Proceedings of the 1987 ERA/APGA Symposium on
MEASUREMENT OF
AND RELATED
AIR POLLUTANTS
-------
Proceedings of the 1987
EPA/APCA Symposium on
MEASUREMENT OF TOXIC AND
RELATED AIR POLLUTANTS
1200 Sixth AvenueySeattle, WA 98101
Jointly sponsored by the
U.S. Environmental Protection Agency's
Environmental Monitoring Systems Laboratory
and
APCA—The Association Dedicated to
Air Pollution Control and Hazardous Waste Management
Research Triangle Park, North Carolina
May 1987
-------
APCA Publication VIP-8
EPA Report No. 600/9-87-010
PROCEEDINGS OF THE 1987 EPA/APCA SYMPOSIUM ON
MEASUREMENT OF TOXIC AND RELATED AIR POLLUTANTS
This document has been reviewed in accordance with U.S. Environ-
mental Protection Agency policy and approved for publication. Mention
of trade names or commercial products does not constitute endorsement
or recommendation for use.
PUBLISHED BY:
APCA
P.O. Box 2861
Pittsburgh, PA 15230
1987
NOTICE
APCA
J FOUNDED IN 1907
The Association
Dedicated to
Air Pollution Control and
Hazardous Waste Management
ii
-------
PREFACE
A joint conference for the second straight year, cosponsored by
APCA's TP-6, TP-7, and ITF-2 technical committees and the Environ-
mental Monitoring Systems Laboratory of the U.S. Environmental Pro-
tection Agency was held in Research Triangle Park, North Carolina, May
3-6, 1987. The technical program consisted of 124 presentations, held in
13 separate sessions, on recent advances in the measurement and moni-
toring of toxic and related pollutants found in ambient and source
atmospheres. Covering a wide range of measurement topics and sup-
ported by 33 exhibitors of instrumentation and consulting services, the
symposium was enthusiastically received by more than 650 attendees from
the United States and other countries. This volume contains the papers
presented. The keynote address to the symposium is also included.
Measurement and monitoring research efforts are designed to antici-
pate potential environmental problems, to support regulatory actions by
developing an in-depth understanding of the nature and processes that
impact health and the ecology, to provide innovative means of monitor-
ing compliance with regulations and to evaluate the effectiveness of health
and environmental protection efforts through the monitoring of long-term
trends. EPA's Environmental Monitoring Systems Laboratory, Research
Triangle Park, North Carolina, is responsible for research and develop-
ment of new methods, techniques, and systems for detection, identifica-
tion, and characterization of pollutants in emission sources and in indoor
and ambient environments; implementation of a national quality assur-
ance program for air pollution measurement systems; and supplying of
technical support to Agency regulatory programs.
This conference, the seventh in a series arranged each year by the
EPA/RTP, but the second as a jointly sponsored conference by EPA and
APCA, was arranged with the following primary objective: to provide a
forum for the exchange of ideas on the recent advances for the acceptably
reliable and accurate measurement and monitoring of toxic and related
pollutants found in ambient and source atmospheres. The growing num-
ber of responses to this symposium represents an encouraging step in the
enhancement of our current measurement and monitoring capabilities.
R. K. M. Jayanty and
Seymour Hochheiser
Technical Program Chairmen
iii
-------
SYMPOSIUM COMMITTEES
TECHNICAL PROGRAM COMMITTEE
Cochairmen
Seymour Hochheiser, U.S. EPA
R. K. M. Jayanty, Research Triangle Institute
Donald F. Adams Sidney M. Gordon Charles H. Lochmiiller
Terry F. Biddleman W. F. Gutknecht Robert McCrillis
R. S. Bramen Judith Harris D. R. Scott
Steve Bromberg Ronald A. Hites Thomas T. Shen
Joseph J. Bufalini J. E. Knoll Franklin Smith
Cliff Davidson D. A. Lane Walter S. Smith
W, J, Dunn J. Lewtas
APCA TP-6 AMBIENT MEASUREMENTS COMMITTEE
Terry Sweitzer, Chairman
Douglas Lane, First Vice Chairman
Thompson G. Pace 111, Second Vice Chairman
R. K. M. Jayanty, Secretary
APCA TP-7 SOURCE MEASUREMENTS COMMITTEE
Billy J. Mullins, Chairman
Mark S. Siegler, Vice Chairman/Secretary
APCA ITF-2 TOXIC AIR POLLUTANTS
INTERCOMMITTEE TASK FORCE
David Patrick, Chairman
Patricia Bartholomew, Vice Chairman
Jitendra Shaw, Secretary
GENERAL CONFERENCE COMMITTEE
Cochairmen
John C. Puzak, U.S. EPA
G. Steve Hart, APCA
RESEARCH TRIANGLE PARK CHAPTER OFFICERS
Gerhard Gschwandtner, Chairman
John Rosen quest, Vice Chairman
Rodney Gibson, Secretary
SOUTH ATLANTIC SECTION OFFICERS
Thompson G. Pace III, Chairman
Elizabeth T, Bar field, Vice Chairman
Cal M. Ogburn, Secretary
Jane Beckett-Camarata, Treasurer
iv
-------
CONTENTS
Keynote Address Vaun A. Nemll 1
MEASUREMENT OF SEMI-VOLATILE ORGANIC
POLLUTANTS-AMBIENT AIR
Separation Sampling for Aerosol and
Gaseous Semi-Volatile Organics — An
Overview and Example Donald F. Adams 5
Further Gap Sampler Evaluation of
Semi-Volatile Chlorinated Organic Com-
pounds in Ambient Air
Field Comparison Study of Poly-
urethane Foam and XAD-2 Resin for
Air Sampling of Polynuclear Aromatic
Hydrocarbons
Non-Occupational Exposure to House-
hold Pesticides
Comparison of Estimated and Actual
PCB Vapor Exposure During a Soil
Excavation Project
GC/MI/FTIR-Improved Capability or
New Dimension in Analyses?
Gas Chromatography/Matrix Isolation
Infrared Spectrometry for the Analysis
of Organic Compounds in Air
Supercritical Fluid Extraction and
Recovery of PAH from Air-Borne Par-
ticulates
Analysis of Air Particulate Samples by
Supercritical Fluid Chromatography
Mobile Sources of Polycyclic Aromatic
Hydrocarbons (PAH) and Nitro-PAH:
Results of Samples Collected in a Road
way Tunnel Bruce A. Benner, Jr. 75
N. D. Johnson 12
Jane C. Chuang 21
Andrew E. Bond 30
James Neely 35
J. W. Brasch 42
Jeffrey W. Childers 57
Steven B. Hawthorne 63
Robert D. Zehr 69
v
-------
MEASUREMENT OF INDOOR TOXIC AIR CONTAMINANTS
Field Evaluation of an Adsorption-
Thermal Desorption Technique for
Organic Contaminants in Indoor Air
Performance Testing of Residential
Indoor Air Cleaning Devices
Evaluation of Organic Emissions to the
Indoor Environment via Small Chamber
Testing
David T. Williams
Bruce Tichenor
Chamber Studies Characterizing Organic
Emissions from Kerosene Space Heaters James B. While
EPA's Indoor Air Quality Test House
1. Baseline Studies
Semi-Volatile Organic Compounds in
Residences
Emission Rates of Volatile Organic
Compounds from Building Materials
and Surface Coatings
Inhalation from Volatile Organic Con-
taminants in Drinking Water
Sampling for Gas Phase Nicotine in
Environmental Tobacco Smoke with a
Diffusion Denuder and a Passive
Sampler
Organic Vapor Phase Composition of
Sidestream and Environmental Tobacco
Smoke from Cigarettes
Studies on Dust Depositions in Certain
Indoor and Outdoor Environments of
Secunderabad
Merrill D. Jackson
David J. Anderson
Lance A, Wallace
Lynn Wilder
Cecil E. Higgins
S. H. Raza
77
Mallory P. Humphreys 85
91
98
104
109
115
123
Delbert J. Eatough 132
140
152
ACIDIC DEPOSITION
Loss of Nitric Acid within Inlet Devices
for Atmospheric Sampling B. R. Appel 158
vi
-------
A Study of the Performance of Annular
Denuders and Preseparators
Evaluation of Methods Used to Collect
Air Quality Data at Remote and Rural
Site in Alberta, Canada
Comparison of Three Aerosol Sampling
Techniques and the Differences in the
Nitrate Determined by Each
Wet Deposition on Forest Canopy at
Mt. Mitchell, North Carolina
Dry Deposition of Oxides of Nitrogen
Establishment of the National Dry
Deposition Network
Measurement of Atmospheric Organic
Acids: Considerations Regarding Sam-
pling Artifacts and Potential Interfer-
ences
Teri L. Vossler
Eric Peake
Stanley A. Johnson
Vinod K. Saxena
D. Stacker
Barry E, Martin
168
174
183
189
195
201
Kochy Fung
208
MEASUREMENT OF VOLATILE ORGANIC POLLUTANTS -
AMBIENT AIR
Trifluorinated Polar Aromatics: Candi-
date Recovery Standards for Combus-
tion Studies
Mobile Tandem Mass Spectrometry for
the Characterization of Toxic Air Pollu-
tant (TAP) Sources
Portable Photoionization Systems with
Communications and Data Retrieval
Capability for Emergency Response to
Toxic Chemical Leaks and Spills
Evaluation of Photovac 10S50 Portable
Photoionization Gas Chromatograph for
Analysis of Toxic Organic Pollutants in
Ambient Air
Halocarbon Analysis of NMOC Air
Samples by GC/Megabore/Electron
Capture Detector
Charles H. Lochmuller 212
B. I. Shushan
218
M. Collins
224
Richard E. Berkley
A. L. Sykes
232
238
vii
-------
A Prototype Sampler for Preconcentrat-
ing and Transferring Volatile Organic
Compounds from Ambient Air to Passi-
vated Canisters
Update on Canister-Based Samplers for
VOCs
National Ambient VOC Data Base
June-September, 6-9 AM, Ambient Air
Benzene Concentrations in 39 U.S. Cit-
ies, 1984-1986
Nonmethane Organic Compound Sam-
pling and Analysis Program
Automated Measurement of Nonmeth-
ane Organic Compound Concentrations
in Ambient Air
Analysis of Volatile Organic Sampling
Train Samples Using Megabore Gas
Chromatography/Mass Spectrometry
Monitoring a Wide Range of Airborne
Organic Contaminants
Preconstruction Monitoring of Ambient
Air for Toxic Non-Criteria Pollutants at
a Municipal Waste-to-Energy Site
Grab Sampling as an Effective Tool in
Air Pollution Monitoring
AM VAN: Delaware's Comprehensive
Air Toxics Monitoring Van
Michael W. Holdren 245
William A. McClenny 253
Jitendra J. Shah 259
Robert L. Seila
265
Robert A. McAllister 271
Dave-Paul Dayton 277
T. A. Buedel 282
William R. Betz 761
Larry D. Ogle 289
Joseph P. Krasnec 294
Terri V. Henry 300
MEASUREMENT AND APPLICATION OF PHYSICAL AND
CHEMICAL PROPERTIES OF TOXICS
Estimation of Vapor Pressures for Non-
Polar Organic Compounds from Gas
Chromatographic Retention Data T. F. Bidleman 305
Estimates of Vapor-Particle Partitioning
from Some Individual Polychlorinated
Biphenyls and n-AIkanes in Denver Air William T. Foreman 311
viii
-------
Measurements of Dibenzodioxins and
Dibenzofurans in Ambient Air
Brian D. Eitzer
316
Measurement of Henry's Law Constant
for Selected Halocarbons in Dilute
Aqueous Solutions
Experimental and Calculated Reactivity
Parameters of Potential Environmental
Significance for Nitro-Polycyclic Aro-
matic Hydrocarbons
A Personal Computer Database for the
Chemical, Physical and Thermodynamic
Properties of the Polycyclic Aromatic
Compounds
Gary B. Howe
Arthur Green burg
Douglas A. Lane
GC/P1D Analysis of Headspace for Vol-
atile Organics Over Drinking Water J. Jerpe
322
329
341
348
DRY DEPOSITION-METHODS COMPARISON STUDY
Overview of the EMSL Methods Com-
parison Study
Reference Measurements of HN03 and
N03 by Tunable Diode Laser Absorp-
tion Spectroscopy
Measurement of Atmospheric Nitric
Acid and Ammonia by the Filter
Method and a Comparison to the
Tuneable Diode Laser Method
Measurements of Nitric Acid and Other
Dry Deposition Components Using the
Annular Denuder System and the Tran-
sition Flow Reactor
Passive Sampling Device Measurements
of N02 in Ambient Air
W. A. McClenny
G. 1. Mackay
Kurt Anlauf
J. E. Sickles
James D. Mulik
Luminox Measurements of Ambient S02 D. K. Bubacz
Comparison Across Different Methods
and Interpretation of Results
358
366
373
Richard J, Paur
379
387
398
404
lx
-------
ATMOSPHERIC TRANSFORMATIONS
The Behavior of Oxy-PAH and Nitro-
PAH on Atmospheric Soot Particles Richard Kamens 411
Comparative Mutagenic Activities of
Several Peroxyacyl Nitrates T. E. Kleindienst 420
An Experimental Technique for Measur-
ing the OH Rate of Attack on Particle
Bound PAH Zhishi Guo 426
Characterization of Organic Species on
Jet Engine Exhaust Particles Michael R. Kuhlman 435
Prediction of Photochemically Produced
Formaldehyde with Chemical Mecha-
nisms Developed for Urban Ozone
Systems Kenneth G. Sexton 442
The Inhibition of Photochemical
Smog —X. Model Calculations of the
Effect of (C2H5)2NOH on NOx~
C2H4-C4H8-2 Atmospheres Julian Heicklen 461
Intercomparison of Methods for the
Measurement of Ambient Levels of
H202 T. £, Kleindienst 467
CHEMOMETRICS AND ENVIRONMENTAL DATA ANALYSIS
Multivariate Calibration Using the Par-
tial Least Squares (PLS) Method W. J. Dunn III All
Multivariate Quality Assurance Program
for Dioxins and Furans Stephen E. Swanson All
Identifying Polycyclic Aromatic Hydro-
carbon Sources from Ambient Air Data
Collected in Urbanized Areas of New
Jersey R. Harkov 483
Use of a Rule-Building Expert System
for Classifying Single Particles Based on
SEM Analysis P. K. Hopke 496
A Composite Receptor Method Applied
to Philadelphia Aerosol Thomas G. Dzubay 502
x
-------
MEASUREMENT OF HAZARDOUS WASTE EMISSIONS
Difficulties and Efficiencies Encountered
with a Bulk-Air Sampling Scheme for
Characterizing Heavy Gas Releases Over
a Grass-Covered Site
Glen A. Marotz
Measurement, Assessment and Control
of Hazardous (Toxic) Air Contaminants
in Landfill Gas Emissions in Wisconsin Julian D. Chazin
Evaluation of the Flux Chamber Method
for Measuring Air Emissions of Volatile
Organic Compounds from Surface
Impoundments
Development of a Sampling Method for
Measuring VOC Emissions from Surface
Impoundments
Laboratory and Field Evaluation of the
Semi-VOST Method for Measuring
Emissions from Hazardous Waste
Incinerators
Evaluation of Microwave Techniques to
Prepare Combustor Source Samples
Development and Evaluation of Analyt-
ical Techniques for Total Chlorine in
Burner Fuels
Alex R. Gholson
Bart Eklund
J; T. Bursey
David A. Binstock
Alvia Gaskill, Jr.
508
516
522
529
5 35
541
547
INTEGRATED AIR CANCER PROJECT STUDY
Overview of the Integrated Air Cancer
Project Joellen Lewtas 555
The Collection of Neighborhood Air
Samples Impacted by Residential Wood
Combustion in Raleigh, NC and Albu-
querque, NM V. Ross Highsmith 562
The Source Apportionment of Carbona-
ceous Combustion Products by Micro-
Radiocarbon Measurements for the
Integrated Air Cancer Project (IACP) G. A. Klouda 573
xi
-------
Volatile Organic Hydrocarbon and Alde-
hyde Composition in Raleiph, NC Dur-
ing the 1985 Woodsmoke Study
Progress Toward Identifying Source
Specific Tracers
The Mutagenicity of Ambient and
Source Samples from Woodsmoke-
Impacted Air Sheds
IACP Emissions: Transformations and
Fate
Roy Zweidinger 579
Raymond G. Merrill 585
Larry D. Claxton 591
L. T. Cupitt 597
SOURCE MONITORING
Chemical Speciation of Nickel in Air-
borne Dusts
Correlation of Micorprobe Analysis
Data for Individual Particle Speciation —
Nickel Compounds from Stationary
Sources
Evaluation and Optimization of a Wet
Chemical Method for Determining the
Hydrogen Sulfide Concentration of Cyl-
inder Gases Employed for TRS CEMS
Quality Assurance
Evaluation of Ethylene Oxide Emissions:
Gas Chromatographic Analysis of
Scrubber Efficiency
Sampling and Analysis Method for
Determination of Cadmium in Station-
ary Source Emissions
A Critique of the Revision to EPA
Method 25
Vladimir J. Zatka
J. T. Rickman
A. K, Jain
Jeremy N. Bidwell
Thomas E. Ward
Kochy K. Fung
605
611
617
624
631
638
GENERAL SESSION
The Pyrolysis of Acetylene-Styrene Mix-
tures Between 450 and 550°C
C. Chanmugathas
642
xii
-------
648
653
659
665
667
671
677
685
691
697
707
Asbestos Fiber Loss from Air Sampling
Cassettes: A Study by Transmission
Electron Microscopy
A Novel 2.5 Cut Virtual Impactor
for High Volume PM10 Sampling
Development of EPA's Air Toxics Con-
trol Technology Center
Detection and Quantification of
Hydroxylated Nitro Polynuclear Aro-
matic Hydrocarbons and Ketones and
Hydroxylated Nitro Aromatic Com-
pounds in an Ambient Air Particulate
Extract
Advances in Continuous Toxic Gas
Analyzers for Process and Environmen-
tal Applications
Monitoring of Cement Kiln Emissions
Through Plants: A Case Study
William E. Longo
Robert M, Burton
Sharon L. Nolen
M. G. Nishioka
Hermann Schmidpott
P. K. Nandi
MEASUREMENT OF WOODSTOVE EMISSIONS
Non-Aqueous Ion Exchange Chroma-
tography as a Preparative Fractionation
Method for Short Term Bioassay Anal-
ysis of Wood Smoke Extracts Douglas A. Bell
Effects of Retrofit to Reduce Emissions
from Existing Wood Stoves R, W. Braaten
The Contribution of Residential Wood
Combustion of Respirable and Inhalable
Particulate Concentrations in Missoula,
Montana James E. Houck
Field Investigation of Catalytic and Low
Emissions Woodstove Particulate Emis-
sions, Efficiency and Safety Paul Burnet
The Environmental Protection Agency's
Accreditation Program for Wood Heater
Testing Laboratories Peter R, Westlin
xiii
-------
ENVIRONMENTAL QUALITY ASSURANCE
Importance and Benefits of Developing
Data Quality Objectives for Air Toxics
Data Collection Activities
Some Neglected Aspects of Environmen-
tal Quality Assurance
Quality Assurance for Precipitation Col-
lection and Measurement: The EPA
State-Operated Network
Quality Assurance Program for the Non-
Occupational Pesticide Exposure Study
(NOPES)
Development of Gaseous Hydrogen
Chloride Standards
Quality Assurance in the Nonmethane
Organic Compound Monitoring
Program
Gary L. Johnson 712
Raymond C. Rhodes 721
W. Cary Eaton 727
Miriam Lev-On 734
Robert B. Denyszyn 745
Robert E. Jongleux 755
Index
771
xiv
-------
THE ROLE OF TOTAL EXPOSURE MEASUREMENT IN RISK MANAGEMENT
KEVNOTE ADDRESS
Yaun A. Newt 1J
Assistant Administrator
Office of Research and Development
I am pleased to open the 1987 Symposium on Measurement of Toxics and
Related Air Pollutants sponsored by the Environmental Protection Agency
ant) the Air Pollution Control Association. I looked over the subjects you
will be discussing over the next few days, and was impressed by the
breadth of coverage and the diversity of topics on measurement systems and
approaches, including techniques for ambient air, indoor air, acidic dep-
osition, and source monitoring,
I believe that measurement serves a critical role in environmental
protection. If we view the dynamics of the regulatory process as a sys-
tem, then measurement of environmental conditions is one important "feed-
back mechanism" by which we can judge the effectiveness of these programs
in reducing the concentrations of pollutants that may come into contact
with man and the environment. Viewed in this way, the data from rrteasure-
ment systems help tell us how much progress we are making in our efforts
to control environmental pollution, and thus provide guidance for us in
modifying our approaches to make them more effective. In addition, as we
face some new chemicals of concern, measurements of pollutant concentra-
tions help us to identify the sources that need to be controlled.
Let me repeat a meteaphor that 1 have heard from monitoring special-
ists: one with which I feel particularly comfortable. The process is
much like a patient visiting a physician *iien he suspects he is ill. The
physician collects data essential in helping him diagnose the problem and
recommending a course of treatment. In environmental regulation, we make
similar measurements to characterize problems. Base upon this characteri-
zation, we develop a control approach. Finally, we use environmental mea-
surements to monitor our progress in correcting these problems. Clearly,
environmental measurements are part of the entire regulatory process.
Existing environmental programs articulate as their goal the protec-
tion of public health and welfare. Public health clearly refers to human
populations, while public welfare has been interpreted as referring to the
nortbuman component of the environment, for example, biota and materials.
If we consider the human component of environmental protection, then we
need to consider both the concentrators and the manner by which pollu-
tants actually reach people. Questions that are fundamental to estimating
the risk of pollutants to the general public are: "How many people are
exposed to the pollutant?" "What are the sources of these exposures?"
"What are the effects of these exposures?" Often we have ample informa-
tion about sources but do not really know the degree to which people were
1
-------
exposed to the contaminants released by these sources. Even if the ef-
fects of a. pollutant at a given concentration were known, we could not
determine the risk to public healtn, because we could not determine how
many people were exposed or to what cancentrations. In essence, our lack
of adequate exposure data prevents us from completing the risk equation.
For the criteria air pollutants, extensive monitoring programs are
maintained to measure pollutant concentrations outdoors. These data have
been useful for guiding existing regulatory programs, but is understood
that the data collected by these networks reflect only outdoor ambient
conditions that may be quite different from the actual concentrations to
which people are exposed. In the l97Q's, for example, a number of small-
scale personal exposure studies showed that pedestrians walking on down-
town streets, bicyclists, and others were exposed to concentrations very
different from the levels reported by official monitoring stations. Even
poorer correspondence was observed for toxic chemicals between outdoor
measurement's and personal exposures. These findings helped raised the
question: "What do fixed monitoring station data mean in terms of the
actual exposures of people?1'
In the 1980's, we have begun to make important progress toward find-
ing the missing exposure data needed to complete the risk equation and
answer these questions. If personal monitors can be used to measure the
exposures of pedestrians walking along downtown streets, then why couldn't
the same monitors be deployed to estimate the exposures of the entire
population of a city? To measure the exposures of everyone in a city of 3
million persons would be expensive, so a simpler approach had to be found.
This search for a better method gave birth to the idea of combining pro-
bability sampling with environmental monitoring. If one could measure the
exposures of a representative probability sample of the population, say
500 persons, then we then should be able to make inferences about the ex-
posures of the entire population. That was our initial reasoning.
A second requirement was a personal monitor that would be carried
easily be every participant. Moreover, we wanted a "smart" monitor -- one
that could distinguish between microenvironments such as commuting, home,
and warlc. A year or tv*o of hard work by ORD scientists designing a com-
bined monitor and manually operated data logger and averaging device led
to the successful construction of the desired instrument.
These concepts led to the carbon monoxide personal exposure monitor-
ing field studies conducted in the winter of 1982-83. In these studies, a
screening telephone survey first was conducted of 5,000 members of a re-
presentative random sample of the population of each city. Then, from
this first stage sample, a second stage probability sample was selected.
Ultimately, EPA collected 712 24-hour CO exposure profiles in Washington,
OC, and 900 24-hour CO exposure profiles in Denver, Colorado. Because the
2"lb. personal monitor measured concentrations on a continuous basis and
because each person carried a diary, it was possible to relate each per-
son's exposures to their activities and the locations they visited. Since
the ultimate goal of the CO standard is to keep rnost people's blood levels
under 2% COHb (carboxyhemoglobln), every participant supplied a breath
sample to provide a direct indication of CO body burden. Extensive sta-
tistical analyses and publications have resulted from these studies, with
numerous findings of importance to the Agency and the field of risk as-
sessment.
The Denver-Washington, DC, carbon monoxide exposure field studies are
illustrative of what is called the "direct approach" iti the newly emerging
2
-------
field of total human exposure assessment. These two field studies on CO
also demonstrated the Total Exposure Assessment Methodology, or TEAM
approach. In a TEM study, one measures directly a set of target pollu-
tants that come into contact with the individual, regardless of whether
they arrive by air, water, food, or skin. For CO, the only route of ex-
posure is by air, so air alone is monitored. For some toxic pollutants,
such as chloroform, the exposure routes include air, water, and bever-
ages.
There are four basic ingredients of a TEAM study: (1) a representa-
tive probability sample of the population of concern, (2) direct measure-
ment of the pollutant concentrations reaching these people through all
relevant media including air, food, water, and skin, (3) direct measure-
ment of body burden, and (4) direct recording of each person's daily
activities through diaries.
All components seek to determine, with known accuracy, the frequency
distribution of exposures of an urban population. At the present time,
the TEAM studies have collected data on the exposures of the population to
important toxic pollutants in four states, representing 700,000 persons in
seven cities. Additional studies now underway in Baltimore covering vola-
tile organic compounds and in Jacksonville covering pesticides will extend
this knowledge data base further. A major TEAM study of particles and
toxic metals has been mandated by Congress, and will be underway in FY
1987.
Direct measurement of personal exposure presents serious technical
challenges to the measurement community. Personal exposure monitors must
be small and unobtrusive, light-weight, quiet, rugged and insensitive to
vibration, able to operate reliably in untrained hands for long periods
without external power, capable of generating observations of known accu-
racy, and, ideally, capable of providing continuous data. The technical
and scientific difficulties in meeting such demanding criteria will be a
serious challenge to environmental measurement specialists. Despite the
many formidable technical obstacles to developing personal monitors, a
number of successful monitors already have been developed and field
tested. Even though some problems remain, notable successes have been
achieved in measuring carbon monoxide, volatile organic compounds, and
pesticides by personal monitoring. These successes not with standing,
additional methods development research is needed to develop continuous
monitors for nitrogen dioxide, passive carbon monoxide monitors, inhaled
(respirable) particles, active and passive monitors for formaldehyde,
volatile organic compounds, pesticides, and polyaromatic hydrocarbons and
related compounds.
Although the direct measurement approach is invaluable for determin-
ing, with known accuracy, the exposures of populations and the sources of
these exposures, an alternative approach -- the Indirect approach -- can
be used to estimate exposures without large-scale field studies. To apply
the indirect approach, one needs data on the microenvlronments people
ordinarily visit and the times people spend there. A mlcroenvironment 1s
a location of statistically homogeneous pollutant concentration, such as
CO concentrations in a parking garage. By combining mlcroenvironmental
concentrations with human activity patterns, it is possible to calculate
exposures. A human activity pattern-exposure model has been developed for
CO and has been used successfully to predict the exposure distribution of
the population of Denver. Despite success with this one model, no vali-
dated exposure-activity pattern models yet exist for any of the other
pollutants. Additional research is needed to develop, test, and validate
exposure models for toxic chemicals. This could yield the data needed for
making reliable risk estimates.
3
-------
Both the direct approach and the Indirect approach have the same
goal: to predict the frequency distribution of exposures of the target
population. With such a frequency distribution, it is possible to state
the proportion of the population above any concentration range. Using
these data, for example, a regulatory program can be designed with the
goal of maintaining some finite percentage of the population below some
target concentration. The data also have important policy imp!ications,
because they show that reducing exposures by strategies previously not
considered often can be more effective in reducing risk than are conven-
tional regulatory strategies. These studies are showing the greater im-
portance of indoor air and human activities in the exposure equation. For
example, recent exposure studies indicate that reducing human exposures to
tetrachloroethylene can be achieved more effectively by reducing the con-
tact people have with freshly dry cleaned clothes than by controlling
industrial point sources. Similarly, human exposures to benzene may be
reduced by reducing the contact that people have with cigarette smoke and
gasoline vapors.
These and many more discoveries are emerging from the new data that
we have been able to collect as a results of our total human exposure
research methodology. For the first time, human exposure to being exam-
ined as one integrated whole, considering the relative contributions by
air pollutants, water pollutants, food, and dermal contact. As the mea-
surement methods have progressed, we have been able to quantify exposure
-- all the sources responsible for exposure -- with a precision previously
never thought possible. That quantification process gets extremely close
to the individual person whose health we are charged with protecting. Let
me illustrate this precision with an example. The new TEAM study of par-
ticles currently being planned seeks to measure the frequency distribution
of tne population to inhaled particles. Thus, it should be possible to
determine the fraction of a city's population subjected to an exposure
exceeding 100 g/m3 concentration of inhalable particulate as a 24-hour
average during a season. These values can be compared with existing moni-
toring station readings and with existing standards. Activity pattern-ex-
posure models, when they are developed, can be used to extrapolate these
findings to other seasons and other cities. Ultimately, using these
models we may be able to estimate the fraction of the U.S. population
whose 24-hour average exposure exceed 100 g/m^ for any 8 a.m. to 8 a.m.
period during the year. Such quantitative goals give us a clearer target
of the regulatory programs required to protect public health.
All of the total exposure field studies conducted thus far have been
alike in one respect: they would not have been possible unless the neces-
sary monitoring systems and techniques were available. Thus, development
of suitable monitoring methods must receive priority, since developments
in this area provide the basic tools needed for all field surveys.
In closing, I hope I have been able to stress the importance of mea-
surement in the environmental protection process. 1 have used the example
of total human exposure measurement, since the technical demands placed on
measurement science are perhaps more demanding here than in other areas.
Although such measurements are extremely difficult technically, they pro-
vide environmental health programs with critical missing data required for
assessment of environmental health risk. Indeed, filling the exposure
knowledge gap is likely to have the greatest benefit of any of the compo-
nents of the overall risk equation 1n reducing our overall uncertainty
about overall risk and the sources of human risk to environmental pollu-
tion. If we are to implement a risk-based environmental management ap-
proach, then we must establish a stronger link between sources of pollu-
tants and the exposures and effects caused by these pollutants. Environ-
mental and exposure measurements will play a critical role in helpinq us
to established this link.
4
-------
SEPARATION SAMPLING FOR AEROSOL
AND GASEOUS SEMI-VOLATILE ORGANICS -
AN OVERVIEW and EXAMPLE
Donald F. Adams, Consultant:
SW 600 Crestview #2
Pullman, WA 99163
INTRODUCTION. Semi-volatile organic compounds (SVOCs) may be pres-
ent in the atmosphere as vapors and adsorbed on suspended particulate mat-
ter. Junge (1) suggested^that most organic compounds in clean air with
vapor pressures below 10 t0E§ exist in the vapor phase and compounds
with vapor pressures above 10 torr exist in the^particlg phase. The
SVOs generally fall in the vapor pressure range of 10 to 10 torr.
The vapor/particle (V/P) partitioning is associated with the vapor
pressures of the SVOs, the size and surface area o£ the suspended particu-
late matter, and the chemical composition of the suspended particulate mat-
ter, i.e. organic or inorganic, and the ambient temperature. The apparent
V/P is defined by the adsorbent-retained to filter-retained ratio (A/F).
How closely hiF represents V/P is not known. Many investigators
believe that A/F overestimates V/P due to the Loss of SVOs from the filter-
collected particles to the sampled air stream. It is also possible that
some of the collected particles may adsorb some vapor from the air samples.
SAMPLING. Even the most sensitive analytical techniques and instru-
mentation suitable for the low levels of SVO concentrations generally
encountered in ambient air samples require preconcertsrat ion from the bulk
air matrix prior to detection and quantification. Thus it is essential to
consider problems associated with available preconcentration techniques
before any sampling program is undertaken.
Sampling includes the collection of representative samples from a bulk
matrix as well as the sample preparation, handling, and storage prior to
the actual analytical measurement. Sampling errors are often large
compared with the errors of the measurement process. A valuable rule,
suggested by Youden, is: "once the measurement step uncertainty is reduced
to a third or less of the sampling uncertainty, further reduction in the
measurement uncertainty is of little value" (2).
5
-------
The significance of this is illustrated by the diminishing return
error model in which reducing errors in the most precise error source
yields a diminishing improvement in the overall error, Table I.
Table I
Diminishing Return Error Model
E /E E
s m o
1 1.40 x E
1.5 1.20 x ES
2 1.12 x E®
3 1.05 x E®
4 1.02 x ES
7 I.01 x ES,
r f 8 •
where: E = total error in the final result, E ™ errors contributed by
the measurement step, and Eg ¦> errors contributed"!^ the sampling steps.
The question of sample integrity is still of major concern in the
determination of atmospheric constituents. The accuracy of the reported
data may be limited by the sample, despite significant analytical advances.
The organic components of an air sample may be concentrated by: (a)
scrubbing in a gas impinger, (b) cryosampling, and (c) adsorption on a
solid or liquid-coated solid.
High-volume filtration has been widely used aitite abovit 1971. How-
ever, filter-derived samples will not provide a true representation of
particle/SVO loadings due to adsorption on collected particles and filter-
catch volatilization problems. Thus, high volume sampler systems usually
consist of a filter for collecting aerosols, followed by an adsorbent bed
or polyurethane (PUF) plug to collect vapors as described below.
In 1967, the use of porous polymer beads such as Porapak Q or S was
suggested for the room temperature concentration of volatile halocarbons
(3). Ihis eliminated the need Eor cryotrapping for many hydrocarbons and
provided for on-colunm gas chromatographic (GC) injection. Frankel and
Black, substituted Tenax GC for Porapak Q in 1976 because of its stability
at 375°C, ease of decontamination as compared with Porapak Q, and its
rejection of water vapor from the sampled ait (A).
PUF plugB were recommended in 1974 for the collection of SVOs (5).
Unfortunately, the polyester form of PUF is not inert and dissolves in
dilute base. Thus, certain classes of compounds may react with the PUF, be
altered, or irretrievably bound (6). PUF plug cleanup prior to sampling is
tedious, especially when electron capture detector/gas chromatographic
analysis is used. PUF plugs may be used over a wide range of sample flow
rates, permitting selection from a range of sampling times and flow rates
to obtain samples of the minimum size required for satisfactory analysis.
Important factors in the selection of adsorbents for sampling include-,
(a) collection selectivity, {b) collection efficiency, (c) recovery
efficiency, (d) sample storage stability, (e) analyte breakthrough volumes,
(£) temperature stability, (g) chemically inert toward analytes (artifact
formation), (h) initial cleanup procedure required, and (i) the influence
of pretreatments and air matrix components on the adsorption, desorption,
and decomposition properties of the adsorbent.
Several criteria for the evaluation of solid sorbents for the collec-
tion of gaseous organic pollutants include the field-spiking of duplicate
6
-------
sorbent tubes, sample blanks, and spiked tubes given identical sampling and
analytical treatment to show contamination and recovery factors (7).
Filter samplers suffer from a variable loss of SVOs from the collected
particulate matter. Sample losses vary with the ambient temperature and
the sampling time following an elevated fumigation episode during an eight-
or 24-hour sampling period because of the continuous and variable loss of
the collected SVOs to the sample air stream during the remainder of the sam-
pling period. The temperature effect is very important, since the vapor
pressures of SVOs nearly double with each 5°C increase. Whether sample
losses or gains predominate depends on how and when these variables change
during a sampling period as well as how the vapor concentrations vary
during each sampling period. Even with a backup vapor collector, the
derived analytical data will not realistically reflect the airborne distri-
butions, although tandem sample collection may efficiently retain both
particle and vapor phases.
Unfortunately, practical sampling methods have not yet been devised
which eliminate the carry through of vapor initially adsorbed on the filter-
collected particles to the backup vapor collector or which eliminate arti-
fact formation during the sampling and sample storage steps.
SEPARATION SAMPLING FOR AEROSOL AND GASEOUS SVOs. Prior to about
1970, separation sampling devices were infrequently used. Aerosols are col-
lected by several methods including filtration, centrifugation, electro-
static precipitation, and impaction. All these techniques suffer from the
the above-noted errors.
A recent separation technique uses a adsorbent-coated annular tube or
tubes followed by a filter and backup adsorbent cartridge or PUF plug (8).
A FIELD STUDY — SEPARATION OF AEROSOL AND VAPOR 2,4-D H0M0L0GUES.
The following discussion describes a practical situation in which separa-
tion sampling is an essential aspect to understanding and reducing the seri-
ous economic impact associated with aerosol/vapor 2f4~D herbicide trans-
port from the area of application to nearby and distant, off-target, sensi-
tive crops. To date, this nearly 30-year old problem in south central
Washington has not yet been definitively solved despite the numerous field
and modeling studies conducted by at least four research groups between
1963 and the present time. It is widely recognized that significant,
damaging quantities of aerially-applied herbicide formulations can drift
away from the target areas. Lesser quantities of off-target drift results
from ground applications. Studies as early as 1972 showed that spray drift
losses of 2,4-D from the target area may range from 40 to 75% (9).
Although the direct determination of the vapor/particle (V/P) ratios
is not yet possible, some estimates of V/P ratios of 2,4-D homologues have
been reported using a low volume rotating disk impactor (RDI) followed by a
midget impinger (10), or an XAD-2 cartridge (11).
A wide variety of 2,4-D homologies have been used to control weeds in
wheat. In the early 1960's, high volatile (HV) isopropyl and butyl 2,4-D
esters were commonly used in eastern Washington and Oregon wheat growing
areas. As 2,4-D injury to off-target crops was documented, HV ester use
was banned in Washington state. Low volatile (LV) isooctyl, propylene
glycol, and butyl ether 2,4-D esters and non-volatile (NV) acids, salts,
and amines of 2,4-D were substituted. However, in south central Washing-
ton, 2,4-D injury to the highly sensitive grape crop continues to be docu-
mented as recently as 1986 despite the use of LV and NV 2,4-D formulations
and changes in commercial application techniques.
7
-------
HV 2,4-D esters are characterized by fewer than six carbon atoms on
the side chain (isopropyl and butyl esters). LV 2,4-D compounds have six
more carbon atoms on the side chain (isooctyl esters and Jty.gher molecular
weight esters). The LV 2,4-D esters are within the 10 to 10 torr
vapor pressure range representative of the SVOs.
Today, there is still a need to determine the physical and chemical
forms of the 2,4-D compounds arriving at off-target, sensitive crops, to
provide formulation selection and application measures and constraints to
eliminate 2,4-D damage to sensitive crops, arid to permit continued use of
2,4-D formulations to control weeds in the surrounding wheat fields.
The 1962 and 1963 field studies of the total airborne 2,4-D-type
herbicide compounds in the south central Washington grape vineyards raised
many questions concerning the mode(s) of herbicide drift or transport from
the surrounding wheat fields to the grape vineyards:
• Is the alleged 2,4-D injury to grapes primarily related to
"over-the- fence" drift of aerosol spray droplets from nearby
wheat fields?
• Are the 2,4-D aerosol droplets volatilizing immediately after
release from the spray nozzles and then being transported to
adjacent or distant, off-target receptor sites in the vapor form?
• Is the herbicide subsequently volatilized from the treated crops
and field surfaces and then dispersed into the atmosphere as the
ambient temperature increases from its minimum during the early
morning application hours to the afternoon maximum temperature?
• Is some of the 2,4-D arriving via long distance transport of
vapor from distant application sites more than ten miles from the
receptor vineyards?
• Is some combination of both modes of transport from target to
receptor sites involved?
• Are herbicides sorbed on soil particles and transported from the
application site to distant receptor sites via wind-blown dust?
• Is the observed crop damage related to illegal application of HV
esters ?
• Is the observed crop damage related to legal applications of LV
or NV 2,4-D forms?
• Is the observed crop damage related to various combinations of
all three volatility classes?
To provide information concerning these questions in 1964, Bamesberger
and Adams (10) designed an air sampling device for the differential collec-
tion of aerosol and vapor fractions of airborne herbicides. This sampling
system consisted of a rotating disk impactor (RDI) for collecting the aero-
sol fraction down to one to three microns diameter, followed by a midget
impinger to collect the gaseous fraction and the small aerosols which pass
through the RDI. The design parameters of nozzle diameter, distance
between the nozzle and the impactor plate, and the sampling rate were calcu-
lated using conventional impaction theory (12). The RDI was based on a
rotary electrostatic precipitator in which sulfuric acid aerosol impinged
on a rotating, stainless steel disk washed with a stream of water (13).
Our design sampling rate was 1 L/min and the impactor jet was 0.55 ran
diam. placed 2 mm from the rotating impactor disk. The aerosol fraction
was retained in n-decane. The RDI was sealed with a press-fit glass cup
which also was the reservoir for the n-decane. The sampled air first
impinged on the impactor disk rotating at 1 rpm into the small liquid reser-
voir for the collection liquid. The impacted particles were retained in
the collection fluid. A Teflon squeege removed any adhering particles and
8
-------
dried the impaction disk surface. This reduced the wetted surface of the
disk and minimized vapor collection in the RDI. The ait sample then passed
into a midget impinger containing n-decane or n-decane/52 aqueous NaHCO^.
In 1974, the press-fit glass cup was redesigned with a threaded-seal
to reduce n-decane loss during handling and transport from field sampling
sites to the analytical laboratory. The midget impinger was replaced with
an XAD-2 filter cartridge to simplify field use (11).
The RDI has variable errors similar to those of filter sampling.
Gases in the sampled air impinge on the dry surface of the rotating disk
and briefly contact the surface of the sample collection fluid, permitting
some vapor phase collection in the aerosol collection stage of the RDI.
The wetted area of the rotating disk was minimized by placing a Teflon
squeege 1 cm above the collection fluid to remove the collection liquid
film from the disk during the remaining 200° rotation of the disk. The
squeege also provides partial separation between the sampled air stream and
the main body of the collection fluid, to reduce solvent evaporation and
minimize absorption of gases from the air sample.
Limited laboratory studies of the RDI showed that about 25% of the
highly volatile isopropyl ester of 2,4-D in a laboratory-generated atmos-
phere was collected in the RDI (10). About 17% of the slightly lower vola-
tile butyl 2,4-D ester (but still defined as an HV ester) collected in the
RDI, Table II (10). The relatively high collection efficiency shown for
the isopropyl ester was not considered a serious design deficiency for its
intended field use, because isopropyl ester had been banned from use by the
Washington State Department of Agriculture, and a ban on butyl ester was
under consideration. The LV isooctyl 2,4-D ester would be the most vola-
tile of the legal 2,4-D esters to be applied to the wheat fields. Subse-
quent field use, however, showed significant concentrations of isopropyl
and butyl 2,4-D esters through 1980, more than six years after HV esters
were banned in Washington.
Table II
Collection Efficiency for 2,4-D Type Herbicides
Compound
Concn..Range,
Mg/m
Midget
Impinger
RDI
Gaseous
0-100
Isopropyl 2,4-D
97.0
26.3
Butyl 2,4-D
97.0
17.1
Aerosol
0-200
Isopropyl 2,4-D
89.0
62.1
Butyl, 2,4-D
85.8
71.2
For comparison, losses have been^ reported^ from filters of about 50%
for anthracene and phenanthrene (10 to 10^ torr)_^to 10 to 35% from
fluoranthrene through benzo(a)anthracene (10 to 10 torr) (15) which
are in the vapor range for SVOs (1).
The following discussion is limited to the 1974 field data because the
data represent the most extensive two-stage sampling data available and the
greatest incidence of grape injury reported between 1964 and 1976 occurred
in 1974.
The cumulative average concentrations for the three volatility classes
for these three months show that n-butyl ester was the dominant 2,4-D
9
-------
compound in the 1974 samples. The RDI collected 522 of the average HV
2,4-D esters found. This compares with the laboratory collection of 26%
and 17% for isopropyl and butyl esters, respectively. During late May an^
early June, 1974, the average butyl e^ter concentration was 0.60 pg/m
and the isopropyl ester was 0.23 pg/m - a ratio of 1:3.5. Using this
ratio and the laboratory-determined collection efficiencies for these two
esters, an average 24% HV collection would be expected - just under half
the observed field collection.
The RDI samples contained only 25% of the total butyl ether and 60% of
the PGBE ether 2,4-Ds recorded, with an average of 56% of the total LV
esters found (10), less than one would expect if the aerosols were greater
than 3 micron diameter.
This wide range of collection efficiencies of the RDI in the field is
probably related to several factors including (a) variations in ambient
temperatures at the time of spray application and sample collection, (b)
distance between the sites of application and collection, and (c) some of
the other questions listed above such as the RDI collection of wind blown
soil particles, etc.
Unfortunately, sampling with the RDI was discontinued following the
1974 field stud^, because of sponsors' financial constraints. Field
samplings between 1975 and 198Q were conducted with single-stage XAD-2
resin samplers. Although this change in sampling technique reduced the
analytical costs by 50%, it precluded the accumulation of adequate
herbicide P/V ratio transport data to provide sufficient information to
assist in understanding the transport questions noted above.
A 1982 study relied extensively on (a) transport modeling within a 10
mile radius, (b) an unlikely time/concentration dose injury threshold rela-
tionship assumption for the more sensitive wine grape varieties, (c) an
underestimated source strength, and (d) zinc sulfide tracer releases (16).
From 1982 through 1987, the major concern has been focused on aerial
applications within 10 miles of the grape vineyards. Many more acres can
be sprayed aerially irt a day than via ground applications, thus increasing
the daily source concentration and emission rate. This raises questions
about the assumed source strength for the model. Despite the conclusion
from the fluorescent tracer studies that "close-in applications" (within 10
miles of vineyards) were primarily responsible for the grape injury, the
question remains whether inorganic zinc sulfide particles appropriately
mimic the many complex actions of semi-volative organic compounds during
field application and for a period of several days thereafter.
Many questions are still unanswered concerning the fate and drift
distances for the LV and NV 2,4-D formulations presently used, which still
cause damage to the grape vineyards. No field sampling studies have been
conducted since 1980 to determine which 2,4-D compounds are present in the
damaged vineyards, and no studies since 1974 have provided information
which might reveal whether the herbicides associated with the damage are
primarily in the aerosol or vapor form, or in some ratio of the two forms.
REFERENCES
1. C. E. Junge, "Basic Considerations about Trace Constituents in the
Atmosphere as Related to the Fate of Global Pollutants," Fate of
Pollutants in the Air and Water Environments, Part X, pp, 7-25.
I. H. Suffet, Ed. John Wiley and Sons, New York. 1977.
10
-------
2. W. J, Youden, "The Role of Statistics in Regulatory Work," J. Assoc.
Off. Anal. Chem.- 50: 1007-1013 (1967).
3. P. W, Williams and M. E. Umstead, "Determination of Trace Contaminants
in Air by Concentrating on Porous Polymer Beads," Ana 1. Chem.. 40:
2232-2234 (1968).
4. L. F. Frankel and R. F. Black, "Automatic Gas Chromatographic Monitor
for the Determination of Parts-per-billion Levels of Bis(chloromethyl)
Ether," Anal. Chem. 48: 732-737 (1976).
5. T. F. Bidleman and C. E. Olney, "High-volume Collection of Atmospheric
Polychlorinated Biphenyls," Bull. Environ. Contam. Toxicology, 11:
442-450 (1974).
B. C. Turner and D. E. Glatfelty, "Field Air Sampling of Pesticide
Vapors with Polyurethane Foam," Anal. Chem. 49: 7-10 (1977).
7. R. G. Melcher, R. R. Langner and R. 0. Kagel, "Criteria for the Evalua-
tion of Methods for the Collection of Organic Pollutions in Air Using
Solid Sorbents," Am. Ind. Hyg. Assoc. J., 39: 349-361 (1978).
8. N. D. Johnson, S. C. Barton, D. A. Lane and W. H. Schroeder, "Further
GAP Sampler Evaluations of Semi-Volatile Chlorinated Organic Compounds
in Ambient Air," paper presented at the 1987 EPA/APCA Symposium on
Measurement of Toxic and Related Air Pollutants. Research Triangle
Park, North Carolina. May, 1987.
9. L. E, Warren, "Volatility and Drift of Herbicides," paper presented at
the Montana Herbicide Workshops, Oct. 1972.
10. W. L. Bamesberger and D. F. Adams, "Collection Technique for Aerosol
and Gaseous Herbicides," Agric. and Food Chem., 13: 552-554 (1965).
11. S. 0. Farwell, L. L. Fox, D. F. Adams and E. Robinson, "Atmospheric
Drift of 2,4-D in the Lower Yakima Valley 1975." Air Pollution
Research Section, Washington State University, Pullman, Washington.
January, 1976.
12. W. E. Ranz and J. B. Wong, "Jet Impactor for Determining the Particle-
size Distribution of Aerosols," Arch. Ind. Hyg. Occupational Med., 4:
462-477 (1952).
13. c. Schadt, P. L. Magill, R. D. Cadle and L. Nye, "An Electrostatic
Precipitator - for the Continuous Sampling of Sulfuric Acid Aerosols
and Other Air-Borne Particulate Electrolytes," Arch. Ind. Hyg.
Occupational Med., 1: 556-564 (1950).
14. "Analytical Data Supplement to the Atmospheric Drift of 2,4-D in the
Lower Yakima Valley - 1974." Air Pollution Research Section,
Washington State University, Pullman, Washington. August, 1975.
15. J. F. Galasyn, J. F. Hornig and R. H. Soderberg, "The Loss of PAH From
Quartz Fiber High Volume Filters", J. Air Pollut. Control Assoc. 34:
57-59 (1984).
16. W. Knuth, H. Girous, D. Moellenberndt, "A Historical Correlation and
Strategy Plan for the Reduction of 2,4-D Damage in Southeast
Washington." Environmental Systems and Service, Feb. 1982. Kelseyville,
California.
11
-------
FURTHER GAP SAMPLER EVALUATIONS OF
SEMI-VOLATILE CHLORINATED ORGANIC COMPOUNDS
IN AMBIENT AIR
N.D. Johnson and S.C. Barton
Ontario Research Foundation
Sheridan Park Research Community
Mississauga, Ontario L4K 1B3
D.A. Lane and W.H. Schroeder
Atmospheric Environment Service
4805 Dufferin Street
Downsvlew, Ontario M3H 5T4
Abstract
Uncertainties exist in defining deposition mechanisms of airborne
semi-volatile chemical species to land and water due, in part, to
existing limitations in sampling methods. A gas and particle (GAP)
sampler has been developed based upon an annular diffusion denuder
technique to separate co-existing phases of high molecular weight
organochlorine compounds occurring in the atmosphere. A series of field
evaluations has been conducted in the vicinity of Lake Ontario to further
evaluate the performance of the sampler and experimentally determine the
phase distribution of target compounds such as hexachlorobenzene (HCB)
and hexachlorocyclohexanes (a- and Y-HCH). During these field
studies attention was given to quality assurance aspects such as
triplicate measurements, routine field blanks and dual capillary column
gas chromatography analysis. Data suggest than an efficient phase
separation of such compounds was achieved by the sampler design and, on
the basis of denuder-d1fference measurements both during the summer and
fall months, the largest fraction of these target compounds existed In
the vapour phase. Methods to optimize the utility of the sampler are
discussed in addition to field results.
Introduction
Atmospheric transport and deposition is considered to be an Important
pathway by which many toxic chlorinated compounds such as PCBs and
pesticides enter the Great Lakes and other water bodies
12
-------
that are remote from Known sources (U. However, Input mechanisms
for such semi-volat1le species that may co-exist in the atmosphere
as vapour and as adsorbed components of particulate matter are not
well defined. Most investigators have concluded that compounds with
vapour pressures in the range of 10~4 to 10-5 mm Hg (at room
temperature) occur predominantly in the vapour state. On the other
hand, because of the numerous factors that can affect
vapour/particle equilibria (eg. temperature, particle loadings) and
the complex nature of airborne particulate matter, it 1s recognized
that currently-available sampling techniques may not adequately
distinguish between phases due to probable disruption of the
equilibria during filtration. For example, conventional sampling
techniques, such as high-volume filtration/adsorption methods, have
been used to operationally derive gas/particle relationships
(Z-4), jj. uncertain whether amounts retained by these
collection media are representative of the actual gas- and
particle-phase distribution since vapour constituents could either
adsorb or desorb from filtered particles during sampling under
certain conditions. A knowledge of the degree of partitioning
between phases In ambient air 1s necessary for adequate assessments
of either potential human health effects or environmental behaviour,
Impact and ultimate fate of such airborne constituents.
A prototype GAP sampler, based on a denuder difference system, has
been developed to simultaneously collect the gas and particle
fractions of several persistent organochlorlne compounds in ambient
air. The Initial development and laboratory evaluation of this
sampler has been reported previously (5>°>. The major objective
of this study was to determine the gas- and particle-phase
concentrations of selected persistent chlorinated organic compounds
1n ambient air 1n southern Ontario and at a remote location 1n
central Ontario. The sem1-volat1le compounds, HCB a- and Y-HCH,
selected for field measurements are often found 1n rainwater and
other environmental samples despite their limited use 1n North
America (7,8).
Methodology
A schematic of the prototype GAP sampler 1s shown 1n Figure 1. This
sampler comprises a multi-annular diffusion denuder assembly, coated
with a binary crushed Tenax/s1l1cone gum coating (to trap the
vapour-phase components of the target compounds), that 1s Integrated
with a dichotomous sampler (to collect fine and coarse particulate
matter). Packed Tenax adsorbers (2 in series) are positioned behind
each filter to collect any material that may escape the filters (eg.
desorbed vapour during filtration). This unit 1s operated in
parallel with a conventional filter/adsorber sampler (I.e. reference
unit to obtain the total airborne constituent concentration). The
net difference in amounts collected by filters and adsorbers of both
units provides the constituent vapour fraction and the combined
filter/adsorber analyses of the dichotomous unit Indicates the
particulate-assoclated fraction. Earlier laboratory studies have
demonstrated practically removal efficiency and retention of HCB and
lindane by the denuder upon exposure to vapours generated at
concentrations that might be expected in ambient air.
13
-------
In this study, field measurements {Involving replicated sampling
over 24 to 48 hour periods) were conducted at several locations In
southern Ontario and at a remote (background) site In the Turkey
Lakes watershed north of Sault Ste. Marie (see Figure 2). The first
set of measurements were taken during July 1986 1n the western
portion of Lake Ontario. Air sampling and meteorological equipment
was operated simultaneously on the shoreline near
N1agara-on-the-Lake and on board the CSS Bayfield (a CCIW Great
Lakes research vessel) while 1t was anchored approximately 5 km
off-shore. Similar on-shore/off-shore measurements were conducted
in the eastern part of Lake Ontario near Sand Banks Provincial Park
and an integrated sample was collected on the ship while it
traversed between the western and eastern sites. Triplicate sample
sets were taken 1n the early autumn at the remote Turkey Lakes area
site. Other field measurements 1n southern Ontario were made at a
suburban site In Sheridan Park, MHsissauga (during autumn, 1986 and
winter, 1987) and 1n Stoney Creek (near Hamilton) at a site
approximately 800 meters from an agricultural chemical producer
(during the early part of the winter).
Analytical methods to determine the target compounds Involved
extraction of filters and adsorbers, extract cleanup and analysis by
dual capillary column gas chromatography with electron capture
detection. In all Instances, the lowest of the two column results
was assigned with the assumption that any differences 1n the values
between columns represented analytical interferences. In many
cases, results between columns were similar while substantial
differences were evident on some occasions. The average difference
between columns for each compound was found to be : 27% HCB, 15%
a-HCH and 57% Y-HCH.
Results and Discussion
Field studies took place over a seven month time period and mean
temperatures during sampling Intervals ranged from -3.7 to 22.9°C
with hourly extremes from -11 to 34°C. Except on two occasions,
under very hot summer conditions, temperature extremes were at or
below temperature conditions selected In laboratory tests 1n which
high vapour collection efficiency for HCB and lindane by the denuder
was demonstrated. The general background nature of the sampling
sites 1s evident from the low suspended particulate concentrations
(1e. <10 ym in size) shown in Table 1. The lowest loadings were
encountered at the remote site as might be anticipated. The small
difference In the overall mean particulate concentrations between
the conventional and GAP samplers (1e. -5%) confirms the earlier
findings that no substantial particle losses occurred on average 1n
the denuders.
Total airborne concentrations of HCB, a- and Y-HCH were
generally in the sub-nanogrwn per cubic metre (of air) range based
on conventional sampler measurements performed at all of the sites
(see Table 2). Although the mean HCB concentration at Turkey Lakes
was somewhat lower than the average amount measured at other
14
-------
locations 1n southern Ontario, results were generally similar
between sites. These findings are consistent with values reported
from other studies. Lindane concentrations were about an order of
magnitude lower than either oc-HCH or HCB concentrations and were
frequently at or near the detection limit of the analytical method
used. The higher abundance of a-HCH relative to lindane may
reflect environmental (perhaps photochemical) transformation of
lindane to the more stable a-1somer. On the basis of measurements
conducted in southern Ontario (including preliminary results of
measurements in earlier studies) at various temperatures, the total
airborne concentrations of all three target compounds were found to.
decrease with decreasing temperatures. The effect of temperature on
airborne concentrations 1s more apparent from data shown in Table 3
where average concentrations were grouped into ~5°C increments.
For example, a 2-3 fold reduction was evident for each compound over
a temperature decrease of ~20°C. It is presumed that the
concentration decrease reflected less volatilization of these
compounds from deposits on surfaces at cold temperature conditions
rather than decreased usage or production during winter.
Individual measurements of the target compounds and the derived
partlculate-associated fractions at each sampling site are shown in
Table 4. In most instances, the values represent an average of two
or three measurement results and appear 1n approximate chronological
order of the study. Generally favourable agreement was found
between duplicate conventional sampler results of the total
concentrations. It can be seen that little difference occurred
between days at each site and results were often similar between
sites when sampled at the same time of year. These data suggest a
a^'al, low-level, ubiquitous occurrence of these constituents in
this study area. Because only one set of samples was taken at each
site during preliminary onshore/offshore measurements, it is
considered premature to deduce any potential trends between the land
and water sites. Lindane concentrations were found to be highest
near one suspected Industrial source that 1s known to formulate
organochlorlne pesticides. It should be noted that sample Sets 1
and 2 at this site were taken during early winter 1985 and Sets 3
and 4 during early winter 1986. At other sites, lindane
concentrations were low with greater variability since amounts were
near the measurement sensitivity. In practically all cases, these
compounds were collected efficiently (1e. by the first of two Tenax
adsorbers positioned In series) 1n the conventional filter/adsorber
sampler and only traces occurred on few occasions in filtered
particulate matter. Except for lindane, amounts in field blank
cartridges usually represented less than 10X of the total amounts
detected. However, spurious results occurred In the occasional
blank cartridge that were sometimes suppressed with the dual column
analytical approach. On the other hand, blank values were of
greater significance in deriving the partlculate-associated
fractions since amounts detected in adsorbers positioned after the
denuder dften were low <1e. approaching blank levels).
15
-------
The average particulate fractions, derived by denuder difference,
also are shown in Table 4 for each site. On several occasions,
replicate (eg, duplicate or triplicate) measurements of this
fraction with the GAP sampler were similar especially when
considering that this fraction is derived by difference from
analyses of several components of the system. From the results of
these air measurements, 1t appears that the substances being
investigated occurred predominantly In the vapour phase during warm
weather as has been found by other Investigators. However, under
winter conditions Cat the Sheridan Park site), a greater fraction of
these high molecular weight, semi-volatile constituents appeared to
be associated with the airborne particulate matter. That is,
greater amounts were detected in cartridges located after the
denuder, which presumably resulted from volatilization from filtered
particles during sampling. Under the cold weather conditions, the
particulate-assoclated fractions were found to be more variable
between days and between replicated measurements. In some
instances, only single column GC results were available or other
sources of contamination in sampling components were possible and
such results (noted with *) were considered to be upper limits.
Nevertheless, these data suggest that the denuder efficiently
removed the vapour fraction during sampling which was found In
earlier laboratory studies.
Conclusions
Specific findings of this field evaluation using an alternative
sampling method for semi-volatile organochlorine constituents are;
- particle losses In the denuder were usually small (~41 on
average) and the denuder was demonstrated by both lab and field
studies to efficiently remove vapours of the target compounds
(HCB, a- and Y-HCH). These data suggest an effective
gas/particle separation by the design,
- measured target compound concentrations (sub ng/m^) at the
sites in southern Ontario were similar to other reported values
and were found to decrease at low temperature conditions.
Except near a suspected source, lindane was normally an order of
magnitude lower than the more stable a-HCH form,
- at moderate temperatures, the mean particulate-associated
fractions in the study area, derived by denuder-difference with
the GAP sampler, were: ~4% HCB, ~7% a-HCH and <10% Y-HCH.
Greater but more variable particulate fractions were evident
under cold temperature conditions at a suburban site. These
field data infer that such constituents occurred predominantly
in the vapour phase at mild temperatures but that particulate
depostlon may be of somewhat greater significance to the Great
Lakes during colder seasons than generally perceived.
Refinements to the sampling system (eg. use of Florlsil cartridges,
simplification by only selective use of the dlchotomous sampler and
thermal desorptlon of the denuder) are presently being tested in
order to Improve the accuracy of measurements. Similarly, tests to
expand the application of the GAP sampler (eg. PCBs) are underway.
16
-------
Acknowledgements
The authors wish to gratefully acknowledge the support of this study
by Dr, Wm, Strachan, National Water Research Institute, and the
technical contribution by Mrs. M.J. Hanley and analytical staff
members of Ontario Research. The study was conducted with the
financial support of the Atmospheric Environment Service of
Environment Canada,
References
1. Great Lakes Science Advisory Board, "A Perspective on the
Problem of Hazardous Substances In the Great Lakes Basin
Ecosystem", Report to the International Joint Commission (1980).
2. Billings, W.N. and T.F, Bidleman, "High Volume Collection of
Chlorinated Hydrocarbons in Urban Air Using Three Solid
Adsorbants", Atmospheric Environment, JLZ (2): 383-391 (1983).
3. Tanabe, S. et al, "Global Distribution and Atmospheric Transport
of Chlorinated Hydrocarbons: HCH (BHC) Isomers and DDT
Compounds In the Western Pacific, Eastern Indian and Antarctic
Oceans", Journal of the Oceanographlc Society of Japan, Ifl:
137-148 (1982).
4. Bidleman T.F., W.N. Billings and W.T. Foreman, "Vapor-Particle
Partitioning of Semivolatile Organic Compounds: Estimates from
Field Collections", Environmental Science and Technology 2Q
(10): 1038-1043 (1966).
5. Johnson, N.D. et al, "Development of a Gas/Part1cle
Fractionating Sampler for Chlorinated Organlcs", Paper 85-81.1
presented at the 78th Annual Meeting of the A1r Pollution
Control Association, Detroit Mich,, June 16-21, 1985.
6- Johnson, N.D et al, "Evaluation of a Denuder-Based Gas/Part1cle
Sampler for Chlorinated Organic Compounds", Paper Presented at
the EPA/APCA Synoposlum on Measurement of Toxic Air Pollutants,
Raleigh, NC., April 28-30, 1986.
7- Strachan, W.M.J, and H. Huneault, "Automated Rain Sampler for
Trace Organic Substances", Environmental Science and Technology
11 (2): 277-278 (1983).
8. Tanabe, S. et al, "PCBs and Chlorinated Hydrocarbon Pesticides
in Antarctic Atmosphere and Hydrosphere," Chemosphere 1Z (2):
277-278 (1983).
17
-------
Train A Trdi* B
(.Reference)
FILT£R(( lOjum)
Backup Adsorber
Diffusion Denuder
Assembly
DfcJy>tomous Sampler
( Virtual Impactor I
1 j Fin* Particle
H Filter U2.5jum)
Z3 Backup Adsorbers
Coarse Particle'
Filter (2.5-IQpm)
1.67 15 L/mtn
FIGURE I SCHEMATIC DIAGRAM OF GAS/PARTICLE FRACTIONATING
SAMPLER.
L - SUPERIOR
SAULT ST. MARIE
HURON
TORONTO
Missiliougo IC^
L . ONTARIO
NIAGARA
L. ERIE
FIGURE 2
SAMPLING LOCATIONS
-------
TABLE 1
Summary of Particulate Concentration UlOvim) and Denuder Losses
Location Part. Cone. Range HFIne
(jjg/m')
Onshore/Offshore 24 16-31 76
Turkey Lakes 6 5-9 50
Sheridan Park 22 8-36 70
Stoney Creek 33 15-53 64
Mean (Conv. Sampler) 21 5-53
Mean
-------
TABLE 4
Summary of
Target Compound Measurements (ng/m3) and Partlculate-Assoclated
FractionsC%)by Denuder Difference
Location Test Series Ave. Total Cone. Ave. Part.Fraction (%)
HCB a-HCH Y-HCH HCB aHCH Y-HCH
Lake Ontario
Land-West
0.3
0.7
<0.03
—
—
—
Onshore/
Ship-West
0.2
0.4
0.06
6
14
46'
Offshore
Land-East
0.3
0.3
<0.03
0
—
—
Ship-East
0.3
0.5
0.07
2
8
-2
Ship-Moving
0.2
0.5
0.06
0
5
-2
Turkey Lakes
Set 1
0.2
0.4
-0.04
0
10
-0
Set 2
0.1
0.3
-0.04
5
5
-0
Set 3
0.1
0.3
0.1*
5
0
-0
Sheridan Pk.
Set 1
0.2
0.4
0.1
17*
18*
-23'
(Mod. Temp.)
Set 2
0.2
0.3
-0.04
4
6
-20
Set 3
0.1
0.3
0.06
14
8
-67
Sheridan Pk,
Set 1
-0.07
<0.03
<0.04
-100*
_ —_
_ _ __
(Low Temp.)
Set 2
0.08
0.1
<0.02
-15
21
Set 3
0.1
0.1
<0.02
-51
35
Stoney Creek
Set 1
0.2
0.1
0.4
60*
____
13
(near source)
Set 2
0.1
0.2
0.5
57*
0
10
Set 3
0.1
0.1
0.1
-73*
28*
10
Set 4
0.2
0.2
0.1
32
1
0
* Contamination suspected
20
-------
FIELD COMPARISON STUDY OF POLYURETHANE FOAM AND XAD-2 RESIN
FOR AIR SAMPLING OF POLYNUCLEAR AROMATIC HYDROCARBONS
Jane C. Chuang+, Steve W. Hannan+, and Nancy K. Wilson#
+ Battelle Columbus Division, Columbus, Ohio 43201
# Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
A study of the sampling efficiency for PAH in vapor using
two adsorbents, XAD-2 resin and polyurethane foam (PUF) was perform-
ed under summer and winter ambient conditions. Two aspects were
investigated:(1) collection efficiency for ambient PAH vapor and
(2) retention efficiency for native and perdeuderated PAH spiked
onto the adsorbents before sampling. Some spiked PAH which were
volatile or reactive were recovered more efficiently with XAD-2
than with PUF. Lower sampling temperature improved the recovery
for volatile PAH using PUF, but the recoveries of reactive PAH,
such as cyclopenta[c,dIpyrene, were not improved at lower temperature
for either adsorbent. An investigation of the stability of PAH
collected on quartz fiber prefilters and XAD-2 or PUF backup traps
as a function of the storage time was also carried out. Storage
at room temperature in the dark for 30 days did not have an adverse
effect on the collected ambient PAH or on the spiked perdeuterated
PAH with XAD-2 resin. However, a decreasing concentration trend
with storage time for naphthalene, anthracene and D^'^enzofaIpyrene
collected or spiked on PUF was found. Most of the particle-bound
PAH collected on quartz fiber filters were not affected significantly
during storage, except that cyclopenta[c,d]pyrene decreased to
approximately half the original level after 30 days of storage.
Investigation of the degradation product of cyclopentatc,dIpyrene,
pyrene dicarboxylic acid anhydride, showed that the levels of
this compound collected on filters increased to about 1.5 times
the original value after storage for 30 days. The distributions
of PAH between filters and XAD-2 resin found in the stability
study corresponded to the volatilities of individual PAH, and
most 2- to 4-ring PAH were found to break through the filters.
This demonstrated that the use of both prefilters and backup adsorbents
is necessary in air sampling for PAH analysis.
21
-------
INTRODUCTION
Polynuclear aromatic hydrocarbons (PAH) have been studied
extensively and have received increased attention in investigations
of air pollution recently because some PAH are carcinogens, mutagens
or both^"5. To understand the human exposure to PAH, reliable
sampling and analytical methodology must be established for monitor-
ing PAH in air. Several studies^"' have shown that filters alone
can not retain all the 2- to 4-ring PAH, The XAD-2 resin and
polyurethane foam (PUF) have been used to collect PAH vapors^"^
by various research groups. The overall comparative effectiveness
of these two adsorbents and the degree to which quantification
of PAH is affected by sampling, handling and storage have not
been fully investigated. This uncertainty precludes an accurate
assessment of the relative overall merit of each adsorbent.
The objective of this study was to compare the sampling efficiency
for PAH vapors of these two adsorbents, PUF and XAD-2 resin, and
to compare the storage stability of PAH vapors collected on PUF
and XAD-2 resin. The storage stability of quartz fiber filters
and investigation of the degradation product of cyclopenta [c,d]pyrene
due to storage were also performed.
EXPERIMENTAL METHODS
Sampling
Modified General Metals PS-1 samplers with General Metals
bypass motors were employed in the ambient air sampling. The
samplers which are described elsewhere^, were placed on the ground
in an open space outside Battelle's laboratory in Columbus, Ohio.
This location can be classified as a medium-size, mldwestern city
with relatively clean air for an urban environment. The sampling
was performed on weekends or holidays to reduce the
contribution of local vehicle exhaust emissions to the samples.
Air was sampled for 24 hrs at & flow rate of 6.7 cfm. Quartz
fiber filters were located upstream of the PUF or XAD-2 in all
sampling experiments.
In the sampling efficiency comparison study, two sets of
three samplers were located in parallel about two feet apart.
The three samplers of each set were separated from each other
by about one foot. The PUF cartridges were used in one set of
three samplers and XAD-2 cartridges were used in the other set.
Prior to sampling, two each of the three PUF and XAD-2 were spiked
with the following PAH (2 |ig of each): Dg-naphthalene,
D^Q-phenanthrene, D^g-pyrene, Dj^'chryBene, D^-benzolalpyrene,
phenanthrene, anthracene, fluoranthene, pyrene, cyclopenta[c,djpyrene,
benz[a]anthracene, benzo[e]pyrene and benzol aJpyrene. These experiments
were performed twice under the same conditions except that during
the first experiment, the ambient temperature ranged from 66-86°F
and during the second experiment, it ranged from 17 to 27°F.
22
-------
In the storage stability study, eight samplers in two sets
of four were used. Sampler location was similar to the comparison
study described above. In the first experiment, PUPs were used
as backup traps, and XAD-2 traps were used in the second and third
experiment. All PUF and XAD-2 from the first and the second sampling
were spiked with only the perdeuterated PAH, noted in the comparison
study, before sampling. After sampling, these PUF and XAD-2 samples
were stored in the dark at room temperature for 0, 10, 20, and
30-day intervals before extraction and analysis as replicate pairs.
The filters from the XAD-2 sampling were similarly stored and
extracted after the same time intervals. The filters from the
PUF sampling were not analyzed. The purpose of the third sampling
was to provide samples for bioassay analysis and to characterize
the polar PAH and possible PAH degradation products. Only the
degradation product of cyclopenta[c,d]pyrene, pyrene-1,10-dicarboxylic
acid anhydride, is reported in this paper since the study is still
in progress.
Analysis
The PUF and XAD-2 cartridges were cleaned before air sampling.
The cleanup procedures are described elsewhere > . Quartz fiber
filters were heated at 400°C for 16 hrs before use. Samples were
Soxhlet-extracted for 16 hrs with 10 percent ether/hexane (PUF)
or methylene chloride (filter and XAD-2). The extracts were concentrated
and analyzed by electron impact (EI) gas chromatography/mass spectrometry
(GC/MS) for PAH. Identification of PAH was based on the GC retention
times of the individual monitored molecular ion signals compared
to that of the internal standard, 9-phenylanthracene. Quantifications
of these compounds were based on comparisons of the respective
integrated ion current responses for the monitored molecular ions
to that of the internal standard using calibration response curves.
The dichloromethane solution of cyclopenta[c,d]pyrene was
spiked evenly onto a quartz fiber filter and was exposed to U.V.
light for 4 days to obtain a mixture of cyclopenta[c,d]pyrene
and pyrene dicarboxylic acid anhydride. The exposed standard
solution was analyzed by both EI and negative chemical ionization
(NCI), GC/MS. The internal standard, Dtpl-nitropyrene, was used
in the NCI method and 9-phenylanthracene was used in the EI method.
Identification of this compound in the sample was based on the
correct GC retention time and the NCI mass spectrum. Since the
accurate concentration of the exposed standard could not be determined,
the response factor of this compound to the internal standard
was assumed to be 1.
RESULTS
Sampling Efficiency Comparison Study
Analysis of the background (non-spiked) XAD-2 and PUF samples
show that PAH recoveries from PUF and XAD-2 are very similar for
all compounds except for naphthalene, the most volatile PAH of
the series. Levels of ambient naphthalene collected on XAD-2
are about 24 and 12 times higher than those on PUF in the summer
23
-------
and the winter experiments, respectively. The summer phenanthrene
levels are slightly higher for XAD-2 resin than for PUF, but identical
levels are observed for both adsorbents in the winter experiment.
This finding reveals that XAD-2 reain has better collection efficiency
for the more volatile naphthalene than doeB PUP and that the collection
efficiency for naphthalene on PUF Improves at lower ambient sampling
temperature.
Generally good recoveries are obtained for most spiked PAH
from both adsorbents after exposure to ambient air at a flow rate
of 6,7 cfm for 24 hrs. The recoveries of Dg-naphthalene from
XAD-2 resin were significantly higher than those from PUF, in
agreement with the result for native naphthalene noted above.
TheBe data suggest that XAD-2 resin is significantly more effective
in retaining two-ring PAH. Similarly) lower recoveries of the
spiked DjQ-phenanthrene aT>d spiked native anthracene were obtained
from PUF than from XAD-2 resin in the summer experiment. As expected,
these PUF recoveries improved significantly in the winter experiment,
becoming comparable to those from XAD-2 resin.
The native phenanthrene recovery data cannot be accurately
addressed in the summer experiment because the ambient background
levels were about 19 times higher than the spiked levels. Thus,
due to the expected variations in sampling and analytical procedures,
negative recoveries were sometimes observed for spiked phenanthrene.
Low recoveries of cyclopentaJc,d]pyrene were obtained with PUF
in both the summer and the winter experiments, 40 and 23 percent
respectively. The loss of spiked cyclopentafc.djpyrene almost
certainly results not from volatilization but from decomposition,
which is supported by the storage stability results (See below).
Sampling Module Storage Stability Study
For XAD-2 resin, no loss of perdeuterated PAH was detected
over the 30-day storage interval. In addition, storage instability
over the 30-day period was not indicated for any of the native
PAH collected on XAD-2 resin. Only trace amounts of cyclopenta[c,dJ
pyrene vere found in XAD-2 resin, and the variability in the data
at these low levels prevent showing the decreasing concentration
trend.
For PUF, the Dg-naphthalene storage stability results are
not meaningful since only 1.2 percent of the spiked Dg-naphthalene
was retained on the PUF at the end of 24-hr sampling. The decreasing
concentration trend for Di2-Benzo[aJpyrene were observed with
PUF. However, this may not cause a serious problem in air sampling,
since benzo[aJpyrene is mainly retained in the filter. It appears
to be a loss on storage of both native naphthalene and anthracene
collected on PUP. Since naphthalene clearly breaks through PUF
during 24-hr sampling, we expect that it may be partially lost
by volatilization during 30-day storage. The greater Iosb from
PUF of anthracene relative to its isomer phenanthrene iB more
consistent with the lower chemical stability of anthracene relative
to phenanthrene than with a loss through volatilization.
24
-------
Storage losses over 30-day period were not observed for any
of the particle-bound PAH on quartz fiber filters except for
cyclopentalc,d]pyrene. The data for cyclopenta[c,d]pyrene are
given in Figure 1. As shown in Figure 1, levels of cyclopenta[c,d]pyrene
decrease to about half of the original levels after storage for
30 days in both experiments. Other research groups 7 have
identified a direct-acting mutagen, pyrene dicarboxylic acid anhydride,
as a possible degradation product of cyclopenta[c,d]pyrene in
airborne particulate material and other environmental samples.
We have investigated the degradation product of cyclopenta[c,d]pyrene
in the third stability experiment.
The EI and NCI mass spectra of pyrene dicarboxylic acid anhydride
are given in Figure 2. As shown in Figure 2, the characteristic
neutral losses of CO2 and CO are found in the EI spectrum. As
expected, only the molecular ion is detected in the NCI condition.
For the detection of this compound, the NCI method Is much more
sensitive than the EI method. Since this compound is highly electro-
negative, it is susceptible to the attachment of a thermal electron
from the NCI reagent gas plasma, which enhances the detection
sensitivity. VJe can detect pyrene dicarboxylic acid anhydride
in the filters only with the more sensitive NCI method but not
with the EI method. Analysis of the NCI data showed an increasing
concentration trend for this compound on filter samples after
30-day storage. The level of this compound from the 30-day sample
was more than 1,5 times of those of the 0-day sample. This finding
suggests that particle-bound cyclopentafc,d]pyrene partially decomposes
to pyrene dicarboxylic acid anhydride after a 30-day storage period.
Distributions of PAH between the filter and XAT>-2 resin from
the stability study in general agree with the volatilities of
the individual PAH. This finding indicates breakthrough of 2-
to 4-ring PAH from the prefliters. Other research groups1 "
have also reported that most of the volatile PAH were not retained
on filters under high-volume sampling conditions. In our study,
medium-volume samplers were used and the flow rate (6.7 cfm) was
lower than a typical high-volume sampling flow rate. Thus, this
observation clearly demonstrates that the use of filters only
to collect PAH from air is not sufficient, and correct quantitative
results, even for species as large as chrysene, cannot be obtained
without the use of back-up adsorbents.
CONCLUSIONS
For air sampling of gas phase PAH, XAD-2 resin seems to have
two performance advantages compared to PUF: (1) 2- and 3-ring
PAH clearly break through standard PUF cartridges in a typical
sampling volume of about 300 cubic meters, and (2) PAH species
that are prone to degradation during sampling or during sample
storage prior to extraction exhibit greater losses from PUF than
from XAD-2 resin. Certainly, the future use of quartz fiber filter/
PUF samplers should involve a minimum of sample handling and storage
to reduce the losses of volatile and reactive PAH. Little is
known about the degradation products of PAH due to sampling and
sample handling. The critical issue of whether a significant
25
-------
part of the mutagenic activity in theae types of air samples is
from sampling artifacts or is actually present in the sampled
air is still not clear, and further investigations need to be
carried out in this area.
ACKNOWLEDGEMENTS
The authors would like to thank Mr. John Koetz and Mr. Warren
Bresler for performing the sampling) and Dre, Samuel V. Lucas
and Larry E. Slivon for technical discussions and advice on man-
uscript revision.
Although the research described in this article has been
funded wholly or in part by the U. S. Environmental Protection
Agency through Contract Number 68-02-4127, 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. Polycycllc Aromatic Hydrocarbonsi Evaluation of Sources
and Effects, national Research Council, National Academy
Press, Washingtonj D.C., 1963.
2. Miller, M., and Alfheiin, I., Environ. Sci. Techno1., 1982,
_16, 221.
3. Dehren, W,, Pitz, N., arid Totningas, R., Cancer Lett., 1977,
4,5.
4. Chuang, C. C. and Petersen, B. A., 1985, EPA-600/4-85-045.
5. Perera, F., Environ. Health Persp., 1981, 42^ 163.
6. Galasyti, T. G., Horning J. F, and Soderberg, R. H.,
J. Air Pollut. Control Assoc., 19&4, .34(1), 57.
7. Cautreela, W., and Vantauwenberghe, K., Atroos. Environ.,
1970, _12_> 1133.
8. Feng, Y. and Bidleman, T. F., Environ. Sci. Technol., 1984,
18, 330.
9. Yamasaki, H., Kuwata, K., and Miyamoto, H,, Environ. Sci.
Technol., 1982, 16, 182. ~
10. Jones, P. W., Wilkinson, J. S., and Strup, P. E,, U.S.
1977, EPA-60/2-77202.
11. Thrane, K. E. and Miksleen, A., Atmos. Environ.. 1981, 15,
909.
12. Chuang, C. C., Bresler, W. E., and Hannan S, W., 1985,
EPA-600/4-85/055.
13. Keller, C. D., and Bidleman, T. F., Atmos. Environ., 1984,
18, 837.
26
-------
Chuang, C. C., Mack, G. A., Petersen, B. A., Wilson, N, K.,
1985 EPA-600/D-85/050.
Lewis R. D. and Jackson, M. D., Anal. Chem., 1982, ^4, 592.
Chuang, C. C., S. W. Hannan and J. R. Koetz, 1985, EPA-600/
4-86/029.
Rappaport, S. W., Wang, Y. Y., Wei, E. T., Sawyer,
R., Watkins, B. E., and Rappaport, H., Environ. Sci- Technol.
1980, 1505.
Stenberg, U. R., and Alsberg, T. E., Anal. Chem., 1981, 53,
2067.
Konlg, J., Balfanz, E., Funcke W., and Romanowski, T.,
Anal. Chem.. 1983, 55, 599.
27
-------
2nd Stability Experiment
I,d Stability Experiment
0 10 20 30
Storage Period, day
Figure 1. Levels of cyclopenta[c,d]pyrene
in quartz fiber filters as a
function of storage time
28
-------
100.0-1
E!
w
114
11.
m
153
300
m
3Q0
tn
roo.o-i
NCI
50.0-
100
JOO
2U M7
11 I
!
290
tH
-r~
100
Ma
r
310
Figure 2. Upper: Electron impact mass
spectrum of pyrene dicarboxylic
acid anhydride
Lovers Negative chemical ioniza-
tion mass spectrum of pyrene
dicarboxylic acid anhydride
29
-------
NON-OCCUPATIONAL EXPOSURE TO HOUSEHOLD PESTICIDES
Andrew E. Bond and Robert G, Lewis
U. S. Hnvirortmentdl Protection Agency
Research Triangle Park, NC 27711
Abstract
The U. 5. Environmental Protection Agency is conducting a study to develop
and validate monitoring methods needed for the determination of human exposure
to common household pesticides. The primary objective is to obtain an estima-
tion of the cumulative frequency of non-occupational exposures to home and
garden pesticides through the air, dermal, drinking water, and dietary routes.
In a nine-home pilot study designed to test methodology, 22 of 28 of the most
commonly used household pesticides were detected. Five of these were measured
in the air in the majority of the households at concentrations ranging up to
several vg/m3. Personal air measurements suggested that 80% of the total
respiratory exposures resulted from Indoor air. Subsequent to the pilot study,
a 1arge-sca^e, multi-season monitoring program involving up to 175 homes in a
high-use southern U.S. city and about 85 homes in a low-use northern U.S. city
was initiated.
in 1985, the United States Environmental Protection Agency instituted the
first attempt to develop methodology for determining pesticides exposures in
the general population of the U.S.A. The project, known as the Non-occupa-
tional Pesticides Exposure Study (NOPES), is designed around the concepts of
the Total Exposure Assessment Methodology {TEAM) studies previously conducted
by EPA (4). The NOPES project will test whether the TEAM approach can be
adapted to develop estimates of exposures to selected pesticides in personal
(respiratory) air, drinking water, and from dermal contact in a stratified
random sample of the population of two urbanized areas in the United States.
The scientific objectives of NOPES are as follows:
o Adapt the TEAM methodology to pesticides by testing and refining
monitoring instrumentation,
o Collect preliminary data relating exposures through each of three
potential routes of exposure (air, water, skin) to pesticide usage
levels.
o Obtain basic knowledge which may be used in future investigations in
four topic areas; (i) human exposure model development, (ii) measure-
ment methods development (iii) rcicroenvironmental field studies, and
(fvj dosage research investigations.
o Investigate causal relationships that explain the variability from
person to person of exposure to pesticides.
o Estimate the relative importance of each pathway of exposure to the
total exposure for each pesticide.
o Determine if a predictive human exposure model can be developed to
relate data on exposures to usage rates and other factors.
30
-------
The study, which is currently in progress, is focused on the measurement
of exposure to 28 of the most commonly used household insecticides, fungacides
and herbicides. Two cities are involved in the study: Jacksonville, Florida,
a warm climate city in which household pesticide use is one of the highest in
the U.S.A. and Springfield/Chicopee, Massachusetts, a cold-climate area char-
acterized by low household pesticide usage. A total of 260 single-family
households will be monitored over three seasons by the completion of the study
in 1988,
During August 1985, a pilot study was conducted over a seven-day period
in Jacksonville to evaluate and select sampling, analytical and survey methods
(2). Nine homes were chosen and presorted into three groups of three homes
each on the basis of the level of indoor use of pesticides. Stratification
was based on the following four variables: (a) the frequency of indoor appli-
cation of pesticides to control insect pests; (b) the frequency of use of
pest control products on pets; (c) the frequency of indoor applications of
pesticides on plants; and (d) the termlticide treatment program of the housing
unit. Usage categories were designated from the above variables as high
3 out of 4, "medium" 2 out of 4, and "low" 1 or less. In each of the desig-
nated low-, medium- and high-use dwellings, 24-hr indoor and outdoor air
samples were taken and one resident carried or kept close by a portable
sampler at all times during the same 24-hr period for the assessment of
personal respiratory exposure. In addition, tap water samples were collected
from each household and dermal (hands only) exposures were monitored when
participants were applying pesticides. Pre-extracted, white cotton gloves
were used for this purpose.
The results of the pilot study demonstrated the utility of the low-volume
air samplers, which pumped air at 3.8 L/min through a 22-mm x 76-mm polyure-
thane foam sorbent trap (3), for indoor, outdoor and personal exposure moni-
toring. The overall conduct of the pf)ot study was relatively trouble-free.
The study showed that the numbers and concentrations of pesticides in residen-
tial indoor air were generally higher by an order of magnitude or more than
those in the envelope of air immediately surrounding the building (the meso-
phere). Of the 33 pesticides and related chemicals shown in 'able I, a total
of 24 were detected in indoor air, 21 in personal air, and 25 in outdoor air.
Concentrations ranged from 1.7 ng/m3 to 15.0 yg/m3. Five pesticides (chlor-
pyrlfos, diazinon, chlordane, propoxur and heptachlor) were found in 80 to
100% of the indoor air samples at mean concentrations of 0.16 to 2.4 ug/m .
These concentrations were generally one order of magnitude higher than those
of the other pesticides measured. In only one Instance, however, were the
guidelines established by the National Academy of Sciences (1) exceeded. On
one of two days of sampling in one high-use household, chlorpyrifos was found
at 15 ug/m3, or about 50% higher than the interim guideline for exposures
to this termlticide not exceeding three years. Personal air measurement cor-
related well with Indoor air measurements, with corresponding concentrations
usually within ^ 50% of each other.
AH tap water samples collected during the pilot study were negative for
all of the target compounds. Detection limits ranged from 0.05 to 1 P9/J-
dependlng on the analyte. Only three dermal exposures were monitored so tnat
there was insufficient data for Interpretation,
The overall conclusions from the pilot study were that (a) monitoring
methodology was effective for monitoring all of the pesticides of interest,
31
-------
(b) most (>80%) of the total personal exposure to household pesticides could
be attributed to indoor air, and (c) the occurrence and concentrations of
pesticides found in indoor air correlated well with the level of usage as
determined by the presampling survey.
The first phase of the full NOPES project was conducted during August
and September 1986 in Jacksonville. There will be subsequent samplings
during March - May 1987 and during January - March 1988. Of 267 households
screened, 66 were sampled (five short of the goal of 70). Duplicate (simul-
taneous) samples were taken at six of these, triplicates at three, and repli-
cates (two different days) at nine. A similar number of households are
scheduled for sampling during the next two sessions. Some of these house-
holds will be monitored once in each of the three seasons, while the remain-
der will be sampled once or twice in one season only. Stratification of the
participating households provides for 50% high-use, 30$ medium-use and 20%
low-use. A total of 864 samples (804 air, 18 dermal, 18 drinking water and 24
field blanks) should be taken in Jacksonville. The Springfleld/Chicopee
study will be smaller, encompassing 85 residences over two seasons.
In addition to direct measurements of air, water and dermal exposure, an
attempt will be made to estimate dietary exposures to the same pesticides.
For this purpose, dietary recall questionnaires will be administered and
pesticide residues published by the U.S. Food and Drug Administration as part
of its annual Total Diet Study will be used.
As of this writing, the analytical results from about 95% of the 256
samples collected in 75 Jacksonville homes are available in partially reduced
form. The air monitoring results for the five most prevalent pesticides are
given in Table II. These results are comparable with those found in the nine-
home pilot study (2). The higher mean concentration and relatively fewer
positive findings of technical ch^ordane may be the result of more rigorous
quality control procedures Instituted for the full study. The stricter QA
resulted in higher detection limits and required more positive confirmation
of the multlcomponent mixture. As in the pilot study, no pesticides have
been detected in tap water samples analyzed to date. Results of dermal and
respiratory exposure measurements are incomplete as of this writing.
References
1. Committee on Toxicology, National Research Council. An Assessment of the
Health Risks of Seven Pesticides Used for Termite Control. Washington,
D.C. National Academy Press (1982).
2. Lewis, R. G., Bond, A. E., Fitz-S1mons, T. R., Johnson, D. E., and Hsu,
J. P. Monitoring for non-occupational exposure to pesticides 1n indoor
and personal respiratory air. 79th Annual Meeting of the Air Pollution
Control Association. Minneapolis, MN. June 1986. Paper No 86-37.4,
3. Lewis, R. G., and MacLeod, K. E. Portable sampler for pesticides and
semi volatile industrial organic chemicals in air. Anal. Chem. 54 (1982),
310-315.
4. Ott, W., Wallace, L., Mage, D., Akland, G., Lewis, R., Sauls, H., Rodes,
C., Kleffman, D., Kuroda, D. and Morehouse, K. The Environmental Protec-
tion Agency's research program on total human exposure. Environ. Inter-
nal. 12 (1986), 475-494.
32
-------
Table 1: Summary of pesticides found in air during Jacksonville pilot study,
August 1985,
Number of Households at Which Detected in Air
Pesticide Indoors Outdoors Respiratory
Chlorpyrlfos
9
7
8
Diazinon
8
7
6
Chlordane
7
3
6
Propoxur
7
4
6
6
Heptachlor
7
5
trans-Nonachlor
6
3
5
Lindane
6
2
3
Heptachlor Epoxide
6
2
3
Aldrin
6
4
3
o-Phenylphenol
5
4
4
r
Dieldrin
5
4
0
Captan
5
2
4
Folpet
5
4
4
1
1
Oxychlordane
5
2
Maiathion
4
2
Bendiocarb
3
1
4
o
a-BHC
3
2
3
Ronnel
3
1
3
Chlorothalonil
3
3
1
0
1
Pentachlorophenol
2
1
1
Qichlorvos
1
1
0
0
0
1
Dicofol
1
2
Methoxychlor
1
1
0
1
1
0
£-£'-DDT
1
cis-Permethrin
trans-Permethrin
0
0
I
0
0
a
Resmethrln
0
Carbaryl
0
0
n
0
n
2|4-0 Esters
0
0
u
A
Atrazine
0
0
0
A
Dacthal
0
0
0
A
Polychlorlnated B1phenyls
0
0
0
33
-------
Table 2: Summary of positive findings for most prevalent pesticides found
in air during the Jacksonville full study, August 1986.
Air Concentration, ug/n\3
Pesticide
Indoor
Outdoor
Personal
Exposure
Chlorpyrifos
Range
Mean3
Positive*5
0.013 to 2.2
0.39
69
0.012 to 0.21
0.043
23
0.014 to 2.8
0.34
66
Diazinon
Range
Mean
Positive
0.059 to 13.7
1.0
42
0.066 to 0.29
0.15
4
0.056 to 4.7
0.63
41
Propoxur
Technical
Chlordane
Heptachlor
Range
Mean
Positive
Range
Mean
Positive
0.019 to 7.9
0.46
53
Range 0.19 to 3.0
Mean 0.78
Positive 29
0.016 to 1.6
0.39
29
0.026 to 0.29
0.092
8
0.21 to 0.63
0.37
7
0.021 to 0.63
0.20
10
0.022 to 3.9
0.29
51
0.17 to 1.3
0.61
25
0.018 to 1.6
0.38
25
a Arithmetic value of all positive measurements.
b Number of dwellings of 75 monitored for which at least one sample contained
detectable levels of the indicated pesticide.
34
-------
COMPARISON OF ESTIMATED AND ACTUAL PCB VAPOR EXPOSURE DURING A SOIL
EXCAVATION PROJECT
James Neely, CIH and Colin Moy
Ecology and Environment, Inc.
160 Spear Street
San Francisco, CA 94105
¦ I
A risk assessment was conducted 'for a heat exchange
which had used the PCBs Aroclor 1248 and 1ZM in^ ^ Qne tQ
fluids. Concentrations of vaporized pC®* we. calculated in the risk
two orders of magnitude greater than had bei liqhter Aroclor
assessment. Further, the PCBs detected be 5equ.tely
1232 or a related compound. Because dust #1 vapor.phase PCBs
controlled, the discrepancy between projeiformation of lighter, more
could only bo explained b, the unforseen 0; 1254,
volatile PCB species from partial evaporation sanples. Partial
or failure to detect Aroclor 1232 on ^!0 contribute to the formation of
photolysis of the heavier PCBs could also
the lower chlorinated PCBs detected in air.
t-hP removal of soil contaminated
In conducting risk assessments for ^ results show
with the most common PCB products 1248 J" ^ or other confirmation
that it is prudent to use on-site: air and the public
techniques to ensure the health and sa t1qate health implications
during excavation. The need to f^rt,h®rpJ"7uVa1n9s isalso identified,
of photolysis products, particularly dibenzofurans,
35
-------
Background
The hazardous waste facility discussed in this report was a
chemical processing plant which utilized heat-exchange fluids containing
PCB 1248 and 1254. following demolition of the plant, it was determined
that the original PCB products or residuals had apparently leaked into
surface soils throughout the 3-acre site. Soil was sampled based on a
grid pattern that divided the site into roughly equivalent sections.
Sampling was continuous, from surface to 6-foot depths, and analysis
continued until the vertical extent of contamination was resolved.
Additional surface samples were collected and analyzed to delineate the
areal extent of contamination, A total of 321 samples were analyzed.
Soil Analyses
Soil samples were analyzed at a laboratory certified by the State of
California for hazardous-waste analyses. The contents of each sampling
sleeve was homogenized and a portion of the the soil sample was extracted
with pure hexane and shaken. Sodium sulfate was also added to absorb
moisture and promote mixing. The extract was then analyzed directly
without a Florisil clean-up step. EPA Method No. 8080 was followed,
as were all required quality-control protocols. Up to 11 of the Aroclor
peaks were used to quantitate the sample. Aroclor type was identified
by matching peak patterns and retention times to internal standards. The
results of this sampling program revealed that site soils contained
variable concentrations of PCBs. The average PCB level in the most
heavily contaminated soils located within 12 inches of the surface was 326
ppm, with a range of <1 ppm to 4,400 ppm, The chromatograph from soil
samples most closely approximated PCB 1254 (see Figure 1).
Exposure Guide!ines
For worker exposure to PCBs, the California OSHA 8-hour Time Weighted
Average (TWA) Permissible Exposure Level (PEL) as well as the American
Conference of Governmental Industrial Hygienists1 TWA Threshold Limit
Value (TLV) are the same: 1.0 mq/m3 for PCBs containing 42% chlorine by
weight (PCB 1242), and 0.5 mg/rrw for PCBs containing 5451! chlorine (PCB
1254). The National Institutes for Occupational Safety and Health, citing
the potential for carcinogenic effects from PCBs, has established a
recommended exposure guideline of 1 ug/rc3 (0.001 mg/m3) as a 10-hour
TWA for any PCB exposure.
A risk assessment was performed to evaluate potential health hazards
arising from direct and indirect human exposure to the site. Because
there are no standards for allowable levels of PCBs in outdoor air, the
risk assessment proposed a safe exposure guideline of 0.101 ug/m3, for
off-site community exposures, based on a calculated allowable daily intake
of 2.5 ug/day. This guideline was extrapolated from a mortality study (1)
and was designed to maintain a lifetime risk of cancer at less than
10-5.
36
-------
Exposure Estimation
PCBs in soils may reside in the vapor, liquid, or adsorbed phase.
Direct evaporation from surface soils, diffusion of vapors from underlying
strata, and wind-born dust may result in airborne exposures. Many
physical and chemical properties of the soil and the contaminant, as well
as external environmental factors, affect the rate of evaporation.
Examples of physical properties of soils include tortuosity, porosity, and
exposed surface area. Chemical properties include moisture content,
organic carbon content, vapor pressure, and solubility. Important
environmental factors include insolation, temperature, and wind speed {2).
In determining vapor emissions, rate-limiting parameters such as
organic carbon adsorption and effective pore space are ignored when
developing worst-case exposure assessments. The authors of the risk
assessment applied Henry's Law in estimating vapor exposure levels in the
immediate site vicinity.
According to Henry's Law;
Cv = He CI
He = 1.73 X 10-7 (3)
CI = 326 X 10-6 g PCBs/g soil-water) X (MW water/MW PCBs)
Cv = 2.3 X 10~12 moie fractions
Using the above values and Henry's Law, one calculates that the
near-surface contamination will contribute 0.022 ug PCBs/M of
air.
Air Sampling
An air-sampling station was positioned at each of the four perimeters
of the site. Depending upon wind direction and locations of s^e
activities, these stations were moved along each perimeter. The
air-sampling pumps (VWR Model 54906) were operated whenever any type of
work took place at the site. Samples were drawn at a fixe" r
-------
Personal air-sample pumps were used to evaluate exposure of equipment
operators to PCBs. Two types of pumps were used: a medium-flow (MSA)
portable pump using the same glass-fiber filter and Florisil tube as the
perimeter sampling pumps; and a low-volume pump (SKC or Sippin) using the
filter and tube sizes specified in N10SH Method 5503. The breathing-zone
samplers were worn by workers throughout their daily shifts.
A total of 72 area samples and 11 employee breathing-zone samples
were collected. The front section of the Florisil sampling sorbent and
the filter were extracted together and analyzed by a Varian 6500 gas
chromatograph with an electron-capture detector and a mixed-phase column
at an isothermal temperature of 200'C. The procedure is described in
NIOSH Method 5503. Aroclor standards and hexane blanks were analyzed
periodically during sample tests as required by the method's quality-
control protocols. Similar quantitation and identification methods were
repeated from the soil testing.
Dust Monitoring
Dust was monitored with a GCA Miniram direct-reading aerosol monitor
(detection limit of 0.001 mg/m^) at upwind and downwind positions
throughout the site. If the wind was negligible, readings were taken at
each of the four perimeters. Readings were taken approximately every 30
to 60 minutes throughout the shift. Downwind readings were compared to
upwind readings to see how much dust on-site activities generated.
Wind speed and direction were also measured, with a wind vane and
anemometer placed on a six-meter pole at the perimeter of the site.
Results
Analytical results of the PCB air samples showed significantly higher
levels of PCB vapors than had been estimated in the risk assessment.
Downwind concentrations averaged .563 ug/m3, with a median concentration
of .768 ug/m3. Employee exposure levels (large tube and filter samples)
averaged 1.845 ug/m3 with a median of 2.180 ug/m3. In addition,
results were being reported as PCB 1232, based on the chromatographic
comparison to standards, rather than the expected PCBs 1248 and 1254 (see
Figure 2). Throughout the project, dust readings on the aerosol monitor
were insignificant. The emissions were therefore believed to be associ-
ated with PCB vapor. This assumption was confirmed by analyzing filter
and tube sections separately. Three of the four samples analyzed
contained PCBs in the front section only.
Pi scussion
Risk analysis and estimation are valuable methods of assessing
health hazards at hazardous-waste sites. Legislative mandates for
performing these evaluations have been incorporated into the Comprehensive
Environmental Response, Compensation, and Liability Act (CERCLA) and its
amendments (SARA). However, this study demonstrates the limitations
associated with estimating risk based on available soil analysis data;
such risk projections may not describe accurately the dynamic field
conditions affecting overall airborne exposure levels. On-site air
sampling is therefore a critical adjunct to the risk assessment process
for PCB remediation.
38
-------
Three possible reasons for detecting only PCB 1232 in ambient air at
a site which was believed to contain only PCB 1248 and 1254 are 1)
masking of the PCB 1232 in the soil sample chromatographs; 2) partial
vaporization of the heavier PCB mixtures; and 3) photodecomposition of
the heavier cogeners during soil excavation.
Masking may occur when the concentration of the undetected species
is much lower than the quantitated species. The analytical procedure for
PCBs utilizes a "best fit" method of determining the cogener or cogeners
in the sample. The peaks from each of the individual cogeners in the
sample may be discernable if their concentrations are within an order of
magnitude; however, the less concentrated cogeners may not be identified
if sample dilution sufficiently lowers their chromatographic response.
Soils containing high levels of PCB 1248 and 1254 and low levels of PCB
1232 could produce vapor emissions identifiable only as PCB 1232, since
its vapor pressure is 10 to 100 times greater than PCB 1248 and 1254 (5).
Partial evaporation of the PCB 1248 and 1254 mixtures could also
result in vapors characterized only as PCB 1232. The more volatile Cl^
to CI4 biphenyls which constitute 96% of PCB 1232 are also present in
PCB 1248 and PCB 1254 at approximate concentrations of 60% and 20%,
respectively (5).
The need to identify adequately all of the PCB cogeners present in
soils at a given site is illustrated in studies performed by Farmer, et
al (6). In their evaluation of evaporation rates of dieldrin from soil,
these researchers determined that extremely low soil concentrations
(25 ppm) of this semivolatile pollutant produced a maximum saturation
vapor density at the soil/air interface equivalent to that of the pure
compound.
GC/MS spectrometry (EPA Method 680) can provide better delineation
of the various PCB cogeners than gas chromatography alone; however, the
higher detection limits associated with the GC/MS method prevent use of
this technique for analyzing environmental samples containing trace
quantities of PCBs. A concentration step, involving a Florisil trap or
similar adsorption media, could remedy the detection-limit constraints.
While partial evaporation of and/or failure to detect the presence
of PCB 1232 in the soil samples are the most likely causes for the
disparity between the soil and air sample results, the potential
contribution of photolysis should be examined. Photolysis has also been
shown to produce partial dechlorination of PCBs, in addition to producing
quaterphenyls, biphenyls and, in the case of hexachorobiphenyl,
chlorinated dibenzofurans (5). These studies dealt with photolysis in
water columns; measured breakdown rates were less than 5% per year.
Photolysis rates in exposed, aerated soils could be much higher. The
health implications of exposure to photolysis break-down products
warrants careful examination of these mechanisms.
39
-------
References
1. D.P. Brown, M. Jones, "Mortality and industrial hygiene study of
workers exposed to polychlorinated biphenyls," Arch. Environ. Health
36: 120. (1983).
2. W.A. Jury, W.F. Spencer, W.J. Farmer, "Behavior assessment model for
trace orqanics in soil: model description," J. Environ. Qual. 12:
558. (1983).
3. W. Piver and F. Lindstrom, "Waste disposal technologies for PCBs,"
Env. Health Persp. 59: 163. (1985).
4. T.M. Miller, "Sampling and analytical methods for PCBs in air: One
Market Plaza, San Francisco," unpublished report submitted to
California Department of Health Services, (1983).
5. M.A. Callahan, et al, Water-Related Environmental Fate of Priority
Pol 1utants, EPA - 440/4-79-029a, U.S. Environmental Protection
Agency, Wash. D.C., (1979).
6. W.J. Farmer, et at, "Volatility of organochlorine insecticides from
soil: effect of concentration, temperature, air flow rate and vapor
pressure," Soil Sci. Soc. Amer. Proc. 36: 444. (1972),
40
-------
8TANDA
AIR SAMPLE CHROMATOQRAM
QURE 2
41
-------
GC/MI/FTIR-Improved Capability or
New Dimension in Analyses?
J. W. Brasch
BATTELLE
Columbus Division
505 King Avenue
Columbus, Ohio 43201-2593
A commercially available interface between a conventional gas
chromatograph and a conventional Fourier transform infrared spectrometer
system provides a means to trap GC effluents in a frozen argon matrix and
to obtain their IR spectra by reflection/absorption techniques.
Chromatographic resolution is maintained by physical separation of the
components in the matrix. The combination of low temperature (10°K),
matrix isolation (MI), and the ability to signal average for substantial
periods of time provides spectra of exceptional quality even with
subnanogram amounts of material. This capability reaches sensitivity
levels commonly used in many GC/MS analyses and permits IR data to be
correlated with GC/MS data under similar, if not identical, GC conditions.
In addition to improved sensitivity, the MI spectral features
are much sharper than those in ambient condensed phase spectra. This
implies a higher level of information content in the MI spectra. This is
indeed the case, and new possibilities for structural elucidation is one
result.
These points are illustrated with data from complex environ-
mental samples.
42
-------
In the course of a research project to evaluate Cryolect
performance in environmental analyses, several samples previously
analyzed by GC/MS techniques have been studied. One such sample provided
a particularly stringent test of Cryolect performance, and the results
clearly illustrate the potential of GC/MI-FTIR techniques for complex,
"real world" samples. Such potential had been suggested by an earlier
study in which MI - FT IR spectra were shown to unambiguously differentiate
all 22 isomers of tetrachlorodibenzodioxins (1).
A Mattson Cryolect interfaced to a Varian 3400 GC was used for
the analysis. The following GC conditions were used: 30 m, .32 mm id,
.25 micron coating, DB-5 column; 1% argon in helium carrier gas, 10 psi
head pressure; injector 290, detectors 300; Initial column temp 160; hold
3 min; 3°/min to 290; hold 30 min.; 2 injected (splitless).
The effluent from the GC passed through a triple splitter, with
80% of the flow going though a heated transfer line (300°C) to the cold
disk (11 °K) of the Cryolect unit. The remaining 20% of the flow was
split equally to an FID detector and an ECD detector.
Infrared spectra were obtained with a Mattson Sirius 100
spectrometer system. Initial spectra were obtained with 100 scans
coadded, and a second group of spectra were obtained by coadding 1000
scans. All spectra were at 4 cm"* resolution.
The portion of the GC-run between 13 and 23 minutes is shown in
Figure 1. The upper trace is the FID chromatogram and the lower trace is
the ECD chromatogram. Seven peaks are numbered in Figure 1. Data
obtained for these peaks are as follows.
The spectrum of Peak 1 is shown in Figure 2 above the MI
reference spectrum of fluoranthene and is easily identified as such.
From retention times and the previous GC/MS analyses,
tetrachloro-dibenzofurans (TCDFs) and tetrachlorodibenzodioxlns (TCDDs)
were expected in the 20-23 minute region.
The infrared spectrum of Peak 2 is shown in Figure 3, above the
MI reference spectrum of 1,3,7,9-TCDD. Again the identification is very
obvious, along with evidence for another component (band at 1170 cm"1).
The infrared spectrum of Peak 3 is shown in Figure 4, above the
MI reference spectrum of 1,3,6,9-TCDD. The distinct pattern in the 1450-
1500 cm"* region makes the identification very certain. Again another
component(s) is present (bands at 1105 and 1380 cm"1)-
The infrared spectrum of Peak 4 is shown in Figure 5, above MI
reference spectra of 1,3,7,8-TCDD and 1,2,4,8-TCDD. The peak is
estimated to be a 60/40 mixture of the two isomers.
43
-------
Figure 6 is the infrared spectrum of the leading edge of Peak 5.
With the exception of the 1405 cm"! band, none of the band patterns can
be matched to any TCDD isomers. This part of Peak 5 is almost certainly
one or more TCDF isomers for which we have reference spectra of only 10
of the 38 isomers. The 1405 cm"* band is due to 2,3,7,8-TCDF.
The spectrum of the trailing edae of Peak 5 is shown in Figure 7
above MI reference spectra of and "C 2,3,7,8-TCDF. The material
was added to the sample as an internal standard during workup for the MS
analysis. While most of the bands in the Peak 5 spectrum are accounted
for by the two TCDF spectra, bands at ~ 1420 , 1210, and 950-900 cm"I
indicate the presence of other materials. The spectrum resulting from
subtracting the and TCDF spectra from the Peak 5 spectrum is
shown in Figure 8. These residual absorptions cannot be matched by any
single or combination of other TCDDs or the ten TCDF reference spectra.
It, too, is almost certainly one or more of the 28 TCDF isomers for which
reference spectra are not yet available.
The infrared spectrum of Peak 6 is shown in Figure 9, above the
MI reference spectrum of 1,2,6,8-TCDD, and is unambiguously identified as
such.
The infrared spectrum of Peak 7 is shown in Figure 10, above MI
reference spectra for *3C and 1ZC 2,3,7,8-TCDD. The l3C material also
was added as an internal standard during the sample workup. The spectra
show that a large percentage of this GC peak is the 13C 2,3,7,8-TCDD
internal standard, and that there is very little of the native
2,3,7,8-TCDD isomer present. However, the spectra also show another
major component(s) with a strong band -1460 cm"1. The result of
subtracting the 13C 2,3,7,8-TCDD spectrum from the Peak 7 spectrum is
shown in Figure 11, above MI reference spectra for 1,2,3,8-TCDD and
3.4.6.7-TCDF. The correspondence with 1,2,3,8-TCDD bands is obvious.
The correspondence with 3,4,6,7-TCDF is far less obvious; however, other
work in this laboratory has proved that 3,4,6,7-TCDF coelutes with
2.3.7.8-TCDD under these chromatographic conditions, and the
identification can be made with confidence.
In summary, the Cryolect data obtained on this complex
environmental sample provided the following information:
1) A typical PAH material was unambiguously identified.
2) Six TCDD isomers were identified and/or discriminated.
3) -internal standards can be used for infrared analyses as
they are for MS analyses, further suggesting that MI - FT-1R
data can be used for quantitative measurements.
4) One, unresolved GC peak was shown to have ^3C and
2,3,7,8-TCDD, 1,2,3,8-TCDD, and 3,4,6,7-TCDF present. GC/MS
data would be unable to discriminate the 1,2,3,8- and
2,3,7,8-TCDD isomers.
44
-------
5) A second unresolved GC peak was shown to have identifiable
2,3,7,8-TCDF (iZC and 13C) as well as unidentified
component(s) whose spectra indicate one or more additional
TCDF isomers. Again MS data could not discriminate these
isomers. Infrared spectra of all 38 TCDF isomers would
permit the discrimination.
References
1) D. F. Gurka, J. W. Brasch, R. H. Barnes, C. J. Riggle, and S. Bourne,
Micro-Diffuse Reflectance and Matrix Isolation Fourier Transform
Infrared Techniques for the Identification of
Tetrachloro-dibenzodioxinT^ Applied Spectroscopy, 40, 978 (1986).
45
-------
FID
C4
ECD
FIGURE 1. PARTIAL CHRQMATOGRAMS OF ENVIRONMENTAL SAMPLE
-------
413340610 RUN1 1/29/86
ABS.
0. 008—
0. 006—
0. 004—
0. 002—
0. 000-
-0. 002—
-0. 004—
1600
1500
1100
1000
1400
1300
1200
900
900
700
FIGURE 2. A) SPECTRUM OF PEAK #1
B) MI REFERENCE SPECTRUM OF FLUORANTHENE
47
-------
413340610 RUN1 1/28/B6
ABS.
0. 0020—
0. 0010—
0. 0000—
-0. 0010—
1300
1400
1200
1600
1500
1100
1000
900
BOO
700
Wavanumbar
FIGURE 3. A) SPECTRUM OF PEAK #2
B) MI REFERENCE SPECTRUM OF 1,3,7,9-TCDO
48
-------
413340610 RUN1 1/23/06
ABS.
0. 0008—
0. 0006—
0. 0004—
0. 0002
0. 0000
-0. 0002—
-0. 0004—
7UU
900
1000
1600
1400
1200
Wavanuinber
FIGURE 4. A) SPECTRUM OF PEAK #3
B) MI REFERENCE SPECTRUM OF 1,3,6,9-TCDD
49
-------
413340610 RUN 1 1/29/86
J L
K-
1600 1500 1400 1300 1200 1100 1000
Wavonuittbor
900 UUO 700
FIGURE 5. A) SPECTRUM OF PEAK #4
B) MI REFERENCE SPECTRUM OF 1,3,7,8-TCDD
C) MI REFERENCE SPECTRUM OF 1,2,4,8-TCDD
50
-------
413340610 RUN1 1/29/86
ABS
0. 0014
0. 0012
0. 0010
0. 0008
0. 000G
0. 0004
0. 0002
vuo
ODD
900
loao
1100
1300
MOO
1COO
1500
Wavenumber
FIGURE 6. SPECTRUM OF LEADING EDGE OF PEAK #5
51
-------
413340610 RUN1 1/20/B6
ABS.
0. 0015
0. 0010
0. 0005
0. 0000
-0. 0005
-0. 0010
-0. 0015
1400
~r~ ~r
1 GOO 1500
1300 1200
i r
1100 1000
7QU
Wavonuaibor
FIGURE 7. A) SPECTRUM OF TRAILING EDGE OF PEAK #5
B) MI REFERENCE SPECTRUM OF 2,3,7,8-TCDF
C) MI REFERENCE SPECTRUM OF 12C 2,3,7,8-TCDF
52
-------
413340610 RUN1 1/28/B6
0. 0010
0. 0008
~. 0006
0. 0004
0. 0002
I
I
-0. 0002
700
800
goo
1000
1100
1200
1300
1600
1500
Wavsnumbar
FIGURE 8. RESULT OF DOUBLE SUBTRACT OF 13C and 12C 2,3,7,8-TCDF
FROM SPECTRUM OF TRAILING EDGE OF PEAK #5
53
-------
413340610 RUN! 1/28/86
ABS.
0. 0030—
0. 0020—
0. 0010—
0. 0000
-0. 0010—
-0. 0020—
1500
1300
1600
1400
1200
1100
1000
90U
EiUO
Wovenumber
FIGURE 9. A) SPECTRUM OF PEAK #6
8) MI REFERENCE SPECTRUM OF 1,2,6,8-TCDD
54
-------
413340610 RUM 1/28/06
ABS.
Aull
• j
1 a /v. A A. JL .Jl_.il..
i A
* . I\ k > 1. ~.j .. ..
1600 1500 1400 13QO 1200 1100 1000 900 BOO 700
Wavanumbor
FIGURE 10. A) SPECTRUM OF PEAK #7
B) MI REFERENCE SPECTRUM OF 13c 2,3,7,8-TCDD
C) HI REFERENCE SPECTRUM OF 12C 2,3,7,8-TCDD
55
-------
413340610 RUN1 1/28/BB
0. 006—
16DQ
1300 1200 1100
Wavanumbar
1000
700
FIGURE 11. A)
B)
C)
RESULT OF SPECTRAL SUBTACT OF 13c 2,3,7,8-TCDO
FROM SPECTRUM OF PEAK #7
MI REFERENCE SPECTRUM OF 1,2,3,8-TCDD
Ml REFERENCE SPECTRUM OF 3,4,6,7-TCDF
56
-------
GAS CHROMATOGRAPHY/MATRIX ISOLATION INFRARED SPECTROMETRY
FOR THE ANALYSIS OF ORGANIC COMPOUNOS IN AIR
Jeffrey W. Childers
Northrop Services, Inc. - Environmental Sciences
P. 0. Box 12313
Research Triangle Park, North Carolina 27709
Nancy K. Wilson and Ruth K. Barbour
U. S. EPA, Environmental Monitoring Systems Laboratory
Research Triangle Park, North Carolina 27711
Abstract
The combination of gas chromatography and matrix isolation Infrared
spectrometry (GC/MI-IR) is currently being evaluated for the characteriration
of semi volatile organic compounds (SVOCs) in air. The ability of GC/MI-IR to
distinguish between isomeric polycycllc aromatic hydrocarbons (PAHs) Is demon-
strated. The ability of GC/MI-IR to identify specific PAHs found 1n complex
environmental samples is illustrated by the analysis of an extract from NBS
standard reference material 1649 (urban dust/organic).
Introduction
Over the past thirty years, matrix isolation has been an important sampling
technique for the infrared characterization of unstable, reactive and transient
species <1). To a lesser extent, matrix isolation Infrared spectrometry (MI-IR)
has been used for the qualitative and quantitative analysis of stable compounds
(2). Recently, MI-IR has been combined with gas chromatography for the analy-
sis of complex mixtures (3). The combination of gas chromatography and matrix
isolation infrared spectrometry (GC/MI-IR) involves the trapping of the GC
effluent in a frozen argon matrix as it is deposited on a rotating gold-plated
cryogenic disk (3). The cryogenic disk is enclosed 1n an evacuated chamber and
is maintained at approximately 14°K. Therefore the separated GC effluents
remain frozen on the disk indefinitely and can be analyzed by Infrared spec-
trometry after the completion of the GC run.
GC/MI-IR offers several advantages over 11ghtpipe-based "on-the-fly" GC/IR
systems. The most significant advantage Is an Increase 1n sensitivity. This
is realized by the ability to signal-average the Infrared spectrum of the
matrix-isolated compound from the GC effluent and by the higher concentration
of that compound in a small cross-sectional area on the cryogenic disk. Matrix
Isolation Infrared spectra also exhibit sharp spectral features, which result
from the elimination of band broadening due to molecular rotations and Inter-
molecular Interactions. Another advantage of GC/MI-IR Is that the separation
57
-------
and the detection of components in a mixture are independent steps. Therefore,
both the chromatography and the spectroscopy can be optimized.
Efforts in our laboratory involve the development and evaluation of
methods for the characterization of semi volatile organic compounds (SVOCs) in
air. The application of GC/MI-IR to the Identification of target polycyclic
aromatic hydrocarbons (PAHs) is illustrated by the analysis of an extract of
NBS standard reference material 1649 urban dust/organic (SRM 1649).
Experimental Methods
All data were obtained on a Mattson Instruments Cryolect system. The
Cryolect is interfaced to a Mattson Instruments Sirius 100 Fourier transform
infrared spectrometer and a Hewlett-Packard 5890A gas chromatograph equipped
with a flame ionization detector (FID), on-column injector, and a Hewlett-
Packard 3392A integrator. GC separations were carried out using a 30 m x 0.25
mm DB-5 fused silica capillary column with a 0.25 urn film thickness. The GC
oven was held at an initial temperature of 40°C for 4 min then temperature
programmed at 8°C/min to 300°C and held at 300°C for 30 min. The effluent from
the GC column was split approximately 5:1 between the cryogenic disk and the
GC/FID. A weighed amount of SRM 1549 was Soxhlet-extracted for approximately 40
h in methylene chloride. The methylene chloride extract was concentrated, the
solvent exchanged to hexane, and then fractionated on a silica gel column. A 1 uL
aliquot of this fractionated extract was injected into the Cryolect system. All
infrared spectra were collected at a nominal 4 cm - 1 resolution and signal-
averaged for 128 scans.
Results and Discussion
The capabilities of GC/MI-IR for the characterization of SVOCs 1n air are
illustrated by the analysis of the SRM 1649 extract, for which a portion of the
GC/FID trace is shown in Figure 1. Previously, a reference MI-IR spectral
library was compiled of target PAHs. The capability of MI-IR to distinguish
between isomeric PAHs is illustrated by the comparison of the MI-IR spectra of
benzo[a]pyrene and benzo[e]pyrene, shown in Figure 2. Other PAH Isomeric pairs,
such as benzo[b]fluoranthene/benzo[k]fluoranthene and chrysene/trlphenylene can
also be easily distinguished by their MI-IR spectra. This is extremely important
considering the wide range of toxicities and the difficulty 1n separating
and/or differentiating PAH isomers by other analytical techniques (e. g.t gas
chromatography/mass spectrometry).
Several specific PAHs 1n the SRM 1649 extract were Identified by GC/MI-Ir,
The MI-IR spectrum of the component which elutes at 25.64 m1n, a retention time
close to that expected for either phenanthrene or anthracene, 1s shown 1n
Figure 3a. A search of that spectrum versus the reference MI-IR spectral
library of several PAHs clearly 1ndent1fies the component as phenanthrene and
not anthracene (compare Figure 3a with Figures 3b and 3c). Similarly, the
compounds elutlng at 29.50 min and 30.20 m1n were Identified as fluoranthene
and pyrene, respectively.
38
-------
The MI-IR spectrum of the component eluting at 37.54 min indicates the
presence of a compound with a carbonyl functional group. Upon closer examina-
tion of its spectrum shown in Figure 4a, spectral features that are character-
istic of benzo[b]fluoranthene are also apparent (Figure 4b). By subtracting a
MI-IR spectrum collected at the tailing edge of the GO peak at 37.54 m1n from a
MI—IR spectrum collected at the leading edge of the same peak, the presence of
benzo[b]fluoranthene is confirmed (compare Figures 5a and 5b), By using either
spectral subtraction or library searches, or the combination of both, the fol-
lowing PAHs were also identified in the SRM 1649 extract: chrysene and tripheny-
lene (34,28 m1n), benzo[e]pyrene (38.39 min) and benzo[g,h,1]perylene (44.25
min),
Conclusions
The combination of gas chromatography and matrix Isolation infrared spec-
trometry has been shown to be a viable technique for the identification of
specific PAH in a complex environmental sample. The ability of GC/MI-IR to
distinguish between isomeric PAHs gives it a tremendous advantage over other
analytical techniques. GC/MI-IR has the capability to identify target compounds
even if they are not completely chromatographically separated from other com-
ponents in the mixture. GC/MI-IR also has the sensitivity required for the
detection of the trace amounts of pollutants typically found 1n environmental
samples. The combination of molecular specificity and high sensitivity indi-
cates a great potential for the utilization of GC/MI-IR for the analysis of
complex environmental samples.
References
1. D. W. Green and G. T. Reedy, "Matr1x-1solatlon studies with Fourier trans-
form Infrared", 1n Fourier Transfrom Infrared Spectroscopy: Applications
to Chemical Systems, vol 1, J. R. Ferraro and L. Ti Basile, Eds., Aca-
demic Press, New York, 1978, pp. 1-59.
2. G. Mamantov, A. A. Garrison and E. L. Wehry, "Analytical applications of
matrix isolation Fourier transform Infrared spectroscopy", Appl. Spectrosc.
36: 339 (1982).
3. S. Bourne, G. Reedy, P. Coffey and D. Mattson, "Matrix Isolation GC/FTIR",
Am. lab. 16: 90 (1984).
59
-------
r»*
Hit
Figure 1. GC/FID trace of SRM 1649 extract.
60
-------
u m ii a*
-I 1 1 L
II H.I ILI || u
—I 1 1 L_
Banxof oJ pyr*r»
.ji
Baniol*] pyrerw
juu
t i i if i i r i ii i n tt i i i 111 11 111 i 11 11 11 i I i i 111 111 11 11 i i
i i T | i r i i pi m ) i i i 111 11 i [ i i 11 |'i i i i ) i i n ( 11 i 11
tWMiiMjaiiaataiMiMiM m
¦MMbar
Figure 2. MI-IR spectra of ben*o[a]pyrene (a) and benzo[e]p.yrene {b).
SRM IMS R«iantlon Tlm« 25.84
. i k
cco
onOinoo«o#
IrAV^1'11"1 ^ ' 1' ^ V| ' H ¦I
i n | n
I
Figure 3. MI-IR spectrum of component elutlng at 25.64 mln 1n SRM 1649
extract (a) compared to MI-IR reference spectra of phenanthrene
(b) and anthracene (c).
61
-------
u
LI II If u u u III n M
J I I I 1 1 1 1... J.
SUM 1S4» Retention Ttma: 37.54
U^LjU
BanzoC U fluoranttww
-I 1 T
->—r
T—r-
tonnflbar
Figure 4. MI-IR spectrum of component elutlng at 37.54 min 1n SRM
1649 extract (a) compared to MI-IR reference spectrum of
benzo[b]fluoranthene (b).
_L
X
»•
_i_
II M
_l _i
SRM 1640 Dlffannc* Sp*etnwn: 37.42 - 37.00
Figure 5.
VoranuMbar
Difference spectrum obtained by subtracting MI-IR spectrum
at 37.60 min from MI-IR spectrum at 37.42 m1n 1n SRM 1649
extract (a) compared to MI-IR reference spectrum of benzo[b]
fluoranthene (b).
62
-------
SUPERCRITICAL FLUID EXTRACTION AND RECOVERY
OF PAH FROM AIR-BORNE PARTICULATES
Steven B. Hawthorne and David J. Miller
University of North Dakota
Energy Research Center
Campus Box 8213
Grand Forks, North Dakota 58202
Supercritical fluid extraction of PAH from particulates has several
potential advantages over traditional methods such as Soxhlet
extraction and sonication with liquid solvents. The solvent strengths
of supercritical fluids can equal or surpass those of liquids, and can
easily be controlled by choice of the supercritical fluid, the
extraction pressure, or the addition of solvent modifiers. Many
supercritical fluids are gases at room temperature which simplifies
analyte concentration steps. We have developed methodology for the
supercritical fluid extraction of PAH from particulate samples that
yields good recovery, requires only small (1-50 mg) samples, and is
simple and rapid to perform. Supercritical ethane, CO2, and N2O (with
and without 5$ methanol modifier) have been evaluated for their ability
to extract PAH from urban dust and fly ash. The most effective
supercritical fluid (N2O with 5% methanol modifier by volume) yielded
quantitative recovery of PAH from urban dust (SRM 1649) In as little as
30 minutes. In general, 60 minute extractions of PAH from fly ash with
supercritical NjO/S? methanol gave better recoveries than four hours of
sonication or eight hours of Soxhlet extraction using either benzene or
methylene chloride. A method for the direct coupling of the
supercritical fluid extraction step with GC/MS and GC/FID has also been
developed that allows for maximum sensitivity by transferring all of
the extracted analytes directly into the gas chromatographic column for
cryogenic trapping prior to gas chromatographic analysis. Analysis of
PAHs from urban dust (SRM 1649) using coupled supercritical fluid
extraction with GC/MS gave good agreement with certified values.
63
-------
Introduction
The rapid arid accurate i dent i ft cat ion artd quantitation of organic
species associated with air-borne particulates is often limited by the
extraction and recovery methods required prior to chromatographic
analysis. Traditional extraction techniques generally require large
volumes of ultra-pure liquid solvents and several hours of sonication
or extraction in a Soxhlet apparatus. After extraction, concentration
steps are generally required, which increases the time required for
analysis and may lead to analyte loss or degradation. Supercritical
fluid extractions have recently been reported to be a powerful
alternative method for the extraction and recovery of organic analytes
such as polycyclic aromatic hydrocarbons (PAHs) from solid samples''^.
Supercritical fluids can have solvent strengths that equal or surpass
those of liquid solvents, and the solvent strength can easily be varied
by changing the extraction pressure (and therefore, the fluid density)
or by the addition of a polarity modifier such as methano'l. Since many
supercritical fluids are gases at room temperature, analyte
concentration steps are simplified and the direct coupling of the
supercritical fluid extraction step with capillary gas chromatography
is feasible.
This report describes the development and use of supercritical fluid
extractions (SFE) for the extraction and recovery of organic analytes
from air-borne particulates. The abilities of several supercritical
fluids to extract PAHs from urban dust and fly ash are compared with
each other as well as with liquid solvent extractions using sonication
and Soxhlet extraction. A method is also presented for the direct
coupling of the supercritical fluid extraction step with capillary gas
chromatography (SFE-GC). SFE and coupled SFE-GC were used for the
extraction of PAHs from National Bureau of Standards SRM 1649 (urban
dust) and quantitative values obtained using GC/MS analysis are
compared with the certified concentrations.
Experimental Methods
Supercritical fluid extractions were performed using on SFT Model
250-TMP supercritical fluid pump (Lee Scientific, Inc.). Five percent
by volume methanol modified CO2 and N2O were prepared by pipetting
methanol Into to the empty pump, then filling the pump with liquid CO2
(or N2O). (Caution, since N2O is an oxidant, extractions of large
quantities of easily oxidized materials may represent an explosion
hazard and should be avoided. As an added precaution for this study, a
pressure relief valve set to vent at 400 atm was installed at the top
of the pump so that both the pump and extraction cells could be rapidly
vented.) Extraction cells were constructed from stainless steel
fittings (Parker brand) and consisted of a 1/16 In. female WT X 1/16
in. tubing union, a 0.5 jjm stainless steel frit, and a 1/16 in. male
NPT X 1/16 in. tubing union. Supercritical pressures were maintained
inside the extraction cells by using 15 to 30 /jm i.d. X 150 jim o.d,
fused silica capillary tubing for outlet restrlctors. Temperature was
maintained during extraction by inserting the celt Into a thermostatted
tube heater. The species extracted during SFE were collected by
inserting the outlet restrictor into a vial containing 2 mL methylene
chloride and 0.5 ,ug of 4,4 -dlchlorobiphenyI as an Internal standard.
PAH quantitations were performed using a Hewlett-Packard Model 5985B
GC/MS in the selected ion monitoring (SIM) mode.
64
-------
The direct coupling of the supercritical fluid extraction step with
the gas chromatographic column was achieved by inserting the SFE outlet
restrictor capillary directly into the gas chromatographic column
through the on-column injection port. Extracted species were
cryogenically trapped in the gas chromatographic column which was held
at 5°C. After the extraction was completed, the restrictor capillary
was withdrawn from the injector and gas chromatographic analysis was
performed in a normal manner.
Results
A comparison of the abilities of supercritical ethane, CO2, N2O,
002/5$ methanol, and ^0/5% methanol to extract PAHs from urban dust
and fly ash is given in Table I. Triplicate extractions were performed
at 300 atm with each fluid of urban dust and fly ash at 45°C (for the
pure fluids) or 65°C (for the methanol modified solvents). The
recovery of PAHs from the urban dust (20 mg samples) was based on the
values certified by N8S. PAH recovery from fly ash (50 mg samples)
was based on deuterated spikes (2 >ig/g each of diQ-phenanthrene, d^Q"
pyrene, and dupery I ene). Extraction conditions were purposely chosen
(i.e., 30 mln extractions using a 20 jrn l.d, outlet restrictor) to
yield less than quantitative recovery in order that comparisons between
the different supercritical fluids could be made. The results of the
comparative extractions (Table I) show that the methanol modified
solvents yielded better recovery of the PAHs from both the urban dust
and fly ash. The replicate extractions showed good reproducibility.
Lower molecular weight species were more easily extracted than higher
molecular weight species in every case, as would be expected based on
the higher solubility of the lower molecular weight species.
A comparison between the extraction efficiencies obtained for PAHs
from fly ash using the best supercritical fluid tested, N20/5X
methanol, and those obtained using sonication and Soxhlet extraction Is
shown in Figure 1. SupercrIticaJ fluid extractions were performed at
350 atm for an hour using a 30 jjm l.d. outlet restrictor. Soxhlet
extractions were performed for 8 hours using 1-gram samples and 50 mL
solvent. Sonlcatfon extractions were performed using 0.5-gram samples
for 4 hours with 10 mL solvent (4,4' -dichlorobiphenyI was used as an
Internal standard In each case). Figure 1 shows that, In most cases,
the supercritical fluid extraction gave better PAH recovery In 1 hour
than either 4 hours of sonication or 8 hours of Soxhlet extraction.
Sonication did not yield better recovery of any of the PAHs, while
Soxhlet extractions yielded better recovery only of the d}2~perylene.
The use of coupled SFE-GC/MS Is demonstrated In Figure 2 by the
analysis of organics from cigarette ash. A 2-mg sample was extracted
for 10 min with 300 atm N2O and the extracted organics were
cryogenically trapped in the chromatographic column (60 01 X 250 pm l.d.
DB-5) as described above. As shown by Figure 2, reasonable
chromatographic peak shapes were obtained using the coupled SFE-GC
technique. The total extraction and analysis time was 40 min.
Coupled SFE-GC/MS was also used to demonstrate the potential for
class-selective extractions using supercritical fluids. A 1-mg sample
of urban dust (SRM 1649) was extracted with 75 atm N2O for 5 min
(extraction 1) with the extracted species being collected directly in
the gas chromatographic column as described above. After the SFE-GC/MS
analysis of this first extract was completed, a second extraction of
65
-------
the same sample was performed using 300 atm N2° f°r 15 min. As shown
in Table It, most of the alkanes were extracted during the 75 atm
ex-traction, while most of the PAHs were not extracted until the 300 atm
extraction. These initial results demonstrate that the coupled SFE-GC
technique can be used to perform class-selective extractions without
the need for any intermediate sample collection and class-
fractionation steps between the extraction and the introduction into
the chromatographic column.
The ability of SFE and SFE-GC/MS analysis to yield quantitative
recovery of PAHs from NBS 1649 (urban dust) is shown in Table III. The
SFE extraction was performed for 30 min with 300 atm of ^0/5% methanol
(30 ftm restrictor). Extracted PAHs were collected in methylene
chloride and analyzed as described above. Coupled SFE-GC/MS analysis
was performed using 20 min extractions with 350 atm of N2O with the
extracted species being collected directly in a 30 m X 320 /im i.d. D8-5
gas chromatographic column. As shown in Table III, both the SFE and
the coupled SFE-GC/MS techniques gave good agreement with the certified
values demonstrating the ability of both techniques to yield
quantitative data.
Conclusions
Supercritical fluid extractions can be used to obtain high
extraction efficiencies of PAHs from air-borne particulates in an
order-of-magnftude faster time than traditional liquid solvent
extraction nrathods, Coupling the supercritical fluid extraction step
directly with on-colurrn trapping of extracted analytes (SFE-GC) can
yield quantitative results with a total extraction and analysis time of
less than 1 hour. Coupled SFE-GC has an additional advantage in that
extracted analytes are quantitatively transferred into the gas
chromatographic column thus yielding maximum sensitivity with minimal
sample size.
Acknow Iedgements
The financial support of the U.S. EPA, Office of Exploratory
Research, is gratefully acknowledged.
References
1. M.M. Schantz, S.N. Chesler, "Supercritical fluid extraction
procedure for the removal of trace organic species from sol Id
samples," J. Chromatogr. 363; 397-401 (19B6).
2. S.B. Hawthorne, O.J. Miller, "Extraction and recovery of organic
pollutants from environmental solids and Tenax-GC using
supercritical CO2," J. Chromatogr. Sci. 24; 258-264 (1986).
66
-------
Table I
Comparison of Different Supercritical Fluids for the
Recovery of PAH from Urban and Fly Ash
Percent
Recovery
from Urban
Oust
ethane C02
n2o
C02/Me0H
N20/Me0H
f1uoranthene
53 79
101
103
100
benzllallanthracene
48 86
104
102
101
benzol!a]pyrene
17 33
40
62
86
benzoCgh i ipery1ene
14 17
22
29
39
i ndenoD,2,3-cdD-
15 20
26
34
43
pyrene
Percent
Recovery
from F1y Ash
ethane CO2
n2o
C02/Me0H
N2O/MSOH
diQ-phenanthrene
33 27
42
58
66
d^-pyrene
8 8
12
23
41
d]2~Perylene
1 1
1
3
4
Table II
Class-Selective Extraction of Alkanes and PAHs
From Urban Dust Using Coupled SFE-GC/MS
Percent In Fraction
Extraction 1
Extraction 2
75 atm N20
300 atm N20
n-a1kanes
C22
95
5
c23
94
6
C24
96
4
c25
92
8
c26
89
11
PAHs
phenanthrene
38
62
f1uoranthene
36
64
pyrene
34
66
ben zC a]an t h r acene
21
79
benzoMpyrene
2
98
indenoEl,2,3-cdUpyrene
ND
>95
be n zoLgh i 3pe r y1ene
ND
>95
Table III
Quantitation of PAHs from NBS SRM 1649 (Urban Dust) Using
Supercritical N20/5? Methanol Extraction and Coupled SFE-GC/MS
fIuoranthene
benz[a]anthracene
benzo|Ia]pyrene
benzodghiJperylene
i ndenoD(2,3-cd3-pyrene
Concentration (jug/g)
Certifled
Valuea
7.1 ±0.5
2.6 ±0.3
2.9 ±0.5
4.5 ± 1.1
3.3 ±0.5
Supercritical
N20/5* MeOHb
7.4 ± 0.4
2.5 ± 0.3
2.9 ± 0.2
3.2 4 0.2
2.3± 0.2
Coupled
SFE-<3C/MSC
7.3 ± 1.0
2.6 ±0.8
2.8 ± 0.5
3.6 ±0.9
3.0 ±0.5
^alue certifled by the National Bureau of Standards.
b8ased on triplicate extractions of 20-mg samples.
cBased on four repl icate analyses of 2-mg samples.
67
-------
d1QPh.n0nthr«,n. (188) M d^pyrena (212) [ZD d10peryl«n« (264)
12
125 - -
100-
Supercritical
-- 125
--100
Q)
>
O
o
Q)
a:
Soxhlet (8 hrs)
Sonication
(4 hrs)
ch.c
60 mm
(SOpvi)
ch2ci2
B«nz«n«
Figure 1. Recovery of PAHs from fly ash using supercritical N20/5%
methanol extraction for 1 hour, Soxhlet extraction with methylene
chloride and benzene for 8 hours, and sonication wi+h methylene
chloride and benzene for 4 hours.
10 15
Retention Time (min)
20'
Figure 2. Coupled SFE-GC/MS analysis of a 2-mg sample of cigarette
ash. Total extraction and analysis time was 40 min. Tentative
identifications were based on their mass spectra.
68
-------
ANALYSIS OF AIR PARTICULATE
SAMPLES BY SUPERCRITICAL
FLUID CHROMATOGRAPHY
Robert D. Zehr
Northrop Servicea, Inc. - Environmental Sciences
Research Triangle Park, North Carolina 27709
Nancy K. Wilson
Environmental Monitoring Systems Laboratory
U. S. EPA, Research Triangle Park, NC 27711
Several types of source samples have been analysed by
packed column supercritical fluid chromatography for
polycyclic aromatic hydrocarbons. The SFC-variable
wavelength UV detector combination provides rapid analyses,
high specificity, and moderate chromatographic resolution
at low separation temperatures. However, low detector
sensitivity makes analysis of ambient air samples difficult.
69
-------
ANALYSIS OF AIR PARTICULATE SAMPLES
BY SUPERCRITICAL FLUID CHROMATOGRAPHY
Introduction
The Environmental Monitoring Systems Laboratory of the
Environmental Protection Agency continually seeks to apply
new techniques to measure air pollutants. This paper will
describe some of our continuing work to apply packed column
supercritical fluid chromatography to analyze trace organic
compounds in air particulate samples.
Experimental Methods
Equipment
The packed column chromatograph is a Hewlett-Packard
Model 1082B Liquid Chromatograph equipped with two high
pressure solvent pumps, a manual valve injector, a heated
column oven, and a programmable variable-wavelength UV
detector. Factory-installed options (1) to permit the use
of supercritical fluids as mobile phases are as follows:
the pumpheads are insulated and cooled by a circulating
ethylene glycol-water mixture at -20C. A heat exchanger is
placed between the column and detector to eliminate light
scattering caused by the turbidity of supercritical fluids.
The flow cell in the UV detector has thick windows to permit
high pressure operation. System pressure is maintained by a
precision back-pressure regulator installed after the UV
detector. Finally, Bystem pressure is monitored by two
pressure gauges: one measures column head pressure before
the valve injector; the other measures system pressure after
the UV detector.
A Vydac TP 201 octadecylsilane (ODS) HPLC column was
used in this work. Solvents were pesticide quality or
equivalent. "SFC Grade" carbon dioxide was purchased from
Scott Specialty Gases, Inc. Methanol-carbon dioxide
mixtures were prepared by weight in a 1-liter stainless
steel sampling cylinder.
When separations required gradient elutions, the
chromatograph was programmed to change the mobile phase
composition linearly from 1% to 10% methanol/carbon dioxide
(w/w). The column oven temperature was set at 65C. The
variable-wavelength UV detector was programmed to change
monitoring wavelengths at preset times to allow selective
detection of each component. Peaks from unknown samples
were tentatively identified by comparison of retention times
at specific wavelengths with those obtained from standard
solutions. Quantitation was performed by mapual
measurements of peak heights, since high noise levels made
the instrument integrator somewhat unreliable.
Attempts were made to install a flame ionization
detector (FID) onto the chromatograph. Accordingly, varying
70
-------
lengths of 5-, 10-, and 20-micrometer ID fused-ai1ica
capillary tubing (Scientific Glass Engineering, Austin, TX)
were connected to the column outlet to provide flow
restriction and to maintain system pressure above the
critical pressure; the free ends were inserted into the
flame tip of the FID. Flow rates greater than 0.2 mL/min of
liquid carbon dioxide either overpressurized the system or
snuffed the flame. The instrument was unable to provide
stable flow rates less than 0.2 mL/min.
Sample Extraction and Cleanup
Samples of National Bureau of Standards (NBS) SRM 1648
(Urban Particulate Matter) were weighed into pre-extracted
cellulose extraction thimbles and extracted with methylene
chloride for 16-24 hours in a Soxhlet extractor. The
resulting solutions were concentrated and solvent-exchanged
into hexane. Aliquots of extracts supplied in methylene
chloride were solvent-exchanged into hexane. Extracts were
fractionated by column chromatography over silica gel.
Standard mixtures were eluted quantitatively with 1/1
methylene chloride/hexane. Chromatograms of unknown samples
were free of obvious interferences.
Results and Discussion
The polycyclic aromatic hydrocarbons (PAH) used as
reference standards in this work are shown in Table I.
These PAH have from 3 to 7 fused aromatic rings and have
molecular weights ranging from 178 to 300. Each is
frequently found in air particulate samples; several are
proven or suspect carcinogens or mutagens.
The PAH fraction from SRM 1648, "Urban Particulate
Matter" is shown in Figure 1. This SRM was collected in St.
Louis, MO, and is certified for various inorganic
constituents. Concentrations for individual PAH compared
favorably to the values obtained by Wise and co-workers
(2). The large amounts of sample used and the high noise
levels shown in the chromatogram indicate that the packed-
column SFC-variable wavelength UV detector combination may
not be well suited to trace organic analysis in ambient
air.
Chromatograms from source samples are shown in Figures
2 through 4. Figure 2 shows PAH in the particulate fraction
from the exhaust of a Volkswagen Rabbit diesel engine. The
pyrene peak represents nearly 14,000 ug per gram
particulate. Higher molecular weight components are present
in significant quantities. The coronene concentration is
approximately 500 ug per gram particulate. For comparison
the chromatogram of a gasoline-powered engine is shown in
Figure 3. The pyrene peak represents approximately 1000 ug
per gram particulate. Coronene is present to the extent of
about 600 ug/gram particulate.
71
-------
To obtain the chromatograms shown in Figure 4, flue
gases from wood stoves were collected using a modified
Method 5 sampling train. Particulate matter was sampled on
a filter. More volatile constituents passed through the
filter and were trapped on an XAD-2 cartridge. The
chromatograms clearly show that the filter trapped the
higher molecular weight hydrocarbons, as expected, and that
the XAD-2 trapped the more volatile PAH.
Conclusions
Several types of air particulate samples have been
successfully analyzed for polycyclic aromatic hydrocarbons
by packed-column supercritical fluid chromatography. The
cleanup scheme provides effective purification and high
recoveries of PAH. The packed column SFC in combination
with its variable wavelength UV detector yields rapid
analyses, with moderate chromatographic resolution at low
oven temperatures. However, low detector sensitivity makes
analysis of ambient air samples difficult.
Acknowledgments
We thank Drs. Roy Zweidinger and Ray Merrill of EPA for
the gifts of source extracts.
References
1. D. R. Gere, R. Board, and D. McManigill,
"Supercritical Fluid Chromatography with Small Particle
Diameter Packed Columns", Anal. Chem. 1982. 54, 736-
740.
2. S. A. Wise, B. A. Benner, S. N. Chesler, L. R. Hilpert,
C. R. Vogt, and W. E. May, "Characterization of the
Polycyclic Aromatic Hydrocarbons from Two Standard
Reference Material Air Particulate Samples", Anal.
Chem. 1986. 58, 3067-3077.
PAH Reference Compounds
Anthracene
Pyrene
Benz[a]anthracene
Benzo[e J pyrene
Benzo[g,h,i]perylene
Fluoranthene
Chrysene
Benzo[a J pyrene
Benzo[b]fluoranthene
Coronene
Table I. PAH used in this work
72
-------
SRM 1648, 0.5798 g
Flash Chromatography
Vydoc—ODS
288 Bar
65 C
1% to 10% MeOH
2 mL/min
Figure 1. Chromatogram of SRM 1648 extract
VW Rabbit - Diesel
Rash Chromatography
Vydac-ODS
285 Bar
65 C
1% to 10% MeOH
2 mL/min
BaA
Figure 2. Chromatogram of VW-Rabbit diesel exhaust extract
73
-------
" * BaP
Ford Van - Gasoline
Flash Chromatography
Vydac-ODS
2B5 Bar
65 C
1% to 10% MeOH
2 mL/min
Cor
Figure 3. Chromatogram of Ford van gasoline exhaust extract
%
a
r. n/ vPy
Standards
r1
Wood Smoke Extracts
Flash Chromatography
Vydac-ODS
28B Bar
65 C
Carbon Dioxide
Density -0.8
2 mL/min
XAD-2
Figure 4. Chromatograms of wood stove emission extracts
74
-------
Mobile Sources of Polycyclic Aromatic Hydrocarbons
(PAH) and Nitro-PAH: Results of Samples Collected
in a Roadway Tunnel,
Bruce A. Benner, Jr. and Glen E. Gordon
Department of Chemistry and Biochemistry
University of Maryland
College Park, MD 20742
Stephen A. Wise
Organic Analytical Research Division
Center For Analytical Chemistry
National Bureau of Standards
Gaithersburg, MD 20899
A recent review article [1] emphasized the need for further
characterizations of the carbonaceous fraction of mobile source
emissions, particularly with the impending removal of lead alkyl octane
boosters and bromine-containing lead scavengers from regular leaded
gasolines. The lead and bromine emitted from the combustion of these
fuels have been used as tracers of mobile source emissions for a number
of years. Single vehicle emission studies have shed light on the
relationship between engine operating parameters and the chemical
characteristics of the emissions but they are not suitable for use in
source apportionment studies which require emission data from a large
number of different vehicles. Air particulate samples collected near a
busy highway or in a roadway tunnel would be more appropriate for use
in estimating the mobile source contribution of organic compounds to a
region.
Suspended particle samples collected in a heavily-travelled
roadway tunnel (Baltimore Harbor Tunnel, Baltimore, Maryland) were
characterized for polycyclic aromatic hydrocarbons (PAH) and some
nitro-PAH by gas and liquid chromatographic techniques. These samples
included those collected on Teflon filters and on glass fiber filters
for investigating any differences in samples collected on an inert
(Teflon) and more reactive (glass-fiber) medium. All samples collected
on Teflon were backed-up with polyurethane foam plugs (PUF) which
trapped any inherent vapor-phase PAH as well as any compounds "blown-
off" the particles during collection. Particle loadings in the tunnel
were high enough for sufficient sample to be collected in 1 h periods
(approx 50 m^ volume collected) so that relationships between the
composition of the emissions and the types of vehicles using the tunnel
75
-------
might be investigated. Two background air samples (Teflon filters +
PUFs) were also collected for comparison of samples resulting from a
mixture of stationary and mobile sources with those purely from mobile
sources.
References
1, Daisey, J. M. ; Chesney, J. L, ; Lioy, P.J. J_,_ Air Pollut. Contr.
Assoc. 1986. 36, 17-33.
76
-------
FIELD EVALUATION OF AN ADSORPTION-THERMAL
DESORPTION TECHNIQUE FOR ORGANIC CONTAMINANTS
IN INDOOR AIR
David T. Williams
Environmental Health Directorate
Health and Welfare Canada
Tunney1s Pasture
Ottawa, Ontario
K1A 0L2
Cecilia C. Chan
John W. Martin
Mann Testing Laboratories Ltd.
5550 McAdam Road
Mississauga, Ontario
L4Z 1P1
This study evaluated, under field conditions, the ability of
the multi-sorbent sampling tube thermal desorption technique
to identify and to provide quantitative data on selected
volatile organic contaminants in indoor air and to provide
information on typical levels of these pollutants in the air
of homes compared to ambient air at the same location. Air
samplers containing multi-sorbent tubes attached to air
sampling pumps set at different flow rates were used to
collect air samples at twelve homes. The sorbent tubes were
analysed by thermal desorption coupled with gas
chromatography-mass spectrometry. The precision and
accuracy of the method under field conditions and the
stability of the organic contaminants during storage of the
sorbent tubes are evaluated. Typical results are discussed
together with some difficulties encountered in the analyses.
77
-------
INTRODUCTION
The adsorption/thermal desorption technique (ATD) has
been widely used, because of its potential sensitivity, for
the analysis of trace levels of organic compounds in air.
Many studies have used Tenax as the trapping media (1-3).
However, we have previously reported on a laboratory
evaluation of a multi-layer adsorption tube for the
collection and analysis of a wide range of organic compounds
in air (4). The present study was designed to evaluate the
use of this technique under field conditions for a selected
group of target compounds.
EXPERIMENTAL
Materials
Sorbent cartridges were obtained from Envirochem Inc.
(Kemblesville, Pa.) and were constructed of pyrex glass (20
cm x 6 mm 0D and 4 mm ID) and packed with sequential layers
of glass beads, Tenax, Ambersorb XE-340 and charcoal
adsorbents. Prior to use the tubes were cleaned by thermal
desorption in a sorbent tube conditioner while being purged
with high purity nitrogen at a flow rate of 50 ml/min.
Apparatus
Analyses were performed using a modified Envirochem
concentrator, Model 780B, interfaced to a Hewlett Packard,
Model 5890, gas chromatograph-Finnigan, Model 3300, mass
spectrometer combination. An Envirochem automatic desorber
unit was attached to the concentrator, with a separate
module controlling the temperature and timing of the
process. Desorption conditions were: transfer lines at
280°C; initial purge time of 2 min; sorbent tube desorbed at
280°C for 12 min with helium flow of 50 ml/min; traps 1 and
2 desorbed at 300°C with flows of 50 and 2 ml/min
respectively onto a J&W DB-5 fused silica capillary column
(30 m x 0.25 mm, 1 um film thickness) maintained at 40°C.
The GC conditions were: column carrier gas (helium) flow of
1.5 ml/min; initial temperature at 40°C for 5 minutes;
temperature program rate at 8°C/min to 2 80°C and held for 2 0
minutes; the MS settings were: electron multiplier at 18 00
eV; electron energy at 70 eV; scan rate of 1 sec/scan; mass
range of 34 to 300 a.m.u.
Sample Collection
Air samples were collected at twelve homes during
November and December 1986. At each home the following
samples were collected: two ambient air samples, four indoor
air samples of various volumes and one indoor air sample
using two sorbent tubes connected in series. The ambient
air samples were collected adjacent to the home and away
78
-------
from any obvious source of pollution. The indoor air
samples were collected on the main floor of the home,
usually in the living or family room. Indoor and ambient
air samples were collected at the same time; a uniform 90
minute sampling time was used and pump flow rates were
adjusted to sample the required volume of air. After sample
collection the sorbent tubes were placed in individual screw
cap glass tubes and then stored in a tightly sealed
container until analysed. Analysis was carried out within
two days.
Sample Analysis
Following thermal desorption the GC-MS program was
begun and full scan mass spectra were collected. Target
compounds were identified by retention times and by full
mass spectra or by the presence of appropriate fragment
ions. Quantitation was achieved by comparison of peak areas
from reconstructed chromatograms of the two or three most
intense ions from each compound with corresponding peaks
from a standard mixture.
QA/QC Procedures
A blank tube was carried to and from each home and
handled and analysed as a sample, except that no air was
sampled through the tube. Each week three tubes fortified
at a low level (approx 70-80 ng) and three tubes fortified
at a medium level (approx 700-800 ng) with a standard
mixture of target compounds, together with a blank tube,
were transported to and from one sampling site and analysed
by thermal desorption GC-MS.
To assess the stability of the organic target compounds
on the sampling tube a limited storage study was carried
out. Triplicate sorbent tubes fortified at low and medium
levels (approx 70-80 and 700-800 ng respectively), together
with a blank tube, were stored for 0, 1, 3 and 7 days under
normal storage conditions and then analysed by ATD/gc-ms.
Preparation of Standards
A stock standard solution was prepared by injecting a
known amount (2 00-400 ul) of each compound into a 10 ml
reactiflask fitted with a Minnert valve. Aliquots of the
resulting solution were injected into a helium filled bottle
equipped with a septum top to produce gaseous standards. To
fortify sorbent tubes aliquots of the gaseous standard were
introduced onto the glass bead layer of sorbent tubes held
at room temperature in the deeorber unit.
79
-------
RESULTS AND DISCUSSION
Most of the target compounds investigated in this study
were detected in inc-oor air at l to 10 ugjts' except fcr the
aromatic and straight chain hydrocarbons which tended to be
present at 10 to 50 ug/nis. In a few houses, relatively high
levels (> 100 ug/ms) of chlorinated hydrocarbons were
detected (e.g. 1,1,1-trichlorcethane). Full data analysis
has not yet been completed but air levels for the target
compounds are shown in Tables I and IX for two houses.
House #4 (Table I) shows typical background levels for the
target compounds except for naphthalene whose level is
higher than in the other homes. These data also show that
outdoor air levels are significantly less than indoor air
levels. The agreement of the data is extremely good ( 70%) and the
precision Ls usually better than 15% RSD. Data from the
storag-e study indicated that there was little loss of the
target compounds during seven days storage.
The ATD GG-MS technique, therefore, shows considerable
promise for indoor air sampling but a number of samples of
different air volumes should be collected to avoid the
technical difficulties discussed in this paper.
B0
-------
ACKNOWLEGDEMENTS
The technical assistance of A.M. Taylor and typing by
V. Schellenberg are gratefully acknowledged.
REFERENCES
1. M. P. Ligocki, J. F. Pankow, "Assessment of adsorption/
solvent extraction with polyurethane foam and
adsorption thermal desorption with Tenax-GC for
collection and analysis of ambient organic vapours,"
Anal. Chem. 57: 1138 (1958).
2. S. Zeldis, A.D. Horton, "Trapping and determination of
labile compounds in gas phase of cigarette smoke,"
Anal. Chem. 50: (1978J.
3. H. R. Langenhove, E. A. Van Wassenhove, J. K. Copplin,
M. A. Achen, N. M. Schamp, " Gas chroiaatography/mass
spectrometry identification of organic volatiles
contributing to rendering odours," Environ. Sci.
Technol. 16: 883 ( 1982).
4. C. C. Chan, J. W. Martin, P. J. Pond, D. T. Williams,
"Development of an adsorption/thermal desorption
technique coupled with GC/MS for the monitoring of
trace organic contaminants in indoor air," Proceedings
EPA/APCA Symposium on Measurement of Toxic Air
Pollutants, Raleigh, North Carolina (1986).
81
-------
table I TARGET COMPOUND AIR LEWI S (ug/K3)- WUSE #4
OUTDOORS | INDOORS
1
cowo
1
31 |
]
1
« i
*i i
«? |
»3 !
>4
SERIES
HI FLOW
TOP
SERIES
HIGH FLOW
eon*
ACFTONE I
j
1
!
1
51 8 |
|
59.3 |
SAT. |
SAT
36.3
i.i
METHYL ETHYL KETONE j
1
1
|
I
1
6.8 |
7.0 i
56 i
6,5
5.9
-
J
DIETHYL EMft j
1
|
[
" 1
f
1
I
1.7 !
1
3.5 |
1.5 i
1.6
1 4
-
1
n-aJTANOC !
i
I
j
|
1
1
i
j
-
-
-
l
9ENZENE ]
|
i
1
3$ j
i
1,8 |
I
8,9 |
5.5 j
6.3
6 5
1.6
DiaiL&OfTWM i
1
I
(
1
1
!
i
1
8.2 !
8,3 !
6,7
7.3
9,1
1 2
1
fiEXANE j
1
1
- !
1
- i
i
5.2 i
i
-
5.5
6.6
6.3
0.9
1
EWL ACETATf |
1
1
1
- i
i
1
3,5 |
i
|
4.2 I
4.6
3,3
-
1
TOLUENE j
1
4.5 )
1
I
6 D |
i
1
29.4 I
24,2 |
29.3 i
30,6
31,4
1 4
1
i,i-oiCHLOfloemaE i
|
1
|
i
I
1
j
j
-
-
-
STYRBiF j
1
1
- 1
1
5.6 I
I
5.6 j
5,3 |
5.5
5,6
-
0 & '"-XYLENE j
|
1
1.7 !
1
!
2.2 j
1
20.9 I
1
20,2 |
19.4 j
21.4
21 9
0.7
1
o-WLENE |
1
!
0.7
|
!
1.1 |
j
!
13.5 I
i
12,4 I
12.9 |
13.4
13 2
0.3
1
Em BENZENE j
1
1
1.0
1
i.3 i
1
M
1
8.5 j
7.9 |
8.1
8.8
-
1
OLOROBOZBiE j
1
1
1
1
- 1
!
i
i
|
|
-
-
-
1
CHOROfW 1
1
1
- 1
1
i.i i
i
j
1.0 ]
1.0
1,2
1.0
1
NAPHTHALENE j
|
3.0 |
j
1
2.7 |
1
76 ? |
1
8? 7 ]
66.9 |
73.0
85.6
3.3
I
TRICHLOfiPeTHYLENF I
|
|
i
f
t
j
|
-
-
-
1.1,1-TWCHLORDETHANE
1
i
1
- !
!
t.5 |
1
4.5 |
3.0 |
3,7
4.7
-
I
p-OICHLOfifflENZENE |
I
- i
i
- i
i
3.3 |
t
j
26 |
1,9
2 3
0 5
CAfiBCN TFTWffllOGIDf |
1
i
- i
i
i
!
i
1
1
|
|
-
-
-
1
n-tNOKWE !
1
i
j
1
- i
1
1
59.9 |
1
47 .5 |
54.9 |
59.0
59.0
1.0
1
TETRAfULOfiOPHYLM |
I
i
i
j
i
i
i
4,7 j
1
j
4.3 j
4.5
4.6
-
I
1
8 5 |
1
20.7 !
1
25.4 |
2.9 >
45,2 [
39.6
22.7
22.7
SAT. » SATURATION
82
-------
TABtE II
IXWOtHO 1
OUTDOORS
1NOOOHS
SERIES
HI FLOW
TOP
SERIES
HI FLOW
BOTTOM
*1 !
«
HI
#2
03 |
U
ACETONE j
- |
4.3
SAT,
233 4
SAT, |
i
SAT.
SAT,
0.9
METHYL ETHVL KETONE |
- |
-
n. 7
33.7
1
15,6 |
15.5
17.8
-
DIETHYL ETHER |
- |
-
-
6,5
1
i
-
-
-
n-RJTANOL |
- |
-
29.2
37,4
1
SAT. |
i
SAT.
42.0
TR
BENrae |
H, 1 |
5.5
9.7
17.4
1
9.6 |
i
11.9
11.6
1.0
DICHLOROMETHANE |
1.6 |
1.2
3.5
4.2
1
2.9 |
j
3.1
4.6
0.2
HEXANE |
?.5 |
t.1
15.4
15.1
1
15,9 (
16.0
17.7
TR
ETHYL ACETATE |
-
-
3.0
-
1
3.6 |
I
3.4
2.7
-
TOLUENE ]
11.1 |
14.4
52.1
58 6
1
51,0 !
1
51,2
65,8
0.9
1,1-DJCHLCKOFT>YLFNE j
- j
0,5
1,0
2.0
1.0 1
1
0,9
1,1
-
STYRENE 3
-
o.e
3.1
2.8
1
3.0 |
3,0
-
-
p i ffl XYLENE |
14.9 |
11.5
55.9
63.1
59.9 j
i
64.0
69.5
0.7
o-XYLENE |
I
?,5
18.9
19.6
1
18.T |
I
18,9
18,3
O.I
ETHYL 00CWE |
3,3 |
2.6
12.0
9,6
I
10.7 |
I
11.1
11,9
0.3
CHLORO0ENZENE I
- j
-
-
-
I
I
-
-
-
CHLOROFORM |
- |
0,2
1.0
-
0.7 |
i
1.0
0,7
-
NAFKTHALENE
1,5 |
1.6
5,2
4,5
E.2 I
1
6. G
5,4
0.2
TRICHLOflOETHYLBft
3.3 I
1.1
0,1
-
(.6 |
l
0,6
0,6
-
1,1,1-TmCHLOMETHANE |
- |
-
6,3
5.6
1
4.0 |
1
5 5
5 8
-
(rCICHLOfCTBCB* |
-
-
1,9
-
1
t.3 |
1
?.3
1.8
0,5
CA80N TFTWCHLORICE j
- |
¦
?2
-
1.0 |
I
1.9
2.3
-
MNOEONE I
- |
0.8
26,1
21,9
1
26 6 |
i
28.3
26,5
-
TETWOK-OROETHYLENE 1
- |
1.1
7 2
56
M |
j
6.4
6.8
-
VOWAt SAMPLED (L): |
e.t |
1
3?. 9
26 2
2.3
47.2 |
1
41.7
23.1
23,1
SAT * SATURATION
TR « TRACE AMOUNT DETECTED
83
-------
TABLE III RECOVERIES (ng) OF FORTIFIED TARGET CWOl/NOS
I"
!
cwpcuno |
1
LOW LEVEL SPIKES
1
|
MED]
N LEVEL SPIKES
FORTIFJCAT
CN (nfl)
1
I ? !
3 i
t
1
2 |
3
LOW LEVEL
~K) LEVEL
j"
ACETCNE j
72.1
71.6 |
i i
1
77.9 |
491.0
567 5 j
584.3
79.1
791 0
METHYL ETHYL KETONE j
i
76.fi
i 1
! 69.8 |
i i
75 4 j
1
601.9
736.7 |
743 6
SO ?
802 0
1
DIETHYL FTHER \
i
60.?
! !
| 50.8 |
i i
1
53.7 |
i
428,0
503.1 |
556.4
59.9
699 0
1
n-ftlTANOL I
t
66.6
1 !
1 1
i |
1
51.3 |
i
620.9
721.3
754.6
81.0
810,0
I
BENZENE |
63.8
| 56.4 |
i i
1
63 5 |
i
623.3
808 4 |
864,0
87 3
873.0
DICHLORCMETHANE f
l
78.0
! 1
| 73.6 |
i )
1
79.0 |
I
460.0
544.6 |
577,0
66.1
661.0
1
HEXANE |
52.0
! !
| 46 6 |
| 1
1
49.1 |
i
468 2
567.3 |
579,4
S3 0
630.0
1
ETHYL ACETATE |
|
89.0
! I
| 100.8 !
1 1
I
94.6 |
1
665 5
836.8 j
871.7
90.2
902.0
!
TOLUBIE |
71.0
1 1
1 67 4 1
i i
68.6 |
1
672,2
886.8 !
977.8
86.3
863.0
1,1-DICHLOROETHYLENE |
45.8
1 1
1 <2 2 |
i i
1
44.8 |
1
409.0
507,4 |
509 4
71.6
716.0
I
STYRENE |
101.0
! !
1 8$ 7
i i
f
93 .5 |
\
724 .0
1005,4 j
1199 0
91 1
911.0
|Wi-mB(E !
|
116.7
I 1
1 129.3 |
i i
I
140.6 t
i
713.0
1634 0 |
1924 6
153.0
1530.0
I
o-XYL0(E |
89 6
1 1
1 76.1 |
I |
1
82.1 (
i
688.4
954.0 |
1089,0
88 4
884 0
1
ETHYL BENZENE |
85.6
j 73,5 |
i i
1
79.2 |
i
655.8
881.4 |
1003.2
84 6
846.0
!
CHLORC8ENZENE |
84.6
! 1
| 63 6 |
I
68 .8 |
643.4
881,4 |
979.4
82.5
825,0
CHLOROFORM j
|
62 3
j 58.2 j
i
60,8
1
526 6
646,6 |
688 6
73 7
737.0
!
TRICHLOROETWLENE |
71 6
1 !
| 62 6 |
i t
1
66.8 |
I
551.2
7015 |
773.4
73.5
735.0
i
1,1,1-TRICHLOROETHANE j
t
56.4
t I
1 si.i 1
¦ i
I
58.1 |
i
487.0
610 C j
627 4
64.9
649.0
I
p-DICHLOROBENZENE I
f
126 5
I 1
1 81.1 1
1 |
1
83 1 !
678.3
1052 8 |
1305.6
81,2
812.0
[
CARBON TETRACHLORIDE |
66?
! I
| 65.0 |
i i
69.8 |
I
578,4
742 0 |
763.6
78 9
789 0
1
TETRACHLOROETHYLENE |
78,2
1 1
| 66 ? |
I
70.4 |
615.3
828 8 |
934.6
80 8
808.0
84
-------
PERFORMANCE TESTING OF RESIDENTIAL
INDOOR AIR CLEANING DEVICES
Mallory P. Humphreys,
Energy Use Test Facility, Tennessee Valley Authority,
Chattanooga, Tennessee
A study is underway by the Tennessee Valley Authority to evaluate the performance of a cross
section of the residential air cleaning devices available on the market. The test procedure being
used in the study follows the guidelines of the Association of Home Appliance Manufacturers
"Standard Method (AC-1) for Measuring Performance of Portable Household Electric Cord-
Connected Room Air Cleaners (1)" and is similar to test procedures employed by Lawrence
Berkeley Laboratories in a 1984 study (2) of room air cleaners. The test procedure involves filling
a room-sized test chamber with a contaminant, mixing to obtain a uniform concentration, and
measuring the contaminant decay rate both with ("induced decay") and without ("natural
decay") the air cleaner operating. Mass balance modeling of the various source and removal
terms involved is then used to calculate effective cleaning rate and removal efficiency for each
pollutant monitored. The pollutants monitored include respirable suspended particulates (RSP
[down to 0.03 micron particle diameter]), generated from cigarette smoke injection; and common
gaseous pollutants (CO, NO2, SO2, and hydrocarbons |HC—expressed as equivalent methane]),
generated from cylinder gas injection. Six portable room air cleaners have thus far been tested in
the study: (1) console electrostatic precipitator (ESP); (2) table-top ESP; (3) high efficiency par-
ticulate air (HEPA) filter console; |4) table-top negative ion generator (NIG); (5) catalytic/HEPA
filter console; (6) sorbent/HEPA filter console. Devices 1 through 4 were tested for removal of
RSP only, while devices 5 and 6 were tested for removal of both RSP and gaseous pollutants.
The devices were tested at both maximum and minimum flow rates. A minimum of four replicate
tests were performed for each test condition.
85
-------
PERFORMANCE TESTING OF RESIDENTIAL INDOOR
AIR CLEANING DEVICES
Introduction
Reduction of residential home ventilation rates through increased weatherization or new con-
struction practices, can result in containment of indoor pollutants within homes. One strategy for
controlling residential indoor contaminants that is gaining recognition is air cleaning. In the past
few years a variety of residential air cleaning devices have become available on the market. This
study was initiated by the Tennessee Valley Authority to evaluate the performance of a cross
section of the air cleaners available to the residential consumer.
Theory
The contaminant decay rate in the test chamber is described by the following differential
equation:
= A + PQfCp - QfCj - KCi - nQjCj
dt V V V V
where
Cj = pollutant concentration inside chamber
t = time
S = indoor pollutant generation rate
V = volume of chamber
P = penetration or transmission factor
Qf = infiltrating and/or ventilating air supply
Co = pollutant concentration outside chamber
K = pollutant reactivity or deposition rate
Qd = volumetric air flow rate for the device
rj = system removal efficiency
This mass balance includes two sources of indoor pollutants: the source inside the test cell (S)
and the outside source term (PQfC0); and three removal rates; removal by exfiltrating air (QfCj),
removal by all other forms [e.g., deposition, reaction] (KCj), and removal by the air cleaner
(iQdCj). For a chamber decay experiment, where there is no internal source and where the out-
door aerosol source is negligible, the mass balance equation simplifies to:
d^ = • Q& - KCi - TQdci
dt V V
If it is assumed that the exfiltration and reactive removal terms are the same for the "natural
and "induced" decay periods of testing, then the difference between the observed decay rate9
represents the air cleaner removal term (^Q^Cf/V). The system removal efficiency (SRE) of the
device is represented by "rj". while the effective cleaning rate (ECR) [or clean air delivery rate
(CADR)] is represented by "rjQd".
Results
A summary of the composite average RSP and gaseous pollutant mitigation results
(representing averages of a minimum of four replicate tests) is presented in Table I and in
Figures 1 through 4. AHAM mathematically simulated clean air delivery rate (CADR
[synonymous with RSP effective cleaning rate]) indices required for 75-percent particulate
removal in 30 minutes from rooms varying in size from 60 to 300 sq ft, are presented in Table II
and are included in Figure 1.
RSP effective cleaning rates observed during testing ranged from 30 cfm (for the table-top
NIG) to 125 cfm (for the console ESP). Relative to the AHAM simulated CADR indices
presented in Figure 1, the table-top models tested would provide adequate RSP removal only for
small rooms (less than 80 sq ft), whereas, the console models would provide adequate RSP
removal for rooms ranging from 160 to 300 sq ft in size. RSP removal efficiencies (Figure 2) rang-
ed from 33-percent (for the table-top ESP) to 82-percent (for a HEPA filter console). Highest
RSP removal efficiencies were observed for the three HEPA filter console models and the console
86
-------
ESP under minimum flow rate operation. RSP removal efficiencies for both electrostatic
precipitators were increased during the minimum flow rate operation, because of the longer
residence time of particles within the devices.
Gaseous pollutant mitigation results for the catalytic and sorbent/ HEPA filter consoles are
presented graphically in Figures 3 and 4. The AHAM Test Standard (1) does not specify deter-
mination of gaseous pollutant mitigation performance; however, the principles of mass conserva-
tion outlined in the "Theory" section of this paper apply to gaseous pollutants as well as par-
ticulate matter. Catalytic device 5 contained a 50:50 mixture of activated carbon and specially
developed room temperature catalyst (composed of a solution of copper and palladium salts on a
porous alumina substrate). The manufacturer claims N02 and SOj removal through chemisorp-
tion, CO removal through catalytic oxidation, and HC removal through adsorption. Device 6 con-
tained a specially developed sorbent material (no specific information is available concerning the
nature of the material). Both devices were effective for removal of NOo and SC>2, but were much
less effective for removal of CO and HC. Gaseous pollutant removal efficiencies for devices 5
and 6, respectively, averaged: 37 and 11 percent for NO2, 54 and 32 percent for SO2, 4.1 and
0.4.percent for CO, and 1.6 and 2.2 percent for HC. Device 5 was more effective in removal of
gaseous pollutants than was device 6 because of the combined effect of the catalyst and sorbent
(activated carbon) materials. The removal efficiencies for both devices were higher during
minimum flow rate operation because of the longer residence time of the airstream within the
devices.
References
1. Association of Home Appliance Manufacturers' (AHAM), "Standard Method (AC-1) for
Measuring the Performance of Portable Household Electric Cord-Connected Room Air
Cleaners," August 1985.
2. Offerman, R.J., R.G. Sextro, et. al., "Control of Respirable Particulates and Radon Progeny
with Portable Air Cleaners," Lawrence Berkeley Laboratory Report No, LBL-16659-UC-11,
Lawrence Berkeley Laboratory, Berkeley, California, February 1984.
3. Tennessee Valley Authority, Internal Engineering Reports, Division of Conservation and
Energy Management, Chattanooga, Tennessee, 1986-1987.
87
-------
i»m i, tunwai of dsp rnt basews pou-uimh miismiw ftEsius
([VICE
HUME*
DEVICE
T*PE
FLON
RME
(CUI
POSER
CWSOT,
(MtUI
1 RSP
: ECR
i SHE
: (cfi)
[ (XI
*03
801
CS
TEhP
(••) i
I AIR E1CMMEC I 1
! HIE OF fiQM 1 1
{ (1/hr) J * ) ! i I I I I
I : 40 aq (t ! 10 m It ! 110 aq It I t44 tq It ! 200 H It ! ISO aq ft I 300 tq It t
1 Rom I Rom I loo* Root I Rom I Rom ! Rom (
; 1
| 1
: 2.s
1 12
1 IS [
25
1 30
( «
1 45 1
U
: 1.0
i 35
1 15 1
SO
1 70
1 SO
1 IM !
121
0.4
: :o
: 33 i
10
i 70
to
! 129 :
130
• ttwiin; I It ctilint
To d»t»niH tha CADR riQiiirtd to acconplUh the aquiviliflt rmvil la 40 aintitaf, Ini.t UM India ly J.
la datrnw tin CM* rufnirti Is jctonpliih tha niij»iliit rtM«tl la II ainvtn, aultiplr CMft loin ty 3.
88
-------
FIGURE I 'Summary of RSP Effective Cleaning Rates
HEPH Filter
Console
mrar
iiiix far
7J* mtril
ia 30
llutn far
ACTH (LOJhri
Catalytic/
HEPfl Filter
Consols
Sorbent/
HCT Fllta-
Consols
Console
TablB-Top
ESP
Tcble-Top
NIG
an ft
U elm
-------
FIGURE 3 :Su(n«ary of Gaseous Pollutant
Effective Cleaning Rates
Sorbent/HEPfl Filter Console
xitaJytic/HEPH Filter Console
MAX FLOW MIN FLOW
Device 5
H NC^ I SO
MflX FLOW MIN FLOW
Device 8
0 CO BHC
FIGURE 4 :Suniaru of Gaseous Pollutant
Reioval Efficiencies
latalytlc/HEPfl Filter Console Sorbent/HERR Filter Console
cn
MRX FLOW MIN FLOW
Device 5
SNO. ISO
MRX FLOW HIN FLOW
Device 6
0 CO 8 HC
90
-------
EVALUATION OF ORGANIC EMISSIONS TO THE INDOOR
ENVIRONMENT VIA SMALL CHAMBER TESTING
4'
Bruce A. T1chenor
Indoor A1r Branch
U. S. EPA, Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
John E. Bunch and Mark A. Mason
Acurex Corporation
Research Triangle Park, NC 27709
The contamination of the Indoor environment has recently emerged as
a major area of concern to health and environmental professionals.
Indoor air pollutants of Interest Include organic and inorganic vapors
and particles. As part of EPA's research in indoor air quality (IAQ), the
A1r and Energy Engineering Research Laboratory is conducting research to
determine the emissions of organic vapors from building materials and
consumer products commonly used 1n homes, offices, and public access
buildings. Samples of the material/product are placed In small environ-
mental test chambers, and measurements are conducted via gas chromato-
graphy to determine emissions rates for various organic species. The
temperature, humidity, and flow rate of the air through the test chambers
are carefully controlled. Data are presented on the organic emissions
from a typical consumer product: liquid floor wax.
91
-------
EVALUATION OF ORGANIC EMISSIONS TO THE INDOOR
ENVIRONMENT VIA SMALL CHAMBER TESTING
Introduction
The contamination of the indoor environment has recently emerged as
a major area . of concern to health and environmental professionals.
Indoor air pollutants of Interest Include organic and inorganic vapors
and particles. A number of Indoor air pollutants have been extensively
studied (e.g., formaldehyde from particleboard, radon from soil, and
numerous contaminants from tobacco smoke). Except for formaldehyde,
however, only limited Information 1s available on organic emissions from
the wide variety of building materials and consumer products found 1n
homes and offices. As part of EPA1s research in Indoor air quality
(IAQ), the A1r and Energy Engineering Research Laboratory Is conducting
research to determine the emissions of organic vapors from building
materials and consumer products commonly used 1n homes, offices, and
public access bufldlngs.
Samples of the material/product are placed In small environmental
test chambers, and measurements are conducted via gas chromatography to
determine em1ssions rates for various organic species. The temperature,
humidity, and flow rate of the air through the test chambers are carefully
controlled. To date, tests have been conducted on the following products/
materials: caulking compound, floor adhesive, partlcleboard. moth repel-
lant, liquid floor wax, and wood stain. Other papers*>2»3»4 have pre-
sented data on several of these materials; the present paper focuses on
the liquid floor wax tests.
Experimental Facility
EPA's Indoor Air Source Characterization Laboratory consists of the
following components: a clean air conditioning and delivery system, an
incubator containing two 166 liter environmental test chambers, sampling
manifolds, and sample collection adsorbers using Tenax and charcoal, a
permeation system for quality control 1s Included. The environmental
variables are monitored and controlled by a microcomputer. Organic
analyses are conducted by thermal desorptlon, concentration using purge
and trap, and gas chromatography (GC) using flame Ionization detectors
(FID). A separate microcomputer provides GC data analysis. All data are
Input to spreadsheets for further analysis.
Experimental Procedures
Headspace Analysis
Prior to evaluation 1n the environmental test chambers, materials
are evaluated by headspace analysis. A sample of material 1s placed 1n a
1 liter Teflon-Hned container which Is purged with nitrogen at 50 ml/m1n
for 30 minutes. The exit flow Is collected on Tenax and analyzed after
thermal desorptlon. Gas chromatography/mass spectroscopy (GC/MS) analysis
provides identification of the compounds contained in the material's
emissions. Approximately 10 of the compounds are selected for measurement
1n subsequent chamber tests.
92
-------
Chamber Testing
The experimental design was developed to provide Insight Into the
effect of air exchange rate, loading, and temperature on the emission
characteristics of the liquid acrylic floor wax (AFW). Table I (Test
Matrix) provides the experimental conditions for the nine tests that were
conducted. All tests were at SOS relative humidity, except test AFW8.1
which was at 202. The floor wax was spread on aluminum plates using a
sponge applicator. The weight of wax applied was determined by weighing
the sponge before and after application. The plates were coated in the
chamber, and the test start (time * 0) was established when the door to
the chamber was closed. The first sample was collected at the 15 minute
mark.
Experimental Results
Organic Compounds Measured
The following compounds were identified by headspace analysis and
were measured during the chamber testing: C-9 substituted cycloalkane,
trlmethyl cyclohexane, methyl octane, C-9 substituted alkane, methylethyl
cyclohexane (2 isomers), nonane, dimethyl octane, ethyl toluene, trlme-
thyl benzene, decane, and undecane. Total organlcs (as decane) were also
measured.
Chamber Concentrations
Figure 1 shows the concentration of total organlcs, nonane, and
ethyl toluene vs. time for test AFW2.1, The symbols represent measured
values; the lines represent the "best fit" of a model5 developed to
analyze the data. This plot 1s typical of all the tests, where the
initial concentration decreases several orders of magnitude (from >100,000
ug/mJ to near the analytical detection limit) over the test duration.
Figure 2 shows how the composition of the emissions changes over time.
Five compounds are plotted vs. time as a percent of total organlcs. Note
that undecane Increases as a percent of the total, Indicating that Its
volatility 1s generally less than that of the other compounds. Methyl
octane and nonane show decreases, providing evidence of higher volatility.
The percentages for decane and ethyl toluene remain relatively constant.
Emission Rates
Figure 3 shows the emission factors for total organlcs and four indi-
vidual compounds for test AFW2.1. Mote that the emission rates decrease
several orders of magnitude over the test duration. Also, the rate of
change of the emission factors varies between individual compounds. This
reflects the differences 1n volatility discussed above.
Effects of Test Variables
The effect of air exchange rate (N} and loading (L) on emission
rates of total organlcs 1s shown In Figure 4. At the beginning of the
test, the higher the N/L value, the higher the emission rate (ER) (i.e.,
for N/L - 16.1, ER - 10,750 ug/cm2-hr; N/L « 8.2, ER - 3620; and N/L «
4.0, ER ¦ 2470). This occurs, In part, because the high air exchange
rates cause reduced chamber concentrations due to dilution which 1n turn
Increases the vapor pressure driving force. After about 1.5 hours, the
emission rates are highest for the low N/L values due to the more rapid
depletion of the source at high air exchange rates.
93
-------
The effect of temperature on emissions was not determined. Only two
tests (AFW8.1 and AFW9) were conducted at high temperature. Visual exami-
nation of the concentrations for these two tests compared to tests AFW2.1
and AFW7 Indicates somewhat higher values for the higher temperature.
Further data analysis is required, ticwewer, before the effect of tempera-
ture can be described.
Quality Control
An internal standard (hexane) 1s used to evaluate the precision of
the organic measurements. For the nine tests, the average percent reco-
very of the Internal standard was 102.2 ±16.9 for 109 samples. The
recovery values for each test were used to calculate the concentrations
of the measured compounds.
Conclusi ons
Emissions from a liquid floor wax cause Initial chamber concentra-
tions of total organlcs >100,000 ug/m3 which decrease to near the analyti-
cal limit over the test duration. Differences in the volatilities of
individual compounds cause the composition of the emissions to vary over
time. Emission factors vary from >1000 to <1 ug/cnt-hr over the first 10
hours. High air exchange rates cause high emission factors during the
first few minutes of the test; by the end of the test the emission factors
are highest for the low air exchange rates. The effect of temperature on
emissions awaits further analysis.
References
1. R. A. Tichenor, M. D. Jackson, R. 6. Merrill, Measurement of organic
emissions from Indoor materials -- small chamber studies, Proceedings
of the 1986 EPA/APCA Symposium on Measurement of Toxic Air Pollutants
EPA-600/9-86-013, Raleigh, NC, 1986, pp. 86-94.
2. B. A, Tichenor, M. A. Mason, Characterization of organic emissions
from selected materials In Indoor use, 79th Annual Meeting of the Air
Pollution Control Association, Minneapolis, MN, 1986, Paper No. 86-16.5.
3. B, A, Tichenor, M. A. Mason, Organic emissions from consumer products
to the Indoor environment, 80th Annual Meeting of the Air Pollution
Control Association, New York, NY, 1987, Paper No. 87-81.8.
4. B. A. Tichenor, Organic emfsslon measurements via small chamber testing,
IAQ'87, Berlin, 1987.
5. J. E. Dunn, B. A. Tichenor, Compensating for wall effects In IAQ chamber
tests by mathematical modeling, 80th Annual Meeting of the Air Pollution
Control Association, New York, NY, 1987, Paper No. 87-83,4.
94
-------
TABLE I. TEST MATRIX - LIQUID FLOOR WAX
Test No.
Temperature
(°C)
A1r Exchange
Rate (hr"M
Sample Area
(c«2)
N/l(®) Duration
(hr)
AFH1
20.9
1.94
200
16.1
70.3
AFW2.1
22.0
0.47
200
3.9
47.2
AFW3
20.0
-------
Total Organics
100,000
Nonane
Ethyl Toluene
1000
100
0.01
Time (hours)
Figure 1. Chamber Concentration vs. Time
Floor Wax - Test AFW2.1
14
| Methyl Octane
12 h
n
.y 10
LW^-
c
a
L.
o
a
J.*
o
8 h
A-v
c
• A
o 4
4J
a.
2 r
Norane
Ethyl Toluene
Undecane
Decane
B 12 16
Time (hours)
Figure 2. Emission Composition vs. Time
Floor Wax — Test AFW2.1
96
-------
10,000
Total Organics
Methyl Octane
1000
Nonane
100
Ethyl Toluene
V.
'X
10
0.01
0.1
1
Time (hours)
Figure 3. Emission Factors vs. Time
Floor Wax - Test AFW2.1
N/L «
16.1
10,000
a.2
or
100
o>
4.0
0
3
4
2
7
8
10
5
6
9
Time (hours)
Figure 4. Effect of N/L on Emission Factors
Floor Wax — Total Organics
97
-------
CHAMBER STUDIES CHARACTERIZING ORGANIC EMISSIONS FROM
KEROSENE SPACE HEATERS
James B. White
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Brian P. Leaderer and Patricia M. Boone
John B. Pierce Foundation Laboratory
Dept. of Epidemiology and Public Health
Yale University School of Medicine
New Haven, CT
S. Katharine Hammond
Dept. of Family and Community Medicine
University of Massachusetts Medical School
Worcester, MA
Judy L. Mumford
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Abstract
This paper reports the results from Phase I of a three phase study to
characterize and model particulate and organic emissions from unvented
kerosene space heaters. Phases I and II Involved test chamber evaluations
of the source, while Phase III will use a test house to validate the
chamber model. Phase I emissions from 12 heaters, covering a range of
design types and Btu ratings, were screened during start-up and three
steady-state operating conditions (normal, low, and high flame settings) to
select candidate heater types for an intensive Phase II evaluation. Aero-
sol mass, size distribution, extractable mass, mutagenicity, and con-
ventional gas (CO, N0X, etc.) measurements were made for all the heaters
tested. Volatile and semlvolatlle organlcs were measured on a subset of
the heaters.
Introduction
Previous chamber and field studies Investigating the nature of air
contaminants from unvented kerosene space heaters and resulting human
exposures, (Ref. 1-3), have focused on the emissions of CO, CO2, N0X, and
SOg. Little has been published on the chemical, biological, and physical
nature of the volatile and semlvolatlle organlcs and particles emitted by
unvented kerosene space heaters. Phases I and II of this cooperative
agreement concentrated on the characterization of kerosene space heater
emissions 1n an environmental test chamber. Phase III will validate the
model derived from the chamber 1n a test house.
Phase I screened 12 commerlcal heaters currently In-use to Identify
4-6 heaters and use conditions for inclusion 1n a Phase II evaluation. The
evaluation criteria were set in Phase I to select at least one heater of
98
-------
each type that produced the most particulate mass and/or mutagenic activity.
In addition, Phase I determined emission factors of the classical air con-
taminants (CO, CO2, N0X, etc.) for a number of kerosene heaters under a
range of fuel consumption rates.
Phase II of the project will provide a detailed characterization of
particle and organic emissions (extractable particle masses, semlvolatlles,
volatlles, bloassays of all extracts, trace elements, add aerosols, ele-
mental/organic carbon, etc.). The results of Phases I and II will be used
to design and conduct the EPA test house studies in North Carolina which
will verify the model developed in the chamber studies. This paper will
describe Phase I and present the results that are available to date.
Methods
The experiments were conducted 1n a fully controlled and well mixed
34 m^ environmental chamber (Ref. 2). The temperature in the chamber
during all of the heater runs was kept below 30OC. Cooling was ac-
complished by passing the recirculated air In the chamber over a cooling
coll. The temperature of the coolant was maintained at approximately 22°C,
well above the dew point of the air fn the chamber, to prevent conden-
sation of air contaminants. The fresh air exchange rate, measured for each
experiment, was typically about 1.2 air changes per hour (ach). Complete
mixing was maintained with a recirculation rate of 20 ach.
Twelve different kerosene heaters were tested: five radiant (R), three
convective (C), three combination radlant/convectlve (C/R), and one double
radiant (R/R). Two heaters were tested twice. The manufacturer-rated
heater outputs ranged from 8,200 to 17,500 Btu/hr (8,650-18,460 kJ/hr).
Most of the heaters were new. Two fuels were tested: one K-l kerosene
(sulfur content of 0.06X and ash content of 0.0022) was used for the In-
itial heater tests and one duplicate run; and the second K-l kerosene
(sulfur content 0.021 and ash content 0.016J) was used for the other
duplicate run.
Each kerosene heater was placed 1n the chamber on a Potter scale at
least a 1/2 hour before the experiment was started, and the ventilation and
recirculation rate of the chambers was set. During the Initial 1/2 hour
before the heater was lit, background levels were obtained of all continu-
ously measured parameters and particle size distributions.
Each kerosene heater was fired up after the first 1/2 hour and run
continuously for approximately the next 5 hours. At the end of 5-1/2 hours
the heater was shut off, and the decay of air contaminants was measured.
During these 5 hours, each heater was operated 1n three modes: normal, lew,
and high flame settings. The first operating mode (normal flame setting)
lasted from heater start-up to 2-1/2 hours into the experiment. Once
steady-state conditions were reached 1n the chamber, the heater flame
setting was adjusted to low, and a new steady-state was reached after 1-1/2
hours. After steady-state was reached at the low setting, the heater flame
setting was readjusted to high, with a new steady-state level achieved
after an additional 1-1/2 hours. The heater was then shut off, and the
decay of contaminants was traced for 1 hour without any adjustment of the
chamber test conditions. Air contaminant gas and particle deposition rates
(removal by the chamber surfaces) were detemlned by comparing the decay
rates of the gases and particles to that of COj.
99
-------
The heater flame settings were determined by visual Inspection uti-
lizing the picture guides contained In the operator's manuals which ac-
company the heaters. Actual fuel consumption rates were recorded with the
Potter scale throughout heater operation. Steady-state chamber conditions
were defined as CO2 levels change at a rate of O.Olt per 5 minute period.
Since CO2 is non-reactive all other gases and particles, whether reactive
or non-reactive, will have reached steady-state by the time CO2 has.
Continuously monitored gases (N0X, SO2, CO, CO2, and non-methane HC),
temperature, dew point, and condensation nuclei (CNC) were measured and
recorded during the full 6-1/2 hours of each experiment. Particle size
measurements (Electrical Aerosol Analyzer--EAA—and Optical Particle
Counter-OPC) were recorded for 10 minute periods before the heater was
turned on, for 30 minutes during heater start-up and shutdown, and for 10
minutes during steady-state conditions.
Integrated total mass samples for trace elemental analysis (XRF) were
collected on up to seven personal pumps spread through the chamber. Total
mass samples for organic chemical analysis and mutagenicity testing were
collected for each experiment using a medium volume sampler (i.e., PM-10
sampler without the sizing head) and pre-extracted filters. The filters
were refrigerated and shipped to the University of Massachusetts for
weighing and extraction. The filters were extracted with dlchloromethane
and solventexchanged to dimethyl sulfoxide. The extracts were then send to
EPA's Health Effects Research Laboratory, RTP, NC, to bloassay for muta-
genicity using a modified Kado micro-suspension assay (Ref. 4). This
modification uses Salmonella typhimurfum TA98 both with and without S9
activation.
In addition, three heater runs doubled as trial runs to develop the
protocol for Phase II. A XAD sample for semlvolatlle analysis and muta-
genicity testing (attached behind the PM-10 sampler) was collected for
these heaters as was a VOC grab sample using SUMMA canisters. Only the
preliminary results on the total mass, size distribution of the particles,
and particulate mutagenicity will be presented and discussed 1n this paper.
Results and Discussions
Table 1 summarizes the particle mass and volume data for each heater
run. The particle volume data Indicate that the highest mass emissions for
radiant heaters generally occur between heater start-up and steady-state at
normal flame setting (buildup). The convectlve heaters tested did not
generally demonstrate a pronounced peak during heater start-up. Steady-
state conditions associated with normal-flame-setting radiant heaters emit-
ted more particles (higher particle volume concentrations) than during
low or high flame settings. Convectlve heaters typically produced the
highest particle volume concentrations during the high flame setting. The
lowest particle volume concentration for all heater types was generally
associated with a low flame setting. Particle size distribution measure-
ments indicate that the particles produced by all the kerosene heaters are
less than 2.0 um with the peak volume occurring in the 0.15 um size range.
Total mass concentrations as measured by the personal pumps exhibited
considerable variability, with a low level of 73 ug/m-3 and a high of 721
ug/m3. No clear pattern is evident for the total mass concentrations In
relation to the type of heater or Btu rating of the heater. One heater
(R/R) produced particle mass levels substantially higher than the others
and with large variability between runs. Since only one heater of this
design was tested, it is not known whether the high levels are particular
100
-------
to this heater or to the general category of heaters with this design.
There was little variability In the particle mass levels associated with
the old and new heater comparison runs (C/R), apparently Indicating little
or no effect by the age of the heater. Particle mass levels for the same
heater with different fuels (repeat radiant heater run) were higher for the
higher ash content fuel.
The emission rates were estimated from the total mass concentrations
as measured by the personal pumps over the full operation period of each
heater, the effective removal rate of particles (ventilation plus loss to
surfaces), and the total fuel consumed during each experiment. The esti-
mated emission rates follow the total mass concentrations. The particle
mass emission rates do not demonstrate any trends associated with heater
type or manufacturer's Btu rating. Excluding the R/R heater, the emission
rates vary by a factor of four. Using this range of emission rates for
a 1,500 ft? (139 m^) house with a 7 ft (2.1 m) celling and an effective
particle removal rate of 1 ach, the estimated range of steady-state
particle mass levels (particles <2.5 um) would be from 12 to 50 ug/m3.
Levels would be considerably higher 1f the emission rate for the R/R
heater were used.
Table 2 summarizes the particulate mass collected by the medium volume
sampler and the mutagenic data derived from the filter extracts. Because
replicates were not available, only general observations can be made. As
a whole, the radiant heaters appear to produce a less mutagenic extract
than the other heaters; the combination (I.e., R/R, C/R) heaters appear to
produce more mutagenic activity than the radiant or convective heaters.
Based on these results four heaters were selected for Phase II analy-
sis. (See Table 2.)
Conclusion
Preliminary results of this research project Indicated that unvented
kerosene space heaters are a significant source of respfrable suspended
particles Indoors. There 1s considerable variability In the emissions that
is not associated with heater type or manufacturer's Btu rating. In addi-
tion certain heater types appear to produce mutagenic particles. The
complete chemical analysis of the particles (the Identity and quantity of
volatile and semtvolatlle organlcs emitted by the kerosene heaters) will be
available soon.
101
-------
References
1. Leaderer, B.P., A1r pollutant emissions from kerosene space heaters
Science 218 (1982), 1113-1115.
2. Leaderer, B.P., et al., Assessment of exposure to indoor air contami-
nants from combustion sources: methodology and application. Am. 0
Epid. 124 (1986), 275-289.
3. Traynor, G.W., et al., Pollutant emissions from portable kerosene-
fired space heaters. Envir. Sci. Tecbnol. 17 (1973), 327-332.
4. Kado, N.Y., et al., A simple modification of the Salmonella liquid in-
cubation assay. Increased sensitivity for detecting mutagens in human
urine. Mutation Research 121 (1983), 25-32.
The objective of this cooperative agreement between the U.S. EPA and the
J. B. Pierce Foundation is to identify and develop emission rates for
important organic air contaminants emitted by unvented combustion sources.
A large scale environmental test chamber at the Pierce Foundation was used
in this work to assess the nature and emission rates of air contaminants
from 12 unvented kerosene space heaters, including radiant and convectlve
heaters. Emphasis was placed on measuring the organic emissions from these
sources as well as determining the mutagenic potential of the emfssions.
Classical air contaminants were also measured. The sources were screened
over a range of conditions (i.e., source types, source ages, and fuel type
and consumption) typically encountered In the real world to identify the
range of individual (or categories of) emissions. The screening runs were
then used to target a subset of heaters for an Intensive evaluation of emis-
sions. The data from these studies are presented and will be Incorporated
into indoor air models for evaluation in EPA's test house.
102
-------
TABLE 1: Particle Emissions from Kerosene Heaters Tested
KEROS.
HEATER
TYPE'
RATED
OUTPUT
Btu/hrb
(kJ/hr)
TOTAL MASSC
(ug/m3)
PARTICLE VOLUMEd
(um3/cm3)
Personal
pumps
Buildup
Normal
Low [High
1
ESTIMATED
EMISSION
RATEe
(ug/g)
R
9,300 ( 9,810)
130
+/- 13
222
47
33
38
62
R
9,300 ( 9,810)
104
+/- 9
168
50
67
58
44
R
10,500 (11,076)
201
+/- 16
745
102
71
100
80
Rf
12,500 £13,186)
73
+/- 14
162
109
56
102
25
Rf
12,500 (13,186)
166
+/- 23
81
72
70
40
44
R
12,500 (13,186)
151
+•/- 12
240
136
70
122
56
R/R9 12,600 (13,291) 533 +/- 23 512 342 211 299 175
R/R9 12.600 (13.291) 721 ~/- 31 812 812 643 1097 260
C
8,200 ( 8,650)
79
+/- 7
45
45
18
52
20
C
12,700 (13,397)
148
+/- 28
137
137
147
214
47
C
17.100 (16,038)
188
+/- 19
—
45
48
46
59
C/R*
9,400 ( 9,916)
91
+/- 7
223
74
28
89
54
C/Rh
9,400 ( 9,916)
98
+/- 15
239
50
26
53
53
C/R
17,500 (18,460)
132
+/- 50
100
100
66
138
27
a R (radiant), R/R (double radiant), C (convectlve) and C/R (combination).
b Btu output as advertised by manufacturer,
c Total mass was collected by from 4 to 7 personal pimps spread throughout
the chamber and a PM-10 sampler without the head.
d Volume is calculated from the EAA particle size data (0.01 - 1.0 urn).
The buildup particle volume 1s the largest volume calculated (2 minute
size distribution measurement) during the period of heater start-up to
normal steady-state. The normal, low, and high volumes are the average of
fifteen 2-m1nute size distribution measurements taken during steady-state
air contaminant concentrations associated with normal, low, and high fuel
burning conditions.
e Calculated from personal pump mass, fuel consumption, and effective
removal.
f Same heater with different fuels (one 0.061 S, 0.0021 ash; and two 0,021
S, 0.0161 ash).
S Same heater, duplicate runs.
h Same heater model (first one 1s new and other 1s old).
103
-------
EPA'S INDOOR AIR QUALITY TEST HOUSE
1. BASELINE STUDIES
Merrill D. Jackson
U.S. Environmental Protection Agency
A1r and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Russell K. Clayton, E. Eugene Stephenson, William T.
Guyton, and John E. Bunch
Acurex Corporation
P.O. Box 13109
Research Triangle Park, NC 27709
EPA has acquired a test house 1n the Research Triangle Park, NC, area
for Indoor air quality research. The test house will be used to measure
the outgassfng of chemicals from common household and building materials,
to determine the effects of residential activities which release organic
emissions, and to evaluate Indoor air quality control technologies. The
results of testing 1n EPA's laboratory size environmental test chambers
will be compared to those obtained under normal household conditions.
The test house 1s a typical three-bedroom, two-bath, one-story frame
house with natural gas heat.
This paper will describe the baseline studies which provide the
required characterization of the test house. These studies Include blower
door and SF6 tests for air leakage and exchange rates; pollutant stratifi-
cation, migration, and surface reactivity evaluations; and analysis of the
baseline Inorganic gases and total organic hydrocarbon levels 1n the house
before any test materials are allowed into the house.
104
-------
INTRODUCTION
Ttie Environmental Protection Agency has an Indoor air quality program
In which common household and building materials are tested fn small and
large chambers for either outgasslng or emitting volatile organic com-
pounds. Examples are the outgasslng of caulk and the emissions from
unvented kerosene heaters. The chambers range from 50-1 Iter laboratory
units to room size chambers. Once the emission concentration values have
been determined in the chamber experiments and models have been developed
for full scale applications, the model must be evaluated 1n a house under
normal usage.
The A1r and Energy Engineering Research Laboratory of EPA leased a
house for the purpose of evaluating these models and Indoor afr quality
control technologies, as well as providing interlaboratory support such
as bloassay sample collection for EPA's Health Effects Research Laboratory
and the evaluation of samplers and methods for Indoor air quality studies
for EPA's Environmental Monitoring Systems Laboratory.
The house meets the requirements for our program Including:
1. Wood Construction
2. One-Story
3. No Basement
4. Single Family
5. Gas Forced A1r Heat
6. Three Bedroom, Two Bath
7. Five to Seven Years Old
8. 1200-1500 Square Feet (100-150 Square Meters)
A multl-storied house was not desired at this time due to modeling con-
straints, and wooden construction was required as the most common type 1n
the nation. The age requirement was based on the need to select a house
which had already outgassed the volatile and semi-volatile compounds from
the original materials of construction, and which was built with the
energy savings features brought on by the oil energy crfsls.
He have the authority to make any changes to the house needed for
the progam such as replacing the subflooHng, carpeting, or cabinets;
adding paneling; or painting or staining the Interior or exterior. An at-
tached garage, running the full width of the house, has been outfitted as
an instrumentation room which 1s attached, yet isolated from the test areas.
DESCRIPTION
Sampling and analylcal instrumentlon for the following Is available
or installed at the test house:
1. Organic Compounds
a. Volatlles
b. Semi-volatlles
c. Total Hydrocarbons
105
-------
2. Inorganic Compounds
a. Sulfur Oxides
b. Carbon Koncxlde
c. Carbon Dioxide
d. Nitrogen Oxides
e. Nitrogen Oxide
f. Ozone
3. Particulate
a. Filters
4. Air Exchange Rates
a. Blower Door
b. Sulfur Hexafluoride
5. Heteorological
a, Temperature
b, Wind Direction
c, Wind Speed
d, Relative Humidity
e, Barometric Pressure
The volatile organics are sampled with Tenax cartridges at flow rates
of 250-1000 ml/min. Semi-volatiles ar'e collected on XAD-2 resin traps using
a modified medium flow (4 cfw or 0.0019 m3/sec) sampler. Inorganics and
total hydrocarbons are measured by continuous monitors, for which a di-
ution system is available for calibration. This all ones a standard curve of
any reasonable range to be run on each instrument and does not require the
storage of numerous cylinders of low concentration gases which have a short
shelf life. Clean air is used as the diluent. Particulate samples are col-
lected using quartz (mass) and Telfon-coated glass fiber (bioassay} filters
using 4-cfm samplers. The blower door method uses a commercially available
unit with standard operating procedures (1), while the SFg method requires
that grab samples be taken hourly after the release of a specified volume
(10-20 ml) of SFg tracer gas. The Tedlar bag samples are analyzed by gas
chromatography with flame ionization detection to determine the rate of
loss 1rr the gas concentration. A meteorological tower 1n the backyard has
a readout fir tte garage.
An IBM-AT computer controls all the sampling times and records the
readings from all the continuous monitors arid the weather data. The control
system can handle up to 10 locations within the house. These locations can
be moved easily. At present eight are inside the house and two are outside
for ambient measurements during experimental runs.
The first products to be tested in the house are unvented kerosene
heaters. In preparation for these tests, certain characterization tests were
made, Including blower door, SFg, and continuous monitor runs both Inside and
outside the house. In the two rooms where the emissions from the kerosene
heaters will be monitored (the den where the heater will be located and the
bedroom farthest from the den), stratification, migration, and reactivity
tests were also performed. The rooms were sampled at elevations of 6, 36,
64, and 94 1n. (0.15, 0.91, l.fi3 and 2.39 m) above the floor. These heights
represent a person lying on the floor, sitting 1n a chair, standing 1n the
room, and the celling level, respectively.
106
-------
The blower door test was run normally with all the doors and windows
closed and then with one window 1n the den open 2 in. (0.05 m). The manu-
facturers of unvented kerosene heaters recommend that one window be opened
slightly during the heater operation.
RESULTS
The results of the blower door tests are in Table 1.
TABLE 1. BASELINE BLOWER DOOR STUDIES
CONDITION ACH ACH EqLA EfLA
50 Pa Norm In.2 m2 In.2 m2
NORMAL 15.28 0.90 256 0.165 133 0.086
WINDOW OPEN 17.36 1.02 319 0.206 176 0.114
ACH 3 Air Changes per Hour
EqLA * Equivalent Leakage Area
EfLA a Effective Leakage Area
The blower door tests showed that the normal ambient air exchange rate
[approximated by dividing the 50 Pa ACH by 17 (1)] for the test house Is In
the generally accepted range for energy efficient houses of Its age. The
equivalent leakage area Is defined as the area of a round sharp-edged
orifice that, at a pressure of 10 Pa, would leak the same as all the house's
leaks at 10 Pa. Effective leakage rate Is defined as the area of a
specially shaped bell-mouthed nozzle that, at a pressure of 4 Pa, would leak
the same as all the house's leaks at 4 Pa.
The results of the stratification test are given In Table 2
There Is no apparent difference between the outside and the inside air
nor Is there any stratification In the two rooms of the house.
The results of the migration and reactivity tests will be Included 1n
a later publication along with the kerosene heater test data.
Organic background tests will be run during the unvented kerosene
heater study as well as further tests on sink effects of the walls,
carpet, and other materials in the house at the time of the tests.
Materials planned for study 1n the future Include moth cakes (para-
dlchlorobenzene), floor wax, wood and wall finishes, and paneling.
107
-------
TABLE 2. STRATIFICATION AND BASELINE CONTINUOUS MONITORING RESULTS
LOCATION
(in.)
CO
C02
NOX
NO
THC
ppm
ppm
PPm
ppm
ppm
OUTSIDE
1.14
398
0.08
0.07
0.09
DEN (6)
1.Z1
422
0.06
0.06
0.10
DEN (36)
1.24
436
0.06
0.06
0.10
DEN (64)
1.23
438
0.06
0.06
0.10
DEN (94)
1.24
439
0.06
0.06
0.10
AVERAGE
1.23
434
0.06
0.06
0.10
BEDROOM
(6)
1.22
455
0.06
0.06
0.09
BEDROOM
(36)
1.20
442
0.06
0.06
0.09
BEDROOM
(64)
1.11
393
0.06
0.06
0.09
BEDROOM
(94)
1.11
393
0.06
0.06
0.09
AVERAGE
1.16
421
0.06
0.06
0.09
SOX = <0.01 ppm for all locations
REFERENCE
1. Operation Manual, Minneapolis Blower Door, 920 West 53th Street,
Minneapolis, MN 55419.
108
-------
SEMI-VOLATILE ORGANIC COMPOUNDS IN RESIDENCES
David J. Anderson and Ronald A. Hites
School of Public and Environmental Affairs
and Department of Chemistry
Indiana University
Bloomington, Indiana 47405
This research focuses on the identification and quantification of
semi-volatile organic compounds (vapor pressures 1 - 10"' mm Hg) in the
Indoor air of single-family dwellings. This general category includes
chlorinated pesticides, polychlorinated biphenyls (PCBs), and polycyclic
aromatic hydrocarbons (PAH). Samples were taken using polyurethane foam
(PUF) as a sorbent with small, low-flow pumps. Extracts were analyzed by
gas chromatographic mass spectrometry (GC/MS) and gas chromatography with
electron capture detection (GC-ECD). Indoor concentrations of these
compounds were elevated when compared to typical outdoor concentrations.
Derivatives of hexachlorocyclopentadiene, such as alpha- and gamma-
chlordane, were detected in the living area of one home at levels up to
60 times the outdoor concentration. The concentration of qamma-chlordane
in the basement of this home was three orders of magnitude higher than
outdoor levels. Indoor sources for these chemicals are Implied.
109
-------
Introduction
Because the average person spends about 90X of his or her time in-
doors, it is important to know the identities and concentrations of in-
door air pollutants, ft chronological summary of major U. S. indoor air
research reveals the neglect of organic pollutants in early indoor air
research (1). There has been some research on volatile organic com-
pounds, buF there are few data on semi-volatile organic compounds (2-4),
Our research involves the identification and quantification of chlo-
rinated pesticides, polychlorinated biphenyls (PCBs), and polycyclic
aromatic hydrocarbons (PAH) in the indoor air of single-family homes.
Indoor concentrations of these compounds were measured in 12 homes and
compared to typical outdoor concentrations. Possible sources of these
chemicals are proposed for some of the compounds.
Experimental Section
Homeowners were solicited by a news release which appeared in the
local newspaper; questionnaires were used to select a set of homes for
sampling. The questionnaires requested general information about the
size and age of the home and about possible sources of organic chemicals.
The majority of the homes were locsted in the Bloomington, Indiana area
and were sampled during May-October, 1986.
Air samples were taken using polyurethane foam (PUF) traps and
constant flow pumps. The PUF was cut into small plugs which were 22 nm
in diameter and 8 cm in length. Two Dupont P-4000A constant flow pumps
(E. I. du Pont de Nemours S Co.) and two SKC Aircheck VII pumps (SKC
Inc.J were used for sampling. The pumps were calibrated with a large
volume soap bubble flowmeter. A nominal flow rate of 3 L/min and sample
volumes of approximately 2 mJ were used. Both brands of pump contain
servo mechanisms to maintain the flow rate at + 5% of the set value dur-
ing sampling. All flow rates in this work were within this tolerance.
The PUF plugs were placed in glass sampling tubes under slight compres-
sion to prevent air seepage around the edges. Sampling was carried out
in duplicate by means of a glass tee connected to the inlet of the pump.
The sampler design was based on the one described by Lewis and MacLeod
(!)•
The PUF plugs were pre-extracted prior to use with glass-distilled
grade solvents. A 24 hour Soxhlet extraction with acetone followed by 24
hours with hexane was used. The pre-extracted plugs were dried under
vacuum at approximately 50° C for 24 hours. The PUF plugs were carried
through the entire pre-extraction and drying steps in groups of three;
this provided two plugs for replicate samples and one blank in each
group.
Samples and blanks were stored at -10° C prior to Soxhlet extraction
with a 1:1 mixture of acetone and hexane. The internal standards for the
PCB analyses (4,4*-dibromobiphenyl) and the PAH analyses (di0-fluorene,
d.g-phenanthrene, and d^Q-pyrene) were spiked directly onto trie PUF plugs
prior to this extraction. The internal standard for the chlorinated
pesticide analyses (2,2',3,4,4',5,6,6'-octachlorobiphenyl) was spiked
into the sample extracts after sample extraction and concentration. The
extracts were concentrated by rotary evaporation and gentle blowdown
under a stream of clean, dry nitrogen.
All pesticide analyses were carried out on a Hewlett-Packard 5985B
gas chromatographic mass spectrometer (GC/MS) using electron capture.
110
-------
negative ion, mass spectrometry (ECNIMS) and operating at an ion source
temperature of 100° C and a pressure of 0.4 torr; 30 m x 0,25 mm DB-5
fused silica capillary columns (J&W Scientific) were used for all separa-
tions. Selected ion monitoring (SIM) was used for quantification, and a
confirmation ion for each compound was also monitored. The confirmation
ions were selected from the same isotopic cluster as the quantification
ions.
PCBs were measured on a Hewlett-Packard 5890A gas chromatograph and
were detected using a Ni electron capture detector (ECO) and a Hewlett-
Packard 3392A integrator. A Hewlett-Packard 7673A automatic sampler was
used for the introduction of all samples and standards. The PCBs were
quantified as Aroclor 1242. The total area of the 16 largest peaks in
the Aroclor 1242 mixture was used. All values were blank corrected.
PAH analyses were done on a Hewlett-Packard 5995A GC/MS system oper-
ated in the SIM mode using electron Impact ionization. The SIM program
used the most abundant ion produced for each PAH (in all cases the mole-
cular ion) as the quantification ion. The identification of isomeric PAH
was verified by matching chromatographic retention times to a standard.
Results
The concentrations of gamma-chlordane, alpha-chlordane, and trans-
nonachlor in the homes are given in Figure 1. kT'\ data are averages of
two replicate samples and are blank corrected. The outdoor concentra-
tions of these compounds are low. The outdoor samples were taken at two
locations in Bloomington. Note that two homes (12 and 63) have indoor
concentrations which are elevated with respect to the outdoor samples and
that the three compounds co-vary with one another; the correlation co-
efficient is at least 0.98 for the three compounds. The most likely
reason for this co-variation is that they have the same ultimate source.
These compounds are all present in technical-grade chlordane, which is a
termiticide.
The source of chlordane in home #63 was investigated using multiple,
simultaneous samples taken throughout the home. The concentrations in
the basement averaged 12 times higher than in the living areas. This
home was treated with chlordane in the late 1970's, according to the
current resident. It seems likely that chlordane has infiltrated through
cracks 1n the basement walls, if sub-surface Injection is assumed to have
been used for application. Large cracks in the basement walls were found
upon inspection of the home. A similar hypothesis is proposed for home
#12 based on Interviews with the homeowner; home #12 was treated for
termites 1n 1974.
The concentrations of heptachlor in the homes are shown in Figure 2.
Heptachlor is also found 1n technical-grade chlordane, but it does not
co-vary with the compounds discussed above (r<0.4). Also, the concentra-
tions of heptachlor are about a factor of 10 greater than the chlordane.
Homes 61 and 62 were treated for termites in 1977 and 1984 respectively.
The exact formulation of the term1t1c1des used 1s uncertain, but there is
a possibility that heptachlor was used. A commercial mixture of chlor-
dane and heptachlor, known as Termide, is available (4). Heptachlor has
a higher vapor pressure than the chlordanes, thus, TC 1s possible that
higher soil outgassing rates have contributed to the generally higher
heptachlor concentrations.
Ill
-------
Dursban (also known as chlorpyrifos) is an organophosphorous pesti-
cide; it is used as a soil insecticide and for the control of household
pests. Owners of the two homes which showed the highest levels of
dursban both indicated the use of dursban on the questionnaire. The
treatment in home #60 was once every two weeks, by the homeowner, with a
product containing dursban. The spraying in home #62 was conducted on a
monthly basis by a professional service.
Aroclor 1242 is the only PCB mixture present in the indoor air of
the homes tested. The concentrations of Aroclor 1242 are presented in
Figure 3. The high indoor concentrations imply indoor sources. However,
some of the homes were built after the use of PCBs was banned in the mid-
1970's. Homes 34 and 52 were built in 1983 and 1985, respectively, and
should not contain indoor sources of PCBs. The presence of electrical
euipment has been suggested as a source of PCBs in indoor air {Tj, and
indeed electrical equipment may be a source in some of the homes in our
study, such as home #62 which was built in 1973. The building in which
our laboratory is located was completed in 1962 and is, no doubt, ser-
viced by electrical equipment which contains PCBs.
The concentration profiles for fluoranthene and pyrene do not reveal
any significant trends; attempts to correlate the PAH data with smoking,
or other combustion sources, failed. Home #62 was sampled on a weekend
when three of the occupants who smoked were home, and the wood-burning
fireplace was in use. There appear to be no elevated levels of PAH in
this home, based on the results from this one set of samples. Similarly,
smoking was evident during the sampling of home #61, but smoking did not
produce indoor concentrations which were elevated over the indoor concen-
trations of PAH in non-smoker's homes. The elevated levels of PAH in our
laboratory are due to an input from the make-up air for our chemical fume
hoods. There is a coal-fired power plant near the building in which our
laboratory is located; the PAH are coming in through the air ducts from
the outdoor air.
Conclusions
The elevated indoor concentrations of these compounds are dependent
upon many factors, one of which is the presence of an indoor source.
Relative volatility of the various compounds may also be important in
determining indoor concentrations. Finally, the rate of exchange between
the indoor and outdoor air is an important variable which must be
measured to understand the role of indoor/outdoor mixing. Indoor concen-
trations of these compounds were not statistically higher in homes which
had the doors and windows closed during sampling when compared to homes
which had the doors and windows open during sampling. But, the quantifi-
cation of ventilation rates may allow the differentiation of changes in
indoor concentrations which occur doe to mixing from those changes which
are related to indoor sources.
References
1. J. E, Yocum, "Indoor-outdoor air quality relationships: a critical
review." J. Air Pollut. Control Assoc. 32: 500. (1982).
2. R. G. LewisTT. E. MacLeod, -'HortaDle sampler for pesticides and
semivolatile industrial organic chemicals in air," Anal. Chem. 54:
310. (1982).
3. K. E. MacLeod, "Polychlorinated biphenyls in indoor air," Environ.
Sci. Techno!. 15: 926. (1981 ).
112
-------
4. C. G. Wright, R. B. Leady, "Chlordane and heptachlor in the ambient
air of houses treated for termites," Bull. Environ. Contain. Toxicol
28: 617. (1982). '
Acknowledgment
We thank Ilora Basu for technical assistance and the U. S. Department
of Energy for support through grant 80-EV10449,
?j«*c-CHLQRDANE
ir H
I;
~niLni.u^n
o,>ic»CHLORDANE
L* o-j g-3 H rm *
s-NOfiA.CHLOR
,LE.1n n n
*!"l '} ' '¦* e \ 1 t'1
1 '"f 2 r * £,1 iJ.fi' ' 6.2 o. ? i. i1 X '•{ 113 c. i a.i
•.* •."> k V' V ^ (P f ^ s * .&
Figure 1. Concentrations of three components of technical-grade
chlordane in the indoor air of hones. (Vertical scales are different.)
Horizontal axis numbers are home Identification numbers. Outdoor samples
are designated "out. 1" and "out. 2". Laboratory concentration is
labeled "lab".
113
-------
(3
o
a
«
h
¦*_>
a
«
o
d
o
u
200
175
150
125
100
75
50
25
0
HEPTACHLOR
1DO
Z-
%
CI CI
68
30
3.7 2.2 5-5 B" 9
P-7-71 r-r-w S A \//A
I
190
1
01
w
4.1 8"7
Y//1
VA
i
i
62
0. 8 0. 2 3- 0
rr7~i
<*¦
\5 \B
0,\ ^ ^y- ^-,7-
<*>
Figure 2. Concentrations of heptachlor in the indoor air of homes.
Horizontal axis numbers are the same as in Figure 1.
800
700
600
500
400
300
200
100
Aroclor 1242
460
380
* go
160
1
I
140
VZi^k
170
1
740
1
310
I
& \u -v> ^ ^ <9-
330
^ # & \
1
%
-------
EMISSION RATES OF VOLATILE ORGANIC COMPOUNDS FROM
BUILDING MATERIALS AND SURFACE COATINGS
Lance A. Wallace
U.S. EPA
Washington, DC
Robert Jungers
Environmental Monitoring Systems
Laboratory
U.S. EPA
Research Triangle Park, North Carolina
Linda Sheldon
Research Triangle Institute
Research Triangle Park, North Carolina
The U.S. EPA measured emissions of 32 volatile organic compounds (VOCs)
from 31 building materials. All materials were collected from one building
for headspace studies; nine materials were investigated In follow-up chamber
studies. Twenty-four building materials emitted measurable levels of the
target VOCs. About 20 VOCs were elevated In the newly-constructed building
by factors of 5-500. Major sources included adhesives, paint, vinyl and
rubber molding, foam insulation, linoleum tile, carpet, and particle board.
lis
-------
Introduction
The U.S. EPA measured air concentrations of 32 VOCs in six buildings
(two offices, two homes for the elderly, one school, and one hospital) in
1984-861>2i3. As part of this study, emissions of all target VOCs from 31
building materials were measured.
Procedures
All emission testing was performed on materials used in the construction
of a new single-story office building in Fairfax, Virginia. Most of the 22
solid building materials tested were subsamples of the actual materials used
in construction and were collected at the sampling site in January 1985. The
remaining solid materials, plus the nine solvent-based materials, were
purchased from the manufacturer. Manufacturing lots were matched where
possible to materials actually used in the building.
Each of the solid materials collected at the building site were temporaril
stored (< 2 weeks) in individual plastic bags (Mobile KorditeR large capacity
1.5 mil) in nonlaboratory areas at Research Triangle Institute. For permanent
storage, the materials were individually wrapped in heavy duty aluminum foil
and placed in an outdoor metal storage shed to minimize the potential for
contamination from laboratory solvents and chemicals. Samples were stored
for a maximum of four months prior to testing. Solvent-based materials were
stored in their containers as received.
Preliminary headspace experiments were performed to determine both the
identities and approximate levels of volatile organic emissions from each of
the 31 building materials. For solid materials, one or two pieces (approximate
2x4 cm) were cut from the material. For solvent-based products, the material
was first mixed, then applied with a Teflon spatula to one side of a clean
glass microscope slide (5.5 x 2.6 cm). The prepared slides were allowed to
dry at 23°C for seven days protected from both dust and laboratory solvents.
During testing, the prepared sample material was placed in a wide-mouth
glass jar, which was sealed with a custonh-built Teflon head. Dry, purified
nitrogen was introduced to the jar at 15 mL/min and allowed to vent for 60 to
90 minutes. After this equilibration period, a Tenax GC cartridge was placed
into the sampling head. For solid materials, the headspace was purged through
the sorbent overnight at 15 mL/min. Solvent—based materials were tested for
approximately 30 minutes. Blank samples were collected from empty jars under
identical conditions to assess contamination.
Nine materials were selected for detailed emission studies using 12-L
chambers under controlled conditions of temperature, humidity, and ventilation.
A sample of material was placed in one of the four chambers and the chamber
sealed. The ventilation rate was set and the samples allowed to equilibrate
overnight at 25°C and 48X relative humidity. Purified air was supplied to
each chamber at a rate of 0.1 L/rain. From each chamber 0.033 L/min of this
air was removed by pumping through an empty sampling tube, while the other
0.067 L/min was vented outside the chamber. A fan blade attached to a magnetic
stir bar provided thorough mixing of the chamber atmosphere throughout the
experiment. After the equilibration period, Tenax GC cartridges replaced the
empty sampling tube to collect organic compounds from the gas stream. Chamber
blanks were prepared by leaving one of the chambers empty and collecting a
116
-------
sample of the air from the empty chamber for 6 hours in order to detect
background contamination.
Tenax GC cartridges exposed during the chamber experiments were loaded
with approximately 250 ng of bromofluorobenzene as an external standard, then
thermally desorbed and injected into a high resolution capillary column.
Samples were analyzed by either GC/MS or GC/FID.
Target chemicals were identified during GC/FID analysis by comparing
relative retention times for the samples and standards. For each material
tested three or four samples were collected with at least one exposed car-
tridge being analyzed by GC/MS. Comparison of these results to GC/FID results
assisted in positively identifying target chemicals. Only those chemicals
identified at high levels without chromatographic interferences were quanti-
tated by GC/FID.
Results
In the headspace experiments, 24 of the 32 target chemicals were emitted
from one or more of the 3i materials.
In general, aliphatic and aromatic compounds were emitted in larger
quantities than halogenated compounds (Table 1). Exceptions included 1,1,1-
trichloroethane from cove (molding) adhesive and tetrachloroethylene from
latex paint.
In the chamber experiments, the nine materials tested emitted 22 of the
target organlcs (Table 2). Chamber results for the six solid materials agreed
well with the headspace results, but were sharply different for the three
solvent-based coatings. Temperature, humidity, time of curing, and the
amounts used all differed between the experiments — apparently in sufficient
amounts to cause order-of-magnitude differences in calculated emission rates.
The building from which the materials were collected was monitored both
Indoors and outdoors within two weeks of its completion, before the occupants
moved in (Table 3).
The monitoring results often compared well with the results of both the
headspace and chamber experiments. For example, the aromatic hydrocarbons
with the highest indoor air concentrations (ethylbenzene, m-ethyltoluene, and
1,2,4-trimethylbenzene), generally show highest emission rates from all of the
building materials, including the carpet, linoleum tile, all of the plastic
materials and the solvent materials such as adheaives, paint, and caulk.
Particle board is probably a major source ofec-pinene. Finally, at the time
monitoring was being performed, vinyl core molding and adhesive were being
applied throughout the building. It is likely these materials are major
contributors to L,1,1-trichloroethane, ethylbenzene, and the xylenes found in
the air samples.
Conclusions
Previous studies by Molhave^ and Wallace^ have Identified and quantified
VOC emissions from building materials and finishings. Results are not fully
comparable because of differences in aging time and chamber conditions.
117
-------
However, it is possible to identify major sources of the 5-500-fold increases
in indoor air concentrations observed in new buildings. These include
adhesives and paints as likely major contributors of many VOCs, and foam
insulation, vinyl and rubber molding, particle board, and telephone cable
as major contributors of individual VOCs.
In view of the great differences observed between headspace and chamber
results for the paints and adhesives, and also of the likely variations in
composition of different brands, it does not seem possible at present to
quantify emission rates from these materials within an order of magnitude.
However, it is clear that elevated concentrations occur in new buildings (and
probably in renovated buildings) and nay cause symptoms such as those observed
in "sick building syndrome."
References
1. Sheldon, L.S. e_t. al, Monitoring In and Around Public Access Buildings,
Draft Final Report, Contract #65-02-4048, U.S. EPA, Research Triangle
Park, NC (1987).
2. Wallace, L., Junger3, R., Pelllzzari, E. , and Sheldon, L., "Volatile
Organic Chemicals in 10 Public-Access Buildings," presentation at the Ath
International Symposium on Indoor Air Quality, Berlin, August 17-21, 1987,
3. Junger8, R. and Sheldon, L., "Characterization of Volatile Organic
Chemicals in Public Access Buildings," presentation at the Ath Inter-
national Symposium on Indoor Air Quality, Berlin, August 17-21, 1987.
4. Molhave, L.B., "Indoor Air Pollution Due to Organic Gases and Vapors of
Solvents in Building Materials," Env. I tit. 8: 117-128 (1982).
5. Wallace, L., Pelllzzari, E., Leaderer, B., Zelon, H., and Sheldon, L.,
"Emission of Volatile Organic Compounds from Building Materials and
Consumer Products," Atmos. Environ. 21: 385-393 (1987).
118
-------
Table I. Summary of Emission Results
Emission Rate
- h)
Aliphatic and
Aromatic
Halogenated
All
Oxygenated
Hydrocarbons
Hydrocarbons
Target
Aliphati c
Compounds
Hydrocarbons
Cove adhesive
287
3800
650a
4737
Latex caulk.
252
380
5.2b
637
Latex paint
111
52
86c
249
Carpet adhesive
136
98
_d
234
Black rubber molding
24
78
0.88
103
Small diameter
telephone cable
33
26
1.4
60
Vinyl cove molding
31
14
0.62
46
Linoleum tile
6.0
35
4,0e
45
Large diameter
4.3f
telephone cable
14
20
38
Carpet
27
9.4
36
Vinyl edge molding
18
12
0.41
30
Particle board
27
1.1
0.14
28
Polystyrene foam
22
insulation
0.19
20
1.4
Tar paper
3.2
3.1
-
6.3
Primer/adhesive
3.6
2.5
-
6.1
Latex paint
(Bruning)
-
3.2
-
3.2
Water repellent
1.5
mineral board
1.1
0.43
-
Cement block
-
0.39
0.15
0.54
PVC pipe
-
0.53
-
0.53
Duct insulation
0.13
0.15
-
0.28
Treated metal
roofing
-
0,19
0.06
0.25
Urethane sealant
-
0.13
-
0.13
Fiberglass insulation
-
0.08
-
0.08
Exterior mineral
board
-
0.03
-
0.03
Interior mineral
board
-
-
-
-
Celling tile
-
-
-
-
Red clay brick
-
-
-
-
Plastic laminate
-
-
-
-
Plastic outlet cover
-
-
-
-
Joint compounds
-
-
-
-
Linoleum tile cement
"
al,1,1-Trichloroethane (minimum value)
bl,1,1-Trichloroethane (A.9) and Trlchloroethylene (0.3) (minimum values)
cTetrachloroethylene
dNo detectable emissions
eTrichloroethylene (3.6)
fTetraehloroethylene (3.8)
119
-------
Table 2. Calculated Emission Rates (ug/m^ - h) of Target Compounds During Chamber Experiments
Vinyl Black Poly- Linoleum Carpet Particle Cove Carpet Latex
Cove Rubber styrene Tile Board Adhesive Adhesive Paint
Holding Molding Foam Glidden
Insulation
Aromatic
Hydrocarbons
Benzene
_a
-
-
-
-
-
-
-
17
Ethylbenzene
1.8
1.7
15
0.45
-
0.14
547
211
12
m-Xylene
5.9
6.9
1.7
0.92
-
0.20
1185
717
50
Styrene
0.14
0.43
6.2
0.63
0.18
—
-
1.9
^-Xylene
1.9
3.9
0.39
0.89
0.83
0.08
202
301
28
Isopropyl benzene
0.38
0.33
1.7
0.41
-
0.15
-
53
8.9
n-Propyl benzene
0.68
0.72
0.86
0.84
-
0.22
-
53
13
wEthyl toluene
3.4
4.6
0.44
3.2
-
-
-
281
57
1,3,5-Trimethyl-
benzene
1.7
2.7
0.08
1.4
-
0.10
-
212
32
_o~Eth y 11 o luene
0.41
2.0
0.07
1.3
-
0.07
-
97
18
1,2,4-Trimethyl-
benzene
2.3
7.1
0.17
4.7
.063
0,20
-
216
33
L ,2, 3-Trlmethyl-
benzerie
1.4
5.5
0.03
2.5
.027
0.06
—
151
20
Aliphatic
Hydrocarbons
®f-Pinene
0.19
1.4
-
-
6.8
-
-
-
i^-Decane
10
14
~
-
0.23
-
-
-
360
ji-Undecane
22
29
-
1.2
7.1
-
-
3245
250
_n-Dodecane
34
9.7
-
0.45
6.4
-
-
2295
110
-------
Table 2. Calculated Emission Rates of Target Compounds During Chamber Experiments (continued)
Vinyl Black
Cove Rubber
Molding Molding
Poly-
styrene
Foam
Insulation
Linoleum Carpet Particle Cove Carpet
Tile Board Adhesive Adhesive
Latex
Paint
Glidden
Oxygenated
Hydrocarbons
jn-Butyl acetate
Ethoxyethyl acetate
Chlorinated
Hydrocarbons
1,1,1-Trichloro-
ethane
Tri chloroe thylene
Tetrachloro-
ethylene
Chlorobe nzene
jr-Di ch 1 o r oben z e ne
o-Di chlorobenzene
0.30
0.12
0.U
1.8
0.42
0.11
0.31
0.97
1.3
1.9
289
0.38
0.47
0.20
0.18
aNo detectable emissions
-------
Table 3. Average Concentration and Possible Sources of Volatile Organics in a Newly Constructed Office Building
Average Concentration (ug/m^)
Over 72-hour Measurement Period
Compound Indoors Outdoors
Aromatic Hydrocarbons
Xylenes 62 5
Ethylbenzene 54 2
Trimethylbenzenes 130 0.5
Ethyltoluenes 40 2
Propylbenzenes 10 0.4
Benzene 3 4
Styrene 3 0.6
Aliphatic Hydrocarbons
n-Decane 520 0.9
n-Undecane 220 0.4
n-Dodecane 180 0.1
ot-Pinene 15 0.1
Chlorinated Hydrocarbons
1,1,1-Trichloroethane 17 1
1,2-Dichloroethane 0.8 0.2
Trichloroethylene 0.5 0.1
Chlorobenzene 0.6 0.02
p-Dichlorobenzene 0.3 0.07
Tetrachloroethylene 0.5 0.4
Carbon tetrachloride 0.4 0.6
Possible Sources (from Tables 1 and 2)
Adhesives, paint, molding, insulation
Adhesives, insulation, paint
Carpet adhesive, paint, molding, linoleum tile
Carpet adhesive, paint, molding, linoleum tile
Carpet adhesive, paint
Outdoors
Insulation
Paint, molding
Carpet adhesive, paint, molding, carpet
Carpet adhesive, paint, molding, carpet
Particle board
Cove adhesive
Linoleum tile
Linoleum tile
Insulation, molding
Molding, Insulation
Outdoors, paint
Outdoors
-------
INHALATION EXPOSURE FROM VOLATILE ORGANIC CONTAMINANTS IN DRINKING WATER
(1) Lynn Wilder, Region III, Technical Assistance Team,
(2) Gerald T. Heston, On-Scene Coordinator
(3) Michael Zickler, Sr. On-Scene Coordinator
(4) Richard Habrukowich, TATL, Technical Assistance Team
An Increasing number of groundwater/residential well water volatile
organic contamination problems has raised questions concerning the amount
of human Inhalation exposure to these organics. Typically, health risk
assessments of contaminated drinking water have only addressed the
exposure for the ingestion of volatile organics. However, over the last
several years, questions have arisen concerning the amount of exposure
occurring through the inhalation route during typical home water use (i.e.
showering). Results from a recent study showed that 80% of trichloro-
ethylene volatilized in a model shower system using room temperature
water. Thus, exposure to volatile organics during showering could make
inhalation an Important factor in health risk assessments of contaminated
potable water supplies.
EPA Region Ill's ERS recently conducted a study under the authority
of SF 104 (b) to go beyond the previous laboratory study and obtain
exposure information from inhabitated homes. In this research, trichloro-
ethylene concentrations were measured in the water and air over the
duration of a typical showering period in several residences with
trichloroethylene-contaminated drinking water supplies. From these
findings, human inhalation exposures are estimated and compared with
exposures from ingestion.
(1 & 4) Roy F. Weston, Inc., Region III Technical Assistance Team,
53 Haddonfield Road, Suite 306, Cherry Kill, NJ 08002
(2 & 3) U. S. EPA, Region III Emergency Response Section, fi!4 Chestnut
St., Philadelphia, PA 19107
123
-------
Introduction
Residential wells (groundwater) in a Pennsylvania community have been
shown to be contaminated with high levels of trichloroethylene (TCE), and
in some cases with low levels of tetrachloroethylene (PCE) and methylene
chloride. As part of an EPA Region III, Emergency Response Section (ERS)
CERCLA Removal, before in-house carbon filtration units were installed, a
study was conducted to determine the amount of TCE that volatilized from
the contaminated groundwater during showering.
Typically, health risk assessments from contaminated drinking water
have only addressed the exposure for the ingestion of volatile organic
contaminants (VOC's). Drinking water standards and health advisory values
for contaminated water are based upon ingestion exposure only. An
increasing number of groundwater/residential well water volatile organic
contamination problems have raised questions concerning the amount of
human inhalation exposure from these organics. Specifically, questions
have arisen concerning the amount of inhalation exposure incurred during
typical residential water usage (I.e., showering).
Physical properties of VOC's such as vapor pressure and limited
solubility in water, make these contaminants favor the air phase in an
equilibrium situation. Thus, greater concentrations will be found in air.
Volatile organics tend to be more of a contamination problem in
groundwater than in surface waters for several reasons. Surface waters
have a large surface area in contact with air. Equilibrium forces tend to
drive VOC's out of the aqueous phase and into the air. Since groundwater
has less contact with air, equilibrium forces tend to keep most volatile
organic contamination in the water phase. In the case of private wells
using groundwater, most of the equilibrium process occurs once the water
reaches a tap or showerhead. At these points, not only does the water
come in contact with air, but the droplet form in which the water leaves
the tap creates a large surface area ratio of air to water. This aeration
process drives most of the volatile organics out of the aqueous and Into
the air phase.
A recent study using a model shower system' has shown that even
with room temperature water, 80% of TCE volatilized from incoming water.
When water temperature was Increased, greater than 90% of TCE volatilized.
Over the duration of the "shower" the concentration of TCE in air steadily
increased, showing no signs of reaching an equilibrium with the water
phase. Thus, exposure to VOC's during a typical showering period could be
quite large, making inhalation exposure a factor in health risk
assessments of contaminated drinking water supplies.
Using the International Commission for Radiological Protection^
rates and volumes for reference air Intake, an estimated 1 to 1.2 cu
meters of air is inhaled per hour for an average adult male or female.
Thus, in a 15 to 30 minute shower, 0.25 to 0.3 cu. meters and 0.5 to 0.6
cubic meters of air is inhaled, respectively.
The amount of retained TCE in the body through inhalation has been
estimated at 36-75%3. Using a TCE retention factor (RF) of 70, 70% of
the TCE inhaled is absorbed by the lungs and the remaining 30% is
exhaled.
124
-------
The Drinking Water Equivalent Level (DWEL) for TCE is 260 ug/L (1.40
mg/ra3). Additionally, the 10-6 cancer risk level of TCE is 2.8 ug/L
(0.02 mg/ra^). The November 13, 1985, Federal Register proposed maximum
contamination levels (FMCLb) of 5.0 mg/L (.027 mg/ra^).
Experimental Methods
Air Analysis
At each residence the MIRAN 1A* portable ambient air analyzer was Bet
up and calibrated. All bathroom doors and windows were closed and any
fanB turned off. One end of Teflon tubing was attached to the MIRAN inlet
port; the other end was placed approximately 1 foot in front of the
showerhead (5.5 feet above the shower floor) above the spray. After a
baseline reading was obtained for 5-10 minutes, the water flow rate was
calibrated and the water was brought to room temperature. The shower was
then turned on and allowed to run for 30 minutes.
Grab air samples were taken for private laboratory analysis after the
shower had run for 15 to 30 minutes. The inlet port of a 1 liter Tedlar
bag was attached to the MIRAN outlet port and allowed to fill. Samples
were placed on ice.
After 30 minutes, the shower water was terminated, the Teflon tube
was disconnected and a final baseline reading was taken for 5 to 10
minutes.
Water Analysis
To calculate the well concentration of TCE, a water sample was
collected in a 40 ml VOA bottle before the shower began. A duplicate
sample was collected for private laboratory analysis. During the shower
run, a water sample was collected every 10 minutes in two locations. One
set of samples were collected near the MIRAN inlet tube, and the second
set was collected near the shower floor.
Samples were analyzed for TCE (headspace analysis) using a portable
GC with a PID detector (Photovac)*. Standards and samples were analyzed
in duplicate. As a QA measure, initial well water duplicate samples were
sent to a private laboratory for analysis (volatile organic analysis by
EPA Method 601).
Reaults
The results of continuous air monitoring in each of the four
residences is shown in Table I. Concentrations were calculated for every
10 minutes of the shower run and after 10 minute after the shower was
terminated.
Laboratory results of grab samples taken at 15 and 30 minutes after
the start of the shower were typically 10-252 higher than these readings.
These differences were probably due to differences in analytical methods
and techniques. Therefore, results of the continuous air monitoring
method (MIRAN) are conservative in nature.
125
-------
From this data, assuming 0.3 and 0.6 m^ of air inhaled for a 15 to
30 minute period, respectively, the TCE concentrations inhaled are
displayed in Table II. The final column in the table displays the amount
of TCE absorbed through the lungs, assuming a 0.70 retention factor (RF).
As Table II displays, depending on the TCE concentration in water, an
exposure range from 1.3 to 17.2 mg occurred over the duration of a 30
minute shower.
Results of grab water samples taken before and during the shower
period are shown in Table III. Split samples analyzed by a private
laboratory agreed closely with the initial water concentrations (time).
Water samples were taken at two locations: at the "head" near the MIRAN
inlet, and at the "feet" at the shower bottom.
From the data in Table III, a rough percent volatilization can be
calculated. These percentages were derived using the following
calculation:
%Volatillzatlon-(lnitlal TCE cone - TCE cone at sampling point)
(initial TCE cone in water)
Results of these calculations are displayed in Table IV.
Conclusion
The results of this study indicate that during a shower, significant
amounts of TCE volatilize. In every case, inhalation exposures after 10,
20 and 30 minutes exceed the DWEL, PMCL, and the 10-6 cancer risk. Table
V compares the amount of TCE absorbed by the lungs with the corresponding
DWEL and 10x-6 cancer risk value for that time period.
The values Of TCE absorbed (Table V) are conservative considering
that the shower water was room temperature. Had the instrumentation used
been able to withstand water vapor, much higher exposure values may have
been found.
Water analyses reveal that at room temperature, from 23-53% of TCE
has volalitlzed by the time it reached the MIRAN inlet sampling port
location (5.5 ft. above the shower floor). Samples collected near the
shower floor indicate that TCE continues to volatilize as contaminated
water droplets fall through the air. At a point just before reaching the
shower floor, from 64-74% of TCE had volatilized from the shower water.
If the water temperature had been increased to normal showering
temperatures, a greater amount of TCE may have volatilized.
Table VI compares the amount of exposure to TCE from the route of
inhalation versus Ingestion. At lower concentrations, inhalation and
Ingestion exposures appear to be equivalent. At higher concentrations,
however, higher exposures occur through ingestion.
When compared to Table VII, Table VI shows that both routes of
exposure exceed all standards and risk values. Therefore, the treatment
method of the contaminated water supply should address both routes of
exposure. If, for example, affected residents were supplied with bottled
water, but were still allowed to continue bathing and showering in the
contaminated water, the amount of exposure through inhalation would exceed
the DWEL, PMCL, and
126
-------
10-6 values. Thus, by not completely addressing the problem, the public
health may remain unprotected.
Inhalation of volatile organics is a significant route of exposure
from contaminated water supplies. Indeed, long-term health risk values
are exceeded through inhalation alone, without even taking exposure
through ingestion into consideration. Therefore, inhalation exposure
values should be considered along with other exposure routes when drinking
water standards and health risk numbers are formulated. The possibility
of inhalation exposure should also be considered when contaminated water
problems are addressed.
References
1. J. B. Andelraan, S. M. Meyers, L. C. Wilder, "Toxic Showers and Baths",
Science News. 130:12. (1986).
2. International Commission for Radiological Protection (1CRP), Report of
the Task Group on Reference Man, Edition No. 23. Pergamon Press, New
York. (1975).
3. United States Environmental Protection Agency, Office of Emergency and
Remedial Response, ECOA, Cincinnati, OH. (1984)
* Reference to products and manufacturers are for illustration only;
such references do not imply product endorsement by the U. S.
Environmental Protection Agency.
127
-------
TABLE I. TCE Concentrations in Shower Air During a 30-Minute Showering
Period.
Resident
Well Water
Time
TCE Cone.
TCE Cone.
Number
Cone., ppb.
Min.
PPm
mg/cu. M.
1
725
0
0
0
10
20
30
0.5
2.7
40
2
7,975
0
0
0
10
2.0
11.9
20
2.9
15.6
30
3.5
21.0
40
1.7
9.2
3
720
0
0
0
10
1.0
5.4
20
1.1
5.9
30
1.3
7.0
40
0.1
0.5
4
21,800
0
0
0
10
6.1
32.9
20
7.2
38.8
30
8.9
48.0
40
3.8
20.5
Dashed lines indicate no readings.
TABLE II:
TCE Inhalation Exposure
Calculations
During a 30-Minute Shower
Period.
Resident
TCE Well Time Amount Inhaled
Amount Absorbed
Total
Number
Cone., ppb. Min.
(mfi)
(mg)
(ibr)
1
725 0
0.0
0.0
30
1.6
1.3
1.3
2
7,975 0
0.0
0.0
10
2.4
1.7
20
3.1
2.2
30
4.2
2.9
6.8
40
1.8
1.3
3
720 0
0.0
0.0
10
1.1
0.2
15
1.2
0.8
30
1.4
1.0
2.0
40
0.1
0.1
4
21,800 0
0.0
0.0
10
6.6
4.6
20
7.8
5.4
30
9.6
6.7
17.2
40
0.8
0.5
128
-------
TABLE III. TCE Concentrations In Shower Water at "Head" and "Feet"
Locations During a 30-Mlnute Shower Period.
Resident Time "Head" TCE "Feet" TCE ,
Number Mln. Cone, (ppb) Cone, (ppb)
1 0 725
10
20 588 264
30 384 140
2 0 7,975
10 5,773 4,328
20 4,920 2,448
30 3,840 1,668
3 0 720
10 704 226
20 440 280
30 504 206
4 0 21,800
10 9,984 4,600
20 9,992 5,076
30 10,924 7,200
Note: TCE concentration in water may fluctuate due to changed
groundwater concentrations.
Dashed lines indicate no readings.
TABLE IV. Percent Volatilization of TCE From Water During a 30-Minute
Shower Period.
Resident
Number
TCE WE11
Cone, ppb.
Average % Vol.
"Head"
Average X Vol.
"Feet"
725
7,975
720
21,800
33.0
39.3
23.7
52.8
72.1
64.7
63.8
74.2
129
-------
TABLE V. TCE Amount Absorbed During Showering Compared with Standards
and Risk Values.
Resident
Number
Time
Min.
TCE Absorbed
DWEL
.(rag)
10x-6 Risk
..
PMCL
(rag)
0
10
20
0
30
1.3
0.6
0.012
0.016
2
0
0
0
0
0
10
1.7
0.2
0.004
0.005
20
2.2
0.56
0.008
0.011
30
2.9
0.84
0.012
0.016
3
0
0
0
0
0
10
0.2
0.2
0.004
0.005
15
0.8
0.42
0.006
0.008
30
1.0
0.84
0.012
0.016
4
0
0
0
0
0
10
A.6
0.2
0.004
0.005
20
5.4
0.56
0.008
0.011
30
6.7
0.84
0.012
0.016
Dashed lines indicate no readings.
TABLE VI. Comparison of Ingestion and Inhalation Exposures of TCE at
Various Concentrations in Potable Water.
Resident Well Water Amount TCE Amount TCE
Number TCE Contamination Inhaled (mg) * Ingested (mg) **
(ppb)
1
725
1.3
1.5
2
7,975
2.9
15.9
3
720
1.0
1.4
4
21,800
6.7
43.6
* After a 30-minute shower.
** Assuming a consumption of 2 liters of water.
130
-------
TABLE VII. Inhalation and Ingestion Values of DWEL, 10-6, PMCL.
Inhalation* (nig)
Ingestion**
DWEL
10-6
PMCL
0.84
0.012
0.84
0.52
0.005
0.01
*
**
Assuming a 30-minute inhalation time.
Assuming a consumption of 2 liters of water.
131
-------
SAMPLING TOR GAS PHASE NICOTINE IN ENVIRON-
MENTAL TOBACCO SMOKE WITH A DIFFUSION DENUDER
AND A PASSIVE SAMPLER
Delbert J. Eatough, Cynthia L. Benner, Jos6 M.
Bayona, Fern M. Caka, Hongmao Tang, Laura
Lewis, John D. Lamb, Milton L. Lee, Edwin A.
Lewis and Lee D. Hansen
Chemistry Department, Brigham Young Univer-
sity, Provo, Utah 84602
Nicotine Is unique to, and a major constituent of, environmental
tobacco smoke. The nicotine in mainstream smoke is present In the parti-
cles, while nicotine in freshly diluted environmental tobacco smoke is
present mostly in the gas phase. Several other organic nitrogen bases
including pyridine, vinyl pyridine, nicotyrine, myosmlne and cotinine are
shown to be present in the gas phase of environmental tobacco smoke by the
use of a benzenesulfonic acid- coated annular diffusion denuder sampler.
Benzenesulfonic acid was also used in a passive sampler to collect basic gas
phase compounds from environmental tobacco smoke. The concentrations of
nicotine and other nitrogen-containing organic bases were determined by Ion
chromatography, gas chromatography and gas chromatography-mass spectrometry.
The passive sampler was calibrated against the annular denuder sampling
system in experiments with samples from a Teflon environmental chamber and
with samples collected in indoor environments with low to moderate concen-
trations of environmental tobacco smoke. The results indicate that nicotine
in the environment is present both in particles and in the gas phase, and
that gas phase nicotine is removed from indoor environments at a faster rate
than other constituents of environmental tobacco smoke, and therefore is not
a surrogate for exposure to other components of environmental tobacco smoke.
Environmental tobacco smoke (ETS) may be a major contributor to indoor
air pollution in environments where smokers are present (1). The identi-
fication and quantification of ETS exposure is important because of irri-
tant and suspected health effects associated with involuntary exposure (1,2)
and the large population which is exposed. Tracers of ETS used in the past
include respirable (or total) suspended particulate matter (RSP), CO, nitro-
gen oxides, nicotine, N-nitrosoamines, aromatic hydrocarbons, acrolein, and
frequency of smoking. Of these various tracers, only nicotine is unique to
Introduction
132
-------
ETS (1,2). While nicotine Is present at relatively high concentration where
smoking occurs, the use of nicotine as a tracer of ETS Is complicated by the
fact that nicotine is found primarily in the gas phase In indoor environ-
ments (3-6). Furthermore, gas phase nicotine may be removed from the en-
vironment at a faster rate than aerosol nicotine or the particulate portion
of ETS (1,7). Thus, the concentration of gas phase nicotine may underesti-
mate exposure to ETS (1,2). In order to determine the appropriateness of
nicotine as a marker for ETS exposure, accurate and sensitive methods are
needed for determining both gas and particle phase nicotine Indoors.
The gas-particle distribution of a semi-volatile compound will be
affected by the collection procedure as a result of volatilization of the
semi-volatile compound from particles collected during the sampling process.
Thus, any sampling procedure which attempts to first remove particles and
then sample the gas phase material after the particle filter will over-
estimate the concentration of semi-volatile components since gases evolved
from the particles on the filter become Indistinguishable from gas in
equilibrium with the particles. The only way to reliably determine both the
gas and particle phase concentrations of a semi-volatile compound is to
first sample the gases and then collect both particles and evolved gases
with an appropriate filter pack system. The gas phase species may be
collected with a diffusion denuder (4) or a passive sampler (9).
This preliminary report describes the use of both a passive sampling
device (PSD) and an annular diffusion denuder sampler to determine both gas
and particle phase nicotine and other nitrogen-containing organic bases In
ETS In an experimental chamber and in indoor environments. There are three
objectives of the research reported here. The first objective is to
develope techniques for the simultaneous determination of gas and particle
phase nicotine and other nitrogen-containing bases in indoor environments;
the second is to develope a PSD for the determination of personal exposure
to gas phase nicotine; and the third is to test the use of these techniques
In typical indoor environments.
g*T*"*lTTltn1
Sampling Gas Phase Nicotine. Gas phase nicotine and other bases were
collected with benzenesulfonlc acid (BSA) coated annular diffusion denuders
(4). Two or three denuder sections were used to check for complete col-
lection of gas phase compounds by the denuder surface.
Particle Sampling. The concentration of particle phase nicotine was
determined by sampling with quartz filters placed after the BSA coated
diffusion denuders. Nicotine lost from the particles on the quartz filter
was collected on a BSA saturated filter placed after the quartz filter.
Passive Gas Phase Nicotine Sanmler. The concentration of gas phase
nicotine was determined using the EPA PSD (9) (Scientific Instrumentation
Specialists) with a 22 mm glass fiber filter saturated with a 1 M BSA
aqueous solution in the central chamber of the PSD.
Environmental Chamber Studies. The 30 m^ Teflon chamber and associated
equipment have been described (4). Replicate experiments were conducted In
which 1, 2, 3 or 4 cigarettes (1R1 Kentucky Reference, University of Ken-
tucky) were burned in the chamber using a standard smoking cycle. After
equilibration of the combustion products in the chamber for a few minutes,
samples were simultaneously collected with both the denuder and passive
sampling systems over a three hour period. The flow rate (10-20 slpm)
through the denuder sampling system was controlled with Tylan mass flow
133
-------
controllers.
Indoor Environmental Studies. Three denuder and passive sampling
systems were used to collect ETS in several indoor environments. Samples
were collected just outside an office in which a single smoker was present
and also in an office lunchroom with a few smokers present for short periods
during the day. Samples were also collected in a home where two of the
occupants were smokers. Samples were obtained over 6-8 hour periods.
Analytical Techniques. Filters and BSA passive samplers were extracted
with HjO in an ultrasonic bath. Diffusion denuders were extracted with 5 mL
of HjO. A fraction of the extracted sample was analyzed for nicotine by ion
chromatography (4) using a UV-absorption detector. An aliquot of the
solution was made basic with NaOH, and the nicotine was extracted into
CH2CI2. The nicotine and other nitrogen bases in the resulting CH2CI2
solution was determined by capillary column gas chromatography and gas
chromatography-mass spectrometry (GC-MS) (10). Extraction efficiencies were
determined both by standard spiking experiments and by repeat extractions.
HbbiiIm. ««» Discussion
The gas-particle phase distributions of various organic nitrogen bases
in ETS determined from analysis of gas and particle samples with the annular
diffusion denuder and from analysis of particles collected on a quartz
filter (10) in the chamber studies are given in Table I. This gas-particle
phase distribution is affected by the concentration of nicotine in the
chamber as shown by the data In Table II. The difference in the mole ratio
of gas phase nitrogen bases to gas phase nicotine is little affected by
changing the concentration of nicotine in the chamber from 4 to 1 ^mol/m3,
Table II, because almost all of the nicotine is present in the gas phase in
both experiments. However, the ratio of gas phase nitrogen bases to
particle phase nicotine is nine times larger when the concentration of
nicotine in the chamber is 1 jimol/m^ compared to 4 pmol/m^, reflecting the
decrease in the percent of nicotine present in the particle phase from 10%
to 2% for the two experiments.
The gas phase nicotine may be reliably collected by the PSD as shown by
the data In Figure 1 for samples collected in both the chamber and indoor
experiments. The collection of nicotine by the passive sampler Is expected
to follow Picks First Law of Diffusion,
M - [C] D A/L
where [C] is the concentration gradient from the PSD surface to the sampled
air, D is the gas diffusion coefficient (0.063 ca^/a for nicotine, ref 4), A
is the cross sectional area of the diffusion channel, and L is the length of
the diffusion channel. For the PSD configuration used in this study, the
expected value for H, the mass flow, Is 30.3 cm^/mln based on previously
reported data (9). We expect that the value for H in the sampler as used by
us will be slightly lower than this value because the glass fiber filter
does not fit snugly into the center cavity of the PSD. The value for the
mass flow determined from the data in Figure 2 is 25.4±0.9 cm'/mln, In
reasonable agreement with the previously reported data. Similar agreement
has been found between empirical and theoretical mass flows for a passive
sampler using sodium blsulfate as the acidifying agent for the collection
surface (6). The data comparing the denuder and PSD for those samples
collected in the Indoor environments are shown in Figure 2. The line in
Figure 2 Is drawn' with the slope obtained from the data in Figure 1 and a
zero intercept. As indicated by the data In Figure 2, the detection limit
for nicotine with the PSD as used in this study is about 5 nmol/m^ (0.5
pg/m ) and the precision Is about ±25% for a sample collected over an eight
134
-------
hour period. The detection limit of the annular denuder for gas phase
nicotine is more than an order of magnitude lower than that for the PSD.
The reproducability of replicate denuder samples is about ±15%, thus making
it possible to obtain accurate data even at low concentrations of environ-
mental gas phase nicotine with sample periods of less than an hour.
Previously reported results using a BSA annular denuder to sample for
ETS nicotine in the environmental chamber (4) showed that 92% of the
nicotine was present in the gas phase, with total nicotine concentrations of
about 2000 nmol/m^. The data obtained with the annular denuder in the
present study was used to determine the gas-particle distribution of
nicotine in both the chamber and environmental studies over a wide concen-
tration range. The results, given in Figure 3, show that the gas-particle
phase distribution of nicotine in ETS in the Teflon chamber varies only
slightly with nicotine concentration below about 3000 nmol nicotine/m^.
Linear regression analysis of this portion of the chamber data gives r -
0.91 with a slope of -0.00004±0.00033 %/(nmol/m^) and an intercept of
97.5±0.8 t. However, the fraction of nicotine present in the gas phase in
the indoor environments is highly variable. This may be due to the adsorp-
tion of nicotine by non-ETS particles or to a more rapid deposition loss of
gas, relative to particle, phase nicotine to indoor surfaces. The loss of
gas phase nicotine in a chamber with metal walls is much more rapid than the
loss of pyridine or vinylpyridlne (7). Information on the relative import-
ance of these two mechanisms can be obtained by comparing the mole ratios of
gas and particle phase nicotine to other gases from ETS.
Data are available in this study to test the relative importance of the
two possible mechanisms which could contribute to the observed change in the
gas-particle distribution of nicotine in the chamber as compared to the
home. The concentrations of gas and particle phase nicotine determined for
a set of triplicate samples collected in the home were 122±43 and 5,0±1.5
nmol/m-*, respectively. The extracts from the BSA denuders for these samples
were analyzed for the various compounds shown in Table I. The mole ratio of
the major gas phase bases to gas and particle phase nicotine for this sample
set, along with the corresponding values obtained from the studies in the
chamber, are shown in Table II. The mole ratios of pyridine and vinyl-
pyridlne to gas phase nicotine are 7 and 4 times larger, respectively, in
the home than in the chamber. In contrast, the mole ratios of these
compounds to particle nicotine are the same, within experimental error, for
both the chamber samples at lower nicotine concentrations and the samples
collected from the home. The results are consistent with the loss of gas
phase nicotine (and myosmine) from the home atmosphere at a rate much faster
than that for the loss of particle phase nicotine. In fact, the data
suggest that particle phase, but not gas phase, nicotine may be a reasonable
tracer for ETS. Additional data are needed to further characterize the
deposition loss of various components of ETS.
Nicotine is a semi-volatile compound, and particle phase nicotine
cannot be accurately sampled on a filter because of loss of nicotine from
particles during sampling. This point Is well illustrated by Figure 4 which
compares the concentration of particle phase nicotine found using the
annular denuder sampling system to that collected on the filter only.
Particle phase nicotine lost from the quartz filters during sampling of ETS
averaged 59128 % of the total for the chamber samples and 41±30 % of the
total for the samples collected indoors. The range in particle phase
nicotine lost during sampling on a quartz filter was 20 to 97 % of the total
for the chamber samples and 4 to 94 % of the total for the samples collected
indoors. Clearly, data obtained using a filter pack sampling system will
not accurately reflect the gas and particle phase concentrations of ETS In
135
-------
indoor environments.
The passive sampling device and annular diffusion denuder sampler
described here can measure exposure to very low concentrations of environ-
mental gas phase (and in the case of the denuder, particle phase) nicotine.
Particle phase, but not gas phase nicotine may be a surrogate for ETS.
Future studies should focus on the variations in the ratios of these poten-
tial tracers as a function of amount of smoking, age of the indoor air mass,
ventilation and types of furnishings, and presence of other sources.
Acknowledgement
Appreciation is expressed to R.J. Reynolds Tobacco Co. for support of
this research through a grant to Hart Scientific Inc., and to John W. Craw-
ford, Gaylon Richards, Von Landon and Brenda K. Sedar for technical assis-
tance .
References
1. NAS (1986) "Environmental Tobacco Smoke. Measuring Exposure and Assess-
ing Health Effects." National Research Council. National Academy Press,
Washington, 337 pp.
2. DHHS (1986) "The Health Consequences of Involuntary Smoking." A Report
of the Surgeon General.. U.S. Dept of Health and Human Services, 332 pp.
3. Eudy L.W., Thome F.A., Heavner D.L., Green C.R., Ingebrethsen B.J.
(1986) "Studies on the Vapor-Particulate Phase Distribution of Environ-
mental Nicotine by Selective Trapping and Detection Methods." Proceed-
ings. 79th Annual Meeting of the Air Pollut. Contr. Assoc. Paper 86-
38.7, 22-27 June, Minneapolis, MN.
4. Eatough D.J., Benner C., Mooney R.L., Bartholomew D. , Steiner D.S.,
Hansen L.D., Lamb J.D., Lewis E.A. (1986) "Gas and Particle Nicotine in
Environmental Tobacco Smoke." Proceedings. 79th Annual Meeting of the
Air Pollut. Contr. Assoc. Paper 86-68.5, 22-27 June, Minneapolis, MN,
5. Hammond S.K,, Leaderer B.P., Roche A., Schenker (1986) "A Method to
Measure Exposure to Environmental Tobacco Smoke." Proceedings¦ EPA/APCA
Symposium on Measurement of Toxic Air Pollutants. ppl6-24, Raleigh, NC.
6. Hammond S.K., Leaderer B.P. (1987) "A Diffusion Monitor to Measure
Exposure to Passive Smoking." Environ. Set. Tech.. in press.
7. Thome F.A., Heavner D.L., Ingebrethsen B.J., Eudy L.W., Green C.R.
(1986) "Environmental Tobacco Smoke Monitoring with an Atmospheric
Pressure Chemical Ionization Mass Spectrometer/Mass Spectrometer Coupled
to Test a Chamber." Proceedings. 79th Annual Meeting of the Air Pollut.
Contr. Assoc. Paper 86-37.6, 22-27 June, Minneapolis, MN.
8. Eatough N.L., McGregor S., Lewis E.A., Eatough D.J., Huang A.A., Ellis
E.C., "Comparison of Six Denuder Methods and a Filter Pack for the
Collection of Ambient HN03(g) and HN02
-------
Table I. Gas and Particle Phase Organic Nitrogen Bases Identified in Envi-
ronmental Tobacco Smoke Equilibrated In a 30 Teflon Chamber.
Chemical
Class
Particles
Bases
Class
wtt of
Particle
10.9±1.2
Identified uMol Compound/Mol CO % Compound
Compounds— Sflg fhflBC Particles In Gas Phase
3.09±0.22g/mol CO
Nicotine
1060011700
9201
80
92
O
1+
to
Myosmlne
3801
30
not
75
78
110
Nicotyrlne
161
6
611
16
21
± 6
Cotlnlne
211
2
841
44
20
±15
Pyridine
4151
18
< 5
100
Vinylpyridine
7521460
< 5
100
Blpyridine
791
7
< 5
100
Table 11. Hole Ratio of Gas Phase Organic Nitrogen Bases from Environmental
Tobacco Smoke with Respect to Gas Phase and Particle Phase Nicotine in
Samples from a 30 Teflon Chamber and from a Home.
Nicotine
Phase
Gas Phase
Particles
Mole Ratio with Respect to Nicotine in Indicated Phase
Sample Pyridine Vlnvlpvrldlne BlPYlTifllne MyPSffllne
Lab High []a 0.039±0.002 0.071±0.043 0.007±0.001 0.036±0.003
Lab Low []b 0.053±0.008 0.066±0.012 0.005±0.006 0.027±0.003
Home 0,30710.059 0.24910.037 0.00710.002 0.037+0.003
Lab High []a 0.451+0.040 0.817+0.50 0.08610.010 0.413+0.048
Lab Low []b 5.20 10.8 6.47 11.2 0.49 10.55 2.65 +0.29
Home 7.36 +1.4 5.97 ±0.88 0.16 ±0.05 0.89 ±0.07
aSamples collected in the environmental chamber.
% total nicotine In gas phase - 92%.
^Samples collected in the environmental chamber.
% total nicotine in gas phase — 98%.
[Nlcotlne(g)] - 4 ^mol/m3,
[Nicotine(g)] - 1 pmol/m3,
137
-------
Figure lf Mass flow rate to the PSD vs the concentration of nicotine in en-
vironmental tobacco smoke in a chamber and in indoor environments.
10
9
8
7
6
5
4
Slope * 25,4 mL/min
3
2
1
0
0 2 4 6
(Thouaonds)
[Nicotine(g)]. r»mol/m3
Figure 2. Mass flow rate to the PSD vs the concentration of nicotine for low
concentrations of nicotine in environmental tobacco smoke in
indoor environments. The solid line is drawn with a zero inter-
cept and the slope for the data in Figure 1.
0.3
0 28
0.26
0.24
0.22 -
Slope =* 25.4 mL/min
O.OB -
0.06 -
0 04
0 02
80
100
160
60
120
180
140
200
40
20
0
[NicoUne(g)], nmol/m3
138
-------
Figure 3. Percent of nicotine present in the gas phase vs nicotine gas phase
concentration for samples of environmental tobacco smoke from a
Teflon chamber,0 , and from indoor environments,0•
100
95
90
V
in
O
£ 85
v>
O
o
c ao
V
c
o 75
Z
K
70
65
60
o
n,
n ~
~
3b
~
a
~
a
o
~
~
>
>
>
)
r i i i
0 2 4
(Thousands)
[Nicotine(g)], nmol/mJ
Figure 4. Comparison of concentration of nicotine in particles determined
using a denuder sampling system and that collected using a filter
for samples of environmental tobacco smoke from a Teflon chamber,0
, and from indoor environments,0 t
120
110
100
90
60
70
Slope = 1
60
50
40
30
20
10
0
120
100
0
40
60
80
20
[Particle Nicotine], nmol/m3, Oenuder
139
-------
ORGANIC VAPOR PHASE COMPOSITION OF S1DESTREAM AND
ENVIRONMENTAL TOBACCO SMOKE FROM CIGARETTES*
Cecil E. Higgins, Roger A. Jenkins, and Michael R, Guerin
Organic Chemistry Section
Analytical Chemistry Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37831-6120
Environmental tobacco smoke (ETS) has received considerable attention
because of its contribution to indoor air pollution. While some studies
have attempted to estimate the exposure o£ humans to ETS constituents by
extrapolating from information gleaned from investigations of sldestream
smoke (SS), few studies have reported a direct comparison between the
composition of SS and that of ETS, ln the study reported her? we
describe the relative compositional similarities and differences between
th. v.f=r ph... .f SS CH.C of TO. SS ... 8«,.,at.d
conditions, Both a new laminar flow chamber, which prevents significant
alteration of the near-cigarette environment, and a modified Neurath
chamber were used for SS generation. ETS samples "eurath
office environment. Vapor phase samples were collected on
~Research sponsored by the National Cancer
Interagency Agreement (NIH) NCI No. Y01-CP-30508 under
Energy Systems, Inc., contract DE-AC-05-840R21400.
Institute under
MartinMarietta
140
-------
Introduction
The impact of environmental tobacco smoke (ETS) on the quality of
indoor air has been an issue receiving considerable attention. While a
number of studies have been performed on indoor air composition (1,2),
data regarding the specific contribution of ETS to the chemical
composition of indoor air has been limited (3). There are significant
difficulties in determining both such a contribution and potential human
exposure to ETS. For example, there axe significant contributions to
constituent levels from other sources, and exposures to ETS are often
episodic and of varying concentration. To this end, investigators have
sought to estimate ETS composition and potential exposures to it by
extrapolating from the composition of sidestream smoke (SS) and mainstream
tobacco smoke (MS), the latter being much more well-understood. There are
at least two deficiencies with this approach. First, in the extrapolation
to SS composition from that of HS, the observed variability of SS
constituent deliveries is much less that those of HS, creating greater
uncertainty in the SS composition than would otherwise be warranted (4).
Secondly, in many procedures used to generate SS for chemical studies, the
generation system may severely alter the environment Immediately around
the cigarette, resulting in uncertainties about the relevance of the
resulting material to that produced under smoldering conditions in a
normal ambient environment.
The purpose of this study was to compare, both qualitatively and
quantitatively, the organic vapor phase composition of SS generated under
maximally relevant conditions with that of the vapor phase of ETS produced
from the same cigarette type in a field situation. A collection system
was employed which provides for trapping of a wide volatility range of
vapor phase constituents. The primary issue addressed was one of
comparability: Is the relative composition of the vapor phase of ETS
substantially different from that of the SS which produces It?
Experimental
The trapping system consists of a stainless steel tube, 20.5 cm long x
0.46 mm I.D., packed with three sorbent materials. Approximately 1.7 mL
of 35-60 mesh Tenax-GC (Alltech/Applied Science) is backed with
approximately 0.8 mL of 20-40 mesh Carbotrap (Supelco, Inc.), *
graphitized carbon black, which is, in turn, backed by 0.3 mL of Ambersorb
XE-340 (20-60 mesh, Rohm & Haas), After conventional resin cleaning
procedures prior to construction, the material is packed in the stainless
steel tubes, separated by small plugs of silylated glass wool, and
desorbed for several hours at 270*C with helium at a flow of 20 mL per
minute. Desorptlon flow is always in the direction of eh* Arabersorb being
the upstream end, while collection flow is in the reverse direction. In
this manner, constituents breaking through the Tenax are retained by the
Carbotrap, and so forth.
Pumps used to collect samples with the triple sorbent traps were flow-
throttled diaphragm pumps (Cole-Parmer Air Kadet). ETS samples wets
collected at flows of 500 mL per minute, typically for 60 minutes. For
collection of the more concentrated SS vapor phase, flows of 20-30 mL per
minute for 10 minutes were employed. In order to minimize collection of
smoke particulate phase constituents on the trap, Teflon-coated glass
fiber filters (Pallflex TX 40 H 120 WW) were placed immediately upstream
of the trap.
141
-------
To generate sidestream tobacco smoke, a laminar flow SS chamber
developed at ORNL was employed (5). The system consists of a tapered
glass chamber, Into which air flows through two stainless steel
laminarizlng screens. The lamlnarized air acts to collimate the
smoldering plume from the cigarette flrecone. Temperature and relative
humidity mapping Inside the chamber indicated that the environment around
the cigarette Is not substantially different from that outside the chamber
in a room-like environment. For these studies, air flow through the
chamber was 30 L/mlnute, equivalent to 2.8 cm/sec in the vicinity of the
cigarette.
All cigarettes used in the study were Kentucky Reference 1R4F, a
filtered, low-tar-delivery reference cigarette, quite similar to
cigarettes currently popular in the U.S. market. The cigarettes were
puffed during SS generation using an ADL/II smoke exposure system (Arthur
D. Little, Inc.) modified to smoke at one 35 mL puff per minute.
Mainstream smoke (MS) particulate matter was collected immediately
downstream of the cigarette butt on a standard Cambridge glass fiber
filter pad (6). MS vapor phase was expelled into a Tedlar gas sampling
bag, so as not to contaminate the Incoming air. For comparison purposes,
some SS was generated using an enclosed, water-jacketed chamber (7),
In order to produce an environment of ETS, 1R4F cigarettes were smoked
on the same modified ADL/II as that described above. Also, as described
above, MS particulate matter was collected on glass fiber filters, while
the vapor phase was expelled into gas sampling bags. Cigarettes were
smoked In a conventional office with a volume of about 37 m3. The
ventilating system was a single pass through type, and air exchange rates
(with the door shut) averaged 5.4 ACH, The door to the office was opened
periodically to simulate human entry and egress. Temperature and relative
humidity were not controlled, although typically averaged 38% at 24*C.
Ambient particle concentrations were monitored using both an automated
reporting piezoelectric microbalance (Model 5000, TSI, Incorporated, St.
Paul, MN) and an ORNL-modified, commercially available forward scattering
particle concentration sensor (Model RAS-1, MIE, Inc., Bedford, MA).
Gravimetric analysis of particulate filters upstream of the triple sorbent
traps yielded particle concentrations generally in good agreement with
those determined with the automated piezobalance.
Structural identification of individual constituents was performed
using a gas chromatograph/mass spectrometer (GC/MS, Hewlett Packard Model
5985). Quantitative analyses were performed by GC (Perkin Elmer Sigma 2),
outfitted with both a nitrogen/phosphorus detector (NPD) and a flame
ionization detector (FID). Data visualization and reduction was performed
using a Maxima Chromatography Workstation. In both cases, the vapor phase
samples were analyzed by thermally desorbing the trapping system at 270*C
with a helium gas flow of 10-12 mL/minute In an oven external to the GC.
Desorbate was transferred via a 30 cm length (heated) of fused silica
capillary to a stainless steel cryoloop (liquid nitrogen) at the head of a
60 m x 0.32 mm ID fused silica capillary column coated with DBS (1.0
micron film thickness). Temperature programming was initiated when the
liquid nitrogen bath was removed from the cyroloop. The oven was held at
0*C for 10 minutes, followed by programming to 230*C at 2.5" per minute,
followed by a 30 minute hold. Column effluent was split iuch that about
40% was directed to the NPD, and the remainder to the FID. Quantitation
was performed using external standards.
142
-------
Results and Discussion
The manner In which sampling and analysis is performed can have a
significant impact on the results obtained. For example, nicotine is
associated with the particle phase of mainstream tobacco smoke. In
contrast, there is an increasing body of evidence (8,9) to indicate that
nicotine resides exclusively in the vapor phase of SS and ETS. This is
due to the more alkaline conditions which exist in the cigarette firecone
during smoldering and the much higher volatility of nicotine in its free
base form. However, because nicotine is retained on filters typically
used to collect smoke particulate phase (9), one might be led to the
erroneous conclusion that nicotine resides in the particulate phase of
these matrices as well. For this study, the vapor phase was defined as
that material which passed through the filters under the conditions of
sampling. While it was presumed that the Teflon coating of the glass
fibers provides for a less absorptive and more inert surface, some
preliminary evidence was obtained which suggests that even these types of
filters retain higher molecular weight nitrogen-containing compounds.
Basifying the filter surfaces with sodium hydroxide increased the amount
of these constituents passing through the filters and being retained by
the vapor phase traps. However, we have not yet developed a consistent
treating procedure, so that the work presented here does not compensate
for retention of vapor phase constituents on filters.
The material used in the traps to retain the species impacts on the
visualized composition of thfe vapor phase. Tenax-GC is used in many
laboratories as & near universal sorbent medium, but it is subject to
relatively rapid breakthrough for C5 and lighter compounds (10). In our
present system, the use of a second stage and a third stage sorbent
significantly Increases the number of constituents visualized and reduces
breakthrough. Comparisons of constituent retention by Tenax-only and the
triple sorbent traps indicated the retention of substantially more
volatile constituents, including acetonitrile, lsoprene, 2-butanone, and
acrylonltrlle by the latter. In addition, breakthrough of most of these
species, under the sampling conditions of the experiments, were on the
order of a few percent. Reproducibility of the sampling system appears to
be very good for most constituents. During simultaneous determinations of
ETS atmospheres, relative standard deviations (RSD) for determinations of
acrylonltrlle, benzene, pyrrole, and limonene ranged from 2-10%. For
compounds such as lsoprene, 2-butanone, and toluene, RSDs ranged from
15-30%.
The system employed to generate the SS for study potentially has an
Influence on the relative composition of the smoke. For example, the
vapor phase composition of SS generated using an enclosed, water jacketed
chamber (7) was compared with that of the SS generated using the QRNL
laminar flow system (5). In order to compensate for differences in the
amounts of material collected, responses obtained for individual
constituents were normalized to those of toluene. For many of the major,
more volatile constituents, relative ratios obtained for the jacketed
chamber were quite similar to those obtained using the laminar flow
system. However, for some of the major, less volatile constituents
(ethylbenzene, m+p xylene, limonene, pyrrole, and pyridine), the relative
ratios were a factor of 2-4 lower. Within this range of variability, the
overall compositional profiles were quite similar.
Deliveries of selected SS vapor phase constituents generated using the
laminar flow system are reported In Table X. The variations in the
143
-------
measured deliveries ranged from a low of 13« for toluene, to a high of 33%
for ethylbenzene. Iaoprene and toluene were the moat prevalent organic
vapor phase constituents eluting prior to limonene (the upper limit of our
study). Propionitrile, 2-butanone, benzene, pyridine, pyrrole, m+p
xylene, and limonene all were delivered at levels in the range of 200-400
ug/cigarette.
In Figures 1 and 2 are compared high resolution GC profiles of SS with
that of ETS, using flame ionization and nitrogen specific detection,,
respectively. Qualitatively, the composition of ETS vapor phase from
acetonitrile through limonene is quite similar to that of ETS. However, a
number of differences can be observed. First, in the FID profile, the
major constituent eluting one minute after benzene appeared in some
chromatogranis, but not all, making it difficult to identify using GC/MS.
The major compound eluting immediately prior to limonene (tentative
structural identification suggests that it may be a diol) appears
predominantly in ETS vapor phase samples. Otherwise, the overall profiles
are quite similar. For the nitrogen containing compounds (Figure 2),
several of the higher boiling constituents, eluting after picoline, appear
to be present in the ETS vapor at proportionately higher levels than they
are in the SS vapor phase. Experiments performed in our laboratory
indicated that this was not an artifact of the sampling conditions for
ETS. We speculate that more volatile constituents initially residing in
the particulate phase of SS immediately following generation may evaporate
with time, to become part of the vapor phase of ETS. We have also
observed this phenomenon with relatively high molecular weight
hydrocarbons, such as neophytadiene, evaporating from the particulate
phase of SS as it undergoes substantial dilution <11),
In Table 2 are listed quantities of several major constituents of ETS
vapor phase for which confirmatory GC/MS and retention time matching using
authentic standards have been performed. Comparison of the relative (to
toluene) composition of ETS vapor phase constituents (in Table 2) with
that of SS constituents (in Table 1) indicates differences of only a
factor of 2-3. All of this data was obtained at relatively high ETS
particulate concentrations (330 ug/m3). While this particulate level
represents a relatively smoky environment, it is well within the range of
levels reported for typical ETS particulate concentrations <12).
Interestingly, for those constituents for which Maximum Allowable Exposure
{MAE) levels have been established by 0SHA, the levels observed in this
study were often two to three orders of magnitude below those limits.
However, the levels determined here are substantially greater than those
which might be predicted from the SS emission rates and SS particulate
deliveries, For example, the amount of SS particulates generated from the
1R4F using the generator described above is about 26 rag/cigarette <5).
This includes any vapor phase constituents (such as nicotine) adsorbed on
the filter. Correcting this apparent particulate delivery for vapor phase
nicotine retained by the particulate filter and extrapolating the 850
ug/cigarette SS toluane delivery to an ambient ETS particulate level of
330 ug/m would suggest an ambient toluene contribution of 12 ug/m , a
factor of 6 less than that observed. Background levels of toluene were
determined to be about 10 ug/m3, or less than 20« of the difference. It
Ls likely that evaporation of lower boiling constituents of SS particles
upon dilution into a larger volume reduces their effective mass
concentration. This, in turn, diminishes their efficacy as a predictor of
ETS vapor phase levels. Wall losses, settling of the larger particles, or
inaccuracies in the determination of the particle concentration by the
piezobalance due to evaporation of the particles during collection may
144
-------
also contribute to Inaccuracies In the estimation. This is additional
justification for direct measurement of the constituent of interest for
the purposes fo exposure assessment.
Conclusions
Methodology for the characterization of the vapor phase of sidastream
and environmental tobacco smoke has been presented, along with the levels
of selected constituents. Examination of the high resolution GC profiles
suggests that the relative composition of the major constituents of ETS
vapor phase is resonably similar to that of the S3 which produces it,
within a range of 2-3. More detailed examination will be required to
determine the relevance of this observation for minor constituents, at
lower ETS levels, or for a wider range of cigarettes. Even at a
relatively high ETS level (as described by particulate concentration), the
concentrations of major vapor phase constituents appear to be much lower
than the Maximum Allowable Exposure limits established by OSHA, However,
levels were determined to be higher than that which might be estimated by
extrapolation from SS particulate deliveries.
145
-------
References
1. L. A. Wallace, E. D. Pellizzarl, and S. M. Gordon, "Organic chemicals
in indoor air: A review of human exposure studies and indoor air
quality studies," Proceedings of the Seventh Life Sciences Symposium,
Knoxville, Tennessee, October 29-31, 1984, Indoor—Air and Human
Health. R. B. Gammage and S. V. Kaye, eds., Lewis Publishers, Inc.,
1985. pp. 361-378.
2. E. D. Pellizzarl, T. Hartwell, H. Zelon, C. Leninger, M. Erickson, and
C. Sparacino, "Total exposure assessment methodology (TEAM): prepilot
study1 - Northern New Jersey," EPA Contract NO. 68-01-3849, Office of
Research and Development, U.S. Environmental Protection Agency,
Washington, DC 20460.
3. The Health Consequences of Involuntary Smoking - A Report of
Surgeon General. U.S. Department of Health and Human Services, Public
Service Centers for Disease Control. Chapter 3, "Environmental
tobacco smoke chemistry and exposure of nonsmokers," pp. 121-174,
1986.
4. M. R. Guerin, C. E. Higgins, and H.. A, Jenkins, "Measuring
environmental emissions from tobacco combustion: sidestream cigarette
smoke literature review," AEfflggphgglfi Envlrgnffignt 21(2), 291-297,
1987.
5. J. H. Moneyhun, R. A. Jenkins, and M. R. Guerin, "Development of an
improved sidestream smoke generation and collection apparatus,"
presented at the 40th Tobacco Chemists' Research Conference in
Knoxville, Tennessee, October 13-16, 1986.
6. W. B. Wartman, Jr., E. C. Cogbill, and E. S. Harlow, "Determination of
particulate matter in concentrated aerosols. Application to 'tar' and
nicotine in cigarette smoke," Anal. Chem. 21 1705-1709 (1959).
7. K. D. Brunnemann and D. Hoffmann, "The pH of tobacco smoke." Food
Toxicol. 12. 115-124, (1974).
8. D. J. Eatough, C. Benner, R. L. Mooney, D. Bartholomew, D. S. Steiner,
L. D. Hansen, J, D. Lamb, E. A, Lewis, and N, L. Eatough, "Gas and
particle phase nicotine in environmental tobacco smoke," 79th annual
meeting of the Air Pollution Control Association, Minneapolis,
Minnesota, June 22-27, 1986 (Department of Chemistry, Brigham Young,
University, Provo, Utah 84602). y v S
9. L. W. Eudy, F. A. Thome, D. L. Heavner, C. R, Green, and B, J.
Ingebrethensen, Studies on the vapor-particulate phase distribution
J^nX t0rUnen^w ti . rie by selected trapping and detection methods,"
I? Toot000 ^_8t8_ Research Conference, Montreal, Canada, October
Z' *R* y R®ynolda tobacco Company, Bowman Gray Technical
Center, Winston-Salem, North Carolina 27102).
10' F' J' King' P' L* Levins, and J. F. Piecewicz,
Characterization of sorbent resins for use in environmental
Sr^7-7.-Sri2SS! *f".
146
-------
References (continued)
11. R. S. Ramsey, Oak Ridge National Laboratory, unpublished work.
12. J. D. Brain and B. E. Barry, "Biological Potential and Exposure Dose
Relationships for Constituents of Cigarette Smoke," in Indogy Air apd
Human Health. R. B. Gamraage and S. V. Kaye, eds,, Lewis Publishing,
Chelsea, Michigan, 1985, pp- 215-225.
13. S. K, Hammond, B. F. Leaderer, A. C. Roche, and M. Schenker,
"Collection and analysis of nicotine as a marker for environmental
tobacco smoke," Atmospheric Environment. 21(2), 457-462, 1987,
147
-------
Table 1. Deliveries of Selected Sidestream Smoke Vapor Phase Constituents
Kentucky Reference 1R4F Cigarette
ORNL Laminar Flow Sidestream Generator
Constituent Deliveries*, ug/cigarette
(mean + one standard deviation)
Nitrogen-specific Detector Flame Ionization Detector
Acetonitrile
1090
+
213
b
Isoprene
a
2494
+
724
Acrylonitrile
182
+
66
176
+
36
Propionitrile
177
±
57
246
+
72
2-Butanone
a
402
+
53
Benzene
a
370
±
75
Pyridine
283
+
79
283
+
69
Pyrrole
301
±
101
372
+
93
Toluene
a
848
±
113
4-Plcoline
126
±
63
b
Ethylbenzene
a
119
±
40
m+p xylene
a
311
+
59
Limonene
a
332
+
94
a.
b.
i,
No detectable response on nitrogen - phosphorus detector.
Not well quantitated using flame ionization detector.
N - 3
148
-------
Table 2. Concentrations of ETS Vapor Phase Constituents
ETS particle level: 330 ug/m3
ETS nicotine level: 120 ug/m3*
Concentration, ug/m3
Constituent Nitrogen-specific detector Flame ionization detector
Acetonitrile
14 ± 3
15 ± 8
Isoprene
a
47 + 6
Acrylonitrlle
2.0 ± 0.6
4.5 ± 0.1
Propionitrile
4.4 ± 0.5
b
2-Butanone
a
48 ± 17
Benzene
a
16 ± 3
Pyridine
11.7 ± 0.6
11 ± 0.6
Pyrrole
9.5 ± 0.7
11 ± 1
Toluene
a
70 ± 22
4-Picoline
8.8 ± 1.1
b
Ethylbenzene
a
8.9 ± 0.9
m+p xylene
a ,
11 + 3
Limonene
a
16 + 2
a. No detectable response on nitrogen - phosphorus detector.
b. Not well quantltatied using a flame ionization detector.
*, ETS nicotine levels determined by the method of Hammond, et al. (13)
using the filters ahead of the vapor phase traps.
149
-------
Sidestream Generator
(FID)
i.
¦
i
*i
il
I!
0.00
4-00 6.00
x 10* minutas
i.
IS
i
i
J
il
!l
«
««
Environmental Tobacco Smoke
(FID)
«.0O-
i ¦II
4.00 1.00
* 101 nlnutflf
9.00
Figure 1 Comparison of chromatographic profiles obtained with flame
ionization detector (FID): sidestream vs, environmental
tobacco smoke vapor phase.
150
-------
Sidestream Generator
(NPD)
TOO-
1.00
I.M
too
0.00
Environmental Tobacco Smoke
(NPD)
too-
I.W
t.H
4.80
> l(M alnulM
Comparison of chromatographic profllaa obtainad with nitrogen/
phoaphotua detector (NPD): aidaatraaa va. anvlrorunantal
tobacco anoke vapor phaaa.
151
-------
STUDIES ON DUST DEPOSITIONS IN CERTAIN
INDOOR AND OUTDOOR ENVIRONMENTS OF SECUNDERAHAD
S.H. Raza, B. Nirmala, M.S.R. Murthy
Dept of Botany
Osmania University
Hyderabad, 500 007, India
The paper presents the measurment of dust depositions in certain
indoor and outdoor environments of Secunderabad. Railway Station plat-
forms and a big domestic mess were selected for indoor studies. Nacharam,
an highly industrially polluted area was selected for outdoor studies.
Size, weight and heavy metal composition from the dust sweeps of indoor
environments were analysed. The concentration of air borne dust particles
was 3.06 - 5.4 g/m2/m in railway station where as in mess it was 1.86 - 2.9
g/m2/m. Dust depositions on the Leaf surfaces of trees growing in indus-
trial areas were quantified. Pithecolobium dulce is found to be a very
good dust monitor. Polyfethia longifolia is susceptible to dust pollution.
152
-------
INTRODUCTION
Duat pollution has been found to be a growing problem in major
cities of India both in indoor and outdoor environments. Dust fall is
undesirable due to its deleterious effects on air quality and human
health (1). The need to monitor and control the elevated levels of
particulate pollution in indoor environments has been subject of great
concern. The provision of many conventional dustfall monitoring stations
in outdoors for a detailed study is found to be very expensive. Phyto
monitoring of dust can serve as a valuable and inexpensive alternative
(2, 3). Hence an attempt has been made to analyse the levels of parti-
culate matter in indoors of Secunderabad and also toinvestigate the
probable use of trees as monitors of dust pollution. For the study the
leaf surface depositions of dust of two different sites were considered
for comparative evaluation.
MATERIAL AND METHODS
Two different types of indoor environments, viz. railway station
platforms and a domestic mess in Secunderabad were selected for the
purpose. Dustfall was measured by glass plate method. 10 cm^ glass
plates coated with thin layer of grease were kept at desired sites
nearly 5 metres above the ground level. Weight of the dust deposition
(g/m2), size of the dust particles (microns) and heavy metals viz. Fe,
Co1, Zn, Mn and Cu were analysed (4) from dust sweeps for 3 seasons. The
leaf samples of 10 trees were collected in 3 different seasons from
Nacharam industrial area (Industrially polluted area) and University
Botanical Gardens (Control area) for the assessment of dust deposition
in the outdoor environments. The leaf samples of trees growing under
isoecological conditions were collected from 10 metre height and were
analysed for the dust depositions as described by Bhatnagar et al (5).
Chlorophyll of leaf samples was analysed (6).
RESULT AND DISCUSSION
The size and weight of dust particleB of the two indoor environ-
ments studied is presented in Table-1. It is evident that the dust
accumulations were more in both the sites in summer. The dust deposi-
tion was also more in the railway station platforms than in domestic
mess. It must be due to the burning of coal and large influx of passengers.
The weight and size of dust particles was found to be increasing consi-
derably as a function of increase in the duration of exposure. There
was nearly 20 times difference in 1 day and 30 day accumulation. It
shows that in the absence of continuous control, the dust depositions
would create problems on human health. The size of the particles was
not in the respirable range i.e., 0,02 - 5 microns. The relationship
between concentration of particulates with particulate diameter was fbund
significant (Y ¦ 0.83 + 0,32x; r " 0.71). The low values in dust
accumulation in rainy season suggest that indoor levels respond very
rapidly to changes in the ambient levels. 8.3-50 mg/m2 indoor deposition
153
-------
rates are reported by Funso Akeredolu (7) in the domestic houses of
Nigeria. The studies conducted by Michael etal (8) in Arizona reported
mostly suspended particulates in the ranges of 25.5 ~ 28.2 (mg/m3 ).
In the present investigation it is found that particulate matter pertai-
ning to concentration of the dust was considerably high which could be
due to the emissions of wood and coal combustion.
The composition of certain heavy metals of dust sweeps is presented
in Table-2. The heavy metal analysis shows comparatively higher concen-
tration in indoor environments of railway station. Mostly the heavy
metals could be emitted from the combustion of coal and dust carried
in by the incoming passengers (9). The sequence of concentration of
heavy metals in the dust was Fe > Mn > Zn> Cu > Co. However the
study showed that the concentration of heavy metals did not exceed the
normal limits in any of Che two sites (9).
The amount of dust deposition on leaves in polluted area is more
than the dust on the leaves of less pollutted area. The same trend
is observed in all 3 seasons. However the deposition of dust was minimum
in rainy months (Table-3). There was a great deal of variation in dust
trapping efficiency of trees. The.range,be|ng 67-88 g/ip2 vaB relatively
rich in pifte colobium dulce, Mangifera indica and Cassia fiatuala.
In CaesaTpina pulcherima and Azadirachta indica 37—60 g/m^ of dust deposi-
tion was observed. In the remaining trees such as Polyalthia longifolia
the deposition was found to be poor, the range being 6-37 g/m^,
Pithecolobium dulce is a fast growing perennial and evergreen plant with
a compact hedged canopy. This could be the reason for its higher
degree of dust trapping efficiency. Cone shaped canopy and deciduous
nature wight be cause of low dust trapping efficiency in Polyalthia.
However Caesalpinia pulcherima. Cassia fistula etc. were found to
be intermediate in nature because of compound leaf character which rather
acts aB a seive for the dustfall.
The concentration of chlorophyll got decreased in the leaves
to the extent of 50% particularly in certain trees like Polyalthia
longifolia, Dalbergia sissoo (Table-3). The reduction of chlorophyll
in the leaves could be due to the deposition of dust in layers. The simi-
lar such observations were made in Cassia fistula. Dalbergia, Polytlttiia
loneifolia ( 10, 11.). The point of interest is that the decline
in chlorophyll concentration in polluted areas would serve as a measure
of ait quality and the degree of susceptibility of trees to dust
pollution.
CONCLUSION
The knowledge on heavy metal composition size and weight of dust
sweeps in indoor anvironment would serve as a measure of particulate
pollution, in minimizing possible damages on human health. Trees
resistant to duat pollution can act as sinks of pollution. The morphology
and canopy of the trees play a greater role in the dust trapping effici-
ency. Chlorophyll content of plants in polluted environment is a
purposeful parameter in caliberating the pollutant concentration in outdoor
environment.
154
-------
REFERENCES
1. K. Jalees, J.M. Dave, " Environmental hazards of particulate
pollution", Indian J". Air Poll. Control. 2 : 96 - 101 (1979).
2. N. Singh, D.N. Rao, "Certain responses of wheat plants to cement
dust pollution", Environ. Pollut. 24: 75 - 81. (1981).
3. E.F. Darley, "Studies on the effect of Cement - kiln dust on
vegetation", J . Air Poll Cont. 16: 145 - 150 (1966).
4. M.L. Jackson, " Soil chemical Analysis", Prentice Hall of India
Pvt. Ltd., New Delhi. 1967, pp 498.
5. A.R. Bhatnagar, A.B. Vora, T.S. Patel, " Measurement of Dust fall
on leaves in Ahmedabad and its effect on chlorophyll,
Indian JAir Poll. Control¦ VI : 77 - 91 ( 1985 ).
6. G.I. Amon, " Copper enzymes in isolated chloroplasts, polyphenol
oxidase in Beta vulgaris, Plant Physiol 24 (i) : 1-15 (1949).
7. Funso Akeredolu, " Indoor/outdoor levels of particulates in
Ile-Ife, Nigeria", Proc. of 7th World Clean Air Congress,
Sydney. 5 : 165 - 167 (1986).
8. D.L. Michael, G. Corraan, M.K. 0 Rourke, C.J. Holberg, "Indoor -
outdoor air pollution, allergin and meterological monitoring
in an arid south west area", JAPCA. 34 (10) : 1035 - 1038 (1984).
9. H.Zeedijk, " Emissions by combustion of solid fuels in domestic
stoves", Proc. 7th World clean air congress, Sydney. 4 : 78-75 (1986).
10. T.M. Das, A. Bhaumik, A. Ghoah, A. Chakravorty, "Trees as dust
filters", Science Today. 19; 19-21 (1981).
11. M. Yusuf and L.N. Vyas, " Effect of cement dust pollution on
some selected plants species growing around Udaipur cement
works", presented at the All India Seminar on Air Pollution,
April, 1982.
135
-------
TABLE - I. Size (microns) and weight(g) of dust collected in two different
indoor environments.
1 day 30 days
Size
Weight
Size
Weight
Railway
W
9.7
0.23
48. 2
5.1
Station
s
10.7
0.27
49.4
5.4
R
9.1
0.25
18.6
3.0
Domestic
W
7.7
0.13
23.3
2.5
Mess
s
8.7
0.15
9.8
2.9
R
7.5
0.21
19.6
1.8
TABLE - II Composition of certain heavy metals of dust sweeps (ppjn)
Fe
Co
Zn
Cu
Mn
Railway
W
25.9
0.27
1.17
1.14
13.3
Station
S
66.0
0.50
1,23
2.11
14.6
R
18.0
0.20
0.95
0.76
10.3
Domestic
W
15.9
0.05
0.95
0.55
6.6
Mess
s
13.9
0.21
2.04
0.84
15.9
R
9.9
0.33
0.95
1.29
6.6
W
S
-
Winter
Summer
R - Rainy
156
-------
TABLE - III Dust deposition on different trees and chlorophyll content.
Name of the tree
Dust
Chi
Dust
Chi
Dust
Chi
Azadirachta indica
17.4
2.3
7.9
13.9
12.4
9.9
Anoaa squa mosa
23.2
3.6
6.3
11.4
16.3
8.4
Caesalpinia pulcherima
34.9
2.4
24.3
15.9
43.3
11.6
Cassia fistula
40.8
3.7
24.3
11.7
31.3
9.3
Dalbergia sissoo
12.6
1.9
4.3
12.0
6.4
8.0
Eugenia iambolana
12.4
2.1
6.1
12.5
9.6
10.1
Manxifera indica
30.4
3.1
8.9
13.0
22.6
7.6
Pongamia glabra
13.0
3.9
8.0
11.8
10.4
7.3
Polyalthia longifolia
12.4
3.1
3-4.
13.6
5.4
8.4
Pithecolobium dulce
54.3
2.5
24.9
16.1
34.4
12,4
LP - Less polluted area;
HP - Highly polluted area;
S - Sumner
R - Rainy
W - Winter
Chi - Chlorophyll (mg/g dry wt)
2
Dust - g/* /m
HP
S
R
W
Dust
Chi
Dust
Chi
Dust
Chi
37.5
1.0
24.3
12.8
32.9
6.5
33.4
1.4
15.0
8.4
15.0
7.0
48.3
1.1
35.4
12.4
37.3
8.0
88.0
0.6
46.3
9.3
54.5
6.4
41.6
0.4
6.0
8.4
4.4
5.4
34.8
1.6
9.6
7.6
16.4
10.0
67.8
1.64
28.6
8.9
36.4
9.0
25.6
2.4
4.5
6.4
15.0
6.4
22,9
1.3
9.3
7.4
12.4
5.3
76.3
2.7
34:3
11.3
48.9
9.8
___
- - i. — — —
_______
J
_
-------
LOSS OF NITRIC ACID WITHIN INLET DEVICES
FOR ATMOSPHERIC SAMPLING
B. R. Appel, V, Povard and E. L. Kothny
Air and Industrial Hygiene Laboratory
California Department of Health Services
2151 Berkeley Way, Berkeley, CA 94704
Seven inlet devices, primarily cyclones, were evaluated for nitric acid
transmission efficiency at simulated day- and nighttime conditions. The
units employed were fabricated from solid PIPE, perfluoroalkoxy-coated
aluminum, PTFE-coated glass, and stainless steel. Transmission ef-
ficiencies for precleaned units ranged from 18 to > 100%, and were
generally lower under simulated nighttime conditions. Residence time and
surface composition were the predominant design parameters influencing
HN0S transmission. Devices with calculated HNOg residence times <0.2 sec
exhibited minimal HNOs loss. Atmospheric sampling provided more effective
conditioning of the coated aluminum cyclone than did laboratory sampling
of nitric acid in purified air. Solid PTFE or PTFE-coated glass inlet
devices are preferable to the coated aluminum cyclone for atmospheric HNOs
sampling.
158
-------
Introduction
Results from the 1985 California Air Resources Board-sponsored Nitrogen
Species Methods Comparison Study (NSMCS)1 suggest that differences in
measured HN03 between methods could be influenced substantially by the
degree of loss of HNOs within sampler inlets. The use of one inlet system
fabricated from polytetrafluoroethylene (FTFE) appeared to result in
substantial HNOs loss. Other systems with FTFE inlet devices showed no
such loss. In an effort to elucidate the problem, seven inlet devices
were evaluated for nitric acid transmission efficiency at their normally
employed flow rates. Most were cyclones employed previously by various
research groups. The devices evaluated were fabricated from solid FTFE,
perfluoroalkoxy (PFA)-coated aluminum, PTFE-coated glass, and stainless
steel.
Experimental Methods
The devices evaluated are listed in Table I together with their internal
volumes. Internal volumes include those of any additional plumbing ahead
of the filter sampler used to collect HNOa. Nitric acid was generated
with a Unisearch Associates permeation tube maintained at 53'C. The acid
was diluted to a total volume of 7 cfm with purified ambient air. The
seven units compared were attached to symmetrically arrayed stainless
steel ports of a cylindrical sampling manifold. The equivalence of the
seven 1.5 in. ID ports was assessed by parallel HNOs sampling with open-
face filter holders connected, to each port containing Nylon filters.
Sartorius 0.65 ftm pore size Nylon filters were used in all cases. A pre-
cision of 7.5% or better was observed at equal flow rates In two trials.
Samplers operating at 16 to 28 1pm sampled from the 1.5 in. ports, and
that at 10 1pm, from a 5/8 in. ID port,. Each device employed a 47 mm
Nuclepore open-face holder and Nylon filter attached at the downstream
end. In addition to the seven devices, two open face Nylon filters
("control filters") operated directly attached to sampling manifold ports,
with one attached to each size port, sampling at 10 and 20 1pm, respec-
tively. Transmission efficiency was evaluated relative to the HN0#
sampled with the control filter from the'same size port. At 50% RH, HNO,
concentrations measured with the control filter at 10 1pm were about 10%
lower than those at 20 1pm, suggesting greater wall loss in the sampling
manifold port. At 80% RH, no significant difference was measured in the
HN0s concentrations at the two flow rates.
All inlet devices were precleaned with a 1-hour, hot water soak followed
by a methanol rinse and drying in particle-free air. Filter holders were
water washed and air dried as above.
Three, 6-h trials were performed at 50% RH, 20*C, and three at about 80%
RH, 13 "C to simulate day- and nighttime conditions, respectively. Second
and third trials were done immediately following flltir changes, and
reflect the effects of conditioning from preceding trials. Following the
third trial at each condition, ttje cyclones were rinsed with two, 5 ml
aliquot* of 2.7 mM HC0s -2.1mM COs ionchroaatography eluent solution
(IC eluent) to assess reeeverakility fre» "the walls. Nylon filters were
extracted in IC eluent and, together With washings, analyzed by IC for
nitrate. Supplementary experiments are described below.
Results
Transmission efficiency results for the 6-h trials ere given in Table II.
These ranged from 18% for the first trial with the stainless Steel cycldne
to 114% for the Teflon-coated glass imftactor. At 50% RH, all inlets
showed increasing transmission with increased oumulative dosage (Figure
159
-------
1) . The AlHL-design2 PFA-coated aluminum cyclone was initially quite in-
efficient, but increased rapidly in efficiency with dosage, never
equalling, however, that for the solid PTFE or Teflon-coated glass
devices. Transmission efficiencies at simulated nighttime conditions were
consistently lower than at simulated daytime conditions, excepting the
stainless steel and PFA-coated cyclones. Host showed higher efficiencies
for the initial trial (Figure 2). Greater retention of the acid by
increased amounts of water remaining in the devices is probable. At both
simulated day and nighttime conditions, the PTFE-coated glass impactor
showed the highest transmission efficiency.
Recoveries of nitrate from the filters plus washings of the devices
following HN03 trials at simulated daytime conditions were about 60% for
samplers a and b, (Table I) , and 90 to 110% for the remaining samplers.
In the simulated nighttime trials, except for sampler d (Table I),
recoveries exceeded 90%. The higher recoveries for simulated nighttime
trials may reflect diminished loss of HN0a by volatilization because of
increased water buildup in the cyclones.
Figure 3 shows nitric acid transmission efficiencies for the first 6-h
trial at 50ft RH against calculated residence times within each device
(including additional plumbing described In Table 1). An excellent
negative correlation was observed regardless of wall composition or
design. Devices with residence times <0.2 sec showed > 60S transmission
efficiency without previous conditioning.
The AIHL-design PTFE and PFA-coated cyclones are nearly identical except
for wall composition, To permit a direct comparison, both units were
operated at 20 1pm without additional plumbing (i.e. with filter holders
containing Nylon filters attached directly to the exit ports of each
cyclone). Two control filters sampled in parallel. Figure 4 shows
results for three successive trials at simulated daytime conditions.
Dosages for each trial were about half those In the preceding experiment
at 50% RH Even without the additional glassware, the PFA-coated cyclone
showed consistently lower transmission of nitric acid. Little or no
conditioning effect was evident for this cyclone, whereas a small effect
can be seen with the PTFS unit. A decreased conditioning effect is
consistent with the lesser dosages employed.
The reproducibility of the behavior of the PFA-coated AlHL design cyclone
was assessed by sampling HN0s in parallel with two such cyclones. Table
III shows the results for this comparison, indicating relatively little
difference in transmission efficiency between the two units. The
agreement observed indicates that damage or imperfections in the PFA
coating is unlikely to be influencing HN0a transmission.
The relatively low efficiency for transmission of HN0s observed with the
PFA*coated cyclone can be contrasted with that following conditioning by
atmospheric sampling. Following the NSMCS, this cyclone and its attached
glassware were evaluated for transmission of laboratory-generated HN0a
without precleanlng. Experimental techniques used were the same as those
described above. The nitric acid concentration, 15.6 ± 2.0
measured with control filters, was similar to that employed in the present
work. However, sampling periods for each of thrfee trials were two rather
than six hours, providing mean dosages for each trial of 36 ftg. For three
trials, the mean nitric acid measured sampling through the cyclone and
glassware was 15.2 ± 1.5 pg/m®. Thus, the average transmission efficiency
was not significantly different from 100%. In addition, comparison of
NSMCS atmospheric nitric acid results obtained with the denuder difference
method*, in which both units were preceded by PFA-coated AlHL cyclones,
showed, on average, daytime nitric acid results which were About 15%
higher than those by Fourier transform infra-red spectroscopy4 ,s.
160
-------
Accordingly, the atmospheric nitric acid results as well as laboratory
trials vith the ambient air pre-conditioned PFA-coated cyclone are in
marked contrast to the present laboratory findings.
Conclusions
1. Transmission efficiencies for nitric acid through initially clean
inlet devices can vary greatly,
2. Residence time and surface composition substantially influence nitric
acid transmission efficiency through inlet devices. Residence times
<0.2 sec minimize HN0S loss.
3. Atmospheric sampling provides more effective conditioning of inlets
for HNOj transmission than does sampling of laboratory-generated HNO,,
4. Solid PTFE or PTFE-coated glass inlet devices are preferable to PFA-
coated aluminum for HN0S sampling.
References
1. S. V. Hering, et al, "The nitric acid shootout: Field comparison of
measurement methods," Submitted for publication in Atmospheric
Environment (1987),
2. V. John and G. Reischl, "A cyclone for size-selective sampling of
ambient air,* . Asgftfl¦ (1980).
3. B. R. Appel, Y. Tokiwa and M. Haik, "Sampling of nitrates in ambient
air," Atmos- Environ. 1£:283 (1980).
4. E. C. Tuazon and A. M. Winer, Private communication (1986),
5. B. R. Appel, Y. Toklwa, E. L. Kothny, S. Wu and V. Povard, "Evaluation
of procedures for measuring atmospheric nitric acid and ammonia,"
Accepted for publication In Atmospheric Environment (1986),
161
-------
Table I. Inlet Devices for Nitric Acid Transmission Efficiency
Device Code
Ism
Device
volume, ml Santpling.RatiB,
a
AIHL-design PFA-coated
aluminum cyclone
475
(with glassware)*
28**
b Andersen stainless steel (with ss extension)! 20
c AIHL-design PTFE cyclone
d RTI PTFE cyclone
e PTFE-coated glass cycloneft
f PTFE-coated glass impactorH
g U. Calgary PTFE cyclone
83
20
60
15.6
75
16
35
16
62
10
(EPA design)
* Volume without glassware is 85 ml.
** The PFA-coated AIHL cyclone was operated at 28 1pm but the nylon
filter sampler downstream sampled at 20 1pm.
f Volume without stainless steel extension is 85 ml.
¦ff University Research Glassware, Carboro, N.C.
162
-------
Table II, Efficiency of Transmission of Nitric Acid Through Inlet Devices
(%)*
Device
Trials at
50% RH. 21'C
Trials at
80% RH. 13*C
Code
A
8
c
A
B
C
a
28.0
54.3
74.1
52.4
46.9
43.4
b
18.0
26.5
38.4
65.4
31.8
27.1
c
85.1
92.7
105
74,8
66.1
63,3
d
81.3
90.5
103
62,1
65.1
59,7
e
91.3
95.5
105
65.0
68.8
59.5
£
104
102
114
81.2
80.5
75. i
g
74.3
88.4
91.9
65.6
58.4
53.2
HHOa-£pataat.tAtlSM. Ug/m!,)
20 lpm:
9.90
11.0
11,2
6.22
5.48!
4,17
10 lpm:
9.07
9,66
11.1
(7.98)** 5.45
4.17
* Conditions: 6-h trials
sampling rates per Table I
* Suspect result, not used for calculation.
Table III. Comparison of Two ,P FA'Coated AlHL Design Cyclones for HNO.
Transmission Efficiency (%)
Devlcff/Trlal
, &
c
l
35.7
40.3
47.9
2
30.9
47.0
46.3
HNO3 Cone, {ftg/m*)
7.f t 0.S
4.6 t 0.4
7.0 ± 0.05
a. Conditions: 3 hr sampling
20 lpm
20 *C
RH Range 53%-64%
m
-------
NITRIC ACID TRANSMISSION EFFICIENCY
50% R.H.
120-
90.0-
EFFICIENCY
(%)
60.0-
30.0-
0.00 --j— i 1 1 1 i [ r (
0.00 65.0 130 195 260
CUMULATIVE DOSAGE (>ig)
¦ PFA-COATED AIHL CYCLONE (ALUMINUM)
© STAINLESS STEEL (SIERRA-ANDERSEN)
O AIHL CYCLONE (PTFE)
~ RES. TRIANGLE INST. PTFE CYCLONE
~ EPA PTFE-COATED GLASS CYCLONE
O EPA PTFE-COATED GLASS IMPACTOR
A U. CALGARY PTFE CYCLONE
325
Figure 1
164
-------
NITRIC ACID TRANSMISSION EFFICIENCY
80% R.H.
100-
75.0-
EFFICIENCY -
(%)
50.0-
25.0-
0.00-
0.00 24.0 48.0 72.0 96.0
CUMULATIVE DOSAGE Oig)
¦ PFA-COATED AIHL CYCLONE (ALUMINUM)
© STAINLESS STEEL (SIERRA-ANDERSEN)
O AIHL CYCLONE (PTFE)
~ RES. TRIANGLE INST. PTFE CYCLONE
~ EPA PTFE-COATED GLASS CYCLONE
O EPA PTFE-COATED GLASS IMPACTOR
A U. CALGARY PTFE CYCLONE
120
Figure 2
163
-------
hno3 TRANSMISSION vs. residence time
50% R.H.
FIRST TRIAL
120-
90.0-
EFFICIENCY j
(%)
60.0"
30.0-
o.oo-^
0.00 0.20 0.40 0.60 0.80 1.00
RESIDENCE TIME (SECONDS)
110 - 96.2X
= 0.94
Figure 3
166
-------
NITRIC ACID TRANSMISSION EFFICIENCY
50% R.H.
100-
75.0-
EFFICIENCY
(%)
50.0-
25.0-
0.00 30.0 60.0 90.0 120 150
CUMULATIVE DOSAGE Qig)
Figure 4
PFA/Aluminum
167
-------
A STUDY OF THE PERFORMANCE OF ANNULAR DENUDERS AND PRESEPARATORS
by
Teri L, Vossler, Robert K. Stevens, and Ralph E. Baumgardner,
USEPA/ASRL, RTP, NC, 27711
INTRODUCTION
For the past five years a number of investigators have been involved in
developing reliable sampling and analysis procedures to measure ambient
concentrations of HNO , SO , nitrates and sulfates, which contribute to acidic
deposition. Filter packs consisting of an inert filter followed by a treated
filter appeared to provide reliable data for sulfates and SO . But measurement
of HNO and nitrate has proven to be more difficult due to losses of HNO in
the sampling system (e.g., inlet) and difficulties in differentiating vapor
phase HNO from UNO produced from the dissociation of NH NO during sampling,
Denuder technology ^ias been shown to circumvent this problem (1-3). In the
California Air Resources Board sponsored Nitrogen Species Comparison Study, a
variety of techniques, including those using denuder technology, were
intercompared. One important finding of this intermethod study was the
apparent loss of acidic gases (e.g., HN03 and SO^) in the Teflon inlet
assemblies used by some of the participants (4). This report describes recent
results from our efforts to design inlets to quantitatively transport HNO and
SO to annular denuder tubes for their collection, while simultaneously
removing coarse particles ( > 2.5 /im) from the air stream. Comparisons of
results obtained with various types of inlets when coupled with annular
denuders to measure S02, HNOa, HNO^, nitrates and sulfates will be discussed.
EXPERIMENTAL
Sampling: Samples were collected using two annular denuder systems (ADS) run
simultaneously. Various combinations of preseparators were employed for each
sampling run to remove coarse particles. Preseparators tested were a Teflon-
coated glass cyclone, Teflon-coated glass impactors and quartz impactors with
no coating. These preseparators, along with the denuder tubes, were fabricated
by University Research Glassware Co., Carrboro, NC. Each preseparator preceded
two annular denuder tubes in series, which were for the collection of acidic
gases, namely SO , HNO^ and HN02- The annular denuder sections were separated
by a 25 mm open tnot annular) cylindrical space in order to allow the flow upon
exiting the first tube to be restored to laminar conditions before entering the
second tube. Details of the annular denuder tube design and operation have
been described previously (2). The purpose of having the second denuder is to
correct for deposition of particles and/or relatively ut\reactive gases which
would have been collected with equal efficiency on the first denuder and which
appear as the ions of interest (i.e., NO , NO , and SO ") in the IC analysis.
All denuder tube assemblies were followed by a filter pack assembly. The
filter packs consisted of a 47 nun 2 pore size Teflon filter for the
collection of fine particles, followed by a 47 mm Nylasorb (similar to Nylon)
filter for the collection of HNO^ formed from the dissociation of NH NO
particles originally on the Teflon filter. Filters were each suppoAet? by a
stainless steel screen and separated by a Teflon spacer. Teflon filters were
weighed before and after sampling using a Mettler ME22 electrobalance in a 50%
168
-------
-------
relative humidity controlled environment to obtain total fine particle mass.
The denuder tubes for the collection of acidic gases were coated with a
1% glycerine and It Na CO solution in a 1:1 mixture of methanol and distilled
water. A 10 mL portion o¥ this solution is introduced into the tube, which is
capped and rotated to completely coat the annular surface. The excess solution
Is then decanted and the tubes dried by allowing clean air to flow through
their.
For five sampling runs a third annular denuder tube was used to
separate out NH from the air stream. Because the diffusion coefficient of NH
is three times fiigher than that of HH03, a shorter (100 mm, vs 200 mm for HNO f
annular denuder tube could be used to obtain 99% collection efficiency of NH5 .
A solution of 2% citric acid in methanol was used to coat the tube. 3
Samples using paired annular denuder systems were collected over a
period of four months in 1986 and 19B7 on the roof of the EPA laboratory in
Research Triangle Park, NC, typically for 21.5 hours starting at 3:00 PM and
finishing at 12:30 PM the following day. Each ADS.assembly was housed in a
separate box designed to shield against precipitation. A dichotomous sampler
was run concurrently with four of the paired samples in order to evaluate the
effectiveness of the ADS inlets to remove coarse particles. The dichotomous
sampler fine particle fraction and one of the paired ADS Teflon filters were
analyzed via XRF analysis so that comparisons could be made.
Sample Extraction and Analysis: The ADS was assembled and dissassembled on a
clean laboratory work table. Nylasorb and Teflon filters were each put into a
clean, separate 100 mL polycarbonate container. Denuder tubes were extracted
with two 5 mL portions of IC solution (0.0045 M NaHCO and 0.0018 M Na CO ).
This was accomplished by capping one end of the tube,3adding 5 mL of t&e3
extraction solution, capping the other end of the tube, rotating the tube to
wet all surfaces, and then decanting the extract to a 25 mL polycarbonate
vessel, Samples were dissassembled and extracted immediately after sampling
was completed. Samples and extracts were stored at 5°C until they were
analyzed by ion chromatography for NO ", NO " and SO " through a 20001 Dlonex
IC unit. Filters were extracted ultf-asonfcally In*10 mL IC solution for 20
minutes prior to analysis. On some occaiaions when the average temperature
during sampling was £ 10°C, the IC chromatogram showed evidence of Incomplete
oxidation of the S0£ collected on the first annular denuder in the form of a
sulfite peak. However, the concentration associated with this peek was small
(< 5%) in relation to the total sulfate measured on the first denuder.
For samples which included a third denuder tube for the collection of
NH , extraction was done with distilled water following tha procedure described
above. Sample extracts were analyzed for Nil by tha method of colorimetry.
3
RESULTS AND DISCUSSION
Ion concentrations measured in the second denuder were subtracted from
ion concentrations measured In the first denuder to correct for unr®active
interfering species, as described previously. This correction averaged 10% for
NO ", 2% for SO " and 16% for NO . The net NO * N0a and SO " concentrations
were convertedVo equivalent concentrations as HNO . HNO and SO ,
respectively. Particulate SO " was obtained from tne Teflon filler extract
results. Particulate NO was* obtained from the sum of Teflon and Nylasorb
filter extract results. The paired SO results differed by an average of 3.4%.
(Percent difference is defined as half: the difference of the two ADS results
divided by the average of che two results). The duplication was not quite as
good for HNQ3 and HNO^, whose paired results differed by an average of 6,44 and
169
-------
8.3%, respectively. One anomolously high percent difference for HN02 and for
HNO are not included in these results. Generally, the high and low values
measured for SO and HNO were- not biased toward one denuder assembly or the
other, as Indicated by tlie average concentration for each denuder. This was
not the case for HNO , as one denuder assembly yielded for most samples an 8%
average higher concentration than the other. The cause of this phenomenon has
not been determined. A summary of these results is presented in Table I.
The impactors designed for and used in this experiment provided a
theoretical D50 outpoint of 1.7 /jib (5), while the cyclone was designed to have
a D50 cutpoint of 2.5 pm (6). The dichotomous sampler has a well characterized
D50 cutpoint of 2.S fim (7), The differences in cutpolnts of the impactors and
cyclone used in this experiment apparently had no effect on the HNO and SO
collection efficiency, as evidenced by the lack of bias in the paired denuder
results,
The results obtained for samples which included a third denuder for the
collection of NH were slightly better than the results given above for HNO^
and SO for the iirst 17 samples. Faired HNO concentrations differed by 5.5%
(compared with 6.3% for the previous samples), and paired SO concentrations
differed by only 1.6% (compared with 3.4% for the previous samples). HNO
concentrations agreed only to within 15%. Comparisons could also be made2for
paired particulate samples with this sub-database, as none of the filters were
set aside for XRF analysis. Fine particle sulfate and nitrate paired results
agreed quite well, to within 5%. The paired NH results agreed to within 16%,
compared with an analytical precision of 4%. It is expected that contamination
plays a greater role in NH^ measurement. A summary of these results is
presented in Table II.
The relative concentrations of gas phase HNO and particle phase NO
are governed by shifts in the equilibrium NH NO (a) - HNO (g) + NH (g) . Table
III shows the average ratio of HNO (gas) to NO " (fine particles)3as a
function of average temperature ancl relative hvraiidity for each sampling period.
For samples where the average temperature was low ( < 6°C) and the average
relative humidity was > 70%, HNO /NO " was < 1; i.e., the NH NO dissociation
constant was low. The gas-to-particle ratio was > 1 for those3low temperature
samples with an average relative humidity < 70%. This is consistent with the
relationship of Stelson and Seinfeld (8) which shows the NH^NO dissociation
constant to decrease with decreasing temperature and increasing relative
humidity. Furthermore, for higher temperature sampling periods (average
temperature > 10°C), the average HNO /NO * was > 2. The lowest values of this
ratio for > 10°C sampling periods averaged < 1 and occurred when the average
relative humidity was > 90%. The gas-to-particle ratio was only 0.4 for a
sample with average relative humidity of 98%, which is where the NHNO
dissociation constant drops off rapidly. Thus, the data presented here
qualitatively follows the relationship presented by Stelson and Seinfeld for
the dissociation of NH NO as a function of temperature and relative humidity.
However, the relationship cannot be verified quantitatively because our data
correspond to nearly a full day of sampling, which results in loss of detail.
The differences in the D50 cutpoint between the denuder inlets and the
dichotomous sampler proved to be a problem when comparing the XRF results of
the fine particle filters from the ADS and the dichotomous sampler. Elements
whose size distributions extend above or below the D50 cutpoint of either
sampler will yield different results for the two samplers. These comparisons
are presented in Table IV. Note that although the ADS inlet for the first
three samples listed had a lower cutpoint (1.7 /jm) than the dichotomous sampler
(2,5 pm), the total fine particle mass measured for those ADS samples was
significantly higher than the fine particle mass measured for the corresponding
dichotomous samples. (Gravimetric mass is measured to within a percent or
170
-------
better), One possible explanation for this result is that coarse particles are
penetrating the inlet and denuder tubes and adding to the fine particle mass.
The impactor stages were not greased for the prevention of particle bounce in
order to avoid the possibility of interfering with the collection of acidic
gases.
REFERENCES
1. R.W. Shaw, Jr., R.K. Stevens, J. Bowermaster, J.W. Tesch, E.Tew,
"Measurements of Atmospheric Nitrate and Nitric Acid: The Denuder
Difference Experiment," Atmoa. Environ. 1£:845-853 (1982),
2. M. Possanzini, A. Febo, A. Liberti, "New Design of a High-Performance
Denuder for the Sampling of Atmospheric Pollutants," Atmoa. Envlrnn
12:2605-2610 (1983).
3. M. Ferm, "A Na CO -Coated Denuder and Filter for Determination of Gaseous
HNO. and Particulate NO in the Atmosphere," Atmoa. Environ. ,£Q:1193-1201
(1986). 3
4. S.V Hering, et.al., submitted to Atmoa. Environ.
5. V.A. Harple, K.L. Rubow, "Cascade Impactor Sampling and Data Analysis,"
Theory and Design Guidelines. Ch. 4, pp 79-101, Amer. Ind. Hygiene Assoc.
Monograph Series, Lodge and Chan, eds. (1986).
6. W. John, G. Reischl, "A Cyclone for Size-Selective Sampling of Ambient
Air," JAPCA 30:872-876 (1980).
7. T.G. Dzubay, R.K. Stevens, "Ambient Air Analysis with Dichotomous Sampler
and X-Ray Fluorescence Spectrometer," Environ. Sci. Technol. £;663-668
(1975).
8. A.W, Stelson, J.H. Seinfeld, "Relative Humidity and Temperature Dependence
of Ammonium Nitrate Dissociation Constant," Atmos.Environ. l£;983-992
(1962).
171
-------
Table I - Average of Results Obtained with ADS
Species
Average
Mg/ni
Range
HNO
2
1,7
0,7 - 3.7
HNO
3
1.6
0.4 - 3.0
SO
2
8.9
0.5 - 20
NO "
3
1.2
0.3 - 2.7
SO "
4.7
0.2 - 10
Average % Difference
10.4% (8,3% w/out outlier)
8.4% (6.4% w/out outlier)
3.4%
From 17 sampling periods of about 22 "hours each between
Oct. 27, 1986 and Dec. 7, 1986.
Analytical precision ranges from <0.5% for all species
at > 1 ppm to 8% for sulfate or 2% for nitrate and nitrite
at < 0.5 ppm.
Table II - Average of Results Obtained with ADS for
Collection of Ammonia
Species
Average
f* g/m
Range
Average % Difference
NH3
0.32
0.10 - 0.57
16%
HNO
2
1.03
0.64 - 1.55
15%
HNO
3
1.37
0.93 - 1.95
5.5%
SO
2
19.67
12.6 - 32.2
1.8%
NO '
3
1.05
0.41 - 2.14
3.6%
SO "
4
2.99
2.38 - 4,68
4.6%
Fine Mass
16.2
13.1 - 21.6
2.0%
From 5 sampling periods of about 21.5 hours each between
Jan. 21,1987 and Jan. 25, 1987.
172
-------
Table III - Atmospheric Conditions and Nitric Acid
to Particulate Nitrate Ratios
Start
Avg.
Avg.
Total
Day
Temp.
RH
Precip.
HN0 /NO.
(1986)
°C
%
mm
3
10/27
15.8
3.76
10/28
11.2
81
3.05
10/29
13.2
83
--
1.08
10/30
13.7
78
--
3.10
10/31
14.6
78
2.74
11/5
16.8
92
--
1.06
11/6
15.9
83
trace
1.26
11/7
14.3
98
trace
0.40
11/16
8.9
--
--
2.57
11/17
11.0
trace
2.97
11/18
11.1
77
trace
1.39
11/19
5.6
67
7.2
1.01
11/20
5.2
88
,1.9
0.81
12/4
3.8
57
--
2.00
12/5
-0.45
65
--
1.69
12/6
-0.88
71
trace
0.64
12/7
3.5
71
trace
0.48
Table IV - Elemental Concentrations of Fine Particle
Fraction from XRF
CONCENTRATION,
ng/m3
i
DATE
SAMPLER
MASS
PB
AL
K
FE
SI
S
10/29
Denuder
33,600
37
267
174
42
241
1970
10/29
Dichot.
30,600
38
208
200
67
235
3030
11/5
Denuder
47,300
34
216
183
20
128
2850
11/5
Dichot.
43,500
43
178
229
52
230
4310
11/6
Denuder
36,000
24
219
84
29
148
2400
11/6
Dichot.
29,100
27
180
95
47
127
3690
11/18
Denuder
15,400
11
167
50
20
234
1600
11/18
Dichot.
11,100
8
133
50
11
209
1530
Impactor preseparator used vith denuder : 10/29, 11/5, 11/6
Cyclone preseparator used with denuder : 11/18
173
-------
EVALUATION OF METHODS USED TO COLLECT AIR QUALITY DATA
AT REMOTE AND RURAL SITES IN ALBERTA, CANADA.
Eric Peake and Allan H. Legge
Kananaskls Centre for Environmental Research
The University of Calgary
Calgary, Alberta, Canada T2N 1N4
As part of a major study on acidic deposition 1n the province of Alberta,
Canada (the Acid Deposition Research Program), an atmospheric sampling
system was designed to collect acidic and basic gases, and fine
particles, for subsequent chemical analysis. The Kananaskls Atmospheric
Pollutant Sampler (KAPS) consists of a series of annular denuder tubes to
collect nitric acid, nitrous acid, sulphur dioxide and ammonia and dual
filter packs to collect fine particles. Large particles are removed by a
cyclone at the Inlet. The system has been 1n operation at a remote moun-
tain site and at two rural locations In Alberta since August of 1985.
Each system Is mounted on an air quality monitoring trailer containing
the necessary vacuum pump, mass flow controllers and a switching system
which allows sequential sampling on four units or repeated sampling on
each of the units over 3 period of several days. A computer controls the
sampling program and records flow rates. Sampling units are prepared and
analyzed 1n a central laboratory.
The KAPS system has proven 1n laboratory studies and 1n field trials at
remote and rural locations to be an effective and reliable means of
collecting acidic and basic gases. Using a 12 h sampling period, atmos-
pheric concentrations of less than 0.009 ppb SOa or 0.01 yg nr» HN0» can
be collected for analysis by 1on chromatography. In the first 19 months
of continuous field operation, data recovery was over 85* with most of
the loss resulting from condensation 1n the unheated system. At a remote
mountain site, where a 48 h sampling period was used, median concentra-
tions of SOa, HN0», HNOa, and NH* were low, 0.22, 0.12, 0.03, and
0.25 ug m~», respectively. The mean SOa concentration was 0.56 v9 m*,
considerably higher than the median, whereas the mean HN0», HNOa, and
NH* values did not differ greatly from the median. Concentrations of
SOa, HNOa, and NH» at the two rural sites, which may be Influenced by
area, point source, and agricultural emissions were five to fifteen times
greater than at the remote site.
174
-------
Introduction
In 1985, a multi-year research program known as the Acid Deposition
Research Program (ADRP) was Initiated 1n the Province of Alberta, Canada.
The ADRP 1s sponsored by the Provincial Government, the Canadian
Petroleum Association, the Energy Resources Conservation Board, and the
Utilities Group, and has the objective of determining the current status
and future potential for regional scale environmental effects of acidic
deposition on ecosystems 1n Alberta. In order to meet this objective, 1t
was necessary to develop atmospheric monitoring systems capable of
measuring low concentrations of a variety of gaseous and particulate
pollutants which are known to affect air quality and ecosystems.
Commercial analyzers are available to monitor some of these pollutants,
such as SOa, NO, NOa, NH», CO, COa, and Oa, but, with the
exception of CO, COa, and 0», they are not sensitive enough to
measure concentrations found 1n areas not directly under the Influence of
urban, Industrial, or agricultural emissions. Furthermore, commercial
monitoring systems are not available to measure nitric acid, which 1s
believed to be a major contributor to acidic deposition 1n some areas of
eastern North America. A number of Investigators have utilized filter
packs and denuders to collect addle gases which are then analyzed by
conventional wet chemical procedures.»,».«.« An Improvement
1n the denuder methods was the Introduction of the annular denuder which
allows sampling at relatively high flow rates facilitating the measure-
ment of low concentrations of atmospheric gases.* This prompted an
Investigation of the annular denuder system for the collection of low
concentrations of gases at a remote location and at sites In rural areas
Influenced by a point source and by area sources. A unique package was
designed, consisting of a combination of annular denuder tubes to collect
SOa, HNOi, HNOa, and NH», and filter packs to collect fine par-
ticles for subsequent analysis by 1on chromatography, by x-ray fluore-
scence spectroscopy, and colorlmetrlcally. This paper briefly describes
the design, testing, and characteristics of the Kananaskis Atmospheric
Pollutant Sampler (KAPS) and summarizes the preliminary results and
experiences of 19 months of continual field use at three sites.
Experimental Methods
Equipment Design
The basic design of the KAPS system is shown 1n Figure 1. A Teflon
cyclone with a particle size cutpolnt of about 2.2 depending upon
the sample flow rate*, removes large particles at the Inlet.' the
flow 1s then split Into two streams each of which Is regulated by a mass
flow control unit (Tylan Corporation, Carson, California or S1«rra
Instruments, Carmel Valley, California). One stream passes Into a filter
pack which collects particles for XRF analysis, the other stream enters
three annular denuder tube; 1n series. Each tube has an inner diameter
of 1.0 cm, an arinulus width of 0,16 cm, and a length of 25 cw. The first
two tubes are coated with sodium carbonate to collect add gases and the
third tube with dtrlc add to collect ammonia. Fine particles, Nhteh
pass through the denuder tubes, are collected on a Teffon filter,
175
-------
followed by a nylon filter to collect any volatile nitrate derived from
the decomposition of particulate ammonium nitrate. The tubes and filter
packs are mounted 1r> a rigid tubular plastic case designed to withstand
high wind^, and then sealed prior to shipment of the unit to the sampling
sites. The units are mounted on a metal stand in groups of four on top
of the air quality monitoring trailers which contain flow control,
timing, and computerized flow monitoring systems as well as Instruments
for collecting related meteorological and air quality data.
Laboratory Tests
A series of laboratory experiments was carried out to determine the opti-
mum operating parameters for the KAPS system with respect to the amount
of material collected for chemical analysis and the efficiency of the
annular denuders.
It was first necessary to determine the detection limit for S0«a~,
NOa", and N0a~ by analyzing coated but unexposed denuder tubes,
determining the standard deviation 1n these analyses, and then calculat-
ing the detection limit. On five different days, a total of 24 tubes
were coated with sodium carbonate and twelve tubes coated with citric
acid. The tubes were extracted twice with distilled water prior to
analysis for anions by ion chromatography (Dionex, Sunnyvale, California,
Model 20201) or for NH.»+ by autoanalyzer (Technlcon instruments
Corporation, Tarrytown, New York).
The capacity of the tubes to collect SOa, HN0», and NH», and their col-
lection efficiency as a function of flow rate was tested. Sulphur dioxide
emitted at the rate of 1.24
-------
Upon completion of the precision studies, KAPS systems were Installed at
three air quality monitoring stations. One station is located 1n a
remote mountain site 1n the Rocky Mountains, 70 km U of Calgary, Alberta;
the second 1s located 1n an agricultural area about 1 km SE of a natural
gas processing plant, 30 km from Calgary; and the third site is 7 km W of
that, plant. These systems have been in operation since August of 1995.
The sampling rate 1s 10 Lprn with either two 12-hour sampling periods
every second day, or a repetitive 12-hour sampling period on each of
four days on a given unit.
Results and Discussion
Laboratory Studies
The mean amount of SO*'" determined by the analysis of 24 unexposed
annular denuder tubes was 0.60 u9 with a standard deviation of 0.40 pg
per tube. Given a detection limit of 4.65 times the standard deviation*,
the detection limit was 1.9 »g per tube. Based on a 12-hour sampling
period at 7 tpm, this corresponds to 0.086 ppb SO* 1n the atmosphere.
The detection limit for N0»" was 0.04 u
-------
the nitrous acid data. When NOa from a cylinder was Introduced through
a series of three tubes, the NOa concentrations, as measured with a NOx
analyzer at the Inlet of the first tube and at the outlets of the first,
second, and third tubes were 302, 189, 175, and 164 ppb, respectively.
The percentage losses of NOa were 37, 7, and 6, on the first, second,
and third tubes, respectively. These data fit the hypothesis assuming
that the source of NOa was Impure and contained about 110 ppb HNOa.
Field Trials
The precision of the method was evaluated using six KAPS units simul-
taneously sampling the ambient air at the University of Calgary. Atmos-
pheric concentrations of pollutants were low but the results presented 1n
Table 1 show reasonable precision. Using the analysis of the second tube
to correct the data from the first tube for NOa and particle depo-
sition, concentrations were: SOa, 3.13 HNOt, 0.18 ug and
HNOa, 0.71 n9 m~* In the Calgary atmosphere on the night of June 17, 1985,
and 3.10, 0.18, and 1.54 tig nr* for SOa, HNO», and HNOa, respectively, on
the night of June 18.
The KAPS system has been 1n operation at three field sites since August,
1985. Median and mean concentrations of species collected on the three
annular denuder tubes contained 1n each KAPS unit are given 1n Table 2.
The median SOa concentration collected on tube 1 at the remote Fortress
mountain site was low, 0.48 m"» when analyzed as S04»~. The mean was
much higher, 1.71 u9 ma~. When outlier values, greater than three standard
deviations from the mean were removed, the mean was reduced to 1.09 yg m~»
still well above the median showing that the data are not normally dis-
tributed. A median SOa concentration of 0.077 ppb was determined by sub-
tracting the tube 2 sulphate concentration from that of tube 1. The cor-
responding median SOa concentration at Crossfleld East, the rural site
closest to a natural gas processing plant, was 1.06 ppb and at Crossfleld
West, 7 km 1 n a predominantly upwind direction, was Q.95 ppb. Median
nitric add concentrations were 0.12, 0.15, and 0.14 ug «"*, median nit-
rous acid concentrations 0,03, 0.14, and 0.22 ug m~*, and median NH» con-
centrations 0.25, 1.14, and 1.44 «g nr* at the Fortress Mountain, Cross-
field West, and Crossfleld East sites, respectively, reflecting the
isolation of Fortress Mountain.
Data recovery over the first 19 months of operation has been greater than
85 percent. The major problems encountered with the denuder tubes were:
wet tubes which resulted 1n a 5.2% loss of data; broken tubes, 1.2X data
loss; and leaks, 2.4% data loss. The problem of wet tubes may be reduced
by a cyclone drain and by heating the system above the dew point. Con-
sidering the rugged terrain over which the units are transported to the
Fortress Mountain site, often under adverse weather conditions, the
breakage of denuder tubes was low. This 1s partly due to the thick outer
denuder tube wall, 2.9 mm, and a rigid case surrounding each unit. Other
problems Included wet or ruptured filters, power failures, computer
failure, pump failure, and problems caused by a lightning strike.
Conclusions
The KAPS system, which contains annular denuder tubes and filter packs,
has proven, 1n both laboratory tests and field applications to be an
178
-------
effective means of collecting both addle and basic gases as well as fine
particles for subsequent chemical analysis. With a detection 11m1t of
less than 0.1 pg nr* for SQa, gaseous nitrate, and gaseous nit-
rite when using a 12 h sampling period. It has the capability of collect-
ing low concentrations of pollutants found in the atmosphere at remote
and rural locations. The reaction of addle and basic gases with collec-
ted particles 1s minimized and the amount of particulate nitrate, Includ-
ing volatilized ammonium nitrate, can be determined. The units, which
are prepared and sealed 1n a central laboratory, have proven 1n field
operations to be rugged and functional 1n design.
Acknowledgements
The authors thank the sponsors of the Add Deposition Research Program
for their financial support and the staff of the Kananaskls Centre for
Environmental Research, especially G. Cross, D. Connery, B. Mottle, and
S. Swacha, for the collection and chemical analysis of samples.
References
1. T. Oklta, S. Morlmota, M. Izawa, S. Konno, "Measurement of gaseous and
particulate nitrate 1n the atmosphere", Atmos. Environ. 10: 1085
(1976).
2. R.W. Shaw, R.K. Stevens, J. Bowermaster, J.W. Tesch, E. Tew, "Measure-
ments of atmospheric nitrate and nitric acid: the denuder
difference experiment", Atmos. Environ. 16: 845 (1982).
3. J. Forest, O.J. Spandau, R.L. Tanner, L. Newman, "Determination of
atmospheric nitrate and nitric add employing a diffusion
denuder with a filter pack", Atmos. Environ. 16: 1473 (1982).
4. C.W. Splcer, J.E. Howes, T.A. Bishop, L.A. Arnold, R.K. Stevens,
"Nitric add measurement methods: an 1ntercompar1son", Atmos.
Environ. 16: 1487 (1982).
5. M. Ferm, A. Sjodin, "A sodium carbonate coated denuder for determina-
tion of nitrous add 1n the atmosphere", Atmos. Environ. 19: 979
(1985).
6. M. Possanz1n1, A. Febo, A. Hbertl, "New design of a high-performance
denuder for the sampling of atmospheric pollutants", Atmos.
Environ. 17; 2605 (1983).
7. W.B. Smith, R.R. Wilson, "A five-stage cyclone system for 1n-s1tu
sampling", Environ. Sd. Techno!. 13: 13 & 7 (1979).
8. C.J. Klrchmer, "Quality control 1n water analyses", Environ. Sd,
Technol. 17: 174A (1983).
9. B.R. Appel, V. Povard, E.L. Kothny, "Loss of nitric acid within inlet
devices for atmospheric sampling", These proceedings (1987).
179
-------
Table 1. Mean concentration and standard deviation of six replicate
samplings of the Calgary atmosphere, flow rate 7 Lpra (yg m~*).
S04a NOa" K0a~
June
n,
16:05 to June
18, 6:10
First Tube
4.95
+
0.35
0.21
t 0.06
0.79
±
0.07
Second Tube
0.21
+
0.06
0.03
± 0.02
0.08
4
0.02
June
18,
17:30 to June
19, 08:10
First Tube
4.90
±
0.2b
0.22
± 0.06
1.73
+
0.11
Second Tube
0.25
±
0.10
0.04
± 0.01
0.19
t
0.05
ISO
-------
Table 2. Median and mean concentrations of species collected by KAPS
annular denuder tubes at three locations (vg nr»).
Tube 1 Tube 2 Tube 3
S04a~ N0*~ NOa" S04a" NO»" NOa" NH*+
Fortress Mountain, n « 341
Median 0.48 0.17 0.07 0.15 0.05 0.04 0.25
Mean 1.71 0.24 0.11 0.29 O.QT 0,0b 0.31
Revised Mean 1.09 0.20 0.10 0.25 0.06 0.05 0.28
Crossfleld West, n - 396
Median 4.49 0.23 0.21 0.41 0.08 0.07 1.14
Mean 6.00 0.41 0.27 0.51 0.10 0.09 1.50
Crossfleld East, n - 397
Median 5.18 0.22 0.32 0.47 0.08 0.10 1.44
Mean 8.08 0.39 0.51 0.75 0.12 0.15 1.89
lftl
-------
n
A. INLET
B. CYCLONE
C. MANIFOLD
E. TEFLON TUBE
F. DUAL FILTER HOLDER
G. NaiCO* ANNULAR UENUUER
H. EXPANSION COUPLER
J. QUICK CONNECT FITTING
(to flow controller and pump)
M. SECOND NaiCOi ANNULAR OENUDER
N. COUPLER
0. CI1R1C ACID ANNULAR OENUDER
P. DUAL FILTER PACK
0 5 -10 15 cm
1 ,..J
Figure 1. Schematic of Kananaskls Atmospheric Pollutant Sampler (KAPS),
182
-------
COMPARISON OF THREE AEROSOL SAMPLING
TECHNIQUES AND THE DIFFERENCES IN
THE NITRATE DETERMINED BY EACH
Stanley A. Johnson and Romesh Kumar
Argonne National Laboratory
Argonne, Illinois
lEfta
vjy
Ambient aerosol samples were collected simultaneously with three different meth-
ods, one a size fractionated impactor sampler and two filter samplers. Infrared spec-
troscopy was used to analyze the collected aerosol using attenuated total internal reflec-
tion for the impactor samples and direct transmission through ultra-thin teflon mem-
brane filters for the filter samples. All samples were analyzed as soon as possible after
collection; some were reanalyzed after being stored in closed petri dishes for up to 10
days.
A major purpose of the study was to evaluate the neutralization of acidic sulfate
aerosols after sample collection by the filtration techniques. No acidic sulfate was found
in any of the samples collected during the field study. However, significant differences
were observed in the nitrate content of the samples collected by the different samplers.
Also, in several samples collected with the ATR impactor the nitrate content decreased
upon storage. In some cases the nitrate absorbance bands diminished to zero; in other
cases the nitrate initially decreased and then remained stable; in yet another case, a
high level of nitrate persisted over several days. The results indicate that two different
types of nitrate compounds were present in the samples, one more "volatile" than the
other, although the infrared spectra were consistent with both of them being ammonium
nitrate.
Introduction
Infrared (IR) spectroscopy has been used for several years for the analysis of ambi-
ent aerosols.1,2 Typically, the aerosol sample is collected by any one of several possible
sampling techniques, transferred to a suitable matrix, and analyzed by transmission,
183
-------
internal reflection, diffuse reflectance, or acoustic Fourier-transform infrared (FT-IR)
spectroscopy. Techniques have been developed recently that permit direct FT-IR anal-
ysis of samples collected by impaction on internal reflection elements,3 or samples col-
lected by filtration on thin teflon membrane filters.4 These newer approaches have the
advantage of minimal sample handling and no special sample preparation, making them
amenable to real time analysis.
The direct infrared analytical methods3'4 were compared during a field study to
evaluate the neutralization of acidic sulfate aerosols after sample collection. Samples
were also collected using a set of denuder tubes ahead of the filter, to determine the
effect on sample chemistry of removing ammonia and acidic gases from the sampled air
stream.
Experimental Methods
The three aerosol sampling methods used were: (1) the "Canadian Filter Pack"
(CFP),5 (2) the Annular Denuder System (ADS),6 and (3) the Attenuated Total Re-
flection (ATR) impactor.3 Aerosol samples were collected during the summer and fall of
1986 at the Argonne dry deposition monitoring site, ~35 km southwest of Chicago, IL.
Because of the greater sensitivity of the ATR technique compared to that of the filter
methods, a much finer time resolution is possible with the ATR impactor technique than
either of the other two techniques. Thus while samples were collected over 24 h periods
for the CFP and the ADS, consecutive 4 or 6 h samples were collected with the ATR
impactor, The CFP collects all particles in the sample air stream. The ADS removes
large particles with a cyclone preseparator and the reactive gases with three consecutive
annular denuder tubes, before the particles are collected on the teflon membrane filter.
The ATR impactor collects only 0.5-1.0 (J.m aerodynamic diameter particles.
The collected samples were analyzed with a Digilab (Cambridge, MA) Model FTS-
14 FT-IR using direct transmission through the teflon membrane filters for the CFP and
the ADS samples, and total internal reflection for the ATR impactor samples collected
on KRS-5 internal reflection elements. Infrared spectra from the filters and the ATR
impactor have been quantified for NHj and SO^~, but not yet for the NOJ. Therefore in
the discussion below, only the infrared absorbance values are used for the NOjj~ analyses.
All samples were analyzed as soon as possible after collection, typically a few hours,
and stored in closed petri dishes. Some samples were reanalyzed after storage times of
up to 10 days.
Results
No acidic sulfate was detected in any of the samples; therefore the potential neu-
tralization of acidic sulfates on filters could not be evaluated. However, some very
interesting changes were observed in the NOJ content of the samples, and these are
discussed below.
Figure 1 shows the FT-IR spectra of concurrent samples collected by the three
sampling methods. The spectra in Figure la were obtained within a few hours of
sample collection; upon reanalysis of the samples five days later, the spectra shown in
Figure lb were obtained. The three spectra in Figure la are virtually identical, i.e., NH^
(1400 cm-1), NOJ (1340 cm-1 and 840cm-1), and SO®- (1100 cm-1 and 620 cm-1); all
exhibit similar relative absorbance. Further, even though the CFP and the ADS samples
include a wide range of particle sizes while the ATR sample covers only submicrometer
184
-------
particles, the major constituents of all samples were the same, ammonium sulfate and
ammonium nitrate. Figure lb shows that after five days of storage all species were
essentially unchanged. Figures 2a and 2b show the spectra obtained immediately after
collection and 6 days later for another set of samples. In Figure 2a the CFP and the ADS
filter samples show a weals indication of NOJ in the 1340 cm-1 region, while the ATR
sample clearly shows strong NO J with both the 1340 cm-1 and 840 cm^1 absorbtion
bands. However, Figure 2b shows that the NOJ disappeared completely from the ATR
spectra after 6 days, while the weak bands persisted in the filter samples.
Nitrate was observed in one or more of the 4 h samples collected with the ATR
impactor during 11 of the 15 days of sample collection. In the CFP samples, NOJ wad
detected on only 7 days; and on the ADS filter NOJ was found on 8 days. The nitT&te
"volatility" was observed only in the ATR impactor samples; in every case where NOJ
was seen on the CFP or the ADS filters, it was persistent upon storage. The rate of
disappearance of the NOJ during sample storage was quite variable. In some cases all
of the NOJ disappeared in one day. In other cases, a rapid initial decrease was followed
by a residual NOJ that was stable with time. This could indicate nitrate from different
sources, such as aerosol present in an aged air mass along with fresh aerogol injected
by local sources. In the one case where a high NOJ persisted in the ATR sample over
several days of storage (Figure 1), the NOJ levd was substantial throughout the 24 h
sampling period.
For the ATR samples that lost NOJ on storage, a corresponding loss of NH^* was
also seen. Calculations indicated that indeed the NOJ and NH^~ losses were equivalent to
a loss of ammonium nitrate, and that the remaining NH« was in the correct proportion
with SOj~ for ammonium sulfate. It was thought that perhaps after ammonium nitrate
was lost from the sample, the residual nitrate would correspond to a different compound.
However, even in these samples enough NHj was still left to balance both the SO*" and
the NOJ. Therefore some other factor must account for why some of the ammonium
nitrate was volatile and some was more stable during storage.
The ATR impactor samples showed distinct diurnal variations, but no regular pat-
tern was apparent. Figure 3 presents spectra from six sequential 4 h samples taken
with the ATR impactor. On this day all major ions changed substantially over the 24 h
period. Note especially the fact that NOJ goes from virtually nothing to a high level
and back to almost zero. Also note that although both the SOj~ and NOJ increased
during the day, the maximum concentrations of the two ions did not occur in the same
time period.
Conclusions
This limited field study shows that samples collected by three different techniques,
one a total filter sample, another a filter sample denuded of ammonia and acidic gases,
and an impactor sample of submicrometer particles only, showed very similar infrared
absorption spectra. Finer time resolution, ATR impactor-eolleeted samples sometimes
showed the presence of 'Volatile" ammonium nitrate, which was not present In any of
the filter-collected samples. In addition to the "volatile" ammonium nitrate, a more
Btable ammonium nitrate was also often observed in these samples.
The ATR impactor samples showed significant diurnal variation in the concentra-
tions of NHj, NOJ,and SO®- from one sample to the next, which variation was lost
185
-------
in the filter samples, These data point up the desirability of conducting fine time reso-
lution sampling, and of developing real time or near real time sampling and analytical
instruments.
Acknowledgments
This work was done under the auspices of the U. S. Environmental Protection
Agency, Environmental Monitoring Systems Laboratory.
References
1. P. T. Cunningham, and S. A. Johnson, "Spectroscopic observation of acid sulfate
in atmospheric particulate samples," Science, 191: 77-79 (1976).
2. P. T. Cunningham, B. D. Holt, S. A. Johnson, D. L. Drapcho, and R. Kumar,
"Acidic aerosols: oxygen-18 studies of formation and infrared studies of occur-
rence and. neutralisation," Chapter 2 in Chemistry of particles, fogs and rain, Jack
L, Durham, Editor, Acid Precipitation Series—Volume 2, John I, Teasley, Series
Editor, Butterworth Publishers, Woburn, MA, 1&84, pp. 53-129.
3. S. A. Johnson, R. Kumar, and F. T. Cunningham, "Airborne detection of acidic
sulfate aerosol using an ATR Impactor," Aerosol Science and Technology, 2; 401-
40G {1983].
4. W. A- McClenny, J, W. Childera, R. Hohlt and H. A. Palmer, *FTIR transmission
spectrometry for the nondestructive determination of ammonium and sulfate in
ambient aerosols collected on teflon filtersA (mas. Environ., 19(11): 1801-1B98
(1985).
5. K. G. Anlouf, P. Fellin, H, A. Wiebe, H. 1. SchifF, G. I. Mackay, R. S. Braman,
tod R. Gilbert, "A comparison of three methods for measurement of atmospheric
nitric acid and aerosol nitrate and ammonium," Atmoa. Environ10: 325-333
(1985).
6. J. E. Sickles II, "Sampling and analytical method* development for dry deposition
monitoring", Pinal Report on EPA Contract No. 68-02-4079, March 1987, in press.
186
-------
01
o
c.
2
l~
o
c
Vji
VJUwL
2400
... 2400T
Frequency, cm
Figure 1. Infrared spectra of samples collected concurrently on
Nov. 5-6, 1986.
A
Figure 2.
400 2400 , 1400 400
Frequency, cm"
Infrared spectra of samples collected concurrently on
Oct. 14-15, 1986.
Comparison of three sampling methods; (a) spectra recorded soon
after collection, (b) spectra recorded from the same sample six
days later. Arrows Indicate the position of the 1340 crn-1
nitrate absorbtlon band.
187
-------
2400
1900
1400
900
400
Frequency, cm"1
Figure 3, Infrared spectra of six sequential samples
collected with the ATR-impactor on
October 27-28, 1986.
Times of collection were: (a) 1330-1730,
(b) 1730-2130, (c) 2130-0130, (d) 0130-0530,
(e) 0530-0930, (f) 0930-1330.
188
-------
Wet Deposition on Forest Canopy at
Mt. Mitchell, North Carolina
Vinod K. Saxena and Ronald E. Stogner
Department of Marine, Earth and Atmospheric Sciences
North Carolina State University
Raleigh, North Carolina
One of the most recent examples of the severe environmental
impact of global and regional air pollution has been identified
to be the forest decline at high elevations in Europe and North
America. First identified as "waldsterben" (death of forest) in
West Germany and now commonly accepted as forest decline has
been noticeable in eastern U. S. beginning 1980. Alarmed by the
West German plight, the U. S. Environmental Protection Agency
launched a project entitled, "Mountain Cloud Chemistry/Forest
Exposure Study" in May, 1985. Under the program, relevant at-
mospheric chemistry and micrometeorological measurements have
already started or are underway at the following locations:
three sites at Whiteface Mountain (New York), two sites at Hub-
bard Brook/Mount Moosilauke Experimental Forest (New Hampshire),
three sites in Mt. Mitchell State Park (North Carolina), three
sites in Shenondoah Park (Virginia), one site at Whitetop Moun-
tain (Southwestern Virginia), and one site in Maine. During
summer, 1986, in the Mt. Mitchell State Park, we erected two
meteorological walk-up towers, one (16.5 m tall) at the Gibbs
Peak (2,006 m MSL on the mountain ridge) and the other (22 m
tall) at the Commissary Ridge (1,760 m MSL on the eastern
slope). Although the southeastern U. S. experienced drought
during the summer, 1986, the Gibbs' Peak tower was exposed to
clouds during more than 70% of days. The direct cloud intercep-
tion by forest canopy is a major mechanism for acidic deposition
at Mt. Mitchell and hydrological and chemical inputs are
governed by sedimentation, impaction and evaporation processes
which, in term, are controlled by the prevailing windfleld and
cloud microstructure.
* The research was supported by U. S. EPA Contract No. CR
812444-02-0 and W. S. Fleming Contract No. 0-31734. Neither of
these organizations have reviewed the paper. The authors are
solely responsible for its contents.
189
-------
Results from initial runs of a direct cloud interception
model indicate that wet deposition from such sources constitute
a significant source of ionic input to the forest canopy.
Values obtained compare favorably with those obtained by other
authors. Our calculations indicate that such sources contribute
on the order of 1 .75 kg ha-1 yr~" of H* ion deposition to Mt.
Mitchell area forest. These numbers are even more significant,
considering the drought conditions during the summer, 1986. Our
results also indicate that short term cloud events (less than
1/3 of a day) may contribute high amounts of ion deposition due
to their high acidity and frequent occurrence.
Introduction
Forest decline at high elevation ecosystems has become in-
creasingly apparent during recent years. Such decline is par-
ticularly evident in mountain spruce-fir ecosystems, which often
remain immersed in clouds for prolonged periods of time. Mt.
Mitchell (2,038 m MSL, 35° 46'N, 82° 16'W) area forest, for ex-
ample are exposed to clouds about 258 days per year on the
average . This prolonged exposure to acidic clouds, may greatly
enhance wet deposition of ionic substances at such sites.
One type of wet deposition, which is parculiar to such
mountain sites, is direct interception of impinging cloud
droplets. Wet deposition amounts, on these mountain area forest
canopies, may be significantly under estimated, if such sources
are not taken into consideration2. Direct measurements of wet
deposition to forest canopies, through techniques such as:
stemflow, throughfall and the use of surrogate surfaces, are
difficult to accurately obtain as well as relate to the complex
forest canopy. A more general and easily applied method is the
use of deposition models. A model to calculate wet deposition,
due to direct cloud droplet interception has been developed and
is currently in use by the Mountain Cloud Chemistry/ Forest Ex-
posure program (MCCP). The model is described in detail by
Lovett. We have used this model to calculate cloud water
deposition rates at Mt. Mitchell.
The model requires wind speed, temperature, relative
humidity, solar radiation, liquid water content and droplet
modal size. All parameters were obtained during the 1986
summer, as part of the MCCP program. The measurements were made
at the Mt. Gibbs site. This paper presents the results from
model runs using the 1986 summer data obtained at Mt. Gibbs.
Experimental Methods
Two research sites were established in the Black Mountains
of North Carolina. Each site is equipped with an aluminum walk-
up tower, which extends about 10 m above the surrounding forest
canopy. The first site is located at Mt. Gibbs. The second
site is located at Commissary Ridge, about 0.75 km southeast of
Mt. Mitchell and some 200 m below the Gibbs summit site at an
elevation of 1,760 m MSL. Table 1 list parameters measured
during the 1986 summer at Mt. Gibbs.
190
-------
Table 1
Parameter
Instruments
Temperature, Relative Humidity
Wind Speed, Wind Direction,
Solar Radiation, Pressure,
Precipitation
Electronic Weather
Station (Operated
continuously)
Cloud Water Chemistry
Active and Passive
Cloud Water Collectors
Precipitation Chemistry
Ozone, SO2 Gas Concentrations
NADP Collector (weekly)
Continuous Gas Monitors
Droplet Size Distribution,
Liquid Water Content
Forward Scattering
Spectrometer Probe(FSSP)
Model Methodology
Meteorological parameters needed for model runs were ob-
tained from data recorded by the electronic weather station
(EWS). One hour averages were used for the model runs. Two im-
portant deviations need to be mentioned here. First, the model
uses values at the canopy-top (about 8 m, in our case), while
our measurements were made at 16.5 m above the ground. Since we
had no real-time turbulence data, to estimate vertical profiles,
we have chosen to use the 16.5 m values directly here.
Secondly, Lovett et. al.2 define Net Deposition as Gross Deposi-
tion (that is all water captured) minus Evaporation. Our ex-
perience has shown that the model's evaporation scheme is ex-
tremely sensitive to relative humidity measurements in the 95-
100% range. Presently, there are no tested, field worthy in-
struments which are capable of accurate measurements of relative
humidity in this range. We have, therefore, chosen to calculate
Gross Deposition, neglecting evaporation. While both of these
points will combine to over estimate the actual deposition,
there are other factors, which will be discussed later, which
will combine to decrease the estimates. The resulting values
may not be far from the true deposition.
The microphysical input parameters needed were obtained
from five minute averages of one second FSSP data. These five
minute averages were used to compute the one hour values that
were used in model runs. Liquid water content (LWC) values were
obtained from Integration of the droplet size distribution data.
The LWC values from our FSSP data compare well with LWC values
obtained by other methods at other MCCP sites.
Ionic deposition rates were obtained by multiplying H+ con-
centration values from cloud water samples collected hourly
using the cloud water collectors listed in Table 1. Data from
the 1986 summer show that there is no statistical difference in
the chemistry of samples (333 hours were sampled) collected from
either the passive or active collectors. We have used samples
from the passive collector. Field H+ concentrations were used
in our computations.
191
-------
Results and Discussion
Droplet size distribution measurements were successfully
obtained during three cloud events of 1986 summer. These events
occurred on the following dates: July 5, August 30 - September 1
and September 10 - September 12. The July 5 event was of short
duration lasting only about five hours. The two remaining
events were long events, lasting more than thirty six hours. We
have used the July and August events for our estimates. The
July event is typical of most short events at Mt. Mitchell in
both physical and chemical composition. The August event is
typical of long term events, physically, however it is about an
order of magnitude less acidic than most long events.
Figure 1 shows microphysical characteristics of the two
events as recorded by the FSSP. The LWC, total droplet con-
centration and average droplet size are shown for the two
events. The July event was a valley fog event, probably en-
hanced by orographic lifting. The event occurred in the early
morning hours and lasted until about noon. The event became ex-
tremely patchy shortly after sunrise, with official cloud water
samples being discontinued at 06 30 EST. A sharp inversion was
evident at sunrise, which completely dissipated by noon. Field
pH measurements for the event ranged from 3.17 at the outset to
2.99 at the end of the event. This drop in pH was probably due
to the dissipation of fog droplets, thereby concentrating their
solution contents.
The August event lasted more than forty eight hours and was
associated with a weak upper level disturbance. The event
yielded light precipitation during the latter stages of the
event. The profiles (Fig. 1) for this period indicate a drop in
all three parameters, this we believe can be attributed to a
shift to a size range larger than the effective cut off diameter
for the Probe. Although the probe is capable of measuring
droplets up to 47 um onboard an aircraft, we suspect the actual
cut-off diameter is smaller (probably about 25 um) when used for
ground based measurements due to lower sampling velocities.
However, no laboratory tests are yet available to support this
contention. The probe, probably will under-estimate the actual
LWC in mature cloud systems. The above factor, coupled with the
lower acidity of the August event will result in lower ionic
deposition estimates for our long duration events. This in com-
bination with the fact that the 1986 summer was one of extreme
drought for the southeast region will tend to cause the es-
timates of normal ionic deposition in the Mt. Mitchell area to
be underestimated. This will be offset somewhat by the factors
(previously mentioned) which tend to over-estimate deposition.
We feel, therefore that, the resulting errors in our estimates
are small.
Wet deposition was calculated by one hour model runs using
the data mentioned above. The data could then be combined with
field chemical data to produce hourly ionic deposition rates.
Finally, knowing the total number of hours of each type of cloud
event that occurred during the summer (May - September) and the
average ionic deposition from these two representative events,
we were able to estimate annual ionic deposition to the Mt. Mit-
192
-------
0 90
~OM
"e
N
° 0.40
I
020
August 30 - September 1, 1906
Liquid Water Content vs, Time
Julys, 1986
liquid Water Content vs. Time
OXKH
MJG 30
OM
0.20
*>
e
X.
e»
O.ISi
¦w
%
0.10-
0.05-
0.00-
AUO 31 SEP 01
Ttm» (EST)
SEP 02
<3 200-
AUO BO
AUO 31 SEP 01
Ttm» (EST)
SEP 02
August 30 - September 1, 1906
Total Concentration vs. Time
AUO 31 SEPOI
Timt (EST)
August 30 - September 1, i960
Average Droplet Size vs. Time
390
~ 300
K)
& 290
8 200
^ 150
3 100
SO
0
10
e
i 4
l41
2
0200 0900
Tim (E3T)
oeoo
July 5, 1906
Total Concentration vs. Time
0200
0900
Tim* (EST)
oeoo
July 5, 1906
Average Droplet Si2e vs. Time
0200 0900
Tim* (EST)
oeoo
Figure 1 Microphysical parameters measured at
Mt. Gibbs, NC. Values shown are five
minute averages of one second FSSP
data.
193
-------
chell area ecosystems. Due to the draught conditions, it is
likely that these annual estimates are somewhat low.
Model runs for the July event indicate hourly wet deposi-
tion rates on the order of 0.13 mm hr from direct cloud inter-
ception and ionic deposition rates on the order of 10 kg ha
hr from such sources. Runs for the August event indicate
hourly wet deposition rates on the order of 0.27 mm hr and
ionic deposition on the order 10 kg ha from direct inter-
ception sources. Extrapolating cloud occurrence data taken from
the summer to the entire year, yields annual ionic deposition
rates of 1.35 kg ha yr-1 for short term events and 0.401kg ha
yr_1 for long term events. The total of 1.75 kg ha yr
compares well with that of Lovett et. al. who provided a value
of 2.4 kg ha yr~ . These values are also obtained from actual
measurements, while Lovett et. al. assumed a typical LWC and
modal droplet size.
Concluding Remarks
Our results indicate that wet deposition from direct inter-
ception sources is a significant source of ionic input to moun-
tain forest canopies. Values estimated from measured parameters
obtained near Mt. Mitchell, N.C. compare favorably with those
obtained by other authors at other locations.
Wet deposition rates from such sources were found to be on
the order of a few tenths of mm hr-^, with resulting ionic
deposition rates on the order of 10 kg ha hr-1 for both
short and long duration events. However, due to the more acidic
nature and more frequent occurrence of short duration events,
these short events may contribute more ionic deposition per year
than do their long duration counterparts. Finally, due to ex-
treme drought conditions which existed during the 1986 summer in
the southeast region, the estimates presented here may actually
be lower than normal for Mt. Mitchell. Further work is in
progress to refine our computations.
References
1. V. K. Saxena, E. B. Cowling, R. E. Stogner, R. V. Crume,
"Cloud chemistry measurements at Mt. Mitchell, North
Carolina", Preprints Conf. on Cloud Physics. AMS,
Boston: 91-94. (1986) .
2. G. M. Lovett, W. A. Reiners, R. K, Olson. "Cloud droplet
deposition in subalpine balsam fir forests; hydrological
and chemical inputs", Science 218: 1303-1304. (1982).
3. G. M. Lovett, "Forest structure and atmospheric
interactions: predictive models for subalpine balsam
fir forest", Phd. Thesis, Dartmouth College: 225 pp.
(1980).
194
-------
DRY DEPOSITION OF OXIDES OF NITROGEN
D. Stockec, D. H. Stedman, B. Evilsizor and M. Burkhardt
Chemistry Department
University of Denver
University Park
Denver, Colorado 80208
Eddy Correlation is the best approximation available as
an absolute method for the determination of dry deposition,
in order to perform an eddy correlation study of the flux of
a pollutant gas, that gas needs to be measured with a time
response of less than one second. We have shown that we ace
capable of making these measurements for ozone and for the
oxides of nitrogen MO and NO,. In order to be sure to be
determining all the important flux of the oxides of nitrogen
out of the atmosphere, it is important to be able to
determine the flux of nitric acid (HNO3). We therefore
propose to build a molybdenum converter which will allow
eddy correlation studies of nitric acid.
Determining the concentration of peroxyacetyl nitrate
has always been difficult. We have developed a fast
portable method which can separate PAN from interferences in
ambient air and monitor its concentration.
Key words: Oxides of Nitrogen, NO, N02, HNO3, PAN, 03,
luminol, molybdenum and Eosin-Y.
195
-------
The dry deposition of many atmospheric species to
various types of ground cover has not been characterized
(1). The equation:
"f * c1 w1
forms the basis for the eddy correlation technique. Average
fluxes are measured directly by correlating the fluctuations
in concentation, c1, with the fluctuations in the vertical
wind, w', without the need for assumptions involving eddy
diffusivities or other inferred quantities (2). This method
requires state of the art fast response instrumentation (> 1
Hz) (3,4) to be deployed in the field along with computers
for data acquisition and reduction. A micro-
meteor o1ogica11y suitable site is required for eddy
correlation studies. Ideally the site should be level, with
a uniform surface (no upwind obstacles which could introduce
a non steady-state turbulent flow). The atmosphere should
also be well mixed so that chemical and dynamic steady-state
prevails.
Meteorological instruments
The (u,v,w) wind components are measured using a Gill
propel lor anemometer mounted at a height of 5-7m on a tower
upwind of the trailer which houses the instruments. Air
samples are drawn from the tower into the trailer using a
high volume sample pump (at a flow rate of ca. 180 1pm)
through 5/8 inch i.d. Teflon tubing, the chemical sensors
draw samples from the main air stream. The time taken for a
sample to travel the length of the tube and reach the
analyzers is < 2 s. This lag is taken into account in the
data analysis. The response frequency of the chemical
sensors sampling through the intake tube is approximately 1
Hz.
The response frequency of the Gill anemometer is also
about 1 Hz, depending on the wind speed 15). The flux
measurements thus sample the frequency spectrum from 1 Hz up
to the averaging time of 25 minutes. At typical wind
speeds, the loss of flux by neglect of longer scale eddies
is probably insignificant. The loss of eddies faster than 1
Hz is predicted to lead to an underestimate of flux not
exceeding 15% (5). This correction can be included in the
data analysis.
A platinum wire thermometer is mounted on the tower in
close proximity to the sample intake. Its response
frequency is about 10 Hz,
«2x
The N0X (NO, NO, and PAN) insturment monitors the
intensity of light eiranftted following the reaction of nitric
oxide with ozone (6,7). The higher oxides of nitrogen
(principally nitrogen dioxide and peroxyacetyl nitrate) are
196
-------
reduced to nitric oxide using a heated molybdenum converter
at 400° C, see Figure 1. The short length of tubing in
front of our convertor has such an affinity for HNO3 that
the response to HNOj changes is over a minute. Using a
radiative heating system, we propose to build a molybdenum
convertor in which 400° C molybdenum is the first surface
which is impacted by the sample air, see Figure 2. This
configuration will allow eddy correlation studies of HNO3.
Chemiluminescent Ambient Air Monitors
By altering the configuration of the Chemiluminescent
Ambient Air Monitor (CAAM), see Figures 3, 4and 5, the
following species can be monitored: 1)n<>2 2)PAN 3)HN03
and 4)03. The first three species are measured by the
chemiluminescence reaction of NO2 with basic luminol
solution (8). Detection of NO2 takes place in the analyzer
cell where luminol solution and ambient air come into
contact. Light produced from the reaction of NOn and
luminol is detected by an EMI 9924B (or Hamamatsu r268)
photomu1tiplier tube (PMT). The photocurrent is converted
by the electrometer to a 0 to 10 volt output signal that is
linearily proportional to the ,NO, concentration. The
response of the chemiluminescent monftors is about 7 Hz.
Ozone is measured by using the chemiluminescent dye
Eosin-Y in ethylene glycol in place of luminol (9).
PAN and HNO3 instruments require special intake systems
and operate at slower response rates. Eddy Correlation
studies are not possible at these rates but the gradient
method could be used to determine the flux of these species.
The concentration of PAN can be found by coupling a CAAM
instrument with a gas chromatograph. PAN and NO? are
seperated using a polyethylene glycol 400 on chromosorb W HP
column with N0„ free ambient air as the carrier gas. The
peroxyacetyl nitrate can be separated and detected once
every 15 seconds. The HNO3/NO5 analyzer is a variation of
the luminol chemiluminescence aetector for NOj, A species-
determinator intake system consisting of a hot-glass beads
converter, a CrO? converter, and filters of nylon and teflon
is used for HNO3 measurement. An automatic timing and
valve sequencing circuit controls the intake air flow that
alternates between readings for NOy(NO, N02, PAN and HNO3)
and for NOy-HNO,. During operation the teflon filter
removes particulate matter from the sample air, HNO3 in the
air sample is then converted to NOj in the hot-glass beads
trap. Excessive reduction is corrected by use of a CrOj
converter just prior to the NO, detector, A valve
periodically switchs a nylon filter inline which selectively
removes HNO3 from the air sample. The value for the HNO3 is
obtained as a difference between the two measurements.
197
-------
References
1. H, Rhode, A. Eliassen, I. Isaksen, F. B, smith, D. M.
Whelpdale, "Tropospheric Chemistry and Air Polution",
lil' 176 (1982).
2. B. B. Hicks, M. L, Wesely, R, L. Coulter, R. l. Hart, J.
L, Durham, R. E. Speer, D. H. Stedman, Bound.-Layer
Meteorol. 34, 103-121 (1986).
3. J, A. Businger, "Evaluation of the Accuracy with Which
Dry Deposition Can Be Measured with Current
Micrometeorological Techniques", J. Climate and Appl.
Meteorol. 25, 8, 1100-1124 (1986).
4. J. A. Eastman, Thesis "Eddy Correlation Measurements of
the Resistance to Vertical Transport of Ozone" (1983).
5. C. J, Moore, Bound.-Layer Meteorol. 37, 17 (1986).
6. A. Fontijn, A. J. Sabadell, R. J. Ronco, Anal. Chem. 42,
575 (1970).
7. B. A. Ridley Atmos. Technol. 9, 27 (1978).
8. G. J. wendel, D. H. Stedman, C. A. Cantrell Anal. Chem.
5_5, 937 (1983).
9. J. D. Ray, D. H. Stedman, G. J. Wendel Anal. Chem.
58,598 (1986).
198
-------
corona discharge
ozone generator
dry oxygen.
limiting orifice
Q-W
iniet
ysten
reaction
chamber
/
•high voltage
•photomultiplier
tube
teflon filter
I
display
system'control
data acquisition
I ozone scrubber v-
. I- | * chart record
vacuum pumps
exhaust
Figure 1 Block diagram of the N0X detector.
L
to zero
gat
supply
IV
)
P7!
To air pump
aspirating Intake
¦£
. . Mo tubing
"XkboocOI Intake and
, flow nstrlctor
rc ^
j Transition a
sadtant tip heater
Transition zone heater
, j^,Convtnt1onil Mo tunings converter
^.Conventional Mo tunings heater
-t*-'
Exit to NO detector
Figure 2 A schematic of the FAST CONVERTER inlet system
for HN03 studies using the NOx detector. The rotary valve
at the left controls whether the intake is pulling in
ambient air or pushing out zero air.
199
-------
air in
housing
W&.
PMT
224
L U&S2
¥
IL.,u
fluid in
Figure 3 Design of the CAAM detection cell that
Eosin-Y in ethylene glycol or luroinol solution.
air in
detector
electrometer
PMT
cell
recorder
air out
Cluid tank
Figure 4 Block diagram for the N02 analyzer.
hot beads
nylon filter
NO,
[detector
converter 400°C
solenoid
valve
teflon
f11ter
air
1 n
Figuce 5 Inlet system foe HN03 detection.
200
-------
Establishment of the National Dry Deposition Network
by
Barry £. Martin & Jack C. Suggs
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S, ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
APRIL 1987
201
-------
Establishment of the National Dry Deposition Network
Introduction
In an effort to coordinate Federal Acid Rain activities, Congress
passed the Acid Precipitation Act of 1980. This Act established a Task
Force to develop and implement a comprehensive National Acid Precipitation
Assessment Program (NAPAP) that would increase our understanding of the
causes and effects of acid deposition. The program includes both research
and monitoring data bases on which administrative decisions can be made.
To fulfill this latter requirement, NAPAP established the National Trends
Network (NTN). Since the NTN measures only the wet component of acid
deposition, the NAPAP assigned EPA the responsibility to design, deploy,
and operate a network to measure the dry components of acid deposition,
EPA delegated this responsibility to the Environmental Monitoring Branch
(EMB), Monitoring and Assessment Division (MAD), Environmental Monitoring
Systems Laboratory (EHSL), Research Triangle park (RTP), NC. The EM8/MAD/
EMSL has designed a network which will consist of up to 100 stations and
will be established over the next 5 years. This network is being estab-
lished to address the following objectives: establish a data base from
which an estimate of dry deposition flux and deposition velocities can be
made; determine the temporal and spatial distribution of dry deposition in
the U.S.; assist in determining the relationship between dry deposition,
wet deposition, and environmental effects; and collect data to test and
improve regional deposition models.
Since there is no reliable mechanism to monitor dry deposition veloci-
ties or dry deposition fluxes, the network will monitor the following
parameters from which dry deposition velocities and fluxes can be estimat-
ed: particulate sulfate {S0a"z), particulate nitrate (NO3"), nitric acid
(HNO3), sulfur dioxide (SO2T, ozone (03), wind speed (WSJ, wind direction
(kfD), temperature (T), delta temperature (aT), solar radiation (SR}, rela-
tive humidity (RH), and precipitation (P). Because of the uncertainties
surrounding some of the sampling techniques and because of the complexity
of the project, a prototype network of six stations was established by EMSL
in 1984. These six stations evaluated the performance of monitoring equip-
ment under expected operating conditions. The experience gained in this
effort has guided EMSL in the final equipment selection and will be used to
refine the station operating procedures. These six stattons have become
part of the 100 station network. The prototype stations are located at
Research Triangle Park, NC; Pennsylvania State University, PA; Oak Ridge,
TN; Whiteface Mt., NY; and West Point, NY. A sixth station is also located
at West Point, NY as a collocated station. These sites were chosen because
of existing monitoring activities for acid deposition at each site and the
Interest in acid deposition of the principal Investigators at the stations.
The stations in the prototype network were Installed and maintained by a
contractor. The operation of the station is by agreement with the princi-
pal investigators at the stations.
A contract was awarded in September 1986 to Environmental Science &
Engineering, Inc. (ES&E) of Gainesville, Florida to assume operation of the
prototype network and to expand the network to 100 stations over the next
five years. The parameters to be monitored are the same as 1n the proto-
type network. Figure 1 shows the proposed locations of the dry deposition
monitoring stations. Due to the need to evaluate the Regfnal Acid Deposi-
tion Model (RADM) by 1989, the first 40 stations will be located in the
eastern United States. The remaining 60 stations will be distributed over
the contiguous United States.
202
-------
Discussion
in the prototype network, the particulate sulfate and nitrate, nitric
acid, and sulfur dioxide were measured weekly using a stacked filter pack
system currently in use in Canada.! This stacked filter pack Is operated
over a 7 day sampling period and consists of two separate samples - one
integrated over the daytime hours and the second integrated over the night-
time hours. Both the daytime and nighttime samples consist of a single
filter holder containing three filters in series. The first filter, made
of Teflon, is for particulate removal and is analyzed for sulfate and
nitrate. The second filter is nylon and 1s analyzed for nitric acid, A
Whatman 41 filter impregnated with «2C03/glycerin analyzed for SO2 1s the
third filter. The flowrate of the system is maintained at 1,5 1pm by a
mass flow controller. The filter holder material is made of Teflon. The
Teflon filter samples are water extracted and analyzed using an ion chroma-
tograph (IC). The nylon filter, which collects gaseous nitric acid, is
water extracted and analyzed by IC. The treated Whatman 41 filter is
extracted with 0.03X hydrogen peroxide and analyzed for sulfate using a
colorimetric procedure. In addition to the Canadian Filter (CF) pack
sampler, two other techniques suitable for routine monitoring applications
were evaluated during 1986 at selected prototype stations. These tech-
niques include the PMjo s^ze Selective Inlet High Volume Sampler (SSI) 4
the Transition flow Reactor Sampler (TFR)2, The filters collected in the
network are supplied by EMSL/RTP, NC and analy2ed by an EPA contractor.
The continuous O3 and meteorological data are collected at the field sites
on data cassettes which are returned from the stations weekly. Data are
reduced and validated by EPA and the EPA in-house contractor. A validated
data set is available from the EMSL/RTP, NC.
One objective of the prototype network was to determine method preci-
sion under actual field conditions. Table 1 gives the precision in terms
of the (X) coefficient of variation (I.e., standard deviation divided by
the mean xlOO). The precision for the CF pack pack and Size Select
Inlet High Volume Sample (SSI) methods was estimated by the yearly average
of individual weekly estimates of the % coefficient of variation (CV)
between collocated Instruments located at the West Point sites. Daytime
and nighttime estimates of precision were calculated separately for each
type of filter used by the filter pack. Alternatively, precision for the
TFR method was calculated as the average % coefficient of variation between
weekly measurements from the left and right channels of the same instrument.
Estimates from different sites were pooled. With the exception of HNO3 the
"within" instrument variation for the TFR was higher {less precise) than
the "between" instrument variation for both the CF Pack and SSI methods.
However, for comparison purposes, neither of the collocated sites employed
a TFR sampler.
Prior to April 1, 1986, the collocated sites (003 and 004) at West
point. New York were equipped with similiar CF Pack samplers. No signif-
icant bias between the two collocated sites was detected during this
period. Beginning in April 1986, The collocated sites at West Point were
outfitted with CF Pack samplers with different filter holders. The CF
Pack at site 003 used an all Teflon filter holder and the CF Pack at site
004 used a Polycarbonate filter holder manufactured by Nuclepore. For
the period beginning April 1986 until January 1987, a slight Increase 1n
the XCV was noted for some pollutants.
A statistical test Indicated that the Increase in 1CV was due to the
following significant biases:
203
-------
1) HND3,(daytime) at site 003 was ^her than site 004,
2} SOk~ (nighttime) at site 003 was 6.2% higher than site 004,
and 3) SO^ (ftigftttirce) at site 003 was 14.9% higher than site 004.
As a result of these biases, CF Pack estimates of precision are given
for the period prior to the change in filter holders.
Table Z summarizes the method comparisons, which are based on weekly
percent differences (i.e., Percent Diff = (Meth 1 - Meth 2)/[(Meth 1 +
Meth 2)/2] x 100, This ratio is approximately equal to the difference
between the logarithms of the method measurements (i.e., In {Meth 1) -
ln(Meth 2)) and, if certain assumptions hold, approximates a normal distrib-
ution centered around zero with standard deviation proportional to the
coefficient of variation^. Statistical outliers were removed and only data
above the minimum detectable range for each method were used in the calcula-
tions, The large difference (156?) between CF Pack and SSI is thought to
be due to the loss irt NO3" on SSI filters due to daytime temperatures. The
CF Pack results were not affected by temperature as much since different
samplers were operated for daytime and nighttime sampling and averaged to
provide a full 24-h measurement for each week. The concentrations levels
(pg/m3) over which these method comparisons apply are summarized in Table 3.
There were not enough weekly pairs to make valid statistical tests for
CF Pack vs TFR or TFR vs SSI using NOg" measurements.
Summary
The experience gained is the operation of the prototype network indi-
cates that the Canadian Filter (CF) Pack sampler provides the best avail-
able approach to monitoring for concentrations of acidic components which
can attritate to dry deposition. The CF Pack sampler was simpler, more
reliable, and more precise than the other units tested. As newer and more
sophisticated designed samplers are developed, EWSL will consider their use
in the NDDN. Before a new sampler or technique will be considered for
inclusion in the NOON, it must demonstrate reliable operation under actual
field conditions.
References
1. K. G. Anlauf, P, FelTfn, H. A. Uiebe, H, I. Scfiiff, G. 1. Msckay, R. S.
Braman, and R. Gilbert, "A Comparison of Three Methods for Measurement
of Atmospheric Nitric Acid and Aerosol Nitrate and Ammonium,"
Atmospheric Environment, Vol. 19, pp. 325-333 (1985).
2. K, T. Knapp, J. L. Durham, and T. G. Ellestad, "Pollutant Sampler for
Measurements of Atmospheric Acidic Dry Deposition," Environmental
Science and Technology, Vol. 20, pp. 633-637 (1986).
3. Suggs, J. C., "The Distribution of the Relative Difference of Paired
f)ata," Proceedings of APCA Specialty Conference on Quality Assurance
in
Air Pollution Measurements, New Orleans, 1979, pp 210-216»
204
-------
Table 1
Precision Expressed as the Average of Weekly Estimates of Coefficients
of Variation (%) for Collocated Data. Canadian Filter Pack Estimates
are Only for the Period Prior to April 1986.
NO3
#WKS
-2
S04
ms
HN03
#WKS
S02
#WKS
%N02
#WKS
Method
Filter
Time
Filter
Teflon
Day
8
6
5
9
m
-
Pack
(FP)
Night
6
6
A
16
m.
m
-
-
Nylon
Day
-
6
If)
9
&
•
—
-
Night
-
•
tfl
ifl
11
6
•
*
-
What-
Time
man
Day
""
*
•
-
-
2
7
"
-
Night
*
¦
•
•*
¦
5
lfl
<•
-
SSI
Filter
Time
SSI
Weekly
9
21
7
37
-
-
-
-
-
-
TFR
Filter
Time
TFR
Weekly
43
10
21
82
18
74
31
67
22
62
205
-------
Table 2, Method Differences Expressed as a Percent (X) of Their Average
Site Weighted
ivera
Pollutant
Comparison
N
1
I
3
4
S
¦~T
HNO3
FMFR
43
-11.9
-12.3
-5.1
N03
FP-SSI
FP-TFR
TFR-SSI
103
4
8
154.6**
157.3
-41.1
154.0**
27.0
146.3
167.5**
163.6**
144.3**
so2
FP-TFR
46
12.1
17.2*
55.7*
S04
FP-SSI
FP-TFR
TFR-SSI
131
56
28
-6.1
22.7
-29.6**
0.6
33.3*
-17.7*
-2.7
-2.7
-3.8
36.5**
Average
155.9**
92.1
5.8
29.1**
-2.7
31.8**
-24.0**
* indicates a statistically significant difference at the a « .05 level
** for the a « .001 level
Table 3. Range of Pollutant Concentrations (yg/m3)
Pollutant
HN03
NO3
-2
S02
S04
N
M1n
Max
Mean
378
.18
19.84
1.86
462
.10
24.60
.83
328
.27
52.59
8.54
684
.30
20.29
5.82
206
-------
Figure 1
~ EXISTING SITES
PROPOSED DRY DEPOSITION NETWORK • NEW SITES
-------
Measurement of Atmospheric Organic Acids:
Considerations Regarding Sampling Artifacts and Potential Interferences
Kochy Fung
ERT, Inc., Newbury Park, CA.
Introduction
Low molecular weight organic acids, primarily formic and acetic acids,
have been measured, although infrequently, in the atmosphere and in pre-
cipitation in both rural and urban areas of the United States. These
organic acids are emitted directly into the atmosphere from sources such
as motor vehicles and can be formed in the atmosphere from the photochem-
ical oxidation of a variety of organic compounds. Organic acids may pro-
vide a significant source of precipitation acidity in rural and some urban
areas. More recently, health effects researchers have hypothesized that
low molecular weight organic acids may be substantial contributors to the
total acidity of fog droplets in urban areas such as Los Angeles, with po-
tentially important implications on human health effects from acidic fogs.
In order to understand more completely the photochemical processes that
can lead to the formation of low molecular weight organic acids in the gas
and aerosol phases during urban photochemical episodes, as well as to
assess the levels, the measurement of organic acids in ambient air is
essential. However, atmospheric concentrations of organic acids have been
measured only on a very limited bases, typically using an alkaline medium
(filter or impinger solution) following a particulate prefilter for the
collection, and analyzing an extract of the filter or the impinger solu-
tion for the acids. For example, Kawamura et al1 used the filter tech-
nique and found an array of Cj - CJ0 organic acids in Southern California,
with acetic and formic being the most abundant. This type of sampling
approach using an alkali medium can suffer from some serious artifacts:
• hydrolysis of peroxyacyl nitrates to the corresponding acids. It is
well-known that these compounds are extremely sensitive to alkali.
Peroxyacetyl nitrate (PAN), a major air pollutant in smog, is hydro-
lyzed to acetate2 or peroxybenzoyl nitrate (PBzN) to benzoic acid.
The hydrolysis of these compounds is complete even in 0.005 N KOH
and has served as the basis for the measurement of ambient peroxyben-
zoyl nitrate3.
• disproportionation (Cannizzaro Reaction). Hydroxides promote the
disproportionation (autoxidation) of aldehydes such as formaldehyde
and benzaldehyde to the corresponding alcohols and acids. The rate
of this reaction is dependent on the strength of the hydroxide.
Since formaldehyde levels may reach as high as 48 ppb in some parts
of Southern California, this reaction, if proceeds rapidly, can lead
to serious errors in the organic acid measurement.
• oxidation of aldehydes by ozone or ^02- Aldehydes, Buch as formal-
dehyde and acetaldehyde, are commonly present in the atmosphere es-
pecially in the urban environment. If these aldehydes are collected
by the sampling medium and are converted to the corresponding acidB
by 03 or HzOz in the sampling air stream, an artifact will result.
However, these reactions are pH dependent, and become less favorable
as the pH of the medium increases.
208
-------
• alkaline hydrolysis of esters. A recent report on hazardous pollu-
tants in Class II landfills4 by the South Coast Air Quality Manage-
ment District indicates that substantial amounts of Ci to C4 alcohols
and the esters formed from these alcohols with Cj to C4 carboxylic
acids, are emitted along with other organics from the five landfills
monitored. However, no free organic acids were found, apparently due
to the abundance of alcohols to esterify these acids. If organic
acid measurements are conducted using alkaline collection medium,
hydrolysis (saponification) of these esters which are also collected
by the medium to the parent acids (carboxylite) may proceed. The
rate depends on the strength of the alkali and the type of esters.
Esters of formic acid, for example, saponify rapidly5.
• volatilization of salts of organic acids from the prefilter to the
collection medium. It is probable under some atmospheric conditions
that particulate formate or acetate on the prefilter are converted to
the parent organic acids which are volatile via reaction with strong
acids (e.g. HNO3, H2SO4 or HC1) in the sampling air stream and be
trapped by the collection medium downstream. However, the formation
of acetate or formate is not likely to occur unless there is an abun-
dance of NH3 to completely titrate all of the strong acids that are
in the air. It will also take a change of wind flow pattern to bring
in sufficent strong acids from outside sources to neutralize the NH3
before this volatization process can occur. Thus, this artifact is
not expected to proceed appreciably.
Experimental Approach
The preceding discussion indicates that serious sampling artifacts are
possible when using the existing technique for organic acids. The basic
problem lies in the use of alkali which promotes the variety of undesir-
able reactions. Sampling may in fact be accomplished without complica-
tion using an impinger containing only distilled water if one considers
the true Henry's Law constants (25°C) for formic and acetic acid are
~2,80G times larger than PAN. This indicates that water is a good collec-
tion medium for the acids, but a very poor one for PAS. An experiment
conducted in the authors laborabory showed that no decrease in PAN concen-
tration was noted by the gaB chromatograph with electron capture detector
when a PAN test atmosphere at 60 ppb was drawn through a distilled water
impinger at 1 1pm. In other words, little PAN is collected with this
technique. Also the other problem associated with using an alkaline col-
lection medium disappear because this collection medium is neutral. Fur-
ther, significant improvement in the collection efficiency can be achieved
if the trapping is conducted at ice temperature because the solubility of
gaseB (the acids) in water are much greater.
To assess the collection efficiency using this approach, tandem impingers
were used to sample at 1 1pm scrubbed air which was drawn through an im-
pinger containg a known concentration of acetic acid in water. An inline
particulate filter prevented any droplets of acetic acid from reaching the
sampling impingers. The front impinger contained 10 ml of distilled water
and it was kept in ice during the one-hour experiment. The back impinger,
not iced, contained 0.05 N KOH and was used for assessing breakthrough.
By varying the strength of the acetic acid solution, different concentra-
tions of acetic acid in air were achieved. Acetic acid was chosen as the
representative organic acid because it is normally present in the atmos-
phere and because PAN is a major interference in its measurement. Upon
completion of sampling, the volume of liquid in each impinger was measured
and its content analyzed for acetate using ion chromatography.
209
-------
Another experiment was performed to determine how the temperature of the
impinger solution might vary as a function of flow rate and inlet air
temperature. It was thought that if the temperature of the collection
medium varied too much, collection efficiency could suffer. A thermo-
couple was used to measure the solution temperature under varying sampling
flow rate (1-5 1pm) while the ambient air temperature was ~23 ± 2°C.
To assess how much interference PAN would have been if the collection was
done using an alkaline medium, an ambient air sample was taken outside of
the laboratory at Newbury Park, CA. A Teflon filter was used to remove
the particles followed by two impingers in series. The upstream impinger
with distilled water was iced. The second impinger with 0.05 N KOH was
at ambient temperature. After sampling, the impinger solutions were anal-
yzed for acetate using ion chromatography.
Results & Discussions
Table 1 shows the results of the collection of acetic acid using water at
~0°C as the collection medium. It is apparent that excellent collection
efficiency was acheived (average 90.5%) even at the highest acetic acid
concentration tested (--640 ppb). The breakthrough as detected by the
second impinger would not occur under ambient sampling conditions, since
the concentration of acetic acid is in the order of a few ppb.
The results in Table 2 indicates that fluctuations in temperature of the
collection medium during sampling would not be a problem since increasing
the flow rate from 1 1pm by a factor of 5 only brought the temperature up
by 4°C.
The ambient air sample collected with the tandem impingers showed ppb
of acetate in the front aqueous impinger and ~13 ppb of acetate in the
back alkaline impinger- Thus ~\9 ppb of acetate would have been collected
if a single alkaline impinger was used, and all of it would have been
taken as acetic acid. But actually, only 6 ppb was acetic acid, as mea-
sured using the present technique. PAN, which would have been a serious
interference, was only collected by the second impinger using alkali.
In conclusion, using ice water as an impinger collection medium for organ-
ic acid is a viable technique and avoids major interferences which other
integrating techniques have suffered. Further development and refinement
of this technique will be pursued to address the other potential, though
considered minor artifacts delineated earlier.
TABLE 1
COLLECTION OF ACETIC ACID BY TANDOM IMPINGERS
MICROGRAM PER SAMPLE
CHaC00H
SOURCE,%
0.1
FRONT BACK
IMPINGER,2 IMPINGER,3
0.05
0.01
84.5 9.6
82.8 7.7
19.2 0.8
20.2 N.D.4
4.2 N.D.
3.9 N.D.
1 1 1pm for 1 hour
2 water, impinger iced
3 0.05 N KOH
4 not detected, <0.1 |jg
210
-------
TABLE 2
CHANGES IN IMPINGER SOLUTION TEMPERATURE AS A
FUNCTION OF SAMPLING FLOW RATES
Flow Rate
lpm
Water Temp.
°C
1
2
3
4
5
1.0
2.0
2.0
3.0
4.0
REFERENCES
1. Kawamura K., L.L. Ng, and I.R. Kaplan, Determination of Organic Acids
(Cj-Cno) in the Atmosphere, Motor Exhausts, and Engine Oils, 1985
Environ. Sci. Technol 19: 1082-1086
2. Stevens, E.R. 1967. The Formation of Molecula Oxygen by Alkaline Hy-
drolysis of Peroxyacetyl Nitrate. Atroos, Environ. 1: 19-20
3. Fung, K, and D. Grosjean 1984. Determination of Particulate Atmo-
spheric Benzoic Acid by Ion Chromatography with Ultraviolet Detec-
tion. Anal. Leters 17 (A6): 475-482
4. Wood, J. and M. Porter 1986. Hazardous Pollutants in Class II Land-
fills Report, South Coast Air Quality Management District,
5. Geissmsn, T.A. 1962. Principles of Organic Chemistry. W.H. Freeman
and Company, San Francisco, p 323.
211
-------
TRIFIUORINATED POLAR AROMATICS: CANDIDATE RECOVERY
STANDARDS FOR COMBUSTION STUDIES
Charles H. Lochmiiller and Anna G. Edwards
Department of Chemistry
Duke University
Durham, North Carolina 27706
Mutagenicity assays indicate that, even in the absence of mammalian tissue
homogenate activation, the moderately polar and highly polar extract fractions of
combustion source samples exhibit high mutagenic behavior. Polar compounds, which could
be employed as internal and recovery standards are needed to assess the results of field
sampling as well as to improve analyte quantitation by GCMS or LCMS. Prior to a field
testing of such compounds, a complete evaluation of their analytical characteristics and
suitability would be required.
The suitability of several triftuoromethylated compounds has been examined in the
work completed thus far and further evaluation studies are in progress to improve internal
standard compound selection from among these choices. Fluorinated compounds were
proposed because they are not widely found in the environment and because they are
uniquely suited to mass spectrometric detection. The proposed compounds were evaluated
for recovery from silica gel fractionation and reversed-phase, HPLC. Those moderately
polar compounds which showed good recovery from silica gel fractionation, reversed phase
column elution, and identifiable MS fragmentation patterns were:
4-amino-3-nitrobenzotrifluoride, 2-amino-5-nitrobenzotri-fluoride, and
4-methoxy-3-nitrobenzotrifluoride. For highly polar fractions, 2-amino-5-
(trifluoromethyl)pyridine and 2-hydroxy-5-(trifluoromethyl)pyridine might serve as
good reference compounds. Data collection for a Principal Components Analysis of the
chromatographic behavior of these compounds is presently underway in our laboratory.
212
-------
Introduction
Results of the Ames Salmonella mutagenesis bioassay on wood smoke1 >2 and diesel
exhaust3 emission extracts indicate that moderately polar and polar compounds account for
substantially more direct-acting mutagenic activity than the relatively non-polar
polycyclic aromatic hydrocarbons. Kamens fiUL1 reported that the direct-acting
mutagenicity of wood smoke particulate extracts was contained in the most polar fractions.
When the extract was exposed to 03 and NOg, In the dark, the mutagenicity increased
substantially and compounds containing nitrogen were detected by GCMS. Kleindiensi &i
aL2 reported that the mutagenicity of both gas and particulate phase components of wood
smoke emissions increased with irradiation in the presence of N02 Schuetzie et a).3
reports that the moderately polar fractions of diesel exhaust particulate extracts exhibited
the highest mutagenic behavior; as much as 30% of the direct-acting mutagenicity of
unfractionated fractions was attributed to 1-nltropyrene. Thus, the potential health hazard
of polar compounds, in general, and possibly of nitrogen-containing polar compounds, in
particular, argues for a sound methodolgy for their assay.
An internal standard compound should improve not only control studies of
environmental monitoring but polar compound assaying as well. An interna) standard should
be chemically similar to the compounds assayed, so thai its response reflects any physical
or chemical perturbations to the system. Yet it should be sufficiently different from the
compounds assayed In order to make Its detection simple and unquestionable. By definition,
an internal standard compound should not be present in the sample. Fluorinated compounds
are proposed here for combustion studies because they do not normally occur in the
environment. The unique isotopic abundance of fluorine should be readily detected in high
resolution mass spectra. In low resolution mass spectra, compound detection should be
possible from the fragmentation pattern.
The candidate compounds were evaluated for their recovery from silica gel
fractionation and from reversed phase HPLC. The column recovery studies were performed
first. Successive elutlon by hexane, methylene chloride and methanol was performed for
silica gel fractionation.
Principal components analysis4 of the candidate compounds based on their
compound/mobile phase covariance in presently underway to further establish the
suitability of a given compound from among the choices. When applied to a large data
matrix, It can determine how many factors aocount for the non-random variation in the data.
Those factors can be used to Identify chemically similar compounds. In the present
treatment, the HPLC capacity factors of the compounds in aqueous methanol and aqueous
acetonitrile mobile phases were measured. By evaluating how the compounds duster with
data already available In our laboratory5, a better evaluation of the suitability of an
internal standard will be possible.
Experimental
The trtfluoromethyl compounds were obtained from several sources. The Eli Lilly
Corp. donated 2>e-dinltro-N1N-dipropyl-4-{trifluoromethyl)benzamlne (trifluralin).
Ishihara Sangyo Kalsha, LTD donated 2-hydroxy-5(trifluoromethyl)pyridine and
2-amino-5(trifluoromethyl)pyrldine. The following compounds were obtained from
Aldrich Chemical Company, Inc. : 4-amlno-3-nltrobenzotrlfluorlde (4A3NBTF);
4-methoxy-3-nltrobenzotrifluorlde (4M3NBTF); 2-amlnO'5-nitrobenzotrifluoride
(2A5NBTF); 2-methoxy-S-nltrobenzotrlfluoride (2M5NBTF); and
213
-------
2-nitro-4(trifluoromethyl)phenol (2N4TFPh). Acetophenone was obtained from Fisher
Scientific. AH solvents were HPLC grade from Mailinckrodt, Inc. Solvents were filtered
through 0.45^m fitters and degassed prior to sample preparation and column elution.
The HPLC instrumental set-up for the column recovery studies included a Waters
6000A Solvent Delivery System, a Perkin-Elmer LC-420B Auto Sampler injection system,
and a Hewlett Packard 1040A Diode Array Detector (HP 1040A DAD). Compound recovery
from a reversed phase C18 column (Perkin-Elmer 5 (im particles, 12.5 cm, 0.46 cm ID)
was performed in 80:20 (v/v) methanol:water. Compound recovery was determined from
comparing pre-and post-column absorptivity ratios. The absorptivity ratio was defined as
the ratio of the slope of the absorbance calibration curve for trifluoromethyl compound
versus the slope of the absorbance calibration curve for acetophenone each at their
respective ultraviolet wavelength maxima. The ratio of the slopes of absorbances of
calibration solutions as measured in 1.0 cm celts in a Gary 2300 Spectrophotometer was
defined as the pre-column absorptivity ratio. The ratio of the integrated areas at each
compounds wavelength maximum on the HP 1040A DAD was labeled the post column
absorptivity ratio. The difference between pre-column and post-column absorptivity ratio
in relation to the precolumn ratio defines the recovery of trifluoromethyl compound. The
column recovery for solutions which gave an absorbance between 0.1 and 0.25 on the Cary
2300 were measured first. Differences of two percent or less at this concentration range
indicated no measureabfe compound column loss. A hundred-fold dilution of these
calibration solutions produced chromatographic peak heights which were approximately
seven times greater than the noise envelope of the HP 1040A DAD. The relative standard
deviation of integrated peak areas varied from twenty percent to five percent at this
concentration level. Recovery measurements by reference to acetophenone were considered
unsound and the difference between measured and expected peaks areas (based on dilution of
the calibration solution) was computed to assess compound recovery. Differences of five
percent or less at this lower concentration level were concluded to Indicate no measure able
compound loss.
Sample recoveries from silica gel fractionation were determined by a procedure
developed by Battelle6. The trifluoromethyl compounds were individually dissolved In
methylene chloride, and solvent exchanged to hexane by rotary evaporation between the
addition of two aliquots of hexane. The sample, now in hexane, was transferred to a silica gel
column which had been prepared In a disposable plpet. The hexane fraction was discarded
and methylene chloride and methanol fractions collected. These solutions were analyzed by
RPLC with gradient solvent programming from 60% to 90% methanokwater.
Data collection for the Principal Components Analysis was collected utilizing the same
approach as that used for the column recovery studies. In addition, the column and efuant
were thermostatted at 22°C . The aqueous methanol and aqueous acetonltrlle solutions were
prepared by volume for isocratic elution of the compounds studied.
Mass spectra of the compounds were obtained from two sources. Mass spectra from a
Finnegan LCMS using water-methanol showed fragmentation patterns which suggest
chemical ionizatlon-like conditions. Mass spectra obtained by direct probe Insertion Into a
Helwett Packard 5988A GCMS gave fragmentation patterns under electron Impact conditions.
Results
Column Recovery
The column recovery results on the candidate compounds are compiled In Table I.
Except for 2N4TFPh, better than 98% recovery for all compounds at the high concentration
214
-------
level was observed. The results for 2N4TFPh are not necessarily indicative of appreciable
compound loss on the C18 column. Studies of other polar phenols7 suggest that these
compounds are appreciably ionized during elution resulting in a concommitant shift in their
wavelength absorption maxima. Because Ion-pairing conditions were not examined in this
study, no additional work was performed on this compound. The difference between observed
and expected integrated peak areas for the diluted solutions are given in Table II. Except for
trifluralin, column recovery was good at the low levels examined. The mass of the
compounds, given in Table II, indicates picogram recovery of these trifluoromethyl
compounds.
Silica Gel Fractionation
The results from silica gel fractionation are presented in Table ill. Except for
2A5TFP and 2H5TFP, all the compounds appeared In the methylene chloride fraction. The
pyridine compounds precipitated from solution when the solvent was changed from
methylene chloride to hexane. Adifferent fractionation procedure would be needed for these
very polar compounds. The recovery of the compounds which eluted in the methylene
chloride fraction was good at both high (10s pg) and low (104 pg) sample levels.
Mass Spectra
While greater fragmentation was observed with direct probe insertion than in LCMS,
loss of fluorine was evident in both low resolution systems. Because direct insertion yields
spectra which would be expected under electron impact conditions, those results will be
discussed here. Except for 2M5NBTF and trifluralin, the Intensity of the (M-F) peak was at
least 15% of the base peak for all of the compounds. The parent peak was the base peak for
4A3NBTF, 2A5NBTF, 2M5NBTF, 2A5TFP, and 2H5TFP. For 4M3NBTF, the parent peak was
56.2% of the base peak at m/z 174. Yet this compound had the most intense (M-F) peak
with a value of 30.6% of the base peak.
Conclusions
The column recovery, silica gel fractionation and mass spectra results indicate that
4A3NBTF, 2A5NBTF, and 4M3NBTF might be suitable internal standards for recovery
studies of moderately polar compounds. Each compound was recovered in high yields from
the methylene chloride fraction. Each compound was recovered from the C18 column at the
lowest quantifiable levels possible on our instrumental set-up. Although 2M5NBTF was
recovered In high yields as well, it was excluded from consideration because its (M-F) peak
was very low. While silica gel recovery results tor 2A5TFP and 2H5TFP were inconclusive,
their recovery from the C18 column and their (M-F) fragmentation peaks suggest that
additional studies on these compounds might be used to determine their suitability as
Internal/recovery standards for highly polar fractions. Principal Components Analysis
based on compound/solvent covariance should aid In deciding If which compound shows better
chromatographic behavior suitability for a given sample type. An analysis if the variation
in capacity factors with solvent composition for proposed compounds in combination with
existing data for nonpolar and polar compounds, such as those available in our laboratory,
may have general value in reference compound selection processes
Acknowledgments
Although the research described in this article has been funded wholly or In part by the U. S.
Environmental Protection Agency through CR8-10744-03-1, 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. The authors would like to thank Mr. Ken Krost of
USEPA/AERL-RTP tor the ICMS data.
215
-------
References
1. R, Kamens, D. Bell, A. Dietrich, J. Perry, R. Goodman, L. Claxton, E. Tejada,
"Mutagenic transformations of wood smoke systems in the presence of ozone and
nitrogen dioxide. Analysis of selected high-pressure liquid chromotography fractions
from wood smoke particle extracts," Environ. Sci. Technol. 19:63 (1985).
2. T. E. Kleindienst, P. B. Shepson, E. O. Edney, L. T. Cupitt, "Wood smoke: measurement
of the mutagenic activities of its gas- and particulate-pliase photooxidation products,"
Environ. Sci. Technol. 20:493 (1986).
3. D. Schuetzle, F. S.-C. Lee, T. J. Prater, S. B. Tejada, "The identification of polynuclear
aromatic hydrocarbons (PAH) derivatives in mutagenic fractions of diesel particulate
extracts," Intern. J. Environ. Anal. Cham. 9:93 (1981).
4. D. L. Massart, L. Kaufman, The Interpretation of Analytical Chemical Data bv the Use
of Cluster Analysis. John Wiley & Sons, Inc., New York, 1983, pp. 39-74.
5. C. H. Lochmuller, "Proceedings of the Malaysian Chemical Conference of 1986."
6. C. C. Chuang, G. A. Mack, P. J. Mondron, B. A. Peterson, "Evaluation of sampling and
analytical methodology for polynuclear aromatic compounds in indoor air," Battelle
Columbus Division, Contract No. 68-02-3487, pp. 7-10.
Tabte I. Reversed-pbase, HPLC Column Recoveries
Compounds
Percent Recovery
4-amino-3-nitrobenzotritluoride
2-amino-5-rtitrobenzotrifluoride
4-methoxy-3-nitrobenzotrifluoride
2-methoxy-5-nitrobenzotrifluoride
trifluralin
2-amino-5{trittuoromethyi)pyr)dine
2-hydroxy-5(trifluoromethyl)pyridine
2-nitro-4(trifluoromethyl)phenol
99.94
99.29
98.80
100.21
98.83
98.80
99.97
43.66
Table II. Column Recovery At Lowest Levels Examined
Compound
Rel. % Diff.
Weight Range (pg)
4-amino-3-nitrobenzotrifluoride
2-amino-5-nitrobenzotrlfluoride
4-methoxy-3-nitrobenzotrifluoride
2-methoxy-5-nitrobenzotrifluorfde
trifluralin
2-amino-5(trifluoromethyl)pyridlne
2-hydroxy-5(trifU/oromethyl)pyridine
-27.0
-0.98
-0.45
5.4
9.2
1.2
0.65
449 - 1124
1560 - 3903
903 - 2258
915 - 2288
1560 -3903
1872 - 4680
1682 - 4206
216
-------
Table 111. Recovery From Silica Gel Fractionation
Compound
4-amino-3-nltrobenzotrifluorlde
2-amlno-5-nltrobenzotrifluorlde
4-meihoxy-3-nitrobenzotrlfluoride
2-amlno-5(trifluoromethyl)pyrldine
2-hydroxy-5{trifluoromethyl)pyridfne
*105 pg
" 104 pg
Percent Recovery
High Level*
-5.2 {+/- 3.1)
-5.9 {+/- 1.3)
-8.2 (+/-0.3)
0.0
0.0
Percent Recovery
Low Level**
-5.2 (+/- 5.1)
-8.8 {+/- 4.6)
-5.0 {+/- 2.9)
0.0
0.0
217
-------
MOBILE TANDEM MASS SPECTROMETRY FOR THE CHARACTER-
IZATION OF TOXIC AIR POLLUTANT (TAP) SOURCES
B.I. Shushan, SCIEX®, 55 Glen Cameron Rd., Thomhill,
Ontario L3T 1P2
G. DeBrou, Ontario Ministry of the Environment, Air Resources
Branch, 880 Bay Street, Toronto, Ontario M5S 1Z8
S.H. Mo & W. Webster, New York State Dept. of
Bnvlronmental Conservation, 50 Wolf Rd., Albany, NY 12233
The recent marriage of tandem mass spectrometry (MS/MS) to mobile Trace
Atmospheric Gas Analyzer (TAGA6) technology has resulted In an analytical
system capable of real-time direct air analysis for TAP's, VOC's and odor
compounds at concentrations as low as a few parts-per-trillion. This
direct analytical approach obvlateB problems associated with conventional
trapping methods. Since their introduction in 1983, the design of mobile
TAGA® MS/MS laboratories has undergone significant development. The
latest design comprises a customized 30 foot transit motor coach confi-
gured with a TAGA® 6000EM MS/MS system and ancillary equipment. This
paper discusses the incorporation of a sorbent cartridge purge unit into
the TAGA® system to assist in rapidly distinguishing between Isomeric
compounds (eg., ethylbenzene and xylene) using GC/MS/MS. Such isomer
pairs are indistinguishable by direct MS/MS techniques. Another recent
design addition is a meteorological station comprising wind speed and
direction monitors. The met. data are acquired and stored by the MS/MS
data system and displayed continuously to the TAGA® operator. The met.
station is .positioned on top of a 30 foot motorized telescoping tower
housed within the mobile laboratory. Combining real-time TAP analysis
with met. data, it was possible to rapidly and accurately characterize
plume profiles and emission frequencies. The mobility of the TAGA®
detector permitted the pin-pointing of point source and fugitive emission.
Examples are provided for industrial emissions of methyl-ethyl ketone and
dimethyl sulfide. A simple method of on-site calibration was used based
upon head space syringe injection. Under intractive computer control,
calibration constants were determined in ca. 5 minutes per compound within
the ambient air matrix. Other recent software developments are presented
including the ability to simultaneously monitor up to 128 different
compounds with real-time output In concentration units.
218
-------
Introduction
The critical need to provide rapid analytical data in the field in
response to the release (accidental or otherwise) of chemicals into the
environment has prompted the development of portable, transportable and
mobile analytical instruments.* Since air Is the most mobile of the
environmental media, it has the most critical needs for timely analysis.
The 'on-site' approach to air analysis has many advantages particularly if
the detector is sensitive, specific, and capable of real-time response.
Mobile mass spectrometers incorporating ion sources capable of direct air
sampling (such as atmospheric pressure chemical ionization APCT) have been
in operation since 1978 and have demonstrated efficacy in a number of
on-site applications including emergency response,2 plume tracking,^
chemical waste dump site monitoring, and odor source characterization^
to name but several. Mobile tandem mass spectrometer (MS/MS) systems have
been in operation since 1983 and have demonstrated a much greater specifi-
city than single MS systemsThe present work discusses
applications of the TAGA® 6000EM mobile MS/MS laboratory for on-site
analysis of toxic-air pollutants (TAP), volatile organic compounds (VOC)
and odor causing compounds.
Instrumentation
A schematic of the TAGA® 6000EM ion optics is shown in Figure 1. A more
complete description of the TAGA® instrument can be found elsewhere;^
briefly, however, airborne compounds are ionized directly in the high
pressure ionization source. Ion source pressures can range from atmosphe»
ric (as in APCI) to as low as 0.1 to 1 Torr (for glow- discharge CI).
Under APCI the ion source Bamples streams of air at rates up to 9 l.aec"1,
These high sampling rates eliminate wall- and memory-effects permitting
instantaneous response to airborne contaminants. Ionization occurs in
this high flow of gas converting compounds into pseudo-molecular ions
(ions indicative of molecular weight). The ions are focussed through an
orifice into the high vacuum analyzer portion for MS or MS/MS analysis.
The glow-discharge CI ion source samples air at rates of ca. 35 ml.
mln"1 through a heated direct air sampling inlet (DASI) which in turn
samples from a high volume flow of air (up to 9 1 sec-*). Once again
this high flow assures instantaneous response and minimal memory effects.
The APCI source is very sensitive to polar compounds whereas the CI
source, using charge transfer from N2+» 02+ and N0+, is
sensitive toward the chlorinated hydrocarbons and aromatics.
Results and Discussion
Figure 2 shows a typical identification of a VOC from a complex chemical
mixture using MS/MS . The complexity of this mixture (ambient air con-
taining auto exhaust) is represented by the number of peaks in the CI/MS
spectrum shown in Figure 2,(top) - each peak represents a pseudomolecular
ion of a neutral introduced to the ion Bouxce. The complexity is typical
of an urban air sample. Using MS/MS a single Ionized component can be
isolated and identified in the following manner; selected with Q1 (the
first quadrupole), fragmented by collisions with Argon gas in Q2, and the
fragment ions analyzed by Q3. The result is a CAD (collisionally activa-
ted decomposition) fragment ion spectrum of the constituent at m/z 106*
(Figure 2, middle). Subjecting this CAD spectrum to a library search
results in an identification of the 106+ ions as those of o-xylene
(Figure 2, bottom). Such an identification takes a matter of seconds to
219
-------
accomplish so qualitative analysis of airborne mixtures can be rapid. In
this case the CAD fragmentation pattern is typical of C2 substituted ben-
zenes, therefore setting Q1 and Q3 to the appropriate masB settings (say
106 and 91 respectively) and continuously monitoring, the MS/MS detector
becomes a real-time specific chemical monitor for C2~benzenes. Up to 128
different parent/fragment ion pairs can be monitored at once in this man-
ner. Real-time quantitation is possible using response factors stored in
the computer. Response factors are derived through one of two calibration
procedures: (i) a simple headspace injection technique which requires
that the vapor pressure of the compound is known and (ii) the use of
precalibrated pressurized bottled gas mixtures containing standards which
are diluted appropriately. The former technique uses user-interactive
software to derive a five- point calibration curve in about 5 minutes.
The quantitative and qualitative accuracy of the TAGA® 6000 mobile MS/MS
has been tested under the auspices of the OS EPA (Environmental Monitoring
Systems Lab) vis-a-via non-real-time techniques (eg., cryotrap GC) and
results of these tests Indicated excellent agreement amongst the technolo-
gies." One of the disadvantages of real-time MS/MS analysis cited by
EMSL on this study was that some isomeric VOG cannot readily be distin-
guished. For example, although o-xylene scored highest in the library
search for the 106"*" ion in Figure 2, 106+ could also be due to ethyl-
benzene, p- or m-xylene or a combination of these C8 aromatics ie., their
CAD spectra are very similar. This limitation has been overcome by the
incorporation of a sorbent tube purge unit (Envlrochem, Unacon thermal
desorber model 810A) which is coupled to the TAGA® MS/MS system's GC
(varian 3400) by a heated nickel transfer line. The GC is interfaced to
the TAGA® by a probe which inserts into the CI ion source. In the present
work a 25m BP-5 (SG&E) capillary column (0.25 mmID coupled directly to the
01 source) was used to chromatographically separate isomeric C8 aromatics
collected by sorbent tube trapping. The GC/MS TIC trace of automobile
exhaust in ambient air is shown in Figure 3. The individual constituents
were identified by their molecular ion mass and MS/MS fragment ion spec-
tra. The sorbent material used was a combination of Tenax® and
Ambersorb®. The thermal desorber-GC/MS/MS combination will obviate many
of the isomer differentiation problems associated with real-time MS/MS
detection. In typical field use after mobile MS/MS monitoring identifies
the location of a plume, sorbent samples can be obtained, capped and
stored in the on~board refrigerator for later analysis by GC/MS/MS. This
scenario could also be performed using SUMMA® cylinders and cryo-trap
GC/MS/MS.
In order to test the quantitative capabilities of this Thermal Desorber-
GC/MS/MS combination, sorbent tubes containing 0,2,10,50 and 250 ng of
chlorobenzene were sequentially desorbed into the GC/MS/MS while a promi-
nent and specific (MS/MS) fragmentation reaction for chlorobenzene waB
monitored (ie., M+ —* [M-C1|+ or 112+-W7+). The chlorobenzene was
one of 10 standard components introduced into the tube via a multiple com-
ponent diffusion source. Because of the specificity of MS/MS, the res-
ponse to chlorobenzene was free of interference from the other 9 constitu-
ents and this peak response was integrated (see inset Figure 4). Figure A
shows the Integrated response vs. amount injected which resulted in a
linear calibration curve (R«0.9997) and a response factor of 3440 counts
per ng. Because we are sampling using MS/MS and pulse counting, the inhe-
rent instrument "noise" is essentially zero (the inset of Figure 4, due to
the 2 ng chlorobenzene desorption, shows a noise level before the GC peak
evolution of 80 integrated counts. The signal, on the other hand, is
220
-------
approx. 8000 integrated counts). If one assumes a sampling volume of 3 1
of air per sorbent tube (1/2 hour sampling time), we are obtaining a S/N
of 100 for a concentration level of 0.13 ppb (v/v) of chlorobenzene and an
inherent detection limit of 4 ppt (S/N»3). The real-time detection limit
for chlorobenzene using the same ion source ia ca. 2ppb in ambient air.
Incorporating trapping, thermal desorptlon-GC/MS/MS, detection limits are,
in this caBe, augmented by several orders of magnitude over real-time
response.
Another relatively recent addition to mobile TAGA® systems is the
incorporation of meteorological data directly into real-time monitoring
data acquisition permitting the observation of meteorological and concen-
tration data simultaneously. Specifically, wind speed and direction are
very useful in determining source location and characteristics. An exam-
ple of such an analysis is shown in Figure 5 for the detection of methyl
ethyl ketone (MEK) emanating from an industrial source. While the meteo-
rological data are being collected the mobile TAGA® laboratory remained
stationary.
The met data are collected using a Lambrecht model no. 1453 meteorological
data collection head mounted on a 30 foot motorized telescoping tower. A
signtfleant TAGA® response to MEK from this source was observed only when
the wind was over 5 knots and out of the East or East-Northeast. Depen-
ding upon meteorological conditions and topography, plumes can be very
well defined and quite narrow. As long as wind is variable (which it
often Is), only with the use of a real-time detector system will an accu-
rate determination of potential peak-exposure be possible to obtain. In
this case, for example, the mean level was 10 ppb while peak-levels inside
the plume exceeded 100 ppb. These peak levels would not have registered
using non-real-time technology (ie., time weighted average).
Another case where meteorological data are also valuable Is when the
source is intermittent. Figure 6 shows the TAGA® real-time response to
dimethyl sulfide (DMS) which is being monitored In the vicinity of a paper
mill. The significant feature to note about this figure Is that the level
of DMS varies considerably whereas the wind speed and direction are
relatively constant.
The large surge in DMS between 6 and 10 minutes is obviously not due to a
meteorological change and must be a property of the source. As it turned
out, this episode of DMS release was due to the purging of the pulp-digea-
ters at this paper mill causing levels of DMS to rise up to approximately
ten times higher than the TWA and, certainly high enough to cause odor
problems In the surrounding neighbourhoods albeit for brief periods* Con-
ventional trapping methods would not have been able to fully characterize
this phenomena the time-weighted-average-TWA-concentration doesn't reveal
temporal distribution. While odor problems may not impact on the health
of the community odors are an important public concern.
Generally speaking, odorous or Irritating substances are polar, or
reactive, or both, and these properties render these compounds difficult
to trap and store especially for subsequent GC analysis. It has been our
experience that direct detection by TAGA® systems using APCI or CI is
often the only way to detect these types of compounds at the low levels
which these chemicals can cause sensory problems. Combining the ability
to detect these 'difficult-to-analyze* compounds with the mobility of the
TAGA® offers an effective and rapid way of pin-pointing the source and
221
-------
eventually of abatement. An example of such a mobile monitoring
application is presented In Figure 7.
A mobile analytical survey wbb undertaken In response to odor complaints
In the vicinity of a rubber-producta manufacturer. The TAGA® detector
Identified aniline and benzothlazole as possible contributors to these
odor complaints by stationary sampling in the neighbourhood. A mobile
survey, was undertaken whereby these and other compounds were monitored
while the mobile TAGA® laboratory criss-crossed streets downwind of the
plant. The mobile portion of the survey took some 17 minutes, a short
enough duration to ensure no great changes In meteorological conditions.
The response to these analytes is shown In Figure 7, top. The distance
away from the suspected source is indicated above each peak response which
signifies a traverse of the plume. From these data it is apparent that
both aniline and benzothlazole originate from the same source (their
plumes track together). The plume concentrations decrease with increasing
distance from the source however, the plume width remains fairly narrow
and well defined even at 800 m. Such dispersion data are important when
planning the location of stationary monitoring stations to continuously
monitor the levels of TAF's emanating from such sites. Plotting up points
of concentration versus sampling location renders a 2-D plot such as that
in Figure 7, bottom. The plume contour is sufficiently well defined to
pinpoint the source of contamination to the very building from where it
originated.
Conclusion
This paper has demonstrated applications of the TAGA® 6000EM for on-site
ambient air analysis for TAP, VOC and odor compounds in the ppt-ppb range.
On-site and mobile analysis offers the flexibility and opportunity to
change sampling protocols in response to changing meteorological condi-
tions. Only the combination of real-time meteorological data with real-
time concentration data provides key information for full characterization
of the source, such as temporal variations. Mobile analysis provides
plume tracking capabilities and spatial distribution of pollutants making
possible rapid location of actual points of emission. Finally, the addi-
tion of ancillary trapping-thermal desorptlon-GC/MS/MS methods overcomes
problems of Isomer differention and augments real-time detection limits by
many orders of magnitude.
References
1. D.R. Cannon, Industrial Chemical News, 7(12),1 (1986).
2. D.A. Lane and B.A. Thomson, JAPCA, 31^(2), 122 (1981).
3. D.A. Lane, Envir. Sci. Technol., 16(2) 38A (1982).
4. N.H. Hljazl and G.B. DeBrou, Transactions, December 1982, pp 100-103.
5. J.B. French, W.R. Davidson, N.M. Reld and J.A. Buckley,
Tandem Mass Spectrometry, Ed. F.W. McLafferty, J. Wiley & Sons,
1983, Chapter 18, pp 353-369.
6. J.J. Zoldak and B.E. Dumdel, Proceedings of the 78th Annual Meeting
of the APCA, Detroit, June 1985.
7. J.B. French, B.A. Thomson, W.R. Davidson, N.M. Reid and J.A. Buckley,
Mass Spectrometry In Environmental Sciences. Eds. F.W. Karasek,
0. Hutzinger and S.H. Safe, Plenum, 1984, pp 101-121.
8. P.9. Dawson, J.B. French, J.A* Buckley, D.J. Douglas and D. SimmonB,
Org. Mass Spectrom, 17(5), 205 (1982).
222
-------
9. Robert Lewis, "Summary Report; Performance Evaluation of the SCIEXfl
TAGA* 6000 Mobile Laboratory for Ambient Air Monitoring", Advanced
Analysis Techniques Branch, Environmental Monitoring Division,
EMSL, RTP, NC, 27711, (1984).
CAD Gas
1 , , * /-CM)
I f [(HchrtlQfina) j „ mJ| /
Vacuui
V- Q2 Q3
rirsnfpBr^rH* quad)
loNIZANoi
INLET IONIZATION TAQA
MODULE SOURCE INTERFACE
MODULE
' TAQA iOOOE MOMS ION OPTICS
/-CEM
/ (pulM counting!
Figure 1: Schematic of the TAQA• 6000EM mobile tandem mass spectrometer
17:
-J
»1
so
qrOLLUQE
,—s
muj->
J-
.-.-J
Schematic of TAQA 6000EM mobile analytical laboratory
Note: Due to lack of apace, a// figures could not be Included. A reprint of this paper
will be mailed to you by contacting: Or.Bori Shushan, do SCIEX,
55 Glen Cameron Rd, Thornhill, Ontario, CANADA 1371P2, (416)881-4646
223
-------
PORTABLE PHOTOIONIZATION SYSTEMS WITH
COMMUNICATIONS AND DATA RETRIEVAL CAPABILITY
FOR EMERGENCY RESPONSE TO TOXIC CHEMICAL
LEAKS AND SPILLS
M. COLLINS
PHOTOVAC INTERNATIONAL INC.
741, PARK AVENUE
HUNTINGTON, NY 11743
R.D. YOUNG
PHOTOVAC INC., Unit 2
134 DONCASTER AVENUE
THORNHILL, ONIERIO L3T1L3 CANADA
In the current environmental, political and legal climate, there is a
pressing need to inprove emergency response at toxic chemical handling and
manufacturing sites. Associated with this is the vital requirement for
sensitive automated field-portable instruments Which can readily be
deployed to site locations in emergency situations where there has been a
leak or spill of toxic chemicals and where this has to be continuously
monitored.
Cue useful instrument which can be stationed at a perimeter boundary
of a plant, landfill or underground storage tank site, is the total
ionizables monitor which utilizes the photoionization detector. This
instrument is microprocessor-controlled, self-calibrating, draws its own
sanples and can sequence by multiplexing through four channels. The unit
can be calibrated to read directly in ppm with programmable "Warning" and
"Danger" levels. Cm-board RS232 interface with auto-dial telephone modem
allows remote data transfer and built-in data logging capability enables
programming of sampling interval and duration.
The total ionisables monitor can be used to establish the presence of
chemical pluties and further identification and measurement of individual
aompounds can be monitored with the microprocessor controlled 10S70
portable photoionization gas chromatograph. This unit also has auto-dial
modem with RS232 interface to transfer chromatographic analysis data. The
analytical capability of the 1QS70 has been significantly enhanced by the
use of wide bore capillary columns which have been encapsulated in an epoxy
resin for field resilience.
224
-------
Automated Total Ionizables Present (AUTOTIP) Air Analyzer
Instrument Features
The basis of the AUTOTIP is a sensitive photo ionization detector (PID)
capable of detecting airborne chemical contaminants from 0.1 - 2Q00 PPM. A
built-in pump draws air saitple through the PID and the extent of ionization
of gases and vapors is measured in terms of a calibrated parts per million
(pE«\} output displayed in digital mode.
The AUTOTIP is contained in a waterproof enclosure approximately 9 in.
high, 12 in. wide and 5 in. deep and the weight of the instrument is
approximately 15 lb. Figure 1 gives a pictorial view of the unit. Figure
2 shows the layout of the program keyboard and indicators. Access to the
keyboard is via a outer door with a window which permits viewing of the
display and indicators. The keyboard and indicators are mounted to the
inner door which can in turn be opened to enable the user to insert or
withdraw various option cards. Pour L.E.D.s are also located on the inner
door. Green indicates "Normal" running, amber and red indicate two alarm
levels "Warning" and "Danger" respectively. Hie fourth L.E.D. signals a
"Fault" with the system. AUTOTIP is based on the 6809 microprocessor with
up to 32K of program memory and up to 16K of data memory. A sample probe
line is connected to the unit and the line can extent up to 1000 ft.
Operation
"Program" and "Option" keys will enter the program mode for chosen
functions and these keys also enable numeric entry. The "Exit" key will
abandon any program changes made and return the user to the monitoring
mode. "Clear" will delete the latest entry and "Enter" will take the user
to the next step.
The top line of the display is used to prompt the next entry and the
bottom line is used to enter the date and time. Switching the AUTOTIP
"On", will prompt the user to enter the calibration interval (in minutesJ
followed by the concentration of the calibrant. the unit will then self-
calibrate at the programmed interval. Various pertinent parameters
indicating the two alarm levels, date and time can be programmed. Using
the "Test" key, and a series of test codes enables eletronic diagnostics on
the system.
The "Option" keys permit data logging over a sample interval and timed
duration. In the multiplexing mode, the frequency of each sanple input and
duration can be entered. For remote surveillance* the baud rate is
entered. With modem operation, the phone number to cadi is entered.
The 10S70 Portable Photoionization G.C.
Hie features of this instrument are shown in Figure 3. The design of
its PID has been reported previously (1,2) and incorporates an ultraviolet
light source excited by electrodeless KF coupling. The PID operates with
air as carrier gas and acheives unprecedented sensitivity to a large number
of gases and solvent vapor (benzene can be detected to concentrations as
low a 0.1 part per billion (ppb) in a 1ml sample of air).
225
-------
Analytical Technique
Hie 10S70 employs the precolumn/backflush technique in order to
perform continuous, unattended monitoring. A Teflon probe line, (which can
extend up to 40 ft), is connected to the instrument. Air sample is drawn
into the saitple loop within the 1QS7G via a srrall internal pump and a
precisly determined portion of the sartple loop is automatically injected
into two columns in series on a timed basis.
The versatility of this precolumn/backflush configuration has been
enhanced by the incorporation of wide bore capillary columns into the
10S70. For reasons of fragility, these columns have found limited use in
field portable G.C.s but, an encapsulation method has recently been
developed which protects the columns against vibrational or frictional
damage. A typical chromatogram of a vapour mixture is provided in Figure
4. Hiis result was achieved using a 10m x 0.53mm i.d. CPSil 5CB, (5u film)
capillary column configured as a precolumn/backflush system.
Connunications Function
The 10S7Q is equipped with an RS232 interface and can be connected to
a remote terminal or specially programmed computer. In addition, the 10S70
has a telephone modem so that remote communications can be performed over a
phone line. By programming the 10S70 with the appropriate telephone
number, the instrument can "call in" to sin off-site location and a
password system prevents access to the instrument by unauthorized
personnel.
A new software package has been developed which allows chromatograms
to be automatically stored on disc and the data to be subsequently
retrieved. A reference chromatogram can be scale expanded and smoothed to
subtract ghost peaks and other artifacts. Figure 5 shows a reference and
last sanple chromatogram superinposed for coitparison and verification of
chemical identity.
Conclusion
With the capability to communicate and be remotely controlled, the
AOTOTIP and 10S70 G.C. enable considerable time saving in terms of the
hours the environmental scientist has to operate at the site. With the
phone line connections, the terminal can be set up off-site at another
faciltiy from which the situation can still be monitored safely.
References
1. R.C. Leveson and N.J. Barker, Proceedings of the 27th Annual ISA
Analysis Instrumentation Synposiun, St. Louis, Missouri, March, 1981.
2. M. Collins, R. Kolaja, R.C. Leveson and R.D. Young, Vth International
Environment and Safety Conference, 1905, September 19, London, England
226
-------
Figure 1 Automatic total ionizables present (AOTOTIP)
air analyzer.
227
-------
Figure 2 Autotip keyboard and indicators.
0.123 PPM
APR. 15/87
11:27
FAULT NORMAL WARNNG
DANGER
t—\ i—» I—\ o
V.
>
DATE
TIME
TEST
1
2
3
ENTER
P
R
0
G
R
A
M
CALIBRATE WARNING
DANGER
4
5
6
CLEAR
DATA LOG MULTIPLEX REMOTE
0
7
8
9
EXIT
P
|
T
I
ANALOG
0
N
m
0
-
ALARM
REPLY
S
228
-------
Figure 3 10S70 Portable photoionization G.C.
229
-------
FIGURE A
SEPARATION OF VARIOUS SOLVENTS USING
AUTOMATIC INJECTION OF A VAPOR MIXTURE
• 1 • 4
Column: 10 m x 0.53 mm CP Sil 5CB
(cut in a 1:8 precolumn:
analytical column arrangement)
Flow rate: 20 mL/min.
Chart speed: 1 cm/min.
Total Analysis Time: Approx. 20 minutes
PHQTOUAC
C
ISOPROPYL ALCOHOL
PROPYLENE OXIDE
JM* U
SOMCf FUIC' Z9
sourct rauex • «¦
swiiU ,
LCUfL
«.•
la.a
C N.
LCilCL
a.a
a.*
tWiH! 9
tsucu
cut»r <
LfUfL
li.a
II.*
j
ii.«
7*9.*
-.|jr « i«4hi.4
Pit*
K. «
WW Rln
JAH
It 17
HNM.VSIS «
a
Kiio iNjfcr
iinrfftiiur*
•J
1. *
•>t»B H
¦UL,
HnrEfiATUPr
a?
fcouc v«nt niw
ALK&i (' Til
1
21 . 4
i?,M
rfrt
LAIN
s
4b4(n«iur<
h
H V. >
»liiAm*.
>
->U.dl
11, M
nn
Off St'T
•M
ViNifni
»
!«>.•»
6. 4'".L 1.1
In
?j. ^ J'i
rm
ftorrtK ofLMT
i*.a
III
ri ,4it.fcr*f
J>
*M«I. 1
* jtfJ
rrn
rinc
i
ma
hi
*. i mi'i**
Vm.*
¦ j. i*i
rwr
ocwc rim
4
fllA
230
-------
FIGURE 5 CHROMATOGRAPHIC DATA RETRIEVED FROM DISC
TUfl Disk Print
14:05
on Show Reference
on Shou Last
off Auto add and zero
off Show setup
off Show peak Info
Copy Last
Smooth Reference
Add Lut
Siitract Last
9BB
Izontal stret
Vertical stretch
Vertical shift
Horizontal shift
Load Reference
Save Reference
Print left window
Printer Is now printing
Press enter to shift lowest point of Reference down to bottow of window.
231
-------
EVALUATION OF PHOTOVAC I0S50 PORTABLE
PHOTOIONIZATION GAS CHROMATOGRAPH
FOR ANALYSIS OF TOXIC ORGANIC POLLUTANTS
IN AMBIENT AIR
Richard E. Berkley
Methods Development and Analysis Division
U.S. EPA, Environmental Monitoring Services Laboratory
Research Triangle Park, North Carolina
Analysis of toxic organic vapors in air is usually done by
preconcentration front air, followed by GC/MS analysis. The
preconcentration step is a source of numerous errors which limit the
credibility of quantitative results. Also, the method is expensive and
slow, A method of getting data quickly and inexpensively for screening
purposes is needed. Also a method which does not involve preconcentration
night provide a valuable supplement to GC/MS data. An instrument which may
be capable of meeting these needs is the Photovac 10S50. It is a portable
gas chroraatograph equipped with a sensitive photo Ionization detector. It
is claimad to be capable of detecting 0.1 part per billion of benzene In
air.
During laboratory evaluation, the benzene detection limit of the
Photovac 10S50 was estimated to be 95 femtograms, equivalent to a
concentration of 0.03 part per billion. The instrument was subjected to
preliminary field evaluation in Houston, Texas. Low levels of apparent
contamination by calibrant were observed. Several unidentified compounds
were present at high levels for which the unit was not calibrated. The
apparent maximum possible level of chlcroethylene was 0.5 parts per
million. Apparent maximum possible levels for multiply-chlorinated
ethylenes were less than 25 parts per billion. Apparent maximum possible
levels of benzene and toluene were less than 10 parts per billion.
232
-------
Introduction
Analysis of vapor phase organic compounds in ambient air is usually
done by gas chromatography using a mass spectrometer as detector (GC/MS),
The principal advantage of GC/MS is its selectivity and range of
applicability. Its principal disadvantage is that preconcentration of
analytes is necessary because GC/MS cannot tolerate large quantities of
oxygen and water and because mass spectrometers cannot detect ambient
levels of analytes in samples small enough to be chromatographed.
Unfortunately, preconcentration produces many errors. Efficiency of
collection and delivery to the GC/MS is low and variable for many
compounds, and numerous artifacts are formed (1). Although
preconcentration and GC/MS analysis of organic vapors in ambient air is at
best a semiquantitative method which is expensive and labor-intensive, its
effectiveness in qualitative analysis makes it virtually indispensable.
However, a method of analyzing pollutants directly without separating them
from air, especially if It were based upon a completely different approach
to sampling and detection, night supply data unspoiled by gross sampling
errors. However, to detect one part per billion by volume of benzene in
one milliliter of air would require the capability of detecting three
pi cograms.
Photoionization detectors are sensitive enough to detect many toxic
organic pollutants without preconcentration. Though more sensitive than
GC/MS they are less selective. For such a sensitive detector,
preconcentration would not be required so most sampling errors could be
avoided. Without sampling errors, a chromatographic peak which appears
within the retention time window of a particular compound must be due to
the presence of that compound or another with similar retention time.
Positive identification is not established from such an observation, but
the "identified" compound cannot actually be present at a higher level than
Indicated without some negative Interference in the sampling or
chromatographic processes, Therefore data for most compounds would provide
credible estimates of the upper limits of true concentrations.
The Photovac 10S50 is a portable gas chromatograph with a highly
sensitive photoionization detector. It is primarily designed for
Industrial hygiene analysis of airborne organic compounds. It is self-
contained, mounted in an aluminum case with internal electric power and
carrier gas supplies. The unit measures 46 X 16 X 34 centimeters (16,25 X
6.25 X 13.25 inches) and weighs 11.6 kilograms (26 pounds). It is equipped
with a microcomputer which controls sampling and analysis and processes
data, and it has a built-in printer-plotter. This unit possesses
significant advantages over conventional gas chromatograph*!
a. It is portable.
b. Its detector Is sensitive enough to respond to ambient background
levels of benzene, toluene, and the ehloroethylenes without
preconcentration. It could be used as-ls to screen for these
compounds In ambient air or near hazardous waste sites. It also
could be used to supplement or coaplenent GC/MS analysis.
c. Typical analyte concentrations are far below levels which could
overload the column and lead to tailing or distortion of peaks,
d. It is tolerant of oxygen and blind to water. Separation of
analytes from air is unnecessary.
e. It can be transported without special preparation.
233
-------
There are also some disadvantages:
a. Detection is not specific. "Identification" of a compound is
evidence, but not proof, of its presence.
b. The unit may only be operated at ambient or slightly elevated
temperatures. This limits the quality of chromatographic
resolution, and only compounds which have high vapor pressures at
ambient temperatures can be analyzed.
c. Calibration standards of known accuracy are difficult and expensive
to prepare,
Exper imentaI
An extrapolated detection '.irt.it for benzene was determined (2).
Aliquots of NBS Standard Reference Material No. 1805 (0.254 parts per
million of benzene in nitrogen) were injected into the unit with gas tight
syringes at a constant gain of 50, and data for each injection were stored
~n disk. Output signal was sampled at intervals of 0.2 second. The
benzene peak was displayed, and start and end points for the p«ak were
determined by inspection and input to the computer. Peak area was measured
by drawing a line between the start and end points and summing the heights
of all data above the line. The standard deviation of the 40 baseline
points immediately before the beginning of the peak plus the 40 baseline
points immediately after the end of the peak was calculated. These data
were stored on disk, and another program was used to retrieve them, plot
peak area (volt*second) versus amount Injected (picograms), and perform a
linear regression analysis. The slope was 0.502 voIt*second/plcogram with
an intercept of 6.4 volt*second and a correlation coefficient of 0.999.
The aggregate standard deviation of baseline points for all peaks was
0.0236. The detection limit was taken to be twice the aggregate standard
deviation of baseline points divided by the slope, or 95 femtograma. It
was equivalent to a concentration of 0.03 parts per billion by volume of
benzene at 25 C, which was lower than the manufacturer's cfalm of 0.1 part
per billion and much lower than required to detect 1 part per billion of
benzene in air. This was an extrapolated detection limit, the smallest
sample actually injected being 1.63 picograms. The detection linit of 95
femtograms was lower than the value of 800 femtograms obtained by Clark and
coworkers who used a Photovac 10A10 chromatograph with the same detector
(31. It was higher than the extrapolated detection limit of 17 fentograms
obtained in our laboratory using the same method with a Photovac 10A10
which was operated at maximum gain.
Field operation of the unit was carried out In Houston, Texas during
March, 1987. The calibration library was generated upon arrival in the
field using as standards commercially prepared mixtures of six compounds in
nitrogen. Host sampling was done from Inside an automobile by using the
sampling pump to draw in air through a length of teflon tubing. Typical
data are shown in TABLE 1. ft is evident that contamlnation of the unit by
compounds present in the Mixture used for recalibration must have
contributed to the levels of some of the compounds observed. The levels
actually present could not have exceeded the amounts shown. They must be
presumed to define true upper limits to the levels of target compounds
which could have been present. Nevertheless, it is apparent that this
unit, as-is, should easily be capable of detecting the part per million or
lower levels of benzene, benzene derivatives, and halocarbons which raise
concern about public health when they are found near hazardous waste sites
and other sources.
234
-------
Results and Discussion
Several problems were encountered during evaluation. The most serious
were related to lack of temperature control In the column compartment.
Sunlight falling on the Instrument rapidly heated the Interior. At a gain
of 100 this caused a large upward baseline drift. When the unit was moved
out of the sun, the baseline drifted downward and disappeared from the
chart for about half an hour. Retention times and detector sensitivity
also vary with temperature. The internal solenoid valves were another
source of heat. They are mounted adjacent to the column enclosure and
generate considerable heat when energized. This can cause significant
temperature excursions over time periods much shorter than the duration of
a run. It was noticed that when the unit was manually recalibrated
immediately after a calibration run, the indicated temperature often
dropped by a centigrade degree during the short interval while the first
quantitation report was being printed. The temperature sensor is not
located inside the column enclosure, so it was not apparent how much column
temperature actually was affected by the solenoid valves.
Mlscalibration (failure to calibrate the right peak) is apt to occur
if automatic sampling is attempted at high gain. Extraneous peaks are seen
at high gain In all calibration gases, even N8S Standard Reference
Materials. The number of them changes with temperature. Because the
callbrant peak Is identified by peak number, the appearance of variable
numbers of extra peaks limits feasibility of unattended operation at high
gain. Uhen a peak other than the callbrant receives the peak number
designated for the callbrant, that peak becomes recognized as callbrant,
and its area Is matched with the concentration Information for the
callbrant being used by the microprocessor. Typically, the peak Is much
smaller than the real callbrant peak, and necessarily has a different
retention time. This puts the unit catastrophically out of calibration
until the next time the true callbrant peak is found during a calibration
run. The only means of preventing such errors Is timely action by the
operator. Mlscalibration could cause serious errors in time-weighted
averages of results obtained during unattended operation at high gain.
It is difficult to provide suitable calibration standards for such a
sensitive instrument. Part per million levels of most potential target
compounds in nitrogen or air can be purchased, some of them traceable to
NBS standards, and these are suitable for compiling calibration libraries.
For field recalibration, levels in the low part per billion range are
needed In order that the Instrument can be operated at high gain. Such
standards are not widely aval tablet they are expensive to prepare)
contaminants are likely to be present at levels similar to callbrant) and
storage stability could be a problem. Such problems are not the fault of
the instrument manufacturer or of the producers of calibration standards.
They are inherent problems of working at such low levels.
Several minor software problems were observed. The unit was not able
to report results below one part per billion. It could be "fooled" into
doing so by entering a calibration gas concentration which was one or more
orders of magnitude higher than the true one and then making corresponding
corrections to resulting data. If practiced regularly, such an expedient
must eventually lead to confusion. It would be better if the software were
adapted to report part per trillion levels. Integrations performed at high
gain may refer to the baseline level at the start of the run, rather than
the baseline at the time the peak eluted. If the baseline has drifted, as
it is likely to do at high gain, a considerable Integration error will
result. Also, the way in which the baseline is drawn in for an Integration
Is dependent in a poorly-predictable way on which integration parameters
235
-------
have been entered. A better integration subroutine or a better explanation
of it would be useful. When entering information into the microprocessor,
mistaken key punches are irreversible. The operator has no option but to
continue with whatever procedure was inadvertantly started. The
consequences can vary from a loss of recaIIbration data to an annoying wait
for the printer to finish printing unwanted information.
References
1. J, F, Walling, J. E. Buagarner, D. J. Drlscoll, C. M. Morris, A. E.
Riley, L. H. Wright. Atmos. Environ. 20. 51 (1966)
2. R. E. Berkley, EPA/600/4-86/041, PB87-132858
3. A. I. Clark, A. E. liclntyre, J. N. Lester, R. Perry. Intern. J.
Environ. Anal. Cheai.. i7. 315 (1984).
Disclaimer
Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
236
-------
TABLE I. SAMPLING AMBIENT AIR IN DEER PARK AT A TAMS SITE, (a)
3-26-07 J, P. Bonnette Junior High School in Deer Park. Partly cloudy.
Wind NE light. Approximately 20 C. Texas Air Control Board
Monitoring Site.
Time of day 15:00 15:13
Chloroethylene * 130
Retention time (seconds) 11.0 11.4
1,1-Dichloroethylene * 2
Retention time (seconds) 43.8 44.5
Benzene * D
Retention time (seconds) 131.6 131.9
Trichloroethylene * 8
Retention time (seconds) 204.8 207.8
Gain 100 100
Unit temperature (C) 32 31
Time of day 15:24 15:37 15:49 16s02
Chloroethylene • 121 93 89
Retention time 10.9 11.5 11.5 11.5
1,1-Dichloroethylene » D ND ND
Retention time 44.5 44.9
Benzene * D D D
Retention time 132.2 132.2 131.9 132.2
Trichloroethylene *854
Retention time 208.8 209.3 209.8 209.3
Gain 100 100 100 100
Unit temperature (C) 32 31 31 32
* Calibration run.
D Detected below O.S part per billion.
ND Not detected.
(a) Concentrations are in parts per billion.
237
-------
HALOCARBON ANALYSIS OF NMOC AIR SAMPLES
BY GC/MEGABORE/ELECTRON CAPTURE DETECTOR
A.L. Svkes.
Radian Corporation
Analytical Chemistry Department
Research Triangle Park, North Carolina
Harold G. Richter
Office of Air Quality Planning and Standards
Office of Air and Radiation
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina
The U.S. EPA has for a number of years sponsored a Nonmethane Organic
Compound (NMOC) air monitoring program. For the past three years Radian
Corporation has conducted the most intensive nation wide sampling and
analysis program under contract to EPA. The primary gas chromatography
(GC) analysis procedure for NMOC measurement has been a cryogenic
preconcentration direct flame ionization (PDFID) method, trapping
approximately 300 mL of air on a glass bead trap, followed by ballistic
heating to 90*C and detection by flame ionization. During the previous
year's effort an additional GC analytical procedure was incorporated into
the program. A 6-port gas sampling valve was used to inject 1 mL of sample
onto a J&W DB-624 Megabore1 fused silica column with electron capture
detection. The DB-624 column was designed to separate 27 volatile
halocarbon priority pollutants without subambient oven temperatures. The
detection limits of these halocarbons ranged from 2 to 35 parts per billion
in each 1 mL of sample. Calibration standards were prepared in the range
of 35 to 50 ppb by dilution of known amounts of each compound in
hydrocarbon free air. Over 170 ambient air samples were analyzed during
the program in addition to 2 EPA audit samples.
t
238
-------
HALOCARBON ANALYSIS OF NHOC AIR SAMPLES
BY GC/MEGABORE/ELECTRON CAPTURE DETECTOR
Introduction
Radian Corporation has assisted EPA for the past three years In
acquiring an excellent database of NMOC data. The PDFID method, with argon
cryotrapping has proven to be a good procedure that can provide NHOC
results from many air monitoring sites. There has been
much interest 1n volatile halocarbons and "air toxics" 1n the urban
environments, but the traditional technique of NMOC does not provide
halocarbon speclatlon or quantification. The work discussed here is a more
in-depth study of NMOC air samples collected in July, August, and September
1986, and has provided EPA and the States with additional information as to
the halocarbon content at various NMOC air monitoring sites across the
nation.
Experimental
Ambient air samples were collected in evacuated 6-liter Summa polished
stainless steel canisters over a three hour period which resulted in
approximately 15 ps1g of pressure. The canisters were previously cleaned
by a series of four evacuations to 5mm Hg, and pressurizing to 40 psig with
cleaned, dry air, These samples were analyzed for NMOC by the PDFID
method. After argon cryotraping the ambient air on a glass bead trap, the
GC oven was bal 11st1cally heated to 90*C, and the response Integrated over
a 2-m1n time period. Propane was used as the calibration standard.
The samples were then analyzed by GC/electron capture detection, which
was selective for halogenated and other electronegative compounds (ie.
oxygen and nitrogen containing). A gas sampling valve was used to Inject
1-mL of sample onto a DB-624 megabore fused silica capillary column. The
megabore column was 30 meter x 0.53mm 1.d., and was developed by J & W
Scientific, Inc. for purgeable organlcs as specified 1n EPA methods 601,
602, and 624-2 These methods are intended for analysis of water samples
using a purge and trap concentration technique, which Radian did not use.
Method 601 also uses the Hall Detector which 1s specific for chlorine and
bromine compounds, and 602 1s for organic purgeables with flame ionization
detection. Method 624 is a GC/MS procedure.
A calibration stock solution (20ppm) in methanol of purgeable organlcs
containing 31 compounds for EPA Method 624 analysis was obtained from ULTRA
Description of Analytical Procedure
Instrument:
Detector:
Data System:
Column:
Carrier gas:
Makeup:
Injection:
Valve Temp:
Inj. Temp:
Detector Temp:
Col. Temp:
N1W Electron Capture 9 range 10
Spectra-Physics Model 4270
J & W, DB-624 Megabore, 0.53mm x 30m fused silica
Helium 0 5 ml/m1n
Nitrogen 9 25 mL/mln
Valco 6-port gas sampling valve with 1.0 mL loop
100'C
100'C
250*C
35*C to 140*C 9 5*C/m1n
Instrument and Conditions
Varlan 3400 gas chromatograph
Calibration Procedures
239
-------
Scientific. Only 24 compounds were resolved or detected using the
ECD/megabore column. A gaseous working standard was prepared by Injecting
a known volume of the stock solution into a stream of zero air flowing into
a 6-liter stainless steel container. The container was then pressurized to
a known volume. The working standard concentrations ranged from 17 ppb to
35 ppb in air. One mL of the working standard was injected using the Valco
valve. The concentration in ppb of each compound was divided by the area
counts to calculate its response factor. These response factors were
multiplied times each compounds' area counts to obtain its concentration.
The detection limits of the compounds of interest were 1 ppbV, except vinyl
chloride and chloroethane, which were 8 and 35, respectively. Gas
chromatography/mass spectrometry was used to verify the elution order of
these compounds. The following Is a list of the compounds and their
retention times:
Compound
RTmln
Chloromethane
3.23
Vinyl chloride
3.52
Bromomethane
4.13
Chloroethane
5.08
1,1-di chloroethylene
6.09
trans-1,2-dlchloroethene
7.54
Chloroform
9.54
1,1,1-trichloroethane
9.93
Carbon tetrachloride
10.33
1,2-dichloroethane
10.71
Trichloroethylene
12.21
1,2-dlchloropropane
12.61
Compound RTnln
Bromodichloromethane 13.28
trans-1,3-dfchloropropene 14.33
cj's-l,3-d1chloropropene 15.63
1,1,2-trlchloroethane 16.03
Tetrachloroethylene 16.47
Dibromochloromethane 16.96
Chlorobenzene 18.68
Bromoform 20.48
1,1,2,2-tetrachloroethane 21.79
1.3-dichlorobenzene 24.22
1.4-d1chlorobenzene 24.54
1,2-dichlorobenzene 25.41
Results and Discussion
Table I summarizes the halocarbon concentration data by compound. The
frequency of occurrence gives the number of samples in which the compound
was found. Only 15 of the 24 target compounds occurred at or above the
method detection limits. The frequency of occurrence of these compounds
varied from a minimum of one for 1,1,2,2-tetrachloroethane to a maximum of
142 for cis-l,3-dichloropropene. The average occurrence was 47. The mean
concentration of the compounds ranged from 1.1 ppbV for carbon
tetrachloride to 434 ppbV for chloroethane, averaging 45.2 ppbV overall.
Chloroethane had the highest maximum concentration of 8327 ppbV. A closer
look at comparisons between frequency of occurrence and mean ppbV
concentrations reveals that only vinyl chloride occurred above the average
for both. Bromomethane had the second highest occurrences, but only a mean
concentration of 8.7 ppbV. Tetrachloroethylene had the fourth highest
occurrence at 92, with a low mean concentration of 2.9 ppbV.
Table II summarizes the halocarbon concentration data by site. The
mean concentrations ranged from 16.8 ppbV for site #8 to 239.9 ppbV for
site #14. The measured maximum ppbV values ranged from 59 at site #3 to
8327 at site #14. The average number of Identifications among all sites
was 47. The summary statistics in Tables I and II are representative of
the samples collected at the sites and apply to those halocarbons
Identified. The statistics should not be used to represent all halocarbon
compounds, or to be representative of average of concentrations across the
country.
Two audit samples, designated "A" and "B" in Table III, containing
seven halocarbons each were provided by EPA-QAD. These audit samples were
used to obtain an estimated accuracy of the sampling and analysis method.
The concentrations in the original samples, and the measured results are
240
-------
given In Table III. The percent bias ranged from -56.4 to -7.7, averaging
-27.0. The absolute percent bias averaged 35.3.
Conclusions and Recommendations
From the results of this work we can conclude that ambient halocarbon
measurements can be easily conducted on NMOC air samples with adequate
detection Units and compound selectivity using a commercially available
Megabore fused silica column with electron capture detection. The Megabore
column allows good compound separation without subambient oven
temperatures. This prevents water vapor in the sample from freezing and
plugging the column. Direct injection onto the column was done without the
need of preconcentratlon techniques.
Recommendations are to expand the current analytical system to measure
air toxics by combining a versatile sample interface and multl-detector
GC/Megabore column analytical system. Radian Corporation is currently
developing and will evaluate such a system which will include
photoionizatlon/flame 1onizat1on/electron capture detectors. The sample
Introduction interface will allow a choice of unconcentrated, adsorbent
trapping, or cryotrapplng techniques to be used with gas chromatographs
with multiple detectors or with mass spectrometers.
Acknowledgments
Radian Corporation, Analytical Chemistry Dept., Research Triangle Park,
North Carolina, conducted work on this project under Contract No. 68-02-
3889 for the Quality Assurance Division, Environmental Monitoring Systems
Laboratory of the U.S. Environmental Protection Agency.
References
1. O&W Scientific, Int., 91 Blue Ravine Road, Folsom, California 95630.
2. Federal Register. CFR Part 136; Vol. 49, No. 209; Friday Oct. 26
1984, EPA Methods 601, 602, 624.
241
-------
Table L 1986 HALOCARBON DATA SUMMARY, BY COMPOUND
Kalocarbon
Frequency
of
Occurrence
Halocarbon
Concentration,
Min Mean
ppbV
Max
Vinyl chloride
102
7.9
78.B
662.0
Bromomethane
127
1.0
8.7
126.0
Chloroethane
37
8.6
434.1
8327.0
l,l-D1chloroethylene
3
2.7
51.9
92.9
Chloroform
35
1.3
8.2
48.3
1,1,1-Trlchloroethane
91
1.0
2.5
5.9
Carbon Tetrachloride
2
1.0
1.1
1.2
Trlchloroethylene
17
1.1
7.0
68.7
c/s-l,3-Dtchloropropene
142
3.2
24.9
66.2
1,1,2-Trlcfrtoroethylene
14
8.7
63.1
ZZB.O
Tetrachloroethylene
92
1.0
2.9
66.2
01bromodichloromethane
2
1.8
8.0
14.2
1,1,2,2-Tetrachloroethane
1
1.2
1.2
1.2
1,3-O1ch7orabenzene (ro)
23
4.5
24.2
163.0
1,4-Dichlorobenzene (p)
22
2.9
37.1
294.0
Overall 1.0 45.2 8327.0
242
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
17
18
Table II. 1986 HAL0CARB0N DATA SUMMARY, BY SITE
Halocarbon
Number of Concentrations, ppbV
Identifications Min Mean Max
44
1.0
60.9
611
39
1.0
55.0
517
24
1.4
17.4
59
34
1.2
28.4
211
39
1.2
13.7
79
40
1.1
26.4
538
45
1.0
39.6
1220
46
1.0
16.8
82
42
1.0
17.4
183
50
1.0
30.0
513
37
1.1
20.0
126
33
1.0
49.7
1188
44
1.1
23.5
294
40
1.0
239,9
8327
35
1.0
75.2
1701
39
1.0
22.1
105
51
1.0
38.6
343
28
1.5
36.3
211
710 1.0 45.2 8327
-------
Table 111. HALOCARBOH AUDIT SAMPLE RESULTS
Audit Sample Measured Percent Absolute
CAacentcatioiu, ppbV Cone«atration®, ppbV Bias Percent Bias
Compound A B A B A B A B
Chloroform
27
13
21
13
-22.2
- 7.7
22.2
7.7
1,1,1-lf ichloroetha&e
*5
21
33
17
-26.7
-19.0
26.7
14.0
Carbon tetrachloride
34
1?
17
9
-50. 0
-47.1
50.0
47.1
1,2-Dlchloroethane
35
27
24
13
-56.4
-5i. a
56.4
51.6
trichloroethjlene
4ft
24
22
11
-54.2
-54-2
54.2
54.2
1,2-DLchloropropane
44
22
57
15
29,5
-31-8
29,5
31.8
Tetrachloroethylene
42
21
30
18
-28,6
-14,3
28.6
14 ,3
Average
Overall average
-21,6 -32.3
-27.0
38.2 32.3
35.3
-------
A PROTOTYPE SAMPLER FOR PRECONCENTRATING AND
TRANSFERRING VOLATILE ORGANIC COMPOUNDS FROM
AMBIENT AIR TO PASSIVATED CANISTERS
by
Michael W. Holdren and Deborah L. Smith
Battelle Columbus Division
505 King Avenue
Columbus, OH 43201-2693
and
William A. McClenny
Environmental Monitoring Systems Laboratory
MD-44
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
INTRODUCTION
The Methods Development and Analysis Division of the Environmental
Monitoring Systems Laboratory (EMSL) of the U.S. Environmental Protection
Agency 1s responsible for the development and evaluation of state-of-the-
art and emerging analytical techniques for the determination of organic
compounds 1n ambient air. Recently a prioritized 11st of volatile organic
compounds has been established by EPA. EMSL 1s focusing on further devel-
opment of analytical methodology for the determination of these compounds.
Developmental efforts during previous work (Contract No. 68-02-3487
(WA-37 and WA-22)) resulted in the following: (1) an automated gas
chromatograph (6C) system that uses reduced temperature preconcentratlon
to collect volatile organic compounds (VOCs) and (2) documentation of
adequate storage of volatile organic compounds In Summa polished canisters.
The automated QC system has operated 1n a sampling and analysis mode at
several field sites using reduced temperature preconcentratlon to enhance
system sensitivity. Other sampling scenarios have involved utilizing the
Summa polished canisters to collect and temporarily store a sample prior
to analysis with the automated GC which was based at the research
laboratory.
The current task 1s a feasibility assessment of non-cryogenic precon-
centratlon techniques. Specifically a prototype sampler was developed to
collect and transfer VOCs to temporary storage vessels (Summa polished
canisters). The sampler incorporates parallel Tenax traps to sequentially
adsorb and desorb organic species In ambient air. Sample enrichment 1s
achieved by utilizing less volume flow 1n the transfer process than In
the collection process. This paper discusses the design, construction
and preliminary evaluation of the prototype system.
EXPERIMENTAL
INSTRUMENTATION
Backflushlng Sample Preconcentrator
A diagram of the prototype backflushlng sample preconcentrator (BFSP)
is shown In Figure 1. All component parts have been mounted onto an
aluminum carrying case and Include the following:
245
-------
(1) An 8-port 2-posit1on switching valve and actuator (Bach
Company).
(2) Two Tenax traps (10" by 1/4" S.S. U-tubes) wrapped with heating
wire. Each trap contains 0.7 grams of 35/60 mesh Tenax GC.
(3) Two temperature controllers (Omega Engineering, Inc.).
(4) Nitrogen (0-10 cc/mln) and air (0-500 cc/min) mass flow
controllers and console (Tylan, Inc.).
(5) A dual diaphragm pump (Thomas Industries, Inc.).
(6) A timer for control of valve actuator and trap heaters (designed
and built at Battelle Columbus 01v1s1on).
The timing cycle for the backflushing sample preconcentrator 1s shown
in Figure 2. At time zero trap No. 1 1s 1n the desorptlon position with
heat being supplied via temperature controller No. 1. Trap No. 2 is at
room temperature and 1s 1n the collection position. After 7 minutes Into
the run, temperature controller No. 1 Is turned off to prepare trap No. 1
for collection which will be Initiated 8 minutes later (i.e., 15 minutes
from time zero). After 15 minutes Into the run the valve actuator 1s
activated so that trap No. 2 is now 1n the desorptlon position with heat
being supplied via temperature controller No. 2. Trap No. 1 which has
cooled to room temperature is now 1n the collection position. Approxi-
mately 1 minute Is required to reach the trap desorptlon temperature
(180°C) while five to six minutes 1s necessary to cool the trap to room
temperature. The valve and heater timing intervals can be varied from
several minutes to one hour. During the actual test runs, the SFSP unit
was operated as follows:
collection cycle: 15 minutes
heat desorptlon cycle: 7 minutes
desorptlon temperature: 180°C
nitrogen flow: nominally 7 ml/mln
air flow: nominally 120 ml/min
Expected concentration enrichment factors were calculated as the ratio of
the measured air and nitrogen flow for each experiment.
Whole A1r Canister Filling System
The canister filling apparatus consisted of the following. A stain-
less steel pump (Metal Bellow Corp., Model MB-158) directed flow through
a mass flow controller (Tylan MC-260) and Into a stainless steel canister.
The mass flow controller was maintained at 20 ml/min during sampling.
Canister Cleaning Apparatus
In preparation for use with the BFSP unit or with the whole a1r
filling system, each canister was sequentially filled (15 pslg) and evacu-
ated (25" Hg) five times using ultra zero air as the flushing gas. After
the fifth evacuation the canister was sealed, transferred to a higher
vacuum system and pumped down to 0.1 torr. A liquid nitrogen trap was
utilized in this sytem to prevent organlcs originating in the pump oil
246
-------
from back-streaming Into the can. Following the final evacuation each
canister was connected to the appropriate filling apparatus.
Automated Gas Chromatographic System
Sample canisters were analyzed with an automated GC system employing
cryogenic trapping techniques. Briefly, the system is equipped with flame
Ionization and mass selective detectors. A 50 m by 0.32 mm l.d., OV-1
fused silica column (Hewlett Packard) was used to resolve the volatile
organic compounds. Optimum analytical results were achieved by tempera-
ture programming the gas chromatographic column from -50°C to 150°C at
8°C/m1n. Zero grade helium (Matheson) served as the carrier gas
(3 ml/m1n). The column exit flow was split so that 1 ml/m1n entered the
mass spectrometer, while the remaining flow passed through the flame
Ionization detector.
A Perma Pure dryer (Model MD-125-48F) with a tubular hygroscopic
1on-exchange membrane was used to remove water vapor selectively from
mixed gas streams. The dryer was purchased 1n a shell and tube configura-
tion. Sample flow through the tube was maintained at 40 mL/m1n with a
mass flow controller (Tylan MC-260). A counter-current flow of dry zero
air (200 mL/min) was used to purge the shell.
PROCEDURE
Laboratory test runs Involved challenging the BFSP unit with a dilute
calibration mixture of 41 organic compounds (2 to 3 ppb each). The BFSP
canister sample results were ratloed with the results from the direct
analysis of the dilute calibration mixture to obtain a measured enrichment
value for each compound. These measured enrichment values were compared
with calculated enrichment values (ratio of airflow/nitrogen flow) In
order to determine col lection/recovery of the 41 compounds with the BFSP
unit.
Two ambient air test runs were carried out at an unoccupied home.
During each run two canister samples were collected; one with the BFSP
unit and the other with the whole air collection apparatus. Again,
measured enrichment values were compared with calculated enrichments In
order to determine collection/recovery efficiencies.
RESULTS AND DISCUSSION
Table 1 shows the experimental results from the two ambient air
tests. Twenty compounds, identified by retention time and quantified
with a flame Ionization detector during the tests, are listed 1n the
table. Concentrations obtained from the analysis of whole air canister
samples are shown, along with corresponding concentrations from the
analysis of can samples collected with the BFSP unit. Enrichment factors
for each compound are determined by ratlolng the two concentrations. In
viewing these ratios It is evident that for n-butane and 1-pentane (com-
pounds No. 1 and 2), compound breakthrough 1s occurring, as Indicated by
the lower ratios.
For the remaining compounds a mean ratio of 17,3 (+ 13.8%) was calcu-
lated from the test No. 1 data while a mean ratio of 1574 (+ 9.9<) was
determined from test No. 2 data. Expected enrichment ratios' were based
upon the actual flow ratio of air to nitrogen during each run. Ratios of
17.4 and 15.7 were determined respectively for test 1 and 2. In comparing
247
-------
these flow ratios with concentration ratios determined experimentally,
excellent agreement 1s observed (17.3 vs 17.4 and 15.4 vs 15.7).
Experimentally determined breakthrough volume for n-pentane 1s 4.5
liters per gram of Tenax (1). In the above experiments 1.8 liters of air
(per 0.7 grams of Tenax) was sampled with each cycle of the BFSP unit.
If one assumes a safe sampling volume to be approximately 50 percent of
the breakthrough volume (I.e., 2.3 1/g), then the above sampled volume
(2.6 1/g) Indicates that n-pentane breakthrough is probably not occurring.
However compounds such as n-butane and i-pentane (i.e., with lower break-
through volumes) will not be efficiently recovered under the experimental
conditions used with the BFSP unit.
Figure 3 shows the results from one of the laboratory test runs which
involved challenging the BFSP unit with a dilute mixture of 41 organic
compounds {2 to 3 ppb each). Concentration enrichment 1s plotted as a
function of compound number. For most of the compounds a consistent
enrichment value 1s obtained. However, for the more volatile compounds
lower enrichment values are observed (compounds 1 through 11). Again, we
believe that these compounds are not efficiently retained on the Tenax
adsorbent under the experimental conditions used with the BFSP unit {Ue.,
breakthrough volumes are being exceeded). A mean measured enrichment of
33.4 Is obtained If one excludes compounds 1 through 11. This value com-
pares well with the calculated enrichment of 35.0. In further examining
the data, one observes that peaks 40 and 41 (1,2,4-trlchlorobenzerie and
hexachlorobutadlene) show somewhat lower than expected enrichment ratios.
Closer Inspection of the BFSP unit during the desorptlon temperature cycle
showed that a maximum bed temperature of only 135°C was obtained at a
temperature set-point of 180°C. Subsequent experiments utilizing a higher
temperature set-point, resulted 1n better recovery of these two compounds.
CONCLUSIONS AND RECOMMENDATIONS
A prototype sampler has been shown to collect and transfer VOCs from
ambient air to temporary storage canisters. The system employs parallel
Tenax traps to adsorb and desorb organic compounds from an air environ-
ment. Sample enrichment has been achieved by utilizing less volume flow
1n the transfer process (desorptlon step) than 1n the collection process
(adsorption step). Preliminary tests of the sampler have indicated that
the system successfully functions as designed. Calculated sample enrich-
ments (15 to 35-fold) based on volume flow ratios were in excellent agree-
ment with experimentally determined concentration ratios.
The results of this study have demonstrated the feasibility of a
non-cryogen1c preconcentratlon technique that can be used 1n the field 1n
conjunction with stainless steel canisters to provide VOC enriched
samples. These enriched samples can then be transported to the laboratory
for subsequent analysis. The primary advantages of this approach over
conventional field sampling with Tenax cartridges are the following:
(1) The somewhat high and variable amounts of individual compounds
(1 to 50 ng per cartridge) normally found 1n field "blanks" are
not observed with the BFSP unit. We suspect that Tenax storage
and handling conditions are prime contributors to the blank
problems. Since the BFSP unit is used on a near real-time
basis, the above problems are not of concern.
248
-------
(2) The enriched canister sample from the BFSP unit can be conve-
niently and repetitively analyzed by one or more Instruments.
In the present study a preliminary evaluation of the backflushing
sample preconcentrator has been completed. Additional studies need to be
carried out using higher enrichment ratios and testing other adsorption/
desorption regimes. Other adsorbents need to also be examined for their
ability to retain the more volatile organlcs. Compound selective
adsorbents, such as the silver/silver oxide sorbent for NO2 (WA-10), and
dinltrophenyl-hydrazine sorbent for aldehydes, could be employed with the
BFSP unit. With slight modifications to the BFSP near real-time monitoring
of select compounds could also be achieved.
ACKNOWLEDGEMENT
The research described in this article was funded by the U.S.
Envlronemntal Protection Agency. However 1t 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 endorse-
ment or recommendation for use.
REFERENCES
1. R.H. Brown and C.J. Purnell, Collection and Analysis of Trace Organic
Vapour Pollutants 1rt Ambient Atmospheres—The Performance of a
Texax-GC Adsorbent Tube, J. of Chrom. 178 (1979), pp. 79-90.
249
-------
Nltrogan
fStalnlm
l Staal
\Cin J
. T*nax
i Trap #2
Pump
Figure 1. Diagram of prototype backflushing sampler.
TCI on
off
TCI en
Trap 1«Collaction
Tr»p 2-0«orption
Trap 1-Datorption
Trap 2-Collaction
0.0 7.0 110 22.0 30.0
Minutai
Figure 2. Diagram of timing cycle for the backflushfng sampler.
250
-------
Measured Enrichment * 13.4
Cafcuteted Enrichment = 310
v X
30
ts>
U»
o
Q>
at
JL
J L
OJ) 10 10.0 1XJ) 20.0 210 30.0 JS.0 40.0 45.0
t. dlcttlorodlflutroaethtn*
Z. «ettiyl chloride
). l,2-e
IT. benzene
IB. Urban tetrachloride
19. 1,2-dlchloroprgpaie
20. trlchloroeUiene
21. c1s-1.3-d1cbV*opree«ne
22. trn-l,Hlililorifr«en
23. 1.1,2-trfditsraethm
2*. toluene
25. 1.2-4
-------
TABLE 1. COMPARISON OF CONCENTRATIONS (ppb) OBTAINED FROM THE ANALYSIS OF
CANISTER SAMPLES DIRECTLY VERSUS THE ANALYSIS OF CANISTER SAMPLES
COLLECTED WITH THE BACKFLUSHING SAMPLE PRECONCENTRATOR
Compound
Retention Time
Minutes
Direct
Test #1
6FSP
Ratio Direct
Test #2
BFSP
Ratio
N»
1.
n-Butane
5.38
12.6
56.4
4.5
4.6
17.2
3.7
2.
1-Pentane
7.86
12.3
143
11.6
3.6
38.0
10.5
3.
n-Pentane
8.73
9.3
151
16.2
2.6
38.6
14.8
4.
2-Methylpentane
10.87
7.3
123
16.8
1.7
25.2
14.8
5.
3-Methylpentane
11.27
7.3
86.9
11.9
*
21.4
*
6.
n-Hexane
11.81
7.3
125
17.1
1.6
25.6
16.0
7.
Methylcyclopentane
12.45
5.3
89.6
16.9
1.2
18.5
15.4
8.
Benzene
13.10
7.3
103
13.6
3.8
50.9
13.4
9.
2-Methylhexane
13.72
5.0
93.5
18.7
1.2
20.9
17.4
10.
3-Methylhexane
13.94
3.6
70.9
19.7
1.0
18.1
18.1
11.
n-Heptane
14.61
4.0
72.8
18.2
1.1
19.9
18.1
12.
Ethylcyclopentane
15.03
3.4
58.1
17.1
0.9
13.9
15.4
13.
Toluene
15.94
15.8
270
17.1
8.6
128
14.9
14.
Tetrachloroethylene
17.09
11.0
190
17.3
6.4
95.4
14.9
15.
Ethylbenzene
18.33
2.3
41.2
17.9
1.2
16.6
13.8
16.
n&p-Xylene
18.54
9.0
142
15.8
5.1
65.8
12.9
17.
Styrene
18.97
0.5
12.0
24.0
0.9
15.6
17.3
18.
O-Xylene
19.08
3.1
56.1
18.1
1.8
26.8
14.9
19.
1,2,4-THwethylbenzene
21.45
2.1
36.8
17.5
1.5
20.9
13.9
20.
1,4-Olchlorobenzene
21.67
0.5
9.0
18.0
2.8
44.0
15.7
*Contaai nation.
-------
UPDATE ON CAMISTER-BASED SAMPLERS FOR VOCS
William A. KcClenny, Joachim D. Pleil, and Thomas A. Lumpkin
EMSL, U. S. EPA, Research Triangle Park, NC 27711
Karen D. Oliver
Northrop Services, Inc. - Environmental Sciences
Research Triangle Park, NC 27709
Over the past year, data interpretation from testing and
comparison studies of canister-based VOC samplers has provided a basis
for recommendations as to the better sampler components and sampling
train configuration. Choice of components is of particular
importance, since the sampled air must pass through the sampling train
before it is collected and stored in the canisters. During this
process, both additive and subtractive changes in the sample content
are possible. Sample contamination from volatilized organics occurs
without adequate cleaning of components. Contaminations, particularly
from the Freona and tetrachloroethylene, have been noted, while losses
of VOCs can occur from samples carried in a dry carrier gas, for
example, from audit gas mixtures. This paper addresses the status of
canister-based sampler design and use.
253
-------
Introduction
Canister-based samplers for the collection of volatile organic
compounds (VOCs) in whole air were designed as a possible alternative
to the use of solid sorbents such as Tenax-GC, The development of
canister-based samplers evolved from the use of SUMMA® polished
canisters in VOC analysis.1 Experience with these samplers in the
field during the past year has pointed up the need for modifications
to the original sampler configuration designed by the Environmental
Monitoring Systems Laboratory (EMSL) at EPA.2 Experience has also
been gained in cleaning and certifying samplers before they are used
in the field. This paper describes the updated sampler design,
details of the certification procedures, and considerations relating
to performing field audits.
Sampler Design
The updated sampler design is shown in Figure 1. The sampler
includes an inlet, 2-pm filter, mass flow controller, pump (optional),
Magnelatcfc valve, pressure gauge, timer, and thermostatically
controlled heater and fans. Modifications to the previous design
include the addition of a blower to continually pull fresh sample
through the inlet manifold at ~12 liters/min. The use of higher inlet
flow rates dilutes any contamination originating in the inlet and
reduces the chances that the sample will be altered by contacting
active adsorption sites on the walls of the inlet. Another
modification was the change from a heated inlet to an unheated inlet.
As long as the interior temperature of the sampler is maintained at a
higher level than the exterior temperature, the heated line was
thought to be unnecessary when a high inlet flow rate is used. The
sampler also contains an optional port for formaldehyde sampling with
Sep-pak cartridges.
Figure 1 shows a vacuum-assisted sampler with an indication of the
position of a pump in the pump-assisted sampler design option. In
such a "vacuum-driven" sampler, the evacuated canister is used to
provide the differential pressure needed to pull the sample. Usually,
flow rates are maintained by an electronic flow controller, which
compensates for the decreasing pressure differential between the
canister and ambient air until this differential reaches about -2 in.Hg
(-6.8 kPa). Removing the pump results in a quiet sampler, which is
desirable when sampling indoors. If positive canister pressure is
required for analysis, the sample can be further pressurized with
zero air after receipt in the laboratory. In a pump-assisted sampler,
whole-air is collected at a flow rate of 10 em3/min for 24 h. This
results in a final canister pressure of about 20 pslg (138 kPa) in 6-L
canisters. A pump-assisted sampler is useful when a larger sample
volume is required, for example (1) for multiple analyses of the
sample, or (2) when long-term, higher flow rate (as compared to a
vacuum-assisted sampler), integrated samples are necessary for a
particular study.
Certification of Samplers
During the past year considerable experience has been gained in
cleaning sampler components and certifying samplers for use in the
254
-------
field. GC-FID humid zero air test results of a badly contaminated
sampler are shown in Figure 2 as an example of the problems
encountered when sampler components are not clean. In addition, the
possibility exists that some VOCs of interest may be removed from the
sample through surface reactions within the sampling train.
In general, components are disassembled and cleaned before the
sampler is assembled. Nonmetallic parts are rinsed with deionized
water and dried in the vacuum oven at 50°C. Typically, stainless
steel parts and fittings are cleaned by placing them in a beaker of
methanol, which is then put in an ultrasonic bath for 15 min. Next,
this procedure is repeated with hexane as the solvent. Then, the
parts are rinsed with deionized water and dried in a vacuum oven at
100°C for 12 to 24 h. Once the sampler is assembled, we have found it
helpful to purge the entire system with humid zero air for 24 h.
The cleanliness of the system is then tested by challenging the
sampler with humid zero air. The humid zero air is pulled through the
sampler and analyzed with an automated cryogenic trap and GC-FID/ECD
or GC-MSD, The sampler results are then compared with control
analyses. Generally, samplers contaminated with ~0,2 ppbv or greater
of target VOCs are unacceptable. If the sampler passes the humid zero
air tests, the next step is to challenge the sampler with ambient air
spiked with selected VOCs at concentration levels expected in actual
field sampling (i.e., 0.5 to 2 ppbv) to test the sample integrity of
the system. A canister is filled with spiked ambient air from the
sampling manifold while periodic control samples from the manifold are
analyzed in real time. The canister sample is later analyzed and
compared to control runs to determine whether any VOCs of interest
were added or deleted from the sample.
Once the sampler has been tested with humid zero air and spiked
ambient air, the sampler is subjected to a second round of spiked
ambient air tests with the interior of the sampler warmed to ~11QeF.
The results of this test are examined to determine whether the
elevated temperature affects sample integrity. For Instance, elevated
temperatures may possibly result in accelerated outgassing of certain
VOCs from sampler components, and this should be investigated before
the sampler is oertified for field use.
Audit Procedures
Another problem encountered during the early field studies was the
poor results of the existing audit procedure for field samplers.
Field audits were performed over -~2 h at an elevated flow rate of 90
cm3/min with no conditioning of the sampler. The audit mixture
consisted of 10 to 50 ppbv of VOCs in dry nitrogen. The FID results
of a simulated audit are presented in Figure 3, and these results are
similar to those of audits conducted in the field. When the sampler
was challenged with a dry audit mixture, significant losses of a
number of VOCs such as o-xylene and ethylbenzene were observed. When a
humidifier was placed in line immediately following a dry audit, the
subsequent humidified audit resulted in higher than expected
concentrations of VOCs. These audit results were the same for the
original sampler and an updated sampler containing a blower pulling
sample at 500 cm3/min through the inlet. When an audit was simulated
255
-------
In the laboratory with only a humidified audit standard, no losses of
VOCs were observed.
Our interpretation of these results is the following: During
normal ambient sampling, VOC concentrations are generally 7 orders of
magnitude lower than the ambient water vapor concentration (i.e.,
1 ppbv vs. 2-5%); thus, the excess water vapor occupies a high
proportion of the available active adsorption sites in the sample path
components, and the VOCs pass through the sampling train relatively
undisturbed. However, when the dry audit gas is applied, the water
vapor equilibrium shifts, and the adsorbed water is removed from the
walls of the sampling path, thus greatly increasing the number of
active sites available to trap VOCs. Because VOC molecules are less
abundant than water molecules, equilibration is expected to take a
long time. Also, when ambient air samples are subsequently taken, the
ambient water vapor displaces the VOCs adsorbed during the dry audit,
which then contaminate the sample. Currently, procedures for
preparing humid audit standards are being developed by EPA.
Conclusions
The original canister-based sampler design has been updated and
includes a blower to pull fresh sample past the VOC sampling port at
elevated flow rates. The use of the heated inlet line has also been
discontinued in favor of an unheated inlet. Use of the pump is
optional on the basis of the user's need for a vacuum-driven sampler
or a pump-based sampler. A certification procedure for samplers has
been developed in which samplers are challenged with humid zero air
and ambient air spiked with target VOCs. Poor results of field audits
of canister-baseci samplers has led to the investigation of humidified
audit standards.
References
1. K. D. Oliver, J. D. Pleil, W. A. McClenny, "Sample integrity of
trace level volatile organic compounds in ambient air stored in
Summa polished canisters," Atmos. Environ. 20: 1403. (1986),
2. W. A. McClenny, T, A. Lumpkin, J. D. Pleil, K. D. Oliver, D. K.
Bubacz, J. W. Faircloth, W. H. Daniels, "Canister-based VOC
samplers," Proceedings of the EPft/APCA Symposium on Measurement
of Toxic Air Pollutants. April 1986, Raleigh, NC. APCA
publication VIP-7:402. Air Pollution Control Association.
256
-------
To AC
intuJattd Encfotur*
Vacuufo/Prtwur*
Cluflt
5ft
Extra Sampling
Z3a Port
flow ControHtr
o
-------
(a) Control-dry gas
u
LL
m
u
u
n
* »
fj •
(b) Dry gas through sampler
t
* *
¦i
r
jjJ—
PJUL
t ¦ s a iv c
A ~ i t is 2
(c) Humidified gas through sampler
* j
r
.Li 1—l-
UU
r,
JL
:
ml
Figure 3. Problems encountered with a dry audit mixture: (a) control; (b) dry
mixture through sampler results in losses of VOCs; and (c) subsequent humidified
samples through sampler show increases in concentrations of VOCs.
258
-------
NATIONAL AMBIENT VOC DATA BASE
Jitendra J. Shah, Emily Heyerdahl, and Ann Batson
Nero and Associates, Inc., Portland, OR 97204
Larry Cupitt, US EPA, Research Triangle Park, NC 27711
Hanwant Singh, Menlo Park, CA 94025
Every year a myriad of sources contribute to the growing body of litera-
ture on ambient Volatile Organic Chemical (VOC) concentrations. Due to the
diversified nature of this information, researchers would be well served by
a uniform data base. To this end, a national data base was prepared for EPA
by collecting, evaluating and consolidating ambient concentrations of VOCs
obtained in the United States for the years 1970 to 1980. It includes 151
potentially hazardous ambient VOCs. The data are evaluated for quality by a
rating and ranking system. In late 1987, the updated data base will be
available on IBM PC-compatible diskettes for use with dBASE III+, a commer-
cially available data base management program.
Summaries of data from the existing data base and potential applications
by other researchers will be presented. For example, 1t could provide a
foundation for evaluating the Implications of state and federal regulations.
The sufficiency of data for a particular chemlcal(s) could be evaluated,
allowing more efficient study design and allocation of resources. Measure-
ments of specific chemicals can be compared as a function of sampling and
analysis techniques. The data base also allows the comparison of air
quality in urban, non-urban, source-dominated and Indoor areas.
The assembled data base will be analyzed to answer basic questions such
as:
o What 1s the likely range and distribution of ambient concentrations for
various chemicals?
o Are there any detectable patterns or trends?
o Are there detectable chemical "hot spots" or are the ambient concentra-
tions relatively consistent over wide areas?
259
-------
Introduction
The anticipation of new regulations for air toxics has placed added
emphasis on the measurement of volatile organic compounds (VOCs) in most
states. As a result, the knowledge about ambient concentrations of VOCs has
increased considerably in the last decade. A significant amount of informa-
tion is now available on many VOCs in the atmosphere. In 1983, a national
data base on VOCs was prepared for and published by EPA.l This data base
critically evaluated and consolidated published data on ambient concentra-
tions of VOCs in the United States for the years 1970 to 1980. It is
currently being updated to include ambient and indoor VOC data, and to
include the years 1980 to 1986. The data base includes information on more
than 150 potentially hazardous VOCs. In late 1987, the updated data base
will be available on IBM PC-compatible diskettes for the use with dBASE III,
a relational data base management system.
The objectives of the task to update the data base are:
o Review EPA1s existing VOC data base and correct any discrepancies found;
o Identify new published and unpublished VOC data, critically evaluate it,
combine it with the previous data, and analyze the resulting data base;
and,
o Provide the resulting VOC data base in a dBASE III format(s) designed to
facilitate searching, sorting, editing, analysis and other manipulations.
Methodology
The existing VOC data base contains comprehensive information on ambient
VOC levels in the U.S. from 1970 to 1980. This data base was checked for
erroneous entries, and converted to dBASE III format to be included in the
updated data base. The major elements of the update procedure are:
o Contact individuals and organizations measuring VOCs
o Perform literature search, obtain data
o Develop data assessment criteria {ranking and rating)
o Review reports, extract and critically evaluate data
o Update the data base
o Perform Quality assurance procedures
o Analyze data base
o Distribute data base
Contact Individuals and Organizations Measuring VOCs. A survey form was
developed and an 0MB clearance obtained. The survey form was sent to
researchers who were identified as potentially having data. Data gathered
through EPA sponsored programs were also obtained. A telephone follow-up to
a small group of researchers was necessary to obtain adequate response.
Literature Search/Obtain Data. After several preliminary searches by
EPA and NAI, a major on-line literature search was performed using the
Chemical Abstracts data base of the Dialog on-line system. The data base
was searched for each of the 151 chemicals 1n the original data base by
Chemical Abstract Service (CAS) registry number combined with key words to
limit retrievals to the ambient and indoor atmospheres 1n the U.S. during
the years of interest (1980 to present).
The search yielded a list of 2769 references of which 600 were considered
potentially useful and were reviewed for useful data. Table 1 lists some of
the major databases which contain VOC information.
260
-------
Develop Data Assessment Criteria (Ranking and Rating). To facilitate
data evaluation, the data quality will be judged and reported in the form of
ranking and rating. Ranking, a mu11i-d1g11 number gives a quantitative
evaluation of each of the chosen factors (Table II) affecting data quality.
Ranking would he useful for data evaluation by researchers familiar with VOC
measurements. Rating, a single letter code, on the other hand, 1s an
overall measure of data quality. The rating is determined from the ranking
of factors and their relative importance. The data quality evaluation will
be performed during the review of reports.
Review Reports, Extract and Critically Evaluate Data. All reports and
papers received from researchers or obtained from the literature were
reviewed and evaluated for measurements of ambient concentrations of VOCs.
The information listed on Table III was extracted from the useful reports
and coded for data entry. Much of the Information from EPA was obtained on
diskettes, thereby reducing data entry time and errors.
Update The Oata Base. The coded data from reports and from EPA disk-
ettes were entered into dBASE III files contents as shown in Table III.
DBASE III permits the specification of ranges of acceptable input values.
This allows for immediate error checking and correction. The existing data
base contains 18,000 records: the update is expected to Include more than
70,000.
Quality Assurance Procedures* The following elements constitute the QA
procedure.
o Only experienced data entry personnel will be used
o Much of the data will be obtained on diskettes and transferred
directly into dBASE III format
o Simple error checking techniques available in dBASE III will be used
as appropriate
o All software will be tested thoroughly before use
o Scatterplots and other techniques for identifying "outliers" as
potentially erroneous data will be performed
o A certain number of reports (>10%) will be Independently reviewed by
two analysts, anomalies found will be discussed and corrected
o A thorough review by prominent researchers 1n this field will be the
most important QC element to Insure data validity
Analyze Data Base. NAI will perform exploratory analysis on the master
data file to address basic questions such as the following:
o What is the Hkely range and distribution of ambient concentrations
for the various chemicals?
o Are there any detectable patterns or trends?
o How does ambient concentrations compare with TLV, or other proposed
ambient guideline levels?
o Which chemicals are Inadequately characterized?
The last data base was widely applied and the new data base 1s expected
to have even wider applications. Some of the specific potential applica-
tions 1n receptor modeling have been discussed elsewhere.2'3
Distribute Data Base. In order to maximize the usefulness of the
resulting data base, it will be available on IBM PC-compatible diskettes in
dBASE III format. The resulting data base will be available from EPA or
NTIS or from NAI 1n late 1987.
261
-------
REFERENCES
1 R, Brodzinsky and H.B. Singh, "Volatile organic chemicals in the
atmosphere: an assessment of available data," SRI Int., Menlo Park, CA,
EPA-600/3-83-027(A} (1983).
2 I. Tombach, "Application of receptor modeling methods to volatile
organic gases," AeroVironment Inc., Pasadena, California, AV-TP-82/592
(1982).
3 J.J. Shah, E. Heyerdahl, and J.E. Core (1987). "Ambient VOC Data Base
For Receptor Models Application," Nero and Associates, Inc., Portland,
OR; DEQ, Portland, OR, April 1987.
4 J.J, Shah, E.K. Heyerdahl, and A. Batson "A national ambient VOC data
base," Nero and Associates* Inc., Portland, OR (in preparation as an EPA
report).
5 G.E. Weant and G.S. McCormlck, "Nonindustrial sources of potentially
toxic substances and their applicability to source apportionment
methods," U.S. EPA, Research Triangle Park, NC, EPA-450/4-84-003 (1984),
6 D.L. Eichler and J.H. Mackey, "The level of certain volatile organic
compounds in the ambient air of the United States," 79th Annual Meeting
of the Air Pollution Control Association, 86-91.5 (1986).
7 J.M. Shackelford and VI.R. Ott, "Bibliographic Literature Information
System (BUS)," U.S. EPA.
8 S. Brown and L. Horn, "Concentrations of indoor pollutants
-------
TABLE I. List of Data Bases Containing VOC Information
Data Bases
National Ambient/Indoor VOC Data Base
(This Work)
J.J. Shah, E.K. Heyerdahl^
% A. fiatson, NA1
National Ambient VOC Data Base
H,B. Singh & R. Brodzinsky, SRI Int.*
Type of Data
Coverage
>70,000 after update* Ambient/
Indoor *
17,700 data records Ambient
Nonlndustrial Sources of Pot. Toxic Subs.^ bibliographic
6.E. Weant & G,S. McCormlck, EPA citations
Source
Occidental Chemical VOC Database
J.H. Mackey » D.L. Eichler
Occidental Chemical Corp.
>10,000 data records Ambient
Bibliographic Lit. Inform. System (BLIS)7
J.M. Shackelford & W.R. Ott, EPA
621 abstracts
Indoor
Concentrations of Indoor Pollutants (CIP)8 283 extended
S. Brown, LBL abstracts
Indoor
Indoor A1r Information Retrieval (IAIR)9 483 bibliographic
D. Chan & O.E, Howes, Jr., EPA citations
Indoor
Interim DB for State & Local VOC Meas.*0 814 data records Ambient/
W.R. Hunt, EPA Indoor
*1ndoor workplace excluded
**List of additional major studies to be Included 1n this data base
TAMS
TEAM (Total Exposure Assessment Methodology) -- Pellizzarl & Wallace
ATEOS
Indoor Study of 40 East Tennessee Homes — Gammage S Hawthorne
EPA/EMSL Chlorinated Hydrocarbon Data
Houston Oxidant Modeling Study 1978
Versar Studies 1n Baltimore 4 Philadelphia (1984)
NMOC -- R.L. Sella, W. L. Lonneman
NE Corridor Regional Modeling Project (NECRMP)
SRI Data 1n EPA/600/3-86/047
Westberg: Milwaukee Ozone Study 1981, and Philadelphia Data Enhancement
Study 1980
Atmospheric Lifetime Study — Rassmussen/Khalil
263
-------
TABLE II. Ranking and Rating
Wei qht Factor
10 Sampling and Analysis
o appropriateness of sampling and analysis methods, storage, etc.
o artifact or other problem with the methods
10 Quantification
o QC
o QA
o calibration procedures
o observed concentration relative to the quantitation limit
o external lab QA
o internal consistency
7.5 Representativeness
o comparisons with others
o internal consistency (ensemble data set)
o spatial, temporal
o quantity of data for purpose
5 Documentation
o objective of monitoring
o meteorology
o comprehensiveness
TABLE III. dBASE III File Structure
CONCENTRATION FILE
Reference Number
Site Number
Chemical Number
Site Type
City
State
Concent ration
Relative Standard Deviation
Number of Samples
Minimum Concentration
Maximum Concentration
Number of Samples Less than
Reference Number
Site Number
Site Type
Latitude
Longitude
Site Address
City
State
SAROAD Number
Reference Number
Site Number
Chemical Number
Sampling Method
Analysis Method
METHODS FILE
Minimum Quantitation Limit
SITES FILE
Minimum Quantitation Limit
Start Date
Start Time
Stop Date
Stop Time
Hours Between Sampling and Analysis
Ranking
Rating
Sampling Duration
Comments
Reference Number
Chemicals Measurement
Principal Investigator
Authors
Title
Citation
Year
Objective
REFERENCE FILE
264
-------
JUNE-SEPTEMBER, 6-9 AM, AMBIENT AIR BENZENE
CONCENTRATIONS IN 39 U.S. CITIES, 1984-1986
Robert L. Sella
Gas Kinetics and Photochemistry Branch
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Ambient air samples from 44 sites in 39 U.S. urban areas
were collected in electropolished, stainless steel canisters on
week days from 6 to 9 am during June through September of
1984, 1985, and 1986. Not all sites were sampled all years.
Samples were analyzed by capillary gas chromatography with flame
ionization detection to determine C2 to C^ volatile hydrocarbon
composition. Benzene, which is carcinogenic and present in
automobile exhaust and gasoline, was present in every sample.
Individual sample concentrations ranged from 1.0 parts-per-
billion as carbon (ppbC) to 273 ppbC. The benzene median
concentrations by site and year ranged from 4.8 ppbC to 35.0 ppbC
with the overall median being 12.6 ppbC.
265
-------
Introduction
Benzene is a known human and animal carcinogen (1) which has
been designated as a hazardous air pollutant under Section 112 of
the Clean Air Act (2). Benzene is present in gasoline and the
emissions from gasoline fueled vehicles. Levels in gasoline
range from 0 to 4% by volume with a national average of 1.3% (3).
The major source of benzene in ambient air is mobile sources. In
1982, for example, mobile sources contributed 85% of benzene
emissions. Of this 85%, 82% was due to exhaust emissions, 17%
from evaporative emissions, and 1% from refueling operations.
Gasoline-powered vehicles accounted for 95% of mobile source
benzene (4). Stationary sources of benzene include fugitive
emissions, coke by-products, chemical production, gasoline
marketing, benzene storage, and solvent use. Benzene is ubiqui-
tous in ambient air. Data on ambient air concentrations of
benzene and other hazardous organic chemicals were assembled by
Brod2insdy and Singh from 241 references covering primarily the
years 1970 through 1980 (5). They reported a median concentra-
tion of 2.8 ppb by volume derived from 2292 measurements in
urban/suburban areas of the U.S. Even with this large number of
measurements, they were unable to perform trend analysis, because
most concentrations were day-time measurements made during the
warmer half of the year.
EPA initiated a multi-year project in 1984 to obtain 6-9 am
total non-methane organic compound (NMOC) concentrations from
selected U.S. cities. Between 10 and 15% of the samples were
analyzed to determine speciated hydrocarbon compositions. Ben-
zene was one of the hydrocarbons determined. Over the project's
three years more than 900 samples were analyzed.
Experimental Methods
Ambient air was sampled daily from 6 to 9 am from June
through September each year. Samples were pumped into evacuated,
electropolished, stainless steel (ss) spheres and air freighted
back to Research Triangle Park for analysis. The sampling proto-
col and equipment are described in detail elsewhere (6,7),
Benzene was determined on a Hewlett-Packard model 5880A gas
chromatograph (gc) equipped with a flame ionization detector
(fid) and outfitted with a dual valve cryogenic preconcentration
system designed to condense 470-ml of air for injection onto a gc
column. C2 to Cj, hydrocarbons including benzene were condensed
in an 18-cm by 3.2-mm o.d. ss trap containing 60-80 mesh glass
beads at liquid oxygen temperature (-183° C) and separated on a
60-m, 0.32-mm i.d, fused silica column with a l.o-um thick film
of DB-1 bonded liquid phase. The column temperature program was
-50° C for 1 min to 200° c at 8°/min. The fid was operated at
270° C. An IBM PC-XT with 64OK RAM was used to receive and
process the gc data. A National Bureau of Standards propane-in-
air standard reference material was used for calibration of all
hydrocarbons. The benzene limits of detection and quantification
were estimated to be 0.04 and 0.12 parts-per-billion as carbon
(ppbc) respectively. Benzene and other hydrocarbons were identi-
fied by retention time.
266
-------
Results and Discussion
Table I is a statistical summary showing the number of
samples, median, minimum, and maximum benzene concentration in
ppbC by year for each site. During the three years of the study
samples were analyzed from 44 sites in 39 cities. The five
cities with two sites are shown with a "-1" or H-2H after the
city in the table. Pour of the Texas cities, Philadelphia, and
Washington, DC were sampled all three years. The other four
Texas cities, Atlanta, Birmingham, Kansas City, and Richmond were
sampled during two of the three years. The remaining 25 cities
were sampled only one year. The total number of analyses for all
cities over all years was 812, although the number per city
ranged widely from only three at Miami to 50 at Dallas.
The individual sample benzene concentrations ranged from £
low of i.o ppbC at Orange in 1984 to 273 ppbC at Birmingham
in 1986. The median concentrations ranged from 4.8 ppbC at West
Palm Beach to 35.0 ppbc at Memphis, with an overall median of
12.6 ppbC for all sites. For 1984 and 1985, the median percent
benzene fraction of NMOC ranged from 1.2 to 3.1 with an average
of 1.9. The variation of benzene relative to NMOC was much less
than the variation of the absolute or median concentrations.
This suggests that benzene concentrations are highly correlated
with total NMOC, which is consistent with the high percentage
mobile source contribution to ambient air benzene concentrations.
A high mobile source contribution to NMOC levels in urban areas
is expected during the 6 to 9 am period.
The contribution of benzene from mobile sources can be
estimated by multiplying measured ambient air acetylene con-
centrations by the benzeneracetylene ratio determined from a high
mobile source area such as a tunnel (8). We calculated the
vehicular benzene contribution for all of the 1984 and 1985
samples using the benzene:acetylene ratio (0.92) that we had
determined for the Lincoln Tunnel in 1982. The results indicated
that mobile source emissions contributed at least 90% of the
benzene present in 92% of the 1984 and 1985 samples. The cities
that showed significant evidence of sources other than mobile
were St. Louis and Indianapolis in addition to four Texas cities,
Beaumont, clute, Orange, and Texas City. Texas City was the most
prominant with 36% of its samples indicating at least a 10%
contribution by stationary sources.
Conclusions
Ambient air benzene concentrations for 6 to 9 am in the 39
U.S. cities reported here were quite low. The site median con-
centrations ranged from 4.8 to 35.0 ppbC and the overall median
was 12 ppbc. The data also indicated that mobile sources were
the major source of benzene in the vast majority of samples.
267
-------
References
1. International Agency for Research on Cancer (IRAC), IRAC
Monographs on the Evaluation of the Carcinogenic Risk of
Chemicals to Humans; Some Industrial Chemicals and Dyestuffs,
Lyon, V2 9, 1982, pp. 93-148.
2. Federal Register, Vol. 49, June 6, 1984, p. 23478.
3. E. M. Shelton, C. L. Dickson, "Motor gasolines, summer 84,"
National Institute for Petroleum and Energy Research,
Bartlesville, OK, 1985.
4. C. L. Gray, Emission Control Technology Division, "Mobile
source benzene emissions and a preliminary estimate of their
health impacts," internal memorandum to R, D. Wilson, Office of
Mobile Sources, U.S. EPA, Ann Arbor, MI, May 15, 1986.
5. R. Brodzinsky, H. B. Singh, "Volatile organic chemicals in
the atmosphere: an assessment of available data," EPA-600/3-83-
027(A), April 1983.
6. R. A. McAllister, D-P. Dayton, D, E. Wagoner, "1984 and 1985
nonmethane organic compound sampling and analysis program,"
Proceedings: APCA/EPA 1986 Symposium on Measurement of Toxic Air
Pollutants, Raleigh, NC, April 1986, pp. 442-457.
7. D-P. Dayton, R. A. McAllister, D. E. Wagoner, F. F. McElroy,
V. L, Thompson, H. G, Richter, "An air sampling system for
measurement of ambient organic compounds," Proceedings: APCA/EPA
1986 Symposium on Measurement of Toxic Air Pollutants, Raleigh,
NC, April 1986, pp. 431-441.
8. W. A. Lonneman, R. L. Seila, S. A. Meeks, "Non-methane
organic composition in the Lincoln Tunnel," Envir. scl. Technol.
20: 790 (1986).
268
-------
Table I. Statistical summary of benzene concentrations.
Concentration, ppbc
City
Year
N
Median
Min
Max
Akron OH
1984
10
6.0
4.6
15.2
Atlanta GA
Atlanta GA
1984
1986
7
14
9.6
11.2
3.7
4.0
53.5
27.8
Baltimore MD
1986
7
16.9
8.2
30.9
Baton Rouge LA
1985
16
8.3
2.4
23.5
Beaumont TX
Beaumont TX
Beaumont TX
1984
1985
1986
9
19
13
10.7
14.1
11.3
S.2
6.8
6.2
33.1
49.4
36.9
Birmingham AL
Birmingham AL
1984
1986
¦ 6
13
8.4
18.7
7.8
6.3
185.7
273.2
Boston HA
1985
8
6.0
3.9
15.9
Bridgeport CT
1986
16
8.9
3.5
33.4
Brooklyn NY
1986
16
13.7
5.0
38.5
Charlotte NC
1984
16
6.9
2.6
25.4
Chattanooga TN
1984
12
22.0
8.4
65.4
Chicago-l IL
Chicago-2 IL
1986
1986
8
14
27.3
20.7
14.1
3.8
48.4
30.3
Cincinnatti OH
1984
7
10.4
6.4
22.5
Cleveland OH
1985
17
23.7
5.2
60.9
Clute TX
Clute TX
1984
1985
10
17
9.4
12.0
3.9
1.3
26.9
34.8
Dallas TX
Dallas TX
Dallas TX
1984
1985
1986
13
23
14
10.0
12.6
11.5
4.5
5.7
4.2
24.4
34.9
21.3
Denver-1 CO
Denver-2 CO
1986
1986
12
13
16.7
24.5
6.3
17.9
39.6
39.5
El Paso TX
El Paso TX
El Paso TX
1984
1985
1986
8
17
9
18.8
16.0
20.0
6.0
6.3
6.9
39.2
39.4
41.6
Fort Worth TX
Fort Worth TX
Fort Worth TX
1984
1985
1986
13
19
16
14.2
13.0
10.5
8.6
5.8
6.2
37.4
26.0
40.4
269
-------
Table I continued
city
Year
N
Median
Min
Max
Houston-1 TX
Houston-1 TX
Houston-2 TX
1985
1986
1986
22
14
12
16.1
22.5
22.7
5.2
7.5
10.9
50.9
112.3
31.7
Indianapolis IN
1984
10
18.9
9.1
76.2
Kansas City MO
Kansas City MO
1984
1985
11
18
12.4
9.7
5.8
4.1
34.8
22.6
Lake Charles LA
1985
16
11.9
3.8
18.2
Manhattan NY
1986
12
10.5
5.3
31.8
Memphis TN
1984
8
35.0
17.8
47.0
Miami FL
1984
3
12.9
7.6
17.3
New Haven CT
1986
16
10.5
2.6
57.5
Orange TX
Orange TX
1984
1985
10
16
8.5
7.8
1.0
1.8
14.5
46.4
Philadelphia-1 PA
Philadelphia-1 PA
Philadelphia-l PA
Philadelphia-2 PA
1984
1985
1986
1985
7
13
14
11
15.2
15.6
6.0
19.6
8.5
6.3
1.9
6.9
27.8
31.0
17.9
51.0
Portland ME
1985
13
10.4
2.4
18.1
Richmond VA
Richmond VA
1984
1985
10
14
6.4
9.8
2.6
4.7
10.0
18.5
Salt Lake City-1 UT
Salt Lake City-2 UT
1986
1986
14
13
21.4
26.1
8.8
9.7
41.8
47.2
Scranton PA
1984
9
7.1
3.2
18.5
St Louis MO
1985
18
11.1
3.8
72.7
Texas City TX
Texas City TX
1984
1985
13
15
12.0
12.6
5.5
3.4
47.3
119.4
Trenton NJ
1986
16
13.1
8.0
36.6
Tulsa OK
1986
12
7.7
2.6
65.2
Washington DC
Washington DC
Washington DC
1984
1985
1986
10
11
11
17.8
15.3
5.6
13.8
7.3
3.7
33.5
23.5
25.3
West Palm Beach FL
1984
8
4.8
2.6
9.7
270
-------
NONMETHANE ORGANIC COMPOUND SAMPLING AND
ANALYSIS PROGRAM
Robert A. McAllister, Dave-Paul Dayton,
Denny E. Wagoner
Radian Corporation
Research Triangle Park, NC 27709
Harold G. Richter
Office of Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
The program to measure nonmethane organic compound (NMOC) concentrations
in ambient air has completed its third year using the cryogenic preconcen-
tration direct flame ionization detection (PDFID) method. Forty-four field
sites were located in (or near) 39 metropolitan centers in the United States.
Twenty-three states in eight EPA regions were involved. Twenty-two sites
participated in the 1984 study, 19 in 1985, and 23 in 1986. Integrated
ambient air samples were taken from 6:00 A.M. to 9:00 A.M. 1n electropolished
stainless steel canisters Monday through Friday from June through September.
Analyses were done at Radian Corporation's Research Triangle Park (NC)
Laboratory within three days of sample collection.
Nearly 5000 separate ambient air NMOC concentrations are included in the
three-year data base. Care was taken to determine the data quality in terms
of precision, accuracy, and completeness. For most of the sites which
participated in the study for more than one year, a lowering trend of the NMOC
concentration with time was detected. In Philadelphia, PA, for example, the
NMOC concentration averaged 1.02 parts per million carbon (ppmC) in 1984,
0.657 ppmC 1n 1985, and 0.450 ppmC in 1986. In 1985 ambient air samples were
tested for their NMOC stability while being stored 1n canisters over fourteen
days. No change in NMOC concentration was discerned over that time period.
271
-------
Introduction
The gas chromatographic, flame lonization-detector (GCFID) method has™ ,
been one of the most accurate methods for determining NMOC concentrations. ' '
The EPA-Quallty Assurance Division (QAD) developed the PDFID method because it
is faster, simpler, and less costly to operate than the GCFID method.
During the summers of 1984, 1985, and 1986, the PDFID method was used
Radian to measure 4773 NMOC concentrations in ambient air, and was shown
to have an accuracy and precision comparable to the GCFID method. The EPA-QAD
monitored the Radian NMOC measurements with a PDFID instrument. The
EPA-Atmospheric Sciences Research Laboratory (ASRL), using the GCFID Method,
measured the NMOC concentration of a number of the same ambient air samples
collected for the NMOC monitoring program.
Data Summary
Table I shows NMOC concentrations at sites in the program for more than one
year In terms of arithmetic averages (means), medians, and number (the count
of valid samples received from the site in one year). The upper part of
Table I shows six sites that participated 1n the NMOC program all three
years. The lower part of the table lists the sites that were in the program
for only two years. Table II gives the same data for those sites in the NMOC
program for only one year.
Comparing the mean NMOC concentrations in 1984, 1985, and 1986 shows an
overwhelming number of cases 1n which the average parts per million carbon
(ppmC) decreased from year to year. There are 22 out of 26 comparisons in
Table I showing a decrease of the annual mean NMOC concentration from year to
year, or between 1984 and 1986. Many of the sites selected were 1n urban
areas, or at locations where the major source of NMOC 1s internal combustion
engines (automobiles, trucks, etc.), or industrial sources.
Precision
Precision was determined by repeated analyses of site samples.
Analytical Precision
Table III shows the results of repeated analyses of site samples in terms
of % difference, and absolute % difference. The Rad1an-vs-Radian mean % 1s
the average difference between the second and the first analysis by a Radian
channel. At Radian, four PDFID channels were in use and designated Channels
A, B, C, and D, respectively. Tests ' " involving analysis of ambient air
samples by all four Radian channels established that there were no significant
differences between analyses among Radian channels. The mean % difference for
the Rad1an-vs-Radian, 1984 through 1986 comparisons was -0.3536%, with a
pooled standard deviation of 15.688%. Mean % difference for the Radian-vs-
Radian comparison was similar to a bias for the second analysis compared to
the first analysis. It was close to zero (- -0.354%), as expected. The
EPA-QAD-vs-Rad1an comparison indicates that the difference averaged -8.875% of
the mean of the individual EPA-QAD and Radian NMOC measurements. The negative
sign indicates that the EPA-QAD NMOC measurements averaged higher than the
Radian NMOC measurements.
272
-------
The EPA-ASRL-vs-Radian comparison shows a positive mean difference of
2.015%, Implying that the Radian measurements averaged higher than the
EPA-ASRL measurements.
The mean absolute % differences given 1n Table III are important
Indicators of precision. They are the expected % differences for the three
comparisons, Radian-vs-Rad1an, EPA-QAD-vs-Rad1an, and EPA-ASRL-vs-Radian NMOC
measurements, without regard to sign. The Rad1an-vs-Rad1an mean absolute %
difference comparison 1s a within-laboratory precision (or analytical
precision) of the NMOC measurement by the PDFID technique. The EPA-QAD-
vs-Radian mean absolute % difference comparison 1s an expected bias for the
PDFID method Including the between-laboratory bias and the within-laboratory
bias. The EPA-ASRL-vs-Rad1an comparison is an expected absolute % bias
between a GCFID method 1n one laboratory and the PDFID method in another
laboratory, thus containing between-method bias and within-laboratory bias.
The standard deviations 1n Table'III are all measures of precision,
either of the % differences or of the absolute % differences. The standard
deviation of the absolute % difference 1s always smaller than the standard
deviation of the % difference. Note also that the Radian-vs-Radian overall
(or pooled) standard deviations are less than the corresponding EPA-QAD-
vs-Rad1an standard deviations or the EPA-ASRL-vs-Rad1an standard deviations.
This follows because 1n the Rad1an-vs-Radian comparison, the standard
deviations (or precisions) are the within-laboratory precisions. For the
EPA-QAD-vs-Radian, and the EPA-ASRL-vs-Radian comparisons, the standard
deviations Include both the between-laboratory (or between-method) and the
within-laboratory components of the precision.
The within-laboratory and between-laboratory components of bias and
precision are fixed factors in this analysis; i.e., they apply to the
particular laboratories involved, Radian (Perimeter Park, NC), EPA-QAD
(Research Triangle Park (RTP) NC), and EPA-ASRL (RTP, NC) laboratories 1n
particular, and not to all laboratories 1n general.
Table IV shows the separation of the between-laboratory standard
deviations (precisions) from the standard deviations given 1n Table III for
the EPA-QAD-vs-Rad1an and the EPA-ASRL-vs-Rad1an comparisons. In all cases
the within-laboratory standard deviations (precisions) were assumed to be
equal to the Rad1an-vs-Rad1an standard deviations (from Table III). The
between-laboratory standard deviations are larger than the analytical
precisions, the within-laboratory precisions.
The mean absolute % difference for repeated analyses In the Radian
Laboratory was 9.8% with a standard deviation of 12.2%. The average %
difference of successive analyses 1n the Radian laboratory averaged -0.354%
with a standard deviation of 15.7%. These values were used for the analytical
precision of the PDFID method.
Between-laboratory % differences were -8.9% for the EPA-QAD laboratory,
and +2.0% for the EPA-ASRL laboratory, both compared to the Radian laboratory,
with standard deviations of 16.3% and 19.5%, respectively. Between-laboratory
absolute % differences were 15.0% for the EPA-QAD laboratory and 13.4% for the
EPA-ASRL laboratory, both compared to the Radian laboratory, with standard
deviations of 14.4% and 15.6%, respectively.
273
-------
Sampling and Analytical Precision
Overall precision, including both sampling and analysis variability, was
determined by analyses of duplicate samples, simultaneously collected 1n two
Identical canisters. The % differences between duplicate sample analyses and
absolute % differences, means and standard deviations are given 1n Table V.
The mean % difference 1s small (-0.263%), as expected, and the standard
deviation of the % difference is 20.139%. The standard deviation of the %
difference is a measure of precision of the duplicate samples and their
analysis and include the between-duplicate variability and the within-
duplicate variability, or analytical error. The expected absolute %
difference for duplicates 1s 11.690% with a standard deviation of 16.433%,
again Including between-duplicate variability and analytical error (or
analytical precision).
Separation of the between-duplicate variability and the analytical is
Indicated in Table V. The calculation for separating the between-duplicate
variability from the analytical variability is similar to the calculations
illustrated 1n Table IV. The assumption Is made that the Radian analytical
precision estimated in Table III applies also to the duplicate sample results.
The between-duplicate variability (or precision) 1s slightly less than the
between-laboratory and between-method variabilities estimated 1n Table IV.
Accuracy
Because the NMOC measurements encompass a range of mixtures of unknown
compounds, 1t was not possible to define absolute accuracy. Accuracy was
determined relative to propane standards with Internal and external audit
samples.
Accuracy for the Radian measurements was monitored throughout the
program. Four days per week a propane sample was prepared 1n-house and
analyzed. The propane used to prepare the in-house quality control (QC)
standard was certified by the EPA-QAD and was referenced to the National
Bureau of Standards, Standard Reference Material 1667b, propane. In-house
propane QC samples- were prepared by diluting the certified propane with
cleaned, dried air. Percent difference between the measured NMOC concen-
tration and the calculated KMQC concentration for the in-house QC standards
showed an overall mean percent bias among Radian channels of 1.18%, ranging
from 0.91% to 1.44% for the four Radian channels. Percent bias was defined as
% bias - ((NMOC (measured) - NMOC (calculated))/NMOC (calculated)) x 100
The results show excellent quality control for each channel considering that
the bias results include errors caused by dilution of the QC sample in
addition to the analytical error.
External audit results were referenced to the EPA-QAD PDFID instrument.
Table VI shows bias figures for all three years of the program for Radian
Channels A,B,C, and D, and for the EPA-ASRL GCFID channel. The Radian percent
biases were all below 5% and were negative, showing the EPA-QAD NMOC
measurement to average higher than the Radian measurements. The EPA-ASRL bias
measurements are positive and are less than 6%. The EPA-QAD NMOC measurements
for propane average greater than those for EPA-ASRL.
274
-------
Completeness
Completeness was defined as the percent of the total scheduled samples
that were successfully taken and analyzed, beginning with the first valid
sample from the site. Completeness 1n 1986 from 23 sites ranged from 87 to
100%, averaging 96.8%. Completeness 1n 1985 averaged 95.8%, and 90.6% in
1984.
Other Results
In 1985, measurements were taken to determine whether the length of
storage of the ambient air samples into the canisters affected the measured
NMOC concentration. For time periods up to 14 days after sampling, no
discernible concentration difference was detected in 26 site samples selected
at random. In the NMOC program, samples were generally analyzed less than 50
hours after sampling, but never more than 80 hours after sampling.
Tables II through VI are available upon request by contacting
Dr. Robert A. McAllister, Radian Corporation, P.O. Box 13000, Research
Triangle Park, NC 27709. (919) 541-9100.
References
1. U.S. Environmental Protection Agency, "Uses, Limitations, and
Technical Basis for Procedures for Quantifying Relationships between
Photochemical Oxidants and Precursors," EPA-450/2-17-021a
(November 1977).
2. U.S. Environmental Protection Agency, "Guidance for Collection of
Ambient Nomethane Organic Compound (NMOC) Data for Use 1n 1982
Ozone SIP Development," EPA-450/4-80-011 {June 1980).
3. U.S. Environmental Protection Agency, "Cryogenic Preconcentration and
Direct Flame Ionization Detection (PDFID) Method for Measurement of
Nomethane Organic Compounds (NMOC),* Environmental Monitoring Systems
Laboratory, Research Triangle Park, NC
EPA-600/4-85-063, October 1985.
4. R. McAllister, D-P. Dayton, D. Wagoner, "Nonmethane Organic Compounds
Monitoring," Final Project Report, Radian Corporation, Prepared for U.S.
Environmental Protection Agency, Research Triangle Park, NC, EPA contract
No. 68-02-3889, January 1986.
5. R. McAllister, R. Jongleux* D-P. Dayton, P. O'Hara, D. Wagoner, "1986
Nomethane Organic Compounds Monitoring," Final Project Report, Prepared
for U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
Contract No 68-02-3889, March 1987.
6. Radian Corporation, "Nonmethane Organic Compounds Monitoring Assistance
for Certain States 1n EPA Regions III, IV, V, VI, and VII, Phase II,H
Final Project Report. Prepared for the U.S. Environmental Protection
Agency, Research Triangle Park, NC, EPA Contract No. 68-02-3513, February
1985.
275
-------
7. R. McAllister, D-P, Dayton, D. Wagoner, F. HcElroy, V. Thompson,
H. Richter, "1984 and 1985 Nomethane Organic Compound Sampling and
Analysis Program," Proceedings of the 1986 EPA/APCA Symposium on
Measurement of Toxic A1r Pollutants, Raleigh, NC, April 1986.
8. D-P. Dayton, R. McAllister, D. Wagoner, F. McElroy, V. Thompson,
H. Richter, "An Air Sampling System for Measurement of Ambient
Organic Compounds," Proceedings of the 1986 EPA/APCA Symposium on
Measurement of Toxic Air Pollutants, Raleigh, NC, April 1986.
TABLE I. HMOC CONCENTRATIONS AT SITES PARTICIPATING MORE THAN ONE YEAR
NWOC
Ambient Air Concentration,
PpmC
1984
1985
1986
Site
Code
Mtan
Median
Number Mean
Median
Number
Mean
Madlan
Nuabex
Beaumont, TX
BHTX
O.B9
0,715
86
1,769
1.679
93
0.796
0.636
89
Dallas, TX
DLTX
0,97
0.905
74
0 .856
0.731
82
0.724
0.648
57
EL Ptso, TX
ELTX
0.93
0.820
69
0.707
0.662
90
0.486
0.413
88
Fort Worth, TX
FWTX
0,97
0,830
69
0.742
0.612
84
0.630
0.568
87
PhLladilphli. PA*
P1PA
1.02
0.920
63
0.657
0.490
71
0.450
0.342
69
Washington, DC
WDC
0.61
0 < 710
63
0.687
0.607
63
0.351
0.289
71
Atlanta, 6A
ATGA
0.79
0.600
55
—
-
--
0.544
0,451
84
BlcmLnghaA, AL
BHAL
0.99
0,705
56
"
-
-
1.019
0.582
92
Clue*, TX
CLTX
0.B2
0.610
72
0.741
0.645
89
"
-
Houston, TXb
H1TX
--
-
-
0.924
0.752
77
1.117
1.011
»1
Kansas City, HO
KCKO
0,79
0.625
68
0.535
0 * 424
92
-
--
-
Wast Gran**, TX
ORTX
0.69
0,650
71
0.586
0.539
92
-
-
-
Richmond, VA
RVA
0.53
0.495
64
0.539
0.450
67
-
-
Texas City, TX
TCTX
0.92
0.780
69
0.603
0.433
&6
--
^Museum location.
bCAHS 1.
276
-------
AUTOMATED MEASUREMENT OF NONMETHANE
ORGANIC COMPOUND CONCENTRATIONS IN AMBIENT
Dave-Paul Dayton, Joann Rice
Robert A. McAllister, Alston L. Sykes
Radian Corporation
Research Triangle Park, NC 27709
Frank F. McElroy, Vinson L. Thompson
Environmental Monitoring Systems Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
An automated preconcentration direct flame ionization detection (APDFID)
method for measuring nonmethane organic compound (NMOC) concentrations in
ambient air has been developed and tested 1n the Radian Laboratory.
The prototype automated system Incorporated the principles of the manual
preconcentration direct flame Ionization detection (PDFID) method and obtained
and analyzed four 9-minute integrated ambient air samples per hour. Bias was
determined on 17 field samples relative to the NMOC concentrations determined
by the PDFID method. The average bias was -6.2%, with an absolute bias of
10.7%.
The relative responses of the APDFID and PDFID methods were determined
for several types of organic compound standards, measured 1n terms of the
ratios of the slopes of the calibration curves for the organic compounds to
the respective slopes for propane. The ratios ranged from 0.953 to 1.422.
Dry standard propane {2.84 ppmC) was humidified by bubbling through
HPLC-grade water. The concentration of the humidified propane was calculated,
assuming that the dry standard had been saturated with water vapor. The
concentration of the humidified propane standard, measured by the APDFID
method showed a bias of -3.22% compared to the calculated concentration.
277
-------
Introduction
Measurements of ambient NMOC concentrations are used by states to develop
emission control measures needed to achieve compliance with the National
Ambient Air Quality Standards for ozone. During the Summers of 1984, 1985, and
1986, analysis of nearly 5000 ambient air samples have confirmed the
usefulness of the PDFID method.
Although accurate and efficient, the PDFID method Is based on manual
operation of the NMOC analytical system to analyze individual air samples. To
monitor diurnal changes In NMOC concentration 1n ambient air and to help
identify NMOC sources, an automated method 1s needed to permit repetitive,
unattended analyses of ambient air for NMOC at field monitoring sites.
This paper describes the development of an automated POFIO instrument for
measuring NMOC concentrations in ambient air. The instrument, operating
procedures, and calibration results with propane and other organic compounds
are described. The effect of water vapor on the system and a comparison of
automated NMOC measurements to manual PDFID results are discussed, A number of
problems have arisen in the course of the development of the instrument for the
automated NMOC concentration measurement. One of the major problems and its
possible solution 1s discussed.
Apparatus and Operating Procedures
While the automated instrument and its operating procedures are still
under development, the initial studies demonstrated that current technology 1s
available to enable automated NMOC measurements successfully and reliably.
The APDFID NMOC measurement is similar to the PDFID NMOC measurement, in
that 1t 1s basically a cryogenic preconcentratlon direct flame Ionization
method. An objective of the APDFID development program was to sample ambient
air 60% of each hour. The analysis instrument selected 1s a programmable
VARIAN Model 3400 gas chromatograph utilizing a modified on-coluenn Injector.
This instrument is operated with a cycle consisting of 9 minutes for ambient
afr sampling and 6 minutes for analysis, resulting in 4 cycles per hour and a
sampling fraction of 60% of each hour.
A six-port valve, mounted in the oven of the chromatograph, routes the
ambient air sample through the cryogenically cooled trap, during trapping, and
routes the carrier gas through the electrically heated trap to the flame
ionization detector during analysis. The APDFID system incorporates an
on-column injector, modified to accept the trap assembly. The trap assembly
consists of a brass block containing the trap, a heater, and a temperature
sensor within the block. The trap assembly is located in an insulated housing.
A solenoid valve controls the flow of liquid nitrogen to the nozzle directing
cryogen spray to the outside of the trap assembly. Both the trapping
temperature (-170 C) and the desorption temperature (130 C) are regulated using
active temperature, control.
278
-------
A vacuum pump pulls the ambient air through the system, and a flow
controller regulates the sampling rate.
The beginning of a sampling/analysis cycle may be chosen as the point at
which the previous sample has emerged from the trap and analysis has been
completed. At that point the trap/block temperature 1s being controlled at
130+1 C, and helium carrier gas passes through the trap to the flame ionization
detector. At that time, cryogen (liquid nitrogen) 1s sprayed onto the
trapplna/desorptlon block, and the setpolnt for temperature control 1s changed
to -170 C. When the temperature of the trap has been controlled at -170 C for
about 30 seconds, the $1x-port valve is switched to draw a sample of ambient
air at 12.1 ml/min for 9 minutes Into the trap, which 1s packed with glass
beads.
The six-port valve Is then switched to bring 1n the helium carrier gas, in
the opposite flow direction, and at the same time, the setpoint of the
trapping/desorptlon block temperature 1s changed to 130 C. The block
temperature is Increased at the rate of 150 C per minute and the released,
revolatHized organic compounds are measured by the FID and the response Is
Integrated. The block temperature 1s then held for about 30 seconds to be sure
that all the organic compounds (and water} 1n the 9-minute sample have been
flushed from the trap. A new cycle Is then ready to begin.
Results
The APDFID Instrument was calibrated dally with two propane standards and
zero air. The propane standards were certified by the U.S. EPA, Quality
Assurance Division (QAO), referenced to a National Bureau of Standards propane
Standard Reference Material No. 1667b. Three types of tests were conducted.
(1) Tests were done to compare the measured APDFID Instrument response for
several organic compounds with propane response. (2) Several canister a1r
samples for which the NMOC concentration had been determined previously by the
PDFID method were analyzed by the APDFID method to compare the results.
Finally, (3) measured concentrations of dry propane and humidified propane were
compared to determine the bias that the presence of moisture has on the APDFID
Instrument.
The organic species results are given 1n Table I. Calibration curves were
generated for the six compounds shown 1n the table, using three concentrations
of each (1.0, 3.0, and 6.0 ppmC) diluted from the original cylinder
concentration. Each concentration was replicated three times. Table I gives
the slopes of the linear regressions resulting from the calibration data for
each compound. The third column 1n Table I gives the slope of the propane
calibration curve that was done the same day as the compound calibration curve.
The ratio of the compound slope to the propane slope, given 1n the
last column, compares the relative response of the compound with the response
of propane 1n the APDFID system. Ideally, all of the ratios should be 1.000,
Indicating that all of the organic compounds produced a response equal to
propane, which 1s the compound of choice for calibrating the NMOC PDFID
instrument and the APDFID instrument. The ratios for propylene and butadiene
were considerably higher than 1.0, which 1s not typical for the PDFID method.
Further work 1$ anticipated to determine the cause of these unexpected results.
279
-------
Analyses of ambient air samples by both the PDFID method and the
APDFID method are reported in Table II. Sample ID Numbers 2899 through 2933
were site samples collected during the 1986 NMQC Monitoring Program and
analyzed first by a Radian PDFIO channel in September 1986 and then analyzed by
the APDFID between November 24, 1986 and December 12, 1986. Sample ID numbers
with an MAB suffix were locally collected and analyzed first on an
EPA PDFID instrument and then on the APDFID Instrument. The NMOC
concentrations are reported in parts per million carbon (ppmC) by volume.
Percent bias in Table II averages -6.2, relative to the PDFID concentration
measurement; absolute % bias averages 10.7.
To determine the effect of water vapor on the APDFID measurement, wet and
dry propane were used in separate measurements. Dry standard propane with a
concentration of 2.84 ppmC was used to determine a calibration factor for
the APDFID instrument. Dry standard propane (2.84 ppmC) was then humidified by
bubbling through HPLC-grade water. The concentration of the humidified propane
was calculated, assuming that the dry standard had been saturated with water
vapor. The concentration of the humidified propane standard, measured by the
APDFID method showed a bias of -3.22%, compared to the calculated
concentration.
Discussion and Recommendations
The preliminary results to date indicate the feasibility of an automated
system to make measurements of NMOC concentrations in ambient air. Both the
instrument and operating procedure are still 1n the development stage.
One of the biggest problems encountered 1n the operation of the Instrument
was that occasfonalTy, 7Yqulof nitrogen would poo? at the bottom of the brass
injector block and would collect between the temperature sensor and its ceramic
packing. This would permit the top of the brass block to heat to a temperature
higher than the control setpoint of -170 C, while the temperature sensor and
the bottom part of the column/block would remain at liquid nitrogen
temperature (-195.8 C). It 1s felt that the temperature control problem can be
solved by adding a second cryogen Injection port 180 from the present
injection point, by enlarging the venting port between the column block and the
chromatography oven, and by adding a small heater in the bottom of the cavity
where the liquid nitrogen is currently collecting.
The data quality may also be improved by optimizing the gas sampling rate
to assure that enough sample is being trapped to give reproducible results with
improved accuracy.
280
-------
Table I. ORGANIC SPECIES RESULTS
Compound
Compound
Slope
Propane
Slope
Compound/Propane
Slope Ratio
WIG
Propylene
468,426
334,665
1.400
Benzene
354,773
350,829
1.011
Ethane
327,805
344,117
0.953
m-Xylene
366,863
344,996
1.063
1,3-Butadiene
521,729
366,988
1.422
Table II.
BIAS OF THE NMOC CONCENTRATION
PDF ID
APDFID
Sample
NMOC,
NMOC,
Absolute
ID
QDKlC
OurnC
% BUS®
% Bias
2899
0.459
0.497
8.3
8.3
2899
0.459
0.513
11.8
11.8
2900
1.679
1.407
-16.2
16.2
2924
1.319
1.324
0.4
0.4
2924
1.319
1.317
-0.2
0.2
2928
0.647
0.382
-41.0
41.0
2928
0.647
0.721
11.4
11.4
2930
1.053
1.095
4.0
4.0
2930
1.053
1.003
-4.7
4.7
2931
0.210
0.178
-15.2
15.2
2933
0.876
0.598
-31.7
31.7
2485A
0.512
0.502
-2.0
2.0
2899A
2.023
1.769
-12.6
12.6
2928A
2.024
1.798
-11.2
11.2
2929A
0.508
0.467
-4.1
4.1
2929A
0.500
0.515
3.0
3.0
2930A
0.496
0.472
-4.8
4.8
Average -6.2
"TCTT"
% Bias - [(APDFID NMOC - PDFID NM0C)/(PDF1D NMOC)] x 100
281
-------
ANALYSIS OF VOLATILE ORGANIC SAMPLING TRAIN SAMPLES
USING MEGABORE GAS CHROMATOGRAPHY/MASS SPECTROMETRY
T. A. Buedel, J. T. Bursey,
J. B. Homolya, R. L, Porch
Radian Corporation
Research Triangle Park, NC
27709
R. G. Fuerst, T. J. Logan, M. R. Midgett
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Analysis of volatile organic sampling train (VOST) samples by capillary gas
chromatography results in increased chromatographic resolution for the
characterization of products of incomplete combustion in stack emission
samples. A megabore column was selected to provide the highest sample
capacity for the wide dynamic range of compound concentrations encountered
in VOST samples coupled with the best possible chromatographic resolution.
VOST samples on Tenax and/or Tenax /charcoal cartridges are thermally
desorbed onto the analytical trap of a commercial purge and trap apparatus,
from which the organic compounds are desorbed directly to the head of a
megabore column operated initially at subambient temperatures. The VOST
target compounds are identified and quantified by a computerized gas
chromatograph/mass spectrometer system. All of the VOST target compounds
exhibited response factors with coefficients of variation of less than 25%
when using triplicate analyses at four concentrations. The megabore column
analysis time for the VOST target compounds 1s approximately 30 percent
faster than the present VOST protocol where a packed chromatographic column
1s used.
282
-------
Introduction
According to the Resource Conservation and Recovery Act (RCRA), owners or
operators of hazardous waste Incinerators must operate these facilities 1n
a manner that does not endanger human health or the environment. The Code
of Federal Regulation (CFR), Title 40, Part 264, states that an Incinerator
must achieve a Destruction and Removal Efficiency (DRE) of 99.99% for every
Principal Organic Hazardous Constituent (POHC) described 1n the Trial Burn
Permit. In the present methodology for the sampling and analysis of
samples taken of the emissions from hazardous waste Incinerators, samples
of stackRgas are collected on a set of sorbent cartridges, one containing
Tenax GC and the other Tenax GC combined with charcoal. The cartridges
are thermally desorbed through 5 mL of water and the organic compounds are
collected on an analytical sorbent trap. The adsorbed organlcs are then
desorbed onto the head of a 3.0 mm x 2 m glass column packed with IX
SP-1000 on Carbopack B and analyzed by a computerized gas chromatograph/
mass spectrometer system.
The analytical methodology for VOST focuses on target compounds In order to
determine the ORE. The key questions 1n a trial burn are, of course: Has
the incinerator performed successfully 1n the combustion of the compounds
of interest? Have these compounds been effectively removed {99.99% or
better) from the emissions? For the limited number of compounds of
Interest, analytical standards can be analyzed to determine retention times
accurately and, unless compounds with Ions of exactly the same masses
coelute exactly, the target compounds can be deconvoluted from potentially
Interfering species. However, when the question of possible formation of
PICs 1s being addressed, the PICs are not target compounds, as their
Identity Is not known 1n advance. The analyst 1s required to perform
qualitative analysis and the quality of the mass spectrum obtained 1s
critical to the success of the mass spectral Interpretation. Automated
deconvolutlon techniques can be untrustworthy and extensive use of manual
mass spectral deconvolutlon techniques is time-consuming and requires
extensive training and experience on the part of the analyst.
Improving the analytical process by Improving the chromatographic
resolution is one route to the production of mass spectra of better
quality: Improved chromatographic resolution decreases coelutlon of
compounds and the mass spectra hence exhibit fewer interfering peaks. The
megabore capillary column (0.53 mm diameter) has sample capacity
approaching that of a packed column while retaining much of the peak
resolution traits of a narrow or wide-bore capillary column. The resulting
improvement in chromatography 1s dramatic: Figure I shows a typical VOST
sample analysis using a packed column, while Figure II shows a comparable
field sample analyzed on a megabore column. An additional benefit of the
use of the megabore column Is evident upon the examination of Figure II:
In the same elution time as the packed column analysis, the megabore
analysis shows the elution of naphthalene and dlchlorobenzenes.
Naphthalene will not elute from the packed column 1n analysis times of an
hour or more; the dlchlorobenzenes do not resolve completely on the packed
column. On the megabore, the three dlchlorobenzene Isomers exhibit base-
line resolution and naphthalene elutes in approximately 20 minutes. The
ability to analyze compounds with good chromatographic resolution on the
megabore 1s limited only by the ability to purge these compounds from the
sorbent cartridges and the water successfully and quantitatively.
283
-------
Experimental
Two GC/MS configurations were used to evaluate the megabore VOST
methodology. Both systems used a quadrupole mass spectrometer. One
system, the Finnigan-MAT 4500 has differential pumping, using diffusion
pumps. The other system, the Finnigan-MAT 5100, has no differential
pumping but uses a turbomolecular pump. The chromatographic column was the
same on both systems, a DB-624 (J&W Scientific), 30 m long with a film
thickness of 3 microns. The column was operated at a flow rate of 2-2.5
ml/min, over a temperature range of -20 C to 295 C, and coupled directly to
the ion source of the mass spectrometer. The subambient temperatures are
used to obtain chromatographic resolution of gaseous target compounds.
Although the VOST method as presently formulated is restricted to compounds
in the boiling point range of 30 C to 100 C, there is widespread interest
in the quantitation of chloromethane, bromomethane, chloroethane, and vinyl
chloride as PQHCs and PICs. Therefore, a goal of the development of the
megabore technigue was to obtain analytical data for the gases with boiling
points below 30 C. The composition of the analytical column of the purge
and trap apparatus was varied as an experimental parameter, since the
analytical trap described in the present VOST protocol entrained too much
water for successful use with the megabore column. The analytical trap
ultimately used was composed of 90% Carbotrap /10% charcoal (w/w), using a
1 minute dry purge on the 5100 but no dry purge on the 4500 system.
Results and Discussion
Using the EPA Method 624 purgeable standards (commercially available),
target compounds were loaded on and purged from a Tenax /charcoal
cartridge. Relative response factors were calculated for the compounds of
interest relative to dg-benzene, in accordance with the present formulation
of the VOST method. Iromofluorobenzene and d.-l,2-dichloroethane were
spiked on the cartridges as surrogates. Reproducibility of these response
factors, as expressed by the relative standard deviation of replicate
determinations over the calibration range, was used to evaluate the
acceptability of the method. Results for both systems, with the target
analytes at a level of 100 ng, are shown in Table I. Coefficients of
variation for the VOST target compounds are all 1n a reasonable range
(5-22%). The coefficients of variation for the gases are higher but all of
the gases are observed. Analytes, surrogates, and quantitation standards
were again spiked onto VOST sampling cartridges (Tenax /charcoal
composition) and thermally desorbed Into the analytical apparatus. The
calibration curve was prepared using an EPA Method 624 standard in tripli-
cate on both systems, with the analytes at concentrations of 25, 50, J0Q,
and 150 ng; the results for both systems are shown in Table II. Repro-
ducibility of the calibration curve for each mass spectrometer, expressed
by the coefficients of variation, is comparable. The chromatographic
peaks, even at the longer retention times, tend to be 8-14 seconds wide.
The goal of improved chromatographic resolution is met with the use of the
megabore fused silica capillary column, and the reproduclbil1ty of the VOST
method using the packed column for chromatographic analyses has not been
sacrificed for the improvement of chromatographic resolution. In addition,
the time required for performance of the chromatography for the VOST assay
is decreased by approximately 30%.
As a final test of the modified method, a blind quality control sample was
prepared and analyzed in triplicate, using the VOST target compounds. The
results are shown in Table III. Reproducibility for the triplicate
analysis was acceptable and all values obtained were within 14% of the
accepted value.
284
-------
Conclusions
The VOST analytical methodology 1s enhanced by including megabore capillary
analysis. Chromatographic resolution is Improved, so the Initial goal of
improving the characterization of PICs should be attainable when the method
1s applied to the analysis of actual field samples. Analysis times using
the megabore column are shorter by approximately 30% by comparison with the
time required to perform the analysis using the packed column. Sensitivity,
as determined by estimated limits of detection in both the packed column
and megabore modes, is improved because the megabore column Is coupled
directly to the Ion source, so the entire sample reaches the Ion source of
the mass spectrometer. Also, capillary chromatographic peaks are sharp, so
the concentration of the compound of interest in the 1on source is higher
at the capillary chromatographic peak maximum. Reproducibility 1s acceptable
for the analytes which are within the purview of the method {boiling point
of 30 C to 100 C), and all of the gaseous compounds of interest are observed.
Several field tests are essential to evaluate the modified methodology with
actual field samples. Parallel tests with the standard methodology and the
modified methodology would be desirable.
This work was supported by the Environmental Monitoring Systems Laboratory,
Office of Research and Development, U, S. Environmental Protection Agency,
under EPA contract No. 68- 02-4119.
Disclaimer
Although the research described in this article has been funded wholly or
in part by the United States Environmental Protection Agency under contract
to Radian Corporation, 1t 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.
Acknowledgements
Contributions by D. Wagoner, N. Hooper, and T. Moody (Radian Corporation)
and E. Johnson (Flnnigan Corporation) are gratefully acknowledged.
References
1. E. M. Hansen, "Protocol for the Collection and Analysis of Volatile
POHCs Using VOST," EPA Report 600/8-84-007, March, 1984.
285
-------
Mil MKIAQKV UCntWB *24 (QflOMI
DRKKHD PNH IttWr^OMWXM. CMRID98
FiitiljgvUg 5100
Ion
fcfnrM sented Oatfflctnc of
Iteox' tarlititft v«rU£Lor. (%r «-#»-
Scmttrd Co*ffiei«* of
DwUtlm VnrUckn (%)
rfcdyl** dhlarUi
•ft
9.C3T
0.0*05
U
l.Xtt
0.170
14
Chlorates
13
0.6160
o.mi
22
0.747
0.074
10
1, 1,1'IttchleteediEW
97
ow
0.0152
21
0.524
0.061
«
Ctrtxo Machloddi
U7
0.3177
0.0665
21
0,321
0.014
5
tkldilerMthm
UO
0.2460
0.027*
10
0.311
O.tt*
1
Inn
71
1.4620
0.2743
19
2.23$
0.486
20
TaoradifaKMtyljra
144
0.TK7
0.0979
K
0.320
0.0M
15
ToUar*
92
0,3717
0.0366
10
0.915
0.111
14
ftlocafcaecari
112
1.9090
0.1A29
I
0,771
0.0(7
9
MyUmm
1»
1.0617
0.1079
10
0,434
0.032
4
f, -p-tylm
106
3.0300
0.3013
10
1.227
0.107
9
206
1.4567
O.U2t
«
0.325
0.071
15
ChWowtm
»
0.2190
0.1190
14
2.146
1.U6
54
¦nenedem
*
0,0360
0.0903
96
o.m
0.0(3
23
VUyl dOorl*
62
0,3090
0-0Q34
U
0.4M
o.oes
6
ChLacOMhr*
64
<7.0953
0.0390
*3
0.272
0.074
27
l.l.?.2-T»craddae»-
•tfvna
93
0.7360
0.0367
S
0.22S
0.067
30
286
-------
T»ku it. utraiti r*cioa CALinun«t com raa tiucw nvtino ut covowot
MfQIUMft PIMM tnUR /1
• »••
• ¦•M
tl
n
».)«
• »ll
M
t.Wt
om
l»
diltrtWuiM
ii*
itn
Ml?
1«
MM
0.10S
14
Mnl^WHW
irt
iJfcfl
*.»»
H.
t.tu
U
a,
199
9.Ml
t.SM
1«
1.000
Ol'l
10
IM
i.m
tm
10
O.fckO
o.o»i
Otlur C«—mil
(kkwaitluai
SO
o w
«.m
tl
J.OJO
t.M
«4
«.»?*
t.tu
«
t.m
*1
Vinyl tkUfM*
II
0 14*
o.tu
' 30
«.*««
• .Ml
IS
AltmUviM
««
0.04)
n
0.0»
»
l.l.l.l'lotiathiara*
lltUM
11
«.W1
OUT
w
»Ui
toil
30
4
*1
• ¦•17
•
oiu
0.01?
I?
lrmtLwitWmM
n
»,»U
u
0-1J0
• or*
2)
^¦Mm*w (lit
**
1.00ft
1000
...
Tri»Ut«t* Mulrtaa «t r*ur umimwIwi », »•« 190, it$ *
4CMfU«i«f» •* njiMiM • itmf-rt nrtiinm ¦ iw
Hans
T*bl. 1(1. BLIND QUALITY CONTROL RECOVERY TEST RESULTS
Avtrtgi
Avtrtgt2
Amount
Amount
Sttndard
P»rc*nt
Compound
Injtcttd
-------
mc'
4i
I
fl
UJWI
¦
«00
1M»
MM
1000
1132
1300 lean
37:30 TWna
Figure 1. Typical Analysis ol a VOST Sample Using an SP-1000 Packed Column
mo
mc-
El
900
TS2
ioi 11 900
1S:M MX
MS Sua
If* TfclM
Figure 2, Chromatograrr si a Field Sample Desorbed From a Tenax* Charcoal
Cartridge and Analyzes Using the Megabore In the 5100 QC/MS Analytical
System
288
-------
PRECONSTRUCTION MONITORING OF
AMBIENT AIR FOR TOXIC NON-CRITERIA
POLLUTANTS AT A MUNICIPAL
WASTE-TO-ENERGY SITE
Larry D. Ogle, David A. Brymer and
Maureen P. Kilpatrick
Radian Corporation
8501 MoPac Boulevard
P.O. Box 9948
Austin, Texas 78766
Ambient air monitoring for toxic compounds prior to the
construction of a Waste-to-Energy facility was used to establish
background concentrations. The data will be used in support of the
design and permitting of the facility. Ambient levels were established
by sampling six consecutive days each quarter for one year.
The non-criteria pollutants of particular interest in this study
included metals, vapor phase chloride and fluoride, polychlorinated
dibenzofurans and dibenzodioxins, polychlorinated biphenyls, polynuclear
aromatic hydrocarbons, and volatile organics. Methodology used was
designed to provide data at levels low enough to adequately define
ambient concentrations. Multiple sampling and analysis techniques were
compared to identify the most useful technique for each analyte. Field
blanks and field Bpikes were used to define the applicability of each
method for the target analytes.
Very low levels of the non-criteria pollutants were observed in the
ambient air. Sample volumes and detection limits established during
this project were sufficient to characterize the compounds of interest
at ambient levels.
289
-------
Introduction
Disposal of municipal waste is an area of increasing concern,
particularly for larger municipalities. One of the options for disposal
of combustible waste is incineration since incineration can reduce
landfill demand by as much as 90%. The resulting heat energy then can
be used to generate electricity providing an additional incentive for
incineration. Waste-to-Energy facilities often are sited close to the
source of refuse to minimize transportation costs, however, this proxim-
ity to heavily populated areas has raised concerns about the environ-
mental impact of such a plant on the surrounding neighborhoods.
The objective of this study was to obtain information on the
baseline (i.e., background) levels of non-criteria pollutants in the
ambient air prior to construction of a waste-to-enetgy facility, A
monitoring site was established approximately four blocks from the
construction site of the Hennepin County WaBte-to-Ensrgy facility in
Minneapolis, Minnesota. The collected data included meteorological,
criteria pollutant and non-criteria pollutant measurements. These data
will be used in support of the design and permitting of the facility.
Meteorological and criteria pollutant data were collected contin-
uously for one year. Non-criteria data was collected during six consec-
utive days each quarter during that year. Quarterly sampling was
desirable to determine if appreciable seasonal variations, both in
content and concentrations, were observable. The non-criteria pollu-
tants of prime interest in this study were: chlorides (as HC1), fluo-
rides (as HF), metals (Sb, As, Be, Cd( Cr, Cu, Hg, Mn, Mo, Ni, Se, Ag,
Sn, V, Zn and Fb). polychlorinated dibenzodioxins and dibenzofursnB
(with emphasis on the 2,3,7,8-tetrachloro-iaomera), polychlorinated
biphenyls, polynuclear aromaties, and volatile organics (with emphasis
on vinyl chloride, benzene, tetrachloroethylene and the dichloroben-
zenes).
Experimental Methods
The design of the experimental methods for the measurement of the
non-criteria pollutants was responsive to multiple objectives: 1)
comparing the performance of ambient air monitoring techniques to
determine which method was the best for the specific needs of this
study; 2) obtaining sufficient data to determine if ambient levels of
some parameters were low enough to delete them from further monitoring
during the course of the program to affect some cost efficiency for the
project; 3) utilizing screening techniques whenever possible to obviate
the need for detailed and expensive analysis if analytes ware not
present at detectable levels; 4) utilizing methods which provided data
on analytes not of specific concern to the project but of potential
interest to the community at large (i.e., broad applicability of method-
ology); 5) utilizing methods which not only were generally accepted but
also were capable of providing data at concentration levels low enough
to adequately define the non-critaria baseline; and 6) defining an
appropriate number of field and laboratory blanks and spikes that the
limitations of the methods could be known.
A test plan was developed which provided sampling and analysis
methods to address the objectives described above. A summary of the
techniques used to measure the non-criteria pollutants is provided in
Table I.
290
-------
Results
The use of the methods (described in Table I) to monitor the
non-criteria pollutants has resulted in a large amount of data. In
general, very low levels of non-criteria pollutants were observed.
Observations on the methodology are: the GC-ECD screening technique for
PCBs was determined to be of limited use; the GC-FID screening technique
for PAHe resulted in a number of false positive identifications; the
XAD-2* resin was preferred for PCB collection over PUF* plugB due to
spike retention; the iopinger technique gave higher values for chlorides
and fluorides than the sorbent tube technique; Tenax* or combinations of
Tenax and Carbosieve S* were not effective in retaining low molecular
weight organics and were very prone to contamination.
Field spikes were used to validate the collection and retention
efficiencies of the methods. Field blanks were employed as a check of
background contamination and to ensure the levels observed were above
the background levels. From the instrument detection limits and the
amount of sample collected, s lower limit of quantitation has been
established. These levels and the field spike recoveries are presented
in Table II.
Conclusions
The data collected during the course of this project have been
sufficient to describe the baseline ambient air levels of non-criteria
pollutants. Sample volumes and detection limits were sufficient to
characterize the pollutants at or below levels of concern. In many
cases, however, the best method was not established. This is particu-
larly true in the case of fluorides and chlorides. Additional method
development and evaluation is needed to establish methods for the
measurement of such air toxica. This development will aid those working
in the area of ambient air measurements, the agencies with the responsi-
bilities to determine if new facilities will provide an environmental
burden, and those designing the facilities and related technology.
291
-------
TABLE I. SAMPLING AND ANALYTICAL TECHNIQUES TOR NON-CRITERIA POLLUTANTS
Target Compound
Group
Sampling
•Method
Analytical
Methods
Particulate Metals
Mercury
Poly c hi or ina t ed
Biphenyls
Polynuclear Aromatic
Hydrocarbons
Polychlorinated
Dioxins and Furans
Chlorides
Fluorides
Volatile Orgaaics
Hi-Vol Filter
Gold Amalgamation
Glass Fiber Filters/XAD-2
PDF Plugs
Glass Fiber Filters/XAD-2
Glass Fiber Filters/XAD-2
Impinger Solutions >
Silica Sorbent Tubes f
Impinger Solutions 1
Silica Sorbent Tubes J
Tenax i
Tenax/Carbosieve S '
Charcoal Tubes
Stainless Steel Canisters
ICAP, GFAA
Vapor Phase AA
GC-ECD
GC-MS
GC-ECD
GC-HS
HPLC-UV and Fluorescence,
GC-FID, GC-MS
GC-MS
Ion Chromatography
Ion Chromatography
Ion Specific Electrode
Thermal Resorption
GC-MS
GC-FID
GC-FID, PID and HE CD
-------
TABLE II. AVERAGE METHOD EFFICIENCIES AND QUANTITATION LIMITS
Target Analytea
Spiking Compound
Amount Spiked
(ug/aample)
Lover
Limit of
Mean Standard Quantitation
Recovery Deviation (ug/m)
PAHs
PCBa (XAD-2)
(PUF)
Nephthalene~d
8
Pyrene-djQ
13
Mono chio to C.bipfceny1
13
Tatraehloro C -biphenyl
13
Octachloro C.-biphenyl
13
Deeachloro Cj^biphenyl
13
Monochloro C, biphenyl
13
Tetrachloro C biphenyl
13
Octachloro C biphenyl
13
Deeachloro C^biphenyl
12
15
15
Q.10
0.25
0.40
0.50
0.10
0.25
0.40
0.50
88
68
65
68
68
68
63
84
63
92
9.0
6.7
24
21
23
23
7.3
0.5
13
7.8
0.1 - 160
0.004
0.0006
Polychlorinatad
Dibenzodioxini
37
C1-2.3.7.8-TCDD
0.015
94
8.5
4 x 10
Volatilea
(Tenaz)
(Tenax/Carboaieve 8)
(Charcoal)
Banaene-d^
Banzene~dg
Methylene chloride
Tatrachloroethylane
1,3-Diohlo robenxene
0.5
0.5
15
17.9
13.5
0
0
106
94
83
0.03 - 0.3*
0.03 - 0.3*
7 - 45*
Stainleaa Steel
Caniatera
2-7*
Chloridi
1.2
Fluoride
Hetale
Mercury
0.9
1.2 - 600*
1.0
'Compound dependent - deacrlbee the range experienced for the varioua analytea.
^Two determination!»
293
-------
GRAB SAMPLING AS AN EFFECTIVE TOOL
IN AIH POLLUTION MONITORIMG
Joseph P. Krasnec
Atmospheric Chemistry Research Group
Scientific Instrumentation Specialists
Moscow, Idaho
Grab sampling of gaseous chemical species has been utilized for a
number of years. Various rigid containers, polymeric film bags and other
devices are used manually, semi-automatically, or most recently with
microcomputer control. The advantages are simplicity, low cost, easy field
use and sample collection, and sample storabllity up to the analysis time.
Disadvantages are high manpower cost for manual sampling in larger field
experiments, sample degradation/contamination in some types of grab sampling
devices, and laok of data on stabllity/storability of a number of species of
interest such as VOC's.
The evolution of rigid grab sampling containers led to an increased use
of sampling devices made from specialized materials, i.e. passivated
stainless steel (PSS) resulting in successful sampling and storage of trace
and toxic gases down to the low ppt (v/v) level for extended periods of
time. Recent advances in the construction of PSS grab sampling containers
and syringes coupled with the improvements in sampling procedures, and
availability of solid research data on stability and storabllity of dozens
of organic toxic air pollutants make this approach to air sampling
increasingly attractive.
On-going automation of grab sampling procedures and subsequent use of
sophisticated, yet readily available analytical techniques such as capillary
gas chromatography (GC) using speclfio detectors (FID, ECD, PID, MSD), or
GC-Uass spectrometry (MS) give the researchers and regulatory personnel a
powerful tool for effective abatement and control of gaseous toxic air
pollutants.
294
-------
Introduction
The rapid industrial development during the past fifty years coupled
with significant population increases brought about a significant
deterioration of localized and global atmospheric environment. The onset of
the era of smog also signifies growing interest and concern about the
sourcesj fate and effects of various air pollutants. The early work in the
1950 * a involved relatively simple experiments, followed by a more thorough
laboratory work geared toward pollutant identification, atmospheric
reaotions and transformations, and exposure/effeots on humans and plants*
This work, in the sixties and early seventies reached a momentum that
brought about an increased public and regulatory agency awareness, concern,
and a clear need for additional theoretical work, research, testing and more
extensive field experiments»
One of the fundamental tasks facing researchers was the need to collect
and analyze samples of ambient or industrial atmospheres. The requirements
included minimal, or preferably no effect on sample composition and
integrity, ability to contain and transport the sample to the laboratory
(clearly the preferred ohoioe to taking; complex and delicate instrumentation
to the field site), low cost and ease of use, initial sampling work utilized
rigid glass, metal and plastic containers, polymeria film bags and various
adsorbents. As the need for more aoourate measurements at lower (sub-ppm)
concentrations intensified, some of the shortcomings of the commonly used
sampling methods beoame quite apparent. Contamination, sample degradation,
rapid loss of collected species thwarted attempts to carry out reliable,
larger scale field experiments. The need to measure background
concentrations of a number of trace atmospheric gases, low level air
pollutants and complex gaseous mixtures found in urban atmospheres resulted
in the prototyping and limited availability of new grab sampling devices.
Specifically, several ressaroh organizations independently developed
versions of a rigid, metal sampling container with different surface
treatments, i.e. electropolishlng, oxidation, coating and passivation (1).
These were used with good to exoellent results for ambient (10 ppt to 500
ppm v/v) sampling of halooarbons, hydrocarbons and a number of stable
atmospheric gases such as CQ2, CHI, N20, CO and others~ By aid-seventies
large number of eleotropollshed and passivated stainless ateel containers
were used for stratoapheria/tropospherio grab sampling of halooarbons from
airborne (ballon, aircraft) and surface (ship, mobile laboratory) platforms.
In ttai« tine period some shortcomings were encountered lit the use of
polymeric sampling bags made from polyethylene, teflon, and mylar (2) and
solid adsorbents (aotlvated charooal, molecular sieve, T«nax) for sampling
of aliphatic and aromatic hydrooarbons CC1-CI2). paasivated stainless steel
containers provided very satisfactory performance for these species.
Increased concerns about tbe ozone levels and Its photochemical reaotions
affecting air quality in larger urban areas intensified the need for
thorough samplIng/monltoring of noa-wthane hydrocarbons (NMHC'b). in the
past several years number of detailed studies oonflrmed validity of grab
sampling using passivated SS containers (3, it, 5, 6).
Grab sampler design and construction oriterla
Most tasks that require monitoring and/or measurement of gaseous
pollutants involve sample collection, storage and subsequent analysis* Air
pollution aontrol authorities need measurements cf a number of toxio gases,
including volatile organla compounds (voC's) to asaure acceptable air
quality levels. In addition, emission Inventory, human exposure data, and
industrial manufacturing prooeas control is needed for tha development and
295
-------
implementation of proper control strategies. Frequently, in-situ monitoring
and measurement of air pollutants is not feasible because of the cost and
complexity of the field deployment of a number of sensitive analytical
instruments. The grab sampling offers a viable alternative with lower cost,
simplicity, and flexibility of use. Properly collected, appropriate siae
single grab sample provides important information on a host of atmospheric
pollutants when analyzed in a well equipped, permanently based laboratory.
.....The most important requirements for the grab sampling containers are
non-contaminating, non-reactive and inert contact surface* small surface to
volume ratio; reusability/durability; optimal construction; relatively small
size and weight; easy sample collection procedures; acceptable cost; safety
and ease of transportation, and application for a wide variety of gaseous
and liquid mixtures.
The construction and pre-deployment handling of the grab sampling
containers, together with the user sampling/recycling procedures determine
the degree of success in using tha.n. While deceptively simple the grab
samplers evolved over a number of years from simple, "home-built" units to
sophisticated, commercially produced sampling instruments. The design
approach, selection of quality components and stringent assembly prrocedures
are of utmost importance. For example, proper forming of hemispheres from
selected grade of stainless steel sheet stock determines the degree of
success in subsequent preparation, cleaning and passivation prior to the
assembly. The passivation procedure is the key step in the sampler
manufacture. Much more effective than electropolishing the SUMMA passivation
process removes surface scale, decreases the oontaot area by smoothing the
SS surface, removes impurities from the stainless steel surface and
effectively oxidizes chromium and nickel in the material itself. The
cleanliness and inertness of the Summa passivated stainless steel surface
has been demonstrated repeatedly by storing ppt levels of trace gases for
several months or longer without any appreciable losses of the stored
species. With proper cleaning the "blanks" of the grab samplers are below
detectable limits of analytical techniques such as GC and GC-MS.
.....The welding procedure is another critical step in the grab sampler
construction. Great attention to proper procedures needs to be exercised,
i.e. amount of current/heat applied to the stainless steel hemispheres,
proper shielding with inert purge ga3, absence of oxygen from the welded
area, penetration/uniformity of the welds, and care in attaching other
components to the finished spherical or cylindrical sampling oontainers.
Complete avoidance of elastomeric materials is an Inherent feature of the
all-welded, or fitting equipped grab samplers. An extensive quality
control/quality assurance program must be an integral part of the entire
manufacturing process to ascertain problem free use. Visual and mechanical
Inspection at every step is supplemented by special leak testing of the
completed grab samplers with a sensitive (1x10-9 std. cc/sec.) mass
spectrometer based helium leak detector. Final cleaning and preparation
utilizes a quadrupole mass spectrometer residual gas analyser (BGA) with a
mass range up to 2^ V4u. This ia.Atraaent allows continuous monitoring and
recording of the entire cleaning process with data on any contaminants
present in fc'i'i system, or degasing from the grab sampling containers. If
necessary, temperature (100-400 C), and/or evacuation level (10-2 to 10-5
torr) can be adjusted to completely remove any undesirable apeaies from the
finished grab samplers. The final testing/blanking involves individual or
batch ECD/FID-GC analysis of the sampling containers readied for shipment.
Grab sampling methods
.....The design of most of the available grab sampling containers allows
296
-------
several sampling modes. The most common approach uses evacuated grab
samplers (care has to be taken to assure zero blanks), a near Instantaneous,
or time-averaged sample is collected. Sampling stops at sub-ambient or
ambient pressure. No pumps or additional hardware is required except a flow
control devioe for the time-averaged sampling. Collected sample ia either
analyzed direotly via an evacuated sample inlet manifold, or the grab sample
is pressurized with zero air. The dilution factor has to be taken into
consideration when low concentration species are sampled. vThe size of the
grab Commonly used
grab samplers range in size from less than one liter to about fifty liters.
The most versatile and widely used size ia al* T-jtara.. Sufficient sample is
available for multiple analyses of low level V0C*s utilizing oryogenio
pre-concentration and GC or QC-MS analysis. The sample inlet, conneoting
tubing and other components (valves, gauges, regulators) must be
meticulously ohecked for contamination to permit trouble-free sampling and
valid data. Most frequently used cleaning procedures use elevated
temperatures (dictated by the component material), evacuation (preferably
with efficient high vacuum, cold-trapped pumps), or purging with zero air or
clean inert gas. Combination of the above techniques can also be used.
Positive pressure grab sample collection is carried out with suitable,
inert sampling pumps with stainless 3teel bellows, or combination of 5S pump
head and Viton diaphragm. The grab sampler can be equipped with a purge "F1
assembly "to"' allow dynamic flow in and out of the sampling container via an
inlet and outlet valve. Sampling system surfaoe equilibration is one of the
benefits of this sampling approach, other advantages are larger sample size,
positive pressure (up to HO psig) which facilitates sample transfer and
analysis, and easy cleaning/recycling with the purge gas. These benefits
are offset by somewhat higher oost, need for additional hardware and
electrical power. However, considering the durability' and flexibility of the
"STSb samplers (particularly with the modular construction where valve(s),
pressure/vacuum gauges and flow control devices can be easily added) the
purge "Tw assembly is a desirable addition to the "basic" grab sampler,
outweighing the initial oost. Again, the users must pay great attention to
the cleanliness of the sampling system components, and assure satisfactory
"blanking" prior to deployment in the field.
Cryogenic pumping/enrichment is used infrequently. Oreat oare and
emphasis needs to be given to the proper cryogenic sampling system
configuration including flow, pressure control and monitoring devioes, and
suitable material selection (most 53 alloys become brittle at low
temperatures). The primary advantage is the ability to greatly increase
sample volume, or to pre-oonoentrate compounds of interest by a pumpless,
cryogenioally induced flow. Additional advantages are stability of reaotive
speaies at low temperatures, and ability to use similar approach for sample
component separation and pre-ooncentration prior to analysis. Disadvantages
include complexity, use of consumables (i.e. liquid nitrogen, oxygen or
argon), and need for skilled personnel,
.....Successful use of grab sampling containers requires a suitable and
reliable recycling system typically consisting of a heat souroe (oven),
vacuum souroe (pump), and clean purge gas (zero air). Sampling system
components must be contamination and leak free, periodio oheoks are needed
to determine the container condition to assure that there is no significant
sample degradation, loss or other undesirable phenomena. A grab sampler use
log is highly desirable.
Currently, the sampling requirements are shifting to longer term,
integrated sampling for periods up to 2D hrs. Clearly, flow control in the
1-100 ml/min, range is required. Several approaches are available. Fixed
297
-------
flow orifices provide a simple, inexpensive flow control but do not allow
full utilization of the grab sampling containers. More sophisticated manual
or eleotrioally aotuated flow controllers are available with better
accuracy, reproducibility and ability to take full advantage of the grab
samplers when used in oonjuction with sampling pumps. Mass flow controllers
are also available but their sensitivity to variations in ambient conditions
(i.e. temperature) and high cost limit their usefulness. Automated
operation of the entire sampling cycle is highly desirable. The timing and
hardware (valves, pumps, controllers) actuation functions are easily
controlled by a small, low cost microcomputer. The unit controls the
sampling start-up time, pump operation, system purge, sequential switching
of individual grab sampler inlet valves for up to twelve grab samplers and
also operation of the flow or pressure control device(s). in addition, delay
between samples, multiple sample collection (duplicates), and other features
can be easily incorporated in the oomputer software. Sample collection
parameters are stored and can be easily retrieved, transferred or printed
out. The main advantages are full automation and integrated, long-term
sequential collection of a number of grab samples. The disadvantages are
higher cost and relative system complexity.
Conclusions
The grab sampling has evolved into a well tested and widely used alternative
to other sampling methods such as polymeria bags and solid adsorbents.
Because of its simplicity, reliability and flexibility of use for a great
variety of sampled species ranging from high ppm level to low ppt
concentrations the grab sampling is rapidly beooming the preferred method
for VOC/NMOC and toxio gas sampling. The automation and significant sampling
capability extension (i.e. long-term integrated sample collection), and the
availability of sensitive and automated analytical methods (7) provide the
reearohers, control and regulatory personnel with comprehensive information
necessary for effective gaseous air pollutant control and abatement.
References
1. D. E. Harsoh, "Evaluation of a versatile gas sampling container design",
Atm. Env. 1H: 1105. (1980).
2. W. A. Lonneman, J. J. Buffalini, R. L. Kurjs, S. A. Meeks, "Contamination
from fluorooarbon films", Env. Soi. Teeh. 15: 99. (1981).
3. K* D. Oliver, j. d. Pleil, W. A. MoClenny, "Sample integrity of traoe
level volatile organio compounds in ambient air stored in Summa polished
canisters", Atm. Env. 20: U»03. (1986).
4. P. F. McElroy, V. L. Thompson, H. 0» Richter, "A Cryogenic
preconoentration-direot FID (PDFID) method for measurement of JWOC in
ambient air", EMSL/EPA report. (1985).
5. R< D. Cox, "Sample oolleotion and analytical techniques for volatile
organics in air", presented at APCA Specialty conference on measurement and
monitoring of non-oriteria (toxio) contaminants in air, Chicago, II. (1983).
6. E. D. Pellizzari, W. T. Gutkneoht, S. Cooper, D. Hardison, "Evaluation of
sampling methods for gaseous atmospheria samples". Final report, Research
Triangle Institute, EPA contract Mo. 68-02-2991. (1985).
7. J. D. Pleil, K. 0. Oliver, W. A. MoClenny, "Time-resolved measurement of
indoor exposure to volatile organio compounds", unpublished nanusoript, to
be presented at Indoor Air '87, Berlin, FRO, Aug. 1987.
298
-------
Table I, Passivated stainless steel grab sampler sampling compatibility list
Type or Group
of Gases to be
Collected
Concentration Recommended
Range (parts: Sampling
Vol ./Vol.) Approach
Comments
-12
Inert, man-made gases;
SF^, Fluorocarbons, etc.
Fluorochlorocarbons,
Halogenated Hydrocarbons
Hydrocarbons (VOC/NMOC)
CH , CO , NO, and
moSt inert gases
Sulfur-containing
gases (is, COS,
mercaptans)
Toxic gases (per
EPA classification)
Reactive process gases
at elevated temperature
10-
-6
¦ 10
-11
10—
¦10
Ambient pressure
Ambient or posi-
tive pressure
Glass/plastio
also suitable
Elastomerlc mate-
rials unsuitable
-10
10—
-10
Positive pressure Elastomers and
-9
-3
10-
-10
or cryogenic
sampling
Ambient pressure
some adsorbents
not recommended
-3
10—
-l|
-10
positive pressure Some limitations
or cryogenic
sampling
because of signif.
sample loss at
low concentrat.
-10 -5
10 10
-8
10-
Ambient/positive Some limitations
pressure sampling w/ solid adsorb's.
-3
-10
Ambient pressure
Avoid condensation
Collection * amb.
temp./press, only
NOTE: Collection of reactive gases 3uch as HCL, HF, NO, NO , so ,
and others should be done under controlled condition^. ^
Figure 1. Illustration of grab sampler sizes, shapes and configurations
299
-------
AM VAN: Delaware's Comprehensive
Air Toxics Monitoring Van
Terri V. Henry, Joseph J. Kliment
Air Surveillance Branch
Division of Air and Waste Management
Delaware Department of Natural Resources
and Environmental Control
New Castle, Delaware 19720
The State of Delaware utilizes a mobile air monitoring van (AMVAN)
for sampling and analyses of non-criteria pollutants. The equipment
involved in the analyses includes a gas chromatograph, meteorological
instrumentation, data logger and computer terminal, radiological equip-
ment and safety equipment. The van is currently used for surveys,
emergencies, and radiological response. AMVAN has been effective in
monitoring for compounds of Interest Including benzene and vinyl
chloride. Future expansion of its monitoring capabilities is presently
under consideration.
300
-------
AMVAN; Delaware's Comprehensive
Air Monitoring Van
Introduction;
AM VAN is Delaware's new mobile air monitoring van designed for rapid
sampling and analyses of air pollutants in the field. The van is
primarily utilized for short and long term surveys of both hazardous air
pollutants and the traditional criteria air pollutants, thus providing
the Division of Air and Waste Management with field data for development
of air toxics monitoring methodologies and environmental risk assessment*
Of particular interest are those air toxic pollutants which have been
designated by the. Environmental Protection Agency as detrimental to the
health of humans.
The van was contracted during the winter of 1985 after air releases at
a petro-chemical complex focused attention on the State's air toxics
monitoring capability. It is the outgrowth of the fixed air monitoring
program vrtiich has been operational in Delafware since the late sixties.
Unfortunately, the major disadvantage of the fixed monitoring system is
the Inability to cover all air pollutant incidents because of the depen-
dence on wind direction. During the early seventies a small van
containing a sulfur dioxide monitor was utilized with considerable
success, while plume chasing to determine ground level concentrations of
sulfur dioxide. The van was subsequently upgraded to a trailer with an
electrical generator and the ability to sample for several air
contaminants simultaneously and continuously. The experience obtained
from this initial work was incorporated into the design of AMVAN
utilizing the latest state of the art instrumentation and data
processing.
Along with the survey sampling, the van is also utilized for any air
related environmental emergencies, radiological emergencies and upgrading
of the emissions inventory for toxic air pollutants. AMVAN is utilized
throughout Delaware with the initial effort being concentrated in the
industrial areas.
Van Design
The van was designed by Environmental Measurement, Inc. around a Ford
Econoline chassis and a specialized Grumnan body. It contains instrument
racks, a laboratory bench, and an operator's desk, along with a
pressurized water system, microwave, refrigerator, and santitation system
to corrpliment on-board analyses. To provide temperature stability for
the instrumentation, two high capacity air conditioner/heater systems are
utilized. A unique internal air filtration system protects the operator
from exposure to hazardous air pollutants and possible contamination when
sampling in the field.2 The system utilizes a combined particulate and
carbon bed adsorption system to cleanse the air.
AMVAN is designed to operate independent of the normal electrical
supply while in the field by utilizing a high capacity electrical ONAN
generator. There is, however, the flexibility to connect to the regular
electrical supply through an umbilical cord for extended sampling
301
-------
periods, providing a special adapter is available. To prevent contamina-
tion of sampled air by the generator exhaust, the air inlet probe is
pointed into the wind and a specially designed catalytic exhaust purifier
reduces the hydrocarbon and carbon monoxide to carbon dioxide and water.
Sampling Equipment
The van's initial instrument complement consists of a Photovac 10570
photoionization (PID) gas chromatograph for detection and analysis of
atmospheric organic pollutants and meteorological sensors for determining
wind speed and direction. The wind sensors are mounted on a telescoping
tower which is raised pneumatically to a height of 30 feet. There is the
flexibility to raise the mast by a hand pump if the pneumatic system
fails. Monitors for measuring criteria air pollutants such as sulfur
dioxide, nitrogen oxides, ozone, carbon monoxide, and suspended
particulates can be installed in a standard instrument rack.
Radiological monitoring instrumentation such as a Geiger-Mueller
counter and Canberra Series 10 Ganina Spectrometer are utilized for radio-
logical emergency sampling.
Consideration is also being given to the installation of a portable
flame ionization gas chromatograph or a mass spectrometer or other state
of the art instrumentation to compliment the PID gas chromatograph for
rapid field analyses of additional organic compounds in the ambient air.
To facilitate field sampling, outside air is drawn through a special
intake probe lined with Teflon and into a manifold system to provide
sufficient outside air to the appropriate monitoring instruments. Gases
for instrument operation and calibration are contained in cylinders in a
special storage compartment inside the van.
Safety Equipment
To provide critical protection to the field monitoring team several
pieces of safety equipment are supplied in the van. Included are self-
contained breathing apparatus (SCBA's), anti-contamination equipment,
full-face masks with canisters for organics and radionuclides, portable
radiation detection kits, dosimeters, and a first-aid kit.
Data Processing and Corounications
Data from the sensors are processed through an Environmental Systems
Corporation on-board data logger and computer with direct read-out on a
Qwint computer terminal. The computer can provide information on
pollutant concentrations, wind speed and direction, along with necessary
quality assurance data. Large volumes of data can be stored directly on
an on-board magnetic tape for subsequent processing on a PDP 11/23 DEC air
quality computer at the laboratory. The van operator has complete
flexibility, through the data terminal, to make program or command changes
or requests for special data output.
For purposes of emergency data reporting, a unique system for transmit-
ting data utilizing a. "cellular" car phone modem is being evaluated. Air
pollution data would be sent via the cellular phone system directly to a
302
-------
terminal thus providing emergency planners with finger-tip information.
Data could also be sent directly to the air quality computer at the
laboratory for subsequent storage and processing capability.
A cellular car phone presently provides direct communication capabili-
ties with Department and State emergency management personnel through use
of the standard telephone system. AMVAN also contains a state communica-
tions radio with the capability of communicating with the Department of
Natural Resources and Environmental Control and Division of Emergency
Planning and operations should an emergency occur.
Field Applications and Experiences
The van is currently surveying a manufacturer who produces chlorinated
benzenes. This involves locating the van upwind, downwind and in densely
populated areas near the plant to monitor for benzene and
monochlorobenzene. In the future, a comparison will be made between
various plant processes with the results of the survey on the fugitive
emissions. This will be incorporated into the State of Delaware's
emission inventory program.
A second on-going seasonal study involves a company which produces
asphalt concrete used as a paving material for surfacing roads. The
plant produces both conventional and recycled asphalt. Odor complaints
have been made by the local citizens during the manufacturing of
recycled asphalt, and as a result a survey has been initiated utilizing
AMVAN to compare the total ionizables calibrated as hexane between the
conventional and recycle processes to determine any major emission
differences between the two products.
AMVAN has also been incorporated into Delaware's radiological
emergency plan.3 It is utilized as an analysis and communications base
in support of radiological monitoring activities. For training purposes
it is involved in a Nuclear Regulatory Commission/Federal Emergency
Management Agency annual drill along with Delaware's Department of
Emergency Planning and Operations (DEPO). These exercises prepare the
Department for Emergency planning in case of an accidential release from
a nuclear power plant located in New Jersey or for a radiological
transportation accident.
The van has been called out on an emergency response incident
Involving a leaking underground gasoline storage tank. Gasoline fumes
had been detected in a private citizens home which were linked to a local
gasoline station. Gasoline from the station's underground storage tank
seeped into the storm sewer and telephone conduit. Ventilation systems
were placed on the storm sewers to relieve vapor pressure which was
building up inside. Grab samples were taken above the ventiliation
systems in response to the local citizens concerns. These samples were
injected into the gas chromatograph located in the AMVAN while on Bite.
Benzene, toluene and xylene peaks were detected but were not significant.
These are only a few examples of the capabilities for a mobile air
monitoring vehicle.
303
-------
Summary
AMVAN provides needed surveys to establish environmental impacts and
upgrade emission inventories. The van also is used during emergencies
and for radiological response. AMVAN has been and will continue to be a
valuable asset to the on-going development of the air toxics program for
the state of Delaware.
Acknowledgement
The authors wish to express their appreciation to Robert R. French,
Manager, Air Resources Section, for the helpful suggestions and comments.
The authors would also like to thank Jeffrey C. Rogers, Chemist, for his
diligence in helping make this project a success.
References
1. Technical Assistance for Sampling and Analysis of Toxics Organic
Compounds in Ambient Air, U.S. Environmental Protection Agency, EPA-
600/X-B3-025-LU3. EPA.
2. Kliment, J.J., "AMVAN is Ready to Roll", Delaware Conservationist
3:24(1986).
3. State of Delaware, "Radiological Emergency Plan", Delaware DEPO,
Dover, Delaware, 1963.
304
-------
ESTIMATION OF VAPOR PRESSURES FOR NON-POLAR ORGANIC
COMPOUNDS FROM GAS CHROMATOGRAPHIC RETENTION DATA
12 2 3
T.F. Bidleman ' , D.A. Hinckley , and w.T. Foreman
1. Department of Chemistry and 2. Marine Science Program,
University of South Carolina, Columbia, SC 29208.
3. Cooperative Institute for Research in Environmental
Sciences, University of Colorado, Boulder, 00 80309.
The vapor pressure of an organic chemical exerts a large influence on
its dispersal in the environment, directly affecting volatilization from
soils and adsorption to suspended particles in air. She Henry's Lav con-
stant, which can be calculated from the ratio of vapor pressure to water
solubility, is a controlling factor in the air-water exchange of vapors.
Vapor pressure also governs the ability of a solid adsorbent to collect
organic vapors during air sampling.
Conventional methods of determining vapor pressures, effusion and gaB
saturation, are time-consuming and require fairly large quantities of
material. Capillary gas chromatography (GC) offers an alternative method
of estimating vapor pressures for non-polar compounds. Several compounds
can be run per day, and only analytical standard quantities of chemical are
required. The latter feature is an advantage when dealing with highly
toxic or costly materials.
in a previous article (1) we presented the theoretical and experimen-
tal details for determining ambient temperature vapor pressures by GC.
Briefly, test compounds and a reference compound having a known vapor
pressure are chromatographed at a series of temperatures on a short fused
silica column coated with a non-polar stationary phase. Vapor pressures of
test compounds are calculated from the relative retention times of test to
reference compounds and the reference compound vapor pressure.
305
-------
When the GC method was applied to 24 test compounds having reported
vapor pressures, we found that the initial estimate of vapor pressure, p:L,
was well correlated with but not exactly equal to the compound's liquia-
phase vapor pressure, p.. One reason for the discrepancy lies in the fact
that many high molecular weight organic compounds are solids at ambient
temperatures, and their (hypothetical) p. must be calculated from the vapor
pressure of the crystalline solid, ps, using;
In p°/p° » (ASf/R)(Tm-T)/T Equation 1.
in Equation 1, T and T are the melting and ambient temperatures (kelvin),
R is the gas constant, and &Sf is the entropy of fusion. Average values of
0S- are 13.5 and 13.1 cal/deg-mol for rigid aromatic hydrocarbons (2) and
polychlorinated biphenyl (PCB) congeners (3), but substantial differences
among compounds occur. For example, the range of OS* for several PCB
congeners is 10-17 cal/deg-mol (3). In this study we have improved and
expanded the GC vapor pressure technique published earlier (1):
1. Instead of assuming an average OS. ¦ 13.5 cal/deg-mol as done
previously (1), we used experimental AS* for as many test compounds as
possible. Experimental p^, were then correlated to p., calculated from
the test compound's pg using Equation 1.
2. A new GC reference confound, p,p'-DDT, was selected to supplement
the n-alkane references eicosane and octadecane previously used (1).
The p- and ASf of p,p'-DDT have been well established by several
investigators. From average literature values and Equation 1, log
p, (torr) - -4640/T + 10.20. (toreover, p,p'- DDT is electron capture
deftector (ECD) responsive whereas a flame ionization detector (FID) is
required with the n-alkanes. Use o£ an ECD is advantageous in that
very small quantities of halogenated test compounds can be chromato-
graphed, reducing the possibility of overloading the short capillary
column.
3. Vapor pressures (p?) of several organochlorine insecticides have
been determined, including endosulfan and its metabolites, major con-
stituents of technical chlordane, and two highly toxic components of
technical toxaphene.
4. The limitations of the GC method when applied to somewhat polar
molecules such as organophosphate and pyrethroid insecticides have
been examined.
Experimental
Test and reference compounds were chromatographed on a polydimethyl-
siloxane bonded phase fused silica column (BP-1, SGE Inc.) 1.0-m long x
0.22 mm i.d. mounted in a Carlo Erba 4160 or Varian 3700 Instrument. A
series of 4-7 isothermal runs in the 70-180 range was made. An FID was
used with hydrocarbons (PAH and n-C3n} and organophosphatesj halogenated
hydrocarbons (pesticides, PCB) were detected with an ECD. Conditions:
carrier gas H, or He at 1-3 mL/tain, injector 150-240 , detector 300-320 .
Samples in n-nexane were injected at a 10-20:1 split ratio. Data treatment
was as described in (1).
306
-------
Results and Discussion
In our original study (1) compounds were chromatographed on a bonded
phase (BP-1) column and a WCOT column containing n-C87 (Apolane-87). The
Apolane-07 column has since been discontinued by its only manufacturer
(Quadrex, inc.), so only the BP-t column was used in the present investiga-
tion. We also switched from He to H, carrier about midway through the
project, and observed improved column efficiency.
Seventeen hydrocarbons and halogenated hydrocarbons having known vapor
pressures (Table 1) were chromatographed along with the p,p'-DDT reference,
and p.. were calculated from relative retention data (1). These p° were
compared to pL of the test compounds, which were calculated from litera-
ture p° using Equation 1. In converting p- to p., we used experimental ASf
values in Equation 1, taken from references (3-7). A plot of log p? vs.
log pjL, showed a close correlation between the two vapor pressures, with a
small^ystematic difference, the regression equation being:
Log p° « 0.991 Log p^, + 0.094, r2 - 0.953 Equation 2.
Equation 2 was used to calculate p° from measured for compounds of
unknown vapor pressure.
vapor pressures (p?) of several organochlorine pesticides, determined
by the GC method, are given in Table 2. Literature values were available
for only two of these, and in both cases the agreement is reasonably good.
The determination of p, for Toxicants A and B illustrates an important
application of the GC method. These toxic components of technical toxa-
phene are not available commercially, and could only be obtained in micro-
gram quantities from laboratories involved in toxaphene research. Their
vapor pressures probably could not have been determined in any other way.
In our previous article (1), p? of several fcb congeners were deter-
mined using an eicosane or octadecane reference and a calibration curve
based on 24 test compounds. ltie calibration curve was prepared using an
average AS. -13,5 cal/deg-mol to convert literature p° to p?.' These data
have been recalculated using experimental AS. for as many test compounds as
possible, which resulted in slight changes in p° for the FCB congeners. A
selection of the recalculated congener data is given in Table 3, along with
recent values determined by Kurphy et al. using a static equilibration
method (8). Die agreement is generally very good, although the p° derived
from GC measurements are slightly higher. The pr of Kurphy et al. (8) were
obtained from commercial Aroclor fluids, and adherence to Raoult's Law was
assumed.
To test the accuracy of the GC method with moderately polar compounds,
p° of ten organophosphate pesticides and flame retardants were deter-
mined vs. n-alkane references. The p„ were generally higher than p?
of the organophosphates, by an averagentactor of 3.8. Since pi. overesti-
mated p°, caution is suggested in applying the GC method to polar com-
pounds. One could continue to use the n-alkane standards, derive a regres-
sion equation for organoghosphates similar to Equation 2, and then calcu-
late p? from measured p—,. Alternatively, an organophosphate of known
vapor pressure could Be used as a reference compound to determine p;L,
which should then correspond more closely to pL, an approach preferred by*
307
-------
Kim et al. (9). Similar difficulties may be encountered with other polar
compounds. The of four synthetic pyrethroid insecticides were deter-
mined vs. p(p'-DDT and pL were calculated using Equation 2. The results
showed a high bias when compared to literature values, by an average factor
of 6.0 (Table 4).
In conclusion, the GC method can provide accuracy within a factor of
two for the pL of non-polar hydrocarbons ang organochlorines. If a BP-1
or equivalent column is used to determine p^, for other non-polar com-
pounds, their p? can be calculated using Equation 2. The technique
should prove userul for other classes of non-polar air pollutants such as
chlorinated dioxins and dibenzofurans.
References
1. T.F. Bidleman, "Estimation of vapor pressures for nonpolar organic
compounds by capillary gas chromatography", Anal. Chem. 56: 2490 (1984).
2. S.H. Yalkowski, "Estimation of entropies of fusion of organic com-
pounds", Ind. Eng. Chem. Fund. 18: 108 (1979).
3. M.M. Miller, S. Ghodbane, S.P. Wasik, Y.B. Tewari, D.E. Martire,
"Aqueous solubilities, octanol/Vrater partition coefficients, and entropies
of melting of chlorinated benzenes and biphenyls", J. Chem. Eng. Data 29:
184 (1984).
4. C. Plato, A.R. Glasgow, Jr., "Differential scanning calorimetry as a
general method of determining the purity and heat of fusion of high-purity
organic chemicals. Application to 95 compounds", Anal. Chem. 41: 330
(1969).
5. C. Plato, "Differential scanning calorimetry as a general method for
determining purity and heat of fusion of high-purity organic chemicals.
Application to 64 compounds", Anal. Chem. 44: 1531 (1972).
6. D. Mackay, W.Y. Shiu, "Aqueous solubility of polynuclear aromatic
hydrocarbons", J. Chem. Eng. Data 22: 399 (1977).
7. J. Tuschall, Northrup Environmental Services, personal communication.
8. T.J. Murphy, M.D. Mullin, J.A. Meyer, "Equilibration of polychlorinated
biphenyls and toxaphene with air and water", Environ. Sci. Technol. 21s
155 (1987).
9. Y.-H. Kim, J.E. Woodrow, J.N. Seiber, "Evaluation of a gas
chromatographic method for calculating vapor pressures of organophosphorus
pesticides", J. Chromatogr. 314: 37 (1984).
10. L.P. Burkhard, D.E. Armstrong, A.W. Andren, "Vapor pressures for
biphenyl, 4-chlorobiphenyl, 2,2',3,3',5,5',6,6'-octachlorobiphenyl, and
decachlorobiphenyl", J. Cnem. Eng. Data 29: 248 (1983).
308
-------
11. F. Barlow, "Laboratory measurements of evaporation and photodecomposi-
tion rates for some synthetic pyrethroids" Fourth Internat. Cong. Pest.
Chem. (IUPAC), Zurich, 24-28 July (1978).
12. E.W. Balson, Trans. Faraday Soc. 43; 54 (1947).
13. B.T. Grayson, E. Langner, D. Wells, "Comparison of two gas saturation
methods for the determination of the vapour pressure of cypermethrin",
Pest. Sci. 13; 552 (1982).
14. D. Wells, B.T. Grayson, E. Langner, "Vapour pressure of permethrin",
Pest. Sci. (in press).
Acknowledgments
Contributions of the following individuals are gratefully acknowled-
ged. Dr. Jack Tuschall, Northrup Environmental Services determined AS- of
several test compounds. Dr. Jay Gooch, Woods Hole Oceanographic Institute,
provided a sample of Toxicant A. Portions of this work were performed
while TFB was a visiting scientist at the Special Analytical Laboratory,
National Environmental Protection Board (PUS-SNV), Stockholm, Sweden. PUS-
SNV also supplied the Toxicant B. This work was supported by the South
Carolina Sea Grant Consortium.
Table 1, Test Compounds for Vapor Pressure Calibration (Figure 2).
-Log p£, torr (25°C)2 -Log p°, torr (25°C)2
HCBz 3.051 p,p'-DDE 4.625
PH 3.125 2,2',4,5,5'-PeCB 4.654
AN 3.153 EICH 4.706
ALD 3.246 o,p'-DOT 4.863
r-HCH 3.321 p,p'-DDD 5.132
PY 3.967 BaA 5.390
DIEL 4.119 2,2',3,3',5,5',6,6'-OCB 5.404
2,2',5,5'-TeCB 4.222 DCB 6.967
FLA 4.322
1. Abbreviations: HCBz - hexachlorobenzene, PH - phenanthrene, AN - anth-
racene, ALD - aldrin, HCH - hexachlorocyclohexane, PY - pyrene, DIEL -
dieldrin, TeCB ¦ tetrachlorobiphenyl, FLA - fluoranthene, PeCB • penta-
chlorobiphenyl, EICH - eicosane, BaA - benz(a)anthracene, OCB - octachloro-
biphenyl, DCB - decachlorobiphenyl.
2. Calculated from literature p2 (1,10) and experimental AS- (3-7) using
Equation 1.
309
-------
Table 2. Vapor Pressures of Organochlorine Pesticides.
Chlordane Components
Heptachlor
a-Chlordene
Y-Chlordene
trans-Oalordane
cis-Chlordane
trans-Nonachlor
cis-Nonachlor
p°, torr (25°)
2.6E-4
2.0E-4
1.4E-4
5.9E-5
4.3E-5
3.5E-5
1.6E-5
p°, torr (25°)
Endosulfans
Endosulfan ether 4.4E-4
Endosulfan lactone 8.1E-5
Endosulfan I 5.6E-5
Endosulfan II 3.5E-5
Endosulfan sulfate 1.5E-5
Miscellaneous
ot-Hexachlosocyclohexane
Toxicant B,
Toxicant A
1.3E-3
2.1E-5
1.2E-5
1. Literature for Endosulfan I (10) - 2.9E-5
2. Literature for ee-HCH (11) » 2.0E-3
3. Components of technical toxaphene
Table 3. Vapor Pressures of PCB Congeners
p°, torr (25°)
This Work Murphy et al. (8)*
2,2',5,6'-TCB 2.1E-4 1.4E-4
2,3',4,4'-TCB 5.3E-5 3.4E-5
2,2',4,4',5-PCB 2.7E-5 2.0E-5
2,2',3,4,5'-PCB 2.1E-5 1.6E-5
2,2',4,4',5,5'-HCB 6.9E-6 3.4E-6
2,2'(3,4,4',5'-HCB 5.3E-6 2.0E-6
2,2',3,4',5,5',6-HCB 1.4E-6 1.3E-6
2,2',3,4,4',5,5'-HCB 5.7E-7 4.2E-7
1. estimated from values at 20°C.
Table 4. Vapor Pressures of Synthetic Pyrethroids
p°, torr (25°C)
ihis Work Literature(11,13,14) Error Factor1
cis-Permethrin 1.9E-7 3.7E-8 5.1 +
trana-Pennethrin 1.6B-7 2.5E-8 6.4 +
Cyperaethrin 4.3E-8 5.2E-9 8.3 +
Fenvalerate 1.6E-8 3.7E-9 4.3 +
1. This work/literature.
310
-------
ESTIMATES OF VAPOR-PARTICLE PARTITIONING
FOR SOME INDIVIDUAL POLYCHLORINATED
BIPHENYLS AND n-ALKANES IN DENVER AIR.
William T. Foreman1*® and Terry F. Bidleman1*3
1. Department of Chemistry.
2. Marine Science Program, and Belle W. Baruch
Institute for Marine Biology and Coastal Research.
University of South Carolina, Columbia, SC 29208.
3. Present address: Cooperative Institute for
Research in Environmental Sciences, Campus Box
449, University of Colorado, Boulder, CO 80309.
ABSTRACT
Polychlorinated biphenyls (PCB), n-allcanes, and other semivolatile organic
compounds (SOC1 are present in air m gaseous and particulate forms. Estimates of
the vapor-to-particle (V/P) distribution for non-pour SOC can be obtained from
high volume air sampling using a glass fiber filter to collect particles and an adsor-
bent trap to collect vapors. Apparent V/P distributions of 10 PCB congeners and
6 n-alkanes (octadecane to trjcosane) from air samples collected in Denver, CO
were estimated using the partition coefficient A(TSP)/F, where A and F are the
adsorbent- and filter-retained SOC concentrations ana T8P is the total suspended
particle concentration. A(TSP)/F wen related to the average sampling tempera-
ture (T, Kelvin) through; Log A(TSP)/F = m/T + b. For both classes of com-
pounds, fitted log A(TSP)/F at 5 C correlated well with the SOC liquid (or sub-
cooled liquid) phase vapor pressure.
INTRODUCTION
Atmospheric residence times of trace organic pollutants depend on the chemical
Mid physical properties of the constituents which control their removal. Removal
processes include destruction mechanisms (i.e., reactions with oxidizing species or
photolysis), precipitation scavenging, dry deposition of particles, ami dry gas
exchange with various surfaces. For semivolatile organic compounds (SOC), a criti-
cal consideration in modeling removal mechanisms is the distribution of these con-
taminants between the vapor and particle phases.1*^ Knowledge of this distribution
311
-------
is also useful when developing sampling and pollution control equipment.
One method of estimating the vapor-to-particle (V/P) distribution, and factors
influencing it can be obtained from high volume air sampling using a filter to col-
lect particles and an adsorbent trap to collect vapors. In recent years, several
research groups have used this approach to estimate the V/P ratio for selected
SOC, including polycyclic aromatic hydrocarbons (PAH), -B n-alkanes, and
organochlorine pesticides and total polychlorinated biphenyls (PCB). The present
study was undertaken to estimate the apparent V/P distribution of selected indivi-
dual PCB congeners and n-alkanes in Denver, Colorado air, and to test the influ-
ence of SOC vapor pressure on the partitioning process.
EXPERIMENTAL
A total of nine high volume air samples were collected in downtown Denver in
October 1985 and January 1986, using a glass Tiber filter-polyurethane foam (PUF)
plug sampling system. Filters and PUF plugs were extracted in a soxhlet extractor
for 24 h, and the extracts cleaned-up and fractionated using an alumina-silicic acid
column chromatography procedure. n-Alkanes were quantified by GC-FID, and
individual PCB congeners by GC-ECD using the method of Capel et al.®
RESULTS AND DISCUSSION
Apparent V/P distributions of 6 n-alkanes (octadecane to tricosane) and 10
cleanly separated PCB congeners (tetra- to heptachorobiphenyls in the Aroclor
1254 fluid region) were estimated using the partition coefficient A(TSP)/F, where
A and F axe the adsorbent- and filter-retained SOC concentrations and TSP is the
total suspended particle concentration. A(TSP)/F of the individual SOC were
related to the average sampling temperature (T, Kelvin) using an expression
derived by Yamasaki et al. :
Log [A(TSP)/F] = m/T + b (Equation l)
Examples of Equation 1 plots for 5 PCB congeners are shown in Figure 1.
Similar plots were constructed for the other 5 PCB and the n-alkanes. Most of the
scatter in these plots is probably related to the use of an average sampling tem-
perature for the 12-24 h sampling period when temperatures fluctuated by as much
as 21°C. Temperature variations can produce changes in the partial pressure of
SOC in equilibrium with particulate matter, resulting in blow-off losses or adsorp-
tion gains to the particles on the filter. Changes in SOC concentration would pro-
duce a similar effect. Samples were also collected during the fall and winter, and
differences in particle size distribution, surface area, and content of carbonaceous
material may have influenced the partitioning. The influence of relative humidity
on SOC adsorption to particles is also unknown, and humidity conditions varied
daily.
For both classes of compounds, log-log plots of the best fit A(TSP)/F at 5°C
versus the liquid (or subcooled liquid) phase vapor pressure (p£) of the SOC were
constructed. A/F partitioning of the n-alkanes (Figure 2) revealed a high degree of
correlation with p£ (r = 0.995). Similarity, Figure 3 shows that partitioning of the
10 PCB congeners also correlated well with p£ (r4 = 0.977), and these results are in
agreement with previous studies suggesting that A/F partitioning of non-polar SOC
312
-------
is largely governed by the compound's liquid phase vapor pressure.7'8
CONCLUSIONS
Despite the inability to control the critical variable of temperature and vapor
concentration, field studies of PCB and n-alkane partitioning from high volume
experiments have provided estimates of the apparent V/P distribution, as deter*
mined from the adsorbent- to filter-retained ratio. This study, along with other
field and laboratory8 investigations of V/P partitioning, have revealed that SOC
volatility, expressed by p£, is the dominating factor governing the adsorption of
non-polar SOC to urban air particulate matter.
ACKNOWLEDGMENTS
We thank the Air Pollution Control Division of the Colorado Department of
Health for providing sampling sites and in their assistance during sampling. This
research was supported by a grant from the Agricultural Research Service • U.S.
Department of Agriculture, Specific Cooperative Agreement No. 58-32U4-4-750.
REFERENCES
1. D. Mackay, S. Peterson, W.H. Schroeder, Environ. Sci. Technol. 20: 810
(1986).
2. J.F. Pankow, L.M. Isabella, W.E. Asher, Environ. Sci. Technol. 18: 310
(1984).
3. H. Yamasaki, K. Kuwata, H. Miyamoto, Environ. Sci. TechflQl, 16: 189
(1982).
4. C.D. Keller, T.F. Bidleman, Atmoe. Environ. 18: 837 (1984).
5. M.E. Ligocki, C. Leuenberger, J.F. Pankow, Atmoe. Environ. 19: 1609 (1985).
6. P.V. Doskey, A.W. Andren, Atmoe. Environ. 20:1735 (1986).
7. T.F. Bidleman, W.N. Billings, W.T. Foreman, Environ. Sci. Technol. 20:1038
(1986).
8. W.T. Foreman, T.F. Bidleman, Environ. Sci. Technol. (in press).
9. P.D. Capel, R.A. Rappaport, S.J, Eisenreich, B.B. Looney, Chemoephere 14:
439 (1985).
313
-------
Figure 1
Plots of log A(TSP)/F versus 1/T (Equation 1) for some
Aroclor 1254 region PCB congeners in Denver air. 95%
confidence bands for the predicted lines are included.
Congener chlorine substitution patterns are PCB #175 -
2,3,4,6,2',3-,5'j PCB *146 - 2 ,3,5,2a,4*,5't PCB #70 -
2,5,3',4't PCB #110 - 2,4,5,2,,5'f PCB #180 - 2,3,4,5,
2',4',5'.
PCB#175
10
10
A3
5„.
i i ' '— E—'—I ' C
3.3 4,4 Ji 3.4 3T 1.8
PCBI146
' ¦ * - ' —l—«—1—¦—I
lo'/T
a.a 3.4 3.8 ).t 3 7 3 1
t03/T
to"
PCB#70
—I—i—t I—i—I i—* - t 10
9,1 3-* 3.S a.* s.t 3.1
PCB#I10
PCB#180
J —I—i,li I ¦ ft - i
10®/T
te*/T
314
-------
Figure 2
Plot of log A(TSP)/F at 5°C versus log p? for the n-alkanes.
95% confidence limits for the predicted points, derived
from the Equation 1 plots, ere shown. 18 ¦ octadecane
through 23 ¦ tricosane are plotted.
4 r-
O
e
ui
5 2
£
w
1
w 1
<9
O
-8
-7
-a
-5
-4
-3
LOO P. , TORR
Figure 3
Plot of log A(TSP)/F at 5°C versus log p° for selected congeners
in the Aroclor 1254 region. 95% confidence limits for the
predicted points, derived from the Equation 1 plots, are shown.
1 - 2, 5,3',4 '; 2 - 2,4,3\4,i 3 - 2,4,5,2',5'j 4 - 2,3,6.3',4'j
5 - 2,3,4,2',3'f 6 - 2,4,5,3',4'» 7 • 2,3,5,2•,4',5•>
8 - 2,4,5,2',4',5'j 9 - 2, 3,4,6 ,2' , 3' ,5* > 10
4 p.
O
P
ui
U. a
a.
to
t 1
a
9 0
r
J t L
2,3,4.5,2',4',5'.
-I -7 -• -4 -a
UO« TORR
315
-------
MEASUREMENTS OF DIBENZODTOXINS AND DIBENZ OFURANS IN AMR IE NT AIR
Brian D. Eitzer and Ronald A. Hites
School of Public and Environmental Affairs
and Department of Chemistry
Indiana University
Bloomington, Indiana 47405
Polychlorinated dibenzo-£-dioxins (PCDD) and polychlorinated dibenzo-
furans (PCDF) are toxic compounds which are transported through the
atmosphere. It has been shown by a comparison of sources and sinks that
during the transport process there is a change, in the homolog profile of
these compounds. To examine this change, a method of measuring these
compounds in ambient air was developed. The method uses a high volune
air sampler equipped with a glass fiber filter and polyurethane foam
plug. This allows for the separate analysis of particulate and vapor
phases. The filter and plug are spiked with isotoplcally labeled
standards prior to extraction and clean-up. The purified extracts are
analyzed by electron capture, negative ion, chemical ionization gas
chromatographic mass spectrometry. The PCDD and PCDF are quantified as
the total concentration of each homolog class for tetrachloro through
octachloro homolog classes. Atmospheric concentrations and vapor to
particle ratios (V/P) for each class have been determined; there is an
inverse relationship between log V/P and temperature.
316
-------
Introduction
The presence of polychlorinated dibenzo-jj-dioxins (.PCDD) and poly-
chlorinated dibenzofurans (PCDF) in remote lake sediments1 indicates that
these compounds are transported through tte atmosphere to environmental
sinks such as lake sediments or human fat.^ The major sources of these
compounds to the atmosphere are waste incineration^"10 or automobile
exhaust.11 The homolog profiles (the relative amount of the compounds
grouped by chlorlnation level) in the sources and 1n the sinks are
different. In the sources, there is a broad range of homolog classes; in
the sinks, the pattern is dominated by octachlorodioxin with heptachloro-
dioxins and furans as minor components. In order to determine what
factors are important during the atmospheric transport process, we have
undertaken a study of PCDD and PCDF in ambient air. The concentrations
of PCDD and PCDF 1n vapor and particulate phases have been determined and
have been shown to be affected by temperature and level of chlorlnation.
Experimental Methods
Air samples were taken at four sites in Bloomington, Indiana. Air
was sampled with a high volume air sampler modified with a 10 x 10 cm
polyurethane foam (PUF) alug as a back-up to a 0.1 um glass fiber filter.
Air was sampled at 0.5 nr/min for a two to three day period. The total
suspended particulate matter was determined by equilibrating the filter
in a desiccator for 24 hrs and weighing it both before and after
sampling.
Prior to extraction, the glass fiber filter and the.-PUF plug were
spiked with per~13C-octachlorodibenzodiox1n and per-13C-l,2,3,7,8-
pentachlorodibenzofuran as Internal standards. It was assimed that any
losses that occurred during extraction and clean-up were equal for the
internal standard and the analytes. This pair of labeled standards was
calibrated with unlabeled standards to develop response factors for each
homolog class.
Air samples were Soxhlet extracted with 300 ml benzene for the glass
fiber filters and 2 L petroleum ether for the PUF plug for a period of 24
hr. After extraction, the solvent was reduced and exchanged to hexane 1n
a rotary evaporator and taken through a two-step chromatographic clean-up
to remove interferences.
The extracts were passed through a 1.5 x 20 cm column packed with
silica gel, activated at 160° C for 16 hr then deactivated with 1* by
weight of water and equilibrated for 16 hr. The sample was eluted with
75 mL each of hexane, 15% methylene chloride in hexane, and methylene
chloride. The dloxins and furans were collected 1n the second fraction.
The solvent of this fraction was exchanged to hexane and reduced to less
than 1 mL. This sample was then passed over a 0.5 x 6.5 cm micropipet
packed with alumina, activated at 250° C for 2 hr, then deactivated with
IS by weight of water and equilibrated for 16 hr. The sample was eluted
with 8 mL each of hexane, 2% methylene chloride 1n hexane, and 40%
methylene chloride in hexane. The dioxlns and furans were present in the
third fraction which was collected in an 8 mL sample v 1 al and concen-
trated to 100 uL by slowly passing purified N2 over the sample. The
sample was then ready for gas chromatographic mass spectrometry (GC/MS).
The samples were analyzed on a Hewlett-Packard 5985B GC/MS system,
A 30 m * 0.25 mm DB-5 fused silica column with hellun as the carrier gas,
was temperature programmed (splitless injection at 60° C, isothermal for
317
-------
2 min., 30o C/min to 210o C, 2o C/min to 280o C, isothermal for 10 min)
to separate the dioxins and furans. The mass spectrometer was operated
in the electron capture, negative ionization mode with methane as the
reagent gas. The iort source pressure was 0.4 torr, and the ion source
temperature was 150° C. To enhance sensitivity, selected ion monitoring
was used. Two ions from the most intense isotopic cluster of each
homolog group were monitored for each of the dioxins and furans. If
these ions were not present in the proper ratio, as predicted by the
isotopic abundances, it was assumed that the peak was that of an inter-
ferent and was not quantified.
Samples were quantified by using response factors for each homolog
class as compared to the labeled standards (one isomer per homolog
class). Concentrations were then determined by ratioing the peak areas,
multiplying by the amount of the internal standard added, dividing by the
sample weight, and correcting for the response factor. Concentrations
are reported as fg/m of each homolog class. It should be noted that the
method might be significantly underestimating the concentration of
2,3,7,8-tetrachlorodioxin because this particular homolog does not
respond well in the negative ion mode.
Results
Since the PUF plugs act as frontal chromatography columns, it is
important that the air sample volume be below the chromatographic break-
through volume. Although the breakthrough volume for these analytes and
plugs was not rigorously measured, two experiments were performed to
assure ourselves that the breakthrough volume was not exceeded. In the
first experiment, a PUF plug was split in two halves (orthogonal to the
flow) which were analyzed separately. No dioxins or furans were found on
the back half of the plug. In the second experiment, two samples were
obtained simultaneously at different flow rates; thus, different volumes
of air were sampled. If the breakthrough volume was exceeded the higher
volume sample should have lower concentrations. This was not the case,
indicating that the sample volume was less than the breakthrough volune.
The average concentrations of PCDD and PCDF in a 40 measurement data
set are shown in Figure 1. It should be noted that the vapor phase shows
increased concentrations of the less chlorinated homologs as compared to
the particulate bound phase. This is expected because these homologs are
more volatile. Also note that the particulate bound homolog pcQfi 1 e Is
quite similar to that seen in other environmental sinks. This
suggests that this phase is important in the transport of these compounds
from their sources to their sinks.
For any given homolog, the ratio of the vapor phase to the particu-
late bound phase is dependent on temperature. This can best be seen in
Figure 2 which 1s a plot of the log of this ratio ^s. Inverse temperature
for the hexachlorodloxins and furans. As the temperature rises, the
vapor phase becomes more prevalent.
Conclusions
We have measured PCDD and PCDF in afr at femtogram per cubic meter
levels. The method we have developed can be used to study these com-
pounds as they are transported from source to sink. The data suggest
that the sorption of PCDD and PCDF to particulate matter plays an
important role In the transport of these compounds to environmental
318
-------
sinks. This sorption is temperature dependent as shown by the inverse
relationship between the vapor/particul ate ratio and temperature.
Acknowledgment
We thank Ilora Basu for technical assistance. This project was
supported by the 1/estinghouse Electric Corporation and the U.S.
Department of Energy (Grant No. 80-EV10449). Some of this information
has been published previously in reference 13.
References
1. J. M. Czuczwa, B. D. McVeety and R. A. Hites, Science 226. 568
(1984).
2. L. Hardell, L. Dome! 1 of, M. Nygren, M. Hansson and C. Rappe, Pre-
print extended abstract ACS Division of Environmental Chemistry,
Miami, FL, April, pp. 167-168 (1985).
3. R. R. Bumb, W. B. Crumett, S. ,S. Cutie, J. R. Gledhill, R. H.
Hummel, R. 0. Kagel, L. L, Lamparski, E. V. Luoma, D. L. Miller, J.
J. Nestrick, L. A. Shadoff, R. H. Stehl and J. S. Woods, Science
210, 385 (1980).
4. H. R. Buser and H. P. Bosshardt, Mitt, Gibiete Lebensm. Hyq. 69, 191
(1978).
5. A. Cavallaro, L. Luciani, G. Ceruni, I. Rucchi, G. Ivernizzi and A.
Gar in, Chemosphere 11, 859 (1982).
6. A. Liberti and D. Brocco, Chlorinated Dioxlns and Related Compounds:
Impact on the Environment (0. Hutzingerj R, W. Frei, E. Merian, and
F. PoccTiTari, eds.) (Pergamon Press, 1982), pp. 245-251.
7. A. Liberti, D. Brocco, A. Ceciata and A. Natalicci, Analytical Tech-
niques in Environmental Chemistry (J. Albaiges, ed.j (Pergamon
Press, f§$2), pp. 281-286.
8. J. VI. Lustenhouwer, K. 01 ie and 0. Hutzinger, Chemosphere 9, 501
(1980).
9. K. 01ie, J. W. A. Lustenhouwer and 0. Hutzinger, Chlorinated
0 ioxi ns and Related Compounds: Impact on the Environment (0.
Hutzinger, R. W. Fre1, E, Merian and F. PocchTarl, eds.) (Pergamon
Press, 1982), pp. 227-244.
10. C, Rappe, S. Marklund, P. A. Gergqvist and M, Hansson, Chlorinated
D1ox1ns and Dibenzofurans in the Total Environment (G. Choudhrv. I.
H. Keth ancT C. Rappe, eds.r~(Bullerworth, iy&S), pp. 99-124,
11. K, Ballschmlter, H. Buchert, R. Niemczyk, A. Munder and M. Swereu,
Chemosphere 15, 901 (1986).
12. J. M. Czuczwa and R. A. Hites, Environ. Sc1. Technol. 18, 444
(1984).
13. B. D. Eitzer and R. A. Hites, Intern. Environ. Anal. Chem. 27,
215 (1986).
319
-------
AVERAGE VAPOR
360 fg/
m
V7,
JZZZZL
^
AVERAGE PARTICULATE
760 fg/
Y//A1 V7/A czzza_^_£^l_^
tfjP ?6cSP tfCCP tfS*>
Figure 1. Average concentrations (in fg/m3) of PCDD and PCDF in air in
Bloomington, Indiana. Top, the vapor phase; bottom, the particulate
bound phase. Total concentrations of each homolog class of the tetra-,
penta-, hexa-, hepta- and octachlorodibenzofurans (TCDF, P5CDF, H6CDF,
H7CDF, and OCDF, respectively) and tetra-, penta-, hexa-, hepta-, and
octachl orod ibenzo-j>-diox ins (TCDO, P5CDD, H6CDD, H7C0D, and OCDD, respec-
tively) are shown. The concentration of the most abundant homolog class
in each group is shown.
320
-------
H6CDD
LOG CV/P)
2 r
1. 5
1
Q. 5
-0
-0. 5
-1
-I. 5
O
° o ° o
r=. 74
O
O
o
o
3. 4
» ° - g
OC£>
o^o
o
^b.
o
- 1
3.5
O
- iO
3.6
100D/T DECREES KELVIN
o
o
^8
3. 7
3. 0
H6CDF
LOG (V/P)
1.5
\
0.5
-0
-0. 5
¦1
-1.5
-Z
-2.5
o
o ° °
o°
r=. 90
O
£
§ ©
3. 4
3. 5
1000/T DEGREES
O
3.6
KELVIN
O
O
3. 7
3.8
Figure 2. Log (vapor/particulate) plotted vj». inverse temperature for
hexachlorodibenzo-jj-dioxlns and hexachlorodlbenzofurans. The correlation
coefficients (r) are shown.
321
-------
MEASUREMENT OF HENRY'S LAW CONSTANT
FOR SELECTED HALOCARBONS IN.
DILUTE AQUEOUS SOLUTIONS
Gary B. Howe
Tony N. Rogers
Research Triangle Institute
Research Triangle Park, North Carolina
Joseph E. Knoll
M. Rodney Midgett
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
Henry's law constant is an Important physicochemical property to con-
sider in predicting the environmental fate of toxic organic compounds. Many
environmental fate assessment models depend heavily upon Henry's law con-
stants as measures of air/water equilibrium partitioning behavior. In addi-
tion, Henry's law constant is an important parameter in the design of han-
dling and treatment processes for toxic organic contaminated water. In this
work, Henry'b law constants were measured for methylene chloride, chloro-
form, and 1,2-dichloroethane using two different techniques. The first
technique used was a static headspace measurement technique known as Equi-
librium Partitioning in Closed Systems (EPICS) in which equal masses of
organic compound were loaded into different liquid volume systems and al-
lowed to equilibrate overnight. The organic concentration in the headspace
of each bottle was then measured by gas chromatography with flame ionization
detection. Measurements were performed at five different temperatures rang-
ing from 10-30eC. In addition, measurements were performed using EPICS for
solutions of 0.5 M aodiun chloride in water containing low concentrations of
volatile organic compound. The second technique used was batch air strip-
ping in which an organic compound in water solution was purged with air
while monitoring the organic in air effluent concentration by GC-F1D.
Henry'b law constant was determined from a linear regression of the loga-
rithm of concentration versus time. Measurements using this technique were
performed at selected temperatures for comparison with the EPICS results.
322
-------
Introduction
In addition to the well established problem of groundwater contamina-
tion frojn toxic organic compounds, air emissions of volatile organic com-
pounds (VOC) from hazardous waste treatment, storage, and disposal facili-
ties has become a major environmental concern. In the case of surface im-
poundments, the VOCa are usually present in water at relatively low concen-
trations and thus Henry's law for ideal dilute solutions can be applied for
vapor-liquid equilibrium calculations.
A technique frequently used to estimate Henry's law constant of a VOC
in water is to calculate the ratio of the pure compound's vapor pressure to
its aqueous solubility limit, as expressed in Equation 1.
psat
Hc = V (1)
As
where; psat * vapor pressure of the pure VOC, atm
xs = ultimate solubility of the VOC in water, mole fraction
This assumes that an incremental amount of VOC added to a saturated aque-
ous solution will form a second pure organic phase that has an equilibrium
partial pressure above it equal to the pure component vapor pressure, To
use this estimation technique, it also must be assumed that the Henry's
Law relationship is valid up to the ultimate solubility limit. And obvi-
ously, this technique will only provide an accurate determination of
Henry's law constant when reliable vapor pressure and aqueous solubility
data are available.
The objectives of this work were: (1) to explore some simple expert-
Rental techniques for directly measuring Henry's law constant for dilute
solutions of VOC in water as an alternative to the potentially unreliable
estimation technique, (2) to test the effects of solution ionic strength
on the Henry's law constant of three VOCs, and (3) to compare experimen-
tally determined values of Henry's law constant for chloroform, 1,2-dl-
chloroethane, and methylene chloride with those calculated from the esti-
mation technique described earlier.
Experimental Methods
The first technique used for directly measuring Henry's law constant
Is known as Equilibrium Partitioning in Closed Systems, or EPICS, which
was originally proposed by Gossett and Lincoff in 1984. EPICS is based on
a closed system mass balance for a given VOC distributed between liquid
and gas phases. The component mass balance takes the form shown in Equa-
tion 2.
M - CLVL ~ CGVG
-------
expression can be written for each system. Equating these expressions and
introducing Henry's law to substitute for the liquid phase VOC concentra-
tions gives an equation relating Henry's law constant to headspace concen-
trations and known volumes. {Equation 3)
tCDl/CG2> VL -VL
Hc = (3)
Vg2 - (cQl/cG2)vGl
where: Hc = djmensionless Henry's law constant
Absolute concentrations need not be measured because any proportional
measure of concentration, such as GC peak areas, will yield the required
headspace ratio for Equation 3.
Prior to performing EPTCS measurements, a saturated stock solution
for each compound was prepared by adding an amount of organic solute in
excess of the solubility limit to pure water in one liter amber glass
bottles. Stock solutions were allowed to equilibrate for several days
before using.
In preparing samples, two pairs of 250 ml amber glass bottles were
filled with 20 ml and 200 ml of pure water, respectively. The same volume
of saturated stock solution was then added to each bottle and the bottles
were sealed with a Teflon lined silicon rubber septum cap. The stock
solution volume added was dependent upon compound solubility and was cho-
sen to produce an Initial liquid phase concentration of 10 milligrams per
liter. This was to ensure that the measurements were made in the concen-
tration range where Henry's law is obeyed. The loaded bottles were shaken
vigorously by hand and placed in a constant temperature water bath for a
minimum of 16 hours before analysis.
After equilibration, headspace samples were withdrawn from the bot-
tles with a 1.0 cm3 gas-tight syringe and injected into a Varian 3700 gas
chromatograph equipped with a flame ionization detector, The GC column
consisted of a 14 in. long X 1/8 in. OD stainless steel tube packed with
60/80 mesh Porasll B. Peak areas were measured by electronic integrator
and the results were transmitted to a personal computer data file. A mean
Henry's law constant and relative standard deviation were determined from
the four measurements by intercomparing the two sets of high volume/low
volume results. Measurements were performed for each compound at five
different temperatures beginning at 10"C and ranging to 30°C in 5#C incre-
ments .
Additional measurements were performed using EPTCS with substitution
of 0.5 M sodium chloride for pure water as diluent. Measurements were
performed for all three compounds at temperatures ranging from 10-30*C.
A second technique investigated for measuring Henry's law constant is
known as batch air stripping. The system we used ifl shown schematically
in Figure 1 and consisted of a water-jacketed glaBS column with a glass
fritted disc press-fit into the column inlet. Air was supplied to the
column from a tank of high purity compressed air. Purge air was passed
through a water-filled impinger prior to entering the column to prevent
water evaporation from the column. A constant flow of temperature con-
324
-------
trolled water was maintained through the column Jacket to ensure isother-
mal operation.
Before beginning a run, the column was loaded with a known volume of
pure water. After temperature equilibration, a quantity of saturated
stock solution was pipetted Into the column and gas flow was Initiated at
a preset rate. The column exit gas was pumped through a 0.5 cm3 stainless
steel sample loop attached to a six-port rotary valve. The valve was
actuated periodically to inject samples into a GC with a flame Ionization
detector. Sample peak elutlon time was typically 30-45 sec which allowed
real-time analysis of column effluent gas instead of collecting liquid
allquots for later headspace analysis.
Henry's law constant was calculated by correlating concentration data
against a solute mass balance equation for the column. The differential
form of the mass balance Is shown in Equation 4.
Y0G - Y(t)G = (C(t)V) (4)
where: Y0 = organic concentration ,at gas
inlet, mol/m3
Y(t) = time-variable outlet gas
concentration, mol/m3
G = volumetric gas flow rate,
m3/min
C(t) = time-variable liquid concentration,
mol/m®
V = total liquid volume, m3
By substituting Henry's law relationship into Equation 4 and integrating,
the expression shown In Equation 5 results.
Hc G
Ln C(t) - Ln CQ + — t (5)
where: t - elapsed time, minutes
According to this equation, a logarithmic plot of liquid or gas-phase
concentration as a function of elapsed time will yield a straight line
with a slope directly related to Henry's law constant. It is not neces-
sary to know the initial batch concentration, C0, or absolute concentra-
tions ln the gas phase since only the y-lntercept will be influenced by
these factors. Therefore, any proportional measure of gas phase concen-
tration such as GC peak area can be inserted for C(t). There are two
primary assumptions in using the batch air stripping technique for mea-
suring Henry's law constant: (1) the gas exiting the column must reach
full concentration equilibrium with the organic laden liquid and (2) there
must be complete mixing of the liquid within the column. These two re-
quirements can create problems in column design since the first require-
ment of equilibrium is best achieved In tall columns whereas complete
mixing is most easily achieved ln shorter columns.
325
-------
Results
Summarized in Table I are the Henry's law constants measured for
methylene chloride, 1,2-dichloroethane, and chloroform using the EPICS
technique. Good precision was achieved with the relative standard devia-
tion less than 8 percent at all temperatures. As expected, the measured
constants increased with the increasing temperature as solubility de-
creases and vapor pressure increases. The constants are reported in units
resulting from expressing the gas phase concentration as a partial pres-
sure in atmospheres and the liquid phase concentration in moles per cubic
meter.
Shown in Table II are selected results from EPICS measurements per-
formed with substitution of 0.5 M sodium chloride for pure water. The
control results are those which were determined earlier for pure water
diluent tests. The constants measured for 0.5 M NaCl systems were consis-
tently higher than the results for pure water systems, indicative of the
"salting out" effect that ionic species have on VOCs. The difference
between the constants for pure water and salt water was statistically
significant in all cases except methylene chloride at 20"C and chloroform
at 30"C.
The 0.5 M salt concentration used is approximately that of seawater
and thus can be considered a worst case situation for natural waters when
assessing the influence of salt concentration on vapor-liquid equilibrium.
Table III shows results from measurements performed by batch air
stripping for methylene chloride and chloroform at two different liquid
depths compared with values for the same compounds at 25*C determined by
EPICS and estimated from literature values of vapor pressure and solubil-
ity. The slight increase in measured constants by batch air stripping in
increasing the liquid depth from 44 cm to 56 cm provides some evidence
that full equilibrium was not achieved for either compound with this col-
umn design. It is not known whether these differences are statistically
significant given the limited number of tests performed.
Literature estimates of Henry's law constant for both compounds are
in relatively good agreement with measured values, suggesting that Henry's
law is obeyed up to the solubility limit and literature values for both
aqueous solubility and pure compound vapor pressure are accurate.
Conclusions
Both EPICS and batch air stripping are suitable techniques for direct
measurement of Henry's law constant for hydrophobic organic compounds.
Some advantages of EPICS are: (l) the mass transfer limitation of the
batch air stripping technique is avoided since the equilibration of VOC
between vapor and liquid phases can take place over several hours, and (2)
the method is relatively slmple~requlring only septum bottles, routine
laboratory glassware, a constant temperature bath, and a gas chromatograph
with flame ionization detector. Some limitations of EPICS are: (1) it is
fairly time consuming since most compounds require at least overnight
equilibration before measurements can be performed, (2) there is the po-
tential for adsorption of VOC to the bottle or septum surfaces resulting
in loss of mass and biasing of headspace concentrations, and (3) a limited
range of sensitivity since at Henry's law constants greater than about 2.0
(dimenslonless units), the headspace concentration ratio becomes nearly
326
-------
constant. However, most VOC's of environmental interest have dimension-
less constants of less than 1.0.
A definite increase was observed in Henry's law constants of the
three compounds in systems containing 0.5 M NaCl, indicating a salting-out
effect, and suggesting that ionic strength should be considered in deter-
mining vapor-liquid equilibrium behavior of VOCs.
Table I. EPICS Teat Results
H„. atm-m3/mol (RSP)
10
1.40
X
10"3(7.0)
15
1.69
X
10-3{1.5)
20
2.44
X
10"3(4.8)
25
2.96
X
10_3(2.6)
30
3.61
X
10-3(1.7)
Chloroform
7.05
X
10-4(4.4)
1.72
X
10-3(4.1)
8.79
X
10-4(4.1)
2.33
X
10-3(2.5)
1.21
X
10-3(7.8)
3.32
X
10-3(2.4)
1.46
X
10-3(1.6)
4.21
X
10-3(5.2)
1.75
X
10-3(6.9)
5.54
X
10-3(7.2)
Table II. Ionic Strength Test Results
Hp (dimensionless1
Compound
Control
0.5M NaCl
Temp..VC
Methylene chloride
0.0600
0.0792
10
0.1015
0.1080
20
0.1452
0.1566
30
1,2-Dichloroethane
0.0303
0.0432
10
0.0504
0.0605
20
0.0705
0.0832
30
Chloroform
0.0742
0.0889
10
0.1380
0.1470
20
0.2228
0.2348
30
Table III. Batch Air Stripping Results
Batch air stripping
Batch air stripping
Henry's
(44 cm depth)
(56 cm depth)
EPICS
Lit. estimate
2.64
X
10~3
2.71
X
10-3
2.96
X
10"a
2.52
X
10"3
law Constant
Chloride
, atm-m3/mol
Chloroform
4.43 x 10~3
4.62 x 10~3
4.21 x 10"3
3.82 x 10~3
327
-------
To Exhaust
1
To Temperature
Control Bath
Water
Jacket
6-Port Sampling Valve
Exhaust
To Gas Chromatograph (FID)
Aqueous Sample
Air Supply
Presaturator
Figure 1. Batch air stripping coluan.
328
-------
EXPERIMENTAL AND CALCULATED REACTIVITY
PARAMETERS OF POTENTIAL ENVIRONMENTAL
SIGNIFICANCE FOR NITRO-POLYCYCLIC
AROMATIC HYDROCARBONS
Arthur Greenberg
Chemistry Division
New Jersey Institute of Technology
Newark, New Jersey 07102
William J. Simonsick, Jr.
Marshall Research and Development Laboratory
E.I. DuPont de Nemours & Co.
3 500 Grays Ferry Avenue
Philadelphia, Pennsylvania 19146
Joel F. Liebman
Department of Chemistry
University of Maryland Baltimore County
Catonsville, Maryland 21228
Abstract
Theoretical and experimental parameters related to the ambient
stabilities of nitrated polycyclic aromatic hydrocarbons are
presented. The calculational method is MNDO, a semiempirical
technique development by Dewar and co-workers. Properties of
interest include gas-phase enthalpies of formation, first and
second ionization potentials (and their difference) and
electron affinities as well as comparisons with PAH. It is
apparent that ionization potentials are affected by the number
of nitro groups rather than by position of substitution. There
also appears to be a "saturation" effect on larger or more
substituted molecules. The IP1 - IP2 difference for the
nitro-PAH barely changes relative to the difference for the
corresponding PAH possibly indicating no dramatic effects for
reaction with singlet oxygen and possibly ozone.
329
-------
EXPERIMENTAL AND CALCULATIONAL REACTIVITY PARAMETERS OF
POTENTIAL ENVIRONMENTAL SIGNIFICANCE FOR NITRO-PAH
Introduction
During recent years there has been increased awareness
of the potential health hazards associated with airborne nitro
derivatives of polycyclic aromatic hydrocarbons(nitro-PAH).(1)
Presently, interest has focused on more polar mutagenic
compounds, many of which are apparently nitrated.(2) A
specific group of compounds, hydroxynitropyrenes or nitropyr-
enols, have been identified in air samples and seem to be
associated with significant biological activity.(3-5) In
trying to understand how polar nitro compounds are formed, one
can consider the reactivity of the PAH themselves. Thus, the
formation of 2-nitropyrene is thought to proceed via initial
attack of OH on pyrene in the presence of oxides of nitrogen
as shown in Scheme 1.(6) 2-Nitrofluoranthene can be produced
similarly.(6) It is also possible that 2-nitrofluoranthene may
be produced during gas-phase reactions of fluoranthene with
dinitrogen pentoxide (pyrene does not yield 2-nitropyrene
under these conditions).(6)
OH
SCHEME 1
OH
/®2
,0H
no2
330
-------
The possibility that a compound such as 2-nitropyren-1-
ol may be formed under these conditions does not appear to
have been considered explicitly. In this case it is clear that
hypothetical formation of this compound would depend on PAH,
not nitro-PAH, reactivity. Photochemical reaction of
1-nitropyrene (Scheme 2) in solution yields the indirect
mutagen l-nitro-2-pyrenol.(7,8) However, this compound does
not appear to be present in ambient airborne particulates
although measurable amounts are observed in particulate matter
associated with diesel emissions.(9,10)
SCHEME 2
OIQ
DC
OIQ
no
Air, UV
»
CH3CN
It is not clear whether nitropyrenols observed in air and
associated with aerosol aging are formed from the nitropyrenes
or from pyrenols# nor is it clear that the other classes of
more polar derivatives of nitro-PAH presumed to be present are
formed from the nitro-PAH themselves. Nevertheless, it seems
useful to summarize what is known experimentally about
parameters related to nitro-PAH reactivity and add to the data
set with calculation. There has been little uniform reported
data on nitro-PAH decomposition. Table 1 reports selected
examples of published data.
Presentation of Results
In Table 2 we present experimental and calculational
data on nitro-PAH and related PAH. The calculational technique
is MNDO developed by Dewar and co-investigators.(11)
Calculations were performed at the Indiana University CDC
Cyber 170/855 and some of these (PAH) have been reported
elsewhere.(12) The experimental ionization potentials ( IP,
adiabatic ) and electron affinities (EA) are from a compendium
(13). It is clear from Table 2 that for the PAH, the
calculated IPs are higher than the experimental values. Part
of this is because the calculated values will correspond to
vertical IPs if one assumes the validity of Koopmans theorem.
Only four values for nitro-PAH (considering benzene a HPAHn)
are known experimentally and the values are also 0.6 - 0.9 eV
too low but the trends ate reasonable. Figure 1 illustrates
that there is something of a "saturation effect" on placing
nitro groups on PAH of increasing size. Thus, the difference
in first IP between the nitro-PAH and the corresponding PAH
appears to decrease as the PAH gets larger. In Table 3 are
listed experimental IPs and EAs for a variety of substituted
331
-------
nitrobenzenes. Although the other monosubstituted benzenes are
not presented, again there is a "saturation effect" in that
the difference in IP between benzene and nitrobenzene is
greater than that between toluene and nitrotoluene. It is also
clear that positional substitution of the nitro group has very
little effect on ionization potentials. The largest effects
are produced as one increases the degree of nitration. The
nitroaniline IPs are qualtatively different from the rest of
the set in that ionization is largely from the amino lone
pair. It is still surprising that direct conjugation of the
pi-donor amino with the pi-acceptor nitro has such a small
effect. Generally, these results agree with the virtual
equality of the IPs of 1-nitro and 2-nitronaphthalene. It is
also clear from Table 3 that there are considerably large
positive electron affinities for nitrated aromatics (they are
strong charge-transfer acceptors) while, in contrast, benzene
radical anion is unbound. In Table 4 we list the first and
second IPs for some PAH and their nitro derivatives. It has
been noted that PAH most reactive to Diels-Alder reactions, a
model for reaction of some PAH with singlet oxygen or ozone,
are the ones having the greatest difference between IP1 and
IP2 even if the IPs are relatively high.(14) It is clear from
Table 4 that nitro groups do not dramatically effect this
difference (although one should keep in mind that these
substituents are not coplanar with the ring). It is possible
that 7-nitrobenz(a)anthracene may be less reactive toward
singlet oxygen than the parent hydrocarbon.
In Figure 2 we have plotted charges and other data for
the nitro-PAH investigated. There is evidence that, for
alternant PAH at least, greater reactivity is associated with
high relative positive charge.(12) If this follows for
nitro-PAH, then 2-nitroanthracene, for example, should be more
reactive than the other two isomers. It would also appear that
2-nitropyrene should be more reactive than the 1-nitro isomer.
The nitro-olefinic nature of 1-nitroacenaphthylene is clear
from this figure and some of this is present in 9-nitrophenan-
threne.
References
1. C,M. White (Ed), Nitrated Polycyclic Aromatic Hydrocarbons.
Huethig, Heidelberg, 1985.
2. J. Siak, T.L. Chan, T.L. Gibson, G.T. Wolff, "Contribution
to bacterial mutagenicity from nitro-PAH compounds in
ambient aerosols", Atmos. Environ., 19: 369 (1985).
3. M.C. Paputa-Peck, R.S. Marano, D. Schuetzle, T.L. Riley,
C.V. Hampton, T.J. Prater, L.M. Skewes, T.E. Jensen, P.H.
Ruehle, L.C. Bosch, W.P. Duncan, "Determination of nitrated
polynuclear hydrocarbons in particulate extracts by
capillary gas chromatography with N selective detection",
Anal. Chem., 55: 1946 (1983).
332
-------
4. M.G. Nishioka, C.C. Howard, J. Lewtas, "Identification and
quantification of 0H-N02-PAHs and N02-PAHS in an ambient
air particulate extract by NCI HRGC/MS. Presentation at
Tenth International Symposium on Polynuclear Aromatic
Hydrocarbons, Columbus, Ohio, October 21-23, 1985.
5. T.L. Gibson, P.E. Korsog, G.T. Wolff, "Evidence for the
transformation of polycyclic organic matter in the atmos-
phere", Atmos. Environ., 20: 1575 (1986).
6. J. Arey, B. Zielinska, R. Atkinson, A.M. Winer, T. Ramdahl,
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", Atmos. Environ.. 20: 2339
(19B6) .
7. A. Yasuhara, K. Fuwa, "Formation of l-nitro-2-hydroxypyrene
from 1-nitropyrene by photolysis", Chem. Lett* » 347 ( 1983 ).
8. G. Lofroth, L. Nilsson, E. Agurell, A. Yasuhara,
"Salmonella/microsome mutagenicity of l-nitropyren-2-ol, a
nitropyrene phenol formed in the photolysis of
1-nitropyrene", 2. Naturforsch.. 39c: 193 (1984).
9. A. Yasuhara, M. Morita, "Analysis of organic substances in
soot", Res. Rep. Natl. Inst. Environ. Stds., Japan, No. 79,
103 (1985).
10. A. Greenberg and Y. Wang, Unpublished results.
11. M.J.S. Dewar, W. Thiel, "Ground states of molecules. The
MNDO method. Approximations and parameters". J. Am. Chem.
Soc.¦ 99: 4899 (1977).
12. R.A. Hites, W.J. Simonsick, Jr., Calculated Molecular
Properties of Polycyclic Aromatic Hydrocarbons. Elsevier,
New York, 1987.
13. s.G. Lias, R. Levin, J.F. Liebman, U.S. National Bureau of
Standards IP and EA Database, J.F.L. Personal
communication.
14. d. Biermann, W. Schmidt, "Diels-Alder reactivity of
polycyclic aromatic hydrocarbons, acenes and benzologs,
J. Am. Chem. Soc., 102: 3163 (1980).
333
-------
Table 1. Examples of reactivity studies on Nitro-PAH
COMPOUND
CONDITIONS t
1/2
PRODS
9-Nitroanthracenea part, carbon (dark)
silica (dark)
O3' no2
anthraquinone
1.2 d phenols, quinones
6 d "
L
1-Nitropyrene UV/DMSO soln
UV/silica
u
3-Nitrofluoranthene UV/DMSO soln
UV/silica
12.5 d
>20 d
II
II
II
II
1,8-Dinitropyrene^ UV/DMSO soln
UV/silica
0.7 d
5.7 d
II
II
*plus significant l-nitropyren-8-ol
a. D.A. Saucy, R.M. Kamens, R.W. Linton, Polycyclic Aromatic
Hydrocarbons, Ninth International Symposium, M. Cooke and
A.J. Dennis (eds), Battelle Press, Columbus, 1986, pp. 811-825.
b. P.C. Howard, M.C. Biaglow, M.P. Holloway, E.C. McCoy, M.
Anders, H.S. Rosenkranz, Presentation at Tenth International
Symposium on Polycyclic Aromatic Hydrocarbons, Columbus, Ohio,
October 21-23, 1985.
334
-------
Table 2. Comparison oE experimental IP and EA data for PAH and
nitro-PAH (eV) with corresponding MNDO values
MNDO MNDO
IP(expt) IP1 IP2 Diff EA
Nitrobenzene (i) 9.86 10.31 10.46 0.15 0.80 {0.97
1-Nitronaphthalene (3) 8.60 9.30 9.81 0.51
2-Nitronaphthalene (3) 8.65 9.30 9.91 0.61
9-Nitroanthracene (£) 7.87 8.72 9.57 0.85 1.59
1-Nitroanthracene (|) 8.60 9.51 0.91 1.47
2-Nitroanthracene (Jj) 8.62 9. 59 0.97 1.49
9-Nitrophenanthrene (7) 9.16 9.24 0.08 1.24
1-Nitroacenaphthylene"(8) 9.41 9.58 0.17 1.82
3-Nitroacenaphthylene ($> 9.40 9.54 0.14 1.67
4-Nitroacenaphthylene (19) 9.35 9.54 0.19 1.60
5-Nitroacenaphthylene (y.) 9.37 9.58 0.21 1.66
1-Nitropyrene (12) 8.62 9.35 0.73 1.57
2-Nitropyrene (ip) 8.58 9.40 0.82 1.47
2-Nitrofluoranthene (li) 9.03 9,21 0.18 1.55
?-Nitrofluoranthene (1?) 9.00 9.25 0.25 1.58
5-Nitrotetracene (16) 8.27 9.39 1.12 1.81
7-Nitrobenz(a)anthracene (17) 8.73 9.11 0.38 1.52
6-Nitrobenzo(a)pyrene (18)"" 8.45 9.11 0.66 1.73
Benzene 9.25 9.39 9.39 0.00
Naphthalene 8.15 8.58 9.11 0.53
Anthracene 7.41 8.05 8.97 0.92
Phenanthrene 7.86 8.48 8.70 0.22
Acenaphthylene 8.22 8.53
Pyrene 7.41 8.03 8.77 0.74
Pluoranthene 7.95 8.47 8.72 0.25
Tetracene 6.97 7.72 8.89 1.17
Benz(a)anthracene 7.41 8.11 8,62 0.51
Benzo(a)pyrene 7.12 7.83 8.65 0.82
.a
A. Experimental value
335
-------
Table 3. Experimental IP and EA Affinities of selected
nitroaromatics.
IP(eV) EA {kcal)
Benzene
9.25
Nitrobe nzene
9.86
22.3
1r2-Dinitrobenzene
410.71
1,3-Dinitrobenzene
10.3
36.2
1,4-Di nitrobenzene
10.43
43.6
1,3,5-Trinitrobenzene
10.96
60.6
2-Nitroaniline
8.27
3-Nitroani1ine
8.31
4-Nitroaniline
8.34
3-Nitrobenzonitrile
10.29
4-Nitrobenzonitrile
10.23
2-Nitrotoluene
9.45
20.6
3-Nitrotoluene
9.48
21.3
4-Nitrotoluene
9.40
20.9
2-Nitro-o-xylene
9.17
4-Nitro-m-xylene
9.10
2-Nitroethylbenzene
9.39
3-Nitroethylbenzene
9.64
4-Nitroethylbenzene
9.71
1,3,5-Trinitrotoluene
3.8
Toluene
8.82
Ethylbenzene
8,77
Table 4. Relationships of IP, and IP_ and their differences for
PAH and nitro-PAH.(MNDO valQes in SV).
PAH NITRO-PAH IP^ ip DIFF
Benzene
Nitrobenzene
9.39
10.31
9.39
10.46
0.00
0.15
Phenanthrene
9-Nitrophenanthrene
8.48
9.16
8.70
9.24
0.22
0.08
Naphthalene
1-Nitronaphthalene
2-Nitronaphthalene
8.58
9. 30
9.30
9.11
9.81
9.91
0.43
0.51
0.61
Benz(a)anthracene
7-Nitrobenz(a)anthracene
8.11
8.73
8.62
9.11
0. 51
0.38
Anthracene
1-Nitroanthracene
2-Nitroanthracene
9-Nitroanthracene
8.05
8.60
8.62
8.72
8.97
9.51
9.58
9. 57
0.92
0.91
0.96
0.85
Tetracene
5-Nitrotetracene
7.72
8.27
8. 89
9.39
1.17
1.12
336
-------
Figure 1.
EXPERIMENTAL ADIABATIC IONIZATION POTENTIALS
OF AROMATICS AND CORRESPONDING NITROAROMATICS
10. 0
2-NN
1-NN
45-\
.50
9-NA
46
BENZENE
NAPHTHALENE ANTHRACENE
337
-------
Figure 2.
4Hf -35.7 kcal
(16.1)
-.Ofel
HOMO -10.3 eV LUMO -0.80
H0M02 -10.5
+.0ZI
-.06f
AH, - 82.0 kcal
HOMO -9.4 eV LUMO -1.67
HOM02 -9.5
+.o n
AHf- 80.5 kcal
8
HOMO -9.4 eV
H0M02 -9.6
-.018
LUMO -1.81
-01+
AH, ¦ 81.5 kcal
f
HOMO -9.A eV LUMO -1.60
HOMO2 -9.5
AH, -82.6 ^c41
V
HOMO -9.4 e
H0M02 -9.6
LUMO -1.66
AH, - 54.7 kcal
1 (26.5)
2
HOMO -9.3 eV LUMO -1.18
HOMO2 -9.8
+.020
AHf= 53.1 kcal
+.03&
HOMO -9.3 eV LUMO -1.2°
H0M02 -9.9
AH. - 73.3 kcal A H
78,0 kca
&Hf - 75.3 kcal
HOMO -8.6 eV LUMO -1.47 HOMO -8.6 eV HOMO -8.7 eV
H0M02 -9.5 LUMO -1.49 LUMO -1.59
HOMO2 -9.6 H0M02 -9.6
338
-------
Figure 2 cont'd
A K ,
72.7 kcal
HQMO -9.2 aV
H0M02 -9.2
+.030
LUMO -1,24
AHf = LOQ . 1. kcal
HOMO -8.3 eV LUMO -l,
H0M02 -9.4
81
A H,
94,4
HOMO -8.7 eV
homo2 -9.1
LUMO -1,52
AH,
101.9
HOMO -8.4 eV
H0MQ2 -9.1
LUMO -1.73
339
-------
Figure 2 cont'd
-.05*5
75 .33 kcal
HOMO -8.6 eV LUMO -1.47
H0M02 9.40
-.06 5
12
A/
A Hf = 77.9 kcal
HOMO -8.61 eV LUMO -1.57
H0M02 9.35
+.0>n
14
/V/
/iHf = 87.2 kcal
HOMO 9.0 eV LUMO -1.55
H0M02-9.2
-.060
AH^ ¦ 88.75 kcal
15
v
HOMO 9.0 eV
H0M02 -9.2
LUMO -1.58
340
-------
A PERSONAL COMPUTER DATABASE FOR THE CHEMICAL,
PHYSICAL AND THERMODYNAMIC PROPERTIES OF THE
POLYCYCLIC AROMATIC COMPOUNDS.
Douglas A, Lane
Environment Canada,
Atmospheric Environment
4905 Dufferin Street,
Downsview, Ontario, M3H
Donna M.A. McCurvin
Ontario Ministry of the
125 Resources Road,
P.O. Box 213,
Rexdale, Ontario, M9W 5L1, Canada.
A Database for the chemical, physical and thermodynamic properties of
the Polycyclic Aromatic Compounds (PAC) has been compiled for use on an
IBM XT or AT (or compatible) microcomputer using the commercially
available SCIMATE data management software. Data, both experimental and
calculated, have been culled from journals, text books, theses and
government and Industry reports for 729 PAC including the unsubstituted,
parent Polycyclic Aromatic Hydrocarbons (PAH), and their alkyl, amino,
oxygenated, and nitro derivatives. Each class of substituted PAH occupies
a separate floppy disk, A separate references disk contains over 440
references.
No attempt has been made to assess the data for accuracy, no data
have been omitted because they appeared to be outlier data, nor has any
data been recommended. In addition, data have been listed as presented in
the literature publications, and where possible, the method used to
determine the value of a property has also been indicated.
From this effort, several observations can be made: (i) there is a
great paucity of information for many properties for most PAC; (ii) the
data reported for a single property for a single compound may vary by
several orders of magnitude; (iii) where several experimental methods
exist to determine a single property, the different methods yield
consistent, but different data; (iv) there are a number of errors in the
literature arising from the incorrect translation of data (reported in a
previous publication) from one set of units to another.
5T4, Canada
Environment,
341
-------
Introduction
In order to assess the potential chemical and biological threat of
the PAC to the environment, to follow their migration through the
biosphere, to perform effective analytical determinations and to
characterize their behaviour in natural and synthetic fuels, it is
necessary that the chemical, physical, thermodynamic, spectral and
biological properties of the PAC be available and readily accessible.
Unfortunately, there is a great paucity of such information in the
literature and, that which does exist, is often widely scattered through
journals, textbooks and government reports (many of which are difficult,
if not Impossible, to obtain). In some instances, data from one report
has been incorrectly quoted (through mistranslation from one set of units
to another) in another publication.
It was for the above reasons that work was started to create a
comprehensive PAC database.
Hardware and Software
The PAC database, to be most useful and most accessible, was created
on an IBM XT personal computer using the SCIMATE (1) data management
software.
Before data can be stored in a record, a template must first be
created to overlay the data which will be input to the records. Each
template may contain from one to twenty fields of data within the record.
After the fields have been defined, there is no limit to the number of
characters per field other than the total record limit of 1894
characters. If it is anticipated that a record will contain more than
1894 characters, two records may be linked together before data is
entered. Linking of records cannot occur after the first edit.
Data in the database may be searched in several ways. Searches may
be conducted on a key word or phrase and the Boolean operators AND, OR and
ANDNOT may be used to facilitate the search procedure. In addition, field
directed searches enable one to search for data in one specific field
only. While the SCIMATE system has some limitations, it has proven to be
quite satisfactory for the purpose at hand. Searched data may be
manipulated directly or transferred to a work file for further
processing. Columnar reports may be generated from the entire user file
or from records retrieved from a search.
The Database
The PAC database currently contains (on 7 floppy disks) over 800
Kbytes of information on 729 PAC taken from over 440 references. A
compilation of data for the aza-arenes has been started but that data is
not covered in this discussion. The fifty-two properties included in the
database are listed in Table I.
It would appear, at first glance, that there is a great quantity of
data about the PAC. However, upon closer examination, one finds that
there is considerable data for a few properties for a very few PAC. This
is well demonstrated in Table II which shows, for 23 of the properties,
the number of compounds in each PAC class for which any data, at all, has
been found. In the final column, the total number of compounds for which
data has been found for a specific property is listed. For example, of
342
-------
729 compounds in the database, melting point data has been found for only
548 (or 75%) of the compounds. Vapour pressures have been found for only
16 (2.2 %) compounds and aqueous solubility data far 47 (6.4 %)
compounds. What is most revealing is the number of zeros appearing in the
table. There is an obvious need for more data in many areas.
A closer inspection of the data for a particular property for a
specific compound reveals some potential problems. For example, consider
the vapour pressure data, shown in Table III, for anthracene. The
reported values range from 2.71 E-04 KPa to 4,02 E-07 KPa, Over three
orders of magnitude separate these 20 values. Which one, if any, is
correct? Further inspection of the vapour pressure data for other PAC
shows that the data reported depends upon the measurement technique used.
Solid phase measurements give results one or two orders of magnitude lower
than those made from sub-cooled liquid phase measurements. HPLC
measurements usually agree quite well with the solid phase measurements
while values obtained from the Antoine calculation are usually higher than
those obtained from solid phase measurements. All of this can lead to
confusion if only one or two values are available. Indeed, the more data
that is available, the more difficult it is to decide which value to use.
Occasionally errors do occur in the journals. The vapour pressure
for Benz(a)anthracene cited in atmospheres (20) was incorrectly translated
to KPa and reported as 6.67 E-13 KPa (2). Six other literature values
averaging 1,99 (± 0.8) E-08 KPa show this result to be erroneous, It
is, therefore beneficial to have several measured values from which to
chose a preferred value.
Conclusions
A computer database for the PAC has been initiated. It has
demonstrated the great lack of chemical, physical and thermodynamic data
available for the PAC, It has revealed a great variation in data for
particular properties of some compounds and has detected some outliers
which turned out to be errors of translation from one set of units to
another.
It is the hope of the authors that this compilation will be useful to
those who require such data and to those who wish to perform experiments
to fill in some of the many gaps in the database.
References
(1) Institute for Scientific Information, 3501 Market Street,
Philadelphia, PA 19104, U.S.A.
(2) D. MacKay, W.Y. Shiu, "A critical review of Henry's law constants for
chemicals of environmental interest," Phvs. Chem. Ref. Data. 10(4):
1175 (1981).
(3) B.J. Zwolinski, R.C. Ilhoit, Handbook of Vapor Pressures and Heats of
Vaporization of Hydrocarbons—and—RejU^d—Compounds. API-44, TRC
Publication No. 101, Texas A&M University, 1971.
(4) F.s, Mortimer, R.V. Murphy, Ind. Eng. Chem.. 15: 1140 (1923).
(5) 0.A. Nelson, C.E. Senseman, Ind. Eng. Chem.. 14: 58 (1922)
343
-------
(6) NRC Secretariat, Polvcvcllc Aromatic Hydrocarbons in the Aquatic
Fnvi rnnment. National Research Council of Canada, Publication No.
18981, 1983.
(7) Ontario Ministry of the Environment, Polvnuclear Aromatic
Hydrocarbons - A Background Report Including Available Ontario Data.
Publication ARB-TDA Report No. 58-79 (1979).
(8) G. Stevens "Vapour pressures and Heats of sublimation of anthracene
and of 9,10-Di-phenylanthracene," J. Chem. Soc.. 2973-2974 (1953).
(9) R.C. Weast Ed., Handbook of Chemistry and Physics. 53rd edition,
1972-1973.
(10) A.B, Macknick, J.M. Prausnitz, "Vapor pressures of high molecular
weight hydrocarbons," J. Ghem. Eng. Data. 24: 175 (1979).
(11) R.D. Bradley, V. Fried, E. Hala, "The vapour pressures and lattice
energy of some aromatic ring compounds," J. Chem. Soc,. 1690-1692
(1953).
(12) J.D. Kelly, F.D. Rice, "The vapor pressures of some polynuclear
aromatic hydrocarbons," J. Phvs. Chem.. 69: 3794 (1964).
(13) H,G. Wiedermann, H.P. Vaughan, "Application of thermogravimetry for
vapor pressure determination," Proc. Toronto Svmp. Therm. Anal.. H.G.
McAdie Ed., 233-249 (1969).
(14) L. Malpasina, R. Gigli, G. Bardi, "Microcalorimetric determination of
the enthalpy of sublimation of benzoic acid and anthracene," J. Chem.
Phvs.. 59: 387 (1973).
(15) S.P Wasik, M.M. Miller, Y.B. Tewari, W.E. May, W.J. Sonnefeld, H.
DeVoe, W.H. Zoller, "Determination of the vapor pressure, aqueous
solubility and octanol/water partition coefficient of hydrophobic
substances by coupled generator column/liquid chromatographic
methods," Residue Rev.. 85: 30 (1983).
(16) American Petroleum Institute, Monograph Series 707, Naphthalene; 708,
Anthracene and phenanthrene; 709, A ring condensed series. API,
Washington, DC, 1979.
(17) B.T. Grayson, L.Y, Fosbraey, "Determination of the vapour pressure of
pesticides," Pestic. Sci.. 13: 269 (1982).
(18) J.W. Taylor, /R.J. Crookes, "Vapour pressure and enthalpy of
sublimation of 1,3,5,7-tetranitro-1,3,5,7-tetraazacyclo-octane
(HMX)," J. Chem. Soc. Faraday Trans.. 72: 723 (1976).
(19) W.H. Power, C.L. Woodworth, W.G. Loughary, "Vapor pressure
determination by gas chromatography in the microtorr range
anthracene and triethylene glycol di-2-ethyl butyrate," J.
Chromatoer. Sci.. 15(6): 203 (1977).
(20) C. Pupp, R.C. Lao, J.J. Murray, R.F. Pottie, "Equilibrium vapour
concentrations of some polycyclic aromatic hydrocarbons, ^s^O,
and SeO„ and the collection efficiencies of these air pollutants,
Atmos. Environ.. 8; 915 (1974).
344
-------
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
TABLE I
List of properties covered in the PAC database.
Appearance
27.
Free Energy of Proton Transfer
Aqueous Solubility
28.
Free Energy of Solubilization
Boiling Point
29.
Heat Capacity
Carcinogenic Potential
30.
Heat Capacity of Solubilization
Chemical Abstracts Number
31.
Henry's Law Constant
Chemical Formula
32.
Hydrogen Affinity
Critical Density
33.
HMO Derealization Energy
Critical Pressure
34.
Ionization Potential
Critical Temperature
35.
Melting Point
Critical Volume
36.
Molar Refractivity
Density
37.
Molecular Connectivity Index
Dielectric Constant
38.
Molecular Weight
Dipole Moment
39.
Name
Electron Affinity
40.
Organic Solubility
Enthalpy of Combustion
41,
Partition Coefficient
Enthalpy of Formation
42.
Phosphorescence Spectra
Enthalpy of Fusion
43.
Photochemical Half-Life
Enthalpy of Ionization
44.
Proton Affinity
Enthalpy of Proton transfer
45.
Refractive Index
Enthalpy of Solubilization
46.
Saturation Vapour Concentration
Enthalpy of Sublimation
47.
Singlet Energy
Enthalpy of Vaporization
48.
Synonyms
Entropy at 298 K
49.
Thermal Reactivity
Entropy of Solubilization
50.
UV Maxima
Equilibrium Vapour
51.
Vapour Pressure
Concentration
Fluorescence Spectra
52.
Van der Waals Volume
345
-------
TABLE II
The Number of compounds within each class for which data has been found
for selected data fields.
Parameter
Parent
Alkyl
Amino
Keto
Nitro
Quinone
Total
Compounds per class
109
355
30
62
92
81
729
Aqueous Solubility
26
21
0
0
0
0
47
Boiling Point
At
92
7
2
10
5
157
Carcinogenic Potential
70
154
1
1
4
5
235
Density
14
26
6
1
1
4
52
Electron Affinity
16
0
0
0
0
0
16
Enthalpy of Combustion
14
5
2
0
0
0
21
Enthalpy of Formation
36
4
2
0
0
0
42
Enthalpy of Fusion
7
2
1
0
0
0
10
Enthalpy of Ionization
14
6
0
0
0
0
20
Enthalpy of Sublimation
25
0
0
0
0
0
25
Enthalpy of Vaporization
30
60
0
0
0
0
90
Heat Capacity
5
3
0
0
0
0
8
Henry's Law Constant
8
2
0
0
0
0
10
HMO Derealization Energy
7
2
2
0
1
0
12
Ionization Potential
59
26
0
0
0
0
85
Melting Point
75
300
30
5
78
60
548
Molecular Connectivity Index
30
13
0
0
21
0
64
Organic Solubility
30
48
23
23
21
7
131
Partition Coefficient
23
20
2
0
0
0
55
Proton Affinity
14
4
0
0
0
0
18
Refractive Index
3
5
0
0
0
0
8
Van der Waals Volume
9
0
0
0
0
0
9
Vapour Pressure
14
2
0
0
0
0
16
346
-------
Table III
Vapour Pressures for Anthracene
Vapour Pressure in KPa
at 25°C
Comments
Literature Reference
2,71
2.09
2.31
2.60
2.60
1.04
3.17
1,44
8.34
8.30
8.60
2.40
1.11
7.99
4.90
4,27
1.00
5.01
3.10
4.02
E-04
E-05
E-05
E-05
E-05
E-06
E-05
E-06
E-07
E-07
E-07
E-06
E-06
E-07
E-07
E-07
E-06
E-07
E-06
E-07
from Antoine Equation
from Antoine equation
from Antoine equation
from liquid phase
from solid phase
from solid phase
effusion method
effusion method
effusion method
microcalorimeter method
HPLC method
said to be literature average
(solid at 20°C)
(at 20°C)
(at 20'C)
(at 20*C)
(at 20°C)
Zwolinski (3)
Mortimer (4)
Nelson (5)
NRC (6)
OME (7)
Stevens (8)
Weast (9)
Macknick (10)
Bradley (11)
Bradley (11)
Kelly (12)
Uiedermann (13)
Malaspina (14)
Wasik (15)
API (16)
OME (7)
Grayson (17)
Taylor (18)
Power (19)
OME (7)
347
-------
GC/PID Analysis of Headipace for Volatile Organic*
Drinking Water
J . Jerpe, EPA Regional Laboratory, Maryland
G. He»ton, EPA/ERS, Philadelphia, PA
L. Wilder, Roy F- Wesion, New Jersey
-------
Introduction
Head space analysis as defined as the static sampling of the vapor phase
in thermodynamic equilibrium with the aqueous phaseiw is a technique which
contrasts with isolation methods of trace gas analysis from water such as
the dynamic organic analysis of gas stripping (purge and trap) procedure
of Bellar and Hchtenberg'2).
It has been shown 1n a lengthy and systematic study(3) that the partition
effect may be greatly modified by addition of salts (sodium sulfate),
temperature, and pH fluctuations. It is therefore possible to choose an
optimum condition for head space analysis whereby drinking water may be
profiled.
The situation exists, however, where potable water contaminated with
trihalomethanes in quite varying amounts presents an immediate and
substantial health hazard. It may be necessary to sacraflce rigorous
quality assurance procedures and the time and effort associated with
dynamic volatile organic analysis for results of a screening analysis of
samples as received. With a portable gas chromatograph of suitable
sensitivity, these testing procedures may be performed in the field with
a minimum of time to reporting concentrations.
It was our intention to evaluate such a method of head space analysis of
organic vapor above drinking water with a minimum of sample preparation.
These results reported as volume by volume were systematically compared
with results reported by other laboratory operators using flame ionization
detection and Ion-trap mass spectrometry. Residential well water (ground-
water) in many areas of Region III have volatile organic contaminants with
levels above the drinking water standards and acceptable human health risk
levels. Once detected, these problems must be addressed In the quickest
manner possible. Initial sampling is extensive and, to ensure safety of
the public health, must continue even after temporary solutions are
implimented (the installation of carbon filtration units). Because of
the large amount and the frequency of sampling required, and because of
the fast turnaround time needed for results, Region III EPA/TAT employed
a portable GC with a photoionlzatlon detector (Photovac, Inc.) for on-site
volatile organic analysis.
Field analysis was first used at a Superfund site after receiving
preliminary sampling results sent to a private laboratory. Eleven
residences at this site were shown to have groundwater contaminated with
various levels of tr1chloroethylene (TCE) and 1,1,1-trlchloroethane (TCA).
In this way, the technique of a rapid and sensitive field method could be
evaluated with attendant precision and accuracy. These water samples were
being consumed by local residents and Immediate screening results were
needed even on a sem1-quantit1ve basis.
When trlchloroethylene and or 1,1,1-trichloroethane were found 1n
appreciable concentration by photo-ionlzation portable gas chromatography
in the field, the specific sample was sent to the laboratory for
confirmation by injecting head space vapor above the solution Into a
Hnnigan Ion-Trap Mass Spectrometer to confirm the screenings analysis.
Furthermore, the mass-spectrometer was calibrated with a fabricated gas
mixture of 1,1,1-trichloroethane, benzene, and trlchloroethylene 1n order
to report as an external standard auto-quant1tat1ve mode concentrations
1n parts-per-b1H1on. Through the use of these techniques, all residential
well water samples were analyzed within 24 hours of collection. Quick
349
-------
turnaround time was necessary for several reasons. First, initial samp 1ing
results indicated which residences were affected and required Immediate
attention. Extent of contamination results also gave a rough indication of
the area of the source of the contamination and the flow direction of the
contamination plume. Knowing plume direction, one could estimate which
homes were most likely to become contaminated in the future. These homes
were sampled periodically to check for the presence of contamination.
Secondly, once carhorv filtration units were installed in individual homes,
periodic sampling and analysis was required to ensure that contaminant
breakthrough had not occurred. Samples collected before the filters (to
monitor changes in groundwater contamination), between the filters, and at
the residential taps were analyzed within 24 hours of collection. If
contaminant breakthrough had occurred at the tap, both filters were changed
immediately, If breakthrough was detected between filters but not at the
tap, the contaminated filter was removed and replaced by the filter nearest
to the tap, and a new filter was put inline nearest the tap.
Photovac detection limits were found to be significantly below below EPA
approved methods. In addition, EPA's Central Regional Laboratory (CRL) In
Annapolis, MD, also analyzed several samples using EPA Method 624 as a
QA/OC measure.
When samples were collected, each VOA bottle was filled half full, allowing
the volatile organics to equilibrate in the headspace/air space above the
sample. Standards of each contaminant were made in a similar fashion.
Known concentrations of compounds were Injected into known volumes of
water and allowed to equilibrate with the headspace. By comparing Photovac
analytical results with those of EPA Methods 601 and 624, the Photovac
headspace method has proven to be a quick, accurate field method of
analysis.
EXPERIMENTAL METHODS
Sample Collection and Preservation:
To rid residential water systems of standing water, each well was purged by
allowing the cold water to run for 10-20 minutes. After purging, the water
flow rate was reduced significantly arid samples were taken. For EPA 601
analysis, 40ml VOA bottles of the same lot number were completely filled
with cold water and capped. For headspace analysis, a 40 mL VOA bottle
was half-filled with cold water and capped. A field blank (distilled,
deionized water} was Included with each sampling round.
After collection, samples were labelled (date, location, and sampling},
placed in zlploc baggies, and then sealed in tin cans and placed on ice.
Samples were analyzed by the lab within 3 days of collection, and with
in 48 hours by the headspace method.
Headspace Analysis:
Instrumentation:
A Photovac Model 10A1Q (a portable GC with a PI0 detector) with a
Hewlett Packard 5990A Integrator was employed for headspace analysis.
A 4' x 1/8" SE-30 column was used with zero grade air (Linde Specialty
Gas UN 1022) as the carrier gas.
350
-------
Standards:
The following Supelco standards were used:
1) 1,1,1-trichloroethane (TCA), 1 ml
200 ug/ml in methanol
Lot No LA14565
Cat No 4-8614
2) Trichlorethylene (TCE), 1 ml
0,2 mg/ml in methanol
Lot No LA13702
Cat No 4-8606
Previous sampling results revealed that concentrations of these compounds
found in residential well water ranged from less than 1 ppb to greater
than 7500 ppb, making several concentration ranges of standards necessary.
Standard concentrations were prepared to encompass less than 1 to 10 ppb,
10-100 ppb, and 100-500 ppb. Any sample that contained greater than 500
ppb of any of the target compounds was diluted until it fell within the
standard concentration ranges.
A known concentration of the Supelco Standard was injected into a known
volume of water and allowed to equilibrate with the headspace. A minimum
of three different concentrations (with three injections each) of a
compound were used to make a standard curve, 0,1 uL of the 2 ppb standard
was injected into a 40 mL VOA bottle containing 20 mL of water to yield a
concentration of 1.0 ppb.
(0.2ug/Lx0.1ul std)/(20ml water) = O.OOlug/ml = l.Oug/L
Samples:
Before injection, samples were removed from ice and allowed to reach room
temperature (the same temperature of the standards). No preconcentratlon
of samples was necessary, as the headspace technique analyzes the headspace
above the water sample. The headspace concentrations are proportional to
the contaminant concentration in water. Once samples reached room
temperature, a known volume of headspace was Injected into the column.
Each sample was analyzed in duplicate to ensure accurate injection
techniques. If a sample contained contaminant concentrations greater than
the standard concentration ranges, It was taken from the full VOA bottle
and diluted.
Sample Splits:
Initially, a duplicate of each sample collected was also sent to a private
lab for analysis (volatile organic analysts by EPA Method 601). After
private laboratory results proved comparable to those of the Photovac,
approximately one out of three samples were sent to a private laboratory
as a QA/OC measure.
To positively identify the gas chromatography peaks of the target
compounds, random samples were injected by the same techniques into a
Finnlgan GC/Ion-Trap Mass Spectrometer and programmed to search and
quantitate 1n an automatic mode. The same standard solutions used for
Photovac screening were Intentionally used to generate calibration
statistics for the quantitative analysis.
351
-------
A formal fabricated gas standard from 99% pure neat solutions of 1,1,1-
trichloroethane, trichloroethylene and benzene prepared in a 1 liter gas
bottle was suitably diluted and injected into the RC/MS system as a cross-
referenced standard to compare the Supelco methanol solution standard
responses.
RESULTS and DISCUSSION:
Sample concentrations were calculated using peak areas of compounds
aluting with the same retention time as the standards. Concentrations of
each multiple injected sample were averaged to yield the final sample
concentration. These results were then compared with analytical results
of EPA Methods 601 and 624. Typical headspace analysis calibration curves
are displayed in Figure 3. A typical chromatorgraph of standard
injections, accompanied by several chromatographs of headspace injections
of samples are shown in Figures 1 and 2.
Over two hundred samples were analyzed using the headspace method: an
average of 30 samples day. Typical results of the headspace method are
shown below.
Tables 1 and 2 display the percent differences between the positive values
of the Photovac headspace method and private laboratory analysis (EPA
Method 601) for 1,1,1-trichloroethane (TCA) and trichloroethylene (TCE).
The mean relative percent difference and the standard deviation for 1,1,1-
trichloroethane are 50.1 and 28.2, respectively. Corresponding TCE values
are 76 and 55.3.
These results compare favorably with those found in the enclosed
Performance Evaluation Study and Interlaboratory Method Evaluation Study
(Method 601) results performed by the EPA Environmental Monitoring and
Support Laboratory Duality Assurance Branch in Cincinnati, Ohio, Results
from their interlaboratory study reveal a standard deviation of 23 and 93
from the mean (based on 100 ppb and 400 ppb), and 12.4 and 61 for TCE
(based on 41,6 ppb and 200 ppb).
The standard deviation of the mean percent differences calculated for the
affected residences are in good comparison with the standard deviations
from the mean of the study (refer to Table 3).
When comparing the variation between Photovac and private laboratory
results, it should also be noted that different instruments and methods of
analysis were used. The private laboratory utilized EPA Method 601 which
involves a purge and trap sample extraction technique followed by
temperature-programmed GC analysis, while the Photovac method Involves
headspace analysis and no temperature programming.
352
-------
References:
1. Philip A. Wylie, "A Comparison of Head Space with Purge and Trap
Techniques for the Analysis of Volatile Priority Polutants"; Hewlett-
Packard Co., Rt. 41 and Starr Road, Avon, Pa. 19311.
2. T. A. Bellar and J. J. Lichtenberg, Journal of American Water Works
Association, 66, 739 (1974).
3. Stephan I. Friant and Irwin H. Suffett, "Interactive Effects of
Temperature, Salt concentration, and pH of Head Space Analysis for
Isolating Volatile Trace Organics in Aqueous Environmental Samples";
Analytical Chemistry, 51, 2167 (1979).
353
-------
TABLE 1. Comparison of Photovac and Private Laboratory Split Sampling
Data for Trichloroethylene (Positive Values).
Photovac
Private Lab
Relative %
(ppb)
(ppb)
Difference
137.2
120
13.4
144.5
120
18.5
15.7
6.8
79.1
20.3
6.8
99.6
150
94
45.9
194
94
69.4
174
100
54.0
224
100
76.5
13.2
5.8
77.9
75.1
69
8.5
2.2
1.3
51.4
35.4
29
19.9
Average Relative Percent Difference: 50.1
Standard Deviation + 28.2
TABLE 2. Comparison of Photovac and Private Laboratory Split Sampling
Data for Trichloroethylene (TCE).
Photovac
Private Lab
Relative %
(ppb)
(ppb)
Difference
0.3
0.5
50
3.6
0.5
151
2.2
0.5
125
3.8
4.7
21.0
5.0
5.0
0
1.0
9.0
160
9.0
7.0
25
Average % Difference: 76
Standard Deviation: + 55.3
TABLE 3. Comparison of Photovac and Private Lab Results With the EPA
Performance laboratory Study,
Compound
1,1,1-Trichloro-
ethane
EPA
(Standard
Deviation)
23
93
Concentration
(ug/L)
100
400
Photovac/Martel
Comparison
Standard
Deviation of
% Dfference
2R.2
Concentration
Range (ug/l)
0 - 200
TCE
12.4
61
41.6
?oo
55.3
0 - 200
354
-------
Ctoonatograii hUfi]«; dWHSTON Awuiwd; SeHM-lMfi f? 20:20
Cornet: C, KANES-HELL 4U NEAMPACE PHOIHAC COHPARISON /3-C S1ANMM
Scan fiarge: 1 - 426 Scan: 79 Int = 0 • 1:20 SIC: 'M\: HB4b
W/
-
J"
t
£
«
|
*
•J
I
te>
e
-t*
w
Figure 1
GC/ion-trap MS Chromatogram
Well water head-space
•)
1 i
£ V £-
^ e t
^ t -t
5 1
100 21
0 300 MB
355
-------
ChwmatogMM _ latafilt: (IMN Acauii*«d: ftug-26-1986 11:3!: 41
Coiwent; 31,0 m 275 FPU SPLlIIiJ.l-ICE/TRICHWROtTHyiJNr EACH PER L
Scan Range; 1 - 406 Scan: 145 Int = 72 0 2:26 RIC: 100/.: 46904
Figure 2
Computor report: Auto-Quan Mode
67146679
19800(274909
200
3:21 5:01
¦amarnrjiM
46612
265159
UJLAl
I I I I I II |
i 111 ] i ri i j 1111111 t t j rrr r^'FT^ ¦ | rw1
No CoMpounA/Standard Hade Exp RT Thrsh Act RT Fit S/N Scant Peak Ana
3:27
4:11
700
700
3:25
4:08
995
986
204
247
198576
265159
356
-------
Calibration Plot (Ext Stds) Filename: ICE Correlation Coeff: 0,964
1,1,1-TRICHLOfiOEIHAME Compound: 1 of 1 Standard Deviation: 8,988
(Peak Ana of Sample) vs (Aiwuftt of Sample Injected) (LinlLin)
KA
3580 •
m-
2500
JS
Figure 3
External Standard Calibration
%
UlT'HI I |T n LIIT11| III III H I j III 111 111111 1111II T| I fill Vl'l IJ1 I'll HI'11 j' 111111 111 j 111 111 111 111 111111
208.888 388.8»d 488.888 588.1
-------
OVERVIEW OF THE EMSL METHODS COMPARISON STUDY
by
W. A. McClenny
Environmental Monitoring Systems Laboratory, U. S. EPA
Research Triangle Park, NC 27711
The question of which of several candidate sampling systems to use for
multicomponent collection at dry deposition network sites has been addressed
by a methods comparison study. The study took place over a thirteen-day period
from 29 September to 12 October 1986, The three leading candidates for
multicomponent collection, i.e., the Canadian filter pack, the transition flow
reactor (TFR) and the annular denuder system (ADS), were operated on a daily
and weekly basis. A tunable diode laser (TDL) system was used as the reference
technique for measurements of nitric acid (HNO3) and nitrogen dioxide (NO2).
Passive sampling devices were used for daily and weekly measurements of NO2
while the Luminox NOg chemi1uminescence monitor was used for real-time monitor-
ing. The study site was located at a prototype dry deposition monitoring site at
the US EPA facility in the Research Triangle Park, North Carolina, U.S.A.
Introduction
The EMSL developmental and evaluation effort for dry deposition monitoring
has recently been focused on the selection of instrumentation for sampling
selected ambient gases and particles. Multicomponent samplers designed for
continuous collection of targeted species, such as sulfate-containing particu-
late, SO2, HNO3, etc., are being considered for long-term use in the 100-sta-
tion national network to be implemented through an EMSL contractual effort.
Specifically, the factors influencing selection of sampling units for weekly
and sub-weekly multi-component collection are under study. These factors
include accuracy, artifact-free collection, reliability and expense. Chief
among the candidate sampling units are: (a) the filter pack (FP) system (usually
referred to as the Canadian filter pack because of their extensive usage*); (b)
the transition flow reactor (TFR)2*3; (c) the annular denuder system (ADS) as
altered from the earlier Italian version ^ to meet dry deposition monitoring
constraints^. These systems plus a tunable diode laser system ® for providing
reference measurements of HNO3 and NOg during common sampling periods, were
brought together for a field monitoring comparison study over a thirteen-day
period from 29 September to 12 October 1986. Supplemental measurements of NO2
were made with the commercially available Luminox® Model LMA3 continuous point
monitor? and with passive sampling badges coated with trlethanolamlne.^ These
systems were assembled at the site of EMSL's dry deposition site 1n the Research
Triangle Park, North Carolina. Routine measurements of meteorological parame-
ters and ozone were also being recorded at this site.
Discussion
The five research groups that were assembled for this study are listed in
Table I. Each of these is represented in the program for this session for the
purpose of presenting and discussing the results of their individual experiments
358
-------
and the reference measurements. The last presentation in the session is intended
to provide a comparison of results across different methods and an identifica-
tion of conclusions and of unresolved issues.
Table 1. Participants and Measurement Methods
Research Group
Unisearch Associates, Inc
Toronto, Canada; G, MacKay
Atmospheric Environmental Service,
Downsvlew, Ontario, Canada;
K. G. Anlauf, H. A. Wiebe,
D. C. MacTavlsh
Research Triangle Institute;
J. E. Sickles, II, L. L. Hodson,
E. E. Rickman, Jr., M. L. Saeger,
D. L. Hardison, A. R. Turner,
S. K. Sokol, E. D. Estes
U.S. EPA and Northrop Services
J. D. Mul1k (EPA) and D. E.
Williams (Northrop Services)
Northrop Services; 0. K. Bubacz,
E. H. Daughtrey, J. D. PI el 1 and
K. G. Kronmiller
Measurement Method; Measured Species
Tunable Diode Laser: NOg and HNO3
Filter Packs with several sampling
periods: NO2, SO2, sulfate, nitrate,
ammonium, HNO3
Annular Denuder System (ADS),
Transition Flow Reactor (TFR);
ADS: SO2, HNO3, MONO, NH3, sulfate,
nitrate, ammonium; TFR: SO2, HNO3,
NO2, NH3, sulfate, nitrate, ammonium
Triethanolamine-coated filter 1n
holder (PSD); NO2
Luminox Model LMA3 real-time measure-
ments ; NO2
The original sampling schedule is shown 1n Table 2. Daily sampling periods
were chosen as the basis for statistical comparisons among the multicomponent
samplers. Only two Instruments were run on a weekly sampling schedule. These
were the ADS and the passive sampling badges. The originally planned TFR and
filter pack weekly samples were to be made by personnel providing the routine
operation of the dry deposition station. However, these samples were not valid
because of problems encountered during a change in operational responsibility
from one group to another. ADS and PSD weekly sample results were compared to
the sum of the dally values for the seven-day period from 30 September until 7
October and for the five-day period from 7 October until 13 October. Both
Luminox and diode laser measurements were averaged on a hourly, dally and also
weekly basis for appropriate comparisons. Dally runs began at 10:30 a.m. and
ended at 8:30 a.m. the following day. The two hour interval from 8:30 to 10:30
was reserved for flow audits, calibration and data processing.
359
-------
Table 2. SAMPLING SCHEDULE
September October
29 30 1 2 3 4 5 6 7 8 9 10 11
TFR (daily) x xxxxxxxxxx x x
ADS (daily) x xxxxxxxxxx x x
Filter Pack (daily) x xxxxxxxxxx x x
Luminox (continuous)
Passive Badges (daily) x xxxxxxxxxx xx
Diode Laser (continuous)
TFR (weekly) H f 1
ADS (weekly) I- 1
Filter Pack (weekly) (¦ 1 -I
Passive Badges (weekly) t- 1 1
The placement of the Unlsearch van (housing the tunable diode laser) and
the AES van, with respect to the station, is shown in Figure 1. Figure 2 shows
duplicate daily TFR, AOS, and FP samplers, as well as weekly AOS and TFR units,
on the roof of the station. Triplicate PSDs were located on the roof also,
while a second set of triplicate PSDs were placed Inside the station and exposed
to ambient air brought Into the station through a large diameter manifold. Two
Luminox NOj monitors and the TDLAS were located inside the Unlsearch van.
Inlets for these units were positioned adjacent to the station near the TFR and
ADS inlets. Short-term filter pack sampling units using an automated sequencer
were located on the room of the AES vans.
A summary of the results to be compared for each of the methods 1s given
1n Table 3.
360
-------
TABLE 3. SUMMARY OF STUDY RESULTS
Species
ADS
TFR
FP
TDLAS
Lunvi nox
PSD
N02
X
X
X
X
X
so?
HNO3
X
X
X
x,xa,xa
xh
X
X
NH3
X
X
Total
NO3-
X
X
X
Total
nh4+
X,XC
X,XC
X
Total
SQU-
X
X
X
a HNO3 computed from TFR results assuming it functions similarly to a FP;
it considers the sum of NOo" recovered from the nylon strip and filter as
an upper limit estimate of I1NO3.
k HNOo from the FP assumes no particle NO3" volatilization and that the NO^"
on the nylon filter represents only HNO3.
c Total gaseous and particulate ammonia and ammonium (Total NH3 + are
also considered,
^ HNO3 was independently measured by Tom Ellestad of the Atmospheric Sciences
Research Laboratory (ASRL) of the US EPA using equipment that was nominally
identical to the equipment used by RTI (supplied to RTI by Ellestad).
Quality control for the study included both flow checks and calibrations
during the daily break in sampling. Standards for HNO3 and NO2 were taken from
calibration systems (permeation tubes) on board the Unisearch van. RTI person-
nel coordinated sample exchanges between their laboratory and the AES labora-
tory for normalization of instrument response with respect to analyses of
extracts from filters and denuders.
Disclaimer
The research described in this article 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 recommend-
ation for use.
361
-------
REFERENCES
1. Aulauf, K. G», Fellfn, P., Wiebe, K. A., Schiff, H. I. MacKay, G. I,,
Braman, R. S., and Gilbert, R., Atmos. Environ. 19, 3Z5 (1985).
2. Knapp, K. T., Durham, D. L,, and Ellestad, T, G., Environ. Scl. and
Techno!M, 633 (1986).
3. Durham, J. L., Ellestad, T. G., Stockburger, L., Knapp, K. T., and
Spiller, L. L., JAPCA, 36, 1228 (1936).
4. Possanzinl, M., Febo, A., and Liberti, A., Atmos. Envrlon., 17, 2605
' (1983).
5. Sickles, J. E., II, Hodson, L. U, Rlckman, E. E., Jr., Saegar, M. L.,
Hardison, 0. L., Turner, R., Sokolt C. K. and Estes, E. D., "Measure-
ments of HNO3 and Other Dry Deposition Components Using the Annular
Oenuder System and the Transition Flow Reactor" in this volume,
6. Schiff, H, 1., MacKay, G. 1. Castled1n1, C., Harris, G, W. and Tran, Q.
"A Sensitive Direct Measurement NO2 Instrument", p. 834 in Proceedings
of the 1986 EPA/APCA Symposium on Measurement of Toxic Air Pollutants,
Raleigh, NC, April 1986.
7. Hastie, D. R., MacKay, G, I., Iguchl, T., Ridley, B, A., and Schiff,
H. I., Environ. Sci. and Techno!. 17, 352A (1983).
8. Mulik, J, D. and Williams, D. "Passive Sampling Devices"* p. 61 in Pro-
ceedings of the 1986 EPA/APCA Symposium on Measurement of Toxic Air
Pollutants, Raleigh* NC, April 1986.
362
-------
Figure 1. Location of sampling stations for the study.
363
-------
Figure 2. Annular denuder system, transition flow reactor, and personal sampling
device placement on dry deposition station.
364
-------
FILTER PACK HOUSING
Figure 3. Filter pack and transition flow reactor (second daily sample)
placement on dry deposition station
365
-------
REFERENCE MEASUREMENTS OF HNQ, AND NCk
BY TUNABLE DIODE LASER ABSORPTION SPECTROSCOPY
G.I. Mackay and H.I.Schiff
Uriisearch Associates Inc.,
S22 Sriidercraft Rd.,
Concordi Ontario, L'tK 1B5
Canada
ABSTRACT
Measurements of HN03 and NOe in ambient air were made using a Tunable
Diode Laser Absorption Spectrometer (TDLAS). This system was used as a
reference technique for the two gases because of the unequivocal identifica-
tion and sensitivity achievable with the TDLAS technique. The two gases were
measured simultaneously under automatic computer control with detection
limits better than 100 pptv and response times of 5 minutes or less.
Average values of HNOa and NO*, were 0.57 ppbv and 8.7 ppbv respectively
and greater than 95Vi data coverage of the two week study was achieved. The
system was calibrated using permeation devices for both gases. Potentio-
metric titrations before and after the study and ion chromatograph measure-
ments of nitrate collected in bubblers containing KOH solutions during the
study indicated that the HMOa device was stable to within + 5'/., The N0e
permeation rate was determined by weighing before and after the study and was
stable to within 3'/.. Overall accuracy is estimated to be + 15'/» for N0B and
+ EOV, for HNOa.
Key word index; trace gas measurements, nitrogen dioxide, nitric acid,
tunable diode lasers.
366
-------
Introduction
Tunable diode laser absorption spectrometry (TDLAS) offers a number of
advantages for measuring atmospheric gases*. It is a universal method
applicable, in principle, to all gases of atmospheric interest. Its high
spectral resolution provides unequivocal identification of the target gas.
When operated with a long path cell it can measure most atmospheric gases at
sub part per billion (ppbv) concentrations# Real time* in situ, measurements
can be made with time resolutions of less than one minute.
It is also ideally suited to serve as a standard against which less
definitive methods, such as filter and denuder techniques, can be compared
and calibrated"*®. Its relatively rapid response permits measurements of
concentration fluctuations not possible with most other methods.
This paper describes an improved version of the instrument and measure-
ments of NO. and HND» made during the Environmental Protection ^9®™^
sponsored Dry Deposition Study at Research Triangle Park North Carolina,
during September—Dctober 19B6.
Description of th» TDLAS BystM
Figure 1 shows a schematic of the TAMS-150, an i«P«ved version of the
system described previously—. Temperature control of the Pb salt diodes in
their operating range of SO to BO K is provided to ± 0.005 K by the combine
tion of a closed cycle helium cryocooler, a heater and a servo temperature
system. Two cryostats are used, each having a laser source assembly contain-
ing ti laser diodes. One laser diode from each assembly can be chosen to
permit two gases to be measured simultaneously. The emitted radiation from
each of the diodes is scanned over the selected absorption feature by chang-
ing the current through the diode.
The laser beam from each head is collected and focussed by all-
reflective optics into the White cell. Mirror, S flips back and forth to
select the beam from each of the diodes in turn.
The beam enters the 1.5 m base path Teflon-lined White cell containing
a corner cube reflector" and undergoes 102 passes before exiting to the
detector. Absorptions down to about 10- can be
detection limits in the range 25 to 100 parts per trillion by volume for most
atmospheric gases.
About 5% of the beam is reflected by the *5- angle of the White cell
entrance window through a cell containing high concentrations of the target
gases onto a separate liquid nitrogen cooled HgCdTe infrared detector. The
output from this detector is used to lock the laser radiation wavelength to
the center of the absorption line.
Sampled air enters through an inlet, flows continuously down the tube
and is exhausted at the other end. A constriction at the inlet end and a
servo valve at the exhaust end maintains a constant flow rate and pressure in
the cell. Maintaining the cell at S5 Torr reduces pressure broadening of the
absorption line which both increases the sensitivity of the measurement and
minimizes the likelihood of interferences from other gases.
The absorption is measured in the frequency modulated, Sf, mode* which
results in an increased signal-to-noise ratio compared to using direct
absorption.
367
-------
The system is under computer control. Laser temperature and current
selections far each species are input to the computer. The system is then
switched to automatic mode and operates unattended for at least 2^ hours.
The computer signals the selection mirror to position itself so that
one of the beams enters the White cell. The line is scanned under computer
control at a rate of v30 Hz for 3 sec (the approximate residence time of the
gas in the White cell) and an accumulated 2f line shape is acquired.
At the end of the 3 sec averaging period the signal on the reference
detector, DS is checked. If the reference channel indicates that the line is
not exactly at the center of the line the computer adjusts the laser
temperature to faring the line back to the center.
The selection mirror is then commanded to bring the second beam into
the White cell and the measurement procedure is repeated far species B. Each
species is thus measured once every 6 sec. Data is accumulated in this way
for a period of * for example* 3 min providing an average value for the 2f
line shapes over that time frame. The data set, which provides one mixing
ratio for each species, is then stored on computer disk.
The computer manipulates solenoid valves and flow controllers to
perform measurement and calibration sequences automatically. In a typical
sequence the valve system is commanded to flow either zero (bottled) air or
scrubbed, ambient air through the White cell. Background spectra of the
components are obtained by scanning over the wavelength regions of the
selected absorption lines. A typical example is shown in Figure 2a.
Calibration gas is then added to the zero (or scrubbed) air flow and
allowed to stabilize for ""1 min. An averaged calibration spectrum is then
obtained for 3 min which also serves as a reference spectrum and the raw data
is archived (Figure 2b).
The background spectrum is subtracted (channel by channel) from the
reference spectrum. This procedure removes any frequency dependent structure
in the background from the reference spectra provided it remains stable for
the averaging period. The result of this subtraction is shown in Figure 2c.
Valves are then reset to admit ambient air, and after another
stabilization period an ambient air spectrum is acquired for 3 min (Figure
2d) and the background spectrum subtracted (Figure 2e), The net ambient
spectra is least square fitted to the net reference spectrum. The same
calibration sequence is followed for both gases being measured.
The frequency of calibration and the time duration of the various
stages of the sequence can be altered by the operator. The choice depends on
the mixing ratios of the species being measured; low ambient concentrations
require more frequent determination of the background and reference spectra.
Application of th» TOLAS Systm to the Meaaurmtent of NCW and HNQ»
The use of the system to measure an atmospheric component has 3
requirements! I) selection of a laser and an absorption feature! 2)
establishment of a calibration procedure; 3) ensuring sampling integrity.
The selection of a laser and an absorption feature represents a
compromise between the absorption line strength, the characteristics of the
laser emission at the wavelength of the line, and absence of interferences at
this wavelength from other atmospheric gases7. Suitable lines were selected
in the 1720 cm-' region for HN0« and in the 1&00 cm-1 region for NOa.
368
-------
The calibration philosophy calls for the addition of a known amount of
the calibration gas to the sampled air stream at the entrance to the sampling
line. The concentration of the "spike" is chosen to provide an increase in
measured mixing ratio comparable to the mixing ratio of the gas in the
ambient air. In this way, any surface effects that may occur will be the
same for the sampled and spiked air and should) therefore, compensate.
In contrast to NOe, which has a response time equivalent to the
residence time of the sample gas in the White cell, the response time to
additions of HNOa is about 5 min to reach 75'/, of steady state. To permit
more rapid calibration for HNOa an alternative method was used. During the
calibration period the calibration gas mixture was redirected into a 15 cm
cell (C in Figure 1) situated in the laser beam in front of the detector.
The pressure in the cell was maintained at 25 Torr to match the conditions in
the White cell and the flow was controlled at a level to provide an
equivalent mixing ratio in the White cell of 6 ppbv. At the beginning and
end of the study the HNOa spike was introduced at the inlet to sampling line
(our preferred method of calibration) to check validity of the optical
calibration. The agreement between the two methods was within 10X.
Permeation devices were used as the calibration sources. The NOb
source was a Metronics wafer device calibrated by weighing before and after
the study to be permeating at a rate of 144 + 2 ng.min-*. The HNOa source
was manufactured by Unisearch Associates and was a special design in which
the carrier gas passes through 3 length of Teflon tubing immersed in a tube
containing a mixture of HNOa and f-fe50«*. The permeation rate was determined
by pH titration before and after the study and by ion chromatography on the
extract from KOH bubblers during the study. From all determinations, an
average value of 267 + 13 ng.min-1 was obtained for the HNOa permeation rate.
Results
Data coverage, as expressed as percentages of the total measurement
period were better than 95V< for both species. The standard measurement
period was 10:30 through 08:30. The period 08:30 through 10:30 was set aside
for calibrations and maintenance of the instrumentation.
Figures 3 and 4 show respectively, the one hour averaged values of N0B
and HNOa taken on October 2. These observations are typical of the study
period. N0b had two daily maxima, morning and evening which coincided with
the peak automobile traffic in the area. 3 min data points ranged as high as
40 ppbv and down to 1 ppbv. HNOa showed a single maximum in early afternoon
which is more typical of photochemical production. 3 minute data points
reached maximums of ^ ppbv during the early part of the study and nighttime
minimum were generally in the 0.1 to 0.3 ppbv range with some periods below
the detection limit of 0.1 ppbv.
Figures 5 and 6 show the corresponding measurements for the entire
study. After the 6th of October the weather deteriorated, becoming mostly
overcast with heavy rain occurring on the 11th. This is evidenced by the
decrease in the HNO« maximum during the latter part of the study. Average
values for HNOb, and NCW over the two weeks were 0.57 ppbv and 8.7 ppbv
respectively.
369
-------
Conclusions
The tunable diode laser absorption spectrometer has been shown to be
capable of providing real time measurements of the diurnal behaviour of NCU
and HNQa in relatively clean air under field conditions. Its high
specificity, good sensitivity and rapid response time makes it a very
suitable standard against which other less definitive methods can be
compared.
Ui -fcera tyx ©. _Q..Li_e_d
1. H.I. Schiff, D.R. Hastie, G.I. Mackay, T. Iguchi, B.A. Ridley,
"Tunable diode laser systems for Measuring trace gases in
tropospheric air"i Enyir„^_Sci._T_e^ 17: 352ft (1983).
~. J.G. Walega, D.H. Stedman, R.E, Shetter, G.I. Mackay, T. Iguchi,
H.I. Schiff, "Comparisons of a chemiluminescent and a tunable
diode laser absorption technique for the measurements of nitrogen
oxide, nitrogen dioxide and nitric acid"* Envir. 8ci. Techno!.
IBj B23 <19841.
3. K.G. Anlauf, P. Fellin, H.A. Wiebei H.I. Schiffi G.I. Mackay,
R.S. Braman, R. Gilbert, "A comparison of three methods for
measurement of atmospheric nitric acid and aerosol nitrate and
ammonium") i?: ^£5 (19S5).
*t. G.I. Mackay» H.I. Schiff, H.A. Wiebei K.G. Anlauf, "Measurements
of NO*, HCHO and HNO» by tunable diode laser absorption
spectroscopy during the 1985 Claremant intercomparison study",
Atfnosptieric EnyijCQIffi? n t. in press.
5, D. Horn, G.C. Pimentel, "2.5-km low-temperature multiple-
reflection cetl't Aool. Optics JO: 199S 11971).
~. J. Reid, K. Garside, J Shewchun, M. El-Sherbiny, E.A. Balik,
"High sensitivity point monitoring of atmospheric gases employing
tunable diode lasers", Appl. Optics 17: 1B06 <197B).
7 F. Slemr, G.W. Harris, D.R. Hastie, G.I. Mackay) H.I. Schiff,
"Measurement of gas phase hydrogen peroxide in air by tunable
diode laser absorption spectroscopy", J. Geophys. Res. 91s 5371
(1936).
Figure Captions
Figure 1. Schematic of the optical system and control electronics
Figure S. Computer screen printouts for N£W> a) raw background spectrum;
b) raw calibration spectrum! c) calibration spectrum, background
subtracted; d) raw ambient spectrum; e) ambient spectrum,
background subtracted
Figure 3. Hourly averaged NQ« mixing ratios at RTP, NC on Oct 2, 1986
Figure
-------
TAMS-150 )«y»» Bwrtr* m*4"
usr.ir1"*
FIG. 3
30 ¦
ki-
lt -
li -
ft -
20 -
1* -
1* -
14 ¦
ia -
«-
«•
EPA DRY DEPOSITION STUDY
ItCSMGH TRMNOU WW OCT 2. 1M«
0 ¦^T"
13
31
371
-------
EPA DRY DEPOSITION STUDY
RESEARCH TRIANGLE PARK OCT 2, I960
0.9 -
O.B •
0.7 -
o.e -
0.3 -
0* -
0.3 •
0.2 -
9
13
17
21
5
1
EPA DRY DEPOSITION STUDY
RCSCAftCH TUtANCU PAHK. WT-OCT 1
35
30 -
25 -
15 -
10
29 301 2 3 * 8 « 7 B 9 10 11 <2
1MB AVlfJIett
EPA DRY DEPOSITION STUDY
RESCAffCH TWANOLE FA*K, SEPT-OCT
2.C
2 4
2 2
2
1
O.B
0.2
0
29 30 1
3
2
6
7
a
6 10 II 12
DATE
— FHR AVESACE5
372
-------
MEASUREMENT OF ATMOSPHERIC NITRIC ACID AND AMMONIA BY THE FILTER METHOD
AND A COMPARISON TO THE TUNEABLE DIODE LASER METHOD
Kurt Anlauf, Davfd MacTavish and Alan W1ebe
Atmospheric Environment Service
4905 Dufferin Street
Downsvlew, Ontario, Canada, M3H 5T4
Gervase Mackay
Unlsearch Associates Inc.
222 Snldercroft Road
Concord, Ontario, Canada, L4K 1B5
ABSTRACT
The measurement of atmospheric nitric acid by nylon filters Is
compared to that by tuneable diode laser spectroscopy over the range
0-2.5 ppb. lh/2h and 22h sampling periods are compared. Furthermore,
the effect of sampling period {up to 22h) on artifact nitric acid 1s also
discussed - any such effects were within experimental error and hence
were minimal under the study conditions. The results of measurements of
atmospheric ammonia (lh-12h sampling periods) by citric acid-Impregnated
W41 filters are also presented. The results of this comparison study
have shown that filter methods for measurement of atmospheric nftrlc acid
and, probably ammonia, may be adequate and reliable for rural conditions
1n N.E. America.
INTRODUCTION
Measurement of water soluble nitrogen and sulfur compounds 1n air
can be readily and economically made by means of filtration methods. By
using combinations of filters, atmospheric species such as particle
-NH$, -N05 and -SQ| and gaseous SO?, NH3 and HNO3 can be routinely
measured in monitoring networks. An eight-station regional network
(CAPMoN), employing filter methods is currently being operated in Canada
by the Atmospheric Environment Service for measurement of atmospheric
particles, SOg and HNO3.
There are potential artifact effects 1n measurements of the
particles and gases when using filter methods, since particles can have
9aseous precursors that exist 1n equilibrium with the particles. These
artifacts are caused by displacement of equilibrium conditions when
373
-------
changes occur In gas concentrations, temperature or humidity during
sampling. For example, the collection and subsequent evaporation of
NH4NO3 during long sampling periods (24 hour) could be an Important
process when filter measurements are made at locations with high
concentrations of NH3 and HNO3, high temperature and low relative
humidity. Furthermore, reactions with particles can result 1n the
evolution of reactive gases, which can be lost from particle filters and
collected by subsequent Impregnated filters.
Comparisons of nitric add and partlcle-NOj measurements by
filter and denuder difference methods^ over 24h sampling periods have
shown that In urban atmospheres, filter measurements were, on average,
about 25$ higher than those by the denuder difference method. This
difference was attributed to systematic artifact particle -NO3
volatilization during 24-hour sampling periods.
A comparison study of methods for measurement of atmospheric
nitrogen and sulfur species was held at Research Triangle Park, N.C. from
September 29 to October 12, 1986. This study provided an opportunity to
evaluate volatile nitrate losses during filter sampling of regional,
rural air masses. For this study, short duration (1h and 2h), extended
duration (4h - 12 h) and daily (22h) filter samples were collected.
Comparison of the averaged short duration with the extended period daily
filter samples should provide Information on artifact processes during
filter sampling for HNO3. It is expected that artifact processes will
be minimized for the short duration samples, since concentrations and
composition of particles, precursor gases, temperature and humidity are
unlikely to change dramatically during sampling. Besides the filter
measurements, a diode laser was operated by Unisearch Associates Inc.'
and provided an Independent measurement for comparison with HNO3
collected by nylon filters. Results of HNO3 and NH3 measurements by
the filter method and HNO3 measured by the aiode laser will be compared.
EXPERIMENTAL
Filter Holder and Sample Analyses. For filter sampling,
triple-stack filter holders of Inner fluorocarbon polymer and outer
polyethylene construction were used. Atmospheric particles were
collected on the front Teflon filter (1 micron pore size, Gelman
Sciences, Inc.). The second filter, made of nylon ('Nylasorb', Gelman
Sciences Inc.) was used for collection of nitric acid. Ammonia or SO2
was collected on a third, Impregnated Whatman 41 filter. For ammonia,
the Impregnation solution was an aqueous mixture of 10$ v/v glycerol and
25% w/v anhydrous citric acid (Baker, reagent grade). Details of
operation of the filter holder system and analyses of the collected
samples are reported in another publication3. The only variation to
these procedures was that 6 mL extraction volumes were used for the lh
and 2h samples and 20 mL for the 22h samples. The 22h filter samples
were originally extracted and analyzed at the Research Triangle Institute
(Research Triangle Park, N.C.). Remaining solutions were then returned
to the Atmospheric Environment Service for re-analys1s. The latter set
of results were used for calculations of atmospheric concentrations used
in this report.
Filter Blanks. Two different levels of water soluble nitrate and
chloride were found in the 13 nylon filters used as field blanks during
the study. These were evaluated as two separate sets. Seven of the
filters contained low levels of chloride (3-5ug/filter) and an average
374
-------
of 0.43m g/f-nter of NO3 (standard deviation -c - 0.10). The
other set of six contained 25-40ug/f1lter CI" and an average of 1.82
m g/f1lter NOg ¦ 0,37). Each of the samples collected on the
nylon filters was corrected by using the appropriate average blank value
according to the level of chloride found 1n the filter. For the high
chloride containing filters, the standard deviation corresponds to 0.06
and 0.03 ppb HNO3 for lh and 2h sample periods respectively (nominal
2.5 and 5 air volumes).
The average NH3 found in 13 citric ac1d-1mpregnated filters used
as field blanks was 0.47M g/filter ir = 0.17), with a range of 0.23
to 0.77m g/fliter. For the Ih and 2h samples, the Standard Deviation
of the average blank corresponds to 0.10 and 0.05 ppb NH3,
RESULTS
Results of filter measurements are expressed In ppb and were
calculated from air volumes referenced to 20°C and 1 atm. Figure 1 shows
HNO3 and NH3 concentrations as obtained from the lh/2h filter
samples. The photochemlcally formed HNO3 shows the typical diurnal
pattern for regional air masses, reaching a maximum 1n the mid-
afternoon. NH3 generally showed the opposite behavior, reaching a
maximum during night-time hours and declining to minimum during the day
(although there were some exceptions).
The results of a linear least squares regression comparison
between the filter and laser measurements are presented 1n Table I. The
95$ confidence intervals about the regression slopes and the intercepts
are Included.
J3
a
a
cn
X
z
-O
a
a
n
0
z
1
k
29 30 1 23456789
September 29 - October 1], 1906
Figure 1. lh and 2h concentrations (ppb) of atmospheric nitric
acid and ammonia during the comparison study period
at Raleigh, N.C.
375
-------
Table I. Linear least squares regression statistics on filter and tuneable
diode laser measurements of HNO3 and NH3 (concentrations in ppb).
n
r?
mean
slope CI y-int CI x
mean
y
lft/2h filter samples fy) vs laser (x) HNO.i
148
0.82
1.12 ±0.08 0.06 ±0.07 0.66
22h filter samples (y) vs laser (x) HNOi
0.80
0.24
11
0.66
0.85 ±0.46 0.09 +0.26 0.58
22h (y) vs lh/2h (x) filter samples HNO3
0.59
0.10
12
0.95
0.86 ±0.13 0.01 +0.10 0.72
4-12h (y) vs lh/2h (x) filter samples HNOs
0.63
0.03
16
0.99
1.00 ±0.07 -0.04 +0.07 1.00
4-12h (y) vs lh/2h (x) filter samples NHi
0.96
0.07
16
0.79
0.87 ±0.26 -0.01 ±0.30 1.10
0.95
0.20
n = no. of samples, r2 * square of the correlation coefficient, CI = 95%
confidence interval, Sy>x = standard error of estimate of y on x
1.
-Q
a ¦
a
w
n .
0
2
1
L
ID
•P
L.
Laser HN03 (ppb)
Figure 2. Linear least squares regression plot of measurement of
nitric acid by the nylon filter (22h sampling period)
and the tuneable diode laser (averaged over 22h)•
376
-------
1h/2h filter samples vs laser HNO3. Over the range of 0-2.5
ppb HNO3, there was more reasonable agreement and the degree of scatter
was considerably less than In a previous comparison . The detection
limit of the tuneable diode laser was about 0.1 ppb and Its accuracy was
estimated to be better than 20% over most of the measuring range2.
This uncertainty, and those In the filter blanks probably account for the
observed differences.
22h filter samples vs laser HNO3. Although there 1s excellent
agreement in the means of concentrations, there are considerable
uncertainties 1n the slope and intercept of the regression line. This 1s
not unexpected, given the few data points and the fact that all, except
one, were in a limited concentration range of about 0.4-0.8 ppb (see
Figure 2).
22h vs lh/2h filter samples HNO3. The average HN03
concentration as obtained by the 22h ¦filters was somewhat lower than for
the 22h-averaged lh/2h filters; likewise the slope was somewhat less than
one. If volatilization of particle nitrate were Important to our
measurements, an HNO3 concentration difference 1n the opposite
direction would have been observed. The observed difference may be
related to two causes: (1) unlike other samples, the 22h samples were
re-analyzed at AES about 3 months after extraction and analysis at RTI,
N.C., and some sample deterioration may have occurred; (Z) variability
In blank corrections for the lh/2h filters.
4-12h vs lh/2h filter samples HNO3. In contrast, to the
previous case, there Is excellent agreement between filters from two sets
of sampling periods. In this case, all filters had I"?0-??
analyzed at AES. Again, volatilization of particle nitrate 1s not
evident.
4-12h vs lh/2h filter samples NH3. Over the range 9f 0-2 p_pb,_
there was little difference In measured^ concentrations (see Table I
for details). Some of the difference may be attributed to variability 1n
blanks. As observed before for HNO3, the measured NH3 dlffererrce
does not suggest volatilization of particle ammonium nitrate as being
Important during our study,
CONCLUSIONS
A comparison of lh/2h and 22-h nylon filter measurements of
atmospheric nitric acid with those from a tuneable diode laser have shown
on average, good agreement at the low »nu
ppb). Varying the sampling period for the fjlte£s
-------
REFERENCES
1. P.A. Mulawa, S.H. Cadle, "A comparison of nitric acid and particulate
nitrate measurements by the penetration and denuder difference methods,"
Atmos. Environ. Jj>: 1317 (1985).
2. S.I. Mackay, H.I. Schiff, "Reference measurements of HNCh and NO? by
tuneable diode laser absorption spectroscopy", Proceedings of the 1987
EPA/APCA Symposium on Measurement of Toxic and Related Air Pollutants, Air
Pollution Control Association.
3. K.E. Anlauf, D.C. MacTavlsh, H.A. Wiebe, H.I. Schiff, 6.1. Mackay,
"Measurement of atmospheric nitric acid by the filter method and comparisons
with the tuneable diode laser and other methods", Atmos. Environ, (accepted
for publication).
378
-------
measurements of nitric acid and other dry
deposition components using the annular denuder
SYSTEM AND THE TRANSITION FLOW REACTOR
J. E. Sickles, II, L. L. Hodson, E. E. Rickman, Jr.,
M. L. Saeger, D. L. Hardison, A. R. Turner,
C. K. Sokol, and E. D. Estes
Research Triangle Institute
Research Triangle Park, North Carolina
R. J. Paur
Environmental Monitoring Systems Laboratory
Research Triangle Park, North Carolina
An intensive field study was conducted in the Research Triangle
Park, North Carolina in the fall of 1986. The main objectives of this
study were to conduct measurements of nitric acid and selected other
airborne gaseous and particulate acidifying species using different
types of samplers and to compare their performance under field sampling
conditions. This paper presents and compares results collected using
the annular denuder system (ADS) and the transition flow reactor
(TFR).
Ambient concentrations of the following constituents were
obtained: nitric acid, nitrous acid, nitrogen dioxide, sulfur dioxide,
ammonia, strong hydrogen ion, and particulate nitrate sulfate, and
ammonium. Concentration and precision results from both samplers are
compared. Statistical analyses of selected results are also presented.
379
-------
Introduction
The phenomenon of acid deposition has received considerable
public, political, and media attention. Dry deposition includes all
processes where airborne contaminants are removed from the atmosphere
at the earth's surface, excluding those processes aided by precipita-
tion. Since dry deposition may contribute substantially to the acidic
deposition burden, the U.S. Environmental Protection Agency (EPA) plans
to implement a dry deposition monitoring network. Several consituents,
including HNO3, HNOj, NOj, S02, NH3, NH^, NO3, S0£ , and H+ may be
important contributors to ecosystem acidification. Currently, three
multiple-constituent sampling systems are available for sampling
selected acidifying constituents from ambient air. They are the filter
pack (FP), the transition flow reactor (TFR), and the annular denuder
system (ADS). The objective of the current study was to collect data
that would permit the comparison of the performance of the ADS and the
TFR under field sampling conditions.
Experimental
A 2-week intensive study was conducted at the EPA dry deposition
monitoring site in the Research Triangle Park, North Carolina. Daily
sampling was performed over 13 consecutive days beginning 09/29/86 and
ending 10/12/86. RTI operated two ADSs and two TFRs each day (22-hour
sample duration) and two ADSs each week (7- and 5-day sample durations)
of the study.
The ADS is described in Reference 1. In this device, sampled air
passes in series at 14 SLPM through: a cyclone to remove large parti-
cles (Dijq = 3.3 ym); two Na2C0-j coated annular denuders to collect
HNOj, HNO2, and SOj; one citric acid coated annular denuder to collect
gaseous NH3; one Teflon filter to collect fine particles; and one
nylon filter to collect volatilized nitrate.
The TFR is described in Reference 2. Air is sampled at 33 SLPM
through a cyclone of the same design employed in the ADS (Djq "1.8
•jra); the sample is then split into two identical trains. A 1 SLPM
bleedstream is drawn from the cyclone base to prevent particle accumu-
lation. This stream passes through a Na2C03 coated filter to
permit resolution of nitrates and sulfates. Each 16 SLPM stream passes
through a tube lined with a 3.2 cm long strip of nylon to collect 8.5
percent of the gaseous nitric acid and a similar strip of Nafion to
collect 17 percent of the gaseous NH^. The sample then passes through
a Teflon filter to collect fine particles, a nylon filter to collect
the remaining HNO3 and volatile nitrate, and an oxalic acid coated
filter to collect the remaining NH3 and volatile ammonium. At this
point a 1.6 SLPM stream is drawn through two triethanolamine (TEA)
coated filters to collect SO2 and NOj. This 1.6 SLPM stream and the
remaining 14.4 SLPM stream then pass through mass flow controllers and
are exhausted through a pump.
380
-------
The samples collected in the ADS and TFR are extracted into solu-
tion where the recovered nitrate, nitrite, and sulfate are determined
by ion chromatography (IC). Ammonium is determined by a colorimetric
method, indophenol blue. Hydrogen ion concentration is determined as
pH. The measured mass of each species is used along with the sampled
volume of air to infer the airborne concentration.
Results
Concentration and Precision Results
Concentration and precision results for the ADS and TFR samplers
are summarized in Table 1. The tabulated parameters include: total
gaseous and particulate ammonia and ammoniun (Total NH^+NH^); fine
particle ammonium (Fine NH^); ammonia (NH3); hydrogen ion collected on
Che Teflon filter (H+ Teflon); total gaseous nitric acid and
particulate nitrate (Total NO3); fine particle nitrate (Fine NO3);
nitric acid (HNO3); nitrous acid (HNOj); nitrogen dioxide (NOj); total
particulate sulfate (Total So|""); fine particle sulfate (Fine S0^~);
and sulfur dioxide (SOj). The term "total" indicates that the cyclone
catch for that species is included. The term "fine" refers to parti-
cles passing the cyclone and entering the Teflon filter in the particu-
late phase. Studywide estimates of mean and standard deviations (SD)
of each parameter are given for each sampler. It should be noted that
tabulated SDs represent day-to-day variability. Within day median
coefficients of variation are also given for the two samplers.
The TFR Fine NO3 results are negative in 24 of 26 cases. As a
result, they do not permit meaningful comparisons and are not consid-
ered in subsequent data analyses.
In general, both the ADS and TFR show good precision for all the
remaining parameters. In all cases, the median difference between the
paired ADS daily samples is less than 0.2 yg/m , and with two excep-
tions, the corresponding results for the TFR+are less than 0.5 yg/m .
With only two exceptions, TFR HNO3 and TFR H Teflon, the median CV's
are less than 20 percent. The median CV's for the Total NH^+NH^, Total
NOj, and Total So|" are all less than 5 percent. This shows excellent
within sampler precision for both the ADS and TFR.
Since the colorimetric ammoniira analysis was more variable than
the IC analyses, the relatively high median CV's for M3 in each +
sampler may be reflecting the large analytical variability for NH^ at
the low concentrations associated with ambient NH3.
The pH analysis of Teflon filter extracts resulted in appreciable
variability of the H+ Teflon in both samplers. This is not surpris-
ing when it is noted that a difference in+pH^of 0.25 at a pH of 5.00
can propagate a difference of 7.8 n mol H /m .
381
-------
Sample Stability
Weekly average concentrations were computed using results from the
appropriate daily ADS samples. These results were then ratioed to the
results of the appropriate weekly samples. Ratios near unity suggest
sample stability across sampling durations of 22 hours to 7 days.
Most of the computed to measured results are near unity. The
ratio for HNOj is high and thought to be so because collected nitrite
is oxidized to nitrate to a larger extent in the ADS over the 1-week
sampling duration than over 1-day periods. The low ratio for HNO3
may also be reflecting this enhanced conversion of nitrite to nitrate
in the weekly samples.
Statistical Analysis of ADS and TFR Results
Correlation coefficients from linear regression analysis of ADS
and TFR daily mean values for the 13-day study are given in Table 2.
The average ratios of daily mean ADS to TFR results for each parameter
are given in Table 2. Differences between the daily mean ADS and TFR
values were used in t tests to determine if the observed differences
between the results from the two samplers are significant at the 95
percent confidence level. Results of these tests are also given in
Table 2,
Behavior of Stable Species. Results of the ADS and TFR are
highly correlated in most cases. In six of seven cases,' correlation
coefficients (r) exceed 0.85. For the two stable species, total
particulate sulfate and total gaseous nitric acid and particulate
nitrate, the results from the two samplers are highly correlated, with
average ratios of concentrations near unity. There is no apparent
difference between the results from the two samplers. This provides
strong evidence that both samplers were operating under known flow
conditions. It also suggests that across-sampler differences in the
partitioning of portions of the Total NO^ or Total SO^ must depend on
unique characteristics of the samplers employed.
Total WO3. As noted above, the measures of total gaseous
nitric acid and particulate nitrate from the ADS and TFR are in agree-
ment. They are highly correlated (i.e., r ¦ 0,96). The average of the
ADS to TFR ratios is 1.00+0.08, and the t test does not indicate a
significant difference between the two measures.
HNO3. The ADS and TFR measures of HNO3 are well correlated
(i.e., r ¦ 0.88). The average of the ADS to TFR ratios is 0.54+0.12,
and the t test indicates that the difference between the measures is
significant at the 95 percent confidence level. The sum of the nitrate
on the TFR nylon strip and filter was used to obtain upper bound HNO3
382
-------
estimates which were also compared to ADS measures of HNO3. Although
not shown in Table 2, these parameters are highly correlated (i.e., r -
0.98), the average of the ADS to TFR ratios is 0.88+0.08, and the two
measures are again significantly different at the 95 percent confidence
level. These results show better agreement between TFR and ADS
measures of HNO3 when the TFR HNO3 is computed under the assumption
that the TFR functions similarly to a filter pacH rather than according
to TFR theory.
Total SO4T As noted previously, the measures of total particu-
late sulfate from the ADS and TFR are in agreement. They are highly
correlated (i.e., r - 1.00), the average of the ADS to TFR ratios is
1.03 +0.04, and the t teat does not indicate a significant difference
between the measures from the two samplers.
SO?, The ADS and TFR measures of S02 ate also highly corre-
lated'TT.e., r - 0.98). The average of the ADS to TFR ratios is 1.32
+0,30 (reciprocal - 0.76), and the t test indicates that the difference
Ts significant. Tests of the recovery of S02 from TEA-coated filters
conducted in our laboratory hive shown approximately 80 percent recov-
ery under humid conditions. The observed discrepancy between the ADS
and TFR SO2 concentrations may be related to reduced sulfate recovery
from the TEA-coated filters in the determination of SO2 with the TFR.
Total NH3+NH4. Total gaseous and particulate ammonium as ^
measured by the ADS and TFR are highly correlated (i.e., r - 0-^5).
Although the average of the ADS to TFR ratios is 0.86+0.13, the t test
does indicate that the results of the two samplers are significantly
different. A portion of this difference may be attributed to under-
sampling of particulate ammonium by the ADS #ince there was no provi-
sion in this sampler to collect any volatilised ammonium downstream of
the Teflon filter. The discrepancy may also be associated with the
previously noted variability in the ammonium determinations.
NH3. The ADS and TFR measures of ammonia are poorly correlated
(i.e., r " 0.49), and the average of the ADS to TFR ratios is
+0.41, The variability of the results prevents detection of a differ-
ence between the ADS and TFR measures of this parameter.
H+ Teflon. Strong hydrogen ion is determined from the measured
pH of the extract of Teflon filters in both the ADS and TFR. ADS and
TFR measures of K+ are well correlated (i.e., r " 0.86), and the
average ratio is 1.59+0.66. In spite of the large variability, the t
test indicates that the results of the two samplers are different at
the 95 percent confidence level. The ADS H concentration may exceed
the corresponding TFR value in part because undenuded ammonia in the
TFR neutralizes a portion of the H+ in that sampler.
383
-------
Conclusions
Conclusions and findings drawn from this study are given below.
o Both ADS and TFR show good precision for all measured species
concentrations except fine particle NO^ (Fine NO3).
o Foj_the two stable species, total particulate sulfate (Total
SO^ ) and total gaseous and particulate nitrate (Total NO3),
the results from the ADS and TFR are highly correlated, have
average ratios near unity, and are not significantly different.
These findings indicate that both samplers were operating under
known flow conditions.
o Statistical comparisons of ADS and TFR. results show no Bignifi-
cant differences for Total NO3, Total S0^~, and NH^; the ADS
estimate to exceed that of the TFR for SO2 and H+; and the TFR
estimate to exceed that of the ADS for HNO-j and Total NHj+NH^.
o Weekly average concentrations computed from measured daily
samples were ratioed to measured weekly results to assess
sample stability in the ADS. The ratio was near unity for most
species. High ratios for HNOj may be reflecting enhanced NO2
oxidation on the denuders over 1-week versus 1-day sampling
periods. The conversion of nitrites to nitrates over extended
sampling periods (i.e., treekly periods) may result in an over-
estimate of ambient HNO3 concentrations and an underestimate
of ambient HNOj concentrations with the ADS.
o R+ concentrations measured with the ADS exceed those with the
TFR because undenuded NH3 in the air sample reaching the TFR
Teflon filter neutralizes a portion of the H+ collected on
the Teflon filter with that sampler.
o SO2 concentrations measured with the ADS exceed those
measured with the TFR by approximately 10 percent. Tests in
our laboratory suggest that the TFR results require a correc-
tion for reduced S0^~ recovery efficiency (i.e., 80 percent)
under humid conditions.
References
1. M. Possanzini, A. Febo, and A. Liberti. "New Design of a High-
Perforraance Denuder for the Sampling of Atmospheric Pollutants,"
Atmos. Environ, 2605-2610 (1983),
2. K. T. Knapp, J. L. Durham, and T. G. Ellestad. "Pollutant Sampler
for Measurements of Atmospheric Dry Deposition," Environ. Sci.
Technol, 20:633-637 (1986).
384
-------
TABLE 1. SUMMARY OF CONCENTRATION AND PRECISION RESULTS FOR ADS AND TFR SAMPLERS
Daily (9/29/86 to 10/12/86)
Mean
+ SDa
Median
, CV,%
ADS
TFR
ADS
TFR
Total NH3+NH£
2. 545_+l. 153
2.911+1.191
3.3
4.6
Fine (as NH4)
1.794+1.055
2.174+1.091
3.2
9.1
nh3
0.717+0.246
0.632+0.176
10.0
13.5
H+ Teflon
29.6U15.31C
22.82+16.74°
11.2
21.8
Total NO3
3.Q86_+0.586
3.141+0.739
4.2
1.3
Fine NO3
0.602+0.559
-0.728j+0.688d
15.7
46.6d
HNO3
-------
TABLE 2. STATISTICAL COMPARISON OF ADS AND TFR RESULTS
a
r
Average Ratio
of ADS to TFRb
t Test
c
Results
Total
NO3
0.96
1.00
+
0.08
no difference
Hno3
0.88
0.54
+
0.12
ADS < TFR
Total
soj"
1.00
1.03
+
0
0
no difference
so2
0.98
1.32
+
0.30
ADS > TFR
Total
NH3 + NH^
0.95
0.86
+
0.13
ADS < TFR
NH3
0.49
1.17
+
0.41
no difference
H+ Teflon
0.86
1.59
+
0.66
ADS > TFR
Correlation coefficient for linear regression of mean daily ADS on TFR
results, n » 13.
^Average of ratio of daily mean results for ADS and TFR one standard
deviation).
cResults of t test for 13 paired results at 95 percent confidence
level.
386
-------
Passive Sampling Device Measurements of NO2 in Ambient Air
James D. Mu11k
Environmental Monitoring Systems Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
Dennis E. Williams
Northrop Services, Inc.
Research Triangle Park, NC 27709
Abstract
Passive sampling devices (PSDs) using triethanolamlne-coated glass fiber
filters as the collection surface and stainless steel mesh as a diffusion
barrier between the afr and the collection surface were deployed as part of the
methods comparison study. The PSDs do not require a pump, and are small (2
diameter cylindrical sections), inexpensive sampling devices. The sampling
rate for the devices with both surfaces of the device collecting has been
established as approximately 154 cc/min under typical ambient conditions.
Three PSDs were placed 1n the ambient air and werfe therefore subject to
any sampling rate variations due to wind, humidity and temperature changes. An
additional three devices were placed Inside the local dry deposition shelter
within a glass cylinder through which ambient air was drawn at a controlled
rate. The outside and inside PSDs averaged 8.84 ppbv and 9.92 ppbv, respective-
ly, and a significant degree of variability was noted among triplicates.
Introduction
Several years ago the EPA began a modest program to develop passive
sampling devices (PSDs) for the collection and analysis of both organic and
Inorganic air pollutants. More recently, however, the EPA has become more
Interested In personal exposure monitoring to support the ambient air quality
387
-------
standards set under the Clean Air Act. Passive sampling devices are ideally
suited for personal exposure monitoring because of their relatively low cost
and small size. Commercially available NO2 passive sampling devices, such as
the Palmes tube (1), lack the sensitivity needed to obtain 8 to 24 hr time
weighted average (TWA) measurements in non-occupational air. The sampling rate
for the Palmes tube is -1.0 cm^/min, which affords a sensitivity of 300
ppbv-hr when analyzed spectrophotometrically. Consequently, 5- to 7-day expo-
sures are required to determine much lower levels of NO2 found in ambient and
indoor air.
PSDs have been used extensively by industrial hygienists to assess the
effects of respiratory exposure to hazardous pollutants of workers. Since
ambient air levels of most pollutants are several orders of magnitude lower
than those normally found in the workplace, more sensitive monitoring systems
are required. These systems are generally not portable and are usually located
at fixed, outdoor sites. Such fixed-site monitoring may not accurately reflect
the average daily exposure of the general populace to air pollutants. This is
particularly true in the United States, where the average person spends an
estimated 90% of the time indoors (2).
To obtain an accurate estimate of individual exposure to air pollution,
the person must carry or wear the monitor. Therefore, the device must be
unobtrusive, and lightweight. It should operate quietly, and place little or
no burden on the Individual. It must also be inexpensive, since many more
units may be required for personal monitoring than for fixed-station monitor-
ing, Active (pump-based) sampling systems used for occupational exposure
assessment can and have been used successfully outside of the workplace (3).
However, the pumps weigh from 0.5 to 1.5 kg and may not be comfortable to wear,
388
-------
especially for small persons. In addition, they must be battery-powered and
most cannot be continuously operated for more than 6 to 12 hr. Passive devices,
which require no pump, are much lighter in weight and are not power limited.
EPA's initial efforts in developing passive sampling devices resulted in
several publications and EPA final reports (4-8). Four N02 passive sampling
methods were studied and were described in detail at the 1986 EPA-APCA Meeting.
Two of the four methods that showed the most promise were:
1. AG/AGO coated on Chromosorb P followed by thermal desorption into a
chemiluminescent N0X monitor.
2. Modified EPA-PSD using triethanolamine coated on glass fiber filter paper
as the sorbent, followed by extraction and ion chromatographic analysis.
Experimental
Further evaluation of the AG/AGO chemiluminescent N02 method showed that
water vapor was a serious Interferent. At relative humidities higher than 30%
the overall sampling and recovery efficiency declines sharply and approaches
30% recovery at 80% relative humidity. The Interference by water vapor would
prevent the use of the sorbent 1n a passive sampling device, but when used in
an active sampling device, this effect could be avoided by removing the water
vapor ahead of the sorbent tube by means of a Naflon dryer. Because of the
water vapor problem with AG/AGO in the passive mode, this work was dropped to
focus on the modified EPA-PSD with Palmes tube chemistry and 1on chromatographic
analysi s.
The EPA PSD that was originally designed for volatile organic compounds
was converted to a PSD for NO2 simply by replacing the Tenax* sorbent with
triethanolamine-coated glass fiber filter paper. Ion chromatography was em-
389
-------
ployed for analysis. The effective samping rate of the modified EPA-PSD is 154
cm3/min making it potentially 150 times more sensitive than the Palmes tube.
The PSD has recently become commercially available from Scientific
Instrumentation Specialists, Inc. (SIS) of Moscow, Idaho. An exploded view of
the device 1s shown in Figure 1. The PSD is 3.8 cm 1n diameter, 1.2 cm in
depth and weighs 36 g. Pairs of 200-mesh wire screens and perforated plates
placed on each side of the sorbent bed serve as diffusion barriers.
The SIS-PSD for NO? was shown to have a minimum detectable quantity of 30
ppb-hr; to show no interference from NO and relative humidities as high as 80%;
and to respond linearly from 20 ug/m3 to 460 yg/m3 (9).
All passive sampling devices have a minimum face velocity above which
they must be used to maintain a constant sampling rate. The curve shown in
Figure 2 1s a plot of the response of the SIS-PSD to NO2 versus face velocity
(cm/sec.). From a practical standpoint, it appears that the SIS-PSO for NO2
samples at reasonably constant rates at face velocities above 30 cm/sec.
The SIS-PSDs for NO2 were tested 1n ambient air by direct comparison with
a tunable diode laser system (10) during a field study 1n the EPA Annex parking
lot at the Research Triangle Park, NC 1n October 1986, The PSDs were exposed
outdoors In triplicate at ambient conditions of wind velocity, temperature and
humidity and were located near the tunable diode laser sampling line. They
were exposed for 22 hours each day of the 13-day study. Table I shows the
excellent comparability between the two methods. In addition to the direct
outdoor exposures, another set of 3 PSDs was simultaneously exposed inside a
chamber through which outside air was passed at a constant known face velocity
(100 cm/sec). Wh1Je these measurements were generally slightly higher than
those made outside the chamber, they agreed within experimental error.
390
-------
Conclusions
Prior to this work, commercially available passive sampling devices for
NO2 could not be used as personal exposure monitors because they lacked the
sensitivity needed for concentrations found 1n non-industrial settings, such as
indoor air and ambient air. The results presented in this paper demonstrate
that the SIS-PSD can be successfully applied to 8 to 24 hr time weighted average
(TWA) measurements of N02 in ambient air. Although there was some variability
between the 22 hr daily averages of the PSDs and the tunable diode User daily
average, the 13 day average agreed within a part per billion.
It should be noted that the PSDs were not meant to provide data that is
accurate to plus or minus a few percent of the real value. They were designed
to provide semi-quantitative data inexpensively and to be user friendly
so that they could be used as personal exposure monitors to obtain the data
that health effects researchers and statisticians need for their models.
The agreement between the two methods is even more remarkable when
considering the complexity and high cost of the tunable diode laser (-0.2
million) and the simplicity and low cost of the SIS-PSD (less than $200). The
cost of the SIS-PSD for NO2 could be reduced drastically by making the device
out of a moldable plastic material. This passive sampler must be further
evaluated for potential Interferences from nitrous add (HONO) and peroxy-
acetyl nitrate (PAN). HONO 1s not expected to be a significant interferent
because of recent work by Blermann et al.
-------
2. PAN would probably not get through the stainless steel diffusion barrier.
3. The 1on chromatographic technique separates nitrate from nitrite.
The SI5-PSD described herein provides a personal exposure monitor that has
the sensitivity and selectivity to monitor NO2 "in non-occupational environments,
thus making it possible to perform large-scale monitoring at a relatively low
cost. It is envisioned that such devices can be tailor-made for other pollu-
tants, such as formaldehyde, ethylene oxide, ozone, etc.
392
-------
REFERENCES
1« Palmes, E. 0., Gunnison, A. F., DiMattfo, J., and Tomczyn, C,: Persona)
Sampler for Nitrogen Dioxide. Amer. Ind. Hyg. Assoc. JJ7:510 (1976),
2. Budiansky, S.: Indoor Air Pollution. Environ. Sci. Tech. 14:1023 (1980).
3. Wallace, L.A. and W. R. Ott: Personal Monitors: A State-of-the-Art Survey.
J. Air Pollution Control Assoc. 32:601 (1982).
4. R.W. Coutant and D.R. Scott: Applicability of Passive Dosimeters for
Ambient Air Monitoring of Toxic Organic Compounds. Environ. Sc1. Tech.
16:410 (1982).
5. Lewis, R.G., R.W. Coutant, G.W. Wooten, C.R. McMilUn and J.D. Mulik:
Applicability of Passive Monitoring Device to Measurement of Volatile
Organic Chemicals in Ambient Air, 1963 Spring National Meeting, American
Institute of Chemical Engineers (March 1983).
6. Lewis, R.G., J.D. Mul 1 k, R.W. Coutant, G.W. Wooten and C.R. McM1ll1n:
Thermally Resorbable Passive Sampling Device for Volatile Organic Chemi-
cals in Air. Ar>a1. Chem. 57:214 (1985).
7. Coutant, R.W., R.G. Lewis and J.D. MuHlc: Passive Sampling Devices with
Reversible Adsorption. Anal Chem. _57^:219-223 (1985).
S. Miller, M.P.: Analysis of Nitrite In NO? Diffusion Tubes Using Ion Chroma*
tography. U,S, Environmental Protection Agency Report No. 600/2-87-000,
Research Triangle Park, fiC, 1n press (1987).
9. Mulik, J.D. and D.E. Williams: Passive Sampling Devices for m2, pp, 61-
70, Proceedings of the 1986 EPA/APCA Symposium on Measurement of Toxic
A1r Pollutants, Air Pollution Control Association Publication VIP-7, U.S.
Environmental Protection Agency Report 600/9-86-013, Research Triangle
Park, NO (1986).
393
-------
10. Schiff, H.I., D.C, Hastie, G.I. Mackay, T. luguchi and B.A. Ridley: Tunable
Hiode Laser Systems for Measuring Trace Gases in Troposphen'i Air.
Environ, Sci. Techno!. 17^:352 (1980).
11. Bfermann, H. W., Pitts, J. N., Winer, A. M.: Simultaneous Spectroscopyc
Determination of Gaseous Nitrous Acid and Nitrogen Dioxide in Polluted
Indoor and Outdoor Environments. Proceedings of the American Conference
of Governmental Industrial Hygienists, Inc., Symposium on Advances in Air
Sampling, in press (1987).
394
-------
TABLE I. Comparison of SIS-PSD with a tunable diode laser for NO2 in ambient air.
Modified PSD
Outside Inside Chamber^ Tunable Diode Laser
Pay Cone, PPB Uncertainty Cone, PPB Uncertainty Cone, PPB Uncertalnt.y
1
11.1 ~
1.3
12.4
3.8
10.2
+ 1.5
2
9.0 +
1.1
9.3
2.9
6.0
+ 0.9
3
10.9 +
1.3
11.2
3.5
7.8
+ 1.2
4
9.6 +
1.2
11.2
3.5
11.3
+ 1.7
5
9.3 +
1.1
8.3
2.6
6.2
+ 0.9
6
6.2 +
0.7
5.9
1.8
4.5
+ 0.7
7
9.1 +
1.1
8.4
2.6
8.4
1 1.3
8
10.3 +
1.2
10.3
3.2
9.2
+ 1.4
9
12.1 +
1.4
14.5
4.5
13.4
+ 2.0
10
7.3 +
0.9
11.1
3.4
12.0
± 1.8
U
7.1 +
0.9
10.6
3.3
12.2
1 1-8
12
9.4 +
1.1
9'9
3.1
7.8
+ 1.2
13
3.6 +
0.4
5.1
1.6
4.7
+ 0.7
Avg
8.8 +
1.1
9.9
2.8
8.8
+ 1.3
A
Ambient outside air passed through chambe
at 100 cm/sec.
395
-------
PERFORATED PLATE
SORBENT
CONTAINMENT
SPACE
INTERNAL 'C' CLIP
200 MESH
DIFFUSION SCREEN
BODY,
ALL MATERIAL STAINLESS STEEL
Figure 1 -- SIS Passive Sampling Device
396
-------
130
120
110
100
0>
> 90
0) 80
sS 70
0s
60
a>
(A
c
Q, 40
(/)
a> 30
0L
20
50
I 1 1 1
-1 1 -
£
l i i 1 1 1 1 1 1
W
-— 9 _—
1 1 1 1
1 f
| 1 1 1 1 1 J J !
Face Velocity (cm/sec)
Figure 2 -- NC>2-P$D:Response vs. Face Velocity
397
-------
LUMINOX MEASUREMENTS OF AMBIENT MO2
D.K. Bubacz, E.H. Daughtrey, J.D, Pleil, and K.G. Kronrailler
Northrop Services, Inc. - Environmental Sciences
Research Triangle Park, NC 27709
Abstract
Two Luminox LMA-3 monitors were run simultaneously to measure
real-time fluctuations of ambient N02- This was done as a part of the
Fall 1986 Methods Comparison Study for measurement of compounds
related to acidic deposition, conducted at the U.S. EPA Environmental
Research Center Annex, Research Triangle Park, NC, from September 29
through October 13, 1986, Although they were purchased from the same
manufacturer (Scintrex, Ltd., Ontario, Canada), the two instruments
employed different temperature-dependence correction methods. One
used temperature-sensing electronics to compensate the output signal;
the second used feedback-controlled heating to maintain the reaction
chamber of the instrument at a fixed temperature.
Data from both LMA-3 instruments were collected in real time by a
Keithly-Compaq data acquisition system, which stored and printed
hourly and daily averages. In this paper we describe our system and
experimental methods, and we use our test results to compare the two
monitors with one another and with the NO2 reference system, a diode
laser operated by another group.
398
-------
Introduction
Between August 29 and September 12, 1986, as part of the methods
comparison study held at the U.S. Environmental Protection Agency
(EPA) Annex in Research Triangle Park, NC, two Luminox LMA3 N02
monitors (Scintrex, Ltd; Ontario, Canada) measured real-time N02 gas
concentrations in ambient air. The two luminol-based instruments
simultaneously sampled outside ambient air in a performance study of
instruments equipped with different temperature compensation devices.
The modifications were implemented to remedy a substantial
temperature-sensitive response problem observed in earlier model
instruments. 1 One LMA3 instrument (HLMA3) was equipped with a
temperature-controlled (30°C) reaction cell, whereas the other
(#2LMA3) contained signal-correction circuitry to compensate for
ambient temperature fluctuations. The data sets from the two
instruments were compared with each other and with the tunable diode
laser (TDL), the instrument used as the reference method for NO2
measurement.
Experimental Methods
Ambient NO2 measurements were made from inside a mobile laboratory
situated next to the dry acid-deposition station in the rear parking
lot of the EPA Annex in Research Triangle Park. The two Luminox LMA3
instruments sampled ambient air through separate 0.25-in.-i.d. Teflon
lines extending from each instrument to a Teflon tee. From th« tee a
common sample line extended to the top of the dry-deposition station
to a vertical height of approximately 10 ft. At the inlet of each
Luminox instrument, t7-mm Teflon filters were installed to prevent
fouling by dust and dirt. Sample air was drawn into each instrument
by a small d.c.-driven, flying-vane-type air pump operating at a
nominal flow rate of 1.5 L/min. Luminol solution was moved into and
out of the reaction cell of each instrument by means of a peristaltic
pump operating at approximately 0.05 mL/min.
Real-time ambient monitoring was maintained continuously for
22 h/day, throughout the study. The 2-h downtime each day was used
for equipment maintenance, data transferal, and instrument
calibration. Real-time data were collected and stored on a Keithly-
Compaq (Keithly Instruments, Inc.; Cleveland, OH) data acquisition
system. The system calculated and stored hourly averages, which were
then manually reduced to daily, 22-h averages.
Multipoint calibration curves generated both before and after the
comparison study were used to arrive at the final average dally
concentrations of ambient NO2.
Results and Discussion
Ambient NO2 values versus time for the two Luminox instruments and
the TDL are plotted in Figure 1. A qualitative comparison of data
from the two Luminox LMA3 instruments indicates that the instruments
tracked each other on a day-to-day basis, showing good agreement on
increasing and decreasing trends. Similarly, good agreement is
evident in the comparison of the Luminox instruments with the TDL with
the exception of day 7 to day 8, when the Luminox instruments both
399
-------
Indicated a decrease while the TDL showed an increase in the NO2 daily
averages. On the eleventh day of the study, the #2LMA3 instrument
showed a substantially higher average daily value than either of the
other two instruments.
Quantitative comparisons are shown in Figures 2 and 3. Figure 2
is a comparison of NO2 daily averages of Luminox instruments #1LMA3
and #2LMA3. A 1:1 correlation line drawn on the plot shows that
#2LMA3 measured higher daily averages than #1LMA3 nearly two-thirds of
the time. Two daily averages from the Luminox monitors show agreement
to within 5% of each other.
Figure 3 compares the daily averages of the #1LMA3 instrument to
those of the TDL instrument. Results indicate excellent agreement
between the two instruments for approximately three-fourths of the
study. In Figure 4, data from the #2LMA3 instrument and the TDL
instrument are compared. Although the #2LMA3 instrument yielded daily
NO2 averages that were consistently higher than averages calculated
for the reference method (TDL), the results roughly parallel those of
the TDL, indicating a possible calibration offset in the Luminox
instrument.
An overall comparison of the two temperature-corrected Lurainox
instruments from this preliminary investigation was made in terms of
agreement with the reference method and in terms of versatility in
field operation. Both Luminox instruments approximated ambient NO2
concentration to a degree comparable to the TDL's performance. The
#1LMA3 instrument showed an overall average concentration of 8.98 ppbv
NO2 while the #2LMA3 instrument showed 10.2 ppb NO2. These values
agreed within 9951 and 86^ of the value obtained with the TDL,
respectively.
The advantages and disadvantages of each temperature modification
design should be weighed before a design is selected for field use.
An advantage of the #1LMA3 design is that the reaction cell
temperature remains constant, thereby reducing interference of ambient
temperature fluctuations with reaction kinetics. On the other hand
the #2LMA3 instrument design incorporates signal-processing
electronics, which may introduce calibration-shift problems.
Portability must be considered for operation of instruments in the
field. When sampling requirements do not exceed 2-3 h, the Luminox
instruments may be operated on a single battery charge. In this
respect the thermostatic control of the #1LMA3 instrument is limited
by its 110-V a.c. power requirement, whereas the #2LMA3 instrument is
capable of independent operation up to 2-3 h. However, the advantage
of independent operation becomes negligible when one considers that
the data recording devices require an external power source to
operate. In addition, most field monitoring studies far exceed the
maximum 2- to 3-h operating time of a single battery charge.
Based on the data from the study, both temperature-correction
modifications show comparable results to those of the reference
method, TDL, with the exception of a slight calibration offset In
#2LMA3. Thus, the temperature-correction design employed in the field
400
-------
should be chosen on the basis of the suitability of Its design for the
conditions under which the instrument is to be operated.
References
1. D.K. Bubacz, J.D. Pleil, "Real-time Measurement of Ambient NOg:
Calibration System Design and Preliminary Evaluation of the
Luralnox LMA-3 NO2 Monitor." Technical Note M2Q-86-01,
Northrop Services, Inc., Research Triangle Park, North
Carolina, 1986.
401
-------
20
18
16
14
12
CONC io
(ppbv)
FALL METHODS COMPARISON STUDY
##1LMA3
-#21MA3
J.TDL
J L
J L
_L
J L
6 7
DAY
J L
10 11 12 13
Figure 1. Plot of daily (J2-h) NO] averages measured with luminoa NOj monitors and the tunable diode
laser, Aug. 2S-Sept. 12, 1986.
#1LMA3 vs. #2LMA3
#1LMA3
(ppbv)
0 2 4 6 8 10 12 14 16 18 20
#2LMA3
(ppbv)
Figure I. Comparison plot of #1LMA3 vs. #JLMAJ dally NOj avenger
402
-------
#1LMA3 vs. TDL
#1LMA3
(ppbv)
B 10 12 14 16 18 20
TDL (ppbv)
Figur* 3. Companion plot of #1LMAI Luroino* monitor vi. tunable diode later dally
NO; averager
#2LMA3 vs. TDL
#2LMA3
(ppbv)
0 2 4 6 8 10 12 14 16 18 20
TDl (ppbv)
Figure *. Companion plot of 41LMAI Lumtno* monitor vs. tunable diode later dally
NO) average!.
403
-------
Comparison Across Different Methods and Interpretation of Results
Richard J. Paur and William A. McClenny
Environmental Monitoring Systems Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Abstract
A thirteen day comparison study of samplers/methods for estimating
concentration of various sulfur and nitrogen containing species was conducted
in Sept/Oct 1986 in Research Triangle Park, NC, The primary objective of the
comparison was to determine whether results obtained from multicomponent
samplers, each representing a candidate method for use in acid deposition
networks, were equivalent, and to further determine whether those results were
in agreement with an independent spectroscopic method for gaseous nitric acid
and nitrogen dioxide. This paper presents selected results from the study and
compares the performance of the various methods.
Introduction
The Environmental Monitoring Systems Laboratory (EMSL) of the US EPA
has been conducting a methods development program for measurement of the atmos-
pheric concentration of the chemical species that are important in acid deposi-
tion monitoring. These species include nitrogen and sulfur containing species
such as SOg, NO2, NO3", SOj", HNOo, NH3, and NH^, and other contributing species
such as ozone and HgOg. The acidity of particulate matter collected on Teflon
filters can be estimated by measurement of the pH of aqueous extracts of the
filters; however, the significance of the measurement is unclear due to problems
1n maintaining the integrity of the sample.
One of the methods examined in the study, the annular denuder system
(ADS), collects acidic species on a carbonate coated denuder assembly. One of
the materials observed during ion chromatographic (IC) analysis of the denuder
extract is NO^" ion; the NOp" ion is present in concentration greater that
that expected for partial collection of NO2, NO or PAN. The source of the
excess nitrite has been tentatively identified as HONO. The ambient concentra-
tions of HONO, HNO3 and NO2 1s compared to provide some insight Into the poten-
tial role of HONO as a nitrogen donor to biological systems and to examine the
plausibility of the identification of the species as HONO by Its correlation
with the concentration of NOg, one of the precursors of HONO in the troposphere.
Many of the results presented here are presented in more detail in the
previous papers given in this session. The intent of this paper is to provide
comparisons between similar measurements from different methods.
The data obtained in the study was reduced to 22 hour average values for
each pollutant and sampler. Replicate values were averaged together to permit
404
-------
statistical analysis of the data in terms of linear regression treatments of 5
or 13 22-hr values from Individual samplers, or, in the case of nitric acid,
from a composite average of 4 different samplers. The uncertainties presented
are those associated with the least squares linear regression and do not include
the uncertainties associated with the chemical analysis or flow rates (see the
preceding papers for estimates of these uncertainties). The error bounds are
calculated for 95% confidence intervals.
In the following discussion the methods for sampling and/or the corres-
ponding measurement results after sampling will be referred to as:
ADS -- annular denuder system
TFR — transition flow reactor
FP « filter pack
-F suffix — denotes measurement results for particle species collected
downstream of an inlet.
»T suffix "¦ denotes sum of jneflsurernent results for psrtlcle species
collected on inlet and downstream of inlet
TFR-RTI — TFR results obtained by Research Triangle Institute person-
nel
TFR-ASRL ~ TFR results obtained by Atmospheric Sciences Research Labor-
atory personnel
TFR-RTI-FP ~ TFR results Interpreted as a filter pack instead of as
recommended by ori9^ator'$ standard operating procedure
TOLAS - tunable diode laser spectrometer measurements
COMP -- composite average of measurements from AOS, FP, TFR-ASRL
and TFR-RTI-FP
Results and Discussion
Sulfate
The sulfate results presented 1n Table 1 locate generally close
agreement between the ADS and TFR samplers for both fine ^action and total
particulate sulfate. For the TFR-ASRL subset of five sampling n'nsj Jh®
was not extracted, so only TFR-ASRL-F can be reported. The FP r®sult^I,*®™
10-15% greater than ADS or TFR results 1n terms of the regression line slope,
this difference 1s probably not significant when flow and are
considered. The Implication of such good agreement for sul fate ts is
that flow rates were accurately known and that any disagreement in results for
other species must be due to causes other than uncertainty in flow rates.
Total sulfate values are evenly distributed over the concentration[range
of 4-14 ug/m3. The difference between fine and total sulfate 1s incremental,
averaging about 5% of the total sulfate; hence, all inlets used were adequate.
Sum of gas phase and particle nitrate
The sum of gas phase and particle phase nitrate obtained by the ADS, TFR-
RTI and FP are In agreement. The relatively large uncertainties associated
«1th the slope and Intercept values are in large part due to the narrow range
of concentrations (2-4 ug/m3) observed during the comparison.
405
-------
Ni t rate
The size range of particles containing nitrate extends from fine parti-
cles to particles well above the cutpoint of the inlets on the AOS and TFR
samplers {3.3 ym and 1.8 ym aerodynamic diameter, respectively). Therefore,
comparisons are complicated and conclusions are less certain than for sulfate.
Table 1 shows that the total particulate nitrate from the ADS and FP are in
agreement. The relatively narrow concentration range observed during the study
is largely responsible for the large uncertainty associated with the slope.
The comparison of ADS-F with FP is not meaningful, since FP results include
large partfcle nitrate and, without exception, are considerably higher than the
corresponding ADS-F values. The comparison between ADS-F and TFR-RT1-F is not
meaningful due to unexplained differences between the nominal and observed
collection efficiencies of the TFR for HNO3 and the fact that fine nitrate is
calculated (not measured) using TFR measurements of HNO3.
The 0.79 ratio between the AOS and TFR-ASRL 1s likely due to the
different cutpoints for the two inlets.
Sulfur dioxide
The sulfur dioxide data shows that the FP, ADS, and TFR-RTI agree well
in terms of slope but that the FP and TFR show an offset of 0.5 and 1.5 wg/m3,
respectively, compared to the ADS. The capture of some of the sulfur dioxide
by nylon filters in the FP and TFR was taken Into account. The offset may be
due to extraction problems, matrix effects with the IC, or blank problems;
however, a cursory examination of the pertinent data does not offer strong sup-
port for any of these hypotheses.
Nitrogen dioxide
Comparison of the TFR-RTI with the tunable diode laser (TDLAS) shows
favorable agreement. The relatively greater estimates provided by the TEA
treated filter 1n the TFR are somewhat surprising in view of laboratory
studies indicating less than 100% col lection/extraction efficiencies for the
TEA filters.
Nitric Add
Nitric acid measurements from the various methods were compared to
a composite average of measurements from the ADS, FP, TFR-ASRL, and TFR-RT1-FP.
Table 1 shows that 4 of the 5 integrating sampler methods agreed with the
composite average to within 0.1 in terms of slope and with less than 0.1 pg/m3
intercept value. The slope of the TFR-RTI vs the Comp regression was 1.17 and
the intercept value was large. This combination causes the TFR-RTI to average
63% higher than the composite average. The cause of the anomalous behavior of
that sampler 1s not understood, 1n spite of a detailed Investigation of its
performance.
406
-------
Ammonium ion
The TFR and ADS comparison yields unity slope but the data shows a
sigrii ficant Intercept. This may be due to positive artifact in the TFR or loss
of ammonium ion from the Teflon filter in the ADS. Limited (4) tests to
determine the losses in the ADS suggest smaller losses than would be required
to explain the observations. Both mechanisms suggested above may be operating.
Ammonia
The results for ammonia show little correlation between the measurements.
The low concentration of ammonia might be expected to produce measurement
difficulties; however, the coefficient of variation of replicates measurement
with the ADS and TFR would suggest that the measurements should agree better
than was observed.
Nitrogen
An interesting aspect of the use of carbonat,e coated denuders In the ADS
is the observation of nitrite 1on in the denuder*ash "SUPf/r«i-^S« wlJri
may be expected from reaction of NO2 and PAN JtilJrt' The e*rp«
denuder tube is used as a field blank to correct for this artifact. The excess
nitrite left after correction for known interferences has been tentatively
identified as due to collection of MONO.
Figure 1 1s scatter plot showing the iconcentruaaV1Thi^f^rSoNn
from nitric acid vs the concentration of nitrogen available from MONO. For the
sue 2nd sampling period of this comparison th.; amount of nitrogen. av»n*b e
from the two sources is roughly equivalent at about 0,3 pg/m , in view of
the^considerable Interest 1n deter.1n1n9 HNO,
rM SlMTi JKS-fc & soipceVf We
excess nitrite. Th,; on,,^.^5^
(sle Figure 2). The nitrite t'hs
5% of the N02 concentration. This 1s In general »9re^ «re take" over
Ass
Interference
from nitrogen dioxide conversion to nitrite 1n the denuder tubes.
Conclusions
The various methods agree well In terms
ments for nitrate are complicated by Part1c1e 2LntifSori »r*
nitrate volatility, but are in generally 9°°d 552 mISI
taken Into account. The sulfur dioxide measurements suggest some blank and/or
extraction problems still exist; these If
samplers which pass the gas stream through nylon fitters prior to collection of
407
-------
the sulfur dioxide. The limited nitrogen dioxide comparison showed close
agreement between the TOLAS and TFR-RTI.
The nitric acid measurements showed close agreement among the FP, AOS,
TFR-ASRL, and the TFR-RTI when treated as a filter pack. When operated according
to recommended operating procedures, the TFR-RTI gave results which exceeded
the composite average by 63% when averaged over the 13-day comparison period.
For ammonium ion, the FP result are about 30% high relative to the AOS and TFR
in terms of regression line slopes. The TFR has a positive offset of about
0.5 pg/m^ relative to the FP and ADS. The TFR and ADS ammonia results were
in very poor agreement over the concentration range observed.
The excess nitrite observed in the ADS contains nitrogen in concentrations
approximately equal to the nitrogen present in HNO3.
Di scl aimer
The research described in this article 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 endoresement or recom-
mendation for use.
References
1. Biermann, H. W., Pitts, J. N., and Winer, A. M.: Simultaneous Spectroscopic
Determination of Gaseous Nitrous Acid and Nitrogen Dioxide 1n Polluted
Indoor and Outdoor Environments. Proceedings of the American Conference
of Governmental Industrial Hygienists, Inc., Symposium on Advances in Air
Sainpfing, fn press (1987). ~
408
-------
Table 1. Linear Regression Parameters for Comparison of Selected Methods/
Chemical Species
Method n R^ Slope* Intercept
*
Sulfate; range = 4-14 yg/m^
TFR-ASRL-F vs ADS-F 5
.993
1.00
+
0.15
- 0.05
+
1,36
TFR-RTI-F vs ADS-F 13
.993
0.95
7
0.06
0.05
7
0.45
ADS-T vs FP 13
.961
0.86
7
0.12
0.74
+
0.90
TFR-RTI-T vs FP 13
.953
0.84
7
0.12
0.77
7
0.96
Sum of gas phase and particle nitrate; range
= 2-4
wg/m3
FP vs ADS 13
.769
0.88
+
0.32
0.30
+
1.01
TFR-RTI vs ADS 13
.914
1.20
+
0.25
- 0.57
7
0.77
Nitrate; range ¦ 2-4 pg/m3
FP vs ADS-T 13
,769
0.88
+
0.32
0.30
+
1.01
TFR-ASRL vs ADS-F 5
.993
0.79
+
0.12
0.13
7
0.13
Sulfur dioxide; range = 2-14 ug/m3
FP vs ADS 13
.994
0.94
+
0.05
- 0.41
+
0.48
TFR-RTI vs ADS 13
.966
1.02
+
0.13
- 1.48
+
1.27
Nitrogen dioxide; range ¦ 8-30 yg/m^
TOLAS vs TFR-RTI 13
.938
1.11
+
0.19
- 0.68
+
3.58
Nitric acid; range * 0.3-3 wg/m^
FP vs Comp 13
.982
1.02
+
0.09
0.02
+
0.16
ADS vs Comp 13
.976
0.95
7
0.10
- 0.08
+
0.17
TFR-ASRL vs Comp 5
.996
1.08
7
0.13
0.04
+
0.24
TFR-RTI vs Comp 13
.804
1.17
+
0.38
0.77
7
0.68
TFR-RTI-FP vs Comp 13
.988
0.98
+
0.07
0.07
7
0.13
TDLAS vs Comp 12
,731
0.65
7
0.28
0.53
+
0.51
Ammonium ion; range s 0.5-5 ug/m^
TFR-RTI vs ADS 13 .987 0.97 + 0,07 0.44 + 0.15
FP vs ADS 13 .933 1.32 + 0.23 0.02 7 0.48
Ammonia; range * 0.3-1.2 pg/m3
TFR-RTI vs ADS 13 .239 0.35 + 0.41 0.38 + 0.31
95% confidence intervals for slope and intercept.
409
-------
. 1
N - HN03 B.34 + B.14
N - MONO 0.27 ~ B.|4
-J-
N11
(HONO ?)
Figure 1. Nitrogen (pg/m ) from nitric acid and from excess nitrite.
t.i
l.S
i.«
.3
Inapt • -.M J .51
Slop* " .K + -*3
R"I » .MS
Calc'4 for tSX C.I.
¦ 1
Nltre|«n 01 an Ida Cug'H3> - TDL
Figure 2. Plot of excess nitrite vs. nitrogen dioxide.
410
-------
THE BEHAVIOR of OXY-PAH and NITRO-PAH on
ATMOSPHERIC SOOT PARTICLES
Richard Kamens1, Hani Karam1, James Fuloher1,
Zhishi Guo1. jean Perry1# n
Daniel Saucy2/ and Leonard stoclcburger
LDept of ESE, Univ. of North Carolina, Chapel
Sill, North Carolina, 27514;
*Arizona State University, Tempe Arizona,
852877 3US Environmental Protection
Agency, RTP, NC,27709
ABSTRACT
A preliminary investigation of the relative atmospheric /0w
stability of selected oxygenated Polyar?m*^H!?£!£SJ5? £«
PAH) and nitrated poly aromatic hydrocarbons (JJitro PAH) on
airborne soot particles exposed to sunlight was undertaken.
Diesel soot particles were used as a *25, SiMrtie?2
wood soot particles as a source of °scy-pAH. EI*SSl'nf i-
systems were exposed to midday natural sun 9 chambers
4 hours in 25m* outdoor Teflon film environmental chambers.
Aromatic oxygenates such as 9/10~an^r*ce"?^?" * _nn*.
cyclopenta(def)phenanthrone appear to be on wood
particles, while polyaromatic hydrocarbons such as benzo(a)pyrene
and benzo(ghi)perylene decay on fresh wood or Jiesel
particles in sunlight; l-nitropyrene and SecSi
freshly emitted diesel soot particles also exhibit decay under
certain outdoor conditions.
411
-------
THE BEHAVIOR of OXY-PAH and NITRO-PAH on
ATMOSPHERIC SOOT PARTICLES
INTRODUCTION
A significant portion of the extractable organic natter from
airborne soot particles is composed of semi-polar and polar
oxygenated compounds. Currently, however, there is almost no
information on the stability of these compounds in the
atmosphere. In this paper we will describe preliminary outdoor
chamber experiments which address the stability of selected
oxygenated polyaromatic hydrocarbons (Oxy-PAH) and nitrated
polyaromatic hydrocarbons (Nitro-PAH) on airborne combustion soot
particles.
EXPERIMENTAL
Chamber Experiments
Wood soot particles were added to two 25m3 outdoor teflon
film chambers in the manner which has been described in previous
documents 11~3'* Diesel soot particles were added to the chambers
directly from the tail pipe of a 1967 Mercedes 200D sedan.
Injection occurred for approximately 6 seconds, after warm engine
startup, with a slight load on the engine. Diesel soot particles
(purchased from the National Bureau of Standards, SRM #1650) also
were injected into the chambers using a liquid nitroqen injection
system, previously described by Saucy and co-workers**'. The
resulting aerosol systems, ranging in concentration from 400 to
2500 ug/m3, were aged for periods of 3 to 4 hours in mid-day sun.
Sample Workup Procedures
Particles were collected at various times during the
experiments on 47 mm Pallflex Teflon-impregnated glass fiber
filters. In selected experiments, an XAD-2 cartridge (4 x 0,9 cm
id SS tube) was placed downstream of the filter to assess the
collection distribution of selected oxygenated PAH between the
particle filter and the XAD trap. The sample flow rate was 18
1/rain.
Filters were extracted with 2 5 ml of dichloromethane (MeC^)
in raicro soxhlet extractors. Material collected on XAD traps was
removed by passing 10 ml of Meci2 through the XAD cartridge.
Extracts were concentrated to approximately 200 ul and
fractionated on a Waters semiprep Poracil column with a mobile
phase flow rate of 2 ml/min ' ' . The solvent program started with
95% hexane, 5% MeCl,. At 12 minutes it was programmed to 100%
MeCl, at 5% min , held at 100% MeCl2 for 1 minute, and then
quickly programmed to 100% ACN and held there for 24 minutes.
The PAH fraction was collected from 8-23 minutes, the Nitro-PAH
from 23-29.5 minutes and the Oxy-PAH fraction from 29.5-42.5
minutes. Recoveries through the fractionation ranged from 60 to
98% depending on the compound and concentration of the sample.
412
-------
Analysis of Oxy-PAH and Nitro-PAH
Analysis of the semipolar Oxy-PAH fraction was performed with
Carlo Erba model #4130 gas chromatograph (FID) with a 30 m fused
silica J&W 1701 column. This column gave better resolution of
the compounds of interest when compared with a less polar 30 m
fused silica J&W DB-5 column. A 30 m J&W carbowax 20 m column
also was evaluated, but its upper temperature limit of 220 to 240
°C and high column bleed prevented analyis of peaks eluting after
9,10-anthracenedione(9,10-anthraquinone). Identification of
selected Oxy-PAH were determined from GC/MS analysis of the Oxy-
PAH fraction on either a Hewlett Packard 5992 GC/MS or a Finnigan
Mat 4500 GC/MS system. Repeated GC/MS analyses of chamber soot
samples of the Oxy-PAH fraction indicated that compounds
identified from authentic standards or compounds tentatively
identified by comparison with library spectra, were free'"J"
contamination by artifact phthalate peaks. J1* J;®"*®^®1* ®nd
authentic standard identified compounds had an
coefficients greater than or equal to <>• J "J***SSiX™
the eight most intense mass sPect"i_pefJjf Sv pah v
The structure and names of sevendi£fe*;?UjLT^blelA
monitored in a chamber experiment are J*"?*' *
reconstructed ion chromatograph of an Oxy-PAH fraction from
chamber wood soot sample is shown in Figure i.
Nitro-PAH were analyzed by reverse d h
fiuoreficent detection using a method originally developed by
£ja£ an5 co-worKers^. A supelco PAH Sum used for
chromatographic separation. Th« «^uting ta col
ca«ly.t The size of
the catalyst particles was estimated to be 40 to lOO^um and the^^
reducer column was maintained at 60 c* nen„tha of 360 & 430
monitored at excitation and emission wave lengths of 360 & 430
nm; 7-nitrobenz(a)anthracene (7N-:BaA> * 6-nitrobenzo(a)pyrene
chrysene (6N-Chry) at 273 & 437 nm; and 6-nitrobenzo(a)pyrene
(6N-BaP) at 420 & 495 nm.
RESULTS AND DISCUSSION
PAH and Oxy-PAH stability on Soot Particles
to determine the recovery of
sampling, clean air, at a flow rate of 18 1/min for 20 minutes,
was drawn through XAD cartridges rations were 100 no Der
mixture of Oxy-PAH. The Oxy-PAH concentrations were 100 ng per
species, and the °^Poun
-------
During experiments with wood smoke (ave temp - 21 °C),
partitioning between the filter and the XAD cartridge was
observed for lower molecular weight oxygenates. For example
34,1% (sd =15.2; n-10) of the recovered 9-fluorenone (mw » 180,
bp - 341.5 °C) was found in the XAD-2 extracts. 'For 9,10-
anthraquinone (mw ¦ 208, b.p. 379.8 °C) 91.1% was recovered
from the filter extracts (sd»7.5%; n«10).
Thus in this paper we have assumed that compounds with
boiling points and molecular weights similar to or greater than
that of 9,10-anthraquinone have a tendency to exist primarily in
the particle phase and are not: appreciably stripped from our
filters during sampling.
Wood soot eystems aged for several hours in the dark show
that PAH and Oxy-PAH concentrations (ng of compound per mg of
collected soot) are stable. When wood smoke with PAH and Oxy-PAH
loadings in the thousands ng/mg was aged in the presence of
midday sunlight (NO.. < 0.07ppm, photochemical o3 < 0.08 ppm), PAH
concentrations slowly declined . Oxy-PAH concentrations however,
appeared to be more stable (Figure 2}. At lower organic particle
loadings and moderate humidities, PAH tend to degrade more
quickly in sunlight. By analogy, we speculate that under such
lower concentration conditions, Oxy-PAH species also would be
more stable than PAH. A possible exception is 7-benzanthrone.
In all of the sunlight experiments it exhibited some decay while
other Oxy-PAH appeared to be more stable.
Based on these experiments it is not possible to determine
the extent to which simultaneous Oxy-PAH synthesis and
destruction processes are occurring as airborne soot particles
age in the presence of sunlight. Our results suggest that if
significant synthesis occurs, then it must be offset by a loss
process in order to maintain the relatively stable time-
concentration profiles that are observed.
NITR0-PAH Behavior
A diesel soot-particle'aerosol free from combustion gases was
established in one chamber using the NBS diesel soot particulate
standard (SRM #1650) and the liquid nitrogen injector. Chamber
Nitro-PAH concentrates from SRM #1650 diesel soot aged in the
presence of sunlight are illustrated in the bottom graph of
Figure 3. The average temperature for this run was 34.9 °c, the
dew point was 17,4 °C and the average total solar intensity was
0.81 cal cm" min . The chamber contained rural background N0X
and NMOC concentrations. Photochemically generated 0-j
accumulated to a maximum concentration of G.06ppra. Figure 3
suggests that most Nitro-PAH are stable on these soot particles
under the noted experimental conditions, although some compounds
(6-NBap) may decay.
Simultaneously, in the second chamber, we conducted an
experiment with automobile diesel emissions. The initial HO and
NO, concentrations were 0.31 and 0.16 ppm. The NMOC was 0.6 ppmc,
ana 0.03 ppm of O3 accumulated. PAH and Nitro-PAH on these
particles exhibited a rapid decay (Figure 3, top graph). He
were unable to determine if nitro compounds that can be generated
414
-------
from secondary photochemical reactions, (e.g., if 2—nitropyrene
or 2-nitrofluoranthene), were formed during these experiments.
Particle measurements taken with a laser optical particle
instrument (LAS-X, PMS, Boulder, CO.) indicate that the Mercedes
diesel particles had a symmetrical particle size distribution
with a mean mass diameter of 0.196 um. This size distribution
did not change significantly over the course of the experiment.
The NBS SRM #1650 particles had a broader mass size distribution
(mean » 0.293). The calculated surface to volume ratios
assuming smooth, spherical particles) for both the fresh and the
NBS diesel soot particles did not change appreciably with time
durinq the course of an experiment. The surface to volume ratio
of NBS diesel soot was approximately 40% lower than the fresh
Mercedes soot (assuming that the paf countor
responded to both types of soot particles in tlhe aame mariner).
It is additionally possible that as NBS Jiesel e°°*
aged (in their sealed vial cont?inersim^^5
experiments, the surface morphology ehJnnSis and
volatile compounds were lost. This may ^ave :
interstices on the particle surface. The net an
additional reduction in the effective 8)£*a°e av?**a^ f?*
reaction with light and other reactive'SJ" " i vtnl
environment of the particle. Lastly, we do not know if suspension
o£ the™*! dl...l pirtlcles In the liquid nitrogen Injector
altered the surface of the particles.
SUMMARY
The preliminary result, reported In this study suggest that
high Oxy-PAH loadings on airborne wood soo't particles and ambient
air fNO < 0.07ppra) appear to be stable in sunlight. Similarly
£lgh'p& concentrations slowly decline. When fresh diesel soot
particles were
decreased in concentration. However, r«n atabie in
sunlight^nder^the^san^outdoor conditions^^Differences^in^l^ ^
Sr^o^c^ic^en^o^ *Jesel
behavior between the same Nitro-PAH on these two types of diesel
soot particles.
acknowledgements
u fhank Mr. Joe McElwee and Ms. Kelly
The authors would like to than* nr. snonaorad hv a
Dix for their technical assistance. 812820-02-0 from
grant, #R8l2256,and cooperative 2®JJ?eIrolSa This caner has
Si undergone S^reWev'for Plication
or recommendation for use.
415
-------
REFERENCES
(1) R.H. Kamens, D,A. Bell, A. Dietrich, J.M, Perry, R.G.
Goodman, L.D. Claxton," Mutagenic Transformations of Dilute Wood
Smoke Systems on the Presence of Ozone and Nitrogen Dioxide"
Environ.SCi.& Technol., 19, 63-69 (1985).
(2) R.M. Kamens, J.M Perry, D.A. Saucy, D.A Bell, D.L. Newton,
Brand,B; "Factors Which Influence Polycyclic Aromatic Hydrocarbon
Decomposition on Wood smoke Particles" Environment
International, 11, 131-136, <1985)
(3) R.M.Kamens, J.N. Fulcher, Z.Guo. "Effects of Temperature on
Wood Soot PAH Decay in Atmospheres with Sunlight and Low NOx",
Atmospheric Environment, 20:1579-1585 (1986)
(4) D.A.Saucy, R.M.Kamens, R.W. Linton "An Aerosol Injection and
Outdoor chamber System for the Study of Atmospheric Gas-Particle
Reactions", Atmospheric Environment, 17,: 2617-2624 (1983)
(5)s.B. Tejada, R.M. Zveidinger, J.E. Sigsby, SEA Technical Paper
Series # 820775, (1982)
Table 1
Structures of Compounds Monitored in Oxy-PAH Fraction
NAME MOLECULAR WT.
o
9-FLUORENONE 180
benzo-ocinnoune /r%jr\ ibo
9.20-ANTHRAQWN0NE (fll Tfjl 208
HYDROXY-FXUORENONE 186
CYCLO FENTA-4ef- J\ 204
FHEMANTHBONG
& ANTHRACENE aoe
CAHBOXAU1EHYDE
"0:GO
9PHENANTHHOL 194
o
BENZANTKBONE 830
416
-------
RIC DATAl F87923 II
81/1V87 IIiMiM CM.Ii F87021 13
SAflPLEi U30 10-30-86
COHDS.i 30N 0B17B1U ,25U 1.2ML/I1IN 138 TO 290<5> 8/T1IN
SCAMS I TO 1500
U 20, 3
1646
1,1500
LABELi
!
1286
20100
99942.
200
3t20
SCAN
TINE
x. Reconstructed ion chromatograph of Oxy-PAH fraction from a
Camber wood soot extract. 30 m 1701 0.25 u fused silica column(J&W). 50
®®c spiit/splitless inj at room temp; program from 130 C to 290 °C.
'innigan Mat 4500 GC/MS. Authentic standards were available for 9,10-
nthraquinone, 9-fluoreone, 7-benzanthrone, and benzanthracene7-
2#dione. other peaks tentively identified by library matching.
417
-------
WOOD SOOT Daytima PAH D«cay
Davtlme PAH Decay
M
So* \
3.1 *»
3-4 -
\
3 -
\
1.» -
3-3 -
\
\
1 A -
3 -
\
%- 1.7 -
\
2-t •
\
I
2.* -
2.4 -
—
%A-
1.3 -
a»r a ^>v
If
*fM' Vv
I| u-
ii-
1.» -
j
u
Z 1 -
t.« •*
8 «-
MCf < r-
1,4 -
OA ¦
u -
QJ •
1 -
04 •
OJ -
¦¦¦ "i i • r —r ,,K" «
O M •
12
twc IN HOWS
TWC IN HOURS
OXY-PAH
•.lO-AMTH
TIME M HWdJ (CST)
OXY-PAH
MAtCH *, IM7
s
to
s?
I!
o.*
a*
6.7
0.1
O.S
0.4
0*
O.J
CP(d«f)PH£N
9,10—AWTH
^^BCNZANTH
SCNZAN—1
9 *
anthald
Tmc in hou*$
Figure 2. Daytime Decay of Oxy-PAH and PAH in Sunlight. Left; Experiment
on February 8, 1987, ave temp = 21 °C; Right, March 4, 1987, ave temp =
16 oC;
-------
Mercedes DIESEL ntfro — PAH IN SUNLIGHT
a -
i -
Aug 26, 1BB6 EAST
7N-BaA
6N-B0P
17
TIME IN HOURS
NBS DIESEL nitro-PAH IN SUNLIGHT
3.9 -4
3
2.S H
ts
TIME IN HOURS
Figure 3. stability of Nitro-PAH on Diesel soot Particles in
Sunlight; 1N-P particle loading divided by io in NBS diesel
experiment(bottom graph), and tentitive identity of 6N-Chry
assigned on the basis of retention time.
419
-------
COMPARATIVE MUTAGENIC ACTIVITIES OF SEVERAL PEROXYACYL NITRATES
T.E. Kleindienst, P.B. Shepson, C.M. Nero, E.E. Hudgens
Northrop Services, Inc. - Environmental Sciences
Research Triangle Park, NC 27709
L.T. Cupitta, J.J. Bufalinia, L.D. Claxtonb
^Atmospheric Sciences Research Laboratory
^Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Salmonella typhimurium strain TA100 was exposed to a series of
peroxyacyl nitrates including peroxyacetyl nitrate (PAN),
peroxypropionyl nitrate (PPN), peroxybutyryl nitrate (PBN),
peroxybenzoyl nitrate (PBzN), and chloroperoxyacetyl nitrate (CPAN).
Gas-phase concentrations for the individual exposures were in the high
ppbv range. The deposition rate (i.e., the dose) could be determined
from the measured decrease of the gas-phase concentration in the
exposure chamber and the exposure time. The mutagenic activity of
each compound was determined from linear dose-response curves. Values
for the mutagenic activity ranged from 250 (PBN) to 7350 (PBzN)
revertants/nmole. The mutagenic activity of PAN as a function of
initial ga3-phase concentration showed nonlinear effects that could
not be attributed to Henry's Law considerations.
420
-------
Introduction
Peroxyacyl nitrates (PNs), RC(0)OONO2i represent oxidants formed
in the polluted troposphere. Aldehydes, except for formaldehyde,
serve as important precursors to PN formation following their reaction
with OH. The peroxyacyl radical formed combines with NO2 to produce
PNs. Thus, acetaldehyde leads to peroxyacetyl nitrate (PAN)
formation, propionaldehyde to peroxypropionyl nitrate (PPN) and
butyraldehyde to peroxybutyryl nitrate (PBN). PAN has long been
associated with possible physiological damage, and it has been shown
to be mutagenic in the Ames Test with Salmonella typhimurium, strain
TA10Q.1 However, quantitative data on the toxicity and genotoxicity
of PAN are fairly limited. Of the PNs only PAN, PPN,2 ancj
peroxybenzoyl nitrate (PBzN) (vide infra) have been observed in
ambient air.3 PAN concentrations in polluted air generally range from
1-20 ppb with the others one to two orders of magnitude lower in
concentration.
Over the past four years, we have examined the mutagenic
activities of mixtures of hydrocarbons (e.g., propylene, toluene) and
oxides of nitrogen that have been irradiated in a smog chamber. In a
number of these mixtures, strong mutagenic activities have been
observed from the products of these oxidations. Although the mixtures
were complex, we have been able to determine that PAN was one of the
primary contributors to the total mutagenic activity in TA1Q0.^ We
are currently examining the mutagenic activities of a series of
peroxyacyl nitrates. These include PAN, PPN, PBN, PBzN, and chloro-
peroxyacetyl nitrate (CPAN). (The last two compounds arise as
products from the oxidation of benzaldehyde and chloroacetaldehyde,
respectively.) The results of these measurements are of significance
with respect to assessing the human health hazards associated with
exposure to urban air.
Experimental Methods
The experimental method involved flowing dilute mixtures of the
test compound in air into an exposure chamber that houses the test
bacteria, S. typhimurium. Several timed exposures were performed for
each compound in order to obtain dose-response curves. The mutagenic
activities were determined from the measured deposition of each PN and
from the dose-response curves. Two methods were employed to generate
the PNs: (1) the appropriate aldehyde was oxidized by NO3 in the
presence of NO2 and air, and (2) a solution-phase synthesis was used
that involved the nitration of a peroxyorganic acid to give the
peroxyacyl nitrate.
In the first technique (which has been previously described with
respect to the formation of PAN^}, the appropriate aldehyde, N02, and
O3 in air were mixed in a 22.7-m3 Teflon reaction chamber. The
chamber was operated in a dynamic mode, which allowed steady-state
concentrations of PNs to be generated in the chamber. The effluent
from the reaction chamber wa3 then fed into a 190-L chamber where the
bacteria were exposed to the gas-phase PNs. This procedure was
421
-------
employed to generate CP AN, PBzN, and PAN. For CPAN and PBzN,
chloroacetaldehyde (50^ in water) and benzaldehyde (neat),
respectively, were added in the vapor-phase to a mixing manifold by
bubbling N2 through the liquid. For the PAN preparation, acetaldehyde
was taken from a dilute tank mixture in N2 (0.8^). The aldehydes were
introduced into the manifold to give an initial concentration of
1 ppm. O3 was added to the manifold from the output of a commercial
ozonizer, and NO2 from a concentrated tank mixture in air at
concentrations of 1 and 2 ppmv, respectively. Dilution air was
provided using an Aadco zero-air generator. All flows were regulated
using mass flow controllers to give the desired chamber concentration.
The average residence time of gases in the chamber was 6.7 h. There
was no evidence of major product formation other than the desired PN.
In addition, there was no evidence of appreciable decomposition of the
PNs.
The PNs were measured directly from the reaction and exposure
chambers by GC as has been previously described for PAN.1 The GC used
to measure PAN was calibrated with a standardized solution of PAN in
dodecane (vide infra). A known quantity of this solution was injected
into air in a Teflon bag to make the gas-phase standard. These
solutions were not synthesized for PBzN and CPAN. For these two cases
it was assumed that the decrease in the aldehyde concentration
measured between the inlet and the effluent of the reaction chamber
represented the concentration of the PN formed. This assumption was
tested for PAN and found to be valid within 15%.
The biological assay used in this work employed the bacteria
S. typhimurium, strain TA100. Tester strains were obtained from
Dr. Bruce Ames, University of California, Berkley, CA, and maintained
by the Genetic Toxicology Division of the Health Effects Research
Laboratory, U.S. EPA Environmental Research Center. The test
procedures were those of Ames etal.5 with modifications as described
earlier (1).
Each exposure chamber (clean air control and effluent) was filled
with 45 petri dishes containing the tester strain and five survivor
plates used as a qualitative check for toxicity. The exposure chamber
was then brought up to the same steady-state concentration of the PN
as in the reaction chamber. The covers on the petri dishes were then
removed to begin the exposure. The total mass of PN deposited into
the plates was determined by measuring the concentration in the
exposure chamber before and after the covers were removed. Exposures
were performed for various lengths of time to obtain a dose-response
curve. We have found that re-covering a dish effectively stops the
deposition of material and that the deposition is linearly related to
the exposure time for PAN. Exposure periods ranged from 15 to 600
min. Following the exposures, the bacteria were incubated for 48 h at
37°C, and the number of revertant colonies/plate was counted. The
exposure of the bacteria to clean air served a3 the control.
The second experimental method (i.e., the nitration of the
peroxyorganic acid) was employed for PPN and PBN. These PNs were
prepared according to the procedure outlined by Gaffney etal.6 The
synthetic procedure ultimately yielded a solution of the PN in
dodecane. This solution could then be used as a steady source of
422
-------
vapor-phase PN. PAN was also synthesized for analytical purposes.
The concentration of PAN in dodecane could be determined by
hydrolyzing PAN to give CH3C(0)0~ and NC>2~ and measuring NO2" by IC.
For PBN and PPN, the analogous conversion to N02~ was significantly
slower, and reproducible results could not be obtained. Thus, the
solution concentration was not determined.
The electron capture GC could not be calibrated directly for PPN
and PBN in air. Thus, these PNs were measured on the NOx channel of a
calibrated NOx analyzer. It was assumed that the response of the
analyzer to PPN or PBN was equivalent to NO2 on a molar basis.
Although this assumption could not be tested for PPM or PBN, it was
tested for PAN since gas-phase standards in air could be generated.
In this case, the magnitude of the response generated by PAN was
identical to that from an equivalent concentration of NO2.
To perform the exposure, the PN in dodecane was added in
sufficient quantity to a bubbler. N2 was bubbled through the
solution, which wa3 thermostatted to 0°C. The N2 stream containing a
equilibrium concentration of the PN was added to a 150-L manifold
where it was mixed with zero air. The contents of the manifold were
flushed through the exposure chamber containing the biological assay.
All further experimental techniques for this method were identical to
those described above, except that the exposure chamber concentrations
were measured with the NOx analyzer rather than by GC.
In each of the experiments, the primary data (i.e., the number of
revertants/plate and the difference in concentration of the test
compound before and after the plates were uncovered) leads to the
calculation of two quantities, the mutagenic activity (revertants/h)
and the deposition rate of the test compound (pmoles/h). The
mutagenic activity was obtained directly from the slope of the dose-
response curve. One such plot has been presented previously for PAN.**
The number of revertants/h (per plate) was determined from the linear
portions of the dose-response curves. Effects of toxicity can lead to
nonlinear plots with negative curvature. All plots obtained in this
study were linear, although PPN did give depressed survivor levels at
the longer exposure times.
The fraction of the gas-phase test compound that deposits into the
plates is used to determine the deposition rate, which is calculated
from the following equation
where [PJV]0 is the initial concentration of the peroxyaeyl nitrate in
ppmv, o is the fraction of the PN depositing into the medium, V is
the flow rate through the chamber (850 L/h, in all cases), and Np is
Results
(1)
423
-------
the total number of plates. The fraction (a) is independent of
concentration, but can vary with flow rate through the chamber.
The results of these experiments have been tabulated in Table I.
The table lists the compounds tested along with the procedure used to
perform the exposure. [PN]0 and (a) were measured during the
exposures, whereas A[PN]/At wa3 calculated from equation (1), The
slope from the dose-response curves is given in the second to last
column. The mutagenic activity (revertants/nmole) is given in the
last column and was obtained by dividing the (revertants/h) by
{jamoles/h).
For the PAN, PBzN, and CPftN exposures, significant concentrations
of the reactants were present in the effluent mixture. The reactants
were thus tested for mutagenic activity by performing an exposure of
the initial unreacted mixture. For PAN and PBzN, the initial mixture
was nonmutagenic. This was not the case for CPAN, because the initial
mixture contained C1CH2CH0, which is a mutagen of TA100. Thus, the
mutagenic activity for the effluent mixture of CPAN needed to be
corrected for the contribution of CICH2CHO. However, the CICH2CHO
contribution to the mutagenic activity was greater than that due to
CPAN, because CICH2CHO is significantly more soluble in the test
medium than is CPAN. Thus, only the upper limit for the CPAN
mutagenic activity was calculated.
In previous studies!,4 we have examined the mutagenic activity of
PAN and found various measurements to differ by up to an order of
magnitude. It was originally suggested1* that PAN decomposition
products might be the cause of the discrepancy. Subsequently, a
series of expoaures was performed at a number of different gas-phase
PAN concentrations. It was found that while the mutagenic activity
(revertants/h) increased linearly with exposure time (i.e., dose), the
activity was largely independent of the gas-phase concentration over
the 500-3000 ppbv range. This observation would imply that the test
medium had reached saturation (i.e., Henry's Law conditions) for PAN.
However, as previously observed (ref 1, Figure 1) even at a gas-phase
concentration of 1 ppmv there is no evidence of PAN saturation in the
medium after a 20-h exposure. In addition, for relatively large PAN
concentrations, the dose-response curve is linear throughout the
exposure, implying that the rate of deposition remains constant and
Henry's Law conditions are not being approached. Therefore, it is
currently unclear why the measured reversion rate should be
independent of the gas-phase concentration. For the present, however,
a reversion rate of 14 rev/h represents an average mutagenic activity
over a wide range of gas-phase concentrations (104-850 ppbv).
Conclusions
Mutagenic activities for a series of PNs have been presented.
PBzN gave the greatest activity being nearly an order of magnitude
greater than any other tested. However, the values for the mutagenic
activity in Table I should be considered preliminary values until it
is determined what effect changing the gaa-phase concentration has on
the mutagenic activities.
424
-------
Disclaimer
Although the research desribed in this article has been funded
wholly or in part by the U.S. Environmental Protection Agency through
Contract 68-02-4033 to Northrop Services, Inc. - Environmental
Sciences, it has not been subject to the Agency's peer and policy
review. Therefore, it does not necessarily reflect the views of the
Agency, and no official endorsement should be inferred.
References
(1) T. E. Klelndienst, P. B. Shepson, E. 0. Edney, L. D. Claxton,
"Peroxyacetyl nitrate: Measurement of its mutagenic activity
using the SaZmonW/a/mammalian microsome reversion assay," Mutat.
Res. 157: 123 (1985).
(2) S. A. Penkett, F. J. Sandalls, J. E. Lovelock, "Observations of
peroxyactyl nitrate (PAN) in air in southern England," Atmos.
Environ. 2: 139 (1975).
(3) G. M. Meijer, N. Nieboer, "Determination of peroxybenzoyl nitrate
(PBzN) in ambient air," in Photochemical Smog Formation in the
Netherlands. R, Gulcherit, ed., TNO, Gravenhage. 1978, pp. 60-63.
(4) P. B. Shepson, T, E. Klelndienst, E. 0. Edney, C. M. Nero, L. T.
Cupitt, L. D. Claxton, "Acetaldehyde: The mutagenic activity of
its photooxidation products," Environ. Sci. Technol. 20: 1008
(1986).
(5) B. N. Ames, J. McCann, E. Yamasaki, "Methods for detecting
carcinogens and mutagens with the Salmonella/manorial ian microsome
mutagenicity test, Mutat. Re3. H: 347 (1975).
(6) J. S. Gaffney, R. Fager, G. I. Senum, "An improved procedure for
high purity gaseous peroxyacyl nitrate production: Use of heavy
lipid solvent," Atmos. Environ. V8: 215 (1984).
Table I. Summary of Experimental Results and the Mutagenic
Activities for the Exposure of Peroxyacyl Nitrates (PNs)
to S. typhimurium, Strain TA100
Mutagenic
Mutagenic
[PN}0
A[PB]/At
Activity
Activity
PN
Method
ppbv
a
(limoles/h)
(rev/h)
(rev/nmole)
PANa
-CHO/NO3
850
0.161
0.0374
14
a
PPN
S0LTN
332
0.133
0.0306
10.8
353
PBN
SOLTN
287
0.160
0.0320
8.1
254
PBzN
CHO/NO3
138
0.070
0.0067
49
7350
CPAN
-CHQ/NO3
106
0.210
0.0155
<10
<645
aSee discussion.
425
-------
AN EXPERIMENTAL TECHNIQUE FOR MEASURING THE
OH RATE OP ATTACK ON PARTICLE BOUND PAH
Zhishi Guo, Richard Kamens and Harvey Jeffries
Department of Environmental Sciences and
Engineering, University of North Carolina
Chapel Hill, North Carolina 27514
last update 5/2/1987 19:20 PM
An experimental technique for studying OH rates of reaction
with particle-bound polyaromatic hydrocarbons (PAH) is described.
PAH on combustion soot particles were reacted directly with OH
radicals in a 200 liter continous stirred tank reactor (CSTR).
To provide a relatively stable source of particles for reaction
in the CSTR, wood soot particles were injected into a 25 m3
outdoor Teflon film chamber. Soot particles were drawn at a
constant flow into the CSTR. The large chamber in effect served
as a storage tank or reservoir for combustion soot particles.
The photodissociation of methylnitrite inside the CSTR was
used to provide a source of OH. PAH concentrations at the inlet,
[PAH]|n, and outlet, [PAH]out, of the CSTR were monitored along
with vapor phase species, such as propylene or ethylene,
[Ref]OU£, which have established literature OH rate constants
(kp). The OH rate constants for selected PAH (kx) was then be
determined from the expression given below if the rates of
reaction with in the CSTR (r^ for PAH and r0 for the reference
compound) can be determined.
rPAH kl fPAH]put
rref k0 tRefJout
Non steady state CSTR equations were developed to calculate
r for both particle bound and gas phase reactants. Other
reactants in the CSTR which result in a loss of PAH (i.e. 03,
N02, H0NO2, photolysis, etc) were considered in the PAH-OH rate
constant estimates.
426
-------
AN EXPERIMENTAL TECHNIQUE FOR MEASURING
THE OH RATE 07 ATTACK ON PARTICLE BOUND PAH
INTRODUCTION
The reaction of vapor phase organic compounds with hydroxyl
radicals(OH) is one of the most significant reaction pathways of
both tropospheric and stratospheric chemistry. The relatively
high OH rates of reaction for two and three ring PAH^J suggest
that OH attack may be the major loss process for these compounds
in the atmosphere. However, virtually nothing is known about the
reactions of OH radicals and polyaromatic hydrocarbons (PAH) with
more than four rings. This is because there has been difficulty
in developing a technique for studying the reaction of particle
bound PAH with gas-phase OH radicals.
GENERAL APPROACH
In this work PAH on soot particles were reacted with OH
radicals in a continuous stirred tank reactor (CSTR), To provide
a relatively stable source of particles for reaction in the CSTR
we injected into one of our large outdoor chambers(25m ),
combustion soot particles from a residential wood stove. The
large chamber was a reservoir for the feed aerosol which could
then be introduced at a constant flow rate into the CSTR (Figure
1) . Real soot particles were used because we have found that
compounds coated onto substrates (silica, alumina, carbon black)
do not necessarily exhibit the same atmospheric reactivity as
these same compounds do on real soot particles.
We have previously shown that particle size distributions do
not change appreciably with time in our outdoor chambersI »3J,
and that the slow loss of particles to the chamber walls can be
characterized. We have also shown that in the dark, particle
bound PAH, nitro-PAH, and aromatic carbonyl mass as measured in
ng per mg of soot are very stable over a period of many
hoursf3'.
The photolysis of methylnitrite (CH3ONO) inside the CSTR was
used to provide a source of OH according to the following
reactions:
CH3ONO + hv —> CH3O + NO
CH30 + 0, —> HCHO + HO2
H02 + NO —> OH + N02
To inhibit the formation of O3, N2°5 an<* N03» excess NO was
added to the inlet flow. PAH concentrations at the inlet and
outlet of the CSTR were monitored along with vapor phase species,
such as propylene or ethylene, which have established literature
OH rate constants. These vapor phase compounds were used as
reference compounds. The OH rate constants for selected PAH was
427
-------
determined by comparing the relative loss of a PAH with that of
the reference compound. This approach is similar in concept to
the relative rate technique developed and used in batch reactors
at the University of California at Riverside^5-'J. In the case
of CSTR, as will be discussed later, the equation for rate
constant calculation is different, but one still need measure
only the decay of the reference compound and compare that to the
decay of a selected PAH to determine the unknown rate constant.
EXPERIMENTAL
The 200-liter CSTR used in this pilot Study has been
previously described by Gery and co-workers'8J. It is surrounded
by thirty-four long-wave UV lamps(GE, F20 BLB). Gas and filter
samples were taken at both the inlet (large chamber) and the
outlet of the CSTR. To aid in the transfer of material from the
large chamber to the CSTR, an airtight in-line fan was used
between outdoor chamber and the CSTR inlet. The inlet/outlet
flow was maintained at 9.24 liters per minute, corresponding to a
residence time of 21.6 minutes. The in-line fan and CSTR outlet
pump were balanced so that one atmosphere of pressure existed in
the CSTR. Particle size measurements (0.09 to 3.0 um) were
performed with a laser optical particle counter (Model Las-X,
Particle Measuring Systems, Boulder, CO). The effects of the in-
line fan on CSTR particles size distributions were negligible.
The method of MeONO preparation in our laboratory has been
previously described'9J. Calculated volumes of ethylene and/or
propylene were directly injected into the chamber and were
monitored by a on-site GC (Carle model 211, isothermal at 60°C,
porapack Q 50-80 mesh 1/8" SS col, FID) . NOx was monitored with
a Bendix model 8101-B chemiluminescent analyzer (Ronceverte, WV
24970, USA). Oj was monitored with a Bendix model 8002
chemiluminescent analyzer.
To avoid any adverse light effect in the large chamber
reservoir, all the experiments were conducted after dark. Wood
smoke emissions were added directly from the chimney of a
residential wood stove to the 25 m3 outdoor Teflon film chamber.
Split pine logs (6 cm x 10 cm x 50 cm) were used as a fuel and
these were burned at a rate of 2 to 6 kg/hr. The resulting
dilute wood smoke had initial particle concentrations of 1500 to
3000 ug/m3. A calculated volume of methylnitrite (500 to 800
c.c.) was flushed into the outdoor chamber by a nitrogen stream.
When the addition of particles from the feed reservoir began, the
lights around the CSTR were turned on. Sampling began 90 minutes
later.
Particle samples were taken from both the large chamber and
the CSTR outlet. Soot particles were collected on 47 mm teflon-
impregnated glass fiber filters (Pallflex T60A20) at a flow of
9.24 liters/min. Particulate masses on the filters were ranged
from 0.2 to 1.0 mg.
428
-------
Filter samples were extracted in 25 ml micro soxhlet
extractors with methylene chloride (Burdick and Jackson, HPLC
grade) 3-4 hours after collection. Sample work up procedures and
analytical techniques used for PAH analysis have been described
elsewhere*- J.
THEORETICAL CONSIDERATIONS
A CSTR consists of a well-stirred tank into which there is a
continuous flow of reacting material and from which the partially
reacted material passes continuously. of CSTR in
steady state has been well established' J . The mathematics
that describes the behavior of the CSTR are based on the
assumption of perfect mixing. A fair approximation to perfect
mixing is not difficult to attain in a CSTR provided that the
fluid phase is not too viscous. The rate expression for a CSTR in
the steady state is:
[C]in - Cc3out . .
r (1)
X
where:
r « the reaction rate for the chosen reactant with the units of
concentration per unit volume per unit time;
[C]in m the inlet reactant concentration;
[C] t® the outlet reactant concentration;
*C ¦ the residence time or holding time, which is defined as 6 »
V/F; where f is the inlet/outlet flow rate in volume per unit
time, and V the volume of CSTR; After seven residence times at a
constant inflow, the CSTR will attain 99.9% of its steady state
concentration.
CSTR in Non-Steady state
In this work gas and particle phase species in the large
reservoir were slowly decreasing. This occurs because the large
reservoir is being slowly diluted by drawing samples into the
CSTR, and because particles are being lost to the reservoir
walls. These loss mechanisms slowly change the rate at which
reactants enter the the CSTR. It is necessary therefore, to
consider the situation of a CSTR under non-steady state. For this
purpose, we have developed two rate expressions.
For gas-phase species, the rate can be determined by
£cUn ~ tC]out
r s (2)
C
where [C]in, [c3oUt' an<* r are same as in Eq. (l) ; term s here
is defined as, s «* d[C]QU^/dt, the rate of change of
concentration in the CSTR. The value of s can be determined by
finding the slope of [C]out vs. time curve. This equation looks
very similar to Eq.(l) except an extra term s. As one can see,
Eq.(l) is actually a special case of Eq.(2). This expression may
be used for reactions of any order.
429
-------
The behavior of reactant molecules on the particle surface differ
from that of gas molecules. Although particles entering the CSTR
can freely move about, the molecules on particle surface can not.
They cannot leave the surface and intermix with other gas
molecules. Each particle retains its identity and acts as a tiny
batch reactor. We call this type of mixing macromixing'-1®*. For
particle bound species and first-order PAH (or pseudo first-
order) reactions, the rate can be determined by
{A}in - (A) ut
r P (3)
X*
where:
{AJj„ and {A)out are the inlet and outlet concentrations of
particle bound species respectively, measured in molecules per mg
of particles, and P the particle concentration in the CSTR during
sampling period, measured in rag of particles per m3 of air.
"C* is the average Residence time for particles collected
from the outlet flow. "C may not necessarily equal X, and its
value is affected by the change of inlet particle concentration
and the particle ljss rate in the CSTR. The experimental
determination of T. is not difficult. The step-by-step
derivation of Eq.(2) and Eq.(3) will be discussed elsewhere.
However, our computer modeling has shown that both equations are
correct and accurate.
Calculation of the OH-PAH Rate Constant
We have assumed that OH attack on PAH and a reference
compound is a first order reaction with respect to OH and the
reactants of interest. In actuality, kj may be a pseudo secondary
order rate constant which embodies other features on the particle
surface which, in turn, promote the reaction of OH with a surface
PAH molecule.
According to the perfect mixing assumption, the CSTR effluent
stream should have the same composition as its contents, rpah is
the rate of reaction of a specific PAH and rref is the rate of
reaction of a the reference compound.
Thus:
rpah = kl £°H3 [PAH]out («>
rref = ko £0H} [Ref]out (5)
Combining these two equations gives:
rPAH _ kl £PA«W_ ^
rref tRef3out
Note that the [OH] terms were canceled out in the above
equation, which means OH concentrations need not be known and
need not remain constant during the experiment. The reference
430
-------
rate constant, kQ, is known from established literature values.
By utilizing Eg.(3) to calculate rj and Eq.(l) or Eg.(2) to
calculate r0, the OH rate constant, k^, can be determined. In the
determination of r^, all reaction rates which lead to a loss of
PAH must be subtracted from r^.
PRELIMINARY EXPERIMENTS AND FEASIBILITY TESTS
To estimate the OH rate of attack of a given PAH on a soot
particle surface, we must also know the reaction rate from other
processes which might result in a PAH loss. These include PAH
reactions with light, 0-j, N02, H0N02, HONO, and N20g. in the OH
experiment to be described we estimated that 0.05 ppm 0-j, 8.6 PPM
of No2, 9.0 PPM of HONO5, 0,4 ppm HONO, 0.003 N20e formed in the
CSTR. Since the data to be presented are very preliminary, we
will report for illustration purposes only the rate data for
benzo(a)pyrene (BaP). It is important to reeraphasize that these
data are very preliminary.
In characterization test of the CSTR we determined the photolysis
loss constant of BaP dus to CSTR lights to be 1.24x10 min .
Base on previous work^ ', the half-life of BaP at 0.05 ppm 03 is
in excess of 10 hours and thus the loss due to was not
considered important. With 23 ppm of NO, + wood soot in the CSTR
(dark), we determined the k»r02 to be 1.28 x 10 cm molecules
sec . Separate CSTR experiments with 4 ppm of nitric acid did
not show a loss of BaP. This is consistent with the work reported
by Lindskog et al.[I2J and Pitts et BaP on wood soot
particles passively exposed to 0.1 ppw HONO did also not result
in any BaP loss"-**'.
When 4.28X10-4 [mg cm"3] wood soot particles, 10 ppm of NO and 24
ppm of CH-iNO, were fed into the CSTR with the lights on, a 33%
reduction in BaP resulted in the CSTR. The CSTR temperature was
296 K and the residence time in the CSTR was 1296 seconds (21,6
minutes). The inlet BaP was 1.06x10 j;molecules mg ] and the
outlet was 7.07x10 . This give a BaP concentration in the CSTR
of 3.03x10* [molecules cm-3]. Ethylene was used as a reference
with a rate constant of 8.5 x 10" [cm molecule sec ).
The calculated gross reaction rate for BaP using Eq.{3) was
1.67x10 . Parenthetically, it is useful to add, that when just
ethylene and propylene were separately run in the CSTR with
CH3ono, the OH rate constant of propylene was 3.08 times greater
than that of ethylene. This is very good agreement with
literature OH rate constants for these two compounds.
In the calculation of rate constant for BaP above losses due
to photolysis, NOj were not excluded. The OH rate may be
calculated however as follows:
OH rate = Overall rate - Photolysis rate - N02 rate
=¦ Overall rate - kphot:o [BaP] - kNQ2 [N02][BaP]
« 1.67x10® - 1.24X10"4 x 3.03X1011
-1.28x10 X 7.38X1013 X 3.03X10
431
-------
= 1.67x10® - 3.75xl07 - 2.86x10®
= 1.27x10® [molecules cm""-* sec"1]
It can be seen that OH rate takes 63% of the overall rate.
This value can be translated to an OH rate constant of k =
7.8x10 fcm3 molecule- second"1]. At an assumed atmospheric
OH radical concentration of 1X10 molecules per cm •- J, the
atmospheric half-life for BaP due to reaction with OH radicals is
146 minutes. Daytime wood soot PAH experiments in our outdoor
chambers (solar radiation = l cal/cmg/sec, N0X=0.06ppm),
typically exhibit (under midday sunlight and moderate humidities
and temperatures) PAH half-lives at 293 K which range from 35-50
minutes™'5J. It is reasonable to assume that OH concentrations
from photochemical smog processes are higher than 1X106
[molecules/cm3J in these outdoor runs, if this is the case, then
a significant portion of the previously observed photo-induced
decay may be attributed to OH attack. We have observed in trying
to characterize the photo-induced PAH loss on soot particles in
our large chambers that this process is influenced by the
intensity of sunlight, humidity, particle loading, and
temperature. Future work will address these factors with respect
to OH-PAH reactions.
ACKNOWLEDGEMENTS
The authors appreciate the very insightfull thoughts of
Professor David Leith. This work was supported by a cooperative
agreement (#CR81280-G2) and a research grant<#RB12256) from the
US EPA to the University of North Carolina
MAJOR REFERENCES
[1] Bierman, H.W., Leod, H.M. , Atkinson, R. , Winer, A.M., and
Pitts, N. Jr., (1985), "Kinetics of the Gas-Phase Reactions of
Hydroxyl Radical with Naphthalene, Phenanthrene, and Anthracene",
Environ. 5ci. Technol., 19:3, 244-248.
[2] Kamens, R., Fulcher, J.N. and Guo, Z.f (1986), "Effects of
Temperature on Wood Soot PAH Decay in Atmosphere with Sunlight
and Low Nitrogen Dioxide", Atm. Environ., 20:8, 1579-1587.
[3] Kamens, R. Perry, J., Saurcy, D.A., Bell, D., Newton, D.L.,
Brand, B., (1985), "Factors Which Influence Polyaromatic
Hydrocarbon Decay on Wood Soot Particles", Environ. Int., 11:131-
136.
[4] Kamens, R., Guo, Z., Fulcher, J., and Bell, D., (1986), The
Influence of Temperature on the Daytime PAH Decay on Atmospheric
Soot Particles, presented at the 79th Annual Meeting of APCA,
June 22-27, Minneapolis, Minnesota.
[5] Atkinson, R. , fit a!.,(1979) "Kinetics and Mechanisms of the
Hydroxyl Radical with Organic Compounds - in the Gas Phase",
432
-------
Advances in Photochemistry, vol 11, John Wiley and Sons, New
York.
[6] Atkinson, R., Aschmann, S.M., and Pitts, J.N., (1982), "Rate
Constants for the Reaction of OH Radicals with a1 Series of
Alkanes and Alkenes at 299 + 2 K", Int. J. Chem. Kinet., 14: 507-
516.
[7] Pitts, J.N.Jr., Winer, A.M., Aschmann, S.M., Carter, W.P.L.,
Atkinson, R., and Bufalini, J.J., (1984), "Experimental Protocol
for Determining Hydroxyl Radical Reaction Rate Constants for
Organic Compounds", Atmospheric Sciences Research Laboratory,
Office of Research and Development, U.S. EPA, Research Triangle
Park.
[8] Gery, M.W., Fox, D.L. and Jeffries, H.E., (1986), " a
continuous Stirred Tank Reactor Investigation of the Gas Phase
Reaction of Hydroxyl Radicals and Toluene", Int. J. Chem. Kinet.,
17: 931-955
[9] Benham, D. and Kamens, R.M.(1985), "OH Rate Constant
Intercomparison Study", Final report, Northrop Services Inc.,
Research Triangle Park.
[10] Levenspiel, 0. (1967), Chemical Reaction Engineering, John
Wiley & Sons, Inc., New York.
[11] Denbigh, K.G. and Turner, J.C.R., (1984), Chemical Reactor
Theory, 3rd Ed., Cambridge University Press, London.
[12] Lindskog, A., et al. (1985), "Transformation of Reactive PAH
on Particles by Exposure to Oxidized Nitrogen Compounds and
Ozone", Environ. Int., 11: 125-130.
433
-------
2$ ™
outdoor
~
OS TfK
:~f~
V i f,ur<> 1
Ttie F.xr'wr i,I,wn
V i lot r>Vudy
434
-------
CHARACTERIZATION OF ORGANIC SPECIES ON
JET ENGINE EXHAUST PARTICLES
by
Michael R. Kuhlman, Jane C. Chuang, and Surendra 8. Joshi*
BATTELLE
Columbus Division
505 King Avenue
Columbus, Ohio 43201-2693
Introduction
The aerosol particles emitted by jet engines operated by the U.S. Air Force
are predominantly of submicron diameter and composed of unburned carbon. Compo-
nents of this aerosol which are very minor, on a mass basis, may, however, assume
great importance in the occupational and ambient environment. Trace elements,
especially heavy metals, which may be present at very low concentrations in the
jet fuel will be concentrated in the small particle fraction of the emitted
aerosol due to the very low volatility of these materials. Reactive hydrocarbons,
such as polycyclic aromatic hydrocarbons (PAH) and nitro-PAH may also be present
in condensed form on the particles emitted due to the low vapor pressures of
these compounds. The very small size of the emitted particles permits them a
relatively long residence time In the atmosphere and enables them to be deposited
1n the Innermost portions of the respiratory tract.
The objective of the current study being performed by BattelJe for the
U.S. Air Force is the characterization of the chemicals found on engine exhaust
particles and assessment of their environmental significance. This type of
measurement of these particle emissions from military jet engines has not been
previously reported.
In this study, the gaseous and particulate-bound organic species 1n the
exhaust from three different jet engines operated at several power settings were
analyzed. The engine exhaust was also injected Into 25 m Teflon-walled outdoor
chambers and the evolution of the resulting photochemical system was followed
with periodic measurements of NO , 0,, and total hydrocarbon concentrations.
X <3
During the multi-hour photochemical experiments, samples of the chamber contents
were collected for subsequent analysis of gaseous and particle-bound organics
using gas chromatography-mass spectrometry (GC/MSJ. In a companion study, the
engine emissions were characterized at various power settings for a more detailed
*Aff1Hated with Tyndall AF8, Florida.
435
-------
analysis of the hydrocarbon species present, especially for the lower molecular
weight compounds.
Experimental Method
Three Teflon-walled chambers were used 1n these experiments. Two of the
chambers were used to age engine exhaust under different conditions, and the
third chamber was used as a control, containing either ambient air or ambient
air plus a mixture of reactive hydrocarbon gases and N0X, but no jet engine
exhaust. The chambers were cylindrical in cross section with a conical top, and
the interior surfaces were Teflon or aluminum to minimize surface reactivity.
The chambers had a diameter of 3 in and a height of the peak of 4 m, which provided
a volume of approximately 30 m . Each chamber was equipped with a low rpm
impeller for mixing, with heated injection and sampling lines, and with a blower
and air inlet for purging.
Prior to use in the engine exhaust study, the chambers were "conditioned"
by injection by high concentrations of Oj repeatedly. Chamber characterization
experiments subsequently performed included measurements of leakage rates and 0^
decay under dark and light conditions, examination of chamber background reacti-
vity by performance of ambient air photochemical runs and matched runs of reac-
tive HC-N0X systems.
The matrix of experiments performed which involved injection of jet engine
exhaust Into the chambers Included three engines operated at a total of three
different power settings of interest. A total of six engine exhaust tests was
run, with 12 chamber conditions. The following factors were examined 1n the
aging experiments: the Influence of photochemical reactions in the light versus
dark chamber experiments, the effect of engine power setting, and the differences
between the engine types.
For each experiment the chambers were purged overnight with charcoal
scrubbed, filtered, shop air which was turned off prior to 0600. Chamber back-
ground measurements and SFg Injections were performed during the time from 0600
until the chambers were moved to the engine test cell to be charged with engine
exhaust. During the period that the exhaust-containing chambers were not being
sampled, the control chamber was Injected with either ambient air or a reactive
hydrocarbon/NOx mixture and monitored exclusively until the exhaust containing
chambers were repositioned for the aging experiment.
For characterization of the organic compounds in the engine exhaust, samples
of the engine exhaust were collected on quartz fiber filters and XAD-2 sorbent
436
-------
traps. For these samples the sampling line was heated to 50 C from the sample
probe to beyond the exit from the test cell. The exhaust gas sample was then
permitted to cool by contact with the walls of a stainless mixing chamber and
drawn through the filter and sorbent trap at a measured flow rate and temperature,
In this manner the aerosol sample was obtained at nearly ambient temperature.
Samples of the chambers contents were collected for later analysis at
several times during the aging experiments. These samples were obtained by with-
drawing approximately 1 of the chamber air through a quartz fiber filter and
XAD-2 sorbent trap. At the end of each aging experiment, a 5 m3 sample was with-
drawn in the same manner. Selected XAD traps were spiked immediately after
sampling to assess the losses of PAH due to sample handling. The XAD traps were
sealed and enclosed 1n aluminum foil after sample collection to protect the
collected sample from light, and the traps and filters were maintained below 0 C
on site until they were shipped to the laboratory for extraction and analysis.
The chambers were monitored at 20-minute intervals for NO, NQX, 0^, THC,
and SFg concentrations. The temperature and relative humidity of the air 1n
each chamber was monitored as well. To monitor these parameters, the air was
drawn through a switching valve which permitted air to be withdrawn from any of
the three chambers or from an ambient air intake located above the mobile analy-
tical laboratory. The sampling lines were either Teflon or glass and were main-
tained above ambient temperature to prevent condensation within the sampling
lines. Each chamber was sampled for a 5-minute interval, and the measurements
made during last 3 minutes were averaged by the data collection/system control
software. Table 1 lists the instruments which were monitored by the Apple lie-
based data collection system.
TABLE i. INSTRUMENTATION USED FOR CHAMBER ATMOSPHERE MONITORING
Parameter Instrument
Total hydrocarbons Beckman 402
Ozone Bendix Model 8002
Solar intensity Eppley UV radiometer
Oxides of nitrogen Monitor Labs Model 8440
Temperature/humidity EG&G Model 911
Chamber temperature Type K thermocouples
437
-------
Analysis of Samples for
PAH and nitro-f'AH
Certain filter and XAD-2 samples from the chambers and engine exhaust were
extracted separately with methylene chloride by Soxhlet technique. The remaining
filters were extracted together with their corresponding XAD-2 sample to maximize
the mass in the extract. Prior to extraction, a standard solution of perdeuter-
ated PAH was spiked onto selected filters and XAD-2 to determine PAH losses
attributable to sample preparation. The extracts were then concentrated by
Kuderna-Danish
-------
hydrocarbons, while alkyl benzenes, 2- to 4-ring PAH, and oxygenated PAH were
also present. Similar classes of compounds were detected in the filter samples,
although the more polar compounds dominated the mass found in these extracts.
Table 2 presents the concentrations of the PAH and nitro-PAH found in the
filter samples collected at the start and at the end of the experiment in which
exhaust from a J-79C (smokeless) engine was aged in the chamber exposed to sun-
light. These values are expressed as concentration of each compound associated
with the suspended particulate matter in the chamber. The initial values in the
chambers represent the engine exhaust diluted by approximately a factor of five.
It is readily seen that these concentrations are quite low and that the decay of
these compounds during the experiment reduces their concentration by one to two
orders of magnitude, in general.
This table also presents the PAH and nitro-PAH concentrations measured at
the start and end of the "light" aging experiment for the TF33-3 engine operated
at idle power. The idle operating power produces the most hydrocarbon laden
exhaust of the power settings examined in this study and the TF 33-3 was seen to
emit higher hydrocarbon concentrations than the other engines studied. Compari-
son of the "initial" concentrations seen here with those in Table 2 permits a
relative comparison of the TF33-3 and J-79 engines to be made. The very exten-
sive decay of the PAH compounds seen above is repeated here.
A final set of results is presented in the table for the TF33-3 engine
operation at 30 percent power. There is an obvious reduction in the exhaust
concentrations of these compounds. There also appears to be a significant les-
sening of the extent of decay of these compounds relative to those seen elsewhere
in the table. The explanation for this result requires further analysis of the
gas concentrations measured in the chambers in these experiments. Sunmary values
such as these may mask the actual dynamics of the formation and decay of the
individual compounds. Only by making comparisons across experiments and with
intermediate values of compound concentrations can one deduce the processes lead-
ing to the changes observed in tables such as these.
Conclusions
The analysis of samples obtained and the Interpretation of the experimental
results of these experiments are currently under way. The results assessed to
date permit us to conclude that the concentration of PAH and nitro-PAH associated
with particulate phase of the engine are quite low, due, no doubt, to the very
large volumes of air processed through these engines. There Is an apparent
increase in the decay rate of the PAH for the chambers exposed to sunlight,
439
-------
relative to those kept darkened, and obvious evidence of the formation of various
nitro-PAH during the experiments. Differences in the results for the different
engines and power settings have been noted at this time and will be better
defined as the analysis proceeds.
Acknowledgments
The support and direction provided by the United States Air Force Engineering
and Services Center, under Contract F08635-85-C-0122 is gratefully acknowledged
by the Oklahoma City Air Logistics Center during the performance of these experi-
ments .
440
-------
TABLE 2. INITIAL AND FINAL CONCENTRATIONS (ng/m3) OF COMPOUNDS IN AEROSOL SAMPLE
FROM CHAMBER CHARGED WITH EXHAUST FROM J79 AT IDLE, TF33-3 ENGINE AT IDLE
AND 30 PERCENT POWER
079 (Idle)
TF33
-3 (Idle)
TF33-3 (30*}
Compound
In i t i at
Final
Ratio
Initial
Final
Ratio
Initial
Final
Rat io
Polynuclear Aromatic
Hydrocarbons
Naphthalene
8.41
0.68
0.081
12.90
1.25
0.104
3.5
0.3
0.09
Dibenzothiophene
32.15
0.54
0.017
82.10
0.45
0.005
11
1.0
0.09
Phenanthrene
153.3
3.49
0.023
757.0
6.45
0.0085
304
15
0.05
Anthracene
11.62
0.56
0.048
101.0
0.41
0.0041
24
2,3
0.10
Fluoranthene
119.0
1.74
0.015
294.0
1.83
0.0062
217
4.8
0.02
Pyrene
138.5
0.98
0.007
410.0
0.97
0.0024
280
2.5
0.01
Benz[a]anthracene
3.96
0.98
0
47.30
<0.22
0.0047
14
<. 3
<.02
Chrysene
9.89
0.21
0.021
25.60
0.37
0.014
22
0.7
.03
Cy c1openta[c,d]pyrene
0.25
NO
0
41.00
—
0
47
ND
--
Benzofluoranthenes
4.70
NO
0
94.70
0.22
0.0023
22
ND
--
Benzo[e]pyrene
2.08
NO
0
34.70
0.26
0.0075
9.4
ND
—
Benzol aIpyrene
1.14
NO
0
37.90
0.22
0.005B
8.4
ND
~
IndenoLl,2.3-c,d]pyrene
2.05
MD
0
56.80
<0.22
0.0039
4.7
ND
-
BenzoEg.h.i Iperylene
5.19
NO
0
78.90
<0.22
0.0028
10
ND
--
Dibenzo[a,h]anthracene
NO
NO
0
1.36
ND
0
c.2
ND
—
Nitrated Polynuclear
Hydrocarbons
2-Nitro-l-naphthol
37.1
0.87
0.023
133.0
0.77
0.006
14
1.3
0.09
4-Hydroxy-3-nitrobiphenyl
1.86
0.31
0.17
5.37
0.12
0.022
2.8
0.25
0.09
1-Nitronaphthalene
2.10
0.0S9
0.028
2.05
0.17
0.083
2.3
0.38
0.17
Nitronaphthalene isomer
1.24
0.049
0.040
0.31
0.045
0.15
0.35
0.16
0.46
9-Nitroanthracene
0.57
0.071
0.12
0.57
0.016
0.028
3.0
0.26
0.09
9-N i trophenanthrene
0.30
0.024
0.08
0.15
0.37
2.5
0.49
0.09
0.18
Nitroanthracene/phenanthrene isomer
0.27
0.021
0.078
0.27
0.011
0.041
0.33
0.04
0.12
3-N i trof1uor ant hene
0.13
0.012
0.092
0.30
0.0071
0.024
0.54
0.57
1.06
i-Nitropyrene
0.69
0.0033
0.0048
0.15
0.0049
0.033
0.18
0.02
0.11
6-Nitrochrysene
0.07
ND
0
ND
ND
--
ND
ND
—
Di-Nitropyrene isomer
NO
ND
--
ND
ND
—
ND
ND
—
ND = Nat detected.
-------
Prediction of Photochemically Produced Formaldehyde With
Chemical Mechanisms Developed for Urban ozone Systems
Kenneth G. Sexton
Dept. of Environmental Sciences and Engineering
School of Public Health, University of North Carolina
Chapel Hill, NC 27514
Abstract
Chemical mechanisms developed for use in Air Quality Simulation
Models for photochemically produced ozone may be used as a
method to estimate formaldehyde yields from the hydrocarbon and
nitrogen oxides precursors. Two chemical mechanisms are
compared in their ability to simulate smog chamber experiments
conducted at the University of North Carolina where formaldehyde
was measured as a photochemical product, A wide range of
experimental conditions are used including simulated automobile
exhaust, simulated urban-like mixtures, and specific
hydrocarbons which are predominant in the atmosphere.
Introduction
Chemical mechanism models have been developed for simulation of
photochemistry leading to the formation of ozone in the
troposphere. Formaldehyde is an important compound in this
chemical system (1). Although successful simulation of the
ozone chemistry requires including the chemistry of
formaldehyde, the mechanisms primary development goal has been
the prediction of ozone. There is concern that these chemical
mechanisms developed for ozone, may be used to estimate
formaldehyde concentrations in the atmosphere. The chemical
mechanisms have not been evaluated as thoroughly for
formaldehyde as they have been for ozone.
There is reasonable concern that the simulation and prediction
of formaldehyde may be less successful than simulation of ozone.
This arises from several problems.
Formaldehyde is formed as a product from the oxidation of
hydrocarbons in the atmosphere. The amount depends on the type
of hydrocarbon and the type of reaction. Often many chemical
intermediates and reactions are involved. Formaldehyde also
photolyzes to yield radicals and carbon monoxide. Many details
of these processes are still uncertain or unknown. The
experimental database of formaldehyde measurements in either the
atmosphere or smog chambers is much less developed and certain
than for ozone and its primary chemical precursors. The
successful simulation of formaldehyde has been only a secondary
goal in the development of these types of chemical models,
mainly to assure the prediction of ozone. Because of limited
resources imposed on mechanism developers, development of a
chemical mechanism often stops when reasonable successful
prediction of ozone is achieved.
442
-------
The goal of this study was to perform a limited modeling program
to determine how successful two mechanisms developed for ozone
in the urban atmosphere are in simulation and prediction of
formaldehyde.
Chemical Mechanisms, Test Conditions and Hethods
Two recent chemical mechanisms were used to illustrate the
ability to simulate formaldehyde in two types of photochemical
systems: UNC Outdoor Dual Smog Chamber experimente and an
"EKMA-type" urban scenario with continuous emissions and
meteorological dilution. The new ERT/ Inc. chemical mechanism
developed by Carter, Atkinson, Lurmann and Lloyd (CALL) (2]
and the recently distributed SAI Carbon Bond IV chemical
mechanism developed by Whitten and Gery (CB4) (3) were used.
The code for the computer-based model PC-PKSS, used to implement
the chemical mechanisms and simulate the different photochemical
systems was written by Harvey Jeffries (4). It is capable of
not only simulating outdoor smog chamber-type experiments with
varying temperature and photolytic rates, but also EKMA-type
simulations with continuous emissions and meteorological
dilution (5). It reproduces EKMA-type simulations performed by
the OZIPM3 (6). The photolytic rates used are discussed in (7).
The UNC Outdoor Dual Smog Chamber has been used to study the
photochemistry responsible for ozone formation in urban systems
since 1974 (8..21). Experimental data including the
measurements of chemical precursors and products, and physical
parameters throughout the experiment have been obtained for a
wide range of conditions of interest to chemical model
developers, over 300 dual-sroog chamber experiments have been
processed and distributed (16,19). The description of the
chamber and the analytical methods are in several reports. The
formaldehyde monitor is an automated reversed West-Gaeke method
sold by CEA Instruments- At its normal operating range (0 to 1
ppm) it has a minimal detection sensitivity of -20 ppb.
Preliminary results show that it compared reasonably well with
other formaldehyde methods in this range in a Intercomparison
study conducted in 1986 (22) .
Several experiments from the UNC Smog Chamber set were selected
which study the reactivity of individual important hydrocarbons
in the atmosphere. These are also either explicitly represented
in the chemical model or are the representative compound used in
the model to represent a class of compounds. Experiments
conducted with hydrocarbon mixtures developed to be eguivalent
to autoexhaust in reactivity, or structured from ambient
measurements were also selected.
In modeling smog chamber experiments, assumptions need to be
made concerning the effect of the chamber on photolytic rates
and type and magnitude of background chamber reactivity.
Discussions of these concerns are in several reports (16,18,21).
443
-------
The chamber teflon material attenuates solar radiation which is
a function of wavelength; and the reflective chamber floor also
must be considered. The area around the chamber also needs to
be considered. Spectral irradiance measurements made inside the
chamber have been used to derive a preliminary set of photolytic
rates for the UNC chamber. These are shown in Table 1. Research
in this area is continuing. Background chamber reactivity has
been studied (1,23,24) and is also continuing. The surface to
volume ratio is lower than most smog chambers because of the
large size. Each side of the dual chamber has a volume of 150,
000 liters. Background reactivity assumed by most modelers is
much less than assumed for other chambers. Normal assumptions
include 0 to 50 ppb of formaldehyde (25 ppb average) and 0 to 25
ppb of N02 source material (10 average) on the walls which
off-gas as a function of light intensity. Also initial HONO
levels of 0 to 3 ppb are normally used: with a maximum of 5 ppb
used for experiments performed after a long series of very
reactive experiments have been conducted. The protocol followed
in modeling a set of experiments is to maintain reasonable
consistency in chamber assumptions through the entire set, using
the same assumptions for both sides of the dual chamber pair of
experiments performed on a given day. The benefit of the
dual-nature of the smog chamber is that the instruments are
time-shared and both chamber sides are prepared (vented and
dehumidified) the same. Both sides are subject to the same
light and temperature; chamber assumptions should be similar.
Even if there is a question of calibration of an analytical
method, at least the relative differences in measured chemical
reactivity should be simulated by a chemical mechanism model.
Results - Smog Chamber Comparisons
Formaldehyde is explicitly represented in both chemical
mechanisms. Four formaldehyde/NOx experiments were simulated.
The dual experiment performed on July 8, 1986 is a matched
initial NOx (-0.17 ppm) and different initial formaldehyde: 1.02
vs 0.54 ppmC. This type of experiment illustrates the benefit
of the dual-nature of the chamber design in that two levels of
reactivity (HC/NOX), with "the same chamber physical conditions
and assumptions are provided for testing. The time-
concentration profile results are shown in Figure 1. The ozone
maximum is better simulated on the 1.02 ppmC side but the
formaldehyde is better simulated on the lower 0.54 ppmC side.
This experiment was part of the formaldehyde intercomparison
study. The other two experiments July 18, 1977 and August 12,
1980 were also simulated reasonably well. These results are not
shown for space considerations.
Ethylene is an dominant hydrocarbon in the atmosphere and is
explicitly represented in each mechanism. Propylene is also
important in the atmosphere. It is the surrogate in the CALL
mechanism for terminal olefins and is the basis for the CB4
olefin chemistry. Four ethylene and two propylene experiments
conducted in three dual-chamber experiment combinations were
modeled: low HC/NOx ethylene vs propylene on July 13, 1986, low
vs high HC/NOx ethylene on July 9, 1986, and high HC/NOx
ethylene vs propylene on October 4, 1984. These cover a range
of reactivity. The two 1986 experiments were also part of the
444
-------
formaldehyde intercomparison study (22) and the formaldehyde
measurements compared well. A consistent set of chamber
background assumptions were used.
In general, the success of simulating these experiments was a
function of HC/NOx, especially for ethylene. The ozone maxima
were simulated fairly well in general but both mechanisms were
late in predicting ozone especially in the lower HC/NOx runs.
Both mechanisms underpredicted ozone for the lowest HC/NOx
ethylene experiment: one side of the July 9, 1986 experiment.
The CALL mechanism overpredicts ozone for the high HC/NOx
ethylene runs. For both mechanisms for all experiments the
hydrocarbon predictions were good to excellent. Formaldehyde
predictions were best for the high HC/NOx experiments for both
mechanisms although both mechanisms underpredicted (15 to 25%
CB4 and 20 to 30% CALL). See Figure 2. For the lower HC/NOx
experiments both mechanisms underpredicted about 45% (see Figure
3) . That is, neither mechanism simulated the relative change in
formaldehyde produced in the delta HC/NOx ethylene experiments
of July 9, 1986. It should be pointed out that the model
simulations of the formaldehyde concentration profiles of the
July 8, 1986 experiment shown in Figure 1 were much more
successful. The formaldehyde monitor was checked for span
calibration at least twice a day. It was very stable during the
entire intercomparison study.
Toluene and ra-xylene are the basis for these two classes of
aromatics for both mechanisms. A dual experiment conducted June
27, 1984 compares matched initial NOx conditions with 2 ppmC
m-xylene on one side and 5 ppmC Toluene on the other. Both
mechanisms simulate the m-xylene NOx, 03 and xylene
concentration- time profiles very well, with the CALL mechanism
just a little slower in reactivity passed the midpoint of the
experiment. Both mechanisms overpredict ozone on the toluene
side (-25% CB4, -33% CALL), but the CB4 mechanism reproduces the
proper shape. While the CB4 mechanism simulates the toluene
profile, the CALL mechanism does not oxidize enough. The CALL
mechanism predicts the measure amount of formaldehyde on the
toluene side and overpredicts about 40% on the xylene side. The
CB4 mechanism overpredicts by a factor of 2 for both systems.
The differences between the chemical mechanism predictions is
from the different representations of the still uncertain
aromatics chemistry. The CB4 mechanism is still under
development with a new release projected for September 1987.
Changes to the toluene chemistry representation are currently
being made (25).
Four hydrocarbon mixtures were used for testing. Mixtures are
good for testing a chemical mechanisms ability to respond
correctly to composition variation. On August 31, 1981, two
matched initial NOx and hydrocarbon concentration conditions
with two different compositions were compared. These are two
olefin/paraffin compositions of different complexity. These
mixtures contain no aromatics. Two experiments with two
mixtures designed to be reasonable approximations of autoexhaust
445
-------
(SYNAUTO) and an urban atmosphere (SYNURBAN) were also modeled.
The design of these mixtures are described in detail (18,20). The
SYNAUTO has been shown to be reasonably equivalent to the
reactivity of real autoexhaust (19).
Both mechanisms underpredict the ozone and formaldehyde maxima
of the SIMMIX side by 20*. CB4 simulates the UNCMIX ozone to
within 10% while the CALL mechanism underpredicts by 2 5%. Both
mechanisms however underpredict the formaldehyde concentration
by 45 to 50%. The SYNAUTO experiment of August 8, 1984 is
simulated relatively well for both ozone and formaldehyde with
both mechanisms with CB4 overpredicting the measured
formaldehyde by -10% and the CALL mechanism underpredicting by
-12% (see Figure 4). The SYNURBAN experiment of August 22, 1984
is overpredicted for ozone by both mechanisms with the CALL
mechanism predicting ozone better although it is too reactive
initially. The formaldehyde is underpredicted by 26 (CB4) to 35
(CALL) percent (see Figure 5).
Results - Urban Simulation
An "EKMA-type" urban simulation is the example 1 in the
"Guidelines for OZIPM3" document (6). This example was
selected for the comparison because it is published and
demonstrates the ability to reproduce the photochemical
simulation. This is a simple trajectory model with initial
chemical precursors including formaldehyde, emissions along the
trajectory path, and with meterological "characteristic-curve"
dilution. The conditions are listed in Table 3.
The original example 1 simulation utilized a default composition
determined by the Carbon Bond mechanism developers. Since then
a new database from 41 cities and rural sites has been used to
determine new default composition for CB4 (26). The same
database was used to determine a default composition by the
developers of the CALL mechanism for urban simulations with the
CALL mechanism (2). Speciktion is different for each mechanism,
and some interpretation is necessary: the developers of CALL
assume 3% formaldehyde in the emissions composition and 2.2% is
used for the CB4. For the entrained aloft hydrocarbon
composition assumptions: 5% formaldehyde is used for CALL and 6%
is used for CB4. These default compositions were used with each
mechanism.
The results of these simulations with both mechanisms for five
levels of control are shown in Figure 6. The agreement between
mechanisms is better for ozone than for formaldehyde, although
the shapes of the ozone profiles are apparent. The absolute
predictions of formaldehyde are shown in Figure 7. These final
concentrations for both pre-control and 100% control are in
agreement with published ambient measurements (27). The
relative effect of control is also shown. This is also
different.
446
-------
Shown in Figure 8 are the production and loss flowcharts for
formaldehyde for the two mechanisms. These are computed from
the integrated rates giving total mass through each of the
different reactions. The reactions shown are often incomplete
for space considerations. The formaldehyde yields from certain
reactions such as the oxidation of principal hydrocarbons is
misleading also since other intermediates produced in the same
reaction go on to produce formaldehyde. The main reason for
showing these flowcharts is for showing the differences in the
representation of the the chemistry in the two mechanisms and to
compute the relative sources of formaldehyde between direct
input and chemical production. Notice that the direct injection
source accounts for 35 to 40 percent of the formaldehyde; only
10 to 13% are from direct emissions and entrainment of aloft
formaldehyde. More than half of the formaldehyde in this
particular urban system is photochemically produced during the
day.
Conclusions and Recommendations
In general, these chemical mechanisms simulate 03 better than
they simulate formaldehyde. Although the simulations were good
for the synthetic autoexhaust and perhaps passable for the
synthetic urban mixtures, the fact that different mechanisms
show different success for simulating the formaldehyde yield
from different important individual hydrocarbons suggest that
the general success with these mixtures is a result of
fortuitous averaging. Other reasonable hydrocarbon mixture
compositions not yet tested may be even less successful. These
mechanisms have not been evaluated for formaldehyde as
thoroughly as they have been for ozone. The formaldehyde/smog
chamber database is not as good or extensive as the general
ozone/smog chamber database. Research is needed to better
understand the aromatics chemistry. As these new kinetic details
are incorporated into the chemical mechanisms, formaldehyde
predictions will become more successful over a wider range of
conditions.
The atmospheric modeling examples illustrate some of the
problems and issues that will arise from different modelers
using different mechanisms and different input assumptions.
Predictions of formaldehyde and control strategies for the
atmosphere are subject to the same uncertainties of the
underlying experimental database used to develop and test
chemical mechanisms.
References
(1) Whitten, G.Z., "The Chemistry Of Smog Formation: A Review Of
Current Knowledge," Environ. Int., V9, pp447-463, 1983.
(2) Lurmann, F.W., W.P.L. Carter, L.A. Coyner, "A Surrogate
Species Chemical Reaction Mechanism For Urban-Scale Air Quality
Simulation Models", Vol I - Adaptation of the Mechanism, Vol II
- Guidelines For Using The Mechanism, ERT Inc., Newbury Park,
CA, and Statewide Air Pollution Research Center, Univ. of
California- Riverside, Office of Research and Development, EPA,
Research Triangle Park, NC 27711.
447
-------
(3) Whitten, G.Z., M. Gery, "Development of CBM-X Mechanisms for
Urban and Regional AQSMs," EPA-600/3-86-012, November 1985.
(4) Jeffries, H.E. , "User's Guide to Photochemical Kinetics
Simulation System PC-PKSS," Dept. of Environmental Sciences and
Engineering, Univ. of North Carolina, Chapel Hill, January, 1987.
(5) Jeffries, H.E., "Procedures and Programs for Calculating
Inputs to PC-PKSS For EKMA-type Simulations", Dept. of
Environmental Sciences and Engineering, Univ. of North Carolina
Chapel Hill, January, 1987.
(6) Hogo, H., Whitten, G., "Guidelines For Using OZIPM3 With
CBX Or Optional Mechanism", Systems Applications, Inc., San
Rafael, CA, January 1986.
(7) Jeffries, H.E., K.G. Sexton, Technical Discussion Related
to the Choice of Photolytic Rates for Carbon Bond Mechanisms in
0ZIPM4/EKMA, Dept. of Environmental Sciences and Engineering,
School of Public Health, University of North Carolina, Chapel
Hill, NC, Prepared For Monitoring and Data Analysis Div., Office
of Air Quality Planning and Standards, Environmental Protection
Agency, Research Triangle Park, NC, EPA-450/4-87-003, February
1987.
(8) Jeffries, H.E., Fox, D., Kamens, R., "Outdoor Smog Chamber
Studies," EPA-650/3-7 5—Oil, EPA, 1975.
(9) Jeffries, H.E. , Fox, D., Kamens, R., Photochemical
Conversion Of NO To N02 By Hydrocarbons In An Outdoor Chamber,
J.A.P.C.A., V26, P480, 1976.
(10) Jeffries, H.E., Fox, D., Kamens, R., "Outdoor Smog Chamber
Studies: Light Effects Relative To Indoor Chambers, ES& T, V10,
pi006, 1976.
(11) Kamens, R., Jeffries, H.E., Fox, D., Alexander, L. , "A Smog
Chamber Study Of The Potential Effects Of Emission Control
Strategies On Nighttime N02," Atmos. Envr., Vll, p225, 1977.
(12) Jeffries, H.E., Kamens, R., Fox, D., Dimitriades, B.,
"Outdoor Smog Chamber Studies: Effect Of Diurnal Light, Dilution,
And Continuous Emission On Oxidant Precursor Relationships,"
Proceedings, International Conf, Photochemical Oxidant, V2,
EPA, EPA-600/3-77-001B, 1977.
(13) Jeffries, H.E., Sexton, K.G., "Modeling Aspects Of Nitrogen
Oxides Using Smog Chamber Data, EPA-600/9-81-025, Formation And
Fate Of Atmospheric Nitrates," pp49-P95, June, 1981.
(14) Kamens, R.M., Jeffries, H.E., Gery, M., Wiener, R., Sexton,
K.G., Howe, G., "The Impact Of a-pinene On Urban Smog Formation:
An Outdoor Smog Chamber Study," Atmos. Envr., V15, pp969-981,
1981.
(15) KAMENS, R.M., JEFFRIES, H.E., SEXTON, K.G., WIENER, R., "The
Impact Of Day-Old Smog On Fresh Smog Systems," Atmos. Envr., V16,
ppl027-1034, 1982.
448
-------
(16) Jeffries, H.E., Kamens, R.M., Sexton, K.G., Gerhardt, A.A.,
"Outdoor Smog Chamber Experiments To Test Photochemical Models "
EPA-600/3-82-016A, EPA, 1982. '
(17) Kamens, R.M., Jeffries, H.E., Sexton, K.G., Gerhardt, A.A.,
"Smog Chamber Experiments To Test Oxidant Related Control
Strategy Issues," EPA-600/3-82-04, EPA, July, 1982.
(18) Jeffries, H., Sexton, K.G., Kamens, R.M., Holleman, M.,
"Outdoor Smog Chamber Experiments To Test Photochemical Models:
Phase II," EPA, EPA-600/3-85/029, March, 1985.
(19) Jeffries, H.E., Sexton, K.G., Morris, T., Jackson, M.,
Goodman, R., Kamens, R.M., Holleman, M., "Outdoor Smog Chamber
Experiments Using Automobile Exhaust," EPA, EPA-600/3-85/032,
March, 1985.
(20) Jeffries, H.E., Sexton, K.G., Holleman, M., "Outdoor Smog
Chamber Experiments: Reactivity Of Methanol Exhaust," EPA,
September, 1985, EPA 460/3-85-009A.
(21) Sexton, K.G., H.E. Jeffries, J.R. Arnold, T.L. Kale, R.M.
Kamens, "Validation Data for Photochemical Mechanisms", EPA,
December, 1986.
(22) Jeffries, H.E., K.G. Sexton, J. Arnold, "Initial Analysis
of Formaldehyde Intercomparison Study at UNC Outdoor Chamber",
November 26, 1986.
(23) Jeffries, H.E., K.G. Sexton, "Smog Chamber Background
Reactivity, Parti: The Influence of NOx from Chamber Walls on
Kinetic Computer Models for Air Pollution", March 1985.
(24) Jeffries, H.E., K.G. Sexton, "Smog Chamber Background
Reactivity, Parti: The Influence of Radicals from Chamber Walls
on Kinetic Computer Models for Air Pollution", March 1985.
(25) Gery, M., personal communication April 1987, Systems
Applications, inc., San Rafael, CA.
(26) Baugues, K., "A Review of NMOC, NOx and NMOC/NOx Ratios
Measured In 1984 And 1985", Office of Air Quality Planning And
Standards, Environmental Protection Agency, Research Triangle
Park, NC, EPA-450/4-86-015.
(27) Turoski, Victor, Editor, Formaldehyde, Analytical
Chemistry and Toxicology, Chapter 17, Exposure to Formaldehyde,
Advances in Chemistry Series 210, American Chemical Society,
Washington, D.C. 1985.
449
-------
Table l. Photolytic Rates for UNC Smog Chamber
UNC Chamber Rates, ES=Aug,A=0.08,Rho=0.4,Tau=0. 7S--0.45,Sky=0.8
ZA 0 10 20 30 40 50 60 70 78 86
N02 LI
0.5731 0.5661 0.5427 0.5045 0.4500 0.3788 0.2887 0.1788 0.0830 0.0153
HCHOR X 1000 L2
1.6522 1.6176 1.5093 1.3364 1.1043 0.8282 0.5297 0.2517 0.0916 0.0227
HCHOS X 1000 L3
2.1693 2.1345 2.0221 1.8398 1.5856 1.2655 0.8874 0.4835 0.2030 0.0547
01D X 1000 L4
1.6878 1.6243 1.4372 1.1576 0.8259 0.4979 0.2294 0.0657 0.0131 0.0016
CCHO X 1000 L5
0.1973 0.1917 0.1747 0.1484 0.1150 0.0786 0.0438 0.0169 0.0048 0.0009
Table 2. Conditions for Urban Simulation Example 1.
Latitude: 39.9 Longitude: 75.1 Date: 6/24/80
Mixing Heights: 250m at 0800, 1235m at 1500, Characteristic Curve
03 aloft: 0.07 ppm, NMOC aloft: 0.04 ppmC, NMOC surface: 0.038
Initial Values: NMOC =1.1 ppm, NOx =0.12 ppm
Future NOx: No change
Emission Fractions of Initial Values:
Hour 8 9 10 11 12 13 14 15
NMOC 0.17 0.17 0.17 0.10 0.02 0.02 0.02 0.02
NOX 0.35 0.35 0.35 0.19 0.03 0.03 0.03 0.07
450
-------
July 8. 1986 (B) 1.02 ppmC HCHO
Vermont* Ot/K/C «***-«1
II*. MMUKS
HOUft. ~ f -*1 «
July 8. 1986 (B) 1.02 ppmC HCHO
r on at 0*/21/»7 t 2 *7 *9
Figure la. Simulations of July 8, 1986 1.02 ppmC HCHO.
-------
Ji:l> i. 1986 {«) 0.53/ ppmC HCHU
P'"C HOO
;o>ton Hixx) fc*s o*/22/^1 i 2 *e
July 3. 1986 (R) 0.537 ppmC HCHO
Cv*v. Uprt. Ltfmgn. CvW H»*«» 4 D*/W*' HUB
July 8, 1986 (R) 0.5J7 ppmC hChO
Cqr^orv Qond Cv^tx1 -31 0*/22/t 7 1 3 *4 00
Figure lb. Siaulatlons of July 8, 1986 0.54 ppmC hcho.
-------
October 11. 19B4 (R) 2 856 ppmC ETHYLENE
CwMT *>l*wiwi, LJo-»W Cho-t*er V«r»46 ppmC ETHtL
C"W«Or tJ "ictrvtwr- ft'SO'i O! L ? Z' A
October 11, 19S4 (R) 2.856 ppmC ETHYLENE
Cfl*. AlWw Ub)i). CWW ««rw < /?•' 13712V
October 11, 1934 (R) 2 056 ppmC ETHYLENE
Cerfea* Be*d ?<**. Che^ber «er»on «j» 04yi*/^T 112* Jfr
rap cura
Figure 2. Simulations of October 11, 19B6 (High Ethylene).
-------
VJ. !0H6 (R) O.fVJ ppmC Ethylene
l*vJ *f*o* H Q*/2S/%7 4T
July 9. 1086 (R) 0.89 ppmC Ethylene
"rl Four. Cluita' Wikwi of 04/23/1' 47
TMC. wn/CS
itooa * as cmt4
¦fc.
July 9, 1936 (R) 0.89 pprriC Ethylene
CoMt» i^noA. Uovd. Uww. CSy^tr.' 4 O*/2J/0? 1 iJ4 17
July 9, 1986 (R) 0.89 ppmC Ethylene
Cwtaft Band F«ur, ClV^ Wfto> at 0«/M/»T 1127 47
0 • -{
O • -
ft n -<
©• -
o y ¦
O.T -
I a*
»
J 0 s -
L
o ^
3.
O
* ...
I
» a.-
<> * -
^
03 -
0 7
0 3 -
a o ->
""1 . . t~ r- ¦ | ^
01 -
OO -
TMC. WWLfltS
MOML ~ r*P CM1«
Figure 3. Simulations of July 9, 1986 (Low Ethylene)
-------
August 3. 1934 3.7 ppmC SYNAUTO (BLUE)
CutNn (kH c«w+«r Werwx * o«/iI J 28 * J
August 8. 1984 3 7 ppmC SYNAIJTO (BLIJE)
Cw«f Altrw Ugy4
•* o*n*/f < j jo »
KJ%
August 8. 1984 3.7 ppmC SYNAUTO (BLUE)
C«ten Send Foi*, C*v*r ' at 0*/u/l* O
August 8. 1984 3.7 ppmC SYMAUTO (BLUE)
Cortv. t#wv Uoyd. Lumw, Owrter * 0*/ • */9' « J 7
4 UP Ml*
Figure 4. Simulations of synthetic Autoexhaust.
-------
22. 1034 .5.2 ppmc SYNURBAN (RED)
'>!-. Lkyy4, Lu»*»»y • »fr»cw m 1«SiM
'"1
3 1 1
O' J
" i
1
4
3. i
/ 1
/ /
i
-i
//
/ /
02 -i
!
0 ' 1 ^
I^Ol/ /
jxL/
August 22. 1984 3.2 ppmC SYNURBAN (RED)
I'd L^o". Oryftt' of 04/'2/fl7 <4S3 22
0. j
0 1 -I
L.
mo
wtm
*no
urn
• 11» |4tt
Aug 22. 1984 3.1 ppmC SYNURBAN (RED)
Canw» hm) cim'w »>i«a ot o*/i j/n^ too* aA
T«*C. MX
«oocx * rr® cmta
Aug 22. 1984 3 1 ppmC SYNURBAN (RED)
Coton Bon) CVr«b«» ot W/1J/87 (00*00
..J
MO
MQOCL
400
fvP DtU
Figure 5. Simulations of synthetic Urban Atmosphere.
-------
0Z1PM Run $\, CALL MECHANISM
ViTH CwSSMnS *HD MjUHOH
o » -
CZli'M Kor. §\, CB4 MECHANISM
with ewrssiown «4D U-jncN
OZtPM Run #1, CALL MECHANISM
MIM EW9MNS mm W-0*0"
DW
4M4
dm a
PODS
OZIPM Run #1, CB4 MECHANISM
WITH CWSBONS **0 CCUttON
OOQ>
D 03 C
0 034
0 001
oaro
ooi*
j
DO" 4
~ oi-j
~ 010
0 000
ODD*
oau
ocua
IW. W
-------
Formaldehyde Control vs HC Control
om ita#i
10
9
e
7
6
5
4
J
:
i
o
2D
<0
0
m
100
rnnOCMBON CONTROL, PEKXMT
D CB4 HCHNtU » QUI NEtMNG*
Formaldehyde Control vs HC Control
OZF* ft* /)
ft
Z
t
u
3
e
t
u
9
J
8
i
0 20 «C K> *0 100
HTMOCAKM C0H1ML. KMXH1
a CB4 HBWH6II * CM1NEOttMH
Figure 7. Pomald«hyd« Control vs HC Control
eo -
-------
OZIPM Run #1, CB4 MECHANISM
<_:64 HCHO PRODUCTION PATHWAYS-
0«-0lE-LVU5 HCHO
OH+OLE—HCHO
O^OLT-O 66 HCHO
O + ETH-HCHCI
tiH + ETH-1 56 HCHO
03+ETH-HCHO
0H+CH4-HCH0
TQL+OH-1.13 HCHO
XYL+OH-0,67 HCHO
0 00*
9 22*
1 07*
fj 02*
10 16%
CHCh
2.57*
6.02*
5.8**
CHtU PROD
64.44*
TOTAL
100.00*
1 33* POHO-HCHO
0.65* RCHC+H02-20 HCHO
27 56* RC03-«-N0—HCHO
0.05* HC03*- RC05—2,0 HCHO
DOS* X02+flC03-'HCHO
6 35*
369*
2552*
DIRECT INPUT
35.56*
EMISSIONS
ENTRAIN WENT
INITIAL HCHO
WfTH EMISSIONS AND DLUTTON
OZIPM Run #1, CALL MECHANISM
CALL. HCHO PRODUCTION PATHWAYS
O+PRPC - 0.4 HCHO
OH+PRPE — HCHO
03+PRFE - O M HCHO
N03 +¦ PfiPE - HCHO
03 + T9LTT -0 3 HCHO
0 +-ETH - HCHO
OH+ETH - 1 56 HCHO
OJ-*€TH - HCHO
N03 + ETH - 2 HCHO
0H+CH4 - HCHO
ALK4 + OH - O 19 HCHO i.JJ*
ALK7 + OH - 0 02 HCHO 0.31*
ALKN + OH - 0 16 HCHO 0 1 1*
CHE 14 PROD
59 37*
0.03*
1 22*
11.68*
0 90*
1.22*
0 1 1*
0.01*
0.06*
.25*
0.01*
27.99*
0.19*
9.92*
D.57*
0.6OK
O 77*
0.00*
0 DO*
10 1 /*
27.33*
DIRECT INPUT
40,63*
TOTAL
100.00*
ALD2 - HCHO
MEK + OH - 0.5 HCHO
ACET - HCHO
GLVX - 0 13 HCHO
MC03 «¦ NO - HCHO
UC03 + PC03 - HCHO
WC03 + R02 - HCHO
M003 + H02 - HCHO
EMSSIONS
ENTRAlNuEffT
INITIAL HCHO
~r
WITH EMISSIONS and dcution
Figure 8a. Formaldehyde Production Flowcharts
459
-------
021PM Run #1, CB4 MECHANISM
CB4 HCHO LOSS PATHWAYS
HiJIU + 0-OH»-HG2"":0 HCH0+N03-HN03+H02+C0 HCH0+0H-H02+CO
0(113 OOK 41 ibr.
HCHCl
36.56X 21 £5K
HCHO—CO HCHO-2 OOH02+CO
TOTAL
100.00*
~l 1
WITH EMISSIONS AND OLimON
OZIPM Run #1, CALL MECHANISM
CALL: HCHO LOSS PATHWAYS
HCH0+H0Z-RQ2R + R02 HCH0+N03—HNOJ++<02+C0 HCH0+0H-H02+C0
J52X 0.Q1X 41,10*
HCHO
3479* 20 59*
HCHO—CO HCHO-2 00H02+C0
TOTAL
100.00*
—( . , 1 t r ~i i I i_
WfTH EMISSIONS AND Di-UTION
Figure 8b. Formaldehyde Loss Flowcharts
460
-------
THE INHIBITION OF PHOTOCHEMICAL SMOG—X.
MODEL CALCULATIONS OF THE EFFECT OF
(C2H5>2NOH ON NOx-C2H4-C4Hg-2 ATMOSPHERES
Julian Heicklen
Department of Chemistry and
Environmental Resources Research Institute
The Pennsylvania State University
University Park, PA 16802
Model calculations are made on the effect of (C2H5)2NOH on
NOx~c2H4~c4H8_2 atmospheres. The critical initial DEHA concentration
needed to keep the maximum O3 and PAN concentrations <1 ppb is 100-200
ppb for slightly or moderately smoggy dayB. It is independent of [NO]o,
[H20]o and nearly independent of [N02)o/[NOX](). It increases dramatically
with dilution rate and increases linearly with hydrocarbon concentration.
The cost for using DEHA as an anti-smog agent would be about $1 per car
per day used.
461
-------
Introduction
In a previous paper,! j have modelled a C2H4-NO atmosphere to test
the inhibiting effect of diethylhydroxylamine (DEHA). The mechanism used
reproduced well the data of Schaal et al.2 In the present paper, the
calculations are extended to see what critical initial concentrations of DEHA,
[DEHA]C, are needed to keep the maximum 03, [03]TOax, and PAN, [PAN]max>
concentrations to <1 ppb for a 12-hour run under typical Los Angeles
sunny-day conditions (sun declinations of 10.53' and latitude of 34*) for a
variety of reaction conditions.
Most of the model runs have been done with C2H4 as the hydrocarbon.
The mechanism used was that used earlier1 with some updating of rate
coefficients* Some calculations were also done with 2-C4H8 as the
hydrocarbon. The rate coefficients were drawn from the same sources as
before with the updating from Finlayson-Pitts and Pitts.®
The updated and added reactions and their rate coefficients are given
in Table I. An explanation of the parameters in Table I is given in my
earlier paper. 1 As before NO, NO2, and H2O are assumed to be in a rapid
heterogeneous equilibrium with HONO. The HONO photodissociation acts as a
homogeneous HO radical source. There are 135 reactions and 60 reaction
species plus N2> which is not reactive but acts as a chape rone.
Results
The results of the model calculations are given in Figures. 1~5. Figure
1 shows the effect of initial NO concentration on a C2H4-NO-H2O atmosphere
with a dilution rate of 7% per hour. The diluent air contains 10* ppm H2O
and 25 ppb 03. In the absence of DEHA, the maximum O3 concentration,
fOsl^nv. goes through a maximum value of 181 ppb as the initial NO
concentration, [NO]q, increases. The critical initial DEHA concentration,
[DEHA]C, needed to keep [03]max and [PANJmax *1 PPb remains unchanged
at 120 ppb, independent of [NOJo-
Figure 2 shows the effect the initial ratio of [NO2] to [NOx] at constant
total NOx concentration. There is no effect on [03Jmax which is <*170 ppb,
but [I>EHA}C increases from 120 to 180 ppb a a [N02)o/l^0x}o increases from
0 to 0,50. This occurs because the NO2 can oxidize some of the DEHA by
generating odd oxygen (0 or O3).
The effect of water vapor concentration is given in Figure 3. Neither
(°3]max nor [DEHA]C is significantly affected by water vapor composition,
as long as it is above 0.3% bo that HONO is present to generate HO radicals.
In Figure 4 we see the effect of dilution rate, which is significant.
With no dilution [Oftlmn* in the absence of DEHA is 287 ppb, but it drops
rapidly with increasing dilution rate, reaching the U.S. federal standard of
120 ppb at a dilution rate of 14%/hour. On the other hand, [DEHA]C rises
sharply from 100 ppb for stagnant air to 165 ppb at a dilution rate of
14%/hour.
The effect of hydrocarbon concentration is shown in Figure 5. Three
series of runs are depicted there. In one C2H4 is the hydrocarbon and
462
-------
[C2H4]0/rNO]o is kept constant at 2.25. Both [0a]mav In the absence of
DEHA and [DEHA]C increase linearly with [C2H4lo with [03lmax being 75 ppb
greater than [DEHA]C. Two series of runs were done with 2-C4H8 as the
hydrocarbon. One of these also contained C2H4. The results are plotted
vs [C2H4lo + 7.46 [2-C4H8J0 because the HO radical reaction is 7.46 times
faster with 2-C4H8 than with C2H4. Since the DEHA scavenges HO radical
it was anticipated that the HO-radical reactivity is the critical parameter in
determining [DEHA]c- For the run with only 2-C4H8 as the hydrocarbon,
the ratio [2-C4H8]0/[NO]0 waa kept constant at 0.56. [03]max in the
absence of DEHA was much lower than in the C2H4 runs, but still exceeded
the U.S. federal standard of 120 ppb in all runs. It increased with
hydrocarbon concentration. Similar results were observed with the series
containing both C2H4 and 2-C4H8. [DEHA]C increased Linearly with
hydrocarbon concentration, but [DEHAJC was lower for comparable HO
reactivity than for the runs containing only C2H4.
Conclusions
In summary about 100-400 ppb of DEHA are needed initially to
completely suppress photochemical smog production, at least for C2H4 and
2-C4H8 as the hydrocarbons. The amount needed is independent of [N0)o,
[H20]o, and is nearly independent of [N02]o/[NOx](j. It increases
dramatically with dilution rate and increases linearly with hydrocarbon
concentration. For slightly or moderately smoggy days where [03]nax 1b
less than 200 or 250 ppb, [DEHA]C = 100-200 ppb. Only for very heavy
hydrocarbon loadings, where the effective equivalent C2H4 concentration
exceeds 1 ppm, does [DEHA]C become larger than this.
With the assumption that about 200 ppb of DEHA are needed to prevent
a typical smoggy day, we can estimate the cost of such a program.
Typically the morning inversion height at the beginning of a smoggy day iB
100 meters. Thus the amount of DEHA needed is 161 pounds per km2. At a
cost of about $3 per pound for the DEHA and its distribution, the cost
becomes $483 per km^. Tjnpically urban centers have a motor vehicle
density of about 500 per km^. The cost for using DEHA as an anti-smog
agent becomes about $1 per car per day used. This can be compared to
the cost of pollution control devices which exceed $100 per car per year
including the price of the device, annual inspections, repairs, regulation,
and the use of unleaded gasoline.
References
1. J, Heicklen, "The inhibition of photochemical snog—IX. Model calcula-
tions of the effect of (C2H5)2M0H on Cgf^-HOx," Atmos. Environ.. 19:
597 (1985).
2. D. Schaal, K. Partymiller, and J. Heicklen, "The inhibition of photo-
chemical smog—VII. Inhibition by diethylhydroxylamine at atmospheric
concentrations," Sci. Total Environ.. 9; 209 (1978).
3. B. J. Finlayson-Pitts and J. N. Pitts, Jr., Ataosoheric Chemistry.
Wiley Interscience, New York, (1986)%
463
-------
Table I: Rate coefficient parameters for updated and additional rate
coefficients*
RCACTTON
03 + HN0 • HCa * NO
i. Of-19 0
0
HO ~ C2H4 m H0C3H4
3. 191-13 0
-411
HO + CH30 - H30 ~ HCO
*01-1 a o
0
0 * C8H4 • CHQ ~ HC0
0. 77C-H 0
§40
03 * CH30 - NOa ~ CH30
¦SB-14 0
13*3
NO ~ CH30 • CH30N0
1.4AI-1! 0
0
NO ~ CH30 » MNO * CH2D
¦.31-13 0
0
NOS ~ CH30 ¦ CH30N03
1, tlf-U 0
0
NQ3 * CK30 ¦ HNOS • CM9B
s 91-13 0
0
CH3CHQ • CH4 * CO
0 3K-4 0
0
0.3000f-04 0.3000f-0fc 0 80001-0* 0. 10001-04
0. 1000f-04 0. IOOOf-64 0.0000**00 <
03 * C3H4 - CH20 * HC03*
1. 31-14 0
3433
Q ~ CH30 ¦ HO * HCO
aac-ii 0
0
HO • 3-C4H# p »-H0C4H§
t.oa-u 0
-94t
3-H0C4HS -03 • 3-MOC4MSQ3
l.OI-tl 0
0
3-H0C4Hf02 • NO « *03 * CH3CH0 •» CH3CH0H
7 ot-ia 0
0
03 * C HOC HON ¦ ClOCHQ • H03
1,01-11 0
0
H03 * 3-HOC4HKJ3 • MOC4HaO&tf
7,71-14 0
0
HOC4HI03H - HSO * 3-C4Hi * 03
1, OK-19 0
0
0 * CSH4 ¦ H ~ CH3CH0
0 4M-II 0
•40
CH3CH0 • CH30 • HCO
I. Otis 0
0
0 •* a-C+Hi • mODUCTI
3.31-11 0
-10
CH903* » CH303
0, 401-13 0
0
CHS03* • COS * H3
0. 13C-12 0
0
CHS03* - CO - H30
o*af-ia 0
0
CHSoa* ¦ n ~ mcoa
0-Q4C*19 0
0
03 * HCD3 ¦ HQS * CO
». Of-tt 0
0
03 - 3-C4H* - CH3CH0 * CH3CH03*
* oac-19 0
113*
CH3CHG3* * CK3CH03
0.40C"ta 0
0
CH3CM0a« • CMS * CO * HO
0. m-ia 0
0
CH3CHOJ# • CH3 » COS * H
0 34K-I3 0
0
CM3CH03* • HCO ~ CH3
0. 091-13 0
0
CM3CH03* • CN4 * CO
0 131-13 0
0
CM303 • NO " CH20 * NO?
7. Of-13 0
0
CH30? * (103 • CH30 • *03
7 Of-13 0
0
CH3CH03 ~ NO - CH3CH0 * M03
7 Of-13 0
0
CH3CH03 * N03 • CH3CH0 • *03
7 Of-13 0
0
HQ * CH30N0 ¦ HMO? * CH30
4.01-13 0
0
HO - CN3QN0 * H30 * CHffO * NO
4.01*13 0
0
HO * CH30N03 * HMD3 * CH30
4 9f-t3 0
0
HO • CH30N03 • H30 * CH30 * NQ3
4. Of-13 0
0
CoeM4
0. ooooc+oo
0. OOOOC+0O 0.
*See Heicklen^ for a list of other reactions and descriptions of the
various parameters.
320r
260 -
240 -
200 -
5 160 —
DC
I-
o 120 -
z
o
u
00 -
40 -
CCfHUlo ¦ 400ppb
CHjOl • I0*ppm
DILUTION ¦ 7 * / hr
° £ Of ]MI
• [0EHA1C
40
80
_L
120
ISO
-L
_L
200 240 280 520 MO 400
[N0]o, ppb
Fig. 1; Plots of concentration vs initial NO concentration for a 12-hour
run at 294K and 1 atm total pressure for a sun declination of
10.53* and a latitude of 34*. Initial concentrations at 6:00 a.m* of
C2H4 and K2O are 400 ppb and 104 ppm, respectively. The
dilution factor ia 7% per hour with air containing the same
composition of H^O vapor and O3 at 25 ppb: 0, fOaU-r in
absence of DEHA; 4, [DEHA]C at 6:00 a.m. to keep fOal—w and
[ PAN ] max ^1 PPb.
464
-------
320
260
240
200
2
o
H
<
tie
K
ac
Ui
u
z
o
o
160
I20<
80
40
O
CCtH<3f 400 ppb
CNOiJo" 69 ppb
CH,0] i I04ppm
DILUTION- 7 %/hr
X
X
1
o c 0,3 „„
• CDEMA]c
I L
0.1
02
03 04 09
CNOt30/ CNO^o
06
ar
0.6
Pig. 2: Plots of concentration vi initial fraction of N02 in the NOx for a
12-hour run at 294K and 1 atm total pressure for a sun
declination of 10.63* and a latitude of 34*. Initial concentrations
at 6:00 a.m. of C2H4, NOx> and H2O are 400 ppb, 89 ppb» and 104
ppm, respectively. The dilution factor is 7% per hour with air
containing the same composition of water vapor and O3 at 25 ppb:
0, [C>3]max in absence of DEHA; •, [DEHA)C at 6:00 a.m. to keep
[°3]max and [PANJqux <1 ppb.
320
880
240
200
100
120
B0
40
1 1 I 1
—
•t 1 r
~
0 0
0
0
•
• •_
•
C ¦ 400 ppb
CNOlo • IT8 ppb
O co,]„.
DILUTION. 7 %/lw
I I —. u L
• C0EHA3,
1 1... I
04
08
1.2
1.6
W»0 J,
2.0 24 2,8 32
Fig. 3: Plots of concentration vs percent H2O in air for a 12-hour run at
294K and 1 atm total pressure tor a sun declination of 10.53* and
a latitude of 34*. Initial concentration® at 6:00 a.m. of O2H4 and
NO are 400 ppb and 178 ppb, respectively. The dilution factor is
7% per hour with air containing the same composition of water
vapor and 03 at 26 ppb: 0, tps^max In absence of DBHA; •,
[DBEAlc at 6:00 a.m. to keep (Oalm.* and [PANImax <1 Ppb.
465
-------
320 r
280T-
240 -
I-
<
a
i-
o
u
120
80
40 -
CC2H,V 400 ppb
[KIODo ¦ 178 ppb
C H,OJ ¦ I04 ppm
J_ J I
• CDEHA]c
S 12 16 20 24
DILUTION RATE, * / hf
28
32
Fig. 4: Plots of concentration vb dilution rate for a 12-hour run at 294K
and X atm total pressure for a sun declination of 10.53* and a
latitude of 34*. Initial concentrations at 6:00 a.m. of C2H4, HO,
and H2O are 400 ppb, 178 ppb, and 10^ ppm, respectively.
Dilution is with air of the same water composition and O3 at 25
ppb: O, [03lmax in the absence of DEHA; •, [DEHA]C at 6:00 a.m.
to keep [03]max and tPANJmax <1 PPb.
TOO
T~
COjJ.o. tDEHA],
eoo
a 500
b
a
z*
400
O
t-
<
E
t-
Z
u
V
z
300
o £00
o
100
CHgO] • to4 ppm , DILUTION > TX/hr
0.2
OS
0.8 10 1.2 1.4 l£
rCtH4]0 +7.46 CC«H,3o, ppm
20
2 2
24
Fig. 5: Plots of concentration vs [C2H4lo + 7.46[2-C4H81q for a 12-hour
run at 294K and 1 atm total pressure for a sun declination of
10.53* and a latitude of 34*. Initial concentrations at 6:00 a.m. of
H2O vapor is 10^ ppm. The dilution factor is 7% per hour with air
containing the same composition of water vapor and O3 at 25 ppb:
0, #, rC2H4to/lNO]o = 2.25, [2-C4Hs3o = 0; 0, ¦, tC2H4Jo/[2-C4Ha]o =
4.0, NO = 178 ppb; a, A, [2-C4H8]0/[NO]0 = 0.56, [C2H4lo = 0. The
open points give [03]max in the absence of DEHA. The filled
points give [DEHA]C at 6:00 a.m. to keep and [PANlma*
ppb.
466
-------
INTERCOKPARISON OF METHODS FOR THE MEASUREMENT OF
AMBIENT LEVELS OF H2O2
T.E. Kleindlenst, P.B. Shepson, C.M. Nero
Northrop Services, Inc. - Environmental Sciences
Research Triangle Parle, HC 27709
R.R. Arnts
U.S. Environmental Protection Agency
Reaearbh Trianglfe Park, MC 2771}
G.I, Mackay, L.K. Mayne, H.I. Schiff
Unisearch Associates, Inc.
Concord, Ontario, Canada L4K 1B5
G.L. Kok, J.A. Lind, A.L. Lazrus
National Center for Atmospheric Research
Boulder, CO 60303
P.K. Dasgupta, H. Hwang
Department of Chemistry and Biochemistry
Texas Tech University
Lubbock, TX 79409
A study was conducted to perform simultaneous measurements of HjOg
concentrations using several techniques! including {1} tunable diode
laser absorption spectroscopy, (2) reaOnance fluorescence from an
enzymatlcally produced complex, and (3) ehemilumlneacence from
reaction with luminol. Sources of H2O2 included surrogate samples in
zero air In the presence and absence of coftmon interferences, that
produced by irradiating hydrocarbon (e.g., C2H4 or CH3CH0)/M0x
mixtures in a smog chamber, and that found in ambient air. The
technique* from four groups were compared with respect to. sensitivity,
selectivity! and dynasiic range for measuring t^jOg cofioentrations.
Measurements of H2O2 in aero air were performed for concentrations
ranging from 0.062 to 130 ppb. The agreement for the four sets of
467
-------
measurements ranged from 14 to 23% when compared to standard values
for the H2O2 concentration. For each technique, there was no
indication of interferences from O3, NO, NO2, HCHO, CH300H, and SO2,
except for the Xuminol technique, which showed a negative SO2
interference. Agreement among the techniques was considerably worse
for the irradiated hydrocarbon mixtures. In these mixtures,
significant concentrations of organic hydroperoxide were also detected
by the enzymatic techniques. For the ambient measurements, agreement
of approximately 30% with the mean values was achieved on average.
Introduction
Hydrogen peroxide (H2O2), an oxidant formed in the troposphere,
has received recent attention due to its proposed role in the
formation of acid precipitation. Produced from the recombination of
hydroperoxyl radicals, H2O2 is extremely soluble in water as it has a
Henry's Law constant of 1.5x105 M atm"1 at 25°C. It is currently
believed that H2O2 is the primary species in the atmosphere that
oxidizes sulfur (IV) (e.g., SO2) to sulfur (VI) compounds in the
condensed phased This has been suggested not only due to the high
aqueous solubility of H2O2, but also because its rate of reaction with
sulfur (IV) increases with increasing acidity. Although in-aitu
mechanisms exist for forming H2O2 in the aqueous phase, absorption
from the gas phase represents the major source of aqueous-phase H2O2.
Thus, to assess quantitatively the role of H2O2 in acid precipitation,
accurate measurements of H2O2 In the gas phase are required.
Few measurements of atmospheric H2O2 in the gas phase have been
reported. Until recently, most attempts have been plagued by
techniques having poor sensitivity or lack of specificity. Most of
these techniques have relied on trapping soluble air components in a
liquid scrubbing medium followed by analysis of the resultant
solution. Many of the previous difficulties have been met with the
use of improved liquid methods.2,3 Direct gas-phase measurements of
atmospheric H2O2 in situ have now also been reported. These recent
developments make it possible to measure H2Q2 accurately under a
variety of atmospheric conditions.
This paper describes an intercomparison study of various
techniques to measure H2O2 at ambient concentrations. The study was
conducted to measure H2O2 in zero air and in the absence and presence
of several interfering compounds. Measurements were also made for
H2O2 formed in surrogate photochemical mixtures and H2O2 produced in
ambient air. Four techniques, three of which were recently developed,
are compared with respect to sensitivity, selectivity, and dynamic
range.
Experimental Methods
The intercomparison study was conducted at Northrop Services, Inc.
- Environmental Sciences (NSI-ES) during the period June 16-26, 1986.
Participants in the study included groups from the National Center for
Atmospheric Research (NCAR), Texas Tech University (TTU), Unisearch
Associates, Inc. (UNI), and NSI-ES. The techniques each of the groups
employed during the study have been presented in the literature and
468
-------
only a brief description is given here. NCAR employed a continuous
scrubbing, fluorometrlc detection (CSFD) technique In which a
fluorescent diraer of p-hydroxyphenylacetic acid is formed by an
enzymatically mediated reaction with H2O2.2 TTU used a diffusion
scrubbing, fluorometrlc detection (DSFD) technique3 that employs the
same basic chemistry as in the NCAR technique to produce a fluorescent
dimer. The method of collecting H2O2 represents the major difference
between the techniques. In each of these enzymatic techniques the
total hydroperoxide and organic hydroperoxide concentrations were
measured. The H2O2 concentration was determined from the difference
of these two measurements. Uniseareh employed tunable diode laser
absorption spectrometry (TDLAS) for the direct measurement of H2O2 in
the gas phase.NSI-ES used the luminol technique based on methods
that were previously presented.5
In the initial phase of the study, pure samples of H2O2 were
generated and mixed with zero air. Vapor-phase H2O2 was generated by
passing zero air through microporous tubing submerged in a solution of
H2O2 in water. The gas-phase concentration of H2O2 emanating from the
generator could be continuously maintained at its Henry's Law
concentration at the flow rates employed. To maintain stability, the
entire generator was immersed in a thermostatted bath. The
concentration in the gas stream was dependent only on the solution
molarity and temperature. The concentrated H2O2 gas stream was fed
into a 1-L mixing manifold where it was diluted with zero air at
30 Lpm. The concentration in the mixing manifold could be set by
regulating the H2O2 flow feeding the manifold. The diluted mixture
was fed directly into a second manifold where sampling took place.
The generator, manifolds, and connecting lines were composed of PFA
Teflon, which has proved to be extremely inert to H2O2 loss, and were
shielded from light.
Undiluted, vapor-phase H2O2 concentrations from the generator were
measured by impinger collection through 28 mM solutions of TiCi.it at
pH 1. Upon reaction with H202» T1C14 forms a yellow complex with a
maximum absorbance of 415 nm. The solution concentration was then
determined spectrophotometrleally using a measured extinction
coefficient of 739 om-1 at 415 nm. This teohnique aan accurately
determine gas-phase H2O2 concentrations in the high ppbv to ppmv
range. With the use of two H202 generator solutions
-------
In a dynamic mode. These types of dynamic systems provide steady-
state concentrations of H2O2 in the presence of many inorganic and
organic species found in polluted air including peroxyacetyl nitrate
(PAN) produced in the CH3CHO/NOX system. The residence time of gases
in the chamber can be varied to allow various extents of reaction to
be obtained. For these experiments, effluent from the smog chamber
passed through 19-mm PFA Teflon tubing and into the sampling manifold.
In the last phase of the study, ambient air sampling was performed
on three consecutive nights and for a 24-h period. The ambient
sampling port, located 4 m above the ground, was connected directly to
the sampling manifold (19-mm PFA Teflon tubing) to ensure that all
measurements were taken from the same air parcels.
Results and Discussion
For the first phase of the study (H2O2 in zero air and in the
presence of possible interferences) the individual sampling periods
were 1 h in length. Collection of CSFD samples involved continuous
collection over the entire period while the DSFD, TDLAS, and luminol
samples were taken in a discrete measurement mode. Measurements were
obtained from a total of 23 sampling periods. In 10 periods, pure
samples of H2O2 in zero air (0.062 to 128 ppbv) were taken. In
another 10 sampling periods, one or more of the interferences were
added to approximately 5 ppbv H202- In the other three periods,
background samples from zero air were obtained including one in which
500 ppbv O3 was added to zero air.
A plot of the H2O2 measurements in zero air against standard H2O2
values (from the TiCl4 measurements) shows good linearity for each of
the techniques over several orders of magnitude. A DSFD value of
70 ppbv for a standard value of 128 ppbv is approximately 40Jt low.
This can be attributed to the fact that calibrations for this system
could only be performed in a range less than 3 ppbv. A luminol value
of 2,4 ppbv at a standard value of 1.1 ppbv appears high since the
technique (as it was used in this study} was operated at its detection
limit (1 ppbv) in this case. A comparison of the individual H2O2
measurements with the standard values gives the following average
percent deviations: CSFD - 18.3JE; DSFD - 21.4j(; TDLAS - 14.3J; and
luminol 23.4J. When averaged together the H2O2 data from the four
techniques showed less than a 3% systematic difference with the
standard values.
The concentration ranges for compounds serving as interferences
are given as follows: O3, 118-456 ppbv; SO2, 6-32 ppbv; HCH0 18-58
ppbv; N0x at 125 ppbv; and CH3OOH at 7 ppbv. The CSFD and TDLAS H2O2
data showed no effect for the addition of the Interferences either
alone or in combination. The luminol data showed a negative
interference due to the presence of S02 although the interference was
removed when SO2 and 03 (150 ppbv) were present together. The TTU
data showed a positive H2O2 interference due to 03 as determined in a
run with 500 ppb O3 in zero air. This signal was apparently due to a
surface effect, and subsequent investigation of this problem has
indicated that contaminated surfaces in the scrubber can lead to
artifact formation of H2O2 by O3.
470
-------
The photochemical systems produced a mixture of species in the
presence or absence of H2O2. The CH3CHO/NOX irradiated system
represents one of the simplest photochemical mixtures, particularly at
small extent of reaction. Under these conditions in which 10-20 ppb
of (JO remain in the mixture, only HCHO, PAN, O3, NO2, CH3ONO2, and
HNO3 are produced^ in significant concentrations. At a larger extent
of reaction for which radical-radical reactions are much more
probable, H2O2, organic peroxides, and organic nitrates are formed in
addition to the products of small extent of reaction. The C2H4/NOX
irradiation at a large extent of reaction produces NQ2, O3, HCHO, and
H20g. For each of the three photochemical mixtures, H2O2
concentrations in the low and sub ppbv range were observed. The H2O2
measurements taken from these photochemical mixtures generally showed
poor agreement (i.e., often worse than a factor of two). However, in
all of these mixtures, organic hydroperoxides (ROOH), as measured
using the enzymatic technique, were produced in yields much higher
than H2O2 Itself. According to the DSFD measurements, ROOH
concentrations were nearly an order of magnitude higher in
concentration than H2O2 for each system (assuming R is CH3). Since
the H2O2 concentration In the enzymatic systems is obtained from the
difference between the total peroxide signal and the organic
component, small relative uncertainties in each of these values can
lead to large uncertainties in the H2O2 concentration. This probably
accounts for the poor agreement in the photochemical systems.
The C2Hu/N0x irradiated mixture showed a large organic peroxide
component even though most current chemical mechanisms for C2H4
photooxidatlon Indicate formation of inorganic peroxide only. It has
been suggested? that hydroxymethyl hydroperoxide (HMP), HOCH2OOH,
could form via the addition of water to the CH2OO radical formed in
the ozonolysis of ethylene. ¦ To test this hypothesis, a C2H4/O3
mixture under dark conditions was tested. In this system, the organic
hydroperoxide as measured by the DSFD apparatus was a factor of seven
greater in concentration than the inorganic component. While the
measurements do not confirm the presence of HMP, there were no
observations inconsistent with its formation.
The ambient measurement of Hg02 generally showed better agreement
between methods than that obtained from the photochemical mixtures.
This is not surprising since the organic hydroperoxide component in
the atmosphere is generally much lower than H2O2 Itself. Mean H2O2
concentrations ranged from 0.1-2 ppbv. Agreement between data sets
was on the order of 3011,
Conclusions
This study has confirmed that reliable methods for measuring H2O2
in the gas phase with high sensitivity now exist. Detection limits of
better than 0.060 ppbv were obtained using the CSFD and DSFD enzymatic
techniques and approximately 0.10 ppbv for the TDLAS. For the lumlnol
system the detection limit was approximately 1 ppbv, though this
system was probably not optimized. The techniques were generally free
of interferences, although a negative interference was observed with
SO2 with the luminol technique. The ambient measurements indicate
471
-------
that the sensitivity of the DSFD, CSFD, and TDLAS is probably
sufficient to measure H2O2 under rural as well as urban conditions.
Disclaimer
Although the research described in this article has been funded
wholly or in part by the U.S. Environmental Protection Agency through
Contract 68-02-4033 to Northrop Services, Inc. - Environmental
Sciences, it has not been subject to the Agency's peer and policy
review. Therefore, it does not necessarily reflect the views of the
Agency, and no official endorsement should be inferred.
References
(1) D. Moller, "Kinetic model of atmospheric SO2 oxidation based on
published data," Atmos. Environ. 14: 1067 (1980).
(2) A. L. Lazrus, G. L. Kok, J. A, Lind, S. N. Gitlin, B. G. Heikes,
R. E. Sketter, "Automated fluorometric method for hydrogen
peroxide in air," Anal. Chem. 58: 59^ (1986).
(3) P. K. Dasgupta, S. Dong, H. Hwang, H-C. Yang, "Continuous liquid-
phase fluorometry coupled to a diffusion scrubber for the
determination of atmospheric formaldehyde, hydrogen peroxide, and
sulfur dioxide," Anal. Chem.. submitted for publication (1987).
(4) F. Slemr, G. W. Harris, D. R. Hastie, G. I. MacKay, H. I. Schiff,
"Measurement of gas phase hydrogen peroxide in air by tunable
diode laser absorption spectroscopy," J. Geophvs. Res. 2J.: 5371
(1986).
(5) G. L. Kok, T. P. Holler, M. B. Lopez, H. A. Nachtrieb, and M.
Yuan, "Chemiluminescent method for the determination of hydrogen
peroxide in the ambient atmosphere," Environ. Sci. Technol. \2\
1072 (1978).
(6) R. Atkinson, A. C. Lloyd, "Evaluation of Kinetic and Mechanistic
Data for Modeling Photochemical Smog," J. Phvs. Chem. Ref. Data
12: 315 (1984).
(7) S. Gab, E. Hellpointer, W. V. Turner, F. Korte, "Hydroxymethyl
hydroperoxide and bis (hydroxymethyl) peroxide from gas-phase
ozonolysis of naturally occurring alkenes," Nature 316; 535
(1985).
472
-------
Multivariate Calibration Using the
Partial Least Squares (PLS) Method
W. J. Dunn III
College of Pharmacy
University of Illinois at Chicago
833 S. Wood Street
Chicago, Illinois 60612
ABSTRACT
The calibration of the gas chromatograph for quantitating the
components of complex mixtures observed in the air quality
monitoring program can lead to difficulties due to nonlinearity
of detector response and colinearity in the calibration data
itself. It is difficult to deal with these two problems with the
software resident on most gas chromatography systems as most
calibration software is based on multivariate regression
techniques. The recently developed partial least squares (PLS)
in latent variables (1) method has been shown to work, quite well
when either or both of these conditions are present is a
calibration problem. This approach and the algorithm on which it
is based will be discussed below.
Introduction
The problem of quantitative analysis of gas chromatography
data obtained from samples of complex mixtures is generally a
calibration problem. In samples in which the number of
components is 30 or more it is not practical to calibrate for the
components as frequently as required sinoe more than 30
calibration samples should be done in order to reduce
collinearity problems. If the detector response is nonlinear in
the concentration ranges of interest the problem is further
473
-------
compounded. A procedure has recently been proposed which can
handle both of these problems (1). This technique is the partial
least squares (PLS) regression method (1) which has been recently
proposed as a method well-suited for multivariable calibration
problems, such as the one discussed here. As the result of
preliminary applications of the method to gas chromatography data
of Aroclor mixtures (2) we have found it to work extremely well.
a. Multivariable calibration
A typical gas chromatography calibration sample is shown in
Figure 1. This reconstructed ion chromatogram was obtained in
the routine calibration of a GC/MS analysis of air quality data.
The known concentrations of the analytes are called the chemical
data and the GC responses are the response data. The chemical
data will be referred to as the dependent variables or y-block
and the response data are the independent variables or the X-
block. The number of calibration samples in a given experiment
is n, the number of components in calibration sample is m and the
number of chemical responses is p. In the case of GC data, m -
p, while in the case of uv/ir data m < p.
Calibration is the process of developing a model which will
allow the prediction of the Y-block from the X-block. This is
sometimes referred to as determining the sensitivity of the
method.
There are two general steps involved in the calibration
process. The first is called calibration and the second is
called testing. There are several data analytic methods which
can be applied in the first, or calibration, step. The method
most commonly used is multivariate regression since it is the
method resident in the computers present on most commercial
instruments. Another method, principal components regression,
has also been used, and more recently, the partial least squares
in latent variables, (PLS) regression has been proposed as a
robust method for this problem.
In the ideal calibration problem, the chemical data are
linear with instrument response. If the problent is multivariable
in nature, there can be multicollinearity in the data. If there
is nonlinearity in the data and/or if there are collinearities in
the data, multivariable regression is not appropriate for
application. Principal components regression can deal with the
collinearity problem but the PLS regression technique gives more
robust results. A discussion of this technique follows.
b. Partial least squares (PLS) regression
The PLS regression method operates by extracting principal
components-like vectors from the X- and Y-blocks, This is shown
in Figure l where the 1 and Q vectors are derived to
simultaneously explain the sytematic variation in the chemical
and response data be optimally correlated through the inner
474
-------
u
Figure 1, Diagram of the PLS regression methods
relation, I ¦ m. This is also shown algebraically in
equations 1-4, below. Here, i is the chemical data index and j
Xfci - X + £ tu Ptl + Bp 1)
yu - y + E u>, + fhJ 2)
U ¦ b t 3)
f ¦ b t q 4)
is the response index* The summations are over the number of
components A where a - 1, 2, .. A. Equation 3 is the inner
relation between the t's and u's which leads to the predictive
relationship in equation 4.
The PLS method uses the Nonlinear Iterative PArtial Least
Squares, or NIPALS, algorithm to extract the T's and U's. The
algorithm is a 7-step iterative procedure which is given below.
Step 1. let u,twt * some y.
Step 2. p' - u"x/u'u (w' - u'X/u'u)
Step 3. normalize p' (normalize w*)
Step 4. t - Xp/p'p (t - Xv/w'w)
Step 5. q• - t'Y/t't
Step 6. normalize q'
Step 7. u - Yq/q'q
On reaching Step 7, compare u with that in Step l. If they are
equivalent atop, otherwise mova the u in step 7 to Step l and
repeat until convergence. After convergence calculate the
weights as shown in steps in parentheses to give orthogonal t's.
Usually the algorithm converges quickly. The t's can then be
used to derive predictive inner relations.
475
-------
Note that in Step 2, the vector p* is calculated so as to
contain information from the X-block and in Step 5 q' is
calculated so as to contain information about the X-block. The
result is that the t's and u's extracted are correlated.
After a component has been extracted, and the y's and x's
predicted, blocks of orthogonal residuals result. From these
addition PLS components can be calculated. Cross-validation (
3) can be used to estimate the number of predictively significant
PLS components that are optimal for a given problem.
Once the t's have been calculated they can be used in the
prediction step to estimate the concentrations. This is done
through the relations shown in equations 3 and 4, above.
PLS regression is a rather new approach to the calibration
problem. It is not well described in the literature but numerous
papers are beginning to appear of its applications in this area.
The method has a number of advantages which make it suitable
calibration problems associated with complex mixtures. One of
its advantages is that it is computationally straight forward and
can be programmed to run on desk top computers.
References.
1. S. Wold, A. Ruhe, H. Wold and W. Dunn, SIAM, J. Sci. Stat.
Comput,, 5 £1984) 735.
2. W, J. Dunn, D. L. Stalling, T. R. Schwartz, J. W. Hogan, J.
D. Petty, E. Johansson and S. Wold, Anal. Chem., 56 <1984)
1308.
3. S. Wold, Technometrics, 10 (1978) 127.
476
-------
MULTIVARIATE QUALITY ASSURANCE PROGRAM FOR
DIOXINS AND FURANS
Stephen E. Swanson, Martin Nygren, Lara-Owe Kjeller and
Christoffer Rappe
Department of Organic Chemistry
University of UtneS
Umeft, Sweden
A quality assurance program has been initiated in this
department which uses multivariate statistics to monitor daily
mass spectrometer performance. With this program, instrument
performance can be determined through the simultaneous evalua-
tion of all 28 analytes in each quantitation standard. This
multi-element calibration is simplified through the use of a
peak detection and quantitation program combined with the SIMCA
multivariate statistics package. This peak detection program
uses peak detection algorithms available in the Finnigan INCOS
software, then evaluates those peaks using techniques developed
in this department. Through the use of these techniques, peaks
are evaluated based on relative chromatographic retention and
relative ratio of ions in molecular ion clusters. Dioxin and
furan peaks can be quantitated or evaluated using multivariate
statistics. Thus instrument response can be accurately eval-
uated regardless of how many different operators are using the
instrument.
Introduction
Frequent monitoring of instrument response is a basic
quality assurance technique which is common practice in most
analytical laboratories and is a minimum requirement of most
standard procedures.! In most laboratories, this is accomp-
lished through maintenance of calibration curves and instrument
response charts for quantitation standards. Frequently these
quantitation standards can contain upwards of 20, 30 or more
477
-------
analytes which must be monitored individually. When instrument
response records are based on methods of linear regression
analysis, a separate response curve and response chart must be
maintained for each analyte. Thus with 20 or more quality
control charts to monitor after every 8 hours of operation,
this minimum quality assurance technique can become quite
tedious. With the use of multivariate statistical methods,
however, the response of these several analytes can be combined
into one instrument file. The result is that response of all
analytes can be monitored simultaneously, this simplifying
record-keeping aiding in interpretation.2
If multivariate statistical methods are to be used
routinely, the standard response data for all analytes must be
processed and combined into a single file for multivariate
analysis. This procedure may require specialized software. The
organic analytical chemistry group at the University of UmeS
has developed a procedure which allows multivariate analysis of
instrument response data collected from a Pinnigan 4500 mass
spectrometer using an INCOS data system. This software was
specially developed for evaluation and processing of MID
(multiple ion detection) results from analyses of Poly-
chlorinated Dibenzo-p-Dioxins (PCDDs) and Polychlorinated
Dibenzo Purans (PCDFs).
Experimental Methods
The Finnigan 4500 mass spectrometer in the organic
chemistry department at the University of UmeA is routinely
used for analysis of PCDDs and PCDFs in a variety of samples.
To achieve maximum sensitivity and selectivity, the mass spec-
trometer is operated in the negative chemical ionization mode
and the mass spectral responses are monitored using the MID
technique on the INCOS data system. The masses monitored and
their theoretical ratios are given in table I.
Special application software has been written by this
group to evaluate and process this MID data. The general pro-
cedure for data processing is shown in Figure 1. Following
GC/MS analysis for PCDDs and PCDFs, the INCOS data system
prints out the chrorn&tograms of ions selected by the MID
program, shown in Table I. This INCOS procedure also quanti-
tates each chromatographic peak and stores the peak heights and
areas in "*.QL" files. The "*.QL" files are then transferred to
an IBM AT personal computer. The IBM then sorts the peak
heights and areas by relative retention times to pair responses
for the molecular ion (M) and the M + 2 ion (see Table I). The
pairs of heights and areas are then evaluated. First, the pairs
are evaluated based on the ratio of M and M + 2 and then on the
relative retention time of the peaks compared to the
established retention time of the PCDD and PCDF* isomers. The
identification of TetraCDFs is shown as an example,
TetraCDF is monitored by response of its molecular ion
(M=303.9) and the M+2 ion (305.9), as shown in Table I. The
peak heights and areas are sorted by the IBM based on relative
retention time so that each peak for mass 303.9 is paired with
a 305.9 peak of the same retention time. These paired peaks are
478
-------
then evaluated based on the following PCDD/PCDF identification
criteria. First the areas of the 303.9 peaks are compared with
the 305.9 peaks. This area ratio is then compared to the theo-
retical ion ratio (76%), shown in Table I. The same comparison
is made for the heights of the 303.9 and 305.9 peaks. Next, the
pairs of peaks are evaluated based on their relative retention
times compared to actual retention times of the TetraCDF
isomers established by analysis of a retention time standard.
Peaks are then flagged as TetraCDFs which meet all three iden-
tification criteria: 1.) peaks which appear at the same reten-
tion time for both M and M+2 ( 303.9 and 305.9)? 2.) peaks which
have the correct ratio of height and area (76 + 10%) and 3.)
peaks which are within 10% of the established TetraCDF reten-
tion times. The results of this sorting and evaluation are
printed out by the IRM computer so the analyst can review the
computer's evaluation. This resulting file may be edited if the
analyst disagrees with the printed evaluations. After the file
is edited, only the PCDD and PCDF peaks are saved.
This PCDD/PCDF raw data file can be of three types; quan-
titation standard, unknown, or retention time standard, as
shown in Figure 1. Each type of file is processed differently.
In quantitation standards, the relative response of the peak
height and area of each compound is calculated compared to the
internal standard response. This information is stored for
later use in calculation of unknowns. This data can also be
stored in a format which can be taken up for multivariate
statistical analysis using the SIMCA-3B program. In the second
type of data file, samples of unknown concentration, the PCDD
and PCDF responses are corrected for internal standard res-
ponse, then are compared with measured responses for the quan-
titation standard and final concentrations are calculated.
These concentrations can be printed out in an annotated data
report or can be saved in a format which can be taken up for
multivariate statistical analysis using the SIMCA-3B program.
The third type of data file, results of analysis of a retention
standard, is used to create a new retention time template which
is then stored for evaluation of subsequent data files.
Principal component (PC) analysis and the SIMCA-3B system
have been described in detail elsewhere.3 Thus a brief summary
of the features of this system will be given here. The PC model
of the data set is calculated, as shown by the following equa-
tion:
xik - *k + a=! fcia * pak + eik
Where x^ is the measured value of the variable k (the height
or area of mass peak k) of object i (the ion chromatorgram i).
The parameters are the averages of the variables k. The PC
parameters t£a» scores, and Pak' loadings, are calculated so
that the residuals are minimized using the least squares
method. The loadings describe the multivariate relation between
variables and the scores describe the multivariate relation
between objects. The number, A, of statistically significant
components (product terms) is determined by cross validation.3
479
-------
Results
The average time for GC/MS analysis of all Tetra- through
OctaCDDs and CDFs is approximately one hour. Subsequent to that
analysis, the printing of the results and creation of "*.QL"
files on the INCOS system takes 45 minutes to one hour. The
remaining steps in the process: transfer of data, sorting,
evaluation, editing and final PC analysis of the standards data
takes another 15 to 20 minutes. In general, little or no addi-
tional editing of the data files are required when quantation
standards are analyzed.
Figure 2 shows a calibration curve which was generated
using this procedure. The results presented are a 6-point
calibration curve which with standards 1 through 6 representing
a concentration range of two orders of magnitude. Though this
appears to be a linear calibration curve, the graph shown in
Figure 2 is actually created from the responses of all M and
M+2 heights and areas, a total of 128 variables.
Conclusion
Through the use of this data processing package, the
results of PCDD and PCDP analysis can be quickly evaluated and
the concentration calculated. This data processing package also
provides the capability for formatting these results in manner
which allows subsequent multivariate statistical analysis using
the SIMCA-3B package. Multivariate statistical analysis of
mixed quantitation standards allows simultaneous evaluation of
the responses of all compounds in the mixed standard, thus
simplifying the evaluation of instrument response.
1. U.S. Environmental Protection Agency,"Waste water analysis
method: 2,3,7,8-tetrachlorodibenzo-p-dioxin. Method 613."
Fed.Reg. 44:69464 (1979).
2. G.U.M. Lindstron, E. Johansson, M. Nygren, "Quality
assurance of multivatiate analytical measurements by
principal components analysis,"Chemoaphere 14: 1695 (1985)
3. S. Wold, et.al.,Chemometrics Mathematics and Statistics in
Chemistry ed. B.R.Kowalski, Reidel Pub.Co., Dardrecht,
Holland. 1984, pp.17-96,
480
-------
GC/MS ANALYSIS |
FILES TO IBM 1
UNKNOWN
SAVE
PRINT
OUT
PRINT OUT
SIMCA
FILE
QUANT
STD.
SIMCA
FILE
RET.TIME STO.
RET.TIME
TEMPLATE
CONCENTRATIONS
QUANTITATION STD
RELATIVE
RESPONSES
PRINT FLAGGED LIST
EDIT LIST
SAVE ONLY PCDDs/PCDFs
SORT G EVALUATE BASED ON:
1. REL. RET. TIME
2. HEIGHT RATIOS
3. AREA RATIOS
4. RET.TIME STO.
Figure 1. Procedure for evaluating, quantitating and storing
PCDD/PCDF data.
ai
-------
Table I. PCDD and PCDF Ions Monitored And Their Theoretical
Ratios
Homolog Masses Monitored
(Theoretical ratio)
TetraCDF
PentaCDF
PentaCDD
HexaCDF
HexaCDD
HeptaCDF
HeptaCDD
OctaCDF
OctaCDD
303.90/305.90
337 .86/339.86
353.86/355.06
371.82/373.82
387.82/389.82
405.78/407.78
421.80/423.78
441.75/443.74
457.74/459.74
(100/76)
(100/61)
(100/61)
(100/51)
(100/51)
{100/43)
(100/43)
(100/87)
(100/87 )
482
-------
IDENTIFYING POLYCYCLIC AROMATIC HYDROCARBON SOURCES
FROM AMBIENT AIR DATA COLLECTED IN URBANIZED
AREAS OF NEW JERSEY
R. Harkov - Office of Science and Research, New Jersey Department of
S. Shiboski Environmental Protection, Trenton, New Jersey.
A. Greenberg - Department of Chemistry, NJIT, Newark, New Jersey.
F. Darack
ABSTRACT - A large ambient air data base for polycyclic aromatic
hydrocarbons has been analyzed to identify potential combustion source
contributors to the measured concentrations. Two related multivariate
procedures, principle factor analysis (PFA) and principle component biplot
(PCBi), were utilized to analyze data collected at three urban New Jersey
sites; Newark, Elizabeth and Camden. Results from this study identified
three to four potential source groupings for the PAH. However, only two
specific source categories were tentatively linked to PAH levels at Newark
and Camden. A major difficulty in the present study is the lack of adequate
PAH emissions data.
INTRODUCTION - Polycyclic aromatic hydrocarbons (PAH) have been
measured in the air environment for more than twenty five years (NAS 1983).
In addition, different profiles of PAH have been developed for various
combustion sources such as motor vehicles, space heaters, power plants, coke
ovens and agricultural and forestry residue removal operations (Grimmer
1983, Daisey et al 1986). Although source and ambient data on PAH have been
available for a period of time, estimating the relative contribution of the
various combustion sources to the measured atmospheric concentrations has
been difficult. This observation is in contrast to the significant progress
that has been made in the apportionment of source contributors for other
pollutants such as particulates (Copper and Watson 1980).
Some of the problems in utilizing ambient air PAH data to apportion the
contribution from various combustion sources include: a) relatively small
data sets, b) incomplete knowledge of emission profiles of various PAH and
the consistency in the emitted profiles, c) poor precision associated with
ambient PAH measurements, d) weak correlations with potential source
tracers such as metals and e) high degree of correlation between PAH within
a data set (Harkov et al 1986a, b, Daisey et al 1986). In spite of this
situation a number of workers have attempted to identify PAH sources
utilizing ambient air data (Daisey et al 1979, Harkov et al 1984, Thrane
1984, Cretney et al 1985, Harkov et al 1986c). Cretney et al (1985) were
successful in resolving transformed PAH data analysed by principle
components analysis (PCA) into the two moat important PAH source* in New
Zealand. Using principle component biplot (PCBi), Harkov et al (1986c) were
able to tentatively identify PAH source# at four New Jersey locations.
Generally, most workers have met with a modest level of success in
identifying PAH source contributors, but have been unsuccessful in
quantitative PAH source apportionment. lite present report utilizes a large
PAH data base from three urban sites in New Jersey to identify combustion
source contributors to the measured ambient PAH levels. la this analysis
483
-------
two related multivariate techniques, principal factor analysis (PFA) and
PCBi are utilized to help resolve PAH sources at the three urban New Jersey
sites.
METHODS -
Sample Collection/Analysis - The PAH data was collected at the three
urban sites (Newark, Elizabeth, Camden) utilized during the New Jersey ATEOS
project (Harkov and Fischer 1982). During the ATEOS project, PAH samples
were collected with size-selective inlet (SSI) PM15 samplers equipped with
pre-ignited glass fiber filters (Gelman-A) through four separate, 6 week
campaigns from 1981-1983 (Greenberg et al 1985). As a follow up study, PM10
samples were collected using Wedding inlets and pre-ignited quartz fiber
filters (Whatman-QMA) every sixth day from 10/83-10/84 at the same three
urban sites. All samples were analyzed in an identical fashion using
soxhlet extraction with cyclohexane, followed by TLC and HPLC. The complete
ATEOS PAH and quality assurance (qa) data and the analytical procedures have
been presented elsewhere (Greenberg et al 198S, Harkov et al 1986a), as have
the subsequent PM10 PAH data with qa results (Harkov et al 1986b).
Data Base Description - By combining both data bases between 200-220
observations are available for analysis for each site. However, since PAH
data sets contain a large number of missing values due to numbers below
detection limits or noise in the analysis (Harkov et al 1986a,b),
multivariate statistical techniques are difficult to perform and interpret
for this pollutant class. Eight PAH were selected for analysis because they
represented the most complete data set of the twenty six compounds measured.
These compounds are shown in Table-1 with some physical characteristics.
Levels of data quality (Harkov et al 1986a,b) and reactivity indices
(Greenberg and Darack 1985) indicate that no systematic patterns exist
within this data base. The multivariate analyses are thus assumed to
reflect source contributions and meteorological changes, rather than the
atmospheric, sampling or analytical fate of the various pollutants.
Statistical Approach
A - Principle Factor Analysis - Principle factor analysis was carried
out with the personal computer software STATGRAPHICS (STSC 1981) utilizing
listwise deletion of missing values and varimax rotations. The Hotelling
T test available in STATGRAPHICS was utilized to identify potential
outliers, and from those observations identified each one was individually
removed to demonstrate its importance on the final model development.
B - Principle Component Biplot (PCBi) - This exploratory data analysis
tool was developed by Gabriel and coworkers at the University of Rochester
(Gabriel 1982, Cox and Gabriel 1982) and was first applied to air pollution
data by Harkov et al (1986c). It is a graphical display of both rows and
columns of a data matrix, i.e. of observations (rows) and individual
pollutants (columns) in the present case. Biplots are useful Cor visual
inspection of data matrices so that outliers or unusual distributions of the
observations or pollutants can be readily observed. For PCBi, the sample
data is approximated by the first two principal components, which generally
484
-------
account for 70-90* of the variance in the model. In the present case, GH'
PCBi has been utilized, since this biplot procedure gives a better fit to
the variables (pollutants) than to the observations (Gabriel 1982). The
theory behind the biplot representation rests on the fact that only matrix x
of n rows and p columns with rank r can be factorized into a product of a
matrix G with n rows and r columns a matrix H with p rows and r columns,
such that X - GH'. The principal component biplot starts with a standard
principal component analysis of the correlation matrix of the data matrix X
and bases G and H on the first two principal component scores and first two
eigenvectors respectively (i.e. r-2). The actual plot Is produced by
plotting G and H in the space of the first two principal components. The n
points In the resulting plots represent the n observations, while the p
sectors represent the p variables of the data matrix X. For the PCBi
display, the angles between any two vectors are Inversely proportional to
their correlation and the length of a vector is proportional to Its
contribution to the principal components. The personal computer version of
the statistical package MIHITAB (Ryan et al 1982)_ was utilized to produce
PCBI. This analysis was accomplished utilizing 'a series of algorlthlms
available in the MINITAB package.
RESULTS
Principle Factor Analysts (PFA) - In the present analysis, PFA
Identified three to four similar factors at each site and only final factor
solutions will be presented (Tables 2-4). Outliers were confirmed for only
the Elizabeth site (N»3) and the Camden site (N"l). The total number of
complete observations for the Kevark, Elizabeth and Camden sites, after
llstwise deletion, were 83, 52, and 79 respectively. The first factor
accounted for approximately 80Z of the variance in the model of each site
and was primarily associated with BkF, BaP, BF and BbF, with lower loadings
of BghiP and ICDP. The next factor was associated with Cor and BghiP and
accounted for approximately 107 of the variance at each site. A third
factor was associated with BeP and accounted for approximately 5SE of the
variance In the models. Finally, at Elizabeth and Camden a fourth factor
was associated with ICDP and accounted for about 4Z of the variance in the
model.
Principle Components Biplot (PCBi) - Since principle components
analysis is based on sample correlation or covarlance matrices, we have
elected to employ pairwlse correlation matrices to increase the amount of
Information utilized in the PCBI analysts. The use of pairwise correlation
matrices Increases the number of samples considered in the analysis by using
all the complete observations on any pair of variables In the computation of
correlation coefficients. However, to actually compute the principle
component scores only the complete observations could be utilised. The
effect of this approach is that the correlation matrices are based on
approximately twice the number of data points when compared with the
observations with no missing values. This method was thought to more
closely approximate the true relationship between variables and provide a
contrast to the approach utilized in PFA.
485
-------
As described earlier, the PCBi technique can allow visual
identification of outliers in a data set. In the present analysis, each
potential outlier was removed from the data set to test its influence on the
biplot results. The testing of outliers from the PCBi technique for the
current data produced the following number of outliers; two at Newark, three
at Elizabeth and five at Camden. Final complete observation numbers for
Newark, Elizabeth and Camden in the biplot analysis were 81, 52 and 75
respectively. Those outliers identified in both PFA and PCBi were
identical, but it appears that the PCBi method is more sensitive to outliers
than the PFA screening procedure utilizing the Hotelling T test. The
information in Figures 2a and 2b provides a comparison of PCBi results with
and without outlier removal at the Camden site.
All three sites had a common cluster of PAH (BkF, BaP, B J, BbF),
(Figures 1-3). At Newark and Camden a common cluster occurred for »ghiP/Cor
and at both these sites BeP was separated from the other seven PAN
compounds. On the Elizabeth biplot BeP/Cor and BghiP/ICDP produced separate
clusters.
Discussion - Profiles of PAH emissions from combustion sources have
been compiled by Grimmer (1983) and most recently by Daisey et al (1986).
Analysis of these compilations indicates that the current state of knowledge
on PAH emission profiles is not adequate for use in receptor modelling.
Because the state-of-the-art for PAH fingerprinting is very immature the
present analysis will rely heavily on additional information pertaining to
the specific airsheds utilized in the current study and previous
apportionments of PM..,., EOM and BaP in New Jersey (Morandi 1985, Harkov and
Shiboski 1986, Harkov and Greenberg 1985).
Both PFA and PCBi identified a group of PAH (BkF, BaP, B.F, BbF) that
were clustered together at the three sites. Since additional PAH had
reasonably high loadings on this factor in PFA and because this factor
accounted for a very significant portion (^80%) of the variance in the PFA
models, it is possible that this cluster was related to general
meteorological conditions and/or a number of non-specific (area) PAH
sources. However, the PCBi identified this cluster as being clearly
separate from the other PAH at all three sites. A review of data concerned
with diesel and spark ignition motor vehicle exhaust, industrial, and oil
and coal combustion emissions does not yield any potential source category
for this cluster (NAS 1983, Grimmer 1983, Daisey et al 1986, Hangebrauck et
al 1967). Relatively high levels of benzofluoranthenes have been detected
in communities impacted by wood combustion and in wood smoke soot (Manning
et al 1983, Soderberg et al 1983, Cretney et al 1985, Sexton et al 1985). A
source apportionment for BaP in New Jersey (Harkov and Greenberg 1985) '
identified wood combustion as the major source of this contaminant during
the beating season (Nov,-March). Because the monitoring locales utilized in
the present study are inner-city sites, any impacts of wood combustion
emissions would probably be small. Since our heating season data set is
insufficiently large after listwise deletion, we could not explore a
winter/summer comparison of this PAH cluster. Thus the actual source
identification for this cluster is uncertain.
486
-------
The PFA identified a Cor/BghiP factor at each site and thia factor
accounted for about 93S of the variance in the model. Both Cor and BghiP
have been used as indicators of motor vehicle contributions to ambient PAH
levels, since PAH profiles from this source category Indicates that
relatively high levels of these two compounds are emitted to the atmosphere
(Grimmer 1983, NAS 1983, Daisey et al 1986). The PCBi analysis also
produced a £or/BghlP cluster at two of the three Bites. In the Elizabeth
PCBi BeP was clustered with Cor and ICDP with BghiP, Thia result is
contrary to what was expected since the Elizabeth site was thought to be the
most heavily impacted by motor vehicles (Harkov and Shiboskl 1986, Harkov et
al 1984). The PAH clusters at a site should reflect; a) source emission
densities and profiles in the vicinity of the locale, b) environmental fate
of the PAH and c) local meteorology. While previous studies in New Jersey
Indicate that motor vehicles are a major source of BaP and extractable
organic matter (Harkov and Greenberg 1985, Harkov and Sbiloski 1986), the
poor resolution of motor vehicles in the Elizabeth PCBi may be caused by two
sources which are located In the same general direction and/or as a result
of strong alternate PAH sources that emit Cor or BghiP. The only potential
source category near Elizabeth which can meet these requirements is a large
oil refinery that is located to the south and west of the site.
Benzo(e)pyrene was identified as a factor at all three urban sites and
independent of the other PAH in the PCBi at Newark and Camden. At the
Elizabeth site, BeP was associated with Cor in the PCBi. Although ve cannot
definitely attribute a source category to BeP, we are tentatively linking it
to fuel oil combustion. Relatively high levels of BeP emissions from oil
combustion have been reported in the literature (Brockhaus and Tomlngas
1976, Hangebrauck et al 1967, Bennett et al 1979), although this emissions
information is far from being mature. Fuel oil Is an important energy
source in inner city areas of New Jersey, although the actual amount of PAH
produced from this source category appears to be small (Harkov and Greenberg
1985).
The factor analysis has produced a factor with high loadings of ICDP at
Elizabeth and Camden. Since PCBi generally indicated that ICDP is
associated with clusters of other PAH and the amount of variance explained
by ICDP in the PFA models is very small (~4X), at the presents time we are
considering this factor as an artifact of the specific statistical technique
and not related to a potential source category.
By utilizing two different multivariate techniques, it has been
possible to provide some Internal checks on the determination of sources of
PAH from this data set. Generally, both methods provided similar results,
with some notable exceptions. One potential advantage of the PCBI procedure
is that the problem of status of pollutants and factor Inclusion In the
final models for PFA Is avoided. Identifying the correct number of viable
factors in PFA is difficult because of the problem of the lack of rigorous
factor cut-off criteria. The problem of factor or cluster identification
appears to be small or nonexistent in the PCBi technique. However, compared
with PFA outlier rejection is a more important step in cluster resolution in
PCBi.
487
-------
Without more complete information of source emissions of PAH it will
remain difficult to utilize ambient data to accomplish source apportionment
for this class of air pollutant. Currently there is a small program
evolving in New Jersey to characterize stationary combustion source
emissions profiles of PAH, NO.-PAH, Quinones and other substituted PAH to
help in both source resolution from ambient data and to understand the
potential impacts from increasing the amount of waste combustion in the New
Jersey environment.
CONCLUSIONS
A multivariate statistical analysis of ambient PAH data was carried out
to identify potential source contributors to the measured concentrations.
Both PFA and PCBi produced very similar results in that some three to four
potential groups of PAH were found to explain most of the variability in the
data from three urban New Jersey sites. Only one grouping of PAH (BkF, BaP,
B.F, BbF) was common to all three sites. Identification of a combustion
source contributor to the ambient levels of these PAH was not possible
utilizing current emissions data. A Cor/BghiP and a BeP group was
identified at Newark and Camden and was thought to reflect motor vehicle and
oil combustion contributions respectively. At Elizabeth two groupings of
PAH, Cor/BeP and BghiP/ICDP, were thought to be produced by the influence of
an additional source co-located in the direction of high source strength
areas for motor vehicles and oil combustion. This source was tentatively
identified as a large oil refinery. Without better emissions information,
source resolution of ambient PAH data sets will be very difficult to perform
successfully.
488
-------
References
Bennett, R.L., Knapp, K.J., Jones, P.W., Wilkerson, J.E., and Strop, P.E.
(1979). Measurement of polynuclear aromatic hydrocarbons and other
hazardous organic compounds In stack gases. In, Polynuclear aromatic
hydrocarbons (eds. P.W. Jones and P. Leber) pp 419-428, Ann Arbor Sci.
Pub., Ann Arbo, Mi.
Brockhaus, A. and Tomingas, R. (1976). Emission polyzyklisoher, Kohlen
wassertoffe bei Vebrennungoprizessen in kleinen Helzungsanlegen and
ihre Konzentration in Der Atmosphere. Staub Reinhalt. Luft 36:96.
Cooper, J.A., and Watson, J.G. (1980). Receptor oriented methods of air
particulate source apportionment. J. Air Pollut. Control Ass.
30:1115-1124.
Cox, C. and Gabriel, K.R. (1982). Some comparisons of biplot display and
pencil-and-paper E.D.A. methods. In, Modern data analysis (eds.
Laurner, R.L. and Siegel, A.F.) Acad. Press. NY, NY 201 pp.
Cretney, J.R., Lee, H.V., Wright, G.J., Swallow, V.H. and Taylor, M.C.
(1985). , Analysis of polycyclic aromatic hydrocarbons in air
particulate matter from a lightly industrialized urban area. Env. Sci.
Tech. 19:397-404.
Daisey, J.M., Cheney, J.L. and P.J. Lioy (1986). Profiles of organic
particulate emissions from air pollution sources: Status and needs for
receptor source apportionment modeling. J. Air Pollut. Control Ass.
36:17-33.
Daisey, J.M., Leyko, M.A. and Kneip, T.J. (1979). Source identification
and allocation of polynuclear aromatic hydrocarbon compounds in the New
York City aerosol: Methods and applications. In, Polynuclear aromatic
hydrocarbons (eds. Jones, P.W. and Leber P.) pp 201-215, Ann Arbor
Sci. Pub., Ann Arbor, Ml.
Hangebrauck, R.P., VonLehmden, D.J., and Meeker, J.E. (1967). Sources of
polynuclear hydrocarbons in the atmosphere. USDHEW-Public Health
Service, Clnn., Ohio 44 pp.
Gabriel, K.R. (1982). Biplot. In, Encyc. Stat. Science. J. Wiley and
Sons, NY, NY. pp 263-270,
Greenberg, A. and Darack, F. (1985). Atmospheric and reactivity Indices of
polycyclic aromatic hydrocarbons. In, Molecular Structure and
Energetics (eds. Liebermann, J. and Greenberg, A.) VCH Pub. Dnrfield,
Pla. In Press
Greenberg, A, Darack, P., Harkov, R., Lioy, P. and Daisey, J. (1985).
Polycyclic aromatic hydrocarbons in New Jersey: A comparison of winter
and summer concentration over a two-year period. Atmos. Env.
19:1325-1339.
489
-------
References (Cont'd)
Grimmer, G. (1983). Environmental Carcinogens: Polycylic aromatic
hydrocarbons. CRC Press, Boca Raton, Fla. 261 pp.
Harkov, R. and 5. Shiboski (1986), Can extractable organic matter be used
as a motor vehicle tracer in multivariate receptor models? J. Env.
Sci. Health. 21:177-190.
Harkov, R., Lioy, P.J., Daisey, J,M., Greenberg, A., Darack, t., Bozfcelli,
J.W., and Kebbekus, B, (1986a). Quality control for non-criteria air
pollutants, J. Air Pollut. Control Asb. 36:388-392.
Harkov, R., Greenberg, A., Darack, F. McGeorge, L. and Pleterlnen, C.
(1986b). Exploring data quality for some Inorganic and organic
constituents of inhalable particulate matter {PMlf)). Presented at
79th Annual APCA meeting, Minn, Minn. #86-46.8.
Harkov, R., Shiboski, S., Greenberg, A., and Darack, F. (1986c). The use
of principle component biplots (PCBi) for exploring polycyclic aromatic
hydrocarbons data from New Jersey. Presented at the 79th Annual APCA
meeting, Minn, Minn. #86-19.3.
Harkov, R. and Greenberg (1985). Benzo(a)pyrene In Mew Jersey. Results
from a twenty seven site study. J. Air Pollut. Control Ass.
35:238-243.
Harkov, R., Greenberg, A., Darack, F., Lloy, P.J. and Daisey, J.M. (1984).
Summertime variations in polycyclic aromatic hydrocarbons at four siteB
in New Jersey. Env. Sci. Techn. 18:287-291.
Harkov, R. and Fischer, R. (1982). Development and initiation of an
integrated state monitoring program for airborne toxic substances.
In, ProcedlngB 78th Annual APCA meeting, New Orleans, La. #82-1.1.
Manning, J.A., Imhoff, R.G, and Akland, G.G. (1983). Wintertime ambient
measurements of particulate polycyclic aromatic hydrocarbons in a
residential community. In, Measurement and monitoring of non-crlterla
contaminants in air. APCA Specialty Pub, #SP-50, 538 pp.
Morandi, M.T. (1985). Development of source apportionment models for
inhalable particulate matter and its extractable organic {Tactions in
urban areas of New Jersey. Doctoral Thesis. NYU Medical Center, NY,
NY 286 pp.
NAS (1983). Polycyclic Aromatic Hydrocarbons. National Acad, Sci., Wash.,
B.C.
Ryan, T.A., Joiner, B,L. and Ryan, B.F. (1982). Mlnitab student handbook,
PWS Pub., Boston, MA. 341 pp.
490
-------
References (Cont'd)
Sexton, K., Liu, K-.S, Hayward, S.B. and Spengler, J.D. (1985).
Characterization and source apportionment of wintertime aerosol in a
wood-burning community. Atmos. Env. 19:1225-1236.
Soderberg, R.H., Hornig, J.F. and Barefoot, A. (1983). Measurements of
polycyclic aromatic hydrocarbons in ambient air particulates in
northern New England. In, Measurement and monitoring non-criteria
contaminant in air. APCA Specialty Pub. #SP-50, 538 pp.
STSC (1981). Statgraphics, Version 1.1. Rockville, MD.
Thrane, K.E. (1984). Application of cluster analysis to identify sources
of airborne polycyclic hydrocarbons. In, 77th Annual APCA meeting,
San. Fran., CA, #84-16.6.
TABLE-1
PHYSICAL DATA
COMPOUND
MP(°C)a,b BP(°C)*,b VP (ATM,25°C)
Benzo(k)fluorantheae
Benzo(j)fluoranthene
Benzo(b)fluoranthene
Benzo(a)pyrene
Benzo(e)pyreoe
Indeno(c,d)pyrene
252
252
252
252
252
275
276
300
215
166
167
179
179
481
480
496
493
534
542
590
i.acw a
-13
1.3x10 f?a
1.9x10 a
Benzo(g,h,i)perylene
Coronene
273
438
a - Santodonato et al (1981)
b - Grisner (1983)
c - Calculated fro* Mackay et al (1982)
491
-------
TABLE-2
FINAL FACTOR ANALYSIS MODEL - ELIZABETH
(ROTATED FACTOR LOADINGS >0.60)
FACTOR 1
FACTOR 2
FACTOR 3
FACTOR A
BkF
BaP
BghiP
ICDP
Cor
BeP
BjF
BbF
0.79
0.67
0.93
0.72
0.67
0.84
0.83
0.92
N-i' 1
TABLE-3
FINAL FACTOR ANALYSIS MODEL-CAMDEN
(ROTATED FACTOR LOADINGS, >0.60)
FACTOR 1
FACTOR 2
FACTOR 3
FACTOR 4
BkF
BaP
BghiP
ICDP
Cor
BeP
BjF
BbF
0.85
0.87
0.61
0.89
0.83
0.67
0.83
0.60
G.94
N=79
TABLE-4
FINAL FACTOR ANALYSIS MODEL-NEWARK
""(ROTATED FACTOR LOADINGS, >0.60)
BkF
BaP
BghiP
ICDP
Cor
BeP
BjF
BbF
FACTOR 1
0.91
0.91
0.76
0.83
0.92
0.88
FACTOR 2
0.55
0.93
FACTOR 3
0.90
N-83
492
-------
Figures
Figure 1 - Principle component biplot - Elizabeth
Figure 2 - Principle component biplot - Camden
a. without outlier removal
b. with outlier removal
Figure 3 - Principle component biplot - Newark
Key for Figures
1 - Benzo(k)fluoranthene
2 - Benio(jJfluoranthene
S - Benzo(bJfluoranthene
4 - Benzo(ajpyrene
5 - Benzo(e)pyrene
6 - Indeno(c,d)pyrene
7 - Benzo(g,hti)pereylene
8 - Coronene
493
-------
coma :
S. 50+
1. 00 +
50+
-£. 00+
7+0 Pi
£«2 A
-1. 80
-2. 70
+COIHD £
1. 60
Figure 1
como 1
3. 00+
. 00 +
-2.0©+ I
flc!4333A
-A.0«+
+—
-£. 50
-1. 10
. 30
1.70
3. 10
+COWB 2
4.50
Figure 2a
494
-------
COMO 1
4. aa-f
,00 +
-4. 00+
3+6ft
-8.00+
-9.00 -6.0® -3.00
—+ + +comp 2
. 00 3. 09 6. 00
Figure 2b
cot1 ¦? :
4. 00+
s«+
-3. 01?+
•£.5®+
hh35h
ft ft V
-6. 00
-4.00
-2.00
8. 00
4.
'+eomp
Figure 3
495
-------
Use of a Rule-Building Eipert System for Classifying
Single Particles Based on SEN Analysis
P.K. Hopke and D.S. Kim
Institute for Environmental Studies
University of Illinois at Urbana-Champaign
1005 W. Western Avenue, Urbana, IL 61801
and
R. Lee and G.S. Casuccio
Energy Technology Consultants
350 Hochberg Road, Monroeville, PA 15146
Computer-controlled scanning electron microscopy (CCSEM) permits the
characterization of size, shape, and composition of individual particles
and thus provides a rich source of data to identify the origin of ambient
airborne particles. The problem is how to make best use of this
information. The general procedure has been to assign each particle to
a class of similar particles based on its x-ray fluorescence spectrum.
The initial efforts developed the class characteristics and classification
rules in an empirical fashion. Recent studies have suggested that
greater specificity and precision in the subsequent class balance
analysis can be obtained if particle classes are more homogeneous. To
obtain the classification of large numbers of particles in an efficient
manner, the x-ray intensity data have been subjected to an agglomerative
hierarchical cluster analysis. The resulting groups of samples are
then used as candidate examples for a rule-building expert system that
provides a decision tree that can be used for subsequent particle
classification. An error analysis approach based on jackknifing has
also been developed. The use of this approach to characterize particle
source emissions will be presented.
496
-------
Introduction
Receptor modeling generally apportions airborne particulate matter
to sources through various statistical methods based on the bulk chemical
analysis of ambient samples. These methods have been important tools for
particulate control strategy over the past 15 years. Among these
techniques, the Chemical Mass Balance (CMB) has been known as a
conceptually powerful technique. However, it cannot be implemented
until accurate source profiles are available. In practice, a CMB
analysis loses power when the sources of particulate matter have similar
compositions. This limitation aan be solved by measuring additional
variables to obtain distinct differences among the source profiles.
Microscopic methods have been used to characterize individual
particles and to identify sources of ambient aerosol mass. Recently,
innovative computer-controlled scanning electron microscopy (CCSEM) has
provided individual particle information, including size parameters and
chemical compositions within a short analysis time so that a large
number of particles can be analyzed. A particle class balance (PCB),
analogous to CMB, is the combination of the statistical and the mioroscoplc
methods and assumes a linearly additive sum of class mass fractions
times a fractional contribution of aerosol mass by the particulate
sources. Johnson and Mclntyre used this technique for aerosol
apportionment study in Syracuse, N.Y. They used twelve source types
from 47 possible source signatures and fitted to the 23 ambient samples
measured. To do this, they created only 25 separate classes of particles
based on the observed relative net x-ray intensities. As they mentioned,
there is a possibility of overlap among 25 particle classes when comparing
signatures of similar sources. Casuccio and Janocko also suggested a
particle balance model based on the segregation of the source and the
ambient samples into classes within various aerodynamia size categories.
The most important thing in the PCB model is to oreate homogeneous
particle olasses to characterize the source profiles. To define the
particle class membership, an agglomerative hierarchical clustering
analysis was used to determine the potential olass membership. Then
chemically homogeneous partiole classes were oreated using a rule-building
expert system by removing outliers from each partiole olass. The mass
fraction of particles belonging to these homogeneous particle classes can
then be used as the source profile.
El Paso Airshed
The City of CI Paso is looated at the tip of western Texas, with the
state of Hew Mexico to the west. There are problems of noncompliance
with both the TSP and lead ambient air quality standards3. The Rio
Grande River flows south of the city, which is the boundary between the
United States and Mexioo. The oity of Ciudad Juarez lies Just to the
south in Mexico across the river. El Paso also has the Franklin Mountains
Intruding from the north,
CCSEM was selected as the primary analytioal method to study source
and ambient partiole samples. Several other methods, Including proton-
induced x-ray emission, ion chromatography, and atonio absorption were
used to verify the CCSEM results. Eaoh source or ambient sample had
more than 700 particles analyzed and eaoh partiole Is oharaoterized by
25 variables; 6 size variables and 19 x-ray intensities. Twenty-one
source samples were used to obtain source composition profiles. More
497
-------
detailed descriptions for analytical methods are available in the
literature
Cluster Analysis
The basic concept of classification is to identify the objects (cases)
that are similar in a pattern space based on a measure of dissimilarity
or distance. The agglomerative hierarchical algorithm, AGCLUS , was
selected for the classification of the CCSEM data since the hierarchical
method is simple in terms of computational scheme, and is flexible in
terms of decision of significant clusters when comparing it with non-
hierarchical methods. The agglomerative method starts from the individual
objects in separate clusters and ends with a single cluster containing
all of the objects. Larger clusters are constructed by merging smaller
clusters. This method requires that once an assignment of objects to a
cluster is made, those objects will remain together at higher and higher
clustering levels. Hierarchical clustering is represented in the form
of tree-shaped dendrograms. Non-hierarchical clustering was used to
confirm the agglomerative analyses'.
It is then necessary to define a "homogeneous" particle class, A
class is determined to be homogeneous by examining the presence or absence
of the related chemical elements for all particles. When all particles
in a class show only the same elements, this class is referred to as
homogeneous. A particle in Al-Si-Fe class will show only Al, Si, and Fe
x-ray intensities. In the sense of this concept, all particles initially
clustered together from cluster analysis can be examined, and thus it is
possible to remove outlier particles for each class.
Expert System as a Classification Tool
A knowledge-based expert system, a class of high performance computer
programs in the area of artificial intelligence, was used to build a
decision tree from examples of particle classes following classification
studies. EX-TRAN® is a series of programs designed to generate a set
of rules based on examples for which various variables are known and
which have known outcomes. It produces a self-contained FORTRAN program
that can implement these rules. EX-TRAN searches the features one at a
time to identify the one for which it can "best" separate one class from
the others based on an extension of Quinlan's ID3 algorithm'.
To perform this study, "representative" examples were obtained from
the homogeneous particle classes. From the AGCLUS cluster analysis, each
data set representing one of source samples was split into 23 to 67
homogeneous particle classes. A total of 11,294 (73t) out of 15,499
particles were placed into one of the homogeneous classes so that they
could be used as examples for expert system implementation. Table I
shows the total number of particles, the total number of particles placed
in homogeneous classes, and the number of homogeneous classes for each
source data set.
In order to create a classification rule from the expert system, the
283 homogeneous particle classes from the 21 source data set were created
after deleting outlier particles. Table £1 shows only a portion of the
homogeneous classes, those which are all the sodium containing particle
classes observed in the individual source samples. As can be seen, some
classes were unique, i.e., observed in only a particular source, and
other classes were observed in two or more sources. Among the whole 283
498
-------
classes, some classes such as Na-Al-Si-Ca» Mg-Al-Si-Ca, Mg-Si-Ca, Al-Si-
K, Al-Si-K-Ca, etc. were frequently observed in more than 12 sources.
Particularly, the Al-Si-K class was the most commonly observed class in
17 out of the 21 source samples.
The derived rules were extensively tested by examining all 15,^99
particles of source samples. Only two <0.01%) out of the total particles
were misclassified. These misclassifications oocurred when the number
of representative examples in a class was much less than that of the active
variables. During the test of the decision rule, it was found that the
classification rules could also detect the misclassification of the
hierarchical cluster analysis, AGCLUS. In the sample T240 from Table I,
501 particles were assigned to homogeneous classes. However, three
Unassigned particles were revealed to belong to one of the homogeneous
classes. In this manner, the total 48 particles from the 21 source
samples were misclassified by the hierarchical cluster analysis. The
misclassified examples were reassigned to homogeneous particle classes
and the information for each class was corrected.
Mass fractions for each souroe sample were calculated. The full set
of source profiles are available on request from the authors. A simple
correlation study showed that profiles for the 21 sources were highly
independent of each other, even for the 7 fugitive soil samples. An
ambient aerosol mass apportionment study will be performed using these
source profiles.
The uncertainties of each class for all the source samples were
calculated. The jackknife approach has been applied in order to estimate
the uncertainties in the mass fractions. One data point at a time is
omitted for all the data points, and the mass fraction is then determined.
From the standrad deviation of these values an uncertainty is estimated.
The method can give a measure of variability regardless of its distribution
function10. Tukey's Jackknife estimate of standard error1^ was used to
estimate the uncertainty of mass fraction in each class.
Generally, as the number of particles in a oertain class increased,
the uncertainty decreased. For a class containing one or two particles,
100$ of mass fraction was used arbitrarily sinoe the method of omitting
one data point cannot be applied. The determination of uncertainty is
essential in the sense of providing credibility of each class membership.
Conclusion
Twenty-one source samples from the El Paso Quantitative Microscopy
Study were explored to define the homogeneous particle classes. Following
cluster analysis, 283 classes were created after deleting outliers. A
decision rule was developed by an expert system and used to check and
correct particle membership. A PCB can give increased source resolution
over that available with bulk chemical data since there are many more
variables available. It reduaes the possibllty of collinearity problems.
Acknowledgements
The work at the University of Illinois is supported by the National
Science Foundation under grant ATM 85-20533*
499
-------
References
1. D.L. Johnson and B.L. Mclntyre, A Particle Class Balance Receptor
Model for Aerosol Apportionment in Syracuse, N.Y., in Receptor Models
Applied to Contemporary Pollution Problems, SP-48, APCA, Pittsburgh,
PA (1982).
2. G.S. Casuccio and P.B. Janocko, Quantifying Blast Furnace Cast House
Contributions to TSP Using a Unique Receptor Model, Paper No*
81-64.2, APCA, Pittsburgh, PA C1981).
3. J.M. Wierseraa, L. Wright, B. Rogers, R. Berta, L. Haeuser, and J.H.
Price, Human Exposure to Potentially Toxic Elements through Ambient
Air in Texas, Paper No. 84-1.2, APCA, Pittsburgh, PA (1984).
4. S. Dattner, S. Mgebroff, G.S. Casuccio, and P. Janocko, Identifying
the Sources of TSP and Lead in El Paso Using Microscopy and Receptor
Modelsi presented at the 76th Annual Meeting of the APCA, Atlanta,
GA (1983).
5. ETC, Identification of the Sources of TSP and Particulate Lead in
the El Paso Area by Quantitative Microscopic Analysis, Vol. I and
Vol. II, Final Report Prepared for the TACB, Austin, TX, Energy
Technology Consultants, Monroeville, PA (1983).
6. D.C. Oliver, Aggregative Hierarchical Clustering Program Write-up,
Preliminary version, National Bureau of Economic Research, Cambridge,
MA (1973).
7. D.S. Kim, P.K. Hopke, D.L. Massart, L. Kaufman, and G.S. Casuccio,
Multivariate Analysis of CCSEM Auto Emission Data, Scl, Total
Environ. 59:141-155 (1987).
8. Intelligent Terminals Ltd., EX-TRAN 7 Users Manual, United Kingdom
(1984).
9. J.R. Quinlan, Learning Efficient Classification Procedures and their
Application to Chess End Games, in Machine Learning, An Artificial
Intelligence Approach, R.S. Michalski, J.G. Carbonell, and T.M.
Mitchell, eds., Tioga Publishing Co., Palo Alto, CA (1983).
10, F. Mosteller, The Jaokknife, Review of the International Statistical
Institute 39(3):363-368 (1971).
11. R.G. Miller, The Jackknife - A Review, Biometrica 61:1-17 (1974).
500
-------
Table I* Source Characterization Jtoaulta for
El Paso Classification Stub/
Ho. or
Hp. of Classified
No. of
Sample ID
Description
Particles
Partloles
Classes
T2J9
UTfPfsoll near monitor)
735
622
31
T220
American Minerals Sample
735
*38
62
Crtn Snelter)
rsw
Highway Enl salon (San Antonio!
735
401
40
T23*
El faso hojk Qvtrry
735
451
i*
Taj5
AS-A3C0 Una Beghc-usa
735
595
18
T235
A&AftCO Copper feverberatory
720
H7)
17
Furnace
T2J7
USAtiCO Converter Baghouse
735
374
53
T238
ASAECO Zinc Plant Baghousa
735
534
H
T2j9
IB 4 WC {near nonltor)
735
421
13
T210
ASAJiCO Lead Blast Furnaoe
735
sot
33
1212
01 t-t Rtwd, Aacerete/Valenaia
735
ioi
m
T2M6
Parlors Brothers ISlag
735
5BM
*3
Crushing Operation
61 a
T256
mm* Etfifl Refinery
750
Ml
tS42(
t:w
2}
Na-Kj-Jl- S- K
12631
O
Ha-dg-Sl- K-Ct
t2}«(
2)
— S i —Ce
t im
*>
tS3Ht
23
tS39(
«)
t2M2(
2)
ka-Hg-3l-Ct-Pb
t!70<
2»
fta-Hg>U-Pt>
t26i(
*>
t26«(
71
t!70(
t!71<
11)
Ma-Mg-Ca
ttiU
1)
na-Ng-re
tJ6»<
5)
t!70(
7)
tS71(
JU-AU&i
t219<
7)
t220C
5)
t239(
10 >
t2H>(
6) tSMBi
»>
t?54(
6)
t26B<
9)
t272I
1D>
Ma-*l~Sl- &
taut
81
Na-Al-SL- 3- K-fa
t!66!
O
W-Al-SI- S-Ce
C221C
7)
t266C
2)
Ml-U- 3-Ca-Ta
tztec
5)
S-fe
Mil
&>
Ni-Jtl-Sl- K
t2l9(
HI
tI2C-(
S)
1221 (
2)
t23«(
«) t!39(
U)
t242(
13)
WW
1)
out
t)
62661
7>
Slttt
9) 62721'
7^1}
*•-11-51- K-Ca
t219(
7)
tS39<
11)
tjtil
*)
t»tl
0 tns (
()
w?a<
«3
«T3(
S)
Na-Al-Sl- K-Ca-fe
1221 (
3)
t tm
11
ti«i(
7)
t24«<
«> t»M(
3)
1272 <
t)
Ha-A1-31* X-Fe
t221f
5)
t£39(
10)
t!4!l
7)
t256<
S> t2t((
3)
tJ72(
3)
Ha-Al-Sl-C*
t219<
19)
t320(
1)
t!21(
H>
t2]4<
6) t239(
16)
t2fl2(
15)
4)
t256<
3)
tS46(
«>
ti63(
13) t272(
15)
t273(
1)
Ma*Al-3l-Ca-Fe
«19<
3>
1239 <
»)
t242(
2>
tm<
3> ti}6(
S)
t273<
«s-W-Jl-Pe
t2J9<
«>
t256(
5)
t266C
2)
*a-«-ai-p»«*i
tttOl
3>
Nt-Al-Sl-Fb
tSSl<
3)
t!71(
ii
Ht-il- !
t!J6{
tl
t!«3(
*>
Na-fll- S- K-Cs
tm!
41
Hi-SI- 3-Ct,
M»2(
3)
»25«
»)
t2S3i
#!
t26i£
S) t272{
3)
Ha-Si- K-C*
«JT(
2)
Mt-Sl- *-Pb
t235C
5)
t!7P(
3)
m-Si-Ca
t2l9<
121
t£2H
«
12 311
17)
t239(
15) tJ56(
2>
t272(
1)
1273 <
#)
Sa-M-6a-?«
tsuai
9>
Ma«Sl-C«-?*.Kn
ta*a<
31
Na-Sl-Fi-W
u»(
5)
Ka-tl-Pb
»36(
tS7^(
5)
10)
tS37(
1)
t»3(
7)
t2S1(
») t2M<
3)
M7«(
10)
Ma- S-Ca
t»j(
10)
t26«(
101
t27K
5)
>1- S-Tt
t»3(
2)
Ka- K-Pb
«35<
t)
Sa-Fe-Hn
t2J« t
i>
Na-Fe-Pb
11
501
-------
A COMPOSITE RECEPTOR METHOD APPLIED TO PHILADELPHIA AEROSOL
Glen E. Gordon, Ilhan Olmez*, and Ann E. Sheffield
Department of Chemistry and Biochemistry
University of Maryland, College Park, MD 20742
Thomas G. Dzubay and Robert K. Stevens
Atmospheric Sciences Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
William J. Courtney**
Northrop Services Inc., Research Triangle Park, NC 27709
A new receptor method that is a composite of chemical mass balances
(CMB), multiple linear regression, and wind trajectory receptor models was
developed to apportion particulate mass into source categories. It was
applied to 156 aerosol samples collected in dichotomous samplers at three
sites in the Philadelphia area and analyzed by x-ray fluorescence (XRF),
instrumental neutron activation (INAA), ion chromatograpy (IC) and pyrolysis.
Major components of mass of <10-um diam particles were 50-55X sulfate plus
related water and ions, 16-27X crustal matter, and 4-7X vehicle exhaust.
Less than 5% was attributed to primary emissions from stationary sources.
Wind-stratified data indicated that 80 ± 20X of the sulfate was from a
regional background. Multiple linear regression attributed 72 ± 8 and 16 ±
5% of S to coal- and residual oil-fired power plants, respectively.
Introduction
The CMB model is useful for apportioning the mass concentration of
suspended particles. However, a difficulty is that the required signatures
are not known accurately for sulfate and vehicle exhaust. For sulfate, there
is an uncertain amount of water. For vehicle-exhaust, the Pb-abundance is
difficult to specify in the United States because measurements are sparse,
and the Pb content in gasoline has been declining rapidly in recent years.
We explore the feasibility of using a composite receptor method to
overcome this difficulty. It was tested using data from a summer, 1982,
study when ambient aerosol was collected for 52 12-h periods in the PM-10
size range at three sites (Figure 1) in the Philadelphia area1. The ambient
samples were collected by dichotomous samplers and analyzed by XRF, INAA, IC
and pyrolysis. Emission source and resuspended soil samples were also
collected by dichotomous samplers and analyzed2,J.
502
-------
We present a mass apportionment baaed on the following steps: (a) Wind-
trajectory analysis4 vas used to identify sources to include in the CMB. (b)
A new method consisting of a composite of CMB and multiple linear regression
(MLR) vas used to determine abundances of Fb in vehicle exhaust and S in
regional sulfate, (c) CMB, based in part on results of the composite method,
vas used to apportion mass Into 9 components, (d) Wind-direction
stratification vas used to detect the relative influences of local and
regional sources of S. (e) MLR of S vs Se and V vas used to estimate
particulate S contributions from coal and oil.
Results and Discussion
Precision and Accuracy. Table I shows correlation coefficients and a
comparison of mean values obtained by XKF, IC and INAA. The paired data are
highly correlated, and the ratios of means agree vithin overall errors for S,
Mn, Fe, Zn, and Sb. The agreement betveen IC and XRF results for S indicate
that sulfate, the species measured by IC, occurred as S.
Wind-Trajectory Method. We used a wind trajectory method4 to select
sources to include in our CMB. Resultant wind direction 6 and standard
deviation 68 vere computed for each 12-h period from hourly data on surface
vlnds. Also, the mean concentration x and standard deviation s vere
calculated for each species. The vind-trajectory method selected events
having &Q £ 20° and x £ x + 2s. Emission sources vere identified when
element abundance ratios matched those in source signatures3. Sources
selected vere a paint pigment plant and a Sb roaster, both south southvest of
Site 28, and municipal incinerators northwest of Site 28. Also selected vere
residual oil-burning and,catalytic cracker emissions2.
Composite of CMB and MU*. In a study in Denver during vinter, Levis et
al. used MLR to determine abundances of elements in vehicle exhaust, wood
smoke, sulfate, and crustal matter5. Because atmospheric K, their tracer
for vood smoke, included a contribution from soil, they used the measured K
concentration minus a correction for soil as a wood-smoke tracer. We applied
a modified form of that technique to Philadelphia aerosol. Modification vas
needed for Pb from vehicle exhaust because a significant portion of Pb
originated from at least two additional sourcesi incinerators and combustion
of residual oil. In our method, non-vehicular Pb vas determined by CMB and
vas subtracted from the measured Pb to yield a tracer for vehicle exhaust.
The CMB method expresses concentration C. of species i as
ci"fAijMj» (1)
vhere AtJ Is the abundance of spades i in component j, and Mj is the total
mass concentration attributed to component j. The unknown valQes H.. are
found by that minimises the expression',
vhere 0t is the observed concentration of species 1, and Ei 3 is the effective
variance, which includes the uncertainties in 0i and A1;).
For each species i in eq 2, two criteria must be met: (a) all major
sources j of the species must be included in eq 1, and (b) its abundance in
each component must be knovn. For our data these criteria vere met for a
fine-particle CMB based on Al, Si» Ca, Ti, V, Ni, Zn, Cd, Sn, Sb, La, Ce and
Sm in the following six components! municipal incinerators*, catalytic
crackers1, residual oil combustion3, Sb roaster2, paint pigment
manufacturing4, and local dust*. Using these 13 elements and 6 source ve
applied the CMB model to data for each 12-h sampling period to obtain 156
sets of M... Then, for mass, S and Pb ve calculated 0>' as'
X" - I l(Ct - OjJ/ljl"
(2)
O'i1'- «1
(3)
503
-------
where Qi' represents |M'], IS'], and [Pb'] which are the portions of observed
concentrations due only to vehicle exhaust and sulfate. Ve applied MLR to
all 156 observations and obtained R » 0.95 (multiple correlation) and
[M'] - 3592 ± 775 ng/m3 + (4.92 ± 0.13)[S'] + (15.0 ± 3.2)[Pb'] (4)
Ve interpret the reciprocal of regression coefficient of [S'] in eq 4 as
the abundance of S in the regional sulfate component, 20.3 ± 1.42. Vater
retained on acidic sulfate particles during laboratory mass analysis may
explain S abundances lover than 24.2% and 26.IX, for S in pure (NH.)2S04 and
NH^HSO,, respectively. Likewise, eq 4 indicates a 6.7 ± 1.5% abundance of Pb
in vehicle exhaust. This falls within the range of results obtained by other
investigators: 6.9% for a Pennsylvania Turnpike tunnel during 19777, and
4.4 ± 0.6* for Denver during winter, 19825.
Mass Apportionment. Fine particle mass was apportioned by CMB into 8
components as in Table II. In addition to the 6 components used in the
preliminary CKB, vehicle exhaust and regional sulfate were included. For
vehicle exhaust, the Pb abundance of 6.7 ± 1.5% was derived from eq 4.
Abundances of other elements in vehicle exhaust (S, Cr, Hn, Fe, Ni and Br)
were taken from Pennsylvania Turnpike tunnel data7 adjusted for the vehicle
mix in Philadelphia. The regional sulfate component included the 20.4 ± 1.4%
S- abundance derived from eq 4. Concentrations of other elements were from
the regional sulfate signature of ref 8 normalized to our S value.
Except for CI, Br, As and Se, the measured and calculated element
concentrations agree fairly well. The model overestimated CI and Br values
because it does not account for losses by volatilization. The regional
sulfate component represents most of the calculated As and Se, but the
signature derived from Shennandoah Valley data8 appears not exactly to
represent As and Se Philadelphia. Deviations for such elements has little
effect on our apportionment of mass.
Ve applied CMB as in Table II to each sample. Averages for each site
were computed and are shown in Table III. Also indicated are estimates of
possible bias due to uncertain abundances of key elements in the source
signatures. Results for the dust signature are labeled "minerals and-coal
fly ash" in Table III because such subcomponents were not resolved. The dust
signature1 that we used is a preliminary one, and, therefore, a large value
of possible bias is shown in Table III. The component labeled "other" is the
Intercept in eq 4 and may represent components not included in the CMB.
The coarse fraction CMB results are based on five components shown in
Table III. Because, we did not constrain our CMB solutions to positive
values, negative values below detection limit were sometimes obtained as
coarse particle results in Table III indicate.
Sources of Particulate Sulfur. Average species concentrations were
calculated for cases of steady wind from NNE and SSV along our line of
samplers. Sampling periods were included if data were available for both
Sites 12 and 34 and if at least 75% of the hourly wind directions were in the
SSV (165° to 255°) or NNE (245° to 75°) quadrants. The ratio of average
upwind to downwind S-concentrations for Sites 12 and 34 was 0.80 ± 0.18.
Because Sites 12 and 34 are at the edges of the Philadelphia urban area, such
ratios suggest that local sources contribute about 20% of the particulate S
and sources beyond the Philadelphia area contribute 80%. In an earlier
study, local and regional components were deduced from sulfate measured at 45
stations in eastern U. S. between 1963 and 1978*. Early in that period,
local sources contributed 45% of total sulfate during summer months. By 1978
the local component had decreased to 25% due to controls on S emissions
applied mainly in urban areas*.
504
-------
Emission inventories for Eastern states indicate that about 90% of the
SOj, the precursor for particulate S, is emitted from coal- and residual oil-
fired power plants. Assuming that Se and V are tracers for coal- and
residual oil-fired pover plants, respectively, ve used MLR to express fine
particle S as
[S] - 380 ± 355 ng/m1 + (1791 ± 201)|Se] + (67 ± 21)(V] (5)
Multiple correlations were R « 0.67 for 156 samples. Because S is
emitted mainly as S0a, atnospheric reactions are required to produce
particulate S. Variability in the fraction of S0a converted to SO could
account for correlations that are only moderately high. Results obtained
by dividing the mean of each tern in eq 5 by the mean fine particle S
concentration indicate that the particulate 5-contributions from coal- and
oil-fired pover plants are 72 ± 8JC and 16 ± 5X, respectively.
Summary
The composite method vas a valuable supplement to the CNB model for
analyzing the Philadelphia data set. The composite method enabled us to
estimate the average Pb abundance in vehicle exhaust when incinerators and
residual oil combustion contributed a portion of the Pb. The combined use of
XRF and INAA enabled us to resolve more sources than possible if only one of
those methods were used. The former provided essential data on Si, S, Nlt
and Pb; the latter provided essential data on Al, V, Se, La, Cs, and Sm.
The wind-trajectory method identified four categories of stationary sources
to include in the CMB. Interpretations reported here are a fev of the
possibilities other investigators may attempt. One may obtain our data on
IBM-PC readable disks by sending one 1.2-Mbyte or tvo 360-kbyte floppy disks
to the principal author.
Although the research described in this article has been funded by the
United States Environmental Protection Agency, 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 commercial
products does not constitute endorsement by U. S. EPA.
References
1. T.G. Dzubay, R.K. Stevens, G.E. Gordon, A.E. Sheffield, I.Olmez, R,
Courtney, Sci. and Technol., submitted (1987).
2. I. Olmez, A.E. Sheffield, G.E. Gordon, J.£. Houck, L.C. Pritchett, J.A.
Cooper, T.G. Dzubay, R.L. Bennett, J.Air Pollut. Cont. Assoc.
submitted (1986).
3. S.A, Batterman, T.G. Dzubay, R.E. Baumgardner, D,L. Johnson, Soil, Air,
and Water Pollut. to be submitted,
S.V. Rheingrover, G.E. Gordon, Aerosol Sci. Technol., in press <1987).
5. C.V. Lewis, R.E. Baumgardner, R.K, Stevens, G.K. Russvurm, Environ.
Sci.Technol. 20; 1126 (1986).
6. J.G. Watson, J.A. Cooper, J.J. Huntzicker, Atmos. Environ. 18I1347 (1984).
7. V.R. Pierson, V.V. Brachaczek, Aerosol Sci. Technol. 226il (1983).
8. S.G. Tuncel, I. Olmez, J.R. Parrington, G.E. Gordon, R.K. Stevens,
Environ. Sci. Technol. 19>529 (1985).
9. A.P. Altshuller, Environ. Sci. Technol. 14>1337 (1980).
'Present address! Massachusetts Institute of Technology, Cambridge, MA
"Present address: International Business Machines Corporation, Dallas, TX
505
-------
Table I. Comparison of ZRF, INAA, and IC Results for 50 Paired Fine Particle
Samples fro* Caaden, NJ {Site 28) between July 14 and August 13, 1982*.
XRF mean Other mean Results Ratio of means
ng/m3
ng/m3
(XRF/other)
R
s
4063
±
20
IC
3719 ±
70
1.09 ± 0.09
0.97
s
4196
±
20
INAA
3863 ±
900
1.09 ± 0.14
0.92
K
101
±
3
INAA
144 ±
<15
0.70 ± 0.10
0.94
Ca
40
±
5
INAA
49 ±
18
0.81 ± 0.11
0.65
V
19
±
4
INAA
13 ±
<1
1.43 ± 0.18
0.86
Mn
6
±
2.3
INAA
6 ±
<0.6
1.09 ± 0.15
0.72
Fe
91
±
3
INAA
94 ±
<10
0.97 ± 0.12
0.89
Zn
82
±
1.5
INAA
92 ±
<9
0.89 ± 0.11
0.997
Se
1.2
±
0.6
INAA
1.8 ±
<0.2
0.67 ± 0.10
0.70
Br
29
±
0.8
INAA
23 ±
<2
1.27 ± 0.16
0.97
Sb
73
±
5
INAA
79 ±
<8
0.93 ± 0.11
0.998
Table II. Mass Balance of Average Fine Particle Composition at (Site 28)a
Meas-
Res.
Paint
Cat.
Munic.
Veh.
Sb-
Regii
Calculated
ured.
Dust
Oil
Pig.
Crack.
Incin.
Exh.
Roast
S
V-C
1027
±
1128
2046
67
15
0
1
4
939
1
0
NV-C
1153
±
584
1867
10
29
0
0
25
1089
0
0
Na
94
±
30
146
12
21
0
1
44
16
0
0
A1 b
46
±
10
53
23
5
2
14
2
0
0
0
Si b
137
±
32
103
65
16
0
43
12
0
0
0
S b
4197
±
403
4196
1
87
0
27
21
47
1
4012
CI
224
±
33
3
1
0
0
0
194
30
0
0
K
63
±
18
" 101
6
1
2
0
54
0
0
0
Ca b
38
±
8
40
12
20
4
0
2
0
0
0
Ti b
17
±
9
15
2
1
13
1
0
0
0
0
V b
11
±
3
13
0
11
0
0
0
0
0
0
Mn
4
±
1
6
1
0
0
0
0
1
0
2
Fe
122
±
11
91
92
11
9
0
2
8
0
0
Co*
684
±
141
483
11
669
0
3
2
0
0
0
Ni b
13
±
3
11
0
12
0
0
0
0
0
0
Zn b
83
±
18
82
1
7
0
0
70
3
0
1
Br
69
±
35
29
0
0
1
0
2
67
0
0
As*
209
±
219
632
0
30
0
0
47
0
0
132
Se*
954
±
233
1800
67
14
0
2
29
0
0
842
Mo*
645
±
396
900
1
372
0
0
272
0
0
0
Cd
2
±
1
2
1
0
0
0
I
0
0
0
Sn b
6
±
4
5
0
0
0
0
5
0
1
0
Sb b
79
±
9
79
0
0
0
0
1
0
77
0
La* b
871
±
187
1120
13
195
0
662
1
0
0
0
Ce* b
712
160
609
26
152
0
534
0
0
0
0
Sm* b
45
±
9
40
3
8
0
34
0
0
0
0
Pb b
249
t
62
249
3
10
0
0
42
179
0
14
Mass
24910
28282
742
723
59
214
717
2683
104
19668
* An asterisk^*) denotes data in pg/mJ} all other data are in ng/m1.
bElements Included in least-squares fit.
506
-------
Table III. Average Mass Balance for July 14 to August 13, 1982, Deduced by
Composite of CMB and HLR Methods.
Size
Possible
Component !
Fraction
Site 12
Site 28
Site 34
Bias, X
Minerals and coal fly ash
fine
1066
751
±
36
597
40
Residual oil fly ash
fine
489
685
±
20
427
30
Ti-rich paint pigment
fine
40
54
±
6
4
30
Fluidized catalyst cracker
fine
144
199
±
6
96
30
Municipal incinerators
fine
503
651
±
31
190
30
Antimony roaster
fine
4
94
±
3
2
10
Motor vehicle exhaust
fine
2388
2757
163
1511
20
Sulfate, cations and water
fine
18635
19396
±
318
18537
10
Other (MLR intercept)
fine
3592
3592
±
775
3592
Minerals and coal fly ash
coarse
7264
9033
±
192
4730
40
Marine
coarse
160
228
±
26
113
50
Sulfate
coarse
440
595
±
67
718
30
Municipal incinerators
coarse
54
-38
±
26
-42
50
Antimony roaster
coarse
15
231
±
14
7
10
Calculated total
fine
26862
28179
±
855
24958
Calculated total
coarse
7934
10048
±
207
5527
Calculated total
PM-10
34796
38227
30485
Measured total
PM-10
34868
39715
33858
PA TURNPIKE
STEEL
WIRE
PVC
INCINERATOR
PHILADELPHIA
OPPER SMELTER
¦ aiai uiuni tun4.
INCINERATOR
mineral!
CAMDEN
IVPjtUM
REFINERY#
GRAIN HANOLINI
INTERNAT^r
Z. AIRPORT
AA TITANIUM. ZINC
ANTIMONY A
'BRAS
A TYPE SHOWN ON MAP
¦ CATALYST CRACKER
• RESIDUAL OIL FIREO BOILER
O COAL FIREO 8CHLER
O AIR SAMPLING SITE
...•••FIBERGLASS
Figure 1. Map shoving sampling site locations and point sources having
particle emission rates above 50 Mg/y.
507
-------
DIFFICULTIES AND EFFICIENCIES ENCOUNTERED WITH A
BULK-AIR SAMPLING SCHEME FOR CHARACTERIZING
HEAVY GAS RELEASES OVER A GRASS-COVERED SITE
Glen A. Marotz
Dennis D. Lane
Ray E. Carter, Jr.
Richard Tripp
Department of Physics
and Astronomy
University of Kansas
Lawrence, KS 660M5
Department of Civil
Engineering
University of Kansas
Lawrence, KS 66015
Department of Civil
Engineering
University of Kansas
Lawrence, KS 660*15
U.S. Environmental
Protection Agency
Region VII, 25 Funston Rd
Kansas City, KS 66115
I
John Helvig
U.S. Environmental Protection Agency
Region VII, 25 Funston Rd
Kansas City, KS 66115
Assessing atmospheric composition using most currently available
techniques is often difficult and time consuming. Difficulties work against
detection and determination of possible toxic gas signatures from industrial
processes, landfills and other release situations. Time constraints
mitigate against rapid response in actually or potentially dangerous
circumstances. Recent work has shown that a bulk air technique can be used
to gather and characterize ambient air samples relatively rapidly. When
coupled with cryogenic-focused GC analysis, qualitative and quantitative
data can be obtained. An approach to characterization of heavy gas releases
using a combination of the whole-air technique, portable meteorological
equipment coupled with a fixed sampling network, and modeling was formulated
and evaluated on a grass-covered site over a six month period. In previous
work, preliminary results from two sample runs were described in which
downwind surface concentrations of two heavy gases were assessed
(volatilized toluene and methyl chloroform emitted at a fixed rate over
short time periods). In this paper, the analysis approach is extended to
data from two additional sample runs that were conducted under different
meteorological conditions. The goal is to further evaluate the three-part
approach as a rapidly deployable scheme for assessing toxic gas releases.
508
-------
Key Words: Heavy Gas, Whole Air Sampling, Field Data, Toluene, Methyl
Chloroform
INTRODUCTION
Characterization of potentially harmful gaseous releases from
industrial activities, or accidental emissions, require a timely response by
organizations with public health responsibilities. The species must be
known or identifiable, and some Judgements about spatial and temporal
patterns must be made in the shortest time period practicable. The problem
becomes more difficult if the gases released have known acute or toxic
effects; yet another layer of complexity is added if the emissions are
heavier than air. While trace gas releases oan be modeled using approaches
based upon generally accepted characterizations of the turbulent structure
of the atmosphere, heavier than air gases (HTAG) pose an additional problem
because they modify atmospheric flow characteristics for a period of time.
Few approaches are established for dealing with HTAG situations.
We have conducted evaluations of a possible set of sampling procedures
based on results from controlled HTAG releases. This set consists of a
gaseous analysis technique that utilizes whole-air sampling and cryogenic-
focused gas chromatography (GC), and possible applications of two HTAG
models for temporal and spatial release description. The results to this
point are summarized, and then the approach is extended to a second set of
release conditions and-aampling/analysia/modeling evaluations.
PRIOR WORK
Lane, et al.1, Pliel and McClenny*, and Tripp' have shown that whole-
air sampling Tbased on stainless steel spheres) and cryogenic-focused GC
analysis yields accurate and precise characterization of ambient volatile
organic compound (VOC) releases. Marotz, et al." hereafter referred to as
MLCTH) used the whole-air sampling/GC analysis technique, combined it with a
meteorological instrument package and a fixed-grid sampling network, and
conducted field evaluations of the approach using a known HTAG source
composition and release rate. Gases used were toluene and methyl chloroform
in a 50/50 liquid mixture by volume. Volatilized emissions were produoed
from a generator designed to operate at different emission rates. The
emission rate was varied between 110 and 200 ml min"1 for a series of eleven
evaluation runs of 30 minutes each spread over a six month period (July
through December, 19861 at a short-grass-covered site in eastern Kansas.
Meteorological and source strength/composition data from two of the
eleven test cases were used as input for HTAG model runs, which formed the
third part of the approach evaluated in MLCTH. Model output was briefly
compared with values obtained from the bulk-air sampling technique.
Based on preliminary analysis of principally surface-collected eamples
and model runs, MLCTH concluded that, overall, the approach yielded
relatively quick and precise results. Questions relative to extension of
the approach to a different set of meteorological conditions, the usefulness
of multiple spheres arrayed at locations other than the gaseous plume
centerline, the capability to model measured concentrations at "fcreatbeable
height", and the usefulness of the available models themselves remain open
and are partially treated in this paper.
509
-------
METHODOLOGY
The tripartite approach (whole-air sampler deployment, meteorological
instrumentation appropriateness, and modeling of results) was evaluated at a
large (approximately 250m by 250m), level, short grass-covered open area
situated to the northwest of two principal transportation arterials.
Meteorological instrumentation consisted of the fewest devices
necessary for provision of needed model input data. Assessed were wind
speed (2m), wind direction (2m), air temperature (1m), surface temperature,
relative humidity (2m), and barometric pressure. Additional data included
soundings (00/12 GMT) from Topeka, KS (40 km from the study site), upper air
(850 mb) and surface synoptic charts.
Available spheres (Demaray Scientific Instrument, Ltd, Model 0647, 6.0L
equipped with Whitey Microvalves, Model SS-21RS2) numbered 13, and were
positioned on the surface at distances of -25, 50, 75 and 100m from the VQC
generator at the intersections of a 30°/60° ray/arc combination (Fig. 1).
Ray directional orientation and angle were determined by the plume
centerline direction.
TM
Two models were used, CHARM (version 4.0) and DEGADIS, to extend
field sample-based characterization of the plume. Both have been tested
against field data, are written in computer interactive form, and are
commercially available. The theory and description for both are presented
in MLCTH, 00ms and Tenneckes5 and Spicer and Havens®.
Conditions for Runs 4 and 5 described here are shown on Table 1.
Stability conditions in both 1 and 5 were neutral to stable, and synoptic
conditions were similar.
RESULTS
Field Data
Surface concentrations ranged from background values to a maximum of
625 and 705 ppb for methyl chloroform and toluene, respectively (Fig. 2a-
2d); in both Runs the plume showed the semi-elliptical shape characteristic
of HTAG releases*. The larger minor axis, higher centerline concentrations
in Run 4 versus 5 for both gases appeared to be a function of four factors!
(a) wind directional* variability, which amounted to about a sixty percent
greater standard deviation value in 4 versus 5; (b) average wind speed,
which tended to maintain concentration along the centerline; (c) a greater
generator emission rate in 5 versus 4; and (d) ambient temperature
differences that may have assisted volatilization (Table 1). Both tests
occurred under similar stability regimes, and surface heating was not a
factor in aiding plume mixing.
Model Runs With Surface Data
Several points are pertinent to the comments in this section. Although
model estimates are presented in Table 2, they are meant for general
reference only. Neither model's predictive capabilities were evaluated In a
proper, complete, and rigorous mannerj to do so would have required much
more intensive sampling, a different sampler array, additional
meteorological information, and many more replications. The sampling
network was intended to provide data principally for the whole-air samplers,
510
-------
not for modeling purposes. The sampling site was large enough for the
former objective, but the limited fetch available may have prevented
complete adjustment of the wind profile to surface conditions, thereby
creating turbulence within the network itself, which could have altered
dispersion in an unpredictable manner. Furthermore, the sampling points
were all relatively close to the source given the way in which HTAG plumes
are gradually modified and disperse.* Indeed, the sampler locations fell
within the distance range where concentration values changed most rapidly
(eg., in Run 5 observed ppb concentrations for toluene fell 68 percent in
50M; Table 2). Rather, interest was in simply taking a brief look at the
role models and modeling might play as part of the approach described in
this paper. Experience does suggest that, overall, both HTAG formulation
efforts are flexible, sensitive to small changes in input values, and that
the potential user has wide latitude in directing the nature of the output
product. The user's application needs, knowledge, and experience with a
particular model will determine its ultimate value.
The user supplies data on a number of meteorological variables for both
models. DEGADIS requires direct input of the stability class, CHARM will
allow the same choice, or will estimate stability from cloud cover and solar
heating estimates. Stability class choice is significant because all
atmospheric mixing conditions are calculated from the stability conditions
selected. The user also must know what gas is being emitted in order to
either select from a,list, substitute one with similar properties, or
provide physical constants if the first two options are unavailable.
Physical constants were supplied to DEGADIS for both gases, a necessity
because the internally supplied values hold only for liquid natural gas.
CHARM™ had toluene on its internally compiled list, but ethelene dichloride
was chosen to represent methyl chloroform, because the latter gas was not on
the internal list. Finally, the source strength and physical
characteristics must be known accurately, or very good estimates must be
made.
Several DEGADIS runs were oompleted using input ohoices that varied
slightly but were based on reasonable judgements about release conditions;
the numbers in Table 2 were the lowest we could obtain without compromising
the model constraints. Ancillary data suggest that the plume shape is
accurately represented, and that concentration gradients are reasonable,
although values show consistent over prediction. DEGADIS provides plume
half width, deviation values, and other information about the emitted plume,
but it does not provide a graphic display,
CHARM yielded predicted values that were oloser to observed values near
the sourae generator in Run ^ for both gases; the displayed concentrations
were larger than observed past 50m for methyl chloroform (ethylene
dichloride}, CHARM also produced a graphio output display of plume width
and length, along with superimposed, user-chosen concentration isoplethio
values. The latter were set at, or close to, observed values at specific
points of interest for this paper, but suoh an approach mitigates against a
view of the concentration value range in the entire plume. The smallest
diameter area that can be displayed consists of a circle of 0.2 km radius,
which forms a background for the predicted plume overlay. Details at one-
fourth the latter radius were obtained by requesting specific concentration
value locations.
SU
-------
CONCLUSIONS
The results of this paper lead to the following conclusions:
The whole-air technique based on data collection using
stainless steel spheres fitted with microvalves appears to
work well; results can be obtained reasonably fast, and
produce accurate and precise representations of the release
situations.
The sampling network is easy to set-up and take-down, and the
arc/ray/height arrangement possibilities allow much
flexibility; the number of samplers and the field arrangement
directly effect the time spent in the field data collection
phase, and a balance must be subjectively reached between
quick response and adequate characterization of the release.
The meteorological equipment needed to provide model input
data need not be complicated or sophisticated; a good
continuous visual display of wind speed and directional data
is helpful as the sampling network is deployed.
In order to better address the applicability of models to
HTAG releases, more investigation of model approaches to the
HTAG problem, the critical Information needed for model
input, and the appropriate sampling ray all required.
REFERENCES
1. Lane, D., R, Carter, Jr. and G. Marotz, "Field Applicability and
Precision of a Whole-Air Sampling Method for Ambient Air Volatile
Organic Compound Determination," Proceedings. EPA/APCA Symposium
on Measurement of Toxic Air Pollutants, Raleigh, NC (1986).
2. Pliel, J. and W. McClenny, "Reduced-Temperature Preconcentration and
Gas Chromatographic Analysis of Ambient Upper-Phase Organic
Compounds: System Automation," Environmental Monitoring Systems
Laboratory, EPA, Research Triangle Park, NC. Internal Report
(1984).
3. Tripp, R., "Automatic Cryogenic Sampling and Gas Chromatographic
Analysis of 'Volatile Organic Compounds," EPA, Region VII, Kansas
City, KS. Internal Report (1984).
4. Marotz, G. , D. Lane, R. Carter, Jr., R. Tripp, and J. Helvig,
"Preliminary Results from a Rapid Deployment Field Study of Heavy
Gas Detection and Dispersion Using a Whole-Air Technique," APCA
(87-103.7) Eightieth Annual Meeting, New York, NY (1987).
5. Ooms, G. and H. Tennekes (eds.), Atmospheric Dispersion of Heavy Gases
and Small Particles. Berlin: Springer-Verlag (198HTT
6. Spicer, T. and J. Havens, "Development of a Heavier-Than-Alr Dispersion
Model for the u. S. Coast Guard Hazard Assessment Computer
System," in S. Hartwig (ed.) Heavy Gas and Risk Assessment-Ill,
Dordrecht: R. Reidel Publishing Co. (T954TT~PP< 74-121,
512
-------
Table 1. Selected Data for Sample Runs
Run Number
Variable 4 5
Date 7-23 11-6
Air T ("C) 29.1 8.6
Surface T (°C) 27.3 7.8
FtH (?) 69 80
Pressure (atm) 1.007 1.003
Wind Direction 345 191
°WD 25 15
Wind Speed (m/seo) 1.1 2.5
Start Time (CST) 9:10 9s12
Arc Angle with centerline ( ") 30/60 30/60
Emission Rate 110 ml/min 150 ml/mln
Table 2
Model Estimated Versus Measured Surface
Concentration Values (ppb) at Centerline Location
Toluene
Run 1 Run 5
Location
Degadis
Measured
TM
CHARM
DEGADIS
Measured
CHARM™
100 M
31821
81
>10, 967M*
2012
225
>200, 71M
75M
5581
150
MOO, 195M
3198
371
>300, 55M
50M
12150
321
>300, 81M
7721
705
>700, 31M
Methyl Chloroform'
Run 1 Run 5
Location
Degadis
Measured
TM
CHARM
DEGADIS
Measured
CHARM™
100 M
3388
70
1871
193
>100, 69M
75M
5960
125
3276
322
>300, 11M
50M
13319
277
>300, 511M
7523
625
>600, 21M
1 Interpolated ppb values
TM
1 k specified concentration value was used as an input to CHARM and the
linear occurrence of that value along the centerline la indicated in
meters
' Ethylene dichloride was substituted for methyl chloroform in CHARM™
S13
-------
100 M
Colocated
Samplers
50 M
Wind Direction
Background
Pig, 1, Schematic Diagram of th« Sampling Network
514
-------
TEST #4
//A
7 A
*.tr
M
IU
Figure 2 a.
u
#4
Figure 2 b.
t ¦
- 75 nl/min
t}ln
U
Figure 2 c.
TEST #S
<•
•25
R 9t4* * V«*
6.48 flpfi
<#
W<
Figure 2 d.
-------
MEASUREMENT, ASSESSMENT AND CONTROL OF HAZARDOUS (TOXIC)
AIR CONTAMINANTS IN LANDFILL GAS EMISSIONS IN WISCONSIN
Julian D. Chazln, Mark Allen and Dennis D. Pippin
Air Monitoring Section
Bureau of A1r Management
Wisconsin Department of Natural Resources
Madison, WI 53707
Vinyl chloride, a known human carcinogen, was found to be present in
landfill gas (LFG) emitted from almost all closed cells at active landfills
or closed landfills which handled municipal waste. Landfills which handled
industrial wastes only (Including pulp and papermill sludge residues) were
free of vinyl chloride in the LFG. The concentrations of vinyl chloride
were found to be significant regardless of the type of venting system
employed. Benzene, another known human carcinogen, was detected In the LFG
of only 3 of 22 landfills studied and at much lower concentrations. Other
noncarcinogenlc hazardous air contaminants were detected but at emission
rates well below those being proposed for regulation under Wisconsin's Air
Management rules.
Measurement of the concentrations of vinyl chloride in ambient air
adjacent to the landfills was not possible due to limitations In the
ambient monitoring procedures used. However, modeling of the emissions
from the vents of several active gas extraction systems indicated ambient
Impacts on adjacent populations to be from 2 to 400 times greater than the
ambient exposure levels for vinyl chloride recommended by the Wisconsin's
Division of Health. As a result, a change In Wisconsin's rules for land
disposal of municipal solid waste is being proposed which would require the
collection and combustion (or adsorption on activated charcoal) of
hazardous air contaminants in landfill gas. The rule, in the series
governing solid wastes and landfills, essentially brings landfills emitting
gas under the same requirements as are being proposed for all point sources
of toxic air contaminants. The rule requires landfills to essentially meet
the requirements of Wisconsin's proposed new air toxics regulations.
516
-------
INTRODUCTION
Landfill gas, composed predominantly of carbon dioxide and methane, 1s
a byproduct of the decomposition of organic wastes deposited In landfills.
In addition to the major fixed gases, other trace gases are released from
decomposition of (or directly from) the waste Itself (for example,
solvents, materials of composition of the waste, hydrogen sulfide 1n
anaerobic decomposition, etc.). Analysis of landfill gas has been
undertaken across the nation and reports have been Issued on the results of
such analysis within Wisconsin as well as from other states.'
In 1985, the City of Madison Division of Health, responding to concerns
of citizens living In the vicinity of Its landfills, conducted some tests
for trace contaminants 1n the landfill gas at four landfills utilizing
length-of-staln (pull) detector tubes. Due to the Indicated presence at
some landfills of vinyl chloride, a known human carcinogen, the City of
Madison sought the assistance of the Wisconsin DNR. Tests were conducted
utilizing Tenax sorbent tubes. Unfortunately, problems were encountered In
the lab analyses of the Tenax tubes and no data could be obtained.
However, because Tenax does not adsorb vinyl chloride, separate tests
utilizing tubes containing activated charcoal were also conducted. The
charcoal tube results confirmed the observations made by the City.
Following the initial tests. It was decided to do a full-blown analysis of
one of the City of Madison landfills as well as a survey of all the
landfills in the city. The study was completed and the results Indicated
a more general problem than was originally believed.
In May, 1983, a Hazardous Emissions Task Force was appointed to provide
guidance to the Department on the regulation of "hazardous air emissions"
(toxic air pollutants). In July, 1985, they published a report designating
specific air contaminants for control and recommending a mechanism to do
so.' Soon after, the Bureau of A1r Management began to develop
regulations based on the recommendations of the task force; the rule
development was occurring simultaneously with the conduct of this study.
PHASE I - MONITORING STUDY
Study Design
In response to our request for assistance, DNR's Residuals Management
Section of our Bureau of Solid Waste Management submitted a list of
possible landfills for a statewide survey of landfills. The (municipal and
Industrial) landfills selected were: (1) either closed landfills or had
closed cells; (2) were known to be generating landfill gas; (3) were Within
1/2 mile of homes; and (4) the cells were closed within the last 30 years.
The area covered Included most of the populated areas of the state.
In addition, a comprehensive study was performed at one landfill in
Madison, the Mineral Point Road landfill. A sampling was conducted over a
period of five days, with the flow being lowered each day in the stack vent
at the tail end of the gas extraction system (by lowering the blower
velocity), from a maximum of about 800 DSCFM to a minimum of 400 DSCFM on
the fifth day. In addition, once flow was stabilized, six one-hour samples
were collected (some 1n duplicate) each day at the blower stack. For all
other landfills In the survey, only two to three samples (with some
duplicates) were collected at each sampling point with the sampling period
ranging from 60 to 80 minutes.
517
-------
Monitoring and Laboratory Analysis Methodology
It was decided that only source sampling would be conducted, whether
from active or passive sources of landfill gas. The sample collection
method included the use of Gillan personal sampling pump and activated
charcoal tubes and was further detailed In an Internal memorandum. The
analysis of samples was conducted by the State Laboratory of Hygiene's
Occupational Health Lab following the NIOSH methods3 with solvent
extraction of front and back halves of a primary and a backup tube,
followed by gas chromatography retention time with a flame Ionization
detector (FID). The FIO was periodically replaced by mass spectrometry as
the detector for confirmatory analysis for each species. In addition to
charcoal tube samples, a combustible gas meter was employed to detect the
presence and relative level of methane as an Indicator of LFG. At select
sites where active extraction systems are vented through stacks, stack
velocities were measured using a standard P1tot tube (for dispersion
modeling purposes).
Quality assurance for the sampling and analysis was provided by:
replicate sampling and analysis; the use of field and laboratory blanks;
the assessment of "breakthrough" on the sorbent tubes by front and back
half analysis (and the use of a "backup" tube); and the confirmatory
analysis by Mass Spectrometry, Flows on the personal sampling pumps were
calibrated by the specified method and were traceable to a primary standard.
Monitoring Results and Discussion
Between April 6, and August 22, 1986, 22 landfills were monitored
throughout Wisconsin. The results of the monitoring were presented 1n a
series of reports beginning on September 22, 1986. A summary of the
results was presented In a report dated September 26, 1986.* If the
detailed results for a specific landfill are desired, copies of the
individual reports for each site are available upon request from the
authors.
The results which included a wide range of volatile organic compounds,
Indicate that for noncarclnogens, the source concentrations are not
expected to exceed emission limitations established in the proposed rule
and when extrapolated to ambient concentrations (by using an arbitrary
1/500 dilution factor) are not expected to exceed 11 of the TLV. The
limiting contaminants were believed to be the carcinogens, benzene and
vinyl chloride. Benzene was found In very few landfills (3 out of 22) and
the concentrations detected were very low ranging from 0.4 ppm (just above
the 0.3 ppm detection limit) to 3 ppm (for a sample pulled directly from a
gas monitoring probe" sunk Into the waste mass). Further consideration of
benzene was put aside pending an evaluation of vinyl chloride.
Table I presents the results for vinyl chloride . The data are
presented in more or less descending order of concentration (while
retaining the classification system presented by the Bureau of Solid Waste
Management). Municipal landfills handling municipal (household) wastes are
the primary emitters of vinyl chloride. Some of these landfills emit no
vinyl chloride or the concentrations are very low. All the others had very
significant concentrations of vinyl chloride In the LFG. In addition,
municipal landfills which accepted predominantly Industrial wastes, and
Industrial landfills which accepted predominantly Industrial wastes or
p&permlll sludge (but no municipal wastes) were free of vinyl chloride
em1sslons.
518
-------
Conclusions - Phase I
It was concluded that vinyl chloride was the limiting hazardous air
contaminant In LFG. It was also concluded that vinyl chloride was present
only in municipal wastes. It was also concluded that, based on the direct
emissions measured and the preliminary estimates of ambient Impact for one
landfill, VC posed a potential threat to human health for persons living In
the vicinity of landfills where LFG was being emitted.
This was based on the CAG risk concentration of 2.6 ug/m3 <1 in 1
million risk)5 as well as some of the acceptable ambient concentrations
listed 1n the NATICH data base4. There however seemed to be some
discrepancy between the available sources. The advice of the Wisconsin
Division of Health was sought because of the discrepancy and because VC was
a known human carcinogen. The interim policy being used In Wisconsin
(before the promulgation of regulations dealing with hazardous air
emissions) required emissions to be limited to ambient impacts which had a
risk of one cancer death per one million persons exposed to a hazardous air
contaminant over a 70-year lifetime.
PHASE II - HEALTH ASSESSMENT
A request for a health assessment was made by the DNR to the Wisconsin
Division of Health and on December 1, 1986, the DOH published (In summary
form) its findings7. The full report was still In preparation at the
time this report was written. The report presented a list of ambient vinyl
chloride guideline levels derived from various sources, ranging from 0.6 to
235 parts per trillion (1.5 to 600 ng/m3), with the lowest value (derived
from U.S. EPA data) and one other <39.2 ppt, 100 ng/m3) representing
one-1n-a-ml11 Ion cancer risk concentrations. The DOH concluded:
"The conclusion of the DOH risk assessment Is that vinyl chloride
venting from landfills would represent a public health threat at the
one excess cancer 1n one ml 111on population level if ambient air
concentrations to nearby residents exceeded 0.0006 ppb as a 24 hour per
day average exposure. The DOH believes that the WI/EPA-CAG cancer risk
model is the most appropriate method for the cancer risk analysis. It
Is the most recent assessment and best reflects the prevailing cancer
state-of-the-art risk assessment process."
It 1s obvious that the one-1n»a~mfIHon risk ambient concentration was
well below that submitted by Region V U.S. EPA £2.6 ug (2600 ng)/m3)] or
any currently being used by other states - from a high of 238 ug/m3 (95
ppb), 8-hour average, for Nevada, to a low of 20 ug/m3 (8 ppt), annual
average, for Illinois'.
One author of the report. Dr. John Olson of DOH, Indicated that recent
literature Indicates onset of the cancers associated with vinyl chloride
are now observed at an earlier age and that "lifetime" exposure to VC may
not be a good predictor of cancer onset. This could mean that shorter term
exposure at higher concentrations could have adverse effects.
PHASE III - MODELING OF SOURCE EMISSIONS
While concentrations measured at closed landfill sites (to which the
general public has access) were above the recommended level, 1t was not
clear that persons living some distance from the LFG source Bright be
continuously exposed to harmful levels* Therefore, it was decided that
dispersion modeling should be performed for landfills which had active
519
-------
extraction systems with conventional sources (stacks) which could be
modeled. In that fashion estimates of ambient concentrations could be
predicted. To perform modeling requires measurements other than VC
concentrations, I.e. gas flow through the stack., temperature of the gas and
stack dimensions (1n addition to information on the surrounding Impacted
terrain). Only four sources were selected for modeling. The other
landfills with active extraction systems contained no VC 1n the LFG.
However, when the monitoring team attempted to measure flow at one of the
landfills selected, they were unable to do so using conventional flow
measurement methods. Therefore, modeling could not be performed for that
source. The modeling exercise therefore was restricted to only three
landfills.
Modeling Methodology
Conventional stack test techniques were used to obtain the necessary
stack parameters, I.e., flow (with a standard Pitot tube), temperature
(with a laboratory grade thermometer), and length (with a standard meter
rule). The measurements obtained for those parameters were In addition to
the average VC concentration obtained from a number of stack samples which
were analyzed by the methodology described earl 1er under Phase I -
Monitoring. Two different modeling exercises were performed utilizing
several different models: 1) U.S. EPA's CUmatologlcal Dispersion Model
(CDMQC) for Mineral Point annual VC concentrations; 2) the singe source
(CRSTR) dispersion model for Mineral Point 1-hour and 24-hour "worst case"
concentrations; 3) the Multiple Point Gaussian Dispersion Algorithm with
Optional Terrain Adjustment (MPTER) model for Delafleld and Verona;
4) preliminary modeling with the CDMQC and ISLCST models for Delafleld and
Verona (not reported).
Modeling Results and Discussion
The resuVts of the modeling were presented 1n three memorandum reports
(available from the authors upon request). It appeared obvious that homes
In the vicinity of Landfill #1 may be exposed to levels exceeding even the
CAG limit. For Landfill #2, which 1s Immediately adjacent to the Dane
County Hospital and Homes, It appeared as If jthe level of exposure 1s 13 to
33 times the DOH limit for residents within the perimeter of the area
modeled. Finally for Landfill #3, with a large numbers of homes within 500
meters, It appeared as If the level of exposure 1s 2 to 7 times the DOH
limit for residents within the modeled area.
Conclusions - Phase III
Based on the results of modeling for the three landfills, It was
concluded that residents In the vicinity of closed landfills or landfill
cells which contain municipal wastes (and which are relatively close to
populations) are being exposed to potentially harmful levels of vtnyl
chloride. While the number of landfills modeled was relatively small
compared to those tested (16 municipal landfills), the levels of VC
detected at almost all of those was believed to be of great concern due to
the comparatively low DOH limit. The difference Is several orders of
magnitude, I.e. 1.00 ppm (lowest detected level) versus the 0.0000006 ppm
limit or one to about 10 million. For emissions from probes or wells on
site, the dilution factor to the fencellne for most landfills would not be
expected to come close to that ratio I.e. would be a much smaller ratio.
Certainly many of these closed landfills are unfenced and people have
access directly to many of them. Therefore, there was general consensus
between Air Management, Solid Waste and Division of Health personnel that
520
-------
there was a need to control landfill gas emissions. Fortunately, almost
all of the largest landfills In Wisconsin already are capturing the
landfill gas and either flaring the gas or are combusting the gas for heat
or for energy.
PHASE IV - CONTROL OF LFG EMISSIONS
Soon after the Initial monitoring performed 1n Madison, the question of
control came to the fore. Is combustion an adequate method for destruction
of VC? DNR's Engineering and Surveillance Section determined that 1t was.
Carbon absorption as an alternative control technique was later determined
to be adequate for removal of vinyl chloride as an alternative. (Reports
available upon request.)
Concurrently, development of administrative rules requiring the control
of Hazardous Air Contaminants (HAC's) was proceeding. Those rules require
the use of best available control technology (without regard to cost) for
known human carcinogen emissions. Since for VC in LFG the control
technology exists, It was decided by mutual agreement between the
respective DNR sections that the specific concepts for LFG be Incorporated
Into revisions of the solid waste rules In the process of being adopted.
The changes in the solid waste rules Include Incorporation of the
definition of HAC's (toxic air pollutants), prohibiting landfills from
emitting HAC's, requiring new landfills to be designed to control HAC's,
and requiring that closure of landfills Include systems for collection and
control (combustion or absorption on activated carbon). The rules restrict
the requirements to landfills greater than 500,000 cubic yards (medium size
landfills and larger). They provide for exemption from collection and
control if the landfill owner can demonstrate (by sampling and analysis of
source emissions and modeling or other methods) that ambient impacts would
meet the HAC rule requirements.
REFERENCES - Due to space limitations could not be Included but are
available from the authors upon request.
Oue to ARCA requirements limiting this paper to six pages, tables,
figures and additional discussion could not be Included; the full text of
this paper is available directly from the authors by writing c/o Box 7921,
Madison, WI 53707-7921.
521
-------
EVALUATION OF THE FLUX CHAMBER METHOD FOR MEASURING AIR EMISSIONS OF
VOLATILE ORGANIC COMPOUNDS FROM SURFACE IMPOUNDMENTS
Alex R. Gholson, John R. Albrltton, and R.K.M. Jayanty
Center for Environmental Measurements
Research Triangle Institute
Research Triangle Park, North Carolina
Joseph E. Knoll and M. R. Mldgett
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
Enclosure methods have been used to measure air emissions of a
variety of compounds from soils, water, sediments, and living organisms.
A flux chamber method, which employs the enclosure method, recently has
been used to measure organic air emissions from hazardous waste treat-
ment, storage, and disposal facilities. Using a simulated surface
Impoundment facility (SIS), this flux chamber method was evaluated. The
liquid surface of the SIS was enclosed so that the total emission rate
from a liquid surface could be determined experimentally. Emission
measurements using the flux chamber method made at several points on the
surface were compared with the emission rate measured for the total
surface inside the enclosure. Both the accuracy and precision of the
flux chamber method were predicted from these measurements. The
Influence of sweep flowrate, emission rate, and different organic com-
pounds on precision and accuracy were Investigated.
The results of this study show that a consistent .negative bias
exists for all the flux chamber measurements. This bias became signifi-
cantly more negative at a low sweep flowrate (2 L/m1n). The bias also
was found to be compound dependent. Precision was less than 5 percent
under all conditions for the single component studies and between 6 and
13 percent for the three component study.
Introduction
The U.S. Environmental Protection Agency (USEPA) has been Instructed
to set air emission standards for hazardous waste treatment, storage,
and disposal facilities (TSDF) by the 1984 Resource Conservation and
Recovery Act amendments. In order to determine the potential health and
environmental effects of these air emissions, methods to measure or
predict them are required. The flux chamber method has been developed
and 1s being used to measure air emission from TSDF to provide a data
base for regulatory decisionmaking and to validate proposed models used
to predict air emissions *,',3,4.
In an attempt to define the quality of the data being produced by
the flux chamber method, this study was made to determine the accuracy
and precision of the flux chamber method for use on surface Impoundment
facilities. The Influence of the experimental parameter of sweep
522
-------
flowrate and the environmental parameters of emission rate and volatile
organic composition on accuracy and precision was Investigated.
Experimental Methods
The flux chamber design used for this evaluation was developed for
the USEPA by Radian Corporation.5 It consists of a stainless steel
cylinder with an enclosure area of 0.13 m2. The top Is enclosed with a
clear acrylic dome fitted with ports for sweep flow Inlet, sample out-
let, temperature probes, and a gas exit. The volume of the enclosure
with a 1 1n. Insertion depth 1s approximately 30 L.
The design for the surface Impoundment simulator (SIS) 1s shown 1n
Figure 1. The surface area of the liquid Is 1.86 m3 with an average
depth of 0.46 m. The surface Is enclosed 1n a shell covered with
Teflon with one end opened and the other end attached to the Inlet of a
blower. Sampling ports are provided for two flux chambers, flow moni-
toring, and sampling the air before the blower.
The volatile organic compounds (VOC) measured for this study were
1,1,1-trlchloroethane for single component studies and a mixture of
1,1,1-trlchloroethane, toluene, and 2-butanone for the three component
study. The organic components were added to the bottom of the tank.
The density of the compounds or mixtures was always greater than
1.0 g/mL to prevent forming an organic layer on the surface. Two Immer-
sion heaters were located Just above the organic layer to control the
tank temperature and Increase the convectlve mixing inside the tajik. A
pump was used periodically to Increase the aqueous concentrations of the
organic components.
All analyses were performed using a gas chromatograph (GC) with a
flame Ionization detector (FID). Flux chamber samples were collected 1n
a syringe and Injected Into a gas sampling valve with a 2-mL loop. SIS
air samples were collected in syringes or aluminum gas cylinders and
preconcentrated on Tenax that was thermally desorbed. SIS water sam-
ples were collected 1n glass vials with no neadspace and analyzed by
syringe Injection.
Results
Single Component Study
Accuracy and precision of the flux chamber method were calculated
from a series of colocated flux chamber measurements made Inside the SIS
containing a single VOC, 1,1,1-trlchloroethane. A series of nine ini-
tial emission measurements were made under similar emission conditions
and the same flux chamber conditions. The sweep flowrate was set at the
5 l/min recommended by Radian's study. Table I shows the variance found
between each duplicate measurement. The coefficient of variance (CV)
was less than 5 percent for all the measurements except for two. Using
a pooled standard deviation, the precision was predicted from these data
to be 3.0 percent. This value was chosen to be the control condition
precision.
523
-------
Colocated measurements, conducted at night, were made at an emission
rate approximately one tenth of the control condition measurements.
Table II compares the precision calculated for these conditions with the
earlier conditions. No change 1n the precision was found for the low
emission rate conditions. The variance found at night was half that
found for the control conditions. The improved precision at night could
be due to an effect of sunlight on the performance of the flux chamber
or the effect of the sunlight on the real emission rate. Emission rates
made In full sunlight were found to be highly variable both from flux
chamber measurements and SIS measurements. All measurements reported
here were made in the shade or on overcast days.
The precision and accuracy of the method also were determined at a
sweep flowrate of 2 L/min and 10 L/m1n. Table II show the results for
three colocated tests at each flowrate. A decrease 1n the precision
value (improved precision) was found at the higher flowrate, and a
slight Increase In the precision value (poorer precision) was found for
the lower flowrate. These results indicate that precision is Improved
by increasing the sweep flowrate possibly due to improved mixing at the
higher flowrate.
The average emission rate calculated from a colocated flux chamber
study was compared with the average emission rate calculated before and
after from the total SIS measurement to predict the accuracy of the
method. Table III lists the average percent bias found between the flux
chamber and SIS values for the control conditions and the other flux
chamber conditions studied. The bias was consistently negative (the
flux chamber values lower than the SIS values). The average bias values
were not significantly different within the 90 percent confidence Hm1t
(CL) for all the measurements except the low sweep flow study, which had
a significantly more' negative bias. This suggests that the accuracy of
the method decreases at the lower sweep flowrate.
Three Component Study
The precision and accuracy study was repeated with three VOC 1n the
tank. These Included 1,1,1-trlchloroethane, 2-butanone, and toluene.
Comparing the results of nine colocated duplicate flux chamber measure-
ments with the control of the single component study reveals several
apparent differences. Table IV shows the precision estimated for each
compound. The values are greater than for the single component control
and vary by almost a factor of two between themselves. The Increased
variance may be due to the large difference 1n emission rates being
pooled.
The accuracy of the three component results showed similar differ-
ences when compared with the single component study. Table V shows the
results of bias calculations between the flux chamber and the SIS emis-
sion results. Three significantly different average bias values were
found, and each one except toluene varied significantly from the single
component results. Of special interest 1s that 1,1,1-trlchloroethane In
the mixture had an average bias of less than one-half of 1,1,1-
524
-------
trlchloroethane 1n the single component mixture. The total emission
measurement bias showed no significant difference from the single compo-
nent results.
Conclusions
The results of the precision and accuracy study Indicate that pre-
cise emission measurements can be made using the flux chamber method.
The consistent negative bias found Indicates that the flux chamber
method may underestimate the emission rate from a surface Impoundment.
Either the flux chamber depresses the emission rate over the area 1t
covers or the total emission rate may not be equally distributed over
the surface with higher emission at the sides.
Of the experimental parameters Investigated, only daylight and sweep
flowrate was found to affect the accuracy or precision significantly.
Results suggest that sunlight may affect the variance between two colo-
cated flux chancers and lower sweep flowrate (2 L/m1n) Increases the
variance between measurement and Increases the bias. Both precision and
accuracy appear to be compound dependent and are dependent on the
matrix.
Studies are recommended to determine the effect of solar Intensity
on both the emission rate and the flux chamber method, the cause of the
compound dependency of the flux chamber accuracy, and the reason for the
consistent negative bias found 1n the results. Plans for determining
the flux chamber precision 1n the field currently are being made.
References
1) C. E. Schmidt, W, D. Balfour, "Direct gas measurement techniques and
the utilization of emissions data for hazardous waste sites", 1n
proceedings of the ASCE National Specialty Conference on
Environmental Engineering, 1983, p. 690.
2) R. R. Dupont, "Measurement of volatile hazardous organlcs emissions
from land treatment facilities," J.A.P.C.A., SL, 168-176, (1987)
3) B. M. Eklund, W. B. Balfour, C. E. Schmidt, "Measurement of fugitive
volatile organic compound emission rates with an emission Isolation
flux chamber," 1n proceedings of the AIChE Summer National Meeting,
Philadelphia, PA, August, 1984.
4) R. R. Dupont, "A flux chamber/sol1d sorbent monitoring system for
use In hazardous organic emission measurements from land treatment
facilities," 1n proceedings from the 79th Annual Meeting of the A1r
Pollution Control Association, Minneapolis, Ml, June, 1986.
5) M. R. Klenbusch, and D. Ranum., "Developing and evaluating test
methods for quantifying air emissions from hazardous waste disposal
facilities: Development and validation of the flux chamber method
for measuring air emissions from surface Impoundments," U'.S. Envi-
ronmental Protection Agency. Contract No. 68-02-3889, Work
Assignment 42, January 1986.
525
-------
TABLE I. FLUX CHAMBER PRECISION AT 5 L/MIN FOR
SINGLE COMPONENT SOURCE
Average
Sample Average surface Surface liquid emission rate,
No, temperature concentration ( q/m1n/m2) X CV
1.
23
130
14,300
2.1
2.
23
130
12,100
0.14
3.
22
130
13,200
1.5
4.
21
16.2
9,730
4.6
5.
21
16.2
9,750
6.0
6.
21
16.2
8,910
0.46
7.
23
93.6
7,470
4.8
8.
24
83.6
7,350
11
9.
24
93.6
7,630
4.8
CV ¦ Coefficient of variance.
TABLE II. RESULTS OF FLUX CHAMBER PRECISION STUDY
FOR SINGLE COMPONENT
Variable parameter Number of replicates Precision*
Control 9 3.0
Low emission rate 3 2.9
Nighttime 3 1.5
2 L/m1n sweep flow 3 4.1
10 L/mln sweep flow 3 1.4
aPrec1s1on
(Xi-X2)2
2n
526
-------
TABLE III. RESULTS OF FLUX CHAMBER ACCURACY STUDY FOR
SINGLE COMPONENT
Variable parameter
Number of replicates
Average bias,
* + 90* CI
Control
Low emission rate
2 L/mln sweep flow
10 L/m1n sweep flow
Nighttime
9 -45.1 + 6.4
3 -67.1 + 16.6
3 -81.5 + 9.6
3 -49.3 + 8.3
3 -57.2 + 21.3
CL ¦ Confidence limit.
TABLE IV. RESULTS OF PRECISION STUDY FOR THREE COMPONENT MIXTURE
Number of duplicate
Compound measurements
Jlange of Precision,
emission rates X
2-Butanone 9
1,1,1-Trlchloroethane 9
Toluene 9
11,000 -» 42,000 6.7
46,000 -> 100,000 10.6
5,100 -» 14,000 13.1
Total 9
65,500 -» 160,000 8.6
TABLE V. RESULTS OF ACCURACY STUDY FOR THREE COMPONENT MIXTURE
Number of
Average bias,
2-Butanone
1,1,1-Tr1chloroethane
Toluene
9
9
9
-68.1 + 3.1
-21.0 + 6.0
-38.2 + 4.4
Total
9
-40.7 + 3.3
CL ¦ Confidence Unit,
527
-------
Aluminum
Flux Chamber
Sampling Porta
sn
Sampling Porta
Inlat
Liquid Surfaca
Iknk Support
Figure 1. Side view of surface Impoundment simulator (SIS).
528
-------
DEVELOPMENT OF A SAMPLING METHOD
FOR MEASURING VOC EMISSIONS FROM
SURFACE IMPOUNDMENTS
Bart Eklund, Michael R, Kienbusch, and David Ranus
Radian Corporation
Austin, Texas
Terry Harrison
U. S. EPA, OAQPS/EMB
Research Triangle Park, North Carolina
This paper presents the results of an EPA-sponsored progran to develop
a sampling method to quantify air emissions of volatile organic compounds
(VOCs) from surface impoundments and other liquid surfaces. The chosen
sampling method used an enclosure device, or flux chamber, to isolate a
defined surface area and collect the gaseous emissions to allow a direct
emission rate measurement. The development program included four distinct
taskBt 1) identify the pertinent design and operation factors of the emis-
sion isolation flux chamber (EIFC) which may aignificantly affect the mea-
sured emission rate, 2) conduct a pilot-scale study to evaluate and optimise
the performance of the flux chamber, 3) develop a standard sampling proto-
col, and 4) demonstrate the method at a field site. Results from each task
are presented. The average percent recovery for the EIFC was 96,7% for a
seven-component standard and the reproducibility was +7,5%,
529
-------
Introduction
The U.S. EPA is interested in measuring the volatile organic emissions
from surface impoundments. Measurement of organic emissions from surface
impoundments has been studied from several different perspectives. Three
general perspectives are: 1) to indirectly measure ambient air concentra-
tions at downwind locations and use meteorological data in conjunction with
dispersion modeling to calculate the emission rate; 2) to measure liquid
phase concentrations and use mass transfer theories and emission rate models
to estimate the emission rate; and 3) to directly measure the emission rate
using an emission isolation flux chamber (E1FC) approach.
The direct approach was chosen for further investigation because it is
relatively inexpensive and offers increased precision estimates over the
modeling and indirect approaches. At the same time, if acceptable, the EIFC
procedure could be applicable in confirming mass transfer and emission rate
modeling theories.
Development of the method consisted of performing four tasks: 1) iden-
tify the pertinent design and operational factors of the EIFC; 2) conduct a
bench-scale pilot-study to evaluate each design and operational factor;
3) write an operations protocol for EIFC applications on surface impound-
ments; and 4) demonstrate the EIFC technique, in this case, at an industrial
wastewater facility.
The purpose of this paper is to present the results from each of the
four tasks performed to develop this technique.
Experimental Design
As previously discussed, four tasks were performed during the method
development study. Each task is briefly discussed below, after a general
introduction to the flux chamber sampling method,
EIFC Theory of Operation
The emission isolation flux chamber is a device used to make direct
emission rate measurements from land or liquid surfaces such as landfills,
spill sites, and surfaefe impoundments. The enclosure approach has been used
by researchers to measure emission fluxes of a variety of gaseous species,
including sulfur and volatile organic species. The approach uses a flux
chamber to sample gaseous emissions from a defined surface area. Clean, dry
sweep air is added to the chamber at a fixed controlled rate. The volumet-
ric flow rate of sweep air through the chamber is recorded and the concen-
tration of the species of interest is measured at the exit of the chamber
after equilibration (steady state). The emission rate is expressed as:
530
-------
Ei = C£R/A (1)
where: = emission rate of component i (ug/m^-min),
= concentration of component i in the air flowing from the
chamber (ug/m3),
R = flow rate of air through the chamber (m3/min). and
A = surface area enclosed by the chamber (m^).
All parameters in Equation 1 are measured directly.
Design and Operational Factors
The first task was to identify the design and operational factors of
the EIFC which may affect the measured emission rate.1 The chamber design
and operating procedures used by previous researchers were noted, along with
prior evaluative and theoretical studies. The theoretical basis for the
applicable emission processes was also discussed in terms of how it impacts
the EIFC design and operation. The design and operation factors chosen for
further evaluation were insertion depth, sweep air flow rate, impeller
rotation rate, internal, pressure, sampling duration, line length, and con-
struction materials. These were selected based upon judgment, prior field
sampling studies, and results from « validation study of EIFC techniques for
soil surfaces,2
Pilot-Scale Testing
Once the design and operational parameters were defined, a pilot-scale
evaluation of selected factors was performed.3 A large (1.5 m x 1.5 m x
0.6 m) galvanized tank was constructed to act as an artificial surface
impoundment. Emission tests were performed using aqueous solutions of
toluene. Toluene represents a typical, moderate-to-low solubility, highly
volatile compound found in many surface impoundments. Four sets of tests
were conducted. A series of calibration tests were performed to set the
operating procedures for the test facility, A set of parametric tests eval-
uated the effect of design and operating parameters on measured emission
rates. A third set of tests examined the reproducibility of the EIFC sam-
pling method. The fourth set of tests evaluated sample recovery using the
preferred design and operating conditions.
Protocol Development
After the pilot-scale validation studies, an operations protocol was
developed.4 The protocol documents the preferred design and operation of
the EIFC method. It provides step-by-step sampling and analytical proce-
dures and provides detailed information on fabrication, sampling strategy,
and data reduction. The sampling strategy is designed to determine an
average emission rate for a given ztone that is +20X of the true value at a
95% confidence interval. The current strategy divides the emission source
into homogenous zones, the zones are then each gridded, and a statistically
531
-------
determined number of grid points are randomly sampled. The protocol iB
currently under review by the U.S. EPA/ RTP,
Field Demonstration
The EIFC sampling protocol was demonstrated at an industrial wastewater
lagoon,^ This study was done in conjunction with another study on the
degree of stratification present at this site. Consequently, the sampling
locations were not randomly selected.
Emission rate measurements were made at five locations on the lagoon.
The EIFC was fitted with three flotation pods to provide buoyancy. The EIFC
was lowered in place using a cherry-picker and operated remotely from the
shore using an umbilical containing the inlet and exit gas lines. At each
point, grab samples of the chamber atmosphere were collected and immediately
analyzed on site for total non-methane hydrocarbons (TNMHC) by gas chroma-
tography (GC). Gas samples were also collected in evacuated, stainless
steel canisters for subsequent off-site GC analysis to determine the emis-
sion rates of individual compounds. Liquid grab samples were collected at
each point for correlation to compound-specific emission rates to calculate
field mass transfer rates.
Results
The results of the pilot-scale testing are summarized in Table I. The
interpretation of results was complicated by the inherent spatial variabil-
ity in emissions of the artificial surface impoundment. The design parame-
ters, material of construction and sample line length were each found to
affect the emission rate measurement by less than 10%. The operational
parameters generally did have a significant effect on the emission rate
measurement. The emission rate was found to increase with an increase in
sweep air flow rate. Use of an impeller increased emissions by about 20%.
Adequate mixing could be achieved through the design of the sweep air inlet
manifold. Therefore, the use of an impeller has been discontinued for the
sake of simplicity. Positive pressure build-up within the EIFC was found to
affect the emission rate, so a pressure relief valve (opening) was added to
the EIFC. Emission rates were found to decrease somewhat with time, prob-
ably due to surface depletion in the column of liquid isolated by the flux
chamber. Results for "reproducibility and standard recovery tests indicated
the accuracy and precision of the method are good. A figure of the EIFC as
currently conceived is shown in Figure 1,
The results of the field study are summarized in Table II. The major
compounds emitted were benzene and toluene. Approximately a dozen compounds
per sampling point could be detected at 1 ppbv or greater in the flux
chamber atmosphere. For TNMHC, the average sampling variability was +25%
and the average analytical variability was +1-22 for the gas-phase analysis.
This ratio of sampling to analysis variability compares favorably vith other
air sampling methods.
532
-------
Conclusions
Each of the four elements of the study was successfully completed. The
pilot-scale examination of the design and operational factors identified
several factors that oust be controlled to ensure a reproducible sampling
technique. These included the sweep air flow rate, induced mixing via sweep
air inlet, the sampling time, and internal pressure. The design of the flux
chamber is also of importance because the sweep air flow rate and induced
mixing are both dependent upon the design. A protocol was developed that
discussed in detail the design and operations of the EIFC for liquid sur-
faces. The protocol is currently under Agency review. A field demonstra-
tion of this method was performed using the protocol as a guide. The field
project showed that the method is feasible for making emission rate deter-
minations.
References
1. B. Eklund, "Evaluation of Flux Chamber Design and Operation Factors,"
Radian Corporation, EPA Contract No. 68-02-3889. Work Assignment 42
(August 1985).
2. "Validation of Flux Chamber Emission Measurements on a Soil Surface."
Radian Corporation, EPA Contract No. 68-02-3889, Work Assignment 18
(June 1986).
3. M. R. Kienbusch, D. Ranum, "Development and Validation of the Flux
Chamber Method for Measuring Air Emissions From Surface Impoundments,"
Radian Corporation, EPA Contract No. 68-02-3889, Work Assignment 42
(January 1986).
4. "Measurement of Gaseous Emission Rates from Liquid Surfaces Using an
Emission Isolation Flux Chamber, Proposed Method.» Radian Corporation,
EPA Contract No. 68-02-3889, Work Assignment 42 (November 1985).
5. D. Ranum, "Supplemental Sampling Data at a Proprietary Facility."
Radian Corporation, EPA Contract No. 68-02-3889, Work Assignment 42
(February 1986).
533
-------
TABLE I. SUMMARY OF RESULTS FOR PILOT-SCALE TESTS
Test
Parametric Tests
tlepth of Penetration
Sweep Air Flow Rate
Impeller Rotation Rate
Internal Pressure
Sampling Duration
Sample Line Length
Construction
Material
Precision Tests
Reproducibility
Recovery Tests"
1.52 m Sample Line
30.5 d Sample Line
Test Range/
Result
1.27-7.62 cm
1.4-21.2 L/min
0-135 rpm
0-0.762 cm H20
0-6.6 hrs
1.52-30.5 m
Steel/Teflon,
Plexiglas/Teflon
+7 .5*
96.7%
104%
Significance
of Test on
Emission Rate
Measurement
None
Yes
Yes
Yes
Yes
None
None
None
None
None
Recommended
Design/Operating
Value
1.27-2.54 cm
5-10 L/min
No impeller
AP <0.127 cm H20
<30 min
<30.5 m
aMulti-component standard,2
•MMWUHM4 •t2CJj"l'"5e
Figure 1. A cutaway design of the emission isolation flux chamber and
support equipment.
TABLE II. SUMMARY OF RESULTS FOR FIELD STUDY
Compound
Methyl Chloride
Benzene
Toluene
Total NMHC
en Grid 2
0.338
9.42
5.14
27.6
0.00480
6.69
2.87
11.6
Emission Rate (g/m^-day)*
OT1 grid A SlSaie—"HeitT
0.0271
3.86
0.851
5.35
0.0290
6.92
1.12
9.25
0.0369
0.758
0.269
1.36
0.0872
5.53
2.05
11.0
534
-------
LABORATORY AND FIELD EVALUATION OF THE SEMI-VOST
METHOD FOR MEASURING EMISSIONS FROM HAZARDOUS
WASTE INCINERATORS
J. T. Bursey, J. L. Steger, M. A. Palazzolo,
D. J. Benson, R. L. Porch, J. Rice,
J. B. Homolya, R. A. McAllister,
J. F. McGaughey, D. E. Wagoner
Radian Corporation
Research Triangle Park, NC 27709
J. H. Margeson, J. E. Knoll, M, R. Midgett U. S. Environmental Protection
Agency Research Triangle Park, NC 27711
Laboratory studies and two field evaluations have been completed to
assess the formulation and performance of the Semi-Volatile Organic Sampling
Train (Semi-VOST) method for measuring concentrations of principal organic
hazardous constituents (POHCs) with boiling points greater than 100 C emitted
from hazardous waste incineration. Laboratory experiments were performed
initially to design and test a dynamic spiking system to allow spiking of test
compounds in the field for a reliable procedure to assess method precision and
bias. Additional laboratory tests were performed to determine analytical
method detection limits to estimate the levels at which deuterium labeled
compounds should be spiked under field conditions. Compounds representative
of a variety of chemical and physical properties were selected as test
compounds to be spiked in a field test executed at a hazardous waste incin-
erator. After collection of the samples in the field, train components were
extracted with methylene chloride, the extracts concentrated, and compounds of
interest were quantified using high resolution gas chromatography coupled with
low resolution mass spectrometry, The first field test showed that method
precision and bias are compound specific, and the distribution and recovery of
POHCs in the various components of the sampling train are related to the
compound chemical functionality, water solubility, and boiling point.
Subsequent to the first field test, an extensive series of laboratory
experiments was performed to modify the dynamic spiking apparatus and proce-
dures for use in the field. Also, since previous experiments had shown
recoveries of certain compounds to be poor or variable, laboratory experiments
were performed to determine the extent and nature of compound interactions
with components of the sampling train. Water soluble compounds may form salts
by reaction with the wet and acidic environment normally encountered in stack
emissions. These water soluble compounds are also collected in the aqueous
portions of the sampling train and may be poorly recovered by solvent extrac-
tion. The analytical procedure was revised by pH adjustment to improve
recovery of basic and water soluble compounds. A screening procedure was also
developed to determine the optimal extraction solvent and to estimate the
535
-------
recovery of test compounds obtained from acidic water condensates ujjlng the
sample preparation methodology presented in the Semi-VOST protocol.
tert-Butyl ether was selected as the optimum solvent for the extraction of
water soluble compounds. For the second field test, five deuterium labeled
gaseous compounds were dynamically spiked into four simultaneously-operating
Semi-VOST trains at a hazardous waste incinerator equipped with a wet
scrubber. Results from the laboratory experiments and the second field test
provided precision and bias estimates for the test compounds, and compound
retention volumes on the solid sorbent were evaluated through a series of
distributive volume experiments. Data were evaluated from gaseous dynamic
spiking and liquid spiking of the test compounds on XAD-2® resin, and^the
effects of sample storage time on sample stability were investigated.
Conclusions reached from the second field test are ; 1) the data from the
dynamic spiking of a variety of FOHCs exhibiting a range o£ boiling points,
water solubilities, and chemical functionalities support the findings of the
previous field test and laboratory studies, 2) the experimental data give the
basis for the use of laboratory-determined retention volumes of POHCs to
predict field breakthrough volumes through XAD-2* resin, and 3) the
experimental results support the selection of the Semi-VOST method as the
method of choice for sampling flue gas emissions of organic compounds with
boiling points greater than 100*C.
Introduction
The Solid Waste Disposal Act, amended by the Resource Conservation and
Recovery Act of 1975 (RCRA), requires that EPA establish a national regulatory
program to ensure that hazardous wastes are managed in a manner which does not
endanger human health or the environment. The statute requires EPA to promul-
gate performance standards for hazardous waste management. Included in the
regulations are provisions for waste disposal by incineration and requirements
that hazardous waste incinerators be so operated that principal organic
hazardous constituents (POHCs) are destroyed and/or removed with minimum
efficiency of 99.99%. To determine destruction and removal efficiency (DRE),
EPA has designed the Semi-VOST Method as the method to measure flue gas
concentrations of POHCs with boiling points greater than 100°C. The Quality
Assurance Division of the EPA Environmental Monitoring Systems Laboratory
(EMSL) has responsibility for evaluating and standardizing EPA source test
methods. In the application of Semi-VOST methodology, gaseous and particu-
late components are lsoklnetically withdrawn from an emission source and
collected in a multicomponent sampling train. Under contract to EMSL, Radian
Corporation is providing technical assistance in evaluating the Semi-VOST
Method, The objective of the Semi-VOST program is to provide data necessary
to determine bias, precision, applicability, and limitations of the method.
Thei technical approach used to collect these data is a multi-task effort
involving literature, laboratory, and field studies. An initial field test,
utilizing a quad-train approach, provided preliminary data on method bias and
precision. As a result of recommendations which evolved from the first field
study, additional laboratory studies and a second field test at a hazardous
waste incinerator were performed.
On the basis of experience gained from the first field test and laboratory
experiments both before and after the field test, a second field test again
utilized a quad-train to evaluate method performance by sampling flue gas from
536
-------
a full scale Incinerator while spiking test compounds into the sampling
trains. The objectives of the laboratory and field program were:
1) evaluate the effect of water solubility and compound functionality
on recovery of selected test compounds;
2) develop a screening procedure for selection of the optimum
extraction solvent and estimation of recovery of water soluble
compounds from acidic water condensate;
3) design, construct, and evaluate the performance of a dynamic spiking
system for introduction of deuterated test compounds;
4) execute a second field test to provide additional data on method
precision and bias for additional test compounds; and
5) determine the utility of using laboratory determined retention
volumes to predict field breakthrough.
Results and Discussion
In the field studies, the sampling system used four simultaneously operating
Semi-VOST trains. To obtain identical samples of the stack gases at a point
of average velocity for each train, a common fixed heated probe system (quad-
probe) was us£d. In the first field test, three test compounds (toluene,
chlorobenzene, and tetrachloroethane) were used for dynamic spiking experi-
ments to determine method bias. Bias values obtained for toluene and
tetrachloroethane were +1 and -16%, respectively. Bias could not be deter-
mined for chlorobenzene since the dynamic spike concentration was low relative
to high levels measured in the stack gas. Method precision for chlorobenzene
could be calculated from unspiked sample train data in three components:
sampling, sample preparation, and analysis, with results of 17,6, 2.3, and
8.9ft, respectively. Overall method precision was 19.9% for chlorobenzene
uncorrected for deuterated spike recovery. Distributive volume experiments
performed to determine sample train breakthrough used chlorobenzene concen-
trations of the stack gases; no breakthrough was observed at a flow rate of
0.5 cfm for a sample collection period of 4 hours.
Laboratory experiments performed using aniline, pyridine, phenol, and
resorcinol showed that basic organic compounds (aniline and pyridine) may
react with the acidic environment to form salts insoluble in the extraction
solvent and soluble in water present in the sampling system. Recovery of
resorcinol, an extremely water soluble compound, requires a more polar
extraction solvent. Based on data obtained from the laboratory experiments,
if no modification is made to the Semi-VOST methodology, compound losses due
to salt formation/water solubility can range from 34 (pyridine) to 73%
(resorcinol). In addition, laboratory experiments showed that it is impera-
tive to extract adsorbent resin samples immediately after removal from the
resin cartridge, as volatilization of compounds from the resin may occur if
extraction is not initiated immediately, Since aqueous components of the
sampling train may contain organic salts and water soluble compounds of
interest, analysis of any water present by high performance liquid chroma-
tography as well as an additional resin extraction using methanol is
537
-------
recommended to retrieve any of these salts or water soluble compounds which
may have been left on the resin after extraction of the resin with methylene
chloride. To assess interferences due to solvent interaction, stabilities of
aniline, pyridine, phenol, and resorcinol in methylene chloride were
evaluated. Concentrations of these four test compounds remained within ten
percent of the original value for up to 118 days; thus, no major
solvent-compound interactions were observed for the usual Senti-VOST sample
storage time.
Breakthrough or specific retention volume is the sample volume required to
elute a compound (adsorbate) introduced onto a resin (adsorbent) bed. Before
breakthrough occurs, the amount of spiked material adsorbed on the bed is
directly proportional to the volume sampled. Therefore, the adsorbate-sample
volume relationship must pass through the origin. After breakthrough occurs,
the amount of adsorbate is no longer proportional to sample volume, but
approaches a horizontal asymptotic value equal to equilibrium loading. When
solid adsorbents are used for trapping and concentrating organic materials, it
is desirable to have some means to estimate the efficiency of the procedure,
to determine the maximum volume of gas that may be sampled before significant
breakthrough occurs. The retention volume determination consists of packing a
chromatographic column with the adsorbent and measuring the retention times of
the analytes at various temperatures (see data shown in Table I), This
procedure was used in the laboratory for comparison with results obtained in
actual field testing. In the second field study, three quad-probe tests were
conducted to compare results of laboratory breakthrough volume studies to data
obtained In the field. During these tests, three trains were spiked with
deuterium labeled compounds using the dynamic spiking system and one train was
operated unsplked. The three spiked trains provided a triplicate set of
samples for the breakthrough analysis and the unsplked train provided a
background sample. Figure 1 shows the results of the distributive volume
study for d„-toluene. Note that once breakthrough has occurred, the adsorbate
versus sample relationship is no longer linear because it leveled off to an
equilibrium adsorbed concentration.
Five tests in the second field study were performed with the quad-probe to
assess method precision and bias. Five deuterated compounds were spiked into
the sampling trains at 100 times the method detection limit level during the
tests to provide data on precision and bias. For the five test compounds, the
following results were obtained: d^-pyridine, precision 17.7 %CV, bias -29.0%;
dg-toluene, precision 14.6 %CV, bias -14.9%; d-jn-o-xylene, precision 8.1 %CV,
bias -0.9%; d.-tetrachloroethane, precision 32.9 %CV, bias -18.5%; d,-chloro-
benzene, precision 13.7 %CV, bias -13.8% (results are summarized in Table II).
Two quad-probe samples were collected to provide information on sample
stability. Dynamically spiked samples were stored for five weeks and nine
weeks and then analyzed. Results for the stored samples were compared with
results for liquid spiked samples to assess compound loss due to sample
storage. The overall conclusion after analysis of these samples is that
holding time (up to 63 days) between sampling and analysis does not signifi-
cantly affect the results obtained. In the field study, the between set
variance in compound recoveries was greatest for duplicate liquid spikes. The
variances obtained when comparing two seta of duplicate dynamic spikes and
dynamic spiking to liquid spiking were smaller. In the field study, all
samples received were spiked in the laboratory with surrogate standards prior
538
-------
to any sample preparation, the surrogate compounds were deuterium labeled
compounds used to monitor the efficiency of the sample preparation process for
recovering compounds from the matrix. Corrections can validly be made to
allow for less than 100% recovery of surrogates and corrected values for
compounds of Interest will reflect recoveries of the surrogates.
Conclusions
The second field test demonstrated that dynamic spiking of the Semi-VOST train
using deuterated compounds is a viable approach for determination of method
bias under field conditions. In the absence of measurable Btack emissions of
POHCs, dynamic spiking can be used to measure method precision. Experimental
data from the field test support the use of laboratory determined retention
volumes to predict field breakthrough volumes. A difference in holding time
before extraction for dynamically spiked reBin samples does not result in a
significant difference in recoveries for all of the compounds spiked. The use
of surrogate standards during sample preparation and analysis is necessary to
obtain the best representation of test compound recoveries. Deuterated
analogs, vhen available, are the best possible surrogates because they behave
exactly the same as the compounds of interest and will be truly representative
of test compound recoveries. Use of deuterium labeled surrogate compounds is
highly recommended. A comparison was made between observed values and
expected levels of five compounds spiked onto resin as quality control samples
provided by the Research Triangle Institute. Bias ranged from -,1.8% to -B.9%
with reference to the RTI values. Average * bias was -4.7%. The agreement is
excellent, shoving that the recovery of organic compounds from XAD-2* resin
using the Semi-VOST methodology is favorable.
This work was supported by the Environmental Monitoring Systems Laboratory,
Office of Research and Development, U. s. Environmental Protection Agency,
under EPA contract no. 68-02-4119.
Disclaimer
Although the research described in this article has been funded wholly or in
part by the United States Environmental Protection Agency by contract to
Radian Corporation, it has not been subjected to Agency review and no official
endorsement should be inferred. '
References
1. J. Bursey, M. Hartman, J, Hamolye, R. McAllister, J. McGaughey, and
D. Wagoner, Contract Bo, 66-02-4119; "Laboratory and Field Evaluation of
the Semi-VOST Method," Volume I, PB 86 123551/AS and Volume II, PB 86
123569/AS, September 5, 1985.
2. U. S. Environmental Protection Agency. Test Methods for Evaluating Solid
tfaste; Physical/Chemical Methods. EPA Report No. SW-846, U. S. Environ-
mental Protection Agency, Washington, D.C,,1982.
3. J. Rice, D. Vagoner, and J. Homolye, Contract No. 68-02-4119, "Laboratory
Evaluation of the Sample Preparation of Semi-VOST Methodology for
Measuring Semi-Volatile Organic Emissions from Hazardous Uaste
Incinerators," interim Report, September, 1986.
539
-------
TABLE I. RETENTION VOLUME
Compound
Stapling Breakthrough T1m for
Flow ' "
¦1/g at 20°C cfwr
Retention
VoIum Flow Rata Breakthrough
fml r (ft3) h
toluene
1,1,2.2,-tatra-
cMorpethana
chloroboniene
pyridine
o-xylana
137 #277
I,430,$72
450*473
68.696
1.021.676
0*50
0.50
0.50
0.50
0.50
2*72 (96)
3.2
28.6 <1011) 33.7
9.00 (318) 10,6
1.36 (46) 1.6
20,4 (720) 24.0
*0.50 cubic foot/minute (cfM 1a equivalent to 0.01416 m3/m1n
TABLE II. SUMMARY OF PRECISION AND BIAS RESULTS
Precision
Compound
Number
of Valid
Sample*
Nun %
Recovery
Percent
Blaal
Standard
Deviation
Percent
Coefficient
of Variation
dg-pyr1d1ne
16
71.0
-29.0
12.6
17.7
dg-toluene
13
85*1
-14.9
12.4
14.6
dc-chloro-
bonzano
13
86.2
-13.8
11.6
13.7
d10-o-xylM»
13
99.1
-0.9
a.o
8.1
d9»tetrach1oro»
ethane
13
81.5
*18.5
26.8
32*9
Percent bias • Moan I recovery - 100
1600 -
LABORATORY RETENTION VOLUME - 2.7 iO.BxIO*
3 4 6 B
IAMPLI VOLUMI 1*10*1, Mm
Figure I. Retention Volume Studies for dg-Toluene
540
-------
EVALUATION OF MICROWAVE TECHNIQUES TO
PREPARE COMBUSTION SOURCE SAMPLES
David A. Blnstock, Peter M. Grohse, Percy L. Swift, and Alvla Gasklll, Jr.
Research Triangle Institute
Research Triangle Park, NC 27709
Thomas R. Copeland
ERCO/A Division of ENSECO, Inc.
Cambridge, MA 02138
Paul H. Friedman, Office of Solid Waste
U. S. Environmental Protection Agency
Washington, D.C. 20460
The use of microwave energy to facilitate sample decomposition prior
to elemental analysis Is now receiving considerable attention* Both wet
and dry digestions are achievable. When microwave energy Is used In com-
bination with add mixtures 1n closed vessels, the combined pressure and
rapid heating can reduce digestion times to a few minutes from hours or
days that may be required for open beaker digestions. This savings 1n time
and labor Is significant and has prompted the Office of Solid Waste to
evaluate this technology as a preparation tool for solid wastes. Of par-
ticular Interest are used oils and other fuels slated for Incineration,
Incinerator ash and particulates from these processes.
This study reports on the evaluation of a commercially available mi-
crowave oven sample preparation system for this application. The effect of
sample preparation conditions, Including the add matrix, heating time, and
pressure were evaluated for fifteen toxic or hazardous elements 1n parti-
culates and ashes. Analyses were carried out by Inductively coupled plasma
emission spectroscopy.
541
-------
Introduction
The techniques that are typically used to prepare Resource Conserva-
tion and Recovery Act (RCRA) wastes for analysis for metals and other ele-
ments are relatively time consuming, requiring several hours to several
days to complete, They also often involve the use of acid digestions and
thermal decomposition steps which may result 1n analyte losses, Incomplete
recoveries, or sample contamination. These limitations are well known to
the analytical community and to the end users of these data 1n EPA, states,
and Industry. The resulting Inefficiency of these techniques reduces lab-
oratory sample throughout, drives up the cost of analytical testing and
Impedes decision-making. Given these concerns, the USEPA Office of Solid
Waste 1s Interested 1n developing cost effective sample preparation tech-
niques for metals and other elements in environmental and process waste
samples. Once developed, these techniques can then be written as methods
for Inclusion 1n "Test Methods for Evaluating Solid Waste, SW-846" and made
available to the user community.
One particularly attractive sample preparation technique that 1s now
receiving considerable attention 1s microwave assisted sample dissolution.
The use of microwave energy to facilitate sample decomposition prior to
elemental analysis has received considerable attention 1n recent years.
Both wet and dry digestions are achievable. When microwave energy 1s used
1n combination with acid mixtures 1n closed vessels, the combined pressure
and rapid heating can reduce digestion times to a few minutes, from hours
or days that may be required for open beaker digestions. This savings 1n
time and labor 1s significant and has prompted the Office of Solid Waste to
evaluate this technology as a preparation tool for solid wastes. Of par-
ticular interest are used oils and other fuels slated for Incineration,
Incinerator ash, and particulates from these processes.
Previous evaluative work in this area was carried out by Nadkarnl1,
Using a commercial microwave oven and an HF/aqua-regla digest, National
Bureau of Standards (NBS) Coal 1632a and NBS SRM 1633a Fly Ash were
solublUzed. Copeland' (1985) used a HNO3/H2O2 microwave procedure to
prepare waste oils for determination of As, Be, Cd, N1, and Pb.
This study reports on the evaluation of a commercially available mi-
crowave oven sample preparation system. The effect of sample preparation
conditions, Including the add matrix, heating time, and pressure were
evaluated for toxic or hazardous elements 1n particulates and ashes. Anal-
yses were carried out by Inductively coupled plasma emission spectroscopy.
Experimental Methods
Microwave Oven
The MDS-81D Microwave system (CEM Corporation, Indian Trail, NC) was
used for this study. The oven resembles a standard microwave oven, but Is
equipped with additional features to facilitate sample preparation. For
example, the Teflon-coated microwave cavity has a variable speed corrosion
resistant exhaust system. The main element of the system couples a precise
microwave variable power system with a programmable micro-processor digital
computer. Other elements include a rotating turntable, Teflon vessels with
caps and a patented pressure relief valve, a capping system, and a cooling
tank.
542
-------
The Teflon sample vessels and caps are designed to withstand pressures
up to 100 ps1 and temperatures up to 200*C.
Inductively Coupled Plasma Emission Spectrometry (ICPES)
All analytical measurements were performed using an Instrumentation
Laboratory 200 ICAP.
Reagents
All inorganic acids used were of "Ultrex" quality, from J. T. Baker
Chemical Co. Other chemicals were of analytical reagent grade quality,
Delonized water of 18 Mfl/cm specific resistivity was used.
Combustion Source Materials
The evaluation of microwave methods was carried out using NBS Coal
1632a and Fly Ash 1633a. Elemental values for these standards were ob-
tained from their certificates of analysis.
Microwave Preparation Procedures
Total Digestion Procedure; A 300 rag sample 1s placed In a 60 mL
Teflon digestion vessel equipped with a relief valve and treated with 1 mL
concentrated HNO3, 3 mL HF, and 0.5 mL HCIO4. The vessel 1s sealed and
heated In the microwave oven at 15 percent power for 5 minutes, followed by
30 percent power for 15 minutes. The vessel 1s cooled, the cap removed, 3
mL HF added, resealed and heated at 20 percent power for one hour. The cap
1s then removed and the contents evaporated until fuming ceases. Six mL of
20 percent HNO3 are added and evaporated to dryness; 15 mL 20 percent HC1
added, the vessel sealed and heated at 10X power until dissolution of the
residue Is achieved. The vessel is uncapped, evaporated to dryness, and 15
mL 5 percent HC1 1s added and heated until the solution clears. The ves-
sel 1s cooled and diluted with delonized water to 50 mL. Total digestion
time 1s two hours.
HF/Aaua-Reala Procedure: A 300 mg sample 1s placed 1n a 60 mL Teflon
digestion vessel equipped with a relief valve and treated with 6 mL aqua-
regla (3HC1:1HN03) and 2 mL HF. The vessel Is sealed and heated 1n the
microwave oven at 100 percent power for 3 minutes. The vessel1s cooled,
the cap removed, the dlgestate filtered and transferred to a 50 mL volu-
metric flask with delonized water.
Results
Total Digestion Procedure
A total digestion (HCIO4/HF/HCI) microwave procedure was used to
prepare NBS SRM 1633a Fly Ash. With the exception of Mn and Pb, recoveries
were all within 25 percent of the NBS values (Table I), in addition,
results for most elements are comparable to those from a more time-
consuming conventional open-beaker digestion of the same material.
543
-------
A modified total digestion procedure was used to prepare NBS SRM 1632a
coal. Due to the potential explosion hazard of the HClO^coal mixture, the
digests were not taken to dryness. Good agreement was obtained for all
elements with the exception of Mn, Pb, and Co (Table II).
HF/Aqua-Reg1a Procedure
NBS SRM 1632a coal was digested using three modifications of the
HF/Aqua-Reg1a microwave procedure (Table III). The modified conditions
were 3 minutes at 100% power (column A), 6 minutes at 75% power (column B),
and a dry ash of 4 hours, followed by 3 minutes at 100% power (column C).
Microwave digestion for 6 minutes at 75% power yielded higher recover-
ies for Co, Cr, Mn, Pb, and V than the other variations, whereas digestion
for 3 minutes at 100% power gave highest recoveries for N1 and Zn. Dry
ashing followed by microwave HF/aqua-reg1a gave the lowest recoveries, with
the exception of Fe, which was higher.
Conclusions
Our preliminary work Indicates that substantial time/cost savings can
be achieved using microwave digestion—particularly with closed vessel
procedures. Table IV Illustrates the considerable time savings for the
three procedures used In this study compared to conventional techniques.
There Is also a potential reduced need for reagents such as HCIO4 and HF.
Future work will Involve refining the acid digestion/microwave condi-
tions for combustion source samples and other RCRA wastes.
REFERENCES
1. R. A. Nadkarnl, "Application of Microwave Oven Sample Dissolution In
Analysis," Anal. Chem. 56: 2233 (1984).
2. T. Copeland, Methods for Determining As. Be. Cd. Cr. N1 and Pb 1n Waste
011s. EPA/0SW Final Report, Contract No. 68-01-7075, WA No. 3 (1986).
3. University of Missouri (1985).
4. E. S. Gladney, Compilation of Elemental Concentration Data for NBS
Biological and Environmental Standard Reference Materials. Los Alamos
Scientific Laboratory (1981).
544
-------
Table I. Analysis of NBS SRM 1633a Fly Ash
(Total Digestion Procedure)
(«g/g)
Element
Mean + S.D. (n ¦ 3)
NBS Values
X Bias
Conventional3
Digestion
Al
160,000 + 28,000
(140,000)
+14
148,000
Be
11.9 + 0,6
(12)
-1
15
Ca
9360 + 1060
11,000 + 100
-15
11,200
Co
49.5 + 0.3
(46T
+8
Cr
169 + 59
196 + 6
-14
160
Cu
90.3 + 3
118 + 3
-24
110
Mn
133 + 2
(190)
-30
175
N1
130 + 25
127 + 4
+2
130
Pb
45.1 + 20
72.4 +"0.4
-38
82
V
313 + 3
(300)
+4
310
Zn
205 + 8
220 + 10
-7
210
n * number of replicates
( ) - uncertified value
Table II. Analysis of NBS SRM 1632a Coal
(Total Digestion Procedure)
(ug/g)
Element
Mean + S.D. (n» 2)
RTI
NBS Values
X Bias
Al
30,200 + 5900
(31,000)
-3
Ca
2050 + 340
2300*
-ii
Co
9.72 + 0.19
(6.8)
+43
Cr
31.2 + 2.8
34.4 + 1.5
-3
Cu
15.6 + 0.6
16.5 + 1
-5
Fe
10,400""+ 600
11,100 + 200
-6
Mg
1030 +"180
„
Mn
19.6 + 0.6
28 + 2
-30
N1
17.3 + 1.8
19.4 + 1
-11
Pb
ND
12.4 + 0.6
Sc
5.45 + 0.56
(6.3)
-14
V
44.2 + 0.7
44 + 3
+0.4
Zn
34.9 + 0.3
28 + 2
+25
n « number of replicates
( ) - uncertified value
545
-------
Table III. Analysis of NBS SRM 1632a Coal
(Modified HF/Aqua-Reg1a Procedures)
Mean + S.D. (n » 3)
(ug/g)
Element
Aa
Bb
CC
NBS Values
Al
24,600
22,700 + 1600
19,600 + 3500
31,000
Cd
<0.5
<0.5
<0.5
0.17 + 0.02
Co
NO
7.37 + 0.73
5.77 + 0.74
(678)
Cr
20.1
24.6 + 2.7
23.8 + 5.3
34.4 + 1.5
Cu
10.3
11.4 + 1.0
10.5 + 4.0
16.5~+ 1
Fe
7270
8940 + 267
9180
11,100 + 200
Mn
19.6
28.2 + 1.9
20.7 + 1.9
28 + 2
N1
18.3
12.9 + 2.2
16.6 + 0.4
19.4 + 1
Pb
7.2
15.4 + 3.2
9.68 + 3.00
12.4 + 0.6
V
33.0
47.6 + 3.1
38.9 + 2.2
44 + 3
Zn
23.7
15.1 + 2.3
13.7 + 5.5
28 + 2
aHF/Aqua-Reg1a, microwave 3 minutes 0 100% power, n ¦ 1
^HF/Aqua-Regla, microwave 6 minutes @ 75% power
CAshed at 400*C 4 hours-HF/Aqua-Regla, microwave 3 minutes @ 100% power
n = number of replicates
( ) - uncertified value
Table IV. Comparison of Digestion Time of
Microwave and Conventional Techniques
Add Digestion
Microwave
Conventional
Nitric acid
40 minutes
4 to 6 hours
Hydrofluor1c/aqua-reg1a
3 to 10 minutes
6 to 8 hours
Total digestion
2 hours
16 hours
546
-------
»
DEVELOPMENT AND EVALUATION OF ANALYTICAL TECHNIQUES FOR TOTAL
CHLORINE IN BURNER FUELS
Alvla Gaski11, Jr.
Eva D. Estes, David L. Hardison
Research Triangle Institute,
Research Triangle Park, North Carolina and
Paul H, Friedman, Office Of Solid Waste,
U.S. Environmental Protection Agency,
Washington, DC
A current EPA regulation prohibits the sale for burning in non-
industrial boilers of used oils and oil fuels contaminated above specified
levels with certain metals and total chlorine. When burned as fuel in a
small boiler, the contaminants may be emitted to the ambient air at
hazardous levels. This regulation establishes a rebuttable presumption
that used oil containing more than 1,000 ppm total chlorine has been mixed
with halogenated solvents and is a hazardous waste. Rebutting the
presumption requires the seller of the oil to prove that this chlorine is
not due to halogenated solvents or other hazardous halogenated organics.
If the rebuttal Is successful, the oil can be sold as fuel up to a level
of 4000 ppm total chlorine.
Analytical techniques for determination of total chlorine were
evaluated or developed to provide regulatory agencies and the regulated
community with appropriate chlorine test methods. The techniques
evaluated included chemical titrations following oxygen bomb combustion,
disposable field test kits, Instrumental mlcrocoulometry, and x-ray
fluorescence spectrometry.
These candidate techniques were subjected to interlaboratory testing
to estimate their precision, accuracy, sensitivity, and susceptibility to
matrix effects. Information on ease of use and analysis costs was also
collected. Based on this pilot study, test methods will be written for
the most promising techniques and subjected to a formal collaborative
study to generate precision and accuracy data for each method. These
methods are to be proposed in the Federal Register as mandatory for
compliance with the existing used oil regulation.
547
-------
DEVELOPMENT AND EVALUATION OF ANALYTICAL TECHNIQUES FOR CHLORINE
IN BURNER FUELS
1. Introduction
More than 1 billion gallons of used lubricating oil are produced In
the United States annually. A significant fraction Is sold for burning In
small nonindustrlal residential, commercial, or Institutional boilers,
typically after blending with virgin nos. 4 or 6 fuel oils. These used
oils frequently arrive at fuel reprocessing or blending facilities
contaminated with chlorinated solvents, lead, cadmium, and arsenic.1
These contaminants may be present as a result of the oil's use or may have
been added through mixing with hazardous waste. Reprocessing, with the
exception of rerefining, typically fails to remove these contaminants.
When burned as a fuel In a small boiler, the contaminants or their
combustion byproducts may be emitted to the ambient air in amounts high
enough to present potential hazards to exposed individuals. This concern
has lead EPA to promulgate a regulation prohibiting the sale for burning
of used oil contantlnanted at levels in excess of those given in Table I.2
When a person first claims used oil fuel meets these specification levels,
he must obtain an analysis or other Information to support the claim.
The final rule (40 CFR Parts 261 and 266) establishes a rebuttable
presumption that used oil containing > 1000 ppm total chlorine is mixed
with halogenated hazardous waste and, therefore, is a hazardous waste.
(Because total halogen content cannot easily be determined In the field or
laboratory, EPA has agreed to interpret "total halogen" in the final rule
as total chlorine3). The presumption may be rebutted by showing that the
used oil has not been mixed with hazardous wastes (e.g. by showing it
does not contain significant levels of halogenated hazardous
constituents). Used oil presumed to be mixed with hazardous waste is
subject to regulation as hazardous waste fuel when burned for energy
recovery.
In addition, the rule establishes a specification for used oil fuel
(I.e. used oil not mixed with hazardous waste) that is essentially exempt
from all regulation and may be burned in nonindustrlal boilers, provided
that allowable levels for designated toxic constituents, flash point and
total chlorine (4000 ppm) are not exceeded. Used oil exceeding any of the
specification levels is termed "off specification used oil" and is subject
to regulatory control. This second chlorine threshold is set to ensure
that harmful emissions of hydrochloric acid are not emitted from the
boiler or allowed to corrode it and reduce its efficiency.
To rebut the presumption, the final regulation requires the seller of
the oil fuel to prove that none of the solvents listed under EPA Hazardous
Waste Numbers F001 and F002 is present at > 100 ppm or that no other
hazardous chlorinated organics are present. These include the degreasing
and spent halogenated solvents methylene chloride, tetrachloroethylene,
1,1,1-trichloroethane, trichloroethylerie, 1,1,2,2-tetrachloroethane,
dlchlorodifluoromethane, and 1,1,2-trichloro-l,2,2-trlfluoroethane.
548
-------
As a consequence of this regulation, the regulated community and the
enforcement authorities will need to determine total chlorine In used oils
and oil fuels. The ideal analytical technique for this determination
should meet several perfornance criteria. The technique should be
accurate such that it can determine if the chlorine content is abo've or
below 1000 ug/g. It should also be precise such that replicate analyses
of a veil nixed sample yield reproducible results. It should be rapid,
field portable and inexpensive. The uBed oil-to-fuel Industry is somewhat
dynamic, requiring in many cases onsite decisions as to acceptability of
oil shipments. Ideally, the analytical determinations should be carried
out in the field to quickly identify contaminated oil and oil fuel.
Finally, the testing procedures should be simple enough so that users over
a wide variety of skill and experience levels can perform the analyses.
These can be expected to include generators, burner*, truck driver*,
laboratory technicians, and enforcement personnel.
To provide the regulatory coiwunity and regulatory agencies with
appropriate chlorine test methods, the EPA Office of Solid Waste (OSW) and
its contractor, Research Triangle Institute (BTJ) hava conducted a pilot
study to identify and evaluate candidate analytical techniques for
determination of chlorine in used oils and oil fuels. The results of this
¦tudy. which are reported here, will be used to develop written teat
methods for the most promising techniques. These test methods Hill then
be subjected to a formal collaborative study to generate precision and
accuracy data for each method. These Methods will then be proposed in the
Federal Register aB mandatory for compliance with the existing used oil
regulation.
2. Approach ,
2.1 selection of Techniques for Evaluation
Candidate techniques were identified from reviews of the literature
and from discussions with testing laboratories, equipment vendors and the
oil Industry. A broad spectrum of analytical approaches was considered.
These ranged from published analytical test methods to prototype
instrumentation and test kits for which development was Incomplete. This
complicated the evaluation of results, but Inclusion of these varied
methods, techniques, and instruments was necessary to assess the options
available to EPA and the regulated community. A brief summary of each
technique is provided below, Details are given in the final report on
this study4.
The bomb combustion technique is based on ASTM Method D808~8i. The
oil sample la pressure oxidized by combustion in a stainless steel
cylinder containing oxygen. The liberated chloride is absorbed in a
sodium carbonate/bicarbonate solution contained in the bomb. The ASTM
Method uses a gravimetric precipitation of chlorine as silver chloride as
the analytical finish. However, this finish was found to lack adequate
sensitivity at the 1000 ug/g level. For this study aeveral alternative
analytical finishes were evaluated, These included Ion chromatography,
nercurlnetric titration, silver nitrate titration, and a ferrioyanide
colorimetrlo method.
549
-------
Instrumental microcoulometry (MCT) Is an automated technique in which
an oil sample is first burned in a combustion chamber under oxygen to
convert the chlorine to chloride. The chloride Is then reacted with
silver ions in a titrating cell, where the silver ions are replaced
coulometrically. The total current required to replace the silver Ion is
a measure of the chloride present in the Injected sample. Three MCT
instruments currently available were evaluated. These are the
COSA/Mltsubishl TSX-10, and the Dohrmann DX-20B and MCTS-20 analyzers.
An automatic instrumental chlorine determinator from LGCO Corp., the
Cl-35 was evaluated. It combusts oil samples under oxygen in a resistance
furnace. The liberated hydrochloride gas is absorbed and measured by a
chloride specific ion electrode.
A prototype field portable Instrument based on the Beilsteln test for
chloride was evaluated. A probe containing copper wool Is Immersed In the
sample and heated In a hydrogen flame. If chlorine is present, copper
chloride emission will occur. The concentration of chlorine in the sample
should be directly proportional to the intensity of this emission and can
thus be quantified.
Two field portable disposable test kits were evaluated. Both kits
operate by first dehalogenating chlorine from chlorinated organlcs
(solvents, PCBs, etc.) by reacting the oil sample with a sodium napthenate
mixture In a diglyme catalyst. The chloride released by this reaction is
then extracted into an aqueous buffer and titrated to the mercuric nitrate
endpolnt.
In the Dexsil Corp. kit, called Chlor-D-Tect 1000, the amount of
sample and reagents are fixed such that a dark blue color results if the
chlorine content is below 1000 ug/g and a yellow color results if the
concentration exceeds 1000 ug/g. There Is a relatively narrow transition
zone over which the color change occurs.
The Chemetrlcs Inc. kit called Quantl-Chlor, involves largely the
same chemistry as the Dexsil kit with the exception that the extracted
chloride is titrated into an evacuated ampule containing mercuric nitrate
solution to a yellow or colorless endpolnt. The chlorine concentration in
the oil over a range of 750-7900 ug/g Is read from a scale on the ampule.
The final technique evaluated was x-ray fluorescence (XRF)
spectrometry. This is a nondestructive technique In which the sample is
placed in a disposable plastic cup covered with an x-ray permeable
membrane and irradiated with x-ray radiation from a source within the
Instrument. The Irradiated sample will then emit fluorescent element
specific x-rays whose intensity can be quantified and correlated with the
element's concentration in the sample.
Several types of XRF systems were evaluated. These included one
field portable energy dispersive (EDXRF) system (the CSI XMET-840),
several laboratory EDXRF systems (Oxford Analytical Lab X-2201, Horiba
MESA 200, Tracor Spectrace 5000, Philips PV 9000), and two wavelength
dispersive (WDXRF) systems (Philips 1400 and Oxford ChemX). These systems
ranged In complexity from a system dedicated to determining only chlorine
and sulfur to multielement systems capable of determining up to 60
elements.
550
-------
2.2 Selection and Preparation of Test Samples
Both virgin and used oils and oil fuels representative of those
subject to this regulation were obtained and characterized for chlorine
content. These Included crankcase, industrial hydraulic, netalworking and
No. 6 fuel oil. Blends of No. 6 fuel oil with used oil were also created
to simulate the final product of the used oil-to-fuel market.
These oils were then spiked with volatile, semivolatile, nonvolatile
and inorganic chlorine over the range 0 to 10,000 ug/g, with most of the
spike levels between 500 to 2000 ug/g. to cover the range of greatest
Interest. Chlorinated solvents were the volatile organics spiked,
chlorinated benzenes the semivolatile organics and sodium chloride, the
inorganic chlorine compound. Chlorooctadecane, a long chain paraffin
hydrocarbon was selected as the nonvolatile spike compound to simulate
similar compounds present as extreme pressure additives in metalworking
fluids.
A total of 40 samples were supplied to each of nearly 20 volunteer
laboratories from industry, equipment vendors, government and commercial
testing laboratories. RTI acted as the referee laboratory and performed
some of the testing by the bomb, test kit and XRF techniques.
Each laboratory was requested to analyse a total of 3 aliquots from
each sample bottle and to report the chlorine content measured in each.
3. Results and Discussion
The accuracy of each technique was assessed on the basis of percent
bias of the average of the three results for each »«»ple fro* of the
expected value based on spiking and background level characterization.
Typical results for several of the techniques are given in Table II
for a used crankcase oil and a 1:9 blend of used crankcase and No. 6 fuel
oil. The LECO and Beilstein techniques were dropped from further
consideration. It was determined that the LECO device has a detection
limit near the 1000 ug/g level and thus suffered from poor accuracy at
this level. The Beilstein device was determined to be unsuitable
determining chlorine in used oils, apparently due to an Interference due
to water. It was also insensitive to all forms of chlorine except:
chlorinated solvents.
The bomb technique with these analytical finishes and the MCT
technique all gave results with ±10 percent of expected. The 1
results were generally within ±25 percent of expected with few false
positives or negatives for the Oexsil kit.
XRP results varied somewhat, although the complexity_of *jje_
instrument was apparently not a factor. In general, the biaseawere
around 30 percent low. probably due to absorption of X-rays bywerin
the oil sample. This is sometimes compounded if the samplebeginsto
stratify in the cell daring the X-ray counting period. It is thought to
be technique and not instrument related. One instrument, the MESA 200
suffers from an interference due to calcium in crankcase oils which
results in high biases near 1000 ug/g-
551
-------
In summarizing the performance of these techniques (Table III),
additional factors such as susceptibility to matrix effects, analysis
time, coat per analysis and field portability were determined or
estimated.
The bomb/wet analytical techniques, the test kits, and the MCT all
gave similar analytical performances, The basic difference is one of
analysis time/sample, and cost/analysis. The bomb methods are tedious and
labor intensive, while the test kits are rapid, have a fixed analytical
cost that is somewhat lower and can be used in the field. The MCT systems
are also rapid and in theory provide relatively low coat analyses.
The XRF technique is subject to matrix effects primarily due to
water. However, the analysis time per sample, and the nondestructive
nature of the technique make it attractive for high sample throughput.
4. Conclusions and Future Work
Techniques have been identified to determine the chlorine content in
used oils in the field and in the laboratory. Based on this pilot study,
written test methods are being developed for the bomb, test kit, MCT, and
XRF techniques. These will be subjected to a formal collaborative study
to generate precision and accuracy data for each method. These methods
are to be proposed in the Federal Register as mandatory for compliance
with this used oil regulation.
As part of this continuing evaluation, a cost effective used oil
screening program is being developed which will combine field and
laboratory testing to minimize costs, turnaround time, and ensure that the
sale of solvent contaminated oil as fuel is prevented.
/6. Acknowledgments
The authors thank the participants in the collaborative study, the
suppliers of the oil samples, and Ms. E. Kittrell for preparing the
manuscript.
7. References
1. Franklin Associates, Ltd. 1984. "Composition and Management of Used
Oil Generated in the United States." Prepared for the U.S. EPA/OSW,
under Contracts 68-02-3173 and 68-01-6467.
2. U.S. Environmental Protection Agency. Federal Register. 50, 49164-
492J1, November 29, 1885,
3. Gasklll, A., RTI to D. Friedman, EPA/OSW, personal communication,
January 29, 1986.
4. Gasklll, A., E. 0. Estes, D. L. Hardison. Evaluation of Techniques
for Determining Chlorine in Used Oils, Volumes I-IV. Prepared for the
U.S. EPA/OSW, under Contract No. 68-01-7073, WA. No. 58, June 1987.
552
-------
TABLE I. USED OIL FUEL SPECIFICATIONS FOR OIL THAT MAY BE BURNED
IN N0N1NDUSTRIAL BOILERS
Constituent/Property Allowable level for burning
without regulation
Arsenic
<5 ppm
Cadmium
<2 ppm
Chromium
<10 ppm
Lead
<100 ppm
Total Chlorine
<1,000 ppma
<4,000 ppmb
Flash Point
>100*F
aLevel presuming mixing with hazardous waste.
bLeve] above which burning 1b not permitted if presumption can be
rebutted.
TABLE II. TYPICAL RESULTS, * BIAS
Technique
Used oil
1320 ug/g
Used oil fuel
blend, 998 ug/g
1. Bomb
IC
-8
-4
Mercuric nitrate
-2
-8
Silver nitrate
-7
-13
Ferrlcyanide
+2
-5
2. Test kit
Fixed point
-25
+ 10
Variable endpoint
-24
+35
3. Mlcrocouloaetry
+11
+9
4. EDXRF
Portable
+5
+6
Nonportable
-21
-16
5. WDXRF
-30
+31
553
-------
TABLE III. TECHNIQUE PERFORMANCE SUMMARY
Evaluation
criteria
Bomb/Wet
Test kits
MCT
XRF
* Bias
<10*
±25*
±10*
-30*
* RSD
<10*
10*
5*
10*
False positives,
negatives?®
Never
Seldom
Seldom
Sometimes
Matrix effects?
Water
Oil type
Chlorine form
Never
Never
Never
Seldom
Never
Never
Never
Never
Never
Often
Sometimes
Sometimes
Analysis time/sample
45 min
10 min
10 min
<5 min
Cost/analysis'5
$25
$11-14
$3-5
$3-5
Portability
No
Yes
No
XMET-840
areBult not within ±25% of the expected value was considered a false
positive or negative determination
bbased on theoretical aample throughput with a technician working full
time (8 hours/day)'
554
-------
OVERVIEW OF THE INTEGRATED AIR CANCER PROJECT
Joellen Lewtas and Larry Cupitt
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
The Integrated Air Cancer Project (IACp) is an interdisciplinary
research program designed to develop the scientific methods and data sets
needed to identify the major carcinogenic chemicals in the atmosphere? to
characterize the emission sources and chemical precursors which give rise
to the identified chemicals; and to improve the methodology and data avail-
able for assessing human exposure and risk due to exposure to airborne
carcinogens' The research effort is focused primarily on characterising
the impact of complex mixtures of products of incomplete combustion, includ-
ing the gaseous, semi-volatile, and particle-bound organic species.
These objectives will be approached in a stepwise manner by conducting
a series of field studies in areas of increasing complexity. The initial
work has concentrated on relatively single airsheds where residential wood
combustion and automobile emissions are the only significant combustion
sources. Future efforts will add other residential combustion sources
(e.g., fuel oil) and industrial sources.
The first phase of this project was conducted at Raleigh, SC and
Albuquerque, NM during the winter of 1984-as. The focus of these initial
efforts was» (1) to develop, evaluate, validate, and document th« sampling
and analysis methods necessary to accomplish the technical research, and
(2) to attempt to extend current source apportionment techniques to account
for the observed mutagenicity of the collected ambient aerosol samples.
During FY-86/87 an extensive field program was conducted in Boise, ID,
555
-------
Introduction
Major scientific questions regarding the relationship between air pol-
lution and human cancer remain unanswered. This is due, in large measure,
to the fact that estimates of human cancer caused by air pollution are
highly refractory to the traditional techniques of epidemiology and exposure
and risk assessment. The length of time (decades) that may pass between
exposures to airborne carcinogens and diagnosis of cancer has precluded
both meaningful prospective epidemiological studies of human populations and
quantitative estimates of the risk from exposure to mixtures of airborne
carcinogens. Retrospective studies are also difficult to perform since one
does not know when the observed cancer was initiated. Because most nonoccu-
pational cancers are not specific and could have multiple causations, the
process of associating an observed cancer with a given putative agent has
not met with great success. In 1977 and again in 1982, international
experts concluded that air pollution arising from combustion products of
fossil fuels, probably acting together with cigarette smoke, have been
responsible for 10% of all cancers for the United States^'2, This estimate
is really a "best guess" based on studies of the urban versus rural cancer
rates. Doll and Peto3, in a review of causes of cancer in the U. S., sug-
gest that a much lower rate of 2% of cancer deaths could be attributed to
pollution. Karch and Schneiderman^ on the other extreme suggest that past
analyses of this problem have overestimated the contribution of smoking and
underestimated the multicausal nature of cancer. They estimate that at
least 11% and more likely 21% of lung cancer is related to air pollution.
Regardless of which of these estimates proves to be the most accurate,
there is certainly reason to believe that the carcinogenic potential of air
pollution may be a serious public health problem. EPA has undertaken a
major research program directed specifically at clarifying the exposure,
risk and sources of carcinogens in the air. This relatively long-term
research program, initiated in 1985, is the Integrated Air Cancer Project
(IACP). As a part of the IACP, a methodology is being developed to allow
an estimate of the future cancer risk associated with human exposures to
the direct and transformed emissions from identifiable sources. The
Environmental Protection Agency has the responsibility and authority to
regulate the emission of carcinogens into the air (e.g., Section 112, Clean
Air Act), however, the identification of which carcinogens and which emis-
sion sources are of greatest potential human risk is not known and remains
a high-priority research effort. The IACP adopts an approach which focuses
on identifying those species actually present in the air which are most
likely to be carcinogenic and attempts to determine how they came to be
present in the environment. The accomplishment of this research requires
the IACP to bring together the multi-disciplinary talents of chemists, phys-
icists, engineers and toxicologists to conduct a series of coordinated
field and laboratory studies of the emission sources, the atmospheric trans-
port and transformation, human exposures, and the toxicological effects, in
order to address the complex issues involved.
Integrated Air Cancer Project Goals and Strategy
tioals
Identify the Principal Airborne Carcinogens. airborne carcinogens
may exist as gaseous or volatile organic compounds (VOCs), as semi-volatile
organic species (SVOCs), or as particulate-bound organic compounds [e.g.,
polycyclic organic matter (POMs)J. The identities and relative contribu-
tions to the total atmospheric burden of carcinogens by many of the organic
chemicals present in the atmosphere are not yet known. The IACP is develop-
556
-------
ing the methods and data necessary to begin to characterize many oC these
species. Within each of these groupings (e.g., VOCs, SVOCs, and POMs), the
specific chemicals or chemical classes which contain mutagenic and carcino-
genic compounds need to be identified, and their concentration in the envi-
ronment needs to be measured. These carcinogens may be direct emissions
from the sources, or they may be the transformation products produced from
reactions of simple hydrocarbons or complex polynuclear aromatic hydrocar-
bons with acidic and oxidizing gases in the atmosphere. The complexity of
the problem requires that an integrated approach be used to address these
goals. Not only must the emissions from the various sources be identified,
but the transformation products from those emissions must also be determined.
Determine Which Emission Sources ftre the Major Contributors of Car-
cinogens to Ambient Air. Before one can reduce human exposure to hazardous
air pollutants, the sources of the carcinogenic compounds themselves, or
their precursors, have to be identified. Field studies with simultaneous
emission characterizations and ambient monitoring, followed by source
apportionment calculations, will be conducted to determine which emission
sources are the major contributors of carcinogens to the ambient air.
Characterization of emissions and ambient samples will be by both chemical
and biological (mutagenicity and carcinogenicity) testing.
Improve the Estimate of Human Exposure and Comparative Human Cancer
Risk from Specific Air Pollution Emission Sources. In order to improve the
estimate of relative human cancer risk from specific air pollution emission
sources, a comparative methodology is being developed to evaluate and
utilize short-term mutagenesis and animal carcinogenesis data on emission
sources. In addition, better total human exposure estimates will be devel-
oped for these complex emission products and individual carcinogens includ-
ing transformation products.
Strategy
The long range goals of this project will require many years to attain.
Approaching the long term goals through field studies of relatively simple
airsheds will, in a stepwise fashion, yield the methodology to understand
the more typical complex airsheds containing emissions from multiple sources
experienced by the general population.
Current data indicate that motor vehicles and residential home heating
sources make a significant contribution to the mutagenic activity of ambient
air samples. Therefore, the initial phases of IACF were aimed at quantify-
ing carcinogens emitted from residential woodstoves and motor vehicles.
Communities were selected with relatively simple airsheds where a signifi-
cant percentage of the homes use wood as the major heating fuel. The
initial IACF studies conducted in Raleigh, NC and Albuquerque, KM emphasized
field and laboratory evaluation to select sampling and analysis methodolo-
gies for a major field study initiated in Boise, ID in the winter of 1986-
87. Transformation studies in laboratory simulations are assessing the
role of atmospheric chemistry in altering the chemical composition and muta-
genicity of the emitted chemicals. Simulations of the field study condi-
tions will be used to determine the critical factors and chemical processes
which transform the species between the source and the receptor.
Technical Approach
Integrated Field Studies
The field studies were designed to simultaneously sample and charac-
557
-------
terize the emissions at the source (e.g., woodstoves or motor vehicles) in
the ambient air near to specific sources and in the ambient air distant from
an aggregate of sources^. Sampling indoors and outdoors of homes both with
and without woodstoves was designed to provide an indication of the total
human exposure. In general, source and ambient emission measurements were
made during heating and nonheating periods as well as separate daytime and
nighttime sampling periods to differentiate between wood combustion emis-
sions and those from motor vehicles- Indoor, ambient, and source emissions
were collected comparably and analyzed identically to provide a valid com-
parison between the fresh emissions and the chemicals at the receptor.
Neighborhood survey data of motor vehicle and residential heating use was
also recorded. Meteorological measurements and tracer studies were con-
ducted throughout field tests as input to the source apportionment and
atmospheric transformation studies.
The initial IACP field study in Raleigh, NC included major components
of methods development and evaluation. Factors such as face velocity,
sample integrity, and sample extraction and preparation methods were evalu-
ated, and standard procedures developed for subsequent studies. Sampling
methods were developed so that comparable samples could be taken simul-
taneously from woodstove sources and ambient air (indoors and outdoors)•
In order to collect large (gram) quantities of ambient air particulate
matter for bioassay directed fractionation/characterization and carcino-
genesis studies, new high volume samplers have been designed^. The sampling
protocols take into consideration techniques necessary to collect particu-
late material as well as volatile, semi-volatile, and condensable organic
material. Mew sampling and analysis methods were developed to prepare,
sample, and extract large quantities of XAD-2 resin for organic characteri-
zation and bioassay studies. The results of the study included sampling
protocols for source, ambient, and microenvironment air collection.
Analysis Strategies
Source Apportionment. Source apportionment is a combination of math-
ematical and analytical procedures which are used to determine the contri-
bution of specific emission sources to measurements of air pollution.
Several methods have been used to apportion the contributions of source
emissions to ambient air quality including emission inventory methods,
source dispersion models and receptor models. The receptor-model approach
to source apportionment which is being used in the IACP is especially
effective where the number of sources are small and well characterized. In
addition, emissions from mobile sources and wood burning contain unique
elemental tracers which improve the accuracy of the source apportionment
calculations•
Source apportionment of mutagenic activity and the use of organics as
tracer species in receptor model studies have not been extensively evaluated
and are a major component of the IACP7. The unique feature in the receptor
modeling studies in the IACP are that several types of information will be
available. The receptor modeling will use a combination of elemental
tracers (e.g., K and Pb), radiocarbon8, organic tracers®, and mutagenicity10
in the source apportionment calculations.
Bloaasay Directed Fractionation/Characterliatlon. This approach has
been successfully usea to identify potential carcinogens in complex mixtures
including synthetic fuels, diesel emissions, and urban air particles^. The
complex mixtures are fractionated and eaoh fraction is bioassayed. Mutagen-
ically active fractions are further fractionated, bioassayed, and character-
ized until the major class or specific compounds responsible for the muta-
558
-------
genicity are identified. The IACP ia using the standard Ames Salmonella
typhimurium plate incorporation mutagenicity assay, together with new
micromutagenesis bioassays, coupled directly to analytical fractionation
procedures1®.
Atmospheric Transformation. A growing number of laboratory and field
studies have suggested that there are unidentified compounds present in
ambient air which cause increased mutagenic responses in both the vajJor and
aerosol phases. Chemical analysis has not fully identified many of these
compounds, but the chemical class information suggests that the target com-
pounds may be the partially-oxygenated or -nitrogenated reaction and trans-
formation products of emitted organics. Recent studies have demonstrated
that exposure of many organic source emissions to photochemical reaction
conditions increases the mutagenicity of the emissions, especially the
gaseous components 1^•
The transformation studies carried out as part of the IACP will utilize
a large reaction and atmospheric simulation chamber to produce irradiation
products of air mixtures like those found in typical U.S. environments and
in the IACP field study. The simulation chamber
will produce consistent conditions which will permit exposure of bioassay
test systems and will provide large quantities of the transformation prod-
ucts for chemical sampling and analysis. Bioassay studies are conducted on
the starting materials and the irradiation products13.
Human Exposure Assessment. Since the initial phases of the IACP are
focusing on characterizing the principal organic mutagens and carcinogens
from products of incomplete combustion (PICs) the exposure assessment
component of the program is designed to characterize the important human
exposure environments indoors and outdoors which are impacted by PICs.
Characterization of the total organic exposures will include measurement of
VOCs via both canister and Tenax sampling and GC analysis14, SVOCs via XAD-2
sampling and analysis, and total particulate organic analysis. Specialized
sampling and analysis of aldehydes was conducted using the 2,4-dinitrophen-
ylhydrazone (DNPH) cartridge method14. Specific analyses are being con-
ducted to quantitate exposures to polycyclic aromatic hydrocarbons (PAHs)®
and selected polar oxygenated and nitrated organics.
Comparative Carcinogenesis and Mutagenesis Assessment. In order to
develop a comparative data base for risk assessment, a battery of bioassays
will be used to evaluate the complex mixtures from the field study filter
samples. A comparative potency method has been developed for cancer risk
assessment based on a constant relative potency
hypothesis and using data from a battery of short-term mutagenesis bioassays
and animal turorigenicity studies on a series of diesel vehicle emissions'^.
The same bioassays have been used to evaluate complex emissions for which
human lung cancer risk estimations are available from occupational and
relatively high dose exposures (emissions from coke ovens, roofing tar
pots, and cigarette smoke) thereby providing a comparative basis for esti-
mating lung cancer risk from emissions for which human data are not avail-
able. Recent EPA studies have expanded the combustion emission data set
for which comparative mutagenesis and carcinogenesis data are available to
include gasoline vehicle emissions, coal, oil, and wood combustion16. The
IACP will further extend this data bose to include comparative mutagenesis
and carcinogenesis studies of composites of ambient particulate organic*
fro.- the airsheds under study.
559
-------
Summary
The IACP Is a long-term research project with goals of (1) identifying
the principal airborne carcinogens, (2) determining which emission sources
are major contributors of carcinogens to ambient air, and (3) improving the
estimate of human exposure and comparative human cancer risk from specific
air pollution emission sources• Ultimately, this research will provide the
methodology and data to estimate the comparative risks associated with
human exposure to various combustion emissions sources and other sources•
Although accomplishment of the long-range goals of the IACP will
require many years to attain, the IACP concept is to approach the goals in
a stepwise fashion to provide short-term data on specific source categories.
Studies are aimed at understanding both the chemistry and toxicology of
complex mixtures of air pollutants focusing first on relatively-isolated
airsheds, and progressing to more-typical and more-complex environments,
with multiple source contributions. During the initial phases of the IACP,
emphasis has been placed on the source categories of residential wood
combustion and motor vehicles. Subsequent phases of this
project will add other residential combustion sources (e.g., oil) and
industrial sources.
References
1. B. Holmberg, U. Ahlborg, "Consensus report: mutagenicity and car-
cinogenicity of car exhausts and coal combustion emissions," Environ.
Health Perspect. 47; 1 (1983).
2. Report of the Task Group, "Air pollution and cancers risk assessment
methodology and epidemiological evidence," Environ. Health Perspect.
22s 1 (1978).
3. R. Doll, R. Peto, "The causes of cancers quantitative estimates of
avoidable risks of cancer in the United States today," J. Natl. Cancer
Inst. 66: 1191 (1981).
4. N. J. Karch, M. A. Schneiderman, "Explaining the urban factor in lung
mortality," National Resources Defense Council, Washington, DC (1981).
5. V. R. Highsmith, R. E. Baumgardner, R. J. Drago, T. A. Lumpkin, D. B.
Harris, R. G. Merrill, R. B. Zweidinger, "The collection of neighbor-
hood air samples impacted by residential wood combustion in Raleigh,
NC and Albuquerque, NM," APCA proceedings this volume (1987).
6. R* M. Burton, V. A. Marple, B. Y. H. Liu, B. Johnson, "A novel 2.5 i»m
cut virtual impactor for high volume PM10 sampling," APCA proceedings
this volume (1987).
7. R. K. Stevens, C. w. Lewis, T. G. Dzubay R. E. Baumgardner, R. B.
Zweidinger, R. V. Highsmith, L. T. Cupitt, J. Lewtas, L. D. Claxton,
L. Currie, B. Zak, "Source apportionment of mutagenic activity of fine
particles collected in Raleigh, N. C. and Albuquerque, N. M.," APCA
proceedings this volume (1987).
8. G. A. Klouda, L. A. Currie, A. E. Sheffield, S. A. Wise, B. A. Benner,
R. K. Stevens, R. G. Merrill, "The source apportionment of carbanaceous
combustion products by micro-radiocarbon measurements for the inte-
grated air qancer project (IACP)," APCA proceedings this volume (1987).
560
-------
9. R. G. Merrill, "Progress toward identifying source specific tracers,"
APCA proceedings this volume (1987).
10. L. D. Claxton, S. H. Warren, V. S. Houk, J. Lewtas, "The mutagenicity
of ambient and source samples from woodsmoke impacted airsheds," APCA
proceedings this volume (1987).
11. D. Schuetzle, J. Lewtas, "Bioaasay-directed chemical analysis ip
environmental research," Anal. Chem. 58: 1060A (1986).
12. L. T. Cupitt, L. D. Claxton, P. B. Shepson, T. E. Kleindienst, "1ACP
emissions! transformations and fate," APCA proceedings this volume
(1987).
13. L. D. claxton, "Assessment of bacterial bioassay methods for Volatile
and semi-volatile compounds and mixtures," Environ. Int. 11i 375
(1985).
14. R. Zweidinger, D. Dropkin, F. Stump, R. Drago, S. Tejada, "Volatile
organic hydrocarbon and aldehyde composition in Raleigh, N.C. during
the 1985 woodsmoke study," APCA proceedings this volume (1987).
15. J. Lewtas, "Development of a comparative potency method for cancer
risk assessment of complex mixtures using short-term in vivo and in
vitro bioassays," Toxicol. Indust. Health 1i 193 (1985).
16. J. Lewtas, "Combustion emissions! characterization and comparison of
their mutagenic and carcinogenic activity," Carcinogens and Mutagens
in the Environment, volume V, The Workplace! Sources of Carcinogens,
CRC Press, Boca Raton, Fl. 1985, pp. 59-74.
Acknowledgments and Disclaimer
The authors acknowledge the critical support and contributions of the
other IACP Steering Committee Members: R. E. Lee, J. Dorsey, G. Tucker,
and R. K. Stevens and Team Leaders: V. R. Kighsmith, C. Rodea, and R. G.
Merrill in providing leadership in planning and coordinating this project
as well as the many scientists and engineers participating in conducting
these studies. The research described in this paper has been reviewed by
the Health Effects Research Laboratory, U.S. Environmental Protection
Agency and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
561
-------
THE COLLECTIOH OF NEIGHBORHOOD AIR SAMPLES
IMPACTED BY RESIDENTIAL WOOD COMBUSTION IN
RALEIGH, NC AND ALBUQUERQUE, NM
V. Ross Highsmith and Charles E. Rodes
Environmental Monitoring Systems Laboratory
Roy B. Zweidinger
Atmospheric Sciences & Research Laboratory
Raymond G. Merrill
Air & Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Pilot field studies were simultaneously conducted in Raleigh, NC, and
Albuquerque, NM, from January through March 1985. Consecutive twelve-hour
sampling (7AM 7PM and 7PM - 7AM) was conducted at two fixed monitoring sites
in both cities. In each city, the primary fixed sampling site was located in
a residential neighborhood heavily influenced by residential wood combustion
emissions. State-of-the-art sampling equipment and procedures were devel-
oped and tested for the collection, Btorage and shipment of field samples.
Semi-volatile and volatile organic samples were collected and analyzed in
Raleigh. Routine criteria pollutant and meteorological parameters were also
monitored at each site. Indoor, outdoor, and source sampling was conducted
at three Raleigh residences with operating woodstoves. Sample analysis
included mass loadings, inorganic ions, volatile/elemental carbon, carbon-
14, mutagenic activity, and organic characterisation.
Both Raleigh and Albuquerque nighttime FINE particle (0-2.5 micron)
mass loadings were more than twice the daytime FINE mass loadings. Approxi-
mately 60%. of the nighttime FINE samples were extractable organics. The
nighttime potassium concentrations from samples collected In both cities
were more than twice the daytime potassium concentrations, suggesting
significant impact by RWC emissions. The FINE loadings immediately outside
the residences correlated well with the fixed site measurements but were
approximately twice the FINE mass collected inside the residences. Overall,
the residential data demonstrates the variability expected with differences
in residence construction and individual activity.
562
-------
INTRODUCTION
Advances in airtight woodstove technology and the relatively low cost
for wood have combined to make residential wood combustion (RWC) an attrac-
tive alternative for home heating. Laboratory studies1"4 have shown RWC
emissions to be rich in polycyclic organic material (POM) and products of
incomplete combustion (PIC's). The concentrations of these potentially car-
cinogenic compounds depend on the woodstove size, fuel source and operating
conditions. Many homeowners operate their woodstoves at very low burn rates
for two reasons: the purchased stove is too large for the installed area, or
in an attempt to maintain overnight stove operation. The PIC's contributed
to the ambient air from residential woodburning significantly increase
under these oxygen-starved conditions. Communities dominated by residential
woodburning report ambient outdoor air quality degradation directly corre-
sponding to RWC emissions.In addition, recent residential studies
contribute increases in indoor air pollutants to the use of woodstoves and
fireplaces for home heating."*^ The potential total human exposure to
RWC emissions, both indoors and outdoors, is a primary area of concern to
the scientific community.
The Integrated Air Cancer Project (IACP) is a long term US Environmental
Protection Agency (EPA) program with three major objectives: identifying
principal airborne carcinogens; determining the major emission sources for
these carcinogens; and improving the estimate of comparative human cancer
risks from specific air pollution emission sources. Initial IACP studies
were planned for simple airsheds in order to develop the procedures needed
to characterize complex airsheds typical of more industrialized urban
areas. Residential Wood Combustion (RWC) was selected as the first source
for evaluation based on its relative use and overall impact on the national
air quality.10 RWC impacted samples, high in mass and organics, would
also support the initial IACP sampling and analytical methods development
initiatives.
Pilot field studies were simultaneously conducted during 1985 in
Raleigh, NC and Albuquerque, NM to evaluate state-of-the-art monitoring for
air carcinogens. The Raleigh study served as the major sampling location
allowing maximum EPA personnel involvement while focusing on sampling and
analytical methods development. Both ambient fixed site monitoring as well
as residential indoor monitoring were conducted in Raleigh. The residential
study was designed to evaluate methodologies for the simultaneous collection
of indoor, outdoor, and RWC source samples* Source apportionment/receptor
model analysis procedures outlined by Lewis and Einfield11 was the primary
focus of the Albuquerque study. Routine particulate and gaseous samples
were collected at each sampling location. The majority of the ambien^par-
ticulate samplers were modified to collect only the FIUE (0-2.5 n») particle
size fraction. Gas and particle phase semi-volatile organic compounds
(SVOC's) and volatile organic compounds (VOC's) were collected indoors and
outdoors in Raleigh but were not collected in Albuquerque. Criteria pollu-
tant and meteorological parameters were also monitored at the fixed sampling
sites.
EXPERIMENTAL
Fixed site ambient monitoring was conducted at primary and secondary
locations in both Raleigh, NC and Albuquerque, NM from January 1985 through
March 1985. The Ralaigh airshed is not significantly impacted by major local
sources but tends to reflect regional source contributions. Local topograph-
ical and meteorological conditions assist itt minimising the- impact of auto-
motive emissions on the Raleigh airshed. Albuquerque, at more than twice
563
-------
the population of Raleigh, is a high elevation city located in the Rio
Grande river basin and backs up against the Sandia mountain range to the
East. As with Raleigh, Albuquerque has no local major industrial sources
but has been shown to be impacted by RWC and mobile source emissions.11
Regional contributions to the local airshed from the Western desert are
minimal. The topography and meteorological conditions in Albuquerque favor
the retention of local emissions especially during the intense winter
inversions.
The Raleigh primary woodsmoke site was located in a residential neigh-
borhood valley away from major roadways. A background site was located at
a rural site approximately 10 miles North of Raleigh to evaluate regional
source contributions to the airshed. Each of the three Raleigh residences
selected for the study had nonsmoking occupants and were located within one
quarter of a mile from the Raleigh primary sampling site. The Albuquerque
primary woodsmoke site was located at the site of the earlier study.11 A
second Albuqerque site was located approximately two miles Northwest of the
primary site at a major roadway intersection. Both Albuquerque sites were in
close proximity (ca. 3/4 mile) of interstate highways.
Each fixed monitoring site was established in accordance with Inhalable
Particulate Network (IPN)1^ and NAMS/SLAMS criteria.13 The types and
numbers of samplers operated at each site are shown in Table I. Uniform
sampling procedures were developed and employed in both cities. Twelve-hour
ambient sampling (changeovers at 7AM and 7PM) was conducted to evaluate day-
time versus nighttime source contributions. Samples for mass and inorganic
analysis were collected on preweighed quartz or Teflon* media. Samples for
bioassay and carbon analysis were collected on Pallflex T60A20 Teflon* im-
pregnated glass fiber (TIGF) and IPN equivalent quartz filter media. Vapor
phase SVOC's passing through particulate filters were collected on XAD-2
absorbent filled canisters installed immediate downstream of the filter.
Cartridges impregnated with 2,4-dinitrophenylhydrazine (DNPH) and evacuated
SUMMA* polished canisters were used to collect aldehydes and VOC's, respec-
tively. Denuder equipped samplers were operated at the two primary sites for
N(>3and HNO3 collection. Criteria pollutant and meteorological parameters
were monitored using reference procedures. SVOC, bioassay and carbon sam-
ples were stored at -80* C immediately following sampling. Particulate
samples for mass determination were conditioned for twenty-four hours fol-
lowed by standard gravimetric analysis. Aldehyde and denuder samples were
stored at -4"C.
Residential sampling followed the ambient procedures outlined above.
Samples were collected on two consecutive nights at each residence, sampling
one residence at a time. Table II lists the samples collected simultaneously
indoors, outdoors, and at the Raleigh primary site. Particulate and SVOC
samples were also collected directly from the operating woodstove flue using
the Woodstove Dilution Sampling System (WSDSS). *it"1E> All the XAD-2 car-
tridges were placed outdoors during sampling to ensure comparability between
residential, ambient, and source samples. Standard SFg tracer release tech-
niques were used to measure residential air exchange rates.
ANALYSIS
The analytical procedures used in this study are referenced in Tables I
and II. A set of fifty ambient site sampling periods (combination of day and
night periods) were selected from each city for detailed analysis. Particle
bound SVOC's were solvent extracted using dichloromethane. Aliquots were
evaporated to dryness for gravimetric determination. Additional aliquots
were solvent exchanged with dimethylsulfoxide (DMSO) and stored at -80* C
564
-------
in preparation for bioassay analysis* The Ames bioaesaay, with and without
the activation agent S9t was ueed to determine sample mutagenic activity.
Residential samples for bioasaay analysis were independently processed by
the laboratory within 48 hours after sample collection. Particle and gaB
phase SVOC's were independently extracted with dichloromethane, AliquotB
of each particulate extract was evaporated to dryness aad quantitatively
determined. The quantity of XAD-2 extracts were determined by gas chroma-
tography and gravimetric analysis. Aliquots of the sample extracts were
DMSO solvent exchanged and stored at -80* C pending bioasaay analysis.
Since the indoor Bample mass was small, the mutagenic activity of all the
residential samples vsb determined using the forward mutation assay method.
RESULTS AND DISCUSSION
The data for PINE particle mass collected iti tha 9C Raleigh and 103
Albuquerque ambient sampling periods are shown in Figures 1 and 2, respec-
tively. FINE mass loadings exceeded 35 ^g/mJ
-------
concentrations with increased RWC emissions. Napthalene, a key component
in woodsmoke, was elevated during RWC impacted periods,
The impact of increased RWC emissions on pollutant concentrations is
shown in Figure 5. Twelve-hour sampling periods and corresponding FINE mass
loadings are also shown. Although daytime lead and potassium levels were
equal (ca, .07 (ig/m^), nighttime potassium levels increased by a factor of
7.5 compared to a fourfold increase in lead levels. This data suggest that
during this period of inversion the Raleigh residential airshed was impac-
ted not only by high RWC emissions, but also by other emission sources
being concentrated by the local meteorological conditions.
The analysis data for the residential measurements are summarized in
Tables IV and V. Four key factors impacted on the data reported. Resi-
dence 1 was noted as having a leaking woodstove which could be detected
from the occasional odor throughout the house. The woodstove plume from
the woodstove chimney at Residence 2 was observed periodically downwashing
on the outdoor samplers. Residence 3 had an operational electrostatic
precipitator in the cold air return duct during sampling. Electrical
constraints at Residence 1 did not allow the operation of the outdoor
dichotonious sampler or CO analyzer.
At the three homes monitored, indoor FINE and PM^q loadings were signi-
ficantly less than the corresponding outdoor loadings, which differs from
previous studies.8The minimal FINE loadings observed in Residence 3
are attributed to the electrostatic precipitator. Outdoor and primary site
FINE and PMiq loadings agreed well and indicate uniform distribution of
emissions over the neighborhood. Indoor COARSE/FINE ratios were larger than
corresponding outdoor values and are attributed to homeowner activities
(i.e., vacuuming, walking, pets, etc.) causing reentrained coarse particles.
Other key inorganic components were uniformly distributed between indoors
and outdoors. The tracer gas studies suggest a complete exchange between
indoor and outdoor air once every two hours. The WSDSS source samples
compared well with laboratory test samples and are reported by Kerrill and
Karris.15
Residence I samples yielded the highest levels of gas phase extractable
organics and napthalene, a key component for RWC. These are attributed to
the leaky woodstove. A alight increase in outdoor extractable organics
observed at Residence 2 is contributed to the dotrawashed plume. Overall, the
mutagenicity of the outdoor residential particulate samples was 3-4 times
greater than the corresponding indoor sample activity. Contaminated XAD-2
field blanks prevented a similar comparison between indoor and outdoor gas
phase mutagenicity. Aldehyde and VOC concentrations were slightly elevated
indoors. In the three homes sampled, the distribution of VOC's did not
change from indoor to outdoor samples nor with increased FINE mass loadings.
The distribution of aldehydes indoors differed from the outdoor levels with
elevated formaldehyde concentrations being observed indoors.
CONCLUSIONS
Many of the procedures tested in these studies for collecting, storing
and analysing ambient, residential and source samples can be used to support
future IACP objectives. The ambient data documented the impact of RWC emis-
sion on both Raleigh and Albuquerque during the 1985 winter heating season
as well as the influence of mobile source emissions on the Albuquerque
airshed. The residential study results provide observations that can be
used to plan future indoor studies. FINE and levels were lower indoors
566
-------
than outdoors suggestuig Chat when operated correctly, an airtight woodatove
should not directly influence indoor air quality. Leaks in the stove or flue
allowing RV1C emissions to enter the home will, however, elevate the indoor
pollutant concentrations. The data suggest that the homes evaluated in
this study were also influenced by other sources such as materials used to
build or furnish the home and homeowner activities. This emphasizes the
importance of preliminary queetionaire information and comprehensive data
collection during exposure studies.
ACKNOWLEDGEMENTS
The authors would like to thaak Ralph Baumgardner, ASRL/EPA for coordi-
nating the Albuquerque field study, Ron Drago, EMSL/EPA for coordinating
the Raleigh residential study, Tom Lumpkin, EMSL/EPA for coordinating
Raleigh continuous- monitoring and Susan Lumpkin, EMSL/EPA and Charles
Weant, Northrop Services, Inc. for conducting the Raleigh ambient study.
DISCLAIMER
The research described in this paper has been reviewed by the Environ-
mental Monitoring Systems Laboratory, US EPA and approved for publication.
Approval does not signify that the contents necessarily reflect the views
and policies of the Agency nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
REFERENCES
1. D, G. DeAngells, D. S. Ruffin, and R, B. Resnik, "Preliminary character-
isation of emissions from wood-fired residential combustion equipment,"
EPA-600/7-80-040 CNTI8 PB 80-182066), US Environmental Protection
Agency, Research Triangle Paxil, NC 27711 (1980).
2. B, R. Hubble, J. R. Stetter, I. Geiiert, J.B.L. HarVness, ant R* D.
Plotard, "Experimental measurements of emissions from residential wood-
burning stoves." In Residential fuels.' Environmental Impacts and
Solutions (edited by J. A. Cooper, and D. Malelt], Oregon Graduate Center,
Beaver ton, OH. (1982).
3. S, s. Butcher, and E. M. Sorenson, "A study of wood stove particulate
emissions," J. Air Poll. Control Assoc.. 29. pp. 724-728 (1979).
4. R* Kamens, D. Bell, A. Dietrich, J. Perry, R, Goodman, L. Claxtoo, and
S* Tejada, " Mutagenic transformations of dilute wood smok* systems in
the presence of ozone and nitrogen dioxide. Analysis of selected high-
pressure liquid chromatography fractions from wood smoke particle ex-
tract*," Environ, fici. Tecitnol* t PP* 63-69 (1985)*
5. J, A. Cooper, " Environmental impact of residential wood combustion
emissions and ita implications," J. Air Poll. Control Assoc., 30: pp.
855-860 (1980).
6. A, E, Romero, R. M. luchan, D. G. Pot, "A study of air pollution from
fireplace emissions at Vail Ski Resort," Environ. Health. 41, 117-119
(1978).
567
-------
7. K. Sexton, J. D» Spengler, R. D. Treitman, W. A. Turner, "Winter air
quality in a wood-burning community: a case study in Waterbury,
Vermont," Atmospheric Environment, 18, 1357-1370 (1984).
8. K, Sexton, J. D. Spengler, and R. D. Treitman, "Effects of residential
wood combustion on indoor air quality: a case study in Waterbury,
Vermont," Atmospheric Environment, 18, 1371-1383 (1984).
9. D. J. Moschandreas, J. Zabransky, and H. E. Rector, "The effects of
woodburning on the indoor residential air quality," Environ. Int., 4^
463-468 (1980).
10. J. Lewtas, and L» T. Cupitt, "The integrated air cancer project: program
overview," 1987 EPA/APCA Symposium on Measurement of Toxic and Related
Air Pollutants, Research Triangle Park, NC (May 1987).
11. C. W. Lewis, and W. Einfield, "Origins of carbonaceous aerosol in
Denver and Albuqerque during winter," Environ. Int., 11, 243-247 (1985).
12. Inhalable Particulate Network Operations and Quality Assurance Manual.
Environmental Monitoring Systems Laboratory, US Environmental Protection
Agency, Research Triangle Park, NC (March 1983).
13. Federal Register. Vol. 49, No. 55, p 10408 (March 20, 1984).
14. A. B. Williamson, R. S. Martin, and D. B. Harris, "Measurement of
condensible vapor contribution to PM-10 emissions," 78th APGA Annual
Meeting, Betroit, MI (June 1985).
15. R. G. Merrill, and D. B. Harris, "Field and laboratory evaluation of a
woodstove dilution sampling system," 80th APCA Annual Meeting, New York,
NY (June 1987).
16. J. N, Pitts, J. A. Sweetman, W. Harger, D, R, Fitz, P. Hanns-R. and A. M.
Winer, "Diurnal mutagenicity of airborne particulate organic matter
adjacent to a heavily travelled West Los Angeles Freeway," J. Air Poll.
Control Assoc. 35, 638-643 (1985).
17. R. K. Stevens, C. W. Lewis, T» G. Dzubay, R. B. Baumgardner, L. T. Cupitt,
V. R. Highsmith, J. Lewtas, L. D. Claxton, B. Zak, and L. Currie,
"Source apportionment of mutagenic activity of fine particles collected
in Raleigh, NC and Albuquerque, NM," 1987 EPA/APCA Symposium on Measure-
ment of Toxic and Related Air Pollutants, Research Triangle Park, NC
(May 1987).
18. R. B. Zveidinger, D. Dropkin, F, Stump, S. Tejada, and R. Drago, "Vola-
tile organic hydrocarbon and aldehyde composition in Raleigh, NC during
the 1985 woodsmoke study," EPA/APCA Symposium on the Measurement of
Toxic and Related Air Pollutants, Research Triangle Park, North Carolina,
(May 1987).
19. J. M. Daisey, J, D. Spengler, and P. Kaarakka, "A comparison of the
organic chemical composition of indoor aerosols during woodburning and
nonwoodburning periods," 4th International Conference on Indoor Air
Quality and Climate, Berlin, West Germany (August 1987).
20. B. N. Ames, J. McCann, and E. Yamasaki, "Method for detecting carcino-
gens and mutagens with the salmonella/mammalian-microsome mutagenicity
test." Mutation Research. 31, 347-364 (1975),
568
-------
21. D. M. Mar on, and B. N. Ames, "Revised methods for the salmonella
mutagenicity test," Mutation Research, 113, 173-215 (1983).
22. D. E. Lentzen, et. al., "IERL-RTP Procedures Manual: Level 1
Environmental Assessment (Second Edition)", EPA 600/7-78-201, NTIS #
PB 293795 (10/78), 26 pages, pp 140-142.
23. L. D. Johnson, R. E. Luce, and R. G. Merrill, "A spot test for poly-
cyclic aromatic hydrocarbons," US Environmental Protection Agency,
Industrial Engineering Research Laboratory, Research Triangle Perk, NC
27711.
24. J, A* Cooper, L. A. Currie, and G. A. Klouda, "Assessment of contempo-
rary carbon combustion sources to urban air particulate levels using
carbon-14 measurements," Environmental Science and Technology. IS,
1045-1050. (1981).
25. L. A. Currie, J, A. Noelces, and D. N. Breicer, Proc. Int. Conf.
Radiocarbon Dating, 9th, 1979, 158.
26. P. J. Groblicki, S, H, Cadle, C. C. Ang, and P. A. Mulawa, "Interlabora-
tory comparison of methods for the anaylsis of organic and elenemtal
carbon in atmospheric particulate matter." General Motors Research
Report No. 4054. General Motors Research Laboratories, Warren, Michigan,
46090 (1963).
27. R. W. Shaw, R. K. Stevens, J. Bowermaster, J. Tesch, and E. Tew,
Measurement of atmosphere nitrate and nitric acid; the denuder differ-
ence experiment, Atmos. Environ.. 16, 845-853 (1981).
28. F. 0. Stump, and D. L. Dropkin, "A gas chromatographic method for
the quantitative speciation of C2-C13 hydrocarbons in roadway vehicle
emissions, Analytical Chemistry. 2629-2634 (1985).
29. S. B. Tejada. "Evaluation of silica gel cartridges coated in aitu with
aciditified 2,4-Dinitrophenylhydrazine of sampling aldehydes and ketones
in air," International J. of Environ. Analytical Chem.. 26, 167 (1986).
30. T. S. Sfcopek, H. L. Liber, D. A. Kaden, and W. G. Tbilly, Relative
sensitivities of forward and reverse mutation assays in salmonella
typhimurium." Proc. Hatl. Acad. Sci. (USA), 75, 4465-4469 (1978),
31. T. R. Skopek, H. L. Liber, J. J. Krolewski, and W. G. Thilly, Quantita-
tive forward mutation assay in salmonella typhimurion using 8-
azaguanine resistance as a genetic marker. Proc, Natl. Acad* Sci. (USA)
75, 410-414 (1978),
32. R. L. Lagus, "Air leakage measurements by the tracer dilution method -
a review," American Society for Testing and Materials Symposium oo
Building Air Change Rate and Infiltration Measurements (March 1978)
Washington, DC, Technical Publication 719, p 36-49,
569
-------
n
200
180
160
E 140
0» 120
» 100
§ 80
2 60
» 40
ill 20
0
J
January
February
lLJudLl.il
lluii
ii
2 6 7 1012 15 17 2022 25 27 30 2 4 6 9 11 13 16 18 20 22
Figure 2. Albuquerque Fine Mass Loadings
SLOPE .0090
r
SLOPE .0018
INTERCEPT .0625
CQRR COEF .8123
!u
I
u
* « m
MUMMOn OU no/m1 fta* Mm
£ '
A
O
&o ©
a a
a
A
AO
OA
A
A$*°
A
& PD
0 i
0 o o1
A -
P o
o
.6°
6.
7 am
7 pm ¦ 7 am
Midnight
7 pm I 7 am
Midnight
February 15 February 16 February 17
Figure 5. Time plot of a woodsnioke' impacted weekend.
570
-------
Residence
Analysis
1
2
3
In
Out
Amblant
In
Out
Amblant
In
Out
Amblant
Fine Mat*, pg/m®
18.0
22,4
27.2
87.2
B.4
2.1
10.0
14.0
Coarse/Fine
Ratio
o.«
0.10
o.a
0J0
0.13
0.70
0.31
0.20
Flna K. fig/m3
008
0.12
0 22
0.29
0.31
BD
0.0B
0.04
Fine Pb. jtg/ros
0.04
0.01
0.22
0.31
0.20
BO
0.(6
0.04
CO ppm
0.6
2.2
2.1
2.0
0.*
0.1
0.0
Air Exchenge
Rat*
0.43
0.30
0.S7
TABLE IV. RESULTS OF INORGANIC ANALYSIS ON RESIDENTIAL SAMPLES
Analysis
Residence
1
2
3
In
Out
Ambient
In
Out
AmMant
In
Out
AmMant
Fin* Masa, w/m1
10.0
22.4
27J
17.2
08.4
2.8
1M
14.8
XAD-2
QMAV mg/sampla
4.0
2.1
1.3
BO
BO
BD
BO
BD
BD
TOO mg/Mmpla
21.1
M
0.3
KM
17J
14.0
M
3.0
1.0
Blosaaay ( + 801
1B.I
47.7
. 47.0
28.2
100.3
18.1
U
18.1
W.4
Flltars mutanta/m*
Total VOC itQ/m*
224
m
101
«24
418
311
204
82
Total
40
14
10
41
34.3
•4
18
1
Aldahydaa ^g/m1
BD-below detection
TABLE V. RESULTS OF ORGANIC ANALYSIS ON RESIDENTIAL SAMPLES
200
January
March
160
S
5
®
c
100
II
15 19
28 « 910 1616 20 38
Figure !• Raleigh Fine Mass Loadings
571
-------
ANALYSIS
SAMPLER
1ALKICB
ALIUOUIKQUI
HlMAgY 1ACKG1QUWP HtlHAKY ROADWAY
ANALYSIS
ursuNcs
KAB&
¦IOASSAY
RESIDENTIAL
SEMI-VOLATILES
lAD-l
Cy/C,
*03, RNO3
ORGANICS
CONTINUOUS
MP
Dichotoaoua
PHj 5 Hi-vol
PMl6 Hi-vol
A CFM with XAD-2
PM2.5 Hi-vol
PHjo Hi-vol
*2.5
Hi-vol
HodifUd
Diehotooou*
Dtnudar
VOC
Aldthyd*
C0,N0Xt03fS02
P «Cat,
MoCafOtology
1
1
YES
YES
YES
1
1
YIS
YES
YIS
YIS CO
YIS YIS
YIS YES
12
20,21
30,31
22»1*
22.23
24,25
16
27
SB
29
TABLE I. AMBIENT SAMPLERS
ANALYSIS
AKALYSIB UTOOOR OUTDOOR PRIMARY REfEREHCE
MASS
Dichotooou*
Dichotooou*
Dichotooou!
12
BIOASSAY
4 CFM
4 CFM
4 CFM
30,31
SEMI-VOLATILE
XAD-2
XAD-2
XAD-2
22,21
ORGAN1CS
VOC
Aldehyde
VOC
Aldchyda
VOC
Aldthyd*
28
29
CRITERIA
CO
CO
CO
AIR EXCHANGE
RATE
BAG
—
—
32
TABLE II.
RESIDENTIAL
SAMPLERS
Raleigh
Albuquerque
Primary
Baekground
Primary
Day
Night
Total
Day
Night
Sample Periods
31
44
82
80
83
Fin*
Mass fta/m*
K pg/nr
Pb itg/mr
18.4
0.06
0.06
42.7
0.24
0.13
18.3
0.07
0.08
27.2
0.10
0.30
87.1
0.18
0.38
Bloauay
EOM MO'nt'
- S9 rev/m1
+88 rav/m®
7.4
10.2
8.8
28.0
22.1
24,8
8.8
18.8
21.2
34.7
21.3
82.1
TABLE III. RESULTS OF ANALYSIS ON AMBIENT SAMPLES
572
-------
"THE SOURCE APPORTIONMENT OF CARBONACEOUS
COMBUSTION PRODUCTS BY MICRO-RADIOCARBON
MEASUREMENTS FOR THE INTEGRATED AIR CANCER
PROJECT (IACP)"
G.A. Klouda, L.A. Currle, A.E. Sheffield,
S.A. Wise, B.A. Benner
National Bureau of Standards,
Gaithersburg, MD 20899
R.K. Stevens and R.G. Merrill
U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711
Abstract
Atmospheric particle samples were collected during the winter of 1984-
1985 in Albuquerque, NM and Raleigh, NC by the EPA for the Integrated
Air Cancer Project (IACP). Selected chemical fractions were analyzed
for to apportion mobile (motor vehicles) and stationary
(residential wood combustion) sources. In addition, these results
were used to validate the EPA Single Tracer Regression Model (STRM),
also a technique for the source apportionment of aerosols.'
Preliminary ^C results for the Albuquerque residential site at night
showed 79% Contemporary Carbon (CC) compared to 95% CC for Raleigh at
night for the total carbon; 88% and 94% of the total-C was organic,
respectively. The Albuquerque traffic site during the day showed 1.4
to 3.9 times less CC compared to the daytime residential site for
total-C. The elemental carbon fraction in all cases showed a lower
percentage of contemporary carbon than the total carbon, which
indicates that this chemical fraction may be an excellent tracer of
mobile sources. These results are consistent with a daytime mobile
source at the traffic site and a nighttime Residential Wood Combustion
(RWC) source at the residential site. Also, the results from the EPA
STRM technique for this study were in good agreement with those
obtained by the 1/,C Direct Tracer (^C-DT) technique for source
apportionment of these aerosols.'
573
-------
Introduction
Concentrations of carbonaceous pollutants in urban atmospheres
have typically been measured over time and space to quantify the impact
of total emission sources at receptor (collection) sites. For
apportioning fossil and contemporary carbon source emissions, the
Direct Tracer (^C-DT) technique has been applied to bulk air particle
samples, which are frequently size segregated (<2.5 Application
of the "C-DT technique to important chemical classes extracted from
the bulk material, e.g. the polycyclic aromatic hydrocarbon (PAH)
fraction, has shown a potential for improving the process of
deconvoluting emission sources.3 Also, the origin of biologically
active (mutagenic or carcinogenic) compounds and/or chemical classes,
e.g. the PAH's, is important from a toxicological standpoint and is one
of the prime objectives of the Integrated Air Cancer Project (IACP).
Additional source information may be gained by comparing the
molecular "fingerprint" pattern of chemical fractions determined from
source samples with that obtained from the ambient samples. Some
compounds may be unique to the combustion source of interest.^1 These
chemical patterns coupled with ^C-BT would potentially discriminate
different combustion processes that utilize identical (in
abundance) fuels, e.g. coal versus auto emissions. This additional
discrimination in source apportionment is important for complex
airsheds involving many sources.
The Integrated Air Cancer Project, with a focus on the
Albuquerque, NM, and Raleigh, NC, airsheds (1985-86), has afforded an
excellent opportunity to utilize the 14C-DT technique on select
chemical classes for the apportionment of stationary (RVC) and mobile
(motor vehicle) sources. Advances in ^C measurements by accelerator
mass spectrometry (AMS) have improved sensitivities to ca. 50 pg-C
quantities^ and have, therefore, extended the capabilities of this
technique to include chemical selectivity and time resolution in
sampling for environmental pollutants. Total carbon (Ct) and
elemental carbon (Ce) from day and night time samples were Isolated for
WC measurements. Preliminary results are presented along with a
brief discussion of the chemical fractionation procedures for isolating
the fractions of interest.
Experimental Methods
lhC - Direct Tracer Model (l4C-DT)
The 14C measurement yields a quantitative measure of both fossil
and contemporary carbon sources which contribute to the chemical
fraction of interest for a given receptor site. For source
apportionment of these particles, knowledge of some critical variables
is required, such as the age of the fuels, usage patterns, time of
sample collection and meteorological patterns. Since one objective was
to evaluate the ^C-DT technique for a simple two-source situation, it
was important to design the sampling for source specificity. Once
evaluated, the ^C-DT technique would be used to validate the results
574
-------
of EPA's STRM. The two techniques would then be applied to samples
collected during the winter of 1986-1987 in Boise, Idaho.
The l^C-DT technique involves the measurement of the
ratio in the sample for AMS or the ^C/^C ratio for the low-level gas
proportional counting (11c) technique. The sample isotope ratio is
compared to the isotope ratio determined for modern carbon, 0.95 x NBS
SRM 4990B Oxalic Acid. These isotope ratios define the fraction of
modern carbon, fj( (Eq. 1). The fraction of contemporary carbon (fc) is '
calculated by adjusting f^j to correct for bomb * C, which is a result
of nuclear weapons testing in the atmosphere during the 1950's and
1960's (Eq. 2).
0.95 x 4990B (Eq. 1)
fc - f„/1.20 (Eq. 2)
fc (meas.) - «auto fCfiUt° + *RWC fCRWC 2.5 jim in
diameter. Two receptor sites were chosen in Albuquerque for maximum
impact of: 1) motor vehicles; San Mateo site and 2) RWC, Zuni Park
site. Samples were collected on a 12-hour schedule beginning at 7:00
am and 7:00 pm. Both day and night samples from Zuni Park, day only
samples from San Mateo, and nighttime samples from Raleigh were
chemically fractionated for measurements. Prior to chemical
fractionation, samples were stored at ca. 1°C.
Chemical Fractionation
Filter aliquots were taken for isolation of total carbon (Ct)
and elemental carbon (Ce). Total carbon was isolated by thermal
oxidation from: 1) 3-24 cm2 aliquots for AMS measurements and 2)
100-280 cor aliquots for lie; the filter area chosen was based on the
particle loading and the concentration of total carbon, The C02
produced was cryogenically collected, distilled and quantified by
manometry in a calibrated volume. Elemental carbon was isolated by
first removing organic carbon by wet oxidation (using HNO3) followed by
thermal oxidation of the insoluble residue and quantification of the
CO2 by manometry as described above. The wet oxidation of organic
matter was taken from the work of Schultz and modified by Sheffield et
al,? for treatment of air particle samples. Filter aliquots of 9 to 34
cm2 were treated to recover Ce for 14C measurements by AMS.
575
-------
Sample CO2 Preparation and Measurements
The procedures for chemical purification of CO2 and
measurement by 11c for fractions containing >5 mg-C are described by
Currie et al. (1983). A few Ct fractions and all Ce fractions
contained only microgram quantities and, therefore, required
transformation to graphitic carbon for measurement by AMS.5'^1^0
Results
Albuquerque, NM
Preliminary results based on -^C measurements of the total and
elemental carbon fractions are summarized in table 1. Nighttime
samples collected at the residential site in Albuquerque showed 79%
Contemporary Carbon (CC) for total carbon and about an equal mix of
fossil and contemporary carbon in the elemental carbon fraction. This
result reflects a predominant RWC source during the night as expected.
The daytime samples reflect a decrease in contemporary carbon, which
would be expected for typically warmer periods (less RWC) and an
increase in automotive activity.
The traffic site during the day was more strongly Influenced by
auto emissions as demonstrated by lower £q values. The elemental
carbon fraction may be more sensitive to variations in fossil source
contributions and, therefore, be an excellent tracer of mobile sources.
Raleigh, NC
Only nighttime Raleigh samples were measured. These results
showed a high contribution from RWC in both the total and elemental
carbon fractions, similar to that identified at the nighttime
residential site in Albuquerque, NM.
Conclusions
The ^C-DT technique applied to samples collected for high
source specificity has demonstrated its power to resolve auto and wood
smoke emissions in Albuquerque, NM and Raleigh, NC given knowledge of
likely sources, usage and meteorological patterns. *^C results showed
that the elemental carbon fraction in all cases contained more fossil
carbon and, therefore, may be a more useful tracer of mobile sources.
These results have been used to validate the Single Tracer Regression
Model (STRM) reported In this same proceedings by Stevens et al.
(1987).
Additional results for Ct and Ce are pending and are
expected to yield more information about source variations for the
study. We are currently isolating the PAH fraction from a few of the
Albuquerque samples which will give more specificity in establishing
the origin of this toxic chemical fraction produced from both auto and
RWC sources. The chemical composition of the FAH fraction coupled with
will reveal the predominant species and their origins. The
576
-------
application of the ^C-DT technique to important chemical fractions,
e.g. the PAH's, has only recently become possible through advances in
accelerator mass spectrometry measurements on 50 pg-C samples.5
Acknowledgment
The authors wish to thank D. J. Donahue, A. J. T. Jull, T. W.
Linick and L. J. Toolin for their participation in the measurements
by accelerator mass spectrometry.
References
1. R. K. Stevens, C. W. Lewis, T. G. Dzubay, R.E. Baumgardner, R. B.
Zweidinger, R. V. Highsmlth, L. T. Cupitt, J. Lewtas, L. D. Claxton, L.
A. Currie and B.Zak, "Source apportionment of mutagenic activity of
fine particles collected in Raleigh, NC and Albuquerque, NM," EPA/APCA
Symposium on Measurement of Toxic and Related Air Pollutants, Research
Triangle Park, NC, May 3-6, 1987.
2. L. A. Currie, G, A. Klouda and K. J. Voorhees, "Atmospheric carbon:
The importance of accelerator mass spectrometry,'1 Nucl. Instr. and
Meth.. 233 [B5], 2, Nov. 1984.
3. S. A. Wise, C, F. Allen, S. N. Chestler, H. S. Hertz, L. R. Hilpert,
W. E. May, R. E. Rebbert and C. R. Vogt, "Characterization of air
particulate material for polycyclic aromatic compounds," U.S. Dept. of
Commerce publication, NBSIR 82-2595 (Dec. 1962).
4. R. G. Merrill, "Progress toward identifying source specific
tracers," EPA/APCA Symposium on Measurement of Toxic and Related Air
Pollutants, Research Triangle Park, NC, May 3-6, 1987.
5. R. M. Verkouteren, L. A. Currie, G. A. Klouda, 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," in press, Nucl. Insr. and Meth.. (1987).
6. H. Schultz, "Studies in radiocarbon dating," PhD. Thesis,
Pennsylvania State University, Dec. 1962.
7. A. E. Sheffield, G. A. Klouda and L. A. Currie, manuscript in
preparation, 1987.
8. L. A. Currie, R. W. Gerlach, G, A. Klouda, F. C. Ruegg and G.B.
Tompkins, "Minature signals and miniture counters: accuracy assurance
via microprocessors and multiparameter control techniques"
Radiocarbon, 25, 2: 553 (1983).
9. A. J. T. Jull, D. J. Donahue, A. L. Hatheway, T. W, Linick and L. J.
Toolin, "Production of graphite targets by deposition from CO/H2 for
precision accelerator *®C measurements," Radiocarbon. 28, 2A: 191
(1986).
10. T. W. Linick, A. J. T. Jull, 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).
577
-------
Table I. Carbon concentrations and the fraction of Contemporary Carbon
in samples collected in Albuquerque, NM and Raleigh, NC.*
Albunueroue¦ NM
"Residential Site"
Night
Total (n-4)
Elemental (n—3)
Carbon fug/m3)**
34 to 71
4.6 ± 0,7
•fck-k
Zc ± SE'
0.79 ± 0.04
0.46 to 0.72
Day
Total (n—2)
Elemental (n-1)
11 to 24
4.8 + 0.7
0.66 ± 0.01
0.25 ± 0.02
"Traffic Site"
Day
Total (n-3) 14 to 24 0.17 to 0.47
Elemental (n-1) 6.1 ± 1.0 0.18 ±0.03
"Residential Site"
RflUlgh, uc
Pay ^
Total (n-4) 23 to 80 0.95+0.06
Elemental (n-1) 3.2 ± 0.2 0.68 ± 0.04
* Results are reported either as an average of "n" observations or a
range when results differ by greater than ca. 10%.
** Average carbon concentrations and standard errors (in for
total carbon were determined by thermal combustion of particles to COj
and detected by NDIR. Measurements were made in triplicate on "1 cm^
filter aliquots. Elemental carbon concentrations were determined by
combustion of the carbon residue, remaining from the wet oxidation
treatment, to CO2. The CO2 was quantified by manometrlc measurement in
a calibrated volume.
*** Is the fraction of Contemporary Carbon and the standard,error
(SE) is lcr-Poisson counting statistics. In the case where "n" is
greater than 1, the error represents the dispersion among the data
points. (Percent contemporary carbon - fc x 100.)
578
-------
VOLATILE ORGANIC HYDROCARBON AND ALDEHYDE COMPOSITION
IN RALEIGH, N.C. DURING THE 1985 WOODSMOKE STUDY
Roy Zwe1d1nger, David Dropkin, Fred Stump, and Silvestre Tejada
Atmospheric Sciences & Research Laboratory
Ronald Drago
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
Initial field tests for collection of volatile organic hydrocarbons
and aldehydes under EPA1s Integrated Air Cancer Project (IACP; were held
1n Raleigh, NC during the winter of 1985. Sampling was conducted during 12
hour (7:00-7:00) daytime or nighttime periods. Hydrocarbons were
collected by pressurizing 6 liter "Summa polished" stainless steel
cylinders. Aldehydes were collected as 2,4-dinitrophenylhydrazone (DNPH)
derivatives using implngers and cartridge collection methods. Sampling
locations Included the primary outdoor site located in a residential
neighborhood, a rural background site and Inside and outside three homes
operating wood stoves. Non-methane hydrocarbon levels at the primary site
ranged from 150 to 850 ppbc while total carbonyls ranged from 3 to 35 ppb
(v/v). Benzene levels ranged from 5 to 32 ppbc while formaldehyde levels
ranged from 2-14 ppb. Hydrocarbon levels and distributions 1n the homes
were similar inside and outside with a few noted exceptions. Carbonyl
levels, however, generally were much higher inside than outside. The
elevated carbonyl levels observed Inside the homes 1s likely due to
factors other than the presence of a wood stove.
579
-------
Introduction
The Integrated Air Cancer Project {I AC P) is a long term
interdisciplinary research program to develop scientific methods and
databases for identifying the major sources of carcinogenic chemicals
emitted into the air and/or arising from atmospheric transformation of
chemicals emitted into the air. Initial field studies were conducted
during the winter of 1985 in Raleigh, NC. Raleigh represented a
simplified air shed impacted mainly by residential wood combustion and
motor vehicle sources both of which can contribute substantially to the
mutagenic activity found in ambient air samples. A limited amount of
volatile organic hydrocarbon and aldehyde sampling was conducted during
February and March 1985 while methods and protocols were being developed.
Most samples were collected at the primary outdoor site located in a
residential neighborhood and both inside and outside three homes operating
wood stoves. A few samples were also collected at a rural background
site. All sampling was conducted during 12 hour {7:00-7:00) daytime or
nighttime periods.
Experimental Methods
Gas phase hydrocarbons were collected in 6 L "Summa" polished stainless
steel canisters using a sampling apparatus similar to that described by
Oliver. The initially evacuated canisters were filled by pumping the
sample through a flow controller at about 20 cc/min. resulting in a final
pressure of about 21 psig after 12 hours. The canister samples were
quantitatively analyzed by GC/FID using a modified Nutech model 320
cold-trap thermal-desorption system for sample concentration and
injection. Typically, a 1.5- to 3.0-L sample was injected. The C2's and
C3's were separated on a 12 foot by 1/8 inch stainless steel silica gel
column maintained at 30°C while the rest of the hydrocarbons were
separated on a Quadrex 100m by 0.5 mm soft glass capillary coated with
7.5% hydrophobic silica using a temperature program from -80 to +140°C.
Details of the hydrocarbon analytical procedures are available elsewhere.
Aldehydes were trapped as their 2,4-dinitrophenylhydrazine (DNPH)
derivatives and analyzed by high performance liquid chromatography (HPLC).
While a few aldehyde samples were initially collected using classical
impinger techniques (acidified solution of DNPH 1n acetonltrile) for
comparison, the majority of samples were collected using a cartridge
technique. Briefly, prepacked silica gel cartridges were coated with an
acidified solution of DNPH in acetonltrile (ACN) and excess solvent
removed by blowing carbonyl-free nitrogen through the cartridges. The
cartridges were used as direct probes and traps for sampling ambient air
when the temperature was above 10°C, and a heated glass probe and manifold
used when ambient temperatures were below 10°C. Sampling flow rates of
lt/min. were controlled using mass flow controllers pr needle valves and a
recording mass flow meter. The cartridges were eluted with 5 mL of ACN
for analysis. Details on the cartridge sampling and analysis procedures
have been reported.
Results
The majority of sampling took place at night (7 P.M.-7 A.M.) at the
Raleigh primary site. The primary sampling site (Quail Hollow Swim Club)
was located in a residential neighborhood about one mile from two major
roadways. Sampling periods encompassed a variety of meteorological
conditions which affected the impact of wood stoves on air quality.
Figure 1 shows the total NMHC levels in ppbc measured for several sampling
580
-------
periods and hydrocarbon distribution by compound class. Both are plotted
against the fine dichot particulate levels which varied from 19-129 yg/m .
The fine particle loadings are an indication of wood smoke impact, with
normal ambient levels in Raleigh being <25 ug/m3. (An atmospheric
Inversion leading to high wood smoke particulate levels does not of course
necessarily mean that the vapor phase was also mostly wood smoke related.
Mobile sources might also be expected to contribute substantially. The
current data set, however, 1s too small to perform apportionment
calculations on the vapor phase.) Total NMHC levels ranged from 150 to
840 ppbc but did not show any particular correlation with particulate
levels although both did exhibit maxima for the same sample period.
Paraffins are the most abundant class of hydrocarbons for all samples
ranging between 49 to 60% of the total NMHC. Aromatic hydrocarbons were
generally the second most abundant class ranging from 18 to 26% of total
NMHC. Olefins ranged from 7 to 24% of total NMHC. Unidentified compounds
typically averaged about 5% of the total NMHC. In general, the
distribution observed for Raleigh 1s similar to those observed in 29 other
urban areas across the country {sampling during summers of 1984 and 1985
between 6 A.M.-9 A.M.). Benzene levels for these same sample periods
ranged from 5 to 32 ppbc and averaged 3.1% of total NMHC.
Carbonyl concentrations observed at the primary site ranged from 3 to
35 ppb (v/v). Figure 2 shows the average distribution of carbonyls
observed at the primary site. Formaldehyde, acetaldehyde, acetone,
butyraldehyde, acrolein, proplonaldehyde and x-butyraldehyde make up 95%
of the total carbonyls. All the carbonyls in figure 2 which are preceded
by an "x" indicate the summation of one or more unknown carbonyl
derivatives which have retention times similar to known standards, e.g.,
one or more unknowns eluting near butyraldehyde between crotonaldehyde and
benzaldehyde have been summed and labeled as x-butyraldehyde. Actual
carbonyl levels observed on a ppb (v/v) basis were: formaldehyde 2 to 14;
acetaldehyde 1 to 8; acetone 2 to 4; and acrolein 0.1 to 2.8. All other
aldehydes generally were less than 1 ppb. Several aldehydes which we
normally monitor were not detected at the primary site, i.e.,
hexanaldehyde, 2,5- dimethylbenzaldehyde and the three tolualdehydes.
Only a few hydrocarbon and aldehyde samples were concurrently
obtained from the background site and primary site. The background site
was located about 20 miles from the primary site 1n a rural area near
Falls lake. All sampling periods at both sites corresponded to particulate
levels <25 ug/m3. Class distribution of the NMHC's were about equal, but
the total NMHC at the primary site averaged twice the background site.
Total carbonyl levels were similar at the both sites for two of three
sampling periods while the primary site level was twice the background
site for the third sampling period. Background formaldehyde levels ranged
from 0.9 to 1.4 ppb while the primary site values ranged from 1 to 2 ppb.
Three homes with wood stoves were selected for indoor
mlcroenvironment studies as part of the Raleigh effort. Concurrent
samples were taken Inside and outside each home to estimate exposures
which may result from an inside source as opposed to intrusion of wood
smoke from outside. All the homes in the study were occupied by non
smokers. Residence #1 had a top loading free standing stove while resi-
dences #2 and #3 both had fireplace inserts. The air exchange rates for
residences #1- #3, determined by SFg, release were 0.42, 0.39, and 0.57
hr-i respectively.
Table I lists some of the carbonyl and hydrocarbon data for the
residence study. All samples were taken during the nighttime sampling
period. Interferences were experienced with the chromatography of
residence #3 hydrocarbon samples which prevented determining accurate
•distributions. The NMHC class distributions were similar for residences
#1 and #2 except for the aromatlcs inside residence #1. This increase is
381
-------
essentially due to a high concentration of naphthalene. Residence #1 had
a noticeable smokey odor inside and the naphthalene levels may be related
to wood stove emissions. Formaldehyde, acetone and acetaldehyde were the
most prevalent carbonyls in all three homes except for residence #3 which
had notably higher levels of two unknowns, x-valeraldehyde and
x-dimethylbenzaldehyde. All three residences had much higher formaldehyde
levels inside than out while both residence #1 and #3 generally had higher
inside levels of all carbonyls relative to the concurrent samples from
outside. Carbonyl samples obtained inside residences #2 and #3 during the
daytime sample period (stoves not in active use) were similar to the
nighttime sample period. This would seem to indicate inside carbonyl
levels may not be effected by wood combustion as much as by other domestic
activities such as cooking and/or particular furnishings or construction
materials.
CONCLUSIONS:
The Raleigh IACP study relative to hydrocarbons and aldehydes
resulted in some general observations with the caveat that the data set is
1imi ted.
1. The distribution of NMHC at the Raleigh primary site was similar
to other U.S. cities.
2. Over 95% of the carbonyl levels measured were due to only 6
compounds of which formaldehyde, acetaldehyde and acetone were the most
prevalent.
3. Carbonyl levels definitely appear higher inside homes and may not
be related to wood combustion.
4. Elevated levels of naphthalene inside one home coincided with a
noted smokey odor.
References
1. J. Lewtas, L, Cupitt, "The Integrated air cancer project: Program
overview", 1987 EPA/APCA Symposium on Measurement of Toxic and Related Air
Pollutants, RTP, NC, May 1987.
2. R. Highsmlth, R. Baumgardner, B. Harris, T. Lumpkin, R. Drago, R.
Zweidinger, R. Merrill, "The Collection of wood combustion samples 1n
Raleigh, NC and Albuquerque, NM", 1987 EPA/APCA Symposium on Measurement
of Toxic and Related A1r Pollutants, RTP, NC, May 1987.
3. Daren K. Oliver, Joachim D. Pleil, William A. McClenny, "Sample
integrity of trace level volatile orgainc compounds in ambient air stored
in "Summa" polished canisters", Atmos. Environ., 1986, pp.1403-1411.
4. Fred D. Stump and David L.Dropkln, "A gas chromatographic method for
the quantitative speciatlon of C2-C13 hydrocarbons 1n roadway vehicle
emissions", Anal. Chem., 1985, pp. 2629-2634,
5. Silvestre B. Tejada, "Evaluation of silica gel cartridges coated in
situ with acidified 2,4-din1tophenylhydrazine for sampling aldehydes and
ketones in air", Intern. J. Environ. Anal. Chem., 26: 167 (1986).
6. William a. Lonneman, "Comparison of 0600-0900 AM hydrocarbon
compositions obtained from 29 cities", Procedlngs of the 1986 EPA/APCA
Symposium on Measurement of Toxic Air Pollutants, April 1986, EPA Report
No. 600/9-86-013, APCA Publication VIP-7.
582
-------
TABLE I. HYDROCARBON AND CARBONYL DATA FROM RALEIGH RESIDENTIAL STUDY
5S538SSSSSSSSSSSSS);SSSSSSSSSSSS53SSSSS388SSaS8B3SBS88SSS3S8Z8S&8SB8SSS8
RESIDENCE #1 RESIDENCE #2 RESIDENCE #3
IN OUT IN OUT IN OUT
% Paraffin
51.66
61.65
54.39
53.42
% Olefin
13.69
14.00
16.65
17.94
• ••
% Aromatic
26.81
14.05
19.84
21.36
---
Total NMHC (ppbc)
434.31
525.99
1081.14
911.28
_ «
Benzene (ppbc)
7.65
8.77
22.01
23.02
8.20
8.73
Naphthalene (ppbc)
30.05
1.57
3.03
2.88
0.36
1.54
Carbonyls (ppb, v/v)
Formaldehyde
13.07
2.86
14.53
7.27
21.98
2.18
Acetaldehyde
5.16
1.34
4.80
4.44
8.46
1.25
Acrolein
1,94
0.31
0.63
1.47
0.74
0.18
Acetone
7,95
1.70
4.72
4.32
6.99
1.81
Proplonaldehyde
0.67
0.22
0.54
0.67
0.91
0.21
Butyraldehyde
0.83
0.41
0.93
1.08
1.40
0.78
Benzaldehyde
0.13
0.05
0.11
0.27
0.41
0.12
x-Butyraldehyde
0.20
0.51
0.55
0.70
0.01
0.23
x-Valeraldehyde
0.37
0.26
0.33
0.40
3.97
0.19
x-D1methy1benzaldehyde
1.07
0.20
0.97
0.65
4.30
0.43
Total Carbonyls
58SSBSSSBS8SSBBC5&SSSSXS
32.5Q
SSISSSS3
8.21
I3SSS3S8
28.18
1S8S8SSB8
22.31
taasssssaxs
49.45
axassesxi
7.78
583
-------
^ 90
»
o
x BOH
a
x
70
ESS X PARAFFIN
EZ2 % OLEFIN
ESS % AROMATIC
¦¦ TOTAL PPBC NMHC
o 60-
19* 19 21 25* 32 43 88 129
FINE OICHOT, UG/CUBIC METER (* - daytime sample)
Figure 1. Total ppbc NMHC and class distribution vs.fine dichot particulate
concentrations observed at the RaTeigh primary site.
8
i
u.
o
K
FORMALDEHYDE
Ui
UJ
03
"T-
ALDEHYDES
Figure 2. Average distribution of carbonyls ot Raleigh primary site
584
-------
Progress Toward Identifying Source Specific Tracers
Raymond G. Merrill Jr.
Roy B. Zwe1dinger, and Randall Watts
United States Environmental Protection Agency
Research Triangle Park, North Carolina 2711
Susan Razor
Acurex Corporation
Research Triangle Park, North Carolina 27709
Identifying organic source specific tracers Is Important to any
Integrated approach to human exposure. Several efforts to find source
specific tracers have been conducted starting 1n earnest in the early
1970's. Emissions from wood heating appliances have been the most
difficult to trace. Using literature data on the concentration of
commonly analyzed compounds such as' those from the class known as
Polycycllc Aromatic Hydrocarbons (PAH) has met with little success
because too few of the members of the class are reported quantitatively
or because only one compound was used to establish the relationship
between the source and Its emissions. Two approaches are currently
being pursued 1n the Integrated A1r Cancer Program to Identify source
specific organic tracers. The use of commonly sought compounds, such as
PAH, 1s being reevaluated using mult1var1ent mathematical techniques.
There appears to be a high correlation between Individual PAH compounds
1n samples acquired directly from wood stoves. These correlations also
appear to occur 1n ambient air samples dominated by residential wood
combustion emissions. A second approach involves Identification of
organic compounds unique to the combustion source of interest. Several
compounds from wood combustion appear to be candidates for unique
tracers. Current progress toward identifying source specific tracers
and quantitative relationships between them will be discussed.
Introduction
As part of EPA's Integrated A1 r Cancer Program (IACP) and EPA's
Vlood Stove Emissions program, several approaches are being Investigated
to find source specific parameters and if possible source tracers from
wood stove emissions. Emissions from wood stoves have been difficult to
relate to human exposure due to the complexity of the source operation
characteristics. Two approaches are being followed to Identify source
tracers from wood stoves. One Involves the apparent association or
correlation between members of the class of compounds known as
Polycycllc Aromatic Hydrocarbons. The other Involves identification of
compounds unique to the wood combustion process* In either case, one of
the major goals of this aspect of IACP 1s to find a unique mathematical
relationship which will allow the contribution of wood stove emissions to
be evaluated 1n a complex air shed.
585
-------
Experimental Methods
Source samples were collected from various types of wood stoves. The
types Included catalytic and conventional noncatalytlc as well as high
technology noncatalytic. Several fuel types were also evaluated Including
pine (cured and uncured), oak (cured and uncured), and douglas fir. Multiple
tests were performed to evaluate the effect of burn rate on the emissions.
For the pine and oak fuel was added to the stove twice during each burn.
At the low burn rate each fuel charge was approximately equal to one third
of the firebox volume. At the high burn rate the fuel charge was increased
to two thirds of the firebox volume. (1) The douglas fir burn rate was
established by the proposed New Source Performance Standard method for
wood stove certification. (2) A single charge was placed Into the hot stove
and burned at one of 4 rates: one low, two medium and one high over a
range of BTll output rates specific to each stove.
Samples were collected from wood stove sources using either of two methods,
Modified Method 5 (MM5)<4) or Wood Stove Dilution Sampling (MSDSS)(4).
Ambient samples were collected using a 10 um H1-Vol sampler (5). Sample
preparation was performed by soxhlet extraction of filters (from all sampling
methods) and porous polymer (from MM5 and WSDSS)(6).
For samples taken directly from the wood stove flue gas by MM5 the combina-
tion of filter and sorbent extract was analyzed as a single sample without
the aid of a cleanup step. For WSOSS samples PAH were Isolated from extracts
using solvent elutlon through silica and analysis was conducted on the
filter and sorbent extract separately. Results for WSDSS are reported as
the combined amounts found 1n both train components.
Analysis for PAH was performed on source samples using gas chromatography/
mass spectrometry (GC/MS){7) and on ambient samples using high performance
liquid chromatography (HPIC) with fluorescence detection.(8,9) Calibration
of each technique was performed with analytical standards for 16 specific
PAH over a range of concentrations encompassing the sample concentration.
GC/MS was chosen for the source samples because the concentration of PAH
from typical samples was well within the range of sensitivity for this
technique. HPLC/fluorescence was used for the ambient samples to allow
quantitation of PAH compounds as much as a factor of 1000 less than the
source samples.
Results
Screening of wood stove emission samples generated a list of tentative
tracer compounds (not Including PAH) shown 1n Table 1. Evaluation of the
presence and concentration of these compounds 1n field collected source and
ambient samples 1s ongoing. Two classes, benzfurans and methoxy phenols,
seem to contain the best candidate tracer compounds.
Combined data from all source sample analyses are shown 1n Figures 1 and
2. Data from the ambient air samples collected In Alaska are shown In
Figure 3, Data for the ambient a1r samples collected 1n Raleigh are
shown 1n Figure 4,
Relations were determined between PAH 1n a pair wise manner using
linear regression statistics. Values for significance were determined
by the number of data points and the correlation coefficient of the
regression Tine. Values greater than .15 were significant above random;
586
-------
however, only correlations w!th correlation coefficients greater than or
equal to 0.8 were used to select Individual pairs of PAH presented here.
Conclusions
PAH Relationships
Combustion of wood, which contains I1gn1n, produces several compounds
which appear to be characteristic of woodstove emission. In addition
there appears to be a high association among several of the PAH In
samples acquired directly from wood stoves. These PAH compounds are
formed fn ratios that seem to be Independent of stove type, wood species
or burn rate. Using linear correlations between members of the set of
PAH provides a better measure of the relationships than simple ratios.
The same PAH ratios found 1n source samples also occurs 1n ambient air
samples dominated by residential wood combustion emissions.
The findings have several potential uses. If the specific tracer
compounds such as the methoxy benzenes prove to be unique tracers for
wood combustion emissions then the concentration of these compounds
can be used 1n classic source receptor/source apportionment modeling.
If the ratio of selected PAH are constant from wood stoves as they appear
to be, and If different ratios exist for different sources then
mathematical techniques are available to apportion the contribution of
wood stoves in a complex airshed sample. They can be used to predict the
concentration of several health related compounds 1n wood stove emissions
with only the data on one in the group. Risk estimates can be refined
by the ability to use the ratio of one PAH to predict the concentrations
of other potentially hazardous PAH compounds in the class. Demonstrated
control of one member of the PAH class can also be used to predict the
control for other members where the relation or ratio exists.
End Notes
References
{1) Leese, K.E., and McCrlllls, R.C., "Integrated A1r Cancer Project --
Source Measurement," 79th Annual Meeting, APCA, Minneapolis, MN, June
1986.
(2) Federal Register 40 CFR Part 60 Standards of Performance for New
Stationary Sources, Standards of Performance for New Stationary Sources,
Standards of Performance for New sources, Residential Wood Heaters;
Listing of Residential Wood Heaters for Development of New Source
Performance Standards; Proposed Rules, Government Printing Office,
Wednesday, February 18, 1987, Vol 52, No. 32. pp. 4994-5066.
(3) Schllckenrleder. L, W,, Adams, J. W., and Thrun, K. E.,
"Modified Method 5 Train and Source Assessment Sailing System
Operator's Manual", EPA 600/8-85-003, February 1985, (NTIS PB8S-169878),
(4) Merrill, R.G., and Harris D.B., "Field and Laboratory Evaluation of
a Woodstove Dilution Sampling System", 80th Annual Meeting, APCA, New
York, New York, June 1987.
587
-------
(5) H1ghsm1th, R., Baumgardner, R., Harris, B.t Lumpkin, T., Drago, R.,
Zwefdlnger, R., and Merrill, R., "The Collection of Wood Combustion
Samples In Raleigh, N.C,. and Albuquerque, N.M." 1n Proceedings of
EPA/APCA Symposium on Measurement of Toxic and Related A1r Pollutants,
Research Triangle Park, N.C,, May 1987.
(6) Lentzen, D.E., Wagoner, D.E., Estes, E.D., and Gutknecht, W.F.,
"IERL-RTP Procedures Manual: Level 1 Environmental Assessment
(Second Edition)", EPA-600/7-78-201, October 1978, (PB 80182066).
(7) McCrllUs R.C., and Merrill, R.G., "Emission Control Effectiveness of
a Wood Stove Catalyst and Emission Measurement Methods Comparison", Presented
at the 1986 National Convention of the A1r Pollution Control Association 1n
Detroit, MI.
(8) Guidelines Establishing Test Procedures for the Analysis of Pollutants
Under the Clean Water Act, Federal Register Vol. 49, No. 209, Friday,
October 26, 1984, pp. 43344-43362.
(9) May, W.E., and Wise, S.A., "Liquid Chromatographic Determination of
Polycycllc Aromatic Hydrocarbons in Air Particulate Extracts" Analytical
Chemistry, 1984, pp. 225-232.
Table 1
Tentative Identification of
Major Compounds/Compound Classes Found 1n Wood Stove Emissions
ssssssBsssBaBStiJiiBssssssasssaissffatfffxffsffftvKffffvMffsirifivcfffviffwiff
Furans Ketones Alcohols
methylfuran
furanyl-
phenol
benzofuran
ethanone
cresol
methylbenzofuran
butanedlone
alkyl-
furfural
hexanedlone
phenols
d1hydro-dimethyl
propanone
ethyl-
furan
methyl furanone
benzenedlol
furanone
ethyl furanone
methoxypropenyl -
bifuran
dimethylbenzo-
phenol
methylfuran-
furanone
hydroxymetho*y-
carboxaldehyde
acetophenone
phenylacetic acid
d1benzofuran
alkyl
methylbenzenedlol
naphthofuran
acetophenones
biphenylol
methyld1benzofuran
hydroxy
nonanedlol
acetophenones
methyoxypropyl-
pentadecanol
alkyl cyclo-
dodecanediol
penteneone
pyranone
hydroxymethyl-
pyranone
fluroenone
phenanthrenedl one
UBS:
588
-------
4-
n
3.8-
¦
3-
e
*4
2.8-
s
*
2-
b
%
1.8-
i
1-
s
0.8-
a.
0
¦ PYRENE
~ K^O^FLUORANTHENE
a 8ENZ0 (WPYRENE
1»
t{:
—i* ¦*
I
A
+
A
-r
3
Fluoranthana Concentration in mg/n?
Figure 1. Fluoranthene vs selected PAH
1000
900
BOO
700
800
BOO-I
400
300
800
100
0
¦
~
~
KNZ (A) ANTHRACENE
CMMEM
BQCZO ® FLUORANTHENE
*
~
~ •
;U*
~
~ s
800
- r—
400
—i—¦
eoo
Banzo(a) pyrana concentration in ig/n3
Figure 2. Benzo (a) pyrene vs selected PAH
589
-------
110
W
& ^
S 70
3 60
BO
30
a. 20
10-
0
¦ BENZ (A) ANTHRACENE
1- CHRY8ENE
~ BENZO (B)FLUORANTHENE
*
*
¦
¦
t
:y ii
Ki *
*m
m
c
* ~
~ ~
10
40
20 ao
Benzo (a)pyrene concentration In ng/tf
Figure 3. Benzo (a)pyrene vs selected PAH
Alaska night data
48-i
40-
">
38-
5
30-
S
28-
1
20-
|
18-
5
10-
a.
8-
0-
¦ BENZ (A)ANTHRACENE
4 CHRYSENE
~ BENZO (B)FLUORANTHENE
*
~
¦
*
i i i i"» i ¦ 'i1 "i i r
2 4 6 8 10 12 14
y
Bonzo (•) pyrene concentration in ng/i
Figure 4. Benzo (a)pyrene vs selected PAH
Raleigh night data
590
-------
THE MUTAGENICITY OF AMBIENT AND SOURCE SAMPLES FROM
WOODSMOKE-IMPACTED AIR SHEDS
Larry D. Claxton1, Sarah H. Warren2, Virginia S. Houk1, Joellen LewtaB1
1Genetic Bioassay Branch
Health Effects Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina
Environmental Health Research and Testing
Research Triangle Park, North Carolina
Objectives of the U. 8. Environmental Protection Agency's Integrated
Air Cancer Project include the identification of major sources that emit
genotoxioants into the air and the identification of these genotoxicants.
As part of the project conducted in Raleigh, NC and Albuquerque, NM, ambient
and source samples from woodsmoke-impacted air sheds were collected, ex-
tracted, and bioassayed for mutagenicity* The assay used was the Salmonella
typhimurlum plate incorporation assay using strain TA98 both with and with-
out exogenous metabolic activation (S9) • In contrast to roadway and back-
ground samples, woodstove source and woodsmoke-impacted ambient samples
that were mutagenic increased in mutagenic response when S9 was present.
The mutagenicity of the laboratory woodstove source samples was very compa-
rable to woodstove source samples collected directly from the homes within
the study. In general, woodsmoke-impacted ambient samples showed more
mutagenicity per cubic meter air when sampled in the evening than when
sampled in the morning hours.
591
-------
Introduction
Air pollutants in the form of particles, vapors, and gases can exist in
any urban environment. It has been recognized for a number of years that
such pollutants, which are often the end products of combustion processes,
are of health concern1'2'3. In the 1960s, it was shown that organic extracts
of urban particulate matter caused skin cancer when painted on mice3'^. The
major anthropogenic sources of this bioactive organic matter in urban areas
appears to be heating sources, power generation, transportation, waste
incineration, and industrial processes5'6'?. Since prokaryotic bioassays
require only a fraction of the material necessary for mouse skin carcinogen-
icity studies, these systems (e.g., the Ames Salmonella mutagenicity assay)
have been used extensively in recent years in the evaluation of airborne
particulate organic matter, especially for the evaluation of potential
carcinogenicity8. The Integrated Air Cancer Project flftCP)9 has taken
advantage of the correlation between the bacterial mutagenicity and animal
carcinogenicity of combustion organics1" by using Salmonella typhinturium
bioassays for mutagenicity in order to compare the genotoxicity of wood-
smoke -impacted airborne particulate samples. This study has compared the
mutagenicity of laboratory source samples, residential source samples, and
ambient air samples.
Experimental Methods
The samples bioassayed were those collected and processed as part of
the IACP studies in Raleigh, North Carolina and Albuquerque, New Mexico.
The collection and preparation of samples is described more fully in a
companion publication1^. The types of ambient air (AA) and source samples
(SS) collected and their designations are as follows* 1) laboratory wood-
stove source samples {laboratory SS), 2) residential (home) woodstove
source samples (residential SS), 3) woodatove-impacted ambient air samples
used for source apportionment studies (woodsmoke AA), 4) automotive emis-
sion-impacted roadway samples (roadway AA), and S) ambient air samples from
a site not highly impacted by woodsmoke (background AA). Samples to be
bioassayed were collected on Teflon-coated glass fiber filters, extracted
with dichloromethane, and solvent exchanged into diraethylsulfoxide. Samples
were bioassayed using the Salmonella typhimurium plate incorporation bioas-
say for mutagenicity as described by Ames, et al., (1975)^2 and Claxton, et
al., (1987)1^. Samples that were to be directly compared were assayed si-
multaneously in the same experiment. Each sample was tested at a minimum
of 5 doses with duplicate plates per dose. Each experiment was replicated.
A strain study demonstrated that S. typhimurium TA98 was the most responsive
strain to each sample type; therefore, results are reported for this strain.
All samples were tested with and without an exogenous activation system
which consisted of 9000-g CD-1 rat liver homogenate {89) as described by
Ames, et al^., (1975) 12. All data was analyzed visually and by using two
statistical packages designed for bacterial mutagenicity data. The two
systems used were those of Stead, et al.,^ and Bernstein, et al.Since
most of the data was linear at the lower doses, all slope values for com-
parative purposes were calculated using the method of Bernstein, et al^, ^
and are reported as either revertants per Mg of organic extract or rever-
tants per m3 air sampled.
Results
The woodsmoke AA samples were those outdoor ambient air samples col-
lected at a woodsmoke-impacted site in each city, figure 1 summarizes the
results for the Raleigh daytime and nighttime woodsmoke AA samples tested
with S9. In general, higher mutagenic activity levels per cubic meter of
592
-------
air were seen for the nighttime (7P.M.-7A.M.) samples than for the daytime
samples. The range of mutagenic activity seen for weekend samples (Figure
1) was very similar to the range seen for weekday samples. In Figure 2,
the Raleigh woodsmoke AA samples are compared to the Albuquerque woodBmoke
AA samples and to the Raleigh roadway AA and background AA site samples.
For Raleigh, the activity seen with and without S9 was very comparable.
For Albuquerque, however, the addition of S9 provided nearly a doubling of
the mutagenic activity level. This may indicate that the Raleigh sample
was impacted by sources of directacting mutagens (e.g., automobiles) to a
greater extent than the Albuquerque samples. Alternatively, it may indicate,
that atmospheric conditions in Raleigh produced a different spectrum of
atmospheric transformation products. Whether examining mutagenicity on
a yg of extractable organic basis or on a per cubic meter of air basis,
the Raleigh woodsmoke AA samples were clearly more mutagenic than the back-
ground and roadway site samples* Source samples were collected directly
from the woodstove stack outlets of 3 homes in Raleigh and from a woodstove
tested under laboratory conditions using the woodstove source dilution
sampler. The range of activity, on a revertant per Ug organic material
basis, was very similar for the home and laboratory source samples (Figure
3). Again, it is quite apparent that the mutagenic response is enhanced
when S9 is added to woodstove organic emissions.
Conclusions
These studies indicate that the particle-bound organic emissions from
woodstoves are mutagenic in a bacterial screening assay. The mutagenicity
of theBe organics is enhanced by the addition of a microsome-containing
exogenous activation system. In general, woodsmoke-impacted ambient air
sampled at night was more mutagenic (revertant per m3) than air sampled
in the daylight hours. These airsheds, in turn, were more mutagenic than
the roadway AA and background AA airsheds. The mutagenicity of ambient air
in Raleigh and Albuquerque during woodburning periods were similar* however,
the Raleigh data demonstrated lesB indirect (S9-requiring) mutagenic activ-
ity. The study also showed that laboratory and home source samples, simul-
taneously tested, gave similar results.
Disclaimer
The research described in this paper has been reviewed by the Health
Effects Research Laboratory, U.S. Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency nor does mention
of trade names or commercial products constitute endorsement or recommenda-
tion for use.
References
1. National Research Council, Airborne Particles, University Park Press,
Baltimore, 1978.
2. S. I. Lamb, C. Petrowski, I, R. Kaplan, B. R. T. Simoneit, "Organic
compounds in urban atmospheres) A review of distribution, collection,
and analysis," J. Air Pollut. Control Assoc. 30t1098-1115 (1980).
3. E. L. Wynder, D. Hoffmann, "Some laboratory and epidemiological as-
pects of air pollution carcinogenesis," J. Air Pollut. Control Assoc.
15:155159 (1965).
593
-------
4. W. C. Heuper, P. Kotin, E. C. Tabor, W. W. Payne, H. Falk, E. Sawicki,
"Carcinogenic bioassays on air pollutants," Arch. Pathol. 74:89-116
( 1962).
5. J. P. Butler, T. J. Kneip, F. Mukai, J. M. Daisy, "Interurban varia-
tions in the mutagenic activity of the ambient aerosol and their
relations to fuel use patterns," In: M. D. Waters, S. S. Sandhu, J.
Lewtas, L. Claxton, G. Strauss, S. Nesnow, Short-Term Bioassays in
the Analysis of Complex Environmental Mixtures, IV, Plenum Press, New
York, 1984, pp. 233-246.
6. J. M. Daisy, T. J. Kneip, "Atmospheric particulate organic matter:
Multivariate models for identifying sources and estimating their
contributions to the ambient aerosol," In: £.S. Macias, P. H. Hopke,
Atmospheric Aerosol: Source/Air Quality Relationships, ACS Symposium
Series No. 167, American Chemical Society, Washington, 1981, pp. 197-
221.
7. L. D. Claxton, J. L. Huiaingh, "Comparative mutagenic activity of
organics from combustion sources," In: C. L. Sanders, F. T. Cross,
G. E. Dagle, J. A. Mahaffey, Pulmonary Toxicology of Respirable Parti-
cles, U. S. Government Printing Office, Washington, 1980, pp. 453-465.
8. L. D. Claxton, "Review of fractionation and bioassay characterization
techniques for the evaluation of organics associated with ambient air
particles," In: R. R. Tice, D. L. Costa, K. M. Schaich, Genotoxic
Effects of Airborne Agents, Plenum Press, New York, 1982, pp. 19-34.
9. J. Lewtas, L. Cupitt, "The integrated air cancer projectt Program
overview," 1987 EPA/APCA Symposium on Measurement of Toxic and Relat-
ed Air Pollutants, Research Triangle Park, 1987. [This volume].
10. J. Lewtas, "Comparative potency of complex mixtures: Use of short-term
genetic bioassays in cancer risk assessment," In: M. D. Waters, S. S.
Sandhu, J. Lewtas, L. Claxton, G. Strauss, S. Nesnow, Short-Term Blo-
assaya in the Analysis of Complex Environmental Mixtures IV, Plenum
Press, New York, pp. 363-375.
11. R. Highsmith, R. Baumgardner, B. Harris, T. Lumpkin, R. Drago, R.
Zweidinger, R. Merrill, "The collection of wood combustion samples in
Raleigh, NC and Albuquerque, NM," 1987 EPA/APCA Smposium on Measure-
went of Toxic and Related Air Pollutants, Research Triangle Park,
1987. [This volume]
12. B.N. Ames, J. McCann, E. Yamasaki, "Methods for detecting carcinogens
and mutagens with the Salmonella/mammalian microsome mutagenicity
test." Mutation Res. 31:347-364 (1975).
13. L.D. calxton, J. Allen, A. Auletta, K. Mortelmana, E. Nestmann, E.
Zeiger, "Guide for the Salmonella typhlmurlum/mammallan microsome
tests for bacterial mutagenicity," Mutation Res., in press (1987).
14. A. Stead, V. Hasselblad, J. Creason, L. Claxton, "Modeling the Ames
test," Mutation Res. 85t13-27 (1981).
15. L. Bernstein, J. Kaldor, J. McCann, M.C. Pike, "An empirical approach
to the statistical analysis of mutagenesis data from the Salmonella
test," Mutation Res. 97:267-281 (1982).
594
-------
FIGURE 1. RALEIGH SOURCE APPORTIONMENT,
DAYTIME (AM) AND NIGHTTIME (PM) SAMPLES
100
TA98 PLATE INCORPORATION (+S9). AM SAMPLES
90
BO
70-
60-
50-
40-
30-
20-
10
Y//A Weekend
Hi Weekday
100
1/31 2/03 a/08 2/00 2/0# 2/10 2/13 2/14 2/18 2/18 2/17 2/20 a/ai 2/27 3/01 3/02
SAMPLE DATE
TA98 PLATE INCORPORATION (+S9), PM SAMPLES
90-
bo -
70 -
60-
50-
40-
20-
10-
R
V\
~
/
~
/
/
/
/
/
~
~
/
/
/
/
/
/
n
/
/
/
/
/
/
/
/
/
/
s
/
/
/
/
/
/
/
/
/
/
/
/
<>
/
/
/
/
*
/
/
/
?
$
~
/
/
~
/
/
~
/
r-
~
<
F77] Weekend
Weekday
SAMPLE DATE
595
-------
FIGURE 2. SALMONELLA AEVERTANTS PER CUBIC METER AIR
FOR SOURCE APPORTIONMENT, ROADWAY,
AND BACKGROUND SAMPLES
S. typhimurlum TA98 Plate Test
SOURCE APPORTIONMENT SAMPLES
38.0
19. B
Albuquerque -88
ROADWAY S MAINSITE SAMPLES
Roadway +89 6
Background +89 SSSSl 0.9
Roadway -89 X//A
Background -89 ^\S 4.7
FIGURE 3. COMPARISON OF HOME SOURCE SAMPLES AND
LABORATORY SOURCE SAMPLE MUTAGENICITY
Hone Source Samples
hom h|HH
+39
hom
hom v|\AAAAAJo.«
Laboratory Source Samples +S9
Uncurad Oak. 4/88 V//////////////////ZZA 0.48
Uncurad Oak, 6/06 KWVWWWWWWV^ 0.4
Uncured Pine. 5/33 Y / / /~X 0. IB
0.62
Home Source Samples -S9
hom hBIo.ii
HOM FE5gOa0.0B
hom vQo.04
Laboratory Source Samplee -39
Uncurad Oak, 4/20 r/////7777) 0.21
uncurad oak, 6/06^0.04
tihcurad Pine, 0/83 0
S. typhimurlum TA98
Plate test
596
-------
IACP EMISSIONS: TRANSFORMATIONS AND FATE
L. T. Cupitt and L. D. Claxton
U, S. Environmental Protection Agency
Research Triangle Park, NC 27711
P. S. Shepson and T. E. Kleindienst
Northrop Services, Inc.
Research Triangle Park, NC 27709
Diluted emissions from wood stoves and automobiles were Irradiated 1n a
Teflon smog chamber to simulate their photochemical transformation 1n the
atmosphere. Throughout the experiments, the chemical composition and
physical properties of the gaseous and aerosol-bound complex mixtures were
monitored. The mutagenicity of the gas-phase components and of the aerosol-
bound chemicals were measured both before and after irradiation. Transfor-
mation of the dilute wood smoke occurred readily under all conditions
tested: with and without added NO, with artificial illumination and with
natural sunlight, at outdoor temperatures or at room temperature. As the
photochemical reactions progressed, the volume of the aerosol-bound organics
was seen to Increase, suggesting that transformation products were condens-
ing to form additional aerosol-phase materials.
The gas-phase reactants and products were tested for mutagenic activity
by exposlnq Salmonella typhimurlum, strains TA 98 and TA 100, to the
filtered effluent. Filter samples of the starting materials and the
transformed products were collected, extracted, and tested for mutagenic
activity 1n the same bacterial strains by using a standard plate incorpora-
tion test. The transformations caused a dramatic change 1n the mutagenic
activity of the gas-phase components. Comparisons of the mutagenic activity
between the gas-phase and aerosol-bound chemicals have been estimated. When
the mutagenicity 1s expressed In units of revertants per cubic meter of air,
the gas-phase reaction products are found to be the most mutagenic.
597
-------
Introduction
The Integrated Air Cancer Project (IACP) has focused on emissions from
residential wood combustion (RWC) and from automobiles. As a part of the
IACP, simulation experiments were conducted to characterize the atmospheric
transformations of these complex mixtures. This research effort focused on
three main questions: (1) Are transformations likely to occur? (2) If
transformations do occur, what changes in chemistry and mutagenicity are
induced? and (3) What 1s the relationship between gaseous and aerosol-bound
mutagens?
Experimental Methods
A schematic diagram of the 22.7 m3 reaction chamber used for most of
these experiments is shown 1n Figure 1. The chamber 1s housed in a truck
trailer and consists of a 7.5 m long cylinder of Teflon film suspended
between two aluminum end plates, each of which is coated with fluorocarbon
paint. Irradiation is accomplished by means of blacklight and sunlamp
fluorescent bulbs which surround the chamber. Oak logs, from the Research
Triangle Park area of North Carolina, were burned 1n a commercially avail-
able wood stove. A portion of the wood smoke was introduced into a dilution
tunnel and mixed with ambient air to cool the mixture. The diluted emis-
sions were then used to fill the chamber to the desired concentrati on.1 A
few, preliminary experiments have recently been conducted using automobile
emissions. The experimental set up is identical to Figure 1, except that
the wood stove was replaced by a 1980 catalyst-equipped Toyota Corolla
operating at high idle conditions and burning a "super unleaded" grade of
gasoline. For a few wood smoke irradiations, a 9.0 m3 outdoor Teflon smog
chamber was used. The outdoor chamber was at ambient temperatures and used
only natural sunlight.
Four chambers were used for exposure of Salmonella typhimurium, strains
TA 98 and TA 100, to the various gaseous mixtures. The chambers are 190-L,
rectangular, Teflon-coated containers capable of holding more than bO test
plates. "Survivor" plates were routinely Included for detection of toxicity
effects. The test mixtures were flushed through the bloassay exposure
chambers at 14 L mln"1, and exposures of the bacteria to the gaseous
mutagens was accomplished simply by uncovering the glass Petri dishes
containing the bacteria and permitting the mutagenic materials to dissolve
Into the plates. If one assumes that the "dose" is proportional to the
exposure period, a type of dose-response curve can be generated conveniently
by allowing various groups of plates to remain uncovered for differing
periods of time. The details of the mutagenicity testing have been de-
scribed elsewhere.1*3 Particulate samples from each of the test streams
were collected on Teflon-coated glass fiber filters identical to those used
in the ambient IACP field study. Mutagenicity testing of the particulate
extracts was accomplished using the Ames standard plate incorporation test.*
Four distinctive air streams were tested: the clean air used to replenish
the chamber as samples were withdrawn, the ambient air used 1n the dilution
tunnel, the reactants (i.e., the diluted wood smoke or auto exhaust prior to
irradiation), and the chamber effluent (i.e., the diluted emissions after
irradiation).
598
-------
In several experiments, two bloassay exposure chambers were used in
series to test the effluent from the reaction chamber. Not all of the
gaseous mutagens entering the first bloassay exposure chamber are removed by
the test plates. Data from the second chamber permits a better estimate of
the total vapor-phase mutagenicity 1n the test air stream. The calculated
gaseous mutagenicity may still be a lower limit estimate, however, since
non-polar mutagens are not likely to be efficiently removed, even by two
chambers In series.
The experiments were conducted by filling the reaction chamber with the
wood smoke or automotive emissions to the desired concentration ("15 ppm
Carbon), adjusting the N0X concentration, 1f necessary, and then Irradiating
the mixture until an ozone maximum was reached. The lights were then turned
off, and the effluent bloassay exposures were begun. For most of the wood
smoke Irradiations, additional NO was added to the system to increase the
extent of reaction and to bring the hydrocarbon to N0X ratio more 1n line
with those measured 1n urban and suburban areas.5 Additional NO was not
needed for the experiments Involving automotive exhaust. Irradiation of the
outdoor chamber was controlled by removal and re-installation of an opaque
cover. The wood smoke Irradiations usually took from one to two hours to
reach the ozone maximum, while the automotive exhaust experiment was
irradiated for about 5 hours. The bloassay exposures typically lasted up to
10 hours. Throughout the bloassay exposure period, the sample withdrawn
from the reaction chamber had to be continuously replaced with clean air.
This meant that the mutagens 1n the reaction chamber became more and more
dilute with time ("4.5% per hour). An "effective" exposure time was calcu-
lated for the effluent bloassay chamber to account for the effects of
dilution.
Results
Changes 1n the chemical composition and mutagenic potency were readily
observed for all Irradiation conditions. Table I lists the initial and
final concentrations of a variety of chemicals observed during three of the
wood smoke Irradiations. Experiment A was conducted with dilute wood smoke
alone, while experiments B and C contained additional N0X. The addition of
N0X to the dilute wood smoke irradiations caused the system to react more
completely and to produce even more mutagenic products than did Experiment
Similar experiments conducted 1ti the outdoor chamber gave essentially
Identical results for both chemistry and mutagenicity, despite differences
In the light intensity and distribution and the somewhat lower (by 9x C)
temperatures. Irradiation of the wood smoke mixture caused the aerosol
volume distribution to Increase, suggesting that transformation products may
be condensing out on the aerosols. The mutagenicity of both the gas-phase
and the aerosol components were measured before and after irradiation. When
two bloassay exposure chambers were used 1n series to measure the gas-phase
mutagenicity, the response 1n the second chamber was about 30* of that 1n
first chamber. This Implies that the bloassay exposure chamber 1s
around 70* efficient at removing the mutagens.
Figure 2 shows the comparison of the mutagenicity associated with the
gas-phase and the aerosol-bound organlcs, both before and after irradiation.
The mutagenicity 1n Figure 2 1s reported 1n revertants m"3 and 1s shown for
direct acting mutagens (I.e., without metabolic activation) 1n two bacterial
599
-------
strains (TA 98 and TA 100). Prior to Irradiation, the bulk of the mutagen-
icity was found in the particulate phase. After irradiation, however, the
gaseous transformation products contribute significantly (80 to 99%) to the
total mutagenic burden in the air. It should also be noted that the
mutagenic products proved to be very stable in the reaction chambers. As
described in the previous section, dose response curves were derived from
exposures which lasted as long as 10 hours after the lights were turned off.
The dose response curves remained linear over this time period, implying
that the mutagenic vapor-phase products are long lived.
Insertion of an XAD-2 trap into the line between the irradiation
chamber and the bioassay exposure chamber caused the mutagenic response to
decrease by more than 80^. Unfortunately, only about 10% of the removed
mutagenicity was recovered in the XAD-2 extract. Hany of the vapor-phase
mutagens removed by XAD-2 may either be unstable on the adsorbent, or may be
lost during the extraction and concentration procedures.
A series of preliminary experiments were also conducted using diluted
exhaust from an idling automobile to fill the simulation chamber. Once
again, the gas-phase transformation products dominated the total mutagenic
burden after irradiation.
Conclusions
Irradiations of complex mixtures involving both wood smoke and auto-
mobile exhaust have demonstrated that chemical reactions are probable and
that the gas-phase products which result from these chemical reactions can
constitute the major portion of the total atmospheric mutagenic burden.
Chemical and mutagenic changes were observed for all tested conditions {with
and without added NQX; artificial or natural illumination; controlled indoor
environment or cooler outdoor temperatures), but the mutagenic response
seems to increase with increasing chemical reaction. The mutagenic gas-
phase products have been shown to be quite stable in the simulation cham-
bers, with lifetimes considerably longer than many residential exchange
rates.6 These results suggest that transformation of the gas-phase and
aerosol components of complex mixtures may contribute significantly to the
total burden of mutagens or carcinogens in the environment and should be
considered in assessing risk.
600
-------
REFERENCES
1. T. E. Kleindienst, P. B. Shepson, E. 0. Edney, L. 0. Claxton, L. T.
Cupitt, "Wood smoke: measurement of the mutagenic activities of Its
gas- and particulate-phase photooxldatlori products," Environ. Sci.
Techno!. 20: 493 (1986).
2. L. D. Claxton, "Assessment of bacterial mutagenicity methods for
volatile and semi volatile compounds and mixtures," Environ. Int. 11:
375 (1985).
3. P. B. Shepson, T. E. Kleindienst, E. 0. Edney, G. R. Namie, J. H.
Plttman, L. T. Cupitt, L. D. Claxton, "The mutagenic activity of
irradiated toluene/N0x/H20/a1r mixtures," Environ. Sci, Technol. 19:
249 (1986).
4. B. N, Ames, J. McCann, E. Yamasaki, "Methods for detecting carcinogens
and mutagens with the Salmonella/tnammalian-microsome mutagenicity
test," Mutat. Res. 31: 347 (1975).
5. K. Baugues, "A review of NMOC, NOx and NM0C/N0x ratios measured 1n 1984
and 1985," EPA-450/4-86-015, U.S. EPA, Research Triangle Park, NC,
1986.
6. R. Zwe1d1nger, S, Stump, 0. Dropkln, R. Drago, S. Tejada, "Volatile
organic hydrocarbon and aldehyde composition 1n Raleigh, NC, during the
1985 woodsmoke study," This volume, (1987)^
601
-------
Table I. IACP Transformation Study: Gas Phase Concentrations of
Chemicals During Wood Smoke Irradiations (ppb, v/v)
Experiment A
Experiment B
Experiment C
Species
(No NOx Added)
(NOx Added)
(NOx Added)
Initial
Fi nal
Initial
Final
Initial
Final
Nitric Oxide
75
0
454
0
461
0
N0X
135
64
657
252
576
259
Ozone
0
79
0
467
0
696
CO (ppm)
33.5
32.0
38.0
33.4
38.7
35.5
Ethylene
702
652
537
313
847
439
Benzene
68
62
62
50
102
68
Toluene
27
16
62
15
24
10
Formaldehyde
325
381
269
365
229
383
Acetaldehyde
140
106
88
109
57
75
PAN
0
52
0
174
0
232
HC (ppm-C)
20.6
19.1
16.4
13.2
17.2
15.0
HC/NOx
153
25
30
Figure Captions
Figure 1. Schematic diagram of the experimental apparatus used in the IACP
wood smoke study.
Figure 2. Graphical comparison of the mutagenicity of the gas- and particu-
late-phase components of wood smoke, before and after irradiation.
(The mutagenicity was measured without metabolic activation and 1s
expressed in units of revertants per cubic meter.)
602
-------
Experimental Schematic of the Wood Stove,
Reaction Chamber, and Exposure Apparatus
Dilution
Air In
0-700 cfm
Turbine
Exhaust
Dilution Tunnel
Particulate
Fitter
Reactants
Biochamber
Metal Bellows
_Q^Pumpi
Particulate
Ambient Filter
Air In I
Ambient Air
Biochamber
1.0% NO
inN2
Particulate
Filter
Teflon
Transfer lines
Lights
Effluent
Bio-
Ahiminum End Plates
Reaction Chamber
22.7 m3
Mixing
Fan
Clean Air
Bio-
chamber
Lights
-------
Gas and Particulate Phase Mutagenicity
of Dilute Wood Smoke + NOx in Air
17500
u 15000
u 12500 -
10000 -
cd 7500
c
to
5000 h
2500 -
0 bi
TA 100
G P
Before After
Irradiation Irradiation
Gas Phase
1 Particulate
-I Phase
i
TA 98
G
G P
] j
Before After
Irradiation Irradiation
-------
CHEMICAL SPECIATION OF NICKEL
IN AIRBORNE DDSTS
Vladimir J. Zatka
Inco L imited
J.Roy Gordon Research Laboratory
Mississauga, Ontario L5K 1Z9
David Maskery
Inco Limited
Central Process Technology
Copper Cliff, Ontario POM 1N0
J. Stuart Warner
Inco Limited
1 First Canadian Place
Toronto, Ontario M5X 1C4
Responsible management of nickel-related health hazards in an indus-
trial environment requires analytical knowledge, both qualitative and
quantitative, of the various nickel species present in the airborne dust of
the workplace. A wet-chemical procedure is described which apportions the
airborne nickel species into four categories: (1) water soluble; (2)
sulfidic; (3) metallic; and (4) oxidic. The procedure involves subjecting
the dust to three sequential leaches with (1) ammonium citrate, (2) hydrogen
Peroxide-ammonium citrate, and (3) bromine-methanol, each of which selec-
tively dissolves a number of related nickel compounds. The three leachates
and the final residue are then analyzed for nickel and possibly for other
elements. Since real samples usually contain a limited number of compounds,
this categorization is often adequate to speciate the sample. The potential
ambiguity of this approach is compensated by the fact that it can be readily
implemented in regular analytical laboratories. The underlying chemistry of
each leaching step is discussed and its selectivity and completeness are
evaluated. The peroxide-ammonium citrate mild oxidative leach is time-
intensive (2 h) and requires pH control and small sample size (<50 mg). The
individual leaching methods were tested in an international interlaboratory
Program and the results are discussed briefly.
605
-------
In the past, an increased incidence of respiratory cancer was associa-
ted with certain nickel refining operations. The compound(s} that may have
been responsible for this risk has not been identified but some researchers
and regulators have hypothesized that all forms of nickel should be consid-
ered to be potential carcinogens. However, extensive epidemiological
studies have shown no nickel-related riBk of cancer in the nickel-using
industries or in the great majority of occupations in the nickel-producing
industry. Furthermore, laboratory experiments indicate significant differ-
ences in the carcinogenic potential of different nickel compounds. Given
the important role nickel plays in our technology, the efficient management
of risk requires the ability to distinguish quantitatively between the
different species of nickel in airborne dusts.
No universal analytical technique capable of quantifying as well as
identifying all nickel species present in minute dust samples is known.
Researchers are currently trying to solve the problem by instrumental and
wet-chemical methods. Instrumental approaches deal with physical character-
istics of atoms, less often of molecules, and usually require a powerful
data processing capability. Nearly all of these techniques examine only a
Bmall area of the sample to the depth of a few micrometers. This limitation
can cause problems during calibration and speciation as minute amounts of
airborne dust are often not distributed uniformly on filters. Last, but not
least, few laboratories have access to the sophisticated instruments and the
researchers with the experience and skills needed to correctly interpret
results. Speciation by wet-chemical methods, on the other hand, exploits
the differences in chemical properties of the various phases found in nickel
bearing dusts. Such methods utilize all of the dust on the filter so
non-uniform deposition is not a problem. However, the overriding advantage
of a chemical approach to speciation is that it can be used by the majority
of normal chemical laboratories.
To be sure, wet-chemical speciation is seldom as simple and accurate
as conventional wet-chemical determinations of total elemental contents.
Speciation has problems of its own especially on a micro scale, such as in
the analysis of personal monitor filters. Of primary concern is finding
conditions which result in highly selective separations. If this is not
possible, some ambiguity in Bpeciation will result. However, knowledge of
the process in the workplace often helps rule out some phases identified as
possibilities by the speciation. Furthermore, the toxic properties of the
compounds of a given metal are probably related to their wet-chemical
behaviors.
Experimental Approach
We developed a procedure for the speciation of nickel-containing
phases based on multiple sequential leaching of a dust sample with reagents
of increasing chemical "power"* Three selective leaches were used - ammonium
citrate, hydrogen peroxide-ammonium citrate, and broains-metbanol solutions
- followed by a final decomposition of the insoluble leach residue by nitric
and perchloric acids. The four solutions so obtained were analysed for
nickel by normal atomic absorption or plasma emission methods. Nickel com-
pounds present in the sample were thus divided into the four groups which
are identified in "Table I. The procedure was tested on small samples
<<30 mg) of fine dusts (<45 jim) from the mill, smelter, and refinery, indi-
vidually and in mixtures. The simple four-fold categorization of the great
variety of industrially important nickel compounds seems adequate as real
dust samples usually contain only a limited number of different nickel-
bearing species and these can usually be deduced from the origin and history
of the sample*
606
-------
Results
Soluble Nickel
Soluble nickel compounds are leached with a 0.1M solution of ammonium
citrate (pH 4) rather than water. The citrate solution acts as a buffer and
so ensures leach reproducibility by minimizing effects caused by foreign
hydrolyzable salts. At the same time, complexing ability of the citrate ion
toward higher-valent elements, such as iron(IIX), prevents unwanted precipi-
tation of hydrolytic products. Citrato complexes of nickel are relatively
weak, the most stable being NiCit" (log K - 5.4), and thus do not change the
water solubility of "insoluble" nickel compounds.
The dissolution of nickel from basic nickel salts, for instance
Nix(OH)y(SO^)z.aq, with water or ammonium citrate is incomplete and decreas-
es with increasing basicity. Up to half of the contained nickel has been
found to be leached from nickel carbonate and nickel hydroxide.
Sulfidic Nickel
This category comprises a large variety of nickel compounds, the most
important of which are sulfides. Leaching with mild oxidizing solutions,
all containing hydrogen peroxide, was proposed1*2'3 for the selective
determination of nickel sulfide in binary mixtures with oxidic nickel.
Others'* reported poor results and incomplete sulfide extractions with one
Buch mixture of hydrogen peroxide and ammonium citrate1 . By extensive
study, we found that a mixture of 1 volume of 30Z hydrogen peroxide and 2
volumes of 0.1M ammonium citrate produced consistent results provided the
solution pH was kept near 4. A minimum of two hour* was usually necessary
to ensure complete leaching of the sulfide phase from a 10-20 mg portion of
dust. The reaction proceeds with formation of thiosulfate a* the product of
oxidation;
Ni3S2 + 5H202 + 3HCit2- - 3NiCit~ + S2O32- + OH" + 6H20 (1)
2NiS + 4H202 + 2HCit2" - 2NiCit" + S2O32" + 5H20 (2)
Of the two nickel sulfides, 8-HiS dissolves more slowly. It is worth noting
that as the sulfides dissolve, the pH would increase if not for the buffer-
ing ability of the leach solution.
Nickel compounds other than sulfides which fall into this category
include arsenides, selenide* and tellurides. All were found leachable by the
peroxide-citrate solution, though at varying rates, as shown in Table II.
Ten rag samples of NiAs, NijjAsa and NiSe dissolved within two hours. NijAsj
and NiTe reacted at a much slower rate and almost 20 hoars were needed to
diasolve 10 mg portions. Fortunately, these compounds are present in small
amounts, if at all, in dusts from the nickel industry.
Tests with pure metallic nickel showed that at pH 4 it was not
attacked by the peroxide-citrate solution. However, it was deemed necessary
to verify that this selectivity was not impaired by side reactions, such as
cementation or surface activation, caused say by the leaching of copper and
"*382 from plant dusts. An in-depth leaching study of metallic phases
(carbonyl nickel and ferronickel powder) showed that only 0.3 to 0.61 of the
nickel dissolved after four hours. The leaching of metallics was not
increased by the addition of either copper(11) or thiosulfate. Similarly
recoveries of teachable sulfidic nickel from synthetic mixtures made of
^382, chalcopyrite concentrate, nickel or ferronickel powders and soluble
cupric salt were satisfactory (92-96%) after two hours of leaching.
607
-------
Metallic Nickel
Separating metallics from oxides by selective leaching has been a long
established practice. A variety of oxidizing solutions ranging from silver,
mercuric or cupric salts to chlorine5 or bromine6 in alcohol were used.
Metallic nickel, pure or as ferronickel, readily dissolves in a solution of
bromine in methanol (3% v/v). Sulfidic nickel species which have not been
removed by a prior peroxide-citrate leach will also dissolve. Precipitated
nickel hydroxide or carbonate should be absent as they are attacked to
varying extents (0.2—5% Ni leached) depending on the history and composition
of the specimen.
Oxidic Nickel
All oxidic forms of nickel remain in the final residue. They are
solubilized by fuming with nitric and perchloric acids. We prefer collec-
ting dust on PVC acrylic copolymer filters, such as the Gelman DM Metricel
membrane, because they are insensitive to humidity fluctuations and, unlike
plain PVC, are easily wet-ashed during oxide dissolution. Complete dissolu-
tion of the oxidized nickel phases can be assured only by fuming with
perchloric acid assisted, in extreme cases, by added hydrofluoric acid.
Analysts occasionally underestimate the refractory nature of oxidized nickel
phases and, upon leaching the sample with aqua regia only or not heating the
perchloric acid to high enough temperatures, report nickel values which
suffer from an extensive low bias.
First Interlaboratory Test Program
The first interlaboratory program for the determination of diverse
nickel phases in airborne dusts was sponsored in 1985 by the Nickel Produ-
cers Environmental Research Association (NiPERA). Twenty laboratories from
eight countries evaluated the individual selective leach procedures on four
bulk samples of dust: two synthetic mixtures and two plant dusts. Each of
the four procedures was tested on two days using duplicate 10 mg sample por-
tions. The small sample size was chosen so as to operate under conditions
more closely resembling those involved in the analysis of loaded filters.
The raw data reported by the participating laboratories was treated statis-
tically according to ISO 57 257 , and the outliers identified at the 95X
probability level were removed. A final summary is presented in Table IIIs.
Considering the small portions analyzed and the inherent problems with
phase analysis, repeatability "r" and reproducibility "R" of the reported
results was generally acceptable. A few results, particularly for total
nickel in the high grade sample ES-12, appeared to suffer from high bias
which could be related to weighing or instrument calibration problems. More
than 85% of all reported data sets passed the statistical tests. It was
concluded that, with due attention by the analyst to the demands of microan-
alytical work, the four leaching methods tested provide a reasonable basis
for the speciation of airborne nickel phases into four categories.
A second international program is now underway to test the sequential
leach procedure using three bulk samples of plant dusts.
608
-------
References
1. E.M. Katsnel'son, E.Ya. Osipova, "Selective determination of nickel
sulfide", Oboaaghchenie Rud 5: 24 (1960); Chen. Abstr. ££:26854g (1961).
2. P.M. Grobae, C.H. Ewald, W.F. Gutknecht, "Determination of nickel
subaulfide in the presence of other nickel compounds", Paper #727,
Pittsburgh Conference, New Orleans, LA, March 1985.
3. J.J. Lynch, "The determination of copper, nickel and cobalt in rocks by
atomic absorption spectrometry using a cold leach", J. Geochem. Explor.
11: 313 (1971),
4. P.R. Klock, G.K. Czamanske, M. Foose, J. Pesek, "Selective chemical
dissolution of sulfides: an evaluation of six methods applicable to
assaying sulfide-bound nickel", Cham. Gaol. ££; 157 (1986).
5. G. Kraft, "Bestimmung von metallischem und oxidischem Nickel in
Huttenprodukten", Fresenius' Z. Anal. Cham. 209: 150 (1965).
6. U. Penttinen, V. Falosaari, T. Siura, "Selective dissolution and
determination of sulfides in nickel ores by the bromine-methanol
method", Bull. Geol. Soc. Finl. 49: 79 (1977).
7. ISO Standarda Handbook 3. 2nd ed., "Statistical Methods1', ISO 5725,
International Organisation for Standardization, Geneva,
Switzerland, 1981.
8. St.J.H. Blakeley, V.J. Zatka, "Interlaboratory test program on nickel
phase speciation in dust samples", Report to the NiPERA Scientific
Advisory Committee, Inco Limited, J. Roy Gordon Research Laboratory,
Mississauga, Ontario, Canada, November 1985.
Table I* Separation of nickel-containing phases by a sequence of selective
leaches
SXSJUL Leach solution
1 0.1H ammonium citrate
Normal salts
(sulfate, chloride, sulfamate,etc.)
Leached nickel phaaaa
2
H2O2 - citrate
Telluride (NiTe)
3 Bromine-methanol
4 Nitric-perchloric acids
Metallic (pure Hi, ferronickel)
Refractory oxide (FiO, Hi ferritt
silicate)
609
-------
Table II. Leaching with hydrogen peroxide-ammonium citrate solution (pH 4).
Compound
Sulfides
Arsenides
Selenide
Telluride
XRD
N13S7
6-NiS
(Fe.NiJgSg
NiAs (72%)
HinAsg (28%)
NijAsj (major)
NinAsg (major) 1
NisAsj (v.minor) '
}
NiSe
NigSes
(major)
(v.minor)
Recovered by leaching
10 me samples. %Hi
NiTe (major)
Iki
98
94
93
100
62
100
100
73
20 hrs
100
99
100
98
100
Table III. Interlaboratory testing of the individual selective leach
procedures on bulk dust samples.
Leach solution:
H2O2+ Brj* HNO3+
Cjttttt citrate methanol BC10/r
Soluble*
Soluble*
sulfidic*
Soluble*
sulfidic* metallic*
Sample
sulfidic
MtftlUf
oxidic
ES-10
Grand mean
.58
1.87
8.02
10.99
(synth.)
Repeatability r
.070
.095
.74
.37
Reproducibility R
.22
.31
2.54
1.17
Data sets accepted
18
17
20
19
ES-11
Grand mean
1.76
4.57
9.02
15.78
(synth.)
Repeatability r
.39
.36
.52
1.02
Reproducibility R
.82
.82
2.72
2.01
Data sets accepted
19
18
19
19
ES-12
Grand mean
1.16
12.55
15.74
72.12
(plant)
Repeatability r
.21
1.02
.89
2.77
Reproducibility R
1.00
2.64
4.22
8.20
Data sets accepted
18
19
19
18
£8-13
Grand mean
2.04
2.25
3.18
27.16
(plant)
Repeatability r
.21
.23
.44
1.27
Reproducibility R
.91
.55
2.07
3.04
Data sets accepted
19
19
20
19
K8-12 - Fluid bed roaster rafter dust
BS-13 - Nickel alloy plant flue dust
610
-------
CORRELATION OP MICROPROBE ANALYSIS DATA FOR INDIVIDUAL PARTICLE SPECIATION
- NICKEL COMPOUNDS FROM STATIONARY SOURCES
0.T. Rickman, I.H. Musaelmanc J.O. Mullis, R.K. Linton
Department of Chemistry
University of North Carolina
Chapel Hill, N.C. 27514
A challenging area of analytical chemistry ia the development of new
techniques to provide apeciation information about elements present In
small quantities within complex environmental samples. Microscopic
chemical analysis techniquesr e.g. laser microprobe mass spectrometry
(LAMMS) and scanning electron mieroscopy coupled to energy dispersive x-ray
spectroscopy (SEM/EDS) combine the advantages of high spatial resolution
and sensitivity. They are used to investigate chemical species and their
Physical distribution between or within individual particle®. The
correlative microscopy studies are part of an EPA-aupported project to
evaluate methods for nickel apeciation in airborne particles produced by
stationary sources.
6i i
-------
Introduction
Aa a consequence of the physical and chemical heterogeneity of
airborne particles emitted by stationary sources, it is valuable to obtain
microscopic analyses of individual particles. For example, the
variabilities in elemental concentrations and speciation, both within and
between particles, are important factors influencing environmental impact.
Of special interest are particles in the respirable size range which also
may be less efficiently captured by pollution control devices.
Consequently, analytical techniques are desired which provide
sub-micrometer spatial resolutions and which are sensitive to chemical
components within analytical volumes corresponding to total masses of lO--^
grams. Two such techniques utilized in this study are laser microprobe
mass spectrometry (LAMMS) and scanning electron microscopy/energy
dispersive x-ray microanalysis (SEM/EDS). The LAMMS technique uses high
power density pulsed laser beams to ablate and ionize single particles.
Mass spectra, including characteristic atomic and cluster ions, provide
information on chemical composition. The SEM/EDS provides information on
particle morphology (by secondary and backscattered electron imaging) and
composition (by analysis of characteristic electron-excited x-ray
emissions). Whole particles and particles viewed in cross-section are
characterized by SEM/EDS to provide information on inter- and
intra-particle compositional variations, respectively. A correlative
regimen also is developed permitting SEM/EDS followed by destructive LAMMS
measurements on the same individual whole particles. The above techniques
are not only complementary, but also may be used to verify information
obtained by more conventional bulk analytical techniques (wet chemistry,
x-ray powder diffraction). The specialized microprobe technques serve,
therefore, to assist in the evaluation of other analytical methodologies
which are more appropriate for applications to routine source or
environmental monitoring of toxic elements and compounds. In other
instances, sample sizes may be inadequate to permit any analyses other than
microprobe studies of individual particles.
To illustrate the combined microprobe techniques, the characterization
of micrometer sized particles produced near a fluidized bed roaster (FBR)
in a nickel refinery will be discussed. The FBR dust is a useful sample
for analytical technique evaluation in that it is known to contain several
major classes of nickel-containing compounds (e.g. oxides, sulfides,
sulfates) which may have differing environmental impacts. The nickel
612
-------
subsulfide of particular interest as a result of its possible
carcinogenic potency in humans. This project is part of an EPA-supported
project to evaluate methods for nickel speciation monitoring in stationary
source emissions. More detailed descriptions of the experimental methods
and results are available elsewhere (1,2).
Experimental methods
Standard and sample preparation
LAMMS and SBM/EDS correlative atudv. The particle Standards Ni, NiO,
Nisc>4, NiS, and NI3S2 were mounted dry on forravar coated transmission
electron microscope (TEH) finders grids. The FBR dust sample was dispersed
in freon in an ultrasonic bath. A drop of the sample dispersed in freon
was deposited on a coated TSM grid and allowed to dry. All samples and
standards were carbon coated to reduce charging in the SEM.
Light microscopy fLMl and SEM/BDS norrelat-lva wfurty. Cross-sections
of the FBR dust sample and nickel standards were obtained by first
embedding the particles in epoxy followed by curing of the epoxy, and final
polishing. These particle cross-sections were carbon coated for SEM/EDS.
Intra-particle chemical variations were investigated by elemental mapping
for Ni and S using characteristic x-ray emissions.
Instrumentation
The SEM,'EDS used was an ISI DS-130 SEM with a Tracor Northern TN-5500
energy dispersive spectrometer and digital imaging system. The LAMMS
analyses were performed using a Leybold-Heraeus LAMMA-500 instrument.
Correlative procedure
LAMMS and SBM/Etaa. Using the SEM/EDS, particles were analyzed at an
electron beam energy of IS keV for 99 seconds live counting time per
particle. Particle analysis software generated size and shape information,
as well as background-subtracted integral x-ray intensities for specified
elements. Grid squares analyzed had an intact formvar coating and
contained well-dispersed particles (particles usually greater than 3
micrometers apart). The same particles were then analyzed destructively by
LAMMS. The laser and mass spectrometer conditions were selected to allow
for complete ablation of each particle and to maximize production of
cluster ions characteristic of individual nickel compounds. Details of
613
-------
SEM/EDS and LAMMS experimental parameters are described elsewhere (2).
in and sEM/EDS. Particle cross-sections were examined using
polarized reflected light, and subsequently mapped for nickel and sulfur
distributions using the SEM/EDS.
Results and Discussion
Standards
LAMMS. Fingerprint mass spectra of Hi, NiO, NiS04, Ni3S2 and NiS
were obtained using LAMMS. Xt was not possible to differentiate between
NiS and Ni3S2 standards, although all other compounds produced
characteristic patterns of positive or negative cluster ions (1,2).
SEM/EDS. A range of +/- sigma values about the average for S/Ni Ka
x-ray intensity ratios for NiS (1.9-2.5), Ni3S2 (1.2-1.5), and NiS04
(1.5-2.6) standards were obtained. Approximately 50 particles were
analyzed per standard to obtain these ranges. Variabilities reflect
fundamental particle size and shape effects on x-ray emissions as described
elsewhere (2), An example of SEM/EDS data is shown in Figure 1 for a
particle in the NiS standard. For comparison, a corresponding LAMMS
spectrum also is presented.
Nickel refinery sample - Fluidized Bed Roaster (FBR) dust
Lamms »i«t fiEM/ttns. of 115 FBR particles analyzed by SEM/EDS and
LAMMS, LAMMS was used to differentiate 57 into either 1] Nix0y/NiS04 or 2]
Ni3S2/NiS classes. Since only positive ion spectra were used, it was not
possible to distinguish NixOy and NiSC>4 solely from the LAMMS data. The
other 58 particles did not produce intense nickel-containing cluster ions
owing to lower nickel concentrations and instrumental factors (2). Using
the S/Ni x-ray intensity ratios from SEM/EDS the 57 particles were further
subdivided into these classes: NiO (17 particles), NiSC>4 (9 particles),
NiS04 or a mixture including NiSC>4 and Nix0y (17 particles), Ni3$2 (6
particles), NiS (4- particles) and a mixture containing NiS and/or Ni3S2
particles).
LM and sum/eds x-rav mapping. The above results on whole particles
614
-------
suggested that many contained multiple nickel species, especially oxide and
•
sulfate combinations. This was substantiated by cross-sectional elemental
mapping by SEM/EDS. Many particles showed large variations in sulfur
distributions, including some with surface enrichments of sulfur. It is
likely that such particles contain layers of nickel sulfate deposited on an
oxidic nickel core. This was further confirmed by surface analysis using
x-ray photoelectron spectroscopy (XPS) showing nickel sulfate as a major
surface component. Additional experiments on the particle cross-sections
are envisioned using ion microscopy to further characterize intra-particle
mixtures of nickel compounds
Acknowledgements
Support of this research under U.S. Environmental Protection Agency
(EPA) Cooperative Agreement No. CR-812908-01-1 with Prank Butler is
gratefully acknowledged. Stuart Warner, INCO, Limited, also is kindly
acknowledged for providing access to particle samples used in this study.
References
1. I.H. Musselman, R.W. Linton and D.S. Simons, "The Use of Laser
Microprobe Mass Analysis for Nickel Speciation in Individual Partioles of
Micrometer Size", Microbeaw An»1v«tn-19flS. 337-341.
2. I.H. Musselman, J.T. Rickman and R.H. Linton, "Fingerprinting of
Chemical Species in Microparticles - Correlative Laser and Electron
Microprobe Studies", Mlernheam An»lyaia-19B7, in press.
613
-------
tn
3
l
ft
NICKEL SULFIDE
M
m/z
N!
N! I
^ I*
10
Entr«v CK»V)
Figure 1. Spectra obtained on NiS standard by LAMMS(top) and SEM/EDS
(bottom).
616
-------
evaluation and oftimizaticn of a wet chemical mehcd
PGR CEIEEMINING THE HYDROGEN SUIfTUE OCNCEMIRATICN OF
CVLCNDER GASES EMPLOYED FOR TBS CEMS QUAUT* ASSURANCE
A. K. Jain, M. D. Marks, L. J. larsan,
H. N. Stryker, R. P. Fisher
National Oouncil of the Paper Industry
far Air and Stream Irprwement, Inc.
Post Office Box 14483
Gainesville, FL 32604
Existing and developing total reduced sulfur (TRS) continuous emission
monitoring system (CEMS) quality assurance programs include, or will
include, requirements for the use of hydrogen sulfide (I^S) calibration
gases. NCASI conducted studies to evaluate and optimize a wet chemical
method for determining the I&S content of calibration gases in the 5 to 1500
ppn range. The study shewed that sanpling tine, saiqple flew rate, holding
time after reagent addition, and regulator ajnditicning ware significant
parameters. Accurate results were obtained following the proposed procedure
upon analyses of 5, 25, 100, 500, and 1500 ppm H2S in nitrogen cylinder
gases.
617
-------
EVACUATION AND OPTIMIZATION OF A NET CHEMICAL METHOD
FOR DETERMINING THE HYEROGEN SULFIDE OONCENIRATTON OF
CYLINDER GASES EMPLOYED PGR IRS OEMS QUALITY ASSURANCE
Introduction
In January, 1985 NCASI published a wet chemical method for determining
the H2S concentration of cylinder gases.1 This method was developed with a
view to assist kraft pulp mills in their total reduced sulfur (IKS)
continuous emission mcriitoring system quality assurance programs by
providing them with an independent means of establishing the concentration
of H2S cylinder gases used to calibrate the TRS analyzer; and it was
validated for an H?S concentration range of 5 to 25 parts per million (ppm)
in nitrogen. Following the publication of this report it was recognized
that there was a need to use this or a similar method for analyzing cylinder
gases containing as ituch as 1500 ppm H?S. these hi$i concentration cylinder
gases are used for calibrating certain IKS continuous emission monitoring
systems which contain integral dilution systems and which are widely used in
the pulp and paper industry. At present no suitable method is available to
establish the H2S content of these cylinder gases. Ihis paper describes a
modified version of the earlier method which has been validated for
analyzing the H2S content of cylinder gases containing 5 to 1500 ppm H2S in
nitrogen.
Method Development Studies
The method fear determining the concentration of H2S in cylinder gas
mixtures that was used in this study is based upon trapping HoS as zinc
sulfide in inpingers containing solutions of zinc acetate, acidifying the
combined trapping solutions in the presence of iodine, and determining
unreacted iodine by thiceulfate titration. During this study a number of
variables' that had an inpact on measured H2S concentrations were examined
and optimized. The results of sane of those studies are described below.
The final method and the required equipment are described in Appendix A.
The details of the method have been published elsewhere.2
Regulator Conditioning
Data in Table I present the results of a series of tests enplcying the
method of Appendix A with a cylinder containing approximately 500 ppm H5S in
nitrogen. The data shew that at a flow rate of approximately 200 ml/mln it
took nearly 3 hours to passivate the regulator. The need to passivate the
regulator was found to be more pronounced for cylinder gases containing more
than 250 ppm H2S in nitrogen which required lower saxiple flow rates.
Consequently, the regulator passivation step has been included in the
procedure.
Holding Time for Sanples Prior to Titration
Tests were conducted to examine the effect of storing samples after
iodine addition prior to their titration with thioeulfate. The results
suggested that a storage time of 30 minutes following iodine addition
allowed the iodine reaction to be complete. Table II presents the results
of one such test.
Loss of Iodine During Transfer of Iitpinger Contents
Tests were conducted to determine the loss of iodine when the aontents
of iipingers containing iodine are transferred to a flask far titration. As
618
-------
an example, the results of one test shewed that the thiosulfate titrant
requirement decreased front an average of 19.41 mL to 19.24 mL when the
contents of inpingers containing Iodine were transferred to a flask for
titration. The method has, consequently, been modified to eliminate the
step requiring transfer of inpinger contents.
Sanple Volume and Sample Flow Bate
Studies showed that highest precision and accuracy were obtained when
the H2S content of the sanple Mas sufficient to consume 35 to 60 percent of
added iodine, and sample flow rate was maintained in a specified range.
Tables III and rv shew the optimum sanple volumes and sampling rates for
different HjS concentration ranges.
Method Performance
The performance of the method of Appendix A was evaluated in the H2S
concentration range of 5 to 1500 ppn. The results of precision and accuracy
tests are presented below.
Accuracy
The accuracy of the method was eetabHnhed by ccnparing the results of
method analyses of seven different cylinder gases with their known
concentrations. For five of the seven cylinder gases the true H2S
concentrations were established by analyzing the cylinder gases with a
permeation tube-calibrated gas ciiranatograph eqpiijjped with a flame
photometric detector. Hie details of this gas chromatographic method are
described elsewhere.1 The other two cylinder gases analyzed by NCASI were
National Bureau of Standards (DBS) Standard Reference Material IfeS in
nitrogen. (Although these NBS reference materials an not yet available
ocmnercially, two of these gas standards ware made available to NCftSI and
wesre used to evaluate the performance of the method under evaluation.)
The results of the accuracy tests are sunmarized in Table V. These
results show that for the H2S concentration range of 5 ppn to 1500 ppn, the
percent difference between the HaS concentration measured according to the
method of Appendix A and the true concentration ranged froo -2 to +3
percent. The average far all seven cylinder gases was less than 0.5
percent. These results suggest that the method is accurate and unbiased.
Precision
As shown in Table V, the relative standard deviation of zcftliosbe
analyses of seven gas mixtures ranged froa 0.9 to 2.9 percent. Th* pooled
relative standard deviation of the test method is estimated to be 1.4
percent for tests conducted within a single laboratory.
Literature References
1. "Wet Chemical Method tar Determining the H-jS Concentration of
Calibration cylinder Gases," NCASX Technical Bulletin No, 450,
NCASI, 260 Madison Avenue, Waw Yoric, NY 10016 (January 1985).
2. "Modified Wet Chemical Method for Determining the H^S
Concentration of Calibration cylinder Gases," NCASI Technical
Bulletin Ho. 519, NCASI, 260 Madison Avenue, New York, Wf 10016
(March 1987).
619
-------
APPE2TOIX A
EQUIPMENT, TEST METHOD, AND CALCUIMTCNS
Description of Equipment
Figure A-l shows the sampling train. The stems of the 100 mL inpingers
(Ace Glass, Type 7541) were lengthened to be approximately 1/2 inch frcm the
bottom in order to increase contact between gas and the trapping solution.
Ihe critical orifice may be fabricated frcm 1/16 inch O.D. stainless steel
capillary tubing.
Sampling and Analysis
Prior to sampling, pressurize the stainless steel regulator and expose
it to test gases for several hours or overnight to passivate activated
sites. Determine the sampling rate, sample volume and sampling tine from
Tables III and IV. Assemble the sampling train and add 50 mL of 0.091 M
zinc acetate solution to each of the first two iapingers. Establish a flow
rate of gas frcm the regulator which is 50 to 100 mL in excess of the
critical orifice flow rate, conduct sanpling far the desired period. While
sampling, titrate 20.0 mL of 0.01 normal iodine solution with 0.01 normal
sodium thiosulfate using starch as indicator. At the end of sampling
pipette 20.0 mL of 0.01 N iodine solution through the stems of the impingers
containing zinc acetate capture solution, dividing it betwaen the tw> (about
15 mL to the first impinger and the rest to the second). Add 2 mL of 2.8 M
hydrochloric acid solution through the stems, dividing it as with the
iodine. Cover the stent openings with paraffin film, stare far 30 minutes in
the dark, then remove the stems with rinsing, and titrate the solution in
the inpiiiger bottoms with 0.01 N thiosulfate. ihe decrease in the amount of
thiosulfate titer corresponds to the amount of H2S captured in the two
lapingera.' Analyze a method blank in a similar manner.
Calculations
Calculate the concentration of H2S (C^ s) using the following equation:
2
S, ppm " K *r (^T, blank ~ VT, sanple)Am, std
2
where K - 12,025 mL (gas)/g eg; Np - normality of standard thiosulfate
solution, g eq/lit; VT blank & % " volume of thiosulfate used in
blank and sample titration, mL; Va 'std " dry standard sample gas volume, L.
620
-------
VENT
ROTAMETER
THERMOMETER
IMPINSER3
NEEDLE VALVE
NATE METER (OPTIONAL)
VACUUM QAUOE
NEEDLE VALVE
PUMP
SILICA OEL \
DRYING TUBE CRITICAL ORIFICE
FIGURE A-l. H2S SAMPLING TRAIN
621
-------
TABLE I. EFFECT OF REGULATOR CONDITIONING ON
MEASURED K^S (XNeEMntATICNS
Regulator Conditioning Measured H_S
No. Period, minutes Concentration, ppm
1 0 508
2 30 507
3 90 518
4 180 526
5 240 521
TABLE II. EFFECT OF HOLDING TIME ON MEASURED H^S (XMCENTRKnCN
Holding Tine
After
Iodine Addition, Measured H_S
Run No. minutes Concentration, ppm
1 15 498
2 20 516
3 30 530
4 40 527
5 60 532
TABES in. OPTIMUM SAMPLE VOILMES
Approximate cylinder Gas Approximate ppm x
HgS Oonoerrtratian, ppti Liters Desired
5 to <30 650
30 to <500 800
>500 to 1500 1000
622
-------
TABLE IV. OPTIMUM SAMPLE FLOW RATES
cylinder Gas
Desired Flew Rates,
KjS Concentration, pjm
xX/nin
5 to <50
1500 p 500
50 to <250
500 t> 250
250 to <1000
200 p 50
>1000
75 J) 25
TABLE V. RESUIilS OF ANALYSES OF H_S CYUNEER GASES
BY THE MEIHDD OF j&RENDIX A
No.
Reference Mathod
Test Results
H-S Cone.,
PI»
Test
Method
Wat Method
Relative
Average standard
I^S Ocnc., Deviation,
ppn Percent
Difference,
Percent
1 7.90
GC/KPD
7.84 (6)*
1.6
-1
2 25.4
GC/FPD
25.5 (6)
2.9
0
3 104
GC/FFD
102 (5)
1.0
-2
4 488
GC/FPD
492 (10)
1.2
1
5 1470
GC/FPD
1480 (6)
1.3
1
6 5.15
NBS
5.32 (3)
0.9
3
7 15.4
MBS
15.5 (4)
1.9
1
*Nuntoers in parenthesis show nurnber of replicate analyses.
623
-------
EVALUATION OF ETHYLENE OXIDE EMISSIONS:
GAS CHROMATOGRAPHIC ANALYSIS OF
SCRUBBER EFFICIENCY
Jeremy N. Bidwell,
TRC Environmental Consultants, Inc.
East Hartford, CT
As Ethylene Oxide {EtO) is now more recognized as a hazard to man and
his environment there is interest in quantifying emissions from a wide
variety of processes. As with any testing procedure, an accurate and
repeatable method of sartpling is needed that is both flexible enough to
accommodate a wide variety of sampling conditions and sample
concentrations, and is straight-forward enough to be adopted by the
analytical community.
Two recent scrubber efficiency tests at hospital supply manufacturers
are examined, where EtO was used to sterilize heat sensitive products
prior to shipping. Samples were drawn utilizing evacuated stainless
steel canisters equipped with low flow regulators to draw 20 minute
integrated samples. Sample analysis was performed on a Gas Chromatograph
using a Flame Ionization Detector.
This technique has shown itself to be a valuable method of assessing
steady state and nonsteady state processes where satrple moisture
concentrations do not exceed 20%, as is the case with most sterilization
processes.
Overview Of Sampling Technique
tony temperature sensitive hopsital products must be sterilized prior
to shipment by exposure to EtO in a sealed chamber. After a
predetermined holding time, the EtO is pumped out of the chamber to the
scrubber, often by means of a water seal vacuum pump. At the outlet of
the pump {which is also the inlet of the scrubber) as well as at the
scrubber outlet, gas stream volumetric flowrates were measured
624
-------
concurrently with EtO concentrations. A Hastings model GSM-IDSK flow
meter was used in conjunction with a atrip chart recorder at both
sampling locations to continuously monitor flowrates.
Gas sanples to be analyzed for EtO concentration were drawn by using
a specially fabricated sampling train, as shown in Figure 1. The
sanpling train consisted of a stainless steel tank of the type used in
EPA Method 25 sampling/ an on/off toggle valve, a small vacuum gauge, a
low flow regulator, and a stainless steel probe. These parts were all
fitted together using 1/4 inch stainless steel Swagelok connection? and
fittings. The tank purge setup (Figure 2) used similar fittings and 1/4
inch Teflon tubing. Each tank was fitted with a quik connect connector
to facilitate easy tank replacement, enabling sanples to be drawn in
quick succession.
Clean sample tanks were stored and transported at a slight positive
pressure to prevent contamination. Just prior to sanpling, a tank was
evacuated; the exact negative pressure (P^) was read from the manometer
on the tank purge setup, and recorded. The tank was then removed from
the purge setup and connected to the sampling train with the toggle valve
in the "off" position. Sanpline; began when the toggle valve was switched
to the "on" position. After sampling, the valve was switched off and the
tank was disconnected from the sanpling train and oonnected to the purge
setup. The pressure
Where Efc ¦ Barametric Pressure
Note that both Pt and P8 are always negative
Pretest Calibrations
Conpressed gas cylinders of several concentration ranges of EtO were
purchased from Scott Specialty Gases of Plumsteadville, Ohio. Fully
Evacuated sampling tanks were filled with these calibration gases and
analyzed using the same method used on the sanples. Standard gases of
other concentration ranges were generated on site by diluting the Scott
gases in a saitpling tank. This dilution was accomplished by filling a
tank to one atm with the gas to be diluted, and fitting the tank with a
septum. Another tank was evacuated and left oonnected to the purge
setup. A glass syringe was used to draw out a small volume (Vs) from the
first tank and inject it in the second tank throught the septum on the
purge setip. this second tank was then pressurised to exactly one atm
and the dilution factor was calculated using equation 2:
Where Vt ¦ Volume of the tank
Dilution factors could be changed by changing the quantity of gas
injected (Vs). Bxanple dilution calculations are presented in Table 1.
625
-------
The low flow regulator on the sanpling train is set at a prescribed
rate (R) before the testing program by use of a bubble flow meter and a
stopwatch. The regulator delivers a constant flowrate until the vacuum
side reaches a pressure of 1/2 atm, which occurs when the tank is half
filled. After that, the sampling rate drops off toward zero. If the
length of the desired sampling period (T) is known, then the optimum
sample flowrate (R) would be:
R = 0.5 V/T (3)
To ensure sanple integrity, a blank check was performed on all sanple
tanks before use by filling each tank with hydrocarbon-free air and then
analyzing a sanple from each tank on the gas chromatograph. Bach
sanpling train was leak checked prior to sampling by connecting a mercury
manometer to the sampling train. With the toggle valve open, the
manometer was observed for a loss of vacuum over a ten minute period.
Criteria for an acceptable leak rate was based on EPA Method 25, which
utilizes a similar apparatus, with less than 2mn Hg vacuum loss over ten
minutes.
Sample Analysis
A Hewlett Packard Model 5710 GC was equipped with a six port gas
sanpling valve and associated equipment as shown in Figure 3, With the
needle valve fully closed, the sample tank was attached to the system by
means of a quick connect fitting. When the needle valve was opened
slightly, the positive pressure inside the sample tank drove gas from the
sanple tank through the whole system, displacing any existing gases in
the sample apparatus with fresh sample gas. This technique used only a
fraction of the pressure in the sanple tank, allowing several duplicate
runs to be made with the same sample. Gas samples of extremely high
concentrations were diluted using the same procedure outlined previously
for the dilution of standard gasses.
The sample injected into the GC had been diluted in the tank.
Therefore, the EtO concentration (C) in the gas stream was found by
multiplying the tank concentration (Ct) by the dilution factor (D) as in
equation 4:
C = (Ct X D) (4)
Efficiency Calculations
For each sample an EtO mass flowrate is calculated using equation 5;
H » (Q x MW X C X K) (5)
Where H - EtO mass flowrate in lb/hr
Q = Gas Flowrate in standard cubic feet per minute
= Molecular weight of EtO ¦ 44.05 g/mol
C - EtO concentration in parts per million v/v
K - 28.3 L 1 npl 60 min 1 lb 10~6 L
1 ft 24.04 L 1 hr 454 g L
K ¦ 15.58 x lO-8
626
-------
For each pair of inlet-outlet EtO mass flowrates, a scrubber
efficiency can be calculated using equation 6-
E » 100 (Hi-HoJ/H (6)
Where E ¦= Percent Efficiency
Hi - EtO mass flowrate at Inlet
Hq - EtO mass flowrate at Outlet
TABLE 1
ETHYLENE OXIDE (ETO) SAMPLING DATA
SAMPLE NO. pj P» Ef fi-
ll- -645 -414 535 5.10
T_0 -645 -188 520 2.55
2_t -645 -1« 722 2.74
2_0 -645 -78 560 2.13
3_I -645 -128 618 2.44
3_0 -645 -80 508 2.04
" Tank initial pressure (mm Hg)
Ps = Tank pressure after sampling (mm Hg)
Pf * Tank final pressure (ran Hg)
TABLE 2
ETO REMOVAL EFFICIENCY DATA
SAMPLE NO.
O
Or
D
c*
H
E
1-1
120
2869
5.10
14,627
11.9
99.59
1-0
120
24
2.55
60.0
0.049
2-1
84.0
8930
2.74
21,111
13.8
99.89
2-0
84.0
12
2.13
25.2
0.014
3-1
60.0
4057
2.44
9397
4.00
97.45
3-0
60.0
124
2.04
248
0.102
Q ¦ Flowrate (SCFM)
Ct ¦ EtO concentration in tank (ppm v/v)
D ¦ Dilution factor .
Cg ¦ EtO concentration in stack (ppn v/vj
H ¦ Mass flowrate {lb/hr)
E ¦ Percent efficiency
627
-------
VACUUM GAUGE
LOW FLOW
REGULATOR
ON/OFF
TOGGLE
VALVE
PROBE
4 LITER
STAINLESS
STEEL TANK
Figure 1. Sampling Train Schematic
628
-------
TO
MANOMETER
A
TO
I VACUUM
PUMP ._L.
ON/OFF
NEEDLE
VALVES
1
SEPTUM
QUICK
CONNECT
COMPRESSED
HYDROCARBON
FREE AIR
SAMPLE
TANK
Figure 2. Tank Purge Apparatus Schematic
629
-------
CARRIER
CARRIER GAS
TO INJECTOR
TO
DUMP
LINE
ROTOMETER
SAMPLE
LOOP
QUICK
CONNECT
TO
SAMPLE
TANK
ON/OFF
NEEDLE
VALVE
Figure 3. Sample Injection Apparatus
630
-------
SAMPLING AND ANALYSIS METHOD
FOR DETERMINATION OF CADMIUM IN STATIONARY SOURCE EMISSIONS
by
Thomas E. Ward
Environmental Monitoring Systems Laboratory
Office of Research and Development
U. S. Environmental Protection agency
Research Triangle Park, North Carolina 27711
and
R. F. Moseman, D. B. Bath, 0. R. McReynolds,
D. J. Holder and A. L. Sykes
Radian Corporation
Research Triangle Park, North Carolina 27709
ABSTRACT
Field evaluation studies were done to assess the usefulness of a Modified
EPA Method 5 sampling train and flame atomic absorption spectrometry for the
measurement of cadmium in stationary source stack emissions• These studies were
performed at a municipal solid waste Incinerator and a sewage sludge Incinerator.
The methodology developed in these studies 1s also for application, subject to
verification, at other sources of cadmium emissions at or above the method detec-
tion limit, A formulation of the methodology was tested previously in laboratory
evaluations 1n which the measured overall accuracy and precision were 89,2 percent
and 1,7 percent, respectively, for the analytical portion of the methodology.
The field sampling valtdatlon phase 1s discussed 1n this presentation. Collocated,
quadruplicate flue gas samples of 30 and 60 dscf 1n one- and two-hour sampling
times were collected to assure an adequate cadmium content, a representative
sample Including volume of stack gas and duration of sampling time, and production
of data to validate the method In terms of between-traln precision. The detection
limit determined in the previous work in this program for the atomic absorption
Instrument was 0,03 pg/mL. The stack sampling method detection limit for a 30 to
631
-------
60 dscf (0.85 to 1.7 dscm) stack gas sample was found to be 0.U5 to 0.025 yy Cd,
respectively, per dscf (1.7 to 0.88 ug Cd per dscm). The percent coefficient of
variation (precision) of between-traln cadmium concentrations averaged 13.5 per-
cent at the municipal incinerator and 3.8 percent at the sewage sludge incinerator.
Separate analysis of the front half (probe and filter) and back half (impingers)
of each of the field samples revealed that all of the cadmium was collected 1n
the front half, based upon the results that all the back half samples were below
the detection limit. Precision of the cadmium results was not affected by varying
the sample size from 30 to 60 dscf. To broaden the scope and applicability of
the method beyond that shown in this work, other source categories would have to
be tested.
INTRODUCTION
The United States Environmental Protection Agency (EPA) has investigated
cadmium emissions from stationary sources as a hazardous air pollutant. Appro-
priate methods of sampling and analysis must be available to quantify accurately
the emission of cadmium 1n stack gases from stationary sources.
The Environmental Monitoring Systems Laboratory (EHSL) of EPA located 1n
Research Triangle Park, North Carolina, has developed and validated a methodology
for sampling and analysis of cadmium emissions. The purpose of this report 1s to
present the results of field studies of cadmium emissions measurement methodology.
The objectives of the study were as follows:
0 Determine the applicability of a Modified EPA Method 5 train and flame
atomic absorption spectrometry for the measurement of cadmium emissions
from stationary sources.
0 Assure that the method has a detection limit sufficient to measure
expected cadmium 1n municipal solid waste and sewage sludge incinerator
flue gas samples of 30 to 60 dry standard cubic feet.
° Combine the results of these determinations to validate the
methodologies at stationary sources.
632
-------
The field studies were conducted at a large municipal solid waste Incinerator
and a sewage sludge incinerator. At the present time no further source evaluations
are planned.
SAMPLING APPROACH METHODOLOGY
The sampling and analytical methods evaluated 1n these field studies were
developed in the associated prior laboratory evaluation phase of the overall pro-
gram. The methodology uses sample collection by a slightly modified EPA Method b
(MM5) sampling train, followed by acid digestion of the sample 1n a Parr bomb and
then analysis by flame atomic absorption spectrometry.
The field tests were conducted at a municipal waste incinerator which had
been tested previously for cadmium emissions and at a sewage sludge incinerator.
Cadmium concentrations reported during the previous testing of the waste Incinera-
tor ranged from 23 to 230 ug/dscm. The range of cadmium found in this study was
of a similar range from 32 to 115 yg/dscm. The sampling method employed was an
MM5 train which used nitric acid In the first two impfngers instead of the water
used normally in the Method 51 train. Separate analyses were performed on the
front (probe and filter) and back (iroplngers) halves of the trains to determine
the collection efficiency of each half. The same set-up was used at the sludge
Incinerator and the range measured there was 836 to 1137 ug/dscm.
In order to assess sampling and sample recovery precision, four "Identical"
samples were collected simultaneously using a quad-probe which consists of four
separate sampling trains operated colncldentally2 at a single sampling location.
Groups consisting of four simultaneous samples were collected at the rate of one
or two groups (quads) each day over a three day period for a total 24 samples at
the waste Incinerator and 16 at the sludge Incinerator. Sampling was conducted
1sok1net1cally at 0.5 dscfm for all test runs.
633
-------
Samples were collected for periods of roughly one or two hours to yield
total sample volumes ranging from 32 to 65 dry standard cubic feet. The sample
volumes were chosen to demonstrate that: 1) cadmium emissions from the incinera-
tors could be measured in samples of this size, and 2) this size range of sample
volume would not affect adversely the analytical precision and accuracy of cadmium
concentration results.
The following procedure was used to recover the cadmium from the sampling
train components. For the front half the probe rinses were combined with the
glass fiber filter for digestion and then analysis. For the back half the impin-
ger solutions and glassware rinses were combined and then analyzed.
RESULTS
The field testing portion of this study was designed to evaluate the precision
of the sampling and sample recovery procedures, the portions of cadmium collected
in the front and back halves of the MM5 train, and the effect of the stack sample
volume on cadmium measurement.
Table I presents the cadmium concentrations and precision assessments for
the quad-train field studies.2 The between-train pooled standard deviations were
12,39 and 38 yg/dscm for the waste and sludge incinerators, respectively, and 1n
terms of percent coefficients of variation, the pooled values were 13.M and 3.a
percent for the waste and sludge Incinerators, respectively, which are measures
of the precisions of sampling and analysis for the two sources.
Contributions to the variables of standard deviation and percent coefficient
of variation shown 1n Table I result from: 1) differences 1n the sampling trains,
2) variations between trains 1n the sample preparation and recovery steps, and 3)
analytical variability. Between-test pooled variability includes all of the
above and: 1) the day-to-day variability of the cadmium concentration 1n the
634
-------
feed, 2) effects of different plant operating conditions and 3) the potential
effect of with1n-run variability on the cadmium collection.
A second major variable which was addressed was the collection efficiency
of the sampling train 1n terms of front and back half portions. Not cadmium was
detected in the back half, indicating that the front half (probe and filter) was
very efficient (>99.99X efficiency) 1n capturing effectively all the cadmium
measured.
A third major variable evaluated in the data set of Table I 1s the length of the
sampling period for each run. About half of the tests were conducted for roughly
one hour of sampling time and the other half of the tests for about two hours.
An analysis of variance confirmed that there was not a significant difference 1n
the variabilities of the'cadmlum concentrations of the one-hour compared to the
two-hour runs, at the 5% level of significance (95% probability). Thus, sampling
times of about one hour will yield concentration data equivalent to that for
two hours at these conditions.
Anomalies 1n the cadmium concentration data set are described here. The
average cadmium concentration level in Test No. 1-M was much lower than for tests
2-M through 6-M. The cadmium concentrations shown for Tests 2-M through 6-M 1n
Table I reflect a reasonably constant average cadmium concentration 1n the stack
gas. however, there Is no reason to suspect that the Test No. 1-M samples were
collected any differently than other test run samples. Therefore, these data
were Included 1n the data analysis. The concentration difference between Run No.
1-M and the other runs 1s due probably to plant operating conditions. Sample B
1n Test No. 6-M shows an extremely high value in terms of cadmium per dscm, but
since the Dixon outlier test Indicates that this value 1s an outlier 1t was not
Included 1n the statistical analysis.
635
-------
CONCLUSIONS
Several conclusions and recommendations are made regarding the proposed use
of a slightly Modified EPA Method 6 (MM5) sampling train and flame atomic absorption
spectrometry for the measurement of stack gas cadmium in stationary sources. Some
of these are Included from the prior laboratory portion of this program. They are:
1. An MM5 sampling train and flame atomic absorption spectrometry were
found to be applicable for the measurement of cadmium 1n stationary
source stack gas samples.
2. The overall accuracy and precision of the analytical steps of the
methodology were 89.2 percent and 1.7 percent, respectively. The
detection limit of the analytical instrument was 0.03 pg Cd/ml of
prepared sample.
3. The method detection 11 mlt for a 30 to 60 dscf (0.8b to 1.7 dscm)
stack gas sample was 0.05 to 0.025 ug Cd/dscf (1.7 to 0.88 ug Cd/dscm).
4. The percent coefficient of variation of between-traln cadmium concen-
trations ranged from 3.4 to 23.1 percent for the six municipal sampling
runs conducted and a remarkably close range of 2.8 to 4.7 percent at the
four sludge runs. Stack gas samples of approximately 30 and 60 dscf
were collected and the cadmium results for the two sample sizes did not
differ significantly; therefore, the method bias was not affected by
these sample volumes. Greater than 99.9 percent of ttie cadmium was
collected 1n the front half of the sampling train for each run. This
suggests strongly that the HNO3 1n the impingers 1s not necessary at
these sources and that water can be used.
5. If a purchased stock solution of cadmium is used for preparation of
working standards, the concentration should be verified against an
independently prepared cadmium standard.
6. At least one sample from each source should be checked using the method
of additions to ascertain that the chemical composition and physical
properties of the sample do not cause erroneous analytical results.
7. To ensure the applicability of the MM5 train for stationary sources with
significantly different processes and properties, further field evalua-
tions should be conducted.
REFERENCES
1. Federal Register; United States EPA Reference Method 5; 40 CFR, Chapter I,
Part 60, Appendfx A, July 1986.
2. Moseman, R, F., D. B. Bath, J. R. McReynolds, D. J. Holder, A. L. Sykes, and
T. E. Ward. Laboratory and Field Evaluation of Methodology for Measurement of
Cadmium in Stationary Source Stack Gases. EPA 600/4-84-04$, becember 1986.
636
-------
Table I. Cadmium Concentrations and W1th1n-Run Precision Assessments
{ug/ds cm)
Test No.
A
Sample Train
B C
D "
Ave£age
X
Standard3
Deviation
Percent
Coefficient
of Variation
(*-CV)
Municipal Waste Incinerator
1-M
32.2
32.8
30.4
32*6
32.0
1.10
3,4
2-M
96.5
85.6
87.2
104.1
93.4
8.63
9.2
3-M
105.5
115.5
73.5
74.9
92.4
21.36
23.1
4-M
76.2
88.0
101.2
76.7
84.8
11.67
13.8
5-M
80.0
111.5
103.3
88.3
95.8
14.24
14.9
6-N
101.4
328.0*5
93.8
95.6
96.93
3.97
4.1
pooled
12.39d
13.640
Sewaqe Sludqe incinerator
1-s
2-S
3-S
4-S
886.2
1,137.1
989.1
1,097.8
927.8
1.133.4
1.088.0
1.041.1
835.5 886.6 884.0
1.G81.9 1.081.4 1,108.5
993.6 1,057.5 1,032.1
1,030.9 1,081.9 1,062.9
37.8
31.0
48.7
32.0
4.27
2.80
4.71
3.01
pooled
38.0°
3.791*
n t » swnyic wiaiu ie»v
cubic meter, yg/dscm:
Drobe rinse + filter ~ implnaers (ua)
total volume of stack gas sampled
bjhis value was excluded from the data analysis because when subjected to a
Dixon outlier teste f0r jest No. 6-M, It did not meet the acceptance criterion
of 126,1 ug/dscm, at the 5 percent level of significance (I.e., 95* probability).
^The Dixon outlier test may be found in Dixon* Mtlford J. and Frank J. Massey, Jr.,
"Introduction to Statistical Analysis", McGraw-Hill Book Company, New York (1057).
spooled ¦
a.
637
-------
A Critique of The Revision to EPA Method 25
Kochy K. Fung, John A. Wood, and Michael L. Porter
ERT, Inc.
INTRODUCTION
G.B. Howe et al.1 have done some very credible work in solving some of the
problems with the original EPA Method 25. The original method was created
by the Los Angeles County Air Pollution Control District, now the South
Coast Air Quality Management District in 1971. The authors of the
proposed new Method 25 have done a good job of using more modern
chromatographic column materials and newer catalysts to improve certain
aspects of the method. The tests they ran were almost exclusively in
lab tests, with a limited amount of field testing.
The authors of this paper wish to share their field experience with
Method 25 to predict some problems which will be encountered with the new
Method 25. Mr. Wood (who Btarted working for the Los Angeles County APCD
in 1970) was one of the originators of Method 25, and has 16 years field
experience with the method. Mr. Porter (who joined the Los Angeles County
APCD in 1974) has 12 years experience with Method 25, and produced a
modification of the method which allows accurate determinations of NMOC on
ambient air. Dr. Fung has over 20 years experience in practical
analytical chemistry methods development, recently built a Method 25
apparatus to the new specifications and used it in the field. All of the
authors are now employees of ERT in Newbury Park, California.
The major problems with the new Method 25 are as follows:
• The method is only suitable for testing emissions of relatively
volatile solvents which are not controlled by combustion, yet this
limitation is not given in the written method.
• The heated filter box is heated only to 121°C, even though many of
the organics encountered in the field will condense at this
temperature.
• The heated filter box may be unnecessary.
• The condensate trap will plug with ice much more easily than the
original trap.
• The sample tank minimum volume of 4 liters is too small.
• The dry ice should be pelletized, not crushed. Crushed dry ice
cakes, and pulls away from the trap.
• The valve on the canisters should be mounted "backwards", so that a
leak in the valve packing will not affect the sample during storage.
• The trapped sample gas should not be flushed into the sample tank.
• The condensate trap should not be warmed slowly to ambient
temperature.
638
-------
• The condensate trap must be heated to tenq>eratures several times
higher than the recommended 200°C.
• The oxidation catalyst is not necessary and causes significant
tailing of the C02. The oxidation tube should be packed with
6-8 mesh crushed quartz and operated at 800°C, or at a temperature
high enough to burn methane.
• Frequent monitoring of the C02 levels by GC while burning the trap is
not necessary.
9
• The flow control valve used to control the pressure in the system
wbile metering sample into the evacuated canister, should be replaced
with an upstream pressure regulator.
• The intermediate collection vessel with a volume of at least 6 liters
is much too large- A two liter bulb is sufficient and provides
better sensitivity.
• The sample tank is pressurized to an excessively high level, causing
a decrease in sensitivity.
• Too many replicates are done, increasing the amount of time required
for analysis significantly.
There are many other minor problems, but space considerations limit us to
discussion of the above problems.
The proposed Method 25 is designed for and has been tested with high
volatility solvents such as toluene, isopropyl acetate and hexane. In the
field, one encounters a very wide range of substances and conditions.
Most printing inks use solvents with a foiling range that goes above
200°C Also, afterburner systems produce organic compounds with
relatively higher boiling points than the laboratory test compounds.
Since the heated filter box is heated only to 121'C, we are certain that
in many cases there will be significant losses of °0^ately J°J®^^
organics. There is even some question as to the necessity of the heated
filter box. In the original Method 25, the probe was part of the trap, so
condensation ^n the probe was not lost The able tip of t e
was an 1/8-inch O.D. stainless steel tube bent 90 . This tip was oriented
to point away from the gas stream. The sampling ratewastypically around
50 to 100 cc/min, producing a linear flow rate in the tube of about
1 4 ft/sec The sample gas stream is flowing in the opposite direction at
\ r&*. This turned out to be a rather effective
particulate filter.
. . j j„ the new Method 25 was developed by
The condensate trap proposed in , ...
G.B. Howe et al.1 as a secondary trap to aid in the removal of C0$ prior
to analysis. It was very successful for the
ccs. " r..s:
i" .Si b, the LO. county APCB In 1970. th...
I T . " " , with plugging when used on afterburner outlets.
£P'trh.;d. tta 25 »r. to hMdl. .
great deal of water without plugging.
The sample collection tank's recommended 4 liter i. too small. The
optimum based on experience gained from testing over 1,000 sources, is
639
-------
between 8 and 12 liters. The sample volume doesn't matter much when it
comes to measuring source levels of COj, CH4 or CO; but it matters a great
deal when measuring NMOC especially at relatively low levels. Unpublished
data produced by the Los Angeles County APCD in 1970 and 1971 showed that
the original Method 25 trap produced good precision when measuring NMOC at
ambient levels when a 50 liter &le was taken. A 2 liter sample
produced erratic results even with stack samples.
The proposed Method 25 specifies the use of crushed dry ice. This can be
used, with care, in an emergency. Pelletized dry ice is far better.
Crushed dry ice packs into a hard "snowball" and evaporates away from the
surface it is supposed to be cooling. The field technician must
constantly breakup the ball with a screwdriver to maintain the proper trap
temperature. Pelletized dry ice doesn't pack like this and requires
relatively little attention.
While it is not mentioned in the proposed Method 25, if the valves are
sealed around the valve stem with a packing should be mounted "backwards".
The flow arrow should point away from the tank so that the integrity of
the sample in transit doesn't depend on the valve packing.
The sample traps should not be flushed into the original sample tank.
This is likely to cause more problems than it is worth. Zn the original
Method 25 system, the loss amounts to less than 1% and the extra handling
is likely to comprise the integrity of the sample. Source test
engineering calculations usually use the CO2 concentration. A small leak
around the fitting in the trap (which is stored in dry ice) often results
in the trap containing nearly pure CO2. If this is flushed into the tank,
it can cause a considerable error.
The condensate trap should not be warmed slowly to ambient temperature
when starting to burn the trap contents. This wastes time and serves no
useful purpose. If the oxidizer is working properly, it can handle the
trap being dipped directly into a furnace. The long wait suggested in the
new Method 25 also requires that more gas be collected, which dilutes the
sample and reduces sensitivity.
The condensate trap must be heated to a temperature several times higher
than the recommended 200°C. Also the trap must have pure oxygen run
through it near the end of the collection cycle. Pyrolysis of the trap
contents is likely when heating the trap enough to remove the less
volatile compounds which experience shows are collected. Pure oxygen
burnt out the carbon produced by the pyrolysis.
An oxidation catalyst is not necessary and actually slows the process. We
found that alumina retards C02, causing apparent tailing on the combustion
process. Crushed quartz, 6 to 8 mesh run at temperatures over 750°C will
give complete oxidation of methane, and therefore, most other organics.
It doesn't cause tailing and allows the trap to be processed much more
rapidly.
Frequent monitoring of the CO2 levels by extracting sample from the
combustion system' is not necessary, since the COg monitor is usually set
to be sensitive enough for this purpose. The recommended monitor is labor
intensive and the extra sample handling increases chances for error.
The proposed Method 25 controls the rate of collection into the
intermediate collection vessel with a flow control valve. The flow
control valve is just a needle valve which must be adjusted manually
640
-------
throughout the burn. The Los Angeles County APCD discarded this approach
in 1971. The needle valve was replaced by an upstream pressure regulator
which is adjusted to maintain atmospheric pressure upstream of the
regulator. It does this with variations under 1 Torr throughout the burn
no matter what flow rates are used. ^
The intermediate collection vessel should be as small as possible to
attain maximum sensitivity. The suggested 6 liters is much too large for
most samples and results in a factor of three loss in sensitivity. We
have found that 2 liters is more than adequate for over 90% of the
samples. In those rare cases that the 2 liter is not big enough, one can
easily switch to another container to finish collecting the sample.
The intermediate collection vessel is also pressurized too much. A final
pressure of 860 Torr provides enough gas to run any analyses, while
pressurizing to 1060 Torr merely dilutes the sample, decreasing
sensitivity.
The proposed Method 25 requires far too many replicate analyses. It
requires that each analysis be done in triplicate. If the samples are
taken in duplicate, you end up with 12 analytical runs plus the burn
progress runs. This totals at least 15 analyses per sample location. The
original method required only four analyses, plus occasional replicates
for quality assurance. We feel that this is sufficient.
REFERENCES
1. G.B. Howe et al.
641
-------
THE PYROLYSIS OF ACETYLENE-STYRENE MIXTURES BETWEEN 450 AND 550*C
C. Chanmugathas and Julian Heicklen
Department of Chemistry and
Environmental Reaources Research Institute
The Pennsylvania State University
University Park, PA 16802
The pyrolysis of acetylene-styrene mixtures has been studied from
450-550*C in a quartz reaction vessel in the absence and presence of O2 or
NO. The rates of disappearance of reactants and formation of adducts are
first-order in each reactant. The major product is polymer with the adducts
accounting for about 2.5* and 6.2X of the styrene removed at 450 and 550"C,
respectively. The acetylene-to-styrene removal ratio is about 27
independent of temperature. The adducts formed are methyl iridene and 1,2-
dihydronapthalene. These are about half-suppressed in the presence of 02 or
NO. The rate coefficients for reactant removal and adduct formation are:
log{k{C2H2}, Af»8-M = 7.53 ± 0.10 - (90.6 ± 1.5)/2.3RT
log{k{CaH8}, /Vis"1} = 6.63 ± 0.60 - (98.5 ± 8.8)/2.3RT
log{k{CioHio}, AMs-1} = 8.27 ± 0.66 - (143.3 ± 9.8)/2.3RT
where the activation energies are in kJ/mole and the uncertainties are one
standard deviation. As the reaction proceeds, the methyl indene and 1,2-
dihydronaphthalene decompose, and indene and naphthalene are formed.
642
-------
Introduction
Pyrolysis or incomplete combustion of hydrocarbons leads to soot form-
ation. The main route to soot formation at temperatures below 1200*C is the
cracking of the hydrocarbons to CzHz which then polymerizes and dehydro-
genates. This process involves aromatic compounds as intermediates (1-3).
In order to investigate the low-temperature pyrolysis of styrene- acetylene
mixture as an avenue for soot or polynuclear aromatic hydrocarbon
formation, the present study was undertaken, paying special attention to
CioHio formation during the early stages of reaction.
Experimental
The experimental apparatus and procedure were similar to those de-
scribed earlier (4-6). Pyrolysis was carried out in a 1090-ca3 spherical
quartz vessel surrounded by a cylindrical resistively-heated oven. Kinetic
data were obtained by both mass spectrometry and gas chromatography.
The mass spectrometric sampling system consisted of a quartz pinhole
(30-jim diameter) through which the reaction vessel contents continuously
bled through a differentially-pumped intermediate region to a modified EAI
Quad 160 mass spectrometer. For quantitative analyses, a small measured
amount of krypton gas was added to the reaction mixture as an internal
standard to minimize the effect of instrumental and sampling fluctuations.
The absolute sum of the products with parent masses at m/z 130 (methyl
indene and 1,2-dihydronaphthalene) could be measured because they have the
same mass spectral sensitivity at m/z 130.
Gas chromatography was used to obtain kinetic data for individual
product isomers. The products were analyzed on a flame-ionization gas
chromatograph (Varian Series 1200) using a 6-ft, 1/8-in stainless steel
column packed with 1QX Silar 10C on chromosorb W/HP. The column was kept at
40*C for 6 min and then temperature programed to 170* at 4'C/min. The He
carrier gas flow rate was 29 ml/min.
When acetylene-styrene mixtures are pyrolysed between 450 and 5B0*C,
there are six heavier products formed in addition to polymer, which
accumulates on the wall of the reaction vessel. In order to identify the
products, gas-chromatograph-resolved mass spectra were obtained. The mass
spectra and gas-chramatographic retention times indicated that PI, P2, P3,
and P5 were respectively indene, methyl indene, 1,2-dihydronaphthalene and
naphthalene.
The minor products PO and P4 were respectively, p-methylstyrene and an
unidentified isomer of naphthalene (mol. wt. = 128). At 560*C traces of
other compounds, including indane (mol. wt. a 188), methylnaphthalene (mol.
Results
U
PI
P2
643
-------
wt. = 142), methyl-substituted dihydronaphthalene (mol. wt. = 144),
xylenes, toluene, benzene, and a product with mol. wt. = 154 were also
seen.
The initial rates of CioHio formation are plotted versus the products
of the reactant concentrations in Figure 1. The log-log plots are linear
with slope 1.0 indicating a second order reaction. Similar plots are
obtained for the initial removal rates of C2H2 and CeHa. The intercepts of
the plots give the rate coefficients, and they are plotted in Figure 2 in
Arrhenius plots. The Arrhenius parameters are listed in Table 1.
Both P2 and P3 formation are partially suppressed by the addition of
NO or O2. Thus both products are formed in part from a concerted molecular
process and in part by a diradical path. Since O2 and NO have comparable
effects, the diradical intermediate is probably a triplet, since Oa may not
scavenge a singlet diradical.
Discussion
The initial products of the reaction are polymer and adducts of
acetylene and styrene. By far, the dominant product is polymer with a ratio
of acetylene-to-styrene of about 27 independent of temperature. The
fraction of styrene consumed which gives CioHio adduct is 2.5% at 450*C and
increases to 6.2% at 550*C.
The reactions producing these products are homogeneous second-order
reactions. The homogeneity is indicated because the rate coefficients have
normal Arrhenius A factors for a second-order addition to produce CioHio or
a radical-chain mechanism to produce polymer. Furthermore, the addition of
excess inert gases did not effect the reaction* Also, Lundgard and Heicklen
(4) have shown that increasing the surface-to-volume ratio by a factor of
59 had no effect on the diaerization of vinylacetylene and only a minimal
effect on its polymerization.
The Arrhenius A factors found for both polymer and adduct formation
are similar to those found for the corresponding reactions in other systems
(4-6), but the activation energies found here for both processes are about
20 kJ/mole higher than for the analogous processes in the other systems.
The secondary products, indene (PI) and naphthalene (P4) come from
decomposition of the primary products methylindene (P2) and 1,2-
dihydronaphthalene (P3), respectively. The methylindene decomposition
presumably proceeds by loss of a methyl group, a conclusion supported by
the observation that p-methyl styrene (P0) was found. P3 decomposition
probably proceeds by loss of Hi or 2 H atoms though we have no information
on the details of the decomposition.
Polym&r formation probably proceeds via a mono-radical chain process.
Initiation could proceed either by self-reaction of styrene or acetylene.
644
-------
References
1. H. B. Palmer and C. F. Cullis, "The Cheaistry and Physics of Carbon,"
Vol. 1, P.L. Walker, Ed., Marcel Dekker, New York, 1966, Chap. 5.
2. M. S. B. Munson and R. C. Anderson, Carbon 1: 51 (1963).
3. B. S. Haynes and H. Q. Wagner, Prog. Energy Coabuat. Sci. 7i 229 (1981).
4. R. Lungard and J. Heicklen, Int. J. Chen. Kinet. 16: 125 (1984).
5. c. Chamugathas and J. Heicklen, Int. J. Chest. Kinet. 17: 871 (1985).
6. c. Chanmugathas and J. Heicklen, Int. J. Chen. Kinet. 18: 701 (1986).
Table I
Arrhenius Paraaeters for Second-Order Rate Coefficients
Rate coefficients
Em. kJ/aole
logU. Af1 sec1)
k{CaHa}
k{CaHa}
k{CioHxo)
90.6 ± 1.5
98.6 ± 8.8
143.3 ± 9.8
7.53 ± 0.10
6.63 ± 0.60
8.27 ± 0.66
645
-------
I I I ' I 1 11 ' I—I—1 I I I 11 1 1—I—I III I
10"® I0"8 10"7 10"
CCj, H23 CSTY3 ( M2
1: Log-log plots of initial rates of Ci0H10 formation obtained by
mass spectrometry versus [CaHa ] [CeHa ]: 0, 450*C; A , 475*C; 9,
500*C; O , S25*C, d, 550*C. Filled points are for Nz present.
646
-------
550
T , °C
500 475
450
425
tCjhy/IOOO.
CW
CC8He2/IOO _
1.40 1.44
I0S/ T( K*"1
Fig. 2: Arrhenius plots for the rate coefficients: A, Cafe removal; 0,
atyrene removal; o, CioHio fornation.
$47
-------
ASBESTOS FIBER LOBS FROM AIR SAMPLING CASSETTES-.
A STUDY BY TRANSMISSION ELECTRON MICROSCOPY
William E* Lcngo & E.J. Jenkins
Micro Analytical Laboratories, Inc.
Gainesville, Florida
The analysis of air samples for asbestos by Transmission Electron
Microscopy (ISM) has increasingly beoone the method of choice for final
clearance samples in post-abatement projects. However, the use of both
polycarbonate (PC) and mixed cellulose ester (MCE) filter cassettes daring
air sampling has raised some concerns about possible fiber loss from the
filter surface due to both Btatic charge problems and rough handling daring
sample shipment to the TEH laboratory. Also, it has been suggested, that
their nay be substantial loss of fibers to the inside of the cowling walls
on the cassette during the actual air sampling process.
This investigation found that all of the air samples examined displayed
fiber loss to the inside cassette walls. This loss was demonstrated for
both the PC and MCE filters. The amount of fiber loss was front 9% up to
over 80% and was seen for all types of sampling cassettes including the new
blade conductive cassettes.
A new washing and re-filtration method was studied that has the
potential to overcome the problems of post-fiber loss after sample
collection. Briefly, the cassette walls and filter surface are completely
washed to remove all of the fibers present and than re-filtered onto a MCE
filter (0.22 mi croc pore size). Hie direct transfer method is used for HM
sample preparation.
648
-------
Introduction
Phase contrast microscopy (PCM) is now the most widely used technique
for the detection of asbestos on air filter samples. The most common uses
for PCM are; (1) general air surveys, (2) air monitoring during abatement
projects and (3) final air clearance samples after abatement. However,
previous studies have shown that PCM is inadequate for both the general air
surveys and final air clearance because it lades specificity for asbestos
and cannot resolve fibers less then 0.25 microns in width.1'2 TEM is the
only technique that now has the capability of positively identifying all
types of asbestos including structures that are less then 0.25 microns wide.
However, economics and slow turnaround times has hindered wide spread
acceptance of T5W analysis.
Proposed Environmental Protection Agency (EPA) regulations that my go
into effect later this year will ban the use of PCM for final air clearance
in school abatement projects. proposed analytical method for final
clearance will be TEM. If these EPA regulations becane law, the use of 101
for final air clearance will probably become the industry standard and gain
a nuch wider acceptance. Since there are only a few qualified IB*
laboratories in the country, moet air filter samples will have to be shipped
to the laboratory for analysis because in moet cases, the logistics of hand
carrying sample cassettes across the country will just not be feasible.
There has been a growing concern that the shipment of the sample
cassettes to the TEM laboratories may cause fiber loss from the filter
surface to the inside of the plastic cowling because of rough handling or
static charge problems, if fiber loss was substantial, it wauld cause
erroneous results from the OEM analysis, allowing in the worst case, the re-
occupation of a building with unsafe asbestos levels.
We became concerned that the actual fiber loss from the filter surface
during sample shipment was probably nuch higher then was oanmanly
speculated. Since to our knowledge, there have not been any reported
studies that actually quantitate fiber loss during shipment using TEH, we
decided to investigate this possibility. This paper describes the
comparative analysis of both the filter surface and cowling walls for
airborne asbestos by TEM. Sample cassettes were examined that wore both
shipped and hand carried to cur facility. Both the clear plastic sampling
cassettes and the new black conductive cassettes were studied. Also, since
fiber loes would be restricted to the inside of the cassette walls, we
studied an alternative sample preparation process that used a washing
technique for recovery of all loss fibers.
Experimental Methods
Air cassettes were chosen for this study that had elevated levels of
asbestos (greater then 0.02 asbestos structuree/oc) when previously analysed
by TEM using current EPA methodology.3'4 From these cassettes, the inside
filter and support pad was carefully removed and archived. The inside
surface of the cassettes was washed with distilled water that contained
0.01% or aerosol solution. The washing process consisted of filling the
cassette with 10-15 ml of the prepared solution, capping the cassette
tightly to prevent water leakage, then shaking vigorously for 20-30 seconds.
?)» wash solution is decanted into a 100 ml beaker and saved, this step is
649
-------
repeated two additional times. The re-filtration procedure for the wash is
based on the EPA "Interim Method for Determining Asbestos in Water". Hie
combined wash solution is then re-filtered onto a 25 ran MCE filter with 0.22
micron pore size and prepared for TEM analysis as described by Burdette and
Rood. Both the (MCE) and the PC filters ware examined. Hie concentration
of the asbestos found on the inside of the cowling walls was compared to the
results obtained from the filter surface only. Air sample cassettes
containing either the MCE or PC filters were run side-by-side during an
asbestos abatement project. Hie filters inside the cassettes were left in
place during the washing procedure as described above. The wash solution
was then re-filtered onto 25 rmi MCE (0.44 micron pore size) filters and
prepared for TEM analysis. The washed filters were dried at 50 C for 1.0 hr
and also prepared for TEM analysis. Results of the two washed filter
surface types were compared.
Results
Shown in T&ble I is the concentration of the amount of asbestos that was
found on the cowling walls for the PC sampling air cassettes. The results
for the air cassettes containing the MCE filters are shewn in liable II. Hie
amount of asbestos found on the inside of the cowling walls of these samples
ranged from 0.033 to 15.150 structures/oc. This represented a total fiber
loss from 9% up to 88% that normally would have been missed in the analysis.
The MCE filters had an average loss of 54% and for the EC filters, the
average loss was 64%. Also, when the cassettes were hand carried to cur
facility, they demonstrated a fiber loss of almost 50% when the cowling
walls were examined.
Tables III & IV show the asbestos removal efficiency of both the MCE and
the PC filters that ware washed with the 0.01% OT aerosol solution.
Throughout this study, we were consistently able to thoroughly wash the MCE
filters while the PC filters were always found to contain sane residual
concentration of asbestos fibers.
Conclusions
All the sample cassettes that we tested at our facility shewed extensive
fiber loss to the inside ocwling walls when examined by TEM. When the
amount of fiber loss was compared between the MCE and the PC filters, the
differences ware minimal. This suggest that for shipping purposes, neither
filter type has an advantage.
The amount of loss from sample to sample was variable tfcich limits the
use of a standard error factor that can be put into the fiber concentration
calculations that are used for these types o£ analysis.
Since the hand carried cassettes also shewed fiber loss it's uncertain at
this tints whether or not the fiber loss is directly attributed to the
shipping or a static charge problem causing the fibers to stick to the
cowling walls during the actual sampling. However, when the anti-static
cassettes were tested, they also showed a high concentration of fiber loss
as shown in Table I. m any case, more experimentation is needed to resolve
the question concerning the source of the fiber loss.
650
-------
Using the proposed washing re-filtration method can overcame both the
shipping problems with rough handling and the suspected fiber loss from
static charge build up that occurs during the actual air sampling process.
This method would allow for the analysis of all the fibers inside of the
cassette during the washing process and not just the fibers found an the
filter. Future work is needed so that a larger statistical population is
examined for a mare valid analysis and the question of possible break up of
bundles and clumps during the washing process also needs to be addressed.
Acknowledgments
We are indebted to both Dr. Baxter at ENTER Environmental Services,
Inc. for his technical assistance and for initially suggesting the use of
the water re-filtration method and Mr. Robert Greene at Law Engineering
Testing Co. for providing us with samples for this study. We would also
like to thank Mrs. Joanne Ake for her help in preparing this manuscript.
References
1. D. A. Falgout, "Environmental Release of Asbestos frcm Commercial
Product Shaping", Office of Research and Development, U.S. Environmental
Protection Agency, EPA/600/S2-85/044, (Aug. 1985).
2. D. L. Keys, M. Beard, J. Breen, "Measuring Airborne Asbestos Following
An Abatement Action",, Office of .Research and Development, U.S.
Environmental Protection Agency, EPA/600/4-85-049, (Nov. 1985).
3. G. Yamate, S. C. Argarwal, R. d. Gibbions "Methodology for the
Measurement of Airborne Asbestos by Electron Microscopy", Draft Report.
Office of Research and Development, U.S. Environmental Protection
Agency, 68-02-3266, (JUly 1984).
4. g. J. Burdette, A. P. Rood "Membrane filter, direct transfer technique
for the analysis of asbestos fibers and other inorganic particulates by
transmission electron microscopy," Envir. Sci. Tectool. 17i 643
(1983). '
5. E. J, Chatfield, M. J. Dillion "Analytical Method for Determination of
Asbestos in Water", office of Research and Development, U.S.
Environmental protection Agency, EPA/600/4-83-043, (Sept. 1983).
651
-------
Table I.
Fiber loss on
inside of cassette walls; Polycarbonate filters.
Sample no.
filter size
Structures/cc an
Structures/cc found
«Fiber
filter surface
on cassette wall
loss
1
37nm
1.372
2.484
64
2
37nm
9.121
15.150
63
3
37nm
3.642
8.486
70
4
37nm
6.325
9.616
60
liable II.
Fiber loss on inside of cassette wall; Mixed cellulose ester.
Sample no.
Filter size
Structures/cc on
Structures/cc found
%Fiber
filter surface
on cassette wall
loss
1
37nm
3.574
4.689
51
2
37nm
16.005
1.604
9
3
37nm
0.016
0.118
88
4*
25nm
0.038
0.034
47
5*
25nm
0.039
0.033
46
6+
25ira>
0.940
3.712
80
7+
25mn
1.304
1.673
56
* Samples
hard carried to
our facility.
+ Cassettes constructed cut of black conductive cowling.
Table ZXI.
Fiber removal efficiency after woshi Polycarbonate filters.
Sample no.
Filter size
Structures/cc on
Structures/oc cn
IRemoval
filter before wash
filter after wash
1
37nm
1.372
0.055
97
2
37am
9.121
0.592
94
3
37nm
12.968
0.683
95
Hable IV.
Fiber removal efficiency after wash; Methylcelluloee
ester filters.
Sample no.
Filter size
Structures/cc on
Structures/oc an
•Removal
filter before wash
filter after wash
1
37rrm
3.574
0.00*
100
2
37nrn
16.005
0.00
100
3
37mm'
0.016
0.00
100
* The concentration of the asbestos present was belcw detection limit.
652
-------
A Novel. 2.5 |im Cut Virtual Impactor for
High Volume PMjq Sampling
Robert M. Burton
Environmental Monitoring Systems Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Virgil A. Marple, Ph.D., Benjamin Y.H. Liu, Ph.D.
and Bradley Johnson, Particle Technology Laboratory
University of Minnesota, Minneapolis, MN
A high volume 40 CFM PMjo Virtual Impactor for collecting large amounts
of particulate matter separated into two distinct fractions (FINE, 0 to 2.5
|xm; and COARSE, 2.5 to 10 |un) has been developed for use in the US Environ-
mental Protection Agency's Integrated Air Cancer Project (IACP).1 Because
studies have shown, in general, that the highest levels of carcinogenic/nu-
tagenic potential associated with ambient particulate aerosols are from
organic compounds found in the FINE fraction,* 3 a requirement for separating
and collecting large amounts of particulate matter below 2.5 microns became
evident as mutagenicity testing progressed in the program. Particulate
emissions from woodstoves, vehicles, and many industrial sources are found
in this size range and can be efficiently separated and collected with the
virtual impactor.
The IACP is a long-term research program with the goal of identifying
carcinogens in ambient air, the sources of these carcinogens, and the associ-
ated potential health risk. As the IACP began its initial phase concerned
partly with the monitoring of residential woodstove emissions, the low
flowrate (16 liter/minute) dichotomous sampler was the only commercially
available collector that would provide separation and collection by filtra-
tion of both the FINE and COARSE fractions of aerosol.4 The IACP required a
method for collecting sufficient amounts of the organic rich FINE fraction
aerosol to be used in chemical analysis and bioassay testing. Consequently,
a 40 CFM virtual impactor was chosen and modified to meet the needs of the
IACP, Emphasis was placed on developing a precise, high flow rate simpler
that could be utilized cost effectively as an add-on to the existing PMiq
Siae Selective Inlet high volume samplers.
The sampler has been tested extensively in the laboratory to define
its collection characteristics and wall losses. Wall losses overall are
1.7? for liquid particles and 0.8J for solid ones. Although the sampler
was designed for a specific application (IACP), it should have wide usage
in the sampling of aerosols where large quantities of fractionated fine
fraction aerosol is desired.
653
-------
A Novel 2.5 p.m Cut Virtual Impactor for High Volume PMiq Sampling
by
R. M. Burton
Environmental Monitoring Systems Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Virgil A. Marple, Ph.D, Benjamin Y.H. Liu, Ph.D.,
and Bradley Johnson, Particle Technology Laboratory
University of Minnesota, Minneapolis, MI
INTRODUCTION
A high volume 40 CFM PMiq Virtual Impactor for collecting large amounts
of particulate matter separated into two distinct fractions (FINE, 0 to 2.5
Ura; and COARSE, 2.5 to 10 (im) has been developed for use in the US Environ-
mental Protection Agency's Integrated Air Cancer Project (IACP).l Because
studies have shown, in general, that the highest levels of carcinogenic
potential associated with ambient particulate aerosols are from organic
compounds found in the FINE fraction,^ 3 an IACP requirement for separating
and collecting large amounts of particulate matter <2.5 microns aerodynamic
diameter became evident as mutagenicity testing progressed in the program.
Particulate emissions from woodstoves, vehicles, and many industrial sources
are found in this size range and can be efficiently separated and collected
with the virtual impactor.
The IACP is a long-term research program with the goal of identifying
'carcinogens in ambient air, the sources of these carcinogens, and the
associated potential health risk. As the IACP began its initial phase
concerned partly with the monitoring of residential woodstove emissions,
the low flowrate (16.7 liter/minute) dichotomous sampler was the only
commercially available collector that would provide separation and collec-
tion by filtration of both the FINE and COARSE fractions of aerosol.^ ^
Other available separators required greasing the collection surfaces which
could potentially contaminate the collected aerosol. The IACP required a
method for collecting sufficient amounts of the organic rich FINE fraction
aerosol to be used in chemical analyses and bioassay testing. Consequently,
a 40 CFM virtual impactor design was chosen and modified to meet the need
of the IACP. Emphasis was placed on developing a precise, high flow rate
sampler that could be utilized cost effectively as an add-on to the existing
PMiq Size Selective Inlet high volume samplers.
EXPERIMENTAL METHOD
The Virtual Impactor Classifier
The principle of operation for the virtual impactor is quite simple.
Particle laden air is accelerated through a nozzle impinging into a receiv-
ing tube that is slightly larger in diameter than the nozzle. The particles
which are larger than the cut size of the virtual impactor have enough iner-
tia to penetrate into the receiving tube and are extracted from the backside
with a small amount of flow (minor flow), which is usually 5 to 10% of the
flow entering the receiving tube. The remaining portion of the flow contain-
ing particles smaller than the cut size (particles with low inertia) reverses
direction in the receiving tube and exits at the outer edges of the entrance
654
-------
to the tube. In this manner, particles both smaller and larger than the
cut size of the classifier are kept airborne and can be collected by filtra-
tion.
The collection efficiency of a virtual impactor is a function of parti-
cle size and is defined as the fraction of particles passing through the
nozzle that penetrates through the receiving tube with the large particle
sample in the minor flow. Most of the smaller particles that do not have
enough inertia to be separated in the receiving tube (those smaller than
the cut size) remain with the major air flow. However, a fraction of the
small particles equal to the fraction of the minor flow penetrate the receiv-
ing tube with the large particles. Therefore, the minimum efficiency of
any particle size is the percentage of the total flow passed through the
receiving tube. This efficiency can be increased by introducing a clean
core of air in the center of the jet providing a small, particle free minor
flow. The addition of the clean air jet, however, does complicate the
design and operation of the virtual impactor.
To achieve 40 CFM flowrate without creating an unsatisfactory pres-
sure drop across the separator, a 12-nozzle design was used. Half of the
virtual impactor stages are located ori each side of the unit with the minor
flows exiting from the receiving tubes into a central common chamber, the
floor of which is the large particle filter. The major flow exits through
two cavities (one on each side of the central chamber) and flows downward
into a common cavity below the large particle filter chamber and above the
small particle filter. The design is shown in Figure 1.
High Volume Virtual Impactor
(top view section)
ACCELERATION ROZZLI
Figure 1
The virtual impactor stages are paired on a common axis so that the
minor flows exiting from the receiving Cubes imping® on each other, aiding
in stopping the large particles before they impact on the opposite wall.
In this manner, the losses of the large particles are minimized. Position-*
ing the large particle filter on the floor of Che chamber also aids in
reducing losses since almost no transport of the minor flow is necessary
before the particles are filtered from the airstream. In a similar manner,
there is little transport distance between the virtual impactor classifier
«nd the filter for the small particles.
655
-------
The High Volume Virtual Impactor as an Accessory to the PMm High Volume
Sampler
The impactor (twelve-nozzle unit shown in Figure 2) has two components:
a classification head assembly and a filter holder assembly. The classifi-
cation head assembly contains the twelve virtual impactor stages and cavi-
ties leading to the large particle and small particle filters. The filter
holder assembly is a casting containing both the small and large particle
filters. The filter holder section is designed in such a manner that the
filters can be assembled in the laboratory, transported to the field test
site, inserted for a run and then returned to the laboratory for disassembly.
In normal operation at least two filter holder assemblies would be used with
one classification head so that the filters can be changed in the laboratory
while the other unit is being used at a field site.
In the high volume sampler the impactor is located between the FM^q inlet
and the high volume blower with the flange of the impactor filter holder
located between the flanges of the PMjq inlet and the flange of the 8 x 10-
inch entrance to the blower (Fig. 2). Because the particles larger than
10 microns have been removed by the PMjq inlet, the particles passing from
the PMiq inlet into the cavity surrounding the impactor will not have enough
inertia to be impacted on the head of the impactor. With this arrangement,
the overall height of the high volume sampler is only increased by about
one inch, which is the thickness of the flange and gaskets.
RESULTS
Calibration
The high volume virtual impactors were calibrated using monodispersed
oleic acid and ammonium fluorescene particles generated with a vibrating
orifice monodispersed aerosol generator (VOMAG)(Berglund and Liu, 1573).
The particles are tagged with a small amount of uranine dye tracer. Parti-
cles of a particular size are passed through the high volume virtual impac-
tor and are collected on either the large or small particle filters or lost
to one of the surfaces inside the impactor assembly. The quantity of
particles being collected on the filters or lost can be determined by
washing the respective surfaces with known amounts of water and measuring
the dye concentration in the wash water with a fluorometer. By analyzing
where the particles are located, the collection efficiency can be determined
as well as the losses and exactly where the losses occur. The collection
efficiency curve for the impactors is shown in Figure 3.
Also shown in Figure 3 is the collection efficiency for the 16.7 Jtpm
dichotomous sampler. Note that the agreement between the two samplers is
very good. Calibration with solid ammonium fluoresceine particles showed
the same collection efficiency as with the liquid particles.
Losses, i.e. particles which have been collected at some point other
than on either of the two filters as a functional particle size, are shown In
Figure 4 for the liquid particles. As expected, the losses are the largest
for particles near.the cut size (2.5 micron) of the impactor and decrease
rapidly for smaller and larger particles * Solid particles have lower
losses than liquid particles due to their ability to bounce from a surface
and be reentrained into the airstream. Since atmospheric aerosols are
generally considered to be solid particles, the lots during ambient sampling
will be similar to that determined by the solid calibration particles.
656
-------
High Volume Virtual Impact or
(lateral taction view)
**** MtlR
uh turn
ITiTITIt
wuiioa mama^S
runt
-------
walls surrounding Che two Bide cavities. If these cavities were larger,
the losses would be reduced, possibly by as much &b 50%, but then the high
volume virtual impactor would not be as compact.
Conclusions
A 2.5 |iin diameter cut high volume virtual impactor consisting of twelve
parallel virtual impactor stages has been successfully designed for applica-
tion with a high volume PMjq inlet. Use of parallel flow virtual impactor
stages and paired opposing receiving tubes has allowed for a design which is
very compact with losses not exceeding 1.7% and 0.8% for liquid and solid
particles, respectively. Since the sampler classifies particles in size
fractions of 0-2.5 [im and 2.3 jim - 10 nm, respectively, the sampler incorpo-
rates two filters: one for the small particles (8 x 10-inch) and one for the
large particles (2 x 6-inch), These filters are contained in a single
filter holder which can be easily transported to and from a laboratory for
filter replacement. Although the sampler is capable of passing 40 cfm of
air flow, it only increases the height of the high volume sampler by approx-
imately one inch. As an option, the virtual impactor can be operated
without a PMio inlet, still providing a 0-2.5 micron fraction pluB a fraction
> 2,5 microns.
The sampler was designed for a specific application in the Integrated
Air Cancer Project; however, it should prove useful in a variety of aerosol
sampling situations where large quantities of fractionated FINE fraction
aerosol is desired. This sampler is advantageous for collecting large
quantities of ambient aerosol (from woodstoves, motor vehicles, etc.) in
the 2.5 uro or less particle size range, for both chemical and biological
analysis.
REFERENCES
1. U.S. EPA. Clean Air Act, Section 112.
2. Huisingh, J. Lewtas. "Bioassay of Particulate Organic Matter From
Ambient Air." From Waters, M.D., et al, Short-term Bioassays in the
Analysis of Complex Environmental Mixtures 11. Plenum Publishing
Corp. (1981).
3. Lewtas, J. Bioassay directed fractionation/characterisation studies
to identify genotoxic compounds in ambient air particles from Washing-
ton, DC, and Philadelphia, PA. A collaborative study. Presented at
conference and workshop on Genotoxic Air Pollutants, Durham, NC, April
24-27 {1984).
4. Loo, B. W., Jaklevic, J. M., and Gouldiag, F. S. "Dichotomous Virtual
Impactors for Large Scale Monitoring of Airborne Particulate Matter."
Lawrence Berkeley Laboratory Report Ko. LBL-3854 (1975).
5. Dzubsy, T.G., and Stevens, R. K. "Ambient Air Analysis with the
Dichotomous Sampler and X-ray Fluorescence Spectrometer." Envir. Sci.
& Tech. 9, No. 7;663 (1975).
6S8
-------
DEVELOPMENT OF EPA'S
AIR TOXICS CONTROL TECHNOLOGY CENTER
Sharon L. Nolen, Gary L. Johnson, and James A. Dorsey
U.S.Environmental Protection Agency
A1r and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
and
Lee L. Beck and John D. Crenshaw
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, NC 27711
ABSTRACT
The A1r Toxics Control Technology Center (CTC) Is a joint effort of
the EPA Office of Research and Development (ORD) and Office of Air and
Radiation (OAR) to provide technical assistance to State and local air
quality regulatory agencies In solving control technology problems related
to air toxics. The CTC will support a wide range of State and local
mitigation activities, including high-risk point source programs (State
initiatives and EPA promoted Initiatives), a high-risk urban toxics
program, State Implementation Plan (SIP) decisions also affecting air
toxics emissions, and Best Available Control Technology (BACT) deter-
minations as they pertain to air toxics control. This program requires
the coordination of research and program office expertise 1n order to
respond effectively to State and local requests for assistance. The CTC
activity in 1987 1s focused on a pilot program to determine and document
the achieved effectiveness of the technical assistance program. This
effort features Implementation of the CTC HOTLINE, a telephone number
which can be used by State and local personnel to gain Immediate access
to EPA expertise In control technology.
The development of the CTC Is described 1n terms of the roles of the
participating organizations, how the CTC 1s operated and requests are
handled, and what types of requests are appropriate for the CTC. In addi-
tion, experience to date with the CTC operation 1s discussed and future
plans for the Center are outlined. Initial response by the State and
local agencies has been encouraging, and both OAR and ORD are optimistic
that the CTC will become a very active part of the Agency's air toxics
program in the years to come.
659
-------
INTRODUCTION
The term "hazardous air pollutants," or more recently, "air toxfcs,"
encompasses a potentially large number of pollutants. Air toxics may be
defined as virtually any substance released Into air media which may
pose an exposure risk to humans. The sources of air toxics are widely
varied and Include traditional air pollution sources such as chemical
plants, metallurgical processes, and motor vehicles as well as non-trad-
ftional sources such as passive tobacco smoke. Indoor radon from soil
gas, and sewage treatment plants.
While existing control strategies are making a positive Impact on
reducing ambient air toxics levels, the growth of the national economy
and a continued dependence on chemical and petroleum products could
worsen the ambient problem and lead to Increased public health risks. In
addition, much of the current and emerging air toxics risk, for example
from home heating and gasoline vapor control, may not be significantly
reducible 1n the near future at costs which society 1s willing to pay.
To address the near-term problem, the Environmental Protection Agency
(EPA) announced In June 1985 a strategy, which encompasses both routine
and accidental release concerns, for a program to reduce public exposure
to toxic air pollutants 1n the ambient air.
The Air Toxics Strategy now being implemented couples the traditional
Federal regulatory programs [e.g., NESHAPs, Mew Source Performance Stan-
dards (NSPS)] with an Increasing State and local regulatory responsibility
to achieve reduced air toxics risk to the public. Over the next 5-8
years, the strategy calls for the States and local authorities to take
the lead regulatory role, with the Agency providing technical and financial
assistance to their efforts. EPA's Office of Air and Radiation (OAR) has
drafted a 5-year A1r Toxics Implementation Plan by which the objectives
of the strategy for routine releases would be accomplished, and which
could serve as a guide to the EPA's Office of Research and Development 1n
developing a more responsive RAD program 1n air toxics.
The Implementation Plan for control of routine releases of air
toxics defines six components to address the Agency's strategy:
o Federal Regulatory Program 1n A1r Toxics (NESHAPs, NSPS, etc.)
o High Risk Point Sources (e.g., non-NESHAP sources with cancer
risk exceeding 10~4 in local situations)
o High Risk Urban Toxics (focused on urban ambient environments)
o Enhancement of State and Local Air Toxics Programs
o Enforcement and Compliance Programs for A1r Toxics
o Management and Coordination of A1r Toxics Programs
As State and local air pollution authorities begin to develop local-
ized air toxics regulations, substantial financial and technical assistance
1s expected to be required. The enhancement of State and local air toxics
programs will provide financial assistance through grants (e.g., Section
105 of the Clean Air Act) and technical assistance through several Agency
660
-------
offices, Including the engineering R&D program. In particular, experi-
ence 1n control technology and source assessment resident in the EPA
will provide a prime source of technical expertise for the State and
local agencies as they develop their air toxics programs.
Development of Air Toxics Control Technology Center
The shift 1n emphasis and responsibility from the Federal level to
the State and local air toxics programs and the need to transfer
expertise from the Federal level to the appropriate level have prompted
EPA's Office of Air Quality Planning and Standards (OAQPS) and the Air and
Energy Engineering Research Laboratory (AEERL) to develop and Implement*
an Innovative technical assistance program. This program Is the A1r
Toxics Control Technology Center (CTC), a collaborative effort among
OAQPS, AEERL, and EPA's Center for Envlronmental Research Information
(CERI). It Is expected to become a principal element In the enhancement
program through the early 1990s. The goals of the Center are to
provide:
o Support to State and local agencies in the implementation of air
toxics control programs through technical guidance and support
on air pollution control technology questions,
o Mechanisms to transfer available engineering Information
of broad Interest to State and local authorities,
o "HOTLINE" telephone access to EPA expertise as an Initial
quick response to Individual problems,
o In-depth engineering assistance to States on Individual
problems, and
o Feedback to EPA on the technical support needs of State and
local agencies.
The types of services to be provided by the CTC can be broken into
three categories: direct technical support, technical guidance, and
requests for assistance. In the area of direct technical support, the
CTC will provide available Information, data, and engineering analysis on
an as-requested basis. Assistance can be provided on a range of topics.
Including:
o Evaluation of source emissions
o Identification of control technology alternatives
o Development of control costs
o Assessing Impacts of control technology on water, solid waste, or
air effluents
o Solving problems with source testing methods
o Advice on permit conditions to ensure good operation and
maintenance of equipment for air toxics control
o Providing expert testimony In support of State and local
regulatory actions.
661
-------
In tl^e area of technical guidance, the CTC win focus on topics of
national Interest that have been Identified by the States, Including:
o Publishing control technology documents which discuss
- Add-on control devices,
- Process modifications, and
- Control of area sources
o Designing and developing microcomputer software to help assess
air toxics control problems and evaluate potential alternative
solutions
o Conducting public seminars and workshops on control technology
pertaining to air toxics.
The CTC will handle requests for assistance through the CTC "HOTLINE."
Through a special telephone number, State and local staffs can contact
EPA personnel who are the most knowledgeable about the requested topic.
The CTC meiy not be able to resolve every technical issue that Is raised.
A primary purpose of the CTC is to provide a quick response based on
whatever control technology expertise or information is available within
the Agency. Usually, this will Involve a series of telephone conversations
and result In the compiling of existing Information from EPA technical
files.
In some cases, 1t may be appropriate to go beyond the rapid-response
level of support and perform additional engineering analyses. Such
requests should be made 1n writing by the requesting agency. Since a
more 1n-depth Involvement of Agency staff may be required here, contractor
resources m«y be used to augment EPA 1n-house support. Decisions on
whether or not the CTC can respond to such requests will depend on a
number of factors, Including public health Implications, usefulness to
other States, funding limitations, and other CTC priorities.
In Its first year, the CTC is being operated as a pilot program.
The CTC 1s a new concept for the air program: no attempt at such
broad-based, cooperative technical assistance has been tried before. At
this time, we have only a limited Indication of the response to this type
of support or of the support areas of most demand. The first year will
be used to determine the best approach to organize and operate the
CTC.
Currently, the CTC Is managed by a Steering Committee with represent-
atives from each participating organization. Plans are currently underway
to expand the membership of the committee by adding several advisory
members, representing Regional, State, and local agencies. In general,
the responsibilities of the Steering Committee Include;
o Assessing technical assistance needs of the States
o Planning activities of the CTC
o Setting priorities for spending CTC resources
o Authorizing decisions on allocation of CTC resources
662
-------
o Developing outreach activities
o Coordinating CTC activities with EPA Regional Offices
o Developing a plan for the permanent operation of the CTC
o Ensuring appropriate coordination of CTC publications or
products.
CTC operation began in the fall of 1986, and the response from the
States has been enthusiastic. A number of telephoned requests, received
through the "HOTLINE," have been addressed. A standard "HOTLINE" form '
has been developed to facilitate gathering the pertinent Information, and
to document the number and nature of Incoming requests to track EPA
response. Since the purpose of the CTC 1s to serve a specific client
audience, efforts have been made to Inform State and local agencies about
the Center and its operation. In October 1986, meetings were held with
members of the Air Toxics Committee of the State and Territorial Air
Pollution Program Administrators (STAPPA) and the Association of Local
Air Pollution Control Officials (ALAPCO) to describe the CTC concept and
to solicit their response to the program end requests for assistance.
Subsequent discussions were held with other STAPPA/ALAPCO members, and
helpful comments were received and factored into the CTC program.
EPA 1s very sensitive about the CTC's ability to respond to State
and local needs in a timely and useful way, Both OAQPS and AEERL have
committed to establishing a program that can satisfy both issues. While
procedures have been developed to expedite the processing of requests,
the proof of the concept lies in the progress actually being made. In
that regard, responses to four requests received in late 1986 Illustrate
the success of the program and its diversified nature.
Current Projects
One of the first projects being conducted by the CTC responds to a
request by the State of Florida regarding the emissions generated by the
air stripping of ground water. Since this problem seems to be of wide
concern, the CTC Steering Committee decided to respond to the request by
producing a guidance document on control of these emissions. Members of
the CTC organizations familiar with ground water stripping have agreed to
produce a document pooling the expertise on the subject from each organ-
ization. It Is expected that the document will be distributed to and
used by those with similar concerns.
Another request originated from the Northeast States for Coordinated
Air Use Management {HESCAUM). The NESCAUM States are faced with a defic-
iency of Information on sampling and analytical {S&A) procedures for
municipal waste Incineration, Although municipal waste fnclnerators are
of increasing concern to the Agency and standards are being considered,
Interim methods are needed to promote consistency among the States* In
this case, the CTC has arranged for the Agency's measurement expert 1n
incineration to assist the group In formulating an Interim sampling
protocol. EPA w1Tl meet with the States to reach a consensus on the SiA
procedures to be used. The results of the meetings will be compiled
for the States' use.
663
-------
The State of Colorado asked for help 1n addressing complaints of eye
and lung Irritation 1n an area surrounding a waferboard manufacturing
plant. The CTC has contacted similar plants to determine 1f the plants
have had a comparable problem and, 1f so, how they have addressed 1t.
The CTC will also conduct a site visit In order to assess the situation
first-hand. The visiting team will consist of EPA personnel and a
knowledgeable contractor representative. The CTC will then provide a
report to the State to characterize the emissions and suggest possible
control options.
A similar request came from Allegheny County, a local agency 1n
Pennsylvania. They were concerned with five plants In their district
which store toluene d1-1socyanate In above-ground containers. The
principal concern was the ability of the plant to deal with an accidental
release of TDI, which could threaten the local population. The CTC was
asked for assistance 1n verifying the effectiveness of the control systems
and techniques 1n place. Again, a team of experienced EPA and contractor
personnel was assembled to evaluate the available Information, visit two
plant sites, and provide a report to the local agency.
Means of Requesting Assistance
Requests for technical assistance are being received, and actions are
being taken on them. Our effectiveness will continue to Improve as we
learn more about the needs of the State and local agencies. Consequently,
we are soliciting State and local agency involvement In this endeavor by
asking that representatives of these groups contact the CTC with their
comments and/or requests for assistance. Formal requests for technical
assistance (e.g., plant Inspections, cost analysis, expert testimony)
should be submitted to:
A1r Toxics Control Technology Center
MD-13
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
or the CTC E-mail box: EPA8390. Informal technical assistance can be
accessed through the CTC HOTLINE. The HOTLINE number 1s (919)541-0800
or FTS 629-0800.
664
-------
Detection and Quantification of Hydroxylated Nitre Polynuclear Aromatic
Hydrocarbons and Ketones and Hydroxylated N1tro Aromatic Compounds 1n an
Ambient A1r Particulate Extract
M. G. Nishioka and D. A. Contos, Battelle Columbus Division
L. M. Ball, University of North Carolina
J. Lewtas, U.S. EPA
Highest percentages of mutagenic activity for urban air particulate extracts
are associated with the polar compound class fractions. Many of the
components of these fractions have not been Identified, and the mutagenic
species, 1n particular, have not been detected. The Identification of
hydroxylated nltro polynuclear aromatic hydrocarbons (OH-NOj-PAH),
hydroxylated nltro polynuclear aromatic ketones (OH-NOg-PAKJ and
hydroxylated nltro aromatic (OH-NOg-Ar) compounds 1s presented here.
Bloassay-directed fractionation through four sequential separations led to
detection of OH-NOg-PAH/PAK In a mutagenic and substantially simplified
subfraction of an urban air particulate (<1.7 diameter) extract.
Identifications have been made from both electron impact (EI) and highly
compound selective negative chemical ionization (NCI) HRGC/MS analyses.
Quantification of these compounds has been achieved using NCI HRGC/NS.
The presence of OH-NO2-PAH in particulate matter has been suggested by
Lofroth1, Nielsen2, and Schuetzle3 to account for mutagenicity in polar
extract fractions. Mechanisms have been proposed 2»3>* by which atmospheric
hydroxylatlon of PAH followed by nitration should yield both NO9-PAH and OH-
NOg-PAH. Similar mechanisms5 have been used to explain the formation of 0H-
NOz-Ar in smog chamber studies of toluene.
The Importance of OH-NO9-PAH became apparent when OH-NOj-pyrenes were first
identified as mutagenic mammalian metabolites of l-NOs-pyrene®. The
Importance of OH-M^-Ar became apparent when N0»-phenols were identified as
phytotoxlc agents' present In rain8. In addition, at least one Mfe-cresol
has been identified as a mammalian metabolite of a NO?-toluene Isomer*.
Increasing evidence suggests that OH-NO^-PAH are not only important products
of metabolism of N02-PAH 1e vivo but may also be important genotoxins to
which humans may be exposed via urban air particulate matter.
(1) lofroth, G. fit il. In "Toxicology of Nltro Aromatic Compounds".
D. Rlckert, ed. Hemisphere Publishing Corp., 1985.
(2) Neilsen, T. fit il. Envir. Health Perspect. 1983, 103-114.
(3) Schuetzle, D. Envir. Health Perspect. 1983, 4Z, 65-80.
(4) Zlellnska, B. fit fil. JACS 1986, lflfi, 4126-4132.
(5) Leone, J. and Seinfeld, J. Int. 0, Chem. Kinetics 1984, lfi, 159-
193.
(6) Ball, I. filil. Carcin. 1984, 5, 1557-1564.
665
-------
(7) Overcash, M. £i al. Water Resources Research Institute, Raleigh,
NC, Report No. 171.
(8) Leuenberger, C. §£ iL, submitted to Nature.
(9) Chism, J. e£ il- Drug Hetab. Dispos. 1984, 12, 596-602.
666
-------
ADVANCES IN CONTINUOUS TOXIC GAS
ANALYZERS FOR PROCESS AND ENVIRONMENTAL APPLICATIONS
-0-0-0-
Hermann Schmidtpott, Ph.D.
Compur Monitors, Munich, W.Germany
Hans J. Brouwers
Compur Liaison Office, Mt. Prospect, II 60056
667
-------
A continuous microprocessor controlled trace gas analyzer foe
toxic gases is developed to monitor processes and/or environ-
ments with a minimum of human interaction. The ensuing
criteria centered around the need to provide a maximum of
safety for the employee in the work place, Hence, the analyzer
must perform the monitoring task with the greatest amount of
accuracy and reliability.
A short response/decay time, wide dynamic range and specificity
are considered the prime criteria to obtain reliable and accu-
rate data. This is achieved by selecting optimum detector
technology for each toxic gas. The gases for which the
analyzer is designed are Hydrogen Fluoride, Fluorine and
Hydrogen Chloride. Phosgene can be detected by means of an
optional photometric detector syBtem. A sample enrichment
system allows detection in the ppb range for most gases.
The basic analyzer system is
designed to require no routine
maintenance. This is
accomplished by reducing the
number of moving parts to a
minimum. The sample is trans-
ported, using an air driven
aspirator.
All critical analyzer para-
meters such as sample and
reagent flowrates, vacuum,
pressure, temperature and
reagent levels are continuous-
ly surveyed and compared to
data stored in the system's
nonvolatile memory. An alpha
numeric display identifies the
cause when one or more
parameters do not conform.
Automatic calibration and
adjustment iB performed when
the analyzer is started up and
at selected time intervals.
Hard copy information of calibra'
tion data, gas concentrations
exceeding present alarm levelB
and their time of occurrence are
available on a built-in printer.
run miTJlH TftBi tl,llVTt<[
•Ir rant £*3"
JWHVrlMui
itrv Mr.
jriKlul.
11>4*66
^^9
-Hfc—J p?
8
/CMnl
rant.
1*
The system has maximum flexibility, as the various parameters
of the analyzer can be modified to adapt to the application.
668
-------
METHOD OF DETECTION
The detection method for HCl, HF and F2 1b the ion specific
electrode. Although the principle of detection is similar for
all these compounds, the examples in this presentation will
concentrate on the method for HC1.
The ion selective electrodes selected for this application are
the Ingold chloride specific electrode assembly.
The calibration curve for this assembly can be shown
as follows:
100 I
i?o
.160
I
10-fold Chang*
In concentration
'2
260
10.00
100
tor4
669
-------
The adsorption solution for a measuring range of 0.1 to 1. mg/m3
is as follows:
147.1 g Sodium Citrate-Dihydrate
58.4 g NaCl
160 mL 1.0 m HC1 solution
10 mL 0.1 m NaF- solution
The above is dissolved in 10 Liters of distilled water. It
provides a value of ul (mV) at the electrode assembly.
The calibration solution is 70% of the measuring range of the
and is made up as follows:
14.7 g Sodium Citrate-Dihydrate
5.8 g NaCl
16 mL 1.0 m HCl solution
70 mL 0.1 m NaF solution
The above is dissolved in 1 liter of distilled water. It
provides a potential of U2 (mV) at the electrode assembly.
Some critical parameters to be aware of are:
1 - "0 - 105 mV
2 - Ph of both solutions should be 5.8 +/- 0.2
3 - Slope of curve should be 56 mV/decade
All measured date are available on the print out, so that
individual critical parameters can be verified at all times.
670
-------
MONITORING OF CEMENT KILN EMISSIONS THROUGH
PLANTS: A CASE STUDY
P.K.Nandi , Madhoolika Agrawal 4 D.N.Rao
Centre of Advanced Study in Botany
Banaras Hindu University
Varanasi 221005, INDIA
#
Environmental Planning 4 Coordination Organization
Prayavaran Parisar, Arera Colony
Bhopal 462 016, INDIA
In certain parts of India, the rapid expansion of cement manufactu-
ring industry has introduced the likelyhood of increase in environmental
problems arising out of particulate emission from cement kilns. The present
study concerns the assessment of dust deposition on leaf surfaces and subs-
quent plant responses to particulate emission from an 1 million tpa capaci-
ty cement plant. The study revealed that the amount of dust deposited on
the leaf surfaces of perennial plants like Citrus medlca. Manqlfera Indlca
and Psldlum guyava were negatively correlated with their distances from the
emission source. While the magnitude of dust deposition was maximum on
C. medlca and minimum on P. guyava, the Ca accumulation was found to be ma-
ximum in C. medlca and minimum in M. Indlca. A species specific direct re-
lationship between the amounts of dust deposition and foliar Ca accumulati-
on was also observed. Moreover, visible injury symptoms were discernible
only on P. guyava and M. indlca growing within a radius of 2 km from the
emission source. The study thus suggests that the differential responses
of plants to particulate kiln emission may serve as a key to use plants for
the detection, recognition and monitoring of particulate pollution and its
biological effects.
671
-------
Introduction
In India increasing gap in demand and supply of cement and the presen-
ce of vast reserves of cement grade limestone have respited In rapid expan-
sion of cement manufacturing industries in recent years with subsequent In-
crease in problems arising due to emission of particulates into the atmosp-
here. Consequently for careful control of particulate pollution in indust-
rial regions, high volume sampler as per specifications of U.S. EPA has
been prescribed as a reference method for the monitoring of total SPM .
Though this method is ideal for measuring quantity and after analysis qual-
ity of particulates in the ambient air, it is far away from indicating or
help modelling biological effects of such pollutants, On the other hand,
leaves of plants may act as efficient dust collectors and sho* differential
responses depending upon the physical and chemical characteristics of part-
iculate matter deposited on them and thus may also act as tools for detect-
ion, recognition and wonitcring of air pollution effects®'-*. When plants
accumulate polluting compounds without changing their chemical nature by
metabolism and the pollutants are easily analysed in sample of plant mate-
rial, they ifiBy be used for monitoring of pollution effects also".
In the present study an attempt has been made to use plants for moni-
toring of particulate emission from cement plants as well as emission indu-
ced biological effects.
Site Description and Sources of Atoaospheric Emission
The cement plant with an annual production capacity of 1 million ton
is located at 21°57' H latitude and 82°20t E longitude on a plain topograp-
hy and at an elevation of 300 m above mean sea level. The minimum and max-
imum temperatures of the area range from 13 to 42°C. predominant wind
direction is towards SE and average speed is 6 Km h" ,
The latest dry process technology has been adopted for Portland cement
manufacture in the plant with inputs of raw materials as cement grade lime-
stone, Clay, laterite, breeze coke and gypsum. The principal sources of
particulate are the raw will, rotary kiln and clinker cooler. Emission
from cement mill and coal mill are relatively low. Generally, the average
particulate emission from the kilns of this type of plants in India amounts
to 0.2% or 2 kg Mt~^ of cement produced and when dust arresting devices
remain uncharged the total emission may increase to ft*. The fallout dust
at the environs of the cement plant thus constitute a heterogeneous mixture
with contribution of raw and clinker phase materials predominating.
Experimental Methods
The study was conducted in March 1966, Three to four years old plants
of Citrus medica, Hanclfera indlca and Psldlum auvava growing at different
directions upto 3 Km from the epicentre of pollution source and upto a
distance of 5 Km downwind were selected for measurement of dust load on
leaves, foliar calcium level and leaf wash suspension pH (Table I). For
the determination of background level of dust a site at a km NW was select-
ed .
672
-------
Plants growing at above mentioned sites were visually examined. Matu-
re leaves were collected, repeatedly washed with delonised distilled water
and leafwasb suspension was dried to determine the weight of solid matters.
Leaf area were measured and then dried In over at 105° C for 8h, Calcium
contents in leaves were determined by flame photometric method after ashi-
ng the dried and powdered leaf samples at 480°C with the addition of satur-
ated magnesium acetate. Leaf wash pH were determined with the help of a
pH meter. For all measurements at least three replicates were maintained.
Results
The dispersion of particulates from the stacks of cement plant as ex-
amined by the measurements of foliar dust load (Table II) and leaf-wash
suspension pH (Table 1} was found to be quite significant upto 3 Km
downwind of the emission source. The magnitude of dust deposition on lea-
ves of plants growing at 1 km NW was found to be comparable to those grow-
ing at distances of 3 and 3.5 Km on SW and 5E directions, respectively.
The levels of leaf surface dust, leaf wash suspension pH and foliar calcium
were found to be maximum at the immediate vicinity of the source but these
values for plants growing at increasing distances from the source were fou-
nd to be significantly low. While the magnitude of dust deposition was
maximum on £. medlca and minimum on P. guyava, the calcium accumulation was
found to be maximum in £. medlca and minimum in H. Indlca (Table II), The
maximum dust load of 49.2 g m-2 leaf are'a of P.guyava was found comparable
to a study conducted elsewhere®. A species-specific direct relationship
between foliar dust load and calcium accumulation was also noted. Chloro-
tJc foliar injury symptoms were only noticed in P. auvava and M.Indlca gro-
wing upto a distance of 2 Km downwind (Table II). On the same site,
lnsplte of thin crust formation on leaves of £. medlca no visible Injury
symptoms were detected in them.
Development of foliar injury symptoms was possibly due to higher dust
load and alkalinity of particulate matter on leaf surfaces. Cxaja® noted a
positive correlation between induction of foliar injury symptoms and the
concentrations of calcium silicate In cement kiln dust vis * via its pH
values. It was suggested that on hydration of cement kiln dust, alkaline
calcium hydroxide solution is released on leaf surfaces which in turn help
saponify the protective cuticle^permitting migration of solution through
the cuticle to mesophyll tissue . This possibly causes accumulation of
calcium in tissue with subsequent changes in plant metabolism and develop-
ment of injury symptoms, resulting In decreases in pbotosynttietic potential,
energy budget and biomass production of planta ' .
Conclusions
The measurements of leaf surface dust load, dust-suspension pH and
foliar calcium accumulation in plants growing at different locations from
the cement plant may help monitoring of particulate emission as well as
modelling its dispersion pattern. Plant responses further help recognition,
detection and monitoring of biological effects of particulate pollutants.
Acknoaledgements
The authors are thankful to the Head, Centre of Advanced study in
Botany for providing laboratory facilities*
673
-------
References
1. "Emission Regulations", Part III, Central Board for the Prevention
and Control of Water Pollution, New Delhi, C0INDS/20/1984-85
(Dec. 1985).
2. S.L. Lerman, E.L. Darley, "Particulates", In J.B.Mudd, T,T,Kozlowski,
eds, Responses of Plants to Air Pollution. Academic Press, New York,
San Francisco, London, 1975,pp.141-158.
3. W.H.O. Ernstf"Monitoring of particulate Pollutants", In L.Steubing,
H,J.Jager,eds, Monitoring of air pollutants by plants. Dr.W.Junk
Publishers, the Hague, 1982,pp.121-128.
4. A.C.Posthumus "Higher plants as indicators and accmulators of
gaseous air pollutants", Environ. Monitor. Assess. 3_ : 263 (1983)
5. B. Lai, R.5. Ambasht, "Impact of cement dust on the mineral and
energy concentration of Psldlum guayava,"Environ. Pollut. 29 :
241 (1982).
6. A.T.Czaja,"Uber die Einwirkung von stauben,speziell von Sementofenst-
anb auf pfflanzen," Angew Bot. 16: 145 (1966).
7. S.N.Singh, D.N.Rao,"Growth of wheat plants exposed to cement dust
pollution, "Water Air soil pollut. 14: 241 ( 1980).
674
-------
TABLE I
Leaf-wash pH of plants growing at different locations around a
cement plant.
Site No.
Direction
Distance (Km)
Leaf-wash pH
5o
-
0
9.2
S!
SE
1
9.0
S2
NW
1
e. a
S3
SE
2
8.8
S4
SE
3.5
8.2
S5
SW
3
8.2
S6
E
3.5
7.8
S7
SE
5
7.8
S8
NW
e
7.6
675
-------
TABLE II
Foliar dust load, calcium contents and injury symptoms in plants
9uyava< CsC. medlca and M=M. lndlca) growing at different
locations around a cement plant.
Site No.
Plant
Dust load
Calcium
Injury
-2
(g m leaf area)
(X)
symptom
P
49.2
4.8
Necrosis
S0
C
M
—
-
-
P
5.8
3.4
Chlorosis
51
C
6.5
3.8
-
M
5.6
3.2
Chlorosis
P
3.9
2.05
S2
C
4.2
2,95
-
i.
M
m
CO
r"\
1.9
P
4.2
2.8
Chlorosis
S,
C
4.56
3.3
-
M
A. 26
2.4
Chlorosis
P
3.95
2.1
S4
C
4.3
3.0
-
M
3.8
1.9
-
P
3.8
1 .95
S5
c
4.3
2.95
-
M
3.8
1.9
-
P
1 .98
1.6
S5
C
2.05
2.5
-
o
M
1 .95
1.2
-
P
2.1
1.62
S7
C
2.2
2.45
-
r
M
1.96
1.3
-
P
2.05
1.6
sa
C
2.16
2.4
-
M
1.90
1.25
-
676
-------
NOH-AQ0BOU8 ION EXCHAHQB CHROMATOGRAPHY X8 A
FREPARATIVB ITIACTIOMATIOM METHOD FOR SHORT
TBRM BXOA88AY ANALYSIS 07 WOOD 8N0KS EXTRACTS
Douglas A. Ball
Hani S. Karam
Richard N. Kamens
Department of Environmental Sciences
and Engineering
University of North Carolina
Chapel Hill, NC 27514
Hood smoke contains significant quantities of mutagenic
and carcinogenic compounds such as polynuclear aromatic
hydrocarbons (PAH). However, PAH account for only 12-25 %
of the mutagenicity of wood smoke particles with the
remainder appearing in the m6re polar fractions. These
polar fractions have proven to be refractory to chemical
characterization and identification of mutagenic species has
not been possible.
Two commonly used separation methods for combustion
emissions, aqueous phase acid/base/neutral and normal phase
HPLC fractionation, have been used on wood smoke but have
resulted in low mass reoovery or poor separation of polar
compounds. In order to address these difficulties, an
acid/base/neutral fractionation method was developed using
non-aqueous ion exchange chromatography. The two columns
utilized were composed of: eation exchange resin, Amberlyst
15, for separating bases; and the anion exchange resin,
Amberlyst 26, for separating aeids. High recoveries, as
determined by GC analysis, were observed for 14 compounds in
a standard mixture. Total mass reoovery for a wood smoke
sample through this scheme was -100 % with the mtutral
fraction accounting for 54 % of the mass. Acid, polar acid,
neutral and basic fractions were collected and following
solvent exchange into DMSO, were tested in the :
Ames/Salmonella mutagenicity assay. The neutral fraction
demonstrated the highest mutagenicity both in the presence
and absence of 89 activation.
677
-------
MON-AQUBOC8 ION EXCHANGE CHROMATOGRAPHY AS A PREPARATIVE
FRACTIONATION METHOD FOR SHORT TERN BI0A8SAY ANALYSIS Or
WOOD SMOKE EXTRACTS
INTRODUCTION
The identification of biologically active compounds in
complex environmental mixtures is a difficult technical
challenge. The polar fractions of wood smoke, which contain
75-90 * of its mutagenic activity, have proven to be
particularly refractory to analysis (1). Preparative and
semi-preparative methods such as aqueous phase
acid/base/neutral separation, open column silica gel
chromatography, and normal phase HPLC silica gel fract-
ionation have been successfully used for separation and
identification of nonpolar polynuclear aromatic hydrocarbons
and partially successful for studying their moderately polar
nitrogen and oxygen derivatives (2,3). However, these
methods have resulted in low mass recovery and/or poor
separation of polar compounds when applied to wood smoke
extract{3,5). In order to address these difficulties, an
acid/base/neutral fractionation method was developed using
non-aqueous ion exchange chromatography.
MATERIALS AND METHODS
Reagents
Chloroform, tetrahydrofuran (THF), and methanol were
purchased from Burdick and Jackson, Muskegon, Mi. Ion
exchange resins were manufactured by Rohm and Hass and
obtained from Alfa, Danvers, MA. Formic acid and isopropyl
amine were obtained from Fluka, Ronkonkona, NY. Source and
purity of standard compounds are listed in Table I.
Preparation of Ion Exchange Resins
Preparation was similar to that of Jewell, et al (6).
Amberlyst 15, the cation exchange resin, was washed with a
methanolic potassium hydroxide solution (5 % by weight) and
then rinsed with methanol. It was then converted to the
acid state by slowly adding a solution of 5 % volume HCL in
methanol and stirring for 30 minutes. The resin was washed
with distilled deionized water until the pH of the washing
water was neutral. It was then soxhlet extracted for 24
hours each with methanol, dichloromethane, and a
chloroform/5 % tetrahydrofuran mixture. The resin was
stored in chloroform until used.
The anion exchange resin, Amberlyst 26, was washed with
a 5 % HCL solution in methanol and rinsed with methanol.
After rinsing, the resin was activated by adding a 5 t
methanolic hydroxide solution (described above). The resin
was washed, soxhlet extracted (no THF was used) and stored
678
-------
as described above. Approximately 1 gram of resin per 10
milligrams of sample was slurry packed in chloroform into 25
or 50 ml burettes which were plugged with glass wool and
fitted with teflon stopcocks. Columns should not be packed
too tightly because the variety of solvents used causes
swelling of the resins during the separation.
Preparation of Fractionation Standard
A high concentration standard was prepared by adding
approximately 60 mg of each of the compounds listed in Table
I to 10 mis of a 10 % THF/Chloroform solution. The
compounds were chosen to reflect the expected distribution
of acidic, basic and neutral components in wood smoke. Many
of these have been identified in wood smoke (see Kamens, et
al, these proceedings).
Preparation of Wood Smoke Extract
Smoke from a residential wood stove burning yellow pine
was introduced into the UNC Combustion Emission Research
Facility chambers and then sampled as described in the
literature (1,4) and elsewhere in these proceedings (Kamens,
et al). The particulate sample, collected on 13 cm Teflon
coated, glass fiber filters (Pallflex T60290), was soxhlet
extracted for 16 hours with dichloromethane, concentrated by
rotary and nitrogen evaporation and a gravimetric mass was
determined for the extract.
Separation Procedure
The general fractionation procedure is outlined in
Figure 1 and is similar to Jewell, et al (6)(but without a
solvent reflux step). 40-100 mg of standard or sample
dissolved in lml of 10% THF/chloroform was placed on the
Amberlyst 15 column. Acids and neutrals were unretained and
were eluted with a 10 % THF/10% methanol/ 80% chloroform
mixture. The modification of the chloroform by adding THF
increased the recovery of acids from the cation exchange
column and the methanol increased elution of the highly
polar compounds in wood smoke. The retained bases were
removed from the resin with 20 % volvtme isopropylamine in
methanol. The acids/neutrals mixture was concentrated by
rotary evaporation to *1 ml and placed on the anion exchange
column. Neutrals were unretained on this column and eluted
in 100 % chloroform. Retained acids were then eluted by
adding a 1% formic acid/methanol solution. A second, more
polar, acidic fraction was collected by eluting with a 20%
formic acid/methanol solution. Each elution was carried out
at a rate of -1 ml/minute and the volume of solvent used was
20-50 mis per fraction (-0.5 mls/mg sample). Basic and
acidic fractions were rotary evaporated just to dryness to
remove any remaining solvent modifier (ie. isopropylamine or
formic acid). They were redissolved in fresh solvent
679
-------
prior to gas chromatography or mass residue analysis.
Gravimetric mass measurements of the fractions were
accomplished by evaporating the solvent from small aliquots
of the sample and weighing the residue.
Sample
A-15 Cation
Exchange
Resin
A—26 Anion
Exchange
Resin
~ f
Neutrais Acids
Figure J Separation scheme.
Base
1
Polar Acids
Gas Chromatography
Chromatography was performed on a Carlo Erba model 4130, gas
chromatograph utilizing a 30m fused silica J & W, DB 1701
capillary column with a flame ionization detector. It was
temperature programed from 130-290°C at 8°C/minute and the
injection was splitless with a 20 second hold. A Shimadzu
integrator was used and the internal standard method was
used for quantitation. Identities of the peaks in the
standard mixture were confirmed using the above column and
program with a 5992 Hewlet Packard GC/MS system.
Mutagenicity Bioassay
Each wood smoke fraction and the neat sample was tested
for mutagenicity in Salmonella typhimurium strain TA98 with
and without a 10% rat liver homogenate mixture (S9) prepared
from Arochlor induced rats as described by Maron and Ames
(7). Due to a limitation of mass, each fraction was tested
with single plates at four doses which were chosen by
estimating the the probable linear dose/response range based
on prior testing (data not shown). Mutagenicity slope
values were calculated by simple linear regression using all
dose levels unless toxicity was observed. Quantitative
slope values were calculated for the purpose of comparison
and the limited number of data points suggests a qualitative
interpretation of mutagenicity results.
680
-------
M8ULTS AMD DISCOSSXOM
Recovery of Fractionation Standard
As is ahovm in Table I, qualitative separation of the
basic standards into the the base fraction was excellent
with no neutral or acid overlap. The quantitative recovery
of bases was greater than 92%. Five neutral compounds
appeared only in the neutral fraction while two compounds,
2,3-dimethyl indole and naphthalic acid anhydride behave
partially as weak acids. Phenol and 1-naphthol, relatively
weak acids, ware split between the neutral fraction {21.5%
and 14.4%) and the acidic fractions. The stronger acids,
benzoic acid and p-nitrophenol, appeared almost entirely in
the acidia fractions. Quantitation of 9-phenanthroi was
unreliable because of solubility problems. There was no
compound which appeared primarily in the polar acid fraction
indicating that there is no representative of this class in
the mixture. Suitable polar compounds which could be
chromatographed and which might appear in this class such as
di-acids or di-hydroxylates were not available. The
material which did elute in this fraction may have been that
which was tightly bound to the anion resin active sites.
Total recovery of standards was generally high {>87%}.
Mass Distribution of Hood Smoke
The particle extract mass which was fractionated and
the mass which appeared in each fraction is displayed in
Table II. The slight over recovery of mass may reflect
measurement error or may possibly be due to the effect of
THF hydrogen bonding with the acids in the sample (8).
Table II. Mass Recovery of Wood Smoke Among Fractions
Mass
Sample Mg Percent
Whole Extract
51.0
100.0
%
Base
2*4
4.7
%
Keutral
27.6
54.1
%
Acid
20.0
39.2
%
Polar Acid
2.8
5.5
%
Sua
52.8
103. S
%
Measurement error
due to balance fluctuation,
*/" 5.
681
-------
Mutagenicity
The mutagenic potency in Salmonella strain TA98 was
measured by performing a simple least squares linear
regression on the dose response data and is expressed as
revertants per ug extract as shown in Table III. The total
revertants per fraction was calculated by multiplying the
slope value of the regression line by the mass appearing in
the fraction. Percent mutagenicity distribution is
calculated by dividing the revertants per fraction by the
revertants in the whole extract.
The unfractionated pine smoke extract was particularly
mutagenic when compared with other samples tested in this
laboratory (typical values for TA98: -S9 - 0.4 rev/ug; +S9 ¦»
1.2 rev/ug). The neutral fraction was the most potent
without adding S9 although not as mutagenic as the whole
extract. The acidic fraction makes a significant
contribution (16.6%) to the mutagenicity of the sample. The
reconstituted sample (reconstituted from the fractions by
mass percentage) also had a lower response than the whole
sample indicating a possible loss of direct acting mutagens
in the fractionation process. This possibility is also
suggested by the 68.6 t recovery of percent mutagenicity
when the mutagenicity from each fraction was summed.
However, given the limited data set, these departures from
100% recovery may simply be due to measurement error.
Indirect acting mutagenicity (+S9) slope values were 2
to 15 fold higher than direct acting values for all of the
fractions. The high potency of the base fraction was
surprising given that no highly mutagenic basic compounds
have previously been identified in wood smoke. However the
contribution that the bases make toward the whole sample is
small. The neutral fraction, in which polynuclear aromatic
hydrocarbons appeared, was the most potent fraction and also
contained 93% of the indirect acting mutagenicity of the
whole unfractionated extract. The acidic fractions make
relatively small contributions to the indirect acting
mutagenicity of the sample. Recovery of indirect acting
mutagenicity as measured by the reconstituted and summed
fractions was 90% and 107% respectively. This suggests
there is good recovery of these mutagens through the
fractionation scheme.
In assessing the usefulness of this method, the good
qualitative separation and the quantitative recovery of
standards demonstrates that non-aqueous ion exchange
chromatography can be applied to complex mixtures. Further
development of the method may result in reducing the overlap
of weak acids into several fractions. Manipulation of
solvent polarity is critical to solving this problem.
682
-------
For a separation method to be used in bioassay directed
fractionation it requires: 1) high recovery of sample mass
through the system in order to make informed decisions about
further analysis; and 2) high recovery of 'weighted'
mutagenicity. While the bioassay evidence supporting the
utility of this fractionation method is limited, the
preliminary results presented here indicate that for a non-
aqueous sample this method measures up relatively well.
This ion exchange separation technique may be especially
useful when the investigator wishes to avoid the addition of
strong acids and bases (ie H2S04 or NaOH) to the sample as
in aqueous phase acid/base/neutral methods.
ACKNOWLEDGEMENTS
The authors would like to thank Drs L. Claxton and J.
Lewtas, us EPA for their continued support of this project.
We would also like to acknowledge the useful suggestions
made by H. Nishioka, Battelle Columbus and the excellent
technical support of J. Fulcher and C. Braun, U. of NC.
REFERENCES
1. Kamens, R. M., D. A. Bell, A. Dietrich, J. Perry, R.
Goodman, L. D. Claxton, and S, Tejada, "Mutagenic
transformations of dilute wood smoke systems in the presence
of ozone and nitrogen dioxide: Analysis of selected HPLC
fractions from wood smoke particle extracts." Bnvlr SeH T
and Tgch, 19: 63-69, 1985.
2. M. Nishioka, C. c. Chuang, B. A. Peterson, A. Austin and
J. Lewtas,"Development and quantitative evaluation of a
compound class fractionation scheme for bioassay-directed
characterization of ambient air particulate matter,"
Environment International> ii»i37-i46, 1985.
3. M. Nishioka, P. Strup, C. C. Chuang, and M. Cooke,
"Fractionation and chemical analyses of wood stove samples,"
Draft final report prepared by Battelle Columbus
Laboratories for ORD-USEPA, contract number 68-02-2686, Task
Directive 134, 1986.
4. D. A. Bell and R. M. Kamens, "Photodegradation of wood
smoke mutagens under low NOx conditions," Atmosnharin
Environment 20: 317-321, 1986.
5. D. A. Bell, R. M. Kamens, L. D. Claxton and j. Lewtas,
"Photoreaction of wood smoke particles: Destruction and
creation of mutagens in sunlight," APCA *86-77.a. presented
at 79th Annual Meeting, Air Pollution Control Association,
Minneapolis, Minn, 1986.
683
-------
6. D. M. Jewell, J. H. Weber, J. w. Benger, H. Plancher and
D. R. Latham, " lon-exchange, coordination, and adsorption
chromatographic separation of heavy-end petroleum
distillates," Analytical Chemistry AH 1391-1395, 1972.
7. D. Maron and B. N. Ames, "Revised methods for the
Salmonella mutagenicity test," Mutation Research. 113: 173-
215, 1983.
8. M. Nishioka, personal communication, 1986.
Table I. Percent Recovery of fractionation Standard
POLAR TOTAL PERCENT
COMPOUND 8ASE NEUTRAL ACID ACID RECOVERY STD.DEV. c
Phenol a
21,5*
12.1*
i.ei
101,1*
i.H
Beoiofc acid
3.91
ite.n
It. 9*
'JJ.tt
11.4*
2,6 Diethyl qufnollne
98.8*
5.2*
Dlbsniofunn
106.8*
106.8*
!.«
2,3 Dimethyl Indol
91. n
s.s*
SO. 5*
2.2*
1 Haphthol
IMt
59,8*
11.51
91.1*
5.1*
9 Fluoranone
104.1*
101.1*
2.5*
1,8 Benioqulnollne
M.I*
92.1*
6.0*
p nltrophanol
(1.9*
0.9*
96.7*
8.5*
Pyr«nt
n.H
$9.4*
1.9*
Naphthallc add anhydride
71.7%
9.All
5.4*
86.5*
1.2*
9 Phenanthrol
23.lt
1.8*
25.4*
14.9*
Pyrtne carboxildahyde b
19.71
89.7*
0.5*
Bent[s]anthrae*n« 1,12 dlona
103.9*
103.9*
1.5*
i All chemicals obtained fro* Aldrlch, Milwaukee, VI, >19* purity unlets noted,
b Alfa > tt*
c Peak area deviation among replicate 6C runs.
Table HI Hutagenlc Potsncy and Distribution
1A99 -S9
TA98 +S9
Revertants Revertants
Percent
Revertants
Revertants
Parcant
Par
Per
Mutagenicity
Per
F#r
Mutagenicity
Seapie
u& Extract Fraction
D1str1but1on
ug Extract
Fraction
Ofstrlbutlon
tfhole Extract
0.62
3 >365
190.0*
r.4t
126595
109.0*
Ease
0.19
453
1.4*
3.04
7308
5 8*
Wtutral
0.54
14911
47.8*
4.21
>1765?
93.1*
Acid
5.16
5200
18.5*
6.40
Mil
6 4*
Polar Add'
(.31
872
2.8*
0.19
2199
1.7*
fteconstltutad
0.44
23352
14,5*
2.16
114114
90.2*
Sua
21512
68.5*
135435
101.0*
Sean (ns3) revertants, controls; -S9 * 23; +55 * 34;
3 uj 2-n1trcfluoren» = 272; 3 ug 2-amlrnMnthreeens « 19(5,
684
-------
EFFECTS OF RETROFIT TO REDUCE EMISSIONS
FROM EXISTING WOOD STOVES
R.W. Braaten and A.C.S. Hayden
Research Engineer and Research Scientist
Canadian Combustion Research Laboratory (CCRL)
Energy Research Laboratories, CANMET
Department of Energy, Mines & Resources
555 Booth Street, Ottawa, Canada K1A 0G1.
Over the past few years, it has become widely recognized that wood
stoves can produce high levels of incomplete combustion products,
becoming, in eome areas, one of the most serious air pollution sources.
Whitehor8e, in Canada's Yukon Territory, has one of the highest levels
of wood stove-generated air pollution in North America. Most efforts to
date, both from the technical and the regulatory side, have concentrated
on developing new, cleaner burning appliances. While this approach is
commendable, and is resulting in b number of cleaner, more efficient
appliances on the market today, the majority of woodstoves installed in
homes in North America are of the older, inferior combustion type. That
most stoves have been installed in the past eight years or less makes
the probability of their replacement in the near term very unlikely.
Unless retrofit technologies or ubo control strstegies are applied in an
effective manner, the emissions from wood stoves will be with us for a
long time to come.
This paper presents some retrofit technologies that offer the
potential to reduce the emissions of incomplete combustion products from
existing wood stoves, either by forcing a change in the rate in which
the wood is actually burned, or by adding on an afterburner-type
catalyst to clean up the emissions before they leave the chimney. In
particular, laboratory results on the use of a retrofit add-on catalyst
fitted to a stove representative of the majority of installations in
Whitehorse, are presented. The add-on catalyst makes a major reduction
in the amount of particulates emitted* but only when the stove is
operated in the updraft position. It is ineffective when the stove is
in sidedraft. On the other hand, the add-on is seen to be quite
sstisfactory in cleaning up carbon monoxide in the exhaust, no matter
what the stove operating mode.
685
-------
Introduction
Over the past few years, it has become widely recognized that wood
atoves can produce high levels of incomplete combustion products,
becoming, in some areas, one of the most serious air pollution sources.
Indeed, Whitehorse, in Canada's Yukon Territory, has one of the highest
levels of air pollution due to wood stoves in North America (1).
Most efforts to date, both from the technical and the regulatory side,
have concentrated on developing new, cleaner burning appliances. While
this approach is commendable, and is resulting in a number of cleaner,
mere efficient appliances on the market today, the majority of
woodstove3 installed in homes in North America are of the older,
inferior combustion type. That most stoves have been installed in the
past eight years or less makes the probability of their replacement in
the near term unlikely. Unless retrofit technologies or use control
strategies are applied in an effective manner, the emissions from wood
stoves will be with us for a long time to come.
Some reduction in emission levels may be possible through consumer
education in proper stove operating procedures and fuel conditioning.
However, changes to the hardware and/or improved feedback to the
operator on the stove's performance offer potentially the most promising
means of reducing emissions. In particular, techniques allowing the
upgrading of existing equipment at reasonable cost offer the best hope
of significant reductions in emission levels through consumer
acceptance.
Any emissions reduction technique, to be effective and acceptable to the
consumer, should have the following characteristics. The primary
requirement is a major reduction in pollutant emissions. At the same
time, there should be a significant gain in fuel economy, leading to a
reduction in wood consumption, to make adoption of the technique
attractive to the user. The technique should have a low capital cost
relative to the cost of a new appliance. It should be easy to install
(preferably consumer installable), again to minimize cost. It should be
uncomplicated to use; coupled to that, a feedback provision should let
the user know that the technique is working, increasing satisfaction and
ensuring proper operation. There should be no degradation of to safety.
Preferably, the technique should even offer some safety benefits, such
as reduced creosote potential, lessening the chance of a chimney fire.
As mentioned, Whitehorse has a serious air pollution problem due to wood
stoves. To attempt to ameliorate the problem, Energy, Mines and
Resources Canada, through its Remote Communities Demonstration Program,
has been carrying out a field study on improved combustion techniques
for wood stoves, over the past heating season (2). The study has
focused on three different methods. The first is a replacement of
conventional, high polluting stoves with new, clean burning appliances.
The second is the replacement of single walled flue pipe with double
walled pipe, to reduce the amount of heat supplied to the house by the
flue pipe, and force the user to run his appliance at a higher firing
rate, hopefully moving the use pattern to the clean side of the critical
burning rate of the appliance (3). The third technique was the use of
add-on catalytic converters, to ignite and burn off the incomplete the
combustion products before they reach the chimney.
686
-------
Experimental Methods
The specific retrofit technique examined in detail for this paper waa
the use of a retrofit add-on catalyst to react some of the incomplete
combustion products, specifically hydrocarbons and carbon monoxide,
which would otherwise condense to form creosote in the flue or be
released as undesireable pollutants to the atmosphere.
Test Stove
The specific stove chosen for detailed testing was one constituting
nearly one-half the installed wood burning appliances in Whitehorse, so
that any successful improvement to this stove would have fairly general
applicability. The unit was a large (0.15 cu.m. firebox capacity)
stove of the sidedraft design (4), which could also be operated in the
updraft mode with the use of a bypass damper.
Retrofit Catalyst
The retrofit catalyst had a ceramic substrate coated with platinum. The
catalyst itself was enclosed in a metal container, with provision made
to manually bypass the catalyst (for reloading of the stove, etc.) by
moving the catalyst so aa to open a passage around the catalyat for the
flue gases. A certain amount of refractory material was built in to
keep the temperatures elevated to ensure catalyst light-off.
Test Procedure
The test results presented in this paper are based on laboratory trials
at CCRL using the C5A/ASTM stack loss/dilution tunnel test method. This
method generally conforms to the draft US EPA teBt requirements for wood
stove emissions, except for the fuel charge configuration and some minor
variations, such as differing dilution tunnel diameter and size of
particulate filtera. For these tests, the Oregon/EPA fuel charge
configuration was used.
Flue gas composition was measured continuously in the following manner:
carbon dioxide with an infrared analyzer, oxygen with a paramagnetic
analyzer, NOx by chemilumineacence; carbon monoxide also using an
infrared analyzer; hydrocarbons using a cold line flame ionization
detector ae well as a Byron hydrocarbons 301 analyzer, in order to
determine some of the non-straight chain hydrocarbons} . Fuel weight
change waa determined continuously using a digital scale, and
temperatures with chromel-alumel thermocouples. All data were
continuoualy recorded and analyzed on-line with a digital computer/data
logging system.
The appliance waa operated over firing rates encompasing its normal
operating range. No attempt waa made to operate at either the minimum
burn rate attainable or the maximum burn rate possible when in the
non-catalyst modB. The same diameter flue pipe as appliance flue exit
diameter was used.
687
-------
The appliance was operated with a section of straight flue pipe
replacing the add-on catalyst for the baseline runs, and using the same
light-off procedure for all the catalyst runs, consisting of running at
maximum air setting for the first five minutes of the run with the
catalyst brought into action after three minutes. After five minutes,
the air setting was changed to its final value, if different than
maximum, and no further changes made for the rest of the run.
Results
Figures la and lb present performance profiles of the test stove for
particulates and carbon monoxide versus firing rate, respectively, with
and without the add-on catalyst and with the stove operated in both the
updraft and the Bidedraft mode.
The emissions of particulates and carbon monoxide for the unmodified
stove are seen to be slightly higher in the updraft mode than in
sidedraft. At a burning rate of 3 kg/h, total CO emissions are about
1600 grams, being about 25SS lower in sidedraft. For total particulates,
as determined with the dilution tunnel, comparable levels are 160 and
120 grams, respectively.
The retrofit catalyst does a good job in reducing carbon monoxide
levels, in both the updraft and Bidedraft modes, to around 500 grams st
3 kg/h, less than 1/3 the normal levels.
The effect of the retrofit catalyst on particulates is not nearly so
straightforward.
In the updraft mode, the catalyst is extremely effective, lowering total
levels to about 20 grams, about 1/8 the normal levels.
However, in the sidedraft mode, the catalyst appears to be ineffective,
probably due to too low temperatures and, perhaps, an inadequate supply
of oxygen.
It is also important to notice that the slope of the curves for both
particulates and CO, using the add-on catalyst and with the stove
operatted in the updraft position, is not nearly so steep for decreasing
firing rats as for the unaltered stove, indicating even better
performance at the lower firing rate8 typically found in dwellings
during the hesting season.
While not reported in depth in this paper, preliminary results from the
field trial in Whitehorse for the double walled flue pipe indicated a
contrary result to what was expected. The double walled pipe used
appeared to result in a marked increase in creososte depostion in the
flue pipe sections themselves, resulting in complaints from the
howowners using the pipe. On closer examination, it was revealed that
the particular double walled pipe chosen had significant slots in the
outside wall at the top and bottom of each section, probably to reduce
heat build-up. It appears that the slots may have been too successful
in this regard, resulting in cold inner surfacss promoting increased
creosote. This has led to a more detailed experimental program at CCRL
to examine in detail the performance of several double walled pipes, and
to formulate an optimized design for wood stove use.
688
-------
Conclusions
1. Laboratory testing indicates significant emissions reduction
potential with the use of the add-on catalyst for the major stove
used in Whitehorse. Reductions in total particulate emissions of
up to 80S CBn be expected, but only when operated in the updraft
mode.
2. The add-on catalyst resulted in a reduction in total carbon monoxide
levels of about 2/3, irrBgardless of whether the sotve was run in
the updraft or sidedraft modes.
3. A good add-on catalyst design can be highly effective in reducing
wood stove emissions on some stoves. However,the same catalyst may
be of little benefit on other stoves, so it is important to be able
to identify the stove or conditions under which the catalyst is
likely to perform well and use it only on those stoves.
Alternatively, add-on catalysts should be designed to function
properly on as wide a selection of appliances as possible.
4. Double walled flue pipe may be a means of improving stove
performance by promoting higher user firing rates, while lessening
creosote potential, but some present pipe designs appear to negate
this advantage by too much heat rejection. There is a need for
further work to recommend optimum designs to achieve all the desired
objectives, including emi88ions reduction. Further experiments are
now being carried out and discussions sre being initiated with the
approval agency and with Canadian manufacturers to attempt to
correct these deficiencies.
References
1. R. McCandless, "Impact of Residential Wood Combustionn on Measured
Particulate and PAH Concentrations in Whitehorse, Vukon Territory",
APCA Paper 84-70.6, Annual Meeting, Air Pollution Control
Association, San Francisco, 1984.
2. Proposal, "Whitehorse Efficient Woodstove Demonstration", submitted
by Omni Environmental LBbs to City of Whitehorse in response to RFP
CR9700-RC12-206 SD of the Remote Communities Demonstration Program,
Energy, Mines & Resources Canada, August 1986.
3. A.C.S. Hayden and R.W. Braatenj "Effects of Firing Retee and Design
on Performance of Wood-Fired Appliancea"j Proceedings, Wood Heating
Alliance Annual Meetingi Louisville KTj March 1983.
4. A.C.S. Hayden and R.W. Braaten}"Effect of Wood Stove Design on
Performance"j CANMET Report 80-46} prepared for Canadian Wood Energy
Institute Annual Meeting} Toronto, Canada} June 1980.
689
-------
Dilution Tunnel Results
300
Carbon Monoxide
ran-avtalyvt «Hrih
uptfrBf* >«¦¦¦ iitalys*
iwiiiMlyii ilMnift
oilttyit sIMrkft
1.5 3 4.5 6
Burn Rita (kg/h db)
7.5
2000
J 600
s
»
&200
w
•
TJ
X
0
1 800
o
JB
*
u
400 »
DM-aiUlytt Ufdrrtt
' »n cstslyvt
'~ ilMrtTt
1.5 3 4.5 6
Burn Rat* (kgsh db)
7.5
Figure 1. Effect of add-on retrofit catalyst on particulate and carbon monoxide
total emissions, when mounted on a sidedraft-design stove, operated
in both the updraft and sidedraft modes.
-------
The Contribution of Residential Wood Combustion to Respirable and Inhalable
Particulate Concentrations in Missoula, Montana
James E. Houck
OMNI Environmental Services, Inc.
Beaverton, Oregon
James H. Carlson
Missoula City - County Health Department
Missoula, Montana
Clifton A, Frazier
Desert Research Institute
Reno, Nevada
Robert A. Cary
Sunset Laboratory
Forest Grove, Oregon
Missoula. Montana, has a long history of air pollution due to a large
extent to residential wood combustion (RWC), road dust, and winter
inversions. Numerous studies have been conducted in the past 10 years
examining various aspects of the problem. During the winter of 1986/1887,
fln extensive ambient and source sampling program was conducted to quantify
the ambient levels of particles attributable to specific sources by using
chemical mass balance (CMB) receptor modeling. The limiting factor in many
CMB studies has been the lack of representativeness in the source profiles
used. This is particularly problematic with RWC and road dust sources. The
chemical composition of road dust can change dramatically over relatively
short distances and the chemical composition of RWC has been shown to be
very variable due to the differences in wood fuel, appliance type, and
operational parameters.
The source sampling strategy and chemical profiles for several Missoula
particulate sources are presented here. Major emphasis was placed on
obtaining representative RWC and road dust "fingerprints". Automobile
exhaust, diesel truck and train exhaust, and several small wood-producte
point sources were also sampled and analyzed.
691
-------
INTRODUCTION
During the period from January. 1984, through July, 19B6, there were
nine violations of the primary 24-hour particulate standard and 60
violations of the secondary 24-hour standard in Missoula, Montana. All
violations except one occurred between December and March. Based on the
fact that the ratio of respirable (<2.5 ju) particulate levels to total
suspended particulate (TSP) levels ranges from near 1:1 to approximately
1:5, it has become apparent that both residential wood combustion (RWC) and
road dust ere important sources of ambient particles, with their relative
impact depending on meterological conditions. (RWC emissions are
predominantly smaller than 2.5 ju in diameter and road dust is predominantly
larger than 2.5 ji in diameter). Other sources of particles in the Missoula
airshed include wood-products industry point sources and vehicular exhaust.
A number of studies have been conducted to assess the sources of
airborne particles in Missoula. The results of the studies have concluded
that RWC is the major source o£ respirable particles and road dust is the
major source of coarser particles. (See for example reference 1.) However,
even previous chemical mass balance (CMB) studies have failed to adequately
quantify the relative contributions of particulate sources due to a limited
number of samples, non-representative source profiles, or inadequate
analytical data.
A key factor in performing CMB source apportionment is representative
source profiles containing data for the chemical species that occur at
highest concentrations in the sources. The sampling strategy,
instrumentation, and analytical techniques used to obtain source data for
the Missoula airshed are discussed here. The chemical compositions of a
number of sources which have been completed to date are also presented.
Selection of ambient samples for detailed analysis is currently in progress.
SOURCE SAMPLING
Several different source sampling techniques were used. Samples for
RWC were obtained in the plume, typically one meter from the chimney top to
permit organic compounds to condense. A specially-designed sampler with
dual inlets was used. One inlet was fitted with a Teflon filter for
inorganic analysis and one inlet was fitted with a quartz filter for carbon
analysis. No effort was made to size categorize the emissions, as the mean
mass distribution of RWC emissions is well less than 2.5 ju. Homes in
Miasoula were selected at random for sampling. Each RWC sample was a
composite of three to five wood-burning appliances. Samples from a hog fuel
boiler, diesel trucks, and a diesel train were also collected with this
sampler.
Road dust samples were collected with a vacuum-cleaner-like device from
three paved streets in the vicinity of the ambient monitoring site. One
street was classified as residential, one as a collector, and one as a main
arterial. Samples before and after application of winter sanding material
were collected. The samples were dried at 110°C for 24 hours, sieved to
less than 38 ju and reeuspended onto both Teflon and quartz filters. A
dichotomous sampler fitted with a PMiq inlet was used for collection of the
resuspended dust, which permitted subsequent analysis an two size fractions
(<2.5 xt and 2.5-10 .u). A grab sample of the city sanding material was also
dried, sieved, and resuspended in the same fashion.
Vehicular exhaust is being ssmpled (currently in progress) with a
mobile dilution sampler. (For a description of dilution sampling see
reference 2.) The vehicular exhaust is being collected onto Teflon and
quartz filters after it is mixed (cooled and diluted) with filtered ambient
air. The motor vehicles are being selected based on the typical make-up of
vehicle types in the Missoula area. Samples are being collected while the
vehicles are being driven in a normal fashion through residential and
692
-------
commercial areas, Aa with the RWC samples, no attempt has been made to
size categorize the emissions, as the predominant maBS of vehicular
particulate emissions is less than 2.5ju.
A wood chip (particle board) dryer/sander dust burner was sampled by
placing two dichotomous samplers (PMjq inlets) into the plume about I meter
from the exhaust ports. Teflon filters were placed in one sampler, quartz
filters in the other. Since this source was a combination of physical and
combustion processes, i.e.* sander dust burner exhaust is used to remove
moisture from wood particles, the size distribution of the emissions could
not be assumed to be predominantly In the fine or coarse range, and hence
dichotomous, rather than single filter, samplers were required.
ANALYTICAL TECHNIQUES
X-ray fluorescence spectrometric (XRF) analysis was conducted for 27
elements on the Teflon filters with a Kevex Corporation Model 700/8000 X-ray
Fluorescence Analyzer. A Cabn 25 electromicrobalance was used to determine
particulate mass. Carbon analyses were performed by the thermal/optical
technique on the quartz filters. Hater soluble metals are being analyzed
(in progress) by inductively coupled argon plasma spectroscopy (ICAP) on the
Teflon filters after completion of the XRF analysis.
SOURCE PROFILES
The RWC profile (Figure 1) represents the mean values from eight filter
sets collected from a total of 35 homes. A variety of woodstoves and
fireplaces are included in this composite. The most commonly burned woods
were Douglas fir, lodgepole pine, larch, and ponderosa pine. If the
homeowner had burned household trash within an hour previous to the sampling
visit no sample was taken. The mean ambient temperature was -6°C during the
sampling. While Figure 1 illustrates the overwhelming importance of
quantifying the organic carbon (OC) and elemental carbon (EC) of RWC source
samples and of the fine fraction of ambient aerosols in areas where a RWC
impact is suspected, the elemental composition of RWC emissions too low to
be seen in Figure 1 is important as well for CUB modeling. The key RWC
elements in decreasing weight percent in the composite fingerprint feres K
(0.35%) s CI (0.18%); S (0.143!); and Ca (0.13%).
The Btreet dust profile is, as mentioned, a composite of three city
streets in the vicinity of the ambient monitoring site* The samples were
collected prior to the time winter sanding was started. Figure 2
illustrates the chemical composition of the coarse (2.5-10ju) fraction. The
principal features of both the fine and coarse road dust fingerprints are
the presence of organic snd elemental carbon and the presence of the
geological species of Al, Si, K. Ca. Ti, Fa, and carbonate carbon (CC). In
general, the chemical composition of the fine and coarse fractions are
relatively similar, except that! (1) There is more than twice as much
organic and elemental carbon in the fine fraction; and (2) Measurable
carbonate carbon waa found in the coarse fraction, while none was found in
the fins fraction. The composition of the city sanding material was very
similar to the road dust. with the exception that the elemental and organic
carbon contents were much lower. This suggests either that much of the road
dust originated from the paved streets waa derived from the previous year's
sanding, or that the chemical composition of the local street dust is
similar for a geological reason to that of the sanding. An increased
organic and elemental content would be expected, with time, due to vehicular
exhaust, tire wear, asphalt, and vegetative fragments. It should b* noted
that the current source of sanding material is blgUy-weathared tertiary
deposits with a high fine content (12.9 percent less Chan 78 ii) which is
predominantly clay and has a low durability index (34).
693
-------
Figures 3 and 4 illustrate the chemical profiles of two small wood-
products point sources in the Missoula airshed. The chemical composition of
emissions from a hog fuel boiler burning pulverized ponderosa pine with a
cyclone emission control unit is illustrated in Figure 3, The composite
fingerprint is the average composition of three samples. The key features
of the fingerprint are the high elemental carbon content and relatively high
organic carbon, potassium, and calcium contents. Figure 4 shows the
chemical composition of the fine fraction of emissions from a particle board
dryer which utilizes the exhaust from a gander dust burner for a heat
source. The system is fitted with several cyclones. The composite
fingerprint illustrated in Figure 4 is the average of two samples collected
from two different cyclone exhaust ports. The key features of its
fingerprint are the high chlorine (chloride) content and the relatively high
organic carbon, sulfur, potassium, and calcium contents. The chemical
composition of both the fine and coarse fractions are quite similar, except
the coarse fraction has more than twice as much organic carbon.
Two types of vehicular exhaust fingerprints are being developed. These
are for heavy diesel trucks (a dies el train vas also sampled) and for
gasoline cars and light trucks. A composite fingerprint for diesel trucks
has been developed. The three principal chemical species are organic carbon
(88%), elemental carbon (8%), and sulfur (0.7%). All other elements were
found to be at levels of less than one tenth of one percent. The gasoline
car and light truck sampling is currently in progress.
AMBIENT SAMPLING
The ambient monitoring site was Rose Lawn Park, which is in a
residential area near a commercial district. Dual dichotomous samplers were
operated simultaneously for 24 hours once every 6 days during the winter.
Teflon filters were used with one sampler and quarts filters with the other.
Based on the particulate mass concentrations determined from these filters
and meteorological records, a subset of filters will be selected for
detailed chemical analyses. The same chemical species as were measured on
the source sample will be measured on the ambient sample to optimize CMB
modeling calculations.
CONCLUSIONS
The chemical profiles which are being developed for principal
particulate sources in the Missoula, Montana, airshed are distinctively
different enough to permit quantitative CMB source apportionment modeling to
be conducted. Chemical species were selected for analysis based on their
usefulness in distinguishing the impact of major sources characteristic of
Missoula. These include 27 elements measured by XKF; carbon species
analysis; and analysis of water soluble metals^ (viz. Na and K), which is in
progress. Unlike many CMB modeling studies, equal resources were dedicated
to measuring the chemical composition of both source and ambient samples in
the airshed.
REFERENCES
1. L. C. Hedstrom, "Particulate Source Apportionment in Missoula, Montana,
a Comparison of Two Winters, 1979-80 and 1982-83," M.S. thesis,
University of Montana, 1986.
2. J. E. Houck, J. A. Cooper, and E. R. Larson, "Dilution Sampling for
Chemical Receptor Source Fingerprinting," paper 82-61M.2 presented at
the 75th Annual APCA Meeting, 1982.
3. J. W. Buchanan, S. Li, and C. Calloway, "A Refinement of the Potassium
Tracer Method for Residential Wood Smoke," paper 86-77.6 presented at
the 79th Annual APCA Meeting, 1986.
694
-------
EC CC Si S K Ti Cr Fe Hi Zrt Se Rb Zr Sb Hg
OC A1 P CJ Ca U Mn Co C« As Br Sr Ci 6a Pb
Figure i. RWC (010100081,-2,-3,-8,-IB,-11,-13,-14)
10-
Q. j-
t-
EC CC Si S X Ti Cr Fe Ni 2n Se Rb It Sb Hg
OC Al P CI Ca U Kn Co Co A* Br Sr W Ba Pb
Figure 2. STREET BUST/COARSE (ST0208802.-1-6)
695
-------
10-
EC CO Si S X Ti Cr Fe Ni Zn Se Rb Zr Sb Hg
OC AI P CI Ca V Mn Co Cu As Br Sr Cd Ba Ph
Figure 3. WHITE PINE HOG FUEL BOILER (018108085,-6,-7)
10-
IC CC Si S X Ti Cr Pe Ni Zn Se ft Zr SI Hg
OC A1 P CI Ca V Mn Co Cit As Br Sr Cd B« Pb
Figure 4. LP CHIP DRYER/FINE
-------
FIELD INVESTIGATION OF CATALYTIC AND LOU EMISSIONS
WOODSTOVE PARTICULATE EMISSIONS, EFFICIENCY AND SAFETY
Paul Burnet
OMNI Environmental Services
10950 S.W. Fifth Street
Beaverton, Oregon 97005
Steve Morgan
Technical Development Corporation
11 Beaoon Street, Suite 1100
Boston, MA 02108
697
-------
EXECUTIVE SUMMARY
As the Northeast Cooperative Woodstove Study in New York
and Vermont nears the conclusion of its second heating season,
interim data and findings from the first heating season have
been analyzed and reported. The purpose of this paper is to
summarize the data and findings from the first heating season
and describe methodological changes undertaken in the study
during the second heating season, completed in April 1987. The
major objective of the Northeast Woodstove Study is to
determine the effectiveness of catalytic and low emissions
stove technologies in reducing wood use, creosote and
particulate emissions. Measurements were taken in 66 homes for
gross wood use and creosote aaoumulation and 32 homes for
particulate emissions and directly measured wood use. These
date and findings are preliminary and for use in indicating
potential directions of, and changes to, the study; the
contents of this paper are preliminary and not intended for
dissemination.
While catalytio and low emission stove technologies reduce
particulate emissions from 75% to 90? compared to conventional
stoves in laboratory tests, results from our first year in the
field performance are not nearly as Impressive. In the first
year of the study, particulate emissions in catalytic stoves
averaged 24.3 grams per hour, or about 35} lower than
conventional stoves. Oregon certified non-catalytic stoves,
based on very limited data, averaged emissions levels about 50}
lower than conventional stoves. The sample of only three
stoves in this group very much reduces the significance of this
finding. The performance of add on/retrofit devices revealed
no significant difference than that of conventional stoves.
Catalytio and low emission stove technologies appear to
reduce oreosote accumulation by approximately a factor of 2 in
comparison to conventional airtight stoves. Average oreosote
depositions in chimneys serving oatalytio stoves were .51 kg
creosote per thousand heating degree days.
Wood ubs in catalytio stoves averaged about 25} lower than
in conventional stoves. This finding is consistent with
laboratory efficiency tests for these technology groups. The
performance of add-on retrofit stoves in wood use was not
significantly better than conventional stoves. Thirty-five
oatalysts were investigated in the first year study.- Of these,
6 physically deteriorated, while another 6 oatalytio devioes
experienced operational and other problems. Laboratory testing
of 3 oatalysts which failed in the field revealed disparate
evidence about the extent to which catalyst combustor, stove
technology, and/or operator behavior contributed most to the
failures.
698
-------
x> Background
The Coalition of Northeastern Governors (CQnEG) oontraoted
OMNI Environmental Services, Inc. to conduct a two-heating
season study of residential woodburning stoves. This study
addresses the performance and efreotiveness of stoves equipped
with catalytic combustors relative to other woodburning
appliances. A total of 66 homes were examined in the study; 33
in the Glens Falls area of Hew York State, and 33 in the
Haterbury, Vermont area. Project sponsors inolude:
o U.S. Department of Energy;
o U.S. Environmental Protection Agency (EPA);
o New York State Energy Research and Development Authority
(NYSERDA);
o Twelve stove and appliance manufacturers;
o Vermont Agenay of Environmental Conservation (VAEC); and
o New York Department of Environmental Conservation (NYDEC).
A variety of performance parameters were examined In the study:
o Emissions characteristics of catalytic and non-catalytic
stoves;
o Wood use characteristics of oatalytio and non-catalytic
stoves;
o Creosote formation in flues serving oatalytio and non-
catalytic stoves;
A variety of integrated oatalytio stoves, add-on/retrofit
devices and low emission non-catalytic stoves were provided by
participating manufacturers to the study for installation in
the study homes. Study participants were volunteers selected
from lists provided by the VAEC and NYDEC. Sixty-six homes
Participated in the study. A group of 32 homes, divided
squally between the two study areas, was designed as Qroup 1.
These homes were equipped with computerized data reoords and
omission samplers. Four additional homes in this group were
available as backups. A group of 21 homes, designated Qroup
II, was monitored for wood use and maintained user log books.
Qroup ill homes, totalling six, had existing catalytic stoves,
and were measured for wood use and emissions for a 1-week
Period. All the study homes had pre*, mid- and post-season
woodpile volume, speolea, and moisture measurements. Creosote
Velghlngs were oonduoted on sweepings from the flu«s of all
study homes.
Most Qroup I homes had nev stoves installed at the
beginning of the study, whioh were to remain in place for the
duration of the projaot. Several existing stoves were left in
place to serve as baseline stoves. Host Qroup II homes will be
Provided with new stoves for the second heating season of the
Project, having used their existing conventional stoves for the
first season. Qroup III homes were existing oatalytio stoves,
and no change of stove technology is to b« oonduoted.
699
-------
The 1985-86 heating season had the following stove
technologies by study group:
Catalvfclo Aflti-On/Retroflt Low Emlaalona Conventional
Group
I
14
12
3
7
Group
II
3
0
0
21
Group
III
6
0
0
0
Total
23
12
~3
28
The stove technology categories were composed of several
different stove models, as identified below:
Stove
Technology
Tvne
Catalytic
Stove
Model
Tvatg
Camtanta
Four manufacturers provided new stoves
for the study. Group III homes
represented three additional models.
Catalyst stoves are defined as having the
oombustor as an integral part of the new
stove.
Add-On
Add-on devices are defined as units which
oan be added to virtually any stove at
the flue collar.
Retrofit
Low Emission
Conventional variety
Retrofit devices are designed to fit one
stove model or design type, and typioally
are close-coupled to the stove.
Low emission stoves are defined for this
study as models which have been certified
under the Oregon DEQ program. A third
stove was to be lnoluded in the study,
but manufacturer withdrew unexpectedly in
November, 1985, reducing the number of
low emission stoves in the study.
Existing stoves in study homes,
representing a range of designs.
Generally oategorlzed as typical of
conventional woodstove technology.
700
-------
II. Ea ul Dm ant
Two major objeotives of the study Involved defining wood
use and atove performance under typical home and activity
conditions. In addressing these objeotives two instruments
were developed: the Data LOQ'r system and the Automated
Woodstove Emission Sampler (AWES),
A. Data LOQ'r System
The Data LOQ'r is a programmable micro-
processor/ controller with the oapability of receiving,
processing, and recording digital and analog signals. As used
in this study, it oan aooomoodate over 30 days of data between
servioing. The Data LOQ'r was programmed to reoord and store
the following information:
o Starting date, time, and unit (Data LOQ'r) serial
number;
o Daily date and time;
o Flue and oatalyst temperatures every 5 to 10 minutes;
o Room temperature, outdoor temperature, and unit
temperatures recorded,and stored in 15-minute
intervals;
o Reoord the alternate home heating system status (on or
off);
o Wood weights and ooalbed condition reoorded on fueling
the atove;
o Reoord oxygen measurements when AWES sampling; and
o Reoord powerllne status every 5 minutes.
The Data LOQ'r was also programmed to control the AWES
sampler, turning the AWES sampler on and off in the proper
sequence.
Wood weights were measured on an eleotrohio soale with an
attaohed woodbasket. Soale readings were reoorded by having a
study participant use an attaohed keypad in a prescribed
sequenoe. flue temperature, oatalyst temperature (where
applicable), ambient room temperature, and auxiliary heating
system on/off conditions were reoorded at 5-10 minute
Intervals. Figure 1 us hows a diagram of the Data LOQ'r and
woodbasket soale equipment.
b. AWES gam pi era
The AWES Sampler is an automated sampler whioh allows the
determination of total particle and/or semi-volatile emission
rates. It oonsiats of a sample probe whioh is Inserted into
the stove or ohimney, followed by a heated filter, an organio
absorbent (XAD-2 resin)* silioa gel, a flow-regulating orifice,
and a pump. Some samplers are equipped with oxygen senior
cells to record oxyg«n concentration in the flue gts. Figure 2
shows a sohematio of the AWES sampler. The samplers used in
*Figure 1 not included in this publication.
701
-------
this study were designed to operate intermittently without
servicing for 1 week, 1 minute on, 29 minutes off. The
resulting sample represents an integration of stove emissions
during the sampling period. Simultaneous samples were obtained
at up to two of four potential looations: before the catalyst,
after the catalyst, at the flue oollar, and at the ohimney
exit. Data from samplers at the flue oollar and after catalyst
locations were compiled for this report for the purpose of
comparing stove technologies.
ixi. Creoaote Measurements
a. Methodology
Dooumentation of creosote accumulation in study home flues
was conducted by measuring the net amount of material removed
from each chimney by sweeping. Chimneys were swept at the
beginning, middle, and end of the 1985-86 study by professional
ohimney sweeps. Creosote dislodged by the sweeping was
colleoted by the ohimney sweeps. The first sweeping was to
establish "clean" conditions, while creosote collected in the
second and third sweepings was weighed. The mass of creosote
collected was then normalized by heating degree days (HDD)
occurring during the creosote accumulation period between
sweepings. Creosote samples from the mid-season sweeping for
all study home chimneys have been sent to the Solar Energy
Researoh Institute (SERI) for ohemioal analysis.
The heating load, in HDD, during the study period was
calculated for each home. Heating degree days were summed for
the period between creosote sweepings, yielding the heating
load for the specific period for eaoh home. Data were used
from the Waterbury and Qlens Calls weather reoording stations
maintained by the Northeast Regional Climate Center. Qlens
Falls data were used for New York homes and Waterbury data were
used for Vermont homes.
B. Results
Creosote measurement data were compiled for individual
homes and by stove technology grouping. Results are summarized
in Table 1. Data are reported in kilograms of creosote (total
mass of material removed from the ohimney by sweeping) and are
normalized by HDD. Frequency distributors of normalized
oreosote accumulations are shown by stove technology grouping
in Figures 3 through 6.*
Both oatalytio and low emission stoves showed similar
results. Mean oreosote accumulations of 0.51 and 0.55 kg/1000
HDD were obtained from the catalytic and low emission
installations, respectively. The oatalytio value exoludes
study homes V05, whloh was disoovered to have an Improperly
seated oombustor, which allowed combustion gases to bypass the
oombustor. The oombustor had apparently been Jarred out of
*Flgures 3-6 not included in this publication.
702
-------
position during stove installation. The creosote aooumulation
value for V05 (with improperly seated combustor) was 2.77
kg/1000 HDD, which is nearly three times higher than the next
highest value from a catalytic stove.
Catalytic stove values showed a minimum of 0.19 and a
maximum of 0.93 kg/1000 HDD (exclusive of V05 data), with a
standard deviation of 0.24. Low emission stove values showed a
minimum of 0.16 and a maximum of 1.33 kg/1000 HDD, with a
standard deviation of 0.51.
Conventional stoves appear to have lower average rates of
creosote aooumulation than the add-on/retrofit devices. Mean
values of 1.07 and 1.26 kg/1000 HDD were recorded for the
conventional and add-on/retrofit groups, with standard
deviations of 0.68 and 0.94, respectively. The conventional
stove groups had a minimum value of 0.06 and a maximum of 2.91
kg/1000 HDD. The add-on/retrofit group had a minimum value of
0.37 and a maximum of 3.60 kg/1000 HDD.
The high degree of variability in oreosote aooumulation
data (as demonstrated by the high standard of deviation values)
is due to a variety of factors, inoluding stove operation
practices, burn rates, and flue height, sise and construction
(inside vs. outside wall, solar exposure, etc.). It should
also be noted that oreosote is primarily condensed organio
material, which may be revolatilized by high flue gas
temperatures. Creosote mass in the chimney can therefore vary
on a continual basis. These values are intended only to serve
as a general indication of creosote aooumulation by the various
stove technology groups.
ZV. Seoond Year Testing
The second year of testing will provide significantly new
and more valuable data for the study. Twenty-four households
switched to a different stove teohnology--oonventional to
catalytio or low emissions or vice versa. By this switch most
of the "noise" created by operator behavior will be eliminated
as a variable. Seoond, the projeot addressed safety
considerations by allowing additional chimney sweeping and
making renovations to ohimneys which might have been hazardous.
Third, the oontraotor rearranged the sampler locations.
Automated Woodstove Emissions Samplers were removed from
ohimney tops and replaoed at the flue collar or above the add-
on oatalysts, depending upon stove teohaology. Finally, dual
samplers at several looations provided greater confidence in
the precision of the AWES data, while still allowing direot
comparison in performance to different stove technologies.
CONCLUSION
In August, CONEG will release the final report of the full
two-year study. At this printing, CONEQ, EPA, and NYSERDA are
703
-------
considering a third year study. The objective of another year
of testing are several in number.
The sponsors wish to investigate further the durability
and degradation of catalysts over time. Not only are the
sponsors interested in absolute longevity of the oatalysts, but
also in the reasons for the degradation. To what extent are
the combustors themselves, their interaction with stove
technologies, and operator behavior responsible for this
degradation? The implications for the EPA enforcement actions
and oonsumer behaviors will be significant over the next
several years.
The field study will also investigate the impact of
operator training on catalytic stove performance, providing an
operator manual and faoe-to-faoe training. Quality training
may have a significant lmpaot on field performance, closing a
considerable gap between laboratory and field testing on
emissions levels in particular.
For the first time, sponsors are oonslderlng an
Investigation of indoor air emissions. Aneotodal evidence
suggesting that low emission non-catalytlo and catalytio stoves
generate as muoh or more indoor air particulates and POMs
prompt sponsor interest in measuring indoor air emissions.
Chimney conditions and their interfaoe with stove technologies
may also undergo study in this Indoor air investigation. Air
changes per hour will also reoelve attention to enable a study
of the impact of house tightness upon indoor air emissions and
stove performance.
The study sponsors are also Interested in testing the
relationship between burn rates and emissions. To do so,
multiple AWES oould be Installed In up to four homes to measure
stove emissions under different operating regimest one AWES
collects during high combuator temperature periods; another
during low oombustor temperatures. A third AWES oould record
when the bypass lever is open or olosed. To improve oatalyst
efficiency calculations, the sponsors are considering the
installation of a second oxygen sensor to an AWES pair, thus
permitting calculation of mass concentration, emissions rates
and emission faotors before and after the combuator.
As EPA moves to implement the New Source Performance
Standard, these and related questions will have a major lmpaot
on the consumer and stove manufacturers. The research
community oan make a substantial impact on issues affeoting the
evolution in stove design and consumer operation of oleaner-
burnlng wood appliances.
704
-------
Table i. Comparison of Performance by Stove Technology Croups
M4*«n/
F»rtscutate
{MtfSitfl*
Cf/lirt
*#.0 -
C*t*lytic
«.• -
I
I
I
\\\\\\N
P»rttc«l**t iitJff i*n*
km* a*t
Cr«MtU
l«M InljfUn
Cenvtntlentl
. I«»4 Oft
(fcf/ROD)
I
- tritttn
(kf/lnOOl
- o.co
I
I
I
I
-------
Figure 2
CHIMNEY EXIT SAMPLER SITE
CHIMNEY CONNECTOR PIPE
AFTER CATALYTIC
ADD-ON SAMPLER
SITE
FLUE COLLAR
SAMPLER SITE
SAMPLER
FIREBOX
SITE
w
ASH CLEANOUT
DOOR
AWES SAMPLER LOCATION SITES
706
-------
THE ENVIRONMENTAL PROTECTION AGENCY'S ACCREDITATION
PROGRAM FOR WOOD HEATER TESTING LABORATORIES
Peter R. Westlin, J.E. McCarley, George W. Walsh
Emission Measurement Branch
Office of A1r Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
In the proposed regulations for certifying new residential wood
heaters, the U.S. Environmental Protection Agency (EPA) 1s planning to
accredit emission testing laboratories. The laboratory accreditation
involves both the EPA and the National Voluntary Laboratory Accreditation
Program (NVLAP), National Bureau of Standards. The NVLAP accreditation
requires the laboratory to complete a technical application which
describes the testing facility, personnel, and training procedures. An
Inspection of the laboratory by technical personnel 1s also required.
The EPA accreditation requires the laboratory to complete a number
of administrative Items and to complete a demonstration of proficiency
using the emission testing procedures defined In the regulation. EPA
will review the capabilities of the laboratory personnel to follow the
test methods and to prepare an accurate and complete report of the
procedures and results.
Laboratory accreditation 1s a new function for EPA and has resulted
1n a number of new tasks. These tasks include administering
applications, scheduling inspection visits, establishing proficiency
demonstration requirements, and reviewing test reports and results for
laboratory evaluation purposes. These tasks are designed to provide
standardization and consistency between testing laboratories In
conducting wood heater certification tests.
707
-------
INTRODUCTION
On February 18, 1967, the Environmental Protection Agency {EPA)
proposed performance standards and rules for new residential wood
heaters. According to the rules, wood heaters sold after July 1, 1990,
must be certified; that 1s, prototype wood heaters representing a model
line must demonstrate specific emission control capabilities over a range
of operations. Only accredited testing laboratories who have
demonstrated minimum administrative and technical capabilities may
perform the certification emission testing of wood heaters.
The purpose of this paper is to outline EPA's role 1n the laboratory
accreditation program and to discuss the experiences EPA has encountered
In implementing the program.
THE REGULATORr REQUIREMENTS
The proposed regulation lists specific requirements for laboratory
accreditation. The 11st includes application to EPA, satisfaction of
some administrative qualifications, and demonstration of testing
proficiency. The Emission Measurement Branch (EMB) of the Office of A1r
Quality Planning and Standards is designated to Implement the
accreditation program and provide the detailed procedures for completion
of these requirements.
A testing laboratory seeking accreditation must send a letter of
Intent to the EMB in Research Triangle Park, North Carolina. Upon
receipt of the letter, EMB will issue an outline of the specific
administrative and technical steps to be completed by the testing
laboratory.
The first step 1s to be accredited by the National Voluntary
Laboratory Accreditation Program (NVLAP), a program sponsored by the
National Bureair of Standards, Department of Commerce. The NVLAP
accreditation Involves completing a detailed application form. The
laboratory must describe the testing facility, the qualifications of the
personnel, the organization for certification testing, and the quality
assurance procedures. A technical expert representing NVLAP will conduct
an on-site visit to review the facility and determine Its compliance with
the requirements of the regulation. Tne NVLAP usually charges a fee for
conducting the accreditation evaluation.
The proposed regulation allows laboratories accredited by the State
of Oregon by January 1, 1988 (grandfathered laboratories} to apply for
EPA accreditation and forego the NVLAP accreditation.
The second step 1n the accreditation process 1s to demonstrate
proficiency with the test procedures used 1n certifying wood heaters.
The proficiency test demonstration requirements have been developed by
EMB, and these are being distributed to laboratories seeking
accreditation. An EMB technical expert will observe the laboratory's
techniques during the proficiency demonstration and determine whether the
specified procedures are followed.
708
-------
Step three 1n the accreditation process 1s to write a complete
report of the proficiency demonstration test and submit this report to
EMB for review. The test report format 1s specified 1n the test
procedures. An EMB technical person will review the test report for
completeness and accuracy and will determine whether the specified
procedures and calculations have been followed.
An evaluation of proficiency based on a comparison of emission test
results with some known results 1s not part of the proficiency
demonstration at this time. However, the emission data contained 1n the
proficiency report are compared. In general, with results from other
laboratories In order that dramatic errors can be Identified, Requiring
a laboratory to retest Is a possibility. A more statistically based
data review will be explored as more data on the predictability of wood
heater performance are developed.
The fourth step 1n the accreditation process 1s an EMB review of the
results of the on-site evaluation, the report review, and the
recommendations of the EMB technical experts Involved in the laboratory
evaluation. If the laboratory's procedures and report are found to be
acceptable, EMB will Issue a notice or certificate of accreditation to
the laboratory. Accredited laboratories then may conduct certification
tests of wood heaters according to the regulations and as long as the
procedures and personnel at the laboratory remain the same as during the
accreditation evaluation. The accreditation notice Includes expiration
dates that Indicate when the accreditation evaluation shall be repeated.
The proposed regulation includes a requirement for accredited
laboratories to participate 1n an annual demonstration of proficiency.
The EMB proficiency test requirements specify when this demonstration
must be completed each year and describe other details of the
demonstration procedures that may change from year to year.
EPA'S ACCREDITATION PROGRAM EXPERIENCES
NVLAP Accreditation
Accreditation of laboratories measuring the emissions from wood
heaters Is not an area with which NVLAP has had a great deal of
experience; NVLAP has accredited laboratories who determine safety
specifications for wood heaters, but has not accredited laboratories who
conduct comprehensive emission testing. The EPA has worked with NVLAP in
developing criteria for evaluating emission testing capabilities within
the usual NVLAP approach.
Administrative complications have developed 1n combining the tasks
of the two accrediting groups, NVLAP and EPA, and in establishing
responsibilities for the various functions. In addition, it has become
apparent that the best qualified persons available for the evaluation of
laboratory capabilities and of the proficiency test results are EPA
emission testing personnel within ENB or emission testing contractors
familiar with EPA test procedures. Because the budget of neither of the
two accrediting groups 1s sufficient to contract all of the accreditation
functions, the burden of the accreditation efforts has fallen upon the
EMB emission testing staff. At this time, EMB is conducting the
709
-------
accreditation program without NVLAP's assistance, and a section of the
preamble to the proposed regulation asks for comments on continuing
1n this fashion.
Test Proficiency Demonstration
The EPA accreditation group has prepared an outline describing the
requirements for demonstrating proficiency with the wood heater Test
Methods 5G, 5H and 28. Seven elements are Identified and defined. These
are:
1. Definition of a Testing Laboratory. The testing organization
and facilities including the key personnel, the testing shop, the
analysis area, and the testing equipment. The testing laboratory may
include capacity for more than one wood heater test at a time provided
the availability of key personnel and equipment reflects this capability.
1, The Demonstration Test. An emission test of a wood heater
selected by EPA and obtained Independently by the testing laboratory.
There shall be a minimum of eight test runs, two in each of four burn
rate categories. The testing laboratory shall follow Method 28 and
either Method 5G or 5H (or both) and operate two particulate sampling
trains simultaneously. The testing laboratory shall use the same
sampling method as will be used for certification testing.
3. Selected Wood Heater. The Catalytic Fir, AK-18, manufactured
by Sweet Home Stove Works, Inc. The criteria for selection includes
current Oregon certification, relatively small firebox volume (for
shorter-run times), catalyst equipped, and a burn rate range that spans
that specified 1n the method.
The wood heater selected for the proficiency demonstration may be
different in succeeding years. Notification of the selected wood heater
and any other changes to the demonstration procedures will be Issued by
EM8 by October 31 of each year, and the annual proficiency demonstration
must be completed by April 15 of the following year,
4. Notification and Observations. Notification to EMB at least
30 days prior to the start of the proficiency demonstration. An EPA
observer may be present during part or all of the demonstration.
5. Reporting Requirements. A complete test report of the
proficiency demonstration as specified 1n Method 28. Results of the
duplicate train values are reported 1n a separate table. The report
should address the criteria for deleting or Including test run results as
specified 1n the methods, but no data collected during the proficiency
demonstration shall be deleted.
6. Participation and Delivery Dates. Required for all testing
laboratories, including the Oregon grandfathered facl1Itles, for the
Initial proficiency demonstration as described. The report of the
Initial proficiency demonstration results for the grandfathered testing
laboratories 1s due April 15, 1987. Other testing laboratories shall
complete the proficiency demonstration and the reporting before an
accreditation certificate 1s Issued by EMB.
710
-------
EPA Response. Review of the test report for compliance with
the requirements 1n the test methods. A written evaluation will be
issued by EMB within 30 days of receipt of the report. A certificate of
accreditation will be issued upon satisfactory completion of the
accreditation requirements.
The four grandfathered testing laboratories and others who have
Indicated an interest In pursuing the EPA accreditation have received
these directions. A number of laboratories have completed the testing
portion of the demonstration.
One purpose of the EPA review 1s to provide guidance to the testing
laboratories 1n achieving consistency and some standardization In the
calculation and presentation of test results. The EMB technical
personnel use a comprehensive check list fn revlewfng the reports. This
check list covers the completeness of the descriptions of the test
facility, the wood heater, the test procedures, and the summary of the
results. The reviewer checks the calculations and results In detail
using the raw data provided 1n each report. The reviewer notes
discrepancies and brings these to the attention of the testing
laboratory. In some severe cases, a testing laboratory may be asked to
submit a revised test report.
Another purpose of the test data review is to satisfy a section of
the proposed regulation that requires EPA to evaluate the precision of
repeatability of wood heater emission test data. By January 1, 1990,
EPA 1s to document the statistical parameters to be used fn reviewing
both accreditation test data and wood heater certification test data; or,
EPA will determine that the data available at that time are Insufficient
to establish such repeatability values.
Review of Proficiency Test Results
The proficiency test program provides EPA with a unique opportunity
to compare results collected with the certification procedures as
conducted by different laboratories under conditions simulating actual
certification tests. The EMB has only begun collecting the emission test
data from the testing laboratories and 1s not yet prepared to conduct
statistical evaluations of the test results. The EMB will likely conduct
a rigorous statistical evaluation of duplicate train results as well as
comparison of results between laboratories. Establishment of acceptance
criteria for laboratory test results will enhance the efforts towards
standardization of wood heater certification testing.
SUMMARY
The proposed wood heater regulation has imposed on the EPA a new
responsibility of testing laboratory accreditation. The EMB of the EPA
has undertaken this task which Includes administering applications,
scheduling Inspection visits, establishing proficiency test demonstration
requirements, and reviewing the test reports and results. The EPA will
also review the proficiency test data fn order to establish acceptance
criteria for the results. These tasks are designed to provide
standardization and consistency between the testing laboratories in
conducting the wood heater certification tests.
711
-------
IMPORTANCE AND BENEFITS OP DEVELOPING DATA QUALITY OBJECTIVES
FOR AIR TOXICS DATA COLLECTION ACTIVITIES
Gary L. Johnson and Stanley M, Blacker
Quality Assurance Management Staff
U.S. Environmental Protection Agency (RD-680)
Washington, DC 20460
ABSTRACT
Implementation of the EPA Air Toxics Strategy during the next five
years will provide several activities which will benefit from the Data
Quality Objectives (DQO) process. The DQO process is an approach to
designing environmental data collection programs that incorporates consid-
eration of the use of the data in decision-making, Che quality of the
data needed for the decision, and the available time and resources for
the data collection. Accordingly, data collection programs will more
likely meet the needs of decision makers in a cost effective manner when
the DQO process is used.
The current Air Toxics Strategy calls for an increased State and
local role in data collection and in regulation, and an increased Federal
role in technical assistance. This paper will describe important elements
of the DQO process and will identify several areas of the Air Toxics
program where the DQO process may be applied. For one selected area, the
paper will address how to apply the DQO process and will discuss the
benefits to be derived by following this process.
712
-------
INTRODUCTION
Environmental data play a critical role in many EPA decisions. In
view of the importance to the Agency of such data, the process used to
design data collection systems should place heavy emphasis on defining
the regulatory (or programmatic) objectives (Why are the data needed?),
the decision to be made from the data (Bow will the data be used?), and
the possible consequences of the decision being incorrect (What is the
risk of making a mistake?). A design process that fails to explore these
issues and focuses only on collecting the "best data possible" can result
in potentially serious problems, particularly if those who define the
"beBt data possible" base their definition on the characteristics of the
data themselves and do not consider the program decision to be made. The
"best data possible" do not always yield the data needed to make the
required program decision.
The Quality Assurance Management Staff (QAMS), in response to a
requirement established by the Deputy Administrator in May 1984, has
proposed an approach to environmental data collection program design
based on the development of Data Quality Objectives (DQOs). This process
provides a logical, objective , and quantitative framework for finding an
appropriate balance between the time and resources available to collect
the data and the quality of the data needed to make the decision.
The DQO process is currently being applied to several new Agency
data collection programs whose results will impact important regulatory
decisions. One of the major environmental issues being addressed by EPA,
and one to which the DQO process may be particularly helpful, is the Air
Toxics program.
In June 1985, EPA announced a strategy for the control of toxic air
pollutant emissions from routine and accidental releases. The routine
releases component of the Air Toxics Strategy focuses on the joint efforts
of Federal, State, and local programs to deal with air toxics problems*
The strategy builds upon existing programs developed under the Clean Air
Act, augmented by expanded regulatory decision making by the States and
the use of other Federal authorities as appropriate* The nature of the
air toxic* problem is complex and, as such, the problem has been addressed
through several comprehensive programmatic approaches. These programs
inc lude:
* Federal Regulatory Program for Air Toxics
* High-Risk Point Sources Program
* High Urban Risk Program
* Enhancement of State and Local Air Toxica program
* Enforcement and Compliance Programs for Air Toxic*
Some approaches will utilise existing data bases to identify
sir toxics concerns; however, there will be needs for additional, spe-
cialised date collection by EPA, the States, and local authorities. As
will be discussed later, new environstenta1 data collection programs
initiated under the Air Toxics Program will be more cost-effective to .
713
-------
Agency and other decision makers when the programs are designed according
to the DQO process.
Overview of DQOs and the DQO Process
Data Quality Objectives are statements of the level of uncertainty
that a decision maker is willing to accept in results derived from envir-
onmental data, when the results are intended for use in a regulatory or
programmatic decision, such as, deciding that a new regulation is needed,
setting or revising a standard, or determining compliance. In order to
be complete, these quantitative DQOs should be accompanied by clear state-
ments of:
* the decision to be made;
* why environmental data are needed and how they will
be used;
* time and resource constraints on the data collection;
* descriptions of the environmental data to be collected;
* specifications regarding the domain or universe of the
decision; and
* the calculations or analyses that must be performed
on the data to yield information in the form needed
for the decision.
These statements suggest that a stepwise process may be defined for each
decision and data collection activity that incorporate each of the items
above. Moreover, this process integrates logically with other, traditional
quality assurance program elements.
Role of DQOs in the QA Program
The development of DQOs should be Che first step in initiating any
significant environmental data collection program. The DQO process helps
to define the purposes for which environmental data will be used and sets
guidelines for designing a data collection program that will meet the
Agency's programmatic or regulatory objectives. Once DQOs have been
developed and the design of the data collection activity has been com-
pleted, DQOs are used to define quality assurance (QA) and quality con-
trol (QC) programs that are specifically tailored to the data collection
program being initiated. From the DQO, a Quality Assurance Project Plan
(QAPjP) is prepared which details the QC criteria for the data collection,
establishes data custody and analysis procedures, and provides for routine
examination of the activity through audits and reviews. Thus, considera-
tion of how. the information will be used in the decision-making process is
an important element of the QA management system that assures Che quality
of the data obtained will be sufficient to support the Agency decision.
More importantly, perhaps, the DQO and the QAPjP derived from the DQO
provide a measure or standard against which progress in the data collec-
tion activity may be evaluated.
714
-------
The DQO Process
The DQO process consists of three stages with several steps in each
stage* The first two stages result in proposed DQOs with accompanying
specifications and constraints for designing the data collection program.
In the third stage, potential designs for the data collection program are
evaluated. Stage III results in the selection of a design that is compat-
ible with the constraints of the program and should enable the program to
satisfy the goals of the DQO. The process is intended to he iterative
among the stages; that is, if the proposed constraints from Stage I, the
proposed DQOs from Stage 11, end the design alternatives analyzed in
Stage 111 are found to be incompatible, the process may be repeated until
the problems have been resolved.
A critical element of the DQO process is the genuine involve-
ment and participation of decision makers. DQOs are developed through a
top-down approach. The initial input and perspective of the decision
maker is essential to the successful development of DQOs. The role of
the decision maker may vary to some degree among programs from providing
input or direction throughout the program to reacting to or concurring
with options presented by key senior staff. Through their personal in-
volvement, however, decision makers can ensure that the DQO process is
used to properly design all significant data collection efforts. The
decision maker can best articulate how the data are to be used and define
any constraints on the data collection that are imposed by the nature of
the decision to be made.
STAGE I: Define the Decision
Stage I is the responsibility of the decision maker. The decision
maker states an initial perception of what decision must be made, what
information is needed, why and when it is needed, how it will be used,
and what the consequences will be if information of adequate quality is
not available. Initial estimates of the time and resources that can be
reasonably be made available for the data collection activity are .alao
presented.
STAGE II: Clarify the Information Needed
This stage is primarily the responsibility of the senior program
stsff with guidance and oversight from the decision maker and input from
the technical staff. The product of the Stage I process is carefully
examined and discussed with the decision maker to ensure that the senior
program staff understand as many of the nuances of the program as possible.
Next, the senior program staff discusses each aspect of the problem,
factoring Into the process any necessary considerations of technical or
policy issues* With the concurrence of the decision maker, specific
guidance for designing the data collection program are generated. The
results of Stage II include proposed statements of the type and quality
of environemtnal data required to support the decision, along with other
technical constraints on the data collection activity that will help to
converge the search for an acceptable design in Stage III. These results
are the proposed DQOs*
It is this stage of the DQO process that is generally overlooked.
715
-------
STAGE III: Design the Data Collection program
Stage III is primarily the responsibility of the technical staff,
but both the senior program staff and the decision maker are involved to
ensure that the outputs from Stages I and II are understood by all partic-
ipants in the process» The objective of Stage III is to develop data
collection plans that will meet the criteria and constraints established
in Stages I and II. All viable options should be considered and presented
to the decision maker for a final selection of the data collection program
design. The decision maker must ultimately determine the best balance
between the time and resources available to collect the data and the level
of uncertainty expected in the final results.
The DQO provides management with an effective tool for improving the
credibility and defensibility of regulatory or policy decisions based on
environmental data collected. The usefulness of such a management tool
can be seen through application of DQOs to one of the Agency's more
complex environmental problems, the routine release of toxic air pollutants
into the air.
Air Toxics Data Collection Activities
The Air Toxics Strategy proposes a significant change from traditional
approaches to air pollution regulations. While the mainstay to date has
been the National Emission Standards for Hazardous Air Pollutants (NESHAP)
program, the current strategy focuses on a joint regulatory role by the
Federal Government, States and local agencies. The State and local roles
will be augmented by focused technical guidance and financial assistance
from EPA. This approach also depends more on the use of existing and
emerging data bases compiled under other authorities, such as New Source
Performance Standards (NSPS) and the Superfuttd Amendment and Reauthoriz-
ation Act (SARA). The federal role will include the identification of
local situations of sufficiently high cancer risk that State or local
regulatory action is warranted. In addition, the problem of toxic air
pollutants in urban areas poses several complicated questions. Additional
environmental data are needed to help determine the contribution to the
"urban soup" of mobile sources versus stationary sources. If risk assess-
ment and other factors indicate that sn urban air toxics problem clearly
exists, then the decision on what to mitigate will depend on how well the
contributing sources to the urban problem can be identified and quantified.
While a number of data collection program# address different aspects
of the air toxics problem, all of them may be improved by application of
the DQO process. As an example to illustrate how DQOs may be applied to
air toxics data collection activities, the Toxics Urban Monitoring Program
will be examined according to the steps of the process outlined previously.
DQO Example: The Toxic Urban Monitoring Program
The Toxic Urban Monitoring Program is one of several efforts
to compile data on the nature of air toxics in urban areas. Some data
are available from earlier monitoring efforts (e.g., SLAMS/HAMS) in
which the samples collected for one purpose were analysed as well for
toxic chemicals. In addition, current programs, such as the Toxic Air
716
-------
Monitoring System (TAM8) network, are helpful, even though the scopes of
such programs do not totally address the needs of the urban program. The
Toxic Urban Monitoring Program is intended to examine the 30 largest
urban areaB in the country to provide additional data on the cancer risks
associated with the urban environment. In addition, this program will
supply additional data on the relative contributions of mobile and sta-
tionary sources to the urban mix. Through application of the DQO process
to this program, it is possible to define the issues that should be
considered so that Agency decision makers may have greater confidence in
the results of the monitoring program and the decision to be made.
STAGE I: Define the Decision
For the purposes of this example, it must be recognised that the
expertise of the Air Office decision makers will not be duplicated. The
discussion that follows is intended OKLY to demonstrate the PROCESS and
STRUCTURE of DQOs as they can be applied to the Toxic Urban Monitoring
Program. The actual design of the desired monitoring program remains with
the Air program office.
In thie particular example, a possible decision may be based on the
general objectives of the Toxic Urban Monitoring Program. For example,
the Federal decision might be stated as whether or not to mitigate the
mobile source contribution, depending upon the relative contributions of
mobile versus stationary sources. The steps in the DQO process and a few
of the issues and questions which may be asked during Stage I based on this
possible decision include the following:
(1) Describe the decision. Are we Really concerned about mobile
sources? Which mobile sources? What age or engine type should
we focus on?
(2) Describe the information needed. What data are needed?
Do we need to monitor source emissions or ambient concentra-
tions? Do we data from street level or building root level
or both? During what time period (i.e., season, time of day,
etc.) must data be collected?
(3) Define the use of the environmental data* How will the
monitoring data be used? Do we want the data to use for
evaluating change* to current regulatory programs that should
be considered. Are there inputs to the decision other than
environmental monitoring data? If so, how important are they
to the decision vis-a-vis the monitoring data?
(4) Define the consequences of staking an incorrect decision based
inadequate environmental data. Under what circumstances could
an incorrect decision be made? What are the implications of
making an incorrect decision? What happens if we decide to
mitigate mobile source* when the environmental data have
incorrectly indicated that a problem exists (i.e., a false
positive)? What happens if we decide not to mitigate mobile
sources when the monitoring data incorrectly indicated that
a problem does not exist (i.e., a false negative)?
717
-------
(5) Identify the resources and resources available. What resources
are available to carry out the monitoring program? Given the
available resources, does it make sense to make measurements
in all 30 urban areas? If not, which urban areas have the
priority?
STAGE II: Clarification of the Information Needed
The information produced by the decision maker should now pass to
the senior program staff. The Stage I outputs should, as a minimum,
include statements of the decision to be made, the information needed for
the decision, why the data are needed, how the data will be used, and the
constraints on time and resources. The senior program staff will do a
more rigorous analysis of the information provided by the decision maker,
even though the decision maker continues to have input to the process.
The DQO steps involved and the issues that may be raised are as follows:
(1) Break down the decision into decision elements. For every
decision, there are multiple factors or elements that influ-
ence the decision. Which of the elements to mitigate mobile
souTces depend on what environmental data? Which elements do
not?
(2) Specify the environmental data needed. For each decision
element dependent on environmental data, what specific data
are needed? Which of the many possible data elements should
be measured? What frequency of sampling is required? What
lab analyses are required?
(3) Define the domain of the decision; that is, the portion of the
environment, delineated by spatial and temporal boundaries,
from which samples will be collected and to which the decision
will apply. Should one focus on the industrial sectors of the
urban area or focus on where the majority of the population
spends most of its day? Should the indoor contributions be
estimated? What spatial coverage of the urban area is neces-
sary? What mix of mobile sources and stationary sources
is needed to differentiate the ambient contributors? What
temporal coverage is needed to distinguish between primary and
secondary formation of species?
(4) Define the result to be derived from the environmental data.
Should the results be a weekly average concentration? Should
the results demonstrate whether the specific analytes of con-
cern are influenced by meteorology? What statistics are
needed to summarise the data?
(5) Determine the desired performance. How certain of reality
does the result have to be for the data to be usable for
distinguishing mobile from stationary contributions.
(6) Determine the need for NEW environmental data. Are there any
existing data that have the desired characteristics?
(7) State the proposed DQOs. What is the level of false positives
and false negatives that is acceptable to the data user?
718
-------
To answer these questions requires an iterative process. This DQO
process should enable the decision maker, the senior program staff, and
the technical staff to think through the data collection requirements.
STAGE III: Design the Data Collection Program
The proposed DQOs produced by Stage II provide the necessary guidance
to the technical staff so they can design the appropriate data collection
and analysis schemes. Any of these schemes should provide the needed
environmental data in a manner consistent with the acceptable uncertainty
in the decision to be made. But, because of possible gaps in the measure-
ment methodology or of resource constraints, the data user may decide not
to choose one of the suggested schemes and may have to relax some of the
data quality requirements in order to achieve the desired product. These
iterations are desired and expected. The DQO process is designed to
encourage such necessary iterations.
At the conclusion of Stage III, a final design of the data collection
plan for the 30 urban areas in the Toxic Urban Monitoring Program will
emerge. The design will identify in quantitaive terms such items as the
number of sampling locations needed, where the locations need to be
sited, the pollutants to measure, the average time for sample collection,
the analyses required, the QC parameters associated with sample collection
and sample analysis, etc. Through such a design, EPA, the States, and
the local agencies should have a comprehensive and affective environmental
data base for assessing urban risk. Moreover, the program will provide the
necessary quality and level of confidence to support the mitigation
decisions that may arise.
Benefits of the DQO Process
Benefits of the DQO process to the staff and to the senior
managers may be summarized as follows:
* For the staff, they have a:
clear statement from the data user on what the user needs,
clear statement of the time and resource constraints, and
clear definition of the data quality needed by the data
user.
* For the manager or decision maker, he has:
thought through whether or not he needs new environmental
data to support his decision,
bought into the effort and is accountable,
a clear understanding of what he is likely to get and
when, and
a clear understanding of the limitations of the data he
is likely to get.
719
-------
Summary
DQOs are statements of the quality of data needed to support a
specific decision or action. The DQO process is an interactive process
that depends upon the active involvement of the decision maker, associ-
ated data users, and the technical design staff. DQOs will provide a
useful aid to managers faced with making important policy and regulatory
decisions by improving the confidence with which those decisions can be
made when they are data dependent.
The current Air Toxics Strategy involves the use of new data collec-
tion activities as well as existing data bases to define the nature of
the air toxics problem. This example of the application of DQOs to
designing the necessary environmental data collection program for the
Toxic Urban Monitoring Program provides a framework to assure that the
proper data quality can be obtained.
The concept of Data Quality Objectives is a fundamental one. No
data should be collected unless there is a rationale for doing so. This
should be the first step in any environmental data collection program,
and, it is for this reason that DQOs have become a central part of the
overall QA program in EPA. Without DQOs, the rest of the Agency's QA
program has no logical, quantitative basis.
720
-------
SOME NEGLECTED ASPECTS OF
ENVIRONMENTAL QUALITY ASSURANCE
Raymond C. Rhodes
Quality Assurance Specialist
Performance Evaluation Branch
Quality Assurance Division
Environmental Monitoring Systems Laboratory
United States Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Progress haB been made during the last decade in implementing quality
assurance In the Environmental Protection Agency. However, because there
are many aspects or elements of a total quality assurance system, It is,
perhaps, understandable that some of the elements have in recent years been
overemphasised at the expense of equally or more important elements.
Considerable efforts have been expended in Improving the quality of
chemical analytical operations la the laboratory* However, not enough
effort has been put Into increasing the quality of the field sampling.
When considering that the entire measurement system includes the sampling
and sample handling (if any) as well as the chemical analysis, the propor-
tion of variability or uncertainty In the final result that is attributable
to the sampling phase of the measurement system la much greater than that
for the chemical analysis* More effort needs, to be placed on reducing the
effects of sampling variables*
Considerable efforts are being expended In establishing goals or
objectives for data quality before a project oc program begins and in the
assessment of data quality during the project or program. More effort
should be placed in comparing the actual achieved quality with the goals
and In determining the cause of significant differences* Further, simply
setting goals and assessing data quality does nothing to improve the
quality. Much more effort needs to be spent in determining the major com-
ponents of variance of the total measurement system and la taking corrective
actions to improve data quality.
721
-------
Progress has been made during the last decade in Implementing quality
assurance (QA) for environmental measurements. Because there are many as-
pects or elements of a QA system, it is, perhaps, understandable that some
of the elements have in recent years been overemphasized at the expense of
other equally or more important elements.
The 23 elements of QA are briefly discussed in the "QA Handbook for
Air Pollution Measurement Systems, Volume I, Principles," EPA 600/9-76-005.
These elements are also categorized and discussed under four major applica-
tion areas — Management, Measurement, Systems, and Statistics — in the
Air Pollution Training Institute course, "QA for Air Pollution Measurement
Systems." The neglected aspects of QA are discussed below by considering
15 myths or misconceptions which seem to be prevalent today as they fall
under the four major application areas.
Systems
Myth 1. Quality Control (QC) is Achieved by Performing QC Checks. To
achieve control of a measurement process, periodic QC checks must be per-
formed. However, the mere performance of the checks will do nothing to
control or improve the process, unless the results are evaluated in timely
fashion and corrective action taken when indicated.
Myth 2. Performance Audits are QA, QA is Performance Audits. There
are those who tend to define QA as primarily the conducting of Performance
Audits — the submittal of carefully prepared and characterized materials
traceable as directly as possible to the National Bureau of Standards or
other recognized primary standard. Thus, the performance audit is an
external check on the accuracy of the measurement system being audited.
And if the auditee produces acceptable results, it implies that the routine
data from the auditees' measurement system is good.
Certainly the conducting of performance audits is a valuable aspect of
QA, and if the results are good, they indicate that the auditee is capable
of producing good results. However, such results should not be considered
as sufficient to Indicate the correctness of routinely obtained data.
Knowing the identity of the audit samples, the auditee may take extraordi-
nary care in their analysis.
To assure that the auditee does in fact produce correct routine data
requires much more than the performance audit data, e.g., It can require a
complete quality system audit, review of procedures and documentation,
review of routine QC data, etc. To do QA adequately requires that the
personnel Involved with the QA responsibilities be thoroughly knowledgeable
of the total QC System. In other words, one cannot perform an acceptable
function In QA without knowing QC.
Myth 3. Corrective Action is a QA Activity. An Important part of a
QA system is that of corrective action. This does not mean that the actual
conducting of an investigation into the causes of the quality problem and
implementing any corrective action is solely QA activities. It is a QA
responsibility to assure that a corrective action system Is established and
is implemented. Certainly, full-time QA persons may identify quality
problems, may analyze data to investigate possible causes, and may recom-
mend corrective actions. However, others may also identify quality prob-
lems, others may more appropriately investigate quality problems, and more
likely than not, others should take the corrective actions.
722
-------
Myth A. The Preparation of SOP's is a QA Activity. QA personnel
should be concerned that adequate standard operating procedures (SOP's) be
prepared for all routine operations of the total measurement system. How-
ever, it is not a QA/QC activity to prepare such SOP's. SOP's naturally
should be prepared by those who actually perform the operations, or by
others thoroughly familiar with all the details of the operations. QA
personnel should be concerned that a system for the preparation and control
of documented SOP's exists, that the SOP's are clearly written, and ade-
quately reviewed and approved and that there be a defined system for revis-
ing and updating the SOP's.
Myth 5. An Observation or Analytical Result That Comes from a Computer
Data Base Is Known Absolutely and Without Error. All too often data gets
into a computer data base without being fully validated, and the data,
replete to seven decimal points, are given to a statistician for statistical
analysis. It should go without saying that the analysis, and conclusions
drawn from the analysis, can be no better than the data. It is possible
that invalid conclusions will be drawn and that incorrect actions or deci-
sions will be made.
Statistics
Myth 6. QA is Statistics, Statistics is CjA. Some consider statisti-
cal QC as synonomous with QA. Statistical QC or statistical process control
has recently been given some emphasis by W. Edwards Deming, the American
who effectively introduced the concept and principles to the Japanese soon
after World War II. Although the application of statistical QC and many
other statistical techniques are an important part of a total QA program,
there are also many at least equally Important but nonstacistical elements
of QA.
Myth 7. The Performance of QA Assessment Checks Assures the Attain-
ment of Precision and Accuracy Goals. It is good to have precision and
accuracy goals, and it is good to perform assessment checks for precision
and accuracy. But the mere performance of the checks does not cause the
goals to be achieved. These checks tell you where you are; but they don't
necessarily get you where you want to be. To improve the precision or
accuracy requires corrective action. Even then you may not achieve the
goals because they have been set unreallstically tight considering the
inherent capability of the total measurement method. For the project
involved, the users of the data must be satisfied with the experienced
quality. In the future, either the method must be improved, or more rea-
listic goals must be set.
Myth 8. You Always Sample 10%. QA Project Plans are filled with
requirements that 102 of the samples be QC samples. The selection of any
such percentage is a cop-out for thinking about what number or percentage
of QC samples of various types should be used. In general, the percentages
of samples to be used for QC purposes should not be decided on a set per-
centage basis or even on a statistical sampling basis, but rather upon the
operational variables involved. These operational variables might Include
how frequently the measurement system is calibrated, how many samples are
Included in an analytical run, the number and type of analytical instruments
being used, the number of analytes being analysed, whether all the samples
of a given batch are being analyzed on the same day or different days, etc.
In short, the frequency of QC samples should have some basis in what In
723
-------
QC parlance is called "rational subgroups," i.e., subgroups of samples that
have been taken or are being analyzed under very similar, or homogeneous,
conditions.
Myth 9. QC Charts are Useful Only for After-the-Fact Data Analysis.
Because of the lengthy time required for some of the more complex and
sophisticated chemical analyses, the analyses of real world samples and the
corresponding QC samples are not available in a timely manner to effect any
real-time QC. Consequently, much QC data are only available long after the
samples have been taken. It Is true that some worthwhile data analysis can
be made long after-the-fact to document quantitatively the QC results of
the past. However, except for possible use of the data for long-term
control purposes, the real purpose of maintaining control charts to effect
real-time control is defeated. Some innovative ideas are needed to control
such measurement systems, such as more emphasis in controlling key opera-
tional variables of the measurement process, or by making some type of
simulation check.
Further, the calculations of control limits for blocks or batches of
data, including outliers or out-of-control data, and applying the limits
after-the-fact Co the game set of data is invalid. The probabilities which
normally would be incurred no longer exist. Naturally, if limits are calcu-
lated and applied as described above, most of the data points, even those
which should really be indicated as out-of-control, will fall within the
computed limits.
Measurement
Myth 10. An Equivalent Method is Automatically a Reliable Method.
The ambient equivalency regulations and procedures are excellent mechanisms
to assure that a measurement method has been demonstrated to be capable of
achieving specified accuracy and precision. However, they do not assure
continued operational success in field performance for extended periods of
time. Much more evaluation of the reliability of the sampling/analysis
operations in field operations needs to be done to determine failure
(malfunction) rates, to identify the major component(s) that fall, and to
determine the major causes of failure.
Management
Myth 11. QA/QC Applies Only to the Chemical Laboratory. There has
been a continuing emphasis on establishing and maintaining an extensive QC
system on the laboratory chemical analysis part of the total measurement
system. In fact, some feel that the sampling activities are not a part of
the measurement system and that QC applies only after the samples reach the
laboratory. The chemical laboratory may very well be in excellent control.
However, the end result depends as much or more so on variables in the
sampling part of the total measurement system.
Myth 12. QA Doesn't Apply to Measurement Method Development. Some
have said that QA doesn't apply to methods development, to research, or to
sampling/measurement equipment and material procurement efforts.
In reality, basic QA/QC principles can be applied to any identified
process having an identifiable product. Basic QA/QC principles are univer-
sal in their applications.
724
-------
Myth 13. Specifying QA Objectives Assures the Attainment of the
Objectives. Simply specifying QA objectives for precision, accuracy, and
completeness will not make it happen. The objectives must be realistically
attainable under the conditions under which the entire measurement system
is operating. The entire measurement system may include sample container
and equipment preparation, sampling, sample handling and preservation,
analysis, computations, and data archival—not just simply the laboratory
analysis part of the measurement s/Btem. QA objectives cannot be baaed
solely on the laboratory analysis or even on collaborative studies of
measurement methods.
Myth 14. Having a QA Project Plan Assures the Execution of QA Activ-
ities Outlined Therein. EPA requires the approval of a QA project plan
prior to the collection of samples for projects involving the collection of
environmental data. This Is good, but independent and indepth system
audits should be conducted to assure compliance to the requirements of the
plan. There has been at least one case where a major EPA contractor as a
part of a contract for a project prepared a very adequate QA Project Plan,
but did not perform in accordance with the plan. When challenged with this
fact, unfortunately after the project was completed, the contractor stated
that the contract only required them to prepare the QA Project Plan; it did
not specifically state that they were contractually bound to comply with
the plant And so it went.
Myth 15. Having a QA Policy< a QA Organization, and QA Coordinators
and Officers Means an Effective QA System Exists. It is good for an
organization to have a QA policy, to have working groups and personnel
labeled as QA. But the policy, the organization, and the personnel will be
ineffective in their QA endeavors unless all management levels from the top
down does more than support QA efforts with only lip service. Further, all
management levels should have some knowledge and training in QA principles
and practices.
All persons working in QA should have completed some formal training
in statistical QC and other QA courses such as quality auditing, QA for
computer software, systems reliability, process capability studies, accep-
tance sampling, design of experiments, procurement QA, quality engineering,
regression analysis, etc.
Much more planning and implementation of QA training is needed in the
environmental organisations* This training should include all of the 23
elements covered briefly in the Principles Volume of the QA Handbook, as
well as related administrative areas such as thi preparation of QA project
plans and SOP's, the development of contractual requirements for quality
and monitoring of QA of contracted projects.
Ideally, persons with a QA title and responsibilities should have had
many years of active experience in QA/QC work. They should have completed
college-level courses in statistical QC and desirably in other related
statistics courses as well. Or else they should have completed other
readily available courses in QA and statistical QC and furthered their
knowledge in all aspects of QA through disciplined self-study and reading.
Because of the relative newness of the field of QA or total QC, very few
universities have yet developed complete curricula toward the granting of
degrees In the field. Persons with assigned QA titles and responsibilities
should be aggressive in their efforts to become more knowledgeable in the
725
-------
field, EPA needs to support much more QA training, particularly for those
having QA titles and responsibilities.
Summary and Recommendations
There should be a very concerted effort in environmental monitoring
organizations to provide training at all levels for these performing QA and
related work. With respect to organization, persons performing QA work
should be those who have the most knowledge, experience, and interest in
QA.
Further, except in very small organizational units, QA officers and QA
coordinators should be dedicated full-time workers in QA without supervi-
sory responsibilities and should report on a staff level as high in the
organization as possible. More systems audits should be conducted of
internal projects as well as contracted projects. The audit team should be
headed by the QA officer or coordinator and assisted by the most QA and
technically knowledgeable persons from various parts of the organization.
More emphasis needs to be given to the QA requirements of contracts for
procurement of sampling and analysis equipment and materials as well as
contracts for conducting research and monitoring projects. Quality incen-
tive contracts should be implemented wherever appropriate. Data validation
techniques need to be further developed and implemented to assure the
validity of data prior to statistical analyses or other uses. Corrective
action efforts need to be strengthened to achieve improvement in data
quality. Without corrective action, setting of data quality objectives,
conducting of performance and systems audits, and assessments of precision
and accuracy are only record-keeping efforts which do nothing to improve
data quality.
The possible application of statistical analyses for QA is unlimited.
More ruggedness tests or other statistically designed studies should be
done for measurement methods development, including field operation vari-
ables. Components of variation studies of the entire measurement system,
including field sampling, Bhould be performed to identify the major sources
of variability so that efforts to improve the measurement system may be
directed to those areas with greatest possible benefit. Continued appli-
cation of statistics in areas such as control charts, process capability
studies, limits of uncertainty, etc., are needed.
More attention needs to be given to the reliability evaluation of
reference and equivalent methods. Evaluations could be made of laboratory-
type operations or better from real-world field operations*
The total QA efforts needs to be more progressive or active rather
than passive. It should be a challenging program with some healthy "bite".
Problem areas should be clearly Identified and vigorously attacked. Lastly,
a successful QA program requires painstaking devotion and attention to
detail by all those who may have an Impact on data quality.
Disclaimers
The opinions and judgments of this paper do not necessarily reflect
those of EPA management. Except where EPA is specifically mentioned, the
observations of this paper relate to Environmental QA in general.
726
-------
QUALITY ASSURANCE FOR PRECIPITATION COLLECTION
AND MEASUREMENT: THE EPA STATE-OPERATED NETWORK
W. Cary Eaton
Research Triangle Institute
Research Triangle Park, North Carolina
Seme I. Bennett
Quality Assurance Division
U. S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Research Triangle Park, North Carolina
Networks that collect and quantify the constituents of precipitation
1n the United States have increased In number and size In recent years.
Several networks have regional or national coverage and sample on a dally
or weekly basis. A number of states (presently, 11) participate In the
EPA-sponsored State-Operated Precipitation Network that collects weekly
samples. As networks have grown, so have the development and application
of quality assurance considerations to ensure that accurate, precise,
complete, and representative data are obtained.
The need for and benefits of on-site quality assurance audits and
technical assistance are discussed. Quality assurance protocols for system
and performance audits of wet deposition collection sites are reviewed.
The protocols address site characteristics, training, precipitation
collection and measurement equipment, sample retrieval and handling, system
audits, and the use of simulated precipitation test solutions.
Results from quality assurance studies of the 27-slte State-Operated
Network are given to illustrate the use of quality assurance protocols.
727
-------
INTRODUCTION
Wet deposition collection networks 1n the United States have grown 1n
number and size 1n the last decade 1n response to the need for data to
determine the extent and nature of addle precipitation. As the networks
have grown, so has the development of protocols and procedures for external
quality assurance. Benefits of quality assurance studies of networks are
these: provide assurance that established protocols for siting, sample
collection, and analysis are being followed; Identify non-complying
equipment to be fixed and operators who need further training; provide a
means of documenting the data base with respect to accuracy, precision,
completeness, and representativeness; allow the data quality of several
networks to be compared 1n a systematic way.
The 27-s1te EPA State-Operated Network, (SON) has been 1n place since
1982. This paper describes quality assurance procedures established for
this network and gives results from visits to the sites 1n 1985 and 1986.
Recommendations for Improvement are also given. The procedures are similar
to those developed and applied earlier bv two other networks - the Utility
Acid Precipitation Study Program, (UAPSP), that collect on a dally basis
(1) and the National Atmospheric Deposition Program/National Trends
Network, (NADP/NTN), that collects weekly samples (2).
QUALITY ASSURANCE PROCEDURES
Siting Criteria Checks
Each site's Installation and nearby surroundings are compared to
criteria given In urban- or regionally-sited station checklists designed
expressly for the State-Operated Network. Important criteria common to
either type station are: (a) type and height of nearby groundcover; (b)
distance of collector from obstructions; (c) distance of the collector from
sources of contamination and pollution; (d) proximity of human or animal
activities to the collector; (e) nearby land and water features.
System Audits
During the visit the following system audit topics are reviewed: (a)
sample collection equipment and procedures (including plastic bag bucket
liners used 1n the SON network); (b) site laboratory facilities and
cleanliness; (c) analysis procedures; (d) communication and recordkeeping;
(e) quality control tests and corrective action procedures. Technical
assistance 1s provided whenever an operator 1s uncertain of procedures or
when equipment Is 1n need of maintenance or calibration.
728
-------
Performance Audits
The accuracy of response of each site's collection and measurement
devices are tested during the visit. The audit techniques and tolerances
for the SON network are described 1n Table I.
RESULTS
Siting Criteria Checks
At least one variation from the designated siting criteria was found
for most of the SON sites. Many were minor and should have negligible
effect on the physical and chemical data base. The most Important siting
exceptions were these:
Collector/rain gauge obstructions. Several collectors and gauges were
near enough to trees, towers, or other equipment such that the object
projected onto the collector at greater than a 45 degree angle. Because
the sample catch and chemistry may be affected, it was recommended that the
samplers be slightly repositioned or that the obstructions be removed.
Proximity to sources* Several sites had sources nearby although the
criteria for separation was minimally satisfied. Examples are: asphalt
plant; animal stockade; and salt water. It was recommended that data users
be made aware of these facts and that the site be moved or the source
removed 1f possible.
Collector/rain gauge positioning. Ideally the collector and rain
gauge orifices should be at the same height above ground (-1 m) and be
separated by at least 5 but no more than 30 m. Several sites had the
collector and gauge positioned within 1 m of each other In the horizontal;
others were separated by several meters In the vertical dimension. It was
recommended that the rain gauges be repositioned and inoperative gauges be
repaired or replaced.
System Audits
System checks were made of the precipitation collector, the rain gauge
sample collection procedures, and the field laboratory. Results are noted
below.
Precipitation collector. Eleven of 27 collectors were higher above
ground than the prescribed 1 meter. This exception was due to placement on
729
-------
platforms or roofs to avoid snowpack, for security, and to avoid
obstructions. The effect Is thought to be minimal. Ten collectors were
Installed with the wet bucket facing N or E rather than the prescribed W.
If rainstorms approach predominantly from the S or SW, the collector itself
may Influence the collection process. Eight collectors lack the event
marker capability. Knowing when and for how long the collector Is open 1s
an Important quality control consideration. All but two sites had properly
seating bucket covers. A dry-side bucket was missing at one site; another
site had a deteriorated liner. Proper seating 1s necessary to keep out
dust.
Rain aauae. Variations from the desired distance from the collector
and height with respect to the collector were noted at about half of the
sites. Six sites do not have recording rain gauges,' several sites had
malfunctioning tipping bucket gauges.
Sample collection procedures. Proper sample collection procedures
were followed 1n virtually all Instances. It was noted that not all
operators visually check the sample for contamination before returning 1t
to the field laboratory. It was recommended this be done; otherwise
contaminants may dissolve and not be noted later.
Field laboratory. The field laboratories had adequate space, were
clean, and temperature-controlled. Operators and analysts were trained and
fatafllar with their duties. Variations in measurement technique were noted
and discussed. It was recommended that smaller sample allquots be used,
that the network's central laboratory have first priority for samples, and
that samples be shipped within three days of collection.
Performance Audits
Performance evaluations were made of: the rain gauge; the
precipitation collector? and the pH meter, conductivity cell, and solution
balance. Results are given 1n Table II and discussed briefly below.
Rain aauae. All thirteen weighing-type recording rain gauges agreed
within 0.1 Inch of the designated audit value over the range 0 to 5
Inches, Eight of thirteen gauges were out of tolerance above 5 inches.
However, this 1s a seldom-used range as the gauge bucket 1s usually emptied
before this depth 1s reached. Most of the gauges were recalibrated to meet
specifications.
Precipitation collector. All collectors operated when the sensor was
activated.The I1d-l1ner/bucket rim seal was adequate. A number of
problems were noted with sensor temperature range and control.
Conductivity meter. Meters at three of twenty field laboratories
varied by more than 4 uS/cm from the designated value. Results were quite
acceptable otherwise. The quality of delonlzed water in use at the sites
also met specifications.
730
-------
dH meter. Only two of twenty-one field laboratories had pH results
that varied by more than 0.1 pH unit from the designated value. The
average absolute variation was 0.054 pH unit.
Mass. Sixteen of the 19 balances checked agree within 1 g of the
designated value at a loading of 1646 g. The worst case would be a 10 g
variation corresponding to 1.2 percent at an 823 g loading or 0.3 percent
at a 3292 g loading.
CONCLUSIONS
The EPA-sponsored State-Operated Network 1s a weekly precipitation
collection network of 27 sites located primarily In the eastern and
southeastern United States. State agencies have provided personnel to
service the sites and laboratories, to analyze the samples, and submit them
to a central laboratory for analysis. The site supervisors, operators, and
analysts were found to be familiar with their duties, handled the
precipitation samples carefully, analyzed the samples accurately in most
cases, and seemed genuinely Interested 1n the network and the data.
A number of the sites should be Improved upon 1n terms of siting and
maintenance of sample collection and analysis equipment. Emphasis should
be placed on proper placement and operation of precipitation collectors,
Installation and proper operation of rain gauges, and standardization of
field laboratory techniques.
REFERENCES
1. W. C. Eaton, K. A. Daum, E. D. Estes, and F. Smith, "Quality Assurance
Results for the Utility Add Precipitation Study Program (UAPSP), 1982
to 1984," Transactions: Quality Assurance in Air Pollution
Measurements, APCA Publication TR-3, Pittsburgh, PA (1985), pp. 488-
499.
2, D, S. Blgelow, "Quality Assurance Report; NADP/NTN Deposition
Monitoring. Field Operations, July 1978 through December 1983."
National Atmospheric Deposition Program, Coordinator's Office, NREL,
Colorado State University, Fort Collins, CO, (August 1986).
731
-------
TABLE I. AUDIT TECHNIQUES AND TOLERANCE LIMITS FOR SON NETWORK COLLECTION AND MEASUREMENT METHODS
Measurement Device and
Audit Observable
1. Recording Rain Gauge
- Accuracy of precipitation
depth on gauge
2. Precipitation Collector
- Lid I1ner-bucket-r1m seal
- Sensor activation and
heating
3. Conductivity Meter
- Accuracy of response to
slMulated precipitation
samples
- Deionized water quality
4. pH Meter
- Accuracy of response to
slwlated precipitation
sables
5. Mass by Solution Balance
- Accuracy of response to
calibrated weights
Audit Technique or Test
- Use calibration weights
simulating precipitation.
- Determine lid drop distance
(bucket in - bucket out).
- Observe open/close cycle.
Determine temperature with
thermistor 5 min. after
activation.
- Have operator determine
conductivity of simulated
sample.
- Have operator determine.
- Have operator determine
pH of simulated precipita-
tion sample.
- Have operator determine
mass of weights In
the range 800 g to 3300 g.
Expected Results
- Agreement within + 0.10 inch of
audit weight.
- D1stance >3 mm.
- Ambient temperature, prior
to activation. Temperature
of 50-7
-------
TABLE II. PERFORMANCE AUDIT RESULTS: EPA STATE-OPERATED NETWORK, 1985-1986
Measurement Device Audited
1. Rain Gauge
- depth
2. Precipitation Collector
- lid liner/bucket
r1« seal
- sensor activation
and heating
3. Conductivity tteter
- accuracy of response
- delon1zed inter
quality
4. pH Neter
- accuracy of response
5. Solution Balance Mass
- accuracy of response
Nuaber
Audited
13
16
27
20
14
21
19
Range of Response;
Average Response-
All within * 0.1 Inch
fron 0 to 51 ndies.
Range 7 to 33
avg. 13.6 m.
Comnents
All respond and open
collector. Range 23
avg.
55° C.
100 C;
0.3 to 5.80 uS/cb; (a)
avg. 1.63 + 1.76 uS/cn.
0.35 to 3.80 uS/cn;
avg. 1.75 uS/ca.
0.01 to 0.18 pH unit; (*)
avg. 0.054 * 0.04 pH unit.
16 of 19 agree within
* 1 g at 1646 g loading.
Out-of-tolerance readings occured
on 8 of 13 but only at 6 Inches
or greater depth, a seldoa-used
range.
All net 3 m criterion of
acceptability.
Eight heated at all tines.
Six did not heat. Twenty-
one heat; of these 10 0
attain temperature <50 C,
three > 7
-------
QUALITY ASSURANCE PROGRAM FOR THE NON-OCCUPATIONAL
PESTICIDE EXPOSURE STUDY (NOPES)
Miriam Lev-On, Joseph C. Delwiche and George Colovos
Environmental Monitoring & Services,Inc.
Camarillo, California 93010
David E. Camann, J. P. Hsu and Herb Schattenberg
Southwest Research Institute
San Antonio, Texas 78284
Andrew E. Bond and Howard Crist
U.S. EPA, EMSL
Research Triangle Park, North Carolina 27711
A field study of non-occupational exposure to pesticides (NOPES) is
being conducted to estimate the exposutes of a statistically representa-
tive sample of urban population to selected pesticides found in the air,
drinking water, and food, as well as through dermal contact. The intended
use of the data collected through this study is to obtain the cumulative
frequency distribution of exposures to selected pesticides for each of
these pathways. The general methodology of the study calls for sampling
in predesignated homes, followed by laboratory analysis of the collected
samples.
The complexity of this study and the difficulties associated with
population surveys, sample collection, chain-of-custody, and analytical
speciation of very low level concentrations of 32 target compounds, has
required careful planning and the development of a detailed quality
assurance plan. This paper presents the QA activities undertaken during
the field sampling and analysis phase of the program in the Summer of
1986 in Jacksonville, Florida. The discussion highlights the data
validation steps, and results are presented for surrogate and matrix
spike recoveries, performance evaluation samples and audits.
*Now with: Roy F. Weston Company, Stockton, California
734
-------
INTRODUCTION
The general public is subjected daily to non-occupational pesticide
exposure via the air, drinking water, food and direct dermal absorption.
The U.S. Environmental Protection Agency is systematically investigating
the various aspects of non-occupational pesticide exposures by conducting
field studies in Jacksonville, Florida for three seasons and in Spring-
field, Massachusetts for two seasons (1-3). The first phase of the study
has been conducted in Jacksonville in the summer (June-August) of 1986.
The main purpose of this study is to estimate the exposures of a
statistically representative tit ban population sample to selected pesti-
cides found in the air, drinking water and food, as veil aa through dermal
contract. Through this study the cumulative frequency distribution of
exposures to selected pesticides wil3 be obtained for each of the above
mentioned pathways. The data from this study will be used to evaluate
seasonal changes, temporal changes and interpersonal usage patterns within
any given home which are considered to be the three basic components of
the variability of exposure.
The design of an effective QA program that is "built in" to the study
design is a key element to the success of a complex multidisciplinary
study such as NOFES. The major quality-related elements of this project
include the followingi
* Population Survey Design
* Field Sampling and Analysis
* Data Processing
* Data Evaluation and Analysis
* Estimation of Exposure Through Multiple Pathways.
METHODOLOGY
The methods discussed below pertain only to field sample collection
and laboratory extraction and analysis. The statistical methods used for
sample selection and household screening are given elsewhere (1).
Sampling Design
The following samples are collected in the participating households
in accordance with the sampling matrices provided in the Detailed Work
Plan for NOPES (4).
1. "Primary" households! One each indoor, outdoor, and personal
exposure 24-hour air sample will be collected at each house-
hold .
2. "Duplicate" householdsi Approximately 11 percent of the "primary
households will be designated as duplicates at vhlch collocated
samples will be collected at both indoor and outdoor locations•
Collocated personal samples will not be collected due to practi-
cal considerations.
735
-------
3. "Triplicate" householdst About 40 percent of the "duplicate"
households will be designated as triplicates, and a third con-
current sample will be collected at both the indoor and outdoor
locations, for independent extraction and analysis by the
Reference Laboratory.
4. "Replicate" households; About 14 percent of the "primary"
households will be designated as replicates. A second set of
(indoor, outdoor and personal) samples vill be collected at
least 5 days after the completion of the first monitoring
period. Also the replicate sample vill be collected on a
different day of the week than the primary sample.
5. Dermal exposure households! Personal air sampling will be con-
ducted during periods that include the dermal exposure measure-
ments performed in about 8 percent of the primary households.
6. Drinking water; Household faucet water will be collected at
approximately 8 percent of the primary households.
Sampling and Analysis Procedures
A general description of the sampling and analytical procedures used
in the NOPBS project are given in the NOPES Work Plan and OA Plan (4,3).
Data Quality Objectives (DQO) that have been established for NOPES are
summarized in Table I. The monitoring effort requires collection and
analysis of air (on Poly Urethane Foam Plugs), drinking water, and dermal
exposure (on precleaned gloves) samples. These procedures were tested
under field conditions during a pesticide exposure pilot study (6).
The PUF plugs, water samples and gloves collected are analyzed by
first extracting with distilled-in-glass solvents. The extracts are con-
centrated and split into two aliquots, one of which is analyzed by GC/ECD
for one set of compounds, while the other is analyzed by GC/MS with mul-
tiple ion detection (HID) for another set of compounds. Compounds which
are determined by GC/ECD are confirmed using a second column with differ-
ent elution characteristics.
All samples collected are stored in the field under dry-ice and are
shipped to the central laboratory within 48 hours. The extraction of all
samples and the concentration of their corresponding extracts for GC/ECD
and GC/HS/MID analyses are completed no later than 7 days from the sample
collection end date, and the analysis of the extracts is completed no
later than 30 days from the date of extraction. Table II provides a list
of the target pesticides for NOPBS together with their Quantitation Limit
Goals.
ELEMENTS OF QA PROGRAM
Design of an effective QA program and its implementation has been
recognized by EPA since the study's inception as a key element to the
success of the study.
The QA program for NOPES consists of the following functionsi QA
Management, Internal QA/QC Operations, and External QA Evaluations.
The program has been organized to meet EPA's QA requirements and to
ensure the generation of data of known and documented quality.
736
-------
QA Management. This function is responsible for coordinating the OA
effort through designated QA coordinators of the various organizations
participating in the study. A major effort vas devoted to developing a
detailed Work Plan and QA Project Plan (QAPP) which serve as working docu-
ments and are being revised as appropriate (4,5). This category also inc-
ludes such activities as qualitative systems audits of participating orga-
nizations providing summaries of problem areas uncovered, and reviewing
data generated in the project. These activities are targeted to provide an
assessment of data quality and adherence to the objectives established.
Internal QA/QC. This function pertains primarily to routine activi-
ties undertaken by each of the study participants to "control" the data
collection and measurement process. Diverse activities such as double
entry of all manually entered data, specific procedures for every step in
the sampling and analysis effort, and thorough validation, review and
flagging of reported data are being used throughout the study. Table III
provides a summary of the various QA/QC samples that have been built into
the measurement system. The results of these QA/QC samples are used for
the routine assessment of measurement system performance and enable imme-
diate corrective action.
External OA. The activities in this function involve independent
quantitative assessment of performance. Performance audits include field
audits to evaluate samplers' performance and the data collection process,
and laboratory performance evaluation, via blind audit spikes provided by
EPA and the Reference Laboratory.
RESULTS
Preliminary analytical data in both diskette and hard copy, together
with copies of field observations and sample collection documentation, are
submitted to the data manager In batches for review and validation.
The data submitted are evaluated as follow*!
* Sample tracking report is generated to identify sampling problems
(duration, flow rate) and adherence to extraction and analysis
holding times.
* Initial calibration linearity and continuing calibration drift
ace evaluated for all runs.
* Matrix spike recoveries and surrogate recoveries are evaluated,
baaed on summary reports generated from the data base.
* Quantitation reports are compared to actual data {reported on (hp
hard copies for a minimum of two complete samples per extraction
batch.
* Air concentrations obtained for indoor, outdoor and personal
exposure are summarized by household and compared for internal
consistency.
* A summary report is generated showing all duplicate and replicate
measurements together with the percent differncts for all
instances and a range of dif£«renc«s observed for each of the
target compounds.
737
-------
Table IV presents a summary of percent matrix spike and surrogate
recoveries for the NOPGS Jacksonville Summer 86 sampling phase. Octa-
chloronaphthalene is used as the surrogate compound and added to each of
the samples prior to extraction to enable tracking the integrity of the
entire analytical system. The results fnr 255 PUF samples shov a mean
recovery of 97% with a median of 94% and a coefficient of variation (Std
Dev/mean) of 0.15.
For each extraction batch one pre-cleaned PUF is designated as the
matrix spike* and matrix spike compounds are added to it prior to extrac-
tion. Three of the compounds added are GC/MS target compounds and five are
targeted for GC/ECD. The mean recoveries obtained vary from a lov of 52%
for propoxur (GC/MS) to a high of 108% for gamma-BHC (GC/ECD). Chlorpyri-
fos is used as a mauix spike compound for both GC/MS and GC/ECD with mean
recoveries of 66% and 73%, respectively. It should be noted that the coef-
ficients of variation for all matrix spikes are in the range of 0.15-0.35.
These results provide an indication of the complexity of the measurement
system, and the variability occurring at the extremely lov concentrations
encountered.
Table V shows a compilation of the percent difference ranges for
duplicate collocated measurements, both indoor and outdoor of sampled
households. The results indicate that the overall precision of the
measurement system varies among households, with the largest contributor
probably being the air concentration levels encountered. This is also
evidenced by the sparsity of observed collocated concentrations, above
the quantitation limit, for the outdoor samplers. This limited data set
provides a semiquantitative measure of precision range and allows the data
user to assign some confidence limits to the total exposure assessment.
SUMMARY
Incorporating effective quality assurance measures as an integral
part of the NOPES program ensures the usability of the data generated and
the ability to estimate overall errors propagated in the system. It ap-
pears that as environmental monitoring studies are becoming more complex
in both statistical design and measurement methods used, veil planned QA
activities and allocation of proper resources are essential to generation
of data of known and documented quality.
The phased (seasonal) sampling design for this study enables tracking
system performance closely throughout a sampling phase, followed by intro-
ducing improvements in procedures for the subsequent phase, whenever prob-
lem areas are uncovered. During the upcoming seasons QA would be improved
by adding more "real time" internal QA/QC activities, in addition to inc-
luding a matrix spike duplicate to enable assessment of analytical preci-
sion. Also, more audit samples will be available from EPA and the Referen-
ce Laboratory to better define analytical accuracy.
REFERENCES
1. F. V. Inunerman, "Preliminary Survey Design Recommendations for the
Pesticide and PCB Exposure Survey", RTI/3080/30-025, EPA Contract
68-01-6826, June 5, 1985.
2. A. E. Bond, D. T. Mage, T. Fitz-Simons, F. V. Immerman and D. V.
Jackson, "Supporting Statement for the Nonoccupational Pesticide
Exposure Study", RTI/3080/30-01F, EPA Contract 68-01-6826, January
1986.
738
-------
3. V. Ott, "Summary of Rationale, Objectives and Study Design of NOPES",
submitted to OHB July 28, 1986; revised by A. Bond, December 3, 1986.
4. M, Lev-On, J. E. Howes, Jr., P. V. Immerman and D. E. Camann,
"Detailed Work Flan for the Non-Occupational Pesticide Exposure
Study NOPES), EMSI-1140.49WD, Revision 1, April 9, 1987.
5. M. Lev-On, J. E. Howes, Jr., J. C. Delviche, D. E. Camann, J, P. Hsu,
H. J. Harding and F.V. Immerman, "Quality Assurance Project Plan for
U.S. EPA Non-Occupational Pesticide Exposure Study (NOPES)", EMS1-
1140.26QA, Revision 2, April 10, 1987.
6. D. E. Johnson, J. P. Hsu, H. J. Schattenaberg, H. G. Wheeler, J. H.
Brewer, H. J. Harding and D. E. Camann, "Sampling and Analytical
Support for the Non-Occupational Pesticide Exposure Study (NOPES)
Pilot Study Conducted in Jacksonville, Florida", Pinal Report on
Work Assignment 28, EPA Contract 68-02-3745, October 1985.
739
-------
TABLE I. DATA QUALITY OBJECTIVES (OQO)
Descriptor
Precision
Accuracy
Represen tat iveness
DQO
25% Analytical
35X Overall
20X Analytical
40-50% Overall
Laboratory Recovery
10% Sampler Flow
Not Quantitative
Comparability 20%
Uniformity of
Reporting Units
Completeness 90%
95*
95*
Method
Matrix Spike Duplicates
Collocation of Samplers
PE Samples for Laboratory
Matrix Spike Recovery
Surrogate Recovery
Mass Flov Controller Audit
Validation of Sampling
Design
Statistically Representa-
tive Population Sample
Probe Placementi Indoor,
Outdoor, and Personal
Split Extracts for Inter-
Laboratory Comparison
3
ng/m -air concentration
ng/L ~ drinking water
ng/glove - dermal exposure
Collection of Targeted
Samples
Valid Rxtraction of
Collected Samples
Usable Data of Extracted
Samples
740
-------
TABLE II. QUANTITATION LIMIT GOALS FOR TARGET COMPOUNDS
Quantitation Limit Goals
Compound
ng on column8
ng/m3'
Chlorpyrifos
0.06
11.0
Pentachlorophenol
4.0
730.0
Chlordane
0.8
146.0
Orthophenylphenol
0.2
18.0
Propoxur*
0.2
18.0
Resmethrin
1.0
45.0
Dicofol
1.0
180.0
Captan
0.3
55.0
Carbaryl
0.5
45.0
T-BHC (Lindane)
0.05
9.1
Dichlorvos (DDVP)
2.0
360.0
2,4-D (methyl ester)
1.0
182.0
Malathion*
0.5
45.0
cis-Permethrin
" 0.4
73.0
trans-Permethrin
0.4
73.0
Heptachlor
0.07
13.0
Aldrin
0.05
9.1
Dieldrin
0.08
15.0
Ronnel
0.07
13.0
Diazinon*
0.6
55.0
Methoxychlor
0.1
18.0
Atraeine*
0.5
45.0
»-BHC
0.04
7.3
Oxychlordane
0.06
11.0
Heptachlor epoxide
0.04
7.3
Bendiocarb*
0.5
91.0
Polpet
0.2
36.0
Chlorothalonil
0.04
7.3
Dacthal
0.05
9.1
Hexachlorobenzene
DDT
0.03
0.06
5.5
11.0
DDE®
0.06
11.0
DDD
0.06
11.0
*to be analyzed by GC/MS/HID
amost quantitation Units above are based on Jacksonville Sumner study
report
bbased on the assumption that extracts vill be concentrated to 10 mL,
split to tvo 5-mL portions, concentrated to 1.0 mL and 0.5 mL for GC and
GC/HS, respectively, and that 2 ul vill be injected to either instrument.
The total volume of air sampled is assumed to be 5.5 cubic meters.
cnot analysed for during Summer 86 phase
741
-------
TABLE III. SUMMARY OF OA AND QC SAMPLES PER SAMPLING PERIODS
Sample Type
Frequency or Number
Function
Unexposed PUP
Field blanks4
PUF
Water
Glove
2%
10
2
1
Contamination check of PUF
batch
Dynamic blank - account for
field contamination
Laboratory blanks**
1 per batch
(20-33 samples)
Method blank
Spiked samples
1 per batch
(20-33 samples)
Analytical accuracy
measure
Duplicate spiked
samples
Same as spiked
Analytical precision
measure
Collocated samples
11% of households
Overall precision measure
GC/ECD evaluation
standards
At the beginning of
each 72-hr analysis
period
Linearity check
GC/ECD individual
standards
At the beginning,
middle and end of
each 72-hr analysis
period
Quantitation
GC/MS initial
calibration
Initially and as
required
Quantitation/calibration
GC/MS continuing
calibration
Every 12-hr
analysis period
Monitor response factors
Surrogate
Each sample matrix
Recovery measure
Blind spike sample
Up to 15
External QA evaluation
ataken to and returned from field.
^solvent blanks per sampling period.
742
-------
TABLE IV. PERCENT MATRIX SPIKE AND SURROGATE RECOVERIES*
(NOPES Jacksonville Summer 86 Sampling)
Compound
N
Mean
Median
Std. Dev.
Ran^re
A. GC/MS
Propoxur
8
52
54
20
16-80
Diazinon
8
70
70
19
31-92
Chlorpyrifos
8
66
68
18
31-87
B. GC/ECD
Chlorpyrifos
8
87
84
15
69-113
Dieldrin
8
89
86
24
46-124
Heptachlor
8
107
114
24
69-133
gamma-BHC (Lindane)
7
108
102
33
73-163
Chlorothalonil
7
62
60
19
39-89
Hexachlorobenzene
3
96
90
13
86-111
OCNb
255
97
94
15
36-169
aAll recoveries reported are for PDF plugs only
^OCN vas used as a surrogate and added to every sample prior to
extraction
743
-------
TABLE V. PERCENT DIFFERENCE RANGES FOR COLLOCATED NOPES SAMPLERS®
(NOPES Jacksonville Summer 86 Sampling)
Compound Indoor Outdoor
(J X Diff Range R X Diff Range
chlorpyrifos
dlazinon
propoxur
dieldrin
ortho-phenylphenol
chlordane
gamma-BHC (Lindane)
bendiocarb
heptachlor
aOnly concentration values above the QAPP designated quantitation
limit have been used in compilation.
6
7-68
3
12-73
3
13-21
2
8-24
2
33-57
1
5
1
10
1
5
1
3
1 14
2 8-18
744
-------
development of gaseous hydrogen chloride standards
Robert B, Denyszyn, Ph.D.
David Pleesor
Karen. Magulre
Scott Specialty Gases
Route 61L
Plumsteadvllle, Pennsylvania
Hydrogen Chloride (HC1) is a very corrosive and reactive gas. Because
of the reactivity of RC1, gaseous mixtures are difficult to prepare and
stability is questionable.
To develop a Btable product at the 100 to SO ppo level, various tech-
nological developments had Co occur. A cylinder which could maintain the
HC1 concentration constant, and an analytical Method capable of quantita-
tively measuring swill changes in concentration had to be developed. Pre-1
vious work at the LBOO to 500 ppm levels had indicated that the use of
either molybdenum or manganese steel alloy cylinders could be used for HC1
applications. A significant amount of development was necessary to decrease
the surface activity of the cylinder. Is order to obtain representative
samples from the cylinders, several delivery systems were tested and a sys-
tem constructed of stainless steel 316 was found to perform satisfactorily.
The basic analytical methodology selected to analyse H<31 was ion
chromatography. To quantitatively analyse the cylinder, triplicate one
liter samples were simultaneously collected in water containing a diluted
solution of sodium hydroxide. The collected solvent* were analysed against
multipoint calibration curve* to ensure the accuracy of th« measurement.
Standard* were introduced periodically in the analytical scheme, and other
measurements ware performed to attsura the accuracy and traceability of the
gaseous standards.
745
-------
I. Introduction
Hydrogen chloride (HC1) is one of many pollutants emitted by the
incineration of both hazardous waste and trash. Several states have
introduced legislation requiring the continuous monitoring of HC1
emissions from trash to steam plants. These requirements have
prompted the Environmental Protection Agency to evaluate the
suitability of various continuous monitors and the calibration of
auch monitors.
To date, the need for quantitative measurements of HC1 in any
process has been lacking. Due to this lack, of interest for such
measurements, little emphasis has been placed on the development of
reliable gaseous standards.
In the case of trash to steam Incineration, the HC1 concentra-
tion of interest is 50 ppm and up. The preparation and quantifica-
tion of HC1 at these concentration levels is a real challenge.
In the summer of 1986, Scott Specialty Gases' Research and
Development department Initiated the development of HC1 in inert
gaseous matrices. Coals of the mixtures were to:
1. Develop a stable 250, 100, and 50 ppm gaseous standard
2. Develop a suitable delivery system.
3. Develop a cost effective analytical method to quantitatively
analyze HC1.
4. Develop gaseous mixtures which have traceability to the na-
tional measurement system.
One of the major goals of the project was to develop stable HC1
gaseous mixtures. It was determined from the onset of this study
that the product would have a stability of +5X/year, meaning that
during the course of a year analytical measurement would not show a
change in concentration of more than +5X.
II. Experimental Work
Development of a Suitable Container
HC1 is a very corrosive gas which, in the presence of bduiII
levels of moisture, will corrode many metal alloys. Specialty gas
manufacturers have a variety of cylinder types available to them
Including: aluminum, steel (manganese and molybdenum/chromium), and
stainless steel.
Due to the reactivity of HC1 with aluminum, aluminum cylinders
were immediately eliminated from the potential list, leaving steel
and stainless steel. Steel cylinders have typically been used to
store pure HC1, and stainless steel would provide the most corrosion
resistant container. Since the requirement is for continuous emis-
sion monitors which are typically routinely calibrated, the amount
of gas required dictates the need for a container which has a total
gaseous volume of 100 - 200 SCT. Even though stainless steel con-
tainers are available in this volumetric range, it was felt that
steel containers could be made to perform in this application. A
major factor was that stainless steel containers are approximately
746
-------
20 times more expensive than steel containers. Steel containers
which are supplied to the specialty gas Industry do have their
limitations with Che major limitation being that the inside is
coated with a thick metal oxide which is very reactive.
In order to have any potential success with such a container,
the thick oxide surface must be removed and the surface modified to
eliminate gas-to-surface interaction. Scott's Aculifetra V process
provided such'modifications of the surface. 'Cylinders treated by
this process were evaluated with gas mixtures containing 1800 and
500 ppm of HC1 respectively with encouraging results (see Table I).
Analytical results seemed to indicate a product which remains
stable for at least six months. The data in Table I does not
reflect the difficulties incurred while obtaining these measure-
ments. Table II is a summary of some of the values obtained when
performing these analyses. The data in Table II shows two problems:
1. The first sample collected and analyzed is almost always
low.
2. The analyses performed on 8/18 and 8/19 are significantly
lower than what would be expected.
Even though the results were not perfect, it was felt that the
Aculife V treated steel cylinders would provide a good starting
point for work at 250, 100, and 50 ppm.
Development of Suitable Delivery System
The system used to generate the data in Table II is shown on
Figure 1. The system was constructed of stainless steel components.
Due to the relatively low results obtained on 8/18/B6 and 8/19/86,
the delivery system was disassembled. Heavy corrosion was observed
at the CGA gland assembly and regulator. It was also noted that the
CGA fitting on the cylinder valve had also developed significant
corrosion. The corrosion was somewhat surprising in view of the
rather low HC1 concentration. The corrosion is a sign of HC1 lost
as well as a sink for further absorbtion. To eliminate the corro-
sion, It was decided that all wetted parts prior to the regulator as
well as the metering valve would be stainless steel 316 or superior
alloys. Other modifications were made to the system resulting in
the system shown in Figure 2 which was used for titration work.
The new system had incorporated a nitrogen purge assembly to
eliminate moisture in the connection and assembly prior to Che
withdrawal of the HC1 mixture. The regulator was also eliminated.
The regulator had provided a lot of surface area for potential ad-
sorption. The system in Figure 2 was satisfactory in as far «s HCl
mixtures could be delivered to the Greenburgh impingers, but in or-
der to analyze 100 and 50 ppm mixtures, 20 - 30 liters of gas were
required to titrate a 0.O1 - O.Q3m8aOH solution at a flow rate of 1
liter per minute. Triplicate samples from one cylinder took 1 - 1.5
hours. The use of ion chromatography as a quantification of Cl~
provided a mechanism to reduce both the volume of solution and gas.
Therefore, a new system was designed and constructed.
Schematics of this system are shown in Figure 3. The system,
theoretically, was to provide triplicate samples simultaneously at
100 ml per minute with a total of 1 liter of gas mixture per sample.
747
-------
Unfortunately, at concentrations of 250 ppm, it was found that even
with a total flow of 300 ml per minute, the system worked Less than
ideally. Titrimetric analyses of a 250 ppm HCL in nitrogen cylinder
indicated the presence of approximately 250 ppm while the Ion
chromatographic system showed little or no HCL. The problem was
narrowed to the following:
1. The system must be heated and purged with nitrogen prior to
and after opening to the atmosphere.
2. A volumetric flow rate through the system of at least 1-2
liters per minute must exist when withdrawing product from
the cylinder.
3. The use of all metal seals and electropolished stainless
steel 316 components provided faster delivery of the gas
mixture to the collection system.
In view of the incurred problems, a modified sampling train
for ion chromatography was designed (see Figure 4). It incorporated
the following features:
1. Delivery system to cylinder valve interface was best if me-
tal seals were used.
2. Delivery from the cylinder was maintained at > 1 liter per
minute.
3. The total volume of gas collected was measured using a mass
flow totalizer.
4. The system was sealed before and after connection to new
cylinders to minimize the effect of moisture Intrusion.
Development of a Cost Effective Analytical Method to
Quantitate HC1 in Nitrogen Mixtures at the 50 PPM Level
A standard titrimetric method was used to initially analyze
mixtures of HC1 at 500 ppm and higher concentrations. The method
consisted of bubbling the HC1 gas mixtures through 200 ml of stand-
ardized sodium hydroxide solution ranging from 0.01 - 0.03 M. The
method was found to be extremely quantitative with a collection ef-
ficiency of >99.92 at a flow rate of 1 liter per minute. The end
point of the titration was detected using a phenolpthalein in-
dicator. A stock solution of NaOH was stored in polypropylene
bottles under Nj atmosphere to minimize the absorption of atmos-
pheric CO2. Major drawbacks of this procedure were:
1. The large volume of standardized solution required in order
to analyze a few cylinders.
2. In order to titrate HC1 mixtures at the 50 ppm level, rela-
tively low concentrations of NaOH solution must be used
(millimolar). These relatively low concentrations make the
transfer of such solutions difficult without having the in-
fluence of atmospheric CO2.
3. Large volumes of gas mixture must be used <30 - 40 liters)
making the analysis time very long.
748
-------
4. The end point detection is poor, impairing the precision of
the method.
Other solutions which can be used to quantitatively collect HCL
solutions ares 0.001 M NiOH, 0.001 M NaHCOj, and 0.001 M Na2C0j,
For the best results from ion chromatography, a solution of
NaHC03/Na2co3 is m<>8t since the elutent used for the
separation has the same constituents. The chromatographic system
used for the analyses consisted of! a Dionex 2000 ion
chromatograph, a Dionex 4270 integrator, and a Dionex autosampler.
Chromatographic conditions were as follows:
Columni HPIC AS4A
Eluent: Na2C0j/ NaHCO^
Flow rate: 2 ml/min
Pressure: 960 psia
Detector Setting: 30 s for less than 100 ppm
100 s for >250 ppm
Precision achieved with 50 ppm equivalent Cl~ concentration is
shown in Figure 5. The coefficient of variance for triplicate in-
jections showed a value of +0.11%. Response was linear for the
equivalent of 10-50 ppm. Analysis time ranged from 3-8 minutes
depending on the analysis performed (manual or automatic).
III. Results To Date
Measurements
A total of 15 cylinders had been prepared for this work; five
at each concentration level (250, 100, and 50). Titrimetric
analyses were performed on all the 250 ppB HC1 mixtures and the
results are shown in Table III. Data indicates relatively good
agreement from cylinder to cylinder. Further analytical work was
performed on A-7778 over a 1.5 month period and our results, as
shown in Table IV, show a gas mixture which is not changing very
rapidly over a short period of time (<10% per month). There seems
to be a bias of 6% between the titrimetric values and the ion
chromatographic values. If we take the mean of both methods, the
concentration in the cylinder can be quantified to 253 ppm +3X.
This value is in good agreement with the gravimetric value which Is
250 ppm.
Due to significant sample lost prior to the impingers, fewer
analyses have been performed on lower concentration cylinders.
Table V summarizes the results obtained to-date on cylinder A-5087
which theoretically contain. 100 ppm. The bias between the two
methods is larger than at 250 ppm (approximately 14X). The mean
concentration of both methods provide us with a cylinder concentra-
tion of 95 ppm. In view of the fact that the gas mixture in the
cylinder was prepared one month earlier, there is a good probability
that this mixture will be found stable.
Results associated with a 50 ppm HC1 mixture (cylinder A-10439)
are shown on Table VI. The bias between the two analytical methods
increased to 38*. Mean concentration after one month was 55 ppm;
again in relatively good agreement with the theoretical gravimetric
value of 50 ppm +10*. It is doubtful that the concentration would
749
-------
be higher than 50 ppm since the mass of gas mixture and pure gas is
large (30 gm) with a sensitivity of +0.2 gnu
Traeeabllity of the Analyses
Traceability to the national measurement system is crucial in
analytical work. In this work, this was accomplished via two
methods.
1. Titrating the HC1 gas in a NaOH standard solution. The NaOH
was standardized with potassium hydrogen pthalate, Standard
Reference Material 84J (Batch 580508). Flow measurements
were made with a wet test meter with an error of 0.5%.
2. In the ion chromatographic method, KC1 standard liquid mix-
tures were prepared using Standard Reference Material 999,
(Lot 589910) in 17 megohm-cm water. An internal standard (2
ppm of KBr certified ACS grade) was also used to keep track
of any water lost during collection of the gaseous sample.
Multipoint calibration curves were generated with both KC1
and KBr to quantify both Cl~ and Br~ concentrations. The
flow measurement devices and rotometera were calibrated
against the same wet test meter used in the titrimetric
method.
Propagation of error analysis has not been performed at this
time but we feel confident that the overall method should he able to
achieve +12 measurement uncertainty.
IV. Conclusion
At this time, the following can be said:
1. We have evidence that HC1 mixtures as low as 50 ppm can. be
prepared accurately.
2. A purgable, electropolished, heated, stainless steel sam-
pling train is necessary to quantitatively analyze HC1 mix-
tures in the low ppm range.
3. A flow rate of at least 1-2 liters must be used through the
delivery system In order to retrieve quantitative samples
from the cylinder.
4. Mixtures at the 250, 100, and 50 ppm level are probably
stable to better than +10% per year.
V. Acknowledgment
This program Is partially supported under E.P.A. Contract No.
4-4321)-3180.
750
-------
TABLE I - Initial Measurements of 1800 and 500 PPM HCL In N2 Mixtures
Theoretical Analytical Results (PPMV)
Cylinder f Cone. (ppwvJ 6/5/86 8/7/86 8/13/86 8/18/86 8/21/86 9/30/86* 4/2/87*
A-17241 1800 1812 1734 1736 1 563 1 687 1668 1584
8/6/86 8/8/86 8/14/86 8/19/86 8/20/86 9/30/86* 4/2/87*
A-16790 500 ppm 506 492 507 446 475 473 430
•date provided courtesy of Entropy, Research Triangle Park, North Carolina
A-17241
A- 16790
TABLE II 1800 and 500 PPM Analytical Precision
Theoretical
Cylinder # Concentration
1800 ppm
500 ppm
8/5/86
8/7/86
8/13/86
8/18/86
8/21/86
1729
1753
1700
1566
1602
1824
1720
1770
1516
1627
1850
1730
1607
1666*
1882
1696*
1773
1695*
1812
1734
1736
1563
1687
8/6/86
8/8/86
8/14/86
8/19/86
8/20/86
501
469
462
367
471.3
505
505
513
447*
476.5
510
494
547
446
476.8
507
498
506
492
50
446
475
Data used to calculate mean
751
-------
TABLE III Titrimetric Analyses of 250 PPM HC1 Mixtures
Cylinder Wo.
A-7778
A-17442
A-17477
A-17508
A-14512
Date Analyzed
3/13/87
3/13/87
3/19/87
3/19/87
3/19/87
Concentration (ppm)
264
264
262
263
256
All cylinders were prepared on 3/12/87
TABLE IV Analyses of 250 PPM HC1 In Nitrogen (A-7778)
Titrimetric 3/13/87 3/26/87
Concentration (ppm) 353 257
4/27/87
260
I.C. 3/20/87 3/24/87 3/25/87 4/15/87 4/21/87
Concentration (ppm) 237 247 248 250
TABLE V Analyses of 100 PPM HC1 Cylinder (A-5087)
Titrimetric 4/10/87
Concentration (ppm) 100
I.C.
Concentration (ppm)
4/27/87
"~T52
4/23/87 4/28/87
"1? 9(5.1
TABLE VI Analyses of 50 PPM HC1 in Nitrogen (A-10439)
Titrimetric 3/31/87 4/10/87 4/27/87
Concentration (ppm) 53T 52 5?
I.C. 4/28/87
Concentration (ppm) 43 49*
*modified Titrimetric sampling train followed by I.C. analysis
752
-------
Stainless Steal Regulator Watering Valva
—&
i
I
o
Qraanburg Impingars
Wat Tast Mater
(1 Utar/Ravolution)
HCLCyllndar
FIGURE 1 - INITIAL SAMPLING TRAIN(Titration)
N» Purga
Wat Tatt Mater
FIGURE 2 - MODIFIED SAMPLING TRAIN (Titration)
753
-------
Rotameters
Metering Valve*
Midget Impingtn
Na Purge
Vacuum
FIGURE 3 - INITIAL SAMPLING TRAIN(I.C.)
V.nt
<3—/
Dryar
Metering VsIvm
Mm Flow M«t«r
Na Purg*
Vacuum Pump
FIGURE 4 - MODIFIED SAMPLING TRAIN(I.C.)
754
-------
QUALITY ASSURANCE IN THE NONMETHANE
ORGANIC COMPOUND MONITORING PROGRAM
Robert F. Jongleux, Robert A. McAllister,
Dave-Paul Dayton, Denny E. Wagoner
Radian Corporation
Research Triangle Park, NC 27709
Frank F. McElroy, Vinson L. Thompson
Environmental Monitoring Systems Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Harold G. Richter
Office of Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
During the summers of 1984, 1985, and 1986, 4773 measurements of
nonmethane organic compound (NMOC) concentrations in ambient air were made
with the Cryogenic Preconcentration Direct Flame Ionization Detection (PDFID)
method. The development of a quality assurance program for determining the
NMOC concentrations with known quality is discussed.
Analytical precision was measured by repeated measurements of site
samples. The mean absolute percent difference was 9.8, with a standard
deviation of 12.2. The average percent difference was -0.354, with a standard
deviation of 15.7. Sampling precision was determined from analyses of
duplicate site samples. Absolute percent differences of the duplicate samples
for 3 years averaged 11.7, with a standard deviation of 16.4, Accuracy, or
bias, was determined from measurements on dry propane standards referenced to
National Bureau of Standards (NBS) Standard Reference Material (SRM) 1667b.
In three years, average biases for Radian analytical channels ranged from
-2.115% to -0.516% relative to EPA-QAD, with a pooled standard deviation of
3.951%.
Data base errors were monitored by a 10% random check of approximately 54
entries for each sample. The error rate in 1986 was 0.015%.
755
-------
Introduction
A number of dispersion models have been developed to describe
quantitatively concentrations of precursor organic compounds and downwind
concentrations of ozone 1n ambient air as a function of time. The Empirical
Kinetic Modeling Approach (EKMA) is a model which requires an average total
nonmethane organic compound (NMOC) concentration from 6 A.M. to 9 A.M. In
parts per mi 111on carbon (ppmC), by volume. An important EKMA application 1s
to determine the degree of control of NMOC required 1n a particular area to-
achieve compliance with applicable ambient air quality standards for ozone. '
Radian Corporation has assisted the U.S. Environmental Protection Agency,
participating states, and the District of Columbia in obtaining the NMOC
concentration data needed to prepare state implementation plans to meet the
National Ambient Air Quality Standards (NAAQS) for ozone. A total of 4773
ambient air NMOC concentrations were measured at 44 sites, representing 22
states and the District of Columbia during the summers of 1984, 1985, and
1986. Because the NMOC data were used for regulatory purposes 1t was
essential that the measurements result 1n reliable data of known quality.
The Cryogenic Preconcentration and Direct Flame Ionization Detection
(PDFIDJ method was developed by the U.S. Environmental Protection Agency to
replace the NMOC 4gasg chromatographic (GC) speciation method. The PDFID
method was shown to have precision and accuracy comparable to the
NMOC GC speciation method, and was less costly to perform.
Quality assurance (QA) and quality control (QC) procedures necessary to
obtain accurate NMOC data of known quality are the subject of this paper.
Data quality parameters considered to be among the more important were
accuracy and precision. Details of the Quality Assurance Project Plan
(QAPP) unique to the NMOC Monitoring Program are described 1n the next
section. There follows QC results of the program to date, and a description of
the ongoing review of the QC procedures and acceptance criteria to Improve the
definition of data quality.
Quality Assurance Project Plan
Quality Assurance Project Plans (QAPP) were prepared according to EPA
Guidelines for each year of the NMOC Monitoring Program. Detailed operating
procedures were developed for the field sampling sites. Radian personnel were
sent to each site to install and test the sampling equipment, and to train
site personnel 1n the sampling procedure and schedule. The sampling schedule
was projected. Integrated 3-hr (6 A.M. to 9 A.M.) ambient air samples were
collected dally, Monday through Friday based on a 17-week program (except for
July 4 and Labor Day), 1n electropollshed stainless steel canisters. All
samples were analyzed by Radian Corporation at the Perimeter Park, NC,
laboratory. Canisters were shipped air freight the day of collection and were
analyzed within two days of collection, except for Friday's samples, which
were analyzed the following Monday.
756
-------
A duplicate sample was scheduled for each site once every two weeks. For
duplicate samples, two canisters were filled simultaneously from the same
ambient air Intake line. The duplicate-sample schedule was arranged so that
the number of duplicate samples taken at each site was equal throughout the
sampling period (June 1 through September 30)* and so that approximately the
same number of duplicate samples was taken each day.
Canisters sent to each site were Initially cleaned in the Perimeter Park
laboratory and sent under vacuum. A Field Sampling Form was enclosed 1n the
canister shipment case for each canister. The site operator completed the
information on the form and returned 1t with the sample. Upon receipt, the
canister was tagged with an Identification number and an analysis sheet was
prepared which accompanied the sample through analysis. The original
field data sheets and analysis sheets became part of the permanent project
files.
All calibrations and audit samples used propane standards referenced
to National Bureau of Standards (NBS) propane, Standard Reference Material
(SRM) No. 1667b., by the U.S. EPA-Quallty Assurance Division (QAD) of the
Environmental Monitoring Systems Laboratory (EMSL) at Research Triangle
Park, NC. All four Radian PDFID channels were calibrated with propane from
0.000 to 9.627 parts per million carbon (ppmC) by volume. "Zero* standard was
obtained using air, filtered through fritted stainless steel filters, dried in
permeation driers, and cleansed of organic content by passing through a
catalytic oxidizer.
Each Radian PDFID channel was calibrated twice dally, once before any
sample analysis for the day, and again after completion of the sample
analyses. From these data percent calibration drift was calculated.
Four days per week 1n-house propane quality control samples were prepared
by the project QC Monitor, and analyzed by each Radian PDFID instrument. On
the fifth day each week, duplicate local ambient air samples were collected
by EPA-QAD and analyzed on each Radian PDFID channel, the EPA-QAD PDFID
channel, and the EPA-Atmospher1c Sciences Research Laboratory
-------
Data were transferred dally after review by the project QA Monitor, from
data sheets, or log books, to a 20-megabyte magnetic data disk, that contains
the NMOC data base. A backup data disk was updated dally. Biweekly meetings
with EPA were held to review the data and to Investigate any unusual data or
data trends. Upon completion of the sampling and analysis for the summer, ten
percent of the data was selected randomly for an audit check. For each NMOC
datum selected for audit, approximately 54 pieces of data from the original
data sheets were checked for correctness and for data transcription errors. A
total of 2351 NMOC measurements are 1n the 1986 data base. Of these, 252
cases were checked for error. Two errors out of 13,608 entries were found and
corrected, for an error rate of 0.015%.
Quality Control Results
Calibration factor drift results are given in Table I in terms of percent
drift and absolute percent drift. The morning calibration factor was used as
the basis for drift because it was the factor used to calculate NMOC
concentrations for the entire day. The average calibration factors for 1985
and 1986 are about equal, but are less than 1n 1984, This result can be
explained by the fact that the analytical instrument was housed In a better
temperature-controlled environment in 1985 and 1986 than in 1984. Absolute
percent differences in 1986 ranged from 0.929 to 1.959% among the Radian
Channels.
Precision of the NMOC measurement was determined from repeated analyses
of site samples. Table II reports precision 1n terms of percent differences
and absolute percent differences for each year of the program and for
three-year pooled values. No trends are evident. The overall average
absolute difference 1s 9.8%, with a standard deviation of 12.2%.
Sampling and analysis precision was determined from duplicate samples.
Table III shows the duplicate sample results in terms of percent difference.
The first run Includes both between-duplicate sample variability and wlthin-
dupHcate variability, or analytical error. The analytical error was obtained
from repeated analysis of site ambient air samples. Note that the standard
deviations In terms of percent difference and absolute percent difference are
about equal for both between-duplicate and w1th1n-dupl1cate variability.
Accuracy 1s reported 1n Table IV in terms of percent bias relative to the
EPA-QAD PDFID instrument. The average bias of the Radian channels ranged from
+2,76% for 1984 to -1.80% for 1986. No significant differences in bias were
noted among the Radian channels. Accuracy was monitored throughout the
program using In-house propane QC samples prepared from propane referenced to
NBS SRM propane. Table V shows the results for 1984, 1985, and 1986 for
Radian Channels A, B, C, and D, and gives pooled values for all four Radian
channels. Significant differences are seen again between the absolute percent
difference means for 1984 and 1985 (or 1986). The in-house samples were
prepared by the project QC Monitor by diluting the certified propane sample
with cleaned, dried air by means of flow controllers. The percent difference
figures include error caused by the dilution process, and analytical error.
After analysis was completed the canisters were cleaned up before being
sent to a site for another sample. The cleanup cycle consisted of first
pulling a vacuum of 0.67 kPa on the canister, followed by pressurizing the
758
-------
canister to 275.8 kPa gauge pressure with cleaned, dried air. This cycle was
repeated three times during canister cleanup. During cleanup, six canisters
were cleaned simultaneously on a manifold. After the third pressurlzatlon,
the canisters on the manifold that had the highest NMOC content before cleanup
was tested for Its NMOC content. If the area counts measured less than
26 (about 0.008 ppmC), the cleanup criterion was met; the canister was
reconnected to the manifold and evacuated a fourth time; all the canister
valves were closed and the canisters were ready for shipment. The average
removal of the NMOC content of the canisters was at least 99.891% in 1986. In
1985, at least 99.898% of the NMOC content was removed. Heating of the
canister above room temperature was avoided to protect the passlvated electro-
pollshed interior surface of the canister.
Ongoing QA Program
Each year, data quality parameters are reviewed and updated. Hew
acceptance criteria are added as needed. The acceptance criteria for the 1987
NMOC program are given 1n Table VI and are based on experience gained in the
1985 and 1986 NMOC Programs. For each measurement listed in the first column,
a critical acceptance region (CAR) 1s given 1n the second column. If the
measurement falls 1n the CAR, no further quality control action 1s required.
If the measurement falls outside the CAR, further action, as listed in the
last column, 1s given. In addition, an ongoing 11st of measurements not
falling within the CAR is maintained to expedite QA review and reporting.
In data validation, the measurement taken 1s the percent error 1n the
data base. For each NMOC concentration measurement, there is an average of 54
Items recorded on original data sheets that need to be checked on the magnetic
disk data base for accuracy of calculation and data transfer. Any errors that
are found are corrected. The data base 1s validated after all NMOC
measurements have been taken for the summer and data transfers to the magnetic
data disks have been completed. If an error rate equal to or greater than
0.03. is found 1n the first 10% of the randomly selected NMOC concentration
measurements, a validation of an additional 10% of the data base 1s required.
If the error rate is still 0.03% or higher when the additional 10% of the data
is validated, an additional 10% will need to be validated.
Tables I through VI are available upon request by contacting
Dr. Robert A. McAllister, Radian Corporation, P.O. Box 13000, Research
Triangle Park, NC 27709. (919) 541-9100.
References
1. U.S. Environmental Protection Agency, "Uses, Limitation, and Technical
Basis of Procedures for Quantifying Relationships Between Photochemical
Oxidents and Precursors," Research Triangle Park, NC EPA-450/2-77-02Ia
(Nov. 1977).
2. U.S. Environmental Protection Agency, "Guidance for Collection of
Ambient Non-methane Organic Compound,(NMOC) Data for Use in 1982 Ozone
SIP Development," Research Triangle Park, NC EPA-450/4-80-011 (June
1980).
759
-------
3. F. McElroy, V. Thompson, H. Rlchter, "A Cryogenic Preconcentratlon Direct
FID (PDFID) Method for Measurement of NMOC 1n Ambient A1r,N EPA/600/4-
85/063, U.S. Environmental Protection Agency, Research Triangle Park, NC,
October 1985.
4. Radian Corporation, "Nonmethane Organic Compound Monitoring
Assistance for Certain States 1n EPA Regions III, IV, V, VI, and VII,
Phase II," Final Project Report, DCN No. 85-203-024-12-01, Prepared for
the U.S. Environmental Protection Agency, Research Triangle Park, NC
(February 1985).
5. R. McAllister, D-P. Dayton, D. Wagoner, "Nonmethane Organic Compounds
Monitoring," Radian Corporation, Final Project Report,
DCN No. 85-203-024-35-01-04, Prepared for U.S. Environmental Protection
Agency, Research Triangle Park, NC (December 16, 1985).
6. R. McAllister, R. Jongleux, D-P. Dayton, P. O'Hara, and D. Wagoner,
"1986 Nonmethane Organic Compounds Monitoring," Radian Corporation,
Final Project Report, DCN No. 87-203-024-93-11, Prepared for
U.S. Environmental Protection Agency, Research Triangle Park, NC (April
1987).
7. U.S. Environmental Protection Agency, "Interim Guidelines and
Specifications for Preparing Quality Assurance Project Plans,"
QAMS-005/80, Office of Monitoring Systems and Quality Assurance, Office
of Research and Development, Washington, DC (December 29, 1980).
760
-------
MONITORING A WIDE RANGE OF
AIRBORNE ORGANIC CONTAMINANTS
William R. Betz,
G.D. Wachob, M.C. Firth,
Supelco, Inc., Supelco Park,
Bellefonte, PA 16823-0048
Airborne organic contaminants at toxic waste sites and indoor air
pollution sites typically range from highly volatile to nonvolatile
compounds. These compounds are also typically present at low levels.
Thus, they must be concentrated for effective qualitative and
quantitative characterization of these atmospheres. Two or more
adsorbent tubes, with dissimilar adsorption and desorption
characteristics, must often be used.
A new multi-bed adsorbent tube containing three Class I,
nonspecific adsorbents effectively adsorbs airborne organic compounds
ranging from volatile (e.g., vinyl chloride) to nonvolatile (e.g., PCBs)
in a size-exclusion mode. Carbotrap C (a graphitized carbon black,
surface area *10 m^/g) adsorbs nonvolatile compounds, Carbotrap (a
graphitized carbon black, «100 mz/g) adsorbs semlvolatile compounds,
and Carbosieve S-III (a carbon molecular sieve, =800 m^/g) adsorbs
volatile compounds. Characteristics of these adsorbents will be
discussed.
A highly efficient thermal desorber is used to ballistically heat
the tube and desorb the trapped compounds. They are transferred, via
gas flow reversal, to a packed or capillary GC column in a discrete
injection slug. This allows the column to refocus the sample and
eliminates the need for a cryotrap. Desorption efficiencies for a wide
range of organic contaminants will be described. Column choice, oven
temperatures, and other key considerations for the chromatographic
analyses will also be discussed.
761
-------
Characterization of both ambient air and indoor air atmospheres has
become the focus of concern for many air sampling analysts• Evaluation
of these atmospheres typically entails qualitative and quantitative
analyses of a wide range of organic contaminants present at "extremely
low" concentrations. Multi-bed adsorbent tubes, used in thermal
desorption modes, are very useful for these analyses. Because the
airborne contaminants exist in many molecular sizes and shapes, and may
possess one or more functional groups, choosing the appropriate
adsorbents becomes difficult. By using Class 1, nonspecific adsorbents
(no ions or active groups), an analyst need not be concerned over which
functional group(s) an adsorbate possesses.
By definition, a Class 1 adsorbent Is without ions or active
groups. Only a Class I adsorbent interacts nonspecifically with all
four typeB of adsorbate molecules (Table I). Therefore, a multi-bed
adsorbent tube containing Class 1 adsorbents adsorbs and subsequently
releases a wide range of organic compounds possessing different
functional groups. Furthermore, an appropriately designed multi-bed
tube will differentiate between molecules of various sizes by
size-exclusion. Vtfhen a molecule enters an adsorbent tube containing
Class I adsorbents of progressively larger surface area (Figure 1), the
tube functions as a small gas chromatographic packed column in which the
migration rate of each adsorbate is inversely proportional to its
molecular size.
Carbotrap™ and Carbotrap C graphitized carbon blacks and
Carbosievem S-III carbon molecular sieve are classified as Class I
adsorbents (1). For these adsorbents, surface area ranges from 10 to
800 m^/gram (Table 11). Because these adsorbents interact non-
specifically with adsorbates, the entire surface of each material is
available for adsorbate/adsorbent interaction. Adsorption is a function
of the surface-surface interactions, or dispersion forces, between the
adsorbate and adsorbent surfaces. Adsorbates having small molecular
size, such as vinyl chloride, migrate quickly through the graphitized
carbon blacks and are adsorbed on the microporous surface of the carbon
molecular sieve. Larger adsorbates, possessing larger surface areas,
interact more strongly with available surfaces. Consequently, larger
adsorbates adsorb to the surfaces of the graphitized carbon blacks.
After they are collected from the air Bample, all adsorbed compounds can
be thermally desorbed by using flow-reversal techniques.
Another key advantage to using a multi-bed adsorbent tube
containing nonspecific adsorbents ia that water adsorption also is a
function of size-exclusion. Larger adsorbate molecules compete more
readily for the available adsorbent surface. Specific retention volumes
(breakthrough volumes) were determined for water and vinyl chloride,
using Carbosleve S-IXX adsorbent and for water and chlorobenzene, using
Carbotrap. In each case, the specific retention volume of the organic
compound is a magnitude of 1000 greater than that of water (Table III),
The surfaces of these adsorbents (and that of Carbotrap C as well) are
therefore considered to be hydrophobic.
762
-------
Specific retention volume data are excellent tools with which to
characterize the adBorptive properties of adsorbents. In this work,
characterization evaluations were performed by following procedures
outlined in EPA Document # 600/7-78-054 (2). These evaluations focused
on adsorbate/adsorbent interactions occurring in the low coverage (or
Henry1s Law) region, using established specific retention volumes,
adsorption coefficients, and equilibrium sorption capacities* Specific
retention volumes (Table IV) were determined at elevated temperatures*
The resulting straight line equation was extrapolated, via linear
regression analysis, to obtain the specific retention volume at ambient
temperature (V| 20°C). This value is expressed in milliliters of
gas per gram or adsorbent, and therefore is useful for determining
adsorbent bed weights.
To establish working ranges for the three Class I adsorbents, the
specific retention volumes of several key compounds, possessing
different sizes and functional groups, were determined. Because the
functional range for Carbotrap overlaps those for Tenax® GC and
Amberlite® XAD-2 adsorbents, comparative evaluations were performed for
these adsorbents (Table V). Note that adsorbent density, as well as
unit surface area, plays an important role In determining the total
available surface area per tube volume (Table V, line 6). Specific
retention volumes, summarized in Table VI, indicate that of the three
adsorbates evaluated, Carbotrap adsorbent possesses the greatest
affinity for most of the adsorbates used. Teaax GC possesses a greater
affinity for n-pentanolc acid because of dipole-dipole interactions
between the acid functional group on the adsorbate molecule and
phenylene oxide groups on the adsorbent's surface.
The working range for Carbotrap C adsorbent was established by
using the same eight adsorbates, plus adsorbates having larger surface
areas. Specific retention volumes for Carbotrap C adsorbent (Table VXI)
are smaller than those for Carbotrap (Table VI), reflecting the smaller
surface available for interactions with adsorbates* In the multi-bed
tube, Carbotrap and Carbotrap C adsorbents will retain these large
molecules. Therefore, only smaller adsorbates were used to characterise
the Carbosleve S-III carbon molecular sieve (Table VIII). The large
specific retention volume for vinyl chloride, 1.26 x 106 ml/gram,
illustrates the adsorbing capacity of this material.
The excellent trapping performance of the adsorbents in the
multi-bed tube was complemented by using thermal desorption to rapidly
transfer the adsorbates to the gas chromatographic column. The thermal
desorptlon unit used is capable of heating the adsorbent tube to the
appropriate temperature in approximately 20 seconds. This rapid
transfer eliminates the need lor cryofocuslng. For example, the rapid
removal of the C2 hydrocarbons at 3306C, versus the commonly used 2Q0"C
desorptlon temperature, provides more effective transfer and improved
resolution of these adsorbates (Figure 2).
The GC column ifl also a key factor in the thermal desorptlon/gas
chromatography analysis. The column must refocus, as well as separate,
the deBorbed analytes. A 6 ft. *1/8" stainless steel column packed
with 60/80 mesh Carbopack1" B packing, at a temperature of 35®C, provided
the best results for the C2 analysis In Figure 2.
763
-------
Both packed and wide-bore capillary columns were chosen to provide
effective refocusing and separation of the adsorbates cited in Methods
TO-1, TO-2, and TO-3 in EPA Document # 600/4-84-041 (3). Two packed
columns were chosen to evaluate the 24 adsorbates cited in these three
methods. A 0.1% SP^-IOOO phase coated on 80/100 Carbopack C was used to
analyze the nonchlorinated and chlorinated hydrocarbons cited in Method
TO-1 (Figure 3). Airborne concentrations of these collected adsorbates
ranged from 50 to 100 parts per trillion in a 10 liter air sample. A
desorption chamber temperature of 330°C and an initial column
temperature of 35°C provided effective chromatography. A 1% SP-1000
phase coated on 60/80 mesh Carbopack B was used to evaluate the
adsorbates in Methods TO-2 and TO-3 (Figure 3). This column also was
effective for evaluating the T0-1 adsorbates.
A 60 meter x 0.75 mm ID SUPELCOWAX*"l0 bonded phase capillary
column (1.0pm phase film thickness) was effective for evaluating the
TO-1, TO-2, and TO-3 adsorbates (Figure 4). A desorption chamber
temperature of 330°C and an initial column temperature of 35°C, the same
temperatures as used with the packed column, were chosen. The
adsorbates cited in all three methods were analyzed effectively under
the same analytical parameters.
Desorption efficiency for these 24 adsorbates was determined by
vapor-spiking the adsorbent tube with known quantities of each
adsorbate. The adsorbates were injected directly through the thermal
desorber/gas chromatographic syBtem to establish the control (1002)
value. Desorption efficiency waB determined by comparing peak areas for
desorbed and directly injected samples. Desorption efficiency was
determined for both the nonchlorinated and chlorinated adsorbates cited
in the EPA methodology (Table IX). Recovery data also were determined
for bromoform and acrylonitrile.
In conclusion, a multi-bed adsorbent tube containing three Class I,
non-specific adsorbents adsorbs a wide range of airborne organic
contaminants in a size-exclusion mode. When this tube is interfaced
with a thermal desorber capable of rapidly transferring adsorbates to a
GC column, accurate quantitative and qualitative data are obtained for
the adsorbates of interest.
REFERENCES
1. A.V. Kiselev and Y.I. Yashin, Gas Adsorption Chromatography, Plenum
Press, Hew York 1969, pp. 11-16.
2. U. S, EPA, Characterization of Sorbent Resins for Use in Airborne
Environmental Sampling, EPA Document # 600/7-78-054. 1978.
3. U. S. EPA, A Compendium of Methods for the Detemnlnation of Toxic
Organic Compounds in Ambient Air, EPA Document $ 600/4-84-041. 1984.
^British Pat. No. 1310422
German Pat. No. 1935500
764
-------
Table I
Classification of Molecules and Adsorbents by Capacity
for Nonspecific and Specific Interaction (1)
Adsorbent Class
Class I
Class II
ClaBS III
(no ions
(localized
(localized
or active
positive
negative
Molecule Type
groups)
charges)
charges)
Group A (e.g., n-alkanes)
spherically symmetrical
shells or o-bonds
Group B (e.g., aromatic hydrocarbons)
electron density locally
concentrated on bonds or
links, II-bonds
Group C (e.g., organometallics)
positive charge localized
on peripheral links
Group 0 (e.g., acids and bases)
functional groups with locally
concentrated electron density
and positive charge on
adjacent links
nonspecific interactions
governed mainly by
dispersion forces
nonspecific
interactions
nonspecific
and
specific
interactions
Table II
Physical Properties of the Class I, Nonspecific Adsorbents
Used in the Multi-bed Tube
Adsorbent
Carbotrap C
Carbotrap
Carbosieve S-I1I
Surface Area
(nr/gran)
10
100
800
Density
(grams/ml)
0.72
0.38
0.61
Table III
Specific Retention Volumes Show Class I Adsorbents
Have Hydrophobic Properties
Adsorbent
Carbosieve S-III
Carbotrap
Adsorbate
water
vinyl chloride
water
chlorobenzene
Specific Retention Volume
(ml/gram)
2.3 * lO^
1.26 x 10®
8.0 * 102
1.58 x 106
765
-------
Table IV
Equation for Calculating Specific Retention Volume (2)
Breakthrough Volume (V|) ¦
calculated volume of gas
ml of gas passing through the system
grams of adsorbent weight of adsorbent (grams)
(tr ~
j Fc
wa
Table V
Physical Properties of Adsorbents
Adsorbent
Property Carbotrap Tenax GC
Surface Area (m2/g) 100.0 23.5
Particle Size (mesh) 20/40 35/60
Density (g/ml) 0.38 0.14
Pore Volume ( cc/g) 0.053
Bed Weight (g) 0.4637 0.1650
Total Surface Area (m2) 46.37 3.878
Amberlite XAD-2
364.0
20/40
0.38
0.854
0.3966
144.4
Table VI
Specific Retention Volumes for Adsorbents
V| 20°C (ml/g)
Adsorbate Carbotrap Tenax GC Amberlite XAD-2
n-Decane 4.79 x 10^ 1.56 x 107 3.36 x 1Q7
Benzylamine 2.23 x 107 3.57 x 106 1.63 x 107
Chlorobenzene 1.58 x 106 1.51 x 105 4.84 x 105
p-Xyleae 4.24 x 107 3.88 x 105 7.95 x 106
p-Cresol 2.06 x 107 1.50 x 107 4.96 x 106
n-Pentanoic Acid 4.31 x 10^ 9.78 x 10^ 1,01 x 10^
Cyclohexanone 2.04 x 106 1.06 x 106 6.27 x 10^
2-Methyl-2-Propanol 6.52 x 1G3 6.86 x 102 5.42 x 103
Table VII
Specific Retention Volumes for Adsorbates on Carbotrap C
Adsorbate V| 20°C (ml/g)
n-Decane
2.97
X
Benzylamine
5.38
X
10*
Chlorobenzene
5.48
X
103
p-Xylene
3.75
X
10*
p-Crasol
1.27
X
105
n-Pentanoic Acid
1.71
X
105
Cyclohexanone
8.28
X
103
2-Methyl-2-Propanol
3.17
X
103
766
-------
Table VIII
Specific Retention Volumes for Adsorbates on Carbosleve S-III
Adsorbate
Chloromethane
Vinyl Chloride
V| 20°C (ml/g)
1.83 x 104
1.26 x 106
Table IX
Desorptlon Efficiency of the Multi-bed Trap
for Adsorbates Cited in EPA Methodology
Desorption
Adsorbate Efficiency (%)
Benzene 100
1-Heptene 99
n-Heptane 99
Toluene 99
Ethylbenzene 100
m-Xylene 100
p-Xylene 99
o-Xylene 99
Iaopropylbenzene (Cumene) 99
1.1-Dichloroethylene 101
3-Chloropropene 100
Chloroform 94
1.2-Dichloroethane 102
1,1,1-Trichloroethane 100
Carbon Tetrachloride 99
1,2-Dichloropropaoe 101
Vinyl Chloride 101
Dichloromethane 98
Chlorobenzene 100
Tetrachloroethylene 99
Trichloroethylene 101
Bromoforra 100
Acrylonitrile 100
767
-------
Figure 1
Multi-bed Adsorbent Tube
a i i i i i i i i i i i i i mm i i i i i i iMj
«*•••#< « e e • • • « ¦
«*•««•« Hp • «Hi ••«•••
• •••••• e^H • »*••• »••«••• ¦
• ••««•• settee
• e e e e e e • ••••• • e e e e e e ¦
....... M. . .
Carbo
Bed
rapC
Carbotrap
Bed
Carbosieve S-I
Bad
Sampling Flow
Deaorptlon Flow
Figure 2
Comparison of Desorption Temperatures
Acetylene/J1Ethan#
Ethylene
Ethylene
Acetylene
200*0
i Ethane
330°C
L_
0 2 4
Mln.
768
-------
Figure 3
Thermal Desorption Analyses on a Packed Column
cs
TO-1 Hydrocarbons
CS
1.
2.
3.
4.
B.
6.
7.
8.
«.
I I I I I I
10 12 14 16 18 20 Mln.
Methanol
(carrier solvent)
Benzene
1 -Heptene
n-Heptane
Toluene
Ethylbenzene
m-Xylene
p-Xytene
o-Xylene
Cumene
(laopropylbenzene)
CS
¦ lliJl
1 I l I I I I
2 4 6 8 10 12 14 Mln.
TO-1 Chlorinated Hydrocarbon!
CS Methanol
(oarrter solvent)
1. Chloroform
2. 1,2-DloMoroethime
3. 1,1,1-Trlchloro ethane
4. Carbon Tetrachloride
B. 1,2-Dlchloropropane
6. TVIehloroethylene
7. Bromoform
8. Tetraqhteroethylene
I Impurity
8. Chlorobenzene
TO-2 and TO-3 Adeorbetea
Vinyl Chloride
Olohtorome thane
Mln.
10
-------
Figure 4
Thermal Desorption Analyses on a Wide Bore Capillary Column
cs
i
0
I
4
I
6
u
7 9
TO-1 Hydrocarbons
1. n-Heptane
2. 1 -Heptane
CS Methanol (carrier solvent)
3. Benzene
4. Toluene
6. Ethylbenzene
6. p-Xylene
7. m-Xylene
8. Cumene llaopropylbenzene)
9. o-Xylene
I Impurity
JL
10 12 14 16 Min.
10
TO-1 Chlorinated Hydrocarbons
1. Carbon Tetrachloride
2. 1,1,1-TMchloroethane
3. Dichloromethane
4. TVfchloroethytene
B. Chloroform
6. Tetrachloroethylene
7. 1,2-Dichloropropane
8. 1,2-Dlchloroethane
9. Chlorobenzene
10. Bromoform
I Impurities
i i i i l
8 10 12 14 16 18 20 Min.
Vinylidene Chloride
Vinyl Chloride
I
T0-2 and T0-3 Adsarbates
I l I I I I I
< e 10 12 14 1fl 18
Min.
770
-------
INDEX
A
Accreditation
707
Acetylene
642
Acidic Deposition
158,168,174,183,189,195,201,208
727
Acidity
671
Acids
541
Add-On-Catalyet
685
Aerosol
183
Agencies
659
Air Cleaning
85
Air Toxics
238,659
Aldehydes
579
Ambient Air
5,12,21,30,35,42,57,63,69,75,238,
253,271,277,294,398
Ammonium
183,379
Aqueous Solutions
322
Aromatics
642
Asbestos
648
Ash
541
Attenuated Total Reflection
183
Automated Monitoring
294
Automotive Emissions
573, See also Diesel Exhaust,
Gasoline Engine ExhauBt
B
Benzene
140,265,516
Bulk Air Sampling
508
C
Cadmium
631
Calcium
671
Calibration
473,477,745
Cancer
555,562,573,579,585,591,597
Canister-Based Samplers
253,271,294,755
Carbon Absorption
516
Carbon Monoxide
685
Cement Kiln
671
Charcoal Sorbent Tubes
516
Chemical Mass Balance
502,691
Chemical Speciation
605
Chemiluminescence
398
ChemometricB
473,477,483,496,502
Chlorine
547
Cigarette Smoke
85,140
Classification
496
Cluster Analysis
496
Coal Combustion
502
Combustion
541
Computer Modeling
461
Conductivity
727
Continuous Emission Monitoring
617,745
Continuous Gas Analyzers
667
Control Technology
659
Creosote
685
771
-------
D
Data Base
259,341
Denuder
379
Design
529
Diesel Exhaust
69
Diethylhydroxylamine
461
Dioxins
316,477
Dispersion Modeling
516
Diurnal Variation
183,277,265
Drinking Water
123,348
Dry Deposition
195,201,358,366,373,379,387,398,
404
Dust
671
E
Efficiency
624
Electron Capture Detector
238
Electron Microscopy
496,611,648
Elemental Carbon
573
Ethylene Oxide
624
Expert Systems
496
Exposure Assessment
35
F
Fiber Loss
648
Field Testing
529
Filters
648
Flux Chamber
529
Forests
189
Formaldehyde
442
Fuels
541,547
Furans
316
G
Gaa Chromatography
140,238,282,305,477,624,755
Gas Collection System
516
Gas Mixtures
745
Gasoline Engine Exhaust
69,420
Grab Sampling
294
Groundwater
123
H
Halocacbons Analysis
238
Halogens
547
Hazard Ranking
665
Hazardous Wastes
35,508,516,522,529,535,541,547
Health Hazards
35
Health Risk
123
Heat Exchange Fluids
35
Henry 1s Law Constant
322
High-Volume Sampling
311,653
Hospital Waste
624
Hydrogen Chloride
745
772
-------
Hydrogen Sulfide
617
I
Incineration
535,541
Indoor Air
77,85,91,98,104,109,115,123,132,
140,152
Industrial Wastes
516
Infrared Spectroscopy
183
Inlet Device*
158
Instrumentation Development
277
Integrated Air Cancer Project
555,562,573,579,585,591,597
Integrated Air Sampling
294
Ion Chromatography
547,677
J
Jet Engine Ejdiaust
435
K
Keynote
1
L
Lanfill Gas Emissions
516
Laser Microprobe
611
M
Maximum Allowable Exposure
140
Megabore Column
238
Metals
'541
Methods Comparison
174,183,358,366,373,379,387,398,
404
Microcoulometric Titration
547
Microwaves
541
Mobile Monitoring
300
Monitoring
671
Multiple Linear Regression
502
Multivariate Analysis
477
Municipal Incineration
502
Municipal Waste
289,516
Mutagenicity
562,591
N
n-Alkanes
311
Networks
727
Neutron Activation
502
Nickel
605,611
Nitrate
183,379
Nitric Acid
379
Hitro-PAH
426, See also Polycyclic Aromatic
Hydrocarbons
Nitrogen Dioxide
379,398
Nitrogen Oxides
461
Nitrous Acid
379
Noncrittria Pollutants
289
Nonmethana Organic Compounds
271,277,294,755
773
-------
0
Oil Combustion
502
Organics
5, 12, 21,30,35,42,57,63,69,75,208
Overview
5,555
Oxygen Bomb
547
Ozone
271,442
P
Particulate Emissions
697,685,691
Particulate Monitoring
671, See also Suspended Particu-
lates, Aerosols
Passive Sampling
132
Pattern Recognition
496
PCBs
35,311
PCDD/PCDF
477
PDFID Method
271,277,755
Performance Audit
727
Pesticides
734
Phase Analysis
605
Photoionization
224,232,348
Photolysis
35
Physical/Chemical Properties
305,311,316,322,329,341,348
Polycyclic Aromatic Hydrocarbons
69,329,341,411,483
Positive Pressure Sampling
294
PreseparatocB
168
Principal Component Analysis
477
Protocol
529
Pyridine
140
Pyrolysis
642
Q
Quality Assurance
712,721,727,734,745,755
R
Radiocarbon
573
Radon
721
RCRA Wastes
541
Real Time Analysis
183,398
Receptor Modeling
496,502
Reference Methods
638
Regulator
624
Residential Wood Combustion
573
Resin Trapping
140
Retrofit
685
Risk Assessment
35,516
Road Dust
691
S
Samplers
245
Scrubber
624
Semivolatiles
5,12,21,30,35,42,57,63.69,75,109,
311
SFC Analysis
69
Siting Criteria
727
774
-------
Smog
271,461
Soil Contamination
35
Soot
642
Source Apportionment
562,573
Source Characterization
218
Source Monitoring
605,611,617,624,631,638
Source Testing
516
Stability Studies
271
Standards
212,745
Static Charge
648
Styrene
642
Sulfate
183
Sulfur Dioxide
379
Supercritical Fluids
69
Surface Impoundment
529
Suspended Particulates
85,152,573,605
T
Technical Assistance
659
Teflon Filters
183
Temperature Compensation
398
Test House
104
Teat Kits
547
Thermodynamic Properties
341
Tracers
585
Transition Flow Reactor
379
Trichloroethylene
123
Trihalomethanes
348
Tunable Diode Laser
398
U
Used Oil
547
V
Vapor Phase
140
Vapor Pressure
35,305,311
Vapor-Particle Partitioning
311
Vegetation Effects
671
Vinyl Chloride
516
Volatile Organics
212,218,224,232,238,245,253,259,
265,271,277,282,289,294,300,522
Volatilization
35,123
W
Waste-To-Energy
289,745
Wind Trajectory Method
502
Woodstove Emissions
420,677,685,691,697,707
X
X-Ray Fluorescence
502,547,611
775
------- |