions. The study
of soil gas movement through greater distances of soil and possible obstructions to flow (such as foundation
joints) is not part of the initial work, but is expected to be dealt with, and will be further discussed at that time.
SUMMARY
A chamber was designed and constructed and experiments designed to characterize the movement of radon
through soil and foundation substructures. Convective and diffusive flow conditions will be monitored along
a two-dimensional vertical plane through the center of the chamber. Pressure, radon concentration, and
temperature in the soil as well as tracer gas times-of-flight are measured. Mathematical models describing
transport and entry have been developed; however, validation of the models is necessary. The project will yield
valuable information about how radon moves through soil and enters homes. This paper describes the chamber,
materials, instrumentation, and measurement methods utilized in this project. Data are currently being collected
for analysis and interpretation and will be addressed elsewhere (5, 6).
REFERENCES
1, Van Der Graaf, E.R., Transport of Radon in Soil Under Controlled Conditions, In: Proceedings of
the Fifth International Symposium on the Natural Radiation Environment, Saltzburg, Austria, September
1991: Radiation Protection Dosimetry (in press).
2. Van Der Graaf, E.R., Heijs, S., de Meijer, R.J., Mulder, H.F.G.N.,and Put, L.S., A Facility to Study
Transport of Radon in Soil Under Controlled Conditions, Presented and published in the "Tagesbericht"
of the Kolloquim Messung von Radon und Radonfolgeprodukten, Berlin, May 6-7, 1991.
3. Mosley, R.B., Model Based Pilot Scale Research Facility for Studying Production, Transport, and
Entry of Radon into Structures., In: Proceedings of the 1992 International Symposium on Radon and
Radon Reduction Technology: Vol. 1.
4. Interim Indoor Radon and Radon Decay Product Measurement Protocols, U.S. EPA, Office of
Radiation Programs, Washington, D.C., EPA-520/1-86-04 (NTIS PB86-215258), April 1986.
5. Mosley, R.B., Menetrez, M.Y., Ratanaphruks, K., and Brubaker, S.A. Jr., Comparison of
Measurement Techniques for Soil Permeability in EPA's Soil-Gas Chamber (presented at AWMA/EPA
Symposium on Measurement of Toxic and Related Air Pollutants, Durham, NC, 5/4-7/93).
6. Mosley, R.B. An Analytical Solution to Describe the Pressure/flow Relationship in EPA's Soil-gas
Chamber (presented at AWMA/EPA Symposium on Measurement of Toxic and Related Air Pollutants,
Durham, NC, 5/4-7/93).
-------
2.Qn
60 ••
50 •
I*"
ui 30 • •
° 20 •
10 ••
2.0n
Figure 1. Chamber Probes and Suction Tubes
9/29/92
"10/9/92
0 5 15 30 46 61 76 81 97 112 127 142 157
DEPTH FROM SURFACE OF SAND (cm)
Figure 2. Soil Moisture with Depth
-------
Moisture Probes
I
H
®
Figure 3. Radon Sampling and Moisture Probe
Location (Chamber Side View)
-135m-
Figure 4. Sampling Probe
-------
Comparison of Measurement Techniques for Soil
Permeability in EPA's Soil-gas Chamber
by: Ronald B. Mosley,
Marc Y. Menetrez, and
D. Bruce Harris
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Richard Snoody,
Krich Ratanaphruks, and
Samuel A. Brubaker, Jr.
Acurex Environmental Corp.
P.O. Box 13109
Research Triangle Park, NC 27709
ABSTRACT
Initial measurements of soil permeability to the flow of air in
EPA's soil chamber yield relatively good agreement between two
methods. One method uses a set of 23 point probes located in a
vertical plane of the chamber. These measurements are similar to the
standard practice of measuring in situ soil permeabilities. The
other method uses an arrangement designed to ensure ideal geometric
flow patterns. It is argued that the latter measurement yields a
better approximation of the effective bulk permeability that
determines the advective flow through distributed entry paths into
buildings such as cracks between floor slabs and walls. Measured
moisture profiles in the soil are also discussed. The permeability
measurements are compared to the predictions of a widely used
empirical model.
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication.
-------
INTRODUCTION
Numerous field studies (1-4) have attempted to correlate
measurements or estimates of radon concentration in the soil with
measured radon concentration in the indoor air of buildings. So far,
no one has provided a correlation with sufficient predictive strength
to serve as a useful tool in projecting the severity of a potential
indoor radon problem for a building yet to be constructed on a
particular site. This lack of a strong correlation has usually been
attributed to uncertainties in the properties of the soil at a given
site and the uncertainties associated with the building structure and
its entry paths.
However, researchers are beginning to question whether the
fundamental assumptions and measurement techniques used to analyze
the radon entry problem are adequate for certain types of soils.
Researchers at Lawrence Berkeley Laboratory (5-8) have reported
several instances in which conventional models underpredict both the
entry rate of radon into structures and pressure fields in the
surrounding soil. One proposed reason for these under-predictions is
that the point probe measurement technique used to determine the
permeability of soil-to-air flow may not adequately measure the
appropriate soil property. The implication would be that the point
probe does not necessarily measure the effective bulk permeability
needed to describe flow that is distributed over extended entry
routes.
In this paper, measurements of permeability obtained from a
number of point probes will be compared with results from a system
specifically designed to eliminate non-ideal effects on the flow
field.
SOIL CHAMBER
EPA's soil chamber was built to study the production and
transport of soil-gas contaminants such as radon, pesticides, and
landfill gases. The chamber is constructed inside a building in
which the environmental conditions can be controlled. The chamber
contains 16 m3 of sandy loam from a phosphate mining area of Florida.
The sandy soil was chosen largely to allow a uniform and isotropic
prism of soil 2 m high, 2 m wide, and 4 m long. The chamber is
designed to simulate full scale phenomena on the dimensions of a
building, but only in two dimensions. A highly porous metal tube
0.05 m in diameter passes horizontally through the center of the
prism. This tube can be used to simulate ideal flow toward a
building feature such as a crack. The effects of the finite width (2
m) is compensated by the use of guard ends on the tube. The details
of the chamber and its instrumentation are presented elsewhere
(9,10). The key point is that the chamber is designed to sustain
ideal flow between a large plane (the soil surface) and a long
parallel cylinder. The bulk permeability of the uniform soil can be
measured under these ideal flow conditions.
10
-------
SOIL MOISTURE
Soil moisture is measured using a Troxler Sentry 200-AP moisture
monitor. The measurements are performed with a sensor inside a
hollow polyvinyl chloride (PVC) tube installed vertically in the
soil. The sensor measures the percent water by volume in the soil to
an accuracy of 0.2% by measuring the effective capacitance of the
soil under an applied electromagnetic field. Moisture profiles are
generated by moving the sensor vertically inside the tube.
Measurements are taken 0.15 m apart. There are four moisture
measurement tubes in the soil chamber. Three tubes are in line on
one side of the chamber at distances of 0.3, 1, and 1.5 m from the
end of the chamber. The fourth tube is at 1.8 m on the other side of
the chamber.
Moisture profiles, expressed as a percent of saturation for the
four tubes, are illustrated in Figure 1. These profiles were all
measured on the same day, but are hardly distinguishable from
numerous sets of profiles measured periodically over 3 months.
Moisture is hardly detectable in the top 0.2 m of soil. It then
increases rapidly to about 15% and remains relatively constant to a
depth of about 1.2 m where it begins to increase rapidly with
increasing depth. The moisture reaches saturation near a depth of
1.75 m where the water level is located.
The permeability of the soil to air has been correlated to soil
moisture by Rogers and Nielson (11). This relationship is
illustrated by the solid curve in Figure 2. The permeability is
relatively independent of depth (moisture) until the moisture reaches
about 25% of saturation at which point the permeability decreases
very rapidly with increasing moisture. The permeability decreases by
2 orders-of-magnitude by the time the soil reaches 80% saturation.
PERMEABILITY MEASUREMENTS
This paper compares measurements of permeability using point
probes with the effective bulk value measured with the system
designed to eliminate non-ideal geometrical factors. However, the
effects of varying moisture on the measured permeability is always of
concern. Fortunately, for the present case, most of the flow occurs
in the region of nearly constant permeability since the collecting
cylinder is located 1 m deep.
The individual symbols in Figure 2 represent measurements by
individual point probes located in the central portion of the chamber
near the location of the second moisture measurement tube. The solid
line corresponds to the Roger's model for which the parameters were
chosen to yield agreement with the bulk permeability in the region
where permeability is constant. The dashed curve represents the same
model adjusted to agree with the average value measured by the point
probes in the region of constant permeability (near the surface).
Qualitatively, the shape of the measured permeability curve resembles
the model calculations. Note that the point probes indicate that the
permeability decreases somewhat faster with increasing moisture
-------
content than the Roger's model. Figure 3 compares the average
measurement of all 23 probes with the Roger's model. The average
values appear to emulate the model calculations somewhat more than
the selected data of Figure 2, probably because of the reduced
scatter. A measurement with an individual probe consists of 11 sets
of pressure and flow values. The slope obtained from a linear
regression of these values yields the measured permeability. The
average standard error of the regression for all 23 probes was 2.6%.
The maximum standard error for an individual probe was 4.4%. The 23
probe measurements are grouped at five depths represented by the
square symbols in Figure 3 . The two model calculations represented
by the curves are the same as in Figure 2 . The curve through the
measurement points is provided to guide the eye, but has no
implication as to shape or functional form. As before, the point
probe measurements suggest that permeability decreases significantly
at lower moisture levels than is indicated by the Roger's model.
Figure 4 shows the flow/pressure relationship from which the
bulk permeability is computed. From equations 6 and 10 in reference
(12)
k_ n [0 . 764 +lnUWJ>2) ] alope
where k (m2) is the permeability, \i (l.SxlO'5 kg m'1 s'1) is the
dynamic viscosity of the soil gas, h (1m) is the depth of the
cylinder in the soil, b (0.0254 m) is the radius of the tube, L
(0.666 m) is the length of the cylinder, and slope (l.eOxlO'6 m4 s kg
1) is the slope of the curve in Figure 4. The above equation yields
a permeability of 3.25x10"" m2. The average permeability measured by
the point probes in the upper part of the soil is 2.76x10'" m2. This
value differs from the bulk value obtained from the slope of the
curve in Figure 4 by only 15%.
DISCUSSION AND CONCLUSIONS
The interpretation of a measurement from a point probe requires
some consideration. The term point probe is used merely to imply
that the probe is small thus approximating a measurement at a point.
The question, however, is to what extent does the measurement depend
upon the exact geometry of the probe. The probes used in the current
measurements were not true cylinders, but were actually truncated
cones made of fritted brass. The slope of the sides of the cones was
about 12 degrees. The analysis of the data treated the probes as
cylinders with an effective diameter to yield the same surface area
as the actual probe. Consistent with the common practice in field
measurements, the geometrical factor associated with the probe
assumed that the probe was quite long in contrast to the reality that
its length is comparable to its diameter. This apparent
inconsistency in 'analyzing the probe data gives rise to potential
questions as to whether the close agreement (15% difference) with the
bulk permeability may be somewhat fortuitous. The answer to this
question may become clearer as future data are analyzed.
12
-------
REFERENCES
1. Eaton, R.S. and Scott, A.G. Understanding radon transport into
houses. Radiat. Prot. Dosim. v. 7, p. 251, 1984.
2. DSMA Atcon, Ltd. A computer study of soil gas movement into
buildings, report 1389/1333, Department of Health and Welfare,
Ottawa, 1985.
3. Nazaroff, W.W, Moed, B.A., Sextro, R.G., Revzan, K.L., and Nero,
A.V. Factors influencing soil as a source of indoor radon: A
framework for geographically assessing radon source potentials,
report LBL-20645, Lawrence Berkeley Laboratory, Berkeley, CA,
1987.
4. Kunz, C., Laymon, C.A., and Parker, C. Gravelly soils and indoor
radon. In: Proceedings: The 1988 Symposium on Radon and Radon
Reduction Technology: Volume 1. Symposium Oral Papers. EPA-
600/9-89-0063 (NTIS: PB-89-167480) p. 5-75 — 5-86, 1989.
5. Nazaroff, W.W., Lewis, S.R., Doyle, S.M., Moed, B.A., and Nero,
A.V. Environ. Sci. Technol. V. 21, p. 459-466, 1987.
6. Garbesi, K. and Sextro, R.G. Environ. Sci. Technol. v. 23, p.
1481-1487, 1989.
7. Revzan, K.L., Fisk, W.J., and Gadgil, A.J. Modeling radon entry
into houses with basements: Model description and verification.
LBL-27742, Lawrence Berkeley Laboratory, Berkeley, CA, 1987.
8. Garbesi, K. , Sextro, R.G., Fisk, W.J., Modera, M.P., and Revzan,
K.L. Soil-gas entry into an experimental basement: model
measurement comparisons and seasonal effects. Eviron. Sci.
Technol. v. 27, no. 3, p. 455-473, 1993.
9. Mosley, R.B. Model based pilot scale research facility for
studying production, transport, and entry of radon into
structures. In: Proceedings: the 1992 International Symposium
on Radon and Radon Reduction Technology: vol. 1.
10. Menetrez, M.Y., Mosley, R.B., Snoody, R., Ratanaphruks, K., and
Brubaker, S.A., Jr. Evaluation of radon movement through soil
and foundation substructures. Presented at the 1993
International Symposium on Measurement of Toxic and Related Air
Pollutants, Durham, NC, May 4-7, 1993.
11. Rogers, V.C. and Nielson, K.K. Correlation of Florida soil-gas
permeabilities with grain size, moisture, and porosity. EPA-
600/8-91-039 (NTIS: PB-91-211904), June 1991.
12. Mosley, R.B. An analytical solution to describe the
pressure/flow relationship in EPA's soil-gas chamber. Presented
at the 1993 International Symposium on Measurement of Toxic and
Related Air Pollutants, Durham, NC, May 4-7, 1993.
-------
0 0.2 0.4 0.6 0.8 1
Moisture saturation
Figure 1. Measured fraction of moisture
saturation at 4 locations.
u-
0.2
0.4
•P0.6
•£•0.8
£ 1 -
g-1.2
Q 1.4
1.6
1.8
9-
i
I
f
A i
• I A
A • //
A «, ,~^~^"
.
0 0.5 1 1.5 2 2.5 3 3.5
Permeability (10 m )
Figure! Comparison of measured and
modeled premeabilities as a function of
depth.
0 0.5 1 1.5 2 2.5 3 3.5
-11 2
Permeability (10 m )
Figure3. Comparison of the average
measured permeability with model
predictions.
0 20 40 60 80 100
Pressure (Pa)
Figured Plot of the total flow/pressure
relationship.
14
-------
RADON IN FLORIDA LARGE BUILDING STUDY
by: Marc Y. Menetrez,
Russell N. Kulp, and
D. Bruce Harris
U.S. EPA
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Felton Perry and
Don Elliot
Acurex Environmental Corp.
P.O. Box 13109
Research Triangle Park, NC 27709
Bobby Pyle and
Susan McDonough
Southern Research Institute
P.O. Box 55305
Birmingham, AL 35255-5305
ABSTRACT
This project is examining how radon concentration and indoor air quality levels are affected by
building ventilation dynamics and building air system conditions, including mixing and leakage rates of
typical residential, commercial, and public structures and heating, ventilating, and air-conditioning system
components. The ventilation dynamics inherent to a building to dilute radon and indoor air pollution and
overcome soil gas entry forces are being analyzed with the Florida Solar Energy Center computer model and
diagnostic and mitigation protocols developed. Two research sites have had newly developed data collection
stations and a weather station installed. Measurements of radon and carbon dioxide concentrations,
temperature, humidity, and pressure within building zones and subslab areas, and outdoor air intake flow
rates are being collected. The outdoor air intake will be adjusted from levels of no outdoor air to those
recommended by the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE)
as a modification of pressurization and dilution of indoor air conditions. Tracer gas active measurement
injection and detection points were placed in all zones. Data from instruments are downloaded by computer
modem connection to allow for prompt evaluation and analysis.
INTRODUCTION
The objective of this project is to continue to develop, validate, and provide guidance for radon
diagnostic procedures and mitigation strategies applicable to a variety of large buildings commonly found
in Florida. To accomplish this, it was necessary to perform detailed field investigations and parametric
studies in a variety of large buildings with elevated radon levels. The detailed investigations evaluate the
nature of radon occurrence, building entry mechanisms, the effects of heating, ventilating, and air-
conditioning (HVAC) system configuration and operation on radon entry, transport, and dilution, and the
significance of HVAC configuration, occupancy patterns, building height, air passageways between floors,
and other building construction features.
The majority of radon research in large buildings to date has focused on developing radon diagnostic
and mitigation techniques for school buildings. Experience in other types of large buildings has been mostly
-------
limited to new construction, rather than existing buildings. However, there are a number of similarities
between school buildings and other similar sized buildings (i.e., construction methods, HVAC systems,
building operation, and maintenance procedures). For this reason it is logical to believe that diagnostic and
mitigation techniques developed by the U.S. EPA for schools can be used as the basis for developing
diagnostic and mitigation techniques for other large buildings.
In accomplishing the project objectives the following activities have been included: 1) identification
of candidate buildings, 2) selection of a representative sub-set of buildings for diagnostics and mitigation
research, 3) developing standard radon diagnostic protocols applicable to large buildings, 4) conducting
diagnostic measurements and research in selected buildings and 5) identifying building mitigation strategies
based on the diagnostic measurements. Of these, only objectives 1,2, and 4 have been accomplished at this
point and are addressed in this paper.
HVAC SYSTEMS
HVAC systems have two distinct primary functions: 1) the provision and maintenance of specific
environmental conditions and 2) occupant or space ventilation for the provision and maintenance of
acceptable indoor air quality (1, 2, 3, 4). Typically the specified environmental conditions are generally for
occupant comfort, but can provide specific environmental conditions for a process or a product. In all cases
the HVAC system must provide the occupant or process with the proper conditions (dry bulb temperature,
relative humidity, air movement, etc.)(5).
The dominating feature of ventilation is the controlled introduction and removal of outdoor air (OA).
Research indicates that the required amount of OA is dependent on the rate of contaminant generation and
the maximum acceptable contaminant level. Understanding this is important to HVAC system designers
since confusion can lead to designs that are energy wasteful (too much OA) or that provide poor indoor air
quality (too little OA). For these two basic functions, there exist a wide variety of design situations. These
include commercial and manufacturing applications, general office space, educational and institutional
facilities, and special purpose space such as laboratories and clean rooms (6, 7, 8, 9, 10, 11 and 12).
INSTRUMENTATION
This project is examining how radon concentration and indoor air quality levels are affected by
building ventilation dynamics and building air system conditions, including mixing and leakage rates of
typical residential, commercial, and public structures and HVAC components. The ventilation dynamics
inherent to a building to dilute radon and indoor air pollution and overcome soil gas entry forces are being
analyzed in an effort to develop diagnostic and mitigation protocols.
Large buildings, being complex in character, raise imposing demands on data needs. The significant
demands of this project to measure many data parameters over time made it necessary to develop a new data
collection station system. Measurements of radon and carbon dioxide concentrations, temperature, humidity,
and pressure within indoor building zones, and subslab areas and outdoor air intake flow rates are being
collected. The OA intake will be adjusted from levels of no OA to recommended ASHRAE levels (3) as
a modification of pressurization and dilution of the indoor air conditions. Tracer gas was measured in all
zones of both buildings studied. Weather station information is recorded continuously. Real time data from
instruments are downloaded by computer modem connection to allow for prompt evaluation and analysis and
minimize on-site time demands.
Measurements have been completed on one large building in Florida, and are nearing completion on
16
-------
a second. The Financial Center North Building (FCN), located in Deerfield Beach, had seven data stations
installed to measure radon, and carbon dioxide concentrations, temperature, humidity, differential pressure
between indoor multiple zones, and subslab areas and the outdoor air pressure. In addition, OA intake flow
rates and supply fan and exhaust fan discharge time of use were recorded. The OA intake was adjusted from
zero to ASHRAE recommended levels of 20 ft3/mm/person*. PFT passive tracer gas emitters were placed
in all rooms, and detector sets were placed in all zones for each OA intake level. Short term EPERM
detectors were also used as integrated samplers for each outdoor air intake level.
Polk County Life and Learning Center (LLC) had five data stations and a weather station installed
and made operational. All stations recorded the same type of data as that collected at FCN, and to improve
data collection technique downloaded by phone modem.
For both buildings a certified test and balance (TAB) company was responsible for generating a list
of system deficiencies (punch list) to be acted on by the building owner/operator, installing a flow control
damper on the OA intake and balancing the HVAC system, and calibrating the damper. After all punch list
items were corrected, the balance was adjusted and tracer gas was measured. Prior to each OA adjustment,
EPERM measurements will be completed, and subslab grab radon will be measured. Tracer gas was
measured in all zones to calculate interzonal mixing (four gases), and to calculate shell leakage and air
exchange rate.
Both buildings had a total of five OA conditions (0, 5, 10, 15, and 20 tf/min/person) tested for
approximately 1 week each. Each week of testing will involve all three fan exhaust conditions( off, on and
intermittent use), during which tracer gas will be measured. If air balancing is required with each OA
condition, at the initiation of that condition the TAB will make these adjustments as soon as possible after
the OA level is changed. During tracer gas testing all thermostat temperatures will be set within the comfort
zone, and all doors will be closed to maintain supply air balance (it is anticipated that testing performed
during weekdays cannot control door use due to normal facility usage). Data will be downloaded at least
weekly and feedback on the success of measurement procedures to test all possible HVAC dynamic
mechanisms will be communicated continuously.
FINDINGS
The FCN building is a three-story office building that measures 46,000 ft2". The HVAC system is
of the unitary system category having 23 separate air handlers (AHs), two OA intakes and contains office
areas on three floors. The systems are fairly typical of speculative office space and along with the provision
of outdoor air, make this facility particularly attractive for the Large Building Study.
The only building site in which data collection has concluded to date is FCN. Data from LLC are
being gathered and are not available for discussion. The current extent of the data analysis is limited to a
qualitative discussion of carbon dioxide levels and a quantitative comparison of radon levels and building
HVAC activity (see Figure 1), with radon correlated inversely to outdoor air levels.
The FCN building had initially exhibited radon levels of approximately 10 pCi/1, during U.S.
Government Services Administration screening measurements, which are above the EPA recommended limit
of 4 pCi/1. In early 1992, Radon Environmental Testing Corporation was requested to provide radon
measurement and mitigation service to the building management. Passive sealing of slab cracks and
penetrations was provided as well as increasing the level of OA by installing supply fans which now inject
0 1 ft'/m = 0.000472 mVs. O 1 ft2 = 0.0929 m2.
17
-------
OA to the AH. This reduced radon levels to below the 4 pCi/1 limit and generally subjectively improved
indoor air quality (IAQ). By intentionally reducing the OA intake, an increase in radon concentrations was
exhibited to a peak level above 4 pCi/1 throughout the building. Distinct average levels of radon can be
identified from the data for a consistent level of OA intake. A clear comparison of radon levels versus
outdoor air intake flow rate is evident in the Figure 1 of integrated EPERM data. The average radon level
measured with no OA (2.98 pCi/1) was over twice that measured with an OA intake of 13.6 ftVmin/person
(1.27 pCi/1). Continuous measurement data including IAQ data is not available at this time.
The LLC is a single story building fairly typical of some of the radon work already undertaken in
other school buildings. It has an area of 18,000 ft2 and has as its HVAC system an all-air system.
Specifically it is a modified central station variable-air-volume (VAV) system. The system controls the
indoor environment by utilizing VAV boxes; however, its configuration cannot allow it to be classified as
a pure VAV system, although it is referred to as such.
Results of measurements at LLC indicate that radon reductions correlate with HVAC on time use.
During week days, on time use of the HVAC system exhibits a lowering of radon concentrations, with the
highest concentrations coinciding with the HVAC system turning on in the morning. Radon concentrations
consistently decrease during the day, and the lowest concentrations of radon are recorded at the time of
system shutdown. Figure 2 displays an example of a graph of daily radon concentrations peaking near 30
pCi/1 versus the OA (16.6 ftVmin/person) rate of flow (as a step function) for LLC having 150 occupants.
During the 12 hours of HVAC system inactivity, radon concentrations steadily increase to the high point (5
to 6 AM). The rate of decrease in radon concentrations is dependent on the rate of OA intake, with greatest
reductions rates correlated to greater rates of OA intake.
DISCUSSION
In both buildings radon results show an inverse correlation between OA intake levels and radon
concentrations. This relationship is visible at every stage of air intake, indicating the greater the level of
OA intake the less concentrations of radon result. The LLC exhibited the highest radon levels at the time
when the AH turned on. A decrease in radon was consistent during the AH on time, and the lowest
concentrations were measured at the time of AH shutdown (see Figure 2).
Tracer gas measurements indicate that zonal mixing, infiltration and overall pressure were dominated
by HVAC control. Thermostat controls were placed on continuous fan operation. However, each of the
23 thermostat control boxes has unrestricted access and many were often found to be reset to cycle the AHs
on/off for temperature control. This could create zones with active AHs which dominate adjacent areas.
Unitary HVAC systems (such as FCN) present unique subjects for investigation.
RESEARCH CONCLUSIONS
The FCN building showed a decrease in radon concentrations with increasing amounts of OA. The
LLC exhibited a daily lowering of radon concentrations with increasing OA levels. Both buildings are being
evaluated and field measurements are continuing on LLC.
The size and relative simplicity of the LLC made it a perfect candidate to begin development of the
HVAC system diagnostic procedures. Particularly with regard to the familiar central station HVAC system.
Evaluating the system, understanding the design intent, reestablishing air balance, and providing pressure
can be controlled with this building and system.
-------
Large buildings are not easily characterized since they are composed of many complex systems. The
two buildings selected represent a start at increasing our understanding of these complex systems, and
attempting to further current knowledge. Evaluation of the research results continues by use of the Florida
Solar Energy Center Large Building Computer Model. Complete study findings are expected to be available
by July 1993.
This research project was partly funded by the State of Florida, Department of Community Affairs.
REFERENCES
(1) ASHRAE 1989 Handbook. Fundamentals. The American Society of Heating, Refrigerating, and
Air-Conditioning Engineers, Inc. Atlanta, GA.
(2) ASHRAE Standard 55-1981. Thermal Environmental Conditions for Human Occupancy. The
American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc. Atlanta, GA.
(3) ASHRAE Standard 62-1989. Ventilation for Acceptable Indoor Air Quality. The American Society
of Heating, Refrigerating, and Air-Conditioning Engineers, Inc. Atlanta, GA.
(4) ASHRAE 1992 Handbook. HVAC Systems and Eauivment. The American Society of Heating,
Refrigerating, and Air-Conditioning Engineers, Inc. Atlanta, GA.
(5) Mcquiston, F.C. and Parker, J.D. (1988) Heating. Ventilating, and Air-Conditioning: Anab/sis and
Design. John Wiley and Sons, New York, NY.
(6) Parker, J.D.(1992) HVAC Systems in the Current Stock of U.S. K-12 Schools. EPA-600/R-92-125
(NTIS PB92-218338).
(7) Colen, H.R. (1990) HVAC Systems Evaluation. R.S. Means Company, Inc. Kingston, MA.
(8) AABC National Standards (1989) National Standards for Testing and Balancing Heating. Ventilating.
and Air Conditioning Systems. Associated Air Balance Council, Washington, D.C.
(9) SMACNA Standard (1990) HVAC Systems-Duct Design. The Sheet Metal and Air Conditioning
Contractors' National Association, Chantilly, VA.
(10) Samfield, M. (1993) HVAC Systems as a Tool in Controlling Indoor Air Quality: A Literature
Review. Report prepared for the State of Florida, Department of Community Affairs. U.S. Environmental
Protection Agency, O-1D1884NATA, Research Triangle Park, NC 27711.
(11) Communication from William Fisk, Radon Group Leader, Indoor Environment Program, Lawrence
Berkeley Laboratory, Berkeley Ca. to David Sanchez and Tim Dyess U.S. EPA, Air and Energy
Engineering Research Laboratory, Research Triangle Park, NC, dated November 3, 1992.
(12) Haines, R.W. (1988) HVAC Systems Design Handbook. TAB Books Inc., Blue Ridge Summit,
PA.
19
-------
loeillon
description
1 1ST RECEPTION
2 TOM SCOTTS OFFICE
3 JOE 8COTT8 OFFICE
4 TOM ANDERSON'S OFFICE
8 JOANN'S BIO OFFICE
8 JOANN'S 2ND OFFICE
7 KEREN'S OFFICE
I FRONT RECEPTION AREA
9 1ST LOBBY
10 RM 101
11 CIO WEST
12 CIDEAST
13 RM 200 SOUTH
14 RM 200 NORTH
18 RM 200 WEST
18 RM 201 EAST
17 RM 201 COMPUTER
18 RM 300 SOUTH
19 RM 300 NORTH
20 RM 301 WEST
21 RM 301 EAST
22 RM 301 COMPUTER
23 ELEVATOR
24 WAREHOUSE
28 ANITA'S OFFICE
23 MAINTENANCE OFFICE
27 JW6COTTS OFFICE
Bu.Un.
ExtoUng 13.8 ofm Oofm 8.5 elm
08/09/92 08/16/92 07/03/92 07/08/92
08/18/92 07/01/82 07/08/92 07/13/92
Av*r«g« Rldon Ltvcl
HlghRudna
Low Reading
222
1.81
1.61
1.44
1.16
128
120
124
0.74
0.80
0.80
1.39
1X0
1.49
1.39
1.80
0.97
0.66
2.61
1.14
129
1.11
0.81
1,32
2.81
0.61
M1
1.81
1.77
1.47
1X3
1X2
1.36
128
129
0.98
1.00
1.03
127
1.30
1X6
US
0.76
0.83
122
0.88
1.01
1.16
128
326
3.67
4.07
3.08
2.85
2.78
2.60
2.17
3.16
3.14
2.67
2X9
2.66
3.38
2.61
2.91
2.24
2.40
2.72
2.06
1.67
1.93
1.80
1.88
1.40
1.44
1.51
1.87
1.64
2.51
2X4
2.42
1.92
1.68
1.83
1.47
0.82
1.99
1.9S
2.00
2.98 1.90
4.07 2.72
2.17 0.62
FIGURE 1. FINANCIAL CENTER BUILDING, AVERAGE RADON CONCENTRATION
02/01
o
01/26 01/28 01/30
01/27 01/29 01/31 02/02
DATE
FIGURE 2. LIFE AND LEARNING CENTER, RADON CONCENTRATION
20
-------
An Analytical Solution to Describe the Pressure/Flow
Relationship in EPA's Soil-gas Chamber
by: Ronald B. Mosley
U.S. Environmental Protection agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
In order to better understand the production and transport of radon
and other potential indoor air pollutants such as pesticides, bacteria,
and landfill gases in soil, EPA has constructed a 16 m3 soil-filled
chamber instrumented to measure such parameters as pressure differences,
air flow rates, and contaminant concentrations. An analytical model
has been developed to describe the relationship between pressure
differential and flow rate. This paper applies the classic method of
images to simulate the boundary conditions imposed by the finite
dimensions of the chamber. The resulting influence on constant pressure
contours and the streamlines are shown.
INTRODUCTION
A number of indoor air pollutants, including pesticides, bacteria,
landfill gases, and radon, originate in the soil surrounding a building.
In order to better understand the processes by which these pollutants
enter the building as well as the most appropriate methods for
preventing their entry, a study of the transport mechanisms in soil was
initiated. Numerous efforts to correlate measured indoor radon levels
with radon values measured in the surrounding soil led to the conclusion
that better experimental control of the parameters that influence the
rate of radon entry would be reguired in order to develop and test
models to predict indoor radon levels. Experimental control of these
parameters is best accomplished in a laboratory environment.
Consequently, a soil-gas chamber for studying radon movement in soil was
designed.
While it has been widely purported (1-8) that pressure driven flow
is the primary process by which radon and other soil-gas borne
pollutants enter buildings, it is becoming more apparent (9-10) that
diffusion may make an important contribution to indoor radon when the
resulting indoor levels are in the modest range (less than 200 Bq m"3) .
While the diffusive mechanisms would be expected to differ somewhat for
the different pollutants, the advective processes would affect most of
the pollutants in a similar manner. This paper will concentrate on the
advective transport processes, specifically on the pressure/flow
relationships.
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication.
21
-------
A previous analytical model was developed (11) to describe the flow
of soil gas into a cylindrical tube buried in the soil. Flow into this
tube can be representative of either flow of soil gas into a drain tile
surrounding the foundation of a building or flow directly through a
perimeter crack between the floor slab and the wall. In reference (11),
an analytical solution was developed for an infinitely long porous tube
buried in the soil parallel to an infinite plane (the surface of the
soil).
DEVELOPMENT OF THE EQUATIONS
Due to the practical limit on its dimensions, the laboratory
chamber, designed to study soil transport characteristics, can not
adequately simulate the infinite dimensions assumed by the analytical
solution. The laboratory chamber is 2 m in height and 2 m x 4 m in
area. The details of this chamber are described elsewhere (12, 13). In
order to modify the analytical solution to match the boundary conditions
of the finite box of soil, the classic method of images is employed.
The method works because the linearity of Laplace's equation allows
superposition of several solutions that can cooperatively satisfy the
appropriate boundary conditions. The solutions that are applicable
arise from a series of virtual images of the original solution whose
origins lie outside the domain of interest. The method is very much
analogous to using images of point electrical charges to simulate the
interaction between a point charge and a grounded conducting surface
such as a plane.
The equations governing transport under steady state conditions can
be written in the form:
V*P 0 , (1)
V=-^Vf , (2)
, (3)
G
where V is the gradient operator, P is the pressure difference
responsible for the flow, v is the superficial velocity, k is the
permeability of the soil to air movement, \i is the dynamic viscosity of
the soil gas, De is the effective diffusion coefficient of radon in the
soil, C is the activity concentration of radon, e is the porosity of the
soil, G is the generation rate of radon activity, and \ is the radon
decay constant. For simplicity in solving these equations, the soil
properties of permeability, porosity, and generation rate are taken to
be uniform and isotropic.
Since the pressures differences involved are a thousand times
smaller than atmospheric pressure, the flow will be considered
incompressible. Equation (1) can be solved to yield the pressure
distribution within the soil. Equation (2) then yields the velocity
distribution throughout the soil. Solution of equation (3) provides the
22
-------
concentration of radon at any location within the soil. Equation (3) is
provided here primarily for completeness. Its solution will not be
discussed in this paper.
The solution of equation (1) developed in reference (11) can be
represented by
(4)
where Pc is the pressure in the cylinder, h is the depth of the cylinder
below the surface of the soil, and b is the radius of the cylinder.
When air flow is restricted to entering only through the finite area of
soil surface defined by the impenetrable walls of a chamber, the
solution becomes
,y)=P/lnf^±
-------
(8)
and
vy(x.y) =
-4*J>
(9)
Note that the infinite sums are also reflected in both the stream
function and the velocity expressions. Equation (9) can be integrated
over the surface of the soil where y = 0 to yield the total flow rate
through the system
(10)
By measuring Q and Pc, this equation and equation (6) provide an easy
method of measuring the bulk permeability of the soil in the laboratory
chamber.
DISCUSSION AND CONCLUSIONS
Figure 1 shows constant pressure contours for the central vertical
plane along the length of the soil chamber. The cylinder is located at
the center of the vertical y-axis. In the absence of the boundaries
formed by the sides of the chamber, the constant pressure curves were
circles centered on the vertical axis (11). Note the considerable
deviation from circles induced by the finite dimensions of the chamber.
This effect occurs because the boundary conditions require that no
horizontal flow results at the impermeable surfaces. Figure 2
illustrates the streamlines. Streamlines which previously extended
toward infinity in circular arcs now flow vertically near the sides of
the chamber (indicating that the boundary conditions are satisfied).
Despite the very considerable modifications in the isobars and the
streamlines, however, the total flow rate is diminished by about only
10% from its value in an infinite block of uniform soil. In both
Figures (1) and (2), only half of the chamber is shown. To the extent
that the soil is uniform and isotropic, the flow pattern is completely
symmetrical about the vertical y-axis. The other half of the chamber to
the left would be represented by the mirror image of the curves shown.
24
-------
REFERENCES
1. Bruno, R.C. Sources of indoor radon in houses: a review. JAPCA, v.
33, no. 2, p. 105-109, 1983.
2. Nero, A.V. and Nazaroff, W.W. Radiat. Prot. Dosim. v. 7, p. 23-39,
1984.
3. Akerblom, G., Anderson, P., and Clavenajo, B. Radiat. Prot. Dosim.
V. 7, p. 49-54, 1984.
4. Nazaroff, W.W. and Doyle, S.M. Health Phys. , v. 48, p. 265-281,
1985.
5. Sextro, R.G., Moed, B.A., Nazaroff, W.W., Revzan, K.L., and Nero,
A.V. In: Radon and Its decay Products: Occurrence, Properties, and
Health Effects: Hopke, P., ed.; ACS Symposium Series 331; American
Chemical Society: Washington D.C., p. 57-112, 1987.
6. Turk, B.H., Prill, R.J., Grimsrud, D.T., Moed, B.A., and Sextro,
R.G. J. Air Waste Manage. Assoc. v. 40, p. 498-506, 1990.
7. Garbesi, K., Sextro, R.G., Fisk, W.J., Modera, M.P., and
Revzan,K.L. Soil-gas entry into an experimental basement: model
measurement comparisons and seasonal effects. Eviron. Sci. Technol.
V. 27, no. 3, p. 466-473, 1993.
8. Nazaroff, W.W., Feustel, H., Nero, A.V., Revzan, K.L., Grimsrud,
D.T. , Essling, M.A., and Toohey, R.E. Radon transport into a
detached house with a basement. Atmospheric Environment, v. 19,
no. 1, p. 31-46, 1985.
9. Rogers, V.C. and Nielson, K.K. Data and models for radon transport
through concrete. In: Proceedings: the 1992 International Symposium
on Radon and Radon Reduction Technology: vol. 1.
10. Tanner, A.B. Radon migration in the ground: a supplementary review,
In: Gesell, T.F. and Lowder, W.M., eds., Natural radiation
environment III, U.S. Department of Energy Report, NTIS CONT-
780422, V. 1, p. 5-56, 1980.
11. Mosley, R.B. A simple model for describing radon migration and
entry into houses. In: Cross, F.T., ed. , Indoor Radon and Lung
Cancer: Reality or Myth?: Part 1, Battelle Press, Columbus, Ohio,
V. 1, p. 337-356, 1992.
12 Mosley, R.B. Model based pilot scale research facility for studying
production, transport, and entry of radon into structures. In:
Proceedings: the 1992 International Symposium on Radon and Radon
Reduction Technology: vol. 1.
13 Menetrez, M.Y., Mosley, R.B., Snoody, R. , Ratanaphruks, K. , and
Brubaker, S.A., Jr. Evaluation of radon movement through soil and
foundation substructures. Presented at the 1993 International
Symposium on Measurement of Toxic and Related Air Pollutants,
Durham, NC, May 4-7, 1993.
25
-------
distance (m)
Figure 1. Calculated pressure contours
CM
0 1
distance (m)
Figure 2 Calculated streamlines
26
-------
Session 2
Air Pollution
Dispersion Modeling
-------
Uncertainties Of Using Short-Term Air Quality Concentrations
To Estimate Annual Average Concentrations
Norman A. Huey
U.S. EPA Region 8
999 18th Street
Denver, Colorado 80202-2466
and
George J. Schewe
Environmental Quality Management, Inc.
1310 Kemper Meadow Drive, Suite 100
Cincinnati Ohio 45240
ABSTRACT
EPA "Guideline" modeling procedures1 allow the use of multiplying
factors to convert 1-hour maximum concentration estimates to averaging
times of 3, 8 and 24 hours. These factors as well as a factor to
convert annual averages2 are intended for inclusion into the T-SCREEN
model3. In an effort to better understand the uncertainties in these
conversions and in concentrations estimations using non-representative
meteorological data, the Air/Superfund program funded a sensitivity
study4 to determine the range of effect that meteorological data might
have on model estimations of 1-hour, 3-hour, 8-hour, 24-hour and annual
average concentrations. Superfund type area and point sources were
modeled using the Industrial Source Complex Short Term (ISC2) model with
input from twenty seven widely spread National Weather Service (NWS)
stations.
This paper presents the findings of the study as they relate to air
stripping of contaminated water. SCREEN5 model 1-hour estimates and
converted 3-hour, 8-hour, 24-hour, and annual average values are
compared to the ISC2 model results.
Understanding the range of uncertainties of the estimated
concentrations is necessary to insure proper interpretation of such
estimates in decision making.
INTRODUCTION
Because it is easier to estimate or monitor short term ambient air
concentrations, short term average concentrations are often used to
estimate long term average concentrations. The SCREEN model was
developed to estimate maximum 1-hour concentrations as a function of
source-receptor distance and meteorological variability. In order to
estimate maximum concentrations with longer averaging times, conversion
factors1"2 were developed (see Table 1) .
Table 1. One hour conversion factors.
Averaging Time Multiplying Factor
3 hour 0.9 (+/-0.1)
8 " 0.7 (+/-0.1)
24 " 0.4 (+/-0.1)
1 year 0.08 (+/-0.2)
29
-------
In order to have widespread applicability and a factor of safety,
these factors must produce conservative estimates of the expected
impacts. Ultraconservative screening Air Pathway Analyses (APAs) may
lead to unnecessary in-dept APAs and are not adequate for decision
making in the early phases of the Superfund process. Not knowing how
conservative the SCREEN model estimations might be, the Air/Superfund
Program decided to do meteorological sensitivity studies in order to
obtain an estimation of the amount of conservatism in APAs based on
screening methods.
STUDY APPROACH
Superfund type area and point sources were modeled using the
Industrial Source Complex Short Term (ISC2) model with meteorological
data from twenty-seven widely spread National Weather Service (NWS)
stations.
Superfund sources are generally low level sources and impacts of
interest are generally close to the source. All modeling for this study
was performed with the ISC2 model for ground level receptors assuming
flat terrain and using rural dispersion coefficients. This paper
presents the sensitivity study findings as they relate to air stripping
of contaminated water.
Tables 2 list the air stripper source input data which were used in
the model.
Table 2. Model inputs for the air stripper.
stack height = 10 m stripper height. 9 m
stack diameter = 0.62 m stripper diameter 3.6m
stack temperature = 293 °K emission rate 1 g/s
stack velocity = 8 m/s
Table 3 contains the ISC2 model output from the Denver NWS
meteorological data set for the air stripper assuming an emission rate
of 1 gram per second. The average annual impact in ug/m3 for any
receptor can be calculated by multiplying the corresponding factor by
the pollutant emission rate in grams per second. For example, the
ambient concentration resulting from an emission rate of 0.2 g/s at the
receptor with polar coordinates of 10" and 50 meters is 0.5 ug/m3 (2.5
x 0.2) .
Maximum concentration factors for each receptor distance,
disregarding receptor direction, are listed at the bottom of the table.
These values are typically the values that screening methods attempt to
estimate.
Table 4 contains the maximum annual impact value from each of the
twenty seven meteorological data sets in descending order of magnitude.
The highest values are listed at the top of the table and the lowest are
listed at the bottom of the table. SCREEN model estimates are listed at
the bottom of the table. There is good agreement between the SCREEN
estimates and the ISC highest maximum estimates (San Juan) for receptor
distances between 100 to 200 meters. When compared to the ISC2 lowest
maximum estimates (Atlantic city), the SCREEN estimates are higher by a
30
-------
factor of 4 to 24.
The model data outputs for each of the other averaging times and
each meteorological data set were searched for the highest maximum
concentration factors and for the lowest maximum concentration factors.
Table 3. Model output from Denver NWS meteorological data
in ug/m3 per g/s emission rate.
azimuthal
direction 50
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
MAXIMUM
2.5
2.2
2.0
1.6
1.2
0.8
0.6
0.7
0.8
0.9
1.1
1.1
1.1
1.2
1.4
1.9
2.3
2.3
1.8
1.5
1.5
1.5
1.7
1.7
1.6
1.6
1.8
1.9
1.9
2.0
2.2
2.2
2.1
2.1
2.2
2.4
2.5
distance from
100
9.8
8.7
7.3
5.3
3.9
2.4
1.7
2.1
2.5
2.8
3.5
3.8
4.0
4.3
4.7
6.3
8.4
8.9
6.6
5.1
4.4
4.6
5.0
4.9
4.5
4.5
4.9
5.1
5.1
5.5
5.9
5.8
5.7
6.2
6.6
8.7
9.8
ISO.
11.0
9.6
7.7
5.5
4.0
2.4
1.6
2.1
2.4
2.9
3.8
4.1
4.5
5.0
5.2
7.1
9.8
10.8
7.5
5.4
4.7
4.9
5.4
5.2
4.7
4.7
5.1
5.2
5.3
5.7
5.9
5.7
5.5
6.5
7.0
9.4
11.0
200
10
8
6
4
3
2
1
1
2
2
3
3
4
4
4
6
9
10
6
4
4
4
4
4
4
4
4
4
4
5
5
4
4
5
6
8
10
.0
.6
.7
.7
.4
.0
.4
.7
.0
.5
.4
.7
.0
.6
.7
.4
.0
.2
.7
.7
.1
.4
.8
.6
.1
.1
.4
.6
.6
.1
.1
.8
.6
.7
.3
.3
.2
source in meters
300
7
6
4
3
2
1
0
1
1
1
2
2
2
3
3
4
6
7
4
3
2
3
3
3
2
2
3
3
3
3
3
3
3
4
4
6
7
.4
.3
.7
.3
.4
.4
.9
.2
.3
.7
.4
.6
.8
.3
.4
.5
.4
.4
.7
.1
.7
.0
.4
.2
.8
.8
.0
.3
.3
.7
.6
.3
.1
.1
.6
.1
.4
400
6.0
5.0
3.6
2.5
1.7
1.0
0.6
0.8
1.0
1.3
1.8
1.9
2.0
2.4
2.4
3.2
4.6
5.4
3.4
2.2
1.9
2.2
2.4
2.3
2.0
2.0
2.1
2.4
2.4
2.8
2.8
2.5
2.3
3.2
3.7
5.0
6.0
500
5.2
4.2
2.9
2.0
1.3
0.7
0.5
0.6
0.8
1.0
1.3
1.4
1.5
1.8
1.8
2.4
3.5
4.1
2.5
1.6
1.4
1.6
1.8
1.7
1.4
1.5
1.6
1.8
1.9
2.3
2.2
2.0
1.8
2.7
3.2
4.3
5.2
750
3.9
3.0
2.0
1.3
0.9
0.5
0.3
0.4
0.5
0.6
0.8
0.8
0.9
1.0
1.0
1.4
2.0
2.4
1.4
0.9
0.8
0.9
1.0
1.0
0.8
0.8
0.9
1.0
1.2
1.6
1.5
1.4
1.3
2.0
2.5
3.4
3.9
1000
3.0
2.3
1.5
0.9
0.6
0.3
0.2
0.3
0.4
0.4
0.6
0.5
0.6
0.7
0.7
0.9
1.3
1.6
0.9
0.6
0.5
0.6
0.7
0.6
0.5
0.5-
0.6
0.7
0.9
1.2
1.2
1.1
1.0
1.5
2.0
2.6
3.0
The highest values come from the meteorological set which represents
poor atmospheric dispersion. The lowest values come from the
meteorological set which represents good atmospheric dispersion. The
-------
SCREEN model was run and the one hour concentration values were
converted to longer concentration averaging times using the conversion
factors. These values are shown in Table 5 and compared to the ISC2
results for the same concentration averaging times. The ratios of
SCREEN to ISC2 concentrations are also shown in Table 5.
Table 4. Maximum annual concentration estimations
for each meteorological data set in ug/m3
per year per g/s emission rate.
meteorological
data set
San Juan
San Francisco
Casper
Salem
Wichita
Helena
Salt Lake C.
Phoenix
Harrisburg
Spokane
Knoxville
Hartford
Atlanta
La Guardia
Philadelphia
Denver
Houston
Baton Rouge
Chicago
Wilmington
Bismarck
Cleveland
Huntington
Raleigh
Birmingham
Charleston
Atlantic City
distance
50
7
4
5
3
3
3
4
5
3
2
3
2
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
100
29
18
18
13
16
14
12
13
13
11
10
9
11
11
10
10
9
8
9
9
9
8
8
7
7
7
7
150
30
21
20
19
19
18
16
14
14
13
13
13
12
12
11
11
11
11
10
10
10
10
10
9
9
9
8
from
200
26
19
17
18
18
17
16
13
13
12
12
13
11
11
10
10
10
10
9
9
9
10
10
9
8
9
7
source in
300
18
13
11
14
13
13
12
10
9
8
10
10
8
7
8
7
8
8
7
6
6
9
9
7
7
7
5
400
13
10
8
10
10
9
10
8
6
6
a
8
6
5
e
6
e
6
5
5
4
7
7
e
6
5
4
meters
500
11
8
6
8
8
7
8
7
5
5
6
6
5
4
4
5
5
5
4
4
3
6
6
5
5
4
3
750
8
5
3
5
5
4
5
5
3
3
4
4
3
2
3
4
3
3
2
3
2
4
4
3
5
3
2
1000
6
3
2
3
3
3
4
3
2
2
3
3
2
1
2
3
2
2
2
2
1
3
3
2
4
2
1
SCREEN
26 28 28 29 28 26 25 26
24
RESULTS
Examination of the ratios for the 1-hour averaging time reveal that
for the location with the poorest dispersion the SCREEN model results
agree quite well with the ISC2 results except possibly for an
underestimation of about 20% of the ISC2 result at the point (distance)
of maximum impact. For the location with the best dispersion
meteorology, the SCREEN model overestimated the ISC2 result at the 1000
meter distant receptor by about 70%.
Examination of the ratios for the 3-hour averaging time reveal that
32
-------
for the location with the poorest dispersion, the SCREEN model results
agree quite well with the ISC2 results. For the location with the best
dispersion, the SCREEN model results overestimate the ISC2 model results
by a factor of 1.4 to 4.0 depending on the receptor distance from the
air stripper.
Table 5. Comparison of ISC2 results to SCREEN results
in ug/m3 per gram per second emission rate.
maximum 1-HR estimat
SCREEN
ISC2 highest maxs .
ISC2 lowest maxs.
SCREEN/ISC2 high
SCREEN/ISC2/10W
50
ions
330
290
270
1.1
1.2
- - -U-Li
100
350
320
270
1.1
1.3
»uaiii;i
150
350
450
310
0.8
1.1
e J.-LUJ
200
360
470
270
0.8
1.3
III HUU
300
340
460
250
0.7
1.4
i(je x.
400
330
390
210
0.8
1.6
11 i[L£:i_
500
310
330
190
0.9
1.6
750
320
300
190
1.1
1.7
1000
290
280
170
1.0
1.7
maximum 3-HR estimations:
SCREEN (0.9)
ISC2 highest maxs .
ISC2 lowest maxs.
SCREEN/ISC2 high
SCREEN/ISC2/low
300
280
150
1.1
2.0
320
310
230
1.0
1.4
320
310
210
1.0
1.5
320
270
190
1.2
1.7
310
260
170
1.2
1.8
300
240
130
1.3
2.3
280
240
110
1.2
2.5
290
240
80
1.2
3.6
280
230
70
1.2
4.0
maximum 8-HR estimations:
SCREEN (0.7)
ISC2 highest maxs.
ISC2 lowest maxs .
SCREEN/XSC2 high
SCREEN/ISC2/10W
230
210
43
5.3
250
250
84
3.0
250
230
77
3.2
250
230
69
3.6
240
200
50
4.8
230
170
42
5.5
220
160
32
6.9
220
150
30
7.3
210
140
24
1C
. _>
8.8
maximum 24 -HR estimations:
SCREEN (0.4)
ISC2 highest maxs.
ISC2 lowest maxs .
SCREEN/ISC2 high
SCREEN/ISC2/10W
130
57
30
2.3
4.3
140
110
70
1.3
2.0
140
140
77
1.0
1.8
140
150
70
0.9
2.0
140
130
51
1.1
2.7
130
98
40
1.3
3.3
120
74
32
1.6
3.8
130
49
19
2.7
6.8
120
38
15
3.2
8.0
maximum 1-YR estimations:
SCREEN (0.08)
ISC2 highest maxs .
ISC2 lowest maxs .
SCREEN/ISC2 high
SCREEN/ISC2/10W
26
7
2
3.7
13
28
29
7
1.0
4
28
30
8
0.9
4
29
26
7
1.1
4
28
18
5
1.6
6
26
13
4
2.0
7
25
11
3
2.3
8
26
8
2
3.3
13
24
6
1
4.0
24
Examination of the ratios for the 8-hour averaging time reveal that
for the location with the poorest dispersion the SCREEN model results
agree quite well with the ISC2 results for receptors near the point of
maximum impact. For the location with the best dispersion, the SCREEN
model results overestimate the ISC2 model results by a factor of 3.2 to
33
-------
8.8 depending on the receptor distance from the air stripper.
Examination of the ratios for the 24-hour averaging time reveal
that for the location with the poorest dispersion, the SCREEN model
results agree quite well with the ISC2 results for receptors near the
point of maximum impact. For the location with the best dispersion the
SCREEN model results overestimate the ISC2 model results by a factor of
1.8 to 8.0 depending on the receptor distance from the air stripper.
Examination of the ratios for the annual averaging time reveal that
for the location with the poorest dispersion the SCREEN model results
agree quite well with the ISC2 results for receptors near the point of
maximum impact. For the location with the best dispersion, the SCREEN
model results overestimate the ISC2 model results by a factor of 3.5 to
24 depending on the receptor distance from the air stripper.
CONCLUSIONS
Data from this study substantiates that in cities with poor
dispersion, the SCREEN model results agree well with the ISC2 model
results at the point of maximum impact.
Data from the study reveals that in cities with good dispersion,
the SCREEN model overestimates ISC2 model results by factors of
approximately 1.1, 1.5, 3, 2, and 4 respectively for concentration
averaging times of 1-hr, 3-hr, 8-hr, 24-hr, and 1-yr at the point of
maximum impact.
Overestimation by the SCREEN model of the ISC2 model results
increases with source-receptor distance and with concentration averaging
time. When 1-yr averaging time concentrations are derived from 1-hr
averaging time values, they may be high by as much as a factor of 24.
Less conservative estimation of air strippers maximum annual
average impacts might be made using factors such as those in Table 4.
Using the San Juan values should result in estimations which are not
conservative beyond a factor of 4. Using the Salt Lake City values
should give estimations within a factor of 2.
The results of this study are based upon modeling of a typical air
stripper and may not be applicable to other types of emission sources.
REFERENCES
1. EPA-450/2-78-027R, "Guideline On Air Quality Models (Revised)",
Office of Air Quality Planning and Standards, Research Triangle
Park, North Carolina.
2. EPA-450/4-92-001, "A Tiered Modeling Approach for Assessing the
Risk Due to Sources of Hazardous Air Pollutants", Section 3.0,
Office of Air Quality Planning and Standards, Research Triangle
Park, North Carolina. March 1992.
3. EPA-450/4-90-013, "User's Guide To TSCREEN", Office of Air Quality
Planning and Standards, Research Triangle Park, North Carolina.
December 1990.
4. EPA Contract No. 68D00124, Work Assignment No 1-127 report;
"Generation of Concentration Factors and Nomographs for Superfund
Point and Area Sources". October, 1992.
5. EPA-450.4-88-010, "Screening Procedures for Estimating the Air
Quality Impact of Stationary Sources". August 1988 (SCREEN Model
User's Guide in appendix A).
34
-------
FURTHER1 DEVELOPMENT OF EMPIRICAL FACTORS FOR
ESTIMATING AIR TOXIC SOURCE IMPACTS
George J. Schewe, CCM
Director of Air Quality
Environmental Quality Management, Inc.
1310 Kemper Meadow Drive, Suite 100
Cincinnati, Ohio 45240
Norman A. Huey
Air Superfund Coordinator
U.S. EPA Region 8
999 18th Street, Suite 500
Denver, Colorado 80202-2405
ABSTRACT
Atmospheric dispersion modeling estimates air pollutant concentrations as a function
of receptor location, source type, and meteorological data set. In the early planning stages of
Superfund activities, specific information about receptor location, source type and
meteorological data is usually not known. This study was performed to further develop a tool
by which gross air quality impacts could be made for specific types of Superfund sources2.
By modeling generic typical sources repetitively with widely spread meteorological data sets,
empirical factors representing conservative estimates of maximum downwind concentrations
were obtained. Empirical factors for five averaging times (1-hr, 3-hr, 8-hr, 24-hr, and annual)
were developed from the results of normalized modeling of generic sources. Source impacts
can be estimated for receptors at various distances by multiplying the emission rate by the
appropriate dispersion factor for each source type/receptor distance combination. Area and
point sources typical of Superfund sites were modeled using the ISCST2 Model and
meteorological data from twenty-seven different locations within the United States. Empirical
factors (dispersion coefficients) for predicting concentrations as a function of receptor
distance were derived for an estimation of the maximum 1-hr, 3-hr, 8-hr, 24-hr, and annual
averaging times at various receptor distances.
INTRODUCTION
Dispersion modeling of nine Superfund-type sources was performed to calculate
normalized (unit emissions) concentration estimates versus distance. This modeling was
performed to supplement and extend former U.S. EPA calculations in "Hazardous Waste
TSDF - Fugitive Paniculate Matter Air Emissions Guidance Document"3 The TSDF modeling
was performed using the ISCLT Model and only annual average concentrations were
estimated. To extend the usability of this type of estimation tool, concentrations were
generated for additional averaging times including 1-hr, 3-hr, 8-hr, and 24-hr. In this analysis
the geometric mean concentration was calculated for the maximum concentrations at all
sites. In addition, the standard geometric deviation was calculated for each set of maximum
concentrations (i.e., at each downwind distance. This combination of statistics and the listing
of the highest and the lowest concentrations are given in the report and can be used to
estimate the potential conservatism in air quality results calculated from the empirical factors.
35
-------
ANALYSIS OBJECTIVES
The purpose of this work was to examine the sensitivity of the Industrial Source
Complex (ISC) Model4 predictions to location specific meteorological data and to develop
empirical factors which will approximate ISC model results. An "empirical factor" in this
analysis is a factor which when multiplied by an emission rate results in a conservative
estimate of the ambient air concentration that would have been estimated by running a more
detailed model. The empirical factors were developed from results of normalized modeling of
generic point and area sources. These factors can be used to estimate Superfund type
source impacts for receptors at various distances by multiplying the emission rate by the
appropriate empirical factor. Maximum annual average concentrations are included because
they are necessary for chronic risk assessment analyses and short-term estimates are
included to assess acute impacts.
MODEL PROTOCOL
Model Methodology
The updated version of the short-term Industrial Source Complex Model (ISCST2) was
selected for this analysis. The ISCST2 Model is similar to its earlier version from 1979 to
1991 but has been reprogrammed and has a new format for imputing data in batch mode
using keywords to indicate specific data components. The ISCST2 Model includes all model
options and algorithms of the former versions and was operational on a personal computer
(486/33 with extended RAM memory). The steps taken to generate the curves of normalized
concentration estimates versus downwind distance using ISCST2 were very specific and
straightforward.
Select Sources
Nine sources representing those potentially found at Superfund sites were selected for
modeling. These included six area sources ranging in size from a small 1 m by 1 m area to
a 500 m by 500 m area. Characterization of the area sources was selected to characterize
similar Superfund-type areas of fugitive emissions ranging possibly from a short-term
accidental spill or dumping operation over a small area to a site-wide fugitive emission due to
an open lagoon, unpaved roads, or open trench air stripping. The other three sources were
point sources related to the representation on air strippers and soil vapor extraction
operations. Emissions from a medium size air stripper and two sizes of soil vapor extraction
units were modeled. Structure downwash of effluent was considered by including the
dimensions of the stripper or extraction units themselves. These structures are typically
adjacent to the vent and are the most likely to cause downwash of effluent plumes. The
dimensions of these structures were taken from representative sizes and shapes of air
strippers5 and SVE units6.
Meteorological Data
Surface meteorological observations of wind speed, wind direction, temperature,
opaque cloud cover and other surface National Weather Service (NWS) parameters were
obtained for 27 surface sites for the analysis. The year of most recent data was used along
with associated nearby mixing height data.
Receptor Locations
In the ISCST2 Model analysis of each of the nine sources, a receptor grid was used
which covered all downwind directions (every 10 degrees) at a number of downwind
36
-------
distances (from six to nine depending on source). The downwind distances were selected to
obtain both a maximum concentration and a profile of concentration versus distance
downwind. Receptor distances for area sources were measured from the center of the area
source.
RESULTS
Table I contains the empirical factors which were derived from this work. The project
report contains additional information which enables the user to understand the amount of
potential conservatism which may be included in estimates made with these factors. Table II
and Figure 1 are examples of the project report graphs which demonstrates the amount of
conservatism which could be in estimates made using these empirical factors.
CONCLUSIONS
The empirical factors can be used to make preliminary estimations of Superfund type
source emission impacts. These determinations will be useful in the Superfund Process prior
to the final design phase when more detailed information about the source becomes
available. Estimations using these factors should be of particular value in the identification of
principal contaminants of concern, in the design of toxic monitoring plans (determination of
needed detection limits), in the selection of remediation alternatives (especially if the
emission control cost might makes the technology less desirable, and in the design of
remediation activities (level of operation). The project report contains information for each
source/concentration averaging time as depicted in Table 2 and Figure 1. This information
can be used to obtain an indication of the potential amount of conservatism in the estimates.
This information will in-turn indicate the probability that more detailed information might be
beneficial before decisions are made on the empirical factor concentration data.
REFERENCES
1. N.A. Huey & G.J. Schewe, Empirical Factor Estimation of Air Toxic Source Impacts",
Proceedings of the HMC/Superfund '92 Conference. Hazardous Materials Control
Resources Institute, Greenbelt, Maryland, 1992, pg. 850-852.
2. EQ 1992, "Modeling Analysis for Idealized Point and Area Sources", Environmental
Quality Management, Inc., Project No. 5025-8, Cincinnati, Ohio, January.
3. EPA 1989, "Hazardous Waste TSDF - Fugitive Paniculate Matter Air Emissions
Guidance Document", EPA-450/3-89-019, U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, Research Triangle Park, NC, May.
4. EPA 1992, "User's Guide for the Industrial Source Complex (ISC2) Dispersion Models,
Volume 1 User Instructions", U.S. Environmental Protection Agency, EPA-450/4-92-
008a, Research Triangle Park, NC, March.
5. EPA 1991, "Estimation of Air Impacts For Air Stripping Of Contaminated Water", EPA-
450/1-91-002, U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC, May
6. EPA 1992, "Estimation of Air Impacts For Soil Vapor Extraction (SVE) Systems", EPA-
450/1-92-001, U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC, January
37
-------
Table I. Empirical Factors in ug/m3 per mg/s emission rate
Small Soil Vapor Extraction Point Source (500 cfm):
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
receptor distance in meters
50 100 150 200 300
13 7.3 5.0 3.8 2.6
5.5 2.8 1.8 1.3 1.0
3.4 1.6 1.0 .89 .70
2.2 1.0 .58 .39 .29
.41 .17 .10 .070 .043
400
2.0
.80
.55
.23
.029
500
1.6
.64
.44
.18
.021
750
1.1
.42
.29
.12
.012
1000
.85
.30
.20
.081
.008
Medium Soil Vapor Extraction Point Source (3000 cfm):
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
Medium Air Stripper
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
receptor distance in meters
50 100 150 200 300
1.0 .98 .99 .84 .63
.58 .59 .49 .44 .32
.49 .45 .36 .31 .21
.37 .32 .30 .26 .17
.075 .066 .045 .032 .018
Point Source (5000 cfm):
receptor distance in meters
50 100 150 200 300
.29 .32 .45 .47 .46
.28 .31 .31 .27 .26
.21 .25 .23 .23 .20
.057 .11 .14 .15 .13
.007 .029 .030 .026 .018
-
400
.53
.25
.16
.12
.012
--
400
.39
.24
.17
.10
.013
500
.44
.19
.12
.083
.009
500
.32
.24
.16
.074
.011
750
.30
.13
.080
.044
.006
750
.30
.24
.15
.049
.008
1000
.23
.10
.065
.030
.005
1000
.28
.22
.14
.038
.006
1x1 Meter Area Source:
averaging receptor distance in meters
time
1-hour
3-hour
8-hour
24-hour
annual
5
3200
2500
1600
690
120
JO.
1400
960
690
280
35
20
510
330
230
92
10
40
160
100
70
28
2.9
100
33
21
14
5.6
.55
50 x 50 Meter Area Source:
averaging receptor distance in meters
time
1-hour
3-hour
8-hour
24-hour
annual
50
9.1
8.1
6.8
3.4
1.1
100
6.8
5.1
3.4
2.0
.40
150
6.3
3.7
2.4
1.2
.21
200
5.8
2.7
1.8
.77
.13
300
4.8
1.8
1.2
.46
.068
400
3.9
1.5
.95
.36
.042
500
3.3
1.2
.80
.29
.029
750
2.3
.86
.56
.20
.014
1000
1.8
.66
.43
.15
.009
38
-------
100 x 100 Meter
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
150 x 150 Meter
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
300 x 300 Meter
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
500 x 500 Meter
averaging
time
1-hour
3-hour
8-hour
24-hour
annual
Area Source:
— receptor distance
100 150 200
3.4 3.4 3.4
2.3 1.8 1.4
1.9 1.2 .96
.95 .69 .56
.29 .17 .11
Area Source:
— receptor distance
150 200 300
2.3 2.3 2.3
1.1 .91 .85
.92 .68 .56
.38 .29 .24
.14 .094 .054
Area Source:
— receptor distance
300 400 500
1.1 1.1 1.1
.43 .42 .42
.27 .25 .25
.13 .12 .11
.040 .027 .020
Area Source:
(values are in
— receptor distance
300 400 500
680 680 680
250 250 250
150 150 150
89 86 85
16 10 6.7
in meters —
300 400 500 750 1000
3.2 2.9 2.7 2.1 1.7
1.2 1.1 1.0 .79 .64
.74 .70 .64 .51 .41
.48 .45 .42 .34 .28
.061 .038 .027 .014 .009
in meters —
400 500 750 1000 1200
2.2 2.1 1.8 1.6 1.4
.79 .78 .69 .59 .51
.49 .48 .43 .37 .33
.22 .20 .16 .13 .12
.036 .025 .013 .008 .006
in meters —
750 1000 1200 1500 2000
1.1 1.0 .99 .92 .80
.40 .39 .38 .35 .30
.24 .24 .23 .22 .29
.11 .10 .090 .080 .068
.012 .008 .006 .004 .003
ug/m3 per g/s emission rate)
in meters —
750 1000 1200 1500 2000
670 650 620 600 580
240 240 230 230 220
150 150 140 140 140
84 82 79 75 69
5.0 3.8 3.1 2.5 2.2
39
-------
Table II. Geometric mean, geometric standard deviation and extreme concentrations
in ug/m3 per gram/second emission rate averaged over all 27 sites for a medium soil
vapor extraction system.
50 100
G-Mean 270 240
G-STD 1.2 1.2
•• 27 (total number of sites)
Receptor Distance (meters)
150 200 300 400 500
Max-Max
Min-Max
Min-Min
370
180
22
320
180
23
180
1.2
300
130
23
140
1.2
260
96
17
92
1.3
170
62
10
63
1.3
120
44
8
46
1.3
83
33
6
750
27
1.2
44
20
3
1000
20
1.2
30
14
2
n = 972 (36 x 27 = total number of receptor locations)
Figure 1. Plot of 24-hour average concentration in ug/m3 per mg/s emission rate as
a function of downwind distance in meters for a medium soil vapor
extraction unit.
Medium SVE System. Geometric Mean
a Max-Max.
0.4 0.6
(Thousands)
Distance from source (meters)
GMean O Mln-Max.
Max-Max curve represents the maximum anticipated concentration in the city with the poorest dispersion.
GMean curve represents the maximum anticipated concentration in the city with median dispersion.
MinMax curve represents the maximum anticipated concentration in the city with the best dispersion.
MinMin curve represents the minimum anticipated concentration in the city with the best dispersion.
40
-------
Evaluating the Impact of Subsurface Contaminants
On Indoor Air Quality Using Estimates
From An Advective-Diflusive Transport Model
Wen-Whai U PhX>. and Marybeth Long
ENVIRON Corporation, 210 Carnegie Center, Suite 201, Princeton, New Jersey 08540
ABSTRACT
Subsurface contamination can affect indoor air quality and may adversely impact human health. Chemical vapors Emanating
from ground water and soil can enter a building through its foundation. The presence of pesticides and other toxic
chemicals in indoor air has been found to cause risks to human health. The ability to accurately predict indoor air
concentrations of air toxics attributable to subsurface contamination is, therefore, important in determining acceptable levels
of residual soil and ground water concentrations beneath buildings.
This paper presents an exact solution to a one-dimensional transport equation for determining the mass flux of
chemicals entering a building from the subsurface. The solution incorporates advection, diffusion and chemical phase
equilibrium. In addition, the paper presents a screening-level procedure for predicting chemical concentrations in indoor air
resulting from subsurface contamination beneath buildings. The screening-level methodology accounts for the advection and
diffusion mechanisms governing chemical transport in subsurface media, as well as gas/liquid partitioning in soil.
Calculations are presented to illustrate exact-solution and screening-level methodology comparisons.
41
-------
INTRODUCTION
Contaminants in the subsurface can infiltrate building substructures, accumulate in indoor air and adversely impact the
health of building inhabitants It is necessary to quantify the impact of these contaminants on indoor air quality, when
performing human health risk assessments, establishing soil and ground water cleanup levels, and evaluating subsurface
remediation alternatives associated with indoor exposure pathways. This paper presents an exact solution to a one-
dimensional fate and transport equation for determining the mass flux of chemicals entering a building from the subsurface.
In addition, the paper presents a simplified screening-level methodology for estimating the indoor air concentrations of these
chemicals, as well as example calculations.
Much of the current literature addressing subsurface contaminant intrusion into indoor environments focuses on the
transport of radon gas. Nazaroff et aLu investigate radon intrusion into buildings and identify pressure difference, radon
generation rate, and soil permeability as the three most critical parameters for estimating the radon intrusion rate. Loureiro
et aL3 present a three-dimensional simulation of radon transport into houses and confirm the conclusion of Nazaroff et aL1*.
For radon entry into houses, pressure-driven advection appears to be the controlling transport mechanism in highly
permeable subsurface media. Phase equilibrium and adsorption onto soil particles are not expected to be signifirant for an
inert radioactive gas.
In an attempt to address the inhalation risk associated with indoor air contaminants, methods used to estimate radon
intrusion into indoor environments via advection have been used to estimate indoor air concentrations of volatile organic
compounds (VOCs) emitted from subsurface media (Johnson and Ettinger" and Little et aL1). Unlike radon transport,
however, VOC transport is influenced by phase equilibrium and adsorption onto soil particles. McCoy and Rolston,6 for
example, demonstrate that hydrophilic and sorptive characteristics tend to retard vapor transport via advection in
unsaturated soil Thus, using radon gas transport models to describe organic vapor transport may result in inaccurate indoor
air concentration predictions.
This paper presents both refined and screening-level models for predicting the infiltration of chemical vapors into
indoor environments. The refined model describes the mass flux of contaminant entering a building from the subsurface.
This model is based on analytical solutions for subsurface vapor transport73'-10 incorporating phase equilibrium and
adsorption phenomena. In the screening-level methodology, a nondimensional parameter (the modified Peclet number) is
used to evaluate diffusive and advective transport mechanisms and determine which of these mechanisms governs
contaminant transport The modified Peclet number describes contaminant movement in terms of advection and diffusion
time scales and vapor-liquid partitic'Hijg. Once the dominant transport mechanism is iHp.nrifip.rfj indoor air concentrations
are estimated using simplified transport equations. Example calculations are presented at the end of this paper to illustrate
comparisons of the exact-solution and screening-level methodologies.
SUBSURFACE CONTAMINANT TRANSPORT
The following sections present models for estimating the contaminant flux, the relative significance of advective and
diffusive transport mechanisms, the indoor-outdoor pressure differential, and the indoor air concentration.
The One Dimensional Fate and Transport Model
In a one-dimensional, homogeneous subsurface porous medium, the generalized mass conservation equation states
that
'*«
where:
Cp total '•"•"""•"I concentrations per volume of soil g/cm3;
A net degradation rate, day"1;
x = soil depth below the surface, cm;
J = fhftmiral mass flux, g/cm2-s; and
t = time, s.
The chemical mass flux can be expressed as:
where:
42
-------
, (4)
Co = chemical concentration in soil gas, g/cm3 air;
CL = chemical concentration in soil moisture, g/cm3 water;
JG = mass flux in vapor form, g/on2-s;
JL = mass flux in liquid form, g/cma-s;
J, = chemical mass flux, g/cm2-s;
J. liquid flux, g/cm2-s;
DG effective air diffusion coefficient, cm2/s;
DL effective water diffusion coefficient, cm2/s;
D.J air diffusivity, cm2/s;
Du water diffusrvity, cm2/s;
i total soil porosity, 0 = 9, + 0m cm3/cm3;
S, air-filled soil porosity, cm3/cm3;
x soil depth below the surface, on; and
v vapor flow as described by Darcy's Law, cm/s.
The effective soil gas diffusion coefficient is equivalent to the air diffusrvity multiplied by a tortuosity factor, 8,wfi/62, to
account for the reduced flow area and increase path length of itiffhcing gas molecules in soil Soil air porosity may undergo
substantial changes over time as soil dries out and when moisture is added by rainfall or by watering. In addition, the void
space in soils may be unevenly distributed in multilayered soils. As a result, accurately accounting for soil porosity in an
analytical model is practically impossible. The use of average values of soil porosity may be the most practical approach.11
By inserting Equation 2 into Equation 1 and
• There is no steady water flux beneath the basement (Le, Jw = 0);
• There is no net degradation loss;
• Advection is driven by the indoor-outdoor pressure difference and the advection velocity is constant throughout
the medium;
• The vapor flow is described by Darcy's Law as;
v - ^ & ; and (5)
|1 2l
• A three-phase equilibrium exists throughout the medium and the sorption equilibrium follows a linear isotherm;
Equation 1 can be written as:
.} fc, (6)
D <» _ » w V _ p I _y — L
^-aT'lT^^r °1*0 ^J ax2
where RQ, RL are the foflowing partitioning coefficients:
Additional parameters used in the above equations are defined below:
k, permeability of soil to soil gas, on1,
P dynamic gas pressure, Pa (1 Pa = 10 g on-s2);
H = dynamic viscosity of soil gas, g/cm-sec.
RG = vapor-phase partitioning coefficient, dimensionless;
R^ = liquid-phase partitioning coefficient, dimensionless;
43
-------
P = soil particle density, g/cm3;
Pi dry bulk density of soil, pt = (1-0) Pt g/cm3;
9m moisture-filled soil porosity, 0m = w^, cm3/cm3;
K^ distribution coefficient, cm3/g; and
H = Henry's law constant, dimensionless.
Since chemical vapor diffusion through soil occurs very slowly, it is assumed that subsurface chemical concentrations in
soil gas (Cp) and soil particles (QJ are in local equilibrium. Thus, the soil concentration can be related to the vapor-phase
concentration, by assuming that K,, for individual contaminants in mixtures approaches that derived in a two-component
system and that all contaminants dissolved in water act according to Henry's Law in multicomponent systems. Therefore,
C - H C
°° - °-
(9)
where:
C.
H
concentration of chemicals in the soil, g/g;
Henry's law constant, dimensionless;
Kj = distribution coefficient, eaf/g, and
Cg = concentration of chemicals in the soil pore spaces, g/cm3.
Using the above parameters, Equation 6 can be solved to estimate concentrations of chemicals in soil, and emission rates
into air for any specific boundary condition.
The initial and boundary conditions used in the current assessment are the same as those used by Hwang et aLu
where,
Initial Concentration: Q- = (H/Kd)C»
Boundary Condition 1: Co
Boundary Condition 2: Q,
@ t = 0, x > 0;
@ t > 0, x = =; and
@ t > 0, x = 0.
The solution to Equation 6 using the stated conditions is (e.g. Jury et aL7):
, fc <> • •£ ^ (l - i ,
-|exp|-
(10)
where erfc (n) is the complementary error function of the argument n, v = - v = — —, and Da = —- * —; and all
u 3x Ra RL
other parameters are as defined above.
Under the stated boundary conditions, Equation ID can be used to estimate the instantaneous mass flux of chemical
vapors mfflrrating the basement, J, (0; t):
(11)
where erf (n) is the error function of the argument a
Indoor Air Concentrations
If a bunding ventilation rate is assumed, the flux calculated using Equation 11 can be used to estimate an indoor air
concentration by applying the following relationship:
where:
= chemical concentration in indoor air, g/cm
J( = chemical mass flux, g/cm2-s
44
-------
A^t area of cracks in building foundation, on2
Q building ventilation rate, cm3/s.
The Modified Peck* Number
A dimensional analysis is necessary to determine the dominant mechanism of soil gas infiltration into buildings.
Equation 6 can be nondhnensionauzed using the following nondimensional variables, commonly used when evaluating inert
soil gas transport in subsurface media (Nazaroff and Sextro1; Johnson and Ettinger4):
P. - A/>/Pr; and
t. =
where Cn Lp, Lp, AP are the characteristic concentration, diffusion length scale, advection length scale, and pressure
difference, respectively. Equation 6 thus becomes:
R D
xa. Lf SP. acg. " V" (1 * ^ D? c?cc. (
^ at. LD &. a*. " t, AP,ID &*
Ignoring the two characteristic length scales (diffusion and advection pathway lengths) introduced by Johnson and
Ettinger4, the right-hand term of Equation 12 shows that the phase equilibrium characterizing subsurface contaminant
transport has an impact on whether advection or diffusion controls the contaminant intrusion into the building. The
partitioning-dependent Peclet number (Pe' ), is a modified version of the Peclet number introduced by Johnson and
Ettinger4. The modified Peclet number accounts for liquid/vapor partitioning as shown below:
Substituting Equations 3,4,7, and 8 into Equation 13, the modified Peclet number becomes
Pe' =
1 * « I 6,
(14)
Du is, in general, four orders of magnitude less than D,,, and Sm is typicaUy in the same order of magnitude as 6,. For
an inert soil gas, such as radon, or many volatile organic which have high Henry's law constants, the Peclet number is
insensitive to the chemical properties of contaminants. Soil gas transport is thus dominated by the pressure-difference-driven
advection in subsurface material that is relatively permeable. However, for hydrophilic chemicals having low Henry's law
constants, such as atrazme, bromacil, and 2,4 D, the advection mechanism is hkery to be superseded by the combined vapor
and liquid diffusion mechanism. Equation 11 indicates that chemical mass flux decreases with a decreasing Henry's law
constant The modified Peclet number (Equation 14), however, provides additional insight into the understanding of the
subsurface contaminant transport. If the modified Peclet number is significantly less than 1, the diffusion mechanism
governs chemical transport. If the modified Peclet number is greater than 1, the advection mechanism governs chemical
transport.
The Indoor Outdoor Pressure Differential
The APr term in the above equations represents the pressure difference between indoor and outdoor environments.
Wadden and Scheff3 present a relationship to calculate the pressure differential due to wind and temperature effects using
45
-------
the local annual average wind speed and temperature. This relationship has also been adopted by the American Society of
Heating, Refrigerating, and Air-Conditioning Engineering14 and is presented below:
(IS)
where:
APr pressure difference between indoors and outdoors, Pa;
P, static pressure over a building, Pa;
APC pressure difference due to a thermal gradient, Pa;
A, area of a building on the windward side, m2;
AL area of a building on the leeward side, m2; and
n empirical exponent, 0.65 (Wadden and ScheEP).
The static pressure over a building can be described by:
P, = 0.6008U2
where u is the wind velocity in m/s.
The pressure difference due to a thermal gradient is given by:
(17)
where:
P = atmospheric pressure, Pa;
h distance from neutral pressure level to the point of interest, m;
T0 = outdoor temperature0, K; and
T, = indoor temperature0, K.
SCREENING-LEVEL METHODOLOGY
If project constraints do not allow for the solution of Equation 11, a screening-level methodology can be applied in
order to obtain order-of-magnitude estimates of indoor air concentrations. The screening-level methodology invokes the
following procedure:
Step 1: Estimate the modified Peclet number using Equation 14;
Step 2: If the modified Peclet number is significantly less than 1, use Equation 20 (presented in the following
section) to calculate the indoor/outdoor air concentration ratio due to diffusive mechanisms;
Step 3: If the modified Peclet number is greater than 1, use Equation 21 (presented in the following section)
to r»lm\atr. the indoor/outdoor ah- concentration ratio due to advective mechanisms; and
Step 4: Estimate the indoor air concentration.
Indoor-Outdoor Pressure Differential
In order to carry out screening-level calculations, it is necessary to determine soil and chemical properties and estimate
the indoor-outdoor pressure differential Once this information is determined, the modified Peclet number can be
calculated according to Equation 14.
The Attenuation Coefficient
The modified Peclet number indicates the driving transport mechanism, and determines the appropriate attenuation
coefficient for approximating indoor air concentrations. The attenuation coefficient (a) represents the ratio of air
concentration indoors to the air concentration in soB pores (Co). Assuming the indoor air concentration (Cn-ow) remains
constant under a steady building ventilation rate (Q,») and a fixed crack space in the building substructure (Aa—), the
attenuation coefficient can be obtained from Equation 1L
46
-------
te],
UJ
(18)
v.Dn
exp
where:
t = time, s
AB,^ = area of cracks in building foundation, cm2
Qra = building ventilation rate, cm3/s
If the soil gas transport is controlled by diffusion, Equation 19 becomes,
(20)
where od is the attenuation coefficient for diffusion-dominated transport scenarios, and all other parameters are defined
above.
If the soil gas transport is controlled by advection, the attenuation coefficient can be obtained from a steady-state
solution of Equation 19, also discussed by Little et aL5 This coefficient is expressed as:
(21)
where a. is the attenuation coefficient for advecbon-dommated transport scenarios, and all other parameters are defined
above.
Application
Equations 10 and 11 provide simple analytical solutions to the chemical vapor transport into the buildings. These
solutions are conservative because they assume that the medium is uniformly contaminated. This conservativeness is
appropriate where the level of contamination beneath a building foundation is poorly understood. Moreover, the medium is
likely to be uniformly contaminated because the seasonal variation of the ground water table, whether the ground water is
contaminated or not, ultimately provides a mixing mechanism hi the subsurface medium. For a rapid assessment or
screening analysis, the modified Peclet number can be used in conjunction with Equations 20 or 21 to estimate indoor air
concentrations.
If ground water is the source of contamination and the vadose zone is considered relatively uncontaminated, Equation
6 can be used in conjunction with different initial and boundary conditions. Solutions to these conditions are available in the
literature (e.&, Carslaw and Jaeger15, Crank", van Genuchten and Alves17, Jury et at7*). For a rapid assessment or screening
analysis, the dominant mechanism from contaminant transport should be first determined by examining the modified Peclet
number. Solutions to a simple steady-state HiBhann equation for diffusion-controlled transport (see Crank16 or Carslaw and
Jaeger13) and Equation 17 for advection controlled transport can be used to yield reasonable estimates of indoor air
concentrations.
EXAMPLE CALCULATIONS
Example screening-level calculations are earned out for soil contaminated with several chemicals including atrazine,
benzene, bromacu, chloroform, dieldrin, Hndane and 2,4 D. The medium is assumed to be a fine sand. Where possible,
chemical and soil properties are cited from Johnson and Ettingef4 and Jury et al7 to facilitate comparisons with literature
values. Screening-level analysis procedures and results are discussed below.
Indoor-Outdoor Pressure Differential
Equations 15, 16 and 17 are used to determine the indoor-outdoor pressure differential. Table I presents assumptions
made in applying Equations 15, 16 and 17, as well as pressure differential results. The pressure differential is estimated to
be apprommatery 10 Pa, or 100 g/on-s2. This differential reflects a 5.5 m/s wind speed and indoor and outdoor
temperatures, of 75°F and 49°F, respectively. The h term in Equation 17 is assumed to have a value of 3 m. This value
represents the distance from the neutral pressure level to the point of interest
47
-------
Pedet Number
Table H presents some of the parameters in determining the modified Pedet number. The soil properties presented
are for a fine sand. Table ffl summarizes Pedet number calculations as presented in Equations 3, 4, 7, 8 and 14. Table HI
also compares the modified Pedet number to the conventional Pedet number as presented in Johnson and Ettinger.' The
conventional Pedet number presented in Johnson and Ettinger* does not account for phase equilibrium and adsorption. It
is defined as:
The modified Pedet number, Pe' , is essentially equal to the conventional Pedet number for relatively hydrophobic
chemicals such as benzene and chloroform. Pe' and Pe values for these chemicals indicate that advection governs
contaminant transport However, the modified Pedet number is notably less than the conventional Pedet number for
hydrophilic chemicals. Pe' , for example, is between 60 and 2600 times less than Pe for atrazine, bromacil and 2,4 D. The
modified Pedet number for these chemicals indicates that diffusion is the governing transport mechanism, whereas the
conventional Pedet number incorrectly suggests that diffusion governs transport The exact-solution attenuation coefficients
discussed in the following section further illustrate the validity of the modified Pedet number.
Attenuation Coefficients
Comparisons between screening-level and exact-solution attenuation coefficients reflect limitations associated with the
screening-level methodology. Table IV presents Pedet numbers and attenuation coefficients for long-term (30-year) and
short-term (1-year) emission scenarios. The attenuation coefficients in columns 4, 5, 7 and 8 are calculated using the
screening-level methodology (Equations 20 and 21). Attenuation coefficients in columns 6 and 9 are calculated using based
on the exact solution (Equation 19).
In cases where the modified Pedet number indicates diffusion-dominated transport (Pe' « 1), the screening-level
diffusion attenuation coefficients dosery approximate the exact-solution attenuation coefficients for both long-term and short-
term emission scenarios. Table IV shows that this is the case for atrazine, bromacil and 2,4 D. The agreement between
exact-solution and diffusion-dominated screening level attenuation coefficients for these chemicals reflects the similarities
between Equations 19 and 20. Equation 20 is obtained from Equation 19 by assuming a negligible indoor-outdoor pressure
differential, and both equations reflect unsteady-state conditions.
It is interesting to note that the conventional Pedet numbers for atrazine, bromacil and 2,4 D incorrectly indicate that
advection is the dominant transport mechanism and result in underestimates of attenuation coefficients (using Equation 21
instead of Equation 20) by factors ranging from 2 to 20. Equation 21 represents a steady-state vapor emission rate where
(1) a constant indoor-outdoor pressure differential exists; (2) the vapor-phase contaminant concentration in soil pores
remains constant; and (3) contaminant depletion via volatilization losses or partitioning in the subsurface medium does not
occur. Equation 21 can be the ideal asymptotic solution of Equation 19 if the pressure differential remains large.
In cases where the modified Pedet number indicates advection-dominated transport (Pe' » 1), the screening-level
advection attenuation coefficient, in general, agrees well with the exact-solution result However, estimates obtained from
the screening-level advection equation could underestimate short-term attenuation coefficients. Thus underestimation is due
to the nature of unsteady-state emissions. For chemicals with strong absorption onto soil and moderate partitioning into
water (e.g, dieldrin, lindane) the attenuation coefficient is governed by diffusion initially and by advection in the long term.
As seen in Table IV, although advection is predicted to control lindane transport, unsteady-state diffusion actually controls
the short-term emission and eventually advection becomes the governing mechanism.
CONCLUSION
This paper reexammes the subsurface transport equation by incorporating phase equilibrium and chemical adsorption
behavior, which is commonly considered in contaminant leaching and volatilization models. With a simple one-dimensional
model, the indoor air concentrations can be estimated using the solution of an advective-dispersive transport equation. For
a rapid assessment or screening analysis, the transport equation can be further simplified to an unsteady-state diffusion
equation or a steady-state advection equation by evaluating the modified Pedet number. It is recommended that the
modified Pedet number should be used instead of the conventional Pedet number.
Example calculations facilitate comparisons between the conventional and modified Pedet numbers, and the screening-
level and exact-solution methodologies. Exact-solution results verify that the modified Pedet number more accurately
reflects contaminant transport for hydrophilic organic chemicals such as atrazine, bromacil and 2,4 D. The use of the
modified Pedet number and screening-level attenuation coefficients for relativery hydrophilic chemicals with high soil
sorption coefficients requires further study.
Furthermore, this paper presents an analysis of one subsurface media and one set of boundary conditions. Studies that
investigate variations in these parameters are necessary before the impact of subsurface contamination on indoor air quality
can be more fuuy understood.
48
-------
REFERENCES
L W.W. Nazarofi, SH. Lewis, SJvI. Doyle, BA. Moed and AV. Nero. "Experiments on pollutant transport from soil
residential basements, by pressure-driven airflow,' Environ. So. Techno! 21(5): 459 (1987).
2. W.W. Nazaroff and R.G. Sextro, Technique for measuring the indoor Rn source potential of soil,' Environ. So.
Technol 23: 451 (1989).
3. CO. Loureiro, LM. Abriola, JJj. Martin and R.G. Sextro, Three-dimensional simulation of Radon transport into
houses with basements under constant negative pressure,' Environ. Sci- Technnl 24(9): 1338 (1990).
4. P.C. Johnson and RA. Ettinger, "Heuristic model for predicting the intrusion rate of contaminant vapors into
buildings," I, Fnvjmn Qnal 25: 1445 (1991).
5. J.C Little, J.M. Daisey and W.W. Nazaroff, Transport of subsurface contaminants into buildings," Environ. So.
Technol 26(11): 2058 (1992).
6. B J. McCoy and DJL Rolston, "Convective transport of gases in soil porous media: Effect of absorption, adsorption,
and diffusion in soil aggregates," Environ. S^i Tpjhnnl 26(12): 2468 (1992).
7. W A. Jury, R. Graver, W F. Spencer and W J. Farmer, "Modeling vapor losses of soil-incorporated Triallate,' Soil So.
SOC.AM.J. 44(3): 445 (1980).
8. WA. Jury, WP. Spencer and WJ. Farmer, "Behavior assessment model for trace organics in soil L Model
description," J. Hnvjmn Qnal 12(4): 558 (1983).
9. WA. Jury, WJ. Fanner and WJ. Spencer, "Behavior assessment model for trace organics in soil: n. Chemical
classification and parameter sensitivity," Environ. ScL Technol 13(4): 567 (1984).
10. WA. Jury, WJ1. Spencer and WJ. Farmer, "Behavior assessment model for trace organics in soil: m. Application of
screening model," J, Rmmrm, Qnql., 13(4): 573 (1984).
1L VS. Environmental Protection Agency (USEPA). Office of Remedial Response. Superfund Fjqpnsnre A^p.^mRnr
Manual Washington, D.C, EPA/540/1-88/001, 1988.
12. S.T. Hwang, J.W. Falco and CH. Nauman, Development of Advisory Levels for Potv-chlorinafed Biphenvl (PCB)
Cleanup. J3PA/600/fr86-002, US. Environmental Protection Agency, Washington, D.C, 1986, pp. A1-A19.
13. RA. Wadden and PA. SchetT. Indoor Air Pollution. New York. John WHev & Sons. 1983.
14. American Society of Heating. Refrigerating, and Air-Conditioning Engineering (ASHRAEX ASHRAE Handbook:
1981 FundamCTitaH New York, 198L
15. US. Carslaw and J.C Jaeger, Conduction of Heat in Solids. 2nd ed, Oxford Science Publications, New York, 1959.
16. J. Crank, Mafrh«"»qtics of T3iffiision. Clarendon Press, Oxford, F.ngfonH^ 1956.
17. M, Th. van Geonchten and WJ. Alves, Analytical Solutions of the One-Dimensional Convective-Pispcrsive Solute
Transport Pjjiiarinn Technical Bulletin No. 1661, U.S. Department of Agriculture, Riverside, 1982, pp. 8-56.
18. D.T. Grismund, MJL Sonderegger and R.C. Sherman, "A framework of a construction quality survey for air leakage in
residential building in Procedures of Thermal Performance of External Envelopes of Buildings," ASHRAE. 422-452 (1983).
19. GH. Zapalac, "A time-dependent method for characterizing the diffusion of Rn in concrete," Health Physics 45(2):
377(1983).
20. US. Environmental Protection Agency (USEPA), Hazardous Waste Treatment Storage and Disposal Facilities Air
Emission Models Documentation, PB88-198619, EPA-480/347-026, December 1987.
49
-------
TABLE I
Summary of Wind and Temperature-Induced Pressure
Differential Calculations
Parameter
Wind Speed
Atmospheric Pressure
Distance from Neutral Pressure Level to Point of Interest
Outdoor Temperature
Indoor Temperature
Static Pressure over a Building
Pressure Difference due to Thermal Gradient
Ratio of Building Side Areas
Empirical Exponent
Indoor/Outdoor Pressure Difference
Symbol
V
P
h
TO
T,
P.
APC
A./A,
n
AP,
Value
5.5
1.013 x Vf
3
282
297
18
1.9
1.0
0.65
10
Units
m/s
Pa
m
K
K
Pa
Pa
dimensionless
-
Pa
TABLE II
Summary of Parameters Used in Modified Peclet Number and
Attenuation Coefficient Calculations
Parameter
Soil Permeability to Vapor
Indoor/Outdoor Pressure Difference
Gas Viscosity
Water-Filled Soil Porosity
Air-Filled Soil Porosity
Liquid Diffusivity
Soil Dry Bulk Density
Fraction Organic Carbon in Soil
Crack Area
Time Period
Ventilation Rate
Characteristic Advection Length Scale
Symbol
t,
AP,
M
«»
0,
DU
/>»
FCC
A^
t
Q
Lp
Value*
10x10*
10"
1.8 x W4
0.12
026
1x10^
1.7
0.002
1,000 c
95x10"
3.4 x W
1,000
Units
cm2
p.
g/cm/s
dimensionless
dimensionless
cm2/s
g/cm3
dimensionless
cm2
s
cm'/s
cm
Notes:
" From Johnson and Ettinger* unless otherwise noted.
i Calculated as shown in Table I.
c Based on Wadden and Scheff."
50
-------
Chemical
Atrazine
Benzene
Bromadl
Chlorofonn
Dieldrin
Lindane
2,4 D
TABLE III
Summary of Modified Peclet Number Calculations
Henry's
La»
Constant
)
5.00-02"
Ue-02c
5.7e-02B
l.le-01D
4.4e-02B
5.0e-OZB
S.Oe-02"
Liquid
DWUiMly
V
(an*/s)
l.Oe-05
1.0e-OS
1.0e-OS
l.Oc-05
l.Oe-05
l.Oe-05
l.Oe-05
Effeclln
AlrDUhilan
Coefllctonl
(cn//s)
3.9e-03
6.9C-03
4.4e-03
8.6e-03
3.4e-03
3.9e-03
3.9C-03
Effeetln
Uqidd
DWuslon
Coefficient
(OB^/S)
5.9e-08
5.9C-08
5.9e-08
,_ 5.9e^8
5.9e-08
S.9e-OB
5.9e-OB
Vapor
Plus.
PaitMonlng
Coeffldenl
"o
(dbnemlonbu)
i7e+06
Z.le+00
9.9e+06
lie +00
6.1e+04
3.4C+04
3.4e+07
Liquid
Plus.
PartWonliuj
Coefficient
"L
(dlmenslonless)
6.6C-01
4.0C-01
3.te-01
2-5C-01
4.1C + 01
4^e-fOO
1.9e-01
Convmtloul
PMltl Number
(dbn.ulonl«9)
1.4e+01
S.lctOO
1.3C+01
6Je+00
1.6e+01
1.4C+01
1.4C+01
ModlBed
Peclet
Number
(dtoenJonku)
2Jc-01
8.1.+00
3.4.-02
6Je+00
1.6C+01
Ue+01
5^e-03
Notes:
A Juty10
" Estimated Value
c Johnson and Bttuujer*
D VSEtA"
Alrazine
Benzene
Btomacil
Chlorofonn
Dieldrin
Lindane
2,4 D
Com
Conventional
Peclel Number
Pe
(dlmensloiiless)
1.4e+01
8-lc+OO
13e+01
6-Se+OO
16c + 01
1.4C + 01
1.4e+01
Notes:
A Based on a crack area of 1000 cm2,13 a
B Based on a crack area of 1000 cm2,13 a
Modified
Peclel Number
Pe'
(dmunslonles!)
23C-01
8.1e+00
3.4e-02
6Je+00
1.6e+01
IJe+Ol
SJe-03
TABLE IV
parison of Screening-Level and Exact-Solution Attenuation Coefficients
Screening-Level
Unsteady-Stale
Diffusion Attenuation
Coefficient''
@ T = 30 yre
(Equation 20)
(dunenslonless)
7.0e-06
63e-08
5.9e-06
7Je-08
7.7e-06
].9e-06
3.7^06
Screening-Level Steady-
Slate
Advection Attenuation
Coefficient"
@ T = 30 yrs
(Equation 21)
(dimenslonless)
1.6C-06
1.6C-06
1.6e-06
1.6e^M
1.6C-06
1.6e-0o
l.oe-06
Eurt Solution
Unsteady-Stale
Dlfluslon&Adveetlon
Attenuation Coefficient
@ T = 30 yrs
(Equation 19)
(dlmenslonless)
7.0e-06
l.fc-06
5.9e-06
1.6C-06
8Jc-06
6.7e-06
3.7C-06
Screening-Level
Unsteady-Slale
Dlfnjslon Attenuation
Coefficient
@T=lyr
(Equation 20)
(dlmenslonless)
3.8e-05
3/«-07
3.2C-05
4.0eJT7
4Je-05
3.2e-OS
2.0e-05
Screening-Level
Steady-State
Advection Attenuation
Coefficient
@T= lyr
(Equation 21)
(dunenslonlesi)
1.6e-06
1.6C-06
1.6e-06
1.6e-06
1.6e-06
1.6e-06
l.oe-06
ventilation rate of 3.4 cm'/s and a time period of 30 years.
ventilation rate of 3.4 cmys and a characteristic advection length scale of 1000 cm.
Eucl Solution
Umtemdy-Slale
Diffusion & AdvecUon
AHenuarlon
Coefficient
@T=lyn
(Equation 19)
(dlmenilomess)
3£e-OS
1.6. -06
3Je-05
1.6e-06
43e-05
3.3e-05
2.0.-05
-------
PRELIMINARY INVESTIGATION OF UNCOMBUSTED AUTO FUEL
VAPOR DISPERSION WITHIN A RESIDENTIAL GARAGE MICROENVIRONMENT
by
Azzedine Lansari1, John J. Streicher"2, Alan H. Huber**,
Gennaro H. Crescenti"2, Roy B. Zweidinger3, John W. Duncan1
1) ManTech Environmental Technology Inc.,
P.O. Box 12313, RTP, NC 27709
2) Atmospheric Sciences Modeling Division,
Air Resources Lab., NOAA; RTP, NC 27711
3) Mobile Source Emissions Research Branch,
AREAL,US EPA, RTP, NC 27711
ABSTRACT
Evaporative emissions from vehicles in an attached garage may represent a significant source of indoor
pollution and human exposure. A pilot field study was undertaken to investigate potential in-house dispersion of
evaporative emissions of uncombusted fuels from a vehicle parked inside an attached garage. In a set of experiments
using sulfur hexafluoride (SF6) tracer gas, the multizonal mass balance model, CONTAM88, was used to predict
interzonal air flow rates and SF« concentration distributions within the garage and house. Several experiments were
included to evaluate the effect of meteorology and mechanical mixing mechanisms on the dispersion of automobile
fuel vapor. Measurements indicated that approximately three percent of the garage maximum concentration was
measured in a room adjacent to the garage. The model successfully predicted garage concentrations under well
mixed conditions, but underpredicted the measured concentrations within various rooms of the house, in which
mixing was incomplete. Multizonal mass balance models such as CONTAM88 may be useful in approximating
contaminant concentrations at various locations within the house.
INTRODUCTION
The introduction of oxygenated auto fuels and fuel additives (alcohols and ethers) into the U.S. motor vehicle
fleet has served to reduce tailpipe emissions of carbon monoxide and total hydrocarbons'". Tailpipe emissions
represent an obvious source of ecological pollution, however evaporative emissions from vehicles in attached garages
may represent an important source of indoor (i.e. microenvironmentat) pollution.
In-house and attached garage concentration of evaporated (uncombusted) fuel species from an automobile's
fuel system may represent a significant component of total human exposure to these chemicals. The use of alcohol
and ether additives increases the fuel vapor pressure, and hence the evaporation rate01. The magnitude of in-house
concentration of a chemical species depends upon the emission rate of evaporating fuel, the concentration of the
component species within the liquid fuel, and the air flow rates between garage and house. Measurement of these
critical variables enables the development of predictive models useful in population exposure assessment.
This study examines the potential for dispersion of evaporative emissions from an auto fuel system into a
residence from an attached garage. A series of field experiments were conducted to obtain estimates of in-house
ambient concentration of fuel vapor components resulting from normal automobile use scenarios. A single family
house with attached garage was selected as the test site. Air flow rates between the garage and various zones within
the house were measured using SFt tracer gas. A multizonal mass balance model was used to predict the spatial
and temporal contaminant dispersion within the house. Previous investigation has shown that multizonal mass
balance models may be useful in designing field study monitoring strategies01. Modeling results were compared to
SF6 measurements in order to investigate the possibility of using the model to predict methanol and/or other
* On assignment to the Atmospheric Research and Exposure Assessment Laboratory, U.S.
Environmental Protection Agency
52
-------
alternative fuel dispersion in the garage and throughout the house. Methanol emissions and concentration
distributions were examined in a later phase of this pilot study. Analyses of methanol concentration and supporting
modeling results will be presented in subsequent report.
MATERIALS AND METHODS
RESIDENCE DESCRIPTION
During the summer of 1992, a house in Raleigh, North Carolina was chosen to study in-house dispersion of
evaporative emissions of uncombusted automobile fuels from a vehicle parked inside an attached garage. This study
determined quantitative flow rates between the garage (pollutant source) and selected rooms of the house. The first
story of this two story house consisted of the master bedroom, a bathroom, a den, the family room, the kitchen and
the dining room (Fig. 1); the second story of the house includes three bedrooms and a bathroom. One second story
bedroom was located directly above the garage. The total volume of the garage was 95 m3. The house physical
characteristics were measured during the first day of the study. These measurements were subsequently used to
model SF6 concentration distribution in the garage and within the house. The SF6 dispersion analysis was used to
calculate interzonal air flows and to calculate contaminant concentrations in order to determine the rooms with
significantly different concentrations.
MASS BALANCE MODEL
The multizonal mass balance model used in this investigation is the National Institute of Standards and
Technology (NIST) model, NBSAVIS/CONTAM88W developed for the Environmental Protection Agency to
simulate transient contaminant concentration distribution in buildings. The model is based on the element assembly
approach, which assumes that a building can be represented as a combination of well-mixed zones linked by flow
and kinetic elements (contaminant mass transport and decay). CONTAM88 solves a set of mass balance and flow
equations. The mathematical formulation of the contaminant concentration is:
[W]C+[M]ljj G (1)
where:
C = Vector containing the discrete concentration values
[W] = System mass transport matrix which contains flow rate data
[M] = System matrix which contains mass (volume) data
G = System generation vector containing kinetics data.
NBSAVIS is a preprocessor to CONTAM88 which allows the idealization of a building through the generation of
a file that describes the building configuration, including indoor and outdoor contaminant sources. Data input to
NBSAVIS is facilitated by a series of data entry screens that allow the user to specify: interior and exterior wall
types; interior and exterior doors, windows, open passageways; filters and fans; room descriptions; and heating,
ventilation, and air conditioning (HVAC) system descriptions. NBSAVIS then calculates the interzonal air flow rates
and system matrices. NBSAVIS was used to build an idealization of the house. The parameters that were measured
in order to run NBSAVIS are:
. The house physical dimensions (including all windows, doors and other openings).
. The house HVAC system output as well as all the locations of the vents with air flow rates.
. The contaminant source information (name, molecular weight, emission rate).
. The location of the source (inside the garage).
. The temperatures of the various rooms in the house.
The outdoor meteorological conditions (temperature, wind speed and wind direction).
53
-------
EXPERIMENTAL DESIGN
Three experiments in which SFS was released in the garage and traced inside the garage and the house were
designed and implemented. SF, was chosen because it is non-toxic, stable with respect to chemical reactions; its
removal due to deposition is negligible and it can be measured in very low concentrations with high accuracy'31.
A diluted mixture of SFe in nitrogen (0.362g SF6 / m3 ) was prepared and used in all experiments. Samples were
analyzed using gas chromatography/electron capture detection (GC/ECD), with a lower detection limit of 20 parts
per trillion volume (ppt). Measurement accuracy was within 5 %. Measurement precision (standard deviation) was
within 5% of the measured concentration. During the entire study the outdoor temperature ranged between 19 C
and 23 C, the wind speed did not exceed 0.5 m/s and the wind direction varied. The low wind speeds were mainly
due to shielding of the house and property by trees.
The first experiment was designed to assess the percentage of contaminant that infiltrates in the house from
the garage. A car was parked in the garage with the front of the car facing an interior wall of the garage. In this
experiment five automated sequential syringe samplers™ were used. Three samplers were placed in the house: one
in the kitchen, one in the den and one in the bedroom above the garage. The remaining two samplers were located
in the garage. One sampler was placed on the car roof, one was placed on the garage floor, next to the car. The
garage door was closed during the experiment. SF6 was released for 20 min at a rate of 1 liter per minute (1/min).
The total mass of SF(l released was 7.23 mg. The SF6 source was located next to the gasoline tank on the passenger
side of the car. A large box fan was used next to the source to create a well mixed condition. Samplers in the
garage were started simultaneously with the SF6 source. At the end of the SF6 release (20 min), the three samplers
inside the house were activated.
The second experiment was designed to investigate how quickly the concentration in the garage drops after
the garage door is opened. In this experiment no car was in the garage. Air samples were taken at two vertical
locations in the garage. The upper sample location was centered 2 feet from the garage ceiling while the lower
sample location was centered 2 feet above the floor. The samples were analyzed in real time. SF6 was released
in the garage for 10 min at a rate of one 1/min for a total mass released of 3.62 mg. The box fan was used to create
a well mixed condition. At the end of the 10 min release time, the fan was turned off and the garage door was
opened.
Experiment three was performed to test the well mixed garage assumption. Five syringe samplers were
used in the garage, one on the car roof, one next to the driver's side, one next to the passenger side, one in front
of the car (back wall location), and one in the back of the car (next to the garage door). No fan was used, and
sampling time was set to 12 min intervals for all the samplers. SF6 was released for 20 min at a rate of 1 1/min.
The SF6 source was located next to the gas tank on the passenger side of the car. The garage door was closed
during the experiment.
RESULTS AND DISCUSSIONS
During the first experiment, CONTAM88 was used to simulate the dispersion of SF6 in the garage and
throughout the house. Good agreement was found between simulated and measured data (Fig. 2). Differences may
be due to a lack of complete mixing in the garage. There were two samplers in the garage; the one on the passenger
side (labeled LEFTGARAGE in Fig. 2) showed a sharp drop of concentration approximately 30 min after the
experiment started. This sampler was located close to the kitchen door, therefore the drop in concentration may
be due to local air exchange. The other sampler (labeled TOPGARAGE in Fig. 2) agrees better with the model,
which assumes a well mixed condition.
Modeling results consistently underpredicted the concentration in all the rooms in the house (Fig. 3). The
highest concentration was measured and modeled in the kitchen; followed by the den and then the bedroom above
the garage. Overall, the model predicted a behavior similar to the data. However, the model underpredicted the
kitchen and bedroom concentrations by approximately 30%, and the den concentrations by approximately 10%.
The model assumption of well mixed zones was not satisfied in the house because of the absence of forced mixing.
The model predicted that the maximum concentration attained in the kitchen would be 2% of the maximum
concentration in the garage, while measurements showed that the maximum relative concentration was approximately
3%.
Following the completion of SF, injection, the concentration in the garage began to decrease, while the
54
-------
concentrations in the kitchen and the other rooms continued to rise. The time delay (phase lag) between the modeled
maximum concentration in the garage and the kitchen was approximately 15 minutes; between the family room and
the garage the time delay was approximately 30 minutes. In the bedroom the concentration was still rising 50 min
after emission stopped (Fig. 3). The measured phase lag in concentrations between the various rooms of the house
is also predicted by the model (Eq. 1) Conservation of mass (SFg) - within the context of a box model - requires
that a loss of mass from one box (e.g. the garage) be accounted for in a net increase of mass in the other boxes (e.g.
other rooms of the house; the outdoors).
In the second experiment, CONTAM88 was used to simulate contaminant build up and decay after the
garage door was open. Modeling results were in agreement with the measurements (Fig. 4). After the garage door
was opened the SF6 concentration decreased to less than 10% of the initial concentration in 3 minutes. Similar
results were found using the decay from the modeling results. This decay rate corresponds to approximately 45 air
changes per hour (ACH). During testing the wind speed was less than 0.2 m/s, therefore this high exchange rate
is likely due to mechanical mixing that occurred when the garage door was opened. These results were supported
by observations during smoke release experiments.
During the third experiment measurements were performed to test the well mixed box assumption. Model
simulation was used to determine locations in the garage where the well mixed condition existed. Simulation results
agreed with the data collected at the car roof, back of the garage, and the passenger side locations. The data
collected near the garage door and driver side locations were approximately half of the model predictions. Total
mixing occurred between 60 and 90 min after the SF6 release stopped (Fig. 5). These results show that sampler
locations cannot be assumed to measure average (well mixed) concentrations, particularly during the period of
contaminant emission. However, for constant sources, a steady state regime may develop in the garage and house,
resulting in quasi-mixed conditions.
CONCLUSIONS
Good agreement was found between modeling and experimental results in the garage when the well mixed
assumption was verified (box fan on). In the rooms adjacent to the garage, the model underpredicted the
concentration of SF6. The model did help assess the broad trend of concentration distribution in those rooms.
Approximately three percent of the maximum garage concentration was measured in the kitchen and 1,5 percent in
the upstairs bedroom. These experiments were performed with the door between the kitchen and garage always
closed. Higher in-house relative concentrations can be expected when this connecting door is opened.
When mixing was not forced (box fan off) within the garage, the well mixed assumption was not valid at
locations next to the garage door and the kitchen door, but the other locations in the garage showed more thorough
mixing. It took 60 to 90 min for total mixing to occur after the release of SF6 stopped. Furthermore, the well
mixed assumption did not hold during the contaminant release time.
Preliminary results showed that multizonal mass balance models such as CONTAM88 may be used to
approximate contaminant concentrations within various locations of the house provided sampling time and location
are chosen judiciously. These models may help identify the locations where mixing occurs and the time duration
to attain well mixed conditions. The models may help choose sampler locations and sampling time during a field
study. Locations of expected high concentration gradients (near locations of local air exchange, such as doors and
windows) must be sampled more intensively than locations of low concentration gradients (remote from locations
of local air exchange). Furthermore, CONTAM88 can be used to successfully model the dispersion of uncombusted
fuel vapor inside a residence and its attached garage. For pollutant emissions in an attached garage, sampling
locations should be next to the garage door and any connecting (interior) doors and windows; one sampler in another
location in the garage is sufficient (car roof or back of garage, etc). Sampling time will be dependent on the
contaminant source release period and the time the source takes to reach steady state.
The potential exists for in-house exposure to emissions (uncombusted fuel vapors and exhaust) from a
vehicle in an attached garage. Therefore, the introduction of alternative fuels and fuel components into the U.S.
automobile fleet will result in residential exposure to these chemicals and their combustion by-products. An
exposure assessment project is ongoing to quantify potential human exposures. The models evaluated here are being
incorporated into this project.
55
-------
DISCLAIMER
This paper has been reviewed in accordance with the U.S EPA's peer review and administrative review
policies and approved for presentation and publication. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
daii Balance Modeling of Benzene Piiperiion in • Private
REFERENCES
(1) BUck.F.M., Control ofMolor Vehicle Emiuiom-The U.S. Experience: Critical Reviewa in Environmental Control, 2I(5,6):373-410 (1991).
(2) Stump, Knapp, and Ray; Seasonal Impact of Blending Oxygenated Organica with Gamline on Moior Vehicle Tailpipe and Evaporative
Emiaaiona: I. Air and Wane Management, 40:872-880 (1990).
(3) Lanaari, A., A. B. Lindatrom, B. D. Templeman, and 1. S. Ifwin. Mull
Reiidence. A4WMA/EPA Inunulionil Sympouum (1992).
(4) CM. R.A. Uaer'a Minual NBSAVIS/CONTAM88. A Ueer Interface for Air Movement and Contaminant Diaperaal Analyaii io Multizone
Building!. National butituu of Sundinli and Technology, Giithenburg, MD. (1991).
(S) Brown, R.M.R. N.DicB and E. A. Core. The Uie of SF6 in Atmogheric Tuiupon ind Difluiion Studiei. I. Geoohvi. Rei 80. 3393-
3398. (1975).
(6) Kraanec, I. P.,D. E. Demany, B. D. lamb, and R. Benner. An Amounted Seauenlial Syringe Samnler for Almomheric Trace Study. Am.
Met. Soc., 1, 372-378. (1984).
-7I-4--
--o-
Froni Cor^h
Floor Plan - 1 st Eoor
1. Residence floor plan. A second
floor bedroom (not shown) was located directly
above the garage.
56
-------
a. 14
I 12
EXPERIMENT 1, 20 min RELEASE
O = 6.03 ug/s
0 10 20 30 40 50 60 70 80
Time (min)
Figure 2. Simulated and measured garage transient concentrations during
and after a 20 min SF6 emission.
EXPERIMENT 1, 20 min RELEASE
0 = 6.03 uq/s
0.30
0.25
S*
•S 0.20
| 0.15
g 0.10
u
*
1/1 0.05
0.00
0 10 20 30
40 50 60 70
Time (min)
Figure 3. Simulated and measured (M) transient in-house concentrations
during and after a 20 min SF6 emission.
EXPERIMENT 2. 1 0 min RELEASE
SAMPLESTAKEN INSTANEOUSLYAT 2 LOCATIONS
0 = 6.03 ug/9
0 2 4 6 8 10121416
Time (min)
Figure 4. Simulated and measured SF6 decay after 10 min SF6 emission
and after the garage door is opened.
EXPERIMENT 3, 20 min RELEASE
0 = 6.03 ug/s
•
•
A
+
— —
Corroof
Backqar
Rtsidgor
Lfisidgor
Gordoor
Simulated
40 60 80 100 120 140
Time (min)
Figure 5. Simulated (well mixed) and measured transient concentrations a
various locations in the garage.
57
-------
Modelling Ozone Deposition Onto Indoor Surfaces
Richard Reiss, P. Barry Ryan, Petros Koutrakis, Sarah Bamford
Harvard University School of Public Health
Department of Environmental Health, Exposure Assessment and Engineering Program
665 Huntington Avenue, Boston, MA 02115
ABSTRACT
Modelling the deposition of pollutants onto surfaces often requires the inclusion of two separate factors:
the transport and surface uptake resistances. In this paper, the surface uptake resistance for ozone onto
several common indoor surfaces was examined by measuring the mass accommodation (or "sticking")
coefficient, which equals the number of "sticks" of a pollutant to a surface divided by the number of
collisions with the surface. It was determined that the mass accommodation coefficients for ozone
deposition onto these surfaces were in the range of 10~5 to 10'6, Given this, an analysis was conducted to
determine the relative effects of the transport and surface uptake resistances. It was found that for the
surfaces tested in this study, the surface uptake resistance is rate limiting for mass accommodation
coefficients on the low-end of the range, while boundary layer resistance is limiting for those on the high
end of the range.
INTRODUCTION
Tropospheric ozone has been identified as a criteria air pollutant by the Environmental Protection
Agency, since it is known to cause acute health effects and there is evidence that it may be responsible for
more severe chronic effects, (1). An individual's exposure to a pollutant such as ozone can be
characterized as the sum of exposures in the microenvironments where the individual spends his or her
time. One delineation of these microenvironments is outdoors versus indoors. It is known that ambient
ozone infiltrates into residences, (2), resulting in indoor concentrations from about 20 to 80 percent of the
outdoor concentration, (3). Considering that individuals spend about 90 percent of their time indoors, (3),
it is clear that the indoor environment may constitute a significant ozone exposure.
One important tool in indoor air exposure assessment is the indoor air quality model. For ozone, one
of the most important variables in these models is the deposition of ozone onto indoor surfaces. This
deposition is the principal reason that indoor ozone concentrations are lower than outdoor concentrations.
In this study, a chamber has been designed to study ozone deposition onto a variety of indoor surfaces.
Also, the factors that affect ozone deposition, including room air flows and surface uptake, are analyzed.
BACKGROUND
Deposition onto indoor surfaces is usually modelled through the concept of the deposition velocity. It
is defined as the rate of decay of the pollutant divided by its mean concentration in air, (4), and can be
written as,
Kd = g (1)
where F is the flux to the surface (units of mass deposited per area per time) and C is the free-stream
concentration, which is assumed to be uniform throughout the indoor space. Use of the deposition
velocity also assumes that the loss process is first-order.
For atmospheric deposition modelling, the flux depends on the convective and diffusive flows
transporting the pollutant to the surface (e.g. ground, building surfaces, etc.) and the uptake of the
pollutant at the surface. For this reason, the deposition velocity can be written as the combination of three
58
-------
resistances: the aerodynamic resistance, ra; the boundary layer resistance, r^; and, the surface uptake
resistance, rc, as, (5):
The first resistance term accounts for the mixing of the pollutant in the core region above the boundary
layer. This term is generally independent of the pollutant. Rather it is determined by the wind speed and
turbulence. The second term corresponds to the movement of the pollutant across the boundary layer to
the surface. This resistance is related to the diffusion coefficient of the pollutant and the thickness of the
boundary layer. Finally, the surface uptake resistance corresponds to the rate of pollutant uptake at the
surface and depends on the nature of the absorbate (i.e. the pollutant) and the absorbent (i.e. the surface).
This analogy can be extended to indoor pollutant deposition with some modifications. Indoor air flows
are often assumed to follow the boundary layer assumption, in which the air away from the surface (the
core region) of an enclosure is assumed to be an ideal fluid while the air near the surface (i.e. in the
boundary layer) is assumed to be a viscous fluid. The air flow in the core region is normally circulating
along the periphery of the enclosure, essentially parallel to the wall. Thus, this air flow does not constitute
a transport mechanism to the surface and there is no aerodynamic resistance. For indoor pollutant
deposition, the important resistances are the boundary layer and surface uptake resistance. Ozone is a
highly reactive gas, and it is assumed that once it is taken up by the surface it immediately reacts. Thus,
the mass accommodation coefficient, which is defined as the number of collisions of a gas molecule and a
surface resulting in a "stick" divided by the number of collisions, will be used to examine the surface
uptake resistance. Low values of the mass accommodation coefficient indicate that the surface uptake
resistance is important The mass accommodation coefficient for several ozone-surface reactions was
determined in this study.
MATERIALS AND METHODS
Description of the Apparatus
The apparatus for the deposition experiments is a laminar flow reactor. Air with ozone (generated by a
UV Photometric Ozone Calibrator) flows through a cylindrical glass test section. Ozone is measured with
Monitor Labs Model 8410 chemiluminescent ozone analyzers before and after the test section to determine
the ozone deposition. Continuous measurements of the temperature and relative humidity in the test
section are made using temperature and relative humidity probes (Omega 41 1). The flow rate through the
test section is monitored with a calibrated rotometer. The test section has an inside diameter of 2.1 cm and
is 30 to 60 cm length. Several materials that are commonly found on indoor surfaces were tested including
latex paint, vinyl wallpaper and paper wallpaper. For latex paint the inside of the test section was simply
painted while the wallpaper was glued to the inside of the test section. At this point, only one brand of
latex paint has been tested. Ozone deposition, defined as the percent of ozone depositing on the test
section, was measured over a range of relative humidities. For a more detailed description of the apparatus
see reference (6).
Mass Accommodation Coefficient Calculation
McMurray and Stolzenburg (7) provide the model for measuring mass accommodation coefficients
from laminar tube flow. For a gas penetrating in a cylindrical tube with fully-developed laminar flow, the
appropriate steady-state convective transport equation is the following,
where C(r,z) is the concentration of the pollutant species, r is the radial distance from the center of the
tube, r0 is the tube radius, z is the axial distance, Dg is the pollutant diffusion coefficient in air and u is the
mean flow speed in the axial direction.
The term on the left side of equation (3) represents the bulk flow while the term on the right side
represents diffusive flows, both radial and axial. However, it can be shown that the axial diffusion is
59
-------
negligible. Porous diffusion into the surface was also neglected because it is assumed that ozone reacts
immediately upon being adsorbed onto the surface. The following boundary conditions apply to our
system, (7),
C(r,0) = C0 (4)
-«
where a is the mass accommodation coefficient and v is the mean thermal speed of the pollutant molecule.
Equation (4) simply states that the inlet concentration is C0 at all r. Equation (5) states that the
concentration is maximized in the radial direction at r=0 (i.e. the center of the tube) and assumes symmetry
around the tube center-line. Equation (6) states that the net diffusional flux to the surface equals the rate of
uptake, which is calculated from the kinetic theory of gases. It also assumes that the re-release of ozone
from the surface can be neglected because ozone is highly reactive. McMurray and Stolzenburg (7)
provide the solution to this system. It was used to determine the mass accommodation coefficient given an
experimentally determined deposition.
RESULTS AND DISCUSSIONS
Table 1 shows selected results from the chamber study. For a more detailed presentation of the
chamber study results see reference (6). The mass accommodation coefficients ranged from about 10* to
10~5. This indicates that ozone reacts once in every one hundred thousand to one million collisions with a
surface. Also, the relative humidity had a large effect on deposition onto latex paint For low relative
humidities, the mass accommodation coefficient was on the order of 1 x 1Q-6, while for high relative
humidities, the mass accommodation coefficient is about 5 x 10~6 to 2 x It)-5. This indicates that at high
relative humidities there is about 5 to 10 times as many "sticks" to the surface per collision than at low
relative humidities. For glass and vinyl and paper wallpaper, the relative humidity had no effect on the
deposition.
Traditional thought regarding surface wall deposition modelling envisions a pollutant from the bulk air
diffusing across the boundary layer to the wall. Modelling this system can be very difficult because of the
complexity of indoor air fluid dynamics. This approach is necessary for large particles, which have small
diffusion coefficients, and, thus, the deposition may be limited by the boundary layer transport. Also,
while small particles have larger diffusion coefficients, they normally stick upon every collision. On the
other hand, gases may stick much less readily, and the uptake process may be the rate limiting step. In
this case, the resistance to uptake at the wall is much greater than the transport resistance to the wall and
the deposition can be modeled by knowledge of only the physics of random molecular motion. This can
be done with the well known laws of the kinetic theory of gases, (e.g. (8)).
As a concrete example, consider the following residential environment: a 5m x 5m room with a 2.5m
ceiling. The side walls are covered with a latex paint All deposition will be assumed to occur on these
walls. The resistances for the two corresponding processes will be calculated. The derivation of these
resistances is provided in reference (9) and references therein.
Boundary Layer Characteristic Time
The boundary layer resistance can be calculated as follows,
5
r» = D-
-------
where S!IL is the dimensionless Sherwood number and L is the length of the surface in the direction of the
air flow. The Sherwood number for flow past a flat plate can be calculated by its relationship to two other
dimensionless numbers, the Reynolds and Schmidt numbers. The Reynolds number is a function of the
air velocity, which will change under different residence conditions. The equations for performing this
calculation are readily available in the literature, (e.g. (11)).
We are concerned with the case where the boundary layer resistance is highest. This occurs when there
is low air flow. Assuming flow parallel along the length of the walls, one can estimate the lowest possible
average air flow for a residence by dividing the infiltration rate by the cross-sectional flow area. A tight
residence may have an air exchange rate of around 0.5 hr1. For our case, the volume of the room is 62.5
m3 (5m x 5m x 2.5m). Multiplying the volume by the air exchange rate gives 31.25 m3/hr for the
infiltration rate. This number is divided by the cross-sectional flow area (12.5 m2) to give 2.5 m/hr or
0.07 cm/sec. From the method above, this translates into a boundary layer of 50 cm and a 323 sec/cm.
However, this situation is seldom, if ever, realized in actual indoor environments. There are typically
thermal gradients within residences that result in much higher free-stream velocities. Matthews et al. (12)
measured a median air velocity in six homes of 5.3 cm/sec. This translates into a boundary layer of 5.7
cm and a resistance of 37 sec/cm. Also, Matthews et al. report that the lowest median air velocity
measured in a single room was 1.1 cm/sec. This air velocity gives a boundary layer of 13 cm and a 84
sec/cm. A high end air velocity for a residence may be about 10 sec/cm, which gives a boundary layer of
4.2 cm and a resistance of 27 sec/cm.
Surface Uptake Characteristic Time
The surface uptake resistance onto a single wall in the enclosure can be calculated as follows,
rs = — (9)
va
where v is the mean thermal speed in the direction of the wall. The mean thermal speed can be determined
from the kinetic theory of gases as, (8),
••V^
where ki, is the Boltzmann constant, T is the absolute temperature and m is the mass of the pollutant
molecule. At 25°C this gives 362 m/sec for the mean thermal speed.
For a low-range relative humidity experiment with a latex paint surface, the ozone mass accommodation
coefficient was determined to be on the order of 1 x 10"6. In this case, the surface uptake resistance is 110
sec/cm. For a mid-range relative humidity experiment (i.e. 50 percent), the ozone mass accommodation
coefficient was found to be on the order of 5 x 10"6. This gives a resistance of 22 sec/cm. Similarly, for a
high-range relative humidity (i.e. 80 percent), the mass accommodation coefficient is about 2.0 x 10"5,
which gives 5.5 sec/cm for the resistance. If this was a glass room, the mass accommodation coefficient
would be about 5 x 1Q-7, which gives a resistance of 221 sec/cm.
Implications of Characteristic Time Analysis
The resistances are summarized in Table 2. The low-range relative humidity experiment has a surface
uptake resistance that is higher than all of the boundary layer resistances, except the theoretical low-end air
velocity case. However, the boundary layer resistances are not negligible for the three air flow cases.
This indicates that the surface uptake process is the major process for deposition onto low relative
humidity latex paint surfaces; however, it is not rate limiting. The boundary layer resistance must be
considered. For the mid-range relative humidity, the surface uptake process is about equally important as.
the boundary layer transport for the median and high-end case. For the high relative humidity case, the
boundary layer resistance is rate limiting. Also, for glass, the surface uptake is rate limiting for the median
and high-end air velocity cases.
61
-------
BIBLIOGRAPHY
(1) M. Lippmann, "Health Effects of Ozone: A fritir.al W(-VJP.W"T Journal of the Air & Waste
Management Association. 39(5): 672-695 (1989).
(2) R.H. Sabersky, D.A. Sinema, F.H. Shair, "Concentrations, Decay Rates, and Removal of Ozone and
Their Relation to Establishing Clean Indoor Air", Environmental Science & Technology. 7(4): 347-353
(1973).
(3) C.J. Weschler, H.C. Shields, D.V. Naik, "Indoor Ozone Exposures", Journal of the Air Pollution
Control Association. 39: 1562-1568 (1989).
(4) R.A. Wadden, P.A. Scheff, Indoor Air Pollution. John Wiley & Sons, New York, 1983, pp 114-
(5) B.C. Spiker, R.P. Hosker, VJ. Comer, J.R. White, R.W. Werre, F.L. Harmon, G.D. Gandy, S.I.
Sherwood, "Environmental Chamber for Study of the Deposition Flux of Gaseous Pollutants to Material
Surfaces", Atmospheric Environment. 26A(16): 2885-2892 (1992).
(6) P.B. Ryan, P. Koutrakis, S. Bamford, R. Reiss, "Reactive Chemistry of Ozone in Indoor
Environments: The Kinetics of Ozone Deposition on Surfaces", To be submitted to Environmental
Science & Technology
(7) P.H. McMurray, M.R. Stolzenburg, "Mass Accommodation Coefficients From Penetration
Measurements in Laminar Flow", Atmospheric Environment 21(5): 1231-1234 (1987).
(8) G.M. Barrow, Physical Chemistry. 2nd ed., McGraw-Hill Company, New York, 1966, pp 28-30.
(9) R. Reiss, P.B. Ryan, P. Koutrakis, Modelling Ozone Deposition Onto Indoor Surfaces, to be
submitted to Environmental Science & Technology.
(10) J.M. Axley, "Adsorption Modelling for Building Contaminant Dispersal Analysis", Indoor Air. 1(2):
147-171 (1991).
(11) C.O. Bennett, J.E. Myers, Momentum. Heat, and Mass Transfer. 3rd ed., McGraw-Hill Company,
New York, 1982, pp 555.
(12) T.G. Matthews, C.V. Thompson, D.L. Wilson, A.R. Hawthorne, D.T. Mage, "Air Velocities
Inside Domestic Environments: An Important Parameter for Passive Monitoring", Proceedings of the 4th
International Conference on Indoor Air Quality and Climate. 1987, pp 21-27.
ACKNOWLEDGEMENTS
The authors would like to thank the Center for Indoor Air Research for funding for this research.
Richard Reiss would like to thank the National Institute of Health for his training fellowship and would
also like to thank Dr. James Axley of the Massachusetts Institute of Technology for helpful discussions.
62
-------
Table 1. Summary of Ozone Deposition Results
Experiment
No.
1
2
3
4
5
6
7
8
9
10
Surface
Latex Paint
Latex Paint
Latex Paint
Latex Paint
Latex Paint
Vinyl Wallpaper
Vinyl Wallpaper
Paper Wallpaper
Paper Wallpaper
Paper Wallpaper
Relative
Humidity
(percent)
4
41-46
53-56
79-85
66-71
6-7
68-75
6-9
67-71
73-77
Ozone
Deposition
(percent)
4
9
20
44
25
15
21
5
4
3
Mass
Accommodation
Coefficient
9x 10-7
2.2 x 10-6
5.6 x 10-6
1.9 x 10-5
7.5 x 10-6
3.9 x 10-6
5.9 x 10-6
1.2 x 10-6
9x 10-7
7x 10-7
Table 2 - Summary of Resistances
Resistance Term
Boundary
Layer
Surface
Uptake
Theoretical Low-End
Air Velocity
Typical Low-End
Air Velocity
Typical Median
Air Velocity
Typical Median
Air Velocity
Low-Range Humidity
Latex Paint
Mid-Range Humidity
Latex Paint
High-Range Humidity
Latex Paint
Glass
Resistance
(sec/cm)
323
84
37
27
110
22
5.5
221
63
-------
EXPOSURE MODELING OF ACID AEROSOLS
Michael P. Zelenka
Atmospheric Sciences Modeling Division, Air Resources Laboratory, National Oceanic and Atmospheric Administration
"On assignment to the Atmospheric Research and Exposure Assessment Laboratory,
U.S. Environmental Protection Agency (MD-56)
Research Triangle Park, NC 27711.
Helen H. Suh
Harvard University School of Public Health, Department of Environmental Health
665 Huntington Avenue
Boston, MA 02115.
ABSTRACT
The U.S. Environmental Protection Agency (EPA) is conducting an intensive characterization and human
exposure monitoring program of acid species and related air pollutants in an urban environment. The EPA's
Atmospheric Research and Exposure Assessment Laboratory (AREAL) hi cooperation with the Harvard School of Public
Health and the Philadelphia Air Management Laboratory is conducting field measurements in Philadelphia, Pennsylvania
to: 1) characterize the spatial and temporal variations of acid aerosol species in an urban environment, 2) investigate the
complex chemistry of acid aerosols and other acidic species, including formation/removal mechanisms, and 3) provide
acidic aerosol and paniculate data base for exposure modeling and a study of pollution health effects.
One of the goals of the EPA's aerosol acidity study is to develop models of human exposure to acid aerosol
species. Exposure models will be used to estimate the distribution of human exposures to acid aerosols. The models
would be an important planning tool for assessing exposures by: 1) determining acid aerosol exposures hi high-risk
groups, 2) facilitating planning of subsequent sampling strategies, and 3) evaluating the effectiveness of proposed or
implemented mitigation efforts on reducing human exposures to acid aerosols.
This paper focuses on issues to be addressed hi developing models of human exposure to acid aerosols. The
intent is to describe a sampling scheme that provides the information needed for development of an acid aerosol exposure
model.
INTRODUCTION
Concern over the health effects from acid aerosols has increased in recent years. Evidence exists linking
exposure to acid aerosols with adverse health effects hi humans.' Especially susceptible are individuals who suffer from
respiratory ailments, including asthmatics and those with chronic bronchitis. Young children and the elderly are also
susceptible.
To address the concern over a health risk from exposure to acid aerosols, the U.S. Environmental Protection
Agency (EPA) is conducting an intensive characterization and human exposure monitoring program of acid species and
related air pollutants in an urban environment. The EPA's Atmospheric Research and Exposure Assessment Laboratory
(AREAL) in cooperation with the Harvard School of Public Health and the Philadelphia Air Management Laboratory is
conducting field measurements in Philadelphia, Pennsylvania to:
• characterize the spatial and temporal variations of acid aerosol species in an urban environment,
• investigate the complex chemistry of acid aerosols and other acidic species, including formation/removal
mechanisms,
• provide acidic aerosol and paniculate data base for exposure modeling and a study of pollution health effects.
Philadelphia is the first city in the multi-city Metropolitan Aerosol Acidity Characterization Study (MAACS) which will
enable characterization of acid aerosol exposures on a regional scale.
One of the goals of the EPA's aerosol acidity study is to develop models of human exposure to acid aerosol
species. Exposure models will be used to estimate the distribution of human exposures to acid aerosols. The models
would be an important planning tool for assessing exposures by: 1) determining acid aerosol exposures in high-risk
groups; 2) facilitating planning of subsequent sampling strategies; and 3) evaluating the effectiveness of proposed or
implemented mitigation efforts on reducing human exposures to acid aerosols.
64
-------
This paper focuses on issues to be addressed in developing models of human exposure to acid aerosols. The
intent is to describe a sampling scheme that provides the information needed for development of an acid aerosol exposure
model.
BACKGROUND
The Philadelphia aspect of the MAACS has four primary parts, all of which are important to developing an
exposure model. The first is to assess the spatial and temporal character of acid aerosol concentrations through a
network of fixed-site monitoring stations that are upwind, downwind, and throughout the Philadelphia metropolitan area.
Information obtained from this aspect will provide valuable input to exposure model development by aiding in the
determination of the effect of population and degree of urbanization on local acid aerosol concentrations. Information on
diurnal variations of acid aerosol concentrations will be an important parameter for model development as well.
Sampling strategies must take into account the variable nature of acidic aerosols where both minimum and maximum
concentrations typically occur between 8:00 am and 8:00 pm.2 Hence, a 12-hour sampling period would span both the
minimum and maximum ambient H+ concentrations.
The second aspect involves detailed measurements of acid aerosol concentrations during the summer months.
Information on formation and removal mechanisms, neutralization of particle strong acid (H+) by ammonia, and
neutralization by ammonium salts are gathered during this period. The information gathered on concentration and
duration of acid aerosol episodes is needed to determine the proper averaging time for sample collection based on health
responses and transport and conversion processes. This information will be vital to developing an exposure model which
accurately estimates a population's exposure when acid aerosol concentrations are typically at their highest levels.
The third component of the study includes the collection of data needed to perform exposure analyses.
Ambient, indoor, and personal measurements of acid species and related pollutants will be collected in a time period that
is consistent with both the reactive chemistry of the pollutants and the health response from exposure to the acid
aerosols. For calculation of exposure, various microenvironments (both indoors and outdoors) are being characterized
for key pollutants to determine levels and an appropriate sampling schedule. Variation in pollutant concentration due to
seasonal and diurnal effects is also being defined. Activity-pattern information can be obtained from previous studies of
human activity,3A5 or by tracking the daily activities of the individuals wearing a personal exposure monitor (PEM) in
each city for a site-specific account of exposure. Of particular importance is the frequency and duration with which
susceptible groups (e.g., asthmatics and individuals with chronic respiratory disease) come into contact with acid
aerosols.
The fourth component of the study will be testing of an integrated weekly sampling approach. Model
requirements will be developed which will be considered in the planning of field studies in the other cities in the
MAACS and future field studies as well. The sampling protocol for these future studies needs to be consistent with the
model requirements to be developed.
Preliminary Modeling of Indoor Acid Aerosol Concentrations
A model for estimating indoor concentrations of H+ was developed from a study that estimated indoor exposures
of children using air pollutant concentrations measured at a single stationary ambient monitoring (SAM) site.2 In the
above mentioned study, the estimated acid aerosol concentrations were compared to measurements collected at the SAM
site. Factors that influenced indoor and personal concentrations were identified and incorporated into the acid aerosol
model.
The predictive model for indoor H+ concentrations is given by:
;) - ([H
IT =[H+Jour- UH+JOUT » kSO^ - UH+JOUT . kSOJ • kNH3 * iNHjU (1)
where H* (nmol/m3) is the estimated indoor concentration, H+om- (nmol/m3) is the measured outdoor concentration,
kSOJ is a dimensionless term representing the fraction of particles that deposit on indoor surfaces or that fail to
penetrate indoors, and kNH3 * NH3 [N is the first-order reaction rate of H+ with NH3. Equation (1) shows that to
estimate indoor H+ requires information on the concentration of outdoor H+, the indoor NH3 concentration, the loss rate
for sulfate, and the reaction coefficient for the reaction of H+ with NH3.
65
-------
Factors Affecting Acid Aerosol Concentrations
A study on personal exposure modeling of aerosol strong acidity in Uniontown, Pennsylvania, during the
summer of 1990 (Suh el. a/.2) found among other results that: 1) there was little to no spatial variation in outdoor H+
concentrations within Uniontown, 2) there was significant diurnal variation in outdoor H+, 3) personal H+ exposure
levels were generally higher than indoor concentrations and lower than outdoor concentrations with the differences being
quite pronounced, 4) outdoor concentrations of H* were unable to account for interpersonal variability in personal H+
exposures, 5) time-weighted microenvironmental models predicted personal exposures for H+ better than outdoor
concentrations alone, and 6) indoor concentrations of H* appeared to be affected by the use of air conditioning because
of its effect on concentrations of gaseous ammonia or ammonium salts.
The reactive chemistry of acid aerosols and ammonia must be included in both indoor concentration and
personal exposure models of acid aerosols. Individual species must be characterized. Ammonia and acid particles do
not coexist, however, they could each be present in appreciable amounts at different times during a multi-hour sampling
period.6 Therefore, an integrated sample collected over several hours may collect both acid panicles and ammonia, thus
hiding the true interaction between these chemical species and making modeling of exposures very difficult.
Health effects associated with acid aerosols are primarily due to exposure to fine (aerodynamic diameter i 2.5
fim) particles. For modeling purposes, it is important to discriminate by particle size. This is particularly true for
indoor concentrations. Particle size and density, deposition on indoor surfaces, reaction rates with other indoor
pollutants, volume of home, and air exchange determine the indoor decay rate of pollutants originating from outdoor
sources.7 Information on housing stock-type and ventilation characteristics may aid in explaining differences between
indoor and outdoor concentrations of acid aerosols and other key pollutants.
For pollutants such as acid aerosols which are characterized by considerable infiltration into homes, it is
necessary to carefully measure the variation of both indoor and outdoor concentrations for use in an exposure
assessment. It has been seen that ambient levels of pollutants with high infiltration rates significantly affect personal
exposures even where indoor sources are present.8 Measurements taken in both indoor and outdoor environments are in
addition to personal sampling. All three types of samples being necessary to characterize and model the spatial and
temporal variation of acid and related aerosols.
An exposure model for acid aerosols must incorporate the reactive chemistry of acid species, particularly with
respect to either co-existence or neutralization by ammonia. Infiltration of fine acid sulfate particles to the indoor
environment needs to be included. Once indoors, heterogeneous processes such as reactions with indoor surfaces and
deposition, needs to be accounted for by the model. Detailed information on the interaction between people conducting
their daily routines and airborne acid concentrations in outdoor and indoor microenvironments is vital for exposure
modeling. Particularly important is time-location and activity data for persons with chronic bronchitis, asthmatics, and
other subsets of the population most susceptible to inhalation of acid aerosols.
RECOMMENDATIONS
Considering the factors from the previous section which affect the spatial and temporal distribution of acid
aerosols, and the findings of Suh and co-workers2 (also in the section above), the following list of recommendations
needs to be considered when planning future exposure studies for acid aerosols.
• Indoor/outdoor (I/O) ratios of acid aerosols are needed to conduct exposure modeling.
Therefore, indoor measurements of acid aerosols as well as outdoor measurements will be required
to model personal exposures to acid aerosols.
• Ammonia levels and sources should be characterized.
• Additional studies of the interactions between H+ and NH3 are needed, especially in the
breadiing zone of a person. Also, information is needed on loss of particles in the immediate
vicinity of a person's body.
• Air exchange and room volume measurements for each dwelling are needed particularly in light
of the fact that H+ particles are primarily from outdoors.
66
-------
• Sampling should take place on a short enough temporal scale to monitor the variation in outdoor
and indoor acid aerosol concentrations.
• Personal exposure sampling should be conducted since interpersonal variability in acid aerosol
concentrations can be large.2
Sampling Protocol
The above recommendations would be expressed through the following protocol of sampling for acid aerosols.
Sample duration:
Indoors: 24-hour sampling with a resolution of no more than 3 hours (1-hour samples would be
preferable).
Outdoors: Same schedule as indoors.
Personal: Samples with the shortest sampling duration achievable (preferably no longer than 3-hour
samples).
The sampling schedules described above may not be appropriate given current measurement capabilities.
Sampling which is not able to achieve the limit of detection (LOD) for a chemical species should not be done. But, as
sampler technology and analytical methods improve, and the LOD for key species is lowered, shorter sampling durations
should be used.
Sample frequency:
Indoors and personal: Every day (24 hours).
Outdoors: Every second day (24 hours).
After analyzing data from the Harvard Acid Aerosol and Six-Cities Studies for seasonal and spatial variability of
aerosol acidity, Thompson et. al.9 determined that outdoor samples collected every other day provided adequate
information on annual averages and medians. Therefore, sampling outdoors every other day would be acceptable.
Primary measurements:
1. Aerosol strong acidity (H+) and ammonia (NH3).
2. SOJ, NO;, NHJ, O3, HONO, HNO3, SO2, and NO2.
3. PM^ and PM10.
Other:
1. Complete characterization of dwelling including: number and orientation of rooms, square footage,
number of windows, air-conditioning, etc.
2. Air exchange measurements across the indoor/outdoor interface for various housing stock and
building types.
3. Ventilation data for indoors.
4. Time/location activity patterns for individuals wearing a personal exposure monitor (PEM). It
would be beneficial to obtain time-activity profiles for representative subgroups of the population
whether they carry a PEM, or not.
SUMMARY
The U.S. EPA's Atmospheric Research and Exposure Assessment Laboratory is conducting research into the
characterization of aerosol acidity and related chemical species with emphasis on their relationship to human exposure.
Exposure modeling of acid aerosol will play an important role in estimating the exposures experienced by urban
populations. The modeling of human exposures to acid aerosols requires detailed information on the spatial and
temporal distribution of H+ and other relevant chemical species. Data on human activity patterns is also required. As
yet there are still many unanswered questions regarding the atmospheric and chemical processes that affect aerosol
acidity and the impact of aerosol acidity on human health. Exposure modeling of H+ will be an invaluable tool for
assessing the distribution of exposures experienced by an urban population and for identifying population subgroups that
are most at risk from exposure to aerosol acidity.
67
-------
In this paper we have tried to identify the physical parameters needed for developing a human exposure model
for acid aerosols. A rigorous sampling protocol has been outlined which is appropriate for collecting the level of
information needed for model input. Future exposure studies of acid aerosols should incorporate as much of the
sampling protocol outlined here as possible in their study design.
ACKNOWLEDGEMENTS
This paper has been reviewed hi accordance with the U.S. Environmental Protection Agency's peer and
administrative review policies and approved for presentation and publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
This paper was written as part of the U.S. EPA's Philadelphia Aerosol Acidity Characterization Study. The
authors thank Dr. William E. Wilson and Mr. Robert Burton of the U.S. EPA for their contributions to the project.
68
-------
REFERENCES
1. An Acid Aerosols Issue Paper. Health Effects and Aerometrics. EPA-600/8-88-005, U.S. Environmental
Protection Agency, Washington, D.C., April, 1989. ,
2. H.H. Sun, J.D. Spengler, and P. Koutrakis, "Personal exposures to acid aerosols and ammonia," Environ. Sci.
Technol. 26:2507-17 (1992).
3. T.R. Johnson, A Study of Human Activity Patterns in Cincinnati. Ohio. Electric Power Research Institute, Palo
Alto, California, 1987.
4. T.R. Johnson, A Study of Personal Exposure to Carbon Monoxide in Denver. Colorado. EPA-600/54-84-014,
U.S. Environmental Protection Agency, Research Triangle Park, 1984.
5. T.D. Hartwell et. al., Study of Carbon Monoxide Exposure of Residents of Washington. D.C. and Denver.
Colorado. EPA-600/54-84-031, U.S. Environmental Protection Agency, Research Triangle Park, 1984.
6. P. Koutrakis, K.M. Thompson, J.M. Wolfson, et. al., "Determination of aerosol strong acidity losses due to
interactions of collected particles: Results from laboratory and field studies," Atmospheric Environment
26A:987-995 (1992).
7. P. Koutrakis, M. Brauer, S.L.K. Briggs, and B.P. Leaderer, "Indoor exposures to fine aerosols and acid
gases," Env. Health Persp. 95:23-28 (1991).
8. D.A. Butler, H. Ozkaynak, J.D. Spengler, and I.H. Billick, "Predicting population exposures to air pollutants
using physical and stochastic models," Presented at the 9" World Clean Air Congress, Montreal, Paper #IU-
23A.07, 1992.
9. K.M. Thompson, P. Koutrakis, M. Brauer, et. al., "Measurements of aerosol acidity: Sampling frequency,
seasonal variability, and spatial variation," Presented at the 84* Annual Meeting and Exhibition of the Air &
Waste Management Association, Vancouver, BC, Paper #91-89.5, 1991.
69
-------
Session 3
Measurement of Semi-Volatile
Organic Compounds
-------
The Distribution of Semi-Volatile Organic Compounds
Between the Vapor and Particulate Phases: Diffusion
Denuder Measurements
Delbert J. Eatough, Hongmao Tang, Wenxuan Cui and
James Machir
Department of Chemistry, Brigham Young University,
Provo,UT 84602
ABSTRACT
Collection of particles on a filter results in under-estimation of paniculate organic compounds due
to losses from the semi-volatile organic fraction during sample collection, i.e. a "negative sampling artifact"
This sampling induced change in the phase distribution of semi-volatile organic material results in the loss of
about half of the paniculate organic material during sampling. These semi-volatile organic compounds lost
from particles can be correctly measured using a diffusion denuder sampling system. A multi-system, multi-
channel, high-volume diffusion denuder sampler has been used for the determination of the particle size
distribution and chemical composition of semi-volatile organic compounds in fine particles in two urban
environments, Los Angeles and Philadelphia. Organic compounds lost from the particles included paraffinic
compounds, aromatic compounds, and organic acids and esters. Underestimation of the composition of semi-
volatile organic compounds in particles is a function of molecular weight, chemical compound class and particle
size. The majority of the organic compounds in fine particles 0.8 to 2.5 (im in size are semi-volatile organic
compounds lost from the particles during sampling onto a filter. The majority of carbonaceous material in
particles smaller than 0.4 |im is not lost from the particles during sampling.
INTRODUCTION
Correct assessment of the exposure of a population to paniculate organic material is in part dependent
on accurate determination of the chemical composition as a function of panicle size for particles present in
the atmosphere. Results obtained from the collection of organic material on a filter indicate that about one-
third of the mass of fine paniculate matter (diameter < 2.5 (im) collected on filters in remote desert regions
of the Southwest U.S. (Macias 1986, Sutherland 1990) and about one-fourth of the fine paniculate mass in
western urban areas such as the Los Angeles Basin (Hering 1993) is organic compounds and elemental carbon.
In the Eastern United States, sulfate is the major component of airborne fine particles. However, based on
filter data, organic material also comprises about one-fourth of the fine paniculate mass in the east (Gebhart
1993). Collection of particles on a filter results in underestimation of paniculate organic material due to losses
from the semi-volatile organic fraction during sample collection, i.e. a "negative sampling artifact". This
sampling induced change in the phase distribution of semi-volatile organic material results in the loss of about
half of the paniculate organic material during sampling (Eatough 1993,1990,1989, Tang 1993). This "negative
sampling artifact" is an order of magnitude larger than the "positive sampling artifact" resulting from the
collection of organic compounds by a quartz filter (Eatough 1993, Appel 1989, McDow 1990).
73
-------
The loss or gain of significant amounts of semi-volatile organic material from particles or the sampling
media during sampling causes errors in the determination of aerosol chemical composition. Accurate collection
procedures for semi-volatile organic compounds must meet the following two criteria.
1. Organic compounds initially present in the gas phase must not be adsorbed onto particles or
the filter during sampling.
2. Organic compounds initially present in the paniculate phase must be captured during
sampling separate from compounds which are present in the gas phase in the atmosphere.
These two criteria cannot be met by any sampling procedure in which the paniculate phase organic
compounds are collected before the collection of gas phase organic compounds because the gas phase organic
compounds and organic compounds volatilized from particles become indistinguishable. Thus, it is necessary
to first remove the gas phase organic compounds and then to collect the paniculate phase organic compounds
with a sampler which will collect all organic material, gas and particle.
This paper described the results obtained using a high-volume, multi-system, multi-channel diffusion
denuder sampling system (Tang 1993) (BIG BOSS) and associated analytical procedures for the determination
of the size distribution and chemical composition of fine paniculate organic material. Details are given on the
chemical composition and concentration of semi-volatile organic compounds retained by and lost from particles
during sampling for samples collected in the Los Angeles urban area. Results obtained from capillary column -
gas chromatography, GC, analysis of collected samples are compared for samples from Los Angeles and
Philadelphia sampling sites. Semi-volatile paniculate organic compounds present in samples collected at Los
Angeles and Philadelphia are identified by GC-MS analysis.
EXPERIMENTAL
The BIG BOSS Sampling System.
The BIG BOSS sampling system has been previously described (Tang 1993). The BIG BOSS uses a
variety of size selective virtual impactor inlets to control the particle size of the particles introduced to the
diffusion denuder sampler. The components of the BIG BOSS are shown schematically in Figure 1. Systems
1, 2 and 3 are used to determine total paniculate organic material after a diffusion denuder which removes
gas phase organic compounds. Total flow through the denuder for each system is 200 L/min. The denuder
for these three systems is preceded by a virtual impactor with particle size cuts of 2.5, 0.8 and 0.4 um,
respectively (Tang 1993). The flow stream after the denuder is split and sampled through two parallel filter
packs. The majority of the flow, 160 L/min, is sampled through a quartz filter followed by an XAD-II bed.
Particles are collected by the quartz filter. Semi-volatile organic compounds lost from the particles during
sampling are collected in the XAD-II bed. The material collected in this part of the sampling system is
analyzed by GC and GC-MS to chemically characterize the organic material present in the panicles and lost
from the panicles during sampling. The remainder of the sample flow, 40 L/min, is sampled through a filter
pack with a quartz filter and a carbon impregnated filter, CIF. These filters are used for quantitative
determination of total paniculate carbonaceous material and semi-volatile compounds lost from the particles
during sampling by temperature programmed volatilization, TPV, analysis (Eatough 1993,1989, Tang 1993).
The fourth system in the BIG BOSS, Figure 1, contains two quartz filters in front of the diffusion denuder.
The data obtained with System 4 is used to correct the data from the other systems for any gas phase organic
compounds not collected by the diffusion denuder.
Sampling With the BIG BOSS.
The BIG BOSS has been field tested in two urban sampling programs. The first sampling program
was completed during June 1992 in the Los Angeles Basin at the South Coast Air Quality Management District,
74
-------
SCAQMD, sampling site in Azusa, CA. Representative results related to the identification of paniculate semi-
volatile organic compounds in samples collected at Azusa were included in the article describing the BIG BOSS
(Tang 1993). Results are reported here for eight-hour samples (10:00 to 20:00 hours) collected on 16
consecutive days and four-hour samples collected twice a day (10:00 to 14:00, and 14:30 to 18:30) during the
last three days of the sampling program at the SCAQMD Azusa sampling site. Samples were then collected
during a one-week period in July 1992 at the North-East Airport Harvard University sampling site in
Philadelphia, PA (Burton 1993). The total number of quality samples obtained at the Philadelphia site was
limited to four samples because of inclement weather during part of the study.
Sample Analysis.
The samples collected during the two field sampling programs outlined above were analyzed to
determine artifact-free paniculate organic material in the particle size ranges of <0.4u.m, 0.4-0.8|im, and 0.8-
2.5|im by both quantitative and qualitative analysis of the collected samples. Quantitative results for total
paniculate organic material in each of the three size ranges were obtained by TPV analysis (Eatough 1993,
1989). Semi-quantitative chemical characterization results for semi-volatile paniculate organic material and
the semi-volatile organic compounds lost from collected panicles during sampling as a function of particle size
have been determined by GC with FID detection (Tang 1993). The specific qualitative identification of the
principal organic compounds lost from particles and semi-volatile organic compounds retained by particles
during sampling has been done by gas chromatography-mass spectroscopy, GC-MS, analysis of combined
samples. Examples of results obtained by each of these analyses are presented in this paper.
RESULTS AND DISCUSSION
Semi-volatile organic material is lost from particles during sample collection on a filter. The amount
of paniculate organic material on filter Q,, (the first quartz filter after the 2.5 u,m inlet and denuder in System
1, Figure 1) agrees with that on filter Qt, (the first quartz filter after the 2.5 urn inlet in System 4) (Eatough
1993, Tang 1993). This agreement shows that the half-life for the loss of semi-volatile organic material from
panicles during sampling is much shorter than the sampling time so that the loss is comparable for particles
collected on a filter before and after a diffusion denuder. This result also shows that the positive artifact
resulting from the absorption of gas phase organic compounds by the quartz filter is minimal and that the gas-
particle equilibrium is not significantly perturbed during passage through the denuder.
The data from the GC analysis of the material collected in the XAD-II beds or the TPV analysis of
the charcoal impregnated filters of Systems 1-4, Figure 1, allow the determination of the particle size
distribution of the semi-volatile organic material which was lost from the particles during sampling. The GC
analysis of the material extracted from the XAD-II sorbent beds by dichloromethane gives a semi-quantitative
measure of the various organic compounds captured by the XAD-II sorbent bed. This is illustrated by the data
in Figure 2 where is shown the GC data obtained from analysis of the four sorbent beds for the sample
collected on 14 June 1992. The amount of material on the XAD-II sorbent bed decreases in going from
samples Xu through X,., Figures 1 and 2. The decreasing amounts seen in going from System 1 through
System 3 reflects the decreasing amount of semi-volatile organic material lost from the particles as the inlet
panicle size cut is decreased from 2.5 |im, to 0.8|im, and finally to 0.4 um. The small amount of material seen
for the Xi,, sorbent bed in Sampler 4 results from the incomplete collection of some gas phase organic
compounds by the denuder. There is a slight tendency for the relative importance of the more volatile semi-
volatile organic compounds lost from particles during sampling to increase with decreasing particle size, Figure
2. The organic compounds with longer GC retention times are more prominent in the Sampler 1 XAD data,
indicating that these com pounds are dominantly present in the atmosphere in the 0.8-2.5 |im particle size range.
Organic compounds with short GC retention times dominate the organic material seen in the X,,, sorbent bed
(Figure 2), consistent with the expected incomplete collection of about 5% of the gas phase organic compounds
by the diffusion denuder (Tang 1993).
75
-------
The TPV analyses of the semi-volatile organic material captured by the GIF corresponding to each
XAD sorbent bed, Figure 1, gives a quantitative determination of the total semi-volatile organic material lost
from the particles and collected by the XAD-II sorbent bed or GIF filter after the denuder. The total
integrated peak area of the GC data for each XAD-II collected sample is directly related to the amount of total
evolved carbon determined from the TPV analysis, Figure 3. The correspondence between these two data sets
provides a calibration of the GC results which may be used to quantitatively interpret the GC results obtained
in the various parts of the sampling system.
The data for the amount of organic material collected by the XAD-II sorbent beds or the CIF sorbent
traps in the first three sampling systems needs to be corrected for the denuder breakthrough measured in the
XAD-II sorbent bed or CIF of the fourth sampling system to obtain the semi-volatile artifact as a function of
particle size. This leads to the data shown as SVOC Lost in Figure 4 for the average particle size distribution
of the semi-volatile organic material lost during the collection of panicles for all of the sampling periods in the
Los Angeles Basin study (Tang 1993).
The organic material determined by the GC analysis of the combined dichloromethane/methanol
extraction (Tang 1993) of the quartz filters in front of the various XAD-II sorbent beds for Sampling Systems
1,2 and 3 (Qu, Qu and Qu in Figure 1) is a measure of the semi-volatile organic material not removed from
the particles during sampling. These results are also given in Figure 4 as Quartz SVOC. The TPV analysis
of the corresponding quartz filters in the CIF filter pack for each system (Qj,i, Qu an(1 QM in Figure 1) gives
the total carbonaceous material remaining on each quartz filter after sampling, Quartz C in Figure 4. The sum
of Quartz C and Quartz SVOC gives the total paniculate C, Particle C in Figure 4.
As indicated in Figure 4, the majority of the organic material in particles 0.8-2.S |im in size is lost from
the particles during collection of the particles on a filter. About 80% of the carbonaceous material in the 0.8-
2.5 urn particles consists of semi-volatile organic compounds which an be stripped from the particles during
sampling. The semi-volatile organic material in particles 0.4-0.8 urn in size is also essentially all lost from
particles during sampling, Figure 4. However, about 60% of the total carbonaceous material in these particles
is retained by the particles during sampling, Figure 4. In contrast, the great majority of the carbonaceous
material in the particles smaller than 0.4 |im is retained by the particles during sampling. These smallest
particles are also the only size fraction with a significant amount of semi-volatile organic material remaining
in the particles after collection of a sample, Figure 4. This regular trend of decreasing importance of the loss
of organic material from particles with decreasing particle size probably results from a combination of two
factors: 1. The concentration of elemental carbon increases with decreasing particle size. The increased
amounts of "soot" in the <0.4 urn size particles, as compared to larger fine particles, can be expected to result
in the retention of some semi-volatile organic material in these particles due to strong absorption of the semi-
volatile organic compounds by the graphitic structure of the soot 2. The concentration of paniculate
secondary organic material produced from photochemical processes probably increases with decreasing particle
size. This material will be rich in oxygen and nitrogen as a result of the photochemistry. The resulting organic
material will be relatively non-volatile and would be expected to be retained by the particles during sampling.
We have previously reported data on the GC-MS identification of the semi-volatile organic material
lost from particles during sampling for the sampling program in Los Angeles (Tang 1993). Compounds are
present from all major organic compounds classes expected to be present in the atmosphere. For each
compound class, the more volatile compounds predominate in the material lost from the particles and collected
in the XAD-II bed during sampling. In contrast, the higher molecular weight organic compounds are retained
by the particles during sampling. For example, paniculate n-tetradecane and n-pentadecane are found only
in the XAD-II bed and not in the particles after sampling. Hydrocarbons lower in molecular weight than these
two compounds are found in comparable concentrations in the XAD-II beds of both Samplers 1 and 4,
indicating they originate mainly from the breakthrough of some fraction of the gas phase component of these
species. In contrast, n-tetracosane and higher molecular weigh! aliphatic hydrocarbons are retained by the
particles during sampling and are not found in the XAD-II sorbent beds. Compounds of intermediate
76
-------
molecular wight, e.g. n-decosane, are partially lost and partially retained by the particles. Also illustrated by
the GC-MS data is the increased tendency for lower molecular weight semi-volatile organic compounds to be
retained by the particles during sample collection as the polarity of a given molecular weight compound
increases. For example, n-heptadecane (MW 226) is largely lost from particles during sampling. However
lauric acid (MW 214) and fluoranthene (MW 202) are largely retained by the particles during sampling.
The GC data obtained to the present indicate that the chemical composition of the semi-volatile
compounds lost from particles during sampling was similar for samples collected at each of the two urban study
sites, Los Angeles and Philadelphia, Figure 5. This probably reflects the importance of organic compounds
from automotive emissions at each of these sites. Semi-volatile organic compounds lost from particles during
sampling at both of the urban sampling sites included paraffins, aromatic compounds and organic acids and
esters.
ACKNOWLEDGEMENT
This research was supported by Southern California Edison, PacifiCorp, and the U.S. Environmental
Protection Agency through a cooperative research agreement to BYU. Appreciation is expressed for the
technical assistance of Laura Lewis, Chris Ogden, and Rebecca Eatough.
REFERENCES
Appel B.R., Cheng W. and Salaymeh F. (1989) "Sampling of carbonaceous particles in the atmosphere~II,"
Aerosol Sci. Tech.. 1ft 2167-2175.
Burton R., Wilson W., Koutrakis P., Allen G. and Hauptman F. (1993) "An overview of the Philadelphia acid
aerosol characterization and human exposure assessment studies," Measurement of Toxic and Related Air
Pollutants. Air and Waste Management Association, Pittsburgh, in press.
Eatough DJ., Wadsworth A, Eatough D.A, Crawford J.W., Hansen L.D. and Lewis E.A (1993) "A multiple-
system, multichannel diffusion denuder sampler for the determination of fine-paniculate organic material in
the atmosphere," Atmos. Environ., in press.
Eatough DJ., Aghdaie N., Cottam M., Gammon T., Hansen L.D., Lewis E.A and Farber R.J. (1990) "Loss
of semi-volatile organic compounds from panicles during sampling on filters," Transactions. Visibility and Fine
Panicles. C.V. Mathai, ed., Air and Waste Management Association, Pittsburgh, PA, p 146-156.
Eatough D.J., Sedar B., Lewis L., Hansen L.D., Lewis E.A. and Farber R.J. (1989) "Determination of semi-
volatile organic compounds in particles in the Grand Canyon area," Aerosol Sci. Tech.. 10: 438-439.
Gebhart K.A and W.C. Malm (1993) "Examination of the effects of sulfate acidity and relative humidity on
light scattering at Shenandoah National Park," Atmos. Environ., in press.
Hering S. (1993) "Descriptive analysis of Los Angeles aerosols during SCAQS" Southern California Air Quality
Study: Data Analysis. Air and Waste Management Association, Pittsburgh, PA, p. 15-20.
Macias E.S., Vossler T. and White W.H. (1986) "Carbon and sulfate fine particles in the Western U.S."
Transactions: Visibility Protection - Research and Policy Aspects. P.S. Bhardwaja, ed., Air & Waste
Management Association, Pittsburgh, PA, p 361-372.
McDow S.R. and Huntzicker JJ. (1990) "Vapor adsorption artifact in the sampling of organic aerosol: face
velocity effects," Atmos. Environ.. 24: 2563-2571.
77
-------
Sutherland J.L. and Bhardwaja P.S. (1990) 'Seasonal fine mass budgets through regression-application to the
Glen Canyon SCENES data,' Transactions: Visibility and Fine Particles. CV. Mathai, ed., Air & Waste
Management Association, Pittsburgh, PA, p 207-212.
Tang H., Lewis E.A., Eatough D.J., Burton R.M. and Farber RJ. (1993) 'Determination of the particle size
distributionand chemical compositionof semi-volatile organic compounds in atmospheric fine particles." Atmos.
Environ., in press.
Legend
ultl-Ch«nnol
Diffusion Donuder
Filter Pick with 1
or 2 Quartz Flltora
Filter Pick with
2 Quartz Filtort
•nd 1 GIF
I -si X»D-IISorb«ntB«d
System 1 System 4 System 2 System 3
2.5pm Inlet 2.5pm Inlet 0.8pm Inlet 0.4pm Inlet
Figure 1. Schematic of the BIG BOSS sampling system, Sampler 1 (denuder,
2.5 ym inlet cut), Sampler 2 (denuder, 0.8 ym inlet cut), Sampler 3 (denuder,
0.4 ym inlet cut), and Sampler 4 (filter/denuder, 2.5 ym inlet cut).
78
-------
SAMPLER 1
SAMPLER 2
JLiki.
SAMPLER 3
SAMPLER 4
Figure 2. GC data for the organic compounds collected in the XAD-II beds after collection of particles
in Samplers 1-4 of the BIG BOSS for samples collected from IftOO to 18:00 pm June 14 in
Los Angeles Basin (Azusa).
79
-------
<2.5ym
<0.8gm
<0.4nm
15
10
20 30
40
50 60
(Thousands)
TOTAL GC AREA, VOLUME NORMALIZED
Figure 3. Comparison of the GC trace peak area for Che organic
material extracted from an XAD bed and the total collected semi-
volatile organic C on the corresponding GIF filter of the BIG
BOSS sampler.
0.8-2.5|jm
0.4-0. 8pm
<0.4um
Quartz C SVOC Loit
SAMPLE
Quartz SVOC
Figure 4. Fine particle size distribution of total carbonaceous
material (Particle C), total carbonaceous material retained by particles
during sampling (Quartz C), semi-volatile organic material lost from
particles during sample (SVOC Lost) and semi-volatile organic material
retained by particles during sampling (Quartz SVOC) for samples collected
in the Los Angeles Basin.
80
-------
A
Los Angeles (Azusa)
mi
Philadelphia
Figure 5. GC data for the organic compounds collected in an XAD-II bed after collection of particles
in Sampler 1 of the BIG BOSS for samples collected in Los Angeles (Azusa) and
Philadelphia.
81
-------
Use of a High-Volume Small Surface Sampler (HVS3)
for the Microbiological Evaluation of Dust from
Carpeted and Non-Carpeted Surfaces
K.E. Leese, R.M. Hall, E.C. Cole, and K.K. Foarde
Research Triangle Institute, Research Triangle Park, NC 27709
and
MA Berry
U.S. Environmental Protection Agency, Research Triangle Park, NC 27709
A High-Volume Small Surface Sampler (HVS3) was previously developed through the
U.S. Environmental Protection Agency (EPA) for the collection and analysis of lead and
pesticides in surface dust (>5um) of carpeted and non-carpeted floors in residential buildings.
At the request of EPA/Environmental Criteria and Assessment Office, Research Triangle
Institute (RTI) has adapted the use of the HVS3 for the collection of surface dust for micro-
biological analyses. The major adaptations involve disinfecting the unit, and using sterile col-
lection bottles. The collected dust samples are weighed, then sieved through a sterile, de-
pyrogenated 250u.m screen. They are analyzed for mesophilic and thermophilic bacteria,
fungi, mite guanine, and endotoxins. Dust samples were collected from routinely maintained
carpet and tile floors in building 'A', and "soiled" carpet from another building "B". Building A
had dust loading levels of 1.3 g/m2 on carpet and 0.1 g/m2 on tile floors. Levels of endotoxin
(0.2 to 6 ng/g of dust) and mite guanine were insignificant. Endotoxin levels were lower in
the tile floor dust than the carpet dust. Mesophilic bacteria colony forming units (CPU)
ranged from 7.0 X 104 CFU/g (tile floor) to 8.8 X 10" CFU/g (carpet), while thermophilic
counts ranged from 5.0 X 105 CFU/g (tile floor) to 7.0 X 103 CFU/g (carpet). Fungi counts
ranged from 5.0 X 103 CFU/g (tile floor) to 1.2 X 10s CFU/g carpet. In building B, dust levels
were found to be higher at 6.2 g/m2 of carpet. Mold and endotoxin levels were found to be
higher in building B at 3.2 x 10" CFU/g and 103 ug/g respectively. The HVS3 can be used to
measure dust microbiological loadings on surfaces. Additionaly, bioaerosol monitoring will
allow the relationship between surface and airborne microbial concentrations to be evaluated.
82
-------
BACKGROUND
The Research Triangle Institute (RTI) is currently performing a study with the
Environmental Protection Agency's Environmental Criteria and Assessment Office
(EPA/ECAO) to characterize biopollutants (both airborne and surface) for exposure assess-
ments inside a building. A part of the study involves the investigation of levels of floor
surface dust, its microbial composition, and its effect on indoor air. A method was needed
for collecting floor surface dusts. A High Volume Small Surface Sampler (HVS3) was pre-
viously developed by the U.S. EPA's Atmospheric Research and Exposure Assessment
Laboratory (EPA/AREAL) for the collection of residential carpeted and non-carpeted surface
dust for pesticide and lead analysis.'2 The HVS3 is currently recommended for use in a
draft ASTM method for the collection of dust for lead analysis. RTI has adapted procedures
for using the HVS3 to collect dusts for microbiological analyses. The major adaptations in-
volve disinfecting the unit, using sterile (steam autoclaved) polypropylene bottles to collect
samples, and using Du Pont Hysurf® bags on the exhaust of the unit. For the first time, RTI
has shown that the HVS3 can be used to collect surface dusts for the analysis of viable
molds, yeasts, bacteria, endotoxins, and dust mite guanine. Preliminary data suggest that
the HVS3 is useful in determining microbiological loadings on carpeted and non-carpeted sur-
faces. Use protocol development is currently ongoing and final results from the building bio-
pollutant exposure study will be reported at a later time. This paper discusses a brief history
of the HVS3, its characteristics, and some preliminary microbial results from two indoor en-
vironments.
INTRODUCTION
The HVS3 utilizes a one horse power Royal" vacuum motor and a specifically designed
nozzle and cyclone trap. The unit has Dwyer Magnehelic* gauges which are used to man-
ually set the flow rate and pressure drop across the nozzle at the monitored surface. At the
specified flow rate and pressure drop, the unit draws air at 20 cfm and has a 50% (D^) cutoff
point of 5 microns. The cyclone effectively collects 99% of the dust lifted by the vacuum.3
The dust collection efficiency has been reported to be approximately 93 to 97% from a
smoothly painted surface, and averages 32%4 on carpeted surfaces, dependent upon
surface loading.2 Du Pont Hysurf® bags, which are 98% efficient at 1 micron, are used on
the exhaust of the unit. The Hysurf® bags are useful in determining whether bio-
contaminants between 1 and 5 microns are exhausted from the cyclone during sampling.
HVS3 OPERATION
Cleaning and Leak Checking
Each part of the HVS3 which may come in contact with dust samples is thoroughly
cleaned with a nylon brush, rinsed with hot water, and disinfected with 70% ethanol for 30
seconds. The pieces are dried with clean gauze and reassembled. A sterile, polypropylene
bottle is attached to the cyclone. A leak-check is performed with a separate, Dwyer
Magnehelic* gauge between a hose connector on the nozzle and one downstream near the
flow control valve. The nozzle is closed with duct tape during the leak check. The unit must
test less than 0.02 inches of water air flow to pass the leak check.
83
-------
Sampling
The area to be sampled is delineated with tape measures, and masking tape is used to
mark the end lines of the area. The end lines are recommended to be at least one meter
apart, and typically are extended by 0.5 meter increments up to 2 meters, depending upon
the available space and the amount of sample required. The end line tapes are then evenly
marked in 75 mm increments (3 inches) to create 'lanes'. The HVS3 is then placed on an
end lane and the wheels are wiped with 70% ethanol. The unit is started, the flow rate and
pressure drop is set, and a timer is started as the sample collection begins. The nozzle is
passed over each lane between the end lines eight times (back and forth four times) at
approximately 2 ft per second. Once the last lane is sampled in this manner, each lane is
then passed over again as the sampler returns to the original end lane. At this point, the
sampler and timer is stopped. The polypropylene bottle is removed, capped, labeled, and
stored in a cooler. A data sheet is used to record the sample number, location, area, time
sampled, and flow and pressure drop settings.
SAMPLE PREPARATION AND ANALYSIS
Sample Mass and Carpet Loading
The surface dust samples are collected in sterile, polypropylene bottles and returned to
RTI laboratories in coolers for processing. The bottles are emptied and the mass of dust is
recorded. The dust is loaded into a sterile, depyrogenated 250 urn stainless steel sieve and
mechanically shaken for 30 minutes. The dust mass fractions greater than 250 urn and less
than 250 urn are calculated. Data from the sampling procedure are used to calculate the
surface dust loading in grams per square meter.
Fungi and Bacteria
The sieved bottom dust (<, 250 urn) is diluted ten and one hundred-fold, and the dilutions
are plated on Sabouraud Dextrose Agar (SDA) plates for fungi (molds and yeasts), and
Trypticase Soy Agar (TSA) for mesophilic and thermophilic bacteria. The SDA plates are
incubated at 25°C for fungi, and the TSA plates are incubated at 32°C for mesophilic, and
55°C for thermophilic bacteria. After incubation, the colonies on the plates are counted and
identified, and colony forming units per gram of dust are calculated.
Endotoxin and Dust Mite Guanine
Endotoxin is a lipopolysaccharide (Ips) component of gram negative bacteria cell walls. It
is a respiratory irritant and can cause toxic effects. For endotoxin, the bottom dust is serially
diluted and aliquots are reacted with an endotoxin specific lysate. The absorbance at 405
nm is read with a microtiter reader and is compared with the absorbance of a standard curve
with known concentrations. The sample concentration is then calculated.
The presence of dust mites in dust samples are measured as a function of the presence
of dust mite guanine. The sieved dust (250 mg) is placed through a series of extractions and
reacted with guanase and xanthine oxidase. Other reagents are added and the absorbance
of the final solution is read at 490 nm against controls with known concentrations.
84
-------
PRELIMINARY RESULTS
Environments
The surfaces of two buildings designated as 'A* and 'B1 were sampled with the HVS3.
The floors in building A are routinely maintained and appear to be clean. Carpeted and tiled
floors were monitored in building A. The carpeted floor surface of Building B did not appear
to be routinely maintained and was noticeably 'soiled.' A carpeted floor was monitored in
building B.
Surface Dust Loading
Figure 1 shows the preliminary results of sampling with the HVS3. Initial carpet dust
loading in building A was 1.3 g/m and the tile dust loading was 0.1 g/m2. The carpet dust
level in building B was higher at 6.2 g/m2.
Fungi and Bacteria
Figure 2 shows preliminary results of fungi in dusts collected with the HVS3. Fungi
concentrations in building A were 5.0 x IfJ5 CFU/g in tile floor dust and 1.2 x 10s CFU/g in
carpet dust. The fungal concentration in building B carpet dust was higher at 3.2 x 10"
CFU/g carpet dust in comparison.
Preliminary mesophilic bacteria concentrations in building A were found to be
7.0 x 104 CFU/g in tile floor dust and 8.8 x 106 CFU/g in carpet dust, while thermophilic
concentrations were found to be 5.0 x 102 CFU/g in tile floor dust and 7.0 x 103 CFU/g in
carpet dust. The microbiological composition of the surface dusts from the buildings is
similar to that of 'ordinary* outdoor dirt. Microbial concentrations have initially been found to
be higher in carpet floor dust than in tile dust.
Endotoxin and Mite Guanine
Dusts in building A had insignificant levels of endotoxin. Figure 3 shows endotoxin
concentrations found in surface dusts. Endotoxin levels found in dusts from building A
ranged from 0.2 to 6 u.g/g of floor dust. Endotoxin levels were lower in the tile floor dust than
the carpet floor dust. Carpet from building B with a higher dust loading was found to have
elevated levels of endotoxin at 103 ug/g. This level is considered to be elevated above
expected levels found in carpet dust. Inhalation of a dose of 20 ug of endotoxin has been
shown to induce a bronchial obstructive response in asthmatic subjects 5.
Preliminary results showed that dust mite guanine was not detected in any of the dust
samples.
85
-------
_
BldgA BldgA
Caipet Tile
BWgB
Carpet
Flgura 1. Surface Dust Loading (g/m1)
BldgA BWgA BUgB
Carpet Tile Caipet
Flgura 2. Fungi Concentrations (CFU/g of dust)
100?
Figured. Endotoxln In Surface Dust
86
-------
CONCLUSIONS
Initial results indicate that the HVS3 can be used to effectively collect dust from carpeted
and non-carpeted surfaces for a variety of microbiological analyses. The unit can be used to
assess the dust loading and levels of microbiological contamination on floor surfaces.
Further surface monitoring with the unit, along with concurrent bioaerosol monitoring, may
yield information on the relationship between the microbial ecology of indoor surfaces and the
microbiological contamination of indoor air.
REFERENCES
1. J. W. Roberts, et al. 'Development and field testing of a high volume surface sampler for
pesticides and toxics in dust," Jour, of Expos. Analysis and Environ. Epidem.. 1(2):143-155
(1991).
2. V. R. Stamper, et al. Development of a High Volume Small Surface Sampler (or
Pesticides and Toxics in House Dusts. EPA Work Assignment No. 11-71, draft final report,
U.S. Environmental Protection Agency, Research Triangle Park, 1990.
3. CS3, INC. High Volume Small Surface Sampler Operation Manual. Serial Number 1812.
March 16, 1992.
4. Recent unpublished results using ASTM Method F-608 indicate that this figure is 70%.
5. O. Michel, R. Ginanni, B. Le Bon, J. Content, J. Duchateau, and R. Sergysels.
Inflammatory response to acute inhalation of endotoxin in asthmatic patients. Am. Rev.
Respir. Pis.. 146:352-357 (1992).
ACKNOWLEDGEMENT
The research presented in this document has been funded by the United States
Environmental Protection Agency under contract number CR-815509-02-0 to the Research
Triangle Institute.
DISCLAIMER
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use by the U.S. Environmental Protection Agency or Research Triangle
Institute (RTI). The views expressed in this document are those of the authors and do not
represent official RTI or Agency policy.
87
-------
Methods for Polycyclic Aromatic Hydrocarbons
and Tobacco Smoke Markers in House Dust: Laboratory and Field Evaluation
Jane C. Chuang, Patrick J. Callahan, and Sydney M. Gordon
Battelle
505 King Avenue
Columbus, Ohio 43201
Nancy K. Wilson and Robert G. Lewis
U.S. Environmental Protection Agency
MD-44
Research Triangle Park, North Carolina 27711
ABSTRACT
Analytical methods were validated to determine polycyclic aromatic hydrocarbons (PAH)
and other semivolatile organic compounds in house dust. We also examined the storage stability of
three potential markers (solanesol, nicotine, and cotinine) for particulate-phase environmental
tobacco smoke (ETS) in house dust. The results showed that less than 10 percent of the spiked
solanesol was recovered from the dust after storage for 7 days in a typical indoor environment
(room temperature and indoor lighting). Under the same storage conditions, after 21 days of
storage, more than 90 percent of the cotinine and approximately 40 percent of the nicotine were
recovered. These findings suggest that cotinine is a better marker for ETS particles in house dust
than nicotine, whereas solanesol is not a suitable marker.
A small field study was conducted to evaluate the role of smokers in the house on the levels
of PAH, cotinine, and nicotine present in house dust. Samples were collected from eight houses,
using the High Volume Small Surface Sampler (HVS3) to collect carpet-embedded dust from
designated areas in the carpet in either the living room or family room of each house. After
collection, the dust samples were separated into fine (< 150 /tm) and coarse fractions. Most of the
dust loading (69-85 percent) was found to be distributed in the fine fraction. The fine dust samples
were analyzed for PAH, nicotine, and cotinine.
INTRODUCTION
Recent studies have shown that significant levels of household pesticides occur in house
dust.' Residues of the pesticides deposited on surfaces and contained in house dust or soil may be
picked up by the skin on contact. Once on the skin, the pesticides may be absorbed directly or
transferred to the mouth and ingested. Because small children spend a great deal of time on the
floor, they are particularly susceptible to exposure to these compounds as a result of dermal contact
with house dust and the frequent hand-to-mouth contact that accompanies their normal play
activities.
Polynuclear aromatic hydrocarbons (PAH) and their derivatives represent another important
group of semivolatile organic compounds (SVOC) that have been found in indoor air2'3'4. Although
PAH have been identified in street dust samples,5 little, if any, information is available about their
occurrence and distribution in house dust. Many of these compounds are known to be carcinogenic
or mutagenic to man, and a wide array of adverse health effects has been linked to exposure to
-------
PAH.4'6 The presence of environmental tobacco smoke (ETS) has been shown to be an important
indoor source of PAH and PAH derivatives.2'3
Note that the particulate portion of ETS may accumulate onto the house dust and contribute
to PAH levels in house dust. It is desirable to identify a suitable marker for particulate phase ETS
in house dust. There are three potential markers for particulate phase ETS, nicotine, cotinine, and
solanesol. Nicotine is a dominant component of ETS, and has been used as a marker for ETS, but
it is primarily in the vapor phase.7 Cotinine can be formed through oxidation of nicotine. Solanesol
has been suggested to be a suitable marker for particulate-phase ETS, but this compound has also
been reported to decompose under UV light, and under the temperatures of an automatic gas
chromatography (GC) injector.8'9 Studies are needed to investigate the analytical method for the
determination of these potential ETS markers (nicotine, cotinine, and solanesol) in house dust
samples, and to determine the storage stability of these compounds for selecting a suitable marker
compound for ETS particles in house dust. In order to make an initial assessment of the relative
importance of house dust to overall indoor exposure to PAH, a reliable analytical method is also
needed to determine PAH and PAH derivatives in house dust.
In this paper we discuss the storage stability for the potential markers (solanesol, nicotine,
and cotinine) for particulate-phase ETS in house dust, and the concentration profiles of PAH,
nicotine, and cotinine in house dust samples from a small field study.
EXPERIMENTAL PROCEDURE
Storage Stability Study
In the storage stability study, eight aliquots (200 mg each) of a smoker's fine house dust
particles (< 150 /im) were used. Each of the four aliquots of house dust were spiked with a known
amount of solanesol and each of the other four aliquots were spiked with a known amount of
nicotine and cotinine. The spiked samples were stored at room temperature, under an indoor
lighting environment for 0 day, 7 days, 14 days, and 21 days. The samples were then analyzed for
solanesol, nicotine, and cotinine after the designated storage time. The analytical method used for
the determination of solanesol consists of extracting the dust sample with dichloromethane (DCM),
derivatizing the DCM extract with N,O-Bis(trimethylsilyi)trifluoroacetamide, and analyzing the
sample extract by on-column injection, gas chromatography/flame ionization detection (GC/FID).10
Analysis of nicotine and cotinine in house dust samples consists of extracting the dust sample with
acidic water (pH = 1.2), and analyzing the extracts by a Trace Atmospheric Gas Analyzer (TAGA)
interfaced with a Battelle-developed non-aerosol vaporization device.10 The analytical methods used
to determine solanesol, nicotine, and cotinine in house dust were validated by determining recoveries
of these target compounds in the spiked house dust samples.
A Small Pilot Field Study
The study was performed at eight homes in Columbus, Ohio. Homes were selected on the
basis of either the presence or absence of ETS in the home. Four homes with smokers and four
homes with nonsmokers were included. The High Volume Small Surface Sampler (HVS3) was used
to collect house dust samples." The HVS3 unit was operated following the manufacturer's
instructions,12 and a draft ASTM standard method.13 The sampling procedure was based on that
followed in the House Dust/Infant Pesticides Exposure Study (HIPES). The collected dust samples
were then sieved into coarse and fine (< 150 jim) fractions. The fine fractions were used for
subsequent analyses for PAH, cotinine, and nicotine. The analytical method for nicotine and
cotinine described in the storage stability study cannot provide adequate detection sensitivity for
89
-------
cotinine in the dust samples. Thus, an alternative method was used, consisting of extracting dust
samples with DCM followed by methanol (MEOH), cleaning up the combined DCM and MEOH
extracts by liquid-liquid partitioning, and analyzing the base fraction by GC/MS. Another aliquot of
each dust sample was extracted with DCM, fractionated by a silica gel column, and analyzed by
GC/MS to determine PAH.10
RESULTS
The recovery data of solanesol spiked onto the smoker's and nonsmoker's house dust
samples were 96 and 92 percent, respectively. This finding indicated that two sequential extractions
with DCM for 20 min with an ultrasonic bath can quantitatively remove spiked solanesol from the
house dust sample matrix. The recovery data of nicotine and cotinine from the spiked nonsmoker's
house, using extraction with acidic water and analyzing by TAG A, were 100 and 65 percent,
respectively. The alternative method consisting of extracting, liquid-liquid partitioning, and GC/MS
analysis also provided quantitative recoveries of nicotine (100 percent) and cotinine (89 percent)
from the spiked house dust sample. Quantitative recoveries of PAH were also obtained from the
spiked house dust sample, and the recoveries ranged from 70 percent of cyclopenta[c,d]pyrene to
100 percent of phenanthrene.
In the storage stability study, we found that solanesol spiked onto dust samples was unstable
and the recovery of the spiked solanesol was only 6.2 percent after 7 days of storage in a typical
indoor environment (room temperature and indoor lighting). Thus, solanesol is not a suitable
marker for the indication of aged particulate-phase ETS in house dust samples. Under the same
storage conditions, the spiked nicotine decreased to about 30 percent of its original value after 21
days of storage, but the spiked cotinine was found to be stable after 21 days of storage. The above
results suggested that cotinine is a suitable marker for ETS particles in house dust.
In the small pilot field study, most of the dust loading (69-85 percent) was found to be
distributed in the fine fraction (< 150 /tm). Table 1 summarizes the measured nicotine and cotinine
results for the fine dust samples. Higher concentrations of both nicotine and cotinine were found in
fine dust samples from smokers' homes as compared to those from nonsmokers' homes. The
highest levels of nicotine and cotinine were found in the smoker's home #H05DS and the lowest
levels of nicotine and cotinine were found in the nonsmoker's home #H07DN. These findings were
in agreement with the storage stability study results described before suggesting that cotinine is a
good marker for ETS particles in house dust. Table 2 summarizes the PAH concentration data from
the small pilot field study. The PAH concentrations ranged from 0.01 ppm (naphthalene) to 90 ppm
(fluoranthene). Note that levels of most carcinogenic 5- and 6-ring PAH in dust are higher than
those of most noncarcinogenic 2- and 3-ring PAH in dust. The known carcinogen benzo[a]pyrene
(BaP) ranged in concentration from 1.8 - 53.8 ppm. The highest PAH concentrations were found in
house dust samples from a nonsmoker's home (#H08DN). That house is within a quarter mile of a
freeway and road construction was performed during the sampling period. These factors may have
contributed to PAH levels. The levels of most PAH in home number H07DN also exceeded those
in the smoker's homes. Therefore, the presence of ETS is not the only important factor to
contribute to the PAH levels in house dust.
CONCLUSIONS
The following conclusions can be drawn from this study:
1. Solanesol is not a suitable marker for aged ETS particles in house dust.
90
-------
2. Higher levels of nicotine and cotinine were found in the smokers' homes as compared
to nonsmokers' homes. However, cotinine appears to be a better marker than
nicotine for aged ETS particles in house dust because cotinine is more stable than
nicotine in the house dust matrix.
3. The presence of ETS is not the only significant factor to contribute to the PAH levels
in house dust samples. Higher PAH levels were found in two nonsmokers' homes
compared with house dust from all four smokers' homes.
DISCLAIMER
The information in this document has been funded wholly or in part by the United States
Environmental Protection Agency under Contract 68-DO-0007 to Battelle Memorial Institute. It has
been subject to the Agency's peer and administrative review, and it has been approved for
publication as an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
REFERENCES
(1) Fortmann, R.C., Sheldon, L.S., Smith, D., Perritt, K., and Camann, D.E. House
dust/infant pesticides exposure study (HIPES), Research Triangle Institute. Final Report
Contract No. 68-02-4544, Work Assignment No. 111-75, U.S. Environmental Protection
Agency, 1991.
(2) Lioy, P.J., Waldman, J.M., Greenberg, A., Harkov, R., and Pietarinen, C. The total
human environmental exposure study (THEES) to benzo[a]pyrene: Comparison of the
inhalation and food pathways. Arch. Environ. Health, 43, 304-312, 1988.
(3) Chuang, J.C., Mack, G.A., Kuhlman, M.R., and Wilson, N.K. Polycyclic aromatic
hydrocarbons and their derivatives in indoor and outdoor air in an eight-home study.
Atmos. Environ., 25B(3), 369-380, 1991.
(4) Chuang, J.C., Wise, S.A., Cao, S., and Mumford, J.L. Chemical characterization of
mutagenic fractions of particles from indoor coal combustion: A study of lung cancer in
Xuan Wei, China. Environ. Sci. Technol., 26(5), 999-1004, 1992.
(5) Takada, H., Onda, T., Harada, M., and Ogura, N. Distribution and sources of polycyclic
aromatic hydrocarbons (PAHs) in street dust from the Tokyo metropolitan area. Sci. Total
Environ., 107, 45-69, 1991.
(6) Paniculate Polycyclic Organic Matter. National Academy of Sciences, Washington, D.C.,
1972.
(7) Eatough, D.J., Benner, C.L., Bayona, J.M., Caka, P.M., Richards, G., Lamb, J.D., Lee,
M.L., Lewis, E.A., and Hansen, L.D. Chemical composition of environmental tobacco
smoke, 1. Gas-phase acids and bases. Environ. Sci. Technol., 23(6), 679-687, 1989.
(8) Tang, H., Richards, G., Benner, C.L., Tuominen, J.P., Lee, M.L., Lewis, E.A., Hansen,
L.D., and Eatough, D.J. Solanesol: A tracer for environmental tobacco smoke particles.
Environ. Sci. Technol., 24(6), 848-852, 1990.
91
-------
(9) Ogden, M.W. and Maiolo, K.C. Comparison of GC and LC for determining solanesol in
environmental tobacco smoke. LC-GC, 10(6), 459-462.
(10) Chuang, J.C., and Gordon, S.M., Determination of solanesol, polycyclic aromatic
hydrocarbons (PAH), PAH derivatives, nicotine, and cotinine in house dust samples. Final
report, U.S. Environmental Protection Agency, Contract No. 68-DO-0007, Work
Assignment 13, 1992.
(11) Roberts, J.W., Budd, W.T., Ruby, M.G., Bond, A.E., Lewis, R.G., Wiener, R.W., and
Camann, D.E. Development and field testing of a high volume sampler for pesticides and
toxics in dust. J. Exposure Anal. Environ. Epidemiol., 1, 143-155, 1991.
(12) High Volume Small Surface Sampler HVS3: Operation Manual. Cascade Sampling Systems
(CS3), Inc., Bend, Oregon, January 13, 1992.
(13) Standard Practice for Collection of Dust from Carpeted Floors for Chemical Analysis.
Draft Standard Method, ASTM, Philadelphia, Pennsylvania, 1992.
Table 1. Nicotine and cotinine concentrations in fine house dust samples
House Code(a'
H01DS
H02DN
H03DN
H04DS
H05DS
H06DS
H07DN
H08DN
Nicotine
90
5.5
4.7
44
430
140
2.2
3.6
Concentration, ppm
Cotinine
0.90
0.31
0.30
5.1
7.1
1.5
0.17
0.20
(a) In H01DS, HOI refers to the first house sampled, D refers to the dust sample, and S refers
to a smoker's house. In H02DN, H02 refers to the second house sampled, D refers to the
dust sample, and N refers to a nonsmoker's house.
92
-------
Table 2. PAH concentrations in house dust samples
Compound
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Retene
Fluoranthene
Pyrene
Benz(a) anthracene
Chrysene
Cyclopenta(c,d)pyrene
Benzofluoranthenes
Benzo(e)pyrene
Benzo(a)pyrene
lndeno(1 ,2,3-c,d)pyrene
Dibenzo(a,h)anthracene
Benzo(g,h,i)perylene
Coronene
H01DS
pg/g*
0.01
0.20
0.28
0.48
6.81
0.73
0.07
12.4
9.39
3.31
5.71
0.21
10.06
3.60
12.7
5.52
1.81
4.57
2.15
H02DN
/jg/g
0.16
0.12
0.09
0.15
2.12
0.25
0.15
3.73
2.87
1.29
2.18
0.17
3.70
1.65
21.1
2.71
0.96
2.36
1.43
H03DN
pg/g
0.09
0.04
0.07
0.11
2.13
0.16
0.18
3.21
2.49
1.04
1.61
0.04
3.52
1.60
1.80
1.79
0.67
1.54
0.70
H04DS
pg/g
0.24
0.04
0.05
0.08
0.98
0.11
0.10
1.77
1.31
0.48
1.00
0.06
1.64
0.70
11.0
1.27
0.37
1.03
0.50
H05DS
pg/g
<0.01
0.02
0.05
0.09
1.10
0.12
0.06
2.24
1.48
0.70
1.14
0.05
2.56
0.25
3.94
0.91
0.21
0.72
0.21
H06DS
pg/g
0.23
0.36
0.15
0.55
6.32
0.69
< 0.01
10.7
8.00
4.08
6.15
0.19
5.64
2.40
1.97
4.09
1.83
3.20
2.05
H07DN
/jg/g
0.06
0.10
0.24
0.46
7.52
1.05
0.23
14.7
11.4
5.07
7.17
0.23
15.4
6.58
7.71
6.86
2.10
5.33
1.89
H08DN
^g/g
0.35
0.52
1.06
2.07
41.0
3.91
0.36
90.2
68.2
24.4
34.4
0.36
103
40.7
53.8
40.8
7.49
34.9
7.22
jtg/g = ppm
-------
SAMPLING, SUPERCRITICAL FLUID EXTRACTION (SFE) AND GC/MS ANALYSES OF INDOOR AIR
SEMI-VOLATILE TOXIC ORGANICS
V.M.Kanagasabapathy', R.W.Bell,P.Yang*, L.Allan*, L.Au*, J.Parmar*, M.A.Lusis, and
R. E. Chapman
Air Resources Branch ('Laboratory Services Branch) Ontario Ministry of the
Environment, 125 Resources Road, Etobicoke, Ontario M9P 3V6
ABSTRACT
People spend most of their time within enclosed air spaces; in indoor
environments most air pollutants are present in higher concentrations than in
outdoor environments. Consequently, indoor air quality should be of greater
concern with respect to respiratory exposure to harmful toxic substances.
Therefore, a method has been developed to sample indoor airborne
carcinogenic polynuclear aromatic hydrocarbons (PAHs) in such a way to extract
them by environmentally safe SFE with less sample handling.
INTRODUCTION
SFE offers cleaner extracts, less sample handling, and equivalent or better
recoveries to conventional technologies. It is cost effective, time efficient and
low in solvent waste generation.
But, supercritical COZ as such does not have the ability to quantitatively
extract PAHs from most of the environmental matrices. Modifying the fluid might
enhance the extraction efficiencies but the performance of the method is highly
matrix-dependent1"3. Therefore, additional developmental work needs to be done.
This work describes method optimization for the extraction of PAHs from an
environmental matrix, XAD-2 (styrene-divinylbenzene polymer) and also sampling
indoor airborne PAHs on teflon impinged glass fibre filter and XAD-2 in such a
way to extract them employing SFE.
Experimental
Sample preparation, extraction, and GC/MS analysis:
Spiking was done by injecting 0.500 mL toluene solution of PAHs standard
into XAD-2. The samples were housed in a glass cylindrical cartridge {13 mm o.d;
11 mm i.d; 51 mm length. An indentation at the lower end of the cylinder provided
a rim to support a 200/200 mesh stainless steel screen that held the sorbent bed.
Atop the sorbent bed, a layer of glass wool and another similar screen were
placed to hold it in place. After spiking, the solvent was allowed to evaporate
on its own.
Extractions were done in a SFX 2-10 extractor with a 260D syringe pump
(Isco, Inc., Lincoln, NE) employing SFC grade CO2 {Airco Special Gases, Riverton,
NJ) . The sampling cartridge was inserted into a stainless steel cylindrical
extraction module (without significant empty space) with an endcap and a filter
element at the bottom. Prior to extraction a modifier (all HPLC grade solvents)
was injected into the spiked XAD-2. Immediately, the upper endcap with a filter
was placed onto the extraction module, hand tightened and the extraction sequence
initiated. Initially the extraction was done with no flow of fluid (static
extraction), which allowed the sample matrix to be steeped in the fluid and the
added modifier. Dynamic extraction (i.e., with continuous flow of the fluid) was
then conducted to collect the analytes. Flow rates were controlled (measured as
liquid fluid at the pump) using 50 urn i.d. fused silica tubing as an outlet
restrictor. The reported flow rate was computed by dividing the total volume of
fluid used, by the duration of dynamic extraction. The extracts were collected
by inserting the outlet end of the restrictor into a culture tube (13 mm i.d.;.
178 mm long) containing 6 mL methylene chloride. To avoid the cooling of the
solvent caused by the expansion of the fluid, the collection system was kept in
a bath of methanol. Because of high gas-flow (especially at higher fluid flow
rates), there was considerable evaporation of methylene chloride. To compensate
the loss, additional methylene chloride (nearly 12 mL) was added in batches into
the collection tube to keep its volume always around 6 ml. The extracts were then
concentrated to 0.5 mL under a stream of pure nitrogen at 45-50°C.
GC/MS analyses were performed, after adding d12-perylene as an internal
94
-------
standard, with a Hewlett-Packard model 5890 GC equipped with a JSW DB-5 (60 m,
0.25 urn film thickness, 250 um i.d.(column, a HP 5970 mass selective detector and
a HP 7673A auto sampler. Selected ion monitoring mode quantitations were based
on the integration of extracted ion chromatogram of the molecular ion of each
PAH.
Results and Discussion
Optimizing SFE conditions:
Effect of modifier:
Of all the modifiers tested (methanol, i-propanol, 2-
methoxyethanol,acetone, chloroform, methylene chloride, toluene, chlorobenzene,
acetonitrile, N,N-dimethylformamide), chlorobenzene showed the highest extraction
efficiency of lower- as well as higher-molecular-weight PAHs. The energy barrier
of desorption between the analyte and the adsorbent can be reduced by selective
interaction of the aromatic ring in the modifier with the matrix-solute complex
since XAD-2 also contains aromatic rings in its structure, as visualized by
Pawliszyn et al*, which explains our result. But, the reason for chlorobenzene
being better than even toluene might be due to very close polarity match between
chlorobenzene and the PAHs. The polarity of the mobile phase mixture must be
optimized to match that of the analyte (s) to attain maximum extractability5.
Methylene chloride, the best solvent for PAHs, was not as good as chlorobenzene,
which demonstrates that the extractability is governed by solute/matrix/fluid
interactions, rather than by solute/fluid interactions.
The results shown in Table 1 indicate that the extractability is governed
by concentration of the modifier too. A uniform increase in extraction efficiency
was evident on progressively increasing the volume of chlorobenzene from 0.5 to
1.2 ml.
Effect of temperature and pressure:
The analyte has to diffuse out of the pores of the matrix into the carbon
dioxide stream before being transported by the bulk fluid out of the extraction
module into the collection vessel1, which could be achieved easily at high
temperatures by decreasing the forces between the analyte and the matrix. In
fact, use of high temperatures has rewarded (Table 1) possibly because the vapour
pressures of analytes increase with temperature. Increasing the temperature from
50° to 200°C, with pure CO2 extractions resulted in two-to six-fold increase in
extraction efficiencies of PAHs from air particulate matter6. But, in our
experiments increasing it beyond 100°C ended in either no effect at all, or
decreased recoveries (Table 1}. Probably decreasing the fluid density too much
(at constant pressure) which happens on increasing the temperature, might reduce
the fluid's solubilizing capacity.
Higher pressures like 450 atm were necessary for the complete recovery of
the PAHs, especially the four and five ring PAHs from matrices like soil7 This
is because raising the extraction pressure leads to higher fluid density which
increases the solubility of the analytes. That is why, most of our extractions
were done at 450 atm (Table 1-4) .
Effect of extraction time and fluid flow rate fluid volume:
The recovery is affected most by extraction time and extraction pressure.
The highest recoveries of PAHs from standard reference soil and marine sediment
were achieved1 at flow rates in excess of about 2 mL.min"1 (as compressed fluid)
and at 90°C. Therefore, simply by exposing the sample to more fluid by increasing
the extraction flow rate, quantitative recoveries of PAHs from environmental
matrices might be possible. However, increasing the flow rate to such a high
level may make the analyte collection more difficult. Moreover, quantitative
collection of relatively volatile analytes is convenient only with flow rates of
up to at least 1 mL.min"1. Obviously, the alternative to maximize recovery is to
lengthen the extraction time. The extraction of PAHs from a railroad bed soil
showed virtually no dependence on the flow rate6, provided it is within 0.3-0.9
mL.min"1. To achieve highly efficient extractions, 3 to 5 void volumes of fluid
must be flushed through the charged extraction vessel".
As shown in Table 1-4, 65 min-long extraction flushing 4 to 5 void volumes
of fluid through the charged extraction vessel helped us to accomplish efficient
extractions. However, to validate our method two sequential extractions of the
same sample was performed; the first fraction being collected during the first
65 min of dynamic extraction, following which the second fraction was collected
95
-------
for further 25 min under the same extraction conditions as those employed for the
first fraction; on anlyzing the fractions separately, nothing was detected from
each one of the second fractions of six replicate experiments (Table 3).
Because of unknown resons, lengthening the static extraction time from 25
to 45 min and increasing the pressure from 450 to 475 atm caused reduction in
recovery in most of the thirty PAHs spiked (Table 3, column # 1 versus I 3
values).
Effect of spiked amount and moisture content of the material:
The precision in the experimental technique includes the precision of the
entire system involving the S'FE, the off-line collection, and the GC/MS analysis.
The precision became poor on reducing the spiked amount (Table 2).
As contemplated by Lopez-Avila et al1, a certain amount of water (10%) in
the sample was observed to be beneficial (Table 4, data in column # 1 versus
either it 2 or 3) especially in terms of extractability of higher-molecular-weight
PAHs. The extractability of higher-molecular-weight PAHs was far better when 10*
water was initially present in the matrix before the addition of the modifier
(data in column # 2 versus # 3) . Saturation of the solid matrix phase with the
entrainers or modifiers like methanol and water, is crucial to obtain high
recoveries of diuron from soil9; is the water modifying the solvent or modifying
the matrix?
Sampling;:
Sampling performed for continuous 24 h periods in a clean room indicated an
absence of breakthrough of spiked compounds, but losses of some spiked compounds,
attributed to reaction were detected (Table 4; data in column #4).
CONCLUSION
A method has been developed to sample, extract, and analyze indoor airborne PAHs.
Acknowledgements
This work is in support of RAC Grant PDF06 and the support of the Research and
Technology Branch is gratefully acknowledged. We are grateful to Mr.E.W.Piche,
Mr.R.Pearson,Prof .R.A.McClelland, Mr.D.Schneeberger and all of our colleagues for
their support.
References
1. V. Lopez-Avila, N.S. Dodhiwala, J.Benedicto and W.F. Beckert,
Supercritical Fluid Extraction of Organic Compounds from Standard Refernce
Materials, EPA-600-X-91-149, U.S. Environmental Protection Agency, Las Vegas,
1991.
2. C.R. Knipe, D.R. Gere and M.E.P. McNally, "Supercritical Fluid Extraction-
Developing a Turnkey Method, " in ACS Symposium Series No. 486. American Chemical
Society, Washington, DC 1992, Chapter 18.
3. S.B.Hawthorne, J. J.Langenfeld, D.J.Miller and M.D.Burford, "Comparison of
Supercritical CHC1F2, N2O, and CO2 for the Extraction of Polychlorinated Biphenyls
and Polycylic Aromatic Hydrocarbons."Anal.Chem.64: 1614 (1992).
4. Z.Miao, M.Yang, J.Pawliszyn et al,"Supercritical Fluid Extraction with
Simultaneous Class Fractionation of PCBs and PAHs from Adsorbent Materials for
Air Pollution Determinations," in Proceedings-Environmental Research: Technology
Transfer Conference. Ontario Ministry of environment. Toronto, 1991, pp.672-675.
5. M.E.P.McNally and J.R.Wheeler, "Increasing Extraction Efficiency in
Supercritical Fluid Extraction from Complex Matrices," J. Chromatogr. 447: 53
(1988) .
6. S.B.Hawthorne, University of North Dakota, Grand Forks, ND, personal
communication, 1993.
7. J.M.Levy, "Important Factors in Enhancing Supercritical Fluid Extraction
Efficiencies for Environmental Applications" in Proceedings-Environmental
Research: Technology Transfer Conference, Ontario Ministry of Environment,
Toronto, 1992, Paper No.D2.
8. J.M.Levy, "Anvances in analytical SFE," American Laboratory (8): 25 (1991).
9. M.E.P.McNally, C.M.Deardorf f and T.M.Fahmy, "Supercritical Fluid Extraction, "
in ACS Symposium Series No. 488. American Chemical Society, Washington, DC 1992,
Chapter 12.
96
-------
Table 1
Extraction Conditions*
d,0-Acenaphthene
d10-Anthracene
d,0-Pyrene
d12-Chrysene
d(2-Benzo [ a ] pyr ene
450 atm
100C
PhCl:
0.500 mL
t 1
69
71
64
60
54
450 atm
100C
PhCl:
1.000 mL
t 2
74/7
70/6
TO/6
72/5
68/5
450 atm
100C
PhCl:
1.200 mL
450 atm
115°C
PhCl:
1.000 mL
t 3
93
94
94
94
83
% Recovery
f 4
73/0
70/1
73/2
73/2
68/6
450 atm
115°C
PhCl:
1.200 nL
t 5
73/4
70/5
69/8
74/10
71/9
Supercritical fluid : COj; modifier : PhCl; matrix : XAD-2 (1.5 g); volume of
extraction cell (vertically kept) : 10 mL; volume of fluid employed : 52-67 mL;
fluid flow rate : 0.59-0.78 mL.min'1; static : 15 min; dynamic : 86 min; amount
spiked : 0.8005-0.8350 ug.
The values under the vertical column # 1 & 3 are from single determination while
those under # 2 & 4 are from three determinations and those under t 5 are from
two determinations.
Routine extractions at 115 C are impractical due to separation of extraction-
cartridge's filter from its joint.
Table 2
Extraction Conditions
Amount
spiked
450 atm
100°C
Static:
15 min
Dynamic:
86 min
450 atm
100°C
Static:
15 min
Dynamic:
65 min
450 atm
100°C
Static:
25 min
Dynamic:
65 min
dlo-Acenaphthene
d,0-Anthracene
d,0-Pyrene
d,2-Chrysene
dI2-Benzo [ a ] pyrene
Acenaphthene
Anthracene
Pyrene
Chrysene
Benzo [ a ] pyrene
0.8310
0.8010
0.8010
0.8005
0.8350
0.2818
0.3082
0.4862
0.5025
0.6677
92.7
93.6"
93.6]
93.7"
82.6"
% Recovery
94.3/1.5
89.2/2.3
89.2/3.7
82.3/4.7
71.7/6.2
93.5/2.8
83.9/5.1
79.2/5.1
76.0/5.0
58.4/6.0
Supercritical fluid : CO2; modifier : PhCl (1.200 mL); matrix : XAD-2 (1.5 g);
volume of extraction cell (vertically kept) : 10 mL; volume of fluid employed :
37-66 mL; fluid flow rate : 0.57-0.99 mL.min"'.
The numbers under the dividing bar represent standard deviations which are for
six replicate extractions at each condition.
These values are from only one extraction.
97
-------
Table 3
2-Methylnaphthalene
1-Methylnaphthalene
2-Chloronaphthalene
Biphenyl
Acenaphthalene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
o-Terphenyl
1-Methylphenanthrene
Fluoranthene
Pyrene
9,10-Dimethylanthracene
rn-Terphenyl
p-Terphenyl
Benzo[a]fluorene
Benzo[b]fluorene
Benz[a]anthracene
Chrysene
Benzo[b]fluoranthene
Benzo[k]fluoranthene
Benzo [ e ] pyrene
Benzo[a]pyrene
Perylene
3-Methylcholanthrene
Indeno [1,2,3-cd ] pyrene
Dibenz [ ac/ah ] anthracene
Benzo[ghi]perylene
t 1
Extraction Conditions
Amount
spiked
(ug)
0.3305
0.3607
0.3237
0.3105
0.2597
0.2818
0.3325
0.3010
0.3082
0.3252
0.4150
0.4165
0.4862
0.3987
0.3082
0.4050
0.6450
0.4060
0.6435
0.5025
0.4030
0.4675
0.3752
0.6677
0.3077
0.3850
0.3457
0.4185
0.4722
450 atm
100C
Static:
25 min
Dynamic :
65 min
450 atm
100C
Static:
30 min
Dynamic:
65 min
475 atm
100°C
Static:
45 min
Dynamic :
65 min
% Recovery*
t 2
f 3
104.9/5.1
94.7/3.4
97.8/4.1
100.9/4.3
90.5/1.9
93.5/2.8
94.7/3.5
93.6/2.8
83.9/5.1
96.3/1.4
92.4/3.8
86.0/3.7
79.2/5.1
34.3/46.5
93.0/5.8
93.0/4.7
91.2/8.5
77.2/9.9
71.7/7.9
76.0/5.0
78.6/19.6
62. 0/2. 9
56.4/7.6
58.4/6.0
53.1/3.3
55.9/11.3
28.0/18.6
41.4/14.1
44.8/5.3
97.3/4.7
90.1/5.6
94.8/2.8
98.2/2.1
84.1/4.7
86.4/3.6
89.7/3.7
82.0/3.4
81.7/3.8
87.1/2.1
77.9/2.4
75.2/4.0
62.7/2.8
76.9/10.0
82.8/3.2
93.4/9.3
82.8/11.2
70.2/4.5
57.8/3.5
51.1/2.8
51.7/6.4
47.1/2.9
42.2/5.0
38.7/8.2
49.3/8.1
0.0/0.0
33.8/12.9
33.1/13.9
32.1/12.4
85.3/5.6
78.2/4.0
127.9/6.4
114.0/7.5
94.7/3.4
95.8/7.1
92.6/5.5
81.1/5.1
74.6/5.2
93.5/5.2
83.4/8.5
71.0/5.6
66.2/6.1
62.2/24.9
83.7/7.0
74.6/3.5
76.3/9.7
72.4/6.1
55.3/5.7
60.5/5.4
43.2/6.3
59.9/10.3
34.1/1.7
45.2/4.4
18.2/21.8
46.2/5.6
38.8/6.0
44.4/7.4
45.3/11.2
Supercritical fluid : CO2; modifier : Phcl (1.200 mL); matrix : XAD-2 (1.5 g);
volume of extraction cell (vertically kept) : 10 mL; volume of fluid employed :
37-59 mL; fluid flow rate : 0.57-0.90 mL.min .
& The numbers under the dividing bar represent standard deviations which are for
five to six replicate extractions at each condition.
Column f 2 values represent the first fraction, collected during the first 65 min
of dynamic extraction, following which a second fraction was collected for
further 25 min in each one of the six replicate extractions under the same
extraction conditions as those employed for the corresponding first fractions;
on analyzing nothing was detected from each one of the second fractions.
98
-------
Table 4
Extraction Conditions
475 atm
100°C
Static:
45 min
Dynamic :
65 min
f 1
85.3/5.6
78.2/4.0
127.9/6.4
114.0/7.5
94.7/3.4
95.8/7.1
92.6/5.5
81.1/5.1
74.6/5.2
93.5/5.2
83.4/8.5
71.0/5.6
66.2/6.1
62.2/24.9
83.7/7.0
74.6/3.5
76.3/9.7
72.4/6.1
55.3/5.7
60.5/5.4
43.2/6.3
59.9/10.3
34.1/1.7
45.2/4.4
18.2/21.8
46.2/5.6
38.8/6.0
44.4/7.4
45.3/11.2
475 atm
100°C
Static:
45 min
Dynamic:
65 min
%
f 2
85.6
86.8
80.3
88.6
89.7
86.6
82.4
85.4
81.8
89.2
93.0
88.4
80.0
125.1
91.2
93.1
87.6
88.9
83.8
79.6
76.2
74.7
70.6
73.2
67.9
81.6
70.9
70.0
62.7
475 atm
100°C
Static:
45 min
Dynamic:
65 min
Recovery*
f 3
87.6/2.8
85.3/3.2
80.4/3.1
87.1/2.5
88.1/1.2
86.7/3.0
80.2/2.3
86.1/3.7
84.3/3.2
87.0/3.2
91.3/2.7
84.1/4.8
71.2/4.5
123.1/6.7
89.6/3.2
90.4/2.4
85.0/2.5
87.0/3.2
70.6/7.7
65.3/7.5
57.8/10.0
58.0/9.7
50.9/10.2
53.7/10.2
47.7/12.1
64.7/10.3
49.1/10.3
49.4/10.0
42.1/9.6
450 atm
100°C
Static:
25-45 min
Dynamic :
65 min
t 4
>100/0
>100/0
>100/0
>100/0
50.1/5.1
92.5/9.6
76.2/6.7
78.6/3.0
0.0/0.0
66.6/8.9
28.9/1.6
8.0/10.7
6.2/8.2
0.0/0.0
0.0/0.0
0.0/0.0
3.6/4.8
0.0/0.0
24.9/4.2
48.4/3.5
77.2/15.2
68.5/2.8
69.3/3.5
18.0/12.0
21.7/14.4
0.0/0.0
56.9/14.8
42.2/16.5
69.2/6.1
2-Hethylnaphthalene
1-Methylnaphthalene
2-Chloronaphthalene
Biphenyl
Acenaphthalene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
o-Terphenyl
1-Methylphenanthrene
Fluoranthene
Pyrene
9,10-Dimethylanthracene
m-Terphenyl
p-Terphenyl
Benzo[a]fluorene
Benzo[b j fluorene
Benz[ a ]anthracene
Chrysene
Benzo[b]fluoranthene
Benzo[kjfluoranthene
Benzo [ e ] pyrene
Benzo[a]pyrene
Perylene
3-Methylcholanthrene
Indeno [ 1,2,3-cd ] pyrene
Dibenz[ac/ah]anthracene
Bnezo [ ghi ] pery lene
" Supercritical fluid : CO2; modifier : Phcl (1.200 mL); matrix : XAD-2 (1.5 g);
volume of extraction cell (vertically kept) : 10 mL; volume of fluid employed :
40-60 mL; fluid flow rate : 0.60-0.90 mL.min'1; amount spiked : 0.2597-0.6677 ug.
The numbers under the dividing bar represent standard deviations; the values
under the vertical columns # 1 S 3 are from five replicate extractions at each
condition while those under columns # 2 & 4 are from one and two determination(s)
respectively; column t 1 values are from the set-up where no water was present
with the matrix while column f 2 & 3 values are from the set-up where the matrix
was made wet with 0.150 mL water, before extraction. However, column t 2 & 3
values differ in the respect that in # 2 case, water was injected in before the
addition of chlorobenzene, while in t 3 case, the reverse addition was made; in
these cases, the extract was dried with anhydrous sodium sulfate before analysis.
After spiking onto an teflon impinged glass fibre filter (47 mm diameter), at
room temperature, clean air was pulled through it and two cartridges (one over
the another, each carrying 1.5 g XAD-2) tightly securing the whole set-up in a
milled aluminum housing, for 24 h; on extracting and analyzing each of the three
components (one filter & two cartridges) individually (for the filter extraction,
2.5 mL extraction cell, 14.9-17.4 mL fluid and 0.23-0.27 mL.min" flow rate was
employed), nothing was detected from the bottom cartridge; column # 4 values
represent total % recovery from the filter and the top cartridge.
99
-------
A DISCUSSION OF VOLATILE ORGANIC COMPOUND
SAMPLING AND ANALYSIS USING A MOBILE GC/MS
H. Yoest and T. Russell
Entropy Environmentalists, Inc.
P.O. Box 12291
Research Triangle Park, NC 27709
and
M. Chu, D. Harvan, and Y. Tondeur, Ph.D
Triangle Laboratories, Inc.
801-10 Capitola Drive
Durham, NC 27713
ABSTRACT
To demonstrate the capability of a mobile gas chromatography/mass spectrometry unit to meet the
criteria of volatile organic compound compliance testing, an experiment was conducted in two phases.
Phase 1, conducted on March 4, 1993, involved tuning and calibrating the mobile unit according to
OSW 846 Method 8240 specifications. For Phase 2, conducted on March 17, 1993, the mobile unit was
moved to a different location. The instruments were retuned, continuing calibrations were performed,
and laboratory and VOST audits were analyzed according to OSW 846 Method 5040A specifications.
Overall, the system tuning, calibrations, and audit results, both laboratory and VOST, were determined
to be successful. Since all requirements for compliance purposes were met during the demonstration, the
mobile gas chromatography/mass spectrometry unit is capable of providing quality data in the field.
100
-------
INTRODUCTION
The need for quick turnaround multi-component emissions sampling and analysis precipitated the
design and construction of a mobile gas chromatography/mass spectrometry (GC/MS) unit. The goal of
this paper is to report the demonstrated success of performing on-site analysis of volatile organic
compounds (VOCs) using a mobile GC/MS. The criteria met during this demonstration are compliance
acceptable but can be used for field screening and engineering studies as well. Analyses were performed
in accordance with the guidelines of OSW-846 Methods 8240 and 5040A.
Method 8240, based upon a purge-and-trap GC/MS procedure, is used to determine and quantify
volatile organic compounds in solid waste matrices including ground water, sludges, liquors, solvents,
oily wastes, tars, fibrous wastes, polymeric emulsions, filter cakes, spent carbons, spent catalysts, soils,
and sediments.
Method 5040A is used to determine and quantify gas stream volatile principal organic hazardous
constituents (POHCs) collected on Tenax and Tenax/charcoal sorbent cartridges using a volatile organic
sampling train (VOST). Because the majority of gas streams sampled using VOST contain a high
concentration of water, the analytical method is based on the quantitative thermal desorption of volatile
POHCs from the Tenax and Tenax/charcoal traps and analysis by purge-and-trap GC/MS.
Method 8240, which is referenced in Method 5040A, specifies the operating parameters and
acceptance criteria for the GC/MS hardware. Method 5040A specifies the acceptance criteria for the
GC/MS analytical procedures used in this demonstration.
MOBILE GC/MS EQUIPMENT TECHNICAL INFORMATION
There were several important design needs that had to be met to ensure the integrity of the mobile
unit. Some of these design needs included measures to reduce the risk of damage to the instrumentation
during transport to the site and ensure the environment inside the mobile laboratory remains as
contaminant-free as possible. Thus, these design needs led to the construction of a prototype air
suspension table to isolate instrumentation from shock and vibration and state-of-the-art temperature
controls operating during transport and while performing on-site analysis. Additionally, the mobile
laboratory includes measures to avoid contamination such as high efficiency particle and hydrocarbon air
filters and a forced air curtain to provide a barrier from air movement in or out of the mobile unit.
Further, to ensure the mobile unit can operate independently, the mobile unit is equipped with two on-
board 10 kVA diesel generators and an un-interruptable power supply (UPS) to provide battery backup.
The system was designed such that the generators can be placed downwind from the mobile laboratory,
ensuring no cross-contamination from the generators.
The mobile laboratory is equipped with an on-board VOST oven and Tekmar Model 2000 Purge
and Trap and Hewlett Packard 5890-71A GC/MS able to quantitate compounds down to the 0.1 jig level.
THE DEMONSTRATION
The main objective of the demonstration was to perform analytical tuning and calibrations of the
instrument, move it to another location, and then retune and check calibrations and recalibrate, if
necessary. The secondary objective was to perform a full array of analytical procedures to demonstrate
101
-------
acceptability compared to the guidance criteria in 8240 and/or 5040A. The demonstration was
performed in two phases: Phase 1 was performed on March 4, 1993 and Phase 2 was performed on
March 17, 1993. During each phase, the recommended GC/MS operating conditions listed in Table 1
were met.
The GC/MS was tuned and calibrated prior to analysis in accordance with 8240. Table 2
presents a summary of the analytical acceptance criteria for VOCs as specified in 8240. Each of the
systematic criteria checks were performed with the mobile unit.
Table 1. Recommended GC/MS Operating Conditions - OSW 846 Method 8240
Electron Energy
Mass range
Scan time
Initial column temperature
Initial column holding time
Column temperature program
Final column temperature
Final column holding time
Source temperature
Transfer line temperature
Carrier gas
70 volts
35-260 amu
1 sec/scan
45 °C
3 min
8 °C/min
220 °C
15 min
According to manufacturer's specs
250 "C
Helium at 30 cm/sec
Phase 1 - March 4, 1993
On March 4, 1993 the instrument was installed in the truck and allowed tune to equilibrate.
Then an initial tuning and an initial calibration were performed.'
The BFB Initial Tuning
The initial tuning involved introducing 50 ng of 4-bromofluorobenzene (BFB) standard as a
reference compound to demonstrate that the GC/MS system was properly mass calibrated and tuned prior
to performing the initial calibration. Specific mass spectral criteria must be met prior to performing
initial or continuing calibrations. As shown in Table 3, the BFB key ions and abundance criteria were
met. The time of starting the data acquisition for this analysis defines the beginning of the 12-hour clock
for a valid set of analyses.
The Initial Calibration
The initial calibration was performed once the BFB tuning was demonstrated. The initial cali-
bration involves the determination of a five-point calibration curve to demonstrate VOST purge and trap,
GC, and MS (the system's) performance. An initial calibration was performed for the 8240 list of
VOCs. The calibration solutions were spiked onto a Tenax/Tenax-Charcoal tube pair and then thermally
desorbed through the purge and trap GC/MS system.
102
-------
Table 2. Summary of Analytical Acceptance Criteria
OSW-846 Method 8240/5040A
Quality Parameter
Initial tuning
Initial calibration
System performance
check
Calibration check
Consistency
Accuracy
Blanks
Method of Determination
Hardware tuning using BFB
Five-point calibration curve
Determine RFs for system
performance check
compounds"
Determine RFs for
calibration
check compounds'1
Internal standard retention
time and area from
calibration check
Spike each fraction with
surrogates
Method blank
Field blank
Frequency
Prior to analyses;
then every 12 hours
Prior to analyses;
then mid-level check
every 12 hours
Prior to use of
calibration curve;
then every 12 hours
After the SPC
Evaluate immediately
after or during data
acquisition
Each fraction
One per set of samples
One per test run series
Criteria
See Table 3.
< 30% RSD of average RF
RF for each SPCC should be at
least 0.30 except 0.25 for
bromofonn
<25% deviation between initial
RF and current RF
Retention time deviation of < 30
sec; EICP area deviation of less
than a factor of 2
50 - 150% recovery
Less than lowest standard
Less than lowest standard
'Chloromethane, 1,1-dichloroethane, bromofonn, 1,1,2,2-tetrachloroethane, chlorobenzene
*l,l-dichloroethene, chloroform, 1,2-dichloropropane, toluene, ethyl benzene, vinyl chloride.
Table 3. Initial Tuning Criteria
BFB Key Ions and Abundance Criteria
OSW 846 Method 8240
Mass
50
75
95
96
173
174
175
176
177
Ion Abundance Criteria
15 - 40% of mass 95
30 - 60% of mass 95
base peak, 100% relative abundance
5-9% of mass 95
< 2% of mass 174
> 50% of mass 95
5-9% of mass 174
> 95% but < 101% of mass 174
5-9% of mass 176
March 4, 1993
19.7
46.9
100
8.3
0.0
81.7
5.5
99.1
6.0
March 17, 1993
19.0
44.9
100
8.2
0.0
91.6
8.2
97.8
5.2
103
-------
Included in the 8240 list were system performance check compounds (SPCC) (chloromethane,
1,1-dichloroethane, bromoform, 1,1,2,2-tetrachlorethane, and chlorobenzene), calibration check
compounds (CCC) (1,1-dichloroethene, chlorofoim, 1,2-dichloropropane, toluene, ethylbenzene, and
vinyl chloride), internal standards (bromochloromethane, 1,4-difluorobenzene, and chlorobenzene-dj),
and the surrogate standards (toluene-dg, 4-bromofluorobenzene, l,2-dichloroethane-d4, benzene-dj, ando-
xylene-dio).
In accordance to Method 8240, the response factors (RF) for each compound relative to one of
the internal standards were calculated at five concentration levels: 0.10 jtg, 0.25 fig, 0.50 ng, 0.75 jjg,
1.0 /ig. An average RF was calculated for each compound. An SPC was made before the calibration
curves were used for analysis. The minimum acceptable average RF for volatile SPCCs is 0.30 for
chloromethane, 1,1-dichloroethane, 1,1,2,2-tetrachloroethane, and chlorobenzene; for bromoform, the
minimum acceptable average RF is 0.25.
As shown in Table 4, bromoform did not meet the 8240 minimum average response factor for the
five calibration points. Bromoform is one of the compounds most likely to be purged very poorly if the
purge flow is too slow. Cold spots and/or active sites in the transfer lines may adversely affect
response.
The RFs from the initial calibration are used to calculate the percent relative standard deviation
(%RSD) for calibration check compounds (CCCs). The %RSD for each individual CCC should be less
than 30 percent. A CC is evaluated after the SPC is met. The CCCs are 1,1-dichloroethene,
chloroform, 1,2-dichloropropane, toluene, ethylbenzene, and vinyl chloride and are used to check the
validity of the initial calibration. As shown in Table 5, the % RSD for the initial calibration (ICAL) met
the calibration criteria of less than 30 percent for all compounds.
Phase 2 - March 17, 1993
On March 17, 1993, the mobile GC/MS laboratory was moved to another location where the BFB
initial tuning and continuing calibration (CONCAL) were performed. In addition, laboratory audits and
VOST audit samples were analyzed.
The BFB Initial Tuning
Another BFB tuning was performed in accordance with Method 8240. Again, as shown in Table
3, the BFB key ions and abundance criteria were all within the acceptance criteria.
The Continuing Calibration
The initial calibration curve for each compound of interest must be checked and verified once
every 12 hours of analysis time. This is known as a continuing calibration and is performed by
analyzing a calibration standard that is at a concentration near the midpoint concentration for the working
range of the GC/MS, which in this case was 0.25 /ig. The SPC acceptance criteria for the response
factors are the same as for the ICAL: 0.30 minimum and 0.25 minimum for bromoform. Again, the
bromoform was not within the acceptable range for Method 8240 (see Table 4).
104
-------
Table 4. System Performance Check Compounds (SPCC)
OSW Method 8240
Compound
chloromethane
1 , 1-dichloroethane
bromoform
1,1,2,2-
tetrachloroethane
chlorobenzene
Minimum Acceptable
Average RF
0.30
0.30
0.25
0.30
0.30
March 4, 1993
ICAL
RFF
0.318
2.24
0.189*
0.520
0.891
March 17, 1993
Concal
RRF
0.337
2.25
0.183*
0.516
0.921
*Not accepted for Method 8240 (not required for Method 5040A)
RF - response factor
Table 5. Calibration Check Compounds (CCC)
OSW-846 Method 8240
Compound
1,1 dichloroethane
chloroform
1 ,2-dichloropropane
toluene
ethyl benzene
vinyl chloride
% RSD
Acceptance Criteria
< 30
< 30
< 30
< 30
< 30
< 30
ICAL
% RSD
6.95
12.4
3.91
6.31
2.27
15.21
CONCAL
%D
0.3
0.4
12.6
4.5
9.9
8.2
105
-------
After the SPC criteria are met, CC acceptance criteria are checked. The CC criteria are
represented in terms of percent difference from the initial calibration to the continuing calibration. If the
percent difference is less than 30%, then the continuing calibration is valid. In each case, the minimum
acceptance criteria of the CC were met (see Table 5).
Consistency
One of the analytical quality parameters to be determined is the consistency of the data. This is
determined by the retention times and internal standard responses evaluated immediately after data
acquisition of the samples. For purposes of discussion, the method and trip blanks, laboratory audits,
and VOST sample audits are considered "samples." The compounds evaluated for this determination
are the internal standards bromochloromethane, 1,4-difluorobenzene, and chlorobenzene-d,.
The criteria for the retention time are based on changes between the retention time measured in
the sample and the retention time from the continuing calibration. If the retention time for any internal
standard changes by more than 30 seconds, the chromatographic system must be inspected for
malfunctions and corrections made. As shown in Table 6, each sample (Method Blank, Trip Blank, Lab
Audit 1, Lab Audit 2, VOST sample Audit 1, and VOST sample Audit 2) are within the 30 second
retention time criteria.
The criteria for the internal standard responses is based on changes between the area measured in
the sample and the area measured for the continuous calibration. If the area for any of the internal
standards changes by a factor of two (i.e., -50% to +100%), the mass spectrometer must be inspected
for malfunctions and corrections made. As shown in Table 6, each sample was within the percent
difference criteria.
Another analytical quality parameter is the accuracy of the data. Accuracy is determined by the
percent recovery of the surrogate compounds. These compounds included l,2-dichloroethane-d4,
benzene-da, toluene-d,, o-xylene-d10, and bromofluorobenzene. A known amount, 0.25 ^g, of each
surrogate was spiked into the samples. As shown in Table 7, the percent recovery criteria of 50 to
150% was met for each compound and sample with the exception of the VOST Audit 2 sample for BFB.
AUDIT RESULTS
Two sets of audits samples were analyzed: (1) laboratory audits 1 and 2 and (2) the VOST sample
audits, Audit 1 and Audit 2.
Laboratory Audits 1 and 2 each consisted of 10.8 liters of a National Institute of Standards and
Technology (NIST) cylinder gas pulled directly onto one VOST Tenax tube. The purpose of these audits
was to identify analytical error. The NIST cylinder gas was certified to one percent purity. The
concentrations in the NIST cylinder gas ranged from 9.63 to 10.2 ppbvd per compound.
The VOST sample audits performed in the shop area purposely were not shielded from cross-
contamination. The audits consisted of 10.618 liters for Audit 1 and 10.336 for Audit 2 of a NIST
cylinder gas. The NIST cylinder gas was certified to one percent purity. The concentrations in the NIST
gas cylinder ranged from 4.7 to 5.3 ppbvd per compound.
106
-------
Table 6. Consistency Results
Retention
Time Limit
bromochlo-
romethane
1,4-
difluoro-
benzene
chloroben-
zene-dj
Area Li mil
bromochlo-
roraethane
1,4-
difluoro-
benzene
chloroben-
zene-dj
Concal v
Standard
13.31
16.63
25.07
Concal
Standard
23,893
101,925
99,903
Blank
time
13.31
16.64
25.06
area
24,956
115,245
103,601
*D
-0.04
-0.12
0.02
%D
-4.45
-13.07
-3.70
Blank'
time
13.31
16.65
25.08
"area- ';
23,527
117,234
88,743
V*B,-
-0.03
-0.15
-0.03
>'
1.53
-15.02
11.17
Lab Audit 1
time
13.32
16.67
25.11
area
25,986
118,716
109,424
%D
-0.07
-0.28
-0.16
%D
-8.76
-16.47
-9.53
Lab Audit 2
time
13.36
16.68
25.11
area
21,340
117,470
105,248
%D
-0.29
-0.37
-0.19
%D
10.69
-15.25
-5.35
VOST Audit 1
time
13.36
16.68
25.11
area
21346
112,621
103,818
%D
-0.40
•0.32
-0.16
*D
10.66
-10.49
-3.92
VOST Audit 2
time
13.34
16.68
25.12
area
25,072
112,587
100,859
%D
-0.26
-0.34
•0.96
%D
-4.93
-10.46
-0.96
Acceptance
Criteria
+/- 0.5 min
+/- 0.5 min
+/- 0.5 min
-50 to
+ 100%
-50 to
+ 100%
-50 to
+ 100%
-------
Table 7. Accuracy Results
Compound
1 ,2 dichloroethane-d.
benzene-da
toluene-da
o-xylene-d,0
bromofluorobenzene
Spiked Amount
0«g)
0.25
0.25
0.25
0.25
0.25
% RECOVERY
Method
Blank
98.00
92.86
105.84
58.79
59.74 •
Trip
Blank
108.49
88.54
116.85
82.92
60.20
Lab
Audit 1
105.50
91.32
101.87
91.40
84.15
Lab
Audit 2
114.95
88.76
100.76
96.78
89.50
Sample
VOST
Audit 1
113.90
90.06
103.68
97.12
87.00
Sample
VOST
Audit 2
104.93
90.58
103.35
96.33
15.32 *
Acceptance
Criteria
Limits (%)
50-150
50-150
50-150
50-150
50-150
Table 8. Audit Results
Compound
Vinyl chloride
Trichlorofluoromethane
Methvlene chloride
Chloroform
1 ,2-Dichloroethane
1,1,1 -Trichloroethane
Benzene
Carbon tetrachloride
1 ,2 Dichloropropane
Trichloroethene
Toluene
Tetrachloroethane
Chlolobenzene
Ethvlbenzeoo
o-Xvlene
Lab Audit 1
23.8
74.8
83.3
86.8
108
84.0
88.5
80.4
80.8
84.8
94.7
96.0
81.3
89.8
79.9
Lab Audit 2
34.9
81.9
99.9
105
132
85.7
88.5
77.4
76.6
84.8
110
97.5
87.9
89.8
82.2
VOST
Sample
Audit 1
75.2
177
229
103
128
130
92.8
88.4
64.2
79.3
126
95.6
68.4
77.2
72.6
VOST
Sample
Audit 2
56.2
129
213
97.5
122
421
89.4
84.7
74.2
81.5
16,700
166
78.6
84.0
78.9
Acceptance
Criteria
50-150%
50-150%
50-150%
50-150%
50-150%
50-150%
50-150%
. 50-150%
50-150%
50-150%
50-150%
50-150%
50-150%
5O-150%
SO-15O%
Audit Pass
Laboratory
1
N*
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
2
N*
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
VOST
1
Y
N*
N*
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
2
Y
Y
N*
Y
Y
N*
Y
Y
Y
Y
N*
N*
Y
Y
Y
*Did not meet mtn
ceptance criteria.
-------
The laboratory and VOST audits were performed independently, each using different cylinders,
different sampling systems, and different technicians collecting the samples. Only the GC/MS
instrument and operator were the same.
As shown in Table 8, the laboratory audits 1 and 2 failed for vinyl chloride; both VOST sample
audits passed. Since the laboratory audit involved the use of one Tenax tube with no charcoal tube, th
loss of vinyl chloride was probably due to poor retention of vinyl chloride on Tenax. All other
compounds were within 50 to 150 percent recovery.
The VOST sample audits realized mixed results. Trichlorofluoromethane failed on Audit 1 and
passed on Audit 2. Additionally, both laboratory audits passed, suggesting there was a sampling or
sample recovery error. However, the VOST tubes used for the VOST audits had been taken to a facil
which made trichlorofluoromethane, or freon-11; therefore, the tubes which were extras were probably
contaminated in the field. Methylene chloride failed both VOST sample audits and passed both
laboratory audits, again suggesting there was a sampling or sample recovery error. Methylene chlorid
is commonly used in the shop, possibly causing cross-contamination to the sampling train. The analyte
1,1,1-trichloroethane, tetrachloroethane, and toluene failed VOST sample Audit 2; the laboratory audit
and VOST sample Audit 1 passed, suggesting a sampling or sample recovery error. Toluene, a comm
shop solvent, failed on the VOST sample Audit 2 at 16,700 percent recovery which was obviously cro
contamination.
DISCUSSION AND CONCLUSION
The demonstration to show the capability of the mobile GC/MS unit to meet compliance criteria
for VOC analysis had two objectives. The first objective, completed in two phases, was to tune and
calibrate the instrument according to Method 8240 specifications, move the mobile unit to another loca
tion, retune, and recalibrate if necessary. The second objective was to demonstrate the acceptability of
the analytical procedures for meeting VOC compliance requirements and was performed during Phase 2.
Phase 1 involved the BFB initial tuning and initial calibration which included a system
performance check and a calibration check. The initial tuning and initial calibration criteria were met by
all compounds except bromoform. Bromoform, which is not required by Method 5040A, did not meet
the requirements of the SPC.
After the mobile unit was moved to another location, Phase 2 was initiated. This part of the
demonstration involved retuning the instrument, checking the calibration, and analyzing the audit
samples. The SPC and CC acceptance criteria for the continuing calibration is the same as for the ink
calibration. Again, bromoform did not meet the Method 8240 specification for the SPC; however, the
CC criteria were met. The first objective of the demonstration was considered a success.
During the analytical part of Phase 2, the consistency and accuracy of the data were determined
by Method 8240 specifications. All of the data passed the consistency criteria for retention times and
area counts. The accuracy criteria measured as percent recovery was met by all of the surrogates exce
bromofluorobenzene during VOST Audit 2. During the laboratory audit, all compounds except vinyl
chloride, which is not usually a target analyte, passed. During the VOST audit, trichlorofluoro-
methane, methylene chloride, 1,1,1-trichloroethane, toluene, and tetrachloroethane failed at least one o
the audits. Sampling and recovery contamination contributed to the failure of most of these compound
Overall, the second objective was considered a success.
109
-------
References
1. OSW Method 8240 Gas Chromatography/Mass Spectrometrv for Volatile Oreanics. September
1986.
2. OSW Method 5040A Analysis of Soibent Cartridges from Volatile Organic Sampling Train (VOS1
Gas Chromatography/Mass Spectrometry Technique. Revision 1, November 1990.
3. OSW Method 0030 Volatile Organic Sampling Train. September 1986.
110
-------
METHOD 301 FIELD AND LABORATORY VALIDATION OF TARGET
SEMI-VOLATILE ORGANIC COMPOUNDS USING SW-846 METHOD 0010
Theresa Russell, Helen Yoest, Jesus Peralta
Entropy Environmentalists, Inc.
P.O. Box 12291
Research Triangle Park, North Carolina
and
Roy Huntley
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
ABSTRACT
Laboratory and field validation experiments were devised for SW-846 Method 0010 at a
non-recovery coke oven. A representative model was selected and sampling was conducted
over a 72-hour period. Using Method 301 criteria, bias and precision calculations were
determined for the target analytes: toluene, xylenes, (o-, m-, p-), phenol, cresol, naphthalene,
and benzo-a-pyrene. Laboratory experimentation validated all analytes except toluene and
naphthalene; field experiments validated SW-846 Method 0010 for all analytes. Since toluene
and naphthalene are used as laboratory solvents, the laboratory validation results for these
analytes were considered skewed due to contamination. SW-846 Method 0010 was judged
validated by Method 301 specifications for representative sampling at non-recovery coke ovens.
INTRODUCTION
Non-recovery coke ovens utilize a coking technology which expends combustibles
extracted from the coal as fuel for oxidation - producing heat, carbon dioxide, and water vapor.
The finished product of the coking process is carbon, a raw material widely used in the steel
industry. Because of this unique re-entrainment technique, non-recovery coke ovens use no
other air emission control devices other than the precise monitoring and control of the coking
process.
Because little data exist to characterize the emissions from this type of process, the U.S.
Environmental Protection Agency (EPA), Office of Air Quality Planning and Standards
(OAQPS), Emission Measurement Branch (EMB), in support of National Emission Standards
for Hazardous Air Pollutants (NESHAPS) selected a representative model for the industry in
order to develop a comprehensive source category document for the coke oven industry; in an
effort to assess the potential environmental impact as directed by the Clean Air Act (CAA) of
1990, this database would serve as a standard from which to determine applicable emission
limits.
Entropy Environmentalists, Inc. (Entropy) conducted a testing program to determine
emissions from a non-recovery coke oven. The purposes of the three-day testing program were
to provide a laboratory and field validation study of SW-846 Method 0010 [Modified Method
Five (MM5)] and to perform an emissions characterization evaluation on the non-recovery
coking process. Data collected should certify the use of Method 0010 sampling trains to
characterize target semi-volatile organic compound emissions from non-recovery coke ovens.
This paper addresses the results of the Method Validation Study.
ill
-------
METHOD 301
Method 301, "Field Validation of Pollutant Measurement Methods From Various Waste
Media," details procedures to be used when a test method is proposed to meet an EPA
requirement in the absence of a validated method. The method includes procedures for
determining and documenting the quality - defined as the systematic error (bias) and random
error (precision) - of the measured concentrations of the source emissions.
In order to determine the bias and precision of a proposed sampling method, known
concentrations of analytes or their surrogates are introduced to collection media and
collocated, simultaneous samples are collected. By analyzing the recovery of the known,
"spiked" compounds, any systematic positive or negative difference between measured and true
values can be determined; these biases are commonly caused by interfering compounds in the
effluent gas, calibration errors, and inefficient analyte collection. Calculated bias correction
factors must fall within the 0.70 to 1.30 range specified by the protocol. By comparing the
results of paired sampling trains, variability in data obtained from the entire measurement
system (sampling and analysis) can be quantified. The precision of the proposed method must
have a relative standard deviation s 50% in order to be accepted as valid.
Validation Procedure
Because the semi-volatile organic analytes to be analyzed are quantified using gas
chromatography/mass spectrometry (GC/MS), Method 301 specifies that isotopic spiking be
used as the framework for this validation study. Sampling train media are spiked with
surrogate compounds prior to field testing. Twelve samples are then collected for analysis
using either paired or quadruplet collocated sampling trains. Paired, collocated sampling trains
are defined as two probes arranged so that the probe tip is 2.5 cm from the outside edge of the
second probe with a pilot tube on the outside of each. Sampling procedures outlined in detail
in SW-846 Method 0010 are strictly followed, with the exception of the paired sampling train
design and inclusion of surrogate compounds. Labeled isotope and/or deuterated mixtures are
introduced at each phase of analysis to pinpoint sources of laboratory loss and thereby
accurately determine the bias and precision of the proposed method.
Validation Calculations
Data obtained from the minimum of 12 runs (six pairs for collocated sampling trains)
are utilized in the following calculations in order to derive the bias and relative standard
deviation quantities and assess the significance of these values.
The sample mean uses the results from the analyses of the isotopically spiked field
samples and is calculated using:
£
xi
where:
Sm = sample mean
X; = ith measurement
n = sample size.
The bias is calculated using the sample mean and the value of the isotopically labeled
112
-------
spike as follows:
B Sm - CS (2)
where:
B = bias
Sm = sample mean
CS = amount spiked.
Next, the sample standard deviation is calculated using the following equation:
SD
n - 1
where:
SD = sample standard deviation
X; = ith measurement
Sm = sample mean
n = sample size.
The standard deviation of the mean is calculated by:
SDM = — (4)
where:
SDM = standard deviation of the mean
SD = sample standard deviation
n = sample size.
The bias of the sample is tested for significance by calculating:
t 1*1 (5)
SDM
and comparing this result to the critical t-value for a two-sided test at the .05 level of
significance, with n-1 degrees of freedom. If the calculated t-value is larger than ty^, the bias
is considered significant and a correction factor must be calculated using the following:
113
-------
CF
1 H- *- S" (6)
cs
0.70 £ CF £ 1.30 for validation
where:
CF = correction factor
B = bias
CS = amount spiked
Sm = sample mean
Finally, the precision, as measured by the percent relative standard deviation, is
calculated by:
or*
%RSD = — X 100 (7)
Sm
%RSD ^ 50% for validation
where:
%RSD = percent relative standard deviation
SD = sample standard deviation
Sm = sample mean.
Sampling and Analytical Procedures
Method 0010 sampling was conducted over a 72-hour period on a non-recovery coke
oven. Thirty-six ovens exhaust into four stacks via a common tunnel; therefore, each stack
approximates the emission's flow rate of nine ovens combined. The coking process operates on
a 48-hour staggered cycle that begins when the even numbered ovens are charged with
approximately 40 tons of coal; oven heat extracts combustibles inherent in the coal and oxidizes
them to produce more heat. Coal bakes in this manner for 48 hours, after which it is removed
as the finished product coke. Twenty-four hours after charging the even numbered ovens, the
odd numbered ovens are loaded with coal and begin the 48-hour cycle. The stack is a vertical
duct with an inner diameter of 96 inches and two ports located 5.2 duct diameters from the
nearest upstream disturbance and 0.82 diameters from the nearest downstream disturbance
(stack exit). In compliance with EPA Method 1 criteria, 24 points were selected for isokinetic
sampling; due to structural obstructions, only one port was employed and all 24 points were
sampled along this single axis.
Because normal operating conditions offered the potential for varying emissions results,
four testing conditions were arbitrarily defined. Condition I was isolated during the charging
cycle; testing began at the beginning of the charge and continued three and a half hours into
the charge for a total sampling time of two and a half hours. Condition II began where
Condition I ended and was allotted six hours; with just enough down time for train
turnarounds, actual sampling time was four hours. Conditions III and IV followed the same
format as Condition II, including sampling and down time allotments.
Laboratory and field sample analyses were performed by Triangle Laboratories, Inc.
(TLJ) using the isotope spiking analytical protocol designed for the target analytes benzo-a-
pyrene (BaP), cresol, naphthalene, phenol, toluene, and xylenes (o-, m-, p-).
Table 1 presents the relationship between each analyte and its respective surrogate and
114
-------
internal standards. The six target analytes were analyzed versus their respective labeled
internal standard; surrogate compounds were then quantitatively compared to the internal
standards. By comparing the analytes and surrogates to the Cl internal standards, losses due
to laboratory handling were corrected before calculations were performed. By comparing Cl
compounds to their respective C3 internal standard, the validity of the analytical recovery was
determined. Bias and relative standard deviation were then calculated from these
measurements.
Table 1. Relationship of analytes, surrogates, and internal standards.
ANALYTE
Toluene
Xylene
Phenol
Cresol
Naphthalene
Benz-a-pyrene
SURROGATE
SOLNC2
Ethylbenzene-d10
Ethylbenzene-dio
2-Chlorophenol-d4
2-Chlorophenol-d4
Acenaphthene-d10
Benz-e-pyrene-d,2
Terphenyl-d14
EXTRACTION
INT STANDARD
SOLN Cl
Toluene ds
o-Xylene-djo
Phenol-d,
o-Cresol-d-TH
Naphthalene-dj
Benz-a-pyrene-d12
ANALYSIS
INT STANDARD
SOLNC3
Dichlorobenzene-d4
Dichlorobenzene-d4
Phenanthrene-d10
Phenanthrene-d10
Chrysene-d,2
Perylene-d,2
Laboratory Validation. The laboratory validation experiment was designed so that samples to
be analyzed closely matched expected field sample concentrations for the analytes of interest.
Front-half, back-half, and aqueous impinger fractions were created using the same reagents and
rinses described in Entropy's test design; for continuity, analysis fractions followed the matrix
employed by Entropy during field sampling and recovery, so that the "front-half1 fraction
included the filter and the probe rinse (methanol/methylene chloride), the "back-half fraction
included the XAD resin and the condenser rinse, and the "aqueous impinger" fraction included
the reagent, moisture catch and rinse recovered from the sampling train impingers. The
separate fractions were spiked with 50 ng of each analyte (ANALYTE) and surrogate (SOLN
C2). XAD resin was also spiked with the pre-spike compound terphenyl-d,4. Laboratory
samples were prepared for extraction by fortifying each segment separately with 100 jtg of
deuterium-labeled compounds identical to the analytes of interest, excluding the isotopic label
(SOLN Cl); extractions were performed and evaporated to 5 mL. Prior to GC/MS analysis,
the three extracted constituents were separately fortified with a second set of deuterium-
labeled compounds (SOLN C3) as internal recovery standards and a 10 mL aliquot was
sampled for analysis.
Using a pre-detennined mass for each compound to generate selected ion current
profiles for quantification, each compound was measured against its respective standard.
Analyte and surrogate recoveries were compared against Cl internal standards and Cl
solutions were compared against C3 internal standards. These comparisons allowed TLI to
correct for analyte or surrogate losses due to laboratory handling. In the same manner, pre-
spiked terphenyl-d14 and laboratory spiked standards recovery were quantified relative to the
recovery internal standards.
115
-------
Field Validation. The field validation procedure involved pre-spiking the XAD resin for all
samples prior to field sampling. Isotopically labeled compounds were selected as surrogates for
the targeted analytes because these compounds were chemically analogous to the compounds
of interest. The relationship between the analytes and the surrogates is shown in Table 1.
XAD resin was pre-spiked with the surrogate solutions and terphenyl-d14 by directly introducing
50 ng of each compound in a methylene chloride solution via a syringe into the inlet end of the
XAD trap before sealing and packing the traps for shipping to the test site. Recovery of the
surrogate compounds was intended to demonstrate the ability of the entire methodology to
achieve the Method 301 criteria.
Field samples were prepared for extraction by fortifying each fraction (front-half, back-
half, and impingers) separately with 50 /ig each of deuterium-labeled internal standards (SOLN
Cl), compounds that are identical to the analytes of interest excluding the isotopic label. The
target analytes were measured versus these internal standards and any loss due to laboratory
handling was corrected during the analytical calculations.
Prior to GC/MS analysis, the separate fractions were fortified with the deuterium-
labeled recovery standards (SOLN C3) listed in Table 1. Internal standards recoveries were
measured relative to these recovery standards.
Method 301 specifies that paired trains be used; however, the validation calculations
given do not pair the data. Therefore, the bias and percent relative standard deviation
verifications required by Method 301 were determined two ways: (1) using valid paired train
data only and (2) using all valid data, excluding field blanks. Results were then compared to
see if different conclusions would be reached. There were 14 measurements or seven pairs of
data for all surrogates except 2-chlorophenol-d4, for which there were 12 measurements or six
pairs of data.
All sample concentrations were given in micrograms (/ig) using Equation (8):
Amount ua = X x Amt Int Std x Dilution Factor
' Area Int Std x RF X
where:
X = Analyte
Int Std = Internal Standard
Amt Int St = Amount Internal Standard = 50 ng
RF X = Response Factor of Analyte X from
following Calibration Point
RESULTS
Laboratory Validation
Method 301 does not address the question of whether to correct the data for blanks
before proceeding with the validation protocol. Therefore, for comparison, analyte and
surrogate results are reported both with and without blank correction. Method 301 specifies
that 12 samples be analyzed; however, because the validation would be determined under a
laboratory control, only three samples were spiked and analyzed. Since the three fractions
were spiked with the analytes, as well as the surrogates, direct validation of both analytes and
surrogates was possible. Therefore, the reported analyte results were not determined by the
corresponding surrogate results. The results of the laboratory validation are presented in
Tables 2, 3, and 4.
116
-------
Front-half Analyses. Without blank correction, all of the analytes and C2 surrogates passed
the validation requirements (see Table 2). The %RSD's ranged from 1.88% to 6.43%. The
analytes toluene, and BaP, as well as the surrogates 2-chlorophenol-d4 and benz-e-pyrene-d12,
required calculation of correction factors of 0.81, 1.30, 1.17, and 1.15, respectively.
With blank correction of the data, the %RSD's ranged from 1.88% to 7.25%, all
meeting the requirement of less than 50%. 2-chlorophenol-d4, benzo-e-pyrene-di2, and benzo-
a-pyrene required correction factors which were 1.17, 1.15, and 1.46, respectively. Thus, BaP
failed Method 301, which specifies that correction factors fall within 0.7 to 1.30, inclusive.
Back-half Analyses. Naphthalene failed validation without a blank correction, having a %RSD
of 53.56% (see Table 3). The other %RSD's ranged from 1.43% to 9.11%. Toluene also
failed validation with a correction factor of 0.61.
With blank correction, naphthalene still failed with a %RSD of 58.86%, which is higher
than when the data are uncorrected. However, toluene now meets both the precision and bias
requirements. This is not surprising since naphthalene is a common contaminant of the XAD
resin and toluene is omnipresent in a laboratory environment. The other target compounds
meet Method 301 criteria.
Impinger Analyses. Without blank correction, all analytes and C2 surrogates analyzed for the
impinger fraction pass the %RSD requirement, with values ranging from 0.73% to 25.36% (see
Table 4). Toluene and benzo-e-pyrene-d12 both failed validation because of their bias
correction factors of 0.27 and 1.31, respectively.
With blank correction, toluene still failed validation with an absolute %RSD of 109.21%.
The adjusted correction factor for benz-e-pyrene-d12 remained 1.31, just outside the range
specified in Method 301.
Field Validation
In order to apply EPA Method 301 procedures for validation, the amounts (/ig) of
spiked surrogates that were recovered were calculated based on the corrected percent recovery
and the amount of labeled compounds spiked (50 /*g). The validation protocol specifies that
paired trains be used for the isotope spiking procedure. However, the bias and precision
calculations do not involve the use of paired data. Therefore, the method validation
procedures were followed using paired and unpaired data in order to compare results. The
results of the field validation are presented in Tables 5 and 6.
Paired Train Results. The paired train calculations validated Method 0010 for the field
experiment (see Table 5). The %RSD's for the surrogates ranged from 5.56% to 36.86%.
Therefore, the method met the precision requirement for each surrogate as specified by
Method 301. Two surrogates, ethylbenzene-d10 and 2-chlorophenol-d4, required bias correction
factors. These correction factors were 0.798 and 0.889, respectively; therefore, the bias
requirement was met by all surrogates as specified by the method. Since the spiked surrogates
were chosen because they are chemically similar to the analytes of interest and since all
surrogates passed validation, all of the analytes passed validation in the field.
117
-------
Table 2. Laboratory validation results - front-half fraction.
FRONT-HALF
UNCORRECTED
ANALYTE
toluene
xylene
phenol
cresol
naphthalene
benzo-a-pyiene
SURROGATE
ethylbenzene-dIO
2-chlorophenol-d4
acenaphthene-d10
benzo-e-pyrene-d12
FRONT-HALF
CORRECTED
ANALYTE
toluene
xylene
phenol
cresol
naphthalene
benzo-a-pyrene
SURROGATE
ethylbenzene-d,0
2-chlorophenol-d4
acenaphthene-d10
benzo-e-pyrene-d,2
PRECISION
%RSD
5.57
6.38
5.57
5.16
5.70
6.43
1.88
2.55
2.93
2.32
6.68
6.36
5.60
5.14
5.70
7.25
1.88
2.56
2.93
2.32
PASS/FAIL
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
BIAS
CORRECTION FACTOR
0.81
NA
NA
NA
NA
130
NA
1.17
NA
1.15
NA
NA
NA
NA
NA
1.46
NA
1.17
NA
1.15
PASS/FAIL
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Pass
Pass
Pass
118
-------
Table 3. Laboratory validation results - back-half fraction.
BACK-HALF
UNCORRECTED
ANALYTE
toluene
xylene
phenol
crcsol
naphthalene
benzo-a-pyrene
SURROGATE
ethylbenzene-d,0
2-chlorophenol-d4
acenaphthene-d,0
benzo-e-pyrene-d,3
BACK-HALF
CORRECTED
ANALYTE
toluene
xylene
phenol
cresol
naphthalene
benzo-a-pyrene
SURROGATE
ethylbenzene-d10
2-chlorophenol-d4
acenaphthene-dla
benzo-e-pyrene-d,2
PRECISION
%RSD
9.11
2.63
2.25
6.63
53.56
2.13
4.19
1.43
3.08
2.73
PASS/FAIL
Pass
Pass
Pass
Pass
Fail
Pass
Pass
Pass
Pass
Pass
BIAS
CORRECTION FACTOR
0.61
NA
NA
NA
NA
1.25
NA
1.19
NA
1.12
PASS/FAIL
FaU
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
1630
2.63
2.27
6.61
58.86
2.16
4.18
1.45
3.10
2.68
Pass
Pass
Pass
Pass
Fail
Pass
Pass
Pass
Pass
Pass
NA
NA
NA
NA
NA
1.25
NA
1.19
NA
1.12
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
119
-------
Table 4. Laboratory validation results - impinger fraction.
IMPINGERS
UNCORRECTED
ANALYTE
toluene
xylene
phenol
cresol
naphthalene
benzo-a-pyrene
SURROGATE
ethylbenzene-d,0
2-chlorophenol-d4
acenaphthene-d10
benzo-e-pyrene-d|2
IMPINGERS
CORRECTED
ANALYTE
toluene
xylene
phenol
cresol
naphthalene
benzo-a-pyrene
SURROGATE
ethylbenzene-d,0
2-chlorophenol-d<
acenaphthene-dro
benzo-e-pyrene-d,2
PRECISION
%RSD
25.36
3.02
1.75
2.88
9.59
1.62
4.92
0.73
2.45
5.60
PASS/FAIL
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
-109.21
2.99
1.74
2.86
9.59
1.60
4.93
0.71
2.44
5.65
Fail
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
BIAS
CORRECTION FACTOR
0.27
NA
NA
NA
NA
1.26
NA
0.77
NA
1.31
PASS/FAIL
Fail
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
NA
NA
NA
NA
NA
1.26
NA
0.77
NA
1.31
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Table 5. Field Validation Results - Paired Data.
SURROGATE
ethylbenzene-dIO
2-chlorophenol-d4
acenaphthene-d|0
benzo-e-pyrene-d,2
ANALYTE
toluene, xylene
phenol, cresol
naphthalene
benzo-a-pyrene
PRECISION
%RSD
5.56
6.00
29.30
36.86
PASS/FAIL
Pass
Pass
Pass
Pass
BIAS
CORRECTION FACTOR
0.798
0.889
NA
NA
PASS/FAIL
Pass
Pass
Pass
Pass
120
-------
Unpaired Train Results. The same validation procedures were followed using all the data that
were followed using paired train data; validation was again supported by the results (see Table
6). The same two surrogates, ethylbenzene-d,0 and 2-chlorophenol-d4, were found to have
significant biases with corresponding correction factors of 0.82 and 0.91, respectively, which fall
within the specified range of 0.7 to 1.3. The %RSD's for all surrogates ranged from 6.79% to
33.26%, all less than the 50% requirement. Therefore, all analytes passed the field validation
according to Method 301.
Table 6. Field Validation Results - Unpaired Data.
SURROGATE
ethylbenzene-d,0
2-chlorophenol-d4
acenaphthene-d10
benzo-e-pyrene-d|2
ANALYTE
toluene, xylene
phenol, cresol
naphthalene
benzo-a-pyrene
PRECISION
%RSD
6.79
7.69
29.19
33.26
PASS/FAIL
Pass
Pass
Pass
Pass
BIAS
CORRECTION FACTOR
0.815
0.906
NA
NA
PASS/FAIL
Pass
Pass
Pass
Pass
CONCLUSIONS
Laboratory Validation
Method 301 specifies that isotope spiking may be performed in field validation
procedures. However, it does not address the use of analyte vs. surrogate in the laboratory.
Since both analytes and surrogates were spiked during the laboratory validation, the
opportunity arose to directly compare analyte and surrogate results.
The method was judged successful without qualification for the analytes xylene, phenol,
and cresol, along with their corresponding surrogates ethylbenzene-d10 and 2-chlorophenol-d4.
However, differences were found between benzo-a-pyrene and its corresponding surrogate
benzo-e-pyrene-d]2 in the separate fractions analyzed. Discrepancies between BaP and its
surrogate BeP-d12 suggest that some interference may have been occurring during the GC/MS
analyses. Since the two correction factors were very close to one another (a difference of 0.05)
and the correction factor for BeP-d12 fell just outside the interval [0.70, 1.30], it appears that
BeP-d,2 is an appropriate surrogate for BaP and that BeP-d,2 may be considered to have passed
validation.
Toluene failed validation completely in the impingers, while naphthalene failed
validation completely in the back-half. Anticipated background contamination made
naphthalene and toluene analyses results unreliable and inconclusive. XAD resin inherently
contains naphthalene which is difficult to completely remove. Toluene is commonly used as an
extraction solvent in laboratories. Every effort was made to reduce all laboratory evaluation
samples and instrumentation exposure to toluene; however, this was judged by TLI to have
been only moderately successful. This contamination may explain why the surrogates
ethylbenzene-d10 and acenaphthene-di0 unconditionally passed validation in all three fractions,
while their corresponding analytes, toluene and naphthalene, did not completely pass validation
in the laboratory.
121
-------
Field Validation
The field validation was approached two ways: (1) using paired data, as per the protocol
and (2) using unpaired data. This approach was taken since the bias and precision calculations
do not involve the use of paired data. In both cases, validation was achieved for all surrogate
compounds, and therefore, for all analytes. The precision of only two compounds was lower
using the paired data; the precision of the other three compounds was higher. The biases of
the compounds were essentially the same between the paired and single data.
The field validation procedures only involved analyzing the XAD for the surrogates
since the XAD was the only fraction that was prespiked with the surrogate compounds.
However, in the laboratory, all three fractions were spiked with the C2 compounds and the
analytes. The variability in the recovery of these compounds from the front-half and the
impingers, as well as the XAD, demonstrates the need for separate extractions and analyses of
the various fractions. Additionally, the recoveries of labeled compounds spiked prior to
extraction were variable. This variability, especially between dual trains, suggests that further
research on this subject is necessary and that dynamic spiking may be needed.
REFERENCES
1. 'Testing Non-recovery Coke Ovens for Standards Development," (Confidential Client),
March 1992. Prepared by Helen Yoest, Entropy Environmentalists, Inc. for U.S.
Environmental Protection Agency, Emission Measurement Branch, Research Triangle
Park, NC 27709.
2. "Protocol for the Field Validation of Emission Concentrations from Stationary Sources,"
Research Triangle Park, NC, April 1991. Prepared by the U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards and Atmospheric
Research and Exposure Assessment Laboratory.
3. Test Methods for Evaluating Solid Waste Physical/Chemical Methods. EPA SW-846,
November 1986, 3rd Edition.
122
-------
HYDROCARBONS IN THE RANGE OF C10-C20 EMITTED FROM MOTOR VEHICLES;
DIESEL VERSUS SPARK IGNITION
B. Zielinska, J. Sagebiel, L.H. Sheetz, G. Harshfield, A.W. Gertler, and W.R. Pierson
Energy and Environmental Engineering Center, Desert Research Institute, Reno, NV
ABSTRACT
To understand better the sources of observed differences between on-road vehicle emissions and model
estimates, a series of experiments was conducted in the Fort McHenry Tunnel, Baltimore, Maryland (June 18-24,
1992). It is a four-bore tunnel, two lanes per bore, carrying the eight lanes of Interstate 95 east-west under the
Baltimore Harbor. Since heavy-duty trucks are directed into the right-hand bore, sampling in both right- and left-
hand bores permitted an assessment of the relative importance of light-duty (mostly gasoline) versus heavy-duty
(mostly diesel) vehicle emissions.
Samples were collected using stainless steel canisters (whole air samples, analyzed for C2-C12
hydrocarbons) and Tenax-TA solid adsorbent cartridges (for semi-volatile hydrocarbons, in the range C8-C20). The
samples were analyzed using high resolution gas chromatographic separation and Fourier transform infrared/mass
spectrometric detection (GC/IRD/MSD) or flame ionization detection (GC/FTD). Comparison of hydrocarbon
concentrations found in the Tenax and canister samples collected in the right- and left-hand bores of the Tunnel
enables an assessment of the contribution of semi-volatile hydrocarbons (C10-C20 range derived from Tenax data)
to the total non-methane hydrocarbons (C2-C20, derived from canisters and Tenax data).
INTRODUCTION
Motor vehicles are a leading (and possibly the major) source of VOC and NO,, the precursors of ozone
formation in urban areas throughout the United States. Yet, current motor vehicle VOC emissions estimates are
low, by a factor of two to four (rngalls, 1989'; Ingalls et al., 19892; Fujita et al., 19923). The reason for the
discrepancy is under debate, but generally it is a difference between real-world vehicles, including the way they are
maintained and driven, and estimates based on dynamometer measurements of volunteer fleets. To provide reliable
real-world emission data for mobile sources, comprehensive monitoring was carried out in a highway tunnel (Fort
McHenry Tunnel, Baltimore, Maryland). The overall objective of the Tunnel Experiment was to procure accurate
data on motor vehicle emissions of carbon monoxide (CO), oxides of nitrogen (NO,), and volatile organic
compounds (VOC) throughout the U.S. One of the specific objectives was to assess the relative importance of
light-duty spark ignition vehicle emissions versus the heavy-duty diesel (HDD) emissions. It has been shown, for
ambient air samples collected in a heavily traveled mountain tunnel with a high percentage of HDD traffic, that n-
alkanes up to C26 could be detected in the gas-phase samples (Hampton et al., 19824; 1983s). Thus, it was
important to assess the contribution of hydrocarbons ranging from CIO to C20 (or higher), so-called semi-volatile
hydrocarbons (SVHC), to total non-methane hydrocarbons (TNMHC) emitted from diesel vehicles. It has been
reported (Zielinska and Fung, 1992s; 1993') that this contribution is rather low for highway tunnel traffic dominated
by light-duty spark ignition vehicles. The traffic from HDD vehicles in the Fort McHenry Tunnel, selected for the
present study, was relatively high. In addition, since heavy-duty trucks were directed into the right-hand bore,
sampling in both right- and left-hand bores permitted an assessment of the relative importance of emissions from
light-duty (mostly gasoline) versus heavy-duty (mostly diesel) vehicles.
EXPERIMENTAL METHODS
Sampling
The Fort McHenry Tunnel, located in Baltimore, Maryland, is the world's largest underwater tunnel
designed for motor vehicle traffic. It is a four-bore tunnel, two lanes per bore, carrying the eight lanes of Interstate
95 east-west under the Baltimore Harbor. Its length is 2,195 m from portal to portal westbound, and 2,174 m
eastbound. The downgrade in the tunnel reaches -3.76% and the upgrade reaches +3.76%, for both eastbound and
westbound traffic. The Bores #1 and #2 are westbound (towards Washington, DC) and #3 and #4 are eastbound
(toward the northeast). The sampling was performed in Bores #3 (light-duty vehicles only) and #4 (light- and
123
-------
heavy-duty vehicles). The sampling sites were located in the west (entrance) and east (exit) portals in the overhead
exhaust ducts of both bores, and in midtunnel beneath the roadway, in order to separate the downhill from uphill
sections of the tunnel. In addition, ventilation air, supplied from entrance and exit buildings was also sampled. The
five measurement stations per bore resulted in 10 canister and 10 Tenax samples per run. Sampling consisted of
11 one-hour runs during the period June 18-24, 1992. Each run was triggered by traffic count and composition (see
Figure 1). Whole air samples, analyzed for C2-C12 hydrocarbons, were collected using the stainless steel canister
sampling method. Semi-volatile hydrocarbons, in the range of C8-C20, were collected using Tenax-TA solid
adsorbent. The canister and Tenax samplers were located side by side in the ventilation ducts of the tunnel, wilh
Teflon sampling lines extending from the samplers through the ventilation louvers into the tunnel area. The Tenax
sampling unit drew two parallel streams of air (at -0.5 L/min and -0.7 L/min per stream). Prior to use, the Tenax-
TA solid adsorbent was cleaned by Soxhlet extraction and thermally conditioned for four hours by heating at 280
°C under nitrogen purge. Approximately 10% of the precleaned Tenax cartridges were tested by GC/HD for purity
prior to sampling. After sampling, the Tenax cartridges were capped and placed in tin containers with activated
charcoal on the bottom, and kept on ice until transported to a laboratory freezer.
Stainless steel Summa-polished canisters of 6 L capacity, employed for volatile hydrocarbon (C2-C12)
collection, were cleaned by repeated evacuation and pressurization with humidified zero air prior to sampling, and
certified as described by U.S. EPA Method TO-148. The sampling procedure essentially followed the pressurized
sampling method described by EPA Method TO-14.
Analysis
Tenax samples were analyzed by the thermal desorption-cryogenic preconcentration method, followed by
high resolution gas chromatographic separation and Fourier transform infrared/mass spectrometric detection
(GC/IRD/MSD; Hewlett Packard 5890 JJ GC with 5979 MSD and 5965B IRD) or flame ionization detection (FID)
of individual hydrocarbons. The Chrompack Thermal Desorption-Cold Trap Injection (TCT) unit, which can be
attached to either the GC/FID or the GC/IRD/MSD system, was used for sample desorption and cryogenic
preconcentration. A 60 m (0.32 id, 0.25 um film thickness) DB-1 capillary column (J&W Scientific, Inc.) was used
and the chromatographic conditions were as follows: initial column temperature of 30 °C for two minutes,
followed by programming at 6 °C/min to a final temperature of 290 °C and held isothermal for five minutes.
Several Tenax cartridges from each sampling period and sampling location were analyzed by the GC/IRD/MSD
technique in order to identify individual hydrocarbons. Identification of individual components was made based on
their retention times and mass and infrared spectra matching those of authentic standards. If authentic standards
were not available, the National Institute of Standards and Technology (MIST) mass spectral library (containing
over 43,000 mass spectra) and the U.S. EPA infrared spectra library were used for compound identification. The
quantification of hydrocarbons collected on all remaining Tenax cartridges was accomplished by the GC/FID
technique. For calibration of the GC/FID, a set of standard Tenax cartridges was prepared by spiking the cartridges
with a methanol solution of standard SVHC, prepared from high-purity commercially available C9-C18 aliphatic
and aromatic hydrocarbons (Alltech Associates, Inc.). Ethylbenzene, 1,3,5-trimelhylbenzene, n-dodecane, and n-
lelradecane were used in the concentration range from -7-8 ng/Tenax up to 200-300 ng/Tenax. The solvent was
then removed with a stream of N2 (5 min, 100 ml/min at room temperature) and the Tenax cartridges were
thermally desorbed into the GC system, as described above. At least four concentrations of standard compounds
were employed. Area response factors per nanogram of compound were calculated for each concentration and each
hydrocarbon and then the response factors were averaged to give one factor for all hydrocarbons measured.
The stainless steel canister samples were analyzed for volatile (< C12) hydrocarbons using high resolution
capillary gas cnromatography with flame ionization detector (Hewlett-Packard 5890 Series H), after cryogenic
sample concentration in a freeze-out loop made from chromatographic-grade stainless steel tubing packed with
60/80 mesh deactivated glass beads. The chromatographic column used for the C2-C12 hydrocarbon analysis was
a 60 m long J&W DB-1 fused silica capillary column with a 0.32 mm inside diameter and 1 um phase thickness.
The oven temperature program was: -50 °C for 2 min, to 220 °C at 6 °C per min. Since the DB-1 column does not
provide complete separation of the light C2 and C3 hydrocarbons, a separate analysis of the canister sample was
performed to obtain accurate concentrations for ethane, ethylene, acetylene, propene, and propane. The
chromatographic column used for this analysis was a J&W GS-Q fused silica capillary column with an internal
diameter of 0.53 mm and a length of 30 m. The GC/FID response was calibrated in ppbC, using MIST Standard
124
-------
Reference Material (SRM) 1805 (254 ppb of benzene in nitrogen). One ppm propane in a nitrogen standard (Scott
Specialty Gases), periodically traced to SRM 1805, was used for calibrating the C2-C3 analytical system.
RESULTS AND DISCUSSION
Figure 1 shows the traffic distributions in Bores #3 and #4 during the 11 one-hour runs (note that the
number of vehicles on the y-axis is given in a logarithmic scale). It can be noted from these figures that the
percentage of HDD traffic in Bore #4 is significant and it ranges from 3.4% for Runs 4 and 5 up to 68.5% for Run
2 and ~68% for Runs 8 and 10. The highest percentage of HDD traffic occurs early at the morning, between 0300
and 0500.
Figure 2 shows the comparison of the gas chromatograpliic traces for two Tenax samples, collected during
Run 2 in Bores #3 and #4 (exit portals). Whereas the chromatograms for the Run 2 canister samples (C2-C12
hydrocarbons) collected in Bores #3 and #4 are not very different (traces not shown), the chromatograms of the
Tenax samples show substantial differences in the intensity of C9-C19 hydrocarbons. The C9-C18 paraffins are
much more abundant for the bore with 68.5% HDD traffic (Bore #4) than for the bore with only 1.8% HDD traffic
(Bore #3).
Table I shows the TNMHC concentrations in the range of C2-C12, as quantified from canister samples, and
corresponding TNMHC concentrations in the range of C2-C20, obtained from combining canister and Tenax data
for Runs 4, 7, 8, and 9 (exit portal data only). In terms of traffic composition, these runs are representative of all
11 runs. The ratio of TNMHC in the C2-C20 range to TNMHC from the canister measurements represents the true
TNMHC relative to the amount obtained from the canister alone. It is seen that if the SVHC are properly
quantified (through Tenax measurements) the TNMHC estimates increase by 8 to 55% relative to what would be
estimated from canisters alone.
Table I. Assessment of the contribution of C10-C20 hydrocarbons to TNMHC in the range of C2-C20 from
canister and Tenax measurements.
Run # Bore #
4 3
4
7 3
4
8 3
4
9 3
4
% Diesel"
0.1
3.4
1.0
30.5
2.0
68.0
1.0
28.2
X Canister HC
C2-C12
ppbC
1606
2280
863
746
453
325
1024
804
TNMHC"
C2-C20
ppbC
1737
2501
963
952
537
505
1172
1135
TNMHC (C2-C20)
I Canister HC
1.08
1.10
1.12
1.28
1.19
1.55
1.14
1.41
" Diesel defined as heavy duty diesel trucks, diesel buses and light duty diesel autos.
b Data for exit portals only.
Table n shows the fleet average hydrocarbon emission rates (TNMHC, aliphatic and aromatic
hydrocarbons) calculated for these four runs using canister data and combined Tenax plus canister data. Percent
change represents the increase resulting from including SVHC in Tenax data. As can be seen from this table,
SVHC account for a significant portion of diesel emissions; for Run 8 (Bore #4) with 68% diesel traffic, the
addition of Tenax data increased the fleet average emission rate by 75%. Both aliphatic and aromatic hydrocarbon
emission rates were increased significantly (70% and 82%, respectively). For runs dominated by light-duty spark
125
-------
Table II. Fleet Average Hydrocarbon Emission Rates [g/vehicle-mile] for Run 4, 7, 8, and 9.
TNMHC
Aromatic
# #
4 3
4
7 3
4
8 3
4
9 3
4
Diesel
0.1
3.4
1.0
30.5
2.0
68.0
1.0
28.2
C2-
C12
0.91
0.78
0.39
0.75
1.29
0.54
0.47
0.74
C2-
C20
0.99
0.85
0.43
0.97
1.52
0.95
0.52
1.04
Change
8.7
8.5
11.4
29.5
17.8
75
10.8
40.7
C2-
C12
0.59
0.51
0.21
0.43
0.68
0.30
0.25
0.48
So
0.60
0.53
0.22
0.51
0.74
0.50
0.26
0.58
Change
1.7
3.6
2.6
18.1
8.6
69.5
3.2
19.4
C2-
C12
0.32
0.27
0.17
0.32
0.61
0.25
0.22
0.25
C2-
C20
0.39
0.32
0.21
0.46
0.78
0.45
0.26
0.46
Change
21.9
17.6
22.7
44.9
28.2
81.6
19.9
81.7
' Aliphatic include paraffmic and olefinic hydrocarbons.
ignition traffic, the addition of the SVHC portion of emissions does not significantly increase the emission rates (8%
to 10% increase). However, the aromatic hydrocarbon emission rate from gasoline vehicles increased more, by -20%,
as the result of including Tenax data.
CONCLUSION
Hydrocarbons in the range of C10-C20 (SVHC) are important components of the total hydrocarbons emitted
from heavy duty diesel vehicles. They should be included in the calculation of diesel vehicle emission rates.
ACKNOWLEDGEMENTS
We would like to thank Dr. R. Zweidinger, Mr. A. McArver, and Mr. J.W. Duncan for their valuable assistance during the Fort McHenry
Tunnel experiment The financial support of Southern Oxidant Study, Coordinating Research Council, AutoOU Air Quality Research
Program, U.S. EPA, and the National Renewable Energy Laboratoiy is gratefully acknowledged.
REFERENCES
1. M.N. Ingalls, "On-Road Vehicle Emission Factors from Measurements in a Los Angeles Area." Presented at 82nd
Annual Meeting, Anaheim, CA. Air & Waste Management Association, Pittsburgh, PA, 1989.
2. M.N. Ingalls, L.R. Smith, and R.E. Kirksey, "Measurement of On-Road Vehicle Emission Factors in the California
South Coast Air Basin - Volume 1: Regulated Emissions." Report No. SWRI-1604 from Southwest Research Institute
to the Coordinating Research Council, Atlanta, 1989.
3. E.M. Fujita, B.E.Croes, C.E. Bennett, et al., I. A&WMA. 1992, 42, pp 264-276.
4. C.V. Hampton, W.R. Pierson, T.M. Harvey, et al., Environ. Sci. Technol.. 1982, 16, pp 287-298.
5. C.V. Hampton, W.R. Pierson, T.M. Harvey, et al., Environ. Sci. Technol.. 1983, 17, pp 699-708.
6. B. Zielinska and K. Fung, "The Evaluation of the Concentration of Semivolatile Hydrocarbons (in the C12-C18
Range) Emitted from Motor Vehicles," in Proceedings of the 1992 U.S. EPA/A&WMA International Symposium on
Measurement of Toxic & Related Air Pollutants. Air & Waste Management Association, Pittsburgh, 1992, pp 883-888.
7. B. Zielinska and K. Fung. Sci. Total Environ., in press, 1993.
8. U.S. Environmental Protection Agency, In Compendium of Methods for Ihe Ceterminalion of Toxic Organic
Compounds in Ambient Air. EPA-600/4-89-017, Office of Research and Development, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
126
-------
10000
1000
100
10
Bore 3 Ft McHenry Tunnel
June 18-24,1992
ullllkkl
1 2 3 4 5 6 7 8 9 10 11 Run No.
Th Fr Fr Sat Sun Sun Mo Tu Tu W Th Day
1230 0300 1600 1200 1200 1600 1100 0300 1330 0400 1600 Hour
Bore 4 Ft McHenry Tunnel
June 18-24,1992
1 23456789 10 11 Run No.
Th Fr Fr Sat Sun Sun Mo Tu Tu W Th Day
1230 0300 1600 1200 1200 1600 1100 0300 1330 0400 1600 Hour
Figure 1. Traffic distribution in Bore #3 and Bore #4 during eleven one-hour runs
(run number, day of the week, and sampling starting time are marked
below the x-axis). Note that the ordinate scales are logarithmic.
127
-------
8ZI
, ethylbenzene
^ m-othy'Hi'^nft
T~^ o-ethyltoluene
g- ^ 1,2,3-trimethylbenzene
1=^— C4-benzene
•^rTi , . ?-mpthylnaphthalpnp
£. 1-methylnaphthalene .,
i-"- n-tetradecane
^— n-pentadecane
%-
\ C1 7 Paraffin
^•phenanlhrene
r*— n-octadecane
I gg"
&— n-nonadecane J2 <•>
f U^ ^
1 °- '
\ CD' H
) (n CD
rn-eicosane — g
x
• n-heneicosane
12
14
18
18
20
22
24
20
3(
3'.
34
36
3
I ethY'benzene m+p.xv,ene
^_
^ o-ethyltoluene
L, n-rterarm ^^1 ,2,4-trimethylbenzene
? ^ 1 ,2,3-trimethylbenzene
s= — C4-benzene
r
b
2-methylnaphthalene
L^.1 -methylnaphthalene
^- n-tetradecane
^— n-hexadecane
^- n-heptadecane
r
K noctadecane
L $°
Pn-nonadecane S 0
b* 0^ CO
y. Tenax artifact
-------
MODELING THE ATMOSPHERIC FORMATION AND DECAY OF GAS AND PARTICLE BOUND NITRO
POLYCYCLJC AROMATIC HYDROCARBONS
Zhihua Fan, Richard M. Kamens, Danhua Chen, Shufen Chen, Stephen Mcdow
Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, NC, 27599-
7400, USA.
ABSTRACT
A method to study and model polycyclic aromatic hydrocarbons (PAH) and nitro-PAH (NPAH) formation
and decay on the gas and particle phase in the atmosphere has been outlined. Diesel exhaust emissions were
added and diluted into a 190 m3 outdoor smog chamber and permitted to age under conditions of daylight and
darkness. Ozone (Oj), nitrogen oxides (NOJ, volatile hydrocarbons in the gas phase were monitored. A
sampling train consisting of annular denuder, filter and polyurethane (PUF) or XAD resin was used for the
collection of gas and particle phase PAH and NPAH. Based upon the results, the current denuder design has
sufficient flow (20 l/min) and adsorption characteristics for collection of PAH and NPAH in the chamber studies.
Sampling artifacts were determined by comparison of the conventional sampling system and denuder system
results. In general, the lower molecular weight PAH, such as naphthalene, acenaphthylene (ACE), fluorene
(FLN) and phenanthrene (PHE) displayed some tendency for volatilization during the sampling process; and
negligible artifacts were observed for higher molecular PAH, such as benzo(a)pyrene (BaP). Three or four-ring
compounds, such as fluoranthene (FL) and pyrene (PY), exhibited both negative and positive artifacts which
depended largely on the sampling conditions and temperature.
A reaction mechanism for PAH and NPAH has been further tested by comparison with chamber results.
Simulations for FL and PY in gas phase were close to chamber observations, but sometimes we could not model
the behavior of particle phase FL, PY or 1NPY. This may occur because PAH and NPAH inside of the particle
and is not available for reaction with sunlight. No particle-associated compounds were found on the denuder,
which implied that 2NFL or 2NPY was deposited on the particle immediately after formation in the photochemical
processes in the gas phase. 2NPY and 2NFL simulations were generally good. Modeling results suggest that
the addition of NOg to the gas phase adduct of FL + OH or PY + OH was the main reaction for NPAH formation.
NPAH loss on particles could be simulated with a photo-induced reaction keyed to solar intensity.
INTRODUCTION
NPAH have been observed in automobile emissions and other industrial processes1-4. It is recognized
that in addition to being directly emitted, NPAH can be formed by the reactions of PAH with gaseous nitrogenous
pollutants5'10. Over the past several years significant advances have been made in our understanding of the
kinetics and reactions associated with the formation and degradation of NPAH in the atmosphere. Most studies
have shown that the gas phase reaction of PAH with OH radicals is the primary processes for the formation of
NPAH in the atmosphere6-7'9'10. To provide a better understanding of PAH and NPAH in the atmosphere, a
mechanism was proposed in 199110 and further modified and evaluated in this study using outdoor chamber
experiments with diesel exhaust (Table 1). Modeling results from this initial effort10 demonstrated that it was
possible to combine reactions of PAH and NPAH with a well established photochemical smog mechanism.
One of the short comings of our initial study was the inability to measure semi-volatile PAH and NPAH
distributed between the gas and "particle" phase14'18. This was because a conventional sampling system with a
filter and a back-up poiyurethane foam (PUF) sorbent was used. Attempts to determine PAH gas/particle
distribution by single channel denuders18 and annular denuders19 have been reported. Vapor phase components
are removed by the denuder prior to the filter, and the paniculate matter is collected on a filter. Any material
"offgassing" from the particulate matter is trapped in an adsorbent which is placed downstream of the filter. More
recently, Gundel et al.20 have developed a denuder coating technique which permits the direct extraction and
analysis of collected organics from the denuder. Based on this work20, we have modified the denuder by adding
more channels and increasing the length to increase the adsorptive capacity. This paper describes: (1) the use of
annular denuders to determine the phase distribution of PAH and NPAH in the smog chamber experiments; (2)
sampling artifacts investigated by comparing the denuder sampling system and the traditional sampling system;
and (3) chamber simulation results with a slightly modified PAH and NPAH mechanism.
EXPERIMENTAL METHODS
Diesel exhaust from a 1967 Mercedes sedan (200D) engine was added and diluted into a 190 m3 outdoor
Teflon film smog chamber located in Pittsboro, NC, USA10. Concentrated diesel exhaust was added to the
chamber for 30-60 seconds. The engine was in an idle mode during the chamber injection process. The particle
concentration in the chamber was in the range of 200-600 ng/m3, and NOX generated from the combustion
ranged from 0.4 to 0.7 ppm. To create active photochemical conditions, extra propylene gas was added into the
chamber after the addition of diesel exhaust. Samples were collected by two 20 cm 4-channel annular denuders
129
-------
in series or a 40 cm 5-channel denuder (ID=1.9 cm with 0.2 cm space), followed by a 47 mm Teflon impregnated
glass fiber filter (type T60A20, Pallflex Products Corp., Putnam, CT) and a 8 x 2.5 cm PUF cartridge or a 4 x 0.9-
cm-i.d. stainless steel tube cartridge filled with XAD-2 (Supelco, Inc., Belfonte, PA; 16/50 mesh). The efficiency
of the 20 cm denuder was greater than 89% for FL and PY at a sampling flow rate of 30 l/min. 85% of FL and PY
were in the gas phase under these conditions. After taking the samples, the denuder was extracted four times
with 30 ml of 2:1 hexane and acetone in field. The denuder then dried with pure nitrogen gas and was reused.
PUF and filter samples were soxhlet extracted in methylene chloride (MeCI2) for 8-12 hours and fractionated via
normal phase high pressure liquid chromatography8-10. The samples subsequently were analyzed on a gas
chromatography/mass spectrometer. The analytic methods are described elsewhere8'10.
RESULTS AND DISCUSSION
Sampling Artifacts
Two afternoon experiments were conducted in July and September, 1992 at our smog chamber facility In
Pittsboro, NC10. The first experiment was performed on July 14,1992. Two 20 cm denuders in series were used
for gas phase PAH and NPAH collection. The sampling flow rate was 30 l/min. The most important observation
was that no 2NFL, 1NPY or 2NPY was found on the denuder. This implied that 2NFL and 2NPY deposited on
particles immediately after they formed in the gas phase. This observation strongly suggested that NPAH
observed on PUF samples in previous experiments resulted from blow-off from the filters. In addition, no other
typically particle associated compounds like benzo(a)pyrene (BaP) and Dibenz(a,h)anthracene (Di-BaA) were
found on the denuder.
A parallel sampling system consisting of a 47 mm Teflon impregnated glass fiber filter and an 8 x 2.5 cm
PUF cartridge was used for the first two samples in this experiment to determine the extent to which sampling
artifacts occurred with the conventional sampling system. The particle concentrations for the denuder system
and the filter/PUF system are compared in Figure 1(a). Extensive volatilization (negative artifact, 30-97%) was
observed for more volatile compounds, such as acenaphthylene (ACE), fluorene (FLN) and phenanthrene (PHE)
(saturated vapor pressure over a pure solid are 6.7 x 10~3 to 6 x 10'4 torr at 25°C); 15-37% negative artifact
occurred for FL and PY (saturated vapor pressure over a pure solid is 9.2 x 10'6 and 4.5 x 10'6 torr, respectively),
and negligible artifacts for higher molecular weight PAH, such as BaP (saturated vapor pressure over a pure solid
is 5.6 x10'9 torr).
Another experiment was carried out on September 13,1992. A 40 cm 5-channel denuder was used for
gaseous PAH and NPAH collection in place of the two 20 crn denuders. Again, in almost all the samples, no
2NFL, 1 NPY or 2NPY was observed on the denuder. The exception was 2NFL, which was found on first denuder
sample. Either gaseous 2NFL formed from some very rapid undetermined process or this sample was
contaminated. No other typically particle-associated compounds were found on the denuder samples. In order to
further evaluate the sampling artifacts, a parallel sampling system with a filter and an 0.9 x 4 cm XAD-2 trap was
used. The filter concentrations in filter+XAD system and in denuder+filter+XAD system are shown in Figure 1 (b),
Negative artifacts were observed for ACE and FLN; and negligible artifacts for less volatile compounds. The
most interesting finding was a 20-80% positive artifact rather than the negative artifacts observed for FL and PY
shown in Figure 1(a). An 18-37% positive artifact was also observed for FL and PY in a winter experiment
(February 10, 1992, temperature was 10°C). Sampling artifacts depend on temperature and sampling
conditions17. The temperature in the September 1992 experiment (30°C) was lower than that (40°C) in the July
1992 experiment; thus, the effect of volatilization was lower in the September experiment. Additionally, the
sampling flow rate was 15 l/min in both the February and September experiments. It was observed that when the
filter face velocity decreased, the adsorption on the filter increased17. Pressure drops across the filter were found
to be linearly proportional to the square of face velocity. As the pressure drop decreased, the adsorption effect
should increase. The face velocity was 28.8 cm/s in the July experiment, and 14.4 cm/s in September and
February experiment. Measurements of the pressure drop across the filter at the 15 l/min and 30 l/min showed
values of 7.35 and 29.1 inches of water, respectively. For this sampling system it appears that for most volatile
compounds, such as ACE and FLN, vapor pressures are so high that volatilization of collected compounds from
filters overwhelms the adsorption effect (positive effect); for moderate volatile compounds, such as FL and PY,
adsorption becomes significant and a positive sampling artifact appears.
Simulation of Chamber Experiments
The first outdoor smog chamber experiment began at 2:30 pm on July 14, 1992. One minute of diesel
soot was injected into 190 m3 smog chamber at 2:37 pm, 0.4 ppmV propylene and 0.6 ppmV NO was added into
the chamber after the addition of diesel soot, and a 260 ng/m3 particle was generated in the chamber. The
maximum concentration of O3 was 0.26 ppmV, and the maximum modeled concentration of OH radical was
1.65x1O'7 ppmV. The experimental data and simulation results for NOX and O3 are shown in Figure 2(a).
A comparison between data and model prediction for gas and particle phase FL and PY is given in Figure
3. Simulation results for gas phase PY and FL agreed very well with observed results, but the model
130
-------
overpredicted the loss rate of particle phase FL and PY. This may be attributed to a nonexchangible fraction of
FL and PY inside of the particles which is unavailable for reaction. Similar behavior for 1NPY was observed in
the chamber. 1 NPY emitted directly from diesel engine and like FL and PY, maybe distributed throughout the
particles. In contrast to 1NPY, particle 2NPY reacted rapidly in the chamber and the simulation results fitted
these data well (Figure 4). This is because most 2NPY was formed in the photochemical processes. After
formation gaseous, 2NPY condensed onto particles and probably remained for the reaction period on or close to
the exterior part of the particles (Figure 4).
To model our experimental data, a model yield for 2NFL of 0.076 from the reaction of FL + OH was used
in the mechanism rather than 0.03 reported by Atkinson et al.11 (Atkinson et al.11, however, reported a factor of 3
uncertainty in this yield). Using this higher yield permitted a lower photolysis rate for 2NFL than previously used10
and this was more consistent with current peri- hydrogen theory of NPAH photostability21. Model results suggest
that addition of NC>2 to the gas phase adduct of FL + OH or PY + OH is the main pathway for 2NFL and 2NPY
formation, and the photolysis is the dominant loss process for NPAH.
Another experiment was carried out on September 13, 1992. Similar initial conditions to the July 14,
1992 experiment were used except higher NOX was added into the chamber. 0.52 ppmV propylene was added
into the chamber at 12:42 pm. Then diesel exhaust was added into the chamber for 2.5 minutes at 1:30 pm, and
600 ng/m3 of suspended particles appeared in the chamber. The total NOX concentration was 1.24 ppm.
Simulation results of the NOX and O3 behavior are shown in Figure 1 (b).
Gas and particle phase PAH simulations are given in Figure 5. Reasonable fits were obtained for gas
phase FL and PY. For particle phase PY, and especially FL, the model predicted much faster decay than what
was observed in the chamber. As in the previous experiment, we have surmised that a significant fraction of FL
might not be available for reaction because it existed inside the particles. In order to confirm this point, we
investigated the degradation of BaA and benzo(b)fluoranthene (BbF) which distribute mostly on the particle phase
(Figure 6). BaA and BbF rapidly decayed at the beginning of the experiment and then appeared to be stable.
This is similar to the behavior of particle FL and PY.
A yield of 0.1 for 2NFL was used in this experiment. Since this yield is at the upper most bound of the
Atkinson proposed yield11, it is possible that some other processes for 2NFL formation is occurring.
ACKNOWLEDGMENTS
This work was supported by a Research Grant (#R816678) from the office of Exploratory Research
USEPA, Cooperative agreements from EPA (#CR819675) and (#CR815152).
REFERENCES
1. D. Schuetzle, F. S-C Lee, T.J. Prater and S.B. Tejada, "The identification of polyruclear aromatic hydrocarbon
(PAH) derivatives in mutagenic fractions of diesel paniculate extracts." Int. J. Environ. Anal. Chem. 9: 93 (1981).
2. T. Nielson, B. Seitz and T. Ramdahl, "Occurrence of nitro-PAH in the atmosphere in a rural area," Atmospheric
Environment 18:2159 (1984).
3. T. Ramdahl, B. Zielinska, J. Arey, R. Atkinson, A.M. Winer and J.N. Pitts, Jr., "Ubiquitous occurrence of 2-
nitrofluoranthene and 2-nitropyrene in air," Nature 321: 425 (1986).
4. J.N. Pitts, Jr., R. Atkinson, J.A. Sweetman and B. Zielinska, "Determination of 2-nitrofluoranthene and 2-
nitropyrene in ambient particulate organic matter: evidence for atmospheric reactions," Atmospheric Environment
19:1601 (1985).
5. J.N. Pitts, Jr., B. Zielinska, J.A. Sweetman, R. Atkinson and A.M. Winer, "Reaction of absorbed pyrene and
perylene with gaseous N2O5 under simulated atmospheric conditions," Atmospheric Environment 19: 911 (1985).
6. J. Arey, B. Zielinska, R. Atkinson, A.M. Winer, T. Ramdahl and J.N. Pitts, Jr., "The formation of nitro-PAH from
the gas-phase reactions of fluoranthene and pyrene with the OH radical in the presence of NOX," Atmospheric
Environment 20, 2339 (1986).
7. J.A. Sweetman, B. Zielinska, R. Atkinson, T. Ramdahl, A.M. Winer and J.N. Pitts, Jr., "A possible formation
pathway for the 2-nitrofluoranthene observed in ambient particulate organic matter," Atmospheric Environment
20:235(1986).
8. R.M. Kamens, J. Quo, Z. Quo and S.R. McDow, "Polycyclic Aromatic Hydrocarbon degradation by
heterogeneous reactions with ^05 on atmospheric particles," Atmospheric Environment 24A, 1161 (1990).
9. J.N. Pitts, Jr., "Nitration of gaseous polycyclic aromatic hydrocarbons in simulated and ambient atmospheres: a
source of mutagenic nitroarenes," Atmospheric Environment 21: 2531 (1987).
10. R.M. Kamens, Z. Fan, Y. Yao, M. Vartiainen, E. Haynes and S.R. McDow, "A methodology for modeling the
formation and decay of nitro-PAH in the atmosphere," Polvcvclic Aromatic Compounds 3: 501 (1993).
11. R. Atkinson, J. Arey, B. Zielinska and S.M. Aschmann, "Kinetic and nitre-products of the gas-phase OH and
N03 radical-Initiated reactions of Naphthalene-do, Fluoranthene-d10, and Pyrene," Int. J. Chem. Kinet. 22: 999
(1990).
131
-------
12. R.J. Zellner, "Recent advances in free radical kinetics of oxygenated hydrocarbon radicals," Chimie. Physigua
84: 403 (1987).
13. R.M. Kamens, H. Karam, J. Fulcher, Z. Quo, J. Perry, D. Saucy and L. Stockburger, Proceedings of the 19B7
EPA/APCA Symposium of Measurement of Toxic and Related Air Pollutants. APVA VIP-8: Pitts, PA, 1987, pp
412-419.
14. B.R. Appel, S.M. Wall, Y. Tokiwa and E.L. Kothny, E.L., "Sampling of carboneous particles in the
atmosphere," Atmospheric Environment 17:1787 (1983).
15. L. Van Vaeck, K.A. Van Cauwenberghe and J. Janssens, "The gas-particle distribution of organic aerosol
constituents: measurement of the volatilization artifact in hi-vol cascade impactor sampling," Atmospheric
Environment 18: 417 (1984).
16. B.R. Appel, W. Cherg and F. Salaymeh, "Sampling of carboneous particles in the atmosphere-ll,"
Atmospheric Environment 23: 2167 (1989).
17. S.R. McDow and J.J. Huntzicker, "Vapor adsorption artifact in the sampling of organic aerosol: face velocity
effects," Atmospheric Environment 24A, 2563 (1990).
18. R.W. Coutant, L. Brown and J.C. Chuang, "Phase distribution and artifact formation in ambient air sampling
for polycyclic aromatic hydrocarbons." Atmospheric Environment 22: 403 (1988).
19. D.A. Lane, N.D. Johnson, S.C. Barton, G.H.S. Thomas and W.H. Schroeder, "Development and Evaluation of
a novel gas particle sampler for semivolatile chlorinated organic compounds in ambient air," Environ. Sci.
Technol. 22: 941 (1988).
20. LA. Gundel, V.Lee. Daisey and K.R.R. Mahanama, Abstracts of the eleventh Annual Meeting of the
American Association for Aerosol Research. 1992, pp 64.
21. J.N. Pitts, Jr., "Formation and Fate of gaseous and particulate mutagens and carcinogens in real and
simulated atmospheres," Environ. Health Perfect. 47:115 (1983).
TABLE 1. PAH AND NPAH REACTIONS
Gas Phase Reactions
1 PY*,n, + OH
2
3
+02
PY(S) + N2°5
4 FL(g) * OH
5 FL(g)-OH + 02
Particle Phase Reactions
8 FL,n, + N,O,
1NPY
(p)
FL(g)-OH + 0.03 x 2NFL
(p)
0.5x2NFL
3NFL,
FL
{+hv}
{+hv}
PY(P) + °3
1NPY(p) {+hv}
2NPY(p) {+hv}
2NFL(p) {+hv>
72000
0.29
0.81
72000
0.29
0.026
0.01
0.01
0.045xkN02
0.02xkN02
0.005
0.005
0.03 x kN02
Reference
11
12
11
11
12
11
8
8
13
13
10
10
10
0.045xk
N02
0.02xkN02
•All gas phase PAH were lost to the walls at a rate of 0.0021 min'1, particle PAH and NPAH were removed at a
rate of 0.0013 min"1.
132
-------
negative
!/
m Filter + PUF in Denuder Syster
en Filter in Rlter + PUF System
July 14, 1992 (a)
tifact
\ no artifact
1 il in an ra-, il^r%
Ba Filter + XAO In Denuder Syst
1=1 Filter In Filter + XAO System
negative artifact S*P<- 1 3. 1 992
-------
18 IB 20
Time in hours (EST)
Time in houra (EST)
Figure 4. Model Simulations and Data of 1 NPY, 2NPY and 2NFL in Particle Phase (July 1 4, 1 992)
Time in houra (EST) Time in houra (EST)
Figure 5. Data and Simulation Results for Gas and Particle Phase FL and PY (Sept. 13,1992)
Time in hours (EST)
Figure 6. Behavior of FL, PY, BaA and BbF
in Particle Phase (Sept. 13,1992)
x
O a.
Time in houra (EST)
Figure 7. Data and Model Simulations
for 2NFL(Sept. 13,1992)
134
-------
EFFECT OF COMBUSTION TEMPERATURE ON THE ATMOSPHERIC STABILITY OF POLYBROMINATED
DIBENZO-p-DIOXINS AND DIBENZOFURANS
Parag Birla and Richard M. Kamens
Department of Environmental Sciences and Engineering
School of Public Health
University of North Carolina
Chapel Hill, North Carolina 27599-7400
ABSTRACT
The incineration of materials containing polybrominated diphenyl ether (PBDPE) flame retardants can potentially
lead to the formation of polybrominated dibenzo-p-dioxins and dibenzofurans (PBDDs and PBDFs). Some of these
compounds may exhibit toxicities similar to those of their chlorinated analogues. Little is known about the atmospheric
stability of PBDDs and PBDFs. In this study PBDDs and PBDFs produced from the combustion of polyurethane foam
(PDF) containing 4.4% PBDPEs were injected into outdoor Teflon film chambers and aged in the presence of sunlight
under typical atmospheric conditions. Incineration experiments with combustion temperatures in the range of 400-450°C
were categorized as "low temperature" experiments and those in the range of 670-780°C as "high temperature"
experiments. PBDDs and PBDFs have been found to occur predominantly in the participate phase, thus filter samples
were collected over time to ascertain the atmospheric stability of these compounds. Production of PBDFs, namely,
tetrabrominated dibenzofuran (TBDF) and pentabrominated dibenzofuran (PeBDF) and decay of tetrabrominated-p-dioxin
(TBDD) was observed in "low temperature" experiments. Production of TBDF and PeBDF is believed to occur from
photolysis of unburned PBDPEs. TBDF.PeBDF and TBDD emissions from "high temperature" experiments were stable.
Validation of PBDF and PBDD behavior was done by collecting paniculate- and gas-phase samples of polycyclic aromatic
hydrocarbons (PAHs). Particle-bound PAHs from "low temperature" experiments degraded while corresponding PAHs
from 'high temperature" experiments were stable. Particle formation and composition from the two kinds of experiments
have been investigated to explain these differences. These observations suggest that under typical incinerator conditions
paniculate-bound emissions of these compounds will be transported over long distances due to long half-lives.
Incineration temperatures around 450°C can lead to unstable emissions with atmospheric half-lives of 1-2 h. The
dependence of the atmospheric stability of incineration-generated pollutants on the combustion temperature is an
important finding of this study.
INTRODUCTION
Incineration has been projected as an attractive waste disposal technology. This projection is underscored by the
fact that since 1979, 3500 landfills have been closed and within the next five years landfill capacity will exist for only 20
million tons for the anticipated 160 million tons of waste generated.1 Concomitant to the likely increased use of
incineration is the growing production of brominated organics, particularly with respect to PBDPEs which is used as fire
retardants in textiles, plastics, carpets and other materials.2 Incomplete combustion of these brominated organics
present in municipal waste can lead to the formation of PBDDs and PBDFs. Therefore, it is expected that atmospheric
concentrations of PBDDs and PBDFs will rise in the future in light of the increasing production of brominated compounds.
While much attention has been given to polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDDs and PCDFs),
animal studies have shown that PBDDs and PBDFs are at least as toxic as their chlorinated analogues.2
Central to biological exposure and uptake of PBDDs and PBDFs is their atmospheric fate and transport.
Atmospheric lifetime, itself can vary depending on the nature of combustion-generated emissions, which is dependent on
combustion variables like combustion temperature, residence time, oxygen concentration and other parameters.
Operating combustion temperatures in incinerators have been found to vary from 300-1300°C, although typical operating
temperatures are around 800°C.3
An important consideration in determination of the atmospheric stability of semi-volatile organics like PBDDs and
PBDFs is their gas-particle partitioning. Theoretical predictions using an equation developed by Junge4 and experimental
observations by Lutes et al.5 revealed that more than 95% of PBDDs and PBDFs occur on paniculate-phase. Thus, the
atmospheric stability of PBDDs and PBDFs is primarily relevant to their occurrence on atmospheric aerosols as opposed
to their presence in gas-phase.
Currently, there exists a paucity of data on photodegradation of PBDDs and PBDFs under realistic conditions.
Laboratory studies have been carried out in organic solutions or on solid substrates. An important observation regarding
these experiments is the rapid decay of 2,3,7,8-TBDD and 2,3,7,8-TBDF in i-octane (half-lives 0.8 min and 0.7 min
respectively) and relative stability on quartz surface (half-lives 32 h and 35 h respectively).6 Given such numbers it
becomes all the more imperative to arrive at representative values for atmospheric half-lives of PBDDs and PBDFs using
outdoor chambers. Preliminary work in this direction revealed that incineration of PDF containing PBDPEs at combustion
temperatures ranging from 640-760°C yielded particulate-bound PBDDs and PBDFs which were relatively stable.5 This
implies that combustion under these conditions yields PBDDs and PBDFs that photodegraded at rates comparable to
solid-phase laboratory experiments. In this study incineration experiments were carried out with PDF containing PBDPEs
135
-------
at combustion temperatures ranging from 400-780°C and atmospheric stability of PBDDs and PBDFs were determined.
PAHs generated from these experiments were also quantified to gain a better insight into the behavior of PBDDs and
PBDFs. Results from a 'low temperature" and a 'high temperature" experiment have been included in this article.
Results were verified by performance of another set of experiments.
EXPERIMENTAL SECTION
Incineration experiments were carried out in an ignition vessel shown in Figure 1. The combustion material
consisted of PUF containing 4.4% w/w of industrial grade DE-71 mixture (40% TBDPE, 57% PeBDPE and 3% HxBDPE).
Experiments carried out with combustion temperatures ranging from 400-500°C were categorized as "low temperature"
experiments and those in the range of 650-800°C as "high temperature" experiments. Emissions from the prototype
incinerator were directed into outdoor smog chambers located in Pittsboro, NC and have been described elsewhere.7
The emissions were allowed to age within the chamber atmosphere for approximately 4 h. During this time, particulate
samples were collected by passing chamber air through a sampling train containing a 47-mm T60 A20 Teflon-
impregnated glass fiber filter, followed by a 4 in. X 1.5 in. PUF cartridge for collection of vapor-phase species. Chamber
ozone, nitrogen oxide, relative humidity and temperature were monitored during the course of an experiment. Total solar
radiation was measured by using a black and white pyranometer and dilution rate was measured by an SFg tracer.
Electrical Aerosol Analyzer (EAA, TSI Inc., MN) and Laser Aerosol Spectrometer (LAS, PMS Inc., CO) were used to
gather data on size distribution of the combustion-generated particles. In addition, nucleopore filter samples were also
collected, which were later analyzed with a Scanning Electron Microscope (SEM) and the resulting photographs were
used to describe particle morphology.
Particulate samples were analyzed for PBDDs and PBDFs in accordance with EPA method 8290.8 Briefly, this
involved extraction of the samples in toluene and enrichment using three sets of columns: silica gel, florisil and
carbon/celite. All laboratory work was carried out under artificial lights, shielded against UV radiation, and samples were
kept wrapped in foil to preclude photodegradation of the analytes in the laboratory. Quantification of f BDDs and PBDFs
was achieved by internal standardization using 13C12 labeled compounds. The samples were analyzed by high resolution
gas chromatography/high resolution mass spectrometry (HRGC/HRMS). The analysis was carried out in the Selected Ion
Monitoring (SI M) mode at a resolution of 10,000. Procedure for PBDPE analysis was similar to the one for PBDD and
PBDF analysis, except that carbon column was excluded in the chromatographic cleanup process9. Samples for PAH
analysis were extracted in dichloromethane and analyzed by HRGC/HRMS in the SIM mode at a resolution of 10,000.
B.B-binaphthyl was used as the internal standard for quantifying PAHs.
RESULTS AND DISCUSSION
PBDD, PBDF & PAH Analysis:
The compounds detected were TBDD, TBDF and PeBDF. Detection of individual isomers was not possible due
to lack of appropriate standards. Hence, results have been presented in terms of the total concentration of a compound
and not in terms of specific isomers. Two "high temperature" experiments were carried out to demonstrate that earlier
results could be repeated. For these experiments, plots of chamber concentrations of TBDD, TBDF and PeBDF versus
chamber aging time had insignificant slopes at a 90% level of confidence. Figures 2a, 2b and 2c include the plots of
these compounds from one of the experiments with a similar trend for the other experiment. Therefore, it can be
concluded that there is no evidence of decay for PBDF and PBDD emissions from "high temperature" incineration
experiments. These results are similar to those obtained by Lutes et al.5 However, in "low temperature" experiments,
there is decay in TBDD concentration yielding half-lives of 1 -2 h. Figure 3a is a plot of TBDD concentration versus time
for one of the "low temperature" experiments. Photodegradation rates in solid-phase and solution-phase laboratory
experiments is somewhat analogous to our "high temperature* and "low temperature" experiments. Figures 3b and 3c
include plots of chamber concentrations of TBDF and PeBDF versus chamber aging time from one of the "low
temperature" experiments. Significant positive slopes imply the production of these compounds over time in the chamber
atmosphere. PAHs exhibited a trend similar to TBDD for "high temperature" and "low temperature" experiments (Figures
4,5,6 &7). In the latter case, PAHs like phenanthrene, fluoranthene and pyrene had half-lives of 1 -2 h. Experimental
conditions for the experiments mentioned in this article are presented in Table 1.
The lack of agreement in TBDD and PAH behavior for the two kinds of experiments may be attributable to
differences in physical and chemical properties of the combustion-generated particles. It was observed that the color of
particles in "high temperature" experiments was dark brown or black in stark contrast to the light brown color in "low
temperature" experiments. Investigators have mentioned that darker substrates can absorb most of the incident light and
reflect little to cause photodegradation of the compounds adsorbed on it.1 ° Also, particles from "low temperature"
experiments may have a higher specific surface area (SSA) to absorb more light, thereby yielding higher rates of
photolysis.11 Drawing an analogy from the laboratory experiments suggests that the presence of an organic layer in 'low
temperature" particles may expedite photodegradation, similar to the solution-phase experiments. An illustration of this
concept is provided in Figure 8. The organic layer may be formed from the PUF vapor that was left uncombusted.
Contrary to degradation of TBDD in "low temperature" experiments, production of TBDF and PeBDF was
observed for these experiments. We believe that this was the result of photolysis of unburned PBDPEs which photolyzed
136
-------
to yield TBDF and PeBDF. Busef12 found that thermolysis of PBDPEs at 510°C caused 10% destruction of the ethers as
opposed to 96-97% destruction at 650°C. Therefore, it is expected that almost 90% of the PBDPEs will be left
uncombusted in our 'low temperature1 experiments. Choudhry et al.13 have reported 14% yield of tetrachloro
dibenzofuran (TCDF) over a 4 h period from photolysis of pentachloro diphenyl ether (PeCDPE) dissolved in cyclohexane.
Extrapolating this to our system facilitated mass balance calculations and revealed that there is degradation of TBDF
(half-life 4h) if the confounding factor due to its production from photolysis of unburned ethers was taken into account.
Comparison of Particle Characteristics:
Analysis of particle distribution data evinced that there was not a significant difference in SSA for the particles
generated from the two kinds of experiments, thus eliminating higher SSA as a plausible explanation for rapid
photodegradation in case of 'low temperature" experiments. Examination of SEM photographs indicated that particle
morphology differs largely for the 'low temperature" and "high temperature" experiments. The former is characterized by
singular and spherical particles, while the latter distinguishes itself by the presence of coagulated masses. This
information was used to prescribe a model for particle formation by taking into account the different combustion
temperatures (Figure 9).^4 According to this model, particle formation in "low temperature" conditions is dominated by
nucleation and condensation. The presence of agglomerates in "high temperature" conditions is the result of formation of
reaction products that coalesced with fly ash or other reaction products to form agglomerates. Such a model for particle
formation in incinerators would also espouse our supposition regarding the presence of an organic layer around "low
temperature" particles.
To further enhance our understanding of particle composition, analysis was carried out to determine fraction of
paniculate mass that is extractable or particulate organic matter (POM). A larger POM should be associated with the
particles generated from "low temperature" experiments as combustion process will not be very efficient leading to a
larger concentration of products of incomplete combustion. POM can be correlated to the presence of an organic layer
around the particles. This draws an analogy from comparison of wood soot (POM 80-90%) and diesel soot particles
(POM 30-40%), wherein the former has a higher extractable mass and probably, an organic layer around it as wood soot
particles lose their integrity when impacted upon a tin foil.15 Percent extractable analysis showed that for "low
temperature" particles the extractable mass was around 46% compared to 30% from "high temperature" particles. A
higher fraction of extractable mass in the "low temperature" particles can be linked to the presence of unburned PUF that
manifested itself in the form of an organic layer around the particles.
CONCLUSIONS
Typical operating conditions in incinerators may lead to air emissions which can be transported over long
distances due to the stable nature of the emissions. This study demonstrates that lower combustion temperatures can
lead to the formation of toxic air pollutants like PBDDs, PBDFs and PAHs that may be relatively unstable in the
environment. The dependence of half-lives of incineration-generated air pollutants on the combustion temperature should
be taken into account for accurate prediction of biological exposure. Experiments should be carried out with a
heterogeneous system as is the case in municipal waste combustors, where the dynamics of particle formation can be
more complex. Characterization of the products of photodegradation will aid in determination of the toxicity of the
products and in postulation of the mechanism of photolysis. It should be stated that while lower combustion temperatures
can be beneficial due to shorter half-lives of toxic air pollutants but they will not lead to significant reduction in waste
volume, one of the primary objectives of incineration.
ACKNOWLEDGMENTS
We want to thank Tom Merz for his help in conducting combustion experiments, G. Dean Marbury for assistance
in mass spectrometry analyses and Dr. Bob Bagnell for assistance in scanning electron microscopy. Funding for this
project was received from The Office of Exploratory Research, Washington, D.C. (Grant* R-817534).
REFERENCES
1. J. E. Wingerter, "Are landfills and incinerators part of the answer? three viewpoints," EPAJ. 15:22-23 (1989).
2. Chlorinated Dioxins and Dibenzofurans in Perspective: C. Rappe, G. Choudhry and L. H. Keith, Eds.; Lewis Publishers
Inc., Chelsea, pp 485-499, 1986.
3. E. Benefati, F. Gizzi, R. Reginato et al., "Polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans in
emissions from an urban incinerator. 2. Correlation between concentration of micropollutants and combustion conditions,"
CJiernpsphere 12(9):1151-1157 (1983).
4.Fate of Pollutants in the Air and Water Environment; I. H. Suffet. Ed.; Wiley Interscience, New York, 1977, pp 7-25.
5. C. C. Lutes, M. J. Charles, J. R. Odum and R. M. Kamens, "Chamber aging studies on the atmospheric stability of
polybrominated dibenzo-p-dioxins and dibenzofurans," Enviro. Sci. Technol. 26(5) :991 -998 (1992).
6. H. R. Buser, "Rapid photolytic decomposition of brominated and brominated/chlorinated dibenzodioxins and
dibenzofurans," Chemgsphere 17(5): 889-903 (1988).
7. D. A. Bell and R. M. Kamens, "Photodegradation of wood smoke mutagens under low NOX conditions," Atmos. Enviro.
137
-------
20(2):317-321 (1986).
8. J. S. Stanley and T. M. Sack, Protocol for the Analysis of 2.3.7.8-Tetrachlorodibergo-p-dioxin bv High-Resolution n^
Chromatooraphv/ High-Resolution Mass Spectrometrv. EPA/600/4-86/004, 1986.
9. L. M. Smith, D. L Stalling and J. L. Johnson, "Determination of part-per-trillion levels of polychlorinated dienzofurans
and dibenzodioxins in environmental samples," Anal. Chem. 56(9):1830-1842 (1984).
10. W. A. Korfmacher, E. L. Wehry, G. Mamantov and D. F. S. Natusch "Resistance to photochemical decomposition of
polycyclic aromatic hydrocarbons vapor-adsorbed on coal fly ash," Enviro. Sci. Technol. 14(9):1094-1099 (1980).
11. T. Behymer and R. Hites, "Photolysis of polycyclic aromatic hydrocarbons adsorbed on simulated atmospheric
particulates," Enviro. Sci. Technol. 19(10):1004-1006 (1985).
12. H. R. Buser, "Polybrominated dibenzofurans and dibenzo-p-dioxins: thermal reaction products of polybrominated
diphenyl ether flame retardants," Enviro. Sci. Technol. 20(4):404-408 (1986).
13. G. G. Choudhry, G. Sundstrom, L. 0. Ruzo and O. Hutzinger, "Photochemistry of chlorinated diphenyl ethers," J.
Aaric. Food Chem. 25(6) :1371-1375 (1977).
14. V. Sethi and P. Biswas, "Modeling of particle formation and dynamics in a flame incinerator," J. Air Waste Manaoe.
Assoc. 40(1)142-46 (1990).
15. S. R. McDow, Q. R. Sun, M. Vartiainen et al., "Factors Influencing Photodegradation of Polycyclic Aromatic
Hydrocarbons in Liquid Organic Media," in Proceedings of the 1991 EPA/A&WMA International Symposium on
Measurement of Toxic and Related Air Pollutants. VIP-21, Air & Waste Management Association, Pittsburgh, 1991, pp
140-146.
TABLE 1. COMPARISON OF EXPERIMENTAL CONDITIONS
COMBUSTION TEMP. (°C)
CHAMBER TEMP. (°C)
RELATIVE HUMIDITY (%)
AVG. TOT. SOLAR RADIATION
(cal./cm^/min.)
[NO] (ppm)
[N02] (PPm)
[03] (ppm)
INJECTION TIME
"LOW"
TEMP.
470
23.9-27.4
42-39
0.86
0-0.016
0.01-0.026
0-0.041
10:15 a.m.
"HIGH"
TEMP.
780
19.5-22.2
10-13
0.63
0-0.074
0.01-0.061
0-0.06
ll:00a.m.
Emissions directed to the Smog Chamber
HOD Grid*
Bunur —
:=
— e-
-*«P
i
3=
-»
is=
SlalnlM. Steel Wall
Insulation
*£
1 U 1 O
TPhsifler
" PUF Sampk to be Combmted
Figure 1. SECTION OF THE IGNITION VESSEL
138
-------
(a) TBDD
0.2 0.8 1 1.4 1.8 2.2 2-8 3 3.4 3.8
TIME SINCE INJECTION (hrs)
• .
! (b)TBDF
0.2 0.8 1 1.4 1.B 2.2 2.6 3 3.4 3.B
TIME SINCE INJECTION (hrs)
3
E 10
I '
(c) PeBDF
02 O.B 1 1.4 1.B 2.2 2.6 3 3.4 3.6
TIME SINCE INJECTION (hrs)
Figure 2. CHAMBER CONCENTRATIONS OF TBDD, TBDF & PeBDF
FOR 'HIGH TEMPERATURE- EXPERIMENT
ff „
i.
ff «
(a) TBDD
TIME SINCE INJECTION (hrs)
(b)TBDF
TIME SINCE INJECTION (hrs)
(c) PeBDF
TIME SINCE INJECTION (hrs)
Figure 3. CHAMBER CONCENTRATIONS OF TBDD, TBDF & PBBDF
FOR "LOWTEMPERATURE" EXPERIMENT
0.2 0.6 1 1.4 1.B 2.2 2.8 3 3.4
TIME SINCE INJECTION (hrs) TIME SINCE INJECTION (hrs)
Figure 4. CHAMBER CONCENTRATION OF PHENANTHRENE VERSUS Figure 5. CHAMBER CONCENTRATION OF PHENANTHRENE VERSUS
AGING TIME FOR 'HIGH TEMPERATURE' EXPERIMENT AGING TIME FOR 'LOW TEMPERATURE" EXPERIMENT
139
-------
1
TIME SINCE INJECTION (hra.)
O FLUORANTHENE+ PYRENE
Figure 6. CHAMBER CONCENTRATIONS OF FLUORANTHENE
AND PYRENE VERSUS AGING TIME FOR
•HIGH TEMPERATURE' EXPERIMENT
0.2
0.6
1.8
'
2.6
3.4
TIME SINCE INJECTION (hra.)
a FLUORANTHENE* PYRENE
Figure 7. CHAMBER CONCENTRATIONS OF FLUORANTHENE
AND PYRENE VERSUS AGING TIME FOR
•LOW TEMPERATURE" EXPERIMENT
•LOW TEMPERATURE" PARTICLE
_ Organic layer with
PBDDs, PBDFs.ete, dissolved
-Aerosol Core
PBDD, PBDF.ete, molecules
"HIGH TEMPERATURE" PARTICLE
PBDD. PBDF.etc. molecules
'adsorbed on the aerosol core
.erosol Core
Figure 8. POSTULATED COMPOSITION OF PARTICLES GENERATED
FROM COMBUSTION EXPERIMENTS
Increasing Temperature
(Saturation Ratio < 1)
Burner
Vaporization of PDF and/or
Formation of Products by
Chemical Reactions
PUFVapor = = =
Fly Ash /„•»
ReactionOo0
Products O O
Decreasing Temperature
(Saturation Ratio > 1)
Formation and Growth
*• oQo Formation by Nucleation
Q'ft " » Growth by Condensation
Growth by Coagulation
Figure 9. SCHEMATIC OF THE BURNER SHOWING PARTICLE FORMATION
AND GROWTH MECHANISMS
140
-------
A METHOD FOR STUDYING HETEROGENEOUS PHOTOCHEMICAL REACTIONS OF
POLYCYCLIC AROMATIC HYDROCARBONS ON ATMOSPHERIC COMBUSTION AEROSOLS
Elizabeth Ann Hayes, Stephen R. McDow, Richard M. Kamens
Department of Environmental Sciences and Engineering, University of North Carolina, CB# 7400,
Chapel Hill, 27599
ABSTRACT
A photochemical turntable reactor was used to study photodegradation of polycyclic aromatic
hydrocarbons in organic coatings on carbonaceous particle surfaces. This required the development of a
technique to coat the particles with an organic layer and to evenly disperse the particles on a filter
substrate. The technique has potential for use in determining differences in reaction rate constants,
reaction products and reaction mechanisms. Previous researchers have noted that polycyclic aromatic
hydrocarbon photodegradation can be strongly influenced by particle surfaces. Preliminary results
suggest that benz(a)anthracene and benzo(a)pyrene photodegradation can also be influenced by other
organic constituents associated with the particle coating.
INTRODUCTION
Polycyclic aromatic hydrocarbons (PAH) on wood smoke react rapidly in sunlight, but are stable
at night1. Photodegradation mechanisms of PAH associated with combustion particles are poorly
understood. Although surface characteristics appear to strongly influence photodegradation of adsorbed
PAH on silica and fly ash2, little is known about the effects of solid elemental carbon surfaces
associated with more common U.S. PAH sources such as diesel soot or wood smoke. Moreover, recent
experiments indicate that organic compounds associated with wood smoke also strongly influence PAH
photodegradation rates. It follows that co-adsorbed reactive organic compounds or a particle organic
layer should also be considered in experiments designed to measure photodegradation in organic
coatings on particle surfaces. This work describes the development of an experimental technique to
compare relative rates of PAH photodegradation in organic coatings on particle surfaces.
APPROACH
A conventional photochemical turntable reactor for comparing relative photochemical reaction
rates in solutions was adapted for studying reactions on atmospheric particles. This was accomplished
by distributing the particulate matrix on filter surfaces and analyzing changes in concentration after
irradiation at certain time intervals. Filters were mounted on specially constructed test-tube shaped
filter holders which were placed in the turntable slots normally used to hold test tubes. This approach
required a separate filter for each sample. For example, if samples were analyzed at one hour intervals
for four hours, comparison of two particle coatings required preparation of eight filters for irradiation.
A significant effort was therefore required to develop a technique to produce even organic coatings on
the model particle surfaces and even dispersal of particles on the filters.
A model system was selected to test this approach. Carbon black particles were coated with an
organic mixture consisting of isoeugenol ( 2 methoxy - 4 propenyl phenol), fluoranthene,
benz(a)anthracene, benzo(k)fluoranthene, and benzo(a)pyrene. The system is simpler than real wood
smoke or diesel soot to simplify measurement of photodegradation rates, analysis of products, and tests
for reaction mechanisms. However, it retains the basic characteristics of many PAH containing
combustion sources because it contains a solid, carbonaceous surface and a reactive organic compound
which influences PAH photodegration rates. Isoeugenol was chosen as the reactive organic species
because of its abundance in wood smoke-* and its known influence on PAH photodegradation4.
141
-------
EXPERIMENTAL
Figure 1 illustrates the methodology and apparatus used in this study. Carbon black particles
coated with PAH or an iseougenol/PAH solution mixture were dispersed on Gelman* Zefluor* Teflon
membrane filters (Fisher Scientific) using vacuum filtration. Five irradiation experiments were carried
out to investigate the influence of isoeugenol on PAH photodegradation in the presence of a
carbonaceous surface. These are summarized in Table 1, which describes experiment duration,
sampling interval, multicomponent mixtures used, and concentrations of the components in the mixture.
The organic mixture consisting of PAH and isoeugenol was dissolved in dichloromethane and
sonicated for 1 minute. The dichloromethane solvent was evaporated, leaving a coating on the particles.
The coated particles were suspended in 25 ml dichloromethane above a 47 mm Zefluor Teflon
membrane filter in a Buchner funnel. The suspension was vacuum filtered, leaving the coated particles
evenly distributed on the filters.
The particle loaded filters were then mounted on an ACE Glassware Photochemical Turntable
Reactor equipped with a water cooled 450W high pressure Hg lamp. After irradiation the filters were
extracted with dichloromethane in microsoxhlet extractors for 12 hours. Extracts were concentrated to 1
ml using a Kuderna-Danish apparatus, transferred to a tared vial, and evaporated to almost dryness
under a stream of nitrogen. 100 to 400ul of acetonitrile were added and the samples were analyzed by
reverse phase HPLC and GCMS.
In three of the experiments outlined in Table 1, photodegradation rates of PAH on carbon black
were compared in the presence and absence of isoeugenol. In the fourth and fifth experiments
photodegradation rates were compared between PAH dissolved in isoeugenol and guaiacol (2 methoxy
phenol) coated on carbon black.
In the first experiment the lamp was placed in a borosilicate immersion well which filtered out
radiation below 300nm. In the second and third experiments the 366 nm line of the Hg spectrum was
isolated using optical filters and a quartz immersion well. The fourth and fifth experiments were
performed using the borosilicate well. Similar experiments were carried out in the dark to verify that
PAH decay was due to photodegradation and not volatilization or other chemical reactions.
Preliminary experiments indicated that PAH photodegradation proceeded more rapidly for PAH
coated on bare Zefluor filters than on carbon black. It was consequently important to verify PAH and
isoeugenol were not simply coated on the filter, but were in contact with the carbon black. To address
this, recovery experiments were designed in which coated carbon black particles were deposited on a
number of filters using the coating technique described above, which were designated as reference
samples. On a second "filter only" set of an equal number of filters the same PAH solution was coated
directly on to the filter by vacuum filtration. The fraction of PAH adsorbed on the filter, instead of the
carbon black, is estimated from a "recovery ratio" defined as the ratio of the fraction of original PAH
recovered from the "filter only" samples to that recovered from the reference samples. Coated samples
were also inspected by scanning electron microscopy (SEM).
RESULTS AND DISCUSSION
During the course of the research it became apparent that the concentrations of PAH and
isoeugenol had a strong influence on whether the PAH solution was in contact with the carbon black or
adsorbed on the filter. The filter/carbon black recovery ratios increased with increasing PAH/carbon
black mass ratios, suggesting that as the capacity of the carbon black to sorb the PAH is approached,
significant filter adsorption occurs. This was supported by SEM photographs, which showed that when
isoeugenol mass was 30% of the carbon black mass, significant amounts of isoeugenol solution was
coated on the filter rather than the carbon black. It was difficult to determine how much isoeugenol was
coated on the filter at lower concentrations, but 30% appeared to be an upper limit for the isoeugenol to
remain in contact with the particle surface. On the basis of the recovery and SEM experiments,
isoeugenol amounts were maintained below 10% of the carbon black mass. The PAH concentration in
142
-------
the stock solution was less than 1000 ng/ul. 400ul of stock solution was applied to Img of carbon black.
The carbon black retains 2-5% of the PAH following vacuum filtration, resulting in an adsorbed PAH
concentration of less than 20 ng/mg.
The results to the irradiation experiments give a preliminary indication that isoeugenol affects
PAH photodegradation rate even in the presence of a carbonaceous surface. In the first experiment,
decay was faster than for the next two experiments because more light was transmitted to the samples
through the borosilicate well than when the optical filters were used. Decay appeared to be initially
very rapid but ceased well before the end of the experiment. The initial rapid decay probably occurred
because of the high light intensity. It is possible that photodegradation stopped early in the experiment
because extractable PAH were hidden from exposure to light in the interstices of the carbon black.
Benz(a)anthracene appeared to react more rapidly in the presence of isoeugenol. Benzo(a)pyrene was
not recovered from carbon black particles which did not contain isoeugenol. In subsequent experiments
a longer soxhlet extraction time of 12 hours enhanced the recovery of BaP from bare carbon black.
In the second and third experiments the observed reaction rates were slower because the optical
filters only transmitted the 366nm wavelength which resulted in a decreased light intensity. The
photodegradation in these experiments is described in figures 2 and 3. The rate constants were tested
for parallelism5 by comparing the first order rate expression for PAH photoreactivity on carbon black
with isoeugenol present to the first order rate expression in the absence of isoeugenol for the same
experiment. Rejection of the hypothesis of parallelism indicates the difference in the rate constant is
statistically significant. In the second experiment the difference was not statistically significant for
benz(a)anthracene (p>.l) but was borderline for benzo(a)pyrene (.05
-------
preliminary experiments reported here concerning filter adsorption, more effort is still needed to verify
that significant photodegradation is not occurring in PAH adsorbed to the filter substrate itself.
CONCLUSIONS
A technique was developed to compare relative photodegradation rates of PAH adsorbed to
particulate matter on filters mounted on a photochemical turntable reactor. The technique can be used
to investigate systems which are simpler than actual combustion particles, but which retain the
characteristics of a solid carbonaceous surface coated with an organic layer. This potentially simplifies
determination of rate constants, reaction products and reaction mechanisms. The technique was tested
by comparing photodegradation of PAH in solutions of isoeugenol coated on carbon black to PAH
coated on carbon black without isoeugenol. Preliminary results suggested that benz(a)anthracene and
benzo(a)pyrene photodegradation was more rapid in isoeugenol solutions even in the presence of a
carbonaceous surface. This is consistent with previously reported results of rapid PAH
photodegradation in toluene solutions of isoeugenol and PAH. This suggests that although particle
surface characteristics can strongly affect PAH photodegradation, organic constituents coated on particle
surface are also important. Several improvements were identified which would further development of
the technique to produce more conclusive results.
ACKNOWLEDGMENTS
This work was supported by a Research Grant, contract number CR-816678 from the Office of
Exploratory Research, USEPA.
REFERENCES
1. R. M. Kamens, Z. Guo, J. Fulcher, and D.A. Bell, "Influence of humidity, sunlight, and temperature
on the daytime decay of polyaromatic hydrocarbons on atmospheric soot particles," Environ. Sci.
Technol. 22: 103-108(1988).
2. T. Behymer, and R. Kites, "Photolysis of polycyclic aromatic hydrocarbons adsorbed on fly ash,"
Environ. Sci. Technol. 22: 1311-1319(1988).
3. S.B. Hawthorne, M.S. Krieger, D.Jf. Miller, and M.B. Mathiason, (1989). "Collection and
quantitation of methoxylated phenol tracers for atmospheric pollution from residential wood stoves."
Environ. Sci. Technol. 23, 470-475 (1989).
4. S.R. McDow, Q.R. Sun, M. Vartiainen, Y.S. Hong, T.L. Yao, E.A. Hayes, and R.M. Kamens,
"Photodegradation of polycyclic aromatic hydrocarbons in hexadecane and methoxyphenols," in
Polvcvclic Aromatic Compounds, P. Garrigues, and M. Lamotte, eds., 1993, pp. 111-118
5. D.G. Kleinbaum, L.L. Kupper, and K.E. Muller, Applied regression analysis and other multivariable
methods. 2nd ed., PWS-Kent, Boston, 1988, pp 266-274.
144
-------
Table 1. Summary of 5 Experiments
Compound Initial Coating Concentration
(ug/mg)
Photodegradation Rate
Constants (hour1)
First Experiment, 8/11/92, Borosilicate Well, l.Shr duration, .5 hr intervals
w/ no iso w/ iso w/ no iso w/ iso
Ratio Significance
of of Rate
Rates Change p<.05
Isoeugenol
H
BaA
BkF
BaP
0
.43
.18
.36
none detected
45
.37
.15
.18
.25
.28
.20
.04
1.7
.69
.27
.10
.18
2.5
1.4
2.5
yes
yes
yes
Second Experiment, 8/28/92, Quartz Well w/ light filter, 3hr duration, .75 hr intervals
Isoeugenol
Fl
BaA
BkF
BaP
w/ no iso
0
4.5
1.8
2.0
3.0
w/iso
13.5
4.1
1.4
1.8
2.9
w/ no iso
.24
.08
.02
.02
w/ iso
.89
.13
.08
.14
.13
0.5
1.0
7.0
6.5
no
no
yes
yes
Third Experiment, 9/10/92 Quartz Well w/ light filter, 6hr duration, 1.5 hr intervals
Isoeugenol
Fl
BaA
BkF
BaP
w/no iso
0
4.8
1.8
2.3
2.8
w/iso
277
5.9
1.7
2.1
2.8
w/no iso
.11
.05
.01
.01
w/iso
.23
.15
.09
.04
.04
1.4
1.8
4.0
4.0
no
yes
yes
yes
Fourth Experiment, 2/05/93, Borosilicate Well, 3hr duration, .33 hr intervals
Guaiacol
Isoeugenol
Fl
BaA
BkF
BaP
w/ guiac
23
0
3.8
1.5
1.7
23
w/iso
0
28
3.9
1.4
1.7
2.5
w/ guiac
.94
.04
.06
.03
.01
w/iso
.57
.21
.18
.05
.02
5.2
3.0
1.7
2.0
no
yes
no
no
Fifth Experiment, 2/19/93, Borosilicate Well, 6hr duration, .5 hr intervals
Guaiacol
Isoeugenol
Fl
BaA
BkF
BaP
w/ guiac
2.5
0
.40
.10
.23
.35
w/iso
0
21
.46
.09
.25
.41
w/ guiac
.44
.07
.05
.01
.01
w/iso
.35
.09
.04
.04
.03
1.3
0.8
4.0
3.0
no
no
yes
yes
145
-------
Figure 1. Apparatus used to prepare, expose to light, and extract organic coated carbon black particles
Buchner Funnel Filtration Apparatus
Figure 2. Photodegradation of BaA and BaP coadsorbed with and without isoeugenol on CB, 8/28/92
BaA ln(c/cO) vs t on CB in TT
8/28/92, quartz well w/ filter
BaP ln(c/cO) vs t on CB in TT
8/28/92, quartz well w/filter
0.8
2.0
time, hours
no too coadsorbed + w/ isouegenol coads
1.8
time, hours
a no iso coadsorbed + w/ isouegenol coads
Figure 3. Photodegradation of BaA and BaP coadsorbed with and without isoeugenol on CB, 9/10/92
BaA Ln(C/CO) vs T on CB in TT
9/10/92, quartz well with filter
BaP ln(C/CO) vs T on CB in TT
9/10/92. quartz well with fitter
no iso coadsorbed + with iso coadsorbed
time, hours
o no iso coadsorbed + with iso coadsorbed
146
-------
Mechanistic and Kinetic Studies of the Photodegradation of Benz(a)anthracene in the Presence of
Methoxyphenols
Jay R. Odum, Steven R. McDow, and Richard M. Kaniens
Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, NC 27599- 7400.
Abstract - Kinetic studies were employed to assess an empirical rate taw describing the rate ofphotodegradation of
polycyclic aromatic hydrocarbons (PAH) in tlie presence of substituted methoxyphenols. A solution of benz(a)anthracene
(BaA) and vanillin in toluene was chosen as the model system. A rate law and corresponding rate constants were
determined for this system: d[BaA]ldt= kA[BaA]+kB[BaA] [Vanillin], where kA=0.024 ±0.005 hr'1 and kB=81.17±3J9M~
1hr'1. Further experiments using structure-reactivity relationships were applied to tlie model system to investigate the
mechanism for BaA photodegradation. Data from t/iese experiments suggest that tlie rate determining step in the
mechanism is hydrogen abstraction of tJiephenolic hydrogen from vanillin.
Introduction
Photochemical reaction is an important removal process for atmospheric polycyclic aromatic hydrocarbons (PAH).-1'2
PAH photochemical reactivity in organic solvents has been studied extensively for over 40 years, and singlet oxygen has
been determined as the most important oxidant.^ There has also been a considerable amount of research conducted on
gas-phase PAH photo reactivity, and it is known that hydroxyl attack is the major oxidation pathway.7 >** However many of
the most mutagenic PAH in the atmosphere are primarily associated with atmospheric participate matter.^ PAH associated
with combustion source particles are also photochemically reactive,10'11 but there has been very little work done in
determining mechanisms of photodegradation for particulate bound PAH because information on particle composition is
generally incomplete and varies considerably from source to source.
Reported photolysis rates for particlulate bound PAH vary widely. 12-14 jj j,as Deen suggested that one reason such
variations exist is that the rates are strongly influenced by the chemical and physical surface properties of the associated
particles." These properties vary significantly depending on the source, fuel, and combustion conditions under which the
particles are generated. For example, 30-60% of the carbon associated with diesel soot is elemental.1*" The rest is
comprised mostly of nonpolar aliphatic and aromatic organics.17 In contrast, as much as 90% of the carbon associated
with wood smoke particles is organic.18 This organic carbon is much more polar than that associated with diesel soot and
contains large amounts (120-300ug/mg of particulate carbon) of methoxylated phenols. Wood soot particles also forma
viscous liquid when collected by impaction and it has been suggested that PAH might be dissolved in a liquid layer
comprised of these organics on the particle.20'21
It was previously reported by members of this group20 that methoxyphenols greatly enhance the photodegradation of
PAH in solution. Thus, since methoxyphenols comprise such a large fraction of the organic material associated with wood
smoke19, it seems reasonable that they may participate in the degradation of PAH associated with wood soot. This is in
contrast to previous ideas that suggest that particulate PAH degradation is due solely to reaction with gasphase oxidants or
simple photolysis. Therefore, it is desirable to try to gain an understanding of the mechanism by which methoxyphenols
participate in this reaction and to formulate a method to obtain a rate law and rate constants for several PAH and
methoxyphenols. In this paper we will discuss the experiments used to develop this methodology and present a rate law,
rate constants, and possible mechanism for a model system~benz(a)anthracene and vanillin in a toluene solution.
Experimental Section
All photodegradation experiments were carried out in a merry-go-round reactor (Ace Glassware, Vineland, NJ)
equipped with a 450 W medium pressure mercury arc lamp in a quartz immersion well and a 366nm filter. The entire
apparatus was submerged in a 18 gal watcrbath that was maintained at a constant temperature of 16.5 ± 0.5°C by pumping
the water (flow rate ~ 75ml/min) through exterior copper coils that were submerged in a 20 gal ice bath. Samples were
irradiated in 13mm x 100mm quartz test tubes (Ace Glassware). Light intensity was periodically measured with a
ferrioxalate actinometer20 and averaged around 2.2 ± 0.3 xlQ!6 photons/sec.
All quantitative analysis was conducted on a Hewlett-Packard 5890 gas chromatograph (GC) equipped with a J&W, 30
m, 0.32 mm i.d., DB-5 column with a 0.25 um film thickness and interfaced to a Hewlett-Packard 5971A Mass Selective
Detector. All injections were performed on column, and the acquisition mode was selected ion monitoring. The
temperature program was: 110° C for 2 min, 110-220° C at 12° C/min, 220-270 at 6° C/min, 270-300° C at 20° C/min,
hold at 300° C for 1 min.
Experiments designed to assess structure-reactivity relationships were performed by preparing four sets of samples
simultaneously. Each set consisted of four samples and one standard. All solutions were prepared in optima grade toluene
(Fisher T291-4). The concentration of benz(a)anthracene (AJdrich B220-9) in all sets was 1 x 10'5 M. The final volume of
147
-------
each samples was 5 ml. Each set contained a different para-substituted guaiacol. Set 1 contained 1 x 10"3 M
acetovanillone (Aldrich Al.080-9). Set 2 contained 1 x 10"3 M 4-hydroxy-3-methoxy benzonitrile (Aldrich 16,260-4).
Set 3 contained 2 x 10"4 M 4-nitroguaicol (Aldrich 32,682-8). Set 4 contained 1 x 10"3 M vanillin (Aldrich V110-4).
Four 5 ml samples from each set were irradiated simultaneously. The rate of decay of BaA was monitored by removing
one sample from each set from the reactor every hour for four hours and quantitating on the GC/MSD.
Rate law experiments were conducted by preparing four sets of samples simultaneously. All solutions were prepared in
optima grade toluene. The concentration of BaA in all samples was 1 x 10"5 M. The concentration of vanilin in set 1 was
2x 10"3 M, in set 2 was 1 x 10"3 M, in set 3 was 5 x 10"4 M, and in set 4 was 2 x 10"4M. The rate of decay of BaA was
monitored by removing one sample from each set from the reactor every hour for four hours and quantitating on the
GC/MSD.
Results and Discussion
It was previously determined by this group2^ that the photodegradation of BaA exhibits pseudo-first order rate behavior
when samples are flooded with vanillin (i.e. concentration of vanillin is at least 20 times higher than the concentration of
BaA). The rate law that describes this behavior is:
d[BaA]/dt=kobs[BaA]
To elucidate the form of kobs, an experiment was performed to examine the relationship between vanillin concentration
and kotls. Four sets of samples, each flooded with a different amount of vanillin, were irradiated in the photoreactor
simultaneously. The data is listed in Table 1.
TABLE I. Correlation between Vanillin Concentration and
SET# [Vanillin] (M) [BaA] (M) knha (hr1)
1 2 x ID'3 1 x 10"5 0.176 ± 0.017
2 1 x 10-3 1 x 10's 0.097 ± 0.003
3 5 x 10"4 1 x lO'5 0.058 ± 0.006
4 2 x 10"4 1 x 10"s 0.044 ± 0.004
Aplotof kot>s versus vanillin concentration gives a straight line (r2=0.995) with a slope of 81.17 ± 3.19 M'1 hr"1 and a y-
intercept of 0.024 ± 0.005 hr"1 (Figure 1). Since this plot yields a straight line with a positive intercept, the rate equation
for BaA photodegradation in the presence of vanillin is:
d[BaA]/dt =kA[BaA] + kB[BaA][Vanillin]
where kA = y-intercept and kB = slope. This type of treatment requires that vanillin concentration be relatively constant
over the course of the experiment. In all four sets vanillin degradation was less than 15%.
The form of this rate law implies that there are at least two mechanisms responsible for BaA photodegradation in this
system. The ratio of the rate constants suggests that the mechanism involving vanillin does not influence the overall rate
unless the concentration of vanillin exceeds 3 x 10"5 M (5 ng/ul). At vanillin concentrations above 3 x 10"3 M (500 ng/ul),
the vanillin dependent mechanism becomes the only important pathway for BaA degradation. Considering that PAH are
known sensitizers of singlet oxygen and that attack of singlet oxygen is the major photodegradation pathway for PAH in
organic solvents3'6, it seems reasonable to assume that the term independent of vanillin concentration is due to singlet
oxygen attack on BaA. Furthermore, benz(a)anthracene-7,12 dione, which is a known product of singlet oxygen addition
to BaA, has been identified as a reaction product by matching its mass spectrum to a standard. However a singlet oxygen
mechanism of this type generally produces second order kinetics.3"6 Thus further study is needed to confirm the
mechanism that is responsible for the first term in the rate law. It is interesting to note that Hawthorne et al22 have
observed the phototoxidation of isoeugenol to vanillin on wood soot particles. This process occurs by singlet oxygen
oxidation in solution.23 Thus perhaps singlet oxygen oxidation is occuring in the organic liquid layer on wood soot
particles despite its insignificant contribution as a gas-phase oxidant.
Given the above rate expression, an attempt was made to determine the mechanism responsible for the second term
in the rate law. In 1972, T. Matsuura documented that triplet sensitizers were capable of abstracting phenolic hydrogen
from catechol and hydroquinone derivatives.23 This precedence suggested that in our model system BaA may be acting as
a triplet sensitizer and abstracting the phenolic hydrogen of vanillin to create a phenoxy radical. Bordwell and Cheng"
showed that electron donating and electron withdrawing para-substituents were able to stabilize phenoxy radicals. They
found that a linear free energy relationship existed between the stabilizing effects of electron donating substituents and
148
-------
their a+ parameters. The o+ parameters are defined as the log of the ratio of the acid dissociation constants for different
para substituted t-cumyl chlorides to the acid disassociation constant for unsubstituted t-cumyl chloride. They are a
measure of a substituent's ability to stabilize an electron deficient reaction site, that is para to the substituent, by coming
into direct, through resonance with the reaction site. The relationship that Bordwell and Cheng found suggest that o+
parameters are measure of phenoxy radicals stability. Thus in order to determine if BaA was abstracting hydrogen from
vanillin to create phenoxy radicals, an experiment was performed to see if BaA rates of photodegradation in various para-
substituted methoxyphenols could be correlated with the a parameters corresponding to those para-subs tituents. However
in this case, electron withdrawing substituents were used instead of electron donating substituents. So o" parameters were
used instead of a+ because a " parameters are a measure of an electron withdrawing para-substituent's ability to come into
direct, through resonance with a para-reaction site.
Four different sets of samples were irradiated in the photoreactor simultaneously. Each set contained BaA and a
different para-substituted guaicol. A rate constant (kg) was obtained for each set. The data is listed in table 2.
TABLE 2. Correlation between BaA Photodegradation in the Presence of Various Para-substituted Methoxyphenols
and the a * Parameters Corresponding to those Para-substituents.
SET# Methoxyphenol a - kR (hr1)
1
2
3
4
Acetovanillone
4-hydroxy-3-mcthoxy
bcnzonitrile
4-nitroguaiaco]
Vanillin
0.84
0.88
1.24
1.03
0.043 ± 0.006
0.032 ± 0.003
0.465 ± 0.025
O.OS3 ± 0.006
A Hammet style plot of log(kg) versus the O~ parameters corresponding to the para-substituted groups yields a straight line
(r^ = 0.96) (Figure 2). The O ~ parameters seem to be a relative gauge of the amount of stability that the corresponding
para-substituents yield to a phenoxy radical. For example, NO2 has a relatively high O" parameter (O"=1.24). This is due
to its strong ability to delocalize the odd electron in the corresponding phenoxy radical by coming into direct resonance
with the reaction site (Figure 3). This delocalization stabilizes the radical product and thus decreases the free energy of the
reaction compared to that of an unsubstituted guaiacol. This decrease in free energy is related to the rate through the
Hammett equation:
AGJ0! - AC; = -2.303RTCT"
cT=log(kN02/kH)
where AG^ is the free energy difference between the ground state complex and the transition state, k^Q2 and kjj are the
second order rate constants for BaA photodegradation (i.e. kg) in the presence of 4-nitroguaiacol and guaicaol respectively,
R is the ideal gas constant, and T is the temperature. This type of relationship shows that the rate increases with O~. As
stated earlier, Bordwell and Cheng observed^ that a correlation existed between phenoxy radical stability and jsc is the quantum yield for
inter-system crossing and I0 is the light intensity at 366nm. The constants ^ and k^ represent the rate of triplet de-
excitation and the rate of hydrogen abstraction respectively.
149
-------
The rate law for this mechanism is also shown in Figure 4. This mechanism can be easily fit to the observed rate law.
The only assumption that is needed is k2 » k3[Van]. By substituting in values for Io, G8"*, (jjisc, kj, and kB, a value of
0.4M'1 is obtained for the ratio of k2/k3. Thus for the vanillin concentrations used in this study (i.e. < 2 x 10~3M), k2 is
much larger than k3[Van]. The value used here for <|>|sc is for BaA in pure benzene -and is assumed to be very similiar to
what it would be in toluene. The variance is surely less than an order of magnitude. So, the assumption needed to make
the proposed mechanism fit the observed rate law is valid.
Summary and Conclusions
The findings of this study suggest that the mechanism by which BaA photodegrades in the presence of vanillin is by
hydrogen abstraction of the phenolic hydrogen of vanillin by excited triplet BaA. The mechanism proposed can be easily
fit to the observed rate law. Thus these findings suggest that if PAH are actually dissolved in an organic layer that contains
large amounts of melhoxyphenols on wood soot particles, then then model system used in this study may be directly
applicable to PAH photodegradation on wood soot particles. If rate constants were obtained for several PAH and
methoxyphenols using the methodology set out in this study, then modeling the atmospheric fate of PAH associated with
wood soot particles would be possible. The rates in solution will obviously be different from those on wood soot. The
quantum yield for inter-system crossing(jsc) and the rate of triplet de-excitation (Jtj) are both medium dependent.
Furthermore, if the organic layer on wood soot particles is extremely viscous then the rate of hydrogen abstraction^) may
be diffusion controlled. This diffusion controlled rate would be temperature dependent and this dependence would have to
be taken into account. Yet despite these differences, if the mechanism responsible for PAH photodegradation on wood
soot particles is known then it would be possible to model their behavior.
This study suggest that further research on heterogenous reactions in the atmosphere should be conducted and that
particular attention be applied to the chemical composition of combustion aerosols.
Acknowledgement
This work is supported primarily by the U.S. E.P.A. office of exploratory research under Contract No. CR-816678 with
significant contributions from Ford Motor Company, and the Finland-U.S. Educational Exchange Commision.
Literature Cited
1. Nielsen, Y.; Ramdahl, T.; and Bjorseth, A. Environ. Health Perspectives. 1983, 47, 103-114.
2. Finlayson-Pitts, B. J. and Pitts, J. N. Atmos. Chem. John Wiley & sons, New York; 1986.
3. Bowen, E. J. Disc. Faraday Soc. 1953, 14, 143-146.
4. Livingston, R.; Rao, V. S. J. Am. Chem. Soc. 1959, 63, 794.
5. Bowen, E.]. Advances in Photochemistry. 1963,1,23.
6. Lee-Ruff, E.; Kazarians-Moghaddam, H.; Katz, M. Can. J. Chem. 1986,64,1297-1303.
7. Sweetman, J. A.; Zielinska, B.; Atkinson, R.; Ramdahl, T.; Winer, A. M.; Pitts, J. N., Jr. Atmos. Environ. 1986,20,
235-238.
8. Atkinson, R.; Arey, J.; Zielinska, B.; Aschmaim, S. M. Int. J. Chem. Kinetics. 1990, 22, 999-1014.
9. Yamasaki, H.; Kuwata, K.; Miyamoto, H. Environ. Sci. Technol. 1982, 16,189-194.
10. Nielsen, Y. Atmos. Environ. 1988, 22, 2249-2254.
11. Kamens, R. M.; Guo, Z.; Fulcher, J.; Bell, D. A. Environ. Sci. & Technol. 1988, 22, 103-8.
12. Kamens, R. M.; Fulcher, J.N.; Zhishi, G. Atmos. Environ. 1986, 20, 1579.
13. Kites, R. A.; Lafiamme, R. E.; Windsor, J. G., Jr. Adv. Chem. Series. 1980, 185, 289.
14. Korfmacher, W. A.; Wehry, E. L.; Mamantov, G.; Natusch, D. F. S. Environ. Sci. & Technol. 1980,14,1094.
15. Behymer, T. D. and Kites, R. A. Environ. Sci. & Technol. 20(1988)1311-1319.
16. Japer, S. M.; Szkarlat, A.C.; Gorse, R.A.; Heyerdahl, E.K.; Johnson, R.L.; Rau, J.M.; and Huntzicker, JJ. Environ.
Sci. & Technol. 18(1984)231-234.
17. D. Schuezle, F.S.C. Lee, T.J. Prater, and S.B. Tejada, International Journal ofEnvr. Anal. Chem. 9 (1981) 93-144.
18. J.A. Rau, Aerosol Science Technology 10 (1989) 181-192.
19. Hawthorne, S. B.; Krieger, M. S.; Miller, D. J.; Mathiason, M. B. Environ. Sci. & Technol. 1989, 23,470-475.
20. McDow, S. R.; Sun, Q. A.; Vartiainen, M.; Hong, Y. S.; Yao, Y. L.; Hayes, E. A.; Kamens, R. M. Polycyclic
Aromatic Hydrocarbons, P Garrigues and M. Lamotte, eds., 1993,111-118.
21. Rounds, S. A.; Tiffany, B. A.; Pankow, J. F. Environ. Sci. & Technol. 1993, 27, 366-377.
22. Hawthorne, Steven B.; Miller, David J.; Langenfcld, John J.; Krieger, Mark S. Environ. Sci. & Technol. 1989,26,
2251-2262.
23. Matsuura, T.; Matsushima, H.; Kato, S.; Saito, I. Tetrahedron. 1972,28,5119-5129.
24. Eskins, K. Photochem. and Photobiol. 1979,29,609-610.
25. Bordwell, F. G.; Cheng, J. P. /. Am. Chem. Soc. 1991, 113, 1736-1743.
150
-------
16 IB 20
Figure 1: Correlation of Observed Rate Constant
with Vanillin Concentration
Figure 2: Hammet Plot for p-Substituted
Guaiaools
,OCH3
Figure 3: Resonance Stabilization of Phenoxy Radicals by Electron Withdrawing (NO2)
and Electron Donating Groups (HC=CHCH3)
151
-------
Figure 4: Hydrogen Abstraction Mechanism
Products
Rate = k3[3BaA"][Van]
Rate = k3[Van]
if K2»k3[Van]
-d[BaA]
[l.<
L "^2
,[BaA
^^[BaA]
+ kg[Van]
[Van] whe
1 1 .1 rr(C [BaA] +€™ [VanD
\A/h^r^ If. — .
vrm 1^ k3 In^aa H*tec ^1
152
-------
Session 4
Measurement and Monitoring
of Toxics, 03 and PAN
-------
Requirements For The Establishment Of
Enhanced Ozone Monitoring Networks
Nash O. Gerald, William F. Hunt, Jr.,
Geraldine Dorosz-Stargardt, and Neil H. Frank
U. S. Environmental Protection Agency
Office of Air Quality Planning and Standards (MD-14)
Research Triangle Park, North Carolina 27711
ABSTRACT
In accordance with the Clean Air Act Amendments of 1990, the Environmental Protection Agency (EPA) has
developed rules for the establishment of enhanced ozone monitoring networks, or Photochemical Assessment Monitoring
Stations (PAMS), in ozone nonattainment areas designated as serious, severe, and extreme. These stations will collect
amhient air measurements for a target list of volatile organic compounds (VOC) including several carbonyls, and oxides
of nitrogen (NO, NO2, and NO,), ozone, and both surface and upper air meteorological measurements.
EPA anticipates that the data produced by the PAMS will enhance the ability of the State and Local air pollution
control agencies to identify and respond to ozone nonattainment conditions by developing and implementing responsible,
cost-effective ozone control strategies. Further, the Agency anticipates that the measurements will be highly valuable in
verifying emission inventories and corroborating area-wide emissions reductions. The data will be used to evaluate,
adjust, and provide input to the photochemical grid models utilized by the States to develop and demonstrate the success
of their control strategies. The PAMS will provide constructive information for the evaluation of population exposure
and the development of ambient ozone and ozone precursor trends. This paper will examine the regulatory requirements
of the rules and the implications for implementation.
155
-------
INTRODUCTION
On March 4, 1992, the Environmental Protection Agency proposed amendments to the ambient air quality
surveillance rules (40 CFR Part 58) to provide for the enhanced monitoring of ozone and oxides of nitrogen and for the
additional monitoring of volatile organic compounds (including carbonyls) to comply with the requirements of Title I,
Section 182 of the Clean Air Act Amendments of 1990. These proposed modifications were proffered to obtain more
comprehensive and representative data on ozone air pollution. Subsequently, following an extended public comment
period, on February 12, 1993, the final rules were promulgated in the Federal Register (58 FR 8452). These
regulations require the affected States to adopt and implement a program to improve ambient monitoring activities and
the monitoring of emissions of oxides of nitrogen and volatile organic compounds and require States to establish
Photochemical Assessment Monitoring Stations (PAMS) as part of their SIP monitoring networks in ozone nonattainment
areas classified as serious, severe, or extreme. Additionally, each State implementation plan (SIP) for the affected areas
must be amended to include provisions for such ambient monitoring. The principle reasons for requiring the collection
of additional ambient air pollutant and meteorological data are, primarily, the historical challenges faced by the State and
Local Government air pollution control agencies in attaining the National Ambient Air Quality Standards (NAAQS) for
ozone nationwide, and secondly, the need for a more comprehensive air quality database for ozone and its precursors to
explain the effects of ozone control strategies.
PROGRAM OBJECTIVES
In formulating the PAMS program, EPA has endeavored to provide a sensible balance between the costs of the
program and the degree to which the program objectives are satisfied. The Agency has maintained that in formulating
the data requirements for the PAMS program, it was necessary to accept some compromises, (i.e., some more crucial
objectives would be better satisfied than other less important objectives). EPA is committed to requiring a program
which would comprise the best technical-fiscal stability to maximize the utility of a variety of program objectives. EPA
has only provided the framework for a minimum required monitoring strategy; States are encouraged to implement
larger, more comprehensive networks if those networks will provide a superior or equivalent database for the fulfillment
of the program objectives.
The primary objective of the enhanced ozone monitoring revisions is to provide an air quality data base that will
assist air pollution control agencies in evaluating, tracking the progress of, and, if necessary, refining control strategies
for attaining the ozone National Ambient Air Quality Standard (NAAQS). Ambient concentrations of ozone and ozone
precursors will be used to make attainment/nonattainment decisions, aid in tracking VOC and NO,, emission inventory
reductions, better characterize the nature and extent of the ozone problem, and prepare assessments of air quality trends.
In addition, data from the PAMS will provide an improved data base for evaluating photochemical model performance,
especially for future control strategy mid-course corrections as part of the continuing air quality management process.
The data will be particularly useful to States in ensuring the implementation of the most cost-effective regulatory
controls.
Specific provisions of the rule require the establishment and operation of up to 5 PAMS stations in each affected
Metropolitan Statistical Area or Consolidated Metropolitan Statistical Area (MSA/CMSA), depending on the population
of the area (See Figure 1). Those stations are identified by number and defined as follows:
o Site tt\ - These sites are established to characterize upwind background and transported ozone/precursor
concentrations entering the area and will identify those areas which are subjected to incoming transport
of ozone. The #\ Sites are located in the predominant morning upwind direction from the local area of
maximum precursor emissions and at a distance sufficient to obtain urban scale measurements.
Typically, these sites will be located near the upwind edge of the photochemical grid model domain.
o Site #2 - These sites are established to monitor the magnitude and type of precursor emissions in the
zone where maximum precursor emissions are expected to impact and are suited for the monitoring of
urban air toxic pollutants. The #2 Sites are located immediately downwind (using the same morning
wind direction for locating Site #1) of the area of maximum precursor emissions and are typically
placed near the downwind boundary of the central business district to obtain neighborhood scale
measurements representative of the MSA/CMSA. Additionally, a second #2 Site may be required
depending on the size of the area, and should then be placed in the second-most predominant morning
wind direction.
156
-------
o Site #3 - These sites are intended to monitor maximum ozone concentrations occurring downwind from
the area of maximum precursor emissions. Locations for #3 Sites should be chosen so that urban scale
measurements are obtained. Typically, these sites are located 10 to 30 miles downwind from the fringe
of the urban area.
o Site #4 - These sites are established to characterize the extreme downwind transported ozone and its
precursor concentrations exiting the area and will identify those areas which are potentially contributing
to ozone transport into other areas. The #4 Sites are located in the predominant afternoon downwind
direction from the local area of maximum precursor emissions at a distance sufficient to obtain urban
scale measurements. Typically, these sites will be located near the downwind edge of the
photochemical grid model domain.
Each station will sample for speciated volatile organic compounds (VOC), often including several carbonyls,
and ozone, oxides of nitrogen, and surface (10-meter) meteorological parameters; the network requirements vary
somewhat with the size of the MSA/CMSA (See Figure 2). Additionally, each area must monitor upper air meteorology
at one representative site. The rule allows a 5-year transition or phase-in schedule for the program at a rate of at least
one station per area per year. Further, the rule provides for the submission and approval of alternative network designs
and sampling schemes. Such alternative mechanisms for compliance with the rules are especially valuable to States
which are currently engaged in some different form of ozone precursor monitoring which has proved adequate for their
SIP needs.
Specific and often different monitoring objectives are associated with each specific PAMS monitoring location.
These monitoring objectives can be summarized into categories to support the following activities: control strategies,
photochemical modeling, emissions inventories, trends, attainment/nonattainment decisions, and exposure analyses. A
monitoring network which adequately supports these six objectives will provide the initial stepping stones that constitute
a pathway toward attainment of the National Ambient Air Quality Standard (NAAQS) for ozone.
Objective #1: Provide a speciated ambient air data base which is both representative and useful for ascertaining
ambient profiles and distinguishing among various individual VOC.
A fundamental objective of the enhanced ozone and ozone precursor monitoring regulations is to provide a
mechanism whereby air pollution control agencies can obtain an air quality database that will assist in evaluating,
tracking the progress of, and, if necessary, refining control strategies for attaining the ozone NAAQS. This
comprehensive data base will allow the States to focus their control strategies where they will be the most beneficial to
attain the NAAQS and to re-evaluate their existing ozone control programs. These PAMS data, especially those
collected at Sites #1 and #2, will enhance the characterization of ozone concentrations and provide critical information on
the precursors which cause ozone. Speciation of measured VOC data and additional NO, data are expected to allow the
determination of which species are most affected by local emissions reductions and assist in developing cost-effective,
selective VOC and/or NOX reductions and control strategies.
Objective #2: Provide local, current meteorological and ambient data to serve as initial and boundary condition
information for photochemical grid models.
The PAMS network requirements are tailored to provide specific data measurements which can be utilized by
photochemical modelers to refine their estimates of initial and boundary conditions, provide a means to evaluate the
predictive capability of the models, and minimize the available adjustment of model inputs. Such information will tend
to increase the probability that the model's calculations will reflect the "right answer for the right reason" rather than the
"right answer for the wrong reason" and reduce the uncertainties associated with estimated model inputs. In fact, the
upwind site (Site #1) and the downwind site (Site #4) are located so as to quantify the atmospheric conditions at the
upwind and downwind extremes of the photochemical modeling domain.
Heretofore, the national air pollution control program has not had the benefit of comprehensive ozone precursor
data as a tool for evaluating, calibrating, performing diagnostics, or otherwise adjusting and conducting reality checks on
the operation of the Urban Airshed Model (UAM). EPA views the PAMS networks as vital steps forward in
complementing grid model applications.
157
-------
Objective #3: Provide a representative, speciated ambient air data base which is characteristic of source
emission impacts.
The emissions inventory serves as an essential element of the air management process as well as a hindameDtal
input for photochemical models. Verification of reported emissions and the tracking of changes in the atmospheric VOC
profiles can assist in the evaluation of control strategy effectiveness. Given that the emissions inventory is the
foundation building block for the entire SIP development process, it is critical that its accuracy be optimized. While the
regulatory assessments of progress will be made in terms of emission inventory estimates, the ambient data can provide
independent trends analyses and corroboration of these assessments which either verify or highlight possible errors in
emissions trends indicated by the inventories. The ambient assessments, using speciated data, can gauge the accuracy of
estimated changes in emissions. The speciated data can also be used to assess the quality of the speciated VOC and NO,
emission inventories. Utilizing other computer modeling techniques, PAMS data will help resolve the roles of
transported and locally emitted ozone precursors in producing an observed exceedance and may be utilized to identify
specific sources emitting excessive amounts of precursors.
Objective #4: Provide ambient data measurements which would allow later preparation of pollutant trends
assessments.
Long-term PAMS data will be used to assess ambient trends for speciated VOC, NOX> and in a more limited
way, for toxic air pollutants. Multiple statistical indicators will be tracked, including ozone and its precursors during the
events encompassing the days during each year with the highest ozone concentrations, the seasonal means for these
pollutants, and the annual means at representative locations. The more PAMS that are established in and near
nonattainment areas, the more effective the trends data will become. Note, however, that in general it will only be
appropriate to combine data from like sites; therefore, trends will likely need to be constructed on a site-by-site or
combination-of-like-sites basis. As the spatial distribution and number of ozone and precursor monitors grows, trends
analyses will be less influenced by instrument or site location anomalies. The requirement that surface meteorological
monitoring be established at each PAMS will help maximize the utility of these trends analyses by comparisons with
meteorological trends, and transport influences.
Objective #5: Provide additional measurements of selected criteria pollutants.
Like SLAMS and NAMS data, PAMS data will be used for monitoring ozone exceedances and providing input
for attainment/nonattamment decisions. Additionally, the NO2 data can be utilized to augment monitoring for compliance
with the NAAQS for NO2 where such data is gathered with the Federal Reference Method (FRM) and is taken on a
year-round basis. Ultimately, the success of any air pollution control strategy is appraised by its ability to achieve
compliance with the NAAQS. (Note that the PAMS will expand the spatial coverage of NAAQS-related monitoring.)
Although the data at any PAMS site can be used for these purposes, it is expected that Site #3 will constitute the
maximum ozone concentration site for comparison with the NAAQS. Further, the additional data will provide an
expanded foundation for developing and administering maintenance plans required by the Clean Air Act.
Objective #6: Provide additional measurements of selected criteria and non-criteria pollutants from properly-
sited locations.
PAMS data can be used to better characterize ozone and toxic air pollutant exposure to populations living in
serious, severe, or extreme areas. Annual mean toxic air pollutant concentrations can be calculated to aid in estimating
the average exposure to the population associated with individual VOC species, which are considered toxic, in urban
environments. Specifically, by measuring the VOC targeted by PAMS, a number of toxic air pollutants will also be
measured. Although compliance with Title I, Section 182 of the Clean Air Act Amendments does not require the
measurement and analysis of additional toxic air pollutants, the Agency believes that the PAMS stations can serve as
cost-effective platforms for a limited enhanced air toxics monitoring program. The adjunct use of PAMS for air toxics
monitoring will allow the consideration of air toxics impacts in the development of future ozone control strategies. The
establishment of a second PAMS Site #2 in an MSA/CMSA will provide an even better data base for such uses. Both
Sites tn. and #3 will probably be the best choices for exposure analyses. EPA notes that the PAMS network is not ideal
as a source of primary ambient air toxics data and regards the collection of air toxics data as an incidental and
secondary, though important, objective of the PAMS system.
158
-------
ALTERNATIVE PLANS
During the process of developing the initial proposal for the photochemical assessment monitoring program,
EPA noted that each air pollution control agency is subject to its own particular set of problems, strategies, limitations,
and authorities. In many cases some of these factors may be sufficiently unique to require tailoring of the PAMS
specifications in order to balance the program objectives for a particular geographic area. Accordingly, in the final rule,
the Agency incorporated provisions which would allow such tailoring via the development of alternative PAMS
monitoring schemes. Such provisions permit alternatives for the number and location of monitoring sites, the
sampling/analysis methodology utilized at each site, the frequency of sampling, the specification of wind directions for
siting purposes, and the stipulation of the monitoring season. In great part, the approval of alternative plans rests upon
the State's ability to provide a balanced response to the program objectives, ensuring a focus on three key elements, i.e.,
development and evaluation of ozone control strategies, data assistance for photochemical modeling, and tracking of
emissions/trends.
Current EPA thinking on alternative plans includes:
o requiring narrative justification for alternatives with supporting data
o entertaining optional programs at Sites #1, #3, and #4.
o establishing higher hurdles for options at Sites #2.
o requiring submittal of historical sampling data, or
o requiring side-by-side sampling and analyses
CONCLUSION
With the promulgation of the PAMS rules, EPA has endeavored to enter a new era of national monitoring
management. The program will not only rely on State and Local air pollution control agencies to operate the monitoring
systems and report the data, as in the past, but will also encourage innovative thinking in the design, operation,
management, and use of proactive monitoring strategies. The Agency believes that PAMS team building will provide
both Federal and State/Local government with a forum to produce an environmental data base which will be unequaled
in usefulness and quality.
BIBLIOGRAPHY
N. J. Berg, et al., Enhanced Ozone Monitoring Network Design and Siting Criteria Guidance Document. EPA
450/4-91-033, U. S. Environmental Protection Agency, Research Triangle Park, 1991.
Code of Federal Regulations. Title 40, Part 58, U. S. Government Printing Office, 1992.
Shao-Hang Chu, "Using Windrose Data to Site Monitors of Ozone and Its Precursors", U. S. Environmental Protection
Agency, Research Triangle Park, Draft, 1992.
Federal Register (57 FR 7687), "Ambient Air Quality Surveillance - Proposed Rule", March 4, 1992.
Federal Register (58 FR 8452), "Ambient Air Quality Surveillance - Final Rule", February 12, 1993.
W. F. Hunt, Jr. and N. O. Gerald, "The Enhanced Ozone Monitoring Network Required by the New Clean Air Act
Amendments", 91-160.3, Air and Waste Management Association, Vancouver, 1991.
M. E. Kantz, G. J. Dorosz-Stargardt, and N. O. Gerald, Photochemical Assessment Monitoring Stations: Program
and Data Quality Objectives. U. S. Environmental Protection Agency, Research Triangle Park, Draft, 1993.
L. J. Purdue, D. P. Dayton, J. Rice, and J. Bursey, Technical Assistance Document for Sampling and Analysis of
Ozone Precursors. EPA 600/8-91-215, U. S. Environmental Protection Agency, Research Triangle Park, 1991.
159
-------
©
MAXIMUM OZONE
SECONDARY \ CENTRAL BUSINESS DISTRICT -
MORNING
WIND
'URBANIZED FHNOE
PRIMARY AFTERNOON
WIND
PRIMARY MORNING WIND
Figure 1. PAMS Network Design
MINIMUM NETWORK REQUIREMENTS
POPULATION OF MSA/CMSA.
LESS THAN 500,000
500,000
TO
1,000,000
1,000,000
TO
2,000,000
GREATER
THAN
2,000,000
FREQ
TYPE
AorC
A/Dor
C/F
AorC
B/E
AorC
AorC
B/E
B/E
AorC
AorC
B/E
B/E
AorC
A«C
SITE LOCATION
(1)
(2)
(1)
<2)
(3)
(1)
<2)
(2)
(3)
(1)
(2)
(2)
(3)
(4)
[ VOC SAMPLING FREQUENCY REQUIREMENTS |:
Type
A
B
C
Requirement
8 3-Hour Samples Every Third Day
1 24-Hour Sample Every Sbdh Dav
8 3-Hour Sampled Everyday
1 24-Hour Sample Every Sixth Day (year-round)
8 3-Hour Sauries 5 Hi-Event/Previous Days 4 Every Sixth Day
1 24-Hour Sample Every Sixth Day
Type
D
E
F
CARBONYL SAMPLING FREQUENCY REQUIREMENTS |j
Requirement
8 3-Hour Sample* Every Third Day
8 3-Hour Samples Everyday
8 3-Hour Samples 5 Hi-Event/Previoua Daya & Every Sixth Day
MINIMUM PHASE-IN
YEARS AFim
PROMULGATION
1
2
3
4
5
NUMBER OF
SITES DICTATING
1
2
3
4
5
OPEKAHNO
arc LOCATION
RECOMMENDATION
2
2.3
1,2,3
1,2,3,4
1,2,2,3,4
Figure 2. PAMS Network Requirements
160
-------
SURVEYS OF THE 189 CAAA HAZARDOUS AIR POLLUTANTS:
I. ATMOSPHERIC CONCENTRATIONS IN THE U.S.
Thomas J. Kelly, Mukund Ramamurthi, Albert J. Pollack, and C. W. Spicer
Battelle, 505 King Avenue, Columbus, OH
Jitendra Shah and Darrell Joseph
G2 Environmental, Portland, OR
Larry T. Cupitt
U.S. Environmental Protection Agency, Research Triangle Park, NC
ABSTRACT
This paper describes a survey of the ambient air concentrations of the 189 Hazardous Air Pollutants
(HAPs) listed in the 1990 Clean Air Act Amendments. The purpose of this survey was to establish typical
ranges and average values for HAPs concentrations in the U.S., as a first step in assessing the exposure of
the U.S. population to these chemicals. This survey found that the 189 HAPs consist of one group of
chemicals frequently measured in many sampling locations, and a second much larger group for which few or
no ambient data exist. For 74 chemicals no ambient data were found, and for 45 others fewer than 100
measurements were found. Oxygenated and nitrogenated organics dominate the group for which no data were
found; this is probably due to the difficulties in sampling and analyzing for these compounds at ambient
levels. For a list of 30 high-priority HAPs, data are more plentiful than for most of the HAPs.
INTRODUCTION
To accelerate the pace of identifying and regulating toxic air contaminants, Congress established a list of
189 chemicals designated as Hazardous Air Pollutants (HAPs) in the 1990 Clean Air Act Amendments
(CAAA). Those 189 HAPs are a diverse group, including volatile organic compounds (VOCs), polar VOCs,
pesticides, semivolatile compounds, and metals. This report presents the results of a survey of ambient
concentration data for the 189 HAPs (1). The purposes of this survey are to provide typical data from
populated areas of the U.S. with which initial health risk assessments may be done for the 189 HAPs, and to
highlight HAPs for which ambient measurements are particularly needed.
SURVEY PROCEDURES
The 189 diverse chemicals designated as HAPs were organized into chemical classes to facilitate
searching for ambient data. This chemical classification was useful because similar chemicals are frequently
measured together, using similar measurement methods. Information on ambient concentrations of the 189
HAPs was located through keyword searches of appropriate computerized databases, in review articles,
reference books, proceedirtgs of relevant air-quality conferences, and in unpublished data sets from recent
urban air monitoring studies. Ambient concentrations for 70 of the 189 HAPs have been compiled through
1987 in the National VOC Data Base (2,3), and that data base was updated concurrently with the present
program (4). For the present study, the ambient data in the 1988 version of the national data base (2,3) were
summarized, and were supplemented with ambient data from recent field studies. The search strategy for the
119 HAPs not included in the National VOC Data Base was somewhat different. The 119 HAPs were the
subject of computerized and manual searches of the literature to locate ambient data. For each chemical, a
keyword search was conducted through the computerized databases of STN International. The databases
searched included the Chemical Abstracts (CA) files from 1967 to the present, Chemical Abstracts Previews
(CAP) current files, and National Technical Information Service (NTIS) files from 1964 to the present. The
search strategy targeted keywords such as "ambient or urban or atmospheric", and "measurements or
monitoring or concentration", and "air". The strategy also specifically excluded keywords such as
"workplace" or "biological" that were not pertinent to this survey. The search was restricted to English
language citations authored in the U.S., in order to focus on data pertinent to toxics exposure of the U.S.
population.
161
-------
The intent of this review was not to catalog every data point or sample, as attempted in the National
VOC Database (2,3). Rather, the aim was to compile information on typical concentrations (i.e., mean
and/or median), the range of concentrations observed, and the number, locations, and time periods of the
measurements. Additional information such as the detection limit of the measurements, the number of results
below the detection limit, and the procedure used for calculation of a mean value, was also recorded when
available. The focus of this survey was on ambient data in populated (urban to rural) areas of the U.S. To
that end, data from remote sites and data indicating direct source sampling were excluded from the survey.
In a few cases identification was ambiguous, and scientific judgment was used to exclude data points which
were notably higher than the upper range of other data. No effort was made to exclude all measurements that
may have been subject to some impact of urban sources, since those data properly represent the upper range
of concentrations to which urban residents may be exposed. Efforts were made to assure data quality by
selecting from well-documented recent measurements.
The list of 189 HAPs includes some redundant entries, in the form of chemical groups (e.g., xylenes,
cresols) and their individual constituent isomers. These chemicals may be used in industrial settings as the
mixed isomers, but are generally measured in the atmosphere as individual isomers. Searches were
performed for both the individual and mixed isomers, but ambient data were found for only the individual
isomers. The HAP denoted as polycyclic organic matter (POM) is comprised of numerous individual
compounds, and the compounds measured are not always clearly defined in reports of ambient measurements.
For consistency, and to emphasize potential health risks from POM, this survey focused on eight individual
POM compounds identified as possible or probable human carcinogens (5,6). Those eight compounds are
benzo[a]pyrene, benzo[a]anthracene, dibenzo[a,h]anthracene, chrysene, benzo[b]fluoranthene,
benzo[k]fluoranthene, indeno[l,2,3,c-d]pyrene, and benzo[g,h,i]perylene. Ambient data were compiled for
the sum of these eight POM compounds.
RESULTS AND DISCUSSION
The results of this survey of ambient concentrations are compiled in a 33-page table, with an associated
list of over 80 citations from relevant literature (1). The data table lists all 189 chemicals in the same order
as in the CAAA and gives the name and CAS number for each compound, the locations and years of
measurements, the number of measurements, the mean, range, and median (if available) of the measured
data, the number of the pertinent reference in the associated reference list, and additional comments on the
data, such as the number of non-detects. In some cases the number of locations and number of samples were
not evident from the literature. In those cases the numbers were estimated. Some studies failed to state the
detection limit, or to define the number of measurements below that limit. The value assigned to non-detects
(e.g., zero, half the detection limit, etc.) in calculating a mean value was also not often clearly stated. These
inconsistencies were addressed by inferring or estimating the detection limits and number of non-detects from
information in the literature. Mean values were then calculated assuming half the apparent detection limit for
the results below the detection limit.
The most noticeable feature of the data is the wide variation in the amount of data found for individual
HAPs. The number of sampling locations varies from zero to over 140, and the number of samples varies
from zero to over 10,000. Of particular importance is that the number of samples is zero for 74 of the
HAPs, i.e., no ambient data were found. In considering the distribution of the HAPs by number of sampling
locations, the greatest frequency is found for zero sampling locations, with the 74 chemicals in this category
comprising nearly 40% of the HAPs list. The second largest frequency is for 1-4 sampling locations, again
indicating the scarcity of data for some compounds. Only 72 chemicals (38% of the list) show data from at
least 5 locations, and only 54 (29%) show data from 10 or more locations.
Figure 1 shows the frequency distribution of HAPs by the number of measurements found, and clearly
indicates the wide range in the availability of ambient data for the HAPs. The 74 chemicals for which no
ambient data were found constitute the largest frequency range in Figure 1, and a total of 119 chemicals
(63% of the list) show less than 100 measurements. However, the second largest frequency range is the 40
chemicals for which over 1,000 measurements were found. This observation illustrates the primary
characteristic of the HAPs list from the CAAA: it is a unique mix of some chemicals frequently measured in
ambient air, and others rarely or never measured.
162
-------
It is instructive to explore what types of chemicals predominate among those HAPs for which no
ambient data were found. That subject is addressed in Figure 2, which shows the total number of HAPs and
the number with no data found, for each of the chemical classes used in the data search. As expected, for
classes such as hydrocarbons, aromatic compounds, and their halogenated analogs, data are available for
nearly all of the chemicals. These compounds are common toxic, relatively non-polar VOCs, and are readily
measured in ambient air by methods such as EPA Compendium Method TO-14. In contrast, no data are
available for most of the chemicals in the nitrogenated and oxygenated organic classes. This fact is
particularly important because together these two groups comprise half of the HAPs list.
Several reasons may exist for the scarcity of ambient measurements of some chemical classes. For the
nitrogenated and oxygenated organics, which collectively fall under the definition of polar VOCs, the most
likely reason is the lack of sampling and analysis methods for these compounds. Due to their water solubility
and reactivity, measurement of these chemicals at expected ambient levels (ppbv to sub-ppbv) is more difficult
than measurement of VOCs, and consequently methods for such chemicals are still in development (7). This
survey indicates, therefore, that method development for polar VOCs in air is crucial if measurement and
regulation of these HAPs are to be accomplished. For other chemical classes, the scarcity of data in this
survey may have other causes. Many chemicals have been measured in the workplace but not in ambient air.
For example, the list designates titanium tetrachloride, elemental phosphorus, and dye intermediates such as
3,3'-dimethoxybenzidine as HAPs. Although the potential toxicity of these chemicals has been established,
their ambient concentrations have not been measured because they have been considered unlikely to be
present at significant concentrations in ambient air. For such compounds, initial ambient measurements
focused in areas of known sources may be preferable to widespread survey measurements, in assessing the
potential for human exposure to these chemicals.
Another reason for the lack of ambient air data for some HAPs is the ambiguous nature of the
identification on the CAAA list. A good example is "coke oven emissions" The emission of a variety of
toxic chemicals from coke ovens is well documented, including sulfur compounds, benzene, other aromatics,
and polycyclic aromatic compounds. However, it is impossible to quantify those compounds originating in
ambient air from coke oven emissions, in the face of other sources of the same compounds, without (e.g.)
detailed source apportionment modelling in the area of a coke oven source. As a result, measurements of
"coke oven emissions" as a chemical group in urban areas simply do not exist.
The representativeness of the HAPs data for use in health risk assessments is an important issue. Some
compounds, such as the chlorinated and aromatic hydrocarbons, have been measured thousands of times in
dozens of locations. The geographic spread of the data is also wide, merely because of the large number of
studies including these chemicals. Thus it may be argued that sufficient data exist to estimate typical and
elevated human exposures to these chemicals. However, as noted above, nearly two-thirds of the 189 HAPs
have been measured fewer than 100 times, and a similar number have been measured in fewer than S
locations. Such small data sets and limited geographic coverage are unlikely to adequately represent the
exposure of the U.S. population to those chemicals. More measurements of these compounds are needed if
health risk assessment is to be conducted adequately.
As an example of the data compiled in this survey, Table 1 summarizes the most recent data found for a
group of 30 high priority HAPs. This list of chemicals was adapted from the draft list of candidate pollutants
in the U.S. EPA solicitation for the National Human Exposure Assessment Survey (NHEXAS). The data
shown in Table 1 are a subset of the complete data sets compiled for these chemicals in the present study.
Shown in the table are the number of study locations, number of samples, mean, range, and years of recent
measurements for the 30 chemicals. The availability of data for these key compounds is generally better than
for the 189 HAPs as a whole. All of the chemicals in Table 1 have been measured recently in ambient air,
and for most of these chemicals several hundred recent samples are indicated. Inspection of the full data set
also indicates that the recent data in Table 1 exhibit means and ranges that are generally lower than those of
earlier data. This difference may indicate decreases in the emissions of these chemicals. However, changes
in the choice of sampling locations may also account for this difference. Site selection in early urban field
studies often emphasized worst-case locations such as urban traffic centers; recent studies have tended to
emphasize sites that are more representative of local population distributions. As a result, the recent data
shown in Table 1 may be useful for initial health risk assessments for these 30 HAPs.
163
-------
CONCLUSIONS AND RECOMMENDATIONS
The conclusions of this study are:
1. The 189 HAPs can be divided into two groups: one comprising roughly 30 percent of the list, for
which previous ambient measurements have been frequent and widespread, and a second much
larger group for which measurements are rare or non-existent.
2. For a core group of key HAPs, recent and earlier ambient data are relatively plentiful, suggesting
that initial health risk estimates can be made with the existing data.
3. The major groups of HAPs for which data are lacking are the nitrogenated and oxygenated
organics; lack of suitable sampling and analysis methods is the main reason for the lack of data for
these chemicals.
The recommendations from this study are:
1. Analytical method development is critically needed for many of the HAPs, particularly for the
nitrogenated and oxygenated organics, which collectively may be called polar VOCs.
2. Additional ambient air measurements are needed for at least 70 percent of the HAPs, to improve
the representativeness of the data for use in human health risk assessments.
3. Efforts should continue to enlarge the present set of ambient air data for the 189 HAPs. The
present survey was not designed to be exhaustive, and inclusion of additional data would be
valuable.
REFERENCES
1. T.J. Kelly, M. Ramamurthi, A.J. Pollack, and C.W. Spicer, Ambient Concentration Summaries for
Clean Air Act Title III Hazardous Air Pollutants, Draft Final Report to U.S. EPA, Contract No. 68-D8-
0082, Battelle, Columbus, Ohio, April, 1993.
2. J.J. Shah, and E.K. Heyerdahl, National Ambient Volatile Organic Compounds (VOCs) Data Base
Update, Report EPA-600/3-88/010(a), U.S. Environmental Protection Agency, Research Triangle Park,
N.C., 1988.
3. J.J. Shah, and H.B. Singh, Distribution of volatile organic chemicals in outdoor and indoor air: A
national VOCs data base, Environ. Sci. Technol.. 22, 1381-1388, 1988.
4. J.J. Shah and D.W. Joseph, National VOC Data Base Update, Final Report to Battelle, Columbus,
Ohio, under Subcontract Number 18723, U.S. EPA Contract No. 68-D8-0082, May, 1993.
5. C.A. Menzie, B.B. Potocki, and J. Santodonato, Exposure to carcinogenic PAH's in the environment,
Environ. Sci. Technol.. 26, 1278-1284, 1992.
6. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Polynuclear
Aromatic Hydrocarbons, Part 1, Chemical, Environmental, and Experimental Data. International
Agency for Research on Cancer, World Health Organization, 1983.
7. T.J. Kelly, P.J. Callahan, J.D. Pleil, and G.F. Evans, Method development and field measurements for
polar VOCs in ambient air, Environ. Sci. Technol.. in press, 1993.
164
-------
1-9 10-99 100-999 >1000
Number of Samples
Figure 1. Distribution of the 189 HAPs by Number of Ambient Air Samples
60+
I I Total in Class
•I No Ambient Cone. Data
Compound Class
Figure 2. Number of HAPs, and Number for Which No Ambient Data Were Found, for Each
of the Chemical Classes Used in the Data Search
165
-------
TABLE 1. RECENT AMBIENT AIR CONCENTRATIONS DATA FOR 30 HIGH PRIORITY HAPs
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Compouod
Benzene
Chloroform
Formaldehyde
Tetrachloroethylene
1,3-Butadiene
Carbon Tetrachloride
Chlorobenzene
Vinylidene Chloride
Ethylene Oxide
Methyl Chloroform
Methylene Chloride
Styrene
Toluene
Vinyl Chloride
Xylenes:
o-xylene
m-xylene
p-xylene'
Lead
Arsenic
Cadmium
Chromium
Mercury
Nickel
POMs
2,3,7,8-TCDD
PCBs
Chlordanet
2,4-Dt
Heptachlort
Hexachlorobenzene'''
Pentachlorophenol
PropoxurT
No. of Study
Areas
14
13
13
13
1
13
13
2
2
13
13
3
14
13
3
3
31
2
2
1
3
2
3
7
5
Great Lakes
2
2
2
2
1
2
NO. or
Samples
5348
4368
804
728
349
728
728
379
>3
728
728
6117
5348
728
4999
4999
785
465
696
349
808
178
664
159
134
Many
301
288
301
301
2
301
Overall
Mean
2.8 Mg/m3
0.4 Mg/m3
3.3 Mg/m3
3.6 Mg/m3
2.3 Mg/m3
1.2 Mg/m3
0.12 Mg/m3
ND
< 1.8 Mg/m3
5.6 Mg/m3
2.2 Mg/m3
0.55 Mg/m3
10.2 Mg/m3
0.96 Mg/m3
2.6 Mg/m3
5.4 Mg/m3
8.7 Mg/m3
9ng/m3
2.5 ng/m3
1.2 ng/m3
3.3 ng/m3
5.8 ng/m3
3.8 ng/m3
8.4 ng/m'
0.04 pg/m3
1.0 ng/m3
17.1 ng/m3
0.003 ng/m3
7.0 ng/m3
0.04 ng/m3
0.92 ng/m3
2.5 ng/m3
Overall Data Range
<0.05 - 67.3
<0.03 - 115
0.12-23.4
<0.69 - 104
<0.02 - 321
<0.06-27.8
<0.09-9.1
< 0. 12, < 0.40 Mg/m3
0.09 - 1.8
<0.28 - 492
< 0.35-1 12
<0.05-35.1
0.11 - 750
<0.08 - 202
< 0.05 -64.1
<0.06 - 127
ND-72
0.4 - 50
1.8-7.0
0.3 - 4.1
< 1 - 30
0.8 - 16
<2-8.7
0.3 - 91
Not available
Not available
<4 - 628
<0.5 - 1.2
<0.5 - 627
<0.5 - 13.0
0.91-0.92
<3 - 286
Study
Years
1989-91
1989-91
1989-91
1989-91
1990
1989-91
1989-91
1989-91
1989
1989-91
1989-91
1989-91
1989-91
1989-91
1989-91
1989-91
1985-87
1989-91
1985, 89
1985
1985,
1988-89
1985;
1984-91
1986-88
Up to 1991
1987-88
1987-88
1987-88
1987-88
1989
1987-88
Mean value* calculated using *0* for non-detected samples; all other cases use 0.5xDetection T imit for non-detected
•amplei.
166
-------
SURVEYS OF THE 189 CAAA HAZARDOUS AIR POLLUTANTS:
II. ATMOSPHERIC LIFETIMES AND TRANSFORMATION PRODUCTS
Thomas J. Kelly, Albert J. Pollack, Mukund Ramamurthi, and C. W. Spicer
Battelle, 505 King Avenue, Columbus, OH
Larry T. Cupitt
U.S. Environmental Protection Agency, Research Triangle Park, NC
ABSTRACT
This paper describes a survey of the atmospheric reactions, products, and lifetimes of the 189
Hazardous Air Pollutants (HAPs) listed in the 1990 Clean Air Act Amendments. This survey focussed on the
primary transformation processes for the HAPs, with the aim of identifying toxic reaction products. For 92
of the HAPs, reaction rate, lifetime, and product information were found, including 13 HAPs for which no
significant transformation is expected. For 85 other HAPs, rate and lifetime data were found, but no
products were identified. For the final 12 HAPs, no reactivity or product data at all were found. Reaction
with OH radical is the most common transformation process for the HAPs, with photolysis, deposition, and
reactions with ozone, nitrate radicals, and water being generally of secondary importance. The most common
HAPs reaction products include low molecular weight aldehydes, alcohols, ketones, organic acids, nitrates,
CO, and CO2. Some non-toxic chemicals in air may also give HAPs as reaction products.
INTRODUCTION
To accelerate the pace of identifying and regulating air contaminants, Congress established a list of 189
chemicals designated as Hazardous Air Pollutants (HAPs) in the 1990 Clean Air Act Amendments (CAAA).
Those 189 HAPs are a diverse group, including volatile organic compounds (VOCs), polar VOCs, pesticides,
semivolatile compounds, and metals. In addition to the concern of human exposure directly to these 189
chemicals, the CAAA also identifies the need for "consideration of atmospheric transformation and other
factors which can elevate public health risks from such pollutants." This survey was performed to identify
the primary removal and transformation processes, products, and lifetimes to be expected in the atmospheric
degradation of the 189 HAPs.
SURVEY PROCESS
For purposes of this survey, the 189 diverse chemicals designated as HAPs were organized into
chemical classes to facilitate searching for transformation data. This classification was useful because similar
chemicals are frequently evaluated together and are transformed via similar reaction mechanisms.
Information on the transformation processes of the 189 HAPs was located using two computerized data bases
and through a general review of articles, reference books, proceedings of relevant conferences, and reports.
One data base reviewed for reaction rate information was the ABIOTIKX software package (1). This program
supplied measured reaction rate constants for the degradation of organic compounds in the atmosphere. Also
provided were literature citations for the information displayed. If the title of the referenced work implied
that reaction products were also identified in the text, then efforts were made to obtain a copy of the
manuscript for in-depth review. The second data base employed consisted of a keyword search conducted
through the computerized databases of STN International. The databases searched included the Chemical
Abstract (CA) files from 1967 to present, Chemical Abstract Preview (CAP) current files, and the National
Technical Information Service (NTIS) files from 1964 to present. The search strategy targeted keywords such
as "atmospheric or air," "reactions or kinetics or removal," and "rates or constants or lifetime." The search
was restricted to English language citations in order to expedite their evaluation. Transformation data were
also obtained from published reviews, reference texts, proceedings of meetings, and reports that were not
identified during the two database searches. These sources reflected general reference materials available at
the time of the survey.
167
-------
ATMOSPHERIC LIFETIMES
If the information found for a particular HAP contained rate constants for atmospheric reactions, then
an estimated lifetime was computed. For the general case of a chemical (x) reacting with an atmospheric
constituent (A), the lifetime (T) of x was defined as TX = l/kCA, where k is the appropriate rate constant and
CA is the concentration of A. To perform this calculation, the following assumed concentrations for the
major gaseous reactants were used:
Concentration
Species (molecules/cm3)
O3 1.5 x 1012
OH 3.0 x 106
N03 2.5 x 10'
H02 l.OxlO9.
These concentrations were meant to represent long-term average concentrations in a relatively polluted
environment (2). Photolysis was also included as a transformation pathway. Three lifetime categories were
utilized: < 1 day, indicating rapid transformation in the atmosphere; 1-5 days, indicating a more persistent
nature; and finally >5 days, generally indicating that the HAP either reacts slowly or not at all under
atmospheric conditions. In some instances, lifetimes indicated in different literature sources for a HAP were
not in agreement. To accommodate this, ranges of < 1 to 1-5 days and 1-5 to >5 days lifetimes were
reported. For chemicals which react slowly in the atmosphere, transport and deposition (wet or dry) were
included as transformation processes since they determined the limiting lifetime of the HAP. The purpose of
this survey was to characterize the dominant features of each HAP's atmospheric transformation, and to
document the likely reaction products. The compilation of reaction rate constants, and the detailed evaluation
of minor reaction pathways, were not the focus of this study. Such efforts are valuable, but are feasible only
when smaller numbers of HAPs are addressed (e.g., 3).
RESULTS AND DISCUSSION
The results of this survey are compiled in a 34-page table, with an associated list of over 140 citations
to relevant literature (4). The data table lists all 189 chemicals in the same order as in the CAAA, and
includes the name and CAS number for each compound, the chemical formula/structure, the major removal
processes, the atmospheric lifetime, the potential transformation products, the references for the data
presented, and any additional comments. Of the 189 HAP compounds, 92 (49% of the list) have data entries
that include reaction rates and corresponding atmospheric lifetimes, as well as identified reaction products.
Included in these 92 compounds are 13 chemicals not expected to undergo significant transformation. For
eighty-five other compounds (45%) reaction process and lifetime data were found, but no identified reaction
products. The remaining 12 compounds were not reported in the literature reviewed; i.e., no information at
all was found on their atmospheric reactivity.
A breakdown of the data obtained during this survey, summarized by the chemical classes used in the
search, is shown in Table 1. This table shows that there were 97 compounds for which no product
information was found (85 no product data + 12 no data at all). The nitrogenated organics and oxygenated
organic compounds comprise nearly two-thirds of the group for which no products were identified. This may
reflect the lack of definitive methods for measuring these polar VOCs and consequently the absence of data
associated with the degradation of these compounds. Methods development for polar VOCs may be necessary
before transformation products can be identified.
The major removal process driving the transformation of the HAP compounds in the atmosphere is the
reaction with hydroxyl radicals, OH. Of the 177 HAPs for which reaction processes are reported, 146 are
primarily attributed to OH. Reactions with ozone, nitrate radicals, liquid and vapor phase water, and by
photolysis are generally secondary to OH oxidation, but account for the major removal process for 16 of the
168
-------
HAPs. For IS compounds, deposition was identified as the major removal process. Fourteen of these latter
HAPs are classified as inorganics and are not expected to undergo rapid chemical transformation in the
atmosphere. Physical processes thus dominate their removal from the atmosphere.
The atmospheric lifetimes of the HAP compounds are shown in Table 2. For 177 of the chemicals
lifetimes are reported. The table indicates that 81 of the compounds are expected to be transformed in < 1
day. Such rapid removal generally reflects rapid oxidation with OH. A total of 34 compounds fall into the
category of being transformed within a 5-day residence time in the atmosphere. In general, these compounds
also undergo OH reaction, only at a slower rate. Five HAP compounds are reported as persisting for a range
of 1 to >5 days. OH is again indicated as the primary removal process. Finally, 57 chemicals are expected
to persist for >5 days. These compounds either react very slowly or are not expected to be transformed at
all and therefore are removed by physical deposition. A review of the lifetime data indicates that
hydrocarbons, nitrogenated organics, aromatic compounds, phthalates, sulfates, and pesticides/herbicides are
generally expected to degrade rapidly in the atmosphere. The oxygenated organics range evenly across the
reported lifetime ranges. The inorganics, halogenated hydrocarbons, and halogenated aromatics are
anticipated to be relatively persistent in the atmosphere.
The reaction products reported for the HAPs reflect a wide range of chemical compositions. In
general, the HAP compounds undergo atmospheric reactions to generate low molecular weight aldehydes,
alcohols, organic acids, ketones, nitrates, carbon monoxide, carbon dioxide, and water. Atmospheric
reactions of many of the HAPs generate other HAP species. Production of formaldehyde in the
photochemical oxidation of many volatile compounds is a good example of this. Other HAP transformations
also produce a variety of stable organic and inorganic compounds that are considered toxic and therefore
contribute to public health risks, but which are not designated as HAPs. For 97 of the HAPs, no product
data were identified. Continued efforts should be made to identify atmospheric reaction mechanisms and
products for these chemicals.
It is important to note that some non-HAP compounds undergo atmospheric transformations to generate
hazardous chemicals. An example is propylene. This common ambient air constituent when irradiated in the
presence of NOX degrades to formaldehyde, acetaldehyde, peroxyacetyl nitrate, nitric acid, propylene glycol
dinitrate, 2-hydroxy propyl nitrate, 2-nitropropyl alcohol, a-nitroacetone, and carbon monoxide (5). Other
as-yet unidentified mutagens are also formed, and likely include organic peroxides and nitrates. Further
investigations of the transformation of propylene with O3 (6) and by hydroxyl and nitrate radical reactions (7)
also identify organic oxygenates being generated although they do not account for the key mutagens
associated with NOX transformation. This work shows that non-HAP compounds, present in ambient air and
considered non-toxic, can go through atmospheric transformations to generate toxic chemicals. This potential
source of toxic chemicals in the atmosphere beyond the HAPs list deserves further consideration.
CONCLUSIONS AND RECOMMENDATIONS
The conclusions of this study are:
(1) Reaction with OH radical is the most common atmospheric reaction pathway for the HAPs.
(2) The most common products generated during atmospheric reaction of the HAPs include low
molecular weight aldehydes, alcohols, organic acids, ketones, nitrates, carbon monoxide, carbon
dioxide, and water. Several of these and other transformation products are considered toxic and
may pose a human health risk.
(3) Atmospheric reaction products are unknown for 97 of the 189 HAPs compounds, and further
research needs to be carried out on these chemicals.
169
-------
The recommendations from this study are:
(1) Analytical methods development followed by atmospheric transformation studies for those HAPs
for which data appear to be lacking.
(2) Identification of reaction products of non-toxic ambient air constituents to define the human
health risk associated with the atmospheric degradation of these non-HAP compounds.
REFERENCES
1. B. Daniel, G. Merz, and W. Klopfer, ABIOTIKX Data-Base, Daniels Electronic, on behalf of Battelle-
Institute.V., Frankfurt, Germany, November 1990.
2. P. Howard, R. Boethling, W. Jarvis, W. Meylan, and E. Michalenko, Handbook of Environmental
Degradation Rates, Lewis Publishers, Chelsea, Michigan, 1991.
3. J. Wrobel, S. Ohshita, and C. Seigneur, Transformations of Trace Metals and Chlorides in the
Atmosphere, paper 93-STP-50.03, presented at the 1993 AWMA National Meeting, Denver, Colorado,
June, 1993.
4. A.J. Pollack, T.J. Kelly, M. Ramamurthi, and C.W. Spicer, Atmospheric Transformation Summaries
for Clean Air Act Title III Hazardous Air Pollutants, draft final report to U.S. EPA, Contract No. 68-
D8-0082, Battelle, Columbus, Ohio, May 1993.
5. T.E. Kleindienst, P.B. Shepson, E.G. Edney, L.T. Cupitt, and L.D. Claxton, The Mutagenic Activity
of the Products of Propylene Photooxidation, Environ. Sci. Teehnol.. 19, 620, 1985.
6. P.B. Shepson, T.E. Kleindienst, E.O. Edney, L.T. Cupitt, and L.D. Clayton. The Mutagenic Activity
of the Products of Ozone Reaction with Propylene in the Present and Absence of Nitrogen Dioxide,
Environ. Sci. Teehnol.. 19, 1094, 1985.
7. P.B. Shepson, E.O. Edney, T.E. Kleindienst, J.H. Pittman, G.R. Namie, and L.T. Cupitt, The
Production of Organic Nitrates from Hydroxyl and Nitrate Radical Reaction with Propylene, Environ.
Sci. Teehnol.. 19, 849, 1985.
170
-------
TABLE 1. DATA COMPLETENESS BY COMPOUND CLASS
Compound Class
Hydrocarbons
Halogenated Hydrocarbons
Aromatic Compounds
Halogenated Aromatics
Nitrogenated Organics
Oxygenated Organics
Pesticides/Herbicides
Inorganics
Phthalates
Sulfates
TOTAL
Complete Data
3
18
14
4
10
18
4
19
0
2
92*
No
Product Data
0
8
3
4
33
21
10
2
4
0
85
No
Transformation Anticipated
0
1
0
0
0
0
0
12
0
0
13
No
Data
0
1
1
0
6
1
1
2
0
0
12
(*) Includes the 13 compounds that are not anticipated to undergo any atmospheric transformations.
-------
TABLE 2. ATMOSPHERIC LIFETIMES OF HAP COMPOUNDS
(in days)
Compound Class
Hydrocarbons
Halogenated Hydrocarbons
Aromatic Compounds
Halogenated Aromatics
Nitrogenated Compounds
Oxygenated Compounds
Pesticides/Herbicides
Inorganics
Phthalates
Sulfates
<1
2
5
12
0
32
10
13
3
2
2
< 1 to 1-5
0
2
2
0
1
3
0
0
0
0
1-5
1
1
2
1
2
15
0
3
1
0
1-5 to >5
0
0
0
1
3
0
0
1
0
0
>5
0
18
1
6
5
11
1
14
1
0
No Estimate
0
1
1
0
6
1
1
2
0
0
Totals
3
27
18
8
49
40
15
23
4
2
TOTAL 81 8 26 S 57 12 189
-------
Session 5
Integrated Air Cancer Project
-------
The Integrated Air Cancer Project: Overview of Roanoke Study and Comparison to
Boise Study
J. Lewtas1, D. B. Walsh2, L. T. Cupitt1
'U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Environmental Health Research and Testing, Inc.
Research Triangle Park, NC 27709
INTRODUCTION
The Integrated Air Cancer Project (IACP) is a long-term EPA research project
designed to improve understanding of the human exposure to and origins of carcinogenic
pollutants. The IACP has focused on products of incomplete combustion (PICs). The
goals of the research program are (1) to identify the principal carcinogens in the air to
which humans are exposed, (2) to determine which emission sources are the major
contributors to the atmospheric burden of carcinogens, and (3) to improve the scientific
capability for estimating both human exposure and the resultant comparative human
cancer risk arising from exposure to air pollution, particularly those from the PICs.
Incomplete combustion products include polycyclic organic matter (POM) primarily
adsorbed to respirable particles. PICs were identified as a major source of carcinogenic
risk in urban areas and constitute a large fraction of the atmospheric burden of pollut-
ants on a national basis1'2. Therefore, the research strategy focused on PICs, especially
those from residential home heating and motor vehicles that are major, ubiquitous
emission sources in populated areas3.
Two residential heating sources have been studied, residential wood combustion
(RWC) and residential distillate oil combustion (RDOC). RWC source was selected
because: (1) it represented a major fraction of PIC emissions on a national basis; (2) it
was under review for regulatory action; and (3) the high mass loadings associated with
wood smoke would ensure that sufficient mass could be collected during the field study
to, in turn, conduct the chemical and biological analyses needed to progress toward the
IACP goals. The first major field study was conducted in Boise, ID, in 1986-1987 where
the airshed contained two major sources, mobile source emissions and RWC4. The
second study, in Roanoke, VA, in 1988-1989 (October to February) was in an airshed
containing three sources, mobile sources, RDOC, and RWC5.
SELECTION AND DESCRIPTION OF STUDY SITES: ROANOKE AND BOISE
Boise and Roanoke were each selected from a relatively large list of towns and
cities initially developed through EPA's National Emissions Data System (NEDS) data
base combined with information on concentrations of particulate matter found in these
cities. Cities with complex industrial sources, unrepresentative meteorological condi-
tions, and low pollution levels were eliminated. In the selection of a city where RDOC
175
-------
was a major contributor to the home heating, the cities were ranked by the quantity of oil
burned per person and per square mile of urban area. A similar ranking was done in the
case of RWC. The top 3-5 locations were site visited by an EPA team and the final city
chosen based on the site selection criteria established for each study. These criteria
included: (1) The residential heating sources being studied (wood combustion in Boise
and oil combustion in Roanoke) were estimated or known to be significant contributors
to high particle loadings which normally occurred during the fall and winter; (2) the
airshed was determined to be relatively simple, with no major confounding emission
sources; (3) there were numerous sampling sites available in the area which met the
criteria for the objectives of this study; (4) the terrain and meteorology was appropriate
for extrapolation to other locations; and (5) the support of local and State government as
well as local civic and business leaders.
These field studies in Boise and Roanoke were conducted in metropolitan urban
areas with populations between 100,000 and 150,000. Both cities are governmental,
educational and commercial centers for their region. Geographically they are both
located in valleys adjacent to or between large mountain ranges. This local topography
and the meteorological conditions for both sites resulted in periods of winter inversions.
There are no large or heavy industrial sources in either city. A detailed description
comparing these two sites and average pollutant concentrations is reported elsewhere5'6.
Important differences between these two locations and studies include: (1) the Boise
airshed contained only mobile source emissions and RWC while the Roanoke airshed
contained a third source, RDOC, and (2) the background pollution levels coming into the
Boise airshed from the surrounding region were much lower than for Roanoke where the
background sulfate concentrations were twice the levels of Boise presumably due to the
influence of industrial emissions on the Eastern US corridor.
SURVEY AND INVENTORY OF POLLUTION SOURCES AND PERSONAL ACTIVI-
TY PROFILES
Surveys were used to learn more about the major sources of PICs in each of these
airsheds. Two surveys were used, the first was a general survey dealing with home
heating and motor vehicle usage. A second survey was administered to respondents who
burned wood (in Boise and Roanoke) and oil (Roanoke only) to determine the type of
heating appliance, usage, etc. In addition to these surveys, a micro-inventory of each
city's potential pollution sources was conducted7'8.
A survey of home heating sources and motor vehicle usage in Roanoke9 and
Boise10 was administered using a probabilistically-derived sample of the housing units
based on block-census data for the cities of Boise and Roanoke. Figure 1 compares the
residential heating sources for these two cities during the IACP field studies. Roanoke
has a substantially lower percentage of .the residences using wood as a heating source
(10%) compared to Boise (62%). Roanoke has an additional source of residential
distillate oil heating (22% of the residences).
176
-------
One of the goals of the IACP is to improve EPA's ability to assess exposure and
risk from airborne carcinogens. To address this issue, the IACP included efforts to
characterize both outdoor and indoor exposure levels and to improve our understanding
of the relationships between indoor and outdoor concentrations of carcinogens. The
residential sampling portion of the field studies are critical to the exposure estimate,
because people normally spend about two-thirds of their time in their houses. The
indoor sampling provides insight and data that can be used to estimate indoor exposures
by the population in Boise. In addition, residents completed daily "diaries" of their activi-
ties during sampling studies at their homes. Diaries were compiled in Boise and in
Roanoke to estimate the fraction of time spent in different locations or activities.
The daily activity diaries from the Boise and Roanoke studies were used to cha-
racterize the average time periods spent in five microenvironments: indoors at houses
(67-69%); outdoors (1-2%); in-transit (3-4%); at the work place (16-18%); and at other
indoor locations like stores, churches, and post offices (8-12%). The percentage of time
spent in each of the five zones for these winter periods was determined from the diaries
and compared to a national survey11 for year-round activity patterns. The time alloca-
tions for both Boise and Roanoke were reasonable compared with the national survey
data. One would expect the time indoors during the winter to be greater than the
national annual average, and for the time outdoors to be less. Boise and Roanoke are
modest sized cities, so the commute time, T, would also be expected to be less than the
national average. These expected differences from the national average were observed,
however the differences were minor (max 4% variation).
The relatively uniform distribution of pollutants across these two airsheds and
across the population distribution facilitates the exposure extrapolation for many of the
PIC pollutants from the relatively small population for which we obtained time-activity
profiles to the general populations in these airsheds. In the Boise study, human exposure
and dose estimates have been completed for the extractable organic mass (EOM) from
fine particles and the exposure apportioned between the two major sources in the
airshed11'12 using methods developed as part of the LACP11'13-
FIELD STUDY DESIGN
The field studies were conducted in the winter months, over a four month period,
during the heating season in Boise, ID in 1986-87, and in Roanoke, VA in 1988-89. The
field program consisted of both ambient and residential sampling. The data generated in
the sampling programs have been detailed in several papers on Boise14 and Roanoke5'6.
The ambient sampling was conducted at three primary sites and four auxiliary sites
in each city as shown in Figures 2 and 3. One primary site in each city was in a
residential area. A second primary site was near well-traveled roadways. A third
primary site was the background sampling location. Four auxiliary sites were also
operated during these studies. Sampling periods were 12 hours long, with changeover
times at 7 AM. and 7 P.M. There were 13 sampling periods scheduled per week, and
one period was dedicated to calibration, maintenance, etc.
177
-------
The residential sampling involved a matched pan- of nearby houses each week.
During the study, ten pairs of houses were sampled. One of the houses in each pair used
a combustion heating device (RWC in Boise and RDOC in Roanoke). The other house
was heated by electricity or natural gas and did not use either RWC or RDOC.
Sampling was conducted in 12 hour periods identical to those at the ambient sampling
sites. Sampling began each Saturday morning and terminated after the nighttime
sampling period, which ended at 7 A.M. Wednesday. For analysis purposes, the eight
sampling periods were combined into four samples: weekend daytime, weekend night-
time, weekday daytime, and weekday nighttime. Whenever samples were collected at the
houses, corresponding samples were also taken at the primary sites. Samples of the RWC
emissions from the wood burning appliances in Boise1 and residential oil heaters in
Roanoke16 were obtained with a dilution sampling system17'18. Each pair of houses was
matched for age, size, etc. None of the residents in the sampled houses were smokers.
The maps of Boise (Figure 2) and Roanoke (Figure 3) show the primary, auxiliary,
and residential sampling locations. Each R number on the map represents a matched
pair of houses. The auxiliary sites were located across the valley in order to examine the
distribution of pollutants across the airshed. Resource limitations prevented the design
of a residential monitoring study sufficiently large to represent the populations statistical-
ly. Although the 10 pairs of houses were not statistically representative of the Boise
population, the data may be used to understand the processes that affect exposures
across the community. In addition, the auxiliary sites provided supplementary data to
support the extension of population exposure assessment across the total population.
CENTRALIZED DATABASE
Data from the sampling, chemical analysis, physical analysis and biological studies
have been integrated into a centralized database. All of the data have been validated by
the EPA scientist or engineer responsible for those measurements. The database is
implemented in a fourth generation, non-procedural, report generation system
(FOCUS*) on the National Computing Center IBM 3090, at Research Triangle Park,
NC. The database for Boise contains approximately 185 unique analysis species and
more than 78,400 data values and the database for Roanoke will be somewhat larger.
The database is described in detail elsewhere4 and is now available from EPA4 in a
format suitable for personal computer use (dBase*).
COMPARATIVE CARCINOGENICITY AND SOURCES OF CANCER RISK
The component of PICs estimated to make the largest contribution to human
cancer risk is the polycyclic organic matter (POM) associated with airborne particles.
The extractable organic matter (EOM) adsorbed to airborne particles contains most of
the carcinogenic POM. Under some ambient conditions, the semivolatile organic
compounds (SVOCs) may also contain polycyclic aromatic compounds. The
carcinogenicity of SVOCs has not yet been studied. The Boise IACP project, for the first
time, both source apportions19 and characterizes the carcinogenicity of ambient POM
from particles using in vivo animal tumor assay19, receptor modeling20'21 and human
178
-------
exposure data12. In Boise, the ambient POM sample containing 33% contribution of
motor vehicle emissions was more than twice as tumorigenic as the ambient sample with
only 11% motor vehicle emissions. Receptor modeling of the EOM in Roanoke is
reported in this volume22 and the animal tumor studies are in progress. Atmospheric
transformations in these airsheds23 are contributing to the presence of air toxics and may
account for a component of the increased tumorigenicity associated with the POM from
mobile sources4. Nitrogen oxides appear critical in the formation of mutagenic
transformation products.
ACKNOWLEDGEMENTS AND DISCLAIMER
The author acknowledges the collaboration of all the other scientists and
engineers participating in the US EPA's Integrated Air Cancer Project and especially R.
Zweidinger, R. Stevens, C. Lewis, D. B. Ray, V.R. Highsmith, A.J. Hoffman, and R.
McCrillis. This paper has been reviewed in accordance with the US EPA's peer review
policies for scientific papers and approved for publication. The contents are not intended
to represent Agency policy.
REFERENCES
1. US EPA, 1985. The Air Toxics Problem in the United States: An Analysis of
Cancer Risks for Selected Pollutants, EPA-450/1-85-001, May 1985.
2. US EPA, 1990. Cancer Risk from Outdoor Exposure to Air Toxics, EPA-450/1-
90-004a, September 1990.
3. Lewtas, J., 1989. Emerging Methodologies for Assessment of Complex Mixtures:
Application of Bioassays in the Integrated Air Cancer Project. Tox. Indust
Health, 5(5):839-850.
4. US EPA, 1993, Integrated Air Cancer Project: Research to Improve Risk
Assessment of Area Sources: Wood Stoves and Mobile Sources: Boise, Idaho,
Office of Research and Development, US EPA, Research Triangle Prk, NC, 27711
5. Highsmith, V.R., Hoffman, A.J., Zweidinger, R.B., Cupitt, L.T. and D.B. Walsh,
1991. The IACP: Overview of the Boise, Idaho and the Roanoke, Virginia, Field
Studies, in Proceedings of the 84th Annual Meeting of the AWMA, paper 93-131.3.
6. Stevens R.K., Hoffman, A.J., Baugh, J.D., Lewis, C.W., Zweidinger, R.B. Cupitt,
L.T., Kellogg, R.B. and J. H. Simonson, 1993. A Comparision of Air Quality
Measurements in Roanoke, VA and other Integrated Air Cancer Project
Monitoring Locations, Proceedings of the 1993 EPA/AWMA International
Symposium on Measurement of Toxic and Related Air Pollutants, this volume.
179
-------
7. Downs, J.L., 1986. Final Report on Integrated Air Cancer Project Site Source
Inventory: Boise, Idaho, EMSI 1135.27FR. Environmental Monitoring and
Services, Inc., Camarillo, CA.
8. Ellis, C, Lowe, S., Ramadan, W, Reece, M. and M. Smith, 1993. Final Report on
Integrated Air Cancer Project Site Source Inventory: Roanoke, Virginia, EPA
Contract No. 68-D9-0173, Atmospheric Research and Exposure Assessment
Laboratory, Research Triangle Park, NC.
9. Johnson, T, 1989, A Survey of Factors Influencing Residential air Quality in
Roanoke, Virginia, Proceedings of the 82th Annual Meeting of the AWMA, paper
89-159.2
10. Cupitt, L.T. and T.R. Fitz Simons, 1988, The Integrated Air Cancer Project:
Overview and Boise Survey Results in Proceedings of the 1988 EPA/AWMA
International Symposium on Measurement of Toxic and Related Air Pollutants,
VIP-10, Air & Waste Management Association, Pittsburgh: pp. 799-803 (EPA
600/9-88-015, NTIS PB90-225863).
11. Glen, W.G., V.R. Highsmith, and L.T. Cupitt, 1991. Development of an Exposure
Model for Application to Wintertime Boise, in Proceedings of the 84th Annual Air
and Waste Management Association Meeting, Manuscript 91-131.7.
12. Cupitt, L., G. Glen, and J. Lewtas, 1993. Exposure and Risk from Ambient
Particle Bound Pollution in an Airshed Dominated by Residential Wood
Combustion and Mobile Sources. Environmental Health Perspectives, in press.
13. Lewis, C.W., R.E. Baumgardner, L.D. Claxton, J. Lewtas, and R.K. Stevens,
1988a. The Contribution of Wood Smoke and Motor Vehicle Emissions to
Ambient Aerosol Mutagenicity. Environ Sci Techno), 22:968-971.
14. Highsmith, V.R., R.B. Zweidinger, J. Lewtas, A. Wisbith, and R.J. Hardy, 1988.
Impact of Residential Wood Combustion and Automotive Emissions on the Boise,
Idaho, Airshed, in Proceedings of the 1988 EPA/AWMA International Symposium
on Measurement of Toxic and Related Air Pollutants, VIP-10, Air & Waste
Management Association, Pittsburgh: pp. 804-813 (EPA 600/9-88-015, NTIS PB90-
225863).
15. Steiber, R.S., R.C. McCrillis, J.A. Dorsey, and R.G. Merrill, 1992. Characterization
of Condensible and Semivolatile Organic Materials from Boise Woodstove
Samples, in Proceedings of the 85th Annual Meeting of the Air & Waste
Management Association, Manuscript 92-118.03.
16. McCrillis, R.C., R.R. Watts, and R.B. Zweidinger, 1993. Comparison of
Residential Oil Furnaces and Woodstove Emissions, Proceedings of the 1993
EPA/AWMA International Symposium on Measurement of Toxic and Related Air
Pollutants, this volume.
180
-------
17. Williamson, A.D., R.S. Martini, and D.B. Harris, 1985. Measurement of
Condensable Vapor Contribution to PM 10 Emissions, in Proceedings of the
78th Annual Meeting of the Air & Waste Management Association, Manuscript
85-14.4.
18. Merrill, R.G. and D.B. Harris, 1987. Field and Laboratory Evaluation of a
Woodstove Dilution Sampling System, in Proceedings of the 80th Annual Meeting
of the Air & Waste Management Association, Manuscript 87-64.7.
19. Lewtas, J., R.B. Zweidinger, and L. Cupitt, 1991. Mutagenicity, Tumorigenicity
and Estimation of Cancer Risk from Ambient Aerosol and Source Emissions from
Wood Smoke and Motor Vehicles, in Proceedings of the 84th Annual Meeting of
the Air & Waste Management Association, Manuscript 91-131.6.
20. Lewis, C.W., T.G. Dzubay, R.B. Zweidinger, and V.R. Highsmith, 1988b. Sources
of Fine Particle Organic Matter in Boise, in Proceedings of the 1988 EPA/AWMA
International Symposium on Measurement of Toxic and Related Air Pollutants.
VIP-10, Air and Waste Management Association, Pittsburgh: pp. 864-869 (EPA60-
0/9-88-015, NTIS PB90-225863).
21. Lewis, C.W., R.K. Stevens, R.B. Zweidinger, L.D. Claxton, D. Barraclough, and
G.A. Klouda, 1991. Source Apportionment of Mutagenic Activity of Fine Particle
Organics in Boise, Idaho, in Proceedings of the 84th Annual Meeting of the Air &
Waste Management Association, Manuscript 91-131.3.
22. Lewis, C.W., R.B. Zweidinger, L.D. Claxton, D.B. Klinedinst, and S.H. Warren,
1993. Source Apportionment of Fine Particle Organics and Mutagenicity in
Wintertime Roanoke, In Proceedings of the 1993 EPA/AWMA International
Symposium on Measurement of Toxic and Related Air Pollutants, this volume.
23. Nishoka, M. and J. Lewtas, 1992. Quantification of Nitro- and Hydroxylated Nitro-
Aromatic/Polycyclic Aromatic Hydrocarbons in Selected Ambient Air Daytime
Winter Samples. Atmos. Environ. 26A:2077-2087.
181
-------
Residential Heating Sources
Boise
Roanoke
Wood--62%
Electric or Gas-66%
Wood-10%
Other-2%
Oil-22%
Electric or Gas--38%
Figure 1. Survey of residential heating sources.
-------
IACP
Sampling Sites
1986-1987
Boise, Idaho
(fC-, Primary Stos
EGP
FS
RCAQ
(A) AuxfflarySte*
FAIR
CBP
WINS
ADAM
A Raskfential Sites
Figure 2. Map of Boise, ID, Showing Sampling Sites
183
-------
IACP ®
Sampling Sites
1988-1989
Roanoke, Virginia
Figure 3. Map of Roanoke, Virginia showing sampling sites.
-------
A Comparison of Air Quality Measurements in Roanoke, VA, and Other Integrated Air
Cancer Project Monitoring Locations
R.K. Stevens, AJ. Hoffman, J.D. Baugh, C.W. Lewis, R.B. Zweidinger, and L.T. Cupitt
U.S. Environmental Protection Agency
Research Triangle Park, NC
R.B. Kellogg and J.H. Simonson
ManTech Environmental Technology, Inc.
Research Triangle Park, NC
ABSTRACT
As part of the U.S. Environmental Protection Agency's Integrated Air Cancer Project (IACP) field air quality
monitoring studies in four U.S. cities were conducted between 1984 and 1989. The chronology of the field studies was:
December 1984 to March 1985 in Albuquerque, NM; (2) January 1985 to March 1985 in Raleigh, NC; (3) December
1986 to March 1987 in Boise ID; (4) October 1988 to February 1989 in Roanoke, VA. Aerosol and gas phase
pollutants collected at these locations were used in receptor models to apportion the contributions of emissions from
wood burning and mobile sources to fine particle organic matter concentrations. This paper compares the composition of
air pollutants measured in the four IACP studies, with regard to temporal variation and geographic location.
INTRODUCTION
Between 1984 and 1990, the U.S. Environmental Protection Agency (U.S. EPA) conducted a series of field
investigations to collect air quality and source emission samples for subsequent chemical characterizations. These studies
were part of the U.S. EPA's Integrated Air Cancer Project (IACP), whose objectives were to identify the principal
airborne carcinogens, determine which emission sources were the major contributors of carcinogens to ambient air, and
to improve the estimate of human exposure and comparative human cancer risk from specific air pollution emission
sources.' Unlike past efforts to identify airborne carcinogens through the use of emission inventories, the IACP took the
approach of measuring the mutagenicity (a surrogate for carcinogenicity) directly and apportioning this property to
source types.2*3
Field air quality monitoring studies in four U.S. cities were conducted between 1984 and 1989. The following is
the chronology of these field studies: (1) December 1984 to March 1985 in Albuquerque, NM, (2) January 1985 to
March 1985 in Raleigh, NC, (3) December 1986 to March 1987 in Boise, ID, and (4) October 1988 to February 1989 in
Roanoke, VA.3'4 In the studies performed in Raleigh and Albuquerque, measurements were made of the concentrations
of selected elemental tracers associated with emissions from wood stoves and motor vehicles. Potassium (K) corrected
for soil content" was measured as a species emitted during the combustion of wood, and lead (Pb) and bromine (Br)
were determined because they are elements emitted from motor vehicles using leaded gasoline. In the studies conducted
in Boise and Roanoke, volatile organic compounds (VOCs) characteristic of mobile source emissions3 were found to be a
replacement for Pb and Br as receptor modeling tracers.
This current work compares results of air quality measurements made in four cities where samples were collected
and analyzed to provide the data to support the IACP study objectives.
EXPERIMENTAL
Overviews of the sample collection and analysis for IACP studies have been given in previous publications.1'4
The basic collection and analysis methods used to develop the data base to support the receptor modeling for the four
IACP studies are given in Table 1.
185
-------
In these IACP field studies, air quality data and samples were collected at three primary sites. One site was at a
background location (except in Albuquerque), and two sites were in urban locations. In each city, one primary site was
a residential location impacted by wood smoke emissions, and the second primary site was located near a major roadway
impacted by motor vehicle emissions. Aerosol and VOC samples were collected at these primary sampling locations at
12-h intervals beginning at 7 a.m. and 7 p.m. in order to represent daytime and nighttime conditions. A complete
description of the monitoring performed at these sites has been described elsewhere.1"4
In the Roanoke study, nonradioactive organometallic rare earths ("°Sm and '* Sm) were added to the residential
home heating oil supplies and to diesel fuel used by the local buses as possible tracers for these sources.' Samples
collected at the residential and roadway primary sites were analyzed for '*Sm and l30Sm content.
RESULTS AND DISCUSSION
Table 2 contains the average composition of the fine particles collected in the four IACP cities. There are several
important features of the fine particles, characteristic of these cities, that are related to the time and location of the
studies. For example, in Albuquerque, the soil-corrected K concentration is substantially lower than that in the other
three cities. This observation is primarily related to the lower K content of the wood fuel (Pinyon pine) typically used in
this city.
The Pb concentration in the fine particles decreased from a high of 237 ng/m3 in Albuquerque in 1985 to 26.9
ng/m3 in Roanoke in 1989. This coincides with the reduction of Pb in gasoline over the past 20 years mandated by
Federal law.
The sulfur content of the fine particles in Raleigh and Roanoke was substantially higher than that in the western
cities. This difference is likely to be in part related to the higher density of power plant activities that use coal and
diesel fuel (which contains sulfur) in the eastern United States as compared to the lower density in Boise and
Albuquerque.
Table 3 contains a comparison of the key receptor modeling species measured at the four study locations. The
variations in the Pb and K values mentioned above are clearly shown in Table 3. In addition, the carbon (volatile carbon
[CJ, elemental carbon [CJ, and extractable organic matter [EOM]) content of the fine particles in Roanoke were
substantially lower than the values measured in the other cities. Correspondingly, the mutagenicity of the EOM samples
from Roanoke was also one half the value or more of the mutagenicity measured at the other IACP locations (Table 2).
In Roanoke, the average fine particle concentration and average coarse particle concentration measured at the
main monitoring site, which was at the convention center near Interstate 581, were 26.3 and 14.4 /tg/m3, respectively.
This was 35 % higher than the average value for the main monitoring site located at Momingside Park (residential
location) and twice the concentration measured at the background location at Carvin Grove Reservoir. The elevated
concentration levels at the convention center location are for the most part related to the high density of motor vehicles
passing the site on Interstate 581 and the close proximity (5m) of the site from the highway.
In Roanoke and Boise, values for the nighttime fine particle mass and extractable organic mass at the urban
primary sites were typically 40% larger than daytime concentrations. For example, in Boise at the residential primary
site, the nighttime fine particle mass averaged 47 fig/m3, whereas the daytime values were 28 /ig/m3. These differences
were associated with the increased use of wood stoves for heating at night and, in some cases, nighttime meteorological
inversions.
A number of the VOCs measured in Boise and Roanoke were highly correlated with the Pb concentrations in fine
particles.1 This relationship has led to utilization of VOCs as a replacement for Pb as a receptor modeling tracer for fine
particle emissions from motor vehicles.7 Table 4 contains the average a.m. and p.m. concentration of selected VOCs
and total nonmethane hydrocarbons (NMHC) measured in Boise and Roanoke. The VOCs with elevated concentrations
shown in Table 4 were species that correlated with elevated fine-particle Pb concentrations.5 This relationship provided
the impetus to test VOCs as candidates to replace Pb as a tracer for fine particle emissions from motor vehicles.7 This
research also has led to the development of chemical mass balance models for VOCs that have the potential to be used to
validate VOC emission inventories.'
186
-------
SUMMARY
The following is a summary of the results obtained from the 1ACP monitoring studies in the four cities where air
quality samples were collected. The observations discussed in this summary are derived from limited, measurements
collected during wintertime periods at selected locations in these four cities, therefore, extrapolation of the data in this
report to assess overall air quality in these cities beyond the period of the studies is not recommended.
1. Fine particle sulfate concentrations in Raleigh, NC, and Roanoke, VA, were twice the levels of those in
Boise, ID, and Albuquerque, MM.
2. Lead concentrations dropped from 237 ng/m3 in Albuquerque in 1985 to 27 ng/m3 in Roanoke in 1989.
3. The fine particle K (soil corrected) in Albuquerque was one third the level in Raleigh, Boise, and Roanoke.
4. Total NMHC in the four cities was approximately the same, 500 ppbC.
5. Extractable organic matter in Roanoke was one half the levels of the other three cities.
6. Nighttime EOM concentrations in all cities were higher than daytime values.
7. In Boise and Roanoke, the mass of fine particulate fraction was 2 to 3 times higher than the coarse fraction,
as measured by the dichotomous sampler.
8. In Boise and Roanoke, VOCs at the roadway and residential sites were dominated by aliphatic hydrocarbons,
which are composed of species related to mobile source emissions.
ACKNOWLEDGEMENT
The authors thank Jan Parsons for assisting in the preparation of this manuscript.
DISCLAIMER
The information in this document has been funded wholly or in part by the U.S. Environmental Protection
Agency to ManTech Environmental under Contracts 68-02-4443 and 68-DO-0106. It has been subjected to Agency
review and approved for publication. Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
REFERENCES
1. J. Lewtas, C.W. Lewis, R. Zweidinger et al., "Overview of the Integrated Air Cancer Project," these proceedings,
1993.
2. C.W. Lewis, R.E. Baumgardner, R.K. Stevens et al., "The contributions of woodsmoke and motor vehicle emissions
to ambient aerosol mutagenicity," Environ. Sci. Teehnol. 22: 968-971 (1988).
3. R.K. Stevens, C.W. Lewis, T.G. Dzubay et al., "Sources of mutagenic activity in urban fine particles," Toxicol.
Ind. Health 6(5): 81-94 (1990).
4. V.R. Highsmith, A.J. Hoffman, R.B. Zweidinger et al., "The IACP: Overview of the Boise, Idaho, and the
Roanoke, Virginia, field studies," in Proceedings of the 84th Annual Meeting of the A&WMA. 1991, paper 93-131.3.
5. C.W. Lewis, R.B. Zweidinger, L.D. Claxton et al., "Source apportionment of fine particle organics and
mutagenicity in wintertime Roanoke,' these proceedings, 1993.
6. J.M. Ondov, Z.C. Lin, W.R. Kelly et al., "Enriched stable isotopes of Sm as intentional tracers of diesel fuel and
residential oil furnace emissions in Roanoke, VA," these proceedings, 1993.
7. R.B. Zweidinger, R.K. Stevens and C.W. Lewis, "Identification of volatile hydrocarbons as mobile source tracers
for fine particulate organics," Environ. Sci. Teehnol. 24: 538-542 (1990).
187
-------
8. C.W. Lewis, T.L. Conner, R.K. Stevens et al., "Receptor modeling of volatile hydrocarbons measured in the 1990
Atlanta ozone precursor study," in Proceedings of the 86th Annual Meeting of A&WMA. Denver, 1993.
Table 1. Collection and analysis methods for four IACP studies.
Sampler
Type
PM10 Dichotomous
PM2.5 Hi-Vol
PM10 Medium Vol
VOC/Aldehyde
Annular Denuder
Continuous Parameters
CO, NOx, 03
Meteorological Parameters
ws, wd, t, RH, SR
Sampling
Medium
Teflon
TIGF
Quartz
Quartz
XAD
DNPH tubes
Canisters
Filters, tubes
Site
Category
primary
auxiliary
residential
primary, aux.
primary
primary
residential
primary
residential
primary
primary
residential
primary
auxiliary
Analysis
mass
XRF
mutagenicity
"C and rare earths
c.,cy
SVOCs
aldehydes
VOCs
acidic gases
Table 2. Air quality aerosol data measurements at primary residential sites: Average
composition of fine particles.
Mass
Al
As
Br
Ca
Cl
Cr
Cu
Fe
K
Mn
Mo
Pb
S
Se
Si
V
Zn
K*
P.
c.
EOM
Mutagenicity
Albuquerque
(1985-1986)
20.6
76.5
(0.7)
84.7
59.0
36.0
(1.8)
(1.5)
44.5
74.2
(1.0)
0.2
237.4
507.1
0.0
76.0
(6.9)
6.6
54.1
2.1
13.2
24.5
40.2
Raleigh
(1985-1986)
30.3
9.4
1.2
27.5
17.8
6.9
0.4
19.8
44.4
158.9
2.7
1.0
95.6
1729.3
1.5
75.8
2.5
15.1
140.3
0.5
10.0
20.1
20.1
Boise
(1986-1987)
35.7
102.3
1.5
14.4
25.8
122.2
0.6
11.3
22.1
144.9
1.7
1.5
45.3
602.7
0.8
69.1
1.3
18.5
128.9
1.7
12.7
22.0
45.8
Roanoke
(1988-1989)
19.9
175.6
1.8
5.4
47.2
52.5
1.1
7.1
113.6
176.8
12.0
1.3
26.9
1177.0
1.8
76.6
4.0
82.8
151.9
1.5
7.3
11.0
19.0
Mass, C., C,, and EOM in micrograms per cubic meter; elements in nanograms per cubic meter;
mutagenicity in revertants per cubic meter.
188
-------
Table 3. Comparison of average 24-h (7 a.m.-7 a.m.) concentrations of key receptor modeling species
at the four cities studied by 1ACP (data from primary residential site locations).
Species
Mass ng/nf
a.m.
p.m.
Pb ng/mP
a.m.
p.m.
K ng/m*
a.m.
p.m.
Cv ng/m1
a.m.
p.m.
Ce jig/m3
a.m.
p.m.
EOM /ig/m3
a.m.
p.m.
Albuquerque
(1984-1985)
20.6
10.7
29.7
237.4
195.1
276.3
54.1
12.6
92.5
13.2
2.1
24.5
Raleigh
(1984-1985)
30.3
95.6
82.0
117.0
140.3
72.0
175.0
10.0
0.5
20.1
Boise
(1986-1987)
35.7
27.8
42.2
45.3
46.9
43.4
128.9
91.5
162.0
12.7
9.4
16.0
1.7
1.8
1.6
22.0
16.2
28.1
Roanoke
(1988-1989)
19.9
17.5
22.0
26.9
24.3
29.2
151.9
125.3
175.2
7.3
5.8
8.5
1.5
1.5
1.5
11.0
7.0
14.9
*K corrected for soil content
Table 4. Comparison of selected VOC values (ppbC).
Species
Acetylene
Benzene
2-Methylpentane
2-MethyIhexane
Toluene
Total Nonmethane
Hydrocarbons
Site*
Residential
Roadway
Residential
Roadway
Residential
Roadway
Residential
Roadway
Residential
Roadway
Residential
Roadway
Average
13.2
19.3
13.5
18.4
9.8
16.0
8.4
13.8
25.5
42.1
455.6
636.4
Boise
a.m.
15.6
20.9
13.0
18.8
9.8
17.0
8.3
14.6
25.3
44.0
453.2
656.7
p.m.
10.8
17.6
14.0
18.0
9.9
14.9
8.4
12.9
25.6
40.0
458.0
615.5
Average
13.4
25.5
7.4
12.9
6.2
11.1
3.6
6.8
15.5
30.4
334.3
624.4
Roanoke
a.m.
11.8
21.0
6.1
10.0
3.0
6.2
2.7
5.0
12.9
23.9
280.0
438.0
p.m.
15.1
29.9
8.8
15.6
4.5
8.7
4.4
8.7
18.3
36.8
390.0
806.0
* Primary sites.
189
-------
Mutagenicity of Indoor and Outdoor Air in Boise, Idaho and Roanoke, Virginia
Debra Walsh1, Sarah Warren1, Roy Zweidinger2, Larry Claxton2, and Joellen Lewtas2
1 Environmental Health Research and Testing, Inc.
Research Triangle Park, NC 27709
2 US Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The mutagenicity of ambient air samples, collected in two large field studies was
examined using the Salmonella microsuspension bioassay.1'2 The first study was conducted
during the 1986-1987 winter heating season in Boise, Idaho. The second study was
conducted in Roanoke, Virginia from October 1988 to February 1989. The study design for
both sampling locations was very similar and has been described previously.2 The field
studies consisted of an ambient air component at stationery primary site locations and a
residential component that involved sampling different homes each week. The primary sites
included a location dominated by residential heating emissions, a site dominated by mobile
source emissions, and a background site. The residential study component consisted of
twenty homes being sampled in matched pairs (10 pairs) over a four-day weekend and
weekday period. Each pair of homes included one home which operated either a residential
oil heater or wood burning appliance (e.g. wood stove or fireplace) and the other home did
not. The Salmonella mutagenicity (revertants/m3) of both particle samples (collected with
PM10 medium flow (0.113 m3/min) samplers) and semivolatile organic compound (SVOC)
samples (collected on XAD-2 cartridges) was determined by simultaneous collection of
indoor and outdoor samples from each pair of homes. The initial results of these studies
have shown that outdoor air significantly influences the mutagenicity of particles indoors.
The concentration of particle mutagenicity indoors is consistently lower than outdoors by a
factor related to the particle infiltration rate. The concentration of mutagenicity from the
SVOCs indoors is greater than outdoors and may be related to indoor sources.
INTRODUCTION
Ambient aerosols (particle matter) in urban areas typically contains condensed
organic matter from combustion emissions. A number of studies have shown that the
extractable organic matter (EOM) from air particles is carcinogenic in animals and
mutagenic in short term bioassays4. Short term mutagenicity bioassays have been used in
ambient air monitoring studies and chemical characterization studies in order to identify the
major emission sources and compounds which contribute to the mutagenic activity of air
paniculate extracts5'6. Small area combustion sources, primarily vehicles and home heating
sources, have been shown to account for most of the mutagenic activity associated with air
particulate matter in several urban areas6'7. These studies were designed to estimate the
190
-------
contribution of combustion sources to the cancer risk in ambient urban aerosol in a relatively
simple airshed.
The US EPA's Integrated Air Cancer Project (IACP) is a multi-year research
program which set as one of its primary goals identification of the major sources of cancer
risk in urban air8'9. The first study in Boise focused on a relatively simple airshed containing
two major area sources (wood burning and motor vehicles) and no major industrial sources.
The second study in Roanoke continued the same design with the addition of residential oil
burning. This paper describes the microsuspension mutagenicity studies conducted on
aerosol samples collected in Boise and Roanoke.
EXPERIMENTAL METHODS
Experimental Design
The overall IACP program8'9 and field study design10 and sampling methodologies11
have been previously reported. Two primary sites were selected to represent a residential
wood smoke impacted area with minimal mobile source impact [Elm Grove Park (EGP) Site
in Boise and Morning Side Park (MSP) in Roanoke] and a site with maximal mobile source
impact [Fire Station (FS) Site in Boise and the Civic Center (CIV) in Roanoke], located
near a major intersection where the mobile sources included a mixture of heavy duty and
light duty diesel and gasoline vehicles.
The residential sampling involved pairs of homes. Each pair of homes included one
home which operated a wood burning appliance (e.g. wood stove or fireplace) or oil burning
furnace and the other home did not. None of the homes used in this study had other
unvented stoves (e.g. kerosene heaters) or tobacco smoking residents. The homes where
matched as closely as possible with respect to age, size and other factors and were located
as closely as possible to each other and the ambient monitoring sites.3'12
The pair of homes were sampled simultaneously. A total of 20 homes, in each study,
were sampled and the homes without the wood or oil burning appliances were also sampled
outside the home . This design allowed a comparison of homes with and homes without
wood burning or oil burning and also comparison of inside and outside the homes. Sampling
started at 7:00 am Saturday and samples were collected every 12 hours until Wednesday at
7:00 am. A total of eight 12-hour samples were collected on two weekend days and two
weekdays at each home. Sampling at the primary sites followed this same schedule.
Samples for bioassay were collected using PM10 medium flow (0.113 m3/min)
samplers with a size selective inlet to exclude particles greater than 10 ju,m. The particles
less than 10 /mi were collected on Pallflex 102 mm T60A20 Teflon-impregnated glass fiber
filters. Semivolatile organic compounds were collected on XAD-2 canisters which were
placed after the filter. The complete description of the entire residential sampling and
analysis is reported elsewhere.12
All filter samples for the bioassay studies were transported on dry-ice and stored at
191
-------
-80°C from the time of collection until extraction. The filters were combined in Soxhlet
devices and extracted for 24 hours with dichloromethane and the extracts filtered (0.2/tm
Millipore-type FG filters). The extracts were concentrated on a rotary evaporator with a
35°C water bath to reduce the volume prior to quantitative transfer to volumetric flasks. A
portion of the extract was used for a gravimetric determination.
Sample Composites
The amount of extractable organic matter (EOM) needed for mutagenicity testing
required the compositing of samples. Weekend daytime (Saturday 7am and Sunday 7am)
were composited for one sample, likewise the weekend nighttime samples (Saturday 7pm
and Sunday 7pm) were composited for one bioassay sample. The weekday (Monday and
Tuesday) samples followed the same design. The composite samples were assigned unique
numbers throughout the study. The filter samples containing the particles and the XAD-2
samples containing the semivolatiles were both composited using this design.
Mutagenicitv and EOM Methods
The Ames Salmonella typhimurium microsuspension assay was conducted by
previously published guidelines1'3 in tester strain TA98 both with (+S9) and without (-S9)
a metabolic activation system provided by an Aroclor induced CD-I rat liver homogenate.
The dose range used was from 0.1, 0.3, 0.5, 1.0, 3.0, 5.0 and 10.0 cubic meters (m3) per plate
with 2 plates per dose. Each assay included negative (solvent controls) and the appropriate
positive control chemicals for TA98 (+ and -S9) for quality control. The initial linear slope
of the dose response curve was analyzed using the method of Bernstein et al12. All samples
were analyzed in two separate experiments and the data were summarized together.
Gravimetric measurements were also performed on each of these samples to
determine /ig of extractable organic material (EOM). The mutagenic response is expressed
as revertants/m3 and the jugEOM/m3 can be calculated.
RESULTS AND DISCUSSION
Mutagenicitv of semivolatile and particle samples
The mutagenicity testing on both the semivolatiles and the particle samples were
conducted on a per cubic meter of air sampled. Separate samples were collected at the
primary sites during the same sampling times to determine fine particle mass, therefore,
percent extractables could be estimated. Table 1 shows the fine particle mass and EOM for
the primary sites at Boise and Roanoke in the daytime samples and nighttime samples. The
fine particle mass collected at the different sites varies. However, the estimated percent
EOM between the daytime and nighttime in Boise varies only a few percent. The daytime
percent extractable at EGP is 55% compared to 53% for the nighttime sample. However,
the difference in the estimated percent extractables between the daytime and nighttime in
Roanoke is greater. The percent extractable at MSP increases from 35% for the daytime
sample to 68% for the nighttime. The fine particle mass for MSP increases from 17/ig/m3
192
-------
to 22/ng/m3, a 29% increase but the EOM increases from 6 pg/m3 to 15/*g/m3, a 150%
increase. This difference in the EOM indicates that the composition of the samples is
different. The higher percent EOM is likely to be due to an impact from wood smoke
emissions that has a much higher % EOM than mobile sources or residential oil burning.
Table 1. Comparisons of Primary Site Day and Night Fine Particle Mass and Particle EOM
in Boise and Roanoke (Micrograms per cubic meter).
city
Boise
Roanoke
She
EGP
FS
MSP
CIV
Mass
AM
27
24
17
21
EOM
AM
15
11
6
8
%
Ext.
55
46
35
38
Mass
PM
43
39
22
29
EOM
PM
23
20
15
20
:% ;
Ext,
53
51
68
69
The mutagenicity of both the filter (fine particles) and the XAD-2 (SVOCs) for both
cities by both daytime and nighttime are shown in Table 2 in revertants per cubic meter of
air. There is little difference in the total mutagenicity between the daytime and nighttime
samples in the Boise airshed. However, the mutagenicity increases in the nighttime samples
in the Roanoke airshed. This increase in mutagenicity between the daytime and nighttime
correlates with the increase in EOM seen in Roanoke.
Table 2. Comparisons of Primary Site Daytime and Nighttime Mutagenicity in Boise and
Roanoke. (Revertants per cubic meter)
City
Boise
Roanoke
Site
EGP
FS
MSP
CIV
Filter
AM
67
73
35
56
XAD
AM
31
26
19
49
Total
98
99
54
105
Filter
PM
67
67
73
119
XAD
PM
33
31
30
40
Total
100
98
103
159
The contribution of the semivolatiles (XAD-2) to the total mutagenicity at the
primary site for the daytime samples is very similar between the residential primary sites in
Boise and Roanoke (EGP and MSP). However, the mutagenic contribution of the
semivolatiles from the daytime samples at the Roanoke mobile source primary site (CIV)
is greater than that of the Boise mobile source primary site (FS).This is shown graphically
in Figure 1.
193
-------
This difference in semivolatile
contribution to the total mutagenicity is not
seen in the nighttime samples in Roanoke.
The nighttime mutagenicity at the CIV in
Roanoke is mostly from the particles on the
filter. This suggests that the composition of
the airshed in Roanoke varies more from
daytime to nighttime than the Boise
airshed. The contribution of the filter and
XAD to the total has a very similar profile
in Boise for both daytime and nighttime.
The nighttime comparison of the primary
sites is shown in Figure 2. The increase in
nighttime total mutagenic potency at MSP
would appear to be related to a different
species collected on the particles. Whereas
the difference in the mutagenic potency of
the nighttime sample at the CIV could be
related to not only the different species on
the filter but also the increase in the mass
of the particles collected.
Primary Site Comparison
Salmonella Microauepvnalon CT*9e *39)
Primary Site Comparison
lion CTA9B -SB)
Figure 1 Daytime Primary Site
Comparison between Boise and
Roanoke (TA98 +S9)
Figure 2 Nighttime Primary Site
Comparison between Boise and
Roanoke (TA98 +S9)
Indoor and Outdoor Mutagenicity
The residential mutagenicity in both
Boise and Roanoke was analyzed to
determine indoor and outdoor relationships.
Comparisons were made between: homes
with wood burning or oil burning (In/With);
homes without these heating sources
(In/Without); and outside the homes (Out).
These samples were composited for the
weekend, weekday, daytime, and nighttime
sampling periods.
The mutagenic potency as measured
in revertants per cubic meter for the homes
in Boise and Roanoke is shown in Table 3.
I'M
-------
Table 3. Comparisons of Residential Sites Daytime and Nighttime Mutagenicity in Boise
and Roanoke (Revertants per Cubic meter)
city
Boise
Roanoke
Homes
Out
In/
Without
In/With
Out
In/
Without
In/With
filter
AM' :
61
21
28
40
27
15
XAD
AM
42
16
22
21
13
23
Total
103
37
50
61
40
38
Filter
PM
61
24
27
54
24
27
XAD
PM
26
21
34
15
15
27
Total
87
45
61
68
39
54
•" In With:
In Without:
Out:
Boise homes - heating source wood; Roanoke homes - heating source oil
Boise and Roanoke homes - heating source either natural gas or electric
Boise and Roanoke — outside the homes using either natural gas or electric
The residential outside mutagenicity in Boise was consistent with the EGP
mutagenicity measured in both the daytime and nighttime. The Boise daytime mutagenicity
was 98 rev/m3 at EGP (Table 2) and outside the homes the average was 103 rev/m3 (Table
3). The correlation in Roanoke during the daytime was still similar with MSP at 54 rev/m3
(Table 2) and outside the homes at 61 rev/m3 (Table 3). The daytime mutagenicity inside
the homes in Roanoke was similar between the homes with and the homes without but was
somewhat lower than the outside air. The daytime mutagenicity in Boise was slightly higher
in the homes with ,in comparison to the homes without but as in Roanoke the outside air
was significantly higher. This relationship in Boise could be influenced by one home where
the wood stove was known to be leaking and therefore increased the inside emissions.
Correlation analysis of the particle mutagenicity showed the mutagenicity inside the
homes was correlated with the mutagenicity of the outside air. When the revertants per
cubic meter were plotted for all the homes on the same graph the inside mutagenicity
increased and decreased with the outside air. This was observed in both the Boise and
Roanoke study.
ACKNOWLEDGEMENTS AND DISCLAIMER
The authors acknowledge the contribution of the IACP field sampling team directed
by V.R. Highsmith and A.J. Hoffman, and bioassay analysis by E. Perry. This paper has
been reviewed in accordance with the US EPA's peer review polices for scientific papers and
approved for publication. The contents are not intended to represent the Agency policy.
195
-------
References
1. DeMarini, D.M., Dallas, MM., and Lewtas, J. (1989) Cytotoxicity and effect of mutagenicity of buffers in
a microsuspension assay, Teratogenesis. Carcinoeenesis. and Mutaeenesis. 9:287-295.
2. Kado, N.Y., Langley, D. and Eisenstadt, E. (1983) A simple modification of the Salmonella liquid
incubation assay. Increased sensitivity for detecting mutagens in human urine, Mutat. Res.. 121:25-32.
3. Highsmith, V.R., AJ. Hoffman, R.B. Zweidinger, L.T. Cupitt, and D.B. Walsh (1991)
The IACP: Overview of the Boise, Idaho and the Roanoke, Virginia, Field Studies, in Proceeding of the
1991 AWMA Meeting, paper 91-131.1
4. Lewtas, J. (1990) Experimental evidence for the carcinogenicity of air pollutants, Air Pollution and Human
Cancer. L. Tomatis, Ed., Springer-Verlag, Berlin, pp. 49-61
5. Lewtas, J. (1988) Genotoxicity of complex mixtures: Strategies for the identification and comparative
assessment of airborne mutagens and carcinogens from combustion sources, Fundamental and Applied
Toxicology 10:571-589.
6. Lewis, C.W., R.E. Baumgardner, L.D. Claxton, J. Lewtas, R.K. Stevens (1988) The contribution of wood
smoke and motor vehicle emissions to ambient aerosol mutagenicity, Environ Sci Technol 22:968-971.
7. Stevens, R.K., et al.(1990) Sources of mutagenic activity in urban fine particles, Tox. Indust. Health 6(5):81-
94.
8. Lewtas, J. (1989) Emerging methodologies for assessment of complex mixtures: Application of bioassays
in the Integrated Air Cancer Project, Tox. Indust. Health S(5):839-850.
9. Cupitt, L. and J. Lewtas, EPA's Integrated Air Cancer Project, Tox. Indust. Health, in press.
10. Cupitt, L.T. and T.R. Fitz-Simmons (1988) The Integrated Air Cancer Project: Overview and Boise Survey
Results, in Proceedings of the 1988 EPA/AWMA International Symposium on Measurement and Related
Air Pollutants. VIP-10, Air & Waste Management Association, Pittsburgh, pp. 799-803.
11. Highsmith, V.R., R.B. Zweidinger, J. Lewtas, A Wisbith, and R.J. Hardy (1988) Impact of Residential
Wood Combustion and Automotive Emissions on the Boise, Idaho, Airshed, in Proceedings of the 1988
EPA/AWMA International Symposium on Measurement and Related Air Pollutants. VIP-10, Air & Waste
Management Association, Pittsburgh, pp. 804-813.
12. Highsmith, V.R., Cupitt, L.T., Zweidinger, R.B., Glen, W.G., Wu, J., and Lewtas, J. (in press)
Characterizing the influence of residential wood combustion and mobile source emissions on the indoor
air quality of selected Boise, Idaho residences, Atmospheric Environ..
13. Bernstein, L. et al. (1982) An empirical approach to the statistical analysis of mutagenesis data from the
Salmonella test, Mutat. Res. 97:267.
196
-------
Radiocarbon Measurements of Extractable Organic Matter from the Integrated Air Cancer Project Study
in Roanoke, VA'
Donna B. Kiinedinst, George A. Klouda, and Lloyd A. Curne
National Institute of Standards and Technology
Surface and Microanalysis Science Division
Gaithersburg, Maryland 20899
Roy B. Zweidinger
U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, North Carolina 27711
ABSTRACT
The use of radiocarbon (I4C) as a unique discriminator of fossil and modern combustion aerosols has
been applied to a number of studies for the purpose of apportioning specific sources of atmospheric pollutants.
It has been used previously as part of the measurement program of the U.S. Environmental Protection
Agency's Integrated Air Cancer Project (IACP). The objective of the work reported here was to determine the
relative contribution of modern and fossil sources on wintertime organic aerosols in wintertime Roanoke, VA
(1988 - 1989), as part of the IACP. Radiocarbon measurements were performed on the extractable organic
matter (EOM) separated from the fine size fraction of paniculate samples collected in Roanoke. Results of
these analyses suggest that biogenic carbon, derived primarily from residential wood combustion, was the
dominate source of atmospheric aerosol EOM at both the residential and rural background sampling sites. The
in town traffic site was more heavily influenced by fossil carbon, derived primarily from motor vehicle
emissions.
INTRODUCTION
The use of radiocarbon (I4C) as a unique discriminator of fossil and modern combustion aerosols has
been applied to atmospheric source apportionment studies of both particles and gases (Klouda, et al., 1986,
Sheffield, et al., 1991). In particular, it has been used in several studies conducted as part of the IACP
(Klouda, et al., 1987, Klouda, el al., 1991). The objective of the work reported here was to determine the
relative contribution of modern and fossil sources to wintertime organic matter extracted from the fine size
'Contribution of the National Institute of Standards and Technology; not subject to copyright.
197
-------
fraction of paniculate samples collected in Roanoke during the winter of 1988 - 1989. In all, radiocarbon
measurements were made on twenty nine samples and nine filter blank EOM extracts. This paper presents the
results of experimental methods, quality control (QC) procedures and 14C analyses performed in support of
this study by us at the National Institute of Standards and Technology (NIST) .
SAMPLE HISTORY
The field sampling and EOM extractions portions of this study were carried out by EPA contract
laboratories. These procedures are summarized here.
Field Sampling: Fine paniculate material, with an aerodynamic diameter of <2.5 urn, was collected on
high purity quartz-fiber filters (ca. 395 cm2) using high-volume samplers equipped with Sierra impactor inlets
in accordance with IACP sampling protocols. The aerosol samples were collected for 12 hour periods
(beginning at 0700h and 1900h) at three separate locations. The sampling locations in Roanoke were (1)
Morningside Park (MSP), an in-town residential area, (2) the Civic Center (CIV), an in-town traffic site, and
(3) Carvin Cove (CC), a rural background site. Subsequently, the EPA's Atmospheric Research and Exposure
Assessment Laboratory (AREAL) selected a subset of these samples and extracted them with dichloromethane
(DCM) to obtain corresponding liquid samples of extractable organic matter (EOM). •
EOM Extraction: Aerosol sample filters and filter blanks were extracted for 21-23 hours with 150
mL of DCM in 35/45 Soxhlet extractors to separate all nonvolatile organic material for I4C analysis. The
extracts were then filtered using 0.2 urn FG type Millipore filters and concentrated to less than 5 mL using
rotary evaporation. After quantitatively transferring the samples to 10 mL volumetric flasks and diluting to
volume with DCM, duplicate 400 uL gravimetric determinations were made. Aliquots of 8 mL were
transferred to sample vials for carbon separation and 14C analysis, and the residual sample volume was
archived at AREAL.
EXPERIMENTAL METHODS AND QUALITY CONTROL
At NIST, Roanoke sample extracts and filter blanks extracts were characterized for their carbon
198
-------
concentration, prepared as AMS targets, and then analyzed for I4C . To accomplish this, three separate
analytical procedures were utilized.
Carbon Separation Procedure: Carbon in the EOM (e.g., carbon emanating from the sample and/or the
DCM and filter blank) was separated from both the samples and filter blanks by evaporating the DCM solvent,
followed by closed tube combustion of the carbon in the sample residue to CO2. First, extract aliquots of 500
- 2000 jil were quantitatively transferred to clean Vycor tubes. Filtered nitrogen gas was passed over the
sample until the aliquot reached dryness. Next, a stoichiometric excess of copper(II) oxide and silver wire
were added to the Vycor tube to provide oxygen for combustion and to remove oxides of sulfur. The sample
tube was evacuated, sealed, and heated at 900 °C for 3 - 5 hours. The evolved CO2 was separated
cryogenically and the mass of carbon was determined manometrically in a calibrated volume. The sample
C02 was then transferred to a sealed sample vial for storage prior to AMS target preparation. A more
detailed description of the evaporation and combustion procedure can be found in Sheffield, et al., (1990) and
Kloudaet a/., (1991).
The following Standard Reference Materials and reference solutions were used to validate the carbon
separation procedure and to maintain quality control throughout sample processing: NIST archived DCM
extracts of 1) SRM 1648, ( St. Louis Urban Paniculate Matter) and 2) SRM 1650, (Diesel Soot Particulate
Matter). These solutions were used to evaluate the precision of the carbon separation procedure. Triplicate
analyses from single extract samples of these two SRM extracts yielded carbon concentration estimates of
10.41 ± 0.44 mg/mL (sd, n = 3), and 12.88 ± 0.23 mg/mL (sd, n = 3), for SRM 1648 and SRM 1650,
respectively. The precision of the carbon separation procedure, in terms of the relative standard deviations
was, therefore, 2 - 4 percent. These results are consistent with those reported for the Boise IACP study by
Klouda, a al., (1991) .
To evaluate the accuracy of the carbon separation procedure, the following reference solutions,
prepared in our laboratory for this study, were used:
PAH #1: C,4 - C22, 0.427 ± 0.01 mg C/mL DCM
199
-------
PAH #2: C|4 - C24, 91.43 ± 0.21 mg C/mL DCM
Pure polycyclic aromatic hydrocarbons (PAH) were weighed and dissolved with DCM, then diluted to a
known volume. The carbon concentration value was obtained by multiplying each PAH concentration by its
respective percentage of carbon and summing the individual carbon concentrations. The average recovery for
these reference solutions was 100.7 ± 1.6% (sd, n = 10), translating into an accuracy of 2%.
Finally, a NIST laboratory DCM Blank was processed for the purpose of evaluating the potential
carbon contamination associated with the evaporation and combustion procedures. For each set of five
samples, a 1 or 2 mL aliquot of High Performance Liquid Chromatography DCM was processed. These DCM
aliquots spanned the volume range of the EOM extract carbon separation aliquots. The average DCM blank,
normalized to 1 mL, was 2.81 ± 0.81 ug C (sd, n = 10). This is less than 2% of the sample carbon and less
than 5% of the filter blank carbon processed for I4C analysis.
AMS Fe-C Target Preparation: The final step of sample preparation was the production of an Fe-C
target for AMS analysis. Target preparation was carried out in a system described by Verkouteren, el at.,
(1987) and Verkouteren and Klouda (1991). Sample CO2 was quantitatively transferred to a 6 mm OD quartz
tube adapted to a vacuum manifold. The CO2 from sample combustion was next reduced to graphite on iron
wool at thermodynamically favorable temperatures in the presence of hydrogen and zinc, then fused into an
Fe-C bead using a hydrogen-oxygen flame.
Prior to transferring sample C02 to the reduction manifold, the sample carbon was once again
quantified by manometry in the reduction manifold calibrated volume. As an estimate of the sample transfer
efficiency, the ratio of this carbon mass determination to that obtained from the carbon separation procedure
was calculated yielding a result of 95.9 ± 5.7% (sd, n = 56). As an additional quality control check of the
reduction procedure, RM 21 (14C blank graphite) was processed in parallel with the filter samples, for the
purpose of estimating the level of the I4C blank resulting from the target preparation and measurement
processes. Additionally, the oxalic acid I4C standards, SRM 4990b (HOx(I)) and 4990c [HOx(II)], processed
at the same carbon mass ranges as those of the samples, were combusted and prepared as Fe-C targets.
AMS Radiocarbon Measurement Procedures: Accelerator mass spectrometry measurements were made
200
-------
on sample Fe-C targets at the NSF-University of Arizona Accelerator Facility for Radioisotope Analysis
according to the protocol outlined by Linick et al., (1986) and Donahue (1991). For each 32 position target
wheel, a minimum of one pair of oxalic acid radiocarbon standards [HOx(I)] and [HOx(II)] plus an RM 21
blank was measured. During each rotation of the sample wheel, individual targets were exposed to the
accelerator's cesium beam for approximately 10 minutes. During this time, the 14C and 13C counts were
measured by cycling between the two for 40 and 4 seconds, respectively. The number of complete wheel
rotations for the Roanoke samples ranged from 3 to 6.
RESULTS
Fraction of Modem Carbon. The measured 14C/13C signal ratios of the HOX(I) and HOx(II) pairs
were converted to the fraction of Modern Carbon (fM ) according to equation 1.
/ Uf I 13/-\
\ *•• ' '-Inarm ilgnal
("C/"C)ffQ<(fl «^.0.
,„
The resulting fM ratio was then compared to the NIST certified value for this ratio, 1.3406 ± 0.0008, to yield
the accuracy of the AMS measurements, expressed as A%. The uncertainty of the fraction of Modem Carbon
for both standards and samples was based on the weighted counting statistics of the measured 14C/13C ratio
signals. The accuracy of the 14C measurements for the Roanoke study ranged from -0.09 to 1.4%, for targets
ranging from 32 to 599 ug C. The mass of 14C blank was calculated from the RM 21 data by multiplying the
fM value of RM 21 by the target carbon mass. The average mass of 14C contamination was 3.9 ± 0.7 ug C
(sd, n = 2). This calculation assumed that the contamination is modern carbon.
The fraction of Modern Carbon (fM ) for samples was also calculated according to Eq. 1 by replacing
the numerator with (14C/'3C ) samp]e signa|- The uncertainty of the calculated sample fM was derived by
propagation of the errors (see Ku, 1966) of the measured 14C/'3C ratio signal of the sample and its associated
201
-------
HOx(I) standard. The precision of the fM results for all samples and blanks was 2% or less.
Blank Corrections. Before these fM results can be used to model atmospheric concentration, a
correction for filter blank contributions must be made and a confidence interval estimated. The correction for
filter blank carbon contributions to the mass and fM of a sample was calculated according to Eq. 2, where
the mean value of the filter blank mass of carbon and fM, corresponding to the sample extract set (1 or 2) was
used.
where:
, mass ft C
= - -
(mass sample C + mass ft C)
The mean filter blank mass and fM for extract set 1 was 69.7 ± 33.0 ug C and 0.54 ± 0.12 (sd, n= 6),
respectively, yielding relative standard deviations of 47% and 22%. Additionally, we noted a linear correlation
between these two variables. The mean filter blank mass and fM of extract set 2 was 27.1 ± 4.7 ug C and
1.22 ± 0.35 (sd, n = 3), yielding relative standard deviations of 17% and 29%, respectively.
Conventional error propagation to estimate uncertainties is a reliable method to propagate errors
through linear functions. When functions are non-linear, such as Eq. 2, and relative errors of the variables are
small, an approximate solution can be obtained by expanding the function using a Taylor's series and assuming
that only the lower order terms are necessary to describe the function. This assumption breaks down when the
relative errors of a variable are large, particularly if they are associated with the non-linear portion of the
function, as is the case for the Roanoke filter blanks. These errors could be evaluated by expanding the
Taylor series function to include higher order terms, however, this would result in a very complicated function.
Instead, Monte Carlo error propagation simulations were used to calculate the uncertainties and construct
202
-------
approximate 90% confidence intervals for the Roanoke t'M (Corr.) data. The results from these simulations
yielded asymmetric confidence intervals for fM (Corr.) in which the width ranged from 0.03 to 0.67. (Note:
confidence intervals routinely calculated as CI - x ± tf*a yield confidence intervals symmetric about the
estimated value of x.) We have chosen an approximate 90% confidence interval width of 0.30 as a cut off
beyond which the results may not be useful for further modeling of the 14C contribution to the EOM in
Roanoke. Seven of the filter samples fell within this category. Two additional samples may also not be
useful for modeling calculations because the recovered carbon mass was inconsistent (e.g., too high or too
low) with the EOM mass data. The results of the fM (Corr.) calculation and error propagation are presented in
Figure 1. The nine samples discussed above are indicated by dashed lines or asterisks. The results are
summarized by sampling site and time.
V.
Residential Wood Combustion (RWC) Contribution: Roanoke, VA is assumed to have three major
carbon based emission sources: 1) residential wood combustion (biogenic carbon), 2) motor vehicle exhaust
(fossil carbon), and 3) residential heating oil combustion (fossil fuel). The two fossil sources cannot be
distinguished by 14C analysis and thus, the model reduces to two components; biogenic and fossil carbon. To
estimate the relative source contributions, blank- corrected fM results [fM ( Corr.)] were normalized to the 14C
abundance of the average age of wood burned during the Roanoke IACP sampling period. Our estimate of the
average age of wood burned in Roanoke during the winter of 1988-89 was 30 - 60 years. The relative
contribution of RWC to the EOM fraction, is given as fRWC according to Eq. 3:
/we = VM EOM) I (fu WOOD) (3)
Note that the calculation of RWC is a systematic correction to the sample fM (Corr.) value, which
depends on the value chosen for fM Wood. Using an equal mass tree ring model, the fM value for wood
between 30 - 60 years is 1.19 -1.40. ( see Klouda, et. al., 1991). The fM value for an average age of 45 years
is 1.26. Thus, the conversion of fM (Corr.) to RWC adds an additional uncertainty to the fM (Corr.) values of
8 to 11%.
203
-------
CONCLUSIONS
Radiocarbon measurements were made on twenty nine EOM sample extracts and nine filter blanks
separated from fine particle filters collected in Roanoke, VA during the winter of 1988 -1989. The fraction of
Modern Carbon was calculated from the measured 14C/'3C signal ratios of the sample and its associated oxalic
acid radiocarbon standard (SRM 4990b). All sample fM values were corrected for filter blank contributions to
the carbon mass and fraction of Modern Carbon. Monte Carlo error simulations were employed to estimate
the uncertainty of the blank corrected fraction of Modern Carbon because of the large relative errors in the
mean values of the filter blank mass and fraction of modem carbon. A total of seven samples have
approximate 90% confidence interval widths that exceed 0.30. Further use of the data from these seven
samples requires rigorous evaluation of the error imparted by these large confidence intervals. Nevertheless,
the results indicate that residential wood combustion was the dominant source of atmospheric aerosols EOM
for both the day and night at the residential site (Morningside Park) and the rural background site (Carvin
Cove). The traffic site (Civic Center) yielded lower fM (Corr.) values overall, due mainly to the impact of
motor vehicle emissions. Additionally, the fM( Coir.) data suggest that on average, the nighttime samples have
higher concentrations of 14C . Use of these results to model modem carbon contribution to the EOM fraction
in Roanoke are reported elsewhere in these proceedings, Lewis, el al., (1993).
ACKNOWLEDGEMENTS
The authors would like to thank D.J. Donahue, A.J.T. Ml, A.L. Hatheway and Dana Bidulth for their
assistance in the measurement of I4C by AMS.
Disclaimer This work was funded in part by the United States Environmental Protection Agency under
Interagency Agreement #DW13935098-01-1 to the National Institute of Standards and Technology. It has
been subjected to Agency review and approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use, nor does it imply that the materials or
equipment identified are necessarily the best available for the purpose.
204
-------
REFERENCES CITED
Ku, H.H., "Notes on the Use of Propagation of Error Formulas", J. Res. NABS. 7QC (Engr. and Inst.1 4: 263-
273 (1966).
Lewis, Charles W., Zweidinger, Roy B, Claxton, Larry D., Klinedinst, Donna B.,and Warren, Sarah H.,
"Source Apportionment of Fine Particle Organics and Mutagenicity in Wintertime Roanoke", Proceedings of
the 1993 International Symposium on Measurements of Toxic and Related Air Pollutants, these proceedings.
Sheffield, A.E., Currie, L.A., Klouda, G.A., Donahue, D.J., Linick, T.W., Jull, A.J.T., "Accelerator Mass
Spectrometry Determination of Carbon-14 in the Low-polarity Organic Fraction of Atmospheric
Particles."Analvtical Chemistry. 62 (19): 2098 (1990).
Klouda, G.A., Currie, L.A., Donahue, D.J., Jull, A.J.T., and Naylor, M.H., "Urban Atmospheric 14CO and
14CH4 Measurements by Accelerator Mass Spectrometry", Radiocarbon, 28(2A): 625 - 633, (1986).
Klouda, G.A., Currie, L.A., Sheffield, A. E., Wise, S.A., Benner, B.A., Stevens, R.A., and Merrill, R.G., "The
Source Apportionment of Carbonaceous Combustion Products by Micro-Radiocarbon Measurements for the
Integrated Air Cancer Project (IACP)", in Proceedings of the 1987 EPA/ACPA Symposium on Measurement
of Toxic and Related Air Pollutants, pp. 573 - 578, (1987).
Klouda, G.A., Barrraclough, D., Currie, L.A., Zweidinger, R.B., Lewis, C.W., Stevens, R.K., "Source
Apportionment of Wintertime Organic Aerosols in Boise, ID by Chemical and Isotopic (I4C) Methods", in
Proceedings of the 84th AWMA Annual Meeting. Vancouver B.C. Canada : paper # 91-131.2 (1991).
Verkouteren, R.M., Klouda, G.A., Currie, L.A., Donahue, D.J., Jull, A.J.T., Linick, T.W., "Preparation of
Microgram Samples on Iron Wool for Radiocarbon Analysis via Accelerator Mass Spectrometry: A Closed-
system Approach", Nucl. Instr. and Meth.. B29: 41 (1987).
Verkouteren, R.M., and Klouda, G.A., "Factorial Design Techniques Applied to Optimization of AMS
Graphite Target Preparation", Radiocarbon. V32(2): 335 - 343 (1991).
205
-------
Roanoke, VA Winter of 1988 -1989
Blank Corrected Fraction of Modern Carbon by Site and Time
at the Approximate 90% Confidence Interval
c
o
•S
CO
O
c
0)
•a
o
1
c
g
l£
0)
ts
*.' *-.*
k' c? x^
^ / /
Fig. 1: Blank corrected fraction of Modern Carbon in Roanoke VA., during the winter of 1988 -1989.
Open symbols represent daytime samples. Filled symbols represent nighttime samples. Error bars
represent the approximate 90% confidence interval. Note the asymmetry of the confidence intervals.
Those samples indicated by dotted lines have confidence interval widths that exceed 0.30. Further
use of these data require rigorous attention to the error structure (see text for details). Two additional
samples, indicated by asterisks, are also not suggested for further use because of inconsistencies
between the EOM and carbon mass results.
-------
SOURCE APPORTIONMENT OF FINE PARTICLE ORGANICS
AND HUTAGENICITY IN WINTERTIME ROANOKE
Charles W. Lewis and Roy B. Zweidinger
U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, NC 27711
Larry D. Claxton
U.S. Environmental Protection Agency
Health Effects Research Laboratory
Research Triangle Park, NC 27711
Donna B. Klinedinst
National Institute of Standards and Technology
Gaithersburg, MD 20899
Sarah H. Warren
Environmental Health Research and Testing
Research Triangle Park, NC 27709
ABSTRACT
During the 1988-1989 winter the U.S. EPA conducted a
comprehensive field study in Roanoke VA as part of its Integrated Air
Cancer Project (IACP). This paper presents results of the source
apportionment of fine particle extractable organic matter (EOM) and
its associated mutagenicity (Salmonella typhimurium TA98 +S9 and
TA98 -S9). The source apportionment methodology is based on multiple
linear regression (MLR) using a variety of tracer species: 14C,
metallic elements and volatile hydrocarbons (VHC) whose ambient
concentrations were measured simultaneously with the EOM and
mutagenicity. The results are compared with those from previous IACP
studies in other locales.
INTRODUCTION
As part of the IACP the U.S. Environmental Protection Agency
has conducted a series of wintertime field studies in U.S. cities to
measure ambient concentrations of fine particle EOM and associated
mutagenicity. Receptor modeling has been employed with these
measurements to determine the quantitative contributions of various
emissions sources to both EOM and mutagenicity. The present work
gives receptor modeling results for the 1988-1989 field study in
Roanoke VA, an airshed whose principal sources of ambient EOM were
anticipated to be woodsmoke, mobile sources and residential distillate
oil combustion (RDOC).
EXPERIMENTAL
Overviews of the Roanoke ambient field sampling program have been
given1'2. The results that follow are from analyses of 12-h fine
particle (0 - 2.5 /m dia) and VHC samples collected simultaneously at
the two primary sites — Morningside Park (residential site) and Civic
207
-------
Center (roadway site). The analyses followed previous IACP procedures
and generated an ambient data set of the following parameters:
elemental composition of fine aerosol from x-ray fluorescence (XRF)3;
fine EOM from dichloromethane Soxhlet extraction3; TA98 +S9 and TA98 -
S9 mutagenicities of the EOM from plate incorporation bioassays3; 14C
content of the EOM from accelerator mass spectrometry4; and VHCs from
gas chromatography-flame ionization detection5.
MLR RECEPTOR MODELING
The same MLR modeling approach was used as in earlier IACP work.
The measured concentration of the pollutant of interest (i.e, EOM or
mutagenicity) is represented by a sum of individual source
contribution terms, with each term being the product of the measured
concentration of a tracer species for that source and an initially
unknown coefficient that is subsequently determined by an MLR
calculation. A non-zero intercept is allowed for, which can be
regarded as the average contribution of additional sources not
represented in one of the explicitly identified source terms.
In previous IACP work the concentration of fine-particle Pb was
used as a mobile-source tracer. In recent years, however, this has
become less tenable, because of the phaseout of leaded gasoline.
During the preceding IACP study in Boise — while Pb was still a
satisfactory tracer — the use of any one of several VHC species as
tracers was shown to give estimates for the mobile source contribution
which were virtually identical to those produced with Pb6. For the
Roanoke data however the VHC species were clearly superior to Pb, as
judged from the quality of the MLR fits that could be achieved
(largest r2 value). The unimpressive Pb-Br correlation (r2 = 0.65)
exhibited by the Roanoke data also suggested that Pb was not a
reliable mobile source tracer. Consequently, the VHC species 2-
methylhexane (2MeHx) was used instead of Pb in the present work, as it
produced a slightly better MLR fit than the other VHCs.
Fine particle soil-corrected potassium (K1) has proven to be very
useful as a woodsmoke tracer in previous IACP work. In the planning
stage of the Roanoke project, it was conjectured that K1 might serve
as a tracer of overall residential heating. This was because RDOC as
well as wood combustion were anticipated to be important contributors
to ambient EOM concentrations, and the expected similar diurnal
emission patterns of these two source categories would frustrate their
separate estimations by an MLR technique. As shown below however
RDOC was so small in comparison to wood combustion that K1 retained
the same role for Roanoke as it had in previous IACP work. Ondov et
al.7 report a similarly small estimate of the RDOC contribution,
through use of an enriched isotope of samarium as an intentional
tracer.
RESULTS
Modeling
Multiple linear regression of the measured concentrations of EOM
(lJ.g m"3) and R^ and RJJ. mutagenicities (revertants m"3) resulted in the
following equations:
[EOM]; = (40 ± 3)[K'], + ( 656 ± 83 )[2MeHx]i - 0.3 ± 0.7 (1)
n = 40;
208
-------
[R«+]| = (32 ± 6)[K']i + (I960 ± 140) [2MeHx]i + 2.6 ± 1.1 (2)
r2 =0.92, n = 37;
[R«-3i = (21 ± 4)[K']j + (1100 ± 130) [2MeHx]j + 4.2 ± 0.9 (3)
r2 = 0.86, n = 38.
The units of the tracer concentrations [K1] and [2MeHx] are jig m"3 and
ppmC, respectively, and
[K1] = [K] - (0.22 ± 0.01)[Fe]. (4)
The numerical coefficient in equation 4 is the average potassium-to-
iron ratio measured in the coarse-particle (2.5 - 10 Jim dia) fraction,
as done previously3.
The cases used in each of the three fits were approximately
equally divided between the Civic Center and Morningside Park sampling
sites, and between day and night. The quality of the fits, as judged
by the r2 values, were similar to or better than those achieved in the
three earlier IACP studies, each of which involved approximately the
same number of cases as in the present study.
Table I shows the average source contributions to the measured
concentrations of EOM and mutagenicity, averaged over both sampling
sites as well as day and night. The source contributions were
calculated by inserting the averages of the measured values of [K1]
and [2MeHx] into equations 1-3. Table I also gives the calculated
mutagenic potency of the EOM associated with each source. The source
potency (revertants (jig EOM)"1) is simply the ratio of the R to EOM
regression coefficient for that source. For comparison Table I also
includes corresponding results from the three earlier studies (Boise,
Albuquerque and Raleigh). The latter results are from Lewis et al.3,
which gives references to the original work. It is important to note
that the source potency values listed in Table I are derived entirely
from ambient measurements.
Validation by 14C
Since 14C is absent from fossil fuels, its presence in fine
particle atmospheric samples is a direct (non-statistical) indication
of the contribution of contemporary carbon sources, assumed to be
essentially only residential woodburning in the wintertime Roanoke
airshed. The 14C-derived estimate of woodsmoke EOM concentration in a
sample i is given by
[Woodsmoke EOM]j = [EOM], (fRWC)i (5)
with fRWC being the fraction of residential wood combustion carbon in
the sample. On the other hand the term in equation 1 involving [K1];
is an independent statistical estimate of the same quantity. Thus the
right side of equation 5 and [K1], should be linearly related, with a
slope that is the same as the regression coefficient for [K1] in
equation 1, and with no intercept. Figure 1 shows [EOM];* (fRWC)i values
for all available data (n = 20) recommended for use by Klinedinst et
al4, plotted vs [K1]. The straight line in the figure is the product
of [K1] and its regression coefficient from equation 1, with the
dotted band representing the coefficient's uncertainty.
209
-------
For the most part the points in Figure 1 cluster around the line,
showing that the potassium-traced portion of EOM is wood combustion-
related, and indicating by inference that the RDOC portion must be
small in comparison. This conclusion is also supported by a Roanoke
TSP emissions inventory specific to the 1988-1989 wintertime period of
the IACP field study which indicates an RDOC emissions rate that is
only a few percent of that for wood combustion8.
CONCLUSIONS
The Roanoke results presented in this paper together with
corresponding results from the previous IACP studies (Table I) show
some very clear consistencies across four geographically scattered
U.S. airsheds: (1) ambient EOM is dominated by woodsmoke, rather than
mobile sources emissions, for these wintertime studies; (2) the
mutagenic potency (Salmonella typhimurium TA98 +S9) of ambient
woodsmoke is approximately 1 revertant per microgram of EOM; (3) the
mutagenic potency of ambient mobile source emissions is about three
times that of woodsmoke. These consistencies span a period of four
years, and depend neither on the type of wood used nor on the choice
of a mobile source tracer (particulate Pb or volatile hydrocarbons) .
The use of 14C measurements have served to confirm the MLR-based
apportionment of EOM.
The woodsmoke domination of EOM may not be surprising for
Albuquerque and Raleigh, since the sampling site for both was in a
residential neighborhood. For Boise and Roanoke however the results
come from the combining of measurements at both residential and
roadway sites, and woodsmoke still dominates EOM overall. The choice
of sampling site locations within the cities presumably had little
effect on the potency values that were obtained.
The Roanoke results for TA -98 mutagenicity and potency are the
first known ambient-derived values for these parameters.
ACKNOWLEDGEMENTS
We thank Alan Hoffman for his supervision of the Roanoke field
study, Bob Kellogg for XRF analyses and Gwen Belk for EOM analyses.
DISCLAIMER
The information in this document has been funded wholly or in
part by the United States Environmental Protection Agency. It has
been subjected to Agency review and approved for publication. Mention
of trade names or commercial products does not constitute endorsement
or recommendation for use.
REFERENCES
1. V.R. Highsmith, A.J. Hoffman, R.B. Zweidinger et al., "The IACP:
Overview of the Boise, Idaho, and the Roanoke, Virginia, field
studies," in Proceedings of the 84th Annual Meeting of A&WMA, paper
91-131.1, 1991.
2. R.K. Stevens, A.J. Hoffman, J.D. Baugh et al., "Air quality
measurement in Roanoke, VA in support of the 1988-1989 Integrated Air
Cancer Project," these proceedings, 1993.
3. C.W. Lewis, R.K. Stevens, R.B. Zweidinger et al., "Source
apportionment of mutagenic activity of fine particle organics in
Boise, Idaho," in Proceedings of the 84th Annual Meeting of A&WMA,
paper 91-131.3, 1991.
210
-------
4. D.B. Klinedinst, G.A. Klouda, L.A. Currie, "Radiocarbon
measurements of extractable organic matter from the Integrated Air
Cancer Project Study in Roanoke, VA," these proceedings, 1993.
5. R.B. Zweidinger, T.E. Kleindienst and E. Hudgens, "Apportionment of
residential indoor VOCs and aldehydes to indoor and outdoor sources in
Roanoke," in Proceedings of the 84th Annual Meeting of AiWMA, paper
91-131.10, 1991.
6. R.B. Zweidinger, R.K. Stevens, C.W. Lewis and H. Westburg,
"Identification of volatile hydrocarbons as mobile source tracers for
fine-particle organics," Environ. Sci. Technol. 24(4):538 1990.
7. J.M. Ondov, Z.C. Lin, W.R. Kelly et al., "Enriched stable isotopes
of Sm as intentional tracers of diesel and residential oil furnace
emissions in Roanoke, VA," these proceedings, 1993.
8. TRC Environmental Corporation, "IACP Source Inventory for Roanoke,"
Contract No. 68-D9-0173, Work Assignment No. 2/236, 1993.
35
15
5-
0.1 0.2 0.3 0.4 0.5
Soil-Corrected K, ug/m3
0.6
Figure 1. MC-determined extractable organic matter vs soil-corrected
potassium. The straight line and its uncertainty band is given by
(40 ± 3) [K1], from equation 1.
211
-------
TABLE I Average ambient concentrations (EOM, TA98 +S9, and TA98 -S9)
and mutagenic potencies attributed to woodsmoke and mobile sources in
four U.S. cities during wintertime. The apparent relative importance
of the sources depends on the the location of sampling sites within
each city and season, and are not necessarily representative of annual
city-wide ratios. Non-Roanoke results are from Ref. 3.
Woodsmoke
Roanoke VA (1988-89)
n = 37 - 40
EOM Conc'n"
R58+ Cone ' nb
R9S+ Potency"
Rjg. Conc'nb
R,8. Potency0
Boise ID (1986-87)"
n = 40
EOM Conc'n*
Rgg+ Conc'nb
R98+ Potency"
d
8.1 ± 0.6
5.9 ± 1.1
0.80 + .16
4.5 ± 0.9
0.53 ± .11
14 ± 2
12 ± 3
0.84 ± .25
Mobile Intercept
Sources
4.0
10
3.0
5.8
1.7
6
18
3.0
± 0.5 -.3 ± .7
± 1 2.6 ± 1.1
±0.4
± 0.7 4.2 ± 0.9
±0.3
±2 2 + 2
±3 3+4
±1.1
Meas.
Total
12
19
15
22
32
Albuquerque NM (1984-85)"
n = 44
EOM Conc'n'
R98+. Cone ' nb
R,8+ Potency0
Raleigh NC (1984-85)
n = 40
EOM Conc'n"
R,8+ Conc'nb
R98+ Potency0
15 + 1
19 ± 2
1.3 ± .2
C
16 ± .5
12 ± 1
0.78 ± .07
3
11
3.7
1
4
3.7
±1 1 ± 1
+ 3 3 ± 3
+ 1.5
+ .3 0.1 ± 1
+ 1 1 ± 1
±1.5
19
32
17
18
_—
" Mg / m3
b revertants / m3
0 revertants / ^g EOM
6 residential and roadway site
e residential site only
212
-------
COMPARISON OF RESIDENTIAL OIL FURNACE AND WOODSTOVE EMISSIONS
By
Robert C. McCrillis
Air and Energy Engineering Research Laboratory
Randall R. Watts
Health Effects Research Laboratory
and
Roy B. Zweidinger
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
DISCLAIMER
The work described in this paper was funded by the U.S. Environmental
Protection Agency (EPA) . This paper has been reviewed by EPA, and approved for
publication. Approval does not signify that the contents necessarily reflect the
Agency's policy and views, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
ABSTRACT
This paper compares the emissions from residential oil furnaces and
residential woodstoves. The comparison is based on a number of analyses that were
performed on the emissions, including total mass, filterable particulate, total
extractable organics, and mutagenicity. The emission samples were gathered as part
of the Integrated Air Cancer Project and included the field testing of oil furnaces
in Roanoke, VA, during the 1988-S9 winter; field testing of woodstoves in Boise, ID,
during the 1986-87 winter and laboratory tests to support both field studies. The
results show that woodstoves produce higher emissions than even poorly tuned oil
furnaces. Mutagenicity of emissions from both types of heating systems is about the
same on a revertants per microgram basis but, because of their much higher emission
rate, woodstoves have a higher mutagenic potential on a per hour of operation basis
and per unit heat input in the fuel. Although there are 2-3 times more oil furnaces
in use in the U.S. than woodstoves, of the two, woodstoves still have a higher
mutagenic potential.
INTRODUCTION
Consumption of wood as a home heating fuel declined markedly during the early
decades of the 20th century as it was displaced by coal which was then displaced by
fuel oil. In the last 20 years, fuel oil has itself been displaced in many homes by
natural gas, so that more than half of all homes are heated with natural gas.
Electrical heating ranks second, followed by fuel oil and lastly wood. Wood
consumption increased markedly during the 1970s with the sharp increase in fossil
fuel prices. Although some wood burners have tired of using wood there continues to
be a market for about 200,000 new stoves each year. In addition, fireplaces
continue to have widespread appeal; approximately 60% of all new homes built today
have a fireplace. Adding a fireplace is one of the most cost effective renovations
one can make to a home.
This paper presents results of continuing work performed under the aegis of
EPA's Integrated Air Cancer Project (IACP) to characterize and compare air emissions
from combustion of wood and fuel oil for residential heating. This paper emphasizes
the analyses of stack emission samples collected from oil furnaces in Roanoke, VA,
during the lACP's winter 1988-89 field study and supporting oil furnace and
woodstove laboratory projects.
EXPERIMENTAL APPROACH
213
-------
Roanoke, VA, Oil Furnaces
Ten homes with oil furnaces in Roanoke, VA, were selected for source testing.
Although a cross section of furnace technologies was desirable, primary emphasis was
on pairing each home with an oil furnace (and not using wood heat) to a non-oil non-
wood heated home in the same neighborhood. Furnace performance was measured using
standard procedures employed by furnace repair firms.
Laboratory Oil Furnace
A single Williamson residential oil furnace was tested in the laboratory,
first with an old technology (pre-1970) atomizing gun burner and then with a modern
retention head burner. The furnace and burners have been described previously1
Laboratory Woodstoves
Two woodstoves were tested at EPA's woodstove test facility at the
Environmental Research Center, Research Triangle Park, NC. One was a conventional,
uncontrolled Lopi model 330/440, and the other was a low emission noncatalytic Lopi
model Answer. The latter stove was certified to EPA's Phase 1 emission standard.
Emission Sampling Equipment
The basic device used for air emission sampling was the Source Dilution
Sampler (SDS) which has been described previously2'3. For the experiments reported
here the basic sampler was modified to include:
• A cyclone located just before the SDS filter designed to remove particles
>2.5 jam aerodynamic equivalent diameter
• Two fine particle samplers in parallel which extract a 4 1pm slip stream
of the diluted stack sample just ahead of the SDS filter. They are
identical except that one uses a Teflon0 filter and the other a quartz
filter for elemental/organic carbon and trace metals, respectively. Both
samplers have a cyclone ahead of the filter to remove particles >2.5 (Im.
• A sampler which extracted a 0.15 1pm sample of the diluted gas just ahead
of the SDS filter for aldehyde analysis. The sample was collected in
cartridges containing dinitrophenylhydrazine (DNPH) coated silica gel.
• An evacuated canister which extracted a 0.5 1pm slip stream between the
SDS filter and XAD-2 cartridge for hydrocarbon analysis.
A block diagram of the sampler with these modifications is shown in Figure 1.
In Roanoke, sampling was conducted at each home for 4 days starting Saturday
at 7 am and ending Wednesday at 7 am. The 4 days were further divided into eight
12-hour sampling periods representing daytime (7 am 7 pm) and nighttime (7 pm - 7
am). In some cases, results are presented as weekday daytime, weekday nighttime,
weekend daytime, and weekend nighttime averages.
RESULTS
Roanoke, VA, Oil Furnaces
The furnace model and burner information, stack carbon dioxide (C02) and
Bacharach smoke number taken before the barometric damper are shown in Table I. No
adjustments were made to the furnaces; however, homeowners were advised if the
furnace appeared to need adjustment.
Chimney exit total extractable organic emissions for each Roanoke home are
presented in Table II. In Table II the results have been aggregated into weekday_
daytime and nighttime and weekend daytime and nighttime averages. Also included is
the average percent of time the furnace was on (burner running) for each aggregated
sample period. The emission results are calculated on the basis of the total sample
period, not just the time the burner was on.
214
-------
exhaust-
XAD-2
cartridge
} filter
dilution
chamber
dilutic
air in
variable
speed
blower
stralghtener
.probe
sample in
Figure 1. Block diagram of source dilution sampler (SDS) .
Laboratory Oil Furnaces
The majority of the laboratory oil furnace test results have been published
previously1 There was a general correlation between smoke number and emission
rate. Even under best tuned conditions the old style burner's emissions were 2-4
times higher than those of the new technology burner. The old burner's emissions
consisted of about equal parts of organic and elemental carbon whereas the new
burner's emissions consisted of 20 parts organic carbon to 1 part elemental carbon.
Bioassay results using the microsuspension assay4 showed that potency ranged from 7
to 18 revertants/microgram (rev/(lg) of organic for both burners. Because the
retention head burner emitted substantially less extractable organics per hour, its
mutagenic emission rate was 0.006 rev/J heat input compared to 0.02 rev/J for the
old style burner.
Laboratory Woodstoves
Laboratory and field woodstove test results have been widely
215
-------
Table I. Summary of furnaces tested in Roanoke, VA, IACP field study.
foci nozzle
Lenr
J31-15. 40 yrs old
RIO Homart furnace and burner, >30 yrs ol
R13 Mueller Climatrol model 227-110, burr
R16 ARCO Flame model Al-3, no burner model
No.
R19 Kevanee model VT-S10, Scries 2X. burner
8.3 9.8
7.5 11.0
B.3 9.6
2.5 4.2
fuel nozzle, 20 yra old
2.5 L/hr fuel nozzle
a One inch of water pressure = 249 Pa
b Insufficient data to calculate efficiency
reported5'6'7 . The
current work can best be
compared to a laboratory
study7 supporting the
IACP field study in
Boise, ID . The combined
Boise and Roanoke
laboratory woodstove
results for total SDS
train catch are shown in
Figure 2. All of the
conventional stove data
(with the exception of
one data point) follow
the expected trend -
higher emissions as
burnrate decreases below
2-2.5 kg/hr. Both of
the low emission stoves
produced lower emissions
compared to the
conventional stove .
They also exhibited a narrower burnrate range. Emission of aldehydes (available for
Roanoke woodstoves only) showed a trend of increasing emission rate with increasing
burnrate. The range was from 0.1 to 0.5 g/hr total aldehydes.. Formaldehyde was the
largest fraction (37%) followed by acetaldehyde (20%) .
60
^ 50
-------
Table II.
Total extractable organics in samples from Roanoke, VA, oil furnaces.
Residence
R01
R04
R07
RIO
R13
R16
R19
R22
R25
R28
Emission rate, g/hr (burner on time, %)
weekend day weekend night weekday day weekd;
1.
4.
0.
0,
1.
1,
1,
1,
4,
6.
90
63
,85"
.85"
.20°
.97"
. 71C
.13
.17
.01
(17.
(19.
(46.
(46.
(47.
(35.
(84.
(24
(35
(72
.3)
.8)
.7)
.7)
.8)
.6)
.3)
.3)
.7)
.6)
1.37
2.66
1.02"
1.28"
0.99"
1.83
1.22
1.17
3.92
5.11
(17.
(29.
(46.
(46.
(34.
(38.
(44.
(24.
(13.
(79.
.3)
.7)
.7)
.7)
.1)
.8)
.4)
.3)
.0)
.4)
no
3.
1.
9.
1.
1.
1.
0.
3.
2.
sample*
45
02"
05"
54
94
81
69
30
44
(18.
(46.
(46.
(40.
(47.
(77.
(24.
(22.
(54.
6)
7)
7)
5)
2)
9)
3)
8)
8)
ay night
no sample3
5.67
1.79"
1.04"
1.47
6.21
3.42
1.13
3.83
4.00
(43.
(46.
(46.
(48.
(53.
(57.
(24.
(6.
(83.
6)
7)
7)
3)
4)
6)
3)
8)
8)
a Samples lost
b Filter samples lost
c XAD-2 total chromatographable organics sample lost
potency with increasing
burnrate, especially when
burning pine. This trend
has been noted previously5
Similarly, a trend of
increasing polycyclic
aromatic hydrocarbons (PAH)
emission rate with burnrate
was seen again which
correlates well with the
mutagenic potential trend if
it is assumed that most of
the mutagenicity is caused
by the PAHs.
Comparison of Woodstove and
Oil Furnace Emissions
?14
a 12
L
, 10
I:
.,.»gp.
1.5 2 2.5
Burnrnte., kg/hr
3.5
Comparing total SDS
emissions in Figure 2 and Figure 3. Mutagenic potency.
Table II, one can see that
the conventional woodstoves
emission rate is 5-20 times higher than the emission rate for oil furnaces.
Emission rates from low emission woodstoves are 2-5 times higher than for oil
furnaces. The dominant component of oil furnace organic emissions is unburned fuel
oil, whereas wood smoke consists of a very broad mixture of organic compounds. For
this comparison the woodstove data in Figure 2 were reduced by 20% to arrive at an
approximation of the extractable organic fraction.
A comparison of overall average woodstove and oil furnace emission mutagenic
factors1 (microsuspension assay) yielded:
conventional woodstove average
low emission woodstove average
0.6 rev/J fuel energy input
0.1 rev/J fuel energy input
217
-------
old technology oil burner 0.02 rev/J fuel energy input
new tecnology oil burner = 0.006 rev/J fuel energy input
SUMMARY AND CONCLUSIONS
Total SDS emission rates from woodstoves vary from as low as 5 g/hr for low
emission stoves to as high as 50 g/hr for a conventional stove. Emission rates from
Roanoke residential oil furnaces tested with the same sampling train ranged from
0.85 to 6 g/hr.
Aldehyde emissions from woodstoves ranged from 0.1 to 0.5 g/hr, increasing
with increasing burnrate.
On the basis of equal heating value, the conventional woodstove mutagenic
emission factor is about 30 times higher than the one for old technology oil
furnaces and about 100 times more mutagenic than the one for new technology
retention head oil burners. The mutagenic emission factor for new, low emission
woodstoves is about 5 times higher than the one for old technology oil burners and
about 20 times higher than the factor for new technology retention head oil burners.
Further analyses are underway to determine the carcinogenicity of oil furnace
emissions.
ACKNOWLEDGEMENTS
R.K. Stevens, C.W. Lewis, and J.O. Baugh, all of the Atmospheric Research and
Exposure Assessment Laboratory, specified the design of the source signature
samplers, assisted during their initial setup, and provided the reduced data frorr1
the samples collected. D.F. Natschke, Acurex Environmental Corp., oversaw the
laboratory studies, providing sample recovery and analysis of samples from the SDS
and final reports from each test series.
REFERENCES
1. Steiber, R.S. and R.C. McCrillis, "Comparison of Emissions and Organic
Fingerprints from Combustion of Oil and Wood," in Proceedings- of the 84th
Annual Meeting of the AWMA, Vancouver, Canada, 1991, Paper No. 91-136.2.
2. Williamson, A.D., R.S. Martin, and D.B. Harris, "Measurement of
Condensable Vapor Contribution to PM10 Emissions," in Proceedings of the
78th Annual Meeting; of the AWMA, Detroit, MI, 1985, Paper No. 85-14.4.
3. Merrill, R.G. and D.B. Harris, "Field and Laboratory Evaluation of a
Woodstove Dilution Sampling System," in Proceedings of the 80th Annual
Meeting of the AWMA, New York, 1987, Paper No. 87-64.7.
4. DeMarini, D.M., M.M. Dallas, and J. Lewtas, "Cytotoxicity and Effect of
Mutagenicity of Buffers in a Microsuspension Assay," Teratoqen.,
Carcinogen, and Mutacren. 9:287 (1989) .
5. McCrillis, R.C., R.R. Watts, and S.H. Warren, "Effects of Operating
Variables on PAH Emissions and Mutagenicity of Emissions from
Woodstoves," J. Air Waste Manage. Assoc. 42:691-694, 1992.
6. McCrillis, R.C. and D.R. Jaasma, "Woodstove Emission Measurement Methods
Comparison and Emission Factors Update," Environmental Monitoring and
Assessment, 24:1, pp. 1-12, 1993.
7. Steiber, R.S., R.C. McCrillis, J.A. Dorsey, and R.G. Merrill,
"Characterization of Condensible and Semivolatile Organic Materials from
Boise Woodstove Samples," in Proceedings of the 85th Annual Meting of
AWMA, Kansas City, MO, 1992, Paper No.92-118.03.
218
-------
Session 6
Indoor Air Quality in
Highly Confined Environments
-------
STRATEGIES USED TO MANAGE AIR QUALITY IN MANNED SPACECRAFT
John T. James
NASA Johnson Space Center - SD4
Houston, TX 77058
ABSTRACT
The quality of respirable ah" hi crew compartments has been a concern since the earliest
days of manned space flight. Air contaminants are controlled by limiting the offgassing of
flight hardware and by removing contaminants using high-capacity air revitalization systems.
Air quality has been monitored retrospectively by ground-based analysis of samples obtained
during flights; however, on-board instruments are under development to measure
contaminants during flight. This requires that spacecraft maximum allowable concentrations
(SMACs) be established for brief and long-term exposures so that appropriate actions can be
taken promptly when contaminants exceed those limits.
At present, SMACs are being set for chemical exposures ranging from 1 hour to 180 days
using a scientific approach developed in cooperation with the National Research Council.
Original toxicity studies are reviewed for quality and completeness and all toxic effects
induced by a chemical are analyzed to determine the most sensitive effect. Human inhalation
data are preferred, but often extrapolations must be made from animal data, from
noninhalation data, or by structure-activity analysis. In addition, the biological changes
caused by space flight, including loss of red cell mass, changes in immune function, and
cardiac dysrhythmias, are factored into the analysis. The approach is illustrated by describing
the results for indole, Freon 113, and octamethyltrisiloxane (OMTS).
INTRODUCTION
Air that is safe to breathe is the most immediate environmental need of space crews.
NASA has learned several important lessons about managing air quality during its experience
with manned space exploration. Chemically induced illness in test subjects during a closed
environmental test led to strict controls on the chemicals used to clean interiors and the
products permitted to offgas into the closed environment1. Review of the cause of the tragic
ground fire during the testing of Apollo 1 resulted in stricter controls on flammability of
materials and reduction of the oxygen percentage of the atmosphere in spacecraft2. The value
of redundancy was underscored during the flight of Apollo 13 when explosion of an external
oxygen tank resulted in the potential for accumulation of life-threatening concentrations of
carbon dioxide. Only an unplanned "jury rig" of the Command Module's carbon dioxide
removal system enabled the crew to maintain a safe atmosphere in the lunar module, where
they had taken refuge, until an Earth landing could be executed2. The historic Apollo-Soyuz
mission nearly ended in tragedy for the American crew because of entry of toxic propellants
(nitrogen dioxide and possibly methylhydrazine) into the cabin during descent through the
Earth's atmosphere3. The design of modern spacecraft and operational procedures after
extravehicular activity (EVA) minimize the risk of crew exposure to propellants. The most
serious air-quality risk in modern spacecraft is the potential for thermodegradation of
polymers associated with electrical hardware. The most significant of these occurred aboard
the Shuttle in June 1991 when a refrigerator motor overheated and caused the release of
formaldehyde and ammonia into the crew compartment. More thermal-overload protection
has been placed on electrical motors since this incident.
221
-------
The Russian experience with air-quality problems is not as well known to us as the NASA
experience; however, there are several interesting references to problems they have
encountered4. During a 150-day mission on Salyut 7 (an early space station), a problem in the
air revitalization system was believed to have generated "acrid odors" that nearly exceeded the
crew's ability to tolerate them. During 140 and 175-day missions cosmonauts were reported to
have developed sensitivity to formaldehyde. The authors concluded that formaldehyde
concentrations in cabin atmospheres needed to be reduced.
The incidents listed above illustrate that air contaminants can originate from a variety of
sources and that unexpected releases must be anticipated. Chemical contamination routinely
enters the spacecraft air from hardware offgassing, use of utility chemicals, and metabolism of
the crew. As a result of accidents, chemicals can originate from payload experiments, fluid
systems, thermodegradation of electrical devices, anomalies in the air revitalization system,
and entry of chemicals that are normally present outside the spacecraft (propellants). In the
past, strategies to manage these risks have depended heavily on Earth-based resources;
however, as missions lengthen and become more remote from Earth, the resources to manage
air quality problems must be available on the spacecraft. A major part of this strategy is on-
board monitoring of air contaminants and the setting of contaminant limits that address both
short-and long-term exposures. The overall strategy of air quality management will be
described first, then the focus will be shifted to methods available to set contaminant limits for
spacecraft air.
STRATEGY FOR CONTROL OF AIR CONTAMINANTS
NASA uses five steps to manage air quality in spacecraft; the steps limit known risks to
air quality, but also recognize the need to be prepared for unexpected releases.
Offgas Testing of Hardware and Materials
The risk of air contamination is severely limited by offgas testing of all nonmetallic
hardware and materials that constitute the cabin interior. Test items are warmed in a sealed
chamber for 72 hours at 120 °F and all volatile contaminants released into the test chamber are
analyzed by gas chromatography and mass spectrometry (GC/MS). No item may release
contaminants that collectively exceed a total toxicity value of 0.5. Total toxicity value (T value)
is defined as the sum of the ratio of each contaminant concentration (indexed to the spacecraft
volume in which it will be used) to its 7-day SMAC. Space modules are also assessed by
mathematically summing the individually-measured products from all hardware in that
module and by measuring contaminants released inside the assembled module before launch.
Typically, the mathematical assessment of a module gives T values an order of magnitude
greater than the T values measured in the module before launch.
Containment of Hazardous Chemicals
NASA classifies the toxic hazard from a chemical into one of three categories. The lowest
category is called a nonhazard and would allow no more than mild, transient eye irritation
that does not require therapy. A critical hazard is one that could cause up to moderate eye
irritation, can last longer than 30 minutes, and may require therapy, but is unlikely to cause
lasting eye damage. Finally, a catastrophic hazard is one that poses a risk of permanent eye
damage, systemic toxicity that impairs performance of the mission, or delayed serious injury.
NASA requires that experiments or systems that pose a critical hazard must have single-fault-
tolerant containment and those that pose a catastrophic hazard must have double-fault-
tolerant containment. Building fault tolerance into hardware can be expensive and consumes
valuable weight and volume resources. Hence, the lexicological assessments must not be
overly conservative.
222
-------
Air Revitalization Systems
Spacecraft air purification systems typically contain elements to replenish oxygen and
remove carbon dioxide, carbon monoxide, trace organic contaminants, and particles. These
systems are designed to handle normal loads on the spacecraft air, with adequate overdesign
margins. However, they are not designed to deal with accidental large releases of
contaminants into the atmosphere. Strategies such as coupling the odor/bacteria filter to a
hand-held vacuum or turning the condensing heat exchanger to full cool have been devised to
decontaminate the Shuttle in the event of a serious contamination problem. The last-resort
solution is to depressurize the contaminated module and repressurize it with clean air as was
done during Skylab.
Monitoring Airborne Contaminants
Until recently, methods that met the severe weight and power constraints of space flight
were not available to measure volatile contaminants on board spacecraft. Traditionally,
contaminants had been trapped in evacuated steel cylinders or on adsorbent resins for
identification and quantitation by GC/MS after the mission. Clearly, this approach is not
useful during a mission. Because of concerns over thermodegradation of electronic devices, a
combustion-products analyzer has been flown aboard the Shuttle since 1990. The device,
which weighs approximately 1 kg, uses electrochemical sensors to measure selected toxic
products of combustion. For high-risk EVA missions, a hydrazines monitor has been available
in the air lock since 1991 to monitor selected propellants at ppb levels using ion mobility
spectrometry (IMS). Instruments under development for Space Station Freedom include an
IMS for rapid, nonspecific detection of volatile hydrocarbons and a GC/IMS device for
periodic monitoring of 20 to 30 targeted chemicals that are often detected in spacecraft air. For
the first time, short-term limits for contaminant exposure will be needed to interpret the
monitoring results.
Setting Airborne Exposure Limits
New SMACs were needed to provide monitoring goals for on board instruments, to deal
with accidents where crew exposures last several hours, and to set limits for long-term
exposures expected on the Station. The existing 7-day SMACs had been set many years ago
without detailed documentation. Much has been learned recently about sett~ \g chemical
exposure limits and more is known about the adverse effects of space flight on humans. The
approaches described below have been developed in cooperation with the National Research
Council Committee on Toxicology.
METHODS USED TO SET SPACECRAFT CONTAMINANT LIMITS
Goals of Short and Long-Term Limits
Setting specific limits in any situation requires that goals of the limits be clearly
delineated. Short-term SMACs (1 and 24 h) are set for rare, accidental releases of chemicals
into the air. Mild transient effects that do not limit a crewmember's performance are allowed.
For example, mild headaches or eye irritation are acceptable; mild visual impairment would
not be allowed. Long-term limits are set to protect the crew from any adverse effect to a
reasonable level of risk. For example, the acceptable level of risk for a carcinogen was taken as
the lower 95% confidence limit of a concentration that would impart a 0.01 % excess risk of
getting cancer as calculated using the linearized multistage model. In setting limits, the goal
was to rely on high quality data, preferably from human exposure studies; however, animal
data, noninhalation data, and structure-activity analyses often must be used to derive
defensible values.
223
-------
The guidelines described below should be viewed as a template for setting exposure
limits. Databases for each chemical differ and it is often possible to argue scientifically from a
database that certain factors or methods are not appropriate. The guideline is meant to be
used when the database on a chemical is limited and no logical basis exists for selecting
alternate factors or approaches.
Threshold-Type Effects vs. Cumulative-Type Effects
Once original references have been obtained via a thorough literature search, each toxic
effect induced by a chemical is reviewed and an acceptable concentration (AC) set for that
specific effect and time of exposure. If an effect does not occur below some threshold
concentration (e.g., central nervous system depression), then the ACs for that effect may be
independent of time of exposure. On the other hand, many effects (e.g., liver damage) become
more severe with prolonged exposure, so the AC will decrease with increasing exposure time.
Because chemicals can induce both threshold and cumulative effects, the distinction between
the two becomes uncertain. For example, a chemical may cause a cumulative effect at high
concentrations, but below some threshold concentration an adverse effect is not induced no
matter how long the exposure.
Application of Human Data
Human data come from controlled exposures in a clinical environment, accidental
industrial exposures, and long-term relatively low concentration exposures. Typically, the
former are useful for setting short-term SMACs if the endpoints measured were relevant and
there were 4 or more test subjects. Such studies are used to derive a no observed adverse effect
level (NOAEL); because only a sample (n) of the population has been tested, the NOAEL is
divided by ^/~n/10. For effects that are tolerated for short times (e.g., headache) a
semiquantitative dose response curve may be useful for estimating the mild-effect level.
Accidental exposures and epidemiology data are seldom used because exposure
concentrations often are unknown, endpoints are not carefully observed, or confounding
factors obscure relationships between variables. Epidemiology data can be used to check
predictions made from other data.
Estimates from Animal Inhalation Data
Animal data must be reviewed with several reservations in mind. The first questions are
whether the effects observed are adverse (or adaptive), and whether the animal model is
appropriate to humans. For example, the hydrocarbon nephropathy induced in male rats
seems to be irrelevant to humans, and hence should not be used for human risk assessment5.
A change in organ weight in the absence of histopathological changes is considered adaptive.
Adverse effects that do not follow a plausible dose-response relationship should be viewed
with suspicion. Finally, some endpoints (e.g., narcosis) are gross indicators of serious effects
and should be used with caution when predicting a NOAEL in humans. Usually, a NOAEL is
estimated from animal data; however, under some conditions (when a dose response curve is
well established), the "bench mark" approach has proven useful6.
The factor typically used in extrapolating an animal NOAEL to a human NOAEL is ten.
The extrapolation is intended to account for "worst case" differences in metabolism and target
tissue susceptibility between the model and humans. The extrapolation is done from the most
sensitive species, although nonhuman primates can be more suitable models than a sensitive
rodent species. However, if metabolic data show that the species-extrapolation factor should
be less than ten, a smaller number is used in the estimate.
224
-------
Noninhalation Data
Inhalation data often are not available for a specific chemical. If toxicity data are available
from other routes of administration, then they are used; when no toxicity data are available,
structure-activity analysis is used. For noninhalation data, assumptions are necessary to relate
the comparative absorptions of the chemical via inhalation and via the available route.
Caution must be exercised if the chemical could induce local effects on the lung rather than
systemic effects that would be evident from the noninhalation data. When no toxicity data are
available on the specific chemical, structurally related chemicals are reviewed. Uncertainty
factors applied to this analysis are based on the structural similarity of the chemicals involved,
the consistency of the toxic properties of the chemicals, and the number of chemicals for which
data are available.
Space-flight Induced Changes
Some of the physiological changes that occur during space flight could leave astronauts
more susceptible to certain toxicants. A safety factor of five has been determined for cardiac
arrhythmogenic chemicals, and a factor of three for hematotoxicants and immunotoxicants.
Cardiac arrhythmias are not common during space flight, but because of their potential
seriousness, a relatively large factor was selected. After a few days in space, an astronaut's red
cell mass decreases by approximately 10%; certain immune functions appear to decrease as
well. Hence, a factor of three has been applied to hematotoxicants and immunotoxicants to
estimate safe ACs for astronauts.
RESULTS FOR SELECTED CHEMICALS
Indole
Indole has never been detected in spacecraft air because of analytical limitations;
however, it is undoubtedly present from metabolic sources. Short exposures cause nausea
(human data) and prolonged exposures cause hematological effects and possibly leukemia
(animal data). The new short-term SMACs were based on avoiding significant nausea and the
longer-term SMACs were based on avoiding hematological effects which had been observed in
mice, rats and monkeys exposed continuously for 90 days. A lower limit of 0.05 ppm on the
long-term SMACs was set because this exposure would increase the normal load of indole
from the gut by only 5% and indole causes only systemic toxicity (i.e.., inhaled indole would
not cause selective lung damage). The leukemogenic effects of indole were demonstrated by
injection only and the data were not considered sufficient to develop a human risk estimate by
the inhalation route,
Freon 113 (l,l,2-trichloro-l,2,2-tiifluoroethane)
Freon 113 is commonly used to clean electronic hardware before launch and is invariably
found in samples of spacecraft air. Like most Freons it is not regarded as very toxic; however,
experiments with dogs have demonstrated that Freon 113 can cause cardiac sensitization to
arrhythmogenic chemicals. Applying a species extrapolation factor of 10 and a space flight
factor of 5 to the 2500 ppm NO AFX observed in dogs gave a SMAC of 50 ppm. This was
considered a threshold-type effect; therefore, the value was applied to all exposure times.
Octamethyltrisiloxane (OMTS)
Siloxanes are common offgas products found in spacecraft air samples. A thorough
literature search revealed no toxicity data for OMTS; however, data were available on
hexamethyldisiloxane (HMDS) and dodecamethylpentasiloxane (DMPS). Comparison of
HMDS and DMPS data by routes other than inhalation showed that HMDS was the more
toxic. An inhalation study of HMDS in rats and guinea pigs showed that 4400 ppm (15 to 20
225
-------
7 h exposures) caused no more than 10% decreases in liver and kidney weights without
histopathological changes. The 4400 ppm NOAEL in rodents, after exposure time and species
extrapolation adjustment, was used to calculate a NOAEL for HMDS. This NOAEL was
reduced by a structure-activity factor of 3 to estimate a NOAEL for OMTS, the compound in
question. The magnitude of this factor was judged from the fact that data were available on
chemicals that structurally "bracketed" OMTS, but inhalation data were not available for one of
them.
CONCLUSIONS
Setting safe contaminant limits is an integral part of the strategy used to manage
spacecraft air quality. A consistent framework has been established to interpret human and
animal toxicity studies and to derive exposure limits ranging from 1 hour to 180 days. Where
appropriate, the potential for increased susceptibility from space-flight-induced physiological
changes has been incorporated into the analysis. Structure-activity analysis has been used
when no specific toxicity data are available on the chemical of interest.
REFERENCES
1. R.A. Saunders, "A new hazard in closed environmental atmospheres," Arch. Environ.
Health 14:380 (1967).
2. J.C Brady et al., Biomedical Results of Apollo (NASA SP 368); R. Johnston, L. Dietlein, C.
Berry, Eds., U.S. Government Printing Office, Washington, D.C., 1975.
3. A.E. Nicogossian et al., The Apollo-Soyuz Test Project-Medical Report (NASA SP 411); A.
Nicogossian Ed., National Technical Information Service, Washington, D.C., 1977, pp 11-28.
4. B.J. Bluth and M. Helpple, Soviet Space Stations as Analogs (NASA Grant NAGW-659);
Reston, VA., 1986, pp 1-90 and 1-137.
5. J.A. Swenberg et al., "The comparative pathobiology of a2u globulin in nephropathy,"
Toxicol. Appl. Fharmacol. 97:35 (1989).
6. K.S. Crump, "A new method for determining allowable daily intakes," Fund. Appl.
Toxicol. 4: 854 (1984).
226
-------
AIR QUALITY MONITORING IN SPACECRAFT: PRESENT AND FUTURE
Thomas Limero and Hector Leano
KRUG Life Science
1290 Hercules Drive, Suite 120
Houston, Texas 77058
J.T. James
NASA - Johnson Space Center
Biomedical Operations and Research Branch
Mail Code SD4
Houston, Texas 77058
ABSTRACT
The quality of the internal atmosphere of the Space Shuttles is maintained by removal of carbon dioxide, carbon
monoxide, and some volatile organic compounds (the latter for odor control), and the replenishment of oxygen and
nitrogen as necessary from storage tanks. The National Aeronautics and Space Administration (NASA) has long
recognized the importance of monitoring spacecraft atmospheres to protect crew health and safety in these partially
closed systems. Until recently, archived samples collected during flight and returned to Earth for subsequent analysis
had been the only means of determining the quality of the Shuttle's atmosphere. Two real-time monitors have been
developed within the past three years for use on the Shuttle, and two additional instruments are being built for Space
Station Freedom (SSF).
This paper will present the methods used by the Toxicology Laboratory at Johnson Space Center (JSC) to
collect and analyze samples from Shuttle atmospheres. Observed trends in constituents from samples of this
"confined" atmosphere will be presented using data from recent missions. Finally, the "real-time" monitoring
instruments currently used on Shuttle and those being built tq monitor the SSF atmosphere will be described as
well.
INTRODUCTION
The partially closed-loop nature of spacecraft environments dictates the need for assessing and monitoring the
air quality of the internal atmosphere. Toward this end, the National Aeronautics and Space Administration's
(NASA) Toxicology Laboratory at Johnson Space Center (JSC) has been investigating and developing techniques to
sample, analyze, and monitor the contaminants in spacecraft atmospheres. Sources of contamination in spacecraft
can include: nonmetal material, human metabolism, hygiene products, food, utility and payload chemicals, fire
extinguishant and clothing. The spacecraft's environmental control and life support system (ECLSS) is designed to
maintain acceptable air quality in the spacecraft by removing carbon dioxide, carbon monoxide, trace volatile
organics, and excess heat and moisture.
Until recently, in addition to temperature and humidity, only oxygen and carbon dioxide concentrations were
routinely monitored in real time during Shuttle missions. However, air samples of the crew cabin are collected on
all Shuttle missions and returned to the Earth for analysis by the JSC Toxicology Laboratory. Concentrations of
methane, carbon monoxide, hydrogen and trace volatile organic compounds are determined from these archived
samples. Additionally, most of the volatile organic compounds are positively identified by gas
chromatography/mass spectrometry (GC/MS). Instantaneous "grab" samples are collected in evacuated cylinders near
the end of each Shuttle mission; during selected missions (i.e. spacelabs, refurbished Shuttles) daily 24-hour
integrated samples are collected on a sorbent resin.
Temporary degradations in air quality have occurred aboard the Space Shuttle' 2 and must be anticipated during
any long-duration space flight Although archival sampling provides postflight information on the severity of the
incidents, the goal of the JSC Toxicology Laboratory is to develop on-board monitors that can detect such events
early, in "real time", and be available to measure the effectiveness of air decontamination efforts. The major
obstacles to the development of real-time air monitors for Shuttle are the stringent requirements placed upon flight
hardware. An air monitor for Shuttle must be compact, lightweight, microgravity-compatible, impervious to the
vibration and shock of launch, and require minimal Shuttle resources such as power and
crew time.
A prototype combustion products analyzer (CPA), the first real-time monitoring instrument developed by the
JSC Toxicology Laboratory, has flown on each Shuttle mission since October 1990. This instrument was
developed in response to minor thermodegradation incidents^ that had occurred during Shuttle missions. A second
real-time monitor was constructed to detect the presence of hydrazines in the airlock following extravehicular
activities (EVA) since there is a risk of contaminating the airlock with these toxic propellants.
227
-------
Consideration of monitoring requirements for longer missions, with an increased risk of chemical accidents, has
led to the development of a prototype total hydrocarbon analyzer (THA) and a volatile organic analyzer (VGA) for
Space Station Freedom (SSF). Both instruments use the emerging technology of ion mobility spectrometry.
ARCHIVAL SAMPLES
Collection
Grab Samples. Volume and weight restrictions dictate the size of the stainless steel cylinders (350 cc). Before
they are stowed for flight, the cylinders are evacuated and certified clean by back-filling the cylinder vith pure helium
and running a GC/MS blank. The cylinder is then evacuated to at least 10"^ torr before being sent to Kennedy Space
Center (KSC) for stowage on board the spacecraft
A prelaunch air sample is obtained just before the launch of each Shuttle, and at least two air-sample cylinders
are flown on each mission. One cylinder is used to collect a sample of the Shuttle's internal atmosphere on the last
day of the mission. The second cylinder is used if the crew detects an unusual odor or experiences symptoms (e.g.
eye irritation) that may be related to degradation of Shuttle air quality. During Spacelab missions, 3 or 4 extra
cylinders are stowed to collect samples of laboratory air when the crew first enters the spacelab module, and again
during the early, middle, and late phases of the mission. The inflight air sample cylinders are transferred to JSC for
analysis immediately upon the conclusion of the mission.
Integrated Samples. In addition to the air-sample cylinders described above an integrated sample is collected
daily on the first flight of new or refurbished spacecraft and on all spacelab missions. The solid sorbent air sampler
(SSAS) obtains integrated samples by pulsed sampling into a tube containing a solid sorbent. The SSAS consists
of 8 tubes (5 in. x 1/4 in.) joined to an 18-port valve. Tube 8, the designated "transport position", is open to the
atmosphere during all non-sampling times; therefore, it is not available as a sample collection tube. Each tube
contains approximately 400 mg of Tenax™ GC. Prior to flight, each SSAS tube is cleaned by heating it to 250'C
while helium is flowed over the sorbent. The cleanliness of all tubes is verified by GC/MS analyses before the
SSAS is sent to KSC for stowage on the Shuttle. The verification procedure involves flowing helium over the
heated tube (200'C) and collecting the volatile organics that are removed by a liquid nitrogen (LN2) cryotrap before
transfer to the GC column.
During flight use, an air pump with a variable duty cycle permits air samples of a specified size to be pulled
over the SSAS tube during a 24-hour period. Typical flow rates allow 1.0 to 1.5 liters of Shuttle air to be collected
per sampling tube. At the conclusion of the 24-hour sampling period, a rotary selection valve permits the
crewmember to switch to the next tube to collect another 24-hour integrated sample. This procedure is repeated each
day of the mission until the last flight day, when the rotary selection valve is moved to the tube "8" position, where
it remains until analyses are performed in the JSC Toxicology Laboratory. A unique feature of the SSAS is that
tube contamination is minimized since the tube is never exposed to the ambient environment except during
sampling.
Analysis
Grab Samples. Air-sample cylinders are generally returned to the JSC Toxicology Laboratory within one or two
days of landing. Sample cylinders stored aboard spacelabs are slower to return, and weeks may elapse before analyses
can begin. Each cylinder is analyzed immediately by GC (Varian 3700) and GC/MS [HP 5890 GC and HP 5970
mass selective detector (MSD)]. Samples are pulled through a gas sampling valve into a GC sample loop by a
vacuum pump. Differential pressures are used to determine GC sample size. A Valco helium ionization detector
(HID) is used to ascertain the concentrations of carbon monoxide and methane. A second GC (Varian) with a
thermal conductivity detector (TCD) determines the hydrogen concentration in the sample. Screening runs on GCs
with flame ionization and electron capture detectors are used to determine the GC/MS sample size.
For GC/MS analysis of organic vapors, the sample cylinder is heated to 50°C for 30 minutes before a sample
volume (determined by the GC runs) is extracted into a (LN2) cryotrap. A 50 m x 0.25mm (i.d.) capillary column
with a 1 |i film of carbowax is used to separate the multicomponcnt samples. The GC column is held at an initial
temperature of 35'C for 10 minutes, then the GC oven is ramped to 175'C at 2'C/min. The MSD detects ions over
a 25-350 dalton mass range; the lower limit permitting the identification and detection of methanol and acetaldehyde.
Integrated Samples. Integrated SSAS samples are analyzed with the same GC/MS method described above.
Each SSAS tube is individually heated to 200"C upstream from the LN2 cryotrap. Helium gas flowing at
approximately 10 cc/min is allowed to sweep the sample out of the heated tube and into the cryotrap. The cryotrap is
then heated to transfer the volatile organics compounds to the GC. The GC/MS conditions are identical to those
228
-------
used for the sample cylinders. A volume correction factor is applied to the final calculated concentrations of some
compounds to account for their relatively low retention volumes in Tenax™ GC.
Mission Results
Typically isopropanol, which used as a cleaning agent before launch, is the only compound found in preflight
samples; this result is testimony to the effectiveness of the prelaunch purge system used to remove the offgassed
organic vapors from the cabin. Results obtained from analyses of samples from the STS-26 through STS-52
missions are shown in Table 1. The many compounds detected (typically >80 each SSAS tube) are grouped into
chemical classes for clarity. Each class is divided further into the total number of times a member of a compound
class (can be more than one per mission) was seen at a concentration above or below 1 mg/m? from STS 26 through
STS 52. With few exceptions, compounds in concentrations above 1 mg/m^ are the focus of the lexicological
assessments of Shuttle air quality. The concentrations (C) obtained on "n" contaminants from each flight are used to
estimate the aggregate toxicity (T value) of the air by comparison to the respective spacecraft maximum allowable
concentrations (SMAC) as follows:
T = Ci/SMACi + C2/SMAC2 + Cn/SMACn
These mission analyses have never shown an unacceptable aggregate-contaminant concentration (T > 1) in the
Shuttle atmosphere.
Methane and hydrogen (Table 2) serve as qualitative internal standards because the concentrations of these
metabolic byproducts continuously increase during the mission since they are not removed efficiently by the
ECLSS. If these compounds are found in unusually low concentrations or are totally absent from the sample, this
suggests a sample-collection problem (i.e., valve leak). The data in Table 1 show that the contaminant classes of
alcohols, ketones, and halocarbons are present in the highest concentrations in the nominal Shuttle atmosphere. The
alcohols emanate from medical and hygiene supplies and the ketones from crew metabolism. Halocarbon sources
include the fire extinguishant and offgassing of residual compounds from manufacturing or component cleaning
processes.
REAL TIME AIR-QUALITY MONITORS
Archival samples unquestionably provide valuable information on Shuttle air quality; however the absence of
real-time data and the inability to detect some compounds such as formaldehyde, hydrazines, and inorganic acids has
led to the development of real-time Shuttle air quality monitors. There have been five minor thermodegradation
incidents during Shuttle flights. Although the crew did not experience any permanent adverse health effects, these
incidents clearly demonstrate the potential for significant air-quality degradation. The combustion products analyzer
(CPA) was developed in response to the early thermodegradation incidents that occurred aboard Shuttle. A hydrazine
monitor was required to detect the presence of hydrazines in the airlock after an EVA.
Combustion Products Analyzer
The JSC Toxicology Laboratory and Exidyne Instrumentation Technology (EIT) have developed a prototype
CPA5 that can alert the crew to incipient thermodegradation events and provide data to determine the effectiveness of
decontamination efforts that follow such an event. The primary function of the Shuttle CPA is to determine when
the Shuttle atmosphere is safe for the crew to breathe after a serious thermodegradation incident
Description and Methods. The CPA (Figure 1) is a hand-held, battery powered instrument containing four
electrochemical sensors that detect and quantify hydrogen fluoride (HF), hydrogen chloride (HC1), hydrogen cyanide
(HCN) and carbon monoxide (CO). These four compounds have been deemed the most lexicologically significant in
view of the Shuttle materials that are most prone to thermodegradalion. The CPA has been exposed to combustion
atmospheres^ in ground-based testing to demonstrate its abilily lo irack ihe concenlrations of the four targeted
compounds.
Each day of the mission, ihe crew destows the CPA, activates the unit, and records the concentration displayed
for each target compound. The CO sensor exhibits a cross-sensitivity to hydrogen (approximately 10:1) that
accumulates in the spacecraft atmosphere from human metabolic processes. Therefore, recorded CO baseline readings
are used to normalize the CO sensor response in the event of thermodegradation. An improved CPA is being
constructed that contains a hydrogen sensor and microprocessor that automatically compensates the CO sensor
display for the hydrogen cross-sensitivity.
Mission RpsnlK The CPA has flown on every mission since STS 41 (October 1990). Typical baseline values
for the four sensors during the course of the missions are shown in Figure 2. The sensors have remained stable
during the mission with the exception of the CO sensor, which is responding to hydrogen accumulations. A slight
increase observed in the HF sensor baseline has not been associated with a contaminant in the Shuttle atmosphere.
229
-------
Hydrazine Monitor
Description and Methods. The hydrazine fuels used in the propulsion systems of satellites, the Shuttle, and
the proposed SSF are toxic compounds; their SMACs are as low as 2 parts per billion (ppb). A crewmember
returning from an EVA could contaminate the airlock if hydrazines from a leaking satellite or spacecraft engine were
to adhere to the suit. A prototype hydrazines monitor, based on doped-chemistry ion mobility spectrometry, has
been selected because of its sensitivity (<10 ppb) and adaptability to microgravity conditions.
Mission Results. Before flight on the Shuttle, the hydrazine monitor was tested*" at the White Sands Test
Facility (WSTF). The hydrazine monitor was exposed to known concentrations of hydrazines under conditions
similar to those that would be encountered in the Shuttle. This testing clearly demonstrated the ability of the
hydrazine monitor to unambiguously detect hydrazines if present in an atmosphere being monitored.
The hydrazine monitor has been used on two EVA Shuttle missions (STS 37 and STS 49) which posed an
increased risk of hydrazines contamination. After the EVA crews returned to the airlock, the hydrazine monitor was
used to ensure that hydrazine had not been inadvertently introduced into the airlock. The hydrazine monitor detected a
very small quantity of ammonia (
-------
afforded by the CPA and THA. Breadboard VOA and THA are ion mobility spectrometry instruments have been
constructed and are presently being evaluated. The IMS characteristics, sensitivity, compact size and minimum
power requirements, are important advantages for spacecraft as well as field screening and indoor air monitoring
applications. Ion mobility spectrometry is an emerging technology with great potential to serve both the
environmental and space communities.
REFERENCES
1. M. Coleman, STS-3S Toxicology Mission Report Memorandum SD4/91-027, NASA, Johnson Space Center
(February 1991)
2. M. Coleman, STS-40 Toxicology Mission Report. Memorandum SD4/91-362, NASA, Johnson Space Center
(October 1991)
3. M. Coleman, STS-6 Toxicology Mission Report. Memorandum SD4/0703-750-03.12, NASA, Johnson Space
Center (June 1993)
4. M. Coleman, STS-28 Toxicology Mission Report. Memorandum SD4/89-316, NASA, Johnson Space Center
(November 1989)
5. T. Limero, J. James, R. Cramer, and S. Beck," A Combustion Products Analyzer for Contingency Use
During Thermodegradation Events on Spacecraft", SAE 21st International Conference on Environmental Systems,
San Francisco, Ca., July 1991 (#911479).
6. N. Martin and H. Johnson, Test Report: Evaluation of Ion Mobility Spectrometer for the Detection of
Monornethylhyrirarint; Hvflrarine. and Ammonia at Reduced Pressures. TR-739-001, Johnson Space Center, White
Sands Test Facility, Las Cruces, NM 88004
231
-------
Table 1: Chemical Contaminants During Missions STS-26 through STS-52
AVERAGE CONCENTRATION RANGES
Contaminants
Alkanes *
Alkenes
Alcohols
Ketones
Aromatic hydrocarbons
Esters
Aldehydes
Silicone compounds
Halocarbons
Miscellaneous
* Excluding methane
Number of Missions (Concentration in mg/m )
lmg/m3
3
0
46
14
2
0
0
6
22
0
232
-------
Table 2: Hydrogen and Methane Contaminants During Missions STS-26 Through STS-52
AVERAGE CONCENTRATION
Hydrogen
8.3 mg/m3
Methane
46 mg/m3
233
-------
HCN
HC1
Outlet
CO
Inlet
HF
Figure 1: CPA Sensor Block
234
-------
25 47 76 95 125 155
MISSION ELAPSED TIME IN HOURS
171
FIGURE 2. Concentrations Displayed by the CPA
During the STS-56 Mission (April 1993).
235
-------
Shuttle, Solutions
There's a gripe about odors in space.
In this era, when astronauts race
'Exercising with strain
So their muscles sustain
ft.ll their tone, in that weightless small place.
This short mission will take But one week^
Physiological tests: at their peak^
With great physical toil
Tons of sweat, a true spoil
This experiment's really quite sleekj.
Tlease remember...our windows are shut
ftnd the doors are well sealed, in this hut
Seven people, here live
fill our faults, we forgive
'We're a team! There's no room for one nut!
There are problems with odors...come...look^
Meet the captain, the crew and the cook^
ft. solution Til pave
To the problem you gave...
It would help...If a shower you took]
Sector Javier Leano
Toxicology Laboratory
Life Sciences
236
-------
SUBMARINE AIR QUALITY: RELATIONSHIP TO
HUMAN BODY BURDEN
Hugh J. O'Neill, Sydney M. Gordon, Louise M. Brousek
and Demetrios J. Moschandreas
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
Douglas R. Knight and Jeffrey S. Bowman
U.S. Naval Medical Research Laboratory
Groton, Connecticut 06349
ABSTRACT
A preliminary survey of the distribution of the volatile organic components (VOCs)
present in submarine air was performed at various stages of an 82 day patrol. Expired air
from one of the crew members was also surveyed during the patrol. Because the chemical
composition of the samples obtained was extremely complex, the data were presented as
the total GC/MS ion count for each of 10 chemical classes. In the air samples, 249
components were identified; in the breath samples, 241. The relationship of the air
composition to the breath samples (body burden) is presented.
INTRODUCTION
The submarine represents a unique environment in which individuals may be confined
for periods of up to 90 days. Thus, any VOC emissions arising from sources as the off-
gassing of equipment components, smoking, cooking, combustion products of fires, and
medical supplies must be controlled. Review articles1'2 summarizes previously published
information on VOCs in submarine air; other articles3-4 focus on atmospheric monitors that
ensure air quality. The work presented here is an extension of a preliminary study5 directed
toward characterizing, in detail, submarine air and assessing the human body burden
resulting from exposure to this environment.
EXPERIMENTAL
The experiments described were conducted at selected intervals during, and upon
completion of, an 82 day patrol. One crew member, a non-smoker, provided samples of the
expired breath before, during, and after patrol. All collections were made while he was in
a fasting, resting condition. The pre-patrol sample (baseline) was collected before exposure
to the submarine atmosphere and the post-patrol sample was collected in a field laboratory
adjacent to the submarine.
Air samples were collected by drawing 20 1 of compartment air at a rate of 300
cc/min through glass cartridges, Vi inch O.D. and 7 inch long, packed with 600 mg of 60/80
mesh Tenax GC. Upon completion of the sampling period, the cartridges were capped with
Teflon plugs, placed in a protective container, and refrigerated.
237
-------
Breath samples were collected by having the subject inhale zero grade air, purified by
a cooled (0°C) charcoal trap and particle filter, and exhaling into a 40 1 Teflon bag. The
subject inhaled through a two-way Rudolph valve and exhaled through a Teflon transfer line.
The contents of the 40 1 Teflon bag was samples by pumping 20 1 of the exhaled air
through the glass cartridge in the same manner.
The collected samples were preconcentrated by transferring the contents of the first
Tenax GC cartridge to a second Tenax GC cartridge. This minimized the quantity of water
vapor collected on the first cartridge. The contents of the second cartridge, dosed with an
internal standard, was then inserted into a desorption oven attached to a computer assisted
gas chromatograph/mass spectroscopy (GC/MS/COMP) system. Desorption was accom-
plished by heating the cartridge (250°C) in an inert gas stream and collecting the volatiles
in a cryogenically cooled trap. The trap was then rapidly heated and the contents flushed
onto a high resolution GC column with the helium carrier gas temporarily by-passed through
the cartridge. The sample on the high resolution GC column was analyzed in a cyclic scan
mode on a MAT 311A mass spectrometer using the Spectro 200 data system (Figure 1).
A spectrum-enhancement algorithm was used to produce a spectrum free of background and
contaminating components. The peak areas obtained were compared to the area of the
internal standard and subjected to a retention-time normalization procedures before an
historical library search routine was performed.
RESULTS
The results presented in Table 1 identify the distribution of VOCs in selected
submarine air samples collected at various stages of patrol. Because the number of
components found was large, the data were reduced to reflect the total quantities of material
in each of 10 major chemical classes8. The values presented represent the total ion count
of all components comprising each class. The last column identifies the number of MS
detector saturated peaks found in each class. To intercompare samples, these peaks were
arbitrarily assigned a value of 2 x 106 since it is not possible to give accurate ion count
values for such detector saturated peaks. This assignment permits comparisons even though
the degree of variation of the air samples (Table 1) compared to the breath samples (Table
2) is biased by the larger number of MS saturated peaks in the submarine air samples.
Therefore, the total ion counts for the submarine sample(s) are very conservative in
comparison to the breath samples.
Table 2 presents the distribution of the various chemical classes of VOCs found in the
baseline breath sample, and in the breath samples collected during and after patrol. A
summary of the total number of components and MS saturated peaks occurring in each
sample set, is presented.
It is clear from the data that the VOC concentrations in the submarine air (Table 1) are
-10 fold larger than the post-patrol expired breath values; and -20 fold greater than they
are in the pre-patrol breath sample. The major classes of VOCs in submarine air were the
alkanes, aromatics, and oxygen-containing organic components. The major classes in the
baseline breath sample were the alkenes and the oxygenated organic components. Of the
18 components present in the alkene class, isoprene represented the major component. In
the submarine sample, methyldecene-2 was the major component in the alkene class and
was the only MS saturated component of the six components comprising this class. At the
start of the patrol, the total VOC levels in submarine air were - 13 fold larger than the levels
238
-------
Table 1. Distribution of chemical classes in submarine air samples
Compound
Name
Alkanes
Alkenes
Cycloalkanes
Aromatics
Oxygenated Cycloalkanes
Oxygenated aromatics
Oxygen containing
Halogen containing
Nitrogen containing
Sulfur containing
Submarine Air*1
Total Ion Counts 103
Pre-
Patrol
54,454
3,105
5,350
14,968
2,196
518
18,203
6,142
6,490
6,000
117,426
Patrol
73,877
8,422
10,757
20,314
2,153
1,975
30,756
14,467
10,165
6,000
178,886
Post- °
Patrol
6,8521
3,118
8,361
28,236
2,267
548
24,186
8,047
12,130
6,000
161,414
Total No.
f Components
Per Class
68
6
18
62
13
20
35
10
14
3
249
No. of MS
Detector
Saturated
Peaks
35
1
5
13
1
0
10
3
6
3
77
ATotal ion counts based on mass spectrum response for 20 1 air sample and corrected for backbround x 103.
"Component peaks that saturated MS detector response arbitrarily given total ion count of 2000 x 103.
Table 2. Distribution of chemical classes in human breath samples
Compound
Name
Alkanes
Alkenes
Cycloalkanes
Aromatics
Oxygenated Cycloalkanes
Oxygenated aromatics
Oxygen containing
Halogen containing
Nitrogen containing
Sulfur containing
Breath Air*-'
Total Ion Counts 103
Pre-
Patrol
847
3,404
402
487
155
606
2,240
299
170
390
9,000
Patrol
29,387
2,052
4,225
4,402
74
361
9,125
6,688
2,064
2,032
60,410
Post-
Patrol
8,250
1,346
284
680
48
435
1,397
2,196
99
86
14,821
Total No.
of Components
Per Class
56
5
16
47
12
15
53
13
16
8
241
No. of MS
Detector
Saturated
Peaks
9
0
2
1
0
0
2
1
1
1
17
njtal ion counts based on mass spectrum response for 20 1 air sample and corrected for background x 103.
Component peaks that saturated MS detector response arbitrarily given total ion count of 2000 x 103.
239
-------
10000
6000 .
2000 .
Submarine Air
c
3
o
u
c
o
5000
3000
1000
1400
1000
600
200
Expired Air/Post Patrol
Expired Air/Pre-deployment
200 600 1000 1400
Scan Number
Figure 1. Typical total ion chromatograms of selected
air and breath samples.
240
-------
in the breath sample; following patrol were -9 fold larger. The breath sample collected
during patrol were higher in sulfur- and halogen-containing components than either the pre-
patrol or post-patrol breath samples.
It is of interest to note that the breath sample from the non-smoking crew member
yielded a benzene baseline ion count of 39, a patrol value of 84, and a post-patrol value of
36, while the submarine pre-patrol air sample yielded a value of 43. However, both the
patrol and post-patrol submarine air samples yielded saturated ion counts for benzene.
Toluene exhibited saturated ion counts for all three sample types (pre, patrol- and post-)
collected from the submarine air and represented the only component in the aromatic class
to exhibit a saturated peak in the patrol breath sample. Why the benzene value was so low
in the patrol breath sample as opposed to toluene is unclear, particularly since smoking is
permitted on-board on a limited basis. However, the patrol breath sample was collected
following a snorkling exercise.
It is also of interest to mention that in the oxygen containing compound class, 6-
methyl-5-heptene-2-one and 6,10-dimetlyl-5,9-undecadiene-2-one were identified in both the
pre- and post-patrol breath samples, but not in the patrol sample. These two compounds
were reported in the breath samples of both normal and diseased subjects, as well as in the
baboon7. Their presence has been proposed as representing products of a possible shunt
pathway in the mevalonic acid pathway. Their absence in the patrol breath sample might
be attributed to the four-fold increase in the total quantities of other oxygenated compounds
present in this sample.
DISCUSSION
The overall composition of submarine air, like all indoor air environments, is a
reflection of various emission sources. The complexity of the chemical species comprising
submarine air is not surprising since many of the emission sources in the submarine also
appear in the workplace and home. However, the total quantities of such components in
submarine air, as would be expected in a closed system, are substantially higher then found
in typical indoor/outdoor relationships8. Therefore, the main emphasis when monitoring
submarine environments should be on the potentially more health related components
present, and on ensuring that their presence is minimized by means of suitable air exchange
rates or efficient environmental control systems. Occupational exposure to the submarine
environment changed the distribution of the total ion counts across virtually all classes of
compounds in the expired breath. For example, alkenes were predominantly desorbed from
the subject before patrol; but after patrol, the distribution of expired VOCs shitted to alkanes
and halogenated substances. This suggests that the body tissues has a greater propensity
to absorb and retain alkanes and halogenated organics than the other classes of VOCs or,
that the other classes of VOCs are more easily metabolized and/or eliminated. Concerns
about the body burden and pulmonary clearance of these compounds, their biotransformation
products, and considerations relating to possible metabolic, physiological, or toxicological
significance must await a more comprehensive study.
REFERENCES
1. P. K. Weathersby, R. S. Lillo, and E. T. Flynn, Air Purity in Diyinq from Submarines.
I. Review and Preliminary Analysis. Naval Medical Research Institute Technical Report.
2. Submarine Air Quality: Monitoring the Air in Submarines, Report National Research
Council, National Academy Press, Washington, D.C. (1988)
241
-------
3. J. J. DeCorpo, J. R. Wyatt, and F. E. Saalfeld, "Atmospheric monitoring in
submersibles," presented at ASME Intersocietv Environmental Systems Confer^™,
San Diego, July 14-17, 198O
4. S. Hunt and R. F. Tindall, "A mass spectrometer atmosphere analyzer for nuclear
submarines, D," Internal. J. Mass Soect. and Ion Proc.. 60:277-287 (1984).
5. D. R. Knight, H. J. O'Neill, S. M. Gordon, E. H. Luebcke, and J. S. Bowman,
>. R. Knight, H. J. O Neill, &. M. Gordon, b. H. Luebcke, and J. S. Bowman, The Body
lurden of Organic Vapors in Artificial Air: Trial Measurements Abroad a Moored
-Submarine. Memo Report 84-4. Naval Medical Research and Development command
Submarine Base, Groton, CT, 1984.
6. H.J. O'Neill, S.M. Gordon, M.H. O'Neill, R.D. Gibbons and J.P. Szidon, "A
computerized classification technique for screening for the presence of breath
biomarkers in lung cancer. Clinical Chemistry 33:16T3-1618 (1988).
7. H.J. O'Neill, S.M. Gordon, B.K. Krotoszynski, H. Kavin and J.P. Szidon, "Identification
of isoprenoid-type components in human expired air: A possible shunt pathway in
sterol metabolism," Biomed. Chromatrogr. 2:66-70 (1987)
8. L. A. Wallace, "Personal exposures, indoor and outdoor air concentrations, and
exhaled breath concentrations of selected volatile organic compounds measured for
600 residents of New Jersey, North Dakota, North Carolina, and California," Toxicol.
and Environ. Chem.. 12:215:236 (19861.
242
-------
CABIN AIR QUALITY ABOARD COMMERCIAL AIRLINERS
Niren L. Nagda, Ph.D.
ICF Incorporated
9300 Lee Highway
Fairfax, Virginia 22031-1207
Michael D. Koontz, M.S.
GEOMET Technologies, Inc.
20251 Century Boulevard
Germantown, Maryland 20874-1192
and
Roy C. Fortmann, Ph.D.
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, North Carolina 27709-2194
ABSTRACT
The purpose of the study, conducted for the U.S. Department of Transportation, was to develop information to
be used for determining health risks from exposure to environmental tobacco smoke (ETS) and other pollutants for
airliner occupants. ETS contaminants (nicotine, respirable suspended particles, and carbon monoxide) as well as ozone,
microbial aerosols, carbon dioxide and other environmental variables were measured on 92 randomly selected flights,
including 69 smoking and 23 nonsmoking flights.
Respirable suspended particles averaged 175 /ig/m3 hi coach smoking sections compared to background levels
of 35 to 40 /ig/m3 on nonsmoking flights. Nicotine levels were 13.4 /ig/m3 in smoking and below 0.3 /ig/m3 in no-
smoking sections and on nonsmoking flights. Measured carbon dioxide levels averaged 1500 ppm, well above the
American Society of Heating, Refrigerating and Air Conditioning Engineers' comfort criterion of 1000 ppm. Levels of
carbon monoxide, ozone and microbial aerosols generally were quite low.
243
-------
INTRODUCTION
The airliner cabin environment has been of great concern for the last twenty years to various elements of the
U.S. Federal Government, special interest groups organized to advocate public or industry positions, and the general
public itself. Passenger complaints about smoking led to segregating smoking passengers in the early 1970s. Later
concerns about stratospheric ozone prompted standards (0.25 ppm maximum instantaneous level and 0.1 ppm as time-
weighted three-hour standard) for the ozone concentration in airliner cabins'. Public Law 100-200, enacted in 1987 and
effective for two years beginning in April 1988, prohibited smoking by passengers on any scheduled domestic
commercial flight of two hours or shorter duration. At the same time, the U.S. Department of Transportation (DOT)
received Congressional approval to conduct a study to resolve certain technical questions related to potential
continuation or broadening of the prohibitions in the law.
The purpose of the study2 was to develop information to be used for determining health risks from exposures to
ETS for nonsmoking airliner occupants as well as risks from other pollutants of concern for all airliner occupants.
Although prior studies3'4 were useful in suggesting ranges of concentrations of ETS tracers encountered in the general
airliner cabin environment, the monitored flights were not randomly selected and the number of observations was
generally small, precluding any generalization of the results. Similarly, determining factors (e.g., smoking rates,
ventilation systems, seating patterns) of ETS concentrations for the general airliner cabin environment were not
investigated in depth.
METHODS
Selection of Flights
The target sample size for the study was 60 to 120 smoking flights on jet aircraft, including some international
flights. A smaller set of 20 to 40 nonsmoking flights was targeted to provide a baseline for comparison. The target
sample size for nonsmoking flights was smaller because flight-to-flight variations in ETS contaminant levels were
expected to be lower than for smoking flights.
A total of 70 airports that collectively accounted for 90 percent of U.S. enplanements during 1987 was used as
the sampling frame for selection of flights to be monitored. Airports of departure were randomly selected for study
flights to provide proportional representation of airports associated with all smoking and nonsmoking flights scheduled
for departure during January 1989, based on computer data files supplied by DOT. The random selections were made
separately for smoking and nonsmoking flights. The specific flights to be monitored were chosen by randomly chaining
together the selected airports of departure, subject to constraints relating to the smoking/nonsmoking status of flights.
Measurements
Air pollutants were selected for monitoring that had known or suspected sources in the aircraft and could be
monitored or sampled in airliner cabins with small, unobtrusive instrumentation. The ETS contaminants monitored
during the study were nicotine, respirable suspended particles (RSP), and carbon monoxide (CO). The other pollutants
that were monitored were ozone and microbial aerosols. In addition, carbon dioxide (COJ was monitored. The
monitoring package configured for the study consisted of instruments and sensors for measurement of time-varying
concentrations of contaminants in addition to samplers for collection of time-integrated samples. It also included a data
acquisition system for recording outputs from the continuous monitors. The instrument was packaged in a single,
compact carry-on bag (approximately 18 inches long, 9 inches wide, and 9 inches high) typical of that carried by airline
passengers.
Nicotine was measured through collection of time-integrated samples and CO was measured with portable
continuous monitors (Table I); RSP was measured both by integrated and continuous methods, with an optical sensor in
the latter case. CO2 and ozone were measured with time-integrated samples whereas short-term samples were collected
for microbial aerosols (bacteria and fungi) near the end of each flight, prior to descent. Temperature, relative humidity,
and cabin air pressure were monitored continuously with portable sensors. Air exchange rates were measured using
constant release and integrated sampling of perfluorocarbon tracers (PFTs). Smoking rates were estimated through
technician observations of the number of lighted cigarettes during a one-minute interval every 15 minutes and collection
of cigarette butts at the end of most monitored flights. All aspects of the measurement protocol were pretested on four
commercial flights that were monitored over a three-day period in March 1989.
The 92 study flights were monitored over a 10-week period between April and June 1989. Monitoring was
performed by each technician at an assigned seat. Based on pretest monitoring at a variety of locations, the following
four locations were chosen for monitoring on smoking flights: coach smoking section; boundary region of the
244
-------
no-smoking section within three nonsmoking rows near the coach smoking section; middle of the no-smoking section;
and remote no-smoking section (i.e., as far as possible from coach smoking, usually near the first-class smoking and
no-smoking sections). Because less substantial variations were expected on nonsmoking flights, two locations (middle
and rear of the plane) were chosen for those flights. ETS contaminants were monitored at all seat locations and other
pollutants were monitored at half of the locations. The instrument package typically was placed on the technician's lap
or lap tray to obtain measurements of contaminants most representative of passenger breathing levels.
RESULTS
Temperature, Humidity, Pressure and Air Exchange
The average measured temperature on study flights was near 76 °F (range: 71 to 81 °F). Measured relative
humidity levels were quite low, ranging from 5 to 38 percent across all flights, but were even lower for smoking
(average of 15.5 percent) than nonsmoking flights (average 21.5 percent). The average cabin pressure was lower for
smoking (635 mm Hg) than for nonsmoking (686 mm Hg) flights. Both the lower humidity and the lower pressure are
consistent with higher altitudes that would generally be reached on the longer-duration smoking flights. For aircraft
without recirculation, the pattern of measurement results indicated that there generally was insufficient mixing of PFTs
throughout the airliner cabin for the results to be indicative of prevailing air exchange rates. In the case of aircraft with
recirculation, the measured air exchange rates generally were consistent with but somewhat higher than nominal rates
based on information provided by equipment manufacturers and airline operators.5
Passengers, Smoking, and Concentrations of ETS Contaminants
The load factor (i.e., percent seating capacity filled by passengers) averaged 76 percent for smoking flights on
narrow-body aircraft (average passenger capacity of 138) and 64 percent for smoking flights on wide-body aircraft
(capacity 288). For the nonsmoking flights (capacity 135), the average load factor was 70 percent. On the average,
there were 18 passengers in the coach smoking section (13.7 percent of all passengers). The average smoking rate per
smoking-section passenger was 1.5 cigarettes per hour (range: 0.2 to 3.5 cigarettes per hour per passenger) during the
period when smoking was allowed. An average of 68 cigarettes per flight was smoked by passengers in the coach
smoking section on the monitored smoking flights.
Average values for various measurement parameters related to particle-phase and gas-phase ETS contaminants
are summarized by monitoring location for both smoking and nonsmoking flights in Table II. RSP concentrations were
highest in the smoking section, averaging near 175 ^g/m3, and results for the gravimetric and optical methods were
highly consistent. In other locations, however, the two methods yielded differing results. There was greater uncertainty
for the gravimetric measurements due to relatively short monitoring durations for a number of flights. For example, 60
minutes of sampling duration (or about 80 minutes flight duration) on a nonsmoking flight would correspond to a sample
volume near O.lm3. For this case, a laboratory uncertainty in mass determination of + 10 /tg could result in
measurement values ranging from -100 to + 100 /ig/m3 for a prevailing concentration near zero.
Cabin-wide optical results were more strongly correlated (r = 0.6) with smoking rates than were the
gravimetric results (r = 0.3), and optical concentrations in the smoking section were also more strongly correlated
(r = 0.6) with nicotine concentrations than were the gravimetric concentrations (r = 0.5). However, because the
gravimetric method has a long history of successful use in various types of environments, neither type of measurement
result can be ignored. The values obtained from averaging the results of the two methods (Table II) indicate that
differences across the no-smoking sections of the aircraft for smoking flights and differences between these no-smoking
sections and nonsmoking flights were less pronounced than differences involving the smoking section. The combined
results for nonsmoking flights are consistent with RSP values that have been reported for other nonsmoking
environments6. The one-minute peak RSP concentrations indicated some migration of ETS contaminants into the
no-smoking sections on smoking flights.
Observed effects of tobacco smoking, based on gas-phase measurements, were more discernible for nicotine
than for CO (Table II). Beyond the marked increase in nicotine in the smoking section, the boundary region of the
no-smoking section was most affected. Differences between nicotine levels for the remaining no-smoking locations and
levels on nonsmoking flights were within the range of measurement uncertainty, but nicotine levels were more often
above detection limits hi the no-smoking locations of smoking flights than on nonsmoking flights. The only discernible
effect for CO was in the smoking section itself. CO levels generally were highest before aircraft were airborne, both for
smoking and nonsmoking flights, due to intrusion of ground-level emissions.
245
-------
Both nicotine and RSP concentrations in the coach smoking section were strongly related to observed smoking
rates in that section7. This relationship also persisted in the boundary region near coach smoking, though not as
strongly. For the other no-smoking sections, there was no apparent relationship between ETS levels and smoking rates,
Within the boundary region, ETS concentrations generally were higher when the technician was seated within one or two
rows of coach smoking than when the boundary seat was three or more rows away.
Concentrations of Other Pollutants
Monitored ozone levels were relatively low (Table III), averaging an order of magnitude below the three-hour
standard of 0.1 ppm and never exceeding this level. Bacteria levels were higher than fungi levels and somewhat higher
in smoking than nonsmoking sections, but the measured bacteria and fungi levels in all cases were low, relative to those
that have been measured in residential environments through cross-sectional studies*.
Relatively high CO2 levels were measured (Table III), averaging over 1,500 ppm across all monitored flights.
Measured CO2 concentrations exceeded 1,000 ppm, the American Society of Heating, Refrigerating and Air
Conditioning Engineers (ASHRAE) level' associated with satisfaction of comfort (odor) criteria, on 87 percent of the
monitored flights. Depending on assumed CO2 exhalation rates, measured levels were as much as twice those predicted
by a cabin air quality model. Even if the measured levels were to be lowered by half, however, C02 concentrations
would still exceed 1,000 ppm on 24 percent of the study flights.
Average CO2 levels measured in the middle of airliner cabins were examined in relation to the type of aircraft,
load factor, air recirculation, and the nominal ventilation rate (expressed as cfm of fresh air per passenger)10. Narrow-
body aircraft had higher CO2 levels (1700 ppm) than wide-body aircraft (1200 ppm). The narrow-body aircraft have a
smaller volume per passenger on the average, and the monitored flights were more fully occupied than the wide-body
aircraft studied. Aircraft with recirculation had slightly higher CO2 levels on the average than aircraft with 100 percent
fresh air. Both the load factor and ventilation rate had very strong relationships with CO2; levels increased as the load
factor increased and decreased as the ventilation rate increased. However, even for a subset of 12 flights with
ventilation rates of 35 cfm per passenger or greater, the average CO2 level (1,237 ppm) was above 1,000 ppm.
CONCLUSIONS
Levels of ETS contaminants monitored during the study were substantially higher in smoking sections of the
aircraft than in nonsmoking areas, and these levels were strongly correlated with observed smoking rates. There was
some evidence of ETS migration to the no-smoking boundary region near the smoking section, particularly for RSP
concentrations in this region that were related to smoking rates and distance from the smoking section. Monitored C02
levels were sufficiently high and monitored humidity levels were sufficiently low to pose potential comfort problems for
aircraft occupants. Ozone levels on all monitored flights were well within existing standards for airliner environments,
and monitored levels of microbial aerosols were below those in residential environments that have been characterized
through cross-sectional studies.
REFERENCES
1. Cabin Ozone Concentrations. Code of Federal Regulations Title 14, Pt. 25.832, U.S. Government Printing Office,
Washington, D.C. (1985).
2. N.L. Nagda, R.C. Fortmann, M.D. Koontz, S.R. Baker, and M.E. Givevan, Airliner Cabin Environment:
Contaminant Measurements. Health Risks, and Mitigation Options. Report No. DOT-P-15-89-5, U.S. Department of
Transportation, Washington, D.C (1989).
3. G.B. Oldaker, and F.C. Conrad, "Estimation of Effects of Environmental Tobacco Smoke on Air Quality Within
Passenger Cabins of Commercial Aircraft," Environ. Sci. Technol.. Vol. 21, No. 10, pp. 994-999 (1987).
4. M.E. Mattson, et al., "Passive Smoking on Commercial Airlines," J.Am. Med. Assoc.. Vol. 261, No. 6, pp. 867-
872 (1989).
5. D.G. Lorengo and A. Porter, Aircraft ventilation systems study, final report DTFA-03-84-C-0084, U.S. Federal
Aviation Administration Technical Center, Atlantic City, NJ (1985).
6. J.L. Repace, "Indoor Concentrations of Environmental Tobacco Smoke: Field Surveys," in Environmental
Carcinogens Methods of Analysis and Exposure Measurement: Volume 9—Passive Smoking. Ed. I.K. O'Niell, K.
Brunneman, B. Dodet, and D. Hoffman, International Agency for Research on Cancer, Lyon, France (1987).
7. N.L. Nagda, M.D. Koontz, A.G. Konheim, and S.K. Hammond, "Measurement of cabin air quality aboard
commercial airliners," Atmospheric Environment 26A, 2203-2210 (1992).
8. R.L. Tyndell, C.S. Dudney, A.R. Hawthorne, R. Jemigan, K. Ironside, and P. Metier, "Microflora of the Typical
Home," Proceedings of the 4th International Conference on Indoor Air Quality and Climate 1:617-621, Institute for
Water, Soil and Air Hygiene, Berlin (West) (1987).
246
-------
9. ASHRAE, Ventilation for Acceptable Indoor Air Quality. ASHRAE Standard 62-1989, American Society of Heating,
Refrigerating and Air Conditioning Engineers, Atlanta, Georgia (1989).
10. N.L. Nagda, M.D. Koontz, and A.G. Konheim, "Carbon dioxide levels in commercial airliner cabins," ASHRAE
Journal 33. 35-38 (1991).
11. N.L. Nagda and M.D. Koontz, "Microenvironmental and Total Exposures to Carbon Monoxide for Three
Population Subgroups," J. Air Poll. Control Assoc. 35(2):134-137 (1985).
12. S.K. Hammond, B.P. Leaderer, A.C. Roche, and M. Schenker, "Collection and Analysis of Nicotine as Marker for
Environmental Tobacco Smoke," Atmos. Environ. 21(2):457-462 (1987).
13. B.J. Ingebrethsen, D.L. Heavner, A.L. Angel, J.M. Conner, T.J. Steichen, and C.R. Green, "A Comparative
Study of Environmental Tobacco Smoke Paniculate Mass Measurements for an Environmental Chamber," J. Air Poll.
Control Assoc. 38:413^117 (1989).
14. H.A. Burge, M. Chatigny, J. Feeley, K. Kreiss, P. Morey, J. Otten, and K. Peterson, "Bioaerosols—Guidelines for
Assessment and Sampling of Saprophytic Aerosols in the Indoor Environment" Appl. Indust. Hyg. 2(5):R-10 to R-16
(1987).
15. J.L. Lambert, J.V. Paukstelis, and Y.C. Chiang, "3-Methyl-2-benzothiazolinone Acetone Azine with
2-Phenylphenol as a Solid Passive Monitoring Reagent for Ozone," Environ. Sci. Technol. 23:241-243 (1989).
16. A.L. Lynch, Evaluation of Ambient Air Quality by Personnel Monitoring. CRC Press, Inc., Boca Raton, FL
(1981).
17. ASHRAE, ASHRAE Handbook - 1985 Fundamentals. American Society of Heating, Refrigerating, and Air
Conditioning Engineers, Atlanta, GA (1985).
18. R.N. Dietz, and E.A. Cote, "Air Infiltration Measurements in a Home Using a Convenient Perfluorocarbon Tracer
Table I. Measurement parameters and methods.
Parameter
ETS contaminants
Carbon monoxide
Nicotine
Respirable particles
(integrated)
Respirable particles
(continuous)
Microblal aerosols
Fungi
Bacteria
Pollutants
Ozone
Carbon dioxide
Other parameters
Temperature
Relative humidity
Barometric pressure
Air exchange
Sample collection method
continuous monitor
sodium-bisulfate treated
filter
filtration with cyclone
separator
continuous monitor
impaction
impaction
MBTH*-coated filter
detector tube
continuous
continuous
continuous
sorbent tube (passive)
Analysis method
solid polymer electrolyte
gas chromatography—
nitrogen selective detector
gravimetry
nephelometry
culture/microscopy
culture/microscopy
spectrophotometry
length of stain
platinum RTD
thin-film dielectric sensor
piezoresistance
gas chromatography of
perfluorcarbon tracer (PFT)
Reference
11
12
12
13
14
14
15
16
17
17
17
18
' 3-Methyl-2-benzotniazolinone.
247
-------
Table II. Levels of ETS contaminants on smoking and nonsmoking flights.
Smoking flights
Parameter
Smoking Boundary
Middle
Remote
Nonsmoking flights
Rear Middle
Particle-phase measurements
Average RSP
(gravimetric) (pg/m3)
Average RSP
(optical) (fig/m3)
Average RSP
(both methods) (fig/m3)
Average of peak RSP
(optical) Oig/m3)
174.6
177.0
175.8
833.4
67.5
39.7
53.6
211.8
42.5
18.8
30.7
68.7
52.1
17.9
35.0
69.6
59.3
10.3
34.8
18.2
69.4
10.6
40.0
16.4
Gas-phase measurements
Average nicotine (/ig/m3)
Percent nicotine samples
below minimum detection
Average CO (ppm)
Peak CO (ppm)
13.43
4.3
1.4
3.4
0.26
54.4
0.6
1.4
0.04
82.6
0.7
1.7
0.05
66.7
0.8
1.6
0.0
100.0
0.6
1.3
0.08
78.3
0.5
0.9
Table HI. Levels of other pollutants on smoking and nonsmoking flights.
Smoking flights
Parameter Smoking rows Middle rows Nonsmoking flights
Average COZ (ppm)
Percent CO2 samples > 1000 ppm
Average ozone (ppm)
Percent ozone samples > 0. 1 ppm
Average bacteria (CFU/m3)
Average fungi (CFU/m3)
1562
87.0
0.01
0.0
162.7
5.9
1568
88.1
0.01
0.0
131.2
5.0
1756
87.0
0.02
0.0
131.1
9.0
248
-------
REMOVAL OF GASEOUS INDOOR AIR CONTAMINANTS
BY COMMERCIAL AIR FILTERS
P.R. Nelson, R.B. Hege, F.W. Conrad, and W.M. Coleman IV
RJ. Reynolds Tobacco Company, Bowman Gray Technical Center
Winston-Salem, NC 27102
ABSTRACT
The use of effective air cleaners provides one means to reduce indoor air
contaminant concentrations. A series of commercial air filters was evaluated to determine their
ability to remove gaseous contaminants from indoor air. Environmental tobacco smoke (ETS) was
used as the test matrix because it contains many of the compounds which are typically present in
indoor air. Concentrations of CO, NO, NOj, Total VOC (by FID detector), formaldehyde,
nicotine, and 3-ethenylpyridine were determined. The data for each filter/compound combination
were analyzed using a recently developed model for determining clean-air delivery rates. Results
of the study demonstrated that significant removal of VOCs from indoor air can be accomplished
through the use of sorbent-based air filtration systems. However, performance of sorbent systems
varied across manufacturer and compound class. A reactive substrate designed for the removal
of formaldehyde was effective, but its use resulted in the oxidation of NO to NO2. These results
suggest that if reactive removal mechanisms are used, the potential effect of reaction byproducts
on IAQ must also be addressed.
INTRODUCTION
Air cleaners and filtration equipment can play an important role in maintaining good indoor
air quality. Filters have long been used for removal of particulate contaminants from indoor air,
and many types of filter media are also available for removal of gas-phase compounds as well.
Filtration equipment can be used to remove general or specific classes of compounds and odors.
Filtration may provide a viable alternative to increased ventilation in areas of a building with
increased contaminant loads or specific problem contaminants or odors.
The ability of a filtration device to remove particulate phase compounds is addressed in
several standards (1,2). One method has been developed to test the ability of an air cleaner to
remove particles under actual use and provides a good estimate of expected performance under
actual operation conditions (1). However, there isn't a similar straight forward method to evaluate
the ability of filtration devices to remove gas-phase compounds from indoor air. To address this
need, a new method to determine one performance criterion, the clean air delivery rate (CADR),
was developed (3). (The CADR of an air cleaner is the equivalent volume of contaminant free air
discharged by the cleaner. An air cleaner with a CADR of 100 CFM would have the same effect
on contaminant concentrations of adding 100 CFM contaminant free air to the room.) The newly
developed method for determining CADR can be applied to vapor phase compounds and is
performed under actual operating conditions. The new method was used to compare the ability
of several commercial air filters to remove gas-phase compounds from the air.
249
-------
EXPERIMENTAL
Determination of Clean Air Delivery Rates (CADR)
Clean air delivery rates were determined by application of a phenonemological model for
the behavior of gas-phase compounds in indoor air. A detailed derivation of the model appears
elsewhere (3). Simply, the model assumes that the concentration of a contaminant at some time,
t, can be described by an equation of the form:
(1)
at
where A(t) is the generation rate of the contaminant at time t.
If one makes the simplifying assumptions that the generation rate is constant and the
component follows a first order decay process, then it is possible to solve for the concentration
at any time, t.
-[l-e"*''] + Coe-*'r (2)
If the generation term is known, then it is possible to determine a solution for k by non-linear
regression. Alternatively, at t-»oo, the rate of generation will be equal to the rate of decay; i.e.,
the system will be at a steady state. The ratios of two steady state concentrations can be combined
to yield the following equation:
C. k.
*'
Cla and Ciro can be determined experimentally. If Ic^ is the decay rate of a component with no
air cleaner in operation, then kr2 can easily be determined. The additional removal rate due to
the air cleaner is simply the difference between krl and kr2. The rate constant also can be
expressed in CFM. Then, the difference between the rate constants is the CADR of the air
cleaner.
Environmental Chamber
The experiments were carried out in a newly constructed 45-m3 environmental chamber.
The chamber is constructed of stainless steel with a baked enamel finish on the interior walls.
Illumination is provided by variable fluorescent and incandescent lights. Air is cooled or heated
and humidified in an air handler within the chamber which recirculates chamber air at =1500
CFM. As much as 450 CFM of makeup air also can be introduced into the chamber through the
air handler. The makeup air is dehumidified and passed through HEPA and charcoal filters
before introduction into the room. Ventilation, temperature and relative humidity can be
accurately and reproducibly controlled within the room. Air is drawn from the room for analysis
through a teflon tube and sampling ports are located along one wall of the room.
Real-time determinations were made of CO (Thermo Electron Model 48, Franklin, MA),
NO, (Thermo Electron Model 42, Franklin, MA), VOCs by FID (Combustion Engineering Model
8401, Ronceverte, WV). In addition, nicotine and 3-ethenylpyridine were collected on XAD-4
sorbent tubes (SKC Inc. Eighty-four, PA) and analyzed by GC with NPD detection (4,5), and
250
-------
formaldehyde was collected on 2,4-DNPH-impregnated Sep-Paks (Waters Chromatography Part
37500, Milford, MA) and determined by HPLC with UV detection.
I Smoke was generated by a smoking machine located in the chamber. Following a 15-
minute background collection period, two pairs of cigarettes were smoked in sequence to rapidly
elevate smoke concentrations in the room to near steady state. For the remainder of the
experiment, individual cigarettes were burned continuously. The smoke concentration in the room
remained steady from 50-140 minutes (see Figure 1). Sampling for nicotine, 3-ethenylpyridine and
formaldehyde was performed during this period. The 50-140 minute period was also used for
determining the average concentration
of analytes measured by real-time
monitors.
Air Cleaners
Smoke particles were removed
from the room by an electrostatic
precipitator. This was done to protect
the filter media from particle build-up.
The fan in the electrostatic precipitator
was also used to force air through the
various filter media. The different
filters used are described in Table I.
Filter media were obtained from three
different manufacturers. All three
panel filters were partial-bypass filters.
Bypass filters allow some air to pass
through, without coming in contact with
the adsorbent media. The bed filters
contain adsorbent-filled panels through
40 60 80 100 120 140
TIME (min)
Figure 1. Real-time CO concentration for the Control
condition. Sampling/averaging time for determining
analyte concentrations is indicated on the figure.
Table I. Description of filters evaluated. Filter type, media, number of tests using filter, and CO
concentration during averaging period with each filter in use are shown.
Identifier
PF1
PF2
PF2b
BC
BCC
BPC
BPA
Type
panel
panel
panel
bed
bed
bed
bed
Filter
Area (ft3)
1
1
1
5.5
11
11
11
Filtration
1" carbon
2" carbon
2" carbon
l/2 carbon
V4" carbon X 2
Va" permanganate
Vi" carbon
V? permanganate
Vi" acid-treated carbon
# of runs
4
2
2
1
5
3
3
CO (ppm)
3.17
3.19
3.31
3.02
3.05
3.02
3.14
Control
N/A
N/A
None
2.95
251
-------
which all the air must pass. The large surface area of these filters decreases the face velocity
across the panels, which results in an increased residence time for air passing through the filter.
The manufacturer of the bed filters provided the electrostatic precipitator and a housing
to hold the media beds. A 12" x 12" box was constructed to hold the panel filters. The bed filters
were installed downstream from the electrostatic precipitator in the order shown in Table I; e.g.,
the permanganate-containing bed was installed upstream from the carbon bed in the BPC
arrangement. "Cleaned" air was directed through a piece of flex-duct into the recirculation inlet
of the air handler. This was done to minimize the likelihood of short-circuiting of air between the
outlet and inlet of the air cleaner. Air flows were determined by performing a five-point traverse
of the inlet duct with a hot-wire anemometer, averaging the five-flow rates, and converting linear
velocity to volumetric flow. The fan velocity was adjusted to give a 500 CFM flow through the
filters. The electrostatic precipitator was not operated during the Control sessions; however, it
was continuously operated during filter test sessions.
Table I also shows the number of runs performed with each type of filter media and the
CO concentration during the averaging period (50-140 min). The CO concentrations in the table
show that similar amounts of ETS were generated in each of the tests. Thus, differences in
concentration of other analytes are predominantly due to the effect of the filtration equipment
rather than differences in smoke production between runs.
RESULTS
Because the removal efficiency of any filter will vary for different types of contaminants,
it is necessary to determine the CADR for each compound of interest. CADRs for each air filter
and analyte tested are summarized in Table II. The efficiency of the filter media can be
determined by dividing the CADR in the table by the flow-rate of air through the filter (500
CFM). It can be seen that efficiency varies with both analyte and type of filter media. Negative
CADRs indicate that the concentration of the given compound increased when that particular type
of filter media was used. Physically, this indicates an effective decrease of ventilation to a space
with respect to that particular compound. The results for each compound are discussed in more
detail below.
Table II. Clean air delivery rates (CADR) for each analyte measured by media type. One-pass
filter efficiency can be calculated by dividing the CADR by 500 CFM.
Filter
PF1
PF2
PF2b
BC
BCC
BPC
BPA
VOC
88
124
197
229
296
255
199
Nicotine
83
111
98
189
235
214
227
3-Ethenyl-
pyridine
120
248
239
389
459
454
391
Formaldehyde
-57
-50
NR
53
89
192
137
NO,
NR
NR
NR
79
91
166
118
NO
39
26
NR
79
93
357
249
NO2
-107
-74
NR
27
65
-87
-92
252
-------
Volatile Organic Compounds
The FID gives a response proportional to the total VOC concentration in the room.
Although this detector is non-specific for particular classes of compounds, the results are indicative
of the filter's ability to remove general organic compounds from the air. Each type of filter media
was effective at removing VOCs from the air in the chamber. In general, increasing filter
thickness and the amount of carbon present in the filter lead to increased efficiency. There may
also be some effect of charcoal type. The use of a permanganate-impregnated media appeared
to remove some VOCs (BC vs. BPC). Phosphoric acid-impregnated charcoal reduced the removal
of VOCs by the filter (BPA vs. BPC). It is likely that impregnation reduced the number of pores
available in the charcoal that could serve as adsorption sites.
Nicotine
Previous studies have shown that nicotine is surprisingly difficult to remove from the air
with filtration equipment (3). For each of the filters tested here, the removal of nicotine was not
as efficient as the removal of general VOCs. The one exception involves the use of acid-treated
charcoal. Nicotine is a gas-phase base; reaction with phosphoric acid in the filter leads to the
formation of a non-volatile salt. However, the use of a permanganate-treated filter appeared to
have as great an effect on the removal of nicotine (BPC vs BC) as the use of an acid-treated filter
(BPC vs BPA). The use of significant quantities of carbon would appear to be the best way to
remove this compound from indoor air.
3-Ethenylpyridine
When tobacco is burned, nicotine is pyrolyzed. One product of this pyrolysis is 3-
ethenylpyridine. Whereas nicotine demonstrated a poorer removal rate than VOCs in general,
all the filters were much more efficient at removing 3-ethenylpyridine than they were at removing
VOCs in general. The removal efficiency exceeded 90% for the BCC filter. Once again, quantity
of charcoal appears to be the most significant factor in predicting the removal of this compound
from the air. Results for this compound further demonstrate that one cannot predict the removal
of specific compounds based on a general indicator.
Formaldehyde
In general, carbon filters are not efficient at removing formaldehyde from the air. This can
be seen in the results presented in Table II. PF1 and PF2 appear to have led to an increase in
the formaldehyde concentration in the room. Some of the increase may be due to slightly higher
smoke levels in the room when this filter was in use (see Table I). However, it also appears that
some off-gassing of formaldehyde from the sorbent or filter construction materials may have
occurred. PF2b did not remove formaldehyde from the air. The carbon-bed filters removed a
small amount of formaldehyde. Permanganate-impregnated filters remove formaldehyde by
oxidation. Reasonably efficient removal of formaldehyde was demonstrated by the permanganate-
containing filters (BPC, BPA). Unlike the plain charcoal beds (BC, BCC), the acid-treated
charcoal was ineffective at removing formaldehyde from the air.
Oxides of Nitrogen
If one examines the CADRs for NO,,, a mistaken impression of the behavior of the filters
can be obtained. Neither the PF1 nor the PF2 filters removed NOX. However, both filters appear
to have removed NO from the air by oxidizing it to NO2. The carbon-bed filters removed a small
amount of both NO and NO2 from the air. The permanganate-containing filters removed NO
efficiently, but they did so in part by oxidizing NO to NO2. Although the permanganate-
containing filter was relatively efficient at removing NOX and quite efficient at removing NO, it
did so by generating a less desirable compound, NO2. A potential exists for generating unknown
and/or undesirable reaction products with oxidative filter media. When using such media, one
253
-------
must be careful to ensure that the removal of one compound does not result in the generation of
less desirable compounds in an enclosed space.
CONCLUSIONS
Air cleaners can be effectively used to remove a wide range of compounds from indoor air.
Commercial filters differ greatly in their ability both to remove various amounts and to select
particular classes of compounds for removal from air. Performance criteria such as CADR can
be used to help determine the best filter for an application. CADRs also can be calculated for
compounds whose concentrations in air cannot be determined in real-time.
Because some air filters/cleaners operate by reacting with species in indoor air, one must
be careful to determine that undesirable species are not being formed as byproducts of such
reactions. A general knowledge of the effect of a filter on indoor air quality does not alleviate the
need for testing a filter to determine its effect on the concentration of specific compounds.
REFERENCES
1. AHAM, Standard AC-1-1986, Method For Measuring Performance of Portable Household
Electric Cord-Connected Room Air Cleaners. Chicago, Association of Home Appliance
Manufacturers, 1986.
2. ASHRAE, Standard 52-1, Methods of Testing Air Cleaning Devices Used in General
Ventilation for Removing Particulate Matter. American Society of Heating, Refrigerating,
and Air Conditioning Engineers, Inc., Atlanta, 1992.
3. P.R. Nelson, S.B. Sears, D.L. Heavner, "Application of Methods for Evaluating Air Cleaner
Performance," Indoor Environ., 2, 111-117 (1993).
4. M.W. Ogden, et al, "Improved Gas Chromatographic Determination of Nicotine in
Environmental Tobacco Smoke," Analyst, 114. 1005-1008 (1989).
5. M.W. Ogden, "Use of Capillary Chromatography in the Analysis of Environmental Tobacco
Smoke," in Capillary Chromatography - The Applications. W.G. Jennings and J.G. Nikelly
eds., Hiithig, Heidelberg (1991) pp. 67-82.
ACKNOWLEDGEMENTS
Credit is due to Mr. Charles Risner for his determinations of the formaldehyde
concentration in the test chamber. Thanks are also due to Dr. Stephen Sears for his assistance
in deriving the comparative decay kinetics in the chamber.
254
-------
Session 7
Measurement Methods Development
-------
FAST ANALYSIS OF C2-C12 AMBIENT AIR HYDROCARBONS USING
A MULTI-COLUMN GAS CHROMATOGRAPHIC SYSTEM
Kochy Fung
AtmAA, Inc.
21354 Nordhoff Street, Suite 113
Chatsworth, CA91311
ABSTRACT
A method has been developed for the speciation of C2-C-)2 hydrocarbons (HC) in
ambient air without employing subambient temperature programming. A ten-port valve was
used for cryogenic concentration of sample air as well as column switching in a three-column
system housed in a single GC oven. The C2-Cg hydrocarbon fraction was analyzed
concurrently with >CQ fraction and detected by separate flame ionization detectors. Analysis
time was approximately 30 minutes.
INTRODUCTION
HC in ambient air are ozone precursors. The main sources of these compounds in an
urban environment are emissions from automotive and stationary sources as a result of
incomplete combustion and evaporation of fuels and solvents. Speciated HC data are
essential to the running of photochemical models that are used by air pollution control
agencies in formulating ozone control strategies. Under the 1990 Clean Air Act Amendments,
the USEPA mandates these agencies in serious, severe, or extreme ozone non-attainment
areas to establish Photochemical Assessment Monitoring Stations (PAMS) which monitor
routinely speciated HC among other parameters.
Gas chromatography has been used for the measurement of ambient air HC. Because
they exist at low levels, it is necessary to concentrate the sample using either solid sorbents
or cryogenic means prior to analysis. Earlier methods employed several chromatographs,
each designed to speciate a certain fraction of the sample HC. In one such approach, the C2-
Cg aliphatic, the C/j-Cg aliphatic, and the Cg-C-) 1 aromatic and aliphatic components were
determined by analyzing separate aliquots of a sample (Lonneman et al. 19741; Westberg et
al. 1984^). Other approaches involved two analyses for the light and heavier fractions
respectively (Dimitriades and Seizinger 19713; Rasmussen et al. 19744, Singh 19805;
Grosjean and Fung 19846; Stump and Dropkin 19857). These approaches have the
disadvantages of needing additional gas chromatographs, sample concentrators, and data
systems in order to process the samples efficiently.
The development of these various approaches reflects improvements in column
technology through the years. Packed and capillary columns with coated stationary phase
have been replaced by high resolution bonded phase fused silica capillary columns. Seila and
Lonneman (19888) performed the analysis of C2-C-|2 HC with a single column albeit needing
sub-ambient temperature programming for separation of the C2 components. However, the
C2 separation is not always ideal as these researchers have experienced. A thick film (low
257
-------
phase ratio) column that is needed to achieve good C2 separation is not efficient for the heavy
components. Consequently, analysis time is prolonged due to the necessity of starting at a
sub-ambient oven temperature (~ -60°C) and finishing at a high temperature. Returning the
oven to the initial conditions also takes some time and increases cryogen use. Recycle time
is in excess of an hour.
There have been continual interests in a gas chromatographic system for the
speciation of ambient HC with minimal need for cryogen. Such systems in automatic form, for
example, would be useful in gathering the much needed HC data in remote areas for the
operational and diagnostic evaluation of large scale acid deposition models like Regional Acid
Deposition Model and Acid Deposition and Oxidant Model. PAMS will also be benefited
because excessive demand for cryogen increases considerably the operating costs and
creates difficulties in supply logistics. The present method was developed to limit cryogen use
and shorten analysis time without sacrificing performance and resolution. The method has
been under evaluation for the past several years, and applied to the analysis of ambient air
samples from the rural areas of the eastern United States and urban samples in California.
EXPERIMENTAL METHODS
Chrompack (Raritan, NJ) pioneered the use of a Porous Layer Open Tubular (PLOT)
column with an internal deposit of A^Os/KCI for analysis of hydrocarbons. The column can
separate methane and the C2 species at ambient conditions. However, its usefulness is
limited by an upper temperature limit of 200°C, at which the >Cg HC can only be analyzed
isothermally. To utilize this column effectively, a three-column switching system incorporating
a 10-port valve (Valco, Austin TX) as shown in Figure 1 was devised. Columns 1A and 1B are
non-polar, methyl silicon capillary columns ( 0.18mm I.D., 0.4u DB-1, 15-m, and 25-m
respectively from J. & W. Scientific, Folsum, CA), and Column 2 is a 50-m 0.32mm I. D.
alumina PLOT column. The freeze-out loop, and Sample In and Out ports in the diagram may
represent any sample concentration device connected to these ports. In our case, a glass-
bead packed freeze-out loop for use with liquid Ar and a volume transfer apparatus (Grosjean
and Fung, 1984^) for assessing the freeze-out air volume were connected. Carrier 1 and 2
were pressure-controlled, and were set to obtain a linear velocity of ~40cm/sec. with
hydrogen. Restrictors were employed to balance the flow streams to avoid baseline
disturbances during valve switching.
The system was operated as follows: A sample aliquot was concentrated in valve
Position 1, and injected into the GC via switcning to Position 2 and heating the loop with
boiling water. With the oven held initially at 23°C, only the heavy HC components in the
sample entering Column 1A were retarded. The fast eluting species in Column 1A were
transferred to, and separated by Column 2 (PLOT). At the appropriate time, the valve was
returned to Position 1, placing Column 1A in series with 1B, thereby allowing the residual
(retarded) components in Column 1A to continue with the separation in Column 1B. Also,
another sample could be concentrated at this point while the analysis was still in progress.
To arrive the proper valve switching time, HC standards were analyzed isothermally
with Column 1A to determine the retention times of components, about which a cut was to be
made. Then the column was installed into the valve and tested again with standards to fine
tune the split time. The analysis was performed with a Hewlett-Packard 5890 Series II gas
chromatograph with dual flame ionization detectors. The GC's Valve Programming feature
258
-------
was used to control the operation of the valve. Data were processed with a dual channel
Shimadzu CR-4A integrator.
Detector response was calibrated using NIST SRM 1805, a 254 ppb benzene in
nitrogen. By altering the valve switch time, benzene can be directed to either the PLOT or DB-
1 column for that purpose. In analysis of ambient HC, it is assumed that all HC respond in the
FID in proportion to the number of carbon atoms, and concentrations of the components are
reported on a ppbC basis.
RESULTS and DISCUSSION
Figure 2 is a chromatogram of ambient air C2-C-J2 hydrocarbons analyzed using a 50-
m 0.32mm I.D. nonpolar (Chrompack CP-Sil 5 CB) capillary column with sub-ambient
temperature programming. The stationary phase was too thin to provide good 62 separation.
The peak at 4.04 min. represents unresolved ethylene and acetylene, and the peak at 4.24
min., ethane. The analysis time is ~50 min., typical for this approach.
With a 15-m short column and an ambient oven temperature, the peaks in the first 20
min. of the chromatogram in Figure 2 would elute unresolved. Compounds up to n-heptane
eluting off Column 1A were found to have transferred to the PLOT column when the valve
was held in Position 2 for 2.5 min. Then the valve was returned to Position 1, and the oven
temperature increased to resolve the components.
The precut after n-heptane represents an optimal choice because there is a wide gap
before the next significant peak, methylcyclohexane, elutes. There is little likelihood of peak
splitting making a cut at that point. Also, the cut allows the PLOT column to be utilized most
efficiently. Unlike a methyl silicon column, which separates compounds by boiling point, the
alumina column separates according to polarity. Branched alkanes of the same carbon
number elute closely as bundles, followed by the corresponding n-alkane, and then the
bundles of corresponding olefins. Benzene elutes after n-heptane. Therefore, unknown peaks
can be readily classify by class and carbon number. Extracting such information is not
possible from an analysis performed with a DB-1 column. To a photochemical modeler,
having unknown peaks classified provides valuable information about the sample, and is,
thus, much better than if the peaks are left unclassified.
The elution of acetylene apparently varies with the state of deactivation of alumina by
water in the sample. The compound has oeen found eluting after propylene to after n-butane.
Under the present analytical conditions, acetylene elutes mostly after n-butane, although
there had been occasions in which it was unresolved from n-butane. Reportedly, Chrompack
has some success in solving this problem by replacing the KCI in alumina with Na2S04.
Since both the DB-1 and PLOT columns were housed in the same oven,
experimentation was needed to arrive a temperature program for optimal elution of the
components concurrently. The analytical conditions were as follows:
Initial: 23°C, isothermal for 2.5 min.
Rate: 10°/min. to 40°
4°/min. to 100°
9°/min. to 195°. isothermal at 195° for 5 min.
Valve: 0 min. ON; 2.5 min. OFF.
259
-------
An ambient air sample from Oildale, CA was analyzed using this approach. The
resulting chromatograms are shown in Figure 3A and 3B. Except for the lack of olefins and
absence of acetylene, this samples, being from a area of petroleum production, is
considerably more complex than a typical ambient air sample. Over 250 peaks were detected.
Excellent resolution was evidenced by the partial resolution of m- and p-xylene. Analysis was
completed in approximately 30 minutes. The remaining few minutes of the temperature
program was to ensure the PLOT column starting the next run at a consistent state of
activation. The shortened analysis time is the direct result of being able to separate the
sample HC into the light and heavy fractions and perform the speciation of each fraction
simultaneously. In addition, by virtue of the valve position, cryogenic concentration of another
sample can occur while analysis is still in progress. This would be important if automated
analyses were to be performed on an hourly basis, as stipulated under the PAMS network.
The method has been used to analyzed standards and ambient air samples in informal
interlaboratory comparisons with several other laboratories, which included the USEPA Gas
Kinetics & Photochemistry Research Branch, and Desert Research Institute. The results were
in good agreement, indicating no deficiency existed with the method.
CONCLUSIONS
A method has been developed for the analysis of ambient air HC without the use of
sub-ambient temperature programming. The method showed comparable results to the
traditional methods, but reduced the analysis time significantly to approximately 30 min.
without sacrificing resolution and performance.
REFERENCES
1 W. A. Lonneman, S.L. Kopczynski, F.D. Sutterfield, "Hydrocarbon Composition of
Urban Air Pollution" Environ. Sci. Technol. 8: 229 (1974).
2. H. Westberg, W. Lonneman, W. Holdren, Identification and Analysis of Organic
Pollutants in Air: Ann Arbor Sciences: Ann Arbor, ML, 1984, pp 323-337
3. B. Dimitriades, D. E. Seizinger, "A Procedure for Routine Use and Chromatographic
Analysis of Automotive Hydrocarbon Emissions" Environ. Sci. Technol. 5: 223 (1971).
4. R. A. Rasmussen, H. H. Westberg, and M. Holdren, "Need for Standard Referee GC
Methods in Atmospheric Hydrocarbon Analyses" J. Chromatog. Sci. 12:80 (1974)
5. H. B. Singh, Guidance for the Collection and Use of Ambient Hydrocarbon Species
Data in Development of Ozone Control Strategies. EPA-450/4-80-008, U.S.
Environmental Protection Agency, Office of Air Quality Planning and Standards,
Research Triangle Park, NC, 1980, pp 29-34
6. D. Grosjean and K. Fung, "Hydrocarbons and Carbonyls in Los Angeles Air", JAPCA
34:537(1984).
7. F. D. Stump and D. L. Dropkin, "Gas Chromatographic Method for Quantitative
Determination of C2 to C-|3 Hydrocarbons in Roadway Vehicle Emissions", Anal.,
Chem. 57:2629(1985).
8. R. L. Seila and W. A. Lonneman, "Determination of Ambient Air Hydrocarbons in 39 US
Cities", in Proceedings of the 81st Annual Meeting. 88-150.8 Air Pollution Control
Association, Dallas, TX 1988
260
-------
FREEZE-OUT
LOOP
POSITION 1
FILL / ANALYZE
POSITION 2
INJECT/ PRECUT
Figure 1: Valve Configuration for Analysis of Ambient Air Hydrocarbons
I ,
Figure 2: Ambient Air Hydrocarbons Analyzed with Subambient Temperature Program.
261
-------
B
Figure 3: Analysis of Ambient Air Hydrocarbons Simultaneously Using an Alumina PLOT and
DB-1 Column: A - PLOT Column; B - DB-1 Column
-------
TESTING AND EVALUATION OF 2 PROTOTYPE DEVICES FOR
DIRECT MEASUREMENTS OF AIR-WATER TRANSFER PROCESSES
INVOLVING TOXIC CHEMICALS
W. Sohroeder, Environment Canada: AES, Downsview, Ontario M3H 5T4
D. Mackay and A. Cassamalli, University of Toronto
Y.-Z. Tang, Q. Tran and P. Fellin, Concord Environmental Corp.
S. Eisenreich and D. Achman, University of Minnesota
ABSTRACT
Among the environmental pollutants of concern in most jurisdictions
in North America are many volatile and semi-volatile organic compounds.
Their vapor-phase fractions partition themselves between the atmosphere
and other environmental compartments (water, soil, biomass). Determi-
nations of the magnitude and direction of the mass transfer fluxes of
toxic chemicals between the atmosphere and the hydrosphere constitute an
important topic in current scientific research. Two prototype devices
for direct measurements of air-water exchange processes have been built
and tested under laboratory conditions. One is a sparger device which
can be used to establish the (truly) dissolved concentration of a given
chemical in water, and hence its potential for diffusive transfer at the
air-water interface. The other device is a flux monitor with which the
chemical mass transfer rate from the water surface to the atmosphere (or
vice versa) can be measured. The paper describes the equipment as well
as the sampling and analytical methods development completed so far, and
presents initial results from laboratory testing of the two devices.
INTRODUCTION
The importance of atmospheric transport and deposition (wet &/or
dry) as environmental pathways for the input of persistent organic
contaminants known to occur in the Great Lakes ecosystem has been
recognized for more than a decade (Eisenreich et al.1; IJC2; Schroeder
& Lane3; Swackhamer & Hites4) . Aside from efforts dedicated to improving
our understanding of vapor phase—particle phase partitioning of semi-
volatile organic compounds in the atmosphere (Bidleman et al. ; Pankow6;
Lane et al. ) researchers in the Great Lakes scientific community (and
elsewhere) are acutely interested in advancing the current state of
knowledge surrounding the mass transfer of persistent organic substances
across the air-water interface (Mackay et al.8; Sproule et al.9;
McConnell et al.10) . Whereas wet and dry (atmospheric) deposition
processes are uni-directional (resulting in the transfer of chemicals
from the atmosphere to the water surface), air-water exchange processes
are bi-directional in nature (mass transfer being either "positive" or
"negative" by convention) with the magnitude and direction of the net
chemical "flux" determined by various parameters, including: (a)
physico-chemical properties of the substance (e.g. vapor pressure and
water solubility — whose ratio make up the Henry's Law coefficient);
(b) the relative concentrations of the chemical in the water and air
compartments (i.e. the degree of "oversaturation" or "undersaturation"
in a given compartment), and (c) environmental conditions (e.g., air/
water temperature, wind speed/turbulence, nature and extent of bubble
formation, presence or absence of a microlayer of surface film) which
influence water-side or air-side mass transfer coefficients. To
understand, and ultimately model or predict, the movement and eventual
fate of toxic chemicals released into the environment (Schroeder &
Lane3) , it is essential that we advance our knowledge of air-water
exchange processes, especially the rates of deposition, absorption,
diffusion, and evaporation/volatilization.
263
-------
Estimation, or (even better) direct determination, of the magnitude
and direction of the fluxes of persistent organic pollutants between the
atmosphere and surface waters (oceans, lakes, rivers, reservoirs,
impoundments) represent an important topic in contemporary scientific
research. On the many lists of priority toxic substances drawn up by
various levels of government and other organizations/institutions or
judicial bodies in the United States, Canada, and other industrial
nations, there are numerous volatile and semi-volatile organic compounds
(VOCs & SVOCs) including chlorinated alkanes/alkenes/benzenes, organo-
chlorine pesticides, and polychlorinated biphenyls (PCBs). Many of
these persistent chemicals are present in natural waters, including
those of the Great Lakes.
To address research goals and objectives outlined in the 1987
Protocal to the U.S.- Canada Great Lakes Water Quality Agreement (Annex
15: Airborne Toxic Substances), a collaborative project involving
government, university, and private sector scientists from Canada and
the United States has been initiated for the purpose of studying air-
water exchange processes involving priority toxic chemicals. In this
project, research is being conducted on the development, evaluation, and
application of two types of experimental apparatus with the potential
for making direct (in situ) measurements of air-water exchange processes
and transfer rates. Two devices — a sparger and a flux monitor — were
built and have been tested under controlled laboratory conditions. This
paper describes the prototype equipment and presents results from
sampling and analytical methods development, as well as laboratory
testing and evaluation of the equipment. The target compounds
tentatively selected for the first phase of our project, namely
chloroform, 1,3-dichlorobenzene, alpha- and gamma-hexachlorocyclohexane
(lindane), and 2,4,6-trichlorobiphehyl, constitute a sub-set of the
compounds of interest for which Henry's Law coefficients have been
provided in Table I.
EXPERIMENTAL SECTION
Equipment and Design Considerations
For the first phase of this project, the sparger apparatus
described by Sproule et al.9 was re-designed to promote: (a) improved
water exchange characteristics (through the flow holes), and; (b) better
internal circulation of bulk liquid so as to prevent significant
depletion of the target chemical(s) from the water column being
contacted ("equilibrated") with the sparging gas (air). Figure 1 shows
the prototype sparger device which was fabricated from virgin grade
Teflon" (to minimize sorption/loss of chemicals on the surfaces of the
construction material) and was subsequently evaluated in the laboratory
(at the University of Toronto). The exit stream passed through a glass
tube containing an adsorbent bed of Tenax T.A. which efficiently retains
the chlorinated organic target compounds.
A flux monitor was designed and constructed for this project by
Concord Environmental Corporation (Toronto). This device is intended to
be used, along with the sparger, in future investigations of mass
transfer of toxic chemicals across the air-water interfacial boundary.
Design criteria established for this device were: (a) large collection
area, small internal volume, and low dilution flow rate so as to
optimize the sensitivity of the system; (b) efficient mixing of target
compounds with purge gas (air) in the monitor chamber; (c) good
stability on an open water surface under moderate wind and wave
conditions; (d) minimal chemical losses on interior surfaces of the
monitor; (e) minimal perturbance of environmental variables such as
light intensity/spectral distribution, temperature, and barometric
pressure; (f) easy to interface with integrative sampling devices
(adsorbent tubes).
264
-------
The prototype unit of the flux monitor is depicted in Figure 2. It
consists of a tubular frame made of 1.27 cm (1/2") schedule 80 PVC pipe
fastened with metal Kee Klamps". The top surface is a 1.27 cm thick
transparent cast acrylic sheet with a hole drilled in each corner for
fitting the PVC pipe. The acrylic sheet is held in place by hose
clamps, which allow the height of the monitor to be adjusted. Attached
to the outside of the 1.27 cm thick acrylic skirt were four floats
("boat fenders") to provide buoyancy. The monitor chamber consists of
a clear Mylar" film which is attached to the acrylic skirt at the bottom
and the acrylic sheet at the top. The truncated pyramid-shaped monitor
covers a surface area of 0.194 m2 and contains a volume of 24.3 L. Ultra
zero air (Matheson Gas Products Canada) was used as the purge gas.
RESULTS AND DISCUSSION
Laboratory Tests
Sparger. Tests were performed with this apparatus in a Nalgene"
tank containing approximately 700 L of water spiked with
monochlorobenzene to a concentration of 10 micrograms/liter. Sparging
gas (ultra zero air) flow rate was 0.3 L/min for experiment durations of
about 30 minutes. Samples were thermally desorbed from the trap into a
gas chromatograph (HP 5890 equipped with a flame ionization detector)
using an Environchem Unacon Series 810 thermal desorption system.
Helium (UHP grade) was the carrier gas. A typical gas chromatogram is
reproduced in Figure 3.
Flux Monitor. Tests were conducted at Concord Environmental Corp.
with 4 target compounds added to a plastic pool containing approximately
100 L of water. The concentrations of the target compounds were:
Chloroform:
Lindane:
258 ng/L
687 ng/L
Dichlorobenzene:
2,4,6 - PCB:
14.8 Jig/L
211 ng/L
After spiking and stirring the water,the flux monitor was placed on the
water surface and a purge flow of 1.1 L/min was used to flush the
chamber. The exit air was split into 2 streams, each of which was drawn
through an adsorbent tube (packed with 300 mg sequential plugs of 20/40
mesh Carbotrap" B and Carbotrap" C in 4 mm (i.d.) x 11.5 cm glass tubes)
at a sampling flow rate of about 0.2 L/min for 60 minutes. Recovery of
target compounds from exposed adsorbent tubes was performed either by
solvent extraction, or by thermal desorption (using a Dynatherm
Analytical Instruments, Inc. model 850 desorber and model 851
temperature controller). Three consecutive sets of duplicate samples
were obtained. A representative gas chromatogram of a pentane standard
solution containing four target compounds is shown in Figure 4.
Sampling and Analytical Methods Development
The target compounds selected span a wide range of volatility, from
extremely volatile (CHC13) to semi-volatile compounds (e.g., PCBs).
Collection on dual adsorbent-bed tubes was considered to be the method
of choice, since few (if any) adsorbents quantitatively retain compounds
with such a broad span of volatility. For recovering target chemicals
from the adsorbents, eight solvents were compared: pentane, hexane,
benzene, toluene, cyclohexane, carbon disulfide, and mixtures of toluene
and pentane or hexane. Pentane proved most effective. Detection limits
(for the solvent extraction - gas chromatography - electron capture
detection method) and pentane extraction efficiencies are given in Table
II. The sensitivity of this method was found to be insufficient for
determination of the target compounds at concentrations likely to be
encountered in field samples. Hence the analytical method was refined
by replacing solvent extraction with a thermal desorption technique
(resulting in a method sensitivity increase of nearly 500 times).
265
-------
CONCLUSIONS
Two distinctly different prototype devices — for application in
future field investigations of air-water exchange processes involving
toxic organic substances known to be present in the waters of the Great
Lakes — have been built, tested and evaluated under laboratory
conditions during the first phase of this research project. These tests
have indicated some areas where improvements can be made, either to the
prototype devices or to the sampling, sample handling/preparation and
analytical methods which are to be carried forward to the next stage of
this project (viz., field testing and intercomparison of the 2 devices).
REFERENCES
1. S.J. Eisenreich, B.B. Looney and J.D. Thornton, "Airborne organic
contaminants in the Great Lakes ecosystem," Environ. Sci.
Technol.15:30(1981).
2. IJC, Summary Report of the Workshop on Great Lakes Atmospheric
Deposition. International Joint Commission, Windsor, Ont., 1987.
3. W.H. Schroeder and D.A. Lane, "The fate of toxic airborne
pollutants," Environ. Sci. Technol.22:240(1988).
4. D.L. Swackhamer and R.A. Kites, "Occurrence and bioaccumulation of
organochlorine compounds in fishes from Siskiwit Lake, Isle Royale,
Lake Superior," Environ. Sci. Technol.22;543(1988).
5. T.F. Bidleman, W.N. Billings and W.T. Foreman, "Vapor-paricle
partitioning of semivolatile organic compounds," Environ. Sci.
Technol.20:1038(1986).
6. J.F. Pankow, "Review and comparative analysis of the theories on
partitioning between the gas and aerosol particulate phases in the
atmosphere," Atmos. Environ.21:2275(1987).
7. D.A. Lane, N.D. Johnson, S.C. Barton, G.H.S. Thomas and W.H.
Schroeder, "Development and evaluation of a novel gas and particle
sampler for semivolatile chlorinated organic compounds in ambient
air," Environ. Sci. Technol.22:941(1988) .
8. D. Mackay, S. Paterson and W.H. Schroeder, "Model describing the
rates of transfer processes of organic chemicals between atmosphere
and water," Environ. Sci. Technol.20:810(1986).
9. J.W. Sproule, W.Y. Shiu, D. Mackay, W.H. Schroeder, R.W. Russell
and F.A.P.C. Gobas, "Direct in situ sensing of the fugacity of
hydrophobic chemicals in natural waters," Environ. Toxicol.
Chem.10:9(1991).
10. L.L. Me Connell, W.E. Gotham and T.F. Bidleman, "Gas exchange of
hexachlorocyclohexanes in the Great Lakes," Environ. Sci. Technol.
In press (1993).
11. Handbook of Chemical Property Estimation Methods; W.J. Lyman, W.F.
Reehl and D.H. Rosenblatt, Eds., McGraw-Hill Inc., New York, 1982.
12. D. Mackay and W.Y. Shiu, "A critical review of Henry's Law
constants for chemicals of environmental interest," J. Phvs. Chem.
Ref. DatalO:1175(1981).
13. W.Y. Shiu, University of Toronto, Toronto, Ontario, personal
communications, 1993.
14. F.M. Dunnivant, A.W. Elzerman, P.C. Jurs and M.N. Hasan, "Quantita-
tive structure-property relationships for aqueous solubilities and
Henry's Law constants of polychlorinated biphenyls," Environ. Sci.
Technol.26;1567(1992).
266
-------
Table I. Henry's Law coefficients for compounds of interest.
Compound
Trichloro-
methane
(Chloroform)
1,3 - dichloro-
benzene
(m-xylene)
1,4 - dichloro-
benzene
(p-xylene)
2,4,4' - tri-
chlorobiphenyl
(IUPAC no. 28)
2,4' ,5 - tri-
chlorobiphenyl
(IUPAC no. 31)
2,4,6 - tri-
chlorobiphenyl
(IUPAC no. 30)
^L- hexachloro-
cyclohexane
o - hexachloro-
cyclohexane
H @ 25°C
(Pa mVmol)
320
283 (20°)
376
160
32
13
67
0.771"
0.532 (20°)
0.356b
0.261 (20°)
Reference
Lyman et al. 11
Mackay & Shiu 12
Shiu 13
Mackay & Shiu 12
Dunnivant et al .
Dunnivant et al .
Dunnivant et al.
McConnell et al.
McConnell et al.
14
14
14
10
10
a: calculated from log H =
b: calculated from log H =
2810/T +
2382/T +
9.31
7.54
Table II. Detection limits and solvent extraction efficiencies
for the pentane extraction - gas chromatography-electron
capture detection method.
Compound
Chloroform
1,3-
Dichlorobenzene
Lindane
2,4-6-
Trichlorobiphenyl
Detection Limit
pg/injection
5
50
1
1
Extraction
Efficiency
(%)
95
86
71
82
267
-------
AIR OUT
TO SORBENT
bCREW K
TEFLON®-*
FLovy.
Mni_e^^
*-»-»»-•
^x^>
ss
ru
0°
°0
FLOTXTIOM
±=3^7 COLLAR.
0^>->
\
PI
AC*
_ SKM
n«-fR"
f
A\fUH
A10 '" °UT
AIR. 4 f
MYLAR*
/FILM
f
PVC
45 CM
Figure 2. Prototype flux monitor.
(Boat fenders not shown).
Figure 1. Prototype sparger.
GC conditions
Column = DB-5 (30 m X 0.32 mm)
Inject, temp. = 300°C
FID temp. = 300°C
Temp, progr. = 65°C + 10°C/min
for 4 minutes
H
CO
| IL
8 " II 16 min
Figure 3. Chromatogram of sample
obtained by sparging 10 L of air
through water containing 10 pg/L
of monochlorobenzene.
("UNK." = unknown chemical).
GC conditions_
Column = DB-17 (30 m x 0.25 ram)
Inject, temp. = 250°C
BCD temp. = 350°C
Carrier (He) = 1.6 mL/min
Temp, progr. = 40°C for 3 min
10°/min to 250°C
Figure 4. Gas chromatogram showing
separation of four target compounds:
chloroform, 1,3-dichlorobenzene,
1 indane, 2,4,6, -trichlorobiphenyl.
(S = solvent (pentane) peak).
268
-------
A UNIVERSAL AIR PRECONCENTRATOR FOR AUTOMATED ANALYSIS
OF AMBIENT AIR, STACK GAS, LANDFILL GAS, AND AUTOMOBILE
EXHAUST USING GC/MS METHODS
Daniel B. Cardin, John T. Deschenes
Entech Laboratory Automation
950 Enchanted Way, #101
Simi Valley, CA 93065
ABSTRACT
Today's Air Analysis laboratory is faced with the challenge of having to analyze a wide
range of sample types and concentrations with limited analytical instrumentation. Most
laboratories do not have a large enough sample load to dedicate a preconcentrator and GC/
MS to a particular sample matrix. This can result in the need to replumb, or reoptimize the
system when switching between ambient and source level samples, or when sample matrices
are high in CO2. In addition, polar compounds may have a different optimum setup from
that of the less polar TO14 compounds. Finally, even after reoptimizing for ambient level
samples, each position of an autosampler manifold would have to be verified as clean before
they could reliably be used for ppt-ppb level work once ppm level VOCs have been allowed
to come in contact with surfaces and fittings.
A single preconcentration system has been developed that can perform analyses of high
(ppm) and low (ppt-ppb) level samples using 16-position automation without the possibility
of system contamination. Totally separate inlet systems are utilized to prevent ambient air
samples from being exposed to surfaces that have contacted high concentration samples.
Absorptive surfaces, such as Nafion dryers and Viton elastomeric valve sets and o-rings have
been completely eliminated from the sample path to further reduce cross-contamination.
Software selection of 1,2, or 3 stage preconcentrations can optimize performance for high or
low level CO2 matrices without any replumbing. Both polar and non-polar compounds are
quantitatively concentrated and delivered into a GC/MS while removing most of the water
and CO2.
SYSTEM DESCRIPTION
The Entech 2000 Preconcentrator was designed to be both modular and software
configurable to permit the analysis of the wide range of sample concentrations and matrices
found in air monitoring. Up to three traps can be installed to concentrate samples, manage
water and CO2, and then focus the concentrate at the head of the analytical column for high
performance GC or GC/MS (FIGURE 2). Currently supported trapping module configura-
tions are as follows:
269
-------
Module 1
Nafion Dryer
Bypass
Glass Bead Trap
Glass Bead Trap
Adsorbent trap
Module!
Glass Bead Trap
Glass Bead Trap
Glass Bead Trap
Low Temp Adsorbent
Adsorbent trap
Module3
On-column focuser
it
ii
ii
Bypass
Each of these trapping procedures can optimized VOC preconcentrations based on technical
as well as practical considerations. For example, when water management is not as impor-
tant (as with GC/FID), a bypass followed by a glass bead trap is sufficient. For heavier,
thermally stable compounds such as BTEX, a single tenax trap operated without focusing or
cryogen will work very well. Cryogenless preconcentration of C2-C10 hydrocarbons and
most of the Air Toxic compounds can be performed using 2-stage adsorbent trapping, first on
a large diameter trap followed by further volume reduction on a narrower trap for faster
desorption and better chromatography.
One hardware configuration listed can perform most of these trapping procedures just by
changing how the preconcentration is allowed to proceed under software control. By install-
ing a glass bead trap in the first stage and a tenax trap in the second stage, the system can be
instructed to either trap initially on glass beads, or alternatively directly on the tenax trap by
keeping the glass bead trap at 60 deg. C. To extend the trapping range of the tenax trap to
include light VOC's, the temperature can be lowered ambient to -50 deg. C, at which tem-
perature vinyl chloride has over a 500 cc breakthrough volume. Optimum configurations for
various sample types and matrices will be discussed individually in greater detail.
FLEXIBILITY THROUGH SOFTWARE CONTROL
The 2000 preconcentrator software allows selection of preconcentration parameters by se-
lecting a particular application. Applications available to the user are, in turn, dependent on
their current hardware configuration. In addition to the selection of trapping modules listed,
the user can indicate in the software which inlets or autosamplers are currently connected to
the 2000 preconcentrator (FIGURE 2). Currently available inlets include:
2016CM 16-Position Tower Canister/Bag Autosampler
2016BCM 16-Position Benchtop Canister/Bag Autosampler
2016LM 16-Position Tower Loop-Injection Autosampler
2016BLM 16-Position Benchtop Loop-Injection Autosampler
2001SSI Single Canister Sample Inlet
TD16A 16-Position Thermal Desorber
Two of these inlets can be connected to the 2000 preconcentrator simultaneously and se-
270
-------
lected through software. Therefore, a system can have both an ambient air inlet and a source
level loop-injection inlet to run both sample types without replumbing when changing back
and forth. In fact, it is possible to load up both autosamplers and switch from one to the
other automatically. In both cases, benchtop versions are usually preferred only when space
is critical (mobile lab) or when the maximum sample number to be run unattended will be
low.
POLAR AND NON-POLAR VOC'S AT PPT - PPB LEVELS
Three stage preconcentrations techniques utilizing this configuration have been described
previously (Automated 2-Dimensional Chromatography and Microscale Purge and Trap (1)).
MP&T, or Microscale Purge and Trap has been found to allow the analysis of all polar and
non-polar VOC's tested thus far. With MP&T, the VOC's, H2O, and CO2 are all initially
concentrated on the glass beads in the first stage. Up to a lOOOcc sample volume can be
preconcentrated resulting in detection limits of 5-20ppt on some of the newer high perfor-
mance quadrupole GC/MS systems (FIGURES 3, 4, & 5). Due to the low affinity of the
VOC's on glass beads at room temperature, these VOC's can be transferred out of the first
stage by desorbing at room temperature. Most of the water is therefore left behind, just as
most of the water remains in the sparge vessel when performing Purge & Trap on water
samples. Retrapping the VOC's on the tenax trap at reduced temperatures allows the CO2 to
pass through while retaining even the lightest TOW target analytes. Subsequent on-column
focusing and desorption yields maximum detection limits for the entire range of VOC's.
Since CO2 is mostly removed, cryogenic cooling of the GC oven is not required to increase
the separation of the otherwise massive CO2 peak and the first eluting TO 14 compounds.
This results in a tremendous savings of liquid nitrogen.
TRACE VOC'S IN HIGH CO2 MATRICES
When it's necessary to perform preconcentrations to detect VOC's at the low ppb level in
matrices consisting of 1-50% CO2 (soil gas, landfill gas, auto exhaust, etc.), straight cryo-
genic trapping on glass beads will not achieve optimum results because the CO2 will concen-
trate right along with the sample. Cryogenically preconcentrating 50cc of a sample contain-
ing 50% CO2 would result in 25cc of CO2 being expanded into the GC/MS at a carrier flow
rate of 1-2 mis a minute. The resulting pressure burst will actually expand the CO2 back into
the GC carrier gas regulation devices requiring up to several hours to bring CO2 backgrounds
down to acceptable levels. Using the universal 3-stage configuration, a "High CO2 Sample"
application can be selected which maintains the first trap at 60 deg. C to force initial
preconcentration to occur on the tenax trap at reduced temperatures. CO2 passes through the
trap much faster than the targeted VOC's and is further eliminated by flushing the trap with
an additional 50-100cc helium. Back-flushing of the tenax trap followed by optional focus-
ing can then be performed.
271
-------
HIGH CONCENTRATION SAMPLES (SOURCE)
Quantification by GC/Quadrupole MS optimally requires 0.1 to 1000 ng of material to be
injected. When sample concentrations reach 1-10,000 ppm, this quantity can be obtained
using loop injection volumes of approximately . 1- Ice. In order to keep these high concen-
trations away from the ambient air portion of the system, a separate 16-position autosampler
is available that only flushes the contents of the loop into the preconcentrator (FIGURE 6).
Transfer of the typical O.Scc sample volume through connective transfer lines and fittings
into the second trap is performed at 60cc/min resulting in sample residence times in fittings
and valves of approximately 0.5 seconds. This isn't enough time for the sample to diffuse
into the dead volume regions of connective fittings, so long periods of flushing to regain a
clean system become unnecessary. Using a separate autosampler also means that each indi-
vidual position won't have to be revalidated for cleanliness from source level to ambient
level analysis, as would be the case if a single autosampler was used for both. A graph
showing reproducibility of the 2016LM Loop-injection Manifold is show in figure 7.
SCREENING CANISTERS OF UNKNOWN CONCENTRATION
To prevent unknowingly exposing the ambient air inlet to high VOC concentrations, canisters
of uncertain concentrations can be prescreened using the loop-injection autosampler. Using a
single isothermal GC/MS run at 180-200 deg. C, each sample on the autosampler can have an
aliquot introduced sequentially onto the GC/MS. Total area counts can then be compared to
determine optimum sample volumes for analysis.
CONCLUSIONS
A single preconcentration system has been developed that allows rapid switching between
different air matrices. Three different autosamplers are available for TO14 (2016CM),
source level (2016LM), and TO1/TO2 thermal desorption (TD16A). Two autosamplers can
be interfaced simultaneously permitting software selection of the active manifold. A single
trapping configuration also allows for optimizing the preconcentration procedure to accom-
modated polar and non-polar VOC's in high or low CO2/H2O matrices. Menu-driven soft-
ware is used to select the preconcentration procedure based on the user selected application.
Default parameters are provided which can be further optimized based on the GC/MS con-
figuration utilized.
272
-------
ENTECH 2000 PRECONCENTRATOR
BLOCK DIAGRAM
Sample
From
Manifold
Figl
Classical Method TO 14 Pathway
UNIVERSAL INLET SYSTEM
USING 1 PRECONCENTRATOR
WITH SEPARATE AUTOSAMPLER
TCH4-Amblent Air
Indoor Air
Ozone Precursers
Fence Une
Fig 2
273
-------
: S«t lUr 30 93 03:33:41 PH ulna; Acqltattotf /ch
: 09*72
«/«trt«Qi L/n,
VJ"WO)
Kloc Info i
Vial
NtoUKt o - Keconsmjcrea
Ion Chromatogram of a
0.2ppb TO 14 standard.
Mlcroscale Purge and Trap
water/CO2 management
was used allowing analysis
of both polar ana non-polar
VOC'S
FIGURE 4 -Selected Ion
Chromatogram (m/z 50)
of chloromettiane at
0.2 ppb.
FIGURE 5 -m/z 91
Selected Ion Chromato-
gram showing toluene (1),
ethylbenzene(2), m+p-
xylene(3) and o-xylene(4)
at0.2ppb.
3300
3000.
4300
*aoo
3900.
3000
3300
1000
1300
IOOO
300
J
uuivunc.
• 3000
•0000
73000
70000
•3000
60000
95000
90000
49000
40000
39000
30000-
2SOOO •
20000
13000-
10000 •
3000-
0-
!!••—> 3.
ill
il
^
10
1
1
1
FIGURE 4
Chloromethane - 0.2ppb
rL— " i-«" • -« '.oo i.oo ».jo
,1
1
10.00
TIC: (BS81JKDLO
II
1
1,
IWuJ
717
^
5 '
%l
/
3»000
3«oo
14000
12000
30000.
2*000 .
aaooo
24000.
12000.
30000
11004 •
noco
14004
13000
1QOOO
•000
•000
4000
3000
FIGURE 5
1
3
2
4
274
-------
2016LM LOOP INJECTION AUTOSAMPLER
SWEEP PURGE GAS
SAMPlfSEECrVALVE
INTERNAL STANDARD
FIG 6
LOOP INJECTION DATA
100000
90000 -
60000 -
70000 -
| 60000-
O 50000 -
j£ -10000 -
30000
20000-
10000 -
0
8:09 8:27
Time
FIG 7
8:45 9:08 9:25 9:43 T0:01
cmpd % RSD
_ AN 1-88
MMA 1.61
0 4VCH 1.93
EB 1.33
, _ _ STY 1.93
X CUM 1 .82
— AMS 1.76
TD16A DIAGRAM
FIGS
*Xl/4«»mOTUlE T-JMWOimOTUBE VOSTSf*3l£ TUBE VOST DUAL TUBE
275
-------
The Development and Evaluation of a Transportable
Fast Gas Chromatograph for the Monitoring
of Organic Vapors in Air.
Jesus A. Gonzalez, Ph.D.
University of Puerto Rico
School of Public Health
G.P.O Box 5067
San Juan, Puerto Rico 00936.
Steven P. Levine, Ph.D., c.I.E.
University of Michigan
School of Public Health
Ann Arbor MI 48109-2029
Richard E. Berkley, Ph.D.
U.S. EPA (AREAL)
79 T.W. Alexander
RTF, NC 27709
ABSTRACT
Gas Chromatography has the potential to be a real-time or near
real-time monitoring method for organics in air. A transportable fast
GC with FID/ECD and PID/ECD configurations has been developed.
Preliminary evaluation has shown that all design features and
improvements of the instrument worked successful.
INTRODUCTION
Gas chromatography is often used for ambient air monitoring.
However, its usefulness for this purpose can be limited by retention
times of several minutes and by the limits of detection.This has
heightened the need for the development of high speed monitoring
systems. Applications requiring real-time or near real-time capabilities
include emergency response activities, stationary-source plume tracking,
and fence line monitoring. In order to meet this requirement, a fast GC
system has been developed and evaluated (1-5).
The most important component of the fast GC is the high speed inlet
system. The electrically heated, nitrogen cooled inlet system produces
significantly narrower injection bands than can be accomplished using
alternate technologies. The retention times of components of mixtures
are thereby reduced to just few seconds. This is 50 to 100 times shorter
than can be otherwise accomplished. Although part-per-billion (ppb)
level limits of detection (LOD) are achieved using this system with only
a 1.0 cm3 sample volume, collection and preconcentration of contaminants
from relatively large volumes of air is possible.
Recently, the design and evaluation of a laboratory-based fast GC
has been reported. These reports included information on instrument
design and the LODs for 41 EPA target analytes (4) . The applicability of
fast GC was also tested for the analysis of components of complex
mixtures of organic vapors in ambient air (5). In that study, mixtures
of up to 34 components were separated in less than 100 second, and up to
1000 effective theoretical plates per seconds was achieved.
276
-------
Additional requirements such as portability have prompted the need
to incorporate new improvements into the fast GC. These efforts have
resulted in the development of a transportable version of the fast GC.
This paper presents the design, development and preliminary evaluation
of the transportable fast-GC.
METHODS
The goal of the design of the transportable fast-GC is to maintain
performance and improve transportability and the logistics of operation.
To reach this goal, several significant improvements were made when
compared to the laboratory-based fast GC. The transportable fast GC was
built by adding a high-efficiency cold trap inlet, a high speed
electrometer, and a laptop personal computer based data system to an HNU
Systems model 311 transportable Gas Chromatograph. A Schematic diagram
of the transportable fast GC is presented in Figure 1. The following is
a discussion of the changes made on the transportable fast GC compared
to the laboratory-based fast GC.
Inlet System
As shown in Figure 1, the inlet system for the transportable fast
GC includes a six port valve which is mounted inside of the oven. This
allows the sample loop to be thermostated without the need for an
additional valve oven. An injector port is also mounted on the top of
the oven for injection of gas or liquid samples via syringe. The outlet
of the injector was connected to a 50 cm long deactivated fused silica
buffer column which is used as a transfer line that carried the sample
to the cold trap. Since the sample is actually injected by the heating
of the cold trap, the presence of dead volume in the transfer line does
not have an adverse effect on instrument resolution.
The cold trap was fashioned from Monel 400 capillary tubing which
was enclosed in a small Teflon chamber and cooled by a flow of cold
nitrogen gas. For the laboratory fast GC system, the inlet system was
built outside of the oven. The sample was injected through the six port
valve to the buffer chamber, and then to the cold trap chamber, and then
condensed inside the cold trap.
In the laboratory system the buffer chamber/cold trap chamber
assembly was made of aluminum, Teflon and copper, and had to be heated
to about 75 °C (depend on sample boiling points) to prevent sample
condensation in that area. Also, at each end of the cold trap chamber,
the trap tubing was clamped between two copper blocks which served as
electrical contacts during the heating cycle. The copper blocks were
heated to 100 °c with 150 W heating cartridges to prevent sample
condensation outside the cold trap.
Comparing these two systems, it is clear to see that the inlet
system in the transportable fast GC system is not only a factor of about
five smaller, but also far more efficient, simpler and more reliable
than the laboratory system.
Cooling system
The basic problem with the laboratory fast GC cooling system was
that it was big, immobile, and inefficient. The design of the
277
-------
transportable fast GC optimized and minimized the use of liquid
nitrogen.
In the old cooling system the liquid nitrogen was stored in a 5 L
wide mouth Dewar. The nitrogen gas was cooled by running it through a
coil of copper tubing immersed in liquid nitrogen, and sprayed into the
Teflon cold trap chamber. The trap temperature was controlled by
adjusting the nitrogen pressure and was monitored with a thermocouple.
For the transportable fast GC cooling system, the trap temperature
can be cooled from room temperature to -100 °C in 15 minutes. Ten liters
of liquid nitrogen will last at least 8 hours, and as long as 16 hours,
of operation. The trap temperature was controlled by a temperature
controller and could be stable at setting temperature with +/- l °c
without adjustment.
For the old system, cooling down the trap temperature from room
temperature to -100 °c took at least 45 minutes. Five liters of liquid
nitrogen lasted at most for 3 hours, with the use of large amounts of
compressed nitrogen gas from the gas cylinder.
Thus, the new design resulted in a size and supply use rate
reduction of an order of magnitude, and a very high degree of
improvement in reliability and accuracy of cooling.
Gas Supply system
For the transportable fast GC system, the gas supplies such as
hydrogen, helium, and argon with 5% methane were supplied in small
rechargeable cylinders which mounted in the lid of the GC. The refilled
gas cylinder could operate at 30 ml/min as carrier gas for at least 16
hours. A small air pump was built on the instrument which could produce
a flow rate of 30 ml/min for use with a flame ionization detector. These
replaced two 1A cylinders. The installation of the air pump in the
system was one of the ways to minimize the size of the instrument. It
also represented an improvement in ease of instrument use.
Electrometer System
The high speed electrometer-amplifiers with a response time of 5 ms
were build for both FID and ECD in transportable fast GC system. These
were developed specifically for this application. A single electrometer
board was used for both detectors. This board replaced two free-standing
electrometer modules that were used on the laboratory instrument. This
reduced the size of the electrometer/amplifiers for the two detectors by
at least an order of magnitude.
Data Collection
For the transportable fast GC system, data were collected using an
80386-SX laptop computer equipped with 120 MB hard disk, 8 mb of RAM and
an A/D board. The use of a computer with 8 mb of RAM allowed all
software and data from an individual chromatographic run to be loaded
into RAM disks, thus reducing the data analysis time.
For the preliminary studies being performed on the transportable
fast GC, the system was equipped with both a standard HNU flame
ionization detector (FID) and an ECD. An HNU Systems-Nordien ECD was
used which had a cell volume of 90 microliters. An alternate
278
-------
configuration of the system that has been used was a photo ionization
detector (PID) and the BCD. The PID is an HNU low-volume cell detector
(40 microliters). A 30 meter, 0.25-mm i.d., 0.1-um bonded methyl
silicone stationary phase capillary column (Quadrex) was used for these
tests. Hydrogen was used as the carrier gas for the PID and FID,
operated at average linear velocities of 60 cm/sec. Make-up gas for the
PID was used at a flow rate of 65 ml/min. The oven temperature was
100 °C.
Figure 1. Schematic of the transportable fast GC system.
(A) Six-port rotatory valve; (B) Injector; (C) Buffer column; (D) Heated copper electrodes; (E) Cold
trap and chamber; (F) Capillary column; (G) Detector; (H) Fast electrometer; (I) A/D and D/A converter;
(J) Capacitor discharge power supply; (K) Laptop computer; (L) Liquid nitrogen Dewar; (M) Heater.
279
-------
FID
13
-vxiyV
15
60
70 §0 9(5"
Retention Time (seconds)
FID
60
70 80 90
Retention Time (seconds)
100
Figure 2. Chromatogram of fifteen-component mixture. The peaks identities are: l.
Benzene(PID only); 2. Toluene; 3. 1,2-Dibromomethane; 4. Tetrachloroethene; 5. Chlorobenzene; 6. Ethylbenzcne; 7.m,p-xylene; 8.
Styrene; 9. o-Xylene; 10. 4-Ethyl toluene; 11. 1,3,5-Trimelhylbenzene, 12. 1,2,4,-Trimethylbenzene, 13. Benzyl chloride, 14. m,p-
Dichlorobenzene; 15. o-Dichlorobenzene.
280
-------
RESULTS AND DISCUSSION
Figure 2 presents the chromatograms of a fifteen-component mixture
of organic vapors in air analyzed with the laboratory fast GC/FID (lower
tracing) and the transportable fast GC/PID (upper tracing). As can be
seen from these two chromatograms, the transportable fast GC/PID
demonstrated similar performance as the laboratory fast GC/FID for this
type of mixture.
However, the PID has evidenced two difficulties as a potential
substitute for the FID. A negative peak is obtained in a retention time
where there may be interferences with the analysis of analytes eluting
just before benzene. This is not observed with the FID. The second
problem is that detectable background contaminants in the system
accumulate with time in the trap. These contaminants are reinjected in
the column and interfere with analytes that coelute with them. In order
to solve this problem, the trap must be cleaned just before introducing
the sample. This strategy represents a ten second delay in the analysis
of the sample.
CONCLUSIONS
In this study, the applicability of the transportable fast GC with
the PID/ECD configuration was tested for the analysis of components of
a complex mixture of organic vapors in ambient air. The separation of
these analytes correlate well with separation obtained with the
laboratory fast-GC, but in a more efficient and convenient way.
REFERENCES
1. Mouradian, R.F; S.P. Levine, R.D. Sacks, "Evaluation of a
nitrogen-cooled, electrically heated cold trap inlet for high speed gas
chromatography," J.Chromatogr.Sci. 28: 643 (1990).
2. Mouradian, R.F.; S.P. Levine, R.D. Sacks, M.W. Spence," Measurement
of organic vapors at sub-TLV concentrations using fast gas
chromatography," Am. Ind. Hyg. Assoc. J. 51 (2): 90 (1990).
3. Mouradian, R.F.; S.P. Levine, HQ. Ke, " Measurement of volatile
organics at part per billion concentrations using a cold trap inlet and
high speed gas chromatography," J. Air Waste Manag. Assoc., 41:
1067-1072 (1991).
4. Ke, HQ.; S.P. Levine, R.F. Mouradian, R. Berkley," Fast gas
chromatography for air monitoring: limits of detection and
quantitation," Am. Ind. Hyg. Assoc. J. 53: 130-137 (1992).
5. Ke, HQ.; S.P. Levine, R. Berkley," Analysis of complex mixtures of
vapors in air by fast-gas chromatography," J. Air Waste Manage. Assoc.
42:146-152 (1992).
281
-------
A CRYOGENLESS AUTOGC SYSTEM FOR ENHANCED OZONE
MONITORING USING A SIMPLIFIED, SINGLE DETECTOR APPROACH
Daniel B. Cardin, John T. Deschenes
Entech Laboratory Automation
950 Enchanted Way, #101
Simi Valley, CA 93065
ABSTRACT
A network of real-time autoGCs for C2-C10 hydrocarbon analysis will be implemented over the
next 5 years to better correlate levels of ozone and C2-C10 "ozone precursors" in non-attainment
areas. These instruments will be operated by state, county, and city agencies with limited
personnel and little or no previous exposure to high performance gas chromatography. To insure
the collection of meaningful data, it will be necessary to reduce the maintenance requirements
and make these systems as reliable and simple to operate as possible.
A system is described that simplifies the cryogenless preconcentration and analysis of the C2-
C10 "ozone precursor" hydrocarbons. The incorporation of a dual sorbent trapping system
reduces sample moisture withoutthe negative effects of using Nafion dryers. In addition, the two
stage trapping procedure eliminates the need for electronic cooling devices that can add cost and
complexity to realtime preconcentrators. Cryogen usage in the GC oven has been eliminated by
using a multicolumn configuration that optimizes separation of the C2-C10 hydrocarbon fraction
at above ambient starting temperatures. A new single detector configuration is also investigated
which could greatly simplify the analysis compared to multicolumn/multidetector approaches.
Single detector compatibility would also permit laboratory-based GC/MS conformations to be
performed using the same chromatographic configuration. The system can operate for weeks
without attention, and can down load information into an Excel Data base for ease of reporting.
Both data retrieval and system operation can be done remotely using modem communication
through a windows operating environment.
INTRODUCTION
Volatile Organic Compounds are generally regarded as those having a vapor pressure of at least
0.1mm Hg at 25 deg. C. In order to analyze this fraction in ambient air, several other chemical
and physical properties must be evaluated before selecting the best possible analytical method.
These properties include boiling point, polarity, polarizability (in the presence of a localized
charge), reactivity, degrees of freedom ( to assess tendencies to adsorb irreversibly), and 3-
dimensional structure. Some VOC's are relatively stable and can be subjected to harsher
conditions during sample preconcentration before performing a GC/FID or GC/MS analysis.
These compounds can be trapped on chemical sorbents, even those partially oxidized by 03 and
NOx exposure, and then recovered at temperatures up to 300 deg. C or higher. On the other side
of the spectrum lie compounds that must be sampled, concentrated and analyzed by the gentlest
conditions possible as in EPA Method TO3 (direct trapping onto a glass bead trap at -180 deg.
282
-------
C). In general, all but the most reactive or highly polar VOCs can be sampled and analyzed using
compromising methodology such as EPA Method TO 14.
Most unsubstituted hydrocarbons and chlorinated hydrocarbons fall under the classification of
relatively stable VOC's. These compounds are less affected by partially oxidized sorbents and
higher desorption temperatures and may therefore be concentrated successfully without the use
of cryogenics or expensive cooling apparatus. Accomplishing this in practice, however, may
require multidimensional trapping followed by multidimensional chromatographic separation to
accommodate the vast boiling point differences of C2-C12 hydrocarbons. A description of a
system that accomplishes this follows.
ANALYTICAL TECHNIQUE
Ambient temperature preconcentration of C2-C12 hydrocarbons followed by capillary chroma-
tography requires at least 2 sorbent trapping stages. The purpose of the first stage is to supply
enough sorbent to keep the C2 hydrocarbons from breaking through the trap before the end of
the trapping procedure and final trap flush. This requires a fairly wide bore trap with a length
considerably longer than classical traps. Since desorption of the first trap can occur with a
relatively small back-flush volume (10-30cc), a much narrower secondary trap con be used to
further concentrate the sample before injection onto a capillary column for separation and
analysis. Rapid injection off the second trap will insure that narrow peaks are obtained in the
chromatogram.
As there is a vast difference in boiling points between C2 and CIO hydrocarbons, no one column
can currently separate the entire range without the use of cryogenic cooling. Therefore two
columns become necessary; one to resolve the light ends and one for the heavier hydrocarbons.
These can be arranged in parallel or in series using either pressure diversion or rotary valve flow
switching. Series operation is generally favorable as it keeps the heavier fraction off of the
column optimized for the light ends which in turn can reduce or eliminate the occurrence of
"ghost peaks" (heavy compounds finally eluting from the column that were injected in a previous
analysis). Each column is usually configured with its own FID thereby generating two
quanlitation files that can be merged together in a spread sheet or data base.
INSTRUMENTATION
The Entech 2000 programmable preconcentrator was used to perform the 2-stage cryogenless
preconcentration. A 2016BCM 16-position autosampler allowed for preprogrammed selection
of a propane standard, retention time standard, blank, and ambient air sample. Primary trapping
was done on a 1/4" X 20" multi-bed sorbent trap after which lOOcc of helium was used to remove
air from the trap before heating (figure 4). Thermal desorption onto a I/16" x 20" multi-bed trap
in the second stage was accomplished with 15-20cc helium under mass flow control. Sample
transfer onto each trap was performed with sorbent temperatures ranging from 25-30 deg. C.
After preheating the second stage, the GC carrier gas was diverted through the trap at higher than
normal flow rates by utilizing Electronic Pressure Control, or EPC (Hewlett-Packard Company,
Palo Alto, CA) in order to improve the peak shape of the lighter hydrocarbons. The GC columns
were configured as shown in Figure 3. The entire C2-C12 fraction was initially injected onto a
283
-------
11 Om DB1 column with the effluent also flowing through a 3 Om GSQ column (J&W Scientific,
Folsom, CA). After the C2-C3 fraction had been delivered onto the GSQ column, GC flow was
diverted directly to the FID thereby isolating the GSQ column. The C4-C10 compounds were
eluted by slowly ramping both the oven temperature and column pressure. After a short bakeout,
'the oven temperature was reduced to 50 deg. C and flow was resumed to the GSQ column to elute
the C2-C3 fraction.
DISCUSSION
Quantitative, reproducible trapping of the C2-C10 fraction was tested using a 56 component
Ozone Precursor Standard (Alphagaz) diluted down to 5 ppbv (lOppbv for aromatics). To
determine the breakthrough volumes for the primary and secondary traps, each was tested
separately by holding all other preconcentration parameters constant. Figure 2 shows the
response of ethane, ethene, and acetylene relative to the propane response for different sample
TABLE 1 - RF's RELATIVE TO PROPANE FOR DIFFERENT
TRAP 1 SAMPLE VOLUMES
COMPOUND 200 400 600 800 1000 1500
Ethane .65 .67 .67 .64 .68 .65
Ethylene .69 .68 .71 .69 .7! .68
Acetylene .63 .65 .66 .65 .64 .55
volumes. As expected from previously reported trapping efficiencies (1), the breakthrough
volumes using such a large sorbent bed were over 1000 cc for all analytes. Table 2 shows the
recovery in trap 2 using different trap 1 desorption volumes. Acetylene, in particular, was
monitored as it had the lowest breakthrough volume on the first trap. Volumes over 20 cc did show
loss of the acetylene in the second stage. However, since recoveries of the CIO fraction did not
appear to increase with desorption volumes over 15cc, the minimum breakthrough volume in
stage 2 was greater than the volume necessary to recover the target analytes from the primary trap.
Figure 7 shows elution of the C2-C5 hydrocarbons directly to the FID without first delivering the
C2-C3 fraction onto the GSQ column. Note that although the C2 and C3 fraction does not resolve
into individual species, the overall peak shape of these unretained analytes is extremely good,
indicating that desorption out of the 1/16" trap using Electronic Pressure Control occurs very
rapidly.
284
-------
TABLE 2 - RF'S RELATIVE TO PROPANE FOR DIFFERENT
TRAP 1 DESORPTION VOLUMES (CC)
COMPOUND 15 20 25 30 40
ACETYLENE .65 .64 .58 42 22
1,3,5-TRIMETHYLBENZENE 3.52 3.56 3.58 3.57 3.58
The C4 hydrocarbons were all baseline resolved on the DB1 column starting at 35 deg. C.
However, some of the C5 and C6 analytes did not fully resolve from one another. In addition,
bakeout of the DB1 column had to be cut short after elution of C12 compounds in order to cool
the oven for elution of the C2-C3 held on the GSQ column before it was time to inject the next
sample (1 hour continuous operation). Separations and column cleanup could be maximized
using this single detector configuration by performing analyses every 1.5 hours rather than
hourly. Otherwise, at the moment the only alternative for hourly analyses is to use the more
classical 2 column, 2 detector configuration. The HP Chemstation can easily support simulta-
neous FID operation.
CONCLUSIONS
The system described was demonstrated to quantitatively trap the complete list of Ozone
Precursors with subsequent injection into a GC/FID. No Nafion dryer was required so partial
loss of the C2-C3 fraction was eliminated. Also, the lack of aNafion dryer is expected to reduce
carryover and artifact introduction during routine operation. Allowable trapping volumes were
2-3 times greater than reported elsewhere with an expected improvement on the quantitation
reliability in the 0.1-0.5 ppbc range. An on-line 16-position autosampler allows several support
gases (Analytical and RT standards, blanks) and field samples to be selected by the system on
an unattended bases. Utilization of an Entech 4510 real-time integrating controller allows
support gases to be sampled and injected while collecting 2-3 hour integrated samples directly
into a canister for subsequent analysis. Hence, 24 hour/day coverage does not need to be
sacrificed in order to run standards and blanks daily. All of the 56 Ozone Precursors did not
separate from one another using the dual column/ single detector GC configuration tested. Until
further investigation proves otherwise, a separate FID detector will need to be allocated to
monitor the effluent from each of the 2 columns in order to maximize target analyte separations
during 1 hour analyses.
This cryogenless configuration is expected to undergo extensive field testing during the 1993
ozone season. Previously obtained Entech preconcentrators can be easily modified to operate
without cryogen due to the modularity of the trapping system.
REFERENCES
1.) Dario A. Levaggi, Walter Oyung and Rodolfo V. Zerrudo, "Noncryogenic concentration of
ambient hydrocarbonsfor subsequent nonmethane and volatile organic compound analysis,
Proceedings of the 1992 U. S EPA/A&WMA International Symposium on Measurement of Toxic
and Related Air Pollutants, RTF, NC, May 1992, pp. 857-863.
285
-------
Entech/HP Ozone Precurser Analyzer
3 Modes of Operation
Cryogenless Dual Sorbent
Trapping Preconcentrator/
Dual Column GC/FID
(Auto GC)
LN2 Based
Preconcentratlon/
Slngls Column
GC/FID (GC/MS)
(Auto GC)
Remote
Unattended
Tube Sampling
with 1600SS
(48/96 hours)
Laboratory
Analysis
Figure 1
Real Time C2-C10 Ozone Precurser
Monitoring System
5-150 seem
Figure 2
Cryogenless GC Separations
30 m GSQ
Cryogenless Ozone Precurser
Trapping
Ambient
Air in To MFC
1 /4" x 20' Multlbed Sorbent
•+]
1/16'x 20" Multlbed Sorbent |
110m, 0.32, mm ID,
1 urn Film, DBI
FID
Focuser
(Not Used)
GC
Regulated
Heliun
1IV x 20" Multlbed Sorbent
1 /16' x 20" Multlbed Sorbent
GC
Focuser
Figure 3
Figure 4
286
-------
Ozone Precurser
"Cryogenless" Preconcentrotion
Events
1 Mln
Trap
40Min
1.5 Mln
OMin
* <$&*
*
"Qaleu^i \
Source- |
*
Trap at
6"100cc/mh
*
Pretieaf
Ml Adsorbent
V
Oesarbvflft
25cctoM2
J,
V
Wai for
©O Ready
1 .5 Mln
^•Realtime Canister
^"••Propane ,. . ..
Cv 3 M n
N>RT Standards
yBlank
^Up to 1 1
Additional c K/|in
rini-! c^,m^iA. O ™lln
Field Samples
5 Mln
0-1 .5 Mln
FIGURE 5
Preroat M2.
Adsorbent
*
tnjecd"
*
Bakeout
- - Traps
J,
V
"•
Coot Down
Haps
*
Watt tor
End of hour
3 In 1 Field Sampler, Autosampler
and Cleaning System
FIGURE 6 - Optional Laboratory
analysis of Ozone Precursors
using automated sampling onto
long sorbent tubes. C2-C10's
are all quantitatively trapped
and recovered
287
-------
FIGURE 7 - Dual stage
cryogenless
preconcentration and
injection of C2-C5 mix
onto a 110m DB1 column
(.32mm ID, 1um Film).
Although the C2 and C3
hydrocarbons did not
separate at the 35° C
starting temperature,
they share the same
peak width as the C4 and
C5 hydrocarbons
attesting to the fast rate
of splitless injection using
EPC.
FIGURE 8 - Cryogenless
preconcentration,
injection and separation
of a diluted (5ppbc)
alphagaz 56 hydrocarbon
standard using two
columns and only one
FID detector. Initial
background
contaminants at 38-44
minutes were eliminated
after a day of operation
(FigureQ).
aooo -
FIGURE 9 - Real time
analysis of Los Angeles
air at the start of rush
hour traffic. C4
compounds and BTEX
were in greatest
abundance. Estimated
detection limits with
SOOcc concentrated and
injected without splitting
is about 0.03-0.05ppbc.
BOOO -
6OOO -
SOOO -
3OOO -
Apr-il 6. 1993
3:33 AM
288
-------
Photoionization Detection of Air Toxics with Microbore
Chromatography Columns
Kenneth R. Carney, Aaron M. Mainga, Edward B. Overton
Institute for Environmental Studies,
42 Atkinson Hall
Louisiana State University
Baton Rouge, LA 70803
ABSTRACT
The microchip gas chromatograph has shown enormous potential for on-site identification and
quantitation of toxic materials in the environment. Detection limits some 2 to 3 orders of magnitude
higher than other portable gas chromatographs, however, limit the utility of the current commercially
available unit. The poor detection limit performance has been widely attributed to the use a micro-
volume thermal conductivity detector (|iTCD). Conventionally, the TCD has been accepted as being
grossly less sensitive than detectors such as the flame ionization detector(FED) and, especially for
portable GC, the photoionization detector(PID). Other work has suggested that the extremely small size
of the microchip TCD results in a much more sensitive TCD, one that is competitive with the FID or PID
in terms of sensitivity. This presentation will contrast the relative merits and drawbacks of the TCD and
PID in the context of field analytical instruments. The results of a direct head-to-head comparison of the
two detectors using a microbore GC column will be presented.
INTRODUCTION
The current micro gas chromatograph has shown precision and accuracy in the field to a degree
unrivaled by other portable chromatographs. The performance is even more impressive when power
requirements are considered. The device can provide temperature and operational controls sufficient to
provide retention precision to within less than 1% r.s.d. and area reproducibility to within less than 5%
r.s.d while drawing approximately 5 to 10 watts, of power from a 12 V power source.
Detection limits from 1 to 3 orders of magnitude higher than other field GC technologies have
hindered the realization of the potential in the microchip GC for field analysis. The relatively high
detection limits are commonly attributed to the use of an insensitive detector. The available unit currently
offers only TCD detection in the form of a micro thermal conductivity detector. There is reason for
questioning the seemingly obvious conclusion that the |oTCD is the limiting factor in the system's
sensitivity; other workers have suggested that miniaturization of a TCD improves its sensitivity. If this
is the case other factors may provide the principal limits for the system's sensitivity. Namely, the use of
small columns reduces the actual amount of material loaded onto the column and subsequently passing
through the detector. The downsizing of the internal volume of the detector may provide advantages
with respect to sensitivity. Furthermore, because the instrument is designed for VOC analysis the
289
-------
detector need not be externally heated. The relatively cooler detector walls that result may provide
sensitivity advantages over what is expected by conventional wisdom.
A distinct advantage of the (iTCD is its relatively constant response for a wide variety of
compounds. While many common detectors respond more strongly to some classes of compounds than
others, the nTCD exhibits relatively uniform response characteristics among numerous classes. Relative
responses for photoionization detectors vary by approximately an order of magnitude between, for
example, aromatic and aliphatic hydrocarbons.1 Similarly flame ionization detectors show very low
response toward perhalogenated compounds.2 Often detector selectivity can be put to good use; neither
the PID nor the FID respond to water, eliminating the difficulty of quantitation in the presence of a large,
ubiquitous background peak. Detector selectivity also eases chromatographic resolution requirements in
predetermined analytical methods. An important disadvantage of relying on selective detectors is that the
methods tend to be brittle, applying only to a given set of analytes in a narrow window of sample types.
If the sample matrix, including interfering analytes, changes greatly then the method often must be
changed to accommodate new interferences. Secondly, the ability to make semiquantitative estimates in
the absence of authentic analyte standards is confounded.
Developments in the application to fast microscale chromatography of well controlled temperature
programming and heated injection systems presage the advent of similarly portable instruments for
semivolatile field chromatographs. These developments will necessarily require that the detector be
heated to substantially higher temperatures than currently required for VOC instruments. This
requirement may make the jlTCD unworkable from the standpoint of both sensitivity and longevity as
higher operating temperatures will certainly have deleterious effects, although to what extent is uncertain.
Two common detectors generally thought to have sensitivity advantages are the photoionization
detector (PID) and the flame ionization detector (FID). Both of these detectors are relatively more
expensive to use and maintain than TCD's. Furthermore both the PID and FID show large variations in
response over different classes of compounds. Nevertheless, the FID and PID seem to be the most
likely alternatives to the TCD. The photoionization detector presents a practical advantage over flame
ionization detectors- it does not require an auxiliary supply of fuel or oxidant gas. Consequently a
system based on PID detection will be more convenient to transport and operate in the field. The small
volume of the (iTCD offers the possibility of routinely using serial detectors to provide additional
confirmation of analyte identities. As an initial step in looking at the feasibility of alternates to TCD as
the detector in microbore GC we have begun to quantify the tradeoffs involved using photoionization
detection with the micro GC.
EXPERIMENTAL
A commercial micro GC unit was obtained from MTI Analytical Instruments. The unit was
equipped with a 4-meter x 0.1 mm i.d. OV-1701 column (df=0.4|i). The detector was the standard MTI
micro thermal conductivity detector (internal volume <1|J.L). The unit was modified by adding a 0.5
290
-------
meter x 0.1 mm i.d. deactivated fused silica transfer line was attached to the outlet of the |iTCD detector
using zero dead volume fittings immediately at the ^.TCD effluent line. This line allowed other detectors
to be used in line with the jiTCD detector. The |iTCD volume was equivalent to less than 1% of the total
column length. Helium was used as the carrier gas. The flow rate for the microbore GC column was set
to 35 cm/second (=0.2mL/minute). Thus the residence time in the transfer line was approximately 1.5
seconds.
The transfer line from the micro GC was connected to the detector inlet of a Photovac 10S50
portable GC. The detector response was recorded from the d.c. recorder output on the 10S50 GC. An
A/D interface that emulated that found in the micro GC was used to digitized the PID detected
chromatograms which were collected using the M2001 software package, designed at the Institute for
Environmental Studies to operate the MTI GC for simultaneous 2-dimensional chromatography.3 The
same software was used to record chromatograms as detected by the (iTCD. This procedure ensured
that both data streams were being treated identically in software.
The uTCD required no optimization or setup other than allowing it to warm up. The principal
optimization variable for the PID was makeup gas. UHP grade nitrogen was used as makeup gas for the
PID. The PID response to mixtures of normal hydrocarbons in air was evaluated as a function of
makeup flow and, under the optimum conditions, that response was compared to the (J.TCD response.
Gas mixtures were prepared by flash evaporation of neat liquids into a flowing air stream. They
were held in Tedlar sample bags for use and fresh mixtures were prepared each day. Samples were
introduced into the micro GC using the standard sample loop mechanism in the commercial unit. A
minimum of 2 ml of sample was flushed through the sample loop to minimize sample loss and carryover;
a 2 ml sample corresponds to about 200 sample line volumes.
RESULTS AND DISCUSSION
Three closely related aspects of PID and |J.TCD sensitivity have been briefly addressed— (1) signal
to noise ratio for individual chromatograms, (2) peak shape, and (3) detection limits for analytes. The
most basic determination is the signal to noise ratio for peaks in a chromatogram. This value is related
most specifically to the detector comparison as the two detectors, connected in series, measure the same
sample. Comparison of the peak shapes at the two detectors indicates the relative the effect on resolution
of the detector system. As detectors rarely have other than deleterious effects on peak shape and
resolution, comparison of PID and |XTCD indicate the relative damage each causes to the
chromatography. Finally, the bottom line in any quantitative analysis system is often "detection limit."
The main question at this point, in gauging the relative performance of the alternative detectors is "does
the use of one detector in the overall system offer a performance advantage over the other detector.
Consequently the detection limits for the system using each of the detectors were compared for a series
of mid- and low-ppbv air samples.
291
-------
Signal to Noise Ratio
Figure 1 shows relative response of the photoionization detector and the micro thermal conductivity
detector to a sample containing 100 ppmv each of pentane and hexane. While the PID response in
Figure 1 is roughly equal to that for the (iTCD, the gain setting for the PID was a factor of 100 below
the maximum. While the |iTCD was operating at maximum sensitivity, the PID was operating well
below its maximum sensitivity. Note that the retention times in have been corrected for residence time in
the transfer line. The PID apparently is capable of much higher sensitivity than the |iTCD. Both traces
have the same limiting noise level; so the relative responses of the two detectors would look the same if
they were compared on the basis of signal to noise ratio. The source of the noise limitation is an
important consideration in extrapolating the results, especially in the projection that the PID may be 100
time more sensitive than the |iTCD. If the noise is in the detector response itself or between the detector
and the electrometer, then increasing the electrometer gain by a factor of 100 will do nothing to increase
analyte detectability; the relative sensitivity of the two detectors will be as illustrated in Figure 1
regardless of the gain setting on the PID electrometer. The source of the limiting noise is currently being
investigated.
Peak Broadening
Another striking feature of the comparison is the peak broadening seen in the PID trace. Figure 2
illustrates the degree of peak broadening seen in the PID result compared with the |lTCD result at several
makeup gas flow rates. The curve for the jiTCD shows the expected (and observed) trend in peak width
as a function of adjusted retention time for the column with little or no post column peak broadening.
This curve intersects the y-axis, representing an unretained peak, at approximately 250 ms, which is the
width of the injection zone. The PID result shows considerably broadened peaks both without makeup
gas, as expected, but also with even large flows of makeup gas. Increases in makeup flow improved
peak shapes only up to makeup flows of about 10 ml/minute. Subsequent increases in makeup flow
rates did not further improve peak shapes. Peak widths for late eluting peaks were comparable for both
detectors, but earlier eluting peaks were progressively more distorted by the PID detector. The peak
width of the PID detected peaks approached a minimum value of 2 seconds. This is consistent with a
response time constant of about 1 second. In other words, an infinitely sharp peak passed through a
detector with a 1 second time constant would have a half height width of about 2 seconds. Internal
detector volume was eliminated as a source of the peak broadening because continued increases in
makeup flow did not further reduce the peak widths. Residence time in the transfer line was too short to
contribute so much to the post column broadening. Replacing the PID with a Hewlett Packard flame
ionization detector verified that the transfer line was not the source of the broadening. The likely source
of the peak broadening in this study was insufficient bandwidth of the Photovac electrometer. In order
to accurately represent the sharp peaks eluting from the short 100|J. columns in the microGC unit, the
detector (including the electrometer) must have a time constant below 0.075 seconds. As with signal to
noise ratio, further investigation is required to quantify the sensitivity comparison inherent in the
detector. A faster electrometer (i.e., larger bandwidth) may mean a higher limiting noise level. A
292
-------
fundamental, direct, relationship exists between the bandwidth of an amplifier and the magnitude of
random noise in the signal. Another way to conceptualize this is to think of a slow electrometer as a low
pass noise filter. Thus it is possible that the some of the apparent two order of magnitude sensitivity
advantage of the PID may disappear when the faster electrometer is used. Again, as with the signal to
noise ratio measurement, much depends on the ultimate source of the noise.
Detection Limit
The calibration curve for a set of standard air mixtures containing n-alkanes at concentrations from
1000 ppbv to 50 ppbv are shown in Figure 3. The standard deviation of peak areas for all compounds
ranged from about 0.5 mV-s at 50 ppbv to about 2.5 mV-s at 1000 ppbv. On the basis of the standard
deviation at 50 ppbv and the calibration curves shown in Figure 3, the detection was estimated at 20 to
50 ppbv for all of these alkanes. Given that the PID is about 10-fold more responsive to benzene than
to n-hexane one may expect that the detection limit for benzene would be in the 2-5 ppbv range. The
caveats of making such an extrapolation were discussed above and apply here as well. Nevertheless,
this is somewhat higher than the extant estimate of 0.1 ppbv benzene detection limit often cited for
portable GC's using photoionization detection. Previous workers have determined detection limits for
the commercial microGC unit to be about 200 ppbv.
CONCLUSIONS
On the basis of signal to noise ratio for a single chromatogram, photoionization detection with
microbore gas chromatography seems to be approximately 100 time more sensitive than the nTCD. The
slow electrometer used here extracted a substantial price in terms of degraded chromatographic
resolution. A loss in that sensitivity advantage may be realized with a faster electrometer because of the
relationship between higher bandwidths (i.e., faster electrometer) and higher levels of random noise.
When the total system was evaluated the advantage in detection limits for the PID over the (iTCD was
only about a factor of 5 for n-alkanes. For benzene the advantage may be as much as 50 if the
extrapolation using the relative response to benzene is to be believed. The PID advantage over the |iTCD
may not be as phenomenal as it first seemed, though it does seem clear that there is some advantage.
Precise quantitation of how much is gained by using a PID with microbore GC must be compared with
the costs of operation, maintenance and development of a PID system for the microGC.
REFERENCES
1. J.B. Barsky, S.S. Que Hee, C.S. Clark, "An Evaluation of the Response of Some Portable,
Direct-Reading 10.2 eV and 11.8 eV Photoionization Detectors and a Flame lonization Gas
Chromatograph for Organic Vapors in High Humidity Atmospheres," Am. Ind. Hyg. Assoc. J.
46(1): 9-14 (1985).
2. M.J. O'Brien, Modern Practice of Gas Chromatographv; R.L. Grob, Ed.; John Wiley and Sons,
Inc., 1985, pp247-258.
3. K.R. Carney, R.L. Wong, E.B. Overton, et. al., Sampling and Analysis of Airborne Pollutants.
E.B. Winegar and L.H. Keith, Eds.; Lewis Publishers, Boca Raton, 1993, pp-21-37.
293
-------
>
~aT
C
o
8
950-1
750-
550-
350-
150-
en
— (iTCD
nTT-t
, C6 ™
|
rfi
a i\ f\'-'
1 \ //\\
r~**J.\ v ^^.^rs^,.,.. ....... JJ, vX^
-jU -t 1 1 1 1
20.00 30.00 40.00 50.00 60.00
time (seconds)
Figure 1: Detector response, lOOppmv n-alkanes.
"o"
^s.
2
r
6000 -I
5000-
4000-
3000-
2000-
1000-
0 •
. •• >i *
•^>D O
0 50
•
D
-
100
•
,_-p
.*-•'"''
-" o
150 200 250 300
none
3ml/min
SmVmin
lOrrJ/rnin
HTCD
RETENTION TIME (sec)
Figure 2: Effect of makeup flow on peak width.
294
-------
">
AREA (m1
M
s
100.00 -
80.00 -
60.00 -
40.00 -
20.00 .
0.00 •
(
^ •'*",, • butane
^**_.--" n hexane
;''.--'' _.. — •*"""" • heptane
,-,<"-' ~ 㣥""" O octane
• ^^^l^r-^ ; ~* • ™
) " 200 400 600 800 1000
CONCENTRATION (ppbv)
Figure 3: PID response vs. concentration with microbore GC
295
-------
NEAR REAL-TIME GC ANALYSIS OF VOLATILE ORGANIC
COMPOUNDS USING AN ON-LINE MICRO-TRAP
Somenath Mitra and Hung Jen Lai
Department of Chemical Engineering
Chemistry and Environmental Science
New Jersey Institute of Technology
Newark, NJ 07102
Merrill Jackson
Source Methods Research Branch
Methods Research and Development Division
Atmospheric Research and Exposure Assessment Laboratory
US Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Micro-traps act as sample pre-concentrators for gas chromatography (GC) that can be used to make
repetitive injections every few seconds. A thermal desorption micro-trap is r-iade from a short segment of
thin tubing containing an adsorbent or a chromatographic stationary phase. A carrier gas containing the
analyte of interest can be introduced into the GC analytical column through the micro-trap which acts as a
sample trap. Rapid heating of the micro-trap releases a "concentration pulse" of the analyte that serves as a
GC injection similar to that from an injection valve. Micro-traps can be used in various applications such as
process stream analysis, fast and multi-input chromatography.
INTRODUCTION
In recent days, the volatile organic compounds (VOCs) have received much attention as air pollutants and
several are listed in the Clean Air Act Amendment of 1990, Title III. The analysis of VOCs is particularly
challenging because they are usually present in very low concentration (ppmv to ppbv level). VOCs in
gaseous matrices such as stationary stack emissions are analyzed using whole air samples collected either
with the Volatile Organic Sampling Train (VOST)1 or with Tedlar bags using EPA Method 182. Both these
approaches attempt to analyze dilute gaseous matrices by concentrating a small amount of analyte from a
large volume of gas. While these methods are quite effective in VOC measurements, they can not be used
for continuous or near real-time monitoring.
The important feature of any continuous, on-line GC instrumentation is the sample introduction system,
that is required to make automatic, reproducible injections. In many chemical industries, process GC is
accomplished by the use of sample valves as injectors. Valves can automatically make injections from a
sample stream onto a GC column. However, sample valves have certain limitations. Being mechanical
devices, they tend to deteriorate during extended operation. Also, sample valves can only handle small
volumes of gas, generally between a few uL and two mL. Injecting a large sample volume causes excessive
band broadening and degrades chromatographic resolution. A small sample volume results in reduced
sensitivity. As a result, sample streams that are at sub part per million concentration levels can not be
effectively analyzed using sampling valves. In many applications, especially in environmental monitoring, low
VOC concentrations are encountered and sample valves are found to be inadequate.
Real-time VOC monitoring (at trace levels) using a GC, requires an automated injection device and a
sample preconcentrator. The research reported here used an on-line micro-trap to serve the dual purpose of
296
-------
sample concentration and injection.
On-line Micro-trap
An on-line micro-trap was made by packing a short (few cm long) piece of metallic or fused silica
tubing with an adsorbent. The sample containing the analyte is introduced into the analytical column
through the micro-trap. The analytes are retained in the micro-trap and can be thermally desorbed by
electrically heating the micro-trap. When the heating is rapid enough, the desorption generates a
concentration pulse that serves as an injection. The different compounds are separated by the column and
analyzed by the detector. The mode of operation for continuous monitoring is that injections (or pulses) are
made at fixed intervals of time and corresponding to each pulse, a chromatogram is obtained. The advantage
of the micro-trap is that it has low thermal mass and can be heated/cooled very rapidly. So, repetitive
injection can be made as long as GC separation is completed. The amount of sample trapped in the micro-
trap is proportional to the concentration of the stream sampled. Consequently the micro-trap response is
proportional to sample stream concentration.
EXPERIMENTAL SYSTEM
The experimental system is shown in the Figure 1. The VOC sample stream was generated by entraining
the analytes from a diffusion tube onto a flow of nitrogen. The analyte concentration was controlled by
changing the diameter and the height of the liquid level in the diffusion capillary. The concentration of the
stream was calculated using diffusion equations.
A Hewlett Packard GC (model 5890) equipped with a flame ionization detector (FID) was used in this
study. The micro-trap was made by packing 0.5 mm ID stainless steel tubing with different adsorbents. The
micro-trap was heated by passing a pulse of electric current (duration 100 to 700 msec) directly through it's
metal wall. The injections were controlled by an IBM compatible personal computer using the digital output
of the analog to digital converter (DAS8-PGA, Metrabyte Corp.) and an electronic switch (OAC5P, Opto 22,
Huntington Beach, CA). The micro-trap was heated at fixed intervals of time. The interval between
injections varied between 5 and 300 sec. A computer program written in Quick Basic was used for making
injections as well as for data acquisition.
RESULTS AND DISCUSSIONS
The operation of the continuous analysis system was demonstrated by continuously monitoring a stream
containing ppbm levels of hexane, dichloromethane, toluene and ethylbenzene. The injection from the micro
trap (referred to as a pulse) was similar to that from an injection port or valve. A series of pulses were
generated at one minute intervals. Each pulse produced a four peak chromatogram as shown in Fig. 2. The
high sensitivity of the micro-trap is quite obvious, chromatographic separation may be reduced by shorting
the pulse interval to 45 seconds or installing microbore columns.
Reproducibility of retention time as well as peak height was very good for the micro-trap and was
comparable to that of an injection port. Since the flow through the micro-trap is very compatible to the flow
through the column and there is practically no dead volume, the micro-trap produces sharp peaks. Linearity
of the calibration curve is also an important consideration for on-line measurements. The amount of sample
trapped by the micro-trap is theoretically proportional to the concentration of sample flowing through it.
Here we found the calibration curve to be linear in the ppb to ppmv range.
REFERENCES
1. Test Methods for Evaluating Solid Waste-Phvsical/Chemical Methods. EPA-SW-846, 3rd Edition,
Method 0030, U.S. Environmental Protection Agency, Washington, DC.
2. U.S. Government Printing Office. Code of Federal Regulations. 40CFR, Part 60, Appendix A, 1993.
297
-------
The information in this document has been funded wholly by the United States Environmental Protection
Agency under cooperative agreement R815734 to Northeast Region Hazardous Substance Research Center.
It has been subjected to Agency review and approved for publication. Mention of trade names or commercial
products does not constitute endorsment or recommendation for use.
298
-------
MICRO-TRAP
VOC CONTAINING
SAMPLE
GC COLUMN
DETECTOR
POWER SUPPLY
SWITCHING DEVICE
COMPUTER
Figure 1. Schematic diagram of the experimental system
-------
2
u
3
LA
3
2
U_
LA
-j
UL__
^
u_
1
:
LA :
(.•
JL_
JL__
^u
2
3
4
U^
P1 p2 P3 P4 P5 P6 P7 P8
1
Hexane
2) DCE
3) Toluene
4) Ethylbenzene
2
8
Time (min)
Figure 2. Continuous monitoring of VOCs using the on-line micro-
trap . P
1'
are the different injections corres-
ponding to which chromatograms were obtained.
-------
THE USE OF ANALYTICAL AND METHOD SURROGATES
IN GC/MD ANALYSIS OF WHOLE AIR SAMPLES
David A. Brymer, Larry D. Ogle, Christopher J. Jones, and Carl L. Shaulis
Radian Corporation
P. O. Box 201088
8501 Mopac Blvd.
Austin, TX 78720-1088
ABSTRACT
USEPA Compendium Method TO-14 describes the general approach necessary to successfully analyze 40
compounds by either GC/MS or GC/MD. The introduction of automated interfaces which concentrate the air
sample prior to introduction into the analytical system provide an opportunity to include additional quality control
measures. An analytical surrogate, a compound added to the preconcentration device during sample concentration,
can help assess data quality during unattended operation. A method surrogate can be added into the canister sample
prior to dilution/analysis to provide an indication of analytical accuracy, including sample dilution and analysis.
Addition of surrogates allows the continual assessment of analytical accuracy and precision during TO-14
analyses. The use of two surrogates introduced at different points within the analytical process help define many
potential problems. A method surrogate can be used to assess the effects of sample matrix and handling while the
analytical surrogate provides a measure of interface, gas chromatograph, and detector performance.
INTRODUCTION
The analysis of whole air samples by GC/MS or GC/MD for volatile organics is a well established
technique. EPA Compendium Method TO-14 describes sample collection, target compound lists, calibration
requirements, and data reduction1. However, the use of surrogates as a means of assessing method/analytical
accuracy and precision as is common in other EPA methodology is not discussed.
Advances in VOC analytical techniques have increased the need for surrogate addition to improve quality
control procedures. An increase in the use of automated interfaces which can analyze up to 16 canister samples in an
unattended mode necessitates this additional quality control to insure proper instrument operation and data validity.
Instrument problems (such as the column temperature not equilibrated prior to injection, significant changes in
carrier or detector gas flows, cryogen outage or failure, and problems with sample loading such as valve failure)
which would normally be apparent to an analyst on a real-time basis can be difficult to assess when data are collected
in an unattended manner.
The same advances in analytical technology which pose this challenge can be used to solve it using surrogate
addition. Several of the commercially available (and privately designed) VOC preconcentration devices provide the
capability to add a surrogate onto the trap prior to sample injection. The ability of instrumentation to add such
compounds in a very precise manner has been described by Dayton and Bursey. Analytical surrogates can be used
to assess proper operation of automated VOC instrumentation.
The assessment of sample handling, storage, and matrix effects are also of interest in evaluating method
accuracy and precision. A method surrogate can be added to the canister either prior to or after sample collection to
assess (post collection) method accuracy and precision. Method surrogates can be especially valuable when dealing
with high concentration samples of unusual matrices (point source samples) which require multiple dilutions and non-
routine analytical approaches.
The goal of this study was to develop techniques for and evaluate the addition of analytical and method
surrogates during VOC analyses. Acceptable surrogate compounds must be amenable to the GC/MD technique
without interfering with potential GC/MS confirmation analysis.
EXPERIMENTAL METHODS
The study was divided into four tasks: surrogate compound identification, analytical surrogate addition,
method surrogate addition, and evaluation. The following paragraphs describe each of these tasks.
Surrogate Compound Identification
The search for viable surrogate compounds was first narrowed to the volatility range of the target compounds
301
-------
(C2 - C12) and organic compounds which respond on multiple detectors (FID, PID, ELCD, OR ECD). It was
desirable that the surrogate chromatographically separate from target analytes as well as any compounds routinely
observed in ambient air samples. Compound stability was also a concern. The compounds chosen for evaluation
based upon these criteria were identified as shown in Table 1.
Each of these compounds was introduced into an evacuated canister and analyzed on a 60m x 0.32mm id
DB-1 chromatographic column using an initial column temperature of -50° C for 2 minutes and then ramping ff C/min
to 175° C. The measured retention time was compared to an ambient air sample and a database of over 200
compounds for potential coelution. Response on each detector was also evaluated.
Analytical Surrogate Addition
The analytical surrogate was added to the preconcentration device to provide data on sample trapping,
desorption, transfer, and detection. Analytical surrogate addition on automated systems necessitates at least one
additional valve to introduce surrogate standard into the system in a controlled manner. Figure 1 presents the design
of the analytical interface used and how the analytical surrogate was added onto the trap. This addition takes place
immediately after sample concentration as a test for cryogen failure or trap blockage. The analytical surrogate
(nominally 13 ppbV) was flushed through the interface lines for 90 seconds, 50 mL were loaded onto the trap, the
lines were flushed with nitrogen and the system was allowed to reach ambient pressure prior to thermal desorption.
Method Surrogate Addition
The method surrogate could be introduced into the canister either prior to or after sample collection. If the
sample matrix was well defined and the measured concentrations did not vary considerably the method surrogate
could easily be added to the canister prior to or post sample collection. If the sample matrix was unknown and the
measured concentrations were expected to vary widely the method surrogate could be introduced into the canister
post collection. In this manner, the sample matrix could be defined through screening to help identify the correct
surrogate and practical spike amounts for each matrix or project.
A fixed loop injection system was used to introduce the method surrogate into the canister. Figure 2
illustrates how the 6 port Valco valve equipped with a 1.1 mL loop made from 1/8" stainless steel tubing was
plumbed into a diluent system. A needle valve was used to control method surrogate flow and a rotometer was
installed downstream of the loop as a visual check of loop equilibration prior to injection into the canister. The
method surrogate, at a nominal concentration of 55 ppmV, was purged through the trap for five seconds. The
surrogate standard was closed and after the trap reached ambient pressure (no flow indicated on the rotometer) the
valve was rotated to the inject position. The diluent gas flushed the surrogate into the canister.
The described techniques were evaluated for accuracy, precision, stability, and practical feasibility. Accuracy
was assessed by comparing a measured concentration to a theoretical concentration (absolute accuracy) or to the
measured standard concentration (relative accuracy). Precision was tested as analytical reproducibility, precision of
the analytical surrogate addition and analytical variability, and method variability. Analytical variability was measured
as an average relative percent difference between the concentration of the method surrogate in 31 pairs of duplicate
analyses. The analytical surrogate data from this same data set represents analytical variability plus analytical
surrogate addition variability. Method variability was estimated by collecting the matrix surrogate into seven canisters
which were subsequently analyzed. This precision estimate reflects method surrogate addition reproducibility,
analytical drift, canister dilution, storage stability, and canister to canister variability.
applications for surrogates in ambient and point source samples was tested for precision and accuracy to vei
usefulness of these techniques in actual samples. The results of the these evaluations are presented below.
RESULTS AND DISCUSSION
The two compounds which best suited all the requirements for surrogates on the DB-1 analytical column
were 1,4-difluorobenzene and 4-fluorotoluene. 4-FIuorotoluene was subsequently selected as analytical surrogate with
1,4-difluorobenzene as a method surrogate for ambient air sample analyses. In this arrangement, 1,4-difluorobenzene
could also be used as an internal standard for subsequent GC/MS confirmation analyses in accordance with existing
and proposed methodology.
302
-------
Relative percent accuracy of the analytical surrogate was determined by comparing response of the surrogate
in humidified air standards containing 15 - 45 volatile organics (n = 192) to the surrogate response in the appropriate
daily method blank. This approach is used for sample recovery and compensates for any slight changes in detector
drift over time. A mean relative accuracy (recovery) was thus calculated to be 95%.
The total variability associated with the analytical surrogate (analytical variability plus surrogate addition
variability) was 5.7%. This relative standard deviation was obtained from standards which were analyzed over a five
day period (n = 39). Analytical variability was estimated at 2.7% based upon the average relative percent difference
of 31 pairs of analytical duplicates. Therefore, the hardware and methodology used to introduce the analytical
surrogate into the system has less than 5% variability.
The analytical surrogate was further evaluated on two different analytical systems over two days using
ambient air samples. The samples used in the evaluation were ambient air samples collected near multiple industrial
point sources located in urban settings. The results shown in Table 2 reflect that both systems show a mean relative
accuracy (sample surrogate AC/method blank surrogate AC) to be between 98 - 104% and precision (relative
standard deviation of the sample surrogate AC) to be less than 8%.
The method surrogate precision (relative standard deviation) was measured to be 1.0% based upon the
analysis of seven replicate standards. The absolute accuracy (100 x [measured concentration - theoretical
concentrationj/theoretical concentration) of method surrogate addition was + or - 7% based upon the mean
measured concentration from the seven replicates. Method surrogate stability data are provided in Table 3 and show
no discernable difference between the mean concentration measured over the study period. Therefore, the method
surrogate is considered stable from the time introduced into the canister through final analysis (at least 24 days).
Method surrogate capability was further tested using synthetic mixtures (n = 192) described in the analytical
surrogate discussion and various point source assessment projects. The method surrogate from the synthetic mixtures
had a mean recovery of 115% when compared to the mean measured concentration of the surrogate standard and a
relative standard deviation of 14%. Method surrogates were utilized on two projects where high level point source
samples were collected and analyzed. One set of samples was characterized by a complex Cl - C7 hydrocarbon
mixture with concentrations ranging from 50 ppmV up to 400,000 ppmV. The samples were collected in canisters and
tedlar bags. After characterizing the sample matrix, c-2-butene was identified as an appropriate method surrogate
due to its retention time and similarity to sample constituents. A method surrogate (12390 ppmV) was prepared and
injected into the sample containers after screening the sample matrix. After multiple sample dilutions, the mean
method surrogate recovery (measured concentration/theoretical spike concentration) was 105% with a standard
deviation of 20.9 (RSD = 19.9%). A large portion of this variability centered around one sample, however, and may
represent matrix effects.
CONCLUSIONS
The use of surrogates can be very important in insuring valid sample handling and analysis when automated
instrumentation is employed. This study has shown that surrogates can be reproducibly added to samples prior to
analysis and at the time of analysis. Two fluorinatcd compounds, 4-fluorotoluene and 1,4-difluorobenzene, are viable
analytical and method surrogates for GC/MD analyses of volatile organics in ambient air. These surrogate data can
be used to validate sample dilutions and proper instrument performance in a timely manner.
The chosen surrogate compounds work well with most ambient samples but may not be feasible for all point
source samples. In such instances, the surrogates must be custom tailored to a specific project or matrix to be
representative of the target compounds and to avoid chromatographic coelution.
REFERENCES
1- EPA Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air, Method
TO-14: EPA-600/4-84-041, U.S. Environmental Protection Agency, AREAL, Research Triangle Park, NC,
1984.
2. Dayton, Dave-Paul and Joan T. Bursey, "Internal Standards Capability of Radian's Interface System" in
Proceedings of the 1989 EPA/APCA International Symposium on Measurement of Toxic and Related Air
Pollutants. VIP-13, Air and Waste Management Association, Pittsburgh, 1989, pp. 106-111.
303
-------
-VACUUM
COLUMN CARRIER
VALVES IN LOAD POSITION
(Selection valve In "Load" position for ECD,
Selection valve in "Inject" position for FID/PID)
MFC = Mass Flow Controller
SAMPLE
-------
VENT
VENT
g"
3
ROTOMETER
, i ,, PRESSURE
\ I GAUGE
N9 In
ROTOMETER
»PRESSURE
' GAUGE
1.1 mL LOOP
1.1 rnL LOOP
LOAD POSITION
INJECT POSITION
-------
Table 1. List of Potential Surrogates for VOC Analyses.
Methyl iodide
Methyl formate
Dimethyl ether
Allene
1-Fluoropentane
4-Fluorotoluene
1,4-Difluorobenzene
Trifluoromethane
3-Chloropropene
2-Bromopropane
Bromofluorobenzene
Chlorobenzene-d5
Table 2. Precision and Relative Accuracy of Analytical Surrogate in Ambient Air Samples.
Method Blank (Area Count)
Sample 1 (Area Count)
Sample 2 (Area Count)
Sample 3 (Area Count)
Sample 4 (Area Count)
Sample 5 (Area Count)
Sample 6 (Area Count)
Sample 7 (Area Count)
Sample 8 (Area Count)
Mean Response (Area Count)
Standard Deviation (Area
Count)
Rel. Std. Dev. (Mean Sample
AC/blank AC)
Mean Recovery
ANALYTICAL SYSTEM 1
Dayl"
7871
7669
7997
8006
8068
7921
7987
9583
8176
634
7.8%
104%
Day 2
11641
11756
11928
11843
12401
12445
12159
12060
11564
12020
308
2.6%
103%
ANALYTICAL SYSTEM 2
Dayl
13560
13898
12131
13837
13816
13355
13070
12881
14010
13374
650
4.9%
98.6%
Day 2
13953
13821
14937
13052
13555
13639
13357
13827
648
4.7%
98.4%
3 Day 1, system 1 response showed higher than expected variability so the load time/volume was increased to produce better precision.
Table 3. Method Surrogate (1,4-difluorobenzene) Stability Data.
Day8
1
7
14
24
Number of
Observations
8
4
4
4
Mean
(ppbV)
19.3
18.4
19.1
19.1
Std. Dev.
(ppbV)
0.644
2.10
0.755
0.882
Rel. Std. Dev.
(%)
3.3
11.4
3.9
4.6
3Elapsed time (days) between surrogate spiking and analysis.
306
-------
MOISTURE MANAGEMENT TECHNIQUES APPLICABLE TO WHOLE AIR
SAMPLES ANALYZED BY METHOD TO-14, II; GC/MS CONSIDERATIONS
Larry D. Ogle, David A. Brymer, Christopher J. Jones and Ruth L. Carlson
Radian Corporation
8501 North Mopac Blvd.
P.O. Box 201088
Austin, Texas 78720-1088
ABSTRACT
Analysis of polar organic compounds collected in canisters using US EPA Compendium Method TO-14 is of
interest to a number of industries and agencies. However, it is commonly known that moisture in the sample can
interfere with the analysis. Most water removal techniques also remove low molecular weight oxygenated organics
which may be of interest. Part I of this work described the use of a selective condensation zone to remove water
during analysis by gas chromatography with multidetectors. This paper will describe the evaluation and continued
development of this technique to remove water after cryogenic concentration to a GC/MS instrument. The method
was determined to effectively remove most of the moisture during GC/MD analysis with little effect on the recovery
of polar organics such as methanol, acetone and diethyl ether. Recoveries of non-polar compounds covering a broad
volatility range, such as butane, benzene, toluene, p-chlorotoluene, n-butylbenzene and n-undecane were unaffected
by this Moisture Management System (MMS). Transfer of the technology to the GC/MS, which is more sensitive to
moisture than the GC/MD system has proven to be more difficult and has required hardware modifications to
remove sufficient amounts of water to achieve reproducible results.
INTRODUCTION
Cryogenic concentration and thermal desorption of water into a chromatographic system during the analysis
of ambient air for volatile organic compounds (VOCs) has been shown to adversely affect chromatography, sensitivity
and/or detector reliability.1"5 When using subambient chromatography, water concentrated from the sample can form
ice plugs in the column resulting in poor peak resolution and retention time shifts, thus making compound
identification difficult. The moisture can also extinguish a flame ionization detector or exceed pressures tolerated by
a GC/MS system during analysis resulting in system shutdown. Nafion membranes and sorbents may be used to
remove moisture from the sample prior to cryogenic concentration, but they have been shown to partially or
completely remove small, polar organic compounds such as alcohols, aldehydes and ketones. These devices may also
be a source of contamination or carryover from sample to sample if not properly conditioned. Another technique
employed to decrease the volume of water introduced into the MS source is the use of a jet separator coupled with a
megabore capillary column. However, this approach yields higher detection limits for smaller target analytes.
A novel approach to the management of water concentrated from ambient samples such that it does not
adversely effect chromatography or detector systems has been developed and is described in Part I of this paper.6 It
is based on condensation of moisture from the saturated carrier gas stream during thermal desorption. A Moisture
Management System (MMS) was installed on a Radian designed and built automated cryogenic interface used to
analyze ambient air samples with GC/MD instrumentation. The MMS system was then evaluated for recoveries of
compounds having a wide range of polarities and volatilities and was found to give excellent recoveries,
reproducibility and chromatography when operated at a temperature of IS" C.
A similar system has been reported by Bernard, et al.7 for the removal of water during purge and trap
analysis. They reported that the amount of water transferred to the column correlates with the vapor pressure of
water in the temperature range of their cryo-focusing module™ (CFM™). The paper reports very good peak shapes
and recoveries for light gases when temperature programming the CFM from -16ff C to -2CFC.
In the first part of this paper, a statistically based experiment was designed to study the inter-relationship
between the Moisture Management System (MMS), relative humidity, compound class, compound concentration,
interface autosampler position, and canister size and the effects of each of these parameters on analyte concentration
measurement precision and accuracy. The results of these experiments led to further studies designed to determine
optimum operating parameters for the MMS.
Based on the results from the initial study reported last year, the technology behind the MMS was evaluated
in conjunction with a Finnigan 4500 GC/MS system. The study objective was to determine if sufficient moisture
could be removed during desorption to utilize a 60 meter by 0.32 mmid column interfaced directly into the detector
307
-------
source. As a general rule, large amounts of water interfere with ionization of target analytes, decrease sensitivity and
increase variability of the analysis.
The wide bore column interfaced directly to the source of the mass spectrometer is vital to achieving the
desired sensitivity, chromatographic resolution and the ability to analyze small polar organic compounds. Two
commonly used approaches to GC/MS analysis of whole air samples, using Nation driers and using Megabore
columns with jet separators, sacrifice some of the desired analytical parameters. The use of Nafion driers results in
the loss of polar organic compounds. The use of a megabore column with a jet separator results in the loss of
chromatographic resolution and higher detection limits.
EXPERIMENTAL DESIGN
The MMS used on the GC/MD systems consisted of an aluminum block which encased a short length of
0.125 inch o.d. tubing between the cryogenic trap and the transfer line to the GC (Figure 1). The device was
passively cooled by nitrogen gas exhaust from the cryotrap as liquid nitrogen was sprayed on the trap during sample
concentration. Temperature of the device was regulated by an 80 W cartridge heater controlled by an Omega®
temperature controller.
The system was configured such that the sample flows through the MMS during concentration. During
thermal desorption, the chromatographic carrier gas flow backflushed the trap transferring desorbed organics and
water vapor through the MMS. Thermal desorption of the cryotrap at 60ff C/minute supersaturated the helium gas
with water vapor which then condensed in the cool MMS region. Through the manipulation of temperature,
desorption time and system configuration, the amount of water removed and the recoveries of organic compounds of
interest were maximized.
A cryogenic interface similar to those used on the GC/MD instruments was equipped with the cryogenic trap
and MMS system as shown in Figure 1. This interface was combined with a Finnigan 4500 GC/MS analytical system
for evaluation. Operating conditions were initially set up to be the same as those which were optimal for the
GC/MD systems (MMS operated at 15° C). The goal of this experiment was to establish a reliable GC/MS analytical
technique capable of accurately measuring low to sub ppb levels of ambient organic compounds from undried air
samples. Columns and conditions used were the same as used for GC/MD analysis, thus providing true
confirmational analysis with similar detection limits and analyte lists.
RESULTS AND DISCUSSION
Analysis of humid standards using the system and conditions described above resulted in very poor
chromatography and sensitivity on the GC/MS instrument. It was obvious that a considerable amount of water was
still being transferred to the column and subsequently into the source of the GC/MS. Analysis of humid samples
with a Nafion drier used to dry the samples before concentration gave very good peak shapes and reproducibility,
indicating that water from the samples was the problem. These results indicate the mass spectrometer, as a GC
detector, is much more sensitive to moisture than the FID and FID detectors on GC/MD systems. Moisture
condensation in the MMS using the same temperature and configuration as with the GC/MD systems did not remove
sufficient moisture to avoid moderation of compound ionization.
Since operation of the MMS at 1? C was optimal for the GC/MD systems, but not optimal for the GC/MS
conditions, the temperature of the MMS was lowered to OP C and the experiments were repeated. At a MMS
temperature of ff C and loading 0.5 L of a 70% relative humidity sample, the chromatograms were much more
reproducible and peak shapes were greatly improved. In addition, polar compounds, such as cthanol, isopropanol
and diethyl ether were observed and peak shape was generally good.
Figure 2 presents an ambient air sample from Houston, Texas analyzed under these conditions. The sample
was diluted by a factor of approximately three with dry nitrogen as is standard operating procedure. A load volume
of 0.5 L of diluted, humid sample was concentrated and injected into the analytical system. As can be observed, the
chromatography for this sample was acceptable and sensitivity was very good.
An attempt was made to analyze the same sample at a load volume of 1.5 L to obtain increased sensitivity
and to determine if load volume had an effect on the GC/MS system. The results of this analysis are shown in
Figure 3. Sensitivity was lost due to the increased amount of water delivered to the mass spectrometer. It was
concluded that the condensation zone of the MMS was of sufficient size to condense the water from a 0.5 L load
volume of this sample, but did not allow adequate surface area or contact time to condense the moisture from a
308
-------
larger load volume. This loss of sensitivity also suggests that the GC/MS system may have a critical mass of
moisture above which sensitivity is rapidly lost.
Since it is desirable to load larger amounts of sample to increase sensitivity, a means of increasing water
removal efficiency was developed. The original MMS system contained a cooled zone approximately 1.5" long in 1/8"
o.d. tubing. The size and length of the cooled zone limits the residence time and surface area available to condense
water. In order to increase surface contact and residence time in the cooled zone, a new design was developed as
shown in Figure 4. The MMS cooled zone consists of an 8 inch piece of 1/16" o.d. stainless steel tubing spiral wound
around an 80 W cartridge heater. Active cooling, as opposed to passive cooling with the original design, was achieved
with a cryogenic solenoid valve and temperature controller. The cartridge heater was used to remove the water from
the MMS system after sample injection by heating to 150" C and backflushing with dry nitrogen for 10 minutes.
A second problem observed with the original MMS design which did not affect the operation of the GC/MD
systems, but is a consideration with GC/MS, is the buildup of water in the transfer lines after multiple analytical
runs. The passive cooling of the original MMS design contributed to this problem by allowing cool nitrogen gas to
contact the sample transfer lines and provide multiple, uncontrolled sites for condensation. This problem was
addressed by sealing the openings in the top of the cryotrap and venting spent nitrogen outside of the interface.
Transfer lines were carefully heat traced to provide only one cool (<80*C) zone, the MMS, where water could
condense.
Initial experiments conducted with the redesigned MMS system indicate that this system may be capable of
removing most of the water from ambient samples and allow loading of larger sample sizes. Operational
temperatures are being evaluated, but temperatures around ff C may be needed to adequately remove the moisture in
most samples. Such temperatures may affect polar compound recovery as observed in the initial study, but allow for
qualitative and semi-quantitative analysis of these compounds. Additional work is planned to determine polar
recoveries and reproducibility.
CONCLUSIONS
The Moisture Management System has proven to be an effective tool for reducing the amount of water
delivered to the column during analysis of VOCs using GC/MD systems. GC/MS instruments have proven to be
more sensitive to the amount of moisture injected. The original MMS design worked well with smaller sample
volumes, but did not allow larger samples to be analyzed, thus limiting sensitivity for undried samples. The
redesigned MMS shows promise of removing additional water and providing larger sample sizes. However, complete
evaluation of this new system has not been completed. As with the original MMS design, the operating parameters
must be optimized to obtain maximum reproducibility and recovery of polar organics. Parameters to be optimized
include MMS operating temperature, transfer line temperatures, bake-out times and temperatures, initial GC/MS
oven temperature, column flowrate and load volumes. Evaluation of these parameters and their effect on the analysis
of undried samples will continue.
REFERENCES
1. J.D. Pleil, W.A. McClenny and K.D. Oliver, "Dealing with water in GC/MS analyses of whole air samples",
presented at the 1989 Pittsburgh Conference and Exposition on Analytical Chemistry and Applied
Spectroscopy, Atlanta, GA, March, 1989.
2. EPA Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air. Method
TO-14; EPA-600/4-84-041, U.S. Environmental Protection Agency, AREAL, Research Triangle Park, NC,
1984.
3. J.D. Pleil, K.D. Oliver and W.A. McClenny, "Enhanced performance of nafion dryers in removing water from
air samples prior to gas chromatographic analysis", JAPCA. 37:244-248, 1987.
4. L.D. Ogle, R.B. White, DA. Brymer and M.C. Shepherd, "Applicability of GC/MS instrumentation for the
analysis of undried air toxic samples", in Proceedings of the 1989 EPA/APCA International Symposium on
Measurement of Toxic and Related Air Pollutants. VIP- 13, Air and Waste Management Association,
Pittsburgh, 1989, pp 824-829.
309
-------
D.B. Cardin and C.C. Lin, "Analysis of selected polar and non-polar compounds in air using automated 2-
dimensional chromatography", in Proceedings of the 1991 U.S. EPA/A&WMA International Symposium
-------
23*
26:%
SUM
25:80 lilt
Figure 2. 0.5 L Ambient Air Sample; MMS at CPC.
J
3986 SC«<
25:68 T1IC
Figure 3. 1.5 L of Ambient Air Sample (same as Figure 2); MMS at (? C.
311
-------
,80 WATT CARTRIDGE HEATER
2
VI
0
8.
-------
DETERMINATION OF VOCS IN AMBIENT AIR AT 0.1 PPBV FOR
THE CLEAN AIR STATUS AND TRENDS NETWORK (CASTNet)
Michael G. Winslow
Dwight F. Roberts
Michael E. Keller
Environmental Science & Engineering, Inc. (ESE)
P.O. Box 1703
Gainesville, FL 32602.
ABSTRACT
The 1990 Clean Air Act Amendments (CAAA) will require laboratories to analyze ambient air
samples for a wide range of volatile hazardous air pollutants (HAPS) at levels below 1 ppbV. One of
the air pollution monitoring requirements of the 1990 CAAA is for state and local air pollution control
agencies to track air toxics emission reductions, develop air toxics emission profiles, and identify
unknown sources of toxic air pollutants. Under the Clean Air Status and Trends Network (CASTNet),
Environmental Science & Engineering, Inc. (ESE) has the responsibility to reactivate and operate the
Urban Air Toxics Monitoring Program (UATMP). This network has been supported by the
Environmental Protection Agency's (EPA's) Office of Air Quality Planning and Standards (OAQPS) since
1988. During 1993-94, ESE will monitor air toxics at several urban sites in the Eastern United
States and analyze samples collected from that network. State and local agencies will be responsible for
collecting 24-hour integrated air samples every 12 days. These samples will be analyzed by ESE for
volatile organic compounds (VOCs), carbonyls, and metals. This paper presents a description of the
analytical approach that will be employed for the determination of VOCs collected in passivated
stainless-steel canisters. The results of system performance and method detection limit (MDL) studies
are also presented.
INTRODUCTION
EPA Compendium Method TO-14, "Determination of VOCs in Ambient Air Using Summa®
Passivated Canister Sampling and Gas Chromatographic Analysis",1 has generally been the reference
method of choice for the sampling and analysis of ambient air samples for a wide range of VOCs. Analysis
by gas chromatography/multidetector (GC/MD) or gas chromatography/mass spectrometry with
selected ion monitoring (GC/MS-SIM) traditionally has been necessary to determine concentration
levels in ambient air as low as 0.1 ppbV. However, both of these approaches limit qualitative and
quantitative information to the target compounds only, and target compound identification is not
positive and often questionable when analysis is performed by GC/MD. Furthermore, the semi-
permeable membrane (Nafion) dryer, traditionally used to eliminate excessive moisture from
samples, limits reliable quantiative and qualitative determinations to non-polar VOCs due to the
coincidental elimination, or partial elimination of polar analytes from the sample stream.
This paper presents results of a preliminary evaluation of an Entech Model 2000 automatic
cryogenic preconcentration system equipped with a 16-position autosampler manifold and interfaced to
a Hewlett-Packard (HP) 5890 GC/Finnigan Incos 50 MS operated in the full-scan mode (GC/MS-
SCAN). This system will be employed by ESE to determine VOCs in ambient air as part of the CASTNet
air toxics monitoring program. The system utilizes a moisture control approach that allows 0.1 ppbV
concentrations levels to be acheived for target VOCs. In addition, it will allow qualitative and semi-
quantitative data to be collected for non-target VOCs, non-polar and polar alike.
EXPERIMENTAL
Canister Preparation
To determine ambient VOC concentrations with acceptable precision and accuracy at
concentration levels as low as 0.1 ppbV, it is imperative that the laboratory establish the cleanliness
and integrity of each sampling canister. ESE utilizes an Entech Model 3000 Canister Cleaning System
for cleaning and leak checking canisters.
Canister leak checking is performed prior to canister cleaning. The canisters are connected to
the system manifold via 1/4-inch stainless-steel tubing and the canister valves are left closed. The
high vacuum pump is engaged and the vacuum pressure is monitored using the in-line pirani gauge (0-
2000 mtorr sensor). If a significant leak exists, it will manifest itself by not allowing the system to
313
-------
quickly evacuate to approximately 0.3 torr, and/or the system pressure will quickly rise once the
high vacuum pump is disengaged. Quick disconnects are situated between each canister and the manifold
to allow easy isolation of individual canisters which may have leaking valves. After the repetitive
fill/evacuate cycling is completed, the high vacuum pump is used for final canister evacuation. Leaks
will not allow the canisters to achieve a vacuum reading of approximately 0.5 torr.
As many as eight 6-liter canisters are cleaned simultaneously over a 4-hour period through
automatic, unattended cycling between canister filling and evacuation modes. Each of the canisters is
heated to 100°C for the duration of the cleaning process. A rough-vacuum oilless diaphragm pump
cycles between eight 8-minute fillings and eight 8-minute evacuations. Nitrogen vent gas from a liquid
nitrogen dewar is used for the fill gas and a humidification chamber containing ASTM Type II HPLC
grade water is used to add moisture to the fill gas in order to assist in the displacement of VOCs from the
interior surface of the canisters. In the fill mode the canisters are pressurized to about 25 psig. After
completion of eight cycles with the low-vacuum pump, the canisters are evacuted with a high-vacuum
oil-based pump for about 75 minutes to an absolute pressure of 0.5 torr, which is measured by an in-
line pirani gauge. A cryogenic trap containing liquid nitrogen is placed between the pump and the
system controller to keep oil vapor out of the system.
At the completion of the cleaning cycle, the canisters are pressurized to 30 psig with
humidified nitrogen and analyzed by GC/MS. No target analyte should be detectable at or above its
reported limit of quantitation. After analysis, each canister is re-evacuated to 0.5 torr.
Standards Preparation
Working calibration standards are prepared in canisters at a minimum of five concentration
levels in the range of 0.1 - 20 ppbV. These working standards are prepared from stock calibration
mixtures containing approximately 100 ppbV of each of the target analytes in dry nitrogen in high
pressure (2000 psi) cylinders. These calibration mixtures are purchased from Scott Specialty Gases
and are certified to + 5 % for each of the VOCs.
The working calibration standards are generated by dynamic flow dilutions of the purchased
mixtures with cleaned, dry zero grade nitrogen humidified with ASTM Type II HPLC-grade water. ESE
utilizes an Entech Model 4560 Dynamic Dilution System for this process. This system utilizes up to
six mass flow controllers for simultaneous blending from multiple cylinders. Canisters are generally
filled to pressures of 15 to 25 psig. Newly prepared calibration standards are allowed to equilibrate at
least 24 hours before analysis.
The 40 compounds listed in method TO-14 (ethyl toluene was not included) were evaluated for
this paper.
Instrumental Analysis
Canister samples are analyzed with an Entech Model 2000 automated VOC cryogenic
preconcentrator equipped with an Entech 2016 16-position autosampler manifold and interfaced to a
HP 5890 GC/ Finnigan INCOS 50 MS operated in the full scan mode (35 to 270 amu). The
preconcentrator system is under software control (IBM compatible) and a QA/QC report is a standard
feature for documentation of actual run conditions.
Canisters are attached to the manifold with 1/8 inch stainless-steel tubing. A leak check of the
system is then performed. Sample flow is set under mass flow control to 150 mL/min. to give an
integrated total volume of 1000 mL. Both subambient and pressurized canisters can be analyzed
because the system employs a mechanical pump to draw samples across the preconcentor. The internal
standards, bromochloromethane, 1,4-difluorobenzene, and ds-chlorobenzene, are also added to each
analysis at 150 mL/min. to a total volume of 100 mL. All rotary valves and transfer lines exposed to
the sample matrix are heated at 70-100°C.
The Entech preconcentrator employed by ESE for analyzing VOCs in amibient air is configured
with a 3-stage trapping approach which allows a large sample volume to be concentrated and the
moisture removed without loss of polar VOCs. The first stage is a high-volume cryogenic trap (1/8
inch nickel tube containing glass beads) cooled during sampling to -150 C with liquid nitrogen. After
cryotraping 1000 mL of sample, the first stage trap is heated rapidly to about room temperature and
gently purged (about 10 mL/min.) with 30 - 50 mL of nitrogen to transfer the trapped VOCs to a
second stage trap (1/8 inch nickel tube) containing hydrophobic Tenax TA held at 0°C. The second
stage trap is then heated to 170°C and backflushed with helium to a third stage megabore focusing
trap cooled with liquid nitrogen to -150°C. After heating and backflushing of the second stage trap is
complete, the third stage trap is then very rapidly heated to above 100°C to allow a rapid injection of
the VOCs onto the GC analytical column.
314
-------
The target analytes are separated on a DB-1 fused silica capillary column, 75 m x 0.32 mm
I.D., with a 1 micron film thickness. The chromatographic run is started at 35°C and held for 5 min.
A subambient starting temperature is not required because of the rapid transfer of VOCs from the
preconcentrator's third stage megabore focusing trap of the Entech preconcentrator to the head of the
GC column. Sharp peaks with 2-5 sec. widths are maintained, even for early eluting compounds. The
column is then temperature programmed at 6°C/min. to 180°C and then at 7.5°C/min. to 225°C.
The total run time for preconcentration of the sample and chromatographic separation is
approximately 60 min.
The MS scan rate is approximately 3 scans/sec. The Incos 50 turbomolecular source pump
draws about 170 U min. This pumping rate effectively prevents source pressure increases from any
residual sample moisture.
System Performance and Method Detection Limit Studies
The reproducibility and linerarity of the analytical system were evaluated over a two-week
period. Eight calibration standards and a blank were prepared as described above in separate 6-liter
canisters. The blank canister was prepared with humidified zero grade nitrogen. The calibration
standards contained the 40 compounds listed in Method TO-14 and were prepared at concentration
levels of 0.1, 0.25, 0.5, 1, 2, 5, 10, and 15 ppbV. Another canister containing the three internal
standards was also prepared. All canisters were attached to the autosampler manifold and analyzed,
after instrument tuning, as described above. Internal standards were added to each analysis at 5 ppbV.
The canisters were analyzed three times, with a week between analytical runs. Initial canister
pressures ranged from 10 to15 psig for the first day's run, from 1 to 2 psig for the second day's run,
and from -1 to -2 psig for the last day's run in order to evaluate reproducibility under different
canister pressures.
After analysis of the calibration standards on the second analysis day, an MDL study was
performed. Lower limits of detection for the target compounds were estimated for each instrument by
determining the method detection limits (MDLs) as specified by the U.S. Environmental Protection
Agency (U.S. EPA).2 Seven canister samples were prepared at about 1 ppbV for each of the target
compounds used the standards preparation system described above. The seven replicate samples were
then analyzed and quantitated against the average relative response factors (RRF) of the five calibration
standards ranging from 0.1 to 2 ppbV. The internal standard method of quanititation was employed.
The MDL for each target compound is calculated by multiplying the standard deviation of the seven
replicate concentration measurements by the appropriate one-sided t-value corresponding to n - 1
(6) degress of freedom. The corresponding t-value for seven measurements is 3.134.
RESULTS
The reproducibility, or precision, of the analytical system was evaluated by comparing the
responses of the target compounds and their corresponding internal standards over the three analysis
days at different calibration standard concentrations. In addition compound retention time variability
was evaluated. Table 1 summarizes response and retention data for vinyl chloride, benzene, and
hexachlorobutadiene (HCBD) at 0.1, 1 and 10 ppbV. These compounds represent low boiling (vinyl
chloride = -13.4°C), middle boiling (benzene = 80.1°C, and high boiling (HCBD = 215°C) VOCs,
respectively. For each of the three compounds listed in Table 1, the relative stardard deviations
(RSDs) of the absolute responses for the three analysis days over the three concentration levels
averaged less than 6 percent. The relative response factors averaged less than 5 percent. The RSDs of
the absolute responses of each of the internal standards for the three analysis days and within each
analysis day over the three concentration levels averaged less than 5 percent. Retention times had
ranges no greater than .03 minutes for the three compounds. Response and retention time results for
the other target VOCs were similar.
Table 2 summarizes calibration data for vinyl chloride, benzene, and HCBD on the second
analysis day. The compound responses were very linear over the 0.1 to 2.0 ppbV range, as measured
by either a linear regression curve or the RSD of the RRFs. Table 2 also shows the corresponding MDL
calculations. Only six of the seven 1 ppbV replicate samples were used because one of the replicates
was improperly prepared. The higher t-value of 3.365 was therefore used . Sample concentrations
were calculated using the RRFs of the samples and the average RRF of the standard curve.
The calculated MDLs averaged 0.13 ppbV for all the TO-14 compounds. Nearly 40% of the
MDLs were less than 10% of the 1 ppbV nominal concentration analyzed. This indicates that the
replicate analysis for those compounds should have been targeted at a lower concentration (probably
0.5 ppbV). The 0.1 ppbV responses for all the TO-14 compounds were well above background noise, in
315
-------
the range of 10 - 50,000 area counts. Compound responses were linear over the 0.1-1 ppbV range,
and generally demonstrated a quadratic fit acoss the full calibration range (0.1 - 15 ppbV), which
would be expected of the typical quadrupole MS.
Polar VOCs
Although this paper does not discuss experimental results for polar VOC determinations using
the analytical system described above, preliminary analysis of eight polar VOCs of interest have been
promising. The compounds that are currently being investigated include acetone, acetonitrile, acrolein,
acrylonitrile, methyl ethyl ketone, methyl isobutyl ketone, methyl methacrylate, and vinyl acetate.
Addition of particular polar VOCs to the canister target analyte list in the CASTNet program will be
dependent upon results obtained from an evaluation of the unknowns data obtained from GC/MS-SCAN
analysis and experimental determinations of their stability in canisters.
CONCLUSIONS
Routine measurement of ambient VOCs at 0.1 ppbV by GC/MS-SCAN can be acheived with
acceptable precision and accuracy using commercially available instrumentation. This has been
demonstrated with a analytical system comprised of an Entech preconcentrator, an HP GC, and a
Finnigan MS. The system will be employed for VOC determinations in canister samples collected for the
CASTNet air toxics monitoring program.
ACKNOWLEDGEMENTS
The authors thank Norm Staubly for his laboratory assistance.
REFERENCES
1. W.T. Winberry, Jr., N.T. Murphy and R. M. Riggin, "Method TO-14," Compendium of
Methods for The Determination of Toxic Organic Compounds in Ambient Air. EPA-
600/4-89-017, U.S. Environmental Protection Agency, Research Triangle Park, NC, 1988.
2. Federal Register. "Definition and procedure for the detemination of the method detection
limit," Code of Federal Regulations. Part 136, Appendix B, Oct. 26 (1984).
316
-------
Table 1: Area Response and Retention Time Reproducibility of
Selected VOCS
Pressurized
2/17/93
10 ppbV
Resp. (x1Q3)
IS Resp. (x1Q3)
RflF
Ret. Time
1 ppbV
Resp. (x103)
IS Resp. (x1Q3)
RflF
Ret. Time
0.1 ppbV
Resp. (x103)
IS Resp. (x103)
RHF
Ret. Time
VC
2058
764
1.35
4.03
278
808
1.72
4.03
29.7
739
2.01
4.02
Bzn
2379
2078
0.57
10.16
302
2143
0.70
10.15
27.1
1968
0.69
10.15
HC8D
883
1665
0.27
28.11
124
1652
0.38
28.11
13.5
1479
0.46
28.10
Atmospheric
2/23/93
VC
2067
702
1.47
4.02
271
762
1.77
4.04
27.9
724
1.93
4.02
Bzn
2256
1947
0.58
10.15
279
2000
0.70
10.16
26.1
1894
0.63
10.14
HCBD
910
1610
0.28
28.11
111
1577
0.35
28.11
13.8
1440
0.48
28.09
Sub- Atmospheric
3/04/93
VC
1996
732
1.36
4.03
290
788
1.84
4.03
31.3
748
2.09
4.02
Bzn
2195
1978
0.56
10.16
300
2094
0.72
10.16
29.8
1985
0.75
10.15
HCBD
843
1582
0.27
28.12
106
1614
0.33
28.11
14.8
1551
0.48
28.11
317
-------
Table 2: Calibration and MDL Results for Selected VOCs
Calibration:
Standard VINYL CHLORIDE
(ppbV) Response RRF
2 515511 1.67
1 270598 1.77
0.5 141763 1.93
0.25 74681 2.03
0.1 27932 1.89
0 0
L.R. Corr. Coeff. 0.9995
y-lntercept 6895
Average RHF 1.86
%RSD of RRFs = 7.5
BENZENE
Response RRF
575498 0.700
279430 0.699
132341 0.672
68616 0.733
26112 0.652
3248
0.9993
-5397
0.691
4.5
HCBD
Response
219910
111351
51646
27152
13810
254
0.9997
264
0.376
13
RRF
0.340
0.353
0.355
0.371
0.461
Method Detection Limit Determinations:
VINYL CHLORIDE
1 ppbV RRF Cone.
1 1.93 1.04
2 1.82 .978
3 1.92 1.03
4 1.92 1.03
5 1.92 1.03
6 1.92 1.03
Average Cone: 1.02
Standard Deviation: 0.022
t-value: 3.365
Calculated MDL: 0.07
(ppbV)
BENZENE
RRF Cone.
0.638 0.923
0.628 0.909
0.626 0.906
0.594 0.860
0.592 0.857
0.599 0.867
0.887
0.029
3.365
0.10
HCBD
RRF Cone.
0.355 0.944
0.337 0.896
0.335 0.891
0.396 1.05
0.404 1.07
0.382 1.02
0.978
0.079
3.365
0.27
318
-------
DEVELOPMENT OF METHODS FOR SAMPLING AND ANALYSIS OF
VENT STREAM EMISSIONS FROM GLYCOL DEHYDRATION UNITS
IN THE NATURAL GAS INDUSTRY
Larry D. Ogle, Curtis O. Rueter, and Dirk L. Reif
Radian Corporation
P. O. Box 201088
Austin, Texas 78720
James M. Evans
Gas Research Institute
8600 West Bryn Mawr Avenue
Chicago, Illinois 60631
ABSTRACT
Emissions of benzene, toluene, ethylbenzene and xylenes (BTEX) and other volatile organic compounds
(VOCs) from the regenerator still vent of glycol dehydration units have become an important environmental concern
for the natural gas industry. The objective of this program was to develop and validate sampling and analytical
methods for the measurement of BTEX and VOCs from these sources. Development of new methods for still vent
streams was necessary due to concerns about the accuracy of existing methods with such a sample matrix. Glycol
vent streams generally have low flow rates and are comprised of approximately 90% steam with a balance of vapor
phase organic compounds. Such a stream does not lend itself to sampling and analysis by standard methodologies,
such as VOST techniques.
A total of five direct measurement sampling techniques and five screening methods for still vent emissions
were evaluated in a preliminary field test conducted in early 1992. The most promising methods were then further
evaluated in extensive laboratory testing to arrive at a limited number of sampling methods for further field
evaluation. Sampling methods evaluated during the field evaluation included Total Capture Condensation (TCC) of
the vent stream, slip-stream condensation of a portion of the vent stream, natural gas collection in canisters, high
pressure rich glycol sampling in stainless steel bombs, atmospheric pressure rich and lean giycol sampling in VOA
vials and a field glycol/gas separation unit.
INTRODUCTION
Glycol dehydration units (typically triethylene glycol) are used to remove water from produced gas streams to
prevent hydrate formation and corrosion in the pipeline. The natural gas contacts the glycol countercurrently in an
absorber column to remove the water. Glycol exits the bottom of the absorber and on most larger units goes into a
flash tank where pressure is reduced and most of the volatile organics, primarily methane, absorbed in the glycol are
removed. The glycol then enters a still and reboiler where water and any residual organics are distilled from the
glycol. Lean glycol is then pumped back into the absorption column.
During the absorption step at high pressure, methane and other organic compounds are absorbed into the
glycol from the natural gas. Aromatic compounds, primarily BTEX have a high affinity for the glycol. Small
amounts are removed in the flash tank, but most along with higher molecular weight aliphatic hydrocarbons, are
removed during the distillation step. These compounds are then potential emissions from the still vent stack which
may be regulated under the 1990 Clean Air Act Amendments.
The purpose of this program was to develop sampling and analytical techniques applicable to the accurate
measurement of the emissions from the still vent stacks under the sponsorship of the Gas Research Institute (GRI).
Emission measurements for glycol dehydration units have been made using a variety of sampling points, sampling
protocols and analytical techniques since standard methods are not available. Most methods used prior the initiation
of this project have focused on the indirect measurement of emissions by a mass balance on the rich and lean glycol
streams. Therefore further evaluation of these methods, plus a direct measurement method of the still vent stream
was needed to accurately determine emissions.
Direct measurement of the organic emissions from the still vent stream is very difficult.1'2 The stream
temperature is in the range of 200-22CT F and contains over 90 percent water with a balance that is primarily organic
compounds (percent levels). Problems encountered when trying to sample this stream include:
319
-------
• Collecting or directly measuring the organics in the presence of extremely high levels of water;
• Collecting a representative sample in a low, variable flow stream that most likely contains condensed
material; and
• Accurately measuring a flowrate in a steam/organic stream.
Since analytical methods are generally well established for analysis of hydrocarbon emissions, generally using
capillary gas chromatography with flame ionization detectors, this paper will primarily focus on the development of
sampling and flowrate measurements for the glycol vents.
Experimental and Discussion
Input to the program is provided by an Industry Working Group which includes representatives from several
industry organizations. Initial methods evaluated included several suggested by the Industry Working Group plus a
number of standard EPA methods and methods from other sources. These methods were evaluated in a preliminary
Held experiment at a eight million standard feet per day glycol unit in South Texas. These methods included direct
measurement techniques for the vent stack and several indirect measurement (or screening) techniques and collection
and analysis of natural gas in canisters for use in modeling the emissions of hydrocarbons. Those methods evaluated
in this effort are listed in Table 1. The methods which were deemed worthy of further study and those eliminated
are indicated. Methods based on absorption of the organics onto sorbent materials, such as with the VOST
technique or NIOSH charcoal tubes were quickly eliminated. Methods based on color changes, such as Drager or
Sensidyne tubes were also eliminated. As a general rule, organic levels in the still vent stack were much too high to
use absorption as a collection technique. Swelling and heating of the material in the tubes caused the tubes to either
crack or the flow to be restricted.
Results from two of the stack methods, the Total Capture Condensation (TCC) and the Air Dilution, agreed
reasonably well and were considered to be an accurate measurement of the still vent emissions. However, it was
determined that in order to obtain an accurate measurement of the emissions using the Air Dilution technique, an
accurate measurement of the stream flow must be made. This measurement proved to be very difficult to make due
to the stream composition. Therefore, an objective of the laboratory development phase was the identification of an
accurate flow measurement technique. A variation of the TCC technique was also to be evaluated in the laboratory
study which would require accurate measurements of the vent flow. This method, slipstream total capture
condensation, was developed as a means of collecting a portion of the total vent flow on units treating more than
approximately 10 MMSCFD and for which condensation of the total vent flow was impractical.
The collection of a small portion of the gas from the still vent in an evacuated canister gave results
approximately seven times higher than TCC. The high results were most likely due to delivery of liquids or aerosols
into the canister. Though this method did not work well in the initial field trials, it was decided to continue working
on it during the laboratory study because it required minimal setup and equipment as compared to the other direct
measurement techniques.
High pressure rich glycol samples are normally collected after the flash tank and before the regenerator to
measure emission of BTEX and hydrocarbons actually removed in the glycol regeneration. The mass balance
between the rich glycol and the lean glycol, after regeneration, is used to calculate emissions. In the initial study the
rich glycol was taken before the flash tank. This did not appear to greatly affect BTEX emission calculations as little
is removed in the flash tank. However, it did appear to bias the total hydrocarbon measurements high as compounds,
such as methane and ethane, normally removed in the flash tank were included in the measurement. Development of
the rich glycol collection in sample cylinders centered around the separation and measurement of volumes of the
liquid and gas phases once the pressure was released from the vessel. Atmospheric glycol laboratory development
centered around the minimizing the loss of volatile gases during collection.
Laboratory Techniques
In the portion of the laboratory studies focusing on the stack sampling techniques, a bench-scale unit
(Figure 1) was constructed to simulate the still vent stream. It consisted of a steam generator that produced 6.8 Ib/h
of steam and an organic injection apparatus. Organic liquid was pumped into a heated injection block to vaporize
the organics. The vaporized organics were swept from the injection zone into the steam by methane. The entire
system was heat traced to minimize condensation before the collection apparatus. Since the flowrate of the steam
320
-------
and the organics was known, the vapor phase concentration could be calculated and used to determine compound
recoveries from each collection technique.
Total Capture Condensation. The total capture condensation apparatus used in the laboratory study
consisted of a 5 gallon container in which a 5 foot long by 1" inside diameter coil of copper tubing was placed. The
copper tubing acted as a condenser for the steam and organics in the simulated still vent stream. A glass collection
vessel was used to collect the condensed liquid. The gas exiting the system could also be collected in Tedlar bags for
analysis. In most instances, the gas phase did not contribute significantly to the calculations as it primarily consisted
of the methane used as the sweep gas and small amounts of pentane and hexane which were not completely
condensed.
A total of six back to back duplicate collections were made using the TCC system. The results from the
collection of selected organic compounds is shown in Table 2. The gas flowrates for the steam and organic phase
along with individual compounds tested are included. These ratios of organics to water and the flowrates are
indicative of those measured in the field trial study. Based on the results of the laboratory study, a field collection
apparatus (Figure 2.) was constructed from a 55-gallon drum and 50' and 1" of copper tubing.
Air Dilution System. The air dilution system was designed to dilute the steam/organic stream by a factor of
approximately 10 and then collect the resulting air, moisture and organic stream in a canister or Tedlar bag for gas
phase analysis. Identical flowrates and concentrations were to be used as tested for the TCC apparatus. However,
during calculations of gas phase concentrations after dilution, it was determined that the concentrations would be very
dose to the lower explosion limit for many of the compounds. Despite the fact that explosion-proof equipment was
used in the initial field trial, it was decided that this method posed too much danger to operators and equipment to
be used as a industry recommended method. Therefore, the air dilution method was dropped from consideration.
Canister Sampling. Canister samples were collected from the simulated vent stream by attaching a canister
with a Veriflow* flow regulator to a heated port in the vent line. In order to avoid the condensation problems
observed in the field trial: the sample lines, filter, regulator and canister valve were all heated with heat tape. In
addition, samples were taken from three different configurations; 1) with the sample probe inserted at a 90 degree
angle and parallel (horizontal) to the vent; 2) with the sample probe 90 degrees from the vent inserted from the top
(vertical) to the vent; and 3) as in #2 except the probe in the stream was bent with the opening facing away from the
direction of flow to minimize aerosol entrainment. The canisters were diluted before analysis by GC/FID. As
before, concentrations were highly variable indicating condensation in the sample lines and canister. This method
was also eliminated as a possible collection technique. Acceptable results also could not be obtained from an on-line
GC which was set up for comparison to the vapor phase analysis of the canisters. It also was plagued with problems
caused by the condensation of water and organics in the sample lines or aerosols carried into the injection loop.
Slipstream Total Capture Condensation. A variation of the TCC technique was evaluated for use of large
units, greater than 10 to 20 MMSCFD. On very large units, condensation of the entire stream is not feasible and a
slipstream from the vent must be used to determine the emission levels. However, in order to determine emissions
from the vents, the stream flowrate must be accurately measured. In the laboratory study, the flowrate is known
since the addition rate of steam and organics is controlled and constant. In a field situation, flowrate measurement is
much more difficult as will be discussed later.
Care was taken to heat the slipstream lines and insulate the still vent to avoid condensation or collection of
aerosols. A total of eight replicate collections were performed, each of which was collected at 0.5 liters/minute and
for 30 minutes. With the calculated flowrates for the total stream flow, recoveries for BTEX, hexane, pentane and
octane average 114% and the recovery of water averaged 96%. The slip-stream technology was considered to be
worthy of additional testing in the field evaluation portion of the program.
High Pressure Glvcol Sampling and Analysis. Glycol samples collected at elevated pressures in high pressure
sample cylinders can release considerable amounts of gas and produce copious amounts of foam upon pressure
release. In order to develop methods for handling the foam, collect gas released from the cylinders, determine losses
of target compounds during atmospheric pressure collection and measure gas volume, synthetic samples of TEG with
BTEX, methane or propane, pentane, and hexane were generated. Methane or propane was added to the cylinder to
pressures around 50 psig to provide a volatile gas. The pressure in the cylinder was brought to 900 psig using an
HPLC pump delivering TEG to which the other compounds and water had been added. Upon release of the
pressure, the volatile organic would degas from the liquid simulating a field sample.
321
-------
Techniques were developed for collection and measurement of the volatile fraction which degasses from the
glycol phase. A positive displacement apparatus consisting of a sealed chamber containing a Tedlar bag for sample
collection and a dry gas meter to measure the volume of gas displaced by the sample gas flowing into the bag was
built. With some minor modifications this system worked well in collecting and measuring gas volumes. In addition,
two techniques designed to help eliminate foaming were evaluated. These included heating the cylinder to reduce
viscosity of the liquid and adding methylene chloride to the cylinder before sample collection to reduce sample
viscosity and act as a keeper for less volatile organics in the glycol (pentane and larger). Heating did not seem to
help, but results from the methylene chloride addition were favorable. Recoveries of the non-gaseous components in
the glycol ranged from 103 to 126%. The high pressure rich glycol sampling technique, with and without addition of
methylene chloride, was also slated for further study in the field evaluation testing portion of the program.
Total Flow Measurement. As mentioned earlier, total flow measurement in the still vent stream is very
important in determining emission rates by any method that does not collect the entire stream. Due to the fact that
the still vent stream is hot and wet, consists of a 2" pipe and has flowrates that can fluctuate wildly due to heating a
cooling cycles of the reboiler, traditional flow measurement devices such as dry gas meters and pitot tubes are not
applicable. In the laboratory study, a hot wire anemometer and a mass flowmeter were evaluated for measurement
of flowrates. The mass flowmeter did not work at all and the hot wire anemometer exhibited a relatively short cycle
fluctuation of less than a minute. However, the mean measurement for the hot wire anemometer during these cycles
was very close to the calculated flowrates. A decision was made to further test the hot wire anemometer in the field
evaluation studies. Orifice plates, which were not evaluated in the laboratory, were also to be evaluated in the field.
CONCLUSIONS AND PLANS FOR FUTURE WORK
Results from the laboratory studies led to the elimination of collection of small samples from the still vent in
canisters and the air dilution of the vent stream. Methods slated for further evaluation at working glycol dehydrators
include Total Capture Condensation, Slip Stream Condensation, Rich Glycol Sample Cylinders and Atmospheric
Rich/Lean Glycol Collection. In addition, two flow measurement techniques, the Hot Wire Anemometer and Orifice
Plates, will be evaluated. A field gas-glycol separation apparatus suggested by a member of the Industry Working
Group is to be evaluated. It consists of two burettes connected in series with a water displacement system. This
apparatus will be evaluated with rich glycol samples as an alternative to the collection of high pressure rich glycol
samples in cylinders. The Total Capture Condensation Method will be used as the baseline method for comparison
of the results from all of the other methods.
The results of these field evaluation experiments and the laboratory studies will be statistically evaluated to
statistically determine the precision and accuracy of each method developed in this program. The method(s) giving
the best precision and accuracy combined with ease of use and cost will be selected as the method of choice for
determining still vent emissions from glycol units. The selected method(s) will be further tested in a series of field
validation experiments to be conducted April through August of 1993. These experiments will be performed at glycol
units operating under different conditions (e.g., a range of treatment rates, different types of glycol, with and without
flash tanks and on sweet and sour gas units). Further modification of the methods developed may be necessary based
on these validation studies.
REFERENCES
1. C.O. Rueter and J.M. Evans, "Development of Sampling and Analytical Methods for Emissions from Glycol
Dehydration Units," in Proceedings of the 1992 GRI Glvcol Dehvdralor Air Emissions Conference. GRI-92-0279,
September, 1992, pp.87-102.
2. C.O. Rueter, L.D. Ogle, D.L. Reif and J.M. Evans, "Development of Sampling and Analytical Methods for
Measuring BTEX and VOC Emissions from Glycol Dehydration Units," presented at the SPE/EPA Exploration and
Production Environmental Conference. SPE No. 25944, San Antonio, Texas, March 1993.
3. PA. Thompson and J.M. Evans, "Estimating Emissions Using GRI-DEHY," in Proceedings of the 1992 GRI
Glvcol Dehvdrator Air Emissions Conference. GRI-92-0279, September 1992, pp. 153-175.
322
-------
Table 1. Preliminary methods considered for glycol dehydrator vent emission monitoring.
Technique
Continue Development?
Comments
Direct Measurement Techniques:
1. Air Dilution
2. Total Capture Condensation
3. Canister Samples
4. Modified VOST
Indirect Measurement Techniques:
1. High Pressure Rich/Atmospheric
Pressure Lean Glycol
Yes
Yes
Yes
No
Yes
Results considered good.
Results considered good.
Results high; focus on collection
techniques.
Serious problems caused by high
water/organic stream.
Focus on development of handling
techniques.
2. Atmospheric Pressure Rich/Lean Glycol
3. Natural Gas Canisters
4. Screening Indicator Methods NIOSH,
Sensidyne and Drager Tubes
Table 2.
Parameter
Steam
Organics (Total)
Benzene
Toluene
Xylenes
Methane"
n-Hexane
Pentane
Octane
Spiked compounds and
Gas Flow
(Liters/Mm.)"
87.40=
9.70
1.07
330
1.36
0.30
1.26
1.16
1.26
Yes
Yes
No
recoveries from the total
Percentage of
Total Gas Phase
Dreamt*1
11
34
14
3
13
12
13
Focus on handling
Used with modelin
techniques.
g.
Problems similar to those with VOST.
capture condensation technique.
Calculated
Recoveries5
98.1
96.8
96.8
99.4
103
ND
92.8
86.3
99.3
Standard
Deviations"
5.41
5.43
7.29
9.84
12.4
ND
9.15
13.3
6.38
"Based on 21?F from the steam generator.
"Based on total capture condensation from field samples. The paraffin hydrocarbons reflect the mass from
unidentified VOCs.
'Based on 6.8 Ibs/hr of water input.
"Methane is 94% commercial grade methane. Impurities include ethane, propane, isobutane, and butane.
'Based on 12 observations.
323
-------
- *- STEAM/ORGANIC
ITAMETER^ [j
ORGANIC LIQUID
PUUP
- HEATER
CONTROLLER
Figure 1. A bench-scale unit simulating the still vent stream.
Figure 2. Total Capture Condensation Sampling Train
324
-------
Session 8
S S Canisters,
Use and Techniques
-------
IMPROVED SENSITIVITY AND QUALITY ASSURANCE IN SUMMA CANISTER
ANALYSIS USING EPA METHOD TO14
Jon Wong Daniel B. Cardin, John T. Deschenes
Curtis & Tompklns, Ltd Entech Laboratory Automation
2323 5th Street 950 Enchanted Way, #101
Berkeley, CA 94710 Simi Valley, CA 93065
ABSTRACT
GC/MS Is the analytical methodology mandated by the USEPA Contract Laboratory Program
Statement of Work. GC/MS provides accurate quantitation and qualitative information using full scan
mode. Method detection limits reported in the CLP SOW are in the range of 0.2 to 1.7 ppbv". Using an
automated air preconcentrator coupled to a high sensitive quadrupole mass spectrometer, method
detection limits are determined ten times lower, exceeding TO14 requirements.
Quality Assurance from both the air preconcentrator and data system is described. Immediate
feedback from the QA reports automatically documents the quality of the analytical data.
INTRODUCTION
The analysis of toxic organic compounds in ambient air by GC/MS techniques requires sample
preconcentration to obtain detection limits in the sub-ppb range. Although such methodology has
been utilized for years, the achievement of detection limits specified in Title III of the 1990 Clean Air
Act Amendment (2-20 ppt) have not yet been reported. This has been due to a combination of
problems concerning moisture and CO2 management, difficulties in sampling large volumes quickly,
and limitations in absolute GC/MS sensitivities.
This paper describes a preconcentrator GC/MS configuration that approaches the low ppt
requirements of Title HI of the CAAA Detection limits for target compounds significantly exceed TO 14
requirements while collecting full scan data using a new, high sensitivity quadrupole mass
spectrometer. Concentration parameters are recorded to provide the operator an operational log during
unattended operation for improved QA/QC. Data system provides automated quality assurance
reports. The system can be configured for research or production oriented laboratories utilizing 1,16,
or 32 position automation.
EXPERIMENTAL METHOD
The system used was an air preconcentrator/autosampler (Entech 2000/2010) coupled to a HP
5890GC/ HP 5972 MSD. Analyses were performed using a Restek 60 m x 0.32 mm ID Rtx Volatiles
capillary column(1.5 micron film thickness). The outlet of the capillary column was directly inserted
into the MSD source. A personal computer controlled the Entech System giving system printouts on
air concentration parameters. Immediate feedback monitors system performance such as initial can
pressure, trap temperatures, actual volume sampled, desorptlon temperatures, and bakeout
temperatures. The MSD was controlled by an Apollo workstation using Target2 software to give
quantitation and quality control reports. QC reports include tune criteria, calibration specifications,
and internal standard retention time/area performance. Optimum parameters such as water
management, cryofocusing, and electronic pressure control improved chromatographlc resolution.
Retention time stability was superior during this study. The HP 5972 allows higher column flows up to
2 cc/min and improved pump oil provides low background. A newly designed electron multiplier
provides excellent signal/noise ratios.
Instrument Settings
Entech Concentrator Glass trap -150°C
Tenax trap 10°C
Cryofocusing trap -150° C
GC Condition.^ 35° C for 5 mins then 6 degree ramp to 210° and hold for 1 min.
EPC Program 2 cc flow (Approximately 15 psijtnitial time 0.5 mins then drop to 1.5
cc(approximately 8 psl) flow and hold at constant flow
MSD Conditions Tune: Std BFB Tune EM = 1650 Volts
Scan parameters 35-300 amu A/D 2A3 (approximately 1 scan/sec)
Solvent Delay 4.5 mins
Threshold set at 100 counts
327
-------
Standard Preparation
The standards were prepared In house and then compared to a Scott-Marrin reference
standard. Percent differences between reference std and In house blend were less than ten percent.
Gas components were purchased as a mix from RestekfVOA CYL IIQand back calculated from ngs to
ppbv. Freon 114 was obtained from Aldrich as pure and tedlar bag blended to a working ppmv level.
Other liquid TO14 compounds were prepared by the static dilution bottle technique2. From the
working ppmv levels, allquots were injected into a precleaned evacuated SUMMA canister and then
pressurized with humidified nitrogen from the gas side of a liquid nitrogen dewar. Cans were
pressurized to approximately 1500 mm Hg (approximately 15 psi) and allowed to equilibrate
ovemlght3.Standards were prepared at 50,10,1,0.2, and 0.02 ppbv.
Calibration
A sample size of 500 cm" was used and a five point calibration was performed at l,5,10,20,and
50 ppbv.(See Figure 1 for TIC of 10 ppbv Standard) Relative standard deviations were less than 10
percent in most cases. Software produces a CLP-like FomiS-Initial Calibration Report seconds after
data Is processed. Tune reports for BFB are automatically produced after acquisition.
Mdl Study
The method detection limit was determined by the procedure of Glaser.et al4, run an estimated
sample at five times the mdl, analyze seven replicates, calculate the mean, standard deviation(SD) and
multiply the SD by Student-t value of 3.143. To accurately simulate 0.2 ppbv sample two method
detection limit studies were performed: first, a 100 cm3 sample from a 1.0 ppbv can and analyzed
seven times(See Table l)and second, a sample size of 500 cm3 from a 0.2 ppbv can and analyzed seven
times.(See Table 2.) Benzene contamination occurred in both studies and in the second study,
contamination of toluene and the high boilers suggest a dirty can.
Extracted Ion chromatograms are provided for the gases at lOppbv, 0.2 ppbv, and 20 pptv(see Figure
2) Note actual amounts of gases in figure 2.
20 Pptv Study
A 20 pptv was prepared and analyzed with a 500 cm3 sample size (See Figure 3 for full scan
spectra)
To enhance the sensitivity of the MSD a 20pptv sample was analyzed with the electron multiplier
operated at BFB tune plus 400 volts and response improved by a factor of ten with some increases in
noise.(See Figure 2 and 3 lower graph)
CONCLUSIONS
Sensitivity is improved by optimizing the following parameters:
1. Water management allowing direct capillary insertion and polar compound analyses
2. Cryofocusing and Electronic Pressure Control improving gas chromatographic resolution
3. Using a low background high sensitivity benchtop quadrupole mass spectrometer
GC/MS full scan data is attainable at the pptv level. Raising multiplier 400 volts above tune enhances
response and accordingly signal to noise ratio. Method detection limits studies performed show
detection limits in the 10 to 200 pptv with most compounds in the 10 to 20 range. Reliable
quantitation level(RQL) suggested by Eaton5 put limits at SOpptv. Improvements in sensitivity call for
extremely clean cans. Contamination of common solvents used in an environmental laboratory is a
challenge for trace analysis. Cleanliness and control of cans is critical in trace analysis. Larger sample
sizes will improve quantitation limits but a larger trap may be needed. A smaller sample size may be
used and still exceed CLP SOW CRQLs and EPATO14.
Quality assurance is available in the system software for immediate feedback on system
performance and quality analytical data. Automation of these reports reduce manual hand
calculations and facilitates operator productivity.
REFERENCES
1. U.S. EPA Contract Laboratory Program. Draft Statement of Work for Analysis of Ambient Air, U.S.
EPA, Office of Emergency and Remedial Response, Hazardous Site Evaluation Division, Analytical
Operations Branch, Washington D.C.(December 1991)
2. Same as reference 1'Volatile Organic Analysis of Ambient Air in Canisters", pp D21-D22
328
-------
3. J.H.M. Stephenson, F. Allen, T. Slagle, "Analysis of Volatile Organics in Air via Water Methods." to
Proceedings of the 1990 EPA/A&WMA International Symposium on Measurement of Toxic and Related
Air Pollutants. USEPA Report Number EPA/600/9-90/026 , pp 194-199
4. J.A. Glaser, D.L. Forest.G.D. McKee. S_A Quave. and W.L. Budde.Trace Analysis for Wastewaters",
Environmental Science and Technology. 15:1426(1981).
5. A. Eaton. "Estimation of Interlaboratory MDLs and RQLs". Environmental Laboratory April/May
1993. pp.10-14
W.T. Winberry, Jr., N.T. Murphy and R.M. Riggln. Compendium of Methods for the Determination of
Toxic Organic Compounds in Ambient Air.. EPA-600-4-84-041, US. Environmental Protection Agency.
Research Triangle Park, 1988
Figure 1. TIC of lOppbv TO14 STD ( SOOcc of 10 ppbv).
329
-------
Table 1. Seven Replicates of lOOcc of 1.0 ppbv.
Compound
Methc Exp
Blank Cone
MDL1 MDL2 MDL3 MDL4 MDL5 MDL.6 MDL7 AVG %REC STD MDL
Freon 1 2
Freon 114
Chloromethane
Vinyl Chloride
Bromomethane
Chloroethane
Trichlorofluoromethane
Freon 113
1,1-Dichloroethene
Methylene Chloride
1,1-Dlchloroethane
cis-1 ,2-Dichloroethene
Chloroform
1,1,1-Trichloroethane
Carbon Tetrachloride
1,2-Dichloroethane
Benzene
Trichloroethene
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
Toluene
trans-1 ,3-Dichloropropene
1 . 1 ,2-Trichloroethane
Tetrachloroelhene
1,2-Dibromoethane
Chlorobenzene
1 , 1 ,2,2-Tetrachloroethane
Ethylbenzene
Xylenes,m&p
o-Xylene
Styrene
4-Ethyl toluene
1 ,3,5-Trimethylbenzene
1 ,2,4-Trimethylbenzene
1 ,3-Dichlorobenzene
1 ,4-Dichlorobenzene
Benzyl Chloride
1 ,2-Dichlorobenzene
1 ,2,4-Trichlorobenzene
Hexchlorobutadiene
0.00
0.00
0.02
0.00
0.01
0.00
0.00
0.00
0.00
0.37
0.00
0.00
0.00
0.00
0.00
0.01
0.46
0.00
o.oo
0.00
0.16
0.01
0.01
0.00
0.01
0.01
0.02
0.03
0.13
0.05
0.02
0.02
0.02
0.04
0.04
0.04
0.06
0.03
0.08
0.02
0.21
0.21
0.52
0.47
0.15
0.39
0.18
0.24
0.25
0.31
0.25
0.25
0.24
0.24
0.21
0.24
0.33
0.24
0.25
0.20
0.43
0.20
0.25
0.25
0.24
0.26
0.28
0.29
0.69
0.32
0.26
0.31
0.31
0.32
0.30
0.29
0.15
0.29
0.25
0.29
0.24
0.19
0.64
0.46
0.17
0.22
0.20
0.25
0.28
0.75
0.27
0.27
0.28
0.28
0.24
0.28
0.75
0.26
0.27
0.22
0.54
0.23
0.27
0.26
0.26
0.27
0.32
0.31
0.79
0.35
0.26
0.33
0.32
0.35
0.31
0.31
0.19
0.31
0.31
0.34
0.25
0.19
0.62
0.45
0.17
0.22
0.21
0.26
0.28
0.86
0.28
0.28
0.29
0.28
0.24
0.29
0.80
0.26
0.28
0.23
0.53
0.23
0.27
0.26
0.26
0.27
0.32
0.31
0.81
0.36
0.26
0.34
0.32
0.36
0.32
0.32
0.21
0.32
0.31
0.34
0.25
0.20
0.62
0.45
0.17
0.22
0.20
0.26
0.28
0.93
0.29
0.28
0.29
0.28
0.24
0.29
0.85
0.25
0.27
0.22
0.55
0.23
0.27
0.27
0.27
0.28
0.33
0.32
0.83
0.36
0.27
0.35
0.32
0.37
0.32
0.33
0.23
0.33
0.34
0.36
0.25
0.23
0.65
0.45
0.20
0.23
0.21
0.26
0.28
0.96
0.28
0.28
0.29
0.28
0.24
0.29
0.83
0.26
0.27
0.23
0.55
0.23
0.27
0.27
0.27
0.28
0.32
0.31
0.81
0.36
0.26
0.34
0.32
0.35
0.32
0.32
0.19
0.32
0.30
0.33
0.24
0.19
0.64
0.45
0.17
0.22
0.20
0.25
0.27
0.87
0.29
0.28
0.29
0.29
0.24
0.30
0.80
0.26
0.27
0.23
0.54
0.23
0.27
0.26
0.27
0.27
0.32
0.31
0.81
0.36
0.27
0.34
0.33
0.36
0.33
0.33
0.22
0.33
0.33
0.35
0.25
0.23
0.62
0.47
0.16
0.23
0.21
0.25
0.28
0.79
0.28
0.27
0.29
0.28
0.24
0.29
0.78
0.26
0.27
0.22
0.54
0.23
0.27
0.26
0.26
0.27
0.33
0.30
0.81
0.36
0.26
0.33
0.32
0.36
0.32
0.32
0.20
0.33
0.32
0.34
0.25
0.22
0.63
0.47
0.18
0.22
0.20
0.25
0.27
0.79
0.29
0.28
0.29
0.28
0.24
0.29
0.78
0.25
0.27
0.23
0.54
0.24
0.27
0.26
0.27
0.27
0.33
0.31
0.80
0.36
0.27
0.34
0.32
0.36
0.32
0.33
0.22
0.33
0.34
0.35
0.25
0.21
0.63
0.46
0.17
0.22
0.20
0.25
0.28
0.85
0.28
0.28
0.29
0.28
0.24
0.29
0.80
0.26
0.27
0.23
0.54
0.23
0.27
0.26
0.27
0.27
0.32
0.31
0.81
0.36
0.26
0.34
0.32
0.36
0.32
0.32
0.21
0.32
0.32
0.34
115%
98%
119%
96%
111%
57%
114%
107%
112%
155%
115%
112%
118%
120%
113%
118%
103%
106%
111%
112%
90%
109%
107%
106%
107%
105%
111%
97%
98%
98%
95%
102%
99%
99%
96%
98%
100%
103%
97%
116%
0,00
0.02
0.01
0.01
0.01
0.00
0.01
0.01
0.00
0.08
0.01
0.00
0.00
0.00
0.00
0.01
0.03
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.01
0.00
0.01
0.01
0.01
0.00
0.01
0.01
0.00
0.01
0.01
0.01
0.02
0.01
0.02
0.01
0.02
0.06
0.04
0.03
0.04
0.02
0.02
0.02
0.02
0.24
0.02
0.02
0.01
0.01
0.00
0.02
0.10
0.02
0.01
0.02
0.02
0.01
0.00
0.02
0.02
0.02
0.02
0.02
0.04
0.01
0.02
0.02
0.01
0.02
0.02
0.02
0.05
0.02
0.05
0.03
* Note percent recovery calculated based on average minus method blank divided by expected concentration
Normal Sample Size SOOcc
Single Point Calibration at 500 cc of 10 ppbv
0.2 ppbv run a sample size of 10Occ at 1 ppbv
330
-------
Table 2. Seven Replicates of SOOcc of 0.2 ppb.
Compound
Freon 12
Freon 114
Chloromethane
Vinyl Chloride
Bromomethane
Chloroethane
Trichlorofluoromethane
Freon 113
1,1-Dichloroethene
Methylene Chloride
1,1-Dichloroethane
cis-1 ,2-Dichloroethene
Chloroform
1,1,1-Trichloroethane
Carbon Tetrachloride
1,2-Dichloroethane
Benzene
Trichloroethene
1,2-Dichloropropane
cis-1 ,3-Dichloropropene
Toluene
trans-1 ,3-Dichloropropene
1 ,1 ,2-Trichloroethane
Tetrachloroethene
1,2-Dibromoethane
Chlorobenzene
1 ,1 ,2,2-Tetrachloroethane
Ethylbenzene
Xylenes,m&p
o-Xylene
Styrene
4-Ethyltoluene
1 ,3,5-Trimethylbenzene
1 ,2,4-Trimethylbenzene
1 ,3-Dichlorobenzene
1,4-Dichlorobenzene
Benzyl Chloride
1,2-Dichlorobenzene
1 ,2,4-Trichlorobenzene
Hexachlorobutadiene
Method
Blank
0.00
0.00
0.02
0.00
0.01
0.00
0.00
0.01
0.00
0.50
0.00
0.01
0.00
0.02
0.00
0.01
0.48
0.02
0.02
0.02
2.01
0.03
0.04
0.05
0.03
0.07
0.14
0.32
1.02
0.36
0.17
0.32
0.14
0.33
0.40
0.41
0.39
0.30
0.68
0.15
MDL1
0.24
0.23
0.55
0.50
0.26
0.
.40
0.19
0,
.28
0.29
0,
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
.99
.29
.29
.30
.36
.24
.31
.39
.30
.31
.23
.98
.20
.29
.33
.26
.34
.46
.45
.15
.49
.32
.52
.50
.60
.52
.52
.31
.55
.14
.82
MDL2
0.23
0.23
0.54
0.48
0.28
0.39
0.19
0.27
0.29
0.95
0.29
0.29
0.30
0.35
0.25
0.30
0.40
0.30
0.31
0.23
0.95
0.20
0.29
0.32
0.26
0.34
0.43
0.45
1.14
0.47
0.30
0.48
0.49
0.56
0.47
0.46
0.27
0.47
1.13
0.82
MDL3
0.23
0.23
0.55
0.49
0.28
0.38
0.19
0.28
0.29
0.93
0.29
0.29
0.30
0.36
0.24
0.30
0.41
0.29
0.30
0.23
0.95
0.20
0.29
0.32
0.26
0.33
0.41
0.43
1.11
0.45
0.29
0.45
0.45
0.52
0.41
0.48
0.22
0.44
1.19
0.87
MDL4
0.24
0.22
0.52
0.50
0.23
0.39
0.19
0.27
0.29
0.90
0.29
0.29
0.30
0.36
0.24
0.31
0.46
0.30
0.30
0.23
0.95
0.21
0.30
0.32
0.27
0.33
0.41
0.43
1.11
0.46
0.28
0.42
0.44
0.51
0.40
0.46
0.18
0.42
1.25
0.89
MDL5
0.24
0.23
0.55
0.49
0.28
0.39
0.19
0.28
0.28
0.91
0.29
0.29
0.30
0.36
0.25
0.31
0.47
0.30
0.30
0.24
0.95
0.20
0.29
0.31
0.27
0.33
0.41
0.43
1.09
0.45
0.28
0.42
0.44
0.52
0.40
0.45
0.15
0.42
1.21
0.93
MDL6
0.23
0.23
0.51
0.48
0.21
0.40
0.19
0.28
0.28
0.92
0.29
0.28
0.30
0.35
0.24
0.31
0.46
0.29
0.30
0.23
0.94
0.20
0.29
0.32
0.27
0.33
0.41
0.43
1.10
0.45
0.29
0.43
0.45
0.51
0.39
0.44
0.13
0.42
1.24
0.95
MDL7
0.23
0.22
0.53
0.47
0.25
0.38
0.19
0.27
0.27
0.89
0.28
0.28
0.28
0.35
0.24
0.30
0.44
0.28
0.29
0.22
0.93
0.19
0.29
0.31
0.26
0.33
0.40
0.43
1.08
0.45
0.30
0.46
0.46
0.56
0.40
0.41
0.15
0.44
1.16
0.94
STD MDL
0.01 0.02
0.00
0.02
0.01
0.03
0.01
0.00
0.01
0.01
0.03
0.00
0.00
0.01
0.01
0.00
0.01
0.03
0.01
0.01
0.01
0.02
0.01
0.00
0.01
0.01
0.00
0.02
0.01
0.03
0.02
0.01
0.04
0.02
0.03
0.05
0.03
0.07
0.05
0.05
0.05
0.02
0.05
0.03
0.09
0.03
0.00
0.02
0.02
0.11
0.01
0.02
0.02
0.02
0.02
0.02
0.10
0.02
0.02
0.02
0.05
0.02
0.01
0.02
0.02
0.02
0.06
0.03
0.08
0.05
0.04
0.11
0.08
0.11
0.15
0.11
0.21
0.15
0.15
0.17
Normal Sample Size 500cc
Single Point Calibration at 500 cc of 10 ppbv
0.2 ppbv run with sample size of 500cc @0.2 PPBV
331
-------
Ion 50X10 amu tram ccs jO.d
; Run ai 01 21 AM PST on Frl M"ar 26, 1993 Peak
Abunoanco tan ez-'ao anu "rom «JO.d 1 Freon 12 10 pp
t Ion 64X10 amu tram ccs lO.d 2 Freon 114 10 ppb
SOOCCO-r 1 ,'l Ion S5.00 amu Irom ccsjo.d 3 Chloromethane 24 ppb
; 100000-= ft ' ', lon 94-°° amo lrom <»J\ \ Chloroethane 38 pptv
loooo^ it M ^
3000T i' \ i' \ • \ A "^ «,
iD 55 8.0 SJ
Figure 2. Extracted. Ion ^Jp^ggjatograms of Gases.
Scan 37 (5.5B4
100000-^
60000-5
20000-3 37 M 47 «
j Q d ' ' ,' / ,1
(H Vinyl Chloride 20 ppb
. 7H 85
j 40 50 SO 70 30 ' 90
' Mass/Charge
Scan 35 (5.563 mm) of 500cc_
Abundance
3000-1
1000-^
j -- - ; „
E_4.d SUBTRACTED SCALED
ej Vinyl chloride 400
40 50 60 70 80 30
Maxs/Charga
Scan 37 (5.534 rain) of 20ppn
' Abundance
T fi
400-^
300-^
I 37
100- «
i ,= , r. ?
J3.d SUBTRACTED SCALED
Vinyl chloride 40 p
2
78
\ 79 91 ^
\ 1
, 40 50 SO 70 30 30
I Mass/Charge
Scan 91 (5^35 mm) ol 500cc_20Dptv-b400.d SUBTRACTED SCALED
Abundance
5000^ 'SL
\ ~ Vinyl chloride 40 p
i *ooo-9 Enhanced at plus 4OO Volt
i 300)4
1 2000-= !
= ; j i '?/
,:? \. ? |
Fignre 3. MaSs Spectra of Hftnyl ChloridStt 4O.O.4.O.3S ppbv and eSianced O n^S,,.!
>ptv
Itv
tv
-------
A SIMPLE, ACCURATE PROCEDURE FOR PREPARATION OF ANALYTI-
CAL STANDARDS FOR TO14 INSTRUMENT CALIBRATION
John T. Deschenes, Daniel B. Cardin
Entech Laboratory Automation
950 Enchanted Way, #101
Simi Valley, CA 93065
ABSTRACT
Several procedures are currently available for preparation of 1014 Volatile Organic Com-
pounds in humidified zero air or humidified nitrogen at the high ppt - mid ppb range. Di-
luted standards can be generated directly from the pure (neat) compounds, or by purchasing
standard mixes at a vendor or NIST referenced concentration for subsequent transfer into a
SUMMA passivated canister. Dynamic dilution can improve the accuracy and precision of
standards preparation relative to static dilution by reducing the number of error causing
variables. Even initial transfer losses can be ignored in a system that is properly constructed
and operated.
A compact, economical dynamic dilution system is presented which simplifies the prepara-
tion of ambient air level standards. The system insures a high degree of reproducibility and
transfer accuracy by maintaining equilibrium conditions during canister filling operations.
Transfer losses are allowed to occur before canister filling (e.g.. before equilibrium is
achieved) and therefore do not affect the intended concentrations in the canister. Standards
can be pressurized into canisters without using a pump, and the relative sample humidity can
even be tailored somewhat.
INTRODUCTION
Calibration standards in the range of 0.5 - 20 ppb are needed for EPA Method TO 14 to
calibrate the GC/MS and verify proper operation of the combined preconcentrator/GC/MS
analytical system. Due to the low concentrations involved, special care must be taken to
insure that standards are prepared accurately and without contamination. Several methods
are currently being used to prepare these standards, as shown in Table 1. Each of these
techniques has its advantages and limitations. For example, any technique using methanol
(P&T or direct injection) may change the final matrix enough that obtained response factors
are not valid for non-methanol containing matrices, such as ambient air. In this case, trans-
mission through a Nation dryer (if used) or even recovery or stability in canisters could
change significantly when saturated with methanol.
333
-------
TABLE 1 - TO14 Standard Preparation Techniques
O Serial Dilution/Flash Vaporization
Dilute into liquid
Dilute into gas phase
O Direct injection of Methanol (P&T) Standard
O Purging of P&T Standard from Sparge Vessels
O Loop Injection of Calibrated Mix
O Dynamic Dilution of Calibrated Mix
Other techniques involving quantitative transfers at non-equilibrium conditions (Serial Dilu-
tion and Loop Injection) could result in fluctuating analyte concentrations due to interactions
with surfaces, especially during gas phase transfers where the matrix ( oxygen, nitrogen, and
argon) has little solvating influence to keep the VOC's off the container or syringe walls.
Dynamic Dilution is unique in that it can be operated fully under equilibrium condi-
tions. It shouldn't matter if analytes are initially adsorbing onto surfaces in
the diluter as long as a ''mass in = mass out" relationship is established
before commencement of canister filling. If this relationship is left undis-
turbed during the entire canister filling operation, the analyte concentra-
tions should remain constant. The following sections describe a simple
implementation of Dynamic/Equilibrium Dilution for preparation of VOC
concentrations in the 0.5 - 100 ppb range.
INSTRUMENTATION
The Entech 4560 Dynamic Dilution System is a 6 channel calibrated
standards generator that was designed to operate fully under equilibrium
conditions (FIGURE 1). Channel 1 is used to introduce a dilution gas,
either humidified nitrogen or zero air, generally at flow rates of 0.5 - 5
liters/min. Depending on the complexity desired for the final diluted
standard, up to 5 more channels can be used to introduced calibrated
standard mixes containing 0.5 to 40 ppm of each analyte. Both Alphagaz
( Morrisville, PA ) and Scott Specialty Gases ( Plumsteadville, PA )
currently have a single mix at 1-2 ppm that contains 40 of the TO14
analytes. Due to incompatibility and storage problems at ppm levels,
however, the 80-90 VOC's that will require monitoring in accordance
with the 1990 Clean Air Act Amendment (Title III) will need to be
blended from as may as 3-5 high concentration cylinder mixes. The
Mass Flow Controllers (MFC's) used for controlling flow rates of
standard are usually chosen to be 10-100 seem (standard cc per
minute), depending on the concentration of the high level mixes.
Mass in
MFC
Mass out
Figure 1
The system operates under equilibrium conditions by maintaining constant flow rate, tem-
334
-------
perature, and system pressure. Flow rates are regulated by individual mass flow controllers
with digital readouts continuously showing setpoint and actual flow rates. Temperatures are
at ambient (20-25 deg. C) and shouldn't change more than a few degrees during standards
preparation. Pressures are maintained relatively constant by providing back pressure regula-
tion at 25 psig. Before withdrawing standard into a SUMMA canister, the combined flow is
passed through a mixing region to insure homogeneity. A low pressure return line is avail-
able for equilibrating the line out to the canister itself before standard introduction.
DISCUSSION
hi order to achieve a mass in = mass out condition, a period of about 15 minutes is usually
necessary to reach a steady state condition where material is coming off of wetted surfaces as
fast as its adsorbing on. Polar and/or heavy VOC's would have the greatest tendency to
interact with surfaces and should therefore take the longest to achieve steady state condi-
tions. During canister filling, the dilution manifold should not experience the initial vacuum
of the canister, otherwise VOC's on the surfaces could be stripped off, resulting in even
higher VOC concentrations than expected. To prevent this, a back pressure regulator is
attached to the high pressure exhaust port. A restricter is then placed between the 4560
Diluter and the canisters so that a maximum flow rate of 500 seem is experienced at the start
of canister filling. In general, as long as at least 500 cc/min total flow is maintained through
the diluter, at least some flow will go through the back pressure regulator during canister
filling thereby maintaining the pressure at 25 psig +-3 psig.
Table 1 shows a typical dilution scheme using a 1.0 ppm TO14 standard from Scott Specialty
Gases. A 0-5000 and a 0-50 seem are used for the diluent and the standard mix, respectively.
After setting flow rates, the system is given 15 minutes to achieve a mass in = mass out
steady state before beginning to fill a canister. Due to the volume of the step down regulator
used for the standard cylinder mix and therefore the extended time necessary to purge it with
fresh standard, the first canister prepared should be the high concentration standard where
the cylinder mix flow rate is at a maximum (50 seem in this case). Equilibration should then
occur in a shorter period of time.
Table 2 shows some polar and non-polar VOC's blended down to 1-50 ppbv and analyzed by
GC/MS. All compounds showed less than 10% Relative Standard Deviation, which is near-
ing the maximum theoretical precision of mass spectrometric detection.
CONCLUSION
The Entech 4560DDS is a commercially available Dynamic Dilution System designed spe-
cifically for preparation of low level TO 14 standards. Reproducible generation of standards
in the range of 0.5 to 100 ppb is easily accomplished with as little as 2 mass flow controllers
when using a calibrated ppm level standard cylinder. Operation under constant temperature,
flow, and pressure insures that the concentrations introduced into the canister while filling
remain constant.
335
-------
TABLE 2- GC/MS Response factors for Dynamically Diluted Polar and Non-Polar
VOC's
ppb
Compound
Methanol
1,1- Dichloroethane
1,1,1-Trichloroethane
Tetrachloroethylene
Bromofluorobenzene
1,3,5- Triethylbenzene
1,2,4- Trimethylbenzene
1
2.59
.78
2.71
.96
.62
1.35
2.97
2
2.67
.69
2.30
.85
.61
1.20
2.78
5
2.83
.66
2.28
.78
.63
1.15
2.55
10
2.30
.68
2.21
.80
.69
1.42
2.61
20
2.32
.65
2.15
.75
.64
1.35
2.48
50 %RSD
2.41
.66
2.12
.76
.63
1.24
2.34
8.4
7.2
9.4
9.5
4.3
7.9
8.5
Data Compliments of J. Stanton, Westinghouse WIPP Program, Carlsbad, NM.
TABLE 3-Typical Dilution of a 1 ppm TO14 Mix
irnc 0-5
Zero Air
in L/min
5.00
2.50
1.99
1.98
1.96
0.95
0.45
L/min ^=f
MFC'S %
of Max
100%
50%
40%
40%
39%
19%
8%
| ([ 0-50
1 ppm Mix
In cc/min
5.0
5.0
10
20
40
50
50
MFC'S %
of Max
10%
10%
20%
40%
80%
100%
100%
Diluted
Cone.
in ppb
1
2
5
10
20
50
100
336
-------
THE ANALYSIS OF CANISTER SAMPLES IN LOUISIANA
BY NON-CRYOGENIC CONCENTRATION AND GC/MS ANALYSIS.
James M. Hazlett
Peggy Hatch
Charles K. Brown
Air Toxics Laboratory
Louisiana Department of Environmental Quality
Baton Rouge, Louisiana
ABSTRACT
The analysis of canister samples by EPA Method TO-14 normally
relies on the use of a cryogenic trap in order to concentrate
pollutants from a sample volume of about 400 ml. However when
analyzing samples collected under high humidity conditions, the water
vapor collected can cause plugging of the cryogenic trap. A nafion
permeable membrane drier can be used to reduce the amount of water
vapor collected. This device however can adversely affect the
collection of polar compounds and other highly volatile compounds.
Additionally, cryogenic concentrators are typically expensive and
require large quantities of cryogenic fluid to concentrate the target
analytes.
The Louisiana Department of Environmental Quality has modified a
sorbent based organic vapor concentrator for the analysis of canister
samples collected at five sampling sites around Louisiana. The
resulting concentrator and analysis system have been shown to generate
excellent analytical results for the compounds of interest. The
analysis system consists of a XonTech 930 Organic Vapor Concentrator
attached to a Hewlett-Packard 5890 Gas Chromatograph and a 5971 Mass
Selective Detector (MSD) operated in the Selective Ion Monitoring
(SIM) mode. The QA/QC procedures used in the analytical procedure and
the results obtained from the ambient air sampling program will be
discussed in this paper.
337
-------
INTRODUCTION
The Louisiana Department of Environmental Quality (LDEQ) has
conducted ambient air monitoring for air toxics since 1984. Most of
that monitoring effort has concentrated upon auto gas chromatograph
monitoring stations fitted with Flame lonization Detectors. While
these monitoring stations can provide a great deal of monitoring data
they are limited in the number and types of compounds which can be
successfully monitored. Since data on many halogenated volatile
organic compounds found in the ambient air in Louisiana was limited, a
canister based sampling program was initiated.
The analytical system selected for the analysis of these samples
was a Hewlett Packard Model 5890 Gas Chromatograph coupled to a HP
Model 5971 mass selective detector. Initially the sample
concentration system consisted of an adsorbent tube thermal desorber
modified to provide cryogenic concentration of the canister samples.
This concentration system was plagued with a several problems. Due to
the high humidity present in the Louisiana ambient air, the cryogenic
trap associated with the concentrator frequently plugged with ice
buildup. A number of techniques were employed to remove the water
from the sample stream but these procedures adversely affected the
recovery of the some of' the target analytes. The liquid nitrogen
consumption rate with this concentrator system was also very high
Additionally the system could not be automated to analyze multiple
samples. Since the financial resources within the department were
limited a low cost alternative method was sought.
The auto gas chromatograph monitoring program conducted by the
department utilizes a XonTech Model 930 Organic Vapor concentrator
which has proven to be very reliable and very cost effective to
operate. This sample concentrator utilizes adsorbent traps for sample
concentration rather than using cryogenics. A study was therefore
initiated to determine if that concentrator system could be coupled to
the GC/MS system and modified to provide the automated analysis of
canister samples.
ANALYTICAL PROCEDURES
The XonTech concentrator is fitted with two multi-bed adsorbent
traps consisting of Tenax-GR/Carboxen-569/Carbosieve S-III packed in a
stainless steel tube (12" length x 1/8" O.D.). The concentrator is
programmed to collect the canister samples over a 50 minute period at
a flow rate of 5 CC/min for a total sample volume of about 0.25
liters. The sample flow through the adsorbent trap is maintained at a
constant rate by a mass flow controller contained within the sample
concentrator. The Organic Vapor Concentrator is connected to the Gas
Chromatograph by means of a heated transfer line.
At the beginning of each desorption cycle the concentrator sends
a remote start signal to the Gas Chromatograph and Computer Data
System. The adsorbent trap is heated to 210 C and the sample is
thermally desorbed at a flow rate of 4 CC/min directly into the Gas
Chromatograph. The capillary column used is a J S W DB-624 75 meter
0.53 mm I.D. with a 3.0 urn stationary phase. The gas Chromatograph is
temperature programmed for a 6 minute hold at 35 C, them ramped @
4.5 /min to 180 C. The total analytical run time is 45 minutes. The
column is connected to the mass selective detector by means of a
capillary splitter which allows only a flow of 1.0 CC/min to enter the
ion source. The MSD is programmed to operate in the selective ion
338
-------
monitoring mode in order to provide maximum sensitivity for the
analytes of interest.
QUALITY ASSURANCE
The instrument is calibrated at five calibration levels from 0.5
ng/1 to 25 ng/1 for all of the target compounds. The calibration and
internal standard mixtures are prepared in 15 liter Summa® passivated
canisters. The humidified calibration mixture is prepared at a
concentration of 25 ng/1 for each compound. Multiple calibration
levels are accomplished by changing the calibration standard sampling
time on the concentrator. Table 1 shows how each calibration level is
programmed.
( Table 1 )
Calibration Levels Utilized For Calibration Curve
Calibration Level Concentration Sample Vol. Sampling Time
Level 1 25.0 ng/L 250 CC 50 minutes
Level 2 10.0 ng/L 100 CC 20 minutes
Level 3 5.0 ng/L 50 CC 10 minutes
Level 4 1.0 ng/L 10 CC 2 minutes
Level 5 0.5 ng/L 5 CC 1 minute
The resulting calibration curves as illustrated in Figure 1 have
been shown to be linear for all of the target compounds with a curve
fit of 0.98 or better. The curves also demonstrate the lack of any
breakthrough in the adsorbent traps at the higher sampling volumes.
The d6-Benzene internal standard and the 4-Bromoflourobenzene mass
standard surrogate are automatically injected into every analysis by
the concentrator. The peak areas for each of these compounds are
recorded and plotted onto control charts.
Spiked canisters are prepared periodically to challenge the
analytical system. The precision and accuracy of the analytical
method were determined by running a series of sixteen analyses of a
spiked canister containing low levels of the target compounds. The
results of this analysis showed that the analytical system was able to
accurately measure all of the target compounds within an accuracy of
+/- 25% and with a. precision of + /- 15%.
Several ambient samples were analyzed by a Saturn Air GC/MS
system recently purchased by the department and the results shown in
Table 3, were compared to the results from the LDEQ canister analysis
system. The results show that the two methods of analysis were
comparable, with all differences within the standard margin of error.
AMBIENT DATA RESULTS
Currently there are six canister sampling sites located around
Louisiana. These sites are located in Baton Rouge, Baker, Dutchtown.
Shreveport, Monroe and West Lake. The canister samples are collected
utilizing XonTech Model 911A samplers. Before sample collection each
canister is evacuated and filled with hydrocarbon free air. The
canister is then blank analyzed to ensure there is no internal
contamination. The canister is re-evacuated and sent out to the
sampling site. About 15 liters of ambient air are collected during a
single 24 hour sampling period. As a result, the canister is
339
-------
pressurized to about 20 psig. After sample collection is completed,
the canisters are transported back to the laboratory for analysis.
Table 2 shows the analytical results of the sampling program From June
1992 through February 1993.
( Table 2 )
Canister Sampling Results
Mean average concentrations in ppb
Freon-12
Chloromethane
Vinyl chloride
Bromomethane
Chloroethane
Freon-11
1,1-dichloroethene
Acetone
Carbon disulfide
Methylene Chloride
t-1, 2-dichloroethene
1,1-dichloroethane
c-1,2-dichloroethene
2-butanone
Chloroform
Methylchloroform
Carbon Tetrachloride
Benzene
1,2-dichloroethane
Trichloroethylene
1,2-dichloropropane
Bromodichloromethane
tl,3-dichloropropene
4-methyl-2-pentanone
Toluene
cl, 3-dichloropropene
1,1,2trichloroethane
Perchloroethylene
2-hexanone
Dibromochloromethane
Chlorobenzene
Ethylbenzene
m+p xylene
o xylene
Styrene
Bromoform
Tetrachloroethane
BAK = Baker,
DCH = Dutchtown
BAK
BTR
DCH
MON
SHV
WSL
1.63
0.73
0.02
0.01
0.01
0.46
0.00
4.10
0.10
0.15
0.00
0.00
0.00
0.57
0.03
0.15
0.16
1.11
0.07
0.06
0.00
0.00
0.00
0.05
1.00
0.00
0.06
0.02
0.04
0.00
0.02
0.16
0.47
0.14
0.05
0.00
0.01
BTR =
MON =
1.92
0.57
0.16
0.03
0.09
0.56
0.00
4.38
0.27
0.12
0.00
0.00
0.00
0.85
0.08
0.88
0.16
1.65
0.17
0.05
0.00
0.00
0.00
0.08
1.65
0.00
0.12
0.06
0.04
0.00
0.09
0.32
1.01
0.31
0.08
0.00
0.01
Baton
Monroe
1.74
0.65
0.03
0.01
0.04
0.46
0.01
2.74
0.14
0.18
0.01
0.01
0.00
0.68
0.03
0.26
0.20
0.48
0.11
0.06
0.00
0.00
0.00
0.02
0.65
0.00
0.01
0.05
0.03
0.00
0.01
0.24
0.43
0.09
0.05
0.00
0.01
Rouge
1.61
0.57
0.00
0.01
0.00
0.43
0.00
4.21
0.25
0.07
0.00
0.00
0.00
0.66
0.02
0.18
0.13
0.60
0.03
0.05
0.00
0.00
0.00
0.08
1.04
0.00
0.02
0.02
0.03
0.00
0.01
0.26
0.68
0.22
0.05
0.00
0.01
SHV =
WSL =
1.89
0.60
0.00
0.01
0.04
0.46
0.01
4.62
0.21
0.15
0.00
0.00
0.00
1.23
0.02
0.23
0.16
1.13
0.03
0.05
0.00
0.00
0.00
0.08
1.97
0.00
0.03
0.04
0.06
0.00
0.01
0.36
0.97
0.29
0.05
0.00
0.01
2.08
0.72
0.01
0.01
0.35
0.45
0.00
5.88
0.19
0.12
0.00
0.01
0.00
2.15
0.05
0.23
0.20
0.80
0.10
0.10
0.00
0.00
0.00
0.12
0.81
0.00
0.42
0.06
0.33
0.00
0.06
0.24
0.49
0.14
0.06
0.00
0.02
Shreveport
Westlake
340
-------
It is very apparent from the ambient monitoring data that the
orientation of each site to the local point sources is most likely
responsible for the differences in concentrations observed between the
several sites. The data shows that the sampling sites located in or
near urban areas (Baker, Baton Rouge, Shreveport) generally have
higher concentrations than those located in more rural areas
(Dutchtown, Monroe, Westlake) . Since the Baton Rouge site is closer
to a major industrial area it would be expected to show higher
concentrations. The Baton Rouge site measured the highest levels of
Benzene, Vinyl chloride, Methylchloroform, and 1,2-dichloroethane.
Except for some slightly higher Freon levels at the Baton Rouge site,
all of the sites had about the same measured levels of the common
background compounds such as Freon-12, Freon-11, Acetone,
trichloroethylene, and Carbon tetrachloride.
CONCLUSIONS
The results of the study showed that the Organic Vapor
Concentrator was able to interface very well with the GC/MS system and
produce quality analytical results. The analytical system has proven
to be very reliable with low operational cost. An analysis of the
analytical results suggests that the system is able to produce data
comparable to cryogenic concentration techniques.
Possible alternative applications for this analytical
technique would include the use in continuous monitoring stations for
air toxics or in mobile laboratories. Without the need for cryogenic
coolants this type of system could operate for extended periods of
time at remote locations.
( Table 3 )
Comparison of Ambient Sample results
Xontech/Hewlett-Packard System versus a Varian Saturn Air Cryogenic
concentration system
Compound
Freon-12
Vinyl Chloride
Carbon disulfide
Chloroform
Methyl Chloroform
Carbon tet.
Benzene
Toluene
1,2-dichloroethane
m+p xylene
o xylene
Styrene
Chlorobenzene
Sample #1
X/HP Sat.Air
2.31
0.00
0.09
0.03
0.43
0.18
0.54
0.97
1.76
0.00
0.05
0.02
0.30
0.11
0.46
0.67
20
32
14
18
38
0.22
0.04 0.17
0.01 0.01
Sample #2
X/HP Sat.Air
2.30
0.02
0.07
0.03
0.42
0.17
0.61
0.99
0.00
0.32
0.14
0.04
0.01
1.41
0.00
0.03
0.02
0.33
0.11
0.50
0.62
0.00
0.25
0.13
0.09
0.01
Sample #3
X/HP Sat.Air
2
0
0
0
0
0
0
0
0
0
0
0
0
.09
.00
.25
.02
.30
.15
.37
.33
.00
.17
.08
.04
.01
1.
0.
0,
0,
0.
0.
0.
0.
0,
0.
0,
0
0.
.41
.00
.12
.01
.22
.11
.37
.30
.00
.20
.11
.10
.00
341
-------
( Figure 1 )
Example Calibration Curves
Vnyl Chloride
Acetone
Chlorobenzene
( Figure 2)
25 ng/L Calibration
Retention Time
342
-------
Session 9
Quality Assurance
-------
Changes to the EPA
Quality System for the Collection and
Evaluation of Environmental Data
Gary L. Johnson
Senior Manager, Quality Assurance
United States Environmental Protection Agency
Quality Assurance Management Staff (MD-75)
Research Triangle Park, North Carolina 27711
Formalized quality assurance (QA) system requirements for the U.S. Environmental
Protection Agency (EPA) have been established for more than ten years. During this period,
many environmental issues and concerns addressed by the EPA have changed. Other issues,
such as hazardous waste clean-up and toxic and related air pollutants, remain a focus of
national environmental concern. These environmental issues are bringing changes to the way in
which the EPA plans, implements, and assesses its program activities. To meet the needs of the
Agency's environmental data collection programs through the end of this century, the QA
system of the EPA has been transformed and revitalized.
The EPA QA system provides the necessary management and technical processes to
effectively plan, implement, and assess the results of work performed. Within this structure, the
program provides managers with "tools" to accomplish each phase of the QA program, including
the Data Quality Objectives (DQO) process, Management Systems Reviews (MSRs), and the
new Data Quality Assessment (DQA) process.
This paper discusses the current state of the EPA QA system implementation, provides a
status report on new QA requirements and guidance documents, and describes some of the new
features of the QA program to improve the quality of environmental data operations.
345
-------
INTRODUCTION
Since 1984, EPA has been applying quality management principles to the collection, analysis,
and use of environmental data. Environmental data help form the basis for nearly all policy,
technical, and regulatory actions at EPA. Therefore, it is vital that the collected data are of the
type, quantity, and quality needed to make decisions with the desired degree of confidence; that
is, to assure management that the data do not lead to an incorrect decision and that the data
can withstand scientific and litigative scrutiny. The quality of environmental data may be
affected by any aspect of data collection and manipulation, including sampling, analysis,
validation of results, and evaluation of the complete measurement system, that may, in turn,
impact the decision-making process adversely.
Quality management, in the context of its use by EPA, may be defined as follows:
Quality Management is that aspect of the overall management systems of an organi-
zation that determines and implements policy regarding the Quality System, including
strategic planning, allocation of resources, and the planning, implementation, and
assessment of programs by the organization.™
The EPA commitment to quality is embodied in EPA Order 5360.1'2', which requires all EPA
organizations collecting environmental data to implement an effective Quality System.
The presence of quality assurance (QA) and quality control (QC) programs in EPA may
be traced back to the early 1970s. By 1979, a formal, Agency-wide QA program was established
and the following year, the Quality Assurance Management Staff (QAMS) issued interim
guidance for preparing two key QA planning documents, QA Program Plans'3' and QA Project
Plans.'4' These two documents became de facto standards and were never updated by EPA for
external use, even though revised internal guidance for QA Program Plans was developed in
1986. The need to provide current documentation has become clear.
The environmental programs being undertaken by EPA in 1993 are vastly different from
those supported by the QA/QC programs of the 1970s and 1980s. Just as the Agency has
evolved to meet changing environmental needs and issues, so must the Agency Quality System
change to assure the continuation of having adequate environmental data for decision-making.
AMERICAN NATIONAL STANDARD FOR ENVIRONMENTAL QUALITY SYSTEMS
Environmental programs are broad in scope and diversity. Over the past twenty years,
EPA and other federal, state, and private sector organizations have developed QA/QC
programs and requirements that fit their particular mission needs. This practice has created a
situation of multiple organizations with different QA/QC requirements, many of which are
conflicting in their specific criteria. As a consequence, unnecessary costs have been incurred by
government and the private sector to satisfy these duplicative requirements.
For the past three years, QAMS has participated with other federal agencies and private
sector companies in an effort led by the American Society for Quality Control to develop an
American National Standard for environmental quality systems. This effort has resulted in a
proposed standard, ANSI/ASQC E4-1993, Quality Systems Requirements for Environmental
346
-------
Programs'5', from the ASQC Energy and Environmental Quality Division which is expected to be
approved and published by the ASQC and the American National Standards Institute (ANSI)
later this summer. EPA has indicated its intention to adopt ANSI/ASQC E4-1993 as the basis
for its internal quality system. Moreover, EPA will seek changes to the EPA Acquisition
Regulations'61 and the EPA Financial Assistance Regulations'7'81 to require that organizations
satisfy the requirements of ANSI/ASQC E4-1993 as a condition for receiving contract or
financial assistance funds.
The impact of ANSI/ASQC E4-1993 on the existing EPA Quality System is not expected
to be great since many of the elements of E4 are already found in the Agency Quality System,
including the Quality Management Plan (QMP), the Data Quality Objectives (DQO) process,
the QA Project Plan (QAPP), the Data Quality Assessment (DQA) process, and the Manage-
ment Systems Review (MSR). While some terminology has changed, greater consistency will
result and unnecessary costs will be reduced.
NEW AND REVISED DOCUMENTATION
In anticipation of the new standard, QAMS has undertaken a comprehensive develop-
ment program for new and replacement documentation to implement upgraded E4-based
quality systems across the Agency. Two kinds of documents are planned: requirements
documents and guidance documents. As the name suggests, requirements documents will define
mandatory criteria. These documents and a revised EPA Order on quality management will be
incorporated into a Quality Manual for internal use, and will be issued as an EPA policy
document following Agency-wide review and concurrence later this year. The requirements
documents will be issued separately for use outside the Agency by contractors and financial
assistance recipients. The guidance documents are non-mandatory and will be issued for general
use both internally and externally.
The availability of the new requirements and guidance documents will commence in a
few months and continue for another 18-24 months. The following is a summary of some of the
documents that are being developed currently:
EPA QA/R-1 EPA Quality Systems Requirements for Environmental Programs
QA/R-1 is the policy document by which EPA will implement the proposed American
National Standard ANSI/ASQC E4, when the standard has been approved. The draft is
complete. It also contains the requirements for the QA Annual Report and Work Plan
(QAARWP). Availability is expected in fall 1993.
EPA QA/G-1 Guidance for Developing, Implementing, and Evaluating Quality Systems
for Environmental Data Operations
QA/G-1 provides non-mandatory guidance to help organizations develop a quality system
that will meet EPA expectations and requirements. Availability is expected by early 1994.
EPA QA/R-2 EPA Requirements for Quality Management Plans
QA/R-2 is the policy document containing the requirements for the Quality Management
347
-------
Plan (QMP). An interim draft was issued in August 1992 to "product test" the approach. Since
QA/R-2 follows ANSI/ASQC E4, it will not be finalized until the standard has been approved.
QA/R-2 is the intended replacement for QAMS-004/80 and the subsequent internal EPA
guidance on QA Program Plans issued in 1987. Availability is expected by fall 1993.
EPA QA/R-2A EPA Requirements for Quality Management Plans for Analytical Laborato-
ries and Facilities
QA/R-2A will provide detailed requirements for environmental analytical labs. Since
there may be a national consensus standard for labs, the content of this document is unclear at
present. This is a planed item.
EPA QA/G-2 Guidance for Preparing, Reviewing, and Implementing Quality Manage-
ment Plans
QA/G-2 provides non-mandatory guidance to help organizations develop Quality
Management Plans (QMPs) that will meet EPA expectations and requirements. The document
will contain tips, advice, and case studies to help users develop an improved QMP. Target
Availability is expected by early 1994.
EPA QA/G-3 Guidance for Preparing, Conducting, and Reporting the Results of Man-
agement Systems Reviews
QA/G-3 provides non-mandatory guidance to help organizations plan, implement, and
evaluate management assessments of their quality systems. The guidance will present a step-by-
step description of the MSR process. Availability is expected by late summer 1993.
EPA QA/G-4 Guidance for Planning for Data Collection in Support of Environmental
Decision Making Using the Data Quality Objectives Process
QA/G-4 provides non-mandatory guidance to help organizations plan, implement, and
evaluate the Data Quality Objectives (DQO) process, with a focus on environmental decision-
making for regulatory and enforcement decisions. The guidance will present a step-by-step
description of the DQO process. Availability is expected in summer 1993.
EPA QA/G-4A Guidance for Planning for Environmental Research Using the Data Quality
Objectives Process
QA/G-4A provides non-mandatory guidance to help organizations plan, implement, and
evaluate the Data Quality Objectives (DQO) process, with a focus on environmental research
decision-making. The guidance will present a step-by-step description of the DQO process as
applied to R&D programs. Availability is expected by fall 1993.
EPA QA/R-5 EPA Requirements for Quality Assurance Project Plans
QA/R-5 is the intended replacement for QAMS-005/80. This policy document will
establish the requirements for QA Project Plans prepared for activities conducted by or funded
by EPA. A revised draft was submitted to the Work Group in January 1993 for final review.
Comments have been received and are being addressed. Closure on the document will be
348
-------
reached by early summer. An interim draft final version will be made available while the
document undergoes Agency "Green Border" review. Availability of the final document is
expected in the fall 1993.
EPA QA/G-5 Guidance for Preparing, Reviewing, and Implementing Quality Assurance
Project Plans for Environmental Programs
QA/G-5 provides non-mandatory guidance to help organizations develop Quality
Assurance Project Plans (QAPPs) that will meet EPA expectations and requirements. The
document will contain tips, advice, and case studies to help users develop improved QAPPs.
Availability is expected by early 1994.
EPA QA/G-8 Guidance for Preparing, Conducting, and Responding to Technical Assess-
ments for Environmental Data Operations
QA/G-8 will provide non-mandatory guidance to help organizations plan, conduct,
evaluate, and document technical assessments for their programs. Such technical assessments
include the Technical Systems Audit (TSA), surveillance, readiness reviews, and the Perfor-
mance Evaluation (PE). The document will contain tips, advice, and case studies to help users
develop improved processes for conducting technical assessments. This is currently a planed
item. QAMS expects to use a Work Group process to develop this guidance.
EPA QA/G-9 Guidance for Environmental Data Quality Assessment: Managing the
Process
QA/G-9 provides non-mandatory guidance for planning, implementing, and evaluating
retrospective assessments of the quality of the results from environmental data operations.
DQA is a statistically-based, quantitative evaluation of the extent to which a data set satisfies
the user's needs (or DQOs). This particular document is aimed at the project managers who
are responsible for conducting the environmental data operations and assessing the usability of
the results. The draft is currently undergoing internal review by QAMS. Availability is expected
by late summer 1993.
SUMMARY
The EPA Quality System is changing to meet the needs of the Agency's changing mission
and priorities. The adoption of the American National Standard ANSI/ASQC E4-1993 will
provide national consistency not only within EPA, but also on a voluntary basis with other
federal and state agencies and across the regulated community. The new quality management
documents will provide current criteria and guidance for planning, implementing, and assessing
quality systems more effectively. They will ensure that environmental programs produce the
type and quality of environmental data needed for key decisions.
349
-------
REFERENCES
1. ANSI/ASQC Q94-1987, Quality Management and Quality Systems Elements - Guidelines,
American Society for Quality Control (1987).
2. EPA Order 5360.1, Program and Policy Requirements to Implement the Mandatory Quality
Assurance Program, U.S. Environmental Protection Agency (April 1984).
3. "Interim Guidelines and Specifications for Preparing Quality Assurance Program Docu-
mentation," U.S. Environmental Protection Agency, QAMS-004/80 (December 1980).
4. "Interim Guidelines and Specifications for Preparing Quality Assurance Project Plans,"
U.S. Environmental Protection Agency, QAMS-005/80 (December 1980).
5. ANSI/ASQC E4-1993, Quality Systems Requirements for Environmental Programs, Ameri-
can Society of Quality Control (DRAFT FINAL, May 1993).
6. "Subpart 1546.2 - Contract Quality Requirements," U.S. Environmental Protection
Agency, EPA Acquisition Regulations, Title 48 Code of Federal Regulations, Chapter 15
(October 1988).
7. "Part 30 - General Regulation for Assistance Programs for Other than State and Local
Governments," U.S. Environmental Protection Agency, Title 40 Code of Federal Regula-
tions, Chapter 1 (July 1990).
8. "Part 31 Uniform Administrative Requirements for Grants and Cooperative Agree-
ments to State and Local Governments," U.S. Environmental Protection Agency, Title 40
Code of Federal Regulations, Chapter 1 (July 1990).
350
-------
The National Performance Audit Program (NPAP)
Elizabeth T. Hunike and Joseph B. Elkins, Jr.
Elizabeth T. Hunike
U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory (MD-77B)
Research Triangle Park, North Carolina 27711
Joseph B. Elkins, Jr.
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards (MD-14)
Research Triangle Park, North Carolina 27711
ABSTRACT
The National Performance Audit Program (NPAP) is one of the major components
in the quality assurance of the Nation's air monitoring program. Over the last
several years and especially in 1993, the NPAP has undergone a metamorphosis. This
paper will explore some of the forces behind this change as well as the resulting
changes which include expansion of the NPAP to cover all the criteria pollutants,
new audit equipment design, more audit equipment, more sites audited, and changes in
the audit site selection process. The paper will include a review of the audit
equipment used throughout the program's existence, an examination, pollutant by
pollutant, of the number of audits performed each year beginning in 1989, and other
enhancements of the expanded NPAP.
Disclaimer
The information in this document has been funded wholly by the United States
Environmental Protection Agency. It has been subjected to Agency review and
approved for publication.
351
-------
INTRODUCTION
The Nation's ambient air monitoring program contains monitors for the six
pollutants that have national ambient air quality standards (NAAQS). These
pollutants are carbon monoxide (CO), lead (Pb), nitrogen dioxide (NO2), ozone (Oj),
particulate matter smaller than 10 microns (PM-10), and sulfur dioxide (SO2). The
pollutants and associated standards are shown in Table 1.
Table 1. National ambient air quality standards.
Pollutant
CO
Pb
N02
o,
PM-10
SO2
Primary standard
(health related)
Std. level
Type of average cone."
8-hrb
1-hr'
Maximum quarterly
average
Annual arithmetic
mean
Maximum daily 1-hr
average0
Annual arithmetic
meand
24-hrJ
Annual arithmetic
mean
24-hrb
9 ppm
(10 mg/m3)
35 ppm
(40 rog/m3)
1.5 f/g/m3
0.053 ppm
(100 fig/nf)
0 . 12 ppm
(225 ^g/m3)
50 vg/tt?
150 fjg/m3
80 /Jg/m3
(0.03 ppm)
365 jjg/m3
Secondary standard
(welfare related)
Std. level
Type of average cone .
No secondary
standard
No secondary
standard
Same as primary
standard
Same as primary
standard
Same as primary
standard
Same as primary
standard
Same as primary
standard
3-hr"
1300 jjg/m3
(0.50 ppm)
"Parenthetical value is an approximately equivalent concentration.
"Not to be exceeded more than once per year.
"The standard is attained when the expected number of days per calendar
year with maximum hourly average concentrations above 0.12 ppm is equal to
or less than 1, as determined according to Appendix H of the Ozone NAAQS.
'Particulate standards use PM-10 (particles less than 10 micrograms in
diameter) as the indicator pollutant. The annual standard is attained
when the expected annual arithmetic mean concentration is less than or
equal to 50 J/g/tn3; the 24-hour standard is attained when the expected num-
ber of days per calendar year above 150 /jg/m3 is equal to or less than 1,
as determined according to Appendix K of the PM NAAQS.
352
-------
There are approximately 4,682 air pollution monitors in the ambient air
network. These monitors comprise the State and Local Air Monitoring Stations
(SLAMS), the National Air Monitoring Stations (NAMS), and the Prevention of
Significant Deterioration (PSD) sites. The distribution of monitors by pollutant
is: S02, 613; CO, 497; NO2, 312; O3, 822; Pb, 426; PM-10, 1359; and total suspended
particulate (TSP) matter (old standards), 653.
The quality assurance/quality control of these monitors has three major
components: the EPA Regional Office Systems Audits; the Precision and Accuracy
Reporting System (formerly PARS); and the National Performance Audit Program (NPAP).
The NPAP is a cooperative effort between EPA's Atmospheric Research and Exposure
Assessment Laboratory (AREAL), the 10 EPA Regional Offices, and the 170 state and
local agencies that operate the SLAMS/NAMS air pollution monitors. Also included in
the NPAP are organizations that operate air monitors at PSD sites. Participation in
the NPAP is required for agencies operating SLAMS/NAMS and PSD monitors as per
Section 2.4 of 40 CFR Part 58, Appendix A and Section 2.4 of 40 CFR Part 58,
Appendix B. The NPAP is operated by the Quality Assurance Support Branch of AREAL.
AUDITS
The NPAP's goal is to provide audit materials and devices that will enable EPA
to assess the proficiency of agencies that are operating monitors in the SLAMS/NAMS
and PSD networks. To accomplish this, the NPAP has established acceptable limits or
performance criteria, based on SLAMS/NAMS and PSD requirements, for each of the
audit materials and devices provided in the program. Any device or material not
meeting these predetermined criteria is not used in the program.
All audit devices and materials used in the NPAP are certified as to their
true value, and that certification is traceable to a NIST standard material or
device wherever possible. The audit materials used in the NPAP are as
representative and comparable as possible to the calibration materials and actual
air samples used and/or collected in the SLAMS/NAMS and PSD networks. The audit
material/gas cylinder ranges used in the NPAP are specified in the Federal Register
{Table 2).
Table 2. NPAP audit material/gas cylinder concentration ranges.
SO2, O3, and NO2
CO
Pb
Audit
level
1
2
3
1
2
3
Audit
level
1
2
Concentration
range , ppm1
0.03-0.08
0.15-0.20
0.35-0.45
3-8
15-20
35-45
Concentration
ranqe, /jcr/strip1
100-300
600-1000
'Federal Register, 40 CFR Part 58, Appendix A, revised
July 1, 1987.
The objectives for the NPAP audits are two-fold: (1) to complete at least 95%
of the scheduled audits by the end of the year, and (2) to determine if the
participants' performance exceeds the limits shown below.
Audit EPA determined limits
High volume/PM-10 (SSI) % difference > ± 15% for 1 or more flows
Dichot (PM-10) % difference > ± 15% for 1 or more flows
Pb % difference > ± 15% for 1 or more levels
SO2, NO2, O3, and CO Mean absolute % difference > 15%
353
-------
The number of NPAP audits
Table 3.
performed from 1989 to the present are shown in
Table 3. NPAP audits 1989 to 1993.
Pollutant
CO
S02
NO part of NO2
N02
03
SSI/hi-vol
Diohot (PM-10)
Labs
Samplers
Labs
Samplers
Labs
Samplers
Labs
Samplers
Labs
Samplers
Labs
Samplers
Labs
Samplers
1989
129
355
135
383
53
98
53
98
14
39
289
1938
13
30
Number
1990
137
345
139
238
59
88
59
88
79
142
293
772
24
44
of NPAP Audits
1991 1992
114
183
123
204
83
119
14
19
43
77
315
612
9
17
134
261
142
279
101
170
14
27
135
340
308
1087
12
24
1993*
154
357
167
355
122
221
122
221
171
482
350
1363
19
54
Pb
Audits
367
297
322
*Audits scheduled.
In 1992 100% of the scheduled audits were completed with the exception of NO2
(15%), O3 (93%), and dichot (PM-10) (66%). These 3 audits had some equipment
problems that are discussed in the following section.
The percentage of 1992 NPAP participants whose performance fell within the EPA
guidelines of 15% of the true values was: CO, 94%; SO2, 89%; NO, 94%; NO2, 79%; O3,
97%; hi-vol/PM-10 (SSI), 91%; dichot (PM-10), 65%; and Pb, 92%. These percentages
have remained similar since 1989 for SO2, NO2, and Pb. O3 (85% to 97%) and NO (81%
to 94%) have increased substantially. CO (98% to 94%) and hi-vol/PM-10 (SSI) (94%
to 91%) have decreased slightly.
NPAP AUDIT EQUIPMENT
Ozone
O3 was added to the NPAP in 1989. The audit device is self-contained with its
own zero-air and ozone generation system. At the present time, this system is not
as field worthy as desired. The equipment is being modified in an attempt to
eliminate downtime due to damages incurred in shipping.
Hioh-Vol/PM-10 (SSI)
The reference flow device (ReF) consists of a modified orifice, a wind
deflector, a manometer, and five resistance plates. The ReF for the PM-10 (SSI)
flow audit is similar except a filter is used as the only flow restrictor.
Dichotomous (PM-10)
The NAAQS for particulates changed from TSP to PM-10 in 1987; the dichotomous
(dichot) audit device was added to the NPAP in 1989. The original audit equipment
included an electronic manometer which was too fragile for repeated shipping. The
audit was suspended in 1991 until new audit equipment could be developed. The
audits resumed in May 1992. The improved dichot audit device consists of an
inclined manometer filled with red gauge oil, an altimeter that measures barometric
pressure in millibars, a small dial thermometer that reads in °F, and the laminar
flow element (LFE) with an air filter on its inlet. The dichot audit device
measures fine flow (15.00 1pm) and total flow (16.7 1pm).
354
-------
Lead I analysis1
The samples are 1.9 cm wide and 20 cm long glass fiber filter strips that have
been spiked with an aqueous solution of lead nitrate and oven-dried. Two filter
strips comprise a sample.
SO,/NO-NO,/CO (gas dilution system)
CO was added to the NPAP in the mid-70's. The original CO audit materials
consisted of a mixture of CO, CO2, methane, and zero air in returnable, 150 ft3
pressurized gas cylinders that simulated ambient air samples. Three cylinders,
representing high, medium, and low concentrations, comprised the set. Due to the
high cost of shipping the cylinders, demurrage charges and lost cylinders ($60K per
audit), the NPAP began using disposable cylinders in 1985. The use of disposable
cylinders reduced the costs of the audit by $45K per audit.
SO2 was added to the NPAP in 1981. From 1981 through 1988, the SO2 materials
consisted of an audit device and a compressed gas cylinder containing SO2 in
nitrogen along with a Size A compressed gas cylinder for zero-air. The zero-air
cylinder was very expensive ($50) and was usually good for only one audit. In 1989
the NPAP began using a portable zero-air system to serve as the dilution air source.
NO-NO2 was added to the NPAP in 1989. At that time, the audit was done using
a gas phase titration system.
Beginning in 1991, one gas dilution system was used for all three audits. The
system consisted of an audit device, one zero-air system, and two cylinders of gas
(NO2 and a blend of SO2, NO, and CO). The EPA went to the gas dilution system to
obtain better results and to save money. Unfortunately, the NO2 would not remain
stable in the cylinder, which caused a problem with the NO2 audits. In 1992
EPA/AREAL scientists combined a Thermo Environmental Instruments Model 165 ozone
calibrator with an EPA-designed gas phase titration and capillary dilution system to
produce an auditing device potentially applicable to all SLAMS gaseous pollutants.
When field and laboratory tests confirmed that this prototype system could be used
to audit SO2, NO2, CO, and O3 monitors, EPA procured 22 additional devices through
the competitive procurement process. These 22 devices arrived in April 1993 and
will be used in a pilot program during the summer and fall of 1993 with the goal of
introducing them into the NPAP in January 1994.
SITE SELECTION
Historically, the state and local agencies have been allowed to select the
NPAP sites to be audited. The 1989 General Accounting Office (GAO) audit raised
concerns about the NPAP site selection process. EPA, henceforth, conducted a review
in 1991 that determined the site selection process used had not biased the audit
results. EPA additionally responded to the GAO comments by developing site
selection criteria (Table 4) that were incorporated into the 1993 NPAP. Priority 1
site selection criteria should be audited annually; priority 2 at least once every 2
years; priority 3 at least once every 3 years; priority 4 at least once every 4
years. All other sites should be audited at least once every 5 years. Based on
this criteria, EPA now selects specific sites that are to be audited in the NPAP.
The criteria are reviewed annually, and site selection is updated accordingly.
CONCLUSIONS
The cornerstone of any data collection system is the quality assurance
component. The data utilization resulting from the Nation's air monitoring network
continues to increase in importance. The strategies developed from the information
can cost millions of dollars. Henceforth, we must remain vigilant in our efforts to
maintain the integrity of this important data set. In these efforts, the NPAP
continues to be refined. The NPAP has expanded to include all the criteria
pollutants. The associated instruments and equipment have been improved to
incorporate the latest technologies. The NPAP continues to respond to comments from
the GAO audits as well as state and local agency contacts. With the increased
interest in the data from the Nation's air monitoring community, it was inevitable
that the site selection process would also be modified. These modifications were
incorporated into the 1993 NPAP.
In summary, the NPAP has expanded to include all criteria pollutants, improved
the associated equipment and instruments, and changed its site selection process.
355
-------
Table 4. Site selection criteria for the NPAP.
Pollutant Priority Criteria
O, 1 Sites that had expected exceedances of the 03 NAAQS
> 1.1 days from 1989 through 1991.
PM-10 2 Sites recording values > the 24-hr. NAAQS.
3 Sites recording values > 80% but < 100* of the 24-hr
NAAQS.
4 Sites recording values > 50% but < to 80% of the 24-hr
NAAQS.
CO 3 Sites recording exceedances of the CO NAAQS from
1990-1991.
3 Some selected sites within CO nonattainrnent areas.
4 Sites recording CO values between 7.5 ppm and 9.4 ppm.
Pb 1 Sites located near sources which are subject to poten-
tial regulatory compliance, out of compliance, and/or
subject to a consent decree.
2 Sites located near sources that are either in compli-
ance with no violations, are closed for business, or
are well above the Pb NAAQS with no significantly
questionable data.
SO2 2 Sites recording values > the 24-hr SO2 NAAQS.
3 Sites recording values between 80% and 100% of the
24-hr SO2 NAAQS.
4 Sites recording values > 50% but < 80% of the 24-hr
SO2 NAAQS.
N02 3 Sites recording values > 50% of the annual NO2 NAAQS
356
-------
Two New Gas Standards Programs
At the National Institute of
Standards and Technology
by
William J. Mitchell, Ph.D.
U.S. Environmental Protection Agency
Atmospheric Research and Exposure
Assessment Laboratory
Research Triangle Park, NC 27711
and
Willie E. May, Ph.D.
National Institute of Standards and Technology
Organic Analytical Research Division
Gaithersburg, MD 20899
ABSTRACT
The 'EPA/NIST certified reference materials (CRM) program is being
terminated and replaced with two new ones: the NIST Traceable
Reference Materials (NTRM) and the Research Gas Mixture (RGM)
programs. These new programs are being implemented to provide
NIST traceability to a wider number of gas mixtures. The NTRM
program will differ from the CRM program in two significant ways:
candidate gas mixtures will not have to be identical to a NIST
Standard Reference Material (SRM), and the producer of the NTRM
rather than EPA will pay NIST to check the concentration of the
gas mixture. In the RGM program, NIST will enter into agreements
with either governmental, commercial or private organizations to
produce gas mixtures for which there are no SRMs or which lie
outside the concentration range of existing SRMs. The details of
these programs are presented in this paper.
357
-------
INTRODUCTION
Many EPA ambient air, stationary source and mobile source
regulations mandate the use of gaseous concentration standards
traceable to the National Institute of Standards and Technology
(NIST) for calibrating the pollutant measurement systems'11.
Standard Reference Materials, (SRMs), which are certified and
sold by the NIST are the highest quality, NIST-traceable
standards available. Presently, NIST supplies SRMs for the
following pollutant gases regulated by the EPA: NO, N02, S02,
CO, C02, O2, and H2S.
The above SRMs were developed by the NIST in cooperation
with the EPA. In anticipation that the NIST alone could not meet
the demand for NIST-traceable standards, EPA and NIST implemented
two other mechanisms to meet this demand. These mechanisms, the
Protocol Gas program and the Certified Reference Materials (CRM)
program were begun to allow the specialty gas industry to provide
gas standards traceable to NIST SRMs at costs less than those for
SRMs. The Protocol Gas program121 has been quite successful. It
is targeted principally to those who use specialty gas mixtures
to calibrate air pollutant monitoring systems, whereas the SRM
and CRM programs are targeted for use by the specialty gas
industry. Protocol Gases cost 10 to 25% of the corresponding
SRM, are certified as stable for periods close to or equal to
those of the corresponding SRMs and are made to order by the
specialty gas industry.
In contrast, the CRM program131 has not achieved its intended
purpose, most likely because CRMs are very close in quality and
other characteristics to SRMs. As shown in Table 1, for the cost
of a CRM, the potential CRM purchaser can have an SRM, a higher
quality standard with a longer stability period. Also, suppliers
of specialty gases are reluctant to prepare CRMs because the lead
time required means they have to prepare each CRM batch on
speculation that they can sell and/or use them quickly after the
CRM batch is certified. They cannot make only the number of CRMs
needed to meet their internal needs, e.g. for calibration and QC
standards, because public funds are used to spot-check the
supplier's assay of the CRM batch. Thus, EPA requires suppliers
of CRMs to make a sufficient number of cylinders in each CRM
batch to be able to sell some of them to others, including
competitors in the specialty gas industry.
NTRM AND ROM PROGRAMS
The CRM program is now being replaced by the NIST Traceable
Reference Materials (NTRM) program, which, NIST and EPA, hope
will correct the deficiencies in the CRM program. NTRMs will
differ from CRMs in the following ways:
1. The NTRM concentration can be at or between those of
358
-------
SRMs for the same analyte and balance gas.
2. NIST rather than EPA, will spot-check the assay of the
NTRM batch.
3. NIST will assign the analyte concentration(s) to the
NTRM.
4. The supplier of the NTRM will pay NIST for the spot-
check of assay.
5. The NTRM will not have to be sold to others since no
public funds are used in the certification.
Table 2 provides a fuller comparison of the NTRM and CRM
programs.
But, what if an organization needs a high quality gas
standard where either (1) the analyte is the same as an SRM but
where the concentration is outside the range of the existing SRMs
or (2) the analyte(s) and/or the balance gas are different from
an SRM? The NIST and the EPA are implementing the Research Gas
Mixture (RGM) program to address these situations. Under the RGM
program, the NIST will enter into an agreement with another
party, such as a government agency, a. trade organization or a
specialty gas company, to produce gas mixtures that have
different analytes, different balance gases or concentrations
outside the range of available SRMs. Since the requesting
organization will fund the work, the RGM will be the property of
the sponsor but any primary standards developed will be the
property of the NIST.
Key features of the RGM program are:
1. Applicable to gas mixtures not covered by the SRM or NTRM
programs.
2. Gas mixture likely can be made.
3. Requestor contacts NIST to discuss procedures and
development costs.
4. Work statement, schedule, milestones, and deliverables are
specified in written agreement with NIST.
5. Requestor funds project up front to amount agreed upon with
NIST.
6. NIST develops new or extends current primary standards
required to perform the analysis of the candidate RGM.
7. NIST analyzes primary standards suite for precision and
accuracy.
8. NIST analyzes candidate RGM to determine analyte
concentration(s) and also the levels of impurities of
interest.
9. If successful, gas mixtures produced are "NIST traceable"
and NIST provides the requestor with a Report of Analyses
and a defined period for which the assigned analyte value(s)
are valid.
10. Requestor takes possession of all gas mixtures in the RGM
batch.
359
-------
NIST and EPA are now preparing documents that will describe
in detail the mechanics of the NTRM and the RGM programs. At
this time, some key features of both programs are not firm. One
of these is the question of "Exclusivity", that is, what rights
will the organization paying for the development of an RGM have
to the exclusive use of the NIST primary standards developed as
part of the work? For example, should they have exclusive use
forever, for 24 months, etc.? If Company B is willing to pay the
NIST to develop similar or identical primary standards to those
developed for Company A, can or should the NIST do this? If many
companies ask the NIST to develop RGMs similar to those developed
for Company A, such that development of an SRM is warranted, can
the NIST develop the SRM? If EPA decides to regulate a pollutant
covered by an RGM, does the NIST have a responsibility to develop
an SRM even if it means the owner of the RGM loses market share?
Two other areas where details are not yet firm are:
1. If the first attempt to develop an RGM fails, but the
information obtained allows NIST to develop that RGM
for someone else, does the first organization have any
rights to the RGM?
2. What should the defined period of stability be for each
RGM and NTRM and how should the stability be monitored?
None of these questions is easy to answer; in the end, they
probably will be decided on a case by case basis.
Disclaimer
The information in this document has been funded wholly by
the United States Environmental Protection Agency. It has been
subjected to Agency review and approved for publication.
REFERENCES
1. Code of Federal Regulations, Title 40, Parts 50, 58, 60 and
75. Office of the Federal Register, Washington, D.C.
2. "Revised EPA Protocol for Assay and Certification of
Compressed Gas Calibration Standards." Draft, April 1993.
Available from the U.S. EPA, Atmospheric Research and
Exposure Assessment Laboratory, MD-77B, Research Triangle
Park, NC 27711.
3. "A Procedure for Establishing Traceability of Gas Mixtures
to Certain National Bureau of Standards SRMs." EPA-600/7-
81-010, Joint Publication of the National Bureau of
Standards and the Environmental Protection Agency. May
1981.
360
-------
Table 1. Comparison between SRMs and CRMs
Parameter
Minimum Batch Size
Cost Each
Gas Volume
Analyte Concentration
Balance Gas
Analyte Assayed Using
Assay Done by
Uncertainty in Assay
Assay Valid For
Assay Certification
Period Begins
Time From Preparation to
Availability
SRM
20
$800 - $1500
30 CU. ft.
NIST's choice
NIST's choice
NIST Primary
Standards
NIST
±1%
24 48 months
When SRM sold
18 36 months
CRM
10
$800 - $1500
150 CU. ft.
Must be within
±1% of SRMs
Must be same as
SRM
SRMs
Supplier of CRM
with EPA spot-
check assays.
Up to ±2%
12 - 24 months
When CRM Batch
is certified
6 12 months
361
-------
Table 2. Comparison between NTRM and CRM programs
CRM
Company decides needs
Company prepares mixtures
Company performs analyses
Data sent to EPA
EPA selects two cylinders for
analysis
Cylinders analyzed by EPA
All data sent to NIST
Audit lab concurs with
concentration
Audit costs covered by EPA
Company provides assay
certificate
Company distributes mixtures
Company responsible for
stability
NTRM
Same
Company contacts NIST; after
agreement, mixtures made
Same
Data sent to NIST
NIST selects cylinders for
analysis
Cylinders analyzed at NIST
Same
NIST assigns concentration
Costs covered by producer
($4500 $6000)
NIST provides assay
certificate
Same
Same
362
-------
A Low Cost Procedure to Make Gaseous
Pollutant Audit Materials
by
William J. Mitchell, Ellen W. Streib and Howard L. Crist
U.S. EPA/AREAL (MD-77B)
Research Triangle Park, NC 27711
and
Ronald Bousquet
ManTech Environmental Technology Incorporated
Research Triangle Park, NC 27709
ABSTRACT
EPA has identified over 130 organic and 20 inorganic gases that
it is now regulating or plans to regulate as air pollutants.
These compounds are covered by a variety of environmental
regulations such as RCRA, NSPS, NAAQS, TSCA, NESHAPS, and CERCLA.
These compounds can be found in the air from 1 ppb to the 1000
ppm range and in different relative ratios. A gas transfer
system is described in this presentation that will allow an
organization to prepare a wide variety of QC and QA materials for
these gases using a small number of stock gases in compressed gas
cylinders. The experiments conducted to characterize the
performance of this system to date and the additional experiments
planned are described.
Disclaimer
The information in this document has been funded wholly by the
United States Environmental Protection Agency. It has been
subjected to Agency review and approved for publication. Mention
of trade names or commercial products does not constitute
endorsement or recommendation for use.
363
-------
INTRODUCTION
Since the mid-70's, the Atmospheric Research and Exposure
Assessment Laboratory (AREAL) of the U.S. Environmental
Protection Agency (EPA), Research Triangle Park, NC, has worked
with the National Institute of Standards and Technology (NIST)
and with the specialty gas industry to develop stable,
accurately-certified compressed gas reference materials. After
the NIST or a specialty gas company completes short term (i.e., 3
to 12 months) stability tests on candidate reference materials,
AREAL takes possession of those that appear stable and subjects
them to longer term (i.e.,1 to ten years) stability testing. When
not undergoing stability checks, AREAL makes the materials
available, free-of-charge, to other organizations to use in
developing, validating, calibrating and quality assuring air
pollutant measurement Ysystems applicable to such EPA regulations
as the RCRA, CERCLA, NSPS, NESHAPS, NAAQS and TSCA.
In 1990, AREAL's inventory contained approximately 500 gas
mixtures, and it met both AREAL's needs (for a statistically
sufficient number of gas mixtures for its stability testing) and
the regulators' needs for calibration and QA/QC materials. When
the Clean Air Act Amendments (CAAA) of 1990 extended the CAAA to
an additional 100 plus compounds, the repository no longer met
the needs of the regulators. The CAAA also caused the EPA to
change how it regulated stationary sources under the CAA.
Historically, EPA had provided the regulated sources with EPA-
validated or, at least well-characterized, pollutant measurement
systems for monitoring the source's emissions. The CAAA's
inclusion of many new toxic pollutants, combined with its
emphasis on Maximum Achievable Control Technology (MACT), caused
the EPA to adopt a performance-based approach to control
stationary source emissions (somewhat analogous to the DRE
approach used for incineration testing under the RCRA and the
CERCLA). In this latter approach, the EPA provides regulated
sources with general information on the test methods applicable
to each regulated pollutant and requires the source to document
that the test method used was accurate.
AREAL's inventory became inadequate because:
1) Most of the 50 newly-identified toxic and 60 oxidant
precursor VOCs were not in the AREAL repository and/or their
stability in compressed gases was unknown.
2) Many of the VOCs were available only at concentrations
significantly higher than those found in the gas streams exiting
from control devices and their stability at these lower
concentrations was unknown or, at best, suspect for extended
storage periods (e.g., 3 to 6 months). (Since the MACT regulation
depends on accurate measurement of the pollutant concentration(s)
before and after the control device, accurately certified, stable
gas mixtures are needed at the lower pollutant concentrations.)
3) Most of the toxic organics in AREAL's repository were
364
-------
available only as binary (VOC plus balance gas) rather than as
multi-component mixtures. (Because the CAAA requires some sources
to simultaneously measure many toxic VOCs, multi-component VOCs
standards are needed.)
Thus, at a time when the test method development and
validation effort required from the regulated community was
increasing significantly, reliable, accurately-certified
reference materials were not available for many of the VOCs. To
address this situation the AREAL : is revising the guidance on
Protocol Gases; has instituted the Research Gas Materials (RGM)
program at the NIST to provide well-characterized reference
materials for situations where the development of SRMs is not
justified111; is continuously auditing the suppliers of Protocol
Gases121; has accelerated its efforts to develop stable,
accurately-certified gaseous reference materials; and is
developing a gas transfer system (GTS) that the regulators and
the regulated can use to prepare small quantities of their own
accurately-certified reference materials.
Using the GTS, organizations can prepare at relatively low
cost a myriad of gas mixtures using a. small number of master gas
mixtures ($200-$10,000 each), a diluent gas ($50-$250 each) and
reusable, low volume compressed gas cylinders ($300-$450 each).
The GTS should be particularly useful when conducting DRE testing
and assessing control device performance testing. It should also
be very useful for gas mixtures whose long-term stability is
known only for concentrations 10 to 100 times higher than those
needed for the measurement systems.
The goal of our program is to provide each EPA regulatory
unit with the design and performance specifications for a fully-
validated, turnkey system they can use to prepare pollutant
standards that meet their needs. A detailed instruction manual
that describes how to operate, clean and maintain the gas
transfer system and also lists the pollutants for which it is
applicable will accompany these specifications. As the AREAL's
research program identifies additional pollutants for which the
system is applicable, this information will be provided to the
EPA regulatory unit(s) affected. The design specifications and
instruction manual will also be provided free-of-charge to any
non-regulatory organization that requests them.
The GTS and the experiments that the AREAL is conducting to
characterize it, are described below.
GAS TRANSFER SYSTEM
Developed in cooperation with Scott Specialty Gases Inc.,
Plumsteadville, PA the GTS employs a stainless steel manifold
(Figure 1) containing eight diaphragm packless valves (items V-l
through V-8 in Figure 1), one bellows valve (item V-9 in Figure
365
-------
1), two pressure gauges (0-400 psi and 0-2300 psi) and fittings
to attach five compressed gas cylinders.
All internal parts of the manifold have been treated with
Scott's Acuclean and Aculife processes. The manifold is mounted
on a 3.2 mm thick aluminum plate which is mounted on a. wall at
ManTech Environmental Technology's Commercial Park West facility
in Research Triangle Park, NC. To maintain a constant
temperature, the GTS manifold is wrapped with heat tape covered
with insulation material. All tubing-to-tubing and tubing-to-
valve connections are either orbitally welded or are of the VCR
type.
The manifold can accommodate two master cylinders, a diluent
cylinder and two receiver (audit) cylinders such as the 11 cm OD
x 25.4 cm long aluminum cylinders we are now evaluating for use
with this gas transfer system. These audit cylinders have been
treated with Scott's Acuclean and Aculife processes and pressure
tested to 3000 psi. Their nominal volume at one atmosphere is
1.5 L, which corresponds to 220 L at 2200 psi. They and the
other compressed gas cylinders attach to the manifold using a CGA
to VCR adapter.
EXPERIMENTS TO CHARACTERIZE THE GTS
Analytical System
The GTS can be used for inorganic and organic gas mixtures.
Presently, we are studying only organic mixtures. Two or three
samples are taken from each gas mixture (replicates) and analyzed
as discrete samples using a cryogenic concentrating system and a
HP 5890 Series II/HP 5970 GC/MSD system. For the gas transfer
efficiency studies, the GC/MSD system uses the master gas mixture
as the reference standard. For most of the compounds in the gas
mixtures being evaluated, the precision of the analytical system
is between 5 and 10%.
Leak-check of GTS
The GTS was found to be leak-free when pressurized to 350
psig with helium and checked for leaks with a helium leak
detector.
Procedure to Clean Audit Cylinders
An eight-cycle pressurization (to 30 psig with VOC-free
nitrogen)/evacuation (to 50 mm Hg) procedure effectively removed
even ppm levels of the VOC checked to date. We have not yet
attempted to optimize the cleaning procedure for the cylinders.
Procedure to Clean Audit Cylinder Regulators
Attaching the Model 19 regulator to a pressurized audit
cylinder and letting the gas flow through the regulator for 70
minutes effectively cleaned even regulators from audit cylinders
that had contained ppm concentrations of VOCs.
366
-------
Check of GTS Manifold Cleanliness
After the GTS system was assembled, a VOC-free gas was
passed through it and collected in two precleaned audit cylinders
and in two Summa polished S.S. canisters. No VOCs were found
when the cylinders and canisters were analyzed. Similar results
were obtained after the GTS had been used to fill audit cylinders
with ppb and ppm mixtures and then cleaned using the eight-cycle
pressurization/evacuation procedure described above. We have not
yet attempted to optimize the cleaning procedure for the
manifold.
Transfer Efficiency Tests
The gas mixtures being used contain VOCs at the
concentrations needed to make gas mixtures that will meet present
regulatory requirements. The efficiency of transfer is being
evaluated as a function of GTS temperature, gas concentration,
VOC, audit cylinder pressure, GTS flow rate settings, and GTS
dilution ratio. Since the precision of the analytical system is
between 5 and 10% for most of the compounds, the target goal for
efficient recovery is 90 to 110% of the nominal concentration.
Three gas mixtures have been studied to date:
Mixture A. A gas mixture containing 14 aromatic and
chlorinated hydrocarbons at 3 to 14 ppm was used to fill two
audit cylinders to 800 psig. The mean percent recoveries for 11
of the 14 VOCs were 90 to 110%; all percent recoveries were
between 85 and 115%.
Mixture B. Three volumes of a gas mixture containing 20
aromatic and halogenated hydrocarbons at 5-10 ppb was blended
with one volume of another mixture containing 9 unique alkane
compounds at 20 ppb and the resulting mixture was placed in two
audit cylinders to 800 psig. The mean percent recoveries for 13
of the 29 VOCs were between 90 and 110%; for 20 of the 29 they
were between 85 and 115%.
Mixture C. A cylinder containing 34 oxidant precursor VOCs
at 45ppb was used to fill two audit cylinders to 1530 psig. The
mean percent recoveries for 33 of the 34 VOCs were between 90 and
110%.
Stability Studies
The effects of cylinder pressure, gas mixture humidity, VOC
concentration and storage conditions will be evaluated for
120-day periods. The following procedure will be used initially.
Two audit cylinders will be filled to 2000 psig using the GTS and
the gas transfer efficiency will be assessed. The two audit
cylinders will be reanalyzed versus the master gas mixtures at 30
days, 60 days, 90 days and 120 days. After the 30-day analysis
the pressure in one of the cylinders will be reduced to 1000
psig; after the 60-day analysis it will be reduced to 500 psig.
Thereafter it will be analyzed without reducing the pressure
after the analysis. Each gas mixture will be studied for
367
-------
stability in the audit cylinders at 1000, 200 and 10 ppb.
Audit Cylinder to Sampling System Transfer Efficiency
If the gas mixtures cannot be quantitatively delivered from
the audit cylinders to the measurement systems under field
condition, the value of the GTS approach for the regulatory
community will be reduced. Available gas delivery systems will
be evaluated as a function of flow rate, VOC concentration, gas
stream humidity/temperature, and audit cylinder pressure to
develop reliable delivery systems for each gas mixture. The
candidate delivery systems will be validated under laboratory and
field conditions.
REFERENCES
1. W. J. Mitchell and W. E. May, "Two New Gas Standard Programs
at the National Institute of Standards and Technology."
Paper presented at U.S. EPA/A&WMA Symposium on the
Measurement of Toxic and Related Air Pollutants. Durham, NC,
May 5, 1993.
2. A. Hines, W. Mitchell, M. Miller and R. Brande, "EPA's QA
Program on the Suppliers of Protocol Gases." Paper
presented at U.S. EPA/A&WMA Symposium on the Measurement of
Toxic and Related Air Pollutants, Durham, NC, May 5, 1993.
Vacuum
V
Low High
Pressure Pressure
-1-1
Figure 1. Gas transfill system
368
-------
Transfill System
Diluent Gas
Low High
Pressure Pressure
Vent
-------
ROUND-ROBIN ANALYSIS OF PERFORMANCE EVALUATION SAMPLES
BY STANDARD ARI 700-88
Shirley J. Wasson and James B. Flanagan
Research Triangle Institute
3040 Comwallis Road
Research Triangle Park, NC 27709
ABSTRACT
This paper describes the results of a round-robin interlaboratory study of analyses for selected
contaminants in reclaimed refrigerant by Air-Conditioning and Refrigeration Institute (ARI) Standard
700-88. Performance evaluation audit samples for contaminant analysis of chlorofluorocarbon
refrigerants were prepared using previously described techniques developed at the Research Triangle
Institute (RTI). Four laboratories participated in the study. Four refrigerants were studied: R-ll, R-
12, R-22, and R-502. Three analytes — impurity refrigerant, moisture, and high boiling residue
(primarily compressor oil) - were studied at two concentration levels. Level one was near the
standard for reclaimed refrigerant and level two was at a higher concentration. For the impurity
refrigerant analyses, the reported range of precision (relative standard deviation) was from 8 to 39%,
with an average bias between -12.7 and +3.7%. For the moisture determination, precision ranged
between 8 to 36% in the different CFC matrices, and bias ranged from -34.9 to +6.7. High boiling
residue analyses yielded a precision range of 6 to 51%, and relative bias was between -18.7 and
59.5%. The major factors affecting total error appeared to be container effects such as cleanliness and
wall adsorption; characteristics of the matrix refrigerants, including vapor pressure and affinity for
water, and the variability among laboratories.
INTRODUCTION
Chlorofluorocarbons (CFCs) have been employed in domestic and commercial refrigeration for
over half a century. As a result of the spectacular success of this technology, a massive amount of
these materials, which have been implicated in stratospheric ozone depletion and global climate
change, is now in the installed base of refrigeration equipment. The existing stock of chlorine-
containing refrigerants must be kept in use until replacement materials can be developed,
manufactured, and phased into the marketplace. Meanwhile, recycling the refrigerants is an important
means of minimizing disruption of the changeover while limiting further manufacture of CFCs. The
Air-Conditioning and Refrigeration Institute (ARI) has published Standard 700-88 which provides
recommended contaminant limits and methods of analysis to assess the quality of recycled refrigerants.
The objective of the study reported here was to assess accuracy and precision of these techniques in an
interlaboratory comparison study.
To determine levels of contaminants in refrigerants used in commercial refrigeration
equipment, ARI conducted a field test in 1991. The purpose of this study was to determine the level
of contamination that operating refrigeration equipment could tolerate before failure occurs. Ten
different categories of refrigeration equipment were studied. Individual units were 3 to 6 years old.
Procedures, methods, and target analyte levels similar to those of ARI Standard 700-88 were used for
370
-------
this study. EPA provided Quality Assurance (QA) assistance to the ARI project by contracting with
RTI to prepare several sets of performance evaluation audit (PEA) samples that contained selected
contaminants of interest and to conduct an interlaboratory study of analytical performance. Clean
refrigerants were spiked with known amounts of specified contaminants to make these samples. Sets
of these samples were to be sent to the laboratory used in the ARI field study and to three other
laboratories who routinely analyze contaminated refrigerants. RTI also analyzed one set and one set
was kept in reserve. After the analyses were performed and the data sets reported, RTI conducted the
statistical analysis and made a final report to EPA which attempted to characterize the components of
variability in the data.
DESIGN OF PROJECT
Four laboratories (RTI and three additional laboratories) participated in the round-robin
analysis. The PEA samples consisted of refrigerants spiked with selected contaminants commonly
found in recycled refrigerants. The contaminants chosen for the project consisted of impurity
refrigerant, moisture, and high-boiling residue, which consists primarily of compressor oil and
breakdown products. The CFC refrigerants chosen for study (R-ll, R-12, R-22, and R-502) were the
same refrigerants used in the ARI field study.
The PEA samples were prepared in RTI's laboratories using the methods described in
"Preparation of Performance Evaluation Audit Samples for the Determination of Impurities in CFCs".1
Each participating laboratory was sent a fully labeled set of audit samples in clean, high-pressure
cylinders. Each set of audit cylinders was accompanied by a protocol for sampling and analysis which
consisted of procedures supplied by the laboratory that analyzed the field samples for the ARI study,
accompanied by notes describing RTFs experience in analyzing its set of audit samples using the ARI
methods.
ANALYTICAL METHODS
Impurity refrigerant was analyzed during the field study using a gas chromatograph/thermal
conductivity detector (GC/TCD) with a packed column of porous polymer beads (Poropak). In the
round-robin study, RTI used a gas chromatograph/flame ionization detector (GC/FID) with a packed
column containing perfluoropolyether (5% Fluorocol) on a graphitized carbon black inert base
(Carbopak B). Either method is allowable under the ARI Standard. The identity of the impurity
refrigerant is determined by its retention time. A multipoint calibration curve was constructed for each
refrigerant for accurate quantitation. Calibration standards were made by weight or volume (using the
ideal gas law), depending on the laboratory. Concentrations of the impurity PEA samples were about
0.2% and 0.4%.
Moisture was determined as described in the method, ASTM E 700-79.2 RTI performed the
moisture analyses in a dry box to prevent uptake of water from the air. Methanol was changed after
every sample. Blanks were determined using virgin refrigerant. Check standards consisted of organic
liquid containing known levels of water.
371
-------
High boiling residue by the Standard 700-88 method requires a volume of refrigerant to be
evaporated and the volume of the remaining residue measured. In the prescribed method, a Goetz bulb
is filled to the 100-mL mark with the refrigerant. After evaporation, the remaining residue flows into
a high-precision calibrated sidearm. Because of the difficulty of filling the bulb with exactly 100 mL
of the highly volatile refrigerant, the amounts of refrigerant and residue were also determined by
weight
CALCULATIONS
Partitioning Corrections
The participating laboratories conveyed the analysis results to RTI who evaluated and reported
them to EPA. Because analyses are performed serially, internal evaporation in the pressurized cylinder
causes the ratio of gas to liquid to shift, resulting in concentration changes. This problem is explained
in detail in references 1 and 3. Accurate treatment of the data requires that the concentration of each
analyte be corrected for the amount of refrigerant that has been removed from the container for prior
analyses. For the data reported here, the magnitude of the correction ranged from negligible to about
6%, depending upon the refrigerant and the contaminant in question.
RESULTS
Impurity Refrigerants
Recoveries of the impurity refrigerants are shown in Table I. Recoveries ranged from
approximately SO to 150%, relative to the PEA preparation data. Individual laboratories tended to
range either high or low, indicative of systematic intralaboratory bias. This interlaboratory variability
was a significant contribution to the total error. The impurity analyses were not corrected for blanks
because observed blank: levels from virgin refrigerant were very small.
Moisture
Moisture recoveries are shown in Table II. No systematic difference in average recovery could
be seen between the two hydrophilic refrigerants (R-ll and R-12) and the two hydrophobic
refrigerants (R-22 and R-502). Moisture blanks were too variable to justify correction of recoveries.
Other factors affecting the moisture analysis may be container effects, difficulties inherent in the Karl
Fischer technique, and properties of the refrigerant compounds.
Some of the problems associated with this analysis were (1) low levels: the moisture analyte
was being measured in an atmosphere containing moisture amounts equal to or greater than the 15- to
200-ppm spike; (2) methanol depletion: the scrubber liquid tended to be depleted easily by the large
volumes of gas passing through it which were necessary for obtaining adequate sample for analysis;
(3) high blanks: there were occasional high blanks due either to incomplete cleaning of the sample
cylinders or moisture with the refrigerant as it came from the manufacturer; and (4) lack of
comparable check standards: standard materials for moisture in refrigerant have not been developed.
High Boiling Residue
Residue recoveries by the volume method are shown in Table III, and ranged from about 50 to
over 250%. Over half of the results were biased high, reflecting systematic bias with the analysis.
372
-------
Two possible explanations for the observed bias are difficulties using the Goetz tube with the highly
volatile refrigerants and container contamination.
Because of the high vapor pressure of three out of the four CFCs tested, a significant amount
of material can boil away before the Goetz bulb can be filled to the prescribed 100- mL volume of
refrigerant To overcome this error, both the cylinder and Goetz bulb were weighed before and after
the ana) /sis. Results were obtained both from the volume analysis (as required by the Standard) and
from the weight analysis. Residue recoveries by the weight method varied less, from about 80 to
about 220% as shown in Table IV. This indicates that one source of bias may have been diminished,
but other sources of error remain. An analysis based on weight measurements tended to lower the
bias, but precision was still not good because of the small differences being calculated in comparison
to the weight of the Goetz tube. Laboratory 3 developed a different protocol with a resulting recovery
range of about 100 to 135%, much better than the other laboratories which followed the unmodified
method.
Dirty sample containers may also have been a factor in the residue analysis. Some cylinders
were not clean as evidenced by residual paniculate matter in the blanks which sometimes equaled the
low level (100-ppm) spikes.
CONCLUSIONS
The primary measures of variability for this project were percent bias (bias%) and coefficient
of variation (CV%) calculations. Bias% and CV% for all the analysis types are given in Table V.
Although the data from this study are limited, some trends emerged. The summary precision
figures for the impurity analysis give an indication that the variation among laboratories was an
important factor. This may be due to variability in methods of calibration standard preparation and to
the use of different analysis methods (e.g., GC/TCD vs. GC/FID). The lack of precision in the
moisture analyses may have been a result of instrumental and procedural difficulties. Precision and
bias was uniformly high for the residue analyses except for the level 2 analyses by weight. Evidently
the amount of spike residue at level 2 was great enough to overcome the errors introduced by use of
unclean cylinders, and the errors involved with calculating small differences between large weights.
The method of analysis by weight may be more accurate and precise than the current volumetric
method.
Partitioning corrections were essentially insignificant to the calculation of impurity analysis
recoveries, but could make as much as a 6% difference in the residue analysis, depending on the order
of analysis. Impurity refrigerant was measured first, when the cylinder contents were nearly all in the
liquid phase. Residue was measured last, when much of the refrigerant was in the gas phase,
enriching the liquid phase in the residue.
REFERENCES
1. S.J. Wasson, S.V. Kulkarni, C.O. Whitaker, and D.L. Harmon, "Preparation of performance
evaluation audit samples for the determination of impurities in CFCs," in Proceedings of the
1992 U.S. EPA/A&WMA International Symposium on Measurement of Toxic and Related Air
Pollutants. Durham, NC, May 3-8, 1992, pp. 1-6.
373
-------
American Society for Testing and Materials, "Standard test method for water in gases using
Karl Fischer reagent," ASTM E 700-79, reapproved 1990.
S.J. Wasson, Validation Study of Analytical Methods for Reclaimed Refrigerant. EPA report
(in publication), U.S. Environmental Protection Agency, Research Triangle Park, NC, 1993.
374
-------
Table I. Impurity refrigerant recoveries, percent (W/W).
Laboratory
Primary
Refrigerant
R-ll
R-12
R-22
Contaminant
R-12 level 1
R-12 level 2
R-22 level 1
R-22 level 2
R-12 level 1
R-12 level 2
1
99.2
86.3
74.2
78.1
116.5
97.2
2
113.3
121.1
101.9
902
89.8
95.7
3
94.9
68.5
80.1
86.1
57.2
57.2
4
ND*
ND
125.5
94.9
151.1
102.8
Note: This table reflects no blank corrections.
°ND = not determined
Table n. Moisture recoveries, percent (W/W).
Primary
Refrigerant
R-ll
R-12
R-22
R-502
Laboratory
Contaminants
H2O, level 1
H2O, level 2
H2O, level 1
H2O, level 2
H2O, level 1
H2O, level 2
H2O, level 1
H2O, level 2
1
90.7
92.3
89.6
65.8
91.4
80.5
87.1
65.5
2
116.8
116.3
81.3
52.1
87.6
89.8
105.3
93.3
3
49.1
53.6
63.9
52.6
101.1
86.9
98.3
88.5
4
125.9
108.7
132.9
90.0
105.7
99.3
136.0
73.6
Note: This table reflects no blank corrections.
375
-------
Table HI. Residue recoveries by volume method, percent (V/V).
Primary
Refrigerant
R-ll
R-12
R-22
R-502
Primary
Refrigerant
R-ll
R-12
R-22
R-502
Contaminant
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Table IV. Residue
Contaminant
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Oil, level 1
Oil, level 2
Laboratory
1 2 3
159 ±76 98 ± 42 72.9 ± .1
52 ±30 78 ± 25 120.2 ± .4
182 ± 40 210 140.4 ± .6
85 ± 14 185 106.1 ± 18.2
68 ±3 176 ±65 175.8 ±48.1
141 ±22 119 104.4 ±14.8
113 ±20 — 142.5 ±5.0
169 ± 19 261 128.9 ± 21.2
recoveries by weight method, percent (W/W).
Laboratory
1 2 3
173 ± 50 121 ± 53 135 ± 8
130 ±14 98 ± 7 124 ±5
218 ± 28 150 120 ± 2
128 ± 28 125 106 ± 4
156 ± 27 133 ± 18 132 ± 8
111 ±11 116 101 ±.4
129 ± 47 133 128 ± 2
104 ±15 115 104 ±3
4
86.3
74.9
105.6
71.3
70.4
92.0
143.9
70.7
4
80.5
108.5
80.0
92.6
118.2
101.5
295.2
80.9
376
-------
Table V. Data quality indicators.
Precision (CV%)'
Contaminant
Impurity,
Moisture,
Residue,
(volume)
Residue,
(weight)
level 1
level 2
level 1
level 2
level 1
level 2
level 1
level 2
R-ll
9.4
29.1
36.0
30.1
36.5
35.0
30.0
12.7
R-12
24.4
8.2
31.9
27.3
28.8
45.4
41.0
14.8
R-22
38.5
23.7
8.7
8.8
503
18.5
11.6
6.8
R-502
19.6
16.1
13.1
50.8
48.2
14.2
R-ll
2.5
-8.0
-4.4
-7.3
4.1
-18.7
27.4
15.1
Bias (%)2
R-12
-4.6
-12.7
-8.1
-34.9
59.5
11.9
42.0
12.9
R-22
3.7
-11.8
-3.5
-10.9
22.6
14.1
34.8
7.4
R-502
—
6.7
-19.8
33.1
57.4
71.3
1.0
CV% = (standard deviation/mean) 100
Bias% = recovery % - 100%
377
-------
EPA'S QA Program on the Suppliers of
Protocol Gases
by
Avis P. Hines and William J. Mitchell
U. S. EPA/AREAL (MD-77B)
Research Triangle Park, NC 27711
and
Matthew Miller and Ronald Brande
ManTech Environmental Technology Incorporated
Research Triangle Park, NC 27709
ABSTRACT
In 1992, EPA's Atmospheric Research and Exposure Assessment
Laboratory initiated a nationwide QA program on the suppliers of
EPA Protocol Gases. The program has three goals: to increase the
acceptance and use of Protocol Gases by the air monitoring
community, to provide a QA check for the suppliers of these
gases, and to help the users of these gases identify suppliers
who can consistently provide accurately certified Protocol Gases.
In this QA program which operates continuously, Protocol Gases
are procured by EPA and the supplier's certification of the
pollutant concentration(s) is verified by EPA. The results are
published on the EPA Technology Transfer Network's electronic
bulletin board. If a supplier's concentration differs from EPA's
by more than 2%, the supplier is notified in writing immediately.
The results obtained for S02, CO and NO Protocol Gases are
presented.
378
-------
INTRODUCTION
The Atmospheric Research and Exposure Assessment Laboratory
(AREAL) of the U.S. Environmental Protection Agency (EPA) has
begun a nationwide audit of the vendors of Protocol 1 (stationary
source) and Protocol 2 (ambient air) Gas Standards. The intent
of this program is as follows:
• Increase the acceptance and use of Protocol Gases as
secondary standards by the air monitoring community.
• Provide a quality assurance check for the vendors of
these gases.
• Assist users of Protocol Gases to identify vendors who
can consistently provide accurately certified Protocol
Gases.
PROCEDURE
Either directly or through third parties, EPA continually
procures Protocol Gases from commercial sources, checks the
accuracy of the vendors' certification of concentration, and
examines the accompanying documentation for completeness and
accuracy.
For Protocol Gases the maximum allowable deviation from the
certified value is 2%. Accuracy of the certification is checked
using Standard Reference Materials (SRMs) If the difference
between the EPA-determined and the vendor-determined
concentration is more than 2%, or if the accompanying
documentation is incomplete, EPA notifies the vendor immediately
to resolve and correct the problem.
Results of EPA certification checks are placed on two
bulletin boards, EMTIC (Emission Measurement Technical
Information Center) and AMTIC (Ambient Monitoring Technology
Information Center), on the Technology Transfer Network of the
EPA Office of Air Quality Planning and Standards. Also included
are notes describing any corrective action taken by a vendor
after being notified of a. problem with a Protocol Gas.
Bulletin board entries are organized in tables by gas (all
nitric oxide results, for instance, are summarized in one table)
and by vendor. Numerical data are supplemented by narrative
footnotes explaining the results of any corrective action taken
by the vendor. Thus the entries provide a continuous record of
all audit activities. Table 1 lists the information presented on
the bulletin boards. The bulletin boards can be accessed by
following the procedure in Table 2.
Users who believe that their Protocol Gas has been certified
incorrectly may contact Ms. Avis Hines of EPA/AREAL
(919-541-4001) to request an EPA certification check. If EPA
379
-------
accepts the gas cylinder for testing, the results of these tests
will also be posted on the bulletin boards.
REQUIRED DOCUMENTATION
The Protocol Gas procedure requires two types of
documentation to accompany the gas cylinder: a Certificate of
Analysis, which may be mailed separately or attached to the
cylinder; and a cylinder tag which must be attached to the valve
under the valve cap. Documentation is incomplete until the
vendor provides every item required to be on the certificate and
on the tag.
RESULTS
This section of the audit report, organized by gas and by
vendor, is updated whenever EPA conducts a new audit or receives
corrective action reports from a vendor. It allows users of
Protocol Gases to easily review the comparative performances of
the vendors.
The standard of comparison used throughout the subsequent
tables is the relative percent difference between the vendor and
the EPA values with the EPA-determined concentration serving as
the reference value.
Tables summarize audit results for each gas, with footnotes
describing corrective actions taken by the vendors. Each vendor
is assigned a footnote letter; all notes pertaining to that
vendor will be listed chronologically in one place. If a vendor
has more than one plant, each plant is assigned its own footnote.
Notes may not be necessary for every vendor on every audit.
For the first three audits (nitric oxide, sulfur dioxide
and carbon monoxide), gas cylinders obtained through third
parties were analyzed in triplicate by EPA and by another
laboratory. Each laboratory used its own SRMs, followed the
Protocol Gas certification procedure, and reported the analytical
results using both the linear regression method and the ratio
method specified in the procedure. Because a statistical
analysis showed that the results from EPA and the independent
laboratory were indistinguishable, only the EPA results are now
shown on the bulletin boards. The other independent laboratory
serves as a referee laboratory whenever differences occur between
results from EPA and the vendors.
Nitric Oxide
The eight vendors passed the analytical part of the audit
for nominally 40 ppm nitric oxide. Five vendors provided
incomplete documentation; upon notification of the deficiencies,
the vendors immediately provided the missing items.
380
-------
Sulfur Dioxide
Only three of the eight vendors had cylinders of nominally
50 ppm sulfur dioxide which passed the analytical part of the
audit. Incomplete documentation continued to be a problem. Only
one vendor had both an acceptable value and complete
documentation.
A review of vendor documentation revealed that cylinders
from three vendors were certified by the same laboratory. All
three, failing the EPA analyses and suspecting a problem with the
standard used, recalled their cylinders for re-analysis.
Carbon Monoxide
The eight vendors passed the analytical part of the audit
for the nominally 40 ppm carbon monoxide. Incomplete
documentation, however, remained a problem five of the eight
vendors did not provide complete documentation until EPA
contacted them.
Disclaimer
The information in this document has been funded wholly by
the United States Environmental Protection Agency. It has been
subjected to Agency review and approved for publication.
381
-------
Table 1. Information presented on the electronic
bulletin boards
1. Vender name and location
2. Cost of Protocol Gas
3. Vendor gas concentration/date, of analysis
4. EPA gas concentration/date of analysis
5. % difference from EPA concentration (Max 2%)
6. Accuracy and completeness of documentation
7. Corrective action taken by vendor
382
-------
Table 2. Procedure for accessing the OAQPS TTN electronic
bulletin boards
Note: One must be a registered user of the TTN to access the
EMTIC or the AMTIC. For information on how to become a user call
919-541-5384.
The procedure to download and subsequently print the
Protocol Gas audit document depends on the communication software
used to connect to the TTN and the software used to work with
text files. However, if you just want to view the document on
the screen follow the steps below:
EMTIC:
1. Connect to the OAQPS TTN.
2. Enter your last name, first name, and
password when prompted.
3. From the Top Menu of OAQPS TTN choose EMTIC.
4. When prompted, choose ontinue on to EMTIC.
5. From the EMTIC BBS Menu choose EMTIC Documents.
6. From the next Menu choose ORD/AREAL Documents
7. A list of files will appear which will include the
audit document called "PROTOGAS.TXT"
8. To view the document type "p 1 protogas.txt".
9. For instructions on how to download the document using
your particular communication program refer to the
OAQPS TTN User's Manual or call their Help line
(919-541-5384)
AMTIC:
1. Same as steps 1 and 2 above.
2. From the Top Menu of OAQPS TTN choose AMTIC.
3. From the AMTIC BBS Menu choose Available Related
Documents.
4. From the next Menu choose ORD/AREAL Documents
5. A list of files will appear which will include the
audit document called "PROTOGAS.TXT".
6. To view the document type "p 1 protogas.txt"
1. For instructions on how to download the document using
your particular communication program refer to the
OAQPS TTN User's Manual or call their Help line
(919-541-5384).
383
-------
ONE SIZE DOES NOT FIT ALL: A PANEL DISCUSSION ON
QA APPROACHES TO AIR Toxic ISSUES
Shrikant Kulkarni
Center for Environmental Measurements and Quality Assurance
Research Triangle Institute
Research Triangle Park, North Carolina 27709
A panel of Quality Assurance managers from various EPA research laboratories and ,
program offices in the Research Triangle Park convened to discuss quality assurance
approaches to air toxic issues at Session 9 of the 1993 AWMA/EPA International
Symposium. The panelists included Ronald K. Patterson (AREAL), Judith S. Ford (AEERL),
Ronald R. Rogers (HERL), Rick Johnson (OIRM), and Gary Johnson (QAMS).
During the introduction, the importance of applying quality assurance/quality control
(QA/QC) to data collection activities was stressed. It was also emphasized, however, that
because of the large diversity in the research activities and objectives of the laboratories and
program offices, the nature of the QA/QC activities varied considerably within each
organization.
After the panel members were introduced, each made a presentation on the mission of
their organization, how QA was currently being used in the organization, and the types of QA
needed. Following is a summary of each presentation.
Ronald K. Patterson. QA Manager for the Atmospheric Research and Exposure Assessment
Laboratory
Mr. Ron Patterson began by giving a description of his organization. The
Atmospheric Research and Exposure Assessment Laboratory (AREAL) operates under
the EPA Office of Research and Development The Laboratory employs 222 scientists
and technicians, mostly in the disciplines of chemistry, physics, meteorology, and
computer science. The basic mission of AREAL is to conduct intramural and
extramural research related to air pollution exposure assessment and the collection and
characterization of air pollutants emitted from stationary, mobile, and biogenic sources.
This mission includes the study of air pollutant formation, transformation, and
transport mechanisms; the study of source-receptor relationships; the assessment of
human and ecosystem exposures to air pollutants; the development and application of
predictive mathematical models; the development of instrumental and analytical
methods; and the study of air pollution trends and patterns. AREAL also provides
support and technical assistance to the regulatory and enforcement sides of the Agency
and to state and local governments.
384
-------
Mr. Patterson explained that to achieve the support aspects of this mission,
AREAL published air pollution QA/QC guidelines in the Code of Federal Regulations
(CFR Part 58) and developed the five-volume "Quality Assurance Handbook for Air
Pollution Measurements" (the "Red Books"). The Laboratory also prepares standards,
conducts, audits on criteria air pollutants, and conducts research and development on
non-criteria pollutant standards, audit devices, and Certified Reference Materials
(CRMs). These activities are the responsibility of AREAL's Quality Assurance and
Technical Support Division.
Mr. Patterson emphasized that the diversity of such a laboratory demands a
flexible QA program. Since the bulk of the activities in the Laboratory are of a
research and development nature, the QA program must focus on more than data
collection. AREAL is beginning to focus on the "quality of science," value-added
quality systems, and up-front planning with clear objectives, well-documented
implementation plans, timely project assessment, corrective action, and peer review of
the final product. These are essential quality systems for AREAL. The Laboratory
also emphasizes the customer/supplier relationships associated with projects and
maintains a strong customer focus. The quality systems approach attempts to foster
teamwork and sharing between the customer and supplier and strives to manage the
quality of project outputs. In conclusion, AREAL has determined that there is no
"cookie-cutter" approach and is very aware that "one size does not fit all."
Judith S. Ford. QA Manager for the Air and Energy Engineering Research Laboratory
The Air and Energy Engineering Research Laboratory (AEERL) develops and
assesses methods and technologies for preventing or reducing the deleterious effects of
air pollutants on human health and welfare, and on the global environment Applied
research is conducted by the Global Emissions and Control Division (GECD) and the
Pollution Control Division (PCD).
Ms. Judy Ford explained that the GECD is primarily concerned with
atmospheric environmental problems. Such problems include global climate change,
stratospheric ozone depletion, ozone nonattainment, air toxics, alternative fuels,
accidental releases, and acid deposition. The other division, PCD, is concerned with
air pollution associated with nitrogen oxides and combustion-related air toxic
pollutants; fine particles and toxic paniculate matter; hazardous, municipal, and
medical waste incineration; radon in new and existing buildings; and indoor air
pollutants.
Ms. Ford stated that direct support to measurement activities directed or
performed by these two divisions is provided by AEERL's QA program. The
program is guided by an Agency-approved QA Program Plan, which delineates
QA requirements of Laboratory measurement activities. Examples of
requirements are each data-gathering project must be planned for and approved
prior to initiation of tests; long-term demonstrations and projects with a direct
385
-------
impact on standard-setting or major decision making must be audited; and the
data quality of results must be documented in EPA publications. Ms. Ford also
informed Session attendants that several of AEERL's innovative QA tools have
been adopted by other groups in the Agency, the two most notably being the
QA four-category approach and the QA tracking system.
In summary, Ms. Ford stated that AEERL's QA program is a structured system
not typical of many research laboratories. This is due primarily to the
multidisciplinary nature of the Laboratory's research programs and its need for
expertise in measurement-related fields, such as the chemical and statistical sciences.
Ronald R. Rogers. QA Manager of the Health Effects Research Laboratory
Mr. Ron Rogers began by stating the Health Effects Research Laboratory's
(HERL's) mission: to perform credible, high quality research that will improve EPA's
ability to assess environmental health risks, and to provide advice on the interpretation
and integration of scientific data for risk assessment and regulatory decisions. He
continued by explaining that HERL is the focal point for lexicological, clinical, and
epidemiological research within the Agency. The research program develops and
applies state-of-the science biological assays, predictive models, and extrapolation
methods which serve as the basis for the Agency's health risk assessments.
To fulfill its objectives, EPA must be both a regulatory and a science
agency. Mr. Rogers described the risk assessment process which the Agency
uses to establish and enforce regulations. This process includes four primary
elements:
• Hazard Identification: Is the agent capable of causing an adverse
effect?
• Dose-Response Assessment: What is the quantitative relationship
between dose and effects?
• Exposure Assessment: What exposures occur or are anticipated?
• Risk Characterization or Estimation: Based on the results of dose-
response and exposure assessments, what is the estimated health risk at
anticipated exposures?
Of the first three, which are the primary data-generating elements, the Agency
has historically focused on data quality issues related to exposure assessment for a
variety of scientific, socio-political, and economic reasons. HERL research, however,
is primarily focused on the areas of hazard identification and dose-response, with some
effort applied to exposure assessment. Logically, says Mr. Rogers, the quality of the
386
-------
hazard identification and dose-response data on which regulatory standards are set is
paramount to the effective and efficient protection of human and ecological health.
Unfortunately, because biological responses and standards against which one can
compare these biological responses are virtually nonexistent, traditional ideas about
evaluating and controlling the quality of such data are often not useful. Mr. Rogers
explained that HERL does utilize laboratory-wide QC activities that are focused on
instrument maintenance and calibration to help ensure the quality of certain types of
measurement data, but must rely on other approaches to confirm the quality of graded
response, dichotomous response, or other types of health effects data. HERL has
chosen to ally available traditional QA activities with an expanded emphasis on review
by peers. HERL also has a strong biostatistics staff to support investigators in the
development of appropriate experimental designs. A prospective, in-progress, and
retrospective peer review program would enable HERL to build quality management
into strategic research planning while also taking better advantage of scientific
expertise to judge the quality of HERL research and the data it produces. Mr. Rogers
also said that HERL is currently working to create a system whereby alliance of peer
review with QA can be as practical as it is logical.
Rick Johnson. Office of Information Resources Management
Mr. Rick Johnson began by describing EPA's development of Good
Automated Laboratory Practices (GALPs). Corruption of computerized data by
its contractors and a follow-up national survey of laboratory data management
practices prompted EPA's Office of Information Resources Management
(OIRM) to develop GALPs. GALPs are recommendations to EPA's national
programs of procedures and practices to follow in developing data for EPA's
use in regulating the environment in laboratories that employ automated data
collection, processing, storage, and retrieval technologies.
Mr. Johnson said that the escalating popularity of the GALPs is a case study
that public and private sector regulatory managers should examine. A conflicting
argument can be made that the need for GALP-like specifications is largely the reason
for this atypical enthusiasm for regulatory guidance. According to Mr. Johnson,
however, an equally compelling statement can be made that the GALPs simply make
good sense for federally regulated industries, for suppliers of laboratory automation
technology, and for national and international regulatory organizations and associations
to adopt.
Mr. Johnson explained that the GALPs are a union of two fundamental sets of
established principles: the Agency's Good Laboratory Practices (GLPs) and its
Information Resource Management Policies (IRMs). The GLPs specify a minimum set
of basically manual procedures for manufacturers of pesticide and chemical products
to follow when submitting health and safety data to the Agency. The GLPs were also
developed after EPA found that laboratory data used by the Agency to regulate
pesticides were unreliable. The Agency's IRM Policies prescribe procedures and
387
-------
practices that must be followed when using automated data processing (ADP)
technology. These IRM standards have been widely used for almost two decades.
Management of EPA's information resources is OIRM's primary responsibility
that it executes through the development of ADP policies and oversight of ADP-
related acquisitions.
Gary L. Johnson. Quality Assurance Management Staff
Mr. Gary Johnson concluded the panel presentations. His organization,
the Quality Assurance Management Staff (QAMS), is directed by EPA Order
5360.1 to establish policies and procedures for planning, implementing, and
assessing the effectiveness of quality systems in support of environmental
programs. Such quality systems include traditional QA/QC activities, as well
as the necessary management systems elements for conducting the quality
management process. QAMS is also responsible for developing the
requirements and guidance documents needed to plan, implement, and assess
the individual quality systems established by EPA program offices, Regions,
and research laboratories. Moreover, QAMS is responsible to senior Agency
management for the oversight and periodic assessment of these quality systems
to ensure their effectiveness and sufficiency.
Mr. Johnson explained that during the past decade, the objectives and
missions of many EPA organizations have changed in order to satisfy evolving
environmental issues and concerns. Similarly, the EPA quality system
requirements have changed to reflect this evolution. He emphasized that the
keystone of all environmental programs is that environmental data must be of
the type and quality needed and expected to support decision affecting
rulemaking, enforcement actions, research, and other needs.
According to Mr. Johnson, this year EPA anticipates the adoption of a
new national consensus standard as the basis for its internal quality system:
"Quality Systems Requirements for Environmental Programs" (ANSVASQC
E4-1993). This standard will be the first American National Standard for
environmental quality systems when it is published. EPA has begun the
development of new and revised requirements and guidance documents to
implement this standard within the Agency. Mr. Johnson said that later,
changes will be sought in the acquisition and financial assistance regulations to
require those organizations who receive contract funds or financial assistance
funds to also comply with this standard. These documents will begin to be
available after this summer and through the next 12 to 18 months. The
documents will become part of an Agency internal Quality Manual, along with
a revised EPA Order on quality management, that will undergo Agency-wide
consensus review and approval later this summer.
388
-------
In a parallel effort this summer, Mr. Johnson announced that QAMS
will implement an extensive management assessment of environmental data
operations across EPA using the Management Systems Review (MSR) process.
The MSRs will examine the quality systems applied to key environmental
programs to determine their effectiveness and provide feedback to management
on opportunities for improvement. Air toxics and related programs will be
among those programs reviewed in the next few years, along with other critical
Agency regulatory and research programs.
Mr. Johnson concluded that change is inevitable and the upcoming changes in
the EPA quality system will provide consistency and sufficiency for the remainder of
the 1990s and beyond.
389
-------
Session 10
Source Emissions
and Measurements
-------
DEVELOPMENT OF A TEST METHOD FOR THE
MEASUREMENT OF GASEOUS METHANOL EMISSIONS FROM
STATIONARY SOURCES
B. A. Pate, M. R. Peterson, and R. K. M. Jayanty
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709-2194
F. W. Wilshire and J. E. Knoll
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Methanol was designated under Title III of the Clean Air Act Amendments of 1990 as a
pollutant to be regulated . The U.S. EPA, through a contract with Research Triangle Institute, has
developed a test method for the measurement of methanol emissions from stationary sources. The
methanol sampling train (MST) consists of a glass-lined heated probe, two knockout traps and three
sorbent cartridges packed with Anasorb ®747, a beaded, activated carbon. Anasorb ®747 samples
are desorbed with a 1:1 mixture of carbon disulfide and N,N-dimethylfonnamide. Samples are
analyzed by gas chromatography with flame ionization detection.
Following laboratory testing, field tests of the MST and the National Council for Air and
Stream Improvement (NCASI) sampling method for methanol were conducted at a paper and pulp
mill. The sampling location was an inlet vent to a softwood bleach plant scrubber. In accordance
with EPA Method 3012, two pairs of trains were run in parallel for six runs, collecting a total of
twenty-four samples by each method. During each test run, half of the trains were spiked with a
known amount of methanol. The average percent recovery of the spike was higher for the MST
method than for the NCASI method. Both methods had a relative standard deviation of less than 5
percent. The practical quantitation limit (PQL) was about 2 ppm for the MST method.
INTRODUCTION
A literature search was conducted to review the chemistry of methanol and the sampling and
analysis methods that are currently used to measure methanol. The methanol sampling train (MST)
was developed from information obtained from the literature search and was evaluated in the
laboratory. The MST (Figure 1) consists of a glass-lined heated probe, two knockout traps in an ice
bath, and three sorbent cartridges packed with Anasorb ®747 (SKC). The Anasorb ®747 (or
equivalent) is used to collect methanol and has a recovery efficiency of 98.1 percent. The knockout
traps, used to remove water vapor, also collect a significant amount of methanol. A 1:1 mixture of
carbon disulfide and N,N-dimethylformamide is used to desorb the Anasorb 747 samples.
Condensate and desorption samples are analyzed by gas chromatography with flame ionization
detection. The MST allowed recovery of about 98 percent of methanol sampled during the
laboratory evaluation.
A field evaluation of the MST and the National Council of the Paper Industry for Air and
Stream Improvement (NCASI) sampling method for methanol (Figure 2) was conducted at a pulp
and paper mill. The sampling location was an inlet vent to a softwood bleach plant scrubber. Four
trains were run in parallel (as two pairs) for six runs, collecting a total of twenty-four samples by
393
-------
each method. Half of the trains were spiked with a known amount of methanol during each test run.
The average percent recovery of the spike was 108.3 percent for the MST method and 81.6 percent
for the NCASI method. Although neither method showed a significant bias at the 95 percent
confidence level, the biases of the two methods were significantly different. The practical
quantitation limit (PQL), as defined in Section 9 of EPA Method 301, was about 2 ppm for the MST
method.
Laboratory Evaluation
Sorbent Tests. The Occupational Safety and Health Administration (OSHA) recently
promulgated a method for methanol (OSHA Method 91) which used Anasorb ®747 as the collection
medium. A literature review indicated that Anasorb ®747 was the inost promising sorbent for the
collection of methanol.
The desorption efficiency of methanol from Anasorb ®747 was determined at three different
loadings. A primary methanol standard (94.8 ug/mL) was prepared by diluting 3 mL of methanol to
25 mL with deionized water. Five grams of Anasorb ®747 was placed in each of three separate
vials, which were then spiked separately with 3 «L, 6 uL, and 9 pL of the methanol standard. The
Anasorb ®747 in each vial was desorbed witli 30 mL of a 1:1 mixture of CS2 and DMF. The
average overall recovery for multiple runs of the three samples was 98.1 percent, indicating that
methanol could successfully be recovered from Anasorb ®747.
Sampling Apparatus. A dynamic dilution system was used for mixing methanol and diluent
humidified nitrogen. The components were mixed in a 1-liter dilution flask. The flowrates of the
test gas and the humidified nitrogen were regulated with Tylan® mass flow controllers. The gas
mixture was passed from the dilution flask to a three-port manifold. The dilution flask and manifold
were enclosed in an insulated, temperature-regulated box. A cylinder containing 250 ppm methanol
was used as the test gas and was diluted to the desired concentration using the system described
above.
The sampling train consisted of two condensate traps in an ice bath, three Anasorb ®747
sorbent cartridges and a pump. The sorbent cartridges were modified VOST tubes where the back
end was replaced with a #7 Ace-Thred® joint. A Nutech® VOST control module was used as the
pump. The sampling parameters were 1.0 L/min for twenty minutes.
Breakthrough of a Methanol Spike. The capacity of the MST was tested by conducting three
runs where 4 ing of methanol was spiked onto the train. All runs were conducted at 1 L/min. For
runs 1 and 2, the front tube contained 3 g Anasorb ®747 and the middle and back tubes contained
1.5 g Anasorb ®747. Thirty liters of gas was sampled for the first run. No methanol was found on
the back tube and less than 5 percent was found on the middle tube. When the sample volume was
increased to 60 L for run 2, more than 4 percent of the methanol was found on the back tube and
nearly 20 percent was found on the middle tube, indicating significant breakthrough. For the third
run, 5 g of Anasorb ®747 was placed in the front tube and 60 L of gas was sampled. No methanol
was found on the back tube and less than 4 percent was found on the middle tube. The third run
showed that the MST could collect a significant amount of methanol without breakthrough. The
average recovery of the spike for the three runs was 97.6 percent.
Analytical Systems and Performance. The methanol samples were analyzed on a 30-m
DB-Wax megabore column with a flame ionization detector. The temperature program started at
394
-------
50°C, then increased 10°C/min to a final temperature of 140°C. A conversation with NCASI
personnel indicated there were five possible interferants: chloroform, acetone, acrolein, methyl ethyl
ketone (MEK), and dichloromethane. Methanol was separated from these interferants by the
employment of a 60-m column. The analytical limit of quantitation was 0.7 ng for the Anasorb
®747 samples and 0.5 ng for the condensate samples.
Sample Stability. The stability of methanol on Anasorb ®747 was tested by spiking twelve
Anasorb 747 tubes. Three tubes and a blank were analyzed on the day of the spike (day 0), and on
days 3, 7, and 14. The recovery of methanol from Anasorb ®747 on days 0, 3, 7, and 14 was 102,
80, 86, and 89 percent, respectively.
Field Testing and Method Validation
Field testing of the MST was conducted at two different sites. The primary objective of the
field testing was to validate the MST in accordance with EPA Method 301 procedures. A secondary
objective was to compare the MST to the NCASI method for sampling methanol. Each field test
consisted of four MSTs run in parallel for six runs (24 samples total). The four trains were run as
two pairs, with the two probes of each pair taped together and placed perpendicular to the vent (or
stack) and to the other pair. The method validation procedure used for the MST was also used for
the NCASI method.
Field Site A: Thennomechanical Pulping Mill. The first field site was the stack from an
atmospheric cyclone at a thennomechanical pulping (TMP) mill. Analysis of the samples from the
first pre-survey indicated that no methanol was collected. After consultation with various plant
personnel, it was learned that the process had two modes: the steam could pass through the cyclone
or it could bypass the cyclone and be forced out the stack. The pre-survey samples had apparently
been collected while the steam was passing through the cyclone, which removed water and methanol.
During a second pre-survey at the TMP site, an attempt was made to collect samples while
the steam was being forced through the vent, bypassing the cyclone. The first run was stopped after
30 seconds because condensed steam had filled the condensate knockout traps. The temperature of
the stack gas was measured at 212°F, indicating that the gas was nearly all water. Because sampling
conditions were unsuitable for both methods, selection of a second field site was necessary.
Field Site B: Softwood Bleach Plant. The second field site was a 3-foot diameter inlet vent
to a softwood bleach plant scrubber at a pulp and paper mill. The vent gas temperature was just
above 130°F.
On the first day of sampling, seven NCASI runs were performed. A seventh run was
necessary because the first run was performed at a flow rate of 0.5 L/min rather than 1.0 L/min.
Each run lasted 30 minutes and a sample volume of 30 liters was collected. Each pair of trains was
alternately spiked with 6 uL of a methanol standard. The methanol standard was prepared by
diluting 3 mL methanol to 25 mL with deionized water, resulting in a concentration of 94.8 ug/mL.
Thus, each spike contained about 569 ug methanol, approximately the same amount as collected
during each run.
The MST method was tested on the second day of sampling. Each MST was to have three
sorbent tubes, packed with 5 g, 1.5 g and 1.5 g of Anasorb ®747. However, one of the nylon
bushings required to connect the sorbent tubes was missing. This resulted in the second train of pair
2 having only two tubes, containing 5 g and 1.5 g of Anasorb 747, for all runs. As a precautionary
measure, a seventh run was conducted to ensure enough valid samples for a meaningful statistical
395
-------
analysis. Each pair was alternately spiked with 6 uL of the methanol standard. The heated probes
were inserted about eighteen inches into the 3-foot, vent.
The paired trains for each method showed good precision with a standard deviation below
4 percent. The average spike recovery for the MST (Table 1) was considerably higher than for the
NCASI method (Table 2): 108.3 percent compared to 81.6 percent. The methods had nearly identical
average unspiked concentrations, but the average measured concentration of spiked samples from the
MST was higher than the average measured concentration of spiked samples from the NCASI
sampling trains. The reason for this difference is unknown. Both methods had a similar, and rather
high, standard deviation of their spike recoveries. The high standard deviation of the spike
recoveries was due to a bias that existed between the two pairs of trains. Pair 2 gave results that
were consistently about 6 percent higher than the results from pair 1. As a result, when pair 2 was
spiked, there was a high spike recovery, and when pair 1 was spiked, there was a low spike
recovery. The spike recovery for pair 2 averaged about 27 percent higher than the spike recovery for
pair 1. This discrepancy was found for both methods and the cause is unknown. Because of the
precision between the trains of each pair, it is unlikely that there was a leak problem.
CONCLUSIONS AND RECOMMENDATIONS
The MST and NCASI methods both met the guidelines specified by EPA Method 301. Both
methods showed good precision and their precisions were not significantly different at the 95 percent
confidence level. Even though the bias of the NCASI and MST methods were not significant, there
was a significant difference in bias at the 95 percent confidence level. While the reason for this
difference is unknown, it is possible that the heated glass-lined probes and heated injection ports
used to spike the MSTs were more efficient at transporting methanol than the unlieated teflon
sampling lines and injection ports used to spike the NCASI sampling trains.
An advantage of the MST is that the composition of the emission source can be determined
by using a mass spectrometric detector to analyze the Anasorb ®747 samples. Emissions at the field
site used in this study contained no compounds at detectable concentrations other than methanol. A
field test at a site with a more complex emission matrix would provide important information about
the capabilities of the MST method. A field test at a site with a higher methanol concentration could
help determine the collection capacity of the MST and if there could be problems with breakthrough.
Although the MST has collected large amounts (5 mg) of methanol in the laboratory without
breakthrough, it is unknown what effect other chemical species in the stack gas may have on this
capacity. The MST also needs to be modified so that it can sample high temperature gas streams
with high water content without flooding the train components.
REFERENCES
1. Clean Air Act Amendments, Federal Register. Vol. 55, p. 26942-26952, June 1990.
2. EPA Method 301 - Field Validation of Emission Concentrations from Stationary Sources,
Federal Register. Vol. 56, p. 27370-27374, June 1991.
396
-------
heated
sampling
probe
condensate
knockout
'raP8 front
A <5S)
,
Anasorb 747 Tubes
middle
(1-5 g)
back
(1.5 g)
i 1 I
ice bath
Figure 1. Methanol sampling train.
unheated
teflon
line
water-filled
impinger dry
(20 mL) impinger
silica gel
tube
(520/260 mg)
ice bath
Figure 2. NCASI sampling train for methanol.
397
-------
Table 1. Percent recoveries of field test spikes for the MST method.
8
9
10
11
12
14
Mean
Std.
deviation
Average
Concentration
of Sample
With Spike
(ppm)
42.98
44.99
43.88
47.94
46.13
44.18
45.02
1.79
Average
Concentration
of Sample
Without Spike
(ppm)
31.27
25.23
29.03
32.42
34.03
30.55
30.42
3.06
Average
Measured
Spike (ppm)
11.72
19.77
14.86
15.52
12.10
13.63
14.60
2.94
Average Spike
(ppm)
13.32
13.85
13.21
13.90
13.02
13.33
13.44
0.36
Percent
Spike
Recovery
87.9
142.7
112.5
111.6
93.0
102.2
108.3
19.5
Table 2. Percent recoveries of field test spikes for the NCASI method.
Run
2
3
4
5
6
7
Mean
Std.
Deviation
Average
Concentration
of Sample
With Spike
(ppm)
48.48
45.52
37.62
40.77
38.40
37.78
41.43
4.56
Average
Concentration
of Sample
Without Spike
(ppm)
37.52
31.37
31.04
27.94
27.34
25.56
30.13
4.25
Average
Measured
Spike (ppm)
10.96
14.16
6.58
12.83
11.06
12.22
11.30
2.60
Average Spike
(ppm)
14.16
13.72
13.46
13.87
13.76
13.94
13.82
0.23
Percent
Spike
Recovery
77.4
103.1
48.9
92.5
80.3
87.7
81.6
18.5
398
-------
Development and Validation of a Source Test Method for
2,4-Toluene Diisocyanate
F.W. Wilshire and J.E. Knoll
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
S.C. Foster and J.F. McGaughey
Radian Corporation
Research Triangle Park, NC 27709
ABSTRACT
Four isocyanates are listed for regulation in the Clean Air
Act Amendments of 1990: hexamethylene-1,6 diisocyanate, roethylene
diphenyl diisocyanate, and 2,4-toluene diisocyanate, each of which
is used in the production of polymers, and methyl isocyanate which
is an intermediate in the manufacture of the insecticide carbaryl
(i.e., Sevin®dust).
To support projected regulations, a study is under way to
produce a source sampling and analysis method for the four
pollutants cited above. In the procedure under development, the
isocyanates are collected in an absorbing solution and derivatized
with l-(2-pyridyl)piperazine and analyzed by HPLC with UV
detection. A system was developed in the laboratory to generate
isocyanate atmospheres for optimization of sampling parameters and
chromatographic conditions. The accuracy and precision of the
method is determined in the field using train spiking and
multiprobe sampling following the procedures outlined in EPA Method
301.
A field test of the isocyanate method, following EPA Method
301 procedures, was performed at a flexible foam manufacturer in
the Greensboro-High Point, North Carolina area. The results were
excellent, with analyte spike recoveries of 91% ± 6%. The method's
limit of quantitation (LOQ) was determined to be 351 ng of TDI/M3.
INTRODUCTION
A class of compounds identified as isocyanates are contained
in the list of 189 pollutants to be regulated by the Environmental
Protection Agency under Title III of the Clean Air Act Amendments1
(CAAA) of 1990. There are four isocyanates of interest in the
CAAA; methyl isocyanate (MI), hexamethylene 1,6- diisocyanate
(HDI), methylene diphenyl diisocyanate (MDI), and 2,4-toluene
diisocyanate (TDI).
Isocyanates are used extensively throughout industry. A few
examples of their use are the production of flexible foam products,
synthetic rubber products, insecticides, enamel wire coatings, and
in the pressed board industry. Foam materials alone are widely
used for such diverse items as toys, bedding, seat cushions,
packing material, flotation devices, and as sorbents in the
environmental field. Because of their widespread use, isocyanates
399
-------
possess the potential to affect many who are sensitive to this
class of chemical compounds. Some of the adverse physiological
effects associated with exposure to isocyanates are severe skin and
eye irritations, eczema, nausea, and bronchial asthma.2 An example
of the concerns expressed about human exposure to the isocyanates
is demonstrated by the NIOSH IDLH level (concentration considered
Immediately Dangerous to Life or Health) for 2,4-toluene
diisocyanate, which is listed at 10 ppm.3
Several critical problems exist when sampling for isocyanates.
They polymerize in the presence of concentrated alkaline compounds,
decompose upon exposure to water and alcohols, discolor upon
exposure to sunlight, and form toxic gases, such as carbon monoxide
and hydrogen cyanide, upon decomposition. Consequently,
isocyanates must be collected rapidly and stabilized immediately
with a derivatizing reagent to insure sample integrity.
The EPA's Source Methods Research Branch, in the Atmospheric
Research and Exposure Assessment Laboratory, Research Triangle
Park, North Carolina through a contract with Radian Corporation,
Research Triangle Park, NC, has developed a method for the
collection, identification, and measurement of one of the
isocyanates of interest; 2,4-toluene diisocyanate. In this method,
samples are collected at the source using a modified EPA Method 5
sampling train. Stack gas is withdrawn at a flow rate of 0.5
ft'/min. for approximately sixty minutes through a heated, glass-
lined probe and into two impingers containing a solution of the
derivatizing reagent, l-(2-pyridyl)piperazine in toluene. The
first impinger is fitted with a water-cooled condenser, on the
outlet of the impinger, to minimize carryover of the organic
solvent into the second impinger. The 2,4-toluene diisocyanate
reacts with the derivatizing reagent to form a stable TDI/urea
derivative. When sampling is completed, the probe and connecting
glassware are rinsed with toluene and acetonitrile and the rinses
are saved for laboratory analysis. Each impinger solution
(TDI/urea derivative) is recovered separately and saved for
laboratory analysis. All samples are stored in a cooler at 0 to 4°
C until returned to the laboratory for analysis by High-Performance
Liquid Chromatography (HPLC with UV detection). Quantitation is by
a standards/retention time comparison procedure.
EXPERIMENTAL
Laboratory Evaluation
The laboratory study was initially set up to meet seven
objectives for the four isocyanates of interest. However, midway
through the laboratory study, budget constraints required the focus
to be shifted to only one of the isocyanates of interest. After
discussions with personnel in the EPA's Office of Air Quality
Planning and Standards, it was decided to focus our efforts on 2,4-
toluene diisocyanate. The seven objectives originally planned for
all of the isocyanates were followed for TDI and are listed in
Table 1. Some of the objectives were also met for the other
isocyanates of interest and are also listed in Table 1.
400
-------
Derivative Formation
Efforts to form a stable isocyanate/urea derivative for all
four isocyanates, using ethanol as the derivatizing reagent were
only marginally successful. An absorption solution was prepared by
adding 1 gram of KOH to 500 mL of 99.9% ethanol. Standard
solutions of each of the isocyanates were prepared by adding the
isocyanate directly to 5 mL each of the ethanol/KOH solution as
follows: 30 mg of MDI; 10 uL of HI, HDI, and TD1. Solid
derivatives for MDI, HDI, and TDI were obtained, but formation of
a derivative for MI was unsuccessful. Chromatograms for the
derivatized and underivatized isocyanates were compared. No
chromatographic peaks were observed for MI or HDI either
derivatized or underivatized, however, peaks were detected for TDI
and MDI.
Previous work by Goldberg, et al.*, using l-(2-
pyridyl)piperazine as the derivatizing agent, investigated
collecting ambient air samples in midget impingers. Since no
current source method for isocyanates exists, the secondary amine,
l-(2-pyridyl)piperazine [1,2PP] was investigated as a possible
alternative derivatizing reagent for ethanol. Using the 1,2PP as
the derivatizing reagent resulted in the formation of solid
derivatives for all of the isocyanates of interest. Each
isocyanate was prepared in a separate 200 mL flask. Approximately
0.2 grams of TDI, HDI, MDI was added to separate solutions of
0.3 mL of 1,2PP and 10 mL of acetonitrile (ACN). The solutions
were allowed to stand for 24 hours to insure enough time for the
reaction to take place. Each derivative was then rinsed with 150
mL of distilled water and allowed to air dry before being
redissolved with acetonitrile and brought to a standardized volume
prior to analysis by HPLC. A derivative for the MI was prepared by
transferring 100 uL of MI to 1 mL of ACN and adding 300 uL of
1,2PP. The solution was shaken for five minutes and then diluted
1:1000 for analysis by HPLC. A 1,2-PP solution was prepared as
previously mentioned for blank analysis on the HPLC. Also, a
solution of the 1,2-PP with MI, HDI, MDI, and TDI was prepared to
determine the retention time of each derivative. The results were
excellent, demonstrating that a mixture of the four isocyanates
could be analyzed with good chromatographic separation and
quantitation.
Isocyanate Generator
An isocyanate atmosphere generator was constructed to provide
a source of isocyanates as a simulated source, for testing within
the laboratory. It is expected that this generator will be
applicable to all four of the isocyanates listed in the CAAA, but
for the reasons explained earlier the generator was tested only for
TDI.
A modified Method 5 sampling train (without the in-line
filter) was set up in the laboratory. Attached to the end of the
probe was a piece of heated 0.5 inch quartz tubing with a stainless
401
-------
steel tee and septum (for introduction of the isocyanate(s)
standard). The temperature of the probe and quartz tubing were
maintained at 120° C. Five impingers were connected in series for
this study. The first impinger was a Greenberg-Smith impinger and
the four following impingers were modified Greenberg-Sraith
impingers (straight stem - no tip). Room air was pulled through a
charcoal scrubber into the heated quartz tubing and subsequently
into two impingers containing the 1,2-PP absorbing solution.
Following the two impingers containing the absorbing solution were
one empty impinger (trap), a silica gel impinger, and an impinger
containing charcoal (scrubber). A TDI standard in methylene
chloride (MeCl2), prepared as described earlier, was introduced by
a motor driven syringe pump, through the septum and into a heated
air stream. Room air was sampled at a rate of 0.5 fts/min for 40
minutes. This flow rate was chosen to test collection efficiencies
at a flow rate expected to be used during the field testing.
Cleanup and analysis procedures were as previously described, using
toluene and ACN rinses and HPLC-UV analysis. Sample breakthrough,
as measured by the recovery in the second impinger, was less than
8 percent. Mean recoveries for seven sample runs were 77 percent
(see Table 2).
When the data in Table 2 was reevaluated, by eliminating the
obvious outliers (Grubbs t-test for multiple outliers)5, the mean
recovery is 98 +/- 15%. An obvious cause for the two outliers was
not determined, since all operating parameters were standardized
for the seven sample runs. As stated by Snedecor and Cochran in
their chapter on regression analysis and outliers", " when no
explanation is found [for the outliers] the situation is
perplexing. It is usually best to examine the conclusions obtained
with the suspect (i) included, (ii) excluded. If these conclusions
differ materially, as they sometimes do, it is well to note that
either may be correct." Even though the Grubbs test for multiple
outliers indicates that both of the outliers are suspect it is
important to note that should one choose to include the suspect
outliers, the recovery data would still be acceptable according to
EPA Method 301 criteria.7
Method Validation
A field test of the method was performed in February, 1993, at
a flexible foam manufacturer in the High Point, North Carolina
area. A modified Method 5 sampling train (with a water-cooled
condenser on the outlet of the first impinger) was used to collect
source gas from the plant's process vent (see Figure 1). The
sample gas stream was passed through a heated glass-lined/stainless
steel probe and through two impingers containing the 1,2-PP
absorbing solution, one empty impinger (carryover trap), one silica
gel impinger, and one impinger containing charcoal (to trap any
toluene vapors before they could enter the meter box). Sampling
was non-lsokinetic at 0.5 ft'/min for 60 minutes. Non-isokinetic
sampling was performed since a presurvey indicated the analyte of
402
-------
Interest was present in the gas phase. Two of the quad trains for
each sample run were spiked with a TDI derivatized standard (22.5
mg TDI/urea derivative in 15 mL of ACN). This standard spike was
the equivalent of 8 mg of underivatized TDI, which was the amount
indicated by the presurvey that we could expect to collect in sixty
minutes of sampling. Impingers and other glassware used in the
sampling train were rinsed first with toluene and then with ACN.
Probe rinse and associated glassware rinses were combined with the
contents of the first impinger for subsequent analysis by HPLC-UV.
Toluene/acetonitrile rinses from the condenser and second and third
impingers were also combined for HPLC analysis. Samples were kept
on ice at 0 to 4° C until returned to the laboratory.
Operating parameters for the HPLC were as follows:
Instrument: Rainin HPXL delivery system with Waters
710B WISP autosampler.
Data System: Nelson 2600 (1 volt)
Column: Zorbax ODS (4.6 mm ID x 25 cm)
Mobile Phase: ACN/0.1M ammonium acetate buffer
Gradient: 25:75 ACN/0.1M ammonium acetate buffer,
pH 6.2, hold 2 min, then to 60:40 ACN/0.1M
ammonium acetate buffer for 19.5 min.
Detector Wavelength: 254 nm
Flow Rate: 2 mL/min.
Results from the field test were excellent (see Table 3). The
mean recovery of the spikes was 91 +/- 6%. Breakthrough, as
measured by the recoveries in the second impingers, were all less
than 2 percent, indicating near complete recovery in the first
impinger. Background or emissions concentrations (as determined by
analysis of the unspiked trains) ranged from 2000 ug/M3 to 7700
ug/lT. The method's Limit of Quantitation (1^) for TDI, calculated
as outlined in EPA Method 3017, was determined to be 351 ng/M3. The
LQ of the method is defined as ten times the standard deviation of
the mean of the data set whereas the method Limit of Detection (L,,)
would be calculated as 3.3 times the standard deviation of the mean
of the data set.
CONCLUSIONS
A method has been developed for the collection and analysis of
TDI. Method validation procedures are still underway, but
preliminary results from the first field test indicate that the
method can be applied with a great degree of confidence to source
emissions for TDI. Other isocyanate compounds (MI, MDI, HDI) have
been or are being studied, and it is hoped that this method can be
successfully applied to them as well. Current plans are for
another field test in the near future, at another flexible foam
manufacturer or other end user. Conditions not experienced in the
sampling of source emissions during the first field test (i.e.,
higher humidity, particulate loadings, and/or warmer stack gas
temperatures) will be investigated in the next field test.
403
-------
REFERENCES
1. Clean Air Act Amendments of 1990, Public Law 101-549, U.S.
Congress, November 15, 1990, 104 STAT., pp. 2533-2535.
2. Material Safety Data Sheet, No. 331, Genium Publishing
Corporation, Schenectady, NY, Nov. 1978.
3. IDLH Levels, National Institute for Occupational Safety and
Health (NIOSH), Publication No. 78-210, 5 Printing.
4. Goldberg, P.A., R.F. Walker, P.A. Ellwood, and H.L. Hardy,
" Determination of Trace Atmospheric Isocyanate Concentrations
by Reversed-Phase High-Performance Liquid Chromatography Using
l-(2-pyridyl)piperazine Reagent". Journal of Chromatoaraphy,
212, 1981, pp 93-104.
5. Grubbs, F.E., "Sample Criteria for Testing Outlying
Observations", Annals of Mathematical Statistics, Vol. 21,
1950, pp. 27-58.
6. Snedecor, G.W. and W.G. Cochran, Statistical Methods, Sixth
Edition, Iowa State University Press, 9 Printing, p. 158.
7. "Field Validation of Emission Concentrations from Stationary
Sources," Method 301 Federal Register, U.S. Government
Printing Office, Washington, D.C., December 1992.
DISCLAIMER
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication. Mention of
trade names or commercial products does not constitute endorsement
or recommendation for use.
404
-------
Table 1. Isocyanate Project Objectives
No.
1
2
3
4
5
6
7
Objective Description
Find one derivatizing
reagent to react rapidly
with all four isocyanates
Set up analytical HPLC
method (for a single
chromatographic run)
Develop instrument and
method detection limits
Determine spike recovery
from derivatizing reagent
construct an isocyanate
generator
Determine recoveries from
spiked Method 5 train
Field test of method and
validation
MI
yes1
yes
no
no
no
no
no
MDI
yes
yes
yes
no
no
no
no
HDI
yes
yes
yes
no
no
no
no
TDI
yes
yes
yes
yes
yes
yes
yes
Yes indicates that objectives have been met. No indicates
that an attempt has not been made to meet the objectives.
Table 2. Recoveries of Isocyanate (TDI) Spikes
Run No.
1
2
3
4
5
6
7
Mean w/outliers
Mean wo/outliers
Spike Amount
(ug)
2.5
2.5
2.5
2.5
2.5
2.5
2.5
Spike Recovery
(ug)
3.05
0.60
2.30
0.65
2.13
2.53
2.23
Recovery
(%)
122
24
92
26
85
101
89
77 +/- 38
98 +/- 15
405
-------
Table 3. Field Test Spike Recoveries
Sample Train
Identification1
1A
IB
2C
2D
3A
3B
4C
4D
5A
5B
6C
6D
7A
7B
8C
8D
Mean
Spike Amount
(ug)
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
7828
Spike Recovery
(ug)
7436
6654
6732
7280
6810
7280
7436
7593
6888
7436
7515
7671
7826
6575
6732
6419
Recovery
(%)
95
85
86
93
87
93
95
97
88
95
96
98
100
84
86
82
91 +/- 6
Sample trains A&B are paired trains, as are sample trains C&D.
406
-------
Absorbing Solution Empty Silica Gd Charcoal
Toluenu/Piiwraana
Figure 1. Isocyanate Sampling Train Configuration
-------
SIMULTANEOUS SUPERCRITICAL EXTRACTION
OF SEMIVOLATILE AND VOLATILE ORGANIC COMPOUNDS
FROM XAD-28 SORBENT FOR AIR TOXICS AND
STATIONARY SOURCE EMISSIONS
Joette Steger
Steve Hoskinson
Radian Corporation
P.O. Box 13000
Research Triangle Park, NC 27709
This paper presents work performed to demonstrate that supercritical fluid extraction (SFE) could be used to
extract large volumes of XAD-2® resins used for stationary source emissions sampling of air toxics. Basic SFE
equipment capable of extracting 30 grams of sorbent was purchased and assembled. Supercritical fluid
extraction conditions for extraction of source sampling resins were determined.
Average recoveries for three to five replicate extractions of XAD-2® analyzed in triplicate ranged from 32 to
162 percent. For 17 of 32 organic compounds tested, average recoveries ranged between 70 and 103 percent.
Compounds that were quantitatively recovered include ethylbenzene, xylene, N-nitrosodi-n-butylamine,
2-methylnaphthalene, 2-nitroaniline, acenaphthylene, and dibenzofuran. Compounds that were not quantitatively
recovered when present in low (0.1 to 0.5 mg) quantities include hexachloroethane, naphthalene, and 2,6-
dichlorophenol. Results of the supercritical extraction experiments of XAD-2® resin are included in this paper.
INTRODUCTION
The first step in analysis of air samples collected on XAD-28 resin involves separating the organic
compounds from the matrix by Soxhlet extraction with methylene chloride. Solvent extraction techniques are
sometimes time consuming, expensive, and methylene chloride is environmentally unfriendly. Supercritical fluid
extraction could significantly simplify and increase the speed of the isolation process. Supercritical fluid
extraction has the possibility of being fast, efficient, convenient, selective, quantitative, and low in cost after the
initial investment in special equipment. Because the solvent strength of a supercritical fluid is directly related to
the fluid density, the solvating ability of a supercritical fluid towards a particular organic compound can easily be
modified by changing the extraction pressure and temperature. Supercritical fluids with different polarities are
available and the polarity of a supercritical fluid can also be changed by using solvent modifiers. Using different
extraction pressures, solvent modifiers, and solvents with varying polarities may be particularly valuable in
achieving class-selective extractions.
Preliminary comparisons between SFE and Soxhlet extraction reported in the literature have shown that
detection limits, reproducibility, and extraction efficiencies are comparable between the two techniques when
small (1-2 g) sample sizes are extracted. Also, SFE is faster, produces an extract with fewer contaminants, and
can be performed in the field. Supercritical fluids have shown potential in quantitatively extracting polynuclear
aromatic hydrocarbons,1 |2'3alkanes,3 polychlorinated biphenyls,3 pesticides,3 halocarbons,3 and polychlorlnated
dibenzo-p-dioxins4from XAD-2® in sample sizes much smaller then those required for stationary source
emissions. Thus, if target compounds typically found in source samples could be quantitatively extracted from
these resins using supercritical fluids, several advantages would be achieved. These include elimination of the
use of methylene chloride (MeCI^, simultaneous sampling and analysis for volatile and semivolatile compounds,
replicate analysis of samples for precision, and concentration of samples containing low concentrations of
analytes. The objective of this work was to determine if SFE of large volumes of XAD-2® resins used for
stationary source emissions sampling of air toxics by Method 0010°was feasible.
EXPERIMENTAL PROCEDURES
SFE Parameters Study
408
-------
The SFE used: a. was capable of obtaining pressures up to 700 kilograms per square centimeter (kg/cm^,
b. an intensifier pump that included an air input pressure regulator for accurate control of supercritical fluid
output pressure and that was capable of providing supercritical fluid flows In the liters per minute range and c.
an extraction cell. The extraction cell was assembled from a threaded preparative high performance liquid
chromatographic (HPLC) column (30 mm by 15 cm) with Kel-F end fittings rated to 350 kg/cm2. Supercritical
fluid chromatographic (SFC) grade CO2was used as the extraction fluid. A polar modifier, methanol (MeOH),
was used, since some of the analytes were polar.
The XAD-2® was prepared following the procedures and quality control criteria in EPA 600/7-78-201.6
Approximately 30 g of resin was packed Into the extraction cell. The remaining space in the cell was filled with
silane-treated glass beads using silanized glass wool to separate the XAD-2® and glass beads. The extraction
cell was packed so that the supercritical fluid passed through the XAD-2® resin first and then through the glass
wool and beads. Two spiking solutions were prepared with 32 organic compounds in MeOH. The compounds
and the levels of each are shown in Table 1. The spiking solution was injected into the center of the resin using
a gas-tight syringe.
After the extraction system was set up and checked, preliminary extractions using XAD-2® were performed
to determine extraction conditions. All preliminary extracts were screened by GC/FID to provide immediate
feedback. Later selected samples were analyzed by gas chromatography with mass spectrometry (GC/MS)
detection.
Initially two extract collection procedures were tried, cryotrapping and collecting the extract in MeOH. The
extract was collected in 2 to 5 mL of MeOH since MeOH was already present as the modifier. MeOH has a low
freezing temperature (-95°C), and most of the analytes were soluble in MeOH. The collection vessel was a 40-
mL vial and the fluid flow rate was adjusted to minimize loss of analytes due to aerosol formation. Flow rate was
controlled using a flow restrictor. Different restrictors were evaluated including crimped nickel tubing (1/16-Inch
outer diameter) and various diameters of fused silica tubing (0,53, 0.1 and 0.05 mm inner diameter).
Recovery and Accuracy Studies
Once the extraction conditions were determined, ten SFEs were performed on XAD-2® to measure recovery
and precision. Two extractions were performed on blank XAD-2®. Five replicate extractions were performed on
XAD-2® at a low semivolatile and high volatile organic compound concentration and three replicate extractions
were performed at a high semivolatile and low volatile organic compound concentration. Analysis of the
supercritical extractions was by gas chromatography with a flame ionization detector (GC/FID). Each extract
was analyzed in triplicate. Because relative and not absolute recovery values were required, the instrument was
calibrated with a one-point standard containing each of the components. The one-point calibration standard
represented the levels expected for 100 percent recovery of each compound. All results were compared to this
one-point standard.
RESULTS AND DISCUSSION
SFE Parameters Study
For the first three experiments, the supercritical fluid extract was eluted through the cryotrap and then
through a bubble trap. The cryotrap was a stainless steel loop submerged in frozen acetonitrile (ACN). The
bubble trap contained MeOH and was immersed in an ice bath. The outlet of the cryotrap was submerged in
the chilled MeOH. At the completion of the extraction, the cryotrap was rinsed with MeOH. Analysis of the
MeOH rinse plus MeOH from the bubble trap showed that no analytes were present. Results from these three
experiments indicated that material either was not deposited in the cryotrap/bubble trap or was not being
effectively removed from the cryotrap/bubble trap. Therefore, the use of the cryotrap was discontinued
since the spike was not detected in the cryotrap.
Additional experiments were performed using a collection system consisting of a midget impinger containing 5
mL of MeOH. The supercritical fluid entered the impinger through a piece of 1/16-inch outer diameter nickel
tubing. Decompression of the supercritical fluid was controlled by crimping the end of the nickel tubing. The
crimped end of the nickel tubing was submersed in the MeOH solution. The impinger was immersed in a dewar
containing a slush of MeOH and liquid argon. Acetonitrile and liquid argon were also evaluated as a low
temperature collection trap bath in a single experiment to arrive at the MeOH-liquid argon low temperature bath.
409
-------
TABLE 1. SUPERCRITICAL EXTRACTION RESULTS
Compound
Volatile Compounds
Benzene and 1,2-Dichloroethane°
1 ,2-Dichloropropane and
Dibromomethane0
Toluene and Trichloroethylene0
Chlorobenzene and
1,1,2-Trichloroethanec
Ethylbenzene
Xylene
1 ,1 ,2,2-Tetraohloroethane
Semivolatile Compounds
Phenol and 2-Chlorophenol°
Hexachloroethane
Naphthalene
2,6-Dichlorophenol
N-Nitrosodi-n-butylamine
2-Methylnaphthalene
2-Nitroaniline
Acenaphthylene
3-Nitroaniline and Acenaphthene0
Dibenzofuran
1 -Naphthylamine and
2,4-Dinitrotoluene°
1,2-Diphenylhydrazine and
Diphenylamine0
Heptachlor Epoxide and
Fluoranthene0
DDE
ODD
Endosulfan Sulfate
Amount
Spiked (mg)
0.432
0.701
0.441
0.489
0.163
0.163
0.307
6.148
1.596
1.376
1.611
1.961
3.572
2.090
1.581
2.964
1.429
2.789
3.572
3.344
1.566
2.090
1.657
Recoverya(%) Amount
Spiked (mg)
66.7
76.8
88.8
86.2
80.6
82.1
85.7
83.8
87.1
86.2
74.5
84.0
87.3
80.8
83.5
82.5
82.3
88.2
81.8
42.2
66.3
60.1
68.1
+ 11.5
± 7.4
± 10.2
± 9.2
± 15.3
± 8.5
± 13.8
± 11.5
± 9.7
± 8.3
± 9.4
+• 10.2
± 8.8
+ 19.0
± 10.1
+ 10.1
± 10.8
± 16.4
± 9.0
± 5.3
± 7.5
+ 7.2
± 8.9
3.600
5.840
3.680
4.080
1.360
1.360
2.560
0.647
0.168
0.145
0.170
0.206
0.376
0.220
0.166
0.312
0.150
0.293
0.376
0.352
0.165
0.220
0.174
Recoveryb(%)
82.8
88.4
88.3
92.2
92.3
91.7
55.4
31.8
36.6
162
0.00
80.0
100
95.5
99.2
90.4
103
69.5
95.7
53.2
137
70.4
94.9
± 10.7
± 12.9
± 11.4
± 10.1
± 10.8
± 11.6
± 16.2
± 11.3
± 5.5
± 61.9
± 0.0
± 13.2
+ 9.2
± 12.9
+ 11.8
± 13.5
± 10.6
± 15.0
± 12.0
± 12.9
± 35.3
± 9.6
± 13.1
"Average of three replicate extractions analyzed in triplicate (n=9).
b Average of five replicate extractions analyzed in triplicate (n=15).
c Compounds coeluted on GC/FID.
410
-------
The crimped nickel tubing allowed a flow rate of 28.57 mL/min of gaseous CO2 This flow was estimated to be
about 30 fjL/min of supercritical fluid. Because the extraction vessel was so large, a flow rate of 1 mL per minute of
supercritical fluid was needed to provide reasonable extraction times (less than two to three hours per sample).
Reproducible control of the flow rate by crimping the nickel tubing was not achievable so various diameters of fused
silica tubing were evaluated.
The 0.53 mm inner diameter fused silica resulted in a flow rate of 5.6 L/mln of gaseous C02. This rate was
estimated to be about 6 mL/min of supercritical fluid which was too high so no extractions were performed using the
0.53 mm ID fused silica. The 0.05 mm ID fused silica provided a flow rate of 570 mL/min of gaseous CO2. This flow
rate was still too slow so a practical internal diameter of fused silica to use for extracting 30 g of XAD-2® was estimated
to be 0.1 mm.
Recovery and Accuracy Studies
Recoveries for the triplicate and quintuplicate replicate extraction studies are reported in Table 1. Average
recoveries for the volatile compounds ranged from 55 to 92 percent. Only benzene/1,2-dichloroethane In the low (0.1 to
0.5 mg) volatile spike and 1,1,2,2-tetrachloroethane in the high (1 to 4 mg) volatile spike had recoveries less than 70
percent.
Average recoveries for the semivolatile compounds ranged from 0 to 162 percent. For the high (> 1 mg)
semivolatile spike, average recoveries ranged from 75 to 91 percent for all of the compounds except for heptachlor
epoxide/fluoranthene, DDE, ODD, and endosulfan sulfate. These four compounds elute at the end of the chromatogram
where the column temperature program is increasing and the column liquid stationary phase begins to bleed into the
detector. The rising baseline from the bleeding stationary phase interferes with the integration and quantitation of these
For the low (< 1 mg) semivolatile spike, average recoveries range from 70 to 103 percent for all of the
compounds except phenol/2-chlorophenol, hexachloroethane, naphthalene, 2,6-dichlorophenol, 1-naphthylamine/2-4-
dinitrotoluene, heptachlor epoxide/fluoranthene, and DDE. Several of these compounds, such as phenol/2-chlorophenol
and 2,6-dichlorophenol, are polar. Supercritical CO2 sometimes does not extract polar materials as efficiently as it
extracts nonpolar materials. Other compounds, such as hexachloroethane, have a high instrument detection limit (IDL)
when using a flame ionization detector.
In Table 2, the total variance is separated among three factors: spiking level, extraction replicate, and analytical
replicate. For ODD, the variance in the analytical replicate contributes 19% of the total variance. A high contribution to
the variance from the analysis was expected from this compound because of the difficulties in integrating the peak
reproducibly due to the column bleed from the temperature program and MeOH solvent used. Except for dibenzofuran,
and phenol/2-chlorophenol, the major contributor to the total variance is the variance from the replicate extractions.
CONCLUSIONS
The study demonstrated that large volumes of XAD-2® could be extracted for volatile and semivolatile organic
compounds by SFE. SFE resulted in quantitative recoveries for 17 out of 32 of the test compounds spiked onto 30
grams of XAD-28. The average recoveries for these 17 compounds ranged from 70 to 103 percent. Recoveries for DDE
(1.566 mg), ODD (2.090 mg), endosulfan sulfate (1.657 mg), 1,1,2,2-tetrachloroethane (2.560 mg), hexachloroethane
(0.168 mg), and 2,6-dichlorophenol (0.170 mg) spiked on 30 g of XAD-2® were less than 70 percent. Recoveries for the
following coeluting compound pairs spiked on 30 g of XAD-2® were less than 70 percent: benzene (0.173)/1,2-
dichloroethane (0.259 mg); heptachlor epoxide (1.550)/fluoranthene (1.794 mg); phenol (0.244)/2-chlorophenol (0.403
mg); 1-naphthylamine (0.154)/2,4-dinitrotoluene (0.139 mg); and heptachlor epoxide (0.163)/fluoranthene (0.189 mg).
Recoveries of hexachloroethane and chlorobenzene/1,1,2-trichloroethane from 30 grams of XAD-2® were comparable at
spike levels of 1 to 3 mg between supercritical fluid and Soxhlet extraction. Recovery for 0.145 mg of naphthalene
spiked on 30 g of XAD-2® was greater than 130 percent because the background contamination of naphthalene in the
XAD-2® was significant at this spiking level (11.8% of the spiked amount or 0.017 mg of naphthalene was detected in the
XAD-28 blank). Recovery of 0.165 mg of DDE spiked on 30 g of XAD-28 was greater than 130 percent because of
challenges in accurately integrating the DDE peak due to the column bleed.
411
-------
Table 2. Determination of Source of Total Variance for Representative Compounds
Spiking Level
Compound
%of
Total
Variance
Prob
of
> F
Extraction
%of
Total
Variance
Prob
of
> F
Analytical
%of
Total
Variance
Prob
of
> F
Semivolatile
Naphthalene
Dibenzofuran
ODD
Phenol/2-Chlorophenol
Volatile
25.6378
51.1728
16.7624
87.3387
0.0001 *
0.0001 *
0.0003 *
0.0001 *
73.8342
40.7863
64.3577
10.1826
0.0001 *
0.0001 *
0.0001 *
0.0001 *
0.5279
8.0409
18.8799
2.4787
-
-
...
...
Xylene
Benzene/1,2-DCEA
1,2-DCPA/DBM
0.0000 0.0001 * 96.7069 0.0001 * 3.2931
32.2888 0.0001 * 60.298 0.0001 * 7.4133
3.489 0.0001 * 95.3676 0.0001 * 1.1434
* Highly significant at the 99% confidence level
Several additional objectives can be pursued including optimization of SFE conditions for extracting stationary
source samples collected on resins, design of a sampling cartridge that is amenable to SFE sample recovery and
stationary source sampling, evaluation of SFE methods on actual source samples collected at an air toxic emissions or
similar site, evaluation of the use of SFE to purify source sampling resins prior to field sampling, extension of current
source sampling methods for collection of larger less volatile compounds and possibly thermally labile compounds by
using supercritical fluid chromatography mass spectrometry (SFC/MS) to separate the analytes prior to detection and to
solve coelution problems, and additional evaluation of SFE for extracting source samples from Tenax9 as used In
Method 0030.5
REFERENCES
1. Hawthorne, S.B. and Miller, D.J. Extraction and Recovery of Organic Pollutants from Environmental Solids and
Tenax-GC Using Supercritical C02. J. Chromatogr. Sci. 24: 258-24, 1986.
2. Wright, B.W., Wright, C.W., Gale, R.W., and Smith, R.D. Analytical Supercritical Fluid Extraction of Adsorbent
Materials. Anal. Chem. 59: 38-44, 1987.
3. Hawthorne, S.B. Supercritical Fluid Extraction. Anal. Chem. 62: 633A-642A, 1990.
4. Alexandrou, N., Lawrence, M.J., and Pawliszyn, J. Cleanup of Complex Organic Mixtures Using Supercritical
Fluids and Selective Adsorbents. Anal. Chem. 64: 301-311,1992.
412
-------
5. Test Methods for Evaluating Solid Waste, Third Edition. Report No. SW-846. U.S. Environmental Protection
Agency, Office of Solid Waste and Emergency Response. Washington, DC: 1986.
6. Lentzen, D.E., Wagoner, D.E., Estes, E.D., and Gutknecht, W.F. IERL-RTP Procedures Manual: Level 1
Environmental Assessment (Second Edition). Report No. EPA-600/7-78-201. U.S. Environmental Protection
Agency, Office of Research and Development. Washington, DC: October 1978.
Although the research described in this article has been fully funded by the U.S. EPA under contract 68-D1-0010
to Radian, it has not been subjected to Agency review and therefore does not necessarily reflect the views of the
Agency, and no official endorsement should be inferred. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
413
-------
AN INVESTIGATION OF PROCESS MASS SPECTROSCOPY
AS A CONTINUOUS EMISSION ANALYZER FOR STATIONARY SOURCES
Laura L. Kinner Ph.D.
Grant M. Plummer Ph.D.
Entropy Environmentalists, Inc.
Research Triangle Park, North Carolina 27612
ABSTRACT
In January 1993, Entropy Environmentalists, Inc. (Entropy) conducted a feasibility study
employing a process mass spectrometer to measure hazardous air pollutant emissions from
stationary sources. Testing was conducted at a small coal-fired boiler in conjunction with a
Fourier transform infrared spectroscopy (FTIR) field validation test, and was supported by the
Emission Measurement Branch (U.S. EPA/OAQPS). Mass spectroscopy was evaluated at the
simplest level, without separation of the constituent gases and without specialized software.
Presented in this paper are the results of this study and recommendations for increasing the
effectiveness this methodology.
INTRODUCTION
Title III of the Clean Air Act Amendments (CAAA) defines 189 hazardous air pollutants
(HAP's) which industrial facilities are required to characterize, quantify and ultimately reduce.
Due to the large number of compounds, and the time constraints for emissions compliance,
sampling and analytical techniques which provide results in a timely fashion are in great need. In
the past, conventional sampling methods have been employed. These methods often require large
testing crews, and are often compound or functional group specific. In addition, these methods
are time consuming, sometimes having two to three week turn around times for analytical results.
For these reasons, instrumental methods having the capability to both identify and quantify
multicomponent HAP mixtures at near real time conditions are being examined. Mass
spectrometry (MS) is one such instrumental technique that has the qualifications for continuous
emission monitoring from stationary sources.
The purpose of this study was to examine the following aspects of direct MS monitoring of
HAP emissions:
• Effectiveness in gathering qualitative data.
• Limitations imposed by large non-HAP effluent constituents.
• Effectiveness of "process" MS software in interpreting mass spectra of multicomponent
mixtures.
• Calibration techniques.
• Quantitative accuracy.
414
-------
EXPERIMENTAL
Process Description and Sample Location
The emissions testing program was performed at a small coal-fired steam generation plant
with an approximate heat input of 68 Mbtu/hr. At this plant, bituminous coal is pulverized and
blown into the combustion chamber by fans that provide simultaneous suspension and combustion
of the coal. The combustion gases and paniculate matter exiting the boiler pass through an air
preheater, baghouse, and an induced draft fan before being exhausted to the atmosphere via a
stack. Sampling was performed at the baghouse outlet. The mobile laboratory housing the mass
spectrometer and the FTIR spectrometer was located within 50 feet of the outlet location.
Sampling System
An extractive gas sampling system was employed to provide sample to the analytical
instrumentation. This system employed a Type 316 stainless steel probe, a Balston® filter, a
Teflon® line and a pump, all heated to approximately 250°F. The sampling system transports the
flue gas sample to the gas conditioning equipment and MS/FTIR systems located in Entropy's
mobile laboratory. Velocity traverses were conducted to find the point in the duct of average flow
rate. These were conducted at the same time as the moisture determinations. All sampling was
conducted at a single point and was non-isokinetic. In addition to the MS/FUR instrumentation,
continuous emission monitors (CEMs) were employed on-site to verify the carbon monoxide
(CO), carbon dioxide (CO2), and oxygen (O2) concentrations.
Two different types of samples were examined with the MS:
• Hot/wet or unconditioned sample, which is considered more likely than conditioned
samples to represent the actual flue gas composition. Only paniculate matter is
intentionally removed from the gas in this type of sample.
• Condenser sample, which a standard Peltier dryer system was used to cool the gas
stream to approximately 3°C. The resulting condensate was removed from the dryer
with peristaltic pumps. This conditioning technique is known to leave the
concentrations of inorganic and highly volatile compounds very near the (dry-basis)
stack concentrations.
A heated distribution manifold within the mobile laboratory was used to monitor and control
sample gas flow to the MS/FTIR instrumentation. Mass spectra were taken in conjunction with
FTIR spectra from both hot/wet and condenser flue gas samples.
Analytical System
The MS system employed for these tests was a Leybold Inficon, 200 atomic mass unit
(AMU), Transpector H200M Quadrupole Mass Spectrometer System. The unit consists of five
main parts: an ion source, a quadrupole mass filter, an ion detector, a vacuum pump, and a data
acquisition system. Such instruments function by ionizing the molecules in the sample and
injecting them onto the axis of a combined radio-frequency/constant quadrupole electric field. As
the quadrupole field is ramped and modulated, ion fragments of only a particular charge-to-mass
ratio maintain stable trajectories as they pass through the quadrupole field, where they are
detected on-axis.
415
-------
This particular instrument is most commonly used in residual gas analysis, where only a few
compounds exist in the gas mixture. For this reason, the data acquisition software typically
provides only minimal spectral manipulation capabilities. The Transpector was chosen for this
testing since its small size and ruggedness allowed for table mounting in Entropy's mobile
laboratory.
The Transpector MS system was positioned in Entropy's mobile laboratory at a point directly
after the gas conditioning manifold. The MS system was maintained at a temperature just above
100 °C to prevent sample condensation; higher temperatures could not be maintained without
damaging certain electronics components. Sample was drawn through a tee in the FUR
absorption cell sample delivery line, a variable leak-rate valve (needle valve), and into the MS
system by a combination of rough and turbo pumps (pumping speed approximately 40 L/sec).
Figure 1 illustrates the sample gas delivery schematic from the source to the instrumentation.
Dynamic Spiking
Cylinders containing up to five HAP's at concentrations of approximately 50 parts per million
(ppm) diluted in nitrogen were prepared for Entropy by Scott Specialty Gases for this test.
These "spike gases" were introduced to the sampling system through a zero to five liter-per-minute
(1pm) mass flow meter. The heated outlet of the flow meter was connected to the sampling
system via a tee located at the inlet of a 1 micron Balston paniculate filter (initial paniculate
filter depicted in Figure 1), which was installed at the outlet of the probe. A 100-foot length of
3/8-inch (O.D) Teflon sample line connected the filter to the sample pump located within the
mobile laboratory. All exposed unions were wrapped with heat tape and insulation to prevent
condensation of the gases. A calibrated orifice located at the outlet of the secondary paniculate
filter was used to measure the total flow of the sample at the distribution manifold. The five-to-
one spike dilution factor was maintained by controlling the spike and total sample flow rates.
416
-------
Heated
Transport
Lines
r
C02
Analyzer
1
O2 CO
Analyzer Analyzer
Spike
Cylinder Gas
Figure 1. Simplified Schematic of the MS/FTIR Direct Gas Phase Sample Delivery System.
-------
Mass Spectral Collection
At the beginning of each day of the field test, the MS was manually "tuned" with
perfluorotributylamine [CF3(CF2)3]3N (PFTBA). This volatile compound was loaded into a small
valved ampule, which was permanently mounted at the sample inlet. These daily adjustments of
the instrument were made to obtain characteristic mass fragments from the PFTBA over a large
mass range, ensuring relatively constant mass fragmentation patterns for other sample molecules.
Background spectra were collected on all testing days by evacuating the system through the heated
sample distribution manifold. These spectra were examined for evidence of contaminants.
Mass spectra of gases in the FTIR sample cell were collected as follows. At a sample
delivery pressure of one atmosphere, the total pressure in the MS system was adjusted to between
IxlO'5 to 3xlO'5 Torr by means of the variable leak valve. The 20 to 200 AMU range was then
scanned; 20 data points per AMU at a nominal resolution of 1 AMU were collected. The scan
time was 2.8 seconds (0.8 milliseconds per data point).
At the completion of the scan, peak areas for each 1 AMU segment were calculated and
stored to disk. The software allowed such data files to be generated either as single spectral
representations, or as multiple sub-files of spectra collected consecutively over a specified time
period. In the latter case, which was selected for this testing, data were automatically generated
at one minute intervals.
The indicated MS system pressure was observed to vary from spectrum to spectrum. Several
possible causes exist for this instability, including the fact that the sample inlet pressure varied as
the FTIR absorption cell was alternately filled with sample and then evacuated. However, it is
also possible that surface effects of water and other major effluent constituents, heightened by the
relatively low temperature of the quadrupole structure, played some role in the observed pressure
fluctuations.
During the first few days of the test, the instrument was accidentally vented several times,
requiring replacement of the filament. Due to this problem, data for most of the hot/wet
sampling is unavailable. An "uninterruptable" power source was installed to power the
instrument's vacuum pumps, which prevented recurrence of this problem.
Data Analysis - Qualitative
The mass spectra recorded with the instrument were extremely complex, particularly at
masses less than 50 AMU. The original software package supplied with the instrument was found
inadequate for performing the spectral subtractions and other manipulations required for handling
such spectra. Therefore, spectral files generated in the field were converted to a format
compatible with a more sophisticated software package available from the same manufacturer.
Due to the complexity of the spectra, initial data examinations focussed on identification of
the spiked compounds present in the sample. Only those compounds exhibiting strong ion peaks
higher than 50 AMU were investigated, since the majority of large peaks belonging to the stack
effluent constituents fall at masses below this level, and tend to interfere strongly with the low
mass spike compound fragments. Additional interferences between the spike compounds
themselves were also encountered, even at the higher AMU values. In order to quantify these
compounds, fragmentation ions unique to individual compounds in the spike cylinders were
sought. Some of the mass fragments were typically not the strongest in a compound's mass
spectrum, and often represented a very small peak near the detection limit of the instrument.
Spectra of the pure spike materials were compared to single-component spectra, obtained from
the National Institute of Standards and Technology, in order to identify the parent molecular
species.
418
-------
Data Analysis - Quantitative
For those compounds identified qualitatively, approximate quantification of gas phase
concentrations were performed as follows. Reference spectra for pure compounds were recorded
from existing 50 ppm gas standards prepared by Scott Specialty Gases. As mentioned above, the
MS system total pressure was not constant for all the sample spectra. To account for possible
effects of these variations, the reference spectra were recorded at the same MS system pressure as
that observed when the sample spectrum of interest was recorded. A response factor was
generated for a selected ion peak of each compound by dividing the observed reference ion peak
area by the compound's reference concentration. Concentrations of the compounds in spiked
samples were then calculated by dividing the observed sample ion peak areas by the response
factors.
RESULTS
The following Tables compare sample concentrations derived from the mass spectrometer
data to those obtained from spike/sample flow ratios. Only compounds which could be quantified
are reported. Due to the complexity of the results from FTIR validation testing, FTIR data are
not presented. However, those compounds marked with an asterisk denote compounds which
were not validated for the FTTR during the course of this test program. The data presented are
derived mainly from the condenser samples, which were generally recorded over longer time
intervals, and at lower water concentrations than the corresponding hot/wet data. Considering
the degree of spectral interference between spikes and stack gas, the results for these compounds
are in fair agreement with the expected concentrations.
419
-------
TABLE 1.
CYLINDER NO. 2 RESULTS
CONDENSER
Sample
GCSfl
Compound
Hexane
Melhylene Chloride
p-Xylene
Mass Spectrometer
Pressure
(10*Torr)
24
24
24
Concentration (ppm)
11.0
4.5
11.0
Expected
Concentration
9.9
10.0
10.0
TABLE 2.
CYLINDER NO. 4 RESULTS
CONDENSER
Sample
GCS04
Compound
Isooctane
Carbon Disulfide
m-Xylene
Mass Spectrometer
Pressure
(lO" Torr)
17
17
17
Concentration (ppm)
9.0
7.0
1.0
Expected
Concentration (ppm)
10.0
9.7
10.0
TABLES.
CYLINDER NO. 7 RESULTS
CONDENSER
Sample
GCS07
Compound
Acrylonitrile*
Allyl Chloride
Ethyl Benzene
Mass Spectrometer
Pressure
(lO-'Torr)
10
10
10
Concentration (ppm)
14.0
2.0
9.0
Expected
Concentration (ppm)
10.0
10.2
10.0
* Denotes compounds not meeting EPA Method 301 criteria for FTIR validation
420
-------
TABLE 4.
CYLINDER NO. 9 RESULTS
CONDENSER
Sample
GCSC9
Compound
Methyl Tertian Butyl
Ether (MTBE)
Trichloromelhane*
(Chloroform)
Cumeoe
Mass
Pressure
(lO'Torr)
12
12
12
Spectrometer
Concentration
4.5
9.0
4.5
Expected
Concentration (ppm)
9.6
9.7
10.0
TABLES.
CYLINDER NO. 11 RESULTS
CONDENSER
Sample
GCS*1I
Compound
Slyrene
1 ,2-Dichloropropane
Trichloroelhylene
Mass Spectrometer
Pressure
(lO-'Torr)
14
14
14
Concentration (ppm)
3.0
9.0
5.5
Expected
Concentration (ppm)
9.6
8.2
9.5
TABLE 6.
HOT/WET
AVERAGE CONCENTRATION
Sample
GHSJI1
Compound
Styrene
1 ,2-Dichloropropane
Trichloroethylenc
Mass spectrometer
Pressure
(10^ Tor()
14
14
14
Concentration (ppm)
15.0
6.0
6.5
Expected
Concentration (ppm)
10.8
10.4
9.6
* Denotes compounds not meeting EPA Method 301 criteria for FTIR validation
421
-------
CONCLUSIONS
In this study, simple mass spectrometry was found to provide useful quantitative and
qualitative concentration data, even in a complicated combustion matrix. The main interfering
compounds are (ion masses in parentheses) H2O (18), CO, N2 (28), NO (30), O2 (32), CO2 (44),
and SO2 (64). These ion peaks dominate the mass spectra, along with their doubly and triply
ionized species. However, the spectral interferences from these compounds do not present an
insurmountable problem, since they appear primarily below 50 AMU. Spectra of particular HAP
compounds, introduced as spikes, were clearly detectable at higher mass levels, despite spectral
overlap from the stack matrix.
The applicability of the technique in HAP emission measurements applications could be
enhanced by a number of possible developments, including the following:
• Provisions in the system's software for handling spectral interferences from high-
concentration sample constituents.
• Increase of the quadrupole structure's temperature (100°C in this instance), to combat
thermal drifts, which were sometimes noticed in high mass peak positions and
intensities, and to lower background PFTBA and H2O levels.
• Provisions for stable sample pressure.
• • Development of accurate quantitative reference spectra for target compounds for
typical operating conditions.
Mass spectrometry could also prove to be extremely powerful in conjunction with other
analytical methods. For instance, MS data could facilitate infrared analyses of heavy compounds,
which possess relatively broad and uncharacteristic infrared absorption bands, but which may
exhibit strong ion peaks in the relatively sparse high mass range.
422
-------
GAS PHASE FTIR SPECTROMETRY AS A METHOD OF MEASURING
HAZARDOUS AIR POLLUTANT EMISSIONS
Thomas J. Geyer, Grant M. Plummer, Thomas A. Dunder,
Scott A. Shanklin, Lisa M. Grosshandler, Patricia Royals, Greg C. Blancshan,
Rick Staughsbaugh, and Mike Worthy
Entropy Environmentalists, Inc.
P. 0. Box 12291
Research Triangle Park, North Carolina 27709
ABSTRACT
Entropy Environmentalists, Inc., (Entropy) conducted a field test using FTIR spectrometry to measure
hazardous air pollutants (HAPs) emitted from a coal-fired boiler facility. Two techniques were employed: (1)
introduction of extracted gases directly to the FTIR absorption cell and (2) sample concentration using Tenax
followed by thermal desorption into the cell. Spiking experiments with HAP compounds were performed using
each technique. Spectra were analyzed to measure spike recovery rates and to determine reproducibility of
results.
INTRODUCTION
The test described in this paper is part of the Environmental Protection Agency's (EPA) FTIR Method
Development project. The purpose of the project is to develop a method to detect and quantify compounds
listed in Title III of the Clean Air Act Amendments (CAM). Prior to this test, Entropy developed a library of
reference spectra, and performed screening tests at several industrial facilities representing a variety of source
categories, including a coal-fired boiler.
The screening tests of the direct gas phase technique provided data on the performance and suitability of
FTIR spectrometry for measuring the emission rates of HAPs. These tests helped determine sampling and
analytical limitations, provided qualitative information on the emission stream composition, and allowed
estimation of the mass emission rates for a number of HAPs and process locations.
At a coal-fired boiler, the usefulness of direct gas phase extraction FTIR for measuring HAP compounds at
concentrations corresponding to 10 tons of emissions per year was found to be limited. This is because of the
rather large effluent flow rates typical of such boilers; even small concentrations (i.e., sub ppm levels) of HAPs
in the effluent output of the boiler can lead to potentially large yearly mass emissions. It was, therefore,
necessary to develop a sample concentration technique before quantification of most HAPs by FTIR
spectrometry could be attempted at such facilities. The screening test at the utility boiler also provided data on
the interferant species and criteria pollutants that one would expect to encounter in a gas stream emitted from
such a source. This information enabled Entropy to prepare for analyzing data before testing commenced.
The testing described in this paper represents a continuation of the method development project. The
analyte spiking procedures of EPA Method 301 served as the model for experiments performed with 47 gas
phase HAPs. Separate procedures were performed to test the direct extractive gas phase and the sample
concentration/thermal desorption techniques. The goal was to determine the bias, precision, and range of the
techniques. The data were analyzed to examine the suitability of the FTIR methods for testing of emissions
from this type of source.
EXPERIMENTAL
Process Description and Sample Point Location
Testing was performed at a bituminous coal-fired steam generating plant with an approximate heat input of
68 MBtu/hr. Bituminous coal is pulverized and blown into the combustion chamber by fans, providing
simultaneous fuel suspension and combustion. The combustion gases and paniculate exiting the boiler, pass
through an air preheater, baghouse, and an induced draft (I.D.) fan before being exhausted to the atmosphere
via a stack.
423
-------
Measurements during this test were made at the outlet location of the baghouse. Direct gas extraction
testing employed 100 feet of heated sample line to connect the sampling probe to the heated manifold In the
FTIR truck. Concentrated samples were obtained through a port adjacent to where the direct extractive probe
was inserted.
Emission Sampling System
The FTIR-based method uses two different sampling techniques: (1) direct gas phase extraction and (2)
sample concentration followed by thermal desorption. The direct gas phase sampling system extracts gas
from the sample point and transports the sample to a mobile laboratory where sample conditioning and FTIR
gas phase analyses are performed. The sample concentration system employs 10 grams of Tenax sorbent,
which can trap some organic compounds from a large volume (typically 300 liters) of sample gas. These
compounds are thermally desorbed into the smaller FTIR absorption cell volume (approximately 7 liters),
providing a volumetric concentration which allows detection of some compounds below 1 ppm in the original
sample. Components of the emission test systems prepared by Entropy for this testing are described below.
Direct Gas Extraction Gas was extracted through an 8-foot heated stainless steel probe. A heated
Balston paniculate filter rated at 1 micron was installed at the outlet of the sample probe. A 100-foot section of
heated 3/8-inch O.D. Teflon8 sample line connected the probe to the heated sample pump located inside the
mobile laboratory. The temperature of the sampling system components was maintained at approximately
150°C. All components were constructed of Type 316 stainless steel, or Teflon®. Digital temperature
controllers were used to monitor the temperature of the transport lines. All points of connection were wrapped
with electric heat tape and insulated to ensure that there were no "cold spots" in the sampling system where
condensation might occur. The sample pump provided approximately 16 L/min of sample gas flow. The
heated sample flow manifold, located in the FTIR truck, included a secondary paniculate filter and valves that
allowed the operator to send the sample gas directly to the absorption cell or through one of the gas
conditioning systems.
The extracted gas sample was treated in one of two ways. Sample sent directly to the FTIR cell was
considered unconditioned, or "hot/wet." This sample is thought to be more representative of the effluent
composition than conditioned samples. The gas stream could also be directed through a condenser to remove
most of the water. The condenser employed a standard Peltier dryer to cool the gas stream to approximately
3°C. The resulting condensate was collected in two traps and removed from the conditioning system with
peristaltic pumps. This technique is known to leave the concentrations of inorganic and highly volatile
compounds very near to the stack concentrations (dry-basis). The condenser was tested in an effort to
ascertain which HAPs can be reliably quantified using this system because the spectra are easier to interpret if
water is removed.
Sample Concentration Four gas samples were collected simultaneously during each run using a quad-
train assembly. Ten ft3 of flue gas flowed through the Tenax cartridge over a sampling time of about 75
minutes.
Components of a single sampling train included a heated stainless steel probe, heated filter and glass
casing, heated teflon connecting line, a stainless steel adsorbent trap in an ice bath, followed by two water-
filled impingers, a knockout impinger, an impinger filled with silica gel, a sample pump, and a dry gas meter.
All heated components were kept at a temperature above 120°C to ensure no condensation of water vapor
within the system. The trap was a stainless steel U-shaped collection tube filled with 10 grams of Tenax
sorbent and plugged at both ends with glass wool. Stainless steel was used for the construction of the
adsorbent tubes because it gives a more uniform and more efficient heat transfer than glass.
Before use, the Tenax was precleaned to remove impurities that might desorb with any sample
compounds collected. The packed tube was heated to 350°C while being treated with preheated nitrogen at 1
to 2 Ipm. The heating and nitrogen flow were maintained for up to 18 hours. Cleaning the desorption tubes
resulted in a decrease in impurity bands that Entropy has observed in spectra from the desorption of new,
commercially precleaned Tenax.
After each sample collection, condensed water was removed before spectral analysis by blowing dry
nitrogen through the Tenax while the sample tube was immersed in an ice bath. The collection tubes were
then analyzed by thermal desorption-FTIR. The tubes were wrapped with heat tape and placed in a heating
chamber. One end of the sample tube was connected to the inlet of the evacuated FTIR absorption cell. Gas
samples were desorbed from the Tenax by heating to 250°C. A preheated stream of UPC grade nitrogen gas
was passed through the adsorbent and into the FTIR absorption cell. Approximately 6 liters of nitrogen were
424
-------
used to carry desorbed gases to the cell and to bring the total pressure of the FTIR sample to atmospheric
pressure. The infrared absorption spectrum was then recorded. The purging process was repeated to verify
that no evidence of additional sample desorption was noted in the Infrared spectrum.
Analytical System
The FTIR equipment employed in this test consisted of a medium-resolution interferometer, a heated
infrared absorption cell, a liquid nitrogen cooled mercury cadmium telluride (MCT) broad band infrared
detector, and a computer. The interferometer, detector, and computer were purchased from KVB/Analect,
Inc., and comprise their base Model RFX-40 system. The nominal spectral resolution of the system is one
wavenumber (1 cm"1). Sample was contained in a variable path white cell, model 5-22 H, manufactured by
Infrared Analysis, Inc. Heated jackets and temperature controllers were used to maintain a cell temperature of
240°F. The absorption path length was externally adjustable from about 2.2 to 24 meters. For the test, the
path length was kept at 22 meters.
Dynamic Spiking
Cylinders containing up to five HAPs at concentrations of approximately 50 (ppm) were prepared for
Entropy by Scott Specialty Gases. The spike gases were introduced through the direct extractive gas sampling
system using a mass flow meter capable of measuring flows in the range of zero to five liters per minute (Ipm).
Spike flow ratios were verified in two ways. First, CTS gas (100 ppm ethylene in nitrogen) was introduced
through the mass flow meter and spike line to the gas sampling system. This was done prior to insertion of
the probe in the sample port so that the CTS spike was mixed with ambient air. Analysis files, which had been
previously prepared, were used to determine the concentration of ethylene in the resulting sample. Another
method of verification used the SF6 tracer contained in each spike cylinder. Any spectrum of a spiked sample
could be analyzed for the concentration of SF6, which should have been approximately 0.2 ppm after dilution.
For testing with the sample concentration system, spike gas was directed through a mass flow meter. The
measured flow of gas was passed through a heated coil to heat the spike gas and then introduced to the
sample system at the inlet to the Method 5 paniculate filter. The combined stream travelled through the cooled
collection tube where compounds could be adsorbed onto the Tenax. The total gas flow was measured with
the dry gas meter. Spiking rates were chosen to provide gas concentrations of approximately 500 ppb in the
stack and about 20 ppm in the FTIR cell.
Testing Procedures
For gas phase testing, a complete data set for a cylinder mixture included 24 samples, half spiked and
half unspiked. The samples were collected consecutively, at about five minute intervals. Complete data sets of
untreated, "hot/wet," samples were collected for all 11 cylinder mixtures. Complete data sets of samples
treated with the condenser were collected for some cylinder mixtures and partial data sets of condenser
treated samples were collected for the remaining cylinders. Within a set of 24 samples, data were collected in
groups of four unspiked, followed by four spiked samples. Alternate groups of four samples were extracted
until 24 spectra were obtained.
Single runs, with the sample concentration system, consisted of sampling 10 ft3 of gas simultaneously
through the four Tenax cartridges of the quad-train assembly. All four probes were inserted in the same
sampling port. The test runs were performed during the day and charged tubes were desorbed during the
evening.
Calculation of Expected Spike Concentrations
Analytical results for samples containing a known amount of spiked material were compared to results for
unspiked samples. It should be stressed that no spectroscopic evidence of the spike compounds was found in
the unspiked samples.
Expected concentrations of HAP spike materials in gas phase samples were calculated on the basis of
flow rates measured during the field test. These flow rates were constantly maintained during the performance
of the field testing for each HAP spiking group, and were verified to within 10% by spectroscopic analysis for
the 1 ppm SF6 tracer gas component in each cylinder.
The value for the expected concentration from the gas phase sampling is given by:
425
-------
c ( FlOWspike
eXP ( Flowtotal
where:
Cexp is the expected gas spike concentration.
Flowspll
-------
Reference spectra for the K-matrix concentration determinations were de-resolved to 1.0 cm"1 resolution
from existing 0.25 cm"1 resolution reference spectra. This was accomplished by truncating and re-apodlzlng
the Interferograms of single beam reference spectra and the corresponding background interferograms. The
processed single beam spectra were recombined and converted to absorbance.
Preparation of Analysis Programs
To provide accurate quantitative results, K-matrix Input must include absorbance values from a set of
reference spectra which, added together, qualitatively resemble the appearance of the sample spectra. For this
reason, all of the Multicomp analysis files included spectra representing interferant species and criteria
pollutants present in the flue gas in addition to de-resolved reference spectra of HAPs used for analyte spiking
experiments.
Prior to the actual field test, synthetic spectra were prepared using sample spectra that had been obtained
during previous testing at a coal-fired boiler. K-matrix programs were then constructed which could adequately
analyze the synthetic spectra. These analysis programs were found to serve as a useful starting point; all the
finalized Multicomp routines are based on the programs prepared using the synthetic spectra.
Preparation of the synthetic spectra proceeded In the following stages. First, Entropy obtained a 1 cm"1
reference spectrum of each cylinder mixture to be used in the analyte spiking experiments. Second, the
cylinder spectra were scaled by a factor of 0.2 to simulate the anticipated dilution. Finally, a computer
generated synthetic spectrum was created by adding a scaled cylinder spectrum to each of several sample
spectra from the coal-fired boiler. This resulted in a set of synthetic spectra (for both "hot/wet" and condenser
samples) representing simulated spiked samples in a stack matrix similar to what would be encountered during
the spiking experiments.
RESULTS AND DISCUSSION
The spectral analysis programs were applied to all sample spectra. All spectra were visually compared to
spectra of the spike cylinder gases to ensure that the resulting concentrations were physically reasonable and
that no obvious spectral interierants had been omitted from the analytical programs. Statistical analysis of the
data was carried out and compounds were evaluated based on the reproducibility of calculated spiked and
unsplked results. A preliminary draft report summarizing the test results has been submitted to EPA, who will
determine which compounds used in spiking experiments can be measured with the sampling techniques
described in this paper.
Spectra and analytical programs Involving compounds for which results were not reproducible were
subject to further scrutiny. In many cases, Improvements in the analytical results were achieved by adjustment
of the spectral region and/or baseline subtraction technique employed in the Multicomp program. Following
these adjustments, the statistical analysis was repeated. Correction of the programs often gave improved
results. Table 1 presents calculated concentrations for ethylene dibromlde, a compound that gave
reproducible results, in spiked and unspiked samples. One data set is for untreated samples and the other
data set is for samples treated with the condenser system.
Several generalizations may be made concerning the difficulty In measuring some species. A small
number of compounds are simply not observable In the spectra of the direct gas samples delivered to the FTIR
Instrumentation. Examples of such compounds are ethylene oxide and propylene oxide; for these compounds,
the observed spiked and unspiked concentrations differed only slightly. This may be due to heavy water
spectral interference, losses in the sampling system, or a combination of both effects. It Is also clear from
visual inspection of the concentrated sample spectra that a large fraction of the spike compounds simply are
not efficiently delivered to the FTIR sample cell. Compounds with low boiling points (< 50°C) are probably not
adsorbed on Tenax.
There is evidence that a few compounds were consistently delivered by the gas phase sampling system,
but the analytical results give consistently low concentrations. There are several possible systematic errors
which could lead to such results, including (1) errors in the reference spectrum gas concentration, spike
cylinder gas concentration, or both; (2) band intensity mismatch between reference spectra and sample
spectra, caused by Instrumental distortion or gas temperature mismatch between reference and sample
spectra; and (3) a consistent loss of a certain fraction of the spike concentration in the sampling system.
Several compounds failed to give reproducible results because of spectral interference with particular
spike or gas matrix species. Other compounds are different, in that visual Inspection of the spectra indicates
that the K-matrix analyses should successfully remove the interferences, but do not. The reasons for these
failures are not fully understood, and are still being sought.
427
-------
TABLE 1.
Preliminary Results: Calculated Concentrations (in ppm) of Ethylene Dibromide for Spiked and
Unspiked Untreated Samples and Samples Treated with the Condenser System.
Condenser
SPIKED' UNSPIKED
9.05
8.90
9.15
9.14
7.25
7.47
7.42
7.50
7.08
7.30
7.49
7.6
7.94
-1.14
-1.05
-0.99
-1.31
-1.45
-1.48
-1.47
-1.47
-1.46
-1.52
-1.54
-1.55
-1.37
Unconditioned Samples
SPIKED UNSPIKED
6.52
6.52
6.51
6.53
6.84
6.91
6.84
6.77
6.58
6.52
6.48
6.44
6.62
-3.27
-3.69
-3.57
-3.57
-3.45
-3.73
-3.88
-3.64
-3.30
-3.40
-3.45
-3.83
-3.56
" Expected concentration (spiked minus unspiked) is 10 ppm.
CONCLUSIONS
The FTIR spectrometric analytical procedures described in this report have been applied to samples
extracted from the flue gas stream of a coal-fired boiler. The results gave information which can be used to
determine the utility of FTIR spectrometry for detecting and quantifying HAPs in a flue gas stream emitted from
a coal-fired steam generation plant.
A total of 47 gas phase HAPs were introduced through the sampling systems and FTIR spectra were
obtained for spiked and unspiked samples. The gas phase extractive system demonstrated utility for
measuring a number of compounds in both "hot/wet" and condenser samples. The large amount of
information obtained has been used to improve analysis procedures and to characterize limitations of each
sampling technique for measuring particular compounds. The direct gas extraction and sample concentration
techniques, when used in conjunction with FTIR spectrometry, provide versatile means of detecting and
quantifying a large number of compounds from about 10 ppm to 500 ppb. Longer sampling times using the
sample concentration technique, can easily provide improved detection limits for species that adhere to Tenax.
The gas extraction technique can provide detection limits of close to 1 ppm, for many compounds, using the
sampling system configuration described in this paper.
REFERENCES
1. G. L McClure (ed.), "Computer-Assisted Quantitative Infrared Spectroscopy," ASTM Special Publication
934(ASTM), 1987.
2. D. M. Haaland, R. G. Easterling, and D. A. Volpicka, "Multivariate Least-Squares Methods Applied to the
Quantitative Spectral Analysis of Multicomponent Mixtures." Applied Spectroscopy. 39(10), 73-84, 1985.
428
-------
FIELD OBSERVATIONS OF COMPLIANCE MONITORING
FOR ETHYLENE OXIDE EMISSIONS FROM HOSPITAL STERILIZERS
by
KEVIN MONGAR
Monitoring and Laboratory Division
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812
ABSTRACT
California Air Resources Board (CARB) Method 431, "Determination of Ethylene Oxide
Emissions from Stationary Sources" specifies an on-site (direct) gas chromatographic
procedure. This method has been applied and evaluated for compliance testing of hospital
sterilizers with catalytic oxidation-, hydrolytic scrubber- and recovery-type control units. In
several cases the control equipment required modification to accommodate the compliance
tests. In several other cases the test method was modified in order to provide useful data.
Proposed modifications to the test method will also be discussed including a Tedlar bag
sampling procedure and an "inlet estimation" technique.
INTRODUCTION
The work presented in this paper is a continuation of work presented at the 1992
EPA/A&WMA International Symposium on "Measurement of Toxic & Related Air Pollutants"1
An "Ethylene Oxide Airborne Toxic Control Measure (ATCM) for Sterilizers and
Aerators" was adopted by CARB (17 CCR, Section 93108) on May 22, 1991. The ATCM
requires control of ethylene oxide (EtO) emissions based on annual usage. The "control
efficiency" is defined as the ethylene oxide mass or concentration reduction efficiency across
a control device, as determined by CARB Test Method 431. The control efficiency is
expressed as a percentage calculated across the control device using equation 1.
EEtO in - ZEtO out x 100 = % Control Efficiency Equation (1)
ZEtO in
CARB Method 431, "Determination of Ethylene Oxide Emissions from Stationary
Sources", was based on work done by Radian Corporation for the United States
Environmental Protection Agency (USEPA)3- The method is applicable to the determination of
EtO emissions from sterilization chambers in pounds per sterilization cycle. The method
requires direct interface GC/FID monitoring of EtO emissions. The ATCM requires that the
inlet and outlet of the control device be sampled simultaneously during testing to measure
the control efficiency. Volumetric flow of vented gas is monitored and total EtO emissions
are calculated for the sterilization cycle using curves of flow and concentration determined
over time. Modifications to Method 431 are now being considered by CARB, including
options that allow the use of Tedlar bag sampling/laboratory analysis as well as an "inlet
estimation" technique.
429
-------
This paper will discuss the application of test procedures to the following types of
control systems: catalytic oxidation- (cat-ox), hydrolytic scrubber-, and recovery-type. The
discussion will primarily be a series of "problems" identified during actual testing.
CARB staff have worked closely with private source testing companies to adapt the
test procedures to the specific types of control devices used in California. In particular, CARB
acknowledges the cooperation of Mark Chips and Jeff Naus of Chips Environmental
Consultants, Inc., whose work provided much of the information presented in this report.
DISCUSSION
Catalytic Oxidation Control Systems
In the typical operation of a cat-ox unit, the EtO laden exhaust air is pulled into the
catalytic oxidizer through a prefilter by a pump, diluted with heated (300 F) ambient air and
"burned" in the presence of a catalyst bed. Typical EtO inlet concentrations after dilution are
about 3000 ppmv with typical EtO destruction efficiencies of 99.9%. Discussed below are
several general observations regarding testing of cat-ox control units followed by
manufacturer/design specific comments.
CARB Method 431 requires the use of Turbine-type or Roots-type meters as the
method was intended primarily to measure emissions flow from acid scrubbers in cubic feet
per minute. However, these types of flow measuring devices can only be used when the
temperature of stack emissions is less than approximately 200°F. Thus the use of Turbine or
Roots-type meters is not appropriate for emissions flow measurement from cat-ox units
where the temperature of emission is typically between 350°F and 450°F. Instead, CARB
Method 2 (type S pitot tube) should be used to determine stack gas velocity and volumetric
flow rate of stacks greater than 12 inches in diameter. EPA Method 2C (standard type pitot
tube) should be used where source stacks or ducts are less than 12 inches but greater than 4
inches in diameter. Typical cat-ox units operate at 50 and 100 scfm. The exhaust ducting of
a typical control unit is 4 to 6 inches and occasionally up to 10 inches in diameter. The larger
size ducting gives very low linear gas velocities (e.g., less than 10 ft./sec.) which are difficult
to measure using standard pitot tube/manometer techniques. A practical solution is to reduce
the diameter of the oversize stack to a temporary 4 inch stack during the test.
Volumetric flow of cat-ox units, corrected for temperature and pressure, is supposed
to be controlled and constant, for example at 50 or 100 scfm, but up to 10% deviation has
been observed during chamber evacuations. Thus the velocity measurements should be taken
continuously or every 2 minutes to get an accurate total volume. While concentration
measurements are obtained at both the inlet and outlet of the control unit, volumetric flow
measurements need only be taken at the outlet (assume equal scfm at inlet and outlet).
Most of the EtO charged to the chamber of a sterilizer with cat-ox control unit will be
delivered to the control unit during the first chamber exhaust. The length of this first
evacuation from hospital sterilizers will typically range between 15 and 60 minutes followed
by a series of air "washes" lasting from 1 to 2 hours. Accurate characterization of this short
emission profile by on-site GC requires a sampling (GC injection) frequency of once every 2
minutes or less.
As stated above, sterilant gases are diluted with ambient air before contacting the
catalytic bed of cat-ox units. This dilution is necessary for efficient oxidation of EtO as well
430
-------
as to prevent damage to the catalytic bed. For the purpose of compliance testing the inlet
gases should be sampled after dilution due to physical sampling difficulties before dilution as
well as for safety of test personnel. If the control unit inlet total mass is measured by on-site
GC or Tedlar bags rather than estimated, testers should have documented evidence that the
inlet probe is placed such that the sampled gases are completely mixed (i.e., chamber
exhaust and make-up air). Monitoring at locations where mixing is incomplete will produce
erroneous results which could be either high or low, depending on the probe location. We
have not seen any cat-ox units with manufacturer installed inlet probes. Testers have had to
use their best judgement to position the inlet probe, which is normally a 1/4" copper tube
centered in front of the catalytic bed. CARB proposed Test Method 422.103 describes
procedures to document correct probe position. We have found that design modifications
were necessary to perform valid inlet testing for Donaldson Company, Inc. and Autoclave
Repair Specialists, Inc. (ARS) sterilization systems.
In addition to dilution with ambient air, 3M, Inc. sterilizer systems use a pulsed
chamber exhaust flow to limit the feed rate of EtO to the cat-ox unit. This pulsed flow
consists of cycling flow "on" for 5.66 seconds and "off" for 3.73 seconds for the first 16
minutes of the first evacuation. This pulsed exhaust flow presents an inlet sampling problem
for on-site GC that is not encountered with integrated Tedlar bag sampling. On-site GC
cannot sample fast enough to accurately characterize this pulsed emission profile. To
minimize the problem of sampling a pulsed flow with on-site GC the sampling train at the
inlet to the control unit was modified to include a 1 liter Greenburg Smith impinger that
functioned as a mixing chamber for the sampled sterilizer exhaust gas prior to introduction
into the GC analyzer. The mixing of the sampled gas smooths out the variable concentrations
associated with the pulsed exhaust flow. Because the exhaust gases are adequately mixed
by passing through the catalyst bed, the impinger is not necessary for sampling at the control
unit outlet.
Many sterilization systems use water ring seal pumps to evacuate the chamber. Some
EtO will be retained in the water as the sterilant gas passes through the pump. Depending on
system design, water ring seal pumps can cause a shift in EtO emission from the initial
chamber purge to the air washes and even into the aeration cycle. In particular this effect has
been noted on MDT Castle, American Sterilizer Company (AMSCO), ARS and Banard and
Associates control units. Testers need to be aware of these emission shifts and compensate
appropriately during tests, especially when performing Tedlar bag sampling. If integrated
Tedlar bag samples are collected, sampling should extend from the beginning of the first
purge to at least 1 hour into aeration to compensate for emission shifts. In addition, the use
of a continuous monitoring (e.g., FID) is strongly recommended to provide documentation
that the bag sampling is representative of the emission curve. The inlet estimation technique
should not be used with sterilization systems using water ring seal pumps.
Testers have speculated that ETO concentrations may, in many cases, be stratified in
the exhaust duct flow from catalytic oxidation control units. Further investigation is
necessary to better define this problem. However, if stratification does occur, some sort of
sample averaging probe would be required to obtain valid test results.
Testers have observed performance problems which cause reduction of the
destruction efficiency of catalytic bed of cat-ox units. The causes of these destruction
efficiency problems are difficult to define precisely. Periodic overheating will cause the
catalytic bed to lose activity with subsequent losses in EtO destruction efficiency. Improper
installation has caused a number of catalysts to fail usually due to liquid water entering the
bed and causing damage to the catalyst during startup. Also, channels or leaks can
apparently develop in the beds. Using direct GC, a leak across the catalyst can usually be
431
-------
identified in that inlet and outlet concentration "track" perfectly at a uniform efficiency (i.e.
3000 ppmv at the inlet gives 15 ppmv at the outlet and 1000 ppmv at the inlet gives 5 ppmv
at the outlet; both pairs give exactly 99.5% efficiency in spite of the varying concentration).
In contrast, for a functioning bed, the EtO destruction efficiency will vary with temperature
(bed temperature changes during the cycle; EtO oxidation is exothermic) and EtO
concentration, throughout the cycle. Routine testing, e.g. yearly, would help identify
declining performance of cat-ox units.
Acid Catalyzed Scrubbers
Acid catalyzed scrubbers hydrolyze ethylene oxide into ethylene glycol by using an
acid catalyst in solution with water. Typically, an acid catalyzed scrubber system is
comprised of a packed tower, a reaction vessel, and a storage tank. After the sterilization
cycle is completed, the sterilization chamber is exhausted and EtO is pumped to the packed
tower where it comes into contact with an acidic water solution. Since EtO is water soluble,
most of it is absorbed into the scrubber liquor which then enters a reaction vessel where the
conversion to ethylene glycol is completed.
Revised Method 431 (422.103) allows the option to measure inlet concentrations
(e.g., with bag sampling or by direct GO instead of using the estimation technique. However,
the concentration of EtO at the inlet of hydrolytic scrubber units will be approximately 27%
and 100% by volume for systems using 12/88 Eto/Freon-12 and 100% EtO sterilant gases,
respectively. Due to the safety concerns associated with the high inlet EtO concentrations,
the estimation procedure is recommended.
The stability of ethylene oxide in hydrolytic scrubber unit emission matrix, in Tedlar
bags, has not yet been demonstrated (by ARB staff). However, a report prepared by Coast to
Coast Analytical Services, Inc.2 for the CARB suggested that acid mists in emissions from
scrubber units might cause decomposition of EtO in whole air samples. The report also
suggests that a sodium bicarbonate cartridge can be used to scrub acid mists from sample
streams without affecting the EtO.
Sterilizers with scrubber units tend to have short primary evacuation times (e.g., 5 to
10 minutes). Thus the direct interface option should only be used to test hydrolytic scrubber
units (inlet and outlet) if sample frequencies are 1 minute or less. Longer sampling
frequencies may not adequately define the emission curves. For sterilizer systems using
12/88 sterilant gas, 1 minute GC sampling frequencies are not possible. Integrated Tedlar
bag sampling is recommended.
Acidic cation exchange resins are often used to further reduce the emissions of EtO
from acid scrubbers. The failure profile of these resin beds is unknown at this time.
Recovery Units
Recovery systems for EtO/CFC mixtures are a version of condensation/expansion
refrigeration systems. At the conclusion of the sterilization cycle, the sterilant gas is drawn
from the chamber through a desiccant bed or dehumidifier and is then compressed. The
compressed and dehumidified sterilant gas reaches the condenser where it is cooled and
condensed to liquid and delivered to recovery cylinders. Recovery cylinders, when full, are
replaced and sent to reprocessing facilities. These systems are generally quite complex and
require trained personnel to keep them operating. Sterilant recovery efficiencies are expected
to be from 80 to 98 percent (for both EtO and CFC). To achieve the high control efficiencies
(99.9%) required by the ATCM reclamation systems would have to include a control device
432
-------
capable of complementing the recycling process. These recovery systems are too complex to
discuss in detail in this report but several design and testing observations are listed below.
To date, only one recovery unit has been successfully compliance tested in California.
The system, made by Joslyn Inc., recovers, under normal operating conditions, up to 98% of
the EtO and CFC. Nonrecovered sterilant gas is routed to a dilute acid scrubber which has
proven adequate even under upset conditions. Testing involved using inlet estimation
technique and Tedlar bag sampling to collect the entire- dmission from the acid scrubber
outlet for on-site GC analysis of EtO. Also, approximately 1.5 liters of EtO contaminated
water are collected in a reservoir each cycle. This water is routed to a heated tank where the
EtO is hydrolyzed before discharge of the water down the drain. Samples of water delivered
to, and discharged from, the heated tank were collected and analyzed for EtO. The system is
set up such that efficiency determination required averaging the results of multiple cycles (6
cycles).
Several other systems have been evaluated but not yet compliance tested. Donaldson
Company, Inc. manufactures a unit which recovers most of the EtO and CFC. Nonrecovered
sterilant gas is routed to a cat-ox control unit for further EtO abatement. No water is
discharged to the drain from this system. Tests will likely consist of inlet estimation along
with cat-ox unit inlet/outlet testing.
Medical Gas Industries Inc. (MGI) makes a system which reclaims most of the CFC
and destroys the EtO. The sterilant gas passes through an acid scrubber to remove EtO
before reclamation of the CFC. In addition, any nonrecovered gases pass through an acid
cation exchange resin bed before emission to atmosphere. EtO contaminated water collected
during the process is routed to the acid scrubber for abatement. The MGI system uses liquid
nitrogen to condense the CFC. Tests will likely consist of inlet estimation with outlet testing
at the resin bed exhaust and nitrogen vent line.
Again, in general, recovery systems are very complex and in many cases the
installation will be site specific. Each recovery system should be carefully inspected during a
pretest site visit to insure that it is testable. Subtle differences in piping can cause endless
problems while attempting to run a test.
"Inlet Estimation"
The amount of EtO charged to the sterilization chamber, and delivered to the control
unit, can be accurately calculated from weight loss in the charging cylinder/cartridge or from
chamber pressure/volume relationships. This estimation procedure assumes that there is no
loss of EtO to the chamber, chamber contents, transfer plumbing or pumps and that there are
no significant leaks before the control unit. Following are several observations made
regarding the application of this estimation technique.
Some sterilization systems add sterilant gas as needed to the chamber during the
exposure stage because the chamber pressure may decrease slightly after initial
pressurization. This addition of make-up gas would, if significant, invalidate the inlet
estimation calculation as with existing systems it would be quite difficult to estimate the
amount of make-up gas added. To minimize this problem, when using the inlet estimation
technique, the test should be conducted with an empty chamber and the exposure stage
should be aborted after no more than 10 minutes.
Since the estimation technique can only be used for empty chamber tests, an exposed
chamber load will not be available if subsequent aeration tests are to be performed. There
433
-------
must be an exposed load in the aerator for a valid test. Thus, an additional sterilization cycle
with unaborted exposure stage would have to be run to provide the materials to be aerated.
Furthermore, the inlet EtO concentrations must be physically measured with Tedlar bags or
direct GC for aeration tests since estimation is not possible. Thus, where aeration tests must
be conducted in addition to sterilizer tests, inlet estimation may not provide any time or cost
benefit.
Use of the inlet estimation technique assumes that the composition of the sterilant
gas is accurately defined and consistent in individual cylinders/cartridges. We have no
information that these assumptions are correct. Thus, a sample from the gas cylinder(s) used
during the test should be analyzed to verify the exact gas composition for the inlet estimate.
Accurate estimates rely on accurate volume measurements and calibrated pressure
gauges. Thus, manufacturer's chamber volume specifications should always be double
checked and system pressure monitoring devices should be evaluated for accuracy.
All sterilizers use some amount of water to activate the sterilization and there tends to
be a residual puddle left in the chamber post-cycle. The amount of EtO in and the volume of
this puddle are variable and unknown but are assumed to be insignificant.
As discussed previously, the inlet estimation technique should not be used with
sterilization systems using water ring seal pumps.
On-Site GC
Several advantages and disadvantages of on-site GC are discussed below.
Direct sampling/analysis will identify a seriously non-complying unit within the first 10
minutes of testing. Thus, the system could be shut down immediately for repairs, minimizing
the hazard associated with excess emissions.
On-site GC readily accommodates differences between control unit models. For
example, the cycle timing and emission curve can vary dramatically between models made by
different manufacturers. Also, some system configurations cause delayed release of sterilant
gas to the control unit (e.g., water ring seal pumps) such that sampling intervals may need to
be extended. With on-site GC, adjustments may be made during the test to produce more
representative results.
At many hospitals, the control unit is not accessible from parking areas (i.e., with 150
foot heated lines to a parked GC-van). Thus, the GC must be physically moved to a location
near the control unit, which may prove inconvenient. Also, adequate power may be difficult
to get at some facilities.
Many testers feel that on-site GC is more expensive and more difficult than container
sampling. In addition to the equipment required, performance of on-site GC requires that an
experienced chemist be involved in the field operations.
Tedlar Bag Sampling
Several advantages and disadvantages of using Tedlar bag sampling for EtO
compliance test are discussed below.
434
-------
Tedlar bag sampling provides a true integrated sample over the duration of the
emission, as opposed to direct GC which provides only a series of data points along the
emission curve. However, some sources such as some acid scrubber configurations may not
be suitable for bag sampling because both concentration and volumetric flow vary with time.
The Tedlar bag sampling procedure is potentially simpler and may be less expensive
than direct GC. However, appropriate quality control, such as blank bag contamination and
leak checks and field spikes and blanks must be incorporated into the test. Also, loss of
samples due to bag breakage could present a problem.
Sampling with Tedlar bags must be planned carefully so that the entire emission curve
is monitored. To provide documentation that the sampling is representative of the emission
curve, it is strongly recommended that a continuous monitor (e.g., FID) be used along with
the bag at the control unit inlet. Sampling times could then be modified as necessary to
account for shifts in emissions.
Using CARB Method 422 procedures, a limit of detection of approximately 1 ppmv
would be calculated for EtO samples collected in Tedlar bags and analyzed using the same
GC analysis as used for direct GC. This level may not be low enough for regulatory purposes.
Thus, some form of sample concentration may be necessary for analysis of EtO from Tedlar
bag samples.
CONCLUSION
As stated in the introduction, this paper has been a report of observations/problems
encountered by private test groups during compliance testing for EtO emissions from hospital
sterilizers. A goal of these tests was to adapt the test procedures for application to specific
control unit types and configurations. In some cases the control equipment required
modification to accommodate the tests. In other cases the test method was modified in order
to provide useful data. In conclusion, it seems evident that sufficient differences exist
between control units (manufacturer, model and site dependent) that close scrutiny is called
for in application of the above described test procedures to obtain useful results.
BIBLIOGRAPHY
1. K. Mongar, "Field Evaluation of Several Methods for Monitoring Ethylene Oxide
Emission from Hospital Sterilizers" in Proceedings of the 1992 U.S. EPA/A & WA
International Symposium on Measurement of Toxic and Related Air Pollutants.
VIP-25, Air and Waste Management Association, Pittsburg, 1992, pp 877 to 882.
2. J. Steger and W. Gergen, Final Report - Sampling/Analytical Method Evaluation for
Ethvlene Oxide Emissions and Control Unit Efficiency Determination. EPA -
68-02-4119, Radian Corporation, Research Triangle Park, 1988.
3. S. C. Havlicek, L. R. Hilpert, D. Pierotti, G. Dai, Draft Final Report - Assessment of
Ethvlene Oxide Concentrations and Emissions from Sterilization and Fumigation
Processes. Coast-to-Coast Analytical Services, Inc., San Luis Obispo, CA, 1992.
435
-------
EVALUATION OF POLYNUCLEAR AROMATIC HYDROCARBONS AND
NITROGEN HETEROCYCLES IN THE STACK EFFLUENT OF
ASPHALT PROCESSING PLANTS
Kin Hudak, Richard Pirolli, and Newt Roue
Connecticut Department of Enviromental Protection
Bureau of Air Management
165 Capitol Avenue
Hartford, Connecticut 06106
ABSTRACT
The Connecticut Department of Environmental Protection (CT DEP) conducted
stack effluent monitoring at several asphalt plants during 1991 and 1992. The
samples were collected in accordance with US EPA SW816 Method 0010 and analyzed
by high resolution gas chromatography/ high resolution mass spectroscopy
(HRGC/HRMS).
The HRGC/HRMS methodology was developed by Triangle Laboratories, Inc. of
Research Triangle Park, North Carolina to satisfy CT DEP's monitoring objectives.
These objectives included determining the feasibility of monitoring and
simultaneously analyzing for both polynuclear aromatic hydrocarbons (PAH) and
nitrogen heterocycles (NHC) in asphalt plant stack samples. The list of target
PAHs was developed by CT DEP to help better define its polynuclear aromatic
hydrocarbon regulations.
The test results show that seven compounds from CT DEP's targeted list of
PAH comprise 98-99% of the identified PAH emissions from asphalt plants. CT DEP
reviewed data from oil and natural gas-fired batch operations as well as natural
gas-fired batch and express plant operations. Comparisons of the results were
made.
INTRODUCTION
The Connecticut Department of Environmental Protection (CT DEP) conducted
stack emissions monitoring at several asphalt plants during 1991 and 1992. Two
types of plants at the Balf Company were tested as part of a USEPA High Risk
Point Source (HRPS) Study to determine qualitatively and quantitatively the PAH
and NHC emitted from asphalt plants. CT DEP intended to use the PAH emissions
data to regulate a specific source as well as support a revision of the existing,
but inadequate PAH regulation.
At CT DEP's request two additional plants were tested. The Astec plant, a
natural gas-fired, 4 ton/batch plant, was tested in fulfillment of a condition
written into Its permit to construct and operate. The other plant, Tilcon, a 5
ton/batch plant, agreed to be tested at the request of the department. For
testing purposes, Tilcon temporarily converted to it's backup fuel, No. 2 oil.
Connecticut has a wide variety of asphalt plants. The majority of Connecticut's
45 plants burn No. 2 oil with natural gas backup. The newer facilities are
required to burn natural gas with No. 2 oil backup. The plants chosen for study
were representative of typical facilities, both new and old. However, the study
does require more testing of the older, oil-burning asphalt plants.
436
-------
The concept for the project was formulated after attempts to estimate PAH
emissions from asphalt plants for enforcement purposes were largely
unsuccessful. Previously, estimated emissions of PAH or polycyclic organic
matter (POM) from asphalt plants were based on limited non-speciated test
data . In addition, CT's existing hazardous air pollutant regulation for PAH
was unclear and difficult to enforce because it was defined as PAH (benzene
soluble). As an interim policy designed to facilitate the permitting of new
asphalt plants, provisions were included in new permits to require testing for
PAHs consequent to the development of an appropriate test method. As a result of
this study, CT DEP now regulates specific PAH compounds and can specify a
particular test method.
EXPERIMENTAL METHODS
Sample Collection
Two of the facilities CT DEP tested were at the Balf Company of Newington,
which operates both batch and express or drum mix asphalt plants. The batch
plant has a maximum operating capacity of 10 tons/batch while burning natural
gas. The asphalt express process is a Gencor-Bituma drum mix asphalt plant.
This plant has a maximum operating capacity of 450 tons/hour while burning
natural gas. Both operations utilize a baghouse for particulate removal.
The PAH/NHC emissions from the plants were sampled in triplicate according
to SW 846 Method 0010. The volumetric flowrate, moisture content, and stack gas
composition were determined during each test run. All testing was performed on
the exhaust stack. In conjunction with PAH/NHC testing, EPA Methods 1 through 4
were performed.
The PAH/NHC sampling method used the modified Method 5 sampling train. A
modification in the sample recovery procedure was used and consisted of replacing
the 1:1 mixture of methanol/methylene chloride rinse with separate rinses of
acetone and methylene chloride. An acetone rinse followed by a methylene
chloride rinse has been shown to be more efficient than the methanol/methylene
chloride (1:1 volume) rinse. The PAH/NHC sampling method also included several
unique preparation steps which ensured that the sampling train components were
not contaminated with organics that may have interfered with analyses.
Triangle Laboratories, Inc. (TLI), Research Triangle Park, North Carolina
performed the preparation of the glass fiber filters and the XAD-2 resin. All
filters were cleaned before their initial use. The methylene chloride extract
from the filter cleaning procedures was anaylzed for PAH/NHCs. If any PAH/NHC
was present in the concentration above the minimum detectable limit, the cleaning
procedure was repeated and the extract reanalyzed until no PAH/NHC was detected.
The XAD-2 resin was placed in a soxhlet and extracted with HPLC grade water,
methanol, and methylene chloride. After extraction, the resin was dried and
placed in the sampling cartridges which were tightly capped with glass plugs.
The extracts were then analyzed for total chromatographic organics (TCO) and
targeted PAH/NHC compounds. If any PAH/NHC were present at a concentration above
the minimum detectable limit and/or the TCO was greater than 20 ug/ml, the
cleaning procedure was repeated until each criteria was met. In addition, the
resin of each trap as fortified with 100 pg/ml of dl4-Terphenyl .
437
-------
Sample Analysis
High resolution gas chromatography/high resolution mass spectroscopy
(HRGC/HRHS) analytical methodology was developed by TLI to satisfy CT DBF's
monitoring objectives 3. These objectives included determining the analytical
feasibility of monitoring for 29 polynuclear aromatic hydrocarbons and five
nitrogen heterocycles in asphalt plant stack samples. These compounds are listed
in Table 1.
Table 1.
List of Target Polycycllc Aromatic Hydrocarbons
and Nitrogen Heterocycles.
Naphthalene 2-Methylnaphthalene Acenaphthene
2-Chloronaphthalene Acenaphthalene Fluorene
Phenanthrene Anthracene Carbazole
Acridine Fluoranthene Pyrene
3-Methyl-fluoranthene Cyclopenta-c,d-pyrene Benz-a-anthracene
Chrysene Perylene Benzo-b-fluoranthene
Benzo-j-fluoranthene Benzo-k-fluoranthene Benzo-a-pyrene
Benzo-e-pyrene 7H-Dibenzo-c,g-carbazole Benzo-[ghi]-perylene
Dibenz-[ah]-anthracene Dibenz-[aj]-anthracene Dibenz-[aj]acridine
Dibenz-[ac]-anthracene Dibenz-[ah]-pyrene Dibenz-[ai]-pyrene
Dibenz-[ae]-pyrene Dibenz-[al]-pyrene Indeno-[1,2,3-cd]-pyrene
7,9-Di-Methyl-benz-c-acridine
No problems were noted by TLI's sample preparation and mass spectrometry
groups while performing the analyses. However, several factors did impact data
validity. One of the factors was matrix related interferences which affected the
quantitation of some analytes. This problem was described as severe in some
samples. Quantitation of naphthalene, chloronaphthalene, acenaphthylene,
acenaphthene, fluorene, phenanthrene, acridine and carbazole could not be
performed in a particular sample as a result of the matrix Interference. TLI did
provide quantitative results for these analytes in that sample which were
believed to be minimum estimates. However, these values were not reliable enough
to support an enforcement action. High recoveries of some internal standards
were calculated in samples as a result of this interference. TLI modified their
cleanup procedures in an attempt to more effectively remove the matrix
interferences.
A second factor was that the results of all samples displayed selective
losses during extraction for the labeled acridine and carbazole Internal
standards. Since the associated analytes act chemically in the same fashion, it
can be assumed that these and other nitrogen heterocycles are selectively lost
during extraction.
Another factor was that "b" and "j" isomers of benzofluoranthene as well as
the "ac" and "ah" isomers of dibenzoanthracene coelute. Therefore, quantitative
results were provided as a total for each isomer pair.
The last factor of concern was that saturated peaks (I.e. beyond
calibration) were noted in all the samples. TLI believed that quantitative
results for saturated signals were "minimum estimates".
The total PAH concentration was calculated for each sample. Naphthalene,
2-methylnaphthalene, and 3-methylfluoranthene were not included in the total PAH
concentrations.
438
-------
RESULTS
CT DEP reviewed data from both oil and natural gas-fired batch operations as
well as batch and express plant operations. Table 2 lists the comparison of data
from asphalt batch and express processing plants firing natural gas. Table 3
shows a comparison of three asphalt batch plants. The Balf and Astec plants were
firing natural gas. The Tilcon plant was firing No. 2 oil.
Table 2.
Natural Gas-Fired Asphalt Batch and Express Plants
PAH Data Excluding Naphthalene (ug/»3).
Analyte BATCH EXPRESS
Acenaphthene 1.75 (21%) 0.364 (9.4?)
Acenaphthylene 0.25 (4*) 1.185 (31*)
Fluorene 2.28 (36%) 0.848 (22%)
Phenanthrene 1.88 (29%) 1.147 (30*)
Anthracene 0.06 (1%) 0.066 (1.7*)
Fluoranthene 0.05 (1*) 0.119 (3.1*)
Pyrene 0.07 C\%) 0.081 (2-1*)
Benzo-a-Anthracene 6.7E-4 (.01?) 0.003 (.08*)
Chrysene 0.01 (0.2?) 0.024 (0.6?)
Benzo-k-Fluoranthene 0 (0*) 0.002 (.04*)
Benzo-j-Fluoranthene 0 (0*) 0 (0*)
Benzo-b-Fluoranthene 0.0017 (.03*) 0.005 (.12*)
Benzo-e-Pyrene 0.005 (.08*) 0.008 (.21*)
Benzo-a-Pyrene 0.0007 (.01*) 0.003 (.08*)
Indeno-123-cd-Pyrene 0.0027 (.04*) 0.003 (.08*)
Benzo-ghi-Perylene 0.027 (.43*) 0.023 (0.6*)
Dibenzo-ah-Anthracene 0 (0*) 0 (0*)
Total PAH 6.4 3-9
DISCUSSION
The majority of the observed PAH emissions were comprised of: acenaphthene
(24*), acenaphthylene (10.7*), fluorene (34.3*)> phenanthrene (22.6*), anthracene
(2.7*), fluoranthene (2.4*), and pyrene (1.8*). These 7 PAHs make up
approximately 98-99* of the PAHs identified in the stack effluent samples.
Naphthalene, 2-methylnaphthalene, and 3-methylfluoranthene were not included.
Emissions from the No.2 fuel oil fired batch process appeared to be twice
the emissions from the natural gas fired process. Emissions from the express
drum mix plant appear to be 40* lower than the batch plant (both plants natural
gas-fired). Although the data base was quite small, it appeared that
acenaphthylene was significantly more prevalent in the express plant samples.
Correspondingly, acenaphthene seemed to be far more predominant in the batch
samples.
CONCLUSIONS
Regulatory Implications
The following analytes were dropped from the target list of PAH/NHC
compounds: 2-chloronaphthalene, carbazole, acridine, cyclopenta-cd-pyrene,
dibenz-aj-acridine, 7,9-dimethyl-benz-c-acridine, perylene,
7H-dibenzo-cg-carbazole, the "ac" and "aj" isomers of dibenzanthracene, and the
"al", "ae", "ai", and "ah" isomers of dibenzpyrene. These compounds were seldom
if at all detected and then, only at trace levels in samples which contained
relatively high PAH levels (especially when compared to ambient concentrations).
439
-------
Table 3
Asphalt Batch Plants PAH Data Eicludlng Naphthalene (ug/m3).
Analyte
Acenaphthene
Acenaphthylene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo-a-Anth:
Chrysene
Benzo-k-Fluo
Benzo-j-Fluo:
Benzo-b-Fluo:
Benzo-e-Pyrene
Benzo-a-Pyrene
Indeno-12;
Benzo-ghl-
Dibenzo-al
Total PAH
HALF
ASTEC
TILCON
Natural Gas-Fired Natural Gas-Fired No. 2 Oil-Fired
10 ton/batch 4 ton/batch 5 ton/batch
1
le 0
2
1
0
0
0
•aoene 6
0
•anthene 0
•anthene 0
•anthene 0
ie 0
le 0
1-pyrene 0
•ylene 0
ithracene 0
6
.75
.25
.28
.88
.06
.05
.07
.7E-4
.01
.0017
.005
.0007
.0027
.027
J_
(27?)
(4?)
(36%)
(29?)
(1?)
(1*)
(1?)
(.01?)
(0.2?)
(0?)
(0?)
(.03?)
(.08?)
(.01?)
(.OH?)
(.43*)
(0?)
1
1
1
2
0
0
0
0
0
0
0
0
0
0
0
0
6
.46
.19
.26
.2
.032
.133
.117
.004
.015
.003
-
.013
.006
.0019
.003
.058
.0014
.5
(22.5?)
(18.3?)
(19.4?)
(33-8?)
(.5?)
(2?)
(1.8?)
(.06?)
(.23?)
(.05?)
(0.2?)
(.09?)
(.03?)
(.05?)
(0.9?)
(.02?)
3-
0.
5.
1.
0.
0.
0.
0.
0.
6.
0
0.
0.
6.
0.
0.
0
n
12
37
18
09
6
36
23
016
086
(28?)
(3.3?)
(47?)
(9.8?)
(5.4?)
(3.2?)
(2.1?)
(0.14?)
(0.77?)
7E-4(.006?)
006
008
(0?)
(0.05?)
(0.07?)
7E-4(.006?)
006
044
_._)
(0.05?)
(0.4?)
(0?)
440
-------
CT DEP now requires SW846 0010 as the PAH test method. Acetone and
methylene chloride are to be used as the sample recovery solvents. Analysis is
performed with HRGC/HRMS. Monitoring for nitrogen heterocycles requires
additional method development and validation as well as separate sampling and
analyses from PAHs. Therefore, CT DEP is not requiring sources to test for NHCs.
The asphalt plants fired with natural gas were generally well below CT DEP's
maximum allowable stack concentration for PAH. The plant fired with No. 2 fuel
oil was approximately 5 times the allowable stack concentration.
CT DEP has developed a proposed regulation for PAH and naphthalene as a
result of this project. The ambiguity of the existing definition and multiple
listings of PAH compounds in the regulations hampered both the CT DEP and the
sources in determining compliance and performing appropriate monitoring.
CT DEP proposed the following definition be added to the regulations: (PAH)
will be defined as the sum of the following compounds: acenaphthene,
acenaphthylene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene,
benzo(a)anthracene, chrysene, benzo(k)fluoranthene, benzo(j)fluoranthene,
benzo(b)fluoranthene, benzo(e)pyrene, benzo(a)pyrene, indeno-1,2,3-cd-pyrene,
benzo-ghi-perylene, dibenzo-ah-anthracene.
CT DEP proposed the following items be deleted from the existing
regulations: benz(a)pyrene, coal tar pitch volatiles, polynuclear aromatic
hydrocarbons (PAH), benz(a)anthracene, benzo(b)fluoranthene, chrysene,
7H-dibenzo(c,g)carbazole, dibenzo(a,h)anthracene, dibenzo(a,h)pyrene,
dibenzo(a,1)pyrene, indenod,2,3-cd)pyrene, and, asphalt (petroleum) fumes. CT
DEP also proposed to redefine "naphthalene" as the sum of both naphthalene and
2-methylnaphthalene.
REFERENCES AND BIBLIOGRAPHY
1. AP-42 Compilation of Air Pollutant Emission Factors Volume 1: Stationary
Sources, U.S. EPA, Office of Air and Radiation, Office of Air Quality
Planning and Standards, Research Triangle Park, HC, October 1986.
2. R. Pirolli, Emission Test Program for Selected Polycycllc Aromatic
Hydrocarbons and Mitrogen Heterocyeles at an Asphalt Processing Facility
Test Protocol, State of Connecticut Department of Environmental Protection,
Hartford, 1991.
3. H. Karam, Proposal in Response to Request From CT DEP Determination of
Polycyclic Aromatic Hydrocarbons and Nitrogen Heterocycles in Ambient Air
and Stack Emission Samples By High Resolution Gas Chromatorgraphy / High
Resolution Mass Spectroscopy, Triangle Laboratories, Inc., Research Triangle
Park, NC, September 1990.
441
-------
Session 11
Measurement of Aerosols
-------
The Summer 1992 PM-10 Saturation Monitoring Study
in the Ashland, KY Area
Eric S. Ringler*
Southern Research Institute
P.O. Box 13825
Research Triangle Park, NC 27709-3825
Van X. Shrieves
USEPA Region IV
345 Courtland Street, N.E.
Atlanta, GA 30365
Neil J. Berg
USEPA OAQPS TSD (MD-14)
Research Triangle Park, NC 27711
A PM-10 saturation monitoring field study was conducted in the Ashland, Kentucky area during July and
August, 1992. The study was sponsored by The U.S. Environmental Protection Agency (EPA), Office of Air Quality
Planning and Standards, in cooperation with EPA Region IV. The study area extended across State and Regional
borders into Ohio (Region V) and West Virginia (Region ni). The purpose of the study was to evaluate the use of
portable PM-10 monitors in examining the potential for exceedances of the NAAQS in the area and to make
recommendations for optimizing permanent monitoring locations. This paper presents the results of the study and
highlights issues of interest for the successful utilization of PM-10 saturation monitoring in PM-10 ambient monitoring
network reviews.
"Formerly with TRC Environmental Corporation
445
-------
INTRODUCTION
A PM-10 saturation monitoring field study was conducted in the Ashland, Kentucky area during July and
August, 1992. The study was sponsored by the U.S. Environmental Protection Agency CEP A), Office of Air Quality
Planning and Standards, Technical Support Division, in cooperation with EPA Region IV. The impact area extends
across State and Regional borders into Ohio (Region V) and West Virginia (Region III). These States and Regions
provided input for the study design and assistance in the study implementation. The purpose of the study was to
evaluate the use of portable PM-10 monitors in examining the potential for exceedances of the NAAQS in the area, and
to make recommendations for optimizing permanent monitoring locations. The portable monitors are not USEPA
reference/equivalent methods, and the data collected cannot be used to support regulatory or enforcement actions. The
primary monitoring objective was to identify high concentration areas. A secondary objective was to characterize the
overall spatial distribution of PM-10 concentrations in the area.
STUDY DESIGN AND IMPLEMENTATION
Network Design
The network provided spatial coverage of the industrialized area centered at the confluence of the Ohio and Big
Sandy Rivers. The network was focused on the industrial sources of PM-10 in the area. This included a steel mill, a
coke plant, an ethanol plant, an oil refinery, a chemical plant, and a carbon reclamation plant. Several large coal tipples
were also considered in selecting site locations. Each major source was covered with upwind and downwind monitors
relative to seasonal predominant winds from the southwest. Upwind monitors recorded background conditions including
emissions from area sources (especially unpaved roads) and more distant point sources.
The network included both elevated (hilltop) and base level sites to cover impacts from stack and ground-level
fugitive emissions. Sites were selected to represent potential PM-10 impacts from both point and fugitive sources near
the site. Existing monitoring data, emissions information, meteorology and topographical influences were all considered
in the network design. In addition, citizen complaint records were used to identify potential high impact areas. Site
locations avoided influences from immediate local sources such as unpaved roads, railroads, or agricultural activity.
Twenty three portable monitors were available for the study. This allowed for establishing 19 sites, given that 2
monitors were reserved as spares and 2 sites were collocated. With this number of sites, it was possible to locate sites in
expected high concentration areas near sources and distribute the sites reasonably uniformly over die study area. Figure
1 shows a map of the study area.
Meteorological data were collected during the study from the national weather service station located at
Huntington Airport (Huntington, WV), the Cooper School (Catlettsburg, KY) meteorological station operated by the
Kentucky Division for Air Quality, and a special station set up for this study at Riverview Nursing Home (Russell, KY).
The Huntington Airport station is located in West Virginia at the eastern end of the study area. The Cooper School
station is located on a hilltop on the Kentucky side of the Big Sandy River, and is collocated with site #15. The
Riverview station was located along the Ohio River at the north end of the study area, and collocated with site #04.
Description of Portable Samplers
The portable PM-10 saturation samplers used in the study were designed and built by the Lane Regional Air
Pollution Authority (Springfield, Oregon). The samplers are battery operated and are typically mounted on a utility pole.
The inlet is designed to achieve the PM-10 particle size cut point at a flow rate of 5 actual liters per minute (1pm). The
flow rate for each sampler is calibrated at the laboratory and checked periodically during operation. Flows are read
from a rotameter at the start and end of each sampling period. Constant flow is maintained by circuitry that adjusts the
pump speed in response to normal variations due, for example, to filter loading during the exposure interval. The
sampler will automatically shut down if it is unable to maintain flow within a specified interval. This could occur due to
an obstruction in the flow lines, excessive filter loading, or if the battery is unable to provide sufficient voltage to
maintain the proper pump speed. The normal exposure interval is 24 hours. The sampler is provided with a
programmable timer and an elapsed time meter to record the exposure interval.
Sampling and Analysis
Sampling was conducted from July 11 through August 18, 1992. This period was selected based on the months
when the highest participate concentrations have historically been recorded in the area. Sampling was conducted on a
daily basis at all 19 sites. Two of the sites were collocated, so a total of 21 filters were exposed each day. Samples
were changed between 4pm and 8pm each day, with the goal of completing sample changes between 4pm and 6pm. The
network was spread over a relatively large area and travel was restricted by bridge locations across the two rivers. Two
446
-------
operators were required to complete the sample changes within a 2 to 4 hour time window. Local operators were
recruited and trained for routine daily operations and quality control checks. Each sample was exposed for
approximately 24-hours. Actual start and end times were staggered according to the time required for the operator to
reach and service the sites. Samples were shipped weekly to the laboratory for weighing.
Filter Media
Both quartz fiber and Teflon filters were used in the study. Many previous PM-10 saturation studies have used
quartz fiber filters. These filters are inexpensive and are well suited for gravimetric analysis; however, they are not well
suited for microscopy or elemental analysis. Several of the study participants were interested in archiving filters from
this study for future microscopy and/or elemental analysis. Therefore, Teflon filters were obtained for use at three
selected sites to provide a higher quality media for future analysis. The sites were selected in areas where impacts from
multiple sources might be expected (sites #05, #09, and #13) since the results from microscopic and/or elemental can be
used to identify source types and relative source contributions.
The quartz filters are extremely friable, and are easily damaged during sample handling, storage, and shipment.
Early reports from the laboratory indicated that 25 to 30 percent of the quartz filters collected during the first two weeks
of the study were damaged during sample handling. The damage occurred because the filter holder compressed the edge
of the filters. The edges would then partially separate during cold storage and shipment. The filters were intact when
removed from the filter holders by the operators. Detecting the problem was delayed because the damage was only
evident after the filters were inspected at the laboratory. The problem was corrected by instructing the operators to
tighten the filter holders no more than necessary. Cold storage of the filters was also eliminated as an added precaution.
These corrective actions were implemented on July 24. After that date, far fewer filters were damaged. Starting August
12, Teflon filters were used at all sites. No filter damage occurred when using Teflon filters. The low pressure drop
type of Teflon filters that were used caused no noticeable decrease in sampler flow rate compared to using quartz filters.
DATA QUALITY AND SAMPLER PERFORMANCE
A rigorous quality assurance and quality control program was implemented for the study. This program was
designed to assure that the study produced data of known quality and to provide information that could be used to
evaluate the performance of the portable monitors. An assessment of data quality and comparability with USEPA
reference/equivalent methods was a key component of the study. In addition to provisions for proper sampling and
analytical procedures, the quality assurance plan provided for collocated sampling, duplicate weighings, collection of
field blanks, and internal and external flow audits. All data were carefully validated before analysis based on laboratory
observations, filed logs, and operator contacts..
Duplicate portable monitors were located at two sites (#12 and #05). Site #12 also contained a single PM-10
reference monitor operating on a 1 in 6 day sampling schedule. Site #05 contained two reference PM-10 monitors that
were operated twice each week on a 5 pm to 5 pm sampling schedule and a continuous PM-10 monitor (TEOM).
Overall precision based on the collocated sample pairs was about 15 % (quartz) and 5 % (Teflon). A comparison of data
from the collocated portable and reference monitors (including the TEOM) was used to assess the performance of the
portable monitors relative to EPA reference/equivalent methods. On average, the portable monitors gave results about 5
/ig/m3 higher than the reference PM10 monitors and about 10 jig/m3 higher than the TEOM.
Two laboratories were involved in the study; the Lane Regional Air Pollution Authority (LRAPA) laboratory
and the South Carolina Department of Health and Environmental Control (SC-DHEC) laboratory. Primary analyses
were conducted by the LRAPA laboratory. The LRAPA laboratory also provided the pre-weighed filters. The
SC-DHEC laboratory performed duplicate weighings (tare and exposed) on about 8 percent of the filters. The difference
between the two weighings amounted to no more than about 1 to 2 jig/m3 at a typical exposure volume (7.5 m3).
Each operator exposed approximately 2 field blanks per week for a total of 24 blanks (15 quartz and 9 Teflon)
for the study. The blanks were exposed by loading a filter into a spare sampler and mounting the sampler alongside an
operational sampler during a normal sampling interval. Blank filters were logged, handled, shipped and analyzed
according to standard operating procedures. The field blanks were used to provide information on the effects of sample
handling, storage, and shipping on PM-10 concentrations. A detection limit taken as the sum of the average blank result
and twice the standard deviation of the blank results is about 12 fig/nf for both quartz and Teflon filters.
A multipoint flow calibration (6 points) was conducted on each sampler at the LRAPA lab before the samplers
were shipped to Ashland. The standards used at the LRAPA laboratory are traceable to NIST volume standards. The
447
-------
set point was established to achieve an actual flow of 5 1pm based on the summer seasonal average temperature and
pressure for the study area (293.1 °K, 739.9 mm Hg). During sampler set-up and check out, the flow rate on each
sampler was checked with a bubble flowmeter (0-30 actual 1pm). An independent flow rate check was performed on 17
of the 23 samplers by the State of Kentucky on August 5. All samplers audited were within about ± 10 percent of 5
1pm. Flow rates were again checked against a standard at the end of the study before they were removed from the area.
All samplers were within ± 10 percent of 5 1pm.
A total of 787 PM-10 samples were collected during the study (570 quartz, 217 Teflon). Overall, 81 percent
(639) of the samples collected were valid; however, 90 percent (196) of the Teflon samples were valid, while only 78%
(443) of the quartz filters were valid. The study goal was to obtain 30 valid samples from each site. This was very
nearly accomplished except at site #10, where sampling was discontinued mid-study so that the sampler could be used to
replace a sampler stolen from site #13.
Sufficient valid data were obtained to satisfy the study objectives. Data quality improved significantly after the
problem with damaged filters was successfully corrected. Further data quality improvements were realized with the
Teflon filters. The samplers performed adequately for the purpose of identifying high concentration areas where there
may be a potential to exceed the NAAQS. With careful sample handling, adequate performance can be attained using
either quartz of Teflon filter media; however, Teflon filters gave better results for all data quality indicators and were
easier for the operators to handle.
DATA SUMMARY AND ANALYSIS
The concentration data were symmetrically distributed with a mean of 37 pg/m3 and a median of 36 pg/m3. The
maximum value was 135 ^g/m3. This value was recorded at site #01 on July 15. This is an exceptionally high value,
given the overall distribution of the data. The second highest value was 82 /ig/m3, and was recorded at the same site on
July 9, 1992.
Figure 2 summarizes the PM-10 levels recorded for each site and date using grouped boxplots. The horizontal
line across each box represents the median of the data. The central 50 percent of the data are represented within each
box. The vertical lines on each end of the box (whiskers) extend to the last data point within 1.5 times the height of me
box on either side of the median. For normal distributions, the whiskers encompass about 95 percent of the data. Data
lying outside the whiskers represent exceptionally high values. Such points are plotted as circles or pluses. The pluses
(" +") represent points that fall outside of an upper limit of twice the box height above the median. The width of each
box is proportional to the square root of the number of points represented. The width of the notches in the sides of the
boxes approximates a 95 percent confidence interval about the median. Two sites or dates can be compared by
comparing the notches. If the notches do not overlap, the median concentrations are different at a 95 percent confidence
level.
All sites are represented by a comparable number of samples (28 to 37), except for site #10 which is
represented by only 14 measurements. The majority of the PM-10 measurements across the study area fall within
approximately the same range (about 30 to 50 (tg/nf); however, PM-10 levels at sites #1, #7 and #18 tend toward the
higher values. The high value recorded at site #01 is omitted from Figure 3 for clarity. Each date may be represented
by up to 21 valid samples; however, most days are represented by about 15 to 20 valid samples. A curve is fitted (cubic
spline fit) through the daily data to represent the general trend of PM-10 levels over time. Peaks in this curve could be
associated with upsets or with meteorological conditions conducive to high concentration levels.
NETWORK REVIEW AND CONCLUSIONS
The study was scheduled to coincide with the time of year when maximum PM-10 concentrations have
historically been measured in the area. Before the existing network's ability to capture maximum PM-10 concentrations
in the area can be evaluated, it should be established that the meteorological and emissions conditions prevailing during
the summer of 1992 are representative of those prevailing in previous years. It should also be considered whether such
conditions are likely to prevail in future years. Such comparisons can be addressed in terms of changes in PM-10 source
emissions, changes in meteorological conditions, and by comparison of study results with ambient monitoring data from
previous years.
Little information was obtained for the study on the emission rates, in recent years, for the various point sources
in the area. Historical data through 1988 were obtained from the report titled Air Pollution Study of Ashland, Kentucky -
448
-------
Huntington, West Virginia - fronton, Ohio Tri-State Area.' In general, emission controls implemented at the various
facilities throughout the 1980's have resulted in improved air quality in the area; however, upsets can still have a
significant impact. No upsets were reported during the study period. The steel mill was in the process of installing new
control equipment during the study period, and emissions from the existing equipment may have been greater than
normal. Immediately following the study, however, operations at the steel plant were greatly reduced for economic
reasons. With the new control equipment in place, and the reduced scale of operations, emissions from this facility
should be much lower in future years.
The weather during the summer of 1992 was unusually cool and fair in the Ashland area, as it was over much
of the United States. The Ashland Regional Office of the Kentucky Division for Air Quality reports that there were no
truly poor air quality days during the study period. Based on previous years, at least several poor air quality days
associated with strong inversions and air stagnation should be expected during July and August. The summertime
median and maximum PM-10 levels obtained at existing sites in the area were reviewed since the initiation of PM-10
monitoring in the area (1986 through 1991). The median and maTinrnm concentrations over this period are consistent
with those obtained for the saturation study. That is, the 1992 saturation study results are representative of the historical
monitoring record for the area.
The Ashland area has historically been subject to relatively intense monitoring activity due to the concentration
of industry and the tendency for strong summertime inversions that concentrate pollutants in the area. The existing
PM-10 monitoring stations span the saturation study area and provide reasonable spatial coverage, but are not as densely
spaced as the saturation network. The existing sites tend to be population exposure oriented, while the saturation study
sites were chosen to represent expected maTirmim concentration areas. In particular, the areas represented by the
saturation study sites reporting the highest overall concentrations are not well represented by the existing network. These
saturation study sites include site #01 (Ironton, Ohio) site #07 (Ashland, KY) and site #18 (Wayne County, WV).
Site #01 is located nearer to point sources of PM-10 than the existing sites in Coal Grove and Ironton, Ohio,
and appears to capture higher PM-10 concentrations than those sites. Site #01 may not be as representative of population
exposure as the sites in Coal Grove and Ironton since it is not located in a primarily residential area, but it may represent
a maximum concentration site for the area. PM-10 monitoring was recently initiated (August 1992) in the area near site
#07. The area around site #18 is not densely populated; however, this location may serve as a maximum concentration
site.
No valid PM-10 measurements exceeding the 24-hour PM-10 standard of 150 ug/nf were recorded during the
study period. Most concentrations were well below the standard. PM-10 levels were fairly uniform across the area
represented by the network and generally fell within the range from 30 to 50 /ig/m3. Because no exceedances of the
level of the PM-10 NAAQS were recorded during the saturation study, and few values approaching the PM-10 standard
were measured, no changes to the existing PM-10 monitoring network in the area were recommended. However,
establishing additional maiinmni concentration oriented sites in the areas represented by saturation site #01 and #18
would improve the ability of the existing network to capture the highest PM-10 levels that occur.
REFERENCES AND BIBLIOGRAPHY
1. Alliance Technologies Corporation. Air Pollution Study of Ashland, Kentucky - Huntington, West Virginia -
Ironton, Ohio Tri-State Area. Final Report. Submitted to United States Environmental Protection Agency,
Region IV, Air, Pesticides and Toxics Management Division, November 1990.
2. Technical Report: Ironton, Ohio-Ashland, Kentucky-Huntington, West Virginia Air Pollution Abatements Activity
Pre-Conference Investigations. EPA-APTD-68-2. U.S. Department of Health, Education and Welfare. Bureau
of Disease Prevention and Environmental Control. Cincinnati, Ohio, May 1968.
3. U.S. Environmental Protection Agency. Muting Heights, Wind Speeds, and Potential for Urban Air Pollution
Throughout the Contiguous United States. (AP-101). Office of Air Programs, January 1972.
449
-------
Hanging
Key to Small Towns
1 - Wunland
2 - Worihington
3-Raccland
4-Flatwoods
5 - Russell
6-BeUcfome
*Mh Point Eftanol Ch«aP°<*'
Figure 1. Map of the Study Area
7/13 W7 T/21 7/M ll» V2 M U1D IA4 IAI
Figure 2. PM-10 Concentration Summary by Site and Date
450
-------
Evaluation of a Real-Time Monitor
for Particle-Bound PAH in Air
Nancy K. Wilson, Ruth K. Barbour, and Robert M. Burton
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Jane C. Chuang and Mukund Ramamurthi
Battelle
505 King Avenue
Columbus, Ohio 43201
ABSTRACT
An instrument for semi-quantitative real-time measurement of polycyclic aromatic
hydrocarbons (PAH) on airborne fine particles was evaluated. The instrument operates
on the principle of photoelectric ionization of PAH adsorbed on particle surfaces, with
resulting loss of photoelectrons and subsequent measurement of the remaining positively
charged particles. We investigated the characteristic performance of the instrument in
both chamber and field studies. This performance included: selectivity for fine particles,
response to PAH only on particles versus response to PAH in the vapor phase, accuracy
compared to integrated sampling, interferences, rapidity of response, limits of detection,
bias, ease of operation, reproducibility, calibration, reliability, and ease of field operation
and maintenance. The instrument performed well and appears to be suitable for
screening air for particle-bound PAH in a variety of microenvironments, as well as for use
in estimating human exposure related to various activities that may generate PAH.
This paper has been reviewed in accordance with the U.S. Environmental Protection Agency's peer
and administrative review policies and approved for presentation and publication. Mention of trade names
or commercial products does not constitute endorsement or recommendation for use.
INTRODUCTION
A newly developed monitoring instrument intended for real-time measurement of
PAH on airborne fine particles, the PAS 1000i (EcoChem Technologies, Inc., West Hills,
California), was evaluated in our laboratories for its potential use in ambient and indoor
air measurements and in human exposure research. To accomplish this evaluation, we
obtained two monitors and carried out a variety of experiments with them in the
laboratory and in several microenvironments.
Principle of Operation
The PAH monitor operates by photoelectric ionization of PAH adsorbed on the
surface of carbon aerosols.1"5 Air is drawn into the instrument by an internal pump at a
flow rate of 4 L/min and passes through an electrostatic precipitator to remove ionized
particles or charged gas molecules. Light at a wavelength of 185 nm from an ultraviolet
lamp then selectively ionizes the PAH on the surface of the particles, while the gases and
451
-------
other aerosols remain neutral. Diffusion or acceleration by an applied electric field
removes the photoemitted electrons and the gases from the airstream. The remaining
positively charged particles are collected on a filter inside an aerosol electrometer, where
the charge is measured. The electrometer current is proportional to the total
concentration of photoionizable particles in the air stream and hence proportional to the
particle-bound PAH concentration.
Because of the lower photoelectric threshold for larger PAH, which results from
their large ^-electron systems, photoionization is most effective for PAH with four or
more rings. Additionally, because of the smaller likelihood of recapture of photoemitted
electrons, photoionization is most effective for PAH <1-2//m diameter. This
effectiveness is enhanced for PAH on carbonaceous (typically combustion-generated)
particles and for PAH with planar molecular structures and packing. For example, it is
more effective for benzo[a]pyrene, which is planar, than for rubrene, which is non-
planar.5 For a given type of combustion, such as heating with fuel oil or automotive
gasoline combustion, the PAH profile is relatively constant.6 This allows an approximate
universal calibration of the monitor current with PAH concentration to be inferred. A
large variety of experiments indicates that the universal calibration is independent of the
type of aerosol within a factor of two.7 Good correlations of the photoemission with
individual PAH concentrations (r = 0.95 for phenanthrene and r = 0.97 for fluoranthene
spiked on carbon particles; r = 0.94 for benzo[a]pyrene in cigarette smoke) measured by
conventional means — extraction and gas or thin-layer chromatography — have also been
demonstrated.8'9 The photoelectric signal is the sum of the signals from the individual
surface-adsorbed PAH.10
To date, the greatest number of applications of photoelectric charging to PAH
measurement have been in the laboratory, and most of the applications have been to
emissions characterization. For example, a study of PAH aerosols in oil stove emissions
and automobile exhaust demonstrated a strong correlation between the photoemission
and the total PAH.11 A few investigations studied PAH aerosols in ambient air. For
example, in downtown Zurich and in Hannover, the signals from the monitor followed the
diurnal patterns and total particle counts in the ambient air; additionally the lower
response of the instrument to aged ambient aerosols demonstrated its greater sensitivity
to fine particles.12
EXPERIMENTS AND RESULTS
Instrument characteristics
Two monitors were operated continuously for periods of 8 hr to 1 mo in our
laboratory. They were collocated and positioned variously: on top of one another,
immediately beside each other, about 5 ft apart at opposite ends of a bench, and in the
middle and near the walls of the room. Stability was excellent; no anomalous results
were observed; set-up was simple; and no repair or re-zeroing of the instruments was
necessary. Excellent agreement between the results from the two instruments was
obtained; a typical correlation factor between the 1-min average concentrations from the
two units was r2 = 0.999 (in this case, for a 48-hr experiment).
To confirm that the monitor responds only to particles, the instruments were
challenged in the laboratory with many different chemical vapors, including naphthalene,
indene, anthracene, and other aromatic vapors. They gave no response, as expected.
Additionally, no response to 200-500 ppb of the PAH indene in the vapor phase was
observed in a chamber in which the particle count was 1-3/cm3.
The physical principles of the PAH monitor should make it more sensitive to PAH
on fine particles. To confirm this fine-particle selectivity, we ran two instruments, one
452
-------
with and one without a 2.5 fjm size-selective inlet,13 for periods of 1-3 days, side by side
in a laboratory served by ca. 50% outside air from vents near the building parking area.
The monitors clearly followed PAH-generating events, including heavy traffic in the
parking area and activity of utility vehicles. No difference was observed between the
responses of the two units.
The sensitivity of the monitor, if one uses the manufacturer's calibration factors, is
1 pA per 1000-3000 ng/m3. We observed a limit of detection of about 10 ng/m3
Because the irradiation from the ultraviolet lamp must be of sufficient energy to
overcome the photoelectric ionization threshold of the PAH and to remove the
photoelectrons from the surface of the particles, a slightly more energetic radiation (185
nm (6.7 eV) is used, compared to the optimum photoabsorption for PAH, which is nearer
to the 254 nm used in some of the early work. Interference is expected from some
metals and organics. However, photoemission from inorganics such as sodium chloride
should not occur (higher energy is required; for example, the work function for sodium
chloride is 8.57 eV). Literature reports indicate that photoemission from sodium chloride
particles does not occur when the salt is coated with aliphatic hydrocarbons, nor does it
occur in the absence of adsorbate.14
However, in our chamber experiments, the monitors responded to sodium chloride
aerosols in three size ranges [0.04, 0.08, and 0.15 fjm concentration mean diameter
(CMD), 22,500-50,000 particles/cm3]. This response was variable and small, compared
to that for typical indoor PAH concentrations. For example, the monitor gave a response
of 0.175 pA to 50,000 NaCI particles/cm3 at the largest CMD above, but for similar
concentrations of environmental tobacco smoke (ETS), with 0.05-0.4 fjm particles, the
response was 2.5 pA, more than an order of magnitude higher. Analysis of the bulk salt
indicated trace amounts of phthalates, suggesting that the adsorbed impurities on the salt
aerosol were responsible. Burtscher15 repeated our work, and he too found a response to
supposedly pure sodium chloride. Heating the salt to 800 C before formation of the
aerosol reduced the response to <8 fA and thus confirmed the role of impurities. It is
expected, therefore, that the instrument may normally exhibit a small positive bias due to
non-PAH impurities.
Little effect of temperature was observed on the monitor response, except that
attributable to changes in the distribution of PAH between vapor and particles as a result
of changes in their vapor pressures with temperature. Possible quenching of
photoemission by high water concentrations has been suggested.8 We are investigating
this possible influence of high humidity.
The electrostatic precipitator (ESP) in the initial section of the instrument is
intended to ensure that the incoming aerosol is electrically neutral. However, operation
of the monitor with and without the ESP made no difference in the results. The ESP may
be needed in the emissions version of the monitor, where particles may initially be far
from charge equilibrium. However, the ESP appears to serve little useful function for
indoor or ambient air aerosols, which are normally at charge equilibrium, and may in fact
cause some particle losses.
Calibration of the instrument is done by the manufacturer, and is limited to
calibration of the ionizing unit based on its ozone production.7 In field experiments, we
used a butane lighter, a kitchen match, or a cigarette to estimate the calibration and test
the response of the instrument. The monitor response agreed with that from collocated
integrated sampling within a factor of 1 to 4, using a calibration factor of 3000 ng/m3-V.
High concentrations of fine-particle PAH required a lower calibration factor; for example
ETS, which is rich in fine particles, required a calibration factor of 1000 ng/m3-V to
produce agreement with the results of integrated sampling. For higher quantitative
accuracy, source-specific calibration is desirable.
453
-------
Instrument field operation
To evaluate the performance of the PAH monitor in field situations, we
incorporated it in several small studies and used it in a variety of environments. These
included long-term evaluation in the laboratory, which we discussed previously; shipment,
field set-up, and operation in a study of indoor air in some newly constructed homes in
Colorado; monitoring fireplace operation and other combustion activities in two EPA
employees' homes; monitoring smoking, traffic patterns, and other activities inside and
outside a veterinarian's office near a busy highway; monitoring fine-particle air exchange
in a house with an attached garage; monitoring smoking and other combustion activities
in shops, bars, and restaurants; and monitoring PAH concentrations in homes with
collocated integrated sampling. Throughout the experiments, the monitors performed
well, survived shipment with no problems, were easy to set up and run, and followed
external combustion events precisely. There was no significant down-time.
The results of monitoring outside a home in a neighborhood heavily impacted by
wood smoke are shown in Figure 1. Over the weekend, many homeowners used their
fireplaces, which produced high concentrations of respirable suspended paniculate
material (RSP), as measured with a piezobalance, and high concentrations of fine-particle
PAH. Then at the beginning of the workweek, the fireplace use diminished, but
automotive traffic increased substantially, leading to increased PAH, RSP, and carbon
monoxide. In Figure 2, the results of monitoring inside a veterinarian's office are shown.
The major peaks indicate smoking by an office occupant. The fine structure of the decay
curve for the cigarette smoke gives a measure of the fine-particle air exchange rate,
which we confirmed in other studies. Minor baseline excursions can be related to other
activities that occurred in the office, for example, arrival of patients, with accompanying
increased outdoor air intrusion and mobile source contributions to the indoor
environment.
DISCUSSION AND CONCLUSIONS
Overall, we found the PAH monitor to be a useful instrument, with potential
applications in a number of areas. These include: (a) Screening microenvironments for
PAH on fine aerosols; (b) Following activities that generate such aerosols; (c) Estimating
human exposure to such aerosols; (d) Measuring phase distributions; (e) Following air
exchange of fine particles and modeling the air flows; (f) Monitoring for excess PAH
aerosols in occupational settings; (g) Monitoring and controlling combustion processes,
using the source version of the monitor;8-16 and (h) Determination of single PAH in
emissions with well-characterized profiles, for example, benzo[a]pyrene in environmental
tobacco smoke.9
Two caveats must be kept in mind in applying this instrument. First, because each
individual PAH has a different photoelectric threshold, and because the distribution of
total PAH among the various compounds will vary with the source, the monitor is most
appropriate as a screening instrument, which produces semi-quantitative results. Despite
this caveat, however, the specific responses in a given microenvironment appear to vary
precisely with the PAH levels and therefore the monitor is an excellent instrument for
following activities. Second, there can be a positive bias from some impurities and non-
PAH aerosols, which is highest at high levels of very fine particles. Therefore, to improve
the accuracy of the measurements in a given type of microenvironment, the response
should be calibrated for the specific source. This can be accomplished by comparisons
with integrated sampling. We are examining other means of calibration.
454
-------
REFERENCES
1. H. Burtscher, "Tutorial/Review: Measurement and characteristics of combustion
aerosols with special consideration of photoelectric charging and charging by flame
ions." J. Aerosol Sci. 23(61: 549-595(1992).
2. H. Burtscher, L. Scherrer, H.C. Siegmann, A. Schmidt-Ott, and B. Federer,
"Probing aerosols by photoelectric charging," J. Appl. Phvs. 53(5): 3787-3791
(1982).
3. H. Burtscher and A. Schmidt-Ott, "In situ measurement of adsorption and
condensation of a polyaromatic hydrocarbon on ultrafine C particles by means of
photoemission," J. Aerosol Sci. 17: 699-703 (1986).
4. R. Niessner and P. Wilbring, "Ultrafine particles as trace catchers for polycyclic
aromatic hydrocarbons: the photoelectric aerosol sensor as a tool for in situ
sorption and desorption studies," Anal. Chem. 61: 708-714 (1989).
5. R. Niessner, "The chemical response of the photoelectric aerosol sensor (PAS) to
different aerosol systems," J. Aerosol Sci. 17(4): 705-714 (1986).
6. W. Schmidt, "Structure and chemistry of PAH's," in Polvcvclic Aromatic
Hydrocarbons and Astrophysics. NATO ASI Series C, Vol. 191, D. Riedel, 1987.
7. E.D. Chikhliwala, E. Pfeiffer, and W. Seifert, "The design, implementation, and use
of a real-time PAH analyzer for combustion products," in Proceedings of the 9th
World Clean Air Congress. Paper No. IU-13C.07, Montreal, 1992.
8. H. Burtscher, A. Schmidt-Ott, and H.C. Siegmann, "Monitoring paniculate
emissions from combustion by photoemissions," Aerosol Sci. Technol. 8: 125-132
(1988).
9. R. Niessner and G. Walendzik, "The photoelectric aerosol sensor as a fast-
responding and sensitive detection system for cigarette smoke analysis," Fresenius
Z. Anal. Chem. 333: 129-133 (1989).
10. R. Niessner, B. Hemmerich, and P. Wilbring, "Aerosol photoemission for
quantification of polycyclic aromatic hydrocarbons in simple mixtures adsorbed on
carbonaceous and sodium chloride aerosols," Anal. Chem. 62: 2071-2074
(1990).
11. S.R. McDow, W. Giger, H. Burtscher, A. Schmidt-Ott, and H.C. Siegmann,
"Polycyclic Aromatic Hydrocarbons and Combustion Aerosol Photoemission,''
Atmos. Environ. 24A: 2911-2916(1990).
12. A. Leonard!, H. Burtscher, U. Baltensperger, A. Weber, A. Krasenbrink, and B.
Georgi, "Ambient aerosol characterization by comparison of particles size and mass
with epiphaniometer and photoemission data." J. Aerosol Sci. 21. Suppl. 1: 5189-
5192 (1990).
13. V.A. Marple, K. Rubow, W. Turner, and J.D. Spengler, "Low Flow Rate Sharp Cut
Impactors for Indoor Air Sampling: Design and Calibration," JAPCA 37: 1303-
1307 (1987).
14. H. Burtscher, R. Niessner, and A. Schmidt-Ott, "In situ detection of photoelectron
emission of PAH enriched on particle surfaces," in Aerosols: Science. Technology.
and Industrial Applications of Airborne Particles: B.Y.H. Liu, C.Y.H. Pui, and H.J.
Fissan, Eds.; Elsevier, New York, 1984, pp 443-446.
15. H. Burtscher, Private communication (1993).
16. A. Zajc, E. Uhlig, H. Hackfort, and R. Niessner, "On line and in situ control of
aerosol emission from hazardous waste combustion," J. Aerosol Sci. 20: 1465-
1468 (1989).
455
-------
Outdoor Concentrations al a Home on Sylvian Way from 2/7/93 (Sunday) to 2/8/93 (Monday)
X
45
40
35
30
25
20
15 •
10 •
5 •
0
PAH (ug/m3 multiplied by 10) RSP(ug/m3)
Sunday
Monday
a -
Figure 1. Polycyclic aromatic hydrocarbon (PAH), respirable suspended particles (RSP),
and carbon monoxide (CO) concentrations as a function of time in ambient air in a
neighborhood in which woodburning fireplaces were in use.
PAH Monitor - Inside
Eno Vet
Cigarette
Cigaretti
1000 1200
Time
Figure 2. Polycyclic aromatic hydrocarbon (PAH) concentrations as a function of time
inside a veterinarian's office in which occupants smoked.
456
-------
Particle and Gas Transmission Characteristics
of a VAPS System and Three Different Inlets
Dennis D. Lane,
Stephen J. Randtke
Professor
Ray E. Carter, Jr.,
Research Assistant
Department of Civil Engineering
University of Kansas
Lawrence, KS 66045
Robert K. Stevens
AREAL
U.S. EPA
Research Triangle Park, NC 27711
Introduction
Ambient air sampling for the criteria air pollutants is a
challenging task. A sampling system capable of collecting both
organic and inorganic gases and size-differentiated particles,
simultaneously, at remote locations would be a valuable addition to
the existing cadre of sampling equipment. An improved sampler
referred to as the Versatile Air Pollution Sampler (VAPS) (see
Figure 1) was designed for this purpose.
Before the VAPS can be deployed at field sites to do routine
ambient air quality monitoring, its physical and chemical
performance characteristics need to be confirmed in carefully
controlled laboratory studies. A sampling system consisting of one
of three different inlets, a VAPS, annular denuders, and filter
packs was challenged with solid uranine particles, sulfur dioxide,
and nitric acid to determine the system's particle loss
characteristics, the 50% aerodynamic particle cut-point (d50) of the
virtual impactor in the VAPS, the d50 of each inlet, and the gas
adsorption characteristics of the entire system. The d50 has been
determined for three different inlets: a glass Dichotomous Sampler
Type Inlet (DSTI) (see Figure 2), the Slotted Cap Inlet (SCI) (see
Figure 3) and a modified version of the University of Minnesota
Inlet (UMI) (see Figure 4) sampling in an average wind velocity of
12.3 mph. The dso of the SCI was also determined while sampling in
average wind velocities of 5.3 mph and 9.3 mph. Particle loss in
the virtual impactor section of the VAPS was determined through
laboratory experiments. Adsorption of HNO3, HN02, and S02 by the
VAPS, by a PTFE-coated inertial impactor, and by a PTFE-coated
metal DSTI was also examined.
457
-------
Beckman Olcnotomous Sampler Inlet Constructed from
Teflon-Coated Glass and Modified for the VAPS (DST1)
ur, i Versatile Air Pollution Simpler (VAPS)
Figure J
Slotted-Cap Inlet (SCI)
University of Minnesota Inlet Constructed from
Teflon-Coated Class and Modified for VAPS (UMI)
458
-------
Experimental Procedures
The particle loss characteristics of the VAPS and the inlets
and the d50 of the VAPS virtual impactor and each inlet were
determined in a series of controlled laboratory experiments. The
VAPS was mounted with each inlet in a specially designed wind
tunnel capable of producing equivalent ambient wind speeds between
1.5 and 12.3 mph. For each test, the system was operated at a flow
rate of 30 1pm. Monodisperse aerosol particles were generated
using a vibrating orifice aerosol generator. The solid uranine
particles were continuously injected into the windtunnel upwind of
the test section for the duration of each test. Aerodynamic
particle diameters used in the study ranged in size from 3.56 ^m to
22 [im. The particles and air stream were thoroughly mixed before
reaching the test section to insure a uniform particle distribution
at the test section. An isokinetic sampling probe was mounted in
the test section to measure the particle concentration for each
test.
After each test, the inlet section, the VAPS body, and the
VAPS top and bottom virtual impactor jets were carefully washed
with a known volume of reagent water. The 47-mm filter in the
filter pack was completely immersed in a known volume of reagent
water. Since the particles used in the study were fluorescent, the
solutions were analyzed for total particle mass using a
fluorometer. The mass of particles collected in each inlet is
reported as a percent of the total particle mass collected by the
sampling system. The particle mass lost in the VAPS was calculated
as a percent of the total particle mass collected in the VAPS and
on the backup filter.
Adsorption of HN03, SO2, and HNO2 (present in the HNO3
solution) by VAPS components and by PTFE-coated DSTI and
inertial-impactor inlets was examined by diffusing the gases into
purified and filtered room air. The surfaces were then extracted
with ion chromatography (1C) eluant and analyzed for nitrite,
nitrate, and sulfate by 1C. Reagent blanks and system blanks were
run and contained negligible amounts of impurities.
For some adsorption tests, the surfaces were lightly coated
with Vaseline by immersion in a 10 gm/L suspension of Vaseline in
pentane followed by air drying. The effect of scratches in the
PTFE coatings was examined by making six deep lengthwise cuts (down
to bare metal) with a sharp knife on both the outer and inner
surfaces of both the top and bottom jets of a VAPS virtual
impactor.
Discussion of Results
The virtual impactor in the VAPS has a d50 of 2.40 Jim
aerodynamic particle diameter. Figure 5 shows a plot of the
459
-------
dso Determination ol Virtual Impactor in VAPS
|- 00
1
> 60
1.0 10.0
Aerodynamic Pamela Diamelef
o Determination of SCI at Three Wind Speeds
2
f
I
~i 1 1—i—\—rr n~
* 5,3 mph
P 9.3 mph
• 12.3 mph
2 5 10 20
Aerodynamic Particle Diameter (jjm)
o Determination of DST) Sampling in a 12J3 mph Wind
2 5 10
Aerodynamic Partds Diameter (pm
Determination ol UMi Sampling in a 12.3 mph Wind
5 10 20
Aerodynamic Panicle Diameter (pm)
f ,
-D
1
: D As Received
I O Scralched
- A Vaseline Coaled °O
D
n D
i 8"D A A
: A A
>A
1 10 100
HNCh Concenlralion fojg/m3)
460
-------
experimental data including all of the particle mass captured in
the VAPS, denuders and filter packs. The d50 is sharp and little
or no particle loss was found on the virtual impactor jets (see
Table 1). Figure 6 shows the d50 experimental results for the SCI
sampling in three different wind velocities. At 5.3 mph, the d50
of the SCI is 10.5 Mm aerodynamic particle diameter. The cut-point
is not sharp (like a jet-type inertial impactor) but it is
distinct. At 9.3 mph, the experimental data show that the dso is
7.0 Mm aerodynamic particle diameter, and the cut-point is sharp,
much like that of a jet-type inertial impactor. At 12.3 mph, the
d50 of the SCI is 6.0 Mm aerodynamic particle diameter.
The d50 experimental results for the glass DSTI and the UMI
sampling at 12.3 mph are incomplete at this time and should be
viewed as preliminary (see Figures 7 and 8). It is evident that
the dso for the glass DSTI will be less than 6.0 /im aerodynamic
particle diameter and the d50 of the UMI will be approximately 7.0
Mm aerodynamic particle diameter. The true shapes of the
performance curves are not discernable from the limited amount of
data reported.
Table l: Particle Losses in the VAPS Virtual Impactor Section
Test #
Aerodynamic Particle
Diameter (Mm)
20
18
15
17
22
16
19
12
14
1.2
2.0
2.2
2.5
3.2
3.4
3.6
4.6
5.7
Virtual Impactor Section
Particle' Losses (%)
Bottom Jet
0
0
0
1
5
0
2
0
0
Top Jet
0
0
0
2
1
0
2
0
0
Body
0
0
0
0
0
0
0
0
0
The first sets of adsorption tests examined the uptake of HNO3
by a VAPS fitted with a PTFE-coated inertial impactor. Significant
adsorption (at times exceeding 20% of the HNO3 entering the system)
occurred and measurable amounts of HN03 were found on all of the
system components, including the inlet, the impaction disk (whether
oiled or not), the virtual impactor jets, and the main block of the
VAPS. Coating of the virtual impactor jets with Vaseline
461
-------
dramatically reduced uptake of HNO3, as did triple-coating them
with PTFE (Table 2).
Table 2: HNO3 ADSORPTION ON
VIRTUAL IHFACTOR JETS
JETS USED
"Original Recipe"
Triple PTFE Coated
Vaseline Coated
(Original Jets)
Avg. HN03-N
uo/m3
25.1
45.5
31.6
Percent
Adsorbed
2.56 ± 0.96
0.18 ± 0.13
0.06 ± 0.09
In another set of tests, the VAPS was fitted with a PTFE-
coated DSTI inlet and exposed to HNO3, HNO2, and SO2
simultaneously. Significant losses of HNO3 and SO2 occurred,
especially at low concentrations (Table 3) ; but no loss of HNO2 was
observed in these or any other tests. The DSTI was responsible for
a majority of the losses.
'Table 3: TRANSMISSION EFFICIENCY AND ABSORPTIVE LOSSES
USING THE DSTI INLET WITH THE VAPS SYSTEM
HNO3-N
Expt.
No.
5
6
7
8
9
10
SO2 as Sulfate
Cone.
/ml
5.8
3.9
0.3
0.3
2.3
0.3
89
90
67
79
90
79
62
70
78
57
75
55
Cone.
ug/m3
615
586
5.4
4.7
1.8
1.5
89
97
77
93
% Loss
on DSTI
96
81
96
88
96
75
A recent series of tests examined the uptake of HNO3 and S02
by VAPS virtual impactor cones tested "as received", deeply
scratched, and deeply scratched but lightly coated with Vaseline.
Adsorption of SO2 was insignificant in all three sets of tests. The
"as received" and deeply scratched jets adsorbed a significant
amount of HNO3, with the data for both lying on the same adsorption
isotherm (Figure 9). Adsorption of HN03 increased with increasing
concentration, as expected; but the relative losses (expressed as
percent loss) were greatest at the lowest concentrations. A light
coating of Vaseline on the scratched jets caused a dramatic
reduction in HNO3 adsorption.
462
-------
Conclusions
The following conclusions can be drawn from the reported
results.
1) The d50 of the VAPS virtual impactor is 2.40 /™ aerodynamic
particle diameter.
2) The particle losses in the VAPS are less than 6% and these
occur on the virtual impactor jets.
3) The d50 of the SCI varies with ambient wind velocity. The
largest d50 value is 10.5 jm at an ambient wind velocity of 5.3
mph. At 9.3 mph the dso drops to 7.0 urn and at 12.3 mph the
dj0 is 6.0 Mm-
4) The results reported for the DSTI and the UMI while sampling
in a 12.3 mph ambient wind velocity are of a preliminary and
incomplete nature. The d50 of the DSTI is less than 6.0 jjm and
the d50 of the UMI is approximately 7.0 ^m.
5) Significant adsorption of HNO3 by the VAPS and by PTFE-coated
DSTI and inertial-impactor inlets was often observed over a
broad range of HNO3 concentration.
6) Virtual impactor jets adsorbed more than 10 times less HNO3
when triple-coated with PTFE and more than 30 times less when
lightly coated with Vaseline.
7) Transmission of HNO3 and SO2 through the VAPS and a PTFE-
coated DSTI ranged from 67 to >99 percent. Values below 90%
were associated with low concentration. The DSTI was
responsible for the majority of the losses. HNO2 was not
adsorbed by any of the surfaces studied.
8) As the concentration of HNO3 decreased, the mass adsorbed on
the VAPS virtual impactor jets decreased, but the percentage
adsorbed increased. Hence, adsorptive losses are most
significant at low concentrations.
9) Deeply scratched VAPS virtual impactor jets adsorbed no more
HNO3 than unscratched jets. Adsorption of S02 was negligible
whether or not the jets were scratched.
10) A light coating of Vaseline reduced adsorption of HN03 by the
VAPS virtual impactor jets by about an order of magnitude.
463
-------
Impact of Changes in Sulfate Aerosol Loading on Greenhouse Warming
V.K. Saxena and J.D. Grovenstein
North Carolina State University
Department of Marine, Earth and Atmospheric Science
Box 8208
Raleigh, NC 27695-8208
ABSTRACT
When fossil fuel is burned, both CO2 and SO2 are injected into the atmosphere. As a result of the
Clean Air Act, emissions of SO2 relative to CC-2 are expected to change. Sulfate aerosols resulting from
the gas-to-particle conversion of SO2 are capable of changing climate by: (1) clear sky radiative effects
and (2) modifying the cloud albedo. The latter results from increased cloudiness and larger
concentrations but smaller diameter of cloud droplets since sulfate aerosols constitute effective cloud
condensation nuclei (CCN). We used diagnostic modeling to simulate the effect of the changes in
sulfate aerosol loading on greenhouse warming. Three scenarios, suggested by the Intergovernmental
Panel on Climate Change (IPCC), were used. They are: growth of emissions by 2% per year, a
reduction by 2% per year, and a "business as usual" scenario of 0% per year. The optical thickness of
the cloud was found to be proportional to Nc^^ where Nc is the number concentration of cloud droplets.
A 2% per year reduction of fossil fuel emissions indicates a 30% reduction in SO2-derived cooling in
the next 50 years. Two percent growth per year for 50 years will cause a 40% increase in SO2 derived
cooling.
INTRODUCTION
Emissions of Sulfur Dioxide (SO2) from the burning of fossil fuels leads to the presence of sulfate
aerosols which influence climate by: (1) clear sky radiative effects and (2) modifying cloud albedo. For
clear sky conditions alone, the cooling caused by current emissions rates is estimated 1 to be 1.0 Wm~2
averaged over the Northern Hemisphere. Observed^ increases in sulfate derived cloud droplet number
concentration indicate a cooling of 2-3 Wm~2 in eastern North America due to increased reflectivity of
clouds. These values are comparable to the estimated-^ 2.5 Wm"2 heating due to anthropogenic
greenhouse gas emissions up to the present. The short residence time of SO2 and the non-uniform
distribution of anthropogenic SO2 emissions produce regional variation in SO2-cooling effects. Future
changes in climate forcing due to SO2 and CO2 will depend on the variation of emissions of these two
gasses relative to one another. Reduction of SO2 emissions as mandated by the Clean Air Act will
cancel cooling due to SO2. Because of the short atmospheric lifetime of SO2, ambient concentrations
will adjust within weeks to changes in emissions. The concentration of CO2, however, will continue to
rise for more than a century even if emissions are kept constant at present levels. It is for this reason the
question has been asked," Will reducing the emissions of fossil fuels cause global warming?"
Wigley^ has attempted to determine the sensitivity of the climate system to changes in the
emission of both carbon dioxide and sulfur dioxide. The Intergovernmental Panel on Climate Change
(IPCC), in their assessment of global climate change, have recognized the importance of carbon dioxide
from fossil fuel emissions in climate change. They have proposed three scenarios of change in
concentration of atmospheric carbon dioxide. These scenarios are growth of emissions by 2 percent per
year, a reduction of two percent per year, and a "business as usual" scenario of zero growth per year.
Wigley^ uses the carbon dioxide concentration scenarios of the IPCC and the forcing-concentration
relationship of the IPCC^ Working Group 1 to determine climate sensitivity to simultaneous changes in
CO2 and SO2. In his analysis, a diagnostic model of global climate change based on direct radiative
effects of SO2 in the clear atmosphere and indirect effects of SO2 in the atmosphere as a modifier of
cloud albedo was developed. The role of SO2 in the formation of hygroscopic sulfate aerosol is well
known as is their capability of increasing the reflectivity of low level clouds0'' >°. This increase in
reflectivity in turn produces a cooling effect on the earth-troposphere system. It is suggested that
regionally SO2 derived cooling is currently counteracting the warming produced by carbon dioxide.
The greenhouse warming of the earth-troposphere system caused by doubling of CO2 could be
counteracted by a meager 2% increase in the shortwave albedo of the low level cloud cover around the
globe. Such a change could be initiated^ by a four fold increase in CCN concentration.
464
-------
CLIMATE MODEL
Wigley4 has developed a simple diagnostic model to determine the sensitivity of the climate
system to changes in the emission of SO2- If dE represents the change in SO2 emissions, the clear-sky
direct radiative forcing dQd may be considered as linearly related to dE:
dQd = adE. (1)
For the indirect SO2 effect through cloud mediated albedo changes, dQj a dC/C, where C represents
CCN number concentration. Assuming that changes in C are linearly related to changes in SO2-related
aerosol concentration, we may write
dQ ; a dE/E = b dE/E. (2)
Here a and b are constants of proportionality whose magnitude should be determined through
experiments. However, uncertainties in a and b are avoided by Wigley4 by expressing future radiative
forcing changes in terms of present manmade SO2-related forcing(5Qo):
where EQ is the natural background emission rate, Ej is the current total emission rate (at time t = 0)
and a is the growth rate of emissions. The growth rate a is taken from the three previously mentioned
scenarios of the IPCC.
The result of Wigley's model (Fig. 1) shows that a reduction in the emission of fossil fuel will
cause a positive SO2-related radiative forcing change. This will lead to global warming by CO2. There
will no longer be a SO2-derived cooling to counteract greenhouse warming by CO2.
In the development of his model, Wigley4 assumes a linear relationship between SO2 emissions,
CCN concentrations, and cloud droplet concentrations. Another relationship that was assumed linear by
Wigley4, the relationship between cloud reflectivity and cloud droplet concentration, is non-linear. We
have developed a new model utilizing Wigley's modeling technique that includes the non-linear
relationship between cloud reflectivity and cloud droplet concentration to determine the change in cloud
radiative forcing due to changes in SO2 emissions.
If, like Wigley, we assume that there is a linear relationship between SO2 emissions, CCN
concentrations, and cloud droplet concentrations, then the size of cloud droplets will become smaller
with increased SC"2 emissions. The reasoning for smaller cloud droplets is that compared to clouds
formed under similar conditions, those clouds formed with air masses of increased CCN content have
smaller cloud droplets because the available water vapor will be shared by more cloud condensation
nuclei that are activated in a cloud. Reduced cloud droplet size in clouds formed with invariant liquid
water content implies increased cloud droplet concentration. Cloud droplet size influences cloud albedo.
A large change in cloud albedo with a small change in cloud droplet radius is due to a non-linear
relationship between cloud droplet concentration and cloud albedo. The albedo of the cloud, Ac, can be
evaluated in terms of the optical depth 10 as
Ac=t/(7.7+r). (4)
Optical depth, t, is related to the number of cloud droplets, Nc, the cloud liquid water content, q c, the
cloud thickness, H, and the density of liquid water pco by the relationship^
T=H[97c(qc)2Nc/(2pw2)]1/3 (5)
As cloud droplet size is reduced (Fig 2), with invariant cloud liquid water, the albedo of the cloud
increases. In Fig. 3 cloud albedo is shown as a function of cloud droplet concentration. Using the non-
linear relationship would alter Wigley's estimation of the future indirect effect of SC-2 on global climate.
The number concentration of cloud droplets, NC is linearly related to the of CCN concentration in
an airmass but the relationship is not a simple one1 1. For our model let us suggest the relationship
Nc = a NCCN (6)
465
-------
where a is a percent of CCN activated as cloud droplets and is a function of the characteristics of the air
mass that formed the cloud.
In the case of conversion of SO2 into sulfate aerosol, Wigley assumes that SO2 will be converted
directly into CCN and ignores the physics of the mechanisms by which this conversion takes place in the
atmosphere. In the aqueous production of sulfate from SO2, there exists a limiting factor, namely, the
pH of the cloud water. If the cloud water pH is less than four the con version 12 of SO2into sulfate will
be negligible, regardless the concentration of SO2- It is with this limitation that we will relate the
number of CCN to the emissions of SO2 by
NCCN = bEsc>2 (7)
where b is the percent of E $o2 that will be converted into sulfate CCN through cloud processes.
With these new considerations we can use Wigley's modeling approach to determine future
indirect SOi-related climate forcing. Let us consider future SO2 emissions as determined by an
equation of the form
E0e<*t (8)
where dEgo, is future emissions, EQ is the current rate of emissions and a is the growth rate.
For the indirect SO2 effect through cloud albedo, utilizing equation 5 we can write
dQ = (dNc)1/3 (9)
Equations (6) and (7) imply that the droplet concentration can be rewritten in the indirect equation as
dQ = (abdEs02)1/3 (10)
and by applying the emissions growth equation (8) we find that
dQ = (a b E0 eat)1/3. (11)
We can rid ourselves of the uncertainties in the values of a and b by expressing the future forcing
changes in terms of the present day SO2 forcing Q0
Q\£ — ^0 ^ \^^/
If we manipulate the equation to
dQ at/3
/~\ ~ C I ij)
VQ
then climate change in the form of cloud climate radiative forcing can be expressed as a percent change
from the present without the need to calculate any current cloud radiative forcing values.
If we apply equation (13), the ratio of the values of future cloud climate radiative forcing to present
day cloud climate radiative forcing can be found for 50 years from the present. Selecting a equal to +2
percent per year and -2 percent per year to represent an SO2 growth scenario and a SO2 reduction
scenario respectively we can examine the sensitivity of the climate system to changes in the emissions
of fossil fuel derived SO2-
CONCLUSIONS
Figure 1. shows the result of the new model with those of Wigley. A two percent per year
reduction in the emissions of SO 2 would reduce the SO2-related radiative forcing by clouds resulting in
warming. A two percent per year growth of emissions would produce a much larger growth in radiative
forcing by clouds, resulting in cooling. The results of the new model differ from the results of Wigley
(1991) due to the application of the non-linear relationship between cloud reflectivity and cloud droplet
concentration. The application of the non-linear relationship decreases the magnitude of the aerosol
contribution to climate forcing. Table 1 shows the combined CO2 and SO2 forcing changes for the
Northern Hemisphere as calculated by Wigley. Table 2 shows the combined CO2 and SO2 forcing
changes calculated by the new model. Wigley estimated, over the next 10-30 years, that it is the aerosol
466
-------
effect that will influence changes in total forcing, to the extent that reduced emissions produce a larger
forcing increase, thus a larger global warming than increased emissions . The new model verifies these
conclusions. It is several decades before any noticeable radiative forcing response to any emission
policy occurs, due to the large response time of the carbon cycle. The aerosol effect delays this
response. The emission control scenarios show increases in forcing changes (warming) depending on
the strength of the aerosol effect, AQO. This tendency would be exacerbated by any SO2-specific
emissions controls as those mandated by the Clean Air Act. Even a small aerosol effect would delay the
response of the climate system to attempts to control or limit the greenhouse problem. Given the
regional nature of SO2 cooling, regional climate is more susceptible to forcing changes than the global
climate.
REFERENCES
1. Charlson, R.J., J. Langner, H. Rodhe, C.B. Leovy, and S.G. Warren, "Perturbation of the northern
hemisphere radiative balance by backscattering from the anthropogenic sulfate aerosols," Tellus.
43AB: 152-163 (1992).
2. Leaitch, W. R., Isaac, G. A., Strapp, C. M., and Wiebe, H. A., "The relationship between cloud
droplet number concentrations and anthropogenic pollution: Observations and climatic
implications," J. Geophvs. Res. 97: 2463-2474 (1992).
3. Climate Change: The IPCC Scientific Assesment: J.T. Houghton, G.J. Jenkins and J.J. Ephraums,
Eds. Cambride University Press, Cambridge, U.K. 1990, pp.365.
4. Wigley, T. M. L., "Could reducing fossil-fuel emissions cause global warming?," Nature 349:
503-506 (1991).
5. Shine, K.P., Derwent, R.G., Wuebbles, D.J., and Morcrette J.-J., Climate Change. The IPCC
Scientific Assesment, In Houghton, J.T., Jenkins, G.L. and Ephraumas, J.J., Eds, Cambridge
University Press, 1990, pp.41-68.
6. Twomey, S., 1977. "The influence of pollution on the shortwave albedo of clouds," J. Atmos. Sci.
34: 1149-1152(1977).
7. Twomey, S. A., "The influence of polution on the shortwave albedo of clouds," J. Atmos. Sci.
34: 1149-1152(1977).
8. Twomey, S., M. Piepgrass, and T.L. Wolf, "An assessment of the impact of pollution on global
cloud albedo," Tellus. 36B: 356-366 (1984).
9. Ghan, S. J., Taylor, K. E., Penner, J. E., and Erickson III, D. J., "Model test of CCN-cloud albedo
climate forcing," Geophvs. Res. Lett. 17: 607-610 (1990).
10. Lancis, A. A. and Hansen J. E., "A parameterization of the absorption of solar radiation in the
earth's atmosphere." J. Atmos. Sci. 31: 118-133 (1974).
11. Shaw, G.E., "The role of sulfur in cloud activation" in Nucleation and Atmospheric
Aerosols/Proceedings of The Thirteenth International Conference on Nucleation and Atmospheric
Aerosols . A. Deepak Publishing, HamptonVA. 1992, pp 369-372.
12. Scott W.D. and P.V. Hobbs, "The formation of sulfate in water droplets," J. Atmos. Sci. 24-
54-57 (1967).
467
-------
Table 1. Combined CO2 and aerosol forcing changes in the Northern Hemisphere for different emissions
growth rates and different strengths of the aerosol effect (AQO).
Emissions Year
growth AQ0
rate (a) (w m-2) 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
+2% yr-1
0% yr-1
-2% yr-1
0.0
-0.5
-1.0
-1.5
all
0.0
-0.5
-1.0
-1.5
0.36
0.27
0.19
0.10
0.30
0.26
0.33
0.40
0.48
0.77
0.59
0.41
0.22
0.58
0.43
0.56
0.70
0.83
1.25
0.95
0.66
0.36
0.83
0.53
0.72
0.91
1.10
1.80
1.37
0.94
0.51
1.06
0.59
0.83
1.07
1.30
2.43
1.84
1.25
0.66
1.27
0.63
0.90
1.18
1.46
3.14
2.36
1.59
0.82
1.47
0.64
0.95
1.26
1.57
3.94
2.95
1.96
0.97
1.67
0.63
0.97
1.31
1.65
4.84
3.59
2.34
1.09
1.85
0.61
0.98
1.35
1.71
5.74
4.22
2.66
1.10
2.03
0.58
0.97
1.36
1.75
6.69
4.75
2.82
0.89
2.20
0.55
0.96
1.37
1.77
7.64
5.26
2.88
0.50
2.37
0.51
0.94
1.36
1.78
For a = 0% yr-1 (middle panel) the aerosol effect is zero and the values shown are therefore those for CC>2
alone. Similarly, for any a value, the aerosol effect is zero for AQO = 0 Wm"2. These rows therefore give the
effect of CC>2 alone for a # 0% yr-1. They also give the changes assumed for the Southern Hemisphere.
Table 2. Combined CO2 and aerosol forcing changes in the Northern Hemisphere for different emissions
growth rates and different strengths of the aerosol effect (AQO).
Emissions Year
growth AQO
rate (a) (w m-2) 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
+2%yr-1 0.0 0.36 0.77 1.25 1.80 2.43 3.14 3.94 4.84 5.74 6.69 7.64
-0.5 0.32 0.7 1.14 1.65 2.23 2.89 3.15 3.42 4.14 4.93 5.75
-1.0 0.29 0.63 1.03 1.5 2.04 2.65 3.05 3.35 3.57 3.75 3.82
-1.5 0.25 0.56 0.92 1.35 1.84 2.4 1.76 1.52 1.38 1.10 0.89
0%yr-! all 0.30 0.58 0.83 1.06 1.27 1.47 1.67 1.85 2.03 2.20 2.37
-2% yr-1 0.0 0.26 0.43 0.53 0.59 0.63 0.64 0.63 0.61 0.58 0.55 0.51
-0.5 0.29 0.49 0.62 0.71 0.78 0.73 0.82 0.82 0.81 0.79 0.77
-1.0 0.33 0.55 0.71 0.83 0.91 0.97 1.00 1.02 1.03 1.04 1.03
-1.5 0.37 0.61 0.8 0.95 1.07 1.14 1.19 1.23 1.26 1.27 1.29
For a = 0% yr1 (middle panel) the aerosol effect is zero and the values shown are therefore those for CO2
alone. Similarly, for any a value, the aerosol effect is zero for AQ0 = 0 Wm"2. These rows therefore give the
effect of CO2 alone for a * 0% yr'l. They also give the changes assumed for the Southern Hemisphere.
468
-------
o
<
o
i.o
0.8
0.6
0.4
0.2
-0.0
-0.2
-0.4
-0.6
-0.8
-1.0'
• 2%
o
•o
a
<
10 20 30 40 50 60
Years From Present
0.
0.5
0.4
0.3
0.2-
0 5 10 15 20 25 30
Cloud Droplet Radius(Mm)
Figure 2. Cloud Albedo vs. Cloud Droplet Radius in micrometers
for a cloud of liquid water content of 0.3 gm'3 and 200m
thickness.
Figure 1. Northern Hemisphere SO^-related forcing cha/iges for
emissions growth rates of+2% yr"' fiower curves) and
-2£'vrl fuppercunes) as calculated by W'igley4. The
full lines represent me indirect (CCN'i effect tor a
present to pre-industnal SO2 emissions ratio' f EI/EQI of
3.6. The dashed lines represent the indirect fCCN) effect
incorporating the non-linear nature of the relationship
between cloud droplet concentration and cloud
reflectivity.
O
•a
01
0.58
0.56
0.54
0.52
0.50-
0.48
100
200
300
400
500
Cloud Droplet Concentration(#/cm )
Figure 3. Cloud Albedo vs. Cloud Droplet Number Concentration
in cm"-* fora cloud of liquid water content ol 0.3 "ii, •'
.Ul(( ""(V Im Ihi.'L n... .
and 2(XJm thickness
469
-------
Testing Of A Triple-Path Denuder Designed For Quantitative
Sulfate And Nitrate Measurements
Briant L. Davis, Yun Deng, Darcy J. Anderson
L, Ronald Johnson, and Andrew G. Detwiler
South Dakota School of Mines and Technology
Rapid City, 3D 57701
Laura L. Hodson
Research Triangle Institute
Research Triangle Park, NC 27711
and
Joseph E. Sickles
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
A triple-path denuder (TPD) assembly has been tested at the Research Triangle Institute
and at the South Dakota School of Mines and Technology. Denuder coatings included oxalic acid
for removal of ammonia, and NaCI for capture of nitric acid; a third path was left uncoated.
Denuder cassettes contained teflon or acid-washed "quartz" filters, and contained either a blank
filter, or were loaded with solid reacting materials such as oxalic acid, NaCI, or MgO (at SDSM&T
only). Primary aerosol concentrations of (NHJSSO^ (17 and 79 /jg m~a), NH^NO3 (10 /jg m"3), and
H^SO,, (10-20 fig m'3) and of gaseous NHa (12-26 /jg nr3) and HNO3 (7-131 pg nT3) were
introduced into the system. Monitoring of generated species was conducted using ion
chromatography, and analysis of reactions on denuder coatings and filter loads was completed by
x-ray diffraction. Results indicate that losses of HN03 were too large to permit using the TPD for
ambient nitric acid reaction product measurements. However, primary sulfate and possibly nitrate
aerosol measurements may be completed with this system. Lower limits of detection (LLD) for
products in the filter loads for this system vary from less than 0.1% by weight, to as high as 24%
for extremely light filter loadings.
470
-------
1. INTRODUCTION
A denuder instrument for stripping of unwanted gaseous species and for protection of
collected particulate from artifact formation was designed and given preliminary testing at the
South Dakota School of Mines and Technology (Davis ef a/.1). Additional testing at the Research
Triangle Institute (RTI) was completed in the summer of 1991 (Hodson, et a/.2) for the purpose of
introducing controlled atmospheres containing particulate ammonium sulfate (Mascagnite -
(NHJ.,SO,, - 'MS'), ammonium nitrate (NHJvlOa - "AN"), particulate H2SO^, and gaseous HNO3,
into the denuder system. The purpose of the RTI test was to generate limits of detection data and
observe artifact reactions for collected species frequently observed in heavily polluted urban
atmospheres sampled at near-ambient concentration levels. The present paper describes a
portion of these tests; those for which ammonium sulfate and ammonium nitrate were generated
for direct collection on teflon membrane filters.
2. DENUDER DESCRIPTION AND EXPERIMENTAL DESIGN
The details of the triple-path denuder (TPD) construction have been presented by Davis et
a/.1. The instrument was modified by outfitting both paths A and C with Possanzini concentric tube
denuders (Possanzini, et al.3); path B consisted of two concentric tubes of equivalent size, but
untreated. The exterior surface of the interior tube and interior surface of the outer tube of the
Path A denuder were coated with oxalic acid for the tests described here. Path C was not used for
the sulfate and nitrate particulate tests.
Teflon filters (PTFE, 37-mm diam.,2-/jm pore) were used exclusively for collections of the
nitrate and sulfate aerosols. The loads were concentrated by means of a machined teflon reducer
into a 2.5-cm diameter central area of each filter. Blank and loaded filters were weighed at RTI (MS
set) and at both SDSM&T and RTI (AN set). For XRD analysis each filter was cemented onto the
2.5-cm diameter spinning pedestal of the x-ray diffractometer. The diffractometer used was a
Philips unit with vertical goniometer, graphite monochromator, and automatic divergence slits.
Scans were made with a CuKa target operated at 40 kV and 20 mA; scan parameters were 0.02-
degree step and 4.12-second dwell.
Figure 1 illustrates the configuration of the aerosol generation tent" and TPD inlet. The
tent was constructed of Tedlar, and was supported on a teflon tubing frame. It contained a mixing
chamber to combine gaseous and aerosol samples prior to entry into the PM10 TPD sampling
head. For quality assurance (QA) purposes RTI located filter pack cassettes on two sides of the
tent; the SDSM&T QA filter pack cassette (CC) was located just above RTI filter cassette B. The
aerosols of MS and AN were generated with a TSI Model 3076 atomizer. Aerosol particles were
passed through a Kr 85 particle charge neutralizer prior to entrance into the dilution manifold (not
shown in Fig. 1).
3. QUALTTY ASSURANCE
QA measures were taken at all stages of the testing. The general features consisted of:
1. Preliminary XRD scans of blank filters and loaded filters (at simulated ambient
concentrations) prior to RTI testing.
2. Novaculite standard QA scans to verify instrument performance during the XRD data
analysis stage of the project.
471
-------
3. Filter weight checks at both the SDSM&T laboratories and at RTI during TPD testing
4. Ion chromatographic (1C) analysis of sampled NO3" and SCV aerosols, of filter blanks,
and of collections at cassette points A and B; and 1C solution checks needed for determination of
RTI laboratory precision.
A study of the repeatability of the electronic balances used demonstrate that the precision
for a given reading may not be better that about 50 micrograms (10 fjg cm'= for the load coverage
here), although the balance scales were readable to 10 pg at SDSM&T and 1 jjg at RTI.
ATOMEER INLET
DBT AM INLET
I MKWO HEAD
PERMEATION TUBE INLET
PM-10 SAMPLING HEAD
n=
EXHAUST
1_
0 30 CM
APPHOZBSATE SCALE
TEBLAR ENCLOSURE
CC
TRIPLE-PATH DENUDES
Fig. 1. Physical Arrangement of Triple-Path Denuder Testing
472
-------
4. RESULTS
4.1 Mascagnite (NHJ,,SO,
Table 1 presents the concentration and flow data for the MS and AN aerosols provided to
the TPD according to RTI specifications. The initial MS series target concentration of 27 jug m~3
was not achieved, apparently because of excessive losses through the nebulizer system; therefore
this target value was increased to the 80 /jg nr3 given in Table 1. The flow rate at the control
cassette ranged from 0.2 - 0.25 m3 hr1 whereas the flow rate for the denuder paths A and B
ranged from 1.6 to 1.8 m3 hr1.
TABLE 1. Delivery and LLD Parameters for Control (CC) and Denuder Samples
PARAMETER
Source Concentration
Target Concentration
Actual Concentration
Control/Path Data
Control (CC)
Path A
Path B
Control (CC)
Path A
Path B
Control (CC)
Path A
Path B
AMMONIUM SULFATE (MS)
0.02M (NHJ2SO^
80 fjg m"3
79.2 pg m"3
Time
1 Hr
1 Hr
1 Hr
2Hr
2Hr
2Hr
4Hr
4Hr
4Hr
Mass
fjg cm-2
7.94
30.4
17.7
7.33
96.8
30.1
17.1
184.2
50.1
LLD
Percent
—
22.3
3.1
._._
18.5
1.5
._.
5.4
0.9
AMMONIUM NITRATE (AN)
0.015MNH^N03
+ 3000 ng min'1 NH3
27 /jg m'3
10.3f/g rrr3
Time
1 Hr
1 Hr
1 Hr
3Hr
3Hr
3Hr
6 Hr
6 Hr
6Hr
Mass
fjg cm'2
BBS1
BBS
BBS
7.74
18.9
BBS
0.2
56.8
22.6
LLD
Percent
—
5.6
—
_.__
5.5
--
—
2.2
—
1 Below Balance Sensitivity (Mass)
Collection was made for 1-, 2-, and 4-hour periods through paths A and B. The ratio of volumes
sampled at the path A and B cassettes to that of the controll cassette (CC) is generally between
6.5 and 9; When this ratio is taken into account the contol cassette mass collected equivalent to
the path A and B collection under identical flow rates was usually significantly greater for the path
B cassette, but often less than that for the path A cassette. For example, for the MS 2-hour test
the equivalent CC mass was 60.1 and 64.1 /jg cm'2 , respectively, for paths A and B, compared to
the actual measured 96.8 and 30.1 (jg crrra for paths A and B cassettes. This suggests either
473
-------
excessively large losses of aerosol along the route to the path B filter or an additional source of
mass arriving at the path A filter. Evidence from the type of compounds observed suggests that
particles from the path A denuder coating were dislodged and collected on the path A cassette
substrate. This point will be discussed further below.
In most cases sufficient collection of MS aerosol at the path A and B cassettes was
obtained to determine the limits of detection for the XRD scanning conditions attempted. The
lower limits of detection (LLD) for a component can be determined by several methods (Davis").
For a single component, such as required in these analyses, the LLD is given by the simple 3-a
criterion
Z,LD=300JV.l— -^ (1)
where !„„•* is the integrated background intensity measured beneath the component integrated
peak intensity I,", and for these samples, W, = 1. The LLD here is given as weight percent of the
specific mass (Mass of Table 1) for each filter collection. Figure 2 presents the LLD values and
intensity variations of the sum of the 102,111 and 103 peaks versus atmosphere sampled and
specific mass for the three test periods.
The initial XRD scans of these collections were made on 9 Sept. 1991, shortly after
completion of the RTI field testing in mid-August. The sample collected for the 4-hour path A run
was scanned again on 17 April, 1992 and finally again on 22 Feb. 1993. This final "aged1 sample
contained 44.7% mascagnite and 55.3% letovicite as determined from the XRD analysis, using
reference intensity methods of Davis ef a/.1. The first scan demonstrated poorly crystalline MS
along with unidentified peaks which were first thought to be an oxalate or oxalate hydrate, possibly
ammoniated, that resulted from spalling from the oxalic acid hydrate coatings of the path A
denuder. Subsequent scans revealed an increase of the MS 102 peak from 5 to 50 to 80 cps
along with sharpening up of all peaks sufficiently to recognize the formation of letovicite,
(NHJ3H(SOJ=, an acid species of ammonium sulfate. The presence of oxalates have never been
confirmed on any of these patterns; The presence of letovicite, however, strongly implies
contamination of the path A mascagnite collection and reaction according to
4 (NHt) 2 (S04) + C2H204-2#20 - 2 (NH4)
MS OAH LV AOH (2)
and where the ammonium oxalate, a transient phase, is presumed to further dissociate to gaseous
components:
(NHt) 2C204-H20 — '2C02t + 2Mf3T + 2H2Ot (3>
We have also discovered that grinding a mixture of mascagnite and oxalic acid hydrate
quickly converts the mascagnite to letovicite. Such a conclusion might even suggest that oxalic
acid hydrate is not an appropriate choice for denuder coatings in contact with freshly generated
ammonium-bearing test aerosols because of possibility of removal of stabilizing ammonia from the
sampled atmosphere.
474
-------
AMMONIUM SULFATE AND AMMONIUM NITRATE ON TEFLON (FTTO
PATHASZSIES
I, (VOL)/ (NH4)aSO, (MASS)
/ '
0 2 4 « 8 10 12 VOL
TBST VOLUME Cm ») AND SPECIFIC MASS COLLECTED (lie/on* )
Figure 2. Diffraction Intensity for Selected Peaks vs. Mass and Volume
and LLD values for Ammonium Sulfate and Nitrate Aerosols.
4.2 Ammonium Nitrate NH^NO3
Table 1 and Figure 2 present the comparable data from these tests for ammonium nitrate
(AN). An attempt was made to maintain AN aerosol stability with an excess of NH3 gas during
loading, even though most of the excess would be lost to the oxalic acid denuder coating for path
A collections. The QA checks of these runs showed a high relative standard deviation for the AN
atmosphere concentration of 34.3%. The final filter weighings showed little or no mass
accumulation above the sensitivity of the balance. However, weak but distinct peaks for the
strongest diffraction spacing of AN were recognizable on most of the filters. LLD values were thus
obtained for the A-path collections. In contrast to the MS series of tests, both ammonium
hydrogen oxalate hydrate (AHOH), NH^HC2O^»H=O, and ammonium oxalate hydrate (AOH),
(NH JaCaO^«H2O, were found to exist with the expected NH,,NO3. Analysis of the path A filters for
these three components is presented in Table 2. For purposes of correction of intensities for
transparency and matrix effects for the 1-hour collection, a specific mass level of 200 jjg cm"2 was
assumed since no weight could be detected from the balance measurements (BBS, Table 1).
TABLE 2. XRD PHASE ANALYSIS OF NH^NO3 SERIES COLLECTIONS
FILTER COLLECTION
PERIOD
1-HOUR
3-HOUR
6-HOUR
COMPONENT WEIGHT PERCENT AND VARIANCE ERRORS
AN
19.9 ± 2.5
26.4 ± 3.1
12.7 ± 1.6
AHOH
48.6 ± 4.7
55.2 ± 4.4
78.7 ± 2.6
AOH
31.5 ± 4.4
18.4 ± 3.6
8.6 ± 2.0
475
-------
These analyses demonstrate conclusively that contamination from the denuder coatings
exists on the teflon filter collections. The one B-path collection showing sufficient mass for XRD
analysis revealed only the AN phase peak present, as expected.
5. DISCUSSION AND CONCLUSIONS
The tests of the triple-path denuder described here have great value in providing direction
for future work with this type of system. First, comparison of control and sample cassette loads
strongly suggests losses of paniculate that may result both from surface scavaging through the
system conduit parts and control losses from incomplete gas-to-particle conversion. Loads
obtained at near-ambient concentrations proved marginal for XRD analysis; masses collected on
many of the filters were below the measurement range of the electronic balances used. However,
XRD patterns, obtained under slow scanning conditions, can still be obtained so that LLD values
could be calculated. Considering the 10-80 jjg m~3 concentrations used here it is apparent that
good XRD patterns can be achieved in an ambient sampling environment only after several hours
continuous operation of the system. Use of oxalic acid as a denuder coating for NH3 removal is
not recommended because of the likely losses from denuder walls to collection cassette.
6. ACKNOWLEDGEMENTS
This research was supported in part by the National Science Foundation, under Grant ATM-
9024485. The authors acknowledge assistance in various aspects of this work from Dr. Robert
Looyenga and Mr. Justin Tomac of the South Dakota School of Mines and Technology.
7. REFERENCES
1. B.L Davis, L.R. Johnson, B.J. Johnson, and R.J. Hammer, "A Triple-Path Denuder
Instrument for Ambient Paniculate Sampling and Analysis", J. Atmos. Oceanic. Tech., 5, 5-15 (1988).
2. L.L Hodson, W.C. Eaton, E.D. Estes, A.R. Turner, and J.E. Sickles II, 1992, "Generation of
Gases and Aerosols for Testing of the South Dakota School of Mines and Technology's Triple Path
Denuder". Research Triangle Institute Report RTI/5064/-01D.
3. M. Possanzini, A. Febo, and A. Liberti, 1983, "New Design of a High-Performance Denuder
for the Sampling of Atmospheric Pollutants'. Atmos. Environ., 17, 2605-2610.
4. B.L. Davis, B.L., The Estimation of Limits of Detection in RIM Quantitative X-Ray Diffraction
Analysis", Adv. X-ray Anal., 31,317-323(1988).
476
-------
Session 12
Risk Assessment
-------
Assessment of Ambient-Air Quality at Landfills 8 and 10 of
Wright-Patterson Air Force Base, Ohio
R.R. Mannar
Senior Engineer
Engineering-Science, Inc.
710 S. Illinois Avenue
Suite F-103
Oak Ridge, Tennessee 37830
Dr. J.R. Tucker
Engineering-Science, Inc.
One Harrison Park
401 Harrison Oaks Boulevard
Suite 305
Gary, North Carolina 27513
and
Teresa S. Finke
645 ABW/EME
5490 Pearson Road
Wright-Patterson Air Force Base, Ohio 45433
ABSTRACT
Five rounds of ambient-air sampling were conducted between July and October 1991 at Landfills 8 and 10 of
Wright-Patterson Air Force Base (WPAFB), Ohio. The primary air monitoring objectives were to determine whether
the landfills are emitting contaminants into ambient air and to assess the environmental impact based on the existing
National Ambient Air Quality Standards (NAAQSs) and the State of Ohio guidelines. WPAFB is located in
southwestern Ohio, east of Dayton (Figure 1). Nine sampling sites, including an upwind and a collocated sampling
site, were used during each round of sampling (Figure 2). Participate matter below 10 microns (PM-10) and selected
metals, volatile organic compounds (VOCs), and semi-volatile organic compounds (SVOCs) were measured at each
sampling site. All the sampling equipment has been designated as the reference method or equivalent by the U.S.
Environmental Protection Agency (USEPA).
Seventeen VOCs, 10 SVOCs, and 10 metals were detected in five rounds of sampling. Of 17 VOCs,
methane was detected at all sampling sites in each round; the highest concentration of methane was detected at
Landfill 10. Relatively high concentrations of acetone and benzene were measured at several sampling sites. Among
10 SVOCs, fluoranthene, naphthalene, and phenanthrene were detected at almost all sampling sites in all sampling
rounds. The concentrations of SVOCs detected at upwind and downwind sites were of the same order of magnitude.
Among the 10 metals, copper was detected at virtually all sampling sites.
In spite of wide variation in the particulate concentrations from round to round, the PM-10 concentration
never exceeded the NAAQS of 50 micrograms per cubic meter (ng/m3) of air.
Based on the results of five rounds of air sampling, no deterioration of ambient-air quality due to emission of
landfill gases was observed.
479
-------
INTRODUCTION
Landfills 8 and 10 are among the several sites targeted for remediation at WPAFB. Both landfills are
located in the southern portion of WPAFB (Figure 1). Landfill 8 was operated from 1947 until the early 1970s,
whereas Landfill 10 was operated from 1965 to early 1970s. Both landfills were reportedly used for the disposal of
general refuse consisting of oily wastes, organic and inorganic wastes, hospital waste, toxic and hazardous waste, and
acid neutralization. Little is known about the waste quantities and disposal dates. At both landfills, waste was
disposed of in a trench-and-cover operation. Both landfills were reportedly covered with clay or soil caps of varying
thickness. The cap over Landfill 8 is moderately vegetated with grasses and varies in thickness from 3 to 12 feet,
averaging typically 8 to 10 feet over the majority of the landfill. The cap over Landfill 10 is sparsely vegetated with
grasses and varies in thickness from 1 to 8 feet, with a typical average of 2 feet over most of the landfill. The
respective elevations of Landfill 8, Landfill 10, and the valley between the landfills are 945, 910, and 860 feet above
the National Geodetic Vertical Datum (NGVD) of 1929.
The natural topography of the area has been extensively modified through waste disposal at both landfills, by
the construction of personnel housing in the area, and the construction of bunkers for ordnance storage at numerous
locations in the valley between the landfills. The bunkers have since been removed. When residential housing was
being built around the landfills from 1971 to 1973, the surface of Landfill 8 was developed into a recreational area.
The valley between the landfills was closed to vehicular traffic and was developed into a wooded park with footpath
access to both landfills.
Parts of both landfill areas were used for recreational purposes from the mid-1970s to early 1985 when parts
of those landfill areas were fenced and posted off-limits. All of the landfill areas and the intervening valley were
fenced in 1986 and warning signs were placed.
Since WPAFB was included on the National Priorities List (NPL), Remedial Investigations (RIs) and
Feasibility Studies (FSs) are being conducted under the guidance of U.S. Environmental Protection Agency (USEPA)
Region V and the Ohio Environmental Protection Agency (OEPA).
Several investigative studies have been performed since 1982 at both landfill sites. Engineering-Science, Inc.,
(ES) undertook the comprehensive RI of Landfills 8 and 10 in the beginning of December 1990. Work was
performed for the U.S. Air Force under contract to the Department of Energy and managed by Martin Marietta
Energy Systems' Hazardous Waste Remedial Actions Program (HAZWRAP). Ambient-air monitoring was a part of
the comprehensive RI program. The description of the sampling methods and the discussion of the sampling results,
which are presented in this paper, are documented in detail in the draft RI report submitted to WPAFB1.
DESCRIPTION OF AMBIENT-AIR SAMPLING
Nine sampling sites, including one upwind and one collocated group of samplers, were chosen on and around
landfills during each round of sampling. Locations of sampling sites varied for each round based on the wind
direction (see Figure 2). The wind direction during Round 1 and Round 5 was predominantly out of the southwest
and during the remaining three rounds it was out of the northeast. The upwind sampling site was located in the open
area across from the Officer's Club Annex Building 189 for Rounds 1 and 5, and during Rounds 2, 3, and 4 it was
placed near the Twin Base Golf Course clubhouse. The three downwind sampling sites (Sites 2, 3, and 4) were
located on Landfill 8 during all five rounds. The first of the three sites (Site 3) was placed in what was determined to
be the center of the downwind boundary of the landfill. The remaining two sites were located approximately 45
degrees northeast and 45 degrees southwest of the central site. One sampling site (Site 8) was located in the open
space of the residential area between Landfills 8 and 10. Three downwind sampling sites (Sites 4, 5, and 6) were
placed on Landfill 10 using logic similar to that used for locating sampling sites on Landfill 8. The remaining
sampling site (Site 9) was collocated with Site 6 on Landfill 10 in Rounds 1, 4, and 5, and with Site 3 on Landfill 8
for Rounds 2 and 3. The downwind sampling sites of both lanHfilk were moved from round to round so that the
samplers were exposed to air after it had traveled across the landfill.
The ambient-air sampling program also consisted of the measurement of several meteorological parameters.
Due to the complex terrain of Landfills 8 and 10, one meteorological station was located on each landfill (Figure 2).
The key parameters measured were wind direction, ambient temperature, barometric pressure, and precipitation.
Wind direction was used to locate sampling sites, temperature and pressure were used to correct the measured
480
-------
sample volume to standard conditions, and precipitation was used to determine when to start or stop the sampling
process. Ambient-air sampling was preceded by at least 24 hours without precipitation, and sampling was stopped
when the rainfall exceeded the measurable quantity of 0.01 indies during sampling. Meteorological data were
collected concurrent with ambient-air sampling during each sampling round.
Paniculate matter below 10 microns (PM-10), 26 VOCs, 32 SVOCs, and 15 metals were measured at each
sampling site in each round of sampling. The parameters were selected based on the historical documentation on
landfill progression and the results of previous investigations. Maximum concentrations of the detected contaminants
at each sampling site during the five rounds of sampling are given in Table 1.
Volatile Organic Compounds
The Model 911A portable canister sampler, manufactured by Xontech, Inc., Van Nuys, California, was used
for collecting VOC samples.
Sampling of VOCs including methane and certain reduced sulfur compounds was performed by pumping
controlled amounts of ambient air into a pre-evacuated 6-liter stainless steel canister. The interior surfaces of the
canister were treated using a special method known as SUMMA passivation, in which pure chrome nickel oxide is
formed on the inside wall of the canister. The inlet of the sampler was placed approximately 2 meters above the
ground surface and about 10 liters of ambient air was collected in each canister over a period of 24 hours. At the
end of the sampling event, valves on each canister were closed, an identification tag was attached, and the canisters
were shipped by an overnight carrier to the laboratory for analysis.
As sampling of reduced sulfur compounds and ketones in a SUMMA canister has not been well established,
stability studies were performed by the laboratory for the three reduced sulfur compounds (hydrogen sulfide, carbon
disulfide, and dimethyl sulfide) and three ketones (acetone, butanone, and 4-methyl-2-pentanone). The purpose of
this study was to determine whether these six compounds, which are among the 26 VOCs selected, are stable in
SUMMA canisters beyond the 24-hour sampling period. Except for hydrogen sulfide, the remaining five compounds
were fairly stable. Therefore, hydrogen sulfide was sampled by a field instrument, the Jerome 631 Gold Film
Analyzer, capable of measuring from 1 part per billion (ppb) to 50 parts per million (ppm).
Semi-volatile Organic Compounds
The Model GPS1 PUF sampler, manufactured by General Metal Works, Village of Cleves, Ohio, was used
for collecting SVOC samples from ambient air.
SVOCs including polychlorinated biphenyls (PCBs) were sampled with a combination of a quartz filter and
adsorbent cartridge and a high volume sampler. The adsorbent cartridge consisted of XAD-2 resin placed between
two polyurethane foam (PUF) pieces in a glass module. The quartz filter was used to separate the suspended
paniculate matter before the air stream was drawn into the adsorbent cartridge. The cartridges were kept
refrigerated in their original containers until needed. Approximately 273 cubic meters (m3) of air were drawn
through the filter and cartridge in a 24-hour sampling period. The air inlet of the sampler was placed approximately
2 meters above the ground surface.
At the conclusion of the 24-hour sampling event, the XAD-PUF cartridges were packed in their respective
containers, placed on ice, and shipped by an overnight carrier to the laboratory.
Paniculate Matter and Metals
The Model G 1200 High Volume PM-10 Sampler was used to sample PM-10 in the ambient air. The
paniculate matter was also analyzed for the metals. This high-volume sampler is manufactured by General Metal
Works, Village of Cleves, Ohio.
The PM-10 was collected on a glass fiber filter with a high-volume electric blower below the filter.
Approximately 1,600 m3 of air was drawn through the filter in a 24-hour period. Similar to VOC and SVOC
sampling equipment, the PM-10 sampler inlet was also set about 2 meters above the ground surface.
481
-------
Following the completion of the sampling event, the fiberglass filters were removed from the samplers,
folded in the center, and placed in individual, scalable plastic bags. The plastic bags were then shipped to the
laboratory by an overnight carrier. Filters were weighed to calculate the PM-10 concentration and the particulate
matter was subsequently analyzed for metals.
DISCUSSION OF RESULTS
To assess the extent of ambient-air contamination, the following standards were compared with the detected
contaminant concentration in each round:
National Ambient Air Quality Standards (NAAQSs) for lead and PM-10;
Maximum acceptable ground-level concentrations (MAGLCs) calculated based on the OEPA Guidelines
for Review of New Sources of Air Toxic Emissions;
National Institute of Occupational Safety and Health (NIOSH) recommended values for polynuclear
aromatic hydrocarbons (PAHs); and
Lower explosive limit (LEL) for methane.
The NAAQSs for lead and PM-10 are 1.5 ug/m3 (based on quarterly average) and 50 ug/m3 (based on
annual average) or 150 |ig/m3 (based on 24-hour average), respectively. Except for lead, none of the detected
contaminants are regulated by NAAQSs.
The MAGLC values are calculated based on the following OEPA procedure:
Determine if a threshold limit value (TLV) developed by the American Conference of Governmental
Industrial Hygienists (ACGIH) exists for the specific contaminant that is emitted from the source.
Then divide TLV by 70 to adjust the standard from the working population to general public.
The calculated MAGLCs for the detected contaminants are given in Table 1.
The 10-hour time-weighted average for PAHs recommended by NIOSH is 100 u.g/m3. Among the detected
SVOCs, anthracene, fluoranthene, naphthalene, phenanthrene, and pyrene are classified as PAHs.
The LEL of a combustible gas is defined as the lowest concentration of gas that will combust hi the presence
of an ignition source and air or oxygen. The LEL for methane is 33 x 106 ug/m3 or 50,000 ppm.
Volatile Organic Compounds
A total of 17 VOCs were detected in the ambient air. Most of the VOCs were found both in upwind and
downwind samples. Among 17 VOCs, only benzene and acetone were measured hi excess of their respective
MAGLCs. Benzene was detected in excess of its MAGLC of 4.3 ng/m3 in upwind and downwind samples, whereas
acetone was found hi excess of its MAGLC of 25,429 ug/m3 at several downwind sites. Acetone was not detected hi
any upwind samples. Toluene and xylene were frequently found in the same range in upwind and downwind samples.
Chlorobenzene and tetrachloroethene (PCE) were each detected once in the upwind sample, and 1,1,2-
trichloroethane (1,1,2-TCA) was also detected once at one downwind site. Methane was detected in all samples in all
sampling rounds. Relatively higher levels of methane were detected in air samples from downwind sites on the
northern portion of Landfill 10. No MAGLC exists for methane. However, the highest detected concentration of
methane is less than 0.01 percent of the LEL for methane. The remaining VOCs were detected hi a limited number
of samples from upwind and downwind sites.
Semi-volatile Organic Compounds
Ten SVOCs were detected hi the ambient-air samples. Concentrations of detected SVOCs in upwind and
downwind samples were generally of the same order of magnitude. Fluoranthene, naphthalene, phenanthrene,
pyrene, and 4-methylphenol were found hi most of the samples. Except for 4-methylphenol, the remaining four are
PAHs. Major sources of PAHs are auto exhaust and emissions from fossil power plants and incinerators.
Diethylphthalate and 2-methylphenol were detected less frequently and the remaining three SVOCs were rarely
-------
detected. No NAAQSs exist for SVOCs. However, N1OSH recommends a workplace standard for coal-tar products
of 100 |ig/m3 based on a 10-hour time-weighted average. PAHs are the principal components of coal-tar products.
The combined concentration of PAHs at each sampling site never exceeded the NIOSH-recommended value. Also,
available MAGLCs were used to compare the SVOC concentrations. None of the respective MAGLCs were
exceeded.
Paniculate Matter and Metals
The maximum PM-10 concentration in ambient air at each sampling site was measured in late August 1991.
This sampling round was preceded by several days of no rain and high daytime temperatures. However, none of the
measured PM-10 concentrations exceeded either the 24-hour NAAQS value of 150 ng/m3 or the annual value of
50 ng/m3.
Ten metals were detected in the ambient-air samples. Concentrations of metals detected in upwind and
downwind samples were generally similar. Copper was detected in almost all samples. Arsenic, iron, and manganese
were found in more than 50 percent of the samples. Beryllium, lead, mercury, and zinc were detected in a few
samples and chromium and antimony were rarely detected. Except for lead, none of the detected metals are
regulated by the NAAQSs. The highest detected concentration of lead is more than an order of magnitude smaller
than the NAAQS of 1.5 (ig/m3. For the remaining nine detected metals, MAGLCs were used for comparison. None
exceed their respective MAGLCs.
CONCLUSIONS
Relatively high levels of methane and acetone were measured at the sampling sites located on both landfills,
indicating the possible emission of landfill gases into ambient air through the landfill covers. However, based on the
NAAQSs and OEPA guidelines, no ambient-air contamination in the vicinity of Landfills 8 and 10 is evident.
Concentrations of most of the detected contaminants in upwind and downwind samples were generally of the same
order of magnitude. Benzene was detected in excess of its MAGLC in both upwind and downwind samples. The
presence of chlorobenzene, PCE, and benzene in the upwind sample implies the existence of another emission source
distinct from the landfills. Chlorobenzene and PCE were not detected in any of the downwind samples.
REFERENCES AND BIBLIOGRAPHY
1. Draft Off-Source Remedial Investigation Report for Landfills 8 and 10 at Wright-Patterson Air Force Base. Ohio.
Prepared by Engineering-Science, Inc., Oak Ridge, Tennessee, 6 October 1992; submitted by Hazardous Waste
Remedial Actions Program of Martin Marietta Energy Systems, Inc., Oak Ridge, Tennessee; submitted to Wright-
Patterson Air Force Base, 645 Air Base Wing, Office of Environmental Management, Wright-Patterson Ah- Force
Base, Ohio.
2. W.T. Winberry, Jr., N.T. Murphy and R.M. Riggin, Methods for the Determination of Toxic Organic Compounds
in Air. EPA Methods; Noyes Data Corporation, Park Ridge, New Jersey, 1990.
3. W.F. Martin, J.M. Lippitt and T.G. Prothero; Hazardous Waste Handbook for Health and Safety: Butterworths,
Stoneham, Massachusetts, 1987.
483
-------
Table 1. Maximum concentration of the contaminants detected during five rounds of ambient air sampling.
(Results are in pg/m3)
Analyte
Metals and Paniculate Matter
Antimony
Arsenic
Beryllium
Chromium
Copper
Iron
Mercury
Manganese
Lead
Zinc
Particulates (PM-10)
Volatile Organics
Acetone
2-Butanone
Benzene
Chlorobenzene
1 ,4-Dichlorobenzene
cis 1,2,-Dichloroethene
Ethylbenzene
Methylene chloride
Toluene
1,1,2-Trichlorethane
Trichloroethene
Tetrachloroethene
Xylene
Carbon disulfide
Dimethyl sulfide
Hydrogen sulfide
Methane
Sitel
(upwind)
---(1)
0.0019 1 (2)
0.0008
0.0058
0.0259
0.4498
0.0001
0.0116
0.025
0.027 ]
43.43
17.3 NJ
3.2 NJ
66.3 N
—
10.9 NJ
45.2 N
16.32 J
25.25 J
1.4
1,184 J
Landfill 8
Site 2
—
0.0019 J
0.0061
0.0198
0.2442 J
0.0104
0.0202
0.0304 J
46.42
38.73 J
16.6 J
—
3.00 J
13.06 J
—
49.0 N
17.3 NJ
46.14
1.4
1,381 J
Site 3
0.0029
0.0019
0.0006
—
0.0141
0.3026 J
0.0001
0.01
0.0163 J
0.0275 J
41.12
500 N (4)
39.62 J
17.06 J
16.2 J
11.32 J
12.48 J
5.09 J
1.4
1,184 J
Site 4
—
0.0019 J
—
—
0.0143
0.2934 J
0.0001
0.0110
0.0163 J
0.030 J
45.60
45,244 J
17.6 NJ
9.82 J
53.6 NJ
3.12 J
1.8
1,578 J
Landfill 10
SiteS
—
0.0026 J
—
0.0556
0.2916 J
0.0001
0.0097
0.0137
0.0255 J
44.74
235,985 J
37.96
12.1 NJ
10.45 J
43.3 NJ
1.4
1.776J
Site 6
0.0028 J
0.0006
—
0.0519
0.3274 J
0.0001
0.0111
0.0166 J
0.0284 J
43.24
3,810 J
7.7 NJ
11.92J
45.96
12.4 N
20.47 J
9.79
3.12 J
5.07 J
1.5
2,302 J
Site 7
0.002 J
0.0006
0.0370
0.3142 J
0.0001
0.0118
0.0156 J
0.0295 J
45.04
14,763 J
11.32J
25.07 J
21.16 J
58.77
68.64
4.33 J
1.6
3,025 J
SiteS
...
0.0019 J
0.0008
0.0314
0.1798 J
0.0001
0.0085
0.0182 J
0.0279 J
38.29
18.08 J
...
—
—
18.89 N
17.41 J
2.88 J
1.4
1,250 J
Site 9
(collocated
site)
—
0.002 J
0.0007
—
0.0865
0.2349
0.0001
0.0096
0.0185 J
0.0293 J
42.83
78,582 NJ
10.58 J
—
—
—
—
—
—
1.5
2,433 J
Calculated
MAGLC
7.1
2.9
0.03
0.7
2.9
3.3
0.14
14.3
1.5(3)
71.4
25,429
8,429
4.3
657
857
11,329
6,200
2,486
2,100
786
3,843
4,843
6,200
443
NA(5)
200
NA
-------
Table 1. Continued
Analyte
Semi-volatile Organics
Anthracene
Fluoranthene
2-Methylphenol
4-Methylphenol
Naphthalene
Phenanthrene
Pyrene
Benzoic Acid
Diethylphthalate
2,4-Dimethylphenol
Sitel
(upwind)
—
0.0092 N
0.0026 N
0.0126 N
0.124NJ
0.0332 NJ
0.0048 N
0.0058 N
Landfill 8
Site 2
—
0.0026 NJ
0.0092 N
0.102N
0.0142 NJ
0.0047 NJ
0.0015 NJ
SiteS
—
0.0031 N
—
0.0093 N
0.1 180 NJ
0.0170 N
0.0018 N
—
—
—
Site 4
—
0.0029 N
0.0162 N
0.148 NJ
0.0159 N
0.0022 J
0.177 NJ
0.0036 N
Landfill 10
SileS
—
0.0027 N
0.0137 N
0.148 NJ
0.0143 N
0.0016 NJ
Site 6
—
0.0034 N
0.0096 N
0.111 N
0.0199 N
0.0022 N
Site?
—
0.0035 N
1.88 N
6.800 NJ
0.151 NJ
0.0166 N
0.0021 N
0.0098 NJ
SiteS
—
0.0021 NJ
0.333N
1.22 NJ
0.1068 NJ
0.0124 N
0.0014 NJ
Site 9
(collocated
site)
0.0125 NJ
0.0031 N
—
0.0715 N
0.1180 NJ
0.0167 N
0.0019 N
0.0061 NJ
—
Calculated
MAGLC
2.9
2.9
NA
314
743
2.9
2.9
NA
71.4
NA
1. - - Indicates that the analyte was not detected.
2. J Indicates that the analyte was positively identified, but the value may not be accurate.
3. No MAGLC is calculated for lead. The given value of 1.5 ug/m3 is the NAAQS based on quarterly average.
4. N Indicates the presence of an analyte for which there is presumptive evidence to make a "tentative identification."
5. NA Indicates that MAGLC was not calculated, since a TLV has not been determined.
-------
Figure 1. Location of Landfills 8 and 10.
-------
FEET
EXPLANATION
© Upwind sampling site and identlfIcatlon.
3
© Dowiwfnd sampling site and identifrcation.
AMBIENT-AIR SAMPLING DATES:
Collocated sampling site.
MeteoroIogi caI stat i on.
Round 1 - 10 to 17 July 1991.
Round 2-01 to 09 August 1991.
Round 3-21 to 27 August 1991.
Round 4 - 16 to 24 September 1991.
Round 5 - 15 To 22 October 1991.
Figure 2. Location of ambient-air sampling sites during five rounds of sampling.
-------
The Use of Air Modeling in the Development of the Air Monitoring Program
for the Dubose Oil Company Superfund Site Remedial Action
Thomas A. Peel, Ph.D., Scott E. Rowden, C.I.H., and Steven Ratzlaff
Engineering-Science, Inc.
57 Executive Park South, NE, Suite 500
Atlanta, Georgia 30329
Air modeling was used to aid in the development of the ambient air monitoring program at the Dubose Oil Products
Company (DOPC) Site. During the remedial action, ambient air monitoring will be performed in order to document Dial
contaminant emissions are not in excess of health-based criteria.
The DOPC Site occupies approximately 20 acres of land in Escambia County, Florida. The site was operated as a
storage, treatment, and recycling facility from January 1979 to November 1981. The remedial investigation identified
volatile organics, metals, polynuclcar aromatic hydrocarbons and pentachlorophenol as contaminants of concern. The
remedial action at the site will consist of contaminated soil excavation and on-site treatment by composling/windrowing.
Site contaminants were ranked based on their relative inhalation risks and those posing the greatest threat were
selected for air modeling. The emission rates of the selected compounds were estimated and dispersion modeling was
performed using the Industrial Source Complex Dispersion Model. The modeling results were used to identify the site
contaminants most likely to exceed health based levels during the remedial action and to identify the most appropriate
monitoring locations.
This approach may also prove useful in the development of air monitoring programs at other Superfund sites.
INTRODUCTION
Purpose
Air monitoring is required during the remedial action at Superfund sites to document that site workers and nearby
residents arc not exposed to site contaminants at concentrations above health based criteria. While ambient air monitoring
is necessary to protect public health, the cost of these programs may be significant The cost of air monitoring is
dependent on the following factors:
number of compounds being monitored;
number of monitoring locations; and
frequency of monitoring.
Air modeling was used on this project to identify the most appropriate compounds for inclusion in the air monitoring
program and to select the best monitoring locations.
Site Description
The Dubose Oil Products Company Site occupies approximately 20 acres of land in Escambia County, and is located
approximately two miles west of Cantonment, Florida. The site currently consists of an open-sided barn, a soil
containment vault, a sump catchment pond, two surface water ponds and an area where soil was excavated before being
placed in the containment vault.
The DOPC site was operated from January 1979 to November 1981 as a storage, treatment and recycling facility for
handling waste oils, petroleum refining wastes, wood treatment processing waste, paint wastes, spent solvents, and spent
iron/steel pickle liquors. Wastes reportedly were thermally treated and/or phase separated.
An emergency response action was undertaken by Florida Department of Environmental Regulations (FDER) during
1984 and 1985 which consisted of constructing an on-site vault for storing contaminated soils for future treatment.
Site Contaminants
Analyses of environmental samples collected during the Remedial Investigation indicated that the extent of
contamination at the site was limited. Contaminants detected at the site include polynuclear aromatic hydrocarbons,
chlorinated and non-chlorinated volatile organics, and pentachlorophenol. Trace metals were not detected in site soils and
groundwater at concentrations greater than those of uncontaminated soil and waters. Contaminants were not detected in air
samples collected at the site.
The highest levels of contaminants were detected in the soil containment vault. Soil samples collected from the vault
indicate that contamination is stratified with the highest concentration of volatile and semivolaUle compounds present at
25-30 feet below the lop of the vault. Volatile organic concentrations in the vault ranged from 22 to 38,270 ng/kg.
488
-------
Polynuclear aromatic hydrocarbons were delected at concentrations ranging from 578 to 122,400 fig/kg and
pentachlorophenol was delected at concentrations ranging from 58 to 51,000 Jig/kg.
Remedial Action
The remedial action at me Dubose Oil Products Company site will consist of the excavation of soils from the vault
and contaminated hot spots and biological treatment of the contaminated soils by composting/windrowing.
Air Monitoring Objectives
During the remedial action, the potential exists for contaminants to become airborne and be transported through the
air to residents on or near the sile. Because of the potential for contaminant releases to the air, the decision was made to
develop an air monitoring program to identify real-time releases of contaminants and use air samples collected for off-silc
analysis to document air quality at the sile.
AIR MODELING PROGRAM
Air modeling was used to estimate expected contaminant concentrations around the site during the remedial action.
These estimated airborne contaminant concentrations were men used to assess the potential health impacts to nearby
residents and to select monitoring locations, target analytes, and monitoring frequency.
Selection of Site Contaminants For Modeling
The inclusion of all sile contaminants in the air modeling study would have significantly increased the project cost
and extended Ihe schedule. The potential airborne hazard posed by a sile contaminant depends upon its potential for
airborne release (i.e., volatility), toxicity, and concentration on sile. A methodology was developed to select contaminants
for inclusion into Ihe modeling program based on their potential inhalation risks. The approach differed slightly for
volatile and non-volatile compounds as described below.
Volatile Compounds
The most important release mechanism for volatiles is vaporization. A Vapor Hazard Index can be calculated as an
indicator of the relative hazard posed by specific sile contaminants. The Vapor Hazard Index is calculated (for a given
chemical) by multiplying Ihe highest concentration found on site by the vapor pressure of that chemical and dividing by the
OSHA Permissible Exposure Limit (PEL) for that chemical compound. By definition a PEL for a specific chemical
compound is that airborne concentration below which it is believed that nearly all workers may be repeatedly exposed
without adverse health affects. The calculated vapor hazard index values for site contaminants were then used to rank the
volatile compounds by their relative hazard (i.e., from volatile compounds presenting Ihe greatest relative hazard to
compounds presenting the least hazard).
Semlvolatlle and Nonvolatile Compounds
These compounds are released from site soils and wastes primarily due to disturbances of site soils and wastes thai
may occur during waste excavation, handling, transport and remediation. For screening purposes, the two factors which
should be considered when evaluating the inhalation hazard of these compounds are concentrations in site soils/wastes and
inhalation toxicity. A hazard index was also calculated for these compounds by dividing the PEL for the compound into its
maximum on-site concentration. The relative hazard index ranking of these compounds was also used to select specific
compounds for subsequent air modeling.
Contaminant Selection Procedure
Based on Ihe relative inhalation hazard indices calculated for site contaminants, five volatile, ten semivolatile and
five metals were chosen for inclusion in Ihe air modeling evaluation. The selected site contaminants were as follows:
• Volatile Organic Compounds benzene, melhylene chloride, trichloroethene, toluene, and
xylene.
• Semlvolatlle/NonvolatUe Compounds benzo(a)anthracene, benzo(b)fluoran(hene,
benzoQOfluoranihene. chrysene, pentachlorophenol,
anthracene, fluoranlhene, fluorene, phenanlhrene, and
pyrene.
• Metals arsenic, beryllium, lead, barium, and vanadium
In general the highest ranking compounds were chosen from each group. However, several exceptions were made.
For example, acetone, elhylbcnzene, aluminum, and iron were deleted from air modeling due to meir low toxicity.
489
-------
Air Modeling Approach
Following the selection of lite chemicals of concern, air dispersion modeling was performed. This modeling involved
a two step process: (1) estimation of air emission rates for the chemicals of concern; and (2) dispersion modeling of
estimated emissions to determine airborne concentrations at potential receptor locations.
Contaminant Emission Estimates
Two different estimated emission rates, emissions caused by volatilization and emissions caused by wind-blown dust,
were calculated for each remedial activity at the DOPC site. Four remedial activities were selected to represent the
primary sources of air emissions. These four sources are described as follows:
• excavation of the vault;
• loading of soil (onto dump truck from front-end loader);
unloading of soil (from dump truck to bioremediation cell); and
• bioremediation.
In order to calculate emission rates from the four sources for the chemicals of concern, a conceptual model was
developed to depict die planned remedial activities at the site.
Dispersion Modeling
The Industrial Source Complex (ISC) Dispersion Model was used to estimate contaminant concentrations at potential
receptor locations around the site. The ISC is actually a set of two computer programs: the Industrial Source Complex
Short-Term Model (ISCST) and the Industrial Source Complex Long-Term Model (ISCLT). These models were used to
calculate contaminant concentrations and deposition values at specified receptor locations for various averaging periods.
The ISCST model calculates concentration or deposition values for a variety of averaging periods including 1,4, 8, 12,
and 24 hours as well as 1 year. The ISCLT model calculates average monthly, seasonal (quarterly) and/or annual
concentration and deposition values.
In this modeling study, the ISCST model was used to determine contaminant concentrations for an averaging period
of 24 hours. For volatiles, only contaminant concentrations due to volatilization were modeled. For semivolatiles, the
average concentrations due to volatilization and wind-blown dust from each source were modeled separately and added.
Meteorological Data
The meteorological data used were obtained from the EPA Scram Bulletin Board for the years 1984 through 1989.
Meteorological data were obtained from the nearest stations to the DOPC site. Surface wind data and mixing height data
for Pensacola and Appalachicola, Florida, respectively, were used.
Model Sources
The ISCST model accepts three different source types: point, area, and volume. The area type source was selected to
model the four sources during the remedial action. The following surface areas were assumed for each source.
. surface area exposure rate during excavation = 300 ydVday (250 m2/day);
• surface area of a front-end loader bucket = 3 yd2 (2.5 m2);
• surface area of soil during unloading from a dump truck = 3 yd2 (2.5 m2); and
• surface area of bioremediation = 2750 yd2 (2300 m2).
Source and Receptor Locations
The ISCST model is capable (in a single model run) of estimating average concentration values from multiple sources
positioned at different locations. The source locations for the dispersion model were selected based on the site layout
developed during the remedial design. Figure 1 shows the source locations. Figure 1 also shows the receptor locations.
To more efficiently conduct the modeling for the 15 chemicals of concern, an approach was used which lakes
advantage of the linear relationship which exists between source strength and resulting concentration values as well as the
linear relationship existing between the emission fluxes of the selected compounds and the four sources. The source ratios
were obtained by dividing the emission flux calculated for each source by the emission flux calculated for the
bioremediation source. These model ratios were used as the source strengths in the model input instructions and a base
emission flux for each compound was defined. In using these ratios as the input source strengths in the ISCST model, the
ambient impact estimated for each receptor represents a base receptor impact for each receptor. Thus, the compound-
specific impacts at each receptor can be obtained by multiplying the base emission flux (for each compound) by the base
receptor impact for each receptor.
490
-------
The maximum 24-hour concentrations of the contaminants of concern at each receptor location were obtained by first
modeling the four sources with respective emission ratios and then multiplying the resulting concentrations by the base
emission flux calculated for each chemical of concern.
Evaluation of Modeled Receptor Concentrations
Two methods were used to evaluate (he air modeled concentrations for the selected site contaminants. First, the
inhalation risks for carcinogenic and noncarcinogenic site contaminants were assessed using the USEPA risk assessment
methodology [1,2].
The USEPA considers carcinogenic risks to be significant if the risk exceeds one excess lifetime cancer case per ten
thousand exposed. The total carcinogenic risk calculated for air releases from the site during the remedial action was 1.8
excess lifetime cancer cases per one hundred thousand which is within the USEPA acceptable range even for the
maximally exposed individual.
Subchronic reference concentration (RfC) values were used to evaluate noncarcinogenic effects because the remedial
action is expected to have a duration of 3 years. To assess the overall potential for noncarcinogenic effects associated wilh
exposure to multiple chemicals, a hazard index approach has been developed by EPA (1). Ratios of inlakc:RfC arc
calculated for each chemical and summed as a hazard index. If the hazard index exceeds 1, it is likely that adverse health
effects will occur. The hazard quotients and hazard index for the predicted airborne concentrations of noncarcinogenic site
contaminants are considerably less than 1 (i.e., hazard index = 0.007). Therefore, a health risk due to inhalation of
noncarcinogenic site contaminants is not indicated.
The second method used to evaluate the model results is based on the Florida Department of Environmental
Regulation (FDER) Draft Air Toxics Permitting Strategy which is described in Chapter 403 of the State Regulations. This
criterion is based on the OSHA permissible exposure limit (PEL) for each compound as follows:
Criteria for Public Exposure = (PEL/100) x (8/24) x (5/7)
The factor of 100 provides a safety factor to protect individuals in the population that are more susceptible than "healthy"
individuals. The remaining factors are used to adjust the occupations standard (i.e., 8 hours/day and 5 days/week) to the
longer exposure periods associated with airborne exposures to the general public (i.e., 24 hours/day, 7 days/week).
Table 1 presents a summary of the maximum carcinogenic risks, the hazard index, and PEL/420 index for modeled
air releases from the site. As shown in Table 1, the concentration of pentachlorophenol is predicted to exceed its PEL/420.
Air Monitoring Program
Based on the results of the air modeling study it was concluded that the airborne concentrations of site contaminants
should not result in unacceptable exposures at the receptor locations. The model results, therefore, supported the
development of an air monitoring program that consists of two components; real-time monitoring with portable instruments
and limited air sampling for off-site analysis.
Real-time monitoring will be performed utilizing HNu®s and MINIRAM®s to measure the concentrations of
photoionizable hydrocarbons and particulates, respectively. The results of the portable monitoring will be used to
implement work practice changes and to trigger additional portable monitoring as well as sample collection for off-site
analysis.
Samples will be collected from three locations around the site on a regular basis for off-site analysis for volatile
organics and semivolatile organics by methods TOM and T013 respectively. The sampling locations are shown on
Figure I. The sampling locations were selected based on the prevailing wind directions the modeling results and the
locations of potential receptors. The frequency of sample collection for off-site analysis is dependent on the work activity
as well as the results of the portable monitoring. If photoionizable hydrocarbons and particulates are not detected above
background levels, samples for off-site analysis will be collected as follows:
• twice per week during the first week of excavation into the vault;
twice per week during the excavation of the lower vault soils (greater than 20 feet); and
• monthly during the remainder of the remedial action.
Based on the air monitoring results, the frequency of air sampling may be revised.
CONCLUSIONS
Air modeling was used to aid in the development of the air monitoring program for the Dubose Oil Products
Company Site Remedial Action. The air modeling results were used lo support a limited air monitoring program and to aid
in the selection of target compounds and monitoring locations.
REFERENCES
[ 1 ] Risk Assessment Guidance for Superfund (RAGS), Volume IA, EPA/540/14-89/002.
491
-------
FIGURE 1
LOCATIONS OF SOURCE AREAS,
RECEPTORS, AND AIR MONITORING STATIONS
TO
BARRINEAU PARK
SCAi£ IN FEET
T
FENCE LINE RECEPTOR LOCATION
_ POTENTIAL ON-SITE
RECEPTOR LOCATION
@Z] IDENTIFIES POTENTIAL RECEPTOR
• AIR SAMPLING STATION
A METEOROLOGICAL TOWER
» CENTROID OF SOURCE AREA
492
-------
TABLE 1
Summary of Air Modeling Results
Chemical
Carcinogens
Benzene
Methylene Chloride
Trichloroethene
Benzo(a)Anthracene
Benzo(b)FluoranUieDe
BenzoOOFluorantbene
Chrysene
Penlachloropbenol
Arsenic
Beryllium
Lead
Noncarclnogens
Toluene
Xylene
Anthracene
Fluorantbene
Fluorene
Phenanthrene
Pyrene
Barium
Vanadium
Hazard Index =
Total Risk =
Max. Predicted
Concentration
In Air
(mg/m3)
6.9E-04
9.4E-02
9.3E-03
2.3E-06
5.2E-07
5.2E-07
6.5E-07
9.6E-03
5.0E-07
8.9E-08
7.9E-06
1.7E-02
1.9E-02
7.9E-05
3.1E-03
3.0E-03
8.1E-04
2.3B-06
7.5E-06
3.5E-06
Hazard Quotient
Chronic Subchronic
(ing/kg/day) (mg/kg/day)
-
-
-
-
-
-
-
--
-
-
--
8.20E-04
6.10E-03
-
-
-
-
-
1.40E-03
-
8JE-03
—
...
—
...
...
...
...
...
—
...
...
8.20E-04
6.10E-03
—
...
...
...
...
1.40E-04
...
7.1E-03
Carcinogenic
Rbk
2.4E-07
1.6E-05
1.9E-06
1.6E-08
3.6E-09
3.6E-09
4.4E-10
...
2.9E-07
8.8E-09
...
lJtE-05
Exposure Index
Cone (air)
PEL/420
9.0E-02
2.2E-02
1.4E-02
4.8E-03
1.1E-03
1.1E-03
1.4E-03
8.0E+00
2.1E-02
1.9E-02
6.7E-02
1.9E-02
1.9E-02
6.6E-04
2.6E-02
2.5E-02
6.8E-03
1.9E-05
6.2E-03
2.9E-02
-------
Risk Assessment Methods for Exposure to
Environmental Substances Found Indoors
Theodor D. Sterling, VVilf L. Rosenbaum,
Chris W. Collett, James J. Weinkam
Theodor D. Sterling is Professor at School of Computing Science, Faculty of Applied Sciences, Simon Fraser
University, Burnaby, B.C., V5A 1S6.
Wilf L. Rosenbaum is Senior Research Associate at School of Computing Science, Faculty of Applied Sciences,
Simon Fraser University, Burnaby, B.C., V5A 1S6.
Chris W. Collett is Director of Environmental Research at Theodor D. Sterling and Associates Ltd., 250 - 1122
Mainland Street, Vancouver, B.C., V6B 5L1.
James J. Weinkam is Professor at School of Computing Science, Faculty of Applied Sciences, Simon Fraser
University, Burnaby, B.C., V5A 1S6.
ABSTRACT
Measures to control possibly hazardous or unwanted substances in public or commercial buildings as well as
residences depend heavily on methods by which risk of exposure to such substances are determined. Two major
approaches have been used: (1) extrapolation from high to low dose of exposure and (2) epidemiologic studies that
determine if there is an elevated risk for exposed individuals, using unexposed individuals as referents. Two important
examples are (1) the recent risk assessment of exposure to environmental asbestos fibers (EAF) by the Health Effects
Institute - Asbestos Research in conjunction with the U.S. Environmental Protection Agency and (2) lung cancer risk
assessment of building occupants exposed to environmental tobacco smoke (ETS), also by the U.S. Environmental
Protection Agency. In practise, the extrapolation and epidemiologic approaches appear to result in risk estimates which
differ substantially. The risk estimate based on extrapolation is considerably smaller (by one or two magnitudes) than
the risk estimates based on epidemiologic studies. Suitability of each method of risk estimate for EAF and ETS (and
other possible substances) and reasons for differences between them will be reviewed.
INTRODUCTION
Two major approaches have been used to estimate risks associated with low dose exposure to substances found
indoors: (1) Downward Extrapolation from high to low exposure. For example, the Health Effects Institute -
Asbestos Research (HEI-AR)' in conjunction with the U.S. Environmental Protection Agency (EPA) used downward
extrapolation for levels of environmental asbestos fibers (EAF) to estimate asbestos disease risk, and (2)
Epidemiologic studies that estimate the relative risk of exposed individuals with respect to unexposed
individuals. For example, the EPA estimated lung cancer and some other risks to environmental tobacco smoke (ETS)
by epidemiologic studies of populations classified into more or less exposed groups by some surrogate of exposure2.
The HEI-AR method assumes that health effects of low level asbestos exposure can be obtained by linear
extrapolation from effects of heavy occupational exposure. In principle, the lifetime excess risk of individuals exposed
to a concentration X times lower than the risk of workers heavily exposed to asbestos is taken to be 1/X times the
asbestos workers' excess risk.
The EPA estimates lung cancer or other risks for ETS by comparing mortality or morbidity of never smokers
exposed to ETS with that of never smokers not so exposed. Exposure is defined by some surrogate which is thought
to be related to dose.
494
-------
EVALUATION OF THE HEI-AR/EPA LINEAR EXTRAPOLATION
The HEI-AR Report's risk analysis relies on four assumptions which are highly questionable or may be
demonstrably untenable in some cases:
1. That the relative risk of asbestos exposure is linearly progressive;
2. That the composition of EAF is equivalent to the composition of asbestos to which workers are exposed
during mining or fabricating operations;
3. That any particular concentration of EAF represents equivalent doses for different exposed populations; and
4. That the health effect of all individuals passively exposed to EAF at the given concentration are
qualitatively the same as those of healthy workers exposed to the same concentrations.
Is the relative disease risk of asbestos exposure linearly progressive?
Human studies are of workers who mine, mill or handle asbestos and are exposed without exception to high doses.
The shape of a dose-response curve for low levels is in many ways difficult to ascertain from human studies,
especially for mesothelioma. However, there are some data from animal and human studies that are measured over a
sufficient number of dose levels and that suggest a curvilinear rather than linear relationship between asbestos disease
risk and exposure.
Malignant mesothelioma can be easily induced following inhalation or intraperitoneal and intrapleural injections of
asbestos fibers in various animals1'4. Table 17, Chapter 6 of the HEI-AR Report clearly shows a very rapid rise in
tumor response to very low doses, reaching an asymptote of approximately 80% tumors in the exposed animal
population. (See our Table 1.) Occupational studies may reveal only the higher range of a dose-response curve of
which the unobserved portion may be similar.
Seidman et al5 also present sufficient human data on mesothelioma to demonstrate the increased risk of
mesothelioma as a function of dose but the number of dose values are insufficient to establish the shape of a low
dose-response function. They have sufficient data on relatively low exposure to plot SMRs for lung cancer and for all
asbestos disease as a function of "estimated fiber exposure starting with less than 6 fibers/cc after 5-40 years since
onset of exposure" and for "length of time worked", starting with a period of less than six months. The data for lung
cancer and asbestos disease by fiber years/cm3 and by length of time worked resembles in a qualitative way that of
animal data for mesothelioma following intraperitoneal injection. Our Figure 1 plots SMRs of asbestos disease and of
lung cancer by estimated fiber exposure as calculated by Seidman et al. The downward progression of SMRs with
decreasing exposure (measured by estimated fiber years of exposure or by length of time worked) appears linear for
heavier exposure but drops off for lesser exposure. These curves clearly show the appropriateness of a curvilinear
relationship. If a straight line would be fitted to SMRs for estimated fiber levels, they would result in SMRs greater
than 2.5 when exposure to asbestos fibers equals to zero. Whatever weakness estimates of fiber count have, an
intercept value along the SMR axis of greater than 2.5 leaves the investigator trapped between the Scylla of positing a
curvilinear function, rapidly dropping off to an SMR value of 1.0 when asbestos dose approaches zero and the
Charybdis of assuming the existence of one or more confounding factors which masks the true relationship between
asbestos exposure and risk. For further discussion see Sterling et al6.
EAF is physically not equivalent to fibers in mining, fabricating or insulating.
EAF varies in distribution of fiber size, type and thickness depending on what material is being released into the
atmosphere and how the release is taking place. EAF may be mixed with other fibers. In most instances, EAF
probably contains additional toxic materials that have adhered to the fibers and add to their toxicity. Moreover, the
average fiber count in a particular building may not be a true indication of actual exposure of occupants and may, in
fact, mislead.
Identical concentrations of EAF do not give the same dose to different exposed populations.
Exposure of industrial workers and of children, females, sick persons, institutionalized seniors and handicapped
persons to the same fiber concentration results in profound differences in delivered dose.
495
-------
The health effect of involuntary exposure may not be equivalent to that of occupational exposure.
It is unreasonable to expect that the health effects of blue collar workers and those of a wide range of individuals
coming into involuntary contact with EAF would be similar for equivalent exposures.
THE EPA'S APPROACH USING EPIDEMIOLOGIC STUDIES.
The EPA2 and NSC/NRC7 used similar methods to estimate the health effects of passive smoking. However, the
EPA Report is the more complete and will be discussed here.
The lung cancer risk for non smokers (or other disease risks) is determined by comparing mortality or morbidity
of non smokers (in most instances spouses of smokers) exposed to ETS with that of non smokers not so exposed or
less exposed.
Exposure is defined by a surrogate that is thought to be related to dose and usually are obtained by responses to a
questionnaire. Surrogates are used to classify exposure levels. For instance, one study defines exposure as "smoking
by any member of the household" while another specifies and classifies exposures by "number of cigarettes smoked by
husband". Thus, different surrogates of exposure reflect different levels and amounts of exposure.
CONSEQUENCES OF THE CHOICE OF RISK ASSESSMENT METHOD.
Risk estimates obtained by downward extrapolation tend to be considerably smaller than risk estimates obtained
by epidemiologic studies. The HEI-AR Report estimates by linear extrapolation that these are 4 to 40 premature
cancer deaths per million persons exposed to outside ambient asbestos levels. This figure is not only impossible to
verify but even to detect an annual increased risk level of 40 per million would take on the order of 200,000 cases and
as many controls. Yet elevated risks of individuals involuntarily exposed to EAF can be determined easily from
relatively small samples of exposed individuals. For instance, Joubert et al.* observed 43 cancer deaths where 25
would have been expected in a sample of 115 decedents who had been involuntarily exposed to asbestos. Even for
such a small sample, the probability of finding such an increase in cancer mortality above that expected by chance is
less than 0.003.
Differences in lung cancer risk estimates associated with ETS between downward extrapolation from active
smoker risk and those derived from epidemiologic (mostly spousal) studies are again large. Estimates of excess
numbers of lung cancer deaths among non smokers in the EPA Draft Report based on epidemiologic spousal studies
of excess lung cancer range from 3000 to 3700. On the other hand, the number of excess lung cancer death derived
from linear downward extrapolation varies between 12 and 354'-'°.
Thus, risks established by the epidemiologic method is significantly elevated while that established by downward
linear extrapolation tends to be small, in fact so small that it cannot be tested by epidemiologic methods.
The low risk levels established by linear downward extrapolation are invalid because health risk associated with
EAF is not linearly progressive; EAF is different in composition and toxicity from fiber encountered in mining and
milling of asbestos; and neither dose levels nor health effects can be compared between populations involuntarily and
occasionally exposed to EAF with those of healthy blue collar workers occupationally exposed to asbestos fibers.
On the other hand, the elevated risk obtained from epidemiologic studies are confounded by socio/economic
factors. Households with spouses and children exposed to EAF or ETS differ in a large number of factors, but mainly
in exposure to toxic materials related to lower socio/economic status, to toxic materials brought home on hair, skin
and clothing of their spouses employed in industry when compared to households where there are no husbands
exposed to asbestos or other toxic substances or no smoking husbands or smoking mothers. Moreover, differences in
socio/economic class between households with and without asbestos workers or smokers are large, and with it are
differences in many health-related factors. This pattern then raises the questions of the extent to which a comparison
of household members with greater exposure (to EAF or ETS) or to other toxic substances) to those with less
exposure also compares groups that differ with respect to indirect occupational exposure to toxic substance and/or to
the associated social class differences in lifestyle and diets and so on that go along with occupational and
socio/economic differences.
496
-------
These considerations are of utmost importance because there is a uniform difference in mortality tied to social
class. Regardless of how social class is measured, by income, by occupation or by socio/cultural factors, there is a
consistent difference in mortality by which lower socio/economic stratas suffer from a significantly higher mortality
than higher socio/economic stratas "•'2'13
Finally, risks from asbestos exposure determined by the HEI-AR method cannot be compared to risks from ETS
exposure determined by the EPA method because the former risk is determined by a method which assures minimal
risk value while the latter is determined by a method which assures maximum risk values.
CONCLUSIONS
Our review reveals the existence of a dilemma. One solution to the dilemma might be to always perform both
types of risk analyses. The better alternative would be to use neither method of risk evaluation. Our brief analysis has
shown that results of neither method of estimating risk lacks the needed validity for drawing conclusions about risks
to health from low doses. Eliminating or at least decreasing as much as possible the risk to low, occasional and
involuntarily exposed populations can be accomplished without knowing for certain the exact values of risk to which
these populations are actually exposed. For instance, maintenance workers will be warned and instructed to take due
caution in removing installed asbestos in the case of repairs and asbestos workers can be trained not to take home
asbestos on their clothing, hair and skin. Similarly ETS exposure can be minimized by proper ventilation and proper
location of smoking permitted areas at the workplace — all this can and should be done quite independently of
establishing a risk of involuntary exposure at any level. In cases such as EAF or ETS, when environmental exposure
is unavoidable, good industrial hygiene practices ought not to depend on risks associated with poor industrial hygiene
practices. Also the need for establishing risk levels for exposure to toxic agents in the workplace for populations
necessarily exposed to high levels of toxic materials is quite a separate matter. Here regulation of occupational
exposure to almost always very high levels of toxic substances require the determination of risks associated with the
occupational environment in order to set standards at levels required to protect exposed workers.
ACKNOWLEDGMENTS
Support for the ETS analysis came in part from several cigarette manufacturers.
REFERENCES
1. Health Effects Institute - Asbestos Research. Asbestos in public and commercial buildings: A literature review and
synthesis of current knowledge. Cambridge, 1992.
2. United States Environmental Protection Agency. Health effects of passive smoking: Assessment of lung cancer in
adults and respiratory disorders in children. (External Review Draft) EPA/600/6-90/006A. Washington, 1990.
3. F. Pott, "Animal experiments on biological effects of mineral fibers". Biological effects of mineral fibers, Vol I;
Wagner JC, Ed. 1980, pp. 261-269.
4. Y. Suzuki, "Comparability of mesothelioma in humans and in experimental animal studies". The third wave of
asbestos disease: exposure to asbestos in place. Ann NY Acad Sci Special Edition: Landrigan PJ, Kazemi H (eds).
pp. 643, (1991).
5. H. Seidman, I.J. Selikoff. "Mortality of asbestos factory workers: dose-response relationships 5 to 40 years after
onset of short term worker exposure". Am J Ind Med 10:479-514 (1986).
6. T.D. Sterling, W.L. Rosenbaum, J.J. Weinkam. "Comments on Health Effects Institute - Asbestos Research:
'Asbestos in public and commercial buildings: A literature review and synthesis of current knowledge' — with special
emphasis on risk assessment methods used" Am J Ind Med. in press, 1993.
7. National Research Council. Environmental tobacco smoke: Measuring exposures and assessing health effects.
National Academy Press, Washington, D.C., 1986.
8. L. Joubert, H. Seidman, I. Selikoff. Mortality experience of family contacts of asbestos factory workers. Ann NY
Acad Sci. Vol. 643, 416-418, 1991.
497
-------
9. A. Arundel, T.D. Sterling, J.J. Weinkam. Never smoker lung cancer risk from exposure to paniculate tobacco
smoke. Environ Intl. Vol. 13, 409-426, 1987.
10. T. Sterling, W. Rosenbaum, J. Weinkam. Result of linear downward extrapolation of ETS-related risk based on the
National Mortality Followback Survey. In press, 1993.
11. T. Sterling, W. Rosenbaum, J. Weinkam. Income, race and mortality. J Natl Med Assoc. in press, 1993.
12. J. Weinkam, W. Rosenbaum, T. Sterling. Computation of relative risk based on the simultaneous surveys: an
alternative to cohort and case-control studies. Am J Epidemiol. Vol. 136, 722-729, 1992.
13. T. Sterling, W. Rosenbaum, J. Weinkam. Risk attribution and tobacco-related deaths. Am J Epidemiol. in press,
1993.
498
-------
TABLE 1. Mesothelioma production in rats in relation to dose from
intrapleural and intraperitoneal injection studies.
(adapted from the HEI-AR Report, Table 6-17.)
DOSE (mg)
15.0
10.0
7.5
5.0
2.5
0.5
0.05
0.01
AMOSITE
79.2
75.0
62.5
70.8
59.4
46.9
25.0
8.3
PERCENT OF
CROCIDOLITE
70.8
41.7
62.5
41.7
56.3
31.3
25.0
0.0
TUMORS
CHRYSOLITE A
79.2
83.3
83.3
79.2
68.8
80.6
37.5
4.3
499
-------
1600 y
1500 1
1400 -j
1300 -I
1200 -
s
M
R
1100 -
1000 -
900 -
800 -
700 -
600 -
500 -
400 -
300 -
200 -
1 nr\ -1
^
.^
/"•"-., / ^^
/ . -^
^ /
. •__-/ r- - Lung Cancer
tl f
>jt — • — Asbestos Disease
r
50
Extracted from Seidman et al, 1986
100 150 200 250
Fibre-Years per Cubic Centimetre
300
350
Figure 1. SMRs of lung cancer and asbestos diseases from 5 to 40 elapsed
years since onset of work, in an amosite asbestos factory,
according to estimated fiber exposure.
-------
A Pilot Study to Assess Personal Exposures to Ozone in the Fraser Valley of British Columbia
Michael Brauer
The University of British Columbia,
Department of Respiratory Medicine / Occupational Hygiene Programme, 2206 East Mall; Room 366A,
Vancouver, BC CANADA V6T 1Z3
Jeff Brook
Atmospheric Environment Service
Downsview, Ontario
Mark Raizenne
Health and Welfare Canada
Ottawa, Ontario
ABSTRACT
Personal exposures to ozone were measured in a pilot monitoring study during July and August of 1992.
Measurements were made in the Fraser Valley, a suburban and rural river valley bordering the metropolitan region
of Vancouver, British Columbia (population approximately 1.3 million). Two groups of 25 healthy individuals
were selected for 14 consecutive days of personal monitoring, based on prior expectations of their activity patterns.
The first group was composed of adult health care workers who were expected to spend a majority of the work day
indoors or commuting. The second group, teenage camp counselors, were expected to spend most of the day
outdoors. Time activity data was collected to investigate the association between activity patterns and ozone
exposures. Personal monitoring was conducted with the nitrite-coated filter passive ozone sampler. Sampler
performance was evaluated and ozone collection rates were determined empirically by collocation of the passive
samplers with continuous ozone analyzers for one month at three fixed sites within the region. Inter individual
variability of ozone exposures was assessed by comparisons between the camp counselors, who spent the majority
of the sampling period in the same general location. Intra-individual variability, as a result of sampler performance,
was assessed by collecting a number of duplicate personal samples. Acid aerosols constituents (H+, SO^", N03"
and NH4+) were also measured for the duration of the study. Although ozone levels were low (< 35 ppb 24-hour
average) during the sampling period, the passive ozone sampler agreed well with co-located LTV photometric ozone
measurements and was found to be feasible for personal sampling. Based on replicate personal samples, personal
measurements differing by more than 7 ppb were associated with true differences in exposure. Differences in ozone
exposures were observed between the two sampling populations, with higher exposures recorded in the group which
spent more time outdoors. Outdoor ozone exposures were estimated to account for 14 - 37 % of the variability in
measured personal ozone exposures. Acid aerosols were not detected and aerosol SO/^'concentrations were low (<
50 nmol m~^).
INTRODUCTION
Recently, concern has been expressed about elevated concentrations of ozone and other constituents of
photochemical smog in the Fraser Valley airshed of British Columbia. The Fraser Valley has experienced many of
the highest ozone concentrations in Canada, and potential for continued adverse air quality is expected to persist as
development in the region continues' While there is a wealth of data regarding concentrations of major air
pollutants, including ozone, little is known regarding the distribution of population exposure to these pollutants and
the mitigating factors resulting in variable exposures in the resident population. Knowledge of population exposure
to ozone will be important in order to assess the effectiveness of future strategies for reducing photochemical
oxidant pollution in the Fraser Valley and to provide useful information for the development of new reduction
strategies. Among other factors, exposures to ozone are likely to be affected by spatial variability in ambient
concentrations, the degree of penetration of ambient ozone into varying indoor environments, characteristics of the
indoor environment which affect ozone reactive decay (residence time, surface materials, other reactive pollutant
concentrations), the relationship between time-activity patterns of the population and ozone concentrations within
specific microenvironments (commuting, workplace, etc.) and the relationship between time-activity patterns of the
population and the diurnal ozone concentration profile.
501
-------
Furthermore, while the health effects of ozone at concentrations experienced in ambient air have been demonstrated
repeatedly in clinical chamber studies the epidemiological evidence supporting ozone health effects is less
convincing^*3. One limitation is the lack of personal exposure data for ozone. These data are limited since
conventional ozone monitoring equipment (UV photometry, chemiluminescence) is bulky and expensive. The recent
development of an inexpensive and simple passive monitor for ozone provides a new opportunity to examine the
distribution of population exposures'*'^. Perhaps a more significant limitation of previous North American ozone
epidemiological studies, particularly those conducted in eastern regions, is the coincidence of ozone episodes with
those of acidic sulfate species^. Human chamber and animal studies suggest that both ozone and acidic aerosols
may be associated with adverse respiratory health outcomes, and evidence for potentiation interactions also
exists^'9. However, from a regulatory and policy viewpoint, control strategies for acidic aerosols and ozone are
distinctly different. Although several SC>2 sources (oil refineries, cement plants) do exist in the Vancouver
metropolitan area, the lack of major oil or coal-fired sulfur emissions sources in the greater Vancouver area make it
unlikely that significant concentrations of acid aerosols will be observed in the Fraser Valley. However, no
measurements have been made to date. By demonstrating that ozone episodes occur in the Fraser Valley in the
absence of elevated concentration of acid aerosols, the potential confounding effect introduced by acidic aerosols
can be reduced eliminated in epidemiological investigations.
METHODS
Aerosol Acidity. Daily ambient aerosol strong acidity monitoring was initiated in mid July 1992 at 3 sites in the
Greater Vancouver Regional District (GVRD): Kensington, Pitt Meadows and Chilliwack. A total of 60 valid 24-
hour samples were collected at each of the three sites (100% valid sample collection). Samples were collected at a
flow rate of 4 L min"' with the Harvard Aerosol Impactor / Aluminum Honeycomb Denuder system. Sampler
inlets were located 1.5 m above all surfaces and at least 1 m from major supports. Teflon membranes were
analyzed for aerosol strong acidity by the method of Koutrakis, et all". Remaining filter extract was analyzed for
804^", NO3~ and NH4+ by ion chromatography (Dionex 2000i). Limits of detection, based on a 24-hour sample
collected at a flow rate of 4 L min'1 are 15, 3.5 and 20 nmoles m"3 for SO42", N03~ and NtL)"1", respectively.
Ozone. To calculate the sampling rate of the passive ozone sampler, passive ozone samplers were collocated with
UV photometric (TECO 49) continuous ozone analyzers at three continuous monitoring sites: Pitt Meadows
Abbotsford and Chilliwack. The Chilliwack site was a rural location at the far end of the Fraser Valley from
Vancouver (approx. 60 miles from the city center). The Pitt Meadows site was a rural-suburban location about 20
miles east of Vancouver. The Abbotsford site was a suburban site approximately 45 miles east of Vancouver.
Due to logistic constraints the passive sampler was suspended near the library entrance (approximately 2m above
ground level) while the inlet for the continuous analyzer was on the roof (approximately 4.5m above ground level).
All continuous analyzers were calibrated weekly and a full site audit was performed monthly. 24-hour average
samples were collected daily for a period commencing in mid July 1992. Passive samplers were placed inside a
wind and rain cover which was hung on a tripod placed at least 1 m from all major supports and building faces.
Except as noted the sampler inlet was approximately 1.5 m above surface level and as close as feasible to the
sampling manifold inlet of the continuous ozone analyzer. Passive samplers were prepared in the laboratory within
four days of their deployment in the field. Sampler preparation and analysis procedures are described elsewhere1*''.
Personal exposures to ozone were monitored in two population groups residing in the Fraser Valley. Personal
samples were collected by clipping the sampler onto their (normal) clothing in the breathing zone. Samplers were
worn continuously for 24-hours, except while sleeping or bathing, during which time the sampler was placed (still
exposed to air) near the individual.. A group of 25 individuals working at a medical clinic in Abbotsford (LFV)
were selected to wear passive ozone samplers daily during a 14-day period. From this group, a total of 320 valid
samples were collected. Hourly time-activity data, recorded by self-administered logs, were completed for each
valid sample obtained. Ozone exposures were also measured in a group of 25 individuals (during a separate two-
week period) selected from the staff of an overnight camp (CLU) located across the Fraser River from Abbotsford.
These individuals were expected to have different activity patterns than the LFV population group, as the majority
of their day was spent outside or in cabins at the camp site. An identical sampling protocol was followed by this
group. A total of 289 valid personal ozone samples were collected from the CLU group. During the period of
personal ozone exposure monitoring, each group of 25 individuals also provided twice daily measurements of peak
502
-------
expiratory flow with Mini-Wright peak flow meters. These measurements were self-administered, following an
orientation, and daily checks with the field technician.
RESULTS AND DISCUSSION
No aerosol strong acidity was detected on any of the samples (LOD = 10 nmoles m'3). Anion and cation
measurements were also low and the results will not be discussed further in this manuscript. Maximum
concentrations of SO^" were below 50 nmol m"3 (mean: 18 nmol m"3) while the maximum concentrations of
NC>3~ and NH4+ were below 35 (mean: 4.5) and 100 (mean: 30) nmol m'3, respectively.
Passive Monitor Sampling Rate. For calculation of the passive sampler sampling rate, data from co-located
passive samplers were compared to data obtained from continuous analyzers (TECO 49) at three sites. Sampling
rates for the passive samplers were determined by goodness of fit for a linear relationship between the passive and
continuous measurements at Pitt Meadows and Chilliwack. The Abbotsford site was excluded from the
determination of the overall sampling rate as the relationship between the passive and continuous measurements
was much weaker than at the other two sites and differed significantly from those previously reported4-5-''. The
slopes determined for the Chilliwack and Pitt Meadows sites were in good agreement with each other and with those
previously reported4-5." The average of the slopes from these two sites was then used as the overall sampling
rate for all three sites as well as all personal samples (25.9 cc/min). Blank values were subtracted from all samples
(mean blank = 1.19 p.g concentration in extract). Using the sampling rate of 25.9 cc/min we then compared 24
hour average ozone concentrations from the passive sampler with the mean of the hourly average concentrations
from the continuous analyzer at each of the three co-location sites:
Chilliwack: Passive = 0.89*Fbted (r2 = 0.88; Intercept: non-significant; N=35)
Pitt Meadows: Passive = 0.76*Fixed + 5.3 ppb (r2 = 0.69; N=34)
Abbotsford: Passive = 0.40*Fixed (r2 = 0.34; Intercept: non-significant; N=34)
Both the Chilliwack and Pitt Meadows sites showed good agreement between the passive and continuous
measurements, while agreement at the Abbotsford location was poor as seen by the low slope and r^ values. The
poor agreement at this site can be explained by the difference in sampler placement. As the passive sampler was
located closer to surface level, ozone concentrations at this altitude are expected to be lower as a result of reaction
between ozone and surfaces. Additionally, as this site was the most urban of the three co-location sites, ozone
reactions with NO from auto exhaust are expected to have reduced surface level ozone concentrations further This
measurable difference between ozone concentrations at near surface level elevations differing by only 2.5m
supports continued personal or microenvironmental measurements of ozone exposures as monitoring network
measurements collected at heights greater than breathing zone heights will tend to be positively biased with respect
to human exposure.
Quality Control. Initial evaluation for differences in blank values from different filter coating batches or field
placements, indicated that blank variation was random. Accordingly, all field blanks (N=53) were pooled together
and an overall blank value was used for correction of sample values (Table 1). Lab blanks (N=15) were not
significantly different from field blanks, indicating that there were no exceptional problems in the field sampling
procedures. All of the blank response was explained by the laboratory portions of the method itself (filter
preparation extraction and analysis). The limit of detection (LOD) of the passive sampler method was calculated
to be 8.1 ppb, based on 3*standard deviation of the field blanks. This LOD is in good agreement with those
reported previously4-5-' *.
Table 1. Personal Ozone Samples Quality Control Parameters.
Parameter
Field Blank
Lab Blank
Duplicates
(difference)
N
53
15
12
Mean
12.5
13.9
3.0
S.D.
2.7
1.7
2.1
Range
64-185
10.4- 17.2
0.5- 76
503
-------
Additionally, a series of duplicate personal ozone samples were collected and analyzed. Subjects were instructed to
wear one sampler on the right side of the chest and one on the left side for identical sampling durations. All
duplicate samples were analyzed during the same sampling run. After subtracting the global blank value from each
of the duplicate samples we calculated the difference in the resulting ozone concentration (ppb) between the two
samples. The mean, standard deviation and range of these differences are reported in Table 1. Using a value of
1.96 * the standard deviation of the differences, we can estimate that 95% of random differences between two
personal exposure measurements will be less than 7.2 ppb. Therefore, with these confidence limits any two
personal exposure samples differing by less than 7.2 ppb (assuming a 24-hour sampling duration) cannot be
considered to be a true difference in personal exposure. Although no replicate results for the passive sampler have
been reported to date, Liu, et al11 estimate the relative error of the method to be 15-25%. Consequently, a 50 ppb
sample would have a relative error of 7.5 - 12.5 ppb. Given the low ozone concentrations during the sampling
period and the estimated method error, our calculation of a 7.2 ppb 95% confidence interval for differences in
replicate personal measurements suggests that no additional error was introduced by collecting personal samples.
Comparisons between the personal ozone samples and fixed-location monitors (continuous at LFV and fixed-
passive monitor at CLU) are shown in Figures 1 and 2. Although ozone concentrations were quite low during the
duration of the monitoring and many measurements were below the LOD, we still observed variability in personal
exposures, beyond that which could be explained by the variability in the method itself. With only a few exceptions,
personal exposures were lower than fixed location monitors, although mean personal exposures generally tracked
along with the fixed-location monitor concentration. The mean differences between continuous ozone and all
personal ozone measurements were 12 and 8.5 ppb for LFV and CLU, respectively. Time-activity data indicated
that the two groups differed significantly in the amount of time spent outdoors. The mean percentage of time spent
indoors, outdoors and in transit were 84.8, 9 and 5.5%, respectively for the LFV group. The CLU group spent, on
average, 72.4, 25.8 and 1.9% of time indoors, outdoors and commuting, respectively. Since we found significant
differences between the time-activity patterns of the two groups, we used the time-activity data to predict
differences in personal exposures. Exposures estimated from time-activity data and fixed-location monitors
(fraction of time outdoors * outdoor concentration) and measured personal exposures showed the following
relationships.
LFV: Estimated Personal Exposure = Measured Personal Exposure * 0.27 (non-significant intercept)
r2 = 0.37 N=232
CLU: Estimated Personal Exposure = Measured Personal Exposure * 0.19 + 2.79 ppb
r2 = 0.14 N=251
The relationship between the measured and estimated personal exposures can be compared with that described by
Liu, et al. for the measurements in State College, PA for 81 samples". These investigators found an overall
model r^ of 0.33 with a coefficient for estimated personal exposures (based on outdoor concentration and time only)
of 0.52. The coefficient for indoor concentration and time was significantly higher (0.82), indicating the
importance of indoor exposures in determining the total personal exposure. In light of the results of Liu, et al., our
estimated exposures based only on outdoor concentration and time show that exposures of different population
groups can be predicted to varying degrees by the amount of time spent outdoors. For these data, although the
personal measurements for the CLU population were closer to the fixed-location measurements (Mean ratios of
personal: fixed location ozone were 0.53 and 0.35 for CLU and LFV, respectively), the fixed location
measurements and the time spent outdoors explained less of the variability in personal exposures. This is likely due
to the higher indoor ozone concentrations experienced by this group. When indoors, these individuals were
primarily in wood cabins with open windows, structures which would be expected to yield elevated indooroutdoor
ozone concentration ratios. In contrast the LFV population, when indoors, were primarily in mechanically
ventilated offices or in their homes, both environments would be expected to have lower indooroutdoor ozone ratios
than the cabins at the CLU site. Accordingly, this population's exposure is dominated by outdoor exposure more
than the CLU group.
504
-------
CONCLUSIONS
Co-location experiments with the passive ozone sampler indicated good agreement with UV photometer continuous
analyzers. The positive response of blank samplers was quite consistent and appeared to be a result of coated filter
preparation and not contamination during sampler assembly storage or analysis. Duplicate personal samples
indicated that no additional measurement error was associated with collecting personal samples although the error.
inherent in the method suggested that only differences in personal exposures greater than 7.2 ppb could be
considered true differences in exposure.
Although ozone concentrations during the sampling period were low, we observed inter individual variability in
ozone personal exposures using a passive dosimeter. Measurements in two populations groups who spent different
amounts of time outdoors indicated that increased time spent outdoors was associated with increased exposure to
ozone, although an improved explanation of the variability in ozone exposures was not associated with time spent
outdoors. This observation suggests that variability in ozone exposure is associated with exposure in other
microenvironments, particularly indoors. Comparisons between fixed-location continuous monitors which were
part of a regional monitoring network and personal monitoring indicated that personal exposures were substantially
lower than reported ambient concentrations. One explanation for the overestimation of personal exposures by the
fixed-site network was suggested in this pilot study to be the elevation of the continuous monitoring network
samplers above the height of the breathing zone. At the height of the breathing zone, ozone concentrations are
expected to be lower as a result of reaction with surfaces, resuspended dust and NO from automobile exhaust.
Additionally, during this limited monitoring period we observed no measurable aerosol strong acidity and low levels
of aerosol SO^~, indicating that epidemiological investigations of effects of ozone exposure on respiratory health
conducted in the Fraser Valley are unlikely to suffer from confounding by acid aerosols.
REFERENCES
1. Greater Vancouver regional District, Air Quality Management Plan Discussion Paper, May 1992.
2. Lippmann, M. "Health effects of ozone: A critical review," J.Air & Waste Management. 39(51:672-695 (1989)
3. Tilton, B.E. "Health effects of tropospheric ozone," Environmental Science and Technology. 23(3):257-263.
4. Koutrakis, P., Wolfson, J.M. and Mulik, J. "Development of an ozone personal sampler", in Proceedings of the
1990 EPA/A&WMA conference on Measurement of Toxic Air Pollutants. Raleigh, NC. 1990.
5. Koutrakis, P., Wolfson, J.M., Bunyaviroch, A., Froehlich, S.E., Hirano, K. and Mulik, J.D. "Measurement of
ozone using a nitrite coated filter," Submitted to Analytical Chemistry (1992)
6. Bates, D.V. and Sizto, R. "Air pollution and hospital admissions in southern Ontario: The acid summer haze
effect," Environ. Res. 43: 317-331
7. Lippmann, M. "Health effects of troposheric ozone," Environmental Science and Technology 25: 1954-1962
(1991).
8. Kleinman, M., etal. "Aerosols formed by atmospheric mixtures." Env. Hlth. Persp. 79:137-145
9. Last, J.A. "Effects of inhaled acids on lung biochemistry." Env. Hlth Persp. 79:115-119
10. Koutrakis, P., Wolfson, J.M. and Spengler, J.D. "An improved method for measuring aerosol strong acidity:
Results from a nine-month study in St. Louis, Missouri and Kingston, Tennessee," Atmospheric Environment. 22:
157-162(1988).
11. Liu, L-S.J., Koutrakis, P., Suh, H.H., Mulik, J.D. and Burton, R.F. "Use of personal measurements for ozone
exposure assessment - A pilot study," Submitted to J. Air and Waste Management (1992)
505
-------
_Q
Q.
C
o
C
0 are shown. Horizontal line denotes limit of detectin of 8.1 ppb for 24 hour sample.
.Q
Q.
O.
o
n3
0)
o
C
o
o
0)
C
o
N
o
;LOD
1 2 3 4 5 6 7 8 9 10 11 12 13 14
DAY
— Fixed-site x Personal
Figure 2. Relationship between fixed location (fixed location passive sampler located at CLU site) and measured
personal exposure samples for CLU study population. Each X denotes an individual sample. Personal exposure
sample concentrations were corrected for concentration of field blanks and therefore only samples where resulting
concentration was > 0 are shown. Horizontal line denotes limit of detection of 8.1 ppb for 24 hour sample.
506
-------
Session 13
Mercury in the Environment
-------
MERCURY DETERMINATION IN ENVIRONMENTAL MATERIALS:
METHODOLOGY FOR INSTRUMENTAL NEUTRON ACTIVATION ANALYSIS
I. Olmez, M. Ames, and N. K. Aras*
Massachusetts Institute of Technology
138 Albany Street
Cambridge, MA 02139
Air paniculate samples have been collected at five sites across Upstate New York since October
1991. Automatic dichotomous samplers with a PM-10 inlet were used to collect fine particles (< 2.5nm)
and coarse particles (2.5 to 10 |im). Vapor phase mercury has also been collected on charcoal sorbants at
the same sites.
Samples were analyzed for their trace element content by instrumental neutron activation analysis,
INAA. In general the determination of mercury in environmental samples is done by irradiating the sample
with neutrons, followed by a radiochemical separation of mercury and subsequent detection of the 279.2
keV gamma ray from Hg-203. This is tedious work if one analyzes many samples. At MIT, we have
found the isotope Hg-197 to be better suited for INAA based on its decay properties and the absence of
any required radiochemical separation.
If the Hg-203 isotope is used in INAA for determination in unknown matrices which may contain
both Ta and Se, the situation is rather complex. The 279.2 keV line used for Hg-203 determination can
not be separated from 279.5 keV line of Se-75. By knowing the branching ratios in the decay of Se-75,
we can determine the amount of 279.5 keV radiation by measuring the 264.6 keV line of Se-75. Yet this
line is not unique as it has interference from 264.1 keV gamma ray of Ta-182, and this should be corrected
through 1221.0 keV line of Ta-182. The three-stage correction introduces errors due to simple
subtractions and uncertainties in branching ratios and detector efficiencies.
Therefore we used 77.4 keV line from Hg-197 instead of Hg-203 to measure total mercury. It
provides a more sensitive measurement because of its much higher production cross section and its shorter
half-life. We have successfully applied this approach to hundreds of environmental samples and verified
the suitability of this methodolgy.
*Fulbright Scholar from, Middle East Technical University, Ankara, Turkey
509
-------
Introduction
Mercury is an element of significant toxicologic importance which may cause well-characterized
health problems in humans. As a result, mercury emissions from anthropogenic sources have become a
major environmental issue. In addition to natural sources of mercury, fossil fuel combustion, waste
incineration, and smelters also add this element to the environment1. Because of its high vapor pressure,
mercury is emitted to the atmosphere primarily as a vapor, as opposed to the aerosol form, and it persists
in this physical state. The tendency in its natural state to volatilize and its touchy both as a vapor and
when incorporated in organometallic compounds have been long recognized1.
In the past a large number of analytical techniques have been applied to determine the Hg content of
a variety of environmental samples^A Among those utilized have been UV absorption
spectrophotometry, atomic emission spectrometry, thin gold-film surface resistance, Instrumental Neutron
Activation Analysis (INAA), Radiochemical Neutron Activation Analysis (RNAA), and X-Ray
Florescence (XRF). In general many of these methods depend upon preconcentratipns of the mercury in
some medium with a subsequent and immediate release of the mercury into a detecting device. Each of
these methods has various advantages and disadvantages^.
Instrumental neutron activation analysis is one of the most sensitive, selective, and reliable
techniques available for the determination of trace elements. It is non-destructive and capable of detecting
up to forty elements at small concentrations.
Assessment of Mercury Determination by INAA
In general the determination of mercury in environmental samples is accomplished by irradiating
the sample with neutrons, followed by a radiochemical separation of mercury and subsequent detection of
the radiation characteristic of a particular mercury isotope. Usually the 279.2 keV gamma ray from Hg-
203 is the detected radiation. At MIT we have found the isotope Hg-197 to be better suited for
instrumental neutron activation analysis based on its decay properties and the absence of any required
radiochemical separation.
As seen in Table 1, two naturally occurring mercury isotopes are suitable for total Hg
determinations, Hg-196 and Hg-202. Although Hg-196 has a lower abundance compared with Hg-202,
Hg-196 is much more sensitive for total Hg determinations. This is mainly due to its higher cross section
and shorter half life. Under identical irradiation conditions, e.g. 12h irradiation with a thermal neutrons,
Hg-197 production is 50 times larger than Hg-203. Additionally, the 77.4 keV gamma line of Hg-197 is
free from interference of other radiation, while the 279.2 keV gamma line from Hg-203 is not.
If the Hg-203 isotope is used for determinations in unknown matrices which may contain both Ta
and Se, the situation is rather more complex. The only line that is used in Hg-203 determinations is at
279.2 keV, which can not be separated from 279.5 keV line of Se-75. In order to know how much of the
radiation measured at "279.2 keV" is from the 279.2 keV line of Hg-203 alone, we must subtract from the
measured intensity the contribution from the 279.5 keV line of Se-75. By knowing the branching ratios in
the decay of Se-75 we can determine the amount of 279.5 keV radiation and the needed correction by
measuring the 264.6 keV line of Se-75. Yet this line also is not "unique" as it has interference from the
264.1 keV gamma ray of Ta-182. However the contribution of this Ta radiation can be corrected for by
measurement of the 1189.0 or 1221.0 keV lines of Ta-182 which are not interfered with by any other
radiation. If we measure total areas under 279,264 and 1221 keV gamma ray peaks determined with a
HPGe detector the above correction can be made by using the following equation:
279(Hg) = 279(Total) — 0.42 [264(Total) — 0.57 x 1221(Ta)] (1)
510
-------
where 279(Hg) and 1221(Ta) are the areas under the peak as at 279 keV and 1221 keV corresponding to
Hg-203 and Ta-182 respectivity. The 279(Total) and 264(Total) are the total areas under the gamma ray
peaks at 279 and 264 keV. The coefficients are calculated taking into account gamma ray intensities given
in Table 1 and relative detector efficiencies of gamma rays involved. For this calculation the ratios of
efficiency are taken as
Eff (264) Eff (1221)
This three stage correction introduces errors due to the simple subtraction and uncertainties in branching
ratios and detector efficiencies. Depending upon the relative concentrations of Hg, Se and Ta in the
sample, it may not be possible occasionally to determine Hg and/or Se with an acceptable accuracy.
Therefore, we prefer to use the 77.4 keV line from Hg-197 in our experiments to determine total
mercury as, unlike Hg-203, it is free from other interfering lines and their concomitant corrections. We
can still utilize Hg-203 for internal quality assurance whenever it is observed.
Experiments, Results and Discussion
We have been collecting air paniculate samples irom five sites across Upstate New York since
October 1991. Samples have been collected for 24 hour intervals for a period of two years. The sites are
maintained and the collectors changed weekly by the New York Department of Environmental
Conservation (NYDEC) and the Adirondack Lakes Survey Corporation (ALSC) personnel^. Automatic
dichotomous samplers with a PM-10 inlet are used to collect fine particles (< 2.5 (im) and course particles
(2.5 to 10 (am). During the 2-year study, 7300 samples (3650 fine and 3650 coarse) will be collected.
We have also analyzed vapor-phase mercury collected on charcoal sorbants at these same sites.
Samples were analyzed for their trace element content by instrumental neutron activation analysis^.
To determine Hg and other elements with long half-lives (ti/2 > 1 day), samples were irradiated at the
MTTR-H Nuclear Reactor for a period of 6- 12 hours at a neutron flux of 8 x 1012 n/sec.cm2. After 2-3
days of cooling, gamma-ray emission from the samples was measured with a high purity Germanium
detector with FWHM of about 1.75 keV for the 1332 keV line of Co-60. For the purpose of this study,
we will only present a few Hg, Se and Ta results from fine particles collected at the Moss Lake (ML) and
Perch River (PR) sites and one charcoal sample (CH) used for vapor phase mercury collection. In Table 2
we present atmospheric Hg, Se, and Ta concentrations in ng/m3 in ML and PR samples and also in our
charcoal in \iglg. In these calculations 77.4 keV, 264 keV and 1221 keV gamma-ray were used for Hg,
Se and Ta determinations, respectively. In Table 3 the areas under the 77, 279, 264 and 1221 keV
gamma-ray peaks are given.
If we apply Eq. (1), we obtain corrected 279 keV (Hg) areas, which are given in the last line of
Table 3. As seen almost all of the 279 keV intensity comes from Se-75 in the case of ML921217 and
charcoal samples. As a result one can not determine Hg using the 279 keV line. Also, for the PR
samples, the area under 77 keV Hg-197 line is about five times more intense than the corrected 279 keV
Hg-203 line. The 77.4 keV line from Hg-197 is interference-free and in many cases is th~ only way to
determine Hg by instrumental neutron activation analysis.
Unlike most other techniques INAA does not require any sample dissolution, addition of reagents
or other pre-analysis sample preparation. Therefore, it is easy also to examine materials that would be
very difficult to dissolve, such as charcoal, rock and mineral samples.
With respect to mercury determinations, the main problem is mercury loss during irradiation.
Mercury has some distinct characteristics under neutron irradiation. Mercury can readily leave the sample
matrix due to several causes: for example, hot atom chemistry, and sample heating in a reactor core. This
heating is produced from fission and also (n,p) and (n, a) reactions occurring in the sample matrix if fast
511
-------
neutrons are significantly present in the beam. To eliminate this problem a common practice is to heat seal
samples in clean pure quartz tubes. Following the irradiation, gamma-ray spectra should also be measured
with the quartz tubes intact
At MIT we use an irradiation port for Hg determination which is cooled to room temperature and
has a very low fast neutron component. MTTs reactor is unique in these respects. These properties
effectively eliminate any loss of mercury due to sample heating and there is no need to encapsulate samples
in quartz tubes.
We have successfully applied this approach to hundreds of environmental samples and have
verified the suitability of this methodology. Filtered atmospheric paniculate material and vapor-phase
mercury collectred on charcoal sorbants were packaged in vials. After the irradiations, and 2-3 days of
cooling, mercury in these samples were determined using 77.4 keV gamma ray from Hg-197. Results of
these experiments will be presented elsewhere.
References
1. L. Friberg and J. Vostal Eds., "Mercury in the Environment", CRC Press, 1972.
2. J. Jaworki, The Determination of Mercury and Its Compounds in "Effects of Mercury in the
Canadian Environments". NRG of Canada Publications. No. 16739, 188-200 1979.
3. I. Olmez and N. K. Aras, "Determination of Mercury in Selected Fishes Living in Turkish Coast by
Neutron Activation Analysis", Radiochem. Radioanal. Lett.. 22, 19-23 1975.
4. I. Olmez, Adirondacks Lake Survey Corporation Trace Metal Study, 1992.
5. I. Olmez, "Instrumental Neutron Activation Analysis of Atmospheric Paniculate Matter" in Methods
of Air Sampling and Analysis. 3rd ed., J. P. Lodge, Jr., Ed., Lewis Publishers, Inc., pp. 143-150
(1989).
512
-------
Table 1. Nuclear Reactions in Hg Determinations
Nuclear Reactions % Abundance of o tl/2
Stable Isotopes (bams) (days)
Ey of Product
Nuclei, keV
(% abundance)
196Rg (n, Y)197 Hg
(n, 7)203 Hg
74Se (n, Y)75 Se
181Ta(n,y)182Ta
0.15
29.8
0.9
99.99
3100 2.672
4.9 46.61
Interfering reactions
52 119.8
20.5
114.5
77.4 (18.2)
279.2 (77.3)
279.5 (24.7)
264.6 (58.6)
264.1 (3.6)
1189.0(16.3)
1221.4 (27.1)
513
-------
Table 2. Concentrations of Hg, Se and Ta in Moss Lake (ML) and Perch River (PR) atmospheric material
fine fractions in ng/m3 and Charcoal sample (CH) in ng/g.
Element/Sample
Hg
Se
Ta
ML921217
ng/m3
0.014 ± 0.003
1.4 ± 0.2
< 0.035
PR921209
ng/m3
0.75 ± 0.02
0.43 ±0.11
< 0.021
PR921202
ng/m3
0.14 ±0.01
1.6 ±0.3
< 0.037
CH
ng/g
29 ±5
11000±2000
280 ±60
Table 3. Areas under 77.4,279,264 and 1221 keV gamma ray peaks for the same sample
Table 2. The 279* represents corrected 279.2 keV peak area from Hg-203.
iles mentioned in
Energy
(keV)
77.4
279
264
1221
279*
Origin of
nuclide
197Hg
203Hg+75Se
75Se+182Ta
182Ta
203Hg
ML921217
Counts
414±111
1288+129
3120± 160
_
-22+166
PR921209
Counts
27896+195
6473±233
502±150
_
6260+250
PR921202
Counts
6201+217
2690±2167
3941+173
_
1035+201
CH
Counts
2910+490
56900+.450
137400±960
4600 ±270
156+780
514
-------
DETERMINING THE WET DEPOSITION OF MERCURY
F WEEKLY
ECTION OF
SAMPLES
A COMPARISON OF WEEKLY, BIWEEKLY, AND
MONTHLY COLLECTION OF PRECIPITATION
Ake Iverfeldt and John Munthe
Swedish Environmental Research Institute (IVL), P.O. Bob 47086, S - 402 58, Goteborg, Sweden.
ABSTRACT
A bulk sampling equipment for total mercury in precipitation has been used in Sweden since
1989. No significant differences in the mercury concentration in precipitation or the deposition rate
are normally found between bulk sampling and wet-only or event samplings. In 1990, a comparison
of different collection periods for the bulk sampling equipment revealed no significant difference
between weekly, biweekly, and monthly sampling.
INTRODUCTION
Today, on a regional scale, atmospheric input of mercury constitutes most of the total load on
our ecosystems. It is also well recognized that mercury is transported by air masses over long
distances.1 In the Nordic countries, a clearly decreasing south-north gradient in mercury
concentration in precipitation exists.2 In Sweden, a seasonal variability with somewhat elevated
mercury concentrations in late winter seems also to be present. 3 In the latter study, bulk collectors
were used to sample over monthly periods, from April 1989 to April 1990. The relative variability
between triplicate sets of collectors range from 2 to 12 %. A geographical distance of 1 km between
two groups of collectors was also included in this variability range. In this paper, we report a
comparison of bulk collectors3-4 operated on alternative weekly, biweekly or monthly basis.
Furthermore, a preliminary comparison of the performance of a bulk collector vs. a wet-only
collector or manual event samples is performed.
EXPERIMENTAL TECHNIQUES
The bulk collectors used have previously been described.3-4 All parts of the collector in
contact with the samples were made of borosilicate glass. The area of the bulk collector funnel was
52.8 cm2.
The wet-only collector was constructed especially for trace-metal sampling. When used for
sampling of mercury in precipitation, all parts of the collector in contact with samples were made of
glass. The area of the funnel was 373.1 cm2. The manual collection of event samples was performed
using a set-up of glass jars with large diameters, which has been described elsewhere.1 All glass
equipment was extensively cleaned according to standard cleaning procedures1.
The effect of collection period of the bulk samplers was studied at the Swedish EMEP-station
Rorvik, located on the west coast of Sweden near Goteborg, during May to October, 1991. In 1989,
the comparison of bulk collectors vs. event sampling was performed at the EMEP-station Aspvreten,
located on the east coast of Sweden, near Stockholm.
The total mercury concentrations in precipitation collected in 1991 were analyzed after BrCl
treatment of the sample followed by SnClj reduction, purging onto gold traps and dual
amalgamation Cold Vapor Atomic Fluorescence detection. An Atomic Emission Spectrometry
515
-------
(AES) detector was used in the analysis of the samples collected in 1989. A representative detection
limit for total mercury determination using CVAFS was about 0.05 ng L'1, while the precision
varied between 5 and 10 %. The detection limit for the AES-method was about 0.20 ng L"1.
RESULTS
Mean deposition rates derived from the comparison of bulk vs. wet-only collectors is given in
Figure 1. No significant difference between the bulk and the wet-only collectors is found.
January x 10
February
Figure 1. Mercury deposition rates for January and February 1989, derived from a
comparison of bulk vs. wet-only collectors or event sampling at Aspvreten.
The January value from bulk collector No. 3, is most probably a result of a contamination.
These outliers are sometimes observed when using bulk collectors. Therefore, it is highly
recommendable to use at least duplicate collectors, in order to be able to find and reject non-
representative values.
No detectable differences in mercury deposition rate is present when comparing the results
from bulk/wet-only sampling to the event based collection. The event deposition rate for February
1989 is calculated from the measured concentrations at the separate events presented in Figure 2. In
this figure, the large variability between mercury levels in various precipitation events, is clearly
demonstrated.
516
-------
Sampling date
Figure 2. Mercury concentrations in event samples collected at Aspvreten
in February 1989.
A comparison of the mercury data from different collection periods for the bulk sampling of
mercury in precipitation revealed no significant differences (Figure 3). No significant dependence
between reproducibility and collection time period was found.
May
June
July
August
September October
Figure 3. Mercury deposition rates for May to October 1991, at Rorvik, based on weekly,
biweekly, and monthly collection periods. The data is derived from a bulk
sampling equipment.
517
-------
In conclusion, the bulk samplers seems most suitable for monitoring time and/or
geographical trends in the atmospheric mercury load to our ecosystems.
REFERENCES
1. Iverfeldt, A. "Occurrence and turnover of atmospheric mercury over the Nordic countries." Water.
Air and Soil Pollution 56: 251, (1991).
2. Lindqvist, O., Johansson, K., Aastrup, M, Andersson, A., Bringmark, L., Hovsenius, G.,
Hakanson, L., Iverfeldt, A., Meili, M. and Timm, B. "Mercury in the Swedish environment -
Recent research on causes, consequences and corrective methods." Water. Air, and Soil
PollutionSS: 261 p., (1991).
3. Iverfeldt, A. "Mercury in forest canopy throughfall water and its relation to atmospheric
deposition". Water. Air and Soil Pollution 56: 553, (1991).
4. Iverfeldt, A. and Jensen, A. "Atmospheric bulk deposition of mercury to the Southern Baltic sea
area". In: Mercury as a Global Pollutant - Toward Integration and Synthesis. Watras and
Huckabee Eds., Lewis Publishers, (1993), In press.
518
-------
WET DEPOSITION OF METHYLMERCURY IN SWEDEN
John Munthe and Ake Iverfeldt
Swedish Environmental Research Institute (TVL), P.O. Box 47086, S-402 58 Goteborg, Sweden
ABSTRACT
Monthly samples of precipitation for measurement of methylmercury (MeHg) have been
collected since 1990 at several different stations in Sweden. Concentrations of MeHg are generally
found to be in the range 0.05 to 0.6 ng L'1. A significant gradient exists for MeHg deposition with
maximum average rates of 0.29 |ig nr2 yr.'1 measured in 1992 at stations on the west coast of Sweden
and 0.07 \Lg nr2 yr.'1 at the northern sites indicating transport from source areas in Northern and
Central Europe.
INTRODUCTION
Environmental contamination by Hg is mainly linked to the appearance of elevated
concentrations of MeHg in freshwater fish. In Sweden, an estimated 10 000 lakes, even in remote areas,
are severely affected making the fish caught in them unfit for human consumption. I-2 Numerous
investigations have been conducted in order to determine the causes of these excessive concentrations
and the mechanisms regulating the local, regional and global cycling of Hg.3-4 The only significant
source of Hg to remote lake systems with low mineral soil Hg content, is atmospheric deposition which
has led to a growing interest in atmospheric concentrations, transformations and deposition patterns of
Hg during the last decades.4-5-6-7 Since the late seventies, a rapid development of techniques for
sampling and analysis of mercury in air and precipitation has led to an increased understanding of the
atmospheric cycling of this element. However, it is not until recent years that methods have become
available for the measurement of MeHg in rain and other natural waters.8-9-10
The predominant form of mercury in the atmosphere is elemental mercury (Hg°).3-6 This species
is relatively stable with an estimated residence time of around one year. The most probable removal
mechanism is aqueous oxidation to water-soluble divalent forms followed by rain-out,7 although
conversion to paniculate phase Hg, followed by rain-out, may also occur. The total deposition of Hg is
not only related to conversion of gaseous Hg" since direct emissions of short-lived gaseous divalent
compounds and paniculate phase Hg from combustion activities also occurs.1'
The predominant form of Hg in fish is MeHg. This form is the most toxic Hg species present in
the environment. Recent investigations of the cycling of MeHg in a forested catchment in SW Sweden
has shown that direct atmospheric deposition to a lake and its catchment may be the dominant source of
MeHg in freshwater fish.12 In the investigated ecosystems, the total soil pools of MeHg are only a
fraction of the pools of total inorganic mercury. Despite this, the net transport of MeHg out of the
catchment is significant due to an apparent greater mobility of this species in soil,13 which could
constitute a direct route for the uptake of MeHg in fish without the need for a microbial methylation of
inorganic mercury. Although the question of the origin og MeHg in fish has not been resolved,
continued research on the atmospheric cycling of MeHg including the identification of sources,
atmospheric transformations and deposition patterns is essential for our understanding of this
environmental problem.
519
-------
EXPERIMENTAL TECHNIQUES
Wet deposition samples were collected at 7 sites in Sweden using bulk samplers. The samples
were preserved by adding 5 mL HC1 to the collection bottle. All glassware in contact with the rainwater
were acid-washed and heat treated.
Total mercury concentrations were analyzed after BrCl treatment of the sample followed by
SnCl2 reduction, purging onto gold traps and dual amalgamation Cold Vapor Atomic Fluorescence
(CVAFS) as described by Iverfeldt.5 Methyl mercury was analyzed using a technique similar to that
described by Bloom9 and Fitzgerald and Bloom.8 This technique is based on the aqueous ethylation of
MeHg Isothermal GC separation followed by pyrolysis and CVAFS detection. The lowest detectable
concentration for a 50 mL sample is about 0.05 ng L"1.
RESULTS
A frequency distribution of measured concentrations of MeHg in precipitation collected during
the period 1990 to 1992 is given in Figure 1.
Figure 1.
[MeHg], ng/L
Measured concentrations of methylmercury in precipitation in Sweden. Samples
were collected at 7 stations located in the South, East, West and North Sweden.
The concentrations found in precipitation are usually in the range 0.05 to 0.6 ng L'1. Occasional
values of over 1 ng L-1 are sometimes found but can in most cases be attributed to contamination from
e.g. insects or bird droppings. Due to this reason, only a small number of the concentrations
> 0.9 ng L"', reported in Figure 1, were used for the calculations of deposition rates. During the same
period, concentrations of total mercury were usually in the range 5 to 30 ng L"1, with occasional high
520
-------
values above 50 ng L"1. No direct correlation between the concentrations of MeHg and Hg in rain has
been observed. In Table 1, the deposition rates at the four different regions are presented.
Table I. Annual deposition of methylmercury in different
regions of Sweden, 1992.
Region
South
West
East
North
Deposition
p.g nr2 yr.'1
0.19
0.29
0.20
0.07
A clear gradient can be seen with higher deposition fluxes in the Southern, Eastern and Western
regions compared to the Northern region. A similar trend has been noted both for gaseous mercury and
deposition of total mercury.5
DISCUSSION
The sources of atmospheric methyl mercury are not known. Attempts to measure gaseous MeHg
have only been partly successful, due to the very low concentrations present in air. Recently reported
concentrations in ambient air are usually below 0.1 ng rrr3.8-14 These concentrations, if true, are
sufficiently high to explain the observed concentrations in rainwater, based on thermodynamic
considerations.
The deposition gradient of MeHg over Sweden is somewhat sharper than the corresponding
gradient for total mercury. This could lead to the suggestion that MeHg is more efficiently washed out
from air. Brosset'5 has suggested that measured concentrations of gaseous mercury in air are a function
of a uniform background concentration with events of higher concentrations related to anthropogenic
activities, such as coal burning in northern and central Europe. Very high gas phase concentrations (1 to
5 ng nr3) of MeHg have been reported from air around coal fired power plants, which indicates that this
may be a significant source.16 However, since other sources are possible, such as formation in
photochemical processes in air, further studies are needed before the origin of atmospheric MeHg can be
resolved
REFERENCES
1. Lindqvist, O., Jernelov, A., Johansson, K. and Rodhe, H. Mercury in the Swedish environment.
global and local sources. SNV PM 1816, Swedish Environmental Protection Agency, S-171 85
Solna, Sweden, 1984.
2. Hakanson, L., Nilsson, A. and Andersson, T. "Mercury in fish in Swedish lakes." Environmental
Pollution 49: 145 (1988).
3. Lindqvist, O., Johansson, K., Aastrup, M, Andersson, A., Bringmark, L., Hovsenius, O., Hakanson,
L., Iverfeldt, A., Meili, M. and Timm, B. "Mercury in th Swedish environment - Recent research
on causes, consequences and corrective methods." Water. Air, and Soil Pollution 55: 261 p.,
(1991).
521
-------
4. Lindqvist, O. (Ed.) "Mercury as an environmental pollutant". Water. Air, and Soil Pollution 56.
(1992).
5. Iverfeldt, A. "Occurence and turnover of atmospheric mercury over the nordic countries". Water. Air
and Soil Pollution 56: 251, (1991).
6. Slemr, F., Schuster, G. and Seiler, W. "Distribution, speciation and budget of atmospheric mercury."
J. Atm. Chem. 3: 407, (1985).
7. Munthe, J. "The aqueous oxidation of elemental mercury by ozone". Atmospheric Environment 26A:
1461, (1992).
8. Bloom, N.S. and Fitzgerald, W.F. "Determination of Volatile Mercury Species at the Picogram Level
by Low-Temperature Gas Chromatography with Cold - vaour Atomic Fluorescence Detection."
Anal. Chim. Acta 208: 151, (1988).
9. Bloom, N.S. "Determination of picogram levels of methylmercury by aqueous phase ethylation,
followed by cryogenic gas chromatography with cold vapor atomic fluorescence detection." Can.
J. Fish. Aq. Sci. 46: 1131, 1989.
10. Lee, Y.-H., "Determination of methyl- and ethyl-mercury in natural waters at sub - nanogram per
litre levels using SCF - adsorbent prodcedure". Int. J. Environ. Anal. Chem.29: 263, (1987).
11. Lindqvist, O. "Fluxes of mercury in the Swedish environment, contributuins from waste
incineration." Waste Management and research 4: 35, (1986).
12. Hultberg, H., Iverfeldt, A. and Lee, Y.-H. "Methylmercury input/output and accumulation in
forested catchments and critical loads for lakes in southwestern Sweden". Mercury as a Global
Pollutant - Toward Integration and Synthesis. Watras and Huckabee Eds., Lewis Publishers, 1993,
In press.
13. Lee, Y.-H., Borg G. Ch., Iverfeldt, A. and Hultberg, H. "Fluxes and turnover of methylmercury:
Mercury pools in forest soils". Mercury as a Global Pollutant - Toward Integration and Synthesis.
Watras and Huckabee Eds., Lewis Publishers, 1993, In press.
14. Fitzgerald, W.F., Mason, R.P. and Vandal, G.M. "Atmospheric cycling and air-water exchange of
mercury over mid-continental lacustrine regions." Water. Air and Soil Pollution 56: 745, (1991).
15. Brosset, C. "Total airborne mercury and its possible origin." Water. Air and Soil Pollution 17: 37,
(1982).
16. Ballantine, D.S. and Zoller, W.H. "Collection and Determination of Volatile Organic Mercury
Compounds in the Atmosphere by Gas Chromatography with Microwave Plasma Detection".
Anal. Chem. 56: 1288, (1984).
522
-------
Session 14
Oxidants in the Atmosphere
-------
PEROXYACETYL NITRATE CONCENTRATIONS AT SUBURBAN AND
DOWNTOWN LOCATIONS IN ATLANTA, GA
Benjamin E. Hartsell and Viney P. Aneja
Department of Marine, Earth, and Atmospheric Sciences
North Carolina State University
Raleigh, NC 27695-8208
Daniel Grosjean, Edwin L. Williams II, and Eric Grosjean
DGA, Inc.
Suite 205, 4S26 Telephone Rd.
Ventura, CA 93003
ABSTRACT
Ambient levels of peroxyacetyl nitrate (PAN) were measured simultaneously in Atlanta at a
downtown and at a suburban location during August 1992. PAN concentrations were typically lower and
daily PAN maxima were observed earlier at the suburban location than at the downtown location. While
maximum PAN concentrations for the entire measurement period were similar for both sites, daily PAN
concentrations showed more variability at the suburban site than at the downtown site. Diurnal profiles
for high ozone (maximum C>3>80 ppbv) and low ozone (maximum O3<80 ppbv) days for the two sites
showed that PAN concentrations were similar for the two sites during high 03 episodes. On low ozone
days, however, PAN concentrations were significantly lower at the suburban site with maxima similar to
those reported for rural Eastern U.S. sites. PAN concentrations at the suburban location for low ozone
days were similar to those reported for both a Northeastern U.S. rural site (Scotia, PA) and a Southeastern
U.S. rural site (site SONIA). On high 03 days the concentrations were similar to those of downtown
Atlanta. Higher order peroxyacyl nitrates (PPN and MPAN) were detected at the downtown location with
some regularity but were detected very infrequently at the suburban location also suggesting that
peroxyacyl nitrate precursor concentrations were more abundant in the downtown area.
Introduction
As part of the Southern Oxidants Study-Southern Oxidants Research Program on Ozone Non-
Attainment (SOS-SOPJVONA), an intensive field study was conducted in Atlanta, GA during July and
August, 1992. As part of SOS-SOPJVONA, peroxyacetyl nitrate (PAN) was measured at two locations in
the metropolitan Atlanta area. PAN is a photochemical oxidant formed by reaction of peroxyacetyl
radical with nitrogen dioxide (NO2). PAN is an eye irritant(y), a phytotoxin(2), and an important
constituent of NO., (NOy = NO + NO2 + HONO + HN03 + N205 + PAN + NO3- + organic nitrates) in
the lower troposphere, a major constituent of NOy in the middle troposphere(J), and is the dominant
component of NOy in the Arctic(-0. PAN decomposes rapidly at elevated temperatures (>25°C) and when
it does so in a NO2 deficient atmosphere will release NO2 and peroxyacetyl radicals. Thus PAN can be
an important early morning source of free radicals which quickly initiate photochemistry(5).
In this paper, PAN concentrations at a suburban location and a downtown location within
metropolitan Atlanta, GA will be examined. The measurements of PAN at two locations in Atlanta
provides a basis for a better understanding of the temporal and spatial distribution of PAN in a large
urban area. PAN and N02 concentrations, in conjunction with reaction rate constants, will be used to
estimate the peroxyacetyl radical concentrations at the suburban site and the results will be compared to
those from two rural sites in the Eastern U.S. Simple and multiple linear regression analysis of PAN on
total non-methane hydrocarbon (TNMHC), oxides of nitrogen (NOX) and 03 concentrations at the
suburban site will be used to estimate the relative dependence of the two photochemical oxidants, PAN
and 03, on their precursors. A better understanding of PAN concentrations and the relationship between
PAN and its precursors are useful in the development of models used for oxidant control.
525
-------
Experimental Methods
PAN was measured at two sites in the metropolitan Atlanta area by two different groups of
researchers. The suburban location was South Dekalb Community College and the downtown location
was the campus of the Georgia Institute of Technology. The two sites are some 20 km apart and are along
the prevailing wind direction when the winds are from the Southeast or Northwest. PAN was measured
by gas chrornatography with electron capture detection (GC-ECD) at both locations. An SRI Model 8610
gas chromatograph and a Valco Model 140 BN electron capture detector were used to measure PAN at the
downtown site and a custom-built gas chromatograph using a Valco Model 140 BN electron capture
detector were used to measure PAN at the suburban location. The two PAN analyzers are described in
detail elsewhere(6,7).
03, NO, NOX, carbon monoxide (CO), sulfur dioxide (SO2), speciated volatile organic
compounds (VOC), and standard meteorological data were also collected at the two locations. The VOC
at the suburban location were measured by NCSU but the other trace gas measurements and meteorology
were collected by the Georgia Environmental Protection Department and the Georgia Institute of
Technology (GIT).
An inter comparison between the two instruments was conducted to validate the comparison of
PAN measurements from the two PAN analyzers. The two analyzers were collocated in a trailer on the
campus of GIT and were set up to sample ambient air through identical lengths of Teflon sample line.
The two instruments had been independently calibrated using different techniques. Chemiluminescent
detection of a PAN sample obtained from the diffusion of liquid PAN into purified air was used to
calibrate the DGA instrument. Bag dilution of a high concentration gas-phase PAN sample; generated by
chlorine atom initiated oxidation of acetaldehyde in the presence of NC>2 and quantified by infrared
spectroscopy; was used to calibrate the NCSU instrument. The inter comparison showed that the two
instruments tracked each other well and were in close agreement on PAN concentration. A linear
regression of the data yielded the following results: NCSUp^N = !-33 DGAPAN' °-04 PPbv- R2 =
0.947 and n = 54. These results show the two instruments to be within 16.5 % of the mean of the
measurements which is within the combined uncertainty (15-20%) of the PAN measurements in the
ambient range found in Atlanta. (6)
RESULTS AND DISCUSSION
PAN concentrations
The composite diurnal profile for the suburban and downtown sites are shown in Figure 1. The
composite diurnal profile is obtained by averaging the PAN concentration for each specific hour of each
day over the entire measurement period. PAN exhibited strong diurnal variation at both the suburban and
downtown locations with pre-dawn morning minima often near the limit of detection and with late
afternoon maxima typically between 1600 and 1800 EDT. PAN concentrations downtown ranged from
below the limit of detection to a maximum of 2.9 ppbv and the average concentration was 0.43 ± 0.47
ppbv (n=817). Average daytime concentration (0900 to 2000 EDT) was 0.71 ± 0.53 ppbv. Daily maxima
exceeding 2 ppbv were observed on eight days. PAN concentrations at the suburban location ranged from
below the limit of detection to a maximum of 3.10 ppbv with an average concentration of 0.23 ± 0.20
ppbv. Average daytime concentration (0900 to 2000 EDT) was 0.39 ± 0.16 ppbv. Daily PAN maxima
exceeded 2 ppbv on two days and exceeded 1 ppbv on 7 days.
526
-------
PAN concentrations on high and low ozone days at the two locations
The composite diurnal profiles of the suburban and downtown sites were segregated into high
ozone (ozone concentration > 80 ppbv) and low ozone (ozone concentration < SO ppbv) days and the
results are shown in Figure 2. Both locations exhibited variation in the shape and magnitude of the PAN
diurnal profile under the two different data sets. At both locations the PAN maximum on high ozone days
was of greater magnitude as would be expected.
The increase in the maximum for the suburban site, however, was much more dramatic than that
for the downtown site. The larger increase in PAN concentrations at the suburban location on high ozone
days may be attributed to the greater variability in the abundance of PAN precursors available at the site.
Figure 3 shows the wind direction for each of the high ozone days at the suburban site. The prevailing
local surface wind direction on the high ozone days was generally from the southwest sector. Atlanta's
Hartsfield International Airport and the surrounding area of heavy anthropogenic impact is ~ 2 miles from
the suburban site in this direction and likely contributes heavily to PAN precursor concentrations when
the wind is out of the southwest The predominant wind direction at the suburban site on low ozone days
was east/southeast putting a predominantly rural low precursor source area upwind of the site. Figure 4
compares the low ozone PAN profile at the suburban site to that of a rural site in North Carolina. The
daily maximum PAN concentrations are quite similar suggesting that when the predominant wind is from
rural areas PAN profiles at the suburban site are indicative of a regional PAN background.
Prevailing wind direction is less important for the downtown location, however, because its
position near the center of the city leaves it exposed to relatively high precursor concentrations from all
directions due to the density of precursor sources surrounding the downtown area. Figure 5 shows the
average PAN daily maximum for the two prevailing wind sectors, southwest and east/southeast for both
the downtown and suburban locations. At the suburban location the average maximum for southeast wind
directions was 0.46 ppbv and for southwest wind directions increased 128.2% to 1.05 ppbv. For the
downtown site, however, the increase in PAN maxima from the southeast to southwest sectors was from
1.25 to 1.58 ppbv, a more modest 26.4 % increase. Other evidence that precursors are less abundant at
the suburban location include the lack of higher order peroxyacyl nitrates at the suburban location. Both
PPN and MPAN were detected at the downtown location but neither were detected at the suburban
location.(6)
Linear regression analysis of the relative importance ofTNMHC and NOX to PAN and O$ production at
the suburban site
Linear regression between the daily PAN maximum, the 9 AM total non-methane hydrocarbon
(TNMHC), the morning NOX maximum, and the daily 03 maximum was performed to assess the relative
importance of VOC and NOX concentrations to PAN and 03 daily maxima at the suburban site. The data
necessary to perform this analysis was only available for the suburban location at the time of this writing.
Also, the TNMHC data used is preliminary and has not undergone extensive QA/QC yet but is deemed to
be representative of the TNMHC at the suburban location. The results of individual and multiple linear
regression of PAN and 03 against TNMHC and NOX are presented in Table I. All regressions resulted in
Revalues explaining significant portions of the variability in the PAN and 03 daily maxima and all were
tested for significance and found to be significant at the 99% confidence level.
PAN and 03 daily maxima were highly correlated to one another as would be expected from
their similar photochemical origins. The correlation of the two photochemical oxidants to TNMHC and
NOX showed interesting differences, however. R^ values for the regression of PAN against TNMHC and
NOX indicate that 40% of the variability in PAN daily maxima can be explained by the variability in the
morning TNMHC levels at the suburban site. Morning NOX concentrations, however, explain 60% of the
variability in the PAN maxima. A multiple regression of PAN against NOX and TNMHC only increases
the explained variability to 61%. It seems then, that the availability of NOX at the suburban location is
more influential on the day's PAN maxima than the TNMHC concentrations.
527
-------
O3, on the other hand, is more dependent on the morning TNMHC concentrations than NOX
availability. TNMHC account for 72% of the variability in the daily 03 maxima while adding NOX to the
regression only increases the variability explained to 77%. It is also interesting to note that the intercept
for O3 in all of the O3 regressions is between 25 and 30 ppbv indicating that this is the background level
of O3 in the suburban Atlanta environment.
CONCLUSIONS
PAN concentrations were generally higher at the downtown location than at the suburban
location and were fairly consistent throughout the study. At the suburban location, however, observed
PAN maxima on days where the O3 concentration exceeded 80 ppbv were similar to those of the
downtown location. On days where O3 did not exceed 80 ppbv PAN concentrations were much lower
than downtown concentrations and resembled those reported for rural areas in the Eastern U.S. The
position of the suburban site near the edge of the metropolitan area subjected it to different upwind
precursor source areas depending on the day's prevailing wind direction. These differing precursor
availabilities strongly influenced the PAN concentration for that day. When the suburban site was
downwind of more rural areas the PAN diurnal profiles resembled those reported for rural sites in the
Eastern U.S. When areas of heavier anthropogenic impact were upwind of the suburban site, however, the
PAN diurnal profiles and daily maxima were similar to those observed regularly at the downtown
location.
Both PAN and O3 daily maximum concentrations were regressed against the morning NOX and
9am total non-methane hydrocarbons (TNMHC) and were found to be strongly correlated to each of these
variables and to each other. PAN was more strongly correlated to NOX while O3 was more strongly
correlated to TNMHC. Multiple regression of PAN against NOX and TNMHC morning maxima
explained 61% of the variability of daily PAN maxima while multiple regression of O3 against NOX and
TNMHC morning maxima explained 77% of the variability in the daily O3 maxima.
The similarity of PAN concentrations suggests that the low ozone day PAN concentrations from
the suburban Atlanta location and the average PAN concentrations reported from rural Eastern U.S. sites
may be indicative of a regional background PAN profile. The high ozone day PAN concentrations at the
suburban location and the regularly higher PAN concentrations at the downtown location are thought to
be indicative of urban 'spikes' of higher PAN concentrations within the regional PAN background.
ACKNOWLEDGEMENTS
This research has been funded by the U.S. Environmental Protection Agency through a
cooperative agreement (S 9153) with the University Corporation for Atmospheric Research as a part of the
Southern Oxidants Study - Southeast Oxidant Project on Ozone Non-Attainment (SOS-SORP/ONA). We
thank Dr. W. Lonneman (U.S. EPA) for providing the PAN analyzer and his advice regarding PAN
measurements, Mr. E.L. Williams, II and Mr. E. Grosjean (DGA, Inc.) for their support and cooperation
during the field study, Mr. M. Rodgers, Mr. J. Wilbum, students and staff of the Georgia Institute of
Technology, Mr. Z. Li and Ms. M. Das for their help in the field, Mr. D.S. Kim for his assistance and
useful discussions, and Ms. B. Batts for the preparation of the draft and final version of the manuscript.
DISCLAIMER
The contents of this document do not necessarily reflect the view and policies of the U.S. Environmental
Protection Agency, the University Corporation for Atmospheric Research, nor the views of all members of
the Southern Oxidants Study Consortia, nor does mention of trade names or commercial or non-
commercial products constitute endorsement or recommendation for use.
528
-------
Table 1.
Regression Equation R'2 n
PAN(pptv) = -0.98CTNMHC) + 6.05(NOx) +19.97 (O3) -531 0.77** 25
PAN(pptv) = 0.42(TNMHC) + 10.17(NOX) -0.48 0.61** 25
PAN(pptv) = 1.73(TNMHC) + 195 0.40** 25
PAN(pptv)=11.88(NOx) + 28.46 0.60** 25
PAN(ppbv) = 0.0199(O3) - 0.47 0.69** 25
03 = 0.07(TNMHC) + 0.21(NOX) + 26.54 0.77** 25
03 = 0.49(NOX) + 31.40 0.59** 25
O3 = 0.097(TNMHC) + 30.5 0.72** 25
O3 = 0.0608(TNMHC) + 20.8(PAN) + 26.46 0.86** 25
** significant at the 99% confidence limit
Diurnal profile of PAN at the
suburban and downtown locations
0 2 4 6 8 10 12 14 16 18 20 22
Tin«e,EDT
Diurnal profile of PAN at the
suburban location on high and low
ozone days
l.SOy
3" 0.50 •• -i**"**^*..
0 2 4 6 8 10 12 14 16 18 20 22
Time, EOT
Figure 1. Diurnal profile of PAN at the suburban Figure 2a. Diurnal profile of PAN at the suburban
(open squares) and downtown (dark squares) location on high ozone (open squares) and low
locations for the entire measurement period. ozone (dark squares) days.
Diurnal profile of PAN at the
downtown location on high and low
ozone days
1.50 y
0 2 4 6 8 10 12 14 16 18 20 22
Time, EOT
0
330345 15 3Q
315 , \ \ / / >. 45
300 ^\A V / //O, 60
285 ^^%&^ 1S
270 i i ijclllj^i i i 90
255 r~~^/*^,JSl^^ 105
240 /ff \\\^ 12°
225 ^ / / \ \ 135
210195 165150
180
Figure 2b. Diurnal profile of PAN at the
downtown location on high ozone (open squares)
and low ozone (dark squares) days.
Figure 3. Wind rose showing the day's prevailing
wind direction at the suburban site for the 7 days
on which the daily PAN maximum at the
suburban location exceeded 1.0 ppbv.
529
-------
Diurnal profile of PAN at the suburban
Atlanta and rural NC locations
0.60-j
0.50
1.0.40;
£ 0.30!l
1 0.20
o.io L
0.00 I i i i i i i i i i i i i i I i i i i
0 2 4 6 8 10 12 14 16 18 20 22
Time, EDT
Suburban Downtown
Location
Figure 4. Diurnal profile of PAN at a rural North Figures. Comparison of average daily PAN
Carolina site (dark squares) and at the suburban maxima for southwesterly winds (open columns)
Atlanta site (open squares) on low ozone days. and for southeasterly winds (dark columns) for
both the suburban and downtown locations.
REFERENCES
1. E.R. Stephens, Adv. Environ. Sci. Technol. 1969,1,119.
2. O.C. Taylor, J. AirPollut. Control Assoc. 1969, 19, 347-351.
3. B.A. Ridley, J.D. Shelter, B.W. Gandrud, L.J. Salas, H.B. Singh, M.A. Carroll, G. Hubler, D.L.
Albritton, D.R. Hastie, H.I. Schiff, G.I. Mackay, D.R. Karecki, D.D. Davis, J.D. Bradshaw, M.O.
Rodgers, S.T. Sandholm, A.L. Torres, G.L. Gregory, S.M. Beck, Journal of Geophysical Research..
1990, 95, 10179-10192.
4. J.W. Bottenheim, A.J. Gallant, K.A. Brice, Geo. Res. Let. 1986.13. 113-116.
5. W.P.L Carter, A.M. Winer, J.N. Pitts, Jr., Envir. Sci. Technol.. 1981,15, 831.
6. E.L. Williams, II, E. Grosjean, D. Grosjean, "Ambient levels of the peroxyacyl nitrates PAN, PPN, and
MPAN in Atlanta, GA", Journal of the Air and Waste Management Association., accepted, January,
1993.
7. B.E. Hartsell, V.P. Aneja, W.A. Lonneman, "Characterization of peroxyacetyl nitrate PAN at the
rural Southern Oxidants Study site in central Piedmont North Carolina, site SONIA", Journal of
Geophysical Research, submitted, March, 1993.
8. M. Trainer, M.P. Buhr, C.M. Curran, F.C. Fehsenfeld, E.Y. Hsie, S.C. Liu, R.B. Norton, D.D
Parrish, E.J. Williams, Journal of Geophysical Research.. 1991, 96, 3045.
530
-------
TRENDS AND ANALYSIS OF AMBIENT NO AND NOy CONCENTRATIONS IN RALEIGH, NORTH
CAROLINA
Deug-Soo Kim and Viney P. Aneja
Department of Marine. Earth and Atmospheric Sciences, North Carolina State University,
Raleigh, NC 27695 -8208
ABSTRACT
Ambient concentrations of NO and NOy as well as 03 and CO were measured during August 19 to September 1,
1991 in downtown Raleigh, North Carolina as a part of the Southern Oxidants Study-Southern Oxidants Research
Program on Ozone Non-Attainment (SOS-SORP/ONA). These measurements were made in an effort to provide insight
into the characteristics of nitrogen oxides and their role in the formation of ozone in the urban environment. NO and
NOy showed bimodal diurnal variations with peaks in the morning (06:00 - 08:00 EST) and in the late evening (21:00 -
23:00 EST). These peaks at this urban site correspond to the coupled effects of rush hour traffic and meteorological
conditions (i.e., variation of mixing height and dispersion conditions). The overall average NO and NOy concentrations
were found to be 6.1 ±5.4ppbv(range: 0 to 70 ppbv) and 14.9 ±8.1 ppbv (range: 0.3 to llOppbv), respectively.
Average daily maxima of NO and NOy (18.3 ppbv and 27.4 ppbv) occured during the morning. 03 showed a diurnal
variation with a maximum in the afternoon between 14:00 and 16:00 EST: mean concentration 20 ± 10 ppbv (range: 1 to
62 ppbv). Maximum 03 and CO concentrations during weekdays result from NO and CO emitted from mobile sources
during the morning rush hour. Background CO concentration at Raleigh was 470 ppbv. A strong correlation between
CO and NOy (r2 = 0.34) was observed. The ratio of CO to NOy (~ 16) at the Raleigh site suggests that mobile sources
are the major contributor to NO and NOy concentrations at the site.
INTRODUCTION
In the lower atmosphere, nitric oxide (NO) is emitted mainly from anthropogenic sources, i.e., stationary and mobile
sources ('•*'. NO, in turn, is converted to nitrogen dioxide (NOi) by reaction with ozone (03) and peroxy radicals (R02).
ROa radicals are produced mostly by the reaction of hydroxyl radical (OH) with reactive hydrocarbons, and the photolysis
of aldehydes which have both natural and anthropogenic origins. NO2 is then photolyzed in the atmosphere, and the
atomic oxygen released combines with molecule oxygen (62) to form 03.
It is believed that a significant traction of the 03 in the lower troposphere is photochemicaUy produced in situ (3,4)
For the past 20 years efforts to reduce ozone concentration, especially in urban areas, have focused on control of
anthropogenic hydrocarbon emissions. There is, however, no clear evidence that the expected decrease in ozone
concentration has been achieved. Over sixty areas in the United States remain classified by the U.S. Environmental
Protection Agency (EPA) as ozone non-attainment areas; twenty four of these areas are in the South. It is not clear
whether the failure to decrease 03 concentration is in the basic control strategy, or an inability to reduce total nonmethane
hydrocarbon (NMHC) emissions by the necessary amount to reduce 03 concentration. The interaction between nitrogen
oxides, NMHC and their intermediate oxidation products is complex. In generally, the relationship between the
formation of ozone and its precusors is non-linear. This non-linearity makes the development of an effective ozone
control strategy difficult Recent work suggests that emissions of other precursor, such as those of the nitrogen oxides,
should be considered in ozone control strategy (5,6). There is also experimental evidence to suggest that ozone
production in the troposphere is limited by the availability of NOX (= NO + NO2) W. Completion of a successful budget
for tropospheric ozone, therefore, requires additional information on the fate and distribution of nitrogen oxides species
(Ridley and Robinson, 1992).
In this section diurnal patterns in the ambient concentrations of NO and total reactive nitrogen species (NOy = NO +
NO2 + NO3 + N2O5 + HNO3 + PAN + HNO2 + NO3' + organic nitrates) measured during late summer at Raleigh,
North Carolina are presented. Temporal variations in the concentration of these 03 precursors is directly related to
mobile sources associated with the morning and evening rush-hour traffic. Statistical correlations between NOy and CO
are also discussed for their source relationsip in Raleigh. This work is part of the Southern Oxidant Study-Southern
Oxidants Research Programs on Ozone Non-Attainment (SOS-SORP/ONA) sponsored by the U.S. EPA. The
SORP/ONA focuses on elucidating the processes responsible for the formation of ozone and other photochemical oxidants
in urban and industrial centers in the Southern United States O. Providing NO and NOy data measured as input and
boundary conditions for photochemical model contributes to develop an effective 03 control strategy.
METHODOLOGY
Sampling site
531
-------
The sampling site is in an open field (~ 1600 m2) on the St. Augustine College campus, which is less than one mile
north of the center of downtown Raleigh, North Carolina (35.9 °N, 78.7 °W, 126.8 m MSL). Raleigh is the capital of
North Carolina and most of the state government office buildings, legislature, and capital buildings are concentrated in a
one square mile area. In additions, a number of municipal buildings and several federal government buildings are
present. Several major state highway arteries criss-cross the downtown area leading to substantial traffic flow throughout
the day as well as during the peak rush hours. Located within 3 miles of the downtown area are several colleges, one
university, the county hospital, as well as a number of county public schools and shopping complexes. Surrounding inner
Raleigh is a four-lane highway complex approximately 10 miles in diameter. This outer loop services traffic from several
interstate highways and local commuters. Ther are no industrial, or municipal power emission sources within 10 nu'less
of the downtown area.
Analvtical instrumentation
Ambient NO and NOy concentrations as well as meteorological parameters (temperature, wind speed and wind
direction) were monitored during the period August 19,1991 to September 1,1991. The hourly averaged concentrations
of CO and 03 were measured the same site by North Carolina Department of Environmental Health and Natural
Resources (NC DENHER). All species were sampled at ~ 10 m height above the ground and all instruments for NO,
NOy, 03 and CO measurement were kept in a temperature controlled mobile laboratory.
NO and NOy were measured simultaneously using a modified two channel commercial chemiluminescent NO/NOX
detector (TECO 14B/E, Thermo Electron) (8'9>. The instrument is based on the reaction between NO and reagent 03
which produces a chemilumunescence that is detected by a photo multiplier tube. The analyzer has an internal heated
(-375 °C) molybdenum catalyst which converts most NOy species such as NO2, HNO3, N2O5 and PAN to NO (*'9\
N2O, N2, and NH3 are not converted to NO to any appreciable degree. For the laboratory operating conditions, the
instrument detection limit was 100 parts per trillion volume (pptv) for both NO and NOy ("•»). Instrument calibration
was accomplished by standard addition of NO in N2 (Scott Specialty Gases, Inc., Plumsteadville, PA) via a Multigas
Calibration System Model 146 (Thermo Environmental Instruments Inc.: Franklin, MA).
RESULTS AND DISCUSSION
Diurnal Variations of NO, NOV and O%
The mean of NO, NOy andO3 during this period were 6.1 ± 5.4 ppbv (n = 306), 14.9 ± S.lppbv (n = 307) and 20.1 ±
10.4ppbv(n = 312), respectively. Concentrations of NO and NOy ranged from less than 1 ppbv to 70 ppbv for NO and
from 1 ppbv to 110 ppbv for NOy. 03 concentrations ranged from 1 ppbv to 62 ppbv. The meteorology parameters were
indicative of late summer at this location with relatively hot afternoon temperatures of-30 °C, relatively low wind speeds
averaging about 1.5 m/sec, and prevailing surface winds from southwest.
Composite diurnal profiles of NO, NOy and 03 are shown in Figure 1. 03 displayed the typical diurnal variation
with maximum concentration in the afternoon between 14:00 and 16:00 EST, and minimum concentration early in the
morning (~ 06:00 EST). The increase in 03 during the morning hours coincides with the decrease in NO concentration.
This behavior is consistent with the photochemical production of 03 from locally emitted precursor species. The
decrease in O3 concentration in the evening probably is the result of dry deposition under the subsiding boundary layer
and titration with NO. The average daily maximum concentration of 03 was 36.4 ± 14.7 ppbv.
Peak in the concentrations of NO and NOy coincided with the morning and evening rush-hour traffic (Figure 1). The
morning peak was relatively brief lasting between 06:00 - 08:00 EST. Average daily maximums of NO and NOy during
this period werel8.3 ppbv and 27.4 ppbv, respectively. The peaks associated with evening rush hour started around
18:00 EST, reaching a maximum at 21:00 EST. Average daily maximums of NO and NOy between 21:00 - 23:00 EST
were 11.5 ppbv and 25.8 ppbv, respectively. The ratio of NO to NOy during the morning peak periods (~ 0.7) was higher
than the ratio for the remainders of the day (ranged from 0.1 to 0.5). NO is emitted primarily by mobile sources during
these peak periods and is converted to N02 in the air. Therefore, one would expect a higher contribution of NO to NOy
during these high NO emssion periods. Lower mixing depth and reduced dispersive conditions during these morning
and night peak periods may also contribute to the increase of NO and NOy for a short time period.
Similar diurnal patterns of NO and NOX have been observed at urban sites such as West London, Glasgow and
Billingham in the U.K. (10), for Atlanta, GA in U.S.A. (data analysis from database of Summer 1990 Atlanta Ozone
Precursor Study conducted by U.S.EPA). In our measurements, NO maximum during evening hours may be associated
with meteorology and chemistry, as well as direct emission from rush hour traffic. Concentrations of both NO and NOy
increase gradually after the evening rush hour (-17:00 EST) and into night (-23:00 EST). The abundance of NO emitted
from rush hour traffic may directly increase NO concentration in ambient air. However, no NO peak was found at the
evening rush hour. This is attributed to immediate loss of NO by reaction with 03 which is abundant during the evening
rush hour. The decrease of 03 after rush hour is thought to be evidence of this fact (Figure 1). The steady increase of
NO until 23:00 EST may indicate the presence of NO sources (probably continued automobile exhaust during the evening)
at the site during the period.
532
-------
To examine the contribution of NO emitted from the rush hour mobile traffic to the NO peak concentration, diurnal
variation of NO. NOy and 03 on weekdays and weekends are presented in Figure 2. On weekends, there were no
significant variations of NO and NOy concentrations. Only a relatively small variation of 03 between day and night were
found. Both NO and NOy mean concentrations were significantly reduced on weekends (0.3 ppbv for NO and 4.8 ppbv
for NOy) from the daily means for the entire measurement period (6 ppbv for NO and 15 ppbv NOy). On the other hand.
significant diurnal variations of NO. NOy and 03 were observed on weekdays. NO and NOy concentrations during
daytime (10:00 to 18:00 EST) were low with average concentrations of NO and NOy less than 0.5 ppbv for NO and about
5 ppbv for NOy. Average NO and NOy for the rest of hours were about 10 ppbv for NO and 20 ppbv of NOy. Peak NO
concentration on weekdays was two orders of magnitude higher than peak NO concentration on weekends. Low levels of
NO and NOy during the day is probably due to photochemistry as well as enhanced dispersive conditions. NO reacts with
peroxy radicals which are formed mostly from the oxidation of CO and hydrocarbons by OH attack and is converted to
NO2. Mixing height also is increased and well-mixed conditions exist during daytime. On the other hand, during night
time, there is no photochemistry and more stable atmospheric conditions exist. A decrease of NO concentration is
observed from 02:00 to 05:00 EST. This decrease is mainly due to the reaction of NO with ozone.
During morning rush hours, the NOy peak occurs about 1 hour later than that for NO; however, NO and NOy
maximum are coincident during nighttime. Examination of NO and NO2 daily peaks at urban cities in the U.K. ('")
showed a similar relationship between daily NO and NOy peaks. The coincidence of NO and NOy maxima during
nighttime reflects the dominant effects of primary emissions of NO and the production of NO2 by ozone titration of NO
together with, possibly, catalytic ambient oxidation at low temperature. These reactions are fast and will not therefore
cause a significant delay between nighttime NO and NOy peaks. During the morning rush hour, however, photochemical
oxidation processes involving OH attack on hydrocarbons to form peroxy radicals are of increased importance. A
contribution from the formation of pcroxv radicals could be a reason for the delay between NO and NOy peaks, as found
previously for the Central London Site ('1).
Daily averaged maximum of 03 concentration on weekdays was found to be —10 ppbv higher than that on weekends.
High NO concentration on weekday mornings may result in greater NO2 concentration by reaction with peroxy radicals at
the site. The level of CO and hydrocarbons in such an urban area may provide enough hydroxyl radical concentration to
oxidize NO. Approximately 1 ppmv of CO concentration was observed during weekdays morning hours. Thus, this
increase of the maximum 03 on weekdays is thought to be due to the high NO emission locally emitted during weekday
mornings.
CO trends and their relation to NOY and 0$ concentrations
Hourly averaged CO concentrations have little variations in the Raleigh measurements. The hourly averaged CO
concentrations during the measurement period vary from 0.2 ppmv to 2.6 ppmv, and the mean and standard deviation was
0.7 ± 0.1 ppmv. CO data measured at the site show the average CO level is higher in the morning and through the night
than during day time. Minimum CO concentration occurs around 16:00 EST when photochemical activity is the
strongest during the day. The minimum value was considered as CO background concentration at the site.
Examination of hourly averaged CO concentrations during the measurement period show apparently different
emission level of CO during morning rush hours. CO concentrations above than 1.2 ppmv were observed during morning
rush hours of the 22nt', 23"' and 29"1 August, 1991 while CO concentrations below than 0.6 ppmv were measured on the
rest of the measurement days. CO should effect on 03 chemistry through oxidation involving radicals and nitrogen
oxides in ambient air (12). Atmospheric chemistry behavior including 03 and NOX may differ depending on CO
concentration, i.e., under high CO conditions and under low CO conditions. High CO conditions are defined for this
study as days where the morning peak time CO concentration is above than 1.2 ppmv. Low CO conditions are defined as
days where the morning peak time CO concentration is below than 0.6 ppmv. The mean of CO concentration during the
high CO conditions (1.2 ± 0.4 ppmv) is about twice that during low CO conditions (0.5 ± 0.1 ppmv). Mean NO and NOy
concentrations in high CO conditions (15.7 ± 14.2 ppbv for NO and 31 ± 20 ppbv for NOy) are significantly higher than
those in low CO conditions (1.2 ± 1.8 ppbv for NO and 6.7 ± 2.9 ppbv for NOy).
High NO concentrations coincident with high CO concentrations are expected because both CO and NO have similar
anthropogenic sources (mainly mobile sources) in urban areas. To examine source relationship between CO and NO,
correlation between CO and reactive NOy concentration measured at Raleigh was shown in Figure 3. A linear regression
of hourly average CO and NOV concentrations was performed and showed a strong correlation during the measurement
period ([CO] = 16.3[NOy] + 470; r2 = 0.53). The regression curve reveals a background CO concentration of 470 ppbv
in the Raleigh urban area. It is not surprising that the background of CO at the Raleigh urban site is much higher than
the value of 200 ppbv in clean air region of the nothern hemisphere*13', because the measurements were made near the
downtown of the city where significant anthropogenic sources for CO are expected. The ratio of CO concentration to
NO., concentration at Raleigh urban site (which is the slope of the regression curve in Figure 3) was 16.3. Average
emissions of CO, NOX and SO2 in eastern United States using NAPAP Emission Inventory for summer were summarized.
and the average emission ratio of CO and NOy in eastern United States during summer was reported as 4.3 <14>. A high
533
-------
emission ratio between CO and NOV (CO/NO,, > 8) indicates mobile sources are dominant while low emission ratios
(CO/NOV < 1) indicate that point sources are dominant. The ratio of 16.3 observed at Raleigh urban site is close to the
emission ratio of CO to NOy (16.5) from only the mobile source dominant Denver metropolitan area. Thus, the high
ratio of CO to NOy at Raleigh urban site suggests that the urban plume in Raleigh is impacted mainly by mobile sources.
Figure 4 shows the diurnal variation of each species for both high and low CO conditions. Maximum ozone in high
CO conditions is about 20 ppbv higher than during the low CO conditions even though levels of NO and NOy in both
cases are similar during daytime hours with NO concentrations of about 1 ppbv and NOy concentrations of about 5 ppbv.
The increase of maximum ozone concentration in high CO conditions is influenced by the following photochemical
factors: (1) significantly higher ambient NO concentration during morning peak time in high CO conditions since NO and
CO share sources and the NO is oxidized to form N02; and (2) high CO levels, since ozone can be produced by oxidation
of CO in sufficient NO concentration. High CO conditions, therefore, are conducive to greater 03 formation leading to
greater 03 maxima. On the other hand, minimum ozone levels are lower in high CO conditions than in low CO
conditions because ozone is destroyed rapidly by reaction with the higher NO concentrations.
The response of daily maximum 03 concentration to maximum NOy concentration, and the ratio of CO cou, •ntration
to NOy concentration ([CO]/rNOy]) at morning peak hours are shown in Figure 5. Daily maximum 03 concentrations in
the Raleigh urban areas were found to be a function of daily maximum NOy concentration and the ratio of CO
concentration to NOy concentration during the measurement period. Positive correlations between maximum 03
concentration and maximum NOy concentration (r^ = 0.34) and between maximum 03 concentration and [CO]/(NOy) (r^
= 0.47) were observed. These observational-based results suggests that local production of NO from mobile sources at
this location contributes to the photochemical production of 03.
SUMMARY AND CONCLUSIONS
The mean NO and NOy concentrations were found to be 6.1 ± 5.4 ppb (ranged from 0 to 70 ppbv) and 14.9 ± 8.1 ppbv
(ranged from 0.3 to 111 ppbv), respectively. Diurnal patterns of NO and NOy observed at Raleigh site is similar to those
were observed at urban cities in the U.K. <10> and Atlanta, GA in the U.S.(Atlanta Ozone Precursor Study, U.S. EPA,
1990). The peak of NO and NOy in the morning was coincident to morning rush hours at the site. The average
contribution of NO to NOy concentration (NO/NOy) was ~ 20 % during the entire measurement period. The high
contribution of NO to NOy concentration (~ 70 %) during morning rush hours was attributed to mobile sources in the
urban area.
Different diurnal patterns of NO and NOy between weekdays and weekends was found and mainly due to emission
from mobile sources. On weekends no morning peak was observed and both NO (0.3 ppbv)and NOy concentrations (4.8
ppbv) were significantly reduced from daily mean concentrations for the entire measurement period (6 ppbv for NO and
15 ppbv for NOy). Daily averaged maximum of 03 concentration during weekdays was found to be 10 ppbv higher than
that during the weekends. Slightly variations of hourly averaged CO concentrations was observed during entire
measurement period (0.2 - 2.6 ppmv; mean and standard deviation = 0.7 ±0.1 ppmv). However, relatively high CO
concentrations above than 1.2 ppmv during the morning rush hours were observed. Increase of maximum 03 in
weekdays and high CO conditions may result from increase of photochemical formation of 03 due to high NO and CO
level during morning rush hours at the site.
Strong correlation between CO and NOy concentrations (r^ = 0.53) was found during measurement period. The
background CO concentration was 470 ppbv. The observed ratio of CO to NOy concentration at Raleigh urban site, 16.3,
was close to the emission ratio of CO to NOy (16.5) which was observed from the mobile source dominant Denver
metropolitan area ^\ This ratio suggests that the air mass at Raleigh site is characterized by mainly mobile sources.
Daily maximums of 03 were correlated to daily maximums of NOy (r^ = 0.34), and the [C01/[NOy) (r^ = 0.47) during
the measurement period. These observational-based results suggest that local production of NO from mobile sources at
this location contributes to the photochemical production of 03.
The results from Raleigh urban site demonstrate that the emission of NO from mobile sources during morning rush
hours may be an important source of atmospheric NO concentration, and that this locally produced NO concentration may
increase the maximum O3 during the day. Without hydrocarbon data analysis, however, there are uncertainties in the
discussion of photochemical production of 03. Thus it is clear that more comprehensive data analysis of nitrogen species
as well as hydrocarbons emitted in urban areas are needed to enhance our understanding of 03 photochemistry in the
urban environment. It is that the analysis and discussions of the nitrogen oxides measurements made at Raleigh site will
contribute to the development of improved methodologies for characterizing the causes of ozone non-attainment in a given
urban area.
Acknowledgements- We acknowledge Dr. W. Robarge, NCSU and Dr. H. Jeffries, UNC at Chapel Hill and members of
our Air Quality group. Benjamin Hartsell. Zheng Li and Mita Das for their assistance and discussions on atmospheric
oxidants: and Ms. B. Batts in the preparation of the manuscript.
534
-------
Disclaimer- The contents of this document do not necessarily reflect the views and policies of the U.S. Environmental
Protection Agency, nor does mention of trade names or commercial or non-commercial products constitute endorsement
or recommendation for use.
REFERENCES
1. J. A. Logan, M. J. Prathe. S. C. Wofsy et al.. "Tropospheric chemistry; A global perspective". J. Geoohvs. Res 86:
7210(1981).
2. J. A. Logan, "Nitrogen Oxides in the troposphere: Global and Regional Budgets". J.Geopvs. Res.. 88: 10785 (1983).
3. W.L Chameides and J.C.O. Walker "A Photochemical Theory of Tropospheric Ozone". J. Geophvs. Res. 78: 8751
(1973).
4. J. Fishman, S. Solomon and P.J. Crutzen "Observational and Theoretical evidence in support of a significant in situ
Photochemical source of Tropospheric Ozone", Tellus 31: 432 (1979).
5. J. B. Milford. A. G. Russell and McRae G. J. ."A new approach to photochemical pollution control: Implications of
spatial patterns in pollutant responses to reductions in nitrogen oxides and reactive organic gas emissions". Environ. Sci.
Technol.. 23: 1290(1989).
6. C. A. Cardelino and W. L. "Chameides Natural hydrocarbons, urbanization, and urban ozone", J. Geophvs. Res. 95:
13971 (1990).
7. University Corporation for Atmospheric Research, Southern Oxidant Study Report. August 1990.
8. A.C. Delany, R.R. Dickerson. Jr. F.L Melchior et al. "Modification of commercial NOX detector for high sensitivity",
Rev. Sci. lustrum.. 53: 1899 (1982).
9. R.R. Dickerson. A.C. Delany and A.F. Wartburg "Further modification of commertcial NOX detector for high
sensitivity". Rev. Sci. Instrum. 55: 1995 (1984).
10. J. S. Bower, F. J. Broughton. M. T.Dando et al. "Urban NO2 concentrations in the U.K. in 1987", Atmos. Environ.
256:267(1991).
11. M. L. Williams, G. F. J. Broughton. J. S. Bower et al., "Ambient NOx concentrations in the U. K. 1976 - 1984 a
summary", Atmos. Environ. 22: 2819 (1988).
12. B. J. Finlayson-Pitts and J. N. Pitts Atmospheric Chemistry. Wiley-Interscience Publication,John Wiley & Sons, New
York, 1986.
13. P. Warneck, Chemistry of the natural atmosphere. Academic Press. Inc., San Diego, California. 1988.
14. D.D. Parrish, M. Trainer. M.P. Buhr. B. A.Watkins et al., "Carbon monoxide concentrations and their relation to
concentrations of total reactive oxidized nitrogen at two rural U.S. sites".J. GeopEv.Res. 99: 9309 (1991).
i 40 r
I r ...-~*
I30- - •
i 2° p*. •* «• i\ •
I'oK^'.. ... >«
s nr \^""""^
1 3 5 7 9 11 13 15 17 19 21 23
* NO — » — NOy — - • O3
Figure 1. Composite diurnal profiles of NO, NOy and
O3 at Raleigh, N.C
S 40 (a) Weekday, ^
IT 3or- .' 'r ."•.
p^-iX.....-:*^
1 3 5 7 9 11 13 15 17 19 21 23
Time (1ST)
* — NO •- NOy — *~- O3
1 40 (b) Weekends
2 T 30 J.J»-»*-»..
f £ in L _*•*•»»•* *-» i<*
s & •* ++** "*
1 »(y^n. ,„„.„,.?; :^5Sr
1 3 5 7 9 11 13 15 17 19 21 23
TJme(EST)
• — NO •" NOy —f— O3
Figure 2. Composite diurnal profiles of NO, NOy and
O3 for (a) weekdays and (b) weken* at Raleigh, N.C
535
-------
10000 £ [CQJ = 16 3* p^) + 470
12 = 0.53
S 1000 t
8 •
100
o o oo oco>oo o
ooocco
0.1
10
N0y
-------
Gaseous Hydrogen Peroxide Concentrations in the
Central Piedmont Region of North Carolina
Mita Das and Viney P. Aneja
Dept. of Marine Earth and Atmospheric Sciences
North Carolina State University
Raleigh, N.C. 27695-8202.
ABSTRACT
Gas-phase total peroxides and hydrogen peroxide were monitored in the ambient
air at Candor, N.C. during the summer of 1991 and 1992. Gas phase Hqp2
concentration measured from below the level of detection (-0.1 ppbv) to about 2.2 ppbv.
Concentrations of hydrogen peroxide showed a diurnal variation with maximum
concentrations in the afternoon (—1.5 ppbv) and was found to be affected by the
concentrations of other trace gases as well as meteorological parameters. Decrease of gas
phase concentrations was observed during wet atmospheric conditions. Evaluation of the
data with respect to meteorological conditions and the concentrations of other pollutants
showed that H2®2 concentrations were associated with long-range transport of polluted
air masses from the surrounding areas. Evidence of night time H2®2 resulting from
breakdown in the stability of the nocturnal boundary layer was found. Linear regression
analysis of hourly average, daytime average and daily maximum 63 on 112^2 indicates
that at this rural site the formation of H2^2 's favored by increasing ozone and an ozone
concentration of ~ 18 ppbv for 1 ppbv of H2C>2 was obtained.
INTRODUCTION
Gas-phase total peroxides and hydrogen peroxide were monitored in the ambient
air at Candor, N.C. as part of Southern Oxidant Study (SOS) project SONIA funded by
the U.S. Environmental Protection Agency during mid July to mid August in 1991 and the
month of June to early July in 1992. The measurements were made using continuous
dual-channel flourometric analyzer based on the horse radish peroxidase method (1).
Hydrogen peroxide is a secondary photochemical oxidant formed by HO2 radical
recombination. Its production is highly sensitive to the competing reaction of NO and
HO2 and thus its production rate is a complex function of the level of photochemical
activity and air mass age (2). Modelling studies have shown that the generation of H2O2
is affected by the presence of atmospheric pollutants such as NOx, VOC, and CO, and by
meteorological parameters such as solar radiation, temperature, and water vapor content
(2-4). It is also highly soluble in water (Henry's Law constant H= -10^ M/atm ) and thus
is easily removed from the atmosphere by wet deposition, including rain scavenging, fog
and mist droplet scavenging (5).
537
-------
In this paper, the role of meteorology on the diurnal variation of HjO2 is
examined. The relationship between 11^2' a"d chemical and physical processes is
examined on the basis of observational based statistical analysis. Finally, the K2^2
concentrations are compared to ozone levels at this rural site .
RESULTS AND DISCUSSION
A statistical summary of H}Q2 concentrations for both years is given in table 1.
H2C>2 exhibits a distinct diurnal trend with significantly higher day time H2C>2 levels than
the night time levels. The daytime mean as well maximum H2O2 levels are higher in 1991
than 1992. But the median is same for both years indicating that during the 1991
measurement period few episodic levels were observed. Fig 1. shows the composite
diurnal profile of both H22 as well as ozone for the entire measurement period during
'91 and '92. The smooth rise during the morning to the afternoon is in accordance with
expected photochemical production. Peak concentrations in both ozone as well H2C>2
were reached during the late afternoon hours (1400-1600 EST) and the minimum was
observed about 0500-0800 EST.
Though the maxima and minima are colocated in time, there is a considerable
amount of background ozone surviving the night whereas H2O2 levels drop to the
detection limits. Also the fall in the H2O2 concentrations is more rapid and takes place
earlier when compared to ozone. Low night timeH2O2 levels have been observed in the
past by many researchers and cannot be accounted for by just surface losses due to dry
deposition. One explanation put forward by Hastie et al., (6) is that during the night time,
aqueous aerosols, formed as a result of water condensing on atmospheric particles at the
onset of nocturnal inversion, efficiently scavenge the highly soluble H2O2- We found a
strong anticorrelation between H2O2 concentrations and the relative humidity ( r= -0.83
and -0.77 during '91 and '92 respectively) and the composite diurnal profile is given figure
2. From an examination of figure 2, it is clear that H2O2 begins to rise only after the
relative humidity has fallen significantly low ~ 70-65% and the rapid fall is coincident with
increasing relative humidity. This apparent higher sensitivity of H2O2 to relative humidity
is explained by its higher solubility. This points to the fact that after sunset, due to the
radiative cooling of the earth's surface, the increased relative humidity effectively
scavenges the gas phase H2O2 trapped in the shallow nocturnal boundary layer. Thus the
concentration of gaseous H2O2 trapped within the inversion is lost by dry as well as
deposition and the concentration H2O2 measured falls without replenishment from above.
Occassionally, however, a secondary night time peak in H2O2 accompanied by a
secondary peak in ozone was observed (figure 3). From an examination of the H2O2
profile together with meteorology, the peak was found to be the result of a weak
nocturnal boundary layer. Downward mixing of undepleted air from aloft under such
conditions can result in high night time concentrations. Evidence of decrease in gas phase
H2O2 concentrations during rain, fog or cloud formation was also found. This has been
attributed to the reduction in solar flux by clouds and also to cloud scavenging of H2O2
due to its high solubility (7).
Linear relationship between H2O2 and other measured physico-chemical variables
were examined. It was found that among all the environmental parameters, H2O2 was
538
-------
most strongly negatively correlated to relative humidity as already explained above. It was
also found to have a strong correlation to solar radiation and temperature (r=0.61 and
0.87 and r=0.68 and 0.67 respectively for '91 and '92). This is consistent with modelling
studies (2,4). High solar radiation levels result in intense photochemical reactions of
chemical species resulting in the formation of free radicals including HO2- ^2^2 was
found to be strongly correlated to ozone (r=0.78 and 0.63 respectively for '91 and '92).
This is not surprising as both are photochemical products in air. Also, in a clean
atmosphere, the photolysis of ozone is the major source of radicals (2), and thus is the
source of HC>2 radicals. During 1992 no significari Correlation between H2O2 and other
primary pollutant species like NOx, CO and SO2 were found. This is expected because
the primary pollutant concentrations are low at this rural site. However, during the 1991
measurement period significant negative correlation between NOx and H2O2 was found
and a surprising positive correlation between H2O2 and SO2 was observed (r=0.45). A
positive correlation between ozone and SO2 was also found (r=0.56) (Fig 4.). This
suggests the role of long range transport of pollutants to this rural site. An examination
of 48 hour back trajectories of the air masses arriving at the site showed significatly higher
peroxide concentration when the air mass was continental in origin than when the air mass
originated over the oceans.
Hourly averaged, daytime hourly averaged and daily maximum 03 was regressed
on H2O2- The results are given in table 2. The slopes obtained from the regression
equation are comparable for both the years indicating that similar conditions favoring the
formation of both ozone and H2O2 prevailed during the two years. For the daytime
averaged data - 18 ppbv of ozone for 1 ppbv of H2O2 was obtained. Also when the daily
maximum was considered, the R.2 obtained is considerably higher. This indicates that the
formation of H2O2 is favored by increasing ozone concentrations in this rural area and
the conditions favoring the formation of ozone also favor the formation of H2O2 .
CONCLUSIONS
H2O2 concentrations exhibited a diurnal profile with maxima during the late
afternoon indicative of photochemical formation in the atmosphere. The minima which
was close to the detection limit of the instrument was observed about 0500-0800 EST
coincident with high relative humidity. It was reasoned that during the night time, in the
presence of a nocturnal inversion, aqueous droplets efficiently scavenged gaseous H2O2
and the concentration dropped due to wet deposition and lack of replenishment from the
mixed layer. However, during conditions of a weak nocturnal boundary layer, downward
mixing of undepleted air from aloft tended to restore the concentration.
H2O2 was found to be highly correlated to solar radiation, temperature and ozone
among the various environmental parameters measured. During the 1991 measurement
period, a high correlation between H202 and SO2 as well as ozone and SO2 was found
indicating the role of long range transport of pollutants to this rural site.
Ozone was regressed on hourly averaged, daytime average and daily maximum H2O2-
When daytime hourly average ozone was regressed on H2O2, it was found that for the
daytime average data -18 ppbv of ozone was obtained for a ppbv of H2O2- Regression of
daily maximum ozone on H2O2 gave a high R.2 indicating that at this rural site the formation
of H2O2 is favored by increasing ozone.
539
-------
ACKNOWLEDGEMENTS
This reasearch has been funded by the U.S. Environmental Protection Agency through a
cooperative agreement (S 9153) with the University Corporation for Atmospheric Research as part of the
Southern Oxidants Study - Southeast Regional Oxidant Network (SOS-SERON). We thank Mr. Eric
Ringler for the QA/QC'd data on trace gas measurements, Dr. Eric Edgerton for the meteorological data
and Mr. Ben Perry for the use of his farmland for our measurement site. We acknowledge the assistance
and support provided by the members of our Air Quality group Mr. Deug Soo Kim, Zheng Li and
Benjamin Hartsell.
DISCLAIMER
The contents of this document do not necessarily reflect the view and policies of the U.S.
Environmental Agency, the University Corporation for Atmospheric Research, nor the views of all
members of the Southern Oxidants Study Consortia, nor the mention of trade names or commercial or
non-commercial products constitute endorsement or recommendation for use.
REFERENCES
1. Lazrus, A.L., G.L. Kok, J.A. Lind, S.N. Gitlin, E.G. Heikes, and R.E. Shelter, 1986, Automated
fluorometric method for H2O2 in air, Anal. Chem. 58, 594-597.
2. Kleinman, L.I., 1986, Photochemical formation of peroxides in the boundary layer, J. Geophys. Res.,
91, 10,889-10,904.
3. Calvert, J.G., and W.R. Stockwell, 1983, Acid generation in the troposphere by gas-phase chemistry,
Environ. Sci. Tech., 17, 428A-443A.
4. Dodge, M. C., (1989), A comparison of three photochemical oxidant mechanisms, J. Geophys.
Res., 94, 5121-36
5. Waleck, C.J., 1987, A theoretical estimate of O3 and H2O2 dry deposition over the Northeast United
States, Atmos. Environ., 21, 2649-2659.
6. Hastie, D.R., P.B. Shepson, S. Sharma and H.I. Schiff, 1993, The influence of the nocturnal boundary
layer on secondary trace species in the atmosphere at Dorset, Ontario.
7. Thompson A. M. and Cicerone R. J., 1982, Clouds and wet removal as causes of variability in the
trace- gas composition of the marine troposphere. J. Geophys. Res., 87, 8811-8826.
Table 1. Statistical summary of measurements made at Candor during '91 and '92
Period
1991
Days
Nights
1992
Days
Nights
N
296
156
140
303
164
139
Mean
0.41
0.64
0.16
0.35
0.52
0.15
SD
0.51
0.56
0.18
0.36
0.36
0.22
Median
0.21
0.42
0.10
0.23
0.43
0.10
Max
2.19
2.19
0.94
1.60
1.60
1.05
Min
-------
Table 2. Regression analysis of hourly average, daytime and daily max H2O2 and
ozone.
Hourly Average
Daytime
Daily Maximum
CANDOR 1991
Regression Eqtn
R2
n
O3=25.26+26.78H2O2
0.60
293
O3=32.79+19.14H2O2
0.57
107
O3=29.17+20.57H2O2
0.69
13
CANDOR 1992
Regression Eqtn
R2
n
O3=34.06+29.19H2O2
0.39
301
O3=43.69+18.23H2O2
0.35
120
O3=42.14+21.42H2O2
0.48
8
1991
H202(ppbl
03(ppb)
1992
1.00-,
Time (EST)
* H202 1
• — 03
Figure 1. Composite diurnal variations of H202 and ozone.
541
-------
1.2
i"
I 0.6
a°-4
0.2
0
100
95 •
90 i
85 i
80 |
75 !
70 i
65 '
60
TIME (est)
l.oo
,0.80
j 0.60
I 0.40
I
' 0.20
0.00
TIME (est)
Fig 2. Figure showing the inverse relationship between H2O2 and Relative Humidity
Fig 3. Figure shoving the secondary night time peak in H2O2 and ozone
H202 (ppbv)
6.00
5.00
4.00
3.00
2.00
1.00
0.00
•
-•—•-
r=0.56
0.00 20.00 40.00 60.00 80.00
OZONE (ppbv)
Fig 4. Positive correlations between H2O2 andSO2 and ozone andH2O2
542
-------
Session 15
FTIR and Remote Sensing
-------
FIELD TESTING OF TWO VOC EMISSION RATE ESTIMATION METHODS
Ray E. Carter, Jr.
Dennis D. Lane
Glen A. Marotz
Department of Civil Engineering
4002 Learned Hall
University of Kansas
Lawrence, KS 66045
Mark J. Thomas
Jody L. Hudson
U.S. EPA, Region VII
25 Funston Road
Kansas City, KS 66115
Charles T. Chaffin
Tim L. Marshall
Mark R. Witkowski
Robert M. Hammaker
William G. Fateley
Department of Chemistry
Kansas State University
Willard Hall
Manhattan, KS 66506
ABSTRACT
Two methods of estimating the VOC emission rate are described. The methods are being field tested at a flat,
grass-covered site by the University of Kansas, in cooperation with Region VII of the U.S.EPA and Kansas State
University. Initial testing is conducted using a single point source; later phases of the study will focus on simulated and
actual area sources, in an attempt to extend applicability to Superfund sites. Both methods use path-integrated VOC
concentrations and a form of the integrated Gaussian dispersion equation to produce an estimated rate. Both methods
attempt to overcome problems associated with the direct application of the Gaussian equation through their use of
meteorological data: Model 1 uses means of wind speed and wind direction for the duration of the test period; Model 2
uses one-minute means of those variables. Based on data from all tests, Model 1 provides an increase in emission rate
estimation accuracy over the direct application of the integrated Gaussian equation; Model 2 provides a slight additional
increase. Based on data from only those tests during which the assumptions used to derive the integrated Gaussian
expression were violated, Model 1 provides a significant increase in estimation accuracy; Model 2 provides a significant
additional increase. Therefore, the models described in this paper allow the integrated Gaussian equation to be applied
more accurately in cases where its underlying assumptions are violated. Estimation accuracy is better under class B
stability conditions than under more stable conditions; Model 2 provides increased accuracy over Model 1 as conditions
become more unstable. Model performance is better at 100 and 150 meters downwind distance than at 50 meters.
Model 1 provides a significant improvement over the direct application of the integrated Gaussian equation at 150
meters; Model 2 provides a significant additional improvement.
INTRODUCTION
The University of Kansas (KU) has assisted Region VII of the U.S.EPA in the development of a volatile organic
compound (VOC) monitoring capability during the last several years'. As a part of that work, KU and Region VII have
field tested and refined a method consisting of whole-air sampling in evacuated stainless steel canisters and gas
chromatographic (GC) analysis. In addition, KU has assisted in the field testing of an open-path Fourier Transform
Infrared spectroscopic (FTIR) method developed by Kansas State University (KSU) and Region VII2. Results have
shown both methods to be viable for ambient air VOC monitoring.
KU, in cooperation with KSU and Region VII, has undertaken to extend the capabilities of the two methods by
using them to estimate VOC emission rates from various types of sources, with emphasis on the applicability to
Superfund sites. To accomplish this goal, KU is conducting field tests of techniques that use VOC measurements and
meteorological data in conjunction with a form of the Gaussian dispersion equation to estimate the emission rate3.
545
-------
The study is divided into three phases, which consist of field testing the emission rate models using only a
single point source during Phase 1, using multiple point sources to simulate an area source during Phase 2, and at
selected actual VOC area and/or point sources during Phase 3. The three phases are being performed consecutively over
a three-year period.
The objectives of the study include the following: (1) the development of field protocols to ensure that data
collected will be of appropriate type and quality to perform emission rate estimations; (2) an assessment of the
performance of the emission rate models for both the single-point-source and area-source cases, including a
determination of statistically appropriate confidence intervals; (3) a determination of the applicability and relative
accuracy of the models as a function of downwind distance and atmospheric stability; and (4) a determination of the
effect of VOCs being released at different heights within an area source.
Data collection for Phase 1 is scheduled to be completed in June, 1993. Data collected through November,
1992, are reported and analyzed in this paper. Because of space constraints, only results obtained for 1,1,1-
trichloroethane (1,1,1-TCA) are interpreted and discussed in detail in this paper. Results for other compounds are
summarized here and discussed in more detail elsewhere4'3.
MODEL DESCRIPTION
Two emission rate models are being evaluated for the single-point-source case in Phase 1. Both models are
based on the premise that integration of the Gaussian dispersion equation in the crosswind direction results in an
expression for the emission rate as a function of the path-integrated concentration3. The models differ in their use of
meteorological data: Model 1 uses values for wind speed and wind direction that are averaged over the duration of the
test period; Model 2 employs one-minute means of those variables4'6. Other models, or modifications of the above
models, will also be evaluated should the results warrant that action.
Justification for using one-minute means of meteorological data in an emission rate model is provided in a
previous study, in which meteorological data were used to estimate the crosswind distribution of material within a VOC
plume6. Based on the results of that study, the use of one-minute means yields a significant improvement in estimating
both the shape of the concentration-versus-crosswind distance curve and the magnitude and location of the highest
concentrations, when compared to the use of meteorological data averaged over the entire test period.
During Phase 1, the effective emission height and the measurement height are both approximately two meters.
With the source and receptors very near ground level, the applicable form of the Gaussian dispersion equation is as
follows • : C(x^ = (Q/TOy0.iU)expr.i/4(y/0.y)Z]j (Equation 1)
where C(x,y) = concentration at (x,y), in mg/m3,
Q = emission rate, in mg/sec,
jy,jz = horizontal and vertical dispersion coefficients, in m,
u = mean wind speed, in m/sec, and
x,y = downwind and crosswind distances to receptor, in m.
Integrating with respect to y, from y = -oo to y= + oo, and rearranging yields
Q = [(2ir)''i/2]Cy(r!u, (Equation 2)
where Cv = crosswind path-integrated concentration, in mg/m2.
This method should produce a reasonably accurate value for the emission rate, given an accurate value of the crosswind
path-integrated concentration and the satisfaction of assumptions made in the development of the above equations.
The assumptions used in developing the Gaussian equation (Equation 1) are seldom rigorously adhered to in
field tests. However, various sensitivity analyses show that meeting some assumptions is more critical than others. For
example, over short diffusion times (approximately one minute) and within small distances (the first tens of meters), the
downwind distribution of material should take the same form as the wind-fluctuation distribution, which approximates a
Gaussian distribution fairly closely9'10. In addition, although the assumption of a Gaussian vertical profile is often
deviated from, especially under convective conditions, deviations are not so serious that it fails as a working
assumption10.
Equation 1 was integrated with respect to y to derive Equation 2. This operation involved an assumption that x
(and, consequently, a, and crj was held constant. Therefore, the path-integrated concentration must be determined for a
546
-------
path normal to the mean wind direction. In addition, the integration was from -oo to +00; thus, all of the plume
constituents at the measurement height must pass within the boundaries of the sampling network. Because of fluctuating
winds, those assumptions are often violated, and in those cases the direct application of Equation 2 will not yield an
accurate emission rate.
The models described in this paper attempt to overcome this problem by using meteorological data to
characterize the configuration of the plume during the test period, allowing Equation 2 to be used more accurately.
Consider Equation 1 in a rewritten form: _ ,_ ... . , . . . , ,21
H Cu/Q = (l/irffyffj exp[-^(y/ay)2]
Summation of Cu/Q values for evenly spaced points along the measurement path yields a relationship between path-
integrated concentration, wind speed, and emission rate for a given stability class and sampling network orientation:
yffJ expl-'^Cy/o-^Vn = E(Cu/Q)/n « Cyu/Q
Values for Cyu/Q can be calculated under the ideal-case conditions used to derive Equation 2 and summed across the
path to yield (Cyu/Q)j. Values can also be calculated using the measured wind direction data from the test period and
summed to yield (Cyu/Q)M.
The ratio of these two values can be used as follows: Cyl = CyM (Cyu/Q), / (Cyu/Q)M
where Cy, is the path-integrated concentration that would be observed under the ideal-case conditions and CyM is the
measured path-integrated concentration. Cy, can then be used in Equation 2 to more accurately estimate the emission
rate. (Cyu/Q)M can be determined either by using means of wind speed and wind direction for the entire test period, or
by using one-minute means of those variables. These two methods of determining (CyU/Q)M give rise to the two
emission rate models alluded to previously4.
Dispersion coefficients (ay and oj were determined from Pasquill-Gifford stability classifications, which were
estimated from the standard deviation of the horizontal wind direction (a0). It is likely that ay and crz can be more
accurately determined by other methods. One such method is the use of a tracer to develop site-specific coefficients.
Another method is to determine ay and
-------
Overall performance of Equation 2, Model 1, and Model 2 for 1,1,1-TCA, toluene, and for a third data set that
includes all other compounds released, is shown in Figure 1. Results displayed in Figure 1 are based on canister-derived
path-integrated concentrations, as generated using a previously developed technique6. Estimation accuracy is expressed
as a percentage of the measured emission rate. Prior to calculating the mean accuracies plotted in Figure 1, outliers
were removed from the data set according to a method by Dixon12. Based on the values shown, Model 1 provides a
slight increase in estimation accuracy over Equation 2; an additional increase in accuracy is provided by Model 2.
Slightly better estimation accuracies are seen for toluene than for the other compounds.
Overall performance of Equation 2, Model 1, and Model 2, as based on FTIR measurements, is shown in
Figure 2. Results are similar to those obtained from canister data, although FTIR estimation accuracy is slightly lower
for 1,1,1-TCA and toluene, and slightly higher for other compounds.
The sensitivity of the models to atmospheric stability can be explored by examining Figures 3 and 4, which
show mean estimation accuracies for tests conducted under class B, C, and D stability conditions. Results indicate that
model accuracy for 1,1,1-TCA is better for tests conducted under class B stability conditions than for those conducted
under more stable conditions. In addition, Model 2 provides increased accuracy over Model 1 as conditions become
more unstable.
Through November, 1992, data were collected at three downwind distances to investigate the influence of this
variable on the emission rate determination. Figure 5, based on canister data, shows that model performance is better at
100 and 150 meters than at 50 meters. Model 1 provides a significant increase in model accuracy over the direct
application of Equation 2 at 150 meters; Model 2 provides an additional significant increase in accuracy. Less difference
among the mean estimation accuracies for the three methods is observed at 50 and 100 meters. Figure 6, based on FTIR
data, shows a slight improvement at 100 meters over 50 meters; note that no FTIR data has been collected at 150
meters.
Analyses of variance were performed in an attempt to provide additional insight into the effects of stability and
downwind distance on each of the estimation methods. Results of these analyses indicated the following: (1) all three
methods show greater sensitivity to stability than to distance, (2) a small contribution to the variation is made by the
interaction of stability and distance; and (3) Model 2 is more sensitive to both stability and distance than are Equation 2
or Model 1. The fact that stability, distance, and the interaction between the two variables all contribute to the variation
observed implies that caution must be used in making general conclusions about model sensitivity based on the limited
data set thus far collected.
Equation 2 was derived using the following two assumptions: (1) the path-integrated concentration must be
determined for a path normal to the mean wind direction; and (2) all of the plume constituents at the measurement height
must pass within the boundaries of the sampling network. In cases where those assumptions are violated, the direct
application of Equation 2 will not yield an accurate emission rate. Model 1 and Model 2 are being evaluated during
Phase 1 of this study to determine the extent to which this problem can be overcome through the use of meteorological
data. Therefore, it would be informative to look more closely at those tests in which one or both of the underlying
assumptions were violated. The overall performance of the estimation methods for those tests is shown in Figure 7 for
1,1,1-TCA. (Note that Assumption 1 was considered to be violated if the mean wind direction was 15 degrees or more
different from the network centerline direction. Assumption 2 was considered to be violated if less than 95% of the
plume constituents at the measurement height passed within the sampling network, based on the meteorological data.)
As seen in Figure 7, Model 1 provides a significant increase in mean estimation accuracy over the direct
application of Equation 2. Model 2 provides a significant additional increase, its mean estimation accuracy being
approximately 17% greater than that of Equation 2 for canister results (approximately 15% greater for FTIR results). In
addition, Model 2 provides large increases in estimation accuracy for a number of individual tests5. Therefore, results
indicate that the use of meteorological data in the models described in this paper allows Equation 2 to be applied more
accurately in cases where its underlying assumptions are violated. Furthermore, the use of one-minute means of
meteorological data (as in Model 2) provides a significantly greater increase in estimation accuracy than does the use of
meteorological data that are averaged over the entire test period (as in Model 1).
CONCLUSIONS
During Phase 1 of this study, two models that estimate VOC emission rate from a single point source, using
path-integrated concentrations and meteorological data, are being evaluated. The performance of Model 1 (which uses
meteorological data that are averaged over the entire test period) and Model 2 (which uses one-minute means of
548
-------
meteorological data) are compared to the direct application of Equation 2 (the integrated Gaussian dispersion equation).
The sensitivity of the models to atmospheric stability and downwind distance is also being evaluated. Results analyzed in
this paper produce the following conclusions:
1. Based on data from all tests, Model 1 provides an increase in emission rate estimation accuracy over the direct
application of Equation 2. Model 2 provides a slight additional increase in accuracy.
2. Based on data from only those tests during which the assumptions used to derive Equation 2 were violated,
Model 1 provides a significant increase in estimation accuracy; Model 2 provides a significant additional
increase. Therefore, the models described in this paper allow Equation 2 to be applied more accurately in cases
where its underlying assumptions are violated.
3. Estimation accuracy is better for tests conducted under class B stability conditions than for those conducted
under more stable conditions. Model 2 provides increased accuracy over Model 1 as conditions become more
unstable.
4. Model performance is better at 100 and 150 meters downwind distance than at 50 meters. Model 1 provides a
significant improvement over the direct application of Equation 2 at 150 meters; Model 2 provides a significant
additional improvement. Much less difference among the estimation methods is observed at 50 and 100 meters.
REFERENCES
1. G.A. Marotz, D.D. Lane, R.E. Carter et a!., "Difficulties and efficiencies encountered with a bulk-air sampling
scheme for characterizing heavy gas releases over a grass-covered site," in Proceedings of the EPA/APCA
Symposium on Measurement of Toxic Air Pollutants. Raleigh, 1987, pp 508-518.
2. M.L. Spartz, M.R. Witkowski, J.H. Fateley et al., "Comparison of long path FT-IR data to whole air canister
data from a controlled upwind point source," in Proceedings. EPA/A&WMA International Symposium on the
Measurement of Toxic and Related Air Pollutants. Raleigh, 1990, pp 685-692.
3. R.L. Scotto, T.R. Minnich, and M.R. Leo, "A method for estimating VOC emission rates from area sources
using remote optical sensing," in Proceedings. EPA/A&WMA International Symposium on the Measurement of
Toxic and Related Air Pollutants. Durham, 1991, pp 698-703.
4. R.E. Carter, D.D. Lane, G.A. Marotz et al., "VOC emission rate estimation from FTIR measurements and
meteorological data," in Proceedings. EPA/A&WMA International Symposium on Measurement of Toxic and
Related Air Pollutants. Durham, 1992, pp 601-606.
5. R.E. Carter, D.D. Lane, G.A. Marotz et al., "Estimation of VOC emission rates, using path-integrated
concentrations and meteorological data," to be presented at the 1993 A&WMA Annual Meeting, Denver, paper
no. 93-W9-102.08.
6. R.E. Carter, D.D. Lane, M.J. Thomas et al., "A method of predicting point and path-averaged ambient air
VOC concentrations, using meteorological data," Journal of the Air and Waste Management Association. 43(4):
pp 480-488 (1993).
7. D.B. Turner, Workbook of Atmospheric Dispersion Estimates. Office of Air Programs, Environmental
Protection Agency, Research Triangle Park, 1970, pp 5-30.
8. S.R. Hanna, G.A. Briggs, R.P. Hosker, Jr., Gaussian plume model for continuous sources, Handbook on
Atmospheric Diffusion. Technical Information Center, U.S. Department of Energy, 1982, pp 25-31.
9. C.C. Lin, W.H. Reid, Handbuch der Phvsik. Vol. 8, Part 2, Springer-Verlag, Berlin, 1963, pp 438-523.
10. G.A. Briggs, Analysis of Diffusion Field Experiments, Lectures on Air Pollution Modeling. A. Venkatram and
J.C. Wyngaard, Eds., American Meteorological Society, Boston, 1988, pp 63-80.
11. R.E. Carter, M.J. Thomas, G.A. Marotz et al., "Compound detection and concentration estimation by open-
path FTIR and canisters under controlled field conditions," Environmental Science & Technology. 26(11): pp
2175-2181 (1992).
12. F.J. Rohlf, R.R. Sokal, Statistical Tables. 2nd edition, W.H. Freeman and Co., New York, 1981, pp 211-212.
549
-------
Figure 1. Estimation Accuracy
vs. Compound (from canister data)
1.1.1-TCA Toluone Other Compounds
Figure 2. Estimation Accuracy
vs. Compound (from FTIR data)
1,1.1-TCA Toluene Other Compounds
Figure 3. Estimation Accuracy vs.
Stability for 1,1,1-TCA (canister)
Class C Class D
Figure 4. Estimation Accuracy vs.
Stability for 1,1,1-TCA (FTIR)
Class B Class C Class D
Figure 5. Estimation Accuracy vs.
Distance for 1,1,1-TCA (canister)
50 Meters 100 Meters 150 Meters
Figure 6. Estimation Accuracy vs.
Distance for1,1,1-TCA (FTIR)
Figure 7. Accuracy for Tests in which
Assumptions Were Violated (1,1,1-TCA)
i 2 • Model 1 B Model 2 \
550
-------
Ambient Air Monitoring Siting Criteria for Open Path Analyzers
Measuring Nitrogen Dioxide, Ozone, and Sulfur Dioxide.
Lee Ann B. Byrd
United States Environmental Protection Agency
Office of Air Quality Planning and Standards (MD-14)
Research Triangle Park, North Carolina 27711
ABSTRACT
Open path analyzers using ultraviolet spectroscopy are commercially available as ambient air
monitoring instruments. Definitive siting criteria would be needed in order to incorporate these
instruments into the State and local ambient air monitoring networks. This paper explores the issues
involved with establishing siting criteria and discusses proposed criteria which parallel existing
requirements for conventional analyzers. Specifically, this paper discusses open path monitor siting for
nitrogen dioxide, ozone, and sulfur dioxide (the three pollutants which are measurable by ultraviolet-
based open path analyzers) that also have National Ambient Air Quality Standards.
551
-------
INTRODUCTION
A new method for monitoring pollutants in ambient air has been developed and introduced to the ambient air
monitoring community.1 This new monitor is a type of open path analyzer that is referred to as a differential optical
absorption spectrometer (DOAS). The DOAS uses an ultraviolet light beam to measure atmospheric pollutant
concentrations in situ over a path of several meters up to several kilometers. In contrast, traditional monitoring
methods, or "fixed point analyzers', measure gaseous pollutant concentrations by extracting a sample of air from the
atmosphere through an inlet probe.
A major regulatory use for ambient air monitoring data is for comparison of measured concentrations against the
National Ambient Air Quality Standards (NAAQS). The NAAQS define levels of air quality which have been
determined by the U.S.EPA as necessary to protect public health and welfare from adverse effects from exposure to
specific air pollutants.2 Six air pollutants, often referred to as the "criteria pollutants", currently have established
NAAQS. The criteria pollutants are ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide, lead, and paniculate
matter (PM-10). Of these six pollutants, ozone, nitrogen dioxide, and sulfur dioxide are measurable by open path
analyzers using ultraviolet based spectroscopy.3
In order for open path ambient air monitoring data to be used for regulatory purposes, appropriate siting criteria
must be developed for these instruments and included in the appropriate regulation. Designing these criteria in a way
that they parallel the existing requirements for fixed point monitors would be prudent to ensure that the open path data
would be compatible with the existing State and local air monitoring data bases. This paper will discuss potential siting
criteria for the DOAS analyzers for measuring ozone, sulfur dioxide, and nitrogen dioxide.
SITING CRITERIA DESIGN
The existing ambient air monitor siting requirements are defined within the air quality surveillance regulations in
terms of monitoring objectives and geographic area, or "spatial scale of representativeness".4 The four basic monitoring
objectives for a given area are to determine maximum concentrations, population exposure, specific source impact, and
background concentrations. The geographic location of each monitoring station within this area is dependent upon the
monitoring objective for that site. To clarify this link between matching a monitoring objective with the appropriate
geographic location, the concept of spatial scale of representativeness, or measurement scale, is introduced. Spatial
scales of representativeness are defined in terms of the physical dimensions of the atmospheric volume to be represented
by a monitoring station. It is assumed that the pollutant concentrations are reasonably homogeneous throughout this air
volume; therefore, a point monitor placed within this air volume would produce concentrations representative of that
measurement scale, provided that the monitor were sited according to specific criteria.
Six measurement scales, with their associated physical dimensions, are defined in the regulations. Of these six
scales, four are listed in the regulation as normally associated with ambient air monitoring of ozone, nitrogen dioxide,
and sulfur dioxide. These scales are described as middle scale, neighborhood scale, urban scale, and regional scale.
Figure 1 illustrates the measurement scales used most frequently by the State and local air monitoring programs.5 Those
measurement scales which are used to monitor ozone, nitrogen dioxide, and sulfur dioxide are identified on Figure 1
beneath the measurement scale dimensions. For the majority of existing ozone and nitrogen dioxide monitoring
applications, the scales of representativeness are neighborhood or urban scale. Both ozone and nitrogen dioxide are
photochemically formed, rather than emitted directly from local sources, and their concentration gradients tend to be
more homogeneous and better represented by the larger scales of measurement. Sulfur dioxide emissions are
predominantly from large point sources; therefore, sulfur dioxide ambient air monitoring is usually conducted on the
smaller scales, particularly middle and neighborhood scales.
Assuming that the pollutant concentration within a particular measurement scale is nearly homogeneous, it
would seem that a path integrated measurement taken within the dimensions of that measurement scale would provide a
concentration descriptive of that area of interest. Open path analyzers appear to provide a more spatially representative
measurement than traditional, fixed point methods because of their inherent path integration measurement capabilities.
Conversely, this spatial averaging technique may not be advantageous when significant nonhomogeneities exist within a
given air mass and the monitoring objective is to obtain the maximum concentration at a certain point. This trade-off
between capturing maximum pollutant concentrations at a point and measuring spatially averaged concentrations is a
fundamental difference between point and open path technologies. This difference should be considered when choosing
the monitoring technology for application in areas with significant concentration inhomogeneities.
552
-------
To ensure that an open path monitor would collect data that is representative of the intended monitoring
objective, specific path siting criteria would be needed. These new criteria should be designed to achieve the same
objectives as the current requirements in order to allow the use of open path data interchangeably with point data within
the same monitoring network. The existing criteria6 could be adapted to accomplish this task. For ozone, nitrogen
dioxide, and sulfur dioxide monitoring, these criteria could be written based on the following siting considerations,
which are analogous to current regulations.
Definition of "Path"
The definition of "path" used within these criteria is the actual path in space through which the analyzer
measurement beam passes and over which the pollutant concentration is measured and averaged. This definition of
"path" may differ from the optical path length used by the analyzer to calculate the path averaged concentration. The
optical path length would be described as the total distance the light beam travels between the transmitter and the
receiver. (For example, when a retroreflector is used, the optical path length is the distance from the transmitter to the
retroreflector plus the distance from the retroreflector to the receiver.)
Horizontal and Vertical Path Placement
For ozone, sulfur dioxide, and nitrogen dioxide fixed-point analyzers, current regulations require that a
monitoring probe be located 3 to 15 meters above ground. The corresponding height above ground requirement for open
path analyzers could require that 80% of the path must meet the 3 to 15 meters above ground criteria. Similarly, in
regard to the horizontal and vertical distance from the path to any supporting structure, the criterion for open path
analyzers could be that 90% of the total path must be one meter, horizontally or vertically, away from any supporting
structure.
Spacing from Minor Sources
The proposed siting criterion which addresses distancing the path from minor sources would be similar to the
current requirements. For ozone open path analyzers, at least 90% of the path should be away from minor sources of
nitrogen oxides. Similarly, 90% of the path used to measure sulfur dioxide should be away from minor sources of
sulfur dioxide, such as furnace or incineration flues.
Spacing from Obstructions
This criterion applies for any open path analyzer, regardless of the pollutant measured. Any path must be clear
of all trees, brush, buildings, plumes, dust, or other optical obstructions, including potential obstructions that may move
due to human activity, wind, or growth of vegetation. This criteria is especially important for open path analyzers
because of their optical requirements. Also, at least 90% of the path should have unrestricted airflow and be located
away from buildings and other obstacles. Obstacles would be defined as any building, tree, or other object that
protrudes above the path at a height greater than or equal to half the horizontal distance from the path to the object.
Spacing from Trees
Trees can provide surfaces for adsorption of ozone, nitrogen dioxide, and sulfur dioxide, and they can obstruct
wind flow. To reduce interference from trees, 90% of the path must be 10 meters or more from the drip line of trees or
brush. Additionally, for any tree(s) that may be considered an obstruction, the path must meet the "Spacing from
Obstructions" requirements listed previously.
Spacing from Roadways
Due to the reactive nature of ozone and nitrogen dioxide when combined with mobile source emissions, certain
limits must be placed on paths that are in the vicinity of roadways or similar heavily trafficked areas. It would be
preferable if the path did not cross any roadway; however, practical siting conditions make this seem unrealistic. Most
ozone and nitrogen dioxide monitors would be placed in areas with small, lightly traveled residential streets in rural or
suburban areas. A more practical requirement would be to allow a path to cross a roadway with an average daily traffic
count of 10,000 or fewer vehicles per day.
Additionally, some consideration should be given to paths in close proximity to larger roadways. Table 1 lists
553
-------
the minimum separation distances between roadways and probes as they are defined in existing regulation. These
minimum separation distances, which vary in size with traffic count, are used to limit the probe's siting within the "area
of potential interference* from vehicle emissions. Similarly, for open path analyzers, it is reasonable to assume that a
portion of the path, while not wholly crossing a roadway of greater than 10,000 vehicles per day, could be located closer
to a major roadway than the minimum separation distance. A criterion to limit this interference would be that 90% of
any path must meet the same minimum separation distances specified for fixed point monitoring probes located near a
roadway. This criterion would be included in addition to the limitation that a path must not cross a roadway with greater
than 10,000 vehicles per day.
Table 1. Minimum separation distance between neighborhood
& urban scale ozone & nitrogen dioxide probes & roadways*
Roadway average daily traffic,
vehicles per day
£10,000
15,000
20,000
40,000
70,000
a 100,000
Minimum separation distance
between roadways & probes
(meters)
10
20
30
50
100
250
"Distances should be interpolated based on traffic tlow.
Cumulative Interferences on Path
Some consideration should be given toward controlling the combined total affect on a path measurement from all
the possible interferences which exist around the path. To control the sum effect of these interferences, a requirement
could state that the cumulative length of a path that is affected by minor sources, obstructions, trees, and roadways must
not exceed 10% of the path. Figure 2 illustrates this concept. In this example, an open path analyzer is sited with two
interferences, a tree and a roadway with an average daily traffic count of less than 10,000 vehicles per day. To
calculate the portion of the path affected by the tree, it is necessary to determine how much of the path would be
obstructed by the tree from the predominant wind flow. This portion is labeled as "Z" on Figure 2. To determine the
portion of the path affected by the roadway, it is necessary to sum the length of the path segments over the roadway with
those over the minimum separation distance on both sides of the roadway, defined by "x" on Figure 2. This portion of
the path is labeled as "Y" The sum of "Z" and "Y" is the total interference on this path. It is this sum which must be
less than 10% of the total path.
Maximum Length of Path
Since the measurement scales are defined in terms of finite physical dimensions, some consideration must be
given to the maximum length of a path used to monitor within each of these measurement scales. For ozone, nitrogen
dioxide, and sulfur dioxide analyzers measuring concentrations based on neighborhood, urban, or regional scales, the
maximum path length could be one kilometer. For middle scale measurements of these three pollutants, a maximum
path length of 300 meters may be appropriate. These path limitations are necessary in order to produce a path
concentration representative of the measurement scale and to limit the averaging of peak concentration values.
In actual siting situations, some slight deviations from these criteria may be necessary to account for actual
siting conditions. Consequently, the waiver provisions allowed for probe siting should also apply for open path
analyzers. Future refinement of these criteria would be possible as monitoring agencies gain more experience from
using the DOAS and other open path analyzers.
CONCLUSIONS
Each monitoring station has an associated monitoring objective and spatial scale of representativeness, or
measurement scale. Given that the appropriate measurement scales for monitoring ozone, nitrogen dioxide, and sulfur
dioxide will range in dimensions from 100 meters to more than 50 kilometers, it is reasonable to state that open path
analyzers would be able to measure pollutant concentrations representative of each of these scales. Specific path siting
554
-------
criteria would be needed to limit path interferences which would cause the analyzer to produce unrepresentative data.
These criteria should be stringent enough to cause the open path analyzers to produce data that would be useful for the
protection of public health, while also remaining flexible enough to encourage the use of this technology.
DISCLAIMER
The information within this document has been subjected to review by the United States Environmental
Protection Agency and approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
REFERENCES
1. T.L. Conner and R.K. Stevens, "Differential Optical Absorption Spectrometry for the Measurement of Ozone
and Its Precursors," in Proceedings of the 1991 U.S. EPA/A&WMA International Symposium on Measurement
of Toxic and Related Air Pollutants. VIP-21, Air & Waste Management Association, Pittsburgh, 1991, pp 31-
36.
2. Code of Federal Regulations. Title 40, Part 50, U.S. Government Printing Office, 1992.
3. A.R. Casavant and C.J. Kamme, "An Inter-Comparison of Sulfur Dioxide, Nitrogen Dioxide, and Ozone with
DOAS," in Proceedings of the 1992 A&WMA International Specialty Conference on Optical Remote Sensing
Applications to Environmental and Industrial Safety Problems. SP-81, Air & Waste Management Association,
Pittsburgh, 1992, pp 241-251.
4. Code of Federal Regulations. Title 40, Part 58, Appendix D, U.S. Government Printing Office, 1992.
5. R.J. Ball and G.E. Anderson, Optimum Site Exposure Criteria for SO2 Monitoring. The Center for the
Environment and Man, Hartford, prepared for the U.S. EPA, Research Triangle Park, EPA publication number
EPA-450/3-77-013, April 1977, p 18.
6. Code of Federal Regulations. Title 40, Part 58, Appendix E, U.S. Government Printing Office, 1992.
BIBLIOGRAPHY
J.L. Hudson et. al., "Evaluation and Demonstration of Open-Path Fourier Transform Infrared
Spectroscopy as a Tool for the In-Situ Measurement of Air Toxics," to be published in the Proceedings of the
A&WMA/U.S.EPA International Symposium on Field Screening Methods for Hazardous Wastes and Toxic
Chemicals. Air & Waste Management Association, Pittsburgh, 1993.
F.L. Ludwig and E. Shelar, Site Selecting for the Monitoring of Photochemical Air Pollutants.
EPA-450/3-78-013, U.S. Environmental Protection Agency, Research Triangle Park, April 1978.
E.S. Ringler and V.P. Aneja, "A Ground Based Inter-Comparison of Path Integrated DOAS
Measurements and Conventional Point Measurements of Ambient Trace Gas Concentrations," in the Proceedings
of the 1992 A&WMA International Specialty Conference on Optical Remote Sensing Applications to
Environmental and Industrial Safety Problems. SP-81, Air & Waste Management Association, Pittsburgh, 1992,
pp 232-240.
R.L. Spellicy et. al., "A Demonstration of Optical Remote Sensing in a Petrochemical
Environment," in Proceedings of the 1992 A&WMA International Specialty Conference on Optical Remote
Sensing Applications to Environmental and Industrial Safety Problems. SP-81, Air & Waste Management
Association, Pittsburgh, 1992, pp 273-285.
K. Weber et. al., "The Measurement of Gaseous Air Pollutants in the Troposphere by Optical
Remote Sensing-Developments and Applications in Germany," in Proceedings of the 1992 A&WMA
International Specialty Conference on Optical Remote Sensing Applications to Environmental and Industrial
Safety Problems. SP-81, Air & Waste Management Association, Pittsburgh, 1992, pp 30-42.
555
-------
Muracale
«100nMttn)
Regional Kale
(>50 kilometers)
03.SO2
Urban Kale
(4-50 kilometers)
O3. NO2. SO2
figure 1 Measurement scales for ambient air monitoring.
(See reference 5 - Ball and Anderson)
Z
Roadway
ADT* 10,000
Portion of path affected by potential or actual interferences. Z + Y < 10% of path
x = 10 meters = Minimum roadway separation distance for O3 and NO2 monitoring.
Figure 2 Example of cumulative effects on path.
556
-------
MONITORING FOR FUGITIVE EMISSIONS AT SUPERFUND SITES
DURING REMEDIATION ACTIVITIES WITH AN FTIR REMOTE SENSOR
Robert H. Kagann, Orman A. Simpson
MDA Scientific Inc
3000 Northwoods Pkwy, 185
Norcross, Ga 30071
j Robert J. Kricks
Blasland, Bouck & Lee
Raritan Plaza, Fieldcrest Avenue
Edison, NJ 08837
INTRODUCTION
An FTIR remote sensor (FTIR-RS) was used to monitor for fugitive emissions during remediation activities at several
contaminated sites. These include the Lapari Landfill Superfund site in soulh New Jersey, the Gulf Coast Vacuum
Superfund site in Louisiana, the Westminister Superfund site in Orange County, California, and a 17 acre surface
impoundment which will undergo remediation by removal and solidification. The FTIR measurements were used in a
plume dispersion model to calculate the emission rates of toxic gases and the downwind impact in areas of human activity.
Figure 1 The FTIR Remote Sensing System. The
..-/^ FTIR-RS uses a single telescope to both
..-•' transmit a 12 inch collimated infrared beam
..-'' to the atmosphere, and receive the infrared
beam after it traversed a round-trip path
through the atmosphere. A 14 inch corner-
cube retroreflector array is used to reverse
the path of the ir beam back to the
1TJT transmitter / receiver telescope.
A recent review describes the use of optical remote sensors, including the FTIR-RS, to measure toxic gases.1 The
FTIR-RS used in the present programs was designed as a unistatic design which uses a single transmitter/receiver
telescope. Figure 1 shows a simple schematic of the measurement setup. The infrared radiation is modulated by a
Michelson interferometer, of a wishbone design, which uses corner-cube retrorefleclors. This retroreflector-interferometer
design eliminates the need to make field alignment adjustments on the interferometer, with changing temperature. The
modulated infrared radiation is collimated by a 12 inch cassegrain telescope and transmitted through the parcel of
atmosphere being measured to a comer-cube retroreflector array. The array reflects the beam back to the single
transmitter/receiver telescope, resulting in two passes through the chemical plume. The radiation is then focused on the
aperture of a mercury cadmium telluride (MCT) detector. The modulated signal, referred to as an interferogram, is Fourier
transformed to produce a spectrum from 700 cm"' to 4500 cm"', with 1 cm"' resolution. This system was designed with
the primary goal of fast and easy setup and alignment. This is particularly important in Superfund and similar applications
involving very contaminated areas. The system can be setup in a matter of minutes. The retroreflector array's orientation
is not critical for proper alignment, so if it needs to be setup in a restricted, highly contaminated area by personnel in
protective garments, it takes only a few seconds to set down and to orient so as to face (within 10 degrees) in the direction
of the telescope.
The FTIR-RS is particularly well suited for monitoring for fugitive chemical emissions during remediation activities to
provide emergency response. These systems can detect and measure the concentrations of a multitude of infrared
absorbing gaseous chemicals in real-time. The conventional Summa canister sampling methods require the subsequent
analysis, usually by gas chromatography / mass spectrometry (GC / MS). This process is too slow for canister collection
techniques to be of use for emergency response. The FTIR-RS and other optical remote sensors measure concentrations
over an integrated path. This type of measurement is extremely useful as an input to a plume dispersion model to calculate
emission rales. If the FTIR-RS measurement is made in the crosswind direction, covering the entire distribution in that
direction, the only dispersion of concern is in the vertical direction. This greatly simplifies the dispersion analysis.
Another advantage of the FTIR-RS technique is that it is an in-silu measurement and is therefore not affected by wall
adsorption effects that occur in collection or extraction systems; the FTIR-RS can directly measure polar molecules.
FTIR MEASUREMENTS
The FTIR-RS is used to make integrated-path concentration determinations of the chemical constituents in a plume.
557
-------
The infrared beam path is set up in the crosswind direction, downwind of the emission source, with the end points chosen
so that the entire plume crosses the beam path. These measurements made downwind of the emission source are referred
to a downwind measurements. The initial spectroscopic data are single beam spectra, I. According to Beer's law, the
concentration-pathlength product, C L, of an absorbing chemical is proportional to the absorbance, A, which is defined as
A(v) = -log(I(v)/Io(v)), (1)
where all three quantities here are spectra and are written as functions of v, the optical frequency in wave number (cm"')
units. Io(v) is the background spectrum taken under identical conditions to I(v), with the exception that the absorbing
chemicals are not present. Beer's law is written as
A(v) = E(v)CL, (2)
where the proportionality constant, e(v) is the absorption coefficient of the chemical. The absorption coefficient is
essentially a spectral shape function and is unique for each molecule, resulting in the unique "fingerprint" shape of the
chemical absorption spectrum used in making qualitative identifications. The quantitative determination is made by
measuring the absorbance, A(v), and determining the concentration-pathlength product, C L, of the absorbing chemical.
The concentration-pathlength products of the absorbing chemicals are calculated using the multicomponent classical
least-squares technique (CLS) developed by Haaland and Easterling.2 This technique involves performing a least-squares
fit of the field spectra to reference spectra (of the absorbing chemical species) using Beer's law,
As(v) = a + bv + ZrYr Ar(v) + e(v), (3)
where As(v) is the total absorbance of the measured atmospheric spectrum, at frequency v, Ar(v) is the absorbance of the
rth reference spectrum and Yr 'S the ratio of the concentration-pathlength product for the r"1 chemical in the field, to the
concentration-pathlength product for the reference spectrum of the r^1 chemical. The first two terms, on the right side of
the equation, perform a linear correction for baseline error. The range of frequencies, v, for the analysis are limited to
regions where the target compound actually absorb the radiation. For each region, the sum in the third term is over all
species which absorb in that particular region. The last term is the error or residual of the fit. The reference spectra are
measurements of the pure gases, usually mixed in one atmosphere of dry air or nitrogen.
EMISSION RATE DETERMINATION
A Gaussian plume dispersion model was used primarily to determine the emission rates of the toxic gases from the
FTIR-RS measurement. The determined emission rates are then used in the dispersion model to calculate the
concentrations at fences lines and at points were human activity occurs. The emission rates were determined from the
FTIR-RS measurements by two different methods, the tracer ratio technique and by plume dispersion calculation using
site-specific coefficients for the dispersion in the vertical direction, a7. When the path integrated concentration
measurement path is in the crosswind direction and the height of the measurement is the same as the height of the emission
source, the general Gaussian dispersion equation (Turner Equation) for the emission rate, Qa simplifies to
Qa= Vn/2 (CL)aCTzUx, (4)
where the index, a, refers to a particular gaseous chemical, (C L)a is the ground-level crosswind-integrated concentration-
pathlength product (path-integrated concentration), in grams / meter^, measured at distance x from the emission source, for
chemical, a, and Ux is the horizontal wind speed in meters per second.
The tracer ratio technique involves releasing a tracer gas, t, (usually carbon tetrafluoride or sulfur hexafluoride) with a
measured emission rate, Qt, and measuring the tracers concentration-pathlength product, (C L)t, with the FTIR-RS, at
distance x from the source. Using Eq (4), the emission rate for unknown chemical emission a, Qa, can be written as
Qa= [ (CL)a/ (CL)t ] Qt. (5)
Thus, the problem reduces to emitting the tracer with a known emission rate and measuring (C L)t simultaneously with the
concentration-pathlength products for the emitted chemicals.
558
-------
The second method of determining emission rates is to use site specific values for the vertical dispersion coefficient, az,
along with the wind speed, Ux, in Eq (2). In cases where the measurement configuration do not satisfy the requirements
for using Eq (2), one uses the general plume dispersion expression, the Turner equation.
THE SITE MEASUREMENT CONFIGURATIONS
Figure 2 shows examples of the FTIR-RS setup configurations used at the four contaminated sites. Figure 2a shows
the Lapari Landfill Superfund Site in south New Jersey. This site contains hazardous liquid contaminants in (he soil, which
could migrate to the ground water. The remediation involves drilling wells into landfill cells, and pumping water into
them to displace the buried the contaminants. The hazardous contaminants are carried through ground water pathways to
interception wells dug at the periphery of the site. The contaminated water in the interception wells were pumped out, to
prevent the contamination from leaving the site in the ground water, and directed to a water treatment facility. The treated
water was men recycled back to the wells in the landfill cells. This flushing procedure continues until the contaminants are
reduced to safe levels. This procedure, and activities such as drilling monitoring wells to obtain ground samples, could
result in the volatilization of the organic contaminants producing air toxics. The FTIR-RS monitored for air toxics during
these activities. Figure 2a shows the FTIR measurement configuration used to monitor for air toxics during the drilling of
Well No. 22. The direction of the wind would carry any resulting emission plume across me infrared beam. The position
of the meteorological tower was located upwind of the infrared beam. The vertical dispersion coefficient, oz, was
estimated from site specific tracer measurements, and the emission rate was calculated using Eq (4). The distance
between the FTIR sensor and the retroreflector array was 100 meters.
Figure 2b shows the configuration at the Gulf Coast Superfund site in Abbeville, Louisiana. At this site, the soil was
contaminated by the petroleum wastes which resulted from past oil drilling operations. The FTIR-RS was introduced for a
pilot scale remediation study to determine the impact on air quality during the various test remediation procedures. The air
emission rates were measured during sludge excavation activities and during the process of solidifying the sludge. The
latter process produces heat which could drive off the more volatile components of the contamination. Figure 2b shows the
FTIR-RS measurement configuration during sludge excavation. The wind here had two prevailing directions (from the
Gulf of Mexico and from the north) and tended to shift from one to the other, so we used a folded beam configuration to
intercept the plume regardless of the wind direction. In this configuration, the tracer gas was released at the excavation
site and the ratio method of determining the emission rate (see Eq (5)) was used. The distance from the FTIR sensor to the
mirror was 60 meters and the total round trip beam path was 240 meters.
Figure 2c shows the configuration at the Westminster Superfund site in Orange County, California where another pilot
scale program evaluated the FTIR-RS for measuring emission rates during excavations. A folded beam path was used to
double the passes through the plume, resulting in four passes, round trip (transmitter/receiver telescope to the retroreflector
array and back). The tracer gas was released at the excavation site and the ratio method, to determine the emission rate,
was used. The distance from the FTIR sensor to the mirror was 32 meters and the total beam path was 128 meters. Figure
2d shows the configuration used at the surface impoundment / RCRA application. The FTIR measurements were made
during sludge dredging activity. The distance between the FTIR sensor and the retroreflector array was 175 meters.
THE MEASUREMENT RESULTS
Two examples of FTIR-RS spectra are shown in Figure 3. Figure 3a shows a field spectrum obtained at the Gulf Coast
Superfund site (upper trace). The middle trace is the laboratory-measured reference spectrum for n-octane (500 ppm-
ineter) and the lower trace is the reference spectrum for methane (81 ppm meter). Aliphatics higher than C-8 were present
at the Gulf Coast site, however, their spectra are similar to octane. The octane reference is used to represent the aliphatics,
C8 and higher. Since octane is the most volatile of these, it is expected to be the most predominant component.
A comparison of a field spectrum taken at the Westminster Superfund site to the reference spectrum of sulfur dioxide is
shown in Figure 3b. The upper trace is the field spectrum and the lower trace is the reference spectrum for sulfur dioxide
(476 ppm meter). A determination of 48.2 ppm meters of sulfur dioxide was made from this field spectrum. The total
round-trip pathlengti was 128 meters, thus the path averaged concentration was 377 ppb. This determination, with the
measured wind speed, and a carbon tetrafluoride tracer determination, resulted in an emission rate value of 0.80 g / sec for
sulfur dioxide.
Examples of the FTIR-RS / Plume Dispersion Model measurement results are shown in Tables 1 to 4. Table 1 shows
the determined maximum fenceline concentrations while one of the wells were being drilled in a landfill cell, in a four day
period, at the Lapari landfill site. The measurement configuration was shown in Figure 2a. The concentrations of the
listed chemicals were below the detection limits of the FTIR-RS during this activity. By a careful measurement of the
MDL's, an upper limit of the path averaged concentrations are determined, resulting in an upper limit on the emission
rates. The plume dispersion model was then used to determine the concentrations at the fenceline, downwind of the
559
-------
drilling activity. The MDL's, which are affected by changes in the humidity and other atmospheric and measurement
conditions, vary from one measurement to the next. The second column shows the range of MDL's, measured during the
four day period represented in Table I.
FT1R-RS
c.
d.
Mellower
•
Figure 2 Field Measurement Configurations at the Contaminated Sites.
a) Lapari Landfill Superfund Site. The FTIR beam path was downwind of drilling activity at well #22. The
tracer gas was released at the well.
b) Gulf Coast Vacuum Superfund Site. A v-shaped beam path configuration was used here because of the
frequent 180 ° change in wind direction. A flat mirror (M) reflected the beam to the retroreflector array at
location R. The tracer was released at the location T.
c) Westminster Superfund Site. A folded beam path was used to double the sensitivity of the measurement
Excavation occurred during this measurement at the tracer release point
d) Surface Impoundment (RCRA) Site. The FTIR beam path was downwind of sludge dredging activity in the
surface impoundment.
Table 2 shows measurement results for emission rates obtained, in a 35 minute time period, during excavation at the
site shown in Figure 2b. The three major components measured here are methane, n-octane, and iso-octane. There was
also present higher chain alkanes but because their spectra are similar to octane's, the n-octane concentration was used to
represent C8 and higher straight chain alkanes and iso-octane was used to represent C8 and higher branched chains
alkanes. The concentrations of all three component were above the detection; thus, these are determinations of actual
emission rates.
Table 3 shows maximum emission rates determined at the surface impoundment / RCRA program. The concentrations of
the compounds listed here were below the MDL's, so these are upper-limit values. Determined emission rates for sulfur
dioxide at the Westminster Superfund site are shown along with a description of the concurrent test remediation activity
560
-------
are shown in Table 4. The sulfur dioxide concentrations were above the detection limits, thus the emission rates were
directly measured.
....tAthh JL.nUiUiht...
2830 2898 2966 3034 3101.9 3169.9 2440 2460 2480 2500 2520 254O
Wavenumbers (cm—1) Wavenumbers (cm —1)
Figure 3 Comparisons of Field Spectra to Laboratory Measured Reference Spectra.
a) A field spectrum of n-octane and methane, measured at the Gulf Coast Superfund Site, is shown in the
upper trace. The middle trace is the reference spectrum for 500 ppm meter of n-octane and the lower
trace is the reference spectrum for methane (81 ppm meter). The total round-trip beam path was 240
meters and the resolution was 1 cm-1.
b) A field spectrum of sulfur dioxide (377 ppb), measured at the Westminster Superfund Site is shown in
the upper trace. The bottom trace is the reference spectrum for sulfur dioxide (476 ppm meter). The
total round-trip beam path was 128 meters and the resolution was 1 cm-1.
DISCUSSION
The measurement data from these four programs are voluminous. The results shown here are small examples taken
from each of the programs, to provide an overview of emergency response applications of FTIR-RS measurements. The
FTTR-RS system provides a real time concentration report which includes a goodness of fit number (GOF) which is equal
to three time the standard deviation of the classical least-squares fit, propagated to the concentration determination. This
number gives a measure of the precision in the concentration determination. It is also used to determine if the CLS result
for the concentration is real or a false positive. If the concentration determination is greater than three times the GOF, the
CLS calculated value is a measure of the concentration of the absorbing chemical. If the concentration determination is
less than three times the GOF, then the concentration of the absorbing chemical is considered to be below the detection
limit. If the chemical is known to be present then two times the GOF is used to determine that the CLS determination is
not a false positive. The minimum detection limit (MDL), which can vary from one measurement to the next (due to
changes in the humidity and other factors), is considered to be equal to two to three times the GOF value. The MDL
determination is very important when the concentrations are below the detection limit because is provides an upper limit to
the concentration values and an upper limit to the emission rates, two very important determinations.
The precision in the concentration measurements at the four sites ranged around three percent. The accuracy of the
concentration measurements are determined by a measurement quality assurance procedure. This involve flowing a
reference gas mixture (in which the component concentrations are known to 1 percent accuracy) through a 15 cm
spectroscopic gas cell which is situated in the infrared beam path inside of the FTIR sensor. The infrared radiation then
passes through both the cell and the atmosphere. The concentration determinations should increase by the concentration of
the components in the cell. The deviation from the expected increase is reported as the accuracy. The accuracy of the
concentration determinations at the four sites ranged from 3 percent to 33 percent, and for most measurements was around
16 percent. The two most major sources of systematic error in the FTIR-RS concentration determinations are error in the
reported concentration-pathlength product for the reference spectrum and baseline errors. The accuracy determinations in
the field programs provide us with feedback to help us Improve the accuracy of the reference measurements.
The error in the emission determinations predominantly come from inaccurate assumptions in the plume dispersion
model, particularly in respect to the vertical dispersion distribution. The ratio technique using Eq (5) gets around this
problem nicely, so that the error in the emission rate is propagated from error in the concentration of the absorbing gas,
the concentration of the tracer gas, the emission rate of the tracer gas, and error due to imperfect co-location of the
tracer emission source to the absorbing gas emission source.
561'
-------
Table 1 Daily Worst-case Maximum Exposure at
the Fenceline of the Lapari Landfill Superfund Site.
These determinations were made from
measurements at which the concentrations were
below the detection limits of the FTIR-RS which in
turn were verified by QA procedures. The full
range of MDL values over the four day period are
given in the second column..
Table 2 Actual Measured Emission Rates
Determine at Gulf Coast Superfund Site.
Time
15:55
15:56
16:01
16:06
16:10
16:12
16:13
16:14
16:18
16:23
16:28
n-Octane
& n-C9+
S/sec
0.034
0.054
0.29
0.55
0.43
0.37
0.58
0.51
0.32
0.65
0.91
iso-Octane
& iso-C9+
g/sec
0.051
0.10
0.072
0.057
0.096
0.071
0.053
0.11
0.20
Methane
g/sec
0.38
0.28
0.30
0.67
0.44
0.54
0.64
0.89
0.21
0.38
0.57
Table 4 Actual Sulfur Dioxide Emission Rates
Determined from FTIR-RS
Measurements at the Westminster
Superfund Site for measurement day 5.
On this day, two different trenches were
dug in an area with subsurface toxic
waste.
Contaminant
Benzene
iis(2-cbloroethyl )elher
Toluene
Xylenes
Ethyl Benzene
1,1,1 -Trichloroethane
1,1-Dichloroethane
1 ,2-Dichloroethane
Methylene Chloride
Chloroform
Carbon Tetrachloride
MDL's
mg / m2)
28 - 120
26- 98
18- 58
25- 59
17 - 120
7 - 30
12 - 52
74 - 290
37 - 106
5.- 13
4 - 9
Max. Fenceline
Concentration (mg/m3)
9/27
1.8
1.5
0.87
0.88
1.9
0.45
0.78
4.3
1.6
0.19
0.13
9/30
1.5
1.8
1.1
1.3
1.8
0.37
0.64
6.3
2.7
0.28
0.20
10/1
1.2
0.82
0.64
0.81
0.54
0.28
0.49
2.4
1.2
0.19
0.12
10/2
0.74
0.77
0.46
0.69
0.47
0.18
032
2.8
1.4
0.16
0.09
Table 3 Maximum Emission Rates Determined for Three
FTIR-RS Events in the surface impoundment (RCRA)
program. The minimum detectable limits (MDL) for FTIR-
RS measurements are given in the column labeled MDL.
Compound
Benzene
Chloroform
1122-
Tetrachloroethane
Tetrachloroethylene
Vinyl Chloride
Meas. Time
Start
13:30
13:45
14:05
14:20
14:25
14:30
16:05
16:20
16:35
16:50
End
13:35
13:50
14:10
14:25
14:30
14:35
16:10
6:25
6:40
16:55
SO2
g/sec
0.55
0.92
0.45
0.73
0.54
1.1
0.07
1.0
0.42
0.41
MDL
mg/m2
105
8.3
43.9
18.3
28.1
Event
B6
g/sec
2.3
0.21
0.90
0.45
0.62
Event
B14
g/sec
1.2
0.11
0.48
0.23
0.32
Event
B17
g/sec
3.0
0.28
1.2
0.58
0.80
Activity in Trenches
Excav, spray foam, encounter waste
Foam for emission suppression
Refill trench
Remove topsoil from second trench
Encounter waste material
Emiss. suppression with C12 sol.
60 buckets of material removed
CONCLUSIONS
The FTTR remote sensor has shown to be a valuable real-time measurement tool for emergency response to
fugitive emissions of toxic air pollutants during remediation of contaminated sites. The use of the FTIR-RS with plume
dispersion models allow calculation of the downwind impact of accidental and routine releases, to determine whether
human exposure to toxic levels have occurred.
REFERENCES
1. Grant W. B., Kagann R. H., andMcClenny W. A. (1992) Optical Remote Measurement of Toxic Gases. J Air Waste
Manage. Assoc. 42,18 - 30.
2. Haaland D. M., and Easterling R. G., (1982) Application of New Least-squares Methods for the Quantitative Infrared
Analysis of Multicomponent Samples. Applied Spectroscopy 36,665 - 673.
562
-------
A New Concept for
Open Path Air Pollution Monitoring
Lyle. H. Taylor
Westinghouse Science & Technology Center
1310 Beulah Road
Pittsburgh, PA 15235-5098
ABSTRACT
This monitor concept combines an acousto-optic tunable filter for emission
spectroscopy (3.5-14.0 j/m) with a mid-IR (4.6-5.4 jum) and far-IR laser (9.2-10.9 jjm) for
absorption spectroscopy in an instrument for the optical remote measurement of ppb-ppm
concentrations of air pollutants. It utilizes commercially available components, is fast (-2
minutes for 120 gases), covers a large area (~6 km path lengths), measures the distance to
any reflecting object, and can take measurements along any lin^-of-sight.
INTRODUCTION
Concern with atmospheric measurements has grown immensely in recent years
with the realization that the quality of human life is significantly impacted by the quality
of the air. Over 40 toxic gases and 172 specific Hazardous Air Pollutants (HAPs) are of
primary concern. To assure a healthy environment for human life, it is necessary to
monitor over large areas the concentrations of these pollutants, most of which are
hydrocarbons which have line spectra in the infrared "fingerprint region" of 8-12 ^m.
For long open-path remote sensing and quantitative measurements of
atmospheric concentrations of trace vapors, differential-absorption lidar (DIAL) is the best
technique. Furthermore, infrared DIAL systems are preferred because they are highly
sensitive to the laser energy, are relatively "eye safe", and, most importantly, are in the
spectral range where most molecular-specific absorption lines occur. Of the available
infrared lasers, CO2 lasers are the best suited for long-path atmospheric monitoring
because they have the highest efficiencies and powers, are easily tuned, and cover the
9.2-10.9 nm range which is very rich in molecular-specific spectra. Furthermore, their
wavelengths can be extended by harmonic generation to cover the 4.6-5.4 um range.
However, all laser systems have limited wavelength coverage. Thus, a DIAL
system should be complemented with a broader wavelength system. An Acousto-Optic
Tunable Filter (AOTF) is a good choice for the complementary system because it is easily
integrated into a DIAL system, it monitors emission spectra passively, it can be quickly
tuned to any desired wavelength, it's sensitivity is easily increased by measuring
derivatives of spectra lines, and it covers two wavelength octaves, e.g., 3.5-14.0 /urn.
POLLUTION MONITOR CONCEPT
System Description
The remote monitor, as shown in Figure 1, is comprised of six key elements: a CC>2
laser, a nonlinear crystal, optics, an AOTF, detectors, and a computer. The CO2 laser is
tunable over 87 lines in the 9.2-10.9 /jm region where a large multitude of hydrocarbons
have absorption spectra. The laser operates at 10 pulses/sec with 1 to 250 mJ/pulse,
depending on the line. The pulse width of the linearly polarized beam is 100 nsec.
563
-------
—-H— =.\ ;=.
Beam ir1"'' i I T..r.-.Ma h
Expander \ /
Tunable
Filter
+
Harmonic
Generator
Detector
_t
Topographic
Target
Computer
Figure 1 — Basic configuration of remote monitor has six key components.
The CC>2 laser frequency is doubled with a nonlinear crystal, TlgAsSe3 (TAS).
TAS harmonic generators, see Figure 2, are completely passive and have produced the
highest measured efficiency in the far infrared, 57%. This crystal will produce pulse
energies from 1 to 15 mJ on 68 lines in the 4.6-5.4 ium region. The transmitter optics then
enlarges the laser beam to a 20 cm diameter to make it eye safe.
Westtoghouse Science & Technology Center
Figure 2 — Harmonic generator and AOTF are small solid-state components.
The receiver optics collects the reflected beam with a 30 cm mirror and focuses it
through the AOTF onto the detectors. The AOTF is fabricated from a TAS crystal, as
shown in Figure 2, and operates as described in Figure 3. The received beam, linearly
polarized as indicated by the arrows, enters the crystal and interacts with a periodically
varying spatial distribution of indices of refraction set up by an acoustic beam inserted via
564
-------
the transducer. Only a narrow spectral band, -10 cm"*, will be phase matched to the
acoustic beam and diffracted out of the main beam, with its plane of polarization rotated
90° since TAS is a bireftingent crystal.
Two detectors are used for improved sensitivity. One detector operates from 7 to
14 Aim and one operates from 3.5 to 7 Mm. The AOTF directs the received radiation to the
two detectors by using one transducer for the 7 to 14 ma portion and a second transducer
for the 3.5 to 7 too. portion. The transducers are placed on orthogonal faces of the crystal
and diffract the two portions of the beam into two different directions. The detector signals
are then analyzed by the computer.
Input Undetected
IU)
Figure 3 — AOTFs are electronically-tunable and have narrow pass bands.
System Operation
The CO£ laser wavelengths are switched in a predetermined pattern, typically
staying on each wavelength for one second. Electronically activated two-position mirrors
direct the CO2 laser beam through the harmonic generator crystal for short wavelength
operation and around the crystal for long wavelength operation. An electronically
controlled gimballed telescope directs the beam to any target in real time. Thus, large
areas can be quickly monitored via several beam paths and the beam paths quickly
changed to respond to fugitive releases wherever they may occur.
The AOTF has two functions. During absorption measurements the AOTF
increases the signal-to-noise ratio by restricting radiation from the atmosphere to a narrow
spectral range, -10 cm" , around the absorption line. During emission measurements, the
AOTF is operated from 3.5 to 14.0 /jm. By careful selection of the acoustic frequency, the
wavelength of the diffracted beam can be centered on key emission lines of specific gases,
such as shown in Figure 4. These key emission lines can be monitored when absorption
565
-------
measurements are not being taken, and if preset thresholds are exceeded the laser can be
activated for detailed measurements. Alternatively, the entire wavelength region can be
scanned to obtain spectra such as the one for benzene shown in Figure 5.
22
2 Chloride
8
9
10
Wavelength ((im)
11
12
Figure 4 — AOTF wavelength is centered on a line with the acoustic frequency.
2.5 |xm
Figure 5 —An AOTF-generated benzene spectrum.
The detectability of sharp emission lines is enhanced by modulating the acoustic
frequency at a fixed frequency, ~1 kHz, as shown in Figure 6. This modulation
sinusoidally shifts the AOTF passband. The modulation does not affect the radiation from
sources which have relatively constant intensities over the AOTF passband, but modulates
the intensity from emission lines narrower than the AOTF passband. A lock-in amplifier
tuned to the modulation frequency gives the first derivative of the spectra within the AOTF
passband. The second derivative is obtained in a similar manner.
The computer determines concentrations and, with a 10 nsec rise time detector,
determines the range to the reflecting target to within 2 m. It then stores and displays the
results. With associated electronics it also controls the operation of the monitor.
566
-------
AOTF
Pass Band
Emission Line
AM Output
Signal
Output Signal
RF Modulation
Figure 6 —AOTF derivative detection is sensitive to narrow-line emissions.
System Performance
The laser power is sufficiently large that signal-to-noise over 6 km path lengths is
not a problem. In this mode of operation the monitor is essentially the same as other CO2
DIAL systems and has the same sensitivities. In the 9.2-10.9 Aim region, the monitor has
the potential to measure concentrations of 101 HAPs and over 40 other vapors of interest.
Detection limits vary from 1 ppb for Freon 12 to 60 ppb for ethyl-mercaptan to 340 ppb for
sulphur dioxide. In the 4.6-5.4 ym region, the monitor has the potential to measure 16
HAPs and over 14 other vapors of interest. Detection limits vary from 0.3 ppb for carbonyl
sulfide to 21 ppb for nitrous oxide to 187 ppb for carbon monoxide.
In measuring emission spectra the large wavelength coverage allows the
monitoring of literally hundreds of gases. However, the sensitivity is lower because the
emitting gas is at or near the same temperature as the atmosphere which is emitting as a
blackbody. Fortunately, atmospheric vapors have narrow line widths which allows the
modulation of the AOTF to increase the sensitivity by obtaining first and second
derivatives of the spectra. This enhancement is shown in Figure 7 for a laser line with 1%
of the spectral radiance of a glow bar in the background. The laser line cannot be seen in
direct detection but when the first derivative is taken the laser line is clearly seen.
The concentration sensitivity of this technique can be in the ppb range but is
wavlength dependent because of the atmospheric blackbody wavelength dependence. The
signal (emission line) to background (atmospheric radiation) ratio is increased, by taking
the first derivative, by 9 at 10.6 urn, by 36 at 5.3 urn, and by 75 at 3.7 urn. The second
derivative gives enhancements of 68 at 10.6 Aim, 1100 at 5.3 /um, and 4700 at 3.7 nm.5
567
-------
A. = 3.39 pirn
I
I
I
I
I
X = 3.39 |xm
I
I
I I I I
46 48 50 52 54 56 58 60 62
Frequency, MHz
46 48 50 52 54 56 58 60 62
Frequency, MHz
Direct Detection Derivative Detection
Figure 7 —Measurements of HeNe laser line with 1% the radiance of a glow bar.
CONCLUSIONS
The extension of the CC>2 laser DIAL system into the 4.6-5.4 jrni region increases
the coverage of some key gases which do not absorb in the 9.2-10.9 Aim region. Designing
the DIAL system with a pulsed CC^ laser and a gimballed telescope allows the range to
any reflecting target to be measured and the monitoring volume to be selected in real time
— major operating conveniences. Absorption spectroscopy over a 6 km path length can
detect concentrations of over 150 gases to levels of 1 ppb to 340 ppb.
The AOTF in the receiver improves the signal-to-noise ratio in the absorption
measurements but its primary advantage is in the 3.5-14 um emission spectroscopy.
Hundreds of gases can be measured in concentration levels down to ppb. These
measurements are possible because of the large enhancements in signal-to-background
ratios obtained by taking spectral derivatives — an easy task with an AOTF.
REFERENCES
1. W. Barnard, U.S. EPA, Research Triangle Park, NC, personal communication, 1992.
2. B. Lee, "Highlights of the Clean Air Act Amendements of 1990," J. Air Waste Mangage.
Assoc. 41: 16 (1991).
3. D. K. Killinger and N. Menyuk, "Remote Probing of the Atmosphere Using a C02 DIAL
System," IEEE J. Quantum Electron. QE-17: 1917 (1981).
4. D. R. Suhre, "Efficient Second Harmonic Generation in TlgAsSe3 Using Focussed C02
Laser Radiation," J. Appl. Phvs. B52: 367 (1991).
5. M. Gottlieb, "Acousto-Optic Tunable Filters," Chap. 5, Acousto-Optic Devices &
Applications. Marcel-Dekker (to be published in 1993).
568
-------
Session 16
Measurement ofVOCs
-------
Recovery After Storage and Desorption Efficiencies
for Volatile Organic Compounds Spiked on Thermal Desorption Tubes
Scott A. Hazard* and Jamie L. Brown
Supelco Inc., Supelco Park, Bellefonte, PA 16823-0048 USA
ABSTRACT
Two very important considerations when analyzing samples collected on thermal
desorption tubes are 1) the conditions under which the tube must be stored prior to analysis, to
assure confidence in the final results, and 2) desorption efficiency of the collected analytes from
the tube.
1) Recoveries were determined for 12 halogenated hydrocarbons desorbed from spiked
thermal desorption tubes after storage at ambient, refrigerated, and frozen temperatures for 3, 7,
and 14 days. We found that three day storage at ambient temperature gave 95% or better
recovery of the analytes. Samples stored for the longer time periods also gave acceptable
results, when stored at reduced temperatures.
2) Thermal desorption efficiencies were measured after spiking adsorbent tubes with 20
halogenated and nonhalogenated hydrocarbons, in volumes simulating 1.0, 2.0, 3.0, 4.0, and 5.0
air liter samples. Recovery values indicate that, with increasing sample volume, desorption
efficiencies decrease for some compounds when the calculations are based on standards
prepared using the flash vaporization spiking technique and a 0.5L sample volume.
571
-------
INTRODUCTION
The collection of air samples on adsorbent(s) contained in thermal desorption tubes is
specified in many USEPA analytical methods [1-3] for the determination of volatile organic
compound (VOC) contamination. One important concern about this type of air sampling is the
integrity of the sample from the time it is sealed after sampling to the time it is thermally desorbed
in the laboratory, to determine VOC concentrations. In our first study, we spiked Carbotrap 300™
multibed thermal desorption tubes, which contain three types of carbon-based adsorbents [4-7],
with twelve common halogenated hydrocarbons listed in USEPA methods. The tubes were stored
at three temperatures: ambient, refrigerated, and frozen, for three time periods: 3, 7, and 14
days. For each time and temperature combination, recovery of the analytes after desorption was
compared to recovery from tubes spiked in an identical manner on the day of analysis.
In a second experiment, we spiked thermal desorption tubes with the same twelve
compounds in a similar manner and allowed 1.0, 2.0, 3.0, 4.0, and 5.0L of inert nitrogen to pass
through each tube before desorbing the analytes to the gas chromatograph. Recovery values
were determined by comparing analyte peak areas for the 1.0-5.OL samples to areas for 0.5L
samples. Recoveries for eight nonhalogenated compounds were determined at the 5.0L sample
volume.
EXPERIMENTAL
Equipment and Materials
A Hewlett-Packard 5890A gas chromatograph (Hewlett Packard, Avondale, PA) equipped
with an electrolytic conductivity detector (Ol Analytical, College Station, TX), a Dynatherm Model
890 thermal desorption unit (TDU), Carbotrap 300 thermal desorption tubes (11.5cm x 6mm OD
x 4mm ID glass) containing SOOmg 20/40 mesh Carbotrap C, 200mg 20/40 mesh Carbotrap B
and 125mg 60/80 mesh Carbosieve™ S-lll, and a VOCOL™ capillary column (105m x 0.53mm ID
x 3.0um phase film) (all from Supelco, Inc., Bellefonte, PA) were used in this work. Analytical
standard mixtures were also supplied by Supelco.
Tube Spiking
Three common tube spiking techniques are cited in EPA methods for calibration in thermal
desorption techniques: permeation device(s), static dilution, and flash vaporization [1,2]. In both
studies we employed the flash vaporization technique, utilizing the injection port and secondary
trapping port of the thermal desorption system (Figure 1). Injections of the analyte mix, 20ng
each compound in 0.2ul of methanol, were made by syringe through this injection port, into an
empty tube in the desorption chamber, heated to 200°C. Inert nitrogen (50ml/min) swept the
volatilized analytes from the desorption chamber to a Carbotrap 300 tube in the secondary
trapping port at the end of this heated pathway. The tube was retained in the secondary trapping
port for the time necessary to achieve the desired sample volume.
Thermal Desorption
Carbotrap 300 tubes containing the adsorbed analytes were desorbed to the gas
chromatograph at 330°C for 5 minutes with approximately 8ml/min carrier gas, with the flow
directed to the capillary column. The long length (105m) and thick film (S.Oum) eliminated the
need for a secondary refocusing device or the use of cryogenics to focus the analytes. An ELCD
detector was used to detect the halogenated compounds because it has high sensitivity for these
compounds and does not respond to methanol. A flame ionization detector (FID) was used to
detect the nonhalogenated compounds.
Tube Storage Study
For each storage time interval, nine identical tubes were spiked with 20ng of each analyte.
Each tube was removed from the secondary trapping port after exactly 10 minutes, giving a
sample volume of 0.5L. Upon removal from the secondary trapping port, the tubes were sealed in
glass storage containers. Three tubes were placed in each of three paint cans containing a small
amount of activated charcoal to minimize the chance of outside contamination. One can was left
572
-------
on the laboratory bench, one can was placed In a refrigerator, and one was placed in a freezer.
The average temperature readings over the storage study were 25°C, 4°C, and -27°C,
respectively. After 3 days storage, three tubes were removed from each paint can and were
allowed to equilibrate to room temperature before desorption. Three standard calibration
analyses were conducted intermittently within the nine sample set (12 tubes total). The peak area
responses were normalized to an internal standard and averaged within each temperature study
(e.g., mean of three normalized area counts for the refrigerated tubes, etc.) and for the calibration
analyses. Percent recovery was then determined, using the mean value for the calibration
analyses as 100%. This procedure was then repeated for the 7 and 14 day storage time periods.
The calculated values of recovery for the analytes are listed in Table 1.
Desorption Efficiency Study
The tube spiking technique described above was also used in this study. A mix containing
40ng of each analyte in 0.2ul of methanol was flash vaporized into a 50/ml/min nitrogen flow and
directed to the Carbotrap 300 adsorbent tube. The tube was retained under the flow conditions
for the time necessary to achieve the desired sample volume. Three analyses each were made
for sample volumes of 1.0, 2.0, 3.0, 4.0, and 5.OL. An internal standard was used to normalize
area counts before the three analyses for each sample volume were averaged. Desorption
efficiencies were calculated by using the mean for three 0.5L analyses as a 100% value.
Separate means for 0.5L analyses were determined for each sample volume study. Data
showing these percent recoveries are listed in Table 2. A 5.0L analysis was done with eight
nonhalogenated compounds; results are listed in Table 3.
DISCUSSION / CONCLUSIONS
Data obtained in the temperature - storage time investigation revealed that, ideally,
adsorbent tubes containing airborne volatile organic compounds should be stored at reduced
temperatures until analysis. A high confidence in sample integrity could be realized, for the
analytes studied, when the analytes were stored on 11.5cm x 4mm ID Carbotrap 300 tubes at
ambient temperature for three days. However, after three days, the tubes must be stored at
reduced temperatures, preferably frozen.
Data obtained in the desorption efficiency study showed that as the sample volume
increases, the amount of a specific analyte desorbed from the tube may become increasingly
less than the amount collected, when compared to calibration standards prepared using a lower
total sample volume.
Analyte desorption efficiency for any sample may decrease with increasing sample volume,
since the analyte could migrate through the adsorbent bed that has the most affiliation for it and
enter a stronger adsorbent which may not release it as efficiently. Breakdown of heat-sensitive
analytes also is more likely with increasing sample volume, because longer retention of the
analyte on the adsorbent is expected as a result of further migration/stronger adsorption within
the adsorbent beds. This increased retention will prolong the analyte's exposure to the heat
required for desorption.
RECOMMENDATIONS
In neither study were the effects of humidity investigated. Humidity effects are being
considered for future work. Alternative storage devices which reduce the amount of dead volume
for analyte migration also are being investigated at this time. This becomes more important for
smaller adsorbent tubes containing less adsorbent material.
The use of permeation devices to spike thermal desorption tubes with constant
concentration over the entire sample volume will be investigated, and results will be compared to
these values obtained by using the flash vaporization method.
573
-------
REFERENCES
1. Winberry, W.T., L. Forehand L, N.T. Murphy, et a/., Compendium of Methods for the
Determination of Air Pollutants in Indoor Air Method IP-1B, EPA-600-4-90-010, U.S.
Environmental Protection Agency, Research Triangle Park, 1990.
2. USEPA Contract Laboratory Program Statement of Work for Analysis of Ambient Air, Exhibit
D, Section 2, Revision IAIR01.2, Draft Method, U.S. Environmental Protection Agency,
Washington, D.C., 1991.
3. Winberry, W.T., N.T. Murphy, R.M. Riggin, Compendium of Methods for the Determination
of Air Pollutants in Ambient Xl/r EPA-600-4-89-017, U.S. Environmental Protection Agency,
Research Triangle Park, 1988.
4. Betz, W.R., G.D. Wachob, M.C. Firth, Monitoring a Wide Range of Airborne Contaminants
Proceedings of EPA/APCA Symposium, Measurement of Toxic and Related Air Pollutants,
Raleigh, N.C., 1987, pp761-70.
5. Heavner, D.L., M.W. Ogden, P.R. Nelson, Multisorbent Thermal Desorption/Gas
Chromatography/Mass Selective Detection Method for the Determination of Target Volatile
Organic Compounds in Indoor Air Environ. Sci. Technol. 26:1737-46, (1992).
6. Betz, W.R., S.G. Maroldo, G.D. Wachob, et a/., Characterization of Carbon Molecular
Sieves and Activated Charcoal for Use in Airborne Contaminant Sampling Am. Ind. Hyg.
Assoc. J. 50(4):181-187, (1989).
7. Betz, W.R., and W.R. Supina, Use of Thermally Modified Carbon Black and Carbon
Molecular Sieve Adsorbents in Air Sampling Pure and Appl. Chem. 61 (11):2147-50 (1989).
574
-------
Secondary Trapping Port
Sample Saver Chamber
Gas Chromatograph
Column
Heated Transfer
Line I
Desorption Chamber
Figure 1. Flow pathways in the thermal desorption unit
-------
Table 1. Mean percent recovery (± standard deviation) of chlorinated hydrocarbons stored on Carbotrap 300 thermal desorption tubes.
Analyte
Chloroform
1,1,1 -Trichloroethane
Carbon tetrachloride
1 ,2-Dichloroethane
Trichloroethylene
1 ,2-Dichloropropane
1,3-Dichloropropane
Tetrachloroethylene
Ethylene dibromide
Chloro benzene
Bromofortn
Bromobenzene
Ambient
95.7 ±1.5
100.9 ±3.2
102.9 ±2.8
97.7 ±1.5
104.0 ±1.6
101. 5 ±3.2
102.1 ±1.8
102.1 ±1.8
108.9 ±4.9
102.1 ±1.4
102.2 ±4.4
104.0 ±0.8
3 Days
Refrigerated
103.8 ±0.7
104.9 ±0.5
105. 8 ±0.4
101 .4 ±0.8
102.5 ±0.9
103.4 ±0.7
102.1 ±0.2
102.1 ±0.2
106.2 ±3.8
99.6 ±1.6
104.3 ±1.9
100.8 ±3.6
Frozen
102.2 ±1.0
103.2 ±0.7
105.8 ±1.4
102.2 ±2.8
103.3 ±1.3
102.7 ±0.4
100.6 ±1.1
100.6 ±1.1
94.1 ±7.8
99.6 ±0.5
104.1 ±3.6
101.7 ±2.5
Ambient
79.5 ±10.2
94.3 ±2.1
98.9 ±1.1
87.3 ±5.4
103.3 ±0.4
100.2 ±3.0
97.9 ±1.9
97.9 ±1.9
95.3 ±1.9
99.0 ±0.7
93.6 ±4.4
98.0 ±1.8
7 Days
Refrigerated
99.1 ±2.1
96.8 ±2.2
99.7 ±1.1
98.7 ±1.6
99.7 ±1.8
100.8 ±2.2
96.6 ±1.8
96.6 ±1.8
100.2 ±5.4
97.1 ±3.0
99.5 ±5.5
97.5 ±3.7
Frozen
103.3 ±1.6
98.8 ±0.6
103.1 ±1.4
103. 5 ±2.3
102.5 ±1.1
100.4 ±1.2
100.1 ±1.4
100.1 ±1.4
91. 8 ±5.8
99.4 ±1.9
95.2 ±2.2
95.5 ±2.2
Ambient
73.8 ±11. 8
96.0 ±2.5
99.7 ±3.7
78.0 ±17.6
105.0 ±1.7
101 .3 ±1.8
100.5 ±1.3
100.5 ±1.3
99.6 ±3.6
102.9 ±1.0
98.1 ±2.7
107.8 ±8.5
14 Days
Refrigerated
104.3 ±1.4
102.5 ±1.7
103.7 ±0.7
98.9 ±1.2
103.4 ±0.6
104.0 ±0.6
102.3 ±1.4
102.3 ±1.4
107.5 ±3.2
103.3 ±1.1
111.1 ±2.3
105.7 ±2.7
Frozen
109.8 ±3. 3
102.2 ±5.5
105.6 ±4.0
104 .2 ±3.0
103.0 ±1.8
100.1 ±4.2
100.9 ±2.0
100.9 ±2.0
108.7 ±7.2
102.1 ±1.3
106.2 ±5.7
102.6 ±1.6
-------
Table 2. Mean percent recovery (± standard deviation)* of chlorinated hydrocarbons
from Carbotrap 300 thermal desorption tubes challenged by sample volumes of 1-5L.
Sample Volume
Analyte
Chloroform
1,1,1-Trichloroethane
Carbon tetrachloride
1 ,2-Dichloroethane
Trichloroethylene
1 ,2-Dichloropropane
1 ,3-Dichloropropane
Tetrachloroethylene
Ethylene dibromide
Chlorobenzene
Bromoform
Bromobenzene
1 Liter
103.4 ±3.9
100.2 ±4.4
100.7 ±5.9
102.5 ±3.6
102.1 ±2.9
101 .2 ±3.1
91.0 ±5.7
98.6 ±4.5
79.0 ±5.4
99.1 ±0.8
94.7 ±6.9
100.5 ±1.2
2 Liters
99.1 ±3.3
94.2 ±3.8
98.7 ±3.4
102.5 ±2.5
103.0 ±2.6
100.9 ±3.4
87.6 ±2.9
101 .2 ±3.3
65.3 ±2.6
102.7 ±2.7
88.4 ±3.7
100.1 ±1.0
3 Liters
94.3 ±2.1
88.5 ±3.9
94.3 ±3.7
99.6 ±2.5
99.3 ±1.9
98.6 ±2.7
83.3 ±2.4
102.1 ±2.6
58.9 ±1.8
102.6 ±1.8
85.6 ±1.1
100.2 ±1.4
4 Liters
85.4 ±8.5
84.8 ±2.7
92.3 ±2.1
101.1 ±1.6
101 .3 ±0.5
97.8 ±1.2
75.8 ±0.2
98.1 ±0.8
48.0 ±0.2
100.3 ±1.3
85.3 ±0.6
100.5 ±0.4
5 Liters
70.4 ±19.1
82.8 ±3.0
90.6 ±3.9
103.2 ±1.2
102.6 ±0.3
99.3 ±0.9
76.5 ±1.5
101 .4 ±0.8
47.1 ±2.5
100.6 ±2.0
79.1 ±6.8
98.6 ±3.5
*Peak areas normalized to internal standard (2-chlorotoluene).
577
-------
Table 3. Mean percent recovery of nonhalogenated hydrocarbons
from Carbotrap 300 thermal desorption tubes (5L samples).
Analyte % Recovery
1-Heptene 99.3
Benzene 100.3
Toluene 100.3
Ethylbenzene 100.4
m-Xylene* 100.5
p-Xylene* 100.5
o-Xylene 99.6
Isopropylbenzene 100.0
"coeluting analytes
578
-------
TEMPORAL AND SPATIAL VARIABILITY OF TOXIC VOC
SOURCES IN COLUMBUS, OHIO
Mukund Ramamurthi, Thomas J. Kelly, and Chester W. Spicer
Atmospheric Sciences and Applied Technology
BATTELLE
505 King Avenue
Columbus, Ohio 43201-2693
ABSTRACT
Source apportionment modeling was conducted on a data set of 142 3-hr integrated air samples
collected at 6 different sites and in 3 separate campaigns during the summer of 1989 in the U.S. EPA's
Columbus (OH) Hazardous Air Pollutants (HAPS) program. Source contributions to 19 light hydrocarbon
and toxic VOC species, including formaldehyde and acetaldehyde, were modeled. Overall, the results
indicated that the mobile and area source categories, the latter defined as fixed, non-point sources, and
specifically, vehicle exhaust and organic solvent usage by small industrial/commercial facilities, were the
main contributors to the major toxic VOCs. A number of observations regarding the short-term variability of
the sources were also made, including diurnal trends in the vehicle exhaust, and gasoline vapor sources.
INTRODUCTION
Exposure to hazardous air pollutants (HAPs), including volatile organic compounds (VOCs), is
believed to result in significant risks to human health. Recent research has shown that many of the
toxic/hazardous VOCs present in urban air, such as benzene, toluene, and xylenes, originate predominantly
from mobile (i.e., vehicular) and area source emissions, rather than from industrial point sources.1"3 These
studies have demonstrated, using chemical mass balance (CMB) source apportionment modeling, that even in
the vicinity of large industrial point sources such as refineries and chemical plants, significant fractions of the
air concentrations of the major VOC species are attributable directly to vehicle exhaust, gasoline vapor, and
area source emissions.
The source-receptor studies conducted thus far, however, have not included the time resolution and
multiple sampling sites needed to assess the variability in VOC source contributions over short temporal and
spatial scales. This information is particularly relevant in urban areas, where time-varying area and mobile
source emission patterns can result in spatial variabilities within and between neighborhoods, and in diurnal
concentration trends and differences between short-term peak concentrations and long-term means. The
modeling conducted in this study attempts to address some of these issues.
Study Background and Measurements
The source attribution study conducted was a part of the broader Columbus HAPS field study
conducted by Battelle4 for the U.S. EPA in June-July 1989. The broader study addressed a number of issues
pertaining to human exposure to HAPs, with a primary focus on the spatial and temporal variability of the
HAPs concentrations in urban areas and of the key area sources of these HAPs.
The Columbus, Ohio, metropolitan area selected for the study has a population of over 1 million
people, thus assuring a sufficient source strength of area and mobile sources in the region. In addition, the
region does not contain any large industrial point sources such as refineries and chemical plants. The absence
of major point sources facilitates the study of the area and mobile sources believed to be responsible for most
of the HAP concentrations in urban areas.
Six measurement sites for the study were selected within the region, based on meteorology,
population density, traffic density, location of residential neighborhoods, and the zones of commercial and
industrial activity. The various sites were selected to provide spatial resolution at site-to-site distances from 1
to 10 km, both in the SW-NE and the NW-SE directions (in and perpendicular to the direction of the
prevailing winds, respectively). One of six sites (Site 1) was located in closest to the center of the city in a
predominantly commercial/business area, and was surrounded by heavily traveled traffic corridors, numerous
large parking lots, small area sources such as printing and publishing facilities, gasoline stations, and small
industries that use various chemical solvents. The other five sites were located in residential neighborhoods,
at differing distances from highways and commercial zones.
579
-------
A variety of samples were collected and analyzed in the study at each of the six sampling sites.4 The
primary sampling protocol at all six sites was integrated sampling over successive 3-hour intervals using
canister sampling for VOCs and dinitrophenylhydrazine (DNPH) impinger sampling for carbonyl compounds.
At all sites, sampling was carried out for eight 3-hour periods each day, during three separate 2-day
campaigns: June 20-22, June 28-30, and July 11-13, 1989. Each 2-day campaign contained 16 3-hour
sampling intervals, for a total of 48 spatial samples at each of the six sites for each of the species measured.
SOURCE APPORTIONMENT
Modeling Method
The Chemical Mass Balance (CMB) modeling approach, involving multiple linear least-squares
regression analysis, was employed in this study for apportioning the sources of the various VOC species
selected from the complete list of VOC species measured in the field study. The application of CMB
modeling to VOC source apportionment has been discussed previously'*-'1>2.
A total of 19 VOC species were selected for inclusion in this modeling study, from the complete list
of VOC species measured at the six sites.4 The list comprised 14 hydrocarbons (ethane, propane, i-butane, n-
butane, i-pentane, n-pentane, acetylene, ethylene, propylene, benzene, toluene, o-xylene, m+p-xylene,
ethylbenzene), 3 chlorinated hydrocarbons (1,1,1-trichloroethylene, perchloroethylene, and dichloromethane),
and 2 aldehydes (formaldehyde and acetaldehyde). A few of the VOC species modeled were not used as
fitting species; i.e, their inclusion in the model did not influence the source apportionment results. These
species included dichloromethane, formaldehyde, and acetaldehyde, which were not fitted because the source
profiles contributing to these species were too few or too uncertain, and acetylene and propylene, which were
often not fitted because of concerns about the data quality for these species.
All of the fitted VOC species are expected to have lifetimes in the urban atmosphere of at least
several hours, and up to a half-day or more in many cases.5 Since the travel times from the VOC sources to
the sampling sites in the study area are likely to be much shorter, on the order of 1 hour or less, the
reactivity of the VOC species fitted in the model was not expected to interfere with the CMB modeling
analyses conducted in this study.
Emissions Inventory
A review of the National Emissions Data System (NEDS)6 and the U.S. EPA's Toxic Release
Inventory System (TRIS)7 listings for Franklin County for 1989 confirmed that Columbus did not contain any
major point sources of VOCs, but rather, contained numerous minor point sources (or "area" sources)
identified in the TRIS listings as emitting various amounts of solvents such as toluene, xylenes, and
chlorinated hydrocarbons. A review of the Greater Columbus Industrial Redliner8 showed that these area
sources were generally scattered throughout much of Franklin County.
Source Profiles
The complete spectrum of source profiles previously developed in the literature for VOC sources was
assembled, including those from the U.S. EPA's SPECIATE database,9 papers by Scheff, et a/.,1 and Sweet
and Vermette,2 as well as several other references cited in these papers. Profiles representing sources that
were not physically present in the Columbus area were first eliminated from consideration. Subsequently, the
most appropriate source categories for modeling the Columbus VOC database were identified by using
various combinations of source profiles to apportion the measured VOC concentrations. A list of five refined
source profiles was thus identified, representing source categories of vehicle exhaust, gasoline vapor, natural
gas, industrial solvent usage (toluene, xylenes) and chlorinated solvent usage, the latter believed to be tied to
(dry)cleaning, degreasing, and wastewater operations (CDW).4
RESULTS
Campaign-Average Data - Overview and Spatial Analysis
Campaign-average data sets were prepared by averaging together the 16 consecutive 3-hr samples
from each site in each 2-day sampling campaign, yielding 18 campaign-average data sets (3 campaigns x 6
sites). The source apportionment results for the composited data showed that vehicle exhaust, gasoline vapor,
and natural gas were the main sources of each of the modeled VOCs at all six sites, with industrial solvents
and the CDW source contributing to specific toxic VOCs. The relative proportions contributed by the various
sources were remarkably similar at the six sites, i.e., spatial differences were small, at least when compared
580
-------
over 2-day averaging periods. This observation confirms that the sources identified are area and mobile
sources, in that they affect VOC concentrations comparably at all six of the widespread sampling sites.
Temporal differences between the various source contributions also appeared to correlate well with the
meteorological conditions in the three campaigns.4
The results from the six sites were composited to provide an overview of the ambient concentrations
and source contributions in Columbus for each of the major toxic VOC, as shown in Table 1. Table 1 shows
that the sum of the source contributions for each species was generally within 100 ± 20% of the measured
average concentration, indicating adequate source identification. The sums of the source contributions for
dichloromethane, formaldehyde and acetaldehyde are low, as expected, since profiles exist for only some of
the sources of these compounds; this study is one of the first to include formaldehyde and acetaldehyde in
CMB modeling for VOCs. Table 1 indicates that on average, vehicle exhaust contributed about 17% of the
measured formaldehyde and 11% of the acetaldehyde. The remaining unexplained portions of these
compounds presumably come from photochemical oxidation of atmospheric VOCs, and possibly from other
as-yet-uncharacterized, direct emissions sources.
Table 1 emphasizes the importance of vehicle exhaust as a source of the most common toxic aromatic
VOCs, namely benzene, toluene and xylenes. Natural gas, an important contributor of the light
hydrocarbons, was crucial to achieving good model fits, but contributes very little to toxic VOCs. Gasoline
vapor likewise contributes substantially to the light hydrocarbons in Columbus, but is also a minor contributor
to the measured toxic VOC concentrations. In contrast, the industrial solvent and CDW sources are minor
contributors to light hydrocarbons, but are important contributors to individual toxic VOCs. The industrial
solvent source particularly contributes to the aromatic VOCs, excluding benzene; the drycleaning source
profile accounts reasonably well for the measured concentration of the chlorinated solvents, and also
contributes substantially to benzene and toluene. These results are consistent with those obtained in other
CMB source apportionment modeling studies of urban areas in the U.S.e'gi '"3
Modeling of 3-Hour VOC Data for Temporal Variability Analysis
The 3-hr VOC data sets collected at Sites 1, 2, and 5 were selected for further CMB source
apportionment modeling. The motivation for modeling the 3-hr VOC data sets was the potential for
identifying temporal, and spatial phenomena from the enhanced time resolution available in these data.
Figures l(a,b) show the short-term temporal trends in the relative source contributions to benzene and
toluene, for the 16 consecutive 3-hr VOC data sets collected at Site 1 during Campaign 1; similar analyses
were conducted for other VOC species, sites and sampling campaigns.4 The results for benzene (Figure la)
reveal that vehicle exhaust was the predominant contributor (>90%) to the measured concentrations.
Gasoline vapor and the CDW composite source together contribute typically 5-10% to the measured levels of
benzene, and natural gas contributes a trace fraction. No diurnal trends were discerned in the modeled
percentage source contributions probably due to the fact that a single source, vehicle exhaust, was responsible
for the predominant fraction of measured benzene levels.
Figure l(b) shows the results for toluene at the same site and campaign. Toluene was attributed
primarily to vehicle exhaust, industrial solvent usage, and the CDW composite source, with a trace
contribution from gasoline vapor. Again, no significant diurnal trends are discernible, although the industrial
solvent contributions to the toluene levels vary sporadically during the campaign. The CDW composite
source contributions to toluene are generally stable.
A number of other observations regarding the spatial and temporal variability of the VOC sources
were also made from the results of modeling,4 including the following:
• For both residentially and commercially-located sampling sites, the absolute vehicle exhaust contributions
were characterized by a peak between 6-9 a.m., followed by lower levels during the day and a second
peak in the late evening hours from 9 p.m.-midnight, and a minimum between 3-6 a.m. These trends are
consistent with the traffic activity patterns that characterize urban areas. The occurrence of the second
peak during the 9 p.m.-12 a.m. hours rather than during the 6-9 p.m. period corresponding to the
evening rush hours may be due to summer-time, evening-hour meteorological conditions that result in
poor dispersion of vehicle exhaust emissions.
• On average, the absolute source contributions from gasoline vapor were similar to each other at the
residential sites, and lower during the daytime hours than during the corresponding night time hours. At
the commercial site, the opposite trend prevailed with daytime absolute gasoline vapor contributions
generally higher than the corresponding night time contributions. The contributions between the hours of
581
-------
midnight-6 a.m. were similar at all sites, but slightly higher at the residential sites. These results suggest
that the gasoline vapor source is correlated to a greater extent with the location and presence of parked or
moving vehicles, than with fixed locations such as gasoline refueling stations, as previously theorized.
Other observations were also made from model results, such as the variability of vehicle exhaust and
gasoline vapor contributions with ambient temperature, and the variability of the natural gas source, industrial
solvent, and CDW source contributions.4
VOC Emissions Estimates
The results from this study can also be used to make estimates of the area-wide emissions from
various source categories for specific VOCs. Columbus is located in predominantly agricultural central Ohio
with no other significant urban areas within a 60 mile radius; the transport of background concentrations of
modeled VOCs into Columbus can thus be reasonably assumed to be negligible. A simple box model analysis
can then be used to characterize the dilution of VOCs in the Columbus area according to the relation, C=
Q/(U L M), where C is the VOC species concentration, U is the wind speed, L is the mean cross-sectional
dimension of the geographical area, and M is the mixing height. For the 6 sampling days in this study, the
average wind speed was — 8 km/hr and the average mixing height was - 850 m.
Table 2 shows the results of calculations using this approach for the most common toxic VOCs based on
the study-average measured concentrations and source apportionment results in Table 1. Table 2 shows that
benzene and ethylbenzene are predominantly emitted by mobile sources; toluene releases are similar for
mobile and area source; the two chlorinated organics originate only from area sources. Table 2 also shows
the fractions of these estimated area source emissions for each species that are represented in the TRIS
listings8 for Franklin County. The small TRIS-reported fractions for all of the species suggest that the
majority of the area source emissions for these VOCs probably originate from the numerous smaller facilities
that are not included in the TRIS listings; these results are similar to those reported by Sweet and Vermette,2
using a box model approach for two study areas located in Chicago and East St. Louis, IL.
CONCLUSIONS
This study addressed the significance and short-term variability of mobile and area source
contributions to urban toxic VOCs, measured at several sites in Columbus, Ohio. The study showed that the
major sources contributing to the measured VOCs in the region are vehicle exhaust, gasoline vapor, natural
gas, industrial solvents, and a composite of cleaning/degreasing/ wastewater activities. The small degree of
spatial variability within the region of the relative source contributions to toxic VOCs suggested the
ubiquitous nature of these sources. Using the 3-hr integrated samples also allowed evaluation of VOC source
variability to an extent not possible with the 24-hr samples normally used in CMB modeling.
The CMB modeled source contributions were combined with a simple box model of the study area to
estimate the emission rates of area sources. The results from this analysis suggest that substantial toxic VOC
emissions from area sources are likely not included in current emissions inventories.
REFERENCES
(1) Scheff, P.A.; Wadden, R.A.; Bates, B.A.; Aronian, P.P., "Source Fingerprints for Receptor Modeling of
Volatile Organics", J. Air Pollut. Control Assoc. 39: 469 (1989).
(2) Sweet, C.W.; Vermette, SJ. "Toxic Volatile Organic Compounds in Urban Air in Illinois", Environ,
Sci. Technol.. 26:165 (1992).
(3) Ramamurthi, M.; Kelly, T.J.; Pollack, AJ. "Source Apportionment Modeling of Toxic Air Pollutants in
Allen County-Lima, Ohio", A&WMA. Kansas City, MO, June 1991, Paper 92-104.09.
(4) Spicer, C.W. a al., "Variability and Source Attribution of Hazardous Urban Air Pollutants - Columbus
Field Study", Battelle draft final report to U.S. EPA, Contract No. 68-D-80082, June 1992.
(5) Finlayson-Pitts; B.J; Pitts, J.N., Jr. Atmospheric Chemistry: Fundamentals and Experimental Techniques
1st Ed.; John Wiley & Sons: New York (1986).
(6) Toxics in the Community: National and Local Perspectives (The 1989 Toxics Release Inventory National
Report), EPA 560/4-91-014, U.S. EPA, Office of Toxic Substances, Washington, D.C., 1991.
(7) National Emissions Data System, Area Source Emissions Report for 1985, Updated October 1987,
obtained for Franklin County, Ohio from Ohio EPA, Columbus, Ohio.
(8) Greater Columbus Industrial Redliner, 1989-90 Edition, Industrial Map Co., Inc. Alexandria, KY (1989).
582
-------
(9) U.S. EPA, Volatile Organic Compound fVOO/Particulate Matter (PM1 Speciation Data System User's
Manual. Version 1.4, U.S. EPA, OAQPS, Research Triangle Park, NC 27711 (1991).
ACKNOWLEDGMENTS
This work was supported by the U.S. EPA under contract 68-D-80082; the project officer for the study
was Dr. Larry T. Cupitt of U.S. EPA/AREAL. This paper has not been subjected to Environmental
Protection Agency review and therefore does not necessarily reflect the views of the Agency, and no official
endorsement should be inferred.
Table 1. Summary of CMB model results from campaign-averaged 3-hr data, composited over all
six sites and all three campaigns.
Modeled Species
Light
Hydrocarbons'"
Benzene
Toluene
Xylenes
Ethylbenzene
1,1,1-
Trichloroethane
Perchloroethylene
Dichloromethane
Formaldehyde
Acetaldehyde
Measured
Average
Cone.
<0g/m3)
33.4
1.5
5.1
4.9
1.1
4.0
1.6
1.5
3.8
2.4
Source Contributions
(% of measured average concentration)
Vehicle
Exhaust
20
96
59
55
56
0
0
0
17
11
Gasoline
Vapor
36
9
2
0.2
0.7
0
0
0
0
0
Natural
Gas
45
2
0.3
0
0
0
0
0
0
0
Industrial
Solvents
0
0
28
42
19
0
0
0
0
0
CDWa
0
10
26
1
0
89
96
42
0
0
Sum of
Source
Contri-
butions
(%)
101
117
115
98
76
89
96
42
17
11
CDWfDrycleaning/degreasing/wastewater composite source.
b Sum of ethane, propane, butanes and pentanes
Table 2. Comparison of annual toxic VOC emissions calculated using a box model and based on
study-average measurements and source apportionment results.
Toxic VOC
Benzene
Ethylbenzene
Toluene
Xylenes
1,1,1-Trichloroethane
Perchloroethylene
Mobile Source
Emissions1
(tons/yr)
2,200
1,400
4,500
4,500
Area Source
Emissions2
(tons/yr)
220 (0.4%)
460 (0.7%)
3,900(3.1%)
3,500 (3.4%)
6,600 (1.7%)
2,600 (2.0%)
ncludes vehicle exhaust and gasoline vapor
2 Percentage values indicate the fraction of the estimated area source release for each species that is
accounted for by the TRIS database
583
-------
Hour 06
Sample Start Hour
Sample Start Hour
Sources:
1= Vehicle Exhaust; 2=Gasollne Vapor; 3=Natural Gas;
•^Industrial Solvent Usage; 5=Cleanlng/Degreaslng/Wastewater;
Figure 1. Modeled percentage source contributions (bottom) during sixteen (16) 3-hour intervals at Site 1 (located in a commercial area of
Columbus) during Campaign 1 to the measured (shown at top) (a) benzene, and (b) toluene concentrations. Values shown apply to 3-
hour intervals in the six sampling days, e.g., "06" applies to the period 0600-0900 hours, etc.
-------
A Method For Separating Volatile Organic Carbon From 0.1 m3 Of Air
To Identify Sources Of Ozone Precursors via Isotope (UC) Measurements"
George A. Klouda, James E. Morris, and Lloyd A. Currie
Surface and Microanalysis Science Division
Atmospheric Chemistry Group
and
George C. Rhoderick, Robert L. Sams, and William D. Dorko
Organic Analytical Research Division
Gas Metrology Group
National Institute of Standards and Technology (NIST)
Gaithersburg.MD 20899
Charles W. Lewis, William A. Lonneman, Robert L. Seila, Robert K. Stevens
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Atmospheric non-methane volatile organic compounds (VOCs) are known to play an important role in
urban ozone formation during the summer. To respond to the need for a direct measure of VOC source
contributions from biogenic (14C/12C= 10'12) and fossil fuel (14C/1JC=0) emissions, a system and protocol are being
developed to separate the total VOC fraction, which would amount to micrograms of carbon, from 0.1 m3 of
ambient air for accelerator mass spectrometry (AMS) "C. The gas separation system developed at NIST allows
for the simultaneous separation of low vapor pressure (LVP) VOCs and H2O, high vapor pressure (HVP) VOCs
and CO2, CO and CH4 through sequential cryogenic separation and selective oxidation techniques. Preliminary
results of this system and procedure for isolating these fractions show a LVP-VOC blank of 2 + 1 /ig C, 95%
confidence limits, which represents the effect of the separation system plus CO2 cross-contamination.
Hydrocarbons having vapor pressures greater than n-decane are not retained at a level of more than a few percent
in the LVP-VOC fraction. The yield for LVP-VOC fraction is greater than that predicted based strictly on GC/FID
analysis of hydrocarbons from the canister given the amount of air processed. Since oxygenated species were
not among those compounds reported, but may exist as some of the unidentified compounds, the question remains
as to whether or not this class of compounds would have a significant effect on the carbon recovered from the
LVP-VOC fraction. An upper limit of the HVP-VOC blank has been estimated at 7 + 8 /xg C. The recovery of
Cj-Cj hydrocarbons in the combined HVP-VOC and CO2 fraction ranges from 27% to 78%. Lower recoveries
were observed for these same compounds in a hydrocarbon standard having 2 orders of magnitude less C02 than
ambient concentrations. It is expected that yields for this VOC fraction will improve substantially by substituting
the existing cryo-trap with a Russian Doll trap.' A subsequent separation of HVP-VOCs from the C02 fraction
will require a preparative gas chromatographic step that has not yet been developed.
* Contribution of the National Institute of Standards and Technology and the United States Environmental
Protection Agency; not subject to copyright.
585
-------
INTRODUCTION
It is well known that atmospheric volatile organic compounds (VOCs) play an important role in urban
ozone (03) formation during the summer through a complex series of reactions that include nitrogen oxides (NOJ
and sunlight. It is believed that these same mechanisms may be responsible for high levels of O3 found over the
South Atlantic as a result of VOCs, CO, and CH4 produced from urban pollution and biomass burning in South
America and Africa.2 For effective strategies to control urban 03, it is necessary to obtain a reliable assessment
of the relative contributions of natural and anthropogenic VOCs in the atmosphere3. One approach to the source
apportionment of atmospheric VOCs requires radiocarbon (14C) measurement of this fraction separated from
ambient air. Radiocarbon measurements of separated VOCs would give a quantitative measure of biogenic
(UC/12C= 10"'2) and fossil (14C/'2C=0) carbon contributions to this fraction. To address this need, a gas separation
(GS) system has been developed at the National Institute of Standards and Technology (NIST) for the
simultaneous separation of VOCs with relatively low vapor pressures (LVP-VOCs), ca. n-decane and above with
a vapor pressure ~0.1 Pa or lower, and H20, VOCs with relatively high vapor pressures (HVP-VOCs), ca. m-
xylene and below with a vapor pressure ~0.1 Pa or higher, and C02, CO and CH4 from ambient air through
sequential cryogenic separation and selective oxidation techniques. Typical concentrations of these compounds
(or chemical classes) and the amount of carbon expected from each fraction by processing 0.1 m3 of air at 20 °C
and 100% humidity are given in Table 1. Subsequent to this bulk separation, a pure HVP-VOC fraction would
be obtained by preparative (prep) gas chromatographic separation of this fraction from the sample C02. Given
that this approach of isolating VOCs for carbon isotope measurements is successful, the method could be applied
to VOCs from remote areas to address issues such as biomass burning as mentioned above.
A preliminary evaluation of the entire measurement process that will lead to 14C measurements of volatile
organic carbon fraction (VOC fraction) is reported here. Important issues that are addressed include: 1) the
ability to isolate a pure VOC fraction from all other gaseous carbonaceous species, 2) the chemical identity of
this VOC fraction, 3) the level of contamination (blank), 4) the recovery efficiency, and 5) the accuracy of 14C
accelerator mass spectrometry (AMS) measurements at the 20 microgram carbon level. These variables are
important to quantify and control so that reliable 14C measurements of VOC fractions can be made and interpreted
in terms of biogenic and fossil carbon contributions to an urban airshed.
A method for the separation of CO and CH4 from 0.1 m3 of whole air for '"C measurement5 has been
modified for this study to include the separation of VOCs. The separation method for CO and CH4 had been
applied to wintertime samples collected in Las Vegas, NV and Albuquerque, NM.5f6 The results showed a
depletion in 14C relative to the activity of modern carbon (ca. IxlO"12 14C/12C) which suggested that both wood
burning and motor vehicle emissions contributed significantly to the excess CO and CH4 concentrations in these
samples. This is an example of how isotope measurements of specific compounds or classes of chemical species
have been applied to apportion sources of urban pollution. Similarly, it is expected that 14C measurements of
VOC fractions from metropolitan areas during the summertime will provide a direct measure of fossil and
biogenic VOC contributions and thus allow for more accurate assessments of the role that anthropogenic and
natural processes play in the formation of O3.
The purpose of this paper is to summarize our current capabilities to separate atmospheric VOCs from
0.1 m3 of ambient air for 14C measurement. Although the GS system is also designed to separate C02, CO and
CH4, discussion will focus only on VOCs. Information about these other carbonaceous species will be included
only as they pertain to steps taken to isolate a pure VOC fraction, void of any significant contamination. The
omission of these species in this discussion does not suggest that 14C measurements of CO2, CO and CH4 are
unimportant, because they may provide additional source information necessary for accurate interpretation of the
14C-VOC measurements. A more complete discussion of the GS system that would include the ability to separate
586
-------
the other carbonaceous species is beyond the scope of this discussion.
EXPERIMENTAL SECTION
The GS system and separation protocol are being evaluated for contamination (blanks) and recovery of
the VOC fractions by processing NIST "primary" standards, NIST Standard Reference Materials (SRMs), zero
air, helium, and real ambient air samples. The air samples were collected during August, 1990 in Atlanta, GA,
as part of the U.S. EPA Ozone Precursor Study.' The results from these experiments allow for close scrutiny
of the sample collection, chemical separation, and sample oxidation processes. The method of preparing Fe-C
targets with microgram (jig) carbon quantities for AMS I4C measurements has been developed by Verkouteren
et al.8 and validated by Klouda et al.9 From the exploratory work on VOCs reported here, our plan is to establish
an effective protocol for the following: 1) separation of the VOC fraction from CO2, CO and CH4, 2) quantitative
recovery of VOCs as CO2, and 3) minimal contamination. These stages of chemical preparation are important
to control to obtain reliable measurements of the 14C abundance of atmospheric VOCs.
Apparatus for Collecting Ambient Air
A prototype field sampling system was designed to collect an ambient integrated fine-particulate sample
on a quartz-fiber filter and an air sample compressed in a 32-Liter SUMMA® polished stainless-steel canister.
The filtering apparatus used a cyclone impactor, operated at ca. 10 Liters per minute (L/min), to remove the
coarse (>2 fan dia) and fine (< 2 jan dia) particles from the air stream while loading the fine particles on a
quartz-fiber filter. All volumes and flow rates are reported at STP, 101 kPa and 273 K. Air was tapped from
the sampling line downstream from the filter and directed to a second system to obtain an integrated air sample
corresponding to the same time period as the filter sample. While maintaining a constant flow with a 0-2 L/min
flow controller, pre-filtered air was compressed to a final sample pressure of 303 kPa (3 atmospheres) in the pre-
evacuated canister. The system was later modified to function under field conditions which included an
environmental enclosure and an on-line bleeder valve for clearing the system line of accumulated water as a result
of high humidity and temperature conditions.10 Seven ambient air samples were collected during August 1990
in Atlanta, GA, using this sampling protocol.'0 Some important issues with regard to sample collection and
storage include: 1) potential loss of VOCs through the blow-off valve designed to remove any water that may clog
the sample line, 2) possible sample contamination due to storage in canisters for extended periods of time, and
3) stability of compounds, in particular oxygenates, in canisters. These questions are intended to be addressed
at a later time.
Reference Gas Mixtures
Reference gas mixtures with known chemical compositions were used to evaluate the LVP-VOC/E^O
fraction and the HVP-VOC/CO2 fraction (For brevity in the following, the labels for these fractions may not
include "H2O" or "C02", depending on the context). One such mixture was made by diluting to ambient
(nanomole/mole level) hydrocarbon concentrations a natural gas sample containing eight hydrocarbons (C,-C6)
of known chemical composition; five of the eight hydrocarbons had also been analyzed for their 13C/12C
abundance. The diluent (balance) zero air was Scott Specialty Gases Ultra Zero Ambient Monitoring (UZAM)
air containing C02 at ambient concentrations (ca. 370 micromole/mole) in air. The presence of C02 in this
mixture enabled us to evaluate the system for separating C02, ~ 17.5 mg C, from LVP-VOCs, -30 jig C; the
amounts expected from 0.1 m3 of air. The fact that C02 was already in the balance gas made it possible to reach
the target concentrations for all species in the final mixture by a simple two-stage dilution. Although the total
VOC concentration of the natural gas reference mixture was too low for evaluating the VOC recovery, the
material was useful for determining the blank for the separation of the LVP-VOC fraction from C02, CO, and
587
-------
CH4. For future separation and recovery experiments, natural gas samples with known isotopic composition will
be diluted with UZAM air to produce ^mole/mole levels. In addition, two separate NIST "primary" standards
were used to evaluate the efficiency of the separation process: #8363 containing CfCs hydrocarbons in air and
#8392 containing C6-C10 hydrocarbons in N2 at nanomole/mole concentrations.
System and Procedure for Gas Separations
System Characteristics. The gas separation system is designed to separate simultaneously the following chemical
fractions from 0.1 m3 whole air: 1) LVP-VOCs and H2O, 2) HVP-VOCs and C02, 3) CO oxidized to C02, and
4) CH4 oxidized to CO2 (Figure 1). The lower half of the system is the sample processing section made of glass
tubing with primarily glass valves. This section contains flow controllers, traps for collecting condensible gases
at cryogenic temperatures, a packed column of Schutze Reagent (I205 on silica gel) to oxidize CO to C02 at room
temperature, and a packed column of Ft on alumina pellets operated at 830 °C to oxidize CH4 to C02. The flow
controllers, limited by the manufacturer's reported precision of 1 %, are calibrated for accuracy against a Califlow
mercury piston and the NIST mercury piston standard.
The upper half of the system is made entirely of stainless-steel high vacuum fittings and valves for
manometry of microliter (micro-line) and milliliter (macro-line) volumes of gas. The micro-line, for manometry
of VOCs, CO, and CH4, oxidized to CO2, uses a 13.3 kPa (100 Torr) pressure transducer and a calibrated volume
of 23.3 + 0.3 mL at 23 °C. All uncertainties reported are 95% confidence limits. The macro-line, for
manometry of the flVP-VOC/COj fraction, uses a 133 kPa (1000 Torr) pressure transducer with a calibrated
volume of 69.5 + 1.1 mL at 23 °C. The confidence intervals reported for these manometer volumes are
dominated by a long term (months to years) uncertainty in the behavior of the pressure transducers. [It is likely
that these volume errors can be reduced by periodic calibration using a single flask volume which has been
precisely calibrated (+ 0.005%) using H2O.] Each manometer system is contained in an individual Plexiglas®
"adiabatic box" to control the temperature during the pressure measurement for subsequent calculation of the gas
volume using the ideal gas law. The pressure transducers are calibrated at NIST using a medium-range ultrasonic
interferometer manometer."
Separation Procedure. A reference gas mixture or ambient air sample, previously analyzed for gas composition,
is adapted to one of the two inlet ports (II or 12) of the GS system (Figure 1). For most experiments, inlet II,
with a 0-5 L/min flow controller, is used at a flow of 0.5 L/min. The 0-200 cnvVmin flow controller is used only
for processing He when evaluating the possibility of system contamination or for on-line dilution of NIST high
concentration Otmole/mole) SRMs. Prior to processing, the entire gas separation line up to the canister valve
is first evacuated with the roughing pump (bottom center of Figure 1), then with the turbo molecular pump (top-
center, Figure 1) to lower than 0.1 Pa (1 mTorr). Next, the three valves for each of the four cryo-traps are
closed and the traps are reduced to their respective operating temperatures. Trap #1, which isolates the LVP-
VOCs and H20 is maintained at -80 °C to -90 °C with an ethanol/dry ice/liquid nitrogen slurry. However, given
the flow rate, trap design, and unpacked trap, the actual temperature of the gas stream at the outlet of the trap
averages around -70 °C for the cryogen at -85 °C. Trap #2, which collects the combined HVP-VOCs and C02,
is operated at -196 °C with liquid nitrogen. Similarly, the sample air stream at the outlet to the liquid nitrogen
trap, under processing conditions, runs about -170 °C. Trap #3, which recovers C02 from the selective oxidation
of CO over Schutze Reagent at room temperature, is operated at -196 °C. Finally, trap #4, which collects C02
and H20 from the high temperature (ca. 830 °C) catalytic oxidation of CH4 to CO2 in the presence of oxygen
(OJ, is maintained at -196°C. Prior to actual processing of SRMs and reference materials, the flow rate is
established through a three-way valve just downstream of the flow controller. For samples, where virtually the
entire volume is to be processed, the flow rate is not established prior to processing. At this point, the trap
isolation valves are opened and the gas is introduced to the system for processing.
588
-------
The cryo-traps were designed with loops that pass in and out of the cryogen to avoid losses due to the
formation of aerosols from the condensation process. Sections of the cryo-traps outside the cryogen were heated
enough to avoid subambient cooling due to the cryogen vapor. Since the automatic liquid nitrogen filling process
produced excessive N2 vapor during filling, it was necessary to occasionally heat the outside coils as high as 100-
150 °C to avoid cooling below ambient conditions. This lack of temperature control was likely the cause of an
observed inefficiency in collecting all the atmospheric CCX in trap #2 which resulted in a COj contamination of
the CO fraction in the Atlanta samples. By replacing trap #2 with a Russian Doll trap, collection efficiencies for
C02 are expected to improve to 100%. Additionally, the efficiency of trap #1 may have been affected by poor
control of the heated portion of the trap outside the cryogen. In the future, this trap will be packed with
deactivated glass wool with better control of the heated portion of the trap in an attempt to improve the efficiency.
Following the separation process, the inlet is closed and the system is pumped to high vacuum using the
roughing pump, then the turbo molecular pump as before. The trap isolation valves are then closed, and the
condensible gases from traps #2, #3, and #4 are individually cryo-transferred to the appropriate section of the
upper manifold for manometry to calculate recoveries for each fraction. The sample fractions, as C02, are then
transferred to individual Vycor break-seals for storage to await the preparation of AMS targets for 14C analysis.
Since it is not possible to get an accurate manometric measurement of. the LVP-VOC fraction in the presence of
1 mL of H20, this composite fraction is expanded into a heated (80 °C) portion of the processing line just to the
right of the trap and completely condensed at -196 °C in a Vycor break-seal for sealing and storage to await
oxidation to C02. As a measure of recovery, the volume measurement of each fraction, except for the LVP-
TOC/H20 fraction, is compared to the expected volume determined for the class of compounds in the canister
or cylinder based upon concentration and sample volume. For real ambient samples, the mass of air processed
and the ambient air density are used to calculate the volume of air processed. Hydrocarbon concentrations are
determined by gas chromatography (GC) with flame ionization detection (FID).
To a first approximation, VOCs are condensed in traps #1 and #2 according to their vapor pressures12
under processing conditions. As such, the VOC fractions are operationally defined as LVP-VOC (trap #1) and
HVP-VOC (trap #2) fractions. This somewhat crude separation step is designed to separate compounds having
higher boiling points from those having lower boiling points. This bulk separation should make it easier to purify
the HVP-VOC fraction from sample C02 using prep GC prior to oxidation of this VOC fraction. The two VOC
fractions would then be combined for a single 14C measurement.
Evaluation of the Separation Process
Composition of LVP-VOC/E^O and HVP-VOCICO? Fractions. Recoveries are studied by determining the volumes
and the chemical compositions of LVP-VOC/H£> and HVP-VOC/C02 fractions separated from dry "primary"
standards and wet ambient samples. Following the separation of these materials, the total condensed volume is
determined by manometry and then transferred to a break-seal vial for storage to await chemical analysis.
Subsequently, the condensates are transferred in vacua from their break-seal storage containers to a GC loop at
-196 °C for injection on column. The column is a Chrompack fused silica open tubular column, 25 m x 0.53
mm ID coated with a lOjim film thickness of AU03/KC1 (PLOT) operated under the following conditions: 45
°C for 10 min, 10 °C/min to 200 °C and hold, 3 mL/min carrier flow rate, 30 mL/min make-up N2, and FID
at250 °C. For calibration of the GC/FID, NIST "primary" standards were cryogenically preconcentrated at -196
°C in the GC loop at a flow of 75 mL/min for 3.3 min, equal to a total injection volume of 250 mL. GC/FID
analysis of the LVP-VOC/H2O fraction from the ambient samples will require purge and trap techniques for
injection on a DB column. The remaining H20 will be analyzed for compounds that may be completely soluble
589
-------
in H2O. This approach provides quantitative yield information from measured concentrations of small volumes
(/iL-rnL) of gases, except for possibly a few very light hydrocarbons, e.g., ethane, that may exhibit loss during
the transfer to the loop, even at -196 °C.
The composition of blank condensates are measured by quadrupole mass spectrometry (QMS) and
GC/FID. The extent that CO2 is entrained in the LVP-VOC/H2O fraction is quantified by allowing the liquid H20
to equilibrate with its vapor in a laser cell while measuring the C02 concentration by diode laser IR absorption
analysis. This latter technique is non-destructive allowing for the quantitative re-recovery of the LVP-VOC/Hf>
fraction.
Volume of CO., from the Oxidation of LVP-VOCs. Subsequent to the separation process, the combined LVP-
VOCIH-fl fraction is subjected to two freeze-thaw cycles in an attempt to separate CO2 that may have been
dissolved or entrained in this fraction during separation. The procedure involves the quantitative transfer of the
LV/"-VOC/H2O fraction from the break-seal to the vacuum line via a metal bellows. A vacuum is drawn on the
break-seal, the tube cracked, and the fraction is transferred quantitatively at -196 °C to a cold finger just to the
right of the micro-line adiabatic box in Figure 1. The fraction is then warmed to -78 °C for 5-10 minutes at
which point any CO2 occupying the head space is condensed at -196 °C in the calibrated volume for manometric
measurement.
The oxidation of the LVP-VOC fraction is accomplished by first cryo-transferring the contents of this
fraction from the break-seal to trap #3. Then, the pre-evacuated trap #4, used previously for collecting CQ
derived from the oxidation of CH4 during the separation process, is reduced to -196 °C and opened to the vacuum
system on the right in Figure 1. With the temperature of the Pt/Al catalyst at — 830 °C, ultra-high purity (UHP)
O2 (99.99%) is introduced at 0.1 L/min through the II inlet, through traps #2 and #3 while at-196 °C, bypassing
the Schutze Reagent (I2O5), and flowing up to trap #3 containing the LVP-VOC/H20 fraction. By opening trap
#3 valves (bypass valve closed), 02 begins to flow through the trap, through the catalyst and into trap #4. After
approximately 3 min, the pressure in the system reaches ca. 13.3 kPa (< 100 Torr) at which time the left valve
to trap #4 is opened to the roughing pump to draw the LVP-VOC/Ef) fraction completely through the catalyst
for oxidation with subsequent trapping of the combustion products plus water in trap #4. This takes
approximately 10 minutes and is estimated to use ca. 1 Liter of UHP 02. Once all the LVP-VOCs are combusted
to CO2 and collected along with the water in trap #4, the 0, source is closed and the residual O2 is evacuated
from the processing line. This total condensate is then quantitatively cryo-transferred using liquid nitrogen to
a cold finger just to the right of the micro-line adiabatic box for distillation of the CO, from the H2O followed
by volumetric determination of the CO,. The blank for this oxidation procedure is evaluated by processing a "no
sample" under identical conditions to that of a LVP-VOC fraction.
The recovery efficiency for separating the LVP-VOC fraction from ambient air samples is evaluated by
comparing the calculated mass of carbon from the measured volume of CO2 recovered from the oxidation of this
VOC fraction to the expected mass of carbon. The expected mass of carbon is based on the volume of air
processed and the total LVP-VOC concentration (as nanomole C/mole) measured in the original air sample from
the canister. GC/FID analyses are performed on the air samples by pre-concentration of 250 mL of gas at liquid
Ar temperature prior to injection onto a DB-1 capillary column.
AMS Fe-C Target Preparation and "C Measurement
Iron-carbon targets for AMS "C measurement are prepared by the Fe catalyzed method for reduction of
CO2 to graphitic carbon.'3 This method has been modified by Verkouteren and coworkers8-1'1 for small
590
-------
(microgram carbon) samples by using hot Zn, replacing Fe powder with Fe wool, and including H2 to minimize
CO recombination thereby reducing CO, to graphitic carbon directly on the Fe wool. Finally, the Fe-
wool/Carbon matrix is fused in an inert atmosphere to a solid solution (bead). The iron serves as a carrier for
the reduced carbon and is believed to be less susceptible to contamination. This "closed-system" technique has
been applied to several studies with sample sizes as small as 16 ^g C.9
The 14C/13C ratio measurements of NIST SRMs as Fe-C beads are performed on the University of Arizona
tandem accelerator mass spectrometer. The measurement sequence or "run" consists of cycling 8 times between
"C (40 s) and 13C (4 s). The mean "C count rate is divided by the mean 13C count rate for each run. The
number of runs varies from 4 to 16 per target. This procedure is applied to as many as 32 targets per target-
wheel, generally 26 samples and 4 oxalic acid standards for radiocarbon dating, NIST SRMs 4990B and 4990C
and 2 NIST RM21 Graphites. The NIST RM21 Graphite is included as a third standard to quantify the amount
of modern carbon contamination. Additional details regarding the instrument and further discussion of the
procedures are reported elsewhere" Replicate measurements of the 14C/13C ratio obtained from the same target
are pooled by calculating a weighted mean and lu-Poisson standard error. The current AMS 14C measurement
sensitivity for modern carbon as an Fe-C bead is 20 /ig-C at the 1.4% precision based on counting statistics.9
Results of RM21 Graphite and SRM 4990B Oxalic Acid measurements show that contamination due to the
reduction chemistry for preparing targets is 1.0 + 0.5 /xg of modern carbon for samples in the 20-50 m> C range.
RESULTS AND DISCUSSION
A series of exploratory experiments was performed on the NIST GS system to evaluate its capabilities
and a proposed procedure to isolate VOCs from 0.1 m3 of ambient air. At the outset, it was believed that the
blank, representative of the entire measurement process, would likely be significant since the amount of total
carbon recovered from the VOC fraction was estimated at around 50 /jg C (see Table 1). As a result,
experiments were designed to minimize and control the blank. For blanks that were significant, the material was
retained in break-seals for manometric, chemical, and, whenever possible, isotopic measurements. The recovery
of VOCs was also studied by processing hydrocarbon standards to evaluate the degree of separation and yield.
Blank Evaluation
Since the method presented for separating and converting VOCs to a suitable chemical form for AMS 14C
measurements requires multiple chemical steps, a thorough examination of the entire process blank is essential.
The following blanks associated with the LVP-VOC fraction are under investigation: 1) the system blank, a result
of the interaction of the sample with the GS system and Vycor break-seals, 2) entrainment of sample C02 (cross-
contamination) in the water (ice) lattice, and 3) the oxidation blank for the combustion of VOCs to CO2. The
AMS target preparation blank, mentioned in the experimental section, has been evaluated separately and will be
reported elsewhere.9 Preliminary results of experiments related to the blank for the LVP-VOCs are summarized
below.
LVP-VOC Fraction Blanks. Reagent helium, a natural gas sample diluted to nanomole/mole concentrations, CO
diluted to the ftmole/mole level, and Atlanta samples (Table 2) were processed to assess the overall blank
contribution to the LVP-VOC fraction. These reference materials included the following characteristics: 1) the
He was expected to be virtually free of all fractions of interest, 2) the natural gas mixture contained C2-C6
hydrocarbons, CH4 and CO2 at or near ambient concentrations, 3) the CO mixture also included ambient C02,
and 4) the ambient samples were chemically characterized. All but the ambient samples were processed under
dry conditions. By processing 0.1 m3 of these materials, blank contributions were evaluated. Analysis of the
trapped condensates were made by manometry, semi-quantitative gas analyses by QMS, GC/FID analysis, and
591
-------
diode laser IR absorption measurements. All blank quantities reported below are in units of fi.L at STP and, for
those fractions that have been chemically characterized, /ig C has been included.
The system blank was estimated by processing pure He. The average total volume of condensate
recovered from trap #1 (LVP-VOC fraction), determined by manometry, was 1.2 + 1.0 /JL. Of the three
replicate system blanks, QMS results of one experiment showed that 95% of the 0.98 /iL recovered was H20,
the other 5% was identified as CO2. This would amount to a 0.05 fig C system blank (0.1 /*L). If the gas
composition from this single experiment was representative of the gases from the other two experiments, then
the effects of processing and storage would not influence the final result in a significant way, assuming of course
that this carbon is of modern activity which has been our past experience. Even if one assumes that the total
blank volume recovered was CO2, this amount of carbon contamination, 0.6 + 0.5 pg C, would be relatively
small (ca. 2.6%) contribution to the total sample carbon expected (ca. 25 tig C).
Next, the possibility of sample C02 cross-contamination, while condensing the LVP-VOCs, was examined
by processing 0.1 m3 of the natural gas mixture which contained CO2 at ambient levels. From this experiment,
a total volume of 1.8 pL was expected given that the combined concentration of the heaviest hydrocarbons (C5's
and C6's) in the mixture amounted to 18 nanomole/mole and assuming 100% trapping efficiency. Later, it was
realized that this expected volume would represent an upper limit since compounds with vapor pressures as low
as ~ 0.1 Pa (1 mTorr) were only partially retained at -80 °C to -100 °C. Of the total condensate from trap #1,
QMS analysis showed that 94-97% was H2O. The other few percent was identified as C02, which amounted to
0.3 + 0.5 iig C. This estimate also included the system blank effect reported above.
The effect of CO2 cross-contamination was also explored by subjecting three Atlanta LVP-VOCIHJ3
fractions, GTS, GT6, and GT7, to two freeze (-196 °C) and thaw (-80 °C) cycles to remove any possible C02
from these fractions. Each fraction was individually treated by adapting the break-seal containing the LVP-
VOC/H2O fraction to the vacuum line by way of a metal bellows such that a vacuum could be established around
the tube. The freeze/thaw procedure was performed as previously outlined. After a thawing for a period of 5-10
minutes at -80 °C, the total volume of head-space gas, assumed to be CO2, was at most 0.03 ± 0.01 /xL of gas
(or 0.02 + 0.01 fig C). It was later suggested that this estimate would be a lower limit since the C02 equilibrium
between liquid and gaseous water would not have been established within the sampling period; it was estimated
that equilibrium would be established after about 24 hours.16 Therefore, any C02 present would have remained
in solution. Given the solubility of 350 /imole/mole C02 in equilibrium with water at 0 °C, the lowest
temperature at which the solubility constant has been measured, the estimated volume of C02 in the liquid would
have been about 0.6 /iL (or 0.3 jig C).
A more direct means of measuring CO2 cross-contamination in the LV7J-VOC/H2O fraction was
investigated using the non-destructive technique of Diode Laser IR second derivative absorption analysis. A
sample was introduced into a 74.8 cm path length cell with a cell volume of 294.8 mL, the total volume of the
system of 427.1 mL, via a break-seal containing the LVP-VOC/Hf> fraction, adapted as described above, and
held at ice bath temperature. The break-seal was then cracked introducing water vapor into the system until a
pressure of 613 Pa (4.610 Torr) was reached, i.e. the vapor pressure of H20 at about 0.2 °C. Next, the
concentration of C02 was measured periodically over 24 h while the system was approaching equilibrium.
Measurements of CO2 concentration were made on Atlanta GT2 and GTS samples relative to a NIST SRM C02
in air certified at 351.2 jimole/mole. Average CO, concentrations measured at equilibrium represent ~ 1 /xg C
contamination from CO2 (Table 3). Henry's Law constant for the equilibrium of C02 and H2O at 0 °C, -log K
= 1.1 moles-Liter'-atm"1,17 was used to calculate the amount of CO2 remaining in the liquid at equilibrium,
approximately 2 orders of magnitude less than that in the gas phase. The median value of carbon contamination
592
-------
measured in these two experiments, 0.9 ng C, has been applied to the correction of the sample (LVP-VOC) carbon
for yield purposes.
The final blank studied was for the oxidation of the LVP-VOC to CO2. Given the oxidation procedure
outlined in the experimental section, the blank was estimated by performing a series of combustions with "no
sample" present. The average total volume of condensate collected from these combustion runs following
distillation from -78 °C was 2.7 +1.5 /nL. Blank results from the oxidation of CH4 using this same catalyst
suggested that this blank would be predominantly CO2. Assuming this to be the case, the oxidation step would
contribute 1.5 + 0.8 jig C. Therefore the combined system and C02 cross-contamination blank (0.9 ng C) plus
the oxidation blank (1.5 ng C) yield an overall LVP-VOC blank of 2 + 1 /ig C.
HVP-VOC Fraction Blanks. The experiments that involved the processing of He, as mentioned above for the
LVP-VOC blank, were also used to evaluate the system blank for the condensation of the HVP-VOC/CO2 fraction
during the bulk separation process. The total condensates recovered from these experiments that would represent
the effect of cryo-trapping this fraction were 7.2, 14.8, and 18.4 /iL. These blank volumes represent an upper
limit of 13.5 + 14.2 jiL, since the He alone may contain trace level impurities of C02, H2O, and hydrocarbons
and the chemical composition of this condensate was not determined. If one assumes that CO2 is the largest
impurity, then the estimated blank would he equivalent to 7 + 8 jug C.
The final phase necessary for isolating a pure HVP-VOC fraction will require prep GC separation of
sample CO2 from this material followed by oxidation of the purified HVP-VOC fraction to CO2. An alternative
approach to Prep GC to obtain a pure VOC fraction may be to preprocess the air from the canister with a LiOH
denuder18 followed by complete condensation of the air at -196 °C into another canister. The HVP-VOCs would
then be separated from whole air using the N1ST GS system. Success of this latter method would greatly depend
on the denuder efficiency, the chemical blank, and whether losses of VOCs occur during the process. If VOCs
are lost during the process, these compounds and their efficiencies must be quantified for an accurate
interpretation of 14C measurements. It is believed that some oxygenated compounds are less likely to survive the
denuder step than are the hydrocarbons." Thus, while the original objective was to measure the 14C abundance
of the total VOCs, the final HVP-VOC fraction may be dominated by the unreacted (primary) light VOCs. If the
denuder is effective at removing the CO2 to an acceptable level, 1 part in ~ 105, but also removes oxygenates in
the process, then a I4C measurement of the "primary" VOCs may also contribute to the understanding of source
contributions to O3 formation.
Currently, the denuder is being evaluated for its ability to remove CO2 from moist air and its effect on
VOCs at the nanomole/mole level." Also, plans are underway to investigate the feasibility of prep GC for
separating C02 from the HVP-VOC fraction and chemically characterizing the LVP-VOC/E2O fraction for
oxygenates with the intention of recovering both VOC fractions for isotope measurements.
Recovery Evaluation
Recoveries of VOCs from the separation process were evaluated by processing two "primary"
hydrocarbon standards, C2-Cg (#8363) with concentrations from 15-25 nanomole/mole and C6-C10 (#8392) with
concentrations from 26-124 nanomole/mole, plus two Atlanta ambient samples (GTS and GT6). For the
hydrocarbon standards, recoveries for both the HVP-VOC/CO? and the LVP-VOC/H2O fractions were obtained
by comparing measured hydrocarbon concentrations in each fraction to their expected concentrations. Only
Atlanta samples were used to determine the recovery for separation and oxidation of the LVP-VOC fraction by
593
-------
comparing the mass of carbon recovered as CO, from the oxidation to the expected mass of carbon from the total
concentration of hydrocarbons (in micromole C/mole) expected in this fraction.
Although the Atlanta samples were measured for their chemical composition after 1.5 years of storage
in canisters at laboratory conditions and the VOC separations took place six months later, this long storage period
was not of major concern, since one objective of the study was to evaluate the gas separation system and
procedure using real ambient samples. But, for quality isotope measurements, sample integrity during the storage
must be well characterized. The GC/FID analyses of the Atlanta samples revealed two unidentified asymmetric
peaks in each sample chromatogram that are not normally observed in routine hydrocarbon analysis of ambient
samples. Since these peaks represent a substantial portion (37-136 nanomole C/mole) of the total VOC carbon
concentration, without identification and quantitation of these compounds their effect on the estimated recovery
and integrity of the VOC fractions is not known. If these peaks represent ambient oxygenated compounds, such
as acetaldehyde or acetone, the recoveries of these compounds may have been affected by the warm canisters
from the field stored under cooler laboratory conditions. Given high humidity, in some cases saturated
conditions, some H20 would be expected to condense on the canister surface and potentially carry water-soluble
VOCs along. Aside from these important but very complex issues, the Atlanta samples afforded a good
opportunity to evaluate the measurement process because they represented the actual gas composition of interest.
GC Analysis of HVP-VOC/CO* and LVP-VOCIH& Condensates. From the processing of two dry hydrocarbon
standards, #8363 and #8392, and two Atlanta samples, GT5 and GT6, an aliquot was taken from each HVP-
VOC/CO2 fraction and analyzed for gas composition by GC/FID. The recoveries for isolating C5-C8 VOCs in
the sample CO2 were determined by multiplying each concentration by the ratio of the total HVP-VOC/CO,
volume recovered to the total volume of gas processed and dividing by the ambient concentration determined
directly from the canister. Assuming that the transfer of an aliquot from the break-seal to the GC loop was
complete, HVP-VOC recoveries from the processing of #8363, GTS and GT6 were 27-78% (Table 4). The
results for n-hexane from the two ambient samples were in excess of 100% which suggest either a problem with
calibration or contamination. Lower HVP-VOC recoveries were observed for #8392 which may result from a
CO2 concentration, 2 ^mole/mole, 2 orders of magnitude lower than in #8363, GT5 and GT6. The presence of
dry ice from the CO2 condensate may be a factor influencing the efficiency for trapping these hydrocarbons
considering the cryo-traps were not packed. Losses of these light hydrocarbons from an inefficiency in the
cryogenic trapping should become insignificant in the future with the implementation of Russian Doll traps that
have been proven to yield 100% recovery for even nanomole/mole concentrations of ethane at -196 °C.' Iflosses
occurred during the transfer of the HVP-VOCICQ? fraction to the GC loop for analysis, then the reported
recoveries represent a lower limit.
Recoveries for C6-C10 hydrocarbons in the LVP-VOC fraction ranged from 0.01% for benzene to 19%
for n-decane. It appears that compounds that have vapor pressures equal to or less than m-xylene or n-decane
and have concentrations greater than 25-50 nanomole/mole would be significantly retained in this VOC fraction.
To date, only one LVP-VOC/Hf) fraction from an Atlanta sample (GT3), has been analyzed by GC/FID for its
gas composition. A —50 jiL aliquot of H,0 vapor was taken by expanding the head-space gases and liquid HjO
from the break-seal to a large (~ 100 mL) volume in the Micro-line section of the GS system (Figure 1) that
included 2 break-seals. Once the H,O vapor reached -2.4 Pa (18 Torr), the vapor pressure of water at 24 °C,
valves leading to the break seals were closed, the total gas condensed at -196 °C, and the tube flame sealed.
Using a PLOT column, GC/FID results of this LVP-VOC/H2O fraction showed several (fa. 25) major peaks
below n-decane, but, for the moment, have not been identified. If oxygenated compounds were retained in this
fraction, given the GC column used, these compounds were not likely to be resolved and may even be retained.
594
-------
Future GC/FID and mass selective detection analysis of LVP-VOCfB2O fractions from Atlanta samples will be
made with a column more suitable for polar compounds and make use of purge and trap techniques to obtain a
more representative sampling of this fraction.
LVP-VOC Oxidation Recoveries. Four Atlanta samples were processed and their LVP-VOC fractions oxidized
to C02 to evaluate the recovery. The yields, corrected for a total process blank, were significantly in excess of
the amounts of carbon expected based on hydrocarbons from n-decane and above measured by GC/FID (Table
5). These results raise important concerns regarding the entire chemical process and the means for evaluating
the yield. If the GC/FID measurements were an accurate means for evaluating the LVP-VOC recoveries, then
the relative amounts of excess carbon prevents one from interpreting the 14C measurements with any sort of
confidence. Alternatively, if the unidentified peaks less than n-decane were condensed in the LVP-VOC/H20
fraction, this carbon mass would account for a significant portion of the excess carbon. In addition, some
oxygenates may not have eluted from the column and, therefore, would also contribute to this excess carbon
mass. AMS "C measurements were performed on LVP-VOC fractions from Atlanta samples GT5, GT6, and
GT7, but, until this excess carbon is completely accounted for, the isotope results remain uninterpretable.
CONCLUSIONS
Preliminary experiments were performed on a gas separation system developed at NIST by processing
0.1 m3 of hydrocarbon standards and ambient air samples for the removal of VOCs for isotope (UC)
measurements. To date, two separate condensates containing VOCs have been isolated: 1) LVP-VOCs
condensible at ca. -85 °C, generally those having vapor pressures less than that of n-decane, plus H20 and 2)
HVP-VOCs condensible at ca. -196 °C, generally those with vapor pressures above that of n-decane, plus CO2.
Results of the LVP-VOC fraction suggest a blank of 2 + 1 /xg C and a 19% recovery for n-decane. Gas
composition analysis by GC/FID of this fraction isolated from one ambient sample collected in Atlanta, GA
detected several compounds below n-decane which have yet to be identified. Masses of carbon recovered from
oxidations of four LVP-VOC fractions from Atlanta samples showed an excess carbon that cannot simply be
accounted for by the blank. We believe that this excess of carbon may be oxygenated VOCs that are not
measured in the original gas composition analysis of the samples.
An upper limit for the HVP-VOC/CO2 blank has been estimated at 7 + 8 jig C, assuming that C02
dominates the volume measured. The recoveries of a few select hydrocarbons from this fraction range from 27%
to 78% and are variable. The recovery of VOCs in this fraction is expected to improve significantly with the
substitution of a Russian Doll trap for cryogenic collection. Investigations are underway to evaluate both
preparative GC and a LiOH denuder to separate CO, from the HVP-VOCs. In spite of the many difficulties
summarized here, isotope measurements of either the total VOC fraction or the two subtractions continues to be
a promising approach for determining the biogenic and fossil contributions to VOC precursors of ozone.
ACKNOWLEDGEMENTS
The authors would like to thank G. Mattingly and P. Baumgarten of the Fluid Flow Group at NIST for
their assistance in calibrating flow controllers. Also, the authors wish to thank D.J. Donahue, A.J.T. lull, L.J.
Toolin, A.L. Hatheway, T. Lange and D.L. Biddtilph for their assistance in the measurement of 14C by AMS.
Disclaimer
The information in this document has been funded in part by the United States Environmental Protection
595
-------
Agency under Interagency Agreement #DW 13935098-01 -3 to the National Institute of Standards and Technology.
It has been subjected to Agency review and approved for publication. Mention of trade names or commercial
products in the text does not constitute endorsement or recommendation for use, nor does it imply that the
materials or equipment identified are necessarily the best available for the purpose.
REFERENCES
1. C.A.M. Brenninkmeijer, "Robust, high-efficiency, high-capacity cryogenic trap," Anal. Chem. 63: 1182
(1991).
2. A.M. Thompson, "The oxidizing capacity of the Earth's Atmosphere: probable past and future changes,"
Science 256: 1157(1992).
3. W.L. Chameides, F. Fehsenfeld, M.O. Rodgers, C. Cardelino, J. Martinez, D. Parrish, W. Lonneman, D.R.
Lawson, R.A. Rasmussen, P. Zimmerman, J. Greenberg, P. Middleton and T. Wang, "Ozone precursor
relationship in the ambient atmosphere," Jour, of Geophy. Res. 97(D5): 6037 (1992).
4. H. Westberg, "Ozone behavior in the combined Baltimore-Washington, DC plume," EPA Project Summary,
EPA/600/S3-85/070, U.S. Environmental Protection Agency, Research Triangle Park, NC, Dec. 1985.
5. G.A. Klouda, L.A. Currie, D.J. Donahue, A.J.T. lull, M.H. Naylor, "Urban atmospheric "CO and "CH4
measurements by accelerator mass spectrometry," Radiocarbon 28(2A): 625 (1986).
6. G.A. Klouda, L.A. Currie, R.M. Verkmiteren, W. Einfeld and B.D. Zak, "Advances in microradiocarbon
dating and the direct tracing of environment carbon," Journal of Radioanalytical and Nuclear Chemistry 123(1):
191 (1988).
7. L.J. Purdue, "Summer 1990 Atlanta ozone precursor study," in Proceedings of the A&WMA 84th Annual
Meeting. Vancouver, British Colombia, paper #91-68.8 (1991).
8. R.M. Verkouteren, G.A. Klouda, L.A. Currie, D.J. Donahue, A.J.T. Jull and T.W. Linick, "Preparation of
microgram samples on iron wool for radiocarbon analysis via accelerator mass spectrometry: a closed-system
approach," Nucl. Inst. and Meth. B29: 41 (1987).
9. G.A. Klouda, L.A. Currie, D. Barraclough, A.J.T. Jull, D.J. Donahue, L.J. Toolin and A.L. Hatheway, "Fe-
C targets for 14C accelerator mass spectrometry: progress at the microgram carbon level," to be submitted to
Nucl. Inst. and Meth. (1993).
10. W.D. Ellenson and J. Adams, Field Study Report: Collection of air samples and air filter samples for
carbon-14 analysis for summer 1990 Atlanta ozone precursor study. Report to the U.S. Environmental Protection
Agency, Research Triangle Park, 1991.
11. S. Dittmann and C.R. Tilford, Report of calibration of pressure transducers. NIST Report by the Vacuum
Group of the Thermophysics Division, National Institute of Standards and Technology, February 22, 1991.
12. T.E. Jordan, Vapor pressure of organic compounds. Interscience Publ., Inc., New York, 1954.
596
-------
13. J.S. Vogel, J.R. Southon, D.E. Nelson and T.A. Brown, "Performance of catalytically condensed carbon
for use in accelerator mass spectrometry," Nucl. last, and Meth. B5: 289 (1984).
14. R.M. Verkouteren and G.A. Klouda, "Factorial design techniques applied to optimization of AMS graphite
target preparation," Radiocarbon 34(3): 335 (1992).
15. T.W. Linick, A.J.T. lull, L.J. Toolin and D.J. Donahue, "Operation of the NSF-Arizona accelerator facility
for radioisotope analysis and results from selected collaborative research projects," Radiocarbon 28(2A): 522
(1986).
16. E. Busenburg, U.S. Geological Survey, Reston, VA, personal communication, 1993.
17. G. Houghton, A.M. McLean and P.O. Ritchie, "Compressibility, fugacity, and water-solubility of carbon
dioxide in the region 0-36 atm. and 0-100 °C, Chem. Eng. Sci. 6: 132 (1957).
18. R.K. Stevens, U.S. Environmental Protection Agency, Research Triangle Park, NC, personal communication,
1993.
19. R.A. Rasmussen, Institute of Atmospheric Sciences, Oregon Graduate Center, Beaverton, Oregon, personal
communication, 1993.
597
-------
Table 1. Expected volume of gas and mass of carbon of each chemical fraction for 0.1 m3 air at 20 °C
and 100% humidity*
Chem. Fraction
H2O
CO2
CO
CH4
vor
Concentration
[/xmole/mole]
14,335
360
I
~2
~1
Volume
1/dJ
1,700
35,000
100
200
100
Mass C
M
-
17,500
50
100
50
* Estimates based on typical summer concentrations in metropolitan areas. The total non-methane volatile
organic compound (VOC) concentration (/^mole C/mole) taken from H. Westberg4 for the summertime in
the Baltimore-Washington, DC area. The compound distribution was comprised of 60% paraffins, 30%
aromatics, and 10% olefins.
598
-------
Table 2. Concentration of species in ambient air samples collected during August, 1990 in Atlanta, GA.
Run*
ID#
Date Time of Day
Concentration [jimole C/mole]
VOC CH4 CO CO2
GT1
GT2
GTS
GT4
GTS
GT6
GT7
3992
3990
3989
3995
3991
3994
3993
6
8
10
10-11
12
13
13-14
0700-1900
0700-1900
0700-1900
1900-0700
0700-1900
0700-1900
1900-0700
0.56
0.28
0.51
1.22
0.52
0.74
0.91
1.93
1.84
1.88
2.59
1.92
1.89
2.51
0.67
0.81
0.75
1.77
0.68
0.76
1.08
387.3
370.0
364.7
426.8
359.6
360.2
389.5
* Total VOC concentrations were obtained from the total area of all peaks detected by GC/FID.
Methane, CO and CO2 concentrations were measured by GC/FID; CO and CO2 were reduced to CH4
prior to the FID. Methane concentrations were also measured by diode laser IR absorption.
All concentrations are relative to standards under dry conditions. Uncertainties are assigned
as follows:
VOC - Total VOC, -5%
CH4- <1%, calibrated against SRMs 1658a,1659a,1660a
CO - <7%, calibrated against SRM 2612a
C02 - ca. 0.3%, calibrated against SRMs 2609, 1672, and two primary gravimetric standards.
599
-------
Table 3. Diode laser IR absorption measurements of CO2 cross-contamination of the
LVP-VOCfRf) fraction separated from Atlanta samples.
ID # CO2 Concentration" Mass C*
[fimole/mole] lug]
GT2 811 + 24 1.23 + 0.04
GT3 335 + 22 0.51 ± 0.05
* Concentrations of C02 measured in 612.6 Pa (4.610 Torr) sample EtjO^ and VOCs
in 427.1 mL at 22 °C in equilibrium with ~ 1 mL of sample H20W at 0.2 °C.
Uncertainties are 95% confidence limits.
t Equivalent mass of carbon calculated from the CO2 concentration measured in the
sample H2OW plus VOCs and the occupied volume.
600
-------
Table 4. Recovery (percent) of High Vapor Pressure (HVP) VOCs trapped with CO2 and Low Vapor
Pressure (LVP) VOCs trapped with H,0 from the gas separation process."
Compound
isopentane
n-pentane
n-hexane
benzene
n-octane
toluene
m-xylene
o-xylene
n-decane
b.p. (°C)
27.8
36.1
68.3
80.1
125.7
110.6
139.1
144.4
174.1
HVP-VOC/C02
#8363
70
72
45
60
33
27
#8392
10
9
6
7
4
ND
GTS
36
78
242
71
ND
72
49
GT6
49
75
158
62
25
59
LVP-VOC
#8363
ND
ND
0.02
0.02
0.2
0.1
1.5
1.8
#8392
ND
0.01
1.4
0.6
16
19
* Data from the processing of NIST "primary" standards #8363 (C2-Cg) and #8392 (C6-C10) plus
Atlanta samples GTS and GT6. Standard #8363 contains hydrocarbons at nominally 20
nanomole/mole and CO, at 226 /^mole/mole in air. Standard #8392 contains CO2 at ~2 /^mole/mole
plus hydrocarbons in N2 at the following concentrations in nanomole/mole: n-hexane, 123.5,
benzene, 102.5, n-octane, 48.2, toluene, 108.3, m-xylene, 47.8, and n-decane, 25.6.
ND = not detected, <0.01%
601
-------
Table 5. Recovery of the Low Vapor Pressure (LVP) VOC fraction of Atlanta samples trapped
with H20 and subsequently oxidized to CCX.
ID # Recovered' Expected" Excess C* Unidentified Ctt
GT2
GTS
GT6
GT7
24.6
22.1
33.6
29.6
0.6
0.4
1.4
2.1
24.0
21.7
32.2
27.5
5.2
11.1
17.9
18.6
* Mass of carbon recovered from the combustion of the LVP-VOC fraction to CO2 corrected
for a total blank of 2 ± 1 /ig C (95% confidence limit); represents system/CO2 blank (n=2)
and oxidation blank (n=3).
** Expected mass of carbon from all peaks above and including n-decane (identified or not),
trap #1 cut off (see text), based on GC/FID analysis of air sample from canister. Concentrations
of unidentified peaks are calculated assuming that their response factor and C-stoichiometry
are comparable to adjacent identified peaks.
t Amount of carbon in excess of that expected for the LVP-VOC fraction based on GC/FID
analysis.
tt Amount of carbon from unidentified GC peaks that eluted before n-decane; if oxygenated,
they may partially account for Excess C.
602
-------
NIST Gas Separation System
HVP-VOCs/CO, LVP-VOCs/H,0
Trap #2 Trap #1
*
PS
-;:
;:
: ; P7 P6
'•I
1
;;
'.' P5
7
-196°C -196°C
CH^ccy ccfccy
Trap #4 Trap #3
Figure 1. Schematic diagram of the NIST Gas Separation system, lower half of schematic, for the separation of the following fractions: a) Low Vapor
Pressure (LVP) VOCs (>C, for hydrocarbons) and H,O in trap #1, b) High Vapor Pressure (HVP) VOCs (
-------
MONITORING VOLATILE ORGANIC COMPOUNDS (VOCs) IN THE GREEN BAY AREA
MARK K. ALLEN and JULIAN CHAZIN
Wisconsin Department of Natural Resources
Madison, Wisconsin 53707-7921
JONNELL HECKER
Wisconsin Occupational Health Laboratory
Madison, Wisconsin 53713
ABSTRACT
In July 1991, the Wisconsin Department of Natural Resources began operating an urban air toxics
monitoring station in the City of Green Bay. The site is a prototype station for a future Wisconsin Air Toxics
monitoring network expected to have 4 to 5 sites. Air samples, collected at the station, are analyzed for volatile and
semi-volatile organic compounds and for nonvolatile elements. This paper will report on the collection and analysis
of the VOCs.
We chose to collect and concentrate VOCs on multi-bed sorbent tubes. After exposure, the VOCs are
thermally desorbed from the tube, focused on a second tube and analyzed by gas chromatography with mass
spectrometry detection. This paper will report on the following;
1. the choice of the adsorbent tubes and the sampling strategy used with the tubes;
2. the analysis technique used and the incorporation of a pre-analysis nitrogen purge to remove trapped water;
3. the validation of the data collected with comments on the data quality; and
4. a preliminary report on the data collected to date at the monitoring site.
INTRODUCTION
In July 1991 the Wisconsin Department of Natural Resources, Bureau of Air Management (WDNR-BAM)
began the operation of a prototype Hazardous Air Pollutant (Air Toxics) monitoring station in Green Bay.
Monitoring at the Green Bay station was designed to be a screening program to determine the concentrations of
selected organic compounds (volatile and semivolatile) and inorganic compounds (metals) present in the urban
atmosphere. The paper will report on the monitoring conducted for volatile organic compounds which began at the
station in August 1991. VOC monitoring was conducted using multi-bed adsorbent tubes to collect the target
compounds in the ambient air. After a start-up problem caused by the moisture (water) collected on the sampling
tubes was resolved, the routine sampling began operation in November 1991. This paper will discuss the VOC
monitoring from November of 1991 to November 1992 and the present state of monitoring operations at the Green
Bay station.
BACKGROUND
The Clean Air Act Amendments of 1991, in Title III, called for the establishment (by the
U.S. Environmental Protection Agency) of a network of monitors in major urban areas to measure the concentrations
of air toxins due to the myriad of "area" sources in such cities, and to determine the risk of the exposed population.
Those sources included such categories as: automobile and truck exhausts; emissions emanating from retail
establishments such as dry cleaners and restaurants; household products such as sprays and application liquids,
insecticides, rodenticides and herbicides, etc. Some of the specific compounds identified included aromatics, PAHs,
POMs, aldehydes, metals, etc.
The WDNR has decided to establish an Urban Air Toxics network to supplement any network to be
established by the U.S. Environmental Protection Agency (USEPA) and to determine the risk to populations in cities
having the greatest potential of exposure as well as significant numbers of people exposed. In addition, concern
604
-------
about industrial fugitive emissions in some cities would not be addressed by the USEPA study. The areas in
Wisconsin which the WDNR selected for Air Toxics stations include Milwaukee, Superior, central Wisconsin, the Fox
River Valley (Oshkosh/Appleton) in addition to Green Bay.
The City of Green Bay was selected as the prototype location because of concern by area citizens, the mix of
urban area sources and industrial sources, and a specific request by WDNR local staff. Green Bay is an industrial
city located on the southern shore of Lake Michigan's Green Bay and at the mouth of the Fox River, which empties
into the bay. The city has a population of approximately 96,500. Industry in the city includes a number of paper
mills. Other large air pollution sources in the city include a power plant and a large wastewater treatment facility.
The monitoring station was originally located at the WDNR's Bay Beach air monitoring trailer, AIRS
No. 55-009-0023, established as a sulfur dioxide monitoring station. The Bay Beach site is located at the north end of
the city where predominate southerly winds carry the pollutants. Pollutants measured at the station are emitted by
both large and small industries, the power plant, the wastewater treatment plant as well as from automobile and truck
traffic.
The prototype monitoring station was intended to collect preliminary data and test sampling methods, with
additional monitoring stations to be established hi future years. The Green Bay monitoring station and future
stations hi the Wisconsin air toxics monitoring network are intended to collect data to provide information to be
analyzed for potential health risks as well as for ambient concentration trends.
MONITORING METHODOLOGY
Sampling
Sampling tubes. Trace organic contaminants in the ambient air, whether toxic or photochemically reactive,
can rarely be directly measured. Sampling and analysis methods must often include steps to concentrate compounds
of interest prior to the actual analysis. Compound concentration hi the field is accomplished by collecting the target
compounds on a filter, adsorbent tube, or other collection media. Alternatively whole air samples can be collected in
bags or canisters and then concentrated at the laboratory on sorbents or by cryogenic methods. Compound
concentration in the field on adsorbent tubes was selected as the primary method for the prototype air monitoring
station. Sampling was conducted using commercially available multi-bed adsorbent tubes tested with a wide range of
compounds including the TO1, TO2, and TO3 compounds1. The adsorbent beds contained graphitized carbon black
and carbon-molecular sieves. These adsorbents tubes trap compounds without regard to functional group using the
principle of size exclusion. In addition, the adsorbents have a hydrophobic nature that is expected to reduce loss to
water molecules. Advantages of this methodology include the following:
- direct exposure of the sampling media to the ambient air, without a sample line;
- the collection of greater air volumes for analysis (22 to 86 liters) than permitted by laboratory cryogenic
methods (usually < 1 liter); and
- the use of a mass spectrometric detection system for compound identification.
Prior to sampling, the adsorbent tubes are precleaned by heating to 330t and purging with an ultra-pure helium flow
of 9 cm /mm for 15 minutes.
The Wisconsin monitoring program does have experience with using canisters collection for photochemically
reactive ozone precursors. The present analytical system for canisters is dedicated to the analysis of ozone precursors
and lacks either multi-detector or mass spectrometer detection capabilities.
Sampler design. The sampler design is based on the TO1 design2, and the sampler used in the original
USEPA Toxic Air Monitoring Station (TAMS) network3. The sampler consists of 4 major parts, a sampling
manifold, flow controlling devices, an air mover, and an electronic timer. The sampling manifold has four
independent channels each with a mass flow controller to monitor and regulate the air flow. A common timer and
pump is used on the sample. All sampler parts are located inside an old high volume sampler housing. The four
teflon vacuum lines extend out from the housing to a hood which shields the sampling tubes. Sampling tubes are
inserted into the teflon lines and are directly exposed to the ambient air. A vacuum recorder provide a means of
monitoring the sampling flow.
Sampling protocol. Twenty-four hour VOC samples were collected on a 1 in every 13 day schedule using a
total of 5 sampling tubes. The primary samples (tubes 1 and 2) were simultaneously collected at flow rates of 15
605
-------
cm3/min and 60 cm3/min. A duplicate set of sample tubes (tubes 3 and 4) were collected to insure a sample should
an accidental loss occur to the first set. The fifth tube served as a field blank.
Unlike whole air sampling techniques (i.e canisters) adsorbent sampling techniques are subject to
complications such as background contamination, compound breakthrough, and artifact formation. Adsorbent
sampling techniques must therefore use sampling schemes that will provide a definition of the sample quality. Such
sampling schemes include the distributed volume technique (parallel sampling) and/or backup tubes (in series
sampling). The distributed volume technique has been reported as a practical technique that is the preferred method
of sampling for adsorbent tubes4. The sampling protocol adopted for this program requires adsorbent tube
distributed volume sampling using 2 sampling tubes with air flows rates in a ratio of 1:4. The ambient concentration
of each target compound collected is reported as the average of the two sample tubes with a Membership Value
(MV) for the distributed volumes. Walling5 has recommended the use of MVs determined by a one dimensional
Gaussian function and evaluated by Fuzzy Set theory. Fuzzy Set theory assumes there is a continuous range of values
reflecting the certainty in a measurement, rather than discrete values of certainty (good vs. not good). A MV of 1.0
indicates a perfect match between the two sampling tubes and a high degree of certainty in the result. As the MV
decreases greater complication between the sampling tube pair is indicated, as is increased uncertainty in the results.
For our data the Gaussian function is scaled to yield a MV of 0.92 for a 10% difference in the data pair. For the
data reported in the Results Section a 25% difference for parallel tubes was used as one acceptance criteria.
However this percentage can translate into different quantities at different levels. For example, at the 150 ppt a 25%
difference is 38 ppt while this same percent difference at 25 ppt is only 6 ppt.
Analysis
The analytical system consists of a Envirochem Model 840 thermal desorption unit integrated with a Hewlett
Packard 5995A gas chromatograph/mass spectrometer. Exposed sample tubes are thermally desorbed by ballistically
heating from an ambient temperature to 270t in 5 seconds; the tubes are held at 270t while a 9 cm3/min helium
sweep gas carries volatile compounds to a 2mm I.D. focusing tube. Focusing on the narrow I.D. minimizes dead
volume and concentrates the columns needed into narrow bands, improving chromatography and eliminating the need
for cryofocusing. The focus tube is spiked with the cc,a,a-trifluorotoluene internal standard used for quantitative
analysis. The focusing tube is then thermally desorbed in the same manner onto the GC analytical column, a
Suplecowax™ 10 glass capillary gas chromatographic column (60 meters x 0.75mm ID x 1.0 |im DF). The sample is
then subjected to a GC/MS analysis using a mass scanning mode ( 29 to 175 AMU).
The analytical method uses a 5 point standard calibration curve. The standards are prepared in a static
bottle from a liquid stock solution of the target compounds. A headspace sample from the static bottle is then spiked
onto a clean 4 mm Carbotrap™ 300 tube. Each analytical standard is desorbed and analyzed in the same manner as
are the ambient samples. Daily operation of the analytical system includes the following: the GC/MS is autotuned
using perfluorotributylamine; a blank with internal standards is run; a check standard is analyzed to verify calibration;
a blind quality control sample is analyzed; and ambient samples are then analyzed.
During analysis of the first few sample sets, the analyst observed that some samples had components whose
retention times were shifted during GC/MS analysis. Additionally, it was observed that the samples during the
focusing step showed a white band which disappeared upon cooling. Water, picked up from the humid ambient air,
was the suspected source of the chromatographic changes. Standards were prepared, spiked with various amounts of
water and analyzed. Analysis of the spiked samples confirmed that the water did cause the retention time shift.
Losses of xylene, cumene, n-butanol, and limonene were seen as well. A pre-analysis nitrogen purge was added to
remove water from the ambient samples. Experiments were conducted to determine the best gas flow rate and total
drying time required. We elected to dry with a nitrogen purge at 20 to 30 cm3/min for 4 hours.
RESULTS AND DISCUSSION
Data Quality
Field blanks. Each complete sample consists of two sampling tubes and a blank. Because the sampling
tubes aggressively scavenge compounds from the air, a field blank is needed to indicate possible contamination. As
an example of scavenging ability, an August 1991 sample blank was contaminated when the transport tube broke in
transit. Upon analysis the tube showed heavy concentration of all compounds ranging from > 1000 ng for hexane to
100 ng for benzene. Examination of the blanks showed that acetone, hexane (used at the site), toluene and xylene
606
-------
were the most commonly found contaminants in the field blanks. In calculating all results, the blank was subtracted
prior to calculation of the ambient concentration.
Replicate samples. In theory, the use of the distributed volume sampling method, will require that every
sampling set contains duplicate samples. In addition to distributed volumes, replicate samples with identical flow
rates were periodically collected for analysis. These duplicates did show relatives average differences ranging from
0% to 100%. A 100% difference denoted sample pairs where a compound was detected in only one tube. The
highest variation was found in isopropanol and the lowest in benzene.
Audit samples. The USEPA supplied the Wisconsin program with two audit samples. These audit samples
consisted of VOC sampling tubes that were spiked with known concentrations of selected compounds. Analysis of
these samples showed the expected loss of the compounds from the more volatile compounds. Preliminary data from
laboratory standards and USEPA audits does however suggest poor recovery of the most volatile compounds (1,3-
butadiene and acetone). Suggesting that the preanalysis nitrogen purge causes loss of these compounds. The
remaining compounds showed better result. Six of 16 compound data points did exceed the 25 % accuracy goal in
the draft Quality Assurance Project Plan.
Monitoring Results
The compiled results of the VOC monitoring from November 1991 to November 1992 are summarized in
Table 1. Analytical results from the two sample tubes, with distributed volumes, were averaged and the data from
the 28 sample sets were evaluated against the following 3 criteria:
- was the compound detected in the sample;
- was the compound detected at concentrations greater than the minimum quantitation limit (MQL); and
- was the MV for the tube pair greater than 0.673 (equivalent to a percent difference of 25%).
Each one of 28 sample sets was analyzed for 20 compounds resulting in a possible 560 data points. The results are
broken down in the following manner:
- 41.6 percent of the time compounds were not detected;
- 22.9 percent of the time detected compounds were less than MQL;
- 223 percent of the time the MV indicated a unacceptable difference in results from the sample pair; and
-13.2 percent of the results were detectable and quantifiable.
Table 1 reports the number of detects and the range of the quantifiable results. There was insufficient data to
calculate an accurate annual ambient mean for most of the compounds monitored. The VOC monitoring data is
consistent with other studies' '7/fhat have found benzene, toluene, and xylene to be among the most commonly found
compounds. Perchloroethylene was the most commonly detected chlorinated compound with detects in 25 of 28
samples. However, only 7 samples had data that meet acceptance criteria for quantitation. Concentrations at the
Bay Beach site have been generally lower than the mean values reported for National Urban Air Toxics Monitoring
Program (UATMP). Comparison of the 1989 and 1990 median concentrations for the UATMP with median
concentration reported in Table 1 show the following: the benzene concentration was 24-33% of the median UATMP
concentrations; the toluene concentration was 21-27% of the median UATMP concentrations; the xylene
concentration was 18-35% of the median UATMP concentrations; and the perchloroethylene concentration was 31-
48% of the median UATMP concentrations.
Careful examination of the remaining monitoring data can provide clues about the ambient concentrations of
other pollutants. Dichlorobenzene is a compound found in extremely low ambient concentrations and the monitoring
results of this compound offer a good example of the additional ambient information monitoring provides. Twelve of
twenty eight samples for dichlorobenzene were reported out as no detects, while 12 were listed as below MQL. In
the latter 12 cases only one of the tube pair was below the MQL. However, because 1 of the pair was below the
MQL, the MV test (the next criteria) could not be applied. If the MV test were applied 3 of the 12 pairs would
showed excellent agreement (MV 0.899 or better) and would represent concentrations of 0.016, 0.015, and 0.029 ppb.
The remaining four BAD PAIR cases reported for dichlorobenzene were monitored at low concentration and despite
percent differences greater than 25% for 2 samples (0.049 ppb and 0.032 ppb) this amounted to only 13 ppb in the
pan- results. Overall we conclude dichlorobenzene was present in the ambient air and in concentrations of less than
50 ppt. Efforts to improve detection of the dichlorobenzene should focus on collection of greater air volumes.
The Bay Beach Monitoring location was chosen because of the predominant southerly wind in the Green Bay
area. Daily vector means of the wind direction data, collected at the site, have shown that the wind direction on 82%
607
-------
of the sampling days falls between 90° and 270a, on 50% of the sampling days the winds were from a sector of 180° to
270°. All VOC sources are to the south of the site and as expected, the highest concentrations of compounds were
also found in winds from the southwest quadrant (see Figure 1). Among the original plans for the data was the use
of multi-variant statistical techniques (principle component analysis and factor analysis) to determine the source of
measured compounds. However, with a large percentage of the data being none detected or below the MQL these
statistical techniques can not be used with confidence. At this time, no attempt has been made to correlate any of
the monitoring data with specific sources in the area or with the WDNR's developing Air Toxics Emissions Inventory.
CONCLUSIONS
Monitoring at the Bay Beach site has met the goals of collecting preliminary data on concentrations of air
toxics in the urban environment. Monitoring has provided concentration estimates for target compounds such as
benzene, toluene, and xylene. In addition, monitoring has provided information on the level of sensitivity that will be
needed to monitor target compounds such as dichlorobenzene. Finally we see that the preanalysis nitrogen purge,
required to dry the adsorbent sampling tubes causes loss of the most volatile compounds, such as 1,3 butadiene. An
alternative method, such as whole air sampling, will be required to monitor this and other highly volatile compounds.
The large number of non detects may be due to the relative remoteness of the Bay Beach site from the
central business district with its many area sources. The demonstration of successful urban monitoring at the Bay
Beach station has facilitated a site move to a new station site located within the city's central business district. While
we will continue to use the adsorbent tube for sampling at the new station, we are in the planning stages of a change
to a whole air sampling scheme using canisters. We believe that whole air canister sampling will be a better choice
for an urban air toxics network for the following reasons:
1. canister sampling simplifies the sample collection and sampling handling;
2. canister sampling will transfers many sample quality concerns from the field to the laboratory; and
3. the sampling method will be consistent with National Urban Air Toxic Monitoring Programs.
REFERENCES
1. W.R. Betz, G.D. Wachob, M.C. Firth, "Monitoring a wide range of airborne organic contaminants",
Proceeding of the 1987 EPA/APCA Symposium on Measurement of Toxic and Related Air Pollutants. VIP-
8 Air Pollution Control Agency, RTF, pp.761-770.
2. W.T. Winberry, Jr.,N.T. Murphy, and R.M. Riggin, Compendium of Methods for the Determination of Toxic
Organic Compounds in Ambient Air. EPA/600/4-S4/041, USEPA, Research Triangle Park ,1984.
3. T.A. Lumpkin and A.E. Bond, Standard Operating Procedure for the Toxic Air Sampler Used in the Toxic
Air Monitoring System (TAMS). EMSL/RTP/SOP-EMD-204, USEPA, Research Triangle Park, 1984, P2-5.
4. J.F. Walling, "The utility of distributed air volume sets when sampling ambient air using solid adsorbents",
Atmos. Envir. 18(4):855-859, (1984).
5. J.F. Walling, "Membership values as indicators of complications in chromatography", J.Chrom 473:267-272
(1989).
6. RA. McAllister, W.H. Moore, J. Rice, D.P. Dayton, R.F. Jongleux, P.L. O'Hara, R.G. Merrill, Jr., and J.
Bursey, 1988 Nonmethane Organic Compound and Urban Air Toxics Monitoring Programs. Final Report.
Urban Air Toxics Monitoring Program, Volume II. EPA 68D80014, Radian Corporation, Research Triangle
Park, 1989, p.11-6.
7. R.A. McAllister, W.H. Moore, J. Rice, E. Bowles, D.P. Dayton, R.F. Jongleux, .G. Merrill,Jr., and J.T.
Bursey, 1989 Urban Air Toxics Monitoring Programs. EPA/450/4-91/001, USEPA,1990, p.1-4.
8. R.A. McAllister, E. Bowles,J. DeGarmo, J. Rice, R.F. Jongleux, R.G. Merrill,Jr., and J. Bursey, 1990 Urban
Air Toxics Monitoring Programs. EPA/450/4-91-0024, USEPA, Research Triangle Park ,1991, p.1-4.
Carbotrap and Suplecowax are registered trademarks of Supleco, Inc., Supleco Park, Bellefonte, PA, 16823-
0048.
608
-------
TABLE 1: Summary statistics for VOC concentrations (ppb) monitored in Green Bay
NAME MEAN MEDIAN MIN
1,3 -Butadiene
2-butanone(MEK) 0.111 0.111
acetone 0.349 0.349
alpha-pinene
benzene 0.348 0.332 0.080
chloroform
cumene 0.020 0.018 0.018
ethanol
isopropanol 0.184 0.184
limonene
methyl chloroform 0.190 0.190 0.176
methylene chloride 0.127 0.127
n-butanol 0.213 0.213 0.149
n-hexane 0.784 0.771 0.353
p-dichlorobenzene
stryrene
perchloroethylene 0.096 0.096 0.038
toluene 0.628 0.519 0.035
trichloroethylene 0.025 0.024 0.018
xylene 0.522 0.351 0.021
Totals
Concentrat
2 o-
I"1 5-
§
Si.o-
\ ' *
§ '
' . • A
" * A¥
, • • 4 • * •
• • *
wSStDonj&Rfvnfinr-S
24 hour vector mean wind
MAX
0.111
0.349
1.145
0.024
0.184
0.204
0.127
0.277
1.393
0.197
1.829
0.033
1.736
i on vs .
*
• * *A
• 1
" " \
t*».
OT en o <- w r
direct Ion
DETECT NON-
DETECT
0 28
1 16
1 8
0 19
18 0
0 19
3 12
0 18
1 17
0 20
2 0
1 13
2 13
5 2
0 12
0 15
7 3
14 0
3 18
16 0
74 233
13.2% 41.6%
VMWD
A
*
m
A *
A
A * • * *
*X
•
uwtMOimini/im
Cfleoreesj
-------
COMPARISON OF NMOC DATA COLLECTED BY TWO METHODS IN ATLANTA
Jack H. Shreffler
Atmospheric Research and Exposure
Assessment Laboratory, MD-75
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The Clean Air Act Amendments of 1990, Title I, call for
"enhanced monitoring" of ozone, which is planned to include
measurements of atmospheric non-methane organic compounds (NMOC's).
NMOC concentration data gathered by two methods in Atlanta, Georgia
during July and August 1990 are compared in order to assess the
reliability of such measurements in an operational setting. During
that period, automated gas chromatography (GC) systems (FIELD
systems) were used to collect NMOC continuously as one-hour
averages. In addition, canister samples of ambient air were
collected on an intermittent schedule for quality control purposes
and analyzed by laboratory GC (the LAB system) . Data from the six-
site network included concentrations of nitrogen oxides (NO^),
carbon monoxide (CO), ozone, total NMOC (TNMOC), and 47 identified
NMOC's. Regression analysis indicates that the average TNMOC
concentration from the LAB system is about 50 percent higher than
that from the FIELD system, and that the bulk of the difference is
due to unidentified NMOC's recorded by the LAB system. Also, there
are substantial uncertainties in predicting a single FIELD TNMOC
concentration from a measured LAB concentration. For individual
identified NMOC's, agreement is poor for many olefins that occur at
low concentrations but may be photochemically important.
Regressions of TNMOC against CO and N0x lead to the conclusion that
the larger unidentified component being reported by the LAB system
is not closely related to local combustion or automotive sources.
DISCLAIMER
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or
recommendation for use.
610
-------
INTRODUCTION
In July and August of 1990, the Atmospheric Research and
Exposure Assessment Laboratory (AREAL) conducted a field study in
Atlanta, Georgia to obtain data on atmospheric pollutants related
to ozone formation in the urban area. Six sites were established
in the Atlanta area where nonmethane organic compounds (NMOC's)
were measured continuously by an automated gas chromatograph on an
hourly basis1, as were meteorological data and other pollutants,
e.g. carbon monoxide, ozone, nitrogen oxides. In addition to the
continuous data, air samples were collected in stainless steel
canisters and analyzed later for NMOC's by a laboratory gas
chromatograph.2 This paper focuses on a comparison between the
NMOC measurements taken by the two methods.
The Atlanta Ozone Precursor Study was conceived, in part, as
a prototypical demonstration for "enhanced monitoring" related to
ozone as called for in the Clean Air Act Amendments of 1990. A
complete description of the measurements taken during the study can
be found in the Data Report.3 A second report4 supplements the Data
Report by giving some alternative perspectives and more in-depth
presentations of the data from the continuously operated sites,
including diurnal curves of all major pollutants.
NMOC MEASUREMENTS
An automated gas chromatographic (GC) system (Chrompack,
Inc.), hereinafter designated the FIELD system, was used to measure
NMOC's each hour and at each of the six sites. A sample was
collected in an absorbent trap for 30 minutes and then analyzed in
the following 30 minutes. The FIELD systems were in place from
July 1 through August 31, 1990, giving 1488 possible hourly
measurements. The final FIELD archive holds, for the various
sites, NMOC concentration values for 50-70 percent of the possible
hours.
Air samples were collected in stainless steel canisters at
Sites 1-6 on a schedule of approximately every other day and at
times that rotated through the diurnal cycle. The canisters were
opened for the first 30 minutes of the hour and so collected a
sample comparable to that taken by the collocated FIELD system.
The canisters were returned to AREAL at Research Triangle Park, NC,
within a few days and NMOC's were measured using a laboratory HP
5890 GC. A total of 174 samples were collected, analyzed, and
archived. Hereinafter, this collection and analysis method will be
referred to as the LAB system.
The archive of the FIELD data contains 47 identified NMOC's as well
as TNMOC, the sum of all identified and unidentified NMOC's (ppbc).
The archive of the LAB data contains comparable NMOC's.
611
-------
COMPARISON OF FIELD AND LAB MEASUREMENTS
There are a total of 174 hourly LAB measurements available.
However, the FIELD system reported measurements at only 85 of these
times and sites. Because the FIELD system, at some sites and
times, reported erroneously high values of isopentane, and because
of evidence that the problem extended to other NMOC's, 21
measurements had to be deleted. In addition, there were several
"outliers." After these deletions, 58 measurements remained to be
compared from the two systems.
Figure 1 gives the regression line (FIELD vs LAB TNMOC) for
the remaining measurements and includes the 90 percent prediction
interval. The slope of the regression line is 0.64 with a standard
error of 0.05. The prediction interval gives a sense of the ability
to predict an individual FIELD measurement from a LAB measurement.
If, for example, there is a single LAB measurement of 400 ppbc
TNMOC, the FIELD system would be expected to return a value between
130 and 380 ppbC nine out of ten times. Obviously the disagreement
between single measurements is considerable.
Simple linear regressions, similar to that shown in Figure 1,
were repeated for the sums of all identified NMOC's and the sums of
all unidentified NMOC's. The regression lines and prediction
intervals are given in Figures 2 and 3, respectively. For the
identified NMOC's, the estimated slope is 0.896 with a standard
error of 0.04 and a non-significant intercept of 2.5 ppbC, showing
that the LAB system tends to report a value only about 11 percent
higher than the FIELD system. For the unidentified NMOC's, the
slope is 0.16 with a standard error of 0.05 and a non-significant
intercept of 32 ppbC. Clearly the major problem lies with the
unidentified NMOC's.
Simple linear regressions were also performed for each of the
47 identified NMOC's, and the results for slope and R2 indicate
that many individual NMOC's do not compare well. Of the 12 NMOC's
having R2 below 0.50, all but two have average concentrations much
less than 1 percent of TNMOC, so the poor explanation of
variability may be due to insufficient instrument sensitivity.
Mobile sources are a principal contributor to NMOC in Atlanta
and other U.S. cities, and therefore one would expect a relation
between TNMOC and other combustion related gases, carbon monoxide
(CO) and nitrogen oxides (NO ) . Figure 4 gives a linear regression
and 90% prediction interval for TNMOC from both systems against CO.
Regressions with similar features are found for TNMOC vs NO . The
Figures reveal a much greater unexplained variability in the LAB
TMNOC as measured by the root mean square error (rmse). Similar
regressions were computed using the sums of identified and
unidentified NMOC's. Considering the identified components,
regressions for the two systems were quite close in all aspects.
The differences in slope, intercept, and rmse seen in Figure 4
612
-------
arise almost entirely from the unidentified components of the two
systems. One concludes that the (larger) unidentified component
that is reported by the LAB system is not closely related to local
automotive or high-temperature combustion sources.
CONCLUDING REMARKS
The 1990 Atlanta Study provides an interesting glimpse at the
problems that may be encountered in implementing routine
measurements of NMOC's in urban areas, where somewhat different
instruments and procedures may be used. The Atlanta study was not
designed as an instrument comparison, with rigorous controls and
side-by-side processing, and in that respect probably more
realistically simulates the type of outcome that may occur in
operational networks. On the other hand, the Atlanta data were
taken over only a two month period, and start-up problems with
emerging auto-GC technology could be addressed in a longer time
period.
There are large differences between the NMOC concentrations
reported by the two GC systems in the Atlanta study, even given
rather careful data editing. The results demonstrate that reliable
NMOC measurements are difficult to obtain and show the necessity
for a strong quality control component in the "enhanced monitoring"
program. This paper presents statistical and graphical techniques
that will be useful in future quality control analyses.
REFERENCES
1. Holdren, M.W.; Smith, D.L. "Performance of Automated Gas
Chromatographs Used in the 1990 Atlanta Ozone Study," Proceedings
of the 1991 U.S. EPA/A&WMA International Symposium, Measurement of
Toxic and Related Air Pollutants, 1:26 (1991).
2. Lonneman, W.A.; Seila, R.L.; Daughtridge, J.V.; Richter, H.G.
"Results from the Canister Sampling Program Conducted During the
1990 Atlanta Precursor Study," Proceedings of the 84th Air and
Waste Management Association Annual Meeting, Vancouver, B.C.
Canada, Paper No. 91-68.2 (1991).
3. Purdue, L.J.; Reagan, J. ; Lonneman, W. ; Lawless, T. ; Drago, R. ;
Zalaquet, G. ; Holdren, M. ; Smith, D. ; Spicer, C. ; Pate, A.; Buxton,
B.; "Atlanta Ozone Precursor Monitoring Study Data Report," EPA-
600/R-92-157, U.S. Environmental Protection Agency, Research
Triangle Park, NC, September 1992.
4. Shreffler, J.H. "A Survey of Data from the Continuous Sites of the
1990 Atlanta Ozone Precursor Study," EPA-600/R-92/172, U.S.
Environmental Protection Agency, Research Triangle Park, NC,
September 1992.
613
-------
800-
600-
9
400H
!!! 200-
0-
T
T
• I • • • • I • • '
200 400 600 800
LAB TNMOC, ppbC
1000
Fig 1. Linear regression of FIELD TNMOC against LAB TNMOC,
including a 90 percent prediction interval.
600-
U
O 400-
O
Q
UJ
0-
600-
I ^^ 'I I
0 200 400 600
LAB ID'ed NMOC, ppbC
Fig 2. Linear regression of FIELD
identified NMOC against
LAB identified NMOC,
including 90 percent
prediction interval.
a.
O
o
400-
g
T
200-
0-
I I I I
0 200 400 600
LAB NON-ID'ed NMOC, ppbC
Fig 3. Linear regression of FIELD
unidentified NMOC against
LAB unidentified NMOC,
including 90 percent
prediction interval.
614
-------
800-
600-
O
.a
-400H
u
o
200-
0-
1000-1
800-
600-
400-
200-
0-
I I
III ^^ | , , . . | .
0.0 0.5 1.0 1.5 2.0
CARBON MONOXIDE, ppm
2.5
Figure 4. Linear regressions of FIELD and LAB TNMOC's against
carbon monoxide, including a 90 percent prediction
interval.
615
-------
VOCs in Mexico City Ambient Air
Robert L. Seila and William A. Lonneman
U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Mail Drop 84
Research Triangle Park, NC 27711
Maria Esther Ruiz Santoyo and Javier Tejeda Ruiz
Instituto Mexicano del Petroleo
Eje Central Lazaro Cardenas No. 152
07730 Mexico, D.F.
Mexico
ABSTRACT
Mexico City, with nearly 20 million people, 3 million vehicles, and 35,000 industrial
businesses, has severe photochemical air pollution.1 The Mexican O3 standard of 0.11 ppm is
exceeded over 300 days of the year. Because of the role of VOCs in the production of O3, a study
of the concentration and composition of VOCs in ambient air was undertaken. From March 6 to 26,
1992, 68 ambient air samples were collected on week-days in passivated stainless steel spheres at
four sites in Mexico City. Most (52) sampling was in the morning from 6:00-9:00 am, however at
2 sites 18, 12:00-3:00 pm afternoon samples were collected. Ten morning samples were also taken
at two background sites north of the city. The samples were analyzed by capillary gas
chromatography-flame ionization detection (gc-fid) for C-2 to C-14 hydrocarbons. CO and CH4
were also determined. Total non-methane organic compound (TNMOC) concentrations were very
high. In the 6-9 am period TNMOC ranged from 1.49 to 6.94 ppm. The afternoon results were
lower, ranging from 0.48 to 3.06 ppm. TNMOC at the two background sites ranged from 0.16 to
1.64 ppm.
INTRODUCTION
Mexico City is within a mountain basin at an elevation of 2234 m, surrounded by peaks
3500 m average height with some peaks reaching over 5000 m. Almost 20 million people, close to 3
million vehicles and a significant portion of the nation's industrial base lie within the Mexico City
Metropolitan Area (MCMA) basin. Because of its rapid growth, 35,000 industries of all sizes and
types are spread over the city with slightly higher concentration in the north and east. The
subtropical climate combines with the geographic characteristics to intensify and concentrate air
pollutants from industrial, transportation, and routine daily activities. The basin is relatively small,
120 x 150 km, however population activities are concentrated in an even smaller area, 2346 kmz, at
the middle of the basin. High solar radiation at the high altitude and strong inversions provide
conditions for the production of photochemical pollution. The Mexican O3 standard, of 0.110 ppm,
was exceeded more than 300 days in 1991.'
A 25 station network monitors O3, SO2, NO2, NOx, CO and meteorological parameters
throughout the MCMA. However, hydrocarbons are measured at only 3 stations by means of
analyzers with photoionization detectors that have highly variable individual hydrocarbon responses.
616
-------
quantitation individual species not detailed enough to be used for the evaluation of hydrocarbon
impact on O3 formation. A field study was conducted in March 1992 to determine C-2 to C-14
ambient total and speciated hydrocarbon concentrations at 4 sites within and 2 sites upwind of the
MCMA. CO and CH4 measurements were also made. This paper presents the results of that field
study.
EXPERIMENTAL METHODS
Six sites were selected for sampling. Three sites were monitoring stations at Xalostoc,
Merced, and Pedregal which lie on a northeast to southwest line through the center of the city. This
line corresponds to the most frequent wind trajectory. Xalostoc is an industrial area, Merced is a
commercial area characterized by heavy traffic, and Pedregal is residential area where the city's
highest O3 concentrations are normally recorded. The station at Tlalnepantla, an
industrial/commercial area to the northeast, was also selected, because the highest O3 readings are
sometimes observed there. All of these 4 sites contained a full compliment of pollutant monitors.
The two upwind sites were at Cuautitlan which is a university campus due north of the city and east
of a major north/south highway and La Reforma which is northwest of Cuautitlan near the small
town of La Reforma.
Ambient air samples were collected on weekdays from March 6 to 26, 1992.
Microprocessor controlled samplers pumped air into passivated stainless steel spheres. Fifty-two
samples were collected from 06:00 to 09:00 at the Xalostoc, Tlalnepantla, Merced, and Pedregal
monitoring stations. At Tlalnepantla and Pedregal 18 midday (12:00-15:00) samples were also
taken. Two upwind sites north of the city, Cuautitlan and La Reforma, were used to collect 5,
06:00-09:00 samples each.
The samples were analyzed at EPA, Research Triangle Park by gas chromatography with
flame ionization detection (GC-FID) using 3 different columns. C-2 to C-14 hydrocarbons were
separated on a 60-m, 0.32 mm i.d. fused silica capillary column with a 1 /j.m thick film of a non-
polar cross-linked liquid phase (DB-1, J&W Scientific, Rancho Cordova, CA). Ethylene, acetylene,
and ethane were separated on a 30-m, 0.53 mm i.d. GS-Q porous polymer open-tubular fused silica
column (J&W Scientific), because they were not consistently resolved on the 60-m non-polar
column. The C-2 to C-14 hydrocarbons were cryogenically preconcentrated in a 11" x 1/8" o.d. U-
shaped trap filled with 60-80 mesh glass beads cooled by liquid argon in a Dewar flask. Injection
was accomplished by switching a 6-port gas sampling valve and heating the trap with near boiling
water. In the case of the C-2 hydrocarbons the trap was filled with Carbowax 20 M on 60-80 mesh
Chromosorb W-AW and heated by room temperature water. The details of these analyses are
published elsewhere.2 CH4 and CO were resolved on a 2-m x 1/8 in od ss column containing 40-
60 mesh molecular sieve. CO was catalytically reduced to CH4 by H2 on a nickel catalyst at 350'C
for detection by FID.
RESULTS AND DISCUSSION
Figure 1 is a box and whisker plot showing the total non-methane organic compound
(TNMOC) results for each site and time period of the study. TNMOC was calculated by summing
all of the individual peak concentrations from the DB1 column except for ethane, ethylene, and
acetylene whose concentrations came from the GS-Q analysis. The center vertical line of a box is
the median. The hinges represent the 25th and 75th percentile values. The inner and outer fences
617
-------
are defined as +1.5 and ±3.0 times interquartile range from each hinge. The whiskers show the
range of values which fall between the inner fences. Values outside the inner fences are plotted
with asterisks, and values outside the outer fences are plotted with empty circles.
The highest median (4.44 ppm) corresponds to Xalostoc which is mainly an industrial area,
although we do not mean to imply an absence or even much diminution of automobile traffic,
because there is significant automobile traffic everywhere in Mexico City. Additionally, at this
station the concentration dispersion is the largest, ranging from 1.93 to 6.94 ppm. The downtown
station, Merced, which is in an area influenced by combined commercial and vehicular activities,
recorded the second highest median (3.50 ppm) with correspondingly lower dispersion, 4.29 ppm.
The urban site of lowest morning median TNMOC reading was Pedregal (1.98 ppm) which is in
primarily a residential area.
The difference in concentration between morning and midday samples is higher at the
northwestern site of Tlalnepantla (1.785 ppm or 55% reduction from morning samples) than at
Pedregal. The lesser midday reduction, 0.443 ppm (22%), at the SW station may be attributed to
the prevailing atmospheric transport, NE to SW, from the central urban area in addition to local
emissions. This fact in addition to Pedregal being the site showing the highest frequency of
maximum ozone readings point to it at an important receptor site.
A comparison of the Mexico City TNMOC results to levels in US cities is also shown in
Figure 1. The US statistics are from a continuing study of 6 to 9 am TNMOC and NOx
concentrations at urban sites in cities that have not met the ambient air quality standard for 03.
Integrated ambient air samples were collected during the late spring and summer months in
passivated 6-1 stainless steel spheres and air freighted to a central laboratory for analysis by the
cryogenic preconcentration direct flame ionization (PDFID) method. Over 10,000 samples from 68
cities were analyzed over a 6 year period. The US boxes in Figure 1 were constructed from the
statistical results rather than from all of the individual measurements.3-4>5'6'7 Therefore values
less than the inner or greater than the outer fences are not shown with the exception of the
maximum values. Although US TNMOC maximums are very high, it is clear that TNMOC in
Mexico City is significantly higher than TNMOC in the US: in terms of medians from 2 to 8 times
higher. This holds true even at the two midday sampling sites as well as the two morning upwind
sites.
The two upwind sites' TNMOC was much lower than that of the urban sites although well
above U.S. urban 6-9 am TNMOC. These high measurements at the outlying sites are at least in
part due to local emissions from nearby sources. The Cuautitlan site was about 200 m from a major
highway, although the highway was not upwind of the prevailing wind direction. The La Reforma
site was near a small town. These measurements are probably an over estimate of the true boundary
conditions.
The most abundant hydrocarbon species are listed in Table 1. The data in this table are
morning results from the 4 MCMA sites: Tlalnepantla, Xalostoc, Merced, and Pedregal. The
mean concentrations are averages of the morning site medians, thus they are not influenced by the
number of samples taken at each site. Each site has equal influence. The maximum concentration
represents the overall maximum for all 6 sites, because no maximums were observed in the
afternoon samples or at the boundary sites. All maximums were observed in the morning at
Xalostoc with the exception of isopentane, benzene, and MTBE (methyl-t-butylether) at Merced.
Compound ratios to TNMOC expressed as percentages are shown in the "composition" column of
618
-------
the table.
The table consists of 7 paraffins, 3 aromatics, an olefin, acetylene, and MTBE. These 13
compounds account for a little more than 50% of the TNMOC. The most abundant hydrocarbon
was propane which is used as a household cooking and commercial fuel. Propane maximum
concentration and highest median, 617 ppbC, were observed at Xalostoc. Although Pedregal had
the lowest propane morning median concentration, 313 ppbC, it nevertheless had the highest
morning mean propane composition, 17.7%. The other compounds in Table 1 are associated with
vehicular related sources including exhaust and evaporation of gasoline. The gasoline additive,
MTBE, was observed in all samples except two at La Reforma. The maximum concentration, 122
ppbC and highest median, 65 ppbC, were recorded at Merced, the site in the area of highest traffic
flow.
CONCLUSIONS
Mexico City TNMOC was very high, on average from 3 to 8 times that of U.S. cities for the
6 to 9 am time period. In addition TNMOC at the background sites was also high compared to
U.S. cities. Propane was the most abundant hydrocarbon followed by paraffins and aromatics
associated with automobile exhaust and gasoline evaporation. MTBE was prominent: it was present
in all but two background samples and was the twelvth most abundant compound in the MCMA
morning samples.
REFERENCES
1. Proqrama Integral Contra la Contaminacion Atmosf erica de la Zona Metropolitana
de la Ciudad de Mexico, Departamento del Distrito Federal, Mexico City, October
1990.
2. Seila, R.L., W.A. Lonneman and S.A. Meeks, Determination of C2 to C12 ambient
air hydrocarbons in 39 U.S. Cities from 1984 through 19S6. EPA/600/3-89-058,
Office of Research and Development, U.S. Environmental Protection Agency,
Research Triangle Park, 1989, p 2.
3. McAllister R.A. (1986) "1984 and 1985 nonmethane organic compound sampling and
analysis program,"in Proceedings of the 1986 EPA/APCA Symposium on Measurement
of Toxic Air Pollutants, VIP-7 Air and Waste Management Association, Pittsburgh,
1986, pp 442-457.
4. McAllister R.A. , Jongleux R.F., Dayton D.,O'Hara P.L. and Wagoner D.E. 1986
Nonmethane organic compound monitoring. Final report for EPA contract number 68-
02-3889, Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, 1987, p 5-2.
5. Radian Corporation, 1987 Nonmethane organic compound and air toxics monitoring
program final report, vol 1--hydrocarbons, EPA/450/4-88-011, U.S. Environmental
Protection Agency, Research Triangle Park, 1988, p 5-2.
6. R.A. McAllister, P.L. O'Hara, W.H. Moore, D. Dayton, J. Rice, R.F. Jongleux,
R.G. Merrill and J.T, Bursey, 1988 Nonmethane organic compound and air toxics
monitoring program final report, vol I: Nonmethane organic compounds, EPA/450/4-
89-003, U.S. Environmental Protection Agency, Research Triangle Park, 1988, p 5-
2.
7. R.A. McAllister, P.L. O'Hara, W.H. Moore, D. Dayton, J. Rice, R.F. Jongleux,
R.G. Merrill and J.T. Bursey, 1989 Nonmethane organic compound and three-hour air
toxics monitoring program. EPA/450/4-90-011, U.S. Environmental Protection
Agency, Research Triangle Park, 1990 p 5-2.
619
-------
Table 1. Most abundant hydrocarbons in Mexico City ambient air.
Compound
Propane
n-Butane
Toluene
i-Pentane
i-Butane
n-Pentane
Acetylene
Ethylene
m & p-Xylene
2-Methylpentane
n-Hexane
MTBE
Benzene
Mean
Cone. ppbC
463
286
170
140
133
128
111
94
77
69
66
47
45
Maximum
Cone. ppbC
852
613
446
309
281
304
263
226
202
170
198
122
100
Composition
% of TNMOC
12.9%
8.6%
5.2%
3.8%
4.0%
3.8%
2.5%
2.9%
2.3%
2.3%
1.8%
1.3%
1.4%
DISCLAIMER
This paper has been reviewed in accordance with the U.S. Environmental Protection
Agency's peer and administrative review policies and approved for presentation and publication.
Mention of trade names or commercial products does not constitute endorsement or recommendation
for use.
620
-------
,ocation
-i
REFORMA
CUAUTITLAN
TLALNEpm
PEDREGALpm
PEDREGALam
TLALNEam
MERCED
XALOSTOC
NMOC89
NMOC88
NMOC87
NMOC86
NMOC8485
(
i i i i i i i
- a—
-Q *
- H 1 1 •
O — •
• 1 1 1 — •
— • 1 1 1 ' *
* ^_r~| h *
-rn o
-m o
>-rn o
— n~i o
— rn o
) 1 2 3 4 5 6 7
TNMOC. ppm
Figure 1. Mexico City TNMOC
_
-
-
-
-
-
-
-
8
621
-------
MONITORING STUDIES OF AMBIENT LEVEL VOCS and CARBONYLs
in PHOENIX, ARIZONA
Carmo Fernandas
James L. Guyton
Cheng Peter Lee
Arizona Department of Environmental Quality
Phoenix, Arizona
Sucha S. Parmar
ENSR Consulting and Engineering
Camarillo, California
1. INTRODUCTION and OBJECTIVES
The Phoenix metropolitan area has been classified by the U.S.
Environmental Protection Agency (EPA) as a "moderate" nonattainment
area for Ozone (O3) . The 1990 Clean Air Act Amendments require
states to submit complete O3 State Implementation Plans (SIP's) by
November 1993. With these objectives in mind, ADEQ prepared a
document entitled Workplan For Development of The Phoenix Ozone
SIP, March 1992. This workplan called for a modeling analysis to
define the relationships between emissions of precursors and the
formation of ozone. The most appropriate model for this purpose is
the Urban Airshed Model (UAM) which requires an extensive air
quality, meteorological and emission database. Part of this
required database is available from air quality and meteorological
networks routinely operated by various governmental and industrial
organizations. However, certain elements of the required database
are not routinely monitored and they include:
• Upper level meteorological and air quality data
• VpC, Carbonyl and NOX air quality data
• Background air quality data
Consequently, ADEQ found it necessary to conduct a monitoring study
to collect these unavailable data elements. Thus, the basic study
objectives can be summarized as follows:
• Collect meteorological and air quality data required to
support a UAM analysis for the Phoenix airshed.
• Collect quality monitoring data for VOCs and carbonyls to be
used in the Phoenix O3 modeling analysis.
2. SAMPLING NETWORK
The VOC/Carbonyl sampling network consisted of a total of six
monitoring sites summarized in Table 1. Three sampling periods, 6
am to 8 am, 10 am to 12 pm, and 3 pm to 5 pm, are scheduled per
each sampling day. The VOC sampling period started from August 10,
1992 to October 30, 1992. In some cases, the sampling was
622
-------
cancelled if ozone did not promisingly approach the violation
limit, 120 ppb. Non-methane organic compounds (NMOC) were
continuously monitored at supersite from Oct. 11 to October 30,
1992.
The carbonyl sampling started from Aug. 26, 1992 for all sites
except Volkswagen Proving Ground, which started by Sept. 9, 1992.
The sampling activities stopped by Oct. 7, 1992.
Table 1
Location Code
1
2
3
4
5
6
Site Name
South Scottsdale
Supersite
Valley National Center
Chandler Heights
Central Phoenix
Volkswagen Proving
Ground
3. LABORATORY ANALYSIS
VOC Samples:
The VOC sample analyses from SUMMA canisters were performed by ENSR
Consulting and Engineering complying with the EPA method TO-14.
The samples were cryogenically preconcentrated and then analyzed by
a GC-FID-ECD-PID system. The target species for the analysis are
listed as follows.
Acetylene
Propylene
Propane
Butane
Pentane
Hexane
Octane
Isoprene
Methylene Chloride
1,1-Dichloroethylene
Chloroform
1,1,1-Trichloroethane
Trichloroethylene
Tetrachloroethylene
Benzene
Toluene
Chlorobenzene
Ethylbenzene
m-Xylene
o-Xylene
p-Xylene
Styrene
Carbonvl Samples:
The analyses of C-18 DNPH cartridges for carbonyls were performed
by ENSR Consulting and Engineering pertaining to the requirements
of EPA method TO-5. The target species for the analysis are listed
623
-------
as follows:
Formaldehyde
Acetaldehyde
Acrolein
Methylethylketone
Acetone
Carbonyl in the ambient air reacted with DNPH to form stable
hydrazone derivatives, which are separated and quantified using
HPLC equipped with a UV detector at 360 nm. The quantities of
hydrazones detected were used to determine the quantities of
carbonyl by comparing with external standards.
4. RESULTS
Figure 1 and Figure 2 summarize the average NMOCs, BETX, and
carbonyl readings. Acrolein was not detected at any site.
Chandler Heights, a suburban location at south-east valley, and
Volkswagen Proving Ground, a suburban location at west valley area,
appeared to be good background sites, whereas other sites represent
areas affected by urban pollution sources.
South scottsdale at an east Phoenix metropolitan area has
appreciable amount of nearby automobile traffic. The average NMOC
is 580 ppb, about 57% more than that at Chandler Heights.
Supersite is located at a central Phoenix residential area. The
average NMOC is 538 ppb, about 45% higher than the average at
Chandler Heights.
Valley National Center at a central Phoenix area collects samples
from a higher elevation than other sites. The NMOC concentration
is 604.5 ppb, about 63% more than that at Chandler Heights. But
the overall readings at Valley National Center are slightly lower
than those at other urban locations.
Central Phoenix site is at downtown Phoenix where heavy bus and car
traffic is within 300 feet of distance. There is a parking lot
within 50 feet to this site. The concentrations of most non-
chlorinated VOCs are significantly higher than those of other
sites. The average NMOC is nearly 14 ppm as compared to a few
hundred ppb at other sites. Central Phoenix also has the highest
average carbonyl concentrations of all sites.
Figure 3 shows NMOC concentrations detected at the monitoring sites
during the sampling period of Aug. 10 to Oct. 30, 1992.
Concentrations detected at Central Phoenix were in the ppm range,
using the right-hand Y-axis. Concentrations from other locations
are mostly in ppb range, using the left-hand Y-axis.
624
-------
Total NMOC
Toluene
Benzene
Xylone/Styrene
Ethy (benzene
ppb
ppm
8. Scottadalo Superalte VNC Chdlr Hgtn Central Phx VW PQ
Figure 1 Distribution of average NMOC(ppm) & BETX(ppb)
-------
Formaldehyde
Methylethylkatona
Y//\ Acetaldehyde
Acetone
ppb
3. Scottadale Supersite VNC Chdlr Hgte Central Phx VW PQ
Figure 2 Spatial distributions of average carbonyl(ppb)
-------
—— 8. Scottadale(ppb) •
-a- Chdlr Hgte(ppb)
PPb
Superalte(ppb) •••*••• VNC(ppb)
Central Phx(ppm) -*- VW PG(ppb)
ppm
32OO
2400
16OO
8OO
8/1O 8/17 8/24 8/31 9/7 9/14 9/21 9/28 1O/6 1O/121O/191O/26
Figure 3 Average daily NMOC at the six monitoring sites
-------
AN AUTOMATED SYSTEM FOR THE ON-LINE ANALYSIS OF OZONE PRECURSORS
I. Seeley, The Perkin-Elmer Corporation, Norwalk, CT and G. Broadway, A. Tipler and E. Woolfenden, Perkin-Elmer Ltd.,
Beaconsfield, Bucks., UK.
1. ABSTRACT
A Perkin-Elmer Model ATD-400 Thermal Desorption System was specially adapted for the collection of whole-air
samples. The €2 to n-Cjo hydrocarbons were trapped using Peltier cooling on Carbotrap®/Carbosieve® adsorbents at -30°
C. Multidimensional chromatography allowed the €2 to Cg compounds to be separated using a porous layer open tubular
column while a methyl silicone column separated the Cg to CJQ fraction. No cryogen was used in the collection or analysis.
An air generator equipped for automatic restart and special software allowed the system to recover from power failures.
Ozone precursor concentrations were detected over 5000 hours of continuous operation at the 0.1 ppbV level. Area
repeatability was below 10%, with retention time precision over 100 hours of approximately 1% RSD for most compounds.
Remote communications allowed chromatography to be monitored and files to be downloaded to a central office at
19.2kbaud. Data in .CSV format was directly assimilated by spreadsheet software for graphing and customized reporting.
Data were archived using external Bernoulli® disks.
2. INTRODUCTION
The 1990 Clean Air Act Amendments require that ambient air be monitored for certain VOCs which catalyze the
formation of ozone in the presence of oxides of nitrogen and sunlight[l,2]. A system has been designed [3] that allows
these VOC species to be sampled from the ambient air on a regular basis to provide a continuous record.
It would be preferable if such a system be capable of running for long periods of time unattended, since, for most of
these sampling sites, no permanent operator is located and relative inaccessibility would make the guaranteed supply of
replacement consumables such as liquid cryogen next to impossible. The design of a system required for such an installation
must therefore provide for a high degree or robustness and automation. Following field trials it was seen that the system
must be able to survive a power outage and bring itself back on line automatically.
This paper describes some ideas and concerns leading to the implementation of such a system, and some experiences
and findings following (in excess of) 5000 hours of continuous operation on one such system.
3. INSTRUMENT DESIGN
A schematic of the system is shown in
Figure 1. A 600mL sample of air
(typically, but not limited to, ambient
air) is drawn by means of a small
sampling pump through the cold trap
of an ATD-400 thermal desorber.
The trap lower temperature is
set to -30°C by peltier effect (electric)
cooling while the air sample is
obtained. The design of this trap has
been previously described [4]. Once
the sample is collected the trap is
heated at 2400degC/min. to 325°C to
ensure that the VOCs are desorbed
rapidly from the cold trap in a fast,
focused band consistent with the requirements of the capillary column. (Figure 2). During this process the desorbed gas flow
through the trap is reversed to prevent the more volatile components from contacting the stronger adsorbent.
Figure 1. Sample Collection
628
-------
4. SAMPLE ANALYSIS
The first GC column is a 50m x
0.22mm ID x l.0\an methyl silicone
(BP-1®). The first (approx. 12
minutes) of compounds eluting from
the BP-1 are switched by a pressure-
balanced Deans' switch[5] to a 50m x
0.32mm ID Al2O3-Na2SC>4 porous
layer open tubular (PLOT®) column
and detected by an FID. Later eluting
components from the BP-1 (approx.
12 minutes to 48 minutes) are
detected directly by a second FID.
The chromatograms obtained from a
600cc sample of a nominal SppbV,
75% RH retention standard canister
are shown in Figure 3.
Figure 2. Cold Trap Desorption
1 2-=
1 o-i
8-
jaa_O3O
1 4-
1 2-
1 O-l
8-i
JJ
ab_O3O
Cohjn»2BP-l
uu
I i i i i i i i i i i i i i i i 1-1 r | r r i i | i i i < | i i < i | ' < < i | i i < <
5 1O 15 2O 25 3O 35 4-O
Figure 3. The 55-Components from the VOC List from Title 1 of the Clean Air Act
Initially a BP-5® (methyl 5% phenyl silicone) column was used for this work[6] to resolve the Cg to n-Cjo hydrocarbons.
However some important coelutions occurred resulting in the final selection of a non-polar methyl silicone column. In this
chromatogram all components are resolved: the component identifications are listed below.
1 Ethane
2 Ethylene
3 Propane
4 Propene
Isobutane
n-Butane
Acetylene
trans-2-Butene
Isobutene
10 cis-2-Butene
11
12
13
14
15
16
17
18
19
20
Cyclopentane
Isopentane
n-Pentane
2-Methyl-2-butene
Cyclopentene
trans-2-Pentene
3-Methyl-l-butene
1-Pentene
cis-2-Pentene
2,2-Dimethylbutane
21
22
23
24
25
26
27
28
29
30
3-Methylpentane
2-Methylpentane
2,3-Dimethylbutane
Isoprene
4-Methyl-1 -pentene
2-Methyl-1 -pentene
n-Hexane
trans-2-Hexene
cis-2-Hexene
Methylcyclopentane
629
-------
31 2,4-Dimethylpentane
32 Benzene
33 Cyclohexane
34 2-Methylhexane
35 2,3-Dimethylpentane
36 3-MethyIhexane
37 2,2,4-Trimethylpentane
38 n-Heptane
39 Methylcyclohexane
40 2,3,4-Trimethylpentane
41 Toluene
42 2-Methylheptane
43 3-Methylheptane
44 n-Octane
45 Ethylbenzene
46 p-Xylene
47 Styrene
48 o-Xylene
49 n-Nonane
50 Isopropylbenzene
51 n-Propylbenzene
52 a-Pinene
53 1,3,5-Trimethylbenzene
54 b-Pinene
55 1,2,4-Trimethylbenzene
The C2S (ethane, ethene and acetylene) have the highest vapor pressure of the target analytes, and normally require
large quantities of liquid nitrogen to trap and chromatograph them. Here ethane elutes at 8 minutes, ethene at 9 minutes
and acetylene at 24 minutes and are thus well resolved. No liquid cryogen is used in this system. Separation of the 55
hydrocarbons is facilitated by the multidimensional column set. By ensuring that no compounds more volatile than hexane
(in the elution order) are presented to the PLOT column this system, when operated continuously, provides optimum
stability. (Note that in this system no component elutes on both columns.)
The chromatographic conditions used for this analysis were:
GC:
Initial Oven Temp.
Initial Oven Time.
Ramp Rate
Oven Temp.
Ramp Rate
Final Oven Temp.
Final Oven Time
Mid-point Pressure
46°C ATD: Column Head Pressure 48psi
15min Collection Time 40min.
5degC/min. Sample Collection Flow 15cc/min.
170°C Cold Trap Low Temp. -30°C
15°C Cold Trap High Temp. 325°C
200°C Trap Hold at High Temp. lOmin.
6min. Cycle Time 60min.
21.5psi
Trap Packing: 44mg Carbotrap® C/60mg Carbosieve® SIII
Moisture management is by means of a semi-
permeable membrane dryer, the performance of which has
been described [8]. The analysis time is 48 min. The
parameters have been carefully optimized to ensure that this
system can meet its primary objective, viz.: a robust, turnkey
system capable of long term continuous operation with the
minimum of attention. It can be seen that, although all
peaks of interest elute by 41 minutes, the run is extended to
200°C and held there for 6 minutes. This is to provide the
PLOT column with a short bake which prepares it for the
next analysis. Similarly, the ATD trap, once it is fired to
325°C, is held at that temperature for 10 minutes to prepare
the sorbent bed for the next sample collection. By these
means it is assured that there is no carryover from the trap
and that optimum stability is obtained from the PLOT
column. A second firing of the trap under such
circumstances is blank[5]. Figure 4 shows the chromatograms obtained when no sample is collected. In this case, the small
low-boiling peaks obtained (e.g. 2mV methane) possibly represent residual sample diffusing through the sample collection
system: no peaks are obtained in the region Cg to CJQ.
12-
9-
5
15-
12-
9-
Ieo-140
PLOT Column
BP-1 Column
Figure 4. Blank Run - No sample collected.
630
-------
Peak area repeatability for successive samples of calibrant is shown in Table 1. Retention time stability data from 105 hours
of sequential sampling is shown in Table 2.
Table 1: Area Repeatability
Name
Ethane
Ethylene
Acetylene
1-Butene
Cyclopentane
1-Pentene
Benzene
Toluene
n-Octane
Ethylbenzene
p-Xylene
Styrene
Isopropylbenzene
Replicates
8
8
8
8
8
8
8
8
8
8
8
8
8
1,3,5-Trimebenzene 8
Mean
76236
69492
47997
118966
150602
128217
298474
122969
130084
54167
110158
106367
121744
140447
SD
3574
3378
2779
5743
5071
3446
10300
4643
2946
3309
4882
4069
9252
4586
%RSD
4.69
4.86
5.79
4.83
3.37
2.67
3.45
3.78
2.26
6.11
4.43
3.83
7.60
3.27
Table 2: Retention Time Stability
Name
Ethane
Ethylene
Acetylene
Butane
Isopentane
n-Pentane
1-Hexene
n-Hexane
Benzene
Toluene
n-Octane
Ethylbenzene
o-Xylene
1,3,5-Trimebenzene
Average RT
8.75
9.47
23.52
22.25
30.61
31.44
41.66
14.07
18.45
26.63
29.48
32.68
34.47
38.85
%RSD
0.27
0.37
0.82
0.45
0.25
0.24
0.23
0.23
0.16
0.10
0.07
0.07
0.06
0.07
5. System Timing for Hourly Analyses
The cycle time is 60 minutes, i.e., each run starts after exactly one hour - say 10 minutes after every hour. The
temperatures and times are selected such that, even under worst case conditions, both the GC and the ATD are able to
stabilize at their initial conditions prior to the next run (i.e., within the 60 minute envelope). The timing of the system is
shown in Figure 5.
LOAD TUBE
LEAKIEST
PURGE TUBE
COLLECT SAMPLE
DESORBTUBE
HEAT TRAP and HOLD
RECYCLE TO -30
GC HEADY
RUNNING
RECYCLE
DATA SYSTEM READY
COLLECTING DATA
PROCESS DATA
12:10 12:51 1:41
1
•_1
•* S
1
tart
I Ho
40
Jr
1
'10
•«-
Start
4B
ol A
4B
alys
40
s
7
, •*
1
1
" 10
Tim
nIF
A
8
48
0
40
epo
n
70
47
\.4
40
11
At
01
\c
7
Ah
'i
c
1
" 10
DL
YL
IN
'U
40
;
48
40
7
1
1
'10
48
48
-a a 10 20 30 40 50 60 0 10 20 30 40 50 60
0 10 20 30 40 50 60 0 10 20 30 40 50 •••
Figure 5. ATD-400 Ozone Precursor System Timing Diagram
The entire sample pathway is leak tested prior to collecting the first sample. Thereafter, a run is performed every 60
minutes. While a run is in progress a fresh sample is being collected. One such system has run 5300 hours on the same cold
631
-------
SSplI
BOpH
Zero Air Compreisor
Generator
Figure 6. Air Supply Plumbing
trap, with an interruption of 56 hours after 2500 hours for firmware updates. At 5300 hours the glass wool used to retain the
sorbent lost its flexibility and had to be replaced.
6. DESIGN CRITERIA FOR LONG TERM UNATTENDED OPERATION
6.1 Consumables
The consumable of most concern is air owing to the utilization of
800cc/min. by the dual FIDs, plus 250cc/min. by the Nafion(®) dryer, plus
20cc/min. by the ATD-400. A single tank of air lasts about 5 days. A tank of
helium carrier gas lasts 3 months or more since the flow is only about 5cc/min.,
and hydrogen is consumed at the rate of 80cc/min. by the dual FEDs so a typical
tank will last approximately 5 weeks. The design goal for this system was two
weeks without a visit.
A special commercial zero-air generator equipped with a self-regenerating
dryer (Balston Inc.) was used since it had a sufficient delivery rate (>lL/min.) to
meet all the requirements of the system and provides air with a dew point of
-100°C, making the Nafion® dryer more efficacious.
This approach suffers from a serious drawback in that it cannot survive a
power failure since the FID flames go out as the air pressure drops. A stand-by air
cylinder was installed teed into the air line as shown in Figure 6. By selecting
suitable pressures as indicated on the diagram the air tank outlet is arranged to be
normally just below the delivery pressure of the zero air generator and thus it stays off-line. If the pressure falls due to a loss
of zero air the air tank will start to supply when its pressure exceeds the compressor pressure. When power is restored the
air tank will take itself back offline after the zero air generator pressure rises. This simple arrangement does not require
power. The receiver tank is included to neutralize pressure surges that might otherwise extinguish the flames.
6.2 Surviving Power Outages
Once the system was configured to survive a power outage and bring itself back on line, the next step was to ensure
that the system would automatically restart sampling at the appropriate time after the top of the hour. This was done by
putting a wait statement into the computer Autoexec.bat startup file. Thus the computer cannot restart until a stipulated
time, at which point it allows the GC to come ready, which in turn allows the ATD-400 to resume sample collection. The
first run following any power failure will therefore be a cleanup run, with the second or third run containing useful data
depending on the period of the outage. A log is kept in the computer of all restarts.
6.3. Data Storage and Archival
The Turbochrom™ data handling system running under Windows® collects two channels of raw data per run at up to
70kbytes total (these are the chromatograms). In addition it is possible to create reports for each run of a further 1 Ikbytes
total. The operator will require ~20 preformatted 1.4Mb floppies to back up the data for two weeks. There is also the
matter of data integrity for Quality Control/Quality Assurance purposes: if the raw data are written to the computer hard
disk and there is a power outage with possible damage to the hard disk, the data are lost. To get around both of these issues
we have elected to use external Bernoulli disk drives. A 90Mb Bernoulli disk drive will hold at least a month's worth of
data. Access is as fast as a hard disk so it can be used by Turbochrom as a directory, unlike a tape cartridge which is purely
a sequential archive device. The data are now recoverable, since in the unlikely event of a head crash the disk may be
recovered using standard repair utilities. Most importantly, simply unplugging a full disk and replacing it with a
preformatted blank disk is all that is required to archive 90 megabytes of data - which is superior to offloading onto a large
number of floppies during a field visit. Alternatively, the data may be actively stored on the hard disk and archived
periodically to the Bernoulli disk.
6.4 Remote Communications
Since the system is running largely unattended, it is likely that the system supervisor will wish to examine the
chromatography of the system periodically. This is possible and convenient using high speed modems (Microm®, Model
QX/4232bis+ MNP-10) with Norton PCAnywhere® communications software. This modem is rated for cellular service and
is better able to tolerate poor quality phone lines. It is found that using regular phone lines access at 19,200 baud is possible
using MNP protocol, with the convenience of remote mouse operation for the manipulation of the data display. Using
PCAnywhere security features, the system can be programmed to call back only certain phone numbers thus limiting access
by unauthorized callers. Scheduling software allows file transfers to be automated such that the bulk of the report data is
transmitted at night on a daily basis to avoid peak calling periods.
632
-------
7. CONCLUSIONS
An automated system has been designed for the hourly detection of the Ozone Precursor compounds stipulated under
the Clean Air Act. Since Peltier cooling is used to trap the VOC species no liquid cryogen is required. Operation for a two
week unattended period is possible by configuring gas supplies for continuous operation. Optimum stability, coupled with
enhanced resolution of the €2 compounds is achieved by multidimensional chromatography. The system may be accessed
and files transferred using high speed data communications. Automatic recovery in the event of a power failure ensures the
maximum data capture rate for remote installations.
References:
1. US. Environmental Protection Agency, Code of Federal Regulations. Title 40 Part 58. Enhanced Ozone Monitoring
Regulations. October 1991.
2. Technical Assistance Document for Sampling and Analysis of Ozone Precursors. US EPA, RTP, NC, EPA Report 600/8-
91/215. Nov. 1991
3. Broadway. G, Woolfenden. E, Ryan.J and Seeley.I. Proc. of the US EPA/AWMA Int. Svmp.. Paper #401, 1992
4. Broadway G., Trewern T, Proc. 13th Int. Svmp. on Capillary Chrom. Vol. 1 pp310-320
5. Deans D., Chromatoeraphia 1 (1968) 18
6. Broadway G., Proc. 14th Int. Svmp. on Capillary Chrom. pp 276-281
7. Broadway G., Tipler A., Seeley I., 15th Int. Svmp. on Capillary Chrom. 1993
8. Pleiel J.D., Oliver K.D., McClenny W.A., JAPCA. 37 pp 244-248. 1987
ACKNOWLEDGMENT
The ATD-400 On-Line Air Monitoring Capability and Ozone Precursor Application was developed by the Perkin-
Elmer Corporation in collaboration with the US EPA Exposure and Assessment Laboratory under an FTTA
agreement
Trademarks:
PLOT
Carbosieve
Carbotrap
Turbochrom
Microcom
Chrompack BV
Supelco Inc.
Supelco Inc.
Perkin Elmer Corp.
Microcom Corp.
PCAnywhere Norton/Symantec
BP-l,BP-5 SGE Corp.
Windows Microsoft Corp.
Nafion Permapure Inc.
Bernoulli Iomega Corp.
633
-------
Field Monitoring of Ozone Precursors Using
An Automated Gas Chromatographic System.
Teni V. Brixen and John K. Stewart
Air Surveillance Branch
Division of Air and Waste Management
Delaware Department of Natural Resources
and Environmental Control
715 Grantham Lane
New Castle, Delaware 19720
ABSTRACT
Title I, Section 182 of the 1990 Clean Air Act Amendments requires states in ozone non-
attainment areas to establish photochemical assessment monitoring stations (PAMS) which include
monitoring for ozone precursors. During the summer of 1992, Delaware was one of three states
along the Northeast Corridor to evaluate the Chrompack automated gas chromatograph in the field.
Air samples were taken through a manifold, concentrated cryogenically, and analyzed for target
volatile organic compounds by capillary gas chromatography with FID detection. This paper will
discuss Delaware's preliminary assessment of the system and initial results.
DISCLAIMER
Although the research described in this paper has been funded in part by the U.S.
Environmental Protection Agency (EPA), it has not been subjected to EPA's peer and administrative
review; it does not necessary reflect the view of the Agency and no official endorsement should be
inferred.
634
-------
INTRODUCTION
On March 4, 1992 the proposed rule to revise the ambient air quality surveillance regulations
to include provisions for the enhanced monitoring of ozone and its precursors was published in the
Federal Register'. These regulations addressed the minimum requirements for the monitoring of
speciated volatile organic compounds (VOC), oxides of nitrogen (NOJ, and meteorological
parameters.
In accordance to the proposed rule, Delaware Department of Natural Resources and
Environmental Control, Air Surveillance Branch established a Type 1 PAMS which characterizes
upwind background and transported ozone precursor concentrations entering the Philadelphia
Consolidated Metropolitan Statistical Area (CMSA). The station is located at Lums Pond State Park
which is approximately 9 miles southeast from downtown Wilmington. The 8 x 12 foot (ft) shelter
is equipped with a Chrompack VOC Air Analyzer, Dasibi ozone analyzer and a Monitor Labs
NO/NO2/NOX analyzer. Meteorology data was obtained from the National Weather Service located
at the Greater Wilmington airport during this study.
EXPERIMENTAL METHODS
Ambient air was sampled 34 minutes during each hour for a minimum of 24 hours, once a
week from June to August 1992. A Chrompack pump unit was used to draw air from the manifold.
This unit included a mass flow controller to control sample flow and a Perma-Pure dryer system to
selectively remove water.
The sample was collected and preconcentrated onto a Carbotrap C/Carbotrap/Carbosieve
adsorption tube (75mm by 6mm OD by 3mm ID) at a 10 ml/min flow rate. The primary trap was
cooled to -21 °C using liquid nitrogen. After desorption from this trap at a temperature of 250°C,
the sample was transferred to the Poraplot U fused silica trap (0.53mm ID, 20 um film thickness)
for cryogenic focusing of components. Liquid nitrogen was used to cool the secondary trap to -
105°C. The cold trap was flash heated to 140°C and then injected onto the column. The cycle of
collection, concentration, drying, desorption, trapping and injection was completely automated and
controlled by the Chrompack auto TCT unit.
A single A12O3/KC1 PLOT capillary column 25m by 0.32mm ID was used for separation.
The column temperature program had an initial temperature of 50°C for 2 min, a rise of 5°C/min
up to 75 °C, 10°C/min up to 125°C, 15°C/min up to 200°C, and a final temperature of 200°C for
30 minutes. The carrier gas was ultra pure grade helium. Analyses of the samples were performed
on a Chrompack CP-9000 GC with a single flame ionization detector. Data acquisition utilized the
Chrompack PCI software. Results were both printed out and also stored on the hard drive and then
transferred to a 3.5 floppy disk.
A 60 component standard supplied by Radian was used for retention time calibration. A 1 %
window was set for compound identification. Single point calibration using benzene as a standard
was done to determine the response factor.
RESULTS
49 of the 60 VOCs in the calibration standard were separated and tentatively identified.
635
-------
methylhexane occurred as one peak as did 2&3- methylheptane and m/p-xylene. Out of 6
calibration runs, the coefficient of variation for retention times ranged from a low of 0.13% for
ethane to a high of 0.68% for n-propyl-Benzene (Table 1). 42 of the 52 target VOC peaks were
detected in the ambient air during the sampling period. These 42 made up 72.75% of the total
measured VOCs. The maximum value measured was above 10 ppbC for the following compounds:
propane, n-butane, 2-methylbutane, n-hexane, 2&3-methylheptane, n-octane, toluene, m/p-xylene
and 1,3,5-trimethylbenzene. The maximum value measured was between 5 and 10 ppbC for the
following compounds: ethane, propene, acetylene, isobutane, isobutene, n-pentane, 2&3-
methylhexane, benzene, and 2,3,4-trimethylpentane. The maximum measured value was between 1
and 5 ppbC for the following compounds: ethene, isobutane, trans-2-butene, cis-2 -butene,
cyclopentane, 2-methyl-2-butene, methylcyclopentane, 2,3-dimethylbutane, 2-methylpentane, 3-
methylpentane, isoprene, 3-methylpentene, 2,3-dimethylpentane, 2,4-dimethylpentane,
methylcyclohexane, n-heptane, perchloroethylene, nonane, ethylbenzene, o-xylene, and
isopropylbenzene. The maximum measured value was less than 1 ppbC for the following
compounds: trans-2-pentene, pentene, and trichloroethene. The following compounds were not
detected: 1-butene, 3-methyl-l-butene, 2-methyl-l-butene, cis-2-pentene, cyclohexane,
trichloromethane, trans-2-hexene, 2-methyl-l-pentene, cis-2-hexene and n-propylbenzene.
Averaging all of the sample data, the 10 most abundant compounds on a carbon basis were in order
of abundance: propane, isobutene, toluene, ethane, 2-methylbutane, n-butane, trans-2-butene, n-
hexane, acetylene and benzene. These 10 compounds made up 53.91% of the total measured
VOCs ( Table 2). Total measured VOCs ranged from 25 to 199 ppbC with a mean of 58 ppbC.
Averaging total VOCs per time of day, total measured VOCs reached the highest level at 0900
(Figure 1).
DISCUSSION
Approximately 50 runs could be obtained using one 160 liter dewar of liquid nitrogen. Use
of liquid nitrogen at a field site in this size container caused logistical problems. Due to a state
contract the exact time of delivery could not be specified. On occasion containers were delivered
several days early and lost much of their contents before they were used. Man-hours necessary to
maintain the system were higher than anticipated due to the frequent need to check on or change the
liquid nitrogen container. Sampling was not done over the weekend because it would have required
someone to change the liquid nitrogen container on the weekend and there were no funds for
overtime pay. Delivery of cylinders to a site that was not always occupied also caused occasional
problems. Expected deliveries were not received when drivers unfamiliar with the site did not know
the combination to the enclosure, arrived in a truck too large to fit on site, and dropped off the
wrong order. Once the cylinders were received, ease of movement was facilitated by a concrete
walkway along the side of the station.
The EPA Technical Assistance Document2 encourages the use of a less complex single-
column configuration to perform the initial system set-up and optimization. This approach worked
quite well and is recommended for the less experienced chromatographer. The Chrompack
A12O3/KC1 PLOT column provided good separation of the light hydrocarbons but many of the peaks
for the C5 - C8 range were very close together and difficult to identify. Some Q hydrocarbons
could be identified but the peaks were very broad. A second column is recommended to confirm
analytical results.
On 4 occasions power interruptions to the station occurred during the sampling period.
When this occurred the system went into the standby mode and did not continue sampling when the
power resumed. 51 hours of intended sampling time were lost because of this. To remedy this
636
-------
problem in the 1993 sampling season, an un-interruptible power supply has been purchased.
Sampling time was also lost due to a missing ferrule on the pump unit when the
instrumentation was moved from the laboratory to the field station. This caused a major air flow
leak into the system. Hardware down time occurred when the main fuse was blown. The system
status indicated " Motor Off ". This was resolved as soon as the fuse was replaced. Detailed
recorded keeping is a must for this type of system. It will be used for both trouble shooting the
hardware and also to provide quality in the data.
The PCI software was used in operation control of the automated GC system and for hourly
data analysis and information storage but was not useful for overall statistical analysis between
groups of the hourly files. Additionally the software occasionally appeared to fail to identify
compounds that were within the window limits. The need for better software was very evident.
For the 1993 season our agency has purchased the Chrompack Mosaic software.
The 8 x 12 ft shelter was barely adequate in size. It was inconvenient when more than one
person was working within the station or when work which required access behind the instruments
needed to be done.
CONCLUSIONS
In a preliminary study, an automated GC system located at a field site was used to analyze
ambient air samples for VOCs. 10 compounds made up 53.91% of the total measured VOCs. The
mean of the total measured VOCs peaked at 0900. Problems with the system included frequent
need to monitor liquid nitrogen use, complete shutdown of the system due to momentary power
losses, and the need for better software. Advantages to the system include hourly data, good
separation of the light hydrocarbons, and minimum variability between calibration runs.
Further technical development needs to address hardware improvements, analytical method
research and development, and quality assurance and quality control.
ACKNOWLEDGEMENTS
The authors wish to express their appreciation to Delaware 's Division of Parks and
Recreation for providing the availability of land for the PAMS site at Lums Pond State Park and
also to the United States Environmental Protection Agency at Region III for both their financial and
technical support in this project.
REFERENCES
1. Ambient Air Quality Surveillance; Proposed Rule, Code of Federal Regulations. Title 40,
Part 58, March 4, 1992, p. 7687.
2. Technical Assistance Document for Sampling and Analysis of Ozone Precursors. EPA/600-8-
91/215, U.S. EPA, Research Triangle Park, NC, October, 1991, p 2-15.
637
-------
Table 1. Mean Retention Times for Calibration Runs
Compound
ethane
ethene
propane
propene
acetylene
l-butane
n-butane
trans-2-butene
1 -butene
l-butene
cls-2-butene
cyclopentane
2-methylbutane
n-pentane
3-methyl-1 -butene
trana-2-pentene
2-methyl-2-butene
pentene
2-methyl-1 -butene
cls-2-pentene
methylcyclopentane
cyclohexane
2,3-dlmethylbutane
2-methylpentane
3-methylpentane
iaoprene
RT mlna
1.613
1.783
2.348
3.656
4.690
4.812
5.137
7.721
7.886
8.388
8.760
9.431
9.560
10.052
11.364
11.671
12.008
12.074
12.219
12.437
13.029
13.229
13.330
13.397
13.470
13.816
SO
0.002
0.004
0.008
0.018
0.024
0.031
0.026
0.034
0.034
0.032
0.033
0.033
0.036
0.032
0.028
0.027
0.027
0.027
0.027
0.026
0.036
0.043
0.026
0.027
0.028
0.029
CV *
0.132
0.204
0.323
0.496
0.503
0.652
0.509
0.443
0.428
0.384
0.372
0.349
0.372
0.320
0.249
0.234
0.222
0.224
0.219
0.206
0.279
0.323
0.195
0.202
0.206
0.207
Compound
n-hexane
trlchloromethane
3-methylpentene
trans-2-hexene
2-methyl-1 -pentene
cls-2-hexene
trlchloroethene
2,3-dimethylpentane
2,4-dlmethylpentane
methylcyclohexane
2&3-methylhexane
n-heptane
perchloroethylene
benzene
2,2,4-trlmethylpentane
2,3,4-trimethylpentane
2&3-methylheptane
n-octane
toluene
nonane
ethylbenzene
m/p-xylene
o-xylene
iaopropylbenzene
n-propylbenzene
1,3,5-trlmethylbenzene
RT mlna
14.007
14.407
14.663
14.711
14.783
15.315
15.561
15.751
15.909
16.074
16.146
16.494
16.663
17.102
17.910
18.332
18.559
19.046
20.118
23.429
24.753
25.289
27.193
30.006
30.620
34.086
SO
0.020
0.029
0.027
0.027
0.028
0.027
0.025
0.029
0.032
0.029
0.027
0.029
0.026
0.029
0.037
0.041
0.043
0.046
0.059
0.109
0.127
0.128
0.147
0.176
0.209
0.226
CV %
0.144
0.204
0.184
0.187
0.186
0.174
0.161
0.181
0.204
0.182
0.170
0.174
0.158
0.170
0.209
0.225
0.233
0.244
0.296
0.464
0.511
0.505
0.541
0.587
0.683
0.664
-------
Table 2. Ten most abundant compounds
Compound
Propane
i-Butene
Toluene
Ethane
2-Methylbutane
n-Butane
trans-2-Butene
n-Hexane
Acetylene
Benzene
Total VOCs
Mean Concentration
(ppbC)
6.09
6.01
3.07
2.98
2.92
2.39
2.18
2.10
2.10
1.62
58.38
Standard
Deviation
8.52
1.81
2.30
1.42
2.96
2.53
1.42
2.54
1.2S
1.21
30.78
% of total
VOCs
10.43
10.30
5.26
5.11
5.01
4.10
3.74
3.60
3.59
2.77
100.00
Figure 1. Total measured VOCS vs Time
120
100 -
0300 0600 0900 1200 1500 1800 2100
Average of 1992 da
639
-------
PERFORMANCE EVALUATION OF THE HP-5971A MSD
FOR ANALYSIS OF VOCs IN AIR
Alston Sykes, Mitchell Howell, William Preston, and Roy Gorman
Geraghty & Miller/Acurex Environmental
Air Toxics Laboratory
P.O. Box 13109
Research Triangle Park, NC 27709
ABSTRACT
The analysis of volatile organic compounds in air by GC/MS using the Hewlett-Packard Model 5971A
Mass Selective Detector (MSD) has been evaluated over the past 18 months. The analytical methods used for
VOC analyses were TO-14, TO-1, TO-2 for ambient air samples, IP-1A and IB for indoor air samples, and
the VOST and Method 18 for source test methods. Practical quantitation limits for the 60 VOCs were 5 ng
on-column, which ranges from 1 to 2 parts per billion by volume, in the full scan mode. Precision and
accuracy was determined by replicate analyses, and interlaboratory comparisons of audit gases. For six audit
canisters (three analyzed one year, and another three analyzed a year later), the average accuracy or bias was
92%, and the average precision (% RSD) for 19 volatile organic compounds was 24%.
INTRODUCTION
Volatile organic compounds are being measured in air to determine emissions from chemical
processes, materials, landfills, and control equipment. VOC measurements are also needed to determine air
pathways, and exposure assessments in ambient and indoor air environments. Samples are collected in Tedlar
bags, Tenax and charcoal sorbents, passivated stainless steel containers, and are analyzed on-site or shipped
to the laboratory for analysis. The HP-5970 mass selective detector has been used successfully for many
years for analysis of VOCs in air, and was upgraded by the introduction of the HP-5971A model. Several
changes that were made were replacing the pumping system from a turbo-molecular to a diffusion pump,
reducing the physical size of the detector, reducing the mass range from 800 amu to 650, and upgrading the
HP ChemStation software to a MS-DOS personal computer system. With the lower cost 5971A model, and
its compact size, it became a very attractive instrument for laboratories to consider for environmental
analyses. Our laboratory purchased a 5971A MSD with the MS-DOS system for air toxics analyses, and
have summarized the performance for VOC analyses. The objectives of our evaluation were to determine
VOC in air analysis precision and accuracy, practical quantitation limits in the full scan electron impact
mode, it's adaptability to modifications, and the system's ease of use and maintenance requirements under
daily routine use.
EXPERIMENTAL
The 5971A MSD we purchased was the most recent version, which had been improved by Hewlett-
Packard by adding vent holes to the ion source and upgrading the mother board. The MSD was added to an
existing HP-5890 gas chromatograph which was connected to a modified Tekmar LSC-2000 purge and trap
concentrator used to concentrate and introduce air samples to the GC. The canister, Tedlar bag, or VOST
cartridges were connected to a heated stainless steel transfer line, 1/16 in. OD, using a leak tight fitting. The
LSC purge line was disconnected from the glass purge vessel, to provide purge gas for VOST samples, and
the transfer line from the air sample was connected directly to the purge vessel. This configuration allowed
the sample to enter the purge vessel and then concentrate onto the trap in the Tekmar apparatus. With
VOST and TO-1 Tenax cartridges, a clam shell type desorption heater was used to thermally desorb the
VOCs through the heated transfer line. For canisters and Tedlar bag samples a known amount of sample
(usually 0.5-1 liter) was concentrated onto the carbon molecular sieve trap in the Tekmar LSC-2000. The
sample flow rate was controlled with a Porter mass flow controller and a vacuum pump connected at the vent
side of the Tekmar unit. The concentrated sample was then thermally desorbed onto a fused silica capillary
column.
640
-------
Tekmar LSC-2000 Conditions:
Purge time:10 min at 50 mL/rain Mount, Transfer Temp: 110°C, 150 °C
Valve Temp: 200°C Desorb Temp: 250°C for 3 min at 10 mL/min
Bake out Temp: 270°C Trap: Carbopack B, 7.6 cm/Carbosieve S-III, 1.3 cm
GC/Mass Selective Detector Conditions:
Transfer Interface Temp: 250°C Injection Port Temp: 200°C
Oven: -50°C for 2 min programmed to 220°C @ 8°C/min
Column: 30m x 0.53mm 3 u, DB-624; He flow rate: 10 mL/min,
SGE Megabore Stainless Steel Jet Separator
Full Scan Mode: 35-260 AMU
HP ChemStation MS-DOS 1990 Ver. 4 Software with Microsoft Excel
A calibration curve consisting of 10 ng, 50 ng, 100 ng, 200 ng, 500 ng, and 1000 ng was constructed
by injecting aliquots of each analyte (Table 1) into the concentrator purge vessel during the purge cycle.
Certified standard mixtures obtained from Absolute Standards at 2000ng/uL in methanol solvent were used.
200uL of the target compound mixture was injected into a 2.0 liter static dilution bulb containing ultra pure
nitrogen. The resulting concentrations were 200ng/mL of gas. The internal standard method of calibration
was used for all target compounds. The following internal standards were used: 1) 1,4-difluorobenzene,
50ng; 2) dj-chlorobenzene, 50ng. The following surrogate compounds were used to determine injection
recoveries: 1) bromochloromethane, 50ng; 2) d4-l,2-dichloroethane, 50ng; 3) dg-toluene, 50ng; 4) 4-
bromofluorobenzene, 50ng. Preparation of the internal standard and surrogate standard gas mixture was done
by injecting 200uL of a lOOOng/uL certified stock solution from Ultra Scientific into a 2.0 liter static dilution
bulb containing nitrogen. The resulting concentrations were lOOng/mL of gas. One-half mL of this standard
was injected with each sample and target standard mixture. The automatic macro quantitative calculation
program in the HP ChemStation software was setup to calibrate using the internal standards and calculate
unknown analytes using the average response factor determined from all points.
RESULTS AND DISCUSSION
The system was tuned daily using BFB and has had no problems meeting the EPA acceptance criteria.
Operator use has been easy, and maintenance has been extremely low. EPA performance audits in SUMMA*
canisters had acceptable precision of 24%, and an accuracy of 92%, at levels of 1-20 ppbv (Table II).
Audit results of Tenax cartridges (Table III) loaded with 13 compounds in the range of 100 ng to 800 ng
demonstrate that average recoveries of 88%, with a standard deviation of 20 can be achieved. Figure 1
presents results of the 1,4-difluorobenzene internal standard, and the surrogate standard recovery plots over
48 injections. The internal standard had an %RSD of 16.3 over a period of three weeks, and the second
internal standard (ds-dichlorobenzene), which is not shown here, performed almost the same with a %RSD of
20.8. The surrogate standards had the following acceptable recoveries, and %RSD's: bromochloromethane,
92%, 7.6; d5-l,2-dichloroethane, 100%, 8.3; da-toluene, 102%, 5.4; bromofluorobenzene, 97%, 8.9.
Additional precision, accuracy, and SUMMA* canister stability results of 21 VOCs at 5 ppbv are presented in
Table IV. The atmospheric pressure 6-liter audit canister was analyzed over a 60 day period with an average
recovery of 97.2%, with a standard deviation of 19.7.
One problem that occurred and was resolved was the effects of moisture. Common to quadrapole
mass spectrometers is the suppression effects of water in the ion source. The water was not a problem in
most ambient air samples using the Carbopack B/Carbosieve SHI sorbent focusing procedure because of its
hydrophobic characteristics. The very small amount of water that did enter the source did not interfere with
any analytes of interest. Unfortunately, the audit samples were fortified with large amounts of water, that co-
eluted with the 1,4-difluorobenzene internal standard, and took all day to come back to acceptable levels in
the MSD. The effects were a 20% reduction of response of the internal standard, but not to the target
compounds, which incorrectly adjusted the analyte values to a higher amount. A 4 min. dry purge was added
to our procedures, and removed sufficient water to maintain consistent responses. We also recalculated the
analyses that showed water effects, using the external standard procedure, with acceptable results. This
further proves the water was the cause of the high recoveries. Further work should be done to fully evaluate
the effects of water and to establish guidelines in dealing with the potential errors involved. Although we
experienced a few problems, we found the system met our performance specifications, and expectations.
641
-------
Table I. VOCs on HP-5971A Mass Selective Detector.
The following compounds have practical quantitation limits of 1 ppbv in SUMMA* canister air samples
analyzed in the full scan mode.
Dichlorodifluoromethane
Chloromethane
2-Methyl-l-Propene
Vinyl chloride
Bromomethane
Chloroethane
Trichlorofluoromethane
1,1-Dichloroethene
lodomethane
Carbon Disulfide
Acetone
Acetonitrile
Methylene chloride
Trans-1,2-dichloroethene
2-Methyl-2-Propanol
Hexane
1,1-Dichloroethane
Vinyl Acetate
2-Butanone
Chloroform
1,1,1-Trichloroethane
Carbon tetrachloride
1,2-Dichloroethane
Benzene
Fluorobenzene
Heptane
2-Chloro-2-Methylpropane
Trichloroethene
1,2-Dichloropropane
Dibromomethane
1,4-Dioxane
Bromodichloromethane
cis-1,3-Dichloropropene
4-Methyl-2-Pentanone
Toluene
trans-1,3-Dichloropropene
Tetrachloroethene
1,1,2-Trichloroethane
Bromoacetone
2-Hexanone
Dibromochloromethane
1,2-Dibromoethane
Chlorobenzene
1,1,2,2-Tetrachloroethane
Ethyl benzene
o, m, p-xylenes
Nonane
Styrene
Bromoform
Cumene
1,2,3-Trichloropropane
1,4-Dichloro-2-butene
1,1,2,2-Tetrachloroethane
Pentachloroethane
1,3-Dichlorobenzene
1,4-Dichlorobenzene
1,2-Dichlorobenzene
1,2-Dibromo-3-chloropropane
Based on a 500 mL air sample
642
-------
Table II. EPA Canister Audit Results, 1992 and 1993.
Canister #
ng/L
Vinyl chloride
Bromomethane
Trichlorofluoromethane
Methylene chloride
t-1 ,2-dichloroethene
1 , 1-dichloroethane
Chloroform
1 , 1 , 1-trichloroethane
Carbon tetrachloride
Benzene
Trichloroethene
1 ,2-dichloropropane
Toluene
Tetrachloroethene
1 ,2-Dibromoethane
Chlorobenzene
Ethyl benzene
o-xylene
Styrene
Average Recoveries
Standard Deviations
%RSD
519
Loaded
3.5
5.6
8.8
5.0
5.5
6.6
7.2
8.1
8.6
5.1
8.7
7.4
6.2
11.0
11.3
7.5
7.1
18.7
2.1
%
Rec
97.1
89.3
34.4
66.0
101.8
78.8
90.3
90.1
96.5
107.8
97.7
89.2
101.6
70.0
74.3
82.7
85.9
147.1
128.6
94.2
19.7
20.9
484
Loaded
4.1
6.5
70.8
5.8
6.4
7.5
8.2
9.3
9.9
5.8
10.0
8.5
7.1
12.7
13.0
8.6
8.1
10.0
2.5
%
Rec
85.4
78.5
28.3
69.0
104.7
77.3
89.0
89.2
92.9
103.4
121.0
85.9
104.2
72.4
73.1
81.4
81.5
138.0
120.0
92.6
19.0
20.5
A06
Loaded
11.4
18.2
16.1
17.9
21.2
23.1
26.0
27.7
16.3
28.0
23.9
20.0
35.6
36.4
24.1
22.7
28.2
6.9
%
Rec
105.3
78.0
98.8
101.7
59.0
87.0
62.3
92.8
93.9
125.4
84.5
80.5
71.3
73.6
78.0
82.8
138.3
115.9
90.5
21.1
23.3
711
Loaded
11.7
15.9
25.2
7.0
7.9
20.0
23.2
28.8
28.0
7.7
25.4
11.3
15.2
31.8
25.1
10.5
10.1
11.9
3.0
%
Rec
123
98.9
100
151
128
99.0
99.0
66.6
41.1
108
98.6
94.1
136
95.3
76.9
98.8
104
79.6
132
104
25.8
24.8
709 %
Loaded Rec
51.3 119.4
117 87.5
66.1 56.2
93.5 86.8
52.2 77.0
85.4
22.8
26.7
740 %
Loaded Rec
25.1 108.9
57.1 92.4
32.3 48.7
45.7 94.5
25.6 82.9
85.5
22.6
26.4
Mean
92.0
21.8
23.8
i Rec is percent recovery; 1992 audits were: # 519, # 484, # A06; 1993 audits were: # 711, # 709, tt 740.
-------
Table III. EPA Tenax Cartridges Audit Results.
Tenax Cartridge #
Fall 1992
Chloroform
1,1,1-TrichIoroethane
1 ,2-Dichloroethane
Carbon tetrachloride
Benzene
Trichloroethene
1 ,2-Dichloropropane
Toluene
Tetrachloroethene
Chlorobenzene
Ethyl benzene
o-Xylenes
Average Recoveries
Standard Deviations
A-122
ng
Loaded
97
131
123
156
172
191
75
198
159
217
142
172
%
Rec
63
56
74
77
75
68
85
101
95
92
113
126
85
21
A-117
"g
Loaded
159
212
199
240
278
309
122
320
257
350
229
278
%
Rec
70
67
82
91
82
74
94
102
104
97
120
131
93
19
A-118
ng
Loaded
372
504
473
625
662
735
290
761
611
834
544
663
%
Rec
66
53
75
88
73
72
91
91
103
91
115
124
87
20
Average
Recoveries
67
59
77
85
77
71
90
98
100
93
116
127
88
20
i Rec is percent recovery.
Figure 1. Internal Standard Response and Surrogate Recoveries.
1,4-Difluorobenzene Internal Standard!
8 450000
I | 350000
i | 250000
<: 150000
*•»•»
fScscsr*!c*)
Injection Number
Surrogate Recoveries I
120
110
100
90
80
70
60
120
100
80
60
40
20
0
--» BCM
> D5-DCE
DS-Toluene — O BFB
644
-------
Table IV. Precision, Accuracy, and Stability of VOCs in a SUMMA Canister Over 60 Days
Date Analyzed 3/1/93 3/11/93
Elapsed Days Day 1 Day
Canister # 745 @ 5 ppbv
March 1-April 29, 1993
Atmospheric press.
Vinyl chloride
Bromomethane
Trichlorofluoromethane
1,1-dichloroethene
Methylene chloride
t- 1 ,2-dichloroethene
1,1-dichloroethane
2-butanone
Chloroform
1,1, 1-trichloroethane
Carbon tetrachloride
Benzene
Trichloroethene
1 ,2-dichloropropane
Toluene
Tetrachloroethene
1 ,2-Dibromoethane
Chlorobenzene
Ethyl benzene
o-xylene
Styrene
ng/L
Loaded
11.2
17.9
27.9
19.8
15.9
17.6
20.8
4.2
22.7
25.6
27.3
16.1
27.6
20.1
19.7
35.1
35.9
23.7
22.4
27.7
6.8
3/11/93
10
3/17/93
Day 16
4/5/93
Day 36
ng/L Found
14.0
19.6
28.6
20.1
16.3
18.3
18.8
4.6
24.1
26.8
18.9
15.3
25.2
20.7
17.8
33.4
36.3
22.4
23.1
21.6
8.0
Average
Standard Deviation
14.0
19.6
28.2
17.6
15.5
18.1
17.8
3.9
23.4
25.2
18.3
14.3
23.7
19.0
16.9
32.0
34.3
21.4
20.8
20.9
7.0
14.0
20.0
28.0
18.0
16.0
19.0
18.0
4.6
23.0
27.0
11.0
15.0
24.0
21.0
18.0
32.0
34.0
22.0
20.0
18.0
6.8
15.0
19.0
32.0
22.0
19.0
19.0
19.0
7.4
24.0
30.0
0.0
18.0
29.0
22.0
22.0
37.0
41.0
25.0
24.0
21.0
8.7
4/16/93
Day 47
df:1.7
10.5
18.4
28.4
17.8
17.3
18.9
19.7
6.0
23.8
25.1
0.0
16.2
25.7
22.6
20.9
32.6
32.7
22.1
20.6
19.0
7.2
4/29/93
Day 60
df:L7
11.4
16.7
33.3
19.6
17.8
17.3
18.0
7.5
21.0
24.3
0.0
16.0
21.4
19.5
16.7
30.7
31.9
16.6
20.2
17.6
6.6
Mean
13.2
18.9
29.8
19.2
17.0
18.4
18.5
5.7
23.2
26.4
8.0
15.8
24.8
20.8
18.7
33.0
35.0
21.6
21.5
19.7
7.4
Stdev
1.7
0.6
1.7
1.9
1.4
0.4
0.8
1.4
0.5
2.0
9.3
1.4
2.1
1.4
2.2
2.1
3.3
1.4
1.7
1.5
0.8
%RSD
13.2
3.3
5.6
10.0
8.1
2.4
4.2
24.7
2.0
7.5
116.2
9.1
8.5
6.6
11.8
6.3
9.3
6.5
8.1
7.8
10.6
Mean
%Rec
vs Load
117.6
105.3
106.7
97.1
107.1
104.8
89.0
133.6
102.1
103.1
29.4
98.4
90.1
103.8
95.1
94.0
97.7
91.0
95.8
71.0
108.1
97.2
19.7
df: dilution factor; % Rec: percent recovery
-------
Whole Air Analysis For TO-14
Low Level And High Level Samples
Richard Jesser
Graseby Nutech
4022 Stirrup Creek Drive, Suite 325
Durham, North Carolina 27703-9000
ABSTRACT
A description of the design and operation of a cryoconcentrator for the Whole Air Analysis of
TO-14 is discussed. Data is presented showing the precision (RSD's) of high level (source) and low level
(ambient) samples and system blanks. Emphasized is the ability of the instrumentation to recover from
both high and low level samples.
646
-------
INTRODUCTION
This talk will address instrumentation designed to accommodate the variety of sample types
occurring in today's TO-14 whole air analysis. Whole air analysis refers to the whole or entire air sample
compound set. This contrasts with analyses performed by sorbent approaches whereby, potentially, part
of the sample can be missing. The specific sample set we will examine in this study is that of TO-14 which
consists primarily of halogenated hydrocarbons.
BODY
Today, Mother Nature gives us approximately 20ppbv of these TO-14 molecules in ambient air
(consequently, the 2ppbv to SOppbv calibration range for TO-14). However, special situations can occur
to cause ambient concentrations to be dramatically higher— the sampling may include source (high level)
material or sample containers may inadvertently be switched in the laboratory. The complication of
mixing source and ambient samples is that concentrations may be different by many orders of magnitude.
With prior knowledge, the careful analytical chemist will group his samples by both concentration and
type. While this will not prevent an accidental mix-up, this approach will minimize the effect of carryover
or cross contamination. Ideally, the analytical system employs a well-engineered design for addressing
these real life complications. The first part of this talk will discuss such a design, and the second part will
show the results from a TO-14 analytical set illustrating the effectiveness of this design.
One approach, in principle, to an effective design is to provide a high level (source) sample
analysis pathway that is separate from the low level (ambient) sample analysis pathway. The analyst could
then channel known source samples through the high level flow path and ambient samples through the
low level flow path. This design offers the flexibility of screening all samples as if they were source
samples, and the apparently ambient samples could be then analyzed through the low level flow path.
Even with this approach, cleanup between samples is required to minimize cross contamination. Cleanup
that is continuously in operation provides a maximum of recovery. The following flow schematics show
such a system and how it processes high and low level samples while maintaining a continuous purging to
ensure a maximum of recovery capability.
CONCLUSIONS
The instrumental design discussed here has established, through the data presented, the capability
of analyzing the wide variety and concentrations of TO-14 samples challenging the environmental
laboratories today.
REFERENCES AND BIBLIOGRAPHY
None cited.
647
-------
Session 17
Acid Aerosols
and Related Pollutants
-------
ACID AEROSOL MEASUREMENT METHODS:
A SUMMARY OF U.S. EPA INTERCOMPARISONS
T. G. Ellestad
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Studies of human exposure to acid aerosols have been underway
for several years, leading to the need to intercompare measurement
methods for acidic aerosol concentrations. Over the past few years
the U.S. EPA has sponsored several intercomparisons of the methods in
use. Most of these methods are similar; nevertheless, some differ-
ences exist among the samplers, as well as in details of their
operation. The studies have included single component aerosols of
sulfuric acid and ammonium bisulfate, photochemical aerosols, a
sequence of compounds chosen to test for on-filter neutralization by
ammonium nitrate, the addition of alkaline dust, laboratory-spiked
samples, and atmospheric aerosol samples. These studies were
designed to establish both intra- and interlaboratory precision and
to attempt to separate the effects of differences in flow rate,
extraction, and analysis. All major groups that are active in acid
aerosol measurement have participated. These studies have not been
done for the purpose of choosing a best method, but rather to
document the current level of performance and to suggest items for
improvement. The results of these studies are that intralab
precision (stated as the coefficient of variation) is typically 5-10
percent for any lab, while total precision among labs is 10-25
percent. However, in most cases of poorer total precision, the cause
is either a systematic bias between two or more labs or inadequate
low-level performance by one or more labs. These findings suggest
that additional standardization in methodology is warranted.
INTRODUCTION
In 1988 EPA's Clean Air Science Advisory Committee recommended1
that existing measurement methods for acid aerosols be evaluated
because acid aerosols were under consideration for possible listing
as a seventh criteria pollutant. Most of the existing methods are
variations of one basic method in which ammonia gas, which could
neutralize acidity collected downstream, is removed by diffusion to
an acidic wall; the particles are removed by a Teflon filter and
later extracted with a dilute solution of a strong acid and analyzed
by a pH electrode. Another method, developed by Brookhaven National
Lab, uses no denuder for ammonia, but collects the particles on a
quartz filter with later extraction and analysis by Gran titration.
Before beginning the studies it was concluded that the most
appropriate indicator of aerosol acidity is fine particle strong
acidity measured as hydrogen ion by either pH or titration2, and that
because no standard for aerosol acidity existed, accuracy could not
be established by these studies. EPA then sponsored a series of
intercomparisons to judge the relative performance of the methods.
SINGLE COMPONENT AEROSOLS
The first study involved three laboratories sampling from a
651
-------
manifold, which was a 3 m long, 15 cm diameter, Teflon-coated
aluminum pipe. None of the samplers used a particle size classifying
inlet. All three samplers were of the same basic type (annular
denuder to remove ammonia, Teflon filter, and pH determination of
acidity). Sulfuric acid aerosol was generated by nebulization first
at a low concentration for three sampling periods, and then at a high
concentration for three more sampling periods. Each lab operated two
samplers for each experiment so that intralaboratory precision could
be determined. The same design was then repeated using an ammonium
bisulfate aerosol.
There was substantial agreement on acidity among the labs
(Figure 1). Each lab's two values for one experiment were often very
close, while there was a greater difference between labs. One lab was
consistently higher than the median, while another tended to be lower
than the median. The intralab precision averaged 5 percent, while
the total precision across all labs was 14 percent.
MULTICOMPONENT AEROSOLS
The second study included the same three participants and the
same physical apparatus as the first study, but used more complex
test aerosols. First was a test of a procedure to correct for the
possible neutralization of acidity on the Teflon filter by ammonium
nitrate aerosol, which if present will pass the ammonia denuder as
particles and then may dissociate on the Teflon filter to release
ammonia. The design was to sample sulfuric acid aerosol and, without
disturbing the samplers, change the nebulization solution and sample
ammonium nitrate aerosol. Reference samplers which were changed
between the two components provided a measurement unbiased by the
possible artifact. This test was done in triplicate but at only one
level. The second part of this study was to generate an artificial
smog by the photochemical reaction of toluene, S02, and NOX. It was
found that not enough acidity was generated by photochemistry alone
without going to undesirably high concentrations of SO2, so most of
the acidity was added by nebulizing sulfuric acid; nevertheless,
there were possible confounding species present in the aerosol,
including ozone and organic acids. These smog experiments were done
in triplicate at each of two concentrations.
The overall results for these experiments showed that, although
most intralab precisions were similar to those of the single
component aerosols, there was about twice as much total variability
(Figure 2). This seemed to be due to a systematic bias of 30-50
percent between one lab and the other two. In the H2SO4/NH4N03
experiments, only about 25 percent of the acidity was found on the
Teflon filter; therefore, the correction was an important factor.
Two of the labs came very close to the reference values. Thus, in
this test the correction scheme appears to work well.
OUTDOOR SMOG CHAMBER STUDY3
Six laboratories, including five that used the basic method and
one that used the Brookhaven method, participated in the chamber
study held at the University of North Carolina's 190 m3 outdoor smog
chamber. Two experiments of each of the following types were
conducted: sulfuric acid only, photochemical smog with added sulfuric
652
-------
acid, photochemical smog with added sulfuric acid and dust, and dust
followed by photochemical smog with added sulfuric acid. Again, each
group ran two samplers in each experiment. Nebulized sulfuric acid
was added in every photochemical experiment due to the inability of
the photochemistry to generate enough acidity. For the experiments
using dust, General Motors fine natural Arizona dust was generated by
a venturi-based deagglomeration system.
Intralab precision for acidity was about 10 percent. The total
precision was 25 percent for all experiments, but reduced to 12
percent for the four highest level experiments. A graph of precision
versus concentration displays the behavior typical of most analytical
methods (Figure 3) and implies that a minimum loading of about 700
nanomoles H* is needed to avoid the level below which interlaboratory
variance increases dramatically.
EXTRACTION AND ANALYSIS STUDY4
The final study was undertaken to examine the possible major
contributors to imprecision, including extraction and analytical
performance, and to examine the methods' performance on real versus
synthetic samples. By having one group collect the atmospheric
samples, factors such as flow rate and sampler differences could be
eliminated as sources of bias. The first test was of spiked filters
in triplicate at six different concentration levels; each lab had to
extract and analyze its filters. The second test was of atmospheric
samples collected by one laboratory; two or more filters for three
sampling periods were extracted and analyzed by each lab. The third
test was of atmospheric samples that were all extracted by one lab,
with the resulting extract solutions for each day combined, spiked,
and divided among the labs for analysis.
Intralab precision was found to be 5 percent, while total
precision was about 10 percent. The lowest level of spiked filters
shows a large variation among labs, whereas the second and others do
not (Figure 4). This implies that with the current level of
performance, one should try to collect at least 400 nanomoles H+ in
order to get good comparability among labs. The ambient filters all
show good interlab precision, indeed somewhat better than the spiked
filters. This implies that analysis of real atmospheric samples is
no less precise than that of spiked samples. Thus, spiked filters
can be used in conducting quality assurance on an acid aerosol
measurement network. The extract solutions have good interlab
precision also, marginally better than the spiked or ambient samples.
This indicates that there is a identifiable but not major effect of
extraction by the different labs.
CONCLUSIONS
1. Intralaboratory precision expressed as the coefficient of
variation was typically found to be 5-10 percent. The total
precision, which includes both intra- and interlaboratory sources,
typically ranged from 10-25 percent. The higher cases were usually
associated with systematic bias or inadequate low level performance
affecting one or two groups. These differences would probably be
reduced by the adoption of a common sampling and analysis protocol,
such as the one the U.S. EPA has recently published5.
653
-------
2. No statement can be made about the accuracy of the methods for
acidity owing to the lack of an accepted standard. Thus, since all
methods generally showed good internal precision, one method of
measuring aerosol acidity cannot be recommended over another.
3. The correction scheme for on-filter neutralization by ammonium
nitrate worked well under the experimental conditions used.
4. Atmospheric acidity samples have a comparable precision to spiked
samples. This supports the use of spiked filters for quality
assurance of acid aerosol measurement networks.
5. Networks involving different groups should design their sampling
strategy (flow rate and sampling period) so that at least 400-700
nanomoles of H+ are collected at the desired minimum detection limit
for the network.
DISCLAIMER
This paper has been reviewed in accordance with the U. S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication. Mention of
trade names or commercial products does not constitute endorsement or
recommendation for use.
REFERENCES
1. Clean Air Science Advisory Committee, Recommendations for Future
Research on Acid Aerosols. EPA-SAB/CASAC-89-002, U. S.
Environmental Protection Agency, Washington, 1988.
2. R.J. Tropp, Acid Aerosol Measurement Workshop: Final Report.
EPA/600/9-89/056, U. S. Environmental Protection Agency, Research
Triangle Park, NC, 1989.
3. T.G. Ellestad, H.M. Barnes, R.M. Kamens, S.R. McDow, J.E. Sickles,
II, L.L. Hodson, J.M Waldman, S.J. Randtke, D.D. Lane, S.R.
Springston, P. Koutrakis, and G.D. Thurston, "Acid Aerosol
Measurement Method Intercomparisons: An Outdoor Smog Chamber
Study," in Proceedings of the 1991 EPA/ASWMA International
Symposium on Measurement of Toxic and Related Air Pollutants.
VIP-21, Air & Waste Management Association, Pittsburgh, 1991,
pp 122-127.
4. T.G. Ellestad, L.L. Hodson, S.J. Randtke, D.D. Lane, G.D.
Thurston, J.M. Waldman, and P. Koutrakis, "Acid Aerosol Measure-
ment Methods: Studies of Extraction and Analytical Effects," in
Proceedings of the 1992 EPA/A&WMA International Symposium on
Measurement of Toxic and Related Air Pollutants. VIP-25, Air &
Waste Management Association, Pittsburgh, 1992, pp 282-287.
5. Determination of the Strong Acidity of Atmospheric Fine-Particles
(<2.5 urn) using Annular Denuder Technology. EPA/600/R-93/037,
U. S. Environmental Protection Agency, Research Triangle Park, NC,
1992.
654
-------
z
o
5
f
o
8
a
_j
o
it
<
O3UU
3000 -
2500 -
2000 -
1500 -
1000 -
500 -
n -
x x $
x
-ISO ugm~3
H2S04 D D
-i.
-125 d +
D
-100
-75
-50
„,- X § fc
~25 B a *
D RTI
+ HWJ
X HAR
X
X
B
10-
a yg m"3
x * H2S04
t *
a ^
n D
X
, 5-
o
0
-250 J
o
E
c
•z.
-200 2
cc
i-
z
-150 B
z
0
o
-100 5
_J
-50 2
JZ
- n
SULFURIC ACID AMMONIUM BISULFATE
Figure 1. Acidities for single component aerosols.
I
o
U
&
<
1000 -
500 -
n _
- 50 ug m"3
H2S04
n
-M
- 25
+
D RTI
1- RWJ
X HAR
Reference
n
-1-
X
+ +
+
a a
D D
D 5-
pgm"3
+ I H2S°4
+ 4-
*
3~
2-
U 1 ~
R D
* +
+
-125
0)
1
-100 1
o
H
-75 |
1
Z
0
-50 £
Z
lu
-25 §
UJ
j;
Tt
CM
-0
SULFURIC ACID /
AMMONIUM NITRATE
SMOG + SULFURIC ACID
Figure 2. Acidities for multicomponent aerosols.
655
-------
70
50
5 40
ffi 30
a.
24 h EQUIVALENT CONCENTRATION, nmol m"3
50 100 150 200 250
20 -
10 -
300
1000 2000 3000
CHAMBER CONCENTRATION, nmol m"3
Figure 3. Coefficient of variation versus level
In the outdoor chamber study.
4000
1ALL MEAN
Ul l\
I i
Ul
O
O
H
RATIO OF LAB MEAr
o
in
i
D RTI
+ KAN
o NYU
A RWJ
x HAR
x
»10% 0 + D n + D n
fi * j j j r4i^:..:i.:z::.i.:i:4iz:
^ -F o *t" * A ,
"-10% A ^ " •>" A
A a
B
SPIKED
AMBIENT
EXTRACTS
Figure 4. Lab means compared to overall means
In the extraction and analysis study.
656
-------
Speciation and Determination of N-Nitrosodimethylamine and NOZ Species
in Ambient Air by Surface Specific Preconcentration
M.S. Thomson and R.S. Braman
Department of Chemistry
University of South Florida
4202 Fowler Ave, SCA 240
Tampa, Florida 33620
ABSTRACT
Recently there has been an increased interest in the quality of indoor air and its
components. Of particular concern are volatile compounds which are either known or potential
carcinogens, such as the nitrosamines. A method has been developed to speciate and quantify N-
Nitrosodimethylamine in the presence of other NOX compounds. Volatile nitrogen compounds are
separated and preconcentrated on a sequential series of specially coated hollow tubes. Sample
analyses are achieved by sequential thermal desorption followed by conversion of absorbed analytes
to NO which is then detected by a reduced pressure chemiluminescence analyzer. Detector
response is linear with sample size and the current detection limit is approximately 0.25 ng/sample
or 0.05 ng/L for a typical ambient air sample. Preliminary ambient air analysis indicates the
presence of N-Nitrosodimethylamine in concentrations in the fractional to low nanograms/L range
which is consistent with results published using other methods.
INTRODUCTION
Concern regarding the indoor environment has been on the increase. Toxic chemicals have
been monitored and acceptable levels have been determined for some. The family of NOX
compounds has been extensively studied and monitored and recently concern has arisen over
exposure to high levels of the volatile nitrosamines. Many nitrosamines have been studied for years
and many have been found to carcinogenic. Without further investigation into the ambient
concentrations of these compounds, it would not be possible to determine an acceptable level for
the volatile nitrosamines, such as, dimethylnitrosamine, diethylnitrosamine, methyl-ethylnitrosamine,
and nitrosomorpholine.1 Nitrosamines form in air when both NOX and secondary and/or tertiary
amines are present. Rapid formation is favored in dark and in humid conditions. Formation also
occurs during the daytime, however molecular photolysis occurs. Many groups have observed half-
lives for the volatile nitrosamines on the order of 30 minutes in sunlight and 60 minutes on a
cloudy day.2-3'4 Investigation into indoor half-lives has not been documented.
The existing methodologies for nitrossamines utilize mostly GC, MS, GC-MS, HPLC, TEA
(chemiluminescence), or a combination.5'6'7'8 The current literature lists TEA as having the most
accurate detection and the highest recovery of the instrumental techniques. The most reported
sample collection technique involves sampling air and collecting on a sobent, such as carbon,
Tenax, or a Thermosorb cartridge.9 After collection freeze trapping, elution, concentration, and/or
reduction are often involved. The published methods also only identify the nitrosamine, and do not
give any information on additional NOX species. The current methods, while they have good
sensitivity and reproducibility, are time consuming, labor intensive, and costly. Analysis procedures
that monitor more than one NOX species, and that are quicker, easier, and less expensive would be
beneficial.
Braman and coworkers have previously reported on the application of coated hollow tubes
for the preconcentration and collection of volatile species, both NOX and organics.10'11 They have
657
-------
monitored HNO3, HNO2, NO2, NH3, NO2, and NO simultaneously. Cobb et al. used coated hollow
tubes for the analysis of specific atmospheric organics. The development of a more viable technique
for nitrosoamine monitoring is needed. Without additional information and data profiles of the
indoor environment, control and/or removal of a potentially carcinogenic source would not be
possible. The focus of this study was to develop the appropriate combination of tubes and to
employ them for the determination of specific NOX species, as well as, volatile nitrosamines in
indoor air. The hollow tube technique allows for quick NOX speciation, and analysis, as well as
avoid the drawbacks of nitrosamine degradation and artifact formation.9'12
EXPERIMENTAL
Instrumental
A Thermo Electron Model 14B/E chemiluminescence analyzer was used and operated in
the manual mode. The automatic functions were disabled and all compounds under analysis were
converted to NO previous to entering the instrument. Calibration of the detector response was done
using either standard calibration gases and/or for the nitrosamines, a permeation tube gas standard.
The critical instrumental parameters were sample chamber vacuum of 25 torr, He carrier gas flow
of 100 ml/min, and an inboard flow of 75 ml/min. With these flow rates the detection limit was
0.006 nanograms as N with a reproducibilty of sample response of + /- 95 %.
Tube Preparation
All chemicals used in the tube preparation were reagent grade and all tubes were made of
6 mm O.D. X 30 cm. quartz. Each tube was etched 5 min. in a 50% HF solution, rinsed, then
boiled in 40% NaOH. After the surface prep, the tubes were rinsed and boiled in DI for 1 hr,
rinsed and stored under DI until needed for coating. Previous to coating the tube was oven dried
and cooled. Without a through surface prep., coatings did not adhere well, blank properly, function
correctly, or last long.
Iron oxide coating. An iron hydroxide gel was made using FeCl3 or Fe(NO3)3 and NaOH
and the mixture was drawn through the tube to give an even wetting of the interior surface. The
tube was dried with He carrier gas flowing through. After drying the tube was heated to approx.
500 °C with O2 passing through. This process expelled both C12 and and/or NOX and a orange-red
crystalline iron oxide coating was formed.
Copper iodide coating. Copper iodide coated tubes were prepared by making a slurry of
Cul from CuCl and KI and pulling this mixture through the tube. The tube was first dried then
heated to 300 - 400 °C, while passing He gas. During the heating some sublimation of I2 may be
observed. The resulting tube has a brittle, translucent tan-white coating.
Carbon and acid doped carbon coatings. Carbon tubes were made by pulling smoke from
a benzene fire through the tube until no bare patches were noticed on the tube walls. For acid
doping, a 1:1 mixture of H3PO4/H2PO4" was quickly pulled through the tube, wetting the carbon
surface, but not removing it. He carrier gas is again used for drying and blanking at 400 - 500 °C.
The heating was done until no more oil-like residue was observed.
Cobalt oxide coating. Tubes coated with cobalt oxide were made by either using a saturated
Co(NO3)2 or by making a cobalt hydroxide gel from Co(NO3)2 and NaOH. As before, the tube was
dried with He and after setting the surface the tube was heated to 500 °C with O2 passing through.
The resulting coating was a glassy blue-black.
658
-------
Conditioning. All tubes required significant conditioning before reproducibility and correct
compound discrimination was observed. After blanking, gas standard samples were cycled on and
off each tube. If tubes had a variance of >5% in their reproducibility, the tube was rejected,
cleaned, and recoated. The cycling was necessary to set the active surface. The surfaces used are
thought to be Fe2O3, Cul, amorphous carbon, and Co2O3. These surfaces, as well as some others,
are presently under further investigation and improvement.
Standards
A permeation tube gas standard for Nitrosodimethylamine, NDMA, was assembled. The
source consisted of a 30 °C water bath with liquid NDMA in a stainless holder with a teflon
diaphragm. A diffusion hole was located in the diaphragm and source was placed hi an additional
holder and submersed. Air was flowing through approximately 25 ft. of tubing before the standard
to ensure a constant temperature. The air flow rate over the standard is > 8 L/min and sampling
was done using a T" port between 0.5 1.0 L/min. The source response was checked versus
standard calibration gases of NO and NO2, and was in good agreement. The source output was
35.63 +/- 3.76 ng/min as total N for over 8 months. The source for HNO2 has previously been
described by Braman et al.13
Air Sampling
Air, both indoor and outdoor, was sampled using a small air pump, that had a previously
calibrated rotometer attached. The flow rate during sampling was between 0.5 - 1 L/min with the
total volume to be analyzed at < 10L. This allowed for duplicate samples and for indoor sampling,
limited inconvenience to the businesses being sampled.The tubes were connected with teflon
connectors and in the order of iron oxide, copper(I) iodide, carbon (or H+ doped carbon), and
cobalt oxide. For transport the tubes were disassembled and capped.
Analysis
Each tube was analyzed individually using thermal desorption with a either a quartz, gold
coated quartz, or no catalyst bed. The catalyst bed temperature ranged between 300 - 500 °C
depending on the output desired. In line after the cat. bed was a H2O bubbler followed by an
empty bubbler and air make-up 'T'. Each individual tube required less than 10 minutes to
completely analyze and reblank and prepare for the next sample.
RESULTS AND DISCUSSION
Individual Tube Response
The copper(I) iodide and cobalt(III) oxide tubes have been characterized in previous work.10
The copper(I) oxide tube collects HNO3, HNO2, and NO2 and in addition to these compounds the
cobalt(III) oxide tube collects NO. The collection of both volatile nitrosamines and amines has not
yet been examined, however collection of these compounds is expected. For this study, instead of
using the mixed potassium-iron oxide tube developed by Cantera10, an iron oxide tube was
substituted. This tube collected HNO3, HNO2, and trace amounts of NO2. The response was in
agreement with the mixed oxide tube, however the conditioning and blanking procedure were far
superior to the other tube. It is thought that there is a combination of chemi and physisorption
occurring on the coated tube surface. The conditioning of the tube most likely either causes a
charge migration of an active species, and/or causes the surface to crack and develop the
topography necessary to hold the analyte molecules. Determination of the active species and
examination of all of the tube surfaces is presently under way.
659
-------
From previous work,10 the catalyst of choice, used in the analysis stream, was gold. For most
of the collection surfaces, a catalyst bed was necessary to ensure complete conversion of the
thermally desorbed species to NO. The gold surface provided adequate reduction sites for the NO^
however, it appeared to be a collector of particulate and organics. The output from this cat. bed
led to peak tailing. When using this catalyst, the surface buildup prohibited a clean release of the
molecules of interest. For this study, superior, cleaner results were achieved using a 15 cm heated
quartz tube. The response using quartz gave sharper peaks with a quick return to the baseline. It
is thought that the quartz surface is smoother and has enough active sites for any needed
decomposition of NOX analytes to NO. All tubes were tested with each surface and with standards.
The comparison between the two types of heated catalyst tubes yielded comparable results.
Both types of carbon coated tubes were found to collect all of the before mentioned NOX
species, except NO, as well as the volatile nitrosamines and volatile amines. Using the permeation
tube NDMA standard, the carbon and H* doped carbon tubes(H-C) were characterized. The
capacity determined previously by Cobb11 was > 500 ng for atmospheric organics. The base capacity
for NDMA, and volatile nitrosamines has not been tested yet, due to the fact that the size of the
samples and amounts recorded for this species are at least one order of magnitude lower. By
employing a dual tube analysis, the scrubbing efficiency for NDMA, for both types of carbon tubes,
was found to be 93.00 +/- 2.54 % in the low ng/L range. Based on the Gormley-Kennedy
equation14, which predicts the efficiency of hollow tubes used for chemisorptive preconcentration,
a calculated value of 94.6 % was found. The experimental and predicted values were within the
standard deviation, and in good agreement. Testing was done to determine the decomposition
lifetime of NDMA when absorbed onto both the C and H-C surface. No detectable change was
observed after 20 hours of storage, however by 66 hours of storage a 5 % loss was noticed.
Catalyst testing. Scavenger solutions and catalyst bed effects were also examined . It has
previously been determined that the weakest bond is the N-N bond that releases the nitrosyl
portion of the molecule. Experiments were run to determine if upon thermal desorption and testing
whether the nitrosyl portion and/or the amine portion were converted quantitatively to NO. Using
the published scavenger solution15 method for analyzing nitrosamines in solution, a loss of standard
was observed, and only 81 - 87 % of the NO was detected. It was found that if the catalyst bed
were omitted, upon thermal desorption from the C or H-C surface, 95 % of the NO group was
detected. Using a heated quartz bed, 98 % of the NO group was observed. If a gold coated bed was
used, not only was response recorded for the NO group, but also for the amine portion. In the low
ng/L, it was determined that 98 - 99 % of the total N was detected. Also, testing of volatile organic
amines, it was found that upon desorption, and only if the gold catalyst bed was used, NO was
produced quantitatively.
Comparison between the C and H-C tube response for NDMA showed that the C tube
desorbed the amine portion, or amines at a slightly lower temperature then the HC tube. This
would imply an acid/base type attraction between the amine/amine portion of the analytes and the
tube coating. With the H-C tube and gold catalyst bed, two unresolved peaks are often observed
for the NDMA molecule. This would also support two types of surface attractions. Attempts to
resolve and quantify these peaks are still underway. The reproducibility on both carbon surfaces
was < +/- 5 %, in the low ng sample size. The 5 % criteria was the limit for initial tube rejection,
so the deviation observed in reproducibility may due mostly to tube manufacture. The estimated
detection limit for NDMA using the C and H-C tubes was found to be 0.0064 ng as N. Based on
the cat. bed response, tandem air samples are run at each location. The H-C tube from each stack
was analyzed using a quartz tube catalyst, which gave only an NO response from the volatile
660
-------
nitrosamines. The carbon tube was analyzed with a gold tube catalyst, and both portions of the
nitrosamine as well as any amine responded. By difference, the level for total nitrosamines and
total amines can be determined.
Air Analysis
Indoor air profiles are presently being done at this laboratory. Central Florida homes and
businesses are being tested. Sampling is done both indoor and outdoor at all locations because
many Florida buildings are not tight. In the cooler months, many building are open, and air flow
is due to windows and fans, not heating and A/C. The preliminary findings, located in Table I,
show levels of NOX and volatile nitrosamines similar to previously published data. Of the locations
tested, indoor and outdoor profiles are also similar, as expected. One area expected to be high in
amines, medical facility, kennel, gave high amine results. This high amine area was an example of
the analysis technique for species separation using the carbon tubes.
Table I: Indoor and outdoor air results for NO,, volatile nitrosamines, and organic amines.
Four tube system, reported as ng/L
Phosphate Plant A
3/09/933 3/23/93"
HNO2
NO2
NO
N
(total N
in R2N2O)
N
(total N
in amines)
lab1
med.2
out.3
lab
med.
out.
lab
med.
out.
lab
med.
out.
lab
med.
out.
3.765 +/- 0.817
2.840 +/- 0.773
3.326 +/- 0.142
8.562 +/- 0.555
9.561 +/- 5.195
6.903 +/- 1.568
5.198 +/- 1.422
4.554 +/- 1.011
2.528 +/- 0.453
1.193 + /- 0.078
0.791 +/- 0.041
0.864 +/- 0.057
< 0.001
8.100 +/- 0.534
< 0.001
6.124 + /- 1.949 surgery
kennel
outside
7.415 +/- 2.459 surgery
kennel
outside
2.284 +/- 1.021 surgery
kennel
outside
1.682 +/- 0.584 surgery
kennel
outside
1.485 +/- 0.445 surgery
kennel
outside
Medical Facility A
4/08/93c
14.98 +/- 1.195
9.552 + /- 1.072
13.91 + /- 2.022
57.33 +/- 10.64
36.85 +/- 5.141
41.81 +/- 3.950
15.57 +/- 0.586
31.08 +/- 2.602
10.89 +/- 1.734
3.367 +/- 0.038
11.30 +/- 0.126
4.526 +/- 0.051
< 0.001
12.63 +/- 0.141
0.210 +/- 0.002
1 - Q/C wet chemistry laboratory
2 - in house medical facility
3 - outside
a - sunny, breezy day in the 70's, A/C on
b - sunny, high 70's, no A/C in buildings
c - partial clouds, high 70's, A/C on
661
-------
CONCLUSION
The use of sequential hollow tube set of iron oxide, copper(I) oxide, carbon or acid doped
carbon, cobalt(III) oxide, has been used to successfully collect and concentrate ambient air species
of NO,, volatile nitrosamines, and organic amines. This tube set, in conjunction with an optimized
chemiluminescence detector, gives an accurate profile of the examined species. This analysis
scheme circumvents the time consuming sample collection, preparation, and analysis, of other
methods. Determination of the ambient nitrosamine concentration is necessary to determine
baseline levels, as well as, defining acceptable levels. This technique has the potential for
monitoring both indoor and outdoor air in the low ng/L level.
REFERENCES
1. D. Grosjean, "Atmospheric Chemistry of Toxic Contaminants. 6. Nitrosamines: Dialkyl
Nitrosamines and Nitrosomorpholine," J. Air Waste Manage. Assoc.. 41: 306-311 (1991).
2. P.L. Hanst, J.W. Spence, M. Miller, "Atmospheric Chemistry of N-Nitrosodimethlyamine,"
Environ. Sci. Technol.. 11: 403-440 (1977).
3. K. Bretschneider, J. Matz, "Occurance and Analysis of Nitrosamines in Air," LARC. 14:395-
399 (1976).
4. C.R.C. Lindley, J.G. Calvert, J.H. Shaw, "Rate studies of the Reaction of the (CH3)2N
Radical with O2, NO, and NO2," Chem. Phys. Lett.. 67: 57-62 (1979).
5. C. V. Cooper, "Gas Chromatographic/Mass Spectrometric Analysis of Extracts of Workplace
Air Samples for Nitrosamines," Am. Ind. Hyg. Assoc. J.. 48(3): 265-270 (1987).
6. D.P. Rounbehler, J. Reish, D.H. Fine, "Some Recent Advances in the Analysis of Volatile
N-Nitrosoamines," in Proceedings of the Vlth International Symposium on N-Nitroso
Compounds. 31, IARC, Lyon, 1979, 403-416.
7. H.J. Issaq, J.H. McConnell, D.E. Weiss, D.G. Williams, "High Performance Liquid
Chromatography Separations of Nitrosamines. I. Cyclic Nitrosamines," J. Liq. Chromatog..
9(8): 1783-1790 (1986).
8. H.J. Issaq, M. Glennon, D.E. Weiss, G.N. Chmurny, J.A. Saavedra, "High Performance
Liquid Chromatography Separations of Nitrosamines. II. Acyclic Nitrosamines," J. Liq.
Chromatog.. 9(12): 2763-2779 (1986).
9. D.P. Rounbehler, J.W. Reisch, J.R. Combs, D.H. Fine, "Nitrosamine Air Sampling Sorbents
Compared for Quantitative Collection and Artifact Formation," Anal. Chem.. 52: 273-276
(1980).
10. R.S. Braman, M.A. de la Cantera, "Sequential Hollow Tube Preconcentration and
Chemiluminescence Analysis System for Nitrogen Oxide Compounds in Air," Anal. Chem..
58: 1537-1541 (1986).
11. G.P. Cobb, R.S. Braman, "Carbon Hollow Tubes as Collectors in Thermal Desorption/Gas
Chromatographic Analysis of Atmospheric Organic Compounds," Anal. Chem.. 58:2213-2217
(1986).
12. J.F. Walling, J.E. Bumgarner, D.J. Driscoll, CM. Morris, A.E. Riley, L.H. Wright, "Apparent
Reaction Products Desorbed from Tenex used to Sample Ambient Air," Atmos. Environ..
20(1): 51-57 (1986).
13. R.S. Braman, M.A. de la Cantera, "Sublimation Sources for Nitrous Acid and Other
Nitrogen Compounds in Air," Anal. Chem.. 58: 1533-1537 (1986).
14. P. Gormley, M. Kennedy, Proc. Roy. Ir. Acad.. Sect. A.. 52(A): 163-169 (1949).
15. L.R. Dix, S.M.N.Y.F. Oh, D.L.H. Williams, "Denitrosation of Nitrosamines- a Quantative
Study. Reactions of N-Methyl-N-nitrosoaniline, N-Nitrosoproline, Dimethylnitrosamine and
N-nitrosoarcosine," J. Chem. Soc. Perkin Trans.. 2: 1099-1104 (1991).
662
-------
MEASUREMENT OF AMBIENT PARTICLE PHASE ORGANIC ACIDITY
USING AN ANNULAR DENUDER-FILTER PACK SYSTEM
Joy Lawrence and Petros Koutrakis
Harvard School of Public Health
665 Huntington Ave., Boston, MA 02115
ABSTRACT
Little is known about the extent to which organic acids contribute to aerosol acidity. This
paper presents the results of a field study to determine the particle phase concentrations of organic
acids using an annular denuder-filter pack system.
Filter samples were cut in half and analyzed first by pH meter to determine strong and total
weak acidity. Samples were titrated to beyond pH 9, and total strong and total weak acidity were
determined by modified Gran plot. Particle phase strong acidity was found to range from ND to
398 nmole/m3. Particle phase total weak acidity was found to range from ND to 167 nmole/m3.
The concentrations of aerosol strong and weak acidity were found to be highly correlated over the
study period (R2=0.93, p <0.001).
INTRODUCTION
The concentration of ambient particulate matter increases during photochemical smog
episodes. Analysis of the organic photochemical aerosols sampled during smog episodes has
identified the presence of carboxylic acids (1-16) and dicarboxylic acids (8, 12-14, 16-23). Mono-
and di-carboxylic acids have also been identified in ambient particulate matter sampled at rural area
receptor sites for regional photochemical pollution from the Ohio River Valley (24, 25).
Monocarboxylic acids have been found ambient aerosols sampled in residential areas (26), rural
areas (5, 27), and remote areas (2, 28, 29).
Carboxylic acids of twelve or more carbons are commonly found in the particulate phase.
The pure compound vapor pressures of C14-C18 acids and C5-C6 diacids are low enough that they
should be found exclusively in particulate phase at ambient temperatures (30). Organic acids up
to C10 have been found in gas samples (31), and organic acids C10 and larger have been identified
in particulate matter (12). The only dicarboxylic acid found in gas samples has been oxalic acid,
reported in several studies (17, 32). Dicarboxylic acids up to C16 have been identified in ambient
aerosol samples (14).
Data are limited, but organic acids seem to account for as much as 18-20% of the organic
fraction of ambient aerosols in urban areas during photochemical smog episodes (1, 5, 15). Smog
chamber studies (33-37) have investigated the formation and growth of aerosols from atmospheric
organic systems (with NOX and/or O3), and found that some classes of organic compounds form
aerosols quite efficiently. Diolefins and cycloalkenes, in particular, are potent organic aerosol
precursors (33-35, 38). The major components of organic aerosols formed in smog chambers from
diolefins and cyclic olefins have been found to be dicarboxylic acids and oxo-carboxylic acids.
There are several known and suggested primary sources of ambient particulate phase organic
acids. Biogenic sources of ambient particulate phase long-chain carboxylic acids include vascular
plant waxes and microbes associated with plant lipids (10-12, 15, 28), the microbiota of marine and
lacustrine environments (12, 15) and soil (6). Long chain organic acids are also emitted directly by
operations involving animal fats and soaps (18), in substantial quantities by charbroiling and frying
of meats (39), and by the combustion of organic materials (6, 12, 40).
Long chain carboxylic acids have several suggested secondary sources. The most generally
accepted (37, 38, 41, 42) is the ozone oxidation of olefins, forming Criegee biradicals which
isomerize to form an acid. A source of the long chain olefin has been suggested to be plant lipid
material (10, 28, 33, 43). Another suggested source of long chain carboxylic acids is the ozone
663
-------
oxidation of aldehydes, via a Criegee intermediate (37).
Dicarboxylic acids have fewer primary sources. The major primary source has been
determined to be motor vehicle exhaust (17, 44). Significant amounts of dicarboxylic acids are also
emitted by charbroiling of meat (39). Plants and soil are not important sources of ambient
particulate dicarboxylic acids (17).
It has been reported that dicarboxylic acids are found almost exclusively in submicron
particles (23), suggesting a photochemical origin. Submicron aerosols are strongly associated with
secondary origin, especially for carbonyl and dicarbonyl compounds (34,38,44). Several studies (18,
21, 22) have observed strong diurnal variation of particulate phase dicarboxylic acids during long
range transport of photochemical pollution from urban areas, well correlated with the ozone
concentration, which also supports a photochemical origin.
There are several proposed mechanisms for the in situ formation of dicarboxylic acids.
Norton, et al. (32) measured oxalic acid in both gas and particulate phase in Colorado air. They
suggested the most likely source would probably be the gas or aqueous phase oxidation of glycoxal
(CHOCHO) by hydroxy radical. Other possible secondary sources of dicarboxylic acids include the
ozone oxidation of unsaturated fatty acids emitted from vascular plant waxes (10, 14, 43) and soils
(14), and the oxidation of diolefins (18, 34, 35).
The oxidation of cyclic alkenes by ozone (or by hydroxyl radical) is a widely accepted
secondary source of dicarboxylic acids (18, 21, 22, 34-36, 45). The reaction proceeds with ozone
addition across the double bond, followed by ring opening. The product of the reaction is a
difunctional molecule, with an aldehyde end, and a Criegee biradical at the other end. The
biradical isomerizes to carboxylic acid, and the aldehyde is then further oxidized to an acid group.
The major source for particulate phase organic diacids has been shown to be photochemical
formation (22), accounting for up to 70% of the particulate diacids. The long chain fatty acids,
however, display a strong even-to-odd carbon preference in ambient particulate matter, clearly
indicating their biogenic origin (10, 43). The importance of aerosol weak acidity is still
uncertain. Few measurements of aerosol weak acidity have been made. Little is known about the
contribution of organic acids to aerosol acidity. Ferek, et al. (25) found that weak acidity comprises
approximately 26% (± 12)of the aerosol total acidity in rural area receptor sites for photochemical
pollution from the Ohio River Valley. They analyzed the samples by ion chromatography (for low
molecular weight dicarboxylic acids) and found that dicarboxylic acids of six or fewer carbons
accounted for 33-127% of the aerosol weak acidity.
Human health effects of particulate weak (or organic) acidity have not been investigated;
there is no evidence that carboxylic acids or dicarboxylic acids are mutagenic (46) and no known
chronic toxicity associated with a long term low level exposure (47). The primary interests in
studying particulate organic acids have included exploring their visibility effects (1) and
understanding their tropospheric chemistry.
MEASUREMENT OF TOTAL STRONG AND TOTAL WEAK ACIDITY
Theoretically, if there is only one acid present, and its dissociation constant is known, the
concentration of the acid can be determined from the pH of the acidic solution by a simple
calculation. Practically, however, the accuracy of this method is not very satisfactory (48). When
more than one acid is present or the K, of the acid is unknown, titration is necessary to determine
concentration.
The potentiometric titration of a weak acid with a strong base can be difficult; the inflection
point of the pH v. titrant added curve can become hard to identify. Further, it has been shown that,
for sufficiently weak acids, the inflection point does not correspond well with the equivalence
volume (49). The most commonly used method to determine the equivalence point in
potentiometric titration is probably the Gran plot (50) or a modified form of the Gran plot. The
664
-------
Gran plot is a graphical linear extrapolation method, which performs excellently when used in the
titration of a moderately weak acid with a strong base (51).
The Gran plot is difficult to apply to titration of weak acids because Gran's approximations
result in curvature near the equivalence point of very weak acids, as well as at the low pH end of
the titration of a moderately strong acid (48, 51-53). To deal with the curvature, Johansson (48)
modified Gran's equations and developed a similar graphical (but still approximate) method that
greatly reduces the curvature in these cases. The advantage of Johansson's modified Gran plot is
that it can be used with mixtures of strong and weak acids, and it has been used successfully in
other studies (25, 54).
PHILADELPHIA FIELD STUDY
Philadelphia is a large metropolitan area, located in the middle of the northeastern seacoast.
During the summer months, the city's air quality is affected by regional photochemical pollution.
The Philadelphia Aerosol Acidity Characterization Study (PAACS) is the first year/ first
city of a multi-year Metropolitan Aerosol Acidity Characterization Study. The program is a joint
effort of US EPA Atmospheric Research and Exposure Assessment Laboratories and HSPH
Environmental Science and Engineering Program to investigate the formation of and human
exposure to acid aerosols and other related pollutants (55). The PAACS is a year-long air pollution
monitoring study designed to investigate the spatial variation of acidity, sulf ates, ammonia, and mass
concentrations in an urban area. During the summer months, intensive monitoring was performed
at one site (N/E Airport), and included additional measurement of elemental carbon concentration,
meteorological conditions, continuous HjSC^/SC},2", particle size distribution (episodic only),
aldehydes (episodic only), and organic acids (gas and particulate phase).
Between June 28 and August 28, 1992, day time and night time organic acid samples were
collected daily at the N/E Airport site. Day time samples were collected between 12:00 and 18:00
EDT, and night time samples were collected between 18:00 and 12:00 EDT. The organic acid
samplers were located 1m above the roof of a trailer, a total of 4m above the ground.
The particulate phase weak acidity during the regular daily sampling was measured using
an annular denuder-filter pack system. One KOH coated annular denuder was used to trap gas
phase organic acidity, and a citric acid coated annular denuder was used to trap NH3, which would
otherwise neutralize particulate phase acidity collected downstream on the filter. The sampler's
impactor is designed to remove particles of aerodynamic diameter greater than 2.1 nm at a sampling
rate of 10 Lpm, and the filter collects the particles smaller than 2.1 ion. The performance of the
K°H coated annular denuder to collect gas phase organic acids has already been described (56).
When returned to the home lab, the filters were cut in half in an ammonia-free hood. Half
of the filter was sealed in a glass 1C vial and stored in the refrigerator at 4° C (with 5 pL methylene
chloride added to reduce biological activity) until ready for extraction, derivitization, and GC/FID
analysis. These analyses will be discussed in a subsequent paper. The other half of the filter was
analyzed (same day) for total strong and total weak acidity.
The half filter was wetted with ethanol and extracted under ultrasonic agitation with 1.7 mL
of solution (9.6 x 10'5 M KC1O4 to reduce CO2 solubility, 0.04 M KC1 to maintain constant ionic
strength, and 3.2 % ethanol to increase solubility of organic acids). The vials were sonicated for
a total of 15 minutes. The pH meter was calibrated with standards at 4.00 and 7.00. A calibration
curve was prepared using a series of 7 analytical standards containing both strong (H2SO4) and
weak (a mixture of even-carbon fatty acids Cg-C^, and diacids C2-C10) acid, covering a range of 0
to 530 jimol H+/L in both strong and weak acid. The initial pH of each standard was measured,
and the standard was titrated beyond pH 9 with 10 iiL additions of 0.004 M NaOH. The titration
was conducted under a purge of CO2-free air, and using a micro magnetic-stirring vane to reduce
the streaming potential of the electrode. Total strong and total weak acid concentrations were
determined using a modified Gran plot (48, 50). The strong acid calibration curve generally had
665
-------
a slope very close to 1.0, an intercept close to zero, and an R2> 0.995. The weak acid calibration
curve generally had an intercept close to zero, a slope of 0.75 to 0.80, and an R2> 0.995.
Samples were extracted and analyzed in the same manner. Total strong and total weak acid
concentrations were corrected using the standard curve. Blank filters and doped filters were used
to provide information about the efficiency of the analytical technique. The filters were doped using
mixtures of the same acids used in the calibration curve. The recovery of strong acid from doped
filters was found to be > 95%, and the recovery of total weak acid from the doped filters was found
to be >80%. A plot of the calculated values for total strong and total weak acidity against then-
known values yields intercepts close to 0 (0.000002 for strong acid and 0.00013 for weak acid) and
slopes relatively close to 1 (1.096 for strong acid, and 0.799 for weak acid); the measured values are
well correlated with their known concentrations (R2 values of 0.997 and 0.996 for strong and weak
acids, respectively).
RESULTS OF THE PHILADELPHIA FIELD STUDY
Figure 1 contains the concentrations of total strong aerosol acidity measured in Philadelphia.
The daytime strong acid concentration ranged from ND to 398 nmol H+/m3, and averaged 38 (±
53). The nighttime concentration ranged from ND to 63 nmol H+/m3, and averaged 19 (± 14).
The overall average was 38 (±53) nmol H+/m3. The day-to-night difference in aerosol strong
acidity concentration was found to be significant (p= 0.048) by a paired sample t-test.
Figure 1 shows the concentrations of total weak aerosol acidity measured in Philadelphia.
The daytime paniculate weak acid concentration ranged from 13 to 167 nmol H+/m3, with an
average of 24 (±24). The nighttime concentration ranged from ND to 15 nmol H+/m3, averaging
6 (±2). The overall average of the observed concentration was 15 (± 19). The daytime
concentration of paniculate weak acidity was found to be significantly higher than the nighttime
concentration using a paired sample t-test (p< 0.001).
Figure 2 presents the ratio of weak to total aerosol acidity. The daytime ratio was found
to range from 0.19 to 0.86, with an average of 0.39 (±0.18). The nighttime ratio ranged from 0.18
to 0.84, with an average of 0.30 (± 0.13). The overall average weak to total acidity ratio was 0.36
(±0.19). Weak acidity was found to be a larger proportion of the total acidity during the day than
at night. The difference was found to be significant, using a paired sample t-test (p< 0.001).
Figure 3 shows the simultaneous day and night concentrations of total strong and total weak
aerosol acidity measured in Philadelphia. Total strong and total weak acidity were found to be
highly correlated (R= 0.93) over the period of the study. Daytime and nighttime concentrations
were correlated equally strongly, though the slope of the regression line for the nighttime
concentrations is smaller than that for the daytime concentrations.
ACKNOWLEDGEMENT
This work was supported by the Electric Power Research Institute (RP1630-59). We would
like to acknowledge the project manager Mrs. Mary Ann Allan for her contribution. Also special
thanks to Mark Davey and Robert Surh for their assistance in field work.
REFERENCES
1. Grosjean, D.; Freidlander, S. K. /. A. P. C. A. 1975 25(10), 1038.
2. Cautreels, W.; Van Cauwenberghe, K.; Guzman, L. A. So. Tot. Env. 1977 8, 79.
3. Cautreels, W.; Van Cauwenberghe, K. Atmos. Env. 1978 12: 1133.
4. Gordon, R. J.; Trivedi, N. J.; Singh, B. P.; Ellis, E. C. Env. Sci. Tech. 1988 22(6), 672.
5. Ketseridis, G.; Hahn, J.; Jaenicke, R.; Junge, C. Atmos. Env. 1976 10, 603.
6. Matsumoto, G.; Hanya, T. Atmos. Env. 1980 14, 1409.
7. Mazurek, M.; Simoneit, B.R.T.; Cass, G.R.; Gray, HA. Int. J. Env. Anal, Chem. 1987 29, 119.
8. O'Brien, R. J.; Holmes, J. R.; Bockian, A. H. Env. Sci. Tech. 1975 9(6), 577.
9. Schuetzle, D.; Crittenden, A. L.; Charlson, R. J. /. A. P. C. A. 1973 23(8), 704.
666
-------
10. Simoneit, B. R. T; Mazurek, M. A. Amos. Env. 1982 16(9), 2139.
11. Simoneit, B. R. T.; Sheng, G.; Chen, X.; Fu, J.; Zhang, J.; Xu, Y. Atmos. Env. 1991 25A(10), 2111.
12. Stephanou, E. G. FKS. J. Anal. Chem. 1991 339, 780.
13. Wauters, E.; Vangaever, F.; Sandra, P.; Verzele, M. /. Chmmat. 1979 170, 133.
14. Yokouchi, Y.; Ambe, Y. Annas. Env. 1986 20(9), 1727.
15. Hahn, J. Ann. N. Y. Acad. Sci. 1980 338, 359.
16. Daisey, J. M. Ann. N. Y. Acad. Sci. 1980 338, 50.
17. Kawamura, K.; Kaplan, I. R. Env. Sci. Tech. 1987 21(1), 105.
18. Cronn, D. R.; Charlson, R. J.; Knights, R. L.; Crittenden, A. L.; Appel, B. R. Atmos. Env. 1977 11, 929.
19. Appel, B. R.; Hoffer, E. M.; Kothny, E. L.; Wall, S. M.; Haik, M.; Knights, R. L. Env. Sci. Tech. 1979 13, 98.
20. Grosjean, D.| Van Cauwenberghe, K.; Schmidt, J. P.; Kelley, P. E.; Pitts, J. N. Jr. Env. Sci. Tech. 1978 12, 313.
21. Satsumabayashi, H.; Kurita, H.; Yokouchi, Y.; Ueda, H. Tellus 1989 41B, 219.
22. Satsumabayashi, H.; Kurita, H.; Yokouchi, Y.; Ueda, H. Atmos. Env. 1990 24A(6), 1443.
23. Schuetzle, D.; Cronn, D.; Crittenden, A. L.; Charlson, R. J Env. Sci. Tech. 1975 9, 838.
24. Barkenbus, B. D.; MacDougall, C. S.; Griest, W. H.; Caton, J. E. Atmos. Env. 1983 17(8), 1537.
25. Ferek, R. J.; Lazarus, A. L.; Haagenson, P. L.; Winchester, J. W. Env. Sci. Tech. 1983 17(6), 315.
26. Cautreels, W.; Van Cauwenberghe, K. /. Chmmat. 1977 131, 253.
27. Yokouchi, Y.; Ambe, Y. Atmos. Env. 1985 19(8), 1271.
28. Simoneit, B. R. T. /. Atmos. Chem. 1989 8, 251.
29. Simoneit, B. R. T.; Cardoso, J. N.; Robinson, N. Chemosphere 1990 21, 1285.
30. Tao, Y.; McMurry, P. H. Env. Sci. Tech. 1898 23, 1519.
31. Kawamura, K.; Ng, L. L.; Kaplan, I. R. Env. Sci. Tech. 1985 19, 1082.
32. Norton, R. B.; Roberts, J. M.; Huebert, B. J. " Geophys. Res. Let. 1983 10(7), 517.
33. Wang, S. C.; Paulson, S. E.; Grosjean, D.; Flagan, R. C.; Seinfeld, J. H. Atmos. Env. 1992 26A, 403.
34. Grosjean, D.; Freidlander, S. K. Adv. Env. Sci. Tech. 1980 Vol. 9 pp 435-473.
35. O'Brien, R. J.; Holmes, J. R.; Bockian, A. H. Env. Sci. Tech. 1975 9(6), 568.
36. Hatakeyama, S.; Ohno, M.; Weng, J.; Takagi, H.; Akimoto, H. Env. Sci. Tech. 1987 21(1), 52.
37. Martinez, R. I.; Herron, J. T.; Huie, R. E. /. Am. Chem. Sac. 1981 103, 3807.
38. Grosjean, D. Atmos. Env. 1992 26A(6), 953.
39. Rogge, W. R; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Env. Sci. Tech. 1991 25, 1112.
40. Cautreels, W.; Van Cauwenberghe, K. Water Air Soil Pollut. 1976 6, 103.
41. Finlayson-Pitts, B. J.; Pitts, J. N. IT. Adv. Env. Sci. Tech. 1977 Vol. 7 pp 75-162.
42. Herron, J. T.; Martinez, R. I.; Huie, R. E. Int. J. Chem. Kin. 1982 14, 201.
43. Simoneit (1991)
44. Pickle, T.; Allen, D. T.; Pratsinis, S. E. Atmos. Env. 1990 24A(8), 2221.
45. Hatakeyama, S.; Tanonaka, T.; Weng, J.; Bandow, H.; Takagi, H.; Akimoto, H. Env. Sci. Tech. 1985 19, 935.
46. Butler, J. P.; Kneip, T. J.; Daisey, J. M. Atmos. Env. 1987 21(4), 883.
47. Lewis, Richard J. (1992) Sax's Dangerous Properties of Industrial Materials. Eighth Edition, Volumes II and III,
Van Nostrand Reinhold, NY.
48. Johansson, A. Analyst 1970 95, 525.
49. Meites, L.; Goldman, J. A. Anal. Chim. Ada 1963 29, 472.
50. Gran, G. Analyst 1952 77: 661-671.
51. Ingman, F.; Still, E. Talanta 1966 13, 1431.
52. Midgely, D.; McCallum, C. Talanta 1974 21, 723.
53. Macca, C. Fres. J. Anal. Chem. 1990 336, 29.
54. Keene, W. C.; Galloway, J. N.; Holden, J. D. Jr. /. Geophys. Res. 1983 88(C9), 5122.
55. Harvard School of Public Health, Exposure Assessment and Engineering Program (1993) "Metropolitan Aerosol
Acidity Characterization Study Year 1: Philadelphia, PA Quality Assurance Project Plan," Prepared for U. S.
Environmental Protection Agency Atmospheric Research and Exposure Assessment Laboratory.
56. Lawrence, J. E.; Koutrakis, P. 1992 In Manuscript, Submitted to Environmental Science and Technology.
667
-------
Figure 1o
Figure 1b
Daytime
Nighttime
Figure 1 (a) Paniculate Total Strong Acidity and (b) Paniculate Total Weak Acidity (nmol/m3) Measured at the Northeast Airport,
Philadelphia, PA, Summer 1992.
Day11 me
Nlahttlme
c/ s\f ^V f\' *\f /\' s\' s\f si' *\f *\S
<§>
-------
Session 18
Indoor Air Quality
-------
Ecology of Fungi in Buildings: Relationship to Indoor Air Quality
Sidney A. Crow, Donald G. Ahearn, Judith A. Noble, Munshi Moyenuddin, and Daniel L. Price1.
Department of Biology, Georgia State University, University Plaza, Atlanta GA. 30303 and
'Interface Research Corporation, Kennesaw, GA.
ABSTRACT
In addition to product deterioration, mold growth in buildings can have aesthetic consequences
(discoloration, musty odors) and health effects. Effects ranging from fungal pneumonia and
pneumonitis to allergies are well documented. We have observed the heavy infestation of register
vents and insulation of heating, air conditioning and ventilation systems (HVAC) sometimes without
concomitant incidence of fungi in air samples. Studies of the mycoflora of in-duct insulation of an
HVAC system demonstrated strains of Cladosporium herbarum and Eurotium herbariorum that grew
on a nutrient medium supplemented with 60% sucrose. Of particular interest was the demonstration
of the in situ production of ascospores by the Eurotium sp. Sexual spores provide the fungus with a
resistant stage to survive temperature and moisture fluctuations hi the system. Studies of the
germination of conidia from several common airborne fungi demonstrated maximal germination on a
simple basal medium containing glucose and peptone.
INTRODUCTION
The presence of mold growth on the internal surfaces of damp dwellings often results in a high
airborne fungal spore count which, in turn, presents a distinct health risk to the occupants.1 In the
United States, approximately ten percent of the population suffers from allergies caused by
environmental inhalants.2 Although these allergies tend to be principally of a respiratory nature in
atopic Individuals, a prevalence of possible toxigenic symptoms including nausea and vomiting,
backache, fainting, and nervousness were reported by adults living in damp, moldy dwellings.3
Common molds (e.g., Aspergillus, Fusarium, Penicillium, and Stachybotrys spp.) produce a wide
variety of mycotoxins which may cause human diseases.4 Fungi capable of producing mycotoxins
may exacerbate the sick building syndrome.5 In particular, the establishment of foci of toxigenic
fungi and the subsequent release of multitudes of spores represent a serious situation. Moreover,
fungal propagules (elements such as spores and mycelia fragments) in high densities in buildings
indicate not only a potential health hazard from hypersensitivity and toxigenic reactions but, on
some occasions, from infections.6 An effective mechanism for monitoring indoor fungal burden and
the foci of any inoculum is therefore essential.
Experimental Methods
Air sampling for microorganisms. Air samples were collected using a Single Stage
Bioaerosol Sampler (Model 10-880, Andersen Samplers, Inc; Atlanta, GA). The Andersen air
sampler delivers a constant air sample of one ACFM (actual cubic foot per minute) onto an agar
plate. We used five different media (Malt extract agar (MEA), Difco; Corn meal agar (CMA),
Difco: Basal media with glycerol (BMGA)7; Czapek with 40% sucrose (CZSA), Difco; Sabouraud
dextrose agar (SAB), Difco) for 1 minute at each site to maximize the isolation of a variety of fungi.
HVAC and building material sampling. Where feasible, the materials were removed as
samples. Materials collected were placed directly in sterile sampling bags and returned to the
laboratory. Portions of this material were examined directly by white light and Nomarsky
microscopy, and other protions stained with fluophores (i.e., acridine orange, fungifluor, Congo
red) and examined at appropriate wavelengths. Smaller portions of the material were inoculated into
671
-------
a variety of liquid fungal growth media to culture the viable components of the fungal commmunity.
Portions of the sampled material were examined using the scanning elecron microscope.
Where it was not feasible to retrieve material directly but there was some access to the
material, we used a modification of the pressure tape method.8 Acetate film tape (Scotch Brand
Acetate Film Tape, No. 800, 3M Company), adhesive side down, was pressed lightly across the
surface of the suspect material. The adhesive tape with attached fungal structures was placed
directly on a labeled slide at the site. These tape mounts were returned to the laboratory in
specialized containers to prevent disruption of the sample. In the laboratory, the tape was removed
partially and a drop of lactophenol cotton blue or other stain was placed on the slide. The tape was
then pressed back over the stain. These semi-permanent mounts are sufficient for both whitelight,
phase contrast, and Nomarsky microscopy.
Culture maintenance and identification. Following initial isolation, all dominant fungi (>
20% occurrence) were identified by conventional techniques. Minor isolates (< 20% occurrence)
were maintained in pure cultures but were not identified. Cultures were maintained on slant of
CMA in screw capped tubes (13 X 100 mm), stored at 5C and transferred at six month intervals
during the course of the study. Fungal colonies on enumeration media were subcultured to SAB to
assure purity. To reduce contamination by rapid spreading at this stage, we incubated all plates at
low temperature (18C) and transferred to final stock culture as soon as identification was confirmed.
Fungal identification was based on observation of classic morphological characteristics.9|10-"
Germination studies. Studies of germination with Aspergillus flaws, A. niger, and
Penicillium sp. were conducted to evaluate their ability to germinate on 45 selected media.
Following germination, the ability of germinated conidia to grow on the media were assayed.
Inocula for each of these studies were prepared as follows. Plates of Sabourauds dextrose agar were
inoculated from a stock culture. Following growth for 96h at room temperature (approximately
23C) the conidia were harvested by adding a 5ml of sterile saline (0.9%) containing 0.5% tween 80.
The saline was agitated gently with a L-shaped glass rod to release the conidia. The conidial
suspension was collected in a centrifuge tube from the petri dish with a pipette. The suspension was
centrifuged at 2000rpm, supernate decanted and resuspended in an equal volume of tween/saline.
This process was repeated 3 times and the washed conidia were suspended to an optical density of
0.6 (107 conidia /ml).
A volume (O.lml) of this conidia suspension was spread on the surface of various media. A
1cm2 piece of inoculated media was removed with aseptic technique and placed on a prescored (T)
microscope slide. Beginning at the intersection of the two lines ten alternate fields were counted at
0, 12, 18, 24 h. The number of germinated and total conidia were tallied for each time. Media
examined included Malt Extract agar (MEA), Potato Dextrose agar (PDA), Tween-80 agar (T-80),
Sabouraud Dextrose agar (SAB), Mycological agar (MA), Czapek agar (CzA), Corn meal agar
(CMA), Water agar (WA), R2A agar (R2A), Basal agar (BMA), Basal agar + 20% glycerol
(BMGA).
Growth. Conidia suspensions (50 /il) were spotted on plates at three positions. The plates
were incubated at room temperature for up to 72h or a until measurable colony. Measurements of
the colony diameter were taken periodically. The growth rate in cm/h was calculated for these time
periods and the maximum rate was recorded. Media for which germination data was available also
were used for the growth studies.
672
-------
Results
From 20 buildings, we have taken 375 samples at 75 sites. The most frequently isolated
have been Aspergillus niger, Aspergillus versicolor, Cladosporium herbarum, Geotrichum sp.,
Penicillium citrinum, Penicillium hirsutum, and Penicillium variotti. Aspergillus flavus, a known
aflatoxin producer, and Aspergillus fumigatus, a respiratory pathogen, were isolated on 20
occasions. Comparisons of five media for the recovery of fungi from samples illustrated that a
basal media as described by Hocking and Pitt7 most frequently yielded (13/46) the highest numbers
(CFU/m3). However, the other media all yielded highest numbers on occasion with Czapek + 40%
sucrose agar, as expected, yielded the highest number in only a few cases (4/46). The fungi
prevalent in ah" were not always the same as those found colonizing indoor substrates.
Studies of the germination of conidia isolated from air samples gave varying results (Table
I). Of 45 media studied for the germination of conidia no single medium consistently gave the
highest germination with the three species tested. Penicillium sp. gave high levels of germination on
all media except water agar. Low levels of germination of Aspergillus niger were observed on
Czapek agar and Mycological agar. Germination of Aspergillus flavus on Sabouraud dextrose agar
was also low. Growth studies suggest similar variation between isolates (Table II). A. flavus grew
best on CzA but also grew well on T80 agar and CMA. In contrast, A. niger grew best on MEA
and grew well on BMGA and CzA. The Penicillium sp. examined in this work grew poorly on
most media. In studies of the survival of conidia under atmospheric conditions, species of
Aspergillus characteristically were recovered hi high numbers after 24, 48, and 96 hours. On the
other hand, conidia of Penicillium species rapidly lost viability following removal of liquid water.
Discussion
The relative ranking of germination of species examined on the media was Basal medium >
Corn meal agar > Malt extract agar > Sabouraud dextrose agar > Mycological > Czapek with
40% sucrose. While growth studies were inconclusive, comparable growth for all three isolates was
achieved on SAB. However, field application of this media often resulted in rapid overgrowth.
Considerations of both efficiency of germination and growth rates are essential in selecting an
appropriate medium.
Indoor ah- samples and concurrent sampling of HVAC insulation and filter material suggest
that these materials contribute to the fungal burden of indoor air under certain conditions. Moreover,
the association of high levels of airborne viable fungi and complaints of poor air quality are not
always consistent.
CONCLUSIONS
Correlations between the perceived quality of indoor air and the presence of measurable
parameters such as total viable fungi and total fungi, volatile organic compounds, the presence of
certain species would be particularly useful in assessing potential indoor air problems. The use of at
least two media (Basal media with glycerol and Czapek with sucrose) to obtain a better indication of
number and diversity of viable fungi are recommended. Furthermore, particulate counts or numbers
of non-viable fungal propagules should be evaluated.
673
-------
Table I. Germination of conidia of Asperglllus flavus, Aspergillus niger and
Penicillium sp. on various media formulations.
MEA
PDA
T-80
SAB
MA
CzA
CMA
R2A agar
WA
BMA
BMGA
A. flavus
96
95
98
73
99
89
98
95
49
100
, 88
A. niger
93
100
93
93
79
24
86
87
4
98
100
Penicillium sp.
100
97
100
100
100
100
100
100
51
99
97
Values shown are % germination [(number of spores germinating/total
number spores) x 100]. Mean of 10 determinations.
Table II.
Growth Aspergillus flavus, Aspergillus niger, Penicillium sp. on various
media formulations.
MEA
PDA
T-80
SAB
MA
CzA
CMA
R2A agar
WA
BMA
BMGA
A. flavus
.33
.21
.44
.36
.27
.51
.39
.18
0.00
.12
.18
A. niger
.47
.33
.30
.36
.36
.41
.35
.33
0.00
.31
.43
Penicillium sp.
.23
.15
.29
.36
.25
.25
.25
.21
0.00
.34
.28
Values are the mean of three determinations.
in cm/h during maximal growth phase.
Table in. Foci of fungal colonization in buildings.
Growth rate is expressed
Substrate
Predominant Organism
Fiberglasss Duct Liner
Filter
Ceiling Tile
Wallpaper
Painted HVAC Vents
Aspergillus sydowii, A. flavus, Cladosporium,
Euratium, Penicillium spp.
Euratium. Acremonium, Penicillium spp.,
Aspergillus spp.
A. niger, Stachybotrys atra
Penicillium spp., Aspergillus, Cladosporium sp.
Cladosporium
674
-------
Acknowledgements
Research was supported in part by cooperative agreement CR-818696-01-0 from the
Environmental Protection Agency and Georgia State University's Chancellor's Initiative Fund 10-26-
41434.
REFERENCES
1. Institute of Environmental Health Officers. 1985. Mould fungal spores - their effects on
health and the control, prevention and treatment of mould growth in dwellings. IEHO,
London.
2. Stites, D.P., J.D. Stobo, H.H. Fudenberg, and J.V. Wells. 1982. Basic & Clinical
Immunology. Lange Medical Publications Palo Alto, Ca.
3. Platt, S.D., C.J. Martin, S.M. Hunt, and C.W. Lewis. 1989. Damp housing, mould
growth and symptomatic health state. Br Med J. 298:1673-1678.
4. Wicklow, D.T., O.L. Shotwell. 1983. Intra fungal distribution of aflatoxins among conidia
and sclerotia of Aspergillus flavus and Aspergillus parasiticus. Can J Microbiol. 29:1-5.
5. Sorenson, W.G. 1990. Mycotoxins as potential occupational hazards. Dev Ind Microbiol.
31:205-211.
6. Landau, J.W., V.D. Newcomer, and J. Schutz. 1963. Aspergillosis: Report of two
instances in children associated with acute leukemia and review of pertinent literature.
Mycopath et Mycol Appl. 20:177-224.
7. Hocking, A.D. and Pitt J.I. 1980. Dichloran-glycerol medium for enumeration of
xerophilic fungi from low-moisture foods. Appl Environ Microbiol. 39:488-492.
8. Roth, F.J. Jr., P.A. Orpurt, arid D. G. Ahearn. 1963. Occurrence and distribution of fungi
ina subtropical marine environment. Canadian Journal of Botany. 42:375-383.
9. Barren, G.L. 1968. The Genera of Hvphomvcetes from Soil. The Williams & Wilkins
Company. Baltimore.
10. Domsch, K.H., W. Gams, and T. Anderson. 1980. Compendium of Soil Fungi. Academic
Press Inc., New York.
11. McGinnis, M.R. 1980. Laboratory Handbook of Medical Mycology. Academic Press Inc.,
New York.
675
-------
Session 19
Pesticides,
Farm Exposure Studies
-------
Mixer-Loader and Applicator Exposure of Nitrapyrin
to Commercial Handlers and Farmers
Richard C. Honeycutt, Ph.D., Jim Vaccaro, Terry Johnson, Ph.D.
Richard C. Honeycutt, Ph.D.
H.E.R.A.C., Inc.
(Hazard Evaluation & Regulatory Affairs Company, Inc.)
3704-B Boren Drive
Greensboro, NC 27407
Jim Vaccaro
The Dow Chemical Company
Dow Chemical Building 1803
Midland, MI 48640
Terry Johnson, Ph.D.
PTRL East, Inc.
3945 Simpson Lane
Richmond, KY 40475
ABSTRACT
Commercial handlers who load N-Serve® (nitrapyrin) into anhydrous
ammonia tanks and mini shuttle tanks, as well as farmers who apply
nitrapyrin to corn using a tool bar were monitored for respiratory and
dermal exposure to nitrapyrin using air sampling and dosimeter
techniques. Fifteen workers loading bulk N-Serve® into mini shuttle
tanks were monitored and found to have an average respiratory exposure
of 0.031 mg/kg BW/day of nitrapyrin and an average dermal exposure of
0.011 mg/kg BW/day. Workers who transferred N-Serve® from mini
shuttle tanks to anhydrous ammonia tanks showed an average respiratory
exposure of 0.018 mg/kg BW/day and an average dermal exposure of 0.094
mg/kg BW/day. Farmers who applied N-Serve® pre or post emergence to
corn showed an average respiratory exposure of 0.0010 mg/kg BW/day of
nitrapyrin and an average dermal exposure of 0.0037 mg/kg BW/day.
This paper reviews the methodology and results of the study discussed
above.
679
-------
INTRODUCTION
N-Serve 24® (nitrapyrin1) is a soil nitrogen stabilizer which can
be applied by pre-plant incorporation TO corn or by a side dressing.
N-Serve is applied at a maximum use; rate of 1 Ib ai/acre, is marketed
in bulk, and applied by soil injection with anhydrous ammonia using a
tractor-drawn tool bar.
Farm worker exposure, especially to mixer-loaders and applicators
who handle agrochemicals, has been studied and reviewed in the
literature extensively (1). While considerable data exists concerning
exposure to mixer-loaders or applicators applying liquid or powder
formulations of agrochemicals using conventional groundboom aerial or
airblast equipment, less data are available concerning exposure to
loaders and applicators for agrochemicals which are applied with
tractor-drawn tool bars.
The Environmental Protection Agency (EPA) requires that exposure
data be available for mixer-loaders and applicators of agrochemicals
in order to perform risk assessments on individual agrochemicals
currently on the market or being developed for market. This study was
designed to determine potential respiratory and dermal exposure of
commercial loaders and farmers who apply N-Serve to corn, either
preplant or as a side-dressing, using a tractor-drawn tool bar. Whole
body dosimetry and air sampling methods were used to accomplish these
goals.
MATERIALS AND METHODS
Chemical Name
Nitrapyrin (2-chloro-6-(trichloromethyl)-pyridine) is the active
ingredient of N-Serve 24® soil nitrogen stabilizer. N-Serve 24® is
22.2% nitrapyrin, 2.5% related chlorinated pyridines, and 75.3% inert
ingredients.
Test Sites
It is important to point out that this worker exposure study was
run "in situ"; i.e., the N-Serve used for this study was sold to
farmers from mini-bulk containers at the loading facility and the
study was conducted with these particular loaders and farmers. There
were no discreet packages of N-Serve prepared especially for this
study since N-Serve application is an intensive, short, seasonal
practice and is dispensed from 250-gallon or larger bulk tanks as the
farmer needs it.
There were several test sites in Illinois and Ohio involved in this
worker exposure study. Sites consisted of 1) dealer loading sites,
and 2) farm application sites. N-Serve is first sold to farmers and
pumped into anhydrous ammonia tanks at the dealership at which time
the anhydrous ammonia tanks are transferred to the farmers's field and
attached to a tool bar applicator prior to application of N-Serve 24®
soil nitrogen stabilizer by the farmer. Application was generally to
pre-tilled, unplanted corn fields. In some instances N-Serve 24® was
nitrapyrin (2-chloro-6-(trichloromethyl)-pyridine)
680
-------
applied as a side dressing to corn plants (4"- 6" high) spaced three
feet apart. In either case, the same application equipment (tool bar)
and techniques were used by the farmer.
Description of Application Equipment
The equipment used in this study consisted of a closed cab tractor
attached to a tool bar to which the anhydrous ammonia tank containing
N-Serve 24® soil nitrogen stabilizer was attached. The tool bar's
nozzles were essentially tubes adjacent to leading bars which were
used to break the earth just ahead of injection of N-Serve and ammonia
through the tubes. The nozzles (tubes) were under a positive pressure
of 40-80 psi. The N-Serve was injected directly into the ground at a
ground speed of about 6 mph. The boom length varied from 15 to about
42 from farm site to farm site.
Test Subjects
There were three types of test subjects in this study: Dealer
Mini-Shuttle (bulk) Tank Loaders (designated DMSTL); Dealer Anhydrous
Ammonia Tank Loaders (designated DAATL); and Farmer Applicators
(designated FLA).
Monitoring Devices
Whole Body Dosimeters. Sears union suits worn under clothing.
Cotton Gloves. Light, white industrial gloves were worn under
chemically-resistant rubber gloves. Chemically-resistant gloves were
removed during exercise of chores other than loading N-Serve 24® soil
nitrogen stabilizer or anhydrous ammonia.
Air Pumps. Chromosorb 106 sorbent tubes attached to Flow-Lite™ air
pumps were placed on each individual.
Description of Sample Handling from Field Through Analysis
Air sampling tubes, whole body dosimeters, and cotton gloves were
taken from test subjects and placed in pre-labeled plastic bags. Air
sampling tubes were capped at both ends prior to placing in the
plastic bags. The bags were sealed, placed in another plastic bag, and
placed in coolers. About 20 Ibs. of dry ice was added to each cooler
and paper placed on top of the dry ice to help prevent dry ice
dissipation. A chain-of-custody form was completed for each shipment.
The form contained each sample code, the date shipped, and signature
of the shipper. The cooler was then sealed with duct tape and shipped
overnight or driven directly to PTRL, Inc., 155 Prosperous Place,
Lexington, Kentucky.
Field Stability Samples
Blank Samples (Negative Controls). One pair of cotton gloves and one
whole body dosimeter were taken directly from their protective cover
and placed on a table at the test site during the performance of a
replicate and left under the environmental conditions of the test
site.
681
-------
Positive Controls. Positive controls (also referred to as field
spikes) were prepared as follows: One whole body dosimeter and one
pair cotton gloves were taken directly from their protective covers
and placed on a table at the test site away from the N-Serve 24® soil
nitrogen stabilizer loading or application area and spiked with a
predetermined amount of N-Serve 24® obtained from a mini-bulk tank
(high spike). Low spikes were from a solution of N-Serve diluted with
solvent. The sample was left out for the duration of the replicate.
Chromosorb 106 Air Sampling Tubes (Positive Control). One sorbent
tube was fortified using a Hamilton® syringe with a known amount of
nitrapyrin from a tank sample of N-Serve 24® soil nitrogen stabilizer
from the test site. The N-Serve was spiked onto the sampling end on
the tube. The tube was then placed on the table at the test site and
left out under the environmental conditions of the test site for the
duration of the replicate.
Analytical Methods. Analytical methods for nitrapyrin on Chromosorb
106 air sampling tubes, whole body dosimeters, and cotton gloves
involved extraction of the nitrapyrin from each matrix and analysis of
the extract by gas chromatography on GC column (J & W 15M DB-1
Megabore, 5.0 pn film).
RESULTS
Method Validation
Analytical methods were validated for Chromosorb 106 air tubes,
cotton gloves, and whole body dosimeters. Concurrent recoveries of
nitrapyrin was run with each sample set. Average concurrent
recoveries for all matrices were between 90-99%.
Storage Stability
Freezer storage stability studies were performed for nitrapyrin on
whole body dosimeters, cotton gloves, and Chromosorb 106 air tubes
stored under freezer conditions. Nitrapyrin was stable under freezer
conditions for all matrices (89%-101% recovery variance over 90 days
in the freezer).
Results of Determination of Respiratory and Dermal Exposure to
Commercial Loaders and Farm Applicators of N-Serve 24®
Fifteen replicates of exposure to each type of worker were
performed. The results of this study are summarized in Table 1.
Dealer Mini Shuttle Tank Loaders (code = DMSTL) experienced an average
potential respiratory exposure of 0.031 mg/kg BW/day and an average
potential dermal exposure of 0.011 mg/kg BW/day.
Dealer Anhydrous Ammonia Tank Loaders (code DAATL) experienced an
average potential respiratory exposure of 0.018 mg/kg BW/day and an
average potential dermal exposure of 0.094 mg/kg BW/day.
Farmers who apply N-Serve® (code = FLA) with a tool bar experienced
an average potential respiratory exposure of 0.001 mg/kg BW/day and an
average potential dermal exposure of 0.0037 mg/kg BW/day.
682
-------
SUMMARY AND CONCLUSIONS
Potential respiratory and dermal nitrapyrin exposure to commercial
loaders and farmers was determined for N-Serve at several locations in
Illinois and Ohio.
Dealer Mini Shuttle (bulk) Tank Loaders (DMSTL) exhibited an
average respiratory exposure of 0.031 mg/kg BW/day and an average
dermal exposure of 0.011 mg/kg BW/day.
Dealer Anhydrous Ammonia Tank Loaders (DAATL) exhibited an average
respiratory exposure of 0.018 mg/kg BW/day and an average dermal
exposure of 0.094 mg/kg BW/day.
Farmers exhibited an average respiratory exposure of 0.001 mg/kg
BW/day and an average dermal exposure of 0.0037 mg/kg BW/day.
REFERENCES
1) Dermal Exposure Related to Pesticide Use; R.C. Honeycutt, Ed.;
187th American Chemical Society Meeting, Division of Pesticide
Chemistry, St. Louis, MO., ACS Symposium Series #273, The American
Chemical Society, Washington, D.C., 1984.
683
-------
Table 1. Summary table.
Average Exposure (mg/kg BW/day)
Worker Type Respiratory Dermal Total
Dealer Mini
Shuttle Tank
Loader (DMSTL) 0.031 0.011 0.042
Dealer Anhydrous
Ammonia Tank
Loader (DAATL) 0.018 0.094 0.11
Farmer (FLA) 0.0010 0.0037 0.0047
684
-------
ANALYTICAL METHOD FOR THE SCREENING OF PESTICIDES AND
POLYNUCLEAR AROMATIC HYDROCARBONS FROM HOUSEDUST
Tapan K. Majumdar, David E. Camann and Paul W. Geno
Southwest Research Institute
P.O. Drawer 28510
San Antonio, Texas 78228-0510
ABSTRACT
Neutral pesticides, PAHs and cotinine from housedust were analyzed using single extraction, followed by
cleanup and finally by GC/MS. Method validation was performed initially using 28 pesticides and 6 PAHs. Later four
more PAHs were added to the list. A 2 g housedust sample was Soxhlet extracted for 16 hours using 200 mL of 6%
diethyl ether in n-hexane. The extract was reduced to 5 mL and analyzed. Matrix effect was observed in the extract
without cleanup. Cleanup through activated alumina showed poor recovery for most of the target analytes. Cleanup
through Florisil showed good recoveries. GPC cleanup showed similar results as Florisil. Cotinine was analyzed
without cleanup because of its low recovery (10-14%) in all the evaluated cleanup methods. There was no significant
difference in results between immediate and 48 hours delayed extraction after spiking the bousedust sample.
Reproducible results were observed when extraction and analysis were repeated on the same housedust samples after
two months. For organochlorine pesticides reproducible results were obtained when analysis was performed by GC/MS
and GC/ECD. High levels of p,p'-DDT and dieldrin were observed in dusts from homes with old rugs, which
confirmed very slow degradation of some organochlorine pesticides in the indoor environment.
INTRODUCTION
Pesticides, polynuclear aromatic hydrocarbons (PAH) and tobacco alkaloids are well known carcinogens and
potential biohazards, which are present in housedust Infants and toddlers are more exposed to such pollutants than
adults because of their frequent contact with the floor and hand-to-mouth behavior. Dust is reported to be a major
source of home exposure to pesticides112 and strong correlations have been found between levels of pesticides in indoor
air and housedust. Track-in is reported to be a major source of pesticides and PAHs in housedust4 and degradation
of chlorinated chemicals in the household environment is slow. These findings have elicited a growing interest in the
analysis of pesticides from housedust.
Despite these concerns, there is very little literature information on the analysis of pesticides and PAHs from
housedust. It is a challenging matrix for the analysis of any target analyte because it contains numerous chemicals,
including foods, cosmetics, chemicals used for indoor pest control, tobacco alkaloids and many other chemicals that
are carried in from outdoors. Monitoring specific chemicals from such a complex matrix is extremely difficult. This
paper reports a method validation for screening of pesticides, PAHs and cotinine (a tobacco alkaloid), analysis results
for some real housedusts, and implications from the analysis.
EXPERIMENTAL
Method validation was performed by spiking 2.0 g of housedust with a series pesticides, polycyclic aromatic
hydrocarbons and cotinine as shown in Tables I and n. Levels of target compounds were varied based on their
response factors in GC/MS. All the target analyte standards were purchased from Chem Service, West Chester, PA.
p-Terphenyl-D14 was used as surrogate standard to monitor the extraction loss. The spiked dust and same amount of
unspiked dust (dust-blank) were Soxhlet extracted in 6% diethyl ether in n-hexane for 16 hours. Then each extract was
blown down to a final volume of 5.0 mL using turbo evaporator at 30-40°C under nitrogen stream. Each of the extract
was then divided into five aliquots and treated as follows:
Aliquot 1 (1 mL): Solvent exchanged to n-hexane and analyzed by GC/MS by the procedure described later.
Aliquot 2 (1 mLI: GPC cleanup - The extract was solvent exchanged to dichloromethane (DCM), GPC
cleanup was performed by dual column (Water Envirogel GPC columns, pore size is of 500 A) using EPA method
3640A. The final volume of eluent was reduced to 1 mL and analyzed by GC/MS.
685
-------
Aliquot 3 (1 mL): Alumina cleanup - Column was prepared by taking 1 g ICN alumina B-Super I (ICN
Biomedicals) in a 5 mL glass syringe of 0.5" i.d. and conditioned with 10 mL n-hexane. One milliliter of extract was
passed through the column and eluted with 10 mL of 10% acetone in n-hexane. The final volume of the eluent was
blown down to 1 mL using turbo evaporator under nitrogen stream and analyzed by GC/MS.
Aliquot 4 (I itiL): Florisil cleanup - The procedure was the same as described in alumina cleanup with the
exception that Florisil (200 mesh, activated Mg-Silicate, Sigma Chemical Co.) was used as column material. This
procedure was used to get the results shown in Table I. The procedure was then changed by using 4 mL Florisil in a
10 mL serological disposable pipet of 8.5 mm i.d., conditioning with 10 mL n-hexane, passing 1 mL extract and then
eluting with 20 mL of 10% acetone in n-hexane and blowing down to final volume to 1 mL for analysis.
Aliquot 5 (1 mL): Kept as reserve.
Table I. Recoveries of neutral pesticides, PAHs and cotinine from housedusL
Compound
Alachlor
Aldrin
Atrazine
Bendiocarb
Captan
Carbaryl
a-Chlordane
Y-Chlordane
Chlorothalonil
Chlorpyrifos
Dacthal
p,p'-DDE
p,p'-DDT
Diaz iron
Dichlorvos
Dicofol
Dieldrin
Folpet
Heptachlor
Hexachlorobenzene
Lindane
Malathion
Methoxychlor
cir-Permethrin
(ranj-Perraethrin
o-Phenylphenol
Propoxur
Resmethrin
Benz(a)anthracene
Benzo(ghi)perylene
Benzo(a)pyrene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Coronene
(-)Cotinine
p-Terphenyl-D14
QL
(ppb)
53
39
39
59
78
78
59
59
59
49
59
39
78
49
78
160
39
78
59
49
59
39
49
230
230
39
39
120
59
59
59
39
39
78
120
20
Spiked
(ppb)
2732
2001
2001
3002
4002
4002
3002
3002
3002
2501
3002
2001
4002
2501
4002
8004
2001
4002
3002
2501
3002
2001
2501
12006
12006
2001
2001
6003
3002
3002
3002
2001
2001
4002
3002
2500
% Recovered
NCL
122
109
119
141
115
144
113
117
109
123
109
103
126
137
118
37
128
107
126
120
113
145
131
121
112
130
134
41
61
109
96
138
29
84
101
113
AL
97
101
78
66
71
109
110
78
103
97
99
111
113
36
60
120
74
109
106
99
72
95
117
114
105
55
73
26
58
99
102
117
85
84
11
108
GPC
118
106
112
122
129
120
114
101
121
100
104
124
127
84
68
132
101
115
107
104
102
123
132
114
104
102
116
46
87
83
90
82
89
46
14
108
FL
116
107
114
129
128
123
121
93
119
100
109
128
130
97
66
127
102
118
114
108
105
122
126
124
117
117
122
31
82
111
109
115
112
104
14
116
686
-------
Table n. Recoveries of neutral pesticides and PAHs extracted 48 hours after spiking
housedust samples.
Compound
% Recovered
Spiked (ppb) HD Sample #1 HD Sample 26
Alachlor
Aldrin
Atrazine
Bendiocaib
Captan
Carbaryl
a-Chlordane
•y-Chlordane
Chlorothalonil
Chloipyrifos
Dacthal
p,p'-DDE
p,p'-DDT
Diazinon
Dichlorvos
Dicofol
Dieldrin
Folpet
Heptachlor
Hexachlorobenzene
Lindane
Malathion
Methoxychlor
cis-Permethrin
iraM-Permethrin
o-Phenylphenol
Propoxur
Resmethrin
Benz(a)anthracene
Benzo(ghi)perylene
Benzo(a)pyrene
Dibenzo(a,e)pyrene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Chrysene
Coronene
Indeno(l,2,3-c,d)pyrene
Dibenz(aji)anthracene
p-Teiphenyl-D14
6435.00
12787.50
7213.75
3500.00
12622.50
3168.00
9405.00
9405.00
5833.35
11725.35
5911.75
8034.65
25166.65
8833.35
3166.65
16333.35
9666.65
25253.75
9435.00
5833.35
14035.85
8060.50
3833.35
19247.20
17550.00
3166.65
1567.50
76666.65
2333.35
7000.00
4666.65
4707.25
3166.65
3166.65
2333.35
4733.35
3200.00
3208.35
3833.35
94
98
97
90
65
99
98
94
1
92
92
92
92
100
70
121
92
89
98
102
96
101
96
89
88
101
112
109
94
90
71
89
81
95
91
88
93
88
90
81
89
94
95
81
132
94
89
101
89
88
86
84
125
75
72
94
88
94
107
90
83
90
86
87
98
103
79
100
119
120
79
95
91
91
84
98
98
87
GC/MS analysis was performed using Fisons MD800 single quadrupole instrument equipped with Lab-base
software. The GC column was DB-5 (J & W Scientific) 30 m long and 0.32 mm i.d. Helium was used as the carrier
gas. Injector temperature was 220°C. Analysis was performed by maintaining the oven temperature at 60°C for 5
minutes, then ramped at a rate of 15°C/min to 200°C, kept for 3 minutes at that temperature and again ramped at a rate
of 15°C/min to 295°C and kept at that temperature for 15 minutes before it went back to the initial temperature of
60°C. Quamitation was performed by operating the mass spectrometer in selected ion monitoring (SIM) mode to obtain
good sensitivity. Two of the most intense ions were recorded for each target analyte and the ion with highest
abundance was used as the quantitation mass. The instrument was used in full scan mode only when confirmation was
necessary for any analysis. lonization was performed by 70eV electron impact.
687
-------
Dual column GC/ECD was performed using Fisons GC 8000 Series instrument. The front column was DB608
(30 m x 0.53 mm i.d.) and the back column was DB1701' (30 m x 0.53 mm i.d.). Helium was used as the carrier gas
and nitrogen was used as the bath gas. The oven temperature program was the same as shown for GC/MS above.
RESULTS AND DISCUSSION
Housedust is one of the most troublesome matrices for analysis of any target analyte. Multiple cleanup is
usually necessary for such a difficult matrix. Our objective was to find a cost effective way to analyze pollutants from
this matrix. Therefore, single step cleanup methods were explored in this study. Percent recoveries of 28 different
pesticides, 10 different PAHs and cotinine after spiking a housedust are shown in Table I. A maximum of 5%
variations in results for some of the target analytes were observed in duplicate analysis. p-Terphenyl-D14 was used
as a surrogate to monitor the extraction loss. The second column in Table I shows the GC/MS quantitation limit in
terms of ppb of analyte in housedust. The high quantitation limits of permethrins were due to the presence of the
compounds in high levels in the mixture. Actual quantitation limits for permethrins were 80 ppb in the instrument used
in this study. The results shown in Table I were corrected for amounts detected in the unspiked housedust sample.
A variation of ± 30% was taken to be acceptable limit for housedust samples. The third column shows the recovery
of target analytes without any cleanup and the results clearly show a large matrix effect, since recoveries for a large
number of target analytes were beyond the acceptable limit High recoveries were observed for bendiocarb, carbaryl,
diazinon, malathion, methoxychlor, propoxur and benzo(b)fluoranthene and low recoveries were observed for dicofol,
resmethrin, benz(a)anthracene, and benzo(k)fluoranthene when the extract was analyzed without any cleanup. Cleanup
through activated alumina lead to low recoveries for a large number of target analytes as well. The last two columns
in Table I show recoveries of analytes after GPC and Florisil cleanup. These two cleanup methods showed good
recoveries for all the compounds except dichlorvos, resmethrin and cotinine. The recovery problem for dichlorvos and
resmethrin was later eliminated in the Florisil cleanup (Table n). Low recoveries were observed for cotinine in all the
cleanup methods discussed above. There is a good analytical method for cotinine alone,5 but our objective was to find
a single extraction and cleanup method for all the target analytes to reduce the time and the cost of analysis.
Considering the importance of cotinine as a marker compound to assess personal exposure to environmental tobacco
smoke5, we decided to analyze it from uncleaned extract (101% recovery, Table I). From the results shown in Table I,
it is clear that either GPC or Florisil cleanup was the method of choice. We decided to accept the cleanup through
Florisil column because it is cheaper and utilized environmentally safe solvent system (10% acetone in n-hexane for
Florisil cleanup and dichloromethane for GPC cleanup). Besides, solvent exchange will be necessary for GPC cleaned
extract if GC/ECD analysis is desired for organochlorine pesticides. Florisil cleanup was used in the rest of the
experiments. Cleanup efficiencies were later checked in the Florisil cleanup by varying the concentration of acetone
in n-hexane and by increasing elution volume of the solvents. Best results (Table H) were obtained by using 20 mL
of 10% acetone in n-hexane as the eluent. However, recovery of cotinine was still low with all the improvements in
the cleanup method.
The results shown in Table I were obtained by extracting the dust immediately after spiking with the target
analytes. Table n shows recoveries of the same target analytes and four additional PAHs [chrysene, dibenzo(a,e)pyrene,
indeno(l,2,3-c,d)pyrene and dibenz(a,h)anthracene] from two different housedusts (Sample 1 and Sample 26) extracted
48 hours after spiking (to insure equilibration) and using the improved Florisil cleanup. Good recoveries were obtained
despite delayed extractioa There was still a maximum of 5% variation for some of the analytes between duplicate
analysis of each housedust sample shown in Table n. HD Sample 1 showed only 1% recovery for chlorothalonil which
was the same in the duplicate extraction. No such problem was observed for HD Sample 26, which can only be
explained as matrix effect.
Reproducibility of the GC/MS results for organochlorine pesticides were checked using dual column GC/ECD
analysis and some of the results are shown in Table HI. The GC/MS quantitation limits were not as low as the
GC/ECD analysis but were very close and results were fairly reproducible.
Table IV shows results for duplicate extractions of four different housedust (carpet) samples. Time difference
between extraction-A and extraction-B was two months. Considering that we were dealing with one of the most
difficult matrices, the reproducibility of the results was acceptable. Sample 27 showed very high level of dieldrin, and
Sample 43 showed very high level of DDT (Table IV). Presence of dieldrin and DDT was confirmed by full scan
analysis of the sample extract followed by NBS library search. DDT has been out of market since the early 1970s and
688
-------
dieldrin has been banned since around 1979. After further investigation on the history of the rugs in the two homes,
it was found that Sample 43 was obtained from a 60-year old oriental carpet that had been in the home for 3.5 years
and Sample 27 was obtained from an 18-year old carpet Therefore, these results clearly reveal that dieldrin and DDT
can be present at very high levels 20 years after use. Followup investigations are in progress in these two homes for
DDT and dieldrin am' their metabolites in the blood of the people living there. Sample 44 showed high levels of PAHs
in the carpet dust (the carpet was 7 years old). Sample 49 showed high levels of carbaryl and permethrins although
the dust was obtained from a 7-month old carpet; it appeared that the carpet was treated heavily with pesticides for
indoor pest control.
Table in. Comparison of GC/MS and dual column GC/ECD results for organochlorine pesticides from
housedust samples.
Compound
a-Chlordane
y-CUordane
Chlorpyrifos
p,p'-DDE
p,p'-DDT
Dichlorvos
Dicofol
Methoxychlor
cu-Permethrin
muu-Permethrin
Sample 43
GC/MS
55
58
309
5297
103377
13
2293
13522
ND
658
GC/ECD
53
61
300
5803
115177
ND
1574
11954
ND
ND
Sample 44
GC/MS
33
ND
195
250
385
12
490
14211
ND
ND
GC/ECD
ND
ND
57
255
223
ND
363
1958
ND
ND
Sample 49
GC/MS
26
24
381
ND
31
ND
ND
ND
588410
299158
GC/ECD
24
124
341
99
40
ND
ND
ND
303836
174229
ND - Not detected
CONCLUSION
Neutral pesticides and pesticides from housedusts can be analyzed by a single sequence of extraction using 6%
diethyl ether in n-hexane, cleanup using Florisil column followed by GC/MS. Cotinine can be analyzed using the same
extract without cleanup. High levels of DDT and dieldrin can be detected from old carpet dust 20 years after
application of such pesticides. Presently a study is in progress for the development of a good analytical method for
the determination of acid herbicides from housedust and soil samples.
ACKNOWLEDGEMENTS
This research was sponsored by the National Cancer Institute under Grant R01-CA56095 to Dr. Jonathan
Buckley, University of Southern California.
REFERENCES
1. J. W. Roberts, W. T. Budd, D. E. Camann, R. C. Fortmann, R. G. Lewis, M. G. Ruby, T. M. Spinier, "Human
Exposure to Pollutants in Floor Dust in Homes and Offices" in J. Exposure Analysis and Environmental
Epidemiology. 1992;2: Suppl. 1, pp 127-146.
2. R. G. Lewis, A. E. Bond, D. E. Camann, R. C. Fortmann, L. S. Sheldon, "Determination of Routes of Exposure
of Infants and Toddlers to Household Pesticides" in Proceedings of the Annual Meeting of the Air and Waste
Management Association. Vancouver, B. C., 1991, Paper No. 62.1.
3. D. E. Camann, R. C. Fortmann, J. W. Roberts, and R. G. Lewis, "Association Between Measured Pesticide
Levels in Indoor Air and Carpet Dust in the Home" in Measurement of Toxic and Related Air Pollutants.
AWMA Publication. 1992, VIP-21, 2, pp 1113-1121.
689
-------
J. W. Roberts, W. T. Budd, J. Chuang, and R. G. Lewis, "Chemical Contaminants in Housedust: Occurrence
and Sources" in Proceedings of the 6th International Conference on Indoor Air. Helsinki, 1993 (in press).
S. L. Kopczynski, "Multidimensional Gas Chromatographic Determination of Cotinine as a Marker Compound
for Paniculate Phase Environmental Tobacco Smoke" in J. Chromatograohv. 1989, 463, pp 253-260.
Table IV. Concentration (ppb) of pesticides and PAHs in 2 g duplicate extractions of housedust samples.
Sample 43
Sample 27
Compound
Alachlor
Aldrin
Atrazine
Bendiocarb
Captan
Carbaiyl
a-Chlordane
Y-Chlordane
Chlorothalonil
Chlorpyrifos
Dacthal
p,p'-DDE
p,p'-DDT
Diazinon
Dichlorvos
Dicofol
Dieldrin
Folpet
Heptachlor
Hexachlorobenzene
Lindane
Malathion
Methoxychlor
cir-Permethrin
nms-Permethrin
o-Phenylphenol
Propoxur
Resmethrin
Benz(a)anthracene
Benzo(ghi)perylene
Benzo(a)pyrene
Benzo(b)fluoranthrene
Benzo(k)fluoranthene
Coronene
Surrogate Recovery:
ND- Not detected
A
ND
ND
ND
ND
ND
ND
ND
ND
ND
1530
ND
ND
ND
ND
ND
ND
138972
ND
ND
ND
ND
ND
656
ND
ND
1118
ND
3279
2330
3428
4223
6334
5216
86
106
B
ND
ND
ND
ND
ND
ND
ND
ND
ND
2293
ND
ND
ND
ND
ND
ND
110188
ND
ND
ND
ND
ND
131
ND
ND
994
10
3281
1584
2552
1882
5655
4551
87
119
A
ND
ND
^ND
ND
ND
ND
55
58
ND
309
ND
5297
103377
ND
13
2293
ND
ND
ND
ND
ND
1705
13522
ND
ND
484
40
ND
4472
4576
6321
7976
5982
454
107
(0.5 g)
B
ND
ND
ND
ND
ND
ND
78
86
ND
306
ND
4166
157783
ND
ND
1728
ND
ND
ND
ND
ND
1125
10345
472
520
352
ND
ND
4423
4670
5655
7194
6475
334
104
Sample 44
A
ND
ND
ND
ND
ND
ND
ND
33
ND
195
ND
250
385
ND
12
490
ND
ND
ND
ND
ND
ND
14211
ND
ND
784
ND
7118
15634
12313
22003
29271
21129
1349
110
B
ND
ND
ND
ND
ND
ND
ND
ND
ND
240
ND
267
1412
ND
ND
452
ND
ND
ND
ND
ND
ND
922
ND
ND
594
14
6094
14174
19753
24326
30373
25503
2078
110
Sample 49
A
ND
ND
ND
262
ND
1163943
26
24
ND
381
ND
ND
31
649
ND
ND
233
ND
ND
ND
ND
ND
ND
588410
299158
ND
26
1467
992
1174
1329
1927
1357
77
97
B
ND
ND
ND
13
1282
9978281
ND
ND
ND
309
ND
86
361
361
ND
ND
157
ND
ND
ND
ND
ND
ND
494305
246331
ND
ND
5723
1092
1243
1485
2321
1360
33
96
B extracted 2 months after A
690
-------
A PILOT STUDY FOR MEASURING ENVIRONMENTAL EXPOSURES FROM
AGRICULTURAL APPLICATIONS OF PESTICIDES: AN OVERVIEW
A. E. BOND, G. G. AKLAND, R. G. LEWIS, AND C. J. NELSON
AREAL, U.S. ENVIRONMENTAL PROTECTION AGENCY, RTP, NC
ABSTRACT
A four-farm pilot study was conducted during May and June 1992
in southeastern Minnesota to test and evaluate a variety of
different sampling methods for determining the potential human
exposure of farm applicators and their families to 26 different
pesticides during routine agricultural pesticide applications.
The study design incorporated sampling methods that included
the potential routes of exposure through inhalation, ingestion, and
dermal adsorption. This paper includes a discussion of the study
design, the sampling rationale, and a profile of the different
types of environmental samples collected from air, soil, house
dust, dermal patches and gloves, diet (drinking water, solid food,
and beverages), and biological samples (blood and urine).
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication. Mention of
trade names or commercial products does not constitute endorsement
or recommendation for use.
BACKGROUND
The US Environmental Protection Agency (USEPA), in
collaboration with the National Cancer Institute (NCI), agreed to
conduct a series of pilot studies to test methods for assessing the
human exposure to selected pesticides commonly used in agriculture.
The principal objectives of the pilot studies are to develop and
improve sampling and analytical methods to be used in a large-
scale, multi-media, 200 farm human exposure study of farm pesticide
applicators and their families. The large-scale study is currently
scheduled to start during FY-94 with the survey of 75,000
registered pesticide applicators selected from North Carolina and
Iowa.1
STUDY DESIGN
The multi-media, multi-seasonal, sampling design developed by
the USEPA for use in the Nonoccupational Pesticide Exposure Study
(NOPES)2 was used as the foundation for the initial approach
to be used in this study. The draft questionnaires evaluated during
the pilot study included pesticide usage and application practices,
health, diet, activity pattern, and safety issues.
691
-------
The initial pilot study was conducted in south central Minnesota
over a three-week period. The farms were selected from a cohort of
licensed farm pesticide applicators being studied by Dr. Vincent F.
Garry of the University of Minnesota. Dr. Garry is Director of the
University's Laboratory of Environmental Medicine and Pathology,
located in Minneapolis, Minnesota. His selection of the four farms
was based on their reported use of the target chemicals (Table 1) ,
their willingness to participate, and their diversity of
agricultural practices, such as: the cultivation of grains, corn,
and/or beans, and the raising of livestock.
SAMPLING STRATEGY
The field sampling of the farms (Table 2) was conducted during
the period of 26 May through 15 June 1992. Five days was required
at each location which included three days of environmental
sampling. The 72-hours of ambient air sampling was divided into
three consecutive 24-hour periods at each location. The five day
activity and sampling schedule as defined below was proposed to be
followed at each of the farms. This complex sampling approach
attempted to include sampling periods identified as pre-
application, application, and post-application for a variety of
different pesticides.
DAY 1 Orientation: The environmental sampling and questionnaire
administration team arrived at the farm and selected the
residential indoor, outdoor, and pesticide storage area or barn
sampling locations, administered the EPA Pesticide Usage and
Activity Pattern questionnaire, set-up the portable meteorological
monitoring station, discussed the proposed dermal and respiratory
sampling plans, explained and demonstrated the operation of the
sampling pumps, determined the carpet area for the collection of
the house dust sample, determined the locations for the collection
of soil samples, explained the proposed three-day sampling schedule
to the family, and answered any questions raised . by the
participants. Air exchange rate emitters and detectors were placed
throughout the house to monitor the air exchange rate over the next
seventy-two hours.
The bio-medical collection team furnished the containers and
the information necessary for the proper collection and storage of
the DAY 2 early morning urine samples from the applicator and
spouse.
DAY 2 Pre-application: The environmental sampling and
questionnaire administration team returned and set-up the
calibrated air samplers required for the collection of ambient
indoor, outdoor (adjacent to the house), and pesticide storage or
interior barn area and initiated sampling. Additionally, the team
provided the applicator with a calibrated personal respiratory air
sampling pump, a pair of pre-cleaned cotton gloves, and gauze
patches. The patches were to be worn on the nape of the neck (on
692
-------
top of the shirt) and on the front of each thigh (on top of the
trousers) . These sampling devices were used to estimate the
applicator's pesticide exposure throughout the day's pre-
application activities (approximately 12 hours). At the conclusion
of the day's activities the farmer provided a hand-wipe sample
prior to performing any routine personal hygiene procedures. The
team made a video and still photo/slide record of the exterior
physical layout of the farm including: buildings, pesticide
application equipment, and residence with relationship to
cultivated and/or livestock areas and a portion of the pre-
application activities the applicator performed during the day. The
pesticide application schedule for the following day was confirmed.
The NCI/EPA Pesticide Usage and Health questionnaire was
administered.
The bio-medical collection team arrived and collected the pre-
pesticide application urine and blood samples from the farmer and
spouse. The urine samples were to be the first specimen of the day.
The team left additional urine specimen containers for the
applicator's initial post application urine specimen (before
retiring on DAY 3) and for the additional specimens scheduled to be
collected on DAY 4 from the farmer, spouse, and up to two children.
The dietary collection team provided the containers and
necessary instructions for the proper collection and storage of the
dietary samples (water and food) scheduled for collection on DAY 3.
The applicator's personal respiratory sample, glove, gauze
patches, and hand-wipe samples were collected at the end of the
day's activities.
DAY 3 Application: The environmental sampling and questionnaire
administration team returned and provided the applicator a
replacement personal respiratory air sample, a new set of pre-
cleaned cotton gloves, and gauze patches. These samples were
provided prior to the initiation of any pesticide preparation or
application activities. The team made a video and still photo/slide
record of each step and the estimated time required in the
handling, mixing, loading, and initial application of the pesticide
formulation. A sample of the formulation was collected for
laboratory analysis at this time and at any time the formulation
was changed until the pesticide application activities were
completed. The DAY 2 indoor, outdoor, and pesticide storage or
interior barn area air samples were collected and replaced with DAY
3 samples. If the pesticide application activity was interrupted
for the dinner meal, the personal respiratory air, gauze patch,
glove, and hand-wipe samples were collected and replaced with new
samples for the remainder of the afternoon pesticide application
activities. In addition to the hand-wipe sample collected from the
applicator, up to two hand-wipe samples were collected from the
children who had volunteered to participate in the remainder of the
study. At the conclusion of the pesticide application activities
693
-------
for the day, a final hand-wipe sample was collected from the
applicator following the removal of the last glove sample, and from
the same children who had volunteered earlier to provide these
samples and a urine specimen on DAY 4. These hand-wipe samples
were collected prior to any hand washing or other normal cleanup
procedures and immediately following all pesticide application
activities.
The dietary collection team administered its questionnaire and
also collected second plate dietary samples of the food and
beverages served during the three meals on this day. Additionally,
they collected a water sample from the main source of drinking
water for the family.
The applicators agreed to provide a post-application urine specimen
prior to retiring for the evening in a container provided by the
bio-medical team and refrigerated until picked-up the next day.
DAY 4 Post-application: The environmental sampling and
questionnaire administration team returned and collected the DAY 3
samples and replaced them with DAY 4: indoor, outdoor, pesticide
storage or barn, personal respiratory air samples, glove, and gauze
patch samples. If the previous day's pesticide activity continued
then an identical procedure was followed from DAY 3 for the
collection of dermal (glove and gauze patches), personal air, and
hand-wipe samples from both the applicator and up to two children.
The house dust sample was collected utilizing the High Volume Small
Surface Sampler (HVS3), Cascade Stack Sampling Systems, Inc. of
Bend, Oregon and the principal entryway and two pathway soil
samples were collected as manual surface scrapings. At the
conclusion of the day's activities the team collected the
respiratory air sample, glove, gauze patch, and hand-wipe samples
from the applicator and the air exchange rate emitters and
detectors from the house.
The bio-medical collection team returned and collected the
post-pesticide application early morning urine and blood sample
from the applicator and spouse, and the refrigerated urine
specimens from up to two different children following the
guidelines used in DAY 2. If the pesticide activity was continued
from the previous day, then an additional post-application final
urine sample was provided by the applicator prior to retiring for
the evening. Regardless of whether or not the DAY 4 activities
included any pesticide application activity, a final post-
application urine sample container was provided to the applicator
for a final urine specimen to be collected on the morning of DAY 5.
The dietary collection team returned and collected the
previous day's diet samples and completed any missing information
required for the dietary questionnaire.
694
-------
DAY 5 Wrap-up: The environmental sampling and questionnaire
administration team returned and collected the DAY 4 indoor,
outdoor, personal respiratory air, and pesticide storage or barn
air samples, associated sampling equipment, and the portable
meteorological monitoring station. The team completed any missing
information required by the EPA and/or NCI EPA questionnaires.
The bio-medical collection team returned and collected any
biological samples obtained during DAY 4 and the final post-
application urine specimen provided by the applicator on the
morning of Day 5.
The environmental sampling and questionnaire administration
team reviewed a written record of any questions raised during the
study by the participants and any responses provided to the
participants. The team then proceeded to the next farm location.
SUMMARY
An analysis of the methods development and sampling design
pilot study results indicated several adjustments to the study and
sampling design were needed to be evaluated during the next pilot
effort. The results indicated that additional methods development
research was required for several of the media, particularly for
soil, house dust, blood and urine. The test questionnaires did not
adequately address the activity patterns of the spouse, children,
or the applicator during non-application periods when linked to the
exposure data for the same periods. Lastly, a more comprehensive
biological and hand wipe sampling approach will be incorporated in
the revised design for the spouse and children. Despite several
minor problems encountered during the initial pilot effort, the
overall study and sampling design proved to be workable and capable
of producing a complex profile of multiple routes of exposures to
pesticides used in agriculture. A more detailed discussion of the
study results are presented in several additional papers and
posters during this symposium.
REFERENCES
1. Agricultural Health Study, A prospective study of cancer and
other diseases among men and women in agriculture. M.C.R. Alavanja,
A. Blair, E. Grose, and D. P. Sandier. Draft Protocol 4 Feb 1993.
2. Nonoccupational Pesticide Exposure Study (HOPES). U.S.
Environmental Protection Agency Final Report. EPA/600/3-90/003, Jan
1990.
695
-------
TABLE 1.
AGRICULTURAL HEALTH STUDY (AHS)
LIST OF TARGET CHEMICALS
HERBICIDE
INSECTICIDE
FUNGICIDE
atrazine
metolachlor
carbaryl
malathion
propoxur
dieldrin
DDE
dacthal
folpet
2,4,-D salts
trif luralin
chlorpyrifos
permethrin
heptachlor
DDT
alpha-chlordane
captan
pentachlorophenol
alachlor
dicamba
phorate
diazinon
aldrin
ODD
gamma-chlordane
dicloran
696
-------
TABLE 2.
AGRICULTURAL HEALTH STUDY (AHS)
FIELD SAMPLING DESIGN
NUMBER OF SAMPLES COLLECTED PER FARM
Matrix
Indoor Air
Outdoor Air
Personal Air
Hand Wipe
Gloves
Gauze Patches
House dust
Soil
Food
Water
Blood
Urine
Storage Building
Formulation
GRAND TOTAL
Day
1
Day
2
1
1
A
A
A
A
A
S
A[AM]
S[AM]
1
11
Day
3*
1
1
A1
A1
c.1 cb'
A1
A1
32
1
A[PM]
1
1
15-21
Day
4
1
1
A1
A1'3
c." q,1-3
A1'3
A1'3
1
3
A
S
A[AM) A3[PM]
g(AMJ
C>[AM] Cfc[AM]
1
I3
14-27
Day
5
A[AM]
1
Sub-
Total
3
3
3-5
2-5
2-8
2-5
2-5
1
3
3
1
2
2
4-5
2
2
3
1-2
41-60
Applicator
S
Spouse
C. = Child a
Cb = Child b
If the applicator comes in for the noon meal and continues the
application in the afternoon, there will be an afternoon sample
collected.
There will be a food sample collected for each of the three meals for
farm number 2 and 3.
If the pesticide activity has continued from the previous day.
Proposed day of pesticide application.
697
-------
ANALYTICAL METHODS FOR ASSESSING THE EXPOSURE OF FARMERS
AND THEIR FAMILIES TO PESTICIDES
Paul W. Geno, David E. Camann and Kevin Villalobos
Southwest Research Institute
P.O. Drawer 28510
San Antonio, TX 78228-0510
Robert G. Lewis
U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, NC 27711
ABSTRACT
In the summer of 1992, the pre-pilot phase of the NCI/EPA Agricultural Health Study (AHS) was initiated
to evaluate environmental sampling and analysis methods for assessing the exposure of farm workers and their
families to selected pesticides used in and around the farm. Sampling media consisted of polyurethane foam
(PUF) for ambient and breathing zone air monitoring; gloves, a-cellulose patches and isopropanol handwipes for
personal exposure monitoring; and soil and housedust samples for environmental monitoring. An initial target
compound list of 26 herbicides, insecticides and fungicides was selected. The target compound list was
supplemented, for selected samples, with two insecticides and two herbicides following field sampling. The
pesticides were separated into three classes: organochlorine pesticides, organophosphates and carbamates, and
acids. The wide variety of matrices and pesticides presented a challenge since no single extraction method or
analytical method could be used for all matrices or for all target analytes. This presentation will describe the
extraction and analysis methodology used for this project. In addition, method validation results, laboratory spike
results and blind external audit sample results are presented.
INTRODUCTION
The Environmental Protection Agency (EPA) and the National Cancer Institute (NCI) have planned a long-
term epidemiologic study of farm workers and their families to identify and quantify cancer risks among farmers
and to develop an exposure assessment strategy concerning agricultural exposures to pesticides. In 1992, EPA
and NCI initiated a pre-pilot study to assess exposure assessment and analytical methods for the proposed farm
worker study.
Specific objectives for the analytical portion of study included: 1) Produce analytical data of good quality
in order for EPA and NCI to accurately evaluate the sampling scheme. 2) Obtain sufficient data to make specific
recommendations as to the appropriateness of the proposed approach. 3) Obtain sufficient data to make specific
recommendations as to where further developmental work is required. 4) Obtain sufficient data to make
recommendations as to how the analysis scheme could be revised to obtain data in a more cost efficient manner
without compromising the quality of the data.
This paper describes the analytical approach, method validation results and blind spike audit sample results
used for the 1992 study.
ANALYTICAL APPROACH
A list of the target pesticides chosen for the 1992 Agricultural Health Study (AHS) pre-pilot study is given
in Table I. hi this table, the analytes are arranged in analytical groups. Organochlorine pesticides were determined
by dual column gas chromatography using 30 m x 0.530 mm i.d. DBS and DB608 columns with electron capture
698
-------
detection (GC/ECD) on a HP 5890A gas chromatograph. Other nonchlorinated neutral pesticides, i.e., carbamates
and organophosphate pesticides, were determined by gas chromatography-mass spectrometry using selected-ion
monitoring (GC/MS-SM) on a Hsons MD-800 GC/MS using a 30 m x 0.320 mm i.d. DB5 column. Confirmation
analysis of neutral pesticides was performed using GC/MS operating in a full scan mode. The acids and phenols
were determined by dual column GC/ECD following diazomethane derivatization. Other pesticides were added
to the target list when opportunities arose to monitor their application and when they could be included into the
existing analytical scheme. Additional analytes included 2,4-D isooctyl ester and lindane, which were grouped
with the other organochlorine pesticides; pyrethrins, which were grouped with the other neutral nonchlorinated
pesticides; and Pursuit® which was derivatized then analyzed by GC/MS-SIM.
Table I. List of target compounds by analytical groups.
Target compounds
Organochlorine Pesticides
Atrazine
Alachlor
Metolachlor
Trifluralin
Aldrin
a-Chlordane
y-Chlordane
4,4'-DDD
4,4'-DDE
Organophosphates and Carbamates
Carbaryl
Chlorpyrifos
Diazinon
Acids and PCP
2,4-D Salts
Dicamba
Pentachlorophenol
4,4'-DDT
Dieldrin
Heptachlor
Permethrin
Captan
Dachtal
Dichloran
Folpet
Malathion
Phorate
Propoxur
Sampling media consisted of a-cellulose patches, cotton gloves, polyurethane foam (PUF) plugs, quartz
microfiber filters, handwipes, housedust and surface soil. Detailed descriptions of the sampling media and
sampling methods have been described elsewhere in this proceeding.1 Sampling media, except handwipes, were
extracted for neutral pesticides using Soxhlet extraction with 6% diethyl ether in n-hexane.2 When required, the
neutral extracts were split into two fractions for organochlorine and organophosphate/carbamate determinations.
Cellulose patches, gloves, PUF, and quartz filters were extracted for acids and phenols using 1:1 diethyl ether in
n-hexane acidified to pH 2 with HC1. Soil and housedust were extracted with a modified version of EPA
Method 8150.
Isopropanol handwipe samples, consisting of SOF-WICK cotton gauze wetted with 50 mL of isopropanol,
were extracted with two 50 mL portions of 1:1 ether in hexane. The extract was then concentrated and, when
required, split into fractions for organochlorine pesticide and organophosphate/carbamats determinations. An
additional fraction was then derivatized with diazomethane for determination of acids and phenols.
699
-------
Since the determination of acid and neutral pesticides required separate extractions for all media except
the handwipes; and the gloves, cellulose patches, PUF plugs and quartz filters could not reliably be separated into
portions for neutral and acid extraction, the analytical scheme was simplified by separating the media into two
sampling groups, i.e., those samples that were associated with a mixer/loader/applicator event (MLA samples) and
those that were not associated with an application event (environmental samples). Specifics of the sampling
scheme are described elsewhere in this proceeding.3 Those samples that were associated with an MLA event and
could not be separated were analyzed only for the analytical group that contained the applied pesticide. When
a mixture of pesticides was applied that contained target analytes in more that one analytical group, the EPA
technical project officer selected the analytical group to be targeted for analysis. Environmental samples that could
not be separated were extracted and analyzed for neutral (organochlorine and organophosphorate/carbamate)
pesticides. Samples that could be reliably separated (e.g., housedust and soil) or media that were efficiently
extracted for all analytical groups (e.g., handwipes) were analyzed for all target analytes.
METHOD VALIDATION
A series of experiments were performed to validate the extraction of selected neutral pesticides and acid
herbicides from PUF, a-cellulose patches and gloves. Three replicates of each medium were spiked with 1.125 pg
each of atrazine, alachlor, metolachlor, trifluralin, dicloran, phorate, dicamba, 2,4-D, and pentachlorophenol (PCP).
The acid herbicides 2,4-D, and dicamba were spiked in their free acid form. Each of the spiked media were then
Soxhlet extracted using 6% diethyl ether in hexane. Results for three replicates of PUF, gloves and body patches
are given in Table II. Results are comparable for all three media. Good recoveries were obtained for most of the
neutral pesticides. Trifluralin results were consistently low for all three media, with essentially no recovery
reported from the a-cellulose patch. Low recoveries were observed for phorate from PUF and cellulose but not
the gloves.
Acid herbicide recoveries for the three media were poor. PCP, which is least acidic of the three acids,
was recovered most efficiently. These results are not totally unexpected since the extraction procedure was not
tailored to acid extraction and indicate that in order to obtain good recoveries for neutral pesticides and acid
herbicides from these personal monitoring media, separate extraction methods are necessary.
In order to validate the extraction method for acid herbicides from personal monitoring media, two quartz
microfiber filters were spiked with 75 ng each of dicamba (as the free acid), 2,4-D dimethyl amine salt and PCP.
The filters were extracted with acidified 1:1 ethenhexane as described above, derivatized and analyzed by
GC/ECD. Recoveries of the acid herbicides were calculated and are reported in Table III. Average recoveries
were greater than 50% for all three target analytes.
Method validation studies were conducted on soil and housedust to determine extraction efficiencies for
selected neutral pesticides (using Soxhlet extraction) and acid herbicides (using the modified EPA method 8150)
as described in the Analytical Approach section above. To test extraction of the neutral pesticides, two 2 g
portions of a housedust sample obtained from a home in San Antonio were spiked with selected neutral pesticides.
The spiked portions and two additional housedust portions were then Soxhlet extracted. To test the extraction of
acids, two 20 g portions of soil, obtained from the SwRI grounds, were then spiked and extracted with two 20 g
soil blanks by Soxhlet extraction. Two 2 g portions of the housedust sample were then spiked with 2,4-D,
dicamba and PCP. The spiked portions and two additional housedust portions were then extracted using the
modified EPA Method 8150. Two 20 g portions of soil were then spiked and extracted with two 20 g soil blanks,
also by EPA Method 8150. A list of target analytes, spike levels, soil and housedust blank results and recovery
results are given in Tables IV and V. All neutral pesticides were recovered with good efficiency from soil with
the exception of trifluralin. The acid herbicides dicamba and 2,4-D were extracted with approximately 50%
efficiency. However, PCP recovery was very low on both replicates (3% and 4%).
Recoveries of neutral pesticides from housedust were good with the exception of trifluralin and propoxur
in Replicate I (25% and 33% respectively) and atrazine in Replicate II (41%). Recoveries for PCP were higher
from housedust than from soil (29% and 24%).
700
-------
Table n. PUF, glove, and cellulose patch recovery results (%) (neutral
extraction).
Target Analyte I II in Mean
PUF
Atrazine
Alachlor
Metolachlor
Trifluralin
Dicloran
Phorate
Dicamba
2,4-D
Pentachlorophenol
Gloves
Atrazine
Alachlor
Metolachlor
Trifluralin
Dicloran
Phorate
Dicamba
2,4-D
Pentachlorophenol
Cellulose Patches
Atrazine
Alachlor
Metolachlor
Trifluralin
Dicloran
Phorate
Dicamba
2,4-D
Pentachlorophenol
118
116
83
59
94
69
ND
2
36
129
95
65
52
87
99
ND
14
57
79
112
84
ND
15
51
3
4
57
116
122
88
60
93
69
ND
2
19
133
80
53
46
68
96
ND
ND
45
95
110
81
4
34
63
3
2
63
123
116
86
58
90
58
ND
ND
41
136
127
88
69
109
104
1
2
76
83
117
86
1
27
68
1
1
65
119
118
86
59
92
65
-0-
1
32
133
101
69
56
88
100
1
5
59
86
113
84
2
25
61
2
2
62
Table III. Recoveries for acid herbicides from quartz microfiber filters.
Theoretical final concentration for all extracts was 0.075 ng/uL.
Target Analyte DBS DB608 DBS DB608
Dicamba 60% 52% 74% 66%
2,4-D Salt 42% 45% 57% 64%
Pentachlorophenol 124% 57% 95% 70%
701
-------
Table IV. Recoveries of selected neutral pesticides and acid herbicides from soil.
Soil Blank Results
(ng/g)
Target Analyte
Alachlor
Atiazine
y-Chlordane
Trifluralin
Chlorpyrifos
Diazinon
Propoxur
Dicamba
2,4-D
Pentachlorophenol
I
0.650
<1.00
0.061
<0.250
<0.500
<0.500
<1.000
<8.34
<8.34
<8.34
II
1.30
<1.00
<0.250
0.050
<0.500
<0.500
0.800
1.667
5.00
<8.34
Spike
Level
(ng/g)
10
15
5
5
10
10
10
166.67
166.67
166.67
Soil Spike Results
(ng/g)
I
10.16
18.00
5.34
2.23
13.07
13.89
15.70
83.83
107.96
5.00
II
10.70
18.50
5.08
2.70
13.34
15.55
15.90
81.22
103.26
6.67
% Recovery
I
95
120
107
45
131
119
157
50
64
3
II
94
123
102
54
133
155
151
48
59
4
Table V. Recoveries of selected pesticides and acid herbicides from housedusL
Blank Results
(Pg/g)
Target Analyte
Alachlor
Atrazine
y-Chlordane
Trifluralin
Chlorpyrifos
Diazinon
Propoxur
Dicamba
2,4-D
Pentachlorophenol
I
0.050
0.025
0.286
<0.125
1.054
1.675
0.350
<0.125
<0.125
0.050
II
<0.125
<0.500
0.182
<0.125
1.088
0.939
0.400
0.050
<0.125
<0.1256
Spike
Level
(ug/g)
5.0
7.5
2.5
2.5
5.0
5.0
5.0
2.5
2.5
2.5
HD Spike Results
(ug/g)
I
3.700
7.989
2.891
0.633
7.712
7.737
1.725
1.18
1.73
0.77
II
3.988
3.050
3.086
1.512
8.436
8.556
4.050
1.03
1.47
0.59
% Recovery
I
73
107
104
25
133
121
33
47
69
29
II
80
41
116
60
147
152
73
41
57
24
An experiment was then performed to determine the recovery of selected neutral pesticides and acid herbicides
from isopropanol handwipes. Three SOF-WICK pads were each moistened with 100 mL of isopropanol. A
volume of a spike solution of acid herbicides and neutral pesticides was spiked into the moistened wipe at a level
of each analyte of 1.125 ug. The handwipes were then extracted as described in the Analytical Approach section
above. One portion of the extracts was then derivatized for analysis of acid herbicides. Recoveries for the target
analytes were calculated and are given in Table VI. Unlike the Soxhlet extraction method, good recoveries were
obtained for both neutral pesticides and acid herbicides in a single extraction. As also observed for personal
exposure media, the average recovery for phorate was low (7%).
702
-------
Table VI. Handwipe recovery results (%).
Target Analyte
I
in
Mean
Atrazine
Alachlor
Metolachlor
Trifluralin
Dicloran
Phorate
Dicamba
2,4-D
Pentachlorophenol
96
116
105
63
85
6
65
83
65
107
116
98
64
79
10
65
82
63
89
107
79
56
70
6
78
98
77
97
113
94
61
78
7
69
88
68
AUDIT SAMPLE RESULTS
Blind spike audit samples were provided by AREAL. Four PUF plugs, three quartz filters, three portions of
housedust from field samples and three portions of surface soil from field samples were spiked, then shipped to
SwRI for analysis using the approach described above.
Each of the four PUF plugs was extracted and analyzed for neutral pesticides. Results and biases are given
in Table VII. For samples PE-PF-1 and PE-PF-2, biases were less than +/- 20% except for a large negative bias
for lindane (-33%) for PE-PF-1. All pesticides reported for PE-PF-3 had a large negative bias suggesting that a
dilution error or other systematic error occurred.
Table VH. PUF audit sample results
PF-PE-1 (ng)
Analyte
Atrazine
Alachlor
Chlorpyrifos
Diazinon
Propoxur
Lindane
PCP
Spiked
100
103
100
99
102
108
100
Reported
106
90
98
101
116
72
NA
Bias%
6.0
-13
-2.0
-9.0
14
-33
PF-PE-2 (ng)
Spiked
498
515
499
495
510
540
500
Reported
520
550
480
495
576
640
NA
Bias %
4.4
6.8
-3.8
-2.2
13
19
-
PF-PE-3 (ng)
Spiked
996
1030
998
992
1020
1080
1000
Reported
770
780
776
741
753
870
NA
Bias %
-23
-24
-22
-26
-26
-19
PF-PE-4
(ng)
Reported
-------
Table Vm. Quartz filter audit sample results.
QF-PE-1
QF-PE-2
QF-PE-3
Analyte Spiked Reported Bias '
Spiked Reported Bias '
Spiked Reported Bias %
Dicamba 0.95
NA
47
42
-11
95
108
14
Results of performance evaluation soil and housedust samples are given in Table IX. Recoveries were
consistently low for the neutral pesticide extraction in both media. Since only one replicate of each medium was
analyzed for neutral pesticides, it cannot be determined whether the compounds were extracted with low efficiency
or an error was made in the final dilution. Large biases were observed for acid herbicides results indicating that
an improved extraction and/or cleanup method is required.
Table IX. Soil and housedust audit sample results (pg/g).
Analyte
Soil
Atrazine
Alachlor
Chlorpyrifos
Diazinon
Propoxur
Lindane
Dicamba
Pentachlorophenol
Analyte
Housedust
Atrazine
Alachlor
Chlorpyrifos
Diazinon
Propoxur
Lindane
Dicamba
Pentachlorophenol
Blank
0.106
0.260
0.135
0.316
<0.002
0.005
0.038
0.012
Blank
<0.013
1.081
<0.026
0.051
<0.052
0.006
0.104
0.219
Spiked
0.100
0.103
0.100
0.099
0.102
0.108
0.095
0.100
Spiked
0.100
0.103
0.100
0.099
0.102
0.108
0.095
0.100
IA-SE-4-PEA
Reported Bias %
NA
NA
NA
NA
NA
NA
0.042 -96
0.024 -88
3-HD-4-PEA
Reported Bias %
NA
NA
NA
NA
NA
NA
0.099 100
0.339 640
Spiked
10.00
10.00
10.00
9.92
10.20
10.80
9.46
10.00
Spiked
10.0
10.0
10.0
9.9
10.0
11.0
9.5
10.0
IA-SE-4-PEN
Reported
7.80
7.75
7.60
6.35
9.05
8.00
NA
NA
3-HD^t-PEN
Reported
4.53
4.32
5.68
5.05
4.95
4.48
NA
NA
Bias%
-23
-25
-25
-39
-11
-26
Bias %
-55
-68
-43
-49
-50
-59
-
-
704
-------
CONCLUSIONS
Neutral target pesticides were generally recovered well from spiked PUF plugs, quartz filters, soil, cellulose
patches and handwipes. Recovery of neutral pesticides from housedust was inconsistent and lacked reproducibility.
Good recoveries were obtained for acid herbicides from quartz filters and handwipes but acid recoveries were low
from soil and housedust. The extraction of acid herbicides from gloves and cellulose patches still must be
investigated. Handwipes appear to be a simple sampling method that provides good analytical results for a wide
range of both neutral and acidic compounds. However, the wipe removal efficiency of these compounds from
hands and other surfaces is unknown and is a vital need to assess the utility of the handwipe method as a tool for
pesticide exposure assessment. Studies have recently been undertaken to improve analytical methods for the
determination of pesticides in housedust. Preliminary results for these studies have been presented in this
proceeding.4
Finally, the utility of separate analytical techniques for organochlorine and other neutral pesticides is
questionable. State-of-the-art GC/MS-SIM instrumentation is capable of detection limits close to those obtained
by dual column GC/ECD. This means that all targeted neutral pesticides can be determined with a single
technique with no significant loss of detection limit.
ACKNOWLEDGEMENTS
This work was funded by the U.S. Environmental Protection Agency (Contract 68D10150). The authors wish
to acknowledge Mr. A. E. Bond and Mr. G. G. Akland of US EPA AREAL for their guidance. In addition, the
project could not have been completed without the technical expertise of Mr. T. Mclntire and Ms. M. Ortiz of
SwRI.
Although the research described was funded by the U.S. EPA, it has not been subjected to the required peer
review and does not necessarily reflect the views of the agency; no official endorsement should be inferred.
REFERENCES
1. H. J. Harding, P. M. Merritt, J. M. Clothier et al., "Sample collection methods to assess environmental
exposure to agricultural pesticides" in Proceedings of the 1993 U.S. EPA/A&WMA International
Symposium on Measurement of Toxic and Related Air Pollutants." Air & Waste Management Association,
Pittsburgh, 1993 (in press).
2. J. P. Hsu, H. G. Wheeler, Jr., D. E. Camann et al., "Analytical Methods for Detection of Nonoccupational
Exposure to Pesticides", /. of Chromatographic Science, 26, 181, 1988.
3. D. E. Camann, P. W. Geno, H. J. Harding et al., "Measurements to assess exposure of the farmer and
family to agricultural pesticides" in Proceedings of the 1993 U.S. EPA/A&WMA International Symposium
on Measurement of Toxic and Related Air Pollutants." Air & Waste Management Association, Pittsburgh,
1993 (in press).
4) T. K. Majumdar, D. E. Camann and P. W. Geno, "Analytical methods for screening of pesticides and
polynuclear aromatic hydrocarbons in housedust" in Proceedings of the 1993 U.S. EPA/A&WMA
International Symposium on Measurement of Toxic and Related Air Pollutants." Air & Waste Management
Association, Pittsburgh, 1993 (in press).
705
-------
NCI/EPA AGRICULTURAL HEALTH STUDY (AHS):
DEVELOPMENT OF THE BIOHARKER QUESTIONNAIRE
C.J. Nelson1, Jacquelyn M. Clothier2, Gerald Akland1, Andrew
Bond1
1 U.S. Environmental Protection Agency
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, North Carolina 27711
2 Southwest Research Institute
PO Drawer 28510
6220 Culebra Road
San Antonio, Texas 78284
ABSTRACT;
The National Cancer Institute (NCI), the Environmental
Protection Agency (EPA), and the National Institute of
Environmental Health Sciences (NIEHS) have planned a long-term
prospective epidemiologic study of men, women, and dependent
children in agricultural areas to identify and quantify cancer
risks that may be associated with pesticide usage. This paper
covers the types of questionnaires that are to be used in the
study. The main topic is the EPA developed questionnaire that was
designed to assess total pesticide exposure to the farm family.
The EPA developed a questionnaire to assess previous pesticide
use and practices, exposure while pesticides are being applied, and
any residual that may find its way into the home. The
questionnaire was administered to 3 farm families during the spring
of 1992. Activity logs were kept by participants during the one
week study period. A food frequency log was kept of all food
prepared during the day of application and information was gathered
on the types of food used by the families, both locally produced
and purchased from retail stores. Problems encountered during the
administration of the questionnaire and steps taken to remedy the
problems will be discussed.
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication. Mention of
trade names or commercial products does not constitute endorsement
or recommendation for use.
706
-------
I. BACKGROUND
The "Agricultural Health Study: A prospective Study of Cancer
and Other Diseases Among Hen and Women in Agriculture" (1) is a
collaborative effort between the National Cancer Institute (NCI),
the U.S. Environmental Protection Agency (EPA), and the National
Institute of Environmental Health Sciences (NIEHS). The major goal
of this study is to establish a prospective cohort that can be
followed for 10 years or more to evaluate the role of agricultural
pesticide exposures in the development of cancer, neurological
disease, birth defects, and other chronic disease outcomes. The
prospective cohort of approximately 68,500 men and 6500 women will
be obtained over a three year period from farmers and commercial
pesticide applicators as they obtain or renew their pesticide
application licenses in Iowa or North Carolina.
Assessment of exposure in most previous epidemiologic research
on agricultural pesticides has typically been limited to the names
of chemicals used, job title, and, occasionally, duration in that
job title. Improvements in the area of exposure assessment is one
of the most important features of this prospective study. A small
subset of the large cohort will be selected for detailed exposure
assessment based on the applicators responses to the 'Enrollment
Questionnaire1 (Q0) .
II. QUESTIONNAIRES USED
Currently there are five different questionnaires to be used
in this study. The "Enrollment Questionnaire" , referred to as Q0,
will be administered when the applicators apply for their
restricted-use pesticide license. The Q,, is designed to collect
personal identifiers, brief pesticide use history, and information
on crops grown and livestock raised. A history of tobacco and
alcohol use, diet and employment will also be collected. Each
private applicator that agrees to participate in the study will be
given a take-home questionnaire, the "Farmer Applicator
Questionnaire" [Qu] and one for the applicators spouse, the "Spouse
Questionnaire" [Qlb]. The latter two questionnaires will seek
detailed information on pesticide exposures, work practices, diet
including cooking practices, and other lifestyle factors and health
outcomes. The "Biomarker Questionnaire" [Q2] will collect
information on environmental factors and pesticide practices during
the current year. Activity diaries for an application event,
participants time during monitoring, and pesticide use in the home
will also be collected. The "Case-Control Questionnaire" [Q3] is
a cancer-site specific questionnaire seeking risk factors relating
to specific cancer types and factors important for evaluating
biologic markers. The biomarker questionnaire is the subject of
this presentation.
707
-------
III. BIOMARKER QUESTIONKAIRE
The Atmospheric Research and Exposure Assessment Laboratory
(AREAL) agreed to collect exposure data on a small subset of
approximately 200 farm applicators and their families.
Environmental measurements will be obtained for the applicator, the
applicators spouse, and up to two children living on the farm
during a three day period. The applicators will be selected based
on their responses to the Enrollment Questionnaire. Environmental
samples will be multi-media and multi-exposure consisting of
personal air, indoor & outdoor air, drinking water, house dust,
food, blood, urine, handwipes, dermal patches, and home surfaces &
equipment wipes. A 4-farm pre-pilot study was conducted in May-
June, 1992, where a prototype of the questionnaire was used. After
the pilot study the questionnaire was extensively revised.
One of the uses of the information collected on the
questionnaire is to aid in the interpretation of the environmental
measurements. For example, we found detectable levels of a
pesticide that was not being applied. This may have occurred when
the farmer was performing repairs to application equipment. The
questionnaire will also aid in quantifying the amount of exposure
during different activities such as mixing/loading/handling and
application. Diaries will obtain information to determine average
times for common farm activities. The average times may then be
used in a variety of exposure models.
QUESTIONNAIRE DESIGN
The prototype questionnaire was assembled from various
sources. The framework came from the Non-Occupational Pesticide
Exposure Study (NOPES) (2) and the House Dust/Infant Pesticide
Exposure Study (HIPES) (3) questionnaires. Some of the questions
came from a questionnaire that was used by Dr. Vince Gary, a
pathologist at the University of Minnesota in Minneapolis, in his
study of farmers who applied pesticide on their farms. Dr. Jack
Griffith, Health Effects Research Lab in RTF, also supplied
questions relating to health issues. In addition, Gerry Akland,
Bob Lewis, and Andy Bond, Atmospheric Research and Exposure
Assessment Lab in RTF, reviewed the questionnaire and had suggested
various additions, deletions, and edits.
There were several problems with the prototype. One problem
was with the children's activity log which was designed for pre-
school children and was not suitable for older children. Also
questions about pesticide inventories, how to record the
application event, and the lack of an orderly progression of
questions that went along with what was actually going on at the
study site needed to be revised.
The current version has been extensively revised and is now
composed of six parts. The first part is designed to collect
708
-------
background information, such as the farm's physical features, past
pesticide use, work practices,and an inventory of pesticides on
hand. The second part collects information on a pesticide
application event. The last four parts are activity logs.
Personal activity logs are designed to collect information on the
activities of the farmer, his spouse, and up to two of his children
during the monitoring period. A house activity log was included to
collect any information relating to pesticide application in or
around the house but not including the monitored application event.
The six parts, called the AGRICULTURAL HEALTH STUDY "Biomarker
Questionnaire" [Q2] will be discussed in their order of use during
a study.
Part 1. AHS BACKGROUND SURVEY
The background survey will collect general information
concerning the family, the buildings on the farm, the history of
the farm, the attitude of the family towards pesticides in general,
previous pesticides used, and an inventory of all pesticides in and
around the house. A record of all agricultural pesticides applied
during the current year will be recorded and the type of
application, number of acres treated, etc. The work practices
section collects information on types of protective equipment the
applicator uses, what actions are taken when he/she is directly
exposed, how the pesticides application equipment is repaired or
cleaned. A series of questions is asked regarding the types and
amounts of food grown and consumed by the family on the farm .
There are lists of both acute and chronic diseases that the
respondent may have or had during the current year. This
questionnaire is completed the day before monitoring begins or the
first day of the monitoring period.
Another section, the household pesticide inventory, is
designed to collect information on all pesticides stored in the
house. This includes disinfectants used to clean bathrooms, as
well as any pesticides used for flies or wasps. There is also a
health and medication diary designed to collect information on any
vitamins or over-the-counter or prescription medication taken by
the participants.
Part 2. APPLICATION EVENT ACTIVITY LOG
The application event log is filled out by an interviewer for
the first pesticide application of the day while the applicator is
in the process of preparing the mixture to be used. The applicator
will be monitored during both the mixing/loading/handling
activities and during the application period. Amount to be
applied, type of pesticide and any additives, type or amount of
crops or livestock to be treated, and type of clothing worn
including protective clothing will be recorded. Information will
also be collected after completion of the application, such as how
709
-------
the pesticide containers and any unused mixture was disposed of,
and information about any required repairs.
Part 3. HOUSE ACTIVITY LOG
The house activity log is used to capture information about
any activities that may impact on the monitoring that is being done
in or around the house. Primarily the information collected is
about any pesticide that may have been used in the house, on the
lawn around the house, or in the garden. Information that may
impact on any personal exposure is collected such as who made the
application, what equipment was used, and any precautionary actions
that may have been taken. One question on the method of
ventilation used each day in the home is asked (i.e., windows open,
air conditioner on, etc.).
Part 4-6. FARMER'S, SPOUSE'S, AND CHILDREN'S ACTIVITY LOG
These forms are an aid to the farmer applicator, his spouse
and up to 2 children for remembering and recording their activities
during the monitoring period. Start and stop time of each change
in activity is recorded along with where they were and any
equipment that may have been in use. The diaries will also collect
information on other possible exposures such as painting, welding,
driving diesel vehicles, etc. These diaries will also aid in the
interpretation and analysis of blood and urine data. Family
members are asked to fill the log out on a daily basis. If the
children are too young to fill in the form, a parent will be asked
to complete the daily log.
The activity logs will be summarized into the major activities
that farm families participate in (e.g., the amount of time
watching TV, working in the garden, in areas where pesticides have
been applied or are being applied, and the amount of time farm
applicators spend in other potential health-risk exposures such as
painting, welding, driving diesel tractors, etc.).
SUMMARY
This paper discusses the development of the biomarker
questionnaire that will be used the Agricultural Health Study. The
questionnaire was designed to collect background information on
previous years pesticide usage, collect information on at least one
application event, and collect information on the activities of the
farm family during the three day monitoring periods. Environmental
samples will be multi-media and multi-exposure consisting of
personal air, indoor & outdoor air, drinking water, house dust,
food, blood, urine, handwipes, dermal patches, and home surfaces &
equipment wipes.
710
-------
ACKNOWLEDGEMENTS:
I would like to thank Michael Alavanja at the National Cancer
Institute, Louise Ritz at SRA Technologies and those previously
mentioned in this paper who had input into the design and the
modification of the questionnaire. Thanks also go to the
reviewers for their helpful suggestions and additions and to Elaine
Grose for her coordination of the US EPA contributions.
REFERENCES:
1. Agricultural Health Study, A prospective study of cancer and
other diseases among men an women in agriculture. M.C.R.
Alavanja, A. Blair, E. Grose, & D.P. Sandier. Draft protocol
4 Feb 93.
2. Nonoccupational Pesticide Exposure Study (NOPES). U.S.
Environmental Protection Agency Final Report. EPA/600/3-
90/003, Jan 1990.
3. Field Evaluation of Methods to Assess the Exposure of Young
Children to Pesticides in the Residential Environment. R.C.
Fortmann, L.S. Sheldon, D. Smith, K Perritt, D.E. Camann, D.O.
Hinton, R. Lewis, and A. Bond. Research Triangle Institute
Final Report, RTI/4657-75/03-FR, April 28, 1993.
711
-------
MEASUREMENTS TO ASSESS EXPOSURE OF THE FARMER AND FAMILY
TO AGRICULTURAL PESTICIDES
David E. Camann, Paul W. Geno, H. Jac Harding, and Nicholas J. Giardino
Southwest Research Institute
P.O. Drawer 28510
San Antonio, Texas 78228-0510
Andrew E. Bond
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Rates of dermal and respiratory exposure of four farmers to four herbicides and two insecticides were
measured during nine mixing/loading/application events. Total dermal exposure (if unprotected by clothing)
ranged from 3 to >4800 mg/h (median 18 mg/h) during prolonged herbicide mixing/applications and from 130
to 140 mg/h during brief insecticide applications. Respiratory exposure (median 0.008 mg/h) was 0.4 to 12% of
handwipe dermal exposure (median 0.8 mg/h). Incidental mg levels were seen on gloves on non-application days,
usually to pesticides applied in the prior month. Occupational exposures of the farmer predominated the residential
exposure of the farm family during the study period, as indicated by child handwipes and concentrations in indoor
air, carpet dust, and drinking water.
INTRODUCTION
A prospective Agricultural Health Study of American farmers and their families is planned to investigate
the observed excess of certain cancers among agricultural workers.1 A small pilot study was conducted at four
farms during the herbicide application season to test environmental and biological sampling/analytical methods
for assessing exposure to selected pesticides in common use. This paper presents measurements of agricultural
and household pesticides on gloves, handwipes, dermal patches and personal air to characterize the personal
exposure of the farmer during mixing, loading, and application. Measurements in indoor/outdoor air, dust, soil,
child handwipes, and drinking water are presented to describe the residential exposure of the family.
METHODS
The study was performed at four family farms in southern Minnesota. Farms were selected from
participants in an on-going surveillance study2 of farm pesticide applicators based on planned agricultural use of
a target pesticide during the monitoring period (May 26-June 15), presence of children in the farm family, and
diversity of agricultural practices. Sampling was performed near the end of the herbicide application season.
Herbicides applied by groundboom tractor during the 72-h monitoring period included alachlor and atrazine on
65 acres of corn at Farms 1A and IB; 2,4-D isooctyl ester mixed with a small amount of trifluralin on 40 acres
of wheat at Farm 2; and imazethapyr on 220 acres of soybeans at Farm 3. Observed insecticide applications were
lindane sprayed on 60 hogs at Farm 1A and natural pyrethrins from an aerosol can sprayed inside a hog barn at
Farm 3. No agricultural pesticides were applied in the first 24 h (Day 1). Application of one or more agricultural
pesticides occurred on Day 2 and often continued on Day 3.
The personal exposure of the farmer was monitored on Day 1 as well as during pesticide application events
on Days 2 and 3. Monitored application events spanned consecutive activities with pesticide exposure potential,
including handling, mixing, loading, application, maintenance, and clean-up. Sampling methods are described by
Harding et al.3 During the event, the farmer wore cotton gloves and patches above clothing on both thighs and
the nape of the neck (to crudely estimate dermal exposure to the hands and rest of body without adjustment for
the protective effect of clothing), and a breathing-zone air sampler (to estimate inhalation exposure). After
removing the gloves at the end of the event, the farmer wiped both hands with two isopropanol-laced gauze pads.
712
-------
Daily 24-h (10 a.m. to 10 a.m.) air samples were obtained on polyurethane foam (PUF) simultaneously
inside the farm house and outside from a porch or in the yard. On Day 3, a dust sample was collected from
carpeted floors in the house while surface soil was obtained at the main entryway (door mat or porch) and along
the pathway from the vehicle parking area to the house. Drinking water was collected from the kitchen faucet.
Handwipe samples were also obtained from each toilet-trained child after a pesticide application event.
Glove, patch, PUF, and fine (<150 urn) dust samples were SoxWet-extracted with 6% diethyl ether/hexane
and analyzed for neutral pesticides (4 herbicides, 16 insecticides, and 4 fungicides). Extracts were quantitated by
gas chromatography using dual-column electron capture detection and mass spectrometry selected ion monitoring
and confirmed by GC/MS in full-scan mode. Extraction and analytical methods are presented by Geno et al.4
Reported results are not adjusted for recovery efficiency. Measurement contamination, accuracy and precision as
indicated respectively by field blanks, spikes, and duplicates are presented by Harding et al.3
RESULTS
A total of 79 field and duplicate samples (18 farmer exposure samples, 53 environmental samples, and
8 child handwipe samples) were analyzed for all principal target analytes. The other 37 field and duplicate
samples, taken to measure the personal exposure of the farmer during agricultural application of a specific
pesticide, were only analyzed for the analyte group which included this target applied pesticide. The most
commonly detected analytes were alachlor (93% of field and duplicate samples), atrazine (91%), pentachlorophenol
(86%), dicamba (84%), 2,4-D (78%), lindane (gamma-HCH) (71%), trifiuralin (68%), captan (56%), chlorpyrifos
(55%), phorate (38%), 4,4'-DDT (37%), dieldrin (36%), propoxur (36%), folpet (35%), and metolachlor (31%).
Occupational Exposure of the Farmer
Farmers 1A and IB received substantial exposure to alachlor during mixing, loading, and application of
the formulated mixture (see event 3C for Farmer 1A and events 3A, 3B, and 4A for Farmer IB in Table I).
Fanner 1A received somewhat less exposure in transferring Lasso® from bulk containers during the baseline
Day 1 (event 1A-2). Farmer 3 also received considerable alachlor exposure during normal activities on Day 1
when no pesticides were applied (event 3-2). A month earlier, Farmer 3 had spent 120 hours applying alachlor
to his corn crop. Farmer 2 did not apply alachlor that spring and had low exposure (Table I).
The measured rates of dermal and respiratory exposure which these farmers received during mixing/loading
and application of alachlor, atrazine, 2,4-D isooctyl ester, imazethapyr, lindane, and pyrethrins to crops and
livestock are presented in Table II. If unprotected by clothing, the total dermal exposures of monitored farmers
ranged from 3 to >4800 mg/h (median 18 mg/h) during prolonged herbicide mixing/application and from 130 to
140 mg/h during brief insecticide mixing/application. The crude rates of unprotected dermal exposure to the hands
and the rest of the body were on the same order of magnitude, except for 2,4-D isooctyl ester, imazethapyr and
a Lasso® (alachlor) mixing accident. The gloves represented from 1% to >99% of the total unprotected dermal
exposure. The highest observed exposure occurred during mixing when Farmer IB accidentally saturated one
finger of a glove while opening a Lasso® container. Handwipes revealed residues of pesticides that were 0.1%
to 85% (median 5%) of amounts found on gloves. The respiratory exposure rate was less than 0.1 mg/h, except
for pyrethins applied indoors by aerosol can (0.44 mg/h). The rate of respiratory exposure ranged from 0.4% to
12% (median 0.7%) of the handwipe exposure rate during mixing/application events.
The exposure of the farmer to a pesticide was generally higher by at least several orders of magnitude
while he mixed and applied the pesticide than on baseline nonapplication days. However, in ten instances, the
farmer's gloves contained mg levels of pesticides that had not been mixed or applied that day; in eight cases, the
high incidental exposure was to a pesticide that the farmer had applied in the previous month.
713
-------
Table I. Alachlor amounts (ug) in personal exposure monitoring of fanners by event and matrix.
Fanner-
event
code
Activity
codes"
Activity
duration
(h)
Glove
pair
Handwipe
Mid
event
End of
event
Three Personal
dermal air cone.
patches
(ug/m3)
Fanner 1A
1A-2 T
1A-3C MA
1A-4B R
8.53
2.60
0.75
18,500
59,800
2,610
270
250
480
92
51
380
17
2.6
13.7
0.49
Fanner IB
1B-3A
1B-3B
MA
MA
MA
3.03
5.38
3.28
105,000
270,000
>2,370,000b
6,380
14,800
9,540
2,050
5,400
1,020
7.6
11.1
20.3
Fanner 2
2-2
2-3 R
2-4
Farmers
3-2
3-3
3-4A
3^B
7.78
3.00
6.40
6.90
7.02
0.05
7.20
11.5
ND
ND
4,470
NA
NA
NA
0.03 0.06
0.44
0.08
36
NA
NA
NA
0.005
0.056
0.043
19.7
NA
NA
NA
ND
0.03
ND
0.43
NA
NA
NA
ND = Not detected
NA = Not analyzed
a T = transfer from bulk container, M = mixing and loading; A -- application; R = residue applied
since trace found in analysis of formulated mixture.
b Gloves removed after one finger of glove was dipped in Lasso® (alachlor) during mixing
30 min after event began. Alachlor result is underestimate, since crystalline precipitate formed
in glove extract.
Residential Exposure of the Farm Family
Measurements indicative of residential exposure to pesticides at Farmhouse 1A are presented in Table ffl.
Air concentrations were generally higher indoors than outside the farmhouse, although similar indoor and outdoor
levels were observed at Farmhouse 1A for the applied analytes atrazine and lindane. Agricultural pesticides were
usually present at higher concentrations in indoor carpet dust than in entryway or pathway soil. The prominent
residues recovered in handwipes of the child were also prominent in the indoor air or dust (alachlor, atrazine,
captan; chlorpyrifos, and pentachlorophenol at Farmhouse 1A).
714
-------
Table II. Rates of dennal and respiratory exposure (mg/h) of farmers to applied analytes from
personal exposure monitoring by event and matrix.
Dennal Exposure
Farmer-event
code
Activity
codes*
Activity
duration
00
Hands
Glove Handwipe
pair6 after event
Rest of
bodyb-°
(dermal
patches)
Respiratory
exposure11
(personal
air)
Alachlor applied to corn from groundboom tractor by Farmers 1A and IB
1A-3C MA 2.60 23.0 0.18 30.7
1B-3A MA 3.03 34.6 2.10 146
1B-3B MA 5.38 50.2 2.75 217
IB^tA MA 3.28 >4740e 2.91 67.2
Atrazine applied to corn from groundboom tractor by Fanners 1A and IB
1A-3C MA 2.60 2.02 0.043 9.21
1B-3A MA 3.03 0.78 0.66 9.77
1B-3B MA 5.38 6.17 0.80 11.88
1B-4A MA 3.28 32.0" 1.15 24.6
2,4-D Isooctyl ester applied to wheat from groundboom tractor by Fanner 2
2-3 MA 3.00 9.77 0.051 0.58
Imazethapyr applied to soybeans from groundboom tractor by Farmer 3
0.025
0.014
0.020
0.037
0.005
0.003
0.004
0.008
0.002
3-3 MA 7.02 3.30 0.023
3-4B MA 7.20 2.86 0.002
Lindane applied to hogs from hose sprayer by Fanner 1A
1A-4B MA 0.75 28.3 2.11
Pyrethrins applied in hog barn from aerosol can by Farmer 3
3^tA A 0.05 50.6 9.0
0.055
0.048
105
88
ND
ND
0.008
0.44
ND = Not detected
a M = mixing and loading; A = application.
b Not adjusted for protective effect of clothing.
c Scaled to total body surface excluding hands from dermal patch amount based on surface
area.
d Based on a farmer breathing rate of 1.8 m3/h, assuming a moderate workload.
e Gloves removed after one finger of glove was dipped in Lasso® (alachlor) during mixing
30 min after event began.
DISCUSSION
This pilot study employed a multiresidue analytical method and a multimedia sampling design to assess the
exposure of the farmer and family to agricultural pesticides. The principal exposure of the farmer was to
pesticides which he mixed and applied. The proportion of unprotected dermal exposure attributed to hand
exposure (7% to >99%) exceeded the 27% to 99% range in studies reviewed by Franklia5 The protection
afforded by cotton work gloves was highly variable: handwipes uncorrected for wipe removal efficiency showed
that 0.1 % to 85% of the residual glove amount had penetrated to the skin and was wiped from the hand afterward.
Cotton gloves provided Farmer IB trivial hand protection during mixing/loading/groundboom tractor application
715
-------
of alachlor and atrazine: handwipes revealed similar residues on bare (event 4A) and gloved (events 3A and 3B)
hands. The multiresidue analysis of glove samples established that the farmer occasionally receives incidental
occupational exposures of 1 mg/h, usually to pesticides which he had applied in the prior month. Uncorrected
handwipe dermal exposure dominated respiratory exposure during every application event, indicating that the
dermal pathway may be the principal route of exposure for these Minnesota farmers, as found in prior studies.5
Table III. Indicators of residential exposure to pesticides at Farmhouse 1A.
Mean (n=3) air
concentration
(ng/m3)
Pesticide
Alachlor
Atrazine
Captan
Chlordane (a + y)
Chlorpyrifos
ZDDT (DDT+DDD+DDE)
Dacthal
Diazinon
Dicloran
Dieldrin
Folpet
Lindane
Metolachlor
Pentachlorophenol
Permethrin (cis + trans)
Phorate
Propoxur
Trifluralin
Indoors
141
12
3.7
0.1
9.0
0.9
0.1
0.3
15
7.1
6.8
0.3
0.7
Outdoors
45
11
0.1
0.1
0.8
0.3
1.0
15
1.0
0.6
0.1
0.1
Concentration (ng/g)
Fine
carpet
dust
2,450
1,150
768
10
90
201
149
12
103
8
18
10
Entry-
way
soil
261
106
0.3
135
131
0.3
316
65
5
12
37
1.3
0.1
Path-
way
soil
163
83
104
0.2
11
2.3
0.3
0.3
Handwipe of
child (age 3)
(ng)
Day 2 Day 3
221 411
199 111
10 322
179 ,
6 5
2
2
99 34
39 26
Drinking
water
cone."
(ng/L)
2
113
2
1
Blank = not detected
a Obtained from faucet at parents' neighboring Farmhouse IB. t
Air concentrations of recently applied herbicides were usually higher inside the farm house than outdoois.
Some elevations of outdoor and indoor air concentrations were concurrent with an agricultural application.6 Indoor
elevation of lindane at Farmhouse IB was traced to volatilization of residues transported on the work clothing or
tracked in on the shoes of Farmer 1A during his visit following lindane spray application to his hogs.6 Aerosol
drift from upwind spraying appeared to elevate the outdoor air concentration in two instances.6
During the study period, occupational exposure of the farmer to agricultural pesticides greatly exceeded
the residential exposure of his family. The personal air concentration breathed by the farmer during application
events exceeded the home indoor air concentration by one to several orders of magnitude. Amounts of applied
pesticides were three orders of magnitude higher in handwipes of the farmer than of his child.
716
-------
The cost of the multiresidue analysis limited the number of samples, so that exposures were determined
over entire events, rather than for component mixing/loading, application, and repair activities. Monitoring data
are needed to assess relative exposure for these component activities. The three patch samples were extracted
together to yield a single crude estimate of non-hand unprotected dermal exposure. Correction for the protective
effects of clothing dictates that representative patches over typical bare skin areas (lower arms/face/neck) and over
clothed areas be extracted separately. A means to adjust clothing surface exposure for pesticide penetration of
clothing is also needed. The use of gloves greatly exaggerates hand exposure due to absorption of liquids7, while
handwipes cannot recover pesticide which has been absorbed or irreversibly bound in layers of the skin.
Experimental determination of wipe removal efficiency for a spectrum of agricultural pesticides may provide a
valid basis to rely on the handwipe in place of the glove sample to estimate hand exposure.
ACKNOWLEDGEMENTS
We acknowledge Vincent F. Garry, M.D. for recruitment of participant farms. This research was funded
by the U.S. Environmental Protection Agency (Contract 68D10150). This paper has been submitted to EPA's peer
and administrative review, but no official endorsement should be inferred.
REFERENCES
1. A. Blair, S. H. Zahm, N. E. Pearce et al.,"Clues to cancer etiology from studies of farmers," Scand. ). Work
Environ. Health 18:209-15 (1992).
2. W. T. Potter, V. F. Garry, J. T. Kelly et al., "Radiometric assay of red cell and plasma cholinesterase in
pesticide appliers from Minnesota," Toxicol. Appl. Pharmacol. 119:150-55 (1993).
3. H. J. Harding, P. M. Merritt, J. M. Clothier et al., "Sample collection methods to assess environmental
exposure to agricultural pesticides" in Proceedings of the 1993 U.S. EPA/A&WMA International Symposium
on Measurement of Toxic and Related Air Pollutants." Air & Waste Management Association, Pittsburgh,
1993 (in press).
4. P. W. Geno, D. E. Camann, K. Villalobos, and R. G. Lewis, "Analytical methods for assessing the exposure
of farmers and their families to pesticides," in Proceedings of the 1993 U.S. EPA/A&WMA International
Symposium on Measurement of Toxic and Related Air Pollutants." Air & Waste Management Association,
Pittsburgh, 1993 (in press).
5. C. A. Franklin, "Occupational exposure to pesticides and its role in risk assessment procedures used in
Canada," Dermal Exposure Related to Pesticide Use; R. C. Honeycutt, G. Zweig and N. N. Ragsdale, Eds.
ACS Symposium Series 273, American Chemical Society, Washington, 1985, pp 429-44.
6. D. E. Camann, P. W. Geno, H. J. Harding et al., "A pilot study of pesticides in indoor air in relation to
agricultural applications" in Proceedings of the Sixth International Conference on Indoor Air Quality and
Climate". Helsinki, 1993 (in press).
7. J. E. Davis, E. R. Stevens, D. C. Staiff, "Potential exposure of apple thinners to azinphosmethyl and
comparison of two methods for assessment of hand exposure," Bull. Environ. Contam. Toxicol. 31:631-8
(1983).
717
-------
Collection and Analysis of Duplicate Diet Samples:
A Pilot Study on Fanner Exposures to Pesticides
Kent W. Thomas, Linda S. Sheldon, Jeffrey T. Keever, and J. Michael Roberds
Analytical and Chemical Sciences
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, N.C. 27709
Paul W. Geno
Southwest Research Institute
San Antonio, TX 78228
Maurice R. Berry Jr.
Environmental Monitoring Systems Laboratory
United States Environmental Protection Agency
Cincinnati, OH 45268
ABSTRACT
A pilot study for the NCI/EPA Farm Occupation Exposure Study (NEFOES), recently renamed the
Agricultural Health Study (AHS), was conducted to provide a preliminary assessment of a multi-media sampling
approach for performing pesticide exposure measurements of farmers. Tests of the dietary component of the
study included collection of 24-hour duplicate diet samples of all farmer-consumed foods and beverages at two
farms. Drinking water was collected at three farms. Food frequency, meal selection, and food diary survey
instruments were tested for recording food consumption and characterizing dietary patterns. Samples were
analyzed for chemicals selected from a target list of 28 insecticides, fungicides, and herbicides. Pesticides
were extracted from aliquots of food homogenates using FDA Method 211.13 and gel permeation
chromatography. Extracts were analyzed by GC/MS in the selected ton monitoring mode and by GC/ECD.
One beverage sample from each farm was extracted and analyzed for acid herbicides using EPA Method 515.
A second beverage sample was extracted by continuous liquid/liquid extraction, then analyzed by GC/MS and
GC/ECD without further clean-up. Water samples were extracted using EPA Methods 507, 508, and 515, then
analyzed by GC/MS. No pesticides were measured above the estimated quantifiable limits in the food and
beverage samples. Atrazine was measured at 0.15 ppb in the drinking water from one farm.
INTRODUCTION
The U.S. Environmental Protection Agency (USEPA), in collaboration with the National Cancer Institute,
sponsored a pilot study to test methods for assessing farmer exposure to 28 selected pesticides (listed in Table
1) commonly used in agriculture. Overall objectives for the pilot study included testing a multi-media sampling
design, several questionnaires, and sample collection and analysis methodology for the inhalation, dermal, and
ingestion exposure routes. The duplicate diet method tested in this study was designed to collect combined
food samples representing total dietary intake of foods and/or beverages over a specified period of time. Only
a few studies have been reported applying duplicate diet (or similar) methods to organic chemical
contaminants1'2. Therefore, the primary objective of the dietary sampling was to make a preliminary test of the
procedures and to identify additional work needed to refine the methods. All foods and beverages consumed
by the farmers at two farms were collected over single 24-hour periods. The composite foods and composite
beverages were homogenized, extracted, and analyzed using gas chromatography/mass spectrometry (GC/MS)
and gas chromatography/electron capture detector (GC/ECD) methods. Preliminary versions of a food diary
and dietary questionnaires were used to identify and quantify the foods collected during the study and to test
methods that may be used to characterize dietary patterns. Drinking water samples were collected at three
study farms using EPA methods and analyzed for most target pesticides.
718
-------
METHODS
Study Design
Four farms were included in the pilot study, numbered 1A, 1B, 2, and 3. Farms 1A and 1B were on
adjacent land that was jointly farmed by a father and his adult son. The dietary component of the study was
meant only to provide a preliminary test of methods, so food and beverage samples were collected from the
farmer only at Farms 2 and 3. Drinking water samples were collected from Farms 1B, 2, and 3 (Farm 1B was
selected because of reported handling of pesticides near the well).
Questionnaires
Two questionnaires were administered to the farmers providing the duplicate diet samples. The food
frequency component of the NCI Hearth Habits and History Questionnaire was self-administered by the farmer
in 25 to 35 minutes. This questionnaire was designed as a dietary recall history, recording the frequency of
consumption of common individual foods3. A meal selection questionnaire was designed to test the ability to
define meals normally eaten by the farmer and the frequency of consumption. It was applied by an interviewer
in 10 to 20 minutes. The purpose in this pilot study was primarily to test burden of administering these
questionnaires. Similar questionnaires may be used in future studies to assess if the collected foods are part of
the normal diet.
Food Collection and Analysis
The farmers at Farms 2 and 3 were instructed to prepare a second plate with the identical type and
amount of food eaten at each meal or snack. Additional servings or leftovers were to be added or subtracted
as necessary. The farmers were also instructed to record the name of each food on the food diary, and
measure the portion size with common household measures (measuring cups, rulers, etc.). The farmers'
spouses were included in the training session and assisted in collecting the foods and completing the food
diary. Duplicate portions of all foods were collected by the farmers from midnight to midnight on the third study
day. At the end of every meal or snack, all foods from the second plate were combined into one 4-L glass jar.
The jar was stored in a foam liner inside a cooler with ice packs. Food samples were transported to the
laboratory within one day and were homogenized within one day of receipt. The entire food sample was placed
in a 4-L stainless steel blender container. Clean water was added incrementally to the sample while blending
15 seconds at a time at the low speed setting. Just enough water was added to achieve thorough
homogenization; the amount depended on the original water content and the weight of the collected foods. A
total of 200 ml was added to the 871 g sample from Farm 2, and 400 mL was added to the 520 g sample from
Farm 3. Samples were blended for 2 minutes at low speed and two minutes at high speed. Aliquots of the
homogenate (50 g) were transferred to glass jars and stored at -20°C.
Three multi-residue extraction methods, selected from the FDA Pesticide Analytical Manual4, were
tested to determine method performance with composite food samples and spiked reagent water prior to
extraction of the farm samples. These methods included 211.13, method 232.4 (modified to include a
petroleum ether extraction), and method 212.13. None of the methods was applicable to the 2,4-D, dicamba,
or pentachlorophenol, which were not measured in the food samples. Method 211.13 resulted in the highest
overall recoveries for most of the target pesticides and was therefore used with a gel permeation
chromatography (GPC) clean-up to extract 50 g portions of the collected food samples. Each food sample was
mixed with potassium oxalate and 50 mL of methanol, extracted three times with 50 mL (1:1) diethyl ether and
petroleum ether. The extracts were dried through anhydrous sodium sulfate, and were cleaned using a
Phenogel Prep-100 GPC column eluted with methylene chloride. High concentrations of fat and other
coexlractive interferences remained in the extracts, which were cleaned again using GPC. The GPC effluent
was concentrated to 2.0 mL and analyzed by GC/ECD. The sample was then further concentrated to 0.2 mL
and analyzed by GC/MS in the selected ion mode. The conditions for GC/ECD and GC/MS were similar to
those described in EPA Method 508s. Samples were fortified with the target pesticides to assess recoveries,
and method blanks were used to assess background levels.
Beverage Collection and Analysis
Beverage samples were collected in a second 4-L glass jar using the duplicate diet methods described
above for foods. Samples were returned to the laboratory, where they were mixed and siphoned into clean
glass jars and stored at -20°C. Sample aliquots (60 g) were extracted for acid herbicides according to EPA
Method 515.15 after dilution to 1.0 L with ether-washed and acidified water (pH < 2). Acid herbicide extracts
were derivatized with diazomethane. Sample aliquots (60 g) were extracted for neutral pesticides after dilution
to 1.0 L with water (pH adjusted to 8.2-8.9). Extraction was performed with methylene chloride in a continuous
719
-------
liquid-liquid extractor for 18 hours. Organochlorine and acid herbicide samples were analyzed by dual-column
GC/ECD. Analysis of carbamates and organophosphorus pesticides, and confirmation analysis of
Organochlorine pesticides and acid herbicides, was performed by GC/MS in the selected ion mode. Samples
were fortified with ten of the target analytes to assess recoveries, and method blanks were used to assess
background levels.
Drinking Water Collection and Analysis
Private wells supplied drinking water at all farms in the pilot study. Drinking water samples were
collected from the kitchen tap at Farms 1B, 2, and 3 and analyzed according to EPA Methods8 507/508 (neutral
pesticides) and 515.1 (acid herbicides). The target pesticides were analyzed by GC/MS after extraction.
Method blanks and method controls were used to assess the background and recovery of the target pesticides.
The EPA Methods did not apply to carbaryl, chlorpyrifos, malathion, phorate, propoxur, captan, dichloran, and
folpet. All except chlorpyrifos and propoxur were tested using the Method 507/508 extraction with GC/MS
analysis. Duplicate water samples were analyzed at a second laboratory by GC/ECD and GC/MS to confirm
the results.
RESULTS AND DISCUSSION
Food and Beverage Collection
Duplicate diet 24-hour samples of foods and beverages were successfully collected by the farmers at
both farms. The 4-L glass jars and coolers were adequate for holding and storing the collected sample
volumes. Food samples are bulky and difficult to store and ship in glass containers, large studies may require
on-site sample processing and homogenization. Additional work should be performed on methods of identifying
and collecting individual food items of local origin, especially those subject to pesticide contamination.
Food Diary and Questionnaires
The farmers (with help from their spouses) were able to complete the food diaries, but providing
detailed descriptions of ingredients and estimating portion sizes with household measures proved to be very
time consuming. The farmers were observed to serve their original plates using measuring cups, possibly
changing the amount of food that would have normally been consumed. It is recommended that the diary
recording and food portion estimates be greatly simplified in future studies. Additional work has shown that the
participant effort can be greatly lessened; foods can be stored separately by the participant, then the
descriptions and size estimates can be performed by the research staff, if necessary. The farmers were able to
complete the food frequency questionnaire with few problems. It was difficult to administer the meal selection
questionnaire in a short time period, and both farmers had trouble defining several typical meals that would
characterize their diets. Neither the food frequency questionnaire nor meal selection questionnaire would be
adequate to assess whether the collected food is typical of the normal diet for a collection period of one day.
Alternative questionnaires that require only 5 minutes to administer, and ask simple questions about dietary
changes and meal consumption frequency, have been developed and tested.
Food and Beverage Analysis
None of the target pesticides were measured in the food or beverage samples above the estimated
method quantifiable limits (EMQL) (Table 1). EMQL values of 3 ng/kg or less were achieved for most of the
target pesticides. Recoveries above 65% were achieved for 14 of the target pesticides from 50 g portions of
the food sample homogenates fortified at 6 jig/kg (approximately two times the EMQL). Recoveries above 50%
were measured for four other pesticides. Low recoveries were noted for atrazine, captan, folpet, and propoxur.
Losses of the more volatile pesticides may have been due to the many sample manipulations required to
remove interferences. Interferences also presented problems in identifying and quantifying heptachlor, cis-
permethrin, and malathion. In most cases the interferents were high concentrations of fatty acids not removed
by the clean-up. Additional method development is currently being directed towards minimizing sample
handling, eliminating or reducing interferents, improving recoveries, and lowering limits of detection. Beverage
samples were extracted and analyzed without the need for time-consuming clean-up steps. The quantifiable
limits were very good for 60 g samples, particularly for the Organochlorine pesticides analyzed by GC/ECD.
Additional recovery studies need to be performed using all of the target analytes in samples with different
beverage compositions, ft is also recommended that if a separate drinking water analysis is performed that
water not be collected as part of the duplicate diet sample.
720
-------
Drinking Water Analysis
Results for the analysis of the target pesticides in the farm drinking water are presented in Table 2.
Atrazine was measured at 0.15 ppb at Farm 1B. Atrazine was reported to have been mixed and loaded into
the sprayer within 6 m of the well at this farm on one or more occasion. No other pesticides were measured at
levels above the EMQL The EMQLs ranged from 0.02 to 0.05 ppb for analysis by GC/MS. Small amounts of
2,4-D and pentachtorophenol were measured in the method blanks. Recoveries from method controls (clean
water fortified with target pesticides) were above 75% for all compounds except pentachtorophenol (62%),
captan (34%), and folpet (3%). Recoveries above 130% were observed for carbaryl and cis-permethrin.
CONCLUSIONS
The primary objective of the duplicate diet component of the NEFOES study was to provide a
preliminary test of the sample collection and analysis methodology. Duplicate diet samples of foods and
beverages were successfully collected and homogenized. Additional method development is needed for the
identification and collection of food items of local origin to identify potential sources of dietary exposure.
Estimating portion sizes with common household measures and providing a detailed food record was very time
consuming for the participant and possibly resulted in nonrepresentative samples. More refined methods have
been tested that will tower participant burden. Questionnaires may be used to evaluate whether the collected
food samples are part of the normal diet for the participant, but if samples are collected for only one 24-hour
period then it is likely that only general questions can be applied with subjective results. No target pesticides
were observed above the EMQLs in the food or beverage samples from two farms. Atrazine was present in the
drinking water at one farm. The extraction and analysis methods for both composite foods and beverages
performed well, with tow detection limits and acceptable recoveries for most target pesticides. Additional
method development is ongoing to reduce interferences, improve analyte recoveries, and to reduce sample
handling steps in the food extraction and analysis methods.
ACKNOWLEDGEMENT
This material has been funded wholly or in part by the Environmental Protection Agency under Contract
68-C2-0103 to Research Triangle Institute. It has been subjected to the Agency's review, and has been
approved for publication. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
REFERENCES
1. P.L. Lioy, J.M. Waldman, A. Greenberg, R. Harkov, and C. Pietarinen, "The Total Human
Exposure Study (THEES) to Benzo(a)pyrene: Comparison of the Inhalation and Food Pathways,"
Arch, of Environ. Health 43(4):304-312. 1988.
2. Y. Ohgane, K. Hiroshima, K. Tamakawa, Y. Mishima, T. Seki, and A. Tsunoda, "Study on the
Daily Intake of Various Chemicals through Daily Foods by the Duplicate Portion Method. II.
Regional Difference of Chtordane Intake." Eisei Kaqaku 35(2):172-176,1989.
3. G. Block, M. Woods, A. Potosky, and C. Clifford, 'Validation of a Self-Administered Diet History
Questionnaire Using Multiple Diet Records." J. Clin. Epidemiol. 43:1327-1335 (1990).
4. Pesticide Analytical Manual Volume I. Methods Which Detect Multiple Residues. Food and Drug
Administration, U.S. Department of Health and Human Services, Washington, D.C., 1989.
5. Methods for the Determination of Organic Compounds in Drinking Water, EPA-600/4-88/039,
U.S. Environmental Protection Agency, EMSL, Cincinnati, Ohio, 1988.
721
-------
TABLE 1. METHOD PERFORMANCE FOR PESTICIDES IN FOODS AND BEVERAGES
Composite Beverage Samples (60 g)
% Recovery
Samples Spiked' EMQL
Pesticide at 1-8 ug/kg, ug/kg
INSECTICIDES
Aldrin
Carbaryl
alpha -Chlordane
garnma-Chlordane
Chlorpyrifos
Diazinon
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Lindane
Heptachlor
Malathion
cis-Permethrin
trans-Permethrin
Phorate
Propoxur
HERBICIDES
Alachlor
Atrazine
Dicamba
2,4-D
Metolachlor
Trifluralin
FUNGiaDES
Captan
Dacthal
Dichloran
Folpet
Pentachlorophenol
37 ± 3
NT
52 ± 2
NT
120
138 ± 2
NT
NT
NT
NT
NT
NT
NT
NT
NT
NT
159 ± 8
137 ± 11
99 ± 24
64 ± 6
42 ± 16
NT
NT
NT
NT
NT
NT
37 ± 6
0.08
1
0.08
0.08
1
1
0.08
0.08
0.08
0.08
0.08
0.08
1
0.08
0.08
1
1
1
1
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.42
0.08
Method
Blank
ug/kg
ND
0.07
ND
ND
0.07
0.05
ND
ND
0.04
ND
ND
0.01
ND
ND
ND
ND
0.03
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Composite Food Samples (50 g)
% Recovery
Samples Spikedb* EMQL
at 6 ug/kg pg/kg
73
52
86 ±
61 ±
83 ±
109
82 ±
102
80 ±
INT
79 i
INT
165
INT
54
58 ±
35
92 ±
2
NT
NT
111
52 t
0 ±
83 ±
57
0 ±
NT
11
9
30
8
26
18
17
3
11
0
12
0
1.5
3
1
2
1
3
1
1
1
1
1
INT
6
NR
NR
1
2
2
NR
NT
NT
2
1
1
1
2
NR
NT
Method
Blank
ND
ND
1.7
0.8
ND
ND
ND
ND
ND
ND
ND
27
ND
ND
ND
ND
ND
ND
ND
NT
NT
ND
ND
ND
ND
ND
ND
NT
ABBREVIATIONS
EMQL = Estimated method quanitation limit ND = Not detected NT = Not tested
INT = Interferent NR = Not reported, low recovery
* Mean and difference for one sample from Farm 2 and one sample from Farm 3.
b Values without differences represent one sample from Farm 3 analyzed by GC/MS.
c Values with standard deviations represent five samples; two each from Farms 2 & 3
analyzed by GC/ECD and one from Farm 3 analyzed by GC/MS.
722
-------
TABLE 2. RESULTS OF GC/MS ANALYSIS OF PESTICIDES IN FARM DRINKING WATER
Pesticide
INSECTICIDES
Aldrin
Caibaiyl
alpha-Chlordane
gamma-Chlordane
Chlorpyrifos
Diazinon
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Lindane
Heptachlor
Malathion
ds-Pennethrin
trans-Pennelhrin
Phorate
Propoxur
HERBICIDES
Alachlor
Atrazine
Dicamba
2,4-D
Metolachlor
Trifluralin
FUNGICIDES
Captan
Dacthal
Dichloran
Folpet
Pentachlorophenol
EPA
Method
508
None
508
508
None
507
508
508
508
508
508
508
None
508
508
None
None
507
507
515
515
507
508
None
508
None
None
515
EMQL
Pg/L
0.025
0.025
0.025
0.025
NT
0.025
0.025
0.025
0.025
0.025
0.025
0.025
0.025
0.025
0.025
0.025
NT
0.025
0.025
0.020
0.050
0.025
0.025
0.025
0.025
0.025
0.025
0.020
Method
Blank
Pg/L
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
NT
ND
ND
ND
<0.05
ND
ND
ND
ND
ND
ND
<0.02
Method Control
% Recovery
(3 or 4 tests)
79
148
91
89
NT
112
94
90
100
92
90
90
116
133
117
88
NT
99
107
77
83
105
75
34
93
97
3
62
±
±
±
X
t
±
1
1
t
±
±
±
±
±
±
±
±
±
i
±
±
±
±
±
t
9.6
26
72
7.8
5.6
10
7.1
8.0
5.1
4.4
8.5
7.8
21
16
5.2
7.5
3.5
2.0
4.0
8.6
9.5
1.9
5.6
0.7
16
Water Samples. ue/L*
Farm IB
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
NT
ND
0.15
ND
ND
ND
ND
ND
ND
ND
ND
ND
Farm 2
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Farm 3
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
NT
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ABBREVIATIONS
EMQL = Estimated method quantitation limit ND = Not dectected NT = Not tested
'Analysis by GC/MS at one laboratory; confirmed by analysis at a second laboratory.
723
-------
Session 20
Acid Aerosols and
Philadelphia Results
-------
EXPOSURES TO ACID AEROSOLS AND GASES
IN SCHOOLS AND YOUTH CENTERS OF PHILADELPHIA
Jed M. Waldman, Chris S-K. Liang and Abdul Kitto
Environmental & Occupational Health Sciences Institute (EOHSI)
Division of Exposure Measurement & Assessment
681 Frelinghuysen Road, Piscataway, NJ 08855
Petros Koutrakis and George Allen
Harvard School of Public Health
665 Huntington Avenue, Boston, MA 02115
Robert Burton and William E. Wilson
U.S. Environmental Protection Agency (MD-56)
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, NC 27711
ABSTRACT
Aerosol species and gaseous components were measured utilizing Annular Denuder Systems
(ADS) during June-August of 1992 inside 15 different microenvironments (schools and youth centers)
in Philadelphia. For two or more weeks each, daily (12-h) samples were collected, starting at 8:00 am.
Infiltration rates were monitored at each site using a perfluorocarbon tracer (PFT) technique. A sampler
was operated on a nearby rooftop to provide matching outdoor data. Indoor levels of aerosol sulfate and
acidity were lower than outdoor concentrations, while indoor NH3 and HONO exceeded their levels
outdoors. Indoor acid aerosols were found to be correlated with the outdoor concentrations, but were
substantially neutralized by indoor ammonia. Air exchange rates were generally hi the range 1-6 h~'
The data will be used to assess whether the magnitude of children's exposures to acid aerosols pose a
potential health risk and to investigate neutralization processes caused by indoor ammonia.
INTRODUCTION
Summer is the time of year when photochemical smog occurs at its highest levels. Along with
ozone, acid aerosols add to the atmospheric burden of lung irritants. While most people would like to
spend then- summer days outdoors, the reality is that most people are indoors much of the day. This
is especially true for young people, who are often being cared for at institutional facilities, such as
schools, daycares or recreational centers.
For most ambient contaminants, indoor spaces are protective, because they can reduce the
exposures by limiting penetration and/or stability. As a secondary pollutant, sulfate aerosols occur in
the accumulation mode, for which deposition is least effective. What has been found is that sulfate
aerosols penetrate indoors with great efficiency. Furthermore, depositional losses are very low, so the
infiltration efficiency for sulfate aerosols is high: frequently, indoor concentrations average close to
100% of the outdoor levels. Sulfate aerosol concentration averaged 0.96 of outdoor levels in Boston
homes (Brauer et al., 1991). Penetration efficiencies for office buildings measured by Li and Harrison
(1990) averaged 0.82. Because aerosol strong acidity is singularly associated with the sulfate component
and with sizes in the 0.1-0.5 pmad range (Koutrakis et al., 1989), it follows that penetration of aerosol
acidity is also very high. However, the effective infiltration depends upon chemical loss terms, since
the physical losses from deposition (as for sulfate) are low.
727
-------
Neutralization of acidic aerosol by ammonia appears to control the levels occurring indoors.
Levels of indoor ammonia are much higher than those measured outdoors. Humans (and pets) are a
principal source of ammonia in occupied spaces; breath and sweat are highly concentrated NH3 sources.
The use of ammonia-containing cleansers can also contribute in some settings. Hence, it is the presence
of people and their activities which creates the ammonia-laden atmospheres indoors. The infiltrated
acidic sulfate aerosols and high indoor ammonia levels have ample time for the neutralization reaction
to go to completion. However, data from indoor studies (Brauer et al., 1991; Liang and Waldman,
1992) indicate the reaction rate occurs in the range of 15 to 90 min, far longer than laboratory data for
pure components suggest (Huntzicker et al., 1980).
U.S. EPA recently began a program to characterize acidic aerosol exposures for metropolitan
areas. In the spring of 1992, Harvard School of Public Health and Robert Wood Johnson started a
multi-site field monitoring program in and around Philadelphia, using measurements for a network of
ambient sites and indoor monitoring in homes, offices and schools. This paper provides a preliminary
report on the measurements for acidic aerosols utilizing ADS during June-August of 1992 inside 15
different microenvironments (schools and youth centers).
METHODS
Acidic aerosol samplers were operated indoors at a set of public school and recreational
buildings. The sites are described in Table 1; their locations are shown in Figure 1. The samplers were
situated inside occupied rooms and operated for 12-h (8 am to 8 pm LDT). Three sites were monitored,
Monday to Saturday, for 2 to 3 weeks at a time. For each interval, an outdoor sampler was operated
on the roof (generally 6-10 m above ground level) at one of the three sites.
"igure 1. Locations of monitoring sites.
Table 1. Sites and Schedules for Summer 1992.
SCHOOLS
NO. PHILADELPHIA June 1 to 15
Lincoln Senior High School HS
Austin Meehan Middle School MS/MO1
Mayfair Elementary School ES
SO. PHILADELPHIA June 15 to 25
South Philadelphia High School SP
John H. Taggert Middle School TG/TO1
Francis S. Key Elementary School FS
RECREATION CENTERS and DAYCARES
CENTRAL PHILA./CAMDEN July 6 to 25
Christian Street YMCA CH
Marian Andersen Recreation Or. MA/BO*
Camden YMCA (6/26-8/1)
W. PfflLADELPIA
Ronald MacDonald House
St. Mary's Nursery School
Penn Children's Center
ROXBOROUGH/NW PHILA.
Green Lane Nursery School
Lyceum Nursery School
Roxborough YMCA
AIR MONITORING STATION
Camden Lab - Daily 24-h
a. Outdoor samplers.
CY/CO1
July 31 to Aug 13
RM
SM
PC/WP
Aug 17 to Sept 1
GL
LS
RY/RO1
June 1 to Sept 15
CA1
728
-------
In addition, daily 24-h samples were collected a Camden site throughout the entire summer study
period. The Annular Denuder System (ADS) was used to collect aerosol along with gaseous
components, using a preseparator cut at 2.5 pmad (Liang and Waldman, 1992). To estimate air
exchange rates, a perfluorocarbon tracer system was used (Dietz et al., 1986); individual measurements
were made with 24-h sampling periods.
RESULTS
Results for aerosol acidity, sulfate and gaseous ammonia are shown in Figure 2. These data are
the average concentrations for 12-h daytime measurements at the indoor sites during the 2-3 week
sampling periods. Following each set of three indoor sites, the results for the proximal outdoor sites
are shown. The indoor concentrations of sulfate were comparable to the outdoor levels. This pattern
is seen in the daily data. Sulfate aerosol concentrations in the northeast U.S. are spatially homogeneous
over a wider area (Waldman et al., 1991; Thurston et al., 1992; Liang and Waldman, 1993). The
correlations among indoor sites were somewhat less than those among outdoor sites. Site-specific
penetration efficiencies, while typically high, contributed more variability to the daily levels among
indoor sites than the spatial gradients over the metropolitan area.
Ambient aerosol acidity was uncharacteristically low during this summer. Outdoor concentrations
(12-h) averaged below 30 nmole nr3, with peak values not exceeding 100 nmole nr3. For data reported
in nearby sites in New Jersey for a previous summer (1989), the average concentration (12-h) was 50,
with maximum levels that reached 200 nmole m3 (Liang and Waldman, 1992; 1993).
Indoor/outdoor ratios for H+ and SO4° are
given in Table 2. The indoor levels were, as
expected, lower than outdoor concentrations.
However, since there was little aerosol acidity
available to drive indoor levels, it is difficult to
interpret the patterns in neutralization. Indoor
concentrations of ammonia and nitrous acid
(MONO) were substantially higher than outdoor
levels.
The indoor/outdoor (I/O) ratios for aerosol
sulfate and acidity are shown in Figure 3. The
patterns are consistent with previously reported
results, indicating fine aerosol (sulfate) had «80%
penetration efficiency. The I/O ratio for acidic
aerosol were much more variable. At some
locations, the results indicate effective
neutralization of the particles that penetrate (e.g.
TG, FS, CY(P1) and CH). For other sites, the
acidic aerosol appears to have the same
penetration efficiency as sulfate, indicating very
little neutralization (CY(P1), CH(P1), and MA).
It is likely that the relatively low acidity levels
has biased some of these relationship, since
neutralization kinetics may be lower at very low
acidity levels.
Table 2 gives the air exchange rates
(AER) measured for the indoors sites. The rates
were generally lower when AC was used. Some
high AER for AC buildings may have been due
Table 2. Acid fraction and I/O ratios.
Period
PI
P2
P3
P4
P5
Site
HS
MS
ES
SP
TG
FS
CY
MA
CH
CY
MA
CH
RM
SM
PC
Periods: PI:
P3:
P5:
H/SO,
0.00
0.06
0.07
0.13
0.08
0.24
0.13
0.25
0.16
0.02
0.09
0.03
0.17
0.06
0.25
NH3 AER
ppb
24
10
25
14
21
15
33
33
53
42
35
37
na
na
na
1/h
1.3-1.8
0.7-0.8
0.6-0.8
1.2-2.4
1.4-2.9
1.0-6.0
na
na
na
8.5-11
3.3-5.0
0.9-3.2
0.4-1.5
0.6-0.8
2.5-3.5
June 1-14; P2: June
June 29-July 12;
P4:
Vent.
OW
AC
OW
OW
OW
OW
AC
OW
AC
AC
OW
AC
OW
AC
AC
15-28;
July 13-26;
July 27-Aug 13.
729
-------
200
160
100
60
(a) Sulfate
I
H8M8E8MO* 8PTQF8TO- CYMACHCO- CYMACWO- RM8MPCWP*
60
40 -
30
20 -
10 -
(b) Acidity and Ammoi
"II
,
IK \
1 i
ii.n
lia
'CT
li , P
15
1 1
1 , MtlfiaJiav
H8M8E8WO- SPTGFSTO- CYMACHCO- CYMACHBO- RM8MPCWP-
Figure 2. Average concentrations for (a) sulfate and (b) acidic aerosol for intervals PI through P5.
730
-------
to direct ventilation into the sampling area (e.g. CY and PC). The differences in aerosol penetration
for buildings with air conditioning (AC) versus windows-open (WO) ventilation were not consistent,
however.
DISCUSSION
Like most buildings, these schools and recreational centers allowed penetration of the ambient
fine sulfate aerosol. Air conditioning did not seem to cause a consistent reduction in indoor levels of
aerosol nor an elevation in ammonia. However, there were site-specific patterns in acidic aerosol
neutralization.
Indoor ammonia (12-h daytime average) was in the 10-50 ppb range, while outdoor levels were
«2 ppb. The rooms monitored were relatively large (100-300 m3). The source strength of ammonia
can be calculated using a simple box model (Liang and Waldman, 1992):
Ej. = R V C,
where E,, is the emission rate (^mole Ir1); R is the ah- exchange rate (Ir1); V is the room volume (m3);
and C is the indoor concentration (pinole m'3). Note: the conversion is 1 ppb ~ 0.04 /jmole m'3. For
the parameters measured in the sites, the ammonia emissions are calculated to be in the range of 100-300
/jmole h"1. Occupancy in the sampling rooms occurred for at least six of the 12-h interval, and 20 or
more children were present. In their discussion section, Liang and Waldman estimate ammonia emission
strengths as 8 pinole h' from exhaled breath and as much as 150 pmole h' from sweat, per adult.
1
t\ o
O.o
0.6
0.4
0.2
• SULFATE E33 ACIDITY \
|
1
1
|
I
1
n
T]
M
J)
\
N
N
N .
\
N N"
-n...
N
\
N
N
N '
N
N
\
V
\
\
\
\
N
\
\
\
\
\
\
s
° HS MS ES SP TQ FS CY MA CH CY MA CH RM SM PC
Figure 3. Indoor/Outdoor ratios for sulfate (light '"uiaded) and acidic (dark shaded) aerosol.
731
-------
CONCLUSIONS
As in previous studies of aerosol acidity, the summer 1992 study in Philadelphia/Camden schools
and recreational centers yielded indoor levels of H+ and SO4° that were lower than outdoor concentra-
tions, while indoor HONO and NH3 exceeded their levels outdoors. Air exchange rates were measured
in the range 1-6 Ir1. In some sites, neutralization was more than 50%; in others, the acidic aerosol was
largely unneutralized. Notwithstanding the mode of ventilation (AC or open windows), the ammonia
concentrations in occupied rooms were consistently high (30-50 ppb). However, the uncharacteristically
low ambient concentrations for acidic aerosols obscured details about the ammonia neutralization
patterns. The data will be used to assess whether the magnitude of children's exposures to acid aerosols
pose a potential health risk and to investigate neutralization processes caused by indoor ammonia.
ACKNOWLEDGEMENTS
The information in this document has been funded by the United States Environmental Protection
Agency, through Cooperative Agreement (CR 819648) with the Atmospheric Research and Exposure
Assessment Laboratory in Research Triangle Park, NC. It has been subject to the Agency's peer and
administrative review, and it has been approved for publication as an EPA document. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use. We would
like to thank the NJ Department of Environmental Protection and Energy, the School District of
Philadelphia and the individual recreation centers for permission to operate our samples at then- facilities.
We would like to acknowledge Jeffrey Myers and C.J. Pemizzi for their heroic efforts in the field and
Bharati Ramesh, Lih-Ming Yiin and Andrew Setikas for then- many hours in the laboratory.
REFERENCES
Brauer M, Koutrakis P, Keeler GJ and Spengler ID, 1991. Indoor and outdoor concentrations of acidic
aerosols and gases. J. Air Waste Manage. Assoc. 41: 171-181.
Dietz RN, Goodrich RW, Cote EA and Wieser RF, 1986. Detailed description and performance of a
passive perfluorocarbon tracer system for building ventilation and air exchange measurements, hi
Measured Air Leakage of Buildings, Trechsel HR and Lagus PL (eds.), ASTM (Philadelphia) STP
904, pp 203-264.
Huntzicker JJ, Cary RA and Ling CS, 1980. Neutralization of sulfuric acid aerosol by ammonia.
Environ. Sci. Technol. 14: 819-824.
Koutrakis P, Wolfson JM, Spengler JD, Stern B and Franklin CA, 1989. Equilibrium size of
atmospheric aerosol sulfates as a function of the relative humidity. J. geophys. Res. 94D: 6442-
6448.
Li Y and Harrison RM, 1990. Comparison of indoor and outdoor concentrations of acid gases,
ammonia and their associated salts. Environ. Tech. 11: 315-326.
Liang CSK and Waldman JM, 1992. Indoor exposures to acidic aerosol at child and elderly care
facilities. Indoor Air 2: 196-207.
Liang CSK and Waldman JM, 1993. Spanning from regional to microenvironmental scales of exposures
to acid aerosols. This volume.
Thurston GD, Ito K, Kinney P and Lippmann M, 1992. A multi-year study of air pollution and
respiratory hospital admission in three New York State metropolitan areas: Results for 1998 and
1989 summers. J. Exposure Anal, and Environ. Epidemiol. 2: 429-450.
Waldman JM, Lioy PJ, Thurston GD & Lippmann M, 1990. Spatial and temporal patterns in
summertime sulfate aerosol acidity and neutralization within a metropolitan area. Ajinns. Environ.
24B: 115-126.
732
-------
SPANNING FROM REGIONAL TO MICROENVTRONMENTAL SCALES
OF EXPOSURES TO ACID AEROSOLS
Chris S.K. Liang and Jed M. Waldman
Environmental & Occupational Health Sciences Institute (EOHSI)
Exposure Measurements & Assessment Division
681 Frelinghuysen Road, Piscataway, NJ 08855
ABSTRACT
To predict human exposure to acid aerosols spanning from regional to microenvironmental scales
is a challenging task. We have investigated spatial patterns of acid aerosols in two field studies (New
Jersey and Atlanta) with sampling sites from 20 to 100 km apart. In both studies, acidity was found
homogeneously distributed among sampling sites during the daytime. The consistency of this finding
indicates that outdoor exposure to acid aerosols throughout these regions could be estimated with one
central monitoring station during the daytime. Sulfate was found homogeneously distributed among
sampling sites for both studies and both daytime and nighttime. The difference in spatial patterns
between sulfate and acidity indicates sulfate may not be a proper surrogate for acid aerosols. In addition,
we conducted measurements inside homes and institutions in each study. The average indoor/outdoor
ratios of acid ranged from 0.13 (air conditioned) to 0.73 (open window) and were not correlated with
outdoor values in most of the cases. Indoor ammonia levels were approximate 4 to 12 times higher than
outdoor levels. The impact of ammonia on acid aerosols was variable; however, a qualitative inverse
relationship between ammonia and acid/sulfate ratio was evident from site to site.
Correlation and ANOVA analyses are proposed to evaluate spatial patterns of pollutants within
sampling programs. This methodology can be used to standardize the evaluation from different research
studies and assist meaningful comparison and interpretation of spatial results.
INTRODUCTION
Acid aerosols have been recognized as one of the regional pollutants that cause respiratory health
effects. Although many studies have been conducted to measure the ambient levels of acid aerosols, only
limited research has focused on regional spatial variations and relationships among sampling sites.
Acidity and sulfate levels are found highly correlated (r^> 0.90) between sites 10-35 km of a rural
community'>2 but correlation strength decreases for distances of 160 to 400 km apart (r2«0.5)3 The
spatial correlation among sampling sites is influenced to different extents by neutralization with ambient
ammonia^, direct emission of acid aerosols, precursors presented in the atmosphere, and photo-chemical
conditions^. These ongoing processes affect the spatial patterns of acid aerosols, which determine the
exposure levels of populations on a community scale.
We conducted two regional sampling programs concurrent with indoor measurements: in New
Jersey (suburban area during the summer of 1989) and in Atlanta (metropolitan area during the summer
of 1990). Our primary objectives herein are to examine the representativeness of outdoor sampling
stations for ambient acid aerosol levels and to propose a methodology to evaluate spatial patterns of air
pollutants. This can be seen as an exploratory approach; the specific results and methodology might be
used to compare other studies in a consistent format. In addition, the understanding of spatial patterns of
acid aerosols will help to identify the location of potential exposures and assist regulatory agencies in
setting up control and monitoring strategies.
733
-------
METHODS
In the New Jersey study, daily 12-h samples were collected from mid-June to early August 1989.
There were three outdoor sampling stations located in suburban to rural areas. The stations were
deployed in northern (CH), central (RC) and southern (ST) New Jersey (Figure 1, left). The distance
from CH to RC is approximately 60 km, while RC to ST is approximately 110 km. Twice daily samples
were collected every day at RC site, while CH and ST sites' samplers were operated on an every other
day schedule. Three indoor sites were selected within 5 km of the RC site, and daily 12-h daytime
samples were collected. The Atlanta study had a similar spatial and sampling design except all outdoor
locations were in an metropolitan area (Figure 1, right). Twice daily 12-h samples were collected at the
outdoor sites (except daytime only at HO) for 28 days in August. Two indoor sites were selected within
the metropolitan area, and daily 12-h daytime samples were collected. Annular Denuder System (ADS)
samplers were used to collect fine acid aerosols (<2.5 nm, H4", SO4=, NH4+) and relevant gaseous
components (SO2, HNO2, HNO3, NH3)6'7.8.
New Jersey 1989
Atlanta 1990
Metropolitan Atlanta
GT-Georgia Tech HO - Grady Hospital
SC-Smyma HI-lndoorHO
TR- Tucker At - Indoor Apartment
Figure 1. Spatial plans of outdoor and indoor sampling locations.
With this sampling design, "the representativeness of central monitoring station over the study
area" is being tested by the following considerations. First, the pollutant of concern is homogeneously
distributed among sampling sites. Second, the uncertainty associated with concurrent collected samples
is small. We define the representativeness of the central monitoring station as satisfying the following
criteria at the confidence level of a=0.05: (a) measurements at one location are "correlated" with
measurements acquired at adjacent monitoring stations, and (b) the central monitoring station will have
the "same mean" as the adjacent monitoring stations over a period of time. The approach methodology is
illustrated in Figure 2.
Correlation analysis. Acidity and relevant species are analyzed for correlation among sampling
sites. Species with non-significant correlation (
-------
the ANOVA analysis are characterized as spatially non-homogeneously distributed among sampling sites.
Chemical species passing both correlation and ANOVA tests are characterized as spatially homogeneous
among sampling sites, and the measurements at central location are considered to be representative over
the study area. In other words, one central monitoring station can be used to represent or estimate air
quality and human exposure over a larger scale.
Correlation Analysis
NO
J
YES
1
ANOVA Analysis
NO
YES
Spatially non-homogeneous
distribution
Spatially homogeneous
distribution
Figure 2. Methodology for spatial evaluation.
RESULTS
The correlation of chemical species among sampling sites for both studies is listed in Table 1.
Acidity was found correlated among sampling sites for both studies during the daytime. At night, acidity
was correlated only in Atlanta study. Sulfate was found correlated in both studies for both daytime and
nighttime periods. The rest of the chemical species were found not significantly correlated at the level a
=0.05. The ANOVA analyses of mean values at each sampling site are tabulated in Table 2 and 3. When
combining the correlation and ANOVA analyses of means, acidity and sulfate were homogeneously
distributed among sampling site during the daytime periods for both studies. During the nighttime
periods, sulfate was the only chemical specie homogeneously distributed among sampling sites for both
studies.
In general, the spatial patterns observed between New Jersey study and Atlanta study were
similar. Acidity was homogeneously distributed during the daytime while sulfates were homogeneously
distributed for both daytime and nighttime. Chemical species such as HNO2, HNO3, and NH3 were
spatially non-homogeneously distributed among sampling sites. Nevertheless, ANOVA analyses in Table
2 & 3 indicated there were differences in spatial patterns between two studies. For instance, the mean
levels of SO2 were found not significant different among sampling sites for both daytime and nighttime
periods in New Jersey study while significant different among sampling sites in metropolitan Atlanta.
This indicated there were more local factors which altered the SO2 levels in the metropolitan area.
There were no obvious sources of acid aerosols in any of the indoor environments in these
studies. The indoor/outdoor ratios of acid ranged from 0.13 (air conditioned type of settings) to 0.73
(open window type of settings). In most cases, acidity was not significantly correlated with
corresponding outdoor levels for the time series measured. Aerosol sulfate demonstrated a strong
penetration capability for all indoor measurements. The indoor/outdoor ratios ranged from 0.74 (air
conditioned settings) to 1.04 (open window settings), and high correlations were evident between indoor
735
-------
and outdoor concentrations. The presence of indoor sources of ammonia was obvious. Ammonia levels
were at least 4 to 12 times higher than corresponding outdoor levels.
DISCUSSIONS
It should be noted that the results of similar spatial studies may vary due to the following factors.
First, the representativeness of outdoor monitoring station is evaluated through correlation and ANOVA
analysis with a=0.05. The selection of a lower a value would be a more stringent test. Spatial
differences in concentrations would be more likely be found. Second, the number and selection criteria of
sampling sites are important factors in determining the spatial patterns. A higher number of sampling
sites means a greater chance these sampling sites would identify the local differences in concentrations.
Ammonia is known as the most abundant alkaline gas in the atmosphere, and the amount present
is associated with various human activities. The detailed kinetic mechanism of ambient acid aerosols
neutralized by ammonia is not entirely clear, and the relationships found for individual sample are not
readily predicted. However, we notice a consistent qualitative relationship between average ammonia
levels and average acid ratios (H+/SO4=) at different locations (Figure 3). This qualitative relationship,
at least, may assist to locate the potential exposure area for health studies.
CONCLUSIONS
The correlation and ANOVA analyses are proposed to evaluate spatial patterns of pollutants with
concurrent sampling programs. This methodology can be used to standardize the evaluation from
different research studies and assist meaningful comparison and interpretation of spatial results.
Acidity was found homogeneously di ' ibuted among sampling sites for both studies during the
daytime but not for nighttime periods. In the other word, the central outdoor monitoring stations could
be used to represent the concentration/exposure over a larger scales during the daytime. This is
important especially since most of the episodic exposure occurred in the daytime. Sulfate was found
homogeneously distributed among sampling sites for both studies and both daytime and nighttime. The
difference in spatial pattern between sulfate and acidity indicated sulfate may not be a proper surrogate
for acid aerosols. Pollutant species such as HNO2, HNO3, SO2 and NH3 were found not
homogeneously distributed among sampling sites for both studies and both sampling periods.
ACKNOWLEDGMENTS
The study in New Jersey was sponsored by the New Jersey Department of Environmental
Protection and Energy/Division of Science and Research. Funding for the Atlanta study was provided by
the US Environmental Protection Agency/Atmospheric Research & Exposure Assessment Laboratory.
Fellowship support for the author Chris S.K. Liang was generously provided by EOHSI and Cook
College. We would like to thank Dr. Premlata Menon for her support in the field, suggestions in the
laboratory and kindness in all ways. We would like to acknowledge David Morris and Lung-Cheng
Hwang for their field and laboratory efforts.
The information in this document has been funded in part, by the United States Environmental
Protection Agency under assistance agreement with Harvard School of Public Health. It has been subject
to the Agency's peer and administrative review, and it has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
736
-------
Atlanta
REFERENCES
Figure 3. Average ammonia versus acid ratio (H~I7SC>4~).
1 W.R. Pierson, W.W. Brachaczek, R.A. Gorse, S.M. Japar, J.M. Norbeck, and G.J. Keeler,
"Atmospheric acidity measurements on Allegheny Mountain and the origins of ambient acidity in the
northeastern U.S." Atmos Env. (23):431-450, (1989).
2 G.J. Keeler, and J.D. Spengler, "Acid aerosol measurements at a suburban Connecticut site.1' Atmos
Env, (25A):681-690 (1991).
3 G.D. Thurston, J. Gorczynski, P. Jaques, J. Currie, and D. He, "Daily Acid Aerosol Monitoring in
Three New York State Metropolitan Areas: Sampling Techniques and Results." Presented at the
84th Annual Meeting & Exhibition of Air & Waste Management Association, Vancouver, British
Columbia. June, (1991).
4 A.M.N. Kitto, and R.M. Harrison, "Processes affecting concentrations of aerosol strong acidity
(H2SO4) at sites in eastern England" Atmos Env. 26A(13):2389-2399 (1991).
5 U.S. Environmental Protection Agency. An Acid Aerosol Issue Paper: Health Effects and
Aerometrics. EPA-600/8-88-005F. Environmental Criteria and Assessment Office, Research
Triangle Park, NC 27711, pp 4.1-4.59, 5.1-5.50, 6.1-6.39, (1989).
*> M. Possanzini, A. Febo, and A. Liberti, "New design of a high-performance denuder for the
sampling of atmospheric pollutants." Atmos Env. (17):2605-2610, (1983).
7 P. Koutrakis, J.M. Wolfson, J.I. Slater, M. Brauer, and J.D. Spengler, "Evaluation of an annular
denuder/filter pack system to collect acidic aerosols and gases." Env Sci & Tech. 22(12):1463,
(1988).
" M. Brauer, P. Koutrakis, J.M. Wolfson, and J.D. Spengler, "Evaluation of the gas collection of an
annular denuder system under simulated atmospheric conditions." Atmos Env. (23):1981-1986,
(1989).
9 R.P. Cody and J.K. Smith "Applied Statistics and the SAS Programming Language" 2nd ed.,
Elsevier Science Publishing Co., Inc. 1987, pp!33.
Tabte 1. Correlation analysis with a=0.05 among three sampling sites. _____
Component
H+
SO4=
S02
HNO3
HNO2
NH3
New Jersey 1989
AM PM
Correlated X
Correlated Correlated
X X
X X
X X
X X
Atlanta 1990
AM PM
Correlated Correlated
Correlated Correlated
X X
X X
X X
X X
X: Non-significant correlated at
-------
Table 2. Results of spatial variation among outdoor sampling sites. (New Jersey 1989)
Species/Parameters Results
Locations (mean cone, in neq/m-*)
AM
PM
Acidity
SO4=
HNO2
HN03
SO2
NH3
Acidity
S04=
HNO2
HNO3
S02
NH3
NS
NS
Significant
Significant
NS
Significant
NS
NS
NS
NS
NS
Significant
ST(59)
RC(117)
RC(29)
RC(79)
RC(170)
CH(117)
CH(41)
RC(87)
RC(47)
CH(20)
RC(114)
RC(119)
CH(52)
CH(IOO)
CH(26)
ST(56)
ST(158)
RC(105)
ST(38)
CH(74)
CH(38)
ST(19)
CH(107)
CH(116)
RC(44)
ST(91)
ST(17)
CH(47)
CH(132)
ST(63J
RC(23)
ST(60)
ST(35)
RC(17)
ST(74)
ST(70)
Note: Locations connected with solid underline are not significantly different in
concentration from each other (a=0.05). NS: Non Significant different.
Table 3. Results of spatial variation among outdoor sampling sites (Atlanta 1990).
Species/Parameters Results
AM Acidity
SO4=
HNO2
HNO3
SO2
NH3
PM
Acidity
so4-
HNO2
HNO3
SO2
NH3
NS
NS
Significant
Significant
Significant
Significant
Significant
NS
NS
Significant
Significant
Significant
Location (mean cone, in neq/m^)
GT(87)
GT(245)
HO(158)
HO(107)
HO(437)
H0(93)
SC(52)
GT(218)
TR(109)
TR(62)
GT(309)
GT(78)
SC(83)
HO(223)
SC(129)
GT(105)
GT(417)
GT(62)
TR(47)
SC(213)
SC(104)
GT(38)
SC(219)
SC(56)
TR(81)
TR(223)
GT(126)
SC(98)
SC(269)
TR(52)
GT(35)
TR(204)
GT(104)
SC(25)
TR(166)
TR(37)
H0(77)
SC(204)
TR(102)
TR(59)
TR(170)
SC(51)
738
-------
Session 21
Acid Aerosol
Field Study Results
-------
Measurement of Acidic Aerosol Species in Eastern Europe
Michael. Brauer,
The University of British Columbia,
Department of Respiratory Medicine / Occupational Hygiene Programme, 2206 East Mall, 3 66A,
Vancouver, BC V6T 1Z3 CANADA
Thomas S. Dumyahn and John D. Spengler
Harvard School of Public Health, Department of Environmental Health
Boston, MA USA
Kersten Gutschmidt, Joachim Heinrich and Erich Wichmann
GSF - Institut fur Epidemiologie
Neuherberg, GERMANY
ABSTRACT
Ambient measurements of acid aerosols were made for approximately one and a half years in Erfurt,
Germany and Sokolov, Czechoslovakia. In both locations, the burning of high sulfur coal is the primary
source of ambient air pollution. In Erfurt, coal-fired power plants and residential coal burning are the
major contributors, while major sources in Sokolov are several power plants and a coal gasification plant.
24 hour average measurements were made for PMig, as well as fine particle (da < 2.5 urn) H+ and
SO^- for the entire study. Additionally, separate day and night measurements of fine particle H+,
SC>42-, NC>3' and NH4+ and the gases, SO2, HNO3, HONO and NH3 were collected over a 7 month
(late winter-summer) period with additional measurements during pollution episodes the following
winter.
In both communities, 24-hour SC>2 (mean concentrations of 20 ppb, with peak levels of > 150 ppb) and
PMio (mean concentration >50ug m~3) concentrations were quite high. However, aerosol SC^-
concentrations (mean concentration of approximately 100 nmol nr^) were not as great as expected given
the high SC>2 concentrations and acidity was very low (mean concentration of <20 nmol m"-*, with peak
levels of only 150 ninol'-5). Low acidity is likely to be the result of NH3 neutralization and slow
conversion of SC>2 to 804^". Measurements indicate the presence of neutralized ammonium sulfate
species as well as NtLtNC^. We use our substantial database of acid aerosol measurements in North
America as well as historical data from London in comparisons with our observations of very low levels
of aerosol acidity in these Eastern European measurements. Additionally, we compare characteristics of
the chemical composition of these Eastern European winter atmospheres to summer episodes in the
eastern U.S and Canada.
INTRODUCTION
With the end of Communist rule in Eastern Europe and the new access to informatioi. it has become
clear that many countries suffered severe environmental neglect. As these countries struggle with the
transition to free market economies there is an urgent need to examine and prioritize environmental health
concerns so that control measures can be taken. As part of this effort an major air pollution epidemiology
study was undertaken in the former East Germany and in the Czech Republic. As part of this study,
ambient air measurements of PMjQ, acidic aerosols and gas species and NH3 were conducted for
approximately 1.5 years at two sites. This monitoring effort provided a unique opportunity to examine
levels of acidic aerosol species in environments that were directly impacted by both local and regional
sources of high sulfur "brown coal." Monitoring was specifically focused on winter inversions during
which pollutant concentrations were expected to be highest. This situation was thought to resemble
historical episode situations encountered 30-40 year ago in London, England, the Meuse Valley in
741
-------
Belgium and Donora PA in which increased daily mortality has been associated with air pollution
exposure'"-'. Although only limited measurements were made it has been argued that the causative agent
in these episodes was aerosol acidity. Here we report measurements of acidic aerosols and gases from an
urban location in the former East Germany and from a smaller industrial town in the Czech Republic.
Comparisons are made between these results and those collected recently in the U.S. and Canada.
METHODS
Site Characterization
Two cities were selected for air monitoring, Erfurt (ERF), in the former East Germany and Sokolov
(SOK) in the Czech Republic. Both Erfurt and Sokolov were reported to be subject to winter inversions
which resulted in poor ambient air quality (high TSP and SC>2 concentrations) and reduced visibility.
Erfurt (population approximately 200,000) is located northwest of Frankfurt, approximately 100 km east
of the former east-west border and is a regional center of commerce. The city has a large older central
area where the primary heating source is individual coal furnaces. The outer areas of the city contain
large apartment complexes with steam heat supplied by a large coal burning power plant located several
km east of the city center. Erfurt is situated on a flat plain bordered by a ridge on the south end of the
city. The sampling site was located less than 5 km from the town center, 15 m from the nearest structure
and 30 m from the nearest major road. Sampler inlets were located approximately 2m from ground level.
Sokolov (population approximately 60,000) is an industrial town located in the coal-mining region of the
Czech Republic, 100 km east of the German-Czech border. There are several power plants and large
industrial complexes in the region, in addition to a coal gasification plant. Sokolov is located in a valley
surrounded by low hills. The sampling site was located on the terrace of a two-story building
approximately 2 km from the central district. Sampler inlets were located 2 m above the terrace surface.
Sampling and Analysis
Paniculate acidity (PM2 5) and PMjQ samples were collected with the Harvard Impactor (with the
addition of a citric-acid coated honeycomb denuder for acidity sampling)^. The Harvard-EPA Annular
Denuder System (HEADS) was used for measurements of gaseous species during some portions of the
study. Sampling and analysis procedures are reported in detail elsewhere^ Daily or every second day
24-hour samples of fine paniculate (da < 2.5 u,m) H+ and SO42" were collected from December 1990 -
June 1992 and 24-hour PMjg samples were collected February 1991 - June 1992. Annular denuder
measurements were made twice daily February 1991 - September 1991 and during episode periods from
October 1991 - April 1992. Detection limits for 12-hour denuder measurements were 6.2 ppb, 1.8 ppb,
16 nmol m-3, 24 nmol nr3, 32 nmol nr3 and 5.6 nmol trr3. for SC>2, NH3, H+, SC>42-, NH4+ and
NO3", respectively. The detection limit for PMjo was 9 ug rrr3. while the Harvard Impactor 24-hour
detection limits were 8 and 15 nmol m"3 for H+and SO^-, respectively.
RESULTS AND DISCUSSION
Summary statistics are presented in Table 1. Aerosol acidity was low in both sites. Figure 1 compares
the annual means measured in Erfurt and Sokolov to those measured in 22 North American communities
as part of a major epidemiological study. The eastern European sites report aerosol acidity
concentrations which are at the low end of the observations in North America. While concentrations of
PMio and SC>2 in Erfurt and Sokolov were significantly higher in the winter, aerosol acidity was slightly
higher in the summer at both sites. This contrasts with observations from North America in which
summer acidity levels are much greater than those measured in the winter. Low aerosol acidity may
result from low SC>42~ production or from neutralization of acidic sulfate species by ambient NHv
Although acid aerosol levels were very low, comparisons to the North American database of SO/)2"
measurements (Figure 2) indicates that the eastern European sites presented much higher concentrations.
A plausible cause for the observations of low aerosol acidity and high SO/)2- concentrations is the
elevated levels of ambient NH3 (Table 1) at Sokolov and Erfurt. Although the record of NH3
742
-------
concentrations is less complete, mean concentrations during the sampling periods were consistently high
(> 2 ppb) throughout the year. Annual averages measured in North American communities are typically
below 1.5 ppb, with only the rural California sites reporting annual means above 2 ppb.
PMjo and SO2 concentrations at both eastern European sites were quite high and were well above
concentrations observed in the North American communities (Figure 3-4). Concentrations were higher
during the winter than in the summer which is consistent with the occurrence of winter inversions and
increased coal burning (particularly in Erfurt) for residential heating.
The meteorological conditions of local winter inversions suggests that SC"2 and consequently
("potential acidity") concentrations and NH3 will peak at the same times, limiting the opportunity for
acidic particles to avoid neutralization. In North American summer episodes it is believed that convective
mixing replaces stagnant surface-level air (high ammonia content) with acid-laden air that has been
transported above inversion layers where it is protected from neutralization^. The situation in eastern
Europe also differs in terms of the proximity of the sources - with local sources emitting SC>2 below
inversion layer heights and therefore in close proximity to ammonia sources. Since the acidic and basic
air masses are not separated there is ample opportunity for neutralization to occur.
Further, the SC>2 conversion reactions are expected to differ between eastern Europe and North America
where photochemical reactions predominate. In North America, emissions are typically above the height
of inversion layers under conditions of low particle concentrations, facilitating transport of gaseous
species. Transport at high elevations provides adequate time for conversion while being protected from
neutralization by surface level sources of NH3 The predominant conversion mechanism in North
American is photochemical, based on the reaction of hydroxyl radical with SO2 in the presence of water.
In eastern Europe, where winter inversions reduce the impact of photochemistry, heterogeneous reactions
are expected to predominate. For example, SC>2 may be absorbed on particle surfaces and catalytically
(by transition metals for example) oxidized to sulfuric acid. An additional pathway involves the diffusion
of SC>2 into a liquid droplet where internal oxidation (by H202 for example) produces sulfuric acid. This
diffusion is controlled by the amount of NIT} in the droplet such that any sulfuric acid produced will be
immediately neutralized in the basic environment. As the droplet becomes more acidic further diffusion
of SC>2 into the droplet is limited. This slow conversion mechanism is likely to be important for moist
environments with high particle concentrations such as those in Sokolov and Erfurt. This aqueous phase
oxidation mechanism may explain our observations of lower than expected sulfate levels at a given SC>2
concentration as well as the high levels of NH4+ ion observed (mean concentrations >230 nmol m~3). In
the winter inversion setting concentrations of oxidizing species may be quite low and conversion of SO2
may be quite limited. Conversion rates are highest (10-40% per hour) for metal-catalyzed reactions,
although there is considerable variability depending upon the pH of the droplet as well as the presence of
inhibitors. Homogeneous gas phase reactions occur at rates of 0.3-2% per hour, while aqueous phase
oxidation is much slower (0.2% per hour).
The extent of SC>2 conversion may be estimated by the [SO42" / SO42' + SC>2] ratio. Much like the
North American situation, the ratios in Erfurt and Sokolov were lower in the winter than in the summer.
Ratios were typically below 0.25 in the winter, and reached peaks of 0.6 or higher in the summer. While
these ratios suggest considerable conversion in the summer, SC»2 concentrations were low during this
period. In the winter, when SC>2 concentrations are elevated, [SC>42~ / SC>42~ + SC>2] ratios were
below those seen in summer acidic atmospheres in North American (mean ratios of 0.4-0.6) (Keeler,
Keeler), suggesting the impact of slower SC>42~ production processes.
Measurements from other coal-burning areas are presented in Table 1 for comparison. Waldman and
colleagues collected a limited number of measurements in Wuhan, China a region which may be
743
-------
expected to be similar to the eastern European sites in that soft coal combustion produces elevated
concentrations of particulate matter and SC"2 The measurements of Waldman and colleagues were
collected in the winter under similar conditions (overcast, cold, high relative humidity) to those observed
in eastern Europe during the winter. While we measured higher peak levels of aerosol acidity than
reported by Waldman, et al. this is likely due to the longer period of sampling. In both situations the
levels of aerosol acidity were surprisingly low, given the high measured concentrations of SC>2 and
PMin. Historical measurements from London were much higher than any present measurements. That
SO^- levels were somewhat higher in Wuhan, than in Erfurt or Sokolov is interesting given the much
higher SC>2 levels measured in Erfurt and Sokolov. SC>2 levels in Erfurt and Sokolov, although well
below those measured in London were significantly higher than those measured in China. This indicates
that conversion of SC>2 to sulfate was relatively low in the eastern European sites. In contrast the Wuhan
environment may have presented more complete conversion, but acidity was controlled by neutralization.
PM]o concentrations were similar to concentrations measured in Wuhan. Although these comparisons
are limited, they suggest that the conditions encountered in Erfurt and Sokolov may be typical of regions
in which high sulfur coal is burned during the winter. In such situations, although PMirj and SC>2 levels
may be much greater than concentrations observed in North America, aerosol acidity levels are very low.
REFERENCES
1. Ito, K. and Thurston, G.D. Characterization and reconstruction of historical London, England acidic
aerosol concentrations. Environmental Health Perspectives 1989, 79:35-42
2. Lioy, P.J. and Waldman, J.M. Acidic sulfate aerosols: Characterization and exposure. Environmental
Health Perspectives 1989, 79:15-34
3. Dockery, D.W. and Speizer, F.E. Epidemiological evidence for aggravation and promotion of COPD
by acid air pollution. Chapter 10, pp 201-225
4. Koutrakis, P., Wolfson, J.M. and Spengler, J.D. "An improved method for measuring, aerosol strong
acidity: Results from a nine-month study in St. Louis, Missouri and Kingston, Tennessee," Atmos
Environ 1988, 22: 157-162.
5. Marple, V., Rubow, K.L., Turner, W. and Spengler, J.D. Low flow rate sharp cut impactors for
indoor air sampling: Design and calibration. J Air Pollut Control Assoc 1987, 37: 1303-1307
6. Keeler, G.J., Spengler, J.D. and Castillo, R.A. Acid aerosol measurements at a suburban Connecticut
site. Atmos Environ 1991, 25A: 681-690
7. Koutrakis, P., Wolfson, J.M., Slater, J.L., Brauer, M., Spengler, J.D., Stevens, R.K. and Stone, C.L.
Evaluation of an annular denuder/filter pack system to collect acidic aerosols and gases. Environ Sci
Technol 1988,22: 1463-1468
8. Spengler, J.D., Brauer, M. and Koutrakis, P. Acid Air and Health. Environ Sci Technol 1990, 24:
946-956
9. Keeler, G.J., Spengler, J.D., Koutrakis, P., Allen, G.A., Raizenne, M. and Stern B. Transported acid
aerosols measured in southern Ontario. Atmos Environ 1990, 24A: 2935-2950
10. Waldman, J.M., Lioy, P.J., Zelenka, M., Jing, L., Lin, Y.N., He, Q.C., Qian, Z.M., Chapman, R. and
Wilson, W.E. Wintertime measurements of aerosol acidity and trace elements in Wuhan, acidity in
central China. Atmos Environ 1991, 25B: 113-120
744
-------
Table 1. Summary statistics of ambient concentrations measured in Sokolov and Erfurt and
comparisons to measuremei i in China10 and historical measurements in London1.
Aerosol Acidity (nmol m"3)
Sokolov
Erfurt
Wuhan, China
London
SO42' (nmol m'3)
Sokolov
Erfurt
Wuhan, China
NH3 (ppb)
Sokolov
Erfurt
S02 (ppb)
Sokolov
Erfurt
Wuhan, China
London
PMjo (ug m"3)
Sokolov
Erfurt
Wuhan, China
Mean
10
8
14
133
100
115
525
3.8
2.4
20
23
16
90
59
66
Peak
166
156
50
2735
379
772
980
34
20
227
274
28
462
247
269
350
50
40
20 -
10 -
\A
v\
7
ii i i i i i i i i i
yor abd pet mty erf liv aok epg eim lem pmbnow dun bib pen zan chv hen elc oak ath mor par unt
CITY CODE
Figure 1. Mean concentrations of aerosol strong acidity (nmol m'3) measured in Sokolov (SOK),
Erfurt (ERF) and in 22 communities in the U.S. and Canada.
745
-------
40 •
30 •
iwhm ehv dun p*r h*n
Figure 2. Mean concentrations of aerosol SO^" (nmol m'3) measured in Sokolov (SOK), Erfurt
(ERF) and in 22 communities in the U.S. and Canada.
pel abd ilc dun chynowlom o*k Ub mly ith (v par tan mot ipfl unl h«n lim pan tok irl
Figure 3. Mean concentrations of PMio
in 22 communities in the U.S. and Canada.
3) measured in Sokolov (SOK), Erfurt (ERF) and
lam dun bib ilc p»i ian pan ilh un( tok i
Figure 4. Mean concentrations of S(>2 (ppb) measured in Sokolov (SOK), Erfurt (ERF) and in 22
communities in the U.S. and Canada.
746
-------
ACID AEROSOL MEASUREMENTS IN EASTERN CANADA
J.R. Bropk and H. A. Wiebe
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario M3H5T4
ABSTRACT
Measurements of fine aerosol strong acidity were collected during 1988-1991 at various Canadian
locations as part of the Harvard School of Public Health and Health and Welfare Canada 24-Community
Study1. Measurement activities have continued since that time as part of the Canadian Acid Aerosol
Measurement Program (CAAMP). Currently 7 sites in eastern Canada are in operation. Filterpack,
denuder and dichotomous sampler or impactor measurements are being collected daily during May-
September and every third day during the rest of the year. The details of the monitoring activities and
preliminary results from 1992 are discussed in this paper.
INTRODUCTION
Over the past 10 to 15 years a significant amount of research has focused on the acid deposition
issue. Knowledge of the sources of acidifying pollutants and the atmospheric processes affecting their
fate has improved substantially as has information on the temporal and spatial variability in their
deposition. In general, more emphasis has been placed on monitoring deposition rates due to concerns
over ecological effects. Consequently, less attention has been paid to the impact of acidic pollutants
while still in the atmosphere. In this form they have the potential to adversely affect human health and
to reduce visibility. These issues are currently receiving more attention in North America.
The Canadian Acid Aerosol Measurement Program (CAAMP), which formally started in 1992,
aims to assist in the study of the health effects of acid rain related pollutants by meeting each of the
following objectives:
1. Collect data, of known quality, on the acidity levels and fraction of aerosol mass in the fine and
coarse mode at multiple Canadian locations.
2. Determine annual and seasonal averages, peak levels (12-24 hr) and temporal variability of aerosol
acidity and fine (PM2.5) and coarse (PM10) masses.
3. Estimate the acid aerosol exposure of the Canadian population.
• to assess the significance of the results of ongoing health effects research in a Canadian context
• insure that the data are appropriate for incorporation in future health studies.
In this paper we describe the details of the sampling activities and present some of the results from the
1992 measurements.
METHODS
At two Canadian Air and Precipitation Monitoring (CAPMoN) sites (Egbert, Ont and Sutton, Que)
and one Ontario Ministry of the Environment site (Windsor) acid aerosol measurement equipment have
been operating since June 1991 or before. During May, June and July of 1992 measurement activities at
four new locations were initiated. All current measurement sites are shown in Figure 1. The operation
of sites in urban areas reflects the interest in estimating the human exposure to acidic aerosols.
Concentrations measured at rural sites are expected to be a better indication of the concentrations within
larger, regional-scale air masses (i.e. the direct effects of urbanization are assumed to be small) thereby
providing information on the spatial distribution of the aerosols. The current rural sites are in close
proximity to the urban sites. This configuration will provide information on the amount of variability in
747
-------
45V
. .Montral
-T5W
Figure 1 The location of the Canadian Acid Aerosol Measurement Program (CAAMP) monitoring
sites during 1992-93.
concentrations over relatively small distances and it may be possible to assess the effect of urban areas
on aerosol acidity.
The main features examined by CAAMP are:
• The strong acidity of the fine fraction aerosols.
• The fraction of mass in the fine and coarse modes.
• The concentrations of the major inorganic aerosol constituents (ions and metals).
• Ambient HNOj concentrations.
In Table 1 the various measurement parameters that are being collected at each site are listed. Four
different sampling apparatus are being used to take these measurements. Both annular denuder systems
(single and double denuder) and the fine and coarse mass particle sampling units (Air Diagnostics and
Engineering 10 1 min'1 10 and 2.5 urn cut size inertial impactors) rely on a MKS 1159B-20.000SV mass
flow controller, calibrated for 1 atm and 0° C, to regulate the flow rate. The flow rate during the 24 hour
sampling duration is 10 1 mur1. The dichotomous samplers rely on their own pump and flow controller
systems which maintain a total flow rate of approximately 16.7 1 mhr1.
The supplies necessary to collect one week of samples and one blank (i.e. filterpacks, filters,
impactor plates and denuders) are shipped to each site and stored at room temperature (-22° C). After
all the supplies in a shipment have been used they are shipped back to the laboratory. Filterpacks are
snipped and stored on site in sealed plastic bags and are only removed during the sampling period.
Similarly denuders are kept capped except during sampling. The filterpacks are loaded with a Gelman
Zeflour, supported PTFE, 2|im pore size Teflon filter which is situated immediately downstream of the
denuders, followed by a nylon filter (Gelman Nylasorb) and a citric acid coated Whatman 41 filter. The
Teflon filters collect the fine particles and the citric acid coated filters capture NH3 resulting from the
748
-------
Table 1 Summary of the particulate and gaseous species measured at the Canadian Acid Aerosol
Measurement Program sites. Type of equipment used and the frequency of the the 24 hour
measurements are also indicated.
SITE
Windsor,
Kejimkujik
Egbert,
Sutton
Toronto,
Montreal,
Saint John
Particulate
H+, sol', NH^, NO;
PM2.5, PM10
trace elements
H+, Sol', NHi, NO;
PM2.5, PM10
trace elements
H+, SO*", NH^, NO;
PM2.5, PM10
trace elements
Gas
HNO3, SO2,
NH3
HNO3, SO2,
NH3
HNO3, SOj
Frequency
daUy/1 in 3 (Oct.-Apr.)
daily/ 1 in 3 (Oct.-Apr.)
Iin6
daily /I in 3 (Oct.-Apr.)
daily/lin3(Oct.-Apr.)
Iin6
daily /I in 3 (Oct.-Apr.)
daily/lin3(Oct.-Apr.)
Iin6
Equipment
3-stage filterpack / 2 denuders
dichotomous sampler
dichotomous sampler
3-stage filterpack / 2 denuders
Air Diagnostics Impactors
Air Diagnostics Impactors
3-stage filterpack / 1 denuder
dichotomous sampler
dichotomous sampler
revolitization of NBfyNC^ particles. In the double denuder system the nylon filter captures HNC>3 also
resulting from the revolitization of NrLtNOj particles. Thus, an H+ correction due to neutralization from
NHj liberated from the NH^NOj particles is possible with the double denuder system. With the single
denuder system the nylon filter provides a measure of both ambient HNC>3, HN(>j from NFLjNHOj
particles and also ambient SC>2.
Samples were collected for 24 hour periods starting at 0800 EST. From May 1 or whenever
sampling commenced to Sept. 21,1992 measurements were taken every day. From Sept. 21,1992 to the
present samples have been collected every third day. All samples are analyzed for the species shown in
Table 1 with the only exception being that elemental analysis using XRF is only carried out on every
sixth sample, however, the remaining filters are being archived.
RESULTS
All sites were operational by early July of 1992. Technical problems resulted in significant down-
time at St. John and thus few daily samples were collected. At the other sites a majority of the days
were successfully monitored. Overall, the summer of 1992 was uncharacteristically ''clean". There were
few periods when the meteorological conditions were favorable for the build-up of air pollutants.
However, there was a measurable amount of variability in the concentrations over time and space and a
number of episodes were observed.
The day to day variation in SO4" and H+ concentrations at Windsor is shown in Figure 2. Typical
for the summer of 1992, the levels of both species tended to be low. There is a clear visual correlation
between SO4 and H+ with H+ concentrations (in nmole nr3) systematically lower than for SO4". Good
correlation was observed at all sites, but there was a noticeable increase in the ratio of H+ to SO4~ at the
sites on the east coast (Maritimes). In Figure 3 the mean distribution (in percent) of the main ions is
shown. In Ontario, H"1" accounted for a relatively small amount of the cations and NH4 a much large
fraction. In contrast, in the Maritimes there was an increase in the amount of H+ relative to the other
ions. This pattern is consistent with the hypothesis that there is less local NH3 in the Maritime region
and potentially less NH/j during transport if winds are across the ocean.
749
-------
u
05-15 05-30 06-14 06-29 07-14 07-29 08-13 08-28 09-12
Figure 2 The day to day variation in 24 hour sulfate and acidity concentrations measured at Windsor,
Ontario, during the spring and summer of 1992.
100%
Figure 3 The site to site variation in the mean distribution of the major ions found on the fine aerosols
collected in eastern Canada.
The changing nature of the aerosol ionic balance can be seen by examining the changes in S04" and
H+ concentration as the episode of Aug. 23-26 is tracked eastward. During this period a large, slow
moving high pressure system moved across the northeast North America. As the southerly winds
associated with this system's "back side", advected the pollutants northward, varying amounts of S04
and H+ were observed at the sites depending upon origin of the pollutants and the transport path. In
Table 2 the concentrations of SO4", H+ and their ratio as a function of site and day are listed. The nature
of the aerosol changed dramatically as the episode formed further east. Concentrations increased in
southwestern Ontario on the 23rd and over Nova Scotia on the 25th of August. While SO4" levels tended
to be less in the east there was no such pattern in H+. With the exception of Windsor on the 23^, the
highest acidity was observed at Kejimkujik. This was due to less neutralization. The molar ratios of H*
750
-------
-.2-
to SO4" were systematically higher at Kejimkujik attaining a value of about 1. This pattern is typical of
the measurements over the summer of 1992 and is reflected in Figure 3. In the east the molar ratios were
close to 1, while in Ontario and Quebec they were significantly less than 1.
Table 2 Fine aerosol sulfate concentrations, acidity concentrations and their molar ratios as a
function of location and date during the episode observed on August 23-26, 1992.
AUGUST EPISODE [H+] (nmole m'3)
SITE
Windsor
Egbert
Toronto
Montreal
Sutton
Keji
22
1
6
13
8
13
1
23
283
52
19
19
7
9
24
23
46
39
106
63
42
25
29
32
25
55
104
194
26
5
19
8
59
47
183
27
5
0
0
0
0
1
AUGUST EPISODE [SO^~] (nmole m'3)
SITE
Windsor
Egbert
Toronto
Montreal
Sutton
Keji
22
14
91
91
86
46
8
23
353
312
195
89
12
9
24
157
257
312
262
152
60
25
137
224
245
219
193
204
26
37
61
59
217
201
178
27
30
7
12
25
6
6
AUGUST EPISODE [H+]:[SO^~]
SITE
Windsor
Egbert
Toronto
Montreal
Sutton
Keji
22
0.05
0.07
0.15
0.10
0.29
0.22
23
0.80
0.17
0.10
0.21
0.58
1.00
24
0.15
0.18
0.13
0.40
0.41
0.70
25
0.21
0.14
0.10
0.25
0.54
0.95
26
0.13
0.31
0.16
0.27
0.23
1.03
27
0.17
0.00
0.00
0.00
0.00
0.13
SUMMARY
Fine aerosol samples were collected at 7 sites in eastern Canada during the summer of 1992 and
sampling is continuing in 1993. There appears to be a clear tendency for less aerosol neutralization over
the far eastern regions. Thus, even though SO4" concentrations are less, the H+ concentrations can be
the same or even greater than those observed in southern Ontario and Quebec.
Measurement activities are expected to continue through the summer of 1993 with the addition of
some sampling in the Vancouver area. Given the relatively large fraction of acid in the aerosols
collected over the Maritimes, additional measurements in this area are being considered. In particular, it
is necessary to collect information on the aerosol acidity in areas closer to the large population centre of
Halifax, Nova Scotia. Over southern Ontario there is a large amount of NH3 and the aerosols tend to be
relatively neutral. In this region it will be necessary to consider focusing on daytime aerosol acidity
measurements. It is possible that much of the neutralization occurs at night when the boundary layer
stabilizes. During the day, when people are more likely to be exposed, the acidities could be higher.
751
-------
REFERENCES
1. Spengler J.D., Koutrakis P. and G. Allen, "Summary of air pollution exposure data for the study of
health effects of ambient aerosols on children in 24 cities in North America." Harvard School of Pqfrlfe
Health Technical Report (199?).
752
-------
DEVELOMENT OF A LOW CUTPOINT VIRTUAL IMPACTOR FOR
COLLECTION OF SEMI-VOLATILE ORGANIC COMPOUNDS.
Constantinos Sioutas, Petros Koutrakis, Steve Ferguson
Harvard School of Public Health
Boston, MA 02115
Robert M. Burton
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711.
ABSTRACT
Traditionally, separation of gas and particulate-phase organics required the use of a diffusion
denuder to collect gases placed upstream of the filter pack. We propose a new sampling scheme that
includes the following components:
(i) a virtual impactor (size cut-off of 2.5 pm) to remove coarse particles from the air
(ii) a slit-nozzle virtual impactor (size cut-off of 0.15 u.m) to separate the fine particles
from the air sample;
(iii) a particle trap that collects all particles between 0.15 - 2.5 ^m; and
(iv) a gas trap that collects all gas phase organics (plus a small amount of particulate mass
dp<0.15 u.m).
A virtual impactor with a 50% cut-off size in the range of 0.1 u.m has been developed. The
impactor consists of an acceleration nozzle with 0.035 cm in diameter and a collection probe 0.05 cm in
diameter. In addition, it operates with a low pressure drop (about 40 inches of water) and with very low
particle losses, averaging to about 7% with a maximum of 14 % at the 50 % cut-off size. The separation
characteristics of the virtual impactor have been evaluated.
INTRODUCTION
Inertial impactors have been widely used for particle collection, mainly because of their sharp cut-off
characteristics. Due to the extensive theoretical work (Marple and Liu, 1974; Marple and Willeke, 1976)
their performance has become well understood and their characteristics can be predicted. The most
important limitations of these instruments are the following (Biswas et al, 1988): i) particles may bounce
from the collection surface upon impaction. ii) collected particles may reentrain iii) wall losses between the
impactor stages may be considerable (Loo el. al, 1976) iv) very large particles may break-up upon
impaction, especially at high impaction velocities. The virtual impactor provides an alternative solution to
the particle bounce and eliminates reentrainment problems.In order to remove the larger particles from
the collection probe, a fraction of the total flow is allowed to pass through the probe, referred to as the
minor flow (typically 10% of the total flow). As a result, the concentration of the larger particles in the
minor flow has increased by a factor of Qi0i/Qmjn, where Qtol is the total flow entering the virtual impactor
and Qmin is the minor flow. In addition to the elimination of the particle bounce problem, the virtual
impactors have the advantage of keeping the collected particles airborne and, as stated previously, by
adjusting the ratio of the nu'nor to the total flow, the concentration of the coarse particles can be increased
to a desired value (Barr, el. al, 1983). This paper presents the development of a small cut-off size virtual
impactor that operates at a small pressure drop. In addition, a detailed measurement of the particle losses
through the system is presented. A belter understanding of the particle losses can enable us to oplimize
Ihe design of small cutpoint virtual impactors.
METHODS
Description of the virtual impactor.
To the best of our knowledge the construction of a 0.1 |im cut-off point virtual impactor has not
been accomplished before. The virtual impactor consisted of three parts made of 6061 TG Al. The first
753
-------
part is the upper plate that includes the acceleration nozzle, where the aerosol enters and gains sufficient
momentum for impaction. Two different upper plates were constructed with nozzle diameters equal to
0.025 and 0.035 cm. The second part consisted of the collection tube whose diameter was chosen equal to
0.05 cm so that the ratio of the receiving to the acceleration nozzle was equal to 2 and 1.5, respectively.
The distance between the two nozzles was kept constant and equal to 0.020 cm. Finally, the third part was
designed as a snap-fit to the second part and consisted of the outlet tube, where the major flow was finally
driven through. The three parts were held together by screws. Leaks were avoided by placing O-rings in
the contact areas of all three parts.
Experimental.
The test system for the characterization of the virtual impactor is depicted in Figure 1.
Suspensions of 2.5% by weight yellow-green latex microspheres (Fluoresbrite, Polysciences, Warrington,
PA) were nebulized by a pocket nebulizer (Retec X-70/N) using room air at 25 psi as described by Zeltner
et. al (1991). The nebulizer was connected to a syringe pump in order to atomize large amounts (60 ml) of
the fluorescent suspension. In addition, the output of the nebulizer was maintained constant to ensure a
stable atomization process. Seven different particle sizes were used: 0.05, 0.12, 0.20, 0.45, 0.75, 1.1 and 2.0
(im in diameter. The particle size range was selected to cover the collection efficiency curve of the virtual
impactor, which was designed to have a 50% cut-off size of approximately 0.15 [im. The aerosol was
subsequently dried in a 6-liter dry-air dilution chamber by mixing with filtered dry air at a flow rate of 10
LPM and passed through a 1-liter chamber where ten Polonium 210 ionizing units were placed
(Staticmaster, NRD Inc.) to neutralize any particle charge. After the neutralizer part of the aerosol was
passed through the test system (Figure 1) which consisted of the virtual impactor and two 47 mm glass
fiber filters connected to the major and minor flows to collect the test particles. Each filter was connected
to a pump with a Matheson mass flow meters in line to control the flows. The aerosol passed through the
test system at a total flow rate varying from 0.30 to 0.65 LPM, depending on the acceleration nozzle used.
The ratio of minor to total flow varied from 0.1 to 0.2. Another part of the test aerosol was driven
through an optical particle size analyzer (model LAS-X Particle Measuring System, Inc., Boulder, CO)
which was used to record the particle size distribution throughout the experiment at a sampling flow rale
of 1.5 LPM. Finally, the pressure drop across the virtual impactor was continuously monitored in every
experiment with a Magnehelic pressure gage (range 0-100 inches of water). In all the experiments the
pressure gage showed a pressure drop varying from 35 to 50 inches of water across the virtual impactor,
depending on the configuration that was tested.
After a sufficient amount of the aqueous fluorescent suspension was nebulized the two glass fiber
filters were extracted using 5 ml of ethyl acetate for each of them, as recommended by the PSL particles
manufacturer (Polysciences Inc.). The quantities of the fluorescent dye in the extraction solutions were
measured by a fluorometer (FD-300 Fluorescence Detector, GTI, Concord, MA) to determine particle
concentration. The efficiency of the virtual impactor was determined by dividing the amount collected on
the minor flow filter to the sum of the amounts collected on both major and minor flow filters.
Furthermore, the acceleration nozzle, the collection probe and the inside surfaces of the virtual impactor
were carefully washed with 10 ml of ethyl acetate each to determine particle losses through the system.
RESULTS AND DISCUSSION
To initiate the parametric study of the virtual impactor characteristics, a main configuration was
chosen. Subsequently, the minor flow ratio, the pressure drop across the virtual impactor and the
collection probe to acceleration nozzle ratio were varied, one al a time. Subsequently, the minor flow
ratio, the pressure drop across the virtual impactor and the collection probe to acceleration nozzle ratio
were varied, one at a time. The main configuration consisted of a jet nozzle diameter equal to 0.035 cm, a
collection probe equal to 0.05 cm (therefore the nozzle ratio was 1.5), a total flow through the nozzle
equal to 0.56 LPM and a minor flow ratio equal to 0.2. The corresponding pressure drop across the
impactor was 35 inches of water. Table 1 shows the results of the virtual impactor characterization. The
50% cutpoint is between 0.1-0.15 urn. As it has been expected, the cutpoint is smaller than the theoretical
prediction for an inertial impactor (0.25 |im). This can be explained by the fact that the minor flow
contains 20% of all the particles smaller than the cutpoint. Particle losses reach a maximum around 0.20
|im. The losses in the collection probe increase with the particle size since large particles cross the
streamlines and enter the collection probe. Particles smaller than 0.25 |im in diameter are primarily lost
on the backside of the acceleration nozzle, as they follow the deflected streamlines of the major flow.
754
-------
Particles larger than 1.1 |im in diameter are lost inside the acceleration nozzle due to the turbulence that
is caused as the flow converges and is rapidly accelerated. This conclusion could be further sustained by
examining the deposition pattern of the fluorescent particles in the nozzle using a magnifying lens.
Effect of collection probe to acceleration nozzle ratio.
In order to test the effect of the collection probe to acceleration nozzle ratio, the smaller of the
two acceleration nozzles (0.025 cm diameter) was used to give a ratio of 2. The total flow through the
nozzle was 0.29 LPM in order to maintain the same velocity in the acceleration nozzle. The minor flow
ratio was 0.2, as in the main configuration. The results of the virtual impactor evaluation is shown in
Figures 2. It can be seen that the collection efficiency of the virtual impactor with the increased
collection-to-acceleration nozzle ratio drops significantly at smaller particles (d <0.25 (im). The most
striking result of the increase in the collection probe to acceleration nozzle ratio appears to be the
dramatic increase of the particle losses (Figure 3). This result is in agreement with the conclusions of Loo
et. al, (1988).
Effect of minor flow ratio.
The effect of reducing the ratio of the minor flow to the total flow from 0.2 to 0.1 on the virtual
impactor characteristics was investigated. The ratio of the collection probe diameter to the acceleration
nozzle diameter was 1.5 and the total pressure drop was 35 inches of water at a total flow rate of 0.56
LPM. The results of this test are shown in Table 2. While there seems to be little effect on the collection
efficiency, the decrease in the ratio of minor to total flow causes a small increase in the overall particle
losses, particularly for particle sizes near the cutpoint. This would be due to the fact that a higher minor
flow results in a higher local velocity around the tip of the collection nozzle, and consequently a stronger
vacuum is applied to the particles as they exit the acceleration nozzle and approach the proximity of the
collection probe.
Effect of increasing the pressure drop across the impactor's nozzle.
The effect of increasing the jet velocity and therefore the pressure drop across the virtual impactor
was investigated. The ratio of the minor to total flow was kept 0.2, the collection probe to acceleration
nozzle ratio was equal to 1.5 (the jet diameter was 0.35 mm) and the total sampling flow rate was 0.7
LPM, resulting to a total pressure drop across the virtual impactor equal to 50 inches of water (or,
equivalently, 0.12 atmospheres). Figure 4 depicts the efficiency curves for the two pressure drops tested.
The collection efficiency increases and the cutpoinl decreases as the pressure drop across the virtual
impactor increases. This is a direct result of the increase in the jet velocity and the subsequent decrease in
the impactor's cutpoint. The particle losses decrease as the jet velocity increases (Figure 5). In the case of
the increased jet velocity, the particle losses become maximum for particles equal to 0.12 |im in diameter,
as opposed to the main configuration, where the loss maximum occurred for particles of 0.20 |im in
diameter.
CONCLUSIONS.
A virtual impactor with a 50% cut-off size in the range of 0.1 |im has been developed. The
impactor consists of an acceleration nozzle with 0.035 cm in diameter and a collection probe 0.05 cm in
diameter. In addition, it ot; ,iates with a low pressure drop (about 40 inches of water) and with very low
particle losses, averaging to ,.bout 7% with a maximum of 14 % at the 50 % cut-off size. The results can
be summarized as follows:
(1) The 50% cut-off size decreases as the collection probe to acceleration nozzle ratio increases from 1.5
to 2. In addition, the particle losses through the impactor increase.
(2) The collection efficiency of the virtual impactor is not significantly affected by a decrease of the minor
flow ratio from 0.2 to 0.1. This decrease in the flow ratio, however, is followed by an increase in the
particle losses, particularly for sizes near the impactor's 50% cutpoint.
(3) Increasing the jet velocity and therefore the pressure drop across the impactor's nozzle results in a
decrease in the 50% cut-off size, as well .j in a decrease in the particle losses across the virtual impactor.
ACKNOWLEDGEMENTS.
The development and evaluation of the low cutpoint impactor was supported by the U.S. Environmental
Protection Agency through the cooperative agrremenl to the Harvard School of Public Health #CR
755
-------
816740. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
LITERATURE CITED.
(1) Barr, E.B, Hoover, M.D., Kanapilly, G.M., Yeh, H.C. and Rothenberg, S.J. "Aerosol concentrator:
design, calibration and use." Aer. Sci. Techn. 1983, 2:437-442.
(2) Biswas, P. and Flagan. R.C. "The •-. Article trap impactor". J.Aer. Sci. 1988: 19:113-121.
(3) Loo, B.W. ans Cork, C.P. "Developrient of high efficiency virtual impactors." Aer. Sci.& Techn. 1988,
9:167-176.
(4) Loo, B.W., Jaklevic, J.M., and Goulding, F.S. in "Fine Particles: Aerosl generation, measurement,
sampling, and analysis." (B.Y.H Liu ed.) 1976.
(5) Marple, V.A. and Liu, B.Y.H. "Characteristics of laminar jet impactors." Environm. Sci.& Techn.
1974, 7:648-654.
(6) Marple, V.A. and Willeke, K. In "Fine Particles: Aerosl generation, measurement, sampling, and
analysis." (B.Y.H Liu ed.) 1976.
(7) Zeltner, T.B., Sweeney, T.D., Skornik, W.A., Feldman, H.A. and Brain, J.D. "Retention and clearance
of 0.9 (jm particles inhaled by hamsters during rest or exercise." /. Appl. Physiol. 1991, 65:1137-1145.
Compressed air
Po-2IO
neutral
Drying
Chamber
T
Compressed
air (20 osi)
I i Nebulizer
D d
,
r .
{
-
h n
PMS
HH^^
1
\—
mp Filter
I
Virtual
Impactor
^
Magnehelic
Gage
Pump
FIGURE 1. Schematic diagram of the test system.
756
-------
Table 1. Particle collection characteristics of the virtual impactor.
Particle
Diameter
(jim)
0.05
0.12
0.20
0.45
0.75
1.10
2.00
(%) in Major
Flow
(0.44 LPM)
45.1
41.4
33.5
4.8
1.9
1.8
1.6
(%) in Minor
Klow
(0.12 LPM)
38.2
50.5
52.6
89.3
95.5
93.6
91.4
(%)
Collection
Efficiency
45.9
55.0
61.1
94.9
98.0
97.1
98.3
(%) Total
Losses
10.4
8.1
13.8
5.9
2.6
4.6
7.0
(%) Losses
on inlet
of nozzle
2.2
1.0
3.8
13
0.2
1.1
3.4
(%) Losses on
collection
probe
3.4
3.8
52
3.1
1.9
2.1
2.8
(%) Other
interstage
losses
1.8
3.3
4.5
1.5
0.5
1.4
0.8
a. The values in the table represent averages of repeated runs.
b. The minor flow ratio was 0.2.
c. The collection probe to acceleration nozzle ratio was 1.4 (nozzle diameter 0.035 cm)
Table 2. Particle collection characteristics of the virtual impactor with a minor flow ratio 0.1.
Particle
Diameter
((im)
0.05
0.12
0.20
0.45
0.75
1.10
2.00
(%) Collection
efficiency
39.9
50.2
58.3
94.5
99.7
99.8
100.0
(%) in Major
Flow
(05 LPM)
55.0
38.8
34.0
5.1
0.3
02
0.0
(%) in Minor
Flow
(0.06 LPM)
36.5
39.1
47.5
87.8
94.5
98.1
98.9
(%) Total
losses
8.4
22.1
18.8
9.0
5.2
1.2
1.1
(%) Losses on
inlet of nozzle
0.9
55
6.4
3.1
3.2
1.2
0.7
(%) Losses on
collection
probe
2.6
10.2
10.9
4.2
1.4
0.0
0.4
(%) Other
interstage losses
4.9
6.4
IS
1.7
0.6
0.0
0.0
a. The values in the table represent averages ot repeatedruns.
b. Collection probe lo acceleration nozzle ratio 1.4.
c. Acceleration nozzle diameter 0.035 cm.
-------
o.a -
o.s-
0.4-
0.2-
0.0 -
« • • •
D
a
[ — .11' • ' ' ' " i ' — r~^1 — *•"•
0.6-
o.s-
a ratio=1.4 5 0.4-
• ralio=2
ji 0.3 -
i! 0.2-
0.'. -
n n
. • .
1 = *
3 " ^^ *
1
- ratio=1.4
• ralio=2
Particle diam«ter ( 'J-
FIGURE 2. Collection efliciency for dilferenl
collection probe to acceleration nozzle ratios.
.01 .' ; '0
Parricle diameter ( a m)
FIGURE 3. Particle losses for different
collection probe to acceleration nozzle ratios.
0.08 atm
0.12 atm
Particle diameter ( u. m)
FIGURE 4. Collection efficiency for different pressure
drops across the virtual impactor.
= 0.08 atm
• 0.12 aim
Particle Diameter (ii.m)
FIGURE 5. Particle losses for different pressure
drops across the virtual impactor.
-------
Session 22
Personal Monitors
-------
Evaluation of a Portable Multisorbent Tube Sampler
for Monitoring Airborne Organic Compounds
Albert J. Pollack and Sydney M. Gordon
Battelle Memorial Institute
505 King Avenue
Columbus, Ohio 43201, USA
ABSTRACT
The Perkin Elmer STS 25 Sequential Tube Sampler is a novel portable sampling device for characterizing
volatile organic compounds (VOCs) in air. The STS 25 is a self-contained unit that is designed to collect air
samples consecutively onto a batch of up to 24 sorbent tubes, one at a time, at a specified flow rate for a
preselected collection time per tube. The device was evaluated under stringent laboratory conditions using a
standard mixture containing 42 nonpolar VOCs and packing the tubes with a three-stage carbon-based
multisorbent bed. Samples were analyzed noncryogenically using a Perkin Elmer ATD 400 Auto Thermal
Desorption system and gas chromatograph.
Results of these analyses indicated that the target compounds were collected with efficiencies generally
comparable with those obtained with the standard (TO-14) procedure for canister sampling followed by cryo-
genic preconcentration and GC analysis. An outdoor sample collected with the device also yielded results that
compared favorably with those obtained using the TO-14 methodology. The portability and ease of operation
of the STS 25 lend itself to use in various microenvironments to provide both temporal and spatial informa-
tion on VOCs in complex indoor and outdoor atmospheres.
INTRODUCTION
Recognition of the importance of trace-level VOCs in breathing-zone air and their potential impact on
human health has stimulated interest in methods for characterizing these compounds. Most of the techniques
in use today for measuring VOC concentrations have their origin in industrial hygiene monitoring methods.
They generally rely on solid sorbents1-2 or containers3 to collect whole-air samples.
A recent development in efforts to measure temporal variability of VOCs is the Perkin Elmer Sequential
Tube Sampler (STS 25), which is designed for use with their Automated Thermal Desorption system (ATD
400) and gas chromatograph (GC). The present study was undertaken to evaluate the performance of the
Perkin Elmer STS 25. Before the evaluations were begun, a method for analyzing TO-14 compounds using a
three-stage carbon-based multisorbent bed for sample collection, along with the ATD 400's electrically cooled
two-stage carbon-based trap for refocusing the desorbed sample, was developed and validated.
EXPERIMENTAL
Sequential Tube Sampler (STS 25)
The STS 25 air sampler is a prototype unit supplied by Perkin Elmer (Figure 1). The device consists of
an air-tight plastic box that contains a sample tube carousel, an electronic timer, and a personal monitoring
pump. A fan mounted in the rear of the unit draws air into the box at a rate of ~ 30 L/min, which is
sufficient to change the air above the sample tubes once every second. The air enters through an opening in
the lid which is positioned above the collection tubes. The air stream is sampled by the one collection tube
that is located in the single position that allows the air monitoring pump to draw the air through a tube.
When not in the sampling position, the tubes are effectively sealed from the air stream through the use of
diffusion-limiting caps.
The sample volume is regulated by time and flow rate. The timing device in the STS 25 will maintain a
tube in the sampling position from 0.1 sec to 10 hr. During this study, each tube was sampled for 75 min.
761
-------
At the end of the sampling period, the carousel rotates and the next tube is positioned to receive an air
sample. An SKC Model 224-PCXR7 personal monitoring pump was used to draw air through the tube and
was operated at a nominal flow rate of -43 cc/min. This flow setting resulted in a 40 cc/min flow through
the packed sampling tube. All flow rates were confirmed using a Buck Model M-5 calibrator.
Power for the STS 25 can be obtained either from a rechargeable 12 v battery or from a DC power
supply/battery charger. The DC power unit was used throughout this study. The STS 25 is also capable of
maintaining a trickle charge to the battery of the personal monitoring pump, which ensures that the perfor-
mance of the pump remains constant over long sampling periods.
During the evaluation of the STS 25, carbon-based sorbents were used to collect the target VOCs. The
three-staged tubes (based upon Supelco, Inc.'s Carbotrap 300 configuration) contained Carbotrap C (300 rag),
Carbotrap (200 mg), and Carbosieve S-III (125 mg). These were maximum mass loadings for the sampler
tubes yet still allowed all of the sorbents to be positioned in the heated zone of the tube during desorption.
Sorbent Tube Analysis
Sorbent tubes used during the evaluation of the STS 25 were processed using the ATD 400/GC system.
Typical operating conditions for the thermal desorption of a tube included desorption of the sample at 325 C
for 10 min with a purge flow of 60 cc/min helium carrier. The refocusing trap was packed with 35 mg of
Carbotrap and 23 mg of Carbosieve S-III. While receiving the organic components from the heated sampler
tube, the trap was held at 4 C, then was thermally desorbed at 325 C for 3 min. A trap desorption flow rate
of ~ 11 cc/min was maintained to accomplish a rapid transfer of the refocused sample material to the GC.
The gas chromatograph was operated with a 50 m x 0.32 mm x 5.0 ^m dimethyl polysiloxane fused silica
capillary column in conjunction with a flame ionization detector. Analytical conditions for the resolution of
the TO-14-based target compounds included an initial oven temperature of 40 C for 15 minutes, then the oven
was heated at 8 C/min to a final temperature of 200 C and held at this temperature for 12 min. A standard
TO-14 cryogenic analytical system was also used during this study as a reference system to compare results
obtained using the carbon-based sorbent traps.
STS 25 Evaluation Experiments
The STS 25 was subjected to three types of tests. Firstly, it was mechanically tested to determine if it
would reliably cycle as specified to collect air samples using sorbent tubes. Secondly, the STS 25 was placed
in a 75-L glass/stainless steel environmental test chamber and programmed to collect blank (zero air) air
samples and samples of TO-14 standard atmospheres at 2 and 10 ppbv concentrations. The unit was
challenged at two temperature levels (25 C and 37 C) and two relative humidity (RH) levels (10 percent and
60 percent). In all cases, 3-L samples were collected. The tubes were analyzed and the results were
compared to those from an initial chamber validation test to determine whether the STS 25 unit was able to
collect laboratory-controlled atmospheres without either contributing to, or removing, any of the TO-14 target
compounds. Finally, a field test was performed to evaluate the STS 25 unit under "real world" conditions.
The STS 25 was set up to collect a 3-L outdoor ambient air sample. Simultaneously, a second sorbent tube
and a 6-L air sampling canister were used to sample the air at the same location. A Tylan-controlled
pumping assembly drew air through this second tube at the same 40 cc/min flow rate as the STS 25 was
using. The pre-evacuated canister was filled using a restrictive orifice that allowed air to enter the canister at
-60 cc/min. This flow decreased to ~40 cc/min as atmospheric pressure was approached during the 75-min
collection period. At the end of the sampling period, the canister was still at subatmospheric pressure (-200
mm Hg). Data from these collection systems were compared to evaluate both the operation of the STS 25
system and the performance of the sorbents used.
762
-------
RESULTS AND DISCUSSION
Mechanical
The STS 25 was operated over two 24-hr periods to evaluate the operational characteristics of the unit.
The device was programmed to collect 75-min and 10-min samples. In both cases, the system performed
with no mechanical problems. The charging system on the STS 25 also performed well. The SKC personal
monitoring pump maintained a constant flow rate (~43 cc/min) over a 24-hr period. No mechanical
problems were encountered with the operational aspects of the STS 25.
Environmental Chamber Tests
Blank Test. The results for the background samples collected by the STS 25 are shown in Table 1A.
With the STS 25 present in the chamber several of the TO-14 target compounds were observed. The 25 C
runs indicate that the concentrations of these compounds change slightly with RH. The 37 C run shows
higher concentrations, which are probably due to compounds being liberated from the STS 25 itself. It is
important to note that the conditions under which all the chamber tests were run reflect a very stringent test
of the artifact characteristics of the STS 25. Since the device is in a relatively small volume that is being
purged at 2 L/min, while the STS 25 is cycling air through itself at 30 L/min, it is evident that any volatile
compounds that may be associated with the materials used in the construction of this sampler are being
concentrated in the chamber atmosphere. Nonetheless, the blank test does indicate that there is the potential
for artifact contribution from the STS 25 during ambient air sampling.
Canister samples were also collected from the chamber while the STS 25 was drawing the sample onto
the collection tube. Differences were identified between the tube concentrations and the canister analysis.
We believe that these discrepancies are indications of the limitations associated with quantifying complex
mixtures with only an FID detector. The differences observed were generally in a direction that indicated
compounds at higher levels with the STS 25 tubes than with the canister samples. This is not surprising
because compounds other than the TO-14 species could be eluting at the same time as one of the target
compounds and are not distinguished on the basis of retention times only.
TO-14 Chamber Tests. The STS 25 results (Table IB and 1C) tend to exaggerate the concentrations
present for several of the earliest eluting species. We believe that this is a further example of artifact
compounds coeluting with compounds of interest. This exaggeration is particularly evident in the 2 ppbv test
where a slight contribution from an artifact results in a pronounced variance in recovery. Relative humidity
does not seem to play as much of a role in liberating light artifacts as does the increased temperature. The
reported concentration values from the STS 25 were confirmed to be due to artifact coelution since the
canister results generally did not reflect increased TO-14 species presence with changes in relative humidity
and temperature.
The canister results for early eluting TO-14 species were generally at expected levels. A noticeable
deviation was that methyl chloride (not shown in the table) was consistently reported with low recoveries.
This could be indicative of compound affinity for the STS 25 sampler.
The intermediate eluting compounds in the TO-14 mixture exhibited predictable recoveries. At 10 per-
cent RH, STS 25 recovery data were generally lower than at the 60 percent RH level at 25 C. This is in
agreement with the results observed earlier where the presence of humidity seems to enhance the desorption
of these TO-14 species from the multisorbent collection bed. Also, agreement with the canister results was
generally more reproducible. It appears that this region of the chromatogram was affected less by artifact
contributions from the STS 25. Therefore, FID quantitation was more reliable and this was reflected in the
reported concentrations for both the 25 C and 37 C tests.
The final set of compounds in the TO-14 mixture were the latest eluting species. From the STS 25 blank
test run it was observed that there was a group of artifact peaks that eluted at the same time as these heavy
TO-14 species. Because of this, artifact coelution, with contribution to the reported STS 25 values, was
763
-------
expected and observed. For the 2 ppbv test, higher concentrations were reported when compared to the
canister results. This was particularly true for the 37 C test. However, canister levels were also elevated in
this region, indicating that the STS 25 at 37 C was liberating TO-14 species along with nontarget compounds.
One particular compound, 1,2,4-trimethylbenzene, was identified by both the tube and canister results as
being a major TO-14 artifact associated with the sampler.
The general statement that can be made from evaluating the STS 25 under stringent laboratory conditions
is that the sampler will collect TO-14 compounds onto sorbent tubes. The Perkin Elmer sampler apparently
does contribute artifacts to the sampling piocess when operated in a confined environment that does not afford
an air changeover rate that is as great as that of the STS 25 itself. Also, the analytical system employed was
limited by the use of only the FID detector, which was affected by coeluting artifacts. Nonetheless, the STS
25 operated acceptably within the confines of the testing environment, sorbents, and analytical system used.
Outdoor Sampling. The outdoor air sampling results, for selected compounds, from the analysis of the
two sorbent tubes using the ATD 400/GC and the canister using the GC/MSD system are presented in Table
ID.
Agreement between the two tube samples was quite good with the same compounds being identified and
comparable concentrations reported. From this information it was evident that when used in a less confined
environment, the STS 25 collects a representative air sample.
The canister samples however did not quantitatively reproduce the tube data. This was particularly true
for the early eluting peaks. Variability is again judged to be associated with the qualitative/quantitative
differences of the FID and MSD detectors and is not necessarily indicative of deficiencies with the STS 25 as
a sampling unit or the sorbents used. In general, analytical agreement for this sampling was quite good and
may be attributed to the less complex nature of the sample allowing better FID identification and
quantification.
A copy of the FID traces from the two collection tubes is provided in Figure 2. It should be noted that
the relative humidity during the collection of these samples was at ~90 percent. It can be seen on both of
the FID traces that there was a negative deflection in the baseline prior to the elution of the VOCs. Although
the flame did remain lit, the conditions may have been close to the tolerable moisture limit for the FID.
CONCLUSIONS
The STS 25 is a portable, easy to use, sampling device which provides an important option for moni-
toring ambient air toxic compounds with carbon-based sorbent sampling tubes. The evaluation of the unit
indicates that it is mechanically sound and able to collect TO-14 VOCs with comparable efficiency to canister
sampling. Limitations of the system are that if it is operated in very confined environments, it is possible that
the sampler may contribute to or alter actual VOC levels. However, when used as intended, sample integrity
is not compromised by the unit. To be able to correctly analyze the sorbent tubes collected by the STS 25, it
is recommended thai GC analysis be combined with mass selective detection, compound specific detectors, or
dual FIDs with two-dimensional chromatography.
ACKNOWLEDGEMENTS
The authors wish to express thanks to Perkin Elmer, both in the United Kingdom and U.S.A., for
assistance during this study. The information presented here has been funded wholly or in part by the U.S.
Environmental Protection Agency under Contract Number 68-DO-0007 to Battelle Memorial Institute.
Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
764
-------
REFERENCES
1 Brown, R. H. and Pumell, C. J., "Collection and analysis of trace organic vapour pollutants in ambient
atmospheres", J. Chromatoga. 178. 79-90, 1979.
2. Krost, K. J., Pellizgari, E. D., Walburn, S. G., and Hubbard, S. A., "Collection and analysis of
hazardous organic emissions", Anal. Chem. 54. 810-817, 1982.
3. McClenny, W. A., Pleil, J. D., Holdren, M. W., and Smith, R. N., "Automated cryogenic
preconcentration and gas chromatographic determination of volatile organic compounds in air", Anal.
Chem. 56. 2947-2951. 1984.
OQ Hi Banwy Pack
(L«8 0002)
S75 Power L0ad
0
0 CMItactor Shield
0 Lead to uw Auxiliary Heater
0 30 AH AutomoUliI Battery
0 Wlusion Limiting Cap*
0 Tube Pumped « IN* Position
0 Lead to Ine Pump Cnerger
(ijj) Pump
(Q Pump Suction Pipe
Figure 1. STS 25 schematic.
LiUL-
,u*jU^«Ju-~--Ji-
Figure 2. FID traces of outdoor air sampling: STS 25
upper, Tylan tube lower.
765
-------
Table 1. Results for selected compounds during the STS 25 evaluation.
Sampler
and
Atmosphere
A. Blank
STS 25
Canister
STS 25
Canister
STS 25
Canister
B. 2 ppbv. TO-14
STS 25
Canister
STS 25
Canister
STS 25
Canister
C. 10 ppbv. TO-14
STS 25
Canister
STS 25
Canister
D. Outdoor Sampling
STS 25
Tylan tube
Canister
Temp
(•C)
25
25
25
25
37
37
25
25
25
25
37
37
25
25
37
37
21
21
21
Relative
Humidity
(%)
10
10
60
60
60
60
10
10
60
60
60
60
60
60
60
60
90
90
90
1,3-
Butadiene
0.1
n.d."
0.4
n.d.
0.9
n.d.
2.3
1.3
3.3
1.8
4.7
2.0
11
13
9
12
3.2
2.7
n.d.
Compound
Benzene
0.4
0.1
1.3
0.1
2.8
0.2
1.9
2.1
3.6
2.0
3.4
2.5
11
12
10
12
1.4
1.8
1.5
Concentration
Toluene
0.5
0.2
0.6
0.1
2.1
0.7
1.6
1.6
2.3
1.9
2.5
1.9
11
13
10
14
3.0
3.5
3.4
(ppbv)
m/p-xylene
0.7
0.4
1.0
0.5
1.9
1.5
1.8
1.6
2.1
1.5
3.1
3.1
9
10
8
15
2.2
2.2
2.4
* n.d. = not detected.
766
-------
Session 23
General
-------
Sensitive Real-Time Monitoring of NO, NO2 and other Nitrocompounds
Under Atmospheric Conditions
Josef B. Simeonsson*, George W. Lemire* and Rosario C. Sausa
US Army Research Laboratory, AMSRL-WT-PC
Aberdeen Proving Ground, Maryland 21005-5066
*NRC Postdoctoral Research Associate
ABSTRACT: A new laser-based technique has been developed recently for the detection of
NO, NO2 and other nitrocompounds in the vapor phase. Atmospheric sampling is
accomplished using a standard pulsed valve/molecular beam apparatus. Supersonic
expansion of the sample cools the gas and facilitates spectroscopic detection. A fraction of
the expanded gas is transmitted in the form of a molecular beam into the analysis chamber
which is differentially pumped. There the molecular beam is probed by a laser beam tuned
to the AV <-X2n (0,0) transition of NO at 226 nm. Molecules of NO in this region are
photoionized by a (1 + 1) resonance-enhanced multiphoton ionization (REMPI) process and
accelerated into a time-of-flight mass spectrometer (TOFMS) for mass selective detection. In
addition to NO species, NO2 and other nitrocompounds are detected efficiently by employing
a single laser operating at 226 nm or 193 nm to both photofragment the target molecule and
detect the NO fragments, produced as a result of the rapid predissociation of NO2. In this
manner, total NOX can be measured simultaneously using a single excitation wavelength for
NO and NO2 detection. Results of studies of the laser photofragmentation/ionization
technique using laser radiation at 226 nm and also at 193 nm are presented and discussed.
INTRODUCTION: Atmospheric NO and NO2 are important participants in several
atmospheric reaction cycles, including the role of NO as a precursor in the production of
atmospheric HNO3 (acid rain) and its role in the photochemical production of O3. Despite the
importance of these species in the atmosphere, they are generally only present at trace
levels making quantitative measurements difficult. Concentrations may range from the parts-
per-trillion (ppt) to parts-per-million (ppm) in parts of the lower atmosphere. A common
method of detection for the NO molecules is the chemiluminescent reaction of NO with an
excess of O3 producing O2, and NO2* which is detected optically. Another method of
detection is laser induced fluorescence which has been developed by Bradshaw, Davis,
Rodgers and co-workers and demonstrated to be an effective method of detection both for
surface and airborne applications.1'2
The detection of NO2 is often accomplished indirectly. Independent measurements of
NO and NOX are performed, and the NO2 is determined from the difference of the two
measurements. NO2 can also be detected directly by long path absorption methods.3 An
indirect method for measuring NO2 is to detect the laser induced fluorescence of NO
molecules which are produced from the UV photolysis of NO2. This method of
photofragmentation-fluorescence using a two-laser pump/probe technique has been used to
detect a number of species (see references 4-15) and has been successfully applied by
Davis, Rodgers and coworkers to detect atmospheric NO2, MONO and other species."'15
The analytical sensitivities that can be achieved by this method approach the ppt level and
769
-------
are superior to other non-laser based methods. Unlike long path absorption techniques, the
photofragmentation/fluorescence method provides a point measurement allowing accurate
spatial profiling of the atmospheric environment.
We have developed recently a technique
for the detection of NO, NO2 and other
nitrocompounds which employs a single laser
operating at 226 nm. The laser radiation is
used for both photofragmentation of
nitrocompounds, which results in the
production of NO fragments, and resonance-
enhanced multiphoton ionization (REMPI) with
time-of-flight (TOP) mass spectrometric
detection of NO molecules via the A2s+ <-X2n
(0,0) band.16 The technique has been
demonstrated using a pulsed valve/molecular
beam time-of-flight apparatus. Supersonic
cooling results in greatly simplified spectra of
both analyte and potential spectral interfered
species thus increasing the selectivity of the
approach. We have extended this technique to
include the use of 193 nm radiation for
TIME-OF-FLIGHT
MASS SPECTROMETER
LENS
LASER (193 or 226 nm)
Figure 1: Schematic of the apparatus
In this paper, we discuss the results of
photofragmentation and ionization of the NO species.
studies performed at both 226 nm and 193 nm.
EXPERIMENTAL: The experimental apparatus has been described previously.16 A
schematic diagram of the salient features of the apparatus is shown in Figure 1. Briefly,
sample introduction was performed at atmospheric pressure where the sample was present
as a minor species in a buffer gas (usually Ar or air). Calibrated mixtures of NO/Ar (0.1%)
and NO2/air (6.2 ppm) were obtained from Matheson and Scott-Marrin respectively. Samples
of nitromethane (from Aldrich), dimethylnitramine (from ARDEC) and nitrobenzene (from
Eastman-Kodak) were introduced in flows of the buffer gas where the concentrations were
calculated using the reported vapor pressures at room temperature. The gas mixtures were
expanded into the analysis chamber using a pulsed valve (R.M. Jordan Co.) operated at 10
Hz which had a nozzle diameter of 0.5 mm. Following expansion, the sample species were
probed by laser radiation at 226 nm or 193 nm which was focused by a lens (f=250 mm or
f=500 mm lenses). The laser radiation induced fragmentation of the parent species to
produce NO fragments which were subsequently ionized within the same laser pulse by
REMPI processes. An excimer pumped dye laser system with frequency doubling (Lumonics
Ltd., Hyper EX-400, Hyper DYE-300 and Hyper TRAK-1000) operated at 10 Hz provided
tunable UV radiation at 226 nm with maximum pulse energies of approximately 200 /J and a
spectral linewidth of 0.16 cm"1. An ArF excimer laser (Lambda Physik Inc. EMG-150)
provided broadband UV radiation (approximately 100 cm"1) at 193 nm with pulse energies in
the chamber of 1-5 mJ. Ion signals from the TOF mass spectrometer were displayed and
monitored on a 125 MHz oscilloscope (LeCroy 9400). Signals were also directed to a gated
integrator (Stanford Research Systems) whose output was acquired by a PC-AT for data
analysis and storage.
The TOF mass spectrometer distinguishes ions of different masses by their different
770
-------
times of arrival at the microchannel plate detector which is located at the end of the flight
tube. Since all the ions produced in the ionization region experience the same potential, they
have the same kinetic energy and are separated in time according to their masses. Although
the overall time of arrival of a given ion depends on the sum of the transit times through
various regions of the TOP mass spectrometer, it is proportional to the square root of the
mass of the ion. The time of arrival t, of mass m, is related to the time of arrival t2
corresponding to mass m2 by the following:
mf
(1)
Thus the determination of any mass, m2 can be performed using an accurate measurement
of t, of a known mass m,. In the current studies, NO+ was used to calibrate the time-of-flight
mass spectral response.
The pulse pressure of the gas in the sample chamber was calculated using the
following equation (adapted from reference 15):
(2)
In the above equation P is the backing pressure, r is radius of the nozzle orifice, R is the
radius of the skimmer orifice and e is the skimmer transmission angle, which is equal to the
arctangent of R over D, the distance of the nozzle to the skimmer. For a backing pressure of
770 Torr, a nozzle diameter of 0.5 mm, a skimmer diameter 3.0 mm and a skimmer to nozzle
distance of 20 mm, the pulse pressure in the chamber is approximately 180 mTorr.
RESULTS AND DISCUSSION: The physical
processes underlying our method for detecting
nitrocompounds by using a single laser may be
understood by referring to Figure 2 which
shows partial energy level diagrams for the NO2
and NO molecules. By restricting ourselves to
using 226 nm radiation, it is found that
nitrocompounds, generalized as R-NO2, which
are irradiated at this wavelength will
photofragment with high efficiency to produce
the N02 fragment and the radical R. The NO2
molecule can then be detected using the same
laser by monitoring its predissociative product,
NO, by (1 + 1) REMPI using the AV-4-X2!!
(0,0) band at 226 nm. Alternatively, NO
fragment detection can be accomplished by LIF
at the same wavelength. However it was
determined in this study that the LIF approach
is not as effective for the NO/NO2 system.
REMPI
120/KX)
100.000
5 80,000
o
(FJ «M°°
cc
LLI
m
20000
0
( 1 + 1 ) ^ NO+ + -'
. W/APMUmi 9.26 «V
(22enm) :
+ - \
2 N
fejjjjjjjjjjj^l 9.76 8V J
hv
(228 nm)
B'B k°<'D> \
/,
r^2r+
i! LIF
^_
PHEDISSOC1ATON
NOXZn +0('D)
hv
(220 nm)
MQjPn
Presented in Figure 3 is a typical mass
Figure 2: Potential energy level diagram
of NO and NO,
771
-------
spectrum obtained when photolyzing a 130 ppm mixture of dimethylnitramine (DMNA) in Ar
using 226 nm radiation. The DMNA molecule was chosen for study because it is a simple
analogue for larger nitramine molecules which are of military interest. The spectrum is
characteristic of all the compounds studied at this laser wavelength in that it shows a
prominent ion signal whose time-of-arrival coriboponds to m/e=30 amu which is that of the
NO+ ion. Similar results were obtained using Ar, air or nitrogen as the buffer gas. No ion
signal was recorded, however, whon the laser was tuned away from resonance. The striking
simplicity of the ion mass spectrum and lack of any other significant signals indicates that the
ionization is the result of a fragmentation/REMPI process. It is important to note that ion
signals corresponding to the mass of NO2 (m/e=46 amu) were not observed in the mass
spectra for any of the nitrocompounds studied. This supports the proposed mechanism
which relies on rapid predissociation of the NO2 to produce NO. To verify the mass spectral
assignment and also to maximize the signal intensity, an ion excitation scan was performed.
This excitation spectrum reveals numerous rotational lines which are attributed unequivocally
to electronic transitions of the NO A2s+<-X2n (0,0) band.
In addition to using 226 nm radiation, we have investigated the use of 193 nm
radiation for photofragmenting and ionizing nitrocompounds. Ionization of NO molecules is
possible by way of the A2s+<-X2n (3,0) and B2n<-X2n (7,0) bands and the D2s+<-X2n (0,1)
vibrationally excited band. The fragmentation
of nitrocompounds at 193 nm to produce NO
occurs in a way that is similar to the
fragmentation processes at 226 nm. While the
precise mechanisms are not known, recent
studies by Houston and coworkers9 of the 193
nm photodissociation of nitromethane indicate
that the NO2 fragment is produced in a number
of excited states which are further fragmented
to produce NO either in its electronic X2!!
ground state or directly in its excited t^s*
state. It is also expected that the NO
fragments have a distribution of energies. The
energy distribution makes it possible for several
of the rovibrational levels in the X2!! state to be
resonant with the relatively broad spectral
output of the ArF excimer used as the
fragmentation/ionization source. The relatively
higher sensitivity that is obtained for NO
pnotofragments from nitrocompounds (as
opposed to pure NO) is a combined result of
the production of electronically excited NO molecules and a distribution of rotationally and
vibrationally excited NO molecules in the X2n state. By comparison, pure NO molecules that
undergo supersonic expansion are restricted to a few rovibrational levels, some of which may
not be in resonance with the ArF laser output. Thus it is likely that a relatively smaller fraction
of the pure NO population is accessible by the laser for photoionization.
Presented in Table 1 are limits of detection (LODs) for the compounds studied
employing the fragmentation/REMPI technique using 226 nm and 193 nm radiation. The
LODs range from ppm to ppb and refer to the gas phase concentration of the analyte prior to
£
55 -40-
IS
z
1 -80-
C3
CO
Z
o
-120
N0+
0 20 40 60 80 100
TIME OF FLIGHT (usec)
Figure 3: Time-of-flight mass
spectrum of DMNA
772
-------
Compound
NO
NO2
CH3NO2
DMNA
Nitrobenzene
LOD at 226 nm
(ppb)
8
240
1000
450
2400
LOD at 193 nm
(PPb)
1200
500
180
510
490
introduction into the analysis Table 1: Limits of Detection at 226 nm and 193nm
chamber. The experimental ^^—"•—•^—^^^^"•••l
conditions for all of these
determinations were identical, thus
any differences in the LODs are
indicative of differences in the
absorption cross sections and/or
fragmentation efficiencies of the
precursor molecule, as well as the
absorption cross section NO
fragment at the respective
wavelengths. For a backing
pressure of approximately 1 atm
(atmospheric sampling
conditions), the calculated gas
volume throughput of the
supersonic nozzle is 6.8 Torr/cm3 ^^mmm^^^^^^^mm^,^^^^^^
in a pulse of about 100 ^s
duration. In this work, a LOD of 8 ppb for NO (for laser excitation at 226 nm) corresponds to
a density of 5x107/cm3 in the probe volume of the laser. Using a laser induced fluorescence
technique, Sandholm et al.14 have reported a LOD of 9.4x107/cm3. For NO2, a LOD of 240
ppb corresponds to a density of 1.4x109/cm3 in the probe region. Sandholm et al.14 have
reported a LOD for NO2 of 4x10e/cm3 using a related two-laser photofragmentation/two-
photon laser induced fluorescence technique. However, this technique requires an
integration period of up to 6 min. in order to achieve its maximum sensitivity while our
present technique is virtually real-time in its response. Plane and Nien3 have used a
differential absorption spectrometry technique operating in the UV to measure tropospheric
N02. They have achieved an LOD of 1.6x1010/cm3 but require an integration time of 4 min
and a pathlength of 5 km. Their technique is also susceptible to absorption interferences
and can only provide an average concentration value over the path of observation.
Concentrations of NO and NO2 in the troposphere may range from the ppt to the low
ppm which requires that a method of analysis possess a linear dynamic range over as much
as 7 orders of magnitude. At 226 nm, the linear dynamic range (LDR) is about 5 orders of
magnitude extends from the LOD to over 100 ppm for NO. At higher concentrations, the
microchannel plate detector saturates resulting in a nonlinear response. The LDR at 193 nm
is about 3 orders of magnitude for NO and is limited by the LOD which is approximately 1
ppm. Similar LDRs are obtained for the NO2 molecule at both wavelengths. With
modifications to the present system design, it is expected that the sensitivity can be improved
upon significantly so that lower LODs and larger LDRs can be achieved at both 226 nm and
193 nm.
Improvements on the current system will include a higher power laser system. It was
determined in these studies that the laser energies used were insufficient to saturate the
observed ion signals. Higher laser pulse energies will lead to saturation of the ion signals
which will enhance the sensitivity and improve the signal-to-noise characteristics of the
measurements. Higher pulse energies will also allow the sample volume probed by the laser
to be expanded leading to an increase in sensitivity. Another improvement to the current
system will be a more efficient sampling apparatus. The pulsed valve used for sample
773
-------
introduction in these studies is not designed for trace analytical applications and is estimated
to transmit less than 1% of the sample gas pulse to the probe region of the apparatus. The
valve could be modified to deliver a much higher fraction of the sample into the apparatus.
CONCLUSION: A new technique for the sensitive and selective detection of NO, NO2 and
other nitrocompounds which employs a single laser for both photofragmentation and REMPI
of the target molecules has been demonstrated. Similar to previous photofragmentation/
fluorescence techniques, the current technique is characterized by real-time response, a
large linear dynamic range and high sensitivity. The photofragmentation/ REMPI technique,
which has been demonstrated at 226 nm and 193 nm using a low energy frequency-doubled
tunable dye laser and an ArF excimer laser, respectively, already achieves ppb to ppm
sensitivity for a number of nitrocompounds, but is capable of significant increases in
sensitivity with modifications to the lasers and sampling apparatus.
ACKNOWLEDGEMENTS: This work was supported by the ARL Environmental Research
Program and the ARL In-Laboratory Independent Research Program (ILIR). Purchase of
equipment through the Productivity Capital Investment Program (R. Sausa) and support from
the National Research Council/Army Research Laboratory Postdoctoral Research Program
(J. Simeonsson and G. Lemire) is gratefully acknowledged.
REFERENCES
1. J.D. Bradshaw, M.O. Rodgers and D.D. Davis, Appl. Opt. 21, 2493-2500, 1982.
2. J.D. Bradshaw, M.O. Rodgers, ST. Sandholm, S. KeSheng and D.D. Davis, J.
Geophys. Res. 90, 12861-12873, 1985.
3. J.M.C. Plane and C.-F. Nien, Rev. Sci. Instrum. 63(3), 1867-76, 1992.
4. P.J. Dagdigian, W.R. Anderson, R.C. Sausa and A.W. Miziolek, J. Phys. Chem. 93,
6059-6064, 1989.
5. S.R. Long, R.C. Sausa and A.W. Miziolek, Chem. Phys. Lett. 117, 505-510, 1985.
6. R.C. Sausa, A.W. Miziolek and S.R. Long, J. Phys. Chem. 90, 3994-3998, 1986.
7. R.C. Sausa, A.J. Alfano and A.W. Miziolek, Appl. Opt. 26, 3588-3593, 1987.
8. E.L. Wehry, R. Hohmann, J. Gates, L. Guilbault, P. Johnson, J. Schendel and D.
Radspinner, Appl. Opt. 26, 3559-3565, 1987.
9. J. Schendel, R. Hohmann and E.L. Wehry, Appl. Spectrosc. 41, 640-644, 1987.
10. M.O. Rodgers, K.Asai and D.D. Davis, Appl. Opt. 19, 3597-3605, 1980.
11. D.B. Moss, K.A. Trentleman and P.L. Houston, J. Chem. Phys. 96, 237-247, 1991.
12. M.J. McQuaid, A.W. Miziolek, R.C. Sausa and C.N. Merrow, J. Phys. Chem. 95, 2713-
2718, 1991.
13. A. Clark, K.W.D. Ledingham., A. Marshall, J. Sander and R.P. Singhal, to appear in the
Analyst, 1993.
14. S.T. Sandholm, J.D. Bradshaw, K.S. Dorris, M.O. Rodgers and D.D. Davis, J.
Geophys. Res. 95, 10155-10161, 1990.
15. M.O. Rodgers and D.D. Davis, Environ. Sci. Tech. 23, 1106-1112, 1989.
16. G.W. Lemire, J.B. Simeonsson and R.C. Sausa, Anal. Chem. 65, 529-533, 1993.
17. J.A. Syage, J.E. Pollard and R.B. Cohen, Space Division Air Force Systems
Command, Report SD-TR-88-13, February 1988.
774
-------
IMPACT OF THE NEW CAA REGULATIONS
ON THE DESIGN AND OPERATION OF
AUTOMATED VOC SAMPLING INSTRUMENTATION
Joseph Krasnec
Scientific Instrumentation Specialists
P.O. Box 8941, Moscow, Idaho 83843
Ludovit Krasnec
Department of Ecosozology, College of Natural Sciences
Comenius University, Bratislava, SLOVAKIA
ABSTRACT
Recent (1990) CAA Amendments and subsequent additions under Section 112
that set standards for 189 hazardous air pollutants require sampling and
analysis of this expanded list of air toxics. This in turn poses new challenges
to the available sampling equipment. While existing instruments continue to be
a viable avenue, new hardware and even more refined sampling techniques will be
needed. Continuosly sampling automated samplers, manual systems, in-situ
continuous analyzers and other techniques (as they become validated) will be
considered for the future sampling projects. Multi-purpose automated VOC
samplers will provide a viable approach based on many years of successful use
of manual or automated single sample systems. Expected requirements for
continuous or intermittent sampling for periods ranging from 1 to 8 hours
during required 24 hr intervals dictate deployment of state-of-the-art sampling
instrumentation. This paper will describe in some detail the design and
performance of a new generation of automated multi-station VOC samplers.
INTRODUCTION
The gradual implementation of requirements for continuous monitoring of
hazardous air pollutants results in a need for acceptable sampling and
analytical procedures. The 1988 introduction of the EPA Method TO-14 outlined
the basic parameters and hardware needs for sampling of the short list of
VOC's. Five years later the needs far outstrip the capabilities of the existing
instrumentation. Around the clock, integrated sample collection and/or analysis
in one or more locations in a multitude of environments place great demands on
the personnel and the instrumentation they use. Several recently introduced
approaches exist, such as use of in-situ, automated GC or GC/MS analytical
systems, remote sensors (IR, laser, microwave) or other alternate methods
currently under investigation. This paper will concentrate on the evolution of
the Method TO-14 instrumentation by utilizing state-of-the-art microprocessor
technology that permits automated multi-mode sampling of VOC's.
EXPERIMENTAL
Early manual sample collection proved the value of Summa passivated
sampling containers for both the qualitative and quantitative work in
monitoring VOC's. As the demands for quality and frequency of sampling
increased some degree of automation was needed. Mechanical, and later digital
timers and electrically actuated valves eased the workload and commitments of
field personnel. Holdren et al. (1) and Krasnec (2) described two automated VOC
multi-container sampling systems in 1989. Several commercially available
775
-------
systems were also introduced, but mostly for single container sampling. As the
requirements for sampling of non-polar, polar (oxygenated/carbonyl group) and
semi-volatile hydrocarbons increase so does the complexity of the sampling
hardware. Additional requirements in the Ozone Non-Attainment Areas for
sampling of the ozone precursors resulted in the establishment of a PAH
(Photochemical Assessment Monitoring) network. Because a number of stations are
already established and more are coming on-line the need for advanced VOC
sampling equipment has greatly increased.
VOC Sampler Considerations
The task of collecting multiple samples in a given 24 hour period with
reliability, flexibility and cost-effectiveness clearly dictates a need for
highly automated, unattended sampling instruments. The reliance on mechanical
and digital timers no longer satisfies the need for control, sampling
monitoring and data gathering functions. Earlier designs that used small hand-
held microcomputers showed promising performance but also some drawbacks (3)
such as sensitivity to environmental effects (temperature, vibration/shock,
electromagnetic fields, etc.), operator errors, power supply variations and
others. Use of standard PC's or even laptop and notebook size PC's is possible
but not ideal because of the cost, difficulty of use, software availability and
user training limitations.
A survey of available electronics, and prototype evaluation in 1991 led
S.I .S. to a full- time effort to design and buil d a microprocessor
(microcontroller) controlled multi-station VOC sampler. The major innovation
lies in the use of a dedicated microprocessor with a built-in real time
clock/timer, 32K RAM and 12 MHz operating speed. While not as powerful as the
latest computers on the market, the microprocessor handles the operation and
control of the sampler hardware very efficiently. It controls up to sixteen
latching solenoid valves, two pumps, up to two mass flow meters/mass flow
controllers and several relays. In addition, the microprocessor will interface
with and control external devices such as pressure transducers, temperature and
wind speed/wind direction sensors, and it will facilitate remote operation via
a modem and a built-in RS-232 interface. Further features of the microprocessor
are its ability to control the entire sampler operation, including the sample
sequencing, sampling cycle monitoring and sampling parameter storage, display
and transfer to external devices, i.e. PC's. Recent work with preproduction
samplers resulted in further enhancements. The most important are the ability
to "re-cycle" after a power failure and to resume its original sampling
sequence. Another feature is a self-calibration ability for the built-in mass
flow meter or mass flow controller. Both Span and Zero functions can be
monitored during the sampling cycle and calibration corrections implemented
while the actual sampling takes place. This, of coure, is much more desirable
than the tedious and unreliable post-sampling calibration and corrections.
Pleil (4) reported that electronic mass flow devices are subject to ambient
temperature changes, vibration and other more subtle environmental variations.
The microprocessor continuously monitors the sample flow, displays it or
alternately provides an output to an external device. At the end of the
sampling cycle it averages the flow over the entire sampling period and
provides the TOTAL SAMPLE VOLUME. Thus, the self-calibration feature is indeed
quite desirable.
Some of the additional features are the electronic duty-cycle control
(choice of sampling from 1 to 100 % of the preset time period), elapsed
(operating) time monitoring, sampling time count-down, display of real-time and
sampling/purge time intervals. Post sampling cycle data is displayed on a large
776
-------
two line LCD. It includes start time, sampling end time, power interruptions,
number of samples collected (sampling times for each sample can be the same or
individually set;, duty cyqle setting (if selected), elapsed time and the
sample flow/volume parameters described above. The microprocessor also monitors
and displays ambient (sampler) temperature. This feature allows electronic
minimum/maximum temperature recording for the sampling record or an additional
input for the sampler "self-calibration" function. Installation of an optional
pressure transducer is possible for either monitoring ambient air pressure or
sample stream pressure. The microprocessor controlled temperature sensor is
also used to regulate the temperature of the sampler enclosure. This is done
through activation of the internal cooling fan or the heating element.
Additional features, such as the sample inlet and/or sample manifold heating
can be incorporated. Wo mechanical thermostats with their inherent operational
hysteresis are used.
These features can be utilized when sampling on solid adsorbents, SepPak,
or PUF (PolyUrethane Foam) cartridges. Possibly the greatest advantage of the
microprocessor control is its inherent ease for alternate modes of sampling
(Summa container, solid adsorbent tube, or cartridge). Only relatively minor
changes in the operating software are required when switching between different
modes of sampling. It is entirely feasible to build a multi-mode sampler, i.e.
Summa container and adsorbent tube, or Summa container and POP cartridge, etc.
version. Of course, the provisions for the proper hardware components (pumpsf
valves, flow meters and the sampling media) will have to be made at the time of
the sampler construction.
The microprocessor programming language (FORTH) is available to users in
the air sampling community. It is expected that the equipment manufacturer and
some of the larger user organizations will have the in-house programming and
software modification capability. The samplers do have a built-in RS-232
interface so that any standard or a compact notebook-size PC can be used for
software loading, modifications, data retrieval and storage and subsequent
sampling report preparation. On the field-user level no programming or even
computer "literacy" is required. All sampling parameter set-up is done in
response to microprocessor prompts. For example, setting sample starting time
requires only that the operator responds to prompts to set the sampling month,
day/date, minute and second individually, and entering these parameters into
the unit. Typically, set-up of a sampling cycle takes less than five minutes.
Once it is completed, the user has an opportunity to review the sampling
parameters and to make corrections, if so desired. After this step, a single
command activates the sampler, and a count-down time to sampling start is
displayed. The actual sampling can begin hours or days later. Prior to sampling
start all valves are closed to prevent system contamination. In the event of
power failure or other sampling interruption the sampler software resets the
sampler and isolates the sampling media from contact with the atmosphere. After
the sampling resumes the system is stabilized for a preset period of time.
Then, the originally set purge period follows. Only after this step does the
actual sampling resume. The time of the sampling interruption is recorded and
reported at the end of the sampling cycle.
The user accessible software is divided into a number of menu sections and
subroutines. For example, the OPERATE menu allows access to the sampling
PROGRAM subroutine. The UTILITY subroutine provides access to the valve and
pump functions, or diagnostic functions. The REPORT subroutine gives all
previously used sampling parameters. Another sections (VARIABLES) is used for
setting important system parameters (rather than the routinely changed sampling
parameters) . This section also offers the calibration feature and is accessed
only through a special code to protect the sampler from unauthorized use.
Ill
-------
VOC Sampler Construction
Typical microprocessor controlled sampler configuration consists of the
sampler control module housed separately from the sample media module. When
used with the Summa passivated SS sampling containers this module houses up to
four 6 liter size containers. Additional two- or four-container modules can be
added as needed (Fig. !•)• All sample path components are stainless steel with
Viton seals (O-rings or diaphragms). The sample module is equipped with a
manifold that permits individual connections to sampling containers. Each
container has a separate on/off latching solenoid valve. The sampler is
equipped with an inert sampling pump; flow control hardware (mechanical flow
controller and a mass flow meter; or a. mass flow controller); sample and pump
pressure gauges, sample system pressure control valve and several ports with
on/off latching solenoid valves. All controls, including the microprocessor
LC display, control key-pad, pressure gauge displays and flow control
adjustments are mounted on the horizontal control module panel. The mass flow
meter display and control is via the microprocessor and its LCD. No separate
control and power module is required. The built-in solid state 12V DC power
supply allows operation from a rechargeable 12V DC battery, or from a regular
115V AC source.
A multi-tube adsorbent sampler configuration has been also investigated.
Up to twelve adsorbent tubes are placed horizontally, side by side on the top
of the control module. Each tube is equipped with a separate inlet on/off
solenoid valve (Fig. 2.). An optional outlet valve assembly will also be
available. All adsorbent tubes have a common inlet and outlet manifold. A
sampling pump is located downstream from the tubes rather than upstream as was
the case with the sampling containers. A flow controller and a mass flow meter
are installed in the sample stream path. Again, the pre-^set sample flow, actual
average sample flow and the total volume of sampled air for each adsorbent tube
is provided by the microprocessor. Similarly, all sampling parameter set-up,
monitoring and sampling cycle data are displayed, stored and retrieved in the
same fashion as with the Summa container sampler. A sample purge fuction is
provided to allow system purge and conditioning. The installation and
replacement of the adsorbent tubes is facilitated with the use of mounting
clips, quick-connects and flexible SS tubing connection(s) to the sampling
manifold.
It is anticipated that a modification of the above configuration can be
made for use with up to eight cartridges. Because of the greater pressure drop
caused by increased flow resistance, a more powerful pump installation is
proposed. In addition, * heated inlet line and inlet manifold zone may be
required. Also, an ozone scrubber installation needs to be considered. A
feasibility study for the modification of the above described adsorbent tube
sampler to a microprocessor controlled cartridge sampler is planned for the
near future.
CONCLUSIONS
The introduction of microprocessor controlled VOC samplers provides a
significant qualitative and quantitative step for use in the currently
implemented 1990 CAA Amendments and Section 112 Additions. Further requirements
for the PAM network, and needs of the ozone non-attainment areas for continuous
sampling of VOC's clearly underline the need for versatile, highly automated
and reliable sampling instrumentation. The flexibility and viability of the
microprocessor controlled sampler has been demonstrated. Its multi-mode
sampling capability and flexibility for collecting integrated 1 hour to 24 hour
samples needs to be demonstrated in field use, and compared to other VOC
monitoring methods.
778
-------
REFERENCES
1. M.W. Holdren, D.L. Smith and J.P. Krasnec, "Sequential canister-based
samplers for collection of volatile organic compounds," in Proceedings of the
1989 U.S. EPA/ASWMA International Symposium on Measurement of Toxic and Related
Air Pollutants, VIP-11, Air & Waste Management Association, Pittsburgh, 1989.
2. J.P. Krasnec, "An Automated Multi-Canister Sampling System for Sampling of
Volatile Organic Compounds (VOC's)," presented at the 82nd APCA Annual Meeting,
Anaheim, California, June 25-30, 1989.
3. C.W. Sweet, "Sampling and analysis of toxic volatile organic pollutants in
ambient air using an automatic canister-based sampler," in Proceedings of_ the
1990 U.S. EPA/A&WMA International Symposium on Measurement .of Toxic and Related
Air Pollutants, VIP-12, Air & Waste Management Association, Pittsburgh, 1990.
4. J.D. Pleil, U.S. BPA,
1988.
Research Triangle Park, W.C., personal communication.
NEEDLE VALVE
INLET BELLOWS VALVE
TWO - WAY LATCHING SOLENOID VALVE
GAUGE
TWO - WAY LATCHING SOLENOID VALVE
FLOW BYPASS / PURGE
FLOW CONTROLLER
MASS FLOW METER
LCD DISPLAY
MICROPROCESSOR
SCIENTIFIC INSTRUMENTATION SPECIALISTS
6 LITER SAMPLING CONTAINER
Legend for Figure 1.
779
-------
S I.S. MICROPROCESSOR CONTROLLED
ADSORBENT TUBE SAMPLER MODEL' ATS - 12 / MPC
MICROPROCESSOR CONTROLLED 8 CONTAINER
VOC SAMPLER MODEL AGS - 8/MPC
PUMP I
TWO - WAV LATCHING SOLENIOD VALVE
FLOW CONTROLLER
MASS FLOW METER
LCD DISPLAY
MICROPROCESSOR
-------
PRESSURE AND TEMPERATURE EFFECTS ON CONCENTRATION OF GASEOUS
CALIBRATION STANDARDS
James Me Andrew, and Keith Rogers and Dmitry Zmimcnskv
American Air Liquide
5230 S. East Ave
Countryside. IL 60525
ABSTRACT
It is well-know n that the concentration of gaseous calibration standards may be affected by the pressure
and temperature of the standard. For example, if the component of interest has a low vapor pressure it
may condense out of a standard prepared at loo high a pressure or exposed to loo low a lempcralure. On
the other hand, very low concentration cylinder standards arc usually only considered reliable if a certain
minimum pressure remains in the cylinder. The maximum allowable concentration before condensation
occurs in a slandard can be calculaled from ihermodynamics. taking into account non-idealities at high
pressures, as will be described. Although this calculation is straightforward enough from a fundamental
poinl-of-view. it is not always understood in praclice and it is necessary lo estimate key parameters for
many species currently of interest in environmental monitoring.
For many systems, il is not sufficient to consider condensation bin one must also take adsorption-
desorpiion phenomena inlo account. The pressure dependence of these phenomena is less well understood
bul a model has recently been proposed by Li et al. The implications of this model for cylinder standards
will be discussed.
INTRODUCTION
In order to make meaningful analyses of trace almospheric pollutants, it is essential to have good
calibration standards. Compressed gas mixtures in cylinders are widely used for this purpose because they
arc convenient, robust and well-characterized. As the range of pollutants of interest increases, more
compounds are added whose stability in cylinders requires special care on the part of the standard
manufacturer, ll is important that the user of these standards have some understanding of factors
influencing the concentration of impurities in calibration standards in order to consistently achieve the
performance they require. In this paper we will briefly review some important fundamentals of vapor-
• liquid equilibria (VLE) and then move on to more recent work in the area of gas-surface interactions in
cylinders. Some of the considerations which will be discussed in this paper for the case of cylinders, are
also relevant to air samples collected in canisters, although those are generally at much lower pressures.
CONDENSATION QUESTIONS IN CYLINDER GASES
As the number of molecules of any component in a gaseous calibration slandard is increased, a condensed
phase will eventually form and limit the gas phase concentration from increasing further. Gaseous
standards are usually prepared in a manner which ensures thai only gas phase species are present as
otherwise the concentration in the gas phase will change dramalically as the cylinder is emptied.
The simplest attempt to calculate Ihe maximum concentration of an componenl which can exisl in a
purely gas phase mixture is to ratio the vapor pressure of the component of interest to the total pressure in
the cylinder. This type of calculation predicts that the maximum possible concentration in a cylinder
decreases monolonically as the pressure increases, or, to put it another way, that the dew-point of a
mixture will increase monolonically wilh temperature. For cylinder gases, however, it necessary to allow
for Ihe effects of interactions of the matrix gas with impurities. Such interactions can be rather large al
high pressures. Recall lhal Ihe Critical Temperature of Nitrogen is -I47°C and that its Critical Pressure is
493 psia. which means that nilrogen in a typical gas cylinder is in a supercrilical condition and thus
781
-------
should often be considered as a solvent rather than as an inert matrix in which the organic molecules of
interest are situated.
If we use the thermodynamic concept olfugacin- (which is closely related to chemical potential), the
fugacity of component i (/,') must be equal in all phases, i.e.
/;"»•""= /:«»< (i)
from v hich it can be show n * lhal the mole fraction of component i. vy, in the gas phase is given by
/>"" -vy an "enhancement factor" which is usually greater than one. The exponential term is known as the
Poynling correction and lakes into account the change in free energy of a condensed phase of molar
volume \-f due to its being at pressure P rather than the saturated vapor pressure. It is usually between 1
and 3. jr, is the mole fraction of i in the condensed phase and will be close to I if component i is the
least volatile present. The fugacity coefficients, tf, arc defined by
ip;"" accounts for the non-ideality of the saturated vapor and. as we are by definition dealing with
impurities of low vapor pressure, is usually close to 1. (p'1(/"J'is the fugacily coefficient of the component
of interest in the vapor phase and is responsible for large enhancement effects (up to several orders of
magnitude).
Y, is the activity coefficient of component i in the condensed phase. If the condensed phase is pure, then
Y, = I • otherwise Jt provides a measure of the interactions between organic species in multicomponent
mixtures. This is the only factor which is likely to result in a reduction of the actual concentration with
respect to the ratio P"' / P For example, if one component were present at a concentration which was
high relative to its vapor pressure, not only might it condense, but other components might dissolve in it,
with the result lhal their concentrations would also be lower than expected.
Equation (2) is useful for calculations below the critical point and allows useful indicators of the direction
of expected trends to be derived. In practice, however, activity' coefficients are not available at high
pressures and other complications arise. It is therefore more effective to use an equation of slate which
can describe both the liquid and the gas phase and solve Equation ( I ) directly. Many such equations exist
(such as the Soave-Redlich-Kwong (SRK) and Peng-Robinson equations) and computer programs capable
of carrying out the calculations are readily available. Sometimes it is necessary to estimate parameters
necessary for the description of more unusual organic molecules or species which are not important in the
chemical industry (where VLE calculations are used extensively).
Application to practical mixtures
It is interesting to note that equation of state calculations such as those discussed above are able to take
interference effects between compounds into account reasonably well. p-Dichlorobeiuene is a solid
organic compound: It is included at I ppm in a 4 1-component mixture sold for TO-14 analyses. The other
components of that mixture are present at 2 ppm and cover a wide range of volatility, but p-
dichlorobeiKene is the least volatile. Figure 1 illustrates the application of VLE calculations to several
mixtures containing I ppm (by volume) p-dichlorobenzene in balance nitrogen. The data are presented
782
-------
- 1600 -
1
Figure 1: Dew Point Curves calculated using the SRK equation of state. Results are shown for
the following mixtures in Nitrogen: 1 ppm p-dichlorobenzene, 1 ppm p-dichlorobenzene 2 ppm
each o-,p-dichlorobenzene, 1 ppm p-dichlorobenzene, 2 ppm each o,-m-dichlorobenzene, 1,2,4-
trimethylbenzene, p-ethyl-toluene, o~,m-,p-xylene, ethylbenzene, 1,1,2,2-tetrachloroethane,
styrene, 1,1,2-trichloroethane, toluene, benzene and chlorobenzene.
in the form of a plot of dew point of the mixture as a function of cylinder pressure. The dew point
increases with pressure at low pressure, as expected, goes through a maximum and then decreases again.
For a mixture containing only p-dichlorobenzene, the dew point never exceeds -30°C. Including the other
isomers of dichlorobenzene has a substantial effect on the dew point. Bringing the total number of
components up to 16 by including more low-volatility compounds has a substantially smaller effect and as
we work through the list of compounds in reverse order of elution (i.e. in order of increasing volatility),
each additional compound has less effect on the dew point of the mixture.
Comparison with experimental results
The thermodynamic description of vapor-liquid equilibria reviewed above has been extensively verified in
practice (see, for example, the examples in reference 1). In particular, the shape of the curves in Figure 1
is typical and similar behavior is observed for a wide range of compounds. However, gas-surface
interactions in cylinders are much less well understood and their description does require proper
incorporation of the effects of high pressure. With this in mind we prepared a set of mixtures which were
deliberately designed to exhibit condensation effects at accessible temperatures. The composition range
of the mixtures was 9-15 ppm benzene, toluene, p-dichlorobenzene, 40-75 ppm chlorobenzene. p-
Dichlorobenzene, a solid compound of low volatility, was intentionally included at a relatively high
concentration. This species is usually supplied at concentrations of 1 ppm or less. The concentration of
chlorobenzene was sufficiently high that we could expect to describe its behavior reasonably well while
neglecting gas-surface effects, at least at high pressure (see the next section). The mixtures were
prepared in disposable "MedGas E" cylinders (supplied by Alphagaz) which are carbon steel.
In order to observe condensation effects, even for these specially prepared mixtures, it is necessary to cool
the cylinder. A range of cylinder temperatures was achieved by using ice or coolant baths .Figure 2
compares the results of several measurements made on these mixtures with the results of equation-of-state
calculations using the Soave-Redlich-Kwong (SRK) equations. Curves based on the ratio of vapor
pressure to cylinder pressure are also shown for the purpose of comparison.
783
-------
For a cylinder containing 9 ppm each of benzene, toluene p-dichlorobenzene and 42 ppm chlorobenzene
in a balance of 1000 psi nitrogen, excellent agreement is obtained with the SRK calculation, as shown in
Figure 2. Note that the concentration at -34°C is substantially larger than predicted by simply dividing
the vapor pressure by lOOOpsi, in accord with the discussion above.
In cases where a condensation problem is anticipated, one may be inclined to use a cylinder prepared at
lower pressure in or "to be safe". This is not always advisable. In Figure 1, however, we see that a rather
small difference in dew point can exist even though the cylinder pressure varies by a large factor (e.g.
between 500 and 1000 psi). For the case of a cylinder prepared with 15 ppm each benzene, toluene, p-
dichlorobenzene, 75 ppm chlorobenzene, the SRK equation-of-state calculation predicts a very small
concentration difference between a 1000 psi cylinder and a cylinder at 500 psi. In Figure 2, these curves
are superimposed and cannot be distinguished. A cylinder was prepared using the above composition at
lOOOpsi and analyzed at 25°C and at 0°C. The lower temperature data showed a decrease in
concentration. The cylinder was allowed to "blow down" to 500 psi at room temperature (using a
sufficiently slow flow rate that no cooling occurred). The measured concentrations were the same as
before at both temperatures, although one might have expected to avoid condensation at 0°C based on the
curve labeled "vapor pressure/500 psia". The error bars (for one standard deviation) on the experimental
data are approximately the same size as the point markers in the figure, so the SRK calculation is
significantly higher than the observed values at 0°C, but correctly predicts the trend and is substantially
closer to the observed values than either vapor pressure ratio curve.. This is an example where the
relatively high concentration of p-dichlorobenzene leads to condensation at 0°C and affects the
concentration of chlorobenzene.
GAS-SURFACE EFFECTS
In practice, calibration standards for air analysis are generally prepared at compositions where
condensation is not an issue. Indeed, from Figure 1, one might suppose that one could use many mixtures
Figure 2: SRK Equation of State Calculations for: (i) a mixture containing 9 ppm each benzene,
toluene, p-dichlorobenzene, 42 ppm chlorobenzene in 1000 psia nitrogen balance (ii) a mixture
containing 15 ppm each benzene, toluene, p-dichlorobenzene, 75 ppm chlorobenzene in 500
psia nitrogen and in 1000 psia nitrogen. Experimental data for both concentrations and for the
75 ppm chlorobenzene system at both pressures are indicated. The result of dividing the vapor
pressure (v.p.) of chlorobenzene by 500 psia and by 1000 psia are included for comparison.
784
-------
at temperatures below freezing without difficulty. Unfortunately, this is not the case because of the
importance of gas surface effects. These can be broadly divided into two categories:
Physisorption: Molecules are bound to the surface by relatively weak physical forces which do not lead to
breakage of any intra-molecular bonds
Chemisorption: Molecules are strongly bound to the surface and intra-molecular bonds may be broken.
The distinction is important in practice because physisorption is reversible: for example, if a cylinder is
exposed to extremely low temperature in storage, some fraction of the impurity molecules may be
physisorbed, but once the cylinder is restored to a more appropriate temperature the original impurity
concentration will be recovered (provided no substantial amount of gas is withdrawn while the cylinder is
cold). Chemisorption, however, is more difficult to reverse. In order to make stable calibration standards
it is necessary to properly prepare the gas cylinder so that the surface does not chemisorb the species of
interest, or, more precisely, so that the number of available Chemisorption sites is small compared to the
number of available gas phase molecules. As the concentration of the calibration standard is reduced this
becomes more difficult.
Model for Gas-Surface interactions in High Pressure Cylinders
Recently, Li et al ^ incorporated the thermodynamic treatment of pressure effects on fugacity as described
in the preceding section into a model which also allowed for adsorption on the cylinder walls according to
a Langmuir isotherm. They applied this to the description of the pressure dependence of moisture
concentration in cylinders of Ultra-High Purity (UHP) Nitrogen. The same model can be applied to
organic species present in calibration standards.
When a cylinder of UHP gas is at full pressure, and if it was properly dried before filling, its moisture
concentration is essentially that of the gas source used, which is usually very low. As the gas is depleted,
the moisture concentration gradually rises due to desorption from the cylinder walls, usually with a sharp
increase at some low pressure (about 100 psi). The detailed behavior depends on the cylinder material
and on the procedure used to dry the cylinder. This behavior is correctly predicted by the model.
0 100 200 300 400 500 600 700 800 900 1000
Figure 3. Effect of Adsorption-Desorption on the concentration of a gaseous calibration
standard. The predictions of the model of Li et al are shown for two sets of gas-surface
interaction parameters (see text). The "low binding energy" curve displays the behavior typically
observed for moisture in a gas cylinder. Experimental data for the four aromatic species in
mixtures (i) and (ii) discussed under Figure 1 are also shown. These mixtures were prepared in
carton-steel cylinders.
785
-------
The adjustable parameters used by the model are the number of surface sites available and the equilibrium
constant for surface adsorption. One of its more interesting features is that it predicts a iinatitatlvefy
different pressure dependence of the concentration for different ranges of values for these parameters.
This is illustrated in Figure 3. The effects of the two parameters are convoluted together and are difficult
to completely separate. If the average gas-surface bond is weak or if the number of sites is small, then the
concentration increases at low pressure, if the average bond is strong, or there are many sites, it decreases.
As discussed above, moisture shows the typical "weak binding" behavior. When the series of cylinders
discussed earlier were slowly reduced in pressure, they showed the opposite, strong binding, effect. The
model's predictions are not in perfect agreement with experiment but it is clearly capable of reproducing
qualitative features correctly. It is striking that the dependence is approximately the same for all four
species despite a wide range of volatilities and the fact that chlorobenzene is present at approximately 5
times higher pressure than the other three species. These mixtures were prepared in carbon steel
cv linders: mixtures prepared in aluminum cylinders show qualitatively the same behavior (in our
experience), but the value of the pressure at which concentration starts to drop off may be somewhat lower
CONCLUSIONS AND PRACTICAL IMPLICATIONS.
This discussion has been far from exhaustive and it should be remembered that other factors can come
into play which were not mentioned here. One obvious candidate is reactions between impurities in the
same calibration standard, but this can generally be avoided as most incompatibilities are understood.
Molecules which tend to polymeri/e give rise to more difficult problems which often require an empirical
approach to the selection of appropriate cylinder treatment, fill pressure and concentration.
In preparing standards, methods for establishing concentrations which will not lead to condensation
problems are relatively well-established and require only access to suitable computer code and the ability
to successfully estimate some parameters. Gas-surface interactions are less well understood, but a model
capable of reproducing observed behavior exists and we can expect our understanding to develop.
Generally speaking, mixtures at higher concentrations and higher pressures give rise to fewer problems.
It is important not to use underestimate the pressure at which condensation will occur for a given mixture
composition, as too low a pressure may give rise to problems due to surface effects.
In using calibration standards it is necessary to take reasonable precautions not to expose them to
extremes of temperature or to deplete them to too low a pressure. If a cylinder is exposed to low
temperature (below 40"F) it should be allowed to warm up before any gas is withdrawn. The cylinder
should not be completely depleted as the concentration will vary. Avoiding pressures below 300 psi
appears to provide a reasonable margin of safety.
ACKNOWLEDGMENTS
Thames to Benjamin Jurcik for man}' useful conversations and instruction in the calculation of Vapor-
Liquid-Equilibria. Thanks to Yao-En Li for explaining his model and providing the computer program
embodying it which was prepared by Malt Giacobbe.
REFERENCES
1. J.M. Prausnite, R.N. Lichtenlhaler and E. Gomes de Axcvedo, Molecular Thermodynamics of Fluid
Phase Equilibria (2nd Edition); Prentice-Hall. Englewood Cliffs, New Jersey. 1986
2. Y.E. Li. J.Rizos, D. Vassallo and M. Giacobbe "The Behavior of Moisture in High Pressure Inert
Cylinder Gases" in Proceedings of the 12th International Vacuum Congress. 8th International
Conference on Solid Surfaces, The Hague. Netherlands.
786
-------
Session 24
General
-------
LABORATORY EVALUATION OF GAS DILUTION SYSTEMS FOR
ANALYZER CALIBRATION AND CALIBRATION GAS ANALYSIS
Robert S. Wright
Robert W. Murdoch
Research Triangle Institute
P. O. Box 12194
Research Triangle Park, North Carolina 27709
ABSTRACT
Gas dilution systems can be used for multipoint calibrations of pollutant gas analyzers and
for the analysis and certification of compressed gas calibration standards. This laboratory evaluation
obtained estimates of the accuracy and precision of four representative gas dilution systems. Diluted
and undiluted gas mixtures containing carbon monoxide, nitric oxide or sulfur dioxide in nitrogen
were sampled by pollutant gas analyzers. Accuracies were estimated from the difference between the
slopes of regression lines from measurements of the diluted and undiluted gas mixtures. These
accuracy estimates ranged from -3.3 to 10.0 percent. Precisions were estimated from the 95-percent
uncertainty for regression-predicted concentrations. These precision estimates ranged from 0.3 to
14.9 percent of the predicted concentration for mid-range dilutions.
INTRODUCTION
Calibration gases often are diluted during the multipoint calibration of pollutant gas analyzers
and during the analysis and certification of compressed gas calibration standards. Gas dilution
systems are devices that allow two gas streams to be mixed together continuously and quantitatively.
Such systems could be used for multipoint calibrations and analyses of gas mixtures. However,
systematic and random errors associated with the dilution would be added to the total measurement
uncertainty. Could one use gas dilution systems for these applications with acceptable levels of
accuracy and precision?
Under U.S. Environmental Protection Agency (EPA) sponsorship, Research Triangle Institute
(RTT) conducted a laboratory evaluation of four representative gas dilution systems. The goal of the
evaluation was to estimate the accuracies and precisions of the systems using measurements by
pollutant gas analyzers. These estimates were obtained from the statistical analysis of multipoint
calibrations using undiluted and diluted gas mixtures.
GAS DILUTION SYSTEMS
The Environics Model 2020 continuous emissions monitoring calibration system uses thermal
mass flow controllers to regulate the flow rates of a compressed gas calibration standard and dilution
nitrogen, which are blended together. The system was configured to operate in both of the dilution
regions that were used in the evaluation (i.e., 1 to 9 percent and 10 to 90 percent of the standard's
concentration). Its specified flow rate accuracy is ±0.5 percent of the full-scale range and its
specified flow rate repeatability is ±0.2 percent of the full-scale range.
789
-------
The Milton Roy Model 82IS gas divider uses 10 identical, parallel capillary tubes and
solenoid values to blend the calibration standard and the dilution nitrogen. The system was
configured to operate in the 10 to 90 percent dilution region. Its specified dilution accuracy is ±0.2
percent of the standard's concentration and its specified repeatability is ±0.1 percent of the point.
The Wb'sthoff Model 1KM67 gas mixing pump uses 2 piston-driven, positive-displacement
pumps to blend the calibration standard and the dilution nitrogen. The standard goes through one
pump and the nitrogen goes through the other pump. Switchable gears vary the pistons' stroke
frequency and thereby vary the diluted gas concentration. The pumps and gears in the instrument
used for this evaluation were lubricated with silicone oil. The system was configured to operate in
the 1 to 9 percent dilution region. Its specified accuracy under calibration conditions is ±1.0 percent
relative to the selected dilution level.
The Wosthoff Model 5KA37 gas mixing pump uses 5 piston-driven, positive-displacement
pumps to blend the calibration standard and the dilution nitrogen. The pumps are an oilless type and
have a fixed stroke frequency. Three stopcocks are used to switch the gas streams entering 3 of the
5 pumps and thereby vary the diluted gas concentration. The system was configured to operate in
the 1 to 9 percent dilution region. Its specified accuracy under calibration conditions is ±0.5 percent
relative to the selected dilution level.
EXPERIMENTAL PROCEDURES
Gas mixtures containing carbon monoxide (CO), nitric oxide (NO), or sulfur dioxide (S02) in
nitrogen were measured during the evaluation. These measurements were made using the analytical
instrumentation listed in Table 1. Several gas dilution systems were compared within a single day's
time. All measurements that were obtained on the same day were considered to be comparable and
no measurements were compared across different days.
Multipoint calibrations were performed using undiluted and diluted gas mixtures. The
undiluted gas mixtures were National Institute of Standards and Technology Standard Reference
Materials (NIST SRMs) and zero-grade nitrogen. These gas mixtures were the reference standards
for the accuracy estimates. The diluted gas mixtures were obtained by dilution of EPA Protocol
Gases, which are prepared and analyzed by specialty gas producers using an EPA-specified protocol.
RTI verified the certified concentrations of these standards relative to the concentrations of the
SRMs.
The calibration data were analyzed using least-squares regression techniques under the
assumption of a constant random error term. For several multipoint calibrations, a quadratic equation
fitted the data better than a straight-line equation. The accuracy of the gas dilution systems was
estimated by comparing the slope of the regression line for the diluted gas mixtures with the
corresponding slope for the undiluted gas mixtures.
Accuracy innT Dil"^ Slope - Undiluted Slope!
\_ Undiluted Slope J
This value is constant across the range of predicted concentrations. Statistical tests were performed
to check if the diluted slope was significantly different from the undiluted slope.
The precision of the gas dilution systems was estimated by the relative uncertainty of
concentrations that are predicted from the regression equation.1
Pr ' ' = inflT^S-Percent Uncertainty for Predicted Cone.
Predicted Concentration
790
-------
Although the 95-percent uncertainty is approximately constant across the range of predicted
concentrations, the precision estimate varies inversely with the predicted concentration. This
variation is illustrated in Table 2 which gives precision estimates obtained from the multipoint
calibration of the NO analyzer on its 0 to 250 parts per million (ppm) range. The precision at the
upper end of the regression line is approximately 0.5 percent, but the precision at the lower end is
approximately 9 times larger. In this paper, precision estimates are generally listed for dilutions of 5
or 50 percent of the diluted standard's concentration.
The precision estimates also vary with the number of measurements in the multipoint
calibration and with the number of measurements of the unknown sample. This variation is
illustrated in Table 3, which gives precision estimates obtained from various subsets of the 250 ppm
NO multipoint calibration. The precision estimates for a 5-point calibration and 1 sample
measurement are approximately 2 to 3 times larger than the precision estimate for a 10-point
calibration and 3 sample measurements. This example demonstrates that precision estimates can be
improved by increasing the number of measurements.
Several unexpected events occurred during the evaluation. Excessive zero drift in the NO
analyzer was discovered and repaired. Several days' data were discarded because of this zero drift.
One of the capillaries in the Milton Roy 821S system became plugged during the evaluation and this
system could no longer be used. The slope of the diluted SO2 regression line for the Wosthoff
1KM67 system was significantly different from the slopes of the undiluted gas mixtures and the
other systems' diluted gas mixtures. This difference may be due to interference of silicone oil from
the Wosthoff 1KM67 system on the ultraviolet fluorescence SO2 analyzer. Additionally, oil was
seen coming from the system's nitrogen pump and a dropout trap was installed to prevent
downstream oil contamination. Worse than expected precision estimates for the Wosthoff 5KA37
system prompted its inspection and repair by its manufacturer. A minor leak was discovered and the
system was reevaluated at RTI after it was repaired.
RESULTS
Several caveats concerning the evaluation should be considered as the results are reviewed.
The evaluation studied only a limited range of experimental conditions. The evaluation did not
attempt to obtain optimum performance (e.g., better precision estimates) from the gas dilution
systems due to time constraints on experiments. Not all of the gas dilution systems were evaluated
in the same dilution region. The gas dilution systems that were evaluated were selected based only
on their availability and their dilution principles. The evaluation did not assess the corresponding
accuracy and precision of undiluted gas mixtures which might be used in multipoint calibrations.
The results of the evaluation are given in Table 4 for the Environics and Milton Roy systems
at the higher-concentration dilutions (i.e., 10 to 90 percent of the standard concentration). The
results for the SO2 and CO multipoint calibrations are given for both straight-line and quadratic
regressions. The SO2 analyzer is slightly nonlinear and the CO analyzer is quite nonlinear.
Accuracy estimates were not calculated for the quadratic regressions. The Milton Roy 821S system
was not evaluated beyond the NO and SO2 calibrations because of the plugged capillary.
The results of the evaluation are given in Table 5 for the Environics and Wosthoff systems at
the lower-concentration dilutions (i.e., 1 to 9 percent of the standard concentration). Results are
given for the Wosthoff 5KA37 system both before and after its repair. The last two after-repair
accuracy estimates may be biased due to a concentration shift in the SRM associated with low
cylinder pressure. The SO2 calibration was replicated to verify the slope accuracy estimate for the
Wosthoff 1KM67 system.
791
-------
CONCLUSIONS
The number of experiments conducted in the evaluation is too small to allow one to draw
any firm conclusions concerning the accuracy of the gas dilution systems. For 12 of the 25 accuracy
estimates, there was no statistically significant difference between the slopes of the regression lines
for the undiluted and diluted gas mixtures. The values of the remaining 13 accuracy estimates
ranged from -4.5 to 10.0 percent although the extreme values are associated with a nonlinear
calibration curve and a possible interference effect The accuracy estimates exhibit some variations
from one day to the next and from one experimental condition to the next. In general, the accuracy
estimates for the four gas dilution systems were similar.
The precision estimates for the four gas dilution systems were also generally similar. The
values range from 0.3 to 14.9 percent for predicted concentrations in the middle portion of the
regression line. The largest values are associated with a leaking system. The precision estimates
exhibited some variations from one day to the next and from one experimental condition to the next
The precision estimates were significantly different in different portions of the regression line. Better
precision can be obtained by making more measurements during multipoint calibrations and sample
measurements.
DISCLAIMER
Although the research described in this paper has been funded wholly by the U.S. EPA
through Contract No. 68-D1-009 to RTI, it has not been subjected to Agency review and, therefore,
does not necessarily reflect the views of the Agency and no official endorsement should be inferred.
Mention of trade names or commercial products does not constitute endorsement or recommendation
for use.
REFERENCES
1 M. G. Natrella, Experimental Statistics. National Bureau of Standards Handbook No. 91, U.S.
Government Printing Office, Washington, D.C., 1963, pp. 5-1 to 5-46.
Table 1. Analytical instrumentation for laboratory evaluation of gas dilution systems.
Gas Mixture Pollutant Gas Analyzer Analyzer Range (ppm)
For Dilution of 10%
to 90% of Std. Cone.
CO/N2
NO/N2
Bendix 8501-5CA
Thermo Electron 10AR
Columbia Scientific 1600
1000
250
For dilution of 1% to
9% of Std. Cone.
50
25 and 100
5
SO2/N2 Thermo Electron 40 500 50
792
-------
Table 2. Comparison of precision estimates associated with different predicted
concentrations for the same multipoint calibration.
NO Analyzer
Range
(PPM)
250
Gas Dilution
System
Environics 2020
Milton Roy 821 S
At Dilution of
90% of Std.
Cone.
0.59
0.50
Precision (percent)
At Dilution of
50% of Std.
Cone.
0.97
0.85
At Dilution of
20% of Std.
Cone.
5.14
4.53
Table 3. Comparison of precision estimates associated with various
number of measurements of calibration standards and sample.a'b
Measured Values Used
in the Regression
Calculations
Odd Values0
Even Values'1
All Values
^Jas dilution system
bAnalyzer range
cOdd values
dEven values
No. of Values in
Regression
5
5
10
Environics 2020
250 ppm NO
10, 30, 50, 70, 90% of std
0, 20, 40, 60, 80% of std.
Precision at Dilution
1 Sample
Measurement
1.85
1.20
0.97
. cone.
cone.
of 50% of Std. Cone.
3 Sample
Measurements
1.24
0.81
0.61
Table 4. Results for higher concentration dilutions (10 to 90 percent of standard concentration).
Analyzer Range Gas Dilution
Gas Mixture
NO/N2
S02/N2
S02/N2b
CO/N2
C0/N2b
(PPM)
250
500
500
1000
1000
System
Environics 2020
Milton Roy 821S
Environics 2020
Milton Roy 821S
Environics 2020
Milton Roy 821S
Environics 2020
Environics 2020
Slope Accuracy
(Percent)
1.60
0.25
2.50a
0.74
—
-4.53a
—
Precision at
Dilution of 50%
of Std. Cone.
0.97
0.85
1.24
2.38
0.76
2.12
10.90
0.47
Statistically significant difference between diluted and undiluted slopes
Quadratic regression
793
-------
Table 5. Results for lower concentration dilutions (1 to 9 percent of standard concentration).
Gas Mixture
NO/N2
NO/N2
(after repair)
N0/N2b
SO2/N2
S02/N2
CO/N2
Analyzer
Range
(PPM)
100
100
5
50
50
50
Gas Dilution System
Environics 2020
Wosthoff 1KM67
Wosthoff 15KA37
Wosthoff 5KA37
Wosthoff 5KA37
Wosthoff 5KA37
Wosthoff 5KA37
Wosthoff 5KA37
Environics 2020
Wosthoff 1KM67
Wosthoff 5KA37
Environics 2020
Wosthoff 1KM67C
Wosthoff 5KA37
Environics 2020
Wosthoff 1KM67C
Wosthoff 5KA37
Environics 2020
Wosthoff 1KM67
Wosthoff 5KA37
Slope Accuracy
(Percent)
-0.72
-0.53
-0.43
1.25a
0.98"
1.08a
2.17a
2.11a
1.39a
0.11
0.34
2.62a
9.87a
0.51
-1.97
10.04"
-1.52
-3.33a
-0.90a
1.09
Precision at
Dilution of 5% of
Std. Cone.
1.01
1.05
14.48
1.52
1.39
2.13
1.38
1.59
0.31
0.59
12.64
3.13
3.81
14.91
2.08
5.22
12.28
0.74
1.11
9.75
Statistically significant difference between diluted and undiluted slopes
Reference standard was diluted for these measurements
Possible interference effect from silicone oil
794
-------
A DUAL PURGE & TRAP / AIR TOXICS PRECONCENTRATION SYSTEM
FOR 16-POSITION GC/MS ANALYSIS OF CANISTERS, WATERS, AND
SOILS.
Leon Levan Daniel B. Cardin
Apollo Analytics, Inc. Entech Laboratory Automation
2960 Airway Ave., Suite B-101 950 Enchanted Way, #101
Costa Mesa, CA 92626 Simi Valley, CA 93065
ABSTRACT
In order to achieve the necessary detection limits for the EPA 500 Series drinking water
methods and ambient Air Toxics monitoring using EPA Method TO 14, the preconcentrators
and GC/MS systems utilized must be optimized. This usually includes implementing a direct
coupling of the analytical column to the mass spectrometer to insure that the entire sample is
introduced into the source. Due to the 1-2 ml/min limitation of current benchtop mass spec-
trometers, an on-column focusing stage is generally incorporated to further reduce the sample
volume for improved chromatography of the lighter target analytes.
An Entech 2000 preconcentrator has been modified to allow its focusing trap to be used for
focusing both canister samples and purge and trap samples, Full contact communication is
provided to an OI 4560 Purge and Trap with 16-position manifold to allow unattended opera-
tion. The interface also permits full use of the GC injector allowing a convenient way to
introduce standards, BFB, and high concentration air samples. The GC injector and purge
and trap are connected in series before the focusing trap on the Entech 2000 therefore elimi-
nating the need for Y connectors and splitters. The system can be set up to switch automati-
cally from waters or soils to canisters after all preprogrammed positions are completed.
INSTRUMENTATION
A diagram of the system is shown in FIGURE 1. A Hewlett-Packard 5890 Series II GC with
EPC was interfaced to an OI 4560 Purge and Trap by connecting 1/16" nickel 200 tubing to
the bottom of the injector to deliver the GC carrier directly into the 6-port rotary valve on the
purge and trap. The 1/16" tubing was heated to allow heavy VOC's to be introduced at the
GC injector (BFB, ppm standards and samples, etc.). The standard heated transfer line from
the purge and trap was routed directly into port 4 of the Entech 2000 Automated
Preconcentrator 8-port rotary valve allowing on-column focusing. The standard heated
transfer line from the Entech 2000 was then connected to the analytical column in the 5890
GC oven using a zero dead volume union.
Purge and Trap of both waters and soils was performed using a multibed VOCARB trap from
Supelco (Bellefonte,PA) allowing samples to be both quantitatively trapped and efficiently
795
-------
dry-purged before desorption to the 2000 Preconcentrator. Focusing was performed on a
deactivated, uncoated megabore column (0.53mm ID) at temperatures below -170 deg. C
followed by a 10,000 deg. C per minute initial temperature ramp upon injection to maintain
narrow peak widths. Separations in the GC were accomplished on a J&W 60m DBS column
(0.32mm ID, lum film) using a 35 deg. C starting oven temperature.
Communication between the GC, Purge and Trap, and 2000 Preconcentrator was performed
using contact closures. Rather than implementing the classical approach of using the oven
ready signal to start sample desorption from the purge and trap, the Aux Out signal from the
Entech 2000 was used to indicate when the focusing trap was cold and ready to accept the
desorbed sample. The focusing trap, in turn, cooled only when both the sample list in the
2000 controller software indicated that another sample remained to be focused and when the
GC ready signal was obtained. The overall operation is described in the flow chart in FIG-
URE 2.
DISCUSSION
The described system uses a totally new approach for interfacing the GC carrier gas to both a
Purge and Trap and Air Toxics preconcentrator. Instead of teeing these systems in parallel,
each component is placed in a loop in series. The GC carrier first passes through the injector
where sample could be introduced manually, allowing direct BFB injections. Source level
samples could also be introduced using gas-tight syringes when higher VOC concentrations
require 0.1 - 2.0 cc sample volumes. The GC carrier is then accessed at the bottom of the
injector to deliver the regulated helium directly into the heated rotary valve on the purge and
trap. Connection at the bottom of the injector eliminates the usual modifications of the GC
(GC/MS) flow regulation pneumatics thereby allowing much easier interfacing to sophisti-
cated flow/pressure controllers such as Hewlett-Packard's Electronic Pressure Control
(EPC). In addition, bypassing of the entire purge and trap/Air Toxics inlet for toubleshooting
purposes is easily accomplished by removing the transfer line at the bottom of the injector
and replacing it with the analytical column.
From there, the GC carrier gas is routed to the purge and trap for VOC recovery from the
sorbent trap when requested. Adequate focusing times during trap desorption are dependent
on GC carrier flow rates and are fully adjustable through the Entech control software. For a
flow rate of 1.5cc/min, 4 minutes worked well. Desorption of the focusing trap occurs very
rapidly providing the narrow peaks given in figures 3 and 4. Both chloromethane and
vinylchloride are shown to have gaussian peak shapes indicating that no dead volume was
experienced during the injection. Peak widths for these early eluters were comparable to
those of analytes eluting later in the analysis.
Finally, the GC carrier is introduced directly into the 8-port valve in the 2000
Preconcentrator to also permit introduction of VOC's obtained from air samples into the
focusing trap. Canisters and tubes are first concentrated in modules 1 and 2 before
backflushing into the focusing trap.
796
-------
CONCLUSIONS
A new Air Toxics/P&T/GCMS interface has been demonstrated that serially connects mul-
tiple inlets into the GC carrier flow path before a final focusing stage. Some of the advan-
tages realized with such an approach include:
1. No hardware changes necessary when switching
from purge and trap to canisters or tubes.
2. Provides maximum utilization of GC/MS hardware
and data systems.
3. Permits laboratories to work their way into the air
market at lower initial costs by time sharing a GC/
MS with purge and trap, canister, and tube
autosamplers.
4. Provides better sensitivity and resolution of
lighter VOC's in water and soil analysis.
5. Forces all retention times and peak shapes to be
identical for a given analyte no matter which inlet
the analyte was introduced from (common focus
point).
6. Allows a convenient way to check method recover-
ies by P&T and TO 14 by comparing responses to
direct injection of concentrated (or methanol)
standards.
7. Focusing is performed after the injector so typical
peak broadening that would occur therein are not
present.
8. The novel design of the Entech 2000
Preconcentrator focusing trap allows it to operate
on about 50 cents worth of liquid nitrogen per
analysis.
797
-------
"Software" Selection of Canister, Tube,
Water or Soil Sample Types
RGl
FIG 2
Control of P&T by Entech 2000
End
Yes
P&T purges sample and waits
for "Focusing Trap Ready"/
0nstead of oven ready)
Entech 2000 gets "Oven Ready"
and cools focusing trap-
Entech sends "Desorb Start"
to start focusing event,
focusing Trap desorbs and
GC/MS Acquisition is started*
I
798
-------
Al\0«Q1003.D
U.
13 Apr M 4<0« fm \
HI 9tM
tJ10<03-«09 ** 1.1994
V9430C1.N
FIGURE 3 - Soil sample
spiked with a methanol
solution containing
chloromethane (1) and
vinylchloride (2).
Peak widths were
comparible to those
later in the analysis.
mmum
500000
450OOQ
400000
300000
250000
150000 •
100000^
50000 •
V A
' V
1
u
TIC. 010100J.D
Z.
1
1
fl
A /I A
\ rJ \ *
\JV^\J\] \ /
"
L -
FIGURE 4-Diesel
contaminated soil
sample containing a
large abundance of
vinylchloride (1).
A:\11010O7.D
U.
33 Apr 13 i:05 pm tuing AcqitaCbad v«4aOel.l
XS 9*71
»10303-001 1.H7 e/lDKL/90TIt
1300000
1100000
1000000
900000
100000
700000
40000O
300000
300000
100000
. /v/J
799
-------
AIR EMISSION RATE MEASUREMENTS OF SURFACE IMPOUNDMENT
QUIESCENT WATER AND SLUDGE SURFACES
William A. Butler, Project Engineer
Du Pont Environmental Remediation Services, Inc.
300 Bellevue Parkway, Suite 390
Wilmington, Delaware 19809-3722
INTRODUCTION
Two 17-acre surface impoundments located at a chemical manufacturing facility are required
to be closed through an administrative consent order with a state environmental regulatory
agency. The impoundments contain approximately 400,000 cubic yards of a 30 percent
solids sludge. The sludge is contaminated with a variety of volatile and semivolatile organic
compounds (VOCs collectively). Chlorobenzene and 1,2-dichlorobenzene are the
predominant organic compounds present at average concentrations of 905 and 1,696 mg/kg
(dry weight), respectively. A variety of heavy metals are also present in the sludge with lead
being predominant at an average concentration of 12,716 mg/kg (dry weight).
Closure will consist of in-situ solidification followed by consolidation into a vault that will be
constructed within the footprint of one of the impoundments. To facilitate the solidification
process, the surface water will be completely removed from the impoundment. This could
potentially result in a release of fugitive air emissions of VOCs from the sludge surface.
State air quality regulations require control of these fugitive emissions if the emission rate
of total VOCs exceed 0.5 Ib/hr. To address the regulatory requirement, a study was
developed to estimate the air emission rate of VOCs from the surface impoundment sludge
surfaces, and determine how the air emission rate from the sludge surface varies with time
after removing the surface water. The results were used to determine whether an emission
control technology would be required to meet the regulatory requirement. The study also
included estimating the emission rate of VOCs from the quiescent water surface for
comparison purposes. This paper discusses the procedures, results, conclusions, and
recommendations from the study.
PROCEDURES
The air emission rate measurements were completed with the use of three emission isolation
flux chambers (flux chambers). A flux chamber is an enclosure device which enables the
user to make direct emission rate measurements of VOCs that are released from a defined
water or land surface area. To measure the emission rate for this study, the flux chamber
was placed over the desired location, and an ultra high purity air (<0.1 ppm total
hydrocarbons) carrier gas was introduced into the flux chamber at a constant flow rate of
5 L/min. The carrier gas flow into the flux chamber enters at several points to facilitate
complete mixing with the emissions from the defined surface area. The carrier gas flow
creates a slight positive pressure within the flux chamber which prevents external air from
entering the flux chamber and possibly contaminating or diluting the exhaust gas. However,
the slight positive pressure is not enough to prevent the emissions from entering the flux
chamber, because the pressure is continuously released through a 3/4-inch diameter vent on
top of the flux chamber. The flux chamber design and operating procedures were obtained
800
-------
from the Measurement of Gaseous Emissions from Land Surfaces Using an Emission Isolation
Flux Chamber - User's Guide (EPA 600 8-86-008).
Samples were collected from the sample outlet, and analyzed for individual VOCs utilizing
EPA Ambient Air Method TO-14. The Summa® canisters used for sample collection had
regulators to limit the sample collection rate to 1 L/min. A portable organic vapor analyzer
was utilized to monitor total organic vapor concentrations and evacuate the sample outlet
prior to sample collection. Samples were collected after five residence times (30 minutes)
to ensure that the carrier gas and emissions were completely mixed. With the analytical
results, the emission rate of individual VOCs were then calculated from the following
equation:
Ej = emission rate of component i (//g/m2-min)
C, = concentration of component i in the sample Gt/g/m3)
Q = total gas flow rate (carrier gas and air emission) (5E-03 m3/min)
A = surface area enclosed by the chamber (0.13 m2)
Emission rate measurements were conducted at three locations within the surface
impoundment. A 10-ft long, 24-in diameter steel caisson was placed at each location to
facilitate the measurements. Caissons A and B were located in areas containing average
concentrations of VOCs, and Caisson C was located in a area known to contain higher
concentration of VOCs. To complete the measurements at each location, the 1.5-feet of
surface water was removed from within the caisson, and the flux chamber was installed in
the caisson on top of the sludge surface. Figure 1 depicts the caisson with a flux chamber
installed. Measurements of the sludge surface were completed initially and after 4, 8, 24,
and 48 hours of exposure. A quiescent water surface measurement was also completed at
each location by floating the flux chamber on the water with the use of a 16-inch inner tube.
RESULTS
Table 1 presents the calculated emission rates in Ib/sqft-hr of total VOCs for each caisson
throughout the 48 hour time period. The results for each caisson are discussed in the
following sections.
Caisson A
For Caisson A, the emission rate of total VOCs ranged from 5.20E-07 Ib/sqft-hr initially to
1.63E-07 Ib/sqft-hr at 48 hours. The lowest emission rate of total VOCs was 1.49E-07
Ib/sqft-hr which occurred at 8 hours. Initially, Freon® 113 was emitted at the highest rate
of 3.49E-07 Ib/sqft-hr (67.1 percent of the total). FREON is not defined as a VOC by the
state's regulations due to its non-photoreactive properties. However, it is included in the
total VOC emission rates discussed in this paper since it appears to be the predominant
compound emitted from the sludge surface. FREON was not analyzed for in sludge sample
analyses, because it was not included as on the standard list of analytes. Toluene and
chlorobenzene were emitted at the next highest rates of 6.42E-08 and 5.23E-08 Ib/sqft-hr,
respectively. At 48 hours, FREON 113 was emitted at the highest rate of 1.12E-07 Ib/sqft-
hr (68.7 percent of the total). 1,2,4-trichlorobenzene and 1,2-dichlorobenzene were emitted
at the next highest rates of 2.49E-08 and 7.11E-09 Ib/sqft-hr, respectively. No significant
changes in the emission rate were observed after 8 hours of exposure.
801
-------
Caisson B
For Caisson B, the emission rate of total VOCs ranged from 7.12E-07 Ib/sqft-hr initially to
1.02E-07 Ib/sqft-hr at 48 hours. The lowest emission rate of total VOCs was 6.25E-08
Ib/sqft-hr which occurred at 24 hours. Initially, FREON 113 was emitted at the highest rate
of 6.84E-07 Ib/sqft-hr (96.1 percent of the total). Chlorobenzene and FREON 114 were
emitted at the next highest rates of 9.58E-09 and 6.50E-09 Ib/sqft-hr, respectively. At 48
hours, FREON 113 was emitted at the highest rate of 9.00E-08 Ib/sqft-hr (88.2 percent of
the total). FREON 114 and chlorobenzene were emitted at the next highest rates of 3.10E-
09 and 2.72E-09 Ib/sqft-hr, respectively. No significant changes in the emission rate were
observed after 8 hours of exposure.
Caisson C
Caisson C had the highest emission rate of total VOCs compared to the other caissons. This
was expected since the sludge in that area of the surface impoundment contains higher
concentrations of VOCs. The emission rate of total VOCs ranged from 2.36E-06 Ib/sqft-hr
initially to 4.94E-07 Ib/sqft-hr at 48 hours. The lowest emission rate of total VOCs was
4.53E-07 Ib/sqft-hr which occurred at 24 hours. Initially, 1,2,4-trichlorobenzene was
emitted at the highest rate of 8.78E-07 Ib/sqft-hr (37.2 percent of the total). Freon® 113
and 1,2-dichlorobenzene were emitted at the next highest rates of 6.84E-07 and 6.26E-07
Ib/sqft-hr, respectively. At 48 hours, 1,2,4-trichlorobenzene was emitted at the highest rate
of 2.35E-07 Ib/sqft-hr (47.6 percent of the total). 1,2-dichlorobenzene and FREON 113
were emitted at the next highest rates of 1.34E-07 and 8.28E-08 Ib/sqft-hr, respectively.
No significant changes in the emission rate were observed after 8 hours of exposure.
Quiescent Water Surface
Table 1 also presents the calculated emission rates in Ibs/sqft-hr of total VOCs from the
quiescent water surface adjacent to Caissons B and C. Data is not available for the
quiescent water surface adjacent to Caisson A due to the SUMMA canister leaking during
transport to the laboratory. The total VOC emission rate for Caisson B was 1.20E-07 Ib/sqft-
hr with FREON 113 being emitted at the highest rate of 8.64E-08 Ib/sqft-hr (72.0 percent
of the total). 1,2-dichlorobenzene and chlorobenzene were the only other compounds
emitted at rates of 2.02E-08 and 1.35E-08 Ib/sqft-hr, respectively. The total VOC emission
rate for Caisson C was 5.09E-08 Ib/sqft-hr with FREON 113 being emitted at the highest
rate of 1.98E-08 Ib/sqft-hr (38.9 percent of the total). 1,2-dichlorobenzene and
chlorobenzene were emitted at the next highest rates of 1.93E-08 and 9.15E-09 Ib/sqft-hr,
respectively.
Overall Air Emission Rate
Table 1 also presents the overall total VOC emission rate from the impoundment. This was
calculated by multiplying the average total VOC emission rate of all the caissons combined
by the surface area of the surface impoundment (17.4 acres = 757,944 sqft). The air
emission rate ranges from 0.91 Ib/hr initially to 0.19 Ib/hr at 48 hours. At 4 hours, there is
a 69.2 percent reduction of the total VOC emission rate. At 8 hours, there is a 76.9 percent
reduction. No significant changes in the emission rates occur after 8 hours of exposure.
The quiescent surface emission rate is 0.06 Ib/hr, which indicates that the surface water is
effectively suppressing the emission rate of VOCs from the sludge surface. Figure 2 is a
graphical representation of the overall emission rate of total VOCs versus time.
802
-------
CONCLUSIONS
The following conclusions can be drawn from the results of this study:
• The emission rate of total VOCs from Caisson C was higher than the emission
rates from the other caissons: This was expected since the sludge in the area
of Caisson C contains higher concentrations of VOCs.
• The quiescent water surface emission rates are less than the lowest sludge
surface emission rates, which indicates that the surface water does effectively
suppress the emission rate of total VOCs from the sludge surface.
• A 69.2 percent reduction of total VOC emissions occurs at 4 hours, and this
corresponds to 87.5 percent of the total reduction from the initial to the 48
hour measurement.
• No significant change in total VOC emission rates occurs after 8 hours of
exposure.
• FREON 113 is the predominant compound emitted from the sludge surface, but
is not defined as a VOC by the state regulations due to its non-photoreactive
properties.
• The initial and highest emission rate of total VOCs including FREON 113 from
the sludge surface is 0.91 Ib/hr.
• Based on these results of this study, emissions control will not be required
during or following the removal of surface water from the impoundment,
because the maximum emission rate of total VOCs excluding FREON 113 does
not exceed the state's regulatory requirement of 0.5 Ib/hr.
RECOMMENDATIONS
The following recommendations are made based on the results of this study:
• A dispersion model should be completed with this data to determine what the
concentrations of VOCs will be at the site boundaries, and whether these
concentrations will affect human health and/or cause odor problems.
• It is recommended that the impoundment be drained slowly in order to minimize
the amount of surface area being exposed at one time; therefore, reducing the
overall initial emission rates from the sludge surface. Table 2 presents the
overall air emission rate of total VOCs from the sludge surface after the water
level is decreased a 0.5 foot per day until no more water is present. The table
shows that even with FREON 113 included, the overall emission rate of total
VOCs would be less than the state's regulatory requirement of 0.5 Ib/hr.
803
-------
FIGURE 1
Field Apparatus Setup
Carrier Gas biia
TABLE1
Calculated Air Emission Elates for Total VOCs
Measurement
Quiescent Surface
Initial
4 Hours
8 Hours
24 Hours
48 Hours
Caisson A
Total VOC
(Ib/sqft-hour)
NA
5.20E-07
Z95E-07
1.49E-07
2.44E-07
1.63E-07
Caisson.B
Total VOC
(Ib/sqft-hour)
1.20E-07
7.12E-07
1.4SE-07
1.02E-07
6.25E-08
1.02E-07
Caisson C
Total VOC
(Ib/soft— hour)
5.09E-08
2.36E-06
6.83E-07
5.64E-07
4.80E-07
4.94E-07
Average
Total VOC
(Ib/soft-hour)
8.5SE-08
1.20E-06
3.74E-07
2.72E-07
2.62E-07
Z53E-07
Overall
Emission Rate
(Ibita)
0.06
0.91
0.28
0.21
0.20
0.19
Notes:
<1)-NA(h
(2J - The ovwal mnlsskin rate ts eafculattd by multiplying the average emission rate by tw surface Impoundment* surface area of 757,944 soJL
804
-------
FIGURE 2
Overall Total VOCAir Emission Rate versus Time
Time (hours)
TABLE 2
Air Emission Rates of Total VOCs
Resulting from Removing the Surface Water
Water Level -i
;(feet) .I"?.:.
4.50
4.00
3.50
3.00
2.50
ZOO
1.50
1.5 (after 24 hours;
Quiescent Surface
''TotalVbG
(ib/hr)
0.06
(757,944 sqft)
0.05
(629,778 sqft)
0.04
(488,498 sqft)
0.03
(303,1 23 sqft)
0.01
(1 57,376 sqft)
0.01
(78,001 sqft)
0.00
(Osqft)
0.00
(0 sqft)
New Sludge Surface
Total VOC
(Ib/hr)
0.00
(Osqft)
0.15
(128, 166 sqft)
0.16
(141 ,280 sqft)
0.22
(185,375 sqft)
0.17
(145,747 sqft)
0.10
(79,375 sqft)
0.09
(78,001 sqft)
0.00
(0 sqft)
Sludge Surface
total VOC
(Ib/hr)
0.00
(Osqft)
0.00
(Osqft)
0.03
(128, 166 sqft)
0.07
(269,446 sqft)
0.12
(454,821 sqft)
0.15
(600,568 sqft)
0.18
(679,942 sqft)
0.20
(757,944 sqft)
Overall
Total VOC
(Ib/hr)
0.06
0.20
0.23
0.32
0.30
0.26
0.27
0.20
Notes:
(1) - 8-S5E-08 tb/sqft-hr was the emission rate used for the quiescent surface (average quiescent surface emission rate).
(2) - 1.20E-Q6 to/sqfl-hr was the emission rate used (or the new sludge surface (average initial emission rate).
(3) -2£8E-07 Ib/sqft-hr was the emission rate used for the sludge surface (average of the 24 and 4B hour average emission rate).
(4) - The surface area for each specflc surface b located below the corresponding emission rate.
805
-------
Characterization of Air Emissions From the
Simulated Open Combustion of Fiberglass Materials
Christopher C. Lutes and Jeffrey V. Ryan
Acurex Environmental Corporation
4915 Prospectus Drive
P.O. Box 13109
Research Triangle Park, NC 27709
Paul M. Lemieux
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
The exposure of persons to fiberglass combustion emissions in structure fires, fires at waste landfills,
and fires at demolition sites has become an issue of increasing concern. This study identifies and quanlitatcs a broad
range of pollutants that are discharged during the small-scale, simulated, open combustion of fiberglass and reports these
emissions per mass of fiberglass material combusted. Two types of fiberglass materials (representing the boating and
building materials industries) were combusted in a controlled outbuilding designed for the simulation of open burning
processes. Volatile, semivolatile, and particulate-bound organics were collected and analyzed by gas
chromatography/mass spectrometry (GC/MS). The emphasis of these analyses was on the quantification of hazardous air
pollutants listed in Title III of the Clean Air Act Amendments of 1990 (CAAAs), although further efforts were made to
identify and quantify other major organic components. Additional sampling and analysis were done for particulate-pnase
metals, respirablc fibers, and hydrogen chloride. Fixed combustion gases (carbon dioxide, carbon monoxide, nitrogen
oxide, oxygen and total hydrocarbons) were monitored continuously throughout the test period. Analytical results show
substantial emissions of a large number of pollutants including carbon monoxide, paniculate, lead, arsenic, benzene,
toluene, styrene, naphthalene, phenol, dibenzofuran, phenanthrene, and benzo(a)pyrene.
INTRODUCTION
Concerns over exposure to air emissions from the open burning of fiberglass in structure fires, and at waste
disposal and demolition sites have been expressed to the Control Technology Center (CTC) of the U.S. Environmental
Protection Agency (EPA). Though little previous research has been done specifically on combustion emissions from
fiberglass, literature does exist relating to the composition of fiberglass, the combustion products of some components of
commercial fiberglass materials, and the suspected health effects of fiberglass fibers. Fiberglass is principally composed
of SiOj (approximately 50% by weight); additional major components are A12O3, BajC^, CaO, and MgO (typically 3 to
20% each) and trace components include F, Fi^O,, K2O, Na2O, SO3 and TiO2 (less then 1% by weight each1).
Additionally, fiberglass materials may contain organics as sizings or binders.1 Organic containing fiberglass materials can
be classified as either epoxy based or polyester based.2 The known combustion products of polyester-based materials
have been reviewed and are known to include acetaldehyde, benzene, biphenyl, carbon monoxide, ethyl benzene,
pentadiene, styrene, and toluene.3"6 Much of the available information on these combustion products has been obtained in
small-scale studies of materials that are likely to be less complex than the commercial materials found in practice.
Therefore, larger scale tests of complex commercial materials promise increased insight
In addition to potential hazards of a chemical nature, the air emissions from open fiberglass combustion may
include fibrous aerosols; the physical nature of fibrous aerosols may lead to additional health hazards. Epidemiological
studies have shown significant increases in non-malignant respiratory disease in populations exposed to glass fibers/
Glass fibers are apparently less harmful to health than asbestos fibers.8 The greatest hazard appears to be related lo fibers
with a diameter less then 1.5 urn and a length greater then 8 urn.8 No measurements of fiberglass fiber emissions from
combustion processes have been found in our literature review to date. An assessment of the concentration and size
distribution of fibrous aerosols produced from fiberglass open combustion processes would be valuable.
In response, through the guidance of EPA's Air and Energy Engineering Research Laboratory (AEERL), a study
was undertaken to measure emissions from the combustion of fiberglass samples from two industries that use fiberglass
extensively. This study included replicate tests of fiberglass materials from the boating industry (polyester based, some
with and some without a gel coating (an epoxy based sealing material)), and the building industry (vinylester based). The
study was designed to collect, identify, and quantify a wide range of air emissions and to report these emissions per mass
806
-------
of fiberglass material combusted. The emphasis of these analyses was on die quantification of air toxics compounds
listed in the Clean Air Act Amendments of 1990 (CAAAs), although further efforts were made to identify and
semiquantify other major organic components. Because of the complexity of this study, this paper can only highlight the
methods used and the results obtained. The reader is urged to consult the forthcoming EPA report on this project for
more complete information.
METHODS
Combustion testing for this study occurred in EPA's Open Burning Simulation facility, which has been
previously described.9 The boating industry fiberglass samples were placed in the facility and ignited using a brief
application of a handheld propane torch which was removed before sampling began. A "hut blank" test in which the
propane torch was briefly introduced into the facility but no fiberglass was combusted was conducted for comparison
purposes. In order to allow an adequate time period for all necessary samples to be obtained, three separate charges of
fiberglass were combusted during each test Combustion of one charge was allowed to go to apparent completion (as
signified by constant weight and near background concentrations of combustion gases) before another charge was
introduced. Attempts to test the building industry fiberglass sample in a~ like manner were unsuccessful because of the
high concentration of flame retardant in this sample. Therefore, the combustion of the building industry material was
supported by a continuous liquid propane (LP) gas flame during sampling. This study design was intended to simulate
the behavior of this flame retarded fiberglass material in the presence of other non-flame retarded combustibles. A
"combustion blank" test in which the LP flame was operated but no fiberglass was present was conducted for comparison
purposes. In addition, various field and laboratory blank samples were collected for each sampling train, as appropriate.
In order to allow an adequate time period for all necessary samples to be obtained, two separate charges of fiberglass
were combusted during each test. Combustion of one charge was allowed to go to apparent completion, as was done for
the boating industry samples before another charge was introduced.
An elemental analysis of the fiberglass samples was performed before combustion using methodologies best
summarized by ASTM methods 3176 and 3172.10 Fixed combustion gases [carbon dioxide (CO,), carbon monoxide
(CO), nitric oxide (NO), oxygen (O^, and total hydrocarbons (THC)] were monitored continuously throughout the test
period through the sampling manifold. Temperatures at relevant locations in and around the test facility and the mass of
fiberglass material were monitored throughout the test period.
Volatile organics were collected from the sample manifold on sorbent tubes (VOST train) and analyzed by gas
chromatography/mass spectrometry (GC/MS).11'12 Since extremely high levels of volatile organic compounds were
observed during early tests of the building industry fiberglass, an additional volatile sampling method was implemented.
Samples were collected in Tedlar* bags and analyzed by GC/MS in accordance with EPA Method 18.13 Samples for
semivolatile and particulate-bound organics and metal aerosol analysis were collected through separate medium volume
PM10 samplers located in the burn hut14 The sample for metal aerosol was collected on a 142 mm diameter quartz fiber
filter and analyzed by inductively coupled plasma-atomic emission spectrometry in accordance with an EPA contract
laboratory program method (similar to EPA method 2Q0.715). The semivolatile and paniculate phase organic sample was
collected with a 142 mm diameter. Teflon impregnated, glass fiber filter and XAD-2 resin sorbent. The filter and resin
were then extracted in methylene chloride, and the pooled extract was analyzed in accordance with EPA Method 8270.16
A realtime photoelectric analyzer designed to quantify total polycyclic aromatic hydrocarbons (PAHs) on submicron
paniculate was also operated using a sample stream withdrawn through the sampling manifold.17'18 A sample probe for
vapor- phase hydrogen chloride was also located in the bum hut These samples were analyzed by ion chromatography.
This sampling was done according to a modified form of EPA Method 26.13
The sampling train used to sample for fiber size and morphology analysis consisted of a 37-mm mixed cellulose
ester filler cassette followed by a low volume sampling pump and dry gas meter. The filter was operated in an inverted
position, parallel to the facility floor during sampling to minimize the collection of paniculate matter through gravitational
settling. Analysis was performed by phase contrast light microscopy (PCM) and transmission electron microscopy
(TEM). For the purposes of these analyses, a fiber was defined as a particle with an aspect ratio of greater than 3:1.
These sampling and analysis methodologies were based on NIOSH methods for asbestos fibers.19
After the completion of the chemical and microscopic analyses, analyte concentration data were coupled with
sample volume, facility air flow, and combustible material mass loss data to derive estimated emissions (expressed as
mass of analyte produced per mass of fiberglass material consumed in the combustion process).
RESULTS AND DISCUSSION
The elemental analysis of the fiberglass materials before combustion (Table 1) indicates that the organic matter
content of the boating industry fiberglass is higher than the building industry material. The substantial halogen
concentration found in the building industry material tends to confirm the manufacturer's statement that the material
contained a brominated fire relardant. The vast majority of the combustion of each charge of boating industry material
was completed in a 20-40 minute time period while the majority of the building industry material in each charge appeared
807
-------
to be consumed in 30-60 minutes. Table 2 summarizes the estimated emissions derived from realtime measurements of
CO, COj, THC, and total PAH bound to submicion paniculate. The substantial observed CO emissions, which are
probably underestimated in this data set, are a concern since CO is believed to be the "primary cause of death of most
fire victims."4 Substantial emissions of paniculate matter were also observed (average values were 117 g/kg for the
boating industry material and 607 g/kg for the building industry material). This is a concern since a majority of previous
studies of combustion products of various polymers have paid little attention to the composition of the paniculate phase.5
The volatile organic data set includes concentration measurements for 35 targeted (the majority of which are
consistently non-detectable) and several dozen tentatively identified species; Table 3 presents average data on several
selected compounds that were among those seen in the highest levels. It appears that the relative ratios of these
components are similar for both the boating and building industry materials but that the absolute emission rate is higher
for the building industry material.
Table 1. Composition of Fiberglass Materials Tested (ail data as % composition).
Carbon
Hydrogen
Oxygen
Nitrogen
Sulfur
Total Halogen (as Cl)
Aluminum
Magnesium
Cadmium
Chromium
Moisture
Volatile Matter
Ash
Fixed Carbon
Boating Industry
Fiberglass -
Without Gel Coat
52.97
4.79
7.1
0.017
<0.03
<0.5
0.28
<0.05
<0.004
0.0091
2.06
63.11
34.83
<0.1
Boating Industry
Fiberglass -
With Gel Coat
55.06
50.1
<0.5
0.015
<0.03
<0.5
0.081
<0.05
<0.004
0.041
1.19
56.24
39.57
3
Building
Industry
Fiberglass
25.33
2.48
10.06
<0.5
<0.5
1.9
2.2
0.12
<0.01
0.086
0.52
35.29
60.23
3.96
-------
Table 2. Combustion Gas and PAH Paniculate Analyzer Concentrations and Estimated Emissions
Date
09/08/92
09/17/92
10/13/92
10/30/92
11/04/92
Test
First Boating Industry Test
Second Boating Industry Test
First Building Industry Test
Second Building Industry
Test
Third Building Industry Test
COasC
Estimated
Emissions
(6*g)'
48.2
55.2
205.9
141.4
163.3
COjasC
Estimated
Emissions
(g*g)'
980.8
953.4
NC
NC
NC
THCas
Methane
Estimated
Emissions
(g*g)*
41.6
40.5
270.3
202.3
232.2
PAH
Estimated
Emissions
(g*g)
1.48
0.85
0.97
0.58
0.59
Key: NC = Not calculated (estimated emissions were not calculated for CO2 in the building industry test
since the LP burner contributed a sizeable and somewhat variable percentage of the emissions).
* = In cases where some measured concentrations exceeded the concentration of the high
calibration point before calculating the averages, the data were truncated to the high calibration
point
Table 3. Estimated Emissions of Selected Volatiles
Compound
Benzene
Ethyl Benzene
Styrene
Toluene
m-p-Xylene
Boating Industry VOST Samples,
Average Estimated Emissions (g/kg)
5.9
0.7
4.5
3.6
0.5
Building Industry Tedlar Bag Samples,
Average Estimated Emissions (g/kg)
34.8
9.3
49.4
18.0
1.1
The semivolatile and paniculate bound organics data set generated from this project includes concentration
measurements for more then 90 targeted (the majority of which were consistently non-detectable) and several dozen
tentatively identified species. Average emission values for a selected set of detected, targeted semivolatile, and paniculate
bound organics are presented in Table 4. Average estimated emissions for these compounds are generally lower then for
the volatile species discussed previously. As in previous measurements, the values obtained in the building industry
fiberglass tests are generally higher than those in the boating industry tests. Preliminary calculations have shown that the
estimated emissions calculated from the output of the real time PAH analyzer (Table 2) agree at least within a factor of
10 with the sum of estimated emissions calculated from the Method 8270 analyses of PAHs that would be expected to be
predominantly in the paniculate phase.
809
-------
Table 4. Estimated Emissions of Selected Semivolatiles
Compound
Anthracene
Benzo(a)pyrene
Biphenyl
Chrysene
2-Methyl Phenol
Phenanihrene
Phenol
Boating Industry Average (mg/kg)
353
86
689
323
125
902
328
Building Industry Average (mg/kg)
202
72
1,936
458
400
2.156
6,830
Fibrous aerosols samples rarely showed significantly more fibers than were seen in blank samples, as shown in
Table 5. However, detection limits were quite high for this analysis, since the maximum feasible loading of total
paniculate on these filters was reached after a very small volume (<20 L) was sampled and it is only feasible to conduct
the microscopic examination on a small representative portion of the filter surface area.
Table 5. Fibrous Aerosol Measurements
Sample
No.
7
8
11
16
17
19
Test
Date
09/17/92
09/17/92
09/17/92
10/13/92
10/13/92
10/30/92
Test
Second Boating Industry
Second Boating Industry
Second Boating Industry
First Building Industry
First Building Industry
Second Building
Industry
PCM Length >5
um Estimated
Emissions
(million S/kg)
1710
4846*
504*
1904*
432*
231*#
ATEM Length >0.5
<5.0 um Estimated
Emissions (million
S/kg)
490*#
1902*#
NFD*#
395*
89*#
120*#
ATEM Length
>5 um Estimated
Emissions (million
S/kg)
420*#
1902*#
NFD*#
395*
89*#
120*#
Key: NFD = No fibers detected, detection limit cannot be stated accurately due to loading problems
PCM = Phase contrast microscopy
ATEM = Analytical transmission electron microscopy
* = Number of observed fibers not greater then 3 times larger of the applicable field and hut blank
values
# = Air concentration of observed fibers not more then 3 times larger then the applicable hut or
combustion blank value
S = Fiberous Structures
The paniculate-phase metals samples were analyzed for 11 elements. Of these, only lead, silver, and possibly
cadmium were detected in the boating industry emissions and only arsenic and possibly chromium were detected in the
building industry emissions. These results are summarized in Table 6.
810
-------
Table 6. Metals Estimated Emissions (mg/kg)
Test
Fust Boating Industry
Second Boating Industry
Boating Industry Average
First Building Industry
Second Building Industry
Third Building Industry
Building Industry Average
Silver
4.41
4.60
4.50
13.38*
14.45*
6.76*#
11.53
Arsenic
0.52*#
0.6*#
0.56
6.13*
48.15
13.05
22.44
Cadmium
0.06*#
0.29
0.18
0.28*#
0.80*
0.24*#
0.44
Chromium
1.57*#
0.48*#
1.03
8.92
14.45*
5.80*#
9.72
Lead
21.00
38.72
29.86
2.79*#
8.03*
2.42*#
4.41
Key: * = Mass of this sample not greater than 3 times the largest of the following: mass in field blank, mass in
hut blank, or mass in combustion blank, as applicable.
# = Sample hut air concentration not greater than 3 times the hut blank air concentration or the combustion
blank air concentration, as applicable.
Vapor-phase hydrochloric acid was not detected in any of the samples analyzed. Detection limits varied from 40
to 260 mg/kg. No other acid gases were analyzed for.
Because of operational difficulties (i.e., sampling media overloading), the sampling periods for various trains
varied widely. Thus, it is not possible to compute an accurate mass balance on this system since the rates of emission of
various pollutants probably vary during various phases of the combustion process. In addition, since the rate of emissions
from a small mass of combusted fiberglass was high enough to threaten overloading of the sampling media, it was
necessary to sharply limit the amount of fiberglass combusted in each test phase. This may introduce a significant source
of error into the data set because the resolution of the balance used to measure the weight of fiberglass lost to combustion
was 0.2 Ib (0.09 kg), [the average weight losses for the samples were: organic semivolatile/particulate train 3.0 Ib (1.4
kg), metals train 3.8 Ib (1.7 kg), VOST train 0.9 Ib (0.4 kg), Tedlar* bag train 2.0 Ib (0.91 kg), CEM train 8.3 Ib (3.8
kg), fiber train 1 Ib (0.45 kg), hydrochloric acid train 7.2 Ib (3.3 kg)].
CONCLUSION
Despite the aforementioned experimental difficulties, this project did succeed in producing estimated emissions
data for a broad range of atmospheric pollutants from a simulated open fiberglass combustion process. Substantial
emissions of a large number of pollutants including carbon monoxide, paniculate, lead, arsenic, benzene, toluene, styrene,
naphthalene, phenol, dibenzofuran, phenanthrene, and benzo(a)pyrene were observed. The health implications of these
emissions in a given situation can be judged if emissions data such as those presented here are combined with fate and
transport, and health effects data to form a detailed risk assessment
REFERENCES
1. H.C.W. Skinner, M. Ross, and C. Frondel, Asbestos and Other Fibrous Materials. Oxford University Press- New York
1988, pp 82-85.
2. Personal communication from D. Preiss to J. Whitfield (EPA/Air and Energy Engineering Research Laboratory), June
1991.
3. D P. Miller, R.V. Petrella, and A. Manca, "An Evaluation of Some Factors Affecting the Smoke and Toxic Gas
Emission From Burning Unsaturated Polyester Resins." Presented at the 31st Annual Technical Conference of the
Reinforced Plastics/Composites Institute of The Society of the Plastics Industry. Inc.. 1976.
4. S.C. Gad and R.C. Anderson. Combustion Toxicology. CRC Press: Boca Raton, 1990, pp 66,155, 176-92.
5. B.C. Levin, "A Summary of the NBS Literature Reviews on the Chemical Nature and Toxicity of the Pyrolysis and
Combustion Products from Seven Plastics: Acrylonitrile-Butadiene-Styrenes (ABS), Nylons, Polyesters, Polyethylenes,
Polystyrenes, Poly(Vinyl Chlorides) and Rigid Polyurethane Foams," Fire and Materials: 11:143-57, 1987.
6. E. Braun and B.C. Levin, "Polyesters: A Review of the Literature on Products of Combustion and Toxicity," Fire and
Materials: 10:107-23,1986.
7. R.I. Mithchell, D J. Donofrio, and WJ. Moorman, "Chronic Inhalation Toxicity of Fibrous Glass in Rats and
Monkeys," JACT, 5:557-575, 1986.
811
-------
8. KP. Lee et at, "Comparative Pulmonary Responses to Inhaled Inorganic Fibers with Asbestos and Fiberglass,"
Environmental Research. 24:167-91, 1981.
9. J.V. Ryan and C.C. Lutes Characterization of Emissions from the Simulated Open-Burning of Non-Metallic
Automobile Shredder Residue. EPA-600/R-93-044 (NTIS PB93-172914), U.S. Environmental Protection Agency,
Research Triangle Park, March 1993.
10. American Society for Testing and Materials, Annual Book of ASTM Standards. 1990.
11. EM. Hansen. Protocol for the Collection and Analysis of Volatile POHCs Using VOST. EPA-60Q/8-84-007 (NTIS
PB84-170042), March 1984.
12. Methods 5040 and 8240 in Test Methods for Evaluating Solid Wastes. Vol. IB. Field Manual Physical/Chemical
Methods. EPA (SW-846 (NTIS PB88-239223), November 1986.
13. 40 - Code of Federal Regulations. Parts 53-60, Revised July 1, 1991, U.S. Government Printing Office, Washington
DC, 1991.
14. A.R. McFarland and C.A. Ortiz," A 10 urn Cutpoint Ambient Aerosol Sampling Inlet," Atmospheric Environment
16: 2959-2965,1982.
15. Method 200.7, in Methods for the Determination of Metals in Environmental Samples. EPA/600/4-91/010, June
1991.
16. Method 8270, in Test Methods for Evaluating Solid Wastes. Vol. IB. Field Manual Phvsical/Chemical Methods.
EPA, (SW-846 (NTIS PB 88-239223), November 1986.
17. E.D. Chikhliwala, J.W. Podlenske, E. Pfeiffer, and W. Seifert, "The Design, Implementation and Use of a Real-time
PAH Analyzer for Combustion Products," Paper presented at the 9th World Clean Air Congress & Exhibition. Montreal.
Canada. August 1992.
18. R. Niessner, "The Chemical Response of the Photoelectric Aerosol Sensor to Different Aerosol Systems," J. Aerosol
Sci. Vol 17, No. 4, pp 705-714, 1986.
19. P.M. EUer, Ed., NIOSH Manual of Analytical Methods. 3rd Ed.. National Institute for Occupational Safety and
Health, Cincinnati, Ohio, 1984.
812
-------
INITIAL IMPLEMENTATION OF THE PHOTOCHEMICAL ASSESSMENT
MONITORING STATIONS (PAMS) NETWORK
Geri Dorosz-Stargardt
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards (MD-14)
Research Triangle Park, N.C. 27711
ABSTRACT
Section 182(c)(1) of the 1990 Clean Air Act Amendments requires the
Administrator to promulgate rules for the enhanced monitoring of ozone, oxides of
nitrogen, and volatile organic compounds to obtain more comprehensive and
representative data on ozone pollution. Following promulgation on February 12,
1993, states with ozone nonattainment areas classified as serious, severe, and
extreme must adopt and implement a program to monitor for such pollutants.
The PAMS network is driven by new monitoring requirements for VOCs,
carbonyls, and meteorological measurements. For the first time. State and local
agencies will be required to report data quarterly to the U.S. EPA for non-criteria
pollutants. There are no reference or equivalent methods available to measure
VOCs nor are there National Ambient Air Quality Standards (NAAQS) for these
pollutants. Agencies will be learning to operate sampling equipment that is often
state-of-the-art and currently undergoing development. Routine sampling for these
pollutants has never been attempted on this scale and will therefore make
implementation of this network challenging, at the least.
813
-------
INTRODUCTION
On February 12, 1993, the Environmental Protection Agency published
amendments to the ambient air quality surveillance rules (40 CFR 58) to provide for
the enhanced monitoring of ozone and oxides of nitrogen and for the additional
monitoring of volatile organic compounds (including carbonyls)(58 FR 8452).
Affected agencies must submit a State Implementation Plan revision to provide for
the establishment and maintenance of a PAMS network within 9 months after this
date or 9 months after the date of redesignation or reclassification of any existing
ozone nonattainment area or the designation of a new area to serious, severe, or
extreme. Even more stringently, these areas must submit a PAMS network
description, including a schedule for implementation within 6 months of the above
mentioned criteria. The first site should be initiated by February of 1994 (within a
year of promulgation), and fully operational by June of 1994.
AFFECTED AGENCIES
There are twenty-two ozone non-attainment areas in the country that have
been designated as serious, severe, or extreme, and are currently affected by this
regulation. The State and local Air Pollution Control Agencies in these areas are
responsible for implementing this regulation by establishing specialized ambient air
monitoring stations. These stations will monitor for ozone, oxides of nitrogen,
meteorology, and most importantly for volatile organic compounds (VOCs) and
carbonyls. The reason that VOCs are so important is that never before have
agencies been required to report to the EPA not only VOCs, but speciated VOCs as
well.
Now, let's consider this from the perspective of a State or local agency. If
EPA were to tell them that they had to monitor for something totally new,
something for which there are no reference or equivalent methods with which to
obtain these measurements, something for which there is no National Ambient Air
Quality Standard (NAAQS), something that has never been measured routinely, is
barely out of the research phase, and they presently don't even have enough staff
to do the job at hand for the criteria pollutants, their reaction would probably be
less than enthusiastic.
EPA's response would then be to remind them that this is not to be taken
lightly and is required by the regulation. This might cause anger and opposition
until EPA advises them that due to the magnitude of the program, they are willing
to assist in the following ways: 1) Funding in the way of grant dollars. EPA
would provide 105 Grant Funds in proportion to the number of sites required in
each Region. The Regions would then disperse this money to the affected
agencies. 2) Training in the way of workshops. EPA has already conducted the
National PAMS Teleconference Workshop which was broadcasted to over 200
downlink sites across the nation. Topics ranged from Program Requirements and
Implementation to Sampling and Analysis to Data Uses and Quality Assurance.
814
-------
3) Technical support in the way of guidance and calibration and audit materials. A
Technical Assistance Document for the Sampling and Analysis of Ozone Precursors
has been distributed, as well as Guidance on the Development and Approval of
PAMS Network Plans, PAMS Program and Data Quality Objectives, and Network
Design and Siting. Calibration and retention time standards for VOCs will be
supplied from a centralized source, and proficiency samples will be disbursed
routinely for quality assurance purposes.
IMPLEMENTATION CONCERNS
Once the decision has been reached to proceed with implementation, a
variety of concerns, questions, and problems arise. The following concerns were
addressed at the PAMS Teleconference Workshop on April 27-30, 1993, which
was sponsored by the Office of Air Quality Planning and Standards and the
Atmospheric Research and Exposure Assessment Laboratory (AREAL).
1. When does the network need to be implemented ?
The network design plan is due in August 1993, and the first
site needs to be fully operational in June 1994. However, the
program was funded a year in advance by EPA 105 grant
money so that the first sites could be operational in June 1993,
and in fact are encouraged to be. Almost all of the 22 affected
areas will be performing sampling, although some on a reduced
frequency, this summer.
2. Has anyone out there done this before ?
There have been numerous research studies conducted in the
past few years by AREAL, the Lake Michigan Air Directors
Consortium (LADCO), and the California Air Resources Board
(CARS), to name a few. Approximately half of the affected
areas were doing PAMS sampling for VOCs on a reduced
frequency during the 1992 Air Toxics Initiative, which was a
program funded to establish volatile organic compound/toxics
and aldehyde monitoring sites in a select number of urban areas
in 1992. The establishment of these sites allowed the early
implementation of baseline monitoring to support the PAMS
program by collecting background data on VOC and aldehyde
concentrations in the atmosphere before gasoline reformulation
and other control programs take effect. They were later to be
supplemented with instruments to sample for pesticides, PAHs,
815
-------
PCBs, and metals.
3. How are samples for VOCs collected ?
Manual methods utilize canisters. Automated methods utilize
the continuous automated gas chromatographs. The automated
methods presently available differ in their abilities to cool the
column ovens by the use of liquid nitrogen or liquid .carbon
dioxide or the use of thermoelectric coolers.
4. How are samples for VOCs analyzed ?
Both methods utilize a gas chromatograph with a flame
ionization detector.
5. What kind of instrumentation is available for automated systems ?
The following vendors have demonstrated that their instruments
are capable of sampling and analyzing according to the
guidelines referenced in the regulation (58 FR 8452):
Perkin-Elmer Corporation
Graseby-Nutech Corporation
Chrompack
Varian Chromatography Systems
Entech Laboratory Automation
6. How is the data going to be used ?
The data can be used in the following ways: 1) As evaluation
tools for control strategies, cost effectiveness, and for
understanding the mechanisms of pollutant transport. 2) As a
baseline for model evaluation and to minimize model
adjustments and reliance on default settings. 3) In analyzing
emissions inventory issues and corroborating progress toward
attainment. 4) In the preparation of unadjusted and adjusted
pollutant trends reports. 5) To make attainment/nonattainment
decisions and to construct NAAQS maintenance plans. 6) To
evaluate population exposure to air toxics as well as criteria
pollutants.
7. How is the data going to be quality assured ?
AREAL will be supplying calibration and retention time
standards from a centralized source for consistency and
comparability of the data nationwide. The agencies will be
paying for these with their grant dollars. AREAL will also be
816
-------
providing materials with which to perform proficiency studies.
Since some programs are overlapping across States and EPA
Regions, these areas will be conducting programs to include
oversight, quality assurance plans, site visits, and audits to
provide for consistency and comparability from agency to
agency.
8. What if an agency wants to do something different ?
The regulation does provide for the submittal of alternative
plans in the areas of siting (number and arrangement),
methodology (sampling and analysis), monitoring season
(months with highest ozone), sampling frequency, and
meteorology (establishing wind directions for siting).
Once some of these initial concerns are addressed, an agency should get
started by submitting their procurement requests for instrumentation and then by
deciding where to locate their monitors. The information gathered to make these
decisions must be packaged and submitted to EPA for network approval. This is
due to the requirements of the regulation (58 FR 8452). The requirements for this
network are the same as for the NAMS network and therefore must be approved in
the same way, by the Technical Support Division Director. The plan is sent from
the States to the Regional Office who sends it to OAQPS. The PAMS network
review committee consisting of the core committee, a meteorologist, a modeler, an
emissions expert, and an AREAL representative, in conjunction with the Regional
Office involved will reach a consensus on the plan and then send their
recommendations on to management for final approval.
CONCLUSION
EPA is currently reviewing plans for the first site location. This location is to
be representative of the maximum precursor emissions of the non-attainment MSA.
Most agencies have already submitted their plans, however a few are still in the
developmental stages. If all goes well, we expect to have some data reported to
the Aerometric Information Retrieval System by January 1994 from the sites that
are operational in June of 1993. During each year of implementation, as another
site is added and data are analyzed from sites already in operation, many of the
problems, concerns, and obstacles dealing with the start of a new monitoring
network will be alleviated. The lessons learned in the initial stages of start-up,
although painful, will prove to be invaluable as the implementation of the network
continues.
817
-------
BIBLIOGRAPHY
N.J. Berg, et al.. Enhanced Ozone Monitoring Network Design and Siting Criteria
Guidance Document. EPA 450/4-91-033, U.S. Environmental Protection Agency,
Research Triangle Park, 1991.
Code of Federal Regulations. Title 40, Part 58, U.S. Government Printing Office,
1992.
Shao-Hang Chu, "Using Windrose Data to Site Monitors of Ozone and Its
Precursors", U.S. Environmental Protection Agency, Research Triangle Park, Draft,
1992.
Federal Register (57 FR 7687), "Ambient Air Quality Surveillance - Proposed Rule",
March 4, 1992.
Federal Register (58 FR 8452) "Ambient Air Quality Surveillance - Final Rule",
February 12, 1993.
W.F. Hunt, Jr. and N.O. Gerald, "The Enhanced Ozone Monitoring Network
Required by the New Clean Air Act Amendments", 91-160,3, Air and Waste
Management Association, Vancouver, 1991.
M.E. Kantz, G.J. Dorosz-Stargardt, and N.O. Gerald, Photochemical Assessment
Monitoring Stations: Program and Data Quality Objectives. U.S. Environmental
Protection Agency, Research Triangle Park, Draft, 1993.
L.J. Purdue, D.P. Dayton, J. Rice, and J. Bursey, Technical Assistance Document
for Sampling and Analysis of Ozone Precursors. EPA 600/8-91-215, U.S.
Environmental Protection Agency, Research Triangle Park, 1991.
National Photochemical Assessment Monitoring Stations Teleconference Workshop.
April 27-29. 1993. Sponsored by Office of Air Quality Planning and Standards and
Atmospheric Research and Exposure Assessment Laboratory, U.S. Environmental
Protection Agency, Research Triangle Park, 1993.
818
-------
Meteorological Considerations in Siting Monitors of Photochemical Pollutants
Shao-Hang Chu
US Environmental Protection Agency
Office of Air Quality Planning and Standards
Technical Support Division
Research Triangle Park, NC 27711
ABSTRACT
Ozone conducive meteorological conditions are identified based on statistics derived from
local meteorological data of 31 eastern US cities in a 10-year (1981-1990) period. A method of
using ozone conducive meteorological conditions to construct a windrose to site photochemical
pollutant monitors is presented. The derived ozone conducive criteria appear to be quite robust,
suggesting that the method may be applicable to site photochemical pollutant monitors in other
areas of the United States.
INTRODUCTION
Ozone is not directly emitted into the atmosphere but is formed photocheraically from two
major primary pollutants, volatile organic compounds (VOC) and nitrogen oxides (NOx). The
accumulation of photochemically produced ozone, however, depends heavily on the dispersion
and transport of pollutants by the wind field. Thus, ambient ozone concentrations can reach high
levels only when the meteorological conditions are conducive both photochemically and
dispersively. For this reason, meteorological conditions observed on high ozone days are often
different from those observed on low ozone days. The ozone conducive meteorological
conditions are high insolation, high temperature, high stability, low winds and low relative
humidity. These ozone conducive conditions can only exist in special weather systems, such as
stagnant high pressure system and quasi-stationary front or trough1'2. Therefore, applying the
conventional method of using the (yearly or seasonly averaged) predominant wind direction(s)
(PWD) as the sole guide for siting ozone and its precursors monitors3 may not be adequate. In
this paper, a modified method is proposed and some general principles and suggestions to apply
this method are discussed.
OZONE MONITORING
A. Concept
Since meteorological conditions are usually different on high ozone days, only the PWDs
deduced from windrose on high ozone days are relevant for ozone monitor siting. For cities
with adequate coverage of ozone monitors, the PWDs for ozone can be deduced directly from
ozone pollution roses. However, for cities without adequate coverage of monitors, the PWDs
for ozone will have to be determined indirectly using ozone conducive meteorological conditions.
The approach suggested here is to
1. use the existing data base to establish criteria for "ozone conducive" days.
819
-------
2. use the identified criteria to extract all "ozone conducive" days for the last 10 years
to derive a windrose for potential high ozone days.
3. use the resulting windrose to select the PWDs for ozone monitor siting.
B. Criteria for Selecting "ozone conducive" days
A review of 31 US cities (see Table 1) east of the 99 W for 10 warm seasons (May-
October) from 1981 to 1990 reveals that three variables have distinct relationship with ozone
concentrations: (1) daily maximum 1-hour temperature, (2) morning (7-10 a.m.) and afternoon
(1-4 a.m.) winds and (3) mid-day (10 a.m.-4 p.m.) relative humidity (RH). Statistics show that
high ozone concentrations are generally associated with high temperatures, low to moderate
winds and low mid-day RH. There are significant geographic differences, too. For example,
in the North, ozone production is most sensitive to high temperatures, while in the South, low
winds and low RH are much more crucial for high ozone.
To select "ozone conducive" days from local climatological data, the values of critical
meteorological variables are estimated from the data base with conditional probability, P,
satisfying
P(y=>y0|x>c) = .95 or .05
Where y is a meteorological variable, y0, the 5th or the 95th percentile value of y, x, a daily
maximum 1-hour ozone concentration variable, and c, a threshold value selected for x. Here,
c is selected to be the National Ambient Air Quality Standard (NAAQS) for ozone, 0.12 ppm.
For example, given maximum 1-hour ozone concentration to be greater than .12 ppm, the
probability for daily maximum temperature to be greater than 80° F is 95 percent.
Following criteria are recommended as a general guide:
1. daily maximum temperature, T > = 80° F,
2. morning 7-10 a.m. wind speed Wra < = 10 kts,
3. afternoon 1-4 p.m. wind speed W^ < = 14 kts, and
4. mid-day (10 a.m. - 4 p.m.) relative humidity, RH < = 68%.
These criteria represent, to a great extent, the necessary conditions for daily maximum
1-hour ozone concentrations to exceed the NAAQS as reflected in the 10-year, 31-city data base.
While these are conditions necessary for high ozone they are not necessarily sufficient. Other
factors such as mixing height, opaque sky cover and topographically induced circulations, etc.,
may increase characterization of the ozone conducive conditions. However, since the intent of
the study is to identify ozone conducive days rather than to predict actual ozone concentrations,
this approach is considered to be adequate. Further, as noted above, T, Wm and W,,,, RH are
the most important as well as most accessible variables, and should be considered first.
Since these are general criteria common to all 31 cities in the eastern U.S., local
conditions could be a little different and are likely to be more stringent. Refinements to criteria
1-4 above should be made to best reflect local characteristics. For example, to account for
climatological differences, higher temperatures should be used hi the South. This does not mean
820
-------
that to produce the same amount of ozone, higher temperature is required in the South than in
the North. It is simply because under the same ozone conducive weather system, the temperature
rises higher in the South than in the North due to latitudinal effect. Also, away from the
transport dominant regions (e.g., the Northeast and the Midwest), the criteria for wind speeds
should most likely be lowered to account for the increasing proportion of local ozone production
under calm and light wind conditions. Further, due to the high temperatures in the South and
plentiful moisture supplied from Gulf of Mexico, the suppression or delaying of the afternoon
moist convective activities is crucial for ozone to reach peak values in the Gulf regions. Thus,
a relative lower RH is needed for the Gulf regions. As a general guide, the following refined
criteria could be applied to broad areas:
la. T > = 80° F for cities north of 40°,
T > = 83° F for most cities in the continental US,
T > = 85° F for cities south of 30°.
2a. Wra < = 10 kts for cities in transport regions,
W.J, < = 8 kts for cities outside transport regions.
3a. WpB1 < = 14 kts for cities in transport regions,
W,™ < = 12 kts for cities outside transport regions.
4a. RH < = 65% for cities north of 30°,
RH< = 60% for cities south of 30°.
C. Constructing Windroses and Siting Monitors
As a starting point, select all days in the last 10 years that satisfy the refined criteria
la, 2a and 4a as a basis for constructing a morning windrose for ozone and subsequent
determination of the morning PWD. Select all days satisfying the refined conditions la, 3a and
4a to construct the afternoon windrose for ozone and deduce the afternoon PWD. Over simple
terrain areas these two windroses and PWDs should be very similar. However, over complex
terrain and coastal areas these two windroses and PWDs could be quite different.
Figure 1 shows examples of windroses derived from ozone conducive meteorological
conditions and those derived from ozone concentrations greater than 100 ppb in the 31 cities.
From this figure, it is obvious that the PWDs selected from meteorologically conducive criteria
are essentially the same as those selected from high ozone days. To further check the robustness
of these criteria, the model has been applied to 10 Western cities outside the study domain. An
example is given in Figure 2. The similarity of the two wind roses shown in Figure 2 suggests
that the model works just as well in the West as in the East. Thus, it is clear that the ozone
conducive critera derived from the 10-year, 31 cities meteorological data are quite robust and
may be applied to other cities of the U.S..
Figure 3 demonstrates how the PWDs can be used in siting the monitors. The morning
PWD is suggested to be used as a guide to site the upwind background monitor (Ml) and to site
the maximum precursor emissions monitor (M2). The afternoon PWD is suggested to be used
to site ozone monitor (M3) at the downwind fringe of MSA/CMSA and to site ozone monitor
(M4) where maximum 1-hour concentrations are likely to be observed. Typically, Ml will be
located along the morning PWD 10 to 30 miles upwind from the city limit and M2 will be
located along the morning PWD at the downwind edge of the CBD. M3 should be located along
821
-------
the afternoon PWD at the downwind fringe of the urban area half way between M2 and M4.
Typically, M4 should be located along the afternoon PWD 10 to 30 miles from the urban fringe
depending on the average the wind speed.
D. Limitations and Suggested Remedies
In many areas, there are three types of high ozone days: namely, overwhelming
transport, weak transport (or mixed transport and stagnation) and stagnation. The windrose
concept to site ozone monitors is only applicable to the transport types, but not applicable to the
stagnation type. In general, transport types dominate north of 40° N, stagnation type dominates
Ohio River Valley and northern Gulf Coast, and a mixture of the two is observed in the rest of
the eastern U.S.. In areas where stagnation dominates the high ozone days, a well defined PWD
may not be available. If no well-defined PWD can be resolved, use the major axes of the
emission sources as the substitutes for PWDs and the ozone monitors should be located along
these axes but no more than 10 miles from the urban fringe. The reasons for this choice are
two-fold: 1) Completely calm condition seldom last more than an hour during the day. Most
stagnation days have light (< 3 kts) but variable winds; 2) Ozone concentrations are likely to
be the highest when the winds are along the axis of emissions because precursor concentrations
are likely to be highest and dispersion is minimum.
For coastal cities, synoptic winds are generally influenced by the Seabreeze or lake
breeze circulations. This is typically reflected in the difference of the morning and afternoon
PWDs. The maximum-ozone monitors should be located at the downwind side of the resultant
winds (i.e., the vector average of the morning and afternoon PWDs) and the monitors should
be located as near as possible to the sea/lake breeze convergence zone.
VOC AND NO, EMISSIONS MONITORING
Exposure to toxic VOC and NOx gases is a prime concern. Thus, the use of the
prevalent wind directions derived from local climatological data is adequate for siting VOC/NOx
monitors (M2a) designed primarily to assess exposure to air toxics. It is not surprising that M2
and M2a are likely to be at different locations since meteorological conditions favorable for high
VOC and NOx concentrations may not be favorable for ozone formation. For instance, NOX
concentrations tend to be much higher on cold stable winter mornings than on hot summer
afternoons when ozone levels are high.
CONCLUDING REMARKS
Meteorological considerations in siting monitors for photochemical pollutants are
discussed. A method using ozone conducive windrose data to site ozone and its precursors
monitors has been presented. The ozone conducive criteria derived from 31-city, 10-year
meteorological data appear to be quite robust, suggesting that this method may be applicable to
site photochemical pollutant monitors in other areas of the US. However, due to the vast
climatological and topographic differences across the nation, the suggested criteria should only
be used as general considerations for siting monitors of ozone and its precursors. For a better
designed ozone monitor network, it is recommended to use any additional credible local data,
if available, to better define the local climatological conditions. Many local data may not be
reported to the National Meteorological Center data base. Understanding the local micro-
climatology and precursors distribution, particularly in the complex terrain area, will greatly help
822
-------
in finding the best monitor locations.
REFERENCES
1. F. M. Vukovich, W. D. Bach, Crissman, and W. J. King, "On the Relationship between
High Ozone in the Rural Surface Layer and High Pressure Systems. Atmos. Environ., 11,
pp. 967-983, 1977.
2. S. H. Chu, "Coupling High Pressure Systems and Outbreaks of High Surface Ozone
Concentration," Proceedings of the 80th APCA Annual Meeting, 87-113.5, 1987.
3. S. H. Chu and D. C. Doll, "Summer Blocking Highs and Regional Ozone Episodes,"
Seventh Joint Conference on Applications of Air Pollution Meteorology with AWMA.
Preprints by American Meteorological Society, pp. 274-277, 1991.
823
-------
Poster Session
-------
ACCUMULATION OF PERSISTENT ORGANIC COMPOUNDS IN SPRUCE
NEEDLES IN RELATION TO CONCENTRATIONS IN AIR AND
DEPOSITION
Eva Brorstrom-Lundln and Anne Lindskog
Swedish Environmental Research Institute
P.O. Box 47086
S-402 58 Gothenburg, Sweden
ABSTRACT
There is an important contribution of persistent organic compounds, POC, to the Swedish
west coast via atmospheric transport and deposition.Uptake by plants may play a significant role
in the circulation of these compounds in terrestrial environment.
The aim of this study is to obtain knowledge about accumulation of POC in spruce needles
in relation to concentrations in air and amounts in deposition (artificial surface). In addition
atmospheric contribution of POC to a forest ecosystem at the Swedish west coast can be
estimated.
Measurements of POC have been carried out in campaigns during different seasons at an
established experimental site, where a forest area is covered with a roof. Needles were collected
both outside the "roof yielding the total deposition and below the "roof where the main
mechanism is gas phase accumulation.
The results indicated that the contribution of POC via deposition from the air to a forest
area is similar to that at a coastal station. No differences in the concentrations of HCH, HCB and
in the needles were obtained between samples collected during a winter period outside and below
the roof but a decrease in the content of particle bound PCB occurred in needles collected under
the roof compared to the total accumulation.
INTRODUCTION
Persistent organic compounds, POC, frequently present in the atmosphere include such
classes as polychlorinated biphenyls, PCB, chlorinated benzenes and pesticides such as
hexachlorocyclohexanes, HCH. Since most of the POC are chemically stable a considerable long-
range transport of these compounds takes place in the atmosphere. i>2.3>4 They exist in the
atmosphere both in vapour phase and bound to particles and the atmospheric life time depends on
factors as reactivity, polarity, etc.5-6
The main sink mechanism for air-borne POC is through deposition to water, soil and
vegetation. The deposition takes place either as dry or wet deposition and includes scavenging of
both the particles and the gas phase. 7-8 Deposition fluxes are highly variable and are controlled
by factors as chemical composition, water solubility, particles size, atmospheric conditions and
type of surface.5 Surfaces such as water surfaces with is lipophilic micro layer and plant waxes
may be good for accumulation. 9-10
Deposition to plant surfaces occurs through uptake of lipophilic compounds in gas
phase, but there may also be a removal in connection with higher ambient temperature or when
the concentration in the air decreases. n POC with low vapour pressure, which mostly exists in
the particle phase in the atmosphere, will also accumulate on the leaf surface. 12
In terrestrial ecosystems vegetation may play an important role in the circulation and
bioaccumlation of POC. 10 In several investigations concentrations in vegetation is used to get a
measure of the loading of stable organic compounds in the environment while the content of
organic compounds in vegetation may change due to biodegradation, reactions and
reemission.13'14.15
827
-------
The aim of this study is to obtain knowledge about the occurrence of POC in spruce
needles in relation to the concentration in air and amounts in deposition (artificial surface). The
uptake in the needles is studied both as a result of accumulation in gas phase and the total
accumulation. In addition, these measurements give information about the contribution of POC
to a forest area at the Swedish west coast.
Measurements of POC at background stations at the Swedish west coast have shown that
this area receives a significant amount of POC as a result of long-range transport but also due to
remission from the sea. 16>17 The compounds studied in this investigation are PCB, HCH and
hexachlorobenzene, HCB.
EXPERIMENTAL
The study was carried out at an experimental site for acidification research, Gardsjon, in
a forest situated about 15 km from the Swedish west coast. 18 At Gardsjon atmospheric
deposition is intercepted of by a transparent plastic roof, 2-4 m above the ground, covering an
area of 6300 m2. The roofed area includes 370 trees, of 15-25 m height with their crowns above
the roof. The roof consists of polycarbonate sheets transmitting 90% of the light.
The sampling of POC was carried out weekly in two campaigns at different seasons, the
first Nov. 19 to Dec. 12. 1991 and the second May 7 to June 4 1992.
POC in air was collected using a high volume air sampler (HVS), which was placed 2 m
above the ground level outside the roof. The sampler was equipped with a glass fibre filter
(Grycksbo Munktell 160 MG) for collection of the particles and an adsorbent of polyurethane
foam (PUR) for the gas phase. The sampling flow rate was about 20 m^ hour-1. The diameter of
the filter and PUR plug was 142 mm and the length of PUR plug 45 mm. Two plugs in series
were used to reduce breakthrough. A third back-up plug was used to control the possible
breakthrough of POC during the sampling. The PUR-plugs were cleaned before sampling by
boiling them in toluene for several hours, followed by Soxhlet-extraction with acetone overnight.
POC in deposition was measured using an open sampler placed 1.5 m above the ground
level outside the roof. Both wet and dry deposition was collected with the deposition sampler,
which consisted of a 1 nfl Teflon coated surface with 10 cm high edges. The bottom inclines
slightly to a central opening an adsorbent consisting of two cylindrical PUR plugs, 4.5 cm * 10
cm, which were used for collection of POC in the precipitation. Particles deposited to the surface
were rinsed from the Teflon area with ethanol, using a Teflon scraper, and collected in a separate
glass bottle. Prior to the sampling the deposition sampler was rinsed with ethanol.
In order to obtain the total accumulation of POC in spruce needles, annual shoots were
collected once a week about 1-2 m above the ground from trees outside the roof. Annual needles
from twigs under the roof were collected to study the uptake of POC in gas phase. The samples
were kept frozen in glass jars until analysis.
Sample Preparation and Analyses
After sampling filters and adsorbents were Soxhlet-extracted with acetone for 24 hours.
The acetone extract was then diluted two-fold with water and extracted twice with pentane/ether
9:1. The extracts were analyzed separately.
The ethanol used for rinsing the Teflon surface of the deposition sampler was filtered
twice using Millipore Teflon filters, 10 and 0.5 /*m, respectively. The filters were then Soxhlet-
extracted with acetone for 24 hours and the organic compounds extracted from acetone into
pentane/ether as described above. After filtration, the ethanol was diluted with water 1:1 and
shaken with pentane/ether as described above. The two organic extracts were then combined.
About 30g needles, fresh weight, were used for the analyses. The samples were Soxhlet-
extracted in portion of 10 g each, first with acetone for 24 hours and then with hexane for 24
hours. The acetone extracts were diluted two-fold with 2 % Na2SO4 and extracted twice with
828
-------
pentane/ether 9:1. The hexane extract was washed with 2 % Na2SC>4 and the organic phases were
then combined.
Internal standards (2,2',3,3t,4,5>5',6,6'-nonachlorobiphenyl and o,p'-DDD) were added
to the combined extracts which then were concentrated to about 3 ml and treated with
concentrated sulphuric acid (1:1). After removal of the sulphuric acid the extracts were
concentrated further and then fractionated on a 2% aluminium oxide column. Three fractions with
increasing polarity were collected, pentane, pentane:toluene 95:5, pentane, toluene: ether
45:5:5.19 In the analytical procedure extraction blanks were used.
The analyses were carried out using a gas chromatograph (Varian 3400) with a SPI
injector, equipped with a 50 m capillary column, 0.25 m i.d., 0.25 /xm film thickness (CP-Sil-8
CB, Chrompack, Holland) and an electron-capture detector (Ni63). The peak areas were recorded
using an integrator (Spectra Physics 4270) and the baselines were corrected using a lab data
system (Winner). The concentrations of the chlorinated compounds were adjusted using the
internal standards and calculated by comparison to certified standards. The concentrations were
also corrected for background from the blank analyses.
Seven individual PCB congeners (IUPAC nos. 28, 52, 101, 118, 153, 138, and 180) were
determined and the sum calculated. Three isomers of HCH were determined, alpha, beta- and
gamma-HCH.
Ambient Conditions
Meteorological parameters such as precipitation and ambient temperature are obtained
via the Gardsjo project. Rorvik, a field station on the Swedish west coast, south west of Girdsjon
is a monitoring station within the European Monitoring and Evaluation Program (EMEP). Data
via EMEP indicate the level of air pollution to the Swedish west coast.
The meteorological conditions, precipitation and ambient temperature for the different
sampling occasions are demonstrated in Table 1. During the first measuring period, Nov. 19 to
Dec. 12. 1991 the average temperature was rather constant ,but the precipitation varied among
the different sampling occasions. _ Data from Rorvik showed that events with increased
concentrations of nitrogen dioxide (NO2), sulphur dioxide (SC>2) and soot occurred several times
during this period, indicating an influence of air pollutants originating from other countries in
Europe (EMEP data).
Some precipitation occurred in the beginning of the second measuring period, May 7 to
June 4 1992, but during most of this period the meteorological situation was characterised by dry
and warm weather. The level of air pollutants measured at Rorvik may be regarded as normal for
this time of the year (EMEP data).
RESULTS
The concentrations of POC in air, spruce needles and deposition are shown in Table 2.
The concentrations of HCB in the air was only measured in the winter samples, G1-G4.
Sampling of the air concentrations was not carried out at Girdsjon during the summer period,
G9-G12, while data from Rorvik have been used for the second period.
Concentrations in Air
The concentrations of PCB in the air was rather constant among the different sampling
occasions during the whiter period, and the sum of the seven analysed congeners varied between
14-18 pg/m3. Higher concentrations of PCB occurred during the summer period, 38-63 pg/m3.
The concentrations of PCB increased with increasing ambient temperature. The distribution
among the different congeners are shown in Figure 1. The PCB components were found almost
exclusively in the adsorbent in the summer samples, while in the winter 10-40 % of the congeners
153, 138 and 180 were found on the filters.
829
-------
Compared to PCB there were greater variations in the air concentrations of HCH among
the different sampling occasions in the winter samples. Total concentrations were at the same
level for summer and winter samples, but the distribution among the different isomers was quite
different.
Concentrations in the Spruce Needles
An average concentration of 2.6 ng/gTS of the identified PCB was obtained in the needles
as a result of total accumulation in the four winter samples, G1-G4. There were only small
variations in the PCB concentrations among the different sampling occasions but there were
differences between the PCB content in needles collected outside and under the roof respectively ,
which is demonstrated in Figure 1. The least volatile PCB species were found in lower
concentrations in the needles collected under the roof.
The PCB content in the needles collected outside the roof was somewhat lower in the
summer samples compared to what was found during the winter period. 1.6 ng/gTS respectively
1.3 ng/gTS of PCB were obtained in the needles collected during the sampling occasion 09-10
and Gll-12
The relative distribution of PCB found in the needles outside and under the roof were
more similar in the summer samples compared to the winter samples. However higher
concentrations occurred in the needles under the roof, especially in the G11-G12 sample.
HCB, which in the atmosphere almost exclusively exist in the gas phase, was present in
the same level in all samples but with somewhat higher concentrations in samples collected under
the roof. This also to some extent applies to HCH in spite of the greater variations in the air
concentrations. However in the summer samples gamma-HCH was present in lower
concentrations in the needles under the roof compared to needles outside the roof.
Amounts in Deposition
The amounts of the PCB identified in the deposition samples G1-G2 varied between 5-8
ng/m^ day. G3 showed somewhat greater deposition, which probably is an overestimate due to
interferences in the analyses of PCB cogeners 28, 52 and 101. Smaller amounts of PCB
deposition was obtained in the summer samples, 3.5 ng/m^ day respectively 0.92 ng/m2 day.
The distribution of the PCB congeners in the deposition is demonstrated in Figure 1. The
deposition contained relatively more of highly chlorinated PCB compared to the air samples
which indicates that a greater share of particle bound PCB are present in the deposition samples.
Some co-variation between the amounts of PCB in the deposition and the amounts of
precipitation was obtained.
The deposition of HCH varied a lot among different sampling occasions and a significant
correlation to precipitation was obtained. The distribution among the isomers was quite different
compared to what was found in the air and the needles.
DISCUSSION AND CONCLUSIONS
Measurements of airborne PCB at an open background station at the Swedish west coast,
Rorvik, during winter have shown that the concentration of the seven identified PCB usually is in
the level of 20-30 pg/m3 (Brorstrom-Lunde'n unpublished data). The atmospheric concentrations
of PCB obtained in the winter samples in the forest area, Girdsjon, were somewhat lower. The
deposition fluxes of PCB obtained at Gardsjon was higher compared to Rorvik, where an average
of 3ng/m2 day during comparable seasons has been obtained. The atmospheric concentrations of
HCH and the amounts of HCH in the deposition were at the same level as those obtained at
background stations at the Swedish west coast.
The results in this study indicate that an important contribution of POC, here represented
by PCB and HCH, to a forest ecological system at the Swedish west coast may take place due to
830
-------
deposition. However, up to now only a few measurements have been carried out at Gardsjon and
deposition measured using an artificial surface which most likely affect the deposition velocity
differently than natural surfaces, as soil, vegetation and water surfaces.
No accumulation of the analysed compounds in the needles was observed during the two
measuring periods. The content of PCS in the needles may be due to uptake of PCB in both gas-
and particle phase. The precipitation did not seem to affect the uptake of POC in the needles and
no correlation between the concentrations of POC in the needles and the measured deposition
was obtained.
The lowest PCB concentrations in the needles were found when the highest concentrations
in the air occurred and in connection with the highest temperature, which indicates that the
equilibrium between POC in the gas phase and the concentration in the needles probably depends
to a great extent on the ambient air temperature.
The higher concentrations of the POC in the needles under the roof compared to the
needles outside in the summer samples are not clearly understood. An explanation may be that the
concentrations probable will be affected under the roof due to evaporation from the soil and /or a
reduction in ventilation.
REFERENCES
1. Gregor, D.J. and Gummer, W.D. Evidence of Atmospheric Transport and Deposition of
Organichlorine Pesticides and Polychloranited Biphenyls in Canadian Arctic Snow.
Environ. Sci. Technol.. 23, No. 5 (1989).
2. Gotham, W and Bidleman, T. Estimating the Atmospheric deposition of Organic
Contaminants to the Arctic. Chemosphere Vol. 22, No. 1 pp. 165-188 (1991).
3. Oehme M. Further Evidence for Long-range Air Transport of Polychlorinated
Aromates and Pesticides: North America and Eurasia to the Arctic. Ambio Vol. 2'0
No.7 pp. 293-297 (1991).
4. Ballschmiter, K. and Wittlinger, R. Interhemisphere Exchange of Hexachlorocyclo-
hexanes, Hexachlorobenzene, Polychlorinated Biphenyles and 1,1,1-Trichloro-
2,2-bis(p-chlorophnyl) ethane in the Lower Troposphere. Environ. Sci. Technol.
Vol. 25, No. 6, 1103-1111 (1991)
5. Bidleman, T.F.Atmospheric processes. Environ. Sci. Technol. Vol. 22, No. 4 pp.
361-367 (1988).
6. Ligocki, M.P. and Pankow, J.F. Measurements of the Gas/Particle Distributions of
Atmospheric Organic Compounds. Environ. Sci. Technol. Vol. 23, No. 1 pp. 75-83
(1989).
7. Ligocki, M., Leuenberger, C. and Pankow, J.F. Trace Organic Compounds in Rain III.
Particle Scavenging of Neutral Organic Compounds. Atmospheric Enviomnment.
Vol.19, No. 10pp. 1619-1626 (1985).
8. Ligocki, M.P. and Pankow, J.F. Measurements of the Gas/Particle Distributions of
Atmospheric Organic Compounds. Environ. Sci. Technol. Vol. 23, No 1 pp. 75-83.
(1989).
831
-------
9. Sodergren, S., Larsson, P., Knulst, J. and Bergquist, C. Transport of Incinerated
Organoclorine Compounds to Air, Water, Microlayer and Organism.
MarinePollution Bulletin. Vol. 21, No. 1, pp. 18-24 (1990).
10. Calamari, D., Bacci, E., Focardi, S., Gaggi, C., Morosini, M. and Vighy, M.
Role of Plant Biomass in the Global Environment Partitioning of Chlorinated Hydro-
carbons. Environ. Sci. Technol.. Vol. 25, No. 8. pp. 1489-1495 (1991).
11. Hortsman, M and McLachan, M. Initial Development of a Solid-Phase Fugacity Meter
Semivolatile Organic Compounds. Environ. Sci. Technol.. Vol. 26, No. 8. pp.
1643-1649 (1991).
12. Paterson, S., Mackay, D and Shiu, W Uptake of Chemicals by Plants: a Review of
Processes, Correlation and Models Chemosphere. Vol.21. No. 3, 297-331(1990).
13. Brorstrom, E. and Skarby, L. Plants as monitoring samplers of airborne PAH.
In 3rd Ec.svmposium. Varese, Italy 1984, pp. 166-175 (1984).
14. Villeneuve, J-P., Fogelqvist, E. and Cattini, C.Lichens as Bioindicators for
atmospheric Pollution by Chlorinated Hydrocarbons. Chemosphere. Vol.17, No. 2, pp.
399-403 (1988).
15. Erikson, G., Jensen, S., Kylin, H. and Strachan, W The pine needles as monitor
atmospheric pollutants. Nature Vol. 341, pp. 42-44 (1989).
16. Larsson, P. and Okla, L. Atmospheric Transport of Chlorinated Hydrocarbons to
Sweden in 1985 Compared to 1973. Atmos.Environ. Vol. 23. No. 8, pp. 1699-1711
(1989).
17. Brorstrom-Lunde'n, E., Lindskog, A and Mowrer, J. Concentration and Flux of Organic
Compounds in the Atmosphere of the Swedish west coast. Atmospheric Environment.
Submitted (1993).
18. Hultberg, H., Andersson, I. and Moldan, F. The covered catchment - An Experimental
approach to Reversal of Acidification in a forest ecosystem. Proceedings from the
International Symposium on Experimental Manipulation of Biota and biogeochemical
Cycling in Ecosystem - Approach, Methodologies, Findings Copenhagen 18-20 May
(1992).
19. Samuel, S., Atuma, A., Jensen, S., Mowrer, J., Orn, U. Separation of Lipophilic
Substances in Environmental Samples with Special Reference to Toxaphene.
Intern. J. Environ. Anal. Chem. Vol. 24, pp. 213-225 (1986).
832
-------
Table 1. Dates for the different sampling occasions,
the amounts of precipitation and the ambient temperature.
SAMPLE
G1
G2
G3
G4
G9-10
G11-12
DATE
Nov1 9-Nov26 1 991
Nov26-Dec3 1991
Dec3-Dec12 1991
Dec 12-Dec18 1991
May7-May21 1992
May21-June4 1992
PRECIPITATION
mm
4.4
10.5
0
22.1
24
0
TEMPERATURE
C
3.0
5.8
1.7
3.9
11.2
19.3
Table 2. Concentrations of POC in air and needles and amounts in
deposition for the different sampling occasions
Sample
AIR
HCB
PCB
HCH
Needles outside the roof
HCB
PCB
HCH
Needles under the roof
HCB
PCB
HCH
DEPOSITION
HCB
PCB
HCH
G1
pg/m3
28
18
201
ng/g TS
0.48
2.5
2.6
ng/g TS
0.55
1.3
2.6
ng/m2day
0.14
6.14
1.6
G2
pg/m3
22
14
74
ng/g TS
0.48
3.0
2.3
ng/g TS
0.57
1.2
2.5
ng/m2day
0.22
5.1
24
G3
pg/m3
44
18
147
ng/g TS
0.48
2.7
2.5
ng/g TS
0.52
1.2
2.8
ng/m2day
0.26
9.4'
1.6
G4
pg/m3
28
17
137
ng/g TS
0.51
2.4
2.6
ng/g TS
0.65
1.3
3.7
ng/m2day
0.39
6.4
30
G9-10
pg/m3*
* *
38
198
ng/gTS
0.29
1.6
2.2
ng/gTS
0.54
2.0
2.0
ng/m2day
0.20
3.5
14
G11-12
pg/m3*
* *
63
242
ng/gTS
0.41
1.3
2.2
ng/gTS
0.52
2.3
1.8
ng/m2day
0.38
0.92
0.06
PCB = sum of 28, 52, 101,118, 138, 152 and 180
HCH = alpha-, beta- and gamma-HCH
* The air concentrations from Rdrvik
'Not determind
833
-------
Figure 1. The distribution among the different PCB
pg/m3
AIR
SPCB28
DPCB52
•PCB101
• PCB118
E3 PCB153
DPCB138
HPCB1SO
ng/m2 day
DEPOSITION
o
O)
o
II PCB28
D PCB52
IPCB101
I PCB118
El PCB1S3
DPCB138
SPCB180
NEEDLES OUTSIDE THE ROOF
NEEDLES UNDER THE ROOF
ng/gTS
ng/gTS
o
o>
S PCB28
D PCBS2
• PCB101
• PCB118
H PCB153
D PCB138
H PCB180
H PCB28
D PCBS2
•PCB101
• PCB11S
HPCB153
DPCB138
SPCB180
-------
Figure 2. The distribution among the different HCH
pg/m3
140
120
100
80
60
40
20
o
F-
~
i
1
=
n
5
1
CM
(3
AIR
|
=
c'; pj
=
=
j
„
=
1
E
= -^
""
E \'
! ^-
— :
[3 alfa-HCH
D beta-HCH
§ gamma-HCH
n
-------
ESTIMATES OF PESTICIDE EXPOSURE FROM THE AGRICULTURAL
HEALTH STUDY (AHS)
Nicholas J. Giardino, David E. Camann, Paul W. Geno, H. Jac Harding,
and Jacquelyn M. Clothier
Southwest Research Institute
P.O. Drawer 28510
San Antonio, TX 78228-0510
ABSTRACT
Several routes of pesticide exposure to fanner applicators were measured and compared. Whole body
(WB) exposure was measured using a-cellulose patches placed on the outer clothing of the farmer, two measures
of hand exposures were taken, one using cotton gloves (GL) and the other using isopropanol handwipes (HA).
Inhalation exposures were measured using a PUF cartridge coupled with a personal sampling pump. The three
highest exposures received in descending order were: WB > GL > HA.
INTRODUCTION
The primary goal of the Agricultural Health Study (AHS) sampling program is to provide quantitative
measures of pesticide exposure from different farm tasks. In this study, the term exposure is defined as what
contacts a person, whether on clothing, skin, or through simple processing by breathing or ingestion. Absorption
and metabolism after this contact takes place is not considered here.
METHODS
All of the methods for collecting the field samples used to assess exposures to pesticides can be found in
Harding et al.1 The analytical methodology used to analyze all field samples can be found in Geno et al.2
RESULTS AND DISCUSSION
Hygiene, Activity Patterns and Demographics
Hygiene. The degree that each farmer practiced good hygiene varied substantially from farm to farm. Farmers 2
and 3 exercised more caution while loading and mixing pesticides. Farmer 2 was the most careful in the use and
application of his pesticides. He did not experience any spills or splashes during loading and mixing of the
pesticides. When applying the pesticide to his crop, he was inside an enclosed cab. A drift retardant was added
to the pesticide application mixture. Farmer 2 also tried to purchase only that amount of pesticide needed for a
specific job reducing leftover waste. He did have to stop work to make a five-minute repair on a leaking hose.
He wore approximately the same clothing as the other farmers. This included a baseball cap, cotton gloves, a
cotton work shirt and pants, and shoes. Farmer 2 washed his hands within 15 minutes after leaving work.
Farmer 3 also exercised good care in handling of pesticides. He wore rubber or leather gloves during all
phases of his work. While applying pesticides, he had to stop work for approximately two minutes to repair a
nozzle. This farmer also did spraying of hogs inside a barn with an aerosol can. He washed his hands within 30
minutes after leaving work. Fanner 3 was observed eating his lunch in the field, during a work break, without
first washing his hands.
836
-------
Fanner 1A also may have contaminated himself while hosing down some hogs with lindane. Farmer 1A
does animal application of pesticides four times yearly. He washed his hands within one to two hours after leaving
work.
Farmer IB poked a hole through a container of alachlor and saturated the finger of one of the cotton
gloves he was wearing. This glove was removed because of the chance of high exposure. Otherwise, Farmer IB
reported little hand contact with any pesticides, since he reported that he washed his hands between each loading.
Farmer IB also applied pesticides to animals six times yearly.
All the farmers changed clothes once a day, and their work clothes were washed once every three days.
Activity Patterns. All the farmers spend as much as 16 hours in the field during a busy season. Crops can be
sprayed every day for up to a two-week period. Fanner 1A reported being active during work and inactive during
leisure hours. Farmer 1A had a three year old child. This child's activities were reported for a total period of
70 hours. Out of this total time of 70 hours only one and one-half hours were spent playing in dirt by the child.
No hand-to-mouth activity was reported although some probably occurred. While not working, Farmer IB spent
most of his time with his spouse indoors during the monitoring period. The family of Farmer 2 all participated
in planting and gardening. The wife of Farmer 2 spent four and one-half hours in the field during the three-day
monitoring period. She spent three and one-half hours of this time in the field while pesticides were being applied.
The family of Farmer 3 also participated to varying extents in lawn and gardening activities.
Demographics. At Farm 1A there was a married couple between the ages of 26 to 45 years old. They had two
children, a son approximately one year old and a daughter three years old. Fanner IB and his spouse were at least
60 years old. Fanner 2 and his spouse were 26 to 45 years of age. They had a 15 year old son and a 13 year
old daughter. Farmer 3 and his spouse were 26 to 45 years old and had a daughter eight years old and a son four
years old.
Occupational Exposure of Farmers During Monitoring Events
Hand Versus Inhalation Exposure. The highest (but not the only) exposures to the hands can occur during
accidental spills and splashes, while loading or mixing pesticide, or during repairs on leaky hoses or nozzles. An
accidental spill with cotton glove and possible hand contamination did occur on Farm IB on Day 4 of the
monitoring period. This accidental spill at this farm highlights the importance of considering hand exposure.
Also, it is unknown whether any of the farmers, while at work, washed their hands before urinating. If not, some
pesticide may have been transferred from the hands to the genital area. This is of special concern because of the
thin dermal layer of the penis and scrotum through which pesticides will transfer readily.
Table I addresses two issues. One concerns the inhalation exposure (BC, see Table I) received by the
farmer during monitored application events. The second is the comparison of total exposure in milligrams received
by the hands (HA, see Table I) as compared to the total milligrams sampled in the farmer's breathing zone. It
is clear that for those cases for which a TLV (Threshold Limit Value) was available, the farmers' inhalation
exposures, while working during the monitoring periods, were far below the TLVs.3 A similar comparison can
be made between the mass of pesticide recovered from the hands and the mass collected by personal air sampling.
The mass of pesticide measured on the hands is far greater than that collected by personal air sampling.
837
-------
Table I. Comparison of handwipe amount to breathing concentration and amount inhaled.
Farm by Event(1>
Alachlor
1B-3A
IB-SB
1A-3C
IB^tA
Atrazine
1B-3A
1B-3B
1A-3C
1B-4A
2,4-D Isooctyl Ester
2-3
Lindane
1A-4B
Pyrethrins
3-4A
Trifluralin
2-3
Time
(h)
3.000
5.400
2.600
3.300
3.000
5.400
2.600
3.300
3.000
1.100
0.045
3.000
TLV
(mg/m3)
NONE
NONE
NONE
NONE
5
5
5
5
NONE
0.5 (SKIN)
5
NONE
BC(2>
(mg/m3)
0.008
0.011
0.014
0.020
0.002
0.002
0.003
0.003
0.001
0.001
0.240
0.001
Mass Processed by
HA0' Inhalation During
(mg) Work (mg)(4)
6.380
14.800
0.480
9.540
2.000
4.330
0.113
3.76
0.152
0.152
0.450
0.023
0.043
0.110
0.065
0.120
0.011
0.019
0.014
0.018
0.005
0.010
0.020
0.005
(1)
All pesticides applied during these events. See Table IV for explanation of
coded events.
BC = The average breathing concentration (mg/m3) during the application of pesticides.
For the HA (applicator handwipe) data, recovery from the skin is unknown. Recovery
from the isopropyl-saturated cotton gauze used to take the handwipes was good. Biases
in the HA data due to skin absorption or other mechanisms is unknown.
Calculated by multiplying time (h) * BC (mg/m3) * 1.8 m3/h (see note below).
Threshold Limit Value (mg/m3), defined as that concentration in air that a healthy
worker may be exposed to for an eight-hour working period presumably without adverse
health effects resulting.
NOTE: A moderate workload was assumed for the farmers with a breathing rate of 1.8 m3/h.
The personal air samplers were run at 0.23 m3/h.
(2)
(3)
(4)
TLV
Inhalation exposure is most likely to occur from overspray while the pesticide is being applied to the
crops. Some small contribution may be received during mixing and loading.
Whole Body Exposure Versus Hand. Another measure of integrated exposure is whole body (WB) exposure.
This was calculated using a scaling factor for each farmer. This scaling factor was calculated by first obtaining
the surface areas of the farmers' bodies off a nomogram.4 This required a knowledge of each fanner's height and
weight. Next, the total area of the analyzed body a-cellulose patches was calculated. Then to get the scaling
factor the estimated surface area of the farmer was divided by the total area for all three analyzed body a-cellulose
838
-------
patches. The results of these calculations are in Table n. To calculate the estimated whole body (WB) exposure,
the corresponding body patch concentration was multiplied by the scaling factor to yield a result in milligrams.
These results were then compared to the measured mass in milligrams on the cotton gloves as is shown in
Table n. Comparing the median values in Table II shows that the estimated WB > GL, but on the same order
of magnitude.
Table n. Comparison of whole body exposure to pesticide retained on cotton gloves.
Farm by Event*
Alachlor
1B-3A
1B-3B
1A-3C
1B-4A
Atrazine
1B-3A
1B-3B
1A-3C
1B-4A
2,4-D Isooctyl Ester
2-3
Lindane
1A-4B
Pyrethrins
3^A
Trifluralin
2-3
DP
(mg)
2.05
5.39
0.38
1.02
0.137
0.296
0.114
0.373
0.00891
0.375
0.023
0.00047
Scaling
Factor
223
223
218
223
223
223
218
223
202
218
202
202
* All pesticides applied during these events.
(1) Finger tip of glove saturated with pesticide.
DP Dermal o-cellulose patches attached to front o
Whole Body Exposure
(WB in mg)
457
1022
83
227
30
66
25
83
1.8
82
5
0.1
MIN=0.1
MAX=1078
MEDIAN=67
GL
(mg)
105
270
59.8
>2370(1)
2.37
33.2
5.25
16
29.3
21.2
2.53
0.252
0.252
2370
25.25
f thighs and to the back of the nape
of the neck.
SCALING FACTOR: Estimated surface area of farmer divided by total area of all
three body a-cellulose patches.
WB Whole body exposure (mg). WB (mg) = (DP)*(SCALING).
GL Cotton glove measurements (mg).
Application Versus Nonapplication Events. Day 2 at all farms and Day 4 at Farm 2 were baseline days during
which no pesticides were applied to the crops. Days 3 and 4 were days during which pesticides were applied to
the crops. It is useful to compare the exposures received by the farmers on Day 2 to those received on Days 3
and 4.
839
-------
It has been shown above that handwipe (HA) data are a good measure of the fanner applicator's exposure.
A comparison of application day exposure versus baseline day exposure is given in Table III for the applicator
handwipe data. Note that the summary statistics in Table III represents all detected handwipe measurements of
all analytes grouped together, separately for application and nonapplication days. The median value for all the
measurements taken on application days is four orders of magnitude above baseline days. The maximum for
Days 3 and 4 is also much greater than that for Day 2. This indicates an increase in exposure due to pesticide
handling and application.
Table III. Farmer handwipe exposure (mg) on application
versus nonapplication days.
Percentiles for Nonapplication Days
10% 25% 50% 75% 100%
0.000015 0.000045 0.000085 0.0029 0.530
N=35
MIN=0.000015
MAX=0.530
MEDIAN=0.000085
Percentiles for Application Days
10% 25% 50% 75% 100%
0.00044 0.0162 0.160 3.76 14.80
N=18
MTN=0.00044
MAX=14.80
MEDIAN=0.16
CONCLUSIONS AND RECOMMENDATIONS
The three highest exposures received in descending order were: WB > GL > HA. The exposure to the
whole body (WB) was comparable to the cotton glove (GL) data. Both represent an integrated average over the
work period. WB exposure is an estimated value which assumes the sampled dermal patch areas are representative
of exposures received by all body surfaces. GL exposure is calculated from direct measurements. Because WB
exposure is an estimated value, this tends to lessen the confidence in the results. GL data may be superior to WB
estimated values, but there were problems with the collection of the GL data. The cotton gloves either were poor
fitting or restricted the farmer in his ability to do his job effectively.
Greater care in exercising good hygiene practices could substantially reduce the farmer applicator's
exposure as well as that of their families. Such practices could include: cautious handling of pesticides, developing
proper hand washing habits, removing shoes prior to entering the home, bathing immediately after work, and
washing work clothes separately in cold water.
840
-------
Table IV. Event ID code.
Farm ID Week Monitored Sampling Period/Event
1A
IB
2
3
Weekl
Week 1
Week 2
Week 3
2 -
3 -
3A-
3B-
4 -
4A-
4B-
Day 2
Day 3
Day 3, 1st event/sample
Day 3, 2nd event/sample
Day 4
Day 4, 1st event/sample
Day 4, 2nd event/sample
ACKNOWLEDGEMENTS
We acknowledge Vincent F. Garry, M.D. of the University of Minnesota for recruitment of participant
farms. This research was funded by the U.S. Environmental Protection Agency (Contract 68D10150). This paper
has been submitted to EPA's peer and administrative review, but no official endorsement should be inferred.
REFERENCES
1. H. J. Harding, P. M. Merritt, J. M. Clothier, D. E. Camann, A. E. Bond, and R. G. Lewis,
"Sample Collection Methods to Assess Environmental Exposure to Agricultural Pesticides," in
Proceedings of the 1993 U.S. EPA/A&WMA International Symposium on Measurement of Toxic
and Related Air Pollutants." Air & Waste Management Association, Pittsburgh, 1993 (in press).
2. P. W. Geno, D. E. Camann, and K. VUlalobos, "Analytical Methods for Assessing the Exposure
of Farmers and Their Families to Pesticides," in Proceedings of the 1993 U.S. EPA/A&WMA
International Symposium on Measurement of Toxic and Related Air Pollutants." Air & Waste
Management Association, Pittsburgh, 1993 (in press).
3. Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices.
ACGM, 1991-1992.
4. Merck Medical Manual. Twelfth Edition, pg. 1840, Figure 57.
841
-------
SAMPLE COLLECTION METHODS TO ASSESS ENVIRONMENTAL
EXPOSURE TO AGRICULTURAL PESTICIDES
H. J. Harding, P. M. Merrill, J. M. Clothier, and D. E. Camann
Southwest Research Institute
San Antonio, Texas 78228
A. E. Bond and R. G. Lewis
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A variety of sample collection media and methods were utilized to assess the environmental exposure of
fanners and their families to agricultural pesticides including application events. Collection media included a PDF
(polyurethane) cartridge and quartz prefilter, cotton gloves, and a-cellulose patches, and handwipes. Low levels of
some target compounds were detected on media field blanks. Field spikes of sample media showed mean recoveries
from 59 to 85%. Collocated PUF cartridges and dermal patches were usually in close agreement. The methodologies
utilized performed adequately to collect different sample matrices for assessing pesticide exposure.
INTRODUCTION
One of the objectives of the method evaluation phase of the NCI/EPA Agricultural Health Study (AHS) pilot
program was to evaluate the proposed multi-media sampling design.1 A number of sample collection media and
methodologies were incorporated into the study design with the primary objective to evaluate exposure of fanners and
their families to agricultural pesticides. Some of the criteria for methodology evaluation from the sampling perspective
included ease of use by field personnel, potential impact on the way the fanner performed his daily activities, and
adequacy to ascertain exposure through the route monitored. This pilot study was conducted on four farms in
Minnesota during the spring of 1992.
SAMPLE COLLECTION METHODS
The multi-media design for environmental exposure incorporated four types of sample media to monitor the
farmer's direct exposure, two types of sample media to ascertain residential exposure for the fanner's family, and two
methods for the collection of samples to explore exposure through track-in of contaminated soil. The fanner's exposure
was monitored widi an air sample for inhalation exposure, body patches for body dermal exposure, and gloves with
a subsequent handwipe for hand dermal exposure. Evaluation of residential exposure was accomplished with air
monitoring at both indoor and outdoor locations and with handwipes on children in the fanner's household. The track-
in phenomenon was explored by collection of surface soils in pathways and carpet dust from the interior of the
farmhouse. The exposure samples were analyzed by methods described by Geno et al.2
Personal Exposure Monitoring of the Farmer
Each fanner was monitored for personal exposure to pesticides via inhalation and dermal routes during fanning
activities which included both general farm activities not involving pesticides and pesticide application events conducted
by the farmer. A pesticide application event included retrieving pesticide formulations from storage, mixing the
pesticide with water and other additives to prepare a sprayable mixture, and finally spraying the mixture. The event
also included cleanup tasks provided they occurred within the monitoring window. Samples collected included personal
air for inhalation exposure, gloves for dermal exposure of the hands, and dermal patches for exposure of other body
areas. Additionally, hand dermal exposure was augmented with handwipes collected at the end of each activity period.
The personal air sample collection system consisted of a cartridge (1" diameter by 5" long) containing a PUF
plug coupled with a quartz fiber prefilter. The PUF plugs were 22 mm diameter and 7.6 cm long cut from
0.0224 g/cm3 PUF foam designated as R45. The cut plugs were pre-exiracted with acetone for 24 hours, hexane for
842
-------
48 hours, and then again with acetone for an additional 24 hours. Plugs were dried with a stream of zero nitrogen prior
to assembly in the glass cartridge. The filter holder was a nylon Swagelok® union for 1" tubing which held a 25-mm
diameter 0.5 urn quartz micro-filter supported by a perforated stainless steel disk. The cartridge assembly was wrapped
in aluminum foil and stored in a tall glass jar before and after use. A black shrink tubing sleeve was placed over the
cartridge for protection from light and physical damage during use. The cartridge assembly was held in the breathing
zone area with a collar clip or Velcro® and coupled with Tygon® tubing to a portable air sampler clipped on a belt
Ambient air was drawn through the cartridge at a nominal 3.8 Lpm (liters per minute) as measured at both the start and
end of each monitoring period.
Hand dermal exposure was determined with a pair of all cotton general purpose work gloves with a plain cotton
cuff (undyed). The gloves were washed with a brightener-free detergent and pre-extracted with solvents using the same
sequence as for the PUF plugs prior to field use. Gloves were stored in glass jars prior to use. Only the fanner would
handle the gloves in the jar and would wear the exposure glove under any type of work glove worn as normal practice
for farming activities performed during the work day. At the conclusion of the monitoring period, the gloves would
be returned to the storage jar.
A set of three dermal patches were used to monitor body exposure. These were located in the center of both
thighs and on the nape of the neck and held in place by pinning the patches to the farmer's clothing. The patches were
precut to 4" x 4" squares from a-cellulose sheets and backed by an aluminum foil square the same size to isolate the
patch from the farmer's clothing. A 3/4" border designated as the handling zone was removed from the patch prior to
laboratory extraction.
Handwipes were collected on the farmer at the end of the monitoring day or application period. Two 4" x 4"
6-ply cotton dressing sponges (Johnson & Johnson SOF-WICK) were used directly from the package after each was
laced with 10 mL pesticide-grade isopropanol while using the wrapper to hold the sponges. The first wipe was a
general wipe of the whole hand while the second wipe was used to more thoroughly wipe each digit. Only the farmer
handled the cotton sponges from preparation to placement in the sample container. A 50 mL volume of isopropanol
was added to the sample container to initiate extraction prior to receipt in the laboratory.
Residential Exposure Monitoring
Indoor and outdoor 24-hour ambient air samples were collected for three consecutive days (monitoring days
2, 3, and 4) at each farm. The sampling train consisted of the same PUF cartridge system for personal exposure
monitoring without the quartz fiber prefilter and holder. The indoor location was an area where the family spent most
of their awake time which was generally in or adjacent to the kitchen or family room. The cartridge inlet was placed
at a height of 1.0 to 1.2 m which approximates sedentary breathing zone height. The outdoor sampler was placed in
an area of outdoor activity, typically where young children would play such as a porch or play area, at a 1.5 m height
above ground.
Child handwipes were collected from eligible children (i.e., toilet trained through age 17) at each farm for each
application day. The procedure followed that used for the farmer.
Surface soil and carpet dust sampling
Exposure from track-in of contaminated soil was investigated by collecting carpet dust from the interior of the
farmhouse and surface soil from three areas where contaminated soil may have been present and potentially tracked
in. These areas included the primary entry way to the house used by the farmer, a pathway location between the house
and the pesticide mixing or use areas (e.g., fields), and the pesticide storage area or an area adjacent to the mixing area.
The carpet dust was collected from wall-to-wall carpeted areas in each farmhouse using the High Volume Small
Surface Sampler (HVS3). Several high traffic areas were identified and then vacuumed with the HVS3 using
appropriate setting for the carpet type to obtain a representative sample and sufficient mass for analysis.
The collection technique used for the surface soil samples was dependent on the characteristics of the surface
sampled. On hard surfaces, such as concrete walks, steps, or compacted soil, a clean bristle brush was used to sweep
843
-------
loose soil and dust into a stainless steel pan. If a step off mat was in the sampled area (i.e., the entryway), the soil was
collected by turning it upside down and gathering the soil that fell from the mat Alternatively, if the mat had a solid,
yet pliable backing, the mat was rolled into a cylinder, turned vertically, and then gently thumped over a stainless steel
catch pan. In gravel areas such as driveways or on bare ground, a clean bristle brush was used to gather the top Vg"
(2 mm) of loose soil.
Application events
Two types of application events were monitored: (1) spraying post-emergent herbicides to crops using a ground
boom sprayer, and (2) applying insecticides in livestock operations. Table I summarizes the herbicide application events
monitored at each farm. Since the fanner did not significantly change formulations during the monitored events, the
herbicide mixture information applies to all events monitored at that farm.
Table I. Summary of herbicide application events.
Farmer-
Event
1B-3A
IB-SB
1A-3C
1B-4A
Period
(h)
3.0
5.4
2.6
3.3
Volume
(gals)
300
600
300
300
Herbicide/Mix Ratio
Lasso® (alachlor)
6 gal/300 gal
Marksman® (dicamba, atrazine)
3 '/4 gal/300 gal
Purpose
Control weeds in corn
2-3
3.0
500
LV-4® (2,4-D isooctyl ester)
5 gal/500 gal
Control of broadleaf weeds on
nonharvestable wheat crop
3-3
3^B
7.0
7.2
2750
1600
Pursuit® (imazethapyr)
100 oz/500 gal
Control weeds in soybeans
Two distinctively different insecticide applications are monitored. Both were for control of insect problems
in hog raising operations, but the specific purpose and method of application were dissimilar. The exposure of the
farmer was monitored with the same methodologies as for the herbicide applications.
Fanner 1A applied lindane with a garden style hose sprayer for control of mange mites on hogs. Lindane was
added to the mixing container on a hose sprayer which was then applied directly to hogs as an aqueous stream
containing approximately 0.06% lindane.
Fanner IB used a natural pyrethriu aerosol spray to control flies in a hog bam. The spray was released in a
manner that created a fog within the bam which was enhanced since doors were closed to minimize ventilation.
Approximately one third of a 32 oz. commercially available can was released in four rooms of the barn for a period
of 2.7 minutes.
SAMPLE COLLECTION QUALITY CONTROL RESULTS
Field blanks, field spikes, and collocated duplicates were used to evaluate the contamination potential, accuracy,
and precision of the sampling data. Results for detected target compounds are reported in the following sections.
Field Blanks/Background Levels
Results from the analysis of field blank samples for five matrices are presented in Table H. Nanogram levels
of target analytes were detected in every field blank matrix: PUF plug, PUF plug and quartz filter, glove pair,
isopropanol-laced cotton sponge dressings used for handwipes, and the ocellulose patches used as body patches. Each
field blank matrix was prepared in the same batches as the corresponding field sample matrix. Field blanks were
exposed to the same set-up and take-down steps as field samples. Thus, for neutral target pesticides, similar levels may
be attributable to brief exposure to surface residues through handling during sample set-up or take-down on the farm.
844
-------
Table H. Results of field blank analyses by matrix (rig/sample).
Cotton Dressings
Analyte
Neutral Pesticides
Alacblor
Atrazine
Captan
Chlordane:alpha+gamma
IDDT=DDT+DDD+DDE
Diazinon
Ljndane
Metolacblor
Propoxur
Acidic Pesticides
2,4-D
Dicamba
Pentachlorophenol
Imazepathyr
PUF
7
78
1
3
3
3
1
NA
PUF + Filter
NA
NA
NA
NA
NA
NA
NA
NA
NA
13
NA
Glove Pair
NA
NA
NA
NA
NA
NA
NA
NA
NA
16
1
370
1
36
11
6
31
2
32
147
8
33
NA
2
NA
NA
NA
NA
NA
NA
NA
NA
NA
350
23
4
NA
a-CeUulose
patches
NA
NA
NA
NA
NA
NA
NA
NA
NA
28
1
63
NA = Not analyzed
Field Spikes
Field spikes of eight target pesticides (six neutral and two acidic pesticides) were generated from three PUF
cartridges and from two isopropanol-laced cotton sponge dressings used for handwipe sampling. Spiked samples were
generated in a clean field setting by spiking a known microvolume of a solvent-based spiking solution onto clean
sample media. The same sample handling protocol was followed after spiking as used with actual field samples.
The average recovery and standard deviation of the field spike results for the neutral pesticides in relation to
the data quality objective of 75% are presented in Table UL Recovery of atrazine, chlorpyrifos, and diazinon from PUF
exceeded 75%, whereas recovery of alachlor, Y-chlordane, and propoxur were somewhat lower. Field spike recovery
of chlorpyrifos, diazinon, and propoxur from the handwipe matrix exceeded 75%, while recoveries of alachlor and
atrazine were slightly lower with Y-chlordane being the lowest of the neutral pesticides.
Table UL Percent recovery of field spikes (mean ± standard deviation).
Spiked analyte
Alachlor
Atrazine
Y-Chlordane
Chlorpyrifos
Diazinon
Propoxur
Data quality
objective
>75
>75
>75
275
>75
>75
PUF plug
(air) (n = 3)
66 ±5
84 ±8
67 ±3
85 ± 15
82 ± 16
59 ±3
Cotton gauze
(handwipe) (n = 2)
71 ±2
71 ±8
64±0
82 ±2
83 ±4
80 ±20
845
-------
Sample Duplicates
Collocated sample duplicates of PUF cartridges and a-cellulose patches were collected to estimate sampling
precision. Results are summarized in Table IV for three pairs of duplicate 24-hour air sample concentrations. The
duplicate air concentrations of target pesticides were usually in close agreement, with ratio, larger/smaller air
concentration, no larger than 1.5 in 24 (75%) of 32 duplicate pairs.
Table IV. Duplicate air sample concentrations (ng/m3).
Farm 2 Indoor Air
Analyte
Neutral Pesticides
Alachlor
Aldrin
Atrazine
Captan
Chlorpyrifos
2,4-D Isooctyl ester
EDDT=DDT+DDD+DDE
Dacthal
Diazinon
Dichloran
Dieldrin
Folpet
Heptachlor
Lindane
Propoxur
Trifluralin
Acidic Pesticides
2,4-D
Dicamba
Pentachlorophenol
Primary
6.2
6.9
0.6
13.2
1.7
0.2
0.3
0.6
0.7
2.7
28.7
2.5
0.6
Duplicate
14.3
17.2
10.6
2.1
0.3
0.3
3.3
0.6
2.9
29.3
2.5
0.9
Farm 2 Outdoor Air
Primary
4.0
4.8
0.4
21.1
0.1
1.1
0.3
0.8
12.8
1.4
0.3
Duplicate
1.9
0.3
2.9
0.4
19.0
0.3
1.6
0.4
14.3
0.6
0.3
Farm 3 Indoor Air
Primary
13.6
3.4
NA
0.2
5.8
0.2
0.2
2.0
3.1
19.5
0.8
0.8
0.6
1.1
Duplicate
13.2
2.0
3.4
NA
0.3
0.2
0.2
2.4
3.5
17.9
0.8
0.3
0.6
1.4
NA = Not analyzed
Farmer 3 wore two sets of dermal patches above his clothing on both thighs and between his shoulders beneath
the nape of the neck during the imazepathyr application event on day 3. The results are reported in Table V for the
collocated duplicate dermal patch samples which were formed by combining alternate patches from the three locations.
The paired dermal patch amount ranged from no difference (3 ng each) for pentachlorophenol, through two-to-three-fold
difference for 2,4-D and dicamba, to a ten-fold difference for imazepathyr. As expected, dermal patch amounts
exhibited more variability than 24-hour fixed location air concentrations.
846
-------
Table V. Duplicate dermal patch sample amounts from Fanner 3
(ug/3 patches)
Acidic Pesticides
2,4-D
Dicamba
Pentachlorophenol
Imazethapyr
Primary
0.170
101
0.003
0.35
Duplicate
0.073
29.4
0.003
3.67
CONCLUSIONS
In general all the sample collection media and methodologies performed adequately to collect the respective
sample matrix for assessing exposure of individuals to pesticides. Specific observations for each of sample media
follows.
The PUF cartridge continues to be a suitable media to collect air samples for inhalation exposure.
• The use of a pre-filter as part of the air sampling train depends upon the sampling environment and target
analyte, but the filter must be supported while in the holder, especially when used for personal exposure
monitoring.
• The limited number patches may not accurately estimate body exposure and more patches appropriately placed
should be used to assess this exposure.
• Gloves for hand dermal exposure may not accurately estimate hand exposure since parts of the glove may
become saturated through accidental contact with the pesticide or work habits may be influenced if gloves are
not normally worn or if normal work gloves are worn over the exposure gloves.
• The isopropanol-laced cotton dressings used as handwipes demonstrated a simple way to collect hand dermal
exposure samples from the farmer and also from young children.
• The collection methodologies utilized for carpet dust and surface soil samples were simple to perform if track-
in is investigated.
ACKNOWLEDGEMENTS
We acknowledge Vincent F. Garry, M.D. for recruitment of participant farms. This research was funded by
the U.S. Environmental Protection Agency (Contract 68D10150). This paper has been submitted to EPA's peer and
administrative review, but no official endorsement should be inferred.
REFERENCES
1. A. E. Bond, G. G. Akland, R. G. Lewis et al., "A pilot study for measuring environmental exposures from
agricultural applications of pesticides: an overview", in Proceedings of the 1993 U.S. EPA/A&WMA
International Symposium on Measurement of Toxic and Related Air Pollutants," Ah- & Waste Management
Association, Pittsburgh, 1993 (hi press).
2. P. W. Geno, D. E. Camann, and K. Villalobos, "Analytical methods for assessing the exposure of farmers and
their families to pesticides" hi Proceedings of the 1993 U.S. EPA/A&WMA International Symposium on
Measurement of Toxic and Related Air Pollutants." Air & Waste Management Association, Pittsburgh, 1993
(in press).
847
-------
COMPARISON OF TRANSFER OF SURFACE CHLORPYRIFOS
RESIDUES FROM CARPET BY THREE DISLODGEABLE
RESIDUE METHODS
David E. Camann, H. Jac Harding, Susan R. Agrawal
Southwest Research Institute
P.O. Drawer 28510
San Antonio, Texas 78228-0510
Robert G. Lewis
U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina 27711
ABSTRACT
Experiments were performed to compare transfers of dried chlorpyrifos residues from carpet by the Dow
drag sled, the California cloth roller, and the Southwest Research Institute polyurethane foam (PUF) roller. On
plush nylon carpet, mean chlorpyrifos transfers were 4.5% by the cloth roller, 1.1% by the drag sled, and 0.65%
by the PUF roller. On level-loop polypropylene carpet, mean transfers were 2.5% by the cloth roller, 1.4% by
the drag sled, and 1.2% by the PUF roller. The cloth roller was found to be less suitable than the other methods
because its transfers exhibited greater variability and were altered by orientation of the roll relative to the lay of
the carpet fibers. Moistening the sampling media increased the transfer by the drag sled and the PUF roller, but
substantially increased the measurement variability of both methods.
INTRODUCTION
Dermal contact with residues of pesticides applied to carpets and subsequent skin absorption or ingestion
through hand-to-mouth activity are routes of human exposure which need better evaluation, especially for young
children. The Dow drag sled1, the California cloth roller2, and the Southwest Research Institute (SwRI)
polyurethane foam roller3 (SwRI invention disclosure #2061, patent pending) are dislodgeable residue sampling
methods which have recently been developed to estimate the transfer of a chemical from a contaminated surface
to the skin. This paper reports three experiments conducted to determine which of the three methods deployed
as currently used by the developer provides the more reproducible and facile transfer of chlorpyrifos residues from
carpet:
Exp. 1. Transfer comparison of the three methods using dry sampling media on new plush cut-pile
nylon carpet.
Exp. 2. Transfer comparison of the three methods using dry sampling media on new level-loop
polypropylene carpet
Exp. 3. Transfer comparison of the better two methods using both dry and moist sampling media on
new plush cut-pile nylon carpet
METHODS
Prior to each experiment, the specified virgin carpet and padding were installed in 1 Vj rooms (1 room
for Experiment 3) of an existing 42 ft x 10 ft 3-room trailer.
848
-------
Dislodgeable Residue Methods
Relevant characteristics of the dislodgeable residue methods are summarized and contrasted in Table I.
The California cloth roller was constructed and the method performed as described by Ross et al. (1991). A soap-
washed and precleaned dry 17 in. x 17 in. cloth of percale bedsheet was placed on the carpet and covered with
a sheet of plastic. A 2 ft long by 4 in. diameter sewer pipe, filled with 25 pounds of lead shot ballast and wrapped
in a sheet of high density PUF, was rolled forward and backward over the plastic/cloth/carpet sandwich ten times.
After the 20 passes, the percale cloth was picked up and analyzed.
The drag sled method was performed using the initial configuration described by Vaccaro and Cranston
(1990). Briefly, a precleaned dry 4 in. x 4 in. denim weave cloth supplied by B. Shurdut, Dow Chemical
Company, was attached beneath foil under a 3 in. x 3 in. plywood block and an 8-lb weight mounted (Figure 1).
The sled was dragged once over a 3 in. x 4 ft carpet strip at 6-8 cm/s. After the single pass, the denim cloth was
removed for analysis.
The original PUF roller sampler (Hsu et al., 1990) was used for Experiments 1 and 2. A precleaned dry
PUF ring (3 in. length, 3.5 in. OD, 1.62 in. ID) was secured on the 8 in. length x 2 in. OD cylindrical 7.2 Ib
stainless steel roller. The new (October 1992) model of the PUF roller sampler was used for Experiment 3. A
precleaned dry PUF ring was secured on the 3 in. length x 1.75 in. OD cylindrical 0.37 kg aluminum roller
(Figure 2). The PUF roller was rolled once over a 3 in. x 1.0 m carpet strip at 10 cm/s once in both directions.
After the two passes, the PUF ring was slit and removed from the roller for analysis.
In Experiment 3, PUF rings and denim cloth were used which had been moistened with deionized water.
A precleaned PUF ring was uniformly moistened with 5.0 ± 0.1 g of water hi the laboratory by spraying the ring
surface with an atomizer, compressing with a squeeze tool to obtain uniform water distribution, weighing and
sealing in a steel canister until use. The sampling surface of the denim cloth was moistened with 0.5 ± 0.1 g of
water from the atomizer and weighed just prior to mounting under the drag sled. When moistened at these levels,
the PUF ring and denim cloth were observed to produce equivalent moisture trails at method pressure on a glass
surface.
Broadcast Application of Chlorpyrifos and Ventilation While Drying
Broadcast application of Chlorpyrifos to test carpeting was conducted by a licensed pest control applicator
according to label instructions for flea control treatment. The formulated product, Dursban® L.O. (E.P.A.
RegistrationNo. 464-571), which contains 41.5% Chlorpyrifos (O,0-diethyl O-(3,5,6-trichloro-2-pyridyl)phosphoro-
thioate), was applied approximately 40 cm above the carpet as a 0.50% aqueous spray (40 mL/3.785 L water) with
a hand-held fan broadcast nozzle attached to an air pressurized tank. Application was accomplished in 2 to 3 min.
The trailer was ventilated for 2 h immediately after application. All windows were opened and window
air conditioning units were operated in fresh return air mode. During the first 30 min and last 15 min of the
ventilation period, both doors were opened and a box fan was operated outside the test room doorway to allow
maximum cross ventilation. Air conditioner units were returned to the usual recirculated air mode just prior to
sampling and remained on throughout the sampling period of Experiments 1 and 2.
Experimental Design
Adjacent samples using each compared dislodgeable residue method (with dry and moist media in
Experiment 3) and a deposition coupon (2 coupons in Experiment 3) were collected sequentially within a
rectangular block of treated carpet. Six replicate blocks were sampled in Experiments 1 and 2; 4 l/2 replicate
blocks comprised Experiment 3.
849
-------
Table I. Characteristics of dislodgeable residue methods.
Property
Sampling medium (material)
Surface of sampling medium
Contact motion
Face (instantaneous contact
California cloth roller
Percale bedsheet (50% cotton,
50% polyester)
Square (42.9 cm)2
Roll
440 cm2 = 42.9 cm x 10.2 cm
Dow drag sled
Denim weave cloth
(predominantly cotton)
Square (10.2 cm)2
Drag
58 cm2 = (7.6 cm)2
SwRI PUF roller
Polyurethane foam ring (polyether,
0.029 g/cm3)
Curved exterior of ring, (OD = 8.9 cm,
length = 7.6 cm)
Roll
38.6 cm2 = 7.6 cm x 5.1 cm
area pressed through sampling
medium)
Mass exerting pressure through 14.4 kg 3.46 kg
sampling medium
Pressure exerted through 2,300 Pa = (14.4 kg)(9.8 m/s2)/
sampling medium [(0.61 m)(0.10 m)]
Sampled carpet area 0.184 m2 = (0.429 m)2
Number of passes over sampled 20
carpet area
Sampling speed over carpet 0.23 m/s 0.07 m/s
3.25 kg;a 3.10 kgb
5,900 Pa = (3.46 kg)(9.8 m/s2)/ 8,300 Pa;a 8,000 Pab = (3.10 kg)
(0.076 m)2 (9.8 m/s2)/[(0.076 m)(0.05 m)]
0.093 m2 = 0.076 m x 1.22 m 0.076 m2 = 0.076 m x 1.0 m
1 2
0.10 m/s
a Original PUF roller sampler
b 1992 model of PUF roller sampler
-------
Weight
•g) .XjJ—Wooden Block
\s^*~~ Cotton Denim
Collection Media
Figure 1. Dow drag sled.
Figure 2. PUF roller sampling instrument
851
-------
Deposition coupons, consisting of absorbent alpha-cellulose pads (4 in. x 4 in.) backed with aluminum
foil were placed on the carpet prior to the chlorpyrifos application and picked up before the adjacent dislodgeable
residue samples from the block were collected. Residues measured on the coupon (pair in Experiment 3) gave
an estimate of the surface loading of residue remaining on adjacent carpeting during sampling in the block.
Field blanks of each method were obtained by sampling on the virgin carpet prior to the chlorpyrifos
application to assess contamination potential during sampling and handling. Deposition coupon(s) were placed
at the designated locations in each sampling block shortly before the application commenced. Field samples were
collected in each blocks upon label allowed re-entry, i.e., when the carpet was dry (which was operationally
defined as 2 hours after application, but checked by hand contact). The dislodgeable residue samples of a block
were collected from specified randomized locations in the block after the deposition coupon(s) were picked up.
All samples were collected hi one block before proceeding to the next block. Spikes of the precleaned
dislodgeable residue media (both dry and moist hi Experiment 3) and of a deposition coupon were made both
before and after the replicate block sample sets were collected; these field spikes were used to assess and adjust
for losses during transport, storage, and extraction.
Sample Analysis
All samples were Soxhlet-extracted with 6% ethyl ether/94% hexane; extraction commenced within
24 hours after sampling. In Experiment 3, the pair of deposition coupons from a block were extracted together.
Extracts were analyzed for chlorpyrifos by GC/ECD on two dissimilar columns and quantitated from the DB-5
column results.
Data Adjustment
Crude results (mg/sample) from each field sample were adjusted for contamination and extraction
inefficiency by subtracting the field blank result and dividing the difference by the mean recovery proportion of
the two field spikes for that method. The adjusted result was divided by the sampled carpet area (see Table I)
to determine the measured transfer rate (mg/m2 of carpet contacted) for dislodgeable residue samples and the
measured surface loading (mg/m2) for coupon samples.
RESULTS
Recovery of chlorpyrifos in field spikes of all sampling media is shown in Table II. Amounts in field
blanks were negligible in Experiments 1 and 3, but within two-to-three orders of magnitude of field samples in
Experiment 2, which was performed six days after Experiment 1.
Table II. Percent recovery of chlorpyrifos in field spikes of sampling media.
Field spikes, Standard
Sampling medium Use Moisture n Mean deviation
a-Cellulose pad Deposition coupon Dry 6 94.2 13.9
Denim cloth Drag sled Dry 5 102.8 11.0
Moist 2 96.6 1.6
Percale sheet Cloth roller Dry 4 96.2 14.0
PUFring PUF roller Dry 4 110.2 4.6
Moist 2 107.7 2.1
852
-------
The surface loading of chlorpytifos on treated carpets determined at the time of block sampling via
deposition coupons is presented in Table III. The chlorpyrifos loading was slightly lower and much more uniform
on the level-loop polypropylene carpet in Experiment 2 than on the plush cut-pile nylon carpets in Experiments 1
and 3.
Table in. Chlorpyrifos surface loading0 on treated carpets during dislodgeable residue sampling11.
Experi-
ment:
Date
1:8/4
2:8/10
3:11/24
Carpet type
Plush nylon
Level-loop
polypropylene
Plush nylon
Deposition
coupons
per
replicate
1
1
2
No. of
replicates
n
6
6
5
Chlorpyrifos loading3 (mg/m2)
Mean
154.1
120.9
183.3
Standard
deviation
94.3
14.0
30.6
Range
74-329
105-135
146-216
Loading
coefficient
of variation
0.611
0.116
0.167
a Loading, mg/m2 = [(deposition amount, mg - blank amount, mg)/(mean spike recovery)]/(coupon area,
m2)
The transfer rates of chlorpyrifos from treated carpet by the dislodgeable residue methods are summarized
in Table IV. Transfers with dry sampling media were highest for the cloth roller, intermediate for the drag sled,
and lowest for the PUF roller, both from plush nylon carpet and from level-loop polypropylene carpet. As shown
by the coefficient of variation, all three dislodgeable residue methods gave more repeatable performance on the
plush nylon carpet than on the level-loop polypropylene carpet. The cloth roller displayed more variation in
chlorpyrifos residue transfer using dry sampling media than did the drag sled and PUF roller, particularly on the
plush nylon carpet. Transfers obtained with the cloth roller differed substantially [8.1 ± 1.5 (n=4) vs. 4.5 ± 0.1
(n=2) for Experiment 1 mean ± std. dev.] for rolls oriented with/against vs. across the lay of the carpet fibers.
Transfers with the drag sled and PUF roller did not vary with the orientation of the drag/roll relative to the lay
of the carpet fibers. The additional transfer variability observed with the cloth roller is largely attributable to the
directional sampling effect.
Transfers using moist media were larger than transfers using dry media for both the drag sled and the PUF
roller. However the measurement variability of both methods increased substantially when moist media were used.
The percentage of the chlorpyrifos loading that was transferred by each method in each experiment is
presented in Table V. The mean percent transfer of chlorpyrifos residue with the cloth roller on level-loop
polypropylene carpet in Experiment 2 (2.5%) was only 55% of its transfer (4.5%) on plush cut-pile nylon carpet
in Experiment 1. In contrast, percent transfers with the drag sled and PUF roller were slightly larger on the level-
loop polypropylene carpet.
DISCUSSION
This research was performed to allow intercomparison of dislodgeable residue transfers obtained by
recently-developed methods in different studies conducted to support registration of pesticides used in the home.
Consequently the cloth roller, drag sled, and PUF roller methods were evaluated as performed by their developers
(Table I), despite differences in properties which are likely to affect transfer including sampling pressure and
speed, sampled carpet area, and number of passes over the area. The methods were compared upon label-allowed
re-entry after application of the pesticide product at the maximum permitted rate, as in pesticide registration
studies.
Serious sampling difficulties were experienced in use of the cloth roller method, but not the other methods
(Table VI). In particular, the sampling cloth tended to bind and shift from its original position with successive
853
-------
passes of the roller, so that the actual carpet area contacted by the cloth differed in magnitude and location from
the nominal sampled area in an unknown and uncontrollable manner. The variation in transfer with orientation
of the passes relative to the lay of the carpet fibers, which was observed only in use of the cloth roller, may relate
to the propensity of the cloth to bind and shift during sampling. In situations where the carpet has a distinct lay
of pile, especially in areas of traffic, the direction of the drag sled may also affect dislodgeable transfer.
Performance of dislodgeable residue sampling using moistened media may provide a more realistic
simulation of transfers experienced by young children who exhibit frequent hand-to-mouth behavior. The goal
for moistness of the sampling media is problematic, since the probability distribution of moisture on children's
hands is broad and decreases with age. Moistening does increase dislodgeable residue transfer, as expected.
However, moistening so exacerbates transfer measurement variability that many more measurements are needed
to obtain the precision in mean transfer achieved with dry sampling media.
Mean percent chlorpyrifos transfer obtained from plush cut-pile nylon carpeting declined markedly from
Experiment 1 to Experiment 3, both with the drag sled (1.12% vs. 0.44%) and the PUF roller (0.71% vs. 0.26%).
The temperature of both the ambient air (Table V) and the room air was higher during sampling in Experiment 1
(performed in August) than in Experiment 3 (in November).
Additional experiments are underway to determine the recovery of residues from the human hand with a
handwipe method, and to determine the effect of sampling variables including air temperature on transfers with
the drag sled and PUF roller. Future experiments will compare transfers by human hand presses to the
dislodgeable residue methods.
Table IV. Comparison of dislodgeable residue method transfer" of chlorpyrifos residue by carpet type and
media moisture.
Experi-
ment Carpet type
Dislodge- Sampling No. of
able residue media replicates
method moistness n
Chlorpyrifos transfer*
(mg/m2)
Standard
Mean deviation Range variation
Transfer
coefficient
of
Plush nylon
Cloth roller Dry
Drag sled Dry
PUF roller Dry
6 6.92 2.20 4.4-10.0 0.318
6 1.73 0.29 1.4-2.2 0.168
6 1.10 0.26 0.8-1.5 0.238
2
3
Level-loop
polypropylene
Plush nylon
Cloth roller
Drag sled
PUF roller
Drag sled
PUF roller
Dry
Dry
Dry
Dry
Moist
Dry
Moist
6
6
6
4
5
4
5
3.00
1.66
1.43
0.81
1.36
0.48
3.91
1.34
0.70
0.56
0.34
0.98
0.07
2.02
1.6-4.9
1.1-2.9
0.9-2.3
0.5-1.3
0.6-2.9
0.4-0.6
0.6-5.4
0.446
0.420
0.393
0.414
0.718
0.141
0.517
a Transfer, mg/m = [(dislodged amount, mg - blank amount, mg)/(mean spike recovery)]/(sampled area,
m2)
854
-------
Table V. Percent2 of cMorpyrifos loading transferred by dislodgeable residue method and sampling
conditions.
Carpet type
Ambient air range
during sampling
Moistness Relative
of samp- Temp humidity
Experiment ling media (°Q (%)
Percent transfer* (mean ± std dev)
Cloth
roller Drag sled PUF roller
Plush nylon
Plush nylon
Level-loop
polypropylene
3
1
2
Moist
Dry
Dry
Dry
19-21
28-32
29-32
24-37
55-67
50-67
0.74±0.53
0.44±0.18
4.49±1.42 1.12±0.19
2.48±1.11 1.37±'0.58
2.13±UO
0.26±0.04
0.71±0.17
1.18±0.46
a Percent transfer = 100 x (dislodgeable residue method transfer, mg/m2)/(surface loading, mg/m )
Table VI. Observations from field use of dislodgeable residue methods.
Strengths
California Cloth Roller
• Simple in design
• Inexpensive to build from available
materials
Dow Drag Sled
• Simple in design
• Inexpensive to build from available
materials
• Simple to use
Weaknesses
Sampling cloth tends to bind and shift from
original position
Plastic bag cover may adhere to PUF sleeve
on roller from static
Difficult to operate due to mass of roller
Operator must contact treated surface
Susceptible to added pressure from operator
Transfer affected by roll orientation relative
to lay of carpet fibers
Drag contact unlike most skin contact with
carpet
Drag contact is potentially directional relative
to lay of carpet fibers
SwRI PUF Roller
• Consistent use across operators due to
few variables
• Relatively simple to use
• Foam roller contact is more like skin
contact
Expensive to build or purchase
855
-------
CONCLUSIONS
1. The California cloth roller is less practical and more variable than the drag sled and PUF roller methods.
2. Transfers as currently performed by the developer were lowest for the PUF roller, intermediate for the
drag sled, and highest for the cloth roller when using dry sampling media on two types of carpet
3. An interaction in transfer percentage exists between carpet type and dislodgeable residue method.
Transfers with the cloth roller are not predictive of transfers with the other methods across different types
of carpet.
4. Transfers with moist media are larger and much more variable than transfers with dry media for both the
PUF roller and the drag sled.
ACKNOWLEDGEMENT
This research was funded by the U.S. Environmental Protection Agency (Contract 68-DO-0007) under
subcontract from Battelle. This paper has received EPA's peer and administrative review, but no official
endorsement should be inferred.
REFERENCES
1. J. R. Vaccaro and R. J. Cranston. "Evaluation of dislodgeable residues and absorbed doses of chlorpyrifos
following indoor broadcast applications of chlorpyrifos-based emulsifiable concentrate", Internal Report,
Dow Chemical Co., Midland, MI, 1990.
2. J. Ross, H. R. Fong, T. Thongsinthusak, et al. "Measured potential dermal transfer of surface pesticide
residue generated from indoor fogger use: using the CDFA roller method," Chemosphere 22: 297-84
(1991).
3. J. P. Hsu, D. E. Camann, H. J. Schattenberg, et al. "New dermal exposure sampling technique," in:
Proceedings of the 1990 U.S. EPA/A&WMA International Symposium on Measurement of Toxic and
Related Air Pollutants. VIP-17, Air and Waste Management Association, Pittsburgh, PA, 1990, pp 489-97.
856
-------
APPLICATIONS OF A CONTINUOUS GAS CHROMATOGRAPH:
AREA MONITORING, SCRUBBERS, AND STACKS
Robert C. Petersen, Ph.D.
TRC Environmental Corporation
Boott Mills South, Foot of John Street
Lowell, Massachusetts 01852
ABSTRACT
The marriage of the computer and gas chromatograph has produced a
very powerful tool in the field of air toxics analysis. Continuous
around-the-clock monitoring is now possible, providing important
data to confirm EPA or OSHA compliance, in regard to the
concentration of hazardous air pollutants. With the aid of the
computer, data acquisition options are limitless. One may wish to
do trend reporting, activate alarms, archive data, or all three.
In the plant, the gas chromatograph can cycle between several
different work areas, in order to assess chemical exposure levels.
Data over an extended period of time can be used to show that the
work place is complying with OSHA requirements.
The area of air pollution control is also well served by the gas
chromatograph. Gas concentrations at the inlet and outlet of a
scrubber can be measured to determine scrubber efficiency.
Finally, the gas chromatograph can be used to monitor stack
emissions, in order to ensure that the EPA requirements, stemming
from the Clean Air Act Amendments, are satisfied.
For air pollutants which do not lend themselves to easy analysis by
conventional continuous emissions monitors, the computerized gas
chromatograph is a powerful tool for the environmental scientist or
engineer, and the use of this instrument will surely increase as
the 21st century approaches.
INTRODUCTION IS GAS CHROMATOGRAPHY BEST FOR YOUR APPLICATION?
This paper is divided into two parts. A discussion of choosing
the right gas chromatograph (GC) and its associated equipment, will
be followed by a presentation of different types of industrial
applicat ions.
First though, you need to establish whether or not your air
monitoring needs require a gas chromatograph or some other type of
gas monitor. Continuous Emission Monitors (CEMs) are readily
available for he most common pollutants (Nitrogen Oxides, Sulfur
Dioxide, Hydrogen Sulfide, Carbon Monoxide, Oxygen, etc.). Other
gas monitors are available to measure specific compounds in indoor
air, with relatively good selectivity and minimum interference from
other components. The detectors on these instruments vary. Some
857
-------
examples are electrochemical cell, semiconductor, photoionization.
infrared, ultraviolet, flame ionization, etc.
However, there will be cases when a conventional gas monitor will
not suit your needs. There will be instances when a monitor is not
available for the specific compound of interest, but a situation
more commonly encountered concerns interferences from other
components. A conventional gas monitor will do you no good if it
is not able to discriminate the analyte gas (the compound of
interest) from the other gases in the sample.
The GC has a superior ability to separate the analyte from the
other gases in the sample, in order to give an accurate assessment
of concentration. For example, many volatile organic compounds
(VOCs) emitted as a result of painting, printing, and lamination
processes can be analyzed as a group by a total hydrocarbon
analyzer or a similar device. However should speciation be
required, the gas chromatograph is definitely the method of choice.
It is able to provide concentration information for several
components during one analysis.
The GC is clearly more expensive and sophisticated than many
conventional monitors. In many cases GC is overkill and a
conventional gas monitor is the instrument of choice. However,
when deemed appropriate, gas chromatography can solve a lot of
analysis problems not addressed by CEMs and other similar
analyzers.
Let us now assume that your application will require the use of a
GC. The following section will aid in equipment selection.
I. CHOOSING THE CORRECT GC INSTRUMENTATION
The market is full of vendors with automated GC equipment. Brand
names will not be discussed, but instead, the purpose of this
section is to help identify specific needs for your application,
hopefully making the decision of what instrumentation to buy,
easier for you.
A. The Analyte(s), and the Sample Matrix
You first need to establish what compound or compounds will be
quantified, and at what concentration levels they are expected to
be found.
Now list all the other gases which will be present in the sample,
and their expected concentrations. In most cases the sample matrix
is air, and the nitrogen and oxygen will not interfere with the
analysis. However, in certain cases, compounds of similar
molecular structure to the analyte, may cause problems with the
analysis, if they are not first identified and dealt with. The
858
-------
temperature and humidity of the sample matrix are also important to
note.
B. The Sample Loop, GC Column, and Detector
The already established expected concentration levels of the
analyte(s) will determine the size of the sample loop used to
inject the sample matrix on to the GC column.
The chemical nature of the analyte(s) will dictate what GC column,
detector, and GC operating conditions (inject, oven, and detector
temperatures, valve timing, etc.) will be used. Techniques
involving the use of pre-column backflushing, and column switching
can also be added to the system if warranted.
C. The Sampling System
Gas samples from different areas are drawn to the GC through
separate tubes by means of a sample pump. Should long tubing runs
be anticipated, a larger pump may be required to achieve sufficient,
draw. Note that the GC will automatically switch from one sample
point to the next for analysis, continuously, 24 hours a day.
Should condensation of the analyte be a possibility (particularly
at high concentrations of low boiling analytes), heat traced lines
all the way to the GC column should, be utilized. In addition, a
particulate filter at the end of each sample line, will prevent
contamination from penetrating the tubing or the GC.
D. Calibration
Proper calibration of the GC is crucial to ensure the collection of
accurate data. Calibrations are normally done automatically at set
time periods. Calibration gas from a compressed cylinder must be
introduced to the GC in a an identical way (same pressures, flows,
etc.) the sample from the sampling lines is. Furthermore, the
concentration of analyte in the calibration gas should be
approximately the same as found in the sampling lines.
E. Computer Operation
Combining the flexibility of a computer with a high quality gas
chromatographic analysis, provides a very powerful tool in the
field of air toxics analysis. The computer allows the GC to run
unattended for long periods of time, performing continuous
monitoring and periodic automatic calibrations.
The computer will not only control the GC, but will accept data
from it, giving you many operational options. For instance, alarms
can be triggered at set concentration levels, trend reporting in
graphical form can be presented (see Figures 1 and 2), and data can
be stored on disk for future use.
859
-------
II. INDUSTRIAL APPLICATIONS OF THE CONTINUOUS GC
Assume now that you have a functional continuous gas chromatograph
with sample lines installed. The purpose of this section is to
present different applications of the GC, which will benefit
operations in an industrial environment.
A. Area Monitoring
Sample points can be located in work areas in order to assess
exposure to the analyte compound. The data generated can be
averaged over an 8 hour period to show that the worker is not
exposed to levels above those set by OSHA. Figure 1 presents an
example of data collected and displayed in graphical form, obtained
from a continuous GC monitoring a particular work area.
Figure 1
C52 Pit Area Concen. Over 2 Months
' I 9 ill il3n4M6
11 14 16 17 19 21 23 25 27 29 31 2 4 6 8 10 12 14 15
Day of the Month, beginning Jan. 1993
If the end of the sample line will be exposed to water (from even
an occasional water hose down), it should be protected in such a
way so that water will never get sucked into the tubing.
B. Scrubbers
If the analyte is being scrubbed out of the air in a scrubber, the
GC is a nice way to determine scrubber efficiency, by measuring the
inlet and outlet concentrations.
860
-------
A word of caution is warranted about sample conditioning. Often
the scrubber air may be very humid, hot, or dirty, and some type of
sample conditioning will be required prior to introduction into the
GC. Make sure the conditioner does not scrub out some of the
analyte, preventing the GC from seeing the true concentration.
C. Stacks
This is one of the most popular applications of the continuous GC.
The Clean Air Act Amendments require strict control of air
emissions, and the GC is a tool used to ensure EPA compliance. The
previous discussion of sample conditioning applies equally for
stacks, and proper precautions should be taken. Figure 2 presents
an example of data collected and displayed in graphical form,
obtained from a continuous GC monitoring stack emissions.
Figure 2
ich Concentration Over 3 Days
r of the Day. begi
Should flow rate data be available from a continuous flow monitor
in the stack, a combination of the GC and flow data can give a
continuous reading of emissions rates (Ib/hr) of the given analyte.
CONCLUSIONS
Continuous gas monitoring can be done by many computerized
analytical methods, one of them being gas chromatography. After
review of your specific monitoring needs, a properly equipped
computer operated GC may be the best alternative in providing the
861
-------
best analytical data available for your analysis.
Whether it be area monitoring to confirm OSHA compliance, or
scrubber/stack monitoring to satisfy EPA requirements of the Clean
Air Act Amendments, the computerized continuous gas chromatograph
provides the user valuable information for all of these industrial
applicat ions.
BIBLIOGRAPHY
R. Annino, R. Villalobos, Process Gas Chromatographv Fundamentals
and Applications, Instrument Society of America, 1992.
862
-------
INTERNAL STANDARD IMPLEMENTATION
IN AIR MONITORING
Sharon P. Reiss
Wendy L. Ballard
GRASEBY NUTECH-RTL
4022 Stirrup Creek Drive
Suite 325
Durham, NC 27703-9000
Abstract
Internal standards are routinely employed in many analytical methods to correct for
fluctuations in system performance. This Poster Session will first discuss the loop injection
technique for introducing internal standards to an air matrix during cryogenic concentration.
Secondly, the advantages of using internal standards in the analyses of volatile organics in air
will be demonstrated by comparative analyses with external standardization.
Introduction
In many environmental laboratories, analytical instrumentation is operated around the
clock to utilize equipment effectively. GC/MS systems in constant use may exhibit changes in
sensitivity as the chromatographic column degrades, the ion source becomes dirty and the
analyzer electronics age. Any changes in response can be corrected by incorporating an
internal standard into the analyses.
The main advantage of internal standards is that the method of quantitation relies on
relative response factors rather than absolute area counts. Relative response factors are
calculated for each analyte from calibration data as follows:
863
-------
RRFX = (Ax) (Cis)
(Ais)
RRFX = Relative response factor for Analyte X
Ax = Area of quantitation ion for Analyte X
Cx = Concentration of Analyte X (units y)
Ajs = Area of quantitation for Internal Standard
C;s = Concentration of Internal Standard (units y)
Internal Standard Introduction
Before incorporating internal standards into the analyses of air samples, the technique
for introduction was evaluated. Internal standard addition is performed on the Nutech 3550A
cryogenic concentrator by loop injection. The measured volume is independent of the
standard vessel pressure, since the loop is allowed to come to atmospheric pressure before the
loop contents are transferred to the cryotrap. A precision study of four compounds
commonly used as internal standards in other volatile analyses was performed. This data was
generated using the Nutech 3550A cryogenic concentrator interfaced with Hewlett-Packard
5890 SeriesII GC/FID. Table I lists raw data area counts for 18 consecutive analyses of the
internal standards. Relative standard deviations (RSDs) for this analysis clearly validate the
high level of precision (>98%) associated with this particular configuration.
Comparative Analysis
To illustrate the effectiveness of internal standard quantitation, one set of data files is
examined using both internal standard and external standard methods. Data was generated
on the Nutech 3550A cryogenic concentrator interfaced with the Hewlett-Packard 5971
GC/MSD. A three point calibration of eight volatile organic compounds (VOCs) in an air
matrix was run in triplicate. Bromofluorobenzene was selected as the internal standard.
Examination of the data files reveals a trend of decreasing area counts that range between five
to ten percent during the twelve hour course of data acquisition.
864
-------
Table n lists results from external standardization. Reproducibility is good with an average
precision of 92% and percent recovery greater than 94%. Table III considers the same
triplicate analysis as does Table II, with the incorporation of an internal standard into the
quantitation. Percent recovery of analyses from actual concentrations averages 97.3%, a gain
of 2.4% over external standardization. Overall precision is two times higher using internal
standards as evidenced by the average %RSD of 4.08%.
Conclusion
The data presented here shows the advantages of using internal standards for air
analysis. Internal standard quantitation shows two times greater precision than does external
standard quantitation. Internal standards allow for better comparison and quantitaion of data
despite variances in the systems performance. Loop injection proves to be a very effective and
consistent means for internal standard introduction.
865
-------
TABLE 1: INTERNAL STANDARD LOOP INJECTION PRECISION BY GC/FID.
IS #3
001F0101
001F0102
001F0103
001F0104
001F0105
001F0106
001F0107
001F0108
001F0109
001F0110
001F0111
001F0112
001F0113
001F0114
001F0115
001F0116
001F0117
001F0118
64906
64812
64884
63824
63916
63943
64432
64034
64556
63969
64624
64509
64531
64307
64076
64838
64838
65533
419001
420052
419779
413789
413275
412562
416540
413561
417270
414169
417617
417753
416351
415812
420588
420297
418463
422836
464096
464243
463047
455360
452873
451748
457305
453972
458086
455505
459448
460215
456141
457812
454389
464005
461594
465264
197972
197323
196659
192026
189709
189047
191512
190399
192392
191571
193319
194187
190963
193227
191108
196396
194884
195986
AVERAGE
StdDev(n-l)
%RSD
64474
456.43
0.71
417206
2940.65
0.70
458617
4306.76
0.94
193260
2740.41
1.42
IS#1 Bromochloromethane
is #2 1,4-Difluorobenzene
is #3 Chlorobenzene-d5
BFB Bromofluorobenzene
866
-------
Table 3: Internal Standard Quantitation.
# Compound
1 Vinyl chloride
2 1,1-Dichloroethene
3 1,1-Dichloroethane
4 cis-1,2-Dichloroethene
5 1,2-Dichloroethane
6 1,1,1-Trichloroethane
7 Dibromoethane
8 Tetrachloroethene
Std Level
ppbv
9.30
11.40
10.50
10.50
10.10
10.40
10.10
9.80
Calibl
ppbv
8.98
14.36
10.99
10.86
10.40
10.86
10.33
10.42
Calib2
ppbv
8.70
12.70
9.90
9.80
9.08
9.62
9.05
8.87
CalibS
ppbv
8.80
10.69
10.10
9.89
9.57
9.88
9.38
8.91
Average
ppbv
8.83
12.58
10.33
10.18
9.68
10.12
9.59
9.40
% Recovery
94.91
110.38
98.38
96.98
95.87
97.31
94.92
95.92
Std Dev
0.14
1.84
0.58
0.59
0.67
0.65
0.66
0.88
%RSD
1.61
14.60
5.62
5.77
6.89
6.46
6.93
9.40
1 Vinyl chloride
2 1,1-Dichloroethene
3 1,1-Dichloroethane
4 cis-1,2-Dichloroethene
5 1,2-Dichloroethane
6 1,1,1-Trichloroethane
7 Dibromoethane
8 Tetrachloroethene
46.50
57.00
52.50
52.50
50.50
52.00
50.50
49.00
47.43
48.39
51.28
51.31
49.74
50.72
50.01
47.87
46.16
62.17
49.80
49.90
47.33
48.74
47.57
45.42
44.76
52.64
49.90
49.77
48.04
49.19
48.26
45.51
46.12
54.40
50.33
50.33
48.37
49.55
48.61
46.27
99.18
95.44
95.86
95.86
95.78
95.29
96.26
94.42
1.34
7.06
0.83
0.85
1.24
1.04
1.26
1.39
2.90
12.97
1.64
1.70
2.56
2.09
2.59
3.00
1 Vinyl chloride 93.00
21,1-Dichloroethene 114.00
31,1-Dichloroethane 105.00
4 cis-1,2-Dichloroethene 105.00
51,2-Dichloroethane 101.00
61,1,1-Trichloroethane 104.00
7 Dibromoethane 101.00
8 Tetrachloroethene 98.00
Average
94.97 93.42 89.53
101.62 113.13 109.41
102.57 102.57 101.91
103.80 103.28 102.18
99.47 99.44 100.60
101.96 101.78 102.17
99.65 100.15 100.07
94.08 93.73 93.25
92.64
108.05
102.35
103.09
99.84
101.97
99.96
93.69
99.61
94.78
97.48
98.18
98.85
98.05
98.97
95.60
2.80
5.87
0.38
0.83
0.66
0.20
0.27
0.42
3.03
5.44
0.37
0.80
0.66
0.19
0.27
0.44
97.26
4.08
867
-------
Table 2: External Standard Quantitation.
Compound
1 Vinyl chloride
2 1,1-Dichloroethene
3 1,1-Dichloroethane
4 cis-1,2-Dichloroethene
5 1,2-Dichloroethane
6 1,1,1-Trichloroethane
7 Dibromoethane
8 Tetrachloroethene
1 Vinyl chloride
2 1,1-Dichloroethene
3 1,1-Dichloroethane
4 cis-1,2-Dichloroethene
5 1,2-Dichloroethane
6 1,1,1-Trichloroethane
7 Dibromoethane
8 Tetrachloroethene
1 Vinyl chloride
2 1,1-Dichloroethene
3 1,1-Dichloroethane
4 cis-1,2-Dichloroethene
5 1,2-Dichloroethane
6 1,1,1-Trichloroethane
7 Dibromoethane
8 Tetrachloroethene
Average
Std Level
ppbv
9.30
11.40
10.50
10.50
10.10
10.40
10.10
9.80
46.50
57.00
52.50
52.50
50.50
52.00
50.50
49.00
93.00
114.00
105.00
105.00
101.00
104.00
101.00
98.00
Calibl
ppbv
8.82
14.13
10.78
10.66
10.22
10.65
10.14
10.22
47.60
48.58
51.43
51.47
49.87
50.89
50.19
48.06
96.07
103.59
104.30
105.50
101.09
103.69
101.21
95.65
Calib2
PDbv
8.93
13.07
10.08
10.07
9.32
9.88
9.28
9.11
47.18
63.78
50.87
50.96
48.33
49.81
48.62
46.47
92.38
112.57
101.93
102.67
98.64
101.05
99.41
93.11
CalibS
PPbv
8.17
9.95
9.38
9.19
8.87
9.17
8.70
8.28
41.09
48.40
45.78
45.72
44.08
45.19
44.31
41.79
79.75
97.77
90.82
90.92
89.49
90.93
89.02
82.97
Average
8.64
12.38
10.08
9.97
9.47
9.90
9.37
9.20
45.29
53.59
49.36
49.38
47.43
48.63
47.71
45.44
89.40
104.64
99.02
99.70
96.41
98.56
96.55
90.58
% Recovery
92.90
108.63
96.00
94.98
93.76
95.19
92.81
93.91
97.40
94.01
94.02
94.06
93.91
93.52
94.47
92.73
96.13
91.79
94.30
94.95
95.45
94.77
95.59
92.43
Std Dev
0.41
2.17
0.70
0.74
0.69
0.74
0.72
0.97
3.64
8.83
3.11
3.18
3.00
3.03
3.04
3.26
8.56
7.46
7.20
7.73
6.11
6.74
6.58
6.71
%RSD
4.75
17.55
6.94
7.42
7.26
7.46
7.73
10.58
8.04
16.47
6.31
6.45
6.32
6.23
6.38
7.17
9.57
7.13
7.27
7.75
6.34
6.83
6.82
7.41
94.90
8.01
868
-------
ASPECTS OF DATA MANAGEMENT FOR AN
LDAR PROGRAM
Ginger Darnell
International Technology Corporation
11499 Chester Road
Cincinnati, Ohio
Kelly Trilk
International Technology Corporation
11499 Chester Road
Cincinnati, Ohio
ABSTRACT
The Clean Air Act Amendments (CAAA) of 1990 that apply to hazardous air pollutants have
produced a set of regulations governing equipment leaks. These regulations were developed through
negotiations among environmental groups, state and local agencies, industry, and the U.S. Environmental
Protection Agency (EPA). Details of these requirements were proposed in the Federal Register in
November of 1992, and proposed in Hazardous Organic National Emission Standards for Hazardous Air
Pollutants (HON). Because some states have decided to adopt these regulations before they are
mandated by EPA, industry must be prepared to meet the requirements of these regulations.
One of the major outgrowths of these regulations is the need for electronic collection and
management of the large amounts of information required to run a successful Leak Detection And Repair
(LDAR) program. In developing its own LDAR computer program, International Technology
Corporation (IT) has explored many possible methods for gathering field data using bar code readers,
punched tags, and data loggers. IT also has investigated the use of personal computers to manage the
information.
Timely reporting to meet the requirements of the regulations has also become a key factor in the
operation of an LDAR program. IT has worked with many facilities to supply the scheduling, field
operation, and data management capabilities needed to comply with these new regulations. This paper
will present the operation of a typical LDAR program's data Management activities as well as some
program options.
INTRODUCTION
The Clean Air Act Amendments (CAAA) of 1990 that apply to hazardous air pollutants have
produced a set of regulations governing equipment leaks. These regulations were developed through
negotiations among environmental groups, state and local agencies, industry, and the U.S. Environmental
Protection Agency (EPA). Details of these requirements were proposed in the Federal Register in
November of 1992, and proposed in the HON. These regulations impose a considerable amount of
recordkeeping and reporting requirements in regard to the Leak Detection And Repair (LDAR) program
that must be performed by the plants that fall under the regulations. To deal with the amount of data
that needs to be kept on site, a well-planned and extensive LDAR database is required. Using a
computer program to help manage the vast amount of data, as well as the reporting and scheduling, has
become necessary to allow the LDAR process to be conducted more efficiently by fewer people. IT
has performed many LDAR jobs using a computer program specifically designed to manage LDAR
projects. This paper will describe a typical LDAR project and address the major decisions that make
869
-------
the difference between the success or failure of an LDAR project.
INITIAL TAGGING AND DATA COLLECTION
The initial step in performing an LDAR program is to identify and tag all of the components in a
plant that are covered by the regulations. Prior to the physical tagging of the plant, several issues must
be addressed. These issues include: what type of tagging and data collection scheme to use, what data
are to be collected, and when will the initial data be collected. The initial data collection, phase is the
most time-consuming process of all phases in an LDAR program, and requires the most accuracy. For
this reason, the data must be collected quickly and accurately.
Tagging
Several options are available for physically identifying the components covered by an LDAR
program. The scheme that we use most often to identify components consists of assigning each
component a unique six-digit number, called a Tag ID. A Tag ID needs to be attached to each
component in the plant so that the component can be identified without the P&I drawings. There are
several different ways that this can be accomplished: 1) The Tag ID can be converted to a bar code and
the bar code attached to the component; 2) The Tag ID can be represented using a punch card code
and the punch card attached to the component; and 3) The Tag ID can be stamped into a stainless steel
tag and the tag affixed to the component. The most commonly used material to affix a tag to a source
is stainless steel wire. The cost is minimal and if properly twisted, it will not be a hazard to plant
personnel. Another option is a nylon or plastic tie. Although these types of ties can be used in some
cases, the majority of tags are placed in environments that will quickly dissolve the ties or make them
brittle. The third type of tie is a lead "head" with a twisted stainless steel wire attached. The wire is
threaded through the lead, which is then crimped on the wire to secure the tag. The extra cost of this
type of tie can be a concern. In deciding what type of tag to use, we need to evaluate the advantages
and disadvantages of each method. We look for a tagging method that is economical, accurate, safe,
speeds up data acquisition, and will survive in a hostile environment.
Bar codes will allow us to identify each component quickly and accurately. However, special
laminates and bases paper are used to prevent deterioration of the tag and fading of the printed code.
These laminates and bases can prove to be quite expensive when applied to tens of thousands of
components. Bar codes also require a special bar code reader to interpret the bar code. To accomplish
this, the bar code reader is connected to a data logger, which is connected to the Organic Vapor Analyzer
(OVA) to record the screening value. The problem with this setup is twofold. One, while the data
logger and bar code reader may be intrinsically safe, when you connect the two together the connection
between them is not. Second, it is dangerous for the field personnel to carry an OVA, data logger, and
bar code reader while climbing a ladder to reach a high source.
Punch tags are small, 2-inch by 4-inch, metal tags that have holes punched in them that represent
a Tag ID. To read a Tag ID, the punch tag must be slid into a reader that determines what the Tag
ID is based on the pattern of the holes. This option, like the bar codes, allows the Tag ID to be read
quickly and accurately. The tags can be purchased relatively inexpensively and are durable enough to
survive in a hostile environment. The problem with this method is that the field person would be
required to carry the OVA, data logger, and punch tag reader. Also, the punch tags must be inserted into
the reader, a two-handed process that can be very difficult for the field person, given the amount of
equipment being carried. Once again there is the question of the connections being intrinsically safe.
Stainless steel tags with the Tag ID punched into them are cheap and durable. This option can be
used in conjunction with a data logger hooked into an OVA. The connection may be intrinsically safe,
and the field person is only required to carry two pieces of equipment. The field person can enter the
Tag ID that is on the tag into the data logger and then take the screening value. Although this method
does not provide the accuracy that the other two methods do, it can reduce data collection time because
870
-------
physical contact with the tag is not required. The data logger can be programmed with the Tag ID's
that are to be screened as well as check to make sure that a valid Tag ID has been entered.
Most data are currently collected on field sheets. Field sheets require that the field person write all
the component and screening information on a form. This option, unlike the above options, does not
incorporate electronic data collection and is time consuming and inaccurate.
Data Collection
After determining how to tag the components in the field, we have to decide what data to initially
collect for our LDAR program. According to the HON, the following information must be kept at each
plant:
o ID numbers of equipment subject to this regulation and a site layout showing the
relative location of the equipment
o a table listing the monitoring frequency of each item of equipment
o H a list of ID numbers of compressors that have been elected to operate at a reading
of 500 ppm or less with the signature of the owner or operator
o ID numbers of pressure-relief devices in gas/vapor service
o date and result of each compliance test, the background reading, and the reading
taken at each piece of equipment
o ID numbers of equipment in vacuum service
o ID numbers of instrumentation system
o ID numbers of equipment in VHAP service for less than 300 hours per year
o list of connectors disturbed since the last monitoring period as well as the date and
results of the follow-up monitoring
o list of reconfigured equipment since last monitoring period in batch process units
o list of valves and connectors removed from or added to the process unit
o documentation of process stream composition
o identification of screwed connectors
o identification of welded connectors, the date of the weld, and the date of monitoring
The extensive amount of information that must be kept available at a plant site requires a computer
program capable of handling the large amounts of data. The Leak Detection And Repair Management
System (LDARMS) that we use for LDAR projects requires a minimum set of data to run. This
minimum data set is the information that we want to collect in the field. These data include: Tag ID,
process unit, component type, manufacturer, service, location, chemical stream, and if it is an exempt
component. The LDARMS program also keeps track of a additional component information which can
be entered at a later date.
COLLECTING THE DATA
Once we determine how we want to tag a plant and what data we want to collect, the next step is
to physically do the tagging and enter the data into our LDARMS program. To collect the data, we
normally use two-person teams: one to tag the component and one to collect the initial data and do the
initial screening. The other option is to use only one person to go through the plant and tag all the
components and collect the component information. This person will then go through the plant a second
time to do the screening. Doing this increases the risk that tags will be missed during the screening
process. After a days worth of data have been collected, the information is taken back to the office
and entered into the LDARMS program. There are two methods for entering the data into the database.
If the data were collected using field sheets, a data clerk must go through all of the field sheets and
manually enter each record into the database. This is a time-consuming and inaccurate process. A more
871
-------
accurate and efficient way to enter the data is to use a data logger to collect the data, and then
electronically transfer the data into the LDARMS program. By using this method, data-entry time can
be reduced from two to three days to five minutes. After the data has been entered into LDARMS, it
will be printed out on paper and check-printed for data entry errors. The data are then ready to be
processed and output in report form.
REPORTING REQUIREMENTS
The reporting requirements for meeting the new HON regulation are quite extensive. Each
connection (valve, flange, etc.) that is in service for over 300 hours/year using chemicals regulated by
the HON will now have to be identified and reported as such to the EPA in an initial statement
submitted within 90 calendar days from the applicability dates. The initial statement will contain:
° process unit identification
° numbers of each equipment types
u method of compliance
° schedule for each phase of the requirements.
The plant technician can no longer go out and check for leaks and 'say' that each connection has
been tested. It now must be documented that each connection has been checked and a value recorded
for each identification number. This information must be reported to the local agency or EPA on a
semiannual basis. The semiannual report will include:
o process unit identification
o frequency of monitoring
o provisions of subpart implemented
For every component type (valves, connectors, pumps, etc.) the following must be included in each
monitoring period:
° number of leakers detected
o percent leaking
° total number monitored
o number not repaired
In addition, an explanation of any repair delays, process unit shutdowns including dates and
durations, changes in processes, and performance test results, or any change in monitoring frequency
must also be included. If leaks are detected during routine screening, the instrument and operator
identification, as well as the equipment ID number, must be recorded. The dates of each attempt at
repairing the equipment must be recorded including the method used to repair the component. The
component rescreening reading must also be documented. If there is a repair delay, the reason for the
delay must be kept on file with the owner or operator's signature. The expected date of repair as well
as the date of successful repair will be recorded. Any process unit shutdowns that occur while the
equipment is unrepaired also must be recorded.
After each screening period, a report is generated that contains all the information gathered during
that period. This report is then kept at the plant site. Without electronic storage of this information, the
data collected could easily take up volumes of paper and shelves of space. The screening data must be
kept for a period of two years, which also increases the amount of space required for information storage.
In all probability there will be civil penalties for not complying with the requirements of the
regulations. Penalties may be assessed for not filing an initial report, or filing an incomplete report, late
872
-------
submission of any report, incomplete records, failure to maintain records for an adequate length of time,
and failure to repair leaks within the specified time. The emphasis on timely submission of reports as
well as their completeness makes the handling of massive amounts of information almost impossible
without the help of some type of computer program. In response to this demand, several programs have
been developed in the past two years. The key to our program is its flexibility and the ease with which
the data can be sorted and checked.
ROUTINE MAINTENANCE
The scheduler within LDARMS schedules component screening according to component type. The
schedule includes any valves, flanges, pumps etc. that are required to be screened in a given quarter.
Once the screening information is entered into the program, a quarterly report is generated that calculates
the emissions from fugitive sources for the plant for that quarter. The information from the screenings
are used to estimate the emissions using either the stratified method or the correlation equation. This
information can then be used in SARA 313 reporting.
SOFTWARE PROGRAM MAINTENANCE
Once the screening information has been entered into the system, additional maintenance must be
performed to ensure the integrity of the data in the program. Repairs made to the leaking components
must be entered, changes in service of a component must be logged, and any changes in the process units
must be tracked. For example, if a line is taken out of one process unit and moved to another process
unit and reinstalled, the program must be updated to reflect the changes in location, service, and
chemical stream makeup. If the program is not updated, it will incorrectly estimate the emissions and
print out the wrong locations on reports. The maintenance of an LDAR software program requires that
a considerable amount of time be spent setting up the program correctly, entering the initial data,
maintaining the data in the program, and generating reports.
The reports submitted to the EPA are not necessarily good reports for identifying ways to reduce
fugitive emissions. A software program must allow the user to access the data in a way that presents
information in an understandable form. The information on which emission reduction plans are based
on at one site may be completely different from the data that is used at another site. For these reasons,
the LDARMS software that we developed allows us, or plant personnel, to sit down at the computer and
design a custom report that contains the required information presented in a way that is easy to
understand.
CONCLUSIONS
In order for facilities to comply with the proposed HON regulations, LDAR programs will have to
be implemented. The basic choices involve the type of tags to be used, the information collected, and
the management this information. Although various tag types can be selected by the plant operator, the
information collected is dictated by the regulations. The most efficient way to handle the large amount
of information that must be kept on site is to store it electronically. IT has developed the LDARMS
program to assist the plant manager in the data management and reporting aspects of an LDAR program.
The ease with which the LDARMS program can be adjusted makes it a valuable asset to the plant
manager.
873
-------
GAS AND PARTICLE PHASE MEASUREMENTS OF ATMOSPHERIC TOXIC POLLUTANTS
Douglas A. Lane
Environment Canada
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario
M3H 5T4
Canada
N. Douglas Johnson
ORTECH International Inc.
2395 Speakman Drive
Mississauga, Ontario
L5K 1B3
Canada
ABSTRACT
To follow the migration of semivolatile pollutants through the environment, and
the atmosphere in particular, it is necessary to know the partitioning of the
pollutants between the vapor and the particle phases. Gas and particle (GAP) samplers
have been deployed at Point Petre and at Little Turkey Lake to measure organochlorine
pesticides such as a- and 7-hexachlorocyclohexane (HCH) and hexachlorobenzene (HCB).
The GAP sampler measurements have shown that the particle fraction of semivolatile
compounds in the atmosphere is greater than that which is determined by the high volume
samplers which are backed up by polyurethane foam adsorbents. Additionally, the
results suggest that the a- and 7-HCH have unexpectedly higher particle fractions
during the summer months. HCB remains constant throughout the year.
INTRODUCTION
The mandate of the Canada/US agreement on the Great Lakes is to determine the
deposition of toxic pollutants, such as a- and -y-hexachlorocyclohexane (HCH) and
hexachlorobenzene (HCB), to the Great Lakes. These semivolatile compounds occur in
both the vapor and particle phases in the atmosphere. Thus in order to assess the
relative importance of wet and dry deposition, and to follow the migration of these
compounds through the environment, it is necessary to know their distribution between
the gas and particle phases.
Traditionally, high volume samplers backed up with a polyurethane foam (PUF)
adsorbent are used to obtain operationally defined particle and vapor phase
measurements. However, due to the pressure drop across the filter and fluctuations in
the atmospheric temperature and in the concentrations of the pollutants, some of the
molecules may leave the particle surface and become entrained in the airflow. The
molecules may, then, either be adsorbed onto the filter itself, or pass through the
filter, be trapped on the PUF adsorbent, and be determined (erroneously) as gas phase
material. In addition, gas phase molecules may be trapped on the surface of the
collected particles or they may be trapped directly on the filter itself.
To circumvent these problems, the GAP (gas and particle) sampler, a multi-
annular, diffusion denuder system, was developed. The GAP sampler reverses the
traditional sequence of air sampling by collecting the vapor phase first, then the
particle phase. The instrument comprises two individual sample collection units which
are operated in parallel. Each unit is fitted with a 10 /im size selective inlet (SSI).
In the "denuder" unit, air passes through the SSI, through a. silicone gura/Tenax-coated
denuder in which the gas phase constituents are trapped and retained. The particles
pass through the denuder and are collected on a glass fiber filter. Adsorbent
cartridges containing Florisil are placed downstream of the filter to trap any material
which might volatilize from the surface of the particles or which might be associated
874
-------
with particles which are too small to be trapped by the filter. The second
"conventional" unit is identical to the first except that it does not possess a
denuder. The "conventional" unit yields the total (gas phase plus particle phase)
loading while the denuder unit yields only the particle phase. Thus, by difference,
the gas phase portion of a compound may be determined.
The GAP sampler was designed to operate in areas where the ambient target
compound concentrations could be described as regionally representative or background
in nature. Air sampling was carried out for (nominally) 46 hours. A description and
assessment of the GAP sampler has been published1 as have papers describing its use in
various field measurement programs2"3.
Field sampling was carried out between 1987 and 1990 near Little Turkey Lake
(north of Lake Superior) and at the Canadian master station (established in response
to the Canada-US agreement on the Great Lakes) at Point Petre on the north shore of
Lake Ontario. The filters and adsorbers were analyzed for a- and -y-HCH, and HCB.
EXPERIMENTAL
Prior to sampling, the denuders were purged with 260°C nitrogen gas for 12 hours,
and then cooled. Whatman GF/A glass fiber filters (47 mm) were baked at 300°C, cooled
and weighed before and after use. Florisil was heated in batches at 650°C and 60 g
portions were transferred to solvent-cleaned glass cartridges for elution with 30%
methylene chloride in hexane (MCH) followed by hexane elution. The adsorbent
cartridges were dried under vacuum and stored at 135°C in a desiccator until ready for
use. The cartridges were capped and transported at -5°C in a portable cooler. Just
prior to sampling, the inside surfaces of the SSI and other air sample transfer tubes
were rinsed with hexane.
Duplicate or triplicate samples, and field blanks were collected during each
sampling period. A Tylan mass flow controller regulated the airflow to 16.7 L/min and
the total volume sampled was determined by a dry test meter which had been calibrated
against a primary (gasometer) standard.
The Florisil cartridges were extracted with MCH. After adding an iso-octane
keeper, the extracts were concentrated to 2 mL for cleanup. The filters were extracted
with methylene chloride and concentrated in the same manner as the Florisil adsorbents.
Sample extract cleanup was carried out on a 2% deactivated Florisil column containing
anhydrous sodium sulphate in the top portion. After elution with MCH, then hexane, the
eluent volume, with additional keeper, was reduced to 2 mL and dispensed in septum
vials for injection into the GC. An Hewlett Packard 5890 gas chromatograph with
autosampler was used for analysis of the samples. Injections were split into two
capillary columns of different polarity (Ultra 1 and Ultra 2) and components were
detected with 63Ni electron capture detectors. When the measured results did not agree
between columns it was assumed that the higher result reflected the presence of
interfering compounds and, the lower of the two results was selected. Differences
between the two column results were usually less than 25% and often less than 10%.
RESULTS AND DISCUSSION
The detection limits (DL) were found by multiplying the standard deviation (SD)
of the recovery efficiency by three and normalizing to a standard air sample volume of
46 m3. Quantitation limits (QL) were similarly calculated by multiplying the SD by 10
and normalizing to the same standard air volume. The resulting DL and QL for the
organochlorines (DCs) were as follows: for HCB DL-7 pg/m3, QL-23 pg/m3; for a-HCH
DL-14 pg/m3, QL-46 pg/m3; for 7-HCH pg/m3, DL-15 pg/m3, QL-50 pg/m3
The results of the field measurements for the target OCs are presented in Table
I. As anticipated, the vapor phase components of the three OCs were marginally lower
than those reported by Bidleman4 who used the hi-vol sampler method. A measure of the
precision between two collocated instruments was obtained by determining the modified
median absolute difference (M.MAD) between the samplers5 and then dividing by the
median of the collocated results and expressing the result as a percent. For three
collocated samplers, the median absolute difference was divided by the median of the
875
-------
average result of the three samplers and expressed as a percent. The precision data
shown in Table 2 is considered to be extremely good.
Figure 1 shows the log of the concentration versus 1000/Temperature (K) for the
HCH isomers and HCB for both Little Turkey Lake and Point Petre. Although the sampling
locations are more than 800 km apart, the data appear to be very similar. The slope
for -y-HCH is almost identical to that reported by Hoff6 and clearly supports the
hypothesis that the presence of this compound is a result of the heating and cooling
of the soils. The virtually flat slope of HCB indicates that it is essentially all in
the atmosphere and that there is little bound up in the soils. Hoff6. however, did not
see any correlation of the a-HCH with temperature.
Figure 2 shows the vapor phase component as a function of temperature. The lower
vapor phase component in summer appeared to contradict the predictions of the Junge
theory7. If the denuder suffered breakthrough under warm conditions, some of the
trapped vapor phase molecules would reach the filter and be determined as particle
phase material. However, as reported previously1, the denuders showed no breakthrough
at 30°C over sampling times exceeding 48 h. HCB and the HCH isomers behave similarly
in the denuder system. If the denuder were acting like a chromatographic column, the
HCB would pass through the column first, followed by the HCH isomers. Since the HCB
does not break through (there is no increase in particle phase at higher temperature
as can be seen in Figure 2c) , we conclude that the HCH isomers are not breaking through
either. Since the phenomenon occurs during both summer periods (1987 and 1990) and at
both sites, we must conclude that an increased particle phase component of the HCH
isomers during the summer months in the vicinity of the Great Lakes is a real
occurrence, and not an artifact of the GAP sampler. Local use of HCH and soil
reentrainment may account for the observed increase in particle phase component.
CONCLUSIONS
The GAP sampler technique was designed to address the uncertainty associated with
ambient air gas/Particle measurements of semivolatile constituents by conventional
filter/adsorber methods. It appears that the a- and -y-HCH have unexpectedly higher
particle fractions during the summer months while the HCB remains remarkably constant.
In general, the denuder system shows that the particle phase fraction of semi-volatile
compounds in the atmosphere is greater than determined by the hi-vol sampler technique.
REFERENCES
1. D.A. Lane, N.D. Johnson, S.C. Barton, G.H.S. Thomas and W.H. Schroeder,
"Development and evaluation of a novel gas and particle (GAP) sampler for semi-volatile
chlorinated organic compounds in ambient air," Environ. Sci. Technol.. 22(8): 941
(1988).
2. D.A. Lane, W.H. Schroeder and N.D. Johnson, "On the spatial and temporal
variations in atmospheric concentrations of hexachlorobenzene and hexachlorocyclohexane
isomers at several locations in the province of Ontario," Atmos. Environ.. 26A(1): 31
(1992).
3. D.A. Lane, N.D. Johnson, M.-J.J. Hanley, W.H. Schroeder and D.T. Ord, "Gas and
particle phase concentrations of a-HCH, -y-HCH and HCB in Ontario air," Environ. Scl,
Technol. 26(1): 126 (1992).
4. T.F. Bidleman, U. Widequist, B. Jansson and R. Soderlund, "Organochlorine
pesticides and polychlorinated biphenyls in the atmosphere of southern Sweden, " Atmos.
Environ. . 21, 641 (1987).
5. R.J. Vet and A. Sirois, "The precision of precipitation chemistry measurements
in the Canadian air and precipitation monitoring network (CAPMoN)," in Proceedings of
the 80th Annual Meeting of the Air Pollution Control Association, paper 87-81.8, Air
Pollution Control Association, Pittsburgh.
6. R. Hoff, D.C.G. Muir andN.P. Grift, "Annual cycle of polychlorinated biphenyls
and organohalogen pesticides in air in southern Ontario. 2. Atmospheric Transport and
sources," Environ. Sci. Technol.. 26(2): 276 (1992).
7. C.E. Junge "Basic considerations about trace constituents in the atmosphere as
related to the fate of global pollutants," Adv. Environ. Sci. Technol.. 8(1): 7 (1977).
876
-------
Table 1. Total concentrations and vapor-phase components for HCB, a-HCH and y-HCH at Little Turkey Lake
and Point Petre.
Date
Avg.
Temp.
HCB
Total
(pg.nr3)
% Vapor
a-HCH
Total % Vapor
(pg.nr3)
T-HCH
Total
(pg.rrv3)
% Vapor
o/y
Little Turkey Lake
87/05/08-10
87/05/10-12
87/05/12-14
87/07/24-26
87/07/26-28
87/07/28-30
87/10/20-22
87/10/22-24
87/10/24-26
Point Petre
88/11/08-10
88/11/10-12
88/11/12-14
88/11/22-24
88/11/24-26
88/11/26-28
89/03/15-17
89/03/17-19
89/03/19-21
89/11/06-08
89/11/08-10
89/12/07-09
89/12/09-11
90/01/23-25
90/01/25-27
90/08/13-15
90/08/15-17
90/08/27-29
90/08/29-31
90/09/17-19
90/09/19-21
11.5
9.4
10.0
19.0
17.5
17.9
1.4
0.9
2.0
11.2
8.0
9.4
1.6
3.9
8.3
-1.4
-4.9
-1.3
4.1
6.1
-9.5
-3.5
0.2
0
17.9
20.7
20.1
17.5
9.9
10.9
97 (2,±3)
81 (2,±4)
71 (2,±1)
79(2,±12)
91 (2,±2)
26 (2+3)
76 (2,±5)
68 (2,±8)
75 (2+2)
82 (3,±4)
74(3,±13)
81 (3,±17)
39(3,±10)
58(3,±14)
82(3,±11)
87(3,±10)
57 (2,±3)
79(3+13)
88(2,±12)
94 (2,±7)
110(2,±10)
86 (2+5)
60 (2+3)
93 (2+3)
125(2,±15)
130(2+0)
120(2+10)
77 (2+2)
43 (2+7)
58 (2+4)
92
96
94
94
96
94
100
100
100
93
96
99
98
92
97
100
96
98
89
85
83
93
100
100
90
90
88
92
100
99
285 (2,±25)
270 (2,±20)
310(2+10)
475 (2,±5)
450 (2+40)
330(2+110)
200(2+10)
210(2+0)
180(2+20)
150(3,±10)
103(3+6)
122(3+28)
100(3,±0)
117(3,±6)
330 (3,+36)
71 (3,±13)
60 (2,±0)
59 (3,±7)
145(2,±15)
140(2,±10)
145(2,±5)
85 (2+7)
ND (2,)
ND (2,)
325 (2+25)
235 (2+5)
122(2+38)
66(2+1)
79 (2,±2)
61 (2,±9)
93
96
92
87
92
90
100
100
100
88
100
100
100
100
94
97
100
100
95
100
100
100
82
73
95
81
90
94
88 (2, ±12)
80(2,±1)
115(2,±15)
[49] (2+4)
53(2+12)
[43] (2,±4)
[23] (2,±3)
[26] (2,±0)
[29] (2,±2)
[18](3,±5)
[11](3,±6)
(16] (3,±9)
[8] (3,±0)
[15] (3+6)
[15] (3+1)
[12] (3+7)
[34] (2,±5)
[8] (3+0)
[13] (2+3)
[11](2,±3)
ND (2,)
ND (2,)
ND (2,)
ND (2,)
127(2,±44)
104(2,±27)
59(1.)
23(2,±1)
27 (2,±0)
[8](1.)
91
93
91
93
77
81
100
100
100
100
100
100
100
100
100
100
100
100
100
100
43
47
93
82
27
100
3
3
3
10
9
8
9
8
6
8
9
8
13
8
20
6
2
7
11
13
2.5
2.4
2-.0
2.9
2.9
7.6
The first number in parentheses is the number of replicate measurements and the second is the range (for
duplicates) or standard deviation (for triplicates) of the measurements. The values in square brackets are above
the detection limit but below the quantitation limit.
Table 2. Precision (%) for duplicate and triplicate sampling measurements of three organochlorines at Little
Turkey Lake and at Point Petre.
Location
HCB
a-HCH
•y-HCH
Little Turkey Lake
(duplicate measurements)
Point Petre
(triplicate measurements)
9.66
15.3
12.9
7.2
17.3
25.5
877
-------
3.4
3.6
1000AT(K)
Figure 1a. Log[a-HCH] vs 1000/T(K)
3.8
,2A'
;1.8 -
§>
1.2 -
0.6-
Log(C) = (-2287/T) + 9.55
Turkey Lake data
Point Petre data
3.4
3.6
1000/T(K)
Figure 1b. Log[g-HCH] vs 1000/T(K)
3.8
o 1
_J
o
6_
n
3.
dd D
• H ^i Bn rji n in «•*
Log(C)= (-131/T) + 2.35
• Turkey Lake data
i
4 3.6
1000/T(K)
n g n D
q- - - D
rj Point Petre data
3.
Figure 1c. Log[HCB] vs 10OOAT(K)
878
-------
a
n 80 -
o
a 60 -
C ,1/1
— 40 -
3?
0 2°
rt 0 -
-1
•LJ LJ— 1_1 y
I
0 -5 C
Q n *• QB ^L. ^
n
• Turkey Lake data n Point Petre data
) ' 5 10 15 20 2,
Temperature (C)
Figure 2a. a-HCH (% in Vapor Phase) vs Temperature
n on _
9- fin -
C Aft
$
T or\ -
0 ^
i n -
Ul U
-1
m 100 -I
.c QO -
n fiO -
$
i- Art -
o
1 0 -
-1
0 -5 C
Figure 2b.
° n C
Q . . .
0 -5 C
_ . . . r* _________
•T
n °
H Turkey Lake data p Point Petre data
) 5 10 15 20 2f
Temperature (C)
g-HCH (% in Vapor Phase) vs Temperature
"^ -a Q ebea 55-^
• Turkey Lake data p Point Petre data
i i i i
5 10 15 20 2!
Temperature (C)
Figure 2c. HCB (% in Vapor Phase) vs Temperature
879
-------
EVALUATION OF POLYNUCLEAR AROMATIC HYDROCARBONS AND
NITROGEN HETEROCYCLES IN THE STACK EFFLUENT OF
ASPHALT PROCESSING PLANTS
Kim Hudak, Richard Pirolli, and Hewt Rowe
Connecticut Department of Enviromental Protection
Bureau of Air Management
165 Capitol Avenue
Hartford, Connecticut 06106
ABSTRACT
The Connecticut Department of Environmental Protection (CT DEP) conducted
stack effluent monitoring at several asphalt plants during 1991 and 1992. The
samples were collected in accordance with US EPA SW846 Method 0010 and analyzed
by high resolution gas chromatography/ high resolution mass spectroscopy
(HRGC/HRMS).
The HRGC/HRMS methodology was developed by Triangle Laboratories, Inc. of
Research Triangle Park, North Carolina to satisfy CT DEP's monitoring objectives.
These objectives included determining the feasibility of monitoring and
simultaneously analyzing for both polynuclear aromatic hydrocarbons (PAH) and
nitrogen heterocycles (NHC) in asphalt plant stack samples. The list of target
PAHs was developed by CT DEP to help better define Its polynuclear aromatic
hydrocarbon regulations.
The test results show that seven compounds from CT DEP's targeted list of
PAH comprise 98-99J of the identified PAH emissions from asphalt plants. CT DEP
reviewed data from oil and natural gas-fired batch operations as well as natural
gas-fired batch and express plant operations. Comparisons of the results were
made.
INTRODUCTION
The Connecticut Department of Environmental Protection (CT DEP) conducted
stack emissions monitoring at several asphalt plants during 1991 and 1992. Two
types of plants at the Balf Company were tested as part of a USEPA High Risk
Point Source (HRPS) Study to determine qualitatively and quantitatively the PAH
and NHC emitted from asphalt plants. CT DEP intended to use the PAH emissions
data to regulate a specific source as well as support a revision of the existing,
but inadequate PAH regulation.
At CT DEP's request two additional plants were tested. The Astec plant, a
natural gas-fired, 4 ton/batch plant, was tested in fulfillment of a condition
written into its permit to construct and operate. The other plant, Tilcon, a 5
ton/batch plant, agreed to be tested at the request of the department. For
testing purposes, Tilcon temporarily converted to it's backup fuel, No. 2 oil.
Connecticut has a wide variety of asphalt plants. The majority of Connecticut's
45 plants burn No. 2 oil with natural gas backup. The newer facilities are
required to burn natural gas with No. 2 oil backup. The plants chosen for study
were representative of typical facilities, both new and old. However, the study
does require more testing of the older, oil-burning asphalt plants.
880
-------
The concept for the project was formulated after attempts to estimate PAH
emissions from asphalt plants for enforcement purposes were largely
unsuccessful. Previously, estimated emissions of PAH or polycycllc organic
natter (POM) from asphalt plants were based on limited non-speciated test
data . In addition, CT's existing hazardous air pollutant regulation for PAH
was unclear and difficult to enforce because it was defined as PAH (benzene
soluble). As an interim policy designed to facilitate the permitting of new
asphalt plants, provisions were included in new permits to require testing for
PAHs consequent to the development of an appropriate test method. As a result of
this study, CT DEP now regulates specific PAH compounds and can specify a
particular test method.
EXPERIMENTAL METHODS
Sample Collection
Two of the facilities CT DEP tested were at the Balf Company of Newington,
which operates both batch and express or drum mix asphalt plants. The batch
plant has a maximum operating capacity of 10 tons/batch while burning natural
gas. The asphalt express process is a Gencor-Bituma drum mix asphalt plant.
This plant has a maximum operating capacity of 150 tons/hour while burning
natural gas. Both operations utilize a baghouse for particulate removal.
The PAH/NHC emissions from the plants were sampled in triplicate according
to SW 846 Method 0010. The volumetric flowrate, moisture content, and stack gas
composition were determined during each test run. All testing was performed on
the exhaust stack. In conjunction with PAH/NHC testing, EPA Methods 1 through 4
were performed.
The PAH/NHC sampling method used the modified Method 5 sampling train. A
modification in the sample recovery procedure was used and consisted of replacing
the 1:1 mixture of methanol/methylene chloride rinse with separate rinses of
acetone and methylene chloride. An acetone rinse followed by a methylene
chloride rinse has been shown to be more efficient than the methanol/methylene
chloride (1:1 volume) rinse. The PAH/NHC sampling method also included several
unique preparation steps which ensured that the sampling train components were
not contaminated with organics that may have interfered with analyses.
Triangle Laboratories, Inc. (TLI), Research Triangle Park, North Carolina
performed the preparation of the glass fiber filters and the XAD-2 resin. All
filters were cleaned before their initial use. The methylene chloride extract
from the filter cleaning procedures was anaylzed for PAH/NHCs. If any PAH/NHC
Has present in the concentration above the minimum detectable limit, the cleaning
procedure was repeated and the extract reanalyzed until no PAH/NHC was detected.
The XAD-2 resin was placed in a soxhlet and extracted with HPLC grade water,
methanol, and methylene chloride. After extraction, the resin was dried and
placed in the sampling cartridges which were tightly capped with glass plugs.
The extracts were then analyzed for total chromatographic organics (TCO) and
targeted PAH/NHC compounds. If any PAH/NHC were present at a concentration above
the minimum detectable limit and/or the TCO was greater than 20 ug/ml, the
cleaning procedure was repeated until each criteria was met. In addition, the
resin of each trap as fortified with 100 pg/ml of dl4-Terphenyld.
881
-------
Sample Analysis
High resolution gas chromatography/high resolution mass spectroscopy
(HRGC/HRMS) analytical methodology was developed by TLI to satisfy CT DEP's
monitoring objectives 3. These objectives included determining the analytical
feasibility of monitoring for 29 polynuclear aromatic hydrocarbons and five
nitrogen heterocycles in asphalt plant stack samples. These compounds are listed
in Table 1.
Table 1.
List of Target Polycycllc Aromatic Hydrocarbons
and Nitrogen Heterocycles.
Naphthalene 2-Methylnaphthalene Acenaphthene
2-Chloronaphthalene Acenaphthalene Fluorene
Phenanthrene Anthracene Carbazole
Acridine Fluoranthene Pyrene
3-Methyl-fluoranthene Cyclopenta-c,d-pyrene Benz-a-anthracene
Chrysene Perylene Benzo-b-fluoranthene
Benzo-j-fluoranthene Benzo-k-fluoranthene Benzo-a-pyrene
Benzo-e-pyrene 7H-Dibenzo-c,g-carbazole Benzo-[ghi]-perylene
Dibenz-[ah]-anthracene Dibenz-[aj]-anthracene Dibenz-[aJ]acridlne
Dibenz-[ac]-anthracene Dibenz-[ah]-pyrene Dibenz-[ai]-pyrene
Dlbenz-[ae]-pyrene Dibenz-[al]-pyrene Indeno-[1,2,3-cd]-pyrene
7,9-Di-Methyl-benz-c-acridine
No problems were noted by TLI's sample preparation and mass spectrometry
groups while performing the analyses. However, several factors did impact data
validity. One of the factors was matrix related interferences which affected the
quantltation of some analytes. This problem was described as severe in some
samples. Quantitation of naphthalene, chloronaphthalene, acenaphthylene,
acenaphthene, fluorene, phenanthrene, acridine and carbazole could not be
performed in a particular sample as a result of the matrix interference. TLI did
provide quantitative results for these analytes in that sample which were
believed to be minimum estimates. However, these values were not reliable enough
to support an enforcement action. High recoveries of some internal standards
were calculated in samples as a result of this interference. TLI modified their
cleanup procedures in an attempt to more effectively remove: the*matrix
interferences.
A second factor was that the results of all samples displayed selective
losses during extraction for the labeled acridine and carbazole internal
standards. Since the associated analytes act chemically in the same fashion, it
can be assumed that these and other nitrogen heterocycles are selectively lost
during extraction.
Another factor was that "b" and "j" isomers of benzofluoranthene as well as
the "ac" and "ah" isomers of dibenzoanthracene coelute. Therefore, quantitative
results were provided as a total for each isomer pair.
The last factor of concern was that saturated peaks (i.e. beyond
calibration) were noted in all the samples. TLI believed that quantitative
results for saturated signals were "minimum estimates".
The total PAH concentration was calculated for each sample. Naphthalene,
2-methylnaphthalene, and 3-methylfluoranthene were not included in the total PAH
concentrations.
882
-------
RESULTS
CT DEP reviewed data from both oil and natural gas-fired batch operations as
well as batch and express plant operations. Table 2 lists the comparison of data
from asphalt batch and express processing plants firing natural gas. Table 3
shows a comparison of three asphalt batch plants. The Half and Astec plants were
firing natural gas. The Tilcon plant was firing No. 2 oil.
Table 2.
natural Gas-Fired Asphalt Batch and Express Plants
PAH Data Excluding Naphthalene (ug/m3).
Analyte BATCH EXPRESS
Acenaphthene 1.75 (27?) 0.364 (9.4?)
Acenaphthylene 0.25 (4?) 1.185 (31?)
Fluorene 2.28 (36?) 0.848 (22%)
Phenanthrene 1.88 (29?) 1.147 (30?)
Anthracene 0.06 (1?) 0.066 (1.7?)
Fluoranthene 0.05 (1?) 0.119 (3-1?)
Pyrene 0.07 (1?) 0.081 (2.1?)
Benzo-a-Anthracene 6.7E-4 (.01?) 0.003 (.08?)
Chrysene 0.01 (0.2?) 0.024 (0.6?)
Benzo-k-Fluoranthene 0 (0?) 0.002 (.04?)
Benzo-j-Fluoranthene 0 (0?) 0 (0?)
Benzo-b-Fluoranthene 0.0017 (.03?) 0.005 (.12?)
Benzo-e-Pyrene 0.005 (.08?) 0.008 (.21?)
Benzo-a-Pyrene 0.0007 (.01?) 0.003 (.08?)
Indeno-123-cd-Pyrene 0.0027 (.04?) 0.003 (.08?)
Benzo-ghi-Perylene 0.027 (.43?) 0.023 (0.6?)
Dibenzo-ah-Anthraeene 0 (0?) 0 (0?)
Total PAH 6.4 3.9
DISCUSSION
The majority of the observed PAH emissions were comprised of: acenaphthene
(24?), acenaphthylene (10.7?), fluorene (34.3?), phenanthrene (22.6?), anthracene
(2.7?), fluoranthene (2.4?), and pyrene (1.8?). These 7 PAHs make up
approximately 98-99? of the PAHs identified in the stack effluent samples.
Naphthalene, 2-methylnaphthalene, and 3-methylfluoranthene were not included.
Emissions from the No.2 fuel oil fired batch process appeared to be twice
the emissions from the natural gas fired process. Emissions from the express
drum mix plant appear to be 40? lower than the batch plant (both plants natural
gas-fired). Although the data base was quite small, it appeared that
acenaphthylene was significantly more prevalent in the express plant samples.
Correspondingly, acenaphthene seemed to be far more predominant in the batch
samples.
CONCLUSIONS
Regulatory Implications
The following analytes were dropped from the target list of PAH/NHC
compounds: 2-chloronaphthalene, carbazole, acridine, cyclopenta-cd-pyrene,
dibenz-aj-acridine, 7,9-dimethyl-benz-c-acridine, perylene,
7H-dibenzo-cg-carbazole, the "ac" and "aj" isomers of dlbenzanthracene, and the
"al", "ae", "ai", and "ah" isomers of dibenzpyrene. These compounds were seldom
if at all detected and then, only at trace levels in samples which contained
relatively high PAH levels (especially when compared to ambient concentrations).
883
-------
Table 3
Asphalt Batch Plants PAH Data Excluding Naphthalene (ug/m3).
Analyte
Acenaphthene
Acenaphthylene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo-a-Anth
Chrysene
Benzo-k-Fluo:
Benzo-J-Fluo:
Benzo-b-Fluo
Benzo-e-Pyrene
Benzo-a-Pyrene
Indeno-12;
Benzo-ghi-
Dibenzo-al
Total PAH
BALF
ASTEC
Natural Gas-Fired Natural Gas-Fired
10 ton/batch 4 ton/batch
1
le 0
2
1
0
0
0
•acene 6
0
'anthene 0
"anthene 0
?anthene 0
le 0
le 0
i-pyrene 0
rylene 0
ithracene 0
6
.75
.25
.28
.88
.06
.05
.07
.7E-4
.01
.0017
.005
.0007
.0027
.027
(27?)
(4?)
(36?)
(29?)
(1?)
(1?)
(1?)
(.01?)
(0.2?)
(0?)
(0?)
(.03?)
(.08?)
(.01?)
(.04?)
(.43?)
(0?)
1.
1.
1.
2.
0.
0.
0.
0.
0.
0.
-
0.
0.
0.
0.
0.
0.
46
19
26
2
032
133
117
004
015
003
013
006
0019
003
058
0014
(22.5?)
(18.3?)
(19.4?)
(33-8?)
(.5?)
(2?)
(1.8?)
(.06?)
(.23?)
(.05?)
-
(0.2?)
(.09?)
(.03?)
(.05?)
(0.9?)
(.02?)
TILCON
No. 2 Oil-Fired
5 ton/batch
3
0
5
1
0
0
0
0
0
6
0
0
0
6
0
0
0
.12
.37
.18
.09
.6
.36
.23
.016
.086
(28?)
(3.3?)
(47?)
(9.8?)
(5.4?)
(3-2?)
(2.1?)
(0.14?)
(0.77?)
.7E-4(.006?)
.006
.008
(0?)
(0.05?)
(0.07?)
.7E-4(.006?)
.006
.044
(0.05?)
(0.4?)
(0?)
.4 6.5 11.1
-------
CT DEP now requires SW846 0010 as the PAH test method. Acetone and
methylene chloride are to be used as the sample recovery solvents. Analysis is
performed with HRGC/HRMS. Monitoring for nitrogen heterocycles requires
additional method development and validation as well as separate sampling and
analyses from PAHs. Therefore, CT DEP is not requiring sources to test for NHCs.
The asphalt plants fired with natural gas were generally well below CT DEP's
maximum allowable stack concentration for PAH. The plant fired with No. 2 fuel
oil was approximately 5 times the allowable stack concentration.
CT DEP has developed a proposed regulation for PAH and naphthalene as a
result of this project. The ambiguity of the existing definition and multiple
listings of PAH compounds in the regulations hampered both the CT DEP and the
sources in determining compliance and performing appropriate monitoring.
CT DEP proposed the following definition be added to the regulations: (PAH)
will be defined as the sum of the following compounds: acenaphthene,
acenaphthylene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene,
benzo(a)anthracene, chrysene, benzo(k)fluoranthene, benzo(j)fluoranthene,
benzo(b)fluoranthene, benzo(e)pyrene, benzo(a)pyrene, indeno-1,2,3-cd-pyrene,
benzo-ghi-perylene, dibenzo-ah-anthracene.
CT DEP proposed the following items be deleted from the existing
regulations: benz(a)pyrene, coal tar pitch volatiles, polynuclear aromatic
hydrocarbons (PAH), benz(a)anthracene, benzo(b)fluoranthene, chrysene,
7H-dibenzo(c,g)carbazole, dibenzo(a,h)anthracene, dibenzo(a,h)pyrene,
dibenzo(a,i)pyrene, indeno(1,2,3-cd)pyrene, and, asphalt (petroleum) fumes. CT
DEP also proposed to redefine "naphthalene" as the sum of both naphthalene and
2-methylnaphthalene.
REFERENCES AND BIBLIOGRAPHY
1. AP-H2 Compilation of Air Pollutant Emission Factors Volume 1: Stationary
Sources, U.S. EPA, Office of Air and Radiation, Office of Air Quality
Planning and Standards, Research Triangle Park, NC, October 1986.
2. R. Pirolli, Emission Test Program for Selected Polycycllc Aromatic
Hydrocarbons and Nitrogen Heterocycles at an Asphalt Processing Facility
Test Protocol, State of Connecticut Department of Environmental Protection,
Hartford, 1991.
3. H. Karam, Proposal in Response to Request From CT DEP Determination of
Polycycllc Aromatic Hydrocarbons and Nitrogen Heterocycles in Ambient Air
and Stack Emission Samples By High Resolution Gas Chromatorgraphy / High
Resolution Mass Spectroscopy, Triangle Laboratories, Inc., Research Triangle
Park, NC, September 1990.
885
-------
ESTIMATES OF PESTICIDE EXPOSURE FROM THE AGRICULTURAL HEALTH STUDY
(AHS)
Nicholas J. Giardino, David E. Camann, Paul W. Geno, H. Jac Harding,
and Jacquelyn M. Clothier
Southwest Research Institute
P.O. Drawer 28510
San Antonio, TX 78228-0510
ABSTRACT
Several routes of pesticide exposure to farmer applicators were measured and compared. Whole body (WB)
exposure was measured using a-cellulose patches placed on the outer clothing of the farmer; two measures of hand
exposures were taken, one using cotton gloves (GL) and the other using isopropanol handwipes (HA). Inhalation
exposures were measured using a PUF cartridge coupled with a personal sampling pump. The three highest exposures
received in descending order were: WB > GL > HA.
INTRODUCTION
The primary goal of the Agricultural Health Study (AHS) sampling program is to provide quantitative measures
of pesticide exposure from different farm tasks. In this study, the term exposure is defined as what contacts a person,
whether on clothing, skin, or through simple processing by breathing or mgestion. Absorption and metabolism after this
contact takes place is not considered here.
METHODS
All of the methods for collecting the Geld samples used to assess exposures to pesticides can be found in Harding
et al.1 The analytical methodology used to analyze all field samples can be found in Geno et al.2
RESULTS AND DISCUSSION
Hygiene, Activity Patterns and Demographics
Hygiene. The degree that each farmer practiced good hygiene varied substantially from farm to farm. Farmers 2 and
3 exercised more caution while loading and mixing pesticides. Farmer 2 was the most careful in the use and application
of his pesticides. He did not experience any spills or splashes during loading and mixing of the pesticides. When applying
the pesticide to his crop, he was inside an enclosed cab. A drift retardant was added to the pesticide application mixture.
Farmer 2 also tried to purchase only that amount of pesticide needed for a specific job reducing leftover waste. He did
have to stop work to make a five-minute repair on a leaking hose. He wore approximately the same clothing as the other
fanners. This included a baseball cap, cotton gloves, a cotton work shirt and pants, and shoes. Farmer 2 washed his
hands within 15 minutes after leaving work.
Farmer 3 also exercised good care in handling of pesticides. He wore rubber or leather gloves during all phases
of his work. While applying pesticides, he had to stop work for approximately two minutes to repair a nozzle. This
farmer also did spraying of hogs inside a barn with an aerosol can. He washed his hands within 30 minutes after leaving
work. Farmer 3 was observed eating his lunch in the field, during a work break, without first washing his hands.
Farmer 1A also may have contaminated himself while hosing down some hogs with lindane. Farmer 1A does
animal application of pesticides four times yearly. He washed his hands within one to two hours after leaving work.
Farmer IB poked a hole through a container of alachlor and saturated the fmger of one of the cotton gloves he
was wearing. This glove was removed because of the chance of high exposure. Otherwise, Farmer IB reported little hand
contact with any pesticides, since he reported that he washed his hands between each loading. Farmer IB also applied
pesticides to animals six times yearly.
886
-------
All the fanners changed clothes once a day, and their work clothes were washed once every three days.
Activity Patterns. All the farmers spend as much as 16 hours in the field during a busy season. Crops can be sprayed
every day for up to a two-week period. Farmer 1A reported being active during work and inactive during leisure hours.
Fanner 1A had a three year old child. This child's activities were reported for a total period of 70 hours. Out of this
total time of 70 hours only one and one-half hours were spent playing in dirt by the child. No hand-to-mouth activity
was reported although some probably occurred. While not working, Farmer IB spent most of his time with his spouse
indoors during the monitoring period. The family of Fanner 2 all participated in planting and gardening. The wife of
Farmer 2 spent four and one-half hours in the field during the three-day monitoring period. She spent tliree and one-half
hours of this time in the field while pesticides were being applied. The family of Farmer 3 also participated to varying
extents in lawn and gardening activities.
Demographics. At Farm 1A there was a married couple between the ages of 26 to 45 years old. They had two children,
a son approximately one year old and a daughter three years old. Farmer IB and his spouse were at least 60 years old.
Fanner 2 and his sp :use were 26 to 45 years of age. They had a 15 year old son and a 13 year old daughter. Farmer
3 and his spouse were 26 to 45 years old and had a daughter eight years old and a son four years old.
Occupational Exposure of Farmers During Monitoring Events
Hand Versus Inhalation Exposure. The highest (but not the only) exposures to the hands can occur during accidental
spills and splashes, while loading or mixing pesticide, or during repairs on leaky hoses or nozzles. An accidental spill with
cotton glove and possible hand contamination did occur on Farm IB on Day 4 of the monitoring period. This accidental
spill at this farm highlights the importance of considering hand exposure. Also, it is unknown whether any of the farmers,
while at work, washed their hands before urinating. If not, some pesticide may have been transferred from the hands
to the genital area. This is of special concern because of the thin dermal layer of the penis and scrotum through which
pesticides will transfer readily.
Table I addresses two issues. One concerns the inhalation exposure (BC, see Table I) received by the farmer
during monitored application events. The second is the comparison of total exposure in milligrams received by the hands
(HA, see Table I) as compared to the total milligrams sampled in the farmer's breathing zone. It is clear that for those
cases for which a TLV (Threshold Limit Value) was available, the farmers' inhalation exposures, while working during
the monitoring periods, were far below the TLVs.3 A similar comparison can be made between the mass of pesticide
recovered from the hands and the mass collected by personal air sampling. The mass of pesticide measured on the hands
is far greater than that collected by personal air sampling.
Table I. Comparison of handwipe amount to breathing concentration and amount inhaled.
Farm by Event(1'
Alachlor
1B-3A
1B-3B
1A-3C
1B-4A
Alrazine
1B-3A
1B-3B
1A-3C
1B-4A
2,4-D Isooctyl Ester
Time
00
3.000
5.400
2.600
3.300
3.000
5.400
2.600
3.300
TLV
(mg/m3)
NONE
NONE
NONE
NONE
5
5
5
5
Bd2'
(mg/m3)
0.008
0.011
0.014
0.020
0.002
0.002
0.003
0.003
Mass Processed by
HA(3' Inhalation During
(mg) Work (mg)(4)
6.380
14.800
0.480
9.540
2.000
4.330
0.113
3.76
0.043
0.110
0.065
0.120
0.011
0.019
0.014
0.018
887
-------
2-3
Lindane
1A-4B
Pyrethrins
3-4A
Trifluralin
2-3
3.000
1.100
0.045
3.000
NONE
0.5 (SKIN)
5
NONE
0.001
0.001
0.240
0.001
0.152
0.152
0.450
0.023
0.005
0.010
0.020
0.005
(1) All pesticides applied during these events. See Table IV for explanation of coded
events.
(2) BC = The average breathing concentration (mg/m3) during the application of pesticides.
(3) For the HA (applicator handwipe) data, recovery from the skin is unknown. Recovery from
the isopropyl-saturated cotton gauze used to take the handwipes was good. Biases in the HA
data due to skin absorption or other mechanisms is unknown.
(4) Calculated by multiplying time (h) * BC (mg/m3) * 1.8 m3/h (see note below).
TLV Threshold Limit Value (mg/m3), defined as that concentration in air that a healthy worker
may be exposed to for an eight-hour working period presumably without adverse health effects
resulting.
NOTE: A moderate workload was assumed for the farmers with a breathing rate of 1.8 m /h. The
personal air samplers were run at 0.23 m /h.
Inhalation exposure is most likely to occur from overspray while the pesticide is being applied to the crops. Some
small contribution may be received during mixing and loading.
Whole Body Exposure Versus Hand. Another measure of integrated exposure is whole body (WB) exposure. This was
calculated using a scaling factor for each farmer. This scaling factor was calculated by first obtaining the surface areas
of the farmers' bodies off a noroogram.4 This required a knowledge of each farmer's height and weight. Next, the total
area of the analyzed body a-cellulose patches was calculated. Then to get the scaling factor the estimated surface area
of the farmer was divided by the total area for all three analyzed body a-cellulose patches. The results of these
calculations are in Table II. To calculate the estimated whole body (WB) exposure, the corresponding body patch
concentration was multiplied by the scaling factor to yield a result in milligrams. These results were then compared to
the measured mass in milligrams on the cotton gloves as is shown in Table II. Comparing the median values in Table
II shows that the estimated WB > GL, but on the same order of magnitude.
Table II. Comparison of whole body exposure to pesticide retained on cotton gloves.
Farm by Event*
Alachlor
1B-3A
1B-3B
1A-3C
1B-4A
Atrazine
1B-3A
1B-3B
1A-3C
1B-4A
DP
(mg)
2.05
5.39
0.38
1.02
0.137
0.296
0.114
0.373
Scaling Whole Body Exposure GL
Factor (WB in mg) (mg)
223
223
218
223
223
223
218
223
457
1022
83
227
30
66
25
83
105
270
59.8
>2370(1>
2.37
33.2
5.25
16
-------
Table II. Comparison of whole body exposure to pesticide retained on cotton gloves.
Farm by Event*
2,4-D Isooctyl Ester
2-3
Lindane
1A-4B
Pyrethrins
3-4A
Trifluralin
2-3
DP
(mg)
0.00891
0375
0.023
0.00047
Scaling
Factor
202
218
202
202
Whole Body Exposure
(WB in mg)
1.8
82
5
0.1
MIN=0.1
MAX=1078
MEDIAN =67
GL
29.3
21.2
2.53
0.252
0.252
2370
25.25
* All pesticides applied during these events.
(1) Finger tip of glove saturated with pesticide.
DP Dermal a-cellulose patches attached to front of thighs and to the back of the nape of
the neck.
SCALING FACTOR: Estimated surface area of farmer divided by total area of all three
body a-cellulose patches.
WB Whole body exposure (mg). WB (mg) = (DP)*(SCALING).
GL Cotton glove measurements (mg).
Application Versus Nonapplication Events. Day 2 at all farms and Day 4 at Farm 2 were baseline days during which
no pesticides were applied to the crops. Days 3 and 4 were days during which pesticides were applied to the crops. It
is useful to compare the exposures received by the farmers on Day 2 to those received on Days 3 and 4.
It has been shown above that handwipe (HA) data are a good measure of the fanner applicator's exposure. A
comparison of application day exposure versus baseline day exposure is given in Table HI for the applicator handwipe
data. Note that the summary statistics in Table III represents all detected handwipe measurements of all analytes
grouped together, separately for application and nonapplication days. The median value for all the measurements taken
on application days is four orders of magnitude above baseline days. The maximum for Days 3and4isalso much greater
than that for Day 2. This indicates an increase in exposure due to pesticide handling and application.
Table HI. Farmer handwipe exposure (mg) on application versus
nonapplication days.
Percentiles for Nonapplication Days
10%
0.000015
25%
0.000045
50%
0.000085
75%
0.0029
100%
0.530
N=35
MIN=0.000015
MAX=0.530
MEDIAN=0.000085
-------
Percentiles for Application Days
10% 25%
0.00044 0.0162
N=18
MIN=0.00044
MAX= 14.80
MEDIAN=0.16
50% 75% 100%
0.160 3.76 14.80
CONCLUSIONS AND RECOMMENDATIONS
The three highest exposures received in descending order were: WB > GL > HA. The exposure to the whole
body (WB) was comparable to the cotton glove (GL) data. Both represent an integrated average over the work period.
WB exposure is an estimated value which assumes the sampled dermal patch areas are representative of exposures
received by all body surfaces. GL exposure is calculated from direct measurements. Because WB exposure is an
estimated value, this tends to lessen the confidence in the results. GL data may be superior to WB estimated values, but
there were problems with the collection of the GL data. The cotton gloves either were ppor fitting or restricted the
farmer in his ability to do his job effectively.
Greater care in exercising good hygiene practices could substantially reduce the farmer applicator's exposure as
well as that of their families. Such practices could include: cautious handling of pesticides, developing proper hand
washing habits, removing shoes prior to entering the home, bathing immediately after work, and washing work clothes
separately in cold water.
Table IV. Event ID code.
Farm ID
1A
IB
2
3
Week Monitored Sampling Period/Event
Weekl
Week 1
Week 2
Week3
2
3
3A-
3B-
4
4A-
4B-
Day2
Day 3
Day 3, 1st event/sample
Day 3, 2nd event/sample
Day 4
Day 4, 1st event/sample
Day 4, 2nd event/sample
ACKNOWLEDGEMENTS
We acknowledge Vincent F. Garry, M.D. of the University of Minnesota for recruitment of participant farms.
This research was funded by the U.S. Environmental Protection Agency (Contract 68D10150). This paper has been
submitted to EPA's peer and administrative review, but no official endorsement should be inferred.
REFERENCES
1. H. J. Harding, P. M. Merritt, J. M. Clothier, D. E. Camann, A. E. Bond, and R. G. Lewis, "Sample
Collection Methods to Assess Environmental Exposure to Agricultural Pesticides," in Proceedings of the
1993 U.S. EPA/A&WMA International Symposium on Measurement of Toxic and Related Air
Pollutants." Air & Waste Management Association, Pittsburgh, 1993 (in press).
890
-------
2. P. W. Geno, D. E. Camann, and K. Villalobos, "Analytical Methods for Assessing the Exposure of
Farmers and Their Families to Pesticides," in Proceedings of the 1993 U.S. EPA/A&WMA International
Symposium on Measurement of Toxic and Related Air Pollutants." Air & Waste Management
Association, Pittsburgh, 1993 (in press).
3. Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. ACGIH,
1991-1992.
4. Merck Medical Manual Twelfth Edition, pg. 1840, Figure 57.
891
-------
Subject Index
Accelerator mass spectrometry (AMS) 585
Accumulation 827
Acenaphthylene (ACE) 129,408
Acetaldehyde 579
Acetone 307,479
Acid aerosols 64, 501,651,727,733,741,747
Acidity 651,733,741,747
Acousto-OpH'c Tunable Filter (AOTF) 563
Activity patterns 836,886
Adsorbents sampling 604
Advection 41
Advecn've flow 9
Aerosol sulfate 727
Aerosols 141,185,197
Agricultural Health Study (AHS) 698,706,712,
718,836,842,886
Air 827
Air analysis 646
Air filters 249
Air monitoring 276,571,863
National stations 351
State and local stations 351
Airpreconcentrator 327
Air quality 221
Measurements 185
Air sampling 679
Air toxics 35,313,408,775
Analysis 857
Air-Conditioning and Refrigeration Institute (ARI)
370
Standard 700-88 370
Air/Superfund program 29
Air-water transfer process 263
Airborne Toxic Control Measure (ATCM) 429
Airliner cabin 243
Alcohols 167
Aldehydes 167
Alkaline dust 651
Allergies 671
Ambientair 161,185,190,257,269,276,313,333,
337,451,457,479,501,537,551,585,610,616,
622,628,640,646,657,741
Monitoring program 488
Sampling 761
Toxics 795
American Society of Heating, Refrigerating and Air-
Conditioning Engineers (ASHRAE) 15,243
Ammonia (Nil,) 470,727,733,741
Ammonium
Bisulfate 651
Nitrate 651
Sulfate 741
Analysis of variance (ANOVA) 733
Analysis methodology 698
Analytical method 698,712
Analytical surrogate 301
Annular Denuder Systems (ADS) 129,663,727
Filter packs 663
Area monitoring 857
Area sources 29,35,545,579
Arizona Department of Environmental Quality
(ADEQ) 622
893
-------
Aromatic compounds 73
Arsenic 806
Asphalt plant emissions 436
Atmosphere 563
Atmospheric Research and Exposure Assessment
Laboratory (AREAL) 64,378
Atmospheric transformation process 167
Atrazine 718
Attached garage 52
Audit 100,351
Materials 363
Auto exhaust 269
B
Benz(a)anthracene (BaA) 141,147
Benzene 307,319,479,806
Benzo(a)pyrene(BaP) 111, 129,141,806
BFB tuning 100
Biocontaminants 82
Biogenic emissions 585
Biological sampling 712
Boating industry 806
Bromine (Br) 185
Building industry 806
Bulk collectors 515
Butane 307
Butylbenzene
N-butylbenzene 307
c
Calibration standards 781,789
California Air Resources Board (CARB) 429
Method 431 429
Canadian Acid Aerosol Measurement Program
(CAAMP) 747
Cancer risk 185
Canisters 313,333,557,585,610,795
Samples 337,761
Capillary column 616
Carbon (C) 585
Carbon molecular sieve 571
Carbonyls 155,313,622,813
Carbotrap300 571
Carcinogenic pollutants 175
Carpet 848
CERCLA 363
Certified reference materials (CRM) program 357
CH4 585,616
Chemical exposure 221
Chemical phase equilibrium 41
Chemiluminescence analyzer 657
Chlorpyrifos 848
Chlorobenzene 800
Chlorofluorocarbon (CFC) 370
CFC-11 370
CFC-12 370
CFC-502 370
Clean Air Act (CAA) 464,628
Clean Air Act Amendments (CAAA) 155,161,167,
257,296,313,327,363,393,399,423,610,622,
628,634,775,806,813,857,869
Section 112 775
Title I 610,634
Title ffl 296,327,393,423,806
Clean Air Status and Trends Network (CASTNet)
313
CO 123,167,227,243,249,378,531,585,610,
616,789,806
CO2 15,94,167,227,243,269,464,585,806
894
-------
Coal burning 741
Coal-fired boiler 423
Combustion 806
Composting 488
Confined environments 221,227,237
Connecticut Department of Environmental Protection
(CTDEP) 436,880
CONTAM88 model 52
Contaminant control 221
Continuing calibration (CONCAL) 100
Continuous
Emissions Monitors (CEMs) 857
Monitoring 296
Sampling automated samplers 775
Copper (Cu) 479
Cotinine 88,685
Cresol 111
Criteria air pollutants 351,457
Cryoconcentrator 646
Cryogen 282,628,634
Cryogenic
Concentration 863
Preconcentration 313,616
Trap 337
Cryotrap 646
Cylinder standards 781
D
Data collection 869
Data management 869
Data Quality Assessment (DQA) process 345
Data Quality Objectives (DQO) process 345
Decane
N-decane 585
Demographics 836,886
Denuder 651,747,874
Deposition 58,167,827
Dermal
Adsorption 691
Exposure 698,712,842,848
Detection limits 289,327,795
Dibenzofuran 408,806
1,2-dichlorobenzene 800
2,6-dichIorophenol 408
Dichotomous sampler 747
Diesel emissions 123,129
Diesel soot 141,147
Diethy ether 307
Differential-absorption lidar (DIAL) 563
Diffusion mechanisms 41
Diffusion denuder 753
Sampling system 73
6 diisocyanate 399
Dislodgeable residue 848
Dispersion 35,819
Distributed volumes 604
Dosimeter techniques 679
Dry deposition 263
Dubose Oil Products Company (DOPC) 488
Dust
Carpet 842
House 88,685,691,698
Mites 82
Surface 82
Dynamic dilution 333
895
-------
E
Emission Measurement Branch 414
Emissions 213
Filterable paniculate 213
Mutagenicity 213
Total extractable organics 213
Total mass 213
Empirical factors 35
Endotoxins 82
Enhanced monitoring 610
Entropy Environmentalist, Inc. 414,423
Environmental asbestos fibers (EAF) 494
Environmental Protection Agency (U.S. EPA) 15,
21,29,58,64,82,155,175,185,197,207,257,
296,301,307,327, 337, 345,363, 378, 384,393,
399,414,423,436,445,479,494, 545, 579,622,
651,679,691,698,706,718,789,795,813,842,
857, 869,880
500 series drinking water methods 795,64,
378
CLP SOW 327
Environmental Criteria and Assessment Office
82
OAQPS 313,414,445
QA system 345
Technology Transfer Network 378
Environmental Science and Engineering, Inc. (ESE)
313
Environmental tobacco smoke (ETS) 243,249
Epidemiologic studies 494,706
Esters 73
3-ethenylpyridine 249
Ethylbenzene 319,408
Ethylene Oxide (EtO) 429
ETS 494
Evaporative emissions 52
Exposure 161,167,185,733
Assessment 698
Limits 221
Human 175,451,651,691
Modeling 64
Extractable organic matter (EOM) 197,207
Extraction method 698
F
Farmer applicator 836,886
Farmer exposure 712,718,842
Federal Register 869
Fence line monitoring 276
FI detectors 257
Fiberglass 806
Fibers 806
Filterpack 747
Flame lonization Detection (FID) 123,289,393,
429,585,616,634
Flash vaporization 571
Florida Solar Energy Center 15
Fluoranthene (FL) 129,479
Fluorene (FLN) 129
Fly ash 141
Food and Drug Administration (FDA) 718
Method 211.13 718
Formaldehyde 249,579
Fossil fuel 464
Emissions 585
Foundation substructures 3
Fourier transform infrared spectroscopy (FTIR)
Remote sensor 557
Spectrometry 423
Test 414
Freon 113 221
Fugitive emissions 557,800
Fungi 82,671
896
-------
Fungicides 698,718
G
Gas and particle (GAP) sampler 874
Gas adsoiption 457
Gas and particle phase measurements 874
Gas chromatography (GC) 257,296,610,616,857
Analysis 761
ECD 685,718
Fast 276
FID/ECD 276
GC/IRD/MSD 123
GC/MD 301,307
Highspeed 276
HRGC/HRMS 436
Microchip 289
Oven 282
PID/ECD 276
Procedure 429
Gas Dilution systems 789
Gas phase 129,135,147,185,423,537,753,827
Gas transfer system 363
Gas vapors 579
Gaseous calibration standards 781
Gaseous components 727
Gasoline emissions 123
Gaussian equation 545
Gel permeation chromatography 718
Glycol dehydration units 319
Graphitized carbons 571
Great Lakes 874
Greenhouse warming 464
H
Halogenated hydrocarbons 571,646
Harvard School of Public Health 64,747
Hazardous air pollutants (HAPs) 161,167,313 414
423,563,579,775,806,869
Hazardous Organic National Emission Standards for
Hazardous Air Pollutants 869
Hazardous waste cleanup 345
HCFC-22 370
Health and Welfare Canada 747
Health Effects Institute-Asbestos Research (HEI-AR)
494
Health risks 167,243
Heating, ventilating, and air conditioning (HVAC)
15
Heavy metals 800
Herbicides 698,712,718
Hexachlorobenzene (HCB) 827,874
Hexachlorocyclohexanes (HCH) 827
a-HCH 874
Y-HCH 874
Hexachloroethane 408
Hexamethylene-1 399
High-Perfbrmance Liquid Chromatography (HPLC)
399
UV detection 399
High Volume Small Surface Sampler (HVS3) 82,
HNO3 457,470,741
HONO 727,741
Humidified
Nitrogen 333
Zero air 333
Humidity 337
Hydrocarbons (HC) 123,257,282,563,579,585,
610,616,628
Hydrogen
Chloride 806
Peroxide (H^) 537
Hygiene 836,886
897
-------
I
ICAL 100
Image method 21
In-situ
Continuous analyzers 775
Measurement 557
Solidification 800
Indole 221
Indoor air 41, 52,249,451,494,501,657,671,712,
727,733
Contaminants 249
Pollutants 21
Quality 58,82,243,249
Samples 640,761
Surfaces 58
Industrial Source Complex Dispersion Model 488
Industrial Source Complex Short Term (ISC2) model
29
Infrared 563
Differential-Absorption Lidar (DIAL) 563
Inhalation exposure 842
Inorganic gases 457
Insecticides 698,712,718
Instrumental neutron activation analysis (INAA)
509
Integrated Air Cancer Project (IACP) 175,185,190,
207
Intergovernmental Panel on Climate Change (IPCC)
464
Internal standards 863
International Technology Corporation (IT) 869
ISCST2 model 35
Isocyanates 399
J
Johnson Space Center
Toxicology lab 227
K
Kansas State University (KSU) 545
Ketones 167
Kinetic studies 147
L
Laboratory spike results 698
Landfill 479,806
Gas 269
Laser 563
Spectroscopy 769
Lead(Pb) 185,800,806
Leak Detection and Repair (LDAR) program 869
Loop injection 863
Louisiana Department of Environmental Quality
(LDEQ) 337
Lung cancer risk 494
M
Management Systems Reviews (MSR) 345
Manned spacecraft 221,227
Manual systems 775
Mass
Accommodation coefficient 58
Balance models 52
Selective detector (MSD) 640
Spectrometry (MS) 94,100,237,269,301,307,
313,327,337,414,557,604,640,646,685,718,
769,795,806,863
Massachusetts Institute of Technology (MIT) 509
Mercury (Hg) 509,515
Metabolites 671
Metallic elements 207
Metals 161,167,313,479,488,806
Heavy 800
Trace 185
898
-------
Meteorological
Condition 531,537
Considerations 819
Data 29,35,545,622
Measurements 155,813
Meteorology 35,52
Methane 479
Methanol 307,393
Dispersion 52
Sampling train (MST) 393
Method
0010 111,436,880
18 296,640
301 111,393,399,423
431 429
5040A 100
507 718
508 718
515 718
8240 100
Surrogate 301
Methoxyphenols 147
Methylene diphenyl diisocyanate 399
Methylmercury (MeHg) 519
Methylnaphthatene 408
Mexico 616
Mexico City 616
Microbial aerosols 243
Microenvironment 727
Microorganisms 671
Microprocessor control 775
Microtrap 296
Mobile sources 185,207,257,531,579
Emissions 175,190
Modeling 35,58
Moisture Management System (MMS) 307
Mold 671
Monitoring 243,563
Air 276,571,863
Ambient 445
Saturation 445
Motor vehicle emissions 197
Multi-mode samplers 775
Multidimensional chromatography 628
Multiple Linear Regression (MLR) 207
Multisorbent collection tubes 761
Multizonal models 52
Mutagenicity 190,207,213
Mycotoxins 671
N
NaCl 470
Nafion permeable membrane drier 337
Naphthalene 111,129,408,479,806
National Aeronautics and Space Administration
(NASA) 221,227
National Ambient Air Quality Standards (NAAQS)
363,445,479,551,813
National Cancer Institute (NCI) 691,698,706,718,
842
National Council for Air and Stream Improvement
(NCASI) 393
National Institute of Environmental Health Sciences
(NffiHS) 706
National Institute of Standards and Technology
(NIST) 100,197,333,357,585
National Performance Audit Program (NPAP) 351
National Research Council 221
National Weather Service (NWS) 29
Natural gas 319
NCI/EPA Farm Occupation Exposure Study
(NEFOES) 718
899
-------
NESHAPS 363
NH 436
Nicotine 88,243,249
MIST Traceable Reference Materials (NTRM)
Program 357
Nitrapyrin 679
Nitrate aerosols 470
Nitrates 167
Nitric acid 470
Nitric oxide 789
Nitro-PAH (NPAH) 129
Nitroaniline
2-nitroaniline 408
Nitrocompounds 769
NiUogen(N) 227,789
Heterocycles 436
Nitrosamines 657
Nitrosodi-n-butylamine
N-nitrosodi-n-butylamine 408
Nitrosodimethylamine
N-nitrosodimethylamine 657
NO 155,249,378,531,657,769,806
NO2 129,155,249,525,551,769
Non-criteria pollutants 813
Non-methane organic compounds (NMOC) 610
Non-polar organics 307
Non-recovery coke ovens 111
Nonattainment areas 155,257,282,622,634,813
Noncryogenic analysis 761
Nonhalogenated hydrocarbons 571
NO., 123,129,155, 525,610,622,628,657,769,
813,819
NOy 531
NSPS 363
o
Octamethyltrisiloxane (OMTS) 221
Offgasing 221,237
Office of Air Quality Planning and Standards
(OAQPS) 313,414,445
Oil burning 190
Oil furnace emissions 213
Olefins 610
Open-path remote sensing 563
Open path analyzers 551
Ultraviolet spectroscopy 551
Organic
Acids 73,167,663
Gases 457
Vapor Concentrator 337
Vapors 276
OSHA 857
Outdoor
Air 727
Exposure 733
Sample 761
Oxygen(O) 227,806
Ozone (O3) 58,123,129,155,243,282, 501,525,
531,537,551,585,610,622,634,813,819
Precursors 155,257,282,628
P-chlorotoluene 307
PAN 525
Paraffinic compounds 73
Particle phase 663
Paniculate 806
Cut-point 457
Matter 147,663
Phase 73,88,129,135,519,753
900
-------
Peltier cooling 628
Pentabrominated dibenzofuran (PeBDF) 135
Pentachlorophenol 488
Perfluorocarbon tracer (PFT) techniques 727
Permeability 21
Measurement 9
Peroxides 531
Peroxyacetyl 525
Persistent organic compounds (POC) 827
Pesticides 161,167,685,691,698,706,712,718,
836,848,874,886
Application 842
Exposure 836,842,886
Phenanthrene (PHE) 129,479,806
Phenol 111,806
Philadelphia Air Management Laboratory 64
Photochemical
Aerosols 651
Grid models 155,257
PAMS 155,257,610,634, 813
Pollution monitors 819
Reactions 141
Smog 501,663
Photodegradation 147
Photofragmentation 769
Photolysis 167
Photo lonization Detection (FED) 289
Plume dispersion model 557
Plume tracking 276
PM10 190,445,479,509,741
Point sources 29,35,545
Polar
Compounds 337
Organics 307
Pollutant gas analyzers 789
Pollution 563
Polybrominated
Dibenzo-p-dioxins (PBDD) 135
Dibenzofurans (PBDF) 135
Diphenyl ether (PBDPE) 135
Polychlorinatedbiphenyls(PCB) 827
Polycyclic aromatic hydrocarbons (PAH) 88,129,
135,141,147,436,451,488,685,806
Indoor-airborne 94
Polycyclic organic matter (POM) 175
Polynuclear aromatic hydrocarbons (PAH) 94,436
Polyurethane (PUF) 135,848
Portable sampling 761
Potassium (K) 185
Precipitation 515,519
Preconcentrators 795
Pressure 781
Pressure/Flow 21
Prevention of significant deterioration sites 351
Products of incomplete combustion (PICs) 175
Protocol gas 357,378
Suppliers 378
Purge & trap samples 795
Pyrene(PY) 129
Q
Quality
Assurance 351,378,384
Control 384
Quiescent water surface 800
R
R-ll 370
R-12 370
R-22 370
R-502 370
901
-------
Radian Corporation 429
Radiocarbon (14C) 197,207
Radium (Ra) 3
Radon (Rn) 3,9,15,21,41
RCRA 363
Real-time 557
Monitoring 227,276,296,451,769
Receptor modeling 185
Recycling 488
Refrigerants 370
Remediation activities 557
Remote 563
Reporting requirements 869
Research Gas Mixture (ROM) Program 357
Residential distillate oil combustion (RDOC) 175,
207
Residential exposure 712
Residential wood combustion (RWC) 175
Resonance-enhanced multiphoton ionization 769
Respirable suspended particles 243
Respitory exposure 679,712
Research Triangle Institute (RTI) 82, 370,393,470,
789
Research Triangle Park (RTF) 384,436,880
Risk 727
Analysis 494
Assessment 41,161,679
Salmonella microsuspension bioassay 190
Sampling 94
Inlets 457
SCREEN model 29
Scrubber monitoring 857
Selective Ion Monitoring (SIM) mode 337
Selenium (Se) 509
Semi-volatile hydrocarbons (SVHC) 123
Semi-volatile organic compounds (SVOC) 73,88,
111,161,167,190,263,408,479,604,753,800,
806,874
Sensitive detection 769
Silica 141
Sludge 800
SO2 378,457,464,551,741,789
SoU 3
Moisture 9
Permeability 9
Surface 842
Soil-gas chamber 9,21
Solanesol 88
South Dakota School of Mines and Technology 470
Southern Oxidants Study-Southern Oxidants
Research Program on Ozone Non-attainment
(SOS-SORP/ONA) 525,531
Southwest Research Institute (SwRI) 848
Space Station Freedom (SSF) 227
Spacecraft maximum allowable concentrations
(SMACs) 221
Spatial
Information 761
Patterns 733
Variations 64
Spruce needles 827
Stack
Effluent monitoring 880
Emission 213,296
Gases 269
Monitoring 857
Standard Reference Material (SRM) 357
State Implementation Plans (SIP) 622
902
-------
Static dilution 333
Stationary phase 296
Stationary sources 257,276,393,408,414
Emissions 190
Still vent streams 319
Styrene 806
Submarines 237
Subsurface contamination 41
Sulfate 733
Aerosols 464,470
Sulfur coal 741
Sulfnric acid 651
Supercritical Fluid Extraction (SFE) 94,408
Superfund 29,35,545,557
Surface sampling 848
T
T-SCREEN model 29
Ta 509
Tagging 869
Tedlarbag 296
Sampling 429
Temperature 781
Temporal 761
Variations 64
Tetrabrominated dibenzofuran (TBDF) 135
Tetrabrominated-p-dioxin (TBDD) 135
THC 806
Thermal
Conductivity detector (TCD) 289
Desorption 571
Time-activity patterns 501
TO-1 640
TO-2 640
TO-14 301,307,313,327, 333, 337,640,646,795,
863
Collection using sorbents 761
Compounds 269
Sampling criteria 775
TO 1 compounds 604
TO2 compounds 604
TO3 compounds 604
Tobacco smoke 88
Toluene 111,147,307,319,806
2,4-Toluene diisocyanate 399
Total Non-Methane Hydrocarbons (TNMHC) 123,
525
Total Non-Methane Organic Compounds (TNMOC)
610,616
Toxic gases 557,563
Trace
Atmospheric pollutants 781
Vapors 563
Tracer gas 3,15
Sulfur hexafluoride (SFj) 52
Tracer species 207
Transport 3,15,21,41,58,263, 515,519, 537, 819,
827
Triangle Laboratories, Inc. 436,880
Triple-path denuder (TPD) 470
TSCA 363
u
Undecanes
N-undecanes 307
University of Kansas (KU) 545
Urban Air Toxics Monitoring Program 313,604
Urban Airshed Model (UAM) 622
U.S. Department of Transportation (DOT) 243
903
-------
V
Vanillin 147
Vapor-liquid equilibrium (VLB) 781
Vapor/panicle partitioning 874
Vapor phase 73,263,319,451,509,769
VATMP 313
Versatile air pollution sampler (VAPS) 457
Virtual impactor 457,753
Volatile hydrocarbons (VHC) 207
Volatile organic compounds (VOC) 41,123,155,
161,167,185,227,237,249,263,269,296,313,
319,333,363,408,479,488,545,571,579,585,
604,616,622,628,634,640,761, 800,806,813,
819, 863
Automated sampling 775
Compliance testing 100
Integrated sample collection 775
Polar 161,167
Volatile organic sampling train (VOST) 100,296,
319,640
w
Wet deposition 263,515,519
Whole air samples 301,307
Whole body exposure 836,886
Wind 545
Windrose 819
Wisconsin Department of Natural Resources
(WDNR) 604
Bureau of Air Management (BAM)
Wood burning 185,190,197
Wood smoke 141,147,175,207
Woodstove emissions 213
Wright-Patterson Air Force Base (WPAFB) 479
X
XAD-2 94,408
Cartridges 190
Xylenes 111,319,408
M-xylene 111
O-xylene 111
P-xylene 111
904
-------
Author Index
A
Achman, D.
Agrawal, Susan R.
Aheam, Donald G.
Akland, Gerald G.
Allen, George
Allen, L.
Allen, Mark K.
Ames, M.
Anderson, Darcy J.
Aneja, Viney P.
Aras,N.K.
Au,L.
B
Ballard, Wendy L.
Bamford, Sarah
Barbour.RuthK.
Baugh, J. D.
Bell, R. W.
Berg, Neil J.
Berkley, Richard E.
Berry, M. A.
Berry, Maurice R., Jr.
263
848
671
691,706
727
94
604
509
470
525, 531, 537
509
94
863
58
451
185
94
445
276
82
718
Birla, Parag
Blancshan, Greg C.
Bond, Andrew E.
Bousquet, Ronald
Bowman, Jeffrey S.
Braman, R. S.
Brande, Ronald
Brauer, Michael
Brixen, Terri V.
Broadway, G.
Brook, Jeff R.
Brorstrom-Lund6n, Eva
Brousek, Louise M.
Brown, Charles K.
Brown, Jamie L.
Brubaker, Samuel A.
Bryd, Lee Ann B.
Brymer, David A.
Burton, Robert M.
Butler, William A.
C
Callahan, Patrick J.
135
423
691, 706, 712, 842
363
237
657
378
501,741
634
628
501, 747
827
237
337
571
3,9
551
301, 307
451,727,753
800
88
905
-------
Camann, David E. 685, £
848,886
Cardin, Daniel B. 269,
Carlson, Ruth L.
Carney, Kenneth R.
Carter, Ray E., Jr.
Cassamalli, A.
Chaffin, Charles T.
Chapman, R. E.
Chazin, Julian
Chen, Danhua
Chen, Shufen
Chu,M.
Chu, Shao-Hang
Chuang, Jane C.
Claxton, Larry D.
Clothier, Jacquelyn M.
Cole,E.C.
Coleman, W. M., IV
Collett, Chris W.
Conrad, F.W.
Crescenti, Gennaro H.
Crist, Howard L.
Crow, Sidney A.
Cui, Wenxuan
198,712,836,842,
282,327,333,795
307
289
457, 545
263
545
94
604
129
129
100
819
88,451
190, 207
706,836,842,886
82
249
494
249
52
363
671
73
Cupitt, Larry T. 161,
Currie, Lloyd A.
D
Darnell, Ginger
Das, Mita
Davis, Briant L.
Deng, Yun
Deschenes, John T. 269,
Detwiler, Andrew G.
Dorko, William D.
Dorosz-Stargardt, Geraldine
Dumyahn, Thomas S.
Duncan, John W.
Dunder, Thomas A.
E
Eatough, DelbertJ.
Eisenreich, S.
Elkins, Joseph B., Jr.
EUestad,T.G.
Elliot, Don
Evans, James M.
F
Fan, Zhihua
Fateley, William G.
Fellin, P.
167, 175, 185
197,585
869
537
470
470
282, 327, 333
470
585
155,813
741
52
423
73
263
351
651
15
319
129
545
263
906
-------
Ferguson, Sieve
Femandes, Carmo
Finke, Teresa S.
Flanagan, James B.
Foarde, K. K.
Fortmann, Roy C.
Foster, S. C.
Frank, Neil H.
Fung, Kochy
G
Geno.PaulW. 685,698,712,
Gerald, Nash O.
Gertler,A.W.
Geyer, Thomas J.
Giardino, Nicholas J.
Gonzales, Jesus A.
Gordon, Sydney M.
Gorman, Roy
Grosjean, Daniel
Grosjean, Eric
Grosshandler, Lisa M.
Grovenstein, J. D.
Gutschmidt, Kersten
Guyton, James L.
753
622
479
370
82
243
399
155
257
718,836,886
155
123
423
712, 836, 886
276
88,237,761
640
525
525
423
464
741
622
H
Hall,R.M.
Hammaker, Robert M.
Raiding, H.Jac 712,836,842,
Harris, D. Bruce
Harshfield, G.
Hartsell, Benjamin E.
Harvan, D.
Hatch, Peggy
Hayes, Elizabeth Ann
Hazard, Scott A.
Hazlett, James M.
Hecker, Jonnell
Hege,R.B.
Heinrich, Joachim
Hines, Avis P.
Hodson, Laura L.
Hoffman, A. J.
Honeycutt, Richard C.
Hoskinson, Steve
Howell, Mitchell
Huber, Alan H.
Hudak, Kim
Hudson, Jody L.
82
545
848,886
9,15
123
525
100
337
141
571
337
604
249
741
378
470
185
679
408
640
52
436, 880
545
907
-------
Huey, Norman A.
Hunike, Elizabeth T.
Hunt, William F., Jr.
Huntley, Roy
I
Iverfeldt, Ake
J
Jackson, Merrill
James, John T.
Jayanty, R. K. M.
Jesser, Richard
Johnson, Gary L.
Johnson, L. Ronald
Johnson, N. Douglas
Johnson, Terry
Jones, Christopher J.
Joseph, Darrell
K
Kagann, Robert H.
Kamens, Richard M.
Kanagasabapathy, V. M.
Keever, Jeffrey T.
Keller, Michael E.
Kellogg, R. B.
Kelly, Thomas J.
29,35
351
155
111
515,519
296
221, 227
393
646
345
470
874
679
301,307
161
557
129, 135, 141, 147
94
718
313
185
161, 167, 579
Kim, Deug-Soo
Kinner, Laura L.
Kitto, Abdul
Klinedinst, Donna B.
Klouda, George A.
Knight, Douglas R.
Knoll, J.E.
Koontz, Michael D.
Koutrakis, Pettos
Krasnec, Joseph
Krasnec, Ludovit
Kricks, Robert J.
Kulkarni, Shrikant
Kulp, Russell N.
L
Lai, Hung Jen
Lane, Dennis D.
Lane, Douglas A.
Lansari, Azzedine
Lawrence, Joy
Leafto, Hector
Lee, Cheng Peter
Leese,K.E.
Lemieux, Paul M.
Lemire, George W.
531
414
727
197,207
197,585
237
393,399
243
58,663,727,753
775
775
557
384
15
296
457, 545
874
52
663
227
622
82
80'6
769
908
-------
Levan, Leon
Levine, Steven P.
Lewis, Charles W. 175,
Lewis, Robert G. 88,
Lewtas,Joellen
Li, Wen-Whai
Liang, Chris S-K.
Limero, Thomas
Lindskog, Anne
Long, Marybeth
Lonneman, William A.
Lusis, M. A.
Lutes, Christopher C.
M
Machir, James
Mackay, D.
Mainga, Aaron M.
Majumdar, Tapan K.
Mannar, R.R.
Marotz, Glen A.
Marshall, Tun L.
May, Willie E.
McAndrew, James
McCrillis, Robert C.
McDonough, Susan
795
276
185,197,207,585
691, 698, 842, 848
175, 190
41
727,733
227
827
41
585,616
94
806
73
263
289
685
479
545
545
357
781
213
15
McDow, Stephen R.
McGaughey, J. F.
Menetrez, Marc Y.
Merritt, P. M.
Miller, Matthew
Milchell, William J.
Mitra, Somenalh
Mongar, Kevin
Moschandreas, Demetrios J.
Mosley, Ronald B.
Moyenuddin, Munshi
Munthe, John
Murdoch, Robert W.
N
Nagda, Niren L.
Nelson, C. J.
Nelson, P. R.
Noble, Judilh A.
Norris, James E.
o
Odum, Jay R.
Ogle, Larry D.
Olmez, I.
O'Neill, Hugh J.
Overton, Edward B.
129, 141, 147
399
3,9,15
842
378
357, 363, 378,
296
429
237
3,9,21
671
515,519
789
243
691,706
249
671
585
147
301,307,319
509
237
289
909
-------
p
Parmar, J.
Parmar, Sucha S.
Pate, B. A.
Peel, Thomas A.
Peralta, Jesus
Perry, Felton
Peterson, M. R.
Peterson, Robert C.
Pierson, W. R.
Pirolli, Richard
Plummet, Grant M.
Pollack, Albert J.
Preston, William
Price, Daniel L.
Pyle, Bobby
R
Raizenne, Mark
Ramamurthi, Mukund
Randtke, Stephen J.
Ratanaphruks, Krish, Jr.
Ratzlaff, Steven
Reif,DirkL.
Reiss, Richard
94
622
393
488
111
15
393
857
123
436, 880
414,423
161,167,761
640
671
15
501
161,167,451,579
457
3,9
488
319
58
Reiss, Sharon P.
Rhoderick, George C.
Ringler, Eric S.
Roberds, J. Michael
Roberts, DwightF.
Rogers, Keith
Rosenbaum, Wilf L.
Rowden, Scott E.
Rowe, Newt
Royals, Patricia
Rueter, Curtis O.
Rufe, Javier Tejeda
Russell, Theresa
Ryan, Jeffrey V.
Ryan, P. Barry
S
Sagebiel, J.
Sams, Robert L.
Santoyo, Maria Esther Rufz
Sausa, Rosario C.
Saxena, V. K.
Schewe, George J.
Schroeder, W.
Seeley, I.
Seila, Robert L.
863
585
445
718
313
781
494
488
436, 880
423
319
616
100,111
806
58
123
585
616
769
464
29,35
263
628
585,616
910
-------
Shah, Jitendra
Shanklin, Scott A.
Shaulis,CarlL.
Sheetz, L. H.
Sheldon, Linda S.
Shreffler.JackH.
Shrieves, Van X.
Sickles, Joseph E.
Simeonsson, Josef B.
Simonson, J. H.
Simpson, Orman A.
Sioutas, Constantinos
Snoddy, Richard
Spengler, John D.
Spicer, Chester W.
Staughsbaugh, Rick
Steger, Joette
Sterling, TheodorD.
Stevens, Robert K.
Stewart, John K.
Stteib, Ellen W.
Streicher.JohnJ.
Suh, Helen H.
Sykes, Alston
161
423
301
123
718
610
445
470
769
185
557
753
3,9
741
161, 167, 579
423
408
494
185, 457, 585
634
363
52
64
640
T
Tang, Hongmao
Tang, Y. -Z.
Taylor, Lyle H.
Thomas, Kent W.
Thomas, Mark J.
Thomson, M. S.
Tipler, A.
Tondeur, Y.
Tran.Q.
Trilk, Kelly
Tucker, J. R.
V
Vaccaro, Jim
Villalobos, Kevin
W
Waldman, Jed M.
Walsh, DebraB.
Warren, Sarah H.
Wasson, Shirley J.
Watts, Randall R.
Weinkam, James J.
Wichmann, Erich
Wiebe, H. A.
73
263
563
718
545
657
628
100
263
869
479
679
698
727,733
175, 190
190, 207
370
213
494
741
747
911
-------
Williams, Edwin L., n
Wilshire,F.W.
Wilson, Nancy K.
Wilson, William E.
Winslow, Michael G.
Witkowski,MarkR.
Wong, Jon
Woolfenden, E.
Worthy, Mike
525
393,399
88,451
727
313
545
327
628
423
Wright, Robert S.
Y
Yang, P.
Yoest, Helen
z
Zelenka, Michael P.
Zielinska, B.
Znamensky, Dmitry
Zweidinger, Roy B.
789
94
100, 111
64
123
781
52, 185, 190, 197, 207,
213
912
-------
Published by the Air & Waste Management Association
The following titles have been published recently by the Association. You may order any of them from the
Publications Order Department, A&WMA, P.O. Box 2861, Pittsburgh, PA 15230, (412) 232-3444, ext 3124.
Optical Remote Sensing
Applications fo Environmental and Industrial Safety Problems
1992. 555 pages. $45 for members, $70 for
others. Order code SP-81
The Response ot Southern Commercial
Forests to Air Pollution
1992. 333 pages. $55 for members,
$85 for others. Order code TR-21.
New Hazardous Air Pollutant
Laws & Regulations
1992. 320 pages. $55 for members,
$85 for others. Order code SP-82.
New RCRA Regulations for Industrial
Boilers, Furnaces and Incinerators
1992.232 pages. $40 for members,
$60 for others. Order code SP-80.
PM10 Standards and Nontraditional
Particulate Source Controls
1992. 1118 pages. $75 for members,
$115 for others. Order code TR-22.
Tropospheric Ozone and the Environment II --
Effects, Modeling and Control
1992. 955 pages. $60 for members,
$90 for others. Order code TR-20.
Measurement of Toxic and Related
Air Pollutants - 92
1992. 1022 pages. $60 for members,
$90 for others. Order code VIP-25.
Emission Inventory Issues
1993. 728 pages. $50 for members,
$75 for others. Order code VIP-27.
Standards, Certification and Training
of Environmental Managers and Auditors
1993. 136 pages. $35 for members,
$55 for others. Order code VIP-28.
Southern California Air Quality Study
Data Analysis
1993. 368 pages. $50 for members,
$75 for others Order code VIP-26.
The User and Fabric Filtration Equipment VI
1993. 324 pages. $55 for members,
$85 for others. Order code SP-84
Continuous Emission Monitoring
A Technology for the 90s
1993. 466 pages. $45 for members,
$70 for others. Order code SP-85.
Waste Combustion in Boilers
and Industrial Furnaces
1993. 328 pages. $45 for members,
$70 for others. Order code SP-86.
The Role of Meteorology in Managing
the Environment in the 90s
1993. 462 pages. $50 for members,
$75 for others. Order code VIP-29.
Municipal Waste Combustion
1993. 1108 pages. $75 for members,
$115 for others. Order code VIP-32.
Field Screening Methods for
Hazardous Wastes and Toxic Chemicals
1993. 1323 pages. $80 for members,
$120 for others. Order code VIP-33.
913
-------
PUBLICATION ORDER FORM
Please complete this form, detach it and mail it in an envelope with payment (in U.S. funds) or credit
card information to the Air & Waste Management Association, P.O. Box 2861, Pittsburgh, PA 15230. If
you require an invoice an additional handling charge will be added to your order.
NAME
AFFILIATION
ADDRESS
CITY
COUNTRY_
PHONE
STATE/PRO V.
ZIP/POSTAL CODE _
FAX
TITLE
ORDER CODE
PRICE PER COPY
Please indicate method of payment
D Check -- payable to A&WMA TOTAL
TOTAL
D Mastercard
DVISA
credit card number_
exp. date
signature
VIP-34
If you are not a member of the
Air & Waste Management Association...
...complete the form below and mail this card to learn how you can
profit from membership in the only technical society that exclusively
serves the air pollution control and hazardous waste technologies.
Postage paid if mailed in the United States.
Name
Affiliation
Address_
City
Country _
Phone
State/Prov.
Zip/Postal Code
Fax
VIP-34
-------
AIR & WASTE MANAGEMENT
ASSOCIATION
PUBLICATION ORDER FORM
Use the convenient order form on the reverse side to purchase Association publications
described in this book.
For member discount price shipment must be made to member address of record.
All others billed at list (nonmember) price. U.S. funds only.
Please complete all sections of the form and mail in an envelope with full remittance
or credit card information to:
Air & Waste Management Association
Publications Order Department
P.O. Box 2861
Pittsburgh, PA 15230
BUSINESS REPLY MAIL
FIRST CLASS PERMIT NO. 9946 Pittsburgh, PA
POSTAGE WILL BE PAID BY ADDRESSEE
Air & Waste Management Association
P.O. Box 2861
Pittsburgh, PA 15230
-------
Now Available...
Field Screening Methods for
Hazardous Wastes and
Toxic Chemicals
1993.1342 pages in two volumes. $80 for Association members,
$120 for others. Order code VIP-33.
The role of and need for field screening methods for
the identification and quantification of contaminants in
environmental media are growing rapidly. Field screen-
ing methods that generate real-time information on the
nature and extent of contamination improve the cost-
effectiveness of remediation.
This proceedings of an international symposium
cosponsored by the EPA Environmental Monitoring
Systems Lab and the Air & Waste Management Associa-
tion presents papers on the state-of-the-art science and
technology of field screening. Case studies on the use of
innovative methods are discussed. Papers from the
United States and around the world provide informa-
tion on recent advances in fiber optics, spectroscopic
techniques and chemical sensors. Everything you need
to know about field screening is available in this one
book.
To order this title use order card inside, or contact
A&WMA, P.O. Box 2861, Pittsburgh, PA 15230.
A list of recent A&WMA publications appears on page 913.
Another opportunity for technical exchange
provided by
AIR & WASTE MANAGEMENT
ASSOCIATION
------- |