EPA-600/2-76-034
February 1976
Environmental Protection Technology Series
DEVELOPMENT STRATEGY FOR
POLLUTANT DOSIMETRY
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five 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 five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on the effects of pollution on humans, plant and animal
species, and materials. Problems are assessed for their long- and short-term
influences. Investigations include formation, transport, and pathway studies to
determine the fate of pollutants and their effects. This work provides the technical
basis for setting standards to minimize undesirable changes in living organisms
in the aquatic, terrestrial, and atmospheric environments.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-034
February 1976
DEVELOPMENT STRATEGY FOR
POLLUTANT DOSIMETRY
J. W. Harrison
D. E. Gilbert
P. A. Lawless
J. H. White
Research Triangle Institute
P. O. Box 12194
Research Triangle Park, NC 27709
68-02-1731
Andrew E. O'Keeffe
Air Quality Measurement Methods Branch
Chemistry and Physics Laboratory
Research Triangle Park, N. C. 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
CHEMISTRY AND PHYSICS LABORATORY
RESEARCH TRIANGLE PARK, N. C. 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for pub-
lication. Approval does not signify that the contents necessarily re-
flect the views and policies of the U.S. Environmental Protection Agency/
endorsement or recommendation for use.
11
-------
ABSTRACT
This report comprises the results of a study to:
(1) Define and place realistic limits upon the needs of
epidemiologists for personal pollutant dosimeters;
(2) Identify pollutant-sensing principles that offer
reasonable opportunity for early development of functioning
dosimeters capable of operating within the limits of (1)
above;
(3) Project the impact of mechanical and electronic
miniaturization techniques upon each of the principal options
disclosed in (2) above;
(4) Assess the cost/benefit aspects of active vs. passive
sampling of the atmosphere, as applied to candidate sensors
identified in (1), (2) and (3) above;
(5) Assess the advantages, if any, of a systems approach to
dosimeter development, in which the design of a sensor for
a given pollutant would be constrained to make maximum use
of components & subassemblies common to sensors for other
pollutants.
The study reported on has been arbitrarily limited to consideration of
devices applicable to the measurement of ozone, sulfur dioxide and
nitrogen dioxide. The principles developed will apply in some degree to
the eventual design of dosimeters for other pollutants.
111
-------
TABLE OF CONTENTS
PAGE
ABSTRACT iii
LIST OF FIGURES viii
LIST OF TABLES xi
SECTION
1.0 INTRODUCTION 1
1.1 Criteria for Method Selection 2
1.2 Study Organization 5
2.0 SAMPLE ACQUISITION METHODS,
2.1 Passive Sampling 6
2.1.1 Flat Plate Absorbers 6
2.1.2 Diffusion Tube Dosimeter 7
2.1.3 Summary 18
2.2 Active Sampling 18
3.0 SORPTION METHODS 21
3.1 Sorption Mechanisms 21
3.1.1 Reversible Sorption 21
3.1.2 Irreversible Sorption 23
3.2 Sorbents for Sulfur Dioxide 25
3.3 Sorbents for Nitrogen Dioxide 30
3.4 Sorbents for Ozone 35
4.0 METHODS FOR CONTINUAL MONITORING 39
4.1 Conduc tome trie Analyzer 40
4.1.1 Prior Use 40
4.1.2 Model 41
4.1.3 Projected Performance 50
4.1.4 Logistical Factors 52
4.1.5 Dose Rate Information 52
4.2 Halogen/Halide Redox Amperometry 52
4.2.1 Prior Use 54
4.2.2 Model 55
4.2.3 Projected Performance 61
4.2.4 Logistical Factors 62
4.2.5 Dose Rate Information 62
-------
TABLE OF CONTENTS (continued)
SECTION PAGE
4.0 METHODS FOR CONTINUAL MONITORING (continued) .....
4.3 Halogen Generation Amperometry ......... 62
4.3.1 Prior Use ................ 62
4.3.2 Model .................. 63
4.3.3 Projected Performance .......... 66
4.3.4 Logistical Factors ............ 67
4.3.5 Dose Rate Information .......... 67
4.4 Membrane Amperemeters .............. 67
4.4.1 Prior Use ................ 6"
4.4.2 Model .................. 69
4.4.3 Projected Performance .......... 73
4.4.4 Logistical Factors ............ 74
4.4.5 Dose Rate Information .......... 74
4.5 Color ime trie Methods .............. 74
4.5.1 Prior Use ................ 75
4.5.2 Model Detectors ............. 78
4.5.3 Projected Performance .......... 91
4.5.4 Logistical Factors ............ 92
4.5.5 Dose Rate Information .......... 92
4.6 Gas Phase Luminescence ............. 92
4.6.1 Prior Use ................ 93
4.6.2 Model .................. 94
4.6.3 Projected Performance .......... 97
4.6.4 Logistical Factors ............ 98
4.6.5 Dose Rate Information .......... 98
4.7 Gas /Liquid Phase Luminescence .......... 99
4.7.1 Prior Use ................ 99
4.7.2 Model .................. 100
4.7.3 Projected Performance .......... 103
4.7.4 Logistical Factors ............ 103
4.7.5 Dose Rate Information .......... 103
4.8 Gas /Solid Luminescence .............
4.8.1 Prior Use ................ 1°4
4.8.2 Model .................. 105
4.8.3 Projected Performance .......... 1°5
4.8.4 Logistical Factors ............ 1°6
4.8.5 Dose Rate Information .......... 106
vi
-------
TABLE OF CONTENTS (continued)
SECTION PAGE
4.0 METHODS FOR CONTINUAL MONITORING (continued)
4.9 lonization and Electrophoresis 106
4.9.1 Prior Use 106
4.9.2 Model 109
4.9.3 Projected Performance 115
4.9.4 Logistical Factors 115
4.9.5 Dose Rate Information 115
4.10 Piezoelectric Mass Monitor 115
4.10.1 Prior Use 115
4.10.2 Model 117
4.10.3 Projected Performance 126
4.10.4 Logistical Factors 127
4.10.5 Dose Rate Information 127
5.0 DATA ACQUISITION 128
5.1 General 128
5.2 Magnetic Tape System 137
5.3 Solid State Memory System 143
5.3.1 MOS-LSI Memory Systems 145
5.3.2 MNOS Memory System 148
5.3.3 Summary 149
6.0 INTEGRATED DOSE METHODS 152
6.1 Solid Adsorbents/Passive Sampling 152
6.2 Solid Adsorbents/Active Sampling 154
6.3 Liquid Adsorbents/Passive Sampling 156
6.4 Liquid Adsorbents/Active Sampling 156
6.5 Packaged Vacuum 157
7.0 DOSIMETRY SYSTEMS 159
7.1 Integrated Dose Methods 159
7.2 Continual Dose Methods 163
8.0 CONCLUSIONS AND RECOMMENDATIONS 167
8.1 Recommendations 169
REFERENCES 171
vii
-------
LIST OF FIGURES
FIGURE PAGE
1.1 Schematic Diagram of a Single Channel Pollutant
Dosimeter for Continual Exposure Measurements 4
2.1 Air Incident upon Absorber and Substrate Only 8
2.2 Air Incident upon Absorber and Substrate Mounted
on Larger Area Carrier 8
2.3 Normalized Concentration in the Diffusion Tube as
a Function of Position along the Tube for Various
Times after a Step Change in Concentration 12
2.4 Normalized Current in the Diffusion Tube as a
Function of Position and Time 14
2.5 Normalized Amount of Material Passing the Inlet of
the Diffusion Tube (x = 0) and Being Collected
(x = L) as a Function of Time 15
2.6 Normalized Response of the Diffusion Tube Sampler to
a Pulse Input, with the Amount of Material Passing
the Inlet (x = 0), as a Function of Time 17
3.1 Adsorption Hysteresis for Water Vapor in Titania
Gel 24
3.2 Relation between Pore Size and Diffusion
Coefficient 24
3.3 Sulfur Dioxide Reaction Mechanisms 25
3.4 Cellulose Nitrate Matrix 28
4.1 Basic Conductance Cell Arrangement 42
4.2 Bridge Arrangement for Conductance Cells 42
O Q
4.3 Conductivity Cell Design of Hall 46
4.4 Change in Conductivity Cell Current for Various
Concentrations of Sulfur Dioxide 49
4.5 Change in Conductivity Cell Current for Various
Concentrations of Nitrogen Dioxide 51
viii
-------
LIST OF FIGURES (continued)
FIGURE PAGE
4.6 Schematic Diagram of Conductance Amplifier
Circuit39 53
4.7 Kromhyr Cell 56
4.8 LindqvisC Cell 56
4.9 Signal Processor for Differential Current
Measurements
59
4.10 Model Cell for Halogen Generation/Amperometrie
Measurement Method 64
4.11 Model Membrane Amperemeter 70
4.12 Model Colorimeter Detectors 79
4.13 Transparent Film Substrate Colorimetric Detector 83
4.14 Tape System Using Fibrous Tape 86
4.15 Measurement Using Surface Reflectance Change
due to Pollutant Exposure 88
4.16 Remission Spectra of Pure CuSO,.5H 0 at Five
Different Crystal Sizes 89
4.17 Fraction R of Linearly Polarized Light after
the Reflection on Paper Dye with Rhodamin,
for Various Angles of Incidence a as Parameter
and Dependent on the Observation Angle Q 90
4.18 lonization/Electrophoresis Apparatus of Vree
and Fontijn 108
4.19 lonization/Electrophoresis Detector 110
4.20 Equilibrium Distributions of Clusters H (HJD)
Predicted by Experimental Data115 . . . 7 .n Ill
4.21 Orientation of AT and DT Cuts with Respect to
Major Axes of Quartz121 118
ix
-------
LIST OF FIGURES (Continued)
FIGURE PAGE
4.22 Effect of Minor Orientation Changes on
Temperature Effects in AT Cut Quartz 122 119
4.23 Family of Frequency vs. Temperature Curves
for AT Cut Quartz122 120
4.24 Zero Temperature Coefficient Operating Temperatures
for AT Cut Quartz as a Function of Orientation122 .... 120
4.25 Detailed Portion of Figure 4.24 Near Room
Temperature122 122
4.26 Typical Frequency Response to a Change in Ambient
Temperature 122
5.1 Generalized Data Acquisition System 131
5.2 Incremental Digital Tape Recorder Data Collection
System 141
5.3 Data Selection Control 144
5.4 MOS LSI Memory System 147
5.5 MNOS Memory System 150
6.1 Pressure Components in Packed Tube Samplers 154
7.1 Integrated Dosimeter Operation Scheme 160
7.2 Dose Measurements from Integral Methods 162
7.3 Continual Dose Rate Measurement Operational Scheme .... 164
7.4 Data Processing for Dose Rate Records from Continual
Monitors 166
-------
LIST OF TABLES
TABLE PAGE
1.1 Performance Goals for Champ Instruments ........ 2
2.1 Characteristics of Commercially Available
Sampling Pumps .................... 19
3.1 Species with Lewis-Acid-Base Character ......... 26
3.2 Typical Chemical and Physical Sulfur
Dioxide Adsorbents .................. 26
3.3 Relative Response of Amines and Non-Amines ....... 29
3.4 Liquid Sorbents for Nitrogen Dioxide .......... 32
3.5 Apparent Sorption Efficiencies ............. 36
3.6 Molar Absorptivities .................. 37
4.1 Projected Performance of Conductivity
Cell Analyzers .................... 50
4.2 Performance Summary for Differential Amperometric
Method for Ozone Measurement ............. 57
4.3 Projected Performance of Differential
Amperometric Cell Analyzers ............. 61
4.4 Projected Performance of Iodine Generator
Amperemeter ..................... 66
4.5 Features of ECT Analyzers ............... 68
4.6 Model SS-310 Interference Data ............. 68
CO
4.7 Advantages and Disadvantages of ECT Analyzers .... °9
4.8 Parameter, Values for Permeation of NOo in
Teflon54 at 25°C ................... 72
4.9 Relative Permeation pf NO, N02 and S02 in
Various Materials5 ................. 72
xi
-------
LIST OF TABLES (continued)
TABLE PAGE
4.10 Projected Performance of Membrane Amperemeters 74
88
4.11 SO Emission Quenching Data 95
4.12 Performance Values for Chemiluminescent
Monitors 98
4.13 Effect of Hydration on Mobility of SO-, NO-,
and 03 Molecules . 7. 112
5.1 Power Requirements and Other Logistic Factors
for Tape Recorder Unit 142
5.2 Comparison of Memory Systems of 64K Bits 151
xii
-------
SECTION 1
1.0 INTRODUCTION
The assessments of the effects of various pollutants on human subjects
requires measurements of the amounts of pollutant to which these subjects
are exposed over some period of time and correlation of these dosages with
the general state of health of these subjects. Dosimetry has been a long
standing required practice for personnel exposed to x-ray, y~ray an(i high
energy particle radiations. Two forms are used. The integrated dosage is
routinely measured by film badges which are periodically collected and
developed to obtain a "read-out" of the cumulative exposure of the indi-
vidual wearer to ionizing radiation. In non-routine measurement situations,
the dose rate is measured continuously, in order to determine the varia-
tions in intensity that may be encountered in a work situation. This is
important when dose rates are apt to be highly variable and/or unpredictable.
Either form of measurement could be contemplated for personal pollutant
dosimeter development. The integrated dose method is relatively simple
and inexpensive; however, it does not give enough definition to the
measurement to allow correlation of dose rate with activity. For example,
a human subject will require greater ventilation under stress. If the
periods of higher ventilation also occur simultaneously with exposure to
high pollutant concentration, the ingestion of pollutants and, therefore,
the dose may be significantly larger than if the subject were able to
"idle" when exposed to high pollutant concentration.
This has been recognized in a discussion concerning a proposed change
in the "significant harm" level and "emergency" action level for photo-
chemical oxidants. Here it was pointed out that although a one-hour
average was being used as a practical basis in assessing monitoring results,
exposure to high enough levels over periods of 15 minutes or shorter could
also lead to significant adverse effects.
In view of this need, a major part of this study has been directed
to determining how present detection methods could be miniaturized to
provide reasonably accurate, portable instruments that could enable
continuous or, more properly, continual measurement of ambient S0?, NO-
and ozone with the data recorded in a format suitable for synchronizing
exposure history with the subject's activity during a realistic time period.
-------
Because economic or other factors might dictate that integrated dose
devices be used instead of, or as a complement to, continual monitors
principles underlying the design of these devices have also been examined,
particularly from the point of view of sample acquisition onto totally
absorbing materials via diffusion transport.
1.1 Criteria for Method Selection
Selection of one or more measurement principles for miniaturization
as a pollutant dosimeter is based upon several criteria:
1. Repeatibility - i.e. calibration stability,
2. Precision and accuracy of measurement,
3. Ease of maintenance and calibration,
4. Convenience of data readout,
5. Portability and ruggedness, and
6. Economic deployment.
These guidelines are interpreted as follows.
Any useful measurement device to be carried around by non-trained,
non-technical people must be repeatable. The zero drift and span change
with time and with ambient conditions must be minimal. This might be
achieved by various compensation techniques in the design or by automatic,
periodic internal re-zeroing and span setting cycles.
Precision and accuracy of these measurements are necessary if the
data are to be useful. Design goals for these have been set for CHAMP
program instruments as +2% precision and +5% accuracy. Table 1.1 lists
these and other performance goals for CHAMP instruments, which could be
also used as design goals for pollutant dosimeters.
Table 1.1. PERFORMANCE GOALS FOR CHAMP INSTRUMENTS
Drift /day
Pollutant Zero Span
Detection
Noise Limit
Response
Precision Time
03 <1%FS <0.5%
N02 <2%FS <5%
S02 <1%FS <2%
<2%FS
<0.5%FS
<2 ppb
4 ppb
5 ppb
5 ppb
2%FS
2%FS
2%FS
10 sec
3 sec
3 min
-------
The design should be amenable to rapid, periodic calibration checks
with an "external" calibration source of pollutants. These checks should
be administered by trained, technical personnel to assure that data
quality is maintained. At the same time routine maintenance should be
performed such as reagent replenishment (if required) and a step-by-step
verification that each functional component is performing within speci-
fications. A modular design concept should be employed that would simplify
functional component checks and replacement, if necessary.
The data format should be convenient for readout and manipulation as
necessary for interpretation. As will be seen later, there are many design
options for data recording and readout. Whichever option is selected will
be influenced by many factors, but the selection should be weighted by
considerations of obtaining the data in a form that enables correlation
of the exposure history with the subject's activity pattern, meteoro-
logical factors, etc., without excessive manpower being required to cast
it into readily usable form. Figure 1.1 indicates schematically a
dosimeter system. The last step from readout to "data ready for use"
should require a minimum amount of skilled manpower, particularly if
the number of subjects and periods of testing are large enough to provide
an adequate statistical basis for projection onto exposure of the general
population.
Portability and ruggedness are necessary design features if the
dosimeter package is to be transported around by normal human subjects.
Ideally, a lapel type device is desired. However, the current state of
the art in miniaturization, particularly when the foregoing constraints
are applied, does not allow this. A briefcase sized instrument package
capable of 10 hours of operation on batteries is a realistic design goal.
The weight of such a package should be less than 20 pounds. It should
be capable of operating unhindered by the impulse forces encountered in
normal activity. The magnitude of these forces is at present unknown,
but they could be estimated from a few acceleration measurements carried
out on a briefcase. It would be desirable if the dosimeter package could
function in any orientation. Certainly it must not be rendered inoperable
by being temporarily placed on its side. This places severe design
-------
Actual
Exposure
Time Reference
r
Transducer
Recorder
T
Readout
t
Data for Use
Periodic Calibration
Exposure
Figure 1.1. Schematic Diagram of a Single Channel Pollutant Dosimeter
for Continual Exposure Measurements.
constraints on those measurement principles which use liquid reagents.
It is possible that this requirement could be relaxed somewhat by
cautioning the human subject to be careful about orientation. However,
it seems unrealistic to expect complete compliance with such a request.
The deployment of a personal dosimeter network and operation of a
system to maintain it, assure data quality and then process the data to
obtain desired research information from it represents a sizable invest-
ment. The anticipated results must justify the commitment. The measurement
methods chosen for development must depend upon existing technology for
miniaturization. It is not feasible to launch a multiplicity of separate
research efforts to obtain miniaturized components. Wherever possible, as
many common components as possible should be used in the design of the
separate dosimeter instruments for S02, N02 and ozone. These components
should, if possible, be obtainable commercially from more than one source.
If manufacture by special order is necessary, the manufacturing operations
-------
required should be routine and use commonly available materials. Only
then can reasonable cost and delivery of initial items and replacement
items be expected. In addition, the cost of maintenance and calibration
must be taken into account as well as the cost of data retrieved. At
this stage of development of the dosimeter concept, only guideline state-
ments can be made about the economics of deployment. Realistic cost
estimates can only be made on the basis of a firm prototype design.
1.2 Study Organization
This report is organized to cover the topic as follows. First,
sample acquisition methods are discussed from the point of view of
passive versus active sampling. Then the various methods in which
pollutants can be extracted from the ambient air are discussed under
the heading of sorption methods. Following this, a review of the prin-
ciples of the methods used for SO , NO- and ozone detection is made
and a mathematical model developed for each where practicable in order
to assess their performance when "scaled down" to a convenient size
for portable monitoring devices. A discussion of state-of-the-art
methods for electronic data recording is next. Subsequently, various
options for deployment and dosage read-out are reviewed. Finally,
recommendations are made as to the methods which appear most feasible.
-------
SECTION 2
2.0 SAMPLE ACQUISITION METHODS
Before a pollutant can be detected or collected to provide information
on concentration or dosage, ambient air must be contacted with the detection
or collection device. The contacting can be brought about either through
natural convection and diffusion, hereinafter termed "passive sampling,"
or through forced convection, hereinafter called "active sampling."
Passive sampling has advantages of simplicity, no power drain and, hence,
low cost. It is suitable for applications where integrated dose measure-
ments are sought and where response time is not an important factor. The
characteristics, merits, and drawbacks are discussed below in Section 2.1.
Active sampling provides a continual supply of air to the detecting
or collecting device. This is required by many measurement methods,
particularly those classified as "dynamic," when the time resolution of
the measurement must be small in order to register transient concentration
changes. Such variations can occur due to "wafting" —i.e., eddy currents
or plumes of relatively high pollutant concentration flowing from some
point source. The exchange for such speed of response is an increase in
complexity, power drain and cost. The characteristics, merits and draw-
backs of active sampling are discussed in Section 2.2.
2.1 Passive Sampling
Transport of vapor species through ambient air occurs through natural
convection and diffusion. The speeds of transport for these two mechanisms
are vastly different, and because of the variability of convective transport
with prevailing weather conditions, the "supply" of a particular pollutant
from a particular source to some point where a standard observer is located
may fluctuate widely. Two modes of passive sampling will be assessed, the
"flat plate" and the diffusion tube, in the light of such fluctuations.
2.1.1 Flat Plate Absorbers
It will be assumed that a specific (for a given pollutant),
totally absorbing material coats an appropriate substrate and is carried
in a holder that permits a well-defined, flat area, A, to be exposed for
some specified period of time, TD, to ambient air. Unless a very high
velocity air stream is directed to impinge normally on this flat surface,
6
-------
there will exist a layer of stagnant air—a boundary layer—between the
surface of the absorber and the ambient air through which the pollutant must
diffuse to be collected. The thickness of this layer, 6, will depend upon
the incident velocity of the air and the size of the body to which area A
is attached. This is illustrated in Figures 2.1 and 2.2. In Figure 2.1,
only the absorber area A is used to intercept the air stream, of initial
velocity v . Depending upon whether or not the air is incident from the
front (absorber coating side) or back, the air flow pattern and, hence
the boundary layer thickness will be different. An even greater effect
is shown in Figure 2.2 where the absorber is carried on a much larger
body that will appreciably alter the air flow pattern.
These are illustrations of size and orientation effects on the air
flow geometry in the vicinity of the absorber. Size will vary from
person to person, and orientation, given the same size absorber and carrier,
may vary from minute to minute. Therefore, for a given pollutant concen-
tration, CQ, in the ambient air stream incident on the carrier, there may
be a large variation in the actual transport rate of pollutant by diffusion,
due primarily to the alteration of the stagnant layer geometry. Conse-
quently, it is very difficult to quantitate the dosage as a concentration
time product.
This difficulty is sidestepped in part by the diffusion tube sampler
which will be considered next.
2.1.2 Diffusion Tube Dosimeter
The diffusion tube dosimeter is a simple device for measuring the
exposure to a particular gas in ambient air. It consists of a tube of
fixed length and area with one end open to the air and the other end
ending in a collecting medium. The collecting medium is assumed to be
specific for the gas to be sampled, and, ideally, will absorb all of the
particular gas that reaches it.
Experimental work on diffusion samplers has been done by Palmes and
2 3
Gunnison and Gunnison, Gunnison, and Palmes for water vapor, S0«, and
N02. The results have encouraged further development of the technique.
The analysis of the behavior of the diffusion tube is simplest in
the steady state. It is assumed that the concentration of the component
-------
Absorber
& Substrate
Figure 2.1. Air Incident upon Absorber and Substrate Only.
Figure 2.2.
Air Incident upon Absorber and Substrate Mounted on Larger
Area Carrier.
8
-------
in the air being sampled is given by CQ, that the concentration in air at
the surface of the absorber is zero, and that the temperature is held
constant.
The equation describing diffusion in this case is
J.-D-fS, (2.D
oX
2
where J is the flux of mass (mole/cm /sec); D is the diffusion constant;
and C the concentration as a function of position within the tube. The
total mass transferred in time t is
C t
Q = JAt = D-A—7- , (2.2)
Li
where A is the area of the tube and L is its length.
The gas which is collected for a time t is analyzed by means
appropriate to the medium and the average exposure (ppm/hr) can be
calculated.
Experimental verification of Equation 2.1 in small diffusion tube
samplers indicates that it is correct over a range of about 3 to 1 in
A
—. The overall accuracy of the collected sample is roughly 90 + 20%,
LI
that is to say, there is a fairly large scatter among various tube sizes.
This may be due to unaccounted for effects of tube geometry or it may
simply represent a certain variability of the collection and analysis
process.
The experimental work has generally been done at constant pressure
and temperature. In real life sampling, these variables will not be
controlled and it is necessary to assess their effects on the collected
samples.
The temperature dependence of the diffusion constant D is T , where
n is between 1.5 and 2.0 for various theoretical models (T is absolute
temperature). For a temperature range of 0°F to 100°F, this gives a
2
range of D from .76 to 1.13 of the value at 68°F (assuming T dependence).
-------
Similarly the pressure dependence of the diffusion constant is P
However, ambient air pressure will rarely change +5 cm.Hg. from the
standard value (76 cm.Hg.)> and this amounts to a variation of +6.6% in D.
It might be expected that air turbulence would seriously affect the
amount of material collected. However, wind tunnel studies have demon-
strated that wind velocities at 140 and 440 ft/min have essentially no
effect at most angles of incidence, with perhaps a small (20%) increase
of material captured at 45°-60°.
Thus, the contributions of these three effects—temperature, pressure
and turbulence—are generally less than the variations of the collection
and analysis process. Of these, temperature has the greatest effect
because the person wearing the sampler can easily be exposed to the
extremes used in the calculations. If the average temperature is higher
than 68°F, then the sampler will overestimate the integrated exposure of
the wearer. If the average temperature is less than 68°F, then the
sampler will underestimate the exposure. However, if the average temper-
ature is known, a correction can be estimated for the exposure which will
reduce the error due to temperature well below the error level of the
sampling process.
In general, the diffusion tube sampler will collect 90% of the
material entering its port, with a scatter of +20%. By standardizing
a particular tube geometry, it is estimated that the statistical errors
can be reduced to about +10% with all effects included.
To be realistic the response of a diffusion tube exposed to con-
centrations that vary with time should be considered. The equation
describing this reponse is, for a one dimensional model,
— - n **- C fj ON
- D 2 ' U>J'
3t 3x
10
-------
It will be assumed that a. step change in concentration occurs at the tube
inlet, which can be described by the following boundary conditions:
C = C
0
C = 0
at t = 0: C(x = 0) = 0 C(L) = 0
at t = »: c(x = 0) = CQ C(L) = 0
The partial differential equation (2.3) is separable and leads to a
solution of the form:
C(x,t) = Cn[(l - f) - \ 2Lj \ cos fp e~n (2L} Dt] . (2.4)
u LI z. - ^ .- £. Z.LJ
TT n=l,3,5... n
The term in parentheses is the steady state solution of Equation (2.1)
Figure 2.3 shows the concentration as a function of position along
the tube for various times after the step change in concentration.
Evaluation of the current J(x,t) can be obtained by differentiating
Equation (2.4) term by term. Thus,
CnD i / -i 2^ N2^
tf X.N 0 n 4 / ^ 1 . mrx -n (-^7-) Dt-, /-0 ,-v
J(x,t) = —=— [1 - — ,oc ~ sin ~^T~ e 2L 1 • (2.5)
L TT n=l,j,.?... n ZL
Again, the first term corresponds to the steady state solution. Here
it is useful to note the behavior of the Fourier components. Because
2
the exponential factor depends on n , and because n is restricted to odd
f\
values, by the time (-rr-) Dt = 1, most of the contribution is from the
n = 1 term.
An effective time constant, T, can be defined by the relation
J(L,T) = .63 J(L,~) (2.6)
which gives
11
-------
4-1
«v
X
N—'
u
1.0
.8
.6
.4
.2
0
0
.4
.8
1.0
JC
L
Figure 2.3 Normalized concentration in the diffusion tube as a function
of position along the tube for various times after a step
IT 2
change in concentration, a = 0^=-) Dt
12
-------
The behavior of J for t > T is essentially exponential in its
approach to the limiting value.
Figure 2.4 shows the current, J, as a function of position along
the tube for various times.
Since the main interest is in the integrated quantity of material
collected, this quantity will be calculated by
J-
J0
dt
GDt -2
0Dt 41 1 . mrx , -na) ] (2.8)
f~*. "a Sin 2lT (1 - *
n=l,3,5... n
where a = (~) Dt .
Again, the first term gives the steady state response. The effect of
the Fourier components is to delay the effective collection of material.
Collection for a time t » T will give Q(t) = QQ(t-T), where QQ is the
amount of material passing through the open end of the tube.
To visualize the size of the numbers, assume a tube about 6.4 cm
2
long and a diffusion constant of .144 cm /sec (NC^ at 68°F). These
parameter values give a T of 140 seconds.
For all practical purposes steady state is reached in 3i (95% response)
or 420 seconds. If the concentration in ambient were to go to zero at time
t,, then the amount of material collected would be essentially constant
at t_ + 420 seconds.
Figure 2.5 shows the amount of material passing the inlet (x = 0)
and the amount of material collected at x = L as a function of time after
the step change in concentration. The amount continuously increases with
time, and eventually the material is collected at the same rate at x = L
as passes through the inlet. Then the amount collected at any time is
equal to the amount which had passed through the inlet T seconds earlier.
Another way of describing the result is that an amount of material equal
to J-T is between the inlet and collection area once the steady state is
reached.
13
-------
CL—°°
0
2C
L
Figure 2.4. Normalized Current in the Diffusion Tube as a Function of
2
Position and Time, a = (-- Dt.
14
-------
3.2
Figure 3. Normalized amount of material passing the inlet of the
diffusion tube (x=0) and being collected (x=L) as a
7T-2
function of time.
Dt. The dashed line is
(ii \
2L}
the limiting amount of material being collected; it
extrapolates backward to zero at a = T.
15
-------
As a final step in the analysis, the effect of wafting will be
estimated by analyzing response to a pulsed input of pollutant. Suppose
at time zero the concentration C_ goes from 0 to some value, and at time
t.. returns to zero. The amount of material collected in the diffusion
tube after the steady state has been finally obtained will be calculated.
Because the diffusion equation (2.3) is linear, solutions may be
superposed to obtain a desired result. In particular, a step to concen-
tration C_ at t = 0 is assumed, and a step to -CQ at t = t, is added.
The superposed responses to these steps lead to
rt/« \ 0 ,- 4 1 / 1. mrx ,, -n a),
Q(N,t) =— [1--- L^ ^sin-^Cl-e ]
n=l,3,5... n
(2.9)
C D(t-t-) , 1 / 1 . mrx . -n B) -,
Q _ 1 ri £ 1. ' ^ — o sin -r::— (1 - e J ,
L l IT B n=l,3,5... n /L
where a = (^=-) Dt and B = (rr-) D(t-t-). This equation is the pulse response
for t > t^ The amount of material collected is given by Q(L,°°). The
exponentials will be zero at t = °° and sin -r~ will alternate + 1 depending
^Ju
on n at x=L. Therefore,
C.Dt , , ^^ ,
nt-t \ 0 Tt ** 1 > 1 . rnr,
Q(L.-) = — [1 - - — T- jLi -3 sin y-]
(r-) Dt n odd n
(2.10)
"J /i. *_ \ ri ^ J- X -1- _j_ **" 1
—r- Sin -T—I
C0D
Q(L,») = -_
This result is independent of the length of time t, in comparison with
T, provided the sampling continues several time constants beyond the
end of the pulse. In fact, if the product C-t., is held constant as the
pulse length varies, then the same integrated dosage will be measured.
This is important, because any random pulse shape can be approximated
with a number of small rectangular pulses. Figure 2.6 shows the amount
of material collected after a rectangular pulse as a function of time.
L6
-------
c0(«)
1
0
1.6 _
x=0
0 .4
Figure 4. Normalized response of the diffusion tube sampler to a pulse
input, with the amount of material passing the inlet (x=0) .
TT
as a function of time a = (TTT) Dt.
17
-------
2.1.3 Summary
The diffusion tube sampler appears to be a good method of passively
acquiring a sample for measuring pollutant exposure. It has a slow tran-
sient response, but measures the integrated dosage reliably. The char-
acteristic time constant is a function of the length of the tube and the
diffusion constant of the gas, but does not enter significantly in the
integrated response. The diffusion sampler is relatively insensitive to
ambient pressure level changes, and to wind velocities. It is somewhat
more sensitive to temperature changes, but corrections could be applied
fairly easily. Comparable or larger errors will probably occur due to
non-specificity of the absorber and in the preparation of the absorbtion
medium for analysis of the absorbed pollutant.
2.2 Active Sampling
Many of the continual monitoring methods discussed in Section 4.0
of this report require acquisition and presentation of either a pulsed
or steady air stream to the sensor. This has the advantages of metering
a fairly precisely known quantity of air during the measurement and of
responding to changes in the ambient concentration due to the continual
supply of fresh sample to the sensor. These advantages are countered by
the increased cost, complexity and, usually, the necessity for a portable
electrical power source.
Because of the increasingly stringent requirements for occupational
and environmental monitoring, many miniaturized air pumps have been
developed which are capable of operating from a battery supply for periods
of typically 8 to 10 hours, at air sample flow rates of anywhere from
20 mA/min to several liters per minute. A representative list of such
commercially available pumps is given in Table 2.1.
Design of the sample air flow system should minimize obstructions
to flow in order to keep the pressure drop, AP, within that rated for
the pump. Otherwise the flow rate will be decreased below the design
level. For some measurement methods—e.g., those in which the sample
air stream is flowed over a sensor and not appreciably depleted by
sorption in the process—this will not be critical. However, for methods
18
-------
Table 2.1. Characteristics of Commercially Available Sampling Pumps
Mfr. -Model
Micronair
(Unico)
Micronair
(Telmatic)
Casella Mk II
Mis co
SKC
Unico C115
Sipin SP-1
Monitair
Spectrex
AS-100,120
Volts
3.75
8.4
6.4
3.0
6.0
6.0
6.0
Amp Hours
1.0
0.5
0.9
0.5
2.0
,32A
Duration in use(hr)
8-10
5
8-10
6-10
8
14
8
8
4
8
AP
in H20
4.0
12.0
17.7
1.4
25.0
4.0
12.0
2.5
12.0
4.0
Rated Flow
mA/min
2000
2800
2000
2800
2000
200
2000
2800
200
2800
2000
1400
Wt(g)
510
1000
600
570
1000
1360
1670
540
740
170
Size (mm)
156x83x64
185x120x100
127x79x57
110x90x30
165x89x45
165x140x63
130x64x38
107x102x51
51x51x51
Remarks
Adjustable
1.5-2.5 A/min
Adjustable
1-5 A/min
Adjustable to
3000 mA/min
Adjustable
40-200 mA/min
Adjustable
to 3.5 A/min
Adjustable
50-200 mA/min
Revised model
Pump only
-------
in which essentially all of the pollutant is removed by sorption as the
sample air is moved through the instrument, change in sample flow rate
can cause a proportionate error in measurement.
20
-------
SECTION 3
3.0 SORPTION METHODS
Most methods of pollutant measurement rely on extraction of the
gas phase pollutant from ambient air by contract with either an absorbing
solid or liquid phase or an absorbing liquid phase. The physical and
chemical nature of these methods is such that the adsorption methods can
be predominantly or partially reversible in nature, but the absorption
method is predominantly irreversible under normal conditions of temperature
and pressure.
3.1 Sorption Mechanisms
It is well known that surfaces of high specific surface (area/unit
mass) are capable of adsorbing gas and vapor phase molecules. The
extent of adsorption (and degree of reversibility) depends upon the
chemical nature of the solid and the physical nature of the solid. The
chemical nature of the solid surface is dependent on the nature of the
unsatisfied chemical bonds of the surface atoms and may also be dependent
upon the general electronic properties of the bulk material. This is
exemplified by many semiconductor materials and transition metals used
for heterogeneous catalysis. The physical nature of the solid surface,
particularly its surface area and porosity, also have a large effect.
3.1.1 Reversible Sorption
When adsorption occurs predominantly due to Van der Waals1 force of
attraction between the adsorbate gas molecules and the adsorbent substrate,
4
the resulting adsorption is termed "physical" adsorption . It is to a
large extent reversible. Because all molecules can exhibit the fluctu-
ating induced dipole moments that lead to Van der Waals attraction, this
type of adsorption is not specific. However, because some molecules
exhibit permanent dipole moments, they are more susceptible to physical
adsorption on most substrates; and if the gas phase concentration is high
enough and the temperature low enough, condensation may occur. A notorious
example is water vapor which is a strong competitor for adsorption sites
on many surfaces.
The Langmuir and BET isotherms, which describe the quantity of gas
adsorbed on a given material at constant temperature as a function of the
partial pressure of the gas, are usually derived theoretically and
21
-------
determined experimentally for only one type of gas present over the
sorbent substrate. A far more complicated situation arises when
several gas and vapor species are present simultaneously. Although it
is conceptually possible to set up a mathematical model describing such
a situation, as has been done to derive the Langmuir isotherm for the
6
case when two gases are being absorbed, the number of unknown parameter
values in such a theoretical model makes the application to a real
situation impractical, if not impossible. At present, one must rely
on an empirical approach when applying most sorbents in a measurement
scheme, particularly when trying to assess interference effects and
the effect of temperature on the amount of a particular gas adsorbed
relative to its partial pressure in the gas mixture in contact with
the sorbent.
Another method of obtaining reversible sorption is to coat a solid
phase with a suitable liquid which can take the vapor or gas species into
solution at some temperature, TI, and release it back into the vapor
phase at some higher temperature, T2, or back into the vapor phase when
the vapor phase concentration has been reduced to a lower value or zero.
The operative principles are diffusive transport from the "bulk" gas
phase across a stagnant boundary layer followed by adsorption at the
surface of the liquid phase and then, usually, absorption into the bulk
liquid. The equilibrium relation between the partial pressure, P.^, of
the species in the gas phase being sorbed and the concentration of this
species in solution, C., is given by Henry's law
P± = HC^ . (3.1)
For small values of P., H is independent of C., but is a function of
temperature.
Equation (3.1) is an equilibrium relation, and time is required
for diffusive processes in the gas and liquid phase to establish this
relation. Typically, gas phase diffusion constants are in the range
2
of 0.1 to 1 cm /sec at 300°K and vary as the 1.5 power of temperature
Q
and inversely with total gas pressure. The boundary layer is not large,
22
-------
being on the order of 0.1 cm. In contrast, the diffusion constant in
liquids is typically 10~ to 10~ cm /sec. Even if a high surface/volume
ratio is obtained for an absorbing liquid there will still be an appreciable
time lag between the establishment of a given partial pressure, P , over
the liquid and the attainment of the corresponding equilibrium concentration
given by (3.1).
The situation can be greatly improved by introducing convective trans-
9
port through turbulence in both the gas and liquid phases. The resulting
eddy-diffusion enables a much higher mass transfer rate. For a discussion
of the theory and practice of eddy-diffusive transport the reader is
referred to Reference 9.
3.1.2 Irreversible Sorption
It is possible for gas and vapor phase species to be irreversibly
(or quasi-irreversibly) taken up by both solids and liquids. Such
irreversible sorption can be due to a combination of both physical and
chemical factors. On micro-porous solids, capillary condensation can
lead to a marked hysteresis of the isotherm. An example, taken from
Reference 4, is shown in Figure 3.1. Here, water vapor adsorbed on
titania gel exhibits a complicated isotherm which is dependent upon the
history of exposure. Even if condensation does not occur it is possible
that very fine pores or intergranular seams can reduce the diffusion
coefficient of gas species. The relationship between diffusion coeffi-
cient and pore size is shown schematically in Figure 3.2, taken from
Reference 4. Although not strictly irreversible, such physical mechanisms
can lead to quasi-irreversible behavior that severely limits the dynamic
response of measurement methods which depend on such sorbents.
Another method through which sorption can be made irreversible is
through chemical reactions of the sorbed species with the sorbent to
chemically bind the sorbed material. Although the chemical reactions
through which this occurs may be strictly reversible, in practice the
reaction rates for the inverse reaction are negligible with respect to
the forward reaction rates. This can occur both in the chemisorption of
gas species on solid surfaces and in the chemical conversion of gas species
in solutions.
23
-------
d)
00
•rl
s
00
o
o
0)
ex
M
-------
3.2 Sorbents for Sulfur Dioxide
Sulfur dioxide, as indicated in Figure 3.3, may be regarded as a
tetrafunctional reagent; that is, it may undergo four modes of reaction.
Lewis base; reducing agent
+ 0*
Lewis acid; oxidizing agent
Figure 3.3. Sulfur Dioxide Reaction Mechanisms.
In the Lewis acid-base system, an acid-base reaction consists of the
donation of a pair of electrons from the base to the acid. In the sulfur
dioxide molecule, the double bond between the sulfur and oxygen atoms
enables this molecule to accept a pair of electrons and, therefore, act
as a Lewis acid. Furthermore, the ability to accept an electron pair
enables sulfur dioxide to undergo an oxidation-reduction reaction by
accepting a pair of electrons from a reducing agent. Likewise, the
unshared pair of electrons on the sulfur atoms enables sulfur dioxide to
function as a Lewis base by donating an electron pair or to act as a
reducing agent in an oxidation- reduction reaction.
The chemisorption of sulfur dioxide is determined by these possible
reaction mechanisms. Table 3.1 summarizes the types of species that would
be expected to exhibit Lewis acidity or basicity. Some examples of Lewis
bases, which will be utilized further in the discussion, are organic amines,
R-NH2 (R = organic group), and ami ne-modif ied silica gels,
a - CH2CH2CH2NH2
where a represents a silicon atom at the surface of silica gel. Another
common Lewis base used as a sorbent is the tetrachloromercurate anion.
25
-------
Table 3.1. Species with Lewis-Acid-Base Character
Lewis Acids
Lewis Bases
Anion
Molecules with double bonds
(except carbon-carbon double
bonds)
Halides in which the central
atom may exceed its ocetet
Cations
Molecules containing fifth- or
sixth-group atoms with an unshared
electron pair
Molecules with carbon-carbon
double bonds
Considering more specific examples of species, a representative list of
materials used as chemical and physical absorbents of sulfur dioxide is
presented in Table 3.2.
Table 3.2. Typical Chemical and Physical Sulfur Dioxide Adsorbents
Chemisorbents
Vanadium Oxide11 (V 0 )
13
Gold J
Gold-Nickel Alloy
13
Pyridine Benzene Copolymers
14
Lead Oxide (PbO )
Hydrogen Peroxide
15
15
Tetrachloromercurate (HgCl. ) Solutions
16
Organic Amines
Physical Adsorbents
Molecular Sieves (Silica Gel)
Activated Carbon
Zeolite10 (Al.O. • nSi00 • xH00 or Na00 • Al.O, • nSiO,
.. _ Z .3 Z Z L £. j t
Dawsonite
Polystyrene Foam Fibers Containing Activated Carbon
12
Nylon Foam Fibers Containing Activated Carbon
12
xH00)
26
-------
Of these materials the most promising appears to be the amines
because they can be selected "as is" or modified to control the
specificity and reversability of the adsorbate-adsorbent reaction.
Amine modification for selectivity or specificity can be demonstrated
by considering several examples. Silica gel, chemically modified by
conversion into
a - CH2CH2CH2NH2
where a represents a silicon atom at the surface of the silica gel,
selectivity adsorbs SO- and C0» in a Lewis acid-base reaction at 25°.
The carbon dioxide is liberated by temperature programmed desorption at
about 100°, but the sulfur dioxide is not released. If this gel is
converted to a-CH2CH2CH2NH Br~, a selective and reversible adsorbent
for sulfur dioxide results. This hydrobromide gel adsorbs sulfur dioxide
at 25° and releases it by temperature programmed desorption peaking
around 92°. Moreover, carbon dioxide is not adsorbed by the amine hydro-
bromide gel nor is nitrogen, oxygen, carbon monoxide, ethylene, ethane,
nitric oxide or nitrogen dioxide.
The amine hydroiodide gel a-CIUCH-CH-NH., I~ adsorbs sulfur dioxide
less strongly, but the adsorption can be increased by methylating the
hydroiodide gel to give a-CH2CH2NH2CH_ I . The addition of alkyl groups,
for example, the methyl group, increases the electron donor capacity of
the amine group by a positive inductive effect or releasing electrons to
the nitrogen. Hence, this alkylation should strengthen the Lewis acid-
base reaction and, as shown, increase the adsorption of sulfur dioxide
relative to the unalkylated amine hydroiodide gel.
The amine hydroiodide gel binds sulfur dioxide with a strength
intermediate between that of the hydrobromide and the hydroiodide gels
while the amine hydrofluoride gel does not absorb sulfur dioxide under
the same conditions.
The amine modified silica gels exemplify the specificity, reversi-
bility and modification potential of amines for pollutant dosimeters.
27
-------
In addition to silica gel amines, various organic amines have been
investigated for S02 sorptidn. The adsorption capacity for sulfur dioxide
of the following amine copolymers decreased in the order of 4-vinylpyridine
divinylbenzene, 2-methyl-5-vinylpyridine divinylbenzene, 2-vinylpyridine
14
divinylbenzene and 5-ethyl-2-vinylpyridine divinylbenzene. The copolymer
with the vinyl group in the number 4 position would be expected to be the
strongest Lewis base and, as seen, should adsorb more sulfur dioxide in a
Lewis acid-base reaction than the similar copolymers.
A comparison of the relative adsorption of several different amines
and several non-amine adsorbents for S0_ is given in Table 3.3. The
relative response is given in terms of resonant frequency change AF of
a quartz piezoelectric crystal when coated with each of the materials and
allowed to adsorb 802 under the same conditions. This measurement tech-
nique is described in detail in Section 4.5 of this report. Two different
3
cell volumes were used, Cell 1 of 2 cm internal volume and Cell 2 of
3
0.1 cm volume. For Cell 1, all the coatings except for the two asterisked
ones, are amines. The first non-amine is an ethylene oxide homopolymer
terminated at one end with an alkylphenol and at the other end with ethanol.
Cellobiose units, shown in Figure 3.4, constitute the matrix of the other
non-amine.
Figure 3.4 Cellulose Nitrate Matrix,
28
-------
16
Table 3.3. Relative Response of Amines and Non-Amines
3
Cell 1 (2 cm volume) AF
Tridodecylamine 160.
Diallylamine 70.
*Alkylphenoxy Poly (ethylene oxy)- Ethanol 69.
(Igepal CO-880-Appl. Sci. Lab.)
*Cellulose Nitrate 35.
Diallyl Melamine 24.
Melamine 4.
3
Cell 2 (0.1 cm volume)
Tripropylamine 225.
Styrene-Dimethylaminopropyl Maleimide 195.
(SDM-Uniroyal)
Polyvinyl Pyridine (PP-2040-Ionic Chem. Co.) 190.
Phenyldiethanolamine 150.
Quaternary Aliphatic Polyamine (PE-100-Ionac 68.
Chem. Co.)
*Polyamide (Versamid 900) 50.
*Vinyl methyl Silicone (UC-W98-Appl. Sci. Lab.) 45.
*2,2-(m-tolylimino)diethanol 3.
For Cell 2 the best adsorbing amines, as with Cell 1, are alkylated;
the first one, by three propyl groups and the second one by two methyl
groups. As expected, of the amines listed, these should be the strongest
Lewis bases and, hence, adsorb more sulfur dioxide.
From this, organic amines appear as promising candidates for a sulfur
dioxide adsorbent which exhibit (1) a large sulfur dioxide adsorption
coefficient, (2) specificity, (3) reversibility and (4) ease of modification
to enhance adsorption capacity, specificity and reversibility.
An irreversible S02 uptake is provided by conversion of S07 to a
sulfate. Such a method is used in the well known "lead candle" method for
IP 90
S02 measurement to obtain a solid phase sulfate. As far as can be
29
-------
determined, no analysis has been made of the detailed mechanism for the
transport of SCL from the gas phase to an adsorbed species which is
subsequently converted to sulfate; and therefore, a theretical relation-
ship between gas phase 802 concentration and solid phase sulfate weight
as a function of exposure history has not been worked out.
This method is sensitive to SO, concentrations up to 700 ppm, but
in general is not a quantitative method. The basic reaction depends
upon a number of factors which influence its completeness: wind velocity,
thickness of the lead dioxide, its wetness, the particle size, and purity*
Candles made from the same, batch of lead dioxide by the same technique
should give reliable relative exposures.
Aside from the positive interferences mentioned, methyl mercaptan
reduces the sulfation rate of reaction, although quantitative data is
not available.
Because of the importance of liquid absorbents in several accepted
methods of S0_ measurement from ambient air samples, many studies of
21
asborption efficiency have been made. Bostrom used an isotopic tracer
method to determine the collection efficiency of TCM and hydrogen peroxide
solutions. Both were found to have about 99% efficiency at flow rates of
1.2 £/min in the range from 0.9 pphm to about 10 pphm when the sample air
was bubbled through a solution column of at least 3 cm. A similar study
22
of hydrogen peroxide solution, carried out in essentially the same manner
also found about 99% collection efficiency.
3.3 Sorbents for Nitrogen Dioxide
The nitrogen dioxide molecule contains an unpaired electron and,
therefore, is paramagnetic. Moreover, this unpaired electron accounts
for the strongly electrophilic, oxidizing, and Lewis acid properties of
this molecule. Nitrogen dioxide is an electron or free radical scavenger
23
with an electron affinity of 3.99 eV and can be expected to react with
electron rich species such as Lewis bases or basic species. The nitrogen
24
dioxide molecule has a dipole moment of 0.58 Debye.
Likely candidate sorbents, then, would be materials containing electron
rich species such as the phenolate oxygen atom in p-cresol, or polar
species, such as a molecular sieve.
30
-------
A study of various absorption liquids for collecting N02 has been
25
reported by Nash. He operated on the hypothesis that since NO^ is such
a good electron acceptor, electron-donor charge transfer type solutions
should provide fast response time and high absorption capacity. He tested
the following groups of sorbents:
a) inorganic salts with reducing or polarizable ions
b) alkaline solutions of phenols and 6-diketones
c) aromatic amines
d) some "control" compounds, e.g. alkali, sulfamic acid
and unionized phenols.
Table 3.4 contains the results of this work and, in addition, pK and E° values.
The terminology C(50) and C(90) represents the corcentrations, in millimoles,
at which 50% and 90% more NO,, was collected than a control of either plain
water or water with alkali added to match the pH of the test absorber. The
terminology pK represents the negative logarithm of the equilibrium dissoci-
ation constant of an acid or base in aqueous solution. E° is the standard
oxidation potential in volts, V. The gas sampling flow rates, 1 £/min, and
concentration of N0_, 3 ppm in air, were the same for test solution and control
solution. By comparison, water with 0.1 N alkali was 14% efficient. A
solution of alkaline guaiacol (0-methoxy phenol (0.05%) in 0.1 N alkali)
was 97% efficient.
Examination of the four classes of liquid sorbents listed in Table 3.4
reveals that the nitrogen dioxide sorption efficiency of the phenols is
superior to the other classes represented.
For the first class of sorbents listed in Table 3.4, the phenols, a
study of the relationship between the sorption of nitrogen dioxide and
the basicity of the phenols is instructive. The pK values listed for the
phenols in Table 3.4 are the negative logarithms of the acid dissociation
constants. As seen in the table, p-cresol is more basic than phenol.
Thymol could be expected to dissociate less than p-cresol because the iso-
propyl group ortho to the hydroxyl should sterically hinder the dissociation
of the hydroxyl proton. Similarly, o-methoxy phenol would be expected to
dissociate less than phenol because a hydrogen bond can form between the
hydroxyl proton and the methoxy oxygen. The electrostatic interaction or
31
-------
Table 3.4 Liquid Sorbents for Nitrogen Dioxide
Concentrations (mM) Required to Collect
50% More and 90% More NO. Than Control
25
Solution
pK Values
24
E°,
alkali (pH = 13)
C(50) C(90)
Phenols
Phenol 5 100
p-Cresol 0.7 8
Thymol 0.1 1
o-methoxy phenol 0.07 0.7
g-Diketones
Barbituric Acid 30
Acetylacetone 4 40
Neutral Solutions
C(50) C(90)
Amines
Aniline 45
p-Anisidine 3
Inorganic Salts
Azide
Thiosulfate
Sulfanilate
Iodide
Sulfite
Ferrocyanide
1800
1250
1000
125
13
13
Blanks indicate 90% collection not
obtained for saturated solution.
9.89
10.17
4.01
94
9.42
9.71
-0.09
-0.535
-0.6
-0.565
32
-------
hydrogen bonding in guaiacol should be stronger than the steric hinderance
in thymol. Consequently, guaiacol should be more basic than thymol. It
appears that for the phenols, the basicity increases in going from the top
to bottom of the list in Table 3.4. Furthermore, as seen in this table the
sorption of the phenols increases as their basicities increase. Therefore,
it is recommended that if a researcher is investigating alkaline solutions
of various phenols for nitrogen dioxide sorption, he study phenol
derivatives with more basic properties than phenol.
The next sorbents listed are the 3-diketones. It is seen from the
pK values of these two acids that the acetylacetone is more basic than
barbituric acid. Again, the nitrogen dioxide sorption of these 3-diketones
increases as their basicity increases. Therefore, it is recommended that
if one is searching for alkaline solutions of 3-diketones for nitrogen
dioxide sorption, he investigate 3-diketones which are more basic
than acetylacetone.
The next class of sorbents listed in Table 3.4 are the amines. The
dissociation of amines in aqueous medium is illustrated in the following
equation,
-NH2 + H20 -NH3+ + OH~ .
As shown by this equation, the more basic or electron rich amine specie
would be the undissociated amine, ^-NH^. Consequently, of two amines,
the amine with the larger pK value would represent the lesser dissociated
molecule or the more electron rich specie. Using this criterion as the
measure of the availability of electrons, p-anisidine is more basic than
aniline and would be expected to sorb more nitrogen dioxide as seen in
Table 3.4. Again, using the principle that the correlation between the
pK values and the sorption of nitrogen dioxide suggests that the same
factors should govern both properties, it is recommended that if solutions
of amines are investigated for sorption properties, the amines with the
higher pK values of dissociation of the base should be chosen.
33
-------
•
The next group of sorbents listed 'in the table are the inorganic
salts. Comparison of the sorption concentrations of these salts with
the standard oxidation potentials shows that a relationship exists
between these two properties. As the oxidation potential of the salts
decrease, their sorption capabilities increase. By this correlation, it
is suggested that the same factors should govern both properties and that
if neutral solutions of salts are investigated as sorbents, the anions
with the lower oxidation potential should be chosen.
Sorption of nitrogen dioxide on solids will now be considered. In
particular, sorption by molecular sieves will be discussed.
Molecular sieves, a group of adsorptive desiccants developed by
Linde, are crystalline aluminosilicate materials, chemically similar to
many clays and feldspars and belonging to a class of minerals known as
zeolites. The outstanding characteristic of these zeolite materials is
their ability to undergo dehydration with little or no change in crystal
structure. The dehydrated crystals are interlaced with regularly spaced
channels of molecular dimensions. This network of uniform pores comprises
almost 50 percent of the total volume of crystals.
The general chemical formula for Type 4A is 0.96 + 0.04 Na20»1.00
A1203«1.92 + 0.09 Si02«xH20. Type 5A is produced from Type 4A through
ion exchange of about 75% of the sodium ions by calcium ions. The general
chemical formula for faujasite Type 13X is 0.83 + 0.05 Na20-1.00 A1203'2.48 +
0.03 Si02»xH20. Molecular sieves have a pH of approximately 10 and are
stable in the 5 to 12 pH range.
An important distinction between the faujasite type zeolites and the
type 5A is the interiors of the small pore system. In faujasite type
zeolite, the interiors of the small pore system coincide with the interiors
of the sodalite units or cages, while in the type 5A zeolite pores contain
calcium as well as sodium cations.
Molecular Sieves Type 4A, 5A, and 13X absorb only those molecules that
are small enough to enter their pore system. Type 4A has a port opening that
will permit only molecules smaller than about 4 Angstroms to enter the
cavities and be absorbed. Type 5A has a slightly larger pore opening which
34
-------
will admit molecules up to about 5 Angstroms in diameter. Type 13X has a
still larger pore opening which will admit molecules up to 10 Angstroms in
diameter.
>jf
Ma and Mancel have performed diffusion studies of nitrogen dioxide
on molecular sieve zeolites by gas chromatography. In this study, they
measured the effective diffusion coefficient of nitrogen dioxide in type 5A
and faujasite sieves. They reported that the effective diffusion coefficient
is approximately the same in both sieve materials. For a given gas, the
pore size of the sieve is normally regarded as an important factor which
affects the diffusional resistance. Since the pore diameter in the faujasite
sieve is twice as large as the type 5A, the fact that the effective diffusion
coefficients were approximately the same can be interpreted as a larger
diffusional resistance for the faujasite zeolite. Recalling that the faujasite
pores contained only sodium cations, while the type 5A contained sodium and
calcium, it is postulated that the increased diffusional resistance is due
to a specific interaction between nitrogen dioxide and sodium cations in
the interiors of the pores. Therefore, it is recommended that if molecular
sieves are investigated as sorbents for nitrogen dioxide, molecular sieves
containing pores lined with sodium cations should be studied.
3.4 Sorbents for Ozone
Ozone is a strong oxidant. This fact complicates its specific sorption
because ozone decomposition is catalyzed by surface contact. An idea of the
extent of the non-specificity of this material can be realized by studying
27
the investigations of Altshuller. He reported that of the various materials
studied he found that only glass or teflon did not catalyze ozone decomposition.
The oxidizing ability of ozone has been utilized as the basis of selecting
sorbents for its collection. Unfortunately, by the nature of its sorption
mechanism, ozone sorption for the most part is irreversible. Let us evalu-
ate and compare several different ozone sorbents.
The classic sorbent is a neutral or alkaline solution of potassium
iodide. The reaction of alkaline potassium iodide with ozone is
2KI + 03 + H20 t- I2 + 02 + 2KOH. This sorbent is sensitive to peroxides
.as well as nitrogen dioxide and sulfur dioxide.
35
-------
28
Another ozone sorbent is reduced phenolphthalein. This sorption
is based on the oxidation of reduced phenolphthalein to phenolphthalein.
The limitation of this sorbent is its sensitivity to any oxidant.
The redox indicator diphenylaminesulfonate is a sorbent superior to
29
iodometric sorbents in two respects. The diphenylaminesulfonate yields
a turquoise color with ozone, a violet color with gaseous chlorine and
peroxides and a yellowish-green color with nitrogen dioxide. In addition
to the different colors produced by interfering gases, the redox potential
is better adjusted to ozone, i.e., 0.85 V instead of 0.59 V for the
30
I2 - KI reaction.
Fluorescein is an ozone sorbent which is oxidized to fluorescein.
30
Fluorescein is reported to exhibit excellent sensitivity. The substance
4,4'-dimethoxystilbene sorbs ozone by cleavage of the double bond to
31
form anisaldehyde.
The collection efficiencies of the various sorbents have been reported.
Table 3.5 summarizes these efficiencies determined at the recommended sampling
rates.
Table 3.5. Apparent Sorption Efficiencies
Sorbent
Recommended Sampling
Rate (liters/min)
% Sorption Efficiency
at 3.0 ppm
Potassium Iodide
Phenolphthalein
Sodium Diphenylamina
Sulfonate
4,4'-Dimethoxystilbene
1.0
0.9
2.83
0.1
97.2
92.0
91.7
97.7
Examination of Table 3.5 reveals that the sorption efficiency is greatest
for 4,4f-dimethyloxystilbene.
The sensitivities of these sorbents can be evaluated by comparing
32
their molar absorptivity values reported in Table 3.6.
36
-------
Table 3.6. Molar Absorptivities
Sorbent
Molar Absorptivity
(absorbance cm mole )
Potassium Iodide
Phenolphthalein
Sodium Diphenylamine Sulfonate
4,4*-Dimethyoxystilbene
24,200
26,900
2,500
15,300
33
A simple, convenient, and useful sorbent is rubber. Standardized
strips of rubber are folded in the middle, and the ends are secured together
to produce a reproducible tension. The ozone exposure time of these
strips in minutes, times the ozone concentration in parts per minute,
gives a constant that has excellent reproducibility under standard
conditions.
The sorbents discussed so far sorb ozone irreversibly. Silica gels
and orthosilic acid polymers have been investigated for reversible ozone
adsorption. It has been shown that adsorption of ozone on silica gel
2 —1
from -117°to 20° with specific surface 360 m g is of the type shown
O /
by low adsorption heat and complete reversibility. No decomposition
34
of ozone on the gel was observed by a spectrophotometer.
The absorption mechanism of ozone adsorbed on silica gel has been
investigated by infrared spectroscopy. It was found that the ozone is
rendered comparatively unreactive as a result of strong hydrogen bond
oc
formation with the surface hydroxyl groups.
Investigations revealed a temperature dependent correlation between
r\f
ozone adsorption and silica gel structure. In an independent study,
37
Reimshuessel attained the same results. He evaluated the adsorption
of ozone at -80° on silica gel, that had been annealed at various
temperatures, as a function of the thermal history of the gel by the
application of the B.E.T. equation. At annealing temperatures of 105° to
108°, maximum values for the monolayer concentrations of ozone were
observed for silica gel that had been annealed at 150°, indicating that
37
-------
the monolayer concentration is affected by the type as well as by
the extent of surface hydration. Two surface silanol groups are
involved in the adsorption of one molecule of ozone and a hydrogen
bonding adsorbate-absorbent reaction is assumed. Annealing at less
than or equal to 105° results in a reduction of ozone adsorption and
in about a two-fold increase of monolayer concentration compared to
the value obtained after annealing at 150°> suggesting that ozone is
adsorbed only onto a water monolayer at these temperatures.
38
-------
SECTION 4
4.0 METHODS FOR CONTINUAL MONITORING
This section surveys principles which have been applied for the
measurement of S02> NO 2 and 0- in ambient air. For each method, the
prior use is reviewed to put into perspective the performance. Then
the measurement principles are mathematically modeled to enable the
calculation of projected performance from a miniaturized version.
Logistical factors bearing on portable operation are surveyed and the
type of dose rate information available is noted.
In order to place the utility of these methods in perspective,
consideration must be given to the size and weight of a convenient instru-
ment package. A "briefcase size" of about 12.7 cm x 30.5 cm x 45.7 cm
3
(5 in x 12 in x 18 in) will have about 16,500 cm of volume available.
A weight of about 7.0 Kg (15.4 Ibs) could be carried. These limits
can be compared to the volume and weight requirements of the methods
reviewed in this section.
Each of the methods requires active sampling. From the data
of Table 2.1 a typical pump with its associated battery would be
3
about 750 cm volume with a weight of about 0.75 Kg. This would operate
for up to 10 hours without recharge.
Many of the methods require liquid reagent, typically at a flow
rate of about 1 ml/min, per pollutant channel. For 24 hour operation
3
per channel, about 1500 cm of solution storage will be required. An
equal amount of storage is required for waste reagent, increasing the
3
required volume to 3,000 cm per channel. Some of the aqueous reagent
will be lost due to evaporation upon mixing with the sample air stream.
As an example, if sample air at 20% relative humidity and 25°C tempera-
3
ture flowing at 100 cm /min is contacted with aqueous reagent flowing
3
at 1 cm /min, the water loss from the reagent is 4%. However, such loss
will be dependent upon temperature and humidity of the sampled air.
Practically speaking, the spent reagent volume storage must equal
the supply volume.
39
-------
Another major volume and weight contribution will be required
for the electronic signal processing and data storage components and
their associated battery supplies. These requirements will depend
upon the design used. The size and power requirements for data storage
systems are discussed in Section 5.0. The largest volume is taken up
by a cassette-type digital tape recorder with its associated circuitry,
3
about 3200 cm . The weight is about 1.36 Kg. Among the electronic
components used in the methods discussed in this section, the largest
3
volume would be occupied by a photomultiplier tube, about 200 cm
including a magnetic shield and its associated inverter-type high
3
power supply, which would occupy about 250 cm . Power for the signal
processing and data collection electronics will require at least two
3
12.2 volt Ni-Cd rechargeable batteries of 205 cm volume, 570 g weight
each.
All of the required support items must be added to the volume
required for the basic measuring device itself in order to see whether
or not the "briefcase size" ideal can be met.
4.1 Conductometric Analyzer
This measurement principle utilizes the increase in electrical
conductance of an electrolyte solvent caused by the absorption of
ionizable chemical species to develop a signal proportional to the
liquid phase concentration of the ionized species. At low concentrations
this concentration can be made to be proportional to the gas phase
concentration of the ionizable species contacted with the absorbing
electrolyte.
4.1.1 Prior Use
Conductance cells were among the first widely applied instruments
for measuring S02 in ambient air. The instrumental construction is
relatively simple, but numerous practical problems limit the credibility
of the measurements obtained and the reliability of instruments used in
40
-------
the field. The most important problem areas are:
1. Susceptibility to interference by other gas or vapor species
which can be absorbed and ionized in the electrolyte.
2. Temperature dependence of conductivity and gas absorption
rate.
3. Establishment of good contact of ambient air sample with
absorbing electrolyte.
4. Necessity for a continuous supply (and disposal means) of
electrolyte of uniform (invariant with time) characteristics.
Similar remarks apply to the application of the conductance method to
NCL measurement.
Ozone does not ionize in aqueous solution. However, its ability
to oxidize halogens to produce ionic species in aqueous electrolytes
can be used to produce conductivity changes proportional to ozone
concentrations.
4.1.2 Model Detector
The model used is shown in Figure 4.1. A coaxial conductance
cell defined by an inner electrode of radius r, and an outer electrode
of radius r-, both of length L, is filled with an electrolyte solution
which flows through the annulus between the electrodes. A voltage
source, E, is connected in series with a current measuring amplifier
which provides an output voltage signal, e , proportional to current
i through the cell. If care is taken to avoid polarization effects
at the solution-electrode interfaces, the current is related to the
supply voltage by
i = GE (4.1)
where G is the conductance of the electrolyte in the annular cylinder.
The conductance is determined by the geometry of the cell and the
conductivity, g, of the electrolyte as
G = g 2irL/£n(r2/r1) . (4.2)
The geometric factors are constant for a given design.
41
-------
Electrolyte Influx
Electrolytt
Outflux
I J
Figure 4.1. Basic Conductance Cell Arrangement.
O"
zero
De-ionizer
Ref. Cell
Meas. Cell
Solvent
Reservoir
Figure 4.2. Bridge Arrangement for Conductance Cells.
42
-------
The conductivity will depend upon the types of ions, their
concentration, the type solvent used, and the temperature. These
factors will be examined. It is convenient to express g in terms
of the equivalent condutance, X, and the concentration, c, of the
ions in the electrolyte, or
g = cX/1000 (4.3)
where c is expressed in gram equivalents per liter of solution. When
there is a mixture of ions,
cX = Z XiCi . (4.4)
i
Since X. is a function of both concentration, in a given solvent, and
temperature, these factors must be examined. As a rule of thumb, the
temperature coefficient is about 2% per degree Celsius.
If it is assumed that water is used as a solvent and the absorbed
species concentrations are small enough to regard as infinite dilution,
standard equivalent conductances can be used to make trial calculations
of the sensitivity expected. Combining equations (4.1)-(4.4) gives
500 £n(r2/r;L)
(Z.c.X.)E . (4.5)
For small changes due to fluctuations in temperature, T, or supply
voltage, E,
3X.
(AI)N = 500 tofa^) (EEiCi §T AT + WlAE)- (4'6)
The fluctuations due to these sources will determine the minimum detectable
change in current. The temperature changes may be compensated by measuring
the solution temperature and adjusting the amplifer gain automatically with
43
-------
a feedback signal dependent upon the temperature signal. Alternatively the
conductance cell can be thermostated or a bridge arrangement used, as
sketched in Figure 4.2. It should be noted also that the temperature of the
ambient air is also important since (assuming the ideal gas law)
x = ~= P/RT . (4.7)
If the partial pressure remains constant, the volumetric concentration
will vary with temperature as
x dT
(4.8)
Ax AT
At T = 300°K, a 1°C change results in
— = -0.00333 . (4.9)
x
The error due to this source must be compared to the error from other
sources to determine whether or not it can be ignored.
Another important point in the design is the "mixer" indicated
schematically in Figure 4.2. This functions to contact the sampled air
stream with the absorbing solvent stream. The efficiency of this contact
and its reliability—in terms of constancy of efficiency with respect to
time and pollutant concentrations—is essential to the accuracy of the
instrument. Denoting by TI the mass fraction of pollutant absorbed into
the solvent stream from the sampled air stream, the mass of pollutant
taken in per unit time in the sample air stream is
dm
44
-------
where M is the molecular weight of the pollutant and Q the sample air
stream volumetric flow rate. The mass rate transferred to the liquid
steam is, by definition of n»
dm
The solvent volumetric flow rate, Q , along with the mass transfer rate
3C
from the air, given by (4.11), determines the mass concentration in the
liquid, and therefore the equivalent concentration
C = ZnxMQ /Q. (4.12)
3. X/
where Z is the number of equivalents per gram molecular weight. Use of
this expression, along with equations (4.5) and (4.6), enables the cal-
culation of current levels and fluctuations expected for certain situations,
For definiteness in modeling, the miniature conductivity cell and
38
gas-liquid contactor design of Hall will be used. These are shown in
3
Figure 4.3. This cell operates with solvent flow rates of 0.1-1 cm /min
3
and gas flow rates of 5-500 cm /min.
Neglecting (for purposes of calculation) the change in cell constant
due to the presence of the liquid exit tube, the value is approximately
. _ TrL TT x 0.794 _ -2
c 500 An(r2/r1) 500 £n(0.10/0.082)
Assume flow rates of
3
Q = 100 cm /min
cl
3
Q = 1 cm /min
with aqueous solvent containing sulfuric acid at a concentration of 5yM
45
-------
B
1
D
1
\
/
0.5 In.
H
0.25 In.
Figure 4.3. Conductivity Cell Design of Hall.
38
46
-------
(to suppress bicarbonate buffering) and hydrogen peroxide at 1%. An
21 22
absorption efficiency of 97-99% can be expected ' but n will be
conservatively estimated at 0.95.
The species to be detected is SO-, which will be assumed to be
present at a level of 20 pphm in air at 300°K. Therefore,
M = 64 grams/mole
x = 20 x io~8 x 4.05 x 10~2mole/liter
and since the reaction in solution is
2H20 + H202 + SO 2 -*• 2H30 + S0~
there will be 1 mole of negative charge per 32 grams of SO,,, or
Z = 2 equivalents/64 gm.
The background conductance due to the H~SO, and H_0 in the water
will be, at 25° C, approximately
8 " I5oo
=] - 4.29 x 10-6
The cell conductance at 25° C will be approximately
G = 2.5 x 10~2 x 4.29 x 1Q~6 = 1.07 x 10~7mho
For a supply potential of 1 volt, the current due to the electrolyte
solution is about 0.1 microampere.
47
-------
Using the atmospheric concentration and parameters assumed above, the
change in solution conductance will be
Ag =
or
Ag = 2 x 0.95 x 8.1 x 10~9 x 100(350 + 79) x 10
which represents a fractional change of
^& = 6.6 x 10~7/4.3 x 1Q~6 = 0.15
g
or 15%. This corresponds to a change in cell current of 16 nanoatnperes.
A plot of change in cell current, A., versus S02 concentration in air at
300°K for the assumed values in the model is shown in Figure 4.4.
The National Air Quality Standards for SO- (secondary standards)
range from an annual arithmetic mean of 2 pphm to a three hour average
of 50 pphm, which would correspond to a range of 1.6 nA to 40 nA current
change in the model conductivity cell.
A similar calculation can be made for N0_. Assuming an absorbing
25
solution of water with 5% efficiency the background conductance o±
the absorbing solution will be
g = 10~3[10~7 x 350 + 10~7 x 199] = 5.5 x 10~8 mho-cm2.
With the same air and solution flow rates assumed for the S02 calculation,
the equivalent concentration in solution for an ambient air concentration
_Q
of 20 pphm will be, from equation (4.12), 4.05 x 10 equivalents per liter.
— 8 2
This will cause a change in conductance of 1.78 x 10 mho-cm , or a
fractional change of
= 1.78 x 10~8/5.5 x 10~8 =0.32
g
48
-------
30 -.
co 20
0)
-------
This represents, for a cell voltage of 1 volt, a current change of
0.445 nanoampere. Calculations for a range of ambient air concentrations
of N02 give the curves shown in Figure 4.5. The National Air Quality
Standard of 5 pphm annual arithmetic mean would correspond to a current
of 0.106 nanoampere.
4.1.3 Projected Performance
The projected performance in terms of sensitivity (slope of
response curve) and fractional current change at 5 ppb desired detection
limit are shown in Table 4.1. The required stability and resolution are
better than 0.5%. It is doubtful that this could be achieved in practice
due to a variety of factors. In the case of the N00 solvent, pure water,
Table 4.1. Projected Performance of Conductivity Cell
Analyzers
Operational Parameters
Cell Voltage
Cell Constant
Absorber Efficiency
Base Current
Sensitivity
Fraction of base
current at 5 ppb
Pollutant
NO 2
1 volt
2.5 x l(f2
0.05
1.37 nA
0.022 nA/pphm
0.008
so2
1 volt
2.5 x i(f2
0.95
100 nA
0 . 8 nA/pphm
0.004
scrupulous care would be required to prevent ionic impurities in the solvent
supply system. Although an increase in the base current due to an increase
in solvent conductivity could be zeroed out with each new batch of solvent,
it would place greater demands on the current resolution of the signal
processing circuitry. In addition, the 2% change in conductance per degree
Centigrade would require sophisticated compensation circuitry. Also, there
would be errors due to changes in temperature of the sampled air, as discussed
previously.
50
-------
3 --
W
-------
More serious problems are created by the occurrence of other pollutant
species which are capable of producing ions, many of which have equivalent
or larger conductance effects. Considering all of these sources of error,
it is doubtful that the conductance cell could meet the zero drift and span
stability requirements in Table 1.1.
One possible design of the potential source and current amplifier
is shown in Figure 4.6. This is based on a versatile, multi-range, fast
response conductivity amplifier, described in Reference 39, designed to
respond to conductivity changes occurring in times ranging from milliseconds
to several hours.
4.1.4 Logistical Factors
The conductivity cell described can be made very compact and can
be operated at very low power levels. However, there is a need for a supply
of absorbing solution. At a flow rate of 1 m£/min continuous flow, about
1.5 H of solution would be required for each cell. In addition a sampling
pump capable of pulling ambient air through each cell at a minimum rate of
100 mA/min would be necessary. These have been described in Section 2.0 of
this report.
4.1.5 Dose Rate Information
A continuous electrical signal, proportional to the conductance cell
current and, therefore, to the ambient air concentration of pollutant would
be available for periodic sampling; digitizing and storage techniques are
described in Section 5.0.
4.2 Halogen/Halide Redox Amperometry
This instrumental method is based on the reduction of molecular
halogen (X«) at one platinum bright electrode and the oxidation of halide
to trihalide at a second platinum bright electrode. An electrical circuit
for measurement consists of the two electrodes immersed in alkalai halide
solutions of different concentrations contained in separate cathode and
amide chambers; an ion bridge connection which allows ionic conduction,
but inhibits diffusive mixing of the two different electrolytic concentra-
tions; and an external connection between the electrodes through a current
monitor to measure the current generated by the reaction of pollutant
species with electrolytic species. This current is dependent on the time
rate of supply of pollutant (input molar flow rate).
52
-------
77777 Seeling ampiifir
Figure 4.6.
Schematic Diagram of Conductance Amplifier Circuit.J9 For
Clarity the Standard Power Supply and Compensation Element
Connections to the Operational Amplifiers Have Been Omitted.
53
-------
Another variation of this method is to use the same electrolyte
concentration of alkalai halide at both electrodes and to provide an
external potential source of about 150 millivolts to drive the reaction
by polarizing the electrodes. Trihalide ions produced by oxidant
reaction with alkali halide reacts with hydrogen. Current flows in the
external circuit to maintain the hydrogen concentration at equilibrium.
4.2.1 Prior Use
Because of its simplicity the KI solution with polarizing potential
method has been widely applied in field instruments to measure ozone.
40
It is subject to many interferences and must be calibrated dynamically.
The relationship of moles of triiodide produced per mole of ozone reacted
depends upon the pH of the KI solution. Usually this solution is buffered
to nearly neutral pH (6.8-7.0). The weight percent of KI used varies,
41 40
ranging from 1% to as high as 20%. At a given KI concentration the
amount of triiodide in solution, produced by a given input of ozone, can
42
vary with time. High concentrations give a maximum ion production
immediately, whereas lower concentrations may require some time to achieve
a maximum. Electrolyte evaporation can be a problem unless means are
taken to replenish from a reservoir.
In addition to ozone, NO,, and S0~ can be detected. When N0? is
absorbed into the KI electrolyte, it reacts in the same manner as ozone.
Hence it can act either as a positive interferent, or if ozone has been
eliminated from the input sample stream by suitable filters, the N0~ can
be monitored. For the KI concentrations and neutral pH solution normally
employed, however, the absorption efficiency of the electrolyte for N0~
is very poor, just a few percent. When S0~ is absorbed into the KI
electrolyte, it acts in an opposite manner to ozone or N02. The SO- is
converted to sulfate ion and in the process produces electrons to reduce
I9 to iodide.
43
Kromhyr has described a cell of the type shown schematically in
Figure 4.7. For ozone the concentration in ambient air is related to
current, i , and the sample air flow rate, Q., by
m A
im = 3.87 x 10"10 (-^) X(03) (4.14)
where P is the sampling atmospheric pressure in millibars, Q. the sample
54
-------
air flow rate in ml/min, T the air temperature in °K and x(O^) the ozone
concentration in pphm. At 5 ppb ozone this corresponds to 65 nanoamperes.
This cell exhibits a background current of about 200 nanoamperes when
air with ozone removed is bubbled through it. In order to obtain a
quantitative reading, this background must be subtracted out. If the
background is not constant in time, the fluctuations may be interpreted as
a signal unless they are dynamically corrected.
44
One method for doing this is to use two cathodes, one of which is
exposed to air from which the ozone has been removed, and one which is
exposed to air in which the ozone remains. A schematic diagram of this
instrument is shown in Figure 4.8. A filter of Cr90~ and orthophosphoric
acid on an inert substrate is used to remove interfering reducing agents
from the inlets of both cathodes. For the reference side, aluminum foil
maintained at 150°C was used to destroy ozone. Sample air is passed
through a humidifier to minimize electrolyte evaporation. Iodine
evaporation is apparently no problem either. The performance of this
instrument, as reported in Reference 44, is summarized in Table 4.2.
Field tests show excellent (0.999) correlation with chemiluminescent
ozone measurements. Operation without electrolyte and reducing filter
renewal for better than 1 year were demonstrated in field tests.
4.2.2 Model
The design features of the differential cell of Lindqvist appear
to be the most suitable for this approach to a dosimeter. This reference
reported that a study of the effect of flow rate, Q , and cathode length,
L, on current, i , generated by the oxidation effect of ozone or the halide
electrolyte showed that above 8 cm length the current is directly proportioned
to the flow rate. A 10 cm cathode length can be accommodated in a briefcase
sized sensor package.
Assuming KI electrolyte, ozone in the sample air produces iodine by
the reaction
2KI + 03 + H20 -»• 2 KOH + I2 + 02 . (4.15)
This iodine is reduced at the cathode by
I2 + 2e 2 2I~ . (4.16)
55
-------
Sample Air Inlet
Platinum
Gauze
Cathode
PTFE
Platinum
Gauze
Anode
Activated
Charcoal
Asbestos Fiber Ion Bridge
Figure 4.7. Kromhyr Cell.
Measuring
ance Cell
obs
1/2 f
obs
Figure 4.8. Lindqvist Cell.
56
-------
Table 4.2. Performance Summary for Differential
Amperometric Method for Ozone Measurement
Flow rates
Span
Span Drift
Zero Drift
Noise Equivalent
Time Constant
Interferences
150 ml/min per cell
3
0-400 u-g/m (0-200 ppb) ozone
5.5% in 22 hrs (possibly due to ozone generator
drift)
3
2.3 yg/m (1.15 ppb) in 62 hours
6 y/m3 (3 ppb) in several weeks
0.6 Jig/m (0.3 ppb)
40 seconds
1. None for concentrations up to of
SO = 4350 yg/nf* (1.5 ppm)
C2H4
= 2660 yg/m° (1.7 ppm)
3
3300 yg/mJ (2.8 ppm)
1-butene = 8800 yg/m (3.8 ppm)
2. For NO. about 0.08 mole equivalent up to
2520 yg/m3 (1.3 ppm)
3. For C19 about 0.01 mole equivalent up to
2500 yg/m3 (0.85 ppm).
4. For PAN about 0.12 mole equivalent up to
1240 yg/m3 (0.24 ppm).
57
-------
A complementary reaction occurs at the anode, which is made of activated
carbon paste, where iodide ions are converted either to triode or to
molecular iodine, depending upon the iodide concentration. Assuming
that the latter is true, the pertinent equation is the inverse of equation
(4.16) above. It should be noted, however, that the concentrations (or
more properly, the activities) of I and !„ can be very much different
around the cathode and around the anode. Therefore, the Nernst equation
for the potential difference will be
m o» (-r~ f,\
E - -0.0295 (^5.) log I?j^y SjiL-^L (4.17)
z. a v.i , A^
where the a's denote activities for the species noted at the anode (A)
and cathodes (C) . If the external resistances, R, shown in the circuit
diagram of Figure 4.8, are not too large, the current flow in the
external circuits between the respective cathodes and the anodes will
tend to equalize the forward and backward rates of equation (4.16),
giving currents which are quantitative for the oxidation due to ozone
(and other oxidizing species).
The necessity for using the resistors to develop a differential
voltage signal is eliminated if a differential current measuring circuit,
using standard electrometer design techniques, is employed. A schematic
diagram of such an arrangement is shown in Figure 4.9. The electrometer
amplifiers, A, and A«, provide essentially zero resistance for the
respective cell currents. Output voltages from these amplifiers are
algebraically summed to provide a net output voltage signal, C , which
is proportional to the pollutant concentration and sample air flow rates.
If the latter are held constant, the signal will be proportional to pollutant
concentration only.
Assuming total cancellation of the "background" current component
by the reference cell, the net current from the differential cell is
2(03) = 3.87 x 10~10 (PQA/T)X(03) . (4.18)
58
-------
rWVn
Cell
Figure 4.9. Signal Processor for Differential Current Measurements.
59
-------
Note that this formula assumes total conversion of the incoming ozone,
which assumption is verified experimentally.
Since NO- acts as an interferent, KI oxidizes according to the
equation
2KI + 2H+ -»• NO + I + H0 + 2K+ . (4.19)
The electrode reactions of (4.16) will occur. The formula (4.18) for the
current due to ozone is changed to
i(N02) = 3.87 x 10~10 n(N02)(PQA/T)X(N02) (4.20)
where the concentration is in pphm of N0_ and where the factor n(NO~)
represents the efficiency of conversion of N02 to oxidant in the electrolyte.
It should be noted that with dissolved oxygen in the electrolyte, NO can be
cyclically converted back to N02, necessitating dynamic calibration of
the system.
By inverting the order of the filters A and B in Figure 4.8 to
eliminate ozone from both sample streams and then reducing agents — assumed
to be primarily S0~ — from the reference stream, a monitor for S0~ can be
obtained. The reactions are
S02 + H20 -»• H2S03 (4.21)
4H+ + 2e (4.22)
followed by reaction (4.16). Each molecule of S0« is capable of producing
one molecule of I2, reducing the background current in the sample cell.
This produces a negative current (with respect to the polarity obtained
for the ozone reaction) of magnitude
i(S02) = 3.87 x 10~10 n(S02)(PQ^/T)X(S02) (4.23)
where the concentrations of S02 is in pphm and the factor n(S02) represents
the efficiency of conversion of S02 to a reducing agent in the solution.
60
-------
4.2.3 Projected Performance
The minimum detectable concentrations for these differential cells
will depend upon the noise levels in the cells themselves. Using the
figure reported by Lindqvist of 4 nanoamperes of background cell noise
current, and requiring a noise ratio of 2, the minimum detectable
pollutant concentrations are as shown in Table 4.3 along with the sensi-
tivity figures. Both rely on the assumed absorption efficiencies shown
in the table and assume that the sampled air is at atmospheric pressure
and 25°C.
Table 4.3. Projected Performance of Differential Amperometric
Cell Analyzers
Operating Parameter
Absorber Efficiency
Sample Flow Rate
Sensitivity at 25°C
and 1 atm total pressure
Minimum Detectable
Concentration
Pollutant
°3
1.0
3
100 cm /min
131 nA/pphm
0 . 6 ppb
N02
0.0?
100 cm /min
6 . 6 nA/pphm
12 ppb
so2
0.95
3
100 cm /min
125 nA/pphm
0.7 ppb
The calibration will change as the atmospheric temperature and pressure
change, but presumably these parameters could be also monitored and compen-
sated for in the signal processor. The sampling rate will also have to be
maintained constant. Since the current signals are directly proportional
to the sample flow rate, any variation in this will cause a corresponding
sized error in the dose rate reading.
44
The time response reported by Lindqvist , about 40 sec, could be
expected for this method.
61
-------
4.2.4 Logistical Factors
Liquid pumping is not required. Makeup water for the electrolyte
would be required, however. A sampling pump would be required for
each cell. If all three pollutants were measured with individual dual
cells, the required pumping of sample air would be 600 ml/min, which
could be handled by most of the pumps listed in Table 2.1. As with the
conductivity cell reviewed in Section 4.1, care would have to be taken
to keep the instrument package vertical in order to avoid leaking and
plugging problems.
4.2.5 Dose Rate Information
A continuous electrical signal, proportional to the ambient air
concentration of pollutant (when compensated for pressure and temperature
changes) would be available for periodic sampling, digitizing and storage
by techniques described in Section 5.0.
4.3 Halogen Generation Amperometry
Another method of amperometric measurement is to automatically
titrate a solution by generating halogen to restore an equilibrium
concentration in an absorbing electrolyte. Alternatively, halogen can
be generated at a constant rate and deviation from halogen/halide
equilibrium concentrations due to pollutant species can be used to
measure the pollutant concentration.
4.3.1 Prior Use
Several different instrumental methods using halogen generation have
been reported. Among these are the following:
1. Absorb pollutant into an electrolyte containing KBr and Br^.
Change in the Br~ concentration develops a potential difference
according to the Nernst equation.^ This potential is used
to drive generating electrodes until the depleted species is
replenished. This generating current is proportional to the
rate at which the pollutant is supplied to the electrolytes,
which—at constant air sampling rates—is proportional to the
ambient air concentration of pollutant. "
2. A constant current source normally oxidizes halide at one
platinum electrode and reduces halogen at a second electrode
in an alkalai halide/halogen electrolyte solution. A third
electrode acts as a sensor which normally does not carry
any current until pollutant is absorbed into the solution
to change the halide/halogen equilibrium point. Then the
sensing electrode will conduct a current proportional to the
concentration deviation and, therefore, proportional to the
rate at which pollutant is supplied to the solution. '
62
-------
3. Alkalai halide reagent is supplied at a constant rate to a
reservoir where a pair of electrodes generates halogen at a
constant rate. This halide/halogen electrolyte is mixed with
sampled air and allowed to collect in another reservoir where
a pair of measuring electrodes offset in potential by a
reference potential source in the range of several tens of
millivolts magnitude. Reaction of pollutant with either the
halogen or halide change the relative concentrations, causing
a change in solution conductance. The unbalance generates
current proportional to pollutant concentration, which can
then be measured. ^8 The alkalai halide reagent is filtered
to remove halogen and is recycled.
Because of its relative simplicity, the third method would probably be
easier to implement in a reduced si2e model.
4.3.2 Model
The model cell is shown in Figure 4.10. Alkalai halide, for example
NI, reagent is metered at a constant flow rate past a pair of platinum
electrodes, G and G«, where a constant current source, I generates iodine
•^ G
through the anode reaction
2I~ + I2 + 2e (4.24)
The solution concentration of I~ depends upon the liquid flow rate Q and
the current magnitude. From Faraday's Law the molar concentration of I-
is related to Q0 and I by
x> (j
[I2] - 0.31 IG/Q£ (4>25)
where I_ is specified in amperes and Q in ml/min.
Cjr X.
Following I» generation the liquid stream is mixed with the sample
air stream to absorb pollutant. From the analysis outlined in Section 4.1.2,
the concentration of pollutant in solution will be given by
C - 1.20 x 10~10 nXPQQT (4.26)
where n is the absorption efficiency of the solution for the particular
pollutant and x is tne concentration in pphm of pollutant in ambient air
sampled at pressure P in millibars and temperature T. The sample air rate
is Q. ml/min and the liquid flow rate is Q^ ml/min.
63
-------
From Liquid
Reagent Supply
From Sampling
Pump
Constant
Current
Generator, I
Coaxial
Generating
Electrodes
I
Pt Sensing
Electrodes
To
Exhaust
To Measuring Circuit
Figure 4.10.
Model Cell for Halogen Generation/Amperometric Measurement
Method.
64
-------
It should be noted that evaporation of the electrolyte will occur
in the mixing process. For example, for a sample air flow rate of 100
ml/min at 20% relative humidity, 25 °C and at a nominal 1 atmosphere
(1013 millibars) contacted with aqueous solution flowing at 1 ml/min,
the water loss can be as high as 4% per pass. If the air sample is
"bone dry" and at 35°C, the evaporation loss is almost 10% per pass.
One method of controlling this is to humidify the sample air. This
can lead to more serious problems, however, when sampling for SO,, since
the loss of S09 on surfaces due to absorption from the gas phase is
49
significantly increased with increasing relative humidity. This
has previously been discussed in connection with the analysis method of
Section 4.2.
Designating the fraction of liquid volume lost by evaporation as
e, the concentration of pollutant and of iodine will both be increased
by a fraction (1-e)
Assuming perfect mixing and reaction rates high enough to totally
react the absorbed pollutant, the reactions will lead either to an
increase in I2 concentration on a 1:1 mole basis for N02 and Oo or
to a decrease on a 1:1 mole basis for S02- Therefore, the change in
1-2 concentration will be
6[I2] = 1.2 x io~10 nxPQA/Q£T (1_e) (4.27)
and the fractional change will be
6[I]
-10
-y= 3.87 x 10 xu nxPQA/IGT (4.28)
Therefore, the fractional change in iodine concentration is independent
of the concentration changes due to evaporation loss of water.
Schulze has reported a linear relationship between conductance
and iodine concentration for a constant value of bias voltage, E in
D
Figure 4.10. The particular conductance levels will depend on the
65
-------
Q. = 100 ml/min
A
IG = 20yA
geometry of the platinum electrodes as well as the bias voltage and
iodine concentration. Typically, the current levels involved are on
the order of 1 vA full scale.
In order to assess the magnitude of the fractional changes in
conductance predicted by (4.28), the following parameter values are
assumed.
n = 1.0
X - 100 pphm
P = 1013 millibar (1 atm) T = 300°K
This gives a change of 65%. A concentration of 5 ppb corresponds to
a 0.325% change in conductance. For 1 yA full scale current, the minimum
detectable current change required is 3.25 nA.
For N02 the absorption efficiency of neutral KI solution is very
low. At 5% efficiency the current change corresponding to 5 ppb N02
in ambient air would be about 0.16 nA. Measurements of currents at this
level require careful circuit design.
As pointed out by Schulze, it is desirable to have steady, non-
turbulent flow of electrolyte past the sensing electrodes, S.^ and S2
in Figure 4.10. The mixing chamber should not introduce bubbles in
the electrolyte, and the liquid flow system should be designed to
eliminate plugging or any condition that would lead to slug flow.
4.3.3 Projected Performance
Based upon the model presented above, the projected performance of
this method is shown in Table 4.4 below. No estimate of minimum detection
table noise is given because there is no data on which to base a calculation.
Table 4.4 Projected Performance of Iodine Generator Amperemeter
Operating Parameter
Absorber efficiency
Sample air flow rate
Sensitivity at 25°C
and 1 atm total pressure
Minimum Detectable
Concentration
°3
1.0
100 cm3 /min
6 . 5 nA/pphm
7
Pollutant
N02
0.05
3
100 cm /min
0.325 nA/pphm
7
so2
0.95
3
100 cm /min
6 nA/pphm
7
66
-------
As is obvious from equation (4.28), the sampling air flow rate must
be held constant. Any variation will cause a corresponding variation in
signal (at constant x)• The total pressure and temperature of the air
will also affect the calibration and should be accounted for if the
results are to be interpreted in volumetric concentration terms.
The time response will depend on the flow rate of liquid and the
hold up volume in the measuring cell. If the latter is 0.25 milli-
liter and the liquid flow rate is 1 ml/min the 90% response time will
be about 35 to 40 seconds.
4.3.4 Logistical Factors
Liquid pumping as well as air pumping is required. Although the
reagent is recycled, makeup water is required to take care of evaporation
losses. Three separate cells, with appropriate filters to take out S02
and other reducing gases and ozone, are required. This brings total air
pumping capacity to 300 ml/min, which can be easily handled by most of the
pumps listed in Table 2.1.
4.3.5 Dose Rate Information
A continuous electrical signal from each cell, proportional to
either pollutant mass concentration, or to pollutant volume concentration
when temperature and pressure are compensated for, would be available
for periodic sampling, digitizing and storage by techniques discussed in
Section 5.0.
4.4 Membrane Amperemeters
This method depends upon the diffusion of the species to be measured
through a membrane chosen to selectively permeate this pollutant while
discriminating against other, potentially competitive species as much
as possible. Following the membrane diffusion, the species goes through
a redox process at a sensing electrode to generate a current which is
related to the gas phase concentration of the pollutant on the atmospheric
side of the membrane. The membrane is used to minimize electrolyte
evaporation as well as to provide some degree of selectivity.
67
-------
4.4.1 Prior Use
Patents ' for electrochemical transducers for S02 and N02 have
been issued. Instruments based on these transducers are commercially
52
available. Table 4.5 shows operating parameter ranges for ambient
air monitoring instruments and Table 4.6 shows manufacturer's data on
53
interference equivalents of 1 ppm of the indicated species.
Some of the operating advantages and disadvantages have been
53
pointed out in a survey report by LBL. These are listed in tabular
form in Table 4.7. The temperature sensitivity must be either compen-
sated for electronically or the device must be kept at a constant
temperature.
Table 4.5. Features of ECT Analyzers
For ambient air monitoring:
Sensitivity
Highest Range
Lag Time
Response Time
Class
Cost
Typical
0.01
5
10
Portable
2200
Range
0.005 to 0.02 ppm
1 to 10 ppm
to sec
20 to 180 sec
950 to 2300 $
Table 4.6. Model SS-310 Interference Data
Interferent
Response to 1 ppm
NO
N02
Aldehydes
Hydrocarbons
CO
C00
< 0.001
< 0.01
< 0.001
< 0.0001
< 0.0001
< 10
,-8
68
-------
Table 4.7 Advantages and Disadvantages of ECT Analyzers
53
Advantages
Disadvantages
1. Fast response
2. Low maintainance
3. Small, light weight
4. Low power consumption
5. Continuous reagent supply
not necessary
1. Periodic electrolyte re-
placement required due to
volatilization
2. Membrane can be separated
from thin film electrolyte
by simple pump suction.
3. Accumulation of adhering
material in sample air-
stream may eventually block
membrane
4. Temperature sensitive
4.4.2 Model
The model used for analysis is shown schematically in Figure 4.11.
A membrane of thickness dm and area A is exposed to air containing the
pollutant to be sensed—e.g., N02 with a gas phase partial pressure P .
The membrane will have a permeability, P, for this species which is dependent
upon its solubility, S ,in the membrane material and its diffusion co-
efficient, D , in the membrane material according to
P = S D
m m
(4.29)
At the back side of the membrane the diffused gas will have a gas phase
partial pres
be given by
partial pressure p^. The flow rate of pollutant through the membrane will
m
S D A
m
(4.30)
As long as the concentration is small, the relationship between p , at the
back face, and the concentration of pollutant at the surface of the thin
69
-------
Membrane, Permeability
P, Area A, Thickness dm
g
m
Sensing Electrode
Electrolyte
Counterelectrode
1
Figure 4.11. Model Membrane Amperemeter.
electrolyte layer between the membrane and the sensing electrode, c, will
be given by
(4.31)
The diffusive flux of pollutant through the electrolyte of thickness d_.
and diffusion coefficient D., is given by
i.
J = -D c/d
(4.32)
if it is assumed that the concentration of pollutant at the electrode
surface is zero. That is, the pollutant species is consumed (transformed
into another species) by the redox process.
It should be noted that equations (4.29)-(4.32) are steady state
equations. For transient analysis, Pick's second law equation must be
applied.
70
-------
Equating the flow rate through the membrane with the diffusive
flow rate through the electrolyte (also of area A) leads to the following
expression for flow rate of pollutant to the electrode:
PSP/d
\
. moles ., „.
p« ~~ ( }
where K is a proportionality constant included to convert the expression
to a molar flow rate. The current generated at the sensing electrode is
T ///moles, ,// equivalent, A\ Coulombs N
I = ( - ) x ( - a - ) x ( - : - = - )
sec sec equivalent
or
I = nFQ . (4.34)
P
The solubility and diffusion parameters of the membrane and electrolyte
will be dependent upon temperature as
P - P exp (-E /RT) -- - (4.35)
o p sec j
cm*- • atm
D = DQexp (-ED/RT) cm2/sec (4.36)
S = S exp (-AH /RT) cm3(STP)/cm3 • atm (4.37)
o s
where EL is the activation energy for diffusion, AH is the heat of solution
for the pollutant species in the medium, R is the gas constant and, for a
given medium
Ep ' ED + AHs ' (4'38)
Values of these parameters are shown in Table 4.8 for the case of N0»
54 ^
permeation in Teflon. Table 4.9 shows the relative permeation of several
53
different membrane materials for SQj, NO, and N02- These data give some
idea of the mass flow rate magnitudes obtainable with membranes.
71
-------
Table 4.8 Parameter Values for Permeation of NO- in Teflon at 25°C
P = 1.57 x 10
D = 3.7 x 10~9
S = 0.43
-9
E
10.67 kJ/mole
E = 58.6 kJ/mole
P
AH = 47.7 kJ/mole
s
Rate of change of permeability with temperature, about 1.4%/°K at 300°K
Table 4.9 Relative Permeation of NO, NO,, and S0_ in Various Materials
53
Membranes
Teflon, 1/4 mil
Polyethylene, 1.2 mils
Polyethylene, 0.3 mils
Polyvinylchloride , 1.5 mils
Pellicon, 0.5 mil
Zitex 12-137B, 2 mils
Cellophane, 1 mil
Polyethylene and Pelicon
Double Membrane, 1.5 mils
Permeation Ratios
so2
0.83
1.00
1.00
1.00
0.94
1.00
1.00
1.00
NO
0.73
0.08
0.08
0.0
0.0
0.30
0.0
0.0
N02
1.0
0.03
0.0
0.0
1.0
0.69
0.0
0.0
The electrolyte used will determine the values of solubility, SF,
and diffusion constant , n. The patents ' on this basic method used IN
£
sulfuric acid as the example electrolyte for both N09 and SO. models and
^ £
claimed the use of aqueous acid electrolytes. This obviously places N0? at
72
-------
a 'disadvantage because of its low solubility in acid solution relative to
S0«. Other electrolytes have been investigated and found more sensitive
55
and selective for NOj-
The type electrolyte used will determine the bias potential, E in
Figure 4.11, which is used to promote the devised reaction while suppressing
competing reactions which occur at higher potentials. However, reactions
which occur at lower potentials can still interfere. The cited patents '
discuss the factors involved in the choice of counter electrode materials.
For a quantitative estimate of the current obtainable from this
transducer, it will be assumed that in (4.33) the membrane and solution per-
meation factors are such that
VE/dE ^ P/dm *
Then
-13 PA
Qp * 4.46 x 10 (~)X. moles/sec (4.39)
where X is the pollutant concentration in pphm. With typical values for N0~
/
in teflon, with a 5 cm diameter teflon membrane, 1/4 mil thick (6 x 10 cm),
and a reaction that gives 2 equivalents per mole, the current is related to
concentration by
I = 8.5 x 10~ x amperes (4.40)
This can be improved somewhat by increasing the area of the membrane or
by finding a membrane material of higher permeability. Either course
will probably also increase the evaporative loss of electrolyte by "reverse"
permeation through the membrane. It appears that a practically sized sensor
will have a sensitivity of the order of 0.1 nA/pphm.
The method has no counterpart for ozone.
4.4.3 Projected Performance
Based upon the discussion in the previous section, the projected per-
formance of these sensors for NOj and S02 is as shown in Table 4.10 below.
73
-------
Table 4.10 Projected Performance of Membrane
Amperemeters.
Operating Parameters
Membrane Area
Membrane Thickness
Sensitivity
Minimum Detectable Concentra-
tions
Values
20 cm2
-4
6 x 10 cm
0.1 nA/pphm
10 ppb
4.4.4 Logistical Factors
In contrast to the previous methods discussed, this transducer
does not require forced air flow although it can be used. Because of
the sensitivity to temperature of permeation rates through the membrane
and electrolyte, the temperature of the device should be controlled.
4.4.5 Dose Rate Information
A continuous electrical signal proportional to pollutant concen-
tration in ambient air is available for periodic sampling, digitizing,
and storage by techniques discussed in Section 5.0.
4.5 Colorimetric Methods
These methods of detection and measurement are based upon the
reaction of the pollutant species with a regent dye to give a char-
acteristic color. It is convenient to divide these methods into two
categories, namely, surface detection and volume detection. In surface
detection, the dye is coated on a solid substrate such as filter paper
or silica gel and the concentration of the pollutant is determined from
the intensity of the characteristic color reflected when the surface
is illuminated with white light. In volume detection a decolorized dye
is converted to a colored species and the amount of light absorbed at
the characteristic color is measured.
74
-------
4.5.1 Prior Use
Many methods have been reported which use surface colorimetric
reactions for S0~ detection and measurement. . Crude applications have
been in the form of indicator papers. Litmus paper has been used as has
congo red paper. Filter paper coated with starch-potassium iodate-potassium
iodide-glycerol, sodium nitroprusside/sodium carbonate, p-phenyl anediamine,
zinc nitroprusside/ammonium sulfite, among other reagents, have been used
to indicate the presence of sulfur dioxide. Generally, these indicator
paper methods have required a relatively large concentration, relative to
ambient air standards, in order to develop a color detectable by the human
eye.
The primary application of surface colorimetric reactions has been in
the area of "detector tubes" designed for occupational environmental moni-
toring. The Draeger, Kitagawa and MSA commercial series include stain
indicator tubes for sulfur dioxide. The use, characteristics and limitations
57 58
of stain indicators has been reviewed by Saltzman and Linch. The support
material for the indicator dye is usually silica gel. Various inorganic
and organic color producing species are used to coat the substrate. Many
of these have been reviewed in Reference 56. As with paper indicators the
usual read-out of these indicator tubes is by the human eye, observing the
length of stained section colored by drawing in a well defined volume of
sample air. However, at least one technique has been devised in which a
59
photocell is used to compare the developed colors with a standard set
Volumetric detection is usually accomplished as a two step process.
The first step requires absorption from the gas phase into a suitable
liquid, and conversion to a sulfite or sulfate ion. Then this solution
is reacted with a suitable reagent to develop a colored species in the
mixtures for analysis via photometric absorption.
Various methods have been used to contact the sampled air stream
with absorbing (or "scrubbing") solution. The simplest is to flow the
air stream across the surface and rely upon diffusion through the boundary
75
-------
layer from the gas phase to the liquid phase. Alternatively, the air
stream may be directed toward the liquid surface (or "impinged"). The
collection efficiency of these methods is highly dependent upon flow rate,
temperature and solution chemistry.
Various studies have been made of the collection efficiency of
bubbler systems to contact a sample air stream with absorbing solution.
3
Sulfur dioxide at 25 to 700 yg/m in air at a sampling rate of 1.4 Jl/min
has been collected at 98.5 to 99.2% efficiency in 50 mi of 1% H202 solution.
At lower concentrations collection efficiency drops to about 95% . Another
study of bubbler efficiency at flow rates of 1.2-6£/min with S09 concen-
3
trations up to 17 pphm (440 yg/m ) indicated collection efficiency of
better than 97% in 1% H?02> tetrachloromercurate solution and water at
pH 5. Efficiency was found to increase with decreasing flow rate and
increasing S09 concentration. One of the problems with bubblers is the
increased chance of NO- interference due to physically entrained air bubbles
in the absorbing solution.
62
A more complicated method has been devised by Lyshkow, who used
a countercurrent of sample air stream with a scrubbing solution stream
and enhanced contact by using rotating discs. This method decreases
physical entrainment of bubbles. A countercurrent absorption method has
also been used to absorb S0_ directly into a color producing reagent—a
63
starch-iodine aqueous solution—in a portable S09 monitor.
A wide variety of color producing chemical reagents has been studied
for sulfur dioxide measurement. Many of these have been reviewed in
Reference 56. The most widely used is the standard West-Gaeke method
modified to eliminate NO and ozone interferences. The procedure, now the
X
Federal Standard Reference method for determining atmospheric S0_, is
described in detail in Reference 64.
Several other dyes have been used or proposed for use in colorimetric
S0» detection. These have been reviewed and compared with pararosaniline
in a recent survey of monitoring methods.
76
-------
One of the major problems with colorimetric reactions is the length
of time required to develop a color. This is due to several factors.
The absorption and mixing step both require some time, and if a sequence
of reactions is used, the mixing and equilibration time of each adds to
the total analysis time. The kinetics of the reactions depend upon the
concentrations. At lower concentrations of reactant more time is required
to reach the end point on color development. The response time can range
from a few minutes to a half-hour.
Many methods have been reported which use surface colorimetric
reactions for N0« detection and measurement. Filter paper has been
impregnated with sulfanilic acid, ot-naphthyl amine, alcoholic diphenyl-
amine, p-anisidine, starch-iodide solution, and acidic pyramidone among
other substances. As with most colorimetric reagents, stability is a
problem and shelf life is very limited. Blue litmus paper has also been
used for NCL detection as has cellophane coated with various dye mixtures.
Silica gel has been coated with diphenylamine/NaCS, solution, 0-tolidine
chloride, p-aminobenzoic acid/nickelous chloride solution, antipyrine,
dimethylaniline and sulfanilic acid/N-(l-naphthyl) ethylene diamine
dihydrogen chloride solutions, among others, to obtain response to NO 2-
As with the filter paper coatings stability and shelf life are usually
limited. Also there are generally many interfering substances which
will also develop the characteristic color of the indicator dye species.
Reagent dye solutions used for N0~ detection by volumetric methods
have included acidified diphenylamine, sulfanilic acid/a-naphthylamine/
acetic acid, phenoldisulfonic acid, indigosol in dilute sulfuric acid
solution, and NEDA/sulfanilamide with phosphoric or tartaric acid.
Detection of ozone through surface colorimetric reactions has been
ported which use filter paper coated with tetramethyl-p-diamidodiphenyl-
amine, benzidine, Kl/starch, thallnium hydroxide, ferric oxide/potassium
thiocyanate, and 4,4'-tetramethyldiainino diphenylamine. Others have used
silica gel coated with liminol/NaOH, fluorescein/KOH, and fuchsin. Many
of these systems are also responsive to other oxidants.
77
-------
Volumetric colorlmetric detection reagents for bZone have included
m-phenylenediamine hydrochloride, fluorescein, indigo/sulfuric acid and
sodium diphenylamine sulfonate/perchloric acid. One of the most widely
used colorimetric reagents for ozone is aqueous potassium iodide. The
Federal Reference method for calibration of ozone instruments uses 1%
(by weight) KI buffered to a pH of 6.8 and assumes a unity conversion
fifi
ratio of ozone to the triiodide ion colored species. This has been disputed.
Higher weight percentages of KI have been used, but there is evidence that
69
the effect of N02 interference increases with increasing KI concentration.
To counter this, however, there is evidence that higher concentrations of
neutral buffered KI give higher absorbances than lower concentration solutions
in response to the same 0» concentration, particularly at low 0~ concentrations.
In addition, the amount of KI used apparently affects the time required to
develop maximum color at 352 ran.
From the point of view of portable instruments for the field the
surface detection reagents are limited by their stability, i.e., by becoming
de-colored or discolored "on the shelf", and the volumetric reagents present
a logistics problem of continual supply and waste removal, as well as sta-
bility problems in many cases. Another problem is that the response times
are relatively long. For "surface" reaction reagents, this is due to the
necessity for the pollutant to penetrate by diffusion into the reagent to
develop color to a sufficient depth to become detectable. For volumetric
reactions there is the necessity to contact the sample stream with the
reagent, following which diffusion must also occur to allow the reaction
to take place in over a volume that will lead to a detectable quantity of
colored species. In the cases where the absorbing reagent and the color-
producing reagent are different, there is also an additional time penalty
for mixing of the two.
4.5.2 Model Detectors
Several possible model volumetric detectors ri:e shown in Figure
4.12. It is assumed that mixing of the air stream has occurred prior
to entry into the analysis cell of each combination. The scheme of
Figure 4.12A could be termed the conventional approach wherein linear
78
-------
Light
Baffle
From Reagent
Supply
PCI
PC2
Air In 1
(A
Contactor °
o
o
Air Out ^
\[
L
1
U
i —
r— I
L
J
Wfft
m
Wk
PCI
From Reagent —
Supply
^ Filter
: LS
D Filter
To
Contactor —
PC2 B
%
<
^^i^^
%
X
\
c=
/
^T» i
H
'
/
Ir
^T
Tr» I»Taof~o
1 From
— — Contactor
Conical
"~->— Reflector
fe^
_ Diffuser Plate
LS
Figure 4.12. Model Colorimeter Detectors.
79
-------
cells are used with a sequential liquid reagent feed in order to eliminate
a pump for this loop. For increased sensitivity the cell length L is
made as long as practicable, since according to the Beer-Lambert
relation
I/I0TL = 10 CCL (4.41)
where I is the light intensity at the photodetector, I the light
intensity at the source, -e the molar absorption coefficient, C the molar
concentration of absorbing species, and L is the path length. The factor
TT is the transmission factor for the liquid in the absence of colored
Li
species developed by the pollutant. The practical limit is the maximum
allowable vertical dimension of the container for the instrument. For
a briefcase sized package, a practical cell length would be about 10 cm
at the most. With a molar absorptivity of the developed colored species
4 -1
of about 10 (M-cm) the intensity ratio is related to the concentration
by
C = -10~5log10(I/I0TL)mole/liter . (4.42)
The detection limit is set by the minimum detectable intensity ratio
change. This is determined by the ability of the signal processor to
discriminate a change which is sufficiently larger—say by a factor of
two—that the normal fluctuations due to photocell detector, amplifier
"noise", light source fluctuations, refractive index fluctuations in the
liquid and liquid absorbance fluctuations. If it is assumed that the dual
chamber/photocell arrangement effectively cancels out light source fluctu-
ations and that the sample air/colorimetric reagent liquid mixing is
thoroughly done to eliminate absorbance fluctuations but in a manner that
does not generate scintillations due to bubbles, the sensitivity limit
80
-------
is set by the system electrical noise. A practical value for this is
o
about 1 part in 10 . This corresponds to a concentration change of
_q
about 8 x 10 mole/liter.
Using an analysis similar to that given is Section 4.1.3 to
determine the amount of material transferred from the sampled air
stream to the liquid, the concentration of colored species will be
given by
—10 97^
C = 4.46 x 10 (4p)Bn X Qa/Q£ moles/liter (4.43)
where T is the air temperature, g is the efficiency for converting one
molecule of pollutant species to one molecule of colored species, n is
the absorption efficiency of the liquid for the pollutant species, x is
the volume ratio of pollutant species in the sampled air (in pphm) , 0 the
3.
air sampling flow rate, and Q^ the reagent liquid flow rate.
Assuming the following values for these parameters
T = 300°K Q = 100 cm3/min
a 3
3 = 0.1 Q = 1 cm /min
&
n = o.i
the minimum detectable air concentration is about 0.2 ppm. In order to
achieve the desired detection limit of 0.005 ppm, both the conversion and
absorption efficiency would have to be increased until their product is
40 times the value obtained from those assumed above. Another possibility
is that a colored species could be found which would have a higher molar
4
extinction coefficient than the value of 10 assumed for this calculation.
The arrangement sketched in Figure 4.12B could use longer cells,
but not more than a factor of two. A multipass arrangement such as the
internal reflection scheme sketched in Figure 4.12C could be used to
obtain longer effective path lengths. The ratio of path length per pass
to cell length is sec 0, where 6 is the angle of incidence and reflection.
The number of passes, N, is determined by this angle and by the width
dimension of the cell, W, from
N = (W/2L')cot 6 . (4.44.)
81
-------
For a specular reflectance, R, assumed constant over both reflecting
surfaces the light intensity at the exit slit, I, is related to that
at the entrance slit, I-., by
= RN i(feCNL'sece . (4.45)
As long as the reflectance remains constant in a time period long
compared to the measurement time, the effect of reflectance changes
can be taken into account by occasional instrument "zero" adjustments.
The major trade-off for the multipass arrangement is in response
time and loss of time resolution versus increased sensitivity for the
dose rate measurement. The larger volume requires more time to fill.
For practical logistical reasons the reagent flow rate cannot be too
large. One mfc/min is equivalent to 1.440 liters (about 1.5 quarts), which
quantity must not only be stored in supply, but also in waste. Assuming
an entrance angle of 2.5° and 12 passes, the width required for the cell
is about 11 cm if the cell length, L1, is assumed to be 10 cm. For a
3
cell thickness of 0.5 cm, the volume is 55 cm . At a flow rate of
o
1 cm /min, almost three hours would be required to achieve 95% response
to a step change in input concentration. This can be compared a single
pass cell, used in the methods shown in Figures A and B, with the same
length, 10 cm, and with a diameter of 0.5 cm. The volume for this cell
3
is only about 2 cm . At the same flow rate of liquid, 95% response
to a step change would require only about 6 minutes. The sensitivity
gained due to increased path length is about a factor of 12, versus a
factor of 27.5 increase in response time.
Some of the response time could be regained by using a zig-zag
chamber with reflecting surfaces at the "elbows". This method would reduce
the volume and hence the response time, but only by the same factor as
the increase in path length.
One possible model measuring device using surface colorimetric
reactions is represented in Figure 4J.3. The method depicted in Figure 4.13
uses a transparent film for transporting the reagent past a nozzle
82
-------
M
To Tape
i| Takeup
Tape Supply
LS
To Signal Processor
Filter
PC
Figure 4.13. Transparent Film Substrate Colorimetric Detector.
83
-------
where sample air is impinged intermittently to provide alternating exposed,
unexposed strips. Because of reagent stability problems upon exposure to
air, and/or the necessity to provide moisture to enhance colorimetric
reactions, it may be necessary to supply either reagent or water to the
tape immediately before exposure to the sampled air. The applicator,
designated M in the schematic diagram of Figure 4.13, is for this purpose.
Spacing a^ is provided between this nozzle and the light source (LS)/
detector (PC) plane so that the tape, moving with linear velocity u,
will have sufficient time to allow color development. This width of the
exposed strip will depend upon the "on" time of the air stream, t , and
s
the tape velocity.
When the exposed strip is positioned between the light source (LS)
and the photocell (PC), the development of color due to reaction between
the reagent and the pollutant will attenuate the light in a given
frequency band. A band pass filter can be used to reduce the incoming
light to only those wavelengths near maximum absorption. Ic may (or may
not) be necessary to add a heater in the separation space a. in order to
speed up the color development. The absorbing species will be distributed
in the reagent layer, thickness y_ with an integrated absorbance of
A = ec(y)dy = ecy . (4.46)
o
In addition to this there will be absorbance due to the supporting film
of thickness y2» characterized by an absorption coefficient a~. The
reagent film which is not exposed to the pollutant will have an absorption
coefficient a.. . The filter, if used, will have a peak transmission, T^,
1 r
at the analyzing wavelength. Therefore in the absence of coloration the
light intensity measured by the photocell, I , is related to the incident
intensity, I , by
T- = logio Tf -[Vi + V21 - (4'47)
o
84
-------
When all of the factors except the colored species concentration are
constant , the ratio of intensities is
(4.48)
Spatial variations of thickness or absorption coefficient of
either the reagent or the support film can occur. From equation (4.47),
the intensity fluctuation will depend on these variations as
(4.49)
where the 6( ) quantities represent small deviations from nominal
values. If the length (in the direction of transport) of these is
enough to spread them over at least one measurement-background cycle,
they will be balanced out. If they are shorter, however, they can
introduce error either through addition to or subtraction from the
net absorption signal. Generally, the reagent depth, y , will be
small compared to tape thickness, y». Also, if the applied reagent
has uniformly wetted the tape surface there should be relatively few
short term fluctuations in the reagent absorption coefficient. This
leaves only fluctuations in tape thickness and absorption coefficient
as possible major sources of error.
A more complex system for analysis is the model shown in Figure 4.14,
The applied reagent — whether applied immediately before measurement
by dispenser M or applied long before during tape manufacture or
preparation prior to loading — is coated on the surface of the fibers
that make up the tape. As shown in the sketch, a light beam coming from
the light source will, more often than not, have to follow a tortuous
path through the tape, undergoing multiple reflections from the fiber
surfaces. Each reflection is through a thin layer of reagent coated
onto the surface. If the reagent has been applied by M there may be
an appreciable amount of liquid filling the interstices, even though
sample air has been drawn through the tape to allow pollutant to develop
the color. The multiple reflections greatly increase the path length
and give much higher sensitivity than the transparent tape scheme
previously discussed.
85
-------
Sample Air In
1
To Vacuum
Pump
PC
-)
To Signal Processor
*
To
* Tape
Takeup
Reel
Light
Source
LS
Band Pass Filter
Photo
Cell
PC
Figure 4.14. Tape System Using Fibrous Tape.
86
-------
A third method of measurement using surface colorimetric reactions
is sketched in Figure 4.15. The multiple reflectance scheme discussed
above also can lead to a large amount of backscattered radiation. This
diffuse reflectance has been long used as an analytical technique. One
particularly sensitive method is to use plane polarized radiation and
select fiber or grain diameters in the diffuse reflecting surface to
optimize the absorption at a particular wavelength. This has been discussed
by Kortum and an example is shown in Figure 4.16 to illustrate. In addition
to polarization and particle or fiber size, the angle of observation, 0, for
the detecting photocell, both measured with respect to normal incidence,
can be selected to enhance the effect of absorption. This is illustrated
in Figure 4.17, also taken from Kortum, where the reflectance as a function
of 0 is shown for various a values at two different wavelengths, one in the
absorption range (550 run) and one outside the absorption range (650 nm) for
rhodamin dye coated on a paper substrate. For example, at a = 60°, 6 = 50°
the fraction of reflected linearly polarized light has increased from 0.25
to 0.75 due to the absorption, a factor of 3, whereas for a = 0° and 0 = 50°,
there has been a factor of 12 increase, although the intensity is lower
than with the large angle of incidence.
A quantitative estimate based on a model of the diffuse reflectance
mode of operation is difficult to obtain. For weakly absorbing powders
the Kubelka
the formula
the Kubelka-Munk theory relates diffuse reflectance to concentration by
c = S(l-r)2/2er (4.50)
where c is the concentration (M) of absorbing species, S(cm ) a scattering
coefficient, e the molar absorptivity (M-cm) , and r is the ratio of
diffuse reflectance with absorber to that without absorber. The value of
S for very fine particles (1 to 4 ym) for the visible region is about 100
to 400 cm . Assuming that a 1% change in reflectance is detectable, and
4
that the colored species has an e value of 10 , the minimum detectable
concentration change is, for S = 200 cm , 1 x 10 M. If these particles
are silica with a specific surface of 1 M per gram, packed with a porosity
of 0,5, and the material converted to colored species has a depth of 0.1 ym
87
-------
Polarizers
Sample Air In
M
00
00
N
To Vacuum Pump
Figure 4.15. Measurement Using Surface Reflectance Change Due to Pollutant Exposure.
-------
16
14
12
10
H08
06
04
02
0
-02
l-2u
(a)
100-150M
10
14
60-75u
-50y
18
22
16
14
12
10
08
06
04
02
0
-02
1-
10
14
(b)
100-150y
60-75y
42-50y
18
22
3 -1
Wave number (10 cm )
Figure 4.16.
Remission Spectra of Pure CuSO, • 5H 0 at Five Different
Crystal Sizes, Measured Against the White Standard of
Zeiss; a) Taken with Polarized Radiation; b) Taken with
Natural Radiation.
89
-------
10
08
06
04
02
X = 550 my
cc=80
10C
0 20° 40° 60° 80°
e
Figure 4.17.
Fraction R of Linearly Polarized Light After the Reflection
on Paper Dye with Rhodamin, for Various Angles of Incidence
a as Parameter and Dependent on the Observation Angle 0.
Lower Part: X = 650 my; Upper Part: X = 550 my (in the
Absorption Range).
90
-------
on the surface of each particle, distributed on particles in a volume
defined by a 5 mm diameter spot to a depth of 10 ym, the quantity of
-13
colored material will be about 2 x 10 moles.
If it is assumed that the efficiency for converting pollutant to
colored species is 1% and the quantity adsorbed represents that obtained
o
from 100 cm of sample (i.e., a sample air flow rate of 100 mfc/min for
1 minute), the minimum detectable change corresponds to a pollutant
concentration in ambient air of about 5 ppb.
This optimistic estimate depends upon the existence of a reagent
which will develop colored species with a molar absorptivity of at least
10 (M-cm) and a reasonable value of the particle scattering coefficient
S. If e is smaller or S larger (or both), there will be corresponding
factors of increase of the minimum detectable concentration of pollutant.
4.5.3 Projected Performance
Using a volumetric absorption technique similar to those analyzed
in the preceding section, except with considerably shorter cells, Lyshkow
72
has claimed a sensitivity of 5 ppb for both NO,, and S0~. The detectors
used were photoresistive. The Lyshkow method used reagent dyes which
developed higher molar absorbance than the 10 (M-cm) assumed in the
foregoing analysis. The absorbing solutions also had a higher value
than the 10% efficiency assumed for the model calculation.
Of the three surface absorption techniques reviewed the use of
reflected polarized light appears to be the most sensitive. If stable
dyes can be obtained, the method should be capable of reaching the
5 ppb detection limit desired.
Sample and zero stability will be difficult to maintain in the dual
sample cell-dual photodetector volumetric method because of differences
in the individual photodetectors. An alternate arrangement is to use a
single photodetector with a mechanical chopper to alternate transmitted
light from the "signal cell" and the "reference cell" onto the detector.
This increases design complexity but has an additional advantage of
enabling the use of synchronous a-c signal processing techniques which
can improve both stability and sensitivity.
91
-------
Such a system is already inherent in the reflective surface, adsorption
technique. Alternation of exposed-unexposed (to sample intake air) surfaces
provides a dynamic reference. However, the tape transport will be relatively
slow, so that for all pratical purposes the photodetector signal will be d-c
amplified unless some mechanical chopping arrangement is used to enable use
of synchronous a-c detection.
For both methods the dynamic range is limited by the background
coloration of the reagent dyes due to factors other than exposure to the
particular pollutant to be detected.
4.5.4 Logistical Factors
The volumetric method requires a liquid supply of about 1.5 liter of
reagent solution per pollutant per 24 hours and an equivalent amount of
waste disposal volume. Sample air pumping requirements are minimal, easily
handled by most of the pumps listed in Table 2.1. Although gravity feed
could be used, it would probably be better to also pump the dye solutions.
Signal processing electronics could be easily accommodated on a battery
supply.
4.5.5 Dose Rate Information
After analog conversion from non-linear forms inherent in light
transmission measurements, a continuous electrical signal proportional
to pollutant concentration in ambient air is available for periodic
sampling, digitizing, and storage by techniques discussed in Section 5.0.
4.6 Gas Phase Luminescence
Gas phase luminescence methods rely on the creation of an excited
state (non-equilibrium energy state) of the pollutant or one of its
reaction products. Spontaneous emission of a photon occurs when the
excited molecule relaxes back to an equilibrium energy state. The wave-
length distribution is characteristic of the molecule (specificity) and
the emitted intensity is proportional to the pollutant concentration in
the gas phase. Creation of the excited state can occur through chemical
reaction (chemiluminescence), direct excitation by light (fluorescence/
phosphorescence), or indirect excitation. In the latter case light excites
a metastable state of some intermediate gas the excited gas molecules of
which are capable of energy transfer upon collision with a pollutant molecule.
Following this collisional transfer, the pollutant molecule can emit radi-
ation via fluorescence.
92
-------
4.6.1 Prior Use
Detection of ozone by chemiluminescence from gas phase reactions has
been based primarily on the ozone-ethylene reaction first used by
"7 O 7 /
Nederbragt and then improved by Warren and Babcock. The method, used
in conjunction with the wet chemical KI verification of the ozone output
from a calibration ozonator. Hodgeson, et al have determined the spectral
characteristics of the reaction and compared the performance characteristics
of this instrumental technique to those of instruments based on the ozone-
NO gas phase reaction and the ozone-Rhodamine B heterogeneous reaction.
The ozone-NO reaction has been studied in detail by a number of
workers. Clyne and Thrush and their coworkers have reported on the
78
spectral characteristics and kinetics of the reaction. Fontijn has
been issued a patent on one configuration that uses the reaction for ozone
detection, which is essentially the instrument compared by Hodgeson et al
to the ozone-ethylene instrument. Turning the reaction around and using
ozone as a reagent, the method of Fontijn has been developed as an NO
79
detector and with the addition of an appropriate catalytic reduction
80
surface to change N0« to NO the instrument has been modified to some
extent and developed for field use by several commercial firms. It has
been reported that commercial NO monitors which rely on reduction of N02
to NO are non-specific but also convert PAN and other volatile organic
81
nitrates to NO.
Both the ozone-ethylene instrument and the 0.,-NO instrument require
a constant supply of gas reagent. In addition, the effluent from the
reaction, excess ethylene or excess NO along with N0«, must be scrubbed
to prevent the production of pollutants.
In addition to the use of the gas phase chemiluminescent reaction
between NO and ozone to detect NO and NO , and ozone-ethylene to detect
X
ozone, the reaction between atomic oxygen and NO or SO has been used by
QO QO
Wooten et al to detect NO,, and S00. Another patent claims the use of
X £.
the NO + 0 reaction to detect many nitrogen containing compounds.
93
-------
The kinetics, energetics and spectral emission of reactions
0 + NO -»• N02 -»- NO + hv
0 + SO -*• S02 -»• S02 + hv
have been studied and reported in a long series of papers by Thrush and
his co-workers. The 0 + NO reaction has been proposed as a standard
87
for determining rate constants for other chemiluminescent reactions.
The method used by Thrush et al to produce these species was to break up
S0_ and N02 into SO, NO and atomic oxygen in a microwave discharge cavity
and then study the recombination emission downstream of the cavity (after-
go
glow) . The method of Wooten et al is to generate atomic oxygen separately
with a pulsed (1-10 pps) electrical discharge. They use a reduced pressure
of 0.5 to 5 torr in the reaction chamber. In this range there are no third
QQ
body quenching effects in contrast to the NO + 0_ and SO + 0., reactions.
Sulfur dioxide has been detected down to a level of about 10 ppb in
89
air by a fluorescent method. Although the method has moved from the
laboratory prototype stage to commercial instruments, it is not yet as
well established as the ozone, NO chemiluminescence methods. Water
' x
vapor quenches the fluorescence, probably both in conventional gas phase
collisional deactivation and by reactions of SO- with adsorbed water
vapor on vessel walls. Collisions with many other types of gas molecules
can also quench the fluorescence. These quenching effects greatly reduce
the excitation efficiency making the emitted light intensity very weak.
The sophisticated detection circuitry required for such low light is not
readily reduced to miniature size. Table 4.11 taken from Reference 88
indicates the quenching effect of various vapor phase species normally
* *
found in ambient air on fluorescent emission by N0~ and S02«
4.6.2 Model
The reaction of SO and NO with ozone give emission intensities of90'91
[A] [o3]
K2 + Zm Ksm
94
(4.51)
-------
where [A] represents the concentration of either NO or SO and [M] is the
concentration of third body species which contribute to quenching through
collision with excited AO species. Since chemiluminescence is intimately
related to the fluorescence properties of the AO species, some idea of
the effect of oxygen and nitrogen on quenching of cheiluminescence is
gained by consideration of their quenching effect on fluorescence of N02
and S02. This has been studied and reported by Metee. The kinetic
studies of References 73 and 74 used argon as the third body. The data
of Table 4.11 indicates that the quenching probability per collision is not
drastically different for either N02 or S02 in their collisions with Ar,
0 or N2. Thus, the rate constants obtained in References 73 and 74 should
be representative when air is the background gas rather than Ar or Ar/02-
When pressures in the order of a torr or larger are used, Zm K3m [M] » K2
and the emission intensity is given by
I = Io [A][03]/[M] . (4-52)
This gives the emission intensity in einsteins per second per liter when
the concentrations are specified in moles/liter.
88
Table 4.11 SO- Emission Quenching Data
Collision
Partner
N2
°2
co2
A
Ne
Collision
Diameter
(nm)
0.386
0.369
0.494
0.360
0.254
Polar izability
(io-3V)
1.76
1.60
2.65
1.63
0.392
Probability for Fluorescence
Quenching per collision
S00 N00
0.28
0.27
0.63
0.21
0.14
t.
0.10
0.13
0.27
0.08
95
-------
The effect of pressure and flow rate on emissions intensity has been
92
studied by Steffenson and Steadman for two different mixing arrangements
79
and reactor configurations. The Fontijn spherical geometry reactor has
93
been analyzed by Harrison and Spangler. Both analyses lead to an equation
for the emission intensity of the form.
[0 ] [NO] Q
^ff-^o LMj ^TR (4'53)
where £ = optical collection efficiency
[NO] = initial NO concentration injected into the reactor
Q. = gas flow rate
A
V = reactor volume
r
TR = lifetime of NO species due to reactions with ozone
in the reaction volume.
Since the hydrodynamic time constant is V /Q (assuming perfect mixing),
it can be seen that the emission intensity is related to the ratio of the
reaction lifetime to the hydrodynamic residence time in the reaction
chamber. The collection efficiency of the integrating sphere of Fontijn1 s
geometry is on the order of 20% whereas the more cylindrical geometries
used in the evaluation of Reference 92 would be expected to be somewhat
lower. Most commercial instruments are designed to give as high detection
(PM cathode) area to volume ratio as possible in order to approximate 2 TT
collection geometry.
As shown in Reference 93, the filter spectral response, photodetector
spectral response and emission spectral characteristic can be combined to
determine an average detector current per unit photon rate, which will
be symbolized as G. This factor will include the photomultiplier gain
factor if a PM tube is used. The signal current due to emission throughout
the reaction volume V will be, when T_ is approximated by (k, [0-])~ ,
K K 1J
where k^ is the rate constant for non- radiative reactions of NO and 0«,
-8
8- (4.54)
96
-------
A similar expression obtains for [SO]. Depending upon the type detector
used and the type of signal processing (e.g., direct amplification or
chopped light with a.c. amplification), the minimum detectable signal level
will be set by dark current fluctuations or shot noise of the signal.
Comparison of the intensity factor, I , for the NO/0., and SO/0.,
reactions indicates a factor of about 4 higher intensity for the latter.
This is counterbalanced by the fact that k.. for the SO/0- reaction is
about 5 times larger than that of the NO/Oo reaction. However, the emitted
73
light for the SO/0- reaction is in the 300-450 nm wavelength region where
photomultiplier cathode efficiencies are much higher than in X > 610 nm
region for the NO/0- reaction. These factors indicate that if some prac-
tical method is found to reduce SO,, to SO in an analogous manner to that
used for the NO- -»• NO reduction, a gas phase chemiluminescent detector
based on SO/0- would be practicable.
The ozone-ethylene reaction is well known and has already been
incorporated into portable instruments capable of monitoring ozone down
to the ppb level.
Because of the low light levels available at low concentrations,
a photomultiplier would have to be used. Presently available commercial
silicon photodiode detectors have a sensitivity of about 0.2 Ampere/watt
of radiant power in the 400 nm region. The sensitivity improves in the
*
red and near infrared region, where the NO- emission occurs, to 0.3 to
0.5 A/watt. This should be compared to the sensitivity of typically
15,000 A/watt available from a photomultiplier in the 400 nm region. Even
though electronic amplification could be used to boost the dynamic signal
from a silicon photodiode, careful design would have to be employed in a
dark current suppression circuit to achieve proper temperature tracking.
4.6.3 Projected Performance
Presently available portable instruments for NO and ozone monitoring
X,
have the specifications shown in Table 4.12. Similar specifications could
probably be obtained in a composite instrument package.
97
-------
Table 4.12 Performance Values for Chemiluminescent Monitors
Pollutant
Parameter
Span
Detection Time
Zero Drift
Span Drift
Sample Flow
Response Time
°3
100 pphm
0 . 1 pphm
15,/min
20 seconds
NO
X
100 pphm
1 pphm
< 1 pphm/ day
< 1 pphm/ day
IJl/minute
1.1 minute
4.6.4 Logistical Factors
The data of Table 4.12 indicates that both instruments require relatively
high flow rates. Equation (4.54) shows that the detector signal current
should be directly proportional to the volumetric flow rate, Q., through
the instrument. The air sampling pumps used must be capable of maintaining
these flow rates at a constant value in order to minimize signal variations
that could be construed as pollutant concentration variations. Several
of the pumps listed in Table 2.1 are capable of 2 to 3 liters per minute,
but at relatively low pressure drops. Stabilizing the sample air flows
would be a problem. The high-flow rate pumps also use more power
and limit the battery operation time to about 8 hours at the most. For
the ozone monitor a source of ethylene is required. This can be supplied
from a lecture bottle sized tank, but the continuous operating time will
be limited to about 8 hours.
4.6.5 Dose Rate Information
A continuous electrical signal proportional to pollutant concentra-
tors in ambient air is available for periodic sampling, digitizing, and
storage by techniques discussed in Section 5.0.
98
-------
4.7 Gas/Liquid Phase Luminescence
This method is based upon chemical reactions in a liquid medium
containing a chemical reagent which is capable of emitting light when
it reacts with species contained in an air stream which is contacted with
the liquid.
4.7.1 Prior Use
The chemiluminescent reaction between ozone and a large number of
94
organic dyes has been reported by Bowman and Alexander. These reactions
consume the organic species, producing reaction products which are
generally incapable of any further useful emission. Life of the dye can
be
by
95
be improved by using excitation transfer reactions in a scheme symbolized
* * *
+ D -> C + D ; D -»• D + hv,
where A is the species to be detected (pollutant), B is an intermediate
reacting species capable of combining with A to produce C in an excited
state which can transfer its excess energy to a suitable chosen dye species
D, creating an excited molecule which is capable of relaxing back to the
ground state via emission of photon hv. This method was used by Bersis
and Vassilious to obtain chemiluminescent emission from Rhodamine B
dye in a solution also containing gallic acid. The gallic acid was
oxidized by ozone to create an excited species wliicli tranferred energy
to the Rhodamine B, which then emitted light.
Another approach is to use the pollutant species to catalyze a
chemiluminescent reaction. Such a principle has been used to develop
an instrument which monitors 0.,, N09 and S09 via a catalyzed H909/luminol
97
(5-amine-2,3-dihydro-l,4-phthalazine dione) reaction. Detection limits
in the range of 1-3 ppb are claimed for the method. Selected filters
are used to absorb out various combinations of pollutants which are
routed to different "microreactors" where they are mixed with the H202/
luminol reagent to produce chemiluminescence. All of these "microreactors"
are closely spaced in front of a two inch diameter PM tube. A rotating
99
-------
shelter sequentially exposes each reactor to allow the chemiluminescence
from each pollutant to be monitored. The stored signals are compared to
a zero air signal and stored calibration signals to obtain a quantitative
readout.
98
A recent Russian patent also uses luminol, modified by a copper-
amine catalyst, flow through a porous disc to provide a continually
fresh surface for exposure to ozone to obtain chemiluminescence.
4.7.2 Model
The miniature micror.eaction cells used in Reference 97 for the
pollutant catalyzed I^C^/luminol chemiluminescent reaction employ very
low reactant flow rates, ranging from 0.1 to 0.5 ml/min and high air
flow rates, typically 2 8,/min. The cells are 0.635 cm (0.25 inch) I.D.,
resulting in a high air velocity which forces the liquid reagent into
a thin film moving along the walls of the reactor tube. Light is emitted
in the thin sheath of the liquid along the wall.
The geometry could be improved to enhance light collection and
reduce the flow rate requirements. The use of a small amount of deter-
gent to enhance uniform wetting of a glass surface by the liquid reagents
would enable a flat surfaced cell with a mirror backing to be used in
order to obtain as large a viewing angle as possible for the photodetector.
In addition such mirroring could be used to isolate adjacent reaction cells
to prevent "crosstalk" from the light generated by the filter-isolated
pollutant/air streams flowing through each cell.
For a quantitative estimate of performance the mechanisms of the
light generation reactions must be known in detail. From the data presented
in Reference 97, it appears that for the reagent concentrations used—
0.25 mg/ml luminol in 0.05N NaOH and 0.6% H20>—the reactions are first
order with respect to pollutant concentration. As shown in previous
sections, the pollutant concentration, C, in an absorbing solution is
related to the gas phase concentration, x (i-n pphm), by
C = 4.46 x 10~10 nQ X/Q0 mole/liter (4.55)
A J6
where n is the absorption efficiency of the solution for the pollutant.
100
-------
If K is the rate efficiency for converting a pollutant species absorbed
into solution into a photon emission, the emission rate will be
I = 4.45 x 10~ KnQAX/Q0 einsteins/liter-sec. (4.56)
A. Jo
The detailed form for K will depend upon the rates of reaction which
cause or compete with (quenching) the emission.
The ratio of responses per pphm of pollutant for the reactant con-
centrations given above and for a Q. of 2000 ml/min and Q of about
1 ml/min are
(I/X)n : (I/X)™ :
-------
Of the parameters involved in equation (4.58) the emission rate ic
is the one most difficult to quantitate. An estimate of the order of
magnitude can be obtained from the data given in Reference 99. The EMI
9558 photodetector in that application is operated at about 1.5 x 10 A/watt
at 430 run. The response to ozone at 1 pphm, 17.9 v (across an anode
load resistor of unspecified size) ratioed to the 0.5 v noise level, taken
—8 —11
to correspond to 10 A, gives an estimated light power rate of 2.4 x 10
watt.
Assuming the following parameter values:
5 = 0.25 n = 1.0 Q = 1 ml/min
-4
V = 1.2 x 10 £ Q = 2000 ml/min
M\ A
gives a value for < of 2 x 10 photons/sec at 300°K.
In order to see whether or not a simpler photodetector can be used,
i.e., a silicon photodiode rather than a photomultiplier, the estimated
light power at 430 nm for 1 pphm of ozone, 2.4 x 10 watts can be used
with a typical sensitivity figure for a commercial silicon photodiode of
99
0.2 Amperes/watt at 430 nm. The photodiode would generate about 5
picoamps. This is well below the nanoampere dark current level typical
in these devices. For a 1 Hertz bandwidth of the signal processing sys-
tem, the shot noise corresponding to this dark current is calculated from
iN = /2eIDAf (4.59)
where e is the magnitude of the electronic charge, I the dark current
-14
level and Af the bandwidth, as 1.8 x 10 amperes. This is about 2
orders of magnitude less than the signal current. The dark current is
very sensitive to temperature, doubling with every 10°C change. It could
probably be zeroed out at a given temperature, but sophisticated circuit
design would be required to make the zero track the dark current as
temperature varied. A signal current level of 5 picoamperes (for ozone,
for which the luminescence is highest) would require a carefully designed
electrometer system, which would be difficult to keep operating properly,
even in a laboratory environment.
Therefore, it appears that a photomultiplier must be used. Several
commercial domestic and foreign firms make 1.27 cm or 1.9 cm diameter
miniature PM tubes that would have sufficient sensitivity for this measure-
ment scheme.
102
-------
The patented method for the H-Oj/luminol measurement technique uses
a 5 cm diameter tube with a rotating shutter which sequentially exposes
microreactors for viewing through 1.27 cm apertures. Another possible
arrangement is to have a single reactor cell facing a single 1.27 cm
diameter end-on PM tube and sequentially valving in a sample air stream
which has been sequentially diverted through appropriate filters. Another
possible arrangement is to use a rotating lens/prism system to sequentially
reflect light from separate reactor cells to a commonly used PM tube.
The third alternative is to use separate PM tubes. Not only would this
be expensive in terms of power supply energy, weight and initial cost,
but the different aging characteristics of different tubes would probably
require frequent recalibration.
4.7.3 Projected Performance
The H-02/luminol method appears to be promising in terms of sensitivity,
detection limits and dynamic range. Assuming that a scaled down version
can approach the performance of the patent prototype, the desired detection
limit will be about 10 ppb. A dynamic range of 400 can be expected. Response
time will depend upon the configuration used, but will probably be about
1 minute if a sequential sampling method is used.
4.7.4 Logistical Factors
Miniature photomultipliers are now being routinely used in portable
equipment. Power demands are moderate, about 1.0 watt, which can easily
be supplied from battery sources. Operating times of 10 to 12 hours
on battery power should be possible. Reagent reservoirs of about li
each would be required for the luminol solution and the H202 solution.
An equal amount of waste storage volume would be required for the spent
reagent. Both solutions are apparently stable for a few days if stored
under low light level exposure. Battery power would also be required
for the air sampling pump and the liquid reagent pumps. An additional
power demand would be imposed if a thermoelectric coder were used to
stabilize the PM tube.
4.7.5 Dose Rate Information
Following PM tube signal processing, the integrated dose signal
per pollutant, per cycle would be available for digitizing and storage
by techniques discussed in Section 5.0.
103
-------
4.8 Gas/Solid Luminescence
In this method the gas phase pollutant reacts with a reagent layer
coated on a solid substrate to produce a chemiluminescent emission which
can be measured by a photodetector and related to pollutant concentration.
4.8.1 Prior Use
In addition to experiments with chemiluminescence of liquid phase
94
organic dyes exposed to ozone, Bowman and Alexander also used dyes
deposited on high area, dry subtrates, such as silica gel. Somewhat
earlier Regener had reported the use of Rhodamine B on silica gel for
atmospheric ozone measurements.
Rhodamine B on silica gel has been used for the measurement of ozone
in ambient air in an instrument prototype developed by RTI for EPA.
The operating characteristics of this instrument have been evaluated
102
and described in detail by Hodgeson et al. These authors have reported
the use of a silicone resin to coat the silica gel substrate in order
to minimize the adsorption of water vapor, which can have a large effect
on both the time response and sensitivity of the chemiluminescent
emission. Another problem, the gradual reduction with time of emission
intensity at constant ozone input, is compensated in the instrument design
of Reference 101 by periodic zero and calibration steps included with a
measurement step in the operational cycle.
103
Guicherit has reported that the use of gallic acid mixed with
Rhodamine B prior to deposition on a silica gel surface precoated with a
hydrophobic material produces very stable emission over a long period of
time. As with the instrument described in Reference 101, Guicherit's
instrument requires periodic exposure to ozone-free air. The sample-zero
cycle time is relatively short, one minute, with 10 to 15 seconds allotted
to sample measurement. The ozone-free air is prehumidified to provide
a moisture content comparable to the sampled air. This prevents the tran-
sient response variation reported by Hodgeson, et al, who cycled dry
zero air versus moist sample air. Because of the stability of the surface
no internal calibration step was required, reducing both cycle time and
instrument complexity. The instrument response was reported to be linear
104
-------
3
over the range 20-2000 yg/m (0.01 to 1 ppm ozone). Normal ambient levels
of NO, N0« and S0« were reported to have no effect. Testing up to 2 ppm
with these as well as H_S, C12, NH~, HC1, HF and H20>2 produced no
interference. However, at much higher concentrations, up to 100 ppm,
N02> Cl~ and S0~ were found to attack the chemiluminescent material.
In field evaluations the calibration of the instrument remained stable
over several weeks of continuous monitoring and showed close agreement
with an electrochemical (galvanic) method.
The use of gallic acid with Rhodamine B by Guicherit in a dry form
was based on a report of the same mixture used in solution form by Bersis
96
and Vassilious. It is one example of a general class of excitation
transfer reactions which have been discussed above in Section 4.7.1.
One of the problems of chemiluminescent methods of detection is to
get adequate mixing of the reagents. For example, the instrument of
Reference 101 uses a spiral pattern of airflow across the RhB/silica
gel disc to establish contact. With this arrangement at constant 0»
concentration the light output is proportional to flow rate for low
flows (less than 200 mJl/min) but tends to saturate at higher flows,
indicating that an increasing proportion of material is going through
102
unreacted. The same phenomena is reported in Reference 103.
104
Lyshkow has recently obtained a patent on a device for producing tur-
bulence in the analysis gas stream by close spacing in the reaction volume
exposed to the detector PM tube. He claims that this is suitable for
both gas phase, e.g., 0«-ethylene, or gas-solid, e.g., NO^-sulfanilimide,
or 0--A1, chemiluminescent reactions.
4.8.2 Model
Miniaturization of an ozone analyzer appears to be a straightforward
application of existing design such as that of Guicherit. Extensive
research would be required to develop methods of extending the techniques
to N02 and S02-
4.8.3 Projected Performance
A dynamic range of 0.01 to 1 ppm can probably be achieved with the
reduced version. The 1 minute cycle time required for period zero air/
sample air cycling limits the response time.
105
-------
4.8.4 Logistical Factors
A sampling pump, an inverter for obtaining high voltage from the PM
tube, and the signal processing electronics would all require battery
power. Switching streams in the alternate zero air/sample air rhythm
would require battery power also.
4.8.5 Dose Rate Information
A digital signal representing the ambient air ozone concentration
would be available once a minute for electronic storage by techniques
discussed in Section 5.0.
4.9 lonization and Electrophoresis
The ionization detector and electron capture detector are
examples of devices which ionize gaseous species and then drift them in
an electric field to obtain a current which can be related to the concen-
tration of the ionized species. The main problem with these simple
devices is their lack of specificity. Many substances can be ionized
by electron extraction or proton extraction, or proton capture or
electron capture. Substances can be ranked in their ability to parti-
^ i. t. - . 107,108
cxpate in such charge exchange or ion-molecule reactions.
However, some separation method such as a gas chromatographic column
must generally be employed to supply only a known species to such a
detector in a given time interval in order to develop a quantitative
signal. Another method is to use the differences in mobility (velocity
per unit electric field) between different ions to separate out a desired
species.
4.9.1 Prior Use
Electrophoretic techniques, using the difference in mobility of
ionic species to separate different species in a drift field (electrical
109
field) while known to physicists for many years, have been employed
only recently for analytical purposes. The most prominent commercial
example is the Plasma Chromatograph. This instrument uses a beta
emitter, Nickel-63, to produce primary ions which are capable of trans-
ferring charge, or participating in ion-molecule reactions, with
molecular species to be analyzed. Identification of species is by means
106
-------
of characteristic drift times of different species, periodically gated
into a gaseous drift field. While a great deal of theoretical work has
been done on ion mobility, practical measurements must rely on instru-
ment calibration with known species under well defined operating conditions,
just as gas chromatography is practiced. This instrument operates at
ambient pressure, making atmospheric sampling much easier than with any
other analytical instruments.
112
Vree and Fontijn have disclosed an electrophoretic device in which
chemionization is used to produce primary ions which are then reacted with
gaseous molecular species to produce ions. These are pulled at a given
stream velocity through a transverse electrical field. A schematic diagram
is shown in Figure 4.18. This is a modification of an old technique known
109
as the "Erickson air blast method". Those ions with a mobility larger
than some limiting value , K , are collected by the electrodes before being
swept downstream. This method serves as an integral device, giving the
number of ions of mobility greater than K , in contrast to the Plasma
Chromatograph which serves as a differential device, giving the mobility
distribution.
Another variation of this approach is to use a radioactive foil to
create ions and then direct these through a series of baffles with poten-
tials maintained on adjacent electrodes of such a magnitude that only
those ions with mobilities in a narrow range, AK, about some central
value R can be deflected along the tortuous path through the baffles
without being discharged at some intermediate electrode and survive to
be collected as a signal current. Devices built on this principle
have been employed as detectors for chemical agents in ambient air by both
11 o
the U.S. Army and Air Force. J
Lovelock has obtained a British patent on the use of the electron
capture detector to quantify the product of reactions which convert S02,
NO and other pollutants (but not ozone) to electron capturing species.
107
-------
SAMPLE AND
X"CARRIERGAS
o
00
DISCHARGE
GAS
16
TO VACUUM
PUMP
GASEOUS
CATALYST
BALLAST
GAS
36
CURRENT
MEASUREMENT
44
1
CB
§
a
« *
J2 E2
J-. S
-------
4.9.2 Model
These brief considerations of ionization/electrophoresis suggest
that a compact instrument could be designed based on the operating prin-
ciples shown schematically in Figure 4.19. The sample air stream would
be diverted sequentially to selective filters FI, F_, F- to pass only
those components to be analyzed to the beta foil analyzer. A polarizing
voltage source, V., would determine the polarity synchronized with that
of V. and w.ould impose a transverse drift velocity V = K E over length L.
A J. TL
Only those ions with a mobility, K, such that their longitudinal velocity,
u, and their transverse velocity transported them to collection electrode,
C, would be registered as a current detected by an electrometer. The air
stream could be moisture conditioned to maintain a constant hydration
of the ionic species generated.
The hydration factor is important. To illustrate this, Figure 4.20 ,
taken from Reference 115, shows the distribution of hydrated protons as a
function of temperature and water vapor partial pressure, when the hydrated
ions are of the form H (H?0) . This hydration can affect the mobility
in two ways-by increasing the mass and by changing the collision cross-
section. The effect on mobility of these parameters can be seen by con-
sidering the theoretical formula for mobility
-8 / (M_ + M_)
_ 2.02 x IQ / UJp TT
Na TM^M^
_o
where N is the molecular density (m ) of the gas through which the ion
2
drifts, a is the collision cross section (m ), M_ and M.^ are molecular
masses (kg) of the pollutant and drift gases, respectively, and T is the
gas temperature. It is also obvious that a change in either gas pressure
(i.e., N) or temperature can also affect the mobility. If the method
is to be used as a measuring device compensation for changes in these
parameters must be provided.
If it is assumed that the effect of adding one H«0 molecule is to
increase the pollutant molecular diameter by one H~0 diameter and that sub-
sequent hydration does not appreciably change the collision cross-section,
109
-------
Polarity Selection Outer Electrode Insulating Rings
Electrode (Grid)/ Radius a
6 Foil
Sample Gas In
I Gas Out
N'H
V.
To Pump
<=>
•OrD
Inner Electrode,
Radius C
To Electrometer
V
B
V,
air/ ————
S02, N02, 03
I Pump
Exhaust
Electrometer
^
Signal Out j
Figure 4.19. lonization/Electrophoresis Detector.
110
-------
10"
10"' )0° 10'
WATER PRESSURE (lorr)
to2 io3
Figure 4.20 Equilibrium Distributions of Clusters H (H20)n Predicted
by Experimental
Ill
-------
the change in mobility at constant total pressure and gas temperature
will be as shown in Table 4.13 below.
Table 4.13 Effect of Hydration on Mobility of S02, N02 and 0-
Molecules
Number of water
molecules, n
1
2
3
Kn/Ko
so2
0.392
0.383
0.375
N02
0.330
0.317
0.310
°3
0.356
0.343
0.332
If this model is representative, the effect of adding the first
water molecule is drastic, reducing mobility to about one third its de-
hydrated value. Thereafter, the magnitude of effect is diminished, but
still amounts to a few percent.
Assuming the use of a cylindrical geometry, as shown in the enlarged
version of the drift tube analyzer sketched in Figure 4.19, with a thin
central electrode of radius c at potential V with respect to ground,
B
a coaxial gas inlet aperture of radius b and a coaxial outer electrode
of radius a, at ground potential ("zero"), the radial electric field
strength at a distance c > r > a is given by
Er = VB ln (?)/r '
The velocity in the radial direction will be
v =
r dt
(4.61)
(4.62)
Substitution from (4.61), separation of variables and integration between
appropriate limits gives
rdr =
At
KV_ In (a/c)dt
D
(4.63)
112
-------
which can be solved to obtain an expression for K, i.e.,
K = (a2 - Y2)/2At VB ln(a/c) . (4.64)
The lower limit of integration along radial displacement in (4.63), y »
represents the uncertainty in starting position of an injected ion. It
will start somewhere between the surface of the center electrode, r = c,
and the outside of the injection aperture, r = b.
For analysis it is desired that only those ions with mobility in a
narrow range AK about a central value K reach the collector electrode which
is located a distance L longitudinally down the drift tube from the injection
plane. The electrode width is 2w, centered on L. If the average longi-
tudinal velocity component is u, the appropriate time factor, At, in
equation (4.64) is given by L/u. Therefore
K = u(a2 - Y2)/2 L VR In (a/c) . (4.65)
Because of the variations of the parameters in equation (4.65) there will
be uncertainties in the precise value of K. This uncertainty, 6K, is
related to the magnitude of parameter variations by
«K o Su _ f 2c2 S 5L
K u 2 2
( a - c
The velocity uncertainty can occur due to variations in flow rate. The
second factor reflects the uncertainty in smarting position, as mentioned
before. The third factor indicates the finite width of the collector
electrode.
A one percent variation in flow rate will cause a corresponding one
percent uncertainity in mobility measurement. For the following assumed
dimensions:
c = 1 x 10~3m a = 2 x 10~2m w = 1 x 10~3m
b = 3 x 10~ 3m L = 1.0 m
the uncertainty due to a spread in starting positions will be 2% and the
range of mobilities collected by the finite width electrode will be one
percent.
113
-------
The amount of current produced at a given pollutant concentration will
depend upon the ionization efficiency of the pollutant. It appears that
negative ions for mobility analysis would be feasible for N02, SO- and 0.,.
The electron affinity of N02 is very large, 4 electron-volts per molecule,
and that of ozone is also large, about 2 electron-volts per molecule.
While quantitative data are not available for S00. it is known that it
£.
does participate in charge transfer from other nejative ions.
A simple model for the charge transfer process from reactant ions
_3
at density ND(cm ) to "target" species (such as N00, S00 and 0_) at an
-3 116
initial density N (cm ), which has been developed elsewhere, shows that
the number of ions formed per second is given by
ni= (AVIKR WAD (4'67)
where A is the cross-sectional area of the ion-molecule reaction region,
V is the ion drift velocity, KR the ion-molecule reaction rate, and T
the drift time in the reaction region. The pollutant concentration can be
written in terms of pphm by volume at 300°K as
NATN = 2.43 x 1011* molecules/cm3 . (4.68)
AD
Assuming only one electronic charge per molecule, and 100% current
collection efficiency, the current developed per pphm of pollutant will be
I/X = 3.9 x 10~8 (A VT ^ TR) NR . (4.69)
values of the parameters can be estimated by assuming that the ionization
region is a tube of 6 mm ID and 2 cm length through which the sample air
Q
stream flows at 100 cm /min. It will be further assumed that the longi-
tudinal drift field which selects the polarity of the ions to be analyzed
is small, on the order of 10 V/cm. The value of Y-. will be taken as the
-93
Langevin limit of 10 cm /sec. The value of reactant ion density can be
estimated from the value of current available from electron capture
—8 9 3
detector sources, about 10 amps, as around 10 cm . Using these para-
meter values in equation (4.69) gives a value for I/X as about 20 nA per
pphm. Since the saturated current value for the ion source was assumed
to be 10 nA, it is apparent that saturation effects would start to limit
current at well below 1 pphm.
114
-------
It is this extreme sensitivity that makes the method subject to
so many interferences. Any species present in the air sample which
tends to form ions of the same polarity as the target species can
interfere. If the interferent has a higher charge affinity than the
target species, or is present in sufficient concentration, the ion
current due to the target species can be severely depleted or suppressed
altogether.
4.9.3 Projected Performance
While the method can be sensitive to very small concentrations of
pollutant it is subject to many interferences and many errors unless a
more sophisticated analysis scheme is used than has been outlined above.
4.9.4 Logistical Factors
The method requires active sampling at a rate which is easily
accommodated by any of the pumps in Table 2.1. The electronic signal
processing circuits and bias polarity sources are easily available from
battery signals.
4.9.5 Dose Rate Information
A continuous electrical signal proportional to pollutant concen-
tration in ambient air is available for periodic sampling, digitizing,
and storage by techniques discussed in Section 5.0.
4.10 Piezoelectric Mass Monitor
The principle of the method is to coat a quartz crystal with a
thin layer of selectively adsorbing material. The coated crystal is
used in an electronic oscillator circuit as the frequency controlling
element. The coated crystal is then exposed to a gas stream containing
the component to be selectively adsorbed. An equilibrium point is
reached after a sufficient time interval at which a certain amount of
mass has been added to the crystal. This causes a reduction in resonant
frequency. A calibration between component partial pressure and frequency
change can be established to obtain a compact, relatively simple measuring
device.
4.10.1 Prior Use
For many years the change in frequency of a quartz crystal due to
mass loading has been used as a method for monitoring the deposition
115
-------
of thin films. King proposed that this principle be used as a
selective detector by using a selective sorbent to coat the crystal
electrodes. For quartz crystals King gave the relation for change
in frequency, AF, with mass, AW, as
AF = 2.3 x 1Q6F2 ~ (4.70)
A
where F is the original resonant frequency of the crystal (in MHz) and
A the area of the coated electrode. At a typical resonant frequency of
9 MHz a frequency change of about 500 Hz per microgram was reported.
The King sorption detection has been reported as used in many
different applications. Pertinent to the present application—to measure
118
NO,,, S0~ and ozone—Frechette and his coworkers have examined various
coatings to determine their applicability in the selective sorption of
S02' They concluded that SDM polymer (a 1:1 copolymer of styrene—
dimentylaminopropyl maleimide) was the best candidate for a selective
sorbent and proceeded to evaluate the performance of a mass monitor using
118
SDM. They reported that both water vapor and N0« could interfere. In
addition it was found that the frequency change, AF, per microgram of SO^
decreased with increasing sample size and that the response thus increased
with increasing sample size.
The longer time response with larger sample size is also not sur-
prising since it is likely that appreciable diffusion of the gas into
the sorbent must occur to allow transfer of more gas across the gas-
sorbent boundary layer.
The non-linear behavior of frequency change with increasing mass is
surprising considering equatic
integral form of this equation is
119
not surprising considering equation (4.70). It can be shown that the
B
1 + (AW)B/F
o
where B is the mass sensitivity at frequency F . If the mass loading of
the sorbent plus the adsorbed gas causes a frequency change that approaches
116
-------
2% of the base frequency, a more detailed look at the relationship between
the deposited mass and resonance characteristics of the crystal-deposit
system indicates that a period measurement is more appropriate than a
frequency measurement, and that it is possible that the acoustic pro-
120
perties of the deposit might have to be taken into account when the
mass loading is very large.
4.10.2 Model
Consideration of the use of piezoelectric mass monitors as portable
continuous detectors is attractive because they can be made very compactly,
use relatively simple circuitry, are relatively inexpensive and com-
paratively rugged. Prior to consideration of the method for continuous
monitoring, however, the limitations of the device must be thoroughly
understood. We now address these under two categories—the acoustic system
(crystal, coating gas) and the adsorbent system (coating, reactant gas).
Acoustic System
The quartz crystal orientation most frequently used in the AT cut.
121
Its orientation relative to the major crystal axes is shown in Figure
4.21. Nominally it is cut at about 35°10' in order to get a relatively
low shift of resonant frequency due to ambient temperature changes. The
effect of orientation on this temperature-frequency relation can be
122
appreciated by considering the curves in Figures 4.22-4.25. At
27°V an AT cut of 35°10' gives a relatively small Af/AT, roughly 25 Hz
for a + 10° change for a crystal resonating at 9 MHz at 27°C. The sign
of the frequency deviation will depend upon the exact orientation of
the crystal cut. For example, the data of Figure 4.22 shows that a
deviation of just 5 minutes of arc from the 35°10' figure can either
rapidly increase the Af/AT (increasing angle). Figure 4.23 shows a family
of such curves. Ideally there is one curve, corresponding to one precise
orientation, that will give zero temperature coefficient at a particular
temperature. This relation is shown in Figure 4.24. A detailed portion
of this curve for AT cuts is shown in Figure 4.25. These data indicate
the desirability of knowing the exact orientation of the crystal to be
used in order to select the operating temperature that gives negligible
temperature effect on frequency for small deviations.
\
117
-------
z1
ELECTRODE
FACE
NODAL LINE
Y'
ELECTRICAL AXIS
Figure 4.21
Orientation of AT and DT Cuts with Respect to Major Axes
of Quartz.121
118
-------
0.005
The Quartz Crystal Oscillator
Frequency constant
/x t= 1670mHz
I 1 i 1 i I
-60 -40 -20
Temp.
\
Figure 4.22 Effect of Minor Orientation Changes on Temperature Effects
in AT Cut Quartz.122
119
-------
THE QUARTZ CRYSTAL OSCILLATOR
0.006
0.004
0.002 ->
s -0.002
-0.004
-0.0061 I
0 20 40 60 80
Temperature, *C
fa)
100
Figure 4.23 Family of Frequency vs. Temperature Curves for AT Cut
Quartz.1^
250
200
150
100
50
°C 0
-50
-100
-150
-200
-250
36° 37" 38* 39" 40* 41*
Angle of rotation about z
(6)
Figure 4.24 Zero Temperature Coefficient Operating Temperatures for AT
Cut Quartz as a Function of Orientation. ^2
120
-------
Another temperature related problem is the effect of a small
123
transient temperature change on frequency. A typical response is
shown in Figure 4.26. This is thought to be due to the presence of
a thermal gradient in the crystal. This type of spurious response
could occur when the gas stream impinging on the crystal suddenly
changed temperature, and could be avoided by controlling the inlet gas
temperature.
The original theoretical justification for equation (4.70) was based
on the assumption that the added mass was "equivalent" to the quartz of
124
resonant crystal, i.e., had the same acoustic properties. More often
than not in applications of this measurement principle, it is not true.
However, formula (4.70) is very closely followed when the amount of mass
added is very small compared to the mass of the quartz crystal. A
theoretical justification, based on a perturbation analysis due originally
125
to Rayleigh, has been given by Stockbridge. This analysis is based
on mathematical arguments that do not rely on a physical model.
A treatment based on a one-dimensional model of acoustic waves
propagating from the quartz crystal into a thin deposited layer of different
I r\f
acoustic properties has provided more physical insight than the
Stockbridge treatment. This model also leads to an equation similar to
(4.70) for the case when the ratio of density-thickness products for the
quartz and the deposited layer is very much smaller than unity. This
treatment has been extended to the case of relatively thick layers, where
it has been shown that when the mass loading shifts the frequency more than
2% the acoustic parameters of the deposit must be taken into account and,
furthermore, the simple relation between frequency shift and deposited
120
mass equation (4.70), no longer holds.
Both of these two-layer models involving the acoustic properties of
the quartz and the deposit assume that the material added to the quartz
will support the shear wave generated by the AT cut quartz crystal.
Furthermore, the material is assumed to be lossless. These assumptions
are justified in the usual applications where such crystals are used to
monitor the deposit of crystalline solids. In the applications initiated
121
-------
35'10'
1
-------
by King, however, where a "selective" sorbent layer is used to coat the
crystal, these assumptions are not justified. Typically such a coating
is an amorphous, visoelastic material. Although the problem of propagation
of shear waves from a stiff medium, such a quartz, to a viscoelastic
127
medium has been previously treated theoretically, the problem of
resonance of such an elastic/viscoelastic system apparently has not been
modeled.
For ambient air monitoring, where the concentrations of pollutants
to be "selectively" sorbed are generally a ppm or less, only a very thin
coating of sorbent is required. Hence, the assumptions of Stockbridge's
perturbational analysis are valid and the linear relationship of frequency
change to added mass, given by equation (4.70), can be expected to hold.
Sorbent System
Providing the sorption is reversible, the thin coating of sorbent
deposited on the surface of the crystal can be regarded as a permeable
membrane characterized by a solubility parameter, S, and a diffusion
coefficient, D. The equation for the mass of material absorbed from the
128
gas phase into the membrane for a one dimensional system is, assuming
that Henry's law is obeyed, the pollutant gas is ideal, and the saturation
concentration has not been attained,
Am = MSA
Sp
1 + Sp/C
!n=oo
-J.E
n=0
exp
-(2n+l)2ir2Dt
(4.72)
where
M =
6 =
A =
P =
S =
Cs =
molecular weight of absorbed gas pollutant
thickness of sorbent
area of crystal covered by sorbent
partial pressure of gas pollutant in atmosphere
solubility constant
saturation concentration of gas in sorbent.
123
-------
It should be noted that the diffusion coefficient of the gas in the
sorbent, D, and also the solubility parameter, S, will be functions of
temperature. As discussed above in connection with temperature effects
on quartz crystal resonance, it is desirable to maintain the temperature
of the quartz crystal constant. Therefore S and D will be regarded as con-
stant. If the absorbed concentration is well below the saturation limit,
or
C = Sp « C (4.73)
S
the amount of mass deposited will be, for all practical purposes, linearly
related to the partial pressure in the gas phase. Also if the sorbent
layer is sufficiently thin, the quantity
T = 62/(2n+l)2ir2D (4.74)
n
will be small, and the response of a step change in concentration, AC « C ,
S
to a step change in gas partial pressure will give a change in mass which
varies with time as
AW = M6AS Ap [l - -^2 e ~t/T0]. (4.75)
IT
Therefore, for all practical purposes the time response is exponential
with a time constant given by equation (4.74) with n set equal to zero.
This treatment has assumed that the absorbed species is relatively free
to move within the sorbent. When the desorption step requires that an
energy barrier (bond breakage) be overcome the effective diffusion
constant, D, becomes much smaller than that operative for uptake of the
gas, giving a much longer time constant for desorption than for adsorption.
It might also be expected that the heat of sorption of the gas in the
sorbent would change with amount of gas adsorbed, similar to the behavior
129
exhibited by water vapor in zeolites. Thus, initially, small quantities
of gas would be transported with a low diffusion coefficient. As the
amount of gas in the sorbent layer increased, the effective diffusion
coefficient would increase. The overall effect would be to alter the
shape of the response from the exponential form of equation (4.75).
124
-------
Using the model developed above, the response of a piezoelectric
mass monitor can be estimated. Using typical dimensions for an AT cut
quartz crystal resonating at 9 MHz, a diameter of 12 mm and a thickness
3
of 0.15 mm, and the density of quartz crystal, 2.65 g/cm , the mass of
the quartz crystal, Mg, would be about 45 milligrams. Triethanolamine
(TEA) will be used as an example of sorbent because it has been used to
collect both N02 and SO-. Assuming a layer of triethanolamine,
density 1.124 g/cm deposited on the crystal to a thickness of one
thousandth of an inch, 25 micrometers, covering a circular region of
diameter 0.6 cm, the added mass is 794 micrograms. This is slightly
under 2% of the crystal mass so the frequency shift will be proportional
to the added mass fraction, or about 640 KHz, according to equation
(4.70).
In order to obtain a quantitative estimate, it will be assumed
that the saturation concentration of TEA corresponds to one molecule of
S02 (or N02) per molecule of TEA. On this basis a total mass of 340 yg
of SO- and 245 yg of NO- could be ultimately absorbed. However, at the
ambient air concentrations normally encountered, the amount absorbed will
be well below this so the approximate formula, equation (4.75), can be used.
Two parameters of interest are the solubility parameter, S, and
diffusion coefficient, D. These can be determined by experiment, using
equation (4.74) with n=0, equations (4.75) and (4.70). Since the density
of the sorbent, pc, is known, the thickness-area product, 6A, can be
O
obtained from the measured frequency shift when the coating is applied
to the electrode, AF , whose area A, can be measured directly. Thus
AA = A(AF )/2.3 x 106F2p . (4.76)
e s
By introducing a step change in partial pressure, Ap, of the pollutant,
e.g., S02, and determining the time constant of the corresponding response
curve, Tn, the diffusion coefficient can be determined from
D = 62/7r2TQ (4.77)
where the value of 6 is obtained from equation (4.76), above.
125
-------
The solubility, S, is determined from the steady state solution of
equation (4.75) with AW determined from the measured frequency, AF ,
resulting from a step change of pollutant gas partial pressure.
An estimate of the response expected from a TEA coated sensor for
measuring S0~ can be obtained by estimating values for S and D. For non-
reactive gases in liquids, diffusion coefficients are on the order of
-5 2 132
10 cm /sec. In the case of water diffusing interstitially in zeolites,
-7 2
the diffusion coefficient is on the order of 10 cm /sec unless the
structure is totally dehydrated, in which case the diffusion coefficient
129
can be two orders of magnitude lower. Assuming that for the S02/TEA
"7 O /
combination D = 10 cm /sec and that the TEA coating is 25 x 10 cm
thick, the time constant will be, from equation (4.74)
TQ * (25 x 10~4)2/Tr2 x 1Q~7 = 6.33 seconds.
From the study by Frechette, £t a.1, if it is assumed that TEA will have
similar adsorptive behavior for SOp as tripropylamine, the frequency change
expected for TEA would be about 9 Hz per ppm.
From a resolution of 1 ppb in the measurement, the circuit for
measuring frequency change would have to be capable of measuring a shift
_2
of about 10 Hz. At the 5 ppb level, this would correspond to a period
difference of about 4 sec. A maximum counting period of 30 sec, corres-
ponding to about 4 ppb, would set the response time. The circuit design
for this would be a straightforward application of current digital tech-
niques.
Although TEA reversjLbly adsorbs S09, development would be required
to find a coating that would reversibly adsorb N0~. Use of selective
filters would prevent interferences. There is no direct way that 0~
could be measured with such a system.
As discussed above, precise temperature control of the crystals and
conditioning of the incoming gas is required to prevent frequency shifts.
4.10.3 Projected Performance
Assuming precise temperature control of the crystals and incoming
sample air stream, the piezoelectric mass monitor should be capable of
a dynamic range of 5 ppb to 1 ppm with a resolution of 1 ppb. A response
time, of about 30 seconds would be required if the measuring and reference
frequency crystals are exact frequency duplicates.
126
-------
4.10.4 Logistical Factors
Active sampling would be required. Flow rates would be moderate,
well within the capabilities of the portable pumps listed in Table 2.1.
Although the response is not dependent upon flow rate from the point of
view of pollutant sorption, it can be indirectly (and drastically) affected
by any temperature shifts caused by flow changes. For this reason flow
control should be employed. A significant amount of power would be
required to maintain a constant temperature, typically 50° to 60°C, for
the crystals.
4.10.5 Dose Rate Information
A digital signal, proportional to pollutant partial pressure, would be
available every 30 seconds for storage by techniques discussed in Section
5.0.
127
-------
SECTION 5
5.0 DATA ACQUISITION
The function of a data acquisition system is to measure one or
more parameters and record the result for future observation or
processing. The data acquisition system for the personal dosimeter
must accomplish this function under several constraints, specifically
low power consumption, low mass and small size. The unit must consume
minimal power to save as much power from the source as possible for use
with the sensor instrumentation. Likewise, mass and weight should be
kept to a minimum since the state-of-the-art allows a much greater degree
of miniaturization in data acquisition system design than in continuous
ambient air monitors.
5.1 General
Types of data acquisition systems adaptable to the personal dosimeter
range from simple systems to versatile, complex systems. At the simplest
end of the scale are data recording mechanisms which are an inherent
part of the sensor. In this case, the sensors permanently record their
measurements in the form of a stable chemical change. Both those instru-
ments which integrate data over long periods of time, such as film badges,
and those which sample continuously, such as tape samplers, fall in this
category. These systems are characterized by specialized data recovery
systems which are dedicated to the recovery of the information stored by
the chemical process, such as photographic processing equipment and
absorbance spectrometers. This type of data acquisition system lacks the
capability of measuring parameters other than the ones for which it is
specifically designed. This has two implications: (1) the system cannot
easily be changed to record different parameters by simply changing sensors,
and (2) the system cannot record parameters from a number of different
sensors but is limited to measuring only one parameter.
A second type of data acquisition system adaptable to small, portable
systems is the electrochemical type. Systems in this category most
frequently consist of strip chart recorders. While these systems definitely
serve a need because of their immediate visual readout of recorded data, the
difficulty in taking a quantity of data off the charts for computer processing
makes their exclusive use in environmental systems inconvenient. However,
128
-------
in systems where weight, size, and power permit, these recorders are useful
for two purposes: (1) as a display which shows past data as well as present,
and (2) as a back-up unit for recording data when the primary system becomes
temporarily inoperative.
The third type of data acquisition system is the electronic systems
which record signals in analog manner, typically on magnetic tape. These
systems, in order to function, must run continuously, which results in two
consequences: (1) if the duration of the data acquisition period is long,
large data storage capacity must be provided resulting in a bulky package,
and (2) due to continuous operation, power must be drawn continuously
resulting in higher current drain than other type systems. These reasons
make the use of this type of recorder inconvenient for personal dosimeter
application where it is not essential to monitor the data continuously.
A fourth type of data acquisition system is the analog/digital system.
This system is the type most frequently found where data is to be processed
by computer at the time of acquisition or later. This system is characterized
by electronic circuitry capable of signal conditioning, multichannel scanning,
analog-to-digital conversion and digital storage. This system has the
advantage of fast, quiet operation, powerful signal processing capability,
high resolution and accuracy (if desired) and output in a form compatible
with computerized bulk data handling systems. Also the systems can be
miniaturized and designed for low power consumption by switching off those
portions of the system when they are not actually needed for operation.
Of the four types of data acquisition systems mentioned here, the
latter, the analog/digital system, is the one most suited to a minaturized,
easily portable system when used with sensors for continual monitoring.
The first system classification, chemical systems, is also suited to small,
portable systems where immediate access of data is not needed and chemical
processing of the storage medium can be accomplished. These systems are
discussed in Section 6. The digital system, however, is a complete self-
standing system capable of use with any sensor or instrument which has
an analog voltage output.
129
-------
A general data acquisition system block diagram is shown in
Figure 5.1. This diagram shows the different sections of a data
acquisition system which utilizes the analog-to-digital conversion
approach. Not all of the blocks shown will exist in every system, for
in some applications not all of them are needed. Signals enter the
system from sensors at the signal conditioner. The signal conditioner
consists of a group of amplifiers, filters, or other signal processors
which convert the input to a form more compatible with the analog-to-
digital converter. Examples of types of circuits found are integrators,
differentiators, resistance-to-voltage converters, amplifiers, low pass
filters and non-linear networks to linearize responses of non-linear
sensors. By preprocessing the data with a signal conditioner, quite
frequently some stringent system requirements are lessened on other
parts of the system. For example, noisy signals when sampled at fixed
intervals can yield misleading averages unless the sample rate is twice
the highest frequency component of the input signal. However, if an
integration or a low pass filter is installed at the input, high
frequency components are removed and the sample rate may be dropped to
the desired observation rate (i.e., typically once every 5 minutes for
environmental data).
After conditioning, the signals are passed to a scanner. This
unit connects each of the active channels to the analog/digital converter
for conversion. Its rates are controlled by the system control unit which
insures that (1) the scanner leaves an input connected to the analog-to-
digital converter (A/D) long enough for the A/D to stabilize and complete
its conversion, and (2) the sequence of scanning all channels is initiated
once at the beginning of each interval of preselected length. The scan
interval is selected at the start of a period of data acquisition based
on the characteristics of the data being monitored and usually is not
changed until the type of test is changed. The number of channels for
which a scanner may be designed is practically limitless. However, for
this application, an 8- or 16-channel system will be adequate to handle
all data inputs as well as some parameters significant to system performance
130
-------
Signal
Conditioning
Scanner
Analog/
Digital
Converter
Data
Formatter
u>
Data
Storage
/ Memory \
( Unit J
Digital Data Entry
Manual
Data
Entry
Control
Circuitry
Clock/Time
Code
Generator
Figure 5.1. Generalized Data Acquisition System.
-------
(for example, power supply voltage). A scanner of this size could be
implemented with various analog switching devices such as relays or
crossbar switch, but for minimal power dissipation on minimum size
switches, the electronic-switch, field-effect transistors appear to
be most feasible. They require less power because no mechanical coils
are used and they occupy less space due to advances in integrated
circuit technology.
The function of the analog-to-digital converter is to convert each
analog voltage level to a digital representation which is compatible
with the storage devices. Currently available devices are low power
A/D converters utilizing Complementary Symmetry Metal Oxide Semiconductor
(CMOS) logic. These units function on a power of only a few milliwatts
while actually converting from analog to digital and consume microwatts
while on standby. Eight-bit models offer resolution of 1 part in 256 which
is adequate for state-of-the-art environmental monitoring and are available
in single CMOS chip packages, making their use in this application very
attractive. However, increased resolution may be obtained by utilization
of 10- or 12-bit units, sacrificing only speed and some size.
Speed of these devices may be obtained at the sacrifice of low power
consumption but scan rates of less than 10 per second are more than ade-
quate for environmental monitoring and are considered very slow for digital
acquisition systems.
The output of the analog-to-digital converter consists of 8 to 12
parallel bits of digital information. Before storage, some information
must be added to this in order to produce a complete, usable unit of data.
Information that must be associated with the data includes source of data
(channel number), time of data, and events which occurred simultaneously
with the data. For units where the amount of storage is not a premium,
the channel numbers of each data point can be stored along with the data.
For systems which use more expensive memory units (cost, power or size),
some scheme may be worked out whereby the data for all channels is formatted
sequentially in the order of the channel numbers. A unique, recognizable
character group is inserted ahead of the group to mark the beginning of the
data. The original channel of the data may then be determined by counting
data points after the special character group.
132
-------
A problem frequently encountered with data recovery from a data
acquisition system is that of correlating the data with events which
have an impact upon the data. Events in the case of personal dosimetry
include location and activity of subject, time, range of instruments,
among others. It is possible to correlate recorded data with events
which occurred as the data was taken by at least two methods, each
with its drawbacks. The first method entails the recording of pertinent
events manually in a log either by the subject or an observer. Disadvan-
tages in this procedure include the possibility of loss of important
data either through loss of the physical data record or neglect of the
person recording data to make the proper entries in the log. Another
drawback of this procedure is the effect on the data brought upon by
the time, effort, and concentration necessary for the subject (or an
observer) to make notes and observations. The second method of correlation
of the recorded data with pertinent events is the analysis of data to
determine the point of occurrence of easily detectable events or changes
of activity. This procedure is usually carried out in addition to log-
keeping. An example would be the change in ambient temperature which
occurs when the subject leaves a protected environment. The disadvantage
here is that the variability of the events is so high that the data must
be processed manually, significantly increasing the amount of human
processing necessary before the data may be processed by computer.
A solution to this problem is the digital data entry system. This
block on the functional diagram allows data already in digital form to
be stored along with the digitized analog data. Data stored typically
comes from an internal real time clock or binary coded decimal (BCD)
digit switches. Prearranged codes allow the switches to be used to input
data pertinent to the test in progress. This data then is located in
chronological order along with the digitized data taken at the same time
and contained on the same storage device. Both data can then be read
directly into a computer and processed routinely without the input of
excessive data from outside the storage device.
The data storage device is the unit which puts the data on the
removable storage record. In the case of tape systems, it is the tape deck.
133
-------
For solid state memories, it may include hardware which accepts a memory
board and stores the data at appropriate addresses on that board. The
data storage device is the device which determines the characteristics
of the data acquisition system as a whole. For dosimeter applications
which are to take data for periods of from 8 to 24 hours, magnetic tape
memories and solid state memories appear to be the most feasible. This
application requires 8 to 24 hours of data to be taken over one interval
continuously. At sample rates of once per minute, at least some magnetic
tape systems have a surplus of data storage capacity allowing storage of
items of lesser importance (more digital inputs, analog signals indicative
of other than primary importance). Solid state memory systems, however,
are expanded in increments of 1024 or 4096 bits at a significant cost
per bit ($0.25/bit for 4K MNOS). Consequently, for those memories it is
more economically profitable to conserve memory space wherever possible.
Digital magnetic tape memories suitable for this application would
have to be small and low power. This narrows down the selection of
recorders to the cartridge and cassette data recorders available today.
At least one manufacturer produces units which use the Phillips Cassette,
a tape housed in a case approximately 6.4 x 10.1 x 1.4 cm (2 1/2 x 4 x 7/16
inches). The unit is an incremental unit which means that the tape is
moved past the recording head a small amount during the recording of each
bit and then stopped during periods of no data. Consequently, power is
saved in two ways: (1) no temporary storage buffer is necessary to save
up data so that it can be placed on tape at a fixed rate while the tape
is moved at a constant speed and stopped between groups of data, and
(2) incremental recording requires that power be drawn only while data is
recorded and no power is consumed while in the standby mode.
The other option for data storage is solid state memory. There
are three types of integrated circuitry memory devices, each with its
own advantages for use in this application. Metal Oxide Semiconductor-
Large Scale Integration (MOS-LSI) has the advantage of high density
packaging so fewer circuits are needed. Complementary Symmetry Metal Oxide
Semiconductors (CMOS) have lower power drain. However, power must be
maintained on both these units or data will be lost unless a non-volatile
memory such as the Metal Nitride Metal Oxide Semiconductor is used.
Typical designs using these devices are described in Section 5.3.
134
-------
As with the design of any system, there are a number of tradeoffs
which may be made in obtaining a system which best satisfies the
requirements dictated by the application. For personal dosimeters,
requirements for small size and low power are more stringent than most
other applications. However, if allowable package size does permit the
use of a miniaturized tape recorder, one is assured that sufficient
storage will be available for monitoring purposes for periods up to 5 days.
However, with the tape system, if less storage is needed, no space can
be saved by sacrificing half of the memory capacity. Solid state memories,
however, can be designed into systems only in the quantity necessary to
accomplish the task. This means the size of the system and power consumption
may be varied depending on the amount of storage necessary. (An exception
arises in that power consumption may not vary appreciably if (1) memory
circuits do not draw substantial power relative to other circuitry or
(2) they are non-volatile circuits such that power may be removed when
data is not being stored. Therefore, additional circuits do not result
in increased consumption.) Therefore, for given maximum size, power
consumption, or cost, the maximum size memory system may be determined.
For a given fixed memory size, tradeoffs now exist for the parameters,
number of channels scanned, resolution, and sample rate. Obviously, the
amount of data recorded is directly proportional to the rate at which it
is taken. Likewise, it is also directly proportional to the number of
inputs which must be scanned each time. The desired resolution also has
an effect on the storage because the greater the desired resolution, the
more bits it takes to represent one reading.
Each of the above parameters may be determined from the application,
and the size of the memory necessary can be computed. If a memory of
sufficient size cannot be accommodated in the allowable weight and power
budget, there are design alternatives which allow a reduction in storage
necessary to record data pertaining to a given experiment.
First, as previously mentioned, either low pass filtering or inte-
gration can be used on the input to reduce the necessary sampling rate,
thus saving memory. Another way of possibly conserving memory arises
in the case where special processing is to be applied to the data. The
processing may be done "on-site" by analog processing before the data is,
135
-------
frequently reducing the sample rate necessary for recording the data.
Examples of this are where rates of change or repetition rates are
desired, as found in heart rate or respiration rate. Here, rather
than recording EKG at a fast enough rate to determine when each pulse
occurs, analog processing can be used to determine the pulse rate and
produce an analog voltage for digitization which is proportional to
the pulse rate. Another approach which can be used where a count of
events is desired is to simply bypass the analog-to-digital converter.
In this case, the count is computed during each scan interval and placed
in a register connected to the digital data entry point. This causes
the data to be placed in memory along with the digitized data. The
savings in memory space arises because the scan rate does not need to
be made fast in order to allow special processing of the data to obtain
a single value for a given parameter from a number of sample points.
It is only necessary that the sample rate be fast enough to insure that
the user can observe it as often as necessary.
Another option which may be exercised in cases where size or power
requirements restrict the allowable size of memory is a histogram type
memory. Instead of saving the measured parameter value for every point
in time (one per minute for each input channel), the electronics package
monitors the level of the input voltage. This unit is connected to a
bank of counters, each counter representing a range of values within the
expected readings. A periodic clock pulse is input to logic circuitry
which determines which of the counters the pulse is used to increment.
The end result at the end of the period of data acquisition is that each
counter contains a number proportional to the amount of time the monitored
value spent in the range of values corresponding to that counter. This
information can then be plotted in the form of a histogram, yielding
information concerning the average value and the variance of the data.
However, information relating the data to time is lost. This is an
example of one tradeoff which may be made at a sacrifice of one group
of information. Substantial reduction in system cost, complexity, and
power consumption may be made.
Unless data acquisition is to be an exercise in futility, some
provision must be made for getting the data from the storage device to
136
-------
a place where it can be inspected or further processed. Typically, data
is transferred into a computer where it is reformatted for inspection and
placed on a computer compatible magnetic tape for further processing.
For magnetic tape systems, the problem of data recovery simply becomes
removal of the tape cartridge, which is always a separate module. There-
fore, the solid state memory must be self-standing once removed. This
means that for volatile memories, some means of power supply (batteries)
must be maintained in the memory modules. Certain memories require pulsing
circuits to keep the memory refreshed as well. Non-volatile memories such
as the MNOS memory do not require either power or refresh circuitry to keep
data and may hold data for periods up to a year with no power applied.
The question of how much circuitry to include in the memory module
other than memory itself is not one with an obvious answer. Some of the
circuitry which is used to interface the memory board to the system has
to be duplicated in the data recovery unit. This circuitry includes
addressing circuitry, the aforementioned refreshing circuitry, and the
memory control. Approaches used may vary from a self-contained complete
data acquisition system to a system with only the memory circuits themselves
in a module. The configuration used would depend pricipally on economic
factors such as number of memory modules needed relative to number of data
acquisition systems.
The following two sections illustrate example designs which could be
implemented using magnetic tape systems (Section 5.2) or solid state
memories (Section 5.3). The designs are intended to be typical for personal
dosimeter application but would probably vary somewhat in specific details
dependent upon application.
5.2 Magnetic Tape System
The configuration of a data acquisition system using magnetic tape
would be exactly as that described in Figure 5.1 with the storage unit
becoming a magnetic tape unit. No other modifications in the system would
be necessary to accommodate the tape unit. The tape unit is a digital
tape unit designed to interface to the digital output of the analog/digital
converter through the data formatter.
Digital tape decks come in several forms ranging from the 7- or 9-track
driyes used in a large computer systems to new cassette drives small enough
137
-------
to be hand-held. Of the systems currently developed, only two general
categories seem to be applicable to personal dosimetry. These two types
are the cassette recorders and cartridge recorders. Cassette recorders
use the Phillips Cassette which is a high quality version of the cassette
used in many portable and home entertainment audio systems. Cassettes
intended for digital use are manufactured to much tighter tolerance some-
what larger than cassettes and use wider magnetic tape (0.633 cm). At
the present time there are a number of different cartridges available,
because different manufacturers have introduced different systems of tape
decks and cartridges within the past few years. One problem with adapting
either of the two types of systems to personal dosimetry is size. While
designers tried to keep the size of a tape cartridge to a minimum to facili-
tate handling, not much effort has been expended to reduce the size of the
deck itself. However, there are some manufacturers who do produce rela-
tively small, low power units, possibly suitable to this application.
Basically there are two different types of systems for transporting
tape past the recording head, incremental and continuous. Incremental
recorders move the tape in discrete steps placing one bit per track at
a time on tape. Gaps are produced by advancing the tape a number of steps
while not putting on any data. The continuous recorders put data on
continuously while running. Consequently, considerations must be made
in the supporting electronics package. Since the tape moves continuously
while recording, room must be left between groups of data recorded at
one time to allow for stopping and starting the tape. In order to make
good utilization of available tape, these blocks of data should be long
so that most of the tape is data separated by occasional gaps and not
mostly gaps with short blocks of data. Since data will arrive at the
tape unit at a non-constant rate, buffer storage must be provided to save
data as it is input to the tape deck until enough is received to place
on tape. The tape is then advanced and data is read from the buffer at
a rate compatible with the tape unit. Gaps (inter-record gaps) are then
left at the end of each block of data stored on tape to allow the recorder
time to stop and start after data ended and before data is started.
Of the two types of recording systems, incremental seems the most
desirable for several reasons. One, it does not require an external
138
-------
buffer for temporary storage of data before placing it on tape because
it simply places it on tape as it receives it without waiting for a whole
block of data to be received. Also, more efficient use of the tape can
be gained because of better control of the tape by the capstan which incre-
ments it past the head.
Magnetic tape systems have several advantages over other types of
data acquisition systems. Some of these are listed as follows:
(1) Magnetic tapes have capability for storing a large quantity
of data on one tape. A cassette in one typical system can
hold 2.2 x 10 bits or 137,000 data points (16 bits each).
(2) Tapes provide a low-cost, high density permanent storage
for data. With magnetic tape systems no additional
translation step is necessary to get data to magnetic tape
for archiving data in a computer compatible format.
Along with the advantages gained by a magnetic tape system come
several disadvantages as well. Some of these are listed below:
(1) For systems where minimum data storage is needed, much
storage space is wasted and the cost of the system per
stored data point becomes high.
(2) It is difficult to miniaturize the system much after the
electronic package is reduced due to the mechanical inter-
face with the tape cartridge including motors, mounting
plates, etc. This could become very significant when gas
analyzer technology allows pocket size dosimeter.
Typical Design
A system suitable for monitoring the outputs of gas analyzer,
physiological parameters, and ambient temperature/humidity is shown in
Some continuous recorders do not use a capstan drive at all but regulate
speed of tape by keeping hub speed constant. Bit density on tapes
produced by this method may vary considerably as tape builds up on the
hub pulling the tape past the recording head.
139
-------
Figure 5.2. This system was designed to measure and store the following
parameters:
Integrated and peak value for ozone (2 channels)
Integrated and peak value for NO (2 channels)
Integrated and peak value for N0_ (2 channels)
Temperature
Dew Point
Heart Rate
Respiration Rate
The system is to be able to store this data for a minimum of 8 hours
and preferably 24 hours. The unit is to be composed of readily available
modules illustrating state-of-the-art in off-the-shelf manufactured
equipment. To keep costs of the designed systems to a reasonable level,
no special developmental items such as further miniaturization of tape
decks were included. Also all components used are proven in everyday
usage, and consequently, the system could be assembled with confidence
that its performance would meet the predicted specifications.
The unit was designed around a commercially available data acquisi-
tion system. This unit, built by DATEL, contains circuitry necessary
to interface with 16 analog channels, digitize the data, and place it on a
cassette via its own internal incremental cassette tape deck; it requires
external clock, control circuitry, manual data entry hardware, and signal
processing in order to make it a complete system. This unit was selected
because it appeared to be representative of a system fabricated from
individual components and its performance specifications determined.
Specifications for the unit may be found in Table 5.1.
Several options are available to modify this system depending on
the specific application. These options include:
Repackaging — The data acquisition could be repackaged in separate
modules or a single module of a different shape. However, a
requirement does exist for the smallest dimension to be at
least 7.5 cm to accommodate the tape unit.
Expansion of input capability — Due to the more than adequate size
of the memory capacity, the input capability may be increased
to 16 channels with no additional hardware. Expansion beyond
that does involve more extensive control circuitry modifications
and additional multiplexors external to the DATEL unit.
140
-------
C/MOS
16 Channel
Multiplexer
Low Power
Sample
And
Hold
Amplifier
C/MOS
8-10-12 Bit
A/D
Converter
Data)
C/MOS
Tape
Formatter
And
Tape Control
C/MOS
Multiplexer Sequencer
System Programmer
Input Storage Register
"006000 ~o
a S
o at
•o k.
O ±J
•8 o
Uf
*J CJ
CO C
VI
(X
Ǥ
a
^
a
X
3 3 3i- M H
ooo u a
fl o o *
^i cu n *-> a v H
O « 3 C H "H O
O U iJQ O U.
l-l C^ 4-1 M
rHl CO O OJ
-------
Table 5.1 Power Requirements and Other Logistic Factors for Tape
Recorder Unit
POWER CONSUMPTION
Telemetry Reception 14 ma at + 12 v
(FM2 Receivers)
Signal Conditioning 0.3 ma at + 12 v 0.3 ma at -12 v
(6 op-amps)
Data Recording record mode 80 ma at 12 v
2 sec/min
(DATEL Unit) standby mode 0.01 ma at 12 v
average current 2.6 ma at 12 v
Control Circuitry 0.6 ma at 12 v
(CMOS Circuitry)
Total Current, average 17.5 ma at + 12 v 0.3 ma at -12 v
peak 97.5 ma at + 12 v 0.3 ma at -12 v
SIZE 10.16 cm x 10.16 cm x 30.48 cm
Includes DATEL unit, 2 control circuitry cords,
2 signal processing cards and 2 FM receivers
WEIGHT Less than 3 Ibs
TEMPERATURE RANGE -10° to 60°C (limited by tape unit)
142
-------
Sample rate — Also because of the abundance of memory space in the
cassette, the rate at which the data is sampled may be increased
without exceeding the memory capacity in an eight-hour day.
Power consumption will be increased somewhat but will still be
low enough to allow battery operation as planned.
Signal processing — Options vary widely for signal processing.
Simplest on the scale of complexity is point sampling in
time. Options exercised in this design include integration
over the sampling interval and peak detection during the
interval.
5.3 Solid State Memory System
Solid state memories are the newest type of memory to be incorporated
into system design. Their use is expanding into new areas such as computer
memories where large quantities of memory storage are needed. They have
also made possible production of memory systems which are smaller, faster,
cheaper, and lower power then conventional memory systems. The major
advantages of solid state memories are low power, fast read and write
speed, and accessibility. The major disadvantages are that they are not
well suited for mass storage of data and most types need power applied
continuously to retain data.
There are several different types of solid state memories, each
with its own advantages. This section will explore two different types
of these memories, the most common type called MOS-LSI, and a newer,
non-volatile type called MNOS. A data acquisition unit using these two
memory types is described below. The unit is made up of two basic parts,
the data control section and the memory section. The memory section is
different for each type of solid state memory device, but the data control
section would be the same for both solid state memories and the magnetic
tape system. The next subsection describes the data control section and
this is followed by descriptions of the two memory systems.
The function of the data control section is to take the data from
the sensors and make it compatible with the memory for storage. A typical
block design is shown in Figure 5.3. The basic parts include a multiplexor
to successively select which data channel to be processed, an analog-to-
digital converter (A/D) to transform the signal to a binary form, and a
143
-------
From
Sensors
Multiplexer
A
Analog
Channel
Select
Counter
A/D
Sample
Sensor
Data
Data Mode
Select
Manual
Data
Manual Data
Switches
Reset
Data Ready
Data Received
i
o
H
Figure 5.3. Data Selection Control.
144
-------
counter/timer section to keep track of when to take new data. Also
included would be a row of switches to be used to enter coded manual
data. This section is basically the same as that described under
magnetic tape systems.
5.3.1 MOS-LSI Memory Systems
The first type of memory system is built around the Metal Oxide
Semiconductor-Large Scale Integration (MOS-LSI) circuits. This name,
MOS-LSI, comes from the process used in their manufacture. The
transistors are formed by depositing a layer of metal over an insulating
layer of oxide on the silicon base. The individual transistors are
so small that the manufacturers can integrate over 25,000 of them on
the same piece of silicon. They then interconnect the transistors to
achieve highly complex functions such as memories and registers. This
is how the term "large scale integration (LSI)" developed.
The basic method of storing data in these devices uses the fact that
once a memory transistor is forced into one of its two stable states, it
stays in that state for a short period of time. To avoid loss of this
information after this brief time, it is necessary to rewrite the data
over itself. This process is known as refreshing and must be done
continually to keep the data. In presently available units, refreshing
must be done 500 times a second. However, a read or write cycle takes
less than one microsecond, so there is still plenty of time between refresh
cycles to record data.
This refreshing process makes it imperative to have a clock to keep
track of when it is necessary to rewrite the data. Since it is impractical
to write new data while refreshing, the control system must have logic
circuitry to decide when to write and when to refresh. It is also necessary
to have two counters to use as address registers. One of them is used to
store the location of the next vacant data block in the memory, and the other
counter retains the location of the next part of memory to refresh. Since
it is convenient to be able to write manually into the memory to log events,
this manual data as well as the necessary logic to control it should be
145
-------
included in the system. Perhaps the most complex part of the memory
system is the timing. This circuitry must properly sequence the necessary
pulses of the correct duration.
Presently it is possible to get MOS-LSI which have 4096 storage
locations in a half-inch by one-inch package. This large memory size
and small physical size make them perfect for portable applications.
The proposed data system has a digital output that is 8 binary digits
(bits) wide. That is to say that one sensor reading is represented as
8 bits in the memory. A memory system would consist of 8 of these memory
packages side by side to accept all the data bits simultaneously. With
an 8-package memory array (the smallest practical size), it would be
possible to record 4096 sensor readings or about 400 readings for each
of the ten sensors. At one reading per minute, this size memory array
would hold 6.7 hours of data. Since this is not enough storage space,
it is necessary to use two memory packages for each of the 8 data bits.
This 16-package memory array would last 13.4 hours at one reading per
minute. If a reading were taken every five minutes, this array would hold
67 hours of data. If a different number of channels or data sampling
rate is necessary, the size memory needed can be found by using the formula:
S = 8 C R H
where S = size in memory (in bits)
C = number of channels
R = number of readings per hour
H = number of hours data is to be taken.
To find the number of packages needed for an arbitrary size memory, it
is necessary to divide the memory size by 4096 and round up to the nearest
multiple of 8.
Figure 5.4 shows a typical block design using the MOS-LSI memories.
The size of the memory (8K x 8) was chosen to be a reasonable size for
this application. The following data is based on this size system. Since
this memory needs a refresh cycle to maintain data, a small battery must
be placed on the circuit board. Average power drain for this typical
memory array is about 250 mw and the necessary control logic would consume
146
-------
/
Data Received
Clock
Reset
Increment
Data Address
Register
/
Data Ready
Data Mode
Select
Increment
Memory
Control
/
Enable
Manual
Data
Sensor
>
Data
>
I
/
Data
Control
/
3
,.\L<
Write
Address
Memory Array
8x2
(4K each)
Data
/
Removable
Figure 5.4. MOS LSI Memory System.
147
-------
about 50 mw. The data selection control would require 20 raw, bringing
the total to 320 raw. If a different size memory is selected, the power
and cost would be the same for the data selection control section. In
the memory array the cost and power is proportional to the number of
packages, and in the memory control section about half the power and cost
is fixed and half is proportional to the memory size.
The part of the circuit on the right of the dashed line in Figure 5.4
is the minimum logic necessary to maintain the data. If this circuitry
and a battery were combined on one circuit board, it could be removed,
another put in its place, and the system would be operational once more.
The circuit board with the data could then be plugged into a readout
device and the data would then be transferred to a computer. It would
probably be easier to include everything except the data selection control
on this removable circuit board. This additional logic would simplify
the readout device and its interface considerably.
5.3.2 MNOS Memory System
The name Metal Nitride Oxide Semiconductor (MNOS) comes from the
fabrication process of the devices. They are made by placing a metal
layer over a layer of silicon nitride over a very thin layer of oxide
which is placed on top of the silicon base. Binary data is stored by
trapping charge in the oxide insulating layer. The charge changes the
threshold of the transistor thus permitting non-destructive readout.
The trapped charge remains in place even with no power applied, making
it possible to keep the data with zero power dissipation. This ability
to maintain information through power interruption gives them their other
name of non-volatile memories. The transistors can be electrically reset
so that they can be used repeatedly.
The control needed for this type of memory is less complicated than
that of the MOS devices because it needs no refresh cycles. Like the MOS
type memory, it needs a data control section to select the data from the
manual switches or the data from the sensors. The control section also
needs timing logic to pulse the memory array with the proper duration.
signals in the correct order to write into the memory. The memory control
section must also confirm to the data section that the data has been stored
in memory. In addition, it must increment the address counter used to store
the next vacant memory location.
148
-------
Presently non-volatile memories are available in a case size of
0.8 x 0.3 inches which contain 1024 memory locations. This is about
half the density of the MOS memories. Although less dense, these
memories make up for some of the additional circuit board space by
requiring less control circuitry and do not require a battery on board.
These non-volatile memories are perfect for small, ultra-low power
memories. To store the same amount of data as in the MOS example mentioned
earlier, an array of these packages 8x8 inches could be necessary. To
determine the memory size for an arbitrary number of channels or length
of time, use the same memory size formula as before. However, it is
necessary to divide this total memory size by 1024 and round up to the
nearest multiple of 8 to determine the number of packages necessary.
Figure 5.5 shows a typical block design using the non-volatile
memories. The 64K memory size was chosen to make it comparable to the
MOS design. Average power drain for the memory array is 3 mw and power
drain for the memory control would be 10 mw. The cost per bit is
relatively high now because the devices are new to the market, but the
cost is expected to drop as volume increases. The data selection control
section would be identical to the MOS type memory with power consumption
at 20 mw, bringing the total power drain to 33 mw. If the cost or power
consumption of a different size memory is needed, the control sections
would remain essentially unchanged, but the cost and power of the memory
array would be proportional to the number of packages.
If a removable circuit board is desired for this memory type, the
minimum circuitry would be the memory array itself. To make it easier
for readout, it would be better to include all of the memory section
except the data control on this board. It would not be necessary to put
a battery on this board and the life of the data would not be limited
by the power of the battery as is the case with the MOS memory.
5.3.3 Summary
The major advantage of solid state memories is low power and small
physical size in small and medium size memory applications. The best
parts of the MOS memory are its small cost and small size. The disadvan-
tages include large power consumption and complexity. The advantage of
using the MNOS memory is ultra-low power consumption, but this is offset
\
\
at present by higher cost.
Table 5.2 lists the major parameters of the solid state memory systems.
149
-------
Reset
Data ^
Received
Data ^
W
Ready
^ Data Mode
^ Select
Manual .
>
Data
Sensor
uai.a
Incremen
1 1
E
Memory
Control
\
t
1— '
1
„ ._ I
1
1
1
^^!
1
Data IS^
Control iX"**^
1
I
„. fr
w' Address
Register
t ^
^_
nable
Frite
HJ
Address
Memory
Array
8x8 (l.K each)
Data
•
Removable
Figure 5.5. MNOS Memory System.
150
-------
Table 5.2 Comparison of Memory Systems of 64K Bits
Section
Data Control
MOS Memory
MOS Memory Control
KNOS Memory
MNOS Memory Control
TOTAL MOS
TOTAL MNOS
Average
Power
mw
20
250
50
3
10
320
33
Equivalent
Circuit
Board Area
sq. in.
18
14
9
32
7
41
57
Hardware
Costs
$
280
430
70
1600
50
780
1930
151
-------
SECTION 6
6.0 INTEGRATED DOSE METHODS
These methods rely upon the collection of the pollutant gas and
retention until analysis can be performed. The gas can be adsorbed
onto a (presumably) specific adsorbent coated on a suitable solid
substrate to give a high surface/volume ratio. Alternatively the
air sample may be contacted with an absorbing liquid, selected for
its collection efficiency and for its compatibility with subsequent
analytical procedures used for readout of the dosage information.
With both types of collection media there are two forms of sample
acquisition that can be used, either the passive diffusion methods or
the active pumping methods discussed in Section 2.0.
These combinations of media/sampling will be briefly discussed
to develop their main features relative to dosimeter development strategy.
More detailed information is contained in the references cited.
Finally in this section the novel techniques of "packaged vacuum"
sampling is briefly discussed.
6.1 Solid Adsorbents/Passive Sampling
As discussed in Section 2.0, the diffusion tube sampler collects
a quantity of material that is directly proportional to the concentra-
tion-exposure time product for the pollutant provided the pollutant
concentration is kept at zero by total adsorption at the collecting
surface. The collecting material can be coated onto a metal screen
3
disc or a silica gel disc kept in position at the end of the diffusion
tube. The mechanical design of the tube should allow rapid, easy
removal of an exposed disc for subsequent analysis and replacement by a
fresh disc.
The minimum detectable quantity of collected pollutant depends upon
the analytical method used for readout. Using equation (2.2) from
Section 2.1.2 and assuming a minimum detectable quantity of 1 ug, a
2
diffusion coefficient of 0.2 cm /sec, and a diffusion tube 1 cm wide
and 8 cm long, collection over a period of 10 hours would enable the
3
detection of an average concentration of 1400 ug/m or about 0.55 ppm
152
-------
of S0_ or 0.75 ppm of NO^. In order to detect a level of 5 to 10 ppb
the analytical method would have to be able to detect 10 nanograms of
pollutant.
An alternative method of achieving diffusive flow is to use a
porous disc containing a large number, N, of small diameter tubes
oriented parallel to each other, i.e., a honeycomb construction. The
formula for diffusive flow through N parallel channels of length I
and cross-sectional area a is
Q = ND(f)C0t . (6.1)
By keeping a/1, the same as A/L in equation (2.2) and making N as large
as practicable, a large enhancement of diffusive flow can be obtained.
As indicated in Figure 3.2, molecular diffusion—such as experienced in
bulk air—can occur in pores as small as 1 ym in diameter at normal
temperature and pressure. Therefore, the use of thin porous discs as
diffusion covers over a suitable absorber looks attractive. The practical
limit to such a reduction is when condensible moisture in the sampled air
can clog up the pores or appreciably reduce their diameter.
Colorimetric changes appear to be more sensitive. A paper
impregnated with pH-sensitive dye has been reported to give a color
change detectable by eye to concentrations of S0~ of the order or 10 ppb
133
10 ppb following an hour's exposure. However, the dye requires a
humectant to provide water for the acid reaction which develops the color.
The degree of development is dependent upon temperature, humidity and—in
the simple flat plate geometry used—at low wind speeds, depends upon
windspeed. This problem with "flat plate" absorbers has already been
pointed out in Section 2.1. It is also probable that the developed color
change is not stable, but fades with time.
A colorimetric indicator for nitrogen oxides using dephenyl benzidene
decasulfonic acid (or an alkalai metal salt of this) on an inert solid
134
support has been reported. Another method for NO^ is to deposit
135
diphenylamineoxalic acid from alcoholic solution onto silica gel.
153
-------
6.2 Solid Adsorbents/Active Sampling
This approach uses a known, constant pumping rate to draw air
sample through a tube containing a packed granular adsorbing bed.
The adsorbed pollutant can be measured either through retention for
subsequent chemical analysis or be development of a color, with the
length of the colored zone being related to the concentration of
pollutant.
Because of normal variation in packing density, the packed bed
cannot be relied upon as the controlling restriction setting flow
136
rate. Such control is usually obtained by using a small diameter
137
limiting orifice on the vacuum pump side of the tube. In order to
prevent plugging of this orifice, a fine pored filter should be used.
The design of the flow system should insure that the pressure drop
across the packed sampling tube and filter are kept as low as possible.
Figure 6.1 shows schematically the components of pressure drop in the
system.
Pre-Filter
Absorber
Air
Intake
Ambient
Pressure AP
Critical Orifice
To Pump
D
Figure 6.1. Pressure Components in Packed Tube Samplers.
154
-------
In order for the critical orifice to function properly, controlling
the flow at the maximum, Q^ , the downstream pressure, P , must be less
than 53% of the upstream pressure, P . The type of pump used will
determine the suction pressure, P , at flow rate Q . In order to
maintain the required pressure drop across the critical orifice, the
sum of the pressure drops must satisfy the relation
&P1 + AP2 + AP3 < PQ - 1.887 PD . (6.2)
Even if this relation is maintained, there are still many sources of
error.
135
The equation for the flow rate of air at critical flow is
= 6.8 x IQ4 CAP /T mJl/min (6.3)
where C = discharge coefficient, usually of unity order,
(dimensionless)
2
A = critical orifice cross section area, cm
PTT = upstream pressure, atmospheres,
T = air temperature, °K.
With a 0.1 mm diameter orifice, at a pressure of 1 atmosphere and air
temperature of 300°K, and assuming unity as the discharge coefficient
value, the flow rate will be about 100 m$,/min. As can be seen from
equation (6.3) a 10°K change in temperature will give about 1.7% change
in flow rates. This assumes that C is independent of temperature, which
is dubious. In addition, PTT will vary as the ambient pressure varies,
and will depend upon the pressure drops, AP , which will vary from tube
Jx
to tube. All of these factors contribute to the uncertainty as to the
volume of air which has been pulled through the packed absorber.
When the concentration of pollutant to be adsorbed is small, the
length of packing, L, which has been coated with pollutant will be
. 138
given by
\ L = KC(1~n)V (6.4)
155
-------
where K = a proportionality constant
C = the volumetric concentration in air
V = the volume of air sampled.
The exponential term n is the same, as that for the Freundlich isotherm
139
equation for equilibrium absorption of SCL on the adsorber.
The collected material may be analyzed by chemical methods, or,
if a dye chemical is used, by photometric analysis of the length of
140
stain. Linch has reviewed the latter practice in detail.
Practical factors affecting the use of non-selective adsorbents
141
such as silica gel and activated charcoal have been discussed by Linch.
Water vapor is usually a problem with such adsorbents. It can also be
a problem on relatively selective adsorbents, although for a different
reason. With the non-selective adsorbents, water vapor tends to
preferentially cover the adsorbent surface and perhaps clog up the
micropores of these materials. This "crowds out" the other molecular
species, and can be construed as a physical process of interferance.
If SQy and N0£ react with adsorbed water to form acids, and these acids
are in turn capable of chemical reaction with some coating material, the
resulting chemical change of the surface might alter the uptake of
pollutant. Such behavior has been noted for SC^ in the presence of water
vapor on SDM co-polymer.
6.3 Liquid Adsorbents/Passive Sampling
The method here is essentially the same as that described above in
Section 6.1 except the collector is a liquid rather than a solid coated
on a support of large surface area. This technique has been used by
142
Palmes and Gunnison to monitor S02 with TCM solution as the collecting
fluid. The method is subject to the same drawback as the solid absorber—
lack of sensitivity. In addition, the cell must remain vertically oriented
in order to avoid spilling the collecting solution.
6.4 Liquid Adsorbents/Active Sampling
Bubblers, with an air sample drawn through an adsorbing solution,
have been used for many years in air pollution monitoring. Spill-proof
143 144
microimpingers have been reported and are available commercially,
matched to suitable miniature sampling pumps to obtain integrating
dosimeters.
156
-------
6.5 Packaged Vacuum
A novel approach to sampling for air pollutant dosimetry has been
145
taken by NRL scientists. This consists of a small cylinder evacuated
to high vacuum. In operation as a sampler, the cylinder is opened to
the atmosphere through a precision orifice. As long as the pressure
in the cylinder remains below 53% of an atmosphere, air flow through the
orifice is essentially constant at critical flow—which was discussed
above in Section 6.2.
The quantities of pollutants collected in the cylinder will be
proportional to the time averaged values obtained during the measure-
ment period. If conditions for critical flow are maintained, the
amount of each pollutant will be given by
m = QMCt (6.5)
where m = weight of material collected, grams
3
Q = flow rate through critical orifice, cm /min
M = pollutant molecular weight, grams
_ o
C = mean pollutant concentration in air, moles/cm
t = duration of sampling.
3
Using a sampling time of 8 hours, with a flow rate of 0.1 cm /min,
and assuming an analytical method capable of detecting 0.25 yg of S0»,
the minimum detectable average concentration is about 2 ppm.
This can be improved by increasing the amount of material collected,
i.e., by increasing the flow rate. The limit on this is imposed by the
size of collection container. Since the pressure in the container at
the end of sampling must be about 0.5 atmosphere, or less, the total
volume sample will be about half of the container volume. Another
method of decreasing the minimum detectable quantity is to increase the
sensitivity of the analytical method. Usually, however, this increases
the cost of analysis.
In addition to these drawbacks, the critical orifice may present
a problem due to the very small size that must be used to restrict the
flow to a manageable value. Partial or total plugging can drastically
\
157
-------
change the flow rate. This could be checked by measuring the final
pressure prior to "readout". Again, this is an additional processing
step which can significantly increase the cost of using this method.
158
-------
SECTION 7
7.0 DOSIMETRY SYSTEMS
Whether gathered in the form of an integrated sample or as a
recorded signal proportional to a time varying concentration, in
order to be of use as an exposure record correlated to activity of
the human subject, a "readout" must be obtained. This section reviews
the operational sequences that might be used in the deployment of
dosimeters and discusses some of the methods that might be used for
readout.
7.1 Integrated Dose Methods
At this point it is important to review the operation of the
dosimeter system to put into perspective the factors which influence
deployment and data quality. Figure 7.1 schematically presents the
operational sequence in the deployment of dosimeters for integrated
dose measurements. The dosimeter is considered to have 3 components:
a "holder" that mechanically protects the collector and enables the
device to be conveniently placed on the subject (weaver); a vessel
which will contain the absorbing reagent; and a standard quantity of
absorbing reagent. The dosimeter design could enable the holder to
stay with a subject in the field, or could require that the holder/
vessel/reagent be assembled in the laboratory each time prior to
delivery to subjects in the field, with the holder and vessel recycled
each time the dosimeter package is returned to the laboratory for
read out.
Presumably in a health effects study the subjects would maintain
an activity log which would be periodically submitted for comparison
and correlation with exposure history. This would probably be supple-
mented by periodic health examinations.
There are many places where errors can be introduced into the
measurement. These are indicated by numerals in parentheses in
Figure 7.1. They include: (1) contaminated reagents at assembly;
(2) spurious exposure of the blank, or partially or totally exposed
measuring reagent due to improper mounting of the dosimeter in the
field; (3) contamination during the decant/demount operation; (4) impro-
per clean-up of vessels and/or holders; (5) contamination or reagent
problems during sample processing; and (6) improper instrument cali-
bration and operation during analysis.
159
-------
^
x Holders
(1)
Absorbing . _
Reagents
(4) •-
Vessel
Cleanup ^
r - ~!
j i
HolHor < J •
J Check ^ j
i _i
jjects
N
in Field Activity
Log
1
f
1
Exposure (2) 1
%
f
Decant/ ,^\
Demount
f
Processing (5)
i
f
Analysis (6)
s
f
Dose Data
Health
Checks
Data Analysis
and Correlation
Figure 7.1. Integrated Dosimeter Operation Scheme.
160
-------
Operating procedures should take these potential error sources
into account and methods should be developed to minimize the possibilities
of introducing spurious exposure data.
Figure 7.2 shows schematically the processing flow for obtaining
dose measurements from the integral dose sampling methods of Section
6.0. Only collections via solid and liquid media are considered. The
dashed line in the figure represents a blank sample which has gone
through the same distribution, wear and collection process as the
measurement sample, but has not been exposed (for a significant length
of time) to air with the pollutant to be measured.
Material from solid adsorbers may be thermally desorbed to allow
analysis or it may be removed by "wet processing". As an example of
the former, assume that a brominated, amine modified silica gel cartridge
(see Section 3.2) is heated to about 95°C for desorption and then
carrier gas is used to conduct the desorbed S02 to a flame photometric
detector. The analog signal from the detector could be used as a
barographic record on a strip chart, or converted to digital form for
printout on a paper tape or put on a magnetic tape in digital form. As
an example of wet processing, assume that PbO, on a suitable substrate,
146
is converted to the sulfate form by exposure to SO . This is then
converted to a barium sulfate colloidal suspension, the concentration of
147 148
which is measured by nephelometer. ' This signal can be used for
barographic chart recording or digitized and further processed as dis-
cussed above.
As an example of processing of material collected in a liquid
absorber, assume that a diffusion tube sampler, or alternatively an
impactor sampler with a pump, has been used to collect S0» in TCM
absorber solution. This solution would be decanted and analyzed by
149
the standard West-Gaeke colorimetric method in parallel with the
blank solution. The signal from the colorimeter could be processed
in the barographic or digital modes mentioned above to obtain a record.
This type of system would be labor intensive, particularly in
the preparation and analysis stages. There would also be a lot of
labor involved in the coding of dose data, activity data and health
status data for subsequent analysis.
161
-------
Solid
Me as
Blank
Thermal
Desorb
Analysis
Dose + Background
Background
I
Wet
Processing
Analysis
Dose + Background
Background
ON
M, .,, fe
Blank
Wet
Processing
__ ?
Analysis
Dose + Background
Background
Figure 7.2. Dose Measurements from Integral Methods.
-------
7.2 Continual Dose Methods
Figure 7.3 schematically presents an operational sequence in the
deployment of dosimeters for continual measurements. The components
of the measurement package are: the instrument package containing all
of the "hardware" for 0_, N02 and SO- measurement; transducers for
activity monitoring and the data collection system; the calibration
reagents; and the measurement reagents. The instrument package with
reagents is delivered to the subjects in the field. Periodically, as
required, calibration and measurement reagents are replenished, either
individually, or possibly with the whole instrument package recycled
(dashed line).
Although the instrument pakcage could—and probably would—have
transducers to monitor subject activity (i.e., through heart rate or
breathing rate measurements), it is likely that the subject would also
be asked to keep an activity log which would be periodically submitted
for comparison and correlation with exposure history, supplemented by
periodic health examinations.
There are many places where error could be introduced into the
measurements. These are indicated by numerals in parentheses in Figure
7.3. They include: (1) contaminated calibration or measurement
reagents; (2) malfunction of instrument components in the field; (3)
spurious errors in the data collection system; and (4) improper clean-
up and check-out of the instrument package and reagent containers during
recycle.
Operating procedures should take these potential error sources into
account and methods should be developed to minimize the possibilities of
introducing spurious exposure data.
The calibration reagents might be employed automatically in the
operating system to obtain span and zero checks several times during a
day, or they might be applied manually at less frequent invervals—
say once a day. The frequency of such checks will involve a trade-off
between system complexity and instrument cost on one hand and data
reliability on the other.
163
-------
Instrument
.Packages
Measurement
Reagents
Subjects in Field
Activity
Log
T
Exposure
(2)
Portable Data
Collection
System
Data Format
Data Readout
Health Checks
Coding
V V
Computer
(3)
Coding
_J
Figure 7.3. Continual Dose Rate Measurement Operational Scheme.
164
-------
Figure 7.4 shows schematically the processing flow for data obtained
in portable collection systems. If "volatile" MOS memories are used,
a "keep alive" power system will be required to maintain the data intact
in memory during transport from the field to the central data collection
system. If MNOS or magnetic tape data storage is used, such an auxiliary
power supply will not be necessary. Each of the three different storage
media will require a unique interface design that will allow insertion of
the memory module and cycling to dump the data into a buffer memory
associated with a central processor (computer). This cycling will insert
the data in a pre-determined format.
As indicated in Figure 7.3 it is probable that a supplementary
activity log and health data will also be used. Data from both of these
sources will have to be coded for entry into the data processing system.
For this dosimeter system the bulk of the labor required should be
in the preparation, delivery and maintenance of the instrument packages
and the required reagents. Because of the complexity of the devices,
skilled instrument technicians or specialists will be required to maintain
the instrument used in the deployment network. This will be true even
when prototype instrument packages have been through a thorough develop-
ment and "shakedown" test phase.
165
-------
Field
"Keep Alive"
Power Source
Interface
Data
Record
From
Continual
Monitors
MOS
t
Format
Interface
MNOS
1
Format
Interface
Mag. Tape
t
Format
Digital
Computer
Analysis
Figure 7.4. Data Processing for Dose Rate Records from Continual Monitors.
166
-------
SECTION 8
8.0 CONCLUSIONS AND RECOMMENDATIONS
A practical pollutant dosimeter system must be designed to be as
simple and reliable as possible commensurate with the amount and quality
of the data obtained. For reasons of economics and deployment logistics
(which could also be translated back to economic terms) the system should
use as many common components as possible. This minimizes initial costs
through reduced design time and longer repetitive production runs. It
reduces spare parts inventory requirements. It reduces the
maintenance and trouble-shooting complexity.
If possible, the sensing principle(s) used should require little or
no reagent supply. This minimizes the logistic burden of continual re-
supply, which is costly not only in terms of raw material costs but also in
terms of the quality control measures required to assure that each batch
of reagent is within specifications.
Finally, the sensing principle(s) should require a minimum amount of
electrical power to be supplied during operation. This reduces initial
costs by reducing battery size or by trading off increased battery size
with longer operating time without recharge. Further, the weight of the
instrument package can be reduced (making subject cooperation easier) if
a smaller battery size is used.
With these considerations in mind, a comparison of the sensing prin-
ciples analyzed in Section 4.0 indicates that only one method offers the
possible combination of the characteristics cited above and at the same
time appears to come close to the accuracy and precision desired (CHAMP
Specifications, Section 1.1). This is the dual-cathode amperoraetric
system modeled in Section 4.2.2. By using suitable adsorbing filters to
select only one pollutant to be introduced to the cell for measurement,
the same basic cell design can be used for ozone, nitrogen dioxide and sul-
44
fur dioxide. Such filters exist and have been described by Lindqvist
97
and Anderson, et al. The system does use a liquid electrolyte and
would, therefore, be orientation limited, but only occasional make-up
water would have to be added to replace evaporative loss.
167
-------
In order to improve the response to N02, if deemed necessary,
research could be performed to determine additives for the electrolyte
which could improve N02 absorption efficiency while not interfering with
the electrode reactions.
The next choice from the list of continual sensors would be a
tape transport colorimetric scheme, such as that outlined on pp. 89-93
in Section 4.5.2. A developmental model for N09 monitoring has been
149
demonstrated by Lyshkow. Further research and development work would
have to be done to improve the mechanical tape transport design and
to determine suitable colorimetric reagents for ozone and S0«. As with
the amperometric system cited above, selective filters could (and would
probably have to) be used to eliminate interferences.
The third choice would be the luminol/lUO,, chemiluminescence
97
system of Anderson, et al. Although it uses a single detector, this
method has the drawbacks of requiring liquid reagent re-supply, a mechanical
device for alternately viewing each micro-reactor cell and a high voltage
power supply for the photomultiplier tube.
All three of these methods require active sampling, but suitable
portable pumps are already commercially available.
Of the data storage methods discussed in Section 5.0, the digital
casette system appears to be the most desirable from the points of
view of storage capacity, non-volatility and ease of entry into a computer
system for subsequent verification and analysis.
Continual monitoring of ozone, nitrogen dioxide and sulfur dioxide
at ambient levels by human subjects carrying a briefcase-sized instrument
package entails the operation of a system of which the instrument packages
are but a part. As discussed in Section 7.2, many operations are required
to deploy instruments in the field and to process the data obtained in order
to get meaningful information about the relation between pollutant exposure
and general health status. Periodic calibration is a necessity. Periodic
routine maintenance must also be performed to keep instrument performance
within specifications. These operations require skilled instrument
168
-------
personnel in addition to the skilled personnel expected to be employed
in the health checks and data interpretation. Adequate staffing must be
considered in addition to hardware costs when cost estimates for system
operation are made.
8.1 Recommendations
Current budgeting might not allow a large scale program to develop
and deploy continual monitoring devices, and might not for some time.
However, research and development of sensors and other components for
these devices should be encouraged. This encouragement can come in several
forms. EPA in concert with other interested governmental agencies can
sponsor forums to discuss the problems associated with miniature sensor
development, inviting industry and university participation. Hopefully
this would stimulate the speculative development of prototypes in these
sectors which could be improved for use either through strictly proprie-
tary in-house development or through partial government funding. The trend
of miniaturization and enhanced portability which began with the develop-
ment of transistors twenty-five years ago, is inexorably and spontaneously
leading to reduced instrument package size on the part of industry. The
trend is most evident in what are regarded as strictly electronic instru-
ments, i.e., voltage and current meters, signal generators and cathode ray
oscilloscopes. In recent years, however, increasing numbers of portable
gas sensing instruments for monitoring in the tens to hundreds of ppm
ranges encountered in occupational environments have been developed and
marketed. Knowledge of a potential market provides a powerful incentive
for private development.
Another avenue is for EPA to conduct in-house research of one or more
sensing principles and then underwrite the development by industry until
viable types are available on the marked. Alternatively, EPA could contract
out the research effort ab initio and then underwrite development or
license for development and marketing.
Whichever strategy is used, it is recommended that at the initial
stages of consideration of a particular sensing method. a modeling effort
such as that carried out in Section 4.0 of this report be undertaken. The
169
-------
better (more complete) the model, the better will be the performance
estimate. Such modeling techniques have been of great value in
electronic systems design and in chemical systems design. In
addition, such modeling would serve to elucidate the design and
operating parameters which influence the performance of the sensing
method, calling to attention those parameters which must be controlled
most closely in order to attain the desired sensitivity and stability
of calibration. Such a technique has been used to advantage by
McClenny et al for a proposed N09 sensing method.
1~~* £
Further, it is recommended that consideration and analysis of
sensing methods, at least mathematically and hopefully experimentally,
be initiated as soon as possible in order to provide time for critical
review and, hopefully, to stimulate alternate, competitive, approaches.
170
-------
REFERENCES
1. Environmental Protection Agency, Russel E. Train, Administrator,
Proposed Rule for Revision of "Significant Harm and Emergency Action
Levels for Photochemical Oxidants" Federal Register 39:9672-9675,
March 13, 1974.
2. Palmes, E.D. and A.F. Gunnison. Personal Monitoring Device for
Gaseous Contaminants. Amer. Ind. Hyg. Ass. J. 34:78-81, February,
1973.
3. Gunnison, A.F., M.E. Gunnison and E.D. Palmes. Personal Samplers
for CO, NO and N0« in Air - Annual Technical Progress. Report to
U.S. Bureau of Mines on Grant No. GO 133066, June 1, 1973 to
May 31, 1974. New York University Medical Center Institute for
Environmental Medicine, June, 1974.
4. Gregg, S.J. and K.S. W. Sing. Adsorption, Surface Area and Porosity.
New York, Academic Press, 1967. p. 9-19.
5. ibid, p. 36-49.
6. Rideal, E.K. Concepts in Catalysis. New York, Academic Press, 1968.
p. 62.
7. Perry, R.H., C.H. Chilton and S.D. Kirkpatrick. Chemical Engineers'
Handbook, Fourth Edition. New York, McGraw-Hill Book Co., 1963, p.
14-12-14-13.
8. Jost, W. Diffusion in Solids, Liquids, Gases, Third Printing with
Addendum. New York, Academic Press, 1960. p. 420-423.
9. ibid, p. JL4-14-14-15.
10. Adsorbents for Gas-Phase Odor Control or Liquid-Phase Heavy Metal
Removal, Patent, Japan Kokai, 72 23,383.
11. Active Carbon for Adsorbing Sulfur Dioxide, Japan Kokai, 72 33,890.
12. Polystyrene or Nylon Foam Fibers Containing Activated Charcoal for
Absorbing Toxic Materials from Air. Japan Kokai, 72 28,214.
13. Salek, J.M.. Adsorption of Gases on Gold Films. J. Phys. Chem.
77:1849-, 1973.
14. Hashida, I. £t aJL, Kagaku Kaishi 1973:1195-, June 1973.
15. Bostrom, C. The Absorption of Sulfur Dioxide at Low Concentrations
(pphm). Studied by an Isotopic Tracer Method. Int. J. Air. Wat.
Poll. 9:333-341, 1965.
16. Frechette, M.W. et al. Evaluation of Substrates for Use on a Piezo-
electric Detector of Sulfur Dioxide. Anal. Chem. 45:1765-1766,
September, 1973.
171
-------
17. Burwell, R.L. and 0. Leal. Modified Silica Gels as Selective
Adsorbents for Sulfur Dioxide. J.C.S. Chem. Comm. 1974:342-343.
18. Thomas, F.W. and D.M. Davidson. Monitoring Sulfur Dioxide with
Lead Perioxide Cylinders. J. Air. Poll. Control Assoc. 11:24-27,
1961.
19. Huey, N.A. The Lead Dioxide Estimation of Sulfur Dioxide Pollution.
J. Air.Poll. Control Assoc. 18:610-611, 1968.
20. Boulerice, M. and W. Brabant. New Pb02 Support for the Measurement
of Sulfation. J. Air Poll. Control Assoc. 19:432-434, 1969.
21. See Ref. 15 above.
22. Bracewell, J.M. and A.E.M. Hodgson. The Hydrogen Perioxide Method
for Sulfur Dioxide in the Atmosphere-Efficiency at Low Concentrations
by a Radioactive Tracer Technique. Int. J. Air Wat. Poll. 9:
431-438, 1965.
23. Pellizzari, E.D. Electron Capture Detection in Gas Chromatography.
J. Chromatography 98:323-361, 1974.
24. Weast, R.C. (ed.). Handbook of Chemistry and Physics, 46th Edition.
Cleveland, Ohio, The Chemical Rubber Company, 1965. p. E-68.
25. Nash, T. Absorption of Nitrogen Dioxide by Aqueous Solutions.
J. Chem. Soc. A 1970: 3023-3024, 1970.
26. Ma, Y. H. and C. Mancel. Diffusion Studies of C02, NO, N02 and S02
on Molecular Sieve Zeolites by Gas Chromatography. AICHE 18:1148-1153,
1972.
27. Altshuller, A.P. and A.T. Wartburg. The Interaction of Ozone with
Plastic and Metallic Materials in Dynamic Flow System. Intern. J.
Air Water Pollution. 4:70-78, 1961.
28. Haagen-Smit, A.J. and O.J. Brunelle. The Application of Phenol-
phthalein Reagent to Atmospheric Oxidant Analysis. Intern. J.
Air Pollution 1:51, 1968.
29. Bovee, H.H and R.J. Robinson. Sodium Diphenylaminesulfonate as an
Analytical Reagent for Ozone. Anal. Chem. 33:1115, 1961.
30. Egerow, M.S. Z. Lebensm. 56:355, 1928.
31. Bravo. H.A., and J.P- Lodge.Specific Spectrophotometric Determination
of Ozone in the Atmosphere. Anal. Chemistry 36:671, 1964.
32. Hendricks, R.H. and L.B. Larsen. An Evaluation of Selected Methods
of Collection and Analysis of Low Concentrations of Ozone. Am.
Ind. Hyg. Assoc. J. 27: 81, 1966.
33. Bradley, E.G. and C.J. Haagen-Smit. The Application of Rubber in
the Quantitative Determination of Ozone. Rubber Chem. Technol.
24:750, 1951.
172
-------
34. Atyaksheva, L.T. et_ al_. Physico-chemistry of Concentrated Ozone.
Zh. Fiz. Khim., 46:2602, 1972.
35. Jonathan, N. Infrared Spectrosocpy Application to Chemical Kinetic
Systems. NTIS AD-683771, 1968.
36. Atyaksheva, L.T. ^t^ _al. Physico-chemistry of Concentrated Ozone.
Zh. Fiz. Khim. 47:996, 1973.
37. Reimshuessle, H.K. and G.A. Mountford.
J. Colloid Interface Sci. 25:558-63, 1967.
38. Hall, R.C. A Highly Sensitive and Selective Microelectrolytic
Conductivity Detector for Gas Chromatography. J. Chrotnatog. Sci.
12:152-160, March, 1974.
39. Knipe, A.C. et^ ad. A Fast Response Conductivity Amplifier for
Chemical Kinetics. J. Physics E:Scientific Instruments 7:586-
590, January, 1974.
40. Tokiwa, Y. et al. Atmospheric Ozone Determination by Amperometry
and Colorimetry. In:Determination of Air Quality, Proc. ACS
Symposium, 1971, Mamantov, G. (ed.). New York, Plenum Press, 1972.
p. 109-130.
41. Boyd, A.W. et al. New Determination of Stoichiometry of the lodo-
metric Method for Ozone Analysis at pH 7.0. Anal. Chem. 42:670-
672, 1970.
42. Schechter, H. Spectrophotometric Method for Determination of Ozone
in Aqueous Solutions. Water Research 7:729-739, 1973.
43. Kromhyr, W.D. Electrochemical Concentration Cells for Gas Analysis.
Ann. Geophys. 25:203-210, 1969.
44. Lindqvist, F. A Differential Galvanic Atmospheric Ozone Monitor.
Analyst 97:549-558 , 1972.
45. Skoog, D.A. and D.M. West. Fundamentals of Analytical Chemistry.
New York, Holt, Rinehart and Winston, 1963. p. 378-380.
46. Philips Model PW9700.
47. Beckman Inst. Co. Model 906A.
48. Schulze, F. Versatile Combination Ozone and Sulfur Dioxide Analyzer.
Anal Chem 38:748-752, 1966; Schulze, F. U.S. Patent No. 3,361,661,
Jan. 2, 1968.
49. Byers, R.L. and J.W. Davis. Sulfur Dioxide Adsorption and Desorption
on Various Filter Media. J. Air. Poll. Control Assoc. 20:236-238,1970.
173
-------
50. Chand, R. and M. Shaw. Apparatus for Measuring Sulfur Dioxide
Concentrations. U. S. Patent No. 3,622,488, November 23, 1971.
51. Chand, R. and M. Shaw. Apparatus for Measuring Nitrogen Oxide
Concentrations. U. S. Patent No. 3,622,487, November 23, 1971.
52. Dynasciences Corp., Los Angeles, California.
53. Environmental Instrumentation Group, Lawrence Berkeley Laboratory.
Instrumentation for Environmental Monitoring AIR, LBL-1, Volume 1.
University of California, Berkeley, CA, May 1, 1972. p. 17-19.
54. Pasternak, R. A., M. V. Christensen and J. Heller. Diffusion and
Permeation Oxygen, Nitrogen, Carbon Dioxide, Nitrogen Dioxide
Through Polytetrafluoroethylene. Macromolecules 3:366-374, 1971.
55. Chand, R. and P. R. Cunningham. Characterization of Organic
Solvents for Electrochemical Air Pollution Sensors. IEEE Trans.
Geosci. Electronics GE-8:158-161, July 1970.
56. Ruch, W. E. Chemical Detection of Gaseous Pollutants. Ann Arbor,
Michigan, Ann Arbor Science Publishers, 1966. p. 168-173.
57. Saltzman, B. E. Direct Reading Colorimetric Indicators. In:
Handbook of Analytical Toxicology,
58. Linch, A. L. Evaluation of Ambient Air Quality by Personnel
Monitoring. Cleveland, Ohio, CRC Press, Inc., 1974. p. 19-78.
59. Patterson, G. D., Jr. and M. G. Mellon. Colorimetric Determination
of Sulfur Dioxide. U. S. Patent No. 2,785,959, March 12, 1957-
60. See Reference 22.
61. See Reference 15.
62. Lyshkow, N. A. System for Analysis of Gases in Trace Amounts.
U. S. Patent No. 3,535,045, October 20, 1970.
63. Cummings, W. G. and M. W. Redfearn. Instruments for Measuring
Small Quantities of Sulfur Dioxide in the Atmosphere. J. Inst.
Fuel 30:628-635, 1957.
64. DHEW, Title 42, Chapter IV, Part 410, Appendix A. Reference
Method for the Determination of Sulfur Dioxide in the Atmosphere
(Pararosaniline Method). Federal Register 36:8187-8189, April 30,
1971.
65. See Reference 53. p. 5-13.
174
-------
66. See Reference 56. p. 140-149.
67. See Reference 56. p. 151-156.
68. Boyd, A. W. et al. New Determination of Stoichiometry of the
lodometric Method for Ozone Analysis at pH 7.0. Anal. Chem.
42:670-672, May 1970.
69. See Reference 40.
70. Shechter, H. Spectrophotometric Method for Determination of Ozone
in Aqueous Solutions. Water Research 7:729-739, 1973.
71. Kortum, G. Reflectance Spectroscopy. New York, Springer-Verlag,
1969.
72. Lyshkow, N. A. Colorimeter and Elements Therefor. U. S. Patent
No. 3,617,136, November 2, 1971.
73. Nederbragt, G. W. et al.
74. Warren, G. J. and G. Babcock. Portable Ethylene Chemiluminescence
Ozone Monitor. Rev. Sci. Instr. 41:280-282, 1970.
75. DREW, Title 42, Chapter IV, Part 410, Appendix D. Reference Method
for the Measurement of Photochemical Oxidants Corrected for Inter-
ferences Due to Nitrogen Oxides and Sulfur Dioxide. Federal
Register 36:8195-8197, April 30, 1971.
76. Hodgeson, J. A. et al. Comparison of Chemiluminescent Methods for
Measurement of Atmospheric Ozone. In: Progress in Analytical
Chemistry, Volume 5, (ed). New York, Plenum Press,
P-
77. Clyne, M. A. A. et al. Kinetics of the Chemiluminescent Reaction
between Nitric Oxide and Ozone. Trans. Fradaday Soc. 60:359-370,
1964.
78. Fontijn, A. Chemiluminescent Method of Detecting Ozone. U. S.
Patent No. 3,528,779, September 15, 1970.
79. Fontijn, A. e£ al. Homogeneous Chemiluminescent Measurement of
Nitric Oxide with Ozone. Anal. Chem. 42:575-579, May 1970.
80. Hodgeson, J. A., et^ al_. Application of a Chemiluminescence Detector
for the Measurement of Total Oxides of Nitrogen and Ammonia in the
Atmosphere. Paper 71-067, Joint Conference on Sensing of Environmental
Pollutants, Palo Alto, CA, Nov 8-10, 1971. Proceedings available
from AIAA Library, 750 3rd Ave., New York, N. Y. 10017.
81. Winer, A. M. et al. Response of Commercial Chemiluminescent NO-NO-
Analyzers to Other Nitrogen-Containing Compounds. Environ. Sci.
Tech. 8:1118-1121, December 1974.
175
-------
82. Wooten, G. W. et^ a]L Chemiluminescent Method and Apparatus. U. S.
Patent No. 3,749,929, July 31, 1973.
83. Benson, S. W. et^ al. Detection Device. U. S. Patent No. 3,647,387,
March 7, 1972.
84. Halsted, C. J. and B. A. Thrush. Chemiluminescent Reactions of SO.
Photochem. Photobiol. 4:1007-1013, 1965.
85. Clyne, M. A. A., et^ _al. The Kinetics of Elementary Reactions
Involving the Oxides of Sulphur. Part I, The Nature of the Sulphur
Dioxide Afterglow. Proc. Roy Soc. (London) 295A: 355-362, December
1966.
86. Clyne, M.A.A. and B.A. Thrush, Mechanisms of Chemiluminescent
Combination Reaction Involving Oxygen Atoms. Proc. Roy. Soc (London)
269A: 404-418, 1962.
87. Fontijn, A., ^t al., Absolute Quantum Yield Measurements of the NO-0
Reaction and Its Use as a Standard for Chemiluminescent Reactions.
J. Chem. Phys. 40:64-70, January, 1964.
88. Mettee, H. Foreign Gas Quenching of Sulfur Dioxide Vapor Emission.
J. Phys. Chem. 73: 1071-1076, 1969.
89. Schwarz, F.P. and H. Okabe. Fluorescence Detection of Sulfur Dioxide
in Air at the Parts per Billion Level. Anal. Chem. 46: 1024-1028,
July, 1974.
90. Halsted, C.J. and B.A. Thrush. The Kinetics of Elementary Reactions
Involving the Oxides of Sulphur. Part III, The Chemiluminescent Reaction
between Sulfur Monoxide and Ozone. Proc. Roy. Soc. (London) 295A:
380-397, December , 1966.
91. See Reference 77.
92. Steffenson, D.M. and D.H. Stedman. Optimization of the Operating Para-
meters of Chemiluminescent Nitric Oxide Detectors. Anal. Chem. 46:
1704-1709, December, 1974.
93. Harrison, J.W. and G.E. Spangler. Evaluation of Chemiluminescent Detection
Methods. In: Comparative Evaluation of Trace Gas Technology Volume II-
Analysis and Evaluation of Instrumental Methods, Final Report on Contract
DAAK02-73-C-0128, Harrison, J.W. (ed). Research Triangle Park, N.C.,
Research Triangle Institute, December 1972. AD777868. p. 495-518.
94. Bowman, R.L. and N. Alexander. Ozone Induced Chemiluminescence of
Organic Compounds. Science 154: 1454-1456 , 1966.
95. E. H. White et al "Energy Transfer in Chemiluminescence" pp. 479-492 in
E. C. Lin (ed.)~Tnternational Conference on Molecular Luminescence,
W. A. Benjamin, New York 1969, 910 pp.
176
-------
96. Bersis, D. and E. Vassiliou. A Chemiluminescence Method for Determining
Ozone. Analyst 91: 499-505, 1966.
97. Anderson, H.H. _et^ jil • System and Method of Air Pollution Monitoring
Utilizing Chemiluminescence Reactions. U.S. Patent No. 3,659,100,
April 25, 1972.
98. Stepanova, A.G. et al. Luminescent Method for Determining Ozone in
Gas Mixtures. USSR Patent No. 329,451, February 9, 1972. Cited in
Chem. Abstr. 77, 1972. Abstract No. 42859.
99. . Silicon Photodiode Application Notes, D3000 B-l. Salem,
Mass, EG&G, Inc. Electro-Optic Div., March, 1973.
100. Regener, V.H. Measurement of Atmospheric Ozone with the Chemiluminescent
Method. J. Geophys. Res. 69: 3795, 1964.
101. Tommerdahl, J.B. et al. Chemiluminescent Ozone Meter for Continuous
Air Monitoring Project. Final Report, Contract No. CPA 70-27. Research
Triangle Park, N.C., Research Triangle Institute, January, 1971.
102. Hodgeson, J.A. et al. Chemiluminescent Measurement of Atmospheric Ozone.
Anal Chem 42: 1795-1802, 1970.
103. Guicherit, R. Ozone Analysis by Chemiluminescent Measurement. Z.
Anal. Chem. 256: 177-182, 1971.
104. Lyshkow, N.A. Apparatus and Method for Analysis of Gas Streams.
U.S. Patent No. 3,712,793, January 23, 1973.
105. Luk&c, S. and J. Sevik. A Discussion of the Detection Mechanism and
the Response Character of the Helium Detector. Chromatographia 5:
258-264, 1972.
106. Pellizzari, E.D. Electron Capture Detection in Gas Chromatography.
J. Chromatography 98: 323-361, 1974.
107. Krejci, M. and M. Dressier. Selective Detectors in Gas Chromatography.
Chromatographic Rev. 13: 1-59, 1970.
108. McDaniel, E.W. &t_ al. Ion-Molecule Reactions. New York, Wiley-
Interscience, 1970.
109. Loeb, L.B. Basic Processes of Gaseous Electronics. Berkeley, Calif.,
Univ. of Calif. Press, 1955. p. 8-39.
110. Registered Trademark of Franklin GNO Corp., W. Palm Beach, Fla.
111. McDaniel, E.W. and E.A. Mason. The Mobility and Diffusion of Ions in
Gases. New York, Wiley-Interscience, 1973.
\ 112. Vree, P.H. and A. Fontijn. Method of Determining Trace Amounts of Gases.
\ U.S. Patent No. 3,540,8451, November 17. 1970.
177
-------
113. Murrmann, R.P. et^ al. Vehicle Detection Classification Using Chemical
Sensors. Sepcial Report 181, U.S. Army Cold Regions Research and
Engineering Laboratory, Hanover, N.H. August, 1972. AD751740. 57p.
114. Lovelock, J.E. Improvements in or Relating to the Detector of Trace
Gases. Brit. Patent. No. 1,266,869, March 15, 1972.
115. Kebark, P. Higher-Order Reactions - Ion Clusters and Ion Solvation.
In: Ion-Molecule Reactions, Volume 1, Franklin, J.L. (ed). New
York, Plenum Press, 1972. p. 346.
116. Harrison, J.W. Analysis of the Plasma Chromatograph. In: Comparative
Evaluation of Trace Gas Technology, Volume II-Analysis and Evaluation of
Instrumental Methods, Final Report on Contract DAAK02-73-C-0128,
Harrison, J.W. (ed). Research Triangle Park, N.C., Research Triangle
Institute, December, 1972. AD 777868. p. 41-101.
117- King, W.H., Jr. Piezoelectric Sorption Detector. Anal Chem. 36: 1735-
1739.
118. Frechette, M.W. et al. Evaluation of Substrates for Use on a Piezo-
electric Detector. Anal Chem. 45: 1765-1766, 1973; .
Simple Piezoelectric Probe for Detection and Measurement of S02-
Environ. Sci. Tech. 7: 1135-1137, December, 1973.
119. King, D. and G.R. Hoffman. Observations on a Quartz Crystal
Deposition Monitor. J. Phys. E., Sci. Instr. 4: 993-998, 1971.
120. Lu, Chih-Shun. Improving the Accuracy of Quartz Crystal Monitors.
Research/Development: 45-50, March, 1974.
121. Mason, W.P. Uses of Piezoelectric Crystals and Mechnanical Resonators
in Filters and Oscillators. In: Physical Acoustics Principles and
Methods, Volume I, Part A, Mason, W.P. (ed). New York, Academic
Press, 1964, p. 375.
122. Warner, A.W. MicroWeighing with the Quartz Crystal Oscillator-Theory
and Design. In: Ultra Micro Weight Determination in Controlled Environ-
ments, Wolsky, S.P. and E.J. Zadnauk (eds). New York, Interscience,
1969. p. 137-161.
123. Warner, ibid., p. 141.
124. Saurbrey, G. Schwingquarze zur Wagung dunner Schichten, Z.Phys. 155:
206, 1959 (German)
125. Stockbridge, C.D. Resonance Frequency Versus Mass Added to Quartz
Crystals. In: Vacuum Microbalance Techniques, Behrndt, K. (ed). New
York, Plenum Press, 1966. p. 193-205.
126. Miller, J.G. and D.I. Boleff. Acoustic Wave Analysis of the Operation
of Quartz Crystal Film Thickness Monitors. J. Appl. Phys. 39: 5815-
5816, 1968.
178
-------
127. Mason, W.P. Physical Acoustics and the Properties of Solids. Princeton
D. Van Nostrand, 1958. p. 27-32.
128. Jost, W. Diffusion in Solids, Liquids, Gases, Third Printing with
Addendum. New York, Academic Press, 1960. p. 303.
129. Jost, ibid, p. 300-304.
130. Bracker , J.H. Triethanolamine for Collecting Nitrogen Dioxide in the
TLV Range. Am. Ind. Hygiene Assoc. J. 34: 390-395, September, 1973.
131. Homolya, J.B., Emission Measurements Research Branch, CPL/NERC,
Research Triangle Park, N.C. Private communication.
132. Jost, ibid, p. 465.
133. Linch, A. L. ibid, p. 3.
134. Brauer, L. W., Indicator for Nitrogen Oxide. U. S. Patent 3,388,075,
June 11, 1968.
135. Rakowski, R. F., Colorimetric Dosimeter for Nitrogen Dioxide.
U. S. Patent 3,574,552, April 13, 1971.
136. Linch, A. L. ibid, p. 31
137. Nelson, G. 0. Controlled Test Atmospheres. Ann Arbor, Mich., Ann
Arbor Science Publishers, 1971. p. 45-49.
138. Linch, A. L. ibid, p. 36.
139. Szepesy, L. op. cit., p. 42.
140. Linch, A. L. ibid, p. 19-73.
141. Linch, A. L. ibid, p. 91-105.
142. Palmes, E. D. and A. F. Gunnison, Personal Monitoring Device for
Gaseous Contaminants. Am. Ind. Hygiene Ass. J. 34:78-81, 1973.
143. Linch A. L. ibid, p. 82.
144. Bendix Environmental Science Div., Baltimore.
145. Williams, F. W. and H. G. Eaton. A New Concept in Personal
Samplers. Paper No. 88, presented at the 1975 Pittsburgh
Conference, March 3-7, 1975, Cleveland, Ohio.
146. N. A. Huey. The Lead Oxide Estimation of Sulfur Dioxide Pollution.
J.A.P.C.A. 18:610-611, September, 1968.
147* Boulerice, M. and W. Brabant. New PbO~ Support for the Measurement
\ of Sulfation. J.A.P.C.A. 19:432-434, 1969.
179
-------
148. Kelly, H. J. and L. B. Rogers. Nephelometric Determination of
Sulfate Impurity in Certain Reagent Grade Salts. Anal. Chem.
27:759-764, 1955.
149. Lyshkow, N.A. private communication.
150. Himmelbau, D.M. and K.B. Bischoff. Process Analysis and Simulation.
New York, John Wiley & Sons, 1968.
151. McClenny, W.A., J.A. Hodgeson and J.P. Bell. Photofragment Detection
of Nitrogen Dioxide. Anal. Chem. 45: 1514-1518, July, 1973.
180
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
-fiOQ/2-7fi-fm
3. RECIPIENT'S \CCESSION-NO.
4. TITLE AND SUBTITLE
5. REPORT DATE
1976
DEVELOPMENT STRATEGY FOR POLLUTANT DOSIMETRY
6. PERFORMING ORGANIZATION CODE
7J. W. Harrison, P. A. Lawless
D. E. Gilbert, J. H. White
8. PERFORMING ORGANIZATION REPORT NO.
3. PERFORMING ORGANIZATION NAME AND ADDRESS
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-1731
12. SPONSORING AGENCY NAME AND ADDRESS
.Environmental Sciences Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report comprises the results of a study to: (1) Define and place
realistic limits upon the needs of epidemiologists for personal pollutant dosimeters;
(2) Identify pollutant-sensing principles that offer reasonable opportunity for
early development of functioning dosimeters capable of operating with the limits of
(1) above; (3) Project the impact of mechanical and electronic miniaturization
techniques upon each of the principal options disclosed in (2) above; (4) Assess
the cost/benefit aspects of active vs. passive sampling of the atmosphere, as applied
to candidate sensors identified in (1), (2) and (3) above; (5) Assess the
advantages, if any, of a systems approach to dosimeter development, in which the
design of a sensor for a given pollutant would be constrained to make maximum use of
components & subassemblies common to sensors for other pollutants.
The study reported on has been arbitrarily limited to consideration of devices
applicable to the measurement of ozone, sulfur dioxide and nitrogen dioxide. The
principles developed will apply in some degree to the eventual design of dosimeters
for other pollutants.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
*Dosimeters
Air pollution
Ozone
Sulfur dioxide
Nitrogen dioxide
06R
13B
07B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
192
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
181
-------
|