United States
            Environmental Protection
             Health Effects Research ani)
             Environmental Monitoring ar, i
             Support Laboratories
             Research Triangle Park NC 2~/ '11
EPA 600/9-79-032
June 1979
            Research and Development
Proceedings of the
Symposium on the
Development and Usage
of Personal  Monitors
for Exposure and  Health
Effect Studies


Research reports of the Office of Research and Development,  U.S.
Environmental Protection Agency, have been grouped into nine series.
These nine broad categories were established to facilitate further
development and application of environmental technology.  Elimination
of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields.  The nine series
     1.  Environmental Health Effects Research
     2.  Environmental Protection Technology
     3.  Ecological Research
     4.  Environmental Monitoring
     5.  Socioeconomic Environmental Studies
     6.  Scientific and Technical Assessment Reports (STAR)
     7.  Interagency Energy-Environment Research and Development
     8.  "Special" Reports
     9.  Miscellaneous Reports.
COVER:  A physiologic monitoring device incorporated into a piece
of jewlery.  The device contains a monitoring electrode that
transmits the ECG through electronic circuitry to activate a
liquid crystal display of the subject's electrophysiologic heart
pattern.  The left-hand picture shows the device in an open
position; the center picture shows it with the electrodes in place
on the body; and the right-hand picture shows the device in a closed
position, the configuration in which it is usually worn.  (Photos
courtesy of Mary Ann Scherr and Dr. George S. Malindzak, Jr.)

                                    June 1979
Proceedings of the
Symposium on the
Development  and  Usage
of Personal  Monitors
for  Exposure  and  Health
Effect Studies
Sponsored by
U.S. Environmental Protection Agency
January 22-24, 1979
Chapel Hill, North Carolina

Symposium Chairmen

David T. Mage, Ph.D.

Statistical and Technical Analysis Branch
Environmental Monitoring and Support Laboratory
Research Triangle Park, North Carolina


Lance Wallace, Ph.D.

Office of Monitoring and Technical Support
Office of Research and Development
Washington, D.C.
Coordination and Editing by
Kappa Systems, Inc.
Arlington, Virginia 22209
U.S. Environmental Protection Agency
Health Effects Research Laboratory
and Environmental Monitoring and Support Laboratory
Research Triangle Park, North Carolina 27711


This report has been reviewed by the Health Effects Research Laboratory, and
the environmental Monitoring and Support Laboratory, U.S. Environmental Pro-
teitlon Agency, and approved for publication*  Approval does not signify that
the <<>nterita necessarily reflect the views and polities of the U.S. Environ-
mental Protection Agency, nor does mention of trade names or commerlral products
constitute: endorsement or recommendation for use.


     Opening Remarks

         Ralph W.  Stacy, Ph.D.
     Role of the EPA Environmental Monitoring and
     Support Laboratory in the Development of Personal
     Exposure Monitors

         Thomas R. Mauser, Ph.D.
     Personal Air Quality Monitors;  Uses in  Studies
     of Human Exposure

         Lance Wallace, Ph.D.
     The Need for Personal Air Pollution Exposure                    19
     Monitors and Applications in Energy Development

         Leonard D. Hamilton, M.D., Ph.D.
     Current NI08H Research on Passive Monitors                      27

         Mary Lynn Woeblcenberg

     Electrochemical Methods for Development of                      35
     Personal Exposure Monitors

          Joseph R. Stetter, Ph.D., Donald R. Rutt,
          and Michael R. Graves
     Portable Ozone and Oxidant Monitors for Random                  47
     Field  Surveys

          Gary P. Heitman and Richard L. Sederquist
     Personal  Sampler  for Measurement of Ambient                     57
     Levels  of N0_

           Edward  D.  Palmes, Ph.D.
      In-Vehicle Air Pollution Measurements:  A User's                65

           Richard A. Ziskind, Ph.D.
      Calibration of Personal  Exposure Monitors for                   77
      Personal Exposure and Health Effect  Studies

           Charles L. Kimbell
      A New Family of Miniaturized  Self-Contained CO                  83
      Dosimeters and Direct Reading Detectors

           Arnold H. Gruber, Arnold G. Goldstein,
           Anthony B. LaConti,  Ph.D.,  Harry  G. Wheeler,
           and John Martin
      Design and Performance of  a  Reliable Personal                  101
      Monitoring System for  Respirable Particulates

           William A.  Turner,  John D. Spengler, Ph.D.,
           Douglas W.  Dockery, and Steven D.  Colome
      Personal  Exposure  to  Respirable  Particulates                   111
      and Sulfates:   Measurement  and Prediction

           Douglas W. Dockery  and John D.  Spengler, Ph.D.

The Tandem Filter Package                                      131

     Robert W. Shaw, Ph.D., and Robert K. Stevens

An Evaluation of Personal Sampling Pumps in                    145
Sub-Zero Temperatures

     Carl D. Parker and Joan C. Sharpe
Electrochemical Air Lead Analysis for Personal                 161
Environmental Surveys

     Francis J. Berlandi, Ph.D., Gerald R. Dulude,
     Reginald M. Griffin, Ph.D., and Eric R. Zink,
Studies of Semiconducting Metal Oxides in                      173
Conjunction with Silicon for Solid State Gas

     Angel G. Jordan, Ph.D., David J. Leary, Ph.D.,
     Gulu N. Advani, and James 0. Barnes, Ph.D.
Microcomputer Control and Information Processing               191
Technology for Semiconductor Gas Sensors

     David T. Tuma, Ph.D., and Paul K. Clifford
Management Strategy in the Design and Use of                   207
Personal Monitors for Environmental Studies

     Ralph W. Stacy, Ph.D.
The Usage of Personal Monitors for Determination               217
of Dosage and Exposure in Health Effect Studies

     David T. Mage, Ph.D.
Carboxyhemoglobin Determinations from Expired Air              229

     Harold W. Tomlinson

     Intelligent Personal Physiologic Monitors in                   237
     Clinical Environmental Health Effects Research
     (Part I)

          Mathew L. Petrovick
     Design of Personal Monitors for External and                   259
     Internal Physiologic Studies in Health Effects
     (Part II)

          Mathew L. Petrovick and Edward D. Haak, Jr., M.D.
     Personal Cardiopulmonary Electrode Monitoring                  281

          Paul N. Kizakevich

     A Respiration-Controlled Personal Monitor                      297

          Donald J.  Sibbett, Ph.D., and Rudolph H.
          Moyer, Ph.D.
     Personal Monitor Cosmetology:  An Aesthetic                    309

           George  S. Malindzak, Jr., Ph.D., and Mary
           Ann Scherr
      'Shirtsleeve Workshop"                                         327
     A New Personal Organic Vapor Monitor with In Situ              365
     Sample Elution

          Donald W. Gosselink, Ph.D., David L. Braun,
          Haskell E. Mullins, and Sandra T. Rodriguez
     A Combination  Sorbent System for Broad Range                   383
     Organic Sampling in Air

          Joseph J. Brooks, Ph.D., Diana S. West,
          Donald J. David, and James D. Mulik

Adapting Commercial Voltammetric Sensors to                    413
Personal Monitoring Applications

     Manny Shaw, Ph.D.
Solid Sorbent for Acrolein and Formaldehyde in Air             425

     Avram Gold, Ph.D., Thomas J. Smith, Ph.D.,
     Christoph E. Dube, and John J. Cafarella
Passive Membrane-Limited Dosimeters Using                      437
Specific Ion Electrode Analysis

     Charles E. Amass
Personal Monitoring by Means of Gas Permeation                 461

     Philip W. West, Ph.D., and Kenneth D.
     Reiszner, Ph.D.
Personal Air Pollution Monitors:   New Developments             473

     Robert W. Miller and Byron Denenberg

A New Sampling Tool for Monitoring Exposures to                479
Toxic Gases and Vapors

     John C. Gillespie and Leah B. Daniel

Panel Discussion                                               493

     The many benefits of our modern, developing,  Industrial society are
accompanied by certain hazards.  Careful assessment  of  the relative risk of
existing and new man-made environmental hazards  Is necessary for the estab-
lishment of sound regulatory policy.  These regulations serve to enhance the
quality of our environment to promote the public health and welfare and the
productive capacity of our Nation's population.

     The Health Effects Research Laboratory, Research Triangle Park, North
Carolina, conducts a coordinated environmental health research program In
toxicology, epidemiology, and clinical studies using human volunteer subjects.
These studies address problems In air pollution, nonlonlzlng radiation, envi-
ronmental carclnogenesls, and the toxicology of  pesticides, as well as other
chemical pollutants. The Laboratory participates  In the development and re-
vision of air quality criteria documents on pollutants  for which national
ambient air quality standards exist or are proposed, provides the data for
registration of new pesticides or proposed suspension of those already In use,
conducts research on hazardous and toxic materials,  and is primarily respon-
sible for providing the health basis for nonlonlzlng radiation standards.
Direct support to the regulatory function of the Agency is provided In the fora
of expert testimony and preparation of affidavits  as well as expert advice to
the Administrator, to assure the adequacy of health  care and surveillance of
persons having suffered imminent and substantial endaangerment of their health.

     Measurement and monitoring research efforts are designed to anticipate
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 monitoring compliance with regula-
tions, and to evaluate the effectiveness of health and environmental protection
efforts through the monitoring of long-term trends.  The Environmental Monitor-
ing and Support Laboratory, Research Triangle Park, is responsible for develop-
ment of:  environmental monitoring technology and systems; Agency-wide quality
assurance programs for air pollution measurement systems; and technical support
to the Agency's operating functions, including the Office of Air, Noise, and
Radiation, the Office of Toxic Substances, and the Office of Enforcement.

     This conference was conducted by the Environmental Monitoring and Support
Laboratory, Research Triangle Park, at the request of the Health Effects Re-
search Laboratory.  Thirty-five papers were presented on advances in personal
monitor designs and applications.  The Environmental Monitoring and Support
Laboratory will play a continuing role in their development through programs of
field testing, quality assurance, and determination of their equivalence to the
EPA-designated instruments for the pollutants to be monitored.
F.G. Hueter, Ph.D.
Health Effects  Research Laboratory
Thomas R. Hauser, Ph.D.
Environmental Monitoring
and Support Laboratory

Opening Remarks
     Ralph W. Stacy, Ph.D.
     Clinical Studies Division
     Health Effects Research Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina
     I trust that no one will dispute my contention that human research on
health effects of air pollutants,  carried out on controlled  and carefully
selected cohorts of human subjects under carefully controlled  conditions,
is entirely necessary in the scheme of research required for the intelligent
determination of ambient air standards.  This type of research is the primary
mission of the Clinical Studies Division of EPA.  Our facilities, located
here in Chapel Hill, N.C.,  are among the most sophisticated  in the world, and
we are currently carrying on a heavy research schedule.

     However, none of us at the Clinical Studies Division labors under the
delusion that our research will supply all of the answers required for air
pollution control.  Our highly controlled scientific studies,  carried out on
selected subject groups with exposures artificially produced in an idealized
time pattern, supply the most accurate data available on the response of
humans to pollution exposure, but  we can never hope to study more than a
small and narrowly defined sample  of the population.  These  studies will
form a data base that is more accurate and more carefully defined than we
could obtain outside the laboratory, and they will probably  function as a
comparison standard for other types of studies.

     Ultimately, however, we must  have information on:  1) the response of
all segments of the population to  pollutant exposure, 2) the amounts and time
patterns of pollutant exposures which actually do or are likely to occur in
the general population, and 3) the correlation of the exposures that occur
with the effects that result.  The exposure data and the health effects data

must be obtained simultaneously  on  people  who  are  of  all  ages,  both  sexes,
well and ill, of all ethnic  groups,  and  in all existing life  situations.
Only when we have  such all-inclusive data  can  we make assuredly correct,
defensible  judgments of  regulatory  requirements.   In  the  final  analysis,  this
is an epidemiological problem, and  it is an extremely important and  difficult
one.  I repeat—this problem is  a very large one;  we  dare not treat  it  lightly.

     Recent clinical researches  on  air pollution effects  have made it abun-
dantly clear that  humans vary enormously in their  physiologic responses to
polluting agents.   For example,  we  have  shown  clearly that in a group of  100
individuals, we can expect  that  between  15 and 30  of  them will  be quite sen-
sitive to photochemical  oxidants (specifically, ozone).   Some 20 to  30  of these
persons will not respond at all  to  the same exposure, and the remainder will
respond to  some degree.   This is actually  a very important observation  be-
cause it means that the  usual scientific experiment—in which the means of a
number of individuals in a  control  group and an experimental  group are  com-
pared—is simply not applicable  to  a true  understanding of pollutant effects.
We expect this variability  to become a great deal  more apparent when we ex-
pand our  studies to include the  very young, the old,  the  ill, and so on.
This means, of course, that we must re-examine our epidemiological and  sta-
tistical  techniques, and that we must be sure  that they are adequate.

     We have also  shown  clearly  that pollutant health effects often  are not
synchronized in  time with pollutant exposure.   An  effect  may  be delayed,  or,
with continued exposure, an effect  may be  appreciably reduced or even eli-
minated entirely.   Thus, we need continuous measurements  over extended  time
periods, and this  imposes additional methodological  requirements.

     In addition,  we are now able to identify  some health effects data  which
are significant  indicators  of pollutant-produced health decrements;  others
are now known  to contribute little  to the  identification  of hazards; and  there
are some whose significance has  not been tested.   Thus, we have some homework
to do before we can head into the final  design of  measuring devices.

     All  of this simply  means that  the kind of information we really must have
is a total  picture of pollutant  exposure and health  effects in the general
population, measured under  normal living conditions.  This leads us  directly
to the conclusion  that the  measuring instrument must  accompany the monitored
individual  wherever he or she goes  and in  whatever he or  she  does.   It  is not
enough to measure  one-time  health status of members  of  the population and to
compare these effects with  pollutant exposure  patterns obtained at some fixed
measurement station years or blocks or miles away.

     It means that we at the Clinical Studies Division are completely con-
vinced of the importance of the development of personal monitors, especially
those in our own field of competence, and that we anticipate that our ex-
posure chambers may well turn out to be the ultimate proving grounds for the
personal monitors developed by our engineers and bioengineers, and for those
monitors developed by other parts of EPA and other Agencies.

     We are under no illusions concerning the scope and difficulty of the
problems of development of personal monitoring equipment and the techniques
involved in their use.  We know that taking on this additional task will put
further strain on our already critical staffing and space problems.  Neverthe-
less, we intend to do our part and to do it to the very best of our ability.

     We welcome this opportunity to play a role in what we feel is one of
the major developments in environmental control of this time period.  We also
recognize the value of this symposium, and we appreciate the contribution it
can make to this important human effort.  We pledge our support and colla-
boration to the development of personal monitors for pollutant exposures and
health effects measurement.

     We look forward to taking part in this symposium, and we hope that all of
you enjoy the symposium and the Carolina setting in which we are meeting.

Ralph W. Stacy, Ph.D.
Clinical Studies Division
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina  27711

Role of the EPA  Environmental Monitoring
and Support  Laboratory in the  Development
of Personal Exposure Monitors
     Thomas R. Mauser, Ph.D.
     Environmental Monitoring and Support Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     Among other programs, the  EPA Environmental  Monitoring and Support
Laboratory (EMSL) at Research Triangle Park (RTP), N.C., has organizational
responsibility for evaluating and selecting air pollution monitoring  instru-
ments for use in studies required by the Clean Air Act as amended in  1977.
In this capacity, EMSL recommends to the Administrator monitoring instruments
for EPA designation, analytical techniques, and instrument performance spec-
ifications, in addition to conducting the air monitoring equivalency  program.
The EMSL/RTP also develops total monitoring systems and is responsible for
the development and operation of the air pollution monitoring quality assur-
ance program.  Along with the EMSL/Las Vegas, the EMSL/RTP is currently
applying its expertise in these areas to study designs using personal air
monitors for human exposure assessment.

     The EMSL/RTP has responsibility within the EPA for the air pollution
 monitoring studies performed jointly with the Health Effects Research Labo-
 ratory/ RTP health studies on human  subjects exposed to both ambient and  cham-
 ber-controlled concentrations of  air contaminants.  These air quality data
 and health data are analyzed by EPA biostatisticlans to determine the degree
 of association, if any, of the measured health effects with the air quality
 data.  Examples of two studies currently underway are measuring the effects

of CO intrusion into vehicles and the effects of short-term exposures to high
levels of N02.

     We recognize that the ambient exposure levels monitored at a single
fixed location may not be representative of the total exposure of residents
in the study areas.  To obtain a better definition of the spatial variability
of pollution within the study community and the variations of exposure re-
ceived by the subjects, a combination of fixed monitoring stations, mobile
monitoring stations, and personal exposure monitors can be used to give a
better estimate.  The personal monitors used  in these studies are tested by
EMSL/RTP for interferences, accuracy, precision, temperature stability, etc.
The results of the personal exposure monitoring data collection can be analyzed
to test the hypothesis that the fixed-station data represent the average
community exposure concentrations within a specified concentration bound
and level of confidence.  If the variations in exposure are significant,
the biostatisticians can relate the measured  variability of exposure to the
health data by statistical  techniques.

     At the present time, the state-of-the-art personal exposure monitors
have been developed primarily for occupational exposures to high levels of
pollutants.  We in EMSL see a need for  increased development of personal
exposure monitors for the lower ambient levels to which the general public
is exposed.

     In summary,  I hope that this conference  will Improve communication
between personal monitor users and personal monitor manufacturers and stimu-
late progress in  their development as reliable instruments for human exposure
studies.  We anticipate that personal monitors will be used with increasing
frequency as research monitoring  tools  for studies of health implications
of air pollution and for relationships  of different monitoring strategies.

Thomas R. Hauser, Ph.D.
Environmental Monitoring  and  Support Laboratory (MD-75)
U.S. Environmental  Protection Agency
Research Triangle Park, North Carolina  27711

Personal Air Quality Monitors: Uses in
Studies of Human Exposure
     Lance Wallace, Ph.D.
     Office of Monitoring and Technical Support
     Office of Research and Development
     U.S.  Environmental Protection Agency
     Washington, D.C.
     I would like  to explain very briefly how this  conference fits into EPA's
R&D Program.  The  Office of Research and Development  (ORD), is one of the
six main offices of the EPA.  The other offices  are often called the program
offices, and are responsible for carrying out EPA's legislative mandate in
areas such as air, water, noise, radiation,  pesticides, solid waste, and
toxic substances.  These offices express their needs  for research to ORD,
which is then expected to find a way to answer those  needs.  This conference
is one way we have chosen to address the need expressed in many different
ways recently, by  groups both inside and outside EPA,  for methods of measur-
ing human exposure directly.

     In the last 2 years alone, four major groups have specifically requested
action on the development and use of personal air quality monitors to measure
day-to-day human exposure to air pollutants.

     The National  Academy of Sciences,  in a  1977 report entitled Enviromental
Monitoring, said:  "We recommend that EPA coordinate  and support a program
to foster the development of small, quiet, sensitive,  and accurate personal
air quality monitors for use in conjunction  with other methods of measuring
human exposure to  ambient air quality"  (1).   In  1977  the Second Task Force on
Research Needs in  Environmental Health  Sciences  recommended that:  "...research
is needed on the application of promising instrumental approaches to the develop-
ment of personal monitors which, through use of  micro-electronic components and
small pollutant sensors, could be used  to monitor individuals within popu-
lations of interest" (2).

     EPA's own task force, the Standing Air Monitoring Work Group, charged with
recommending an Agency-wide 5-year monitoring strategy, stated:  "As a concluding
observation, SAMWG believes that EPA  should take a more active research role In
determining the usefulness of personal air pollution  samplers to measure the
exposure dose to representative individuals'* (3).  And, a congressional commit-
tee in April of 1978 authorized a large increase in appropriations for studying
effects of criteria pollutants, stating that:   "...the Agency needs to improve
the data base relating  actual personal exposures to monitored ambient air pol-
lution levels in order  to increase the reliability of epidemiological studies.
This should be done by  development of personal  dosimeters..." (4).

     The last major conference on personal air  quality monitors was held 3 1/2
years ago at Brookhaven National Laboratory.  There, many possible technologies
were ranked according to  their probability of payoff  within a few years, and
a national R&D program  of about $1.5  million a  year for 5 years was recommended

     Such a program has not materialized  to date, but some activity is taking
place.  For example,  the Environmental Sciences Research Laboratory here in
Research Triangle Park, N.C., has a project to  develop a portable monitor for
organics.  This project will be further described later in this conference.
This same laboratory  supported a feasibility study (6) for personal monitors
several years ago, and  is now working on  a monitor for respirable particles
and gases, which will be  described in the paper by Dr. Robert Shaw and Mr.
Robert Stevens.  A second ORD laboratory, the Health Effects Research Labora-
tory, has a briefcase-sized monitor for SO-, CO, and N0_ that it has developed
over the past 2 years and may use in  epidemiology studies.  A third ORD labora-
tory, the Environmental Monitoring and Support  Laboratory in Research Triangle
Park, has used a commercially available CO dosimeter  in a congressionally man-
dated study (7) of CO intrusion into  sustained-use vehicles such as buses,
taxis, and police vehicles.  This will be described further in the talk by
Dr. Richard Zisklnd.  Also, a fourth  ORD  laboratory,  the Environmental Moni-
toring and Support Laboratory in Las  Vegas, is  administering an interagency
agreement with the National Bureau of Standards (NBS) to develop one or more
personal monitors for a pollutant or  pollutants yet unnamed.  Dr. Jimmie
Hodgeson from NBS is here to give more information on NBS experiences with
personal monitors.*  Thus, I believe  we are seeing the first stages of an effort
that may lead to far more detailed knowledge of human exposure than we dreamed
of 5 years ago.
^Editor's Note:  A copy  of Dr. Hodgeson's  symposium  paper  was  not  available
at  the  time of  printing  and  thus  does  not  appear  in  these  Proceedings.


     For this effort to continue, however, even at a level far below the major
national commitment called for by the Brookhaven workshop, it is necessary to
keep its benefits clearly in mind.  What are the benefits of a balanced program
of development, evaluation, and use of personal monitors?  I can see the
following six major uses:

     1)  More detailed knowledge of individual, day-to-day exposure.  Obtaining
this knowledge requires that a large sample of people within a community be
equipped with personal monitors and keep records of their activities.  Together
with fixed-station measurements, traffic and commuting records, and other
personal activity models, the monitoring results will eventually allow esti-
mates to be made of the frequency distribution of exposures for the entire

     2)  Identification of high-risk subpopulations and quantification of
their exposures.  The exposures at the high end of the frequency distribution
are the controlling factors in determining the adequacy of our standards.
The more precise our knowledge of the exposures of the sensitive portion of
the population, the more precisely we can set our standards, avoiding both
unnecessary economic impact if the standard is too stringent, and inadequate
protection of human health if it is too weak.

     3)  Measurement of exposures during episodes.  Although it is generally
accepted that certain infamous air pollution episodes have cost scores or even
thousands of lives (or at least they have hastened deaths that were imminent),
no actual measurements have been made of individual exposures during episodes.
To effectively measure exposures during episodes, which last only 3 or 4
days, it would be necessary to have a team of trained volunteers and the means
of supplying them with monitors within a day of the beginning of the episode.

     4)  Validation of diffusion models and activity pattern models.  Use of
personal monitors in the field can provide data from hundreds of spatially
dispersed points, multiplying by many times the number of points available
from fixed-station locations.  The same data can be used to validate the
models of personal behavior—including commuting, shopping, and recreational
travel—that will be necessary to develop as part of our total exposure models.

     5)  Calibration of fixed-station readings.  A very valuable contribution
of personal monitor studies would be the establishment of a solid relation
between the fixed-station readings in a community to the exposures of at least
some population subgroups in that community.  If this is possible, then it
would be conceivable that fixed-station readings in other communities would

bear  a  similar  relationship to exposure.   This would allow estimates  of ex-
posure  to  be made for more cities than those in which personal  monitor studies
had been carried out, and would also open up the possibility of using his-
torical records of fixed-station measurements to estimate past  exposures of
community  groups.  Since chronic respiratory diseases depend on exposures over
a  number of years, this information would be valuable to the epidemiologist.

      6) Development of dose-response curves.  All five of the  above  purposes
can be  achieved without any concurrent measures of effects; they are  pure
exposure studies.  Such studies presuppose that the effects at  various concen-
trations—the dose-response curves—have been adequately established  by clini-
cal or  epidemiological studies, and have been documented in the criteria
documents  or other publications.  Of course, for many pollutants, this is not
the case.   For those pollutants, full-scale epidemiological studies,  utilizing
personal monitors for measuring dosages and lung-function tests or other med-
ical  end points for measuring responses, are necessary.

      At present, the studies just outlined have yet to be carried out.  As
an illustration of the likely cost and time involved in carrying out  such a
 study,  I borrowed some equipment—the CO dosimeter (9000 series) of Energetics
 Science and the Dupont personal sampling pump—and measured my  exposure while
 commuting  from Reston, Va., to Washington, D.C., by bus for a period  of 4
weeks last summer.  This period proved sufficiently long to characterize my
 average exposure to within 10 percent.  The instrument used is  a small, light
 electrochemical device that can fit into a pocket or hook onto  a belt.  The
 Dupont  pump is attached to the sensor by a flexible tube and can be inserted
 in a  nearby pocket.  A somewhat larger readout device can be placed in one's
 lap to  allow minute-by-minute records to be kept, or it can be  kept at the
 office  or at home to give a total exposure reading.  I preferred to keep a
 continuous record, which required leaving the dosimeter plugged in to the
 readout device while casting periodic glances (about every 5 minutes) at the
edgemeter  to document the changes that occurred in ambient concentrations.
Figure  1 shows a typical exposure record obtained using this technique.  You
can see the high concentration in the bus as I boarded it, at a pick-up
point where the engine had been idling for about 5 minutes.  The concentration
falls rapidly on the expressway (between 15 to 20 miles from D.C.), and then
begins  to  rise as stop-and-go traffic is encountered near the city.  It falls
again while crossing the Potomac River, rises to a peak of about 30 to 35 ppm
in the  center of the city, and falls to 5 ppm or so as I walk to work.  The
smooth  curve is the integrated exposure for the length of the trip, measured
in ppm-hours.  The total exposure was 35 ppm-hrs, over a total  travel time of
53 minutes, which results in an average concentration for the trip of 40 ppm*


    CO concentration
                              (36)   (80)
                                                                   (Puff from
                                                                   bus exhaust)
CO exposure
8:10      8:20      8:30     8:40      8:50      9:00      9:10      9:20     9:30 A.M.
FIGURE 1.  Exposure to carbon monoxide during  commuting:   sample trip.

     About 30 one-way trips were  taken, and  the range of exposures varied be-
tween 5 and 35 ppm-hrs, except for one trip.  Figure 2 shows the concentrations
encountered on this trip.  The instant I got  on the bus, the dosimeter sent
the edgemeter needle right off the scale (max = 100 ppm).  Thinking it might
be defective, I replaced it with  another one  from  the kit, but this one
behaved equally startlingly.  The bus was  air-conditioned with all of the
windows closed, so 1 opened a window to test  whether outside air would make
a difference.  (At 90° F on a bus full of  hot commuters, this is not an act
undertaken without careful deliberation.)  Within  a minute, the needle fell
to values between 25 and 75 ppm,  oscillating  swiftly in  response to traffic
conditions.  Closing the window resulted in  a rapid buildup to values MOO ppm.
The total exposure for a 1-hour trip was 65  ppm-hrs, and would have been >100
ppm-hrs if I had left the window  closed, as  I would have done had I not been
measuring the CO concentrations.  For a normal 8-hour working day, the driver
would have been exposed to between 800 and 1,000 ppm-hrs, which would be con-
siderably greater than the Occupational Safety and Health Administration
limit of 400 ppm-hrs.

     This exact bus  (the identical vehicle)  had been ridden the day before
without displaying excessively high concentrations.  Thus, an intriguing
possibility—possible high exposures to vehicle exhaust  gases due to mainte-
nance problems—presents itself.  This type  of observation was in fact the
precipitating factor in leading Congress to  mandate the  CO intrusion study
described earlier that our laboratory in the Environmental Monitoring and Sup-
port Laboratory, Research Triangle Park, is  carrying out.

     Figure 3 compares the average concentrations  measured inside the buses
with those measured  at the Community Air Monitoring Program (CAMP) station in
the central city.  The values encountered  inside the vehicles were typically
two-to-four times those measured  in the city. There was no correlation be-
tween the ambient and in-vehicle  measurements.

     One unusual finding here was that air-conditioned buses appeared to have
CO concentrations 60 percent above nonair-conditioned buses, the difference
being statistically  significant.  Since the  buses  were diesels with a separate
gasoline-powered jeep engine driving the air-conditioner, it seems possible
that the exhaust of  the air-conditioner motor may  have accounted for the higher
concentrations.  At  any rate, the value of the personal  monitor in such studies-
first in quantifying exposure, second in detecting possible unsafe levels,
and third in suggesting hypotheses for the source  of the in-vehicle pollution—
has perhaps been adequately demonstrated (8).


                                                                              Instantaneous Concentration

                                                                                     Integrated Exposure
                                                                                           C- Windows Closed
                                                                                           0- Windows Open

12:30 P.M.    12:40i    ,12:50'     1:00      1:10        1:201
                                                                         1:30       1:40 P.M.
FIGURE 2.   CO  exposure during  commuting:  worst  case  (air-conditioned  bus).

                                                               Carbon Monoxide
                                                               Concentrations In Vehicle
                                                            Ambient CO Concentrations (ln City
                                                            During Commuting Hour*
                               I  / /    1  i I I  I  1 I i 1 1  I  I  I
 7 8 9 1011 12 13 14 15 16 17 18 192021 2223 /    1345678910111213141516
                 July                              August

FIGURE  3.   Commuter exposures to carbon monoxide:  Washington, D.C., summer


     My final topic deals with the  future directions of research in this  field.

Four major questions require answers:

     1) What pollutants are of highest  priority?

     2) What technologies are most  promising?

     3)  What is the optimum balance among development, evaluation, and
utilization of personal monitors?
     4)  How should this research be funded?

     To the first question, I can bring responses recently solicited from EPA
representatives of relevant program offices.  The air program office has men-
tioned fine (inhalable) particulates as a high priority.  We are in the midst
of preparing for a possible standard on Inhalable particulates, with a decision
to be made in the early 1980's.  A monitor capable of being deployed in
fairly large numbers to establish dose-response curves and exposures to
sulfates, nitrates, trace metals, and other components of the total respir-
able particulate load, would be highly desirable.  Both the air office and
the Office of Toxic Substances would also favor a monitor capable of collect-
ing organics for later analysis in the laboratory.  The extent of nonworkplace
exposure to organics and to carcinogenic vapors is only hazily comprehended
as yet.  Other air pollutants of high priority to EPA will be discussed at
the Tuesday evening workshop.

     I will not attempt an answer to the second question:  "What technolo-
gies are most promising?"  It will be the major topic of our workshop session
Tuesday evening and of the summary panel discussion Wednesday afternoon to
which all are invited to attend.

     The question as to what is the optimum balance between development and
evaluation of personal monitors and their utilization in exposure studies may
be partially attacked in the following manner:  since the seven criteria
pollutants have already had their health effects quantified, relatively in-
expensive pure exposure studies—involving no concurrent health effects meas-
urements and utilizing the most reliable commercially available instruments—
should be of high priority.  These studies will have the immediate benefit
of providing the first reliable estimates of the exposures of an entire com-
munity population to a given criteria pollutant.  This far more detailed
knowledge of the population at risk could allow more efficient protective
measures to be taken.  Of course, a necessary prerequisite to such pure ex-
posure studies would be the rigorous evaluation by EPA laboratories of the
commercial instruments.

     On the other hand, the development of new instruments must look beyond
the criteria pollutants to the problems of the future—in particular, the
toxic organics and fine particulates.  This is not to say that development
of instrumentation for S02, N02, ozone, and CO should not be supported; only
that limited resources will require hard choices to be made.


     This  brings me  to  the final question—also called the  bottom line—the
question of  funding.  I have for several years advanced the thesis that  in
the absence  of  a mass market for personal monitors,  the Federal  government
must support their development.   At the same time,  I felt that it could  be to
industry's advantage to fund development, since the  more precise knowledge
of exposures would allow fine-tuning ambient standards, with a probable  re-
duction of economic  inefficiency associated with too-stringent standards.
A paper published by the National Academy of Sciences in 1977 called  for an
industry/government  coalition on a development program (9).

     Now,  as a  representative of government, I would make the same appeal—
that we work together on developing and using the methods for determining  more
exactly our  day-to-day  exposure  to air pollutants.   I am very much heartened
by the excellent  response of the industry to this symposium, and I am certain
we have a  stimulating 3 days ahead of us.

  1.   National Academy of Sciences.  Environmental Monitoring.   Studies  for the
      U.S.  Environmental Protection Agency, Vol.  IV.  Washington, D.C., May 1977.

  2.   U.S.  Department of Health, Education, and Welfare.   Human Health and
      the Environment—Some Research Needs.  Report of the 2nd  Task Force on
      Research Planning in Environmental Health Sciences.   NIEHS,  HEW Publi-
      cation #77-1277.  Washington, D.C., 1977.

  3.   Environmental Protection Agency.  Air Monitoring Strategy for State
      Implementation Plans.  Office of Air Quality Planning and Standards,
      Research Triangle Park, N.C. EPA 450/2-77-010.   June 1977.

  4.   U.S.  House of Representatives.  Authorized  Appropriations to the Office
      of  Research and Development, Environmental  Protection Agency. 95th
      Congress, 2nd Session.  Report #95-985, p.  7.  March 17,  1978.

  5.   Morgan, M.6., Morris, S.  Individual Air Pollution Monitors:  An Assess-
      ment  of National Research Needs.  Report of a workshop held at Brookhaven
      National Laboratory, July 8-10, 1975.  BNL  50482.  Energy Research and
      Development Corporation.  January 1976.

  6.   Harrison, 0.  et al.  Development Strategy for Pollutant Dosimetry.
      EPA,  Research Triangle Park, N.C., 1976.

  7.   Zisklnd, R. et al.  Final Report:  Carbon Monoxide Instrusion into the
      Passenger Area of Sustained-Use Motor Vehicles.  Report to Congress
      by  the Environmental Protection Agency.  Washington, D.C., July 31, 1978.

 8.  Wallace, L.  Use of Personal Monitor to Measure Commuter Exposure to
     Carbon Monoxide in Vehicle Passenger Compartments.  Paper submitted to
     1979 Annual Meeting of the Air Pollution Control Association.  June 1979.

 9.  Wallace, L.  Personal Monitors.  In:  Supplement to Vol. IV, Environ-
     mental Monitoring.  National Academy of Sciences.  Washington, D.C., 1977.

Lance Wallace, Ph.D.
Office of Monitoring and Technical Support
Office of Research and Development
U.S. Environmental Protection Agency
401 M St., S.W.
Washington, D.C.  20460
     NADER:  John Nader, EPA, Research Triangle Park.  I would gather concern-
ing these personal monitors that they will be switched on and off as one en-
ters from, say, environmental exposure outdoors to any kind of exposure in-
doors.  Is this correct?

     WALLACE:  In the particular case that I was talking about, the monitor
had enough range so that it was sensitive to both the concentrations that I
was meeting on the bus and in the office.  So,  I left it on in the office for
the first few days until I was determining that my exposure there was so low
that it added essentially nothing to my exposure in the community.  After that,
I used the time to charge up the battery.

     NADER:  But the intent of personal monitors for environmental exposure
would be to separate the exposure outside from the exposure inside;  is that

     WALLACE:  I think it would be valuable to be able to make that separation.
On the other hand, there may be cases in which it is far cheaper to collect
something over the full 24-hour day, say, and send it off to the laboratory
to look at.

     In that case, you would only get an integrated exposure,  and you won't
be able to differentiate.  But the cost benefit may make that worthwhile in
some cases.

     NADER:  Doesn't that raise a question on the validity of the data so far

 as CO  is  concerned?   For example,  you might get considerable CO exposure in-
 doors  under smoking  conditions,  and this would affect your data?

      It would give you the true  fact of exposure.   But as to attributing
 that exposure to outdoor levels, of course, that would be flexible.

      WALLACE:  Presumably one could get around that if, in the case  you men-
 tioned, we  had a cheaper way of  getting the total integrated exposure.
 For  statistical purposes, we could equip a large number of people with  moni-
 tors with the analysis done the  cheaper way, and then equip a smaller statisti-
 cal  sample  for continuous exposure such that we could differentiate  the ex-
 posures due to various components of the data.

      MAGE:   Dr. West?

      WEST:   Phil West of Louisiana State University.  I think John has  already
 raised part of the question.  Was there much smoking in the bus when you meas-
 ured  these  levels that we are talking about so far?

      WALLACE:  In this case there was no smoking.

      JOSEPHSON:  Julian Josephson, Associate Editor, Environmental Science and
 Technology Magazine (ES&T).  Do  you believe that the source of the CO was the
 gasoline-powered air conditioning rather than the diesel engine of the  bus,
 or is  this not really determined?

      WALLACE:  It is not determined.  But I have talked to people at the De-
 partment  of Transportation, and  they tell me that diesels do not put out much
 CO and that gasoline engines do, and that their guess was that it was the
 air conditioning.  They also point out that these were older buses and  that
 most newer buses operate the air conditioner off the same power plant—the
 diesel power plant.   So, this particular problem may be seen less often in
 newer  buses.

      JOSEPHSON:  You are talking about the famous Reston commuter bus—right?

      WALLACE:  That is right. They have about 70 buses.  They are all  diesels,
 circa  1955, and they have over 700,000 miles on them.

      DOCKERY:  Doug  Dockery, Harvard School of Public Health.

      What are the policy implications if we go out and do personal monitoring
 and  find  that personal exposures are not related at all to what the  outside
 environment is actually telling  us they are, and then we relate that back
 to health effects?  What is EPA  going to do about setting standards  and chang-
 ing  their policy?

      WALLACE:  That  is a very serious policy question about the Indoor  aspect
 of exposures.  Many people feel  that it has fallen through the crack with
 EPA doing outdoors,  OSHA doing the workplace, and indoors and inside vehicles
 being  not quite covered by any group.

The Need for Personal Air Pollution
Exposure  Monitors  and  Applications in
Energy Development
     Leonard D. Hamilton, M.D., Ph.D.
     Biomedical and Environmental Assessment Division
     Brookhaven National Laboratory
     Upton, New York
     There has been much research in  the past few years devoted to air pollu-
tant transport and chemistry.  These  efforts provide a. much better basis for
relating ambient  pollution levels to  sources.  Knowledge of the relation
between ambient air pollution levels  and individual or population exposure,
however, is still very limited.  People move around, spend most of their time
indoors, and are  exposed to pollutants from smoking, faulty auto exhaust
systems, gas ranges, the workplace, and other sources.  Even in air pollution
epidemiology, where medical surveillance may be extensive, exposures are too
often estimated from fixed monitoring stations serving neighborhoods or whole

     In investigating health-damage functions for use in comparing the health
implications of various energy alternatives, our group concluded that the
weakest link in determining such damage from air pollution is the accuracy
of quantitative estimation of individual or population exposures to pollutants.

     In 1975 we convened a workshop of health and instrumentation specialists
to consider the role of personal pollution exposure monitors in air pollution
health effects studies, and to assess research needs in this field.

     Two early conclusions of the workshop were that "the importance of popu-
lation exposure estimates in air pollution epidemiology makes it imperative
that future epidemiological studies include exposure estimates more repre-
sentative of what people actually breathe," and that "the use of individual
air pollution monitors is a necessary factor in the design or performance of
definitive studies of the health effects of air pollution" (2,3).


      The Brookhaven workshop produced a first-order ranking, according to
 promise of ultimate performance, of a number of the candidate instrumentation
 technologies.   These rankings, shown in Table 1, reflect a loose consensus
 of  the  participants at the time of the workshop.

      For each technology, the workshop developed a first-order estimate of
 the research needs.

      It seemed advisable to support development work on several of the more
 promising technologies in parallel before concentrating on a smaller number
 for further development.

      Since the workshop, the Department of Energy has funded several projects
 aimed at developing personal exposure monitors.

      Recently, Brookhaven was asked to organize and chair a working group to
 assess industrial hygiene needs for coal conversion and oil-shale industries,
 with an eye to the possible expansion of an instrumentation program already
 underway.  A preliminary report of this working group has been issued (5).
 This has given us an opportunity to review the rather considerable advances
 in instrumentation made over the last 3 years.

      Coal conversion plants will emit conventional combustion products (parti-
 culates, SO., NO ) from the boilers needed to supply process heat to the
            fc    *V
 conversion units.  On an energy-produced basis, these pollutants are emitted
 at much lower levels than from coal-electric power plants.  Conversion pro-
 cesses, however, involve the production of much more hazardous materials.
 These include heterocyclic compounds and reduced sulfur compounds.  Facili-
 ties are designed to prevent the escape of these materials, but exposure of
 workers (particularly maintenance workers) must be considered.  Since full-
 scale plants have not yet been built, the potential extent of public exposure-
 cither routinely or in accident situations—cannot be well defined.  Some
 surveillance among the public, at least during the start-up and early opera-
 tion of such plants, seems warranted.  Gammage et al. have pointed out the
 gap between the sophisticated analytic tools used in the laboratory to char-
 acterize pollutants from synfuels operations and the crude techniques avail-
 able for measuring polycyclic materials in the field (1).

      Among the range of heterocyclic compounds, the question of what specifi-
 cally to monitor is especially important for field monitoring of population
 exposure.  The need to develop one or more indices is important.  Use of in-
 dices was strongly recommended by an interagency-supported investigation on
 health and safety guidelines for coal gasification (4).


                                                 TABLE  1

           Tentative Ranking Developed by Participants in the Brookhaven Workshop on Individual
           Air Pollution Monitor* of Candidate Gas- and Particle-Sensing Instrumentation Techniques

           (Performance criteria applied in the ranking were the development of a prototype device
           of lunch box size or less with annual support of $200,000/year or less. The rankings are not
           based on a full quantitative analysis and should  be viewed only as a qualitative guideline.)
Probability of Payoff
            Within <3 yr
Within 3 to 6 yr
    Fair to poor
                   GAS METHODS*
Gel tape colorimetry
Electrochemical cells
Coulometric and amperometric devices
Fluorescence devices (SG^)  I
Nondispersive spectroscopy (CO)
Chemiluminescence (Os, NO,)

                  PARTICLE METHODS6
Virtual impactor (M, 2D, C)
Filters (M, —.V.—.D.C)
Diffusion batteries with filters (C(D))
ft meter (Af)
                    GAS METHODS
Microwave devices (CO)
                                            PARTICLE METHODS
                           Diffusion battery with
                             condensation nuclei counter (M, D)
                    GAS METHODS
Gas chromatography
Cryogenic sampling
Electronic solid state devices
Nondispersive spectroscopy
   (gases other than CO)
Piezoelectric gas sensors
Microwave devices
   (gases other than CO)

                  PARTICLE METHODS0
Piezoelectric mass balance (M)
Optical scattering (~JV, D)
                                                                        Sorption methods
                                                                        Colorimetry (tape based)
                                                                        Electrochemical cells
                                                                        Coulometric and amperometric devices
                                                                        Fluorescence devices (SO2)
                                                                        Nondispersive spectroscopy (CO)
                                                                        Chemiluminescence (Os, NO*)
                                                                        Virtual impactor (M, 2D,C)
                                                                        Filters (M, -JVT, ~D, C)
                                                                        Diffusion batteries with filters (C(£»))
                                                                        Optical scattering (A', D)
                                                                        ft meter (M)
                                                                        Electronic solid state devices
                                                                        Nondispersive spectroscopy
                                                                          (gases other than CO)
                                                                        Piezoelectric gas sensors'
                                                                        Microwave devices (CO)
                                                                        Gas chromatography
                                             Virtual impactor (Mt >2D,C)
                                             Diffusion battery with condensation
                                               nuclei counter (JV, D)
                                             Piezoelectric mass balance
                                                (M, ~JV, ~fl)
                                             Optical scattering (~N, D)
                                                                        Cryogenic sampling
                                                                        Bioluminescent methods
                                                                        Optoacoustic methods
                                                                        Microwave devices
                                                                          (gases other than CO)
   •When the classification depends on the gas of interest, this has been indicated in parentheses. Techniques which
 were discussed but which me group did not feel qualified to evaluate included fluid-immersed solid state sensors and
 gas-sensitive liquid crystals.
   bAf indicates device appropriate for measurement of total suspended paniculate mass; JV, for measurement of total
 number of particles; and D, for measurement of particle site distribution. ID indicates that sizing into two categories
 such as >3 urn and  <3 /im is possible. C indicates device appropriate for collecting samples of particles for chemical
 analysis, and C(D), that chemical composition can be estimated for two or more separate ranges of particle size.
   ^Other candidate technologies considered but viewed as unsatisfactory or inappropriate for use in individual air pol-
 lution monitors included liquid impingers, centrifugal  separation, conventional impactors,  thermal precipitation,
 electrical mobility, and microwaves.

     A single index may not be appropriate for all situations.  Gasification
and liquefaction have different characteristics, and even different exposure
situations from the same plant may be best evaluated with different indices,
due to variations in the mix of compounds in  the pollutant stream.  Identi-
fication of indicator compounds needs to consider both the chemical charac-
teristics of the mix of compounds and their biological activities.  A problem,
especially difficult in personal samplers, is that PNA's are likely to be in
aerosol form or in vapor form subject to condensation in the sampling train.

     From our perspective, other recommendations for future efforts include:

     1)  Personal air pollution exposure monitors are an essential step in
quantifying individual and population exposure for protection, epidemiology,
and regulation.

     2)  Passive samplers offer a cheap and convenient tool.  They should be
exploited much more extensively in population studies.  New substrates should
be developed to expand the diversity of measurable compounds.

     3)  Passive monitors are inadequate to fulfill the full promise and need
from personal exposure monitoring.   Development and demonstration of active
devices which can sample and record  short-time exposures are important.

     4)  Samplers must be field-tested under  various conditions of inter-
ferences, temperature, pressure, humidity, etc.

     5)  Personal monitoring of particulate exposure is an area of great
need, but it does not seem ripe for  exploitation.  Areas where continued
support seem warranted are:  a) improving efficiency of collection; b) improv-
ing extraction efficiency and separation techniques; c) chemical specification;
d) in situ separation of soluble and insoluble particulates.

     Regulatory  decisions  on air  pollution  control—which involve direct and
indirect costs of  billions of dollars—are  being made without adequate know-
ledge of the  health  impacts of air  pollution.   The weakest link in the studies
of health effects  is our knowledge  of  individual exposures.  The importance
of population exposure  estimates  makes it imperative that future studies in-
clude exposure estimates more representative of what people actually breathe.

     Personal air pollution exposure monitors that are easily carried or
worn are  essential for the design or performance of all definitive studies
of  the health effects of air pollution.  Such monitors also have prevention
and regulatory uses in public and occupational situations.

     Given the enormous costs of air pollution control and regulation and the
considerable uncertainties in our present understanding of the health Impacts
of  air pollution, a coordinated national program of support for the develop-
ment and  use of personal air pollution exposure monitoring instruments should
be  initiated without delay.

 1.  Gammage, R.B., Vo-Dinh, T., Hawthorne, A.R., Thorngate, J.H., Parkinson,
     W.W.  A New Generation of Monitors for Polynuclear Aromatic Hydrocarbons
     from Synthetic Fuel Production.  In:   Jones, P.W., Frudenthal, R.I.,
     Carcinogenesis, Vol. Ill, Polynuclear Aromatic Hydrocarbons.  New York,
     Raven Press, 1978, pp. 155-173.

 2.  Morgan, M.G., Morris, S.C.  Individual Air Pollution Monitors.  An
     Assessment of National Research Needs, BNL 50482.  January 1976.

 3.  Morgan, M.G., Morris, S.C.  Individual Air Pollution Monitors, 2. Examin-
     ation of Some Non-Occupational Research and Regulatory Needs, BNL 50637.
     January 1977.

 4.  U.S. Department of Health, Education, and Welfare and EPA.  Recommended
     Health and Safety Guidelines for Coal Gasification Pilot Plants, DHEW
     NIOSH Pub. 78-120/EPA-600/7-78-007.  January 1978.

 5.  White, 0.  Working Group on Assessing Industrial Hygiene Monitoring
     Needs for the Coal Conversion and Oil Shale Industries, BNL 24925.
     August 1978.

Leonard D. Hamilton, M.D., Ph.D.
Biomedical and Environmental Assessment Division
National Center for Analysis of Energy Systems
Brookhaven National Laboratory
Upton, New York  11973

     JOSEPHSON: Josephson, ES&T Magazine.  Do you  foresee that the present
effort concerning the anti-inflation  fight might have an adverse effect on
the development of this  instrumentation  and coordinating activity that you

     HAMILTON:  No;  I frankly  don't  think so, because from what my job is and
from the point of view of  inflation,  one is thinking In terms of controls,
as they apply  to plants  10 or  15  years from now.

     I am not  quite  sure what  those  capital requirements for those plants are—
whether the capital  has  to be  on  line for many  years In advance or not.

     But it seems to me  that what we  are talking about is a few millions of
dollars.  It is so insignificant  when I  compare it  to the gross national
product of the United States that I  can  say that it would have actually no
relationship whatsoever*

     The only  thing  is this:   If  we  did  have a  better understanding of what
the dose effect relations  were, we would feel more  confident as to whether
or not we were justified In  spending  these large sums on control.  Let me
Just say this:  we are going to spend these large  sums on control anyway.
We are committed to  this even  In  the  present state  of uncertainty.  For ex-
ample, the health effects  of something such as  sulfate are uncertain.  Al-
though we can't say  with 95  percent  certainty that  sulfate is causing an in-
creased mortality, certainly we can  say  this with 60 percent certainty,
and under these circumstances, we have to go ahead  with these expensive con-
trols.  What I am saying is  that  from the point of  view of public policy,
this tiny Investment In  air  pollution and personal  monitors is kind of a
dreary thing for many people.  I  don't think they  like to support the idea
of developing  measuring  Instruments.  I  think this  is one of the reasons why
it possibly hasn't been  supported.

     ZISKIND:  Richard Zlsklnd from  Science Applications.  You mentioned the
problem for the need to  develop indices  If one  wants to deal with the events
called commercial processes.   What are the limitations right now as you see
them in developing such  Indices?   Are there monitoring problems or Identifi-
cation of levels or  substance  identification that  should be of concern?

     HAMILTON:  I frankly  don't think there are any problems.  A more likely
problem is apt to be in  having to develop a good personal monitor.  I am sure
we can raffle  off a  number of  possible indices. I  have already mentioned one
of them.

     What we need la a monitor for a  simple heterocycllc compound characteris-

tic of what might come out of the process.  I feel that the problem  lies
In the development of the actual good personal Instrument rather  than  In  the
choice of the Indices.

     But again, that Is the sort of subject I feel one could develop In the
process of discussion.

     CHASE:   Paul Chase of Beckman Instruments.   You mentioned the DOE Is
sponsoring the development of some Instruments.   Is that only your shop, or
did I misunderstand what you said?  How about a  portable gas chromatograph?

     HAMILTON:  The pocket-sized gas chromatograph?  I am not sure who Is
actually doing that*   However,  It Is almost at the commercial stage.   But
I am not sure who that would be.

     I think that Beckman would be very interested In commercializing It.

Current  NIOSH Research on Passive
     Mary Lynn Woebkenberg
     Measurements Research Branch
     Division of Physical Sciences and Engineering
     National Institute for Occupational Safety and Health
     Cincinnati, Ohio
      I would like to take this opportunity to describe to  you  the  work  that
 is currently being undertaken by the National Institute for  Occupational
 Safety and Health (NIOSH) and its contractors in the area  of passive monitors.
 We are studying not only points of current interest, but also  future applica-
 tions of  passive personal monitors.

      For  lack of any other, I would like to offer the following  definition
 of passive personal monitors. (Please keep in mind that this definition is
 dependent upon the interests and responsibilities of NIOSH in  the  area  of
 personal  exposure monitoring.)  A passive personal monitor is  a  device  worn
 on an individual for the purpose of measuring—without the use of  an active
 flow mover—personal exposure.  The key aspects of the definition  are three-
 fold: 1) the device is worn on the individual—preferably within  the breath-
 ing zone; 2) the device uses no active flow mover, such as the typical  pump
 that is necessary with the various solid sorbent tubes; and  3) the device
 measures  personal exposure.  These characteristics are the opposite of  those
 of an area monitor.

      There are several advantages to passive monitors.  This type  of monitor—
 interchangeably referred to as a badge—is small, lightweight, and easily worn
 by any individual.  This is an advantage over personal sampling  pumps,  which
 can weigh up to 2 pounds each.  The monitors are free of liquid  content, which
 makes them preferable to impingers and bubblers.  The subsequent chemical
 analysis  of passive monitors is generally similar to existing  methods.  The
 badges, however, are not without disadvantages.  Whether the badge uses the

principle  of  diffusion or permeation through a membrane,  there is an extra
factor  (not necessary with the use of sorbent tubes and pumps),  which is
either  the diffusion coefficient of the species through air,  or  the  permea-
tion  constant of  the species through the membrane material used  in the badge.
This  constant must  be known for each compound.  In addition,  there is evidence—
even  with  sorbent tubes—that high humidity alters the adsorption of various
solvents.  In the case of sorbent tubes, desiccant tubes may  sometimes be
used  preceeding the sorbent tube.  It is not yet apparent what can be done
to  correct for high humidity effects in passive monitors.

      The emphasis of NIOSH's current research with passive monitors  centers
on  the  passive organic monitors.  Most of the studies involve the Gasbadge,
marketed by Abcor Development Corporation, and the Organic Vapor Monitor,
marketed by the Minnesota Mining and Manufacturing Company.  One study also
includes the  MiniMonltor, marketed by the Reiszner Environmental and Analytical
Labs, Inc.

      The Abcor Gasbadge is 8 cm long, 5.7 cm wide, and 2 cm deep.  It weighs
approximately 40 g and consists of seven parts:  the sliding  cover;  the front
plate of  the  badge, which has a 4.4 cm x 3 cm opening to allow diffusion  of
gases;  a protective screen; a draft shield; an open grid that defines the
diffusion  geometry; the collection element; and the back plate of the badge.
One unique feature of the Gasbadge is that the newer-designed case allows for
placement  of  a second collection element which can be used to take a second
sample or  can be used as a field blank.  The Gasbadge is reusable by replacing
the collection element.

      The 3M Organic Vapor Monitor is a circular badge that is 10.2 cm long
(including the clip), 4.4 cm wide at its widest point, and 1.2 cm deep.  The
sampling opening is circular with a 3-cm diameter.  This badge weighs 13.5 g.
During sampling,  the unit consists of six pieces:  the outer  rim; the draft
shield, which is held in place by the outer rim; an open grid that defines
the diffusion geometry; the collection element; and the solid back piece
of  the monitor.  The sixth piece is the clip for attachment to the person.
The 3M monitor is the only passive monitor that we are currently using that
allows  for in situ sample elution.  The 3M badge is not reusable.

      The MiniMonitor, which was developed by Philip West at Louisiana State
University, is also a circular badge.  Its diameter is 5.0 cm, it is 0.625 cm
thick,  and it weighs 35 g.  There are two features unique to  the MiniMonitor:
the badge  works on 1) the principle of permeation of contaminant gases and
vapors through a membrane; and 2) adsorption of the pollutant(s) onto


approximately 1.35 g of PCB activated charcoal.  This is free-flowing, as
opposed to the collection elements of the two previously mentioned badges.  The
MiniMonitor case is reusable by introducing a fresh supply of charcoal.

     The most basic ongoing research with the passive monitors is an evalua-
tion of the passive monitors that is being conducted at the Utah Biomedical
Test Laboratories under contract to NIOSH.  The purpose of this study is to
establish the reliability and the working characteristics of the Gasbadge, the
Organic Vapor Monitor, and the MiniMonitor.  This effort consists of eight
tests that are designed to study various properties of the badges.  All of the
tests are being conducted in a specially designed and constructed exposure
chamber with accurate control and monitoring of temperature, humidity, concen-
tration of the contaminant being used in the study, face velocity, and exposure
time.  Even though independent checks are being constantly run on the concen-
tration—i.e., via GC, PID, or other calibrated instruments—in all of the
tests in this study, charcoal tubes are being used in side-by-side sampling
with the badges.  The charcoal tube sampling is used as a reference because
the method represents the state-of-the-art in sampling for organic vapors and
is the best "primary standard" available.

     In the first test with the badges and the tubes, we will consider the
precision of the measurement with the badges, and the comparability of the
badges with the charcoal tubes.  Under conditions of constant temperature,
humidity, face velocity, and time, the badges and charcoal tubes will be
simultaneously exposed in 12 tests.  The 12 tests are comprised of four con-
taminants at three concentration levels each.  The contaminants to be used in
this study are vinyl chloride, acetone, toluene, and a mixture of n-hexane
and acetone.

     The effects of storage on the sample accuracy will be studied.  In this
test, one set of monitors and charcoal tubes will be exposed to methylene
chloride.  A matching set of monitors and charcoal tubes will be exposed to
toluene.  Monitors and tubes from each exposure will be divided into three
groups to be analyzed at three times—within 1 day of exposure, on the 7th
day following exposure, and on the 14th day following exposure.

     Perhaps the most difficult test is that which will consider maximum and
minimum levels of quantitation and the capacity of the monitors.  These tests
have been designed to determine over what concentration range and exposure
times the sampling and analytical methods for the monitors will provide a pre-
cision and accuracy of + 25 percent of the "true" value for 95 percent of the

      A fourth study will look at the effects of face velocity on the badges.
 The face velocities will be varied from essentially stagnant conditions to
 400 fpm.  Orientation of the badges will also be studied, with the contami-
 nant approaching from the front, side, and back of the badges in three sepa-
 rate tests.

      Effects of temperature from 0° to 40° C and humidity from 10 to 80 per-
 cent relative humidity will be studied in two different series of tests.

      Out-gasing or back-diffusion from the badges could be a serious short-
 coming of the monitors.  It is possible that back-diffusion could occur in a
 situation where a worker would be exposed to a very high concentration of some
 contaminant and then move to an area of no exposure.  Back-diffusion could
 occur, due to the concentration gradient, depending upon how tenaciously the
 contaminant is adsorbed to the collection element.  To study this phenomenon,
 three tests have been designed that look at nonconstant concentration exposure
 and continually varying concentrations.

      The final test in this contract involves exposing the monitors to a com-
 plex mixture, such as refined petroleum or coal tar products, and comparing
 the badge results to the charcoal tubes.

      The Division of Surveillance, Hazard Evaluation, and Field Studies of
 N10SH is undertaking an effort to accumulate comparative field sampling data
 between the Gasbadge, the Organic Vapor Monitor, and charcoal tubes.  This
 effort will result in a report later this year that contains 500 points of
 comparison data between passive monitors and charcoal tubes—all from field

      We are also engaged in research that involves possible future applica-
 tions of passive organic monitors.  Our Organic Methods Development Section
 is attempting to develop a passive monitor for aldehydes.  Their approach is
 based on a derivative formation.

      Another of the more complex problems with the passive monitors is the
 question of saturation.  How do we know when the collection elements have
 reached their capacities for adsorption of contaminants?  In the laboratory—
 under controllable conditions—it is easy to determine the point of satura-
 tion.  But what happens in the field if the exposure of a worker is being
 measured and the passive monitor saturates?  To address this question, we are
 undertaking an effort to look at two-stage passive monitors.  In this study,
 we determine the capacity of a collection element by plotting exposure time


versus percent recovery.  Saturation of the collection element is indicated
by the drop off in the curve from the theoretical 100 percent recovery, and
the capacity of the element can be calculated from this exposure time.

     What we hope to be able to detect is the breakthrough of the contaminant
through the first collection element and to determine the validity of the
sample.  When the first collection element is saturated, there will be a con-
centration gradient to the back of the badge in the direction of the second
collection element.  There will probably also be a less dramatic gradient
between the first collection element and the ambient air.  Because of these
developing gradients, we hope to be able to detect the contaminant on the
second element after the badge has been exposed up to its saturation point.
This will be a great aid in using the passive monitor in the field.

     Another type of passive monitor is a personal atmospheric gas sampler
with a critical orifice which was developed through an interagency agreement
between NIOSH and the Naval Research Laboratory.  This whole air sampler uses
the leak through a critical orifice to meter the air flow into an evacuated
container.  The critical orifice admits air at a constant rate up to one-half
atmospheric pressure.  An 8-hour sample can be taken.  For analysis, the sam-
pler is connected to a gas chromatograph, through a gas sampling loop inter-
face.  The pressure of the gas sample is measured using a manometer placed in
the gas sampling loop.  The advantages of the whole air sampler—besides being
small, lightweight, portable, and requiring no personal pump—are that the
sample taken never undergoes a phase change, the sample is never handled,
and there are never breakthrough and desorption problems.  One disadvantage
to the whole air sampler is that it does not provide an enrichment step, and,
therefore, sample analysis requires sensitive analytical instrumentation.
Substances with low-level standards may not be sampled in this fashion.

     Passive monitors are currently available for NO , SO,,, Hg, and the organic
           —*                                        «W    fc
solvents.  The U.S. Army Environmental Hygiene Agency has been working on a
passive monitor for ammonia.  Soon, independent companies will make available
passive monitors for NH,, S02, HC1, and NO , as well as marketing uniquely
designed passive monitors for organic solvent vapors.

     NIOSH is planning to look further into passive monitor use by testing
different solid sorbents as the collection element and perhaps looking into
some electrochemical applications of passive monitors.  As far as their
application to ambient personal monitoring goes, the present state-of-the-art
passive monitor has some major drawbacks:  1) the monitors generally lack
specificity (a drawback of some other sampling techniques as well); and


2) the detection limits of the monitors—at the low end of the scale—may not
meet the needs of ambient sampling.  While it Is true that collection on the
element is an enrichment step, it could take quite a long time to accumulate
a detectable sample from ambient air.  With certain new products that are
coming out, and as the technology allows ue to develop more effective monitors,
passive monitoring will be useful in the ambient situation.

Mary Lynn Woebkenberg
Measurements Systems Section
Measurements Research Branch
Division of Physical Sciences  and Engineering
National Institute for Occupational Safety and Health
4676 Columbia Parkway
Cincinnati, Ohio  45226
     WALLACE:   Lance  Wallace,  EPA.   Is  that  cold air sampler going into com-
mercial  development?

     WOEBKENBERG:   MDA markets It.

     WEST:   Phil West, Louisiana State  University.  I can make a comment,
which  is not a question.   One  is that the  permeation-type badge does handle
liquids. In fact,  they use that on a chlorine badge that we have just de-
veloped. In that  case, incidentally, it is  specific for chlorine if the
ozone  does  not Interfere.

     I might also  mention  that so far humidity has no effect on the badges
that we  have developed based on permeation.

     WOEBKENBERG:   Some of the studies  that  we have run have shown a humidity
effect.   The data  have not been worked  up  yet, but we have seen a minor humid-
ity effect.

     But that  is true about the liquids.   There are some passive monitors
that will be available.  In fact, there are  some that are already available
that do  use  liquids as the collection element.  Some of them will be talked
about  today.  And  some of  them are not  yet ready to be spoken of.  But there
are some liquids also available.

     THOMAS:  Charles Thomas.  You mentioned that some of  the monitors were
specific and some were not.  If they aren't specific, how  do you  get  around the
problem of determining what the lowest level of exposure should be  if the col-
lection efficiencies vary from gas to gas?

     WOEBKENBERG:  Well, I said that the monitors are not  specific.   Gener-
ally, I mean in the case of the organic vapor monitors.  The charcoal col-
lection elements will trap organic vapors, and in that case, they are not

     In order to resolve the question of specificity, then you have to  rely
on your analytical method to do that for you.  However, in the case of  some
of the passive monitors—as Dr. West mentioned, they have one for chlorine
and 3M has one for mercury—there are certain monitors that are specific.
But in the case of nonspecificity,  they have to rely on the analytical  method
to resolve that problem for you.

     THOMAS:  Isn't it a collection device and not a monitor?

     WOEBKENBERG:  That is true; yes.

Electrochemical  Methods for Development
of Personal Exposure Monitors
     Joseph R. Stetter, Ph.D., Donald R. Rutt, and Michael R. Graves
     Energetics Science Division
     Becton Dickinson and Company
     Elmsford, New York

     Recent social and legal requirements have given great  impetus for the
development of personal exposure monitors for a variety of  gaseous pollu-
tants.  Since the exposure of a normally active individual  is not accurately
represented by a fixed-site monitor, there is a need for personal monitors
in order to measure the actual human exposures.

     The application of our electrochemical sensing method  has resulted in a
variety of practical monitors for  the measurements of CO in ambient air.
This established technique has been applied to personal dosimetry for CO in
the workplace environment.  An evaluation of this device demonstrates that
the performance characteristics of the larger fixed-site instruments can be
achieved in a compact, lightweight personal exposure monitor.

     This instrument approach to dosimetry overcomes many of the historical
disadvantages of integrated dosage measurement.  The CO Dosimeter, Ecolyzer
9000 Series, enables accurate measurement of dosages over short-time intervals
and correlation of exposure and physical activity.

     The success of the electrochemical method for the measurement of carbon
monoxide in the workplace environment is well known.   In the  sensor, carbon

monoxide is electro-oxidized to carbon dioxide at a potentiostatically con-
trolled  platinum electrode in aqueous electrolyte and follows the reaction

                       CO + H20 •*•  C02 + 2 H+ + 2 e~

The  signal produced upon exposure of the sensor to an air mixture is propor-
tional to the partial pressure of carbon monoxide in the gas which is analyzed.

      This current flows between the sensing and counter electrode, and the
reaction at the counter electrode is

                         02 + 4 H+ + 4 e~ •*•  2 H20

therefore yielding an overall cell reaction of

                           2 CO + 02 •*-  2 C02

      This sensing method has been developed and adapted for personal monitor-
ing.  In order to accomplish this task, it was required that the sensor be
miniaturized and designed for ruggedness.  Further, accurate measurement of
the  TWA while worn by the worker performing routine activities necessitated
that the instrument performance be insensitive to both orientation and

      The gas is Introduced into the sensor (shown schematically in Figure 1)
by means of a sampling pump.  The sensor is a miniaturized three-electrode
electrochemical cell.  The Teflon-bonded diffusion electrodes are in con-
tact with the aqueous electrolyte, which Itself is suspended in an immov-
able matrix.  The matrix allows the sensor to be operated in any special

      The sensor is interfaced with an instrument system, which is described
schematically in Figure 2.  The compact system design has a volume of approxi-
mately 300 cc and a weight of 0.43 kg (12 oz).  The critical working elements
consist of pneumatics, sensor and electronics for power control, amplification,
and  storage.  During operation, an external pump draws a sample of ambient air
through the pneumatic system at a constant flow rate.  The sensor continuously
monitors the concentration of carbon monoxide in the sampled air stream.  This
concentration-related signal is amplified and converted into digital pulses
for  storage in a digital circuit.  The battery pack is capable of supplying
power in excess of 2 years of dosimeter operation.  A flow-adjust valve allows
for  fine control of the sample flow rate.  The prefilter removes particulates

        ^—FILTER AND
            MANIFOLD BLOCK
FIGURE 1.  Dosimeter schematic.



FIGURE 2.  Instrument system schematic.

 and Interfering gases, giving a high degree of specificity to the analysis.
 Zero and span controls allow for convenient calibration of the continuous CO
 monitor section of the dosimeter.

      A side connector is made so that it can be plugged into a readout device.
 The dosimeter readout is a self-contained rechargeable instrument which is
 used for dosimeter calibration and also for dosage measurement.  The data col-
 lected in the dosimeter is digitally transferred to the readout by depressing
 the read button.  This process does not clear the dosimeter, and the same dosi-
 meter can be read any number of times within any given time period.  Clearing
 of the stored dosimetric data is accomplished by depressing the "clear" button
 with simultaneous depression of the "read" button twice.  This virtually elimi-
 nates accidental clearing of the integrated CO reading stored in the dosimeter.

      Typical response of the dosimeter to a variety of CO concentrations is
 illustrated in Figure 3.  For this particular instrument, 98 percent of signal
 is obtained in less than 45 seconds.  For 100 instruments studied, we have found
 that 100 percent of signal is obtained between 20 and 60 seconds after initial
 gas exposure.  This means that calibration can be accomplished in approximately
 1 minute and that the dosimeter will accurately record fluctuations in the
 ambient concentration.

      For this instrument, the magnitude of the response can be expressed by
 the equation

                            ppm = 0.98x - 0.38

 where ppm and x represent the instrument signal and actual CO concentration,
 respectively.  This excellent linearity of response allows the instrument to
 be calibrated at a single point and still record dosage accumulations with
 accuracy over a wide range of exposure concentrations.

      One hundred instruments were studied by placing them in an environmental
 chamber and introducing an 84 ppm CO/air mixture Into each dosimeter at
 100 cc/min.  The gas mixture was drawn continuously through the dosimeter,
 and dosage readings were taken at hourly intervals over the course of a 6- to
 8-hour exposure.  During this exposure, the temperature was changed from
 40° C to 22° C to 0° C.  At each temperature the instrument span and zero
 were recorded.  Analysis of this data revealed the instrument performance
 characteristics at three temperatures and over a long time exposure.

      The data is summarized in Table 1 for 100 instruments.  The system error

             80 - _
             70 - -
             60 - -
             50 4-


             40 -

             30 J-    /
                                                                    RECORDER SCALE
                                                   SIGNAL  CO (PPM)   (PPM FULL .SCALE).










                      400 _





                0                1          2
                             TIME (MIN.)

FIGURE 3.  Dosimeter response versus time  at various CO  concentrations.

                                    TABLE 1
                         Ecolyzer Performance Summary
                                 100 Dosimeters

Std. Dev.
22° C
+ 0.46
+ 3.0
40° C
+ 0.96
+ 4.9
0° C
+ 2.5
+ 5.3
+ 2.10
+ 3.41
  Percent deviation from theoretical dosage
  For 25 instruments, random selection
 was calculated by comparing the actual dosage measured with the theoretical
 dose exposure.  To obtain a theoretical exposure, the instrument span at 22° C
 was multiplied by the total time of exposure.  Excellent stability is found
 for these 1-hour exposures at each temperature.  It is significant to note that
 the mean deviations of the dose from theoretical values are within+^ 2.5 per-
 cent, even when operated at a temperature 20° C different from the calibration
 temperature.  The standard deviation of this 100-unit sample indicates that
 the accumulated dosages are within the design goal of + 15 percent error with
 an extremely high reliability.  Errors not included in this analysis, however,
 are those due to the operation, such as calibration, and those due to equip-
 ment failure, such as the sampling pump and calibration gas.  However, it
 should be possible to keep such fluctuations well within +^ 5 percent by using
 appropriate precautions and following proper procedure.

      Examination of the individual variations of instrument stability with
 temperature is shown in Figure 4.   The error bars are two standard deviation
 units wide and demonstrate that the instrument's sensitivity to temperature
 changes is small.  Operation can occur outside the 0° to 40° C range and will
 result in only slightly higher errors than those measured here.  The sensor
 can be operated at temperatures as low as -10° F before failure is observed.

      In order to demonstrate the instrument analog/digital characteristics
 under various concentration/temperature conditions, we performed the experi-
 ment summarized in Table 2 and Figure 5.  Here, five dosimeters were simul-
 taneously exposed to:  1) 85 ppm CO at 24° C, 2) 50 ppm CO at 24° C, 3) a
 temperature change to 40° C with 50 ppm CO exposure, 4) continued 50 ppm


























         "<	I	1	-f9-
   - J.^
 10       15

35      4(4.
FIGURE  4.  Dosimeter zero  and span versus temperature.

                                                          TABLE  2

85 84 -1.2
142 140 -1.4
196 190 -3.2
246 236 -4.1
335 320 -4.5
86 85 -1.2
142 140 -1.4
197 190 -3.6
247 235 -4.9
336 318 -5.4
87 88 -1.2
143 140 -2.1
197 189 -4.1
247 234 -5.3
338 315 -6.8
89 86 -3.3
144 137 -4.9
198 186 -6.1
248 231 -6.9
339 314 -7.4
90 90 0
145 144 -0.7
199 196 -1.5
249 245 -1.6
341 333 -2.5

       FIGURE  5.   Dosimeter  analog  output  versus  time.

 exposure,  5)  85 ppm CO at 40°  C,  and 6)  zero ppm CO at 40°  C.   The  recorder
 trace shows the analog output  for each instrument for a 1-minute duration
 once each  5 minutes.  In Table 2, a comparison of theoretical  and experimental
 dosages (digital outputs) is summarized;  the dosimeters had a  mean  error of
 -3.4 percent +2.1 percent.  One observes that the instrument is capable of
 following  dosages accurately with a single calibration at the  initiation of the
 test and in the presence of 20° C temperature changes and concentration changes.

      Other personnel monitors that are based upon this electrochemical sensing
 method are available for hydrazines (1,2), NO- (3),  and H.S (4).   These instru-
 ments have not been miniaturized to the same degree  as the previously described
 carbon monoxide dosimeter; however, they are portable analyzers weighing 5  to
 10 kg.  Further, other devices are possible with existing technology for
 gases such as NO and SO..

      The future holds great promise for development  of personal monitors of
 miniature size using Energetics Science electrochemical gas sensing methods.
 This technique offers the advantages of small size,  light weight,  low power,
 stability over long sampling periods, ease of operation, sensitivity, and
 ruggedness.  Primary areas which will require development to provide electro-
 chemical personal monitors are related to:  1) improved sensitivity and sta-
 bility to widen the application range without reducing accuracy;  2) development
 of auxiliary instrument systems related to pneumatics and electronics;  and  3)
 improved specificity for expanding the application to a wider range of con-
 taminated environmental samples.


 1.   Stetter, J.R., Blurton, K.F., Valentine, A.M.,  Tellefsen,  K.A.  The
      Electrochemical Oxidation of Hydrazine and Methylhydrazine on Gold:
      Application to Gas Monitoring.  J Electrochem Soc 11:1804-1807, 1978.

 2.   Stetter, J.R., Tellefsen, K.A., Saunders, R.A., DeCorpo, J.J.  Electro-
      chemical Determination of Hydrazine, Methyl- and 1,1-dimethylhydrazine
      in Air.  Talanta (in press).

 3.   Ecolyzer Model 6632 Analyzer, Energetics Science Division, Becton
      Dickinson and Company, 85 Executive Boulevard,  Elmsford, N.Y.  10523.

 4.   Ecolyzer Model 6430 Analyzer, Energetics Science Division, Becton
      Dickinson and Company, 85 Executive Boulevard,  Elmsford, N.Y.  10523.


Joseph R. Stetter, Ph.D.
Energetics Science Division
Becton Dickinson and Company
85 Executive Boulevard
Elmsford, New York  10523
     ZISKIND:  Richard Ziskind of Science Applications.  I wonder if you are
concerned about electromagnetic interference with these devices, such as
radio transmission—something like that?

     STETTER:  Yes.  The circuitry is susceptible, and there are a lot of
wires and things.  We don't find high-frequency interference on most of our
monitors because they have metal cases.

     On these particular small ones, they have light plastic cases.  And we
found it necessary to coat the inside of the plastic case with a metallic
coating which is grounded.  We had one of these successfully tested at the
Kennedy Space Center.  And it shows indeed that the HF sensitivity is very

     SPENGLER:  John Spengler, Harvard School of Public Health.  A couple of
questions:  one is about cell life, and two, have you ever actually field
tested this alongside of an ND1R CO monitor?  Because 1 am concerned about
the highly fluctuating nature of real CO exposure.

     STETTER:  Early in the game, we compared the electrochemical technique
and NDIR for CO measurement.  We found that by and large our response times
were much better because they have a lower purge rate for the larger cell.
You can purge this cell very rapidly so you get a very good response charac-
teristic.  In terms of cell life, we warranted these kinds of cells for 6
months.  And generally speaking, with proper care and normal ambient con-
ditions where the relative humidity is, you know, 40 to 80 percent, you should
get a lifetime in excess of a year.

     SPENGLER:  What is the cost of the cell and the unit?

     STETTER:  The instrumentation comes in a kit of five which gives you
the readout and five dosimeters for about $3,000, give or take a little bit.

     WILLIAMS:  Marsha Williams.  Do I understand that it was tested with
flow rates, temperatures, and humidity checks like that at ambient levels
which are more typical in the environment that EPA would like to monitor?

     STETTER:  There is a field  test now; at  least  these devices have been
and are actually being used in the  field.

     WILLIAMS:  I don't mean  field  testing.   I am talking about your sensiti-
vity test to see how sensitive they are  to  flow  rate.

     STETTER:  Right.  In the laboratory, we  of  course did the flow rate study,
which was easily accomplished.

     WILLIAMS:  Have you done any studies at  low level concentrations to see
whether or not there is a difference in  effect?

     STETTER:  Yes.  And there is no difference  in  effect because the re-
sponse is linear.  It  is a percent  of  span.   In  other words,  if you have
50 ppm and you drop 1  ppm or  1 percent of span where you measure 100 ppm
at the same  flow rate, you would get 98.  That would be 2 percent of the
signal.  It  is a percent span change with flow rate—that is  the character-

     Now, we have done relative  humidity at all  three temperatures, at 0 to
100 percent.  The only effect you get  is dilution of your calibration gas.
It is the only effect  you can see,  and there  is  no  signal to  monitor at all.

     HODGESON:  Jlmmie Hodgeson, National Bureau of Standards.  Can you give
us any more  details—is there an active  relation between your electrical
signal and electrical  return  in  the amount  of CO going through the sensor?

     STETTER:  Yes.  There is a  signal relationship.  For example, if you
went to higher flow rates, you can  get higher currents—put more CO through
it.  So, in  terms of more detail:   yes,  in  effect.

     SPENGLER:  You didn't show  any data on interferences.  I know there are
several with the Ecolyzer such as NO.  And  it is the same principle, I assume?

     STETTER:  Well, there is not an interference due to NO with a proper
filter.  The filter has scrubbers and  adsorbents in it which  take out NO,
N0_, S02, H2S, and other electrochemically  reactive gases.  So, you get a high
degree of specificity. Without  the filter  in place, you can  get signals
from other electrochemical reactive gases.

     SPENGLER:  Would  that be a  concern  with  personal monitoring as opposed
to stationary monitoring?

     STETTER:  Yes.

 Portable Ozone  and  Oxidant Monitors  for
 Random Field Surveys
     Gary P. Heitman and Richard L. Sederquist
     Mast Development Company
     Davenport, Iowa

     Small,  portable electrochemical instruments capable of measuring ozone
(O^) at the  parts-per-billion (ppb)  level  are widely used in industry as
safety monitors  near ozone generators and  ozone-producing processes,  and in
quality control  of ozone-producing appliances such as photocopy machines and
electronic air cleaners.  Similar instruments, or measuring techniques,  may
be suitable  for  use in ambient air monitoring at remote locations, complement-
ing the existing fixed-station network.

     While the  once-popular colorimetric  potassium iodide (KI) analyzers
are no longer manufactured, amperometric  KI  instruments are currently avail-
able in several convenient configurations.   Similar in design to the well-
known Mast Ozone Meter (Model 724-2),  which  was widely used for ambient air
pollution studies  prior to 1975, Mast  Series 724 Oxidant Monitors utilize
an electrochemical sensor with buffered 2 percent KI reagent.  Details on  the
detection technique have been given by Mast  and Saunders (4).  Recent instru-
ments, while retaining the original detection  technique, offer several design
improvements, including a signal amplifier which allows multiple-range direct
readout, increased sensitivity at low  Og  levels, and zero/span adjustments.
Reagent is supplied from a single reservoir  with enough capacity for up to
30 days continuous operation.  Two models are  appropriate for ambient level

 03 monitoring—Model 724-5, a.c. powered (see Figure 1),  and Model 724-4,
 battery powered (see Figure 2).  The battery-powered model was designed
 according to National Institute for Occupational Safety and Health (NIOSH)
 guidelines for construction and performance standards for compliance moni-
 toring in the industrial environment (6) and will operate 20+ hours on a
 fully charged battery.  Both instruments weigh less than 12 pounds and are
 easily transported.

      Although relatively low in cost and simple to maintain and operate,
 certain electrochemical monitors can exhibit exceptional performance.

      A Mast amperometric KI oxidant monitor was evaluated against analyzers
 using chemiluminescent and ultraviolet absorption measurement methods  in  a
 field study at Los Angeles and St. Louis during 1970-1972 (3).  Long-term
 calibration stability and general operational characteristics were compared.
 Results are shown in Tables 1 and 2.

      More recently, a Model 724-5 monitor was evaluated by a project team at
 Harvey Mudd College, Claremont, Calif., for potential use as a commercial
 aviation 0_ monitor.  The instrument was calibrated prior to shipment  using
 the 1 percent neutral buffered potassium iodide (NBKI) procedure.  After  30
 days of continuous use at the college, the instrument's calibration was
 checked at the State of California Air Resources Board station at El Monte.
 The Air Resources Board (ARB) uses a Dasibl UV absorption monitor and  a
 special three-meter ultraviolet instrument for comparison in calibration  (11).
 The Mast monitor, Serial Number 2370, agreed with the ARB standards within
 six-tenths of 1 percent (0.6 percent) (11).

      Parker (6) reported similar calibration stability in a study for  NIOSH
 and also determined that neither changes in relative humidity (0 to 100 per-
 cent) or changes in ambient temperature (10° to 40° C) significantly affect
 the measuring efficiency of the Mast monitor.  Other operational character-
 istics which make the Mast monitor suitable for field use are convenient  size,
 quiet operation, low maintenance, and long-term unattended operation.

      Obviously, there must be some disadvantages in the use of an ampero-
 metric KI system.  Photochemical oxidants have been traditionally defined


 FIGURE 1.   Model 724-5 Oxidant Monitor,  front door open.  Oxidant measuring
 ranges are  0-0.1 to 0-1 ppm.
FIGURE 2.  Rear view of Model 724-4 Oxidant Monitor.   High-capacity  lead  acid
battery powers monitor for 20+ hours.

                                    TABLE 1

               Long-Term Calibration Summary  for 0, and 0  Instruments
                                                    J      X
Bendix Env. Sci.
Bendix Process
RTI (Solid Phase)
Mast b
Number of
zero drift
span drift
• Correlation
a.  Reference 3
b.  Mast QX, Model 724-2
                                    TABLE 2
           General Operational Characteristics  for 0_ and 0  Instruments
                                                      J      A

Bendix Env. Sci. 03
Bendix Process 03
RTI (Solid Phase) 03
Dasibi 03
Mast O^
Technicon 0^
Beckman 0^
Number of
days tested
tire (%)
, 94.2
time (%)
   a.  Reference 3
   b.  Mast Ox, Model 724-2

as substances which will oxidize a KI reagent, which responds to 03 with  nearly
100 percent efficiency, and to a lesser extent, to peroxyacetal nitrate (PAN)
and hydrogen peroxide—all components of the photochemical oxidant mixture—
and also to nitrogen dioxide (NO.).  Sulfur dioxide (S02) reacts negatively,
reducing an oxidant-caused signal.  The most important disadvantage of the
amperometric KI system, therefore, is its nonspecificity.  Sulfur dioxide is
a major interferent because it reduces iodine formed in the KI reaction from
contact with oxidant back to potassium iodide, producing a strong negative
interference.  A SO  filter (Model 725-30) is available for the Mast instru-
ment.  Consisting of glass fiber paper impregnated with chromium trioxide and
sulfuric acid, as described by Saltzman (7), it has proven to be effective in
over 7 years of field use and experimental evaluation (5,7).  A potential dis-
advantage in the use of this filter is its ability to oxidize nitric oxide (NO)
to N0_, causing additional interference in 0« determination.  This may not
present a problem, however, since 0« and NO rarely coexist.  Use of the filter
would be disadvantageous in only a few areas (such as Los Angeles), where
concentrations of NO are relatively high and those of S02 low (7).

     The Mast monitor will also respond to approximately 12 percent of PAN
present in a sample and to 9 percent of hydrogen peroxide (2).  Since these
compounds generally exist at low levels in ambient air, compared to ozone
levels (1), interference potential is usually insignificant.

     Assuming S0_ is filtered or is not present, the major interferent to
contend with is N0>.  The efficiency of the Mast monitor in measuring NO^
is approximately 6 percent.  The efficiency drops at high ozone levels.  For
instance, if an 0_ level of 0.25 ppm is being sampled concurrently with a
0.55-ppm level of N02, instrument response will be 0.28 ppm 0  (9), an error
of only 5 percent.

     In actual field use, an electrochemical KI oxidant method can show
equivalent performance when compared with a chemiluminescent and UV photo-
metric 0. method.  Wendt (10) reported that the California ARE compared,  at
three separate stations, a total oxidant analyzer (colorimetric KI),  an
ethylene chemiluminescent ozone monitor, and an ultraviolet photometric
ozone monitor.  The experimental results showed the total oxidant,  ultra-
violet photometric, and ethylene chemiluminescent monitoring methods to
be equivalent over a range of 0 to 500 ppb at the three sites.  Other recent
comparisons also show, in general, small differences between KI oxidant data
and ozone data, especially when instruments are standardized with the same
calibration atmosphere and when appropriate corrections for major interferents
are considered (1).


      A miniature KI oxidant sensing device currently used in a Mast Brewer
 ozonesonde may be adaptable for use as a personal 0, exposure monitor.  Figure
 3 shows the major components of a Mast bubbler/pump assembly, and the general

      Packaging, including electronics, readout, and housing, would produce a
 miniature monitor about the size of a cigar box weighing less than 1 pound.
 (See Figure 4.)

      Table 3 gives estimated performance specifications.  Other parameters
 have not been fully evaluated.  (See Figures 5 and 6.)

      The bubbler sensor is a small electrolytic cell which contains a few
 milliliters of KI reagent.  A miniature pump forces ambient air continuously
 through the cell at a constant rate of 200 cc/min.  The cathode is fine plati-
 num gauze, and the anode is a silver wire.  The arrangement of the cell is
 such that the bubbling action also provides the maximum possible flow of the
 solution through the cathode gauze.  When a potential of 0.42 V is applied
 across the cathode and anode, it balances a spontaneous emf associated with the
 detector, and no current will flow unless free iodine is present.  As the air
 containing 0  bubbles through the electrolyte, free iodine (I ) is formed
             X                                                £,
 which contacts the cathode screen, through diffusion and the sensor bubbling
                                 TABLE 3
        Estimated Performance Specifications of Mast Miniature Bubbler
 PERFORMANCE PARAMETER                          UNITS          OXIDANT

 Range                                           ppm           0-0.5
 Noise                                           ppm           0.005
 Lower detectable limit                          ppm           0.009
 Zero drift, 8 hour                              ppm           0.005
 Span drift, 8 hour                              ppm           0.005
 Rise time, 90 percent                         Minutes           5
 Fall time, 90 percent                         Minutes         2.5

   Tested on bromine permeation tube standard

              -5 cm
                FIGURE 3
                                9 cm
1 —

_ — ~.
~ —




-r I — IT






— - —




— .~

^— —
- •

- - -

~— 1 -

0 6


o : 5




.". :.




• - -

0 - 3
— '


!; 1

• • PP


_ .",.


3 — 1

- -- -
0 C
= I



3~- 1

— :-

— _^.-


0 --2

— ^_



- - .








— —

0 1 4


- - -


0 5



0^1 7



0 8






- -

0 -10C

: ;-


-- ;
                 FIGURE 4
FIGURE 3.  Miniature oxidant sensor.
FIGURE 4.  Experimental bubbler-monitor weighs less than 16 oz.
FIGURE 5.  Typical chart recording from oxidant bubbler.  Bubbler is ranged
for O-.l ppm 0  full scale.  Chart speed is 5 cm/hr.
FIGURE 6.  Chart shows Mast bubbler output over a 6-hour period.  Oxidant
source level is 0.084 ppm.  Chart speed is 2 cm/hr.

 action, and  is  reconverted  to  iodide  (21  ).   At  the  anode  the  I.  is  re-
 formed, but  it  is  prevented  from  recirculating  through the formation of
 practically  insoluble  silver iodide (Agl).   In  principle,  each ozone mole-
 cule entering the  sensor  gives rise to  a  flow of two electrons in the
 external  circuit of  the sensor.

      The  electrochemical  processes of the Mast bubbler are  such that  an
 insoluble layer  of  silver iodide is formed on the silver anode:

                              Ag + I  -e -*-  Agl

 As  the  silver  iodide  (Agl)  is relatively inert to the electrolyte,  the
 amount  of iodide plated on the  anode should be equivalent to the amount of
 oxidant which  has passed  through the cell.

      Due  to  the  small concentration of oxidant being sampled (less  than 1 ppm)
 and  to  the relatively low flow  rate of the pump (nominally  200  cc/mln), the
 amount  of iodide plated on the  silver in light hours is too small to  be
 determined gravimetrically.  However, Perone et al.  (8) have reported using
 cathode stripping voltammetry for the determination  of iodide using silver

      In the  determination of iodide, the iodide is first electro-deposited
 on  the  silver  anode in a  pre-electrolysis step.  The actual analysis  occurs
 when the  iodide  is  stripped from the electrode.  In  the bubbler, the  chem-
 istry is  much  the same as in the pre-electrolysis step.  The amount of iodide
 on  the  electrode can  then be determined using a cathodic stripping  method.
 Due  to  the stability  of the silver iodide, the stripping analysis may not
 need to be done  immediately, but may be done at a later time.

      The EPA  has  suggested  (1)  that,  ideally,  0_  should be measured  at  loca-
 tions downwind  from  the  source-intensive center-city area, if  peak 0- concen-
 trations are  to be recorded.   Increased rural  surveillance has also  been
 suggested.  Since concentrations of NO  in an  urban air mass decays  rapidly
 as the air moves  away  from  the city (1), nonspecific KI oxidant monitors
 may be suitable for  this task.   The cost-ratio for chemiluminescent  Og  or

UV photometric 0^ methods versus the amperometrlc KI (Mast)  method is nearly

     A miniature KI oxidant instrument,  currently used for atmospheric 0^
research, may be useful as a personal exposure monitor,  or,  when used with
a tethered radiosonde, a tool for tropospheric mixing and  transport studies.

     Amperometric KI oxidant monitors and  sensors are rugged,  small,  low in
cost, and exhibit excellent stability.   These  versatile  instruments may prove
useful in supplementing data collected at  fixed-monitoring stations.

 1.  Air Quality Criteria for Ozone and Other Photochemical Oxidants.
     U.S. Environmental Protection Agency.   EPA-600/8-78-004.  April 1978.

 2.  Bufalini, J.J.   Gas Phase Titration  of  Atmospheric Ozone.  Environ Sci
     Technol 2:703-704, 1968.

 3.  Decker, C.E.,  Royal,  T.M.,  Tommerdahl,  J.B.  Field Evaluation of New Air
     Pollution Monitoring Systems,  Final  Report.  Research Triangle Institute,
     EPA Contract No. CPA 70-101.   1972.

 4.  Mast, G.M.,  Saunders,  H.E.  Research and Development of the Instrumenta-
     tion of Ozone  Sensing.   ISA Trans 1:325-328, 1962.

 5.  Miller, D.F.,  Wilson,  W.E., Jr., Kling, R.G.  A Versatile Electrochemical
     Monitor for Air-Quality Measurements.   J Air Pollut Control Assoc
     21:414-417,  1971.

 6.  Parker, C.D.  Evaluation of Portable, Direct-Reading Ozone Meters.
     Research Triangle Institute,  NIOSH Contract No. HSM-99-73-1.

 7.  Saltzman, B.E.,  Wartburg, A.F., Jr.  Absorption Tube for Removal of
     Interfering Sulfur Dioxide  in Analysis  of Atmospheric Oxidant.  Anal
     Chem 37:770-782, 1965.

 8.  Shaln, I., Perone, S.P.   Determination  of Iodide Using Stripping Volt-
     ammetry at the Silver Electrode.  Anal  Chem 33:325, 1961.

 9.  Tokiwa, Y.,  Twiss, S.,  deVera, E.R., Mueller, P.K.  Atmospheric Ozone
     Determination  by Amperometry  and Colorimetry.  In:  Determination of Air
     Quality.  New York, Plenum  Press, 1972, pp. 109-130.

 10.   Wendt, J.F.,  Torre, K.J.  Ozone Measurements, Field Instrumentation and
      Comparative Data.  Unpublished paper presented at Conference on Air
      Quality  Meteorology and Atmospheric Ozone, ASTM Committee D-22.  Boulder,
      Colo., 1977.

 11.   Private  communication from Steven C. Runo, Harvey Mudd College, Clare-
      mont,  Calif., June 26, 1978.

 Gary P. Heitman and Richard L.  Sederquist
 Mast Development Company
 2212 E. 12th Street
 Davenport,  Iowa  52803

Personal Sampler for Measurement of
Ambient Levels of NO2
     Edward D. Palmes, Ph.D.
     Institute of Environmental Medicine
     New York University Medical Center
     New York, New York

     The NO- sampler to be described  is  a  "passive" device.  The only driving
force for the collection of molecules of the contaminant gas is the random
motion of the molecules themselves.   There are, therefore, no requirements
for pumps,  flow regulators, or batteries.  Because of this, the devices  can
be made very lightweight, inexpensive, and maintenance-free.  These charac-
teristics make them ideally suitable  for the purpose for which they were de-
signed; i.e., the measurement of N02  in  underground mines.  Prototypes of this
kind of sampler were reported by Palmes  and Gunnison (4) in 1972 and published
(5) in 1973.  The subject device was  published by Palmes et al. (6) in 1976,
a publication which describes the N02 sampler both from the practical and
theoretical standpoints (the reader is advised to consult this publication
for details of construction, operation,  and application to the workplace
environment). An attempt will be made in  this paper to describe the general
features of the device and the theoretical background, as well as to demon-
strate the applicability for monitoring  NO. at the community and the indus-
trial levels.

     The sampler consists of a 3/8-inch  I.D. acrylic tube, approximately
3 inches long.  One end of the tube is permanently closed with a close-fitting
polyethylene cap, inside of which are three  stainless steel screens coated

 with triethanolamine (TEA), which is an excellent absorbent for NCL,  as
 reported by Levaggi et al. (2) and by Blacker (1).  During sampling,  the
 other end of the tube is open to the environment to be sampled.  Since the
 TEA maintains the N0_ concentration at the closed end of the tube at  or very
 near zero, a concentration gradient is set up between the open and closed ends
 of  the tube.  This causes a net flow of NCL from the environment to the TEA,
 where it is captured.  For storage both before and after sampling, both ends
 of  the sampler are capped.  The complete sampler weighs approximately one-
 quarter ounce.  The materials for the construction are inexpensive, and all
 of  the components except the polyethylene caps are cleaned and reused.  It is
 estimated that a technician could either prepare or analyze approximately
 100 samplers in a working day.

      The quantity of NO- transferred by diffusion during an exposure can be
 calculated using Pick's First Law of Diffusion.  This law states that if there
 is a concentration difference between the two ends of a diffusion barrier,
 there will be a flux of the gas species for which the concentration difference
 exists.  This flux will increase linearly with the concentration gradient
 and with the cross-sectional area of the diffusion path, and will decrease
 linearly with the length of the diffusion path.  The quantity of gas trans-
 ferred by diffusion will also depend on the time of the diffusion flux.  This
 can be expressed in the equation
                             n    = D — ct
                             %)2   U L CC

 where QNO  = quantity of NO- transferred during exposure (moles)
       D    = coefficient of diffusion of NO- in air (moles/cm )
       A    = cross-sectional area of tube (cm )
       L    = length of tube (cm)
       c    = concentration difference between open and closed end (moles/cm )
       t    ° time of exposure (sec).

 Thus, it is seen that the sampler depends on three constants, D, A, and L,
 and on two variables, c and t.  Since for a particular tube the dimensions
 are known and the coefficient of diffusion can be estimated from values in
 the literature, when one measures the quantity of N02 trapped and knows the
 time of exposure, he/she can calculate the time-weighted average concentration.


     To determine the accuracy and precision of the samplers, it was necessary
to set up known environments of N02 and to expose the samplers for definite
times to these environments.  Known concentrations were produced by diluting
the output of an N02 permeation tube maintained at a constant temperature.
The constants for the sampler described were:

                      DNo2-Air = 0'154 cm2/sec'

                             A - 0.71 cm2,
                         and L = 7.1 cm.

     Substitution of these constants and conversion to more conventional
units gives the expression:

                      QNQ  = 2.3 (ppm N02 x hrs) x 10

Average N0~ concentration was calculated from QN_  and time of exposure.

     The agreement with permeation tube values, and between individual sam-
plers receiving identical exposures, was excellent.  The calculation of results
was greatly simplified because the N02 was apparently converted quantitatively
to nitrite in the final determination, which was essentially that of Levaggi
et al. (2) and Blacker (1).  It was found more accurate and convenient, however,
to combine all of the reagents and to add a single volume of combined reagent
directly to the sampler.  It was then transferred to a cuvette and read at
540 nm.

     A number of trials were conducted in which samplers were mailed to coop-
erating laboratories, exposed by them, and returned to NYU for analysis.  Re-
sults reported to the cooperating laboratory in ppm N02 x hrs and the NYU
value obtained by dividing by the time of exposure were compared with the
measured concentrations at the cooperating laboratory.  All of the compari-
sons gave very good agreement, even one in which the samplers were mailed to
Johannesburg, South Africa, and returned to New York for analysis.  The con-
centrations used in these comparisons varied from fractional parts of a
million up to several parts-per-million.


      In the course of the earlier work, it had been demonstrated that the
 samplers were quite stable both before and after sampling.  This made possible
 the mailing of samplers to distant points while still obtaining satisfactory
 results.  The standards for occupational exposure are, however, approximately
 100 times higher than those for ambient air pollution by N02.   It was con-
 cluded that the sampler would be satisfactory for measuring lower levels,
 however, if the time of exposure was increased from 1 working  day to a full
 week.  Using a 1-week exposure, a study (7) was carried out on 10 dwellings
 equipped with gas stoves and 9 with electric stoves.  Since all of these
 dwellings were in the metropolitan New York area, all of the homes were cen-
 trally heated.  It was found that the homes with gas stoves had very signi-
 ficantly higher NO- concentrations than did those with electric stoves.  This
 was true both in the kitchen and nonkitchen areas.  It should  be pointed out
 that the level found in the kitchen of the gas stove dwelling  was 49 ppb,
 which compares to the U.S. Primary Air Quality Standard of 50  ppb (annual
 average).  Electric stove kitchens in this study, however, showed an average
 of about 8 ppb.  A second study carried out in cooperation with a group in
 London (3) showed very similar results in two kitchens with gas stoves and
 two with electric stoves.  A further study was carried out on homes in the
 Gainesville, Fla., area and included homes with 1) no unvented gas appliances,
 2) gas stoves, 3) gas space heaters, or 4) combinations of gas stoves and gas
 space heaters.  The results of this pilot study (to be published elsewhere)
 would indicate that the unvented gas space heater can be a significant con-
 tributor to indoor NO,, concentrations.

      Finally, outdoor air has been monitored on a weekly basis for 2 years on
 the 2nd and 14th floors of a building in Manhattan; these results are being

      It  is concluded that the N02 sampler can be a very effective tool for the
 monitoring of this gas at ambient levels.

      The device is light, inexpensive, maintenance-free, and can be satisfac-
 torily exposed by untrained personnel.

      A 1-week sampling period appears to be suitable for air pollution


     The stability of the samplers, both before and after sampling, makes
feasible their distribution by mail and return to a central laboratory for
analysis.  They therefore appear to be very practical for simultaneous mea-
surement in a number of areas of either indoor or outdoor N(>2 concentrations.

     This work was supported by Grant No.  G0133066 and No.  G0177042 from the
U.S. Bureau of Mines, and is part of a center program supported by Grant No.
ES00260 from the National Institute of Environmental Health Sciences.

1.   Blacker, J.H.  Triethanolamine for Collecting  Nitrogen Dioxide in the
     TLV Range.  Am Ind Hyg Assoc J 34:390-395,  1973.

2.   Levaggi, D.A., Siu, W., Feldstein, M.   A New Method for Measuring
     Average 24-hour Nitrogen Dioxide Concentrations in the Atmosphere.
     J Air Pollut Control Assoc 23:30-33, 1973.

3.   Melia, R.J.W., duV. Florey, C., Darby,  S.C.,  Palmes,  E.D.,  Goldstein,
     B.D.  Differences in NO, Levels in Kitchens with Gas or Electric  Cookers
     Atmos Env 12:1379-1381, 1978.

4.   Palmes, E.D., Gunnison, A.F.  Personal Monitoring  Device for Gaseous
     Contaminants.  Presented at AIHA Annual Meeting, San Francisco, 1972.

5.   Palmes, E.D., Gunnison, A.F.  Personal Monitoring  Device for Gaseous
     Contaminants.  Am Ind Hyg Assoc J 34:78-81, 1973.

6.   Palmes, E.D., Gunnison,  A.F., DiMattio,  J., Tomczyk,  C.   Personal Sampler
     for N02.  Am Ind Hyg Assoc J 37:570-577,  1976.

7.   Palmes, E.D., Tomczyk, C., DiMattio, J.   Average N02  Concentrations in
     Dwellings with Gas or Electric Stoves.   Atmos  Env  11:869-872,  1977.


Edward D. Palmes, Ph.D.
Institute of Environmental Medicine
New York University Medical Center
550 First Avenue
New York, New York  10016


      MAGE:   Before I take any questions, I would just like to make a comment.
 I think the diffusion coefficient is proportional to the square root of ab-
 solute temperature.  So, you would expect the 1 percent increase with a 10°
 Fahrenheit  rise.  That would be about 2 percent change in temperature on
 the absolute scale.  So, that would be a 1 percent increase.

      PALMES:  Per 10 degrees?

      MAGE:   You said Fahrenheit, and you got 1 percent.

      PALMES:  Well, the thing is that as the temperature rises, the volume
 also increases.  That is_one power.  So, it actually goes T % divided by
 T  , and so it goes by T A  It is thus absolute temperature to the one half.
      ZISKIND:  Richard Ziskind, Science Applications.  I had never seen it
 addressed in health studies.  Maybe you can tell me if any account was taken
 of the NO- level in the interior of the homes when they classified this as
 a high exposure/low exposure district.  1 wonder if that might be a real
 confounding factor that wasn1t taken into account?

      MAGE:  I believe the classifications were according to the ambient levels
 of N02.  If there is anybody here who would like to correct me, please do so.

      WHITE:  Otto White, Brookhaven Laboratories.  It appears from the equa-
 tions that if you increase the area or decrease the length that you increase
 the sensitivity.  Is it that simple?

      PALMES:  It is that simple up to a point.  If one gets down to a very
 short tube, then it starts behaving as if it were not quite so short.  We
 have taken these down as low as 1 centimeter by using much smaller diameter
 tubes than this.  But we have actually changed just the length in a tube of
 the constant diameter, and this has a direct linear effect.

      If,  however, you change both the diameter and the length, then we found
 that so long as we stayed above about 1 centimeter in length, then transfer
 could be  predicted quantitatively.

      GOLD:  Avram Gold,  Harvard School of Public Health.  Have you ever con-
 sidered the possibility of losses of the NO. onto the acrylic?

      PALMES:  We have certainly considered that, and we have tried the ef-
 fect of storage on exposed samples and then analyzing them.  We have done
 this for  at least a couple of months after exposure and have obtained the
 same results.

     In other words, once sampled, we can then measure for a very  consider-
able period after exposure and still get the same results.

     GOLD:  I think that that was what I was wondering—essentially whether
you have lost NO. into the wall during the sample that you would never get

     PALMES:  You are talking about the acrylic, not the plastic.  Well, we
get the theoretical transfer.  Now, there might be a monolayer of NO,, laid
down onto the acrylic.  I don't know.

     Incidentally, we get exactly the same results with glass as we do with

     GOLD:  Glass can soak it up.  I have had some experience with this.

     PALMES:  I want to finish my story on this one.  The first time we took
these into a. mine, they said we couldn't take them below ground.  They wouldn't
allow any glass down there.  So, we stuck them on the wall.  But after I got
down there and saw these big loaders running around when we had only silly
little lights on our heads—because otherwise they would never know we were
there, and there is no place to hide—I thought, boy, glass or plastic is the
last thing I would worry about.

     DEBBRECHT:  Fred Debbrecht, AID.  Relative to the length versus diameter
again, do you have a gut feel for the ratio that they must be to still pre-
serve the diffusion effects?

     PALMES:  The A to L ratio is going to determine the sampling rate with
the given coefficient of diffusion.  Our sampler would have the equivalent
sampling rate of approximately 50 ml per hour.  The 3M sampler, which is again
based on an air gap, has a sampling rate, as I recall, of around 30 ml per

     But in any event, it is almost a factor of 60 or so difference in this
area to length ratio, so that it looks like it can vary quite widely.  Now,
If one is going to get involved with that, I think that the effect of turbu-
lence is going to become much, much greater—that is my gut feeling on it.

     SPENGLER:  John Spengler.  I am really fascinated by your measurements.
Some of the measurements Inside gas-cooking homes are very similar to the ones
we have been making In our health study.  But I encourage you to make out-
door measures simultaneously, because really the relative difference seems
to be important.  And have you noted whether you have seen an effect with
venting In the kitchen?  Because this apparently will change the value drasti-
cally in the gas-cooking homes that we have been measuring, as to whether
there was venting or not in these kitchens.

     PALMES:  We did not check whether or not the gas stoves were vented in

 those  which we  measured.   What we find generally Is that people use the vent
 only when they  are frying Chinese food or something that smells or gets
 a  lot  of splashing.   In other words, It is not used to control the combustion
 gas  part of It.   We  have not done this, but It certainly would make a dif-
 ference;  there  Is no question about that.  Now, we have used this for out-
 door measurements, but have been having difficulty finding good values with
 which  to compare ours.  We have 2 years of measurements In New York.

     TURNER:  Bill Turner.  Do you know If your method Is at all susceptible
 to NO  Interference?

     PALMES:  It is absolutely not susceptible to NO interference.  We have
 modified this,  and the article is in press now.  So, we can use it for NO
 by converting the NO to N0« with the dichromate.  But if you let it go
 through the screen,  bounce off the dichromate, and then try to bounce back,
 it gets caught.

     So, we can determine, then, both NO and N0~ simultaneously.  Of course,
 there  are different coefficients of diffusion.  But if you run one of these
 samplers along  with an analytical sample, then you can determine their dis-

In-Vehicle Air Pollution Measurements:
A User's  Perspective
     Richard A. Ziskind, Ph.D.
     Science Applications, Inc.
     Los Angeles, California

     Science Applications, Inc. (SAI) conducted  measurements of gaseous
pollutants in heavy-duty diesel tractors;  transit,  school, and mini-buses;
taxlcabs;  and police  cars.  In performing  these  research  programs and in
planning future ones, we have concluded that in  order  to  accomplish study
objectives, a good  deal of thought must be given to the selection of instru-
mentation and that  compromise is expected.  This manuscript will review our
user experience in  formulating and conducting in-vehicle  air pollution meas-
urements as they relate to the development and usage of personal exposure

     The basic elements of two in-vehicle  monitoring projects are summarized
below.  General conclusions are made about the objectives and requirements
of such programs.

     An experimental  program was conducted for the  U.S. Department of Trans-
portation (DOT)  to  measure cab concentrations of toxic gases.  Driver com-
plaints, accident autopsy results, and very limited testing results motivated
the DOT to sponsor  this scoping study.  Program components were to:
     •  Survey the  heavy-duty diesel truck population and measure cab
concentrations of toxic gases under a variety of conditions;

      •   Interpret  the results In terms of short- and long-range driver
health  and  performance effects.

      This  study,  sponsored by the Environmental Protection Agency,  was
specifically to look at the problem of carbon monoxide (CO) exposure in
sustained-use  vehicles such as school buses, taxicabs, police cars, etc.
Here,  the  motivation for the study was similar in that incidents of drivers
or  passengers  being overcome by CO had been reported,  and concentrations
in  the 100-ppm range measured.  The components of the study were to:
      •  Conduct a literature survey dealing with issues related to  the
problem—i.e., concentrations measured, causal factors of source and path-
ways,  measurement approaches (biological and air concentration), and the
concentration/effects relationship;
      •  Delineate factors affecting CO exposure, including vehicle  design,
operation  control conditions, utilization characteristics, maintenance,
and ambient control techniques;
      •  Conduct a limited sampling of vehicles in order to develop  test proto-
col for a  full-scale sampling program, observe the effects of postulated
controlling factors, identify a worst-case subpopulation, evaluate  instrument
performance, and establish the range of CO concentrations within and among

      Such studies can easily be seen to have a number of objectives.   Phrasing
 these as  questions,  we may ask the following:
      1)   To which substances may a driver or a passenger be exposed?
      2)   How are these measured satisfactorily, including both biologically
 by  performance or by instrumentation?
      3)   What are the sources of this exposure?  How does it get to the
 driver or passenger?
      4)   What conditions control or affect exposure?  Is vehicle self-
 contamination important?
      5)   Is any segment of the population at risk?   If so,  who and to
 what  degree?
      6)   Are additional regulations needed?  What form should they take?

     Other questions could be posed; however, all of the above  directly
Impact on Instrumentation requirements.  Specifically, In order to  determine
which substances may be present in the vehicle interior (Question 1), we
could draw up the body of emissions measurements made up principally of exhaust
analyses and urban ambient monitoring.  This does not directly  consider fuel
evaporation and blowby.  Nor does the emissions literature deal with differ-
ences in the source to interior transfer function among pollutant species.
Secondary pollutant formation and photochemical or other conversion reaction
rates are not well established between pollutant production and occupant expo-
sure for the same vehicle.  It would appear to be necessary to  initially con-
duct a screening of the principal Identified toxic emissions.  This requires
a variety of portable instrumentation which is preferably direct reading.
Additionally, we must have a good deal of confidence that the error caused by
interference of other substances in a detector is minimal or well known.

     This brings us to Question 2.  If substances act synergistically, is
it more appropriate to measure performance indicators or physiological func-
tions such as respiratory parameters?  Even for a single pollutant such as
CO, it is more direct to measure exposure effect or risk by blood carboxy-
hemoglobin (COHb) level through expired air or blood analysis sampling.  How-
ever, in order to identify exposure levels for individual substances, and to
determine pathways and dependence upon postulated factors,  specific air
sampling instrumentation is needed.   In our initial effort to survey instru-
mentation for the continuous monitoring of CO, NO, NO-, and hydrocarbons,
161 instruments were evaluated for 22 different performance characteristics.
These 22 characteristics could be grouped into six general categories:
simplicity, portability, dependability, reproducibility,  ease of calibration,
and stability.

     Specifically, for our diesel tractor exposure study,  we evaluated the
particular needs of this program,  such as the ability to  quickly and easily
perform a large number of idling and on-the-road measurements and the par-
ticular requirements of the diesel cab testing environment.   We established
the following requirements that the measurement system  had  to meet:
     1)  The instrumentation system had to be integrated  into a single, rela-
tively compact, and lightweight package with self-contained  power supplies
sufficient for continuous operation for at least 10 hours.
     2)  A system had to be capable of measuring ambient  and truck cab level
concentrations of carbon monoxide (CO), nitric oxide (NO),  and nitrogen
dioxide (N02), with required accuracy (+ 2 percent).

      3)  The system had to be capable of accurately and automatically record-
 ing the CO, NO, and NO^ concentrations measured  on a  permanent  record,  so  that
 the operating technician would have  sufficient time to note and record  the
 other important test variables and parameters (such as cab entry configurations,
 traffic and road conditions, and  temperature and elevation),  to readily sample
 ambient air, and to generally control the  time-wise flow of the test program.
      4)  The instrumentation system  and concentration levels  recorded had
 to be insensitive to the vibration levels  encountered in the  diesel truck
 cabs, and they had to be relatively  position- and orientation-independent
 (i.e., not require a flat horizontal surface for accurate instrumentation
      5)  The system had to be capable of measuring any particular  conditions
 (e.g., temperature and atmospheric pressure) that would affect  the concentra-
 tion readings.  Power supplies for all components for the instrumentation
 system had to be quickly and easily  rechargeable overnight by simple plug-in
 to any available 120-VAC source (e.g., a motel room outlet).
      6)  The system had to include internal pumps for continuously drawing
 the sample air through the measuring instruments, and to have a convenient
 set of lines and manifolding to enable easy sampling  of the air at any  loca-
 tion within the truck cabs (including sleeper cabs) and of the  external
 ambient air environment.  All such sampling lines and manifolding  had to be
 constructed of materials that would  not significantly alter the concentrations
 of CO, NO, or N0« in the air passing through them.  The system  had to include
 appropriate chemical scrubbers in the input to each instrument  to  remove
 potentially interfering constituents.
      7)  The system had to include small portable subsidiary  devices for the
 measurement of external factors such as wind speed and direction,  particularly
 for idling testing.
      8)  All electrical cables and devices carrying the concentration data
 signals from the three measurement instruments to the final recording point
 had to be shielded from spurious  electrical signals (truck-generated and
 external) , and the instrumentation system  had to be fully capable  of operation
 on either the self-contained power or external 120 VAC.
      9)  The instrumentation had  to  be able to respond quickly  enough to
 track transients in the cab due to temporary conditions such  as cigarette
 smoke, tunnel passage, road grades,  etc.
     10)  The cab environment should  not be contaminated by the  instrumen-
 tation exhaust.

      The specialized measurement  system designed and  constructed to meet the
 above requirement is shown in Figure 1 and is schematically illustrated in
 block diagram form in Figure 2.   Air to be sampled is pulled  in (from a cab

FIGURE 1.  Continuous sampling instrumentation assembly.
                      100    80    60  I  40    20    0
FIGURE 2.  Field measurement  system  and  data  output  schematic.

 Interior location  or  the  ambient  external  environment)  through  sampling  lines
 and manifolding of Teflon and 316  stainless  steel  construction.  The  instru-
 ments are each direct  reading,  portable AC,  or battery-operated  gas analyzers
 with self-contained pumps,  batteries,  and  battery  chargers.   The sensors are
 of the electrochemical voltametric type with linear output and  direct reading
 in parts-per—million  by volume.   The  instrumentation was  verified  to  be  posi-
 tion/orientation-independent and  Insensitive to electrical signal  output
 vibration.  The output signals  were transmitted through shielded coaxial
 cable to a  special multiplexing switch in  order to allow  the  three levels to
 be recorded on one single-channel, lightweight, portable  stripchart recorder.
 The instruments could  be  automatically sequentially cycled over  an adjustable
 period or manually controlled.

      We believe this  type of approach is a suitable compromise  when the  study
 focus is divided between  gathering survey  data on  exposure levels  in  a number
 of vehicles and attempting to establish source importance, mechanisms of in-
 trusion, and parameters which affect  interior concentrations.

      In order to consider the previously posed fifth question—whether any
 segment of  the population is at risk,  and,  if so,  to what degree—a distinctly
 different testing  approach should be  considered.   Specifically,  as we formu-
 lated a sampling program  to assess the CO  problem  in sustained-use vehicles,
 we developed a multisegmented monitoring approach.  Initially,  we  surveyed
 the large variety  of  instruments  and  devices available  for the  measurement
 and detection of CO in air at ambient occupational levels.  For  the purpose
 of this study, we  restricted consideration of relatively  nonportable  labora-
 tory instruments to serve only  as standards  or for comparison purposes.
 Measurement approaches reviewed included portable  CO instruments/detectors,
 biological  monitors,  tracer gases, and exhaust analyzers. Although all  these
 measurement approaches are incorporated in the protocol,  only the  first  cate-
 gory, portable CO  instruments/detectors, will be considered here.

      Portable CO Measurement/Detection Devices. These  devices  are identified
 by characteristics of  size, weight, power  requirement,  and insensitivity
 to position, movement, vibration,  and vehicle acceleration/deceleration.
 Suitable specifications for the class could  be taken as less  than  1 ft  volume,
 15 Ib weight, and  requiring only  external  DC power (if  any) or  internal  bat-
 teries with minimum 8-hour life between charging.   Instrument bulk temperature
 range should span 40°  to  over 100° F,  relative humidity up to 95 percent, and
 in-vehicle  pressure range from  up to  7,000-ft elevation,  equivalent to slight
 overpressure due to ventilation blower. Accurate  correction  factor/charts

 should be available.  Temperature correction/stability  should  separate  sampled
 air  temperature (0° to 100° F) in addition to the bulk  instrument  range.
 Suitable devices should ideally be free from significant interferences  due  to
 other gas species and from intermittent interferences caused by  specialized
 conditions such as the electromagnetic wave environment due to CB  radios, etc.,
 operating in the near vicinity.

     Classification of Portable CO Devices for Ambient-Occupational Use.  Al-
 though maximum values of characteristics such as sensitivity, accuracy, etc.,
 are  nearly always desirable and sought after, the practical and  sufficient
 level—from both test purposes and cost viewpoints—depends greatly on meas-
 urement purpose and objectives, the ultimate use of the data and overall re-
 sults, and, of course, the present state of technology.  For example, the least
 sensitivity and accuracy would be required of simple indication/detection
 devices used in large population sampling, or of limited detection uses where
 the  device would be affixed to the vehicle interior to indicate  the presence
 of CO concentrations exceeding some limit value, such as 35 ppm.  Such a de-
 vice would require minimal sensitivity of approximately the limit value—for
 example, 35 ppm—and an accuracy of perhaps +_ 25 percent.

     In order to span the broad use/purpose and performance characteristic
 spectrum of portable CO measurement/detection devices of potential in-vehicle
 use, the devices have been categorized in four use/effectiveness classes.
 The  classes are:  continuous analyzer, cumulative dosimeter, intermittent/
 spot sampler, and limit detector.   The basic use, measurement techniques,
 and  the approximate overall ranges in physical characteristics, performance
 specifications, and costs for each of the four classes are listed in Table 1.

     The continuous analyzer class includes relatively high-performance
 portable instruments capable of continuous CO concentration measurements and
 of providing output for continuous recording.  Some are position/orientation
 sensitive due to the particular design of the liquid electrolyte or use of
 humidifying reservoirs in the internal sampling stream.   This type will be
 used in a selected subgroup of the sustained-use vehicles to conduct a detailed
 investigation of particular vehicles to determine effects of various factors
 (exhaust system condition, ventilation effects,  leak paths, ambient levels,
 the  spatial concentration distribution in the vehicle, etc.).   These instru-
ments should approach performance  characteristics of stationary monitoring/
 laboratory instruments, such as 1  or 2 ppm sensitivity and + 1  or 2 percent

                                    TABLE 1
                   Classification of Portable CO Measurement/Detection Devices
                              and Summary of Characteristic*
Measurement Type
Measurement Princl-
Catalytic Combustion
Gas Filter Correlation
Stain Tube/Pump
Colorimetric Tube/Pump
Stain Tube
Blec trochemlcal
Stain Tube/Band Pump
Colorimetric Tube/
Limit Detection
Indicator Badge
Ceramic Disc





  Position Sensitivity


  Cost Range

  Typical Cost
+1 to + 5 percent

0.1 to 2 ppm

0-50 to 0-1,000 ppm

0.1 to 1 ft3

3 to 20 Ib



$600 to $4,000

$600 Catalytic,
$1,200 Electrochemical,
Higher for Others
^15 to + 25 percent
1 to 100 ppm-hr
1-100 to 10-3,000 ppm
1/2 hr to 7 days
0-999 ppm-hr

Shirt Pocket
1 oz to 1 Ib

$1 to $600
$500 Electrochemical,
$250 Tube/Pump,
$1 Tube
(Plus reading device)
Hand Pump
+2 to + 25 percent

2 to 10 ppm
0-50 to 10-1,000 ppm

Pocket to 0.2 ft
1/2 to 2 Ib

$250-$!,000 Electro-
$90 Tube/Band Pump
+ 25 percent at beat
50 to 200 ppm

Pocket, Badge
1/2 to 2 oz

$1 to $5

       Portable CO  measurement devices  of the  cumulative  dosimeter  class are
  distinguished by  providing  a cumulative measure of CO exposure.   These personal
  air quality monitors are  designed  to  be attached to an  individual,  without
  Interfering significantly in normal activities, sampling  air in the breathing
  zone, and providing a cumulative CO dosage measurement  over the sampling time.
  Several  different measurement principles have  been employed, including:
       •   Electrochemical cell types with solid  state electronics,
       •   Stain tubes with  air throughput by pocket-size  sampling pumps, and
       •   Nonaspirated stain  tubes.
  Sample  times can  range from one-half  hour to days among the categories.
  Position insensitivity is critical, as is the  operating time between battery
  recharges or replacements.   This equipment class will provide an  important
  part of  the program for several reasons.  They measure  actual CO  exposure and
  dosage  for the  individual during normal activities.  They do not  require con-
  tinuous  technician presence.  They can provide nearly simultaneous  data to
  allow comparison  across several groups.

       Additionally, we will  utilize nonaspirated stain tubes to measure inte-
  grated  exposure in a large  number  of  vehicles.  These inexpensive devices

will  be used as a screening  tool  to  survey  randomly  selected  vehicles in
various fleets and locations  in order  to investigate for  hot  spots and
establish sampling strata for more expensive and  intensive  subsequent testing.

      Devices of the intermittent/spot  sampler class  provide the  capability
of measuring the instantaneous CO concentration at the specified  sample  loca-
tion.  They differ from the continuous analyzer class, which  also  includes
an instantaneous measurement capability, by the fact  that they are typically
more  portable, less expensive, and do  not provide a  continuous recording

      The limit detector class includes devices which  can be attached  to
an individual or a fixed location.  They provide a positive indication of
exposure to CO concentrations above some design levels.  Some school  buses
are currently equipped with such devices, which are checked periodically to
see if a particular CO level was exceeded.

      In Summary.  Devices in each of the four classes described above may
have  an application in a measurement program.  The selection of an appropriate
class of device or combination of classes will depend on the measurement pur-
pose  and objectives, the ultimate use of the data in overall results, and, of
course, the study's size and funding.

     Needs that we have identified in personal samplers, related to their
utilization for in-vehicle pollution studies, include a sensitive  range,
nitrogen dioxide (N0«) personal cumulative dosimeter, and a respirable
fraction particulate personal sampler.

Richard A. Ziskind, Ph.D.
Science Applications, Inc.
1801 Avenue of the Stars
Suite 1205
Los Angeles, California  90067

     MEIER:  Eugene Meier, EPA, Las Vegas.  Why do you say that you need
very sensitive N0_ dosimeters, in this particular case from the health point
of view?

     ZISKIND:  In our diesel truck study, of the gases that we measured, we
found NO. levels to be the closest to the levels of known toxic effects.

     MEIER:  Was it exceeding  these levels or just close to them?

     ZISKIND:  Well, it can exceed a criterion level momentarily.  But in
general, it was highly time dependent and condition dependent.  That is, we
could see that it would be very useful to have a personal sampler rather than
an electrochemical detector or instrument so that we could do more of a sur-
vey to really identify more explicitly the conditions of N02 emission leading
to these levels.

     MAGE:  David Mage, EPA.   What standard was being violated?  Was it an
occupational standard or  an ambient standard?

     ZISKIND:  It was an  occupational standard.  If I am contrasting it with
the NIOSH proposed standard of one part-per-million and a half part-per-
million action level ceiling,  then that was commonly found to be violated
where we considered the ambient and the vehicle contribution.

     If we considered the public secondary or primary standards, that was
violated all the time.
:   So you are saying it was a relatively high concentration  of NO-?
     ZISKIND:  We are  talking  about  trucking operations that in this case
are located  in urban areas.  And  specifically, we did the Southwest, and then
we are driving along on  crowded freeways.  We are getting a lot of NO..

     MEIER:  Then relate that  back again  to buses.

     ZISKIND:  We haven't made any NO. measurement on buses—only on diesel

     MEIER:  So, it is primary occupational at this time in terms of truck

     ZISKIND:  In terms  of truck  drivers, yes.

     We are  taking ambient levels, and we are taking in-vehlcle levels.  We

are trying to look at the difference.  And we are  also  looking  at the people
who are driving along right next  to  these trucks.  And  they are both the cause
of emissions, and, also, the people  are getting  some N0_  from the trucks.

     WILLIAMS:  Marsha Williams,  EPA.  I am curious about  the CO results on
the diesel trucks because I would expect, just due to the  emissions from the
diesel trucks, that you would see much higher interior N02  levels in trucks
than you would CO, relatively speaking, because  of what  they are putting out.

     ZISKIND:  That is right.

     WILLIAMS:  So, I am curious  about what the  CO results  were.   And the
other thing is:  What kind of data capture did you have on  the  instrumen-
tation of your trucks?  You lose  a tremendous percentage of  your  data just
due to instrument problems with vibration and such.

     ZISKIND:  Well, we had the equipment shock  mounted.  And we  were monitor-
ing continuously.

     WILLIAMS:  Do you know what  kind of capture rate you had,  then?

     ZISKIND:  We manifolded these things.  So,  we alternated one,  two,  and
three in the Instrumentation.  We came back to each one.  Not too much happens
between readings.  We get a pretty good square wave.  It is  constant  for 20
seconds.  Sometimes you get fluctuations when something drives  by.

     What we were looking for was driving along  and setting  up  a  different
condition.  The condition may be speed; it may be vents, heater,  air  condi-
tioning, or something like that.  We were looking at what the effect  is.

     So, we looked for the equilibrium to be reached, and then  it is  reached.
And then we say go on to another condition.   We got essentially 100 percent
capture, if I can be so bold as to make that statement.   But we didn't have
any loss on that method.

     You are right about the conclusion.   Carbon monoxide in diesel  trucks
did not appear to be as significant as in gasoline vehicles.

Calibration  of  Personal Exposure Monitors
for Personal  Exposure and  Health  Effect
     Charles L. Kimbell
     Houston Atlas, Inc.
     Houston, Texas

     Standard reference gas mixtures are required  to calibrate personal  ex-
posure monitors.  This paper describes a device whereby a low parts-per-
million (ppm) mixture of carrier gas and pollutant can be generated in minutes
with + 1 percent accuracy.

     In the  past, compressed cylinder mixtures, permeation tubes (2),  dynamic
flowmeter blending, or exponential dilution techniques were used.  These
methods require long lead times of several weeks for preparation and careful
adjustments, and they cannot easily be converted for use over wide ranges
or to use with different gases.  Permeation tubes  require close temperature
control and  are subject to  inaccuracies which have not been adequately de-
scribed in the literature.

     The data shown in Figure 1 illustrate errors  associated with permeation
devices. Long-term stress  on Teflon causes creep, resulting in a change in
permeation rate over a period of days.  Temperature changes cause excessive
permeation rate change, except in controlled laboratory environments and with
good temperature controls.

     This paper describes a Standard Reference Sample Generator (1) that is
ideally suited to preparing ppm  samples of any trace gas  in any carrier gas.


                      DAYS   2       4      6       8      10     12      14

 FIGURE 1.   Plot of  ppm  generator by Model  601 and by permeation  tube.
                            ROLLER BEARINGS

                          CARRIAGE RODS

                        SYRINGE CLAMP


                                                    STEPPER-MOTOR and HOUSING

                           FORWARD BUTTON

 FIGURE  2.    Precision  syringe  drive

            FAST/SLOW   x    SWITCH

             PULSE INDICATOR
 FIGURE 3.   Laboratory  Analyzer, Houston Atlas Model  825R.

The equipment can be removed from storage, and, within minutes, a contin-
uously flowing gas mixture is available either for short-term use or for
continuous flow for an indefinite period.  The only materials required are
reagent-grade pure trace gas and a carrier gas of a volume slightly greater
than the amount of sample required.

     This convenient method of preparing any type of ppm mixture is well
adapted to calibration of personal exposure monitors and instruments, and
for generation and maintenance of known atmospheres in environmental cham-
bers.  When an enclosure is purged, with a flowing trace gas mixture of four
times the enclosure volume, then the trace gas in the enclosure will be with-
in 2 percent of the original ppm mixture.  The trace gas generator can con-
tinue purging the enclosure and will eliminate any deterioration of the en-
vironmental chamber contents by absorption or adsorption.

     The Standard Reference Sample Generator is basic in its operation.  A
volumetrically metered minute quantity of trace gas is mixed with a continu-
ously flowing fixed quantity of carrier gas.

     Minute quantities of gas are very difficult to accurately produce with
the usual flow devices.  The sample generator uses a microliter sampler to
incrementally inject a fixed volume of trace gas into the flowing carrier.
Accurate timing and rapid injection result in ppm mixtures that are accurate
to + 1 percent.  Labyrinth mixing results in a uniformly blended sample.
An alternate to the microliter sampler method of injecting trace gas is to
use a precision-type syringe drive actuated by a synchronous stepping motor
(Figure 2).

     To determine accuracy and long-term stability of the Reference Standard
Generator, a stable analyzer was required.  A Model 825R Houston Atlas hydro-
gen sulfide (H2S) analyzer was used.  (See Figure 3.)  The unique first-de-
rivative principle of operation provides stable zero and span even at low
ppm.  (See Figure 4.)  Linearity of response and accuracy (Figure 5) allowed
accurate measurements.  The data shown in Figure 1 above were obtained in this

                        Tungsten Lamp
                        Focusing Lens
Photocell Fine Focus
Balancing Lent
                                                        Reference Photocell
                                                        Measuring Photocell
                                              -Sample Chamber

                         Reaction Window	'  PATENTED PRINCIPLES OF OPERATION
 FIGURE 4.   Rate  reading  analyzer principle.
I i
Comparison of Photorateometric
Method and Manual Tit rat ion.


. Manual umpl
reeding n ihi

gave no




                                  40      60      80
                                       PPB H2S IN AIR

FIGURE 5.   Comparison of  Houston Atlas Model 825R H^S Analyzer  and manual

titration analysis.
               TEST: Model 601  PPM Reference  Standard  Generator

 FIGURE 6.   Sample  generated by Houston Atlas,  Inc., Model 601 Reference

 Standard Sample Generator.



     Chart recordings verified the repeatability, accuracy, and long-term
stability of this method for generating trace gas mixtures.  Figure 6  is a
chart recording of the sample generated by the Houston Atlas, Inc., Model
601 Reference Standard Sample Generator.  The analyzer provided sample and
hold readings every 3 minutes.  This recording shows a 5-hour run with a
1-hour shutdown period.  Note that no overshoot occurs and that a stable
reading is obtained in minutes.  This proves that long-term storage does not
interfere with ready availability for quick use.

     This system easily lends itself to automatic verification and cali-
bration of continuous monitors.  Several installations in the field have
proven dependability during more than a year of unattended operation.

     This parts-per-million generator can also easily provide parts-per-
billion samples intermittently or continuously.  The simplicity and ease of
operation allows handling by laboratory technicians.  No delicate accessory
equipment, such as a microgram balance, is required.  This enhances portabil-
ity and field worthiness.

     Personal monitors can be calibrated and their accuracy verified by
feeding sample directly to the monitor or by placing it in a chamber whose
environment is maintained constant by continuous flow from the ppm generator.

     This equipment removes the calibration and verification of equipment
from the realm of the research laboratory to that of a practical field tool
for day-to-day use by technicians in the field.

 1.  Kimbell, C.L., Method of Testing an Analyzer to Determine the Accuracy
     Thereof and a Volumetric Primary Standard Apparatus for Doing the Same,
     U.S. Patent No. 4,114,419.

 2.  O'Keeffe, A.E., Gas Dispensing Device.  U.S. Patent No. 3,412,935.

Charles L. Kimbell
Houston Atlas, Inc.
9441 Baythorne Drive
Houston, Texas  77041

A New Family of Miniaturized Self-Contained
CO Dosimeters and  Direct Reading
     Arnold_H. Gruber, Arnold G. Goldstein, Anthony B. LaConti, Ph.D.,
     Harry G_.  Wheeler, and John Martin
     Direct Energy Conversion Programs
     General Electric Company
     Wilmington, Massachusetts

     The Direct Energy Conversion Programs component of General Electric (GE)
Company has been  engaged since 1958  in  the development of solid polymer elec-
trolyte technology for use in water  and brine electrolysis,  oxygen concen-
trators, fuel  cells, and toxic gas sensors.  In 1962-1966,  fuel cells were
produced for the  Gemini spacecraft program.  In 1964-1968,  an improved ver-
sion of the Gemini fuel cell was used on the Biosatellite spacecraft program.
As GE fuel cell technology advanced, other applications of  this technology
were developed.

     In 1972,  using this technology, General Electric introduced a small
commercial hydrogen generator for laboratory instrumentation and chromato-
graphic applications.  This small (<1 cu ft), high purity hydrogen generator
produced laboratory hydrogen safely  and on demand, eliminating the need for
hazardous hydrogen cylinders in the  laboratory.

     The feasibility of a miniaturized  hydrated solid polymer electrolyte
sensor cell was demonstrated with hydrogen in 1965 (1).  Polarographic and
potentiostatic techniques of maintaining the sensing electrode at a fixed
voltage versus a  stable reference electrode were shown.  This effort was
extended to the detection of carbon  monoxide and other toxic gases, as re-
ported in 1971 (2).  Prototype carbon monoxide (CO) instruments specifically
designed for mine use were developed under a Bureau of Mines contract (3,4).

      In a CO sensing instrument, a measurable  electrical  current  is generated
 through electrochemical reaction of  the  contaminant gas at  the  potential
 controlled electrode.  The sensor can be designed  to  selectively  detect NO,
 N°2' H2' H2S> S°2' ethan°l> an<* other gases, depending upon the biasing volt-
 age placed on the cell, the sensing  electrode  composition,  and  the scrubber
 system for removal of interfering gases.

      The electrochemical  cell  is particularly  adaptable to  small, portable
 personal dosimeters (5) and detectors, as the  cell can be miniaturized and
 can be used with solid state electronics,  lightweight pumps, and  rechargeable
 battery systems.  The major significance of  the  specific  electrochemical
 cell as a sensor is the reduction in size and  weight  of the instrument.  Now
 possible are pocket personal dosimeters  and  small, lightweight  belt or strap-
 carried detectors weighing 1 pound or less;  the  final unit  size is limited
 by the additional features of  readout, alarms, etc.,  rather than  sensor size.

      Figure 1 shows a photograph of  the  new  family of miniaturized portable
 instruments developed by  GE for the  detection  and  monitoring of CO levels in
 air.  From left to right  are the support console,  personal  dosimeter, hand-
 held direct reading detector,  and detector case  with  carrying straps.

      The heart of any gas  detection  device  is  the  sensor or  transducer.  These
 instruments utilize a hydrated  solid polymer electrolyte membrane cell  (Figure
 2) in an electrically biased  fuel  cell  mode.   The  cell generates an electrical
 signal proportional to  the CO concentration of the sample gas  in the cell,
 and displays excellent  stability over long  operating  periods.  Circuit  compen-
 sation is provided to correct for  shifts  in sensor span signal due to tempera-
 ture changes; no zero compensation is required.

      The fuel cell utilized is  packaged in  a housing  1.51 in wide x 1.0 in
 deep x 0.92 in high.  The  solid polymer electrolyte membrane,  which is  ap-
 proximately 0.01 in thick, serves  as both an electrolyte and separator.
 Bonded to the membrane  on  one side are  the  gas sensing and reference elec-
 trodes (the sensing electrode is directly exposed  to  the sample), and the
 current collecting counter electrode on the other  side.  The second side
 is also exposed to a renewable  supply of  distilled water from  the reservoir,
 since the membrane functions  only  when  hydrated.   Lexan® plastic support
 members hold the membrane  cell  in  position, while  directing  the gas flow over

FIGURE 1.  Carbon monoxide instrument family.
FIGURE 2.  Detector and dosimeter cell.

 the sensing electrode and the water flow to the counter electrode.  The
 sensing electrode of the cell is held at a fixed predetermined anodic volt-
 age using an efficient electronic control circuit.

      As with conventional fuel cells, the relevant reactions are:

      1) Sensor electrode

                             CO + HO  •*•  CO. + 2H~*~ + 2e

      2) Counter and reference electrodes

                            2e + 1/2 02 + 2H+  •*-  H20

 Note:  the counter electrode reaction is dependent upon the cell configuration
 and design, and does not affect output signal.

      The hydrated solid polymer is extremely efficient, durable, stable in
 long-term response to CO, and has a low background (zero CO) current.  Unlike
 cells employing liquid acid or alkaline electrolyte, there is no need for
 electrolyte normality control; the hydrated cell unit is stable to extreme
 variations in humidity.  No corrosive electrolytes are added to the cell;
 the only routine maintenance needed for cell operation is the periodic fill-
 Ing of the water reservoir with distilled water.


      A system schematic for the direct reading detector and personal dosimeter
 is shown in Figure 3.  In both units, the CO containing gas sample flows
 through the dust filter and gas scrubber chamber, then through the pump, sen-
 sor cell, and rotameter (flow indicator), and is subsequently vented.  The
 pump speed is set to 60 cc/min by a trim adjustment.  Although the sensor
 response is somewhat sensitive to sample flow rate (see the Operational Test
 Results section), field experience has shown the flow system to be remarkably
 stable, and trimming is required very infrequently.

      The filter and gas scrubber serves to remove small particles of dust
 from the flow system as well as removing interferant gases which might affect
 the output of the cell assembly.  Potassium permanganate on alumina is used
 for gas scrubbing; the effectiveness of this scrubber for potential inter-
 ferant gases is shown in Table 1.  The scrubber material should be changed




•__ -_ _ _

n, i



1 i 	 1 *
-> I

                    *  DETECTOR ONLY

FIGURE 3.   System schematic.
                                      TABLE 1
            Effect or. Instrument CO Response by Potential Interferants
                   H20 VAPOR

50 to 100% RH
16 to 90%
50 PPM
10 PPM
25 PPM
100 PPM
100 PPM
10 PPM
100 PPM
       (PPM) OF CO
                  *KMn04 ON ALUMINA (PURAFIL) FILTER

 when the color changes from red-purple to brown.  As can be seen from the
 table, only ethylene and acetylene (of the listed gases) provide any signifi-
 cant interference to CO with the use of the filter.

      The power supplies for the instruments are rechargeable 5.2 V, 250 MA-hr
 nickel-cadmium battery packs, which allow a minimum of 10 hours continuous
 operation after an overnight battery charge.  As previously noted, the current
 resulting from the electrochemical oxidation of CO is directly proportional
 to the concentration of CO in the gas stream.  The ensuing output signal is
 then amplified for visual readout (direct reading detector only) or visual/
 audio alarms, and is simultaneously stored in an integrating coulometer for
 later determination of time-weighted average exposure.  The detector also
 has an output jack for providing a signal to an external recorder (0 to 1

      Operational specifications for both instruments are presented in Table 2.

      Both the direct reading, hand-held detector and the personal dosimeter
 utilize many common primary components (i.e., pump, sensor cell, flow indi-
 cator) and have similar features (i.e., 5.2 V nickel-cadmium battery pack,
 filter system, alarm system, accumulated CO readout); however, each unit was
 specifically designed to handle different  tasks.  A tabulation of instrument
 particulars is presented in Table 3.  Details of the design characteristics
 of both instruments follow.

 Direct Reading Detector

      The miniaturized portable self-contained direct reading detector was
 developed for use  in a wide variety of applications, including:  1) to be
 worn by Individuals for immediate personal assessment of carbon monoxide
 in working areas such as mines, factories, motor vehicles, etc.; 2) to be
 used as a portable continuous area monitor; 3) to be used as a personal do-
 simeter for obtaining time-weighted average exposure to CO; 4) to be used in
 conjunction with a recorder for obtaining a permanent, continuous record of
 CO exposure.  The detector (Figure 4) has many features, including a one-
 half-inch LCD digital readout (0 to 1,000 ppm CO) which is readily visible
 at the top of the case.  The instrument is lightweight (1.1 Ib), small
 (5.9 in high x 3.8 in wide x 2.3 in deep), and is totally self-contained with

                            TABLE 2
          Carbon  Monoxide Instrument Specifications

          ACCURACY           ± 10% FOR RANGE 0-500 PPM CO UNDER
                             THE FOLLOWING CONDITIONS:

                             TEMPERATURE: 1-40°C

                             ORIENTATION WITHIN 45° OF VERTICAL

                             RELATIVE HUMIDITY: 0-100%

                             DAILY ZERO AND SPAN CALIBRATION

                             OXYGEN LEVELS ABOVE 16%

                             INTERFERENCE COMMONLY FOUND IN
                             AMBIENT AIR, NO. N02, H2S, S02,
                             NH3, C02. DUST, ETC.










 0-1000 PPM




 ± 10% (1 WEEK)

 ± 2 PPM (10 HOURS)

 1 TO 50°C





                                           TABLE 3
                                   Instrument Particulars
Hand Held Detector

5.9 in.
3.8 in.
2.3 in.

5.3 in.
2.9 in.
1.4 in.
              2.  WEIGHT

              3.  BATTERY PACKS

              4.  USEFUL RANGE

              5.  LOWER DETECTABLE

              6.  DIRECT READING LCD
                 IN PPM

              7.  ACCURACY
                   DIRECT READING
       1.1 Ibs.            0.63 Ibs.


    0-1000 PPM CO        0-1000 PPM CO
     1 PPM CO
                      1 PPM CO
 < * 10% OVER 0-500 PPM RANGE

              8.  ALARMS
              9.  SELF-CONTAINED
              10. ADDITIONAL FEATURES
                                         RED LIGHT
                                         WARNING AND AUDIBLE
                                         ALARM @ 200 PPM CO
                     AMBER LIGHT
                     100 PPM CO

                     RED LIGHT WARNING
                     AND AUDIBLE ALARM
                     @ 200 PPM CO
                                       (NO ADDITIONAL PUMPS, FILTERS. BATTERIES, ETC. FOR
                                        REMOTE READOUT
                                        EARPLUG ATTACHMENT
                                        LOW BATTERY SIGNAL
  its own  pump,  filter  system,  and  rechargeable battery pack.  A leather  carry-
  ing case with  shoulder strap  or belt mount  is available.

       The carbon monoxide detector controls,  adjustment points, and performance
  observation windows are shown in  Figure 5.   The  unit contains both visible
  and audible alarms set normally for 200 ppm CO,  and an earphone  connection

FIGURE  4.  Detector  and  case.
                                                    SENSOR CELL
                                                    WATER FILL
               PUMP SPEED
                ZERO ADJUST
               SPAN ADJUST
                                                                   COULOMETER READOUT
                                                                   AND INPUT TO REMOTE
                                                                   BATTERY CHARGER
FIGURE 5.   CO  detector  controls  and  adjustment points

 for operation  in noisy areas.   Windows  are available in the case for  obser-
 vation of  pump  speed  on a rotameter,  the color of  the filter material,  and
 the distilled water level in the sensor cell.   Distilled water fill and air
 inlet sample ports are visible on the top of the case.   Conveniently  located
 on the side of  the case are the off-on  pushbutton  switch, the coulometer
 readout  and input to  a remote  recorder  or readout  (Q to 1 volt), the  earphone
 connection, and the battery charger connection. Also located on the  side
 of the case for convenient setting are  the pump speed adjust, and zero  and
 span adjust screws.   In addition to a direct LCD readout of CO in ppm,  the
 unit contains  an integrating coulometer for time-weighted average data, which
 can be read out on the Support Console.  The unit  also features a self-test
 circuit  which  enables the alarms and amplifying system to be checked  each time
 the unit is turned on.   In addition,  approximately 1 hour prior to the  need
 for battery recharge, a low battery warning is provided by the appearance of
 three decimal  points, a semicolon, and  an arrow on the LCD display.   The
 alarms are momentarily activated, and the instrument display automatically
 turns off  when  the battery voltage falls below a predetermined level.

      Calibration of the direct reading  detector is carried out as shown in
 Figure 6.  A zero and span gas cylinder kit is optionally available from
 General  Electric Company.  The regulator on the calibration kit is set  for
 150 cc/min.  The pump on the unit will  draw 60 cc/min of zero or span gas.
 Adjustment of  the direct readout is made by setting the zero and span pot
 adjust screws.

 Per s ona 1 Dosimeter

      The personal dosimeter was designed for daily use by a worker in a haz-
 ardous area.  The dosimeter (Figure 7)  is extremely compact—being pocket-
 sized (4.75 in  high x 2.75 in  wide x 1.38 in deep) and light (0.6 Ib).
 No external pumps are required;  the unit is completely self-contained with
 built-in pump,  5.2 V  rechargeable nickel-cadmium battery pack, and alarms.
 The small  filter attaches to the unit air input and clips to the worker1s
 collar so  that  the input taken is in the worker's  breathing zone. The  in-
 strument has a  coulometer for  recording the worker's exposure to CO during
 the working period.

      The unit also features alarms for  the safety  of the worker which respond
 as follows:  1)  an amber LED will light if the instantaneous concentration
 of CO exceeds 100 ppm,  and 2)  a red LED and buzzer will alarm if the  instan-
 taneous  concentration of CO exceeds 200 ppm.   The  alarm levels are preset

FIGURE 6.  Calibration of direct
reading detector.
FIGURE 7.  Carbon monoxide do-
        FIGURE 8.   Carbon  monoxide
        dosimeter,  rear  view.

 but  can  be modified  by  factory adjustment.   A self-test  circuit  is  available
 for  the  alarms,  so  that when the unit  is  turned  on the alarms  will  illuminate/
 buzz  for approximately  15  seconds to advise that the  alarms  are  functioning.

       The dosimeter  unit has  all of its adjustment controls recessed in  the
 face  of  the  unit and covered by a key-locked door. This feature is provided
 to minimize  the  possibility  of the worker tampering with the controls during
 operation.   Visible  to  the worker (on  the rear of the unit,  Figure  8) are
 windows  showing  the  level  of distilled water in  the cell, and  a  rotameter
 showing  the  pump speed  (and  that the pump is working).  A pump speed adjust
 pot  is also  located  on  the rear of the case.

       Controls  and adjustment points for the personal  dosimeter are  shown in
 Figure 9. Behind the keyed  door of the dosimeter are the battery and pump
 on/off switches, the zero  and span adjust screws, and the cable  connection
 for  readout  of the  coulometer at the support console  and charging of the
 battery. Both a gas sample  inlet port and a distilled water fill port  are
 available on top of  the dosimeter case.

 Support  Console

       A support console  (6  in wide x 6  in high x  6 in  deep) is  required  in con-
 junction with  the personal dosimeter and can also be  used with the  direct
 reading  detector.  The  console digitally displays the total  cumulative  carbon
 monoxide dosage  during  the prescribed  operating  period.   The console also
 provides a digital  display of the instantaneous  carbon monoxide  level when
 used  in  conjunction with the dosimeter.  This latter  display is  used for
 calibration  of the  dosimeter, as shown in Figure 10.

       The support console also provides a battery charging capability for the
 dosimeter.   A  single and six dosimeter charging  support  console  (Figure 11)
 is available.  The  multiple  charge support console also  has  the  capability
 of testing the state of the  battery charge, and  it has a jack  plug  for  use
 with  a external  recorder.


       A table of  maintenance  recommendations for  the detector and dosimeter
 units are presented  in  Table 4.

FIGURE 9.  CO dosimeter controls and adjustment points,
FIGURE 10.  Dosimeter and support console.

         LINE CORD

                                                                  CO  DOSAGE
  FIGURE 11.  A single  and six dosimeter charging support console.
                                       TABLE 4
                               Function and Maintenance

        QUIRED. (WEEKLY)

        BROWN. (WEEKLY)






     Typical results of operational testing of the dosimeter and detector
units are presented in the next series of figures*

Response versus Concentration

     The indicated response of the units to various concentrations of CO in
air (at 60 cc/rain pump rate, 24° C) are shown in Figure 12.  The data follow
the linear response curve anticipated over the entire 0 to 1,000-ppm range
of concentration.

Response versus Time

     The indicated response of a specific CO detector unit to CO gas concen-
trations at various time intervals is shown in Figure 13.  In these tests,
three different CO concentrations (97 ppm, 460 ppm, and 1,040 ppm) were sep-
arately fed to the instrument for 200 sec (at 60 cc/min, 24° C).  At the end
of 200 sec, CO-free air was fed to the instrument.  The response time obtained
lies well within the specified response time.

Response versus Temperature

     The indicated response of dosimeter and detector units to various con-
centrations of CO (21, 58, and 97 ppm) at different temperatures is shown in
Figure 14.  In these tests, the sensing instrument was left at the specified
temperature a minimum of 60 minutes prior to obtaining the reading.  The
results obtained are within specifications established for the instrumen-

Response versus Flow Rate

     The indicated response of the detector to variations in sample flow rate
with 97 ppm CO calibration gas, after calibration at 60 cc/min and 24° C, is
shown in Figure 15.  The results show some sensitivity of output response to
flow rate change.  The pump is, however,  extremely stable in operation and
will maintain the pump rate within + 2 percent in normal daily operation.
Variations encountered as a result of normal flow rate changes lie well with-
in the specifications established for instrumentation.

          MO  -
          700  -
                           FEED FLOW - 60 CC/MIN
                           TEMPERATURE - 24*C
          600 -
            0 100    300    600    700     900
                  CONCENTRATION OF CO IN AIR (PPM)
 FIGURE  12.   Response versus  concen-
 tration for  a General Electric port-
 able CO detection instrument.
                                                                 1040 PPM CO IN AIR
                                                  (PPM CO IN AIR)
                             FEED FLOW 60 CC/MIN
                                                           zoo -
                 CO-FREE AIR

                 460 PPM CO IN AIR
                                                                 97 PPM CO IN AIR
                                                                                  CO-FREE AIR  -
                  80     160     240     320
                     RESPONSE TIME (SECONDS)
FIGURE 13.   Response versus time for
a General Electric  portable CO  detector
to  CO in  air  feeds.





1 T 1 1 1

ff^**"^ \~~*****+

1 1 1 1 1





0 10 20 30 40
FIGURE 14 . Response versus tempera-
ture (° C) for General Electric CO gas 10
detection instruments.

I I I I I I I ^
"~ .^^^ ~
_ ^^^ —
~- ^^ — -
_ * —

— ' —

97 PPM CO AT 60 CC/MIN _
1 i i i i i i i
                                                                      SAMPLE FLOW (CC/MIN)
                                                 FIGURE 15.   Effect of flow on response
                                                 for  a General Electric CO  detection


     A new family of miniaturized self-contained carbon monoxide dosimeters
and direct reading detectors have been described.   The units are extremely
lightweight, versatile, self-contained, and meet the requirements for in-
trinsically safe operation.  The cell needs no acid;  only distilled water is
required.  The units are an outstanding example of the combination of func-
tional design engineering and technical experience based on fuel cell tech-

 1.  LaConti, A.B., Dantowitz, P., Kegan,  R.,  Maget,  H.J.R.   Hydrogen Sensors,
     Electrical Bias Approach.  TIS Report #66 ASD3,  General  Electric Company.

 2.  LaConti, A.B., Maget, H.J.R.   Electrochemical Detection  of H,,  CO and
     Hydrocarbons in Inert and Oxygen Atmospheres.  J Electrochem  Soc 118:
     506, 1971.

 3.  Schnakenberg, G.H., LaConti,  A.B.  Improved CO Detection Systems.
     Proceedings, 17th International Conference,  Bulgaria, 1977.

 4.  Dempsey, R.M., LaConti,  A.B., Nolan,  M.E., Schnakenberg, G.,  Chilton, E.,
     Torkildsen, R.  Gas Detection Instrumentation Using Electrically-Biased
     Fuel Cell Sensor Technology.   PB 254823,  NTIS, Springfield, Va.  1978.

 5.  Jones, R.C., LaConti, A.B., Nuttall,  L.J.  Carbon Monoxide Monitor
     Features Hydrated Solid  Polymer Electrolyte Cell.  Proceedings,  American
     Industrial Hygiene Conference, New Orleans,  La,, May 1977-

Arnold H. Gruber
Direct Energy Conversion Programs
General Electric Company
50 Fordham Road
Wilmington, Massachusetts  01887

      ZISKIND:   Richard  Ziskind,  Science Applications.  How much field experi-
 ence  has  been  accumulated  on this  instrument?

      GRUBER:   The dosimeter has  been in the  field  for well over a year.
 The detector is being  introduced at  this  point  in  time.  The circuitry and
 design of both the detector and  the  dosimeter are  quite  similar.

      Before we introduced  the product, we at GE did  considerable testing of
 the product, both in-house and with  some  field  testing as well.  Both units
 have  been tested for at least 2  years, the detector  being tested for at least
 1 year prior to its introduction.

      ZISKIND:   The other question  I  had was  the question that was asked before
 about electromagnetic  interference.

      GRUBER:   As a matter  of fact, the unit  also has a plastic case.  The
 plastic case is metalized.  One  of our specifications is no electromagnetic
 interference,  and the  units will meet that specification.

      WILLIAMS:   Marsha Williams.  To what extent has the instrument been test-
 ed at temperatures below zero Centigrade?

      GRUBER:   As with  some of the  other electrochemical  cells used below zero,
 we have a water cell.   The water will freeze, and  you can deteriorate your
 cell  at low temperatures.   This  is the same  with all of  the electrochemical

      All  of the electrochemical  cells have specific  temperatures to which
 they  can  go down.   Regarding a water cell, obviously you wouldn11 want to go
 below zero degrees Centigrade.   With some of the acid cells—they can go down
 below zero Centigrade.

      MEIER:  Could you  say something about cost?

      GRUBER:   Vic Socol, would you like to say  a couple  of words?

      SOCOL:  The dosimeter costs $595.  There are  various options on the sup-
 port  console,  which cost from $695 to $825,  and a  direct reading indicator,
 at $945.

      McSTRAVICK:   John  McStravick, Energetics Science.   Just for the record,
 the current Energetics  Science dosimeter  sensor can  go to -17° Centigrade.

Design and Performance  of a Reliable
Personal  Monitoring  System  for  Respirable
    William A. Turner, John D. Spengler, Ph.D., Douglas W. Dockery, and
    Steven D. Colome
    Department of Environmental Health Sciences
    Harvard School of Public Health
    Boston, Massachusetts

     Personal  exposure to respirable  particulates and sulfates is being meas-
ured as a part of a long-term prospective epidemiological  study of the res-
piratory health effects of air pollution, the Harvard Six  City Study (1).
The purpose of this monitoring program is to develop better  estimators of
actual personal exposure from comparison of the direct measurements of
personal exposure with simultaneous measurements of the normally measured out-
door air, the  air inside each participant's home, and records of the daily
activities of  each participant.   Results are reported elsewhere in these
Proceedings (2).

     Respirable particulates and  sulfates were selected for  monitoring be-
cause they were considered to be  important potential toxic agents.  Further-
more, for practical considerations, reliable sampling equipment which modeled
the definition of "respirable particulates'' (3) was commercially available
and readily adaptable for use in  the  community.

     The commercially available samplers have been designed  to meet Federal
requirements  for respirable dust  exposures in mines (4).  In this program,
MSA Model G and Bendix BDX-31 samplers were used.  The sampler consists of
a 10-mm nylon  cyclone presampler  to remove the nonrespirable particles,
followed by a  filter to collect the respirable fraction.  Air is drawn through
the sampler by a battery-operated diaphragm pump which is  equipped with

 a stroke counter and miniature  rotameter.   The  pump  is  designed  to  operate
 from the batteries  for  a normal 8-hour  work shift.   Flow is  controlled  by a
 needle valve.  Normal operation calls for  overnight  charging of  the batteries.
 At the beginning of the sample, the  pump is turned on and allowed  to run for
 approximately 15 minutes to  eliminate initial battery voltage surge before
 setting the  flow to 1.7 liters-per-minute  (1pm).  At the end of  the sample,
 the flow is  checked and the  total  number of strokes  recorded as  an  indication
 of total sample volume.

      In applying these  samplers in the  community, the first  problem we  en-
 countered was unacceptable noise.  The  situation was somewhat improved  by
 placing the  pump in a hard plastic box  lined with sound-absorbing material.
 The box was  equipped with a  shoulder strap, and the  cyclone  was  mounted on
 the outside  of the  box. Problems  arose with sampling lines  being crimped
 within the box with this setup.

      The second major problem was  with  the limited sampling  time.   A sampling
 time of 24 hours was desired for comparison with regular State and  Federal
 air pollution sampling  programs, and also  to ensure  sufficient collected
 mass for analysis.  Given a  nominal  battery life of  8 hours, a supplementary
 power source was required.   The battery chargers operate in  two  modes—16-
 hour and 64—hour recharging. At the high  rate, 16 hours, the chargers  supply
 enough power so that there is only a slight drain on the batteries.  Thus,
 the participants were instructed to  place  the monitor in some fixed location
 whenever they were  in a room for some extended  period—for example, on  their
 desk while working  in their  offices. The  charger was then plugged  into 120
 VAC outlet and then into the pump.

      This procedure proved to be less than ideal.  The  pumps were  still noisy.
 The chargers were bulky and  inconvenient to carry around. In the  first sam-
 pler program in Watertown, Mass.,  37 people agreed to participate.   Thirteen
 did not last through the first  sampling day. In the end only 18 highly moti-
 vated people remained.  In addition, we encountered  several  electrical  prob-
 lems.  First, if the MSA chargers  were  plugged  in while the  pumps were  op-
 erating and  the charger was  on, they blew  their internal fuse.   Secondly,
 the higher voltage  supplied  by  the chargers quickly  deteriorated the motor
 brushes, and we had a very short motor  life. The charger connected to  the
 pump was not secure so  that  the plug often slipped out  without the  partici-
 pant's knowing it.

      The final problem  was variation in the flow rate.   The  actual  flow was

proportional to the voltage supplied to the motor.  Since  even with the
chargers attached, the pumps still drew power from  the batteries,  the  flow
declined during the day as the battery voltage decayed.  On  the  average,
the end flow was 80 percent of the nominal initial  setting.

     The commercial samplers were modified at several steps  along  the  way,
but it was readily apparent that these samplers were not acceptable for our
program.  As a result, the Harvard/EPRI sampler was developed.

     The design criteria were:
     1)  Provide a constant flow rate of + 0.1 liter-per-minute  (1pm)  over
the range of 0.5 to 3.0 1pm;
     2)  Include the self-contained capability to be recharged overnight and/
or to run off line voltage;
     3)  Produce a noise level acceptable for home and office  use;
     4)  Operate for 12 hours or more on batteries;
     5)  Portable, lightweight, rugged, yet simple to operate;
     6)  Provide a sample comparable to our fixed-location indoor/outdoor
monitoring system; i.e., system operating where no pulsation is  present.

     One hundred personal monitoring pumps have been built and have passed
initial testing.  They are presently being field tested and will be put into
full use during a personal monitoring program during the spring  of  1979.

Size:  6 in x 6 in x 3 in
Weight:  4 Ib
Pump:  "Brailsford" Brushless TD-3LL or TD-3L
Battery Pack:  "Gould" 12 V-1.2 SC cells Ni-Cad
Flow Control:  Variable Constant Voltage
Range of Operation:  0.5 to 3.0 1pm
Case Metal:  Aluminum
Cyclonic Separator:  10-mm nylon with filter
     •  Self-contained battery charger
     •  No warm-up time required to reach stable flow
     •  14 to 20 hours sample time on battery mode, indefinite from line voltage
     •  Minimum maintenance brushless pump (10,000 hours before service)
     •  Quiet operation.


      Air is drawn through the 10-mm nylon cyclone, a 1.0 y pore size Fluoro-
 pore filter, and a pulsation damper by a constant displacement pump (Figure
 1).  The flow control through the system is controlled by varying the speed
 of the pump instead of by conventional throttling methods.  This allows for
 precise flow control with minimum energy loss due to pumping losses induced
 by throttling.  The variable constant voltage power supply allows the pump
 to be set at a given flow rate and will maintain the flow + .05 1pm under
 normal sampling loads and 4^ .1 1pm under severe filter loading conditions.

      The Harvard/EPRI sampler has a six-position switch as follows:
      1)  Battery Run
      2)  Line Run
      3)  Line Run and Charge Low
      4)  Charge Low (36 to 40 hours)
      5)  Charge High (12 to 14 hours)
      6)  Off.

      These six positions allow the unit to serve as a  portable personal
 monitoring device or a stationary 24-hour sampling unit.  The unit has a
 detachable 8-foot, three-wire safety power cord and a  detachable 48-inch
 carrying strap.  The respirable sampling cassette and  cyclonic separator can
 be detached and located apart from the main body of the sampler.

 Flow Stability

      Laboratory and field tests conducted between October and December of 1978
 have shown the flow to be stable to +_  .05 1pm under normal filter loadings
 for 12 to 24 hours, after an initial break-in period for each unit (72 hours).
 This break-in period is required to seal the pump valves and to establish a
 diaphragm set.  On some units it has been necessary to install a restricting
 orifice in line to allow the pump to obtain lower flow rates.  This does not
 affect the stability or designed sampling life  times while operating from

 Effects of Pulsation on Cyclone Efficiency

      The effects of air flow pulsation on the cyclone collection efficiency
 in personal sampling devices have been investigated by numerous researchers


                                      LINE VOLTAGE
                                   VOLTAGE REGULATING BCARD

                                   AND CHARGING CIRCUIT
                               BATTERY PACK
FIGURE 1,  Schematic of flow system,

                                  -AP TRANSDUCER
FIGURE 2.  Test apparatus.

 (5,6,7).  To determine the flow characteristics of the Harvard/EPRI pump,
 a Fleisch Pneumotachograph (#7316[000]) and a pressure transducer were coupled
 to an amplifier and oscilloscope.  Figure 2 is a schematic display of the
 test apparatus.  For comparison with other pumping systems, an index of
   ,  ..    ,D   [MAX FLOW - MIN FLOW].    ,  .   _           .    ...  .
 pulsation  (P = -»•	MEAN FLOW	 ' a     e fre
FIGURE 3A.  H/E without damper.
FIGURE 3B.  H/E with damper.

 In addition, the self-contained charging circuitry and lower noise  level have
 eliminated the nuisance characteristics that prevented broader participant
 acceptance.  The Harvard/EPRI pump has been designed to meet the  particular
 sampling requirements of assessing community personal exposures in  an  epi-
 demiologic air pollution health study.

      Steve LeMott, Harvard School of Public Health,  Department of Physiology,
 for aid and use of the pulsation measurement equipment.
      Pat Quinlan, for pump assembly and circuit testing.
      Electric Power Research Institute, Palo Alto,  Calif.,  Contract No.
      National Institute of Environmental Health Sciences, Grant No. ES01180.

  1.  Air Pollutants and Health:  An Epidemiologic Approach.   Environ Sci
      Technol 11:648-650, 1977.

  2.  Dockery, D.W., Spengler, J.D.  Personal Exposure to Respirable  Partic-
      ulates and Sulfates:  Measurement and Prediction.   Proceedings  of  the
      Symposium on the Development and Usage of Personal Monitors  for Exposure
      and Health Effect Studies, Chapel Hill, N.C., January 22-24,  1979.   (EPA

  3.  Llppmann, M., Harris, W.B.  Size-Selective Samplers for Estimating
      "Respirable" Dust Concentration.  Health Physics 8:155, 1962.

  4.  Federal Coal Mine Health and Safety Act of 1969.

  5.  Caplan, K.T., Doemeny, L.J., Sorenson, S.D.  Performance Characteris-
      tics of the 10 mm Cyclone Respirable Mass Sampler:  Part I.  Monodisperse
      Studies.  Am Ind Hyg Assn J 38:83-95, 1977.

  6.  Caplan, K.T., Doemeny, L.J., Sorenson, S.D.  Performance Characteristics
      of 10 mm Cyclone Respirable Mass Sampler:  Part II.  Coal Dust  Studies.
      Am Ind Hyg Assn J 38:162-173, 1977.

  7.  Blackman, M.W., Lippmann, M.  Performance Characteristics of the Multi-
      cyclone Aerosol Sampler.  Am Ind Hyg Assn J 35:323-325, 1974.


 William A. Turner, John D. Spengler,  Ph.D.,
 Douglas W. Dockery, and Steven D.  Colome
 Harvard School of Public Health
 Department of Environmental Health Sciences
 665 Huntlngton Avenue
 Boston, Massachusetts  02115
(NOTE:  The discussion for this  paper appears following the text  of  the  next
paper, "Personal Exposure to Respirable Particulates and Sulfat^s:   Measure-
ment and Prediction.")

Personal  Exposure to Respirable Particulates
and Sulfates: Measurement and  Prediction
     Douglas W. Dockery and John D. Spengler, Ph.D.
     Department of Environmental Health Sciences
     Harvard School of Public Health
     Boston, Massachusetts

     Epidemlological studies of the health effects of air pollutants have in
the past frequently depended on ambient air quality measurements at a single
site within a community for correlation with observed health effects.  Given
that Americans normally spend 95 percent of their time Indoors (1,2), such
measurements of outdoor air pollution  levels may not represent the actual
air pollution exposure of the sample population.

     Jackson and Newill (3) noted that "perhaps the weakest  link in our at-
tempts to establish the exposure-response relationship is in obtaining an
accurate quantitative estimate of exposure."  Morgan and Morris (4), in a
study of the needs for personal monitoring, concluded that "the importance
of population exposure measurements in air pollution epidemiology makes it
imperative that future epidemiological studies include exposure estimates
more representative of what people actually breathe."

     A monitoring program to measure personal exposure to respirable parti-
culates and sulfates was undertaken as part of a long-term epidemiologic
study of the respiratory effects of air pollution (5).  These observations
were compared to measurements of the ambient outdoor environment and of the
indoor environment in each participant's home.  In a similar study, Binder
et al. (6) had 20 school children carry a suitcase sampler with them for 1
day.  They concluded that outdoor measurements do not accurately reflect the
particulate exposure of individuals who live in the area of  sampling.

      With observations of actual personal exposure to respirable particulates,
 we hope to develop a method to characterize the personal air pollution expo-
 sure of individuals in terms of their personal activities, characteristics
 of their home and household, and ambient pollution level.  Average exposure
 can be estimated by a simple time-weighted model of the form

                               ~C »** t c  / *• t
                               L  i Eici ' i Ci

 where C is the average exposure, t  is the time spent in activity i, and c.
 is the air pollution concentration associated with that activity.

      Ideally, the air pollution concentration associated with each activity
 would be known, so that the exposure of any individual could be estimated
 directly from his or her activities.  It Is unrealistic to expect that we can
 specify air pollution concentrations for every activity, but hopefully we
 can make reasonable estimates for broad classes of activities.  While not
 specifying exposure exactly, these estimates should be significantly better
 than the outdoor values alone.

      The method used, therefore, was to measure the integrated exposure of
 individuals with different activity patterns.  Comparison with simultaneous
 measurements outside and inside their homes allowed the most important factors
 in determining exposure to be identified and quantified.

      Volunteers were asked to carry a small, personal particulate monitor
 with them for 24 hours as they went about their normal daily activities.
 They were also asked to keep a diary of their activities that day.  Similar
 monitors were run concurrently at fixed sites in their home and at several
 locations outside in the community.  Samples were taken in Watertown, Mass.,
 every 6th day for 2 months in the summer (July and August 1975) and 2 months
 in the winter (mid-January to mid-March 1976).  The study was repeated in
 Steubenvllle, Ohio, in June 1976.  In this case, 12-hour samples were taken
 on Monday, Wednesday, and Friday each week.

 Volunteer Selection

      Volunteers were solicited at random from census lists.  The study was
 described to the potential participants in a letter followed by a phone call*

Those interested in the study were visited and shown the equipment.  Although
the intent was to draw a representative sample of the population,  this  did
not occur.  In general, anyone doing physical work could not conveniently
carry the monitor.  Most working people with manufacturing jobs were not al-
lowed to bring the monitors to work by their employers for fear that they would
uncover occupational health violations.  Many potential participants found
the program too burdensome.  The final set of volunteers provides  a very
wide range of exposures and activities, but was not representative of the
sample population in the health study.

     Whenever possible, we attempted to have both the husband and  wife  in a
family participate.  A total of 37 people participated, 18 in Watertown and
19 in Steubenville.  There were five smokers in the Watertown sample, and
one in the Steubenville sample.  Twenty-one of the participants were women,
and 23 were employed away from home.  All those working had white  collar jobs
except for one self-employed plumber.

Pollutant Measurements

     In order to characterize those particulates which are most likely  to
have a significant respiratory health effect, we chose to measure  only  those
small enough to penetrate deep into the lungs.  The AEC Los Alamos criteria
(7)—which specify that 50 percent of the 3.5-um particles penetrate deep
into the lungs, while particles larger than 10 ym will not—was used.   This
also provides a convenient starting point for measuring individual exposure
to air pollution in the community in that sampling devices had been developed
for the industrial environment based upon these criteria.

     The sampler consists of a Dorr-Oliver 10-tnm diameter hyd roc lone pre sam-
pler, followed by a 37-mm Fluoropore filter with a 1-ym pore size.  Air is
drawn through the sampler by a Mine Safety Appliance Portable Pump—Model
G was used in Watertown, and a Bendix BOX Super Sampler in Steubenville.
A flow rate of 1.8 liters-per-minute has been shown by Seltzer, Bernaski, and
Lynch (8) to deposit particulates in the cyclone presampler in close approxi-
mation to the Los Alamos curve*  This was the flow used in this sampling.
The American Industrial Hygiene Association (9) has recommended a  flow  of
1.7 llters-per-minute (1pm) for the 10-wm cyclone.  Knight and Lichti (10)
have shown essentially no change in the mass collected on the filter between
1.3 and 2.65 1pm.  It is felt, therefore, that this difference in  flow  rate
does not compromise the samples.

     The pumps are battery-operated, with a nominal operating time of 8 hours.

 It  was possible to operate the pumps for longer periods by connecting them
 to  their battery chargers with only a small drain on the batteries.  Since
 initial tests showed that the major obstacle for people participating in
 the study was the noise of the pump, the pump was placed in a sound-insulated
 box with a strap so that the monitor could be carried over the shoulder, as
 shown in Figure 1.

 Sampling Plan

      In Watertown, samples were collected for 24-hour periods every 6th day
 to  coincide with Federal and State air quality sampling programs.  In Steu-
 benville, 12-hour samples were taken on Monday, Wednesday, and Friday, be-
 ginning at 8 a.m.  At that time,  the volunteer turned on his personal monitor
 and a similar one placed in the home in the room with the most activity (ex-
 cluding the kitchen).  At the same time, outside monitors were turned on by
 automatic timers.  Thus, we had three measurements for each sample—personal,
 inside the home, and outside.

      We assumed that the particulates were uniformly distributed over volumes
 the size of a room.  The actual placement of the monitor on the person was
 therefore not critical, and breathing zone samples as required in industry
 were not necessary.  Volunteers were instructed to connect the pumps to the
 chargers whenever they were in one area for an extended period.  In this way,
 we  were able, in most cases, to run the samplers for up to 24 hours.  (This
 type of operation is not recommended by the manufacturer.)  The volunteers
 were also asked to keep a diary of their activities during the sampling period,
 including how many cigarettes they were exposed to each hour.

      Outdoor monitors were located throughout the community.  The outdoor
 monitoring was part of a continuing program to evaluate the indoor/outdoor
 pollution relationships.  The outdoor data were pooled to give the mean out-
 door value for each day with observations.  Hereafter, this pooled outdoor
 mean will be referred to as the "outdoor concentration."  This is the concen-
 tration which would be obtained by an ambient air pollution monitoring site
 using this sampling technique.

 Sample Analysis

      Filters were preweighed on a Cahn 4100 Electro-Balance in a humidity-
 controlled room (RH <50 percent)  and sealed in plastic sampling cassettes.
 The volunteer was visited after each sampling day, the filters were changed,
 and the operation of the pump was checked.  Flows were checked periodically,

FIGURE  1.  Personal  respirable  particu-

late monitor showing  sound deadening

box and normal  carrying mode.
                                                                           PERSONAL raMITORI'IO DATA SliEET
&>c AW«
Smoker £no)
—-~^—^—~ Ti^f

Mode (circle one)
(^"Personal-' Fixed (describe location —
TTrtTT'SldlT. > etc.)
1/2 pk/day
1 pk/day
1 1/2 pk/day
2 pk/day

Filter '.'

Period oo
— '• • J ••-. i-'. 8" - ° Vv\.
\- f \ . ^ f\f\ o "* ?^ ^-^ ^ - rvi-
t. Pre

/ /• T UF/n3
              2+  pk/day
Summary of Activities
  In transit  car
             other  (descr-.be)

  Home  outdoors
         kitchen  area
         living rn, den, etc.
                                                                                      Indoors (Hrs/Dav)   Outdoors  (Hrs/Dav)
  Shopping/Recreational outdoors
    Grand Total (better be
                close to 2l»)

Summary of Exposure
  Hear or In smoke
  Time you were smoking or It of
  ."Jear or In dusty area
    road dust

Summary of Transportation

  1. Label nap with  sampling date  and filter nunbers and  name.

  2. Locate home (H)  and work (W).  Trace transit route.

  3. Trace and/or locate general travel (non-core-.ute).
                                                      FIGURE 2.   Example of  personal monitoring  data sheet used
                                                      in 1976 winter sample.

 but not after each sampling day.  The used filters were returned to the lab-
 oratory and conditioned for 24 hours before reweighing.  The filters were cut
 in half, one half being stored for future analysis.  The other half was ex-
 tracted in 1,100 pi of water with 50 pi of ethanol as a wetting agent in
 an ultrasonic bath.  The soluble sulfate was then measured by a modified
 turbidimetric technique (11).

      The measured concentrations were then analyzed to compare the personal
 exposures, the indoor, and the outdoor measurements.  The activity patterns
 from the diaries were used to evaluate the proposed linear time-weighted model
 for exposure.

 Activity Diary

      Each participant was asked to keep a diary of his or her activities during
 the sampling period.  Each was asked to make special note of the number of
 cigarettes smoked in his or her vicinity, or of any other source of pollution
 of which he/she was aware.  Figure 2 is an example of the activity diary
 used in the first sampling period.  The volunteer was asked how much time was
 spent indoors and outdoors at home, work, and other locations, plus the time
 spent in transit.  There was space for the volunteer to make notes on the ac-
 tivities hour by hour, but this was only to assist him or her in determining
 the total time spent in each activity.

      The diary entries were aggregated to give the total amount of time spent
 indoors at home, at work, and in other places—e.g., shopping.  The amount of
 time outdoors was determined at home, at work, in transit, and all other
 places.  Table 1 shows the mean time spent by the Watertown participants in
 each of these activities for the two seasons.  As would be expected, the total
 amount of time spent indoors was significantly higher in winter (95 percent)
 than in summer (86 percent).  Although the amount of time spent outdoors in
 winter is about one-third that of summer, in both seasons an average of 1
 hour was spent in transit; that is, traveling by auto, public transportation,
 foot,  or some other method.

      These results are similar to other time-usage studies (1), which have
 shown people in 44 U.S. cities to spend 94.5 percent of their time indoors—
 70.9 percent in the home.  These studies also found that an average of 1.25
 hours per day were spent in transit.

                                    TABLE 1
                Mean Dally Time Spent In Various Activities for
                      18 Participants In Watertown, Mass.
Location Mean (SD)* %
Home 16.8 (6.5) 71.9
Work 2.6 (3.9) 11.1
Other 0.6 (1.8) 2.6
Total Indoors 20.0 (5.9) 85.6

Home 1.7 (4.7) 7.3
Work 0.1 (0.3) 0.2
Transit 1.0 (1.2) 4.1
Other 0.7 (1.8) 2.9
Total Outdoors 3.4 (4.9) 14.5

Total All 23.4 100
Mean (SD)* %
18.2 (6.4) 78.5
2.8 (4.7) 12.0
1.0 (2.1) 4.2
21.9 (4.3) 94.8
0.0 (0.3) 0.2
0.1 (0.9) 0.4
1.0 (1.2) 4.2
0.1 (0.5) 0.5
1.2 (1.5) 5.2
23.1 100
*SD « Standard Deviation
     One volunteer in Watertown attempted to evaluate the relative impacts
of different modes of transportation.  One day, he took his car to work.
The next sampling day, he bicycled.  The third day, he canoed down the Charles
River.  Unfortunately, any differences which might have existed were masked
by the high partlculate levels from his smoking.


      In Watertown, a total of 287 personal and 251 indoor samples were col-
 lected.  Over half the potential samples were missing because of equipment
 problems, vacations of the participants, voided observations, and scheduling
 problems.  In Steubenville, 245 personal and 131 indoor samples were collected.
 The data-retrieval rate there was higher because of the shorter sampling time
 and closer supervision.

 Sulfate Relationships

      There are two questions to be answered from this data.  First, is person-
 al exposure related to the normally monitored outdoor air pollution levels?
 Secondly, if not, what is the best estimator of personal exposure, and what
 is the minimum information required for that estimate?

      We will consider only comparisons of mean levels for each individual in
 the study.  Observations are included only when we have a complete set of
 personal, indoor, and outdoor measurements of concentration for that individ-
 ual and that day.  While cutting down the number of available observations,
 this ensures comparability.  We will also require each individual to have at
 least three complete sets of observations in order to be included.

      Consider the respirable sulfates first.  Although sulfate particulates
 are generated by matches (12) and the open burners of gas stoves (13), the
 contribution is small compared to the concentrations of sulfate in the outdoor
 air in Watertown and Steubenville.  We can then easily explore the relation-
 ship between personal exposure and outdoor concentration.  In Figure 3, mean
 personal exposure levels are plotted against the mean matched outdoor concen-
 tration for each individual.  The points are clustered into two groups:
 Watertown observations with the lower values (mean outdoor « 5.8 yg/m ),
 versus Steubenville observations with higher values (mean outdoor - 17.6 yg/m )

      The correlation coefficient (r) between the outdoor and personal means
 is 0.81, but from visual inspection it is clear that the outdoor means are
 not good estimators of personal exposure.  There is considerable scatter,
 especially in Steubenville, about the perfect prediction line, the diagonal
 in Figure 3.  The proportion of the variance of the mean personal exposures
 explained by a linear regression on mean outdoor concentrations (r ) is 65
 percent.  But considering the mean outdoor values to be a direct estimate of
 personal exposure, a more appropriate statistic is the standard error about
 that estimate,  which is 4.86 yg/m .


S  15.0*
*H    _
S  10.0+
              *  •  •
              • t
          ? •*  2*
                                        5.0*      2
                                               23  2
                                                 2 »
     0.0         10.0
               Outdoor Mun Suim.
FIGURE 3.  Relationship between
mean outdoor  and mean  personal  sul-
fate levels (ug/m ).
                                                     10.P         20.n
                                                5.0         15.0         25.0
                                                                      Exposure Model
                                   FIGURE  4.  Relationship  between  mea-
                                   sured personal  mean sulfate levels
                                       m ) and predicted  personal mean
                                               sulfate  levels from time-weighted
                                               exposure model.
                           * 2  •
 2  »
•2 »
3 •
]    -
1    .
5  40.*
                                                             »   • •   *
      0.          DO.           80.
            20.           60.          100.
           Outdoor Hun Ruipirabl* P«rtlcul«t«

FIGURE 5.  Relationship  between
personal mean and outdoor mean
respirable particulate levels
                                              20.          60.          100.
                                                 TlM-N*ight«d Expoiur* Model

                                   FIGURE 6.   Relationship  between
                                   measured personal mean respir-
                                   able particulate levels  (yg/m )
                                   and predicted personal mean
                                   respirable particulate levels
                                   from time-weighted exposure

      We can test these pairs for a significant difference between them with
 a paired t-test.  In this case, the mean difference among the 37 observations
 is 3.06 yg/m .  The t value is 4.87, which implies a significant difference
 at the 0.1-percent level.

      We can also test the variance about the estimator—that is, the out-
 door means—versus the variance of the personal means.  The ratio of these
 gives an F of 0.625, which is significant.

      Assuming that there are no significant sources of respirable sulfate
 particles indoors, the personal exposure should be equal to the mean of the
 concentration outside weighted by the time spent outside, and the concen-
 tration inside weighted by the time spent indoors.  Assume that the indoor
 concentration for each individual is represented by the level measured inside
 his/her home.  Figure 4 compares the mean personal levels for each individual
 with the mean time-weighted model estimates of exposure.  Comparison with Fig-
 gure 3 shows less scatter of the data about the diagonal.  The r  has increased
 to 0.765, and the standard error of the estimate has decreased to 3.75 yg/m .
 The paired t for this data was 2.61, which implies the difference was still
 significant at the 5-percent level.

      Considering the fact that such a large fraction of the time is spent
 indoors, it is reasonable to ask how good the indoor value by itself is as
 a predictor of personal exposure.  Table 2 compares the correlation, paired
 t, and standard error of the estimate using outdoor, the time-weighted model,
 and indoor measurement as estimates of personal exposure.  The time-weighted
 model does not perform significantly better than the indoor measurements alone.
                                     TABLE 2
          Comparison of Outdoor Means, Time-Weighted Exposure Model, and
        Indoor Means as Estimator of Personal Exposure for 37 Observations


Time Model

Time Model

Respirable Particulate Relationships

     Whereas  the mean personal  sulfate exposures are determined  by  outdoor
sources, personal exposure  to respirable partlculates Is determined by  both
outdoor and Indoor sources.  The Watertown personal monitoring program  demon-
strated the importance of smoking, as one source, in determining  personal
particulate exposures (12).

     Comparison of the mean personal exposure with the mean simultaneous
outdoor measurements (in Figure 5) shows more scatter than with  the sulfates
(Figure 2).   In this case,  r  = 0.479, and the standard error of  the estimate
is 20.18 yg/m .  Again we see the clustering of the Watertown data  at lower
levels (outdoor mean - 17.5 vg/m ), and the Steubenville data higher (outdoor
mean » 67.3 wg/m ).  There  Is a tendency for higher personal values compared
to outdoor values in Watertown, where smokers were included in the  samples.
In Steubenville, where smokers were not Included, the outdoor mean  values
are more representative of  actual personal exposure.

     The paired t of -0.851 indicated no significant difference, but this
apparently results from the large variance.

     Personal exposure was  estimated by a time-weighted combination of  indoor
and outdoor concentrations  plus a time-weighted parameterization of  smoking
exposure indoors away from home.  Smoking within the home is intrinsically
included by the concentrations measured indoors.  Smoking outside the home
was estimated from the number of smokers each participant reported  he was ex-
posed to away from home.  Regression analysis indicated that the impact of
each smoker added 20 yg/m   to the average exposure.  The resultant  model was,

          "C + [t.     . c. .     + tt.c+t.    . (c, .
              1 home    indoors    out    out    other     indoors
                         + "smok ' 2° rt'l ' 'tot

where *C is estimated exposure to particulates, t,     is the time spent at
home, c.  .     is the concentration measured indoors at home, t  .  is the
       indoors                                                 out
time spent outdoors, c    is the concentration measured outdoors, t  .    is
the time spent indoors away from home, N   v is the number of smokers the
participant was exposed to away from home, and t    is the total sample

      Figure 6  compares mean personal exposures with mean values estimated
 by  this method.   The r  has increased slightly to 0.570, and the standard
 error of the estimate has decreased, to 14.3 yg/m .  The paired t test still
 shows no significant difference.   Personal values were regressed against
 just the indoor  values to assess  the improvement in prediction of the time-
 weighted model.   Table 2 compares the statistics for all three methods.
 While using indoor values significantly improves prediction of personal values
 over Just the outdoor values,  the addition of a smoking variable gives a much
 better estimate.  It is clear also that a best fit to the data in Figure 6
 would give a non-zero intercept,  indicating that there are other contributing
 sources and exposure factors that have not been included.

      Although personal exposure to respirable particulates and sulfates is
 determined in part by outdoor levels, indoor concentrations are,  because of
 the large fraction of time spent indoors, a much better estimator of exposure.
 In the case of respirable particulates, passive exposure to smokers contributes
 significantly to personal exposure.  In addition, other activities still to
 be determined are adding to the particulate exposures.  Further studies are
 planned in the relatively clean ambient environment of Topeka, Kansas,  this
 spring in order to define these contributions.

      This work has been supported by NIEHS Grant No.  ESDI180.   The study was
 possible only through the understanding and cooperation of our volunteers in
 Watertown and Steubenville.

  1.   The Use of Time:   Daily Activities of Urban and Suburban Populations
      in Twelve Countries.  (Szalai, A., ed.).   Paris, Mouton, 1972,  p.  114.

  2.   Chapin, F.S.   Human Activity Patterns in the City.  New York,  Wiley-
      Interscience,  1974.

  3.   Jackson,  D.L., Newill,  V.A.  The Strengths and Weaknesses of Population
      Studies in Assessing Environmental Health Effects.  In:   Proceedings
      CEC-EPA-WHO International Symposium on Environment and Health,  Paris,  1974

 4.  Morgan, M.G., Morris, S.C.  Needed:   A National R&D  Effort  to Develop
     Individual Air Pollution Monitor Instrumentation.  J Air  Pollut  Control
     Assoc 27:670-672, 1977.

 5.  Air Pollutants and Health:  An Epldemiological Approach.  Environ  Sci
     Technol 11:648-650, 1977.

 6.  Binder, R.E., Mitchell, C.A., Hogein,  H.R.,  Bouhuys,  A.   Importance  of
     the Indoor Environment in Air Pollution Exposure.  Arch Environ Health
     31:277-281, 1976.

 7.  Lippman, M., Harris, W.B.  Size Selective  Samplers for Estimating  "Res-
     pirable" Dust.  Health Physics 8:155,  1962.

 8.  Seltzer, D.F., Bernaski, W.J., Lynch,  J.R.   Evolution of  Size-Selective
     Presamplers, II. Efficiency of the 10-mm Nylon Cyclone.   Amer  Ind Hyg
     Assoc J 32:441-446, 1971.

 9.  AIHA Aerosol Technology Committee:   Guide  for Respirable  Mass  Sampling.
     Amer Ind Hyg Assoc J 31:133, 1970.

10.  Knight, G., Lichti, E.K.  Comparison of Cyclone and Horizontal Elutri-
     ator Size Selectors.  Presented Annual Meeting, Amer  Ind  Hyg Assoc,
     Denver, 1969.

11.  Tabatabai, M.A.  A Rapid Method for  Determination of  Sulfate in Water
     Samples.  Environ Letters 7:237, 1974.

12.  Dockery, D.W., Spengler, J.D.  Personal Exposure to Respirable Partlcu-
     lates and Sulfates Versus Ambient Measurements.  Paper #77-44.6, Air
     Poll Contr Assn Meeting, Toronto, 1977.

13.  Hollowell, C.D., Budnitz, R.J., Traynor, G.W.  Combustion Generated In-
     door Air Pollution.  Presented, 4th  Intl Clean Air Congr, Tokyo, May
     16-20, 1977.

Douglas W. Dockery and John D.  Spengler, Ph.D.
Harvard School of Public Health
Department of Environmental Health Sciences
665 Huntington Avenue
Boston, Massachusetts  02115

     (NOTE:  The  following discussion  took place after the presentation of
the papers  "Design and Performance of  a Reliable Personal Monitoring System
for Resplrable Participates" and  "Personal Exposure to Respirable Particulates
and Sulfates:  Measurement and Prediction.")

     McCURDY:  My name is Tom McCurdy.  I work for the Pollutant Strategies
Branch, EPA, and  my comments and  question are for Mr. Dockery.

     Our office is doing some time averaging in high pollution exposure
monitoring.  We are into human activity.  1 also have a couple of comments.
One, Chapin is not a sociologist.  He  is an urban planner here at the Univer-
sity of North Carolina.  He is a  physical planner.  1 think he would reject
being called a sociologist.

     Number two,  there is a Dr. Robinson of Cleveland State University who
has some very good time activity  data.  Another comment is that for modeling,
we need hourly human activity data, not daily data like you aggregated.
That may be true  only for pollutants with a 1-hour averaging time.  Have you
published any of  your summary of  the activity data anywhere?

     DOCKERY:  We have presented  papers at meetings in the past couple of years.
It hasn't been directly published.

     On the averaging time, we are doing measurements on the health effects
every 3 years.  And even if you get down to 24-hour average exposure, it is
much finer  detail than we needed  in determining what our health effects are
going to be, although it depends  on whether the argument is the long-term
average health exposure or the peak levels that are really causing the
health damage.

     We are trying to keep information down to the hourly level, but at least
in terms of exposure, or exposure models, we are using 24-hour average values.

     McCURDY:  I  realize that.  But for our purposes, we would really like to
have some hourly  data.  So, if you are running it, try to keep in mind the
most desired aggregated time we could get for a lot of people probably is
on the order of 1 hour.

     One further  question:  I don1t understand what you are saying about your
regression  line and your coefficient of determination.

     You were putting up data that had less of an R-square value, but you
were saying that  you had less data in that sample.  It seems to me your data

were contradictory with what you were  saying.

     DOCKERY:  When you use this correlation, you are  looking  at  an arbitrary
fit to any particular line that you can  fit  through  there.  The only differ-
ence in the one that you have compared this  back to, the  one-to-one reversing
line, is that we are comparing his actual measurements versus  some  predictor,
just to see how well that works.

     In reality, what we are looking at  is F testing,  Student's t-testing,
things on that order.  But I didn't want to  get that deep into the  statistics
today.  I was just using the R-square  as an  indicator, trying  to  get an in-
dicator that everybody was familiar with.

     The only point I really want to make is that when we have gone to  more
complicated models beyond just looking at the indoor measurements,  we have
seen significant Improvement in the statistics.

     MEIER:  Gene Meier, EPA.  The first question I have  for you  is on  the
type of instrumentation used.  Your indoor personal measurements  were taken
with your instrument.  The outdoor numbers were obtained  with  what  type of

     DOCKERY:  It is the same cyclone.  And  the only difference is  we used  a
critical orifice to control flow instead of voltage control.

     MEIER:  The other question I had was, have you done  any comparisons  to
ensure that you are getting the same data with the two different  types  of
instrumentations so that that is not affecting your correlation?

     TURNER:  You are talking about Indoor-outdoor monitoring  done  with both
of the main orifices.

     The personal exposure data were done with the commercially available moni-
toring unit I described.

     MEIER:  So, indoor-outdoor measurements were taken with the  same in-

     TURNER:  Yes.

     DOCKERY:  I may have to qualify that a  little bit, but we will  talk about
that later.

     MEIER:  The other question I have is in terms of your correlation  of
indoor-outdoor data.  You have got a selected population  group depending on
your volunteers.  The homes that you are looking at may not be the  average
home.  So,  even if you did find a correlation,  that correlation may not hold
for the population as a whole.

     DOCKERY:  Well, I think the jump you are making is that we can relate
personal monitoring to the Indoor pollution.  OK;  I am not saying what  the
relationship is between the indoor and outdoor pollution  at this  point.  But

with our more extensive  indoor-outdoor  network,  then we  can  make  that  ad-
ditional step.

     MEIER:  The  thing that we  are  looking  at  is trying  to make the  corre-
lation of the outdoor measurement to  the  indoor  measurement—that  is where
we would like to  go  eventually.

     According  to your data, we  cannot  do that because  they  don't  correlate
well.  And my question is  this:  whether  or not  in  the  population  subgroup
that you selected, you selected  a very  narrow  group of  people  or homes that
are either well-insulated  because of  their  economic capability, with air
conditioning, etc.,  the  house is well-sealed,  compared  to what you would ex-
pect in the average  population  subgroup?

     DOCKERY:   I  think that you  have  a  legitimate point  there—that  we don't
have a completely unbiased sample of  what the  population is.   You  know,^in
our indoor-outdoor monitoring,  we tried to  select homes  based  somewhat on
household characteristics  and on variables  that  we  are  looking to  better
define from this  study.

     As Jack Spengler said, we  have gone  through 2  years of  this.  And what
we are doing now  is  shutting down that  monitoring,  going through an  analysis
phase, and trying to determine what the variables are  that we  have good handles
on, and what things  we don't know about.  Next year we  will  begin  a  new phase
of indoor-outdoor monitoring to  pick  up some of  the other variables  in the
household mix.

     RICKEY:  John Hickey, Research Triangle Institute.  You mentioned that
whenever any of your volunteers  went  to work,  the company industrial hygien-
ists threw them out.  For  whatever  their  motives, this  may have been a bless-
ing in disguise,  because the occupational limit  for respirable dust  is 5,000
micrograms.  That may have completely masked any of your other respirable
dust data, plus the  fact that the occupational exposure  would  be a site-specific
type of dust and  not comparable  to  what they would  be getting  the  rest of the

     DOCKERY:  That  is very true.   It is  just  a  little  unsatisfying  to us
because we have to characterize  everyone  in our  population.  We are  not
selecting a population because  they are working  or  nonworking  people,  and
especially in Steubenville, Ohio, we  have a lot  of  steel workers,  for  example.
We actually did get  some measurements in  the steel  mills, but  I didn't report
on those.

     PETERS:  Edward Peters.  I  still don't quite understand your  outdoor meas-
urements.  Did each  of your individuals have a personal  meter  and  was  it
directly associated  with an outdoor meter?

     DOCKERY:  Each  person has a personal meter,  and each has  one  inside his
or her home.  In  addition, we had our indoor-outdoor network of monitors
running at the same  time in the  community.

     So,  we had five homes in each  of these communities  where  we were  taking

our measurements at the same time.  These were independent,  in general,  of
the taking of the personal measurements.

     It gave us a spatial distribution across the community, and  that  is what
we were trying to define—whether there were big spatial distributions,  so
that if a person moved from one area of the community to another,  he was
getting a different outdoor exposure.

     WHITE:  Otto White, Brookhaven Laboratories.  My question is, what  is
the personal exposure to the individual obtained from the personal dosimeter
as opposed to a personal sample?  By going to a weight-height measurement,
a lot of the activities performed in the home are done at that particular
height—the preparation of food, the dispensing of dry detergents.

     DOCKERY:  We have done some studies on this, looking at the distribution
of particulates throughout the home, looking at the vertical distribution
within the home.  We really haven't seen any significant differences in  any
of the homes.

     I am sure these differences exist on a short-term basis.  But when  you
integrate that exposure over a long period of time—you know, when somebody
is cooking in the kitchen, you are going to get a high peak—but when you
average that out over 24 hours, the differences don't show up anymore.

     WHITE:  There is quite a bit of industrial data, obtained by putting  dupli-
cate samples on a worker, that show a considerable gradient exists.

     DOCKERY:  Clearly, it is a different case in the industrial environment.
I mean, we are talking about levels that were two or three orders of magnitude
lower than what you see in the industrial environment, and you know, there is
rapid turnover there—air changes on the order of 1 to 2 air changes of  the
whole building air per hour.  There is very rapid mixing within the home.

     WHITE:  I would probably agree with that if that sample were taken  in
the respirable zone—in the breathing zone.

     DOCKERY:  This system measures the respirable fraction of less than
3.5 microns, and you know, they have residence times on the order of days
because their settling velocities are very small and they are very rapidly

     I think the key to understanding the distribution is that there are
multiple sources in the home—smoking, vacuuming, and everything else that
can generate particulates outside the kitchen area.  The monitoring we have
done to look at that specific question on distribution of particulates in
the home shows that it Is well mixed.

     Now, we are facing contradictory evidence if you look at N0?> which
Ed Palmes points out is twice as high in the kitchen area as it Is in any
other living area of the house.  So, you have that conflicting evidence  ver-
sus what we see in the fine particulates, which is well mixed; no significant
differences between rooms in the house.

     We also looked in the vertical,  too, and we can't see any differences
in the vertical.

     I am not really that bothered by the difference between gases and the
particulates, because when we look at the outside air pollution data, we see
well-mixed particulates where you will  see  large gradients in the gaseous
pollutants in the vertical, especially  in the vertical distribution.

     SHAW:  Bob Shaw from EPA.  Isn't is possible that if you took into account
the lag between the time in which the outside levels reach a certain point
and the time it takes for it to leak  into the hpuse, that your correlation
might be improved?  Because you could have  an excellent correlation, but it
may be displaced by some amount of time.

     DOCKERY:  Clearly, there is a buffering capacity in the home.  We might
see the peaks following each other.   In fact, other studies have shown about
a 2-hour lag in the particulates penetrating into the home.

     SHAW:  That would take care of it?

     DOCKERY:  Yes, but if that thing is happening—if we are getting a sharp
peak—potentially that is getting smoothed  out.  Lag is something we really
can't define with our sampling methods.  We would have to go to the continuous

     LIN:  Chin-I Lin, Lawrence Berkeley Laboratory.  I would like to know
more about the measurements associated  with the study, since one of your
slides showed indoor-outdoor pollution correlated better in wintertime than
in summertime.  I would have expected it to have been the other way around.
Do you have any comment on that?

     DOCKERY:  I have looked at these data  by season, and they showed much
better correlation in summer than winter.   These indoor-outdoor data that we
presented are, you know, gross means.  I don't think they are a good repre-
sentation of what the real situation  is.

     When you look at individuals, however, the correlation is much, much
higher in the summer than in the winter.

     LIN:  Do you make ventilation rate measurements in these study homes?

     DOCKERY:  We don't have a good indicator of ventilation.  That is also
in the planning stage; we are trying  to include it in our next series of meas-
urements.  That is a real key.

     WALLACE:  Lance Wallace, EPA.  Have you looked at the levels in traffic
as compared to indoors or outdoors?   At least, I have noticed in my CO data,
for example, that I have got more CO  in the 2 hours I commuted than in the
other 22 hours per day.

     DOCKERY:  Let me comment on that.  I philosophically don't believe it
is important, because when you look at  the  amount of time you are spending

In traffic, we are showing, and other studies have shown, that an hour  per day
does not have a really significant impact, unless you are going to have 8
times or 20 times as much exposure in traffic.

     We have got some data that show considerably higher particulate levels
in traffic.  But I don't think they are large enough to really be of signifi-
cant impact.

     I think that the key right now is the ventilation rate, as Dr. Lin men-
tioned—trying to determine that and relate that back to the data.  We  can
get information on the sources within the home, but if we don't know the ven-
tilation rate of the home, then we aren't able to parameterize that out.

     SPENGLER:  I feel like you do.  This is where we have our difference.
We have done a lot of condensation nuclei counts in traffic.  The highest
counts I ever see are associated with roadways and in automobiles*  The order
is a magnitude higher.  1 think that is one exposure category with associated
exposure patterns that we have to define more carefully.

     WALLACE:  One might expect that the dust in traffic, particularly  the
lead, and the gases have to be more toxic than the pollution in the home.

The Tandem  Filter  Package
     Robert W. Shaw, Ph.D., and Robert K. Stevens
     Inorganic Pollutant Analysis Branch
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     The use of chemically treated filters  to  absorb gases of interest from
an air stream passing through the filter  is not new.  A computer-assisted
search of the literature has turned up about 20 research papers in which rec-
ipes for filter treatment are reported—the oldest of which was published in
1954.  References are given here only  to  those papers appearing in the pub-
lished literature (1-14).

     Almost all of this reported work  is  concerned with SO- and/or BLS.  The
recipes for SO. collection include carbonates  or bicarbonates (seven papers),
hydroxides (six papers), and zinc acetate,  ferrous sulfate, tetrachloromer—
curate, and hydrogen peroxide (one paper  each).  The recipes for H~S collec-
tion include lead acetate (four papers),  silver metal membrane (one paper),
and silver nitrate (three papers).  Only  a  few of these papers report the
use of a prefliter for aerosol particles.

     In addition to the work cited above, atmospheric analysis using chemi-
cally treated filters is reported by the  British Factory Inspectorate for 26
dangerous gases, fumes, and dust in a  particularly interesting series of book-
lets (22).  These methods are intended to detect levels higher than the thres-
hold limit values.

     The initial motives of the EPA's  Inorganic Pollutant Analysis Branch
(IAPB) in developing the tandem filter package were:  1) to provide a means of

 separating and collecting various particulate and gaseous components of the
 atmosphere;  2) to provide separation of the various chemical species of sulfur
 in gaseous compounds (e.g., S02, HjS, etc.); 3) to permit the sample collec-
 tion analysis to be carried out as simply and cheaply as possible with a
 minimum of operator intervention but with sufficient sensitivity to allow
 measurements at low (e.g., rural) levels; and, 4) to collect these components
 in a manner suitable for elemental analysis using our X-ray fluorescence
 (XRF) spectrometer.

      The tandem filter package (TFP) consists of a series of filters in tandem,
 each of which collects a different component from the air stream passing through
 it.  Aerosol particles are collected on porous, nonreactive filters through
 which gases will pass (ideally) without effect.  Gases are collected on filters
 which are individually treated with reagents which will absorb and fix the
 gas of interest.  The following sections describe the TFP which we have devel-
 oped.  It separates and collects coarse and fine aerosol particles, SO. and
 H.S.  Analysis for elements present in the collected aerosols is carried out
 by X-ray fluorescence.  Analysis for sulfur is by X-ray fluorescence or by
 wet chemical extraction of the collection filters; and analysis of the ex-
 tract is by the Thorin-Brosset colorimetric procedure, by ion chromatography,
 or by flash pyrolysis into a flame photometric detector.  Development and
 evaluation of the system for other atmospheric gases (e.g., NEL, nitrogen
 oxides) are continuing.

      The choice of filters for collection of aerosol particles is based on
 results by workers in the IPAB and their collaborators (15), and others (16).
 Briefly, one uses a front filter of large-diameter pores which collects larger
 particles, and a second filter of smaller-diameter pores to collect small
 particles.  We currently use a Nuclepore filter with 8- or 12-ym diameter
 pores,  and either a Nuclepore filter with 0.4-ym diameter pores or a Fluore-
 pore (Teflon) filter with 1-ym diameter pores.

      The collection efficiency of Nuclepore filters of large-pore diameter
 has been studied in great detail by Parker (17), and that for small-pore
 diameter filters by Liu and Kuhlmy (16).  Their results show that collection
 efficiency depends on many factors:   velocity of the air stream through the
 filter,  particle density, filter porosity, filter flow resistance, etc.
 Roughly speaking, however, the large-pore diameter filter collects particles
 larger  than 3-ym diameter, and the small filter collects particles at least

down to 0.03-wn diameter with nearly 100 percent efficiency.  The minimum
in the well-known bimodal mass distribution for typical atmospheric  aerosols
occurs for particles having diameters around 3 ym; and the deposition  of
particles during breathing occurs such that, very roughly speaking,  particles
larger than 3 ym do not reach the lower areas of the human respiratory system.

     Recent work by John et al. has shown that the tandem filters separate
solid and liquid particles with significantly different efficiencies (23).
These results, and our own experience (26), indicate that tandem filters  give
only an approximate separation of fine and coarse particles.

     In choosing a pumping system for the TFP, one is faced with the usual
compromise between the desire for the highest possible sensitivity and the
requirements that the system be relatively portable, compact, inexpensive,
etc.  The specific requirements placed on the pumping system for rate of flow
through the filter depend on:  1) the atmospheric abundances of the elements
of interest, 2) the time permitted for sampling, and 3) the sensitivity
of subsequent analysis.

     To illustrate how the above conditions dictate our choice of pumping
systems, let us say that we desire to be able to detect the gases SO. and H-S
at the 1 ppb (V/V) level and that the sampling time should be no longer than
8 hours.  Our X-ray fluorescence analysis system has a sensitivity to S such
that about 1 yg can be detected quantitatively (to 10 percent).  At 25° C
and 1 atmos, the concentration of S in H_S or SO- at 1 ppb is 1.3 yg/m .
Hence, to accumulate 1 yg of S on a filter of 100 percent efficiency, we must
pass about 3/4 m  through it, and, to achieve this in 8 hours, the pump must
have a rate of about 1.6 1/min.

     A pump, which will have the necessary capacity determined by considera-
tions as outlined above, must be able to sustain the pressure drop which
occurs across the filters at the desired rate of flow.  Liu and Kuhlmey (16)
have measured coarse and fine particle collection efficiencies as a function
of pressure drop for various commercially available filters.  The pressure
drop across the 1 ym Fluorepore filters used in this work for both fine
particle and gas collection is 10 Torr at a face velocity of 13 cm/sec.  For
an exposed filter area of 8 cm , this face velocity corresponds to a flow
rate of about 6 1/min.  For lower flow rates the pressure drop is corre-
spondingly lower, but Liu and Kuhlmey do not present any data for pressure

  .rops less than 10 Torr.  We conclude that for a filter pack consisting of 3
 1 wn Fluorepore filters, the pump must be capable of sustaining a pressure
 drop of 30 Torr if the flow rate is to be 6 1/min.  For lower flow rates the
 pumping capacity is lower, and, hence, the requirements on the pumping system
 are rather modest.

      The pressure drop results quoted above are due to the resistance of the
 filters to flow.  As particles are collected on the aerosol filters, this
 "loading" will increase filter resistance and, for very heavy loadings,
 affect the sampling rate.  For Teflon filters, the Impedance doubles at a
 loading somewhat less than 100 ug/cm  (18).  Recent analyses of air samples
 in major U.S. cities gave total suspended participates in the 20 to 60 yg/m
 range (19).  We conclude that filter loading is unlikely to cause excessive
 demands on the pumping system for the TFP.

      Size and weight limitations for a personal dosimeter create a severe
 limitation on suitable pumps.  A survey of personal sampling pumps by Parker
 et al. (24) shows that none of a range of commercially available personal
 sampling pumps can support the TFP given the demands outlined above.  We
 have used the Brallsford pump with the TFP in the field with good results,
 and a compact and portable personal monitor using this pump has been developed
 by Turner et al. (24).

      The treated filter development for separation and collection of SO.
 and H_S is based largely on the work by Huygen (13,14), who found that the
 rate of uptake of these gases in the filter media was increased with increased
 humidity.  Huygen explained this behavior by attributing the rate of uptake
 to diffusion into the liquid phase created by condensed moisture on the filter.
 We add triethanolamine (TEA) to our collection solutions to provide a hygro-
 scopic material which will keep the filter surface wet.

      Sulfur is a relatively light element, and its X-rays are, correspondingly,
 of relatively low energy.  Consequently, if the sulfur is collected in the bulk
 of the filter rather than on the surface, severe attenuation of the emitted
 X-ray intensity will occur during X-ray fluorescence analysis.  In order to
 collect the sulfur gases on the surface of the filter, the collection solution
 is applied to the filter surface in an aerosol spray.  The filter medium used
 is Teflon (Fluorepore), and, in order to prevent the aqueous solution from
 accumulating on the hydrophobic Teflon surface in large drops, sufficient

 isopropanol  is  added  to  the  solution to  permit  the  solution to wet the filter
     The absorbing  and  fixing  agents  for SO- and H-S are NaOH  and
respectively.  The  absorption  of  SO-  into basic solutions  is well  known.
In a basic environment,  the  sulfurous acid equilibrium  is  strongly shifted
to the  ionized state, and  the  escape  of SO- is hindered.   Eventually,  the
bisulfite or sulfite  ion presumably undergoes further oxidation, possibly by
atmospheric oxygen, to  SO,,  and is fixed in the filter  solution.   The  reaction
of H_S  with metals  and  metal salts in the presence of water is also well
known.  The reaction  causes  the formation of unsoluble, often  colored  sulfides,
which serves to  fix the sulfur.   Other workers have used Ag or Pb  for  this
purpose; however, these elements  have X-ray transitions which  can  interfere
with detection of the S  X-ray  and, hence, are not suitable for X-ray emission
analysis on our  instrument.  We have  chosen a salt of Cu which is  tarnished
by hydrogen sulfide and  which  has no  interfering X-ray  transitions.  The  H-S
collection solution does not collect  (CH-)-S.

     The solution recipes  for  the SO- and H-S collection filters are:

               2M NaOH(aq)     0.1 M Cu(N03>2 (aq)*     TEA Isopropanol

S02                 55%                —               27%     18%

H2S                —                 45%              18%     37%

*The H2S solution is acidified to 0.01M with HN03

     These recipes  have  not  been  systematically optimized, and adjustments
in relative amounts of  their components may improve their  collection proper-
ties.  Work is continuing on the  evaluation of the chemically  treated filters
under varying conditions; e.g., gas concentration and humidity.

     Analysis for sulfur as SO. can be carried out using ion exchange chro-
matography, or the Brosset-Thorin method (25).  Filters are extracted by
placing them in a polypropylene bottle along with 10 to 20 ml of 5 x 10  N
perchloric acid.  The bottle is put in a sonic bath for 20 minutes.


     The spectrophotometric-Thorin method  was  first  presented  by Persson
(21) and later modified  by  Brosset et  al.  (22).   With  two  exceptions,  this
method is  identical  to  that proposed by Brosset  et al.;  instead of  using
Dioxane or acetone as  a  solvent,  isopropanol  is  used,  and  sulfanazo is used
to obtain  a reference  intensity level.   Before analysis  with  the spectrophoto-
meter, the extracted solutions are put through an ion  exchange column  using
Dowex 50W-X8 50-to-lOO-mesh ion exchange resin.   The columns  are 15 cm long.
The  spectrophotometer  used  is an Hitachi Model 101.  Calibration is accom-
plished using a  primary  solution of 0.1N H-SO^ (Fisher)  and diluting to cover
the  range  10 N  to 10   N in five steps.  Five  readings are taken at each
concentration, and  the data are fit to a linear  function by the method of
least squares.   The  calibration is done once  for each  set  of  samples or at
least once each  day.  Quality control  solutions  are  run  for each calibration
curve.  Analysis of  data shows the minimum detectable  level for this method
to be 5 neq/ml.   Concentrations can be determined to within +_ 5 percent.

Ion  Exchange Chromatography

     The  second  method used to determine sulfate is  ion  exchange chromatog-
raphy.  The chromatograph used is a Dionex Model 14.  If the  sample has been
extracted  using  an acid, it is spiked  with a  base ( .003M NaHCO. + 0.02 M Na.CO-)
before  running  on the chromatograph.  Calibration is accomplished by mixing
sodium  salts (Fisher)  with a buffer.  Retention  times  and  peak areas are ob-
tained  from these standards, and a calibration curve is  drawn. Analysis of
sulfate data shows the minimum detectable level  to be  10   neq/ml.   Concen-
trations  can be  determined to within +^ 10 percent.

     Several SO- collection filters have been analyzed using  the above proce-
dure.   Table 1  shows the results for sulfur determination  for three SO- collec-
tion filters by  X-ray fluorescence analysis (XRF) and  subsequently  by extrac-
tion of  the filter and ion exchange chromatography (1C).
                                     TABLE 1

                   Comparison of XRF and 1C for Sulfur Analysis
Sulfur (ug)

     Chemical analysis  of S  on  the H_S  filters  can be carried out using ion
                                  = fc                       at
exchange chromatography, but the S  must  be  converted to SO,.  We are currently
investigating conversion procedures—in an initial experiment,  we used a
solution of H-O- for extraction and conversion,  and observed a peak corre-
sponding to SO? of about the expected size on the  chromatograph.   A major
interference occurs, however, because NO- from  the H_S collection solution
elutes from the column  just  before the  appearance  of  SO,,  and the tail from
      —                                                H           =
the NO, peak interferes with the quantitative determinations of SO, •   This
interference could be avoided by using  a  different acid  and  copper salt (e.g.,
Cu acetate and acetic acid)  in  the collection solution.

Flash Pyrolysis with Flame Photometric  Detector

     A third wet chemical analysis for  S  is  by  sudden (flash) pyrolysis of
the extraction solution on a heated metal strip, and  detection  of S in a flame
photometric detector.  Initial  experiments using the  extraction solutions
from H2S filters indicated the  feasibility of this technique, but it  has not
been developed further.

Summary of Chemical Analysis of Filters

     Sensitivities of these  various analyses for S  are shown  in the following
table, which assumes extraction of the  filters into 10 ml  of  solution.   The
following amounts can be determined to jh 20 percent or better:

     Brosset-Thorin           800 ng
     ion chromatography         16 ng
     pyrolysis-FPD            320 ng

     We have designed a filter holder that is based on two principal criteria:
1) maximum simplicity and 2) minimum expense.  An exploded view of the TFP
is shown in Figure 1.  The filter frames are aligned by the four bolts which
hold the assembly together.  The functions of the "0" rings as shown in Fig-
ure 1 are to hold the filter in the frame and to provide a pressure seal to
the next filter in line or to the holder assembly.  It is sufficient to turn
the screws down finger tight to provide an adequate seal.  In Figure 2 we show
a schematic drawing of a tandem filter package mounted on a pump assembly.
In this arrangement, the flow rate is controlled by a critical (limiting)
orifice.  Although these orifices provide flow control in a very simple way,

                                                              FILTER ALIGNMENT
                   AIR STREAM
                                 \ VfllM

                       THE FILTER IN THE FILTER FRAME
FIGURE 1.   Tandem Filter Pack,  exploded view.
                                       1 - 1 - LV
             MOUNTED •*.
                            s / / / / / *
                                                         RAIN SCREEN
                                                        FILTER HOLDER
                                                      "O" RING
                                                                     ,   PUMP
 FIGURE 2.  Tandem  Filter Pack and  pumping  system.

they have the disadvantage of requiring applied vacua of 300 and 550 Torr
for 1.0 liter/minute (1/mln) and 14 1/mln orifices,  respectively.  A simple
alternative to the critical orifice is a calibrated  rotameter (flow meter)
with a limiting needle valve.  Our experiments indicate that the filter
frame-filter "0" ring assemblies will not tolerate flow rates greater than
2 1/min (filter face velocity of 250 cm/min).   At higher flow rates, the
filters pull away from the frames.  This can be avoided by using an adhesive
to seal the filters to the frames.  We are now investigating the use of Teflon
filters bonded to the frames.

 1.  Sensenbaugh, J.D., Hemeon, W.C.L.   A Low-Cost  Sampler  for Measurement
     of Low Concentration of Hydrogen Sulfide.   Air Repair  4(1):1-4,  1954.

 2.  Lorenzen, J.A.,  Environmental Monitoring  Device  for X-Ray Determination
     of Atmospheric Chlorine, Reactive Sulfur,  and  Sulfur Dioxide.  Adv  in
     X-Ray Anal 18:568-578, 1975.

 3.  Okita, T., Lodge, J.P., Jr., Axelrod,  H.D.   Filter Method for  the Meas-
     urement of Atmospheric Hydrogen Sulfide.   Environ Sci  Technol  5(6):
     532-534, 1971.

 4.  Natusch, D.F.S., Konis, H.B., Axelrod,  H.D., Teck, R.J., Lodge,  J.P., Jr.
     Sensitive Method for Measurement of Atmospheric Hydrogen Sulfide.  Anal
     Chem 44(12):2067-2070, 1972.

 5.  Axelrod, H.D., Hansen, S.G.   Filter Sampling Method for Atmospheric Sulfur
     Dioxide at Background Concentrations.   Anal Chem  47(14):2460-2462, 1975.

 6.  Pate, J.B., Lodge, J.P., Jr., Neary,  M.P.   The Use of  Impregnated Filters
     to Collect Traces of Gases in the  Atmosphere.  II.  Collection of Sulfur
     Dioxide on Membrane Filters.  Anal Chim Acta 28:341-348, 1963.

 7.  Matsuda, Y., Mlzohata, A., Mamuro,  T.   Simultaneous Analysis of  Gaseous
     and Particulate Sulphurs in the Atmosphere  by  X-Ray Fluorescence Spec-
     trometry.  J Japan Soc Air Pollution 9(2):190, 1974.

 8.  Amaya, J.  Simple Methods of Measuring  Air  Pollutants  by Mini  Samplers,
     J Japan Soc Air Pollution 9(2):192,  1974.

 9.  Johnson, D.A., Atkins, D.H.F.  An  Airborne  System for  the Sampling and
     Analysis of Sulfur Dioxide and Atmospheric  Aerosols.   Atmos Env  9(9):
     825-829, 1975.

10.  Liyv, R., Ott, R., Luyga, P., Pikkov, V.  Cumulative Measurement of Sul-
     fur Dioxide and Fluorides in Ambient Air.  Izv Akad Nauk Est SSR, Khim
     Geol 23(3):208-213, 1974.

11.  Adams, D.F.,  Bamesberger, W.L., Robertson, T.J.  Analysis of Sulfur-
     Containing Gases  in the Ambient Air Using Selective Pre-filters and a
     Microcoulometric  Detector.  J Air Pollut Control Assoc 18:(3):145-148,

12.  Chamberland,  A.M., Bourbon, P., Malbosc, R.  Collection of Atmospheric
     Sulfur Dioxide on Impregnated Fiberglass Filters and Colorimetric Meas-
     urement.  Int J Environ Anal Chem 2(4):303-311, 1973.

13.  Huygen,  C.  The Sampling of Sulfur Dioxide in Air with Impregnated Fil-
     ter Papers.   Anal Chim Acta 28:349-360, 1963.

14.  Huygen,  C.  The Sampling of Hydrogen Sulfide in Air with Impregnated
     Filters.  Anal Chim Acta 30:556-564, 1964.

15.  Parker,  R.D., Buzzard, G.H., Dzubay, T., Bell, J.P.  A Two Stage Res-
     pirable  Aerosol Sampler Using Nuclepore Filters in Series. Atmos Env
     11:617-621, 1977.

16.  Liu,  B.Y.H.,  Kuhlmey, G.A.  In:  X-Ray Fluorescence Analysis of Environ-
     mental Samples (Dzubay, T., ed.).  Ann Arbor, Mich., Ann Arbor Science
     Publishers, Inc., 1977, pp. 107-119.

17.  Parker,  R.D.  A Fundamental Study of Particle Deposition onto Large
     Pore  Nucleopore Filters.  Ph.D. Dissertation, Duke University, Durham,
     N.C., 1975.

18.  Goulding, F.S., Jaklevic, J.M., Loo, B.W. Private communication.

19.  Stevens,  R.K., Dzubay, T.G., Russwurm, G., Rickel, D.  Sampling and
     Analysis of Atmospheric Sulfates and Related Species.  Presented at In-
     ternational Symposium on Sulfur in the Atmosphere, Dubrovnik, Yugoslavia,
     September 7-14, 1977.

20.  Persson,  G.A. Automatic Colorimetric Determination of Low Concen-
     trations of Sulfate  for Sulfur Dioxide in Ambient Air.  Air and Water
     Pollut Int J  10:845-852, 1966.

21.  Brosset,  C.,  Andersson, K., Fern, M.  The Nature and Possible Origin
     of Acid  Particles Observed at  the Swedish West Coast.  Atmos Env 9:
     631-642,  1975.

22.  HM Factory Inspectorate:  Methods for the Detection of Toxic Substances
     in Air.   Her  Majesty's Stationary Office, London, P.O. Box 569, SE19NH.

23.  John, W., Reischl, G., Goren,  S., Plotkin, D.  Anomalous Filtration of
     Solid Particles by Nuclepore Filters.  Atmos Env 1555-1557, 1978.

24.  Turner, W., Spengler, J., Colome, S., Dockery, D.   Design and Performance
     of a Reliable Personal Monitoring System for Respirable Particles*
     Proceedings of the Symposium on the Development and Usage of Personal
     Monitors for Exposure and Health Effect Studies, Chapel Hill, N.C.,
     January 22-24, 1979.  (EPA symposium.)

25.  This work was developed in the laboratories of Northrop Services,  Inc.,
     by Russwurm, 6., Kistler, C., Stikeleather, J.

26.  McClenny, W.A., Shaw, R.W., Baumgardner, R.E., Paur, R.J.,  Coleman, A.,
     Braman, R.S., Ammons, J.M.  Evaluation of Techniques for Measuring Bio-
     genie Airborne Sulfur Compounds.  EPA-600/2-79-004, January, 1979.

27.  Parker, C.D., Lee, M.B., Sharpe, J.C.  An Evaluation of Personal Sampling
     Pumps in Sub-Zero Temperatures.  NIOSH Report 78-117.


Robert W. Shaw, Ph.D., and Robert K. Stevens
Inorganic Pollutant Analysis Branch
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina  27711
     TRAYNOR:  Gregory Traynor, Lawrence Berkeley Laboratory.  You  said  that
not having a fine cut point may be a disadvantage in  the  tandem  filter.
Wouldn* t it be an advantage since your nose and  lungs  don't  have a  fine  cut

     SHAW:  Well, you could take that point of view.   And  in fact there  was
a big discussion recently at EPA having to  do with  the question  of  what  sort
of samplers should be used in order to determine inspired-respired  particu-
lates in the atmosphere.

     I guess that my understanding of the discussion  is that because people's
lung capacities are so different, and because it makes a difference whether
you are mouth-breathing or nose-breathing,  and your level  of ventilation,
that the inspiration curves are quite different  between individuals.  It
may be better to settle on some precise separation  scheme  rather than trying
to limit the human lung.

     TRAYNOR:  There is something that we do know.  We know  that the human
lung is not precise.

      SHAW:   Yes.   Well,  I can't say any more than I did.   There Is no average
human lung,  so to speak.  If there were, then you would want to repeat the
behavior—no doubt about it.  But since different lungs act so differently,
the  decision has  been made to separate the particles according to their two
modes.  There is  more information on that.

      GUDAP:   Bill Gudap, Research Triangle Institute.  Bob, do the essen-
tially damp  or wet filters cause any problems?

      SHAW:   No, because the layer is very thin.  So, they go right through
the  X-ray spectrometer with no trouble.  Actually, water  doesn't adsorb X-rays
very strongly.  And so that thin layer doesn't cause any  trouble.

      BROOKS:  Joe Brooks, Monsanto Research Corporation.   I have a question
concerning  the pumps.  Do you know what kind of pressure  drops these pumps
will tolerate?

      SHAW:   Yes.   The Brailsford can support 50 mm of Hg, pressure drop at
about 4 liters a minute.  We use it to pump across some sampling devices which
have a much bigger pressure drop than this.  This device  has a very small
pressure drop.  I don't know what it is, but it is much less than 50.

      WHITE:   Otto White, Brookhaven Laboratories.  I am not sure that I under-
stand the comment about the comparison between the radiation exposure where
you  wouldn't see the data for several days.  I think my colleagues might
be at odds with me if I didn't comment, but there are pocket dosimeter devices
which can alarm.   I think In this particular area—at least in the industrial
area—that you have similar type instrumentation where you want some type of
alarming device—such as the CO monitor talked about in a paper presented

      And you may have an accumulative device.  So, I am not sure that I un-
derstand your meaning.

      SHAW:   Yes;  I remember from the old days that when I would walk out of
the  room where the scattering chamber was, I would occasionally wonder how
much exposure I had gotten while I was In there.  We didn't have the kind
of dosimeters that you are  talking about.

      I was just trying to make a point that this particular system requires
that it be sent back to the lab for analysis.

      One could develop chemically treated filters that would respond color-
imetrically so that you could compare right on the spot with a prepared scale.
But  the system that we have now doesn't do that.

      WHITE:   I think both types of systems are appropriate.  And whether or
not  you want that type of immediate response depends upon the need for a gross

      ZISKIND:  Richard Ziskind, Science Applications.  You are going to get

sulfate dioxide converted to sulfate.  You are also going to get sulfate from
the air.  You can't discriminate between those, can you, with the dosimeter?

     SHAW:  If you remember, the first couple of filters took out the par-
ticles and let the gases go through.  And the next filters down the line
took out the gases according to their chemistry.  So,  that is how you get
the separation.

An Evaluation of Personal Sampling Pumps
in  Sub-Zero Temperatures
     Carl D. Parker and Joan C. Sharpe
     Research Triangle Institute
     Research Triangle Park, North Carolina

     The development of the large oil reserves along the northern slope  of
Alaska's Brooks  Range and nearby Prudhoe Bay  typifies the industrialization
currently taking place in cold environs.  These developments—precipitated
by America's growing dependency on foreign oil, with its inherent threat of
crippling embargoes—are taking place in a severe environment which tests the
endurance of both men and machines.  The growth of industrialization in  these
and other cold regions has generated a growing need for air sampling instru-
mentation and methodology suitable for industrial hygiene investigations in
sub-zero temperatures.  With a view toward these needs, the authors have in-
vestigated the suitability of currently available personal sampling pumps
for use in cold  environs; i.e., to temperatures as low as -50° C.  This  paper
is a brief presentation of the results of these investigations.  A detailed
description of these investigations is available in the Project Final Report

     The pumps evaluated were tested at temperatures between 25° and -50° C.
While many environmental factors or stresses  act in synergism with tempera-
ture, most are of significance only at high temperatures and were not incor-
porated into this experimental program.  An environmental factor which does
have an important synergistic relationship with temperature at low tempera-
tures is humidity.  Absolute humidity is low  in cold environs; thus, air
sampled during these tests was obtained from  a dry air manifold.

      The personal sampling pumps evaluated can be categorized into two differ-
 ent groups.  One group is the Coal Mine Dust Personal Sampling Units (CMDPSU),
 for which a certification program has been defined (2).  In order to be cer-
 tified, the CMDPSU pumps must be capable of operating from their internal
 battery packs at a nominal flow of 2 liters-per-minute against a resistance
 of 4 inches of water for not less than 8 hours, presumably at room tempera-
 ture.  (There are other requirements.)  This performance requirement was
 used as a test condition for the evaluation tests described herein.

      The second class of pumps is used with a variety of sample collection
 devices under a variety of conditions.  This suggests—and it is true—that
 these pumps, in the aggregate, provide industrial hygienists a wide range of
 sampling capability.  A given pump, in contrast, may only function over a
 narrow region of the aggregate range.  Thus, it is impractical to define a
 single set of test conditions which will be meaningful in terms of the capa-
 bility of all of these pumps.  The practice adopted for these evaluations
 was to test each pump under a set of conditions which was somewhat representa-
 tive of each pump's capabilities as stated by the manufacturer.  Consequently,
 the data reported herein are not intended to be used to compare one pump
 against another.  Instead, these data are intended only to demonstrate how
 each pump's operation is affected by a low-temperature environment.

      Since it was the rapid industrialization of the Alaskan oil fields that
 largely precipitated these studies, it is of interest to briefly characterize
 the Alaskan environment.  North and central Alaska are divided into two cli-
 matic divisions by a 600-nile stretch of low but rugged mountains known as
 the Brooks Range.  This expanse of mountains forms a natural partition be-
 tween the Interior Basin to the south and the Arctic area to the north.  The
 Alaskan oil fields are located along the harsh, sub-zero environs of the
 north slope of the Brooks Range; this is an area of primary concern.

      The north slope of the Brooks Range is located in one of four extremely
 cold winter regions.  (The others are Greenland, northeast Siberia, and the
 Antarctic.)  In this region, the average temperature in January is -24° C,
 and in July is 8° C.  The temperature in some areas is below -40° C at least
 10 percent of the time in the coldest month.  The diurnal cycles are very
 weak, and very low temperatures can persist for several days.  At Pt. Barrow
 in the Arctic area, for example, temperatures below -34° C have persisted

for as long as 12 days, and temperatures below -40° C have persisted  for  2
days.  Temperatures of approximately 24° C have been observed at Pt.  Barrow
in midsummer, but the average midsummer temperature at Pt. Barrow is  only
2° C.

     While extremely low temperatures largely characterize the Alaskan environ-
ment, there are other features which may also affect the responses of personal
sampling pumps.  The absolute humidity, for example, is very low in extremely
cold environs.  (The relative humidity is usually very high.)  The average
precipitation is low, but it is usually in the form of solid precipitates.

     Most of the low-temperature tests described were conducted in a test
chamber which was cooled with liquid nitrogen.  Figure 1 is a schematic of
the basic test facility.  A pump under test sampled air metered from a dry
air manifold.  The sample air passed-through a heat exchanger inside the test
chamber so that it was also cooled to the test temperature.  The pressure
drop across the pump was measured with a water manometer located outside the
test chamber.  The sample air was exhausted inside the chamber at the test
temperature.  As illustrated, provisions were made in each pump to substitute
a power supply for the pump's batteries when the batteries became a perform-
ance-limiting factor.  The volumetric change in the air mass between the roto-
meter (at room temperature) and the pump (at test temperature) is accounted
for in the data presented herein.  Corrections for pressure variations were
small and are neglected.

     In the following paragraphs, some typical test results are presented
for the pumps that performed best.  The reader^is cautioned that there is
some variance between pumps of a given model and some day-to-day variance
for a single pump.  These data are presented as being somewhat typical al-
though they are from a single unit of each pump.  As many as five units of each
pump were tested.

Mine Safety Appliances Company (MSA), Model G

     The Model G is a certified CMDPSU pump and was tested with CMDPSU re-
quirements in mind.  Figure 2 is a data plot that illustrates the flow rate

                         BATTERY / PUMP
 FIGURE 1.   Schematic of the basic test facility.
 stability of the Model G over an 8-hour period.   The  pump  "load" or  pressure
 drop was approximately a constant and was 4 inches  of water.  During these
 tests, the flow rate was sometimes adjusted to a maximum,  and a power supply
 was sometimes substituted for the pump's battery pack.  These events are  clear-
 ly marked in Figure 2 and in other data plots  which follow.

      The data in Figure 2 show a marked flow rate dependency on temperature.
 They show a significant pumping capability at  -11°  C  and even at -22° C with-
 out any adjustments.  All were marginal at -30°  C,  and  none were even margin-
 ally satisfactory at temperatures below -30° C.

      In Figure 3, flow rate is shown as a function  of pressure differential
 across the pump with temperature as a parameter  for a different Model G.
 These data were acquired after the pump equilibrated, nonoperating,  at the
 test temperature for 1.5 hours.

                •+t*C  A -If »C

                0-8*0  X-80»C

                                                                M*A HOOCL 0


                                                                       o .i«c
FIGURE 2.   Flow rate stability, corrected  for temperature, Model G.

                                                                  MSA, Model G



                                 0*C CORRECTED

                                     » 24* C
                                       (  In. of H20 )

FIGURE 3.  Flow rate versus pressure differential,  Model G.

 Mine Safety Appliances Company (MSA), Model S

      The characteristics of  the MSA Model S did not differ significantly
 from the Model G, and, consequently, no Model S data are presented.  Both the
 Model G and Model S performed well as compared to other CMDPSU pumps tested.

 Research Appliance Company (RAC), Model 2392PS

      The Model 2392PS personal sampling pump is configured for use as a 2 1/min
 pump; i.e., it has a rotometer calibrated between 1.6 and 2.0 1/min with a
 resolution of about 0.1 1/min.  Consequently, it was tested similarly to the
 CMDPSU units.  The RAC unit  also has a similar series arrangement of rotometer,
 control valve, damper, and pump as the CMDPSU units.

      Figure 4 is a plot of the Model 2392PS measured flow rate, corrected for
 the  temperature, as a function of both time and temperature.  As illustrated,
 the  room temperature performance was excellent.  At 0° C, the flow rate was
 significantly reduced but was stable over the 8-hour period.  At -10° C, its
 performance was marginal, and both a flow rate adjustment and a power supply
 substitution were made in an attempt to maintain a satisfactory flow rate.
 At low temperatures, flow rate adjustments have little effect on the Model
 2392PS, and the power supply substitution was also of marginal value.  At -20° C,
 a flow rate adjustment at 1.5 hours had no effect at all.  The pump continued
 to operate from its battery  pack at about 1.4 1/min (measured at room tempera-
 ture) for about 5.5 hours.   At -30° C, the flow rate was adjusted to a maximum,
 and  a power supply was substituted for the battery pack after 1.5 hours; thus,
 a flow of about 1.2 1/min measured at room temperature was maintained.

      The flow adjustment on  the RAC unit is best described as a pinch or clamp
 mechanism that clamps and restricts the sample flow tubing.  At low tempera-
 tures the tubing tends to remain '"set," and releasing the clamp mechanism
 does not provide positive control over the flow.

      Figure 5 is a plot of the flow rate of the RAC unit as a function of
 the  differential pressure across the pump with temperature as a parameter.
 The  pump was initially adjusted for a flow of 2 1/min at 4 inches of water
 at 21° C.  As with the Model G, it was equilibrated, nonoperatlng, at each
 test temperature for about 1.5 hours.

      The data in Figure 5 includes 21°, 0°, -10°, and -20° C curves.  Curves
 corresponding to both the measured flow rates (measured at room temperature)
 and  curves corrected to correspond to the test temperatures are included.



                              (I)  ADJUST FU)W TO MAXIMUM
                              (2)  SWITCH FROM BATTERY TO POWEfl SUPPLY

                                                        o   o—o——-o    o  0* C
                                           »«^ f 121
                                      4        8

                                   TIME (HOURS)
      Flow rate stability, corrected  for  temperature, RAC  Model  2392PS,
                                                          o    o°C
                                                          o    0° C Corrected

                                                                  C Corrected
                                                              20" C Corr«ct«d

                              AP  (In. H20)

      Flow rate  pressure differential,  RAC Model 2392PS,

 Mine Safety Appliances Company (MSA). Model C-200

      The MSA Model C-200 personal sampling pump is designed for a flow range
 of 25 to 200 ml/min through a flow resistance of up to 2.5 inches of H20.
 It is nominally adjusted by the manufacturer for a flow rate of 200 ml/min
 at 1.5 inches of H.O, and most tests were conducted at this setting.  The
 pump incorporates a counter which indicates the number of pump strokes—thus,
 the volume of air pumped.

      Without altering the flow rate adjustment as received from the manufactur-
 er, the Model C-2001s were operated for 8 hours against a flow restriction of
 1.5 inches of lUO at several test temperatures.  The results of these tests
 for a typical pump are plotted in Figure 6.  These curves show generally
 excellent performances at room temperature at -2° C.  At -10° C, the perform-
 ance is marginal.  In other units, the 10° C performance was unsatisfactory.
 In no units were the performances acceptable at temperatures lower than -10° C.

      A significant factor in the C-200's failure to perform satisfactorily
 at low temperatures is that a mercury (Hg) reference cell is used in the pump's
 electronics.  Two voltages are available from the Hg cell, and these were ob-
 served to decrease significantly with a decrease in temperature to limit low-
 temperature performances.

      The flow rate versus pressure-drop characteristics of the Model C-200
 are shown in Figure 7 for room temperature and 0° C.  The flow rates com-
 puted from the counter are also included.  The constant flow rates indicated
 by the counters show the pump speed to be constant and indicate that a very
 stable constant flow rate can be anticipated for a given temperature and
 pressure differential.  The measured flow rates at 25° and 0° C are observed
 to vary reasonably linearly with pressure differential.

 DuPont Model P-125

      The DuPont Model P-125 sampling pump incorporates an electronic flow
 controller that senses the air flow and controls the speed of the pump motor
 to achieve a constant flow.  Other distinctive features include a battery
 charge status indicator, and a flow monitoring indicator that trips and latches
 if the flow is interrupted or cannot be maintained near the calibrated value
 for a period of time.  The P-125 is specified as adjustable to flow rates
 of 25 to 125 ml/min with a pressure drop of up to 25 inches of H^O.  Signifi-
 cantly, the P-125 is specified for an operating temperature range of -7° to
 49° C.

           -32 °C
       0    X    I
                                           4        5

                                        TIME (HOURS)
FIGURE 6.  Flow rate stability,  corrected  for temperature, Model C-200.
                                                         • +-25° C
                                                           Computed from Pump Counter
                                         AP (In. of  H20)

FIGURE 7.  Flow rate versus differential pressure, Model C-200,
                                                                              0" C

     The flow rate stability of the P-125's was measured at 100 ml/min and
10 inches of H20; i.e., somewhat near  the middle of the specified capability.
Data from a typical unit are shown in  Figure 8.  As with the other pumps, flow
rate measurements were made at room temperature and corrected for the differ-
ence between room and test temperatures.  The performances at room temperature
are notably stable.  They are also generally good at a nominal 0° and -22° C.
At -22° C, the  temperature-corrected flow rates tended to be high by about 15
to 25 percent,  but they were stable over an 8-hour period.

     It is notable that three of the five P-125's maintained a useful flow
rate at -30° C  for an 8-hour period, and that useful flow rates could be main-
tained at -30°  and -40° C by substituting a power supply for the battery pack.
Thus, it is evident that the battery pack is a principal performance-limiting
factor in the DuPont P-125 at -30° C and -40° C.  Therefore, useful sampling
at these temperatures may be possible  with a larger battery pack.

     The flow rates of a P-125 as a function of pressure differential, with
temperature included as a parameter, are shown in Figure 9.  In the aggregate,
these data show remarkably linear characteristics between +25° and -20° C.
These results are further evidence of  the P-125's generally excellent perform-
ance at temperatures as low as -20° C.

Accuhaler Model 808

     The configuration of the Model 808 differs significantly from the other
pumps evaluated.  It utilizes a limiting orifice that attaches integrally
to the pump to  control the flow rate.  The pump motor drives a pin-cam mechanism
that, in turn,  initiates the following cycle of events.  An exhaust valve to a
pump cavity opens, and a diaphragm (i.e., a piston with a flexible diaphragm
seal) is driven into the pump cavity,  causing it to exhaust.  At maximum
compression—i.e., minimum cavity volume—the exhaust valve closes, and the
piston/diaphragm drive mechanism springs back to a null position, opening the
motor circuit.   The piston/diaphragm mechanism, driven by a spring, returns
to an extended,  maximum-volume position as air bleeds into the cavity through
the limiting orifice.  At the extended, maximum-volume position, the motor
is re-energized to repeat the cycle.   Each cycle pumps a volume of air equal
to the displacement of the piston-diaphragm assembly, and the cycle rate is
largely controlled by the limiting orifice.  Several sets of orifices are
available, and  the orifice is readily  changed with a wrench or nutdriver.

     The flow rate stability of the Model 808's was measured using the test

                                  (I) SWITCH FROM BATTERY TO POWER SUPPLY

                                              DUPONT P-129

                                                                                -X -30 °C
.0 -2 *C
                                           4         5
                                        TIME (HOURS)
 FIGURE 8.  Flow rate  stability,  corrected  for temperature,  Model P-125.
                                                    DUPONT  P-125
    200 -
                                     -20° C
                                     -20° C Corrected
     50 -
A •"" "^^ ™*A ^^^^^ *£*™ " *"*JJ,****™^ o •«•
0 0 13 ft T

i i
A^*"^^» * 	 M

0°C Corrected ~ ~"^~S
•"" ' •*"~~»"~~" »— —• •• •" ••• I.,.. ^ ..•-^-^».
*25° C
1 ( i 1 » i
68 10 12 14 16
                                           AC (In. of H/»

 FIGURE 9.  Flow rate versus differential  pressure, Model P-125,

 apparatus  illustrated  in Figure 10.   (The sample flow through  the Model  808
 pulsates and  cannot  be measured with a rotometer.)   The  restricting  valve  was
 set  independently to drop 1.5 inches of H_0  at  a flow rate  of  100 ml/min.
 Sample  air was drawn through the restricting valve  from  the dry air  manifold.
 When flow  measurements were made,  the sample was temporarily drawn from  the
 soap bubble meter.

       The limiting orifice that attaches to the  pump is the  only flow control
 option  available to  the user.  For a given orifice, the  stability of the pump
 is likely  to  be enhanced if the pressure drop across the  orifice is  larger than
 the  drop across the  sample collection medium (i.e., the  pressure regulatory
 valve in Figure 10.)

       The flow rate stability data  for a typical Model 808 with a 100 ml/min
 orifice is shown in  Figure 11.  (The flow rates were measured  at room  tempera-
 ture, but  the values plotted are corrected to the respective test temperature.)
 At 25°  C,  the flow rates are reasonably uniform and tend  to be within  about
 5  to 10 percent of 100 ml/min.  The pumps did not operate for  an 8-hour  period
 at 100  ml/min.  It is  clear that a lower flow rate  would  tend  to extend  the
 operating  period significantly.  The pump operates  on a  periodic basis—a
 quasi "duty cycle"—in that the battery circuit is  closed and  the motor  cycles
 the  cam.   During the "off" period, there is  no  load on the  batteries.

       All of the pumps evaluated were powered  by  secondary  (i.e., rechargeable)
  sealed, sintered plate nickel cadmium (Ni-Cd)  batteries.   These batteries
  have  very low internal resistance and can deliver  high currents with little
  loss  of voltage.  A very significant feature  is  that when  delivering moderate
  currents, they will perform satisfactorily at  very low temperatures.  These
  cells can supply useful but reduced energy at  temperatures as low as -40° C (3).

       The capacity of a Ni-Cd battery depends  largely upon  the manner of battery
  operation.   Cells are usually characterized at 25° C and at a stated discharge
  rate; e.g., C/5 or C/10, where C is the  rated  ampere-hour  capacity.  The actual
  capacity is a function of the discharge  rate  and tends to  decrease signifi-
  cantly with an increase in the discharge rate.  A  cell characterized at a C/5
  (5 hour) rate will yield about 80 percent of  rated capacity at a C/l (1 hour)
  rate, and about 108 percent of rated capacity  at the C/10  rate.  The cell
  capacity is also influenced by temperature. At  0° C, for  example, a Ni-Cd
  cell  will deliver approximately 90 percent of  its  room temperature capacity (3).


                          TEST  CHAMBER
                        SOAP BUBBLE
                        FLOW METER
FIGURE 10.  Test apparatus for the Accuhaler Model 808.

          MDA ACCUHALER 808
                                 3         4
                                  TIME (HOURS)
FIGURE 11.  Flow rate stability, corrected  for temperature, Model 808.

      There  is  very little data available that  is  descriptive  of  the  reliability,
 i.e., cycle-life,  of  Ni-Cd cells at low temperatures.   Data that are available
 show a  seriously reduced cycle-life at -18°  and -34° C; however, these  results
 are attributable to the establishment of unsatisfactory operational  conditions
 rather  than any failure to the cell itself (3).   The reader should be mindful
 that many factors other than temperature influence  the  cycle-life and perform-
 ance of Ni-Cd  batteries.  These include use  factors such as routine  depth
 of discharge,  recharge practices, and discharge rates.

      The results of these evaluations suggest  that  the  Ni-Cd  batteries  used
 in personal sampling pumps yield good performances. While they  are  inadequate
 for the desired performances at low temperature,  the problem  is  due  largely
 to pump characteristics.  Indeed, battery capacity  is reduced at low tempera-
 ture; however, the pumps demand more current.  Thus, the batteries must dis-
 charge  at the  higher rate, which further reduced  their  capacities.

      The battery pack should be used with care regardless of  temperature.
 Recharging should be carried out with the charger supplied by the manufacturer
 according to instructions and at room temperature.  In  lieu of this  option,
 the batteries  should be charged at a constant  C/10  rate for about 15 hours,
 or at a C/20 rate for much longer periods.  (C is the ampere-hour rating of
 the battery.)   The battery should not be recharged  unless a significant portion
 of its  capacity has been dissipated; e.g., C/3 ampere-hours.

       All of the personal sampling pumps evaluated  are  sensitive  to  temperature
  changes between 25° C and -50° C.  There is considerable variance in  the  sensi-
  tivity of the different models evaluated,  and some variance within  a  single
  model.  All of the pumps evaluated are useful sampling instruments  at 0°  C,
  and  none are functional at -50° C.  Between these  extremes, the  different models
  are  affected differently by temperature.  From among these pumps, the industrial
  hygienist can select a pump suitable for most any  application to a  low tempera-
  ture of -20° C.  At lower temperatures, however, the options are severely
  limited.  For the pumps evaluated, reasonable low-temperature limits  are  tabu-
  lated as follows:

        MSA Model G                          -20
        MSA Model S                          -20
        RAC Model 2392PS                       0
        MSA Model C-200                        0
        DuPont Model P-125                   -30
        MDA Model 808                        -20

     It should be emphasized once again that a comparison of the pumps evaluated
was not an objective of this effort.  The pumps are rated or specified dif-
ferently by their manufacturers, and, consequently, comparison is somewhat un-
fair.  Moreover, they were tested differently.  This is especially true of
the pumps which were not tested as CMDPSU units.  However, there is an obvious
quality of construction and room-temperature performance about some of these
pumps which should be acknowledged.  The MSA Models G and S, the MSA Model C-200,
and the DuPont P-125 perform well at room temperature and give an impression
of quality construction.  The Model 808 has a different operating mechanism
which could have some advantages.  It should, for example, be relatively in-
sensitive to battery status until the pump fails to operate at all.  The Model
2392PS is similar in many respects to the Models G and S.  It is lighter, has
a smaller battery pack, and performs well for 8 hours at room temperature.

     One objective of the evaluation effort was to evolve a set of recommended
standards for personal sampling pumps for use in cold environs.  We concluded,
however, that the personal sampling pumps evaluated were not designed for use
in sub-zero environments.  Instead, they were designed and optimized for use
in temperatures near 25° C.  We conclude further that, in response to the
solicitation for pumps for this evaluation, standard models were submitted
without significant changes to optimize these units for low temperatures
because of insufficient time and incentive.  Consequently, standards based
upon the results of these evaluations will tend to be premature and somewhat
pessimistic.  Standards recommended in the Final Report should be considered
as preliminary, subsequent to further investigations and developments (1).

     This study would not have been possible without the cooperative support
of the manufacturers of the pumps evaluated.  These manufacturers made pumps
available for testing at no cost or risk to the authors.

     This study was sponsored by the National Institute for Occupational
Safety and Health through Contract No. 210-76-0124.



 1.  Parker, C.D., Lee, M.B., Sharpe, J.C.  An Evaluation  of Personal Sampling
     Pumps.  Final Report, NIOSH Contract No.  210-76-0124, DREW Publication
     No. (NIOSH) 78-117, September 1977.

 2.  Code of Federal Regulations, Title 30, Part 74.

 3.  Bauer, P.  Batteries for Space Power Systems.   (NASA  SP-172) TRW for
     NASA, Redondo Beach, Calif., 1968.

Carl D. Parker and Joan C. Sharpe
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, North Carolina  27709
     TURNER:   Bill Turner from Harvard.  Were you able to make any pulsation
measurements  in  any of your tests?

     PARKER:   No, we did not.  One of the pumps that we looked at was very
unique  in  that it had a critical orifice that controlled the  flow.  The
pump exhausted the cavity of the diaphragm.  The electrical system became
inoperative at that point.  It came to a stand position, and  then a spring
turned  the diaphragm to its stand position, opening the cavity and drawing
the sample through a critical orifice.

     This  pump had a very unusual pumping mechanism.  It would take the sample,
pull it through  the orifice, and then recycle the diaphragm.  But all these
were tested at relatively stable constant flow rates.

 Electrochemical Air Lead Analysis for
 Personal Environmental Surveys
     Francis J. Berlandi, Ph.D., Gerald R. Dulude, Reginald M. Griffin, Ph.D.,
     and Eric R. Zink, Ph.D.
     ESA Laboratories, Inc.
     Bedford, Massachusetts

     Exposure to lead via  airborne particulates  has become a cause of in-
creased concern in the United States.  The U.S.  Government, via the Environ-
mental Protection Agency,  has recently proposed  an ambient air lead standard
of 1.5 jig/m (1).  The justification for the  enactment of this standard has
been the increased lead  exposure from both mobile and stationary lead sources
by the general populace.   This has been documented by the monitoring of blood
lead levels in both children and adults in high- and low-exposure zones.   The
recommended method of measurement for determining air lead exposure is the
collection of the air particulate material on a  filter medium, as in the
normal TSP methodology.  The lead is then extracted from the filter into
a solution and measured  by a chemical technique.

     Since 1968, the use of anodic stripping voltammetry (ASV) to measure
blood lead levels in both  children and adults has become quite widespread.
Over 20 percent of the medical and clinical  screening facilities in the
United States use anodic stripping voltammetry as their major technique for
blood lead analysis (2).   This paper describes the use of a rapid electro-
chemical technique for the measurement of air lead using ASV.

     The method (3) is summarized as follows. The air filter to be measured
is placed in a test tube or another suitable container.  The filter is soaked
for 10 minutes in a Metexchange® reagent. The lead particulate is eluted
and/or dissolved from the  filter matrix with mechanical agitation.  The sample

 reagent mixture containing both the solution and the filter is placed directly
 on the electrode assembly on the ESA Model 3010A Analyzer.  The analysis se-
 quence is initiated, and, approximately 90 to 150 seconds later, a quantita-
 tive measure of the amount of lead on the filter is available to the operator
 as a digital display.  A permanent record is also available in the form of
 a recorder tracing.

      The basic advantages of the technique include:  a practical lower limit
 of measurability for lead in the 1 to 10 nanogram range; accuracy and preci-
 sion which compares favorably to a standard reference technique; and a conven-
 ience of analysis which is both rapid and simple.

      Anodic stripping voltammetry  is an electrochemical technique involving
 two steps.  The first step  is  the  plating or reducing of metal ions into a
 mercury electrode by holding the electrode  potential at a more negative value
 than is necessary to cause  the reduction of those metal ions of interest*
 The potential of the electrode is  then rapidly and linearly varied in a posi-
 tive or anodic direction.   When the potential of the test electrode reaches
 the formal potential of a metal ion/metal couple, that particular metal will
 reoxidlze back into solution,  and  a current will flow in the measuring circuits
 external to the electrochemical cell.  Because each metal has a unique formal
 potential, a number of different metals can usually be analyzed simultaneously.

      Figure 1 is a simplified  picture of the two-step process called ASV.
 In the first step, the metals  are  being reduced by holding the electrode at
 a controlled potential.  If the metal ions were present in relatively high
 concentrations, it would be possible to quantitatively measure them during
 the reduction step by integrating  the area under the current-time curve dis-
 played in the top half of Figure 1.  When dealing with parts-per-billion
 (ppb) metal ion concentrations, the contribution from reduction of all re-
 ducible metal ions present  to  the  current-time integral is small (as repre-
 sented by the cross-hatched area of the plot).

      The major fraction of  the current-time integral will be due to reduction
 of impurities (such as hydrogen ion) present in excess of the metal ions.  If,
 at the end of the plating step, the potential of the electrode is varied in
 a positive direction, the current  signal obtained will be almost entirely due
 to reoxidation of each reduced metal present in the electrode.  This step is
 depicted in the lower half  of  Figure 1.


i M



Controlled Potential Electr.





Stripping Voltammetry


.•• r
                                   FIGURE 1.  Plating step and stripping
-175 -340 -530
                            FIGURE 2.  ASV scan.

                             matrix:  4F LiCI, 0.5F NaAc
                                 phh  4.5
                          plate pot'I-.  -1100mv
                          plate t ime: 30 minutes
                         sweep rate •. 60 mv  sec
                         metal concentrations 20 ppb
                              Stripping Peak Potentials
                                vs. the Ag /AgCI reference

      When the test electrode is composed of a thin film of mercury deposited
 on an inert substrate, such as the waxed graphite electrode, the result of
 the anodic stripping step will be a very sharp and well-resolved current
 peak for each metal deposited (into the thin film of mercury) during the
 plating step.  A typical ASV scan is shown in Figure 2.

      Peak potentials serve to identify the mercury-soluble metals in the
 sample.  The height of each peak is proportional to the amount of metal ion
 reduced into the mercury and is, therefore, proportional to the amount of
 metal present in the sample.  Figure 3 illustrates the kind of calibration
 curve linearity that is regularly obtained with the composite mercury graphite
 electrode (CMGE).

      Most analytical techniques (4) for air lead analyses require that the lead
 particulate be extracted, dissolved, or totally digested from the air filter
 matrix.  The validity of the rapid, ASV analytical methodology is established
 by comparing the measured lead values from this -new- methodology to a second
 set of values obtained by using an EPA-recommended, acid extraction method
 on the same or duplicate samples.  The EPA methodology is fully described in
 the Proposed Ambient Air Standard for Lead and is compatible with a final lead
 determination by atomic absorption on anodic stripping voltammetry.  At this
 point, we define the ASV wet digestion as denoting an ASV analysis performed
 on conventionally digested samples, whereas ASV extraction will denote the
 rapid and relatively newer ASV techniques using samples treated with Metex-
 change® reagent.

      Since most of the comparison data developed in this study compare the ASV
 extraction technique to the ASV wet digestion methodology (5), Table 1 summa-
 rizes the ASV wet digestion performance on standard NIOSH lead air filter
 sets from 1974 to 1977.  The data demonstrate an excellent agreement of ASV
 wet digestion data with results obtained from a variety of other methods such
 as AAS and dithizone.  Three years of comparison studies on ESA laboratory
 performance on both air filter and blood lead proficiency standards provide
 additional data as to the inherent validity of the ASV wet digestion method.

      The ASV wet digestion methodology is also compared for agreement with AA
 results obtained on duplicate samples of extract from hi-vol filters.  The
 summary of results is shown in Table 2.  Based on the results shown in both

      1200 —
      600 —
      200 —
200     400    600    800    1000
     nanograms jnetal
FIGURE 3.  Typical ASV calibration curves.

                                     TABLE 1

                         NIOSH "PAT" Summary for Air Lead
 Number of value pairs = 64
 Intercept = .00087
 Slope = 1.01140
 Correlation coefficient » .9914
 If = .10057                         "PAT" result average

 Y = .10086	ESA reported average	

                                     TABLE 2

 	Hi-Vol Filter Results ASV Wet Digestion versus AAS	

 Number of value pairs = 20
 Intercept = .235
 Slope « .969
 Correlation coefficient = .998

 X" - 12.65                          ASV wet digestion

 Y = 12.50	AAS	

 Tables 1 and 2, the use of the ASV wet digestion methodology as a reference
 method in the development of the ASV extraction screening approach is techni-
 cally sound.  In addition, it can be shown that the ASV extraction results
 can be compared to either AAS or ASV wet digestion-generated data to establish
 the validity and accuracy of the methodology.


      The previous section has established the technical basis for the use of
 ASV wet digestion in lead analysis for those not familiar with the electro-
 chemical methodology.

      This section summarizes the data obtained in the direct comparison of
 the ASV extraction air filter technique to both EPA filter standards and ac-
 tual field samples.

      Recovery of lead on air filters with this direct ASV extraction method
 was checked using EPA standard air filters against the atomic absorption tech-
 nique.   Results are shown in Table 3.

                                     TABLE 3

                           EPA Filter Strip Comparison
EPA STANDARD  FILTER                ASV  EXTRACTION                 AA

   ug Pb/filter	ug  Pb/fllter	ug  Pb/filter
       0                                  0                        0.1
     115                                108                      110
     300                                284                      270
     400                                409                      365
     675                                695                      595
                                    TABLE 4

	Industrial Environment Filter Study
Number of pairs «=  12
Slope -  .984
Intercept »  .96
Correlation coefficient »  .999

IT « 73.17                          ASV wet digestion

Y • 72.95	ASV extraction	

                                    TABLE 5

                            Oxide Dissolution Study
1 mln
3 min
6 min
10 min
15 min
     Also, air filter pairs were checked from an industrial collection site
against total digestion procedures.  Duplicate air particulate samples were
collected from each of several work environments.  In each case, the dupli-
cate pairs were collected simultaneously with equal collection efficiencies.
One filter was analyzed using the direct ASV technique, and one was analyzed
using a total digestion ASV technique.  Results are shown in Table 4.

     Dissolution studies on the more common oxides of lead that could occur on
the air filter matrix were evaluated.  Measured amounts of PbO, Pb02, and Pb30,
were placed in Metexchange* reagent.  The amount of lead dissolved versus
time was measured at various time levels.  Since the particulate size used in

                                     TABLE 6
                         Volumetric Air Lead Calculations
5 min
30 min
60 min
8 hr
0.1 yg/m
Air Concentration (yg Pb/m )
3 33
1 yg/m 10 yg/m 50 yg/m
Filter Filter Filter
100 yg/m
 the study was greater than 50 microns, the results are a conservative estimate
 of the rate of dissolution for typical lead species found on a filter substrate.
 Table 5 summarizes this data.  Dissolution was complete in 6 to 10 minutes.

 Discussion and Results

      The advantage of both the ASV wet digestion and ASV extraction to air
 pollution measurements is their ability to accurately measure lead down to a
 level of 1 to 10 nanograms.  The implications of this statement can be seen in
 the namagraph of Table 6.  Using a sample time of 5 minutes at an air flow of
 2 liters/minute will provide the user with 10 nanograms of collected lead on
 the filter if the air lead level is 1 yg/m .

      The utility of the ASV extraction, rapid lead screening technique is best
 demonstrated in a limited field study conducted using OSHA-type portable air
 samplers.  The study was conducted on December 8 and December 9, 1977.  The
 purpose was to derive data on the relative lead exposure to a typical suburban
 commuter during a typical work day.  A series of samples were collected during
 the work-day period.  The chronology is summarized as follows:

      SAMPLE 1:  Air sampling in suburban industrial facility in Bedford, Mass.,
 adjacent to no known air lead sources except limited vehicular traffic.

      SAMPLE 2:  Air sampling during drive from Bedford to Winchester, Mass.,
 during the evening commuter traffic.

      SAMPLE 3:  Air sampling .during drive from Winchester to downtown Boston
 after major commuter rush hour.

      SAMPLE 4:  Air sampling at a professional society meeting held in a
 second-floor conference room of an office building in downtown Boston.

     SAMPLE 5:  Air  sampling during  drive  from Boston  to Winchester,  Mass.,
in the midevening traffic.

     SAMPLE 6:  Air  sampling during  drive  from Winchester  to Bedford,  Mass.,
on the following morning during  commuter traffic.

     The exposure study results  are  presented in Table 7 and demonstrate
the flexibility of the analysis  to respond to a variety of sampling times.
The data also indicate the consistent air  lead levels accompanying the driving
segments of the day.

     The results demonstrate that the lead exposure on this day during transit
in a vehicle is much higher than the lead exposure measured in the suburban
industrial facility or in the downtown Boston office building.  Plans  are under-
way to extend the effort to monitoring pedestrian and nondriving exposure to
lead while in residence at various types of roadway conditions.  Due  to the
inherent simplicity of the method, the onsite evaluation of lead exposure is
possible using an ASV analyzer and an appropriate power source, although the
data presented here represents laboratory analysis.

     A second series of samples  was obtained on February 28, 1978, and analyzed
using a wet digestion methodology.  The sample summary is as follows:

     SAMPLE A:  Air sampling during drive to Waltham, Mass., from Bedford on
Route 128.

     SAMPLE B:  Air sampling during meeting in office building one-half mile
from Route 128.

     SAMPLE C:  Air sampling during drive to Bedford from Waltham on Route 128.
                                    TABLE 7
                           December Monitoring Study
1 1013
2 1710
3 1755
4 1835
5 2100
6 0810
127 min
30 min
40 min
145 min
23 min
25 min
.04 yg/m
4.90 yg/m
4.40 yg/m
.69 yg/m3
4.46 yg/m3
4.06 yg/m3

      SAMPLE D:  Air sampling In office building in Bedford.

      SAMPLE E:  Air sampling during drive from Bedford to Winchester, Mass.

      SAMPLE F:  Air sampling during drive from Winchester to Cambridge, Mass.

      SAMPLES G & H:  Air sampling in two meeting rooms at MIT.

      SAMPLE I:  Air sampling during drive from Cambridge to Winchester.

      SAMPLE J:  Air sampling during morning commute, Winchester to Burlington
 to Bedford.

      The exposure results are presented in Table 8 and show a similar pattern
 to the December exposure data.

      This second set of data is consistent with the December exposures ob-
 tained using  the rapid screen methodology.  In both sets, the use of short-
 term samples  to document the exposure during specific event periods provides
 the experimenter with a more accurate reconstruction of exposure events.

      In conclusion, the sensitivity and convenience of the ASV rapid screen
 methodology offer extreme flexibility in the design and implementation of an
 ambient, air  lead, field study program.
                                     TABLE 8
                            February Monitoring Study
17 min
39 min
12 min
57 min
24 min
20 min
79 min
157 min
30 min
27 min
3.2 ug/m3
.9 ug/m3
9.2 ug/m3
•5 ug/m
4.7 ug/m
2.1 ug/m3
•7 ug/m
.4 ug/m
3.8 ug/m3
1.7 ug/m3
in bldg.
in bldg.
in bldg.
in bldg.

 1.  Proposed National Ambient Air Quality Standard  for Lead.   Federal Register
     42, #240, p. 63076.  December 14,  1977.

 2.  Blood Lead Proficiency Testing Summary.   HEW-CDC  Report.   December  1977.

 3.  Direct Determination of Lead in Air Filters  by  Anodic Stripping Voltam-
     metry.  ESA Methodology, TMA-17.

 4.  NIOSH Manual of Analytical Methods, Volume I. P & CAM 173.  April 1977.

 5.  Procedure for Analysis of Aerosol  Pb and  Cd  Particulates by ASV.  ESA
     Methodology, M-30.

Francis J. Berlandi, Ph.D.
ESA Laboratories, Inc.
43 Wiggins Avenue
Bedford, Massachusetts  01730
     SCHEIDE:  Gene Scheide, Environmetrics.  In  the automobile, did you use
leaded or unleaded gasoline?

     BERLANDI:  I used leaded gas;  we  haven't gone  to an unleaded car yet.
I have a Volvo that is kept tuned up pretty well.

     SCHEIDE:  It would be interesting to  compare to see how much emission
you are getting from other vehicles on the road.

     BERLANDI:  Right.  As I said,  these are representative of a variety of
driving conditions.  I suspect that either I am entraining the dust, or it
is coming from my car.  I don't know.   But certainly it is something that
has to be investigated.

     As I said, this is done on our own.   We would  certainly like to pursue
this and do the study the way it should be done for documentation of this
type of exposure.

     CLIFFORD:  Paul Clifford, Carnegie-Mellon University.  If you are using
a filter of 0.8 microns for a size  cut, how much of the lead is smaller than
that?  How much of it are you missing?

      BERLANDI:  That  is  one  thing  that  you  have  to  remember about  a 0.8 micron
 filter.  It doesn't mean that  particles less  than 0.8 get  through.  The filter
 is sort of like a  sponge.  To  my best knowledge, from the  tests  that I have
 run on particle entrapment,  I  think  you are getting down to maybe 0.02, 0.01

      People have tried bubbler techniques behind filters in order  to capture
 the particle  lead  or  any of  the tetraethyllead.  And from  what they have been
 able to determine, 97 to 99  percent  of  the  lead  is  captured on the filter  in
 these cases.

      CLIFFORD:  So most  of the combustion product of the tetraethylleads
 are actually  very  large?

      BERLANDI:  Yes.  They can end up on the  filter strip  particles.  Maybe
 someone else  here  would  want to talk about  vehicle  exhausts.  I  am not an
 expert on it.  But I  know that those who have looked for the gaseous lead
 compounds, using the  filter, have  not seen  significant  transport to the in-
 teriors through the filter.

Studies of Semiconducting Metal  Oxides in
Conjunction with  Silicon  for Solid State Gas
     Angel G. Jordan, Ph.D., David J. Leary, Ph.D., Gulu N. Advani, and
     James 0. Barnes, Ph.D.
     Carnegie-Mellon University
     Pittsburgh, Pennsylvania

     This paper reports work at Carnegie-Mellon University dealing with semi-
conducting metal oxides, such as SnOj  and ZnO, that hold promise  as base ma-
terials for gas sensors.  The research alms at the understanding  of the role
played by dopants properly selected to enhance the sensitivity of the materials;
at the understanding  of the mechanisms of electron exchange between bulk and
surface when the gases to be detected  are sorbed onto the surface; at the under-
standing of the transport of charge carriers; and at the understanding of the
mechanisms involved in the preparation of the materials and devices, and the
characterization of such materials and devices.  The research  purports to study
the feasibility of materials* deposition and/or growth on conventional semi-
conductors, such as silicon.  The work is being carried out by an interdisci-
plinary team from the Departments of Electrical Engineering, Metallurgy and
Materials Science, and Chemical Engineering.

     An integral part of this research program is work aiming  at  the under-
standing of the most  prominent commercially available devices,  and the incor-
poration of these devices and the devices emerging from our own research in a
portable air pollution/combustible gas monitor.  This part of  the research is
presented in a separate paper by D.T.  Tuma and P.K. Clifford in these Pro-

     This paper starts with a section  on SnO. as a material for gas sensors.

 Details of sample preparation by Radio Frequency Sputtering techniques are
 given,  and analytical characterization of the Sn02 thin films is described.
 The electrical characterization of the films, with a variation of the oxygen
 pressure within the sputtering ambient, is presented.  A scheme is proposed
 which incorporates a stable SnO- film and silicon bipolar transistors in an
 integrated structure.

      This paper proceeds with a section on ZnO as a material for gas sensors.
 It focuses on polycrystalline thin films and the strong dependence of film
 resistivity and gas sensitivity on grain size.  The role of certain dopants in
 controlling the stability of the films and the gas sensitivity to certain
 gases is reported.  Response times, reproducibility, and long-term stability
 of films investigated appear to be adequate.

      The last section of the paper deals with H_ and H_S detectors utilizing
 MOS structures.  These devices are Pd-SiO.-Si Schottky barrier diodes and
 MOS capacitors.  Both appear to be very sensitive to H_ and H_S.

      A scheme is presented for incorporating Zeolite sieves on metal oxide
 devices for separating gases before detection by the sensor.  This technique
 is also amenable for incorporation in Pd-SiO^-Si devices.

 Sample P r e pa ration

      Thin films of tin dioxide are prepared by Radio Frequency Sputtering.
 The apparatus consists of a Randex 2400-J Sputtering system equipped with an
 He closed-cycle cryopump with variable power splitting facilities for the
 sputtering process.  The target material is a sintered disc of SnO. prepared
 by pressing high-purity SnO_ material obtained from Alfa Ventron Corporation.

      A base pressure of 3.5 x 10   Torr is routinely obtained.  The sputtering
 is performed by plasma composed of high purity (five 9's) argon and oxygen
 mixtures in dynamic equilibrium within the system.  A water-cooled substrate
 table with tungsten-halogen lamps provides a deposition temperature range
 from about 120° to 450° C.  The RF power supply is coupled to the anode (sub-
 strate)-cathode (target) assembly through a tunable impedance matching network.
 Typical RF powers for a deposition are in the range of 500 to 700 watts.
 Similar techniques are used for ZnO depositions.

Analytical Characterization  of  SnO  Thin Films

     Sn and 0_ combine  to  form  several compounds, some of which are SnO,  SnO_,
and Sn^Og.  The  tetragonal structure of SnO- has a lattice structure given
by a - b « 4.738 X, and a  c/a ratio of 0.675.

     Films are analyzed for  their morphology, structure, and composition.
The Scanning Electron Microscope (SEM) and the Transmission Electron Microscope
(TEM) complement each other  in  supplying information of grain size.  In gen-
eral, it is found that,  for  depositions carried out in an ambient of Oy^r
at a pressure of 25 mTorr, the  grain size is typically of the order of a  few
hundred angstroms.

     Structural  information  is  obtained from TEM and X-ray diffraction studies.
The information  from these data suggests that the films are polycrystalline
in nature, with no particular orientation with respect to the plane of the

     Elemental analysis  of the  films is carried out with the Auger Elemental
Spectroscopy techniques  (AES).  The results are to be interpreted with caution
because of the difficulties  with the measurement system; however, preliminary
results show that the Sn-to-0 ratio for stoichiometric SnO. is about 1.60.

     Figures 1 and 2 summarize  the results of analytical characterizations
on a Sn02 thin film.

Electrical Characterization  of  Sn02 Films

     Figure 3 shows a plot of the room temperature measurement on as-sputtered
films of Sn02 with a variation  of the oxygen pressure within the sputtering
ambient.  As can be seen,  the films for 0 percent and 100 percent 02 pressures
(corresponding to 100 percent Ar and 0 percent Ar, respectively) possess high
conductivities.  The curve goes through a maximum in resistance, presumably
at that point where the  stoichiometric composition of SnOj is reached.   More
studies are underway to  clarify this aspect.

Gas Sensitivity Response of  SnOg Films

     Several samples of  SnO« have been studied for their response to oxygen
and hydrogen.  Studies with 0- reveal information regarding chemisorption
phenomena.  The tests are carried out in a clean vacuum system that can be

               Sputttrvd  Film
 FIGURE 1.  Typical X-ray diffraction pattern for a thin film  of  SnO-.  The
 figures correspond to  the Bragg angles  (26)  as specified In the  ASTM cards.
 The  spread in the peaks is probably due to  a strain in the lattice.
                                                      SnO     2.00
                                                      Conlftrllt 1.72
                                                      NESA gtau 1.80
 FIGURE 2.   Typical Auger  spectrum of SnO_.  The  relative ratio  of  Sn-to-0
 peak height remains substantially constant, although their absolute values
 differ from one region to another depending on  the morphology.  The table shows
 the relative ratios for various Sn/O combinations.

                               Sputtering  Ambient
                               oxygen / argon
   10   -
                                                                  Gas Modulated
                                                                    I" 1

FIGURE 3  (left).   Variations of resistance of sputtered SnO_ films  as  a function of the oxygen percentage
in the sputtering ambient.  FIGURE 4  (top right).  The proposed  integrated gas sensor.  The  Sn02  serves
as a gas-modulated base resistor  that alters the quiescent level, and  hence the output voltage.  FIGURE 5
(bottom right).   Bipolar transistor  structure incorporating an isolation region for thermal  shielding.
The right-hand  portion of the figure  has a heater and the gas-sensitive oxide film.  Figure  used  in compu-
ter study of  heat analysis to evaluate the temperature distribution on the chip.

 pumped  down to 1 x 10   Torr.  The sample is heated in this pressure range for
 an  hour,  after which 0. is introduced into the system.  Very precise amounts
 of  the  gas can be introduced by means of a special injection port.   Readings
 of  the  resistance as a function of time, when plotted on semi-log paper, illus-
 trate a linear dependence, which is typical of an Elovic plot.   This, in turn,
 suggests  a surface reaction.  The sample is allowed to come to  equilibrium
 in  this environment, after which controlled amounts of H. are introduced in
 the system.  A plot of the fractional change in resistance versus the prepa-
 ration  conditions of the sample shows Increasing sensitivity for samples pre-
 pared in  decreasing amounts of 0. in the sputtering ambient.

      However, stability studies show exactly the opposite effect.  Thus, the
 most stable films are those prepared in 100 percent 02.  The opposing nature
 of  these  two results suggests that there is an optimum sputtering environment,
 which probably agrees with the optimum doping levels appropriate to the grain
 size of these films.


      The  results of the previous section indicate that some sacrifice can be
 made on the sensitivity of the films in order to considerably enhance the film
 stability.  An increase in the sensitivity may, however, be made by coupling
 the film  to an amplifying device such as a bipolar transistor.   Figure 4 shows
 such a  device, where it may be seen that the SnO_ layer serves  as a variable
 base bias to the transistor Ql.  Figure 5 illustrates a prototype of such a
 device  which is currently being developed in our laboratory. The need for
 integrating such a device calls for some isolation between the  heater and the
 transistor.  Such an isolation has been attempted with some success.  A com-
 puter simulation of the heat transfer in such a device has also been carried
 out,  suggesting the feasibility of the idea.  We should be reporting on these
 aspects in short term.

 Polycrystalline Film Studies

      Electrical properties of ZnO surfaces are very sensitive to ambient con-
 ditions,  and  high-purity, single crystal specimens are valuable for studying
 such processes.  Gas-solid interactions by chemisorption processes involve
 charge  transfer of electrons to or from surface layers.  Figure 6 illustrates
 a  specific  example of oxygen chemisorption on n-type ZnO.   Oxygen acts as an


           of oqrflM
FIGURE 6 (left).   Surface band  bending in polycrystalline ZnO has a spatial
extent of the order  of  a  Debye  length, L_.   FIGURE 7  (right).  Conduction
process in polycrystalline ZnO  where adsorbed  oxygen  plays a role at the
intergranular contact.
acceptor, introducing  surface  acceptor states  that  deplete electrons from the
solid.  The energy band bending  depicted  in this  figure  extends over a distance
of the order of a Debye length,  L_,  into  the solid.

     Conductivity measurements on  single  crystal  specimens,  however, will not
reflect any surface  sensitivity  if L   is  small  compared  to the  sample thickness.
In polycrystalline films,  the  surface  charge transfer  process is still at
play, and, with small  crystallite  sizes,  the surface layers  dominate the
measured conductance.  Figure  7  is an  illustration  of  possible  processes in
polycrystalline films  where  individual grains are connected  to  form a conduc-
ting medium.

     For the first situation,  we show  adsorption  of oxygen at free surfaces.
This establishes a depletion region accompanied by a large reduction in free
charge.  The depletion region  can  extend  through  the intergranular contact,
as shown, pinching off the conduction  path  between grains.   In  the bottom
figure, oxygen is adsorbed along grain contacts as well,  resulting in a poten-
tial barrier to carrier flow between adjacent grains which is modulated by the
amount of oxygen on  the surface.   The  expression  for the  measured resistance
of such films is

                               R     = g/
                                meas   ° n  --qv ,,
                                          effM eff

 where n    is the effective carrier concentration in the individual grain, q
 is the electronic charge, u .. is the effective mobility in the intergranular
 region, and g is a geometrical parameter.

      Studies by previous workers on films with this geometry have dealt with
 the situation by considering a highly conductive grain to be separated from
 its neighbors by a high-resistance phase which effectively determines the
 measured resistance.  Such an oversimplification does not deal with the elec-
 tronic processes directly, and cannot give more than qualitative information
 if applied to gas sensitivity.

      A more fundamental approach recently applied to this situation has pro-
 vided a semiquantitative description of electronic processes and a guide for
 materials processing to optimize gas sensitivity of polyerystalline metal-
 oxide semiconductors.

      A chemisorption charge transfer model would be expected to depend critical-
 ly on grain size and the corresponding number of free carriers taking part in
 conduction and surface reactions.  For the case of small crystallites, the
 volume concentration of surface acceptors can be comparable to or even greater
 than the volume density of free carriers inside.  Such a case requires simul-
 taneous solution of Poisson's equation and the charge neutrality equation
 for each grain.  We apply this approach to determine the crystallite surface
 potential, and, then, the effective carrier concentration as a function of
 the density of ionized surface acceptor states for a particular set of mate-
 rials parameters (e.g., grain size and geometry, doping concentrations, donor
 and acceptor energy levels distributed in the bandgap, capture cross section
 for surface acceptor states, etc.) (1).

      For the case of ZnO, the effective carrier concentration versus grain
 size is plotted in Figure 8 for a given surface acceptor density, N  .  The
                                                                    D 8
 results of the computer solution suggest the existence of a sharp transition
 in the effective carrier concentration between a highly conductive grain with a
 large radius and a nearly fully depleted grain of small radius.  The transition
 occurs for small changes in grain size and, for a given doping level, is a
 function of temperature and concentration of surface acceptor states.  The
 result suggests that material parameters might be adjusted such that small
 ambient induced changes in surface acceptor density could give rise to large
 changes in n  . and the measured resistance of the film sensitivity.  The


      T   10"
                  Comprfw ShniMkin
                                                     T «600K
                      Grain Sla R (nml
O'         10*
Groin Dionwler (run)
       FIGURE 8  (left).   Computer solution of effective carrier  concentration in a single  crystallite plotted
       versus crystallite radius.  FIGURE  9 (center).  Computer  solution of effective carrier concentration
       plotted versus surface acceptor  density on a single crystallite of ZnO.  FIGURE 10  (right).  Measured
       film resistivity of as-sputtered undoped ZnO versus diameter.   After heating in air,  all films have
       resistivity greater than 10  ohm-cm.  Solid line is computer  fit.

 effect of a distribution in grain size about a mean and a distribution in
 energy of surface acceptor levels and bulk donor levels is shown to smear
 out this sharp transition.

      The situation is illustrated in Figure 9, where n -- (from the computer
 solution) is plotted versus N  .   The surface acceptor concentration (for ZnO
 material parameters) is listed in the figure.  For the particular material
 parameters used, an optimum sensitivity of n __ to N   is determined for a
                >     f                *             S8
 discrete grain size of 35 nm and doping level N  * 1 x 10   cm

 Experimental Confirmation of the Transition Region

      In the effort to confirm the existence of a transition region in RF sput-
 tered ZnO, and to justify the application of the model to gas sensitivity of
 these films, resistivity measurements were taken.  Grain size of sputtered
 films was determined by TEM and SEM techniques.  Grain size could be controlled
 by varying the film thickness or substrate temperature during deposition.
 The details of deposition and characterization of RF sputtered ZnO appear
 elsewhere (1) .

      Figure 10 shows data obtained for several films exhibiting the depend-
 ence of resistivity of as-spurtered ZnO on grain size.  The data are fitted
 loosely to a computer-generated curve (solid line), where the fit was obtained
 for N  = 3.125 x 1016 cm"3, E  - .05 eV, N   = 2.2 x 1012 cm"3, and E  - 0.8
      u                       OSS                         A
 eV.  The Hall mobility could be measured in several of the lower resistivity
 samples.  Its value was less than 1 cm /V-sec.  Yeh (2) reports that the
 mobility in such films is independent of temperature and oxygen partial pres-
 sure, whereas n f, changes with these parameters.  We therefore infer that the
 measured resistivities reflect changes in n ff.  This suggests that the data
 of Figure 10 provide a semiquantitative confirmation of the dependence of the
 effective carrier concentration on grain size in polycrystalline ZnO.

 Gas Sensitivity Testing of Sputtered ZnO Films

      Gas testing has been carried out with the specific goals of 1)  evaluating
 the feasibility of this technology for personal monitors;  2)  establishing op-
 timal film doping and/or grain size;  3) determining sensitivity as a function
 of temperature for various gases;  and 4) exploring the use of additives such
 as palladium or platinum which are known to enhance sensitivity to particular
 gases.   The response to hydrogen,  carbon monoxide, and changes in the nitrogen/
 oxygen ratio,  are measured.

     Hydrogen  testing  is  carried  out by passing a mixture of  forming  gas
(5 percent 1^ 95 percent NZ) and synthesized air over the  sample.  Concen-
trations in the range  0 ppm  to 9,000 ppm are employed.  Carbon monoxide test-
ing at present is conducted  under static conditions, using  concentrations
in the 100-ppm range.  The nitrogen/oxygen ratio is varied  from 100 percent
N_ to 100 percent 0^.  In all cases, the film temperature is monitored  with
a chromel-alumel thermocouple.  Film resistance is measured using an  AC volt-
age and current; the AC technique is necessary because of the high thermo-
electric power of ZnO, which results in sizable DC thermoelectric voltages.

     The resistance measurement is performed with an AC digital voltmeter.
However, a DC voltage  proportional to the sample resistance is also derived
and recorded with a pen recorder.

     Figure 11 shows the  hydrogen response of two Ga-doped ZnO films  which
were both sputtered onto  substrates heated to 480° C.  Both films are seen  to
quite closely follow a dependence on the square root of hydrogen concentration,

                                 f  - A [H//2

where G  is the conductance  in air and A is a constant.  Therefore, calibration
of a particular sensor is reduced  to specification of the constant A  for that
device.  Comparing the two curves of Figure 11, it is seen that Device B has
a somewhat greater sensitivity.   This device is similar to A, except  that a
500 X film of palladium was  sputtered onto the ZnO film in the last step of
fabrication.  In Device B, palladium is probably acting to catalyze the dis-
sociation of hydrogen—a  necessary step in the reaction of hydrogen with the
sensor surface.  Based on the scatter in the data, a detection limit of 10
ppm H2 is indicated for Device B.

     The response of a sensor to  hydrogen is indicative of the response of
the device to a number of gases containing hydrogen;  i.e., CH^, HgS,  and NH.-.
In these cases, the parent molecule undergoes dissociation,  neutral hydrogen
is produced, and the latter  reacts with the sensor surface.   Of particular
practical interest are detectors  for CH^ and H2S.   In the latter case, a palla-
dium device similar to Device B is particularly attractive,  since palladium
is known to catalyze the dissociation of HjS (see below).  However, ZnO devices
must be protected against direct exposure to sulphur compounds, which poison
this material by forming surface ZnS layers.   Two solutions to this problem are

                  [*"*• Concpif rof ion J
                                                           to * \
                                                                      III pp» CO
                       FIGURE 11
                                                 FIGURE  12
Temperature  Dependence
of Hydrogen  Response of
Device A.
         280     300    320     340    360
                   Temperature (*C>
                   FIGURE  13

  FIGURE 11.   Hydrogen  response  at 330° C.
  FIGURE 12.   CO response of Device A at
  330°  C.  FIGURE 13.   Temperature de-
  pendence of  the hydrogen response of
  Device A.  FIGURE 14.   Reverse  leak-
  age  current  density of a Pd MOS diode
  and  flat band voltage of a Pd MOS
  Capacitor plotted against hydrogen con-
                                                                         . In otr  T.JOO-K
                                                 K>   NO   WOO
                                                 Hg CorvCMMflo. too)  b dr
                                                 FIGURE 14

 being  evaluated.   One  Involves  the  use  of  a  Zeolite molecular  sieve  (see
 below) over  the  film.   The Zeolite  is chosen to  allow  passage  of  hydrogen
 but not  sulphur.   Palladium is  added to the  outer  layers  of  the Zeolite  through
 sputtering or  other  means  such  that the dissociation of H-S  is promoted.
 Since  this device  will also respond to  hydrogen, it must  be  paired with  a
 device not sensitive to H2S,  but  instead to  hydrogen.  The latter could  be  a
 Zeolite-covered  film without  palladium  added to  the Zeolite.   The other  solu-
 tion being investigated involves  cladding  the ZnO  film with  palladium, which
 is relatively  unaffected by sulphur compounds and which is transparent to
 hydrogen.  In  this case, the  palladium  acts  to screen out sulphur.

     Figure  12 is  an example  of the response of Device A  to  111 ppm CO.
 Device B showed  negligible response to  CO, presumably because  the overlaying
 Pd film prevented  the  CO molecules  from reaching the ZnO  film.  Palladium or
 platinum additives,  therefore, are  useful  for detectors designed  to respond
 to small concentrations of methane  or other  gaseous hydrocarbons  in the pres-
 ence of large  concentrations  of CO.

     In addition to  the selectivity afforded by catalytic additives, the  tem-
 perature-sensitivity relationship may be a means of enhancing  selectivity.
 Figure 13 shows  the  temperature dependence of the hydrogen response of Device
 A.  The temperature  dependence is believed to be a unique characteristic  of
 a particular gas.  In  the  case of hydrogen compounds such as CH,  or KLS,  the
 temperature  dependence  will reflect dissociation of the parent molecule,  as
 well as reaction of  the film  to hydrogen, and, therefore, should be unique for
 a particular gas.  Use  of  temperature as a means of selectivity will require
 processing information  from an array of  sensors operating at different tempera-
 tures.  This approach  is favored over scanning a single sensor through a  tem-
 perature range because  of  response  time considerations.

     The response  time, reproducibility, and long-term stability of films
 investigated thus  far have been adequate for most monitoring applications.
 The response time  for exposure to a gas is on the order of seconds;  this meas-
 urement may reflect  the response of the gas testing system itself and is  there-
 fore an upper  limit  of  the film characteristic.   Recovery following removal of
 the gas appears to be somewhat slower,  with the present sensors requiring sev-
 eral minutes.  Stability has been investigated over periods of 1  week, and the
 conductance in air of a given film appears stable within a 10-percent range
 for a film cycled over  a range of temperature and exposed to a number of gases.
Long-term stability  is, of course, critical to the successful operation of
gas sensors.

 Schottky Barrier Diodes and MOS Capacitors

      Thin metal films which dissociatively dissolve H-, such as Pd and Pt,
 have been incorporated in integrated diodes and capacitors on passivated
 silicon substrates (3,4).  Changes in work function at the metal-SiC^ inter-
 face induced by ionized HZ and gases such as H2S and NH3 can give rise to large
 changes in the reverse leakage current of a Schottky barrier diode or changes
 in the flat band voltage of MOS capacitors.  H. sensivitity is shown in Figure
 14 with the structures operated at room temperature.  The figure illustrates
 diode sensitivity in terms of changes in reverse leakage current and changes
 in capacitor flat band voltage versus hydrogen concentration.  One to 10 ppm
 H_ in air are detected, with the upper limit on sensitivity being near 1 per-
 cent H..

      MOS capacitors can be operated at elevated temperatures, which Improves
 response times while the sensitivity to H_ is preserved.  Both structures
 are sensitive to BUS, but operation at room temperature results in a short-
 term deterioration in response due to sulfur poisoning of the metal surface.
 However, the sensitivity is entirely reversible at elevated temperatures above
 160° C in air.  Diode structures cannot be used at elevated temperatures,
 whereas MOS capacitors essentially retain their characteristics in this tem-
 perature range.  As an alternative to measurements of changes in flat band
 voltage with H- or H_S, the actual change in capacitance can be measured at
 a given gate voltage.  Figure 15 Illustrates the result of such a measurement.
 The figure shows a change in differential capacitance versus H.S concentration
 at elevated temperature in air.  One to 10 ppm H.S can be determined.  Device
 sensitivity can be improved by minimizing the SiO. dielectric thickness.

 Device Selectivity

      The aspect of selectivity has been approached from several angles.  Fig-
 ure 16 constitutes a specific illustration where the addition of a catalyst
 as an impurity shifted the ZnO temperature distribution for the response of
 the film in the case of CO.  In general, the sensitivity of a metal-oxide is
 temperature dependent, and conceivably an array of devices operating at dif-
 ferent temperatures with the associated signal processing could be used to
 discriminate several gas constituents.  Alternatively, preprocessing of the
 gas ambient by filtering of the gas components—analogous to preprocessing

               C (fftl

    >„ • «0ni»
   V -IJ Volll
                 WO   W*    K>'   rf
                 MjS CONOOmunON (pom) hi ok
                FIGURE 15
                                                    1*0 Vf Se
                                         225     300
                                    FIGURE 16

   I  io*
          Ziollt* 3A
RF SpuHt'frt ZnO -
                            with 3A
                       * .0*
              wf     10*     10*
                Conctntrotlon ppm

                FIGURE  17
                                                   Zeolll. 3A
                                                                TGS*8I3  (SnOj
                                        10      100
                                   Conecnlration  ppm

                                    FIGURE  18
FIGURE  15.  Differential capacitance  in a Pd MOS capacitor plotted  versus
concentration for a  negative gate  bias of -1.5 volts.   FIGURE 16.   Percent
change  in resistance of ZnO film exposed to CO in  air  plotted versus temperature.
The presence of Pd in the film acts to shift the temperature dependence of the
response  to lower temperatures.  The  sensitivity is  improved.  FIGURE 17.  3 X
pore  size Zeolite incorporated on  ZnO sensor effectively screens HjS separating
the response to H? and CO from H^S.   FIGURE 18.  3 X pore size Zeolite on SnO.
is also used to separate H. from H2S.

of communication waveforms  before  detection—can provide selectivity,  as  well.
We have been working  to  incorporate Zeolite molecular sieves  with commercially
available devices  and with  our  own integrated versions.   Figure  17 demonstrates
the results of  the use of a 3 angstrom pore size Zeolite deposited on  ZnO as
a filter for CO.   H.S, which is a  poison to ZnO, is screened  out because  its
molecular size  is  too large to  pass through the Zeolite.  Responses to H_S
without the Zeolite are  shown.   With the filter, H-S is  effectively screened
from  the sensor.   The currently employed deposition process for  the Zeolite
is not fully compatible  with standard planar processing  techniques and, before
this  technology can be fully Implemented,  thin film deposition of molecular
sieve material  must be developed.

      Figure 18  illustrates  the  utility of  3A Zeolite for separation of H_ and
ELS.  The results  shown  here are for a commercially available SnO.-based  de-
tector.  Two Identical detectors,  with one incorporating the  Zeolite,  can
effectively discriminate the response to HLS from that to H2-  This is also
of interest for the diode and capacitor structures referred to earlier.

      Radio Frequency Sputtering is a powerful,  versatile technique for the
 preparation of thin films of SnO-  and ZnO to be used  as sensors  for certain

      In conjunction with Sn02,  the fractional change  in resistance versus
 the preparation conditions shows increasing sensitivity for samples prepared
 in decreasing amounts of 0_  within the sputtering  system.   However,  the most
 stable  films are those prepared in 100 percent 0«.  Thus,  there  appears to  be
 an optimum sputtering environment  for both a reasonably stable,  and yet an
 adequately sensitive, film.

      Reasonably stable films of SnO_  (albeit not very sensitive) may be in-
 corporated with bipolar transistors in an integrated  structure,  with the gas-
 sensitive  portions  being operated  at  elevated temperatures while the bipolar
 transistor remains  at near room temperature.

      In polycrystalline films of ZnO  (and also SnO.),  the  grain  size depends
 on deposition temperatures.   There appears to exist an optimum grain size for
 which the  sensitivity of the film  is  significantly enhanced.

      Ga doped ZnO films are  stable, and with the addition  of Pd  can be used

for the detection of EZ and hydrogen bearing gases (CH,, H2S, NHj) with good
sensitivities.  In the case of H.S, the ZnO must be protected agaitx'st direct
exposure to sulfur compounds which poison the material.

     In addition to the selectivity afforded by catalytic additives, Ga
doped ZnO devices can be designed to operate at elevated temperatures, so
as to have an enhanced sensitivity to gases.

     Schottky barrier diodes and MOS capacitors, both incorporating Pd-SiOj-Si
layers, can be utilized as very sensitive H. and hydrogen-bearing gases such
as H~S and NH«.  In the diodes, their reverse current sets a limit to the
maximum temperature of operation.  The capacitors, however,  can operate at
sufficiently high temperatures (of the order of 160°  C)  so that the molecule
of the detected gas may be dissociated.

     Zeolites molecular sieves may be incorporated in the devices above to
afford selectivity—by acting as filters against certain gases while passing
some others which would dissociate on the surface and modulate the electrical
properties of the sensor.

     This research was supported in part by the Department of Energy under
Contract No. EE-77-S-02-4346.  Support of certain portions of the work by
Mine Safety Appliances Co. Is also acknowledged.

 1.  Leary, D.J., Barnes, J.O., Jordan, A.G.   Characterization of RF Sputtered
     ZnO Films (to be published).

 2.  Yeh, K.W., Muller, R.S., Washburn, J.  Characterization of RF Sputtered
     ZnO Piezoelectric Films using TEM.  33rd EMfJA Meeting, Washington, D.C.,

 3.  Lundstrom, I., Shivaraman, M.S., Svensson,  C.  A Hydrogen Sensitive
     Pd-Gate MOS Transistor.  Journal of Applied Physics, 46:3876, 1975.

 4.  Lundstrom, I., DiStefano, T.  Influence  of. Hydrogen on Pt-SiOj-Si Struc-
     tures.  Solid State Communications 19:871-875, 1976.


Angel G. Jordan, Ph.D.
Carnegie-Mo lion University
Pittsburgh, Pennsylvania  15213
     MAGE:  I wo>uld like to ask that the  questions about this paper be held
until after the next paper, because it will be on the same topic from the
same department.

     (NOTE:  The discussion for this paper appears following the text of
the next paper, "Microcomputer Control and Information Processing Technology
for Semiconductor G.as Sensors.")

Microcomputer Control and Information
Processing  Technology for Semiconductor
Gas Sensors
     David T. Tuma, Ph.D., and Paul K. Clifford
     Carnegie-Mellon University
     Pittsburgh, Pennsylvania

     A primary goal of our research is the eventual development of a portable
pollution monitor.  The ideal monitor would be small, lightweight, safe,
rugged, inexpensive, and capable  of continuous operation throughout the work-
day.  As a gas-detection system,  it would need to  respond reliably to com-
bustible and polluting gases over a dynamic range  of several orders of magni-
tude, with a time resolution of a few minutes and  reasonable sensitivities
determined by flammable and toxic limits.

     A gas-detection technology that seems suited  to the eventual fulfillment
of these goals is that of semiconductor detectors.  However, to date,  semi-
conductor detectors have been characterized by low reproducibility, drift,
and very poor gas selectivity.  These limitations  cannot be overcome until
much more is known about the nature of the detection mechanism in the semi-
conductor material.  Some of these limitations, however, appear to be in-
herently tied to the nature of the materials used  in such a way that it is
unlikely that they will be completely overcome by  materials research in the
near future.  Although we will see the rapid evolution of a wide variety of
semiconductor sensors in the near future, they will remain broad-spectrum
devices.  Aside from this lack of selectivity, it  is likely that semiconductor
sensors will continue to exhibit  highly nonlinear  responses which are not
amenable to present "direct reading" instrument design.

     If the computational power which is now becoming available at low cost in


microcomputers  can be used to compensate for the Imperfections of present
sol.id-state  detectors,  then the development of a portable gas-detection In-
strument  may be possible.   "Smart"  instrument design through the use of
microprocessors may obviate the need for a simple linear response and provide
a means  to  enhance selectivity of existing or future detectors.  This is
made  possible only by the  significant decrease in both size and cost of com-
puter systems over the past few years.

      At  Carnegie-Mellon University, we have been empirically investigating
the basic response mechanisms of semiconductor detectors.  That is,  we deter-
mine  the  dependence of the semiconductor material properties on temperature,
gas concentrations, fouling,  past history, etc.   The purpose of this investi-
gation is twofold.  First, we seek to provide insight into the characteristics
which would  be desirable in future detectors, thereby aiding in the  develop-
ment  of  higher-performance detectors.  Secondly, we are determining  which
properties  of the sensors  are able to be exploited, by using innovative signal-
processing  techniques,  to  overcome their seemingly Inherent limitations.  A
synergistic  cooperation between semiconductor and microcomputer, broad-spectrum
sensors  and  computational  selectivity enhancement, may be necessary  to make
the genesis  of a truly personal, portable gas-detection system feasible.

Instrumentation Design

      The  design of personal detection systems to date has concentrated on
direct reading sensors with a simple linear response.  They are most suitable
to  portable  technology. Unfortunately, direct reading sensors are not avail-
able  for  the plethora of environmental contaminants which need to be monitored;
the gap  between specific sensors and monitoring  needs is widening.  For this
reason,  instrumentation design will need to adapt its methods to sensors of
low gas  specificity and nonlinear responses.  Fortunately, this is possible
with  the  advent of miniaturized computers.  A microcomputer-based detection
system would have the following attributes which may become indispensable in
a portable  personal gas monitor.

      Display.  An Instrument  for hazard detection would need to provide a
warning signal  when a dangerous level of a particular pollutant is present*
A microcomputer could provide a hazard signal when either an instantaneous
dosage or cumulative exposure exceeds a preset level.  The threshold levels
could be  easily programed  and may even be allowed to depend on other gas con-
centrations.  In addition  to  warning alarms, a microprocessor-based  instrument
could be  easily interrogated  by the wearer to provide a display on demand of
instantaneous,  average,  and cumulative exposures to one of several gases*


     Memory.  Increasingly,  the  focus of personal pollution monitoring  Is
shifting from hazard  detection to  trace gas monitoring.  For  this,  automated
data analysis and  data  recording is needed to permit the handling of  the
large amounts of data generated  during routine monitoring.  We need an  in-
strument not only  capable  of direct gas analysis and small enough to  be carried
by a worker, but it must also maintain a memory of gas exposure  throughout
the day.  This memory can  then be  transferred to a permanent  record in  a
centralized monitoring  system.   Such a personal monitor necessitates  the
use of microprocessor technology with its attendant semiconductor memory.
Present processor  technology enables us to visualize an instrument  small
enough to be carried  which would provide a hazard signal if one  of  four or
five gases exceeded safe levels, and which could maintain a memory  of the day's
profile of gas concentrations.   At the end of the day, that memory  could be
dumped into a larger  computer facility, providing permanent and  detailed re-
cords of personnel exposures.

     Control.  A microprocessor*  s capability for real time control  extends
from the mundane to the esoteric.  At one level, the microprocessor may be
used to reduce power  consumption by intermittently operating  the sensors.
Sensors may be sampled automatically, with power-conserving dead space between
samples.  More Interestingly, the interactive control of semiconductor sensors
may extract much more information from the sensor than could be had statically.
Feedback to and control of the electrical bias point or device temperature,
based on computed results, may be needed to optimally exploit the semiconduc-
tor's detection properties.

     Signal Processing.  Information from a bank of sensors—each sensor of
which in itself is unselective but exhibits differing relative gas  sensitivities-
can produce a system of nonlinear equations.   These may be solved numerically
by a computer.  Microcomputers having the computational ability and memory for
a task such as this (i.e., nonlinear regression of sensor information to pro-
duce gas concentrations), are now available in sizes and costs amenable to the
needs of personal dosimetry.  Current trends in processor technology Indicate
that their signal processing  abilities might well continue to grow;  increasingly
complex algorithms may be  implemented in real time.

Differential Sensitivities

     Sufficient information must be provided to the microprocessor so that

 It is able to deconvolve the gas concentrations of interest.  This may be done
 either by using several detectors with different relative sensitivities to
 the gases, or by using one sensor whose sensitivities to the gases may be
 changed in time by computer manipulation of a controlling parameter.  The
 following methods may be used to gain differential sensitivities from otherwise
 broad-spectrum sensors.

      Detector Construction.  Gas detection by semiconductor sensors is a com-
 plex process involving absorption and desorption of gases on the semiconductor
 surface, surface chemical reactions, diffusion of gases along grain boundaries,
 and interaction of adsorbed species with the electronic structure of the ma-
 terial.  Layer thickness, grain size, heat treatment, dopant concentrations,
 and any physical or chemical modification of the catalytic activity of the
 surface may be used to change gas sensitivities.  Although it may be difficult
 to attain—for example, absolute selectivity by a change in dopant concen-
 tration—at least the relative sensitivities to several gases may be suffi-
 ciently changed to allow use of the device.

      Filtering Techniques.  Various filters may be used in conjunction with
 wide-spectrum, metal oxide semiconductor detectors to render very different
 relative gas sensitivities.  In particular, we have had success with Zeolite
 molecular sieves.  This technology may be merged with silicon integration
 techniques to yield single chip sensors composed of a dozen or so differently
 filtered, metal oxide semiconductor devices.  Such ultra-miniaturization may
 soon produce a very powerful, general-purpose personal gas monitor.

      Manipulation of Bias Conditions.  Some semiconductor gas detectors (e.g.,
 Pd-SiO_-Si Schottky diodes) have electrical bias conditions which may be made
 subject to microcomputer control.  Depending on the physical and chemical
 properties of the particular gas species being adsorbed, a change in bias
 of the semiconductor device may change its gas sensitivity.  Hence, a change
 of electrical bias by the microprocessor yields different relative gas sen-
 sitivities.  Simplicity of control, rapidity of response, and ease of inter-
 facing, make electrical bias change a desirable form of control.  Unfortunately,
 there are few sensors which behave reproducibly and whose relative gas sensi-
 tivities have yet been reliably measured.

      A semiconductor gas sensor* s thermal bias—its temperature—is subject to
 easy manipulation.   Because the mechanisms of chemlsorption, gas diffusion,
 chemical reactions,  and gas-bulk electronic interaction are all strongly
 temperature-dependent,  a metal oxide semiconductor's sensitivity to a gas is
 a very strong function of temperature.  A single sensor, if operated sequenti-

 ally at  several  temperatures,  will  exhibit  different  relative  gas  sensitivities.
 Each gas will  have  its  own "eigenfunction"  of  sensitivity  relative to tempera-
 ture.  By  changing  the  temperature  of  a  device, a  processor  may  gain enough
 independent  sensitivities  for  the determination of several gas concentrations.

 Development  of the  Detection Algorithm

     After a bank of  sensors exhibiting  different  relative sensitivities  to
 gases has  been assembled by any  of  the above techniques, signal  processing
 can be used  to deconvolve  the  individual gas concentrations  from the collec-
 tive response  of the  sensors.  To construct the deconvolution  algorithm,  a
 knowledge  of the exact  nature  of the response  of all  the detectors  to all the
 gases is needed.  Quantitative expressions of  detector response  can be deter-
 mined empirically,  guided  by theoretical knowledge of the  reaction  mechanisms
 and a suitable experimental data base.   It is  Important to note, however, that
 a full understanding  of a  device's  nature is not necessary for using that device.
 A phenomenological  knowledge of  its behavior,  derived empirically,  is suffi-
 cient.   The  microprocessor has the  benefit of  a collection of  responses either
 gained from  a  bank  of sensors  or a  single sensor whose operating conditions
 are changed  sequentially in time (or any combination  thereof):

                             Sl ' fl «V G2' - V
                            Sm - f  (G , G , ... G )
                             m    m   1   /       n

where S.       is the response in voltage or current;
      f,       is the predetermined functional dependence of response on gas
               concentration; and
      G, ...G   are the concentrations of ambient gases.
       I    n

     If m >^ n — that Is, if there are enough detector responses with different
relative gas sensitivities — then the gas concentrations, G. , G2,  ... G , can
be determined.  Of course, the f's will not in general be linear  functions,
and the above system of equations will not be explicitly solvable analytically.
Numerical techniques, however, can be developed and used to solve such a
system.  The efficiency of those techniques will depend on the specific
nature of the nonlinear functions.  If m > n, then, as well as gas concentra-
tions, the solution can include a figure of merit or confidence interval for
each determined gas concentration.  This is very desirable, since a personal

 pollution  monitor will often be exposed to gases which it was  not  designed  to
 measure.   Even though these extraneous gases will interfere  with the  gas  meas-
 urements,  the confidence interval produced with each gas concentration will
 signal  its validity or error.

      A  technique of selectivity enhancement is presently being developed  for
 use  with commercially available gas sensors.  We are in the  process of deter-
 mining  the functional dependence of device response on gases in the ambient.

      Our research effort has concentrated on the testing of a variety of  in-
 house gas sensors as well as commercially available sensors.  The Taguchi Gas
 Sensor (TGS), which is manufactured by Figaro Manufacturing Company of Japan,
 typifies the state-of-the-art in gas detection.   Our purpose in studying  it
 is twofold.  Its behavior is characteristic of many other metal oxide semi-
 conductor detectors, so that knowledge of its parameters may guide research
 and development of new detectors.  Being commercially available,  it is manu-
 factured reproducibly; we can test large numbers of them inexpensively.  The
 other purpose of the study is the identification of parameters of the Taguchi
 Sensor which may be conducive to microcomputer selectivity enhancement.  It
 may be that an already-existing detector which is used in only smoke detection
 may be suitable for use in gas identification.

      The TGS is constructed of a highly porous,  n-type,  tin oxide semiconduc-
 tor whose conductivity is greatly affected by the gases  present at its surface.
 The surface reactions which determine the sensor's detection properties are
 strongly temperature-dependent.  To control the  sensor temperature, a heating
 coil is imbedded within the sensor.  A power of  600 mw dissipated in this coil
 is capable of heating the TGS to about 400° C.  Heating  in order to achieve
 reasonable sensitivities to gases is a method likely to  be used by most
 detectors we plan to test.

      When heated, the Taguchi Gas Sensor is responsive to a wide variety  of
 gases.  Because of the complexity, nonlinearity, and nonspecificity of the
 Taguchi Gas Sensor's response to gas concentration, these devices are used
 only for smoke detection or explosive level indication;  they are not used for
 low-concentration detection or gas identification.

 Measurement Apparatus

      A dual microcomputer data acquisition and analysis  system developed  by


 the authors  is  used  for all  measurements.  It  consists  of  three  parts:   a
 clean gas  system,  a  dedicated  microcomputer data acquisition  system,  and a
 data analysis microcomputer.

     Gas System.   The  gas  system  consists of several exponential dilution
 bottles in which  the semiconductor  sensors are mounted.  Gas  concentrations
 in these bottles  can be maintained  accurately down to a few parts-per-million.
 It is possible  with  these  systems to produce slowly changing  concentrations
 of gases so  that  the data  acquisition system can take data unattended over
 long time  periods.   There  are  several dilution bottles so that some may  be
 dedicated  to an automatic  data collection system while others can be  used for
 rapid manual testing of detectors.

     Data Acquisition.   A  real-time data acquisition computer system  measures
 and records  the resistances  of a bank of sensors mounted in the  clean gas
 system.  The data  can  be output in readable form to the teletype or,  under
 program control, be  transferred to  the data analysis computer.   The forte of
 this microprocessor-based  data acquisition system is that it can log  huge
 amounts of important data  in an automatic, reproducible manner over long
 periods of time, thereby freeing the experimenter from this costly task.

     Data Analysis System.  The heart of the data analysis system is  a SOL-20
 (Processor Technology  Corp.) 8080 microprocessor-based computer  system.   This
 system has a keyboard  and  video display for human interface.  A  floppy disc
mass storage system  gives  the  computer sufficient memory capability to run
an Interactive  high-level  language.  The functions of the data analysis
system are the  following:

     1)  To  control  the  microprocessor*based data acquisition system.  Control
parameters are  passed  to the data acquisition system, telling it when to  take
data and how to change  the sensor heater voltage.  The accumulated data  is
passed to the main computer, where it can be processed.

     2)  The raw data  can  be stored on a floppy disc, from which it can  be
retrieved at will.

     3)  An  interactive high-level language (BASIC) can be used  to program
the data analysis functions.  The data analysis will consist predominantly of
curve-fitting techniques.  Trial functions will be fit to the raw data,  and
the goodness of fit  will be tested.

     4)  Outputs—such as best fits of curves, functional parameters, or inr-


 portant  intermediate results—can be delivered to either a teletype  or video

      Experiments.  Data acquisition and analysis experiments are of  two broad
 classes:  investigation of the steady state response of the detectors  to am-
 bient gas concentrations; and determination of the dynamic response  of the
 sensors  to a step change in gas concentration or temperature.   Most  experiments
 are done automatically by the computer system so that they may be reliably
 repeated at a later time or for other conditions.


      Steady State Response.  When TGS sensors operating at 5-V heater  voltage
 are placed in pure N_, their resistances drop from "air" values of approxi-
 mately 10 to 20 K« to very low values (.5 to 2 K«).  In addition, the  sensors'
 resistances are observed to drift over a large range (of up to a factor of 2)
 over time periods of hours to days in a constant environment.   If 0_ is added,
 the resistances increase—nearly following a power-law relationship—until in
 pure 0_  they are the greatest observed (500 Kfl to flm).  Figure 1 shows this
 power-law response for several sensors.  Resistances of individual sensors
 start at vastly different values, but all show similar power-law slopes.

      If  one starts with a high 0_ concentration (for instance, pure  02 or air)
 and adds either H», CH,, wet CO, or HjO, very similar curves are produced
 (as shown in Figure 2).  The resistance versus concentration is again  de-
 scribed  by essentially a power law, but now the resistance decreases.   The
 remarkable feature of these curves is that their slope is the negative of the
 slope of increasing resistance in 0_ concentrations.

      An  equation which fits these curves very well is found to have  the
 following form

                               R  =  (KQ + K[C])~6

 for H2,  CH^, CO, H20; and for 02, replace - 8 by + 8.

 Where R   =  resistance of sensor
       K   =  a detector dependent constant
             which describes the finite
             resistance at zero gas


       1000. .
                                OXYGEN CONCENTRATION (ppm)
    FIGURE 1.  TGS resistance versus oxygen concentration  in ni-
    trogen.  The response  approaches a power law for high  con-
    centrations.  R/'KfO-J   where 8 is the power-law exponent.

       1000- •
                                            wet CO
                            GAS CONCENTRATION (ppm)
   FIGURE 2.  TGS resistance versus  reducing gas concentrations
   in  an air ambient.  For high concentrations, the response
                           •* A
   is  power law:  RrK.fC.]  .  Differing  gas sensitivities,
   K's,  are evident in the separation  of  the curves.

       K  -  a constant which Is specific
             to a sensor and the type of gas
      [C] «  concentration of gas (ppm)
       3  «  power-law slope.

      In the above equation, the B is very similar from sensor  to sensor or gas
 to gas, but it changes sign for concentrations of 0..  The gas sensitivity
 coefficient, K, for a particular gas is in general independent of  the con-
 centrations of other gases.  However,  there are important exceptions*  In parti-
 cular,  the sensitivity to carbon monoxide is dependent on the  concentration
 of water vapor.  Its K is the  product  of a temperature-dependent term and the
 water partial pressure.  For all other gases measured, the K's are all inde-
 pendent of gas concentrations.  In general, the gas sensitivity coefficients
 change  from sensor  to sensor,  gas to gas, and temperature to temperature.
 It is this temperature variation that  will need to be exploited if these sensors
 are to  be used to selectively  identify a gas.  Imbedded in the K   term are
 the drift effects.

      For a single sensor operating at  constant temperature, there  is a day-to-
 day variation in resistance, even in a very "clean" atmosphere.  If the depend-
 ence of sensor resistance on CH, concentration in air is measured  on different
 such days, a family of curves  is obtained, each with a different starting
 point but with other essential features much the same.  If these curves are
 optimally fit to

                                R = (KQ + K[C])~B

 g is found to be constant to + 25 percent and K constant to +  10 percent.
 Almost  all variation is found  in the K term.  We feel this variation is due
 to slow chemical reactions, the products of which foul the surface.

      Based on preliminary results, we  feel that the dependence of  a sensor
 resistance on the simultaneous presence of concentrations of several reducing
 gases can be expressed by

                           R =  (KQ +Z   Kf [C^)"8

 where [C ] is the concentration of the i   reducing gas, and K. is a strongly
 temperature-dependent gas sensitivity  coefficient for the i   gas.

      This expression suggests  that if  there is enough temperature variation

 among  the  sensitivity coefficients K  ,  then  it may be  possible  to  produce
 gas  specificity by  varying  the  temperature.  Long stabilization times  and
 hysteresis  effects  greatly  complicate data collection  when  the  temperature  is
 varied.  If,  however,  at  different temperatures  the resistance  versus  concen-
 tration characteristic is measured, and  then the corresponding  relationship
 between resistance  and temperature at constant concentrations is obtained,
 then a graph  similar  to that  shown in Figure 3 would be obtained.  Here,  we
 can  see that  temperature  dramatically affects the zero-concentration resistance
 as well as  the  sensitivity  to the gas.   If such curves for  different gases
 show different  sensitivity  functions  versus  temperature, then,  conceptually
 at least, we  have a means of  improving the specificity of the sensors  to
 various gases.

     The dependences  of the sensitivity  coefficients on temperatures have not
 as yet been well determined for any gas  other than methane.  Preliminary data
 suggest, however, that they are characteristically different functions, as
 shown in Figure 4.  Experimental effort  is continuing to characterize  these
 functions further.

     Transient Response.  When the temperature of a Taguchi Gas Sensor is
 abruptly changed, its  resistance exhibits a  very complex transient response,
 as shown in Figure  5.   There  is a sudden peak (or dip, depending on the direc-
 tion of temperature change) and a slow relaxation towards the initial value.
 This is followed by a  very  slow drift in resistance extending from 10 minutes
 to 3 hours*  The sudden peak  and its  relaxation can be understood in relation
 to the chemisorption  processes occurring at  the surface of the material.  Other
 metal oxide semiconductors  show very  similar behaviors.  The very slow change
 occurs as the material  reaches a new  equilibrium with the oxygen ambient.
 This interpretation is  confirmed by studies  using Auger Elemental Analysis,
which show changes  in  the oxygen content and stochiometry of the tin oxide
material with changes  in  temperature.   Measurement of the transient response
 for many step changes  of  temperature  in the 400° to 800° K range, and subse-
 quent analysis of the  time  constants  observed,  are yielding valuable insight
 into the evolution  of  chemisorptive models for device behavior.

TGS Selectivity Enhancement

     The general response of a Taguchi Gas Sensor to the reducing gases H-,
CH, , and CO is:

                     R - (K  + KHJ  + K[CH4]  + K3[CO])"8

                   FIGURE 3
               1000 -

                                      0 ppm
100 •
                                                      100 ppo

                                   TEMPERATURE (*K)
                                                                      A 300
                                                                      E ZOO
                                                                        500 -H
                                                        400 "
                                                                                                         FIGURE 4
                                                                                         800' K
                                                                                                         FIGURE  5
                                                                           -I	1—

                                                                            100      1000

                                                                             TIME (sec)
        FIGURE 3.  Detector resistance versus temperature in various concentrations of methane  in air.   The detec-
        tor's sensitivity steadily  increases with increasing temperature.  FIGURE 4.  Variation of gas  sensitivity
        coefficients  with temperature  (not to scale).  The sensitivity to each gas is a characteristic  function
        of temperature.   FIGURE 5.  Transient response of the TGS to a step  change of temperature.  The initial
        peak and relaxation reflects a change in chemlsorptive equilibrium at the semiconductor surface.  The
        subsequent slow drift, due  to  chemical and stochiometric change, may persist for several hours.

Because the K's  are  strongly  temperature-dependent,  the  response  may  be
measured at several  temperatures  and  the  system of equations  solved numerically
for the gas concentrations.   A complication arises with  the gas CO.   Sensitiv-
ity to it depends  on water concentration; an Independent measurement  of  water
vapor pressure may be  necessary.  It  is possible  that in an atmospheric  ambient
of normal relative humidity,  the  error introduced by ignoring  the dependence
on water vapor will  be negligible.  Measurement of the gas sensitivity coef-
ficients will reveal whether  this is  the  case.

     Another important complication is the long stabilization  time introduced
by an abrupt change  in temperature.   This prevents the measurement of response
by sequentially  changing  the  temperature.  We are rapidly gaining knowledge
of the causes of long  stabilization times and expect to be able to computation-
ally extrapolate stable values from initial transients.  This wilj. enable the
manipulation of  temperature in real time, producing detection  times suitable
for personal dosimetry.  An alternative would be to use several sensors, each
of which operates  at a different  temperature.  Because the temperature of each
is constant, the detector's behavior  is stable.  The only disadvantage to
using several sensors  is  the  power consumption needed to maintain the high
temperatures of  the  devices—about one-half watt each.

     Of course,  filters may be cascaded around several Taguchi Gas Sensors,
expanding the above  response  equation into a system of independent equations.
The filters need not select perfectly for and against specific gases;  they
need only produce  sufficiently different, effective sensitivity coefficients
for a computer solution to be possible.  A disadvantage of this solution is
the bulkiness of the sensors and filters and the resulting power consumption.
On the development front, we expect the integration of several sensors and
filters on one chip, obviating many of these difficulties.

     Materials research will continue to alter the catalytic activity of semi-
conductor surfaces so that solid state gas detectors will become more selec-
tive.  But, at the same time, pollution research monitoring needs are evolving
from the measurement of hazardous levels of a few simple gases to the deter-
mination of less than a part per million of many highly complex organic vapors.
To keep up with these needs, we will not only need more sensitive and selective
sensors, but also increased intelligence in the measurement instrument.

     The newly developed microprocessor technology can meet this need.


Microcomputers of  sufficient  capability and small enough size for personal
dosimetry are now  being marketed as hand-held data-entry terminals for inven-
tory control and costing  (due  to their massive memory and market considerations)
at about $900.  This  cost could easily drop, for a mass-produced instrument,
to the $30  range.   Fully  programable  scientific calculators are now available
for under $30.  The projected  downward cost of microprocessors and their
peripherals should make the microprocessor-based system increasingly attractive
for personal monitor  applications.

     We  have shown that,  at constant  temperature in an oxygen-rich ambient,
the TGS  resistance, R, varies  with  reducing gas concentration [C] as R K[C]  ,
where K  is  a sensitivity  coefficient  peculiar to the reducing gas and the op-
erating  temperature,  and   is  a power-law exponent.  It seems possible that
the dependence of  the sensitivity constants K for various gases on detector
temperature can be utilized to enhance the TGS selectivity in gas detection.
The major complication to this scheme would be the long stabilization times
involved when  temperature is  changed.  Nevertheless, a synergistic combination
of semiconductor  sensors  and  microcomputer-based signal processing is expected
to help  overcome  the  most significant limitation of present sensors—their
lack of  selectivity.

      The  authors  acknowledge  the  support of Dr. A.G. Jordan.  This research
 was  supported  in  part  by  the  Department of Energy under Contract No. EE-77-

 David T.  Tuma, Ph.D.,  and Paul K.  Clifford
 Carnegie—Mellon  University
 Pittsburgh,  Pennsylvania  15213
     (NOTE:  The  following discussion took place after the presentation of
 the papers  "Studies of  Semiconducting Metal Oxides in Conjunction with Silicon
 for Solid State Gas Sensors" and  "Microcomputer Control and Information Pro-

ceasing Technology  for Semiconductor Gas Sensors.")

     SHAW:  Bob Shaw  from EPA.  I would expect that water vapor would  be
tenaciously absorbed  on  these surfaces—much more strongly absorbed  than  the
molecules that are  of interest.  And since there can be several percent of
water vapor in a  typical atmospheric sample, the water vapor could very likely
be taking up most of  the sites on the surface.  Would you have to dry  the

     CLIFFORD:  Most  of  the metal oxides1 detectors were achieving the detec-
tion efficiency at  very  high temperatures.  At very high temperatures, there
is not much of anything  really adsorbed sitting on the material at one time,
so that the surface coverage is much less than a monolayer for water and  all
of the other gases  of interest.

     Now, we may  have trouble with a lot of water sitting on the material
condensed, as well  as maybe some other gases at the lower temperatures.   So
far we have been  operating at high temperatures, and it is not a problem.

     There is a sensitivity to water just as there is a sensitivity to other
reducing gases.

     SHAW:  I guess I missed hearing about the mechanism of operation.  It
seems to me that  if water vapor is more strongly adsorbed than many of these
other gases—and  yet  if  it is the modulation of the adsorbed oxygen that
gives you the change  in  resistivity—I just don11 see how raising the  tempera-
ture will correct this?

     JORDAN:  Of  course.  We operated at temperatures on the order of 200°,
300°, 400" Centigrade.

     SHAW:  But then  all these other gases will be desorbed way below  those

     CLIFFORD:  They  are desorbed at specific temperatures.  The surface  of
the material is very  heterogeneous.  There are adsorption activation energies
that cover a wide spectrum of energy, so that in any given temperature range
a great many gases  are adsorbed in different ways and at different activation

     SHAW:  What I  should have asked, I suppose,  is have you seen that relative
humidity does not affect the behavior of your detector surfaces?

     JORDAN:  Of  some temperatures, more;  of some temperatures, less.  Three
hundred or 400° Centigrade temperature is the range for the device with which
I am working.  The  effect is not deleterious.

     CLIFFORD:  Also, water vapor is actually detected by the sensor.

     JORDAN:  It  is another component of the mixture which you have to reckon

       BURTON:  Bob Burton, EPA.  What is the probability of getting your sen-
  sitivity down by a factor of 10?

       JORDAN:  The ppb*s rather than ppm*s?

       BURTON:  Yes.

       JORDAN:  It is fair to say that at the moment our lead oxide, tin oxide
  for leading devices, was in ppm's.  I think you can go to lower.  1 think it
  would be coupled under there.

       SCHEIDE:  Gene Scheide.  One of your original advantages was to use a
  single broad purpose sensor to sense many gases.  But in the end, are you
  saying that you have to use one sensor for each gas operating under a differ-
  ent condition?

       CLIFFORD:  We won't need one sensor for each gas.  What we need are a
  couple of independent sources of information.  That can be three or four sen-
  sors or one sensor operated at three or four temperatures.

       SCHEIDE:  You need a different sensor operating on a parameter for each

       CLIFFORD:  For each gas; that is right.

       SCHEIDE:  Do the number of variables have any relation as to the equal
  number of gases being analyzed?

       CLIFFORD:  Yes; we need at least that many degrees of freedom or more
  for independent methods.  The advantage of deconvolving this information
  through the computer is also that when we have more sources of information, we
  determine, as well as the gas concentration, the degree of validity of that
  concentration—an error down the line for every gas measured.

       JORDAN:  But remember this is not going to be coupled.  But they are
  beginning to be coupled already to the silicon chip.

       CLIFFORD:  So, we can see one sensor that eventually has mounted on it
  12 different filtering arrangements or 12 different sensors.  But by being
  swept in temperature, it gives us essentially 100 temperatures now.

       SCHEIDE:  But another one of the problems was temperature stability—

       CLIFFORD:  Yes.  Right now there arc metallurgical problems with that.
  Also, we don't really know a lot about the temperature reactions.  It may be
  that once we know a little bit more about them, that just by looking at the
  initial transient when the temperature is changed, we can predict what the
  stable value would be an hour later.

Management Strategy in the  Design and
Use of Personal Monitors for Environmental
     Ralph W. Stacy, Ph.D.
     Clinical Studies Division
     Health Effects Research Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina
     It is not  difficult to develop justification and enthusiasm for the
development and application of personal monitors for advanced  epidemiological
studies on environmental pollution problems.  Indeed, by this  time most of
the participants at this symposium are probably at a fever  pitch in their
realization of  the Importance of the whole thing.  In a situation like this,
it would be all too easy for us to go screaming off in all  directions, wasting
an enormous amount of time and effort and requiring us to come back later and
do the careful  preprogram planning that we should do at this time.  This is
a plea for careful attention to such planning at the management level.

     To begin with, what we are proposing to do is to develop  a system for
measuring and recording a variety of parameters (exposure and  health effects)
that are not easy to measure accurately, even under carefully  controlled lab-
oratory conditions.  We proposed to design these measuring  and recording devices
into a single package, small enough to accompany an individual human through-
out his or her  entire day's activities, cheap enough so that we can afford to
place thousands of these in the field, yet powerful enough  so  that we can be
thoroughly confident that the measurements made with the system are of such
quality and validity that they can be defended against the  most vigorous attack.

     One of the most important factors in this development  is  that when these
personal monitoring devices are placed in the field, they must be of one single
design for any  one complete study.  It may be that we will  eventually end up
with several devices from which we can choose one to be used in a particular
study; a little later on, I will espouse the value of the development of a

modular  system  that  can be adapted to a particular study*   In any case,  we
cannot depend on haphazardly dissociated competitive development for this
particular  task.

      To  be  sure, the individual devices or components will  surely be developed
by various  agencies—governmental, university,  and private  industry.  But in
the  long run, these  Individual components must  be assembled into a personal
monitor  within  a single, well-defined program unit.   An  agreed-upon set  of
Quality  Assurance criteria must be met, and there must be a carefully designed
and  implemented performance testing program.   Therefore,  the planning phase
of the entire development program is at least as important  as the actual devel-
opment.   This planning is the responsibility of management, and  management
must be  made aware of this responsibility at the outset  of  the program.

      I suspect  that  almost all of us here are aware  that  the program we  are
discussing  is no small effort.  Indeed, when one carefully  considers the re-
quirements  of the ultimate development and compares  these with the state-of-
the-art  in  the  several fields of instrumentation involved,  the whole affair
becomes  a little frightening.  My thesis is that we  must  somehow manage  to
impart  this appreciation of the magnitude of the task facing us  and the  diffi-
culties  involved to  the management components of every agency to be involved.
Only with the  intelligent and enthusiastic participation  of management can
the  program be  mounted at all.

      All of us  who work daily with the generation and measurement of air pollu-
tants and air pollutant mixtures are constantly aware that  accurate measure-
ment of  those pollutants is a continuing problem, even in the fixed-laboratory
situation.   The amount of time and effort spent in checking, calibrating, and
adjusting the measurement instruments is large.  Now we  are proposing to de-
velop a  system  which will measure and record those values without constant
attendance  over a considerable period of time,  and which  will be small enough
and  portable enough  so that it can accompany the individual human subject as
he or she moves around in his or her normal daily routine.

      The state-of-the-art in pollutant measuring devices  can only be described
as being in the early developmental phase.  To  be sure,  there are a number of
devices  in  existence which give promise of being useful  for personal monitor
development. When these devices are finally evaluated for  use under the con-
ditions  I have  described, some of them will in  fact  prove to be  useful,  and
others will not.  I  am virtually certain that some measurement components
will have to be developed almost from scratch.

     We remember, of course, that the personal monitor is intended for  the
continuous measurement of not only the air pollutant levels, but also those
health effect parameters that are indicative of deleterious effects of  the
pollutants.  And if we are completely honest about it, we have to admit  that
we may face more difficulty in this phase of the development than in the
pollutant-measurement phase.  The physiological parameters that are easy to
measure—the EGG, body temperature, respiratory movements, and so on—are
unfortunately not useful as indicators of health effects decrements produced
by ambient levels of air pollutants.

     In fact, the physiological variables which are indicative of pollutant
effects are very difficult to measure with the accuracy required, even in the
laboratory environment.  Let us examine just a few of these to support this

     It is well established that the most detectable physiological changes
associated with exposure to most pollutants are changes in lung mechanics,
and that most significant among these is the resistance of the airways.
This effect of pollutants on airway resistance is especially important in
people who are already suffering from pulmonary disease—asthmatics,  people
with emphysema, people with black lung disease, byssinosis, silicosis, and
so on.  Unfortunately, the measurement of airway resistance is a relatively
difficult process, especially under the conditions under which the personal
monitors are expected to be used.

     There are two ways of approaching the airway resistance problem.   One
of these is direct measurement, and the other is derived measurement  from
the analysis of air flows in special respiratory maneuvers.  For field epi-
demiological studies, the direct measurement of airway resistance is  hopeless
at the present state-of-the-art.   There are only two commonly accepted methods
for this; one is the use of a body plethysmograph, which by no stretch of the
imagination could be made into a personal monitoring device.   The other  re-
quires the measurement of the intraesophageal (representing intrathoracic)
pressure, and this involves the subject having an intraesophageal balloon
in place.  It is inconceivable that general population subjects could  be in-
duced to have such a balloon introduced and left in for the required  length
of time.  Therefore, unless someone comes up with an entirely new technique
for direct measurement of airway resistance,  we can forget that approach.
I will not say we are completely dead in this area.  It is barely possible
that the relatively noninvasive but not thoroughly proven forced oscillation
technique might be developed to the point that it could become useful  in per-
sonal monitoring.  In its present condition,  this would not be true.

      The indirect measurement gives more promise for personal monitor use.
 This indirect measurement is actually a measurement of the maximum rate at
 which the subject can expel air from the lungs.  This maximum flow rate can
 be measured with a device called a pneumotachometer, which can conceivably
 be developed into an instrument that could be clipped or pinned to the sub-
 ject and used only at intervals.  My conception of the ultimate form of this
 pneumotachometer is shown in Figure 1.  Its output is electrical, and thus
 is amenable to recording and/or analysis in portable monitoring equipment.

      The use of this pneumotachometer unfortunately requires that the sub-
 ject perform a very special kind of respiratory maneuver.  He or she must
 expel the air from the lungs at a maximal rate.  In the laboratory, he is
 coached into putting all his effort into the air expulsion, and even with such
 coaching, the maneuver is not always successful.  For field use, the subject
 would have to be carefully trained to perform the maneuver in the right way,
 and would somehow have to be programed so that he would make a maximal effort
 at every measurement time.  There would be no attendant coaching him or ob-
 serving his performance.  If he does not perform quite to the maximum, the
 indicated maximum air flow is reduced, and this sort of reduction is exactly
 what we would be looking for if a pollutant effect were experienced.  There-
 fore, it is going to be very difficult to tell whether an apparent decrement
 in air-flow capability is due to a pollutant effect or to a failure on the
 part of the subject to exert maximal effort.  It is obvious that even in this
 relatively simple sort of measurement, its adaptation to personal monitoring
 is highly questionable.

      What else can be measured that might give some indication of deleterious
 effects of air pollution?  Not much.  We have reason to believe that some
 pollutants have effects on the body's immune systems, but these cannot be
 measured in the personal monitoring situation.  Some of them may (and probably
 some do) have carcinogenic effects, but these cannot be measured in personal
 monitoring.  It is possible that air pollutants may affect cardiac performance
 in those individuals (cardiac patients) who do not have a reserve on which
 they can call in pollution-stress situations.  Therefore, the EGG probably
 should be measured and/or recorded.  Unfortunately, EGG interpretation is not
 amenable for use in the personal monitoring situation; it still depends almost
 entirely on the use of pattern recognition by trained humans.  There are, of
 course, computer programs that can be adapted to this analysis, and I am sure
 that these could be developed to make EGG analysis more feasible for personal
 monitoring purposes.

                                                      floorer cuf*
                   PNEt/MOTACUOMETER (M/T DESIGN)
FIGURE 1.  Diagram and sketch  of one way to design a transducer for use  in
a personal monitoring system.  Currently used pneumotachometers have the
pneumotach tube and screen,  the differential pressure manometer, and the basic
circuitry as separate components, and are very large and cumbersome.
     There are some fairly interesting developments occurring in electrical
impedance plethysmography which could be of value in personal monitoring.
This field, and the techniques associated with it, would have to be explored
thoroughly.  I am sure this would provide a way for continuous recording of
general respiratory movements, so that we could monitor coughing frequency,
etc.  In general, electrical impedance-measured parameters are at best only

      The measurement of body temperature is well within the state-of-the-
 art,  but is of doubtful utility in personal monitoring for air pollution
 health effects.  It might help us to detect those individuals who have an
 infection, and there is always an outside chance that ambient levels of some
 as yet unknown pollutant may have a body temperature effect.  I doubt that
 it would cost us much money or effort to include this measurement capability.

      One other possibility is the development of some measure of mental
 alertness, responsivity, etc.  Certainly some attention should be paid to
 this, for behavioral effects always lurk in the background as possible pol-
 lutant deleterious effects.

      Thus, there are not many meaningful measurements that can be made in
 the personal monitoring situation on the human himself.  We who are working
 all the time in the field are constantly on the lookout for significant meas-
 urables, and we may be required to make a concerted effort in this direction.
 Again, laboratory management components should be aware of the importance of
 this kind of development and be prepared to support it with funds, space, and

      As we all know, technology has come a long way in the production of
 integrated circuitry, both analog and digital, and in the development of
 microcomputers, digital memory, and both analog and digital recording.  My
 next thesis probably needs little defense:  the state-of-the-art is such
 that with reasonable effort and fund expenditure, all components of personal
 monitoring systems that follow the transducer can be developed in relatively
 short order.  Therefore, a major portion of the development effort will have
 to be toward the development of transducers and the procedures for their

      My own conception of the personal monitor is of one utilizing full
 modularity.  With chip design and production as it now exists, it should
 be possible to design and produce a series of integrated circuits, including
 amplifiers, signal modifiers, A-D converters, telemetering circuits, counting
 circuits, smoothing and differentiating circuits, and so on down the line.
 These would be designed to be incorporated into a packaging unit having pins
 on the bottom and sockets on the top, 30 that they could be stacked to pro-
 duce any combination of circuits desired.  One row of such connections would
 carry through the entire stack, providing power from the bottom or base unit
 to all the circuits in the assembly; another row would effectively serve as
 a signal bus, carrying analog or digital signals as required.  All package

modules would be  identical  and  could  be assembled in any combination  desired.
I visualize these as  being  approximately 1 inch X 2 inches X  1/4 inch.

     Power modules would necessarily  be larger than the circuit modules.
Thus, I have generally considered  these as being the bottom units.  The power
modules could be  snapped together  horizontally, so that several different
monitors could be incorporated  into a single package.  This general scheme
is visualized in Figure 2.

     I have also considered that there will probably need to be two types of
terminal (top) units.  One  of these would be capable of recording or  storing
in digital form the data derived throughout the period of monitoring.  This
will become more and  more in the realm of possibility as promising new de-
velopments in storage memory come  to  fruition, and when it becomes possible
to store millions of  bits of information in a couple of cubic inches  of space.
This type of storage  would  be limited to relatively small data productions,
for even if it is decided to accept 8-bit digital data, the number of data
words that can be stored has a  quite  finite limit.  This concept is illus-
trated in Figure 3.

     The other possibility  (and I  suspect this will emerge as the most im-
portant possibility)  is the telemetering of Information from the unit worn
on the person to a receiver/computer/recorder combination housed separately
in a briefcase-type of enclosure that could be set anywhere within a few
hundred feet of the monitored individual.  This unit would only need to be
moved if the individual himself moved beyond the range of the telemetering
device.  This unit could also be the  source of an alarm signal which would
alert the individual  to the fact that it was time to perform a required
respiratory maneuver, etc.  This unit might also include a voice tape record-
er on which the subject could note any special activities he engaged in—
again, by telemetering to the recording unit.

     Thus, those of us in clinical research do have a major interest in
personal monitoring, and we have some ideas for the design of such moni-
tors which we think may be of value.  At this time, we are pursuing these
ideas at a low level, and we hope  to  increase our effort in the near future.

     Ultimately, it is likely that the greatest contribution the Clinical
Studies Division of EPA can make to the personal monitoring development pro-
gram may be in the Quality Assurance procedures which will be required before
the monitors can be accepted for field use.  Not only will it be necessary

                                                     tlUf CMHKTIOHS

                                                     FtMALf CONNECTIONS
                                           PEK5OML M0ni7O&S_
 FIGURE 2.  Conception of a  modular system for design of personal  monitors.
 Each monitor assembly would serve for monitoring one parameter.   Multiple
 parameter monitors would consist of several of these fastened  together side-
                        x urrriA MODULE
                      TOiBltrCAM SUffOBTUNrr
                                                          IN MTA COLUCT VAN
                      ANALVZCO LOCALLY BY UlCftOCOMniTIM
                             ALTtRKATIVECCMEnALSCHEMO fOfl PtMQMAt MOMITOHINO
 FIGURE 3,  Two  possible ways  of handling  the problem  of data storage and/or
 recording in the  personal monitoring situation.  In Alternative  1,  the sub-
 ject would "wear"  the transmitting monitor,  but recording would  be  carried
 out in a larger,  heavier but  still portable  briefcase-size package.  In Al-
 ternative 2, which depends on not-yet-available digital storage  capacity,
 the data are stored in the monitor itself.

to show that  the  individual monitors can measure air pollutants or  health
effects parameters with acceptable accuracy and dependability, but  it will
also be necessary to  show  that  the monitors can be worn by the subject all
day long without  causing him or her undue discomfort or without restricting
his/her movements.  The monitors must be tested with all components  (environ-
mental and health) operating in a situation where actual changes in  pollutant
levels are occurring.  This kind of testing—the final step in certification
of quality assurance—can  only  be carried out in exposure chambers of the type
that we have.

     In the final analysis, only the clinical scientist can make the most
important judgment of all—whether observed alterations in the monitored
health variables  can be considered to be indicators of decrements in human
health.  That is  what the  whole program is about.  The Clinical Studies
Division of EPA wants and  expects to play a significant role in the personal
monitor development program, as well as in the ultimate program of use of
the monitors.

     The personal monitor  development program will surely involve the efforts
of many individuals from many different fields.  If any major effort were
ever multidisciplinary, this one is.  Management components of the agencies
involved, whether they are government laboratories, university laboratories,
private research  organizations, or industrial groups, must be aware of the
total requirements of the  program, and must be made to recognize that a new
and unique type of effort  is required in this program.  There must be a de-
gree of organization of the effort which has not been put together up to this
time, and this will require new management techniques.

     Without complete management involvement and collaboration in this de-
velopmental effort, the tasks will be much more difficult.  On short notice,
there must occur  some major reprograming of funding and staff assignments.
I believe that each of us must carry this message to our management, and
those of us in management must launch a major search for ways and means of
accomplishing the important job needed for the development of personal en-
vironmental and health monitoring systems.


Ralph W. Stacy, Ph.D.
Clinical Studies Division
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina  27711

The  Usage of Personal Monitors for
Determination of Dosage and  Exposure in
Health Effect Studies
     David T. Mage, Ph.D.
     Statistical and Technical Analysis Branch
     Environmental Monitoring and Support Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     The health effects of a pollutant are a function of  the amount of the
material absorbed  by the human protoplasm through exchange processes in the
lungs, the digestive system, the skin, and other organs.  For the gaseous
pollutants such as carbon monoxide  (CO) and ozone (0-), the health effects
are primarily produced from exchange processes in the lungs, although eye
irritation may be  caused by absorption directly into  the  eye.  For ozone,
eye Irritation and lachrymation may be dependent on the exposure to concen-
tration of 0_ at the eyeball surface.  However, for lung  irritation and
respiratory effects, the health effect may be dependent upon dosage of mate-
rial absorbed into the protoplasm at the gas exchange Interface.  In design
of epidemiological studies to test  hypotheses of health effects produced by
pollution, it is important for us to consider both dosage and exposure as
variables upon which the health effect is dependent.

     The following sections describe the technical considerations for studies
to assess exposure and dosage and the application of  personal monitors in
these studies.

     A health  effect of oxidants such as ozone and peroxyacetylnitrate (PAN)

 is  eye  irritation.   We may hypothesize that the severity of  the  effect  during
 the study period is a function of the number of oxidant molecules  which are
 absorbed into the eye fluids per unit time.  The rate of mass  transfer  from
 the ambient air into the eye fluid may be modeled as a function  of the  am-
 bient concentration and the physical properties of the fluids  such as solu-
 bilities, diffusion coefficients, etc.  Because these physical properties
 are relatively constant for exposed individuals, the only necessary variable
 to  be measured is the ambient concentration.  The chemical properties of
 the eye fluids will vary between individuals and cause different responses
 to  a constant exposure.  In this case, the dosage to the eye is  directly
 dependent upon the ambient concentration, and a correlation  can  be made be-
 tween the measured health effects and the exposure.

      During an epidemiological study, a person is exposed to air pollution
 in  many different ways.  For simplicity, four exposure categories  may be
 defined as follows:
      1)  Ambient (outside buildings)
      2)  Residential (indoors at home)
      3)  Occupational (at work, indoors, or out-of-doors)
      4)  Miscellaneous (in a vehicle, at a theater,  etc.).
 The average exposure X_, in the period T, can be represented by  Equation 1.

                                     = |/T Xdt                          [1]

 where X represents the concentration to which the person is  exposed in  each
 of  the four categories described above during the study period T.

      Assuming a control is available for the experiment by a measured re-
 sponse to low-level exposures, X^aO, the excess health effect  can  be defined
 AHE=f(XE)-f(0).  The excess health effect AHE can be plotted versus X£  and
 a correlation developed.  Figure 1 shows the expected results  from an experi-
 ment in which the underlying relationship is linear.  The data points will
 scatter about the regression line, and the correlation coefficient will be
 close to 1.  These results are obtainable if the exposures of  all  the experi-
 mental subjects are computed by Equation 1.  In practice, however, this may
 not be possible.  Several epidemiological studies have compared  the excess
 health effects observed in the study populations, A HE,  with  the  average am-
 bient or outside air concentration of the pollutant, X..   If we  were able
 to  measure Xg for all study subjects, the distribution of the  integrated
 exposures would have a finite variance, and X.  may or may not  be close  to
 the average exposure concentration of the n subjects in the  health study

                                  hypothetical  relation
                                  of health to exposure
                        EXPOSURE TO  POLLUTION   X
FIGURE 1.  Linear relation of excess health effect to pollution exposure.

of personal
                      EXPOSURE  X
FIGURE 2.  Hypothetical frequency distribution of population exposure to

defined as 5^  where  X£  = S  XE /n.    A distributional relationship may be as
             n         n j=l    j
shown  in  Figure  2.  The  data  from an ambient exposure experiment can lead to
a valid conclusion when  AHE is functionally related with JL  if Xf JL .   This
is  demonstrated  by the following analysis.   We can define the probability
of  a health  effect for the ith individual exposed to air pollution as the
integral  over all possible exposures of the probability of receiving an expo-
sure,  X^, times  the probability that a health effect will occur at exposure X_,
This relation can be  expressed as:

                              p(H)1 = /"p(X) H(X)  dX                     [2]

where:     p(X)  is the probability of exposure Xp,
           H(X)  is the probability of a health effect given exposure X_,
                 is the total  probability of health effect in the  individual  i.
      Let us assume H(X)=k X_.,  where k.  is a complex function of  other  vari-
 ables such as age, sex,  race,  smoking habits,  physiological  state,  and past
 exposures to pollution of the  ith individual.   Substitution  into Equation 2
                                       OD  _   —   —
                              p(H)i - /  k1XEp(XE)dXE                      [3]
 If  the health study examines n individuals,  we may estimate  the  total  number
 of  individuals m who exhibit a health effect as:
                                  m = £    p(H)                          [4]
                                   11 oo   _   _   _
 and                           m =  IS  k^pCX^dXg.                      [5]

 Since  k.  is a function independent of
 and                                 m =
 where  k is  the average value of k.  in the population n

     If  the  average ambient  air concentration, X., is approximately equal
to the mean  value Xg  ,  (X^Xg  ), then the experiment can lead to a valid
                    n        n
relationship between health  effects and measured exposures.  In the case where
   * Xj, the relationship will either underestimate or overestimate the
effect of the pollutant on human health unless there is a known consistent
relation between X. and X- .


     The design of an experiment to relate population exposure to health
effects  requires knowledge of  both the average value of k (k) and the average
value of XgCX., ).  The value of k can be determined from clinical evaluations
and knowledge of the age-health relations of the population.  The value of
Xg  requires measurement of the distribution of Xg.  One method for estimating
the mean exposure of the community is to use the technique of random sampling
(1).  An outline of a study design contains the following elements:

     1)  Estimation of the variance of the exposure distribution.  A pilot
study of selected Individuals may give estimates of the minimum and maximum
exposures as well as the variance (o ) of the distribution.

     2)  Detailed Study Design.  Based upon the estimate of the variance,  a
detailed sampling design can be made to achieve an estimate of the true mean
exposure with any combination of accuracy and confidence level desired.  The
constraints of time, cost, logistics, etc., are all considered,  and the
number of individuals to be sampled is chosen.  The result is a study de-
signed to measure the mean population exposure within AX ppm at a y percent
level of confidence.  In general,  the measured mean of  a set of n random
samples, in the limit as n becomes very large (»), is a member of a normally
distributed population with mean Xg  and variance o /n.   For small n,  the
Student's t-distribution is required to set the levels  of confidence.   For
example, if we estimate the variance of the distribution of exposures as 16
ppm  and we take 36 random samples, the mean of the 36  samples will be with-
                   1.35 ppm of the true mean, 95 percent of the time.

      3)  Conduct the experiment.  Sample the exposures of n individuals chosen
 in accord with the study design.

      The dosage, D, is the mass of pollutant that the subject accumulates
 during the study period T.  For this discussion we will assume that inhalation
 is the only significant pathway for the material to enter the body during the
 period of the study.  The dosage can be computed by Equation 8.
                             D =  Z/ T(X -X  )v#v)dt                    [8]
                                 i=l    1  eq

 Where:    X  is the exposure in the ith category, |»g/m
            body fluids
X   is the concentration of pollutant in equilibrium with the
           v is the ventilation  rate, m /min
           <^(v) is the  fraction  of  the theoretical mass transfer which is ob-
           tained, 0«£(v)X. (2).  The health effect may also be depen-
 dent on the dose received  per unit body weight, M.  For the purpose of this
 analysis, we  will assume that <£(v)=1.0, X  =0, and that AHE is dependent
 on dosage D,  although  in practice  one should compare the results of the ex-
 periment by correlating AHE with both D and D/M.

      As in the case of linearly related exposure health effects, it can also
 be shown that if excess health  effects are linearly related to D, then meas-
 urement of D  produces valid results when correlated with AHE.

      In the past, several  epidemiological studies have been performed as
 exposure experiments when  the excess health effect observed was dependent
 on dosage.  In these cases, the health effects were correlated with exposure
 instead of dosage, and, as a result, the true relation may have been hidden.

      To demonstrate the difference between X. and D, let us imagine a group
 of outdoor tennis players  (A) being observed by sedentary spectators (B)
 with a constant k for all  (A) and  (B).  During the match, the (A) group has


a ventilation rate of 30 liters/minute, and the (B) group has a ventilation
rate of 10 liters/minute.  Both (A) and (B) have identical exposure in terms
of Xj, but (A) has three times the dosage of (B).  On this day, if we study
the health of these type (A) and (B) people, during a period AT, we might
get results as shown in Figures 3A and 3B.

     The plot of p(H) versus X^ shows variable health effects with a constant
exposure, but the plot of p(H) versus 5 shows increasing health effects with
an increase of dosage.  Experiments have shown this type of relationship,
where increasing exercise levels at constant concentration leads to increasing
health effects of the subjects.  The difference between X.  and D can be
significant.  Respiration and ventilation are at a minimum during sleep,
which usually occurs at night.  During the day, when active, the respiration
and ventilation rates are higher.  This relation is directly analogous to the
diurnal variation of ambient pollution, which is usually minimum at night
and maximum during the day.  Consequently, there is a positive correlation
between v and X  and in general D/v T>X_,  where v is the average ventilation
rate.  If AHE is analyzed with X.  instead  of iL , the difference in results
will be even greater.                          n

     In order to estimate the dosage D, it is necessary to monitor the sub-
ject's activity and exposure to pollution.  A table which could allow the
computation of D for an individual may look something like Table 1.

     By estimating the ventilation rate, v, and pollutant exposure, X ,  at
each location, the integral of Equation 8  can be solved for D over the per-
iod of the study.  A random sample of people can be studied, and as their
exposures and activities are determined, the distribution p(D) can be esti-
mated.  As described previously for exposure, the necessary sample size  for
the dosage experiment is a function of the expected variation of X. and  v
throughout the study community, the desired accuracy, and the designed con-
fidence level.  By taking a random sample  of location of exposure and activity
pairs, we can determine the mean dosage of the subjects, which is an estimate
of the distribution mean.

     The treatment of health data as being correlated with X.  instead of D
may have contributed to the number of studies which produced nonstatistically
significant results.  The treatment of AHE as a function of X. instead of D
may also have contributed to the difficulty in reproducing results of epi-
demiological studies.

                        Difference  between  relations
                       of dose and  exposure to  health
               EXPOSURE  X
                  DOSAGE   D
 FIGURE 3A.   Relation of excess health   FIGURE 3B.  Relation of excess health
 effect to pollution exposure.           effect to pollution dosage.
                                   TABLE 1
                  Hypothetical Table of Activity and Location
   In car  to work
   At office—indoors
   In car  to home
Light activity
Light activity
Very active


      The  design of  epidemiological experiments needs to consider whether
 exposure  or  dosage  of  pollution is the independent variable.  When dosage
 is  the governing variable,  the correlation between exposure and ventilation
 rate  should  be  accounted  for.  Personal monitors can be used effectively to
 define the exposure distribution by a random sampling procedure, and venti-
 lation can be estimated by  knowledge of activity patterns.

 1.  Mage, D.T., Ott, U.R.  Random Sampling as an Inexpensive Means of Meas-
     uring Annual Average Air Pollution Concentrations in Urban Areas.
     Paper #75-143, 68th APCA, Boston, Mass., June 1975.

 2.  Ott, W.R., Mage, D.T.  Interpreting Urban Carbon Monoxide Concentrations
     by Means of a Computerized Blood COHb Model.  J Air Pollut Control Assoc
     28:911-916, 1978.

David T. Mage, Ph.D.
Statistical and Technical Analysis Branch
Environmental Monitoring and Support Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina  27711

     BRAUN:  David Braun, 3M Company.  I enjoyed your paper very much.   I
would like you to comment, if you would, on the lung capture efficiency,
because not all of the pollutant that we ventilate is captured by the  lungs.

     MAGE:  That is correct.  It is a variable for some pollutants.  As  a
matter of fact, if your blood has higher COHb than the COHb which would  be
an equilibrium with the ambient CO, you won't take any CO in;  you will exhale

     And the efficiency of captures of particles,  for instance,  is a function
of the ventilation rate.  1 have oversimplified here purposely in that I
assume my efficiency capture unit is constant.  It is a variable that  would
have'to be built in.

     ZISKIND:  Richard Ziskind,  Science Applications.  Do  you  think there
might be any circumstances  with  something  like  sulfur  dioxide  or another
pollutant that the  body may respond  to by  developing brochoconstriction or
something like that,  so  that you need to  take  that  type  of thing into  account
in the deposition of  these  pollutants?

     MAGE:  I am not  a  physiologist  or a medical  doctor, but I believe that
those things are considered in all the analyses of  the health  effects  which
are produced.

     Is  there anybody here  from  health effects  who  would like  to comment  on
that question?

     STACY:  Would  you  repeat the question?

     ZISKIND:  Dr.  Mage is  describing how  you  can measure  ventilation  for at
least respiration  rate  and  utilize that  to help you come up with dose.
And I was wondering that when you do get  into  an  acute exposure, an episode
of some  sort, whether there are  adequate  indices  or just that  particular  case
where bronchoconstriction can occur  or something  like  that that may alter
deposition  even  of  gases?

     STACY:   I think the answer  to that  is in  the kind of  information  that
does not exist.  The biomathematics  of the whole  situation, the whole  dosage
situation,  is very  poorly understood.  We  don't know how much  is caught up
in the  termination  of noses, etc.

     We  know  that  in animals breathing ozone,  part  of  it doesn't get down
into the lungs.

     With  participate matter, it depends  entirely on the size  of the par-
ticulate and  the amount of ventilation.   We just  don't know how to model  It

     ZISKIND:  If  you don't know how to  model  deposition,  you  can't go to
the modeling  of  the effect of deposition.

     STACY:   The biomathematics  of that  situation is extremely crude at this

     SCHEIDE:  Gene Scheide, Environmetries. You also have synergistic ef-

     WHITE:   Otto  White, Brookhaven Laboratories.  Two comments:  one  is  that
comparison  in your  model for exposure with the radiation data  and to see
whether  or  not you  can predict a thousand  dollars per  person as being  pro-
jected  as  the risk value in the  radiation Industry; the  second comment deals
with the linear  dose response—I guess  through all  your  relationships  that
there is no threshold effect.  Could you  comment  on that?

     MAGE:  I haven't looked at  the  radiational calculations that you  pointed

out, but  I would  be  pleased  to  if  we  can get  together later.

     As far  as  a  zero  threshold, this is just my assumption.   There are people
who see hockey  sticks,  that  is, no effect below a certain value.   It is quite
controversial whether  there  is  such a thing as a threshold.

     When you consider  that  you have  a control group  which can have positive
and negative deviations about their average response,  and you compare it to
people with  low exposures, sometimes  it is very,  very  hard to differentiate
between no effect  or no statistically significant effect.

     Many times we have to have very  large populations in order to get the
power of  the test  to discern it.   I have just chosen  the  simplest  model.

     STACY:  Dave, 1 would like to hear you defend the statement you made a
little earlier—that because most  people are  awake in  the daytime, it would
be better to have  high  pollution at night and low pollution in the daytime.

     MAGE:   OK, 1  will  defend it to you as a  member of the population who
is concerned about the  total number of people in  the country  who are exposed.

     STACY:  This  is the question  I am really asking:   Should we be concerned
about the average  individual and his  average  exposure? Or should  we be con-
cerned about the worst  case  exposure?

     MAGE:   Let me give you  a very clear definite answer:  yes and no.   Yes;
we should as human beings.   No; not if we are starting to  look at  economic
evaluations, cost, and  benefit.  I mean,  if there is only  one person in the
whole United States  exposed  to a particular level and  it  produces  a health
effect, it might be  cheaper  to give him or her $50,000 to  move across town
rather than  spend  a  billion  dollars to clean  up the stack.

     These are  decisions that somebody other  than myself will have to make.
But these are also considerations  that we  have to look into.

     DOCKERY:   Doug  Dockery, Harvard.   You are concerned about the positive
correlation  between  inhalation and air pollution  levels in that you would
breathe harder  during high exposure.

     I can't quantify this,  but looking at our indoor-outdoor data and  our
personal  monitoring  data, in my experience people are  involuntarily respond-
ing to the bad  air pollution levels by doing  things to clean  up their homes.

     And  what we are seeing  on the whole  when we  look  across  all our cities
where we  supposedly  have a large difference in the air pollution levels—when
we look at the  indoor levels and what  their exposure is—we see very, very
small differences  between this, because  there  has been this response of
people not wanting to dust as often.   So,  they are closing up their houses
and making other changes in  their  lifestyle to ameliorate  their air pollution

     And  I think the actual  case is just  the  opposite  of  what you  are con-
cerned about.


     MAGE:  The  positive  correlation I  referred  to arises  because  people  are
awake during  the day  and  breathe  at a faster rate.   When pollution is  higher
during daylight  hours,  there is a positive correlation.

     CHAPMAN:  Robert Chapman,  EPA.  Dave, I was concerned where the previous
discussion  got left off by leaving in people's minds the impression that  in
certain  types of epidemiological  studies,  it wouldn't be very useful to have
an ongoing  indication of  what people's  respiratory rate  was.

     I think  it  would be  useful—very useful—in certain types of  epidemio-
logical  studies  for this  reason:   no matter whether one  gets  a clear indication
of absolute dose from multiplying exposure times respiratory  rate,  one still
can, 1 think, get  some decent indication of relative dose  between  or among
the people  in the  study on the grounds  that the  same type  of  pollution—and I
realize  what  I am  about to say is challengeable—within  limits can be  expected
to affect all study participants  in more or less the same  way.

     I know that there are a lot  of holes in that statement.   Still in all,
I think  that  the uniformity of response would be enough  so that if you do
concentration measurements over the course of these people's  exposure  time,
as well  as  their respiratory rate over the same  time, you  would get a  con-
siderably better indication of relative dose than we are able to get right

     MAGE:  Yes; I agree.

Carboxyhemoglobin Determinations from
Expired Air
     Harold W. Tomlinson
     ladec, Inc.
     Albany, New York
     Carbon monoxide  is the most widely distributed and the most commonly
occurring air pollutant.  Total emissions of  carbon monoxide to the atmos-
phere exceed those of all other pollutants combined (1).  Carbon monoxide
combines with hemoglobin to form carboxyhemoglobin (COHb).  A new Instrument
called the Carboximeter was designed to measure  this.

     Hemoglobin has an affinity for carbon monoxide over oxygen of greater
than 200 to 1;  thus,  it displaces oxygen, resulting in an interference with
the oxygen transport  system of the body.  It  also moves the oxygen dissoci-
ation curve to the left, binding the remaining oxygen more tightly.  The pre-
dictability of carbon monoxide absorption and elimination in man makes the
test for COHb a method of estimating ambient  carbon monoxide levels to which
a subject has been exposed.  DHEW Secretary Califano's Office on Smoking and
Health has started emphasizing the adverse health effects of carbon monoxide
from tobacco smoke, which was highlighted in  the recent Surgeon General* s
report.  COHb and  smoking are dose related, and  the Office on Smoking and
Health now uses a  Carboximeter to demonstrate this immediate biological ef-
fect of smoking.

     The heart appears to be the most vulnerable target organ for elevated
COHb.  Medical research has provided a new role  for low-level COHb in heart

     Some of these adverse health effects have been measured at only slightly
elevated COHb levels  (2).  For Instance, levels of about 2 1/2 percent have

been shown to aggravate angina and depress  the  forcefulness  of  contraction.
Levels of about 3 percent have been  shown  to  aggravate  intermittent  claudi-
cation.  Electrocardiograms  have  been  altered in  normal subjects  at  about
8 1/2 percent COHb.  Thus, with this kind  of  medical  evidence available, the
measurement of COHb  has become increasingly important.   Most of the  well-
known symptoms of carbon monoxide poisoning such  as headache, nausea,  etc.,
do not occur below the 15 to 20 percent  COHb  level.   Thus, we are usually
unaware of elevated  levels in  these  lower  ranges,  and a diagnostic test  is
necessary to make that determination.

     Traditionally,  the test for  COHb  has  been  difficult to  make.  It  re-
quires drawing blood and performing  blood  gas analysis  on equipment  that re-
quires expert calibration and  highly trained  personnel  to operate it.  It is
usually in a location remote from the  patient,  resulting in  delays from
sample to results.   It may not always  be convenient to  take  a blood  sample,
and  the sample must  be properly prepared and  transported.  All  in all, these
traditional  tests are not suitable  for field  use  or for applications where
portability  or on-the-spot results would be necessary.   Thus, the rarity of
obtaining COHb measurements, even when they might be  useful. An  alternate
method of obtaining  rapid COHb levels  is clearly  needed. The well-known
Haldane equation  expresses the relationship between alveolar CO and  COHb.
This relationship was used by  Dr. Sjostrand (3) to determine COHb from al-
veolar CO.

      In 1958, Dr. Jones (4)  demonstrated that CO  tension in  the alveoli  will
approximate  the CO tension in  the pulmonary capillaries using a 20-second
breath holding technique.  In  this  technique, the air in the lungs is  ex-
pired, then  a deep breath is held for  20 seconds.  It is necessary to  expire
all  the air  from  the lungs and collect the last portion of air  from  the  al-
veoli.  The  alveolar CO is then measured.   Using  this technique,  Jones demon-
strated an almost linear relationship  between alveolar  CO and COHb.  Several
other studies using  the Jones  20-second  breath  holding  technique  have  cor-
roborated this relationship.  Two curves derived  from such studies are shown
in Figure 1 and compared to  the curve  of Dr.  Jones.   One is  from  Dr. Peterson
(5)  and the other from Dr. Stewart (6).  These  are best-fit  curves from  mul-
tiple data points.   Dr. Peterson  derived the  formula  shown,  and it is  this
formula which we have selected to use  for  conversion  of alveolar  CO  to percent
COHb.  With  this alveolar CO-COHb relationship  well established,  it  was  nec-
essary to develop an instrument that would  measure alveolar  CO  accurately
and  reliably, not only under lab  or  clinical  conditions, but in field  situ-
ations where portability is  important.

                                               JONES, R.H.«al
                                     ff *COHb ' V 108.08 + 7.6 COA -11.89
                                   PERCENT COHb
FIGURE 1.   The  relationship between alveolar CO and COHb after 20-second
breath holding.
     Theoretically,  any instrument that can measure CO in air could be used
to measure COHb  by the expired-air technique.   However, there are many po-
tential areas  for error if an ambient CO monitor is used.  The Ecolyzer,
with its accuracy, ease of calibration, and portability, has become one of
the most popular instruments for measuring CO  In ambient air.  Indeed, it
has been used  to measure expired air CO for conversion to percent COHb.

     The sensor  has  excellent linearity accuracy and repeatability, and a
minimum detectable sensitivity of .5 ppm.  Thus, it has the ability to re-
liably read  low  concentrations and to measure  normal COHb levels.  For ex-
ample, 1 percent COHb—the approximate average normal for nonsmokers—repre-
sents only about 7 ppm of CO in expired air.   This is a result of red cell
destruction  and  breathing naturally produced CO in ambient air.

     Recognizing the need for a new instrument to measure COHb by the expired-
air technique, we joined with Energetics Science to design this new instru-
ment with the  advantages of the sensor but without the disadvantages of the
ambient air  monitor  for this procedure.  We designed out several areas of

 potential  error possible when using ambient air monitors for making the
 COHb  test.

      Figure 2 is a flow diagram of the system.  The sample chamber is de-
 signed  to  collect the last 20 cc's of air blown into it.  This is important
 because the accurate conversion of alveolar CO to percent COHb requires a
 good  alveolar air sample.  Using bags for the collection of expired air
 samples risks dilution of alveolar air and exposing the sample to ambient
 CO.   Our system eliminates the use of bags.  The breathing technique for
 blowing the sample into the sample collection system is also simplified.  How-
 ever,  nonalveolar breath samples taken without the breath holding technique
 are  usually about 10 to 15 percent lower—close enough to make the Instrument
 useful  in obtaining immediate confirmation of CO exposure in uncooperative

      Our readout is directly in percent COHb.  We eliminate the need to con-
 vert the readings of ppm of CO to percent COHb, thus reducing the potential
 for  error.  When using the ambient air monitor, we found it quite common for
 a reading of, say, 20 ppm of CO to be recorded as 20 percent COHb.

      We have added an automatic scale change on our dual scale.  One scale
 is zero to 10 percent COHb, and the other is zero to 33 percent.  When the
 reading goes over 10 percent, the needle will automatically move to the larg-
 er scale,  and a light will indicate that that scale should be read.  When
 the  instrument is reset, it automatically goes back to the smaller scale.
 This of course reduces the risk of reading the wrong scale.  A sample-and-
 hold circuit allows for proper reading and recording results.  The peak CO
 in the sample is measured and the meter indicates percent COHb, holding the
 reading until the instrument is reset.

      We isolate the alveolar air sample by use of a zero filter on the out-
 board end of the sample chamber.  Thus, when we pump the sample from the
 sample chamber to the sensor, all ambient CO is removed, and the only CO to
 reach the sensor comes from the lungs.

      We have moved the high voltage outside the case with a UL-approved
 battery charger, but we have retained the portability with rechargeable
 batteries.  The instrument can be used in its portable mode for up to 8

      There are other design considerations that I will mention briefly.

             SAMPLE INPUT

                                                     f OVERFLOW EXHAUST


                                                     f^tf+W*1 ••••••••••Mi
                                      SAMPLE    pi i TCD
                                      CHAMBER   FILTER
                            ANALYZING MODE
FIGURE 2.   Flow diagram of the system.

We  have  angled the meter to make it easy to read and have added a carrying
compartment in the cover.   We have gone to a simple pushbutton operation,
and we have even added a 20-second timer on the face of the panel so  that
the 20-second breath holding technique can be performed without a clock or
watch.   We calibrate with 20 ppm of CO, which converts to 4.3  percent on
our top  scale.  The instrument weighs only 10 pounds, and it is still pos-
sible to use the bag-sampling technique if the instrument and  the subject
cannot  be brought together.

      Some of the uses for which the Carboximeter is being used include:
measuring shift workers before and after work to determine GO  exposure on
the job, measuring the rate of heme catabolism, counseling smokers, and
correcting for COHb error from total saturated blood.  It is important to
remember that even a small dose of CO can be detected if it is combined with
hemoglobin.  Carbon monoxide in the ambient air never hurt anyone. When
attached to hemoglobin, it becomes very dangerous.   The Carboximeter  can
be carried to any location to determine not how much CO is in  the ambient
air around the subject, but how much CO was actually inhaled by the subject
and combined with his or her hemoglobin.

      In summary, this Carboximeter development project has resulted in an
 instrument that is lightweight and portable, is accurate even  at low  COHb
 levels,  and will readout directly in percent COHb with a simple 30-second
breath test.

  1.  Department of Health, Education, and Welfare.  Air Quality Criteria for
      Carbon Monoxide.  March 1970, 2-1.

  2.  Goldsmith, Jr., Aronow, W.S.  Carbon Monoxide and Coronary Heart Disease:
      A Review.  Environ Res 10:236-248, 1975.

  3.  Sjostrand, T.  A Method for the Determination of Carboxyhemoglobin
      Concentrations by Analysis of Alveolar Air.   Acta Physiol Scand 16:
      201-297, 1948.

  4.  Jones, R.H. et al.  The Relationship between Alveolar and Blood Carbon
      Monoxide Concentrations During Breathholding:  Simple Estimate of COHb
      Saturation.  J Lab Clin Med 51:553-564, 1958.

  5.  Peterson, J.E.  Postexposure Relationship of Carbon Monoxide In Blood
      and Expired Air.  Arch Environ Health, 21(9), 1970.

 6.  Stewart, R.D.  et  al.  Rapid Estimation of Carboxyhemoglobin Level  in
     Fire Fighters.  JAMA 235(4), 1976.

Harold W. Tomlinson
ladec, Inc.
1A Lincoln Avenue
Albany, New York  12205
     MAGE:  Is there any restriction in the amount of expired breath that is
necessary?  I am interested in school children; how small a child can use
this device and this technique?

     TOMLINSON:  One of the reasons we made the sample chamber so small is
for that very reason—so that we would not require a large sample.  Almost
anyone who is capable of breathing into the instrument, regardless of the
size of his lungs, can get a big enough sample, a 20 cc sample, out of it.

     As a matter of fact, this is being used in schools right now for smoking
education.  And the kids love it.  They have no problem using it.  They blow
into it, and we are getting excellent results in that kind of application.

     ZISKIND:  Richard Ziskind, Science Applications.  A couple of questions:
first, could you tell us anything about the school children levels of CO
you are defining compared to an adult level?

     TOMLINSON:  We have found that generally speaking, the carboxyhemoglobin
levels are—with the kids who smoke—above what we would find in adults.
I don't have all the data with me, but that is the overall look of it.

     What is amazing is that the smoking age keeps going down.  We get kids
smoking in the fourth, fifth, and sixth grades.

     ZISKIND:  Another question:   I have seen one study that indicates a varia-
tion in regression coefficient between alveoli CO level and carboxyhemoglobin
which suggests that before doing any survey—especially if you are going to
look at many people and contrast, say, two groups such as control and ex-
posure—that they would recommend direct blood analysis be made at least on
some proportion of that population that they are going to look at.

     Would you think that is not necessary?

     TOMLINSON:  Well,  we  have done quite a number of tests  on fairly large
groups of people.   We have done bloods on the original development of this
with very good  correlations.

     As a matter of fact,  Dr.  Anderson at the University of  Louisville has
two of these  instruments.   He  reported to me last April that he was getting
a  .96 correlation  of bloods done on the new IL-282.   The old 182 was not very
reliable for  COHb,  but  the 282 is pretty good.

     ZISKIND:   What is  the price?

     TOMLINSON:  $3,500.

     SPEN6LER:   I  have  done a lot of carboxyhemoglobin tests as you described
with an Ecolyzer and found that 1 had a lot of problems with someone who had
recently used an alcohol  mouthwash.  If any volatile stuff was ingested pre-
vious  to the  test,  an instrument would give a positive reading for a long
period of  time—sort of poison the action for a while.

     Have  you eliminated  that problem?

     TOMLINSON:  The interference filter in the mouthpiece assembly is pri-
marily for alcohol.  It will absorb a lot of it.  There is a simple test to
determine  when the filter begins to pass alcohol, but that is in calibration
gas  through it.

     But you  are right; it will do that.  If a guy has a "two martini lunch,"
even the mouthpiece filter will not take it out if it is used after the in-

     That  is  a problem.  Particularly in exhibitions, in trade shows, people
come around after  lunch and say, "I want to blow into your machine."  And
it is very difficult to explain their responses.

 Intelligent Personal Physiologic  Monitors in
 Clinical  Environmental Health Effects
 Research (Part I)
     Mathew L. Petrovick
     Physiology Branch, Clinical Studies Division
     Health Effects Research Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     The lack of a definitive data base in urban environmental health ef-
fects research (2,9,12,13,18) today is based to a large extent on systems
and devices representing  extensions of hospital-oriented biomedical instru-
mentation*  These devices are frequently designed as vital sign detectors
for patients with well-developed physiologic symptoms; i.e., known cardiac
history, emphysema,  post-surgical recovery, and the acutely ill.  Devices
of this type serve in life-saving roles and need not have the sensitivity
to detect subtle physiologic changes.  Detection of subtle physiologic
changes in a normal,  healthy, urban population requires considerable clinical
judgment as a means  of screening out false positives.  As such, conventional
biomedical instrumentation may not be adequate for assessment of subtle
changes in normal, mobile, healthy adults exposed to environmental pollutants.
Conventional biomedical Instrumentation systems may be useful in clinical
health effects research as baseline reference measuring devices.  However,
the subject must be  brought to the clinic, thus limiting the population
under study.  The problem in urban personal monitoring requires incorporating
the biomedical and pollutant instrumentation on-board the attire of the test
subject as a means of data collection (12).  The bulk, size, weight, power,
and lack of subtle detection sensitivity, and the lack of portability of con-
ventional hospital-like instruments, are incompatible with clinical epidemio-
logic research.   Ideally,  health effects protocols designed to search out
dose-response relationships must be extended from the clinical research lab-

 oratory to the field through implementation of properly designed "personal
 monitors."  The rationale of present-day clinical protocols does not lend it-
 self to personal monitoring, in view of the high degree of test-subject
 mobility and normal function in daily activities.

      The above limitations have precluded the design and implementation of
 clinically useful systems and, in effect, have limited the urban health ef-
 fects data base.  In order to circumvent the above limitations, a quantum
 jump in the scientific design of biomedical instrumentation, transducers, and
 pollutant dosimeters is imperative.  Successful clinical epidemiologic re-
 search, therefore, must be based on highly individualized noninvasive tech-
 niques, interactive software systems, transducers, and highly sophisticated
 systems managers in the form of microcomputers, all on-board the mobile ur-
 ban test subject as exemplified in Healthe-Belt (12).

      It is, therefore, the objective of this paper to establish a series
 of basic and fundamental guidelines toward the design of environmentally
 useful personal monitors, as opposed to classic hospital-like devices used
 in clinical research.  In view of the limitations of previous technology
 (9,13), this design initiative is set forth herein through illustration of
 several design examples requiring a significant change in operational and
 clinical philosophy.  Several new and original concepts are introduced which
 take advantage of the modern measurement technology, very large scale inte-
 gration (VLSI), and microcomputers (14,15,17), today and in the near future.

      Dose-response measurements on a normal individual performing daily ac-
 tivities or work tasks within his/her environment must be based on two major
 elements in order to represent credible and scientific health effects data.
 These elements are the simultaneous measurement and documentation of pollut-
 ant dose level and concomitant physiologic response.  This philosophy must
 be recognized as fundamental to any health effects research (9,12,13).
 Further, this concept is  the cornerstone of any management strategy in the
 design of any personal monitor system In support of scientifically credible
 legislation.  Since design of pollutant dosimeters is an extensive subject
 beyond the scope of this  paper (18), only the physiologic data acquisition
 system designs will be addressed herein.

      Clinical epidemiologic requirements for personal monitors are broad and

extensive.  In order  to  introduce  and  establish  the  roots  of  a  new design
philosophy (4,11,12),  a  series  of  design  priorities  are  identified to  illus-
trate the need, criteria,  and methodology in attaining a quantum  jump  (11)
in system design  required  for EPA  to fulfill its  part in any  Public Health
Initiative through  intelligent  personal physiologic  monitors.   Design  of
personal monitors must include  the following priorities:
     •  Test subject  comfort and wearability;
     •  Design of clinical data base;
     •  Design of new physiologic  parameters;
     •  Design of new multimode transducers;
     •  Design of new algorithms to replace conventional electronic circuits;
     •  Design of clinical field logistics and the means to clinically vali-
date and characterize  system performance  with clinical and air  pollutant

     The EPA/Health Effects Research Laboratory response to the Office of
Health Effects mandates imposes the need for significant scientific flexi-
bility in collection of desired data base knowledge in support of preventive
environmental medicine and proposed legislation.  This requirement, coupled
with the advancements of medicine and technology, creates the need for the
ever-changing variety of multlvariable protocols which have become a fact
of life in health effects research.  Conversely, no organization can afford to
re-tool its scientific resources each time a new research issue or pollutant
episode develops.  This is particularly critical in assessing environmental
episodes and concomitant health effects data.  Clinical research in the past
has advocated the design of new systems and devices to suit specific protocols
wherein the design cycle time is Incompatible with pollutant episodes and the
need to characterize related health decrements.  Frequently, when such systems
transfer from the development stage to the clinical-use stage, they are often
obsolete or the protocols have been changed to answer new research questions.
Therefore, utilizing conventional hospital biomedical instrumentation in this
type of research climate would render these systems almost useless.  In order
to illustrate the type of design flexibility envisioned by the clinical in-
vestigator, the following criteria depict a few guidelines necessary for a
quantum jump in design of IPPM systems.

User Design Criteria

     1)  Design a legislatively defensible data base as a first-line scien-


 tiflc priority before any system designs are attempted.

      2)  Design means for software  flexibility  in protocol changes so as  to
 support the data base product.

      3)  Provide protocol flexibility  through adaptive software designs and
 elimination of hardware wherever possible.

      4)  Incorporate modularized prescription design for ease of protocol

      5)  Provide means of protocol  replication  by others through very large
 scale  integrated circuit memories erasable  programmable read only memories

      6)  Design anatomically  comfortable transducers which provide a natural
 interface  to male  and female  test-subject anatomy.

      7)  Design multimode transducers  to acquire more than one parameter
 from a given anatomic site  and  reduce  subject encumbrance.

      8)  Design "intelligent  transducers" which are capable of sensing phys-
 iologic data and converting these data into meaningful clinical measurements
 prior  to reporting on call  to a higher-level system.

      9)  Incorporate multimode  statistical  data base at the system level  to
 avoid  extensive redesign of data base  at the end of a given study.

     10)  Design standardized  modular systems and sensors which are Inter-
 changeable and adaptable to several protocols and/or research questions.

     11)  Maintain  fixed hardware while software provides the key flexibility
 and is varied to suit new research  questions.

     12)  Ensure that system size, weight, power, and interfacing are com-
 patible with normal human activities and acquisition of scientific data.

     13)  Ensure cost effectiveness  through  anticipatory design in support of
 high-volume disposal IPPM systems.

     14)  Replace conventional (TTL) circuit logic with microcomputerized
 algorithms performing circuit logic functions wherever possible.


    15)  Design  all  systems  in  direct  anticipation  of  changing  needs  and
minimize technological  obsolescence when  protocols  or  programs  are  changed.

     Intelligent  personal  physiologic monitors represent a significant  depar-
ture from conventional clinical instrumentation in that IPPM devices (11,12)
are designed  to respond to noninvasive environmental health effects needs
and not life-saving  functions.  The key design factors identified above are
reflected and  illustrated  in  the design approach described here.

     The design approach to IPPM systems  and devices must be based on multi-
level modular  designs geared  to the ever-changing and accelerated technology
of today and  to that of the next 5 years.  It is not uncommon today to initi-
ate a given systems  design effort only to find protocols were changed or be-
came obsolete  18 months later when the device was completed.  It is, therefore,
imperative that the  design structure advocated must be constantly and contin-
uously adaptable  to  changing needs throughout the design and development
cycle.  The most  effective approach to adaptive design is through flexible
microcomputer  software.  Previous technology (9,10,13) has not been respon-
sive to this design  philosophy and has resulted in a multitude of surplus
instruments or hardware designs for each  protocol or each new research ques-
tion.  Conversely, hardware logic has its place in many endeavors.  However,
the bulk, size, power, and weight of sensors and instrumentation would pose
considerable constraints on human anatomy by conventional design methodology.
Clinical criteria, changing protocols, new research questions, cost effective-
ness, and the  need for mass clinical epidemiologic screening of the population
demand a "revolutionary change" in overall design approach of physiologic data
acquisition systems  (PDAS) for any clinical research.  The major attributes
of any PDAS system design  (12) responsive to the above criteria and, specifi-
cally, to health effects personal monitors, are "clinical flexibility,"
"test subject comfort and  acceptance," and "cost effectiveness."  The design
of these attributes  begins with interfacing the human anatomy to the phys-
iologic transducer (8).

     Considering today's technology, particularly very large scale integra-
tion of the next 3 to 5 years (14,15), the most promising approach in design

 of personal monitors  is  clinically interactive  software  which emulates  hard-
 ware logic functions  in  concert  with human judgment.   By imparting inter-
 active human  judgments or  clinical rationale to the processing of  health
 effects data, we  impose  a  level  of human intelligence  as an inherent  function
 of the system design.  Hence,  the name  "intelligent personal physiologic

      The  efficacy of  any PDAS  system is based on the  type of analog signals
 available which  represent  the  range of  human variance  to be analyzed  and
 documented in a  given data base  design.

      IPPM devices are unique in  that five major constraints are imposed on
 the  designer  and  the  clinical  researcher.  These are:

      •  All  transducers  and measurements must be noninvasive and painless.
      •  The  need  for  subject freedom, comfort,  and  ease  of wearability
 without  loss  of  scientific data  is imperative to the  success of IPPM.
      •  Significant limitations  exist in the interfacing of noninvasive
 transducers  to  both male and female anatomy in  search  of valid physiologic
      •  Size  versus function is  a matter of technological trade-offs  as
 contrasted against a  definitive  data base design.
      •  Cost  effectiveness must  permit  mass production in order to reach
 statistically significant  populations in support of any  EPA Public Health

      The  significant  constraints of human anatomy demand and impose the ac-
 quisition of  multimode physiologic transducers  from the  minimal number  of
 anatomic  placements;  i.e., multimode transducers (11)  should provide  two
 or more  parameters from  a  common transducer.  Hence,  time-shared transducers
 represent a  realistic approach in multiparameter sensing,  which may jointly
 aid  in characterizing physiologic decrements resulting from air pollutants.
 However,  multimode transducers Impose a dilemma of  excessive amounts  of data,
 that are  frequently beyond human comprehension,  being  produced.  The  signifi-
 cant need herein  for  an  IPPM is  the ability to  screen  multimode parameter
 variance  based on previous human judgment and clinical criteria to the  extent
 of reducing  transducer outputs to an interim-processed level.   This endeavor
 may be feasible  through  specifically designed "intelligent physiologic  trans-
 ducers" (IPT) (4,11).  IPT devices serve to screen  out irrelevant  data  and
 only utilize  responses which have met a level of interactive Judgment through
 the rationale of  clinical  software protocols.  The  criteria and clinical
 rationale would be a  function  of the microcomputer  algorithms  within  the


 transducers  themselves.   Hence,  in higher-level PDAS memory systems,  such an
 IPPM would not  be saturated with useless data, thus clogging the support
 of an effective and  defensible data base—as  is the case  in some clinical
 laboratory PDAS systems  in existence today.

 The Concept  of  Transphysioware

     The  design of IPT devices advocated here is based on transducing  (8)
 a given physiologic  signal from  the human anatomy  into its  analog  form.
 However,  the need is to  convert  the analog signal  into some meaningful clini-
 cal measurement which requires a form of clinical  judgment;  i.e.,  transduce
 a physiologic function signal  into a meaningful clinical  measurement.   Im-
 parting the  latter judgment to a transduced signal can be accomplished,  within
 limits, through an appropriate algorithm.  This process is  defined here as
 "transphysioware" (11).   This  function is advocated at the  transducer  level
 by incorporation of  a microcomputer within the transducer itself.   Ideally,
 limited screening of data would  occur at the  transducer level, thus simpli-
 fying the process function of  the IPPM whose  priorities and memory capacity
 is spread over  several other physiologic and  air pollutant  dosimeter para-

     Advocating a specific philosophy in any  endeavor is  best  illustrated by
 realistic examples which address the principles, problems,  and efficacy of
 design integrity.  The key design factors identified above  are reflected  and
 illustrated  in  the following example designs.

Multlmode Respiratory Intelligent Personal Physiologic Monitor

     Preliminary pulmonary function studies within EPA suggest that exercise
respiratory tidal volume and heart rate provide a means of characterizing
pulmonary reactivity to pollutants to the extent that subtle changes in over-
all lung function are detectable.  Clinical studies currently underway in  the
EPA CLEANS facilities have served to develop the clinical rationale in res-
piratory tidal volume measurement.  However, these protocols are precluded
from use at an IPPM level.  The design issue is:  how to extend a respiratory
tidal volume protocol to a personal monitor level.  A preliminary design of
such a system at an IPPM level is illustrated in the subsequent design.

      A dedicated single chip 2K memory microcomputer (Intel 8748/8749) and
  supportive preprocessing are employed in a preliminary IPPM design as follows:

      1)   A pneumotachograph for measurement of respiratory flow is incorpo-
  rated  into a construction worker's helmet.

      2)   The transducer is normally recessed in a frontal storage area and
  is  pulled out with the index finger by the subject on software auditory
  beeper command.

      3)   The subject places the pneumotachograph in his mouth (for a pre-
  determined period) and continues performing his duties.

      4)   The microcomputerized IPPM is preprogramed to present the subject
  with a previously validated protocol based on a sequence command after the
  pneumotachograph is placed in the subject's mouth.

      Figure 1 illustrates a timing chart of multimode parameters desired in
  a typical exercise tidal volume protocol wherein only one transducer is used
  for pulmonary measurements.  (With multimode parameters, one anatomic place-
  ment is used.)  The clinical interactive rationale provided in the software
  algorithm provides four parameters; i.e., respiratory rate, volume, flow, and
  tidal  volume, or a total of four parameters from one transducer placement.
  The algorithm converts respiratory flow to volume by mathematical integration
  of  the flow signal.  Each flow breath rate and volume signal is stored in
  RAM memory and displayed in an accumulative form over a 1-minute period,
  resulting in a measure of tidal volume (see Figure 1).  On a separate channel,
  heart  rate trend is recorded and constantly compared to the prescribed clini-
  cal safety level within the algorithm for a given subject's age and maximum
  heart  rate.  Hence, the 8748 microcomputer algorithm performs an interim
  level  of processing, thus providing the IPPM with trend information permitting
  clinically meaningful judgment in detection of subtle changes; i.e., a trans-
  ducer  has sensed the physiologic signal, and a microcomputer algorithm has
  converted one signal into four meaningful clinical parameters (transphysioware)
  as  health effects indicators.  If preset trend levels exceed prescribed levels
  of  heart rate, a flag and alarm to "halt" the test subject from further ef-
  fort are initiated, and the IPPM accesses the respiratory IPT for relevant
  data in support of a dose-response data base.

      Figure 2 illustrates a preliminary record of the actual 8748 microcom-
  puter  controlled analog signals and digital measurements.  Figure 3 illus-

                                        PULMONARY TIDAL VOLUME PROCESSOR.
                                                %/AI I'rTATiriM AKIAI r\f* CIOMAI C
                                                VALIDATION ANALOG SIGNALb
                 TIDAL VOLUME
              RESPIRATORY RATE-
                   HEART RATE-
                    INCREASED TIDAL-
                 *-WORKERS' EXERCISE-*'
                    STRESS PERIOD   J

                   DECREASED      ,'
                   TIDAL VOLUME    i
                                 •  DECREASED VOLUME
                                 '  INCREASED RATE
                                                     INCREASED FLOW RATE
                     (*—ALARM PERIOD—*j
                                           MAXIMUM HEART
                                           RATE THRESHOLD
FIGURE 1.  Ideal pulmonary function parameters from one transducer, 8748
microcomputer algorithm process.  Heart rate, separate transducer.

                        i M -I I I f 41 I  i !- i '  ! '  : :
                        -SPEED 1 mm/sec -^-
                   t i
                                  III  HIM  i!  II  '
                         TIDAL VOLUME
FIGURE 2.  Actual human  pulmonary record.  Top,  respiratory  rate  (digital).
Middle, respiratory flow.  Bottom, respiratory tidal volume.   See calibration
curve, Figure 3.

      7.5 T
L    2.5
          ..   I
                    0   20  40  60    80  100  120  140  160  180  200
                              ASR-33 DECIMAL TELETYPE OUTPUT
 FIGURE 3.  Respiratory pneumotach calibration based on Fisher-Porter Flow
 Meter Model 10A1027A.  System calibration  includes A/D Converter and 8748
 FIGURE 4.  Graphically derived 8748 microcomputer output for 1 minute of
 tidal volume at rest.

trates the calibration  of  the  pulmonary tidal volume IPPM system used  to  pro-
duce the record of Figure  2.   Figure 4 illustrates the microcomputer output
parameters computed during a 1-minute sample.  Figure 5 illustrates a  possible
configuration of a pulmonary IPPM built into a construction worker's head-
gear.  Figure 6 illustrates design details of the exercise pulmonary tidal
volume processor contained in  a construction worker's helmet.

Multlmode Cardiopulmonary IPPM

     Static cardiovascular measurement assessment of young normal test sub-
jects exposed to air pollutants may be difficult in terms of detecting subtle
change responses.  Conversely, as in exercise pulmonary function, potential
for cardiac dose-response processing may be feasible through the appropriate
selection of the parameters and trend analysis.

     Several cardiac parameter measurements (1,3) may be required in order
to accumulate a sufficient battery of data to establish correlation of subtle
trends or a dose-response index following pollutant episodes.  As an example,
a current EPA system development study involving acquisition of cardiac
(treadmill exercise) systolic time intervals (STI) (a measure of myocardial
contractility) has demonstrated that STI measurements may be highly applicable
for field IPPM systems.  The clinical validation of cardiac impedance meas-
urements has been well documented in the literature by Kubicek, Klzakevich
(5,6,7), and others.  Definitions are beyond the scope of this paper and are
contained in the referenced literature (5,6,7,16,19).  The multiplicity of
cardiac parameters available from impedance cardiography currently in de-
velopment at EPA/CSD clearly indicates the potential for use in IPPM systems.
(See Table 1.)  Figure 7 illustrates a male electrode configuration for im-
pedance cardiography wherein STI provides an index of aortic impedance blood
flow velocity and aortic left ventricular ejection time (LVET) (5,6,7).
From these two basic cardiac measurements, it is possible through clinical
rationale to develop several parameters which give further insight to vari-
ations in normal cardiac performance.   Figure 8 illustrates a real time
microcomputer display of heart rate during a "Bruce" exercise protocol.  The
display incorporates a medical-legal safety monitor for heart rate (see 85
percent max) and a heart-rate trend throughout the protocol period.

     An Intel 8010 single board microcomputer functions as a system controller

                                 (CONSTRUCTION WORKER)
               PNEUMOTACH ^*




  FIGURE 5.   Proposed  tidal  volume IPPM configured into construction worker's
  helmet.  Note  details.
                                    MARK I PROTOCOL
                 I 8748
                 MEMORY & SIGNAL
                    ALL CIRCUITRY IS BASED
                    ON VERY LARGE SCALE
                    INTEGRATED CIRCUITS
                    (VLSI) TECHNOLOGY.
                    PERFORMED BY SOFTWARE
                    RATHER THAN HARDWARE
                    LOGIC WHEREVER POSSIBLE.
                                          • MODULARIZED SOFTWARE
                                          VLSI CIRCUITS

                   INPUT/OUTPUT AND
                   INTERFACE CIRCUITRY
                   FOR POST EXERCISE
                   INTERROGATION BY
                   HIGHER LEVEL
  FIGURE 6.   Design details  and  features of  construction helmet.


                                             PAIRED ALUMINUM ELECTRODE ALSO
                                             SERVES AS SYSTOLIC TIME INTERVAL
                                   ECG ELECTRODES
                    ALGORITHMS REDUCE DATA ON
                    BOARD THE SUBJECT AND STORE
                    IN NITRATE METAL OXIDE (NMOS)
                    32k MEMORIES FOR LATER RETRIEVAL
                                                    INTELLIGENT PHYSIOLOGIC

                                                       PERSONAL MONITOR
                                       1. (4) ELECTRODE (3M) ALUMINUM TAPE
                                         AND ECG PROVIDES STI
                                         PHYSIOLOGIC VARIABLES.

                                       2. MICROCOMPUTER ALGORITHMS TIME
                                         SHARE TRANSDUCERS IN A SEQUENTIAL
                                         SAMPLING THROUGHOUT ANTICIPATED
                                         FIELD EPIDEMIOLOGIC PROTOCOLS; I.e..
                                         REST, DAILY ACTIVITY, AND EXERCISE

                                       3. TRANSDUCERS SERVE IN MULTIMODE
                                         FUNCTIONS PROVIDING MORE THAN
                                         ONE PHYSIOLOGIC PARAMETER FOR
                                         SINGLE ANATOMIC PLACEMENT.

                                       4. ALL MEASUREMENTS ARE NONINVASIVE.

Typical  STI cardiac  Impedance electrode  configuration,  male sub-
wherein the "R" wave  of the electrocardiogram (ECG)  triggers an  algorithm
to perform ensemble averaging of  the  aortic impedance signal.  Figure 9
illustrates a typical output supported by multiple parameters depicting an
index  of myocardial contractibility through computation of systolic time-
interval measurement  during treadmill exercise conditions.  Figure  9  illus-
trates an ensemble-averaged aortic impedance signal  recorded under  conditions
of walking, jogging,  and deep knee bends—conditions thus highly applicable
to mobile urban subjects.  A microcomputer algorithm eliminates  the classic
artifact, resulting in a clean, clinically useful waveform.

     Figure 10 illustrates a trend plot of selected  variables which have po-
tential for detection of subtle changes induced by air pollutant exposure.

     Figures 8 through 10 illustrate  the measurement of STI's on a  male sub-
ject.   However,  the efficacy of a scientific statistical data base  must

                                     TABLE 1

                        Impedance Cardiographic Parameters
 PARAMETER                          DEFINITION
 LVET                               Left ventricular ejection time
 PEP                                Pre-ejection period
 Ratio                              PEP/LVET, myocardial contractility
 Heart rate                         Heart rate trend
 Cardiac impedance                  1st derivative aortic flow velocity
 EMS                                Electromechanical systolic
 R-R mask                           Expected R-R interval trend
 R-R mean                           Actual data used
 Av rate                            R-R rate
 Av ctr                             No. cardiac cycles used
 Efficiency                         Ratio—average over total cardiac cycles
 Q-wave                             Component ECG waveform
 T-slope                            Time to peak Z slope
 TZ max                             Time to peak Z
 DZA                                Time to diastolic peak
 TZ rise                            Time to PEP to DZ max
 TZ fall                            Time from max to DZX
 include the measurement of identical parameters in the female population.
 The transducer-to-anatomy interface on female subjects poses a unique set
 of design problems, as indicated in the user design criteria.  Further prob-
 lems are reducing user resistance and enhanced wearabllity.  Figures 11 and
 12 illustrate an STI electrode configuration for a female subject wherein
 comfort and natural anatomy-to-transducers interface is accomplished with
 a brassiere and a choker necklace.  To the left in Figure 12 is a male shirt
 collar with STI electrodes built in.  Figures 13 and 14 illustrate the in-
 corporation of IPPM devices within shoewear for both male and female popu-
 lations, as a means of Including these systems in a natural way without sacri-
 fice of scientific data.  (Note that both male and female measurements must
 be identical in support of valid statistics.)  In both users, comfort and
 interface play a major role in valid physiologic data acquisition.  In both
 the shirt collar and choker necklace, an infrared bolometer transducer is
 incorporated to monitor thermal blood flow rate; i.e., multimode transducers
 or an illustration of transphysioware.

      Finally, in the design of IPPM systems, every possible avenue of incor-
 porating systems into clothing, headwear, shoes, belts, and jewelry must be
 considered in light of the need of portable, humanly compatible IPPM systems

               FIGURE  8
                                                  I HEART RATE
                      Kt/L-UK. EXERCISE DATA TO DISK
                        MVANCE TO NEXT STAGE
                       lEWIIHATE TREAOHILL EXERCISED
I   38  M 98  121  151 IN
               FIGURE  9
                                             TYPE CONTROL/I; TO COHTIHUEI
                                  188  158  288   258  388  358  488  458 588
                          1   775   772    77    64    66    %   1*
                        (HMWE   PEP  TSLOPE  TZHAX  EHS   DZA  TZRISE  TZFALL LICT
                         28    98   117   135   368   4
               FIGURE  10
                       SELECT STI TREND DISPLAY
                       STI <»>, STIXEHS <1>, EXIT 
                                                     ELECTRODE WIRES
             STI ELECTRODES
              CONNECTOR TO
      ECG "P" WAVE
FIGURE  11.   Cardiac Impedance  transducer-to-anatomy interface.   Statistical
data  base requires identical measurements on male and female  subjects.
         HEART RATE
                             SYSTOLIC TIME INTERVAL
                             IMPEDANCE CARDIOGRAM
                             SYSTOLIC TIME INTERVAL
                             IMPEDANCE CARDIOGRAM
             HEART RATE
           SHIRT COLLAR
                             FOR PERSONAL MONITORS
FIGURE 12.   Cardiac impedance  transducer incorporated into normal  dress wear;
i.e., reduction of user resistance.

                             TO TRANSDUCERS AND
                             OTHER SYSTEMS
              FIBER OPTIC DATA
              STOCKING SEAMS
                INTERFACE DESIGN
                                              SECLUDED ACCESS PORT
                VLSI CIRCUITRY
                                          COMPLETE IPPM (PDAS)

FIGURE  13.  IPPM devices must be  incorporated within natural human attire.
                  INTERFACE DESIGN
                                                   EACH 4K MEMORIES.
                                                  COMPLETE VLSI IPPM (PDAS)
                                               SINGLE CHIP

                                       MALE WORK BOOT
                                        1. HEAVY BOOT. HAS IPPM MODULE AS
                                           OUTBOARD SYSTEM.
                                        2. IPPM IN LEFT BOOT, POWER PACK IN RIGHT
                                           BOOT. USES FIBER OPTIC DATA AND POWER
                                           LINKS BETWEEN BOOTS THROUGH CLOTHING
                                        3. IPPM MODULE RETURNED BY MAIL FOR
                                           MAINTENANCE. FIELD REPLACEMENTS AT
                                           MODULAR LEVa.
FIGURE  14.   IPPM  system for  construction worker.

      With software modification of existing STI systems and retaining the
 common STI electrode configuration, it is possible to record respiratory
 rate.  Additionally, adding an infrared transducer or liquid crystal to the
 neck electrode may provide an index of carotid blood flow through I-R thermal
 radiation of blood which has a spectral emissivity at 9.3 microns.  Hence,
 infra-red blood flow index, ECG, STI, and respiratory rate measurements are
 all feasible from the common placement of the STI electrodes.  A further
 enhancement of respiratory volume measurements is possible with impedance
 pneumography, if calibrated against the pneumotachograph as illustrated in
 the construction worker's tidal volume system.  Combining STI and tidal vol-
 ume measurements could be accomplished on a single individual through dedi-
 cated IPT systems operating in a multimicroprocessor mode.  The data array
 made possible by this combination is listed below.  This is a more specific
 example of multlmode transphysloware.

      Cardiopulmonary Parameters
      Respiratory rate
      Respiratory volume
      Respiratory flow
      Respiratory tidal volume
      Heart rate
      I-R blood flow rate
      All STI and impedance measurements.

      It should be noted again that the STI Intel 8010 microcomputer is serving
 as an intelligence converter reducing the abundance of cardiac data into a
 meaningful clinical measurement.  Additionally, multimode parameters are dis-
 played in the form of trend data (Figure 10) as an index of subtle response.
 The stand-alone capabilities of the microcomputer (4,11,15,17) permit direct
 support to a clinical data base without extensive statistical support of
 large ADP systems such as the UNIVAC.  This example illustrates the cost
 effectiveness of "intelligent microcomputerized blomedical Instruments"
 (IMBI).  With 64K single chip microcomputers being introduced in 1979 (15),
 the complete STI system software and ensemble averaging illustrated above
 could be packaged (less graphics display) within the subject's belt, shoes,
 or hat.  With the abundance of high-density memory, nitrate metal oxide field
 effects (NMOST), long-term memory storage (1 to 10 years) is also feasible.
 NMOST memories could be interrogated in post-data collection analysis to
 review graphics and trends for the generation of an urban data base.  Ex-
 tension of this design philosophy is the basis of IPPM, IPT, multimode trans-

ducers, and multlmicrocomputer  systems.  Furthermore, this design  philosophy
is expandable  to include  future designs adaptable to changing protocols
(through changing software, not hardware) within internal and external IPPM
devices presently beyond  the state-of-the-art as of this writing.  Addition-
ally, the design concepts presented here could well be the forerunners of
future internal and external IPPM and IPT 3 to 5 years hence.

     Any new approach such as IPPM noninvasive physiologic measurement always
contains a question as to the efficacy of the parameter under analysis.  The
clinical validation of IPPM systems advocated here are based on the use of
the EPA Clinical Environmental Laboratory (CEL) as an "IPPM test center" for
the evaluation and validation of all personal monitors.  The efficacy of all
IPPM and dosimeter designs would be dynamically exercised under physiologic
and pollutant performance conditions prior to use in any epidemiologic urban
health effects program (13).  As such, the CEL provides a unique controlled
clinical setting sufficient for semi-invasive and external physiologic data
acquisition system investigations.

     Once preliminary IPPM systems become available, pilot studies in both
air pollutant and physiologic data acquisition may be conducted and evaluated
against known clinical and air chemistry standards.  Use of the CEL as a
''national standards resource" for design, development, and evaluation of
IPPM would provide a common data base for the design of such systems.  Further,
the CEL facilities would provide a basis for validating clinical and pollu-
tant dosimetry protocols advocated in IPPM systems prior to field epidemio-
logic studies.  Hence, the urban health effects data base would be acquired
on a realistic basis.  This validation approach is the only method that will
ensure a scientifically defensible data base in support of legislation or any
public health initiative.

     It is advocated here that all IPPM systems would be designed and de-
veloped in this context, wherein the clinical and environmental standards
of the CEL would serve as the EPA national standard resource for those inter-
ested in certification of such systems and devices.

     The advocacy of new personal monitor designs has been presented through


implementation  of microcomputer technology and  interactive  protocols.  In-
corporating  clinical  rationale and Judgment within software algorithms at
the clinical level  has  been demonstrated through design examples.  Extend-
ing this  technique  to the personal monitor level in population  studies re-
mains  to  be  implemented.   However, with the acceleration of high-level VLSI
technology,  microcomputers, and submicron systems in the immediate future,
the attainment  of a clinical environmental preventive medicine  data  base
as an  urban  early-warning system is a realistic possibility.  Conversely,
if epidemiologic research retains present methodologies, the results may
continue  at  present levels of knowledge.

     A major thrust in the design and development of IPPM and IPX devices is
imperative if useful  epidemiologic health effects data is desired.   Develop-
ments  of  this type  must be designed to maximize clinical flexibility in terms
of changing  environmental needs, assessment of  new pollutant episodes, and
adaptability to new research protocols.  A significant barrier  presently
exists in terms of  adequately designed physiologic/anatomic transducers which
are compatible  with male and female subjects.   Subject comfort  and ease of
use are paramount if  physiologic data are to be collected in their natural
state. The  concepts  proposed here advocate a new design approach in physio-
logic  transducers and systems which must be Incorporated within clothing,
shoewear, headwear, underwear, jewelry, and cosmetics.  Additionally, trans-
ducers must  function  at an intelligent level through microcomputer software
as a means of reducing size, weight, and power  while improving  comfort and
clinical  utility.

     Finally, it is envisioned that any design  initiated today  must  be flex-
ible enough  to  be adaptive and useful to new protocols 2 to 3 years  later,
when priorities will  most certainly change. It is,  therefore,  the responsi-
bility of systems designers and program managers to invest  in appropriate
research  and development efforts in the implementation of IPPM  systems in
response  to  an  effective public health initiative.   Part II of  this  paper
addresses future needs through anticipatory research.

  1.   Computers in Cardiology, IEEE Computer Society #75CH1018-1C.  The
      Netherlands, October 1975.

  2.   Costle,  D.M.  Health and the Environment.   EPA Journal, July-August

 3.  Harrison, D.C. et al.  Cardiovascular Imaging and  Image Processing.
     The Society of Photo Optical Instrumentation Engineers, Vol. VXXII,

 4.  Hyman, W.A., Lively, W.M.  Workshop Report:   Microprocessor Impact  in
     Health Care.  Proceedings of the 12th Hawaii International Conference
     on Systems Sciences, pp. 272-275, January 1979.

 5.  Kizakevich, P.  Personal Cardiopulmonary  Electrode Monitoring.  Proceed-
     ings of the Symposium on the Development  and Usage of Personal Monitors
     for Exposure and Health Effects Studies,  January 22-24, 1979.  (EPA

 6.  Kizakevich, P., Gallon, F., McDermott, J., Aranda, J.  Continuous Non-
     invasive Cardiac Monitoring of Cardiac Function: Its Applications in
     Exercise Stress.  British Heart Journal,  1978 (in  press).

 7.  Kubicek, W.G.  Development and Evaluation of an Impedance Cardlographic
     System to Measure Cardiac Output and Other Cardiac Parameters.  NASA
     Contract #NAS-9-4500.  June 30, 1969.

 8.  Lion, K.S.  Instrumentation in Scientific Research.  McGraw Hill Book
     Company, Inc., 1959, pp. 74-84.

 9.  Love, G.J., Shy, C.M., Calafiore, D.C., Benson, F.B., Finklea, J.F.
     The Strategy for Determining the Effects  of  Environmental Pollution on
     Human Health.  Environ Letters 3(1):13-20, 1972.

10.  Medical Electronics Fifth International Conference, Liege, Belgium, 1964.

11.  Petrovlck, M.L.  Patient Owned Systems and Devices In Health Care.
     Presented at the Texas A&M Conference on  the "Impact of Microprocessors
     in Health Care," September 18-19, 1978.   (In preparation.)

12.  Petrovick, M.L., Malindzak, G.S., Jr., Strong, A.T.  The Role of Personal
     Monitors in Environmental Health Effects  Research.  Proceedings of the
     IEEE 1974 National Aerospace and Electronics Conference, NAECON 74,
     pp. 152-159.

13.  Shy, C.M., Finklea, J.F.  Air Pollutant Effects on Community Health.
     Environ Sci Technol 7:204-208,  1973.

14.  Soucek, B.  Microprocessors and Microcomputers.  New York, Wiley -
     Interscience Publication, John Wiley  and  Sons, 1976.

15.  Strader, N.R.  The Microprocessor Transducer Function.  Proceedings
     of the 12th Hawaii International Conference  on Systems Science, pp.
     256-265, January 1979.

16.   Steigbigel, J.  et  al.  Quantitative Evaluation of First Derivative
      Impedance Cardiogram.  Circulation 42(11):67, 1975.

17.   User's Guide, M6800  Exerciser.  Motorola, Inc., 1975.

18.   Wallace, L.  Personal Air Quality Monitors:  Past Uses and Present
      Prospects.  American Chemical Society 4th Joint Conference on Sensing
      of  Environmental Pollutants, pp. 390-394, 1978.

19.   Welham, J.C. et al.  First  Derivative of  Transthoracic Electrical Im-
      pedance as  an Index  of Changes  in Myocardial Contractility in the Anes-
      thetized Dog.   Intensive Care Med 4:43-50, 1978.

 Mathew L.  Petrovick
 Health Effects Research Laboratory
 Clinical Studies Division
 Physiology Branch
 U.S.  Environmental Protection Agency
 Chapel Hill, North Carolina  27514
      CHAPMAN:  Robert Chapman,  EPA.   Matt,  I got a little confused.  In the
 middle of your talk, you showed a system on a male in which he had three
 EGG leads on, and he had a couple of straps around his chest, and he also
 had some straps around his neck.  And then  two  slides later you showed a
 picture of a shirt collar with some  things  In it.  Would this now be part
 of the same system, or are we sort of jumping ahead to an even smaller sys-

      PETROVICK:  No.  This system requires  two  sets of electrodes—a pair
 on the neck and a pair at the thoracic cage.  The slide with the shirt collar
 only indicated how that electrode could be  packaged so it would be functional
 and aesthetically pleasing.  The other electrodes would be contained under
 the shirt.  So, there are four electrodes,  and  you seclude them in clothing
 and such.

Design of Personal Monitors for External
and Internal  Physiologic Studies in  Health
Effects (Part II)
     Mathew L. Petrovick and Edward D. Haak, Jr., M.D.
     Clinical Studies Division
     Health Effects Research Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     Scientific assessment of urban health decrements resulting from air pol-
lutant exposures has eluded clinical Investigators  to the extent that a de-
finitive urban health effects data base remains  a high priority in support
of EPA legislation.   More Importantly, a definitive urban clinical data base
would provide a means for establishing early-warning guidelines in preventive
environmental medicine as a public health initiative.  However, the need
for an early-warning data base has not been uniformly recognized in terms
of epldemiologic research, due to the obvious lack  of properly designed
transducers, clinical instrumentation, and logistic resources.

     The key significance of an urban early-warning data base is predicated
on the ability to characterize dose-response relationships at a highly in-
dividualized level through personal monitoring.  Systems which are capable
of accurately documenting both physiologic and air  pollutant variables simul-
taneously are essentially nonexistent.  It is, therefore, incumbent upon
clinical and environmental scientists to systematically assess the urban
epidemiologlc needs  toward the design, development, and Implementation of
personal monitors capable of simultaneous physiologic and air pollutant data

     Such a system,  Healthe-Belt, was advocated  by  Petrovick et al. in 1974
(7).  It was the opinion of several peer groups  (in 1974) that technology
could not support a  personalized pollutant and physiologic data acquisition

 system design.   It is still common to hear various researchers suggest the
 improbability of detecting dose-response variance in normal mobile adults
 resulting from pollutant exposures.  Unfortunately, such assessments of per-
 sonal monitors are speculative and not based on credible scientific investi-
 gation.  Conversely, the latter rationale is reasonable and consistent with
 antiquated clinical technology.  Today, new technology (10,12,13) further
 confirms the Healthe-Belt concept in that through appropriate research and
 development, such a goal is entirely feasible.  Specifically, the explosive
 advances (10) in microcircuitry, submicron components, very large scale in-
 tegration (VLSI), microprocessors, and microcomputers can be implemented
 and configured for simultaneous pollutant doslmetry and physiologic data
 acquisition systems.  In view of the accelerated pace of VLSI technology and
 the need for constantly changing health effects protocols, it is imperative
 to design personal monitors (PM) with maximum flexibility toward adaptation
 to changing clinical needs.

      Part I of this two-part report sets forth design guidelines and criteria
 for personal monitors using today's technology.  This report is Intended to
 set forth an entirely new philosophy 3 to 5 years beyond that advocated in
 Part I, as a means of personal monitor design that is adaptable to and anti-
 cipates changing clinical, epidemiologic needs.  Accordingly, a research
 rationale must be developed such that any R&D of PM attempted must anticipate
 clinical needs and advances of modern technology in order to effectively respond
 to the urban epidemiologic needs.

      The following fundamentals are therefore proffered, somewhat specifically
 in the area of clinical physiology and high-level technology as supportive
 resources to a public health initiative capable of characterizing urban
 health effects resulting from air pollutants.  As such, these fundamentals
 may be used as basic research guidelines in the design and development of
 internal and external intelligent personal physiologic monitors (IPPM) in
 search of new knowledge.

      It is imperative to:
      1)  Increase new knowledge and understanding of external and Internal
 physiology in order to avoid present-day limitations in parameters, transducers,
 techniques, and technology.

      2)  Conduct exploratory studies In the search of new physiologic para-
 meters specific to assessment of urban health effects.


     3)  Conduct exploratory  research  in wearable external and  Internal
IPPM devices.

     4)  Conduct exploratory  studies in the design and development of new
transducers, systems, and devices  to permit assessment of internal and ex-
ternal physiology as a basis  for increasing.new knowledge from which to
draw upon in the development  of new parameters.

     5)  Apply the knowledge  of existing and future technology—such as very
large scale integration, microcircuitry, microcomputer techniques, software
engineering, intelligent transducers,  and interactive clinical Judgment—as
a means of developing new methodologies specific to urban epidemlologlc
assessment of health effects  research.

     6)  Support an in-depth, comprehensive budget structure that is respon-
sive to cost effectiveness and long-term research support for the design
and development of new tools  for this  endeavor.

     Implementation of the above rationale has considerable potential for the
evolution of several new physiologic parameters, transducers, and techniques
as basic research spinoffs.  This  potential can be realized in terms of
productive research that is "made  to happen" by departure from traditional
technology and clinical techniques.

     The test subject transducer criteria (Part I) impose a significant
problem in the physical packaging of IPPM systems design.  Use of multimode
transducers will require the basic analog sensors, electrodes, etc., followed
by a level of intelligent signal processing, artifact screening, rejection,
and simplification of data in response to call from higher-order multiprocessor
systems.  Incorporation of these features can only be accomplished through a
revolutionary change in technology, such as implementation of micro- and sub-
micron-component technology in IPPM design.

     In the search for new physiologic parameters and techniques to supplement
conventional physiologic data acquisition systems (PDAS) measurements, a new
design Initiative is Imperative to overcome the extreme limitations of con-
ventional noninvaslve clinical research.  The development of new techniques
must permit feasibility studies for design and development of new systems.

 Feasibility prototypes must be investigated, constructed,  and clinically vali-
 dated  before use in epidemiologic studies.

     Noninvasive physiologic parameters have predominantly been associated
 with some form of interfacing transducers to the external anatomy (8,12).
 Few, if any, options exist for internal physiologic measurement, or it seems
 that way in the sense of classic design rationale.  However,  properly designed
 systems at the Internal level may serve as prototypes for future external
 physiologic data acquisition systems.  The latter is the basis of internal
 exploratory research via IPPM PDAS designs described in this  paper.

     Physical size and available real estate (on-board the subject) for
 transducers cannot in all cases be based on discrete component (TTL) logic,
 in view of bulk, power, and physical encumbrances to the subject.  However,
 it is  feasible to perform many discrete circuit functions, such as signal
 processing, with software rather than hardware logic; I.e., a first-derivative
 amplifier function can be accomplished with an algorithm corresponding to a
 low-pass (hardware) filter function.  Logic process control functions can well
 be performed with software logic, thus eliminating the need for TTL circuit
 logic  devices without loss of basic function.  Essentially, transducers de-
 signed to incorporate software as a process function become uniquely compatible
 in microcomputer-controlled personal monitors.  It is immediately recognized
 that software rather than hardware logic will require considerable memory
 in each IPPM.  However, with VLSI technology today, 64K microcomputer memory
 modules (10) in a 40 pin dip package can easily accommodate transducer functions,
 as well as most interactive protocols and emulated circuit logic.  Herein is
 the key significance of microcomputer technology.  Transducers incorporating
 microprocessors integral to their function may serve as IPPM  system peripherals
 in a multiprocessor mode (8,10).  Each transducer could communicate through
 an I/O port as a direct memory-addressable DMA peripheral. This approach
 would  free the bus master to operate on higher-level priority functions, while
 system interrupts from peripherals would be initiated only when relevant sub-
 tle changes occur.  Incorporation of microcomputers, microprocessors, and
 peripheral devices into transducers and IPPM systems cannot be accomplished
 effectively unless a quantum jump (8,9,10) in component size  and high-density
 packaging is attained within industry.  Furthermore, test subject space is
 needed for air dosimetry sensors and devices on the same test subject to
 ensure simultaneous PDAS and air pollutant assessment.  The VLSI techniques
 definitely are a move in the proper direction.  However, the  design of IPPM
 external and internal systems advocates a combination of VLSI and submicron
 component technology based on standardized modularized subsystems.

     To  illustrate  the  design principles  and  problems  advocated  in IPPM
systems,  several  design examples  are  proffered as initial design concepts.

     Design of IPPM systems  requires  a high degree of  anticipation of  clinical
need (8)  prior to the principal investigator's own recognition of a given
need.  As such, maximum software  flexibility  to changing protocols with mini-
mal hardware  changes is a  significant requirement in the design  of IPPM de-
vices.   The design  philosophy expounded in this paper  is based on today's
needs with tomorrow's technology.  This approach ensures design  flexibility,
cost effectiveness,  and elimination of technological obsolescence at the  com-
pletion  of a  given  design.

     It  is Immediately  recognized that several hardware and software devices
advocated here are  not  currently available.  Therefore, these designs  are
considered to be  design goals and examples that should be attained in  response
to urban epidemiologic  needs.  It is  further recognized that availability
of such  devices are  possible,  and that these designs can be made  to happen
if a realistic public health initiative is advocated and supported.  Designs
initiated at  the  present time will require a lead time of 18 months to 3
years, based  on the  acceleration of VLSI technology today.  Hence,  it  behooves
program managers  and designers to plan urban epidemiologic health programs
with these factors well in mind.  The need for micro- and submicron technology
(8,10) in intelligent physiologic monitors is of significance in  that  urban
health effects personal monitoring cannot be successfully accomplished without
this resource.

VLSI and Submicron Technology

     With the packing density of VLSI, and with submicron technology increas-
ing and the physical size decreasing, standardized subsystems such as 8-16
bit microprocessors, I/O devices, static RAM, ROM, tri-state buffers 8, 16,
32, 64K memories, and system controllers could be manufactured as precanned,
modularized system wafers.  Additionally, selection of individualized micro-
computers at the 4 to 16 bit level could be designed quickly and simply if
precanned hardware and software were available in modular wafer form.  (See
Figure 1.)  Modularized subsystems would be used as if they were discrete
(TTL) components, and IPPM designing would be accomplished at systems level
by utilizing prevalidated systems and software protocols.  This design capa-
bility would provide significant clinical flexibility in adapting any system
modules to individualized IPPM systems.


                                                                 1.  PHVSICIANS PRESCRIPTION DEFINES THE PROTOCOL
                                                                    AND PARAMETERS OF MEASUREMENT!
                                                                 2.  SOFTWARE ENGINEERING ANALYZE* REQUIREMENTS,
                                                                    DRAM ON STANDARDIZED ALGORITHMS FOR BIO-
                                                                    MEDICAL MEASUREMENTS.
                                                                 3.  SYSTEMS ENGINEERING FILLS PRESCRIPTION THROUGH
                                                                    ASSEMBLY TO MEET PHYSICIANS NEEDS.
                                                                 4.  0-A BIOLOGICAL SAFETY CHECKS PERFORMED.
                                                                 S.  DELIVER TO CLINIC.
                                                                 6.  PHYSICIAN ADMINISTERS TO PATIENT.
                                                                 7.  UPON RETRIVAL, DATA INTERROGATION PROCESS
                                                                    FEEDS RESULTS INTO DATA RESPONSE BANK.
                                                                 I.  PHYSICIAN REVIEWS RESULTS, ADMINISTERS
                                                                    FURTHER HEALTH CARE MANAGEMENT OF
                                                                    PATIENT. RELEASE FROM HOSPITAL
  FIGURE  1.    Modular wafer  subsystem components  used as  discrete  systems  rather
  than  the  classic TTL logic  approach.
                                                   ELEMENT ASSEMBLY
                                                ASSEMBLED CATERPILLAR
                                                PROCESSOR PACKAGING
                 END VIEW
                LSI 8K RAM
                             I/O BUS
                                                                   FIBEROPTIC DATA BUS

                                         POWER *.1 VOC

                                          MAGNETIC INTERLOCKING
                                          MNOS-UKE IK STATIC
                                          RAM WAFER
                                          LSI CHIP

                                          INTERFACE CONNECTORS
                                                                           GROUND BUS

  FIGURE  2.    System  wafer  modules high density  packaging  for  internal/external
  physiologic exploration.

     Ideally, microprocessors  used with IPPM  should  take  the  form  of stand-
ardized devices (8,9,10) with  specific capabilities  in  a  physical  configu-
ration, thus ensuring maximum  flexibility  to  individually designed systems.
Additionally, the mechanical and  electronic packaging should  be  in a form
which enhances all  power,  I/O  data bus, control bus, and  Interface connections
(Figure 2).  The physical  shape of modules used in any  system should be  easily
adaptable  to individual subject configurations, thus enhancing ease of as-
sembly by  semitechnical personnel.

     Figure 2 illustrates  an example of two adaptive wafer designs that  snap
together in serial  fashion.  This packaging scheme would  permit  various  shapes,
curvatures, or caterpillar-like segmented designs as an example.   Ideally,
several wafer modules could be assembled to suit a particular study protocol
for urban  health effects screening.  Both pollutant dosimeters and PDAS  systems
could be designed by this  process so as to ensure miniaturized systems com-
patible with personal mobility.

     Finally, the physical hardware should be designed  at a modular level
such as a  wafer configuration.  Each module would serve as a  subsystem such
as MCPU, I/O, RAM,  ROM, and interface devices.  This would permit  the physi-
cian to prescribe a system design at a block diagram level, thus avoiding
unnecessary design  detail.  The physician would select  his/her design based
on a diagnostic data base matrix (8) (see Figure 3), and  the  protocol he/she
feels is most beneficial in a  specific urban epidemiological  study.  Design-
ing an IPPM with pretested, modularized systems would eliminate  the  use  of
time-consuming discrete components, connections, and hardware debugging, while
ensuring standardization of selected protocols.

     Clinically, the physician's physical examination and diagnosis  of the
subject is at a physiologic-systems level, including, for example,  judgments
on the EEC, blood pressure, blood chemistries, etc.  Why not, then,  tailor
the IPPM to follow  a similar analytic rationale in collecting physiologic
data and provide for a total systems approach to health effects  research.
Hence, upon completion of preliminary studies, the physician  could  write a
prescription to design an external IPPM to suit a specific protocol  for  a
selected population group and  related pollutant (which  is a form of  pre-
scription  engineering).  The protocol would ensure exact measurements to be
made by previously  proven algorithms which, in the final analysis,  provides
the physician with  data useful in the final data base.  Semitechnical person-
nel would  construct the IPPM in accordance with the prescription,  and they
would return it to  the physician to administer to the population group.






i— —

                     ADAPTIVE DESIGN
                   INTELLIGENT PERSONAL
                   PHYSIOLOGIC MONITOR
                                                                   > MANUFACTURER'S STANDAROIZED IVISI)
                                                                    CW ARMY BUS STRUCTURE AND
                                                                    ASSEMBLY CODE
                                                               STANDAROIZED MICROPROCESSOR
                                                               PREPROCESSOR AND ALGORITHM
 FIGURE  3.  This matrix  is used to  rationalize  an IPPM design configuration

 in  a short period  of time by  utilizing proven  protocols and  systems.  This

 approach is  much more responsive in assessment  of pollutant  episodes•

                                  Traimrertiia thoracis
                        fntfrnal thnratie ttmth \
                 Left jihrenir
           Pulmonary pleura

              Coatal pleura
                           Sum pathetic trunk

                                 Thoracic duet
     Atugot tein

Vaout nervet
 FIGURE  4.  Transverse  section of  the thorax showing  anatomic relationship of
 the esophagus to  heart  and lungs.

     In the search and exploration  for new physiologic parameters  and  trans-
ducers, semi-invasive (internal) studies hold potential for  the development
of new measurements and techniques  for use in IPPM external  systems  through
new knowledge and techniques.  However, suitable transducers and subsystems
do not presently exist on  the shelf to permit such investigations, and,
therefore, must be specifically designed.  The design of IPPM systems  thus
advocates individualized systems which permit the charting and charac-
terization of limited internal physiologic function.  Hence, a whole new  phi-
losophy in the acquisition of external and internal physiologic data must
precede the design of IPPM's.

     Clinical gastrointestinology (1) has in the last two decades developed
a wide variety of techniques for swallowable instrumentation devices in the
form of EGG electrodes, temperature sensors, pressure transducers, telemetry
systems, and balloon-mounted devices, each representing discrete components.
To this date, there has been little effort toward the design of coordinated
internal multiparameter systems.  Gastroenterology techniques, however, have
previously demonstrated that a number of physiologic parameters can be meas-
ured internally from the esophagus through and to the GI tract.  Figure 4
Illustrates an anatomic transverse section of the thorax relative to the
heart, lungs, and the esophagus or survey site for semi-invasive IPPM for use
in clinical exploration.  Table 1 illustrates an example of parameters detec-
table internally by the designs proposed herein.

     The semi-invasive parameters available for clinical study and evaluation
in the design of IPPM systems provide a significant potential for a cardio-
pulmonary data base.  Initially, the study of these parameters must be per-
formed under careful clinical management, such as in the "Clinical Environ-
mental Laboratory" or CEL (Strong 1978) (11).  The major impasse, however,
is the lack of properly designed IPPM systems for internal data acquisition.

     Utilizing the principles of the PDBM and VLSI technology, a brief illus-
tration of internal IPPM systems design is presented as follows.

     Utilizing the IPPM wafer systems modules developed from the PDBM, a
500-mg envelope (Figure 5) is instrumented with two inflatable elastomer

                                      TABLE 1

         Physiologic  Parameters Detectable from Internal Esophageal PDAS

                           Respiratory rate
                           Respiratory volume
                           Respiratory tidal volume
                           Cough rate and intensity
                           Heart rate stress—medical-legal safety
                           EGG "P" wave arrhythmia detection
                           Heart sounds—cardiac valves
                           New parameters
  docking rings,  as a means of fixing and stabilizing the IPPM within the esoph-
  agus at the level of the cardiac orifice (Figure 6).  Each docking ring
  contains a pair of transducers—one pair for pulmonary and one pair for car-
  diac measurements.  Measurement of respiratory volume, rate, and tidal volume
  are all feasible from the piezoelectric (6) transducers ( Z ) located within
  the inflatable  docking rings located in the esophagus.  Clinically, a flex-
  ible tube (track) similar to an esophageal balloon leader is inserted through
  the nasal-pharyngeal cavity until it is visible orally at the posterior pharyn-
  geal wall of the mouth.  The IPPM is contained within a 500-milligram envelope
  and is attached through the mouth to the flexible track and then passed down
  the track or to the monitoring site of interest.  Once located anatomically,
  the physician inflates the elastomer docking rings with the syringe, thus
  placing the pulmonary and EGG transducers against the inner walls of the
  esophagus at the level of the cardiac orifice or posterior of the heart.
  Functionally, the piezoelectric transducers ( Z ) are used in a multlmode
  transphysioware, wherein one transducer performs the initial anatomic in-
  terface, and a  software algorithm performs multiparameter acquisition of
  cardiac parameters.

       The piezoelectric (6) transducer lends itself to a wide variety of phys-
  iologic measurements, such as low- and high—frequency pressure, and acoustic
  sensing.  It is well known that shunting a specifically cut piezoelectric
  sensor with external resistance or capacitance can dramatically change the

                               INFLATABLE ELASTOMER
                               ESOPHAOEAL DOCKING
                CLEAR AIR SPACE MINI-
                MIZES AIRWAY RESIS-
                TANCE WITHIN ESOPHA-
                                                                     m« ENVELOPE
                                                              PHONOCARDIOORAPHIC SENSOR (CVS)
                                  INFLATABLE ELASTOMERS ARE MOLDED TO
                                  PREDETERMINED SHAPE AND TAKE ON A
                                  SET CONFIGURATION UPON INFLATION
                         • ECO ELEC-
                    SENSOR WIRES FROM OUTER

                MANUAL SYRINGE
                INFLATE UPON INITIAL

                                1. ANATOMIC LOCATION AT CARDIAC ORIFICE TO
                                 DETECT ECG AND HEART SOUNDS, !.«.. ~P" WAVE
                                 DETECTION ARRYTHMIA PROCESSING. MVO-
                                 CARDIAL INFARCT.
                                2. UPON COLLECTION OF CARDIAC DATA. THE SYS-
                                 TEM IS REMOTELY LAUNCHED BY THE INVESTI-
                                 GATOR BY DEFLATING THE CAPSULE AND THE
                                 BREAKAWAY PNEUMATIC TUBE.
                                3. UPON SYSTEM LAUNCH. THE CAPSULE TRAVERSES
                                 ITS ROUTE THROUGH THE G.I. TRACT COLLECT-
                                 ING VARIOUS PHYSIOLOGIC DATA.

                                4. VALIDATION STUDIES TO BE CONDUCTED
                                 IN LABORATORY ANIMALS.
FIGURE  5.  This  device is  supportive  to basic physiology  in the search  for
new parameters and transducers.
characteristics  of the  Zk  element.   Changes in "Q,"  the  equivalent circuit
                              t                           p
merit,  and  the frequency response can cause the  Z   sensor to  readily  respond
to  and  detect DC to high-frequency information.

      As a considered design approach,  one  could attain multimode  functions
through remote tuning  of a  piezoelectric transducer  with  a software algorithm.
In  this design,  the objective is to  remotely change  the   Zfc characteristics
within  the  swallowable microcomputer,  upon software  command,  to sense  the
following parameters:
      •   Gastric  motllity
      •   Gastric  pH
      •   Cough rate and intensity
      •   Respiratory rate
      •   Heart sounds
      •   ECG  "P"  wave-arrhythmias
      •   Heart rate.

                                     INTERNAL EXPLORATORY
                             PHYSIOLOGIC DATA ACQUISITION SYSTEM
                 VOCAL FOLDS
                                                                      INFLATE SYRINGE
                                                           DEFLATE PINCH-OFF
                                                           LAUNCH CONTROL
                 OWNED SYSTEM PDAS
                                             CAPSULE INFLATED TO HOLD
                                             POSITION WITHIN ESOPHAGUS
                                       <.   CARDIAC
                                        •s  ORIFICE
                        ENTRANCE TO -
                                             1.  SYSTEM INSTALLATION SIMILAR TO SWALLOWING
                                               ESOPHAGEAL BALLOON.
                                             2.  CAPSULE IS ATTACHED TO THE LEADER LINE TRACK
                                               AFTER PASSAGE THROUGH NASAL PHARYNGEAL WALL.
                                             3.  PHYSICAL ATTACHMENT THROUGH ORAL CAVITY AT
                                               POSTERIOR PHARYNGEAL WALL AREA.

                                             4  APPROPRIATE ANESTHETIC IS USED TO DESENSITIZE
                                               GAG REFLEX - A NORMAL PROCEDURE.
                                             &  AFTER THE SYSTEM IS INFLATED AND INSTALLED.
                                               THE LEADER LINE TRACK. SYRINGE. AND PHYSICIAN'S
                                               LAUNCH CONTROL IS REMOVED.
                                             6.  THE LAUNCH CONTROL REMOVAL PINCHES THE
                                               PNEUMATIC LINE SO AS TO PROVIDE A SLOW LEAK
                                               ALLOWING SPECIFIC TIME BEFORE LAUNCHING
                                               INTO THE STOMACH.
FIGURE 6.   Internal PDAS is an  investigative  tool in  support of external IPPM.
                              Z   is  interfaced to  a specifically addressed input
                                            A controlling  algorithm applies  a limited
      Conceptually,  the
port of the microcomputer device
voltage to  a tri-state  buffer-like and ladder  network  device.
                                                                              By  this  process,
 the  Z  is remotely tuned  through a common MOST/FET-like  voltage-controlled

 resistor  which changes  the  "Q"  of the   Z.  for the measurement  of a  specific
 physiologic parameter.   This design advocates the use  of  one   Z  sensor which

acts as a peripheral device containing a specific  input  port  address.   Access
of this input port  to  the  control bus is through a form  of  trl-state buffer.
Each of the above parameters may be detected by a  change of shunting impedance
to the  Z , thus altering  the  low/high frequency-response to  detect the above
parameters.  This process  is repeated in sequence  to perform  the above  meas-
urements from internal firmware.  Each sequence algorithm, of course, identi-
fies in memory the  sensor  mode, parameter, data, and subject  ownership  while
in use.  Figure 7 illustrates  how the same IPPM microcomputer can through
prevalidated (EPROM)-like  algorithms, remotely tune or change Z  character-
istics for use in pulmonary rate and volume measurements.

     By this process, it may be feasible to detect two or more physiologic
measurements from a single transducer (multlmode).  The  related algorithm, of
course, would preprocess these signals into meaningful clinical measurements—
i.e., transphysioware.  Another option is to retrieve data from long-term
memory (1 to 10 years), such as Is possible with the new metal-nitrate-oxide
slllcone transistor (MNOST) semiconductor memories within a given IPPM.  Data
interrogation of the IPPM  by high-level computer would permit a form of data
base development in the off-line mode through interactive graphic analysis,
as illustrated in Part I.

     In another mode, the  platinum electrodes on the 500-milligram PDAS,
(Figure 5) can detect heart rate and "P" wave arrhythmia through associated
clinical rationale in the  form of a prevalidated algorithm designed for this
purpose.  A further mode time shares the platinum electrodes  for measurement
of esophageal pH or gastric pH if the IPPM is inserted beyond the cardiac
orifice and into the root  of the stomach.  If the IPPM is located at the
cardiac orifice, respiratory rate, cough rate,  and cough intensity are also
possible to detect.  The acoustic cough signal would be converted through
software (RMS signal processing) to a DC voltage and compared with the dia-
phragmatic pressure recoil during a cough.   If both signals occurred simul-
taneously, a software logic algorithm would verify the cough  rate, intensity,
and duration (as opposed to artifact).  This would be of particular value in
exercise pulmonary ozone exposure studies,  in which frequently there is a
coughing response after 4 hours of ozone at .6  ppm.

     Taking full advantage of semi-invasive internal IPPM, it is feasible to
perform "cardiac Impedance systolic time interval" (2,3,4,5) measurements

                                     MOST-LIKE VOLTAGE
                                     CONTROllEO VCR fET




                                                             REMOTE MUITIMODE
                                                            PIEZIOELECTRIC ELEMENT
                                                 EQUIVALENT FET RESISTANCE
                                                CHANGE FREQUENCY RESPONSE
                                                     OF VCR-FET
                                              REMOTE CONTROL WORD
                                              SELECTS VCR VOLTAGE
                           REMOTE TUNING CONTROL
                           WORD FOR MUITIMODE
                            NOTE  COMMON PZT TRANSDUCER

                            t. VCR RANGES

                             «.  .1 • II Hi GASTRIC MOTIIITY

                             1.  .1-10 Hi RESPIRATORY RATE

                             C.  IN  500 Hj HEART SOUNDS
                               SWALLOWABLE MICROCOMPUTER  WITH
                                  REMOTE MULTI-MODE PHYSIOLOGIC

 FIGURE 7.   When  space  constraints  exist,  and  multiple parameters are desired,
 this  concept can be very effective.
                POLLUTANT MONITOR
                             2" x 3- x %
                       BUS MASTER
               IPPM SYSTEM
               DATA AND POWER LINES
               RUN UNDER SUBJECT'S
               ARM TO SHOE OR
               BELT PDAS/PEP SYSTEMS
                       MICROCOMPUTER CONTROLLED
                       1024 x 1024
                       LIQUID CRYSTAL
                       PICTORIAL DISPLAY
                          8-BIT DATA WORD

                                                           SUBJECT RESPONSE
                                                           CONTROL SWITCHES
                                                               NOTE: THIS ILLUSTRATION DEPICTS
                                                                   THE INTEGRATION OF IPPM
                                                                   SYSTEMS IN PART I AND PART II
 FIGURE 8.   IPPM  display provides  means  of implementing  systematic  field pro-
 tocols in  the absence  of supervisory clinical personnel*   Pretraining of

 subjects is required.

as an index of myocardial  contractility.  Impedance STI measurements  permit
an array of cardiac  performance measurements similar  to those  reported  in
Part I.  However,  if an  internal IPPM  system for  this purpose  is desired,
the system design  would  require a specific configuration as  follows:  a
conventional esophageal  balloon is instrumented to include two  foil electrodes
around its inner surface,  while two additional electrodes are  placed  around
the subject's neck at  the  carotid artery level.  The neck electrodes  are
excited with a 100-KHz carrier signal, while the esophageal  balloon electrodes
are positioned near  the  cardiac orifice or the level of the  aortic root of
the heart.

     As each cardiac cycle occurs, the electrical impedance  carrier signal
is modulated by the  aortic flow velocity (first derivitive impedance) at  the
aortic root level, and an  analog impedance signal is detected.  Each  cardiac
cycle is preprocessed by ensemble averaging and stored in memory.  If immedi-
ate STI data is sought,  signals are passed out through the nasal pharyngeal
cavity by fiber optic data lines to an external microcomputer.  This  process
would permit real  time exercise analysis of STI, as illustrated in Part I.
If it is desired to  collect and store periodic internal STI data for  later
interrogation and  analysis in an off-line fashion, this may be accomplished
by storing data in RAM within the IPPM and analyzing them later.  The resulting
cardiac Impedance  signals  would then be stored in RAM for subsequent analysis
after exploratory  surveys  have been completed.  This technique would  provide
a means of cross-validating the external IPPM/STI device prior to field
epidemlologic use.

     Upon the successful development of the modularized wafer systems (Figures
1 and 2), the packaged systems used for internal exploratory research may
also be applied to external personal monitoring.

     A critical need in field epidemlologic health effects is the use of the
physician's assistant or clinical technician to ensure that field protocol
techniques are properly adhered to.  Conversely, it is totally impractical
to provide clinical technicians with each IPPM during data collection.  How-
ever, with implementation of microminiaturized visual display screens in the
form of a wrist-mounted device (WMD), a form of visual and auditory communi-
cation with the trained subject is feasible (as described below).


     The use of  liquid  crystal  pictorial  displays  (LCPD) have  been utilized
in recent military aircraft  radar systems for  target  identification  and de-
struction as part of  the  "Smart Bomb" weapons  systems  (14).  LCPD modules
are available  as 1-inch-square  100 x 100  visual  matrix devices which,  through
character generator  techniques, appear  as a  postage stamp-sized TV set.
Each matrix contains  1,024 x 1,024 digital encoded pictorial elements, which
can be  electronically stacked to form a 2 x  3-inch visual  display (as  illus-
trated  in Figure 8).   With precanned protocols built  into  on-board EPROM
devices in an  IPPM,  such  a design could be used  as a  communicative prompting
device  (STI, Part I)  to have the subject  follow  specific protocol routines
without the assistance of a  medical technician.

     Operationally,  the LCPD would have three  basic modes  of operation, as
     •   Systems  management process controller
     •   Auditory attentional command and  response  functions
     •   Visual prompting  and interactive  instructions.

      A single microcomputer located  in the  wrist-mounted  device will  function
 as  a  systems  management  processor which will control  all  messages  to  and  from
 the subject via the LCPD.   The SMPC  will function  as  a  bus master  in  a multi-
 microprocessor environment.  One microcomputer will function  as a  physiologic
 data  acquisition system  (PDAS) located in the  subject's right shoe (heel)
 or  belt  (Part I).   A second microcomputer is a pollutant  exposure  processor
 (PEP)  located in the left  shoe (heel)  or belt  (Part I).   Each communicates
 on-call  to the bus master  in the wrist-mounted device.  The PDAS and  PEP
 systems  will  function as a multimode bus master/slave system.  The bus master
 will  periodically  "poll" the PDAS and  PEP throughout  protocol test periods,
 thus  accessing the two slave systems for relevant  subtle  response  data.

      The PDAS and  PEP systems will be  dedicated as IPPM systems and will
 contain  precanned  protocols relative to the detection,  processing,  and low-
 level  diagnosis and/or interpretation  of events.   If  no subtle effects within
 the PDAS or PEP systems  occur, these systems would report to  the bus  master
 only  on-call  in a  routine  ''polling'1  mode.  However, if  a  subtle change occurs
 in  either IPPM,  either one or both PDAS'  or PEP's  could interrupt  the bus
 master for "relevancy analysis"  by the bus  master.  In  such an event, the

bus master will weight  out  the  significance  of  the  physiologic  or  pollutant
effect, and sound  an attentional  tone  to  the subject  to  look  at the LCPD
for further instructions.

     Based on a low-level analysis of the PDAS or PEP, the bus master  will
sound an auditory  tone (three beeps) to get the subject's attention.   The
test subject then  responds, acknowledging his/her attention by depressing the
"yes" button on the wrist-mounted device.

     Upon test subject attentional response indicating "yes," the bus master
is told that the test subject's attention is available for interactive mes-
sages in that the subject is viewing the LCPD and awaiting information.

     The VPII and LCPD are designed to display precanned messages stored  in
an on-board EPROM device.  An array of interactive prompting messages are
called by the bus master and displayed to the subject, depending on the cir-
cumstances.  In the event of performing specific protocol functions based
on previous training in the EPA "CLEANS" facility (i.e., pulmonary function
exercise by a construction worker, as in Part I), the subject can be in-
structed via the LCPD messages and paced throughout a short protocol.  As an

     1)  An auditory tone beeps three times and gets the subject's attention.

     2)  The subject responds by depressing the "yes" button on the WMD.

     3)  The subject reads the message on the LCPD.

     4)  The prompting message will instruct the subject to:
         a)  Place pneumo in mouth;
         b)  Continue rest breathing until 1-minute tone;
         c)  Increase exercise until 5-minute tone, then stop.

     5)  If during the pulmonary exercise protocol the subject's heart rate
or other parameter exceeds the maximum level, the PDAS system alarms and  in-
terrupts the bus master.  The bus master automatically sends three auditory

 tone beeps to get the subject's attention.  The subject responds with the
 "yes" button, the bus master replies with the message "STOP"—"STOP," "GO TO
 REST-STATE."  After a period of 1 minute, the bus master polls the PDAS and
 PEP again.  If the subject's heart rate is in excess of prescribed limits,
 the bus master sends three beeps, and the subject responds with a "yes."
 The bus master sends the message, "LEAVE WORK AREA"~"SEE PHYSICIAN IMMEDIATELY.'

      6)  If only the PEP alarms without a physiologic response, the bus master
 sends three beeps, the subject responds with "yes," and the message display
 indicates "Pollutant Alarm" and continues to flash.  A second message is dis-
 played, "LEAVE AREA IMMEDIATELY," and also flashes.

      With the combination of protocol-prompting of a trained subject, and low-
 level diagnosis of PDAS or PEP as an "interactive intelligent personal monitor,"
 IPPM can serve as a low-level adjunct human supervisor in support of and lack
 of a medical technician of the test subject.  Other algorithms and protocols
 may also access the test subject's own known judgment through prompting mes-
 sages.  This approach permits a joint man-machine interactive rationale as an
 "intelligent personal monitor" (Part I).

      A wide variety of protocols, alarm messages, displays, etc., are feas-
 ible through this approach.  The functional features of an IPM design of this
 type are within the state-of-the-art today.  However, the submicron technol-
 ogy of microcomputers, with the size and packaging density advocated here,
 will require considerable research and development.  Conversely, devices
 such as the LCPD currently exist in military fire control systems.  The
 design issue requires access and design integration of a LCPD with an IPM

      The design examples and guidelines for external and Internal IPPM systems
 have been presented through implementation of VLSI, submicron components,
 and  microcomputer technology.  Through this process, the concept of the
 "physician's data base matrix" has been presented as a means of developing
 the  principles of "prescription engineering" in IPPM devices.  Prescription
 engineering provides a key mechanism in the selection of prevalidated pro-
 tocols and design of adaptive IPPM systems in response to changing research
 needs, clinical feasibility, cost effectiveness, and the development of an
 urban clinical data base.   The prescription engineering approach to clinical

epidemiologic  research  provides  the means of replacing discrete  circuit logic
with interactive  intelligent  software logic as a means of reducing  physical
size, weight,  and power.  This approach  enhances subject acceptance through
microminiaturization  and  elimination of  technological obsolescence  due  to
changing health effects mandates.

     Prescription engineering by the physician and his or her support team
illustrates the principles and problems  involved in the Interfacing of  human
anatomy to a physiologic  data acquisition system.  The anatomic-to-transducer
interface is the  greatest Impasse in IPPM system design, along with concomitant
development of new health effects parameters.  Further, the importance  of
common clinical protocols for male and female populations Imposes severe
but not insurmountable  design obstacles.  Resolving such design  obstacles is
illustrated by design examples in Parts I and II through Implementation of
the concept of multlmode  transducers.  Multimode transducers provide the  means
of minimal physical transducer placement and maximizing the number  of clini-
cally useful parameters.  The cardiopulmonary transducers and algorithms
proposed here  speak to  this design philosophy without encumbering the test
subject or sacrificing  clinical  data.  This approach also provides  the  means
of using standardized VLSI subsystems and microcomputers, which  in  quantities
needed in clinical epidemiologic research could be more cost effective  than
any other approach currently available.

     Finally,  it  is fully recognized by the authors that the level  of the
VLSI technology required  in the  proposed designs is not currently available.
The design concepts presented here have been utilized as building blocks  in
establishing design principles for the implementation of external/internal
IPPM systems in the near  future, 3 to 5 years from today.  Systems  of the
type advocated here will  not be  in the making in the foreseeable future,
unless clinical and environmental scientists are enlightened by  the potential
of new microcomputer  technology and apply modern methods to the  solution  of
health care and environmental epidemiologic problems.  Conversely, we must
recognize that microcomputers are not "cure all" devices and that their
implementation must be  based on a balanced engineering approach.  If we fail
to adopt the design principles similar to those defined in Parts I and II,
the lack of a  realistic,  simultaneous air dosimetry and physiologic data
acquisition system will continue to stagnate environmental science.  A new
strategy for the  design of personal physiologic monitors is required as a
means of implementing a public health initiative.  The philosophy of internal
semi-invasive  data acquisition systems is introduced as a means  of  conducting
exploratory physiologic research in support of this strategy.  Through  this
approach, it is anticipated that a new prospective will spawn new knowledge,

 parameters,  and techniques that may be extrapolated to external intelligent
 personal physiologic monitors.

      It is abundantly clear that through technology of today and in the next
 5  years, simultaneous pollutant and physiologic personal monitors shall be
 implemented  through the design philosophies and principles presented in Parts
 1  and II.   The issue is not how, but when, will clinical and environmental
 researchers put forth the necessary research and development effort to bring
 IPPM systems into being.  Again, it is fully recognized that several concepts
 and devices advocated here do not presently exist; they must be made to
 happen, as was done in exploration of the moon.  For this reason, appropriate
 design goals presented here must be established.  The internal/external ap-
 proach to IPPM systems represents the ultimate design.  Critics will scoff and
 allude to such innovations as being products of technological intoxication.
 However, we need only look back at the progress of personal monitor design
 in the last 5 years to realize that personal pollutant and physiologic monitors
 were advocated in 1974 Healthe-Belt, but still do not exist today;  hence, a
 loss of 5 years that could have better served the public.

      A significant problem in epidemiologic urban health effects data ac-
 quisition is lack of supervisory clinical technicians to administer the pro-
 tocol as in the EPA "CLEANS" facility.  The liquid crystal pictorial display
 device is not a complete solution to the problem.  However, such a system
 could serve as an interim solution to the field logistics of urban epidemiology.

      These proposed concepts and design philosophies will in all probability
 meet with similar pessimism, as in the past, without the benefit of scientific
 research.   In considering the explosive technology of today and the next 5
 years, the design cycle time for an external IPPM is 18 to 24 months•  This
 period of time must be followed by appropriate clinical validation studies
 for an additional year.  Once such a system is clinically accepted for field
 epidemiologic use, a minimum of a 2-year field study would be required.
 Hence, the design-to-data-base cycle time would not be unusual at 5 years.
 The key question is:  can an urban public health initiative afford to wait
 another 5  years (a total of 10 years) before appropriate R&D support is im-
 plemented  to provide a supportive legislative health effects data base?

  1.   Daniel,  E.E.,  Chapman,  K.M.  Electrical Activity of the Gastrointestinal
      Tract as an Indication of Mechanical Activity.  The Am J of Digestive
      Dis  8(1):54-102,  1963.


 2.  Kizakevich, P., McDermott, J., Gollan, F.  An Automated System for
     Systolic Time Interval Analysis.  Proceedings of Digital Equipment
     Computer Users Society, pp. 795-798, May 1976.

 3.  Kizakevich, P., Gollan, P., McDermott, J., Aranda,  J.   Continuous Non-
     invasive Cardiac Monitoring of Cardiac Function:  Its  Application in
     Exercise Stress Testing.  British Heart J, December 1978 (in press).

 4.  Kizakevich, P.  Personal Cardiopulmonary Electrode  Monitoring.   Pro-
     ceedings of the Symposium on the Development and Usage of Personal
     Monitors for Exposure and Health Effects Studies, Chapel Hill,
     N.C., January 22-24, 1979.  (EPA symposium.)

 5.  Kubicek, W.G. et al.  Development and Evaluation of Impedance Cardio-
     graphlc Systems to Measure Cardiac Output and Other Cardiac  Parameters.
     NASA Contract #NAS-9-45.  July 1967.

 6.  Mason, W.P., Thurston, R.N.  Use of Piezoresistive  Materials in the
     Measurement of Displacement, Force, and Torque.   J  Acoustical Soc of
     America 29, 1957.

 7.  Petrovick, M.L. et al.  The Role of Personal Monitors  in Environmental
     Health Effects Research.  Proceedings of the IEEE National Aerospace and
     Electronics Conference, NAECON 74, pp. 152-159.

 8.  Petrovick, M.L.  Patient Owned Systems and Devices  in  Health Care.
     Presented at Texas A&M Conference on "The Impact of Microcomputers
     in Health Care," September 18-19, 1978.  (In preparation.)

 9.  Soricek, B.  Microprocessors and Microcomputers.  New  York,  Wiley—
     Interscience publication,  John Wiley and Sons, 1976.

10.  Stroder, N.R.  The Microprocessor Transducer Function.   Proceedings  of
     the 12th Hawaii International Conference on Systems Science,  pp.  256-
     265, 1979.

11.  Strong, A.T.  Description of the CLEANS Human Exposure System,  EPA R&D
     Report #600/1-78-064.   November 1978.

12.  Van DeWater, J.M. et al.  Monitoring the Chest with Impedance.  Chest
     64:597-603, 1973.

13.  Wallace, L.  Personal  Air Quality Monitors:   Past Uses and Present
     Prospects.  American Chemical Society 4th Joint  Conference on Sensing
     of Environmental Pollutants, pp.  390-394,  1978.

14.  Winner, R.N.  A Study  of the Application of Reflective Displays to
     Synthetic Array Radar.   Hughes Aircraft Company,  Technical Report AFAL-
     TR-73-155.  August 14,  1973.


Mathew L. Petrovick and Edward D.  Haak,  Jr., M.D.
Health Effects Research Laboratory
Clinical Studies Division
U.S. Environmental Protection Agency
Chapel Hill, North Carolina  27514

Personal Cardiopulmonary Electrode
     Paul N. Kizakevich
     Research Triangle Institute
     Research Triangle Park, North Carolina
     Environmental health effect  studies are usually designed to correlate
controlled pollutant dose with health  effect decrement.  In the experimental
laboratory, accurate dose measurements are compared to standard cardiopul-
monary parameters.  Typical physiological measurements Include respiratory
rate, respiratory volume, alveolar-arterial oxygen gradient, heart rate,  and
electrocardiogram stress response (30).  While such experiments provide use-
ful and reproducible information,  they could be supplanted with studies in-
volving subjects performing their normal daily routine by employing personal
exposure and physiological data monitors.

     The initial target organ for airborne pollutants is the lung, and, con-
sequently, respiratory function measurements are commonly examined.  Adverse
pulmonary function may affect oxygen transport with resulting changes in
cardiac performance, especially under  Increased workloads.  As heart and
lung disease are prominent in our society, a significant subpopulation is
sensitized and susceptible to adverse  cardiopulmonary effects.  Although
pollutants which permeate the pulmonary membrane may be transported to all
organ and tissue systems, it is unlikely that such health effects could be
studied with personal monitors.

     Cardiopulmonary measurements are, therefore, the most promising choice
for personal monitor health effect studies; however, the selection of suit-
able nonlnvaslve cardiopulmonary measurement techniques is limited.  The
necessary physiological data must  be acquired with ease, comfort, and accuracy

 in order to gain favor with both the investigator and subject volunteer.

      Using only surface electrodes, the electrocardiogram (EGG) is routinely
 used for monitoring changes in heart rate, rhythm, and muscle physiology.
 A second electrode technique, thoracic electrical impedance (TEI) measure-
 ment, is an experimental method which detects changes in cardiac muscle per-
 formance and pulmonary physiology.  Together, the EGG and TEI provide a
 comprehensive assessment of cardiopulmonary function and physiology which
 appears well suited for personal monitoring and environmental health effects

      The physiology and measurement of  the ECG are well known and routine in
 modern clinical practice (7,12).  Contraction and excitation of cardiac mus-
 cle are accompanied by tissue currents  which give rise to millivolt potential
 differences at the body surface.  The normal ECG presents a rhythmic waveform
 pattern which is fairly reproducible among different subjects with the same
 electrode placement.  Changes in the waveform may occur with drug therapy,
 exercise, cardiac ischemia, and other physiological inputs.

      Although the ECG electrode placement may vary, three-lead differential
 amplifier input is generally used to provide high common-mode noise rejection.
 The V5 precordical lead measures the ECG close to the heart, yields a large
 amplitude with minimal motion artifacts, and is often used in exercise studies
 (Figure 1).  The primary electrode is placed on the mldclavicular line over
 the fifth intercostal space, the reference electrode above the sternum and
 the indifferent electrode on the lower  right thorax.

      In the apparently healthy individual, the ECG is primarily used for heart
 rate measurement.  Since systemic oxygen delivery is determined by cardiac
 output (stroke volume * heart rate), the heart rate is a factor In determin-
 ing the cardiac response to a given workload.  Maximal stroke volume is
 roughly attained at 40 percent of the individual maximal oxygen uptake (1,
 14), and the remaining work response. Is accommodated by increasing heart rate

      While the ECG has been studied for nearly a century, TEI measurement is


FIGURE 1.  Typical configuration for the  TEI  current  source  (solid), TEI
pickup (dotted), and EGG (dotted)  surface electrode Interface.
FIGURE 2.  Composite sketch of TEI  waveform  components:  mean Impedance  (Z  ),
low frequency respiratory waves (DZ),  and  high  frequency cardiac  waves (ICG).

 a relatively  recent  development.   First  measured  in 1932  (2),  continued  re-
 search of  the cardiopulmonary  impedance  effects (27,33) led  to NASA  instru-
 mentation  and clinical  studies support (28).  The fundamental  properties are
 well reviewed (11);  however, the  number  of  recent clinical publications
 lends itself  to  further presentation.

      A variety of  electrode  configurations  have been reported, and each  tends
 to enhance a  specific TEI  physiological  component.   For cardiac  and  mean
 thoracic impedance measurement, the  four annular  electrode system is most
 popular (Figure  1).  A  very  high  frequency  carrier signal (4 ma  at 100 KHz)
 is applied to the  outer electrodes and passes through the thoracic volume.
 At the designated  frequency, the  applied carrier  current  is  completely safe
 (11).  The carrier signal, modulated by  intraelectrode volume  conduction
 change, is monitored by the  inner electrodes and  demodulated to  obtain three
 physiological components (Figure  2). These are the mean  or  base impedance
 (Z ), the  low frequency impedance (AZ),  and the high frequency Impedance (ICG)

      The DC component (Z ) or  mean thoracic impedance is  a measure of intra-
 thoracic fluid volume.   In animals subjected to acute Intravenous hypervol-
 umia and transient saline  injections, a  fall in Z  accompanied Increases in
 intrathoracic fluid  volume (6,17,31).  The  noninvasive impedance measurement
 generally  preceded invasive  central  venous  pressure measurement  by as much
 as 45 minutes (6,37).   Investigators evaluating Z  changes in  human  clinical
 studies have  confirmed  that  mean  thoracic Impedance accurately follows the
 clinical course  of pulmonary congestion, edema, and intrathoracic volume
 overload and  forced  diuresis (20,38,44,45).

      Changes  in  Z  can  occur rapidly for both healthy and unhealthy  individ-
 uals, as was  demonstrated  by a 14-minute treadmill exercise  test (5).  An
 increase in circulating blood  volume and cardiac  stroke volume presented a
 3 percent  Z  decrease during the  healthy subjects'  exercise.   The unhealthy
 subjects,  suffering  from ischemic heart  disease,  had a reduced stroke volume
 reserve and therefore could  not accommodate the increased blood  volume.   Con-
 sequently,  they  had  progressive Intrathoracic fluid retention, as indicated
 by a progressive Z  decrease to a 15 percent total reduction.

      Considering the magnitude of these  Z  changes and that  patients suffering
 pulmonary  edema  may  have a 25  percent Z   reduction,  it is encouraging that
 the normal  Z   may  be predicted to 4.5 percent of  the actual value by modeling
 the thoracic  geometry (21).  Since a decrease in  cardiac  performance was de-

tectable under  exercise  stress, changes in Z  under transient  environmental
stress might  be detected.  Furthermore, by computing an expected Z  value,
thoracic fluid  accumulation  due to chronic exposure may be demonstrated  with-
out requiring a 10-year  study.

     The low  frequency TEI waves (AZ) due to respiration have  been extensively
studied (11).   The  thorax consists of tissues, fluid, and air.  Since air
is an insulator,  the volume  of air respired (AV) is related to the change
in impedance  (AZ).  While the ratio (AZ/AV) differs with individual vari-
ance in size, thoracic circumference, and subcutaneous fat (18), Improved
electrodes and  individual splrometer or pneumotach calibration allow subse-
quent continuous  resplred-volume measurement.

     With calibration, AZ may be used for monitoring respiratory rate, as
is commonly employed for neonatal apnea detection.  Regional lung pathology
is detectable by  comparing AZ when each lung is monitored separately.  The
subject thereby serves as his own control, and, again, calibration is not
required (8).

     The high frequency TEI waves, synchronous with the cardiac cycle, are
the electrical  impedance cardiogram (ICG) and the ICG first derivative (DZ/
DT) (Figure 3).   Over 25 papers and abstracts have been published during
1976-1978 concerning the genesis and application of these waves.

     Considerable evidence indicates that the cardiac impedance pulse re-
flects left and right ventricle ejection into the aorta and pulmonary artery
(3,19,35); however, the actual ventricular volume change contributes less
than 10 percent of  the total Impedance change (35).  The aortic contribution
was predominant in  one study (19) and less than 30 percent in another (35).
Experiments in  a  calf with a totally Implanted cardiac prosthesis presented
an aortic contribution of 60 percent and a pulmonic contribution of 40 per-
cent (3).  Clearly, the origin of the high frequency component is complex and

     A close relation between DZ/DT and aortic blood flow has been observed
(28) and verified (43,46).  Both the amplitude and time-to-peak derivative
of left ventricular pressure (DP/DT) have correlated with their counterpart
in DZ/DT (40,41,46).  However, in one study,  the relation of DZ/DT amplitude
to afterload (aortic dlastolic pressure) depended upon the experimental  in-
tervention (32), which indicates that DZ/DT may be a reliable index of myo-
cardlal contractility under controlled conditions (41,42,46).

  FIGURE  3.   Illustration of temporal relationship  among  the  electrocardiogram
  (ECG),  impedance  cardiogram (ICG),  ICG  first  derivative (DZ/DT), and heart
  sounds  (HS).

*" BUH-hK





HR, :
1 HR

                             LABORATORY SYSTEM

 FIGURE 4.  Functional data flow for laboratory  systolic  time  interval pro-

     Impedance  cardlography  has  long  been a promising method  for  continuous
measurement  of  stroke volume (27,33).  Early studies had resulted In Imped-
ance-calculated volumes  of higher value  than obtained by comparative methods
(11).  Recent Improvements In the Impedance model and Its application have
reduced  the  differences  to within experimental error (9,16,19,26,28,47).
Thus, by combining  the EGG and TE1, heart rate, stroke volume, and cardiac
output are measurable.

     The sequence of cardiac mechanical  events, collectively  called  systolic
time Intervals  (STI), correspond to DZ/DT waveform features (29).  An  auto-
mated system for STI measurement using DZ/DT and the heart sounds was  first
developed In 1974 (22).  The excellent correlation of DZ/DT STI features  to
aortic and carotid STI features  Is now well established (4,10,15,39).  DZ/DT
STI measurements have proved reliable In exercise (13,24,36) and  postural
(42) stress  testing as indices of myocardial stress response.

     The necessity for EGG and TEI automated signal processing arises from
the complex nature In which the clinical information is contained.  Certain
waveform features must be identified, interfeature amplitude and time meas-
urements made, and relevant clinical indices computed.  As noninvasive
methods, both signals are subject to physiological artifacts, transductlon
artifacts, and uncertainty resulting from their indirect and remote trans-
ductlon of the physiological source.  The TEI is sensitive to a variety of
physiological effects which must be isolated prior to measurement.  These
challenges and their solution have been addressed by analog filter prepro-
cessing, cardiac component ensemble averaging, and waveform feature extrac-
tion algorithms (23,24).

     The acquisition and analysis of the EGG, Z , and DZ/DT signals have been
the author's principal interest in this field.  Recently, a laboratory-based
system for cardiac monitoring during treadmill exercise using these signals
was developed under EPA Contract No. 68-02-2772 (25).  As an example of the
signal processing methods, the ensuing discussion will be limited to the
current system design rather than to project-specific processing requirements
for a personal monitor.

     The EPA Systolic Time Interval Processor (STIP) uses analog signal
conditioning prior to digital signal processing and analysis.  The EGG is

 FIGURE 3.  Illustration of  temporal relationship among the electrocardiogram
 (ECG), impedance  cardiogram (ICG), ICG first derivative (DZ/DT), and heart
 sounds (HS).

* BUhl-tH






HR, ;



                            LABORATORY SYSTEM

 FIGURE 4.   Functional data flow  for laboratory systolic time interval pro-

      Impedance cardiography has long been a promising method for continuous
 measurement of stroke volume (27,33).  Early studies  had resulted in imped-
 ance-calculated volumes of higher value than obtained by comparative methods
 (11).  Recent  improvements in the impedance model  and its application have
 reduced  the differences to within experimental  error  (9,16,19,26,28,47).
 Thus, by combining the EGG and TEI,  heart rate,  stroke volume,  and cardiac
 output are  measurable.

     The sequence of cardiac mechanical events,  collectively called systolic
 time intervals (STI),  correspond to  DZ/DT waveform features  (29).   An auto-
 mated system for STI measurement using  DZ/DT and the  heart sounds  was first
 developed in 197A (22).   The excellent  correlation of DZ/DT  STI  features  to
 aortic and  carotid STI features is now  well established  (4,10,15,39).  DZ/DT
 STI measurements have  proved reliable in exercise  (13,24,36)  and postural
 (42) stress testing as indices of myocardial stress response.

     The necessity  for EGG and TEI automated signal processing arises from
the complex nature  in which  the  clinical information is contained.  Certain
waveform features must be identified, interfeature amplitude and time meas-
urements made, and  relevant  clinical  indices computed.  As nonlnvaslve
methods, both signals are subject to  physiological artifacts, transduction
artifacts, and uncertainty resulting  from their indirect and remote trans-
duction of the physiological  source.  The TEI is sensitive to a variety of
physiological effects which must be isolated prior to measurement.  These
challenges and their solution have been addressed by analog filter prepro-
cessing, cardiac component ensemble averaging, and waveform feature extrac-
tion algorithms (23,24).

     The acquisition and analysis of  the ECG, ZQ, and DZ/DT signals have been
the author's principal interest in this field.  Recently, a laboratory-based
system for cardiac  monitoring during  treadmill exercise using these signals
was developed under EPA Contract No. 68-02-2772 (25).  As an example of the
signal processing methods, the ensuing discussion will be limited to the
current system design rather  than to  project-specific processing requirements
for a personal monitor.

     The EPA Systolic Time Interval Processor (STIP) uses analog signal
conditioning prior  to digital signal  processing and analysis.  The ECG is

 passed through an analog bandpass filter (0.05 to 50.0 Hz) to remove elec-
 trode potential baseline shifts and to reduce 60 Hz and high frequency muscle
 noise.  The filtered ECG is passed through a QRS event matching filter (7.2 to
 23.4 Hz) to produce the RWAVE event detection signal.  The TEI signals (ZQ,
 AZ, DZ/DT) are bandwidth filtered within the commercial impedance instrument
 and only require amplitude and baseline adjustment.

      The ECG, DZ/DT, and RWAVE are sampled at a 400-Hz rate and held in  a
 sample buffer (Figure 4).  Each new cardiac cycle is detected by an adaptive
 trigger algorithm that adjusts the trigger level to the running average
 RWAVE height.  Continuous RWAVE detection is thus assured without manual
 intervention.  With each new cycle, Z  is sampled and stored with DZ/DT  and
 the ECG QRS complex in a cycle buffer.  At the same time, the instantaneous
 cardiac cycle duration is used to compute and queue the instantaneous heart

      When a cardiac function measurement is requested, the remaining processes
 are activated.  In order to further isolate the TEI cardiac component and
 to improve both the QRS and DZ/DT signal-to-noise ratios, an ensemble of the
 acquired cycle buffers is averaged.  Since the systolic DZ/DT waveform is
 dependent upon the instantaneous cardiac cycle duration—a natural heart rate
 variance—premature ventricular contraction or other anomalous ventricular
 event would invalidate the ensemble average.  By selectively averaging cycles
 of similar duration, these problems are reduced.

      To do this, the heart rate buffer is analyzed for the mean expected
 heart rate (HR), and a 12.5 percent acceptance range (MINHR, MAXHR) is deter-
 mined.  Each  successive cycle is compared to these criteria prior to Inclusion
 in the  ensemble summation process.  The resultant averaged ECG and DZ/DT
 waveforms are then smoothed and displayed (Figure 5).

      The major clinical endpoints for the EPA system are the cardiac systolic
 time  intervals which are relative timing measures between waveform features.
 Each waveform feature (0,  ..., 4) is identified in several stages:  1) the
 average heart rate is used to estimate the feature waveform segment; 2)  the
 segment waveshape pattern is classified; and 3) the feature is identified
 using an algorithm based on the pattern classification.  A list of quality
 control parameters and selected clinical indices is then displayed.  Finally,
 the  investigation may accept, reject, or flag the cycle for further analysis.

      The data base management design has a comprehensive data structure  suited

                                   roe cmmi/t TO
             I    I    I    f    I    T   I    I    I    I    I
             •   3lllli5l2*2»3»35l4ll«5l3»
              I. MM* RMEM  MIMTE  «UCTI CYOB EFFCY.  90
              3   tf2   Ctt    tt    M    II    IN   07
             M   PEP  TOOK  TZMX  BK   03  TZINE T2NLL  L«T
             29   HI   139   197   IB   441    47   179   IB
FIGURE 5.  Laboratory system graphic display of averaged QRS and DZ/DT waveforms,

waveform feature cursors, and resultant data list.


         79 .
"^       ^A
         91-1 98 -J
        STI , STILUS <\>, EXIT 
 for experimental laboratory analysis.  Over a 12-minute acquisition period,
 1,570 cardiac cycles may be saved on double-density floppy disc with heart
 rates ranging from 40 to 175 beats/minute.  Along with each sampled waveform
 is a data vector containing instantaneous cycle duration, relative time,
 exercise stage, experimental protocol, pollutant, and other individual cycle
 data.  By keying on a specific data vector element, the raw data are easily
 recalled for automatic waveform analysis.

      The waveform analysis output data are held in a memory array and stored
 on floppy disc.  Editing, transfer, and listing of the disc data are executed
 via the memory array.  For qualitative review of the treadmill exercise re-
 sponse, a graphics procedure recalls select cardiac indices and displays
 their response trend (Figure 6).  Grouped statistical analysis and distributed
 processing data transfer are further capabilities which could be implemented.

      The EPA STIP laboratory system described in the last section will receive
 clinical and environmental health effect evaluations during the current year.
 While a reduction in size and function may be a useful initial approach for
 a personal monitor design, an expanded outlook for research and development
 may present a better formulation.

      The body-electrode Interface, electrode placement configuration, and
 ECG and TEI signal transduction comprise the electrode interface system
 (Figure 7).  Each of these topics must be reviewed in the biomedical engi-
 neering literature and alternative technologies studied for optimum trans-
 duction of the desired cardiopulmonary signal components.  Since no signifi-
 cant change in TEI electrode technique has emerged in the last 10 years,
 it is likely that improved electrodes and impedance detection electronics
 could be developed using today's technology.  The current STIP system would
 be used as a basis for controlled clinical and subsequent environmental health
 effect studies in the evaluation of electrode interface system developments.

      The commercially available instrument for TEI measurement occupies a
 volume of 1 cubic foot.  This instrument must be redesigned as a functional
 electronic component within the overall system concept.  In this way, the
 electrode interface and control, ECG and TEI transduction, TEI component
 isolation, analog preprocessing, and digital signal processing could be in-
 tegrated as a single hybrid microcircuit module.  The output data would



& I/O
                          SIGNAL ANALYSIS & SYSTEM CONTROL
FIGURE 7.  Functional  diagram of proposed personal cardlopulmonary monitor.
represent analog  and  digital preproceseed waveform complexes ready for pattern
recognition and feature analysis.

     Specific physiological waveform processing,  data base input/output, and
overall system characteristics in  communicating with other monitoring devices
are general systems engineering problems  which require study and formal
specification.  Current techniques for automated  waveform analysis would be
reviewed, and—based  upon the intended application—appropriate software
would be designed.  The data base  and system functions would be governed
by the monitoring application, yet would  be designed in a standard format.
Implementation of the signal analysis and system  control module would rely
on standard microcomputer components.  As a result, reduction of these system
functions from the laboratory system would be straightforward.

     The clinical validation is a  necessary aspect of any biomedical system
development.  At  the  very least, the instrument must operate as specified
and record valid  physiological measurements.  Using controlled environmental
response studies  and  Inpatlent clinical observations, the cardiopulmonary
electrode monitoring  concept would be tested.   Based on known and expected
response patterns, experimental studies would be  designed, performed, and
evaluated within  the  environmental laboratory setting.  In this manner, the

sensitivity, reproducibility, and accuracy of the personal cardiopulmonary
electrode monitor would be validated.

     The need for simultaneous personal monitoring of physiological response
and pollutant exposure was addressed, and a method for the required physio-
logical data acquisition was presented.  An overview of the physiology,
transduction, signal processing, and significance of the electrocardiogram
and thoracic electrical impedance were discussed, along with research and de-
velopment requirements for a personal cardiopulmonary monitor.  The further
investigation of these topics and the development of the suggested monitor
would provide new insight into environmental health effects and most likely
transfer its application to noninvasive monitoring of critically ill patients.

 1.  Astrand, P., Cuddy, T.E., Saltin, B., Stenberg, J.  Cardiac Output During
     Submaximal and Maximal Work.  J Appl Physiol 19(2):268-274, 1964.

 2.  Atzler, E., Lehmann, G.  Uber ein Neues Verfahren zur Darstellung der
     Herztatigkeit (Dielektographie).  Arbeitsphysiol 5:536, 1932.

 3.  Baker, L.E.  Thoracic Impedance Changes During Ventricular Ejection.
     Fed Proc 36:544, 1977.

 4.  Balasubramanian, V., Mathew, O.P., Behi, A., Tewari, S.C., Boon, R.S.
     Electrical Impedance Cardiogram in Derivation of Systolic Time Inter-
     vals.  British Heart J 40:268-275, 1978.

 5.  Balasubramanian, V., Boon, R.S.  Changes in Transthoracic Electrical
     Impedance During Submaximal Treadmill Exercise in Patients with Ischemic
     Heart Disease—A Preliminary Report.  Am Heart J 91(l):43-49, 1976.

 6.  Berman, I.R., Scheetz, W.L., Jenkins, E.B., Hufnagel, H.V.  Transthoracic
     Electrical Impedance as a Guide to Intravascular Overload.  Arch Surg
     102:61-64, 1971.

 7.  Burton, A.C.  Physiology and Biophysics of the Circulation.  Chicago,
     Year Book Medical Publishers, 1965.

 8.  Cathignol, D., Fourcade, C., Jossinet, J., Lavendier, R., Wiesandanger, T.
     Interface for Pulmonary Exploration by the Real-Time Treatment of Curves
     Obtained by Transthoracic Impedance.  Med & Biol Eng & Comput 16(5):
     459-470, 1978.

 9.  Denniston, J.C., Maher, J.T., Reeves, J.T., Cruz,  J.C.,  Cymerman,  A.,
     Grover, R.F.  Measurement of Cardiac Output by Electrical Impedance
     at Rest and During Exercise.  J Appl Physiol 40(l):91-95,  1976.

10.  Ewing, K.L., Poder, T.C., Baker, L.E., Rubal, B.J.,  Gutgeshell, E.G.
     Measurements of Left Ventricular Systolic Time Intervals by Electrical
     Impedance.  Federation Proceedings 36:518, 1977.

11.  Geddes, L.A., Baker, L.E.  Principles of Applied Biomedical Instrumen-
     tation.  New York, Wiley & Sons, 1975.

12.  Geselowitz, D.B., Schmidt, O.K.  Electrocardiography.  In:  Biological
     Engineering (Schwan, H.P., ed.).  New York, McGraw-Hill, 1969.

13.  Gollan, F., Kizakevich, P.N., McDermott, J.  Continuous  Electrode Moni-
     toring of Systolic Time Intervals During Exercise.   British Heart J
     40:1390-1396, 1978.

14.  Hermansen, L., Ekblom, B., Saltin, B.  Cardiac Output During Submaximal
     and Maximal Treadmill and Bicycle Exercise.  J Appl  Physiol 29(1):82-
     86, 1970.

15.  Hill, D.W., Merrifield, A.J.  Left Ventricular Ejection  and the Heather
     Index Measured by Non-invasive Methods During Postural Changes in Man.
     Acta Anaesth Scand 20:313-320, 1976.

16.  Hill, D.W., Thompson, F.D.  The Importance of Blood  Resistivity in the
     Measurement of Cardiac Output by the  Thoracic Impedance  Method.  Med &
     Biol Eng 13(2):187-190, 1975.

17.  Hukushlma, Y.   Physiological Identification of Variation Sources of
     Transthoracic Electrical Impedance During Breath Holding.  Jap Heart J
     11(1):74-90, 1970.

18.  Hull, E.T., Irie, T., Heemstra,  H., Wildevuur,  R.H.  Transthoracic
     Electrical Impedance:  Artifacts Associated with Electrode Movement.
     Resuscitation 6:115-124, 1978.

19.  Ito, H., Yamakoshi, K., Yamada,  A. Physiological and Fluid-Dynamic
     Investigations of the Transthoracic Impedance Plethysmography Method
     for Measuring Cardiac Output:  Part 2—Analysis of the Transthoracic
     Impedance Wave by Perfusing  Dogs.   Med & Biol Eng 14(4):373-378, 1976.

20.  Keller, G., Blumberg, A.  Monitoring  of Pulmonary Fluid Volume and Stroke
     Volume by Impedance Cardiography In Patients on Hemodialysis.  Chest
     72(l):56-62, 1977.

21.  Khan, M.R., Tandon, S., Guha, S.K., Roy,  S.B.   Quantitative Electrical-
     Impedance Plethysmography for Pulmonary Oedema.  Med & Biol Eng & Comput
     15:627-633, 1977.

 22.  Kizakevich,  P.N.,  Gollan,  F.  On-Line Measurement of Systolic Time In-
      tervals  Using  the  First Derivative Impedance Cardiogram and the Heart
      Sounds.   Proc  Ann  Conf Eng Med Biol 16:132, 1974.

 23.  Kizakevich,  P.N.,  McDermott, J., Gollan, F.  An Automated System for
      Systolic Time  Interval Analysis.  Proc Digital Equip Users Soc 2(3):
      795-798, 1976.

 24.  Kizakevich,  P.N.,  Gollan,  F., McDermott, J., Aranda, J.  Continuous
      Non-invasive Monitoring  of Myocardial Function:  Its Application in
      Exercise Stress Testing.   Proc Symp Comp Appl Med Care, IEEE Catalog
      No. 77CH1270-8 C 5:   325-336, 1977.

 25.   Kizakevich,  P.N.,  Beadles, R.L.  Feasibility Design of an Impedance
      Cardiogram for Exercise  Systolic Time Intervals.  Technical report pre-
      pared for Environmental  Protection Agency contract #68-02-0722.  June

 26.   Kobayashi, Y., Andoh, Y.,  Fujinami, T., Nakayame, K., Takada, K.,
      Takeuchi, T.,  Okamoto, M.  Impedance Cardlography for Estimating Cardiac
      Output During  Submaximal and Maximal Work.  J Appl Physiol: Resplrat
      Environ Exercise Physiol 45(3):459-462, 1978.

 27.   Kubicek, W.G., Karnegis,  J.N., Patterson, R.P., Witsoe, D.A., Mattson,
      R.H.  The Development and Evaluation of an Impedance Cardiac Output
      System.  Aerosp Med  37:1208, 1966.

 28.   Kubicek, W.G., Patterson,  R.P., Witsoe, D.A., From, A.H.L.  Impedance
      Cardiography as a  Noninvasive Method of Measuring Cardiac Output and
      Other Cardiac  Parameters of  the Cardiovascular System.  NASA, Houston,
      Texas, 1969.

 29.   Labadidi, Z.,  Ehmke, D.A., Durin, R.E., Leaverton, P.E., Lauer, R.M.
      The First Derivative Thoracic Impedance Cardiogram.  Circulation 41:
      651, 1970.

 30.   Linn, W.S., Hackney, J.D.  Health Effects of Air Pollution.  J Cardio-
      vascular & Pulmonary Technol 7(2):50-55, 1979.

 31.   Luepker, R.V., Michael,  J.R., Warbasse, J.R.  Transthoracic Electrical
      Impedance:  Quantitative Evaluation of a Non-invasive Measure of Thoracic
      Fluid Volume.   Am  Heart  J 85(l):83-93, 1973.

 32.   Moore, A.G., Baker,  L.E., Rubal, B.J., Poder, T.C.  The Relationship
      Between  the Maximum  Rate of  Change of the Thoracic Impedance Waveform
      and Afterload.  Federation Proceedings 36:518, 1977.

 33.   Nyboer, J.  Electrical Impedance Plethysmography.  Springfield, 111.,
      Charles C. Thomas, 1959.

 34.  Patterson, R.P.  The Use  of Signal Averaging of the Electrical Imped-
     ance Signal  to Determine  Cardiac Timing Information During Uninterrupt-
     ed Exercise.  AAMI  13th annual meeting, Washington, D.C., 1978.

 35.  Patterson, R.P., Kubicek, W.G., Prom, A.H.L., Witsoe,  D.A.  Studies on
     the Effect of Controlled Volume Change on the Thoracic Electrical Im-
     pedance.  Med & Biol Eng  & Comput 16:531-536, 1978.

 36.  Poder, T.C., Gutgesell, H.P., Baker, I.E.  Application of Impedance Car-
     diography to Cardiovascular Stress Testing in Children.  Fed Proc 37:
     776, 1978.

 37.  Pomerantz, M., Baumgartner, R., Lauridson, J., Eiseman, B.   Thoracic
     Electrical Impedance for  the Early Detection of Pulmonary Edema.   Surgery
     66:260, 1969.

 38.  Ramos, M.U., LaBree, J.W., Remole, W., Kubicek, W.G.   Transthoracic Elec-
     tric Impedance.  Minn Medicine 58:671-676, 1975.

 39.  Rasmussen, J.P., Sorensen, B., Kann, T.  Evaluation of Impedance  Car-
     diography as a Noninvasive Means of Measuring Systolic Time  Intervals
     and Cardiac Output.  Acta Anaesth Scand 19:210-218,  1975.

 40.  Rubal, B.J., Poder, T.C., Baker,  L.E.  Correlations  Between  t-max DP/DT
     and t-max DZ/DT.  Federation Proceedings 37:776, 1978.

 41.  Siegel, J.H., Fabian, M., Lankau, C., Levine,  M.,  Cole, A.,  Nahmad,  M.
     Clinical and Experimental Use of Thoracic Impedance Plethysmography in
     Quantifying Myocardial Contractility.  Surgery 67:907,  1970.

 42.  Smith, J.J., Porth, C.J., Reinke, J.A., Ebert,  T.J., Tristani,  F.E.
     Impedance-Derived Cardiac Indices in Coronary Heart Disease.  Physiolo-
     gist 20:88, 1977.

43.  Steigbigel, R.T., Babbit, H.I.,  Fox,  S.M., Warbasse, J.R.  Quantitative
     Evaluation of the First Derivative of the Impedance Cardiogram.   Cir-
     culation 42(111):65, 1970.

44.  Tempel, G., Jelen,  S.,  Hundelshausen, B.V.  Transthoracic Electrical
     Impedance in Anaesthesia and Intensive Care.   Resuscitation 6:97-105,

45.  Van De Water, J.M., Mount, B.E.,  Barela,  J.R.,  Schuster, R., Leacock, F.S.
     Monitoring the Chest with Impedance.   Chest 64:597-603, 1973.

46.  Welham, K.C., Mohapatra,  S.N., Hill,  D.W., Stevenson, L.  The First
     Derivative of the Transthoracic Electrical Impedance as an Index  of
     Changes in Myocardial Contractility in the Intact Anaesthetised Dog.
     Intensive Care Med  4:43-50,  1978.

47.  Yamakoshi, K., Ito, H., Yamada, A., Miura, S., Tomino, T.  Physiological
     and Fluid-Dynamic Investigations of the Transthoracic Impedance Plethys-
     mography Method for Measuring Cardiac Output:  Part 1—A Fluid-Dynamic
     Approach to the Theory Using an Expansible Tube Model.  Med & Biol Eng
     14(4):365-372, 1976.

Paul N. Kizakevich
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, North Carolina  27709
      PETROVICK:  Matt Petrovick, EPA.  Paul, could you give the group an
 idea  about  the  practicality of the impedance cardiograph measurements in
 a  roving  mobile subject?  For example, let's take this scenario: when you
 get up in the morning, you hop in your car or you walk to work, you get on
 a  bus,  you  sit  down, you go to lunch, and you do all your daily activities,
 and we would  like  to record periodically the measurements of systolic time
 intervals through  this process.

      How  practical do you think this technique is with regard to the severe
 environment of  the artifact from the electrodes?  Would you please address

      KIZAKEVICH:   As I have already demonstrated, we have been able repeatedly
 to monitor  patients doing treadmill exercises.  In fact, from the work in
 Miami, we have  a paper published in the Heart Journal, December 1978, demon-
 strating  the  expected response of systolic time intervals in treadmill ex-
 ercise through  a stage four program.  And that is about as severe as you are
 going to  get.   You know, that is equivalent to jogging—much faster than jog-
 ging  down the  street.  I think there is no problem.

 A Respiration-Controlled Personal Monitor
     Donald J. Sibbett, Ph.D., and Rudolph H. Moyer, Ph.D.
     Geomet, Inc.
     Pomona, California
     Although a considerable variety of instrumentation such as impingere,
cascade impactors, battery-powered air samplers, diffusion collectors, and
gas-stain detector tubes is available for  applications relating to personal
monitoring in industrial and ambient environments, few of the results obtained
by such devices are easily relatable to the exposure or dosage of individuals
to noxious airborne components.   Continuous, electrically powered, and diffu-
sion-controlled collectors fail  to measure the varied physiological sampling
rates imposed in response to diverse demands for oxygen during a range of
activities*  Thus, exposure to pollutants  during periods of high activity
which result in deep and rapid inspiration are weighed equally, in a statis-
tical sense, with periods of shallow breathing, by collection devices uti-
lizing uniform sampling rates.  For accurate correlation of exposure with
health phenomena, it is highly desirable that air sampling for analytical
chemical determinations be proportional to the ventilation rate of Individuals
under study.

     The Geomet Personal Sampler provides  a method for sampling air in pro-
portion with the respiration of  the wearer.  The method is characterized
by its use of a carefully designed air  pump which is supported next to the
diaphragm.  The pump is activated by the expansion and contraction of the
thoracic cavity during respiration and  draws air through a collector which
may be varied in accordance with monitoring requirements.

     The Geomet respiration-controlled  personal monitor consists of five com-
ponents:  1) the pump, 2) a harness or  vest, 3) a sample volume accumulator,

4) a separated digital readout, and 5) a pollutant collector.  With the digital
readout excepted, the total assembly weighs slightly over  1  pound  in its current
configuration.  The pump and harness weigh 293 g; the battery  and  accumulator,
207 g; and a  typical gas sampler assembly weighs 30 g.  A  particulate sampling
head such as  the Bendix Sampling Assembly, Part No. 3900-90, weighs 128 g.

     Figure 1 shows the pump, harness, and signal accumulator  which are worn
during sampling.  The pump shown in the center of the picture  is made from
black ABS plastic.  This version, which normally pumps  in  the  150  to 250
ml/minute range, is 15.5 cm in height, 13 cm wide, and  3.7 cm  thick when
expanded.  Compressed, it is less than 2 cm thick.

     The  pump housing is connected to a lightweight Velcro harness by three
easily connected and adjusted plastic buckles.  The harness  is passed around
the wearer's  abdomen and connects to the side ears on the  outer  pump panel.
Over—the-shoulder straps, connected to the circumferential band  in the rear,
support  the  inner pump panel at its top.  The harness may  be adjusted for
abdominal measurements up to whatever dimensions are necessary.  By use of
Velcro  tabs,  the excess strap is fastened back on itself.  Dimensional ad-
justment  of  the  harness is made at the buckles; the Velcro fastenings are
only  used to  avoid  dangling ends.

      The  pump performance in terms of air displacement  is  monitored by use  of
an electronic accumulator which is shown at the right side of  Figure 1.
This  cigarette  package-sized instrument contains a 9-volt  battery  and elec-
tronic  circuitry to monitor the motion of a linear potentiometer located with-
in the  pump  housing.  The signal from the pot is transmitted through a shield-
ed lead  which fastens to one end of the accumulator through  a  miniature
clamped  hexagon  connector.  The accumulator measures and stores  potentiometer
motion  in both  directions.  The absolute magnitude of this signal  is estab-
lished  by screwdriver adjustment of a trim-pot through  the package wall.
After  calibration and adjustment, the accumulator may read out in  any desired
volume  units.  As normally calibrated, the readout unit operates in milli-

      A closer look  at the pump is shown in Figure 2.  The  cross  section en-
titled B  on  the  right shows the extended position of the pump.  The pump Is
returned  to  this position by the spring, which is shown schematically at the
bottom of the figure.  All working components are contained  within the molded
housing  and  are  protected by it from accidental damage.

FIGURE 1.  The Geomet  Personal Monitor.  Pump,  harness, accumulator, and  a

typical harness.
                          ) 1   Molded Housing
                                           Valve \
                                                          to Accumulator
 FIGURE 2.  The Geomet Personal Monitor pump,

     Abdominal wall motion moves the base section of the  housing  to  Impart
a pumping motion.  On expiration, the spring drives the two  housing  sections
apart and two flapper valves control the air intake.  On  expiration,  the
silicone rubber diaphragm is extended, drawing air through a sampling tube
Into the pump body through a tubing connector visible at  the lower center
of the diagram.  On compression of the diaphragm or expansion of  the  thoracic
cavity, the inlet valve closes and the exhaust valve—located between the rub-
ber diaphragm and the pump cover—opens.  Repeated flexing of the diaphragm
produces pumping action.  The signal-generating linear potentiometer  is lo-
cated on the left side of the housing.  The signal is passed to the  accumu-
lator by a cable extending from the housing on the right-hand side.

     This device, in the described size, pumps in the range  from  150  to 250
ml-per-minute during most activities.  Changed sampling rates may be  achieved
by modification  of the housing size.

     To use  the  system, the harness is placed over the head,  the  vee-shaped
section is connected to the top coupling, and the circumferential belt is
connected to  the sides of the pump.  Appropriate tension  is  achieved  by pull-
ing  the side  belting until the pump diaphragm is just slightly compressed dur-
ing  normal exhalation.  Harness tension should be barely  noticeable.   Excessive
tightening of  the harness will result in poor performance of the  pump*  With
such incorrect adjustment, the pump will bottom (or close) during inhalation.
The  harness  may  be held down by clips to a belt, if appropriate.

     The procedure to determine the volume of air sampled during  any  interval
is  carried out by use of a digital readout unit.  This device, measuring
13 x 11 x 16.5 cm, is utilized at the start and termination  of any operation.
At  all  other  times,  it is disconnected from the sampling  device.  Figure 3
shows  the readout unit.  Each switch has two operating positions; the mode
switch  on the  left permits direct recovery of the signal  stored in  the ac-
cumulator or gives a continuous monitoring of that storage.   The  up  position
gives  the stored value at the time that the center switch is depressed.  In
the  other position,  the mode switch gives continuous readout, which  is basi-
cally useful  for demonstrations or short-calibration measurements.  In the
depressed position,  the clear switch zeros the storage system when  the center
switch  is activated.  Elevated (the normal experimental position),  the stored
signal  is displayed on the digital meter.  This procedure does not destroy
the  storage.

     The digital readout displays in six digits, which in current usage is
sufficient to sample 1 cubic meter less 1 ml.  It is connected to a 115-V


FIGURE 3.  Digital readout, connected to electronic accumulator.
                                               FIGURE 4.  Personal monitor
                                               in use.

60-Hz  source  by means  of  the power line from the back of the case.  The
five-prong miniature hexagonal connector without the clamp is connected to
the  matching  terminal  on  the accumulator for use in measuring operations.
The  readout device can be modified for operation on battery power.

     At  the start of any  sampling operation—after the pump, harness, and
accumulator are adjusted  on a wearer—the accumulator is connected to the
readout  unit. With the clear switch in the depressed position and the mode
switch elevated,  the middle switch is activated.  This zeros the accumulator
and  constitutes  the beginning of the sampling interval.  The accumulator is
then disconnected from the readout.  The accumulator commences operating as
soon as  the  zeroing process is completed.  However, the clear switch position
should be changed immediately to avoid loss of future data when another
challenge is  made.

      The personal sampler assembly may be used for collection of a wide
variety of airborne components.  Figure 4 shows a typical example of the
 full assembly set up to collect a gaseous component on a coated tube col-
 lector.   The  collector in this case consists of a 20-cm ceramic tube which
 is located under  the collar clip.  It is protected by a tygon sheath.  Air
drawn from the neighborhood of the wearer's head passes through the tube.
The desired  pollutant  is  collected on the inner surface of the tube by an
 appropriate  coating.  For example, sulfur dioxide may be collected on a coat-
 ing of sodium carbonate,  and nitrogen dioxide on triethanolamine.

      At  the  end  of the collection interval, the coated collector tube is
 placed in a  screw-capped  test tube and transmitted to a laboratory for anal-
 ysis.   Table  1 shows a selection of components which have been sampled by
 the personal  monitor technique.

      Mercury  is  quantitatively removed by a short silver wire absorber packed
 randomly in  a 5-cm section.  It is desorbed from the silver in an induction
 furnace and  passes directly into a UV photometer for quantitation.  This
 technique is  also used to calibrate the sampling volume of the collector/
 accumulator  components.  For this purpose, a constant concentration mercury
 source was set up which gave 5 nanograms-per-liter of mercury vapor.  The
 silver plug  collector was used to sample this stream by operation of the
 personal monitor  pump. The accumulator pot was adjusted until the digital
 readout  was  1 ml-per-digit.

                                     TABLE 1

                        Applications of Personal Monitor

Mercury                          Silver                  UV Photometry
Hydrogen Chloride               Sodium Carbonate         Luminol Chemilumines-
                                                          cence (LC)
Nitric Acid                      Sodium Chloride          LC
Sulfur Dioxide                   Sodium Carbonate         LC or Flame Photo-
                                                          metric Detection
     The  second  example utilizes  collection  of hydrogen chloride on a sodium
 carbonate coating  placed on a  small-bore  collector  tube which is part of  the
 gas inlet line to  the  pump.  The  acidic gas  diffuses to the wall during
 passage through  the  collector  and is  retained with  high efficiency (>95 per-
 cent)*  The collector  tubes are returned  to  the laboratory for analysis•
 We have utilized a procedure which involves  rinsing the carbonate/chloride
 coating off the  tube with a solution  of potassium iodate in sulfurlc acid.
 The evolved gas  may  be quantitated by use of the Geomet lumlnol chemilumines-
 cence instrument.  For hydrogen chloride, collection of about 50 nanograms
 is required.

     Similar techniques are  available for the collection and analysis of
 nitric acid and  sulfur dioxide.   Nitric acid is collected on a sodium chlo-
 ride coating.  In  the  laboratory,  the analysis uses a sulfuric acid rinse
 and the luminol  chemlluminescence method for quantitation.  Sulfur dioxide
 is collected on  sodium carbonate  and  may be analyzed by a similar luminol
 chemiluminescence  technique  or by flame ionization  photometry.  Nitrogen
 dioxide may be collected  on  a triethanolamine coating,  diazotized,  and read
 out spectrophotometrically.  These methods require  collection of 10 to 40
 nanograms of the pollutant.

     Other applications  of the collection technique may involve small col-
 umns of absorbent  in place of the  coated diffusion collector tube.   Table
 2 shows a number of  potential applications.  For example,  a variety of organic
 compounds may be collected on styrene-dlvinylbenzene copolymer spheres such
 as Chromasorb 102  or Tenax-GC.  Thermal desorption  Into a gas chromatograph
or a gas chromatograph/mass  spectrometer combination provides analysis.

                                     TABLE  2

                       Applications  of Personal  Monitor

Selected Organic Compounds    Styrene-Divinylbenzene   Gas Chromatography
                                                        (GC) or GC/Mass
                                                        Spectrometry (MS)
Vinyl  Chloride                Charcoal                  GC
Pesticides                    Supported  Ethylene       GC/MS
                              Polyurethane Foam
Carbon Dioxide                Hydrazine-Crystal        Spectrophotometry
      Likewise,  vinyl  chloride may be collected on charcoal and desorbed into
 a gas chromatograph for analysis.  Pesticides  may be collected on a variety
 of materials and  analyzed similarly.  A novel  procedure for carbon dioxide
 utilizes a tube coated with hydrazine-crystal  violet followed by a specto-
 photometric determination of the remaining  colored species.

      In general,  procedures can be devised  which are specific and adequately
 sensitive for most industrial gaseous species.  However,  care must be exer-
 cised to match the sample collector, sample volume,  air concentration,  and the
 sensitivity of the analytical laboratory technique.

      Although this version of a personal monitor may be used without corre-
 lation with the respiration rate of the wearer,  a number of applications are
 of interest in which  dosage or exposure data may be  obtained.

      Table 3 shows a  test which compares the volume  sampled by the pump with
 splrometer measurements of the wearer's respiration.  In these tests a plas-
 tic bag was filled with an air sample and pumped out by the personal monitor
 pump, which was activated in position on a  wearer.   The exhaled volume was
 measured by a simple  spirometer.   As the data  show,  in this test about 4.3
 percent of the  exhaled volume was sampled by the pump.

      A second test was run on the same subject after the harness was removed,
 replaced,  and readjusted.   Table 4 shows the data from the second test.  In
 this series, about 4.8 percent of the respired volume was sampled.

      The difference in the fraction sampled in the two tests was 10.8 percent.


                                     TABLE  3
                             Respiration Correlation
                                    TEST NO.  1
Volume (1)

Volume (1)

Mean 0.0431
C.V. 4.3 percent

Respiration Correlation

Volume (1)

Volume (1)
5.2 percent
This may be considered to be a tentative estimate of the reproducibility of
the sampling technique on a single wearer.  Obviously, considerably more
data is needed on various individuals in a variety of activities before the
limits of the correlation are established.

     Patent applications on this device are pending.


 Donald J. Sibbett, Ph.D., and Rudolph H. Moyer, Ph.D.
 Geomet, Inc.
 2814-A Metropolitan Place
 Pomona, California  91767
       PETROVICK:   Matt Petrovick, EPA.  I had a question regarding your  com-
  ments on the placement  of  the  device over  the abdominal wall.  I wonder if
  you have had a problem  in  subjects who might be abdominal  breathers  In  one
  case, and in the other  case, thoracic breathers; and how you covered both
  of those kinds of breathers?

       SIBBETT:   Actually, our anticipation  is that  the correlation might be
  established for each individual.  A short  calibration with a spirometer or
  something of that type  would be necessary  at the beginning In  order  to  have
  the dosage correlation  at  the  end.  In short, I don't know how good  a gen-
  eralized statement is on that  point.

       STETTER:   Joe Stetter, Energetics Sciences.   In terms of  the way your
  pump operates, does it  pump when the subject is inhaling or exhaling, or at
  the same time?  And if  so, would you give  us an idea of what the flow rate
  through the sampling tube  would look like  as a function over time?

       SIBBETT:   In general, each stroke is  worth about 30 milliliters in each
  direction if it is a full  stroke.  In normal respiration—meaning with  the
  subject at rest—about  40  percent of the time is actually  involved in move-
  ment of the pump.

       Roughly 60 percent of the time the pump is inoperative.

       BEADLES:   Bob Beadles, Research Triangle Institute.   You  have an inter-
  esting portable, self-powered  device.  Do  you have any idea of what  the elec-
  trical equivalent power might  be with it?  You can use it  in a variety  of ways.
  But in a very simplistic way you can drive a super miniature electrical
  generator with it.  What is the equivalent power?

       SIBBETT:   I don't  know.   We have used these accumulator modules with the
  9-volt batteries since  last September.  The 9-volt charge  does quite well for
  the data storage.

       MAGE:   Dave Mage.  Bob, you were asking if there were an  electrical
  conductor moving in a magnetic field could you power your  microprocessors?

     BEADLES:  Absolutely.  That  Is  the  question.

     SIBBETT:  Almost  anything  Is possible with microelectronics.

     WOEBKENBERG:  Mary Lynn Woebkenberg, NIOSH.  I was wondering if you  had
tried the pump with  the commercially available sorbent tubes and charcoal

     SIBBETT:  If you  are talking about  the kinds of things that adsorb
gases, I did try them.  The pressure drop is too great.

     WOEBKENBERG:  What about the smaller charcoal tubes, where the drop
isn* t quite as large?

     SIBBETT:  Actually, in principle, of course, here you would like to
limit the pressure drop as much as possible through the adsorbent.  We have
done a fair amount of work for  NASA  on dosage measurements in the field.
We find that—in regard to dosage measures on chemically coated tubes—that
a simple coating, say 5 percent sodium carbonate in the case of an acidic
component, lasts almost indefinitely.  We have never really faced a case  of
the concentrations to be collected being high enough to exceed the capacity
of the tube.

     DOCKERY:  Doug Dockery, Harvard.  Did you say who is supporting this
research and what the purpose of  it  is?

     SIBBETT:  Up to this point,  nobody has really supported It.  The purpose
of it is quite clear:  to develop a  unique individual monitoring or sampling
procedure which may be of use in  pollution studies of the type that we have
talked about earlier today.

     The company I work for is  involved in a fair amount of computer modeling
studies of exposures.  This is  just  sort of a natural attachment to some  of
the other things that they do.

 Personal Monitor  Cosmetology:
An Aesthetic  Approach
     George S. Malindzak, Jr., Ph.D.
     Northeastern Ohio Universities College of Medicine
     Rootstown, Ohio
     Mary Ann Scherr
     Kent State University, Kent Ohio; and
     Parsons School of Design, New York, New York
     Currently, environmental and human health effects data are collected and
evaluated by techniques involving fixed-monitoring  stations.  The advantages
of and requirements for personal monitoring devices are being considered to
enhance the current devices to  perform the environmental and health effects
surveillance function of air quality evaluation programs.  An ideal personal
monitor system should: 1) record both physiologic and environmental para-
meters simultaneously; 2) be reasonably priced, personalized, lightweight,
portable, and suitable for use  by human subjects; and 3) contain a solid state
memory with short-term data storage and data transfer capability.  Few such
functional personal monitors exist in the desired configuration, although the
technology for development and  fabrication is well  known.  The role of per-
sonal monitors in environmental and health effects  research has been reported
earlier (1).  The purpose of this paper is to 1) place in perspective the
physiologic and health consequences of environmental pollution in terms of
the pollutant sources, 2) develop guidelines as to  how the physiologic moni-
toring process might take place; and 3) present and display prototypes of
"personalized" personal monitors, particularly for  the female segment of the
population to be studied.

     There are several  types of pollutants which  have been shown to affect
the environment as well as the health of animals  and humans.  These include
atmospheric gases, aerosols, participate matter,  pesticides, toxic substances,


 odors,  noise,  and radiation.   Gaseous pollutants such as carbon monoxide (CO),
 sulfur  dioxide (S02),  nitrogen dioxide (N02), nitric oxide (NO), peroxyacetyl
 nitrate (PAN), and hydrocarbons (HC) are products of automobile exhausts,
 power plants,  and industrial  exhaust gases due to the combustion of fossil
 fuels such as  coal and oil.   The pollutant gases are generated as the by-
 products of combustion and incomplete combustion, such as the generation of
 S0_ from oxidation reaction of sulfur in fossil fuels.  Complete combustion
 of the  fuels would produce carbon dioxide and water, which are not considered
 to be pollutants.  Carbon monoxide and hydrocarbons are products of incomplete
 combustion. Nitric oxide is  produced when the nitrogen is oxidized by the
 high temperatures of combustion.  The internal combustion engine of the auto-
 mobile is the  principal source of atmospheric NO and N0« pollution.

      The ultraviolet energy of sunlight induces the oxidation processes to
 form the photochemical oxidants.  The photo-oxidation converts NO into N0_, 0-
 into 0-, sulfur dioxide (SO^) into sulfur trioxide (SO*), and generates the
 unstable PAN compounds due to the irradiation of nitrites in air or oxygen (2),

      The formation of pollutants within the environment is the product of a
 complex atmospheric chemical  process.  The atmospheric level of each pollu-
 tant is dependent on atmospheric conditions, number and quality of combustion
 sources, and the surrounding  topography.  The conversion of photochemical
 oxidants is directly related  to the amount of available sunlight and tempera-
 ture of the polluted air mass.  The available light is not only dependent on
 cloud cover, but the amount of photochemical aerosols in a given polluted air
 mass, which are themselves by-products of the interactions of the photochemi-
 cal oxidants.   The dilution of air pollutants is affected by winds, tempera-
 ture inversions (which prevent pollutant dissipation into the atmosphere), and
 elevation or air density.  Other types of airborne pollutants include aerosol
 sprays (hair,  paint, and agricultural pesticides), industrial solvents, and
 fine particulate matter (coal dust).

      Noise sources have been  shown to be harmful above 80 decibels A-weighted
 sound level (dBA), and those  annoying to the major segment of the population
 are considered by the Environmental Protection Agency as pollution (3).  En-
 vironmental noises are produced by a wide variety of sources; for example,
 automobiles, construction equipment, transport trucks, commercial aircraft,
 and steel mills.  Noise sources may be fixed, moving, transient, or temporary,
 and may range  in sound level  from an average of below 35 dBA found in the
 suburban home, to over 110 dBA measured in and around mineral mine operations.
 Two hundred feet from a jet at takeoff, the sound level is 120 dBA (4,5,6).
 Many industrial and transportation noises are above the 90 dBA maximum 8-hour

 Occupation,  Safety,  and Health Act (OSHA)  exposure  level.   The  reduction of
 excessive  noise levels at the source,  and  providing adequate hearing  protection
 to  the  exposed  population,  is clearly  of interest to the health sciences as
 well as to the  regulatory agencies.

     Human exposure  to electromagnetic radiation is extensive when  the expo-
 sure from  industrial,  medical, and scientific devices is combined with the
 exposure to  home equipment,  radio, FM, TV  stations,  military and civilian
 communications  equipment, navigational aids, and radar.  The sources  that
 produce ionizing radiation are X-ray and radioisotope equipment, nuclear
 power plants, atomic blasts  and weapons testing, and solar  radiation.  Although
 the physical characteristics of electromagnetic and  ionizing radiation are  well
 understood,  their Impact as  environmental  pollutants is less clear.

     Environmental  pollutants  can affect the health of individuals in a
variety of ways, ranging in  severity from physiologic changes of uncertain
consequence to pathologic changes leading to disability and, sometimes,
death.  The more severe  effects, such as chronic long-term illness and
death, are manifest in a relatively small portion of the population over a
long period of time,  and they  are difficult to quantify because other
complex factors difficult to quantitate are usually associated with death.
At the other end of the  spectrum, slight physiologic and pathologic changes
may not be detectable with the degree of certainty desired because of the
unique ability of the human body to respond to and recover from most types of
environmental insults.   Adverse health effects may result from short-term or
long-term pollutant exposure.  Many factors contribute to differences in the
distribution of diseases associated with air pollution exposure, such as
age, sex, race, occupation, cigarette smoking habits, geographical location,
etc.  The health effects discussion that follows is based on appropriate
statistical adjustments  for these factors in assessing their environmental
health effect impact.

     The vast majority of environmental pollutants are airborne and gain ac-
cess to the body via  the respiratory tract (i.e., the lungs).  Depending on
the particle size of  the  pollutant, its chemical composition and activity,
the effect of the pollutant may be restricted to the respiratory tract, or
it may be absorbed via the blood stream and exert its effect indirectly on
another target organ.  A few pollutants have a direct effect on body tissue
and organs, such as,  for example, radiation and environmental contaminants


 that irritate the external  layers of the eyes.   Other pollutants appear to
 affect the brain and central nervous system and to indirectly produce changes
 in behavior,  control,  and coordination of human activity.  There are a wide
 variety of human health effect responses that are yet to be identified, be-
 cause they have not been examined thoroughly and systematically.

      Basically, when foreign matter is introduced into the human airways, the
 natural respiratory response is to expel it through the action of hairlike
 cells sweeping the material out, washing the surfaces with mucus which is
 constantly being secreted,  or by coughing the material into the mouth for ex-
 pulsion or elimination.  Environmental pollutants act to increase the secretory
 activity of the mucous cells (or inflame them), and inhibit the action of the
 special hairlike "scrubbing" cells, thus producing an increase in frequency,
 duration, and intensity of  the compensatory cough mechanism.  Irritation of
 the mucous cells causes them to swell and partially occlude the airway, making
 breathing more difficult (i.e., Increasing airway resistance).  Decreasing the
 activity of the "scrubber"  cells produces a stasis of mucus material which
 constitutes an excellent culture medium for the growth of bacteria and infec-
 tious agents within the lungs and airways.

      The site of irritation may be in the upper airways, small (lower) airways,
 or within the lung tissue itself.  If the inflammation reaches the gas ex-
 change tissue (alveoli), normal exchange of needed 0. and unloading of C0« is
 impaired.  The greater the  area of inflammation, the greater will be the re-
 sulting decrease in lung function, or impairment of gas exchange.  Chronic
 inflammation increases tissue susceptibility to infection with pathogens.
 N0_, for example, has been  shown to be related  to an increase in incidence of
 respiratory infection in humans (6).

      Basically, any irritation which produces an enlargement and inflammation
 of mucous cells within the  airways and increases secretions will decrease the
 potential air volume which  can be taken in and  expelled and will affect normal
 respiratory function.   Focal irritation of large airway channels may produce
 a substantial reduction in  the lumen at that point.  Air moving through this
 constricted area can produce turbulence which in turn can be heard through the
 chest wall as "wheezes" and "rhonchi."  The greater the luminal constriction,
 the more intense the turbulence and subsequent  sound produced for a given
 respiratory effort.

      Environmental pollutants like N02, 0-, and SO,, produce changes in respi-
 ratory airway membranes which in turn alter pulmonary function (6,7).  S02,
 for example,  is highly soluble in water (forming ^2S03 mlst^ and is usually


 absorbed  quickly in the proximal parts of the  respiratory  systems;  hence,
 principal effects are seen in the upper respiratory  tract.   NCL and 03,  on
 the other hand,  are not as soluble in water and  in the  upper respiratory tract.
 Their effect  is  seen in the smaller airways and  in the  lung  tissue  itself.
 Epidetniologic studies have shown a greater incidence  of respiratory infections
 and a decrease in pulmonary function in people living in an  environment  high
 in N02, 0^, and  S02,  compared with those living  in a  relatively pollutant-free
 environment (6).   0«  is also a powerful chemical  oxidant and recently has  been
 shown to  produce a radiomimetic (radiation-like)  effect  on peripheral blood
 cells.  Ozone has been shown to produce an increase in  chromosomal  breakage  of
 peripheral leucocytes—about 10 times that of  nonexposed subjects.

     Carbon monoxide  traverses the alveolar membrane  and combines with the
 hemoglobin in the blood to form carboxyhemoglobin.  As a consequence, hemoglo-
 bin—the  iron component of blood,  normally used to carry oxygen to  the tissues—
 is rendered less  effective.   The net result is that the  end  organ or tissue,
 dependent on  the  oxygen carried by the hemoglobin, is less efficient at  perform-
 ing its principal function.   Recent studies have  shown  that  the heart must work
 harder in order  to do the  same amount of  work  following  exposure to carbon
 monoxide  than before  carbon monoxide exposure.  In healthy young people, these
 changes go almost undetected,  due  to the  intrinsic ability of the heart  to
 compensate for the slight  compromise in oxygen-carrying  capacity.   In subjects
 with pre-existing coronary vascular disease, the  changes produced in cardiac
 function  are  more obvious  and  detrimental  to the  individual's health (8,9).

     Radiation has been shown  to produce  a  direct effect on  cellular function
 because of its ability to  penetrate surface and deep structures.  Excessive
 ambient noise  levels  have  been shown to  produce an effect on human  performance,
 presumably due to interference in  the  reception sensory  nervous signals by the
 brain from multiple pathways  (i.e.,  auditory,  optic,  tactile, etc.) during the
 performance period.

     The major adverse  health  effects  observed thus far  have been associated
with the  respiratory  system, mainly because of the respirable nature of air-
 borne environmental contaminants and,  to a  limited extent,  because of the
cardiovascular system and  central  nervous system.  It is unclear (and indeed
unknown)  to what  extent and  in what manner  environmental contaminants affect
other organ systems,  such  as the kidney, the reproductive system, and skeletal
muscles.  Eventually,  each possible health  effect should be  explored in a
 short-term and long-term exposure  schedule  in  order to obtain the most complete
 information regarding the  relationship  between environmental pollutants and
adverse human  health  effects.


      Currently,  environmental  and  health effects  data are collected and evalu-
 ated  through a combination of  techniques,  including data from fixed environ-
 mental  monitoring  stations.  Increasing  public  concern and national priorities
 have  dictated the  need  to  expand  the  current  data collection base to include
 more  detailed surveillance of  a wider range of  pollutants and a more comprehen-
 sive  evaluation  of individualized  health effects.  In this context, the develop-
 ment  of personal monitors  which measure  physiologic and pollutant events
 simultaneously are being proposed. Comprehensive personal monitor capability
 will  permit accurate monitoring of pollutant  data and the concomitant physio-
 logical response within the immediate vicinity  and physical location of the
 subject at all times.

      Personal monitors  are envisioned to serve  as a means of providing a
 more  meaningful  scientific relationship  between the physiologic and environ-
 mental  pollutant exposure  in human subjects.  More specifically, a need appears
 to exist for an  individualized "early warning health effects surveillance
 system" designed to sense  and  process physiologic responses and environmental
 pollutant data simultaneously. Ideally, a complete physiologic monitor system
 would include invasive  and noninvasive clinical evaluation techniques.  Prac-
 tically, however,  only  noninvasive techniques are being considered.

      An ideal personal  monitor should be able to  process and correlate simul-
 taneous physiologic and environmental pollutant measurements at a scientific
 laboratory level of accuracy coupled  with short voice comments from the sub-
 ject.  It is imperative that multilevel  feedback capability be available to
 document the subject's  assessment  of  discomfort relative to environmental
 conditions, such as symptom reports,  location,  and general environmental con-
 ditions.  Such feedback capability has been shown to be invaluable in the
 assessment of data collection  and  analysis programs in the past.

 Physiologic Monitoring

      With the many physiologic variables to monitor, it is unreasonable to
 consider burdening the  subject with an excessive  number of transducers and
 associated equipment which may be  cumbersome  and  physically unattractive.
 The transducer-to-anatomy  interface and  the transducer-to-pollutant interface
 must  be considered in any  final personal monitor  design.  Ideally, transducers
 could be optimally placed  on male  and female  subjects to provide for the
 measurement of multiple physiologic parameters.  For instance, a piezoelectric
 transducer designed to  respond to  acoustical  energy might also measure


 simultaneously  pressure transients,  voice, ambient noise, heart sounds, heart
 rate, and cough.   The  Incorporation  of  appropriate electrodes within  the
 piezoelectric transducer housing  can provide  the measurement of the electro-
 physiologic activity of the  heart (electrocardiogram).  From specific signal
 conditioning and  filtering  techniques of  the  complex acoustic (pressure) and
 electrophysiologlc signals,  a  variety of  additional relationships may be derived
 which are not otherwise available from  a  single transducer.  This design pro-
 vides the basis of an  "integrated transducer" which would have the capa-
 bility of recording more than  one phenomenon  from a single transducer at a
 single anatomic site.

     1)  Respiration-Pulmonary.   Pollutants may cause irritation of the mucous
 cells which line  the respiratory  tract, resulting in the reflex production of
 a cough.  Increased mucosal  irritation, in turn, may cause narrowing of airways,
 resulting in respiratory wheezing produced from airway turbulence.  Cough events
 may be detected and quantltated from the  placement of a piezoelectric trans-
 ducer over the  lower abdominal wall  to  register cough diaphragmatic pressure.
 Additional respiratory  parameters, such as the ventilation rate (breaths per
 minute) and indirect respiratory  volume,  may be detected by changes in thorac-
 ic geometry by  electrical Impedance  or  pneumographlc techniques.

     Transducer-to-anatomy interface differences that exist in males and
 females may be  circumvented  by installing the appropriate transducer within
 the structure of  conventional garment wear.  For example, in female subjects,
 the transducer  for heart sounds may  be  placed in the base elastic strap of the
 bra over the anatomic apex of the heart;  for wheeze registration, the trans-
 ducer may be placed within the rear  bra strap area of the shoulder blades;
 coughs may be recorded  from  a transducer  placed within the abdominal wall
 panel of panty  hose.

     It is well known that some people are "chest breathers," while others
are "abdominal  breathers."   In either case, the respiration rate transducer
might be a small  pneumatic tube connected to a semiconductor transducer.  For
 "chest breathers,"  the  transducer might be Incorporated in the base elastic
 strap of the bra  (for female subjects);  for "abdominal breathers," the trans-
ducer might be  incorporated  in the abdominal panel of panty hose.

     2)  Cardiovascular-Heart.  The  personal monitor system should provide
 the capability  for sensing and characterizing the electrophysiologic and
hemodynamic events of the vascular system and heart.  Typically, electro-
physiologic information is obtained  through the electrocardiogram (EGG), and
hemodynamic information is usually obtained from the arterial blood pressure

      The  ECG measurements may be made with as few as a single pair of bipolar
 surface electrodes,  or with as many as 12 electrode combinations, depending on
 the  thoroughness of  the electrocardiographic examination.  The ECG waveform is
 characterized by the P-QRS and T features which correspond to specific electro-
 physiological events throughout the cardiac cycle.  From this electrical infor-
 mation, basic heart  rhythm (heart rate),  and the position of the S-T segment of
 the  waveform with respect to the isoelectric baseline, are important determi-
 nants.  Often, abnormal rhythms (arrhythmias) and shifts in the S-T segment of
 the  waveform occur under a variety of conditions, particularly when oxygen
 (carried  by the blood) is unavailable to  the contracting heart.  Recent studies
 (8,9) have  shown that exposure of subjects to 100 ppm (parts-per-million)
 carbon monoxide (which decreases effective oxygen delivery to heart tissue)
 produces  a  detectable shift in the S-T segment of the ECG during exercise in
 normal healthy subjects.

      Quantitation of the cardiac rhythm and the degree of S-T segment devia-
 tion from the baseline may be accomplished with a special-purpose S-T segment
 microprocessor integrated within the personal monitor system.  Specific features
 such as S-T segment  duration, S-T segment amplitude (positive and negative),
 and  rate  of change (first derivative) of  the S-T segment are most important
 and  significant in assessing specific adverse cardiac health effects in response
 to pollutant exposure*

      The  measurement of blood pressure may be accomplished by affixing a
 piezoelectric transducer to the radial artery of the arm in such a way as to
 detect maximum (systolic)/mlnimum (diastolic) arterial pressures, pulse
 pressure  (systolic minus diastolic), and  heart rate.

      Valuable cardiac performance information may be derived from the directly
 measured  cardiovascular parameters.  For  example, heart rate from the ECG
 through a conventional cardiotachometer,  and systolic time intervals (ST1)
 from the  combination of the phonocardiogram, the electrocardiogram, and blood
 pressure.  STI's provide information relative to the rate and duration at
 which the heart is contracting and relaxing.  The triple product combination
 (heart  rate, mean blood pressure, and ejection time of the heart) serves as an
 index of  the amount  of oxygen the heart is consuming during the process of

      3)   Central Nervous System.   The central nervous system activity (electro-
 encephalogram, EEC)  is one of the most difficult signals to interpret in that
 responses are often  subtle and difficult  to quantitate.  Detection of behavioral

 patterns  associated with changes in the  EEC or  the electro-oculographic  poten-
 tials  (EOG)  may be accomplished, but not without carefully planned and executed
 experimental programs.

     Central auditory responses are related to  the temporal lobe of the  brain;
 recording of the EEC over the temporal area of  the skull may provide an  index
 of auditory  time discrimination judgments  by a  given subject.  Recent studies
 suggest that auditory time discrimination  shows a decrement after exposure to
 100 ppm of carbon monoxide for 4 hours.

     The  human  eye is (as is  the lung) directly exposed to environmental con-
 taminants.   Eye irritation can produce increased frequency of eye blinking,
 fatigue of ocular musculature,  and  pain  sensation.  Eye blinking activity
 may be monitored with silver  chloride electrodes placed laterally and slightly
 posteriorly  to  the eyes.   Appropriate signal conditioning of the EOG signal
 could provide a reliable  electrical representation of an eye blink, blink
 rate, and duration.   Additional conditioning of this electric signal (such as
 the first derivative) may provide information of ocular fatigue relative
 to normal and abnormal environmental pollutant exposures.

     4)   Other.   Tracking the location of  the subject with a personal monitor
within a  given  geographic area  will provide an additional means of referencing
and characterizing pollutants in a  given area at any time of day, in contrast
to a fixed-station operation.   Tracking may be accomplished as a function of
verbal or coded  location  reports  into the  system memory from the subject.  The
monitoring of radiation,  for  example, might be coupled with location and
radiation film  badge  exposure levels through a similar process.

Data Acquisition System

     The  data acquisition system  (DAS) design for personal monitors may be
implemented  in a  variety  of ways, any one of which appears to require state-
of-the-art techniques.  Physiologic transducers and air pollutant transducer
signals are  different in  nature; hence, a properly designed DAS must take these
differences  into  consideration.  Most physiologic responses are dynamic,
rapidly changing,  and almost  periodic functions of time.  Air pollutant levels
are usually  static  or change  slowly In time.  Therefore, the dynamic frequency
response  of  each  transducer is an important consideration.

     All  transducers are  subject  to various forms of  artifact (physiologic,
60 Hertz, movement artifacts, pollutants, interference from other contaminants,
etc.).  The  judicious rejection of  these artifacts is an important factor for


 reliable DAS design.  To ensure maximum DAS  stability  and  reliability,  it is
 important  to select a physiologic  event as a time  reference with which  to
 sense and  reject any artifact  from the system signals.   For example,  in
 detecting  a cough, a microphone or another appropriate  acoustic  transducer
 may  be  employed.  Coughing  is  an acoustical  signal produced during  the  expira-
 tory phase of  the respiratory  cycle.  Any acoustical signal sensed  by the cough
 transducer during inspiration  (such as talking,  ambient noise, etc.)  must be
 rejected.  However, during  expiration, the expiratory  gate would open for any
 possible cough data that might occur.

     All data  collected on  board the personal monitor  might be stored in
 digital form in solid state memories for subsequent playback, processing, and
 display.   The  system should have the capability  of recording data for 24-hour
 periods, after which the data  could be read  out  into a long-term storage device
 for  inspection and analysis.   The  solid state memory readout could  be accom-
 plished by a variety of ways;  i.e., dataphone interface to local or regional
 fixed stations.

     A  complete system design  must consider  playback,  display, and  processing
 functions, which in turn are determined by the nature  in which the  data are
 recorded.  The system design should consider options to perform  a wide  variety
 of information processing procedures, such as signal averaging,  digital filter-
 ing, tread analysis, and interactive graphics.  Ideally, provisions to  process
 the  personal monitor information should include  on-line and off-line  capabili-
 ties.   This capability is currently available with low-cost, high-efficiency

      Personal monitors are  envisioned  to  provide more  reliable  physiologic
 and environmental  information on human pollutant exposures  beyond  the  range
 and capability of  community health surveillance stations.   Specifically,  a
 need appears to exist for individualized  surveillance  systems to sense and
 process physiologic  and environmental  pollutant data simultaneously.   Devices
 have been developed  to satisfactorily  detect  environmental  contaminants and
 their physiologic  consequences.   Fabricating  these  detection devices to
 ensure user cooperation has not  been a simple task.  Compliance and user
 resistance problems  are particularly sensitive when dealing with the female
 segment of the population to be  studied.

     An attempt  was made approximately 9 years ago  to incorporate the  (then)
 state-of-the-art technology related to turning simple physiological and en-
 vironmental monitors into devices that were  portable, lightweight, and
 reasonably priced.   Pilot feasibility studies were  conducted to evaluate  the
 effectiveness and practicality of personalized personal monitors, particularly
 those which female  subjects would be required to carry or wear.  The remainder
 of this presentation will deal with the results of  the feasibility studies, in
 which specific pollutant and physiologic monitoring devices were incorporated
 with specific pieces of  jewelry which are not only  scientifically functional,
 but also attractive.

     1)  Oxygen  Pendant  (Figures 1  and 2).  A photo cell mounted in the pend-
 ant senses the continuous samples of ambient air.  When the level of environ-
 mental contamination exceeds a preset threshold, an audio device is triggered
 electronically which alerts the subject to the presence of pollution.  The
 pendant also contains a  mask and a  10-minute supply of oxygen.   This device
 may be especially useful to subjects with a pre-existing respiratory disability
 such as asthma.   In these cases,  the oxygen canister might also contain a
 bronchodilator to relieve the distress  and discomfort associated with an
 asthma attack.

     2)  Heart/Pulse Sensor Bracelet (Figures 3 and 4).   A piezoelectric
 transducer relays the peripheral  arterial (radial) pulse to circuitry which
 activates a light emitting  diode display device.  Heart rates above the preset
 threshold for an individual enables  an  oscillator to emit an audible frequency,
 the output of which is connected  to  an  earphone speaker to produce a warning
 sound which alerts  the subject  to check his/her activity level, take appro-
 priate medication,  or seek  professional assistance.

     3)  Body/Air Sensor Belt (Figures  5 and 6).  This belt is  made of stain-
 less steel and liquid crystals  designed to monitor toxic gases, ultraviolet
 radiation, and air  and body temperature.  Increasing doses of X-ray radiation
 progressively lower the  color play range of cholesteric-phase liquid crystal
materials.  The  effect is enhanced when an effective amount of  an iodine-
 containing compound is used  in  the liquid crystal material.  Novel iodine-
 containing compounds in  the  appropriate proportions give direct indications
 of the dose of X-ray radiation  that  has been received.

     A layer of  water-soluble plastic material is first  deposited on a limb
 or torso to serve as a barrier  layer.  A cholesteric layer is placed over the
barrier layer.   A chromatic  display  of  temperature of the limb  or torso will

 FIGURE 1.  Oxygen pendant (normal
 closed position).
FIGURE 2.  Oxygen pendant (open
position with mask and canister
 FIGURE 3.  Heart/pulse bracelet in
 place on hand.

FIGURE 4.  Heart/pulse bracelet
with electronics and compartments

      FIGURE  5.   Body/air sensor belt.
6.   Body/air sensor belt and arm  bracelet.

 be visible to the naked eye,  including minute variations.   Vasoconstriction
 in the extremities in response to smoking and the vasospastic effects of
 peripheral vascular disease may be clearly observed.

      4)  Liquid Crystal Heart Beat Display (EGG) (Figures  7,  8. and 9).
 A monitoring electrode is attached to the body to transmit the EGG through
 electronic circuitry to activate a liquid crystal display  of  the subject's
 electrophysiologic heart pattern.  Abnormalities in either the rhythm (or
 rate) or the waveform will be manifest in an aberrant chromatic pattern.

      5)  Tracheotomy Necklace (Figure 10).  An adorned standard medical device
 fits into an opening, which is surgically performed on the trachea, to permit
 ventilation.  The tracheotomy device is available in sterling silver and
 silicone plastic through conventional medical supply sources.  The tracheotomy
 necklace is constructed of gold, silver, gemstones, chain, and other materials
 which may change the clinical appearance of the equipment, permitting it to
 be viewed as a necklace rather than as a prosthetic device.

      Each of these devices was designed utilizing the technology of approximately
 5 years ago.  Each device is large and obvious to attract  attention.  Each
 device was (and can be) designed to suit the individual's  taste and physiog-
 nomy.  Since the time of the original design and feasibility  studies, further
 developments in this area have nearly ceased due to the lack  of appropriate
 interest and support.  However, the needed facilities and  interest to continue
 with these developments still exist.  Current technology and  fabrication tech-
 niques could substantially reduce the size as well as the  weight and cost of
 each of these devices.

           The Northeastern Ohio Universities College of Medicine is part of a
 consortium of three universities—Kent State University, the  University of
 Akron, and Youngstown State University.  The College of Medicine serves as the
 base from which the basic medical science studies are conducted.  Kent State
 University has a Liquid Crystal Institute at which further and continued
 studies with the liquid crystals may continue.  The University of Akron has
 an excellent School of Engineering, within which exists a  competent Electrical
 Engineering Department capable of developing and implementing any micro-
 circuitry and microprocessor-based system required for further development in
 this area.

FIGURE 7.  Liquid crystal heart beat
display necklace.
FIGURE 8.  Liquid crystal heart beat
display necklace (electronics open
and exposed).
FIGURE 9.  Liquid crystal heart beat
display necklace (electrode placement
and electronics exposed).
FIGURE 10.  Tracheotomy necklace.


     This work was  supported  in  part  by  the Northeastern Ohio  Universities
College of Medicine Intramural and  Biomedical Research  Development Funds and
by  the American Heart Association.  Body monitor  and  jewelry prototypes and
research efforts have been  predominantly funded by  the  personal  resources of
Mary Ann Scherr, who is also  a consultant  designer  for  jewelry and a metal-
smith in New York City.
 1.    Petrovick, M.L., Malindzak,  G.S.,  Jr.,  Strong,  A.A.,  Burton,  R.M.   The
      Role  of  Personal Monitors  in Environmental  and  Health Effects.   Proceed-
      ings,  National  Aerospace Electronics  Conference (NAECON),  pp. 152-159,
      May 1974.

 2.    U.S.  Department of  Health, Education, and Welfare.  Air Quality Criteria
      for Photochemical Oxidants.   National Air Pollution Control Administration
      Publication No. AP-63.  Washington, D.C., U.S.  Government  Printing Office.
      March 1970.

 3.    United States Clean Air Act  (42  U.S.C.  1857 et  seg.)  includes the  Clean
      Air Act  of 1963 (PL 88-206)  and  Amendments  made by the Motor  Vehicle
      Air Pollution Control Act  -  PL 89-272,  October  20,  1965, the  Clean Air
      Act Amendments  of 1966 - PL  89-675 (October 15,  1966), the Air  Quality
      Act of 1967 - PL 90-148 (November  21, 1967), and the  Clean Air  Amendments
      of 1970  - PL 91-604 (December 31,  1970).

 4.    Botsford, J.H., Cudworth,  A.L.   Fifth Institute on Noise Control Engineer-
      ing.   Institute on  Noise Control Engineering, Bethlehem, Pa., August 1972.

 5.    Peterson, A.P.G., Gross, E.E.  Handbook of  Noise Measurements,  Sixth
      Edition. West  Concord, Mass., General  Radio Company, 1967.

 6.    Shy,  C.M., Finklea, J.F.   Air Pollution Affects Community  Health.
      Environ  Sci Technol 7(3):204-208,  1973.

 7.    Cohen, C.A., Hudson, A.R., Clausen, J.L., Knelson, J.H.  Respiratory
      Symptoms, Spirometry and Oxidant Air Pollution  in Nonsmoking  Adults.
      American Review of  Respiratory Disease  105(2),  1972.

 8.    Anderson, E.W., Andelman,  R.V.,  Strauch, J.M.,  Fortuin,  N.J., Knelson,
      J.H.   Effects of Low-Level Carbon  Monoxide  Exposure Onset  and Duration
      of Angina Pectoris. Annals  of Internal Medicine 79:46-50, 1973.

 9.    Haak, E.D., Jr.  CO and Cardiac  Performance. In:  Clinical Implications
      of Air Pollution Research  (Finkel, A.J., Duel,  W.C., eds.).  Publishing
      Science  Group,  Inc., 1976, pp. 113-117.



 George S.  Mallndzak, Jr., Ph.D.
 Department of Physiology
 Northeastern Ohio Universities
 College of Medicine
 Rootstown,  Ohio  44272

 Mary Ann Scherr
 School of  Art
 Kent State University
 Kent,  Ohio  44240
     PETROVICK:  Matt Petrovick, EPA.  Dr. Malindzak and Ms. Scherr, I would
 really like  to commend  you  for  a very unusual and very well designed approach
 to looking at physiologic measurements in the female.  The female happens
 to be one of the most Ignored subjects in terms of science and in terms of
 data acquisition.

     I believe your  design  approach is going to have a significant impact
 in placing our health effects data base into balance, which presently I don't
 think it is.

     WALLACE:  Who supported this development?

     SCHERR:  This work?  I do.

     MALINDZAK:  Totally and completely.

     SCHERR:  I am an Associate Professor of Art at Kent State University.
My interest  in this  subject started with seeing a woman who was very disturbed-
cosmetically disturbed—with the hole in her throat, and who wore many scarves
to mask herself.  And I managed to see in my mind in one brief moment this
piece of equipment,  and I asked her if she would like me to design something
that would be a little bit more interesting.  And I did.  And that set the
stage for another whole wave of research.

     I think that maybe I am more or less a scientist, although I could never
compete in this area.  I am a reactor.  And in reacting to this lady, along
with the television  programs of space technology and watching the space pro-
gram, I decided at this time to get involved in programs where maybe we
could monitor ourselves.  And that was my own reaction to it.

     I started then  in all of these areas,  trying to find devices that would
monitor anything.  So, this is why is all started.

 "Shirtsleeve Workshop"
 An informal session designed to arrive at a consensus on present abilities of
 commercially available instrumentation and promising avenues for future research
     DR. DAVID MAGE:  This evening we would like to discuss informally some
of the aspects of personal monitoring that we have gone over more  formally
during the meeting.

     We have heard several papers about personal monitors for exposure
and for physiological variables.  Some of us have been trying to act  as a
kind of marriage broker to see if we could merge the two to fit into  the same
package—we really see that as a necessity.  We would like to have some dis-
cussion as to the needs that the different Government agencies have in this
area.  We will have people from NIOSH, EPA, and other agencies speak  on what
they perceive as being the needs for their agencies.

     We would like to make it apparent to the people in industry as to where
their markets might be, and we want to stimulate competition to produce the
devices that we need to do our job.

     The bottom line, I guess, is that,  as Congressman George Brown said
in his keynote address, we have to show that our approaches to air quality
standards are reasonable, let alone correct.  In order to do that,  we  have
to be able to demonstrate that people are being exposed and people are having
health effects from that exposure.  It is one thing  to put someone in  an
environmental chamber and give him 0.4 ppm ozone while walking on  a treadmill
and then document that he has had some kind of physiological response.   But
then somebody can ask us, "Well,  how many people on  treadmills in  Los  Angeles
are breathing 0.4 ppm ozone," and, you know, we can't say 1,  10, 100,  or 1,000,

It becomes very difficult to quantify.  As soon as it becomes difficult  to
quantify, then it can also become very difficult to establish that we  are
being reasonable.  Can we state beyond the shadow of a doubt that there  are
100, or 1,000, or 10,000 people who are having health effects?

     If industries talk about their cost of pollution control,  they  talk in
units of billions of dollars.  Can we justify this cost  on  the  grounds that
there is a benefit to the public in terms of billions of dollars?

     What I would like to do now is turn this portion of the meeting over to
Dr. Lance Wallace for some informal remarks.  Some of the things we  would like
to do is discuss with you some of the needs that we perceive for developments
of personal monitors.  We would like people to let us know  some of the prob-
lems that they perceive; for example, that perhaps we haven1t been funding
you when you  have been submitting proposals to develop those monitors  that
we would like to have.

     I would  like to get the discussion out informally so that  we can  have
some good interchanges among us.

     DR. LANCE WALLACE:  I feel that my horizons have been  opened up quite
a bit by this meeting.  I came here thinking of personal monitors basically
as being chemical and physical sensors, and by the end of 2 days, I  have
realized that there are physiological measurements that  can be  made.  There
are  the cosmetological aspects.  There are the microcomputer aspects.  It
appears that  we are on the verge of something considerably  larger than I had

     So, with that as a bit of an apologia, I will mention  that what I have
done for the  past 2 weeks before coming here was to ask  the different  offices
within EPA what chemical and physical measurement devices they  felt  were nec-
essary and what pollutants they felt were Important to develop  personal
air quality monitors for.  As I mentioned in my talk, the Office of  Research
and Development acts as a service organization to some of the other  program
offices within EPA.  They express their needs for research, and we try to
accommodate them.

     I spoke  to several of the offices that are the most involved with this
topic, including the Office of Air, Noise, and Radiation, which includes
the Office of Air Quality Planning and Standards and the Office of Mobile
Sources.  These offices have somewhat different areas of responsibility, and
different needs for personal monitors.  I have obtained  from the people  in

 those offices  lists of the pollutants for which they  see  a  high  priority
 for  developing personal monitors*

     I have also spoken to the Office of Toxic  Substances and  received  the
 same sort  of list from them,  and to the other offices within the Office of
 Research and Development and  have  received their nominations for high priority

     But as a  sort of  straw man, it might be desirable for  us  to look at  the
 lists that we  have come up with and using those lists to  investigate the
 possibilities  for using monitors that are already available, that are al-
 ready out  there and can do ambient level measurements, to determine areas in
 which further  research is needed.

     I would like to pass out these lists,  and  I will read  them  over with you.

     Within the Air Office, the Office of Air Quality Planning and Standards'
 staff members  were mostly concerned with the criteria pollutants.  The  pol-
 lutants that they found to be most important were carbon monoxide and nitrogen
 dioxide.   They desired that field  studies be done as  soon as possible for
 these two  pollutants if there were any personal monitors available that were
 capable of making ambient level measurements.

     From what I  have  heard in the past 2 days, it seems very possible  that
 we do have personal monitors  that  are capable of being used in field studies—
 the ESI monitor and the GE monitor,  both for CO.  Dr. Palmes'  NO, device  also
 seems possible.

     Perhaps before 1  go any  further,  I could ask if  there are comments about
 just these two  pollutants as  to other devices that appear as possibilities
 for immediate  use in field studies?   Are there any comments on the Energetics
 Science's Ecolyzer 9000 series and the GE device that we have?

     DR. MANNY SHAW:   I am Manny Shaw of  InterScan.  We aren' t ready right
 now to give you a dosimeter.   We are tooling up on a different type of dosimeter
 that will be of  interest.  And when I  say different "type," I don't mean
 necessarily a  ppm-hour dosimeter.   I mean a dosimeter which will alarm at the
TLV.  1 would  say that within 6 months  to a year we will be able to field
 test the InterScan TLV alarm  dosimeter.

     DR. DONALD  SIBBETT:   Donald Sibbett, Geomet.  I feel that with a short
validation program,  our personal monitor  could be used with NO,,.   It is

 a resplrable pump driven by the abdomen.  It can be used, I think,  rather
 readily with N02«

      DR. JOHN BACHMAN:  One thing that would be of use as far as we  are  con-
 cerned is if some studies could be done on these two pollutants—CO  and  N0_—
 at the same time.  We want these data for two reasons:  one is to help us  set
 better air quality standards for these pollutants, and the second is to  do
 a better job of assessing total exposure.

      One of the reasons those two pollutants had high priority among the
 people in our office was that the concentrations of these pollutants tend
 to vary greatly over an urban area relative to certain other pollutants
 because of the nature of the emissions and the formation.
      NO- happens to be a secondary pollutant, as everyone knows.  CO  is  a
 primary pollutant.  But they vary with the automobile.  It would  be of inter-
 est to us to do both of these studies at the same time.

      DR. FRANCIS BERLANDI:  Francis Berlandi from ESA.  Looking at the phi-
 losophy of the development of the pollutants of interest, I notice that  all
 the pollutants are associated with air quality standards.  And noticeably,
 there is one pollutant here for which there is an air quality standard,  and
 yet there is no interest in seeing that a personnel monitor is available for

      In particular, I mean lead.  In the future we are  talking about  arsenic
 and cadmium.  I was curious as to whether people didn't feel that there  was
 a monitor capable of making these measurements and that is why it is  omitted,
 or if nobody is really interested in the variety of sources that  are  available
 to the person in his or her daily environment where he  can actually uptake

      BACHMAN:  I think we feel that saying fine particulates is our shorthand
 for saying anything you can measure on fine particulate filters.  But in the
 specific case of lead, right now we feel that the best  personal monitor  is
 blood lead.  There are other indicators.  Because of the total body burden,
 we have to measure the various pathways.

      We felt first of all that a system that was capable of measuring for
 fine particulates ought to be able to do lead.  Second, right now we  have some
 fair indication of total body burden from the human body.

     Concerning  the other pollutants,  by  the way, as Lance will  go  on  to
talk about, vapor  phase organlcs  and  some of the fine  particulates  cover  a
whole host of  things which are not related to  air quality standards, but  are
related to possible hazardous air pollutant standards.

     BERLANDI:   The reason I  bring up  the point about  lead is  that  yes, I
understand that  the body burden is a good indicator of lead exposure.  But it
hardly provides  sufficient data for the state  to implement plans  to control
the lead.  We  really don't know where  individuals are  picking  it  up in their
daily lives, whether it be in the home, whether the kids are getting it on
the streets or empty lots, or whether  they are breathing it in, as  opposed
to secondary entrained  dust.

     There could be some very costly strategies implemented whl«.h could back-
fire in the long run.   This is why I mentioned the fact that I think the
technology is  available,  and  maybe that formally it should be  put down here.

     BACHMAN:  Again, from the program office  point of view, talking to the
people who set the lead air quality standard,  they didn1 t feel it would be
particularly useful in  that application at this point.

     DR. EUGENE  MEIER:   Gene  Meier, EPA,  Las Vegas.  Another point  is that
when our request went out,  we were referring to immediate programs.  "If
we had a monitor available, which one would you want to use now?"  And some
priority had to  come back.  These are  the first-thought priorities.  These
are where we actually have programs now and where we feel we need monitors.

     Your point  is well taken.  If we had the programs going on or  the people
available, we  would probably  be working in those areas.  But right now, the
other compounds  listed  are those  which have first priority.

     MR. PAUL  KIZAKEVICH:   Paul Kizakevich from Research Triangle Institute.
Admittedly I don't have much  of a background regarding the physical monitoring
parameters and how they are measured.  But in the sense of the physiological
measurements that  we have  talked  about, it is expected that the changes we
would like to  look at would either be transitory or perhaps accumulated over,
say, an 8-hour period,  over a workday.

     My question is;  Are  there any physical parameters that can  be projected
to be measured and assessed so they can be compared to such dynamic studies
as a physiological response?   The accumulation of 24 hours of  the pollutant

 is not going  to give  very much  information  in relation to a person's change
 in respiration rate or  something  else  during  his  time  at work.

      WALLACE:  At  least for  the CO,  it did  not depend  on a 24-hour average—
 we had a  continuous readout.  The N0_  is  another  matter.

      KIZAKEVICH:   Is  that  the only parameter  that can  currently be measured in
 that sense?

      WALLACE:  In  a continuous  way?

      KIZAKEVICH:   In  a  relatively continuous  or dynamic fashion.

      11AGE:   Well,  I think  that  of all  of  the  criteria  pollutants, ozone
 needs  to  be monitored continuously.  Our  air  quality standard may be changed
 in  the future to another averaging time,  but  it would  still have to be
 monitored continuously.

      KIZAKEVICH:   You are  not  talking  about the same thing, now.  If you put
 a monitor on an  individual  over an 8-hour workday and  you expect to indicate
 a health  effect  perhaps by  the  severe  pollutants  in the working environment,
 are  there any physical  measurements to compare the physiological measurements
 with at all,  other than to  say  that the guy went  to work for 8  hours, and as
 a result  during  that  work  of 8  hours he just  changes his physiology?

      WALLACE: You mean rather  than CO?

      KIZAKEVICH:   Anything.  I  don't care what it is.

      DR.  EUGENE  SCHEIDE: Gene  Scheide, Environmetics.  Can you include phys-
 iological measurements  with pollutant  measurements on  the same  scheme?  Or
 are  you comparing  apples and oranges?   But  is he  saying should  we also be
 monitoring other  physiological  responses  as well  as physical parameters—is
 that correct?

      KIZAKEVICH:   Well, gathering pollutants  over a 24-hour period is certain-
 ly  not the same as measuring the  distribution across the population, because
 it depends on whether the  person  moves.  I  am interested in knowing which
 parameters can be  measured  without resorting  to gas samples over a 24-hour
 period.   That would provide no  information  on health effects whatsoever other
 than ambient  levels.

     MEIER:   It  really depends on the objective.   In most cases,  the  objectives
are  to profile the  pollutant of interest,  because  we already have established
that there is a  health effect.   Now we are trying  to find out what its  con-
centration is in various  locations in the  environment so that we  can  determine
whether our standards  are proper and our stations  are located in  the  proper

     In addition to that, information provided by  personal monitors can give
us a better idea of where to locate a limited number of stations  in our sam-
pling network.   If  management comes in and says, "You want 10 stations.
We are only going to give you 2,"  where do we best put those 2 monitors
that we have  to  get the most effective data to rate the health effects?

     MR. MATHEW  PETROVICK:   Matt Petrovick, EPA.  That is a rathpr interesting
comment you made.   I wonder  if  you might enlighten me on which specific known
health effects you  are alluding to?

     MEIER:   In  terras  of  what?   The standards?

     PETROVICK:   In terms of the standards  and/or any pollutant?

     WALLACE:  Well, we have six or seven  criteria pollutants now.  The
criteria documents  point  out the levels at  which health effects have been
noted for those  pollutants.

     PETROVICK:   That  is  my  question.  Which health effects and which associ-
ated pollutants?

     WALLACE:  One  for each  of  these  six—the six criteria pollutants.  It is
a different health  effect in each  case.

     PETROVICK:   Is  it a  lung  function effect, or what?

     BACHMAN:  It runs from  asthmatic attack, bronchitis, general heart and
lung disease, excess mortality,  eye  irritations.  There are all sorts of ef-
fects listed, and each pollutant has  its own—CO has its own special effects.
Each one has  a number  of  effects,  and these are documented.   This is prob-
ably not the  proper  place to go into documentation of any specific health
effects for each of  the pollutants.

     But I think that  the statement made about our personal monitor needs
was put well, at  least  for CO and  N02.  What we were asking was, having estab-


 lished  standards  already,  an initial reason to do this monitoring is to do
 a  better  job  of assessing  the exposure.   And that will affect in turn the
 standards that are  processed.

      Meanwhile, a second  reason is to use these things later in health effect
 studies.   And therefore, you might want  to measure the physiological para-
 meters.   But  when we  were  asked what we  wanted right away, we didn't speak
 to that  issue.

      WALLACE: That is not to say that the other effects are known at certain
 standards. We have a standard.  We have documents that were written back in
 1970.  With the knowledge  that was current then, there was a standard set.

      So,  for  those  six, one can do pure  exposure studies and determine the
 frequency distribution of  exposures for  a given population and find out what
 percent  are,  say, exposed  to levels above what those criteria documents found
 to be important.  That is  one aspect.

      The  other aspect is  for those pollutants for which you do not have stand-
 ards.  One is obviously not going to be  able to stop at pure exposure studies.
 And then, your question about the concurrent measurement of physiological
 parameters becomes  more important.

      PETROVICK:   One last  comment.  1 have a very strong feeling that of the
 effects  that  have been identified with the associated pollutants, most of
 those relationships probably have not been measured in a dynamic sense,
 such as  Mr. Kizakevich was alluding to.

      The  issue is how the  effect was determined and relative to what pollutant,
 and was  it a  dynamic measurement or was  it a static one?  I have a strong
 feeling that  the  effect is probably defined on the basis of an interview with
 a  group of panelists or subjects or through medical forms, rather than from a
 dynamic  active measurement right off a person in the mobile environmental

      MEIER:  That is true.  But you don* t go out and kill someone to find out
 what the  LD 50 is.   You go collect the data that you have available and make
 your best judgment  based  on those data.   Then you go out with your personal
 monitor to find out if indeed people are being exposed to those concentrations.
 And if so, then you go back for further  studies, as you might want to do in
 your case, and look at the physiological factors as well as the concentrations.
 Then once you get a strong correlation,  you are justified in using the other


type of monitor.  But  right  now,  the  only  physical  fact that you know is  the
concentration level  associated  with a health effect.

     WALLACE:  Maybe I could suggest  this:  we do have a physiological con-
tingent of sorts here.   If they could get  together  and make some specific
recommendations as to  the pollutants  for which physiological measurements
would be best suited to be taken  concurrently with  the measurements of the
concentration involved, that that could be an input to our workshop tonight,
and could be reported  on tomorrow in  a panel discussion by the chairman or
some representative  of the group  who  has expertise  in taking physiological

     Is that a possibility?

     BACHMAN:  Lance,  wouldn't  it be  more  productive rather than saying for
which pollutants, to first of all define the purpose of what we are doing?
I presume it would be  to learn  more about  health effects, not just to catalog
exposure.  And that  means you are doing an epidemiological study of some sort,
unless you are talking about clinical studies, which is another breed of
cattle altogether.

     In that situation,  I think we have some people here who also know what
kinds of general, almost nonspecific,  kinds of respiratory symptoms and other
symptoms people experience from air pollution.  Also, let's talk about what
kinds of techniques  and physiological measurements  would be useful on a con-
tinuous and on a personal basis in that regard.

     In a general sense, we  are getting guidance on what is the state-of-
the-art in conducting  epidemiological studies with  respect to physiological
measurements on a personal basis; something that is a little more general
rather just related  to  CO or any  other single pollutant.

     WALLACE:  Let me  go on  to  discuss the entire list.

     Another section of OAQPS mentioned two noncriteria pollutants as being
high priority—vapor phase organics and fine particulates.

     But in the case of  the  fine  particulates, the OAQPS added the reservation
that this would be important only if  the health effects groups felt that con-
current epidemiological  studies could be conducted.  I did speak to the
representatives of the  health effects  group, and they thought that a personal
monitor for fine particulates would be important and useful.


      For  the  fine  particulates,  as  you know,  we  have had some discussion as
 to  the  proper size cutoff  for  fine  particulates.  For reasons that I do not
 fully understand myself, there are  cutoffs  that  have been chosen at 2.5 and
 15  microns.

      BACHMAN:  The reason  for  that  is because we are going to revise all of
 the air quality standards  by 1980,  including  the particulate standard.  And
 in  doing  that, all of  us knew  that  we should  look at particle size instead
 of  just garbage particles  again—perhaps including chemical composition.

      In looking at the array of  definitions of respirable particles and fine
 particles, we found so many different definitions that we felt that another
 hard look at  it was in order,  and in protecting  the public health of even
 the most  sensitive segments of the  population—given that that is our mandate—
 we  asked  our  health effects group to take a look at what really is a respi-
 rable particle.

      They put together a team, which came back with its findings.  For ex-
 ample,  if we  want  to protect even the mouth-breathing asthmatic, we should
 look at particles  less than 15 micrometers.

      The  2.5  micron cutoff comes from the apparently natural distribution of
 particles in  the atmosphere, where  certain  particles that are formed from gases
 or  condensates fall into a range less than  2.5 microns in diameter, but
 those particles that are kicked  off the ground or are mechanically generated
 are greater than 2.5 micrometers.   That is  one basis for the 2.5.

      A  second basis, which is  a  little more tenuous, is that is seems to be
 an  area where most subjects tested  fall into  a similar pattern.

      This is  an important  definition as far as we are concerned.  EPA is set-
 ting up a network  of some  300  stationary samplers for inhalable particulates.
 It  is a strong movement on our part toward  measuring the impact of respi-
 rable particles on ambient air quality.

      So,  for  people who are making  these instruments and are interested in
 future  epidemiological studies,  we  are looking at these numbers.

      The  second factor is,  as  I  mentioned before, the chemical composition.
 We  feel that  it is important in  doing these kinds of studies to look at the
 chemical  composition,  as well, in terms of  epidemiological studies.  So, what-
 ever system is chosen  should be  amenable to some of the kinds of techniques


 that are becoming routine,  such as X-ray fluorescence  to do  elemental analysis,
 sulfates, or  other important components of  pollutants.

     WALLACE:   It does appear,  just on my understanding from the  first  2  days
 of this symposium—and I  know we have our organics contingent coming  in to-
 morrow—that  in these two cases of the vapor  phase organics  and the fine
 participates, we do not presently have a working personal monitor capable
 of doing ambient level measurements.

     MAGE:  What averaging  time are you talking about?  Twenty-four hours?

     WALLACE:   I think I  would  even be happy  with a week averaging time if
 that were necessary.

     BACHMAN:   There  are  two different needs.  Doing epidemiological  studies,
 it depends on the nature  of the study that  you are talking about, a 1-hour
 averaging time  if you are looking for acute effects or a longer term.  But
 with the organics,  we are talking in  reality  about carcinogens right  now.
 That is what we  care  about,  at  parts-per-billion and maybe even lower con-
 centrations.  We have a list of hot numbers right now which we are looking at.

     So, there  are some specific compounds  that we are interested in  in terms
 of only semiquantitative  analysis to  start  with, long-term kind of exposures
 and the lifetime exposures.   We need  to  know  what is out there in terms of the
 kinds of compounds.

     We would accept  something  that is  sensitive to low concentrations of a
 number of compounds,  10 or  12 or 20,  maybe.   It would be useful in just get-
 ting the gross  idea of  exposure.

     Now, there  are some  techniques that  are  being used now in terms of stat-
 ionary sites, but  the idea  of doing it with personal exposure would be some-
 thing, of course.   And  later, as  we become  more sophisticated in what we are
doing, we would  be  more interested  in maybe very quantitative or more quanti-
 tative methods  for  specific  compounds and specific studies.

     SCHEIDE:  There  are  some techniques available.  There is charcoal tube
 sampling.  There are  passive  badges.  These are concentrator devices, and they
can be worn very easily.

     BACHMAN:  I am not sure  that they are  good for ambient concentrations,


 however.   Certainly  they haven't  been adequate in ambient samples.  Charcoal
 has  been  a big  problem.

      We are not sure that the technology that is useful in the occupational
 setting immediately  translates to the ambient area.   The ambient area is a
 lower order of  magnitude in terms of concentration.

      SCHEIDE:   You are talking about longer sampling times, too.

      BACHMAN:   That  is true.

      DR.  JOSEPH BREEN:  Joe Breen, Office of Toxic Substances.  At the lower
 concentrations, you  are getting quantitative recovery samples.  But I believe
 that people are looking at other  absorbents that are being used.

      DR.  JOSEPH BROOKS:   Joe Brooks, Monsanto Corporation.  I will be report-
 ing  tomorrow on the  work that we  are doing with a combination sorbent collec-
 tion system. We will be working  and looking at 20 carcinogens in a study
 next year.  We  are on step right  now for validating this sampling system aga-
 inst the  20 carcinogens.

      I would say that in certain  cases the technology is probably available
 now.  When you  say vapor phase organics, you are talking about such a broad
 range of  compounds,  and it is going to be a long while before that whole thing
 is sorted out.

      MS.  MARY LYNN WOEBKENBERG:  Mary Lynn Woebkenberg, NIOSH.  We are cur-
 rently involved in an evaluation  program of several sampling techniques.
 Since you mentioned  carcinogens,  we are looking at polynuclear aromatics.

      We are looking  at a mine sampling train which is for particulates and the
 vapor phase of  any PNA's that come through.  They are looking at Tenax.  We
 have gotten back one set of samples which will give us approximately 400
 data points, which will compare several aspects of the sampling trains.

      We are looking  at the possibility of using the glass fiber filters and
 silver membrane filters.  We  are  looking at the need for using the silver mem-
 brane filters to see if  the Tenax trap behind it will effectively trap all
 of the vapor phase PNA.

      We are looking  at various slides in the 37-millimeter cassette 13.  We
 are  looking at  the 10 cc cyclone.   We are also doing a particle size distri-
 bution study.

     Most of  the  study that we are doing is taking  place in foundry operations,
 because we know that there does exist a reasonable  concentration of polynu-
 clear aromatics after the mold-pouring process  and  also  after  shake-out.

     So, it is  not of the ambient situation,  but  we are  currently doing an
 evaluation.   With the rapid-scanning fluorescence detection that we have  avail-
 able to us, we  will be able to look at 18 of  the  PNA's in relatively short

     MR. OTTO WHITE:   Otto White, Brookhaven  Laboratories.   Perhaps later on
 during the course of the evening, I will get  a  chance to give  you some  high-
 lights on a project that Brookhaven has been  involved in;  that is,  assessing
 the industrial  hygiene monitoring for the coal  conversion oil  shale industry.
 The question  of PNA's came up there,  and the  approach that  the committee  has
 recommended in  the process of getting the final report in is the use of so-
 called indicators or proxy compounds—pretty  much like SO.  has been used  as
 an indicator  of air pollutants in that the total  number  of  carcinogenic PNA's
 that are available are so large that  you have to  settle  on  looking  at 10  or
 20 compounds  as indicators of the environment's conditions.

     MR. DOUGLAS  DOCKERY:   Doug Dockery,  Harvard.   I am  a little confused.
 I am trying to  understand what personal monitor data have to do  with standard
 setting.  I can understand,  and I am  a little provincial  about this,  in that
 we are doing  active epidemiological research.   I  can see  where there  is a
 direct need for personal monitoring in terms  of health effects research.   But
 in terms of defining  how many monitors we need  or where  they need  to  be placed,
 I don't see the utility of that.   The only  use  I  see for  these devices  is in
 determining health effects.

     I would  like  to  know what the motivation is  for trying  to get  these  data.

     WALLACE:   As  I mentioned in my paper,  of my  six uses  for  the  personal
monitors, five  did not require the health effects measurements.  That is,
 those five uses were  simply  to gather exposure  data  for  those  pollutants,
six or seven  in number,  for  which the health  effects have already  been  deter-

     The standard  setting  is affected very  strongly by the  frequency distri-
bution of exposures,  because the  standard setting is based  on  the  sensitive
portion of the  population.   If  indeed one finds that the  sensitive  portion
of the population  is  a rather small group that  could perhaps be  protected by

other means, then a serious and strong change might be called  for  in  the
standard setting process.

     UK. ROBERT CHAPMAN:  What other means, Lance, do you have  in  mind?

     WALLACE:  Well, for example, there was a paper study by Morgan and Morris
that took seven subgroups of the population and estimated their exposures  to
SO.., as I recall.  Only one of those seven subgroups was found  to  be  anywhere
near the level set by the standard.  That one subgroup was construction workers
who lived in the center of the city and worked outdoors in the  daytime.  I
haven't investigated this, but it does seem possible that respirators or some
other sort of individualized protection for them might be a possible  alternate
reaction to the present approach of setting single, uniform, nationwide stand-
ards.  However, there could be other possible approaches to the standard-
setting process—it could be regional or it could be more closely  identified
with that portion of the population that may be at risk but for which we have
very little data now.

     UK. CHIN-I LIN:  We are here to talk about personal sensors,  and yet we
are talking about a very wide spectrum.  I wonder if EPA has released some  sort
of tentative priority list.

     WALLACE:  I guess one of the problems is we have too many  lists.  We
have a series of lists put out by different offices.  And with  new informa-
tion coming in all the time from testing and so on, there is a  standard
saying around EPA about the "pollutant of the month."

     DR. PHILIP WEST:  Phil West, Louisiana State University.   I admit that
I am among those who are confused.  Among other things, we see  distributed
here a list of toxins that are of prime interest.  Carbon monoxide appears
all the way down through the list.  We have heard about carbon  monoxide in
so many ways.  What percent of the population, however, intermittently, al-
most every day in their lives, are inhaling 300 to 400 parts-per-million of
carbon monoxide from the moment they wake up until they go to sleep at night?
Why are we worried about the range of 10 to at the most 40 parts-per-million?
Why do we say we must have catalytic converters on automobiles,  because we
can destroy carbon monoxide, when even in tunnels the levels nowhere  approach
the concentrations that are present in each and every puff of a  cigarette?

     It seems to me that the priorities are a bit distorted here.  We  are over-
looking things that are much more important, it would seem to me,  than some-
thing like carbon monoxide.


     BACHMAN:  The  real problem is,  you  know, whatever people care  to do  to
 themselves—quick or slow suicide—in terms  of voluntary risks from day-to-day,
 the general population should be protected from involuntary risks imposed upon
 them from others  and from themselves in  their automobiles.

     Smoking—I always admit  that it is  a very real risk that smokers get.
 While it may not  kill them on the spot,  it certainly kills a number of them.

     To say that  many people  are exposed to  high levels does not say that
 the lower levels  in the ambient area don11 cause a real risk and pose a ser-
 ious threat to the  rest of the population.

     That is why, indeed,  we  care about  much lower concentrations.  Of course,
 you could use the same argument about people in their occupations, I suppose,
 who receive much  higher concentrations.

     Some of the  people who don* t smoke  can  also be quite sensitive with re-
 spect to carbon monoxide or perhaps  other pollutants at lower concentrations.
 And perhaps if they smoked, they would drop  off much faster.

     So, the basic  premise is philosophy—the will of the people expressed
 in the Clean Air Act.   We  can't have people  imposing involuntary risks on the
 general public just because a large  percentage of the population—some 30 to
 50 percent—is imposing on itself a  much, much greater risk.

     WALLACE:  Perhaps I could continue  going down the list of personal monitor
 priorities expressed  by EPA offices.

     The mobile sources portion of the air program has a somewhat different
 interest than the rest of  the air program.   Of course, they are most interested
 right now in the  diesel question.  The automobile manufacturers need a decision
 some time around  the  end of 1979  and the early part of 1980 as to health ef-
 fects of diesel particulates.   So, a personal monitor for diesel participates
 is one of the high  priorities for the mobile sources group.

     They have also mentioned nitrogen dioxide, transportation-related pollu-
 tants, including  lead,  and again,  the ever-present organics.

     Within ORD we  have several  laboratories that have cited their priorities.
The Environmental Sciences and Research Laboratory, which is responsible for
development of instrumentation  for air pollution measurements, feels rather
 strongly that the sort of  work  that  we do on personal monitors should build on

the work that has already been done; that pretty much limits  them  to  the  cri-
teria pollutants.  Among the criteria pollutants, they have chosen ozone, with
CO as a close second, as their top priorities.

     The next list is a list of pollutants selected by our Health  Effects
Research Laboratory here in Research Triangle Park.  However,  the  list  is
dated March 1978, and EPA being such a volatile agency, I would be glad to
hear anything newer that our HERL representative, Dr. Chapman, has to give
us.  The old list that we have here is basically the criteria pollutants  in
a particular order.

     CHAPMAN:   I am Robert Chapman from EPA.  I might say just to  start off
that I am an epidemiologist concerned at the present time mostly with studies
of air pollution in communities as opposed to occupational studies of pollution

     1 have been asking myself and some colleagues back at work throughout the
conference what is perhaps a more fundamental question than questions that we
have all been dealing with up until now (that is from the standpoint  of
epidemiologists interested in community health effects and air pollution
exposure):  Where can personal monitors be appropriately used and  where can1t
they be used?   I wonder if we could grapple with this question a while?

     When I think of health effects of air pollution and other types  of
pollution, I am inclined to divide them into effects of long-term  exposure and
effects of short-term exposure.  1 am having a very difficult time seeing how,
from the standpoint of epidemiologists, personal monitors are going to  be
particularly helpful in our studies of long-term exposure effects.

     1 would be delighted to be talked out of the way I am thinking right
now.  But up until now, 1 have not seen any way in which we would  be  able to
put them to any particularly useful application in such studies, which  leaves
me with the notion that 1 think they would probably be most useful—and I
am sure that I am not saying anything terribly new—in studies of  relatively
short-term exposure.

     This leads me to a couple of questions, one of which is:  From a phys-
iologic standpoint, what kind of averaging time would be most relevant  for
us to have a personal monitor that would be of any substantial use to us?

     It seems to me that the type of monitor that we would be most interested
in would be one able to characterize, at the most coarse level, 24-hour aver-


 age exposures;  and considerably finer gradation  within  that would be useful
 to us, largely  because we really don't have  a good handle, as I think some
 of you think we do,  on short-term health effects of the pollutants.

     We do have some fairly confident estimates  of the  types of health effects
 that are being  promoted by pollution exposure, but very, very few available
 data can be characterized as being solid.

     So, the finer the gradation of exposure that we can get a picture of
 within the day, the  better we are going to like  it.  I would say first that
 we should have  certainly 24-hour average exposures available to us for what-
 ever pollutants we use personal monitors for.  I would also say, in my mind
 and I think in  some  of my colleagues'  minds, that the second most desirable
 type of exposure averaging time would probably be the daily maximum hourly
 average within  a day.   The third most desirable—which is practically equal
 in my mind to the second—would be some index of the maximum instantaneous
 concentrations  of the  pollution in question  for  each day.

     I don* t think that exposure times over  a period longer than a day without
 data for shorter exposures are going  to be terribly useful to us, to be per-
 fectly honest with you,  because of the types of  effects or the types of bio-
medical variables that I can see us measuring in those studies*  I don't think
 it is terribly  plausible to think that that  kind of variable would be affected
 by pollution exposures a week ago as  opposed to  today or yesterday.

     I don't really  see  much point in a big  effort being made toward develop-
 ing monitors whose finest capability  is a week-long exposure.

     WALLACE:   Can I ask about  the lead exposure time, which is a month?

     CHAPMAN:   I would have to  admit,  Lance, that I am not very good on that.
There are a lot of people who can tell you more than I can.  I should probably
preface what I  am  saying:   I  think that  what I know best is criteria pollu-
tants—criteria pollutants  as they would have been defined up to a year ago.

     Yes; you are  quite  right.   I  am  trying to generalize here.  There would
certainly be some  exceptions.   I  think, however,  that the exceptions as far
as I know aren't  going to  really  have me change my mind too much, even though
individual ones—other individual  ones—might well be thought of.

     1 have been  wondering  whether I am  right, for example, for a lot of the
organic substances.  And  I  must  say I am wondering, again from the point of


view of  a  community  kind  of  epidemiologist as opposed to an occupational one,
what kind  of  short-term effects  can we  hope to see in response to exposure
to  some  of the  organics?

      I must say I  really  don* t know the answer.   1 have come up with the
question that is rather large in my mind:   How helpful to us are the various
organic  types of personal monitors going to be?   I won't say very helpful,
and 1 won't say not  very  helpful, but I will certainly say that the answer is
highly uncertain in  my mind.  1  will have  to also say that 1 can't really come
up  with  a  specific application in an epidemiologic study of personal monitors
for organics.

      I must say that the  kind of monitors  that I  can best get a handle on
the use  of, a)  because 1  know this kind of pollution best, and b) because I
think there may be something to  it, too, is the criteria pollutants, again
excluding  lead.

      WALLACE:  You raised two questions in my mind, one of which is, among
the criteria  pollutants,  do  you  have a  ranking or do you have studies that
are ongoing now that could be improved  or  enhanced by the addition of personal
monitors to those  studies?

      CHAPMAN:  Well, I think there are  studies ongoing now for really all
of  the criteria pollutants,  excluding lead, that  would be enhanced by appro-
priate personal monitors. And 1 must say  that I  want to stress the word

      1 am  very  impressed  by  a lot of the technology I have seen in the last
2 days,  but I have to voice  some skepticism as to whether this technology
has really received  the full gamut of laboratory  and field testing that would
make us  confident  of using it.

      1 say this largely because  of assurances that we have gotten over the
years of the  big instruments—the instruments that are put in ambient situ-
ations that are not  worn  by  people.  Even  these instruments, to my way of
thinking,  are not  entirely debugged as  yet.

      1 think  it would be  an  unrealistic request of any maker of a personal
monitor  to ask  within 2 or even  3 years of the conception of the instrument
to  have  it entirely  ready for field use.

      1 think, on the other hand,  that it would not be fair to us epidemiolo-


 gists  to  plunge into the field use of such instruments  until  they  were  fully
 verified  both in the lab and in the field.

     WALLACE:   That is a perfectly reasonable  request to make.  Laboratories
 within ORD would have the responsibility for evaluating these instruments
 in the field.   I am referring to the environmental monitoring units at  our
 laboratories,  either here or at Las Vegas.

     CHAPMAN:   That is really what I am  coming down to, if I could just
 say another  sentence or two.   I really can't determine rank within the  cri-
 teria  pollutants, and therefore would be inclined to go along with what I
 think  you said about the ESRL,  and that  is  that 1 think we would feel the
 most confident using first the monitor that presently is closest to develop-
 ment.  I  think we could come  up with an  interesting and relevant application
 for it, as long as we were convinced that it was really going to work.

     That is more important in my mind than a specific pollutant which is
 chosen to be measured.

     DR.  JOHN  McSTRAVICK:   John McStravlck, Energetics Science.  You bring
 up a very interesting point that is something we haven* t discussed here in
 the past  2 days—that is,  performance specifications.  What are you looking

     OK.  You  make a very  general  statement which I accept.  But what kind
 of accuracy do you need?   What  kind of configuration do you need?  These are
 the things that we have to get  sent back to us as well.   I am hoping that this
will come out  here tonight at the  workshop.

     CHAPMAN:   I think that that is unrealistic.  I think this is the kind
 of thing  that  we would  have to  sit down  together to discuss for several hours
 for each  substance,  really.

     WALLACE:   John,  correct  me  if I am wrong, but it does seem to me that
 I remember you or someone  at  ESI stating  that one of the reasons that you de-
veloped the 9000 series was because of Anthony Cortese1s studies in Boston
 in which  he used the Ecolyzer and  stated  that he needed  something that was
more easily portable.

     McSTRAVICK:   Absolutely.

     WALLACE:   That  seems  to  me  to be an extremely important point, which is

that industry will develop the sorts of things for which they are given  some
indication that there is a need for.  And it is, I think, a challenge  to EPA
to come up with the specifications that we need.

     For example, do we need tremendous accuracy in personal monitors  for
some of the uses we are talking about?  1 am of the opinion that we do not.
We do need sensitivity.  We need to go down to ambient levels.

     But the sheer number of data points that we can get may make it possible
to compensate, by means of a large N, for the fact that they are not as  accu-
rate as the large fixed-station monitors.

     MEIER:  Could I read off something that Joe Behar and I put together
last night?  We did just what you asked.  Don* t hold me to what I will read
to you as  representing all of EPA.  This is simply what we at Las Vegas
called desirable characteristics.  It is quite a long list, but we worked  it
out among  ourselves and thought that we might present it here, which we  have
not done so, so far.

     Let me read some of the things to you, and they kind of answer your
questions, I think:
     1)  Lightweight—light enough to be worn easily by the general public;
less than  1 pound, including power supply, is desirable*
     2)  Durable—able to withstand treatment received by personal equipment
such as a  wrist watch.
     3)  Dependable power supply—in our terms, a power supply must be able to
provide constant power over a minimum of 24 hours.
     4)  Easily changed.
     5)  Tamper proof.
     6)  No exposed controls that could be easily adjusted by the wearer.
     7)  Accurate—an accuracy of plus or minus 20 percent at 2 sigma  is desired.
     8)  Sensitive—should be able to detect the parameter of Interest at
one tenth  of the most stringent air quality standard.
     9)  Sensitive to range of concentration expected in air.
    10)  Easily calibrated.
    11)  Separate readout—hourly and dally Integrated readings are desired.
    12)  Reliable—less than 1 failure for 10 units.
    13)  Free of common Interferences—total interferences less than 10  percent
of total signal.
    14)  Traceable to a standard method If possible.
    15)  Inexpensive—less than $500 per instrument is desired.
    16)  Stability—capable of operating for a minimum of 24 hours without
changing components.

     Note  that that was just a quick list that we  put  together.

     DOCKERY:   I would like to state a different viewpoint  from what  you were
saying about  the instruments.   We  are kind of  working  at  the edge  of  the
science  here  in trying to measure  health effects for new  pollutants that are
coming along,  and things that  we have little knowledge about.  The instrumen-
tation we  are having to use in doing this is really not field  tested,  hard
and fast instrumentation, but  Instruments that are basically laboratory  in-
struments  that we are trying to apply in the field.

     We  have  a lot of failures in  going out and trying to use  these.   I  think
if you are looking for a market where you are  going to go out  and  sell a lot
of these monitors, it doesn't  exist.   There are going  to be a  couple  of
e p id end o log leal studies, and then  we  are going to go on to something  else.
I don't  think the market exists.

     McSTRAVICK:   In the Industrial workplace,  it does exist.

     DOCKERY:   In the industrial workplace, it  does, but not in the ambient

     McSTRAVICK:   Well,  I can  address myself to both of them.

     MR. DAVID BRAUN:   Dave Braun, 3M Company.  I am an engineer.  I really
don't understand  or know that  much about  epldemlological work.  But going
back to your  original idea of  short-term  and accessible effects that you can
look at, it seems  to me  that with systemic  toxins, don't you really want to
have a data base  for which the  time interval is not so important, even if it
is a month?  Doesn' t that fall  into the  realm  of the kind of work you do?

     CHAPMAN:   Are you talking  about  things, exposure  to which—such as vir-
tually any metal—would  be manifestly  bad  for you?

     BRAUN:  No,  but let's say  carcinogens or anything where the long-term
effect is  important.   When you  were saying  that you really don't see the
reason for extended period sampling,  1 think I would submit that that is one
place where—or am I understanding your work incorrectly?

     CHAPMAN:   No,  I don't think so.  Let me try and give you some feeling
of why I think it  is a limited  application.

     One problem is that we  are absolutely proscribed from doing prospective

studies in cancer, I think, given both the necessary sample sizes and re-
sources—financial resources—that would be required, even in the best of
all possible worlds.  Given, at least for us Government researchers, the exi-
gencies involved in launching sustained studies, I think it is unrealistic
for a prospective seller of personal monitors, say for carcinogenesis, to
launch a heavy research and development program in expectation that scien-
tists in EPA are going to be able to monitor a truly big prospective study—
that would be a mistake.

     1 think it is too bad, but 1 think it is true.  I wish we could do it.
But up to now, we are just playing.  We have not been able to because of
shifting tides that are out of our control.

     Another thing that gives me trouble is that I am not really addressing
the physiologic part of the question right now.  A more scientific than phys-
iologic reservation in my mind is the difficulty of using monitors now to
try and determine what exposures have been in the past.  Because we are not
able to do prospective studies, we are more or less stuck doing retrospective
ones; that is, finding pockets, let's say, of high cancer Incidence, and then
trying to determine in a retrospective fashion what the people have been ex-
posed to.

     Now, monitors might be useful to give us an indication of substances in
the town now that we didn't really expect were there, or that we thought maybe
were there but we didn't know about.  But beyond that fact, in a retrospective
study—a fact that it seems to me could be turned up with a stationary monitor—
a very limited number of stationary monitors may be rotated throughout the
town.  1 really don't know what personal monitors could add.  Do you see what
I mean?

     BRAUN:  Yes.

     BACHMAN:  Let me add something to that, though, because we made a request
to put vapor phase organics on the list.

     1 agree with your comments with respect to epidemiologic studies.  We
didn't put it on there because we saw a use in epidemiology.

     What we see is a use in exposure monitoring, and stationary monitoring
is good and useful for this purpose; emissions monitoring is good and neces-
sary, and modeling is going to be used.

      In the standard setting with carcinogens, however,  it  is going to be im-
 portant to us to have some studies of total body burdens and  various exposure
•regimes to validate our models and to develop better exposure models than we
 have  now.

      And we wish to measure what is in the air now so that  we can  regulate
 the emissions that are coming out now to prevent future  problems—not wait
 to see  the dead bodies 20 years from now.

      So, the idea is to get these monitors used in standard settings.   So,
 we do see  a use for these instruments, but certainly not in epidemiology.

      CHAPMAN:   So, you are talking about studies in which there are  no specific
 biomedical measurements?

      BACHMAN:   That is exactly right.   We are assuming a risk of cancer.
 We can  do  some extrapolations and get some ideas just from  the exposure.
 This  is the use that we see for it.

      WEST:   First, I swear that I had no input whatsoever to  what  Gene said
 a  few minutes ago about certain specifications that would be  desirable for
 personal monitoring.  He was obviously talking about passive  units.  And so,
 three of us in the room right now bless him because he summarized  what we
 would summarize in our talks tomorrow.

      I  would like to comment on the incremental measurements  of hazardous
 materials.   Personally, I have been involved in this work for some 27  years.
 We started on S02 and worked on the acute toxins.   These were obvious  and
 spectacular,  and people were worried about them.   Apparently,  we are  still
 worried about them.

      I  sometimes think that this is in error,  because regarding a  few  toxins,
 if they don't kill you, don11 worry about them.   If  they do kill you,  of
 course, someone else is going to worry about them.

      You have the acute toxins and the cumulative toxins.  And there  is another
 group that seems to be overlooked,  at  least in most  classifications—that would
 be additive toxins.  In the case of additive toxins,  it  doesn't matter very
 much  about  how much you get in any one increment.   But everything  points to
 an end  point  at which you have a certain liability that  you must pay off.

      Concerning radiologic exposures,  the same  thing  applies.  You can* t

backtrack after you have had it.  You smoke cigarettes.  Now, loosened up by
martinis a good many years ago, someone who was very knowledgeable in the
industry confessed to me that he knew that a person could smoke one pack of
cigarettes a day for 40 years, and there was, say, a 20 percent chance of
dying of lung cancer.  Or two packs of cigarettes a day for 30 years would give
you exactly the same chance.  You could have four packs a day for 10 years.
You come to the same thing.  So, it was strictly additive.

     So, if that is the case, then 1 should think that we would be interested,
not in breaking this down to a daily exposure or even a weekly exposure, if
we just had the time range averaged for a month for some things.  Suppose we
say vinyl chloride or benzene.

     Now, 1 don't think that this is necessarily significant as long as it
is sublethal.  It could be; we don't know.  But it could be the additive
effect—that is, the total insult from all of these supplemental increments
that add up over 5 years or 10 years, or if you don't have very much of it,
it may take 30 or 40 years.  But somewhere along the line, there is going
to be some sort of an accounting.

     Now, if that is the case—then some type of monitor that is simple or
eloquent in its simplicity—truly 1 think a passive type monitor is perfectly
acceptable, and the accuracies certainly are good.  1 might add—and 1 will
point this out tomorrow—that we have monitors that will give us a time-
weighted average for a 30-day exposure.

     I might mention also that we have personal monitors that are passive for
tetraethyl and tetraethyl lead that we could easily apply in the ambient
atmosphere and have 1 month exposures and get the time-weighted average for

     So, these are possibilities that I think should be explored.  Now, un-
fortunately, funding agencies have been very parsimonius.  We developed a
vinyl chloride monitor for $43,000, and we begged and cried to get that.

     So, with a little money, I think there is a lot that can be done.  But
I do think that extended periods are justified in some things.

     CHAPMAN:  I think that is right.  I think it is right for compounds of
low concentrations.  I might say that I would be delighted to eat the words
that I spoke at the outset of my remarks, especially if what we are constantly
doing in community health research and air pollutants is associating short-

 term—say  dally or shorter—pollutant exposure with physiologic changes  of
 uncertain  significance.

     People  will cough,  for example.   Some  people may wheeze.  Other people
 may develop  eye irritation.  We  have  no  real  sense that a great many of  these
 symptoms that  appear to  be reversible are more or less innocuous.  We have
 no sense of  the degree to which  these are harbingers (after repeated exposures)
 of some overt  disease developing over the long term.

     If, with  respect to a lot of the compounds, in particular the organics,
 certain short-term symptomatic or biochemical or hematologic responses could
 be elicited  and tied to  the development  of  some real illness condition down
 the road,  I  would be more than happy  to  take back in a minute what I said

     One thing that gives me a good deal of trouble, though, is the overuse
 of scientific  energy,  or technology development to the exclusion, let's say,
 of determining what the  real meaning  of  some of these shorter-term responses
may be, be they after a  month or after a day.

     WEST:   But not eliminate either  one.

     CHAPMAN:   That is right.  Bear in mind the necessity of the concurrent

     DR. LARRY LOCKER:   Larry Locker, Solid State Sensors.  I just want to
make the comment  that  the people who  are interested in personal monitoring
and selling  equipment  would like very much to see Government agencies in-
crease the criteria pollutants by hundreds, by thousands.   And the instru-
ments that you need—the analytical instruments you need—to get the infor-
mation that  you require,  are, I  think, quite different from what the prospective
market needs.

     They want something  that is  low  in cost, light in weight, and which they
don't have to  change every  day.  They want something that can be used over
a long term.   It  is only  a  question of reliability of statistics, the way
I see it.  For you  to  get statistics on health effects, you need a different
type of instrumentation  than would be ideal for someone who wants to monitor
in an occupational  situation.

     CHAPMAN:  We certainly want  some common features without extreme unob-
trusiveness.   We  just  really  can't use them in our studies to any particularly

good degree.  That is every bit as important as the accuracy of the instru-

     Maybe 1 shouldn't say this since by doing so I am pointing out the
limitations in our field.  But I am inclined to agree with you, Lance, that
extreme accuracy isn't all that important.  My first reaction is that the
precision of instruments, reproducibility of response from one instrument
to another, may be more important for us than very delicate accuracy.

     DR. EDWARD McMADDEN:  Ed McMadden, Gastek.  As a manufacturer, 1 guess
I can say in all honesty that we are quite pleased with the possibility of
EPA1s getting into this kind of instrumentation.

     But we seem to be talking a great deal about organics.  I don't believe
there would be a challenge if I were to say that within the criteria that
were set down, which Include hourly accumulation, that there is nothing
presently in the wind other than 24 badges which are changed on the hour
at the stroke of the clock which are going to do that job for organics.

     Rather than spending a lot of time chasing off on that tangent, it ap-
pears that within the criteria pollutants—although the instrumentation avail-
able right now does not meet all your specifications—in CO, it is quite close.
Out of our total of 16 entries for pollutants to be measured, CO constitutes
a quartet of those.  The same basic type of technique can be used to measure
another group of the nitrogen oxides, sulfur dioxide, and so forth.

     So, that takes care of a total of half the entries.  Can we now eliminate
some of the things that probably are at best well beyond the horizon at the
present time ami concentrate particularly on CO because we could have something
reasonable with the technology that is available now?

     I am not sure that your $500 criteria will be met.  But be that as it may,
what numbers are we talking about?  Are we talking about dozens?  Are we
talking about hundreds or thousands?

     This will be helpful to us if we can get specific and right down to the
point where we are close today as manufacturers.

     WALLACE:  Two comments in response to that.  One is that that is a fair
question about the numbers.  I can probably give you some response to that.
The other is that we really have two areas involved here.  First, we have
studies that we could do almost immediately with existing (properly evaluated)

 instruments.   Secondly, we look into the future and  decide  what  we  need to
 develop for  the  next few years or the next dozen years  or so.  And  there,
 we do have to  look beyond what is available now.

     BACHMAN:  I am afraid there is a misconception  here.   There are  several
 kinds of needs.   Chad was talking about epidemiological needs.   When  he says
 an hour, that  is one thing, and it is useful in that short-term  system.
 You are asking for organic monitoring,  and 1 think OTS  and  the others on the
 list—by the way,  there are four CO entries and four organic entries—any-
 way, the other people are also asking for organic measurements.  They are
 talking about  the same kinds of problems we are:  carcinogens and these long-
 term effects.  What you are talking about is apropos, and maybe  we  are not
 so far over the  horizon.   That is important to  us.   But I think  it  is a real
 need you will  have.   A number of people are obviously interested in it.  So,
 I don't want to  turn off anybody.

     McMADDEN:   1  am not trying to down play organics.   What I am saying is
 that we have in  one area an approach that is reasonably close to your  desires,
 What I am saying is that if this group  cannot come up with  some  sort  of meet-
 ing of minds where we are close,  I think it will be  less productive to get
 driven off on  other things which really are much further away.

     Granted,  that for the long term,  it is useful.  But for manufacturers,
 that would be  another thing.

     MAGE:  I  would like to ask Otto White if he  wouldn't mind briefing us
 on what was discussed at  Brookhaven.  1  think maybe  some of the  very  same
 questions have been handled there,  and  perhaps  it would  be  beneficial  for  us
 to get this input  from their point of view.

     WHITE:  Brookhaven was asked by the Department  of Energy to assess  the
 industrial hygiene monitoring needs of  coal  conversion  in the oil shale  in-

     The way we  attacked  that was to formulate  what  we  interpreted  to  be an
authoritative  group,  consisting of  names within the  field of industrial
hygiene.  The  committee met three times.   The final  meeting concluded  with
a symposium similar  to this,  in which the theme of the  first day was  to
 identify the industrial hygiene monitoring problems  for  the emerging  possible
 field industries.  On the second  day, the theme was  to address the  state of
 the technology for performing these monitoring  needs.

     We identified at that time a number of monitoring needs for the occu-
pational environment.  They include things like the evaluation of health
effects, monitoring performance of engineering controls, monitoring trends
within those controls, developing designed requirements for doing the controls,
documentation of exposure levels, testing for compliance purposes, measuring
process losses, selection of suitable respiratory detector devices, identi-
fication of contamination sources, and assessing emergency conditions*

     When we started attacking the problem, one of the things that turned out
to be extremely important were the polysaccharide hydrocarbons.  If you look
at the process characterization—the effluent characterization studies—you
find some thousand compounds for which toxicological studies indicate that
some neoplastic or carcinogenic properties are associated.

     We also associated the industries with their use of some very common
materials, a number of which appear on your list.

     So, the first thing the working crew did was to categorize the hazardous
potentials of some of these materials.  We developed four categories:  the
first was materials immediately hazardous to life and health.  The second
category was high risk but not immediately hazardous.  The third was a moder-
ate risk, but not highly hazardous.  Fourth was short-term, nonroutine, high
hazard conditions.

     We had to ask ourselves some questions.  The questions that required
answering were:  What compounds were present in each hazard category class?
Is the monitoring technology currently available for that particular pollutant?
Is the current technology compatible with future needs?  If not, is technology
being developed to meet that projected need, and when is the technology needed?

     In the high hazard category we listed six compounds, hydrogen sulfide,
carbon monoxide, the oxides of nitrogen, oxygen deficiency, and coal and
shale dust, as well as hydrogen.

     We said these materials were present in such concentrations within these
processes that acute irreversible effects could result from exposure.  The
working group believed that the technology was available to provide equipment
which could be used as personal and area monitors for continuous and direct
reading and long capabilities for hydrogen sulfide, carbon monoxide, oxygen,
and hydrogen, of course.

     However, the technology was lacking for monitoring exposure levels of

coal dust and nitrogen oxides in a  format  that would give the personal capa-
bility.  We also did  not  believe that  these  six substances were unique for  the
coal conversion oil shale industry.  So, this particular group recognized
no special priority from  this standpoint for these particular forms of in-

     The second category  that we looked at was the high risk but not immedi-
ately hazardous category. We classified within this category materials which
cause irreversible effects but were not present in sufficient quantities to
produce acute toxicologlcal  effects.   The carcinogenic, mutagenic, and tera-
togenic organics are  included in this  group, as well as some of the trace
metals, dust, and fibers.

     I don11 think it would  be fair to you to read through the draft of the
report, but I will point  out some of the specific research suggestions that
the committee will be recommending  to  the Department of Energy.

     In terms of personal monitors for gases, the working group encouraged
that there be additional  evaluation of current instrumentation under various
conditions of Interference,  temperature, pressure, humidity, etc., to ascer-
tain the appropriateness  for future needs.

     It is likely that passive monitors will remain a primary monitoring tool.
Improvements needed include  the  development of a new substrate to expand the
diversity of measurable compounds and  to allow higher flow rates.

     We believe that  this area is likely to provide high payoff in a few years.

     We also recognized the  need to increase personal monitoring pump flow

     And that has been identified as one requirement for obtaining sufficient
samples to facilitate  bioassay samples.

     Both benzene-soluble fraction and benzo(alpha)pyrene have been recognized
as potential indicators in the coal conversion industries.  There appear to
be some drawbacks in  use  of  these particular tools as monitors in that you
need a little more material  to run your sample than is currently available,
say, for dust samples.

     So, we recognized that  you  need to have personal-type pumps, similar to
the MSA Model G.  The  committee  looked at second generation instrumentation.

The working group encouraged  the  development  of  new sampling  instruments which
will  facilitate  automatic  analysis  and  data  recording  to  permit  the handling
of large amounts of data.  And  the  working group therefore  encouraged  the
development of the personal gas chromatograph.   The Stanford  research  group
has a pocket-sized gas  chromatograph which can monitor some 10 compounds and
give  you a time-weighted average  or the cumulative  concentration or dose at
the end.

      We would encourage this  type of instrumentation,  which presently  has
been  funded by NIOSH, but  DOE should look at  this for  the polysaccharide hydro-
carbons which may require  another type  of detector  and maybe  some different
substrates in terms of  the column.

      Additionally, we felt that real time data reporting  is necessary  only
for the Class A  compounds; that is, the carbon monoxide,  hydrogen sulfide,
and those  immediately hazardous compounds.  We felt that  in other cases, the
most  attractive  approach is a personal  sampler,  which  either  can report re-
sults of real time analysis over  the course  of a day or collect  a sample over
the course of a  day, and at the end of  the day the  sample is  inserted  into a
device which either analyzes  or records the  exposure automatically.

      We also found that there is  a  need for  looking at nitrogen  compounds.
There are  no good sampling techniques for ammonia and  hydrogen cyanides, as
well  as nitrogen, which are of  particular interest  in  the oil shale industry.

      In the area of aerosols, our comment here is that this is an area where
there are  many needs, but  that  it is one that does  not appear to be right for
exploration.  For example, there  is a need to do in situ  chemical characteri-
zations, with no avenues for  research in this area  being  evident, and  a size
cutoff at  15 microns and 2 microns, consistent with the recent EPA finding
should be  made.  Current technology is  adequate  for this, and development of
more  sophisticated devices is not recommended at this  time.

      We believe  that in terms of  particle or  aerosol monitoring,  we need to:
improve the efficiency  of  collection; improve the extraction  efficiency for
the separation techniques, chemical specification,  and in situ separation of
soluble and insoluble particulates; and to review the  state-of-the-art used
for personal particulate samples  to determine their needs in  the future

      In the area of metals, we  identified possibly  five metals of concern.
They  include arsenic, cadmium,  chromium, and  nickel.   Lead  was not included
in our particular list.


     However, we  are concerned that the toxlclty  of metals  varies  with their
biological availability.   A classic example is  chromium, and also  chemical
compounds.  Our current technology does not provide insights on  these  factors,
and research is needed to provide this  additional information.

     One area of  exposure in terms of these new industries  is the  contamination
of surfaces—that neither are there safe permissible levels for  skin contami-
nation, nor are there good techniques for evaluating skin contamination.   We
recognized a need in that area.

     One area where  the committee wanted to provide some insight was in ana-
lytical development.   We  thought  that we should encourage the refinement of
analytical tools  and techniques for materials which are identified by  those
characterization  studies—that is, process  streams and air pollutant characteri-
zations.  But we  did not  believe  that it would  be appropriate at that  time
to dilute the available funds to  develop a  few  high technology devices.  There
are a number of papers on the Auger spectrometer.  We thought that that wasn* t
the particular analytical device  that would receive funding in other areas.

     Basically, that summarizes the committee's concern.  We felt that person-
al monitoring had its place.   We  were concerned that these industries were
going to be approximately 1  square mile in  terms  of area.  The use of area
monitors probably should  be  evaluated.   But certainly for operations similar
to all refinery-type operations,  personnel  monitoring is probably going to
be one of the major  tools.

     It is interesting to find that some of the compounds that we are concerned
with are compounds that EPA  is concerned with in  the environment.

     MAGE:  Thank you,  Otto.   In  your view,  as  a  representative of the com-
mittee, did it also  envision personal health monitoring and personal exposure
monitoring simultaneously?

     WHITE:  The  area of  personal health monitoring was not really addressed
by the committee.  The committee  was  concerned  about a time some 20 years
from now, having  a Mancuso-type concern in  terms  of this particular industry.
And so, the need  to  get a handle  on all the parameters that are available
la paramount.  We  did not address the physiological monitoring question; no.

     MAGE:  I would  just  like  to  say  that we are  thinking of epldemiological
studies of a thousand subjects and a  thousand personal monitors, and that  is
not the only use  of  the personal  monitor.


      Everybody knows that there are economic considerations involved in setting
 air quality standards and regulations*  We have just witnessed the Automobile
 Manufacturers Association's ability to get a relaxation of our standards on
 vehicle emissions set back year after year.  These are economic considerations.

      One of the economic considerations that we have to take into account
 is the number of people who are exposed to particular pollution levels.
 We are setting standards which are required to be met by the States through
 implementation plans.  The States have to reduce emissions in order to bring
 air pollution down below the standards.

      It is going to be very important for us to prove that not only are those
 measurements representative of communities, but indeed that there are a
 significant number of people who are exposed.  Also, we will have to answer
 the question:  "Are the effects of in-home and nonambient environments more
 pertinent for us to worry about than the 1 or 2 hours of ambient exposure
 that we might have outdoors?

      So, there are many reasons why we should look at personal monitors other
 than for the purpose of conducting epidemiological studies.

      I look forward to the day that the personal physiological monitors are
 inexpensive, handy, and reliable, so that we could put one out with each per-
 sonal exposure monitor.

      But I think until that day comes, 1 still see a use for personal exposure
 monitors simply for characterization of exposure, if for nothing else.

      WALLACE:  All it requires is a redefinition of the word ambient, or rather,
 going back to what it originally meant, which was surroundings—the air that
 surrounds you in the vehicle and the home is ambient air.

      DR. VIC SOCOL:  I think we have a basic question here.  I think we have
 the manufacturer represented and the scientific community represented.  Who
 is going to lead the way?  Is the manufacturer going to establish what he
 can build, the criteria that he can build it to, and tell the scientific
 community, "This is what we can do"?  Or is EPA or the scientific community
 going to tell the manufacturer what is wanted both qualitatively and quanti-

      It is the question of the cart leading the horse or the horse leading the

     WALLACE:   I  think it is exactly that  question.

     SOCOL:  That is  what I hope will come out of  this whole meeting.

     WALLACE:   Yes.   We have one manufacturer asking to tell him what our
plans are.  I  think that it is the sort  of challenge that we ought to try  to
meet.  In response, you can tell us whether we are off target, and perhaps
we can come together.

     SOCOL:  I think  a manufacturer can  do whatever the scientific community
requires him to do, although not necessarily at the cost that you have stated.
It is a question  of numbers.   It is a question of cost.  It is a question
of what you really need.

     DR. GEORGE MALINDZAK:   George Malindzak from Ohio.  I represent another
level of confusion, in that I presume that the majority of the group is talk-
ing about personal exposure monitors.  I represent those individuals who are
concerned with the physiologic monitors.

     My confusion is  associated with the realistic need for collecting personal
exposure information  in the absence of the physiologic consequences.  Where
does that get  you?  I  don't understand the purpose of putting all this effort
in collecting  exposure information if you  don't understand what the human
response is going to  be.

     WALLACE:   There  are  two  possible answers:  one is that it is expensive
to collect extra  information  on health effects along with exposure; and the
second is that  if you  already have established health effects for given levels
of the criteria pollutants,  then the exposure measurements satisfy certain
other needs and thus  should be extended.

     For some  noncriteria  pollutants, such as carcinogens, there may be some
purpose in collecting  exposure information without concurrent physiological
responses, since  they  have  already been shown to be carcinogenic in some
clinical or toxicological way.

     I am not  trying  to say we should not  collect physiological evidence.
I am in fact in favor  of  that.   There are  questions of funds that have to be
considered also.

     MALINDZAK:   Technologically,  it is probably simpler to collect physio-
logical information than exposure  information.


     WALLACE:  There is also a factor of ignorance of the possibilities of
collecting physiological information on the part of the people who are in the
air pollution field, in many cases.  I certainly speak for myself about not
knowing what is possible and what is not possible in that area.

     I would really be anxious for the insights of the physiological contingent
to be brought forward.

     DR. THOMAS FEAGANS:  Tom Feagans, EPA.  I would like to support Lance and
Dave in their thought that this second kind of information on personal exposure
would be very valuable to us in OAQPS.  Whatever your level of knowledge
about health effects of a given pollutant, it helps us to estimate what the
consequence of various possible standards we might set would be, no matter
what the exposure levels would be.

     MAGE:  Even though 1 believe that the Clean Air Act says that the Ad-
ministrator of EPA is supposed to set the standard irrespective of economics,
I might say that economics do play a role.  It is a significant question.

     Now, in order to evaluate the economic effect, which of course is in
units of millions of dollars, we have to know what level of exposure the
general public receives, and we have to know, if we set a standard, how many
people we are protecting.

     Thus, even though we may not be measuring the health effect directly in
an epidemiologic study, it doesn't preclude the need to measure the exposure.

     On the other hand, we have to recognize that our standards are living
standards.  They are not dead and fixed in concrete.  The Clean Air Act re-
quires us—mandates us—to review our standards every 5 years at a minimum,
and the Administrator can change a standard at any time that he feels there
is new evidence to justify it.

     There is a long-range payoff for control of air pollution or relaxation
of standards if you can justify it, so that it does justify expenditures of
millions of dollars in order to save billions.  But it is hard to get those

     I think that it will be very beneficial that we are doing the development
work on the personal physiological monitors, because 1 don't think that the
last epidemiological study has been done yet.

     CHAPMAN:   I  certainly hope not.

     SCHEIDE:   I  have one comment.  We  have to  keep  in mind why we monitor.
The reason we monitor is to identify  problems.  Whether  the action level  is
set low or high is another question that  Government  organizations have  to deal
with.  No matter  where it is set, we  still  need to monitor to identify  pro-
blems.  That is what we have to keep  in mind.

     MALINDZAK:   But a problem represents the differential between what you
expect and what you find.  If you never look at the  physiologic information,
you are never going to change that  standard with  respect to physiologic con-
sequences.  If  you are going to devote  all  your effort to exposure monitors,
you are never going to change the physiologic consequence which you learn to
establish the standard to begin with.

     My point is  that there is a maldistribution  of  effort; that all the  ef-
fort seems to be  going into the exposure  monitor.  If you want to characterize
the population  exposure,  you don't  need personal  monitors.  You very well
should have fixed stations.

     But the object is to characterize  the  physiologic response to this ex-
posure, not just  to collect exposure  information.

     FEAGANS:   I  agree with one point you made, but disagree with another.
I agree with the  point that personal monitors can certainly play a role in
epidemiological studies to help us  learn  more about what patterns of exposure
cause what effects.

     But I don't  agree with the second  point about a stationary monitoring
network being sufficient  to tell us all we  need to know about those patterns
to make the estimates that I mentioned  the  last time I spoke.

     MALINDZAK:   I  said if you want to  characterize the exposure of population,
you don't need  personal monitors.  The  purpose  was to characterize the popu-
lation exposure,  not  to know all about  the  exposure of that population.

     FEAGANS:   But  I  think the personal monitors  give you a finer grade of
characterization  that is  very  helpful to  us in making the estimates we need
to make.

     MALINDZAK:   There is no question about that.  I don't argue with that
at all.


     McKINNON:  Not only a finer grade.  I think  it  is necessary  to char-
acterize the exposure initially.  I think several people  in  the room will
agree that the air monitoring that is done in  fixed  sites certainly doesn't
account for a lot of population factors  that determine exposures,  even  in
groups with large numbers of people.

     MALINDZAK:  I would like to make the record  clear.   1 support personal
monitors, but for different reasons.  I  support your personal  physiologic
monitors to run concurrently with exposure monitors  so that  the response to
these exposures can be characterized in  a realistic  and scientific manner, so
that we can respond to them in an intelligent  way.

     Right now we don* t collect the physiologic information  concurrently with

     MAGE:  Another reason for the use of physiological monitors  is that we
must, as an agency, keep progressing the state-of-the-art.  We are running
Clinical studies in parallel with epidemiological studies.  We are doing new
and different things.  We are discovering new  things all  the time.

     A colleague of mine told me that there are some 8,000 medical journals
published throughout the world on all aspects  of  medicine.  It is  very  possible
and very probable, that somebody is submitting for publication right now,
in some place in the world, something which will  define a heretofore unobserved
physiological effect.

     Suddenly we are going to have thrust upon us a  request  to replicate that
study or a need to go out and monitor.   If these  personal physiological moni-
tors aren't being developed parallel with all  our thrusts in epidemiology,
in clinical studies, and personal exposure, we might find ourselves again with
our pants down, 3 years behind, simply because we didn't  have  the  foresight to
recognize that although there might not  be some kind of acute  symptom that
we can see from the outside, perhaps some of the  devices  that  George and Matt
developed could see from the inside, and had we spent the money, we might
have had those devices.

     So, we need people who are looking  ahead  in  the same way  that we need
people who are looking today, to keep our feet on the ground.  We  need  these
people who know the state-of-the-art of  the developing technology, who  see
the abilities of micro circuits and also know  the physiology,  and  who can say,
"This is a possibility."  Then, it will  be here when we need it.

     Although it might be impractical at the moment to put a thousand of  these
new devices out with a thousand subjects in a particular health study, we
can* t simply say we don't need it.

     I think we need both long-terra and short-term research.

     I have now heard about eight or nine Government agencies who are working
in this area and not all of them are talking to each other.  We have EPA,
NIOSH, the Departments of the Interior, Commerce, and Energy; all kinds of
agencies.  There should be some kind of coordinated plan.  Perhaps Congressman
Brown's idea, of a bill which would establish a monitoring program, would
provide a mechanism for long-range research as well as short-range development.

     Even though there is no immediate payoff from development c2 personal
physiological monitors in epidemiological studies scheduled for this year,
I think we should make certain people in the Government agencies recognize
that it is something that we definitely need in the future.

     I can't imagine our not needing it in the future, as our ability to see
Inside the human body progresses.

     SOCOL:  I would like to change some of the things that I was going to
say based on what Dave Mage said.

     When you go into personal monitoring and physiological monitoring,  the
effect is going to be different on me—who weighs 130 pounds—than on a man
who weighs 180 to 200 pounds.  It is going to be different on a child and
different on an adult.  It is going to be different by age.  It is going to
be different by sex.

     I feel that personal monitoring is going to become more effective than
fixed-station monitoring, where you don't know what effect you are getting on
what particular individual.  I think that you can achieve more physiological
information by the personal method than by the fixed method.

     I am borrowing from what Dave has said to respond to your remarks.

     WALLACE:  OK, thank you very much.

A New Personal Organic Vapor Monitor
with In Situ Sample  Elution
     Donald W. Gosselink, Ph.D., David L. Braun, Haskell E. Mullins, and
     Sandra T. Rodriguez
     Occupational Health and Safety Products Division
     3M Company
     St. Paul, Minnesota

     In considering a new approach to monitoring personal exposures to organic
vapors, the steps  in an overall monitoring  program can be Individually ad-
     1)  Development of monitoring plan
     2)  Purchase  of supplies
     3)  Preparation of equipment
     4)  Sampling  of personal exposures (includes recording data during sampling)
     5)  Preparation of sample for analysis
     6)  Analysis  and TWA calculation
     7)  Report  communication and follow-through.

     In examining  and improving current methods for carrying out each of the
above operations,  we have evolved a monitoring system which simplifies the
work while maintaining desirable objectives and monitoring requirements.
The monitoring system which evolved is the  3M Brand Organic Vapor Monitor No.
3500 (see Figure 1).

     We can now  consider the way in which this monitor interacts in each of
the above seven  areas.

     Step one, monitoring planning, is simplified because the 3500 system
allows the hygienist  to work through trained aides in remote locations.  It
also helps increase the number of samples taken in the case where the hygienist


FIGURE 1.  3M Brand No. 3500 Organic Vapor Monitor.
FIGURE 2.  Use  instructions  for  the 3M Brand No. 3500 Organic Vapor Monitor.
(The  numbers  in  the  illustration  refer to the steps outlined in the text.)

is present for the monitoring,  thereby  lowering hygienist  cost-per-sample.
Purchase of supplies, step  two,  is  reduced,  since only a single  product  is
needed.  Step three  is virtually  eliminated, because  the 3500 monitor  needs
no preparation or calibration  by  the  user.

     In the conduct  of step four, or  the actual exposure sampling,  the monitor
is readily accepted  by employees  wearing it.  Monitors are  individually  identi-
fied, and the exposure time is  recorded directly on the monitor.  The  package
can be used to record comments  and  other data.  A cap terminates sampling and
hermetically seals the sample.

     Sorbent transfers and  other  analytical  preparations normally comprising
step five are eliminated  because  the  capped  monitor is already a discrete
sample ready for elution.   Step six,  the actual analysis,  is aided  by  in situ
elution of the contaminants from  the  sorbent.  Manual or automatic  gas chro-
matographic analysis is also facilitated by  the monitor design.

     Efforts in preparation of  reports, step seven, are reduced by  having all
the needed information concerning the sampling on the monitor package  for
calculation and reporting.   In  the  next section, the  objective is to techni-
cally describe the 3500 system.

     A pictoral  representation of  the  monitoring  and  analysis method using
the 3M Brand No. 3500  Organic  Vapor  Monitor  is  shown  in Figure 2.  Referring
to the numbered  steps  as  illustrated:
     1)  The monitor  is  removed from the  package.  Exposure start time is re-
corded on the back of  the monitor.
     2)  The monitor  is  attached near  the breathing zone  of the worker.  After
exposure, the exposure end time is  recorded  on  the back of the monitor.
     3)  The white face  of the monitor is removed.
     4)  The elutriation  cap is snapped on,  and  the ports are sealed.
     5)  The sampling  information  is recorded on  a package, and the monitor
is inserted.  The package is sealed  with  the tubular  closure.
     6)  The eluent  is added through the  center  port,  and the port is  imme-
diately resealed.
     7)  After 1/2 hour  (with  occasional  gentle  agitation), the eluent is
decanted into the sampler vial, or  an  aliquot is  removed  from the center port
for manual  injection.
     8)  The gas chromatographic sample analysis  is carried out by conventional

     The time-weighted exposure level is then calculated using the following
               mg/m  = Corrected weight on monitor (nanograms)	
                       Sampling rate (cm /min) x sampling time (mln)

                 ppm - mg_   22.4/mole    T° K      760 mm Hg
                       m3 X m.w.g/mole x 273" K x  Pmm Hg

The corrected weight on the monitor is obtained from the sample analysis after
corrections are made for blank samples and desorption efficiency.  The sam-
pling rate for each contaminant is determined and published by 3M Company, as
discussed in the next section.

Diffusion Equation

     The relationship between environmental contaminant  concentration and the
weight of material collected by  the monitor is given by  Equation 1:

                                    M = lAjt                          [1]

where     M = total mass of contaminant collect (nanograms)
          D = molecular diffusion coefficient (cm /sec)
          A = diffusion path area (cm )
          L = diffusion path length (cm)
                                                           3        3
          C = environmental contaminant concentration  (mg/m  - ng/cm )
          t = sampling time (sec).

     Equation 1 assumes that the contaminant concentration at the  face of the
monitor closely approximates the environmental concentration, that the monitor
wall effects are negligible, and that the collection efficiency of the sorbent
element is unity.

     In normal use, environmental concentration (C) is to be determined,
mass collected (M) is determined via gas chromatography, and the monitoring
time (t) is measured.  Therefore, a value for the D^- term is needed to solve
the equation.  Substitution of the appropriate units for D, A, and L reduces
to cm /sec units, which defines  the sampling rate for  the contaminant and shows

the correspondence  between diffusion-controlled monitors  and  conventional
pump methods.  The  diffusion-controlled monitors  do,  in effect,  sample  the
contaminants in air at  a  useful,  calculable  rate.  The driving  force  for
sampling is the concentration  gradient between the collection face  and  the
sorbent layer  in  the monitor.

Monitor Calibration

     Parameters Calibrated.  Monitor  calibration  data are provided  by 3M
Company for many  organic  vapors of  industrial hygiene importance.   The  items
which are provided  are:
     •  Sampling  rate (D=- from equation)
     •  Recovery  coefficient
     •  Exposure  limit
     •  Environmental effects
     •  Sample storage  recommendations
     •  Concise equations for  calculation of results.

     Calibration  Apparatus.  Monitor  calibration  requires that accurately
known exposures be  made.   Typical equipment  for generating known, time-
weighted, average contaminant  concentrations is shown in Figure 3.

     The compressed air from the  gas  cylinder is  passed through a charcoal
bed to remove any organic material  which may be present.  The air flow  is
integrated by a calibrated flow meter.  The  organic liquid to be studied is
placed in the bottom of the diffusion column, which is divided as shown  by
a porous membrane.   The organic vapor diffuses into the dilution air stream
at a calculable rate, and a wide  range of contaminant concentrations are
attainable by varying diffusion column diameter and length, dilution air
flow rate, or diffusion column temperature.  The  monitor exposure section
is a metal conduit  with a series  of ports into which monitors fit snugly.
The design permits  easy exposure  replication to give good estimates of monitor
precision.  A simple insert makes adaptation to charcoal tubes possible  for
comparison studies.   The  air stream finally enters a large collection chamber
where the relative  humidity is monitored and the  contaminant  concentration
verified with a calibrated hydrocarbon analyzer.  The monitors exposed  in
series each removes a small fraction  of the  contaminant (usually less than
0.5 percent), but the amount removed  is calculable, and corrections are  easily
made.  The dilution air stream can  be humidified  by several means between
the flow meter and  diffusion column.

                                              FIGURE 3  (left).   Monitor calibration ap-
                                              paratus .
FIGURE 4  (right).   Ethyl  acetate
exposure  limit.
                                                          1000       2000      MOO

                                                                EXPOSURE (PPH-HOUftS)
                                                                                    «000      5000
  PRODUCT NO. 3500
 Oualineallon Dili tot
Monitor Sampling Rata        »0 cmVmfti
RKOvtry CoaHlclant         Q.»*a.M _
Upptr Expotura Until1        >1MOppm-Houn
Motoeular Walght          n.11  _
Vapor Prawura (JO*C)        J7.B»niH»Jte
PubOthad OKItnlon ContUnt'    fl.08» eaf/tue
Otiatty            •
T.WJk. (TLV)
1. Ocltnnim ma* at uMlylt In mteroflmni.
2. DittrmlM npowra Urn* In mlnutw.
3. Calculate conctnlrallon from:

mfl  =  30.X (mlerogrmu)    =   30.1
"Si1" (noevtry eMlfletenl) (mbiuSt)     InOnuttt)
ppm  «  MXjmlerearamt)    =   t.UO (naemgnmi
   (raeewHy eotHleltnt) (mbnttet)     (mbuto)
 t. Tht Uppar Eipoturt Limit at MM R.H. It
 t PuWterwd dlMutlon eonaldiU an provUad (or ganaral
  Inttml only. -Monitor Sampllnfl Mf OJU lor taeh
  compound It dttarmlnad by 3M Company.
 9. ppm OMtd on M'C. 760 mm Ho.
                                                        T = Charcoal Tube
                                                        OVM = 3M Organic Vapor Monitor

                                              FIGURE  6  (above).   Sampling  sequence
                                              gasoline  exposure.
FIGURE 5  (above).   3M  Brand Or-
ganic  Vapor Monitor represen-
tative qualification sheet.


     When a substantial  amount  of  liquid  Is  Introduced  Into  the  diffusion
column, the system Is  capable of equilibrating  to generate stable, known  con-
centration environments  for  long periods  of  time.  More  typically, a known
quantity of organic  liquid Is added  into  the diffusion  column and allowed to
evaporate to depletion as indicated  by  the hydrocarbon  analyzer.  Using this
method, an accurate  time-weighted  average contaminant concentration can be
determined by only two measurements; a  gravimetric weight determination of
organic liquid (mg)  and  a calibrated flow meter reading  (m ).  As the organic
liquid evaporates, the concentration varies with time,  but the time-weighted
average concentration  is known  at  the end of the experiment.  Both diffusion-
controlled monitors  and  pump methods are  capable of integrating  variable  con-
centration, so this  factor is of no  consequence.

     Calibration Techniques;

     1)  Sampling rate
The sampling rate of the 3M  Organic  Vapor Monitor is determined  experimentally
by generating known  concentrations of each contaminant,  exposing and analyz-
ing the monitors, and  solving Equation  1  for the D=- term.  The published
sampling rate is generally based on  three separate determinations in the
0.25-2 TLV, 8-hour exposure  range.   This  experimental determination is con-
sidered necessary since  desired diffusion coefficients (1,2) are not always
available, and their accuracy is not usually known.

     2)  Recovery coefficient
The recovery coefficient is  determined  by vapor-state spiking of monitors.
This is accomplished by  removing the white barrier film  of the monitor, plac-
ing a filter paper on  the diffusion  spacer plate, snapping on the elutriation
cap, and injecting a known quantity  of  the organic material onto the filter
paper through the center port.  The  monitor  is  aged 16  to 24 hours before
elution to allow total transfer of the  organic  material  from the filter
paper to the sorbent.  The published number  represents  the mean + 2 standard
deviations from the  tests conducted  by  3M.  It  is, however, recommended that
recovery coefficients  be at  least  verified by the user,  since techniques
and presence of multiple contaminants can affect recovery coefficients.

     3)  Exposure limit
The capacity of the  monitor  for each individual compound is based on molecular
structure, vapor pressure, environmental  conditions, etc.  The exposure limit
is specified in lieu of  a need  for a "back-up"  section.  This is done by
3M by plotting the weight of the contaminant collected  by the monitor versus
exposure.  (See Figure 4.)


     The exposure limit is specified as the weight of contaminant collected
when significant deviation from linear adsorption is noted.  For strongly
adsorbed materials, this limit is usually of the order of 15 mg.

     4)  Environmental effect
The ultimate vapor capacity of the monitor is sometimes affected by relative
humidity.  The exposure limit test described above is repeated at 80 + percent
relative humidity, and deviations due to relative humidity are noted.

     5)  Sample storage
Monitor samples which have received duplicate exposures are stored and ana-
lyzed after 1,7, and 14 days.  If deviations are noted, refrigerated storage
is recommended.

     Overall information and recommendations are published on qualification
sheets.  Figure 5 shows a representative qualification sheet for benzene.

Benzene Monitoring

     Using equipment as pictured in Figure 3, known benzene vapor concen-
trations were generated in air and monitored using the published benzene
sampling rate (33.0 cm /min).  The analytical procedures outlined in the
"Product Description" section were followed.  Table I is a summary of results
comparing the 3M Organic Vapor Monitor to the known generated benzene concen-
trations.  Tabulated values  for the 3M monitor represent the mean of four
separate determinations + 2  standard deviations.  The values fall well with-
in the accuracy requirements contained in the OSHA standard.

Complex Mixture Analysis

     It is recognized that in real field-monitoring situations, combinations
of organic vapor contaminants are typical, and that any feasible monitoring
system must be capable of complex mixture analysis.

     Gasoline was used to test the capability of the 3M Organic Vapor Monitor
to sample complex hydrocarbon mixtures.  It was reasoned that the high boiling
components of gasoline would not be present during a typical exposure.  To
simulate a more likely field situation, a fraction of gasoline was distilled

               TABLE 1
        Benzene Monitoring Test

 3M Monitor
  No. 3500
 (ppm ± 20)
 7.75 ± 0.43
 4.12 ± 0.23
 2.22 ± 0.06
 1.23 ±0.04
 0.91 ± 0.06
               TABLE 2
    Results of Gasoline Exposure Test
3M No. 3500
(PPM ±20)
5.832 1.408
0370 t .045
0.307 i .047
0.892 t .160
1.710 t .119
1.230 ±.086
Charcoal Tube
(PPM i 2o)
5.053 t 300
0.578 i .082
0.329 t .056
0.873 i .180
1.S52 t .220
1.034 i .160
               TABLE 3
  Integration of Variable Concentration
„. JL__-
„. ill
ItHt '<
IHT ; i>r
Mm* in

from a bulk sample.  The  fraction  used was  the  first 80  percent  of  distillate
collected.  The  remaining 20 percent was  discarded.  A syringe was  used  to
inject 1.24529 grams of the light  fraction  through  a septum  capped  port  into a
receiving vial.  An air flow rate  of 9.1924 liters-per-minute was allowed to
pass over the open vial until  all  the gasoline  had  evaporated into  the air
stream.  At the  conclusion of  the  test, the total air  volume was determined by
the change in readings of a Model  175-S Rockwell flow  meter.  A  total of 11.581
cubic meters of  air had passed during the 21 hours  required  for  the gasoline
distillate to evaporate.  A flame  ionization detector  was  used to verify the
test end-point.  Charcoal tubes were used to collect samples for comparison and
correlation of test data.  The tubes and  3M monitors were  randomly  selected but
placed in a set  pattern on a sampling channel.  Two charcoal tubes  were  placed
side-by-side in  the first sampling port,  followed by four  3M monitors in ports
two through five.  Port six contained two more  charcoal  tubes, again placed
side-by-side (see Figure  6).

     The charcoal tubes were positioned vertically  to  prevent channeling, and
the 3M monitors  were positioned so that the air stream would be  flowing  parallel
to the face of the monitor at  approximately 100 fpm air  velocity.   One and one-
half milliliters of spectroquality carbon disulfide were used to desorb  the
collected gasoline vapors.  The samples were allowed to  set  for  30  minutes with
occasional agitation, and were then transferred to  a vial.   Withdrawal of a 2
microliter aliquot from the sampling vial,  and  injection into a  Hewlett  Packard
5840A gas chromatograph,  was accomplished by using  a Hewlett Packard automatic
sampler (Model 7671A).  The analytical column was a 7 .5  ft x 1/8 in S.S. tube
packed with 15 percent Carbowax 20M on 80/100 mesh  Chromosorb W. A temperature
program was used, starting at  70°  C for 5 minutes and  increasing to 125° C at
the rate of 4° C per minute.   A hold of 8 minutes was  used on the final  temper-
ature.  Helium (30 cc/minute)  was  used as the carrier  gas.

     Quantification of pentane, heptane,  octane, benzene,  toluene,  and xylenes
was accomplished by preparing  four calibration  standards for each component.
Each standard was individually prepared by  injecting a known volume of each
component into a known volume  of carbon disulfide.  The  four data points were
used to establish a best  fit line  from the  linear regression method. The
weight of each component  was determined by  comparing the peak area  of the sample
to the calibration line.  This weight was corrected for  desorption  efficiency,
and the time-weighted exposure level was  calculated in milligrams-per—cubic-
meter and converted to ppm.  Figure 7 shows a replication  of the chroraato-
graphic analysis made of  the parent, 3M monitor, and charcoal tube  samples.

                                                 FIGURE 7.   Chromatographic anal-
                                                 ysis gasoline exposure.



,- —..BAFFLES — —
o O
o O
o O
-» /"N
o O
o O
o O
/ \


Outlet •

                                                 FIGURE 8.  Monitor exposure

                                                 chamber air  velocity  study.
« «i
i • •
I .
§ -10 ,
§ -11

• *





                  100       110
                AIR VELOCITY (FT/MM)
                                           FIGURE 9.   Air velocity depen-
                                           dence (parallel-to-face) .

     The chromatographs are  identical except  for  peak amplitude, which  illus-
trates that both monitoring  techniques  sample each vapor as  if  it were  pre-
sent Individually.
     The desorption efficiencies  for components of gasoline  were determined
in the presence of one another.   The recovery tests were done on both the
charcoal tubes and the 3M monitors.  A  known  amount of  each  component was
injected onto 100 mg  sections  of  the charcoal tubes or  onto  a piece of  Whatman
filter paper 2.5 cm in diameter in the  3M monitor, as described earlier.

     The analytical results  of the charcoal tubes and 3M monitors agree very
well with each other, and with the actual test concentration (Table 2).

     In all cases, the 3M monitors gave lower standard  deviations and better
agreement with the challenge environment.  The concentration of the selected
components in gasoline was determined by injecting known volumes of the
light gasoline fraction into known volumes of carbon disulfide  and analyzing
in the same manner as the monitor samples.

Integration of Concentration Spikes

     To verify the ability of  the No. 3500 Organic Vapor Monitor to integrate
high peak concentrations, several test  exposures  of 1,  3-butadiene were made
using an exposure system as  described in Figure 3, except  that  an injection
septum was placed in  series  between the flow  meter and  exposure channel.
In each case, charcoal tubes were used  to collect samples  for comparison and
correlation of test data.  During sampling, a randomly  selected charcoal tube
was positioned vertically and  upstream  of a 3M Organic  Vapor Monitor.   The
3M monitor was positioned in such a way that  the  air flow  was parallel  to the
face of the monitor.  The charcoal tubes had  a sampling rate of 52 cm -per-
minute.  The sampling rate of  the 3M monitor  is 41.3 cm -per-mlnute for the
collection of 1, 3-butadiene.

     Three different  concentration levels were used to  challenge the tubes
and 3M monitors, and  each level was replicated three times.  The first  con-
centration used was a single spike generated  by injecting  5  ml  of butadiene
vapor into the air stream at a flow rate of 9.774 liters-per-minute, thus
giving a TWA of 528 parts-per-million as determined by  the total air volume
(flow meter) and the  weight  of 1, 3-butadiene injected.  The next concentration
(1,585 ppm) was generated by three injections of  5 ml each.  The last test
concentration (3,170  ppm) was  produced  by three injections of 10 ml each.
The spike generations were monitored by a total hydrocarbon  analyzer equipped

with a flame ionlzatlon detector.  The signal output from  the detector  was
connected to a millivolt recorder  to give a chromatogram of each  injection.
The sampling probe  for the  total hydrocarbon analyzer was  placed  downstream
from the charcoal tubes and 3M monitors such that no interferences would  be
caused during the sampling.  It was noticed from the recorder chart  that  each
spike was generated  in approximately 6 seconds.  This means that  the actual
peak concentration was in the range of half a percent (5,000 ppm) for 5-ml
injections, and 1 percent (10,000  ppm) for 10-ml injections.  The results
are shown in Table 3.

     Clearly, both  the 3M monitors and the charcoal tubes with pumps effectively
integrated the abrupt concentration changes to give a good estimate of  the
time-weighted average concentration.

Air Velocity Effects

     Diffusion-controlled environmental monitors require a slight air move-
ment across or impinging upon the  face of the monitor.  Use of Equation 1
assumes that the contaminant concentration at the face of  the monitor is rep-
resentative of the overall  environment.  In stagnant air,  the contaminant
is removed from the  atmosphere by  the monitor.  This results in a depleted
zone at the face of  the monitor and an underestimation of  the true contaminant
concentration.  Previous work (3)  studied air velocities above 50 fpm.

     Experiments were conducted to determine the air velocity requirements for
the 3M No. 3500 Organic Vapor Monitor.  The exposure apparatus shown in Figure
3 was employed, but  the monitor exposure channel was designed as shown in
Figure 8.

     The air velocity in the exposure channel was controlled by adjusting air
flow rate.

     Air velocity was calculated from the flow rate and the exposure channel
cross-sectional area.  Uniformity  of air velocity was also measured using
an anemometer probe  in the  ports indicated.  Several exposures to known
benzene concentrations were made,  and the sampling rate was determined.  A
total of six monitors was exposed  at each air velocity level.  Elution and
analysis were done  following standard procedures outlined  previously.  Figure
9 is a plot of the  results.

     The plot shows  a significant  drop in benzene sampling rate below 10 ft/rain

face velocity.  The + 2 standard  deviation  error  bars  indicate  considerable
variability at low air velocity.  This  reflects the  difficulty  in  achieving
reproducible, uniform, low velocity  in  the  exposure  channel.  The  mean values
do, however, conform well to  the  published  sampling  rate  for  benzene  at  the
higher velocities, and the results are  believed to be  valid.  It should  be
noted that this experiment tested only  parallel-to-face air velocity  effects,
which is  the worst case situation.

     As a result  of these data, it is believed that  the air velocity  require-
ments of  the 3M Organic Vapor Monitor are minimal and  of  no concern for  per-
sonal monitoring.  When used  as an area sampler,  however,  placement of the
monitor should be considered  so that stagnant air areas are avoided.  This
situation could potentially exist against walls,  in  corners,  etc.

Temperature and Pressure Effects

     Both temperature and pressure affect the diffusion rate  of gaseous  con-
taminants in air.  This could, in principle, lead to substantial monitoring
errors using diffusion-controlled monitors.  However,  the net effects of tem-
perature  and pressure are offset  to  some degree- by corresponding changes in
contaminant concentration (4).  In the  case of atmospheric pressure,  the
change in diffusion rate and  concentration  offsets each other exactly, and no
correction is required.  The  effect  of  temperature on  diffusion coefficient
and concentration is given by Equations 2 and 3.
                                  D  -  ttT"/*;                            [2]

                                     C  -
                            Weight  collected -

     The  net  temperature  effect  can be  compensated,  if  desired,  by increasing
the  experimentally  determined  concentration in mg/m   units  by 1.08 percent
for  every 10° F above 70° F (and reducing  by 1.08  percent for every 10°  F
below 70° F).

 1.  Gilliland, E.R.  Diffusion Coefficients  in Gaseous  Systems.   Ind  and
     Eng Chemistry, June 1934.


 2.  Nelson, G.O.  Controlled Test Atmospheres.  Ann Arbor Sci Publ, 1971.

 3.  MeGammon, C.S., Woodfin, J.W.  An Evaluation of a Passive Monitor for
     Mercury Vapor.  Am Ind Hyg Assoc J 38:378, 1977.

 4.  Palmes, E.D.  Personal Sampler for Nitrogen Dixoide.  Pres Am Ind Hyg
     Conf, Minneapolis, Minn., June 1975.

Donald Gosselink, Ph.D.
Occupational Health and Safety Products Division
3M Company
St. Paul, Minnesota  55101
     PETROVICK:  Matt Petrovick, EPA.  I think you have a very nice little
exposure meter.  The size particularly appeals to me from the standpoint
of human mobility.  1 think it is something that is easy and nice to wear,
and light.

     It is the kind of thing that makes room for physiologic data acquisition
systems which may be bulky.  And, together, I think they fit nicely.  That's
one of the prime advantages that I think you have.

     The other thought that occurs to me is this:  Has there been any plan-
ning or any designing in an attempt to try and convert some of the measured
exposure levels to an electrical signal—for example, such that you may be
able to look at the integrated accumulative effects of these pollutants?
In other words, the transfer function from the electro-organic chemistry
level to some measurable signal that might be able to be placed into a micro-
computer or memory, so that if the level is exceeded, an alarm can be issued
so that the person who is exposed might be made aware of it?  Is there any-
thing like that on the horizon?

     BRAUN:  I am an engineer at 3M.  Sometimes I can understand what our
people in Central Research are doing, and I wouldn't be a bit surprised if
they are, but my own work has been mostly in the sampling and the diffusional
aspects.  It could be done.  I think you conveyed some very Interesting

     PETROVICK:  I have a very strong feeling that that's the way to go if
you want to alarm a person.  It is no use to send the device to the lab, and,
you know, 4 days later find out he was endangered.

      BRAUN:  That's a good point.

      LOCKER:  Larry Locker of Solid State Sensors.  I think that's an ele-
 gant way, but you look at the compounds like mercury, benzene, trichloroethyl-
 ene—I think the long-term effects of low levels are what are of concern.
 I think that the idea here is that you can monitor those for an extremely
 low cost.  And usually 4 days is not so bad to know that someone was exposed
 to 10 or 20 parts-per—million of benzene*  He's not going to die on the spot.

      BRAUN:  I would like to just comment, too, that we are really building
 a record of safe operations, and that's a good thing to be doing; it really

      PETROVICK:  But to respond to your comment:  What about the person work-
 ing in a factory who has been there 10 years, and he has an accumulative re-
 sidual of this chemical, whatever it might be, and one of these days, he
 crosses his tolerance threshold, and it is that day he might sicken or drop dead?

      LOCKER:  That's why he should have periodic monitoring, according to the
 N10SH regulations.

      PETROVICK:  That's right.  But, wouldn't it be nice for him to have some
 early warning capability, in terms of something he can hear, saying some-
 thing like:  "Hey, get the heck out of the room"?  That's the kind of ration-
 ale I had in mind.

      JENSEN:  Janice Jensen, with the EPA in Washington.  What kind of in-
 terferences are there with a benzene monitor?

      BRAUN:  Well, you see—the analysis is by GC or GC mass spectrometry
 or whatever method you utilize; because the monitor, up to the limits we are
 talking about, would just collect most organics.  We haven't worked with all
 of them, but, you see, the answer is that the interferences, if there are
 any, are resolved by the analytical instrument.

      DEBBRECHT:  Fred Debbrecht, AID.  I didn't quite understand the calibra-
 tion system that you used.  Are you counting on that benzene evaporating

      BRAUN:  No, we don't care.  When you inject the benzene through the
 septum, you are going to get a rise and then a gradual drop off.  The area
 under the curve that the monitor integrates is milligrams-per-meter cubed.
 And we don't care if the curve is level or not.

      DEBBRECHT:  Is the system reasonably similar to charcoal tubes?  I assume
 you have got some adsorbent there?

      BRAUN:  Yes.

      DEBBRECHT:  Certainly the efficiencies that charcoal tubes work at are
 subject to the concentration of material and amount adsorbed, as opposed to an
 integrated effect.  They are not 100 percent efficient in capturing the pol-


     BRAUN:  As a matter of  fact, the concentration profile is quite  flat  up
until you enter the device—we know about where it is within 10 percent,
so that we can Just operate  reasonably.  The main concept there is that  it
is a simple, repeatable way  to give a known time-weighted average; that's  all
I am really saying*

     HODGESON:  The main question about  the calibration system is are you
collecting all organic data  with your exposure?

     BRAUN:  No; not at all.

     HODGESON:  How do you know how much; what's the efficiency?

     BRAUN:  Well, it is moving past the device, and that's just like the
configuration we use in the  environment.  In fact, about 1 percent is all we

     HODGESON:  One other question.  You didn* t say much about the actual
construction of the device.  Is it diffusing through a membrane?

     BRAUN:  Yes, and then through a static layer of air.

     GOLD:  I was curious to know whether you looked at the effect of water
vapor and temperature?

     BRAUN:  Yes; part of that qualification procedure is that we analyze
results at both high and low humidities.  Usually the effect of water vapor
is to decrease the range of  utility of the monitor by a little bit.  Yes, we
do that.

     GOLD:  How about temperature?

     BRAUN:  Over the temperature range  of normal human existence, this ef-
fect is negligible—at least for the compounds we are working on.

     BREEN:  I am not sure if I understood—do you have any data on parts-
per—billion exposures?

     BRAUN:  No.  The data I showed were at the 450 parts-per-billion level
corresponding to occupational exposures.

     WHITE:  Are you limited in terms of the sensitivity of the GC mass

     BRAUN:  That's correct.

     WHITE:  If you improved the sampling efficiency, as we were discussing
with Ed Palmes, you might be able then to go to the ambient exposure levels.

     BRAUN:  Yes, and then you can make  it bigger, like a big sunflower.

     BREEN:  Well, isn't it  part of the  problem at the lower levels that you
may get quantitative desorption?


      BRAUN:   We are not sure whether that occurs.  That's why we would like
 to look into it;  absolutely.  I would like to work with any interested
 people at this conference to develop this device for use with daily ambient

A Combination Sorbent System for Broad
Range Organic  Sampling in Air
     Joseph J. Brooks, Ph.D., Diana S. West, and Donald J. David
     Monsanto Research Corporation
     Dayton, Ohio
     James D. Mulik
     Organic Pollutant Analysis Branch
     Environmental Sciences Research Laboratory
     U.S. Environmental Protection Agency
     Research Triangle Park, North Carolina

     The use of solid sorbent materials as sampling media  is rapidly becoming
the method of choice for the collection and preconcentration of organic
pollutants in air.  Unfortunately,  no  single sorbent material is available
that is universally applicable to the  broad range of organic compounds that
may be present in environmental and workplace samples.   For this reason, ad-
ditional sorbent materials are needed  to supplement the sorbent character-
istics of the more commonly used materials (e.g., Tenax-GC and activated char—
coals) in applications requiring broad range sampling efficiency.

     In light of this need, a program  has been initiated to evaluate the
sorbent characteristics of some commercially available sorbent materials
with the aim of finding a suitable  combination of sorbents that is capable
of efficiently collecting compounds representing a broad range of volatilities
and polarities.  The ultimate objective of this EPA-sponsored program (Con-
tract No. 68-02-2774) is to develop a  portable, sorbent-based collection
system that is capable of sampling  a wide variety of carcinogenic vapors in
ambient atmospheres.

     This paper describes the evaluation and selection of  the sorbent materials
and suggests a technique for estimating the applicability  of sorbent materials
for sampling a particular compound  based on certain compound properties.
The anticipated performance characteristics of the selected combination sorbent

system are also discussed.


Selection of Candidate Sorbents

     Before initiating studies to evaluate and select sorbent materials, a
comprehensive review was conducted of available literature on the use of solid
sorbent materials for sampling organic vapors.  Information was collected on
specific sampling applications for over 110 compounds using more than 30
different sorbent materials.  This survey indicated three major classes of
sorbent materials:  1) porous polymers (e.g., Tenax-GC, Porapaks, Chromosorbs,
XADs); 2) carbonaceous materials (e.g., activated carbons and charcoals,
graphitized carbon black, Ambersorbs); and 3) others (e.g., molecular sieves,
silica gel, liquid-coated solid supports).

     Generally, the porous polymers were found to have the most desirable
properties for air sampling, including low background and low reactivity,
as well as high capacities for many compounds.  Unfortunately, the porous
polymers were found to have little capacity for the more volatile compounds.
On the other hand, the carbonaceous materials were noted to have much better
capacities for volatile compounds, but are plagued with reactivity problems
and problems associated with hydrophilicity.  The other sorbent materials
generally had intermediate capacities and various associated sampling problems,
and they have not been widely utilized in air sampling applications.  Specific
candidates were selected from these three groups of sorbent materials.

     The goal of this selection was to choose a collection of commercially
available materials that, when used in an appropriate combination, showed
promise for quantitatively trapping compounds varying widely in volatilities
and polarities.  To do this, it was necessary to compare the trapping capa-
bilities (capacities) of a sorbent for various compounds with parameters that
are indicative of volatility and polarity.

     Consistent trends in sorbent capacities were seen when related to com-
pound volatilities, as indicated by either vapor pressure or boiling point,
with boiling point being the more convenient parameter.  Unfortunately, find-
ing a single polarity parameter that resulted in consistent sorbent capacity
trends was not as easy.

     After considerable effort, an empirically derived relative polarity
ranking of compound functionalities was arrived at based on a weighted  com-
bination of polarity parameters,  including dipole moment, solubility parameter,
hydrogen bonding index, and  polarity index.  The order of polarity obtained
by this scheme was hydrocarbons < halogenated compounds < aldehydes w ethers «
esters « ketones < nitro compounds < nitriles « strong acids < amines <
alcohols « phosphates < water.  This functionality ranking was used in  plots
of boiling point versus relative  polarity for a variety of compounds, and
consistent trends were observed for most of the available sorbent capacity
data when superimposed on  these plots.

     On the basis of these plots  and related capacity information, Tenax-GC,
Forapak N, Forapak R, Ambersorb XE-340, and SKC activated charcoal were se-
lected for further evaluation.

     These five sorbents represent three porous polymers and two carbonaceous
materials that have a variety of  individual properties and the potential,
in some combinations, to effectively sample compounds of a wide range of
polarities and volatilities.  Tenax-GC was selected for the collection of
moderately high to high boiling compounds, since it has a high temperature
limit.  Tenax also has a slight preference for nonhydrogen-bonding compounds.
For moderately boiling compounds, the literature showed Porapak R to have the
greatest capacity, although  retention time data indicated that Porapak N
(for which there were no capacity data) might actually have greater capacity.
These two Porapaks also show a preference for polar compounds.  For highly
volatile compounds, a conventional activated charcoal (SKC coconut) was chosen
for evaluation, since activated carbons have traditionally been used in such
sampling applications.  However,  a new sorbent material, Ambersorb XE-340
(Rohm and Haas) was also included since it showed promising capacities for
highly volatile compounds  (e.g.,  81 1/g for butane [3]) without some of the
undesirable characteristics  of charcoals (e.g., hydrophilicity).

Selection of Test Compounds

     In order to evaluate  the selected sorbent materials for their sampling
capabilities, it was necessary to establish a matrix of test compounds that
sufficiently describe the  ranges  of volatilities and polarities that could
reasonably be anticipated.  Entries were sought for a matrix described by
low, medium, and high polarity in one dimension and low, medium, and high
volatility in the other.   The ranges were established on the basis of relative
boiling points for volatility and compound functionality, as previously

                                      TABLE 1

                               Test  Compound Matrix
            Hydrocarbons and
Halogenated compounds,
  aldehydes, ethers,
  ketones and esters
Nitro compounds, nitriles,
amines, and strong acids,
 alcohols and phosphates
n-Hexadecane [1]
Phenanthrene [10]
M6dium NaphSile^e [13]
„. . n-Butane [7]
Benzene [16]
Hexachloro-l,3-butadiene [2]
4-Bronvodiphenylether [11]
bis-(2-Chloroethyl) ether [5]
1 , 2 , 4-Tr ichlorobenzene [14]
Propylene oxide [8]
Benzyl chloride [17]
Succinonitrile [3]

                                                      RETENTION TIME
FIGURE 1.  Elutlon analysis method for sorbent capacity determinations.
detector (FID).  Sample  sizes were within  the Henry's Law region such that
retention times were not  affected by variations.  The data obtained included
the retention time (RT),  which was considered to be the amount of time from
injection to peak maximum, and the first elution time (FET), which was the
                                    TABLE 2
                       Retention Times and Capacities for
                      Selected Test Compounds on Porapak R

tur« (*C) n-Butan*
RT "
50*Vq "
222 PR-
'"v, '
RT •
182 50*Vg "
PR -

RT •
150 50%Vg "
PR -

RT •
100 M*Vg *
PR •

0.17 min
8.22 x 10*
0.04 Bin
1.94 x 10*
0.26 min
12.6 x 10*
0.10 min
4.84 x 10*
0.50 min
24.2 x 10*
0.30 Bin
14.9 x 10*
2.11 Bin
102 x 10*
1.6 Bin








Propylene oxide
RT -
50%Vg "
RT -
PR -
RT •
5Wv, •
PR •

RT •

PR -
77.4 x 10» t/g FRy •
0.20 min
9.68 x 10* t/g
0.05 min
2.42 X 10* t/g
0.33 min
16.0 x 10* t/g
0.15 min
7.26 x 10* t/g
0.68 min
32.9 x 10* t/g
0.40 min
19.4 x 10* t/g
3.49 Bin
169 x 10* t/g
2.9 min
140 x 10* t/g


RT -
50%Vg '
RT -
50%v, '
PR •
RT •
M%Vg "
PR •
RT «
50%V, "
PR •
mvg "
0.24 min
11.6 x 10*
0.12 min
5.81 x 10*
0.46 min
22.3 x 10*
0.35 min
16.9 x 10*
1.00 min
48.4 x 10*
0.8 min
38.7 x 10*
6.35 min
307 x 10*
5.4 min
261 x 10*









RT «
50»Vg '
RT "
0.85 min
41.1 x 10*
0.40 min
19.4 x 10*
2.67 min



50% - 129 x 10* 1/g
RT •
2.0 min
96.8 x 10*
10.33 min


50«y > 500 X 10* t/g
PR »
8.45 Bin

PR • 409 X 10* t/g
RT •
*o«v9 •
PR «
150 Bin
7,260 x 10*
125 Bin
6,050 x 10*



RT - 0.47 min
SOly - 22.7 x 10* t/g
PR - 0.25 min
PR • 12.1 x 10* t/g
RT » 1.12 min
50^ • 54.2 x 10* t/g
PR • 0.80 min
FBTy - 38.7 x 10* t/g
RT - 3.23 Bin
50*y • 156 x 10* t/g
PR - 2.60 Bin
PRy - 126 X 10* l/g
RT - 28.6 Bin
50^ - 1,380 x 10* t/g
PR • 23.5 min
mv - 1,140 x 10* t/g















        FIGURE 2.  Plot  of  log  50%v
        versus 1/T for selected test
        compounds on Porapak R.





                                                      _  LO







FIGURE 3.   Plot of log FETy
versus 1/T for selected test
compounds  on Porapak  R*





                                               ' 3.0







                                                                                            BENZYL CHLORIDE
                                                                                    FIGURE  4.   Plot of log
                                                                                    versus  1/T  for selected test
                                                                                    compounds  on Porapak N.

                                    TABLE 3
                       Volumetric Capacities for Selected
                      Test Compounds on Porapak R at 20° C
                       COMPOUND NAME          FET   (1/g)
                       n-Butane                   4.93
                       Propylene oxide           12.4
                       Acrylonitrile             16.8
                       iso-Octane             3,550
                       Benzene                  175
time from injection  to peak initiation.  These points are indicated in Figure
1.  By multiplying the retention times and first elution times by the flow
rate, and dividing by the amount of sorbent within the chromatographlc column,
the 50 percent (50%  ) and FET (FET_ ) volumetric capacities, respectively,
were obtained.  Sample injections were then made at a variety of column
temperatures to obtain similar data.

     An example of the experimental information obtained with Porapak R is
given in Table 2.  These data were used to obtain an Arrhenius plot (log Vg
versus 1/T).  Figures 2 and 3 are examples of these plots for the 50%   and
FET   values, respectively, for Porapak R.  The FET..  values were then ex-
trapolated  to 20° C  to obtain the approximate ambient volumetric capacities,
FET.. ,2o° c*t of  this sorbent for the  test compounds.  Table 3 gives this
information for Porapak R.

     The volumetric  capacities, for as many of the test matrix and low molec-
ular weight aliphatic compounds as possible, were determined for the other
sorbent materials in a similar manner  from the plots shown in Figures 4, 5, 6,
and 7.  They are  compiled in Table 4.

     The volumetric  capacities (1/g) determined in these evaluations estimate
the volume  of air that can be sampled  per gram of sorbent materials before
sample breakthrough  (i.e., exceeding capacity).  This was determined to be
the most applicable  measure of capacity for our studies since we wish to
sample quantitatively (i.e., no breakthrough).

     Methods (2)  are available for determining saturation (or weight) ca-
pacities, but for this project it was  felt that sampling techniques, such as

                              low Poumirv
                            A MEDIUM POLARITY
                            O HIGH POLARITY
                    aw  j.o
                        im°m i vf





g* 2.0





                                                       METHANE •
                  l/Tt°KI 1103










       FIGURE  5.  Plot of log FETy
       versus  1/T for all test  compounds
       on Tenax-GC.   Numbers correspond
       to bracketed  numbers in  Table 1.
FIGURE 6.  Plot of  log FETy
versus 1/T for standard  gases  on
Ambersorb XE-340.
FIGURE 7.   Plot of  log FETy
versus 1/T for standard  gases
on  SKC charcoal.

                                                     TABLE  4
                          Compilation of  Capacity and Solubility Parameter Data




Propylene oxide
Benzene '
1 BO- Octane
Bthylene glycol
Benzyl chloride
Bis- (2-chloroethyl)
1,2, 4-Sr iohloro-
Hexachloro-1 , 3-












1.04 x 10"
1.02 x 10"

1.06 x 10"
2.9 x 10"
1.14 -x 10"
1.9 x 10«
1.3 x 10°

1.57 x 10"

2.4 x 10«

1.15 x 10"












Porapak N
3.9 0.591
17.6 1.246
69 1.839
239 2.378
2900 3.462
3 x I01ab 12.477b
6.6 x 10s 5.820

2.8 x 10s 6.443



— —
Ambersorb XE-340
FBTVg(20«C) 10*
U/q) FETVq
0.009 -2.046
0.229 -0.640
5.33 0.727
106 2.025
7500 3.875
4 x 10s 5.602




- -
SKC charcoal
FBTVg(20'C) log
0.047 -1.328
2.40 0.380
162 2.210
12 x 10" 4.079
2 x 10s 5.301
_ _




Potentially large experimental error.

the use of backup  tubes  and varying  sampling  volumes,  would be an appropri-
ate manner of  obtaining  valid  samples  in the  event  of  highly concentrated

     As it is  a  long  and arduous  task  to determine  the capacities of a large
number of compounds on several sorbent materials by the method just described,
it was thought that a means of correlating  sorbent  capacities with compound
properties would be highly desirable.   Such a means of correlation would per-
mit the estimation of the FET   ..-„  .  of any compound on any sorbent by
knowing the value  of  the particular  property  or properties, and how this prop-
erty is related  to sorbent capacity.   Then, if a change or addition should
occur in the list  of  compounds of interest, one could  state whether the new
compound is probably, possibly, or not likely to be quantitatively trapped
by a particular  sorbent  under  certain  conditions.  If  sorbent materials were
used in series for sample collection,  one would also have an idea of on which
sorbent the majority  of  a certain compound  is likely to appear.  The general
advantage of a means  of  correlating  sorbent capacity and compound properties
would be to eliminate the necessity  of experimentally  determining the capacity
of every compound  on  every sorbent.

     The forces  that  determine the interaction of an organic compound with
other materials  (neglecting vapor-to-liquid phenomena  of nucleation and growth)
are:  E., the  dispersion or London forces;  E  , the  permanent dipole/dipole
forces; and E    the hydrogen-bonding forces.   These forces are additive, and
the total interaction energy may  be  expressed as E  « E . 4- E  + E, .  Dividing
both sides by  the  nuclear volume  of  the material, one  obtains:
                                     E, + E  + E.
Since                    6 + \\i]  - cohesive energy density

where  6  is  the  solubility parameter,  then
     The  solubility parameter has been well described in the literature (1).
Many experimental  methods  exist to determine $ for a compound,  and many more
methods exist  for  estimating  6 from compound properties such as surface tension.
A very useful  method of  estimating  6  from the structural formula, density,
and molecular  weight of  a  compound has been developed by Small  (4).

     Accordingly,  6^,  6    6^,  and  6 were examined for a number of compounds.
No correlation of  the  separate or  total forces with the FET  (2QO  . was  found,
One would expect a correlation with the dominant force factor, if one should
exist, and a correlation with  6 would seem even more reasonable.  However,
of two compounds with  similar  6 values, the compound with the higher molecular
weight (or size) would be  expected to have a larger FET  (2QO  . on a given
sorbent due to its lesser  volatility.  Therefore, various molecular weight
adjustments of 6 were  tried.   The  simple expedient of multiplying 6 by the
molecular weight gives a relative  size dependent factor, 6  - 6 x MW, which
has the units of (g  x cal)^/(cm  x mole) .  The molecular weight modified
solubility parameters  ( fi^) for the test matrix and low molecular weight ali-
phatic compounds,  both from literature and calculated (2) solubility parameter
values, are given  in Table 4 along with sorbent capacity data.  The data  for
«  are plotted versus  the  log  FETVg/2o° c) for SKC charcoal» Amb' rsorb XE-340,
Porapak N, and Tenax-GC  in Figure 8, and for Porapaks R and N in Figure 9.
The correlation  coefficients  for the "best" lines in these figures are given
in Table 5.

     Note that the most  divergent results occur for a few halogenated com-
pounds on Tenax-GC.  Unusual  behavior of such compounds on Tenax-GC had been
noted previously during  capacity studies.  It was postulated that some sort
of solubility effect was possible occurring, since Tenax-GC has a solubility
parameter similar to that of  halogenated compounds and is known to be soluble
in such halogenated solvents  as chloroform, carbon tetrachloride, and methy-
lene chloride.

Desorption Properties

     A number of experiments  were conducted to determine how amenable the
candidate sorbents are to thermal desorption.  These experiments were of a
preliminary nature, since desorption properties are compound dependent, and
the test matrix  compounds are not necessarily the compounds that will be of
interest in future field studies.  The preliminary experiments that were con-
ducted attempted to evaluate  the sorbents for efficient compound desorption
and extraneous background due to contamination and/or sorbent characteristics.

     For the desorption  efficiency experiments, 6-in long, 1/4 in O.D., 4 mm
I.D. glass sampling tubes were packed with the sorbent materials listed in
Table 6 and evaluated on the  desorption/analytical system depicted in Figure
10.  Note that the Chromalytics Concentrator® oven contains a six-port, two-
position valve which is  sketched in Figure 11.  This valve offers the alter-

                                                            o TENAX
                                                            O AMBERSORB XE-340
                                                            a SKC ACTIVATED CHARCOAL
                                                                m FROM LIT 5

                                                            * HALOGENATED
                                                            * POTENTIALLY LARGE
                                                               EXPERIMENTAL ERROR
 FIGURE 8.   Correlation  of 6  and log FET_, ,«ft0 «\  for  test compounds on  sor-
                               m              VgllU  UJ
 bent materials.   Entries correspond to  key in Table 4.
                                                             * POTENTIALLY LARGE
                                                                EXPERIMENTAL ERROR
 FIGURE 9.   Correlation  of «B and  log FETy ,20«  c»  for  selected  test compounds
 on Porapaks R and N.  Entries correspond to key in Table 4.

                                    TABLE 5
              Correlation Coefficients  for the  "Best" Lines of
versus Log FET
                                         Plots (Figures 8 and 9)
                                    FETVg(20° C)
                              CORRELATION COEFFICIENT
         SKC charcoal
         Ambersorb XE-340
         Porapak N
         Porapak R
         Tenax-GC (excluding
           halogenated  compounds)

native of direct syringe  injection of a standard or thermal desorption of
a sampling tube.  Standards were  prepared with a 1,000 pg/ml concentration
in acetone and stored  in  septum-capped vials.  Chromatographic conditions
were established by direct injections of standards using a 7 ft, 1/4 in O.D.,
2 mm I.D. glass analytical column containing 1.125 g of Tenax, which was
conditioned at 320° C  overnight.  For statistical determination, replicate,
direct injections (
                                              TABLE  6

        Sampling  Tube Data (All  Sorbent  Filled Tubes  Have Glass  Wool  Plugs)
Sorbent material
Tenax (60/80 mesh)
Tenax (60/80 mesh)
Tenax (60/80 mesh)
Porapak R (50/80 mesh)
Porapak R (50/80 mesh)
Porapak N (50/80 mesh)
Porapak N (50/80 mesh)
SKC activated charcoal
SKC activated charcoal
SKC activated charcoal
Ambersorb XE-340
Ambersorb XE-340
Ambersorb XE-340
Ambersorb XE-340
extraction information
Overnight with MeOH
Overnight with each
MeOH, EtOAc and pcntane
Overnight with MeOH
Overnight with MeOH
Overnight with each
MeOH and distilled H30
Amount of Manufacturer's
sorbent temperature Conditioning Desorption
material limit temperature temperature
(g) ("C) (»c) (*C>
400 325
400 325
400 325
250 175 and 240
250 175 and 240
190 175
190 175
                                              TABLE  7

                        Chromatographic  Conditions and Raw Data  for
                        the  Determination of  Desorption Efficiencies
 Avaraaa  Avaraga «r««           Com).
 W Of    COIt/Ill    SaBpHna    taBp.
atandarda for atandard	tuba	(»C)
taap.       Ml
I'd  Trial »td.
 Araa   affielaney
oounta    {%)

TaBparatura PrograBBad
froB SO'C t 2-C/Bln
for 10 Bin than t
!~" 32»C/Bln to a final
octMn* 250*C taaparatura
and hald thare for
alght Blnutaa

TeBparature PrograBBed
frOB 150'C to 250*C
8 8*C/Bln and hald
temperature for
4 Binutaa

TeBp. Prog. froB 200 -C
Phananthrana to 100*c ( 12'C/Bin and
hald 4 Binutaa
TaBparetura PrograBBed
n-Hexadacana fr~ '^.nS n!?d%
2SO*C for 4 Blnutaa

2.98 15694

(Ha carrier)
1.14 9376

3.32 14713

10.85 346086

11.33 350729

11.30 341814

7.72 205658

9.00 291098

8.95 292206

Tenax 17 300 300

Porapak H 235 150

Porapak N 175 150

Tanax 17 300 300

Porapak R> 235 220
Porapak N* 175 150

Tenax 17 300 300

Tanax 17 100 300











>30 Bin

>30 Bin



_ _

_ _





    *Mo Indication of a aaapla daaorptlon undar axparlBantal eondltlona




           CONCENTRATOR SYSTEM (200 °C)


                                NITROGEN CARRIER GAS (30 min/ml) INTO CHROMALYTICS

                                CHROMALYTICS* OVEN/GC INJECTION PORT (250 °CI

                                GCOVEN CONTAINING 2.1 m.0.6cm O.D..GLASS
                                  COLUMN WITH L125 g OF TENAX (60/80 MESH)

                                FID DETECTOR 1350 °C)

                                GC CABLE TO CENTRAL 3554 HEWLETT PACKARD
                                  COMPUTER SYSTEM
FIGURE  10.   Desorption/analytical  system used in  desorption  efficiency  evalu-

                                            SORBENT TUBE IN TUBE FURNACE
                           SWITCH IN "TRAP"
                              (DIRECT) INJECTION PORT
                                      CURRENTLY UNUSED (CAPPED)
                                        CARRIER GAS SOURCE   TOGC
                                          DIRECT INJECTION OF STANDARDS INTO CC
                                            SORBENT TUBE IN TUBE FURNACE
                                                                   CURRENTLY UNUSED (CAPPED)
                              (DIRECT) INJECTION PORT

                                        CARRIER GAS SOURCE   TOGC

                                          THERMAL DESORPTION OF TUBES INTO GC
 FIGURE  11.    Diagram of  six-port,  two-position  valve within Chromalytics® oven.


      Since  these  experiments were of a preliminary nature,  only a few of the
 test  matrix compounds  were used in the evaluation of the desorption proper-
 ties  of  the sorbents.   However, by relating desorption efficiency to reten-
 tion  time data found previously (for capacity studies using 3  ft, 2 mm I.D.,
 1/4 in O.D.  glass columns filled with the sorbents) at or near desorption
 temperatures,  an  idea  of the types of compounds that are readily thermally
 desorbed from  the selected sorbents may be obtained.  Table 7  gives the
 chromatographic conditions and raw data for the determinations with the porous
 polymer  type sorbent materials, while Table 8 presents retention time data
 along with  experimentally determined desorption efficiencies.   Note that
 several  values are above 100 percent efficiency.  This may be  due to either
 background  contributions to peak areas (especially for Porapaks), or slight
 differences in peak shapes between direct injections and tube  desorptions
 resulting  in different peak areas.  In relationship to the indicated retention
 times,  the  desorption  efficiency data suggest potentially easy quantitative
 desorption  of  compounds with retention times (Table 8) of less than a few
                                 TABLE 8

           Retention Time Data and Desorption Efficiencies at or
             near Desorption Temperature for Sorbent Materials
Test compound
Propylene oxide
Ethylene glycol
Benzyl chloride
bis- (2-Chloroethyl) ether
1,2, 4-Trichlorobenzene
4-Bromodiphenyl ether
Hexachloro-1 , 2-butadiene
Retention time (min)
. Porapak N Porapak R
m @ 150°C @ 150°C @ 222°C
395 0.51 0.50 0.17
594 1.03 0.68 0.20
557 1.93 1.00 0,24
719 4.48 3.23 0.47
788 12.1 10.33 0.85
(113±l%)a (108±4%) (>132%)
906 -
1073 -
1176 -
1253 110
1269 -
(?)C (?)°
1402 165
1657 -
1684 -
1747 -
1811 -
2014 -
2616 -
2783 -

@ 302 °C
lumbers in parentheses are desorption efficiencies.
Significant contribution from sorbent background.
"No indication of sample desorption after 30 min at  experimental

                                                           TABLE 9

                            Qualitative Evaluation of Two Sorbent Properties Compared to 6m

                                           to Determine Ranges of Sorbent  Utility
6 6
m m
Key Name (lit) (calc)
A Methane 89
B Ethane 180
C Propane 282 282
D(7) n-Butane 395 401
E n-Pentane 505 519
8 Propylene Oxide 534
9 Acrylonitrile 557 573
F n-Hexane 629 638
16 Benzene 719 719
4 iso-Octane 788 834
6 Ethylene Glycol 906
3 Succinonitrile - 1073
18 Phenol " 1176
17 Benzyl chloride - 1253
13 Naphthalene 1269 1410
5 Bis-(2-chloroethyl) 1402 1402
12 o-Nitroaniline - 1657
15 m-Nitroanisole - 1684
10 Phenanthrene 1747
1 n-Hexadecane 1811 1766
14 1,2,4-Trichloro- - 2014
11 4-Bromodiphenyl - 2616
2 Hexachloro-1,3- - 2783
Trap Des
/ 1
^ 1

Porapak 1
orb Trap De:
/ 1
/ )
/ )
/ J
/ )
/ >
/ J
/ >

% Porapak 1
sorb Trap De!
? /
? /
f J
? /
C / 3
C / )
C / )
C / )
C / )
t / )
C / J

* Ambersorb
sorb Trap Desorb
00 1
00 !
00 I
00 !
/ 1?
/ 1?
/ 1?
/ 1?
l_ / 1?
/ X
^ X
/ X
7 / X
? / X
I? / X
? / X
C / X
C / X
t / X
C / X
C / X
t ^ X
t / X

SKC Charcoal
Trap DesorJ
00 1
00 1
00 1
/ 1?
/ !?
/ I?
/ 1?
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X
/ X

                                                                 Key to Table

                                                                    Probably  Possibly  Not Likely
Trap Quantitatively
Desorb Quantitatively
00 (Vg<480 1/g)

                                    TABLE 10
                Ranges of Utility of Solid Sorbents Based on  fim
                    SORBENT      6  (Trap)    To   6  (Desorb)
                                 m                 m
SKC charcoal
Porapak N
Porapak R
materials may prove  to be more desirable.

     In addition  to  desorptlon efficiencies and retention time data at or
near desorption temperature, Table 8 also gives 6  values for the compounds
listed.  This table  indicates a  relationship between these retention times
and 6  . especially for Porapaks  R and N.  This relationship suggests that 6
     m                                                                     m
may be used to determine which compounds will be quantitatively desorbed from
sorbent materials.

     Since a fair correlation has been demonstrated for 6  and sorbent capacity,
and a  relationship potentially exists for 6  and desorption efficiency, then
6  may be useful  in  determining  which compounds are likely to be quantitatively
adsorbed on and subsequently desorbed from a particular sorbent material.
A qualitative comparison of 6  with  sorbent capacity and desorption efficiency
is given in Table 9.  The entries in this table are defined in the key beneath
it and are based  upon a combination  of experimental data, demonstrated cor-
relations, and chemical intuition.   The enclosed regions indicate compounds
which  are probably or possibly quantitatively trapped at a high-volume condi-
tion (480 1, corresponding  to 1  1/min for 8 hr) and are probably or possibly
quantitatively desorbed from the sorbent materials indicated.  These regions
correspond to ranges of 6  values which are listed in Table 10.

     Based on these  results, a sorbent material may be used to sample most
compounds with £  values  falling within its fi^ range.  Alternatively, a
sorbent material  may be selected for sampling studies with a particular com-
pound  which has a 6   value  within the 6m range of utility of the sorbent.
For example, a compound with a 6 value of 2,000 should probably be sampled
using  Tenax-GC.   A compound with a «m value of 1,400 falls within the ranges
of utility of Porapak N, Porapak R,  and Tenax-GC.  Since 1,400 is close to

the 1,500 6  desorption limit of the Porapaks, Tenax-GC would again be the best
sorbent (of these five) to evaluate first for sampling this  compound.  However,
a compound with a 6  value of 1,000 would probably be easier to sample using
one of the Porapaks, since this value is near the 6  trapping limit of Tenax-GC,
A problem potentially exists for compounds with  6  values around 750, since
this is the 6  desorption limit of Ambersorb XE-340 and the  6  trapping limit
of Porapaks R and N.  Therefore, for a complex organic sample, some overlap
exists for the ranges of utility (and the types  of compound  sampled) of
Tenax-GC and the Porapaks, but a potential gap exists between the ranges of
utility of these polymeric adsorbents and the carbonaceous adsorbents.  It
should be emphasized that these ranges are only  estimates of sorbent utility,
and it is likely that some compounds may not adhere to this  scheme.

     The other area of interest during the evaluation of the thermal desorp-
tion properties of the selected sorbent materials was potential extraneous
background resulting from contamination and/or sorbent characteristics.
Sorbent backgrounds can be composed of a variety of contaminants from a
number of different sources.  Depending upon the degree and  nature of these
contaminants, they can lead to complications during both sample collection
and analysis.  In order to minimize problems from background contamination,
it is best to know as much as possible about what types of contamination can
arise, how they usually occur, and how best to minimize all  sources of con-
tamination.  In this way, complications—such as in situ reactions and inter-
ference during analysis—can be minimized and possibly avoided.

     Generally, all commercial sorbent materials as received from suppliers
contain significant levels of background contamination as the result of pro-
duction, packaging, and/or transportation.  Although most of this initial
contamination may be removed by solvent (before  packing in tubes) and/or
thermal (usually after packing in tubes) conditioning, such  conditioning may
also Introduce contaminants.  Backgrounds may be introduced  from trace im-
purities in the solvent and the gas stream, and  also from decomposition pro-
ducts of the polymer.  Obviously one does not heat a polymer above its tem-
perature limit or extract it in a solubilizlng solvent, but  some polymeric
breakdown can occur under other situations as well.  Each sorbent will have
thermal properties that are unique, and, therefore, conditioning procedures
should be validated and optimized for each sorbent material.  As a precaution,
one should also be aware of the types of compounds each sorbent produces as
the result of thermal and/or chemical degradation.  Table 11 lists several
of the major thermal decomposition products that could be expected from the
selected sorbent materials, as well as the chemical compositions and tempera-

                                      TABLE 11
                     Expected Thermal Decomposition Behavior
                          of Selected Sorbent Materials
                   Chemical composition
Temperature     Major thermal
limit (°C)  decomposition products
       Porapak N
       Porapak R
               N-vinyl pyrrolidone
               N-vinyl pyrrolidone
                     2 , 6-Diphenyl-p-phenylene oxide
       Ambersorb XE-340 Carbonized styrene-divinyl
SKC activated
                     Carbonized organics




Vinyl pyrrolidone
Vinyl pyrrolidone
Alkyl benzenes
Alkyl phenols
ture limits of the sorbents.   Since the temperature  limit of Ambersorb XE-340
was not  stated by its manufacturer, a freshly  packed 3  ft, 1/4 in O.D., 2 mm
I.D. glass column of  this sorbent was prepared,  conditioned, and analyzed by
GC/MS  to verify that  thermal  decomposition was not occurring at the condition-
ing temperature of 350°  C. This column was  placed in the GC oven of a Hewlett-
Packard  Model 5982A GC/MS system and subjected to stepwise temperature in-
creases  (50° C intervals) up  to 250° C.  The effluent from the column was
continuously monitored  by mass spectrometry  and showed  none of the expected
thermal  decomposition products of the styrene-divlnyl benzene polymer.  In-
deed,  the only effluents observed were C02,  S02» argon, and traces of a few
low molecular weight  aliphatics.  Therefore, thermal decomposition of Amber-
sorb XE-340 does not  occur at 350° C, which  is above the anticipated operating

     Once the conditioning procedure for a  particular sorbent has been estab-
lished,  care must be  taken to prevent additional contamination during sub-
sequent  handling.  The handling and storing  of sampling tubes should be
optimized without being unreasonably complicated or  expensive.  The  techniques
developed should also ensure the  integrity  of  all samples collected  (i.e.,
no sample loss).

     Finally, backgrounds from other sources in the  desorption/analytical
system should be eliminated or reduced  to  the  greatest extent possible.
These  can arise not  only  from the usual  sources of chromatographic  background

contamination (e.g., column and septum bleed), but also  from  sources  that are
unique to the method of desorption and sample  introduction.   These less obvious
difficulties could include dust being introduced  into  the desorption  chamber,
or fingerprints on the sample  tube, depending  upon the desorption chamber
design.  Generally, background contamination introduced  by a  desorption/analyt-
ical system are best minimized by using  forethought when designing the system.

     A series of experiments was conducted  to  evaluate the background levels
of the materials being studied and possible sources of background within the
desorption/analytical system.  Sampling  tube backgrounds were evaluated by us-
ing the desorption/analytical  system sketched  in  Figure  12 and  the sampling
tubes listed previously in Table 6.  The following are some of  the findings
determined during these evaluations of sorbent backgrounds.


     •  No difference was noted between  solvent extracted and nonsolvent ex-
tracted sorbents (for the sorbents studied  and the experimental  system used).

     •  For Porapak R, conditioning at 240° C  produced less background upon
desorbing at 170° C than with  conditioning  at  175° C and desorbing at 170° C.

     •  Thermal conditioning removed a great deal of contamination, most with-
in the first hour.

     •  After conditioning, the backgrounds of Porapak R and  N  tubes  were above
the background of a blank (empty) control tube, while  the backgrounds of
Tenax-GC, Ambersorb XE-340, and SKC activated  charcoal matched  the background
level of the blank control tube.

     Handling and Storage.

     •  A desorption system designed to  eliminate the  exterior  of the sampling
tube from analytical desorption greatly  diminishes problems with handling and

     •  Capping tubes with metal fittings and  storing  them in culture tubes
with Teflon-lined caps seems to be the best method of  tube storage to elimi-
nate extraneous contamination.

     •  Glass sampling tubes offer a more inert surface  than  stainless steel,

                              © 571QA HEWLETT PACKARD GAS CHROMATOGRAPH
                              © 7132A HEWLETT PACKARD CHART RECORDER
                              ® CHROMALYTICS THERMAL CONTROLLER TO 1047
                                 CONCENTRATOR SYSTEM
                              © CHROMALYTICS TUBE DESORPTION CHAMBER TO
                                 1047 CONCENTRATOR SYSTEM
                                 CHAMBER TO THERMAL CONTROLLER
                              © DESORPTION CHAMBER ENTRANCE AND (0-RINGI SEAL
                              © NITROGEN CARRIER GAS (30 ml/mini
                              © DESORPTION CHAMBER/GC INJECTION PORT (300°CI
                              ® GC OVEN 1320 °C) CONTAINING EMPTY a 9m, a 6 cm O.D..
                                 2 mm I. D. GLASS COLUMN
                              ® FID DETECTOR (350 °C)
                              ® GC CABLE TO CENTRAL 3554 HEWLETT PACKARD
                                 COMPUTER SYSTEM

 FIGURE  12.    Desorption/analytlcal  system used  in tube
 background  evaluations.
                                                                                             NOTE:lPSIG = 6.9xl03Pa
                                                                                                                                  IN SERIES
   SS*.'.S^                            	B——
&•&            	e	   GLASS w
>•%-	r	:	_	•—
                                           AMBERSORB XE-340
FIGURE  13.   Pressure drop versus  flow  rates  for
15  cm, 0.6  cm  O.D.,  4 mm  I.D. glass  sampling
tubes  filled with sorbent materials.

and have fewer contamination and  reactivity problems.

Evaluations of Sampling Tube Pressure Drops and Design

     One limitation  that automatically applies to  any portable  air  sampling
system is the ability of a portable  pump  to pull air through  sorbent-filled
sampling tubes at a  desired rate.  This ability is directly related to  the
pressure drop across the sampling  tube.   For a particular  pump  to pull  air
through a sampling tube system at  a  specific rate  (0.5  to  3 1/mln for this
project), the pressure drop across these  tubes must be  equal  to or  less than
a certain maximum value.  This corresponds to the  rated capacity of the pump
at a particular pressure drop.  Sampling  tube parameters which  affect the
magnitude of the pressure drop include:   type of sorbent material packing,
mesh size of sorbent material packing, amount of glass  wool plug, diameter
of tube, and amount  of sorbent material packing.

     Of these parameters, the types  and amounts of sorbent material packings
will be mainly determined by the  capacity of these sorbents for atmospheric
pollutants under the anticipated  sampling conditions.   The mesh size of the
sorbent materials offers some versatility, but there are limitations to the
mesh sizes available.  The amount  of glass wool used to contain the sorbent
material within the  tube can be a  critical factor  in determining the pressure
drop, so that care should be taken to ensure that  the minimum amount of glass
wool is used.  In larger diameter  tubing, a method of restraining the glass
wool plug may be necessary when minimal amounts of glass wool are used.
Tapering the ends of the tube or  inserting a stainless  steel  screen are two
possible ways to secure the glass wool plugs.

     The experimental parameter that offers the most flexibility in determin-
ing the pressure drop is the tube diameter.  The only restriction on the di-
ameter of the tube is the necessity  to achieve rapid and uniform heating of
the sorbent bed during the thermal desorption process.  This  becomes increas-
ingly difficult with increasing sampling  tube diameter.

     The pressure drops across sampling tubes of various shapes, diameters,
and sorbent beddings were measured using  a differential pressure gauge.  Flow
rates were set by using a Brooks Model 5841 thermal mass flow sensor/con-
troller.  An example of the type of  data  generated during  these pressure stud-
ies is given in Table 12 and depicted in  Figure 13.

     These pressure drop studies led to the conclusion  that a 10-mm diameter

                                           TABLE  12

                Pressure Drop Data  for Glass Sampling Tubesa (pslg)

              Tenax GCTenax GCPorapak NAmbersorb XE-340Glass tube~
             60/80 mesh  35/60 mesh  50/80 mesh     (not sieved)     without    Tenax GC (35/60),
      Plow    0.1953 g    0.2014 g    0.2018 g       0.1964 g      sorbent     Porapak N, and
      rate   w/34.3 mg   w/34.3 mg   w/38.9 mg      w/36.1 mg      w/36.2 mg  Ambersorb XE-340
3.0 1.1 to 1.2
1.6 0.6
0.4 0.1

     Other conditions

       Transfer  lines only at 3 i/rain show M) psig pressure drop.
       Inlet pressure:  62 psig; downstream pressure: 9 psig.
                                          TABLE  13

                              Adsorbent  Properties Chart
Taap. Oond. Daa.
llaUt t««p. tan*. Ctwalcal
Adaorbant CO I'CI (*CI eoBwwitlon
Tanu-GC 400 120 100 2,6-oiphanyl-
(15/60 a»an) p-phaaylana

Porapak P. 250 215 150 n-Vlnyl
(50/10 aaah) to ryrrolidoaa

Porapak H 190 175 150 a-Vinyl
(50/10 Man) Pyrrolldona

ABfearaorb* >400 120 100 carbonlaad
O-140 ftyrana-

•XC Aoti- »400 120 loo carbonltad
vatad Organlea

Hijor tharaal
Alkyl ununa
Alkyl Phanola

Vinyl Pyrroli-

Vinyl ryrroli-
Pyrri lidiana

nona obaar-
vad laftar
at JJO'C •» ob-
aarviBg on

Background ap*
GOOD 160
(nona dataotad
afaova ayataat

Attar condl- ^.l"
tinning at
2J5-C, back-
ground upon
daaorbiag lat
(wall abova
ayat. baok-
PAIH 0 150*C
ayat. baok-
poo* 1.1
(^•ii abova
ayataai back-

0000 u>ft
(DOM datac-
tad abova
ayata* baek-
(noon datao-
tad abova


Would afCUiantly
trap IntanMdlataly
(t laaa) volatila
oowounda wlttt
allghtly laaa affi-
nity for polar OOB-
•bould affielantly
trap latanadlauly
(( laa»l voUtlla
coipounda vith
allghtly graatar
affinity for polar

Should afflctantly
tup intaraadiatatly
(t laaa) volatlla
coapounda with
aligbtly graatar
affinity for polar
fbould afflalaBtly
trap bigbly (t all
laaa) volatlla oora-
P^jmt^ vitb graatar
affinity for polar
•bould affioiaotly
trap highly (t all
laaa) volatila eo*~
pouaoa Mtn aucn
graatar affinity far
polar compnuiKla.
Vary aaanabla to
tnarMal daaorp-
tlon for Intar-
••diataly (t all
hlgbar) volatila

Vary MiniliU Co
thttn.*..! dvatorptioo
COT lAfcMaMdiAMly
U *11 blflhw)
volatila ooapounda*

Vary laanihla to
tbanal daaorption
for lataraadlataly
(i all higbar)
voUtlla oMPnuMa.

aainibltlty to
tbanal daaorption
for all but higply
volatlla bydro-
bllity to tnaoal
daaorption for all
but Uglily volatila

Rang* of utility
v»io to >J«00

•V750 to «15OO

M50 to 1-1500

i rat.

 represents a reasonable compromise between the requirement to have the small-
 est possible diameter for efficient thermal desorption and yet maintain an
 acceptable pressure drop across sorbent-filled tubes.  Two designs, straight
 and tapered, with 10-mm cross sections, were evaluated taking into account
 the desirable design features of the miniature collection system.  Although
 the tapered tubes had higher pressure drops than the straight tubes, the
 difference was not significant.  Furthermore, the tapered design offered ad-
 vantages for the containment of the sorbent, since smaller amounts of glass
 wool can be used with no "blowouts" of sorbent material.  Finally, lighter,
 less expensive (one-quarter inch) fittings can be used with the tapered tubes,
 which is an Important factor in the design and construction of a miniature
 collector.  The specifications for the cartridge design determined to be most
 desirable is given in Figure 14, along with a sketch of a sampling tube.

 Evaluation of Sorbent Package

      Five sorbent materials have been evaluated to determine their basic
 properties, including special abilities and potential problems.

      Table 13 is a compilation of the major sorbent material properties eval-
 uated.  Along with the name of the adsorbent is included:
      •  The manufacturer's stated temperature limit (or >400° C).
      •  The temperature used for sorbent conditioning.
      •  The temperature used for sorbent desorption.
      •  Some of the major thermal decomposition products.
      •  A qualitative description of the background to be expected from the
 sorbent material.
      •  The pressure drop across tapered tubes containing 1.0 g of adsorbent
 at a 3 1/mln flow rate.
      •  A qualitative description of the type of compounds that will potenti-
 ally be quantitatively trapped by the adsorbent (capacity >480 1/g).
      •  A qualitative description of the type of compounds that will potenti-
 ally be quantitatively thermally desorbed for analysis.
      •  The range of utility based on fi  (described previously).

      Of these five sorbents, a combination of three has been selected for
 use in the miniature collection system.  These three sorbents are listed be-
 low along with the major reasons for their selection.

      Tenax-GC.  The only high-temperature (400° C) adsorbent available which
 allows the quantitative thermal desorption of low volatility organic compounds.

- 7
• /.
12 mm 0. D.

CiH *
10 mm I.D.
%a ^^^^M^S
4mm 1.
                            SORBENT MATERIAL
                  GLASS WOOL
FIGURE 14.  Specifications for tapered sampling  tube  design.
     Porapak R.  One of the highest-capacity polymeric absorbents with a
reasonable background level (better than Porapak N) and with an overlap in
range of utility (6 ) with Tenax-GC.

     Ambersorb XE-340.  Less difficulty anticipated with  the desorption of
compounds of intermediate volatility,  fewer detrimental effects by water and
reactivity with collected samples than with charcoal.  Also, its range of
utility (6 ) leaves the smallest gap  between polymeric and carbonaceous ad-
sorbents in the types of compounds collected.
     These sorbent materials will be packed  into  three  separate tapered glass
tubes for sampling, with flow being directed either  in  series or in parallel.
If sampling is done with the sorbent tubes in series, the arrangement will
be Tenax-GC at the air intake, Ambersorb XE-340 at the  air exhaust to the pump,
and Porapak R in the middle.  In terms  of total range of utility (6 ) of the
sorbent package, it should quantitatively trap and desorb compounds ranging
from ^450 to >2,800 (fi ) with a possible "qualitative only"
               gap at *750  (6 ).
     This research was sponsored by the U.S.
under Contract No. 68-02-2774.
Environmental Protection Agency


 1.  Burrell, H., Immergut, B.  Solubility Parameter Values.   In:   Polymer
     Handbook (Brandup, J., ed.).  New York, Interscience Publishers, 1966.

 2.  Gallant, R.F., King, J.W., Levins, P.L., Piecewicz, J.F.   Characteri-
     zation of Sorbent Resins for Use in Environmental Sampling.   EPA-600/7-
     78-054, pp. 79-92.  March 1978.

 3.  Holzer, G., Shanfield, H., Zlatkis, A., Bertsch, W., Juarez,  P.,
     Mayfield, H., Liebich, H.M.  Collection and Analysis of Trace Organic
     Emissions from Natural Sources.  J Chromatogr 142:755-764, 1977.

 4.  Small, P.A.  Some Factors Affecting the Solubility of Polymers.  J Appl
     Chem 3:71-80, 1953.

Joseph J. Brooks, Ph.D., Diana S. West, and Donald J.  David
Monsanto Research Corporation
Station B, Box 8
Dayton, Ohio  45407

James D. Mulik
Organic Pollutant Analysis Branch
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina  27711
     SCHEIDE:  Gene Scheide, Environmetries.  What was the temperature  used
in the thermal preparation of your adsorbent material?

     BROOKS:  It varies with sorbent material.  Of course, it is  based  on
what the upper temperature limit is.  We used about 350 degrees Centigrade
for the Tenax, for the Ambersorb and the charcoal.

     We used about 175° C for the Porapak N, and about 240°  C for the Porapak
R.  Again, these are related to what their upper thermal limit is.

     I should say we will be evaluating these sorbents in the second year
against 20 possible and probable carcinogenic compounds that have been  iden-
tified as being of interest.

     SPENGLER:  I understand you have not used all  of  these  in  the  field,
yet.  The Tenax is currently being used by Lawrence Berkeley Labs on  indoor-
outdoor measurements, to look at organics.

     But we would like  to  relate someone else's  experience:  that of
Peter Mueller from Environmental Research and Technology who also used  a
Tenax collector in ambient conditions.  He related  that they had some ser-
ious problems with blank contaminations with  the Tenax, and  they were totally
frustrated with the its use.

     I don't know if you have any comments on what  would cause  that, but he
said that oftentimes the blanks would have more  material on  them than the
exposed Tenax.  They are disenchanted with it.

     MEIER:  Gene Meier, EPA, Las Vegas.  The problem with Tenax traps  is
aging.  As your trap ages, it degrades, and you  can't hold a trap for more
than 3 weeks; otherwise, you get a significant hydrocarbon background from
two sources—either just contamination from the  normal environment  on your
Tenax, or also it can be the composition of the  Tenax itself.

     If you look at your gas chromatogram, you will see a typical hydrocarbon
response, along with the chromatogram of the various alkanes as they break
down off the Tenax itself.  That's why I would like to see some of  the  data
on the Ambersorbs, because we have done some work at Las Vegas—not our lab
personally, but the people at the University—and the Ainbersorb seems to be
a lot cleaner than the Tenax in trapping all  the organics from water samples,
which I would expect would hold true here.

     BROOKS:  The reason that we are using three different materials is be-
cause we have to cover  the range of compounds.   You might have problems re-
covering a lot of the heavier, less volatile materials—if you were just
using Ainbersorb.

     MEIER:  Have you actually physically collected the samples and tested,
or are these hypothetical  results?

     BROOKS:  We have done a limited amount of sampling with test compounds.
Right now, in the second year, we will be working with the actual 20 com-
pounds that we have selected to determine what the  collection and adsorption
efficiencies are.

     MEIER:  The people out at Research Triangle Park have been using the
Tenax traps for their monitoring program and have been quite successful in
trapping a wide range of hydrocarbons, including your heavier molecules.

     BROOKS:  With Tenax,  right, but I am saying that with Ambersorb, you
are not going to get them  back.

     MEIER:  I realize  the differences there, but I am saying that  you  work
in a whole series and one  of the questions I am  asking is why is the Tenax
acceptable, while you are  looking for the three  solvents?

     BROOKS:  Because when you are talking about  sampling,  say, at a liter-
per-tninute for 8 hours, you are going to get significant breakthrough of the
higher volatility compounds—they will not be  trapped.

     MEIER:  Well, they have been collecting 100  liter  samples on 5/8-inch
diameter and I would say about 4 inches long tubes, and have been quite suc-
cessful.  They use it in the quality assurance program  where they spike
their traps before they go out in the field, so they can verify when they
have a breakthrough.

     BROOKS:  I doubt that they are successful for  the  really highly vola-
tile compounds.  Let me say just one other thing.  We do have first annual
reports available on two contracts that relate to this  type of sampling and
that will be out very shortly.

     MAGE:  I just might add from my own personal experience that so much
depends on the method of preparation.  One batch  of Tenax may not be the same
as the other.  There is a lot to be done to maintain quality.

Adapting  Commercial Voltammetric Sensors
to Personal  Monitoring Applications
     Manny Shaw, Ph.D.
     InterScan Corporation
     Chatsworth, California
     Of recent concern to industrial  hygienists and safety engineers is the
exposure of workers  to harmful levels of noxious gases and vapors.  Apart
from the survey type or fixed-station monitor, there is a need for an in-
expensive but accurate personal monitor with real time indication—one that
is wearable without  being burdensome. The one proven analytical technique
meeting these requirements is the electrochemical voltammetric method of
instrumental analysis.

     The voltammetric method differs from other instrumental methods of gas
analysis, such as  the coulometric,  conductimetrie, or colorimetric techniques,
which require that the sample gas react with chemical reagents.  In the volt-
ammetric technique,  sample gas is electro-oxidized (or electroreduced) di-
rectly at an electrode.  This makes it possible to design a simple and prac-
tical portable monitor using a dry technique.

     The voltammetric sensor operates under diffusion-controlled conditions.
The gas sample is  drawn through the sensor, passing over an electrocatalytic
sensing electrode  of the gas diffusion type.  Some of the gas molecules dif-
fuse into and are  adsorbed on the electrode, where they are electrochemically
oxidized or reduced  (depending on the sensor) at an appropriate electrode
potential.  This electrochemical reaction results in an electric current,
directly proportional to the gas concentration.


     To understand the concept of a diffusion-controlled reaction, picture
a single pore within the porous surface of  the sensing electrode.  Within each
such pore, a thin film of electrolyte  constitutes a microscopic meniscus in
contact with the sample gas, creating  a three-phase junction of gas, liquid,
and solid (electrode surface), shown schematically in Figure 1.  The total
electro-oxidation reaction of a sample gas  includes dissolution of gaseous
molecules in this thin liquid film, diffusion or mass transfer of the dis-
solved gas through the film, adsorption of  the gas on the electrode, and trans-
fer of the electron charge to the metal surface.  The reaction becomes dif-
fusion-limited when the gas is consumed at  a greater rate than can be re-
plenished at the reaction site; that is, when the charge transfer reaction
is faster than the mass transfer reaction.  The limiting diffusion current
is linearily proportional to the gas concentration according to the simpli-
fied equation
where i is the limiting diffusion current  in amps, F is the Faraday constant
(96,500 coulombs), A is the reaction interface area in cm , n is the number
of electrons per mole reactant, L is the diffusion gradient in cm within the
thin liquid film, C is the concentration (moles/cm ) of gas molecules dis-
solved in the thin liquid film, and D is the diffusion coefficient (cm /sec)
of the gas.  To be an analytical device having linear readout with gas concen-
tration, the voltammetric sensor must be biased so that the electrode reaction
occurs within the diffusion-limited region of the voltammetric curve.  This
is illustrated in Figure 2, showing typical voltammetric curves for several
gases.  Electrochemical current is plotted as ordinate, and the sensing
electrode potential as abscissa, relative  to the reversible hydrogen elec-
trode (RHE) at zero potential.  Increasing oxidation potential (more positive
values—Stockholm-Gibbs convention) is towards the right.  Oxidation current
is arbitrarily chosen as positive.

     Taking SO. as an example, no reaction occurs until a minimum oxidation
potential is reached at a, at which point current begins to flow, increasing
until it attains a limiting value at b.  Since the region b to c represents
the diffusion-limited region of the current, the sensing electrode in the SO-
sensor is held at an oxidation potential of V. volts.  This is accomplished
by applying the appropriate bias voltage to the sensing electrode relative
to the reference counterelectrode.  Similar voltammetric curves are shown
for H.S and CO oxidation.  NO- is unique in that it exhibits both oxidation
and reduction characteristics.

   PHASE -
            FIGURE 1.  Electrode-electrolyte
FIGURE 2. Voltammetric curves of several gases.

     These curves also show that In certain cases, interferences may be elimi-
nated by biasing the sensor at an oxidation potential below that of the inter-
fering gas.  For instance, if the H2S sensor is biased at V2t below the oxi-
dation potential for S(>2, then S02 will not interfere in the HgS sensor.  HjS
will, however, interfere in an S02 sensor.  The extent of this interference
depends upon the specific output of the sensor in its response to the inter-
fering gas.  In certain cases, choice of electrode material may virtually
eliminate interference from certain gases.  Selective filters are generally
used when a potential interferent cannot be electrochemically scrubbed.

     Figure 3 is a schematic representation of a voltammetric sensor used in
present commercial survey monitors.  Sample air, drawn into the sensor by a
pump, streams over the backside of the sensing electrode.  A portion of the
sample diffuses into the body of the electrode, adsorbing on its electro-
catalytic surface moistened with electrolyte.  The bulk of the electrolyte
is immobilized in a matrix comprising the reservoir, which also contains the
reference counterelectrode.  A small banana jack in the sensor head and a
metal strip fastened to the sensor near its bottom make connections to the
sensing electrode and to the nonpolarizable reference counterelectrode,
respectively.  The sensor is secured by a clamp permanently mounted to the
base of the portable analyzer, the clamp also serving as an electrical con-
tact to the metal strip attached to the reference counterelectrode.

     The use of a nonpolarizable counterelectrode allows it to serve the dual
function as reference electrode, so that a third electrode is not required.
The three-electrode sensor offers no advantages in sensitivity, response time,
drift, linearity, or interference rejection—these parameters being a function
of the sensing electrode only.

Response Time

     Response times of properly designed voltammetric sensors are relatively
fast, ranging from 3 to 20 seconds to 90 percent of final value, depending
on sensor type.  Rise and fall times are approximately equivalent, although
the fall time of CO sensors tends to tail the last 5 to 10 percent.  Excessive

              •  • • <
              • • •
        ELECTROLYTE  ,
        (150 ml volume)
                 ft'fi'!'!'!'!'!'ffi'!-!yffi         • " • *
                  •" ~ • • ^^^^9t^^^T^9^9^9t9mm 9 9 9
                                               SE BANANA JACK
                                                 EXTERNAL REC
                                                 CONTACT TO
                                                 SENSOR CLAMP
FIGURE 3.  Voltammetric sensor in portable monitor.
FIGURE 4.  Background current of sensor,

drying out of some sensors, particularly the CO type, Increases the response
time, due to significant  increases  in  internal resistance.

Sensor Range

     The specific output  of voltammetric sensors, using high reaction rate
sensing electrodes, is high—on  the order of microamps-per-ppm.  This permits
measurements over a broad range.  For  practical reasons, determined by zero
drift and interference, the lowest  recommended range is 0 to 1 ppm full scale.
The upper limit is determined by the current and voltage characteristics of
the current amplifier.  Excessive current, due to gas concentration well in
excess of the meter range, may cause amplifier saturation, clipping the
output voltage—resulting in a fixed,  meaningless meter reading.  The measur-
ing range of the instrument is determined by the appropriate feedback resistor
associated with the current amplifier.  Because of power supply limitations,
feedback resistors of different  values are used to cover a broad range of
concentration.  Multlrange units have  a range switch to select the particular
feedback resistor.  Special sensors can be formulated to give a low specific
output, permitting their  use in  high-concentration applications.

Zero Drift

     A key performance parameter of a  practical sensor is its zero drift
characteristic.  In voltammetric sensors, this phenomenon is caused strictly
by the temperature effect on the background current.  Figure 4 shows the back-
ground current of the sensor as  a function of oxidation potential.  The rising
positive current at higher potentials  is due to electro-oxidation of the elec-
trocatalyst, followed by  oxidation  of  water to oxygen at the highest potentials,
The rising negative current at lower potentials is caused by electroreduction
of dissolved oxygen in the electrolyte.

     The zero control in  the instrument provides a compensating voltage, which
nulls out this background signal.   With change in temperature, this current
increases or decreases, resulting in an uncompensated signal.  This drift is
measurable in microamps and is a sensor phenomenon only, remaining constant
irrespective of the amplifier gain  characteristics and, hence, the meter
range.  Percent drift must therefore increase with decreasing full-scale con-
centration range.  As an  example, a sensor has an output of 10 yA/ppm and a
zero drift of 0.1 nA per  ° C, equivalent to 0.01 ppm per ° C.  On a 0 to 10
ppm range, this corresponds to a change of 0.05 ppm for a 5° C change in
temperature, giving a zero drift  of 0.5 percent of full scale.  This corre-

sponds to 5 percent drift on the 0 to 1 ppm range and 0.5 percent drift on  the
0 to 100 ppm range.

Sample Flow Rate

     Sensor output is flow dependent at flow rates below 2 liters-per-
minute, as illustrated in Figure 5.  This is not a pressure effect, as demon-
strated by data obtained by pushing the sample through the sensor from a
pressurized cylinder, or pulling it through using the internal pump and a
sample bag.  This flow effect is a result of the rapid consumption of gas at
the high reaction rate sensing electrode.  As the flow rate is decreased,
there is less tendency for new gas molecules to replenish those consumed at
the electrode; i.e., the flow rate becomes less and less sufficient to fully
sustain the reaction.  Figure 5 shows that a decrease of flow rate from 940
to 740 ml/minute decreases the sensor output by 5 percent.

     Actually, the flow rate effect is an advantage because no purge of the
sensor is required.  One merely shuts off the pump for zeroing.  The portable
InterScan analyzers have a constant flow pump so that the flow rate is fixed,
independent of the state of battery charge.  Calibration with a sample bag
ensures that the same flow rate will be used for the sample measurements.

     The electrochemical voltammetric technique is highly suited to real
time air quality monitoring and integrated exposure measurement.  It is prob-
ably one of the few available techniques offering a combination of small
size, low cost, minimal maintenance, and accurate linear output.  The scaling
down in size from a portable monitor to a personal monitor requires no basic
development effort, but is rather one of design engineering.  A large re-
duction in sensor volume can be accomplished with only a relatively small
reduction in area of the active elements—the electrodes.

     The sensor in the portable, survey-type monitor has been reduced from
                 3        3
a volume of 24 in  to 1 in  for the personal monitor.  The total volume
                                          3                     3
of the personal monitor is less than 20 in  compared with 525 in  for the
portable monitor.  This drastic reduction in size in done by reducing the size
of the reservoir.  The trade-off is the necessity for the frequent addition of
water to the personal monitor reservoir.  (No addition of water is required
in the sensor shown in Figure 3, which has a sufficiently large reservoir
to give a sensor life of a year or more of normal use.)


                                                      Internal pump -f
                                                        sample bag
                   1000              1500
                   FLOW RATE IN ML/MIN
 FIGURE 5.  Effect of  sample flow rate on sensor output.

                                          CO DOSIMETER RESPONSE CURVE
                                            1 INCH= 30 SECONDS
 FIGURE 6.  CO response in personal monitor.

     Such reduction of size does not affect sensor performance.  Figure 6
shows the response curve of the 1 cubic inch sensor in the CO personal monitor,
Rise and fall times are rapid.  Rise time is 8 seconds to 95 percent (chart
speed shown is 1 inch equals 30 seconds).

     The voltammetric technique is particularly suitable for personal dosi-
meters which give an accumulated exposure over a time period.  Using coulo-
metric or electronic integrators, the total dosage in ppm-hours is obtained
in a separate read-back unit.  Read-back may be done at the end of an 8-hour
shift, and the time-weighted average (TWA) calculated from the ppm-hours.  If
desired, read-back can be done at any time during a given period to identify
the existence of high-dosage periods.

     The disadvantage of this approach is that in many cases the read-back
reading is an after-the-fact event, the wearer of the dosimeter already having
been exposed to a TWA greatly in excess of the threshold limit value (TLV).
Even though some dosimeters have an alarm indicator, it is only for indicating
the ppm level at a given moment.  A more practical device would be a dosimeter
capable of being set to alarm at a TWA equivalent to the TLV.  This concept
has been developed, and is presently being used by a major steel manufacturer
as an area monitor equipped with a TLV alarm.  The electronics necessary to
accomplish this can be packaged in a pocket-sized dosimeter.  The availability
of low-current logic components permit their usage in battery-powered circuits.

     Figure 7 shows a block diagram of a personal dosimeter using this new
concept.  Shown are the more standard components such as reference voltage,
zero and span circuits, amplifier, power supply, and low-battery indicator.
A plug-in meter is used for calibration and ppm level measurement.  The block
diagram entitled TWA & ALARM CIRCUITRY represents the logic package designed
to accomplish the TLV alarm.  This is detailed in Figure 8.

     The analog output voltage is converted to a digital format.  A clock
(square wave oscillator) regulates the feeding of information through the
gate into memory storage.  If and when the TLV is exceeded, a visual (LED)
and audible (buzzer) alarm is actuated.  The TLV is set at the factory, specif-
ic for that type of dosimeter.  A second circuit permits a second alarm to
activate, based on exceeding the Short-Term Exposure Limit (STEL).  This is
Indicated by a different color LED and the buzzer.

                                                             To Calibration Indicator
Sample  ]•-——-

 FIGURE  7.   Block diagram of TLV alarm dosimeter.
                                                                       To Optional
 FIGURE  8.   Block diagram of  TWA Integrator.

     An important feature of this dosimeter is the TEST circuit, which allows
the user to check the operation of the TLV alarm.  By flipping a switch, a
signal is generated, which is designed to actuate the alarm in a given specif-
ic time, such as 30 seconds.  The timing of the alarm is precise, within
+ 1 second.  The timing is so precise that the test should be repeated three
or four times to be certain that the timepiece is being read correctly.  (The
RESET control is used to clear the memory.)  If the alarm is actuated too soon
or too late, relative to the 30+1 seconds, the unit is faulty.  This simple
test, performed in the dosimeter itself, gives an excellent quick check for

     The concept of a TLV alarm dosimeter has obvious advantages.  No auxil-
iary read-back unit is required—the dosimeter is complete in itself.  A re-
set button in the dosimeter allows its transfer from one worker to another
during a shift, without the necessity of going to a separate erasing unit.
Although an optional TWA unit will become available, there is the practical
question as to its need.  The main purpose in determining the ppm-hours of ex-
posure, or the TWA, is to prevent a worker from being exposed to a dosage in
excess of the TLV.  If such is the case, then it seems that a logical way to
do this Is with a TLV alarm dosimeter.

Manny Shaw, Ph.D.
InterScan Corporation
9614 Cozycroft Avenue
Chatsworth, California  91311
     PETROVICK:  You illustrated a slide there where you have the electro-
lytes bound within a matrix.  Could you give me an idea why you used a ma-
trix and what its particular features are in that design?

     SHAW:  Well, I am an old battery man besides being a fuel cell man,
and I don't see the logic of taking the sensor and putting in free acid,
because if the electrodes which are rupturable diaphragms are ruptured,
you have got free acid all over the Instrumentation.

     I used to meet electronic engineers and when they would look at  such
devices they would exclaim:  "Oh, gee, we would never put  that  into our
instrumentation—having an acid electrolyte in there."  They don't stop  to
think:  they do the same thing with NiCd batteries all the time.  It  has
got strong alkali like the material you clean up a toilet  with.  It's the
same type of alkali; it is that strong—23 percent potassium hydroxide.  But
it isn't sitting around there loosely, slushing around.  It is  in a matrix,
bound up.  So, we simply make the sensor the same way.

Solid Sorbent for Acrolein and
Formaldehyde  in Air
     Avram Gold, Ph.D., Thomas J. Smith, Ph.D., Christoph E. Dube, and
     John J. Cafarella
     Kresge Center for Environmental Health
     Harvard School of Public Health
     Boston, Massachusetts

     Acrolein and formaldehyde are toxic and irritating volatile aldehydes
of increasing interest In industrial hygiene and environmental  settings.
Outdoor sources of acrolein  are combustion and diesel exhaust,  while indoor
sources may be cooking and smoking.  Formaldehyde exposure in residences
from particle board and blown-in urea-formaldehyde foam have recently gen-
erated concern.

     Current sampling  methods for both aldehydes (7,8) involve  the use of
liquid-filled impingers and  are not particularly well-suited to personal
sampling.  Additionally, the acrolein analysis must be performed within sev-
eral hours after sampling, and the effect of storage on formaldehyde samples
is unknown.  Studies on stability of formaldehyde solutions of  low concen-
tration, however, strongly suggest that storage will present problems (11).
Since our initial interest in the aldehydes was for inclusion of a sampler
in a package to be used by Boston firefighters on-the-job, we desired a
solid sorbent which would have a large specific capacity and provide sta-
bility for stored samples.

     Activated molecular sieves  have been used extensively in industry  to
purify gas streams (1,6) and,  unlike currently available porous polymers,


 have  large capacities for adsorbates—20 to 30 percent by weight.   Despite
 the clearly Indicated potential for air sampling,  molecular sieves have not
 found much application for this purpose.  We therefore undertook an investi-
 gation to determine the suitability of molecular sieves for sampling low
 molecular weight aldehydes in air.   For this study, 13X sieves were chosen
 because large pore size would ensure efficient mass transfer.   Sieves were
 ground to 12/30 mesh and activated  by heating slowly to 400° C under vacuum and
 maintaining temperature and vacuum  overnight.  Because acrolein is more toxic
 than  formaldehyde and the problem of obtaining constant concentrations in air
 much  more straightforward, verification of sieves  as a sampling medium for
 acrolein received priority.


      Since some labile compounds react on sieves even at ambient temperatures
 (2,9,10), efforts were initially directed at determining the yield of recovery
 from  sieves.   When known amounts of acrolein were  loaded onto sieves as vapor,
 recovery by desorption with chilled distilled water (sieves added to water)
 was quantitative:  97 + 11 percent  for 3 to 8 ug acrolein/g sieves, 90+7
 percent for 60 to 200 yg/g.  Analysis of desorbate was performed by gas chro-
 ma tog raphy on Tenax-GC with a flame ionizatlon detector.  Sample tubes loaded
 with  equal volumes of vapor, stored in a refrigerator, and analyzed serially
 over  4 weeks showed no losses.

      The performance of sieves was  assessed by determining the breakthrough
 characteristics of acrolein.  We determined that the Theory of Statistical
 Moments (TSM) (3-5) would describe  the behavior of acrolein on sieves and
 extrapolated breakthrough performance from moderate to low concentrations
 by TSM.  In this manner, problems inherent in working with low concentrations
 of a  reactive compound were avoided.  The apparatus in Figure 1, shown in a
 configuration for breakthrough testing, was capable of producing constant
 concentrations in the range 400 to  3,000 ppm for periods of up to 4 hours.
 Typical breakthrough data obtained  with 1 g of sieves in a 10 mm I.D. sampling
 tube  at a sampling rate of 1 1/min  are given in Figure 2.

      As the solid curves indicate,  breakthrough is satisfactorily described
 by curves constructed from the o's  (standard deviation) and tn  's (break-
                                                              u »j
 through time  of 50 percent of input concentration) derived from the data.
 Table 1,  Column 3, indicates that the relative standard deviation a* (*0/tQ e)
                              25° Bath
FIGURE 1.  Dynamic uiluiion apparatus

                                                  Sampling pump
                                                            To exhaust
     80        100
       Time (min)
FIGURE 2.  Representative breakthrough experiments for the following  input
concentrations:  a) 6.31 mg/1, b) 4.68 mg/1, c) 3.08 mg/1, d) 2.79 mg/1,
e) 1.66 mg/1.

                         TABLE 1

Breakthrough Data  for Acrolein  and Water on  13X Sieves

c^ mg/L
Gravimetric Desorption GC
1.66a 2.45
2.79a 2.73 2.25
3.22* 3.68
3.53a .3.39 3.52
4.68 5.55
4.81* 5.11
5.08* 4.24
6.31* 6.94 6.42

A, mg/g
Gravimetric Desorption


169* 166

179* 197


a* = o/t0>5




F, L/min

2.0 •
         used in least squares fit to Langmuir isotherm.  Accuracy of results was assumed to be gravimetry > desorption
   > GC.

through data (Table 1) by a least squares fit to the best Langmuir  isotherm.
The fit is quite good, with a correlation coefficient of 0.94 and standard
error of estimate +_ 19 percent.  In addition, the calculated saturation  capa-
city (dashed line, Figure 3) of 236 mg/g is in good agreement with  the experi-
mental value of 226 mg/g determined in static experiments.

     With the assumption from TSM of concentration independent a*,  the break-
through curve for any concentration can be constructed from the corresponding
tQ - determined from the isotherm by the relation

                                        .£    —
                                 C0.5 " GI X F

where     a - capacity, mg/g
         c. « input, concentration, mg/1
          w * weight of sieve bed, g
          F - flow rate, 1/min
      and o « a* x tn e
                    u «->
     Calculated 5 percent breakthrough times for various acrolein concen-
trations sampled in dry air at 1 1/min with 1 g sieve beds are given in Table
2.  Since 2 ug acrolein/g sieves can be readily determined by GC analysis, the
sensitivity for acrolein in dry air is 0.01 ppm for an 80-liter sample.

     Because water will displace acrolein from the sieves, the effect of hu-
midity on acrolein breakthrough had to be defined.  Breakthrough of 1,500
ppm acrolein at 30 percent relative humidity superimposed on the water break-
through curve is shown in Figure 4.  Figure 5 shows results of a similar
experiment at 100 percent relative humidity.  The acrolein elutes from the
sieve bed with the water front in the form of a GC peak, having a variance
similar to the water breakthrough curve and a concentration maximum at ap-
proximately tn _ of water.  At 100 percent relative humidity, 5 percent
breakthrough occurs in *8 min.  With a sensitivity of 2 ug acrolein/g sieves,
a concentration of 0.1 ppm should be detectable in an 8-liter sample on 1 g of
sieves.  Since the capacity of sieves for water is constant to very low rel-
ative humidities, sample volume and therefore sensitivity will increase
directly in proportion to decrease in relative humidity.


     Formaldehyde is more difficult to handle.  To date, the recovery of








                          •i	^	1-
                                               ;l	1-	1-
                          I	|	|	|	|	|	|	|	|	

                         2.0   4.0   6.0  8.0  10O  12.0  W.O   16.0  18.0  200
              FIGURE 3.   Dynamic adsorption isotherm of  acrolein at 21°  C.
                                                       Time (rmn)

              FIGURE 4.   Acrolein  breakthrough, 30  percent relative  humidity.
              FIGURE  5.  Acrolein breakthrough, 100 percent relative humidity.

                                    TABLE 2

   Breakthrough Times of 5 Percent of Acroleln Input Concentration In Dry Air

                                                 t     (mln)
                      4.57                          20
                      2.29                          43
                      0.229                         72
                      0.023                         82
                      0.0023                        82
                      0.0002                        82
formaldehyde from sieves has been established when It Is loaded as vapor from
formalin headspace.  Relatively high concentrations were used so that analysis
of desorptlon solutions could be performed by GC on Carbowax 20 M with a flame
lonlzatlon detector.  The chromatograms of formalin and the headspace show
two peaks in addition to methanol (present as a preservative).  The major
peak, preparatlvely chromatographed and collected In water, gave a chromo-
phore with chromotropic acid and was assumed to be monomeric formaldehyde
(and/or the hemiacetal with methanol).  This peak was used as a standard for
determination of formaldehyde in desorption solutions.  The presence of the
monomeric aldehyde in headspace was qualitatively demonstrated by observing
the formaldehyde carbonyl absorbance bands in the UV on bubbling headspace
into heptane in a sample cuvette.

     Recovery of formaldehyde from sieves was determined by comparing total
formaldehyde content of distilled water through which a known volume of dilut-
ed headspace had been bubbled with that in desorbate from sieves loaded with
an equal volume of diluted headspace.  For loadings of 100 to 200 yg formalde-
hyde/g sieves, recovery was 98+5 percent; for 0.9 to 1.1 mg/g, 108 + 5 per-
cent.  It was also determined that formaldehyde in desorbate could be analyzed
colorimetrically with chromotropic acid, but only with desorption volumes >IQ
ml/g sieves.  More concentrated desorbate gave irregular color development
with yellow interference.  For a 10-ml desorption volume and the sensitivity
reported in Reference 8 for the final colorimetrie solution, 0.2 ppm formal-
dehyde could be detected In an 8-liter sample on 1 g sieves at 100 percent rel-
ative humidity.  We have assumed that breakthrough characteristics of form-
aldehyde will be similar to those of acrolein; however, work is currently in
progress to describe breakthrough behavior.

  FIGURE 6.  Gas  chromatogram of a 2.5-liter field  sample collected over 12 min at
  a  sampling rate of 208 ml/min.  The  following concentrations were calculated
  on the basis of a comparison of peak heights with those of standard solutions:
  1)  methanol, 110 ppm; 2) acetaldehyde,  10 ppm; 3)  ethanol, 3 ppm;  4) acrolein,
  16  ppm;  5) acetone, 66 ppm.


i 5.



                                          "  O
            ••SHORT Te?n EXPOSURE LIMIT-
                1     10       25    5C     75       90     99
                                                                      99.9  99.99X
 FIGURE 7-  Frequency  distribution of acrolein concentrations at  fires, total
 118  samples.

Field Measurements

     Molecular sieve sampling tubes were contained In a sampling package
deployed on the rescue squads of the Boston Fire Department.   A gas chro-
matogram typical of the sample desorbates is shown in Figure  6.  Formaldehyde
did not elute from Tenax-GC, but its presence was confirmed by chromotropic
acid analysis of the desorbate.  To our knowledge, these data are the first
confirmation of the presence of acrolein at urban residential fires.  The dis-
tribution of acrolein concentrations in fire samples, shown in Figure 7.
indicates that acrolein exposures are of concern as a potential hazard.
Formaldehyde was not present in sufficient concentration to justify allocation
of technician time for routine analysis.  Analyses were performed on a spot
basis on samples which indicated severe exposure to other contaminants.   Max-
imum recorded formaldehyde exposure was 11 ppm.

 1.  Anderson, R.A.  Mol. Sieves II.  ACS Symp Ser 40:637,  1977.

 2.  Barrer, R.M., Reucroft, P.J.  Proc R Soc London,  Ser A 258:431,  1960.

 3.  Grubner, 0., Zikanova, A., Ralek, M.  J Chromalogr 28:209,  1967.

 4.  Grubner, 0.  Adv Chromatogr 6:173, 1968.

 5.  Grubner, 0., Underbill, D.  Sep Sci 5:555, 1970.

 6.  Lee, H. Adv Chem Ser 121:311, 1973.

 7-  NIOSH Manual of Analytical Methods, 2nd Edition.   U.S. Department of
     Health, Education, and Welfare, Center for Disease Control,  NIOSH.
     Cincinnati, Ohio.  Publication #77-157-A P and CAM 118.

 8.  Ibid., P and CAM 125.

 9.  Poutsma, M.L.  Zeolite Chem and Catalysis, ACS Monogr 171:529, 1967.

10.  Venuto, P.B., Landis, R.S.  Adv Catal 18:259, 1968.

11.  Walker, J.F.  Formaldehyde.  New York, R.E. Krieger Co., 1975 Ch. 3, 5, 8


 Avram Gold, Ph.D.
 Harvard School of Public Health
 Kresge Center for Environmental Health
 665 Huntlngton Avenue
 Boston, Massachusetts  02115
      DEBBRECHT:   I would like to Just make one comment  and  possibly a  sugges-
 tion.  The one comment Is:   I think you are very justified  In watching out
 for the volatile fume that  obviously you became aware of.   You  had to  put
 these materials  on your molecular sieve when the vapor  condenses  to a  liquid

      I don't know how well  known this fact is, but  one  runs into  the same
 problem with charcoal tubes in industrial hygiene work.  It makes a difference
 between your efficiencies and recovery.

      The second  thing Is that certainly the flame ionization has  extreme im-
 portance to formaldehyde.  One possibility is using the  technique that is
 currently available on a GO system for carbon monoxide,  which is  also  not
 detected by flame.   And that is to convert the carbon monoxide  to methane
 at  the GC separation, when  that occurs, using a nickel and  hydrogen atmosphere.
 This  should certainly work  with formaldehyde and at least give  you methane-
 sensitive formaldehyde.

      GOLD:   Yes,  there are  several possibilities.   They  are interesting
 possibilities, and  Dr.  Lin  here has suggested,  for  example,  a photo ionization
 detector.   We  do  not have that capability and would not  have it in the near

      DEBBRECHT:   1  don*t think the photo ionization detector has  enough energy
 to  get  the  formaldehyde;  I  might be wrong.

      GOLD:   1  have  not  had  any personal experience  with  that.   I  understand that
 it  is high  enough in energy,  so that it would attach hydrocarbons, if  the
 formaldehyde is present  as  an aldehyde,  when it enters the  detector—we have
 some  evidence  that  it is.   We are very worried about the form of  the form-
aldehyde  because  of the  fact  that if it appears in  a gas stream as a mono-
mer—or anything  that has a molecular diameter smaller than the pores  of the
sieves—it won't  be collected;  it will pass  right through.   So, we were very
worried about whether we  had  monomer or polymer.  We had some interesting
experiences  with  the  effluents  from permeation tubes, using  paraformaldehyde.

     Also, there is a possibility perhaps of doing the analysis by high
performance liquid chromatography.  One has a very good chromophore in  the
dinitrophenolhydrazine group.  The concentration of the desorbate which
would occur in a prechromatographic step on some sub-packing could also be
used.  I don't think there is any doubt that the method could be improved

     I also have talked to other people who have had problems with the chromo-
trophic acid determinations.  We used it, because at least as far as our
verification experiments were concerned, we knew we were dealing with form-
aldehyde or something like formaldehyde.

     I can't remember whether I have said that we do have some evidence that
the formaldehyde in the headspace of the analyzer contains at least some of
the aldehyde, because we bubble the headspace vapor into heptane in a UV
spectrophotometer, and in fact, we're able to see the fine structure of
formaldehyde in the UV.

     So, at least some formaldehyde is present as monomer.

     MEIER:  I would caution you about extrapolating from high concentrations
to low concentrations.  Past experience on other studies that we have done
indicate that the small amounts or irreversible adsorption that do take place
on things such as molecular sieves become more and more significant as your
concentration becomes less and less, and can result in errors of 100 to 200
percent or false-negative results indicating nothing is there when indeed
there is.

     GOLD:  1 guess my answer to that would be that I think that for the ap-
plications which most people would be interested in, errors are tolerable of 100
percent at really low levels, if you are talking parts-per—billion.

     MEIER:  If you are talking about criteria level—and I am just using a
hypothetical case—of parts-per-million, and you irreversibly adsorb at that
level so that you are actually getting results of zero, it becomes very

     GOLD:  That's true.  At those parts-per-million levels we are OK.  We
can take enough of a sample so that irreversible adsorption in the small
pores, or whatever, of the sieves would not significantly impair our accuracy.

     If one worries about trace analysis, that is another story, and at this
point, I wouldn't care to comment.

Passive Membrane-Limited Dosimeters
Using  Specific Ion  Electrode Analysis
     Charles E. Amass
     Orion Research, Inc.
     Cambridge, Massachusetts

     In recent  years, many workers have described  systems using permeation
techniques for  detecting and determining toxic gases in the atmosphere.
Kenneth E. Reiszner and Philip W. West (8) described a method using  a  sill-
cone rubber membrane with a liquid collection medium for measuring sulfur
dioxide at ambient levels (7).

     Similarly,  B.I. Ferber, F.D. Sharp, and P.W.  Freidman at the Bureau of
Mines used a silicons membrane, a liquid collecting medium, and an ion spe-
cific electrode readout for determining oxides of  nitrogen (2).

     Methods have been developed for vinyl chloride, carbon monoxide,  nitrogen
oxide, sulfur dioxide, hydrocarbons, and ammonia utilizing diffusion prin-
ciples (1-8,10).  This report describes the development of a system  for
detecting a number of inorganic gases in the air and a rapid method  for de-
termining their levels.

     In designing  the systems, a number of requirements were formulated  as
being necessary for successful performance.  The membrane must be mechanically
strong, uniform, and be free of reaction with solution chemistries.  The

  membrane permeability must be uniform from piece to piece and slow enough to
  be  the limiting layer;  that is,  free from effects of changes in flow pattern
  around the dosimeter.  In addition,  the permeability should be relatively low
  for water (and high for the contaminant), combined with a rapid time response
  for integrating peak values,  and low sensitivity to temperature change.   Per-
  haps most importantly,  the membrane/solution combination should exhibit
  linearity with respect  to both time  and concentration.

       The solution chemistries must be stable for 6 months before exposure,
  and the stoichiometrically formed products must be invariant for at least 2
  days.   A rapid and complete reaction to form the product is necessary to ensure
  proper integration—that is,  maintain a zero partial pressure for the contami-
  nant within the dosimeter.  In addition,  the reagent and analytical method must
  be  selective,  sensitive, and  capable of covering a wide concentration range.

  Theoretical Background

       Although  much attention  has been given to a theoretical model for geo-
  metrically limited diffusion  for dosimeter designs (3,5,6,9), no work has
  clearly outlined the theoretical design criteria for a  successful membrane-
  limited dosimeter.

       The past  several years have seen a rapid advance in gas-permeable mem-
  brane  technology.   This technology has been applied with great success at
  Orion  Research,  Inc., in the  development  of a large number of gas-sensing
  electrodes.  The use of these membranes in dosimeters,  while apparently  quite
  different,  involves the same  problems and theoretical analysis which were
  developed for  the  gas-sensing electrodes  (9).

      On exposure to an  atmosphere containing the sensed species,  a steady
  state  will  be  attained  in a short time (less than 1 minute), and steady  state
  solutions to Pick's Law of Diffusion can  be applied.  Concentration gradients
  of  a diffusing  gas under steady  state conditions are shown in Figure 1.
  There  are three  diffusion phases:  a boundary layer outside the membrane of
  thickness 1D which depends on the aerodynamics of the sample atmosphere;  the
  actual  membrane  whose thickness  is 1..;  and an internal  absorption liquid.
 Within  any  phase,  the concentration  gradients are linear,  and the flux of the
 diffusing species  per cm  is  given by:

                             »b-*f— 'm—H
                                    k. rm
FIGURE 1.  Membrane-limited  diffusion.model.  Cs atmospheric concentration, 1.
boundary layer,  1   membrane  layer, Ca concentration in boundary layer  at mem-
brane surface,  Cm  concentration  in membrane at outer surface.
where F is the flux in moles/cm /sec
      0 is the diffusion coefficient in cm /sec
      1 is the phase thickness in cm
      AC is the concentration difference between the phase boundaries  in
      moles/cm .

     A concentration discontinuity may exist at the outer surface of the mem-
brane because of the partition of the diffusing gas between air and  the mem-
brane phase*  Such partition reactions are rapid for most gases, and it is
permissible to assume that equilibrium exists at the interface;  i.e.,
                                  C  = k C
                                   m      a
where C  and C  are the boundary  concentrations of the diffusing gas  in the
       m      a
membrane and in air, respectively, and k is the partition coefficient (i.e.,
the Henry's Law constant for the  gas expressed in dimenslonless units).

       At the membrane-absorbing solution Interface, the concentration of the
  diffusing gas is zero, since the absorbing solution is designed to contain
  a sufficiently high concentration of a fast-reacting species to destroy every
  molecule of diffusing gas which reaches the absorbing solution surface.

       The overall flux of diffusing gas into the dosimeter can be obtained
  by combining Equations 1 and 2 and noting that the flux in the boundary layer
  must be equal to that in the membrane in the steady state.

                                      Dmk         C,                    [3]
                                   1m + D kl.
                                         m  b

  where D  and D  are the diffusion coefficients of the gas in the membrane
  and in the air, respectively, and C  is the sample concentration in moles
        3                            ®
  per cm .

       If the gas reacts with the absorbing solution to produce s moles of
  product per mole of gas, then the final concentration of the product after
  an exposure time t is
                                    P- v
                                             Fdt                        [4]
 where V  is  the  volume of the absorbing solution,  and C  is the concentration
 of  the product  sensed by the ion specific electrode in the absorbing solution.
 Thus, C   is  proportional to the integrated exposure to the reacting gas
 concentration C  in the environment.

      The constant  of  proportionality in Equation 5  includes a term involving
 1. , the boundary layer  thickness.   In normal use, the dosimeter will be ex-
 posed to atmospheres  which vary  from completely static to high flow-rate
 situations.  As a  consequence  of the variable rates of stirring, the effective
 boundary layer thickness JL will vary.   Virtually no design data are avail-
 able regarding air flow patterns in the Immediate neighborhood of a human


subject, and exact predictions of 1, are therefore not possible*  Measurements
In our own laboratory, however, under static conditions yield values of 1,
on the order of 10   cm.  Extrapolations of published engineering data on
"gas side resistance'' in absorption columns suggest that 1, has a value around
10   at flow velocities of 1 mph, and that this thickness decreases rapidly
with further increase in flow velocity.  It was necessary, therefore, to
select a membrane with appropriate  values of 1 , D , and k, such that
                                              m   m

                       m  *. ,n  b   for 10 percent accuracy            [6]
                      D~k ? 10 D~
                       m        a

Values of D  for gases of intermediate molecular weight (20 to 100) in air at
atmospheric pressure at room temperature are on the order of 0.1 cm /sec,
and for the membrane, D^k = 2 x 10   cm /sec for SO-, MR., and H2S and
1  " 2.54 x 10*  cm.  Thus, substituting for D , D k, 1 , and for 1. , the
 m                                     ^i     a'  m *  m'          b*
worst case static condition value of 10   cm, we have:

                            2.54 x  10"3     IQCIO"1)                  [7]
                            2 x 10~4          .1
                              thus 13>10

The condition on accuracy imposed by Equation 5 is therefore met, and dosi-
meters of the type proposed can operate with negligible interference from
variable aerodynamic states in the environment.  Laboratory tests have con-
firmed this result.

Sulfur Dioxide System

     An acid buffer assures  that all sulfur dioxide is trapped on entering
the solution in  the form of  bisulfite ion.

                         H20 + SO, + NaX •*• HX + NaHSOg                 [8]

Sulfur dioxide is complexed  in such a way  that an ion which can be sensed by
electrode  is stoichiometrically released.

Ammonia, Hydrogen Fluoride,  Hydrogen Chloride

     The dosimeter filling solution consists of a buffer which provides  the


  necessary trapping pH and antibacterial action.  The pH is on the acid side;
  this converts all ammonia entering the dosimeter into ammonium ions.  An am-
  monium specific ion electrode using a macrotetralide antibiotic as the ion
  selective element is used in combination with a single Junction reference
  electrode to complete the measurement cell.  Similarly, hydrogen fluoride
  or hydrogen chloride are also trapped at this pH, and readout is by means of
  the corresponding ion specific electrode.

  Hydrogen Sulfide

       The approach taken is to utilize the extreme insolubility of a metal
  sulfide, even in the presence of an enormous excess of a strong complexing
  agent.  Hydrogen sulfide dissolves to form hydrosulfide ion which reacts with
  metal ion.  An electrode sensitive to the metal ion is used for the readout.
  This method is a so-called subtractive one, and in order to obtain a usable
  balance between sensitivity and range, the metal concentration of the dosi-
  meter is set at two times the expected consumption for the point of maximum
  required accuracy.

  Hydrogen Cyanide

       The hydrogen cyanide filling solution consists of a known amount of a
  cyanide complex dissolved in a buffer at a high pH.  At this pfl, any hydrogen
  cyanide entering the solution will form cyanide ion.  Cyanide ion further
  complexes the small amount of uncomplexed metal in the solution in proportion
  to the cyanide added.  The readout cell consists of a metal ion electrode and
  a  reference electrode.

  Dosimeter Design

       The prototype dosimeter units used in our experiments are shown in Fig-
  ures 2A and 2B.  All pieces except the 0-ring were machined from linear poly-
  ethylene.   The 0-ring is solid Teflon.

       The diffusion-limiting membrane construction is still proprietary at
  this time.   The dosimeter is designed to contain 2 ml of absorbing solution,
  which leaves a small headspace.  Full contact with the membrane is not neces-
  sary for accurate integration.

2.2ml VOLUME
1.247" DIA.x
.115" DEPTH
                                                     SECTION A-A
                                           TEFLON -0-RING
                                           .115" DEPTH
                                           1.247" DIA.
                                           2.2cm3  VOLUME
 FIGURE 2.  Prototype dosimeters:   (A) with grid over membrane and locking no-
 turn washer, and  (B) simplified design incorporating Teflon 0-ring.

  Equipment for Generating Test Atmosphere

       A schematic diagram of  the  test equipment is shown in Figures 3A and 38.
  Permeation tubes were used to generate  test atmospheres for sulfur dioxide,
  ammonia, and hydrogen sulfide.   For hydrogen chloride, a gas phase permeation
  device was used.  The gas phase  permeation device consists of a stainless
  steel cylinder with connections  to a HC1 tank.  Nitrogen gas or air is passed
  through a length of porous Teflon tubing which enters the cylinder and exits
  from it.  For this type of device, an air oven was used for temperature control
  because of the danger of water contamination and the relative insensitivity of
  the gas phase device to temperature changes.

       The dosimeter test chamber  consisted of a pyrex autoclave (2,200 ml) with
  the seal modified by incorporation of a large 0-ring and retainer.  A blending
  device (Figure 3B) allowed mixing either dry or humidified diluent air with
  the permeation tube output, and  then splitting the gas stream into two parts
  to allow for varying the flow rate through the chamber at constant concen-
  tration.  All connections were made with ball and socket joints with a thin
  film of vacuum grease.  Bubblers on all downstream lines were provided to give
  the complete integrated analysis of the contaminant atmosphere concentration
  during the course of the experiment*

       The entire chamber was placed inside a constant-temperature air oven
  to allow temperatures in the range of 0° to 50° C to be generated.  It was
  found that even a short length of tubing would equilibrate the sample gas to
  the temperature of the test chamber.  Humidification was supplied by passing
  compressed air through a bubbler in a water bath at reduced temperature and
  then through a suitable filter to remove micro droplets, or by a diffusion
  tube technique.


       A volume of the dosimeter solution is pipetted into the dosimeter body
  and sealed in place.   The dosimeter is then stored with a protective cover
  until  exposure.   After exposure,  the cover is removed and a pair of electrodes
  are placed into  the solution.  A potential is developed between the electrodes
  which  can  be  measured  by a specific ion electrode meter.  A calibration curve
  is  constructed  (Figure  4)  by  adding known amounts of the ion of interest to
  a  sample of  the  dosimeter filling solution.   A stable reading Is usually ob-
  tained in  less  than 1 minute.  In the hydrogen sulfide system,  solutions are

                                                                                                                       VENTED TO
                                          ALL THERMOMETERS
                                          .fC DIVISIONS)
                       HIGH PRECISION
                                                                                                                  SOAP BUBBLE
                                                                             TEST CHAMBER
                                                                                                      BUBBLER STAGES
        WATER BATH «
                                                . o o . o
                                          0 AIR OVEN
                                                                       m i mi • mt i §• i !• • tm i §• im
                                                              DRYING TUBE
                                                                                                              LIQUID TRAP
        FIGURE 3A.   Test apparatus—permeation  tube and temperature control system.

                                        GAS BLENDER
                                                   SPLIT STREAM OR
                                                   BUBBLER TAKE-OFF

                                         WATER BATH
 FIGURE 3B.  Test  apparatus—gas blending  and humidification  system.
+70    +90   +110
                                                     30   *I50   +170
FIGURE  4.   Electrode  calibration curves  for sulfur dioxide  (0)  -1 x 10   M,
reagent (background for current OSHA standards); (a) 1 x 10  M. reagent
(background for current EPA standards).

prepared with progressively lower concentrations by diluting the  filling
solution with ion buffer.


     Regular checks of permeation tube output are compared to the bubbler-
integrated value to determine system holdup or leaks.  Calculation of ppm
exposure is based on the bubbler values corrected by the amount of gas trapped
by the dosimeters and the volume of gas remaining in the chamber which does
not pass through the bubblers.  A permeation rate was calculated according to
the formula

   P « 1   lO^ (ml gas [RTF]) x (cm thickness of membrane)	
               (seconds exposure) (cm  area membrane) (cm Hg partial pressure
               of contaminant - 0).

Permeation rates calculated in this fashion should be constant, and indepen-
dent of variations in time, concentration of exposure, and pressure.  Once
such permeation values have been obtained for a series of different exposure
parameters, the range in these values is reduced using an average permeability
to decrease the error for measurements under unknown conditions.  Thus, in
actual use, an average permeation value is selected for the particular gas of
interest, and that value is used as a constant to convert the moles of species
found in the dosimeter to ppm-hours.  Dividing by the exposure time then gives
the average ppm level.

Membrane Selection

     The membrane used  offers  a  significant  advantage over  the  silicone mem-
branes used  by  other  workers.  Although  it is  somewhat  less permeable  than
silicone rubber membrane,  we gain the  advantage  of  less interference from
windage effects.  In  addition  to this, however,  the membrane's  mechanical
properties allow it to  be  used without the individual calibration  of each dosi-
meter.  If reasonable care is  used in  assembly,  the values  obtained with a
sample of the membrane  are reproducible  without  prior calibration.  All that  is
necessary to know is  the membrane area and the solution volume.  This  membrane
area-to-solution volume ratio  is the only determinant of the concentration  to

  be  found  In  the  dosimeter,  for  a given  exposure  time and concentration of gas.
  This  substantial advantage  allows mass  production of the dosimeters without
  individual testing.  Therefore, they can be disposable, since their manufacture
  is  not  labor intensive.

        All  data in this  report were obtained using a  fresh piece of membrane
  for each  exposure.  Thus, any variation in membrane permeability due to thick-
  ness  or composition is inherent in  the  data.


       System  linearity  is a  function of electrode slope and the permeation
  rate variation with  concentration, time,  temperature, or humidity.  Typical
  calibration  curves are shown in Figure 4  for  the ion selective electrode
  used with  the SO- filling solution.  Figures  5A, 5B, and 5C show linearity
  with respect to  time at various selected  constant concentration levels for the
  ammonia, hydrogen chloride, and hydrogen  sulfide systems.  The regression line
  calculated for the hydrogen sulfide system  suggests a high blank.  Actually,
  the deviation is caused by  a slight fall  off  in permeation rate at longer
  exposure times (discussed below).  Each point represents a group of from four
  to six dosimeters.

       Concentration linearity at various typical constant exposure times are
  shown in Figures 6A, 6B, and 6C) for the  ammonia, hydrogen chloride, and hydro-
  gen sulfide systems.   The regression line for the hydrogen sulfide system
  was calculated using only values below 25 ppm.  The high exposure levels
  produce a falling off  in permeability, probably due to back diffusion (see
  discussion below).

       An overall response to variation of  temperature, concentration, and ex-
  posure time is shown for sulfur dioxide in Figure 7.  This graph of concen-
  tration of exposure versus  concentration  predicted by the dosimeter shows a
  close correlation with the  theoretical 45° slope.  Reproducibility of the
  systems is indicated in Table 1, where statistical data are presented for each
  of the systems currently tested.  For this table, an average permeation rate
  was selected based on  the entire group data,  and the value was then used to
  calculate ppm levels.

                      O MEAN

I                        ERROR RANGE AT THE 95%
                        CONFIDENCE INTERVAL FOR A
                        GROUP OF 4-6 DOSIMETERS
                7 2
                                      TIME (MRS.)
FIGURE 5A.   Time  linearity—exposure time versus  dosimeter concentration,
5 ppm ammonia 20°  C.
                       O MEAN

                       T ERROR RANGE AT THE 95%
                        CONFIDENCE INTERVAL FOR A
                       1 GROUP OF 4-6 DOSIMETERS

FIGURE 5B.   Time  linearity—exposure  time versus dosimeter concentration,

10  ppm hydrogen chloride 20° C.
                        ERROR RANGE AT THE 95%
                        CONFIDENCE INTERVAL FOR A
                        GROUP OF4-6 DOSIMETERS
FIGURE 5C.   Time linearity—exposure  time versus dosimeter  concentration,^
20  ppm hydrogen sulfide 20° C.

                     o 5
                          O  MEAN
                             ERROR RANGE AT THE 95%
                             CONFIDENCE INTERVAL FOR A
                             GROUP OF 4-6 DOSIMETERS
   FIGURE 6A.  Concentration linearity—exposure time versus  dosimeter  concen-
   tration, 1 hour ammonia*
                            ERROR RANGE AT THE 95%
                            CONFIDENCE INTERVAL FOR A
                            GROUP OF 4-8 DOSIMETERS
   FIGURE  6B.  Concentration linearity—exposure time  versus  dosimeter  concen-
   tration,  4 hours hydrogen chloride.
                          O MEAN
                            ERROR RANGE AT THE 95%
                            CONFIDENCE INTERVAL FOR A
                            GROUP OF 4-6 DOSIMETERS
  FIGURE 6C.   Concentration linearity—exposure time versus dosimeter concen-
  tration, 8 hours hydrogen sulfide.


                   I 20
       O MEAN

                                   12   16   20   24   26
                                   EXPOSURE CONCENTRATION (PPM)
FIGURE 7.   Overall  system  response for sulfur dioxide—test atmosphere  concen-
tration versus dosimeter calculated value  using  average  permeation rate.
                  1 200

                  d 160
    - O
                                        & 185
               O 164
                                          D(2.9»IO"* )
                         O.I   0.25 09   I
                                               I  I I	L
                                               10 1520      100
                                         TIMEIMIN )
FIGURE 8.   Response times for hydrogen  sulfide.   (0) 2.5  x 10    M. metal Ion,
(Q)  2.5 x  10~* M.  metal ion, (A)  2.5 x  10   M. metal ion. (•)  dilute  buffer,
(  )  concentrated buffer.

                          J	L
                                     O MEAN
                                     T ERROR RANGE AT THE 95% CONFIDENCE
                                      . INTERVAL FOR A OROUP OF 4-6 DOSIMETERS
                                            9   «
                                          TIME (MRS)
 FIGURE  9.  Membrane  response for hydrogen  sulfide—exposure time versus  ppm-
 hour found in dosimeter.                                                        451

                                      TABLE  1

                                Systems  Performance
S02 a:
H S cr
HC1 d:
-10.8% to 10.0%
-12.9% to 5.1%
-11.6% to 5.2%
-11.1% to 5.5%
  a =  four  exposures  of  five  dosimeters  each  at 0% humidity  range, 1 to 9 ppm
      range and  20° to 40°  C  temperature range.

  b =  four  exposures  of  five  dosimeters  each  at 0% humidity  range, 5 to 70 ppm
      range and  20° to 40°  C  temperature range.

  c =  18  exposures of five  dosimeters  each  at 0% humidity  range, 4 to 210 ppm
      range and  20° C temperature  range.

  d =  12  exposures of six dosimeters each at  0% to 45% humidity range, 1.5 to
      26  ppm  range and 5° C to 40° C temperature range.
 Response Time

      Other  investigators have  suggested  response  time for silicone rubber
 membrane in the range of 30 seconds  to 10 minutes  for various gases (1,5,8).
 Hydrogen sulfide was selected  for  response  time tests since  the solution used
 adapted itself well to adjustment  for quantitating low concentrations, and the
 OSHA ceiling limit for H_S requires  a sampling time of under 10 minutes.  A
 tank of 210 ppm hydrogen sulfide in  nitrogen was used to generate the test
 atmosphere  in the test chamber.  A 28/15 socket on the test chamber was adapt-
 ed to an earlier prototype dosimeter by means of a standard clamp.  Dosimeters
 could be placed over the orifice and clamped in place for a set period of time.
 When a dosimeter was about to  be placed over the orifice, the exiting gas (300
 ml/min) would ensure that rapid contact between the test atmosphere and the
 dosimeter membrane would occur.  A bleed was provided for escaping gas during
 exposure.   For the extremely low levels, multiple exposures of the dosimeter,
 with a relaxation time between exposures, were used to build up a sufficient

level of the sulfide in the test solution.  Figure 8 illustrates  the  results
of these tests.  It should be noted that the extremely short-range  tests
were done with ion buffering at two concentrations.  The lower values  for
permeation rate at short times can be related to back diffusion of  H.S  from
the dosimeter before reaction could occur, since higher permeation  rates were
obtained over the same exposure times (.1 to 1 minute) by adjusting the buffer
concentration, which, by reducing complexation, raises the available metal
ion concentration to that used for times from 5 to 20 minutes.  This change
in concentration increases the rate of reaction with hydrogen sulfide dis-
solving in the solution.  The high gas concentration utilized, over 20  times
the TLV, could readily exhaust a thin film of dilute metal ion concentration
near the membrane.

Trapping Efficiency

     As Indicated above, for accurate integration the partial pressure  of
the contaminant in the dosimeter must be zero, or substantially 100 to  1,OOOX
lower than the partial pressure of the contaminant in the atmosphere.   The
hydrogen sulfide system Indicates some of the problems that can be  encountered.
With one membrane construction, a somewhat erratic and large variation  in
permeation rate is found, as shown in Figure 9, where exposure time versus
concentration found in dosimeter is graphed for 25 ppm H.S.  The deposition
of a precipitate on the interior of the membrane causes a decrease  In permea-
tion with time.  Modification of the membrane construction corrected most of
the change, as shown in Figure 5C, although, as pointed out above,  some de-
crease is noted at the high end.  It should be noted that the ppm-hours for
typical government limits occur In the linear portion of the curve.

     In addition, the hydrogen sulfide system showed some lowering  of  trap-
ping efficiency with high concentration of H_S at short exposure  times, as
shown in Figure 10, where the variation of permeability as a function of ppm
level at constant ppm-hours is shown.  High levels of H.S probably  exhaust a
thin layer of buffer solution near the membrane surface and cause measurable
back diffusion.  The effect is minor in light of the low standard of accuracy
proposed for dosimeters.


     Figure 11 shows the water loss  from  the  dosimeters for an 8-hour  exposure
period at various  relative  humidity  levels as  a percentage of total solution
volume.  Unlike systems  that  use  solid absorbents, one of  the main  limitations

                   £ 200


                          I    I    I    I   I    I    I    I
                          10  20  30   4O   50
                                                70  80 *2IO
                                     PPM H2S
FIGURE 10.   Membrane  response—ppm  exposure  level versus  membrane permeability
at  constant  ppm-hours.
                                  I3*C   ZO-C   Z5'C
FIGURE 11.  Water loss  of prototype  dosimeter—temperature versus percent
water loss.   (0)  0 percent relative  humidity,  (D) 40 percent relative  humidity,
(A)  100 percent relative  humidity.   Membrane area 8.6 cm  ,  solution volume
2.2  milliliters.
                i 300
                       O MEAN
                       T  ERROR RANGE AT THE 95%
                         CONFIDENCE INTERVAL FOR A
                   ,[_   1  GROUP OF4-6 DOSIMETERS
                        5*C    IO«C   IS'C   20"C   25*C
                                                      ,	i _.
 FIGURE 12.   Permeability of hydrogen chloride—temperature versus permeati

of the Orion system is water loss with a resultant change in dosimeter com-
position.  Even under the most rigorous conditions, water loss is acceptable.
In addition, a simple means of correction can be provided by weighing the
dosimeter at the end of the exposure period and correcting the concentration
by an appropriate factor.

     The literature has indicated that membrane permeabilities can be affected
by humidity.  The hydrogen chloride dosimeter was selected as a worst case
test for humidity problems with the dosimeter.  The average permeability of
all hydrogen chloride tests was 440.  For a test at 40 percent relative hu-
midity, the permeation rate was 415~a small difference from the average value.

     It might be noted that the dosimeters themselves can generate local
humidity levels in the chamber.  Flow rates were kept high in mc^t tests to
keep the relative humidity below 5 percent.

Windage Effects

     As noted above, the dosimeter theoretically would have very little wind-
age effect.  Tests on the sulfur dioxide system at face velocities of between
30 to 16,000 cm-per-minute showed no effects, other than slightly higher
H_0 loss.  A test on the hydrogen sulfide system in which dosimeters were
twirled in the chamber at speeds of about 5 miles-per-hour showed no differ-
ence with those exposed to a flow of 0.001 miles-per-hour.  On the other hand,
dosimeters using microporous Teflon as the only diffusion-limiting layer
showed erratic response in a similar test.

     Dosimeters containing a grid (Figures 2A and 2B) did show about a 10 per-
cent across-the-board decrease in permeability versus the dosimeter without
such a covering (Figure IB).  This effect is ascribed to an increase in L, ,
the static boundary layer (discussed above).

Temperature Effects

     The literature has  indicated that the effect of the permeation rate with
temperature varies with  the gas of  interest.  Reiszner (8) indicated sulfur
dioxide  to have a negligible  temperature coefficient in silicone rubber, for
example, while carbon monoxide (4) had a large  temperature dependence.

     The hydrogen chloride system permeation  rate at constant ppm levels
versus temperature  is  shown  in Figure 12.  The  variation  is within other

                                     TABLE 2

                     Permeation Rate Variation with Temperature
            FOR:	Px	5°  C	20° C	30° C	40° C

            S02         330        -         285           -        370

            NH3         290        -         300           -        275

            H2S         220                 220           -

            HC1         440       530      >440          475      <440
 experimental variables.  Less  data  is  available  on  other  systems.  Spot tests
 are  summarized  in Table  2,  where  permeabilities  at  various  temperatures are
 shown versus  the average permeability  for  the  systems.


      The hydrogen sulfide  system  was designed  to avoid  interference by HC1.
 HCN would  be a  strong  interference  for this  system.  The  H_S dosimeter was
 tested  in  ozone at 2.5 ppm  for 8  hours and nitrous  oxide  at 22.8 ppm for 8
 hours with no change in  the dosimeter  solution.

      The ammonium electrode is somewhat sensitive to other  cationic species*
 Hydrazine  and other low-boiling amines can pass  through the membrane, but  no
 data are available at  this  time on  this interference level.  Buffering pre-
 vents changes in the hydrogen  ion concentration—the only other significant
 variable caused by dissolution of acid gases.

      The sulfur dioxide  system is sensitive  to hydrogen cyanide and hydrogen
 sulfide gases,  both of which can  liberate  the  sensing ion.  However, trap-
 ping efficiency for these gases will be very low due to the low pH of the
 trapping solutions relative to the  pKa's.  >

      The hydrogen chloride  system would be affected by  hydrogen cyanide,
 hydrogen sulfide, and ammonia.  Hydrogen sulfide and hydrogen cyanide, however,
 would not be trapped efficiently  by the buffer.  Ammonia  would be trapped  as
 ammonium ion, which does not interfere.

     The hydrogen cyanide system would have a slight  Interference  from hy-
drogen sulfide.  The hydrogen fluoride system has no  Interferences.

Storage Stability

     A significant problem in designing dosimeter systems of this  kind is  the
necessity for extremely dilute reagents and standards to remain stable for
periods of 6 months or more.  All the solutions used  in our program have been
tested in polyethylene for at least 6 months, including the standards  which
are generally at the 10*  1(T  M. level.  It is hoped that some of the solutions
may be stable for as long as a year.  The main limitation is water loss in
the dosimeter, which, due to the experimental nature  of the dosimeter  pro-
totypes, has not been significantly tested.  A good seal is all that is re-

Limit of Detection

     The limits of detection for the current dosimeter systems are outlined
in Table 3.  Being of special interest to ambient air quality, an  attempt
was made to extend the sulfur dioxide system to its ultimate limit within
the confines of the current geometry.  By adjusting the internal solution
concentration downward, the sensitivity was increased as shown in  the  cali-
bration chart, Figure 4 above.  It is possible and convenient to quantitate
at the .1-ppm level for a 24-hour exposure or at the  .5-ppm level  for  a 3-
hour exposure, as suggested on the graph by the indicated levels of concen-
tration to be expected.

     The sensitivity of the SCL system can be increased by a factor of 2 by
heating the dosimeters.  Water loss at low humidities and high temperatures
would put a time limitation on the device, for 24-hour exposures.

     It might be possible to reduce water loss for a  special application by
use of a nonaqueous solvent or addition of neutral salts to the dosimeter

     A passive membrane dosimeter has been developed  that  requires  no  individ-
ual calibration,  is  lightweight,  durable,  and  has  no  moving  parts.  When com-
bined with various  solution chemistries and specific  ion electrodes, the dosi-
meter provides a  system for integration of trace quantities  of  hydrogen sul-
fide, sulfur  dioxide,  ammonia,  hydrogen cyanide, and  hydrogen fluoride.   The

                                      TABLE 3

                                Sensitivity Limits*
                 pom hours
  S02               .25

  HC1              1.6

  NH                 R

  H2S             16d

  a)  using 2.0 ml solution and a membrane area of 8.6 cm
  b)  1 x 10   molar reagent
  c)  1 x 10   molar reagent
  d)  2.5 x 10   M. reactant

  e)  2.5 x 10   M. reactant
  analysis method is simple enough for less skilled personnel and can be done
  in a few minutes.   The basic method can be extended to carbon dioxide and
  other gases amenable to electrode or other methods of analysis.

       Exposure tests for various concentration, time, humidity, flow,  and
  temperature conditions have been performed on hydrogen sulfide, sulfur dioxide,
  ammonia, and hydrogen chloride with results more accurate and sensitive than
  the current monitoring procedures required in government standards.

       The  author wishes to acknowledge the work of other past and present  Orion

staff members:  Martin Frant, who originated the concept and developed many
of the chemistries; James Ross, who provided the theoretical work on diffusion;
Jon Soderberg, responsible for development of the hydrogen sulfide and sulfur
dioxide systems; and Nelson Perry, who patiently carried out the test program
for the other gases and provided the graphs for this manuscript.

 1.  Bell, D.K., Reiszner, K.D., West,  P.W.  A Permeation Method  for the De-
     termination of Average Concentrations of Carbon Monoxide in  the Atmos-
     phere.  Anal Chim Acta 77:245-254, 1975.

 2.  Ferber, B.I., Sharp, F.A., Freedman,  R.W.  Dosimeter for Oxides of  Ni-
     trogen.  Am Ind Hyg Assoc J 37:32-36, 1976.

 3.  Frant, M.S., Ross, J.W., Riseman,  J.H.  Electrode Indicator  Technique
     for Measuring Low Levels of Cyanide.   Anal Chem 44:2227-2230,  1972.

 4.  Mazur, J.F., Bamberger, R.L., Podolak, 6.E.,  Esposito,  G.G.  Develop-
     ment and Evaluation of an Ammonia Dosimeter.   Am Ind Hyg Assoc J 39:
     749-753, 1978.

 5.  Nelms, L.H., Reiszner, K.D., West, P.W.  Personal Vinyl Chloride Moni-
     toring Device with Permeation Technique for Sampling.  Anal  Chem 49:
     994-998, 1977.

 6.  Palmes, E.D., Gunnison, A.F.  Personal Monitoring Device for Gaseous
     Contaminants.  Amer Ind Hyg Assoc J 34:78-81, 1973.

 7.  Palmes, E.D., Gunnison, A.F.  Personal Sampler for Nitrogen  Dioxide.
     Am Ind Hyg Assoc J 37:570-577, 1976.

 8.  Reiszner, K.D., West, P.W.  Collection and Determination of  Sulfur
     Dioxide Incorporating Permeation and West-Gaeke Procedure.   Environ
     Scl Technol 7:526-532, 1973.

 9.  Ross, J.W., Riseman, J.H., Krueger, J.A.  Potentiometric Gas Sensing
     Electrodes*  In:  International Symposium on Selective Ion-Sensitive
     Electrodes (Moody, G.J., ed.).  Cardiff, The  University of Wales, 1973,
     pp. 473-487.

10.  Tompkins, F.C., Jr., Goldsmith, R.L.  A New Personal Dosimeter for  the
     Monitoring of Industrial Pollutants.  Am Ind Hyg Assoc J 38:371-377,


 Charles E.  Amass
 Orion Research, Inc.
 380 Putnam  Avenue
 Cambridge,  Massachusetts  02139
      SCHEIDE:   Do you see any advantages  that you would have on using this
 liquid solution collecting medium  as opposed to a solid collector?

      AMASS:  We could use a solid  collector.  We just felt that from the point
 of view of analysis,  it  would be easier.  We were looking, possibly, even at
 having field people do this work rather than having to train personnel to do

Personal Monitoring by Means of Gas
     Phlljp W. West, Ph.D., and Kenneth D. Reiszner, Ph.D.
     Environmental Sciences Institute
     Louisiana State University
     Baton Rouge, Louisiana
     There is an obvious need for personal monitors that are suitable for
determining exposures to toxic agents  in  residential areas as well as in the
workplace.  For such broad applications,  convenience and low costs, together
with sensitivity, accuracy, and reliability, are prime requirements.  Passive
monitors are especially attractive because they can be produced at a relatively
low cost and are the ultimate in simplicity.  Passive devices for sample col-
lection employing gas permeation through  polymer membranes are especially
attractive because environmental variations Induce little if any error,  and
the total analytic process can be so designed as to accommodate the study of
almost any and all gaseous hazards.

     Although permeation-type monitors are the subject of this discussion,
there are other passive systems that have merit for certain applications.
Monitors based on effects response can be used where qualitative results are
sufficient and low cost and convenience are important.  Although quantitative
sampling is not possible, relative concentrations are indicated by the degree
of the induced effect.  A second type  of  passive sampling is that based on
the use of evacuated vessels.  With such  devices, a known volume of sample
can be collected by simply opening the evacuated bottled or flask in the
atmosphere under study to provide a grab  sample.  Such an approach can also
be used for the collection of integrated  samples over a predetermined time
if the sample inlet is provided with a rate-controlling orifice.  Evacuated
vessels are limited as to the volume of sample that can be collected and
therefore lack the necessary sensitivity  for  most work that is involved  in
personal monitoring.


      Diffusion has been introduced as a convenient means of sampling.   The
 diffusion technique was first introduced as a means of preparing standard
 gas atmospheres (1), and where conditions can be kept reasonably constant,
 such as in the laboratory, this approach has proven to be quite successful
 (2,3).  For the preparation of standard gas mixtures, the liquid form of the
 reference material is allowed to diffuse into a dilute matrix system.   The
 resultant conversion of a concentrated reference material to a dilute standard
 mixture can be reliable and convenient.  The application of diffusion to the
 reverse problem—that is, the sampling of a dilute system with the quantitative
 collection of a minor constituent—has been advocated by the 3M Company
 and the Abcor Development Corporation, as well as by the British Defense
 Ministry at Porton Down.  To date, the diffusion-type samplers have employed
 adsorption for the capture of the diffused sample.  After the sample is col-
 lected, it is desorbed by suitable means and determined by conventional an-
 alytic processes such as gas chromatography.  The badge is usually discarded
 after a single sample collection.  In our experience, sample collection by
 means of diffusion is subject to some degree of error due to variations in
 humidity as well as by variations in the relative rates of movement of the
 collection device and the ambient atmosphere.

      Permeation was introduced by O'Keeffe and Ortman (4) in 1966 as a means
 of preparing  standard gas mixtures.  This technique has been widely accepted
 as a method of choice for accurately Introducing known amounts of sample gas
 into appropriate volumes of diluting atmospheres.  The reverse of this process
 was introduced by Reiszner and West in 1973 (5) as a means for quantitatively
 collecting sulfur dioxide from ambient atmospheres.  In this approach the
 ambient air was sampled by means of a simple cell containing a solution of
 tetrachloromercurate(II),  stoppered at the top and enclosed on the bottom
 by means of a  permeable membrane of a silicone polymer.  The permeated 802
 was absorbed  and fixed as the sulfItodichloromercurate(II) complex which was
 subsequently  analyzed by the well known West-Gaeke procedure (6).  By prior
 determination  of the calibration constant for a given membrane, the rate of
 permeation for SO-  was determined;  and,  thus, the time-weighted average con-
 centration for SO-  for a known exposure could be established.

     Permeation of  a given gas through a given membrane is a function of the
 concentration  of  that gas.   In general,  permeation of a gas can be compared
 with dissolution.   That  is,  a gas molecule coming in contact with a liquid
 dissolves  and  permeates  into that liquid without the benefit of stirring.
 Instead  of  liquids,  polymeric membranes  can also serve as a permeation medium.
Gas molecules  coming  into  contact with such membranes dissolve and pass into

the barrier until saturation is achieved.  If a receiving system  is  introduced
on the other side of the barrier, equilibrium is disturbed and  the permeating
molecules are removed, thus providing a means for quantitative  mass  transfer.
Because of the fixed relationship between a given gas and a given membrane,  the
passage of a gas through membranes can be described by the equation:
where  k = the  permeation constant
       C « concentration of the gas of interest, ppm
       t « time of exposure
       w - amount of gas absorbed, yg.

For use in personal monitoring, the above concepts can be applied, and knowing
the permeation  constant, or more  properly, the calibration constant for an
individual gas, sampling can be achieved and the concentrations of the gas
of interest  in  unknown atmospheres for a given exposure period can be calcu-

                                      C = wh

where  C = the  time-weighted average  for the gas, ppm.

     At this point  it  is proper to point out the difference between diffusion
and permeation.  Although  there are similarities, there are subtle differences
which may be significant in adjudging characteristics and, thus, applications.
Unlike permeation, which is essentially a solution process as discussed above,
diffusion represents  the passage  of molecules  through holes in a discrete
barrier where turbulence is minimized.  Beyond  the barrier, mass transfer
across a stagnant  space  to a collecting surface provides a means for quanti-
fying the transport  process.   The quantitative  relationship is given by Pick's
Law of Diffusion,

                                N - - DA (DC/DX)

where  N » rate of diffusive  transport, moles/ sec
       D » diffusivlty of  species, cm /sec
       A » area of diffusion  path, cm
       C = concentration of  species,  moles/cm
       X = path length,  cm.

      There are a number of general advantages to the permeation technique for
 sample collection.  A unique advantage that has been disclosed over the past
 8 years of laboratory and field experience is the almost indestructible na-
 ture of permeation-type devices.  Since there is nothing to wear out or other-
 wise degrade,  months or even years of service can be expected.  Only the sorp-
 tion medium needs to be replaced, and thus the cost for collecting any in-
 dividual sample amounts to only a few cents.  The long-term performance, to-
 gether with the low cost of the original devices themselves, make this the
 lowest cost approach for reliable monitoring of gaseous pollutants.  It is
 important to note also that the badge-type monitors are the ultimate in con-
 venience and that they present no more hazard than the typical identification
 badge worn by plant personnel.  This is in contrast to the impediment to mo-
 tion and the danger of entanglement presented by the conventional battery,
 motor, pump, and tube personal monitors.

      The versatility of the permeation approach for personal monitors pro-
 vides a variety of options for the solution of different monitoring problems.
 Solid adsorbers such as silica gel and activated charcoal are of course well
 suited for use in the monitors.  In addition, liquid systems are readily us-
 able, as, for example, utilization of tetrachloromercurate(Il) for S0« sam-
 pling and the use of solutions containing bromide and fluorescein for the
 simultaneous collection and determination of chlorine.

      Permeation-type monitors have been widely field tested over the past
 6  years,  and in all cases excellent results have been obtained.  Colleagues
 in the United  Kingdom, Europe, Canada, and various parts of the United States
 have cooperated in the field tests, and their results have generally confirmed
 the  results of the extensive testing that has gone into the original develop-
 ment of the individual monitor types.

      Field use of any type of monitor emphasizes problems associated with
 environmental  variations and stresses.  The first problem faced in the develop-
 ment of permeation-type devices was that of temperature effects*  The char-
 acter of  most  polymeric membranes is very susceptible to changes in tempera-
 ture,  but silicone polymers were found to exhibit minimiurn variation over
 normal temperature ranges, and for some gases, such as vinyl chloride and
 chlorine,  no errors were found over the range of 0° to. 40° C.  In general,
 therefore,  temperature effects were found to be of little significance when
 the  proper membrane was used.   In general,  the presence or absence of sun-
light  has  been without effect,  and humidity and relative air movement have
been  of no significance.   The  presence of a wide variety of copollutants

in industrial atmospheres has presented no problems with any of the monitors
developed so far.

     The following discussion summarizes the essential make-up of five dif-
ferent devices that have been developed for the monitoring of sulfur dioxide,
chlorine, alkyl lead, benzene, and vinyl chloride, respectively.  In each
case, the monitor itself is basically a hollow badge having a selected mem-
brane enclosure and containing a sorption medium chosen especially for the
collection and stabilization of the toxic gas of interest.

       Membrane—dimethyl silicone, single backed
       Sorber—sodium tetrachloromercurate(II), 1M
       Analytic finish—conventional West-Gaeke spectrophotometric
       Temperature effect	0.5 percent * C increase
       Humidity effect—none
       Relative air movement—no effect
       Sensitivity for 8 hours exposure—0.01 ppm
       Remarks—a catalytic stabilization can be used that permits sample
       collection over l-to-4-week periods.

     CHLORINE (7)
       Membrane—dimethyl silicone, single backed
       Sorber—buffered (pH 7) solution of fluorescein (0.005 to 0.0005 percent)
       and sodium bromide (0.31 percent)
       Analytic finish—spectrophotometric or colorimetric measure of the
       color produced by the Cl_-oxidation of bromide and resultant bromination
       of fluorescein to produce eosin
       Interferences—none of significance
       Temperature effect—none
       Humidity effect—none
       Relative air movement—no effect
       Sensitivity for 8 hours exposure:  0.013 ppm
       Remarks—sampling and simultaneous estimation of exposure results
       are possible by on-the-spot colorimetry.

     ALKYL LEAD (8)
       Membrane—dimethyl silicone, unbacked
       Sorber—silica gel with desorption by means of IC1

        Analytic finish—spectrophotometric measure of color produced with
        dlthlzone or an atonic absorption spectrophotometric determination
        Interference—none; errors due to lead Impurities In reagents minimized
        by use of silica gel rather than charcoal
        Temperature effect-- -0.66 percent * C for TEL and +0.81 percent
        per * C for TML, thus essentially compensating for the usual plant
        Humidity effect—none
        Relative air movement—no effect
        Sen8itivity~0.2 yg, range 5 to 300 yg Pb/m
        Remarks—analysis time only a fraction of that required in previous

        Membrane—dimethyl sllicone, single backed
        Sorber—activated charcoal with desorption using CS.
        Analytic finish—gas chromatography
        Temperature effect—none
        Humidity effect—none
        Relative air movement—no effect
        Sensitivity—0.02 ppm, range linear to at least 50 ppm
        Remarks—widely field tested with excellent results.

      BENZENE (10)
        Membrane—-sllicone polycarbonate copolymer
        Sorber—activated charcoal with desorption using CS-
        Analytic finish—gas chromatography
        Temperature effect— -1.2 percent/* C
        Humidity effect—none
        Relative air movement—no effect
        Sensitivity—0.02 ppm 8 hours, linear to at least 20 ppm
        Remarks—analysis dependent on the purity of the CS. used for

     Passive sampling  of toxic gases for monitoring personal exposures can
 be accomplished  reliably,  conveniently,  and at low cost by means of badge-type

                                    TABLE 1
                         Reaulta of Typical  Field Teata
                     Sulfur Dioxide (area monitoring),
Permeation Device
                              Vinyl Chloride,  ppm
Permeation Device
 0.12; 0.11
 0.05; 0.08
 0.92; 0.97
                         Pump-Charcoal Tube
                        Chlorine, ppm (preliminary data)
Permeation Device
0.52; 0.67; 0.56
1.10; 0.20; 1.15
0.44; 0.49; 0.52
2.70; 2.60
2.80; 2.10

*NI08H methyl orange procedure
 Sane aa above ualng saturated aolutiona

                   FIGURE  1.   The  MinlMonitor.

devices  utilizing  permeation  for  quantifying  sample  collection.   Samples  col-
lected over  periods  of  hours,  weeks,  or  in  some  cases  even months,  are  an-
alyzed by  conventional  means  to give  the time-weighted average  exposures.

     Permeation-type monitors  are indefinitely  reusable because  there are
no moving  parts to wear out,  nor  is  there anything  to  degrade with  time.
The  collected  samples are adsorbed or absorbed  by  suitable means, removed
after  the  sampling period, and analyzed.  All  that  is  required  for  repeated
use  of the monitor is to  replace  the  sorber after  each exposure;  thus,  these
devices  are  the ultimate  on the basis of convenience,  cost,  and  reliability.
The  convenience is obvious because the monitors  are  single units  having a
size and weight comparable to  that of an identification badge.  The cost
advantage  is based on an  initial  cost of less  than $80,  which can be amortized
over months  or years of service—the  only operational  cost being  a  few  cents
for  recharging with  the appropriate  sorber  plus  the  cost of  the  subsequent
analytical procedure.   The reliability is attested  to  by extensive  laboratory
experience verified  by  widespread, independent,  cooperative  field studies
(Table 1).
Permeation-type  personal monitors,  "The MiniMonitor"  (see Figure  1),  are
available  for a  number of  toxic gases with  additional  monitors  currently
under development.  REAL,  P.O. Box  3341,  Baton Rouge,  La.  70821.


 1.  McKelvey, J.W.,  Hollscher,  H.E.  Anal Chem 29:123, 1959.

 2.  Allshuller, A.P., Choen,  I.R.  Anal Chem 32:802, i960.

 3.  Saltzman, B.E.,  Gilbert,  N.  Am  Ind Hyg Assoc 20:379, 1959.

 4.  O'Keeffe, A.E.,  Ortman, G.C.   Anal Chem 38:760, 1966.

 5.  Reiszner, K.D.,  West,  P.W.   Environ Sci Technol 7:526, 1973.

 6.  West, P.W., Gaeke, G.C.   Anal  Chem 28:1816, 1956.

 7.  Hardy, J.K., Dasgupta, P.K., Reiszner, K.D., West, P.W.  Submitted to
     Anal Chem, 1978.

 8.  Dasgupta, P.K.,  Pitts, W.D., Reiszner, K.D., Wolcott, J.W., West, P.W.
     Submitted to Anal Chem,  1978.

 9.  West, P.W., Reiszner,  K.D.   Am Ind Hyg J 39:645, 1978.

10.  Reiszner, K.D.,  West,  P.W.   In preparation.


Philip W. West, Ph.D., and Kenneth  D. Reiszner, Ph.D.
Environmental Sciences Institute
Chemistry Department
Louisiana State University
Baton Rouge, Louisiana  70803
     SYKES:  Dr. West, would you comment on the calibration  procedures in
these permeation-type adsorbers—specifically,  on your  system  for calibrat-

     WEST:  We have standard gas atmospheres.  So, we generate our  standard
atmospheres by diffusion for the alkyl lead, and for the  rest  of them I think
we uniformly used permeation tube standards for generating our standard

     We have a dynamic system in which the standard gas passes through the
calibration device, which is something that we have published. If  you look

  back at  some of the publications you can see the details on that.   We just
  simply insert or expose our monitors or our membranes to the dynamic standard
  gas  atmosphere, and from that,  determine the calibration constant.

       MAGE:   Dr. West,  does any  Government agency support your research right

       WEST:   Yes, I must give credit to the National Institutes of Health.
  We have  had  some support from them, and we have had some support from industry,
  and  the  university also supported it.   Unfortunately, the support we have
  from N1H will be terminated in  August.  I am ready for any support  you could
  give me,  and I am open to any suggestions as to what you would like to have
  worked on.

       BERLAND1:  Dr.  West, there has been a number of these gas badges at the
  conference that you have talked about.  Are there any physical features of
  the  badge that you have developed that allows it to volumetrically  sample
  more air, or are all these badges essentially similar?  Is the membrane some-
  what unique  in that it allows for more precision?

       WEST:   Incidentally, to impress some people, we can provide them with
  gold plating,  so it will cost a little bit more if that is important.

       At  any  rate,  as far as the unique features of this particular  badge
  go,  yes,  we  have more  capacity.  If we are using adsorbents such as charcoal,
  we have  somewhat more  charcoal, and therefore greater capacity.  There is no
  breakthrough,  by the way, on these.

       I certainly am  not running down the other badges, but, the  thing is
  is that  these  badges cost more  than the 3M badge or the Gasbadge.   These will
  run,  say, anywhere  from $40 to  $80.  But they are usable indefinitely.  We
 have  had  one  or two  of these successfully performing,  I guess,  after more than
 4 years.  There is  nothing to wear out,  and since the sorber is  poured out
 for  the analysis  and then replaced,  that is the only cost after  the initial
 cost  of the  badge.   Then you pay a few pennies for whatever sorber  you have
 to use in refilling  it.

      PETERS:    Edward Peters,  York Research.   Perhaps I missed something
 earlier,  but I  notice  that  there is no effect from the movement  of  air.

      WEST:  That  Is  right.

      PETERS:   The way  I  see it,  the higher the movement of air,  the more one
 might ingest these gases,  and yet  your instrument shows a constant  level.
 Can you comment on that?

      WEST:  Well, what we  are doing  is that  we are collecting the air at that
 particular spot and we are  getting  a time-weighted average of that  particular
 exposure. Now, the  rate  at which  the  sample is collected here corresponds
 quite closely with the sample rajte  obtained,  say,  with a charcoal tube and
 pump.  So, any variation  that might be due  to  that would be,  I think,  char-
 acteristic of any sampling  system.

     SCHEIDE:  I think the point to make is that you want  those  badges  to
be diffusion or permeation limited so that if the person is sitting  still
or walking down the hall, it should sample at the same rate.  That,  in  effect,
is what the membrane does.

Personal Air Pollution Monitors:
New  Developments
     Robert W. Miller and Byron Denenberg
     MDA Scientific, Inc.
     Park Ridge, Illinois
     With the advent of increasing  awareness by the research community and
the public at large of the potential  short-term and long-term health hazards
associated with airborne toxic  contaminants, great emphasis has been placed
on the development of new instrumental monitoring techniques to measure these
hazards.  In many cases, laws have  been established to limit human exposures
to various contaminants, regarding  both occupational and ambient levels.

     Historically, most ambient monitoring has been area monitoring, done
with relatively sophisticated,  expensive, and sensitive analytical devices
that are hardly one step removed from the typical laboratory devices that
require great care and attention in calibration and operation to obtain
optimal performance.  The use of this kind of analytical equipment for pollu-
tion monitoring in ambient air  is dictated by the low concentration being

     In the occupational environment, quite different approaches have been
undertaken.  Typically, the toxic levels being measured are one to three
orders of magnitude higher than permissible ambient levels, and, therefore,
the choice of analytical technique, with adequate sensitivity, is greater.
As additional toxicological and clinical evidence becomes available, there is
a constant downward trend in permissible exposure levels of various contami-
nants.  Typical examples of this process are vinyl chloride, acrylonitrile,
and benzene.  There is an ever-increasing body of data suggesting strongly
that a very large percentage of the ambient and occupationally exposed popu-
lation exhibits a multitude of  subtle deleterious symptoms of exposure.

 Many instances are documented of the effects of very low levels of pollutants
 causing significant changes in behavior and hormone-enzyme balance that re-
 sult in numerous symptoms that are difficult to treat because the industrial
 physicians involved are not aware of the causative agent.

      The need to accurately define what an individual is actually exposed to,
 and to comply with OSHA requirements, has added Impetus to develop personal
 monitoring systems for toxic substances.  Oftentimes, standards are based
 on the limitation of sampling techniques, as well as toxicologlcal data*
 For example, most OSHA exposure limits are time-weighted averages, based on
 averaging air concentration over a prolonged sampling period—i.e., 15 minutes,
 2 hours, 4 hours, etc.  Obviously, liquid implnger and charcoal tube tech-
 niques with typical sampling times, as suggested above, will have poor time
 resolution of changes in concentrations.  The use of continuous, instantaneous
 readout monitors for substances (for which they are available), Indicates
 that it is not unusual to see marked concentration flux over periods of 15
 minutes or less in the occupational environment, as well as in ambient en-

      Since a short-term peak concentration may be as Important as total dose
 in its effect on humans,  it becomes Important to define these exposures,
 where possible,  and to provide immediate indication of concentration so that
 protective action can be taken to avoid excessive exposures.  A number of
 passive monitoring techniques have recently been developed, including small,
 lightweight adsorption badges.  Although these devices can be useful as diag-
 nostic or screening agents, they cannot provide an exposure profile, and,
 most importantly, they only provide historical information—i.e., after it is
 too late to do anything about the exposure.  To detect peak concentration
 requires,  first,  a sensing device small and lightweight enough to be readily
 worn by an individual in or near his or her breathing zone, with the capa-
 bility of  providing maximum time resolution and a continuous concentration
 output.  A number of analytical techniques can meet these criteria, but not
 without limitations.   Electrochemical cells and semiconductors are attractive
 techniques for personnel  devices but oftentimes are limited by sensitivity
 and/or specificity problems.   In these cases,  the application often dictates
 success  or failure of measurements relative Co possible interferences.

     One technique that offers great promise for further development and ap-
 plication  is  the  detection of  toxic gases using chemically impregnated paper
 tapes.  In  this case,  contaminants which are collected on the tape react with
 the chemicals  on  the  tape  to  produce a stain proportional to the mass con-
centrations of the  contaminant.   The strength  of this technique is that not


only can it provide time resolution (as short as 30 seconds),  but  also that
the chemistry can be manipulated to obtain remarkable  sensitivity  (low ppb)
and a high degree of specificity.

     A personal sampling system using this detection principle (the MCM)
has been commercially available for several years, and  is capable  of  provid-
ing both a time versus concentration curve and total dose information for
accurately defining personnel exposures.  A modified MCM system, referred
to as the Chromatox System, is currently in the final  stages of development;
it will expand even further the time resolution capabilities,  as well as  the
overall versatility In utilizing new tape detection systems as  they are de-
veloped from an ongoing research program.  This system  is analogous to hi-fi
tape recorders, with the difference being that a chemical signal instead  of
an audio signal is recorded and played back.

     To immediately increase the capabilities of the MCM and other personal
detection devices that have a continuous output, a unique mircoprocessor  pro-
file alarm dosimeter (Chronotoximeter) has been developed.  The Chronotoxi-
meter is designed to accept an output signal from any detection system.   The
microprocessor takes a reading every second (this feature is adaptable to
the sensor) and stores that datum  in a temporary memory.  At the end  of 90
seconds, the data are processed and a 90-second average value  is stored in
a permanent memory.  At the end of the exposure period, the Chronotoximeter
Is then interrogated to provide a  time versus concentration curve and/or  a
digital printout of the stored data.  The microprocessor can be programed
to accept the input from other detection systems, in addition  to the  MCM.
In fact, it has been designed with this intent so that  a user  would have  maxi-
mum choice of detectors for data gathering.  As detectors are  changed, a
"FROM" (programmable read  only memory), calibrated to  the detector calibration
curve, is inserted into the Chronotoximeter, and the system is then totally

     For ambient personal  monitoring,  the  tape detection technique offers
the greatest promise because  of  the ppb sensitivity and because of the in-
finite possibilities of chemical reactions  to choose from to obtain appro-
priate specificity.  An ongoing  research program is continuing to  develop new
detection systems.  The capability to  begin meaningful surveys of  general
population exposures to compounds  such as  CO, N02» and SO,, with the Chronotox
System should provide a fruitful ground for research in the next several  years.

     The Chronotoximeter will also incorporate digital real time readout
with high level and total  dose alarm  capability  In one embodiment.  This


  further  enhances  its usefulness, from a protection standpoint, to the in-
  dividual wearing  the device*  Later phases of the development program will
  provide  telemetry output of the Chronotoximeter so the data produced can be
  telemetered back  to a central data center and appropriate warnings given to
  the wearer, when  warranted.  Such telemetry systems are already operational
  for measuring individual motion and heartbeat in hazardous occupational

 Robert W. Miller and Byron Denenberg
 MDA Scientific, Inc.
 808 Busse Highway
 Park Ridge, Illinois  60068
      PETROVICK:  Would you tell us what you are using and what the memory
 capacity of it is?

      MILLER:  Both answers at this point are proprietary.

      PETROVICK:  Another comment:  I am pleased to see that someone is moving
 in that direction, because of the extrapolations from your hydrobodies to a
 true, physiological monitoring and and response type capacity.  I commend
 you for that;  I think that is the way to go*

      MILLER:  Just to comment in reference to what you say:  I really didn't
 come here to talk about the physiological approach, but associated with this
 system, there  is also a radio-operative system, and that radio-operative
 physiological  system is one that an individual can wear, which will provide
 a heartbeat, that is tied into this complete system where you can radio the
 signal  back to a central location.

      So,  if you have an individual  out in an industrial environment—which
 would be  a primary application here—where they may encounter an extremely
 high concentration of something that they cannot physically remove themselves
 from immediately,  you will get an alarm signal as well as a direct heartbeat
 and  respiration signal,  sent  back to a central control panel where individuals
 are  continually monitoring, and then you can immediately take steps to send
 someone up.  That  system also now exists and is available,  in conjunction
 with this.

     PETROVICK:  I would like to come back to that.  That  is very  interesting,
because if you pick up this week's US News and World Report, there  is  an  ar-
ticle in there on cardiac heart attacks.  It alludes to a 23 percent drop in
fatalities from coronary attacks in the last 5 years with  the  implementation
of some kinds of technology, particularly the emergency health care of the
kind where help can get to the person within an hour.  It  is that crucial
hour-long-period that I think you have*  And if someone may have problems,
even as simple as a bee sting, there are enzymes within the heart muscle
that reside there that release themselves and partly are responsible for  a
coronary event.  That is not my opinion—it is in the US News and World Report.
I urge you to read it.

     LIN:  Will particulates be a strong interfering agent in  the tape sampler?

     MILLER:  Generally speaking, we have found that not to be the case—
actually, even in foundry operations where you have high ambient dust  levels.
And the reason is that the tape is continuously moving.  It is not stopped
on one spot.  So primarily the stain that develops there is a function of
the organics or whatever compound you are measuring and not of the particu-

     Also, most of the heavies drop out because of the relatively low  flow
rates that we are using, so you don't pick up a lot of the heavy particulates.
Most of the submicron particles, because of the tape movement, don't create
enough darkening of the tape system to make a significant difference.

     Now, realistically speaking, that is not universally  true, because I
have found one or two situations where in fact the ambient dust level  was
so high that it did cause a significant background reading.  In which  case
you had to correct for it by compensating your baseline.   But  that also,
you know, can be done.

     I might also add that in  those situations the ambient dust level  was  so
far above anything remotely approaching permissible exposures  that it  was  a
necessity, really, to correct  the ambient dust levels before they worried
about some of the other things they were sampling.

     SHAW:  What is the operating principle of this cell?

     MILLER:  I think it is best construed as an amperometric  cell.

     WEST:  I believe you mentioned you have gotten away from  the  lead ace-
tate.  What do you use?

     MILLER:  I will  tell you  that we  do use lead acetate.  However, a little
modern chemistry and  engineering has,  let's say,  taken  the basic lead  acetate
and made  it workable,  in effect, in a  real-world  environment.

     There are units  available that use  the old  lead acetate  that  require
all kinds of unbelievably complex humidification  schemes.  Our system  does
not require that.  Again, it  is  proprietary, but  suffice it  to say that we
build the water into  the  tape.


      WEST:  We'll  take  your  word  for that,  because that is  important.

      MILLER:  That  is critical.   The sensitivity is tremendous,  but  it  has
 always been the  humidity  effects  that have  really been the  problem one  has  to
 cope with.

A New Sampling Tool for Monitoring
Exposures to Toxic  Gases and Vapors
     John C. Gillespie and Leah B. Daniel
     Abcor Development Corporation
     Wilmington, Massachusetts

     Air sampling devices are needed to provide a variety of exposure infor-
mation, including:

     1)  Occupational exposure information concerning employees  in a plant
and environmental exposure information of the population in an area sur-
rounding a given plant.

     2)  Personal and fixed-station (area) exposure information.

     3)  Information concerning immediate (acute) versus long-term (chronic)
health hazards.

     4)  Peak versus time-weighted average exposure data (or episode versus
average concentrations)•

     5)  Continuous versus integrated sample measurements.

     These various applications require a number  of different sampling de-
vices, each having certain advantages and limitations.  A comprehensive sum-
mary of the uses of air quality monitors in measuring the exposure of the gen-
eral public to air pollution is explored by Lance Wallace in Reference 1.

     The purpose of this paper is to discuss a passive  sampling  device, the

 GASBADGE   dosimeter, and  Its applications to date (primarily In occupational
 health monitoring), and  to suggest which of the exposure Information needs
 presented above It can best  satisfy*

      The GASBADGE organic vapor dosimeter is a 1 and 1/2-ounce sampling device
 which can easily be worn by an Individual to collect time-weighted average
 exposures to organic vapors.  Since it requires no air sampling pump, it is
 often referred to as a "passive" sampler.  The GASBADGE dosimeter Is analogous
 to and as convenient to wear as the radiation film badge.

      The dosimeter (Figure 1) contains an activated carbon collection element
 which is removed after exposure for subsequent analysis.  To determine the
 time-weighted average exposure, the element is removed from the dosimeter and
 placed in a sealed vial.  The organic material collected on the element is
 then desorbed using carbon disulfide or another appropriate solvent and an-
 alyzed by gas chromatography.  The dosimeter is reusable and can be easily
 reloaded with a new collection element.  Additional elements (other than char-
 coal)  can be used to collect other gases and vapors.

      The components of the dosimeter (Figure 2) include (from bottom to top):
      1)   A sliding cover to prevent contamination, or exposure, for short
 periods  of time.
      2)   A dosimeter front with a carefully defined opening to allow diffusion
 of gas or vapors into the dosimeter.
      3)   A draft shield made from a nonreactive porous material.
      4)   An open grid to define the diffusion geometry and minimize internal
      5)   A replaceable collection element.
      6)   A caseback with a spring clip for attachment to the individual.

      Generally,  the first reaction to the GASBADGE dosimeter centers around
*GASBADGE  is  a  trademark and registered servicemark of Abcor Development
Corporation,  Wilmington, Mass.

FIGURE 1.   The GASBADGE Organic

Vapor  Dosimeter.
FIGURE 2  (right).   GASBADGE  compo-

nents  and  specifications.
                                                            Sliding protective cover prevents contam-
                                                            ination or exposure for short periods of
                                                            Badge front with opening allows diffusion of
                                                            gas or vapors into dosimeter
                                                            Draft shield made from nonreactive porous
                                                            Open grid defines diffusion geometry,
                                                            minimizes internal mixing
                                                            Replaceable collection element
                                                            Dosimeter back with spring clip

                                                         Collection Element:
                                                         Analysis Method:

                                                         Sampling Time:
                                                         Sampling Range:

                                                         Shelf Life:
                                                         Air Velocity:
                  2x2 9/16 x 5/8 in.
                  (5.1 x6.5x 1.6cm)
                  1.5oz.  (43g)
                  Activated carbon
                  NIOSH P&CAM 127
                  8 hr nominal
                  0.2-160 ppm/8-hr TWA
                  + 25%
                  2 years
                  15fpm (0.08 m/s)
                  minimum.   Normal
                  industrial plant
                  air velocity exceeds
                  20fpm (0.1  m/s)
                  U.S. Patent 3,985,017

  the absence of an air sampling pump.  The reaction is either positive or neg-
  ative.   The positive reaction is:   "Now I won't have to spend all my time
  calibrating my pumps, charging batteries, or Just trying to keep the pumps
  running."   The negative side of the "no pump" response is this:   "If it doesn't
  have a  pump to control the air flow into the device, how will I  know the volume
  of  air  sampled?"

       What  these two responses point out is that:
       1)  There is a fundamental difference between collecting samples using
  sampling tubes and an air sampling pump and using the GASBADGE dosimeter be-
  cause the  GASBADGE dosimeter does  not depend on the volume concept associated
  with air sampling pumps;  and
       2)  Since the GASBADGE dosimeter has no rotometer, makes no noise, and
  has no  moving  parts, it is impossible to physically see the sample collection
  in  process.

       The GASBADGE dosimeter is based on controlling the random but predictable
  molecular  movement of gas molecules rather than controlling air  flow.  What
  Abcor has  done with the organic vapor dosimeter is to utilize physical chem-
  istry principles of vapor diffusion to control the amount of material col-
  lected  by  the  collection element.

       Figure  3  shows a comparison of the two equations used to calculate the
  ppm concentration of an organic vapor exposure using either a diffusion dosi-
 meter or a pump sampler.   Both equations are time dependent In that more ma-
  terial will  be collected  for longer exposures at  a given concentration (up  to
  the  capacity of the device).

       The first two  parts  of each equation are basically the same.  The dif-
 ference  between the two equations  is in Section III, or what is  labeled the
 "Transport Force."   In Section III,  the flow-rate or volume component of the
 pump  sampler equation is  replaced  by three physical constants which control
 the mass of  material  collected by  a diffusion-based sampler, namely:
       1)   The length of the diffusion path;
      2)   The cross-sectional  area  onto which the  gas or vapor is diffusing;
      3)   The diffusion rate  of the gas or vapor in air, determined by ex-
 perimental methods.

      To  accurately  determine  measurements of employee exposures  using a char-
 coal tube sample collected with an air sampling pump,  the flow rate must be
 precisely controlled at a  known  constant  level throughout the sampling period.

Measurement of
Collected Sample
Diffusion Dosimeters
desorptlon efficiency
Pumped Sampler
desorptlon efficiency
Standard Conditions and
Molar Conversions
1 *«**» , , T°K 760mmHg
•* 	 __, v *JV Afm\ f**»/mnlM v " v *.
' mw X 22''100cc/mo"' x 273o K x PmmHg *
— r 224S/mol0 * ^ K « 7""'inH8
mw x22'4B/m°'«' x2?30|C x pmmHfl >
Transport Force
X(cm) .
CD/cm2\ Afcm2)t««o)
Flow Rate(fi/min) x T(mln)
X -  Diffusion Length (cm)
D -  Diffusion Coefficient
A "  Cross-sectional Area (cm2)
FIGURE 3.  Diffusion and air sampling pump  equations*
                                    8     12    16
                                      HiLLICRAMS ADSORBED
FIGURE 4.  Percent recovery versus milligrams adsorbed.  (Source:   Mazur, J.F.
et  al.  A New Personal  Sampler for Organic Vapors.   Paper presented at the AIHA
conference, May 1977.)

  If the flow rate changes during sampling,  or it is not verified before and
  after sampling,  the calculated results will be in error.

       Passive samplers, operating on the diffusion principles,  are dependent
  not on flow but  on careful definition of the physical dimension of the sam-
  pling device and on making certain that the only force causing the collector  to
  operate is gaseous diffusion.   The GASBADGE dosimeter does this by creating
  a dead air space that is relatively unaffected by changes in air velocity
  between the draft shield and the collection element.   The GASBADGE dosimeter
  works, in effect, because the  draft shield prevents air flow into the  device,
  thereby ensuring that the amount of gas or vapor collected is  determined  only
  on the diffusion rate (molecular movement) of gas or  vapor in  the air  being

      Two  types  of performance testing  are:
      1)   Testing under carefully controlled  laboratory conditions  with a
 known concentration and environment.
      2)   Testing primarily in industrial  (occupational)  environments  under
 field conditions which present more variability and  essentially unknown

      The  advantage  of  performance testing under controlled  laboratory con-
 ditions is that  one can measure the accuracy and precision  of  the  sampling
 method when  compared to a  known concentration.   The  disadvantage is that
 testing under laboratory conditions does not necessarily duplicate the real

      While field  testing is the real world,  determination of accuracy and
 precision is much more  difficult  because the actual  concentration  is  generally
 unknown and varying and can be  measured only by the  use  of  other secondary
 sampling devices such as charcoal  tubes which also have  variations in their
 accuracy.   In other words, when comparing two sampling methods  under  field
 conditions, one may be  comparing  one sampling method with a potential error
 of plus  or minus 25 percent with  another device which has a similar error
 range.   This means one  could expect variations  between the  results of two
 such sampling methods of as much as plus or  minus 50 percent.

      For these reasons, Abcor has relied on  controlled exposure  testing  by
 Abcor and  other  groups to determine the accuracy of  the GASBADGE dosimeter

for various organic materials. Correlations between side-by-side  field  samples
collected using organic vapor dosimeters and other suitable  sampling methods
verify that there is general agreement between two (or more) sampling methods
under varying field conditions*

     The U.S. Army Environmental Hygiene Agency (USAEHA) conducted an exten-
sive evaluation of this monitoring device (2).  Table 1 shows the percent
recovery and relative  concentration obtained by the USAEHA for seven different
organic compounds.  The compounds tested represent seven different classes
of organlcs; namely, an aromatic, an alkane, an alcohol, a ketone, an ester,
a halogenated alkane,  and  a halogenated alkene.  The compounds were Individu-
ally sampled to determine  percent recoveries for each in the 1/4, 1, and 2x
the TLV concentration  range, and to determine if the dosimeter compared favor-
ably with the recoveries previously obtained by the USAEHA for charcoal tubes.
Since the dosimeter data were so reproducible, the U.S. Army Environmental
Hygiene Agency felt that the results could easily be corrected for the  slight-
ly lower percent  recovery  levels for absorption/desorption efficiencies,
just as charcoal  tube  data are  corrected.

     Figure 4, again from  the USAEHA study, shows a plot of  percent recovery
versus total mg adsorbed for iso-octane, trichloroethylene, methyl chloroform,
n-butanol, and MIBK.   The  results of this experiment indicated that the per-
cent recovery was relatively independent of ambient concentration when  less
than 10 mg of material is  loaded on  the collection element.  For  this reason,
Abcor recommends  that  the  maximum loading of GASBADGE collection  elements not
exceed 10 mg  total  solvent.

     Many customers  have asked  if  the  GASBADGE dosimeter measures mixtures as
well as single-component vapors.  The  data  in Table 2,  again from the USAEHA
study, show  that  diffusion operates  effectively  for mixtures of organic com-
pounds as well as when the compounds  are  each monitored separately.  This
holds true as  long  as  the  mixture  does not  exceed  the maximum recommended
loading of 10 mg  total solvent.

     Tables  3  and 4  show the accuracy and  precision of  the GAS BADGE dosimeter
for benzene  and  acrylonitrile monitoring.   Table 3 was  compiled  from  performance
testing conducted by Abcor for benzene exposures varying from 0.28  ppm  to
4.69 ppm.  It shows the mean dosimeter concentrations,  systematic error,  and
relative  standard deviation for the GASBADGE organic vapor dosimeter  at var-
ious levels.   The pooled values of the individual test  levels are a measure
of  the overall accuracy of the device for all exposure runs combined and  in-

                                      TABLE  1

                   Percent Recovery Versus Relative Concentration
n- Butyl
0.076(a) »
11.5, .
4.7, .
15.8 c
(b) 76
94 + 4
89 + 4
104 + 3
101 + 3
84 + 3
84 + 5
79 + 4
80 + 6
80 + 2
93 + 3
88 + 4
84 + 4
85 + 2
77 + 3
90 + 2
90 + 3
89 + 3
88 + 5+"
      Too low for accurate quantitation.
 '    Percent recovery improved by Abcor to 94 percent by modifying the an-
 alytical technique used in the USAEHA study.
 ( c j
 v '  Overload conditions.
 +  From the literature.
 Source:  Mazur, J.F. et al.  A New Personal Sampler for Organic Vapors.
 Paper presented at the AIHA conference, May 1977.

                                    TABLE 2

                Percent Recoveries for a Four-Component Mixture
Butyl Acetate
 Percent recovery In the absence of other components
Source:  Mazur, J.F. et al.  A New Personal Sampler for Organic Vapors.
Paper presented at the AIHA conference, May 1977.
dlcate that the GASBADGE dosimeter Is within plus or minus 20 percent at a 95
percent confidence level.

     Similar accuracy  for  the GASBADGE dosimeter is demonstrated in Table 4,
which summarizes data  collected  by Dow Badlsche (3) for monitoring exposures
to known concentrations of acrylonitrile.  After conducting laboratory meas-
urements to determine  the  accuracy of the GASBADGE dosimeter, Dow Badische
performed  field tests  on a side-by-side basis with charcoal tube samples.
The data from  the  field testing  demonstrated that the GASBADGE organic vapor
dosimeter  is comparable to the charcoal tube technique.

     The major uses of the GASBADGE organic  vapor  dosimeter  have centered
around potential  occupational exposures,  including:

     1)  Monitoring employee exposures to benzene  in petroleum refineries,
benzene manufacturing facilities,  pharmaceutical companies,  laboratories, and
other similar industries.

     2)  Monitoring exposures of field personnel to benzene  in remote  pe-
troleum pipeline  and pumping station locations where the dosimeters are dis-


                               TABLE  3
GASBADGE™  Organic  Vapor  Dosimeter Performance Data

                        Benzene  Monitoring
                     (8-Hour  TWA Exposures)

0.28 ppm
0.5 ppm
0.5 ppm
0.56 ppm
0.56 ppm
0.97 ppm
1 .90 ppm
4.69 ppm

Number of

Mean Dosimeter
0.25 ppm
0.56 ppm
0.49 ppm
0.64 ppm
0.60 ppm
0.98 ppm
2.24 ppm
5.25 ppm

0.01 ppm
0.01 ppm
0.01 ppm
0.03 ppm
0.04 ppm
0.04 ppm
0.10 ppm
0.30 ppm

- 2%
+ 7%
+ 1.0%
+ 8.5%*
Error Range to
a Confidence Level
of 95% Based on
Systematic Error
-19.0% to- 3.0%
+ 8.6% to +15.4%
- 6.0% to + 2.0%
+ 4.6% to +23.4%
- 6.4% to +20.4%
- 7.2% to + 9.2%
+ 8.9% to +26.9%
+ 0.5% to +23.3%
- 1.1% to +18.1%
                                Systematic Error
                                                  Mean Dosimeter Concentration - Exposure Concentration
                                Relative Standard Deviation
              Exposure Concentration

              Standard Deviation

          Mean Dosimeter Concentration
                                 'Calculated using a weighted systematic error.
                                *'Calculated using the technique specified by NIOSH for evaluating the accuracy of methods for the standards completion pro-
                                  gram. Summary of Statistical Terms and Formulas, S112-D1 (Attachment D). The mathematical equation is expressed as:
                                 Z   fi (RSDj)2
                                                        f j

                 Pooled Relative Standard Deviations (Coefficient of Variation)
                 degrees of freedom, equal to number of observations minus one. at the Ith level
                 Relative Standard Deviation of the observations at the Ith level
                 Z   fj

                                                                                     TABLE 4
                                                    GASBADGE*  Organic Vapor  Dosimeter Performance Data
                                                                        Acrylonitrlle Monitoring
Exposure Number of Mean Dosimeter Standard
Concentration Dosimeters Concentration Deviation
0.74 ppm 10 0.75 ppm 0.03 ppm
8.9 ppm 16 8.67 ppm 0.83 ppm
19 ppm 7 19. 14 ppm 2.41 ppm
Pooled Values
This data demonstrates that the GASBADGE organic vapor dosimeter meets
Standard for Acrylonitrile issued by OSHA on January 17, 1978.
4 Error
-1.0%* 9.1%**
the accuracy requirements of the
Error Range to
a Confidence Level
of 95% Based on
Systematic Error
±2 RSD
- 6.6% to + 9.4%
-21. 8% to +16.6%
-25.9% to +24.5%
-19.2% to +17.2%
Temporary Emergency
                              The above data is a summary of results of a study conducted by the Dow Badische Company, exposing the dosimeters in a dynamic test
                              chamber for 6-8 hours. As a further check, eighteen GASBADGE dosimeters were worn, side-by-side with charcoal tube samples under
                              field conditions. The data from both the dynamic testing and field testing show the GASBAOGE technique is accurate, reproducible
                              and comparable to the charcoal tube technique. A copy of the complete test data is available from Abcor Development Corporation or
                              W. O. Calhoun, Quality Control Laboratory, Dow Badische Company, Post Office Drawer D, Williamsburg, Virginia 23185.
                              Systematic Error
                                                 Mean Dosimeter Concentration - Exposure Concentration
                              Relative Standard Deviation
     Exposure Concentration

     Standard Deviation
Mean Dosimeter Concentration
                               •Calculated using a weighted systematic error.
                              • 'Calculated using the technique specified by NIOSH for evaluating the accuracy of methods for the standards completion pro-
                                gram. Summary of Statistical Terms and Formulas, SI 12-D1 (Attachment D). The mathematical equation is expressed as:
                                                                          RSD  =  /.£  fj (RSDj)2

       Pooled Relative Standard Deviations (Coefficient of Variation)
       degrees of freedom, equal to number of observations minus one, at the itn level
       Relative Standard Deviation of the observations at the itn level

  tributed to the field,  reloaded in the field,  and samples sent in vials
  to central laboratories for analysis.

       3)   Monitoring employee exposures to various chlorinated solvents used
  in metal degreasing and other cleaning operations, such as trichloroethylene.

       4)   Monitoring employee exposures to acrylonitrile from the manufactur-
  ing of the basic material through its  use in ABS injection molding operations.

       In  addition to its use in a traditional industrial hygiene sampling
  program  of a large plant under the supervision of an industrial hygienist,
  the GASBADGE dosimeter  lends itself to "self-administered" sampling of per-
  sonnel at remote locations or in plants where  there is no industrial hygien-
  ist.   The GASBADGE dosimeter makes sampling easier and even possible in areas
  where air sampling pumps cannot practically be sent or used due to the lack
  of trained personnel.

       Additional areas in which the GASBADGE dosimeter has or will be used

       1)   Evaluations of hospital personnel having exposure to various anes-
  thetic gases and other  gases and vapors.

       2)   A study of vocational education students to determine qualitative
  exposure  information for a variety of  organic  vapors found in painting and
  decorating,  plastic technology, and auto body  shops.

       3)   An  evaluation  of the exposures of Individuals to various dry clean-
  Ing solvents

       4)   A study of the exposure of college art students to various solvent
 vapors present  in silk  screening operations.

      The GASBADGE dosimeter provides a  sampling device  which can be used to ob-
 tain personal exposure information related  to  health  effect studies for en-
 vironmental as well as occupational exposures*

      The primary purpose of this sampling device is for personal monitoring,
 although it could be used to obtain area monitoring information.  Since it


Is not direct reading and has no alarm to warn of overexposure, it should
be used for monitoring chronic rather than acute health hazards.  While it
is effective for short-term monitoring (1 hour or less),  it should normally
be used for obtaining time-weighted exposure information over several hours.
It should be used for obtaining integrated sample measurements rather than
as a continuous monitor to measure changes in concentrations.

     Since the GASBADGE dosimeter does not require an expensive air sampling
pump nor the calibration and training requirements associated with pumps,
it can be distributed and collected in a number of ways (mail, a central
distribution point, or by personal visit).  In other words, it is both eco-
nomical and relatively easy to use in a large-scale population study.

     Performance data indicate the success of passive samplers for monitoring
occupational exposures to a variety of gases and vapors.   It would be ap-
propriate and necessary to conduct similar performance tests to determine
the effectiveness of the GASBADGE dosimeter for monitoring exposures over
days or weeks (rather than 8 hours) at levels in the parts-per-billion rather
than the parts-per-million sampling range.  We believe such testing will
demonstrate that this sampling approach has the ability to provide valuable
exposure Information for a variety of population groups in the home and out-
side environment as well as in the workplace.

 1.  Wallace, L.  Personal Air Quality Monitors:  Past Uses and Present
     Prospects*  Paper presented  at  the 4th Joint Conference on Sensing of
     Environmental  Pollutants, American Chemical Society, 1978.

 2.  Mazur, J.F. et al.   A New Personal Sampler for Organic Vapors.  Paper
     presented  at  the American Industrial Hygiene Conference, New Orleans, La.,
     May  1977.

 3.  Calhoun, W.O., Silverstein,  L.G.  Personal Monitoring for Acrylonitrlle
     Using  the  Abcor GASBADGE".   Dow Badische Company, Williamsburg, Va.,
     September  1977.


John C. Glllespie  and Leah  B. Daniel
Abcor  Development  Corporation
Wilmington, Massachusetts  01887

Panel Discussion
     Panel members for the following discussion were:   Dr. David Mage, En-
vironmental Monitoring and Support Laboratory, U.S.  Environmental Protection
Agency (EPA); Dr. Lance Wallace, Office of Monitoring  and Technical Support,
EPA; Mr. Otto White, Brookhaven National Laboratories;  Mr. Mathew Petrovick,
Health Effects Research Laboratory, EPA; and Dr. Manny Shaw,  InterScan Cor-
     MAGE:  Thank you very much for returning this afternoon.  We are going
to try to summarize for the benefit of the record, as well as  for the dis-
cussants, our impressions and conclusions of what we have learned,  and  the
directions that we feel we should be going in.

     You will have an opportunity to make remarks for the record after  the
panel members have given  their summary.

     I am going to start  off by recognizing the panel members—Dr.  Lance  Wallace,
Mr. Otto White, Mr. Matt  Petrovick, Dr. Manny Shaw, and finally, myself.

     WALLACE:  I don't have any prepared remarks.  I am still  trying to come
to grips with everything  that I have learned  the  last few days,  which has
been a large amount of information  to  try to  take in.

     Frankly, my horizons have been broadened.  I came into this conference
thinking of personal monitors basically on  the physical-chemical side of it


 and of making  that  type of measurement.  I now see that the advances in the
 methods of measuring physiological responses are great enough that they have
 opened up to us a possibility of dose response measurements that I hadn* t
 really envisioned before.

      The microcomputer aspects, that would enable us to take these continuous
 measurements of both dose and response and provide complete 8-hour or 24-
 hour readout,  appear to be extremely promising for the future.

      So, these three aspects—the physical-chemical side, the physiological
 side, and the microcomputer side—appear to be very promising.

      Among the monitors that are available now, 1 see several obvious repre-
 sentatives:   one is the electrochemical family for CO for which we already
 have the ESI instrument, the GE Instrument, and we have a promise from our
 industry representative of an InterScan instrument in the next year or so.

      For NO., the Palmes effort has been field tested and is getting indoor
 levels in the 50 parts-per-million range near gas stoves.

      In another area, the organics area, it is clear that we have several
 specific monitors.  Dr. West has benzene, toluene, alkyl lead, vinyl chloride,
 and halogenated hydrocarbons.  These monitors have been used in both the
 workplace and home environments.

      Both Minnesota Mining and Manufacturing (3M) and Abcor have special
 activated charcoal elements that can adsorb any organic materials that can't
 be absorbed  on activated carbon.  Those dosimeters are available right now
 so that  we can monitor out in the field.  That is what they have been doing—
 monitoring a  wide  range of organics.

     Each badge is not  specific  for a particular organic.   You can use one
 badge  and monitor  a wide variety of them.  You just break it down in the analy-
 sis.   So,  those are  available now for workplace monitoring.

     These badges  have  not been  tested  at the parts-per-million levels found
 in ambient air,  since  there  has  been  no need  to do so up to this time.

     I have the  feeling  at  the moment that  we do not  have available personal
monitors for ozone,  although  we  did see  the Mast instrument, which is a port-
able one.

     In regard to fine particulates, the Harvard group does  have  a  set  of
instruments that they have adapted themselves.  I am not certain  whether
we can call these instruments working personal monitors or not, because they
are rather bulky.  In their study that they described previously, one third
of the persons in the first sample group who tried it gave it up  on the first
day.  Apparently, more work is needed here.

     In regard to S02> it is no problem at all.  Dr. Philip  West  has been
doing that for some  time, not really as a personal monitor,  but for workplace
and ambient area monitoring in general.  Personal monitoring can  be done with
the permeation approach, and it can be done over long periods of  time.
Stability is ensured by converting the SO,, to sulfuric acid, and  the analysis
proceeds from there.  This provides a completely stable system.   These  badges
can be used for levels of 0.01 ppm and below.

     The one remaining thing that I would like to mention is the  prospects for
the future in the development of personal monitors.  It appears from what
we have heard in the last 2 or 3 days that the most emphasis and  the largest
payoff in the future may be the area of fine or respirable particulates.
We have a great interest in that within EPA.  A particulate  dosimeter pre-
sents a possibility  of collecting the material and then subjecting  it to analy-
sis for a spectrum of individual pollutants.  And the other  main  aspect would
be the organic monitors  that we heard about  from Dr. Joseph  Brooks,
Leah Daniel, Dr. Phil West, and other people here.  We may be able  to do
classes of organics. We may be able to do individual organics of interest.
Monsanto staff are going  to check 20 carcinogens.  They have a 3-year program
of which they have completed the first year.  This may be an area in which we
can expect great advances  in the future.

     MAGE:  Thank you very much, Dr. Wallace.

     WHITE:  Certainly,  this week  has  been enlightening  to me, and  hopefully
the information from this  meeting  will  help  formalize  the  final  report  that
my committee is working  on.  Certainly we are  aware  of pollution conditions
that exist, not only in  the occupational  environment,  but  also in the general
environment, that resulted in  the  conditions of black lung,  the  Donora, Pa.,
episode, and the London  air  pollution situation.

     One does  not have  to ponder long  in thinking in terms of  why we need  to
monitor from  the  epidemiclogical standpoint or the compliance  standpoint.
We need to determine whether or not the workplace or the environment is in-
deed safe.  These represent  challenging areas for not only  the industrial


 hygienlst, but  also  for  the  environmentalist—that  is,  to develop the instru-
 mentation  to  perform that monitoring.

      As I  see it, we  are still on  the  threshold of  trying to develop instru-
 mentation  for some of  the more common  pollutants such as CO and oxides of
 nitrogen.  We have yet to begin  tackling what may Indeed be the more difficult
 components of the environment.   We can look at the  incidence of cancer per
 county as  published  by HEW.

      It has been reported in the literature that a  high percentage of cancers
 are environmentally  or anthropogenically induced.   This represents a more
 challenging area of monitoring than our traditional oxidants, NO's, CD's,
 total particulates,  and hydrocarbons.

      So, 1 think yes,  there is a need  for defining  where the instrumentation
 development should come from and what  should be analyzed.  1 think there has
 been a lot of discussion in terms of the physical-chemical aspects of moni-
 toring.  There has been some discussion about monitoring the physiological
 stress on the individual.

      1 personally am concerned about long-term effects, as to whether or not
 that kind of data per individual would allow us to  explain some of the long-
 term effects.   Maybe it is going to be necessary to perform these exposure
 monitoring activities for a long period of time.   I guess 1 am Just not in-
 formed as to what health effects you are going to be looking for over a short
 term for some  of the ambient ranges of pollutants.

      But 1  also  think we should  probably search the horizon for the types of
 tools  that  we  are looking for.   Almost no comment has been made on some of
 the  biological systems, bioassay systems.  I recall about 12 years ago when
 1 was  here  at  the University of North Carolina,  and we were looking for bio-
 Indicators  for oxidant and  ozone, using white pine  seedlings.

     Although  the Ames test  was not intended to  be a bioindicator, I certainly
 think  it  represents  an area  where research and development may be warranted
 in terms of defining  whether or not that tool can be used as an indicator.
 That may be of particular interest  for some of these organics where there
 is a potential of there being a large host of compounds.  Some decision is
going to have  to  be made  as  to  which  compounds represent types of interest.

     Maybe what  is significant  is a measurement of biological activity of the
atmosphere  as opposed to  measurement  of benzopyrenes and benzene soluble

     There are a lot of biological screening tests.  At  Brookhaven we have
been using tradescantia, which is a plant system.  Our tradescantia has been
cloned to be very sensitive  to chemical mutagens.  It has  been used in a test
system trailer which has been going around  the country with  an EPA air pollution
monitoring system.  That is  an area where there may be some  development in
terms of a personal monitor.

     1 know that you can take the buds from the tradescantia system and wear
it on the lapel.  As a proposal, maybe this is something—maybe we could de-
vise a corsage or something  that could be worn on an individual1s  lapel and
after a week of use and with some hydroponlc solution to keep this system wet
and moist, you would be able to score this.  I think this  is a very easy
system to score by continuing visible mutations.  This may indeed  give you an
indication of the biological activity of the atmosphere.

     I think that in terms of my own concerns for identifying those kinds of
pollutants that are of interest for  the coal conversion  and  oil shale industries,
we are looking at the occupational environment now.  Unless  these  systems are
quite tight and enclosed,  the emissions may be your CO and oxides  of nitrogen
of the future.  The commercialization and applications of  these systems are
expected to come on line within the  next 4  or 5 years, and they thus represent
a concern to us all.

     The monitoring  technology  is  there.  We are  in  trouble  in terms of de-
fining what to measure and  in  terms  of  Identifying health  effects.  If we are
going to try to eliminate Workmen1s  Compensation  claims  that are comparable
to the claims for black  lung, which  have been  reported to  be $1 billion annually,
then we are behind  in  our  efforts  to develop  the  technology.

     In regard  to  the  comments  made  by  the  developers  and  manufacturers of
these personal monitors,  one question that  came up  repeatedly is,  "Where is
the market?"  I think  that as  pollutant standards are  developed,  in terms of
defining what  is  a  safe  environment  and the frequency  and  the type of precision
that is required  in order to ensure  that environment,  that at least in the
occupational environment the instrumentation need is going to be  quite high.

     So, I think  that  there is that particular market  area.   There is a man-
date to make sure  that instrumentation is simple, uncum her some for the individ-
ual to wear, and  cheap.   If you are talking about sampling some representative
fraction of  the work force, you can't afford to stay in business  and buy
$5,000, high  technology devices to monitor a representative component of that
work force.  But  you are going to have to  have simple, lightweight devices.
That is where  I  think the passive dosimeters are needed.

       I think,  however,  that there are other  areas  where you need more in-
  formation than passive  dosimeters can provide,  such  as special survey devices
  for high hazards and potentially acute exposures.

       And there are areas where area monitors are appropriate.  If the work-
  place conditions are fairly stable and you are  protected, say, from micro-
  meterological  episodic  conditions, area monitors can provide the type of
  monitoring  that is required.

       This particular meeting  has been enlightening for me.  It has provided
  me  with an  insight in terms of the fact that not only in the occupational
  environment, but also in the  ambient environment,  are there instrumentation needs.

       I think there should be  a unified effort addressed to the suppliers so
  that  what technology is developed is useful  to  all sectors.

       MAGE:  Thank you very much,  Otto.  Our  next speaker is Matt Petrovick,
  from  EPA's  Health Effects Research Laboratory.

       PETROVICK:   I would like to reflect a little on my impression of this
  conference.  I  think I  agree  with the other  panelists in that it has been
  quite  an  enlightening and an  exciting educational experience.

       I  see  a number  of  opportunities  in so many areas that we are not going
  to be  able  to talk about them all.   But I would just like to reflect on a few.

      Specifically, in this  morning's  initial  session, Dr. Mage mentioned a
 wide variety of  the  types of  sensors  that we  are going to look at.  He touched
 on the  topic of  agricultural  pesticides.  Perhaps I might have missed a few
 sessions, but I didn't  hear too much  other comment in that direction.  Maybe
 I have a lot to  learn on the  sensors  that are used in this area.

      However,  I would like  to  identify  for this group an area where the badge-
 type membrane devices and sensors  that I have heard about here today can be
 very effectively applied  in terms of manufacturers' motivation, health effects
 monitoring,  and dosimeter measuring.  And that is specifically in the area of
 agricultural spraying; that is, agricultural spraying with aircraft.  If I
 can  just briefly give you some background information on this.  The Federal
 Aviation Administration  (FAA)  happens to be in a great state of alarm at the
 present time due to the  increased incidence of agricultural aircraft accidents.

     Not only are the pilots having problems  in the flying of these aircraft,

but the manpower that happens to be coordinating with these people have
problems.  The farmer's wife standing at the end of a row that is to be
sprayed, a coworker filling the spray tanks of the plane, and the field workers
are all exposed to the atomized droplets by direct skin contact.  When the
pilot finishes a summer of spraying, he might have acquired a significant
build-up of these pesticides, without being aware of it.  It is suspected
by the FAA that the pilots' perceptive and cognitive judgments may suffer.
For example, there could be central nervous system disorders which have to
do with motor reaction, sensory perception of the visual tract, and judgment.
It is that particular aspect that  the FAA feels might be responsible for the
pilot losing control of the aircraft and going in.  That is only one aspect.

     The other aspect is that there is not enough attention being given to
the person on the ground who is getting belted with a lot more pesticide than
the pilot.  Over a summer  of this  kind of activity, there is widespread op-
portunity to look at a population  that is exposed and isn't being monitored
very well.

     On  the other side of  the coin, from the  research standpoint and from the
clinical epidemiologlcal standpoint, EPA has  invested a large sum of money
into design and development  of  a clinical-environmental laboratory here in
Chapel Hill—a clean air facility  and  a controlled atmosphere facility.  It
just seems illogical that  we can*t transpose  some form of clinical protocol
out of a laboratory  of  this  kind and,  in  this specific instance, into a small,
single-cockpit laboratory  where that  pilot  is sitting.  These are air condi-
tioned cockpits which  leak like a  sieve.  The air conditioning flows back
through, picks up the  pesticide,  passes  it  through  the cockpit,  and the pilot
sits there and breathes  it.

     Some of  the  pilots  wear oxygen masks;  some of  them  don't.   But it seems
to me  that we  can create an effective payback to the taxpayer for  the invest-
ments  in these kinds of  research facilities by addressing problems  like  that
and actually  performing  an epidemiologic study in the area of pesticides where
a man  sits in a  fixed  chamber and you have the opportunity to look at phys-
iological  responses*

     As  experimenters, we are very fortunate.  The man is sitting in a  seat
relatively  still.   The EEC, respiration, and cardiac functions can be monitored
quite  easily.  The  Air Force has 30 years of clinical background experience
in  this  kind  of  an area.

      I am  advocating that we take other Interdisciplinary sciences and join


  them into the health effects area to look at the types of dosimeters and ex-
  posure devices we have here today, to merge medicine into computer science
  and into epidemiology, and to bring these disciplines together in a coordinated
  fashion in a public health initiative.

       I would like to identify that one area that stands out here, because I
  see a lot of opportunity for a simple study with immediate payoff.  1 see a
  lot of technology here, and it gives me the Impression that it is untapped
  and unapplied, yet the potential is staring us in the eye.  I think it is
  incumbent upon this conference to somehow pull these attributes together.

       A second point might be:   How could we pull these attributes together;
  that is, to take advantage of the knowledge that we all have to offer and that
  we  all have to gain and learn?  I might like to make a suggestion here because
  1 have had the feeling that many of us are doing our own thing.  Many of us
  appear to be running in opposite directions.  We have the organic chemist who
  is  in the corner doing his thing, and we have the physiologist in his corner
  doing his own thing.  Wouldn't it be more effective if we could all communicate
  in  a professional manner through the formation of—for a better choice of words—
  a society for personal monitoring in health effects?  Why not have an organi-
  zation of the interdisciplinary sciences of medicine, epidemiology, computer
  science,  biomedical engineering,  organic chemistry, biochemistry, and micro-
  biology?   Why would we advocate such a position?  1 can reflect over the last
  20 years.   1  have seen many new disciplines come into being,  which showed
  the  same  symptoms that this conference is showing here today;  that is, there
  being  a  somewhat  unfocused direction and the embryonic signs that here is a
 whole  new  discipline emerging  before us.

      So, I would  think that one of the ways to pull this together would be
  to establish  such a  society and to follow it up with a professional journal,
 where  the  experience  and  research results can be published,  at least on a
 quarterly basis.   It  would  also help to have the different Federal agencies
 and  industries provide  a  society  of  this kind with  seed  money  to help it along
 and  to further implement  the educational processes  of that new interdisciplinary
 hybrid, the environmental  scientist.   For example,  the environmental scientist
 who  may have an organic chemistry background may also have an  epidemiologic

      So, if we can shape and train these people in  an academic environment,
 10 years from now we will have  the kinds  of  disciplines  and  the kinds of talent
 out  in the field that can effectively  respond  in a  coordinated fashion to
 solving problems in environmental health effects.

     It is rather obvious to me that we are not going to come up with  solutions
overnight.  I see this as a long-term effort.  And the need will exist as  long
as technological man walks the earth.

     So, I think it is incumbent upon this conference and future conferences
to seriously consider formally organizing such a society so we can  provide
feedback to the Federal agencies which might be providing the funds  to do
this kind of work and to do this kind of research for the increase of  knowl-
edge in health effects.

     I must also Include the area of microcomputers.  I think that  there
lies in this area a great opportunity for all of the disciplines, particularly
in organic chemistry and membrane technology.  From that standpoint, we should
be looking at the other disciplines to learn what has been done in  the last
20 years.

     For example, there has been a lot of membrane technology displayed here
today.  But if we look into the field of medicine, you will find that  in the
design of artificial kidneys,  in the design of blood oxygenators and artificial
hearts, there is a widespread  knowledge of membrane technology and  ion specific

     In summary, that  is my recommendation and observation.  Thank  you.

     MAGE:  Thank you very much, Matt.  Our next speaker is Dr. Manny  Shaw
of the InterScan Corporation.

     SHAW:  I do have  a  few  things  that I want  to  say.  But just before I
get into  that, I would like  to say  that in  regard  to  the SO. dosimeter—as
soon as we have a CO dosimeter, we  will have  an SO- dosimeter.  The S02
instrument we are now  making  is a  portable  monitor.   It is our second  largest
selling instrument.   So,  there is  no problem—not  just  for us, but  for some
of the  people who are  making  this  type of  a device.   It is here  now if you
want it.

     It has been  stated  by  a few  of the  people  here who are  the  Instrument
makers  that they  had  hoped  that the purpose of  this meeting would  be to make
a beginning of an understanding between EPA and  we who  have  to  make, develop,
and market  these  instruments.  I  think they are leaving today with a notion
that that hasn't  been done as yet.   Hopefully,  in the last hour or so  that
remains with  us,  and  possibly with the idea that Mr. Petrovick had expressed
of working  together through an actual society,  that might be the way  to go.


        Even prior to that,  however,  there are some things  that  can be  done.
   Now,  I  have been in the instrument world for some time.   Long before I  ever
   got  into  the instrument world,  I was in the technical  research world.   One of
   the  things that I  have noticed  in  particular with instruments, going back a
   decade  in regard to instruments that have been developed for  air monitoring,
   is that there is a very large credibility gap between  such  agencies  as  EPA
   and  other agencies who are  responsible for this type of  monitoring and  the
   industry  which is